text
stringlengths 1
1.05M
| meta
dict |
|---|---|
1. Introduction {#sec1-sensors-20-00472}
===============
The relationship between water contents in the plant and yield has been well established \[[@B1-sensors-20-00472],[@B2-sensors-20-00472]\]. The use of standardized parameters based on physiological thresholds that lead to photosynthetic impairment if surpassed have been successfully used \[[@B3-sensors-20-00472],[@B4-sensors-20-00472],[@B5-sensors-20-00472],[@B6-sensors-20-00472]\]. Nevertheless, assessing the water status of a few leaves seldom represents the water status of the field, and thus irrigation decisions would benefit from noninvasive and nondestructive methods covering larger ranges \[[@B7-sensors-20-00472],[@B8-sensors-20-00472]\]. The use of remote sensing for assessing plant water status on different scales, providing near real-time assessments, might become an adequate option. Notwithstanding, the approach has to be tested under field conditions with crops sensitive to water deficiency. Potato (*Solanum tuberosum* L.)---considered to be the third most prevalent edible crop in production after wheat and rice \[[@B9-sensors-20-00472]\]---was used as an example. Its sensitivity to water scarcity (highly associated with a shallow root system) \[[@B10-sensors-20-00472]\] and its cultivation in drought-prone zones caused by climate change \[[@B11-sensors-20-00472]\] are the drivers spurring research to find ways to maximize potato yields while reducing water resources. The use of physiological thresholds has contributed to this goal; thus, Ramírez et al. \[[@B6-sensors-20-00472]\] recommend keeping leaves at maximum stomatal conductance (associated with leaf pore openness and gas exchange) at saturated light \>0.15 mol H~2~O m^−2^ s^−1^ to guarantee appropriate tuber yield whilst saving water. However, the use of this kind of physiological traits for water status inspection requires expensive, exhaustive, invasive, yet microscale (leaf) assessments. Thus, infrared thermography is seen as a promising technology that can detect variations in the canopy temperature ($T_{canopy}$), and ultimately, estimate the crop water stress index (CWSI), an indicator highly correlated with stomatal conductance \[[@B6-sensors-20-00472],[@B8-sensors-20-00472],[@B12-sensors-20-00472]\] (see [Section 2.4](#sec2dot4-sensors-20-00472){ref-type="sec"} for further details).
On the basis of the black body radiation theory, an infrared thermography (IRT) camera can detect the spectral radiance emitted by an object due to its temperature \[[@B13-sensors-20-00472],[@B14-sensors-20-00472],[@B15-sensors-20-00472]\]. Such a radiance, which ranges from 7 to 14 $\mathsf{\mu}$m, is converted into electrical signals and then displayed as a 2D array or IRT image. One of the shortcoming of current IRT image sensors is their typical resolution of 320 × 240 pixels, which precludes appropriate identification of the spatial structure in the captured scene. Unlike its CCD -based visible RGB counterpart, IRT sensors are based on a microbolometer, a sophisticated array of small thermal sensors with a complicated manufacturing process \[[@B16-sensors-20-00472]\]. To overcome this problem, companies recently started to offer IRT camera systems that include a fixed visual (RGB) camera to help identify regions of interest through additional image processing as well as to avoid convoluted image registration methods \[[@B17-sensors-20-00472],[@B18-sensors-20-00472]\].
Numerous agricultural studies have reported the use of infrared thermography along with particular acquisition methodologies to analyze different plant parameters and conditions. For example, Pitarma et al. \[[@B19-sensors-20-00472]\] evaluated tree health through the use of IRT images. In this study, identification of healthy and unhealthy tissues in the tree's surface was performed through an analysis of infrared thermographic data. Similarly, in \[[@B20-sensors-20-00472]\], a review of pest detection applications based on infrared thermography was reported. This comprehensive study includes the detection and identification of the main pest in plant crops such as maize, rice, and soybean. With regard to water-stress estimation in plants, Möller et al. \[[@B21-sensors-20-00472]\] conducted an investigation on the water status of grapevines using thermal and visible images acquired from a platform located 15 m above the canopy (nadir-pointing view). This study aimed to analyze the correlation between the calculated CWSI and the measured leaf stomatal conductance, resulting in a correlation of $R^{2} = 0.91$. Image processing steps in the estimation of $T_{canopy}$ include image registration (i.e., alignment of multisensor images), which requires the placement of aluminum crosses over the field, and the selection of canopy pixels in the thermal image based on the transformation of the RGB image to hue-saturation-intensity color space followed by a manual thresholding procedure. A similar study was conducted in \[[@B22-sensors-20-00472]\] for cotton crops where leaf water potentials showed a linear relationship with CWSI values ($R^{2} = 0.816$). Here, a nadir-pointing view IRT camera was located 5 m above the crops. Since there was no additional sensor to detect where the canopy was located in the scene, canopy pixels detection was based solely on estimated temperature thresholds, thus avoiding the use of an additional sensor. In the context of potato crops, Prashar et al. \[[@B23-sensors-20-00472]\] implemented a high-throughput field phenotyping methodology to estimate CWSI using IRT images. Image acquisition was performed from a fork-lift at a height of about 8 m, with an angular inclination to cover up to 27--36 plots in the scene. Then, manual selection of the central regions in the plots was performed to estimate the CWSI. Similarly to the previous study, the inclusion of canopy pixels in the analysis was based on temperature thresholding.
Consequently, several factors can influence the accurate estimation of $T_{canopy}$ and the subsequent calculation of the CWSI. First, the use of an independent, nonfixed sensor imposes an image registration problem where convoluted image processing algorithms could be required. The use of control points in the scene could alleviate the computational load while increasing the overall acquisition time. Second, the optical setup, involving the camera viewing angle ($\theta$) and the distance to the object ($d_{object}$), defines the canopy scene to be analyzed, and an appropriate combination of such parameters is needed to reduce the number of nonleaf pixels. Third, since the pixel resolution of thermal cameras is notably low, a single pixel can detect both soil and leaf thermal radiation such that thresholding based solely on temperature can yield a high level of uncertainty in the estimation of $T_{canopy}$. Finally, the studies mentioned above do not provide a detailed explanation regarding the various image processing procedures involved in the use of thermal images; this is possibly due to the dependence on commercial software.
This research has the following specific objectives: (1) To provide a thorough description of infrared thermography for the estimation of canopy temperature, and subsequently the CWSI, in the context of potato production; (2) to generate an open-source software to perform such tasks that can be freely used by the agricultural remote sensing community; (3) to demonstrate that thermographic sensors along with the described acquisition methodology are appropriate for defining irrigation thresholds, which can reduce water consumption.
2. Materials and Methods {#sec2-sensors-20-00472}
========================
2.1. Plant Material and Study Area {#sec2dot1-sensors-20-00472}
----------------------------------
The potato cultivar used was UNICA (CIP code: 392797.22), an early variety adapted to warm and dry environments and slightly tolerant to salinity \[[@B24-sensors-20-00472]\]. Two field trials were performed at the International Potato Center (CIP) and the National Agrarian University---La Molina (UNALM) experimental stations in Lima, Peru (12.08° S, 76.95° W, 244 m.a.s.l.) during October 2017--January 2018 (first experiment---CIP, E1) and June--September, 2018 (second experiment---UNALM, E2). The study site is characterized by a semi-warm and humid climate \[[@B25-sensors-20-00472]\].
During the growing season, the minimum temperature ($T_{min}$), maximum temperature ($T_{max}$), relative humidity ($RH$), solar radiation ($R_{s}$), and maximum vapor pressure deficit ($VPD_{max}$, estimated using the equation used by Ramírez et al. \[[@B26-sensors-20-00472]\]) were 16.4 ± 0.19 °C, 23.2 ± 0.23 °C, 86.5 ± 0.45%, 16.0 ± 0.43 MJ m^−2^ day^−1^, and 1.06 ± 0.03 kPa, respectively (in E1); and 14.2 ± 0.05 °C, 18.7 ± 0.19 °C, 89.7 ± 0.04%, 7.5 ± 0.46 MJ m^−2^ day^−1^, and 0.64 ± 0.03 kPa, respectively (in E2). Specific values for environmental conditions per experimental period are provided in [Table 1](#sensors-20-00472-t001){ref-type="table"} and [Table 2](#sensors-20-00472-t002){ref-type="table"}.
2.2. Experimental Design and Crop Management {#sec2dot2-sensors-20-00472}
--------------------------------------------
The experimental units (EU) were plots of 3.6 × 12.5 m^2^ (with 120 plants) and 4.5 × 15.8 m^2^ (with 180 plants) in E1 and E2, respectively. Since the experiments were conducted in two facilities with different total areas but identical weather conditions, the plot sizes are different. However, to overcome any variability in the results, we adopted the same population density of 3.7 plants m^−2^ per plot in both cases. A completely randomized design, with three irrigation timing treatments repeated in five EUs, was established in both experiments. Two water restriction levels defined with thresholds values of CWSI (T2: CWSI \< 0.4 and T3: CWSI \< 0.7) were compared against a control (T1: fully irrigated). The reader is referred to [Section 2.4](#sec2dot4-sensors-20-00472){ref-type="sec"} for a detailed description of the CWSI. The water restriction started when the tuber initiation occurred (33 and 31 days after planting (DAP) in E1 and E2, respectively). In both experiments, the drip irrigation system was comprised of a nonpressure compensating emitter and distributed drip-tapes (Toro Aqua-Traxx PBX). There were two drip-tapes per furrow separated by 0.2 m. In addition, the emitter flow rate (the distance between emitters was 0.20 m) was 1.27 L h^−1^ at 0.055 MPa, and the manufacturing coefficient of the variation of the emitters was below 3%. Irrigation pulses (up to field capacity) were supplied to maintain the soil moisture above 70% of field capacity (in T1), and each time the crop reached the established threshold value of CWSI (0.4 and 0.7 in T2 and T3, respectively).
The fertilization application (N:P~2~O~5~:K~2~O) consisted of 180:100:160 and 160:80:180 kg ha^−1^ supplied as NH~4~NO~3~:(NH~4~)~2~HPO~4~:K~2~SO~4~ and NH~4~NO~3~:H~3~PO~4~:K(NO~3~)~2~ in E1 (at sowing and hilling) and E2 (in four periods from emergency to tuber initiation), respectively. Additionally, doses of CaO:MgO (60:30 kg ha^−1^) were supplied as Ca(NO~3~)~2~:Mg(NO~3~)~2~·6H~2~O in E2. Chromatic and pheromone traps were used as the ethological control in both experiments. Recommended doses of chemical products with different mechanisms of action were applied in E1: 0.50 L ha^−1^ of Movento 150 OD (Bayer AG, Leverkusen, Germany), 0.55 L ha^−1^ of Sorba 50 EC (Farmagro, Lima, Peru), 0.60 kg ha^−1^ of Evisetc-S (Arysta Life Science, PA, USA), and 0.15 kg ha^−1^ of Trigard 75 WP (Farmagro, Lima, Peru). These were rotated weekly from 30 to 75 DAP. In E2, a single chemical application composed of 0.40 L ha^−1^ of Confidor 350 SC (Bayer AG, Leverkusen, Germany) was performed at 66 DAP.
2.3. Image Acquisition and Analysis {#sec2dot3-sensors-20-00472}
-----------------------------------
The thermal camera used in the experiments was the FLIR E60 (FLIR Systems Inc., Täby, Sweden), which is a specialized device for hand-held acquisition due to its robust and light-weight design. It includes both IRT and RGB sensors. The IRT camera lens has an angular field-of-view (FOV) of 25° × 19° and focal length (*f*) of 18 mm. The IRT sensor has a spatial resolution of 320 × 240 pixels, a thermal sensitivity of \<0.05 °C in the 7.5--13 $\mathsf{\mu}$m spectral range, a thermal response time of ∼8--12 ms, and an accuracy of ±2 °C or ±2% when reading for an ambient temperature from 10 to 35 °C. The RGB camera lens has an angular FOV of 53°× 41°, and the RGB sensor has a spatial resolution of 2048 × 1536 pixels.
Each pair of images was manually acquired with an object distance $d_{object}$ of ≈3 m (defined as the distance between the camera's lens and the plant canopy), a horizontal inclination or viewing angle $\theta\mspace{600mu} \approx \mspace{600mu} 60^{{^\circ}}$, with respect to the zenith and in position opposite to the sun (see [Figure 1](#sensors-20-00472-f001){ref-type="fig"}A). Viewing angle $\theta$ selection relied on the following: (1) The assumption that the potato canopy behaves as a reflective uniform surface wherein higher emission can be acquired when $\theta$ is similar to the angle of incidence of solar rays ($\theta_{i}$); (2) the fact that the most suitable period for water status characterization is around 14:00 \[[@B8-sensors-20-00472]\]; and (3) the need for having 75% of the potato canopy in the scene when combined with IRT's FOV and $d_{object}$. Finally, the acquisition was conducted between 13:00 and 15:00 ($\theta_{i} \in \left\lbrack 60,80 \right\rbrack^{{^\circ}}$) with a constant $\theta$ and during clear-sky days to meet the previous requirements and ensure reproducibility. As a result, six plants (effective area ≈ 1 m^2^) were captured in the IRT scene. The acquisition period was previously determined in \[[@B8-sensors-20-00472]\], in which leaf temperature and stomatal conductance were monitored during the entire day (from 07:00 to 18:00) for the same variety of potato (UNICA), comparing well-irrigated and water-stressed potatoes. It was also demonstrated that this variety showed a strong stomatal sensitivity closure after noon due to abrupt changes in the meteorological variables: VPD and solar radiation.
The total thermal radiation received at the detector includes both the radiation emitted by the object ($T_{object}$) and the radiation originating in the surroundings, which is reflected by the object ($T_{reflected}$) \[[@B27-sensors-20-00472]\]. Therefore, the estimated temperature from the IRT cameras needs to be corrected to obtain an accurate value for $T_{object}$. The correction procedure, which is explained below, relies on the a priori estimation of $T_{reflected}$. For this, an IRT image from a low-emissivity cardboard panel with a uniform surface was acquired before each acquisition campaign and the value of $T_{reflected}$ was estimated by strictly following the provided instructions in the FLIR E60 user manual. Additionally, the calculation of CWSI (explained in [Section 2.4](#sec2dot4-sensors-20-00472){ref-type="sec"}) requires the measurement of the temperature of a wet leaf reference ($T_{wet}$) \[[@B28-sensors-20-00472]\]. As recommended in \[[@B6-sensors-20-00472],[@B21-sensors-20-00472],[@B29-sensors-20-00472],[@B30-sensors-20-00472]\], a 1 mm double piece white cotton cloth around a piece of polystyrene foam floating in a 0.32 × 0.22× 0.10 m^3^ plastic tray was used as an artificial wet reference surface (AWRS) ([Figure 1](#sensors-20-00472-f001){ref-type="fig"}C). It was present in every IRT image acquired in this study (see [Figure 1](#sensors-20-00472-f001){ref-type="fig"}B).
The proposed methodology to align the RGB and IRT images, extract the canopy area, and calculate the average temperature of the potato canopy in the scene is depicted in [Figure 2](#sensors-20-00472-f002){ref-type="fig"}. For this, let the RGB visible image be defined as $I_{RGB} \in \mathbb{Z}_{+}^{W_{RGB} \times H_{RGB} \times 3}$ and the IRT image as $I_{IRT} \in \mathbb{R}_{+}^{W_{IRT} \times H_{IRT}}$, where $W_{RGB} \times H_{RGB}$ = $2048 \times 1536$ pixels and $W_{IRT} \times H_{IRT}$ = $320 \times 240$ pixels are their resolutions, respectively.
1. Geometric Transformation: The IRT and RGB images acquired with the FLIR E60 camera have a horizontal and vertical displacement vector $\left( b_{x},b_{y} \right)$ (units in pixels) due to the shift between both image sensor's optics. In addition, there is an additional scaling factor $\left( S_{x},S_{y} \right)$ (unitless) as a result of the different lenses' FOV and the corresponding sensors' spatial resolution (pixel quantity and size). Such parameters can be calculated a priori using a one-time calibration protocol based on the methodology described in \[[@B13-sensors-20-00472]\]. Fifty pair of IRT and RGB images acquired with $\theta\mspace{600mu} \approx \mspace{600mu} 60^{{^\circ}}$ and $d_{object}\mspace{600mu} \approx \mspace{600mu} 3$ m over the potato crop, and including the AWRS in the scene, were utilized for calibration. A MATLAB script was developed to allow the user to manually select correlated control points in both images through visual inspection, and consequently, provide averaged displacement vector and scaling factors ([Figure 3](#sensors-20-00472-f003){ref-type="fig"}). The presence of the AWRS in both images serves to locate additional control points due to its higher contrast (see [Figure 3](#sensors-20-00472-f003){ref-type="fig"}). The number of image pairs (50) helps overcome the inherent human error in the selection, and thus, reduce the measurement uncertainty. The resulting scaling factor $\left( S_{x},S_{y} \right) = \left( 0.367,0.375 \right) \pm 5.40\%$ is firstly applied to the RGB image ($I_{RGB}$). Bi-linear interpolation in $I_{RGB}$ is allowed since it only serves to determine the canopy location in the scene. Then, the displacement vector $\left( b_{x},b_{y} \right) = \left( 219,189 \right) \pm 2.33\%$ is applied to the IRT image ($I_{IRT}$) with respect to $I_{RGB}$. The coordinate origin $\left( 0,0 \right)$ is assumed to be the top left corner of $I_{RGB}$. Furthermore, no scaling was applied, i.e., the interpolation process is carried out over $I_{IRT}$ to avoid altering the IRT sensor measurements. [Figure 2](#sensors-20-00472-f002){ref-type="fig"} (Step 1) shows the resulting scaled $I_{RGB}$ and translated $I_{IRT}$ as a single false-color image, which also indicates the overlapping region to be analyzed.
2. Color-Based Threshold Calculation: The green--red vegetation index (GRVI) is used to determine the presence of the potato canopy in the scene. As was thoroughly explained and assessed in \[[@B31-sensors-20-00472]\], GRVI can serve as a threshold to determine the leaves' location in an RGB image. After the geometric transformation procedure is performed, the overlapped region from the scaled RGB image is extracted; it is referred as $I_{RGB}^{*}$ from now on. Then, the GRVI is calculated for each pixel ($h,w$) using the following equation: $$GRVI = \frac{I_{RGB}^{*}\left( h,w,2 \right) - I_{RGB}^{*}\left( h,w,1 \right)}{I_{RGB}^{*}\left( h,w,2 \right) + I_{RGB}^{*}\left( h,w,1 \right)},$$ where $I_{RGB}^{*}\left( h,w,2 \right)$ indicates the pixel located at the 2D coordinate $\left( h,w \right)$ in the $green$ layer ($red = 1$, $green = 2$ and $blue = 3$). In \[[@B31-sensors-20-00472]\], a value of $GRVI$ = 0 can detect the early phase of leaf green-up over several forest species, such as deciduous broadleaf and deciduous coniferous, as well as grassland and a rice paddy field. In this study, additional experiments were conducted with the potato canopy, and it was found that a threshold value of $GRVI = 0.04$ can determine the presence of the potato canopy with a high level of accuracy when it is not affected by the crop senescence. Finally, the logical mask $M \in \left\{ 0,1 \right\}^{320 \times 240}$ is generated and contains 1s when a given pixel yields a $GRVI \geq 0.04$ and 0 s otherwise.
3. Morphological Operations: As a result of the linear interpolation of the RGB image $I_{RGB}$, individual pixels do not conserve their spectral information, and the GRVI cannot detect them as part of the canopy, ultimately creating small holes (a group of 0s) in the mask *M*. For this reason, a set of mathematical morphological operations is used. First, dilation is applied to eliminate the noiselike structures over the potato canopy. Second, erosion is used to fully cover those regions that do not belong to the canopy, as shown in see [Figure 2](#sensors-20-00472-f002){ref-type="fig"} (Step 3). Only those values with 1 s (white in the image) are used to calculated the averaged $T_{canopy}$. Additionally, since the size of the image is fixed (320 × 240 pixels), a kernel size of 4 × 4 pixels was used for both operations. Finally, small regions with $1/10$ of the total mask area are removed.
4. Correction with the FLIR Metadata and Average Temperature Calculation: The IRT image $I_{IRT}$ provided by FLIR E60 has units of Kelvin and its generation considers the total IR radiation that reached the detector during acquisition. Such radiation is composed of two components: the thermal radiation originated from the object and the radiation originating in the surroundings and reflected by the object. The fraction of the reflected radiation depends on the emissivity of the object $\varepsilon$, specifically, when $\varepsilon < 1$, and should be removed from the measurement \[[@B7-sensors-20-00472]\]. In order to perform the correction, the FLIR E60 provides the estimated total temperature in raw 16-bit format *S* and additional factory calibration parameters. Additionally, $T_{reflected}$ and $\varepsilon = 0.96$ \[[@B6-sensors-20-00472],[@B8-sensors-20-00472]\] are utilized to estimate $T_{object}$ as follows: $$RAW_{reflected} = \frac{R_{1}}{R_{2}\left( exp\left( B/T_{reflected} \right) - F \right)} - O,$$ $$RAW_{object} = \frac{S - \left( 1 - \varepsilon \right)RAW_{reflected}}{\varepsilon},$$ $$T_{object} = \frac{B}{\ln\left( R_{1}/\left( R_{2}\left( RAW_{object} + O \right) \right) + F \right)},$$ where $R_{1}$, $R_{2}$, *F*, *O*, and *B* are the so-called Planck calibration constants for FLIR cameras (and are set upon manufacturing), which can be extracted by the publicly-available EXIFTOOL application from the IRT image metadata. Furthermore, $T_{reflected}$ is the reflected temperature in Kelvin, $RAW_{reflected}$ is a 16-bit calculated value, *S* is the 16-bit raw value provided by the FLIR R60 camera, and $T_{object}$ is the object temperature in Kelvin. This procedure is performed for each pixel in the IRT image. Such values are then multiplied with the logical mask *M* and averaged to obtain the average canopy temperature $T_{canopy}$. It is noteworthy that the noise likely induced by manual acquisition is highly reduced through the averaging of multiple pixels in five (5) IRT canopy images for a given treatment.
The selected emissivity value $\epsilon = 0.96$ was thoroughly studied in \[[@B6-sensors-20-00472],[@B8-sensors-20-00472],[@B32-sensors-20-00472]\], and in this study, it was considered constant throughout the complete growing season. As pointed out by Usamentiaga et al. \[[@B33-sensors-20-00472]\], the emissivity of a real object is variable and dependent on wavelength; however, it can be assumed constant when the wavelength interval is short. Since the IRT image spectral width is 6 $\mathsf{\mu}$m, the assumption of it being constant $\epsilon$ remains valid. Finally, the described algorithm has been implemented as a software package named the "Thermal Image Processor" (TIPCIP) using QT Creator v5.0 for Windows operating system and under an open-source and free-access policy. The download links for the executable and source files are included in the Reference section.
2.4. Response Variables {#sec2dot4-sensors-20-00472}
-----------------------
In E1, six thermocouples (TT-T-36-SLE, Omega Engineering Inc., Manchester, UK) with an accuracy greater than 0.5 °C or 0.4% (above 0 °C) were attached to each plot into the abaxial surface of the leaf center of target plants using surgical tape \[[@B34-sensors-20-00472]\]. In both experiments, thermal images were acquired (interdaily from 34 to 82 DAP) and processed according to the methodology described in [Section 2.3](#sec2dot3-sensors-20-00472){ref-type="sec"}. Data from thermocouples and IRT images correspond to the same plots. Canopy temperature data from IRT images were used to estimate the CWSI following the empirical method used for potatoes in \[[@B12-sensors-20-00472]\]:$$CWSI = \frac{T_{canopy} - T_{wet}}{T_{dry} - T_{wet}},$$ where $T_{canopy}$ is the measured canopy temperature, $T_{wet}$ is the AWRS measured temperature, and $T_{dry}$ is the artificial dry reference surface estimated temperature. $T_{dry}$ was considered 13 °C \[[@B8-sensors-20-00472]\] and 7 °C \[[@B6-sensors-20-00472],[@B12-sensors-20-00472]\] over the air temperature in E1 (hot season) and E2 (wet season), respectively.
Before the irrigation was applied in each treatment, soil samples were collected at 0--0.35 m depth (where \>80% of root-zone biomass is concentrated) in each plot to estimate volumetric water content ($\theta_{v}$, in %) according to Ramírez et al. \[[@B6-sensors-20-00472]\]. The irrigation time and irrigated water quantity (IWQ, in mm) was estimated in each treatment from $\theta_{v}$, $\theta_{v}$ at field capacity (32.7% and 28.4% in E1 and E2, respectively), root-zone width (0.40 m), root-zone depth (0--0.35 m), and drip-tape flow rate (8.5 and 7.7 L m^−1^ h^−1^ at 0.05 MPa in E1 and E2, respectively) (see the equations in \[[@B6-sensors-20-00472]\]). Before treatment onset, five intense irrigation pulses were realized by furrow irrigation in E1 (in all treatments) at 1, 8, 12, 19, and 25 DAP to every plot with ≈ 36 mm per irrigation. In this case, the irrigated water volume was estimated as the product of the flow of the main water channel of CIP's experimental station (≈0.75 m^3^ min^−1^) and the time required to flood each plot. In the case of E2, the drip-tape irrigation system was utilized as this facility (UNALM) did not have a water channel for crop flooding. Thus, the irrigation (in all treatments) before tuber initiation (31 DAP) occurred when soil moisture achieved 70% of field capacity. A uniform-distributed irrigation utilized $1131.6$ m^3^ ha^−1^ of water.
Total IWQ (IWQ~*T*~) is the sum of IWQ (each irrigation) during the growing season. Four center plants (in which the CWSI evaluations were performed) were individually harvested and separated into leaves, stems, and tubers at 102 and 92 DAP in E1 and E2, respectively. Fresh tuber yield (FTY, in t ha^−1^) was calculated from the average value of tubers biomass per plant and plant density.
2.5. Statistical Analysis {#sec2dot5-sensors-20-00472}
-------------------------
The effects of irrigation treatments on CWSI for each assessment were tested with one-way ANOVA. Fisher's least significant difference (LSD) test was performed to determine differences among irrigation treatments on FTY and IWQ~*T*~. Pearson correlation coefficient ($r_{Pearson}$) was calculated to determine the accuracy of leaf temperature estimations. The significance of all statistical tests were assessed at *p* \< 0.05 and *p* \< 0.01 using R v3.6.1 software \[[@B35-sensors-20-00472]\].
3. Results {#sec3-sensors-20-00472}
==========
3.1. Accuracy of the Canopy Temperature Estimations {#sec3dot1-sensors-20-00472}
---------------------------------------------------
The range of values of $T_{canopy}$ recorded by the thermocouples was between 23.0 and 39.6 °C. The overall average values of $T_{canopy}$ estimated by the TIPCIP software showed a positive linear relationship with the average $T_{canopy}$ values by thermocouples ($r_{Pearson}$ = 0.84, *p*-value \< 0.01) ([Figure 4](#sensors-20-00472-f004){ref-type="fig"}). In addition, 87.0% and 82.6% of the $T_{canopy}$ values using TIPCIP, corresponding to the T1 and T2 treatments, underestimated $T_{canopy}$ as compared to the thermocouple values, in a range from −2.05 to −0.15 and −3.27 to −0.61 °C, respectively ([Figure 4](#sensors-20-00472-f004){ref-type="fig"}). Furthermore, 60.9% of T3 treatment values overestimated $T_{canopy}$ as compared to the thermocouple values, in a range of +0.21 to +3.80 °C ([Figure 4](#sensors-20-00472-f004){ref-type="fig"}). The accuracy of $T_{canopy}$ estimations was better in T1 ($r_{Pearson}$ = 0.90, *p*-value \< 0.01), followed by T2 ($r_{Pearson}$ = 0.85, *p*-value \< 0.01), and T3 ($r_{Pearson}$ = 0.83, *p*-value \< 0.01).
3.2. CWSI and Irrigation Treatments through the Growing Period {#sec3dot2-sensors-20-00472}
--------------------------------------------------------------
During E1, 14, 9, and 1 irrigation treatments of between 4 and 14 mm, 5 and 16 mm, and 25 mm ([Figure 5](#sensors-20-00472-f005){ref-type="fig"}A) were established for T1, T2, and T3, respectively. The maximum average value of IWQ~*T*~ and FTY in this trial were 3238.7 ± 95.80 m^3^ ha^−1^ and 38.9 ± 5.94 t ha^−1^ corresponding to T1 ([Figure 6](#sensors-20-00472-f006){ref-type="fig"}A,B) with no significant (*p*-value \> 0.05) reduction in FTY with T2 of 34.5 ± 3.72 t ha^−1^, which showed a −10.5% decrease in IWQ~*T*~. T3 significantly (*p*-value \< 0.05) reduced by −37.2% and −61.4% the IWQ~*T*~ and FTY, respectively, in comparison to T1 ([Figure 6](#sensors-20-00472-f006){ref-type="fig"}A,B). During E2, 17, 10, and 4 irrigation treatements of between 1.7 and 9.2 mm, 3.7 and 18.7 mm, and 3.7 and 16.5 mm ([Figure 5](#sensors-20-00472-f005){ref-type="fig"}B) were established for T1, T2, and T3, respectively. The maximum average value of IWQ~*T*~ and FTY in this trial was 1738.5 ± 44.35 m^3^ ha^−1^ and 49.5 ± 3.37 t ha^−1^ achieved by T1. T2 showed a no significant (*p*-value \> 0.05) reduction in FTY in comparison to T1 of 44.3 ± 2.66 t ha^−1^, but showed a significant −29.7% (*p*-value \< 0.05) reduction in IWQ~*T*~ ([Figure 6](#sensors-20-00472-f006){ref-type="fig"}C,D). T3 significantly (*p*-value \< 0.05) reduced by −55.0% and −39.2% in terms of the IWQ~*T*~ and FTY, respectively, in comparison to T1 ([Figure 6](#sensors-20-00472-f006){ref-type="fig"}C,D).
The CWSI values ranged between 0.25 and 0.64, 0.31 and 0.60, and 0.39 and 0.82 (in the hot season---E1; [Figure 5](#sensors-20-00472-f005){ref-type="fig"}A), and 0.14 and 0.59, 0.12 and 0.73, and 0.12 and 1.0 (in the wet season---E2; [Figure 5](#sensors-20-00472-f005){ref-type="fig"}B) in T1, T2, and T3 respectively. The average value of CWSI before the irrigation treatments in T1, T2, and T3 were 0.34 ± 0.02, 0.45 ± 0.02, and 0.72 ± 0.0 (in E1); and 0.34 ± 0.04, 0.56 ± 0.05, and 0.65 ± 0.03 (in E2), respectively. In E1, the CWSI values were close to the established threshold (0.4 and 0.7 for the T2 and T3 treatments, respectively). In E2, the CWSI values exceeded the threshold after 60 DAP in the T2 treatments (five of nine measurements; [Figure 5](#sensors-20-00472-f005){ref-type="fig"}B) and T3 (two of nine measurements; [Figure 5](#sensors-20-00472-f005){ref-type="fig"}B).
4. Discussion {#sec4-sensors-20-00472}
=============
4.1. Acquisition Configuration and Comparison with Instrumental Measurements {#sec4dot1-sensors-20-00472}
----------------------------------------------------------------------------
With the advent of relative inexpensive thermographic imager systems such as those that include both IRT and RGB sensors, the analysis of larger crop areas can be attained with a single pair of images. The additional RGB sensor provides the capability to determine specific structures in the scene through appropriate image processing techniques. Notwithstanding, several factors have to be considered when working with these images, such as the field-of-view of the optical system, the distance to the object, and the viewing angle, as they could insert uncertainty in the temperature estimation.
The selection of an adequate viewing angle requires the analysis of the three-dimensional structure of the canopy. For a highly heterogeneous canopy, the best viewing angle is parallel to the solar beam line-of-sight (LOS) since the image will mostly include well-illuminated leaves and reduce the number of shaded, cooler leaves that ultimately would affect the estimation of $T_{canopy}$ \[[@B7-sensors-20-00472]\]. On the contrary, a potato canopy, during 30 to 80 DAP, can be assumed to be an almost planar surface where leaves have quasi-uniform heights from the floor. The selected viewing angle of 30° inclination from the horizon not only allows one to acquire this uniform structure in a single image with a reduced number of shaded leaves, but also prevents the presence of shadows in the scene, which can appear if the image is taken parallel to the solar LOS. In addition, since the fixed camera's FOV is narrow, substantial variations in the reflected emission from different parts of the generated image are not expected. However, the assumption of a planar canopy is not valid for the days before 30 DAP as a result of the growth stage, nor for days after 80 DAP as a result of the senescence period. It is noteworthy that Luquet et al. \[[@B36-sensors-20-00472]\] investigated the impact of the viewing angle on the estimated temperature of different canopies, presenting variations of up to 1.5 °C for viewing angles higher than 45° with respect to the ground plane normal and opposite to the sun. According to \[[@B37-sensors-20-00472]\], such reported temperature variations could be governed by the changing proportion of background (soil in the scene) viewed at different angles. As this is still an open discussion, further analysis should be done for plant canopies whose 3D structure significantly deviates from planar.
The spatial resolution (i.e., the size of a resulting pixel in the image typically in units of centimeters or meters) varies with the camera--object distance $d_{object}$. Short distances yield a more extensive area of coverage for a pixel, which could include both canopy and soil, thereby hindering the correct temperature estimation. This issue is aggravated when a nadir view is utilized to acquire the images from a sparse crop plot \[[@B37-sensors-20-00472]\]. Subpixel methods, which can estimate the fraction of the pixel only occupied by the leaf, have been thoroughly studied in \[[@B38-sensors-20-00472]\]; however, the high complexity of the algorithm and the computational cost of its implementation limit its utility for campaigns that can span several months. In this study, the use of a viewing angle with inclination from the horizon along with a $d_{object} \approx$ 3 m and the given IRT FOV resulted in $> \approx 75\%$ of pixels with canopy. Although the image acquisition procedure was conducted manually, instead of using a fixed platform, a high correlation between thermographic estimated temperatures and in-situ measurements from thermocouples was attained. Such a result can be leveraged by newly introduced thermographic systems, such as the FLIR ONE (FLIR Systems Inc.,Täby, Sweden) in which a thermal camera can be attached to a smartphone. In this context, this research also hopes to make the technology available to smallholder farmers, who can acquire these inexpensive devices, acquire data manually, and use the developed software to, ultimately, conduct simple irrigation management.
Additionally, a comparative study between thermography and in-situ measurements was conducted in this research. It is noteworthy that physical techniques for measuring leaf temperatures are different. Thus, thermocouples utilize thermal conduction through physical contact with the abaxial surface. Transmission of heat from the plant towards the sensor is carried out through the movement of excited electrons from the plant to the sensor, which are ultimately converted into voltage and mapped to a temperature value. On the other hand, infrared thermal cameras measure thermal emissions from the adaxial surface, which is exposed to the convection effect (air circulation) and direct incident light. Through the Planck's law equation, IR radiation is then converted into temperature values. Although a large temperature gradient between adaxial and abaxial surface due to its millimeter thickness is not expected, the use of different methods imposes additional uncertainty, which is finally revealed in the correlation calculation.
4.2. Thermography Usefulness for Irrigation Scheduling in Potato Crops in Humid Environments {#sec4dot2-sensors-20-00472}
--------------------------------------------------------------------------------------------
Humid environments, characterized by low VPD, impose technical (emission detection, image processing of thermal and visible images; \[[@B37-sensors-20-00472]\]) and physiological (stomatal openness sensitivity detected in the potato; \[[@B8-sensors-20-00472]\]) challenges for plant water status using thermography. The dry temperature in the CWSI calculation represents the temperature that no transpired leaves achieve \[[@B28-sensors-20-00472]\], and from an empirical perspective ($sensu$ \[[@B12-sensors-20-00472]\]), it is calculated adding a *X*-value to the air temperature (Equation ([5](#FD5-sensors-20-00472){ref-type="disp-formula"})). *X*-value can be experimentally determined as the maximum temperature achieved by leaves previously covered with petroleum jelly to avoid transpiration \[[@B28-sensors-20-00472]\]. For the potato, seven \[[@B6-sensors-20-00472],[@B12-sensors-20-00472]\] and 13 °C \[[@B8-sensors-20-00472]\] have been used as *X*-values in different areas. The potato has an acute stomatal closure sensitiveness in humid areas during the day, depending on various thresholds of VPD and radiation, which promotes an important increase in leaf/canopy temperature in relation to the atmospheric temperature \[[@B8-sensors-20-00472]\]. In this study, different *X*-values were used in the same area depending on the season, which was characterized by different average values of VPD and radiation. Thus, the hot and wet season showed values (99% of the data) of solar radiation of 14.2--17.9 and 3.7--6.2 MJ m^−2^ day^−1^, with average VPD values (during the evaluations) of 0.90 ± 0.01 and 0.39 ± 0.01 kPa; 13 and 7 °C being the *X*-values used for each season, respectively. Because of the close relationship between the "standard" and maximum stomatal conductance at light-saturated and tuber yield, CWSI has been proposed as a good indicator of water status in potato crops \[[@B6-sensors-20-00472],[@B8-sensors-20-00472],[@B12-sensors-20-00472],[@B39-sensors-20-00472]\]. In agreement with other works \[[@B6-sensors-20-00472],[@B8-sensors-20-00472]\], this study confirms the use of CWSI \<0.4 as a threshold for irrigation in potato crops, which allowed us to save water (341.6 ± 63.7 and 515.7 ± 37.7 m^3^ ha^−1^ in E1 and E2, respectively) without a significant (*p*-value \> 0.05) reduction in tuber yield. Our study area belongs to the South American arid diagonal \[[@B40-sensors-20-00472]\], a hyper-arid zone distributed along the Peruvian coast, characterized by a higher atmospheric humidity during the winter season (May--August) promoted by fogs that come from the Pacific Ocean \[[@B41-sensors-20-00472],[@B42-sensors-20-00472],[@B43-sensors-20-00472]\]. This wet season is the most appropriate for potato production in this zone \[[@B44-sensors-20-00472]\], and the most challenging for the use of thermography methodology (see discussion in [Section 4.1](#sec4dot1-sensors-20-00472){ref-type="sec"}). Our second trial was carried out during this season, obtaining the potential tuber yield reported for the UNICA potato variety in the literature (50 t ha^−1^; \[[@B24-sensors-20-00472]\]) with a significant (*p*-value \< 0.05) water saving (515.7 m^3^ ha^−1^) in relation to the control using 0.4 CWSI threshold value (T2, [Figure 6](#sensors-20-00472-f006){ref-type="fig"}D). In comparison to the hot season trial, the wet season was monitored three times per week (Mondays, Wednesdays, and Fridays), and in some cases, our target thresholds were mainly surpassed after 60 DAP ([Figure 5](#sensors-20-00472-f005){ref-type="fig"}B) i.e., after maximum canopy cover and during the senescence stage. Ramírez et al. \[[@B6-sensors-20-00472]\] caution to avoid potato plants close or above a CWSI of 0.6 (known as the "severity threshold") because of the potential of affecting tuber yields through oxidative damage. However, it seems that this severity threshold could be achieved after maximum canopy cover or during senescence in the wet season without affecting the tuber yield. The Peruvian central coast will be likely affected by an increase in temperatures in the future because of global warming \[[@B45-sensors-20-00472]\]. The hot season trial simulated the potato production under these future scenarios, which will demand more water and potential yields in this crop will be difficult to achieve.
4.3. Advantages and Disadvantages of the Developed Software (TIPCIP) {#sec4dot3-sensors-20-00472}
--------------------------------------------------------------------
Several agricultural analyses based on IRT imagery and image processing did not provide a user-level software; this is because the authors were mainly concerned about validating the physiological meaning of IRT-based treatments \[[@B21-sensors-20-00472],[@B23-sensors-20-00472],[@B37-sensors-20-00472]\]. As a result, similar software packages were implemented but not distributed to the scientific community. In this context, this research not only reveals the various steps involved in canopy identification and temperature calculation using IRT images, but also offers the open-source TIPCIP software to perform those tasks. Hence, TIPCIP allows the user to automatically identify the canopy section in the scene and provides the average temperature, processing up to 50 image pairs (RGB-IRT) in a single run, and reporting formatted (CSV, XLS, among others) results. Furthermore, as a result of the available access to the source code, a scientific developer could include additional image processing steps to adapt the software to other crops or to the use of NDVI images instead of RGB images, thus increasing the accuracy of the canopy detection. Notwithstanding, TIPCIP is currently limited to working with FLIR-format images for the extraction of raw data as well as calibration parameters. However, a developer could include additional libraries or functions facilitating it to read other formats and increase the current structure of the software.
In addition, commercial thermal image processing tools for general purposes can be found on the internet. For example, the FLIR Tool Plus (U.S. $\$ 250$) enables the user to visualize, edit, and analyze IRT images acquired with FLIR cameras. However, such analyses are conducted only with thermal images, and consequently, image classification/segmentation is solely performed based on temperature thresholds. A zero-cost programming tool, which is still in development, can be found at <https://github.com/micasense/imageprocessing> for Micasense cameras. The Altum Micasense multispectral camera includes an IRT sensor and allows the users to generate their code based on the Python programming language. The user requires specific programming skills to interpret the examples but then can construct software and include segmentation as well as generate several vegetation indexes.
In sum, TIPCIP provides an option for image processing related to CWSI generation based on FLIR-format images. Its source code allows for modification but requires programming skills in the C++ programming language.
5. Conclusions {#sec5-sensors-20-00472}
==============
The accurate estimation of a plant water stress index via thermal imaging amounts to the need for a simple and efficient image processing algorithm. In this study, the Thermal Image Processor (TIPCIP) software was implemented, and through image segmentation and morphological operations, determined the canopy region and its average temperature. A comparison with in-situ measurements using thermocouples showed that our estimated values of $T_{canopy}$ yield were in good agreement ($r_{Pearson}$ = 0.84). On the basis of these results, an irrigation schedule was implemented using a CWSI-based threshold, specifically CWSI \< 0.4 and CWSI \< 0.7 for T2 and T3, respectively, in two experiments carried out during the hot and wet season. A water saving without yield penalization was found for T2 (341.6 ± 63.7 and 515.7 ± 37.7 m^3^ ha^−1^) in E1 and E2, respectively. Finally, the authors have made the developed TIPCIP software available under the Creative Commons (CC 4.0) policy and encourage the agricultural remote sensing community to assess it and provide feedback (the executable files can be downloaded from \[[@B46-sensors-20-00472]\] and source code files from \[[@B47-sensors-20-00472]\]).
The authors thanks Nikolai Alarcon and Jesus Zamalloa for their technical support.
D.A.R. and R.Q. coordinated the entire project; D.A.R. and H.L. conceived the baseline of the project, G.C.-P. developed the software tool (TIPCIP), J.R. and D.A.R. designed the field experiments in Lima, Peru, J.R. and J.N. conducted the imagery acquisition in the potato fields, J.R. obtained and process the statistical values reported in this manuscript. G.C.-P., D.A.R., J.R. and J.N. wrote the draft of the manuscript and G.C.-P., H.L., J.R., J.N., R.Q. and D.A.R. reviewed and expanded it. All authors have read and agreed to the published version of the manuscript.
This work was supported by Programa Nacional de Innovacion Agraria (PNIA) with the Project grant N° 016-2015-INIA-PNIA/UPMSI/IE "Uso efectivo del agua en el cultivo de papa en zonas aridas: Mejorando el manejo del riego mediante monitoreo de estatus hidrico para enfrentar el Cambio Climatico". The research was conducted under and received complementary funds from the CGIAR Research Programs on Root, Tuber and Bananas (RTB).
The authors declare no conflicts of interest. The funding sponsors had no role in the design of the study; in the collection and interpretation of data; in the writing of this manuscript; or in the decision to publish the results.
![Acquisition procedure of infrared thermography (IRT) and red-green-blue (RGB) images using the FLIR E60 model thermal camera, according to the procedure of Rinza et al. \[[@B8-sensors-20-00472]\] (**A**). The IRT and RGB images acquired in A (**B**). AWRS---the artificial wet reference surface (**C**).](sensors-20-00472-g001){#sensors-20-00472-f001}
{#sensors-20-00472-f002}
{#sensors-20-00472-f003}
{#sensors-20-00472-f004}
{#sensors-20-00472-f005}
{#sensors-20-00472-f006}
sensors-20-00472-t001_Table 1
######
Environmental conditions during the experimental period 2017--2018 (E1). Average daily values ± standard error. VPD---Vapor pressure deficit.
October November December January
------------------------------------ ------------- ------------- ------------- -------------
Minimum Temperature (°C) 14.6 ± 0.08 15.4 ± 0.19 17.5 ± 0.18 19.7 ± 0.10
Maximum Temperature (°C) 21.5 ± 0.27 22.2 ± 0.26 24.1 ± 0.32 27.5 ± 0.21
Average Temperature (°C) 16.8 ± 0.13 17.8 ± 0.18 20.0 ± 0.22 22.7 ± 0.13
Relative Humidity (%) 83.4 ± 0.49 80.6 ± 0.54 81.0 ± 0.73 74.4 ± 0.66
Solar Radiation (MJ m^−2^ day^−1^) 17.2 ± 0.63 15.9 ± 0.73 14.4 ± 0.95 18.5 ± 0.65
Average VPD (kPa) 0.35 ± 0.01 0.42 ± 0.01 0.48 ± 0.03 0.76 ± 0.03
Maximum VPD (kPa) 0.84 ± 0.03 0.92 ± 0.03 0.99 ± 0.05 1.51 ± 0.05
sensors-20-00472-t002_Table 2
######
Environmental conditions during the experimental period 2018 (E2). Average daily values ± standard error. VPD---Vapor pressure deficit.
June July August September
------------------------------------ -------------- -------------- -------------- -------------
Minimum Temperature (°C) 14.8 ± 0.09 14.4 ± 0.06 13.9 ± 0.07 14.3 ± 0.08
Maximum Temperature (°C) 17.8 ± 0.28 18.6 ± 0.37 18.9 ± 0.27 20.4 ± 0.28
Average Temperature (°C) 15.8 ± 0.09 15.7 ± 0.10 15.5 ± 0.10 16.3 ± 0.11
Relative Humidity (%) 85.9 ± 0.56 84.9 ± 0.61 82.5 ± 0.64 80.2 ± 0.59
Solar Radiation (MJ m^−2^ day^−1^) 3.2 ± 0.38 5.2 ± 0.53 7.5 ± 0.66 10.7 ± 0.68
Average VPD (kPa) 0.26 ± 0.01 0.28 ± 0.01 0.32 ± 0.01 0.39 ± 0.01
Maximum VPD (kPa) 0.47 ± 0.03 0.58 ± 0.05 0.68 ± 0.04 0.85 ± 0.04
| {
"pile_set_name": "PubMed Central"
} |
Background {#Sec1}
==========
Target condition being diagnosed {#Sec2}
--------------------------------
Despite successful vaccination programmes, meningococcal disease (MD) remains a leading infectious cause of septicaemia and death in children worldwide \[[@CR1]--[@CR5]\]. Early diagnosis of MD significantly improves outcomes with reduced morbidity and mortality \[[@CR4], [@CR5]\]. The challenge is, however, that during the prodrome invasive MD is indistinguishable from many self-limiting viral infections \[[@CR4]--[@CR6]\]. This invariably leads to a very cautious approach to the management of these children with many receiving parenteral antibiotics pending culture results \[[@CR7]\]. Despite this cautious approach, children are still being diagnosed late due to the difficulties in identifying those children who are infected with MD from those who have a simple viral illness \[[@CR4], [@CR7]\].
Over the years, a number of studies have explored the value of widely available biomarkers including the use of CRP, Procalcitonin and white cell counts in the initial diagnosis of possible MD \[[@CR8]--[@CR12]\]. Whilst these tests have value, none of them have the necessary diagnostic accuracy to allow them to be used as rule out tests at presentation \[[@CR8]--[@CR12]\].
Loop-mediated-isothermal AMPlification (LAMP) for MD is a rapid form of PCR that targets the ctrA gene sequence. The ctrA gene sequence is genetically conserved across all pathogenic (capsular) stains of the bacterium *Neisseria meningitidis* that is responsible for MD \[[@CR13]\]. This technique is faster than traditional PCR techniques and requires much simpler equipment \[[@CR13]--[@CR17]\]. It is possible that LAMP technology could be used as a rapid point of care test (POCT) for the early diagnosis of MD in children. This could be achieved through the rapid testing of blood samples or throat swab specimens in emergency departments, primary care facilities or pharmacies.
Clinical pathway {#Sec3}
----------------
It is very difficult to diagnose early meningococcal disease with current guidance recommending decision-making based on the clinical presentation and laboratory results \[[@CR7], [@CR12]\]. Unfortunately, no single biomarker, combination of biomarkers or clinical guideline has been found to be ideal \[[@CR8]--[@CR12], [@CR18]\]. This has resulted in a very cautious approach resulting in the overtreatment of many children \[[@CR8], [@CR11], [@CR12], [@CR14]\]. Despite such a cautious approach, children are still being diagnosed late \[[@CR12]\].
LAMP could potentially be used at two points within existing care pathways. Firstly at presentation to identify early invasive meningococcal disease in children who present with a minor illness \[[@CR14]\]. Alternatively, LAMP could be used in place of the current gold standard (quantitative PCR or sterile site culture) to quickly confirm or exclude the diagnosis allowing for a more tailored treatment including early ambulation.
The development of LAMP technology for the diagnosis of early MD could therefore represent a significant breakthrough that could alter the care of thousands of children every year worldwide.
Why perform this review? {#Sec4}
------------------------
This systematic review is required because there are a growing number of individual studies that have reported on the diagnostic accuracy of LAMP technology in diagnosing MD \[[@CR13], [@CR14], [@CR19]\]. These studies have used a similar approach; LAMP directed at the conserved CtrA region of the *bacteria Neisseria meningitidis*, but in different populations using different specimens, i.e. blood, CSF and throat swabs \[[@CR13], [@CR14], [@CR19]\]. There are, to our knowledge, no existing systematic reviews. This review may help researchers and policymakers identify the most suitable sample (blood, CSF, throat swab), and it may help to determine the role of LAMP within the existing diagnostic pathway.
Objectives {#Sec5}
----------
The objective of this systematic review is to determine the diagnostic accuracy of LAMP technology in the diagnosis of invasive meningococcal disease in children (\< 18 years of age).
Methods/design {#Sec6}
==============
We will perform a literature search for relevant studies and then screen and select studies for inclusion against eligibility criteria. Data extraction will be performed in duplicate on the selected studies with meta-analysis and report writing.
We will adhere to standards of the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) in reporting the findings of this review \[[@CR20]\]. The content of this protocol follows the Preferred Reporting Items for Systematic Review and Meta-Analysis Protocols (PRISMA-P) recommendations \[[@CR21]\]. (Please see the Additional file [1](#MOESM1){ref-type="media"} for the completed PRISMA-P checklist.) This review is registered with the International Prospective Register of Systematic Reviews (PROSPERO) \[[@CR22]\]. The registration number is \[CRD42017078026\].
Inclusion criteria {#Sec7}
------------------
Table [1](#Tab1){ref-type="table"} below outlines the inclusion criteria for this review. These criteria are discussed in more detail below.Table 1Eligibility criteriaStudy characteristicsInclusion criteriaPopulationChildren \< 18 years of age with suspected meningococcal diseaseIndex testsLoop-mediated-isothermal AMPlification for *Neisseria meningitidis*Reference testQuantitative PCR and/or culture of sterile site (blood and/or CSF) specimensOutcomesTrue and false positives, true and false negativesStudy designsAll prospective, retrospective and randomised control studies that report measures of diagnostic accuracy of Loop-mediated-isothermal AMPlification for *Neisseria meningitidis*
### Types of studies {#Sec8}
All prospective, retrospective and RCT studies that assess the performance of LAMP in assessing children (\< 18 years of age) with potential invasive meningococcal disease will be included. There are no language restrictions.
### Participants {#Sec9}
The participants were children (\< 18 years of age) with signs or symptoms of invasive meningococcal disease.
### Index tests {#Sec10}
The index test being investigated is the LAMP test for meningococcal DNA. For the purpose of this protocol, this is further defined as LAMP testing specific to the ctrA gene of *Neisseria meningitidis*. Index testing can be performed using blood, cerebrospinal fluid and throat swabs. Commercially and non-commercially available tests will be considered.
### Target conditions {#Sec11}
Meningococcal infection (invasive meningococcal disease) is the target condition.
### Reference standards {#Sec12}
The reference standard used to confirm the presence of the target condition in this study is quantitative PCR to detect *Neissseria meningitidis* DNA in a sterile site sample (normally blood or CSF). A positive blood or cerebrospinal bacterial culture of *Neissseria meningitidis* will also be used. No other reference standard will be accepted.
Exclusion criteria {#Sec13}
------------------
Studies that only assess carriage rates in healthy children will be excluded.
Search methods for identification of studies {#Sec14}
--------------------------------------------
### Electronic search strategy {#Sec15}
An electronic search strategy has been developed in collaboration with the Queen's University Belfast Medical Librarian (RF). To identify all prospective, retrospective and RCTs, we will search MEDLINE, Embase, Web of Science, Scopus and The Cochrane Library inclusive of Cochrane Controlled Trials Register. An example Medline search strategy is attached as Additional file [2](#MOESM2){ref-type="media"}. There are no language restrictions for this review.
### Searching other resources {#Sec16}
In addition, we will hand-search reference lists of relevant articles. A targeted grey literature search will include contacting the manufacturers of commercially available LAMP test for meningococcal disease and a search of conference abstracts.
Data collection {#Sec17}
---------------
### Selection of studies {#Sec18}
Two reviewers (TW, MDS) will independently screen the study eligibility and extract data. Screening will be a two-step process with initial title/abstract screening followed by full-text screening. Disagreements among reviewers will be resolved through consensus or third-party reviewer (DF). Reports that are duplicates or co-publications of studies will be identified. Following full-text screening, a list of excluded studies with reasons for exclusion will be provided in an appendix of the final report. We will begin with screening published and unpublished records and select those that meet the inclusion/exclusion criteria. Our search of literature will involve both primary studies and systematic reviews. The latter will be used only to identify additional primary studies.
### Data extraction and management {#Sec19}
TW and MDS will develop a data extraction form, and this will be piloted initially to achieve a good level of agreement between the data extractors. The following data will be extracted in duplicate by TW and MDS:Study characteristics: author, year of publication, country, design, sample size, clinical setting, number studied, number of drop-outs with reason, and funding source.Population characteristics: inclusion/exclusion criteria and patient demographics such as age and gender.LAMP testing: timing of sampling, method of sampling (e.g. throat swab, blood or CSF), time to result and commercial availability of the test.Gold standard: Quantitative PCR (e.g.TaqMan® PCR) or sterile site bacterial culture (i.e. blood/CSF)Outcomes: From this 2 × 2 table, we will calculate true positives, false positives, true negatives, and false negatives.
### Assessment of methodological quality {#Sec20}
The risk of bias of each article will be evaluated independently by two investigators (TW, MDS) and reported according to the Quality Assessment of Diagnostic Accuracy Studies (QUADAS-2) tool \[[@CR23]\]. The most likely bias will be spectrum bias with participants not representing the population of interest. Early MD is very difficult to diagnose with typically vague and non-specific symptoms. Depending on the population, rates of MD can vary from 1 to 33% \[[@CR8]--[@CR11], [@CR14]\]. We are primarily interested in LAMP testing as an early test to identify those children who will go on to deteriorate rather than considering its use in very unwell child in whom treatment will be given irrespective of the test result. Studies that focus on very sick children or studies that have an especially high rate of MD may not reflect the target population for the routine use of this test in the future. A commentary on this potential source of bias will be included, and where possible, we will compare the performance of LAMP as a diagnostic test in both low and high prevalence populations---discussed below. Disagreements between the two investigators (TW, MDS) will be resolved by consensus or arbitration by a third party (DF).
An assessment of publication bias will not be performed. There is no evidence of publication bias in systematic reviews of diagnostic accuracy, and methods for detecting publication bias are unreliable when applied to diagnostic accuracy studies \[[@CR24]\].
Statistical analyses and evidence synthesis {#Sec21}
-------------------------------------------
We will present an overview of the available studies summarised in two tables. The first table will summarise the study designs, participants, index tests details, sample types and the reference standards. The second table will summarise the details on study quality relating to QUADAS-2.
LAMP test result data will be compared to the reference test. Data for 2 × 2 tables of index test against reference standard will be extracted from each study. The true positive, true negative, false positive and false negative rate will be recorded. If these data are not provided, they will be calculated from raw data wherever possible. A summary table of evidence will be produced, and individual studies represented using forest plots displaying the sensitivity and specificity values of the LAMP test with 95% confidence intervals.
Inferred statistics: LAMP testing is binary with either a positive or negative result. There is no cutoff and the bacteria are either present or not. In the meta-analysis, we will therefore use a bivariate random-effects model with covariates such as age, sample type, disease incidence, variations in index tests and study type included to report summary statistics of sensitivity and specificity. This approach is recommended by the Cochrane collaboration and is best suited for the meta-analysis of diagnostic tests where there is no cutoff or the cutoff is very similar between studies \[[@CR24]\]. All analyses will be performed in duplicate by TW and MDS using SPSS 23 and STATA.
In addition to the covariates included in the bivariate random-effects model already discussed, we will also perform subgroup analyses on the following groups:Infants (less than 1 year of age)Pre-school children (less than 4 years)School-age children (4--11 years)Adolescents (11--18 years)Sample type (throat swab, blood, CSF)High disease prevalenceLow disease prevalence
Investigations of heterogeneity {#Sec22}
-------------------------------
We will investigate the heterogeneity by incorporating covariates into the random-effects models as discussed above.
Discussion {#Sec23}
==========
Meningococcal disease is a notoriously difficult disease to diagnose in the early stages, and as such many children are treated with precautionary intravenous broad-spectrum antibiotics. Despite this approach, children are still being diagnosed late with MD. Advances in nucleic amplification technology mean that it is now possible to perform LAMP testing with results being available in under 30 min. What we do not know, however, is how these tests can be used by clinicians in real-world situations. What is the diagnostic accuracy of LAMP testing for MD and how is affected by sample type and age of the child?
By answering these questions, we will have a better understanding of where LAMP testing could fit into the current clinical pathway for the diagnosis of MD. In particular, could it be used as a rapid point of care test to diagnose early MD, or could it have role in rapidly confirming diagnosis after the administration of antibiotics? This information will be of value to policy planners and researchers in determining where in the diagnostic pathway for MD the LAMP test could be trialled.
Additional files
================
{#Sec24}
Additional file 1:PRISMA-P Checklist. (DOCX 33 kb) Additional file 2:Example search strategy. (DOCX 99 kb)
CRP
: C-reactive protein
CSF
: Cerebral spinal fluid
DNA
: Deoxyribonucleic acid
LAMP
: Loop-mediated-isothermal AMPlification
MD
: Meningococcal disease
NPV
: Negative predictive value
PCR
: Polymerase chain reaction
PCT
: Procalcitonin
POCT
: Point of care testing
PPV
: Positive predictive value
RCT
: Randomised control trial
**Electronic supplementary material**
The online version of this article (10.1186/s13643-018-0747-0) contains supplementary material, which is available to authorized users.
Richard Falls (AF) QUB librarian assisted with the design of the literature searches.
Funding {#FPar1}
=======
TW is a doctoral fellow at Queen's University Belfast. TWs fellowship funded by the Health and Social Care Northern Ireland Public Health Agency Research and Development Office. The funder has played no part in the conception or design of this review protocol, and they will have no involvement with the systematic review.
TW, DF, FL, BB and MDS were involved in conception and design of this review protocol. All authors have read and approved the final version of the manuscript.
Ethics approval and consent to participate {#FPar2}
==========================================
Not applicable.
Consent for publication {#FPar3}
=======================
All authors consent for publication.
Competing interests {#FPar4}
===================
Derek Fairley is a shareholder and non-exec Director of Hibergene.
Publisher's Note {#FPar5}
================
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
| {
"pile_set_name": "PubMed Central"
} |
Introduction
============
Haptoglobin (Hp) is a plasma protein that binds extracorpuscular hemoglobin (Hb) and prevents it from inflicting iron‐mediated oxidative tissue damage.^[@b1]^ The Hp gene has 2 major alleles, Hp1 and Hp2. Of these, the latter is a mutated version of the former and unique to humans.^[@b2]^ This common polymorphism (rs72294371) leads to 3 structurally and functionally distinct proteins, Hp 1‐1, 2‐1, and 2‐2, with reported genotype frequencies in subjects of Western European descent of 16%, 48%, and 36%.^[@b2]--[@b3]^ While Hp 1‐1 forms dimers, Hp 2‐1 and Hp 2‐2 form larger linear and cyclic polymers, respectively.^[@b2]^ Hp 2‐2 was reported to be less efficient than Hp 1‐1 in binding Hb and preventing oxidation by stabilizing heme iron within Hb, thereby displaying lower antioxidant activity.^[@b4]--[@b5]^
Based on these biological mechanisms, it has been suggested that the Hp 2‐2 genotype may confer an elevated risk of developing cardiovascular diseases (CVD). Epidemiological studies conducted in general populations, however, have shown contradictory results. On one hand, Hp 2‐2 was linked to the severity of myocardial infarction^[@b6]^ and to more frequent (but less severe) peripheral arterial occlusive disease.^[@b7]^ On the other hand, a large longitudinal population‐based survey has reported a ≈2‐fold risk for coronary artery disease (CAD) death in people with the Hp 1‐1 genotype.^[@b8]^ Two smaller case‐control studies found that the Hp 1 allele was more frequent in patients with CAD^[@b9]^ and lacunar stroke,^[@b10]^ compared with healthy controls.
In the past years the focus has shifted towards the role of Hp genotype in CVD of diabetic patients. Because diabetes is featured by a high level of oxidative stress, relevance of the antioxidant Hp may be considerably higher in diabetic patients.^[@b11]^ Actually, in a meta‐analysis of 5 studies involving 1829 patients with diabetes, the pooled odds ratio (95% CI) for CVD was 2.03 (1.46, 2.81) in a comparison of the Hp 2‐2 genotype with other genotypes.^[@b12]^ In line with this finding, Hp 2‐2 was associated with incident CVD in type 2 diabetes mellitus patients (odds ratio \[95% CI\], 4.96 \[1.85, 13.33\]). In conflict with these findings, the Framingham Offspring Study reported that Hp 2‐2 was associated with significantly lower CAD prevalence among diabetic patients than Hp 1‐1 (odds ratio \[95% CI\], 0.43 \[0.21, 0.90\]).^[@b13]^
As a potential explanation for the substantial heterogeneity in the Hp ‐ CVD association between the various studies effect modification by levels of HbA~1c~ has been proposed.^[@b14]^ Indeed, the reported dysfunction of Hp 2‐2 in preventing Hb‐mediated oxidative damage was found to be accentuated with glycosylated Hb (HbA~1c~) in cell culture experiments.^[@b4]^ Testing this hypothesis in humans, Cahill and colleagues^[@b14]^ reported the Hp 2‐2 genotype to be associated with elevated CAD risk among subjects with elevated HbA~1c~ in the Nurses\' Health Study (odds ratio \[95% CI\], 10.12 \[1.08, 94.97\]) and validated their finding in a cohort of type 2 diabetic subjects (hazard ratio \[95% CI\], 7.55 \[2.79, 20.47\]). Based on this and earlier studies, Hp genotyping among diabetic patients and antioxidant treatment in diabetic patients with the Hp 2‐2 genotype has been propagated ^[@b2],[@b12]^ but further confirmatory data are required to change clinical routine. Given its potentially large impact on diabetes management, we investigated the association of Hp genotype with CVD conditional on elevated HbA~1c~ in the prospective, population‐based Bruneck Study.
Methods
=======
Study Population and Data Collection
------------------------------------
The Bruneck Study is a prospective, population‐based survey on the epidemiology and pathogenesis of atherosclerosis and cardiovascular disease.^[@b15]--[@b18]^ At baseline in 1990 the study population comprised an age‐ and sex‐stratified random sample of all inhabitants of Bruneck (125 men and 125 women from each of the fifth through eighth decades of age, for an age‐ and sex‐stratified random sample of n=1000, all of western European descent). No subjects were enrolled after study initiation. In 1995, 826 subjects participated in the first quinquennial re‐examination and DNA samples for genotyping were available in 816 individuals. During follow‐up from 1995 to 2010, detailed information about fatal and nonfatal new‐onset CVD was carefully collected. Follow‐up was 100% complete for clinical endpoints, which was made possible by the extremely low population mobility of 0.2% in the Bruneck area. The study protocol was approved by the ethics committees of Bolzano and Verona and conformed to the Declaration of Helsinki. All study subjects provided written informed consent. Risk factors were assessed by means of validated standard procedures as described previously.^[@b15],[@b19]^
At the baseline and follow‐up examinations in 1995, 2000, 2005, and 2010, venous blood was sampled in the morning after an overnight fast for laboratory measurements, including fasting plasma glucose and HbA~1c~ (Diabetes Control and Complications Trial‐aligned assay; equipment and reagents from BioRad, Milan, Italy, at both baseline and follow‐up examinations). In accordance with the study of Cahill et al, HbA~1c~ was dichotomized according to a cut‐off of 6.5% (48 mmol/mol, International Federation of Clinical Chemistry units).
Diabetes mellitus was coded present for subjects with fasting glucose levels ≥7 mmol/L (≥126 mg/dL) or a medical record confirmed prediagnosis of definite disease status. Based on the literature^[@b14]^ we anticipated few subjects with the Hp 1‐1 genotype and, to maximize statistical power, decided prior to analysis to pool the Hp 1‐1 and 2‐1 genotypes, forming a group of Hp1 allele carriers, which is a common approach.^[@b14]^
We ascertained leisure time‐related physical activity by a standardized questionnaire^[@b20]^ rating the intensity of activities according to the compendium of physical activities^[@b21]^ and calculated average metabolic equivalent hours per week to estimate long‐term physical activity. We assessed food intake by a standardized food‐frequency questionnaire (FFQ) based on the gold standard Harvard FFQ by Willett and colleagues and adapted the FFQ to the dietary peculiarities in the survey area. We validated our FFQ with a dietician‐supervised short‐term assessment of food intake. Validity was high and similar to that previously found for the same FFQ in other populations.^[@b22]^ The Alternative Healthy Eating Index (AHEI) was calculated from these data as a quantitative measure of healthy dietary behavior.^[@b23]^ Body mass index (BMI) was calculated as weight in kilograms over height in meters squared.
Haptoglobin Genotyping
----------------------
The Haptoglobin genotype was determined by PCR as described previously^[@b24]^ with slight modifications. In brief, 20 μL reactions contained 20 ng genomic DNA, 2 units Qiagen Taq Polymerase (Qiagen), 1× Qiagen PCR buffer (Qiagen, Hilden, Germany), 1× Q solution (Qiagen), 200 μmol/L of each dNTP (Peqlab) and 0.25 μmol/L of each primer (A/B or C/D). Oligonucleotide primers A and B were used for amplification of a 1757‐bp Hp 1 allele‐specific sequence and a 3481‐bp Hp 2 allele‐specific sequence. Primers C and D were used to amplify a 349‐bp Hp 2 allele‐specific sequence. The primers A (5′‐GAGGGGAGCTTGCCTTTCCATTG‐3′), B (5′‐GAGATTTTTGAGCCCTGGCTGGT‐3′), C (5′‐CCTGCCTCGTATTAACTGCACCAT‐3′) and D (5′‐CCGAGTGCTCCACATAGCCATGT‐3′) were purchased from Microsynth. The amplification reactions were conducted on a DNA Engine Cycler (BioRad) under the following conditions: initial denaturation 3 minutes 94°C; 94°C 30 seconds, 57°C (primers A/B) and 62°C (primers C/D) 30 seconds, 72°C 2 minutes, 35 cycles; final extension 10 minutes 72°C. After amplification 8 μL PCR product A/B and 2 μL PCR product C/D were mixed and separated together on a 1% agarose gel.
Endpoints
---------
The composite CVD endpoint included incident fatal and non‐fatal myocardial infarction and stroke. Presence of myocardial infarction was assessed by World Health Organization criteria (definite disease status),^[@b25]^ while stroke was classified according to the criteria of the National Survey of Stroke.^[@b26]^ Events were ascertained by careful review of medical records provided by general practitioners, death certificates, and Bruneck Hospital files. A major advantage of the Bruneck Study is that virtually all inhabitants of Bruneck are referred to 1 local hospital that cooperates closely with the general practitioners. This allows retrieval of complete medical information.
Subjects who had experienced CVD events before the study baseline in 1995 were included in the main analysis and observed for recurring incident CVD between 1995 and 2010, and excluded in a sensitivity analysis.
Statistical Analysis
--------------------
Hardy‐Weinberg equilibrium was tested by a permutation‐based chi‐squared test. Associations with incident CVD were assessed by Cox proportional hazards regression with HbA~1c~ or diabetes status (present/absent) as time‐varying covariates in a granularity of 5 years. This approach relates the most current measure of glycemic exposure to incident CVD and avoids potential confounding due to reliance on a single baseline measurement. Effect modification of Hp 2‐2 by HbA~1c~ category was tested using an appropriate interaction term. To simultaneously obtain marginal effects of Hp genotype for both HbA~1c~ groups as well as the interaction effect, a linear combination of model parameters was made. The proportional hazards assumption was tested by computing the correlation coefficient of survival time with scaled Schoenfeld residuals and was met. Base models were adjusted for age and sex (model 1) and 2 multivariable models with progressive adjustment were employed (model 2: additionally for current smoking, systolic blood pressure, low‐density lipoprotein (LDL) cholesterol, and high‐density lipoprotein (HDL) cholesterol; model 3: additionally for metabolic equivalent hours, alternative healthy eating index, statin use, and body mass index). Last observation carried forward or next observation carried backward imputation reduced the proportion of missing HbA~1c~ values from 5.9% to 0.4%.
Sensitivity analyses excluded subjects with missing HbA~1c~ values, used age as the time scale and focused on individual disease endpoints. All tests were 2‐sided and *P* values smaller than 0.05 were considered significant. All analyses were performed using the R statistical package.
Results
=======
Characteristics of the study population are shown in [Table 1](#tbl01){ref-type="table"}. Hp genotyping resulted in unambiguous results for 810 of 816 subjects for which DNA samples were available (Call rate, 99.3%). Of these, 4 had missing values in HbA~1c~ concentration, which resulted in a baseline study size of 806 subjects. Duplicate measurement of 24 DNA samples yielded concordant findings in all cases. Genotypes were distributed as follows: Hp 1‐1, 10.3% (n=83), Hp 2‐1, 41.7% (n=336), Hp 2‐2, 48.0% (n=387) and were in Hardy‐Weinberg equilibrium overall (P=0.46), in subgroups of subjects with diabetes (*P*=0.13) and without (*P*=0.79), and subjects younger than 65 (*P*=0.52) and subjects at least 65 years old (*P*=0.80). There were no significant differences in prevalent diabetes (1995 baseline) or in incident diabetes (follow‐up 1995‐2010) between dichotomized Hp genotypes (*P*=0.71 and 0.92, respectively). For repeated measurements of HbA~1c~ concentration we found an intra‐class correlation coefficient \[95% CI\] of 0.60 \[0.56, 0.63\].
######
Characteristics of the Study Population According to Haptoglobin Genotype
Haptoglobin genotype *P* Value
------------------------------ ---------------------- ---------------------- ---------------------- -------
*Measurements taken in 1995*
n (%) 83 (10.3) 336 (41.7) 387 (48.0)
Age, y 63.6±12.6 62.1±10.3 63.0±11.4 0.417
Male sex, n (%) 40 (48.2) 175 (52.1) 180 (46.5) 0.342
HbA~1c~, % 5.3 (5.1 to 5.7) 5.4 (5.1 to 5.8) 5.4 (5.1 to 5.8) 0.496
HbA~1c~≥6.5%, n (%) 3 (3.6) 15 (4.5) 23 (5.9) 0.575
Diabetes, n (%) 12 (14.5) 31 (9.2) 44 (11.4) 0.481
Hemoglobin, g/dL 14.1±1.3 14.3±1.2 14.3±1.3 0.122
Glucose, mg/dL 98.0 (91.0 to 109.5) 97.0 (90.0 to 104.0) 98.0 (91.0 to 107.0) 0.140
Current smoking, n (%) 13 (15.7) 68 (20.7) 74 (19.5) 0.714
Total cholesterol, mg/dL 227.7±41.3 225.6±38.9 234.4±45.5 0.036
HDL cholesterol, mg/dL 57.4±13.9 58.5±15.7 59.3±17.0 0.623
Systolic BP, mm Hg 145.3±21.4 148.1±20.3 149.00±20.9 0.225
Current smoking, n (%) 70 (84.3) 268 (79.8) 311 (80.4) 0.707
Body mass index, kg/m^2^ 26.0±4.4 25.9±3.6 25.4±3.9 0.214
AHEI, score 40.6±9.00 39.2±8.8 39.2±8.6 0.317
MET, h/wk 42.0 (30.4) 42.0 (35.6) 42.0 (38.3) 0.782
Statin use, n (%) 1 (1.2) 7 (2.1) 16 (4.2) 0.170
Prior CVD, n (%) 8 (9.6) 21 (6.2) 19 (4.9) 0.304
*Measurements taken in 2000*
n (%) 68 (9.7) 300 (42.8) 333 (47.5)
Age, y 60.7±11.5 60.7±9.7 61.1±10.6 0.895
Male sex, n (%) 27 (39.7) 154 (51.3) 144 (43.2) 0.067
HbA~1c~, % 5.7 (5.5 to 5.9) 5.7 (5.5 to 6.0) 5.7 (5.5 to 6.0) 0.155
HbA~1c~≥6.5%, n (%) 3 (4.4) 17 (5.7) 34 (10.2) 0.064
Diabetes, n (%) 8 (11.8) 33 (11.0) 39 (11.7) 0.952
*Measurements taken in 2005*
n (%) 61 (10.1) 256 (42.6) 284 (47.3)
Age, y 59.6±10.9 59.7±9.5 59.3±9.7 0.826
Male sex, n (%) 25 (41.0) 124 (48.4) 122 (43.0) 0.327
HbA~1c~, % 5.5 (5.4 to 5.7) 5.6 (5.4 to 5.9) 5.6 (5.4 to 5.9) 0.539
HbA~1c~≥6.5%, n (%) 3 (4.9) 20 (7.8) 28 (9.9) 0.328
Diabetes, n (%) 6 (9.8) 34 (13.3) 36 (12.7) 0.787
Values are presented as mean±standard deviation, median (interquartile range), or count (percentage). *P* values are for tests of any difference between the 3 groups under adjustment for age and sex. AHEI indicates alternative healthy eating index; BP, blood pressure; CVD, cardiovascular disease; HbA~1c~, glycosylated hemoglobin; HDL, high‐density lipoprotein; MET, metabolic equivalent.
The analyses reported in the following were adjusted for age and sex, unless specified otherwise, and utilize updates of HbA~1c~ levels during follow‐up (time‐varying covariate). From 1995 to 2010, 123 subjects experienced at least 1 CVD event (48 suffered myocardial infarction(s), 68 stroke(s), and 7 both). Only the first CVD event was considered in the main analysis. No significant differences emerged when examining the risk of CVD by Hp genotype (hazard ratio \[HR\] \[95% CI\] for Hp 2‐2 versus Hp 2‐1/1‐1: 0.98 \[0.69, 1.40\]; *P*=0.912). However, when taking into consideration HbA~1c~ category, Hp 2‐2 subjects with HbA~1c~≥6.5% showed a non‐significant trend towards lower CVD risk than Hp 1 carriers (Hp 2‐1/1‐1 subjects) with HbA~1c~≥6.5% (HR \[95% CI\], 0.47 \[0.19, 1.13\], *P*=0.092) ([Figure](#fig01){ref-type="fig"}). This result did not materially change upon progressive adjustment for HDL and LDL cholesterol, systolic blood pressure, and smoking (model 2 HR \[95% CI\], 0.48 \[0.18, 1.26\], *P*=0.136) and metabolic equivalent hours, alternative healthy eating index, statin use, and body mass index (model 3 HR 0.52 \[0.19, 1.39\], *P*=0.189) ([Figure](#fig01){ref-type="fig"}). No risk differences were detected among subjects with HbA~1c~\<6.5% (HR \[95% CI\], 1.10 \[0.75, 1.62\], *P*=0.629; Hp 2‐2 versus other) resulting in a near significant interaction between dichotomized Hp genotype and HbA~1c~ category (*P* for interaction=0.082) as illustrated in the [Figure](#fig01){ref-type="fig"}.
{#fig01}
Finally, Hp 2‐2 was not associated with CVD risk in subjects with diabetes (*P*=0.944) nor in those without (*P*=0.950), and no statistical interaction existed between diabetes and Hp genotype (*P*=0.927).
Sensitivity analyses excluding subjects with missing HbA~1c~ values, excluding subjects with prior CVD or using age as the time scale (instead of time‐on‐study) yielded similar findings ([Table 2](#tbl02){ref-type="table"}) as did analyses on individual disease endpoints (HR \[95% CI\] for stroke: 0.48 \[0.16, 1.47\], *P*=0.198; HR \[95% CI\] for myocardial infarction: 0.59 \[0.16, 2.20\], *P*=0.433; Hp 2‐2 versus other). Subgroup analyses according to sex and age ([Table 2](#tbl02){ref-type="table"}) should be interpreted cautiously given limited sample sizes in these groups.
######
Subgroup Analyses, Endpoint‐Specific Analyses, and Sensitivity Analyses
Model 1 Model 2 Model 3
---------------------------------- --------------------- --------------------- --------------------- --------- --------------------- ------- ------- --------------------- ------- -------
Full sample \<6.5% 1.10 \[0.75, 1.62\] 0.629 0.082 1.11 \[0.74, 1.66\] 0.623 0.117 1.13 \[0.75, 1.69\] 0.563 0.149
≥6.5% 0.47 \[0.19, 1.13\] 0.092 0.48 \[0.18, 1.26\] 0.136 0.52 \[0.19, 1.39\] 0.189
*Subgroups*
Age\<75 y \<6.5% 0.98 \[0.59, 1.63\] 0.949 0.715 0.95 \[0.57, 1.58\] 0.834 0.767 0.98 \[0.58, 1.64\] 0.928 0.564
≥6.5% 0.79 \[0.26, 2.35\] 0.666 0.77 \[0.23, 2.65\] 0.682 0.66 \[0.19, 2.31\] 0.511
Age≥75 y \<6.5% 1.22 \[0.65, 2.27\] 0.536 0.035 1.35 \[0.68, 2.68\] 0.390 0.081 1.30 \[0.65, 2.60\] 0.451 0.083
≥6.5% 0.11 \[0.01, 0.95\] 0.044 0.18 \[0.02, 1.55\] 0.118 0.17 \[0.02, 1.53\] 0.115
Men \<6.5% 0.98 \[0.59, 1.62\] 0.938 0.986 0.94 \[0.55, 1.60\] 0.810 0.821 0.93 \[0.54, 1.60\] 0.795 0.796
≥6.5% 0.97 \[0.24, 3.92\] 0.963 0.78 \[0.17, 3.64\] 0.748 0.75 \[0.16, 3.60\] 0.719
Women \<6.5% 1.36 \[0.72, 2.55\] 0.339 0.018 1.46 \[0.76, 2.79\] 0.258 0.074 1.59 \[0.82, 3.09\] 0.171 0.110
≥6.5% 0.27 \[0.08, 0.88\] 0.030 0.40 \[0.11, 1.40\] 0.151 0.47 \[0.13, 1.77\] 0.264
*Individual endpoints*
Stroke \<6.5% 0.87 \[0.53, 1.45\] 0.601 0.337 0.94 \[0.56, 1.58\] 0.823 0.645 0.96 \[0.57, 1.62\] 0.883 0.691
≥6.5% 0.48 \[0.16, 1.47\] 0.198 0.67 \[0.18, 2.56\] 0.563 0.72 \[0.19, 2.76\] 0.630
Myocardial infarction \<6.5% 1.51 \[0.84, 2.71\] 0.165 0.201 1.43 \[0.78, 2.63\] 0.246 0.100 1.43 \[0.78, 2.62\] 0.254 0.081
≥6.5% 0.59 \[0.16, 2.20\] 0.433 0.38 \[0.09, 1.64\] 0.197 0.35 \[0.08, 1.52\] 0.160
*Sensitivity analyses*
Age as time scale \<6.5% 1.09 \[0.74, 1.61\] 0.659 0.114 1.10 \[0.74, 1.66\] 0.630 0.146 1.13 \[0.75, 1.70\] 0.553 0.176
≥6.5% 0.50 \[0.21, 1.21\] 0.125 0.51 \[0.19, 1.34\] 0.171 0.54 \[0.20, 1.46\] 0.226
No HbA~1c~ imputation \<6.5% 1.08 \[0.72, 1.61\] 0.708 0.255 1.10 \[0.73, 1.67\] 0.640 0.252 1.10 \[0.73, 1.67\] 0.649 0.346
≥6.5% 0.60 \[0.24, 1.52\] 0.281 0.59 \[0.21, 1.60\] 0.297 0.64 \[0.23, 1.82\] 0.408
Subjects with prior CVD excluded \<6.5% 1.24 \[0.80, 1.90\] 0.332 0.130 1.23 \[0.79, 1.92\] 0.357 0.233 1.27 \[0.82, 1.99\] 0.287 0.233
≥6.5% 0.55 \[0.21, 1.44\] 0.222 0.63 \[0.23, 1.73\] 0.373 0.65 \[0.23, 1.81\] 0.409
HR \[95% CI\] and *P* value are for the comparison of Hp 2‐2 versus other genotypes among subjects with glycosylated hemoglobin (HbA~1c~) \<6.5% or HbA~1c~≥6.5%, respectively. P~interaction~ is for a difference in the effect of dichotomized Hp genotype between those with HbA~1c~\<6.5% and those with HbA~1c~≥6.5%. Model 1: adjustment for age and sex; Model 2: as model 1, with additional adjustment for current smoking, systolic blood pressure, LDL cholesterol, and HDL cholesterol; Model 3: as model 2, with additional adjustment for metabolic equivalent hours, alternative healthy eating index, statin use, and body mass index. CVD indicates cardiovascular disease; HbA~1c~, glycosylated hemoglobin; HR, hazard ratio.
Finally, among those with HbA~1c~≥6.5%, risk estimates were almost identical for Hp 2‐1 compared with Hp 1‐1 subjects (HR \[95% CI\], 0.97 \[0.25, 3.76\], *P*=0.964) providing a post‐hoc justification for the dichotomization of Hp genotypes applied in the current study (Hp 2‐2 versus 2‐1/1‐1).
We also investigated differences in blood lipids by Hp genotype. Total cholesterol and LDL cholesterol were higher in Hp 2‐2 subjects than in those with other genotypes when adjusting for age, sex, and body mass index (*P*=0.008 and *P*=0.016, respectively). These results are shown in [Table 3](#tbl03){ref-type="table"}.
######
Blood Lipid Levels (mg/dL) by Haptoglobin Genotype
Hp 1‐1 Hp 2‐1 Hp 2‐2 *P* ~any\ difference~ *P* ~Hp\ 2‐2\ vs\ other~
------------------- ------------- ------------ ------------ ----------------------- --------------------------
n 83 336 387
*Lipid parameter*
Total cholesterol 227.7±41.3 225.6±38.8 234.4±45.5 0.028 0.008
HDL 57.4±13.9 58.5±15.7 59.3±17.0 0.788 0.970
LDL 144.1±35.6 141.6±35.7 148.5±40.3 0.053 0.016
Triglycerides 143.8±116.1 126.7±66.7 133.3±80.6 0.075 0.189
Values are presented as mean±standard deviation. *P* values are adjusted for age, sex, and body mass index. HDL indicates high‐density lipoprotein cholesterol; Hp, Haptoglobin; LDL, low‐density lipoprotein cholesterol.
Discussion
==========
Our data do not confirm the recent report of Cahill and colleagues that identified the Hp 2‐2 genotype as a cardiovascular risk factor among subjects with elevated HbA~1c.~^[@b14]^ Part of this discrepancy may derive from differences in study design and population. Cahill used a nested case‐control design with 1:1 matching in female nurses with 14‐year follow‐up for their main analysis, and data from a clinical trial with 18 months of follow‐up performed in type 2 diabetic patients for confirmation, whereas we used a 15‐year prospective observational design on a random sample of the general population. Importantly, Cahill excluded participants with prior CVD, which we did not, and used a coronary heart disease (CHD) endpoint, while we used a compound of myocardial infarction and stroke in the main analysis. Another potential explanation for the discrepancy between our and Cahill\'s results is survival bias. However, we observed a higher frequency of Hp 2‐2 genotype (48%) than most previous studies in Caucasians did (36 to 40%),^[@b2]--[@b3],[@b14],[@b27]^ which is the opposite of what would be expected in the presence of a survival disadvantage. Furthermore, genotypes were in Hardy‐Weinberg equilibrium overall as well as in the younger and older half of our sample separately, which argues against the existence of significant survival bias.
Our results are unexpected because Hp 2‐2 as a risk genotype among diabetic subjects would be backed by a biologically plausible rationale based on its weaker antioxidant properties.^[@b2],[@b28]--[@b30]^ However, not all aspects of this rationale are fully consistent. In particular, the Hp 2‐2‐Hb compared with the Hp 1‐1‐Hb complex has been reported to be taken up into macrophages via the CD163 scavenger receptor at higher^[@b31]^ as well as at lower^[@b4]^ rates. No differences in Hb binding and antioxidant potency between purified human Hp 1‐1 and Hp 2‐2 were recently found in an animal model of hemolysis.^[@b32]^ In addition, Hp has numerous functions apart from Hb clearance, taking part in inflammatory pathways (prostaglandin synthesis, cathepsin B activity, endothelium‐dependent vasodilation)^[@b3],[@b33]^ and interfering with the functions of immune cells.^[@b34]--[@b35]^ Hp phenotypes differ in some of these functions.^[@b3]^ Notably, Hp 2‐2 is the most angiogenic phenotype,^[@b36]^ which was discussed as an explanation for longer walking distance in Hp 2‐2 peripheral vascular disease patients (*P*\<0.05),^[@b7]^ and Hp 1‐1 was linked to decreased endothelial repair potential in lacunar stroke patients.^[@b37]^
In this study, subjects with the Hp 2‐2 genotype had elevated total and LDL cholesterol ([Table 3](#tbl03){ref-type="table"}), which has been reported previously.^[@b38]--[@b39]^ This finding may be explained by the close genetic linkage between Hp genotype and a single nucleotide polymorphism in the gene encoding haptoglobin‐related protein, which is associated with levels of total but not HDL cholesterol via apolipoprotein L.^[@b27]^ However, adjusting for LDL levels did not appreciably change risk estimates ([Table 2](#tbl02){ref-type="table"}), which argues against the possibility that CVD risk differences due to Hp genotype are mediated by LDL levels.
Strengths of the present study include the virtually complete, long‐term follow‐up, detailed characterization of the study population and repeated measurements of HbA~1c~. Another major strength of the Bruneck Study is its high representativity for the general population. It comprises predominantly low‐ and medium‐risk individuals, which are of the foremost public health interest since most CVD events happen in such individuals.^[@b40]^ One downside of this is a comparatively low prevalence of elevated HbA~1c~, which precludes more complex analyses like 3‐way interactions between HbA~1c~, diabetes, and Hp genotype. It merits attention that while we found a trend towards protective effects of the Hp 2‐2 genotype among subjects with elevated glycohemoglobin, this result was statistically not significant, and hazardous effects would also be compatible with our data. The largest hazardous effect that we cannot refute at the α=0.05 level is a risk elevation of ≈26% (multivariable model 2, [Figure](#fig01){ref-type="fig"}). Larger studies are required to draw definite conclusions.
In summary, this study does not confirm that the Hp *2‐2* genotype is associated with a higher CVD risk in subjects with elevated HbA~1c~.
Willeit, Kiechl, and Weiss are supported by the FWF (Fonds zur Förderung der wissenschaftlichen Forschung) grant TRP 188. The Bruneck Study is funded by the Pustertaler Verein zur Prävention der Herz‐ und Hirngefäßerkrankungen, Sanitätseinheit Ost, Assessorat für Gesundheit, province of Bozen, Italy.
None.
| {
"pile_set_name": "PubMed Central"
} |
Cortical oscillations in the beta frequency band (15 to 35 Hz) are thought to play a central role in cortical processing of information related to movement and cognition ([@r1], [@r2]). In Parkinson's disease (PD), cortical beta oscillations become abnormally synchronized through their interaction with the dopamine-depleted basal ganglia network ([@r3], [@r4]). These pathological beta oscillations are suppressed when continuous high-frequency deep brain stimulation (DBS) is applied to either subthalamic nucleus (STN) or the internal segment of the globus pallidus ([@r5][@r6]--[@r7]), suggesting a strong dependence on a distributed subcortical network architecture for their generation and maintenance. A central goal in the field is to define the neuronal interactions through which abnormally strong and sustained beta oscillations emerge following dopamine depletion. Addressing these pathologically exaggerated activities may also give key insights as to how oscillations are transmitted in the healthy brain ([@r1]).
The spiking activity of neurons across the basal ganglia locks to specific phases of cortical oscillations in the parkinsonian brain ([@r8][@r9][@r10][@r11]--[@r12]). The high stability of these conditions of synchronization is likely determined by synaptic connectivity between different neuronal populations in the network, be it physiological, pathological, or compensatory. Elucidating the mechanisms of network synchronization is made harder by the large variance in baseline firing rates and presence/absence of autonomous firing in different basal ganglia cell types. Several hypotheses have been raised as to which network connections are necessary and sufficient for cortical oscillations to be abnormally propagated and/or amplified. These theories include 1) the excitatory--inhibitory coupling between the STN and the external segment of the globus pallidus (GPe) ([@r13][@r14]--[@r15]); 2) alteration of striatal output ([@r12]), possibly due to interneuron dysfunction ([@r16][@r17]--[@r18]); and 3) enhanced pallidostriatal positive feedback ([@r19]). These hypotheses have been based around the idea that oscillatory dynamics are relatively stable. However, recent studies have revealed that beta band activity in healthy ([@r20]) and parkinsonian basal ganglia local field potentials (LFPs) and frontal electroencephalograms (EEG) ([@r21], [@r22]) occurs in transient beta bursts (β bursts). It is currently unclear whether cortical and basal ganglia phase locking occurs predominantly during such bursts. If this is the case, the question of how beta oscillations are propagated and amplified can potentially be approached by identifying the neuronal interactions that occur before, during, and after these transient events. Such questions have assumed clinical significance due to the development of closed-loop stimulation approaches to DBS ([@r23]), as they have the potential to reveal the most effective way to disrupt the emergence of pathological oscillations.
Here we use multiple neuronal signals and analytical approaches in PD patients and dopamine-depleted rodents to demonstrate that cortical β bursts are associated with highly stable cortical and basal ganglia phase locking. These conditions of synchronization are established significantly earlier than the threshold commonly used to define the onset of β bursts using oscillation amplitude. Moreover, this initial period of synchronization is associated with the cell type-specific phase trajectories of basal ganglia spikes on the cortical beta oscillation. These findings have important implications for identifying the process through which neuronal oscillations propagate through different brain areas and for the design of closed-loop DBS.
Results {#s1}
=======
The definition of a β burst varies extensively across different studies, in terms of both the threshold used to categorize an epoch of activity as a β burst, as well as the temporal dynamics of the identified burst ([@r20][@r21]--[@r22], [@r24]). In this study, the occurrence of a cortical β burst was inclusively defined as any period lasting \>50 ms when the instantaneous beta amplitude of the Fz--Cz EEG (in the case of PD patients) or frontal electrocorticogram (ECoG, in parkinsonian rats) exceeded the 75th percentile of the amplitude calculated across the entire recording ([@r21], [@r22]). While frontal EEG signals may include contributions from subcortical sources, in this context they have been widely used as a read-out of cortical activity and provide similar information to ECoG ([@r11], [@r25]).
PD Patients. {#s2}
------------
### Corticosubthalamic phase locking follows the time course of EEG β-burst amplitude in PD patients. {#s3}
β bursts have been extensively reported in the STN LFPs of PD patients, but it is unclear to what extent they reflect STN unit activity. To address this issue, we utilized the background unit activity (BUA) signal recorded in and around the STN of PD patients undergoing intraoperative functional mapping for subsequent DBS therapy. The BUA signal is a continuous time series (like EEG/ECoG) and provides insight into the synchronous spike discharges of local neuronal ensembles ([@r12]). BUA recorded from the STN of PD patients exhibited enhanced rhythmic activity at beta frequencies ([Fig. 1 *A* and *B*](#fig01){ref-type="fig"}). Moreover, subthalamic BUA was strongly coherent with the EEG in the beta frequency band ([Fig. 1*C*](#fig01){ref-type="fig"}). Beta oscillations and significant coherence with EEG were confined to BUA signals within the physiologically defined boundary of the STN ([Fig. 1 *B* and *C*](#fig01){ref-type="fig"}). Having established that synchronous beta oscillations were present, we used these signals to address the question of whether the phase locking of subthalamic neuronal activity to the EEG followed the time course of EEG β bursts. To this end, after filtering EEG and BUA in the beta frequency band (EEG~β~ and STN-BUA~β~) we computed phase locking, as measured by the phase synchrony index (PSI) of the Hilbert transform-derived phase of these signals, around the onset of the EEG β bursts (hemispheres = 13, patients = 7, *n* = 18 recordings). Significant modulation of the time course of all PSI analyses was assessed using a cluster-based test to compare the burst-aligned data with data that were randomly selected with no relation to burst onset (see [*Methods*](#s13){ref-type="sec"} for details). PSI in time, which measures the strength of phase locking over short epochs (50 ms) across each individual burst ([*SI Appendix*, Fig. S1*A*](https://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1819975116/-/DCSupplemental)), was elevated around the β-burst onset, and the duration of locking increased with longer burst durations ([Fig. 1*E*](#fig01){ref-type="fig"}). Importantly, the onset of phase locking (in time) began on average −115, −114, and −35 ms relative to the onset of short (50 to 150 ms), medium (150 to 250 ms), and long (250 to 350 ms) bursts, respectively, suggesting that increased corticosubthalamic synchronization actually precedes the amplitude-based burst threshold. PSI across bursts, which measures the consistency of the phase relationship between bursts at each individual time point ([*SI Appendix*, Fig. S1*B*](https://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1819975116/-/DCSupplemental)), began on average −80, −74, and −79 ms for short, medium, and long bursts relative to the EEG β-burst threshold ([Fig. 1*F*](#fig01){ref-type="fig"}). The time course of PSI across bursts followed that of the amplitude less closely than that of PSI in time ([Fig. 1*F*](#fig01){ref-type="fig"}). These findings demonstrate that in PD patients, corticosubthalamic synchronization precedes the onset and follows the time course of EEG β bursts, and the conditions of this transient synchronization are consistent across bursts.
, for schematic explanation of analysis methods). Significance for each burst length was determined using a cluster-based analysis ([*Methods*](#s13){ref-type="sec"}) which tested the difference in the modulation of PSI over time in the burst-aligned data to that from randomly selected data that had no relationship to burst onset. Significant increases with respect to baseline are indicated with horizontal bars, color matched to the burst duration. *B* and *C* show mean and SEM across hemispheres. In *E* and *F*, shaded regions indicate the SEM across all recordings.](pnas.1819975116fig01){#fig01}
6-OHDA--Lesioned Rats. {#s4}
----------------------
Electrophysiological recordings in the basal ganglia of PD patients are mostly limited to the STN. To define how other areas of the basal ganglia network are modulated by cortical β bursts, we used multichannel silicon electrodes (or probes) to record unit activities in the GPe, striatum (Str), and/or STN of parkinsonian 6-OHDA--lesioned rats; frontal ECoGs were simultaneously recorded with these basal ganglia activities ([@r9], [@r12]). As for the EEG in the patient recordings, the Hilbert transform-derived phase and amplitude were extracted from the beta-filtered ECoG (ECoG~β~ phase and amplitude; see [*Methods*](#s13){ref-type="sec"} for details of filtering). [*SI Appendix*, Fig. S2](https://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1819975116/-/DCSupplemental), shows that ECoG β bursts of different lengths had broadly equivalent characteristics across these datasets ([*SI Appendix*](https://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1819975116/-/DCSupplemental)).
### ECoG β bursts are associated with increased cortical and basal ganglia phase locking and local oscillations in the basal ganglia. {#s5}
As shown in PD patients**,** STN BUA~β~ in 6-OHDA--lesioned rats exhibited increases in phase locking to ECoG~β~ slightly before and during the ECoG β burst ([Fig. 2](#fig02){ref-type="fig"}). This transient synchronization extended to the GPe and Str BUA~β~ phase~,~ suggesting that it is a network-wide phenomenon ([Fig. 2 *A* and *C*](#fig02){ref-type="fig"}). Across all structures and burst lengths, increases in both PSI over time and PSI across bursts preceded the ECoG β-burst threshold ([Fig. 2](#fig02){ref-type="fig"} and [*SI Appendix*, Tables S1 and S2](https://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1819975116/-/DCSupplemental)). The same analyses, aligned to burst offset, demonstrated that the time at which ECoG~β~ amplitude dropped below the β-burst threshold correlated with the loss of phase locking with the basal ganglia BUA~β~ signals ([*SI Appendix*, Fig. S3](https://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1819975116/-/DCSupplemental)). These results demonstrate that transient increases in cortical and basal ganglia phase locking occur in individual bursts, with stable parameters of synchronization between bursts across the BG. To test this further, we triggered raw, unfiltered BUA signals at the burst threshold, after adjusting the trigger to the nearest peak of the beta phase, whereby changes in phase alignment or frequency of the signals would weaken the timing of activity in relation to time and/or phase. In each structure, the alignment of the raw BUA with the ECoG β phase around the time of the β burst was consistent enough, across bursts and recordings, for the beta oscillation to emerge simply from summing the unfiltered activity around a single time point ([*SI Appendix*, Fig. S4](https://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1819975116/-/DCSupplemental)). As in the analysis of PSI, this phase-driven modulation appeared to start 1 or more cycles before the burst threshold in all structures ([*SI Appendix*, Fig. S4](https://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1819975116/-/DCSupplemental)).
{ref-type="sec"}) which tested the difference in the modulation of PSI over time in the burst-aligned data to that from randomly selected data that had no relationship to burst onset. Significant increases with respect to baseline are indicated with horizontal bars, color matched to the burst duration. (Shaded regions indicate the SEM across all recordings made from a given structure.)](pnas.1819975116fig02){#fig02}
To establish whether this increase in cortical and basal ganglia phase locking was associated with a concurrent increase in local beta oscillations in the basal ganglia structures, the evolution of BUA~β~ amplitude (i.e., the magnitude of the Hilbert Transform) in STN, GPe, and Str was computed in relation to ECoG β-burst onset for the 3 burst durations ([*SI Appendix*, Fig. S5](https://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1819975116/-/DCSupplemental)). In GPe, longer ECoG β bursts were clearly associated with longer periods of increased BUA~β~ amplitude following ECoG β-burst onset ([*SI Appendix*, Fig. S5](https://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1819975116/-/DCSupplemental)). In Str and STN, BUA~β~ amplitude generally increased following the β-burst threshold, but the relationship with burst duration was less clear ([*SI Appendix*, Fig. S5](https://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1819975116/-/DCSupplemental)). Overall, the time course of ECoG~β~ amplitude during β bursts was associated with parallel modulation of cortical and basal ganglia phase locking and basal ganglia oscillation across the network.
#### Conditions for cortical and basal ganglia phase locking are established before the ECoG β burst. {#s6}
The analyses above demonstrate that phase locking begins significantly before the ECoG burst threshold and that the conditions of synchronization (i.e., the phase difference between the signals) are consistent between bursts. Next we tested whether it was possible to detect the point at which these stable phase conditions are established. To address this question, we calculated the probability of phase slips, defined here as abrupt changes in the phase difference of the ECoG~β~ and basal ganglia BUA~β~ phase, in relation to the ECoG β-burst onset ([Fig. 3](#fig03){ref-type="fig"}). The average probability of a phase slip in the 200 ms before β-burst onset was greater than the average probability of a phase slip occurring in the following 200 ms (paired *t* test *P* \< 0.001 for all regions). To determine whether these phase slips were predictive of the onset of stable synchronization, we compared the conditions of phase alignment immediately before and after the slip to those that occurred within the β burst itself. We observed that the angle of the phase difference between the ECoG~β~ and BG BUA~β~ immediately before a phase slip was inconsistent across β-burst epochs, as revealed by low PSI values, in all structures ([Fig. 3*B*](#fig03){ref-type="fig"}). Immediately after a phase slip, the PSI values more than doubled ([Fig. 3*B*](#fig03){ref-type="fig"}). Crucially, we found that the precise angle of the phase difference occurring immediately after the phase slip was the same as that within the burst itself ([Fig. 3*C*](#fig03){ref-type="fig"}, as indicated by a difference of 0°). In contrast, there was a weak relationship between the preburst phase difference and that within the burst, with a slight tendency for the opposite alignment ([Fig. 3*C*](#fig03){ref-type="fig"}, as indicated by a difference of 180°). These findings were repeated across all BG structures and were consistent with an increase in the variance in BUA~β~ frequency before the burst threshold, which was significant in Str and GPe ([*SI Appendix*, Fig. S6](https://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1819975116/-/DCSupplemental)). The same analysis of the differentiated unwrapped Hilbert phase of each of the ECoG~β~ and BUA~β~ individually showed that with the exception of the striatal BUA, phase slips were also more likely to occur in the 200 ms before the burst threshold in all structures ([*SI Appendix*, Fig. S7](https://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1819975116/-/DCSupplemental)). While the alignment of phase slips was more pronounced in cortex, any comparison must take into account the caveat that the BUA signals are likely produced by more local ensembles of neurons. In addition, repeating these analyses following the removal of trials where the instantaneous frequency had a negative value (i.e., where the oscillatory cycle completely breaks down) did not significantly change the results shown in [Fig. 3](#fig03){ref-type="fig"} ([*SI Appendix*, Fig. S8](https://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1819975116/-/DCSupplemental)), suggesting that complete breakdown of the oscillatory cycle did not underlie these findings. Overall, these analyses highlight that at the level of local ensembles of neurons, abrupt changes in the alignment of BG spiking activity and cortical beta oscillations precede transiently stable conditions in cortical and basal ganglia phase locking.
{#fig03}
#### What is the relationship of transient increases in cortical and basal ganglia phase locking to the firing rate and pattern of basal ganglia neurons? {#s7}
BUA provides an effective way of measuring recruitment of ensemble activity within a given structure to cortical activity but could mask important differences in specific populations of neurons within those structures ([@r9], [@r26]). In addition, BUA signals do not allow the delineation of changes in firing rate and recruitment of neurons to the oscillation. To address these issues, we performed complementary analyses using sorted action potentials of single neurons and distinguished between the 2 major populations of neurons in the GPe, based on well-established criteria ([@r9]). GP-TI neurons fire most around the inactive phase of the cortical slow oscillation and have been shown to correspond to prototypic GPe neurons that project to the STN. In contrast, GP-TA neurons fire most around active phase of the cortical slow oscillations and correspond to arkypallidal neurons that project only to Str ([@r26], [@r27]). The mean phase angles calculated over the entire recording are shown in [*SI Appendix*, Fig. S9](https://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1819975116/-/DCSupplemental), and firing rates of these neurons are described in [*SI Appendix*, Table S3](https://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1819975116/-/DCSupplemental). The subset of striatal neurons used here (*n* = 104), which were significantly locked to cortical beta, were likely to be indirect pathway spiny projection neurons ([@r12]).
To evaluate how the rate and pattern of action potentials in the basal ganglia were modulated by cortical β bursts we first computed triggered averages, similar to the widely used spike-triggered average, using the burst onset points throughout the recording as the triggers ([*SI Appendix*, Fig. S10*A*](https://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1819975116/-/DCSupplemental)). As with spike-triggered averages, this simple technique can reveal both oscillatory and nonoscillatory modulations of instantaneous firing rate on multiple time scales. We first triggered unit activity by the burst onset points defined only by amplitude. This method did not result in any modulation of firing rate in any of the BG populations at any time scale ([Fig. 4*A*](#fig04){ref-type="fig"}), demonstrating that overall rate of action potentials is not altered on the time scale of whole bursts (see [*SI Appendix*, Fig. S10*A*](https://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1819975116/-/DCSupplemental), for example of how the opposite result could occur). When the burst triggers were adjusted to nearest oscillation peak, however, there were clear oscillatory modulations in the firing pattern of each population at beta frequency in line with the cortical β-burst time course ([Fig. 4*B*](#fig04){ref-type="fig"}). This demonstrates that modulation of the burst-triggered raw BUA ([*SI Appendix*, Fig. S4](https://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1819975116/-/DCSupplemental)) is likely underpinned by the timing of action potentials from single neurons. In line with this level of explanation, individual GP-TI, GP-TA, Str, and STN units exhibited clear oscillatory firing at the onset of ECoG β bursts ([Fig. 4*C*](#fig04){ref-type="fig"}). For individual GP and STN units, it is noteworthy that each β burst (i.e., trial on the raster plot) could evoke a highly consistent pattern of oscillatory firing for over 200 ms ([Fig. 4*C*](#fig04){ref-type="fig"}). Simulations of this analysis, under the assumption that the cortical amplitude reflects the probability that a unit will fire at a specific phase of the same oscillation, suggest that this method is resistant to changes in frequency, burst length, and firing rate within the range of variance present in the real data ([*SI Appendix*, Fig. S11](https://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1819975116/-/DCSupplemental)).
{#fig04}
### Temporal evolution of ECoG β bursts is associated with cell type-selective phase-locking trajectories of the basal ganglia neurons. {#s8}
Finally, we aimed to identify the precise temporal profiles and conditions of synchronization of action potentials recorded from specific populations of BG neurons as the amplitude of the ECoG β burst evolved. To this end, we divided each ECoG β burst into individual cycles around that containing the burst threshold (cycle 0). By using the β phases of all action potentials fired by a single neuron within a given cycle, across bursts, we could calculate cycle-by-cycle circular statistics in relation to burst onset ([*SI Appendix*, Fig. S10*B*](https://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1819975116/-/DCSupplemental), and [Fig. 5](#fig05){ref-type="fig"}). In each neuronal population, the vector length across cycles was significantly modulated by the position of cycle in relation to the burst threshold (Kruskal--Wallis ANOVA, *P* \< 0.002). GP-TI neurons (*n* = 151) displayed a significant increase in vector length 2 cycles before the burst onset ([Fig. 5*A*](#fig05){ref-type="fig"}), with around 20 and 40% of neurons significantly phase locking in the prethreshold cycles, respectively ([*SI Appendix*, Fig. S12*A*](https://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1819975116/-/DCSupplemental)). Notably, GP-TA neurons (*n* = 31) followed a similar pattern, but phase locking started after and ended before that of the GP-TI neurons ([Fig. 5*B*](#fig05){ref-type="fig"}). STN neurons (*n* = 19) also showed an increase in vector length 1 cycle before burst onset ([Fig. 5*C*](#fig05){ref-type="fig"}) but disengaged 1 cycle earlier than both types of GPe neuron ([Fig. 5 *A--C*](#fig05){ref-type="fig"}). For striatal neurons (*n* = 20), although vector length was significantly modulated by cycle position (Kruskal--Wallis ANOVA, *P* = 0.002), no individual cycle was significantly different from those preceding it ([Fig. 5*D*](#fig05){ref-type="fig"}). Despite this, around 20% of striatal neurons were significantly locked in the cycles surrounding the burst threshold, and this proportion decreased to 0 around the edges of the burst ([*SI Appendix*, Fig. S12*A*](https://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1819975116/-/DCSupplemental)). Cycle-by-cycle analysis confirmed that none of these neuron types displayed modulation of firing rate in relation to the burst onset ([*SI Appendix*, Fig. S13](https://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1819975116/-/DCSupplemental)). The spikes of GPe and STN neurons could phase lock to several consecutive cycles of the burst, suggesting oscillatory firing, whereas striatal neurons mostly locked to 1 or 2 cycles ([*SI Appendix*, Fig. S12*B*](https://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1819975116/-/DCSupplemental)).
{#fig05}
Previous studies have provided detailed information regarding the mean ECoG beta phase (mean β phase) of spikes in each of these BG neuronal populations ([@r9], [@r12], [@r26]). However, our analysis of phase slips ([Fig. 3](#fig03){ref-type="fig"}) suggests that the alignment of spiking to the ECoG moves in phase to achieve the stable conditions of synchronization that occur during bursts. We hypothesized that phase slips before and stable conditions of synchronization during bursts would be reflected in the phase at which spikes locked to the ECoG β phase as the amplitude of the burst evolved. This proved to be particularly prominent for the spikes of GP-TI neurons, the preferred β phase of which moved around 3/4 of the oscillation cycle over the rise and fall of the ECoG β amplitude, settling for only a few cycles at their mean phase during the maximum amplitude of the burst ([Fig. 5*E*](#fig05){ref-type="fig"}). In contrast, the mean phase of GP-TA and STN spikes shifted dramatically between the cycles before and containing the burst threshold but then stayed at or close to their mean phase angles for the remaining cycles with increased vector length ([Fig. 5 *F* and *G*](#fig05){ref-type="fig"}). Despite their burst-related increase in vector length not reaching significance, the angle of locking of striatal action potentials still showed distinct structure, moving to the mean phase before and during the burst ([Fig. 5*H*](#fig05){ref-type="fig"}). When viewed together, the preburst phase locking angles of GPe and STN spikes were all at the descending phase of the oscillation, before moving to the mean β phases at point of burst onset ([*SI Appendix*, Fig. S14](https://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1819975116/-/DCSupplemental)). Notably, the phase trajectory of GP-TI spikes meant that they were within 1/4 of a cycle of those of GP-TA, STN, and striatal neurons immediately before phase-locking magnitude across all neurons dropped back to baseline ([*SI Appendix*, Fig. S14](https://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1819975116/-/DCSupplemental)). The analyses of striatal neurons ([Fig. 5 *G* and *H*](#fig05){ref-type="fig"}) was likely underpowered due to their low firing rate and tendency to lock to only 1 to 2 consecutive cycles ([*SI Appendix*, Fig. S10*B*](https://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1819975116/-/DCSupplemental)), which meant that few neurons accumulated enough spikes in each cycle to calculate circular statistics. To address this, for each population we generated resampled data where 500 spikes were randomly selected across the population for each cycle, and this was repeated 100 times. This allowed us to pool data from a larger number of neurons (i.e., spikes could be taken from neurons that did not have \>20 spikes on each cycle). This analysis suggested that while smaller in magnitude than the other populations, the vector length of striatal spikes did significantly increase, with spikes firing at a consistent phase angle at or close to their mean β phase over the same time course as GP-TI neurons ([*SI Appendix*, Fig. S15](https://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1819975116/-/DCSupplemental)).
These findings suggest that temporal evolution of synchronization and oscillation in basal ganglia neuronal ensembles during cortical β bursts could be underlain by the precise phase relationships of specific populations of neurons to cortical beta oscillations and to each other.
Discussion {#s9}
==========
Beta oscillations in the cortex and basal ganglia occur in transient bursts of increased amplitude ([@r20], [@r21], [@r24], [@r28]). Here we show the time course of cortical and basal ganglia synchronization follows that of the cortical beta amplitude at the level of ensembles and single neurons and that the conditions of synchronization are consistent for each burst. The initiation, peak, and offset of cortical bursts is associated with cell type-selective trajectories of spiking in relation to the cortical phase. Thus, the generation, maintenance, and termination of network synchronization is likely underpinned by the synaptic interactions between these connected groups of neurons that evolve over the time scale of the burst.
Transient Cortical and Basal Ganglia Phase Locking Occurs Independently of Firing Rate. {#s10}
---------------------------------------------------------------------------------------
Oscillations and synchronization have become increasingly prominent in explaining the pathophysiology of the parkinsonian brain; however, there is considerable evidence that changes in firing rate across the basal ganglia play also an important role ([@r29]). Distinguishing changes in the timing of activity from those of firing rate/excitability is therefore crucial for delineating the neuronal activity that leads to parkinsonian symptoms and for the role of these activities in healthy processing. In the context of PD, the Str is particularly important in this regard as dopamine-based changes in excitability of spiny projection neurons are central to rate-based models of pathophysiology ([@r30]). The striatal neurons that contributed most heavily to the striatal BUA here were likely to be GPe projecting, indirect pathway spiny projection neurons (iSPNs), which become hyperactive following dopamine depletion ([@r12], [@r31]). Importantly, the magnitude of hyperactivity of iSPNs is positively correlated with the strength of their phase locking to cortical beta oscillations ([@r12]), suggesting that modulation of excitability is necessary for iSPNs to propagate cortical rhythms. Here we found that striatal neurons increased their cortical phase locking during β bursts, without any change in firing rate over the same time scale. Together, these findings suggest different time scales for pathophysiological changes in rate/excitability and phase locking and that the former is not the result of the latter. Despite this, the very low firing rate of SPNs under heathy conditions means that the tonic, background increase in iSPN excitability following dopamine depletion could still be crucial for these neurons to propagate cortical oscillations downstream ([@r12]). In contrast, in STN and GPe, much of the firing is driven by intrinsic currents ([@r32]). As a result, small perturbations to timing and synchronization of spiking can have a bigger impact on output than changes in excitability ([@r28]). Thus, for these populations it is less surprising that profound modulations in timing of action potentials can occur independently of rate changes, and this is consistent with evidence that modulation of oscillation, rather than rate, best predicts the effectiveness of therapy ([@r7]). Overall, our results suggest that changes in the excitability of basal ganglia neurons do not interact with beta oscillations and synchrony on the subsecond time scale.
Identifying Cortical and Basal Ganglia Interactions at the Onset of β Bursts. {#s11}
-----------------------------------------------------------------------------
Since the discovery that beta oscillations are abnormally synchronized in PD patients, a central question in the field has been how these activities are generated and propagated within cortical, basal ganglia and thalamic circuits. In patients, there is significant evidence that cortical oscillations lead those in basal ganglia ([@r11], [@r33][@r34][@r35]--[@r36]). For this reason, we used the EEG or ECoG amplitude as our reference point for the timing of other activities. However, given the looping nature of cortical and basal ganglia circuits, it seems likely that even if the cortical oscillations provide an initial reference frame, the subsequent time course of cortical amplitude is the result of interactions across the network that evolve over time. We provide an approach to this issue that utilizes the transient nature of β bursts to examine network events that occur around their onset.
Phase locking of each BG population started before the ECoG burst threshold. We do not interpret this as the BG driving the cortical burst but rather that there is a period before the burst threshold where the spikes align to a specific phase of the cortical oscillation. This early alignment of cortical and basal ganglia activity could provide the initial conditions through which a beta oscillation can be amplified and propagated across the network. Thus, it is noteworthy that phase locking in GP-TI neurons started 1 cycle before that of STN neurons. For the burst initiation, this result is at odds with experiments in MPTP-lesioned primates suggesting that beta is propagated through the corticosubthalamic hyperdirect pathway ([@r14], [@r37]). Rather, it suggests that in rats, cortical oscillations initially influence GPe neurons via iSPNs, their other main source of input. In line with this observation, highly oscillatory corticosubthalamic oscillations fail to pattern GPe unless dopamine is depleted and/or striatal neurons are also coupled to the oscillation ([@r38], [@r39]). Our recent finding that identified iSPNs lock to cortical beta oscillations provides further evidence for a possible role for striatal outputs ([@r12]). Here analysis of striatal neurons did not provide conclusive evidence to support or refute this hypothesis. Despite recording hundreds of striatal neurons, their low firing rate meant that only a fraction could be analyzed using the cycle-by-cycle method (i.e., having enough spikes in each cycle to calculate a reliable vector length). Indeed, consistent with their low firing rate, they tended to lock to only 1 or 2 cycles of the β burst. This low rate and sparse nature of striatal firing, combined with the huge volume and number of neurons in the structure, means that even higher-density recordings may be necessary to fully evaluate their role. Whatever the role of striatum, our results are consistent with a key role for GPe in network synchronization ([@r26], [@r27]). Even if STN is influenced by the indirect before the hyperdirect pathway, however, it could still be the case that convergence of the 2 pathways on to STN and GPe neurons could promote the highest levels of synchronization and oscillation amplitude. It is also noteworthy that GP-TI neurons were also a cycle ahead of GP-TA neurons (which correspond to striatal-projecting arkypallidal neurons), demonstrating that neuronal populations can have their own temporal dynamics with respect to afferent input and/or network activities even within a basal ganglia nucleus.
At the population level, the onset of phase locking could be detected as a slip in the phase relationship between the ECoG and BUA, after which the conditions of synchronization remained constant until burst offset. It seems likely that these population level events were underlain by the shifts in the preferred angle of phase locking that we observed in the spiking of all of the BG populations close to the burst threshold. In the case of STN and GP-TA neurons, the phase of spikes shifted abruptly over 1 cycle, before settling at the mean phase for those populations. In contrast, the preferred phase of GP-TI spikes moved steadily along the oscillation as the amplitude evolved across cycles, settling only briefly at the peak of the burst amplitude. Notably, the progression of the GP-TI phase meant that it eventually reached the mean phase of the other populations, at which point phase locking to cortex was lost. The axons of GP-TI neurons are in the unique position to inhibit all of the other populations simultaneously ([@r26]). Thus, the movement of GP-TI spikes between antiphase and in-phase relationships with their targets could determine whether network was more or less likely to enter and/or maintain a synchronous oscillation over time. If this is the case, an important question raised by our findings is how and why the spikes of GP-TI neurons continue to change their preferred cortical phase over time, while the other populations were relatively stable. The answer to this question might reveal how and why some β bursts last longer than others.
Implications for the Development of Closed-Loop Stimulation. {#s12}
------------------------------------------------------------
In addition to being of importance to uncovering pathophysiology mechanisms, these findings have important implications for the development of closed-loop stimulation in PD and other brain disorders. One of the most promising closed-loop approaches is adaptive DBS, where high-frequency STN stimulation is delivered when instantaneous beta amplitude crosses an amplitude (i.e., burst) threshold ([@r23]). Here we demonstrate that the commonly used burst threshold is a good predictor of the time point at which basal ganglia populations have reach phase-locking stability at their mean preferred phases, providing physiological rationale for this biomarker. Our findings suggest that it may be possible to go a step further than adaptive DBS and interact with neuronal events that precede these stable dynamics and disrupt their generation. A tractable approach to achieving this would be to use phase, rather than amplitude, as a biomarker to control the timing of stimulation as phase alignment consistently preceded amplitude changes across different signals and analyses. Phase-dependent stimulation using peripheral signals can reduce the amplitude of tremor by up to half ([@r40], [@r41]). While beta oscillations are more transient than tremor and require a central signal to track phase, we and others have already provided proof-of principle evidence that this approach can also suppress beta oscillations in the parkinsonian brain ([@r7], [@r42]). Stimulation of 1 of the BG structures at a specific cortical beta phase could prevent BG spikes from entering the phase-locking conditions associated with network β bursts, either by preventing firing at some phases and/or driving spikes at others. Given the ubiquitous nature of transient oscillations in the cortex and basal ganglia, the ability to selectively alter the events underlying network oscillations could provide a powerful scientific and therapeutic tool.
Methods {#s13}
=======
Intraoperative Recordings. {#s14}
--------------------------
### Patient details and clinical scores. {#s15}
The present study was conducted in agreement with the Code of Ethics of the World Medical Association (Declaration of Helsinki, 1967) and received local ethics approval (Ethik-Kommission Ärztekammer Hamburg). Written informed consent was given by all patients who participated in this study. Cortical EEGs and microelectrode recordings from the STN were recorded from 7 patients (4 female, 3 male; age 67 ± 3 y) with advanced idiopathic PD with a disease duration of 17 ± 8 y, during bilateral implantation of DBS electrodes in the STN, guided by microelectrode mapping. Clinical details are summarized in [*SI Appendix*, Table S4](https://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1819975116/-/DCSupplemental).
### Microelectrode recordings. {#s16}
Microelectrode recordings were acquired from 5 electrodes using a MicroGuide interoperative recording system (Alpha-Omega). While the central electrode aimed at the theoretical target ([*SI Appendix*, *Supplementary Materials*](https://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1819975116/-/DCSupplemental)), the remaining 4 electrodes were placed 2 mm around it. Signals were amplified (×20,000), sampled at 24 kHz, and filtered between 300 and 6,000 Hz. Electrode positions were classified as being in the STN using several previously described criteria: a well-defined elevation in the background activity ([@r43], [@r44]), together with presence of action potential discharge patterns with tonic irregular, oscillatory, or burst-firing features. These features allowed for a distinction to be made from the neighboring substantia nigra pars reticulata thalamus/zona incerta neurons. Cortical EEG signals were simultaneously recorded and sampled at 3,005 Hz, amplified, and filtered (×4,000; bandpass, 0 to 400 Hz); see Moll et al. ([@r45]) and Sharott et al. ([@r11]) for details. EEG signals were recorded approximately from the Fz and Cz using Ag/AgCl cup electrodes filled with conductive gel (Nicolet Biomedical). In this study, the Fz was referenced to the Cz.
Recordings from Parkinsonian Rats. {#s17}
----------------------------------
### In vivo electrophysiological recording with multielectrode arrays. {#s18}
Simultaneous extracellular recordings of unit activity and LFPs were made from numerous sites in the dorsal Str, GPe, and STN of urethane-anesthetized 6-OHDA--lesioned rats. Striatal recordings were made without any other electrodes in the brain (*n* = 7 animals, *n* = 89 recordings). In some cases, STN and GPe recordings were made simultaneously (*n* = 9 animals, *n* = 25 recordings), and in others, GPe was recorded alone (*n* = 9 animals, *n* = 19 recordings). All signals were recorded together with coincident epidural ECoG, which was recorded with a 1-mm-diameter screw above the frontal, motor cortex. Further details on 6-hydroxydopamine lesions and silicon probe recordings can be found in [*SI Appendix*, *Supplementary Materials*](https://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1819975116/-/DCSupplemental).
Data Analysis. {#s19}
--------------
Recordings were analyzed using custom-written software in MATLAB. Throughout the paper, the beta frequency band is defined as 15 to 35 Hz. From the recordings described above, in this study we have made use of the following types of signals.
### BUA. {#s20}
In experimental parkinsonism, BUA was derived from raw probe recordings by high-pass filtering the recordings at 300 Hz using a third-order Butterworth filter. We then identified spiking activity by setting a threshold at mean + 4 SD of the recording, removed a 4-ms segment around each epoch crossing this threshold, and replaced it with part of the recording that did not contain any spiking activity. Following removal of spiking activity, the data were rectified and low-pass filtered at 300 Hz using a third-order Butterworth filter. For STN microelectrode recordings from PD patients, the signal was recorded using a band-pass filter (300 to 6,000 Hz). The other processing steps were identical to processing of BUA in experimental parkinsonism. BUA activity in the beta frequency band (BUA~β~) was derived by filtering (second-order Butterworth filter) the BUA activity ±5 Hz around the frequency with highest coherence with Fz--Cz EEG/frontal ECoG within the beta frequency band.
### Single-unit activity. {#s21}
Single-unit activity was extracted from the band-pass filtered (500 to 6,000 Hz) wide-band probe signals according to the following criteria: 1) signal/noise ratio of \>2.5 and 2) spike sorting using methods such as template matching, principal component analysis, and supervised clustering (Spike2 Cambridge Electronic Design Limited). Units were classified as single units if a distinct refractory period in the interspike interval histograms could be identified.
### Spectral analysis of EEG and STN BUA in PD patients. {#s22}
EEG and STN BUA signals were downsampled to 512 Hz (MATLAB function "resample"). BUA power and EEG-BUA coherence were evaluated with fast Fourier transform-based methods using the MATLAB toolbox Neurospec, as described previously ([@r46]). BUA power and coherence between the EEG and each coincident BUA signal were calculated with a FFT size of 512, giving a frequency resolution of 1 Hz. Each BUA power spectrum was normalized using the mean SD across high-frequency spectral values (100 to 150 Hz) to convert each 1- to 80-Hz power value into a z score. The normalized power spectra were then averaged for each hemisphere. [Fig. 1*B*](#fig01){ref-type="fig"} shows the mean and SEM of these z-scored spectra for each hemisphere. For each coherence spectra, significance in 1- to 80-Hz bins was evaluated using 95% confidence limits based on the number of segments used and was independent of frequency ([@r46]). For each hemisphere/STN, we calculated the percentage of significantly coherent EEG/BUA pairs for each frequency. [Fig. 1*C*](#fig01){ref-type="fig"} shows the mean and SEM of the percentage of significant bins in a given frequency across hemispheres. BUA signals were assigned as being inside or outside the STN based on the physiologically defined borders of the structure for that electrode ([*SI Appendix*, *Methods*](https://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1819975116/-/DCSupplemental)).
### β-Burst definition. {#s23}
β-burst onset was derived from the Fz--Cz EEG in PD patients and from frontal ECoG in 6-OHDA--lesioned rats. An epoch was labeled as a β burst if the instantaneous beta amplitude of these signals exceeded the 75th percentile of amplitudes over the entire recording for at least 1 beta cycle (i.e., 50 ms). β-burst onset and offset (i.e., first and last points above the 75th percentile threshold) were adjusted to the nearest 0° of the cortical beta phase (within ±30 ms of the threshold crossing) unless otherwise stated. Bursts with a duration of 50 ms or less were excluded from further analysis due to ambiguity associated with the nature of transients that last less than 1 beta cycle ([@r47]). Cortical beta phase was derived from the angle of the Hilbert transform.
### Phase locking and PSI. {#s24}
For all PSI analyses, basal ganglia BUA and EEG/ECoG were filtered ±5 Hz around the peak BUA-EEG/ECoG coherence frequency (BUA~β~ and EEG~β~/ECoG~β)~. Instantaneous phase was derived from the Hilbert transform of the BUA~β~ and EEG~β~/ECoG~β~. Phase alignment (ϕ~j~) was computed by subtracting the instantaneous phase of the BUA~β~ from that of the EEG/ECoG~β~. Phase alignment values were corrected to be within ± π by adding 2π when phase alignment was less than −π or subtracting 2π when phase alignment was greater than π. PSI in time was computed by dividing the burst period into overlapping 50-ms epoch and deriving the PSI for each epoch using the following formula: $\left| \left( {\sum\limits_{\text{j} = 1}^{\text{j} = \text{N}}\text{expe}^{- \text{i}\varnothing_{\text{j}}}} \right) \right|/\text{N}$, where N was the number of samples corresponding to 50 ms (see [*SI Appendix*, Fig. S1*A*](https://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1819975116/-/DCSupplemental), for schematic description). PSI across bursts was computed by deriving PSI across β bursts using the following formula: $\left| \left( {\sum\limits_{\text{j} = 1}^{\text{j} = \text{N}}\text{expe}^{- \text{i}\varnothing_{\text{j}}}} \right) \right|/\text{N}$, where N was the total number of ECoG/EEG β bursts of a certain duration (see [*SI Appendix*, Fig. S1*B*](https://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1819975116/-/DCSupplemental), for schematic description). For both methods, PSI values from each BUA-EEG/ECoG pair were first z-scored and then averaged across different recording pairs. Periods of significant phase locking between BUA-EEG/ECoG were determined using a cluster-based, nonparametric statistical test (described below), which compared the time course of phase locking values aligned to the onset of a EEG/ECoG β bursts (condition A) to that calculated using randomly selected epoch of the same length taken from anywhere in the recording (condition B). The onset of phase locking was defined as the first point when the time course of phase-locking value between BUAβ and EEG/ECoG had a positive derivative before a significant peak (as defined by the cluster-based statistic) in PSI. BUA-EEG/ECoG pairs were included in analysis only if the summed coherence between 15 and 35 Hz was greater than 10% of the total coherence from 1 to 500 Hz.
### Assessment of modulation of BUA~β~ amplitude. {#s25}
ECoGs and BUAs were down-sampled to 1,000 Hz and filtered ±5 Hz around the peak ECoG-BUA beta coherence using a second-order Butterworth filter (after which they are referred to as ECoG~β~ and BUA~β~). Instantaneous amplitude in the beta frequency band was derived from the magnitude of Hilbert transform of the ECoG~β~ and BUA~β~ signals. A moving average filter of length 50 ms (box filter) was applied to the instantaneous beta amplitude since any amplitude changes in the beta frequency band would occur at a much slower rate than the average period of the rhythm of interest (i.e., 50 ms). Changes in BUA~β~ amplitude were assessed with respect to the median BUA~β~ amplitude −500 to 0 ms before a reference time point (i.e., onset of a cortical β burst). Significance was determined in the same way as described for PSI, by comparing the amplitude modulation to that of randomly selected data using the cluster-based, nonparametric statistical test for statistical comparison of time series described below.
### Cluster-based nonparametric statistical test for statistical comparison of time series. {#s26}
Statistical difference between 2 time series (condition A and condition B) was determined using the ft_timelockstatistics function from the MATLAB toolbox FieldTrip. Briefly, a paired *t* test across subjects was computed to compare condition A and condition B. Neighboring samples where the *t* value exceeded that corresponding to a predefined cluster-forming threshold (*P* \< 0.05) were considered as clusters, and *t* values inside these clusters were summed, generating the cluster-level *t* statistic. The procedure was repeated 2,000 times while randomly exchanging labels between the 2 conditions for each subject. The values of summed *t* statistics for the clusters of the original condition assignment were compared with the distribution of the maximal summed *t*-statistic values collated across the 2,000 random reassignments. A cluster would be labeled as significant if its summed *t*-statistic value exceeded 97.5% of the randomized distribution. This would correspond to *P* \< 0.025 which we use to correct for the fact that we tested for both positive and negative differences.
#### Burst-triggered averaging of spiking activity. {#s27}
Spiking activity was converted to a binary time series sampled at 1 kHz, with a 1 at the time of the peak of the each spike. To investigate the phase organization in single units, this binary signal was convolved with a Gaussian filter (reciprocal of SD 25 ms) and averaged across all ECoG β bursts, following realignment to the onset of each ECoG β burst. If a unit was not significantly phase locked to the cortical beta (Rayleigh test, alpha = 0.05), or the maximum burst-triggered spiking activity from a probe was less than 1.96 (z score) (within ±200 ms of burst onset), burst-triggered spiking activity derived from that unit was excluded from further analysis. To differentiate between phase-driven modulation and that due to slower changes in firing rate, the same analysis was performed using ECoG β-burst onsets that were not phase adjusted. (See [*SI Appendix*, Fig. S10*A*](https://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1819975116/-/DCSupplemental), for schematic description.)
### Analysis of phase slips. {#s28}
Instantaneous phase was derived from the Hilbert transform of either the filtered BUA or the filtered ECoG, and phase alignment was computed and corrected as described above ([*Phase locking and PSI*](#s24){ref-type="sec"}). We then differentiated the unwrapped phase alignment and subsequently calculated the z score for each BUA and ECoG pair. When z-scored difference between adjacent unwrapped phase alignment values was greater than 1.96 or less than −1.96, this sample was labeled as a phase slip. Therefore, phase slips included unwrapped phase alignment values that had either a positive derivative greater than 1.96 (z score) or a negative derivative less than −1.96 (z score). Samples labeled as a phase slip were set to 1, while other samples were set to 0, creating a binary time series. Finally, phase slips derived from each BUA-ECoG pair were averaged across recording pairs for the 25 longest cortical β bursts. To compute the time evolution of phase alignment between a BUA-ECoG pair around a phase slip, we computed the average phase alignment within a 50-ms window (i.e., average in time). The angular difference between the average phase alignment between a BUA-ECoG pair −50 to 0 ms before a phase slip and the average phase alignment 0 to 50 ms after the cortical β-burst onset was calculated. This was compared with the angular difference between the average phase alignment between a BUA-ECoG pair 0 to 50 ms after a phase slip and the average phase alignment 0 to 50 ms after the cortical β-burst onset ([Fig. 5*C*](#fig05){ref-type="fig"}).
### Analysis of BUA frequency stability. {#s29}
BUA frequency stability was derived by filtering the BUA ± 2.5 Hz around the frequency which was most coherent with the ECoG (f~coherent~) and then computing the Hilbert transform of the filtered BUA to derive instantaneous phase. We unwrapped instantaneous phase derived from the Hilbert transform and differentiated the unwrapped phase. We computed instantaneous frequency by scaling the differentiated unwrapped phase by 1,000/(2π). Using a moving window of width 50 ms, we computed the SD of the instantaneous frequency. The moving window was advanced every 1 ms and was not tapered. SD of the instantaneous BUA frequency was aligned to cortical burst onset and averaged across different bursts. Only BUA recordings, which had the summed ECoG-BUA coherence between 15 and 35 Hz greater than 10% of the total coherence between the 2 sites, were included in the analysis. This procedure was repeated for f~coherent~ − 10 Hz and f~coherent~ + 10 Hz. Time course of the SD of the instantaneous BUA frequency, aligned to burst onset (condition A), was compared with the time course of the instantaneous BUA frequency that was not aligned to burst onset (condition B). The cluster-based nonparametric statistical test, described above, was used to determine when these 2 time courses differed significantly.
### Cycle-by-cycle phase-locking analysis. {#s30}
The instantaneous phase and power of the ECoG were separately calculated from the analytic signal obtained via the Hilbert transform of 5-Hz-wide filtered signals (Butterworth Filter, second order) across the beta frequency range, with center frequencies of 17.5 to 32.5 Hz and a 2.5-Hz overlap. The phase-locking strength of each spike train across the whole recording was assessed by computing Rayleigh's test and the vector length using the ECoG β-phase values in each frequency band at the time of every spike. If phase locking was significant at 1 or more beta frequency ranges (Rayleigh's test, *P* \< 0.05), that spike train was included in cycle-by-cycle analysis using the ECoG~β~ phase leading to highest vector length for that neuron. The same frequency range was then also used to define the ECoG β-burst onset. Once the locking frequency had been selected, the minima of the derivative of the ECoG β phase were used to define the indices of the start and end of each cycle around each burst threshold. The cycle containing phase-adjusted burst threshold (i.e., the time point of the oscillation peak of this cycle) was designated cycle 0. Subsequent analyses were conducted on this cycle and 10 cycles on either side. For the spike train of a given neuron, we then extracted the ECoG β phase at the time of every spike occurring in each cycle across all of the ECoG β bursts for that recording (see [*SI Appendix*, Fig. S10*B*](https://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1819975116/-/DCSupplemental), for schematic description). These phase values were then used to compute the vector length and mean angle of locking for each cycle with respect to the ECoG β-burst threshold. For each group of neurons (STN, GP-TI, etc.), a Kruskal--Wallis ANOVA was used to determine whether vector length across neurons was significantly modulated by cycle position with respect to ECoG β-burst threshold (*P* \< 0.0125 with Bonferoni correction for number of groups).
If a neuron was significantly modulated by ECoG β-burst cycle, differences between the vector lengths of progressive cycle positions were examined. The first cycle with a significant increase in vector length was defined as that with a significantly higher vector length than any of the preceding cycles, as measured by post hoc Dunn--Sidak tests. Significance on subsequent cycles was assessed relative to the cycle preceding the first significant cycle. For each individual cycle (for 1 neuron), Rayleigh's test (*P* \< 0.05) was used to define whether spikes were significantly phase locked to the ECoG~β~ phase (i.e., the phases had a nonuniform distribution). Across a population of neurons, the results of these tests were used to calculate the number of neurons with significant phase locking for a given cycle and the maximum number of consecutive cycles with significant locking across the 21 cycles for each neuron. For a given neuron, if the spikes in a given cycle were significantly locked, the angle of locking for that cycle was calculated as the circular mean of the phases for all spikes in that cycle. The circular mean of these values (across neurons) for each cycle was used to calculate the mean angle of locking in a given cycle for a group of neurons. The Watson--Williams test (*P* \< 0.01) was used to test whether there was a significant difference in the mean angle of locking between consecutive cycles. All circular statistical measures were only calculated if there were \>20 spikes/phases for a given neuron or cycle. All circular statistical measures were calculated using the CircStat toolbox for MATLAB ([@r48]).
Supplementary Material
======================
We thank all the patients who participated in this study. We thank F. Vinciati for acquiring some of the primary data in the striatum of rodents. Also, we thank Ida Cagnan Duff for her patience. This work was supported by the Medical Research Council UK (MRC; award MC_UU_12024/1 to A.S. and P.B. and awards MC_UU_12020/5 and MC_UU_12024/2 to P.J.M.), by Parkinson's UK (grant G-0806 to P.J.M.), and by a grant of the German Research Council (SFB 936, projects A2/A3, C1, and C8 to A.K.E., C.G., and C.K.E.M., respectively). H.C. was supported by MR/R020418/1 from the MRC.
The authors declare no conflict of interest.
This article is a PNAS Direct Submission.
This article contains supporting information online at [www.pnas.org/lookup/suppl/doi:10.1073/pnas.1819975116/-/DCSupplemental](https://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1819975116/-/DCSupplemental).
[^1]: Edited by Nancy Kopell, Boston University, Boston, MA, and approved June 24, 2019 (received for review November 22, 2018)
[^2]: Author contributions: H.C., M.W., C.G., A.K.E., P.J.M., P.B., and A.S. designed research; H.C., N.M., C.K.E.M., A.G., W.H., and A.S. performed research; H.C., A.B.H., and A.S. analyzed data; and H.C. and A.S. wrote the paper.
| {
"pile_set_name": "PubMed Central"
} |
1. Introduction {#s0005}
===============
Anti-N-methyl-[d]{.smallcaps}-aspartate receptor encephalitis (ANMDARE) is a disorder presenting with subacute onset of seizures, psychosis, memory and language deficits, abnormal movements, and breathing and autonomic disturbances [@bb0005], [@bb0010]. Although initially identified in association with ovarian teratomas, in a majority of children and young adults, ANMDARE occurs without tumors [@bb0010]. Three-fourths of patients with ANMDARE make substantial recovery with early diagnosis, tumor resection and immunotherapy, but relapses may occur in a quarter of them [@bb0005].
In 2012, Schmitt et al. [@bb0015] described an electroencephalographic (EEG) pattern characterized by rhythmic 1--3 Hz delta activity with superimposed bursts of rhythmic 20--30 Hz beta activity overriding on each delta wave, which they named 'extreme delta brush' (EDB). This pattern since then is believed to be diagnostic of ANMDARE, occurs in 30% [@bb0015] to 58% [@bb0020] of patients with ANMDARE, and is associated with abnormal magnetic resonance imaging (MRI), high seizure burden, prolonged hospitalization and poor outcome [@bb0015].
We describe an adolescent girl with non-paraneoplastic ANMDARE, who despite persistence of EDB for nearly 2 years had a speedy and sustained response to immunotherapy.
2. Case report {#s0010}
==============
A 16-year-old right-handed girl presented in November 2016 with the history of seizures and episodic confusion of 3 months duration. She had a few seizures at the age of 5 years and was diagnosed with benign rolandic epilepsy, for which she received carbamazepine for over one year. The drug was discontinued and she remained seizure-free from the age of 6 until seizure recurrence at the age of 16. The new onset seizures were characterized by motor restlessness and confusion lasting for nearly 2 min and occurred 2 to 3 times per week, and failed to respond to adequate doses of levetiracetam and clobazam. She was a brilliant student until the recent seizure recurrence, after which there was a dramatic decline in her school performance.
On neurological examination, the findings were confined to the higher mental function testing. She had profound difficulties with writing and calculation. She had right--left disorientation and could not identify her fingers. The rest of the neurological and systemic examinations, including the optic fundi, and motor and sensory systems were normal. A diagnosis of Gerstmann\'s syndrome due to a rapidly evolving epileptogenic lesion involving the left angular gyrus region was considered.
A 3 Tesla MRI of the brain was normal ([Supplementary Fig. 1](#f0020){ref-type="graphic"}). Her EEG revealed polymorphic left hemispheric slow activity, maximally over the left parietal region with EDB morphology ([Fig. 1](#f0005){ref-type="fig"}). Both the serum and cerebrospinal fluid (CSF) tested strongly positive for NMDA receptor antibodies (by indirect immunofluorescence on transfected cells at our laboratory). There was no evidence of ovarian lesions on ultrasound of the abdomen and pelvis. Brain fluorodeoxyglucose-positron emission tomography (FDG-PET) was normal ([Supplementary Fig. 2](#f0025){ref-type="graphic"}).
We treated the patient with injections of methyl prednisolone 1 g/day along with intravenous immune globulin (400 mg/kg/day) for 5 days. She had a seizure recurrence 10 days later characterized by a painful sensation that started in the right upper limb spreading to the right lower limb, along with posturing of the right upper limb with progression to a generalized tonic--clonic seizure. Chronic second-line immunotherapy with injections of rituximab 1 g/m^2^ was started and the same dose was repeated after 2 weeks. There were no further episodes of seizures on oxcarbazepine monotherapy. Her scholastic performance came back to baseline within the next 3 months, and she completed her 12th grade examination with high scores 6 months after the initiation of immunotherapy. Her serial EEGs done at nearly 4--6 month intervals are depicted in [Fig. 2](#f0010){ref-type="fig"}. They revealed persistent focal electrical dysfunction over the left centro-temporo-parietal region with extreme delta brushes over the left fronto-centro-temporal region and infrequent left central spikes. In view of the persisting EDB, we continued biannual injections of rituximab (two doses of 1 g/m^2^, 2 weeks apart). Her serum and CSF anti-NMDA-R antibody remained positive when repeated in November 2018, at 2 years following commencement of immunotherapy. The EEG performed during the same time continued to show EDB ([Fig. 2](#f0010){ref-type="fig"}D). She continued to take oxcarbazepine 300 mg in the morning and 450 mg at night.Fig. 1The first EEG shows 1--2 Hz, polymorphic delta activity over the left hemisphere, with maximum expression over the centro-temporo-parietal region, along with superimposed rhythmic beta activity, which is better expressed over the fronto-centro-parietal region (EDB pattern). (High frequency filter = 70 Hz, low frequency filter = 0.5 Hz, notch filter = off, sensitivity = 10 μV/mm, and display speed = 30 mm/s.)Fig. 1Fig. 2Serial EEGs depict the persistence of the left hemispheric polymorphic slow activity and EDB pattern. The EEG dated May 22, 2018 (C), in addition, shows left central spike discharges. (High frequency filter = 70 Hz, low frequency filter = 0.5 Hz, notch filter = off, sensitivity = 7--10 μV/mm, and display speed = 30 mm/s.)Fig. 2
3. Discussion {#s0015}
=============
We have described here a confirmed case of non-paraneoplastic ANMDARE in an adolescent girl, who despite complete resolution of neurological symptoms and signs within 3 months of initiation of immunotherapy, had persistence of EDB and serum antibody positivity for over 2 years. Our patient presented with the four symptoms that are diagnostic of Gerstmann syndrome: finger agnosia, acalculia, left--right disorientation, and agraphia. Although the MRI and FDG-PET turned out to be normal, the EEG disclosed a focal disturbance of electrical function over the dominant hemisphere associated with EDB. Our case illustrates the immense value of scalp EEG, even in this era of advanced neuroimaging, providing not only the localization, but also the first hint to the etiology of the lesion.
Although ANMDARE is typically recognized as a multistage disease characterized by prodromal flu-like illness, followed by acute onset of neuropsychiatric manifestations, difficult to control seizures, movement disorders and autonomic instability [@bb0005], [@bb0010], atypical forms of the disease with isolated focal manifestations are being increasingly reported both in children [@bb0025] and adults [@bb0030]. To the best of our knowledge, our patient\'s presentation with Gerstmann syndrome is unique.
While brain MRI is unremarkable in 50% of patients with ANMDARE [@bb0005], [@bb0010], EEG is abnormal in most [@bb0020], [@bb0035]. In a recent study of 53 patients (35 adults and 18 children), only 2 (4%) patients had normal EEG [@bb0040]. Focal disturbance of electrical function was encountered in 73% patients. Overall, close to one-third of patients with ANMDARE show a EDB pattern, although its incidence has varied widely in different case series depending upon patient selection, clinical state and timing of EEG [@bb0015], [@bb0020], [@bb0035], [@bb0040].
Generally, with immunotherapy, the EDB pattern gradually becomes less prominent and less frequent, and resolves completely within a few weeks or months [@bb0020], [@bb0045]. In a serial EEG study of 5 children with ANMDARE and the EDB pattern, it resolved within 6 months in 3 children, but persisted in 2 of them [@bb0045]. While an infant of 6 months at onset, in whom the EDB pattern on EEG persisted for 10 months, was left with significant cognitive and motor impairment despite control of seizures with medication, a 14-year-old boy had the EBD pattern persist for 17 months and made a complete recovery [@bb0045]. To the best of our knowledge, our patient had the longest persistence of EDB reported to date.
Several studies have linked markedly abnormal initial EEGs, especially the presence of the EDB pattern, to prolonged hospitalization, high seizure frequency, abnormal MRI, poor response to immunotherapy and unfavorable outcome [@bb0015], [@bb0035], [@bb0040]. However, the adverse prognostic implication of this pattern is less apparent in some recent studies [@bb0020], [@bb0050]. Because of its frequent co-occurrence with brief rhythmic discharges, and electrographic and electro-clinical seizures, EBD has been labeled as an ictal rather than an interictal phenomenon [@bb0055]. Our patient did not have any of these clinical or EEG characteristics, despite the early presence of EDB and its persistence for 2 years.
EDB may not be unique for ANMDARE and has been reported in febrile infection-related epilepsy syndrome, hypoxic encephalopathy, brain tumor, focal cortical dysplasia, stroke and glycine receptor antibody-associated epilepsy [@bb0020], [@bb0060]. Furthermore, benzodiazepine and barbiturate induced beta activity may be misread as EDB. EDB has also been documented in association with methotrexate toxicity [@bb0065]. Our patient received none of these drugs. A recent report suspected that EDB might represent an electromyographic artifact due to rhythmic contractions of the frontalis muscle, occurring in synchrony with frontal delta activity [@bb0070]. However, its unilaterality, stereotyped morphology in multiple EEG recordings and fronto-centro-temporal distribution would strongly argue against this possibility in our patient.
Although immunotherapy for autoimmune encephalitis remains to be standardized, a recent systematic review concluded that patients treated with dual first-line immunotherapeutic drugs fared better and had lesser chance of relapse in the long-term [@bb0075], which prompted us to treat our patient aggressively. It is unclear whether ANMDARE patients with persistence of EDB, despite complete resolution of symptoms, need prolonged immunotherapy to avert relapse. In the absence of reliable guidelines, although our patient was free of neurological symptoms for over one and a half years, in view of persisting EEG abnormalities and positive serum and CSF anti-NMDA-R antibody, fearing relapse, we decided to continue immunoprophylaxis with rituximab. Our case report emphasizes, although the presence of the EDB pattern on EEG supports a diagnosis of ANMDARE, its presence and persistence may not be a reliable biomarker to forecast the response to immunotherapy and prognosis.
The following are the supplementary data related to this article.Supplementary Fig. 13 T MRI, axial and coronal sequences, shows no structural abnormalities.Supplementary Fig. 1Supplementary Fig. 2Normal FDG-PET axial sequences.Supplementary Fig. 2
Declaration of Competing Interest
=================================
None of the authors have any conflict of interest to declare.
| {
"pile_set_name": "PubMed Central"
} |
Background {#Sec1}
==========
The Istituti Fisioterapici Ospitalieri (IFO) comprise two scientific institutes: the Regina Elena National Cancer Institute (IRE) & the Dermatological Institute S. Gallicano (ISG), both located in Rome, Lazio, Italy. Despite the overwhelming challenges the entire healthcare community faced due to coronavirus disease 2019 (COVID-19) diffusion in our Region, IFO confirmed its mission to provide expert care without compromising continuity of assistance for its patients. Here we describe the strategic decisions made and the plans developed to limit the spread of COVID-19, while taking the opportunities going along with the outbreak. We also report on the effects of our actions, hoping that our experience could be helpful in promoting resilience, emergency plan and continuity of care provision within a public healthcare system.
Activation of a crisis unit and adoption of general measures {#Sec2}
------------------------------------------------------------
Since the beginning of the COVID-19 outbreak in Italy, we designed institutional protocols and internal guidelines early on to guarantee cancer treatment for our patients while minimizing COVID-19-related risks ensuring a safe environment for care providers. The first measure was the activation of a Crisis Unit, chaired by the Medical Director and including the main representatives of different units involved in the virus containment[1](#Fn1){ref-type="fn"}, that consisted in meeting once a week to discuss the current clinical scenario, and updating institutional protocols and internal guidelines through a shared decision-making process. The weekly meeting of the Crisis Unit was organized more frequently when needed until 30 May 2020 when institutional activities were fully restored. Afterwards, the meetings, held via a web-based conference platform, were attended every 14 days.
Protective and organizational measures have been implemented to provide healthcare professionals with the necessary training needed against the spread of the virus, including prevention, containment measures and hygiene recommendations: use of personal protective equipment (PPE), wearing PPE equipments in healthcare settings, PPE disposal in biohazard containers, use of screening questionnaires for patients admitted to hospital. The training of healthcare providers and PPE supply had a key role.
Further measures to preserve safety of our employees have been also implemented, including the use of smart working via a dedicated platform developed upon the healthcare emergency. Until 18 May 2020, about 250 employees were currently on smart working. Residential meetings were replaced by video conferences.
Our Institute adopted internal and external communication measures to highlight the health risks and behavior necessary to manage those risks. The Press Office and Communication Office worked synergistically to inform and constantly update both healthcare providers and patients regarding the virus prevention and containment measures by using printed brochures, visual alerts and poster distributed in the hospital.
Filters to and within the hospital {#Sec3}
----------------------------------
Restrictions of patients and employees' access to the hospital, employ assistance and cancer center continuity of care support were implemented.
### Patient admission to the hospital {#Sec4}
A triage was applied before entering the hospital. Patients were receiving a phone call 24--48 h before hospital admission to monitor their health status and to potentially identify suspected cases. If tested positive, patient was not admitted to the hospital. Furthermore, no visitors were allowed to IFO for the duration of this crisis. Figure [1](#Fig1){ref-type="fig"}a shows a flowchart pertaining to patients admitted to the hospital and encompassing four filters. The prerequisite for triage was the presence of acute symptoms, defined as fever (\> 37.5 °C), assessed by a thermal scanner, suspicion of respiratory infection, and at least one of the following data: respiratory rate \> 30/min, oxygen saturation (SpO~2~) \< 0.95% without oxygen supplementation, and dyspnea. The patient was also interviewed about potential contact with confirmed or suspected COVID-19 patients. The triage allowed us to detect an increasing proportion of suspected patients over time, ranging from 0.3--1.2%, up to June 2020, who were either referred to hospitals dedicated to COVID-19 treatment or sent home, thus avoiding their admission to the wards whenever possible (Fig. [1](#Fig1){ref-type="fig"}b). As of 30 June 2020, we identified 756 suspected patients. Restriction measures dropped visitors and/or caregivers percentage proportion from 42.9% in March 2020 to 9,15% in June 2020. Over time, the parameter most frequently associated with the condition of "symptomatic case" and/or a "case with critical and/or risk contacts" changed from being in contact with a COVID-19-confirmed case and showing dyspnoea during the early stages of the outbreak to the presence of oxygen saturation \< 0.95% (Fig. [1](#Fig1){ref-type="fig"}c). Fig. 1Flowchart of patients admitted to the hospital (**a**), number of hospital admissions during lockdown (**b**) and number of cases with symptoms or with critical contacts (**c**)
The nose--throat swab is the current standard procedure to confirm a COVID-19 diagnosis. Relying on a Microbiology and Virus Unit and associated laboratories, our Institute has been part of the Lazio CoroNET network and has analyzed almost 21,576 nose--throat swabs (from 30 March to 30 June 2020). Eighty-seven nose-throat swabs have been analyzed for an external company located in our Region. To optimize the access to swabs while preserving health system sustainability and limiting unjustifiable requests, we have defined prescribing criteria for nose--throat swabs by identifying seven potential case scenarios as illustrated in Fig. [2](#Fig2){ref-type="fig"}. Fig. 2Prescribing criteria to request nose--throat swabs
### Active surveillance and epidemiological survey of healthcare workers {#Sec5}
Healthcare personnel taking care of patients with cancer need to be safeguarded thus following PPE guidelines. A surveillance registry has been established to identify those, who may be categorized as contacts at risk and subsequently monitored by the hospital risk manager via a risk assessment questionnaire. As a result, healthcare workers can be categorized as low and high risk. Low-risk healthcare workers undergo active surveillance for 2 weeks monitoring their body temperature and symptoms daily. High-risk healthcare workers are tested for COVID-19 by nose--throat swab and if they test positive, they are quarantined at home; if they test negative, they undergo active surveillance for 2 weeks. Figure [3](#Fig3){ref-type="fig"} shows a flowchart pertaining the actions taken within IFO to manage contagion containment and isolation of suspected cases. Fig. 3Flowchart pertaining the actions taken within IFO to manage contagion containment and isolation of suspected cases
A total of 370 subjects underwent surveillance and 2,151 were screened between 18 March and 30 June 2020. Reverse transcription polymerase chain reaction (RT-PCR) was positive in 8 healthcare workers and in 16 patients (seven inpatients and nine outpatients). A total of 1,598 serological tests have been carried out of which 1,563 for employees and 35 for homeless people. Overall, 51 employees were found positive and underwent nose-throat swabs with only one being positive and subsequently undergoing active surveillance.
Reorganization of care provision to make our hospital closer to patients {#Sec6}
------------------------------------------------------------------------
Our institution reshaped the provision of care to address the urgency of not deferrable procedures (either surgeries, biopsy or chemotherapy/radiotherapy) while banning the access only for follow-up visits, according a case-by-case evaluation of the risk/benefit ratio \[[@CR1]\].
To reduce potential viral exposure at pharmacy departments and avoid in-person hospital visits to refill prescriptions, at home delivery service has been quickly implemented for those receiving either oral anticancer therapy or biologics. During the national lockdown, we have reached 119 patients with about 450 delivered treatments, thus markedly reducing hospital visits while promoting better patients' quality of life and favoring improved adherence to therapy.
Patients who are not currently receiving active therapy may be well-suited for telemedicine consultation and/or routine follow-up. A switch to telemedicine has been recently recommended in breast cancer patients \[[@CR2]\] and virtual visits have been proposed as promising tools to ensure continuity of cancer services and assistance to patients \[[@CR3]\]. We have rapidly developed a dedicated telemedicine platform, named *\#IFOConTeOnline*, through expedited physician credentialing, training and modification based on changing regulations. With 9,178 web interactions on our platform, we have reached more than 2,000 patients and provided telemedicine services, consultations, symptom control and counselling to both cancer patients and their caregivers as well as to external patients with mild and ordinary dermatological conditions \[[@CR4]\].
To support people living with cancer during COVID-19 isolation and to mitigate the effects of quarantine on mental and physical well-being, we relied on digital technology to implement two phone helplines to promote patient mental health and coping with the pandemic crisis.
Finally, we have realized that ensuring psychological support of frontline workers is also necessary to prevent burnout and attrition.
Conclusions {#Sec7}
===========
The COVID-19 outbreak has presented both challenges and learning opportunities for cancer centers \[[@CR5]\]. As suggested by Trapani et al., the key to success in COVID-19 and cancer is to ensure a continuum of healthcare never disconnecting the cancer cause from the population health needs \[[@CR6]\].
Overall, the proactive management and containment measures, including the triage protocols, have significantly aided the identification of as many as 756 patients with suspected symptoms related to COVID-19, thus limiting their admission to our cancer center. Furthermore, the activation of the active surveillance registry and the epidemiological survey allowed us to closely monitor 370 employees, to test over 200 of them and guarantee to those tested positive the immediate treatment and evaluation in referral hospitals for COVID-19. A total of 1,598 serological tests have been carried out of which 1,563 for employees and 35 for homeless people. Overall, 51 employees were found positive and underwent nose-throat swabs with only one being positive and subsequently undergoing active surveillance. We have reshaped our provision of care by preserving elective and day surgeries and urgent ambulatory services and deferring only routine follow-up visits. Nevertheless, we acknowledge that, in the long term, the capacity to undertake diagnostics and elective surgery should be further expanded to address a growing backlog of delayed diagnosis and unmet needs.
IFO
: Istituti Fisioterapici Ospitalieri
IRE
: Istitute Regina Elena
ISG
: Istitute San Galllicano
PPE
: Personal protective equipment
SARS-CoV-2
: Severe acute respiratory syndrome coronavirus 2
SpO2
: Oxygen saturation
Medical Director, who was the chairman, Pharmacy Unit Director, Economic Resources Director, Medical Dept Unit Director, Health Professional Unit Director, Human Resources Director, Quality Assessment and Clinical Risk Manager, Nurse Reference for prevention of care-related infections, Responsible of the Project "Welcome in hospital", Responsible of the Diagnostic Team, Education Reference Unit, Directors of three Clinical and Research Departments.
**Publisher's Note**
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Editorial assistance was provided by Polistudium Srl (Milan, Italy).
AD, CD, TL and GC contributed to manuscript draft and data analysis. AM, FR and BV critically revised the manuscript. All authors read and approved the final manuscript.
No external funding.
All data in our study are available upon request.
Not applicable.
Not applicable.
The authors declare that they have no competing interests.
| {
"pile_set_name": "PubMed Central"
} |
Introduction {#Sec1}
============
The ocean acts as a buffer against global warming^[@CR1]^. However, understanding all processes and fluxes that lead to carbon sequestration is very complex and acts at multiple scales, from genes to ecosystems *via* physiological processes^[@CR2],[@CR3]^. In particular, the ecosystem community structure and composition, from large organisms to molecular descriptions, is essential when studying biogeochemical processes and their variability^[@CR2]^. In that community, gelatinous zooplanktons are still poorly represented compared to other groups^[@CR4]^ such as silicifiers, calcifiers, or crustacean. In particular, their role in biogeochemical processes is still debated mainly due to the lack of information on the fate of the gelatinous biomass, even-though several recent studies support their inclusion in ecosystem models^[@CR5]--[@CR7]^. While there is no evidence that their abundances are globally increasing^[@CR8]^, one observes their impact on carbon fluxes^[@CR9]--[@CR11]^ and the structuration of the trophic food webs^[@CR12]--[@CR14]^. Modeling the ecosystem structure to better represent biogeochemical processes in the ocean is, therefore, of primary importance. However, due to scaling issues and lack of holistic information on the plankton community, marine ecosystems are still very simplified compared to their terrestrial counterparts^[@CR15]^. Today, the new generation of Ocean Biogeochemical Models called Dynamic Green Ocean Models allows resolving the biological complexity of marine ecosystems better thanks to the inclusion of multiple plankton functional types^[@CR15]^. Similarly, trait-based modeling uses functional traits such as body size, shape, with a particular emphasis on trade-off to represent the ecosystem functioning^[@CR16],[@CR17]^. Overall, general plankton compartments such as macrozooplankton or microphytoplankton summarize ecosystems in order to facilitate the parameterization of non-linear parameters. Nonetheless, such parameters remain difficult to determine because of their multi-scale implication, from physiology to ecosystems.
One of the main challenges in ecological modeling consists of acknowledging the whole biological complexity while remaining computationally tractable. To handle both opposite constraints, one advocates that a formal selection of the modeled species is a reliable solution^[@CR18]^. The sensitivity of a model to one parameter might depend on the value of other parameters. In this setting, analyzing the sensitivity of the model to single parameters in isolation is not satisfying. Moreover, standard sensitivity analyses do not consider the number of simulations that are needed to obtain accurate results^[@CR19]^. Following previous applications in engineering, the use of the Statistical Model Checking Engine (SMCE) overcomes this weakness by bringing formal confidence (trust) in the results while enhancing the range of sensitivity analysis towards considering global dependencies between parameters. First, instead of fixing parameter values to their mean observed values and performing sensitivity analysis of one parameter at a time, SMCE embeds uncertainty on parameter values inside the proposed models by assigning a probabilistic distribution to each parameter value (*i.e*., uniform distribution per default). SMCE performs model simulations by picking parameter values within their attached distributions (*i.e*., by considering the variances of parameters) and executing standard simulations. Thus, following several simulations, which implies considering several and distinct parameter choices, the SMCE performs a generalization of standard sensitivity analyses, not by analyzing the sensitivity of a single average simulation but rather by analyzing all feasible simulations and proposing general statistics of the whole; *i.e*., accurate statistical guarantees to perform predictive simulations while taking into account experimental uncertainties^[@CR20]^. Overall, considering a predictive goal at both physiological and ecosystem levels, the SMCE produces a global set of parameter values that guarantees that the model matches experimental observations despite slight parameter variations.
In the Mediterranean Sea, *Pelagia noctiluca* is the most abundant scyphozoan species^[@CR21]^. This holoplanktonic species is present all year long and has been observed without interruption since 1994^[@CR22]^. Mostly present offshore, *P. noctiluca* reaches coastal waters thanks to wind events or sea currents variations^[@CR23],[@CR24]^. Moreover, this species can realize nycthemeral migrations between 0 and 300--500 m deep. The development of the jellyfish *P. noctiluca* from oocyte to adult is strongly related to temperature and environmental food conditions^[@CR25]--[@CR30]^. It is a non-specific predator^[@CR21]^ that responds rapidly to changes in the biotic and abiotic environment^[@CR31],[@CR32]^. Indeed, like most of the scyphozoan species, *P. noctiluca* has been known to shrink its body mass when prey concentration becomes limiting^[@CR33]--[@CR37]^.
Because *P. noctiluca* is a multi-scale player (*i.e*., strong interaction from microbial communities to high trophic level), estimating its model parameters is difficult. The goal of this paper is to apply the SMCE to infer the parameters of an ecophysiological model and discuss the putative importance of a jellyfish in marine ecosystems and biogeochemical processes in the Mediterranean Sea. According to our knowledge, the SMCE approach was never used before in this context. To this purpose, we (*i*) build an ecophysiological model for *P. noctiluca* to describe the fundamental physiological processes that are involved in carbon fluxes, (*ii*) infer the model's parameters using the SMCE and (*iii*) discuss the potential contribution of *P. noctiluca* egestion to POC fluxes despite missing knowledge. The benefits of the SMCE consist not only in performing an accurate parameter estimation on several scales (from laboratory physiological experiments to *in situ* biomass distribution) simultaneously but also in emphasizing correlations between parameters of its ecophysiological model through automatic analysis. Here *P. noctiluca* is used as a scaling-up example of how one can apply state of the art verification methods in computer sciences to better estimate parameters of an ecophysiological model.
Results {#Sec2}
=======
Conceptual model {#Sec3}
----------------
To reproduce the jellyfish growth and degrowth in captivity and wild conditions, an ecophysiological model was built based on seven physiological processes (predation, ingestion, assimilation, respiration, excretion, reproduction, egestion) and constrained by 17 parameters (see Fig. [1](#Fig1){ref-type="fig"} for illustration and methods for details). Eleven parameters (*p*~*max*~, *b*~*p*~, *t*~10*p*~, *R*~*o*~, *b*~*r*~, *t*~10*r*~, *α*, *β*, *a*~*re*~, *b*~*re*~, *W*~*e*~; Table [1](#Tab1){ref-type="table"}) were defined by previous experimental data whereas six unknown parameters (*k*~*p*~, *a*~*max*~, *k*~*a*~, *c*~*re*~, *spn*, *c*~*e*~; Table [1](#Tab1){ref-type="table"}) were inferred using SMCE. During the simulations, the jellyfish carbon mass (CM) prediction was constrained by two forcing variables: temperature and zooplanktonic biomass. In controlled conditions, the temperature was fixed at 18 °C following Lilley, *et al*.^[@CR38]^ whereas prey concentration was null or estimated from a prey concentration range (F~lab~; Table [1](#Tab1){ref-type="table"}) in degrowth and growth experiments, respectively. In the *in situ* conditions, these variables were obtained by an annual climatology of sea surface temperature and zooplankton concentration from 2011 to 2015 in the bay of Villefranche-sur-Mer, France.Figure 1Conceptual diagram and equations used in the *Pelagia noctiluca* ecophysiological model. Arrows between compartments represent the biological carbon transfer following the ecophysiological processes. Symbols and units of the different variables are described in Table. [S1](#MOESM1){ref-type="media"}. The red and green colors represent all parameters deduced from the literature dataset and Statistical Model Checking Engine (*SMCE*) respectively.Table 1Values, symbols, description and units of the different parameters used in the model.Ecophysiological processParametersValuesUnitsDescription*Predation (P)p*~*max*~0.1399*gC.ind*^*−1*^*.d*^*-1*^Theoretical predation rate of 1 g CM individual at 0 °C and with unlimited food; log(C) = b.log(B) + a (adapted from Acuna, *et al*.^[@CR66]^)*b*~*p*~0.8856dimensionless*k*~*p*~\[3 × 10^*−*5^: 2 × 10^*−*5^: 2 × 10^*−*4^\]*gC.L*^*−1*^Half saturation coefficient for predation (this study)*Respiration* (*R*)*R*~*o*~2.80*μmolO*~2~.d^*−*1^Theoretical respiration rate of 1 g WM individual at 0 °C (adapted from Lilley, *et al*.^[@CR34]^)*b*~*r*~0.934dimensionlessAllometric exponent for the effect of individual mass on respiration rate (adapted from Lilley, *et al*.^[@CR34]^)*Reproduction* (*Re*)*a*~*re*~0.07dimensionlessAllometric exponent for the production of eggs (adapted from Lilley, *et al*.^[@CR34]^)*b*~*re*~4.66dimensionless*W*~*e*~1.52 × 10^*−*6^*gC.eggs*^*−*1^Eggs carbon mass^[@CR34]^*c*~*re*~\[1: 0.1: 3\]dimensionlessProportion of mucus production during spawn (this study)*spn*\[0.4: 0.1: 1\]*d*^*−*1^Spawning rate (this study)*Assimilation* (*A*)*a*~*max*~\[0.5: 0.1: 1\]dimensionlessMaximal assimilation rate (this study)*k*~*a*~\[1.5 × 10^*−*5^: 0.5 × 10^*−*5^: 5 × 10^*−*5^\]*gC.L*^*−1*^Half saturation coefficient for assimilation (this study)*Excretion (Ex)c*~*e*~\[0: 0.2: 2\]dimensionlessProportion of excretion production (this study)*Conversion factorβ*447 × 10^*−*3^*J.μmolO*~2~^*−*1^Coefficient to convert from oxygen units to energy units^[@CR67],[@CR74]^*α*2.28 × 10^*−*5^*gC.J*^*−*1^Coefficient to convert from energy units to carbon units^[@CR67]^*t*~10~1.066dimensionlessTenth root of the Q~10~ coefficient which describes by how much as a rate changes with a 10 °C increase in temperature (adapted from Lilley, *et al*.^[@CR34]^)*Food concentrationF*~*lab*~\[0.7 × 10^*−*5^: 0.1 × 10^*−*5^: 1.2 × 10^*−*5^\]*gC.L*^*−1*^Prey concentration in laboratory condition (this study)
Implementation of the statistical model checking engine {#Sec4}
-------------------------------------------------------
The SMCE is a new mathematical approach based on probability ($\documentclass[12pt]{minimal}
\usepackage{amsmath}
\usepackage{wasysym}
\usepackage{amsfonts}
\usepackage{amssymb}
\usepackage{amsbsy}
\usepackage{mathrsfs}
\usepackage{upgreek}
\setlength{\oddsidemargin}{-69pt}
\begin{document}$${\mathbb{P}}$$\end{document}$). The core of this method consists of a combination of the Monte Carlo method^[@CR39]^ and the Sequential Probability Ratio Test (SPRT)^[@CR40]^. Implemented within a computational framework, the SMCE allows simulating models or programs with uncertainties (Fig. [2](#Fig2){ref-type="fig"}). In particular, the SMCE allows a modeler to search for the optimal parameter values concerning experimental data. Initially, the modeler assigns a virtual search vector, with upper bounds, lower bounds, and resolution chosen according to prior knowledge, to each unknown parameter (step.1 in Fig. [2](#Fig2){ref-type="fig"}; Table [1](#Tab1){ref-type="table"}). This vector allows defining a global search domain for optimal parameter values as the Cartesian product of the individual search vectors of all unknown parameters. Consequently, the size of the search space, as well as the computing time, increases proportionally following the width and the resolution of the individual vectors of each unknown parameter (*i.e*., the size range of values subjects to investigation). In addition, instead of considering exact parameter values for each sample in the search space, SMCE will take into account uncertainty by using standard deviations around a sample before measuring its adequation to experimental datasets. Briefly, the essence of the SMCE is as follows: for each set of parameter values *Param* chosen from the sample space, a pseudo standard deviation (*std*) is used to create a new interval \[*Param* -- *std*: *Param* + *std*\] (step.2 in Fig. [2](#Fig2){ref-type="fig"}), yielding a probabilistic model (*i.e*., a model where the parameter values are chosen according to a uniform distribution on their intervals). The size of this deviation is chosen according to prior knowledge of the variability of each unknown parameter.Figure 2The activity diagram of the *SMCE* used to find the best combination of values for the unknown parameters in the ecophysiological model.
With such a probabilistic model, one can define a probability measure on the outputs of its simulations. Ideally, the score obtained by SMCE should reflect the probability measure of the set of outputs that best match experimental datasets. Unfortunately, computing the exact measure (written $\documentclass[12pt]{minimal}
\usepackage{amsmath}
\usepackage{wasysym}
\usepackage{amsfonts}
\usepackage{amssymb}
\usepackage{amsbsy}
\usepackage{mathrsfs}
\usepackage{upgreek}
\setlength{\oddsidemargin}{-69pt}
\begin{document}$${\mathbb{P}}$$\end{document}$) is very costly. Therefore, the SMCE resorts to statistical analysis in order to compute an estimation of $\documentclass[12pt]{minimal}
\usepackage{amsmath}
\usepackage{wasysym}
\usepackage{amsfonts}
\usepackage{amssymb}
\usepackage{amsbsy}
\usepackage{mathrsfs}
\usepackage{upgreek}
\setlength{\oddsidemargin}{-69pt}
\begin{document}$${\mathbb{P}}$$\end{document}$. In order to do so, the SMCE performs many simulations of the model according to the chosen parameter set. At each simulation, random values of the parameters are chosen from their interval, which takes into account variability, and implemented in the model before running it (step.3 in Fig. [2](#Fig2){ref-type="fig"}). Model outputs are then compared to observations before running the SMCE to inform the latter whether the predictions were correct or not (*i.e*., *True* or *False*) (step.4 in Fig. [2](#Fig2){ref-type="fig"}). From those random simulations, the Monte Carlo technique allows computing a value Θ which is an estimation of the measure of $\documentclass[12pt]{minimal}
\usepackage{amsmath}
\usepackage{wasysym}
\usepackage{amsfonts}
\usepackage{amssymb}
\usepackage{amsbsy}
\usepackage{mathrsfs}
\usepackage{upgreek}
\setlength{\oddsidemargin}{-69pt}
\begin{document}$${\mathbb{P}}$$\end{document}$, with a degree of precision ε that depends on the chosen number of simulations and represents the adequation of the obtained prediction concerning the given data (step.5a in Fig. [2](#Fig2){ref-type="fig"}). As a consequence, obtaining an estimation Θ of the measure $\documentclass[12pt]{minimal}
\usepackage{amsmath}
\usepackage{wasysym}
\usepackage{amsfonts}
\usepackage{amssymb}
\usepackage{amsbsy}
\usepackage{mathrsfs}
\usepackage{upgreek}
\setlength{\oddsidemargin}{-69pt}
\begin{document}$${\mathbb{P}}$$\end{document}$ ensures that $\documentclass[12pt]{minimal}
\usepackage{amsmath}
\usepackage{wasysym}
\usepackage{amsfonts}
\usepackage{amssymb}
\usepackage{amsbsy}
\usepackage{mathrsfs}
\usepackage{upgreek}
\setlength{\oddsidemargin}{-69pt}
\begin{document}$${\mathbb{P}}$$\end{document}$ lies in the interval \[Θ − ε; Θ + ε\]^[@CR41]^.
In order to spare computing time, the Monte Carlo technique has been combined with the SPRT method. This approach allows stopping the estimation of $\documentclass[12pt]{minimal}
\usepackage{amsmath}
\usepackage{wasysym}
\usepackage{amsfonts}
\usepackage{amssymb}
\usepackage{amsbsy}
\usepackage{mathrsfs}
\usepackage{upgreek}
\setlength{\oddsidemargin}{-69pt}
\begin{document}$${\mathbb{P}}$$\end{document}$ (hopefully before all the simulations have been performed) if the simulations that have already been performed ensure that this probability cannot be greater than a threshold (γ) (step.5b in Fig. [2](#Fig2){ref-type="fig"}). When all the necessary simulations have been performed, the SMCE either returns the estimation Θ of $\documentclass[12pt]{minimal}
\usepackage{amsmath}
\usepackage{wasysym}
\usepackage{amsfonts}
\usepackage{amssymb}
\usepackage{amsbsy}
\usepackage{mathrsfs}
\usepackage{upgreek}
\setlength{\oddsidemargin}{-69pt}
\begin{document}$${\mathbb{P}}$$\end{document}$ if it is high enough or a statement that $\documentclass[12pt]{minimal}
\usepackage{amsmath}
\usepackage{wasysym}
\usepackage{amsfonts}
\usepackage{amssymb}
\usepackage{amsbsy}
\usepackage{mathrsfs}
\usepackage{upgreek}
\setlength{\oddsidemargin}{-69pt}
\begin{document}$${\mathbb{P}}$$\end{document}$ is below the threshold otherwise. Such estimation is performed for each set of parameter values from the search space (step.6 in Fig. [2](#Fig2){ref-type="fig"}). The estimated measures of all sets of parameter values are finally compared, which allows identifying the set of parameter values that gives the best prediction for the experimental data.
To summarize, the SMCE estimates the probability $\documentclass[12pt]{minimal}
\usepackage{amsmath}
\usepackage{wasysym}
\usepackage{amsfonts}
\usepackage{amssymb}
\usepackage{amsbsy}
\usepackage{mathrsfs}
\usepackage{upgreek}
\setlength{\oddsidemargin}{-69pt}
\begin{document}$${\mathbb{P}}$$\end{document}$ for each set of parameter values (before adding the pseudo *std*) thanks to the Monte Carlo or SPRT approach by comparing individual simulations of the model with our experimental data. One accomplished such a comparison *via* a "tunnel" selection based on the standard deviations of our experimental datasets. Considering a tunnel as the range of acceptable model outputs, a simulation of a model for a given set of parameter values is "correct" if the outputs of this simulation fit inside the tunnels (see step.4 in Fig. [2](#Fig2){ref-type="fig"} for illustrations).
Calibration of the ecophysiological model {#Sec5}
-----------------------------------------
The use of SMCE allows defining vectors of optimal parameter values (*i.e*., central values for the distributions of each parameter) within the parameter space. While originally designed as a deterministic model with a mechanistic description of the biological processes, SMCE provides an estimation of a set of vectors of optimal parameter values that transforms the model into a probabilistic one (*i.e*., parameter values are described as a distribution of values). The set of vectors could change if one considers the field and/or laboratory conditions (see the color-filled zones in Fig. [3](#Fig3){ref-type="fig"} for illustration).Figure 3Vectors of the central points of the optimal parameter distributions for the growth rates measured in the laboratory (**a**), *in situ* conditions (**b**) and according to both (**c**). Color areas represent upper and lower bounds of the central points for the 50 best vectors of optimal parameter distributions. For each parameterization condition, the best vector is depicted by red lines.
The best parameters set selected in controlled condition corresponds to a spawning frequency of 0.7 days (*spn* = 0.7 ± 0.05), a maximum fraction of assimilated food of 80% (*a*~*max*~ = 0.8 ± 0.05), a half-saturation assimilation and predation constants of 5.10^−5^ (*k*~*a*~ = 5.10^−5^ ± 5.10^−6^) and 1.9.10^−4^ (*k*~*p*~ = 1.9.10^−4^ ± 1.10^−5^) gC.L^−1^ respectively, a very low excretion rate (*c*~*e*~ = 0 ± 0.025), a carbon cost for mucus production equal to 2.7 times the carbon-cost of eggs production (*c*~*re*~ = 2.7 ± 0.05) and a prey concentration around 9.10^−6^ gC.L^−1^ (Fig. [3a](#Fig3){ref-type="fig"}). The representation of 50 random simulations from the best parameter set highlights that the ecophysiological model correctly represents the jellyfish growth and degrowth in laboratory conditions (Fig. [4](#Fig4){ref-type="fig"}). In starved conditions, the model shows that the jellyfish bell diameter decreases exponentially due to carbon losses *via* respiration, excretion, and reproduction (Fig. [4a](#Fig4){ref-type="fig"}). At the beginning of the degrowth experiment, respiration, excretion, and reproduction are responsible respectively of 47%, \<1% and 53% of the total organic carbon loss, compared to 62%, \<1% and 38% at the end of simulations. However, the model slightly underestimates the biomass losses (\~11% compared to the average of the measurements). In contrast, in feeding conditions (Fig. [4b](#Fig4){ref-type="fig"}), jellyfishes grow exponentially throughout the somatic phase, reaching saturation when sexually mature though prey concentration stays constant in the environment. This equilibrium suggests that the amount of organic carbon gained by predation is equal to losses promoted by excretion, egestion, and respiration. Thus, along with other marine species, *P. noctiluca* growth can be represented by a sigmoid curve.Figure 4Comparison between the results of the laboratory experiments and the outputs of the ecophysiological model, based on fifty random simulations within the best parameters combinations defined by the *SMCE*. Subpanels show degrowth and growth simulation in (**a**) starved and (**b**) feeding conditions respectively.
In the *in situ* conditions, the 2011--2015 annual climatology of temperature and zooplankton biomass in the northwestern Mediterranean Sea (Fig. [5a](#Fig5){ref-type="fig"}) showed significant seasonal variations. In winter, both sea surface temperature (SST) and zooplankton biomass are low (13 °C, and 1.10^−6^ gC.L^−1^, respectively). In early spring, heating of the water column leads to a stratification allowing phytoplankton growth which, in turn, generates an increase of zooplanktonic biomass (5-6.10^−6^ gC.L^−1^). During summer, the zooplankton biomass decreases to 1.10^−6^ gC.L^−1^ while the SST increases to 22 °C, before decreasing below 18 °C in autumn. According to our observations (*n* = 1734), the size of *P. noctiluca* ranges between 2.1 and 21 cm throughout the year and is strongly related to environmental conditions. Three phases characterize *P. noctiluca* growth: (*i*) a slow growth in winter and early spring, (*ii*) an exponential growth as soon as the temperature and food availability increase in mid-spring, and (*iii*) a degrowth phase during summer and autumn associated to massive gametes emission^[@CR35],[@CR36],[@CR42],[@CR43]^ and respiration, which is not compensated by feeding on the scarce food available.Figure 5Comparison of jellyfish size variations between *in situ* observations and outputs of the model, based on fifty random simulations within the best parameters combinations defined by the *SMCE*. Subpanels show: (**a**) the annual sea surface temperature (blue) and annual zooplankton biomass (black) climatologies (from August 2011 to December 2015) that were used as model inputs; and (**b**) the jellyfish size variation obtained from *in situ* observations (dots) and model simulations (red lines). The blue color represents the tunnel of confidence for our *SMCE* decision.
Compared to laboratory conditions, the best parameter set selected corresponds to spawning frequency of 0.8 (*spn* = 0.8 ± 0.05), a maximum of assimilated food around 100% (*a*~*max*~ = 1.0 ± 0.05), a half-saturation constant assimilation and predation near 4.10^−5^ (*k*~*a*~ = 4.10^−5^ ± 5.10^−6^) and 1.9.10^−4^ (*k*~*p*~ = 1.9.10^−4^ ± 1.10^−5^) gC.L^−1^ respectively, an excretion rate close to 0% of the respiration rate (c~e~ = 0 ± 0.025) and mucus production equal to 2.5 time the carbon-cost of eggs production (*c*~*re*~ = 2.5 ± 0.05). The representation of 50 random simulations from the best parameter set shows a good agreement between adult size observations and simulations (Fig. [5b](#Fig5){ref-type="fig"}). However, an unusual event at 100^th^ Julian day could not be captured by the simulations and by any parameter sets. This mismatch occurs during the early spring season when observations of jellyfish are difficult because of sea state conditions. In addition, these abiotic perturbations can also impact the vertical distribution of the jellyfish leading to potentially biased observations. Interestingly, Milisenda *et al*.^[@CR44]^ showed the occurrence of different cohorts in the Messina Strait throughout the year. Compared to the size measurements carried out in this study, all observations captured mainly the size distribution of the highest cohort (adult) except in April where the observations were similar to the intermediate cohort. This result suggests that simulations do not overestimate and reproduce correctly adult size distribution while the April observations could be biased by a younger cohort.
Finally, satisfying both laboratory and *in situ* conditions, we do not introduce more variability on each individual parameter but found a compromise for the best parameter set for both conditions (Fig. [3c](#Fig3){ref-type="fig"}). Here, the goal of SMCE is to identify a set of parameters that matches all observations (laboratory experiments and *in situ* observations). In this context, the errors are the sum of errors from laboratory and *in situ* conditions. Special attention has been given to the time resolution in order not to advantage of one condition compared to the other.
Discussion {#Sec6}
==========
In the last few years, computer sciences promoted the use of the SMCE method for verifying large software models that are out of reach of standard verification methods. Its purpose is to analyze a software model in order to prove (or disprove) that it satisfies desirable properties. Although marginally applied to non-software systems (*e.g*., aeronautics^[@CR45]^ or gene regulatory networks^[@CR46]^), its use has always required extensive computational skills (*i.e*., use of dedicated modeling languages). Here we propose (*i*) to apply SMCE to biological models *per se* with no pervasive modification and (*ii*) to reach parameter estimation expectations while (*iii*) allowing a gain of global knowledge on the models besides the verification of predefined properties. While other efficient methods exist for parameter estimation^[@CR47]^, such as MCMC^[@CR48]^ or others that consider uncertainties^[@CR49]^, SMCE brings several novel aspects that cannot be obtained through existing methods. In particular, SMCE, while being inherently probabilistic, can be applied to models originally designed as deterministic ones by using distributions on parameter values. While deterministic models tend to represent the behavior of an *average system*, their probabilistic version allows representing the whole community instead. Thus, probabilistic models take into account the inherent variability of parameter values that inevitably emerges when extracting a trait parameter value from experiments that, by essence, consider a community of individuals rather than a single individual. Aside from these experimental uncertainties, the use of probabilistic models also allows taking into account uncertainties that arise from incomplete knowledge such as those produced by incomplete mechanistic descriptions in the model, or ambiguities about initial conditions^[@CR50]^.
While standard methods, such as MCMC parameterization techniques, are highly efficient, their purpose remains to find one vector of optimal values within the parameter space. Aside from the fact that the resulting parameterized models will remain mainly deterministic, these methods also do not benefit from an extensive analysis of the parameter space. On the contrary, SMCE uses such an analysis to bring additional insights. Indeed, SMCE provides an estimation of a set of vectors of optimal parameter values (again, central values for the distributions of each parameter) that could change if one considers *in situ* or laboratory conditions (see the color-filled zones in Fig. [3](#Fig3){ref-type="fig"} for illustration). Each vector of optimal parameter values is certified, and Fig. [3](#Fig3){ref-type="fig"} presents one of them in red for the sake of illustration. However, we advocate herein to consider the whole set of certified vectors to learn dependencies between parameters, as pictured in the correlogram Fig. [6](#Fig6){ref-type="fig"}, that are usually out of reach of standard sensitivity analysis. The benefits of this analysis are multiple. By embedding uncertainties with the models, one can (*i*) link parameters that concern different biological scales within the same model analysis, and (*ii*) identify independent parameters that are insights for reducing the complexity of models during their design.Figure 6Correlogram representing Spearman's correlation rank between input variable (F~lab~) and parameters (*k*~*p*~, *a*~*max*~, *k*~*a*~, *c*~*e*~, *c*~*re*~, *spn*), output variables obtained by the SMCE (Match, Score growth, Score *in situ*), and the contribution of POC exported due to jellyfish (Median export). Color scale and circle size indicate the strengths of the correlation.
For the sake of model validation, the best parameter values found by the SMCE are comparable to the ones that have been found previously, therefore confirming those isolated observations. Concerning reproduction processes, the spawning behavior of *P. noctiluca* has been widely studied^[@CR26],[@CR29],[@CR34],[@CR51]^ showing that all sexually mature organisms spawned daily until their death. This gamete emission occurs 3 hours after the start of the light and for 30 minutes. However, this behavior could also be the result of laboratory conditions. Our new model parameterized with SMCE suggested that an almost daily spawning frequency is optimal for error minimization both in the laboratory and field conditions. Spawning activation was mostly discussed for the species *P. noctiluca*, but spawning among scyphozoans seems to be related to environmental stress such as light variations^[@CR34],[@CR52]--[@CR55]^ temperature^[@CR56]^, and food availability^[@CR26],[@CR56]^. Our model supposed that mucus production during spawning represented respectively 67.5%, 62.5%, and 67.5% of total spawn in the laboratory, *in situ*, and both conditions compared to literature, which estimated at 52%^[@CR34]^. Our modeling results confirm that scyphozoan produced massive amounts of eggs daily to promote growth population and survival^[@CR57]^.
To our knowledge, no experiment investigates the DOC excreted by *P. noctiluca*. Our simulations estimated an excretion rate close to 0 in all conditions. In comparison, two different studies measured excretion rates for *Mnemiopsis leidyi* (Ctenophora) ranging between 0.4 and 61.6 μmolC.gDW^−1^.h^−1 [@CR58]^ or 0.18 and 0.86 μmolC.gDW^−1^.h^−1 [@CR59]^ with temperature ranges of 18--27 °C and 10--24 °C, respectively. *Chrysaora quinquecirrha* and *Aurelia aurita* scyphomedusae, which are closer to *P. noctiluca* from a phylogenetic point of view, released high quantities of DOC with estimates spanning 1.3--46.4 μmolC.gDW^−1^.h^−1 [@CR58]^ and 1.2--6.7 μmolC.gDW^−1^.h^−1 [@CR60]^ in temperature ranges of 14--27 °C and 16--20 °C, respectively. Most of the time, jellyfish release a large amount of highly labile DOC, which can be easily metabolized by bacterioplankton with uptake rates two or six times that bulk of DOM^[@CR10]^. Moreover, the authors showed that specific bacterial groups in the water column successfully used this matter and suggested that jellyfish promote fundamental transformation in the biogeochemical functioning and microbial loops.
In addition to rejecting specific parameter sets, we used the SMCE approach to evaluate the entire structure of models. The limits of the ecophysiological model used in this study were highlighted by the assimilation efficiency and excretion rate, which tend toward the extreme values (100% and 0% respectively) in both conditions simultaneously. Even if *a*~*max*~ values were unrealistic for *in situ* and both conditions simulations, it reached 80% in laboratory experiments. Moreover, the relationship between "Score *in situ*" and "Score growth" with *a*~*max*~ parameter had opposite correlations (Fig. [6](#Fig6){ref-type="fig"}), which means that to reduce simulation errors in the laboratory (*in situ*) condition, the assimilation efficiency values need to be lowest (highest). Here, the SMCE output analysis raises the common problem of modeling between parameterization in the field and controlled conditions.
Regarding the excretion rate, the SMCE results showed that the variability of size measurements during the degrowth experiments did not allow to constrain it, which was also confirmed by the correlogram where *c*~*e*~ does not affect laboratory score but seems to be the most influential parameters for the field conditions. To conclude, the correlogram showed that a good "Match" mostly depends on the growth in laboratory conditions, which itself largely depends on the laboratory prey concentration. These results suggest that, in addition to *c*~*e*~ and *a*~*max*~, the vector of the laboratory prey concentration tested and adapted following publication^[@CR42]^ need to be estimated more precisely in the future experiments.
The present ecophysiological model provides carbon fluxes for predation, ingestion, assimilation, respiration, excretion, reproduction, and the egestion of *P. noctiluca*. Combining the modeled egestion with *P. noctiluca* abundances in the North-West Mediterranean Sea and estimated remineralization, and sinking rates of their produced mucus could enable estimating their contribution to carbon export in the region in the future (Eq. [9](#Equ9){ref-type=""}). For example, considering that in 2013, *P. noctiluca* abundances ranged from 0 to 3.45 ind.m^−2^ with a median of 0.018 ind.m^−2^ (1^st^ and 3^rd^ quartiles: 0.003 and 0.1 ind.m^−2^ respectively; *n* = 1,371). We could estimate an average mucus export at 200 m depth between 3.10^−3^ and 0.2 mgC.m^−2^.d^−1^ in October and April, respectively. In comparison with the total carbon export at DYFAMED station^[@CR61]^ (ranging from 1.53 mgC.m^−2^.d^−1^ in July to 14.01 and 13.55 mgC.m^−2^.d^−1^ in February and April, respectively), this suggested that the contribution of *P. noctiluca* (%POC~jelly~) at 200 m in the region could range from 0.01% to 2.31% in October and August respectively, while in summer this contribution could vary between 0% and 288%, with a median value of 1.01% (first and third quartile: 0.15% and 5.6% respectively; Fig. [7](#Fig7){ref-type="fig"}). While this example shows the potential of the common use of the ecophysiological model and experiments to estimate the jellyfish contribution for carbon fluxes, it still major pitfalls that advocate for further modelings. The current model does not consider mortality and jelly falls, which could all impact fluxes estimates. Also, local POC~jelly~ predictions are compared to the monthly climatology of total POC fluxes obtained at the DYFAMED station^[@CR61]^, which is the only time series available for this region. The Supplementary Information thoroughly discusses these hypotheses and potential impacts (section *Model uncertainties and putative refinements*).Figure 7Spatiotemporal variability of the percentage of POC exported at 200 m depth due to the scyphozoan egestion in the Ligurian Sea, according to the ecophysiological model and compared to the total POC measured at the DYFAMED station in Ramondenc, *et al*.^[@CR61]^. The red and black numbers represent the mean of POC percentage exported at 200 m depth and the number of observations respectively.
This probabilistic modeling, initially designed as a determinist model, could suggest a weak impact of *P. noctiluca* on carbon fluxes, but the high biomass and prevalence of this species make it an excellent candidate for the jelly-POM concept^[@CR11]^. Indeed, similarly to what has been found for Thaliacea^[@CR62]--[@CR64]^, the relatively high abundances and predation rates of scyphozoans can impact planktonic communities locally^[@CR65]^, which therefore needs to be integrated into pelagic ecosystems modeling studies. Also, carcass decomposition coupled with excessive jellyfish excretion, can increase dissolved inorganic carbon concentration in the water column, and thus promotes the microbial food web^[@CR10]^. Overall, this modeling exercise allows us to scale up from physiological properties, to test for potential biogeochemical implications, integrating laboratories, and *in situ* observations of the jellyfish *P. noctiluca*.
Methods {#Sec7}
=======
Conceptual model {#Sec8}
----------------
To understand the role played by jellyfish for carbon cycling, it is essential to trace each physiological process where carbon is implicated. An ecophysiological 0-D model was built based on three forcing variables, which are temperature (T), food concentration (F) and initial jellyfish carbon mass (CM). The growth (G; gC.d^−1^) of any organism depends on the balance between the quantity of assimilated preys (A) and losses due to respiration (R; gC.d^−1^), excretion (Ex; gC.d^−1^) and reproduction (Re; gC.d^−1^). After ingestion (I; gC.d^−1^), unassimilated predated preys are considered as egestion (Eg; gC.d^−1^). In this way, the scope of growth and Eg could be expressed respectively by equations Eq. [1](#Equ1){ref-type=""} and Eq. [2](#Equ2){ref-type=""} below. Each variable was determined separately.$$\documentclass[12pt]{minimal}
\usepackage{amsmath}
\usepackage{wasysym}
\usepackage{amsfonts}
\usepackage{amssymb}
\usepackage{amsbsy}
\usepackage{mathrsfs}
\usepackage{upgreek}
\setlength{\oddsidemargin}{-69pt}
\begin{document}$$\varDelta G=A-(R+{Ex}+{Re})$$\end{document}$$$$\documentclass[12pt]{minimal}
\usepackage{amsmath}
\usepackage{wasysym}
\usepackage{amsfonts}
\usepackage{amssymb}
\usepackage{amsbsy}
\usepackage{mathrsfs}
\usepackage{upgreek}
\setlength{\oddsidemargin}{-69pt}
\begin{document}$${Eg}=I-A$$\end{document}$$
To estimate the assimilation rate, an estimation of ingested prey itself subject to the predation rate (P) and prey concentration (F) is necessary. The total amount of predated food is a function of an allometric relationship that depends on individual carbon mass (CM), temperature (T; °C) and food concentration, as follows:$$\documentclass[12pt]{minimal}
\usepackage{amsmath}
\usepackage{wasysym}
\usepackage{amsfonts}
\usepackage{amssymb}
\usepackage{amsbsy}
\usepackage{mathrsfs}
\usepackage{upgreek}
\setlength{\oddsidemargin}{-69pt}
\begin{document}$$P=1/(F+{k}_{p}).{p}_{{\max }}.{{CM}}^{{bp}}.{{t}_{10}}^{T}$$\end{document}$$where *p*~*max*~ is the maximum net feeding rate, *t*~10~ the tenth root of the *Q*~10~ which describes by how much rate changes with a 10 °C increase, and *b*~*p*~ is the regression coefficient. The parameters *p*~*max*~, *b*~*p*~, and *t*~*10p*~ were calibrated following the relationships between carbon weight and clearance rate defined for 8 different jellyfish species by Acuna, *et al*.^[@CR66]^ (Fig. [S1](#MOESM1){ref-type="media"}). Michaelis-Menten kinetics were performed in the predation equation, with a half constant saturation of feeding rate (*k*~*p*~). However, the parameters *p*~*max*~ and *k*~*p*~ are interconnected. To minimize the time needed for the new parameterization approach, we assumed that the most laboratory observations^[@CR66]^ corresponds to food saturating condition (*i.e*., two times higher than the maximum food concentration recorded in the *in situ* condition; *F*~*sat*~ = 0.00012 gC.L^−1^) accordingly to which the relationship between *p*~*max*~ and *k*~*p*~ were identified, such as *p*~*max*~ = 0.1399.(1 + *k*~*p*~/*F*~*sat*~).
In this model, the ingestion rate is expressed by the simple product between prey concentration and predation rate, as follows:$$\documentclass[12pt]{minimal}
\usepackage{amsmath}
\usepackage{wasysym}
\usepackage{amsfonts}
\usepackage{amssymb}
\usepackage{amsbsy}
\usepackage{mathrsfs}
\usepackage{upgreek}
\setlength{\oddsidemargin}{-69pt}
\begin{document}$$I=P.F$$\end{document}$$
However, the assimilation rate is different than the ingestion one and the model assumes that the fraction of food ingested and effectively assimilated decreases with increasing zooplankton concentration following a Holling type II Michaelis-Menten relationship:$$\documentclass[12pt]{minimal}
\usepackage{amsmath}
\usepackage{wasysym}
\usepackage{amsfonts}
\usepackage{amssymb}
\usepackage{amsbsy}
\usepackage{mathrsfs}
\usepackage{upgreek}
\setlength{\oddsidemargin}{-69pt}
\begin{document}$$A=I.(1-{a}_{{\max }}.F/(F+{k}_{a}))$$\end{document}$$where *a*~*max*~ is the maximum fraction of assimilated food and *k*~*a*~ is the half-saturation constant for assimilation.
To estimate the organic carbon losses due to the jellyfish metabolism, we considered four main processes: respiration, excretion, reproduction, and egestion rate. Many authors investigated jellyfish respiration rates^[@CR34],[@CR66]^ but we choose to adapt the approach of Lilley, *et al*.^[@CR34]^, which performed an allometric equation on *P. noctiluca* species such as:$$\documentclass[12pt]{minimal}
\usepackage{amsmath}
\usepackage{wasysym}
\usepackage{amsfonts}
\usepackage{amssymb}
\usepackage{amsbsy}
\usepackage{mathrsfs}
\usepackage{upgreek}
\setlength{\oddsidemargin}{-69pt}
\begin{document}$$R={R}_{o}.{{CM}}^{\mathrm{br}}.{{t}_{10}}^{T}$$\end{document}$$where *R*~*o*~ is the theoretical respiration rate for 1 g CM individual at 0 °C, *b*~*r*~ is the allometric exponent for the effect of individual mass on respiration rate and *t*~*10*~ is the tenth root of the *Q*~*10*~ coefficient. To convert oxygen to energy units and then energy units to carbon units, two parameters *α* (J. μmolO~2~^−1 [@CR67]^;) and *β* (gC.J^−1 [@CR36]^;), were used respectively. The parameters *R*~*o*~, *b*~*r*~, and *t*~*10r*~ were calibrated according to log-log relationship between carbon weight and respiration rate obtained by Lilley, *et al*.^[@CR34]^ (Fig. [S2](#MOESM1){ref-type="media"}).
Jellyfish excretion has been largely studied^[@CR10],[@CR58]^. DOC concentration from excretion activity was considered to scale with respiration rates, which is representative of the overall metabolic rates, as follows:$$\documentclass[12pt]{minimal}
\usepackage{amsmath}
\usepackage{wasysym}
\usepackage{amsfonts}
\usepackage{amssymb}
\usepackage{amsbsy}
\usepackage{mathrsfs}
\usepackage{upgreek}
\setlength{\oddsidemargin}{-69pt}
\begin{document}$${Ex}=R.{c}_{e}$$\end{document}$$
with c~e~ being the scalar factor.
Concerning the reproduction, when *P. noctiluca* is sexually mature, each individual spawning releases hundreds of eggs bound into mucus. In our model, eggs production was estimated thanks to the power-law relationship developed by Lilley, *et al*.^[@CR34]^ (Fig. [S3](#MOESM1){ref-type="media"}) and two additional parameters *c*~*re*~ and *spn*, that characterize mucus (as the proportion of eggs carbon) and spawning frequency (spawn per day) respectively following the equation Eq. [8](#Equ8){ref-type=""}:$$\documentclass[12pt]{minimal}
\usepackage{amsmath}
\usepackage{wasysym}
\usepackage{amsfonts}
\usepackage{amssymb}
\usepackage{amsbsy}
\usepackage{mathrsfs}
\usepackage{upgreek}
\setlength{\oddsidemargin}{-69pt}
\begin{document}$${Re}=({a}_{{re}}.{{CM}}^{{bre}}).{W}_{e}.{c}_{{re}}.{spn}$$\end{document}$$
where *a*~*re*~ and b~re~ are coefficient parameters and *W*~*e*~ is the egg carbon mass^[@CR38]^. Sexual maturity allowing egg production was attained when jellyfish size reached 4 cm^[@CR29]^.
To estimate the contribution of POC~jelly~ to the carbon pump, the sinking speed (*w*; m.d^−1^) and the remineralization rate (*k; %*.d^−1^) were implemented in our model after estimation by laboratory experiments. The amount of carbon sequestered by jellyfish at a given depth was then modeled as follows:$$\documentclass[12pt]{minimal}
\usepackage{amsmath}
\usepackage{wasysym}
\usepackage{amsfonts}
\usepackage{amssymb}
\usepackage{amsbsy}
\usepackage{mathrsfs}
\usepackage{upgreek}
\setlength{\oddsidemargin}{-69pt}
\begin{document}$${C}_{{seq}}={Eg}.w\mathrm{.k}$$\end{document}$$
In order to adapt the equations found in literature and switch easily between morphometric units, the model uses two conversions developed by Lilley, *et al*.^[@CR34]^. Thus, the jellyfish carbon mass represents 0.36% of the wet mass. Moreover, the wet jellyfish mass and carbon mass follow an allometric law with the size of *P. noctiluca*, such as:$$\documentclass[12pt]{minimal}
\usepackage{amsmath}
\usepackage{wasysym}
\usepackage{amsfonts}
\usepackage{amssymb}
\usepackage{amsbsy}
\usepackage{mathrsfs}
\usepackage{upgreek}
\setlength{\oddsidemargin}{-69pt}
\begin{document}$$\mathrm{WM}=0.075.{\mathrm{BD}}^{2.993}\mathrm{and}\,\mathrm{CM}=0.26.{\mathrm{BD}}^{3.017}$$\end{document}$$
with BD, WM, and CM representing respectively the bell diameter over lappets (*cm*), wet mass (*g*) and carbon mass (*mg*).
Model execution and observation data {#Sec9}
------------------------------------
To correctly represent growth and degrowth of *P. noctiluca*, model outputs were compared to our laboratory and *in situ* measurements as well as those carried out by Lilley, *et al*.^[@CR38]^. According to Lilley, *et al*.^[@CR34]^, degrowth and growth experiments were carried out in a room maintained at 18 °C. Regarding food concentration, no prey was used in degrowth experiments whereas jellyfish were fed *ad libitum* during growth measurements. Facing important uncertainties that represent the term "*ad libitum"*, a food concentration vector specific to laboratory condition was tested with the SMCE approach (*F*~*lab*~, see Table [1](#Tab1){ref-type="table"}).
An annual climatology based on weekly measurements of sea surface temperature and zooplankton concentrations, recorded from plankton imaging, were calculated for the 2011--2015 period. Zooplankton biomass was estimated in carbon units by converting zooplankton biovolume to biomass following the linear model of Alcaraz, *et al*.^[@CR68]^, which was developed for the northwestern Mediterranean Sea. As previously mentioned, *P. noctiluca* undergoes nycthemeral migrations from the surface towards the bathypelagic^[@CR69]^. This behavior was integrated by calculating ecophysiological rates for both sea surface temperature and bathypelagic temperature (*i.e*., 13 °C in the Mediterranean Sea). Then, the average of each ecophysiological rate at time *t* was computed to represent the average daily rates.
SMCE processing {#Sec10}
---------------
We ran 500 simulations for each parameter set which gave a 5% estimation precision and an error rate of 1% on the probability that the parameter set is correct (*i.e*., qualified as *Match*). The experiment was performed on a 64 cores CPU, 512 Go Ram, 13 To HDD computer, and ran for 8 080 minutes (*i.e*., 6 days). In the end, the SMCE returned all parameter sets that presented a *Match* greater than 70%. Additional characteristics were also returned, such as the average number of simulations that were found outside the confidence interval defined from data observation (*Score*), and the distance to the interval (*Dist*) together, for laboratory and *in situ* conditions. The ecophysiological model implementation is available at <https://gitlab.univ-nantes.fr/delahaye-b/Pelagia-Noctiluca>.
Abundance and biomass estimation {#Sec11}
--------------------------------
The Ligurian Sea is an area in the northwestern Mediterranean Sea that presents its own cyclonic circulation^[@CR70]^. Indeed, two different Modified Atlantic Waters (MAW) are mixed around Corsica and formed the "Northern current". This strong flow progresses anticlockwise eastward along the Italian and French coasts. The Northern current exhibits specific physical and biological conditions and splits the basin into three different hydrographic zones: (*i*) the peripheral zone, (*ii*) the frontal zone and (*iii*) the offshore central zone. Our study monitored the distribution of *P. noctiluca* from 2012 to 2015^[@CR71]^ with a quantitative method. Briefly, thanks to the French ship ([www.alchimie-mediterranee.fr/](http://www.alchimie-mediterranee.fr/), Alchimie), each cruise was carried out at night from Villefranche-sur-Mer (43°41′N, 7°18′E) to Calvi (42°34′N, 8°45′E). During this transect, only adult *P. noctiluca* were recognized and counted every 10 minutes. The jellyfish biomass was estimated thanks to the wet mass from sampled adult individuals.
Sinking and remineralization rate experiments {#Sec12}
---------------------------------------------
Adult *P. noctiluca* were collected at the surface with a dip net (1 mm mesh size) close to the frontal zone. Each individual jellyfish was placed in an 8 L plastic bucket filled with filtered (100 μm) *in situ* seawater. Few hours were necessary for the gelatinous organisms to egest their gut content in the form of digestive mucus. The excreted particulate matter was individually incubated in sealed vials of 33.4, 67.9 or 136.5 mL filled with filtered (0.2 μm) seawater to determine the remineralization rate. Each vial contained small optodes foil glued inside their glass wall^[@CR72]^. Optodes excited by light pulses emit a fluorescence pulse in response that depends on the oxygen concentration of the solution (quenching). The phase delay of the light response provides oxygen saturation rates for the incubator. This system (optical electrodes Presens ©), is a non-intrusive and precise method (precision: 0.4% of O~2~ air saturation) but it is sensitive to temperature and pressure. For this reason, it is necessary to calibrate each vial under controlled conditions before experimental measurements (need to re-calibrated every 2 months). In parallel, other vials were prepared without particles to obtain the control condition. Average respiration from control measurements was subtracted to experimental measurements with gelatinous mucus. All incubations (*n* = 33) ranged between 9--20 h. At the end of the experiment, each particle of mucus was frozen at −60 °C for CHN analysis. The degradation rate of organic matter was calculated thanks to a linear regression based on the decrease of O~2~ in time. Thanks to the final amount and degradation rate, both carbon content at the initial time and remineralized can be estimated. In another way, sinking speed rate experiments were conducted in the laboratory. The mucus remineralization rate was estimated at 0.034 d^−1^ function to the CO~2~ consumption measurements (Table [S1](#MOESM1){ref-type="media"}).
All other individual digestive mucus was used to estimate particles sinking speed, inspired by the previous study^[@CR73]^. One by one, each particle (*n* = 19) was placed in the first centimeters of a large graduated plastic bucket, which was 38 cm tall and 30 cm in diameter, full with filtered *in situ* seawater (0.2 μm) in order to conserve identical water density. The time taken by the particles to reach between 15 and 20 cm was used to estimate the sedimentation rate. The sinking rate of the mucus produced by *P. noctiluca* ranges between 384 m.d^−1^ and 1329 m.d^−1^ with a median equal to 751 m.d^*−*1^ (Table [S2](#MOESM1){ref-type="media"}). The sinking rates were positively correlated with the mucus size (r^2^ = 0.6, *p* \< *0.05*) and showed similar properties to marine snow. Finally, based on these two estimates, a mucus sinking between 0 and 200 m is expected to lose 1% of its carbon weight.
Supplementary information
=========================
{#Sec13}
Supplementary information.
**Publisher's note** Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Supplementary information
=========================
is available for this paper at 10.1038/s41598-020-62357-5.
This research was partly supported by MEGALODOM and PENELOPE, two CNRS MITI funded projects and the ModelOmics project of the Emergence program of Sorbonne Universités. This work was in part developed during Simon Ramondenc's thesis (Analyse des variations spatio-temporelles du zooplancton gélatineux et son effet sur les flux de matières à l\'aide d\'une approche combinant expérimentation et écologie numérique, 2017) funded by Sorbonne Universités.
S.R., D.E., L.G., F.L., and B.D. designed and performed the study; S.R. and B.D. analyzed data; S.R., D.E., L.G., F.L., and B.D. wrote the paper.
The authors declare no competing interests.
| {
"pile_set_name": "PubMed Central"
} |
Background {#Sec1}
==========
According to the statistical report released from Primary Brain and Central Nervous System Tumors Diagnosed in the United States in 2015, thirty out of every 100,000 people in the general population suffered from a primary intracranial tumor requiring craniotomy for tumor resection. Glioma accounted for 60.8% of all primary intracranial tumors \[[@CR1]\], of which most of these operations require general anesthesia. These patients have vulnerable central nervous systems because of their intracranial tumors (e.g., gliomas are aggressive lesions that invade central neuronal structures anatomically, and at the neuropathologocial level disturb neuronal connections to impair neural function). Compared to intact brains, neuronal network connection disruptions from intracranial lesions and central nervous system inhibition by anesthetics yield a "double insult" to neurosurgical patients, so to speak, and may make them more sensitive to general anesthesia and surgery, leading to both short and long-term neurologic and neurocognitive dysfunction. Anesthesiologists attempt to choose the anesthetic regimen cautiously, taking into account anesthetics' effects on cerebral physiology (e.g., intracranial pressure and cerebral perfusion pressure), as well as to minimize the seeming disturbance to neuronal function by facilitating rapid emergence and orientation during recovery. The less postoperative neuronal function is affected by anesthetics, the more accurately the patient's intracranial disease-related status is reflected, and this is beneficial to neurosurgeons so as to correctly evaluate neural function. The optimal postoperative treatment based on neurological evaluation contributes to reduction of neurologic complications and is critical for long-term quality of life. Therefore, it is important to investigate how different mechanisms of general anesthesia affect postoperative neural function and anesthesia recovery in patients with supratentorial (frontal-parietal-temporal) gliomas in both a qualitative and quantitative (over time) manner, as this may guide anesthetic choice in supratentorial craniotomies for gliomas to ensure a reasonable, safe, and economical perioperative anesthetic regimen.
Inhalational anesthesia and intravenous anesthesia are widely used in current neurosurgical anesthesiology \[[@CR2]\], and sevoflurane combined with remifentanil or propofol combined with remifentanil are both accepted as general anesthesia strategies. Comparative studies have been done to investigate the effects of different anesthetics on craniotomies \[[@CR3]\], both primary and secondary outcomes focused on the changes in intraoperative physiological parameters \[[@CR3], [@CR4]\], laboratory results \[[@CR3]\], and anesthesia recovery indices during emergence \[[@CR5], [@CR6]\]; however, neuronal disease-specific study is lacking, and there is no evidence related to how postoperative neurological function and neurocognitive outcomes are affected by different general anesthesia strategies \[[@CR7]\]. Patients with neurological diseases, e.g., brain tumors, are usually excluded in studies examining the correlation between anesthesia and neural function \[[@CR8]\], as those diseases are major confounders leading to nervous system dysfunction, however, anesthetic effects on neural function cannot be ignored in neurosurgical patients, let alone the previously mentioned "vulnerable" brain tumor patient who is even more susceptible to anesthetics \[[@CR9]\]. Recent findings have reported patients with supplementary motor area (SMA) lesions presenting with intraoperative neurological deficits for awake craniotomy which could not be explained anatomically by their intact corticospinal tracts as detected by cortical stimulation or cortical mapping \[[@CR10], [@CR11]\], and these deficits were reversible after operations over time without further intervention. Although this finding may have been related to lesion location, which is called "SMA symptom", this phenomenon cannot exclude the possibility of residual anesthetic-induced neurological deficits in both the intraoperative and early postoperative periods.
It has been reported that in some supratentorial brain mass patients, a small plasma concentration of sedative could significantly worsen neurologic deficits before any operative intervention \[[@CR9]\], but there is no evidence related to how residual anesthetics affect neurologic deficits after operations. Repeated neurological evaluations after neurosurgeries can help to assess whether surgical interventions are successful and whether long-term neurologic outcomes are primarily related to diseases and surgeries, while short-term neurologic outcomes are subject to perioperative care beyond the above factors; despite only a few hours of surgical intervention for brain tumor, the overlap of general anesthesia and its effects, after surgery is completed, can linger for hours to days, which may worsen neurologic function and confuse providers. However, in the early postoperative stage, it is unknown whether sevoflurane combined with remifentanil and/or propofol combined with remifentanil could result in unexpected neurologic function deterioration, and whether one strategy is inferior or superior to another.
In this study, we will use sevoflurane combined with remifentanil or propofol combined with remifentanil to maintain anesthesia during craniotomy in patients with supratentorial gliomas to determine whether they have comparable effects on neurologic function and neurocognition in the early postoperative period. Since both sevoflurane and propofol are GABAergic anesthetics, we hypothesize that these two general anesthesia methods equivalently affect early postoperative neural function in this group of patients.
Methods {#Sec2}
=======
Study purpose {#Sec3}
-------------
\(1\) Primary purpose: To investigate neurologic function in the early postoperative period for patients with supratentorial intracranial gliomas under inhalational general anesthesia compared to intravenous general anesthesia. The hypothesis is that the difference in postoperative National Institute of Health Stroke Scale (NIHSS or modified NIHSS) score changes under the two general anesthesia methods are not different, with the score difference within − 1 to 1.
\(2\) Secondary purpose: To compare the effects of inhalational general anesthesia versus intravenous general anesthesia on neurocognition, hemodynamics, cerebral physiology, anesthesia recovery quality, pain scores, anesthesia expenses, and stress responses in patients with supratentorial intracranial gliomas undergoing elective craniotomy.
Trial design {#Sec4}
------------
The study is a prospective, single-center, open label, randomized parallel arm equivalent clinical trial comparing sevoflurane and propofol, both combined with remifentanil, general anesthesia (Sevoflurane-remifentanil versus Propofol-remifentanil group) in patients with supratentorial gliomas; it will be carried out in the neurosurgical operation room in Beijing Tiantan Hospital (a large city), Capital Medical University, Beijing, China.
Populations {#Sec5}
-----------
Adult male and female patients scheduled for elective craniotomy under general anesthesia with supratentorial (frontal-parietal-temporal) gliomas diagnosed by magnetic resonance imaging (MRI) are eligible for the study. All patients must sign institutional approval and informed consent before enrollment, the written informed consent will be obtained 1 day or a few days before operation, when patients are seen by anesthesiologists in the ward. The information and reasons why eligible patients are not recruited to the trial will be documented. The inclusion and exclusion criteria are listed in Table [1](#Tab1){ref-type="table"}. Table 1Inclusion and exclusion criteriaInclusion CriteriaPatients aged between 18 to 65 years old with American Society of Anesthesiology (ASA) status I \~ II who are scheduled for elective craniotomy for the treatment of supratentorial gliomas must fulfill the following:1Frontal-Parietal-Temporal glioma diagnosed by preoperative MRI2Glasgow score of 15 without preoperative symptomatic elevated intracranial pressure3New and/or recurrent intracranial gliomas are allowedExclusion Criteria1Unable to comprehend and cooperate with the neurologic examination2Emergency craniotomy or changed to emergency from elective craniotomy3Insular lobe is invaded by glioma4Scheduled intraoperative motor evoked potential monitoring5Patients with traumatic brain injury, intracerebral hemorrhage, or cerebral vascular diseases6Patients with prolonged emergence, postoperative mechanical ventilation, and/or sedation dependence due to a definite reason (e.g., surgery itself or tumor location)7Hypothalamic dysfunction8Radiotherapy and/or chemotherapy before surgery9Uncontrolled hypertension or severe heart disease that impairs cardiac function (New York Heart Association Functional Classification ≥ III)10History of related anesthetic allergy11Severe endocrine system dysfunction that impair metabolic index12Impaired mental status13Drug and/or alcohol abuse14Pregnant and/or lactation period patients15Neuromuscular diseases16Infectious and/or immune diseases with positive biomarker(s)17Body mass index \> = 35
Patients will be recruited in Beijing Tiantan Hospital, Beijing, China. The potentially eligible patient will be screened and contacted by a trial team member who explains the study and ascertains the patient's interest 1 or 2 days before the scheduled operation day. If interested in enrolment, the patient will receive the detailed trial explanation and written consent form.
Randomization and blinding {#Sec6}
--------------------------
Permuted-block randomization will be used with a block size of 4 and an allocation ratio of 1:1 to either sevoflurane-remifentanil group or propofol-remifentanil group. Random allocation sequence will be based on a computer-generated random digits table. One investigator who will not participate in anesthetic management or follow-up will implement the randomization and enroll patients, and allocation will be concealed in a sealed opaque envelope until patient enters the operating room. Randomization will occur for patients conforming to the above criteria, and written informed consent will then be obtained from themselves or their next-of-kin. Since the two general anesthetic administration routes are distinct (inhalational versus intravenous), patients and anesthesiologists cannot be blinded. An independent team who is not involved in the intraoperative management will be in charge of postoperative follow-up and is blinded to the intervention.
Interventions {#Sec7}
-------------
### Standard anesthetic management {#Sec8}
On the day of operation, patients will be admitted into the operating room to be randomly assigned into either the sevoflurane-remifentanil or propofol-remifentanil group. Vital signs including electrocardiography (ECG), blood pressure (BP), heart rate (HR), pulse oxygen saturation (SpO2), end-tidal carbon dioxide (ETCO~2~), body temperature, and urine output will be monitored throughout the study. 100% fraction of oxygen will be provided by mask to the patients for preoxygenation for 5 min prior to anesthetic induction. No premedication will be given.
All patients will be induced with 0.3 μg/kg sufentanil, 2--2.5 mg/kg propofol, and 0.7--0.8 mg/kg rocuronium or cisatracurium. After tracheal intubation, mechanical ventilation will be established, at a tidal volume of 8-10 ml/kg, respiratory rate of 12--15/min, inspiratory:expiratory ratio of 1:2, fraction of inhaled fresh oxygen as 60%, and flow rate of fresh gas as 1.5 L/min. After induction, anesthesia will be maintained according to one of the two group allocations: (1) 6-8 mg/kg propofol with 0.05--0.2 μg/kg remifentanil for the propofol group or (2) 1.3--1.5 Minimum Alveolar Concentration (MAC) sevoflurane with 0.05--0.2 μg/kg remifentanil for the sevoflurane group. The dosage of anesthetic will be adjusted according to the bispectral index (BIS) value, which will be maintained between 40 and 60. 0.5% ropivacaine 1 to 2 mL for each injection point will be used for scalp nerves block before the start of surgeries. Sufentanil 5--10 mcg can be given, by anesthesiologist discretion, to alleviate potent stress responses when head pins are placed or scalp incision is performed, based on the hemodynamic parameters. Additional sufentanil bolus at the dose of 5-10mcg during operation is allowed, the last bolus should be given ahead of at least 60 min before expected end of surgery. The muscle relaxant cisatracurium will be infused at 0.1 mg/kg/h during the operation for all patients and stopped once the bone flap is secured. Propofol and sevoflurane will be reduced according to BIS and hemodynamic parameters once the bone flap is fixed and stopped at skin dressing.
At the end of the operation, ondansetron 4 mg will be prophylactically administered to all patients to prevent nausea and vomiting, and tramadol 1.5-2 mg/kg will be given if patients are experiencing rigors/chills. Neostigmine 20-40mcg/kg and atropine 0.5-1 mg will be available to antagonize residual muscle relaxation when deemed necessary by train-of-four (TOF) twitch monitoring. Peripheral nerve stimulator will be performed for TOF monitoring, achieving TOF ratio of 1.0 at the ulnar nerve/adductor pollicis is considered an adequate sign of recovery from muscle relaxant.
Adverse hemodynamic responses will be recorded and classified as hypertension, hypotension, tachycardia, or bradycardia, which require standard treatment according to the protocol (Table [2](#Tab2){ref-type="table"}). These episodes of blood pressure and heart rate changes and the medication administered during their treatment will be recorded as "hypertension" (MAP \> 20% above preoperative baseline value) or "hypotension" (MAP \< 20% below preoperative baseline value), Tachycardia and bradycardia are defined as HR \> 100 bpm and HR \< 45 bpm, respectively. Standard treatment for hemodynamic disturbances are shown in Table [2](#Tab2){ref-type="table"}. Table 2Treatment for hemodynamic disturbancesHemodynamic FluctuationDefinition\
(if any of the below changes are sustained for equal and/or longer than 5 min)Standard Treatment AlgorithmHypertension EpisodeMAP \> 20% above preoperative baselineIncrease propofol or sevoflurane concentration according to BIS, increase remifentanil infusion rate, or administer 5mcg sufentanil; if correction still not achieved, nicardipine will be given as bolus and/or infusionHypotension EpisodeMAP \< 20% below preoperative baselineDecrease propofol or sevoflurane concentration according to BIS, give adequate volume loading; if correction still not achieved, vasoactive agent administration (dopamine, norepinephrine, or phenylephrine) will be given with the dose and infusion rate adjusted according to the blood pressure responseTachycardiaEpisodeHR \> 100 bpmEsmolol bolus and/or infusion according to heart rate responseBradycardiaEpisodeHR \< 45 bpmAtropine administration*MAP* mean arterial pressure, *HR* heart rate, *bpm* beat per minute
Patient-controlled intravenous analgesia (PCIA) will be used for postoperative pain control with 100mcg sufentanil and 16 mg ondansetron diluted to a total volume of 100 ml within normal saline being prepared, with a bolus dose set at 0.5 ml, a lockout time set at 15 min, and a background infusion of 2 ml/h. PCIA will be started after the patient is discharged from the operating room.
### Anesthesia monitoring {#Sec9}
Anesthesia management will aim to achieve targeted physiological parameters. The blood pressure will be monitored by radial artery catheter placement, and the target value is defined as the range of ±20% of the baseline MAP value, which is defined as the average MAP of the first three values measured after the patient enters the operating room and before induction. When the blood pressure is out of this range, measures will be taken such as changing the infusion rate of crystalloid or colloid and remifentanil, or giving bolus injection of sufentanil or vasoactive agent (such as 0.5 μg/kg/min phenylephrine or 0.01 μg/kg/min norepinephrine). Heart rate will be maintained between 50 and 90 beats per minute. SpO2 will be kept ≥98% during the operation. End-tidal CO~2~ will be maintained between 30 and 35 mmHg by adjusting ventilation parameters. Core body temperature will be kept between 36 and 37 degree Celsius. Plasma glucose concentration will be maintained between 5.0--7.8 mmol/L.
Discontinuation criteria {#Sec10}
------------------------
### Massive hemorrhage or massive transfusion {#Sec11}
Massive intraoperative bleeding can occur, and this study defines massive bleeding as blood loss exceeding the entire blood volume within 24 h from the start of operation or 50% of circulating blood volume loss within 3 h during operation requiring emergency intervention. Intraoperative massive transfusion (transfusion of more than 10 units of packed red blood cells) may be needed if massive hemorrhage is encountered. If bleeding occurs to this extent and meets either or both of the above criteria, the trial will be discontinued for this particular patient.
### Anaphylaxis during the operation {#Sec12}
Anaphylaxis rarely but occasionally occurs intraoperatively, especially in the setting of antibiotic and muscle relaxant administration, but also with the administration of other medications, and normally can be treated quickly. If such a reaction occurs and is severe, necessitating discontinuation of the operation, the patient will be withdrawn from the trial. Anaphylaxis will be reported as an adverse event.
### Venous air embolism {#Sec13}
Significant venous air embolism has been observed in rare cases, but neurosurgery carries the potential risk for this sometimes intractable complication, especially when the operative intervention occurs near venous sinuses or in the sitting position \[[@CR12]--[@CR14]\]. Venous air embolism may result in severe hypoxia and hypotension which can impair cerebral perfusion and oxygenation. If this happens, timely intervention and resuscitation needs to be carried out and the patient will be withdrawn from the trial. This will be reported as an adverse event.
### Postoperative coma {#Sec14}
If the patient is in a coma (nonresponsive) after operation for any reason, the subject will not be able to cooperate with any neurological assessment, the subsequent follow-up will need to be discontinued, and this will be noted in the neurologic complication section. The patient will be withdrawn from the trial.
### Outcomes {#Sec15}
Patients will have been screened 1 to 2 days before operation, and the demographic characteristics, clinical manifestations, and past history obtained at this time. Detailed descriptions of the lesion in MRI will be obtained. The lesions' pathologic diagnoses will be obtained 2 weeks after the tumor removal. Neurologic function will be assessed using the National Institute of Health Stroke Scale (NIHSS/mNIHSS) at baseline and in the postoperative period. NIHSS scores 15 items including consciousness, visual function, facial and motor function, ataxia, sensory and language function, and attention, and the total score ranges from 0 (no deficit) to a maximum of 42. A modified version of the NIHSS (mNIHSS) abridges to 11 items by deleting level of consciousness, facial palsy, limb ataxia, and dysarthria; the sensory item is collapsed from 3 to 2 choices \[[@CR15]\]. Since there is no reliability and validity evidence comparing NIHSS and mNIHSS use in the intracranial tumor population, we will score both as primary outcomes in this study. The scoring system is shown in [appendix](#Sec26){ref-type="sec"}. Mini-mental state examination (MMSE) will be used to assess the cognitive performance of patients at baseline and after operation. During anesthesia recovery, Anesthesia Recovery Quality and the modified Aldrete scale will be used to assess the recovery condition of patients. Visual Analog Scale (VAS) will be used to assess postoperative pain. QoR-40 (Quality of Recovery from Anesthesia) will be used at 2 days after operation to assess the recovery condition and the satisfaction of the patient related to anesthesia. All adverse events will be recorded. The schedule of data collection is shown in Fig. [1](#Fig1){ref-type="fig"}. Fig. 1Schedule of enrollment, interventions and assessments for the SPRING trial
Study endpoint {#Sec16}
--------------
The primary endpoint will be NIHSS score change within 4 h after general anesthesia by sevoflurane-remifentanil or propofol-remifentanil when observer's assessment of alertness/sedation (OAA/S) reaches 4. The time to reach OAA/S score of 4 at the first time point will be recorded.
Secondary endpoints will be as follows: The NIHSS score change at postoperative day 1 and postoperative day 2 from the baseline, the level of OAA/S in these two time points will be recorded.Hemodynamic stability including blood pressure, heart rate, episodes of hypertension, hypotension, tachycardia, and bradycardia;Intraoperative brain relaxation (Brain relaxation assessed immediately after opening of the dura on a scale ranging from 1 to 4 (1 = perfectly relaxed, 2 = satisfactorily relaxed, 3 = firm brain, 4 = bulging brain);Quality of anesthesia emergence including time of eye opening, time of emergence, coughing during extubation, postoperative nausea and vomiting (PONV), shivering, and agitation;Quality of anesthesia recovery assessed by QoR-40 at postoperative day 2;Changes of cognitive function at postoperative day 2 assessed by MMSE;Postoperative pain at PACU, 1, and 2 days after surgery evaluated by VAS;Postoperative neurological complications (e.g., intracranial hematoma, infection, seizure etc.);Anesthesia expenses and in-hospital total expenses;
Data collection and management {#Sec17}
------------------------------
Anesthesiologists on the follow-up team are all trained and received certification from the NIHSS official training program website ([www.nihstrokescale.org](http://www.nihstrokescale.org)). Data collection ends when patients are discharged from the hospital. All the information will be recorded on a Case Report Form (CRF), and raw, non-numerical data are coded for data storage, review, tabulation and analysis. Data will be entered at our medical center and stored and monitored securely in an electronic databases. Each of the data collection forms and the detailed information will be discussed on an item-by-item basis. Double data entry will be used, different data entering individuals will use standardized terminology and abbreviations, training will be performed regarding entering data on forms, and we will respond promptly to data discrepancy queries and general concerns about overall quality. Any missing data or errors in the data will be summarized along with detailed descriptions, and will be queried by checking the original forms. Data safety and monitoring inspectors will evaluate the trial safety, efficacy, and any ethical issues. The data to be collected and the procedures to be conducted at each visit will be reviewed in detail. There will not be a formal data monitoring committee (DMC) in this study, because the intraoperative anesthesia management is standard with minimal risks and the follow-up duration is short for this trial.
Paper case report forms are stored in numerical order and kept in locked cabinets. The electronic data will be saved in a database with password protection, and the passwords will be changed on a regular basis. Database backup will be performed once a month. All the original files will be maintained in storage for a period of 5 years after completion of the study.
Participant retention {#Sec18}
---------------------
Once a patient is enrolled in the study, the team will make every effort to follow the patient for the entire study period. For patients whose consent is withdrawn, who are lost to follow-up, or discontinue for any reason, the reason will be recorded in the CRF in order to better interpret the results. The strategies that are used to improve participant retention are: provide letters regarding the evaluation time schedule prior to data collection to remind patients and relatives of each assessment and test;provide adequate communication when visiting patients in the ward to let them be informed of upcoming data collection;limit patient burden related to follow-up visits to the greatest extent possible;patients who withdraw for follow-up assessment for one of the primary or secondary outcomes can continue with assessments for the other outcomes.
Sample size calculation {#Sec19}
-----------------------
Sample size calculation was conducted by PASS software (NCSS, LLC, USA). The mean difference of NIHSS change between the two types of anesthetics was estimated as 0.5, the estimated within group standard deviation was 1.5, and 1.0 was set as the equivalent margin based on clinical significance. Z-test was used as power analysis of equivalence tests of two independent proportions to detect equivalence. The minimum number of cases was 160 per group to achieve 80% power to detect a possible equivalence between the two groups. Considering the possibility of early termination during the study and allowing for a 10% drop-out rate, this number was raised to 176 subjects for each group. As two groups were to be studied, we aim to enroll 352 subjects.
Statistical analysis plan {#Sec20}
-------------------------
Kolmogorov-Smirnov testing will be used for evaluating if the continuous variables have a normal distribution or not. Descriptive statistics will be reported as means with standard deviation for normally distributed data; medians with interquartile range will be described for non-normally distributed data; for categorical variables, count (percentage) will be described. The difference in postoperative NIHSS scores at certain time points between the propofol and sevoflurance groups will be analyzed by t-test or Mann-Whitney U test depending on normal or non-normal distribution, respectively, and 95% confident interval will be determined for the NIHSS score change and difference between the two groups. The changes in VAS scores, intraoperative parameters, and stress response biomarkers over the observation time will be compared using repeated measurement, and post hoc analysis will be used for sensitivity analysis. For other secondary outcomes such as brain relaxation, MMSE score, anesthesia quality score and medical expense, the difference between the two groups will be analyzed by t-test or Mann-Whitney U test. All the categorical variables will be analyzed by chi-square testing. Missing data will be adjusted using inverse probability weighting and worst-case imputation scenarios. Statistical significance is defined as a type I error of 0.05, and the nature of the testing is two-tailed. Analyses will be conducted using Statistical Package for the Social Sciences (SPSS) version 17.0 (Chicago, IL, USA).
All data will be analyzed according to an intent-to-treat, such that all randomized patients originally allocated to the propofol or sevoflurane anesthesia arm at the time of randomization will be used for efficacy analysis. No interim analysis is planned. Worst observation carried forward method will be used to handle missing data.
Reporting of adverse events {#Sec21}
---------------------------
All adverse events associated with this trial will be closely monitored until the adverse events are resolved, stable, or confirmed as having no relation to the trial. Once adverse events occur, they will be immediately reported to the research department and to the principal investigator to determine the severity of the adverse events and the consequences of the injury. All adverse events associated with this study will be recorded and reported to the Ethics Committee within 1 week, which will be part of the annual report. The principal investigator will be responsible for all reported adverse events.
Protocol amendment {#Sec22}
------------------
The principle investigator of the SPRING trial will be responsible for any decision to amend the protocol. If there is any modification that impacts the conduct of the study and affects potential safety or benefit to the patient, such as trial objectives, design, patient population, sample sizes or study procedures in the protocol, the principle investigator will notify and gain approval from the China Ethics Committee of Registering Clinical Trials prior to implementation, and the related trial registration information will be updated in the Chinese Clinical Trial Registry.
Trial results and publication {#Sec23}
-----------------------------
Publications and presentations related to this trial will always maintain and protect the integrity of the major objectives of this study. Each paper or abstract will be reviewed and approved by the principle investigator before being submitted. The SPRING trial may terminate at a planned target of 6 months after the last patient has been randomized. We expect to take about 4 to 6 months to compile the final results for publication in an appropriate medical journal. The results will be reported and disseminated to the public regardless of the magnitude or direction of effect. The principle investigator of the study should be considered as the lead author. All professionals that have participated in the SPRING trial for a minimum of 3 months will be listed in the authorship, and those who do not fulfill such criteria will be acknowledged in the publication.
Discussion {#Sec24}
==========
In current neurosurgical anesthesia practice for elective craniotomy for brain tumor resection, either an inhalational agent or intravenous propofol in combination with a short-acting opioid and muscle relaxant may be used. For patients with normal intracranial pressure and those without need of intraoperative neurophysiological monitoring, there does not appear to be one technique that is deemed superior to another in terms of neuroprotective efficacy, anesthesia maintenance efficacy, or postoperative recovery quality. With respect to neurofunctional outcome, so far, no clinical evidence exists to elucidate whether there is a difference or equivalence between these two general anesthetics. Since both sevoflurane and propofol are GABAergic general anesthetic that have similar mechanism, we hypothesize that these two general anesthesia methods equivalently affect early postoperative neural function in patients with supratentorial gliomas.
We chose sevoflurane in one arm of this study as it is one of the most frequently used inhalational agents in our medical center, and worldwide. Compared to isoflurane and desflurane, it has the least vasoactive (vasodilatory) effect and can preserve cerebral blood flow best up to 1--1.5 MAC, while maintaining cerebral autoregulation \[[@CR16]--[@CR19]\]. Propofol is the only available intravenous general anesthetic for maintenance in both TIVA and combined inhaled-intravenous anesthesia. It is pharmacologically difficult to use only a single anesthetic agent to maintain general anesthesia during surgery, as usually an intraoperative analgesic is necessary as well, and so we selected an ultra short-acting opioid (remifentanil), which has a very short elimination half-life, so as to avoid respiratory depression postoperatively, which is especially important for neurosurgical patients.
The 15-item National Institutes of Health Stroke Scale (NIHSS) was developed for and is widely used to evaluate neurologic function \[[@CR20]\]. Although the modified NIHSS (mNIHSS) appears to be a more easily-used measure that is identical with the original NIHSS in its validation and reliability when evaluating the severity of stroke \[[@CR21], [@CR22]\], the deletion of limb ataxia in the mNIHSS has significance in patients with intracranial tumors based on our previous study \[[@CR9]\], and there is no evidence showing the superiority of one to another in the intracranial tumor population. We therefore will choose to calculate both the NIHSS and the mNIHSS for comprehensive neurologic function evaluation. The Chinese version of NIHSS has been validated in a previous study \[[@CR23]\].
Motor evoked potential (MEP) monitoring is very important for some supratentorial craniotomies if the tumor is located in or near the motor strip or descending motor pathways near the prefrontal cortex. Clinical evidence shows that sevoflurane suppresses MEP amplitudes in a dose-dependent manner \[[@CR24], [@CR25]\]. When using a sevoflurane-propofol combination in clinical practice, sevoflurane concentration is usually maintained below 0.4 MAC; while in this study, sevoflurane will be the only general anesthetic to maintain the desired anesthetic depth in the sevoflurane arm, the concentration will be much higher and vary among individual patients, resulting in failure of MEP monitoring. Therefore, patients who require intraoperative neurophysiological monitoring will be excluded from the study.
The current study aims primarily to elucidate the possible differential effect of inhalational sevoflurane versus intravenous propofol anesthesia on the short period of post-operative neurologic function in patients receiving supratentorial tumor resection in a randomized controlled trial. As the repeated neurologic function assessments are important to evaluate the patients' recovery from surgical intervention and the presence of any complications, the results of this trial would help to interpret anesthetic residual effects on postoperative outcomes, and perhaps help to determine the intraoperative general anesthetic that affects neurologic function with the least detriment and therefore optimizes anesthetic management in this specific patient population.
Trial status {#Sec25}
------------
This registered study (ChiCTR-IOR-16009177) started recruiting on May 28th, 2018, and is planned to complete recruiting on December 31st, 2020. The protocol version number is 01 (September 1, 2016).
Appendix {#Sec26}
========
Table 3NIHSS and mNIHSS score systemItemNIHSSmNIHSSScore1aLOC\[deleted\]0 = Alert1 = Arousable by minor stimulation2 = Arousable by strong/repeated stimulation3 = Unresponsive1bLOC questions0 = Answers both correctly1 = Answers one correctly2 = Answers neither correctly1cLOC Commands0 = Performs both tasks correctly1 = Performs one task correctly2 = Performs neither task2Gaze0 = Normal1 = Parial gaze pulsy2 = Forced deviation3Visual0 = Normal1 = Partial hemianopia2 = Complete hemianopia3 = Bilateral hemianopia4Facial Palsy\[deleted\]0 = Normal1 = Minor2 = Parial3 = Complete5aMotor Arm (left)0 = No drift1 = Drift before 10s2 = Falls before 10s3 = No effort against gravity4 = No movementUN = Amputation or joint fusionexplain:5bMotor Arm (right)6aMotor Leg (left)6bMotor Leg (right)7Limb Ataxia\[deleted\]0 = Absent1 = In one limb2 = In two LimbUN = Amputation or joint fusion8SensoryIn NIHSS:0 = Normal1 = Mild-to-Moderate loss2 = Severe to total lossIn mNIHSS:0 = Normal1 = Abnormal9Language0 = Normal1 = Mild-to-Moderate aphasia2 = Severe aphasia3 = Mute or global aphasia10Dysarthria\[deleted\]0 = Normal1 = Mild-to-Moderate2 = SevereUN=Intubated11Extinction and Inattention0 = Normal1 = Visual, tactile, auditory, spatial, or personal inattention2 = Profound hemi-inattention or extinction to more than one modality*NIHSS* National Institute of Health stroke scale, *mNIHSS* modified National Institute of Health stroke scale, *LOC* Level of ConsciousnessThe total score in NIHSS and mNIHSS are 42 and 31 respectively
BP
: Blood pressure
BIS
: Bispectral index
ECG
: Electrocardiography
ETCO~2~
: End-tidal carbon dioxide
GABA
: Gamma-aminobutyric acid
HR
: Heart rate
LOC
: Level of consciousness
MAC
: Minimum alveolar concentration (the concentration of an inhalational anesthetic in the alveoli of the lungs that is needed to prevent movement in 50% of subjects in response to surgical stimulus.)
MAP
: Mean arterial pressure
MEP
: Motor evoked potential
MMSE
: Mini-mental state examination
MRI
: Magnetic resonance imaging
NIHSS
: National Institute of Health stroke scale
mNIHSS
: Modified National Institute of Health stroke scale
OAA/S
: Observer's assessment of alertness/sedation
PCIA
: Patient-controlled intravenous analgesia
SMA
: Supplementary motor area
SpO~2~
: Pulse oxygen saturation
VAS
: Visual analog scale
**Publisher's Note**
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Yan Xing and Nan Lin are co-first authors who equally contributed to this trial.
Not applicable.
NL contributed to the study design and concept, conduct, as well as being involved in the statistical analysis planning, writing the protocol draft, and revising it critically and approving the final version. YX contributed to the study design and conduct, data collection, and writing the protocol draft. YP was involved in the data collection coordination and contributing to the statistical plan. RH was involved in coordination and supervision of data collection and reviewing the protocol. JFB contributed to the protocol writing, reviewing, and revising. XL, YL, JD and MZ were involved in data collection and protocol writing. JL, MZ, and LN were involved in follow-up data collection. All authors have read and approved the manuscript as submitted.
\*Yan Xing and Nan Lin are co-first authors who equally contributed to this trial. Nan Lin and Ruquan Han are co-corresponding authors who are equally responsible for this protocol.
The trial design and data collection are funded by 'Beijing Municipal Administration of Hospitals 'Youth Program' (No.QML20160503), 83 Shuangqinglu Rd., Haidian District, Beijing, 100085, China; data analysis and interpretation are funded by the National Natural Science Foundation of China (Grant No.81701038), 83 Shuangqinglu Rd., Haidian District, Beijing, 100085, China; and Beijing Municipal Administration of Hospitals Clinical Medicine Development of Special Funding Support (ZYLX201708), 83 Shuangqinglu Rd., Haidian District, Beijing, 100085, China supports the data collection and manuscript writing.
The datasets used and analyzed during the current study are available from the co-corresponding author (Nan Lin, linnan127\@gmail.com) on reasonable request.
This study protocol was approved by China Ethics Committee of Registering Clinical Trials (ChiECRCT-20160051). All ethical aspects were considered. Written informed consent will be obtained from all participants.
Not applicable.
The authors declare that they have no competing interests.
| {
"pile_set_name": "PubMed Central"
} |
The Mediterranean region has unquestionably played a central role in human history, partly deriving from the navigable nature of the sea that connects southern Europe, western Asia, and North Africa. This unique setting has led to its being one of the most important and dynamic areas throughout prehistory.
The period of the Mesolithic to the Bronze Age (in western Eurasia) covers two major cultural shifts that are arguably among the most important transitions in human prehistory, heralding the change from hunter-gatherer subsistence to food production and later the emergence of metallurgy, changes that fundamentally transformed human culture. Recent large-scale studies of ancient human genomic variation (e.g., refs. [@r1][@r2][@r3][@r4][@r5][@r6][@r7]--[@r8]) have focused mainly on central and northern Europe and have revealed that changes during the Neolithic and later during the Bronze Age were driven by population movements into Europe from the southeast and east, first by early farmers from Anatolia and the Levant ([@r1][@r2]--[@r3], [@r9], [@r10]) and second by herders from the Pontic-Caspian steppe ([@r4], [@r6]). These migrations profoundly reshaped the genetic and cultural landscape of Europe. However, studying the genetic impacts of these cultural transitions in southernmost Europe, especially the Mediterranean, has usually focused on single time periods ([@r10][@r11][@r12][@r13][@r14][@r15]--[@r16]).
The full Neolithic package reached the Iberian Peninsula and northern (modern-day) Morocco *ca*. 7,500 Cal BP, with the Cardial pottery culture coming from the central Mediterranean ([@r17]). This was rapidly followed by a regional diversification of ceramics and lithics with the Cardial pottery type present in most of the Mediterranean fringe and the interior of the Iberian Peninsula represented by the Boquique type (e.g., El Portalón de Cueva Mayor) ([@r18]) potentially introduced through the north via the Pyrenees ([@r19]). In southern Iberia (Andalusia), however, the early Neolithic is characterized by a type of impressed non-Cardial ceramic decorated *a la Almagra* ([@r20]). This type of pottery culture reached central Andalusia by 7,300 Cal BP, soon replacing the Cardial pottery, and is found at the Murciélagos de Zuheros cave ([@r21]). It has been proposed that North Africa played a significant part in the origins of the Neolithic in southern Spain ([@r22]), although this has recently been challenged ([@r23]). The prehistory of the Iberian Peninsula as a whole is of particular interest, given its specific geographic location at the westernmost edge of the continent, naturally making it the furthest point from the documented prehistoric migrations originating from eastern Eurasia. This location holds potential for a complex demographic history with migrations from diverse sources, as it is connected with mainland Europe in the north, is surrounded by two potential maritime migration routes along the Mediterranean Sea and the Atlantic Ocean, and furthermore is in close geographic proximity to North Africa. Previous studies on early Iberian farmers have shown that these populations represent the descendants of migrants from Anatolia ([@r6], [@r13]) followed by admixture with local hunter-gatherers ([@r5], [@r6], [@r9][@r10]--[@r11], [@r13], [@r24]). Furthermore, modern-day southwestern Europeans are genetically closer to Early and Middle Neolithic Europeans than are modern-day central Europeans, who are more closely related to Late Neolithic and Bronze Age populations ([@r1], [@r2], [@r4], [@r6], [@r11], [@r25][@r26][@r27]--[@r28]), suggesting diverse and regionally distinctive demographic histories.
To investigate the demographic history of the westernmost edge of the prehistoric Eurasian migrations, we have sequenced the genomes of 13 individuals excavated from six prehistoric Iberian sites in the north and south of modern-day Spain ([*SI Appendix*, Section S1 and Table S2.1](http://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1717762115/-/DCSupplemental)). These sites cover the Neolithic, Late Neolithic/Copper Age (LNCA), and Bronze Age chronologies between 7,245 and 3,500 Cal BP ([*SI Appendix*, Fig. S3.2 and Table S3.2](http://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1717762115/-/DCSupplemental)), including the oldest sequenced genome in southern Iberia, from the Murciélagos de Zuheros cave. This individual is directly dated to 7,245--7,024 y Cal BP and represents the first genome of an individual from the Neolithic Almagra Pottery Culture, the early agriculturalists of the south of the Iberian Peninsula. For the El Portalón cave, we generated additional DNA sequence data for published individuals ([@r11]) as well as sequencing five additional individuals, enabling the genomic analysis of a population that spans a temporal sequence comprising the Neolithic, Copper Age, and Bronze Age periods (directly dated to between 7,165 and 3,500 y Cal BP). These genomic data, combined with published data ([@r6], [@r9], [@r11], [@r13], [@r24]), allow us to comprehensively study demographic changes through time in the Iberian Peninsula in general and in one single location in particular. We contrast these developments with contemporary populations in other parts of Europe where two major population turnovers took place during this time span ([@r1], [@r4], [@r6], [@r25]). Our genomic analysis was combined with stable isotope analysis to investigate the role of aquatic resources in the diets of the individuals tested here. Previous genomic studies have revealed increasing amounts of genetic hunter-gatherer admixture in the farmer population after the initial arrival of Neolithic people in Europe ([@r1][@r2]--[@r3], [@r11], [@r13]), implying the continued survival of hunter-gatherer populations or at least their lineages. In contrast, paleodietary stable isotope studies of the Iberian late Neolithic--Early Bronze Age have indicated that aquatic resources are not abundant in the diet, despite their likely availability at many sites ([@r29][@r30][@r31][@r32][@r33][@r34]--[@r35]). Detecting aquatic resources can be difficult using only bulk isotope analysis, and it has been posited that freshwater and terrestrial C3 diets can be more easily distinguished using stable carbon isotope values of amino acids ([@r36], [@r37]). Ten of the sequenced individuals were investigated for amino acid stable carbon isotopes of bone collagen to reconstruct paleodietary preferences ([@r38], [@r39]), in particular the presence of aquatic food intake ([*SI Appendix*, Section S3](http://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1717762115/-/DCSupplemental)) ([@r36], [@r40]).
Results and Discussion {#s1}
======================
We sequenced the genomes of 10 individuals from northern and southern Spain either contextually or directly radiocarbon dated to the Neolithic, LNCA, and Bronze Age ([Fig. 1*A*](#fig01){ref-type="fig"} and [*SI Appendix*, Table S4.1](http://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1717762115/-/DCSupplemental)), and we increased the sequencing depth of three individuals from a previous study (ATP16, ATP2, and ATP12) ([@r11]) using additional bone material ([*SI Appendix*, Table S2.1](http://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1717762115/-/DCSupplemental)). Altogether, our 13 sequenced genomes range from 0.01× to 12.9× ([*SI Appendix*, Table S4.1](http://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1717762115/-/DCSupplemental)) with six individuals having \>2.0× genome coverage. Our sequence data show postmortem damage and fragmentation, as expected from endogenous ancient DNA (aDNA) molecules ([@r41]). Eleven individuals were genetically inferred to be males and two to be females. We obtained contamination estimates based on the X chromosome in male samples ([@r42]), which were all ≤5% or lower ([*SI Appendix*, Table S4.1](http://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1717762115/-/DCSupplemental)). Mitochondrial contamination estimates based on reads mapping to the mitochondrial genome ([@r43]) suggest \>5% mitochondrial contamination for two samples (POR003 and ATP019) ([*SI Appendix*, Table S4.1](http://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1717762115/-/DCSupplemental)), and the sequence data from these individuals were subsequently filtered to retain fragments displaying postmortem damages to conservatively restrict the analysis to these sequences ([@r44]).
) are labeled. (*B*) Enlarged section of the PCA plot ([Dataset S2](http://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1717762115/-/DCSupplemental)) showing the part of the PC1--PC2 space occupied by the ancient Iberians as well as other ancient groups. BA, Bronze Age; BB, Bell Beaker; CA, Copper Age; CWC, Corded Ware Culture; EN, Early Neolithic; HG, hunter-gatherer; LBK_EN, Linearbandkeramik_Early Neolithic; LNCA, Late Neolithic/Copper Age; MN, Middle Neolithic.](pnas.1717762115fig01){#fig01}
Nine of the 13 ancient Iberian individuals were found to carry mitochondrial haplogroups associated with European early farmers, namely K, J, N, and X ([*SI Appendix*, Table S4.1](http://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1717762115/-/DCSupplemental)), distributed throughout the Early Neolithic to the Bronze Age ([@r6], [@r45]). Two individuals have haplogroups HV0 and H, known in both European early farmers and hunter-gatherers ([@r25], [@r45]) and are present during the LNCA. Further, haplogroup U5, characteristic of hunter-gatherers ([@r46], [@r47]), is found in a Late Copper Age individual. Consistent with the mitochondrial haplogroup composition of the ancient Iberians, the Y chromosome composition ([Dataset S1](http://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1717762115/-/DCSupplemental)) displays a mix of haplogroups associated with both European farmers and hunter-gatherers. Among the Early Neolithic individuals, we find the European farmer-associated haplogroup G2a2 ([@r9]) and the European farmer-associated haplogroup H2 ([@r1], [@r6]), while in the LNCA we observe haplogroup I2, previously found in both hunter-gatherers and farmers ([*SI Appendix*, Table S4.1](http://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1717762115/-/DCSupplemental)) ([@r1], [@r6]). Both Bronze Age males carried haplogroup R1b-M269 ([*SI Appendix*, Table S4.1](http://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1717762115/-/DCSupplemental)), which is frequent among Late Neolithic and Bronze Age samples from other parts of Europe ([@r4], [@r6]). This uniparental marker composition is in agreement with the well-known admixture between resident hunter-gatherers and incoming farmers.
To obtain an overview of the genetic variation within prehistoric Iberia, we performed principal component analysis (PCA) on a reference panel of 26 modern-day populations from western Eurasia ([@r1]) on which we projected our 13 ancient individuals, together with relevant Mesolithic (*n* = 17), Neolithic/Chalcolithic (*n* = 98), and Late Neolithic/Bronze Age (*n* = 78) genomes from Europe and Anatolia ([Fig. 1*B*](#fig01){ref-type="fig"} and [Dataset S2](http://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1717762115/-/DCSupplemental)). This PCA replicates previous findings: (*i*) a clear genetic distinction between early farmers and resident hunter-gatherers, (*ii*) affinity of the former with the southwestern modern-day European variation, and (*iii*) an increased affinity of prehistoric farmers to western hunter-gatherers over time due to increased admixture between the two populations ([@r1][@r2]--[@r3], [@r11], [@r25], [@r26]). Despite the geographic proximity of southern Iberia to northern Africa, we do not see substantial affinities of any individual to modern-day African populations, but the lack of ancient North African genomes limits our abilities to test these connections ([*SI Appendix*, Fig. S5.1](http://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1717762115/-/DCSupplemental)).
Zooming into the genetic variation within prehistoric Iberia, we do not find a geographic stratification pattern between North and South Iberian populations; instead, we observe stratification directly associated with chronology. Three clusters are identified among the Iberian farmers on the PCA. The first cluster comprises the early Neolithic Iberians ([Fig. 1*B*](#fig01){ref-type="fig"}, yellow triangles) that falls within the modern-day Sardinian genomic variation showing the highest affinity to Sardinians among all early European farmers. The second group includes the Middle Neolithic ([Fig. 1*B*](#fig01){ref-type="fig"}, blue triangles and squares) and LNCA ([Fig. 1*B*](#fig01){ref-type="fig"}, red triangles and squares) populations falling within the modern-day southern European variation but differentiating from the early Neolithic Iberians, a pattern which can be explained by the subsequent admixture with local hunter-gatherers ([@r6], [@r11], [@r24]). These Iberian farmers were likely the group that later developed Bell Beaker pottery (widely found in Western Europe during the third millennium BC), which then spread without major migrations toward central and northwestern Europe ([@r48]). The third cluster encompasses the Bronze Age individuals ([Fig. 1*B*](#fig01){ref-type="fig"}, turquoise triangles and squares) and is differentiated from the other two groups, showing the highest affinity to modern-day Iberians. There is a strong genetic difference among Iberian populations dating to the Mesolithic, early and middle Neolithic, LNCA, and Bronze Age; this genetic pattern is likely the result of numerous migrations and admixture events over time.
A striking feature of the PCA ([Fig. 1*B*](#fig01){ref-type="fig"}) is the genetic difference between Iberian and central European farmers (orange, purple, gray, and black diamonds in [Fig. 1*B*](#fig01){ref-type="fig"}). This division could represent slightly different gene pools of the migrating populations along the two different routes for early Neolithic farmers: one following the Danube river into central Europe and a second along the Mediterranean coast, which has been suggested based on the analysis of smaller datasets ([@r5], [@r12], [@r24]). To formally test this separation, we calculated f~4~ statistics to investigate if Iberian farmers form a clade to the exclusion of central European farmers. The statistic of the form f~4~(Chimp, X; Central_LBK_EN and Iberia_EN) measures whether an individual X shares more genetic drift with early Neolithic central Europeans (if the value is negative) or early Neolithic Iberians (if the value is positive). For this analysis, we only used SNP-captured individuals for the reference populations (Central LBK_EN and Iberia_EN) to avoid spurious affinities between references and X due to technological artifacts ([@r49], [@r50]). In contrast to prehistoric populations from central Europe, this statistic is consistently shifted toward positive values for prehistoric Iberian groups from different chronologies ([Fig. 2*A*](#fig02){ref-type="fig"}) and is qualitatively similar when using other individuals as reference ([Dataset S4](http://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1717762115/-/DCSupplemental)). This pattern does not seem to be driven by hunter-gatherer--related admixture into the farming populations ([Dataset S5](http://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1717762115/-/DCSupplemental)). This observation suggests that all Neolithic Iberians trace most of their ancestry to the first Neolithic migrants arriving in the peninsula and that later contributions from contemporary central Europeans were only minor. The overall pattern is consistent with two independent Neolithic migrations of genetically slightly different populations that spread farming practices across Europe. The Mediterranean route migrants show a strong connection with modern-day population isolates in southwestern Europe. Modern-day Sardinians have been suggested to be relatively direct descendants of the early Neolithic individuals ([@r27], [@r51]), and modern-day Basques also trace a high proportion of their ancestry to the first Mediterranean farmers with only minor additional admixture since the Neolithic ([@r11]).
 and [S5](http://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1717762115/-/DCSupplemental)). (*B*) Estimates of ancestry proportions in different prehistoric Europeans as well as modern southwestern Europeans. Individuals from regions of Iberia were grouped together for the analysis in *A* and *B* to increase sample sizes per group and reduce noise.](pnas.1717762115fig02){#fig02}
A large migration of Pontic-Caspian steppe herders (the Yamnaya culture) during the Late Neolithic/Early Bronze Age has been found to have a substantial impact on the gene pool of central and northern European populations ([@r4], [@r6][@r7]--[@r8]), but the impact of this migration on contemporary southern and western Europeans has been unclear. To estimate the genetic contributions of different prehistoric groups (hunter-gatherers, early Anatolian farmers, and Steppe herders) to other ancient populations, we inferred admixture fractions using both unsupervised (ADMIXTURE, ref. [@r52]) ([Dataset S3](http://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1717762115/-/DCSupplemental)) and supervised approaches (qpAdm and ADMIXTURE, refs. [@r6] and [@r52]). Neolithic European populations share different proportions of hunter-gatherer and Neolithic farmer genetic material, with a tendency toward more hunter-gatherer--related ancestry in later groups ([Fig. 2*B*](#fig02){ref-type="fig"}) ([@r11]). Late Neolithic and Bronze Age north-central Europeans display substantial fractions of Pontic Steppe ancestry (up to 71% estimated with qpAdm; 93% based on supervised ADMIXTURE) ([*SI Appendix*, Fig. S5.3](http://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1717762115/-/DCSupplemental)) at the onset of the Bronze Age. However, steppe ancestry in Bronze Age individuals from Iberia (13%; 18%) and Hungary (21%; 38%) is lower than in their north-central European counterparts ([Fig. 2*B*](#fig02){ref-type="fig"}), a pattern previously suggested but not directly quantified ([@r24]). The estimates for Bronze Age Iberians are close to the 15% steppe ancestry estimated for the modern Spanish population ([Fig. 2*B*](#fig02){ref-type="fig"}). Consistently, Bronze Age populations from Greece and Anatolia also show a limited increase in steppe ancestry compared with their Neolithic ancestors ([@r15]). This reduced impact of Steppe herders on these populations could reflect a decrease in the number of migrants or a dilution of Steppe ancestry during this process. In contrast to the events in north-central Europe, the arrival of most of the Yamnaya-related ancestry in Iberia postdates the onset of Bell Beaker pottery in Iberia, suggesting that the Bell Beaker culture spread culturally ([@r48]), while steppe ancestry was brought into Iberia through later migrations. Notably, both male Bronze Age Iberian individuals in this study as well as all three Iberian Bronze Age males in ref. [@r24] carried R1b-M269 Y chromosomes ([*SI Appendix*, Table S4.1](http://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1717762115/-/DCSupplemental)) also found with high frequency in individuals associated with the Yamnaya culture, the source population of steppe ancestry ([@r4], [@r6]), indicating a continuing male-driven migration from central Europe into southwestern Europe ([@r8], [@r24], [@r53]).
To obtain additional insights into the demographic development of prehistoric populations---their diversity as well as their effective population sizes---we estimated genetic diversity through time in all shotgun-sequenced Iberian, Anatolian, and central European farmers as well as in Mesolithic hunter-gatherer populations ([Fig. 3](#fig03){ref-type="fig"}). As previously observed ([@r2]), Mesolithic hunter-gatherers had the lowest diversity of the groups tested. Interestingly, the group with lowest genetic diversity among farmers was early Neolithic Iberians. There is a significant increase in diversity from the early Neolithic to the middle Neolithic in Iberia, a pattern not observed in central Europe. The low diversity levels in early Neolithic Iberian farmers could potentially reflect a bottleneck resulting from the initial migration along the Mediterranean coast. The subsequent increase in genetic diversity over time represents a recovery from this bottleneck and is likely due to an increase in population size and admixture with local hunter-gatherers.
![Conditional nucleotide diversity ([@r2]) for different prehistoric populations. Each population is represented by the two shotgun-sequenced individuals with the highest sequencing coverage. Error bars show two SEs estimated using a block jackknife. The estimates for early Neolithic Hungarians are based on the shotgun data published in ref. [@r25].](pnas.1717762115fig03){#fig03}
Paleodietary analysis ([*SI Appendix*, Section S3](http://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1717762115/-/DCSupplemental)) of these same individuals shows that they had a remarkable degree of homogeneity in their diets, temporally and geographically. The analyses indicate that the individuals maintained a C3 terrestrial diet and that neither freshwater nor marine sources of protein significantly contributed to the diet, in agreement with previous studies (e.g., refs. [@r29] and [@r30]). Even though the Early Neolithic individuals from northern and southern Iberia were found \>600 km apart, there is a clear similarity in dietary preferences for terrestrial foods. In addition, although there is evidence of contact between groups that have different subsistence strategies (an increasing hunter-gatherer genetic component through time), dietary preferences remain constant from the Early Neolithic to the beginning of the Bronze Age.
Conclusions {#s2}
===========
We present a comprehensive biomolecular dataset spanning four millennia of prehistory across the whole Iberian Peninsula. Our results highlight the power of archaeogenomic studies focusing on specific regions and covering a temporal transect. The 4,000 y of prehistory in Iberia were shaped by major chronological changes but with little geographic substructure within the Peninsula. The subtle but clear genetic differences between early Neolithic Iberian farmers and early Neolithic central European farmers point toward two independent migrations, potentially originating from two slightly different source populations. These populations followed different routes, one along the Mediterranean coast, giving rise to early Neolithic Iberian farmers, and one via mainland Europe forming early Neolithic central European farmers. This directly links all Neolithic Iberians with the first migrants that arrived with the initial Mediterranean Neolithic wave of expansion. These Iberians mixed with local hunter-gatherers (but maintained farming/pastoral subsistence strategies, i.e., diet), leading to a recovery from the loss of genetic diversity emerging from the initial migration founder bottleneck. Only after the spread of Bell Beaker pottery did steppe-related ancestry arrive in Iberia, where it had smaller contributions to the population compared with the impact that it had in central Europe. This implies that the two prehistoric migrations causing major population turnovers in central Europe had differential effects at the southwestern edge of their distribution: The Neolithic migrations caused substantial changes in the Iberian gene pool (the introduction of agriculture by farmers) ([@r6], [@r9], [@r11], [@r13], [@r24]), whereas the impact of Bronze Age migrations (Yamnaya) was significantly smaller in Iberia than in north-central Europe ([@r24]). The post-Neolithic prehistory of Iberia is generally characterized by interactions between residents rather than by migrations from other parts of Europe, resulting in relative genetic continuity, while most other regions were subject to major genetic turnovers after the Neolithic ([@r4], [@r6], [@r7], [@r9], [@r25], [@r48]). Although Iberian populations represent the furthest wave of Neolithic expansion in the westernmost Mediterranean, the subsequent populations maintain a surprisingly high genetic legacy of the original pioneer farming migrants from the east compared with their central European counterparts. This counterintuitive result emphasizes the importance of in-depth diachronic studies in all parts of the continent.
Materials and Methods {#s3}
=====================
Archaeological Samples. {#s4}
-----------------------
Thirteen individuals from northern Spain (El Portalón, San Quílez, and Cueva de los Lagos) and Andalusia in the south (Murciélagos de Zuheros, Cueva de los Cuarenta, and El Pirulejo) were sampled for aDNA analyses. All six sites cover a chronological period from the Early Neolithic to the Bronze Age. Eleven samples were directly radiocarbon dated using accelerator mass spectrometry, and the remaining two were associated with archaeological contexts. See [*SI Appendix*, Section S1](http://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1717762115/-/DCSupplemental).
Sequencing. {#s5}
-----------
DNA was extracted from bones and teeth ([@r54]); DNA extracts were converted into blunt-end Illumina libraries ([@r55]). All samples were prepared in dedicated aDNA facilities at the Evolutionary Biology Center in Uppsala, Sweden. The libraries were sequenced on Illumina HiSeq platforms 2500 or X Ten at the SNP&SEQ Technology Platform at the Science for Life Laboratory Sequencing Centre in Uppsala. All 13 samples were screened for human DNA and yielded over 1% human DNA content; thus, all were included for downstream analysis. See [*SI Appendix*, Section S2](http://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1717762115/-/DCSupplemental).
Stable Isotopes. {#s6}
----------------
A subsample of the individuals studied for DNA underwent amino acid stable carbon isotope analysis of bone collagen to determine the long-term dietary preferences of these individuals. Bone collagen was isolated from bone samples following a modified Longin method ([@r56], [@r57]), and amino acids were prepared via hydrolysis of collagen. The δ^13^C values of the amino acid fractions were measured using a Thermo Fisher LC-isotope ratio MS (LC-IRMS) system following methods similar to those described in ref. [@r38]. These data were supplemented by bulk δ^13^C and δ^15^N values provided with the radiocarbon dates of the bone collagen. See [*SI Appendix*, Section S3](http://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1717762115/-/DCSupplemental).
Next-Generation Sequencing Data Processing and Authentication. {#s7}
--------------------------------------------------------------
Overlapping paired-end reads were merged, the remaining adapters were trimmed ([@r58]), and the fragments were mapped to the human reference genome using bwa ([@r59]). Fragments with identical start and end positions were considered PCR duplicates, and all duplicates were collapsed into consensus sequences. Contamination was estimated based on heterozygous sites on the X chromosome in males ([@r42]) and in the mitochondrial genome ([@r43]). All samples show indications of characteristic aDNA damage, and for samples with high levels of mitochondrial contamination (\>15%), the analysis was restricted to fragments indicating postmortem damage ([@r44]). See [*SI Appendix*, Section S4](http://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1717762115/-/DCSupplemental).
Uniparental Haplogroups. {#s8}
------------------------
HAPLOFIND ([@r60]) was used to infer the most likely haplogroup for mitochondrial consensus sequences. Y chromosomal haplogroups were assigned by investigating up to 732 haplotype-informative single base substitutions obtained from the Phylotree version of March 9, 2016 ([@r61]). Sites presenting more than one allele were not taken into account for the classification. Derived sites for all samples are shown in [Dataset S1](http://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1717762115/-/DCSupplemental). See [*SI Appendix*, Section S4](http://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1717762115/-/DCSupplemental).
Reference Datasets. {#s9}
-------------------
Newly sequenced individuals were analyzed with a large set of published prehistoric European genomes as well as 203 modern populations from the Human Origins panel ([@r1]). At each SNP position, a single read (minimum mapping and base quality of 30) was randomly drawn to represent the ancient individual. Transitions were coded as missing data to exclude potential postmortem damage. See [*SI Appendix*, Section S5](http://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1717762115/-/DCSupplemental).
Population Genetic Analysis. {#s10}
----------------------------
For each ancient individual, a PCA was conducted together with modern Europeans from the Human Origins panel using smartpca ([@r62]). All ancient individuals were projected on one plot using Procrustes analysis. We used popstats ([@r63]) to calculate *D* and *f* statistics ([@r64]) to estimate shared drift between populations and to test tree topologies. To estimate genetic diversity ([@r2]) within prehistoric groups, we calculated conditional nucleotide diversity for shotgun sequence data and at transversion polymorphisms ascertained in Yorubans. See [*SI Appendix*, Section S5](http://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1717762115/-/DCSupplemental).
The model-based clustering approach implemented in ADMIXTURE ([@r52]) was run with all modern individuals of the Human Origins dataset and all ancient individuals. Common modes among the different runs were identified, and clusters were aligned across different values of K using pong ([@r65]). The full unsupervised ADMIXTURE results are shown in [Dataset S3](http://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1717762115/-/DCSupplemental). We used supervised ADMIXTURE and qpAdm ([@r6]) to estimate the ancestry proportions attributable to early Anatolian farmers, hunter-gatherers, and steppe herders in Neolithic and Bronze Age Europeans ([*SI Appendix*, Section S5](http://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1717762115/-/DCSupplemental)).
Supplementary Material
======================
We thank everyone involved in the excavations for their contributions, Rafael Carmona at Museo de Priego de Córdoba for facilitating access to material, Arielle R. Munters and Robin Olsson for initial processing and curating of next-generation sequencing data, and Elena Santos for editorial assistance with text and figures. This project was supported by grants from the Transforming Human Societies Research Focus Area and Bridging Fellowship from La Trobe University (C.V.), the Wenner-Gren Foundation (T.G.), the Knut and Alice Wallenberg Foundation (M.J. and A.G.), the Swedish Research Council (M.J. and A.G.), the European Research Council (M.J.), Australian Research Council Grant FT0992258 (to C.I.S.), the Junta de Andalucía Research Project of Excellence Grant HUM-1510 (to R.M.M.-S.), and Spanish Ministerio de Economía y Competitividad Grant CGL2015-65387-C3-2-P (MINECO/FEDER) (to C.V., I.U., R.R.-V., E.I., L.R., J.-M.B.d.C., E.C., J.-M.C., and J.L.A.). Sequencing was performed at the National Genomics Infrastructure, Uppsala, and all computations were conducted through the Uppsala Multidisciplinary Centre for Advanced Computational Science under projects b2013203 and b2013240.
The authors declare no conflict of interest.
Data deposition: The sequences reported in this paper have been deposited in the European Nucleotide Archive (accession no. [PRJEB23467](PRJEB23467)).
This article contains supporting information online at [www.pnas.org/lookup/suppl/doi:10.1073/pnas.1717762115/-/DCSupplemental](http://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1717762115/-/DCSupplemental).
[^1]: Contributed by Juan Luis Arsuaga, February 1, 2018 (sent for review October 18, 2017; reviewed by Richard Edward Green and Emilia Huerta-Sanchez)
[^2]: Author contributions: C.V., T.G., C.I.S., and M.J. designed research; C.V., T.G., J.C.V.-R., I.U., E.I., R.R.-V., L.G.S., R.M.M.-S., E.M.S., H.M., L.R., A.A., J.A.H.V., J.-M.C., J.L.A., and C.I.S. performed research; J.-M.B.d.C., E.C., and A.G. contributed new reagents/analytic tools; C.V., T.G., and C.I.S. analyzed data; and C.V., T.G., J.L.A., C.I.S., and M.J. wrote the paper.
[^3]: Reviewers: R.E.G., University of California, Santa Cruz; and E.H.-S., University of California, Merced.
[^4]: ^1^C.V. and T.G. contributed equally to this work.
| {
"pile_set_name": "PubMed Central"
} |
Introduction
============
Over the course of a chronic illness, patients face challenges on many fronts. On a basic level, they endeavor to understand what is happening to them and deal with their illness. This may include navigating the health care system and understanding their medication regimen. They interact with information that may change their abilities to engage in these behaviors and make health decisions. While existing literature has investigated how people seek health-related information, there is a need for additional research on how information facilitates changes in patients' understanding of their health, which may in turn lead to long-term changes in health management.
This study investigated the relationship between information and health management of those with a chronic condition---fibromyalgia. Fibromyalgia is a complex, poorly understood condition characterized by chronic widespread pain, joint stiffness, and systemic symptoms (eg, mood disorders, fatigue, cognitive dysfunction, and insomnia) \[[@ref1]-[@ref3]\]. Due to the diversity of symptoms and problems that patients experience, fibromyalgia has an impact on multiple facets of patients' lives \[[@ref4]-[@ref6]\].
Because fibromyalgia patients often appear healthy and their symptoms are invisible, patients continually struggle with stigma, social isolation, and a search for legitimacy \[[@ref7]-[@ref10]\]. In addition, patients struggle with the medically unexplained nature of the syndrome \[[@ref6],[@ref7]\]. In the case of many illnesses, diagnosis can serve to give meaning to an illness experience, but with fibromyalgia, initial relief is replaced with the realization that the diagnosis does not lead to increased understanding, treatment options, or respect from others \[[@ref11]-[@ref13]\].
This is where information might potentially play a role. Though patients with fibromyalgia have shown little long-term improvement \[[@ref14]\], previous research has shown that becoming knowledgeable about one's condition is an important factor in acceptance or coming to terms with pain \[[@ref15]\], and pain acceptance is associated with less pain, disability, symptoms, mood disturbance, as well as better general health, functioning, and greater well-being (eg, \[[@ref16],[@ref17]\]). Because fibromyalgia is a condition for which there are limited treatment options, self-management is increasingly being recommended \[[@ref18]\].
Fibromyalgia patients consult many sources to try to understand their condition and possible treatments, including health care professionals, the Internet, health organizations, magazines, television, radio, support groups, and other people \[[@ref19],[@ref20]\]. Fibromyalgia patients and other patients with chronic conditions may use online resources such as online discussion forums and blogs to exchange information, understand their illness, and offer social support \[[@ref21]-[@ref23]\]. Online participation may lead to benefits such as reduction of social isolation \[[@ref22]\], patient empowerment \[[@ref24]\], and improved psychological, social, and cognitive health \[[@ref25]\].
Previous research has also reported that fibromyalgia patients' information needs change over the course of the illness \[[@ref26]\]. At first, individuals may be preoccupied with finding a cure. Searches for information on exercise, medications, and research increase over time. However, it is unclear what drives this evolution in information behavior, and moreover, what information behaviors may lead to successful self-management. The motivation for the current study was to provide insight concerning this gap. This paper explores three aspects of fibromyalgia patients' illness journeys: (1) health management, (2) information seeking, and (3) information consumption and use processes.
Methods
=======
Sample and Recruitment
----------------------
Multiple mechanisms were used to recruit a convenience sample that self-identified as having fibromyalgia (*N*=23). A recruitment goal was established to recruit a sample that was diverse in terms of three characteristics: age (˂47 years and ≥47 years), illness duration (≤4 years and ˃4 years), and social media participation style (non-user/lurker, infrequent participator, active participator), with substantive representation in each of the subcategories per category. A lurker was defined as someone who read social media content but did not author content themselves, an infrequent participator was someone who authored social media, but infrequently, and a frequent participator was someone who authored social media content quite often. These definitions are based on those in previous studies, with modifications to account for participation on other types of social media \[[@ref27],[@ref28]\].
The decision to focus on these dimensions was based on previous work that showed there was great variation in the age and illness duration of fibromyalgia patients and that social media participation style was significantly associated with other aspects of illness adjustment \[[@ref29]\]. The age threshold was based on the mean age in prior studies \[[@ref19],[@ref26]\], and the illness duration threshold was set in order to emphasize the first several years after onset.
The recruitment mechanisms included an email contact list from a previous survey \[[@ref26]\], a university staff and student listserv, face-to-face support groups, health-related discussion forums, and Twitter ([Table 1](#table1){ref-type="table"}). In the case of face-to-face support groups, the leaders of support groups for fibromyalgia, chronic pain, and chronic fatigue syndrome were contacted, and permission was sought to visit the support groups to introduce the study and invite members to participate. The health-related discussion forums included websites such as Reddit, HealingWell, and ProHealth, which feature forums dedicated to fibromyalgia and other conditions that are often co-morbid with fibromyalgia, such as chronic fatigue syndrome. In each case, a description of the study and an invitation to participate was posted in relevant forums. In the case of Twitter, users who self-identified as having fibromyalgia were contacted and invited to participate.
######
Recruitment mechanisms and participants recruited.
Recruitment mechanism Participants, n
--------------------------------------- -----------------
Participant pool from previous survey 4
Listserv^a^ 7
Social networking sites 5
Face-to-face support groups 6
Twitter 1
^a^Includes those referred by someone on the listserv.
Interviews
----------
The first interview focused on participants' health history and information seeking and use. Participants were also asked to draw a timeline representing their illness journey ([Figure 1](#figure1){ref-type="fig"}). Timelines have been used in previous health-related research (eg, \[[@ref30]-[@ref32]\]). When the exercise was introduced, participants were asked to think about their illness journey and "to draw something that represent\[ed\] it." They were told there were no rules as to what they drew and that the timeline need not be a line. The aim of the prompt was to leave the activity as open as possible, so that participants would feel free to depict the journey as they experienced it. The purpose of this activity was to help participants access their memories of their illness history.
The second interview was used to explore participants' social media participation histories, using an interface called the Online Scrapbook that was designed for the study. This interface enabled participants to view their social media participation over time. There were multiple reasons to incorporate the interface, including reminding participants of what they had previously authored, as well as providing them a fresh look at it through an interactive visualization. As this paper focuses on thematic analysis of the interview content, the interface will not be discussed in further depth. The interview guides for the two interviews have been included in the [Multimedia Appendices 1](#app1){ref-type="supplementary-material"} and [2](#app2){ref-type="supplementary-material"}.
Participants were interviewed either once or twice, depending on the extent to which they participated in social media and their geographic proximity. If participants participated only minimally in social media or lived far away, they were usually interviewed once, and the questions from the second interview were integrated into the first interview. Three interviews were conducted via Skype or phone due to issues of geographic proximity. All other interviews were conducted in person. To ensure that participants were comfortable during the interview, the location for the interview was left up to the participant, and almost all interviews occurred either in participants' homes or in coffee shops. Altogether, the study involved 37 interviews with 23 participants, and the mean total interview time per participant was 2 hours and 26 minutes.
{#figure1}
Data Analysis
-------------
The analysis method was derived from two approaches: interpretative phenomenological analysis and constructivist grounded theory. The primary aim of interpretative phenomenological analysis is to explore how participants make sense of their world and focuses on participants' interpretations of the object or event \[[@ref33],[@ref34]\]. Grounded theory focuses on how social and psychological processes occur in a given environment or situation \[[@ref35],[@ref36]\]. These two foci facilitated a study that investigated lived experience as well as social interactions and context.
The interview transcripts and a purposively sampled subset of posts authored by participants who engaged in online discussion forums such as Reddit served as the basis for the analysis. Because some participants were extremely prolific in their social media content production and it was not possible to manually analyze all of the posts, it was necessary to select a subset of posts that provided a sense of the diversity of each participant's social media production. The posts that participants authored were analyzed in the context of the threads, or dialogues, in which the posts were embedded.
The content was analyzed using Atlas.ti Version 1.0.1. In order to protect the identities of the participants, each participant was assigned an identification number. There were four pilot participants; thus, the 23 participants in the study will be referred to as P05-P27.
The analytic procedure involved initial line-by-line coding, followed by conceptualization of codes as a nested hierarchy, as is customary in grounded theory \[[@ref37]\]. In interpretative phenomenological analysis, a similar process is performed in which the codes are thought of as belonging to themes and subthemes \[[@ref33]\]. The themes and subthemes relating to health management and information behaviors are reported in this paper.
Exact prevalence of themes has not been reported in this study for two reasons. First, due to the sample size and recruitment strategy, the sample is not necessarily representative of fibromyalgia patients as a whole. Second, though prevalence of a theme may be an indicator of its significance, simply because a behavior occurs often does not immediately render it important \[[@ref38]\]. However, it is understandable that a reader would like to acquire a sense of how common a behavior is from the reading, and thus, consistent conventions of "a few," "some," "many," "almost all," and "all" have been used, as in previous research \[[@ref38]\].
The study protocol was approved by the Institutional Review Board at the University of North Carolina, Chapel Hill. All participants gave written informed consent for their data to be used in publications.
Results
=======
Participants
------------
The sample included 23 individuals who self-reported that they had fibromyalgia. The majority of the sample was white women ([Table 2](#table2){ref-type="table"}). The participants resided in nine different states; Washington, DC; and Australia. Because recruitment occurred using multiple mechanisms including several social media channels, the sample was naturally geographically diverse. The use of multiple recruitment methods also led to a sample that varied in terms of age (range 21-79 years), illness duration (1-58 years), and social media participation style. Though diversity was achieved in all three target categories, those with short illness durations were underrepresented.
The sample was highly educated, with the majority holding at least a 2-year or 4-year degree, and approximately half holding graduate degrees. Potential reasons for this bias were that a university listserv was used for recruitment and that those who had graduate degrees might have had a greater appreciation for the contribution of research to health care and thus volunteered for the study.
Health Management Features of the Illness Journey
-------------------------------------------------
This section focuses on the health management aspects of the journey in three parts: (1) moving from onset to diagnosis, (2) acceptance, and (3) development of an acceptable level of self-management.
### Moving From Onset to Diagnosis
Though for some participants, fibromyalgia onset coincided with an event such as a surgery or immunization, a more common pattern was experiencing symptoms for some time, before recognizing that the symptoms were not to be ignored. Most participants were diagnosed years after onset. There were various reasons. In many cases, participants did not seek medical assistance right away. P09 experienced fatigue for years, but she did not seek help because she thought she was just being lazy. It was not until she started experiencing pain that she sought the opinion of a physician. P27 had a fast-paced lifestyle full of events that she was committed to, so she ignored her symptoms until "\[her\] body was forced to stop." In retrospect, she said:
> Thinking back on how long I've been feeling a little bit tired or feeling a little bit achy and thinking of all the ways that I made excuses for that, I realized that probably my symptoms have been going on a lot longer than I thought...But I just said, well this is what it's like to be an activist...you're just always tired because you're always doing stuff for the movement or for the community, so just push through it.
>
> P27
There were other reasons why diagnosis took a long time. Physicians tended to diagnose the condition after excluding other possibilities, so many participants experienced a period of uncertainty in which they saw multiple health care practitioners and underwent many lab tests before being diagnosed. P05 remarked somewhat facetiously, "It was really crazy...a lot of ER visits, um, got to know doctors very well---all kinds of specialists...and I sit there, and I was like, I should have been a doctor. Because at this point, I have done almost every test you can think of."
Participants' responses to diagnosis varied, but many did not want to be diagnosed with fibromyalgia due to their impressions of the condition. P15 said of the moment when her physician made her preliminary diagnosis:
> I remember thinking, "No!" That's one of the things I never wanted to have, because...it's like, Chronic Fatigue Syndrome or back problems...you can't really see it, and nobody believes it's real. And it's one of those things that makes people out all the time from work, and...people think you're faking it...I thought, "Oh God! Of all things!"
>
> P15
A few participants mentioned that their mother or other relative had fibromyalgia or chronic pain, and the possibility of a genetic basis to fibromyalgia has been hypothesized in previous literature \[[@ref39]\]. P20 remarked that she had not really believed her mother: "I thought that something had just happened to my mom, that she was making it up. Because this was in the '70s, when she was diagnosed with fibromyalgia. And I'm like, what kind of made up stuff is that?" So when P20 was diagnosed, she thought: "Please don't give me that. Anything but that."
Being diagnosed with fibromyalgia could be bittersweet. At 24, P11 thought that she might have had fibromyalgia and went through the next several years having lab tests and seeing different doctors. In a Reddit post, she wrote of her diagnosis at 30: "For me, it was relief, to finally have an answer (an answer I thought was RIGHT, as I'd thought it was fibro for a while but had never brought it up) but also a bit sad as I'm stuck with this for the rest of my life!" P24 also went through a period of frustration at the lack of answers, and she said that when she and her physician finally found a treatment regimen that worked, "that visit with her was kind of bittersweet because it was like, great, something's working, I have an answer---but it's fibromyalgia, and I'm going to have this for the rest of my life."
The path to diagnosis was often long and stressful. Nevertheless, being diagnosed was important because the diagnosis enabled participants to move forward in terms of figuring out how to manage their condition.
### Acceptance
After participants were diagnosed, it often took time for them to move towards acceptance, which consisted of two parts: acceptance of the diagnosis and acceptance of the illness. In terms of the diagnosis, participants seemed to come to accept it because their symptoms matched clinical descriptions of fibromyalgia. P10 initially did not believe that she had it, but "as time progressed, and I had other symptoms, like migraines...that went along with fibromyalgia, as I read more about fibromyalgia, I kind of accepted that I had fibromyalgia."
The second part involved an acceptance of the illness as being there to stay. In the beginning, some participants felt that the illness was temporary, and they were looking for a way to "fix it," for the "magic pill" \[P05\] or the "magic bullet" \[P17\]. P15 described her experience:
> 2013 was going to be my year of...health...it was going to be the year that I got myself back. Yeah right...at a certain point, I adjusted to a year of concentrating on wellness, instead of just, "Oh, I'm going to be...cured." Because I kind of felt like, "I'm going to cure myself!"...I think it's kind of like a pipe dream that some people like me will cling to and...I need to accept...I think you have to grieve like you have to grieve any other loss or death...you have to go, "Okay that's the old me," and "this is the new me."
>
> P15
### Self-Management
Achieving an acceptable level of self-management was often multidimensional, including both symptom and emotion management, and was predicated upon acceptance:
> For once in a very long time, I felt like I could handle the fibromyalgia\...I was starting to realize: ok, it's a part of my life, and I started noticing some of the triggers a little bit better, also the best way to manage some of the symptoms, and also not being so mad at myself or my body.
>
> P05
The concept of self-management does not mean the elimination of symptoms, but rather, reaching a point where patients believe that they are able to manage their symptoms or that the extent to which they experience symptoms is "acceptable." For many, this may mean "wellness." Patients may still experience "flares," but to a lesser extent. The level of functioning for each individual might differ significantly, but there is a pragmatism to it in the sense of finding solutions that fit people's lives: "...basically what I have learned is that you just manage your life" \[P21\].
All participants made changes to their lifestyles. Many made dietary changes to avoid trigger foods; others had strategies such as having nuts on hand to avoid hypoglycemic episodes. Many found that exercise was helpful, particularly yoga. Participants also used alternative therapies such as massage therapy, acupuncture, meditation, and hypnotherapy. Participants reported that meditation and hypnotherapy were effective for both pain management and emotion regulation.
P21's timeline aptly illustrates the main health management features ([Figure 1](#figure1){ref-type="fig"}). She started out her journey with a gas tank that was always low, meaning that she was constantly fatigued and struggling with different health issues. Along the way, she saw multiple doctors (represented by the stethoscopes). Towards the end of her journey, she encountered and tried multiple alternative modalities (represented by the trees). At the end, she developed an effective management strategy and was able to consistently maintain a fuller gas tank.
Information Seeking
-------------------
In terms of information seeking, the predominant pattern was a move from active information seeking to monitoring information sources with intermittent focused searching.
######
Participant characteristics.
------------------------------------------------------------------------------------------
Characteristic Category n \%
-------------------------------- --------------------------------------------- ---- ------
**Age**
\ 21-30 4 17.4
\ 31-40 4 17.4
\ 41-50 2 8.7
\ 51-60 7 30.4
\ 61-70 5 21.7
\ 71-80 1 4.3
**Gender**
\ Female 22 95.7
\ Male 1 4.3
**Race/Ethnicity**
\ White 20 87.0
\ Black 2 8.7
\ Asian 1 4.3
**Education**
\ Some college 2 8.7
\ 2-year or 4-year college degree in progress 1 4.3
\ 2-year or 4-year college degree 8 34.8
\ Graduate degree 11 47.8
\ Graduate degree in progress 1 4.3
**Employment status^a^**
\ Student 3 13.0
\ Employed full-time 11 47.8
\ Not employed 1 4.3
\ Retired 9 39.1
**Received disability**
\ Yes 8 34.8
\ No 15 65.2
**Illness duration**
\ ≤4 years 3 13.0
\ ˃4 years 20 87.0
**Social media participation**
\ Non-user/lurker 9 39.1
\ Infrequent participator 5 21.7
\ Frequent participator 9 39.1
------------------------------------------------------------------------------------------
^a^Participants may belong to more than one category.
### Active Information Seeking
In general, participants engaged in active information seeking toward the beginnings of their illness journeys. Prior to knowing what they had, participants looked for conditions that had similar symptoms; others used symptom checkers. Some participants suspected that they had fibromyalgia and discussed it with their doctors. Participants employed a diverse array of information sources including print (newspapers, magazines, books) and digital media (Twitter, informational websites, discussion forums, blogs, webinars, e-newsletters, and e-books), people (health care practitioners, authors, family, and friends), informational and emotional support venues, patient education courses, and others (commercials, radio, and television).
Once participants had an idea that fibromyalgia was what they had, they would engage in more extensive information seeking. When P17 was first diagnosed, she read "everything \[she\] could get her hands on." Prior to her diagnosis, P11 sought answers both online and through her doctors, but after she was diagnosed, she "googled everything on creation" ([Figure 2](#figure2){ref-type="fig"}).
A period of confusion often followed the diagnosis as participants searched for information and found no clear answers. Many participants said that they saw multiple doctors and had numerous lab tests. P19 had "this battery of tests for, you know, we call it the symphony of catastrophic diseases? Lupus, MS..." P05 said that one doctor thought it was fibromyalgia; another said arthritis, and "they're like, we're not exactly sure. So it was a very uncertain time in terms of figuring it out because it wasn't getting any better" \[P05\]. P06's encounters with health care practitioners did not appear to be leading towards a resolution. Thus, she ended up trying to figure things out herself: "I was doing research on my own and realizing that there was no medical consensus about what fibromyalgia was or how to treat it, so I really ended up doing a lot of research on my own" \[P06\].
Some participants found the diagnosis empowering in that it enabled them to do something to help themselves. In the case of P09, it helped in her search for information: "Once I had a name, I searched for fibro communities. Before that, I kind of trawled the Internet looking for places where other people talked about it...I searched for the symptoms, ignored what the forum was, and just sort of talked in various forums." Participants found it frustrating to not know what was happening with them, and active information seeking often continued through the point of diagnosis, until they developed an effective management approach.
{#figure2}
### Transition to Monitoring
Eventually, study participants came to accept their illness and learned to manage it. Along with these changes, participants also engaged in less information seeking. There were various reasons. First, because information seeking was often symptom-driven, as participants were able to get their symptoms under control and/or learned to manage their condition better, they felt less of a need for information. When asked if she currently searched for information, P09 responded: "The symptoms fit the diagnosis, and while it's frustrating to have that diagnosis and I don't like it, and I wish it were something that were curable, I've mostly stopped, um, because the current treatment regimen is helping." Many participants reached a point where they just wanted to move on:
> For me, you just kind of reach a point where it's like (sigh), "It is what it is. I'm going to continue to eat as healthy as I can. I'm going to continue to walk every day. I'm going to continue to lead as healthy a life as possible." But I really am kind of done trying to figure it all out. I just want to get on with my life...When I have a bad day, "Oh well," I don't really question it anymore.
>
> P14
Participants also stopped seeking information because they felt that they knew what was out there, and there was no new information to be found. P06 said, "I occasionally get back online and kind of look up and see where the research is at. Every now and then I'll see a research study highlighted about fibromyalgia and I'll read it just to see if there's any major breakthroughs \[chuckle\], but there never are." Most participants tried to maintain some level of awareness of fibromyalgia research through "monitoring," that is, subscribing to e-newsletters from services such as Medscape. There is a lingering sentiment of wanting to know: "I'm still signed up for a couple of newsletters, but I don't look at them on a regular basis. They flood...they come to my inbox and I don't want to...unsubscribe...there are moments...I just want to see latest conversations, and latest research\...So I keep them...they all go into a folder" \[P05\].
Searching for information on an as-needed basis also served to reduce information overload: "\[I\] started looking for specific things for specific problems...I wanted to piece \[together\] what would cause different areas of the symptoms, instead of looking at it as a whole, 'cause then it was just overwhelming" \[P05\]. Thus, on the whole, participants eventually settled into a pattern of monitoring. But once in a while, the appearance of new symptoms, serendipitous encounters with new information, and other events might trigger some focused searching.
If a patient develops a new condition, they may cease monitoring and cycle back to active information seeking. For example, after P17 developed fibromyalgia, she engaged in active information seeking and participated in online discussions, but eventually her participation waned. She started seeking information and participating again, after being diagnosed with alopecia.
Information Consumption and Use Processes
-----------------------------------------
Participants engaged in information consumption and use processes throughout the course of their journey. Several key processes emerged: forming a coherent representation of the conceptual space, evaluation and synthesis of information, taking charge of one's health care, and iterative problem solving.
### Forming a Coherent Representation of the Concept Space
Over time, participants came to understand the concept space in different ways. P05 became acquainted with the scientific explanation: "I know the biology and the science behind fibromyalgia, what they say about it, the causes they don't really know about it..." Many came to know fibromyalgia in terms of the symptoms: "Mostly when I was still learning what fibromyalgia is, I was looking at symptoms and stuff like that... at all these different websites explaining what is going on, and what is its effect, and how people with fibromyalgia are going to feel" \[P13\]. P06 surveyed the online space and selected treatments based on her own comparison of patient reports: "I felt like I was doing all the major recommendations...massage, getting a lot of rest, the one particular drug that a lot of people have had success with, the guaifenesin." There are differences in the ways that participants represented this space, but each formed an understanding of the space that they could accept, in other words, that was coherent to them.
At some point in their illness journeys, many participants came to a point of saturation, where they felt as if there was "nothing new" \[P20\] and that they knew "all there was to know" \[P25\]. This coincided with the transition to monitoring described earlier.
### Evaluation and Synthesis of Information
As they were coming to understand the concept space, participants continually encountered and evaluated information. Many participants read extensively about fibromyalgia and synthesized across sources. They developed their own heuristics for evaluating the quality of information. One common rule was that they dismissed information that "promised a cure" \[P09\]. P22 looked "to see if they're accurate about the basic mechanics of how it \[fibromyalgia\] works." Others looked for consensus across multiple sites.
P13 and P27 engaged in another type of synthesis, involving comparison of explanatory perspectives on fibromyalgia. P13 quickly realized that allopathic medicine's explanation of fibromyalgia did not satisfy him and moved on to study Chinese medicine. P27 said, "As soon as the doctors are thinking, okay, this might be fibromyalgia, I started doing research on the Internet, but then also checked out at least ten fibromyalgia books from the library, just to read different perspectives on what fibromyalgia is and, like, differing ways that you can treat it."
### Taking Charge of One's Own Health Care
Over time, many participants realized that they needed to take charge of their own health care. They showed this initiative in various ways. P15, like many others, went through a prolonged period of lab tests and consultations prior to being diagnosed. During that time, she realized that even if she were diagnosed, she would refuse medication, so she needed to take matters into her own hands: "I was not getting what I wanted to get with the doctors and all that, so I thought, 'Well, what would you do differently if you had the diagnosis?...Whatever that is, you need to start doing it now'" \[P15\]. Then she began an elimination diet, which involved progressively removing items from her diet until she figured out what she was sensitive to.
Though health care practitioners may be experts in their respective areas, patients are likely to have a more intimate understanding of their own body. P12 explains the rationale for taking charge: "Doctors know some things, and you know some things. And you've got to have somebody who lets you put that together, 'cause you're the expert on your body, and they might be the expert on some treatments, but then you're the one that has to sort of be your own case coordinator, and monitor your body."
This was not a role that participants naturally took on: "It took me a long time to be the manager of my own health system. I expected doctors to kind of manage my life for me. It took me a long time to realize that, no, I'm in charge of this. The doctors that work for me are a team, and I manage that team" \[P14\]. P14 depicts herself as a manager of her multidimensional care team ([Figure 3](#figure3){ref-type="fig"}, right).
{#figure3}
### Iterative Problem Solving
Over the course of their illness, participants also engaged in episodes of problem solving to address the physical problems that they experienced.
Figuring out a management strategy was often an iterative process involving trial and error. For example, in the case of exercise, participants often needed to experiment to find the right type of exercise and amount that they could handle. P17's story is a case in point: "I got myself an exercise bicycle...I worked up to ten minutes a day...but then the other 24...23 hours a day I was in the bed. So I figured that's not going to work." She said that eventually, "I learned to evaluate myself, how tired I was getting, and how I was feeling, so I could better pace myself and rest when I needed to, and that made a huge difference\...And yoga helped a lot in helping me become self-aware" \[P17\].
Another common issue was identifying food sensitivities and allergies. Some participants, such as P15, researched how to do this on their own. P24 used a mobile app called Pain Coach to track her food intake. This enabled her to figure out that gluten was causing her a problem, and then she eliminated it from her diet. P26 underwent lab tests and a rotation diet to figure out the foods to which she was sensitive.
Participants progressively made adjustments and/or engaged in additional information seeking based on evolutions in their thoughts and often, in their symptoms. P15 described this as "peeling back layers":
> Like with the nutrition, say I have an issue with certain types of foods...that's one thing, and then I am getting better nutrition, not the processed things...more organic stuff, and that's peeling off another layer...and getting better exercise and more movement, of the proper kinds of movement that don't cause me to have pain. That peels off another layer. And that exposes something else.
>
> P15
Not all of participants' energies were engaged in problem solving via information seeking; there was also internal sense-making. Participants found it frustrating that there were so many unexplained symptoms, and they were constantly trying to figure out the root cause of their problems: "If you have any intelligence at all and you want to get better, you want to try to figure out why you woke up feeling so bad. So I would go back and go through all the foods that I ate and go through everything. Did I go through these stresses?" \[P14\].
Discussion
==========
Principal Findings
------------------
This paper reported the findings of a qualitative study of fibromyalgia patients' illness journeys. At the outset, there were health-related features: moving from onset to diagnosis, acceptance, and development of an acceptable level of health management. Information seeking changed over time, particularly in terms of a transition from active information seeking to monitoring. Last, patients engaged in information consumption and use processes: forming a coherent representation of the conceptual space, evaluation and synthesis of information, taking charge of one's health, and iterative problem solving.
Aligning the themes from the interview content temporally affords an integrated conceptualization of how health management and information behaviors are related over time ([Figure 4](#figure4){ref-type="fig"}). Patients engage in active information seeking at the beginning of their journeys, beginning with cognizance of their condition and extending through diagnosis and acceptance. As they begin to develop an acceptable level of management, their information seeking tapers to a pattern of monitoring with intermittent focused searching. The spacing of the four phases (ie, onset, diagnosis, acceptance, and management) is intended to reflect their relative temporal differences, though this may change if we observe a decrease in times to diagnosis.
Alongside these developments, participants continually engage in information consumption and use. They form and refine their interpretations of the concept space. They also engage in information evaluation and synthesis activities, which become more sophisticated over time but decrease due to lessened need. They learn to take charge of their health care. Periods of iterative problem solving to address issues such as irritable bowel, fatigue, and sleep problems also occur. The information consumption and use processes might also be thought of as activities that individuals become increasingly skilled at over time. A set of arrows from monitoring back to active information seeking indicates that patients may traverse the process again for a new condition.
The journey described in this paper reflects the predominant patterns among study participants. Though most participants engaged in extensive information seeking, a few participants did not. P18 said, "I didn't really look into it \[fibromyalgia\]...I just kind of accepted it for what it was and didn't worry about it." P07 said that she was never really a very curious person. Both participants ended up learning more about fibromyalgia serendipitously, decades after they were diagnosed, when they engaged in exercise that inadvertently led to health information encounters. For P10 and P25, who had conditions that pre-dated fibromyalgia, there also seemed to be less of an impact.
{#figure4}
Comparison With Prior Research
------------------------------
Though there has been considerable work investigating the lived experiences of fibromyalgia patients (eg \[[@ref5],[@ref40]-[@ref42]\]), at least to this author's knowledge, there is no existing work that has taken a qualitative approach to investigate the relationship between health management and information behavior in fibromyalgia patients over time. The results of this study illustrated that, over time, participants became clearer and more accepting of their condition. These developments were accompanied by an increased awareness of and ability to use information sources to improve their health management, as well as improved communication with physicians and other health care providers.
The study findings share similarities with existing research in health literacy and self-management in chronic illness. In this study, participants developed skills that have also been identified in a meta-synthesis of self-management in chronic illness (eg, taking ownership of health needs, adjusting to illness, and activating resources) \[[@ref43]\]. Additionally, the study findings showed that participants developed skills that have been associated with increased health literacy, such as becoming more active in clinical consultations, and greater autonomy and empowerment in decision making \[[@ref44]-[@ref46]\]. Previous research testing an Internet-based self-management intervention for fibromyalgia has also demonstrated that increased knowledge can lead to better health outcomes \[[@ref47]\]. This convergence in the study findings and the self-management literature suggests that some of the lessons learned from this study might be incorporated in the design and delivery of self-management education.
Implications for Information System Design
------------------------------------------
There are various ways that information technology could support fibromyalgia patients' information needs. First, to make sense of their condition, participants attempted to synthesize information across diverse source types and from multiple explanatory perspectives. Because they are not focusing on a single information source, tools that help patients make sense of and compare information sources could be particularly helpful. Building interfaces that enable patients to discern and make sense of explanatory perspectives has also been suggested in a previous study concerning information about Lyme disease \[[@ref48]\]. Systems that help users understand and evaluate health information from different perspectives could be invaluable for conditions in which there are many treatment options, multiple alternative perspectives, and unclear treatment protocols.
Participants also expressed frustration because there were so many factors that could be influencing their health, and it was difficult to disentangle them (eg, P14). Thus, the development of tools to analyze different types of personal health data and the integration of knowledge bases to provide additional information are important directions for future development. Fibromyalgia patients are also more likely to have comorbidities \[[@ref49]\], which can make it seem like work to track data \[[@ref50]\].
Last, once patients start developing a picture of the concept space, they experience less of a need to seek information but are still interested in keeping up with the research. Patients who engage in monitoring could benefit from tools that automatically process newsfeeds and flag articles for perusal.
Limitations and Future Directions
---------------------------------
Though this study provided valuable knowledge about how people use information in the context of chronic illness, there are limitations. The participants in this study had all reached acceptance, most had achieved a stable level of health management, and almost all had stopped actively seeking information. As such, the study afforded a glimpse of participants' journeys through their eyes, at a particular point in their journeys. There are potential problems with retrospective recall, including errors in memory, lack of clarity about past events, and differences in interpretation of one's history over time. Some persons may never reach acceptance or develop effective management strategies; their perspectives are not reflected in this sample.
Thus, an important priority for future research is to work with patients earlier in their illness journeys. Though participants in this study reported developing a more coherent representation of the concept space, more familiarity with and ability to evaluate information sources, and increased ability to take charge of their health care over time, there is still much that we do not know about how these skills evolve. Additional research focusing on critical time periods such as postdiagnosis, and important activities, such as problem solving and sense-making about symptoms, could inform the design of patient education programs.
The author would like to thank Dr Barbara M. Wildemuth for her guidance and suggestions through the process of conducting this research, and the study participants for sharing their experiences.
Conflicts of Interest: None declared.
Interview 1 Guide.
Interview 2 Guide.
| {
"pile_set_name": "PubMed Central"
} |
Background {#Sec1}
==========
Enhancing the grain yield potential of wheat is a key focus of wheat breeders. Grain yield is the product of various yield components. Wheat grain yield per unit area is the product of grain yield per spike (GYS) and the number of spikes per unit area. The latter depends on sowing density and is highly affected by environmental factors. The GYS is determined by grain number per spike (GpS) and thousand grain weight (TGW), which are variably correlated in different wheat collections/populations. A significant negative correlation between these two traits has been reported in bi-parental populations \[[@CR1]--[@CR3]\], but no significant correlation was observed between TGW and GpS in collections of Chinese landraces \[[@CR4]\], French winter wheat cultivars \[[@CR5]\] and CIMMYT-derived spring wheat collections \[[@CR6]\]. On the other hand, a significant positive correlation between TGW and GpS was reported in modern Chinese cultivars \[[@CR4]\].
TGW in wheat has been one of the target traits for selection during domestication and breeding \[[@CR7], [@CR8]\]. For example, in China, an increase in wheat yield potential from \~1 T ha^−1^ in 1992 to \~5.4 T ha^−1^ today is mainly due to the genetic increase in TGW from \~20 g to \~45 g, respectively \[[@CR9]\]. However, genetic gain in TGW has not reached its limit and thus provides an opportunity to increase yield potential \[[@CR9]\]. It is estimated that an increase in yield of 140--160 kg ha^−1^ can be obtained by just a 1-g increase in TGW \[[@CR10]\]. However, genes and their roles in controlling TGW in wheat are still largely unknown. In wheat, TGW is a quantitative trait controlled by several genes/QTL distributed on all chromosomes \[[@CR8], [@CR11]\]. For example, Su et al. \[[@CR12]\] discovered eight TGW-related QTL on chromosomes 2D, 4B, 5A, 7A and 7B, explaining up to 16.2% of the phenotypic variation. Similarly, four QTL for TGW on chromosomes 1D, 2A, 5D, and 6A explained 5.9 to 20.1% of phenotypic variation in different environments \[[@CR13]\].
Nevertheless, from this plethora of QTLs, few loci/genes have been cloned by map-based cloning approaches mainly because of the large and complex hexaploid genome (\~17 Gb) that consists of three homeologous genomes (A, B, D) and an abundance of repeat sequences (80%) \[[@CR14]\]. Studies on comparative genomics have shown high synteny and collinearity among different grass genomes, such as those of wheat, barley, rice, millet, maize and sorghum. This pattern of genome organization in the members of the grass family provides a powerful approach for gene discovery in common wheat \[[@CR15]\]. A large number of genes have been discovered in common wheat by synteny-based cloning, in which the gene sequences of model crops such as rice and barley have been used as references to identify orthologous genes in wheat. For example, the genes *TaTGW6* \[[@CR16], [@CR17]\], *TaCwi-A1* \[[@CR18]\], *TaSus2-2B* \[[@CR19]\], *TaSus2-2A*, *TaSus1-7A* \[[@CR20]\], *TaGW2-6A*, *6B* \[[@CR9], [@CR12], [@CR21]\], *TaCKX6-D1* \[[@CR22]\], *TaSAP1-A1* \[[@CR23]\], *TaGS1a* \[[@CR24]\], *TaGS-D1* \[[@CR25]\], and *TaGASR-A1* \[[@CR26]\] were discovered using rice-wheat synteny and using molecular markers in marker-assisted wheat breeding. Hence, the isolation and characterization of genes controlling grain size in common wheat will help breeders maximize yield potential by establishing gene-based breeding programs.
The *FLOURY ENDOSPERM2* (*Flo2*) gene is a member of a conserved gene family in plants. In rice, this gene has been shown to have a tetratricopeptide repeat (TPR) motif consisting of 3--16 tandem repeats of 34 aa residues that mediate protein--protein interactions in the nucleus \[[@CR27], [@CR28]\]. The *OsFlo2* gene was cloned in the *indica* cultivar 'Kasalath'; this gene was found to have 23 exons and 22 introns and coded for a protein consisting of 1720 amino acid residues that had three TPR motifs in the middle \[[@CR27]\]. The expression of *Flo2* was constitutive in both vegetative tissues and developing seeds, and the expression was relatively high level in developing seeds. The *flo2* mutants exhibit a significant reduction in amylose content and grain weight and exhibit altered expression of various starch synthesis-related genes, indicating its key role in regulating rice grain weight and starch quality \[[@CR27], [@CR28]\]. In this article, we report the rice-wheat synteny-based isolation of *Flo2* orthologs in hexaploid wheat, the association of *TaFlo2-A1* sequence polymorphisms with TGW and the comparison of temporal expression profiles of *TaFlo2-A1* haplotypes in flag leaves and developing caryopses.
Methods {#Sec2}
=======
Plant materials {#Sec3}
---------------
For cloning the *TaFlo2* gene in hexaploid wheat, Chinese Spring (CS) and two sets of cultivars with lower and higher TGW were used; the set of cultivars with higher TGW included Dixiuzao (49.5 g), Enmai4 (49.2 g), Liying 5(49.4 g) and Laizhou 953 (52.2 g), and the set with lower TGW included Jinyang 60 (23.5 g), Baihuamai (24.1 g), Sanyuehuang (25.2 g) and Zipi (25.5 g). The Chinese Spring nulli-tetrasomic lines were used to assign *TaFlo2* genes to wheat homeologous chromosomes. The Chinese Micro Core Collection (MCC, 262 accessions) and Pakistani wheat collection (130 accessions) were used to confirm the association between *TaFlo2-A1* haplotypes and TGW. To avoid the effect of population structure, normalized MCC subpopulations were used for association analysis \[[@CR19], [@CR29]\]. The Pakistani collection was selected based on previous reports \[[@CR30], [@CR31]\] considering the effect of population structure on association analyses.
Cloning and characterization of *TaFlo2* sequences {#Sec4}
--------------------------------------------------
The genomic sequence of the rice *OsFlo2* gene (NC_008397) was used as a query for BLAST searches against the wheat sequences database in the URGI (<https://urgi.versailles.inra.fr>/). All wheat scaffold sequences with high similarity to the rice *OsFlo2* sequence were assembled to construct a putative *TaFlo2* gene using DNAMAN (<http://www.lynnon.com>). Based on the scaffold sequences, six conserved primer pairs were used to specifically amplify *TaFlo2* coding and promoter sequences from the three wheat sub-genomes: A, B and D (Table [1](#Tab1){ref-type="table"}). The *TaFlo2* mRNA of 4902 bp was cloned in Chinese Spring using three primer pairs designed from the predicted mRNA sequence (Table [1](#Tab1){ref-type="table"}). Genomic DNA was extracted from young seedlings of each line using the CTAB method \[[@CR32]\]. A 20-μl reaction volume comprising 0.5 μl (5 μM) of each primer, 2× Taq mix (GenStar, Beijing, China) and 100 ng of DNA was used for PCR amplification that consisted of a cycle profile of 5 min at 94 °C; 35 cycles of 30 s at 94 °C, 30 s at 60 °C and 4 min at 72 °C; and a final extension of 10 min at 72 °C. The PCR products were detected by electrophoresis in 1% agarose gels with nucleic acid dye (Tiangen, Beijing, China), and gel images were captured using a UV spectrometer (BioRad, Hercules, CA, USA). The targeted PCR products were obtained from the agarose gels and purified using the TIANgel MIDI Purification Kit (Tiangen, Beijing, China). The purified PCR products were then ligated into the pGEM-T Easy cloning vector (TransGen Biotech, Beijing, China). The ligation product was transformed to 50 μl of Trans1-T1 competent cells by the heat shock method (Tiangen, Beijing, China). Positive clones from each transformation were selected based on positive PCR tests and were sequenced (Beijing Genomics Institute). The sequences were analyzed using DNAMAN software (<http://www.lynnon.com>).Table 1Primer sequences used in this studyPrimer namePrimer sequence (5′-3′)Position on scaffold sequenceAnnealing temperature (°C)PCR product sizeFunctionFlo2-1FTGTGCTGGAATCACCCACTC793--812601061cloning TaFlo2 /polymorphism detectionFlo2-1RGCGCGGCGAAAACTAATCAT1853--1844Flo2-2FGTGCCGTCCATAATCGTTGC1546--1565601781cloning TaFlo2 /polymorphism detectionFlo2-2RCATGTGCGGCAAAAGACACA3326--3307Flo2-3FAACGGGCATGTGTCTTTTGC3299--3318603025cloning TaFlo2 /polymorphism detectionFlo2-3RCGACGCAGCTCTGAAAATCG6332--6313Flo2-4FCGCTTAGCAGTGGATTTGCC5719--5738603948cloning TaFlo2Flo2-4RATCCAACAAACAGGTGCCCA9667--9647Flo2-5FTTGCGGAAGCCCATCATTCT8387--8406603836cloning TaFlo2Flo2-5RTGACCTTCTGCGGATGCTTT1222--12,203Flo2-6FCAGAACAGGGCCGGTACAAT11,368--11,387602600cloning TaFlo2Flo2-6RCGCTCATCTGGATAGGGCAA13,967--13,948TaFlo2-InDel8FACCCCTCCTCCGTTATCGTC1337--135660145/1538-bp InDel polymorphism in *TaFlo2-A1*TaFlo2-InDel8RCCTCCTTCTTCTTGCGGTCG1470--1489Flo2-A1FGTGCTCCGATCCGATGTGCAGTTAT5387--5411585872A specificFlo2-A1RGTGCACAACCAAGTAAAAGG5973--5954Flo2-B1FGTC ATC ACTAGAGGA ATTTTCC6851--6872589022B specificFlo2-B1RCTCTCAGAACTGTGGAT7752--7736Flo2-D1FCTGTATCTGTAATTTGTTCCG5378--5398583262D specificFlo2-D1RCTTCCGAAAAATGTGGGG5704--5687mFlo2-1FTAACGGTGGTGCACTTGTGT--581868Cloning mRNAmFlo2-1RTCAGCCGCAAGTTATGCTCA--mFlo2-2FTGCGGACGAGATGGAAAACA--581809Cloning mRNAmFlo2-2RAGCAGTCAGCCGATGGTATG--mFlo2-3FATGCGTACTCCCTAAGCGTG--581889Cloning mRNAmFlo2-3RCACGAAGTGCTGCTTGCTTT--eTaFlo2FCCATTCGGCTTTCGTGCAAA--55134Expression analysiseTaFlo2RTGTTTTCCATCTCGTCCGCA--ActinFAGCCATACTGTGCCAATC--55134Internal controlActinRGCAGTGGTGGTGAAGGAGTAA--
Characterization of *TaFlo2-A1* haplotypes and development of haplotype-specific markers {#Sec5}
----------------------------------------------------------------------------------------
The 262 MCC and 130 Pakistani varieties were genotyped with the primer pair TaFlo2-InDel8, and PCR product was run on 8% polyacrylamide gels. Based on TaFlo2-InDel8 scoring, the MCC and Pakistani accessions sorted into two groups according to their haplotypes (*TaFlo2-A1a* or *TaFlo2-A1b*) for the *TaFlo2-A1* gene. For MCC, the average values of TGW of the two haplotype groups were calculated using replicated data collected from 3 years (2002, 2005, 2006) of plants in Beijing \[[@CR19]\]. For Pakistani varieties, the average values of TGW of the two haplotype groups were calculated using replicated data from 2 years (2009, 2010) of field trials at the University of Agriculture, Faisalabad. The resulting values were then compared and statistically analyzed using SPSS 13.0 for Windows (IBM, New York, USA).
Quantitative RT-PCR analysis of *TaFlo2-A1* haplotypes {#Sec6}
------------------------------------------------------
The Yangmai 19, Chinese Spring, Pubing3228, Shannong23 and Zhengmai9405 varieties were sown at the experimental station of the Institute of Genetics & Developmental Biology, CAS in Beijing, China in October 2014; three rows of each variety were planted. The length of each row was 2 m, and the row-to-row distance was 20 cm. The plants were managed in accordance with standard agronomic practices; irrigation and fertilizer were supplied for optimal growth. Twelve- day-old flag leaves of five plants from each variety were harvested and stored at −80 °C. Unfertilized grains were collected from each variety 1--2 days before flowering (DBF). Fertilized grains were collected from each variety at 5, 10, 15, 20 and 25 days after flowering (DAF). The flag leaf and developing grain samples were processed for the preparation of total RNA as described previously \[[@CR33]\]. Three biological replicates that were collected from different plants were analyzed separately for each variety for quantitative RT-PCR evaluation. For *TaFlo2-A1* transcripts analysis, the primer set eTaFlo2, which is specific for *TaFlo2-A1* (Table [1](#Tab1){ref-type="table"}), was designed and used. Quantitative RT-PCR was then carried out as described by Feng et al. \[[@CR34]\]. The wheat actin gene was used as an internal control. The relative expression level of *TaFlo2-A1* in each flag leaf and in each sample of developing grains was calculated using the data of three technical replicates as described previously \[[@CR35]\]. Statistical comparisons of *TaFlo2-A1* expression levels (presented as the mean ± SD) among different samples were made by ANOVA using SPSS 13.0.
Bioinformatics comparison of nucleotide and protein sequences {#Sec7}
-------------------------------------------------------------
Nucleotide and protein identities among the compared sequences were calculated using DNAMAN (<http://www.lynnon.com>). Amino acid sequence alignment was accomplished using ClustalW2 in EMBLEBI ([www.ebi.ac.uk/Tools/msa/clustalw2](http://www.ebi.ac.uk/Tools/msa/clustalw2)). Potential signal peptide sequences in the deduced proteins of TaFLO2 and its homologs were predicted using Softberry software (<http://www.softberry.com/berry.phtml>). The predicted TaFLO2 protein was BLASTed both in the NCBI smart blast system (<http://blast.st-va.ncbi.nlm.nih.gov/smartblast>) to search for homologous proteins and in the NCBI CD system (<https://www.ncbi.nlm.nih.gov/Structure/cdd/wrpsb.cgi>) to search for conserved domains.
Results {#Sec8}
=======
Cloning and characterization of *TaFlo2* genes {#Sec9}
----------------------------------------------
To select potential candidate *TaFlo2* genes, the rice *Flo2* sequence (NC_008397) was used as a query against the wheat genome sequences database in the URGI (<https://www.ncbi.nlm.nih.gov/Structure/cdd/wrpsb.cgi>). Three bread wheat scaffolds (IWGSC_chr2AL_ab_k71_contigs_longerthan_200_6436403, IWGSC_chr2BL_ab_k71_contigs_longerthan_200_7959819, IWGSC_chr2DL_ab_k71_contigs_longerthan_200_9909583) with high similarity (*E* value = 0 and similarity \>73%) were identified as potential orthologs to the rice *Flo2* gene. The sequences of these scaffolds were downloaded and assembled with DNASTAR (<http://www.dnastar.com/>) to construct a putative *TaFlo2* sequence.
To search *TaFlo2* homologs and predict their deduced protein sequence and structure, Softberry (<http://www.softberry.com/berry.phtml>) and NCBI (<https://www.ncbi.nlm.nih.gov/>) tools were used. The deduced proteins of *TaFlo2-A1* (1734 aa), *TaFlo2-B1* (1698 aa) and *TaFlo2-D1* (1682 aa) were highly similar (\>94% identity among themselves) and exhibited \>77% similarity with the rice FLO2 protein. Like in the rice FLO2 protein \[[@CR27]\], four tetratricopeptide repeat (TPR) motifs were observed in the deduced TaFLO2 protein at the positions of 947--988, 1032--1072, 944--1017 and 1028--1106 amino acid residues. Furthermore, three mitochondrial CLU domains were also observed at 737--878, 50--162 and 357--401 amino acid residues (Fig. [1a](#Fig1){ref-type="fig"}). The TaFLO2 protein showed high similarity with *Aegilops tauschii*, *Brachypodium distachyon* and long-grain rice proteins (Fig. [1b](#Fig1){ref-type="fig"}).Fig. 1**a** Putative structure of the OsFLO2 and TaFLO2 proteins. Clu_N (mitochondrial function, CLU-N-term), CL (clustered mitochondria domain), CLU-center (an uncharacterized central domain of CLU mitochondrial proteins), TPR (tetratricopeptide repeat). **b** Similarity between the TaFLO2 protein and proteins from related plant species
To clone the full-length genomic sequence of *TaFlo2* in Chinese Spring, six conserved primer pairs were used (Table [1](#Tab1){ref-type="table"}). The assembly of sequences with the six conserved primer pairs yielded three fragments, 14,009, 14,078 and 13,814 bp. Based on alignment with wheat scaffolds in the database and genome-specific primers, the fragments were designated *TaFlo2-A1*, *TaFlo2-B1* and *TaFlo2-D1*. The open reading frames of *TaFlo2-A1*, *TaFlo2-B1* and *TaFlo2-D1* were 12,183 bp, 12,270 bp and 12,022 bp in length, respectively. The *TaFlo2* mRNA of 4902 bp was cloned with three primer pairs (Table [1](#Tab1){ref-type="table"}). Based on the prediction and alignment with cloned mRNA, the cloned genomic sequences from 2AL, 2BL and 2DL consisted of 23, 23 and 24 exons, respectively (Fig. [2a](#Fig2){ref-type="fig"}). Among the three homoeologs, the sequence and size of the first four exons were conserved, whereas the size and sequence of the other exons varied.Fig. 2Polymorphism, molecular marker and ORF structure of *TaFlo2* homoeologs. **a** Exon and intron pattern in the ORFs of *TaFlo2-A1*, *TaFlo2-B1* and *TaFlo2-D1*. The length of each ORF (between the start and stop codons) is shown to the right of the graph. The number of nucleotides in each exon or intron is indicated. CDSf: First coding sequence, CDSi: Internal coding sequence, CDSl: Last coding sequence. **b** Alignment of the part of cloned *TaFlo2* orthologs. Polymorphism both in the promoter and first intron is indicated by stars. **c** PCR product of molecular marker TaFlo2-InDel8 discriminated by PAGE in 60 MCC accessions
Polymorphism detection in *TaFlo2-A1* {#Sec10}
-------------------------------------
To detect polymorphisms in the putative *TaFlo2* sequences between high and low TGW accessions, three conserved primers that covered scaffold segments from 793 to 6332 bp were used (Table [1](#Tab1){ref-type="table"}). Polymorphism in *TaFlo2-A1* sequences between high and low TGW accessions was observed between 1396 to 1791 bp, while no sequence variation was observed between high and low TGW accessions in *TaFlo2-B1* and *TaFlo2-D1* (Fig. [2b](#Fig2){ref-type="fig"}; Additional file [1](#MOESM1){ref-type="media"}: Figure S1). The conserved sequences of *TaFlo2-B1* and *TaFlo2-D1* in all the higher and lower TGW accessions implicated non-functional nature of these genes. An 8-bp InDel was identified in *TaFlo2-A1* sequences from 1396 to 1403 bp which was −17 to −10 bp upstream of the first coding sequence (ATG) at position 1417--1419 bp (Fig. [2b](#Fig2){ref-type="fig"}). Five SNPs (G/C, A/G, C/T, C/G and −/T) were observed at 1514, 1538, 1545, 1727 and 1791 bp. From the start codon (ATG), the positions of these five SNPs (G/C, A/G, C/T, C/G and −/T) were in the first intron at 98, 122, 128, 311 and 375 bp, respectively (Fig. [2b](#Fig2){ref-type="fig"}). The 8-bp InDel and the five SNPs together formed the two haplotypes designated *TaFlo2-A1a* and *TaFlo2-A1b* (Fig. [2b](#Fig2){ref-type="fig"}). From the position 1792 to 6332 bp, no polymorphism was observed in the *TaFlo2-A1* sequence.
Molecular marker development and characterization of *TaFlo2-A1* haplotypes {#Sec11}
---------------------------------------------------------------------------
To characterize the observed *TaFlo2-A1* haplotypes in large wheat populations, a molecular marker based on the 8-bp InDel observed in the promoter region was designed and named TaFlo2-Indel8 (Table [1](#Tab1){ref-type="table"}). The forward and reverse primers of TaFlo2-Indel8 are located at −80 bp and 72 bp from the start codon, respectively. The PCR products of TaFlo2-Indel8 in the accessions with or without the 8-bp InDel have lengths of 153 bp and 145 bp, respectively. The bands of 153 bp and 145 bp were easily discriminated by polyacrylamide gel electrophoresis and represented the haplotypes *TaFlo2-A1a* and *TaFlo2-A1b*, respectively (Fig. [2c](#Fig2){ref-type="fig"}).
Chromosomal location of *TaFlo2* genes {#Sec12}
--------------------------------------
To assign chromosomal locations to *TaFlo2* genes, genome-specific primers and a set of Chinese Spring nulli-tetrasomic lines were used. The *TaFlo2* genes *TaFlo2-A1*, *TaFlo2-B1* and *TaFlo2-D1* were found to be located on chromosomes 2A, 2B and 2D (Fig. [3](#Fig3){ref-type="fig"}). The cloned sequences of *TaFlo2-A1*, *TaFlo2-B1* and *TaFlo2-D1* showed \>99% similarity with 2AL, 2BL and 2DL scaffolds (IWGSC_chr2AL_ab_k71_contigs_longerthan_200_6436403, IWGSC_chr2BL_ab_k71_contigs_longerthan_200_7959819, IWGSC_chr2DL_ab_k71_contigs_longerthan_200_9909583). Further analysis revealed that the *TaFlo2-A1* gene was located on deletion bin '2AL1--0.85-1.00'.Fig. 3Assignment of *TaFlo2-A1*, *TaFlo2-B1* and *TaFlo2-D1* to wheat chromosomes 2A, 2B and 2D, respectively, by PCR mapping with the genomic DNA of Chinese Spring (CS) and derivative nulli-tetrasomic lines (N2AT2B, N2BT2A and N2DT2A). The size (kb) of DNA markers is shown to the left of the image
Association of *TaFlo2-A1* with thousand grain weight {#Sec13}
-----------------------------------------------------
To associate *TaFlo2-A1* with TGW, two natural populations, the Chinese Micro Core Collection (MCC) and the Pakistani collection, were used. In the MCC, the homozygous *TaFlo2-A1a* haplotype was found in 219 (83.5%) accessions, whereas the *TaFlo2-A1b* haplotype was found in 43 (16.5%) accessions. In the Pakistani wheat collection, the number of accessions carrying *TaFlo2-A1a* and *TaFlo2-A1b* were 85 (64.6%) and 45 (35.4%), respectively. Both in the MCC and Pakistani collections, the positive haplotype *TaFlo2A-A1b* had a lower frequency, which showed the scope of improving grain weight.
The difference in TGW between the haplotypes *TaFlo2-A1a* and *TaFlo2-A1b* was statistically significant in both populations (*P \< 0.05*, Table [2](#Tab2){ref-type="table"}). In the MCC, the mean difference in TGW between the groups of accessions having *TaFlo2-A1a* and *TaFlo2-A1b* was significant (*P* ≤ 0.05) across the 3 years of data. The mean differences in TGW between the two haplotypes in 2002, 2005 and 2006 were 7.00 ± 1.12 g, 7.80 ± 1.11 g and 8.40 ± 0.94 g, respectively. Consistent with the results of the MCC, the mean difference in TGW between groups of accessions having *TaFlo2-A1a* and *TaFlo2-A1b* was also significant (*P* ≤ 0.05) across both years of data in the Pakistani wheat collection. The values of the mean difference between the two haplotypes (*TaFlo2-A1a* and *TaFlo2-A1b*) in the Pakistani wheat population were 4.50 ± 0.71 g and 5.20 ± 0.72 g for 2009 and 2010, respectively. The phenotypic variance for TGW explained by *TaFlo2-A1* haplotypes was 6.19% in 2002, 7.76% in 2005 and 8.37% in 2006 in the MCC. In the Pakistani collection, the phenotypic variance for TGW explained by *TaFlo2-A1* haplotypes was 4.42% in 2009 and 5.11% in 2010 (Table [2](#Tab2){ref-type="table"}). Moreover, to determine whether *TaFlo2-A1* also affects grain number per spike (GpS), an association analysis was performed for GpS in both populations. However, the differences in GpS between the haplotypes *TaFlo2-A1a* and *TaFlo2-A1b* were not significant in either population (*P* \< 0.05, Table [2](#Tab2){ref-type="table"}).Table 2Association of TGW and GpS with *TaFlo2-A1* in the Chinese Micro Core Collection and Pakistani wheat collectionsNatural populationsYear (number of accessions)*TaFlo2-A1a*\
Mean ± SE^a^ (number of accessions)*TaFlo2-A1b*\
Mean ± SE (number of accessions)Mean difference ± SEPVE (%)^b^ TGWTGWGpSTGWGpSTGWGpSChinese Micro Core Collection2002 (137)33.6 ± 0.54(98)50.8 ± 1.2(98)40.6 ± 1.2(39)49.7 ± 1.5(39)7.0 ± 1.1\*\*1.06 ± 2.1^ns^6.192005 (169)30.7 ± 0.52(128)43.1 ± 0.8(128)38.5 ± 1.1(41)40.5 ± 1.1(41)7.8 ± 1.1\*\*2.6 ± 1.5 ^ns^7.762006 (185)32.8 ± 0.43(141)51.4 ± 0.7(141)41.2 ± 0.9(44)48.6 ± 1.2(43)8.4 ± 0.9\*\*2.7 ± 1.5 ^ns^8.37Pakistani collection2009 (130)40.6 ± 0.43(85)46.6 ± 1.1(85)45.1 ± 0.55(45)44.4--1.8(45)4.5 + 0.71\*\*2.2 ± 1.9 ^ns^4.422010 (130)40.5 ± 0.47(85)47.8 ± 1.2(85)45.7 ± 0.46(45)44.9 + 1.6(45)5.2 ± 0.72\*\*2.9 ± 2.0 ^ns^5.11^\*\*^indicates significant differences, and ^ns^ indicates non-significant differences (*P \< 0.01*; Student's t-test) among groups carrying different haplotypes^a^Standard error^b^Percentage of phenotypic variance explained by association analysis
Collectively, our data demonstrated that *TaFlo2-A1*, like the *OsFlo2* gene in rice, is associated with TGW in wheat. Furthermore, the lack of association of *TaFlo2-A1* with GpS suggests that the high TGW of the examined genotypes is primarily due to the positive haplotype (*TaFlo2-A1b*) for high TGW instead of loci for low number of kernels per spike and/or low grain yield.
Expression of *TaFlo2-A1* is positively related to TGW {#Sec14}
------------------------------------------------------
To observe the contrasting effects of *TaFlo2-A1a* and *TaFlo2-A1b* on TGW at the gene expression level in flag leaves and developing grains, two polymorphic accessions were used. The expression level of *TaFLO2* was positively correlated with TGW, which is consistent with previous results in rice \[[@CR27]\]. The haplotype *TaFlo2-A1a*, which exhibits low expression levels, represented the group of accessions that have low average TGW, and the haplotype *TaFlo2-A1b*, which exhibits high expression levels, represented the group of accessions that have high average TGW in both Chinese and Pakistani wheat populations. Quantitative RT-PCR assays showed that for both types of haplotypes, the expression level was maximum in 12-day-old flag leaves followed by expression in developing grains sampled at 5 DAF. However, the expression of both types of haplotypes decreased rapidly in the fertilized caryopses collected at 10, 15, 20 and 25 DAF. The expression level of *TaFlo2-A1b* was higher than that of *TaFlo2-A1a* at all tested stages but significantly differed only in flag leaves and developing grains at 5 DAF (Fig. [4a](#Fig4){ref-type="fig"}). Furthermore, the expression level was positively correlated in Chinese Spring and three cultivars (Pubing3228, Shannong23, and Zhengmai9405) in developing grains sampled at 5 DAF. The expression level was lowest in Chinese Spring (TGW, 21.3 g) and highest in the cultivar Zhengmai9405 (TGW, 64.1 g) (Fig. [4b](#Fig4){ref-type="fig"}). All these cultivars contained the positive haplotype *TaFlo2-A1b*. Together, these results suggested that the relative expression level of *TaFlo2-A1* was highest in flag leaves but started to decrease in developing grains. However, the expression in developing grains at 5 DAF was positively correlated with TGW in cultivars carrying the positive haplotype.Fig. 4**a** Expression of *TaFlo2-A1* in flag leaves and developing grains. **b** Expression level of *TaFlo2-A1b* in cultivars with different TGW values at 5 DAF
Discussion {#Sec15}
==========
Rice-wheat synteny-based gene cloning in wheat {#Sec16}
----------------------------------------------
The rice *OSFLO2* orthologs *TaFLO2-A1*, *TaFLO2-B1* and *TaFLO2-D1* were cloned, characterized and found to be located on homeologous chromosome group 2 in wheat. Sequence polymorphism observed in the promoter region of *TaFlo2-A1* was associated with TGW. Thus, *TaFLO2-A1* is a yield-related gene, and its manipulation could be useful for improving the grain yield potential of bread wheat. Many genes related to TGW and grain yield have been isolated and characterized in wheat using rice-wheat synteny \[[@CR15]\]. The success of rice-wheat orthology-based gene cloning in wheat is due to high nucleotide and amino acid similarity between the corresponding orthologous genes. For example, with their respective rice orthologs, *TaTGW6* has 71% nucleotide and 68% amino acid similarity \[[@CR16], [@CR17]\]; *TaGW2* has 98% nucleotide and \~ 87% amino acid similarity \[[@CR12]\]; *TaCKX6-D1* has 66% amino acid similarity \[[@CR22]\]; *TaGS-D1* has 75.5% cDNA and 72.2% amino acid similarity \[[@CR24]\]; and *TaGASR-A1* has 88% amino acid sequence similarity \[[@CR26]\]. These data provide a genetic framework for marker-assisted selection (MAS) to pyramid positive alleles for TGW and yield during cultivar development. However, there are still many important genes that have been characterized in rice that are not being used as template for cloning their orthologs in wheat, e.g., *OsTB1* \[[@CR36]\], *GW5* \[[@CR37]\], *GS5* \[[@CR38]\], *GW8* \[[@CR39]\], *GW7/GL7* \[[@CR40], [@CR41]\], and *OsAGSW1* \[[@CR42]\]. Thus, comparative genomics approaches between rice and wheat will remain useful in discovering orthologs of rice genes in wheat and will continue to enhance our understanding of the genetics of yield potential in wheat.
The *TaFLO2-A1* gene is related to TGW in wheat {#Sec17}
-----------------------------------------------
Map-based cloning using QTL mapping approaches is an important strategy to isolate loci and genes controlling genetic polymorphism \[[@CR43]\]. However, progress on map-based cloning in wheat has been relatively slow compared to that in rice, and very few QTLs have been subjected to fine mapping in order to isolate candidate genes, mainly due to the complexity and large genome size of wheat.
In our study, *TaFlo2-A1* was found to be associated with TGW and explained from 4.42% (in the Pakistani collection) to 8.37% (in the MCC) of phenotypic variation. The TGW-related QTL identified on 2AL includes '*Xgwm339*-*Xbarc311*' in 139 RILs between two hard red spring wheat lines \[[@CR44]\]; *QTgw.ipk*-*2A* (*Xgwm372*) in 111 BC~2~F~3~ lines derived from the cross 'Flair × XX86' \[[@CR45]\]; *QGwt.crc-2A* (*Xgwm558-Xgwm294*) in a double-haploid population generated from the cross 'RL4452 × AC Domain' \[[@CR46]\]; *QTkw.sdau-2A* (*Xwmc181a*-*Xubc840c*) in 131 RILs derived from 'Chuan 35050' × 'Shannong 483' \[[@CR13]\]; *QSZ.uaf-2A.1* (*Xwmc455*) in natural populations of 108 CIMMYT and Pakistani spring wheat accessions \[[@CR47]\]; and *QTkw.hwwgr-2AL* (*Xgwm312*--IWA6090) in 127 RILs derived from 'Ning7840' × 'Clark' \[[@CR42]\]. Furthermore, three QTL on 2AL that were stable across five trials were detected in the same MCC (262) used in present study \[[@CR29]\].
These QTL on 2AL are located between *Xgwm71.2*/*Xgwm558* and *Xgwm294*, with an interval of 22 cM according to the consensus map of Somers \[[@CR48]\] or 16.1 cM according to the ITMI map (<http://wheat.pw.usda.gov/ggpages/SSRclub/GeneticPhysical/>). From this TGW-QTL-rich region, only one gene, *TaCwi-A1*, has been isolated thus far between the *Xgwm 71.2* and *Xbarc15* deletion bin 'C-2AL1--0.85', which is adjacent to the centromere \[[@CR18]\]. By integrating the information from the ITMI (<http://wheat.pw.usda.gov/ggpages/SSRclub/GeneticPhysical/>) and 'Yu 8679 × Jing 411' SSR + SNP \[[@CR49]\] maps, the location of *TaFlo2-A1* was inferred on deletion bin '2AL1--0.85-1.00'. Hence, the *TaFlo2-A1* is a TGW-related gene located on the distal deletion bin of chromosome 2AL, and the molecular marker 'TaFlo2-InDel8' is an addition to the kit of wheat breeders for marker-assisted selection.
Relationship between TGW and GpS {#Sec18}
--------------------------------
The relationship between the number of grains per spike (GpS) and the TGW was traditionally found as being negatively correlated \[[@CR1]--[@CR3]\]. However, the simultaneous selection of favored haplotypes for one plus neutral ones for the other or otherwise favored haplotypes for both traits has changed the correlations from negative to neutral or even positive \[[@CR4]\]. Therefore, no significant correlation was observed between TGW and GpS in the collections of Chinese landraces \[[@CR4]\], French winter wheat cultivars \[[@CR5]\] or CIMMYT-derived spring cultivars and lines \[[@CR6], [@CR50]\], while significantly positive correlations were observed in Chinese modern cultivars \[[@CR4]\].
In many genome-wide association studies (GWAS) for TGW and GpS, many loci were found to be associated with only one of the traits \[[@CR4], [@CR29], [@CR50]\]. The favored haplotypes at these loci should increase the phenotypic value of the one trait without negatively affecting the phenotypic value of the others. Thus, selection of such QTL was likely a major factor in changing the relationship between TGW and GpS over time. Similarly, selection of the favored haplotype (*TaFlo2-A1b*) identified in this study would help to increase TGW without reducing the average GpS in wheat.
Effect and putative mechanism of *TaFlo2-A1* in the determination of TGW {#Sec19}
------------------------------------------------------------------------
*TaFlo2-A1*, which is represented by two haplotypes in our study, was found to be significantly associated with TGW. Polymorphisms of an 8-bp InDel in the promoter and of five SNPs in the first intron were observed in *TaFlo2-A1*. The orthologs *TaFlo2-B1* and *TaFlo2-D1* lacked sequence variations associated with TGW (Fig. [2b](#Fig2){ref-type="fig"}). The association analysis of the Chinese Micro Core Collection (MCC) and Pakistani accessions indicated that *TaFlo2-A1b* was the superior haplotype for TGW. Nevertheless, some accessions that contained *TaFlo2-A1a* also had high TGW. This is mainly because the effect of *TaFlo2-A1* is likely masked by other genes associated with grain size \[[@CR20]\].
In wheat, *TaFlo2-A1* consists of 23 exons that encode 1734 amino acids with four TPR motifs at the positions of 947 to 988, 1032 to 1072, 944 to 1017 and 1028 to 1106 amino acid residues. Furthermore, three mitochondrial CLU domains were also observed at 737--878, 50--162 and 357--401 amino acid residues (Fig. [1a](#Fig1){ref-type="fig"}). The rice *OsFlo2* gene also consists of 23 exons and encodes 1720 amino acids with three TPR motifs at the positions of 933--966, 975--1008, and 1017--1050 amino acid residues \[[@CR27]\]. However, no mitochondrial CLU was reported in rice FLO2 by She et al. \[[@CR27]\]. To confirm the absence of mitochondrial CLU in rice FLO2, we BLASTed the rice FLO2 protein (accession: CAE03171) in an NCBI CD search. The results of the rice FLO2 protein (accession: CAE03171) query using the NCBI CD system showed the presence of two mitochondrial CLU domains at the intervals of 52--124 and 721--863 and four TPRs at the intervals of 932--973, 1017--1057, 929--1002 and 1013--1091 amino acid residues (Fig. [1a](#Fig1){ref-type="fig"}). Thus, the prediction of wheat and rice FLO2 protein structure using NCBI CD indicates high similarity between their structures.
Flo2 is considered to be a member of a conserved gene family in plants \[[@CR27]\]. *TaFlo2-A1* is abundantly expressed in flag leaves and in developing grains at 5 DAF stage, and the expression level of the positive haplotype (*TaFlo2-A1b*) was higher than that of the negative haplotype (*TaFlo2-A1a*). The phenomenon of higher expression of the positive *TaFlo2* haplotype is consistent with the results for rice *OsFlo2*, in which the overexpression of the positive haplotype significantly increases grain size \[[@CR27]\]. In rice, the *flo2* mutation in the promoter and in the open reading frame hinders the expression of genes involved in the synthesis of starch and protein \[[@CR27], [@CR28]\]. In rice cultivars that have different genetic backgrounds, some *flo2* mutations negatively affect grain quality attributes such as amylose content, grain appearance and physiochemical properties despite maintaining or increasing grain size \[[@CR27], [@CR28]\]. Based on these similarities between *OsFlo2* and *TaFlo2-A1* at the sequence, structure and expression levels, the 8-bp InDel mutation in the *TaFlo2-A1* promoter likely regulates grain size by affecting the expression of genes involved in the synthesis of starch and protein in wheat grains. Therefore, the increased expression of *TaFlo2-A1* has a positive effect on grain yield but may have a negative effect on some grain quality attributes in wheat, which shall need further investigations.
The newly developed molecular marker 'TaFlo2-InDel8' is an addition to the kit of wheat breeder for marker-assisted selection. This study likely lead to additional investigations to unveil the exact regulatory mechanism of the *Flo2* gene in wheat.
Conclusions {#Sec20}
===========
The *Flo2* orthologs in hexaploid wheat were cloned, and *TaFlo2-A1* was found to be associated with TGW but not with grain number per spike (GpS) in both the MCC and Pakistani wheat collections. The frequency of *TaFlo2-A1b* (positive haplotype) was low in commercial wheat cultivars; thus this haplotype can be selected to improve grain weight. This study likely lead to additional investigations to understand the regulatory mechanism of the *Flo2* gene in hexaploid wheat. The newly developed molecular marker 'TaFlo2-InDel8' could be incorporated into the kit of wheat breeders for use in marker-assisted selection.
Additional file {#Sec21}
===============
Additional file 1: Figure S1.a. Part of *TaFlo2A1* sequence in eight accessions. The first four sequences are of accessions with low TGW and other four sequences are of accessions with high TGW. b. Part of *TaFlo2-B1* sequence in eight accessions. The first four sequences are of accessions with low TGW and other four sequences are of accessions with high TGW. c. Part of *TaFlo2-D1* sequence in eight accessions. The first four sequences are of accessions with low TGW and other four sequences are of accessions with high TGW. (PDF 377 kb)
Aa
: Amino acid
ANOVA
: Analysis of variance
Bp
: Base pair
CIMMYT
: the International Maize and Wheat Improvement Center
CS
: Chinese Spring
CTAB
: Cetyl trimethylammonium bromide
DAF
: Days after flowering
Flo2
: Floury endosperm2
GpS
: Grain number per spike
GYS
: Grain yield per spike
InDel
: Insertion and deletion
ITMI
: International Triticeae Mapping Initiative
MCC
: Micro Core Collection
NCBI
: National Center for Biotechnology Information
QTL
: Quantitative trait loci
RIL
: Recombinant inbred line
TGW
: Thousand grain weight
TPR
: Tetratricopeptide repeat
URGI
: Unité de Recherche Génomique Info
**Electronic supplementary material**
The online version of this article (10.1186/s12870-017-1114-3) contains supplementary material, which is available to authorized users.
We thank two anonymous reviewers for their critical review and suggestions.
Funding {#FPar1}
=======
This research was financially supported by National Natural Science Foundation of China (31571643) and the Chinese Academy of Sciences (XDA08010104).
Availability of data and materials {#FPar2}
==================================
All the data supporting our findings is contained within the manuscript.
MS and XM carried out most of the experiments. MS wrote the manuscript. MS performed quantitative RT-PCR and analyzed the data. SHK and WY grew the plant samples, YS collected all the phenotypic data, and AM and DL conceptualized the experiments and revised the manuscript. All authors read and approved the final manuscript.
Ethics approval and consent to participate {#FPar3}
==========================================
Not applicable.
Consent for publication {#FPar4}
=======================
Not applicable.
Competing interests {#FPar5}
===================
The authors declare that they have no competing interests.
Publisher's Note {#FPar6}
================
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
| {
"pile_set_name": "PubMed Central"
} |
{#sp1 .245}
{#sp2 .246}
| {
"pile_set_name": "PubMed Central"
} |
Introduction {#S0001}
============
Dry eye disease (DED) is a multifactorial disease of the ocular surface characterized by a loss of homeostasis of the tear film, and accompanied by ocular symptoms.[@CIT0001] Being the most common reason for seeking medical eye care, DED is a common public health problem, with the prevalence ranging from approximately 5--50% worldwide.[@CIT0002]--[@CIT0004] Although the pathogenesis of dry eye disease is not fully understood, it is recognized that inflammation has a prominent role in the development and propagation of this debilitating condition. DED inflammation disrupts the normal homeostasis of the ocular surface, resulting in a vicious circle of ocular surface injury, and anti-inflammatory and immunosuppressive drugs are considered the pharmacologic agents of choice for controlling the inflammatory cascade of this disease.[@CIT0005]--[@CIT0007] Even so, there is still a great potential for the development in anti-inflammatory and damage-repair treatments of dry eye disease.
Vitamin B12 (VB12) is a dietary essential nutrient and is important for metabolic functions of the nervous system, whose deficiency is associated with an impairment of sensory innervation, and may cause optic neuropathy, eye movement disorders, and corneal epitheliopathy with decreased vision and photophobia.[@CIT0008]--[@CIT0010] Recently, several studies have reported the parenteral and topical use of VB12 for ocular pain and symptoms in dry eye treatment.[@CIT0011],[@CIT0012]
Oxytocin (OXT), a nonapeptide produced in the hypothalamus, exerts a wide spectrum of central and peripheral effects. Nasal spray of OXT, which has been proved to be safe in children,[@CIT0013] is currently suggested as a potential treatment for various psychopathologies of social cognition and behavior in humans.[@CIT0014] In addition to its reproduction and psychiatry-related functions, OXT has also been proved to display a potent anti-inflammation and wound-healing effect in various inflammatory disease,[@CIT0015],[@CIT0016] as well as an analgesic and nerve healing effect in animal models.[@CIT0017],[@CIT0018] However, the anti-inflammatory and nerve healing effect of OXT has yet to be applied in ocular diseases.
Nebulization, applied mainly in pneumology and rhinology diseases, is a widely used means of drug delivery to the upper and lower airways. Nebulizers are devices that convert a drug liquid in solution or suspension into small droplets.[@CIT0019] Its theoretic advantage over classic means of delivery is that, as a non-invasive painless treating system, it delivers drug and humidifies the target region simultaneously, enables dose modification and dose compounding, and directly reaches the target organ, which avoids systemic side-effects and enhancing local efficacy.[@CIT0020] To date, no applications of nebulization therapy havebeen reported in ophthalmic diseases.
In the light of the facts mentioned above, we developed a new ocular nebulization therapy delivering vitamin B12 and oxytocin separately for the treatment of DED. The main objective of this study was to estimate the efficacy of ocular nebulization in dry eye patients, as well as to compare the characteristic of VB12 and OXT aerosols using both clinical and in vivo confocal microscopy (IVCM) evaluations.
Materials and methods {#S0002}
=====================
Study design and participants {#S0002-S2001}
-----------------------------
This prospective cohort study included 38 patients with DED recruited from the outpatient department of the Department of Ophthalmology at Peking University Third Hospital between August 2017 and June 2018. The study protocol was approved by the Human Research and Ethics Committee of Peking University Third Hospital, and the research was adhered to the tenets of the Declaration of Helsinki. Written informed consent was obtained from each participant before enrollment.
Inclusion criteria included adult patients 1) with a diagnosis of DED in accordance with the criteria proposed by the Tear Film and Ocular Surface Society (TFOS) Dry Eye Workshop (DEWS);[@CIT0001] 2) willing to cooperate during the examination and treatment procedure; and 3) with the ability to participate in follow-up examinations for at least 3 months during nebulization treatment. Exclusion criteria consisted of the following: a) patients with active allergies, infections, or inflammatory diseases of the ocular surface unrelated to dry eye within 6 months; b) patients who currently used treatments for DED (other than artificial tears); c) patients who took oral neurotrophic and anti-inflammatory drugs in the past 3 months; d) a history of ocular trauma or surgery within 1 year; e) patients with alterations in the lacrimal drainage system such as punctal occlusion; f) patients who used contact lenses within the past month; g) patients with uncontrolled systemic diseases; and h) pregnant or nursing mothers.
Clinical evaluation {#S0002-S2002}
-------------------
Each participant had three clinic visits: before starting the nebulization therapy (baseline), 1 month after treatment, and 3 months after starting treatment. Both the patients and the observers were blinded of the nebulization group throughout the process. During each visit, all participants had a complete masked ophthalmic evaluation conducted in the following order: evaluation of symptoms using the Ocular Surface Disease Index (OSDI) questionnaire, along with a self-assessment of light sensitivity and dryness (each on a scale of 0--4, with 0 for never and 4 for severe), a complete examination of the ocular surface of both eyes, including tear meniscus height (TMH),[@CIT0021] tear break-up time (BUT),[@CIT0001] and corneal fluorescein staining (CFS) evaluated using the Van Bijsterveld scale,[@CIT0022] and IVCM analysis of the central cornea. Additionally, adverse events during the treatment, including pain, pressure, dizziness, blurred vision, and abdominal discomfort, were recorded during the 3-month follow-up.
TMH {#S0003}
===
The central lower TMH were measured using a slit lamp microscope (with a graticule in 0.05 mm units).[@CIT0021] Three consecutive readings were obtained, and the final results are presented as medians.
BUT {#S0004}
===
A total of 5 mL of 2μ sodium fluorescein was instilled onto the bulbar conjunctiva using a micropipette, without inducing reflex tearing. The patient was asked to blink naturally without squeezing three to five times and was then asked to stare straight ahead without blinking under the cobalt blue light until he or she received other instructions. A stopwatch was used to record the time between the last complete blink and the first appearance of a dry spot or disruption in the tear film.[@CIT0001] The procedure was repeated three times, and the final score is presented as an average value.
Corneal fluorescein staining {#S0005}
============================
For the corneal staining evaluation, the cornea was divided into four sectors. Each sector was graded from 1 to 3 using the following criteria: 1 representing few separated spots; 2 representing many separated spots; and 3 representing confluent spots.[@CIT0022]
In vivo confocal microscopy {#S0006}
===========================
All participants underwent laser IVCM of the central cornea in both eyes by a masked investigator (XL) using Heidelberg Retina Tomograph 3 with the Rostock Cornea Module (Heidelberg Engineering, Heidelberg, Germany). Each image represents a coronal section of 384 × 384 pixels which is equivalent to 400 × 400 μm of the cornea. Before each examination, a drop of 0.4μ oxybuprocaine chlorohydrate (Santen, Osaka, Japan) was instilled into the lower conjunctival fornix. The examination was conducted approximately at the corneal apex.
For each eye, 1 image of basal epithelial layer, and 3 most representative images of the sub-epithelial layer were selected by a single masked investigator (XL). Three independent observers (JF, YL and YZ) measured each image, among which the mean value of series of the 3 sub-epithelial layer readings was recorded, and the average of the measurements by the observers was used for further analysis.
Evaluated parameters included the following:Basal epithelial cell density (cells/mm^2^): basal epithelial cells were immediately above the Bowman's layer and identified by the lack of a visible nucleus, polygonal cell shape, hyper reflective cell borders and a relatively higher density of cells per frame.[@CIT0023] The counting was carried out within a region of interest of standardized size (region of interest \[ROI\] = 100 × 100 μm) under a manual cell counting procedure conducted using ImageJ software (<http://imagej.nih.gov/ij/)>, in which cells that were partially within the area analyzed were counted only along the left and upper margins.[@CIT0024]Dendritic cell (DC) density (cells/mm^2^): Dendritic cells were morphologically identified as bright individual dendritiform structures with cell bodies,[@CIT0025] which was manually counted in each corneal section at the level of subbasal nerves, selecting the images in which there was a greater density of DCs. The counting was carried out using the manual cell counting procedure same as basal epithelial cell density in the ROI of 400×400 μm.Nerve tortuosity: the grade of nerve tortuosity was classified in four grades according to the tortuosity grading scale reported previously.[@CIT0026]Sub-basal nerve density (mm/mm^2^): defined as the total length of all nerve fibers within a frame (400×400 μm). It was traced and calculated using NeuronJ (<http://www.imagescience.org/meijering/software/neuronj/>), a semi-automated nerve analysis plug-in of ImageJ.[@CIT0027]
To note, change of variables (Δ) at each follow-up was defined as the value at a certain follow-up minus that at baseline (eg, ΔOSDI at 1 month=OSDI at 1 month-OSDI at baseline).
Nebulization treatment {#S0006-S2001}
----------------------
Recruited subjects under the age of 50 were assigned to the VB12 group due to the possible effects of OXT on reproductive age patients, and others were randomly assigned to receive 4 mg (28 mL) of VB12 or 30 International Unit (23 mL) of OXT[@CIT0028] (both as saline solution) nebulization twice weekly for 3months with a VGR-001 Ultrasonic Nebulizer (INSPIRED, Guangdong, China) and an eye mask (3M, St. Paul, MN, USA). This nebulization system enabled room-temperature nebulization at the rate of 1 mL/min, providing particles with a diameter of 5±1.25 μm. Nebulization was terminated when the nebulizer chamber was empty, usually within 10 minutes. A photo of the ultrasonic nebulization system used in this study is shown in [Figure 1](#F0001){ref-type="fig"}.Figure 1Ultrasonic nebulization system developed in this study.
For drug conservation, VB12 (Kingyork, Tianjin, China) was preserved at room temperature, and OXT (Harvest, Shanghai, China) was stored at 4°C due to its poor thermal stability. After being taken out of the refrigerator, OXT was used for nebulization within 30 minutes. Altogether, two nebulization systems were used following routine sterilization protocol and cleaning of equipment prior to each use.
Statistical analysis {#S0006-S2002}
--------------------
For variables other than the three symptom indicators (OSDI, light sensitivity, and dryness), patient's data for both eyes were calculated (n=76), while only one value was applied for analysis for these three factors (n=38).
Statistical analyses were conducted using SPSS software version 22.0 (SPSS, Inc., Chicago, IL, USA). The basic characteristics of the two treatment groups were described using descriptive statistics including means and standard deviations for continuous variables, and frequencies and proportions for categorical variables. An independent *t*-test was used to analyze the continuously numeric variables, and a chi-square test was applied in classified variables in the comparison of basic characteristics between groups. The paired *t*-test was applied to assess the differences of all variables at each follow-up from baseline in the two groups separately. Furthermore, an independent *t*-test was implemented to determine differences in all variables at each time point, as well as their changes from baseline between the two groups. Two-sided *P*-values \<0.05 were considered statistically significant for all comparisons.
Results {#S0007}
=======
Basic characteristics {#S0007-S2001}
---------------------
A total of 38 DED patients were enrolled, with 19 patients in the VB12 group and 19 in the OXT group, and both eyes of these patients were evaluated. Of these subjects, 18 patients in the VB12 group and 17 patients in the OXT group finished the follow-up at 1 month after treatment, and eventually 14 patients in the VB12 group and 15 patients in the OXT group completed the 3-month follow-up. As shown in [Table 1](#T0001){ref-type="table"}, patient characteristics were matched between VB12 and OXT at baseline, except for a higher age in the OXT group. No adverse events were observed during the 3-month therapy, either in VB12 or OXT group.Table 1Clinical characteristics and IVCM results at baseline in VB12 and OXT groupsObjectivesVB12 (n=19)OXT (n=19)*P*-valueDED history, months17.2±26.224.3±25.20.262Age, years59.7±16.966.5±11.90.045\*Sex, n (%) Male4 (21.1%)1 (5.3%) Female15 (78.9%)18 (54.7%)0.150OSDI46.6±21.549.1±19.00.705Dryness3.5±0.63.7±0.60.285Light sensitivity2.2±1.02.4±0.90.502TMH, mm0.12±0.090.14±0.080.315BUT, seconds3.03±1.003.07±1.760.911CFS0.7±1.11.4±2.30.121Basal epithelial cell density, cells/mm^2^3685.7±843.13888.2±873.00.360Dendritic cell density, cells/mm^2^41.3±37.954.1±39.70.191Nerve tortuosity2.7±0.82.7±0.50.875Sub-basal nerve density, mm/mm^2^14.94±5.9613.48±4.730.272[^2]
Clinical and IVCM results at each follow-up {#S0008-S2002}
-------------------------------------------
Patients receiving VB12 nebulization, as shown in [Table 2](#T0002){ref-type="table"}, achieved significant improvements in clinical and IVCM parameters in the first month of treatment, except for DC density and nerve tortuosity. Significant differences were demonstrated from 1 to 3 months of nebulization therapy in OSDI, dryness, BUT, basal epithelial cell density and, notably, DC density. Overall, there were statistically significant improvements in all signs and symptoms of DED, except for nerve tortuosity during the 3-month treatment.Table 2Clinical and IVCM results at each follow-up in the VB12 groupObjectivesBaseline1 month3 months*P*~1~*P*~2~*P*~3~OSDI46.6±21.534.8±17.328.7±13.30.001\*\*0.005\*\*0.002\*\*Dryness3.5±0.62.8±0.52.1±0.8\<0.001\*\*0.005\*\*\<0.001\*\*Light sensitivity2.2±1.01.3±0.80.7±0.90.001\*\*0.0530.002\*\*TMH, mm0.12±0.090.22±0.070.23±0.07\<0.001\*\*0.123\<0.001\*\*BUT, seconds3.03±1.004.19±1.425.48±1.01\<0.001\*\*0.002\*\*\<0.001\*\*CFS0.7±1.10.1±0.300.002\*\*0.3310.012\*Basal epithelial cell density, cells/mm^2^3,685.7±843.14,526.4±562.54,978.3±717.1\<0.001\*\*0.005\*\*\<0.001\*\*Dendritic cell density, cells/mm^2^41.3±37.939.8±38.317.9±12.80.5550.015\*0.033\*Nerve tortuosity2.7±0.82.7±0.62.7±0.40.2620.3310.437Sub-basal nerve density, mm/mm^2^14.94±5.9619.04±6.0420.11±5.590.026\*0.093\<0.001\*\*[^3]
[Table 3](#T0003){ref-type="table"} shows the results of subjects treated with OXT, among which all clinical and IVCM variables improved significantly within 1 month, except for nerve tortuosity. There were significant improvements in OSDI, TMH, BUT, and sub-basal nerve density at 3 months compared to that of 1 month, whereas nerve tortuosity appeared to be increased. Apart from BUT and nerve tortuosity, clinical and IVCM results demonstrated significant improvements at 3 months from baseline.Table 3Clinical and IVCM results at each follow-up in the OXT groupObjectivesBaseline1 month3 months*P*~1~*P*~2~*P*~3~OSDI49.1±19.036.1±18.722.9±14.1\<0.001\*\*0.033\*0.002\*\*Dryness3.7±0.62.4±0.92.1±1.0\<0.001\*\*0.0510.001\*\*Light sensitivity2.4±0.91.4±0.91.2±1.0\<0.001\*\*0.1040.004\*\*TMH, mm0.14±0.080.19±0.050.23±0.05\<0.001\*\*0.003\*\*0.009\*\*BUT, s3.07±1.763.40±1.254.33±1.040.002\*\*0.018\*0.231CFS1.4±2.30.5±1.200.003\*\*0.6180.039\*Basal epithelial cell density, cells/mm^2^3,888.2±873.04,709.3±777.64,364.3±981.1\<0.001\*\*0.6440.020\*Dendritic cell density, cells/mm^2^54.1±39.718.8±18.526.1±26.8\<0.001\*\*0.4170.001\*\*Nerve tortuosity2.7±0.52.5±0.63.0±0.60.0630.017\*0.060Sub-basal nerve density, mm/mm^2^13.48±4.7315.87±6.4818.92±3.110.003\*\*0.014\*\<0.001\*\*[^4]
[Figure 2](#F0002){ref-type="fig"} represents the IVCM change at each follow-up in two groups. Both VB12 and OXT nebulization induced improvement in DED conditions. [Figures 2A](#F0002){ref-type="fig"} and [G](#F0002){ref-type="fig"} represent the corneal basal epithelial layer at baseline in VB12 and OXT groups, with [Figures 2B](#F0002){ref-type="fig"}, [C](#F0002){ref-type="fig"}, [H](#F0002){ref-type="fig"}, and [I](#F0002){ref-type="fig"} showing increased basal epithelial cell density at 1 month and 3 months after treatment. Loss of nerves and multiple DCs of dry eye patients can be seen in [Figures 2D](#F0002){ref-type="fig"} and [J](#F0002){ref-type="fig"}, while [Figures 2E](#F0002){ref-type="fig"}, [F](#F0002){ref-type="fig"}, [K](#F0002){ref-type="fig"}, and [L](#F0002){ref-type="fig"} demonstrate elevated sub-basal nerve density, and decreased DC density throughout the 3-month treatment. Figure 2IVCM images at each follow-up between VB12 and OXT groups.**Notes:** Frames show basal epithelial cell with bright polygonal borders and nonhomogenous cytoplasmic reflectivity immediately anterior to Bowman layer, sub-basal nerve plexus fibers running roughly in parallel and dendritic bright cells (DC) with different densities (arrows). VB12 group: (**A**--**C**) basal epithelial cell layer at baseline, 1 month, and 3 months after treatment: showed an increased basal epithelial cell density. (**D**--**F**) DC and sub-basal nerve plexus at baseline, 1 month, and 3 months after treatment: showed a decreased DC density and a notably increased sub-basal nerve density. OXT group: (**G**--**I**) basal epithelial cell layer at baseline, 1 month, and 3 months after treatment: showed an increased basal epithelial cell density. (**J**--**L**) DC and sub-basal nerve plexus at baseline, 1 month, and 3 months after treatment: showed a notably decreased DC density and increased sub-basal nerve density and tortuosity.**Abbreviations:** OXT, oxytocin; IVCM, in vivo confocal microscopy; VB12, vitamin B12.
Differences of clinical and IVCM results between VB12 and OXT groups {#S0008-S2003}
--------------------------------------------------------------------
The comparison of clinical and IVCM data between the VB12 and OXT groups is shown in [Figure 3](#F0003){ref-type="fig"}. Although obvious improvements were detected in OSDI, dryness, light sensitivity, TMH, CFS, and sub-basal nerve density in both groups during the ocular nebulization process, no significant difference was evident between the two treatment groups. TheVB12 group revealed a significantly higher BUT at 1 month and 3 months (*P*=0.041 and *P*=0.005, respectively), along with a higher basal epithelial cell density at 3 months (*P*=0.043) compared with the OXT group. Asignificant lower DC density was observed in the OXT group after 1 month of nebulization treatment (*P*=0.021).Figure 3Clinical and IVCM results at each follow-up between VB12 and OXT groups.**Notes:** \**P*\<0.05 between groups. \*\**P*\<0.01 between groups.**Abbreviations:** BUT, tear break-up time; IVCM, in vivo confocal microscopy; OSDI, Ocular Surface Disease Index; OXT, oxytocin; S, seconds; TMH, tear meniscus height; VB12, vitamin B12.
[Figure 4](#F0004){ref-type="fig"} describes the changes in clinical parameters and IVCM results of each follow-up from baseline between the two groups. Although no difference was found at 1 month (*P*=0.723), Δbasal epithelial cell density in the VB12 group was significantly higher at 3 months (*P*=0.022). Moreover, patients treated with OXT nebulization showed a significantly higher DC reduction at 1 month (*P*\<0.001) compared to the VB12 group, with no significant difference at 3 months between groups (*P*=0.136). As for Δnerve tortuosity, OXT treatment induced a significantly greater reduction at 1 month (*P*=0.049), in contrast to a notable increase at 3 months (*P*=0.040), corresponding to the nerve tortuosity increase from 1 to 3 months mentioned above.Figure 4Changes in clinical and IVCM results at each follow-up between VB12 and OXT groups.**Notes:** \**P*\<0.05 between groups. \*\**P*\<0.01 between groups.**Abbreviations:** BUT, tear break-up time; IVCM, in vivo confocal microscopy; OSDI, Ocular Surface Disease Index; OXT, oxytocin; S, seconds; TMH, tear meniscus height; VB12, vitamin B12.
Discussion {#S0009}
==========
This study proposed a new approach of ocular nebulization for the treatment of dry eye disease, which appeared to be effectively proven by clinical and in vivo confocal evaluations. Two drugs, vitamin B12 and oxytocin, were tested, and both proved to be efficient in relieving DED symptoms and signs with several differences in certain aspects.
Nebulizers have long played an important role in the treatment of diverse diseases. Based on the aqueous-deficiency and inflammation-related nature of DED, we considered nebulization on the ocular surface to be a feasible therapy and, in combination with VB12 and OXT, might achieve a remarkable relief in signs and symptoms of DED. This is the first research, to the best of our knowledge, describing a nebulization therapy in the treatment of dry eye disease, as well as the first attempt of oxytocin application on ocular surface.
In this study, our objective was to demonstrate the changes occurring in the ocular surface after nebulization treatment, using clinical evaluations and IVCM study. As it is increasingly used in in vivo study, IVCM has been proven to be capable of identifying different corneal structural aspects that are considered pathognomonic in DED,[@CIT0029] including a reduced number of corneal epithelial cells, increased number of dendritic cells, reduced sub-basal nerve density, and increased nerve tortuosity.[@CIT0030]
The neurotrophic effect of vitamin B12 has long been known in clinical practice.[@CIT0011],[@CIT0012] Recently, Macri et al[@CIT0031] indicated that VB12 eye drops improved oxidative stress as well as symptoms of dry eye. Hence, in our study, it is noteworthy that DED patients responded well as the clinical signs and symptoms improved notably within 1 month after the initiation of VB12 nebulization, accompanied by a significant improvement in IVCM results ([Table 2](#T0002){ref-type="table"}). Previous studies have shown that VB12 is able to promote corneal re-innervation and re-epithelization after mechanical injury in a rat model.[@CIT0032] In agreement with this, our study revealed a continuous increase in basal epithelial cell density and sub-basal nerve density at 1 month and 3 months after VB12 nebulization. This might be explained by the evidence that vitamin B12 plays an important role in the enhancement of β-III tubulin expression in neurons,[@CIT0033] in addition to the regulation of neurotrophic factors synthesis, such as an increased expression of TNF-α and decreased neurotrophic epidermal growth factor (NEGF),[@CIT0034] which support neurite outgrowth and survival and, in turn, promotes epithelial wound healing.[@CIT0035]
Similar to VB12, our study revealed a significant effect of OXT nebulization with regard to DED symptoms and signs ([Table 3](#T0003){ref-type="table"}). The mechanism underlying this finding is not yet fully understood, and we assumed that there could be four mechanisms of OXT contributing to the DED healing process. First, as has been well described in numerous studies, OXT treatment displays an anti-inflammatory effect by i) stimulation of nitric oxide release, leading to inhibition of the adhesion and aggregation of neutrophil leukocytes;[@CIT0036] ii) decreased release of IL-6 and increased release of prostacyclin, inhibiting platelet aggregation, which has been suggested to promote the inflammation process;[@CIT0037],[@CIT0038] iii) decreased M1 macrophages differentiation with no effect on the M2 population, thereby leading to an anti-inflammatory phenotype;[@CIT0039] and iv) increasing corticosterone levels acutely, which is capable of inhibiting neutrophil extravasation in response to different stimuli.[@CIT0040] This was consistent with our results, as DC density rises rapidly at 1 month after the application of OXT. The second possible role of OXT, described in a rat sciatic nerve damage model,[@CIT0017] is that it resulted in an accelerated nerve recovery, probably by increasing nerve growth factor (NGF) and IGF-1, which enhance nerve regeneration and promote axonal growth rate.[@CIT0041],[@CIT0042] In our study, we believed the increased sub-basal nerve density after treatment to be related to this nerve-healing effect of OXT. Another finding was the relieved dry eye symptoms including dryness and light sensitivity, which could probably be explained by a potential relationship between OXT and pain due to an activation of OXT receptors located superficially in the dorsal horn, a possible combination of OXT with opioid receptors, and a decrease of pain sensitivity by improving mood, relieving anxiety, and mitigating the stress response.[@CIT0018] The fourth mechanism is that OXT mediates a myoepithelial cell-driven acini contraction that contributes directly to the function of the lacrimal gland,[@CIT0043] corresponding to our results of TMH and BUT elevation during the OXT nebulization process.
In the comparison of VB12 and OXT nebulization, one might notice the mismatch of age at baseline, with a range of 27--81 years old in the VB12 group in contrast to 50--87 years old in the OXT group. This was due to a physiological consideration in the central and peripheral actions of oxytocin, including parturition, lactation, maternal behavior, erectile dysfunction, and ejaculation.[@CIT0044] Although oxytocin is a small peptide drug with only nine amino acids, there was a possibility of systemic administration through ocular route.[@CIT0045] It is, therefore, necessary to avoid OXT application in patients of reproductive age due to ethical considerations.
After both were demonstrated to be efficient in the treatment of DED, the characteristic of VB12 and OXT nebulization were further compared due to their different mechanism. Our study did detect their differences in certain aspects ([Figures 3](#F0003){ref-type="fig"} and [4](#F0004){ref-type="fig"}). In this study, individuals treated with VB12 had a better improvement in BUT and basal epithelial cell density than those who received OXT nebulization. Tear film breakup occurs mainly as a result of tear film evaporation, and BUT represents stability of tear film,[@CIT0046] whose homeostasis is achieved reflexly by the lacrimal functional unit (LFU). LFU consists of the ocular surface, its secretory appendages, and the connecting innervation, where the trigeminal innervation of cornea, conjunctiva, and lid margins provides the afferent limb of the feedback loop whilst the secretomotor innervation of the lacrimal gland, meibomian glands, and the conjunctival goblet cells provides the efferent limb.[@CIT0047] We suspect that, due to its remarkable effect on corneal re-innervation, VB12 restores corneal sensitivity and, therefore, induces secretion of the ocular appendages,[@CIT0047],[@CIT0048] thereby leading to the restoration of LFU function and tear film stability. Also, corneal nerve regeneration, as mentioned above, is a key factor in the renewal of a normal epithelium during DED treatment. At the IVCM level, our study revealed a rapid reduction of DC density at first month along with a notable increase in nerve tortuosity at 3 months among OXT patients. This is probably the result of the previously described anti-inflammatory effect of OXT that leads to a quick decrease of DC, which was proved to have a critical function in activation of the immune system in the ocular surface.[@CIT0049]
Therefore, in addition to the promising effect of both nebulization therapies adopted in this study, drugs are recommended to be selected according to patient conditions. In this regard, we highlight the application of OXT nebulization for DED patients with inflammation-related signs and symptoms, and VB12 for those with severe epithelial damage as well as patients of reproductive age.
Our results contributed to open interesting prospects for applying ocular nebulization therapy in the treatment of DED. Key advantages of this new therapy are that: a) it enables continuous delivery of low doses of an aerosolized drug to its site of action for a relatively more stable effect; b) it is fully acceptable for DED patients due to its non-invasive and painless feature; c) it simultaneously achieves humidification, anti-inflammation, and neurotrophic effect in correspondence to the diverse etiology of DED, with a further potential of combining with other drugs available; d) it is a possible way of providing a home-based treatment for DED patients, although droplet sizes and the therapeutic effects of portable nebulizers are still being tested as a result of different mechanisms between commonly-used and marketed portable nebulizers.
This preliminary research has a few limitations. First, there was a mismatch of age between the two treatment groups, which had been explained earlier in the discussion. Second, this study is in the absence of a control sham group receiving, for example, nebulization with sodium chloride. Third, this was a single center research study in exploration of a new treating method of DED, and dry eye subtypes weren't classified due to the relatively small number of enrollees. Fourth, dose and frequency of treatment were certain in each nebulization group and further dose gradient this could be estimated in the determination of an optimal therapeutic regimen. In the inspiration of the proven efficacy, future studies should focus on an effect confirmation of nebulization in the delivery of multiple kinds of drug compared with artificial tears, as well as a sub-group analysis in a diverse type of DED patients, and the key mechanisms of ocular nebulization therapy on DED requires further investigation.
Conclusion {#S0010}
==========
In conclusion, this study described a new ocular nebulization therapy delivering vitamin B12 and oxytocin, which was demonstrated to be efficient for dry eye treatment. Both groups exhibited parallel improvements with respect to clinical and confocal parameters, with a relatively stronger effect of BUT elevation and epithelial repair in VB12 and DC reduction in OXT nebulization, which might mainly be explained by the neurotrophic effect of VB12 and anti-inflammatory effect of OXT.
This study was supported by the Capital's Funds for Health Improvement and Research under grant number 2018-2-4093 (Xuemin Li) and the Peking University Medicine Seed Fund for Interdisciplinary Research under grant number BMU2018MX016 (Weiqiang Qiu).
Disclosure {#S0011}
==========
The authors report no conflicts of interest in this work.
[^1]: These authors contributed equally to this work
[^2]: **Notes:** \**P*\<0.05 between groups.
**Abbreviations:** BUT, tear break-up time; CFS, corneal fluorescein staining; DED, dry eye disease; IVCM, in vivo confocal microscopy; n, number; OSDI, Ocular Surface Disease Index; OXT, oxytocin; TMH, tear meniscus height; VB12, vitamin B12.
[^3]: **Notes:** *P*~1~: *P*-value between baseline and 1 month; *P*~2~: *P*-value between 1 month and 3 months; *P*~3~: *P*-value between baseline and 3 months. \**P*\<0.05 between follow-ups. \*\**P*\<0.01 between follow-ups.
**Abbreviations:** BUT, tear break-up time; CFS, corneal fluorescein staining; DED, dry eye disease; IVCM, in vivo confocal microscopy; n, number; OSDI, Ocular Surface Disease Index; TMH, tear meniscus height; VB12, vitamin B12.
[^4]: **Notes:** *P*~1~: *P*-value between baseline and 1 month; *P*~2~:*P*-value between 1 month and 3 months; *P*~3~: *P*-value between baseline and 3 months. \**P*\<0.05 between follow-ups. \*\**P*\<0.01 between follow-ups.
**Abbreviations:** BUT, tear break-up time; CFS, corneal fluorescein staining; DED, dry eye disease; IVCM, in vivo confocal microscopy; n, number; OSDI, Ocular Surface Disease Index; OXT, oxytocin; TMH, tear meniscus height.
| {
"pile_set_name": "PubMed Central"
} |
Protocol
========
To guarantee high quality and reproducibility, we recommend the use of Standard Operating Procedures (SOP). In this video, published SOPs as developed and used in our laboratory are applied. ^1^
1. Middle Cerebral Artery Occlusion
-----------------------------------
1. Mice are anaesthetized with an appropriate anaesthetic *regime* in consult with veterinary staff. (E.g. induction with 1.5 - 2 % Isoflurane and maintainance with 1.0 - 1.5 % Isoflurane in 2/3 N2O and 1/3 O2 using a vaporizer). Body temperature of the mice is maintained at 36.5°C ± 0.5°C during surgery with a heating plate. A feedback controlled heating pad, which warms according to the rectal temperature of the mouse, is highly recommended.Disinfect the skin and surrounding fur with an appropriate agent (e.g. 70% ethyl alcohol) and dry it afterwards.
2. A midline neck incision is made and the soft tissues are pulled apart.
3. The left common carotid artery (LCCA) is carefully dissected free from the surrounding nerves (without harming the vagal nerve) and a ligature is made using 6.0/7.0 string. 5.0 string can also be used.
4. The left external carotid artery (LECA) is then separated and a second knot is made.
5. Next, the left internal carotid artery (LICA) is isolated and a knot is prepared with a 6.0 filament.
6. After obtaining good view of the left internal carotid artery (LICA) and the left pterygopalatine artery (LPA), both arteries are clipped, using a microvascular clip.
7. A small hole is cut in the LCCA before it bifurcates to the LECA and the LICA. A monofilament made of 8.0 nylon coated with silicon hardener mixture (see below) is then introduced into the LICA, until it stops at the clip. Attention has to be paid not to enter the occipital artery. (Figure 1)
8. The clipped arteries are opened while the filament is inserted into the LICA to occlude the origin of the LMCA in the circle of Willis.
9. The third knot on the LICA is closed to fix the filament in position.
10. The mice receive saline 0.5 mL subcutaneously as volume replenishment. For pain relief, Lidocaine gel is topically applied in the wound.
11. If reperfusion is intended, mice stay for 30 - 90 min occlusion in a heated cage, the wound could be closed with a small suture clip. Afterwards, a second anaesthesia is performed, the third knot on the ICA is momentary opened and the filament withdrawn.
12. The remaining sutures are shortened and the skin is adapted with a surgical suture.
13. All animals receive a second volume replenishment as described above.
14. Animals are put in a heated cage for two hours to control for body temperature. The animals must be checked daily after surgery for signs of discomfort. The mice could show some post surgical weight loss. They receive mashed food in a petri-dish placed on the floor to encourage eating. The food is replaced daily for seven days.
2. Sham Operation
-----------------
1. For sham operations the filament is inserted to occlude LMCA and withdrawn immediately to allow instant reperfusion (1.8). The subsequent operation is identical to that performed on the animals undergoing cerebral ischaemia (1.9 - 1.14) including a second anaesthesia.
3. Preparation of the Filament
------------------------------
1. Sterility of the filament should be considered. The use of sterilized equipment as well as an appropriate handling of the filament afterwards is a prerequisite for a sterile surgery. Disinfection of the filament is difficult, since many of the common sterilization methods may worsen the quality of the filament. However methods such as radiation, for example with UV light or γ rays, or chemical sterilization, for example with highly reactive gases such as ethylene oxide, are applicable.
2. 8.0 nylon filament is cut into lengths of 11 mm under the microscope
3. The filament tip must be coated completely and evenly over a length of 8 mm with a hardener mixture of Xantopren M Mucosa and Activator NF Optosil
4. Representative Results
-------------------------
Depending on the duration of blood flow restriction, different motor and behavioural deficits result. Both after 30 and 60 min of cerebral ischaemia, animals in most cases show decreased resistance to lateral push and circling due to disturbance in locomotion. Milder lesions manifest as a flexor position in the front limbs. These easily observable signs can be used as a basic score for the success of the operation.^2^
Morphologically the lesion can be assessed using either histology or Magnetic Resonance Imaging (MRI). Sixty minutes occlusion of the middle cerebral artery produces tissue pannecrosis in an area including both the striatum and the neocortex, whereas 30 minutes of ischaemia cause mainly neuronal cell death limited to the striatum. ^3^ ( Figure 2) In terms of infarct volume, we expect a standard deviation lower than 30% in a set of operations. Mortality depends on the occlusion time with around 5 % after 30 min of ischaemia and 10 - 20 % after 60 min.
Another minimal invasive possibility is the use of Laser Doppler flowmetry (LDF) during the operation, which allows a direct control of its success. In an individual animal, reduction to 10 - 20 % of preocclusion values clearly indicates successful induction of focal cerebral ischemia. ^4^ However, LDF cannot be used as a method for interindividual comparisons, since LDF can only measure quantitative changes (in percentages) of blood flow within a small and limited tissue sample volume. It does not give information about the spatial extent of blood flow reduction. ^5^
There are several tests to assess behavioural aspects after stroke, including gait analysis ^6,7^, Rotarod ^8^, Pole test ^9,10^, adhesive removal test ^11,12^, staircase test ^13,14^, ladder rung test^15,16^ and Morris water maze^17^. In all these tests, mice subjected to MCAo perform less successfully than control animals.
**Figure 1.** Scheme of the vessel architecture supplying the brain (depicted in background) in the mouse. Different strains may show variations, for example the occipital artery sometimes leaves from the internal carotid artery.
**Figure 2.** Schematic illustration of typical lesion sizes after different time points of reperfusion in the proximal MCAo model. In the middle, the typical course of functional activity and cerebral blood flow after MCAo are depicted. (MCAo: Middle cerebral artery occlusion, LDF: Laser Doppler flow measurement)
Discussion
==========
The model of transient, proximal MCA occlusion^18,19^ presented here mimics one of the most common types of ischaemic stroke in patients.^20^ Based on variable reperfusion times, the model offers different grades of damage ranging from transient ischaemic attack (TIA) to large infarcts including major parts of the ischemic hemisphere. This allows the researcher to study different pathophysiological mechanisms after stroke. ^20,21^
Surgery can be performed in a short time period and produces highly reproducible lesions. Nevertheless, this requires a thorough control of confounders.^22^ Small differences in operation technique may account for different effects on the infarct.^23,24^ Furthermore, due to variances in cerebral vascular anatomy, different mouse strains show a different outcome.^25,26^ Body temperature affects neurological damage, with hypothermia leading to smaller lesions^27^ and hyperthermia to more severe deficits.^28^ Accordingly, temperature control and maintenance is highly relevant in this model.^29^ In addition, blood pressure and blood gases are important confounders of outcome and need to be monitored.^30,31^ The use of quick and minimal invasive methods (non-invasive blood pressure measurement, suitable and easy accessible blood collection sites) is recommended. The choice of the anaesthetic is also highly important, as some may have neuroprotective effects, and/or be vasodilators, as for example Isoflurane.^32^ Consequently, exposure to anaesthesia should be as short as possible and standardized. We exclude animals which have undergone surgery for longer than 15 min.
Shaving the surgical site produces microabrasions and inflammation and releases hair fragments. This may further increase inflammation and foster local infection, which may impact on stroke pathophysiology. Housing conditions, especially the use of enrichment, may affect stroke outcome and should be standardized and described in research reports. ^6,33,34^ The use of environmental enrichment and its effect on reproducibility is a matter of discussion.^35^ Another important confounder is the stroke induced risk for infection especially after longer times of ischaemia^36^, which leads to additional morbidity and increased mortality. ^37,38^ As infections become symptomatic at around day 3 to 5, this has important consequences for long term follow-up in such models.
To produce results relevant for the development of novel treatment strategies for stroke, standardization and quality control is highly important in preclinical translational stroke research. ^39^ \"Good Laboratory Practice\" ^40,41^ mandates:
1. an appropriate and detailed description of animals used;
2. sample size calculation and reporting on the expected effect size;
3. inclusion and exclusion criteria, set up before the study;
4. randomization of the allocation into groups;
5. allocation concealment with respect to the investigators;
6. reporting animals excluded from analysis;
7. blinded assessment of outcome;
8. reporting potential conflicts of interest and study funding.
Disclosures
===========
Experiments on animals were performed in accordance with the guidelines and regulations set forth by Landesamt fuer Gesundheit und Soziales, Berlin, Germany.
This work was funded by the European Community\'s Seventh Framework Programme (FP7/2007-2013) under grant agreement n° 201024 and n° 202213 (European Stroke Network). Additional funding was given by the Bundesministerium für Bildung und Forschung (Center for Stroke Research Berlin), and the Deutsche Forschungsgemeinschaft (Exzellenzcluster NEUROCURE).
The authors would like to thank Mareike Thielke (Dep. for Experimental Neurology, Charité Berlin) for the support during the operation and Elke Ludwig (Charité Video Services) for producing the animation.
[^1]: Correspondence to: Ulrich Dirnagl at <ulrich.dirnagl@charite.de>
| {
"pile_set_name": "PubMed Central"
} |
{#sp1 .162}
{#sp2 .162-1}
{#sp3 .163}
{#sp4 .164}
{#f1 .162-1}
| {
"pile_set_name": "PubMed Central"
} |
Although the CA2 region was first described by Lorente de Nó in 1934^[@R8]^ relatively little is known about its functional properties and behavioral role. To examine the importance of this region, we generated a transgenic mouse line (*Amigo2-Cre*) that expresses Cre recombinase predominantly in CA2 pyramidal neurons (PNs) in adult mice ([Extended Data Fig. 1](#F5){ref-type="fig"}). Because this line expresses Cre throughout the brain during early development, as well as in certain limited areas outside of CA2 in the adult, we stereotactically injected Cre-dependent adeno-associated virus (AAV) into the hippocampus of adult *Amigo2-Cre* mice to limit viral expression to CA2 pyramidal cells.
To determine the specificity of CA2 expression in the transgenic line, we bilaterally injected into dorsal hippocampus a Cre-dependent AAV to express yellow fluorescent protein (YFP) in Cre^+^ cells ([Fig. 1a](#F1){ref-type="fig"}). We observed selective and robust YFP expression in CA2 PNs throughout dorsal hippocampus^[@R9]-[@R11]^ ([Fig. 1b](#F1){ref-type="fig"}; [Extended Data Fig. 2a](#F6){ref-type="fig"}). We confirmed that the Cre^+^ cells were indeed CA2 PNs by demonstrating co-staining for RGS14^[@R12]^ (97.38 ± 0.31% overlap; *n* = 4 mice, 2546 cells; [Fig. 1c-e](#F1){ref-type="fig"} and [Extended Data Fig. 3](#F7){ref-type="fig"}) and other known CA2 PN markers ([Extended Data Fig. 2](#F6){ref-type="fig"}). In contrast, there was no co-staining for a CA1 PN marker ([Extended Data Fig. 2](#F6){ref-type="fig"}). Additionally, the electrophysiological properties of the YFP^+^ neurons differed significantly from those of CA1 PNs ([Extended Data Table 1](#T1){ref-type="table"}) and largely matched the values previously reported for CA2 pyramidal neurons^[@R7]^. Only a minute fraction of YFP^+^ neurons were also GABA^+^ (0.16 ± 0.16%; *n* = 3 mice, 1539 cells), demonstrating the specific targeting of CA2 excitatory PNs ([Fig. 1f, g](#F1){ref-type="fig"} and [Extended Data Fig. 3](#F7){ref-type="fig"}). Finally, our AAV injections resulted in the targeting of the vast majority of CA2 PNs in the dorsal hippocampus, measured by the percentage of RGS14^+^ cells that were also YFP^+^ (82.33 ± 2.37%, *n* = 4 mice, 2992 cells).
Next, we mapped CA2 synaptic input and output using viral tracing strategies that take advantage of the genetic targeting of CA2 PNs in the *Amigo2-Cre* mice, and largely confirmed results of previous studies using conventional^[@R13]^ and genetic-based^[@R14]^ approaches. Monosynaptic inputs to CA2 PNs were determined using an EnvA pseudotyped ΔG rabies virus strategy^[@R15]^ ([Extended Data Fig. 4](#F8){ref-type="fig"}). Unilateral viral injections revealed bilateral inputs from CA3 and CA2 ([Fig. 2a, b](#F2){ref-type="fig"}) and strong unilateral input from both lateral and medial EC layer II neurons ([Fig. 2c, d](#F2){ref-type="fig"}). In addition, synaptic inputs were detected from medial septum and diagonal band ([Fig. 2e](#F2){ref-type="fig"}), median raphe nucleus ([Fig. 2f](#F2){ref-type="fig"}), and the supramammillary nucleus of the hypothalamus ([Fig. 2g](#F2){ref-type="fig"}).
Surprisingly, we observed only sparse labeling of EC layer III neurons with the rabies virus approach. Our laboratory previously concluded that EC LIII axons provide strong excitatory drive to CA2 PNs, based on the finding that large excitatory postsynaptic potentials (EPSPs) are evoked in CA2 PNs with a focal stimulating electrode placed in the stratum lacunosum (SLM) of the CA1 region^[@R7]^, where axons from LIII EC neurons are thought to provide the predominant source of excitatory inputs. Our present results, combined with recent results^[@R13],[@R14]^ suggest that these synaptic responses recorded in CA2 PNs may result from activation of LII fibers that course through or near SLM in CA1.
Output projections from CA2 were determined by expressing YFP in CA2 PNs (as in [Fig. 1](#F1){ref-type="fig"}) and examining brains for YFP fluorescent axons. Unilateral viral injections resulted in strong bilateral labeling in hippocampal CA1, CA2, and CA3 regions, with densest projections observed in stratum oriens (SO) and weaker projections detected in stratum radiatum (SR) ([Fig. 2h, i](#F2){ref-type="fig"}). We did not observe extra-hippocampal outputs.
These anatomical results generally support previous^[@R13],[@R14]^ findings. However, we failed to observe vasopressinergic input to CA2 from the paraventricular nucleus of the hypothalamus^[@R13]^, which may reflect an inability of the transsynaptic rabies tracing system to label peptidergic inputs^[@R15]^. We also did not observe CA2 output to the supramammillary nucleus^[@R13]^, suggesting this output may represent an inhibitory projection from CA2 as we selectively labeled PNs. Moreover, we did not observe CA2 output to EC layer II^[@R16]^, perhaps because the anterograde tracing failed to detect weak connections.
To examine directly the functional and behavioral relevance of CA2, we employed the *Amigo2-Cre* mouse line to inactivate output from CA2 PNs selectively. We injected into the dorsal hippocampus of the *Amigo2-Cre* mice a Cre-dependent AAV to express tetanus neurotoxin (TeNT) light chain fused to green fluorescent protein (eGFP-TeNT) in CA2 PNs to block their synaptic output. We first verified the efficacy of this approach and characterized the influence of CA2 on its CA1 PN targets by co-expressing the light-activated cation channel channelrhodopsin-2 (ChR2)^[@R17]^ with either TeNT or YFP using Cre-dependent AAVs. Low intensity illumination (using 2-ms pulses of 470 nm light at 3 mW·mm^−2^) focused on CA2 reliably triggered action potentials in CA2 PNs, as seen by the presence of fast action currents in cell-attached patch clamp recordings ([Fig. 3a-c](#F3){ref-type="fig"}). Similar spiking was seen in neurons that co-expressed either YFP ([Fig. 3b](#F3){ref-type="fig"}) or TeNT ([Fig. 3c](#F3){ref-type="fig"}) with ChR2, indicating that the TeNT did not inhibit excitability.
Next, we determined the strength of synaptic transmission from CA2 to CA1 PNs using whole-cell current-clamp recordings to measure light-evoked postsynaptic potentials (PSPs) in CA1 PNs from hippocampal slices in which ChR2 and YFP were expressed in CA2 PNs ([Fig. 3d](#F3){ref-type="fig"}). In agreement with anatomical mapping ([Fig. 2h, i](#F2){ref-type="fig"}) and paired recordings^[@R7]^, focal photostimulation delivered to CA1 SO and SR regions evoked robust monosynaptic PSPs (mean latency 1.22 ± 0.06 ms, *n* = 119 observations) in nearby CA1 PNs ([Fig. 3e](#F3){ref-type="fig"}). Increasing the light intensity recruited progressively larger PSPs, presumably due to an increase in the number of optically-activated CA2 axons ([Fig. 3e, f](#F3){ref-type="fig"}). In stark contrast, in slices in which TeNT was co-expressed with ChR2 in CA2 PNs, illumination over a wide range of intensities produced little or no synaptic response in CA1 neurons ([Fig. 3e, f](#F3){ref-type="fig"}), demonstrating the efficacy of the TeNT lesion.
What are the behavioral consequences of inactivation of CA2? To address this question, we compared the behavior of control mice (CA2-YFP) with mice in which CA2 PNs were inactivated (CA2-TeNT), using viral injections in dorsal hippocampus^[@R11]^. Functional inactivation of dorsal CA2 did not alter locomotor activity or anxiety-like behavior ([Extended Data Fig. 5](#F9){ref-type="fig"}). Surprisingly, CA2-inactivation also did not significantly alter hippocampal-dependent spatial memory assessed by the Morris water maze (although there was a trend for the CA2-inactivated mice to learn the task more slowly; [Extended Data Fig. 6](#F10){ref-type="fig"}). Nor was there any change in hippocampal-dependent contextual fear memory or amygdala-dependent auditory fear memory ([Extended Data Fig. 7](#F11){ref-type="fig"}).
The finding that CA2 PNs integrate synaptic input from lateral EC (which conveys non-spatial information^[@R18]^) with subcortical input from both the serotonergic median raphe nucleus^[@R19]^ and the hypothalamic supramammillary nucleus^[@R20]^ suggests a potential role for CA2 in non-spatial hippocampal tasks. Previous studies have shown that the mRNA for the vasopressin 1b receptor (*Avpr1b*) is strongly expressed in CA2^[@R21]^ and that unconditional deletion of this gene impairs social recognition memory^[@R22],[@R23]^. However, *Avpr1b* mRNA is also expressed outside hippocampus^[@R21]^ and its deletion results in changes in non-hippocampal dependent behaviors, including reduced aggression and decreased sociability^[@R22],[@R23]^, raising questions as to the selective role of CA2 in the knockout phenotype^[@R24]^.
To assess directly the role of CA2 in social behavior, we first compared the performance of CA2-YFP versus CA2-TeNT mice in a three-chamber test of sociability^[@R23]^, which examines the normal preference of a subject mouse for a chamber containing a littermate versus an empty chamber ([Fig. 4a](#F4){ref-type="fig"}). In contrast to the effect of *Avpr1b* deletion, selective silencing of CA2 did not alter sociability as the CA2-TeNT and CA2-YFP groups displayed a significant and similar preference for the compartment containing the littermate ([Fig. 4a](#F4){ref-type="fig"}).
In contrast to their normal sociability, CA2-TeNT mice displayed a profound deficit in social recognition as determined by a three-chamber social novelty test^[@R23]^ ([Fig. 4b](#F4){ref-type="fig"}). In this test, social recognition was measured by the increased time a subject mouse spent interacting with a novel unrelated mouse compared to the time it spent interacting with a familiar co-housed littermate. Multiple comparison testing revealed that the CA2-YFP control group demonstrated a significant preference for the compartment containing the novel animal whereas the CA2-TeNT group did not ([Fig. 4b](#F4){ref-type="fig"}). Moreover, the difference score (time spent exploring the novel mouse minus time spent exploring the familiar mouse) of the CA2-TeNT group was significantly less than that of the CA2-YFP group ([Fig. 4b](#F4){ref-type="fig"}). This deficit was not due to a lack of interest in novelty, *per se*, as the CA2-TeNT mice demonstrated normal preference for a novel object as assayed by two different novel object recognition protocols ([Extended Data Fig. 8](#F12){ref-type="fig"}).
As the social novelty test does not incorporate a defined learning phase or delay period, we next conducted a more specific test of social memory, the direct interaction test^[@R25]^. For this test, a subject mouse was exposed to an unfamiliar mouse in trial 1. After a 1-h inter-trial interval (ITI), the subject mouse was either re-exposed to the same mouse encountered in trial 1 ([Fig. 4c](#F4){ref-type="fig"}) or exposed to a second unfamiliar mouse ([Fig. 4d](#F4){ref-type="fig"}). Social memory, measured as the decreased time a subject mouse spends exploring a previously encountered mouse, was fully suppressed by CA2 inactivation ([Fig. 4c](#F4){ref-type="fig"}). In contrast, CA2 silencing did not alter sociability as evidenced by the equal exploration times for trials 1 and 2 when a subject mouse encountered two different unfamiliar mice in successive trials ([Fig. 4d](#F4){ref-type="fig"}).
We next conducted a more stringent 5-trial social memory assay^[@R26]^ to confirm that CA2-inactivation abolishes social memory. In this assay, a stimulus mouse was presented to a subject mouse for 4 successive trials. On the fifth trial, a novel stimulus mouse was introduced ([Fig. 4e](#F4){ref-type="fig"}). The CA2-YFP control group of mice displayed normal social memory, as evidenced by a marked habituation (decreased exploration) during trials 1-4 and a striking dishabituation (increased exploration) upon presentation of a novel animal on the fifth trial. In contrast, the CA2-TeNT group showed no significant habituation during the four exposures to the stimulus mouse or dishabituation to the novel stimulus mouse, thus confirming the necessity of CA2 for social memory.
As olfaction is crucial for normal social interaction^[@R27]^, we examined whether CA2 silencing influenced the detection or recognition of non-social or social odors. CA2-TeNT mice showed no loss in the ability to detect the presence of food buried under a deep layer of cage bedding, a test of non-social odor detection ([Extended Data Fig. 9a](#F13){ref-type="fig"}). Next, we used an olfactory habituation/dishabituation test ([Extended Data Fig. 9b](#F13){ref-type="fig"}) and found that CA2 inactivation also had no effect on the ability of the mice to detect or discriminate either non-social or social odors. Thus, we conclude that the deficit in social memory in the CA2-TeNT mice was not due to a defect in sensing social or non-social odors.
Here, we developed and validated an *Amigo2-Cre* mouse line that enables precise genetic targeting of excitatory CA2 PNs, allowing us to map selectively the inputs and outputs of this largely unexplored region and demonstrate that the CA2 subfield is essential for social memory. Although we observed a fairly specific deficit in social memory following inactivation of dorsal CA2 pyramidal neurons, our results do not rule out the possibility that CA2 may participate more generally in hippocampal-dependent memory tasks. Thus, other regions of hippocampus may be able to compensate for loss of any role that CA2 may normally play in performance of the water maze or contextual fear conditioning tasks. Alternatively, CA2 may be selectively required for performance of more demanding non-social memory tasks.
The importance of human hippocampus for social memory is famously illustrated by the case of Henry Molaison (patient H.M.), who, following bilateral medial temporal lobe ablation, could not form new memories of people he had worked with for years^[@R28]^. Lesions limited to the hippocampus also impair social memory in both humans^[@R1]^ and rodents^[@R25]^. As a number of neuropsychiatric disorders are associated with altered social endophenotypes, our findings raise the possibility that CA2 dysfunction may contribute to these behavioral changes. This possibility is supported by findings of a decreased number of CA2 inhibitory neurons in individuals with schizophrenia and bipolar disorder^[@R29]^ and altered vasopressin signaling in autism^[@R30]^. Thus, CA2 may provide a novel target for therapeutic approaches to the treatment of social disorders.
Methods {#S8}
=======
Generation of *Amigo2-Cre* mouse line {#S9}
-------------------------------------
Selective expression of *Amigo2* in the CA2 region of hippocampus was identified based on GENSAT^[@R31]^ and Allen Brain Atlas^[@R32]^ data. The RP23-288P18 bacterial artificial chromosome (BAC) that contained the *Amigo2* gene and its surrounding regulatory elements was obtained from the BACPAC Resource Center^[@R33]^. Recombineering with galK selection and the SW102 bacterial strain^[@R34]^ was employed to seamlessly modify RP23-288P18 so that a Cre-HSV-polyA cassette was inserted at the translation start site of the *Amigo2* gene. Specifically, the Cre expression cassette was PCR amplified from pLD53.SC-Cre^[@R35]^. The homology arms used for the recombineering were 5\' arm: 5'-ATTGGTGGGAGACTGAGCTGATGAGAAGCGACTGGCAAGAGACTCAGAGGCGACCATA-3' and 3' arm: 5'-ATGTCGTTAAGGTTCCACACACTGCCCACCCTGCCTAGAGCTGTCAAACCGGGTTGCAGAGA-3'. This modified BAC was injected into B6CBA/F2 pronuclei and embryos were implanted into pseudopregnant females. PCR was used to identify the offspring that were Cre-positive. These founders were crossed to the Ai14 Cre-reporter line^[@R36]^ to examine the specificity of Cre expression. At 12 weeks of age, the Cre^+^ offspring were transcardially perfused with 4% paraformaldehyde (PFA) in phosphate buffered saline (PBS) and expression of tdTomato was examined in 50 μm coronal slices. CA2-specific expression of tdTomato was not observed in any of the founder lines. However, injection of the EF1α-FLEx-eYFP-WPRE-hGH Cre-reporter adeno-associated virus (AAV) into the hippocampus of adult mice (\> 8 weeks old) revealed CA2-specific expression in 1 of the 6 founder lines. This line was used for all of the studies presented here. The line was backcrossed to C57BL/6J a minimum of 6 times before any behavioral or physiological experiments were performed.
Subjects {#S10}
--------
The *Amigo2-Cre* line was maintained as a hemizygous line on the C57BL/6J background by breeding Cre^+^ males to C57BL/6J females. Only Cre^+^ males were used for these experiments. Mice \> 8 weeks old were injected with virus under stereotactic control into the hippocampus proper to avoid *Amigo2-Cre* expression in mossy cells of the dentate gyrus. All anatomical, behavioral, and physiological experiments were conducted 2-4 weeks following injection. All procedures were approved by the Institutional Animal Care and Use Committee at Columbia University and the New York State Psychiatric Institute.
Virus constructs {#S11}
----------------
AAV5- EF1α-FLEx-eYFP-WPRE-hGH (4 × 10^12^ virus molecules ml^−1^) was injected to label CA2 PNs and trace their axons. (EnvA)SAD-ΔG-mCherry (1 × 10^8^ infectious particles ml^−1^) pseudotyped rabies virus was produced as previously described^[@R37]^ and used to label monosynaptic inputs to CA2. This virus can only infect cells expressing the TVA receptor^[@R37],[@R38]^. Before rabies virus injection, AAV5- EF1α-FLEx-TVA-mCherry-WPRE-hGH^[@R38]^ (3 × 10^12^ virus molecules ml^−1^) was injected to express TVA in CA2. To permit retrograde synaptic transport of the ΔG virus, AAV5- CAG-FLEx-rabiesG-WPRE-hGH^[@R38]^ (2 × 10^12^ virus molecules ml^−1^) was co-injected with the TVA virus to express G in CA2. The aforementioned AAVs were obtained from the University of North Carolina vector core. To specifically excite CA2 PNs, AAV5-EF1α-FLEx-hChR2(H134R)-EYFP-WPRE-hGH (2 × 10^12^ genome copies ml^−1^) was injected to express ChR2 in the CA2 neurons. This vector was obtained from the University of Pennsylvania (UPenn) vector core. To ablate CA2 pyramidal cell output, tetanus neurotoxin light chain (TeNT) was expressed selectively in these cells. A Cre-dependent AAV vector carrying eGFP-TeNT was created by PCR amplifying eGFP-TeNT from pTRE2-eGFP-TeNT-PEST^[@R39]^ and subcloning it into pAAV- EF1α-DIO-hChR2-mCherry-WPRE (Addgene plasmid 20297) between the NheI and AscI sites in the inverse orientation. The resulting vector, pAAV- EF1α-FLEx-eGFP-TeNT-WPRE-hGH was sent to the UPenn vector core for custom production of AAV5- EF1α-FLEx-eGFP-TeNT-WPRE-hGH (1 × 10^13^ genome copies ml^−1^).
Stereotaxic injection {#S12}
---------------------
Mice were anesthetized with isoflurane (2-5%) and placed in a stereotaxic apparatus (Digital Just for Mice Stereotaxic Instrument, Stoelting). The head was fixed, and the skull was exposed. Burr holes were made and a glass micropipette (Drummond Scientific) was slowly lowered into the dorsal hippocampus at −1.6 mm anteroposterior, ±1.6 mm mediolateral, and −1.7 mm dorsoventral relative to bregma. The pipettes were formed with 20μm diameter tips using a P-2000 laser puller (Sutter Instrument). For the mouse line validation, anterograde tracing, and behavioral experiments, 180 nl of virus was pressure injected into each hemisphere. For the retrograde tracing experiments, 180 nl of a 1:5 mix of the TVA and rabies G AAV vectors was injected unilaterally. For the electrophysiological experiments, 360 nl of a 1:1 mix of ChR2 AAV and either YFP or TeNT AAV was injected. After injection, the pipette remained in place for 5min and then was slowly retracted. The mice were placed on a heating pad (TR-200, Fine Science Tools) throughout the duration of the surgery. Following injection, the scalp was sutured, saline was administered subcutaneously, and buprenorphine (0.05-0.1 mg/kg) was administered intraperitoneally for analgesia. The mice were placed under heating lamps during recovery from anesthesia. For the retrograde tracing experiments, these procedures were repeated two weeks after the initial AAV injection to inject 360 nl of (EnvA)SAD-ΔG-mCherry rabies virus. To test the specificity of the rabies virus, a subset of animals was injected with 360 nl of the (EnvA)SAD-ΔG-mCherry rabies virus without prior injection of AAVs expressing TVA and G. All injections were verified histologically. No injections were mistargeted, hence no subjects were excluded from analysis due to injection failure.
Immunohistochemistry and confocal microscopy {#S13}
--------------------------------------------
Mice were administered ketamine/xylazine (150 mg/kg, 10 mg/kg) and transcardially perfused with ice-cold PBS followed by ice-cold 4% PFA in PBS. Brains were postfixed overnight in 4% PFA in PBS and 50μm slices were prepared (Vibratome 3000 Plus, The Vibratome Company). Antigen retrieval^[@R40]^ was performed for RGS14 staining. Briefly, free-floating sections were incubated at 80°C for 30 min in 50 mM sodium citrate (pH = 8.5). Slices were permeabilized with 0.2% Triton X-100 in PBS and blocked with 10% goat serum in PBS. The sections were incubated at 4°C overnight in primary antibody (1:50 dilution, NeuroMab, 73-170). For PCP4^[@R10]^, STEP^[@R41]^, WFS^[@R42]^, and GABA staining, sections were permeabilized and blocked as above and then incubated in PCP4 (1:200 dilution, Sigma-Aldrich, HPA005792), STEP (1:500 dilution, Cell Signaling Technology, 4396), WFS1 (1:250 dilution, Proteintech, 11558-1-AP), or GABA (1:500 dilution, Sigma-Aldrich, A2052) primary antibodies at 4° C overnight. Sections were washed the following day and incubated for 2h with Alexa 555 or 647 secondary antibody (1:500 dilution, Invitrogen, A21422, A21428, or A21245) and NeuroTrace (1:250 dilution, Invitrogen, N21479 or N21483). Slices were then mounted with either Prolong Gold (Invitrogen, P36930) or VECTASHIELD (Vector Laboratories, H-1000) and imaged. An inverted laser scanning confocal microscope (LSM 700, Zeiss) was used for fluorescence imaging followed by analysis in ImageJ^[@R43]^. For cell counting experiments, every fifth slice throughout the rostral half of the hippocampus (5 slices in total) was examined. The first section was randomly chosen, and cells were assessed for double labeling in a single optical section taken near the middle of the slice.
Electrophysiology {#S14}
-----------------
2-3 weeks following AAV injection, mice were anesthetized with isoflurane (5%) and transcardially perfused with an ice-cold dissection solution that contained (in mM): 10 NaCl, 195 sucrose, 2.5 KCl, 10 glucose, 25 NaHCO~3~, 1.25 NaH~2~PO~4~, 2 Na Pyruvate, 0.5 CaCl~2~ and 7 MgCl~2~. The hippocampi were dissected out and 400-μm thick slices were cut (VT1200S, Leica) perpendicular to the longitudinal axis of the hippocampus. The slices were then transferred to a chamber containing a 1:1 mixture of dissection solution and artificial cerebrospinal fluid (aCSF). The aCSF contained (in mM): 125 NaCl, 2.5 KCl, 22.5 glucose, 25 NaHCO~3~, 1.25 NaH~2~PO~4~, 3 Na Pyruvate, 1 Ascorbic Acid, 2 CaCl~2~ and 1 MgCl~2~. Slices were incubated at 30°C for 30 min and then at room temperature for at least 1.5 hr before recording. Slices were transferred to a recording chamber (Warner Instruments), perfused with aCSF, and maintained at 33° C. All solutions were saturated with carbogen (95% O~2~ and 5% CO~2~). Whole-cell recordings were obtained from PNs with a patch pipette (3--5 MΩ) containing (in mM): 135 KMeSO~4~, 5 KCl, 0.1 EGTA-Na, 10 HEPES, 2 NaCl, 5 ATP, 0.4 GTP, 10 phosphocreatine at pH 7.2 and osmolarity of 280--290 mOsm. Series resistance, which was always less than 30 MΩ, was monitored and compensated throughout the experiment. Cells with a 15% or greater change in series resistance were excluded from analysis. To activate ChR2, 2 ms pulses of blue (470 nm) light (M470L2-C1, Thor Labs) were delivered through a 20X objective. Light power from the objective was measured with a power meter (PM100D, Thor Labs). The objective was centered on the neuron that was being recorded during the experiment. For the CA2 cell-attached recordings, a gigaohm seal was made and action currents were measured in voltage-clamp mode (cell clamped at −70 mV) while 5 pulses of blue light were delivered. For the input-output curves, whole-cell recordings were made from CA1 PNs in current-clamp mode and the objective was centered on the patched CA1 neuron. This provided illumination over stratum oriens (SO), stratum pyramidale (SP), and stratum radiatum (SR) thus activating the CA2 projections to CA1 that course through SO and SR.
Behavioral tests {#S15}
----------------
Mice were housed 2-5 per cage and were given *ad lib* access to food and water. They were kept on a 12 h (6 A.M. to 6 P.M.) light--dark cycle in a room maintained at 21°C. All tests were conducted during the light cycle. Mice were habituated to handling and transport from the colony room to the behavioral room for 3 days before behavioral tests were begun. Mice were given 1 h to habituate after transport to the behavioral room before any tests were conducted. The experimenter was blind to the treatment groups. The control group (CA2-YFP) was injected with AAV5- EF1α-FLEx-eYFP-WPRE-hGH while the CA2-inactivated group (CA2-TeNT) was injected with AAV5- EF1α-FLEx-eGFP-TeNT-WPRE-hGH. To blind the experimenter and randomize the treatment groups, virus aliquots were stored as pairs of coded cryotubes. Half of the mice in each home cage were injected with the YFP virus, while the other half were injected with the TeNT virus. The identity of the groups was revealed only after testing was completed. For the elevated plus maze, novel object, Morris water maze, and 3-chamber tests, mice were tracked with an overhead FireWire camera (DMK 31AF03-Z2, The Imaging Source) and ANY-maze (Stoelting). Freezing during fear conditioning was tracked with a Fire-i (Unibrain) camera and analyzed with ANY-maze (Stoelting). All apparatuses and testing chambers were cleaned with 70% isopropanol wipes (VWR) between animals unless otherwise indicated below.
Open field {#S16}
----------
Mice were placed in an open field (ENV-510S, Med Associates, Inc.) for 30 min and locomotor and rearing activity was monitored via IR beam breaks and recorded by the Activity Monitor (Med Associates, Inc.) software. The entire apparatus was enclosed in a sound attenuating cubicle.
Elevated plus maze {#S17}
------------------
Mice were placed in the center of a maze (Stoelting) constructed in the shape of a plus with two enclosed arms (15 cm high walls) and two open arms. The maze was elevated 40 cm from the ground. Mice were allowed to explore the maze for 8 min. Entry into an arm was scored only after 85% of the animal\'s tracked body area was in the arm.
Novel object {#S18}
------------
Two variations of the novel object task were run. Both were conducted in a 50 cm long × 25 cm wide × 30.5 cm high arena. For both tests, the snouts of the mice were tracked and object interaction was measured as time spent with snout within 2 cm of the object. The objects (a glass chess piece, a small metal lock, and a small plastic box) were secured to the arena with neodymium magnets to render them immovable. In the first variation, mice were habituated to the arena and objects 1 and 2 over the course of four 5 min trials separated by an ITI of 10 min. Mice were then tested for object recognition memory 1 h after the fourth trial during the 5 min-long fifth trial. Either object 1 or object 2 (counterbalanced) was swapped for object 3 during the fifth trial. In the second variation of this test, the mice were habituated to the empty arena for 10min each day for 3 consecutive days. On day 4, the mice were exposed to a pair of either object 1 or object 2 for 5 min. Object recognition memory was tested 1 h after this trial by exposure to objects 1 and 2 for 5 min. In both protocols, object recognition memory was measured as the increased time spent investigating the novel object.
Morris water maze {#S19}
-----------------
The Morris water maze task was run over the course of 14 days in a 120 cm diameter pool filled with water that was opacified with non-toxic white paint (Prang tempera paint, VWR). The water was maintained at 19-20° C. Four 1 min trials were administered per day, and mice were run in groups of 8. On days 1-2, cued learning was conducted. During this procedure, mice were trained to find a circular platform (10cm in diameter) submerged 1 cm below the surface of the water and marked with a flag. Distal cues in the room were obscured by a black curtain that encircled the tank. The mouse was removed from the tank and returned to its home cage 15s after locating the platform. If a mouse failed to locate the platform during the minute-long trial, it was gently guided towards the platform. The mice were released from different start points at the beginning of each trial, and the platform location also varied between trials. On days 3-7, the flag was removed from the platform, rendering it hidden, and the curtains were removed, which allowed the mice to now use distal cues to locate the hidden platform. The platform was kept in the middle of the SW quadrant of the maze during days 3-7. On day 8, spatial memory was assayed with a 1min probe trial in which the platform was removed. Reversal training was conducted on days 9-13 with the platform now hidden in the NW quadrant. Spatial memory of the novel location was tested with a 1min probe trial on day 14. Release and platform locations were adapted from previous studies^[@R44]^.
Fear conditioning {#S20}
-----------------
A 3-day delay fear conditioning protocol was employed to test hippocampal-dependent contextual fear memory and amygdala-dependent auditory fear memory. On day 1, the mice were placed in an enclosure (17 cm × 17 cm × 25 cm) with a steel grid floor. This enclosure was located in a sound-attenuating chamber that contained a FireWire camera, light, and speaker. On day 1, the enclosure was outfitted as context A which consisted of 3 plexiglass walls and 1 opaque wall with black and white stripes. 1% acetic acid was placed as the dominant odor, and the house fan was turned on. The enclosure was cleaned with 70% isopropanol between animals. Mice were moved from their home cage to a transfer cage with no bedding and after 15-20s were placed in the fear conditioning chamber. After 150s, a tone (30 s, 2.8 kHz, 85 dB) was played and co-terminated with a shock (2 s, 0.7 mA). Mice were removed from the chamber 30 s after the shock. On day 2, contextual fear memory was assayed by placing the mice back in context A for 300 s. On day 3, the mice were brought to the testing room that was now dimly illuminated with red light. The mice were placed in context B, which consisted of an enclosure with 3 solid gray colored walls, 1 plexiglass wall with a circular door, and a red, flat plastic roof. The floor of the enclosure was a white piece of plastic, 0.25% benzaldehyde was the dominant odor, and the enclosure was cleaned between animals with Vimoba. Mice were first moved from their home cage to a circular bucket and then to the testing chamber. After 180 s, the tone from day 1 was sounded for 60 s. Percent time spent freezing (defined as the absence of all movement except for respiration) was measured throughout these experiments and served as an index of fear memory.
Sociability and social novelty {#S21}
------------------------------
This test was performed as previously described^[@R23]^. Briefly, mice were placed in an arena divided into 3 equal-sized compartments by plastic mesh. On day 1, a 5 min sociability trial was conducted. A littermate was placed in the left or right compartment (systematically alternated) and the test subject was placed into the center compartment. The time the test subject spent investigating each compartment (snout within 2 cm of the mesh barrier) was measured and a difference score was computed. On day 2, a 5 min social novelty test was conducted in which a littermate was placed in either the left or right compartment, and a novel animal (C57BL/6J, 3-month-old, male) was placed in the other compartment. The test subject was placed in the center compartment, investigation time was measured, and a difference score, determined by subtracting the time spent investigating the two compartments, was computed.
Direct interaction {#S22}
------------------
This test was adapted from Kogan *et al.*^[@R25]^. Under low light (12lux), mice were placed in a standard clean cage and a novel mouse (C57BL/6J, 4-5-week-old, male) was introduced. Activity was monitored for 5 min and scored online for social behavior (anogenital and nose-to-nose sniffing, following, and allogrooming) initiated by the test subject. After an ITI of 1 hr, the test was run again with either the previously encountered mouse or a novel mouse. The time spent in social interaction during trial 1 was subtracted from the social interaction time during trial 2 to obtain the difference score.
5-trial social memory assay {#S23}
---------------------------
This test was run as previously described^[@R26],[@R45]^. Briefly, subject mice were individually housed for 7 days prior to testing. On the day of testing, the subjects were presented with a 10 week old CD-1 ovariectomized female mouse for 4 successive 1 minute trials. On the fifth trial, a novel stimulus animal was presented.
Buried food test {#S24}
----------------
To ensure palatability of the food, mice were given 1 g reward treats (F05472-1, Bio-Serv) in their home cages one day before testing. All pellets were consumed. The mice were then food deprived for 18 h before the test to improve sensitivity^[@R46]^. A treat was hidden under 1.5 cm of standard cage bedding, a mouse was placed in the cage, and the latency to consumption of the treat was recorded.
Olfactory habituation/dishabituation test {#S25}
-----------------------------------------
This test was run as previously described^[@R46]^ with the exclusion of the first 3 trials in which a water-soaked cotton swab is presented. A trained observer measured and recorded olfactory investigation of the odorant-soaked cotton swabs.
Statistical Analysis {#S26}
--------------------
Prism 6 (GraphPad) was used for statistical analysis and to graph data. Statistical significance was assessed by two-tailed unpaired Student\'s t-tests, 2-way ANOVA, or 2-way repeated measures (RM) ANOVA where appropriate. Significant main effects or interactions were followed up with multiple comparison testing using Holm-Sidak\'s correction. Results were considered significant when P \< 0.05. α was set equal to 0.05 for multiple comparison tests. Sample sizes were chosen based on previous studies. Data met assumptions of statistical tests and variance was similar between groups for all metrics measured except for AP duration ([Extended Data Table 1](#T1){ref-type="table"}), social novelty difference score ([Fig. 4b](#F4){ref-type="fig"}), and direct interaction difference score ([Fig. 4c](#F4){ref-type="fig"}).
We thank T.R. Reardon for providing the rabies virus and Justine Kupferman and Franklin Lema for experimental assistance. We thank Christine Denny, Zoe Donaldson, René Hen, Josh Gordon, Jayeeta Basu, and Marco Russo for helpful discussions and comments on the manuscript. This work was supported by a Ruth L. Kirschstein F30 NRSA from the NIMH (F.L.H.) and the Howard Hughes Medical Institute (S.A.S.).
**Author Contributions** F.L.H planned and performed the experiments, analyzed the data, and wrote the manuscript. S.A.S oversaw the overall execution of the project, contributed to the experimental design and the interpretation of the data, provided financial support, and helped write the manuscript.
The authors declare no competing interests.
![Genetic targeting of the CA2 subfield using the *Amigo2-Cre* mouse line\
**a**, Bilateral hippocampal injection (*n* = 64) of Cre-dependent YFP AAV in *Amigo2-Cre* mice resulted in specific expression of YFP (green) in CA2 PNs. **b**, Extent of transduction. Left, adapted reference atlas images^[@R9]^. Center, YFP expression. Right, mm from bregma along rostrocaudal axis. **c-g**, Magnified images of boxed area in (**b**). **c**, YFP (green). **d**, RGS14 staining (red, *n* = 4). **e**, Merge of (**c**) and (**d**) showing YFP and RGS14 overlap. **f**, GABA staining (red, *n* = 3). **g**, merge of (**c**) and (**f**) showing no GABA and YFP overlap. Panels show coronal sections with Nissl counterstain (blue). Scale bars, 1000 μm, 400 μm, 200 μm in (**a**), (**b**), (**c-g**), respectively.](nihms-555416-f0001){#F1}
{#F2}
{#F3}
{#F4}
{#F5}
![*Amigo2-Cre* mice express Cre in a genetically defined population of CA2 PNs\
Coronal sections of hippocampus from *Amigo2-Cre* mice injected in dorsal hippocampus with a Cre-dependent AAV to express YFP (shown in green) in CA2. **a**, Coronal section of ventral hippocampus (\~2.8 mm caudal to bregma, see Figure 54 of Franklin & Paxinos^[@R9]^ for reference image) showing CA2 axons (green) from dorsal CA2. Note absence of YFP in ventral CA2 neurons (RGS14 stain in red). **b**, 97.22 ± 0.46% of YFP^+^ cells (*n* = 4 mice, 2948 cells) express the CA2 marker PCP4 (red). **c**, 98.45 ± 0.33% of YFP^+^ cells (*n* = 4 mice, 2870 cells) express the CA2 marker STEP (red). **d**, Nearly no YFP^+^ cells (0.17 ± 0.13%; *n* = 4 mice, 2870 cells) express the CA1 marker WFS1 (red). **e-f**, Magnification of boxed area in **(b)** showing YFP signal (**e**) PCP4 staining (**f**) and a merge of the two (**g**). **h-j**, Magnification of boxed area in (**c**) showing YFP signal (**h**) STEP staining (**i**) and a merge of the two (**j**). **k-m**, Magnification of boxed area in (**d**) showing YFP signal (**k**) WFS1 staining (**l**) and a merge of the two (**m**). Nissl stain shown in blue. Scale bars, 400 μm (**a-d**) and 100 μm (**e-m**).](nihms-555416-f0006){#F6}
{ref-type="fig"}. **b, e**, Magnification of area boxed in (**a**). **c**, RGS14 staining of section shown in (**b**). **d**, Merge of (**b, c**) demonstrating YFP and RGS14 co-labeling. **f**, GABA staining of section shown in (**e**). **g**, Merge of (**e, f**) showing no overlap of GABA and YFP. **h**, Reproduction of section −1.46 mm shown in [Fig. 1b](#F1){ref-type="fig"}. **i, l**, Magnification of area boxed in (**h**). **j**, RGS14 staining of section shown in (**i**). **k**, Merge of (**i**, **j**) demonstrating YFP and RGS14 co-labeling. **m**, GABA staining of section shown in (**l**). **n**, Merge of (**l, m**) showing no overlap of GABA and YFP. **o**, Reproduction of section −2.18mm shown in [Fig. 1b](#F1){ref-type="fig"}. **p, s**, Magnification of area boxed in (**o**). **q**, RGS14 staining of section shown in (**p**). **r**, Merge of (**p, q**) demonstrating YFP and RGS14 co-labeling. **t**, GABA staining of section shown in (**s**). **u**, Merge of (**s, t**) showing no overlap of GABA and YFP. Scale bars, 200 μm. Nissl stain shown in blue.](nihms-555416-f0007){#F7}
{#F8}
{#F9}
{#F10}
{#F11}
{#F12}
{#F13}
###### Electrophysiological properties of Cre^+^ neurons
The electrophysiological properties of Cre^+^ neurons (Column 1) closely matched the properties previously reported^[@R7]^ for CA2 neurons and significantly differed from the properties of CA1 neurons (Column 2). Two-tailed unpaired t-tests were used to assess significant differences between the neuronal populations. The P values are shown in Column 3. Whole-cell recordings of Cre^+^ (*n* = 5) and CA1 (*n* = 9) neurons were conducted to measure input resistance, capacitance, resting potential, action potential (AP) amplitude, AP duration, and sag. Input resistance and capacitance were measured with a −5 mV pulse. The AP amplitude and duration were measured during a 500 ms depolarizing pulse and the sag resulting from activation of I~h~ was measured during a 500 ms hyper-polarization from −70 to −100mV. Smaller sag in Cre^+^ neurons compared to that previously reported^[@R7]^ is likely due to differences in extent of whole-cell dialysis resulting from differences in recording protocols.
Cre^+^ neurons CA1 P value
------------------------ ---------------- -------------- -----------
Input Resistance (MΩ) 68.3 ± 3.03 90.0 ± 6.65 0.039
Capacitance (pF) 296.0 ± 18.68 140.7 ± 8.02 \< 0.0001
Resting Potential (mV) −76.3 ± 0.63 −72.8 ± 0.92 0.024
AP Amplitude (mV) 90.81 ± 2.17 99.15 ± 1.36 0.007
AP Duration (ms) 0.83 ± 0.02 1.06 ± 0.06 0.031
Sag (mV) 1.92 ± 0.50 7.55 ± 0.85 0.0006
| {
"pile_set_name": "PubMed Central"
} |
Introduction {#Sec1}
============
Water is a seemingly simple yet practically complex liquid with extraordinary phase behavior, which enables many of life's intricacies. While water is possibly the most studied liquid, there remain many areas where its behavior is still mysterious^[@CR1]^. A prime example for this is the freezing and the supercooling of water occurring in our daily lives and scientific research^[@CR2],[@CR3]^. Ice formation and the preceding supercooled state of microdroplets in atmospheric clouds are crucial elements for precipitation and reflection of solar radiation^[@CR4][@CR5]^. Furthermore, chilling, freezing, freeze avoidance, and supercooling are important strategies to combat cold environment for ectothermic animals^[@CR6],[@CR7]^, treat malignant diseases via cryotherapy^[@CR8]^, and preserve food and various biological samples, such as cells, tissues, and organs^[@CR9],[@CR10]^.
Recent advances have shown that supercooling can be a promising alternative approach for the preservation of cells, tissues, and especially organs^[@CR11]^. Nevertheless, an important hurdle for supercooling preservation, as well as other applications of supercooling, is that simultaneous low temperature (\<−10 °C), large volume (\>1 ml), and long period (\>1 week) of supercooling for aqueous solutions cannot be readily achieved^[@CR12]--[@CR14]^. High-pressure-based approaches have provided supercooled states of water down to −92 °C briefly^[@CR1]^, according to the water phase diagram. They are, however, expensive, might further complicate preservation of biological samples, and their long-term fate is unknown. Few experiments have unstably supercooled large volumes, several hundred milliliters, of water to −12 °C^[@CR15]^, albeit also for periods on the order of seconds. Similarly, in Dorsey's classical work on freezing of supercooled water, he was able to achieve a temperature of −19 °C for a few milliliters of water very briefly during his constant cooling experiments^[@CR16]^. A method that overcomes these hurdles and enables long-term supercooling of large aqueous samples at low temperatures could find applications in biopreservation, as well as many other areas which have previously been practically prohibited.
Under normal atmospheric conditions, ice melts at 0 °C, the ice-water equilibrium temperature (*T*~e~). Nevertheless, the observed freezing temperature (*T*~f~) for pure water could fall below *T*~e~ since successful ice nucleation at 0 °C can take a long time. Water, in the liquid phase, below the equilibrium temperature is said to be "supercooled" where Δ*T* = *T*~e~ − *T*~f~ measures the degree of supercooling. Supercooled water is intrinsically metastable and can spontaneously transform to lower-energy-level ice crystals through the formation of ice nuclei, which can be readily achieved by ice seeding^[@CR17]^, ultrasonicating^[@CR18]^, or presenting ice-nucleating agents^[@CR19]^. On the contrary, it is very difficult to maintain supercooled water unfrozen, especially for a large volume, under a high degree of supercooling, or for a long period, as each of these increases the possibility of ice nucleation and water freezing (Supplementary Note [1](#MOESM1){ref-type="media"}). For instance, Δ*T* of a water droplet decreases logarithmically with increasing volume under a constant cooling rate^[@CR20]^. Similarly, supercooling frequency (*f*~s~, *f*~s~ = number of unfrozen droplets/number of total droplets) of an ensemble of droplets decreases exponentially with increasing droplet volume, storage time, and nucleation rate (*J*)^[@CR21],[@CR22]^, while *J* itself increases exponentially with Δ*T*^[@CR23]^. Consequently, simultaneous long-term (\>1 week), large volume (\>1 ml), and deep supercooling (DSC) (Δ*T* \> 10 °C) of water has not yet been achieved.
There are two general ice nucleation mechanisms, homogenous and heterogeneous crystallization. Homogeneous crystallization occurs due to random aggregation of interior water molecules to create a critically large nucleus of ice crystal, which could only be achieved and observed below −20 °C^[@CR24]^. Heterogeneous crystallization, on the other hand, stems from ice nucleus formation catalyzed by a substrate and/or with the aid of foreign objects at much higher temperatures^[@CR25]^. Consequently, water freezing is generally initiated by heterogeneous nucleation, and the water/air interface is the primary nucleation site as revealed in theoretical^[@CR26],[@CR27]^, experimental^[@CR14],[@CR28]^, and numerical^[@CR29],[@CR30]^ studies.
Here, we describe an unexpected method based on sealing of the water surface by an immiscible hydrocarbon-based liquid, such as oils, pure alkanes, and pure primary alcohols. This method, as we demonstrate through a series of experiments, enables stable supercooling of large volumes of water for long periods at temperatures well below −10 °C by eliminating the primary ice nucleation site on the water/air interface. The supercooled water can withstand vibrational and thermal disturbances with all sealing agents, and even ultrasonic disturbance if it is sealed by alcohols. In addition, we utilize this DSC approach to preserve human red blood cells (hRBCs) for as long as 100 days.
Results {#Sec2}
=======
Water deep supercooling via surface sealing {#Sec3}
-------------------------------------------
When water molecules aggregate on the water surface (water/air interface) to form an ice nucleus, they need to overcome an energy barrier *γ*^ia^ − *γ*^wa^ (*γ*: interfacial tension, symbols i, w, and a refer to ice, water, oil, and air, respectively) per unit area as the ice/air interface replaces original water/air interface. In comparison, the energy barrier for homogeneous ice nucleation within bulk water is proportional to the water/ice interfacial tension, *γ*^wi^. This interfacial tension can be expressed via the Young's equation as *γ*^wi^ = *γ*^ia^ − *γ*^wa^ cos*θ*~iwa~ ≥ *γ*^ia^ − *γ*^wa^ (*θ*~iwa~: water contact angle on ice/water/air interface, Supplementary Note [1](#MOESM1){ref-type="media"} and Supplementary Figure [1a](#MOESM1){ref-type="media"}). This inequality indicates that heterogeneous ice nucleation on the surface is thermodynamically more favorable than homogeneous nucleation in bulk as complete wetting (*θ*~iwa~ = 0°) is not generally observed^[@CR27]^, and a receding contact angle of 12° has been reported^[@CR31]^. Therefore, if the water surface is sealed by an oil phase, the energy barrier of ice nucleation at the water-oil interface would be *γ*^io^ − *γ*^wo^ (symbol o refers to oil phase). Similarly, the homogenous nucleation energy barrier can be now expressed in terms of another triple interface, namely the oil/water/ice as *γ*^wi^ = *γ*^io^ − *γ*^wo^ cos *θ*~iwo~, where *θ*~iwo~, for many oils can be nearly 0° as they are very repellent to ice (Supplementary Note [1](#MOESM1){ref-type="media"} and Supplementary Figure 1b)^[@CR28],[@CR32]^. In the case of *θ*~iwo~ ≅ 0, the energy barrier approaches the limiting case *γ*^io^ − *γ*^wo^ ≅ *γ*^wi^. This analysis indicates that the energy barrier of heterogeneous crystallization at the surface is elevated almost to the level of homogeneous one when the water/air interface is replaced by an oil/water interface. Accordingly, we hypothesized that surface sealing of water with an appropriate oil phase could suppress primary heterogeneous ice nucleation at the surface and enable extended storage of deeply supercooled water.
Water deep supercooling via surface sealing with oils {#Sec4}
-----------------------------------------------------
First, we cooled a large ensemble of polystyrene tubes containing 1 ml of ultra-pure water to −13 °C (Fig. [1a, b](#Fig1){ref-type="fig"}). This resulted in \>90% of samples to be frozen after 24 h and nearly all samples to be frozen after 5 days. In contrast, the ultra-pure water samples could be kept in the liquid phase for a week, at the same temperature, if their surfaces were sealed by various types of immiscible oils, such as light mineral oil (MO), olive oil (OO), heavy paraffin oil (PO), and nutmeg oil (NO). Interestingly, the curdling of OO during DSC does not trigger water freezing, though the cumulative freezing frequency (*f*~f~, *f*~f~ = 1 − *f*~s~) increases significantly compared to water sealed by other oils (Fig. [1a](#Fig1){ref-type="fig"}). In supplementary experiments, we observed that the water degassed by vacuuming for 24 h, has similar *f*~*f*~ as normal water, with or without oil sealing (Supplementary Figure [2](#MOESM1){ref-type="media"}). These experiments indicate that air dissolved in the water does not play a major role in ice nucleation in our experiments. Given this result and the consistent efficacy of surface sealing by different oils on freezing reduction, we infer that the air-water interface is the primary nucleation site.Fig. 1Deep supercooling of pure water enabled by surface sealing with oils. **a** Cumulative freezing frequency (*f*~f~) for 1 ml water at −13 °C over 7-day deep supercooling (DSC), without sealing (W/O seal), with surface sealing by light mineral oil (MO), olive oil (OO), heavy paraffin oil (PO), and nutmeg oil (NO). Number of independent experiments *n* = 6, number of total tested samples for each case *N* = 56. NS: *p* \> 0 .05; \*0.005 \< *p* *\<* 0.05; \*\*1.0 × 10^−6^ \< *p* *\<* 0.005, \*\*\**p* *\<* 1.0 × 10^−6^. **b** Corresponding samples of (**a**) post 1-day storage. **c** *f*~f~ of DSC water of various volumes post 1-day storage at −13 and −16 °C. *n* = 7, *N* = 272, 145, 336, 123, and 125 for 3, 30, 200, 1000, and 10000 μl water, respectively. **d** *f*~f~ of 100,000 μl water with different sealing oils and temperatures post 1-day storage, *n* = 7, *N* = 35. Error bars represent standard deviations
We also examined the influence of water volume on the efficacy of oil sealing for freezing inhibition. We studied the two most promising oils, MO and PO, at −13 and −16 °C for ultra-pure water ranging from 10^0^--10^5^ µl (Fig. [1c, d](#Fig1){ref-type="fig"}, Supplementary Figure [3](#MOESM1){ref-type="media"}). We found that MO sealing can effectively suppress water freezing for water volumes up to 10^4^ µl at −13 and −16 °C. PO sealing was even more effective with a low *f*~f~ throughout the entire volume range at −13 °C, and only 45.8% of samples frozen at −16 °C for the 10^5^ µl samples. In addition, 8 out of 35 (22.8%) samples of 10^5^ µl water were kept in the supercooled state at −16 °C for 100 days without any freezing event after Day-3 (Supplementary Figure [3b](#MOESM1){ref-type="media"}). While further investigations might be necessary, these observations are incompatible with conventional stochastic freezing processes (Supplementary Note [1](#MOESM1){ref-type="media"}), which implies exponential decrease of *f*~s~ with time^[@CR21],[@CR22]^. Alternatively, the freezing of DSC water sealed by oil could be depicted as "case-specific" that some of sealed water samples are more susceptible to crystallization than others. A reconciliation of these two cases might lie in the fact that those samples that do not freeze within our observation period have much fewer impurities and thus a much smaller exponential for the decay of *f*~s~ than those that freeze within 3 days.
In order to further support our hypothesis that the water/air interface plays a dominant role in ice nucleation and subsequent freezing, we measured water freezing frequencies under differential degrees of surface sealing by MO, ranging from (I) unsealed (0 oil), (II) ring sealed along the contact line between water and tube wall (0.01 ml), (III) partially sealed with partial exposure to air (0.1 ml), (IV) critically sealed with water surface just completely covered (0.15 ml), (V) normally sealed (0.5 ml), and (VI) over sealed with excessive oil mounted on water surface (3.5 ml) (Fig. [2a, b](#Fig2){ref-type="fig"} and Supplementary Figure [4](#MOESM1){ref-type="media"}). The results indicate that the capacity of freezing inhibition increases with the degree of sealing, with a statistically maximum plateau achieved by critical sealing (Fig. [2a](#Fig2){ref-type="fig"}). Ring sealing (0.01 ml) that nullifies the triple solid/water/air contact line has a mild effect on freezing inhibition at high temperatures (−10 °C) but is not effective below −13 °C. Taken together with partial sealing results (0.1 ml), this result implies that the contact nucleation at the air/water/solid triple interface is not as dominant as that at water/air interface especially at low temperatures. Considering the crystallization efficiency depends on the integration of nucleation probability *J* and nucleation length (or area), the triple contact line of short length would provide smaller crystallization efficiency than the air/water interface even though it has higher *J*^[@CR33],[@CR34]^. Overall, we confirmed that the water/air interface is the primary ice nucleation site for DSC water, and surface oil sealing that removes the water/air interface can effectively inhibit ice nucleation and water freezing.Fig. 2Dependence of water freezing efficiency on the volume and viscosity of sealing agents. **a** Effect of sealing oil (MO) volume on *f*~f~ post 1-day DSC at different temperatures. *n* = 6, *N* = 70. **b** Side view of corresponding samples of **a**. MO includes Oil Red O for staining and imaging. 0 oil, 0.01 ml, 0.1 ml, 0.15 ml, 0.5 ml, and 3.5 ml indicate no seal, ring seal, partial seal, critical seal (just complete surface seal), standard seal, and over seal by MO, respectively. **c** Effect of viscosity of sealing agents on *f*~f~ post 1-day DSC at −16 °C. The sealing agents are hydroxy (OH) terminated polydimethylsiloxane (PDMS) of different chain lengths and viscosities. *n* = 5, *N* = 56. Error bars represent standard deviations
We also observed that oil addition beyond the critical sealing has a statistically negligible effect on freezing suppression. This indicates that additional pressure and dampening effects, associated with a long-column of viscous oil phase, have a negligible effect on freezing inhibition. In order to further test this, we examined the effects of viscosity of the sealing agents where we used hydroxy (-OH) terminated polydimethylsiloxane (PDMS) of different chain lengths (Fig. [2c](#Fig2){ref-type="fig"}). In a similar fashion, we did not observe statistically significant differences in the capacity of freezing inhibition of PDMS with a viscosity range of 1--5 × 10^5^ cP, with the exception of 3500 cP PDMS that has almost no freezing suppression effect. We hypothesize that this odd behavior is likely due to the formation of an ordered structure between water and this particular PDMS on the interface through hydrogen bonding, which closely matches the lattice of hexagonal ice^[@CR35]^.
Water deep supercooling via surface sealing with alkanes and alcohols {#Sec5}
---------------------------------------------------------------------
Most oils are complex mixtures of alkanes, saturated cyclic alkanes, alkylated aromatic groups, and fatty acids among other hydrocarbon compounds. In an effort to more systematically study the observed freezing inhibition effect of supercooled water sealed with an immiscible hydrocarbon phase, we studied two prototypical families of hydrocarbons: linear alkanes and their corresponding primary alcohols of different lengths (Fig. [3](#Fig3){ref-type="fig"}). Specifically, we have studied alkanes (C~*m*~H~2*m*+2~, denoted C~*m*~, *m* = 5 -- 11) and primary alcohols (C~*m*~H~2*m*+1~OH, denoted C~*m*~OH, *m* = 4 -- 8) as the sealing agents for DSC water at −20 °C. Since linear alkanes have very low polarity, they have weak interaction with polar water molecules. On the other hand, the primary alcohols, which are amphipathic, can form strong hydrogen bonds with water through their hydroxyl group (hydrophilic end) and even stable ordered interfacial structures. However, a binary combination of an alkane and an alcohol should not be utilized as a sealing agent as it forms cooperative hydrogen bonding, heterogenous microdomains, and interfaces^[@CR36],[@CR37]^, pumping water into the mixture to produce milky emulsions on top of water (Supplementary Figure [5](#MOESM1){ref-type="media"}).Fig. 3Deep-supercooled water sealed with linear alkanes and primary alcohols. **a** *f*~f~ of 1 ml deep-supercooled (DSC) water at −20 °C. *n* = 7, *N* = 87. Error bars represent standard deviations. When *m* \> 11 for linear alkanes and *m* \> 8 for primary alcohols, the sealing agents are frozen at −20 °C and cause DSC water frozen. When *m* \< 5, the linear alkanes are gaseous under atmospheric condition and not suitable for sealing. When *m* \< 4, the primary alcohols are miscible with water and not suitable for sealing either. **b**, **c** Schematic configurations of alkane/water (**b**) and alcohol/water interface (**c**), respectively. The alkane and alcohol molecules are displayed without aliphatic hydrogen atoms and colored in light green. The O and H atoms in hydroxyl group of alcohol and water are shown in red and white dots, respectively
We found that *f*~f~ of DSC water, at −20 °C, sealed with alkanes decreases monotonically with increasing carbon number *m* and chain length *l* (Fig. [3a](#Fig3){ref-type="fig"}). The capacity of alkanes in freezing inhibition matches that of MO (Fig. [2a](#Fig2){ref-type="fig"} at −20 °C) at *m* \> 9. This coincides with the observation that mineral oils tend to have hydrocarbons with alkane chain lengths above 10. While oils comprise of many different hydrocarbons, alkanes make up a major fraction of their composition. Accordingly, the trend of higher freezing inhibition with longer alkane chain lengths, might also partially explain the differences in *f*~f~ between PO and MO (Fig. [1d](#Fig1){ref-type="fig"}) among other effects from other hydrocarbon groups that we have not yet studied. PO likely consists of longer carbon chain alkanes than MO based on their densities (PO \~ 0.855 -- 0.88 vs MO \~ 0.838 g ml\^−1) and dynamic viscosities (PO \~34 vs MO \~23 cP^[@CR38]^). On a molecular level, the mechanism for this trend might lie in the structure of the alkane/water interface. It has been observed that an interfacial electron depletion layer with a thickness *δ* exists between water and hydrophobic alkane chains by both X-ray reflectivity (XR) measurements^[@CR39]--[@CR41]^ and atomistic molecular dynamics (MD) simulations^[@CR42],[@CR43]^. The few water molecules in the depletion layer (electron density \< 40% that of bulk water^[@CR44]^) can buckle in the intermolecular space near the ends of alkane molecules (Fig. [3b](#Fig3){ref-type="fig"}), and create a template for the formation of an ice nucleus^[@CR29]^. The alkane chains adjacent to the water molecules preferentially have their longest axis parallel to the water interface with a tilt angle *β*^[@CR39]^. This tilt angle increases with *m* and *l*, resulting in a more parallel orientation for longer alkanes^[@CR39]^. Accordingly, longer alkane chains are expected to reduce the corrugation and roughness of the interface on the side of alkanes. This, consequently, would decrease the number of buckled water molecules and nucleus templates, and thus lower the probability of heterogeneous ice nucleation on that layer^[@CR29]^. These expectations are in line with our observations of decreasing freezing frequencies for longer alkane chain lengths. From the perspective of thermodynamics, longer and flatter-oriented alkanes results in fewer and sparser buckled water molecules in the interface serving as nucleation template, which implies smaller contact region between ice embryo and sealing alkanes, smaller *θ*~iwo~ (even though they are already much smaller than *θ*~iwa~), and higher energy barrier for heterogeneous ice nucleation (Supplementary Figure [1b](#MOESM1){ref-type="media"}).
On the other hand, *f*~f~ of DSC water, at −20 °C, sealed with alcohols increases with *m* and *l*. For example, *f*~f~ equals 4.4% and 21.4% after 1-day DSC for C~4~OH and C~8~OH, respectively. C~4~OH has a higher freezing inhibition capacity compared to C~5~OH (*f*~f~ = 4.4% for C~4~OH, *f*~f~ =12.8% for C~5~OH sealing). Nevertheless, C~4~OH has a small but relatively higher solubility in water than C~5~OH, and accordingly C~5~OH might be the optimal choice for sealing. The different behavior of alcohols with respect to chain length might be due to the different structures of the alcohol/water interface compared to that of alkane/water interface (Fig. [3c](#Fig3){ref-type="fig"}). Unlike the alkanes which prefer a parallel orientation, the primary alcohols orient perpendicularly to the interface with a small *β* (usually less than 30°)^[@CR45],[@CR46]^. The primary alcohols align their hydroxyl (-OH) heads toward the interface to form hydrogen bonds with water molecules. Accordingly, no depletion layer of interfacial water exists as in the alkane/water interface. The 2D layer of interfacial water molecules are strongly hydrogen-bonded to the hydroxyl groups, with their H atoms pointing toward alcohol as revealed by heterodyne-detected vibrational sum frequency spectroscopy (SFG)^[@CR47]^. Therefore, structures and dimensions of the contacting layer of amphilic alcohols essentially determine the distribution and arrangement of interfacial water molecules, and the formation of heterogeneous ice nucleus^[@CR35],[@CR48]^.
Experimental measurements via grazing incidence X-ray diffraction (GIXD) and MD simulation of ice nucleation in droplets under monolayers of long primary alcohol chains with 16 ≤ *m* ≤ 31, revealed a very low tilt angle *β* (\~7.5 -- 12°) and a very good lattice match between hexagonal ice and the alcohol structure for 29 ≤ *m* ≤ 31^[@CR45],[@CR46],[@CR49]^, resulting *T*~f~ as high as −1 °C for these longest chains. As *m* and *l* decrease, *β* increases up to \~19° for *m* = 16^[@CR46]^. In conjunction, a greater lattice mismatch between hexagonal ice lattice and ordered alcohol layer at the interface along with a lower ice nucleation efficiency and *T*~f~ were observed^[@CR35],[@CR45],[@CR46],[@CR49]^. For shorter alcohols (4 ≤ *m* ≤ 8) in this study, larger tilt angles would ensue as evidenced by *β* = 28° for *m* = 6 and *β* = 30° for *m* = 5^[@CR46]^, causing greater lattice mismatches between hexagonal ice and ordered alcohol structure given the general structural similarity of primary alcohols. Compared to longer alcohol chains, the interfacial -OH groups anchored to smaller alcohols have stronger in- and out-of-plane fluctuations at the same temperature. We, therefore, expect that lattice mismatch and the --OH group fluctuations can destabilize any ordered domaine of crystalline water and impede the formation of ice nucleus of critical size^[@CR49]^. Given that both effects are larger with smaller chain lengths, we expect that higher nucleation inhibition can be achieved by smaller primary alcohols, in line with our experimental observations. From the perspective of thermodynamics, greater lattice mismatch and interface fluctuation associated with shorter alcohol molecules directly reduce the probability of the formation of icing template of critical size for successful nucleation, which indicates smaller stable contact area between ice nucleus and sealing alcohols and thus, higher free energy barrier for heterogeneous ice nucleation at the interface. Once again, we observed that there is no significant difference of *f*~f~ between 1-day and 7-day storage when sealed by either alkanes or alcohols. This further suggests the case-specific, rather than stochastic, nature of water freezing with oil sealing that we have previously discussed.
Stability tests for deep-supercooled water {#Sec6}
------------------------------------------
Having established the efficacy of the DSC approach using either oils or pure alkane and alcohol phases, we then studied its stability under vibrational, thermal, and ultrasonic disturbances. Vibrational disturbances were introduced by placing DSC water onto a shaking plate with various shaking speeds and frequencies. When the DSC water (−20°) is sealed by MO, its *f*~f~ is 0% and 5.6%, respectively, under 0.84 g and 2.1 g centrifugal acceleration (Fig. [4a](#Fig4){ref-type="fig"}), which are much higher than ac/deceleration forces of a commercial airliner (0.2 -- 0.4 g) during potential transporation. Thermal disturbances were induced by putting the DSC samples into 37 °C incubator or plunging them into 37 °C water bath with warming rate of 10^0^ C min^-1^ (heated by natural convection in air) or 10^2^ C min^−1^ (heated by forced convection in water), respectively. Very few (0% for gas warming, 2.5% for water warming) of the samples freeze under these thermal fluctuations. In contrast, these samples cannot endure ultrasonication in 40 kHz ultrasonic water bath (Fig. [4a](#Fig4){ref-type="fig"} and Supplementary Movie [1](#MOESM3){ref-type="media"}), with *f*~f~ of \~ 84%. This is probably due to the vigorous collapse of cavitation bubbles in water during ultrasonication^[@CR18]^, which would cause ultrahigh local pressure (\>1 GPa)^[@CR50]^, and therefore, significantly increase equilibrium temperature *T*~e~ and the degree of supercooling Δ*T*.Fig. 4Stability tests for 1 ml deep-supercooled water at −20 °C. **a** *f*~f~ of deep-supercooled (DSC) water sealed by MO under various disturbances. Vibrational disturbance was imposed by shaking plate with different shaking frequencies and centrifugal forces (i.e., 0.84 g or 2.1 g). Thermal disturbance was imposed by placing or plunging the DSC tubes into 37 °C incubator (37 °C gas) or water bath (37 °C water). Ultrasonic disturbance was introduced by putting the DSC tubes into 40 kHz ultrasonic water bath. *n* = 6, *N* = 48. Error bars represent standard deviations. **b** *f*~f~ of DSC water sealed by linear alkanes and primary alcohols under 40 kHz ultrasonic disturbance. *n* = 3, *N* = 24 (except for C~5~, *N* = 8). **c** Representative image sequences of ultrasonication tests for DSC water sealed by linear alkanes or primary alcohols
Upon the instability of DSC water sealed by MO under ultrasonication, we further tested its stability sealed by pure alkanes and primary alcohols. DSC water sealed by alkanes freeze immediately upon being ultrasonicated (Fig. [4](#Fig4){ref-type="fig"}b--c and Supplementary Movie [2](#MOESM4){ref-type="media"}), which is consistent with previous observation of MO sealed water since MO has a high content of various alkanes. On the contrary, none of the samples freeze upon ultrasonification if they were sealed by any of the primary alcohols (Fig. [4b](#Fig4){ref-type="fig"}). Instead, the sealing alcohols would be emulsified with supercooled water, starting from the interface and then evolving toward supercooled water (Fig. [4c](#Fig4){ref-type="fig"} and Supplementary Movie [3](#MOESM5){ref-type="media"}). The exact mechanism of the freezing resistance of DSC water sealed by alcohols to ultrasonic disturbance is still unknown, and one hypothesis would be that ultrasound preferentially transduces its energy into joint molecular motion at interface due to the hydrogen bonding between water and amphilic alcohols to form nanoemulsion^[@CR51]^, rather than cavitation bubbles for ice nucleation in DSC water.
Deep supercooling of hRBCs for extended preservation {#Sec7}
----------------------------------------------------
In addition to DSC of pure water, we have utilized the surface sealing method to achieve DSC for aqueous solutions and cell suspensions to demonstrate extended supercooling preservation of biological specimens (Fig. [5](#Fig5){ref-type="fig"}). The current clinical standard for preservation of hRBCs is via conventional cold storage at 4 °C with the CP2D + AS-3 (anticoagulant citrate phosphate double dextrose supplemented with additive solution 3) solution^[@CR52]^. This standard approach for hRBC preservation can provide storage for a maximum of 42 days^[@CR53]^, beyond which the cells experience irreversible storage lesions including hemolysis as shown in Fig. [5a](#Fig5){ref-type="fig"}. According to the Arrhenius relationship, preservation at deep subzero temperatures would slow down cellular metabolism and decay rates, extending this biopreservation period. Accordingly, we have preserved hRBCs at as low as −16 °C for an extended period of 100 days. Specifically, using surface sealing by PO, we successfully supercooled 1 ml suspensions of hRBCs in either conventional cold storage solution, CP2D + AS−3^[@CR52]^, or University of Wisconsin solution supplemented with 5% (w/v) trehalose (UW + Tre), for as long as 100 days at −7, −10, −13, and even −16 °C (see Supplementary Figure [6](#MOESM1){ref-type="media"} for *f*~f~). The suspension volume could be extended to the magnitudes of 10^1^ even 10^2^ ml as 5 out of 6 vials of 30 ml hRBC suspension (500 million cells) kept unfrozen over 365-day supercooling at −13 °C in our pilot experiments. In CP2D + AS-3 solution, hRBCs experience remarkable hemolysis as shown by the presence of dark spots (RBC debris without holding hemoglobin) in the micrographs of Fig. [5a](#Fig5){ref-type="fig"} and reduced recovery rates of hemoglobin in Fig. [5b, c](#Fig5){ref-type="fig"}, especially at deep subzero temperatures. Surviving hRBCs become spherically shaped, losing the distinctive biconcave disc form of fresh hRBCs (Supplementary Figure [7](#MOESM1){ref-type="media"}). However, in UW + Tre solution, hRBCs do not undergo noticeable hemolysis, and the recovery rates of hemoglobin at −16 °C (94%) is higher than that of conventional storage at 4 °C in CP2D + AS-3 solution post 100-day storage (76%, Fig. [5c](#Fig5){ref-type="fig"}). In addition, they can also maintain their intact, shiny, and discoid-shaped phenotype at DSC temperatures (the second and fourth rows of Fig. [5a](#Fig5){ref-type="fig"}), resulting in higher percentages of normal morphology (\>88%, exclusion of serrated, spherical, swollen, or shrunk cells) than cold storage at 4 °C (49%, Fig. [5d](#Fig5){ref-type="fig"}). These results demonstrate that DSC via surface sealing can effectively prolong the storage time of hRBCs to 100 days combined with the optimization of preservation solutions.Fig. 5Cold storage and deep-supercooled preservation of human red blood cells. **a** Preservation of human red blood cells (hRBCs) for 42 and 100 days at various temperatures in conventional anticoagulant citrate phosphate double dextrose (CP2D) supplemented with additive solution 3 (AS-3) (CP2D + AS-3), and UW solution supplemented with 5% (w/v) trehalose (UW + Tre). Black dots are the debris of lysed RBCs without holding hemoglobin. Successful DSC for 1 ml suspension of 10 million hRBCs in polystyrene tubes was achieved by surface sealing with 0.5 ml PO. Scale bar is 50 µm. **b** Hemoglobin recovery rate of RBCs post 42-day storage. *n* = 3. **c** Hemoglobin recovery rate of RBCs post 100-day storage. *n* = 3. **d** The percentages of RBCs with normal morphology and size post 100-day storage in UW + Tre solution. Bright disk-shaped RBCs with smooth plasma membrane and diameter 5 µm ≤ D ≤ 10 µm are regarded as normal morphology. *n* = 3. The total number of counted cells *N* = 1800 ± 50 for each case. Error bars represent standard deviations
Discussion {#Sec8}
==========
In the preceding, we demonstrated a seemingly counterintuitive approach to achieve long-term DSC of large volume water by using a hydrocarbon-based immiscible phase to seal the water surface. Our initial observations with laboratory grade oils demonstrated that replacing the water/air interface, which is the primary ice nucleation site, with a water/oil interface dramatically inhibits stochastic freezing processes. The seemingly time independent nature of the freezing frequency of oil-sealed water suggests that its freezing might be case-specific rather than stochastic. Our studies with linear alkanes and primary alcohols suggest that freezing inhibition can be achieved by surface sealing with starkly different interfacial structures and microscopic mechanisms, which results in different trends of inhibition capacity correlated to the chain length. While all sealed DSC water show great stability under vibrational and thermal disturbances simulating normal storage and transportation conditions, only the primary alcohol sealed supercooled water can withstand ultrasonication. While we have hypothesized about possible macroscopic (thermodynamic) and microscopic mechanisms that might explain our observations, further studies are warranted to test, confirm, and improve upon them. In addition, electric field on interfacial water introduced by surface sealing by alkanes and alcohols also could contribute to the inhibition or promotion of ice nucleation^[@CR54]^. But the direction of electrical field ("positive" or "negative") and induced orientation of water molecules ("hydrogen down" or "hydrogen up") are not consistent even contradictory according to SFG measurements and MD simulations^[@CR40],[@CR47],[@CR55]^, whose effects on water freezing require further investigation. Especially, since most existing literature focuses on longer alkane and alcohol chains at the water interfaces and the resulting molecular structures, computational and experimental studies with short chains might prove useful. Similarly, careful measurements of interfacial properties and structures at low temperatures with mixed and pure hydrocarbons can shed further light on why some oils are more effective than others.
Given that monolayers of primary linear alcohols of long chains (*m* ≥ 16) have been historically used to initiate ice nucleation, our results with the short chains to prevent nucleation expand the use of alcohols to provide a robust control mechanism over the temperature at which nucleation can be achieved in an aqueous solution. Further studies with different families of hydrocarbons and their mixtures will be aimed to expand this robust control of supercooling to enable various applications. Beyond its fundamental implications, DSC of large volumes of aqueous solutions can enable previously prohibitive applications, and provide unprecedented biopreservation methodologies for cell, tissue, and organ engineering and transplantation, as well as other areas, such as food preservation. Given our prior experience and interest in both organ and cell preservation using supercooling and the limitations we have previously encountered in terms of temperatures, volumes and durations for preservation, we believe that the DSC via the surface sealing with immiscible phases will be vital in advancing these applications forward. The immediate goal is to translate this approach to preservation of other types of cells that are amenable to DSC preservation as we demonstrated for hRBCs here, and then translate such results to the clinic. We will then explore tissue and organ preservation with DSC approaches.
Methods {#Sec9}
=======
Experimental materials {#Sec10}
----------------------
For all experiments in this study, DNase/RNase-free distilled water (Life Technologies/Thermo Fisher Scientific, USA) was used to minimize potential pollutants or ice-nucleating agents, except DSC trials of 100 ml water where deionized (DI) water (resistivity *R* = 18.2 MΩ) produced by a deionizing water system (METTLER TOLEDO Thornton, USA) was used. All water containers (dishes, 96-well plates, round-bottomed tubes, and bottles, Corning, USA) were made of polystyrene, and clean and sterile before experiments. All oil phases, such as light mineral oil (MO), heavy paraffin oil (PO), olive oil (OO), nutmeg oil (NO), alkanes, and alcohols, used for water surface sealing were purchased from Sigma-Aldrich, USA, and their purities were at least 99%.
Water supercooling procedures {#Sec11}
-----------------------------
The loading of water into containers was performed in a chemical hood to avoid contamination of the samples by pollutants or dust particles in the air. Water of small volume (\< 1 ml) was loaded into containers (dishes or 96-well plates) using clean and sterile tips (Thermo Fisher Scientific, USA) and calibrated pipets (PIPETMAN, Gilson, USA), while that of large volume (≥ 1 ml) was loaded into containers (round-bottomed tubes or bottles) using serological pipets (Thermo Fisher Scientific, USA) by pipette filler (Drummond Scientific, USA). We note that as water droplets smaller than 10^0^ µl are subject to significant evaporation during long-term DSC experiments, and those bigger than 10^5^ µl (100 ml) beyond the volume capacity of the freezing chamber, they were not investigated in this study. After loading water samples into the containers, oil phase was gently added onto the water surface using serological pipets, trickling down along the wall of containers to avoid splashing or trapping air bubbles at the interface. The water-laden containers (with or without sealing oil) were transferred into portable temperature-controlled freezers (Engel MHD-13, Engel, USA) that were placed in 4 °C cold room to minimize temperature fluctuations, or stored in −20 °C freezer (Thermo Fisher Scientific, USA). The temperatures within these freezers were verified by Toluene-filled low-temperature thermometer (Sigma, USA).
Water degassing for supercooling tests {#Sec12}
--------------------------------------
To examine the effects of dissolved air in water on ice nucleation and water freezing, the water was vacuumed at a pressure below 10^−4^ atmosphere for 24 h to extract dissolved air molecules. The degassed water was, then, gently pipetted into tubes and sealed with mineral oil (MO) for supercooling tests at −16 °C. The air content of the degassed water is significantly lower than that of normal water without degassing, as no air bubbles emerge from the degassed water (second row of Supplementary Figure [2b](#MOESM1){ref-type="media"}) under vacuum. The same procedure was carried out for normal water for comparative purposes, and several big air bubbles can be observed after 3-hour degassing (first row of Supplementary Figure [2b](#MOESM1){ref-type="media"}).
Stability tests {#Sec13}
---------------
To test the stability of DSC water sealed by oil phase, three types of disturbances, vibrational, thermal, and ultrasonic disturbances, were studied. For vibrational disturbance test, DSC tubes were placed on shaking plate (Labline 4625 titer shaker, Marshall Scientific, USA) with shaking speed 500 and 800 rpm for 30 s, which give rise to the centrifugal acceleration of 0.84 and 2.1 g (*g* is gravitational acceleration), respectively. To prevent heat transfer, the tubes were wrapped with thick tissue paper in tube racks, all of which had been previously cooled to −20 °C in freezer. The temperature of the DSC water would not change noticeably during experiments given the brief shaking period and thick insulation layer. For thermal disturbance test, the DSC tubes were put into 37 °C incubator (warmed by air) or plunged into 37 °C water bath (warmed by water). Therefore, DSC water would experience different warming rates and temperature gradient. For ultrasonic disturbance test, DSC tubes would be plunged into 4 °C ultrasonic water bath. The sonicator (Branson B-200, TMC Industries, USA) generates 40 kHz ultrasonic wave with power 30 W. The freezing of static DSC water can be determined by visual inspection for the change of sample transparency (from transparent to opaque, Fig. [1b](#Fig1){ref-type="fig"}), while for DSC water in stability tests, the occurrence of freezing can also be determined by tube tilt as frozen water cannot flow freely (Fig. [4c](#Fig4){ref-type="fig"}).
Supercooling preservation for hRBCs {#Sec14}
-----------------------------------
Fresh hRBCs were purchased from Zen-Bio (NC, USA) and utilized immediately upon arrival. They were suspended in either anticoagulant citrate phosphate double dextrose (Haemonetics, MA, USA) supplemented with additive solution 3 (Haemonetics, MA, USA) (CP2D + AS-3), or University of Wisconsin solution (Bridge to Life, SC, USA) supplemented with 5% (w/v) trehalose (Sigma-Aldrich, USA) (UW + Tre), with a concentration of 10 million/1 ml. Next, 1 ml hRBC suspension was transferred into 5 ml round-bottomed polystyrene tube, sealed by 0.5 ml paraffin oil (PO), and stored in portable temperature-controlled freezer for desired periods. Post certain storage time (42 or 100 days), samples were warmed up in 37 °C incubator, followed by the removal of sealing oil via aspiration. Then hRBCs were resuspended into phosphate-buffered saline (PBS) for the measurements of hemoglobin recovery rate and examination of cell morphologies. To release the hemoglobin in intact hRBCs, deionized (DI) water was added in RBCs to burst cells via hypotonic osmolarity, followed by vigorous vortex for 3 min. The concentration of hemoglobin in solution was determined by direct spectrophotometry, utilizing formulas employing absorbance at 415, 380, and 450 nm ("Allen correction" formula^[@CR56],[@CR57]^) i.e., *C*~he~ (mg L\^−1) = 1.68*A*~415~--0.84*A*~380~--0.84*A*~450~, where Che is the concentration of hemoglobin, *A*~415~, *A*~380~, and *A*~450~ are the absorbance values at 415, 380, and 450 nm, respectively. The recovery rate of hemoglobin in hRBCs for each sample was calculated by *η* = *C*~he,pe~/(*C*~he,pe~ + *C*~he,su~) × 100%, where *η* is the recovery ratio of hemoglobin in RBCs, *C*~he,pe~, and *C*~he,su~ are hemoglobin concentrations of lysed RBCs in the pellet (hemoglobin is not released by hemolysis during storage or processing) and supernatant solution of RBCs (hemoglobin released during storage or processing). In addition, hRBCs suspended in PBS were observed and imaged by phase-field microscopy. To obtain the rate of normal morphology (bright disk-shaped cells with smooth membrane rather than serrated, spherical, swollen, or shrunk cells) in preserved hRBCs, around 600 cells (total 1800, *n* = 3) were counted and the rate of normal morphology was calculated.
Data analysis {#Sec15}
-------------
All data were organized and reported as the mean ± standard deviation from at least three independent runs of experiments (*n* \> 3); further information on sample numbers are disclosed in figure captions. The statistical significance of mean values between two groups was determined by Microsoft Excel based on Student's two-tailed *t*-test, assuming equal variance. Although a *p*-value less than 0.05 is generally regarded as statistically significant, different ranges of *p*-value (NS: 0.05 \< *p*, \*0.005 \< *p* \< 0.05, \*\*0.005 \< *p* \< 10^−6^, \*\*\**p* \< 10^−6^) were provided to show different degrees of significance.
Data availability {#Sec16}
-----------------
All relevant data are available from the authors upon request.
Electronic supplementary material
=================================
{#Sec18}
Supplementary Information Peer Review File Description of Additional Supplementary Files Supplementary Movie 1 Supplementary Movie 2 Supplementary Movie 3
**Publisher\'s note:** Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
**Change history**
9/5/2018
This Article was originally published without the accompanying Peer Review File. This file is now available in the HTML version of the Article; the PDF was correct from the time of publication.
Electronic supplementary material
=================================
**Supplementary Information** accompanies this paper at 10.1038/s41467-018-05636-0.
We are grateful to Prof. Mehmet Toner for the very helpful discussions and his suggestions. We would also like to thank the NIH for funding this work through grants no. 5P41EB002503 (BioMEMS Resource Center), 1R21EB020192, 5R01EB023812.
H.H., M.L.Y. and O.B.U. conceived the project. H.H performed experiments and wrote the manuscript draft. All authors analyzed the data and revised the manuscript.
Competing interests {#FPar1}
===================
The authors declare no competing interests.
| {
"pile_set_name": "PubMed Central"
} |
1.. Introduction {#sec1}
==================
The technique of single-crystal X-ray diffraction enables the routine and precise structure determination of biological macromolecules at high resolution (Blow, 2002[@bb9]; Rupp, 2010[@bb51]). It has been applied extensively to proteins, DNA and RNA, with the PDB recently celebrating the amazing milestone of 100 000 deposited structures. This is essentially the result of a multitude of innovations and technological developments spanning the last few decades (Abola *et al.*, 2000[@bb1]; Fersht, 2008[@bb22]). Nonetheless, behind this success hides the struggle to produce purified samples, to obtain crystals with suitable diffraction quality and to subsequently reproduce/optimize them (Bergfors, 2009[@bb7]; Khurshid *et al.*, 2014[@bb31]).
The poor yield of suitable crystals can be explained by the concept called the 'curse of dimensionality': there are so many dimensions associated with the parameter space to be explored that it is problematic or impossible to perform an analysis that has any statistical significance. The underlying reason is that the combinations of reagents employed in crystallization trials alter the combinations of variables associated with the main parameters of crystallization (McPherson *et al.*, 1995[@bb39]); for example, the variables that are related to the nature of the protein and the experiment, the type of protein--protein interactions and so on. Hence, initial crystallization screening is a stochastic process (Carugo & Argos, 1997[@bb13]; Lomakin *et al.*, 1999[@bb33]) that usually requires various approaches and conditions.
A crystallization condition traditionally contains a precipitant, a buffer and an additive. There are now hundreds of well known crystallization reagents and the possible combinations used to formulate conditions in a systematic manner has grown to a very large number that cannot be captured in any practical way because the amount of sample and the screening technology are limiting (Carter & Carter, 1979[@bb12]; Gorrec, 2014[@bb28]). As a consequence, many laboratories have chosen an approach with a minimum number of conditions. A widespread minimal approach is to employ a set of conditions selected empirically to form a 'sparse-matrix' screen (Jancarik & Kim, 1991[@bb30]). In the last two decades, advances in automation and the increase in the number of crystal structures solved and deposited have stimulated the optimization of sparse-matrix screens, mostly in the form of sets of 96 conditions, as this is an automation-friendly format (Kimber *et al.*, 2003[@bb32]; Rupp & Wang, 2004[@bb52]; Newman *et al.*, 2005[@bb55]; Stock *et al.*, 2005[@bb54]; Newstead *et al.*, 2008[@bb42]; Fazio *et al.*, 2014[@bb21]).
Nevertheless, we have argued that the minimal approach may mean undersampling of conditions and therefore essential hits could be missed (Gorrec, 2013[@bb27]). Subsequently, novel formulations should still be investigated when possible. Besides, screens should evolve in parallel with the increasing complexity of the samples and the technical difficulties encountered during the process of structure determination. Notably, the demands of cryo-crystallography (Petsko, 1975[@bb46]; §[](#sec3.1.1){ref-type="sec"}3.1.1) as well as current and future solutions to the phase problem (Taylor, 2003[@bb56]; §[](#sec3.1.2){ref-type="sec"}3.1.2) should be taken into account.
Previously, we presented an innovative approach in a screen formulation called MORPHEUS (Gorrec, 2009[@bb26]). In order to reduce bias towards a subset of samples, the conditions were formulated *de novo* by integrating a larger number of reagents than traditionally employed. For this, novel mixes of reagents were investigated. Reagents that aided protein stabilization, crystallization and crystal cooling were included in the final formulation. The multiplexing of reagents has been performed previously for cryoprotectants (Garman & Mitchell, 1996[@bb24]), precipitants (Majeed *et al.*, 2003[@bb36]), buffer systems (Newman, 2004[@bb40]) and additives (McPherson & Cudney, 2006[@bb38]) (§[](#sec3.1.3){ref-type="sec"}3.1.3). The original MORPHEUS combined all of these innovations. It is worth highlighting two other particular design principles of MORPHEUS. Firstly, the inclusion of PDB-derived small molecules (as potential ligands) that were gathered into mixes of additives, sorted according to their chemical nature to avoid incompatibilities. Secondly, mixes of precipitants, additives and buffers were combined within a 96-condition three-dimensional grid screen using fixed ratios to facilitate easier screen preparation and follow-up optimizations.
Here, we present MORPHEUS II, a 96-condition protein crystallization screen formulated in continuity with previous work. MORPHEUS II follows the design principles of MORPHEUS; however, we introduced less common additives, for example metals that are amendable to the collection of anomalous data sets directly from screening conditions. We also included nondetergent sulfobetaines (NDSBs), polyamines, amino acids and monosaccharides, which are known to enhance the solubility and stability of many proteins. To complete the formulation of MORPHEUS II, four unusual 'glycerol-like' polyols have been added as cryoprotectants to aid flash-cooling. Finally, innovative buffer systems were included as part of the formulations. The suitability of the resulting conditions is shown by the crystallization of eight different protein samples and their efficiency is compared with commercially available conditions (§[](#sec3.2){ref-type="sec"}3.2).
2.. Materials and methods {#sec2}
===========================
2.1.. Screen formulation {#sec2.1}
--------------------------
The mixes of ligands, precipitants and buffers were combined using a fixed ratio of volumes for the stock solutions as employed in the original MORPHEUS screen: 0.5 stock precipitants + 0.1 stock additives + 0.1 stock buffer system + 0.3 water. Methods used to select the PDB-derived ligands, to design the screen and to prepare the stock solutions were also as described previously (Gorrec, 2009[@bb26]). Further details can be found in the Supporting Information concerning the four precipitant mixes (Supplementary Table S1) and the three buffer systems (Supplementary Table S2).
Although the additive-to-protein ratio preferably needs to be maximized (Danley, 2006[@bb17]), the concentrations of the mixes Divalent cations II, Akalis, Oxometalates and Lanthanides had to be lowered compared with other, more soluble and less reactive additives, such as monosaccharides and carboxylic acids. Relatively low concentrations (around 1 m*M*) are suitable for these particular additives according to others (Petsko, 1985[@bb47]; Trakhanov & Quiocho, 1995[@bb59]).
By empirical experimentation, stable and suitable combinations of reagents were found. Unfortunately, some traditional heavy atoms had to be excluded, such as those of the platinum group (chloride salts of platinum, osmium, iridium, ruthenium, rhodium and palladium) since they could not be solubilized and/or were unstable in solution. Of course, they can still be tested later on crystals already formed when necessary. The buffer system was removed from conditions B5--B8 to avoid precipitation (probably owing to formation of a chelate between a divalent cation and one of the corresponding buffers).
Also following an empirical approach, it was found that four small polyols that are not currently found in any commercially available screens vitrified samples during flash-cooling as efficiently as glycerol (*i.e.* typically 20--25% required to cryoprotect conditions): 1,2,4-butanetriol, 1,2,6-hexanetriol, 1,5-pentanediol and 1,1,1-tris(hydroxymethyl)propane. They were thus integrated into the precipitant mixes (Table 1[▸](#table1){ref-type="table"} and Supplementary Table S1). X-ray diffraction tests with the mixes of polyols and polyethylene glycols (PEGs) flash-cooled in cryoloops were then used to adjust the concentration of cryoprotectants and ensured that the resulting diffraction patterns were free of background from ice. NDSBs are another group of reagents often used in sample preparation and crystallization additive screens. They have a relatively low frequency of occurrence in crystal structures, which suggests that their role may be less specific and therefore they have been integrated to the precipitant mixes.
2.2.. Crystallization experiments {#sec2.2}
-----------------------------------
The protein samples can be briefly described as follows: concanavalin A ('Con', molecular weight 27 kDa, concentration 13 mg ml^−1^, Sigma catalogue No. L7647 dissolved in 0.1 *M* Tris pH 8.5), polymerase III clamp--exonuclease complex ('Pol', 80 kDa, 10 mg ml^−1^; Rêgo *et al.*, 2013[@bb58]), ESCRT-II complex ('E2H', 115 kDa, 7 mg ml^−1^; Teo *et al.*, 2004[@bb57]), bar domain ('Bar', 6 mg ml^−1^, 29 kDa; Peter *et al.*, 2004[@bb45]), HIV capsid ('HIV', 25 kDa, 32 mg ml^−1^; Price *et al.*, 2014[@bb48]), coiled-coil domain of the cytosolic nucleic acid sensor LRRFIP1 ('CCD', 12 kDa, 9 mg ml^−1^; Nguyen & Modis, 2013[@bb41]), ubiquitin--protein ligase ('UPL', 21 kDa, 9 mg ml^−1^; Elliott *et al.*, 2014[@bb19]) and mRNA nuclear-export factor complex ('NEF', 68 kDa, 8 mg ml^−1^; Aibara *et al.*, 2015[@bb2]).
The original MORPHEUS screen was purchased from Molecular Dimensions Ltd ('MORPHEUS I', Lot No. 021-1-46) and the sparse-matrix screen 'The JCSG+ Suite' was purchased from Qiagen (Lot No. 54806713). Triplicate droplets of 300 nl final volume with a 2:1 protein-to-condition ratio were formed in MRC plates at 20°C using a Mosquito robot (TTP Labtech). Single droplets of 200 nl final volume (1:1 ratio) were prepared similarly for the 'Bar' sample only. The plates were swiftly sealed and centrifuged (1000 rev min^−1^, 1 min) and then kept at 18°C. Droplets were visualized with a stereo microscope after one week. Only obvious hits were taken into account (*i.e*. drops with crystals larger than 5 µm and with sharp edges). Details of the corresponding crystallization results can be found in the Supporting Information (Supplementary Table S3).
3.. Results and discussion {#sec3}
============================
3.1.. Formulation of MORPHEUS II {#sec3.1}
----------------------------------
The formulations of the 96 MORPHEUS II crystallization conditions are listed in Table 1[▸](#table1){ref-type="table"}. The 35 PDB-derived ligands selected to formulate MORPHEUS II can be found in Table 2[▸](#table2){ref-type="table"}. The recipes for preparing the eight additive mixes are listed in Table 3[▸](#table3){ref-type="table"}.
### 3.1.1.. Integration of cryoprotecting agents {#sec3.1.1}
To obtain vitrification after flash-cooling, crystals are usually soaked briefly in a solution containing a cryoprotectant (most commonly glycerol). Many crystals are lost using this approach owing to the extensive handling and/or resulting variations in the composition of the mother liquor. A logical remedy for these issues is the use of crystallization conditions that are already cryoprotected. From this perspective, the amount of glycerol needed to successfully vitrify the conditions of Jancarik & Kim (1991[@bb30]) was determined by Garman & Mitchell (1996[@bb24]). Later, another study expanded these data with PEG 400, ethylene glycol and 1,2-propanediol (McFerrin & Snell, 2002[@bb37]). Adding the cryoprotectant directly to the formulations seems straightforward; however, the impact of introducing an additional reagent at high concentration on the yield of quality crystals was not investigated.
Cryoprotection should not bias formulations towards only a few cryoprotectants as main reagents since this would contribute to a further undersampling of initial screen conditions. We thought that more cryoprotectants should be tested as part of the development of novel conditions. Therefore, we integrated other polyols into the MORPHEUS II screen. Using polyols as cryoprotectants is beneficial since they are typically easy to handle and they display other interesting properties. For example, polyols are somewhat hygroscopic (Cohen *et al.*, 1993[@bb15]) and hence they also act as precipitants, altering both the hydration of proteins and the kinetics of vapour-diffusion experiments (Forsythe *et al.*, 2002[@bb23]; Collins, 2004[@bb16]). Finally, it should be pointed out that different cryoprotecting solutions cause different degrees of contraction upon flash-cooling and also affect cooling rates (Berejnov *et al.*, 2006[@bb6]; Alcorn & Juers, 2010[@bb3]). These parameters will affect the differential contraction between the macromolecular crystal and the mother liquor surrounding it, and hence the quality of the diffraction data obtained.
### 3.1.2.. Importance of additives {#sec3.1.2}
Additives may alter the parameters of crystallization experiments in a myriad of ways. If used correctly, they can increase the chances of obtaining useful crystals. One approach is to test how reagents alter the stability and solubility of the sample prior to crystallization assays (Ericsson *et al.*, 2006[@bb20]; Izaac *et al.*, 2006[@bb29]). Also, binding to the protein may be investigated (Boggon & Shapiro, 2000[@bb10]). Nevertheless, other essential parameters are specific to crystallization. For example, it has been demonstrated that crystal growth can be altered by additives such as divalent metal cations (Trakhanov & Quiocho, 1995[@bb59]). Therefore, it is probably advisable to integrate as many additives as possible into our initial screen.
The positive impact of additives on the yield of crystals can be explained through the formation of new crystal contacts (Carugo & Djinović-Carugo, 2014[@bb14]). More complex molecules, for example polycarboxylic acids, sugars and polyamines, can bind to pockets in macromolecules and stabilize them or help them to adopt a particular conformation (Arakawa & Timasheff, 1982[@bb5]; Sauter *et al.*, 1999[@bb53]; Maclean *et al.*, 2002[@bb35]).
Polyamines were originally used with polynucleotides as they form favourable electrostatic interactions with DNA and RNA, leading to stable complexes (Bolton & Kearns, 1978[@bb11]; Drew & Dickerson, 1981[@bb18]). However, polyamines are now also regularly observed in complexes with proteins (*e.g.* spermine, PDB residue ID SPM, 105 occurrences in the PDB, Table 2[▸](#table2){ref-type="table"}). Zwitterionic organic chemicals, such as NDSBs (Table 2[▸](#table2){ref-type="table"} and Supplementary Table S1) and also HEPES-like buffers (*e.g.* MES) can be used as additives/buffers for solubilization and may prevent aggregation or polymerization (Vuillard *et al.*, 1996[@bb60]). This may be explained through the abilities of molecules to shield specific apolar surface patches (Pusey *et al.*, 2007[@bb49]).
### 3.1.3.. Heavy atoms {#sec3.1.3}
In order to solve a structure by SAD, some early work suggested that derivatization should be performed by soaking the crystals since crystal growth may be altered by heavy-atom binding and hence prevent lattice contacts or eventually produce a different crystal form (Petsko, 1985[@bb47]). However, the increasingly challenging nature of proteins studied with crystallography means that different crystal forms occur less frequently. In addition, it is well known that when heavy atoms are tested for derivatization by soaking crystals (Garman & Murray, 2003[@bb25]) the diffraction usually worsens or may even be lost (as for cryoprotectants; §[](#sec3.1.1){ref-type="sec"}3.1.1).
Integrating heavy atoms into the initial screen is therefore very desirable. This was demonstrated when crystals of the nitrogen regulation-related protein NreA grew in two similar MORPHEUS conditions but with two different mixes of additives. Subsequently, closely related crystal forms were obtained that contained either iodide (NreA--I; PDB entry [4iuh](http://scripts.iucr.org/cgi-bin/cr.cgi?rm=pdb&pdbId=4iuh)) or nitrate (NreA--NO~3~; PDB entry [4iuk](http://scripts.iucr.org/cgi-bin/cr.cgi?rm=pdb&pdbId=4iuk)) and enabled structure determination *ab initio* with experimental phasing (Niemann *et al.*, 2014[@bb43]). Although we cannot yet share similar results from MORPHEUS II, it is worth mentioning that we did not observe issues with diffuse scattering caused by conditions containing heavy atoms (Luft *et al.*, 2014[@bb34]), probably because of the very small proportion of the corresponding heavy atoms in the samples.
Finally, before thinking about derivatization, our goal was to increase our yield of initial (and novel) crystals. In this context, it is worth mentioning other work suggesting that multivalent metal ions such as yttrium can modulate protein--protein interactions and even mediate crystal contacts that help to form the crystal lattice: they specifically bind to acidic surface patches as well as bridging acidic side chains from neighbouring subunits (Zhang *et al.*, 2011[@bb62]).
### 3.1.4.. Integration of mixes of additives {#sec3.1.4}
Use of mixes of additives during initial crystallization screening has been tested before with success (McPherson & Cudney, 2006[@bb38]). This strategy carries the risk that one component of a mix might have a deleterious effect and thereby mask the positive contribution of another. Nevertheless, if one of the additives from the mix participates in specific effects or interactions, the less specific addtive should be less pronounced. Also, by selecting components that have been regularly observed as ordered parts of crystal structures, the chances of incorporating molecules that play a positive role should be increased (Gorrec, 2009[@bb26]).
Finally, more than one type of additive may be required for crystal growth, as many structures contain multiple additives. This was demonstrated when the original MORPHEUS screen was used to crystallize the human endoplasmic reticulum aminopeptidase HERAP2 (PDB entry [3se6](http://scripts.iucr.org/cgi-bin/cr.cgi?rm=pdb&pdbId=3se6)), where four additives and a buffer component are part of the crystal structure (Birtley *et al.*, 2012[@bb8]).
3.2.. Crystallization experiments {#sec3.2}
-----------------------------------
The overall yield of crystals was 11.3% with JCSG+, 17.2% with MORPHEUS and 16.5% with MORPHEUS II (Supplementary Table S3). Some may argue that our panel of test proteins may be more likely to crystallize under conditions containing high-molecular-weight PEG as the main precipitant (MORPHEUS screens; Page & Stevens, 2004[@bb44]). Different proteins may have specifically required relatively high salt concentrations and hence preferably crystallized in a sparse matrix (JCSG+). Others could also argue the most efficient pH range could have been anticipated (Kantardjieff & Rupp, 2004[@bb71]). Ultimately, the results will strongly depend on the subset of proteins tested.
The underlying problem is the numerous biases participating in the curse of dimensionality, notably those associated with the preparation of the screens/buffers (Wooh *et al.*, 2003[@bb61]), the type/number of protein samples selected (McPherson & Cudney, 2006[@bb38]) *etc*. In the end, no matter how sophisticated the statistical analysis and data mining of crystallization space, any of the approaches will only provide a basis for increasing the probability of crystallization success, but will never guarantee success for any particular protein (Rupp, 2003[@bb50]).
As a consequence, no strict conclusions should be drawn when comparing efficiency between screens. Surely, we have demonstrated the suitability of nontraditional and *de novo* formulated MORPHEUS II conditions for protein crystallization, at least for samples with a propensity to crystallize in conditions with high-molecular-weight PEG precipitant, while two other screens widely used by the protein crystallography community were used as controls. Nevertheless, we hope that MORPHEUS II will efficiently extend the range of available conditions and hence enable the crystallization of recalcitrant samples. A cryoprotected screen certainly reduces reproducibility issues observed during the cryoprotection of crystals.
4.. Conclusions {#sec4}
=================
We have demonstrated that unusual and under-represented reagents can be combined to formulate suitable and useful conditions for protein crystal growth. The resulting screen is based on principles that have proven to be successful previously. The strategy of formulation reduces bias towards a subset of conditions or samples and integrates new mixes of reagents. MORPHEUS II has already increased our overall effectiveness with *de novo* structure determination enabled by including more heavy atoms in our initial screen.
Supplementary Material
======================
Crystallisation results with MORPHEUS II and other tables.. DOI: [10.1107/S2053230X1500967X/en5565sup1.pdf](http://dx.doi.org/10.1107/S2053230X1500967X/en5565sup1.pdf)
I would like to thank Jan Löwe and Olga Perisic for their support in developing innovative screens and the running of the crystallization facility at the Laboratory of Molecular Biology. This work would not have been possible without the support of colleagues who shared samples/results, notably Andrzej Szewczak, Rafael Fernández Leiro, David Jacques, Shintaro Aibara, Yorgo Modis and Paul Elliott. I hereby state a conflicting commercial interest since MRC Technology will commercialize MORPHEUS II under exclusive licence to Molecular Dimensions Ltd, UK. Contact at MRC Technology: Karen Law (<karen.law@tech.mrc.ac.uk>).
###### Formulation of MORPHEUS II
A mix of precipitants includes a high-molecular-weight PEG and a cryoprotectant (small polyol). Two precipitant mixes also include NDSBs. The formulations of the eight additive mixes can be found in Table 3[▸](#table3){ref-type="table"}. The formulations of the three buffer systems can be found in Supplementary Table S2. The roman numeral II was used to distinguish the new mixes from similarly named ones in the original MORPHEUS screen. AMPD, 2-amino-2-methyl-1,3-propanediol; BES, *N*,*N*-bis(2-hydroxyethyl)-2-aminoethanesulfonic acid; bis-tris, bis(2-hydroxyethyl)aminotris(hydroxymethyl)methane; GlyGly, glycylglycine; HEPES, 4-(2-hydroxyethyl)piperazine-1-ethanesulfonate; MES, 2-(*N*-morpholino)ethanesulfonic acid; MOPSO, 3-morpholino-2-hydroxypropanesulfonic acid; TEA, triethanolamine.
Well Mix of precipitants Mix of additives Buffer system
------ ------------------------------------------------------------------------------------------ ------------------------------------ ------------------------------
A1 15%(*w*/*v*) PEG 3K, 20%(*v*/*v*) 1,2,4-butanetriol, 1%(*w*/*v*) NDSB 256 0.03*M* of each LiNaK 0.1*M* MOPSO/bis-tris pH 6.5
A2 12.5%(*w*/*v*) PEG 4K, 20%(*v*/*v*) 1,2,6-hexanetriol 0.03*M* of each LiNaK 0.1*M* MOPSO/bis-tris pH 6.5
A3 10%(*w*/*v*) PEG 8K, 20%(*v*/*v*) 1,5-pentanediol 0.03*M* of each LiNaK 0.1*M* MOPSO/bis-tris pH 6.5
A4 5%(*w*/*v*) PEG 20K, 25%(*w*/*v*) 1,1,1-tris(hydroxymethyl)propane, 1%(*w*/*v*) NDSB 195 0.03*M* of each LiNaK 0.1*M* MOPSO/bis-tris pH 6.5
A5 15%(*w*/*v*) PEG 3K, 20%(*v*/*v*) 1,2,4-butanetriol, 1%(*w*/*v*) NDSB 256 0.03*M* of each LiNaK 0.1*M* BES/TEA pH 7.5
A6 12.5%(*w*/*v*) PEG 4K, 20%(*v*/*v*) 1,2,6-hexanetriol 0.03*M* of each LiNaK 0.1*M* BES/TEA pH 7.5
A7 10%(*w*/*v*) PEG 8K, 20%(*v*/*v*) 1,5-pentanediol 0.03*M* of each LiNaK 0.1*M* BES/TEA pH 7.5
A8 5%(*w*/*v*) PEG 20K, 25%(*w*/*v*) 1,1,1-tris(hydroxymethyl)propane, 1%(*w*/*v*) NDSB 195 0.03*M* of each LiNaK 0.1*M* BES/TEA pH 7.5
A9 15%(*w*/*v*) PEG 3K, 20%(*v*/*v*) 1,2,4-butanetriol, 1%(*w*/*v*) NDSB 256 0.03*M* of each LiNaK 0.1*M* GlyGly/AMPD pH 8.5
A10 12.5%(*w*/*v*) PEG 4K, 20%(*v*/*v*) 1,2,6-hexanetriol 0.03*M* of each LiNaK 0.1*M* GlyGly/AMPD pH 8.5
A11 10%(*w*/*v*) PEG 8K, 20%(*v*/*v*) 1,5-pentanediol 0.03*M* of each LiNaK 0.1*M* GlyGly/AMPD pH 8.5
A12 5%(*w*/*v*) PEG 20K, 25%(*w*/*v*) 1,1,1-tris(hydroxymethyl)propane, 1%(*w*/*v*) NDSB 195 0.03*M* of each LiNaK 0.1*M* GlyGly/AMPD pH 8.5
B1 15%(*w*/*v*) PEG 3K, 20%(*v*/*v*) 1,2,4-butanetriol, 1%(*w*/*v*) NDSB 256 0.5m*M* of each Divalent cation II 0.1*M* MOPSO/bis-tris pH 6.5
B2 12.5%(*w*/*v*) PEG 4K, 20%(*v*/*v*) 1,2,6-hexanetriol 0.5m*M* of each Divalent cation II 0.1*M* MOPSO/bis-tris pH 6.5
B3 10%(*w*/*v*) PEG 8K, 20%(*v*/*v*) 1,5-pentanediol 0.5m*M* of each Divalent cation II 0.1*M* MOPSO/bis-tris pH 6.5
B4 5%(*w*/*v*) PEG 20K, 25%(*w*/*v*) 1,1,1-tris(hydroxymethyl)propane, 1%(*w*/*v*) NDSB 195 0.5m*M* of each Divalent cation II 0.1*M* MOPSO/bis-tris pH 6.5
B5 15%(*w*/*v*) PEG 3K, 20%(*v*/*v*) 1,2,4-butanetriol, 1%(*w*/*v*) NDSB 256 0.5m*M* of each Divalent cation II
B6 12.5%(*w*/*v*) PEG 4K, 20%(*v*/*v*) 1,2,6-hexanetriol 0.5m*M* of each Divalent cation II
B7 10%(*w*/*v*) PEG 8K, 20%(*v*/*v*) 1,5-pentanediol 0.5m*M* of each Divalent cation II
B8 5%(*w*/*v*) PEG 20K, 25%(*w*/*v*) 1,1,1-tris(hydroxymethyl)propane, 1%(*w*/*v*) NDSB 195 0.5m*M* of each Divalent cation II
B9 15%(*w*/*v*) PEG 3K, 20%(*v*/*v*) 1,2,4-butanetriol, 1%(*w*/*v*) NDSB 256 0.5m*M* of each Divalent cation II 0.1*M* GlyGly/AMPD pH 8.5
B10 12.5%(*w*/*v*) PEG 4K, 20%(*v*/*v*) 1,2,6-hexanetriol 0.5m*M* of each Divalent cation II 0.1*M* GlyGly/AMPD pH 8.5
B11 10%(*w*/*v*) PEG 8K, 20%(*v*/*v*) 1,5-pentanediol 0.5m*M* of each Divalent cation II 0.1*M* GlyGly/AMPD pH 8.5
B12 5%(*w*/*v*) PEG 20K, 25%(*w*/*v*) 1,1,1-tris(hydroxymethyl)propane, 1%(*w*/*v*) NDSB 195 0.5m*M* of each Divalent cation II 0.1*M* GlyGly/AMPD pH 8.5
C1 15%(*w*/*v*) PEG 3K, 20%(*v*/*v*) 1,2,4-butanetriol, 1%(*w*/*v*) NDSB 256 1m*M* of each Alkali 0.1*M* MOPSO/bis-tris pH 6.5
C2 12.5%(*w*/*v*) PEG 4K, 20%(*v*/*v*) 1,2,6-hexanetriol 1m*M* of each Alkali 0.1*M* MOPSO/bis-tris pH 6.5
C3 10%(*w*/*v*) PEG 8K, 20%(*v*/*v*) 1,5-pentanediol 1m*M* of each Alkali 0.1*M* MOPSO/bis-tris pH 6.5
C4 5%(*w*/*v*) PEG 20K, 25%(*w*/*v*) 1,1,1-tris(hydroxymethyl)propane, 1%(*w*/*v*) NDSB 195 1m*M* of each Alkali 0.1*M* MOPSO/bis-tris pH 6.5
C5 15%(*w*/*v*) PEG 3K, 20%(*v*/*v*) 1,2,4-butanetriol, 1%(*w*/*v*) NDSB 256 1m*M* of each Alkali 0.1*M* BES/TEA pH 7.5
C6 12.5%(*w*/*v*) PEG 4K, 20%(*v*/*v*) 1,2,6-hexanetriol 1m*M* of each Alkali 0.1*M* BES/TEA pH 7.5
C7 10%(*w*/*v*) PEG 8K, 20%(*v*/*v*) 1,5-pentanediol 1m*M* of each Alkali 0.1*M* BES/TEA pH 7.5
C8 5%(*w*/*v*) PEG 20K, 25%(*w*/*v*) 1,1,1-tris(hydroxymethyl)propane, 1%(*w*/*v*) NDSB 195 1m*M* of each Alkali 0.1*M* BES/TEA pH 7.5
C9 15%(*w*/*v*) PEG 3K, 20%(*v*/*v*) 1,2,4-butanetriol, 1%(*w*/*v*) NDSB 256 1m*M* of each Alkali 0.1*M* GlyGly/AMPD pH 8.5
C10 12.5%(*w*/*v*) PEG 4K, 20%(*v*/*v*) 1,2,6-hexanetriol 1m*M* of each Alkali 0.1*M* GlyGly/AMPD pH 8.5
C11 10%(*w*/*v*) PEG 8K, 20%(*v*/*v*) 1,5-pentanediol 1m*M* of each Alkali 0.1*M* GlyGly/AMPD pH 8.5
C12 5%(*w*/*v*) PEG 20K, 25%(*w*/*v*) 1,1,1-tris(hydroxymethyl)propane, 1%(*w*/*v*) NDSB 195 1m*M* of each Alkali 0.1*M* GlyGly/AMPD pH 8.5
D1 15%(*w*/*v*) PEG 3K, 20%(*v*/*v*) 1,2,4-butanetriol, 1%(*w*/*v*) NDSB 256 0.5m*M* of each Oxometalate 0.1*M* MOPSO/bis-tris pH 6.5
D2 12.5%(*w*/*v*) PEG 4K, 20%(*v*/*v*) 1,2,6-hexanetriol 0.5m*M* of each Oxometalate 0.1*M* MOPSO/bis-tris pH 6.5
D3 10%(*w*/*v*) PEG 8K, 20%(*v*/*v*) 1,5-pentanediol 0.5m*M* of each Oxometalate 0.1*M* MOPSO/bis-tris pH 6.5
D4 5%(*w*/*v*) PEG 20K, 25%(*w*/*v*) 1,1,1-tris(hydroxymethyl)propane, 1%(*w*/*v*) NDSB 195 0.5m*M* of each Oxometalate 0.1*M* MOPSO/bis-tris pH 6.5
D5 15%(*w*/*v*) PEG 3K, 20%(*v*/*v*) 1,2,4-butanetriol, 1%(*w*/*v*) NDSB 256 0.5m*M* of each Oxometalate 0.1*M* BES/TEA pH 7.5
D6 12.5%(*w*/*v*) PEG 4K, 20%(*v*/*v*) 1,2,6-hexanetriol 0.5m*M* of each Oxometalate 0.1*M* BES/TEA pH 7.5
D7 10%(*w*/*v*) PEG 8K, 20%(*v*/*v*) 1,5-pentanediol 0.5m*M* of each Oxometalate 0.1*M* BES/TEA pH 7.5
D8 5%(*w*/*v*) PEG 20K, 25%(*w*/*v*) 1,1,1-tris(hydroxymethyl)propane, 1%(*w*/*v*) NDSB 195 0.5m*M* of each Oxometalate 0.1*M* BES/TEA pH 7.5
D9 15%(*w*/*v*) PEG 3K, 20%(*v*/*v*) 1,2,4-butanetriol, 1%(*w*/*v*) NDSB 256 0.5m*M* of each Oxometalate 0.1*M* GlyGly/AMPD pH 8.5
D10 12.5%(*w*/*v*) PEG 4K, 20%(*v*/*v*) 1,2,6-hexanetriol 0.5m*M* of each Oxometalate 0.1*M* GlyGly/AMPD pH 8.5
D11 10%(*w*/*v*) PEG 8K, 20%(*v*/*v*) 1,5-pentanediol 0.5m*M* of each Oxometalate 0.1*M* GlyGly/AMPD pH 8.5
D12 5%(*w*/*v*) PEG 20K, 25%(*w*/*v*) 1,1,1-tris(hydroxymethyl)propane, 1%(*w*/*v*) NDSB 195 0.5m*M* of each Oxometalate 0.1*M* GlyGly/AMPD pH 8.5
E1 15%(*w*/*v*) PEG 3K, 20%(*v*/*v*) 1,2,4-butanetriol, 1%(*w*/*v*) NDSB 256 0.5m*M* of each Lanthanide 0.1*M* MOPSO/bis-tris pH 6.5
E2 12.5%(*w*/*v*) PEG 4K, 20%(*v*/*v*) 1,2,6-hexanetriol 0.5m*M* of each Lanthanide 0.1*M* MOPSO/bis-tris pH 6.5
E3 10%(*w*/*v*) PEG 8K, 20%(*v*/*v*) 1,5-pentanediol 0.5m*M* of each Lanthanide 0.1*M* MOPSO/bis-tris pH 6.5
E4 5%(*w*/*v*) PEG 20K, 25%(*w*/*v*) 1,1,1-tris(hydroxymethyl)propane, 1%(*w*/*v*) NDSB 195 0.5m*M* of each Lanthanide 0.1*M* MOPSO/bis-tris pH 6.5
E5 15%(*w*/*v*) PEG 3K, 20%(*v*/*v*) 1,2,4-butanetriol, 1%(*w*/*v*) NDSB 256 0.5m*M* of each Lanthanide 0.1*M* BES/TEA pH 7.5
E6 12.5%(*w*/*v*) PEG 4K, 20%(*v*/*v*) 1,2,6-hexanetriol 0.5m*M* of each Lanthanide 0.1*M* BES/TEA pH 7.5
E7 10%(*w*/*v*) PEG 8K, 20%(*v*/*v*) 1,5-pentanediol 0.5m*M* of each Lanthanide 0.1*M* BES/TEA pH 7.5
E8 5%(*w*/*v*) PEG 20K, 25%(*w*/*v*) 1,1,1-tris(hydroxymethyl)propane, 1%(*w*/*v*) NDSB 195 0.5m*M* of each Lanthanide 0.1*M* BES/TEA pH 7.5
E9 15%(*w*/*v*) PEG 3K, 20%(*v*/*v*) 1,2,4-butanetriol, 1%(*w*/*v*) NDSB 256 0.5m*M* of each Lanthanide 0.1*M* GlyGly/AMPD pH 8.5
E10 12.5%(*w*/*v*) PEG 4K, 20%(*v*/*v*) 1,2,6-hexanetriol 0.5m*M* of each Lanthanide 0.1*M* GlyGly/AMPD pH 8.5
E11 10%(*w*/*v*) PEG 8K, 20%(*v*/*v*) 1,5-pentanediol 0.5m*M* of each Lanthanide 0.1*M* GlyGly/AMPD pH 8.5
E12 5%(*w*/*v*) PEG 20K, 25%(*w*/*v*) 1,1,1-tris(hydroxymethyl)propane, 1%(*w*/*v*) NDSB 195 0.5m*M* of each Lanthanide 0.1*M* GlyGly/AMPD pH 8.5
F1 15%(*w*/*v*) PEG 3K, 20%(*v*/*v*) 1,2,4-butanetriol, 1%(*w*/*v*) NDSB 256 0.02*M* of each Monosaccharide II 0.1*M* MOPSO/bis-tris pH 6.5
F2 12.5%(*w*/*v*) PEG 4K, 20%(*v*/*v*) 1,2,6-hexanetriol 0.02*M* of each Monosaccharide II 0.1*M* MOPSO/bis-tris pH 6.5
F3 10%(*w*/*v*) PEG 8K, 20%(*v*/*v*) 1,5-pentanediol 0.02*M* of each Monosaccharide II 0.1*M* MOPSO/bis-tris pH 6.5
F4 5%(*w*/*v*) PEG 20K, 25%(*w*/*v*) 1,1,1-tris(hydroxymethyl)propane, 1%(*w*/*v*) NDSB 195 0.02*M* of each Monosaccharide II 0.1*M* MOPSO/bis-tris pH 6.5
F5 15%(*w*/*v*) PEG 3K, 20%(*v*/*v*) 1,2,4-butanetriol, 1%(*w*/*v*) NDSB 256 0.02*M* of each Monosaccharide II 0.1*M* BES/TEA pH 7.5
F6 12.5%(*w*/*v*) PEG 4K, 20%(*v*/*v*) 1,2,6-hexanetriol 0.02*M* of each Monosaccharide II 0.1*M* BES/TEA pH 7.5
F7 10%(*w*/*v*) PEG 8K, 20%(*v*/*v*) 1,5-pentanediol 0.02*M* of each Monosaccharide II 0.1*M* BES/TEA pH 7.5
F8 5%(*w*/*v*) PEG 20K, 25%(*w*/*v*) 1,1,1-tris(hydroxymethyl)propane, 1%(*w*/*v*) NDSB 195 0.02*M* of each Monosaccharide II 0.1*M* BES/TEA pH 7.5
F9 15%(*w*/*v*) PEG 3K, 20%(*v*/*v*) 1,2,4-butanetriol, 1%(*w*/*v*) NDSB 256 0.02*M* of each Monosaccharide II 0.1*M* GlyGly/AMPD pH 8.5
F10 12.5%(*w*/*v*) PEG 4K, 20%(*v*/*v*) 1,2,6-hexanetriol 0.02*M* of each Monosaccharide II 0.1*M* GlyGly/AMPD pH 8.5
F11 10%(*w*/*v*) PEG 8K, 20%(*v*/*v*) 1,5-pentanediol 0.02*M* of each Monosaccharide II 0.1*M* GlyGly/AMPD pH 8.5
F12 5%(*w*/*v*) PEG 20K, 25%(*w*/*v*) 1,1,1-tris(hydroxymethyl)propane, 1%(*w*/*v*) NDSB 195 0.02*M* of each Monosaccharide II 0.1*M* GlyGly/AMPD pH 8.5
G1 15%(*w*/*v*) PEG 3K, 20%(*v*/*v*) 1,2,4-butanetriol, 1%(*w*/*v*) NDSB 256 0.02*M* of each Amino-acid II 0.1*M* MOPSO/bis-tris pH 6.5
G2 12.5%(*w*/*v*) PEG 4K, 20%(*v*/*v*) 1,2,6-hexanetriol 0.02*M* of each Amino-acid II 0.1*M* MOPSO/bis-tris pH 6.5
G3 10%(*w*/*v*) PEG 8K, 20%(*v*/*v*) 1,5-pentanediol 0.02*M* of each Amino-acid II 0.1*M* MOPSO/bis-tris pH 6.5
G4 5%(*w*/*v*) PEG 20K, 25%(*w*/*v*) 1,1,1-tris(hydroxymethyl)propane, 1%(*w*/*v*) NDSB 195 0.02*M* of each Amino-acid II 0.1*M* MOPSO/bis-tris pH 6.5
G5 15%(*w*/*v*) PEG 3K, 20%(*v*/*v*) 1,2,4-butanetriol, 1%(*w*/*v*) NDSB 256 0.02*M* of each Amino-acid II 0.1*M* BES/TEA pH 7.5
G6 12.5%(*w*/*v*) PEG 4K, 20%(*v*/*v*) 1,2,6-hexanetriol 0.02*M* of each Amino-acid II 0.1*M* BES/TEA pH 7.5
G7 10%(*w*/*v*) PEG 8K, 20%(*v*/*v*) 1,5-pentanediol 0.02*M* of each Amino-acid II 0.1*M* BES/TEA pH 7.5
G8 5%(*w*/*v*) PEG 20K, 25%(*w*/*v*) 1,1,1-tris(hydroxymethyl)propane, 1%(*w*/*v*) NDSB 195 0.02*M* of each Amino-acid II 0.1*M* BES/TEA pH 7.5
G9 15%(*w*/*v*) PEG 3K, 20%(*v*/*v*) 1,2,4-butanetriol, 1%(*w*/*v*) NDSB 256 0.02*M* of each Amino-acid II 0.1*M* GlyGly/AMPD pH 8.5
G10 12.5%(*w*/*v*) PEG 4K, 20%(*v*/*v*) 1,2,6-hexanetriol 0.02*M* of each Amino-acid II 0.1*M* GlyGly/AMPD pH 8.5
G11 10%(*w*/*v*) PEG 8K, 20%(*v*/*v*) 1,5-pentanediol 0.02*M* of each Amino-acid II 0.1*M* GlyGly/AMPD pH 8.5
G12 5%(*w*/*v*) PEG 20K, 25%(*w*/*v*) 1,1,1-tris(hydroxymethyl)propane, 1%(*w*/*v*) NDSB 195 0.02*M* of each Amino-acid II 0.1*M* GlyGly/AMPD pH 8.5
H1 15%(*w*/*v*) PEG 3K, 20%(*v*/*v*) 1,2,4-butanetriol, 1%(*w*/*v*) NDSB 256 0.01*M* of each Polyamine 0.1*M* MOPSO/bis-tris pH 6.5
H2 12.5%(*w*/*v*) PEG 4K, 20%(*v*/*v*) 1,2,6-hexanetriol 0.01*M* of each Polyamine 0.1*M* MOPSO/bis-tris pH 6.5
H3 10%(*w*/*v*) PEG 8K, 20%(*v*/*v*) 1,5-pentanediol 0.01*M* of each Polyamine 0.1*M* MOPSO/bis-tris pH 6.5
H4 5%(*w*/*v*) PEG 20K, 25%(*w*/*v*) 1,1,1-tris(hydroxymethyl)propane, 1%(*w*/*v*) NDSB 195 0.01*M* of each Polyamine 0.1*M* MOPSO/bis-tris pH 6.5
H5 15%(*w*/*v*) PEG 3K, 20%(*v*/*v*) 1,2,4-butanetriol, 1%(*w*/*v*) NDSB 256 0.01*M* of each Polyamine 0.1*M* BES/TEA pH 7.5
H6 12.5%(*w*/*v*) PEG 4K, 20%(*v*/*v*) 1,2,6-hexanetriol 0.01*M* of each Polyamine 0.1*M* BES/TEA pH 7.5
H7 10%(*w*/*v*) PEG 8K, 20%(*v*/*v*) 1,5-pentanediol 0.01*M* of each Polyamine 0.1*M* BES/TEA pH 7.5
H8 5%(*w*/*v*) PEG 20K, 25%(*w*/*v*) 1,1,1-tris(hydroxymethyl)propane, 1%(*w*/*v*) NDSB 195 0.01*M* of each Polyamine 0.1*M* BES/TEA pH 7.5
H9 15%(*w*/*v*) PEG 3K, 20%(*v*/*v*) 1,2,4-butanetriol, 1%(*w*/*v*) NDSB 256 0.01*M* of each Polyamine 0.1*M* GlyGly/AMPD pH 8.5
H10 12.5%(*w*/*v*) PEG 4K, 20%(*v*/*v*) 1,2,6-hexanetriol 0.01*M* of each Polyamine 0.1*M* GlyGly/AMPD pH 8.5
H11 10%(*w*/*v*) PEG 8K, 20%(*v*/*v*) 1,5-pentanediol 0.01*M* of each Polyamine 0.1*M* GlyGly/AMPD pH 8.5
H12 5%(*w*/*v*) PEG 20K, 25%(*w*/*v*) 1,1,1-tris(hydroxymethyl)propane, 1%(*w*/*v*) NDSB 195 0.01*M* of each Polyamine 0.1*M* GlyGly/AMPD pH 8.5
###### Selection of the 35 PDB-derived ligands selected to formulate MORPHEUS II
Number of structures (*i.e.* occurrences in the PDB) showing a main chemical as an ordered ligand (as of December 2014). Note that a few reagents that were not found in the PDB were integrated to complete the formulations (the four polyols used as cryoprotectants and five of the six buffers used to formulate the buffer systems).
Chemical name Type PDB ID (main) No. of structures
------------------------------------------- ----------------- --------------- -------------------
Lithium sulfate Common salt LI 51
Sodium chloride Common salt NA 4726
Potassium sulfate Common salt K 1638
Manganese chloride tetrahydrate Divalent cation MN 1938
Cobalt chloride hexahydrate Divalent cation CO 474
Nickel chloride hexahydrate Divalent cation NI 699
Zinc acetate dihydrate Divalent cation ZN 8413
Barium acetate Alkali BA 91
Cesium acetate Alkali CS 75
Rubidium chloride Alkali RB 34
Strontium acetate Alkali SR 101
Sodium chromate tetrahydrate Oxometalate CR 7
Sodium molybdate dihydrate Oxometalate MOO 20
Sodium orthovanadate Oxometalate VO4 73
Sodium tungstate dihydrate Oxometalate WO4 47
Erbium(III) chloride hexahydrate Lanthanide ER3 2
Terbium(III) chloride hexahydrate Lanthanide TB 11
Ytterbium(III) chloride hexahydrate Lanthanide YB 57
Yttrium(III) chloride hexahydrate Lanthanide YT3 33
Xylitol Monosaccharide XYL 25
[D]{.smallcaps}-()-Fructose Monosaccharide FRU; FUD 36; 4
[D]{.smallcaps}-Sorbitol Monosaccharide SOR 12
*myo*-Inositol Monosaccharide INS 16
[L]{.smallcaps}-Rhamnose monohydrate Monosaccharide RAM 43
[DL]{.smallcaps}-Threonine Amino-acid DTH; THR 23; n/a
[DL]{.smallcaps}-HistidineHClH~2~O Amino-acid DHI; HIS 24; n/a
[DL]{.smallcaps}-5-HydroxylysineHCl Amino-acid n/a; LYZ 0; 7
*trans*-4-Hydroxy-[L]{.smallcaps}-proline Amino-acid HYP 149
Spermine4HCl Polyamine SPM 105
Spermidine3HCl Polyamine SPD 33
1,4-Diaminobutane2HCl Polyamine PUT 22
[DL]{.smallcaps}-OrnithineHCl Polyamine ORD; ORN 3; 56
NDSB 256 Surfactant DMX 4
NDSB 195 Surfactant NDS 7
Bis-tris Buffer BTB 114
###### Formulation of the eight additive mixes of MORPHEUS II
Note that the mix called Alkalis includes two of the alkali-earth metals, strontium and barium, and two of the alkali metals, cesium and rubidium. Also, the mix called Lanthanides includes one of the rare-earth elements chemically very similar to the lanthanides (yttrium).
Row Mix name Chemicals
----- --------------------- -----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
A LiNaK 0.3*M* lithium sulfate, 0.3*M* sodium sulfate, 0.3*M* potassium sulfate
B Divalent cations II 5m*M* manganese chloride, 5m*M* cobalt chloride, 5m*M* nickel chloride, 5m*M* zinc chloride
C Alkalis 10m*M* rubidium chloride, 10m*M* strontium acetate, 10m*M* cesium acetate, 10m*M* barium acetate
D Oxometalates 5m*M* sodium chromate, 5m*M* sodium molybdate, 5m*M* sodium tungstate, 5m*M* sodium orthovanadate
E Lanthanides 5m*M* erbium(III) chloride hexahydrate, 5m*M* terbium(III) chloride hexahydrate, 5m*M* ytterbium(III) chloride hexahydrate, 5m*M* yttrium(III) chloride hexahydrate
F Monosaccharides II 0.2*M* xylitol, 0.2*M* [D]{.smallcaps}-()-fructose, 0.2*M* [D]{.smallcaps}-sorbitol, 0.2*M* *myo*-inositol, 0.2*M* [L]{.smallcaps}-rhamnose monohydrate
G Amino-acids II 0.2*M* [DL]{.smallcaps}-arginineHCl, 0.2*M* [DL]{.smallcaps}-threonine, 0.2*M* [DL]{.smallcaps}-histidineHClH~2~O, 0.2*M* [DL]{.smallcaps}-5-hydroxylysineHCl, 0.2*M* *trans*-4-hydroxy-[L]{.smallcaps}-proline
H Polyamines 0.1*M* spermine4HCl, 0.1*M* spermidine3HCl, 0.1*M* 1,4-diaminobutane2HCl, 0.1*M* [DL]{.smallcaps}-ornithineHCl
| {
"pile_set_name": "PubMed Central"
} |
The content published in Cureus is the result of clinical experience and/or research by independent individuals or organizations. Cureus is not responsible for the scientific accuracy or reliability of data or conclusions published herein. All content published within Cureus is intended only for educational, research and reference purposes. Additionally, articles published within Cureus should not be deemed a suitable substitute for the advice of a qualified health care professional. Do not disregard or avoid professional medical advice due to content published within Cureus.
Introduction and background
===========================
Baastrup\'s disease was first described in 1933 as a cause of postural back pain, which was thought to be related to the adjacent \'kissing\' of the spinous processes \[[@REF1]\]. The diagnosis was based on \'symptoms\' of positional back pain with extension, and plain radiographs showing close approximation of the spinous processes (Figure [1](#FIG1){ref-type="fig"}).
{#FIG1}
Gradually throughout the years, individual case reports in patients with clinical symptoms of Baastrup\'s disease were published describing ligamentous stenosis and large dorsal epidural cysts causing neurogenic claudication, as well as interspinous fluid collections believed to be related to the epidural cyst. Review of these individual case reports clearly demonstrates a spectrum of abnormalities involving weakening of the interspinous ligament that can lead to the formation of interspinous fluid clefts, facet and dorsal epidural cysts, intervertebral disc narrowing, and spinal instability with anterolisthesis and ligamentous posterior spinal stenosis \[[@REF2]-[@REF3]\]. By relating the biomechanical role of the interspinous and supraspinous ligaments and the different types of radiologic findings seen in Baastrup\'s disease, interspinous bursitis and segmental instability this review highlights the underlying interrelationship of what is often different stages of the same lumbar segmental degeneration.
Review
======
Biomechanical studies show that the interspinous ligament works in combination with the supraspinous ligament and the ligamentum flavum to provide flexion resistance to the lumbar spine \[[@REF4]\]. The interspinous ligament is one of the mechanisms that maintains sagittal stability of the spinal segment. Inflammatory reaction with fluid and cyst formation within the interspinous ligament can result from chronic repetitive weakening and stretching of the ligament. The ventral area near the posterior epidural space is the weaker section and where cysts develop \[[@REF5]\]. Degeneration and weakness of this ligament leads to spinous process approximation, but more importantly a tendency to develop instability and anterolisthesis, or forward shifting of the superior vertebrae over the inferior vertebrae, at the involved segment \[[@REF6]\]. The different pathologic changes identified on magnetic resonance imaging (MRI) scans with both Baastrup's disease, interstitial bursitis and dorsal epidural cysts are part of the degenerative and biomechanical process that occurs after the interspinous and supraspinous ligaments deteriorate and lose tensile strength \[[@REF7]\]. There are obviously very few pathologic studies of Baastrup\'s disease but bursas are commonly found in the interspinous space. In autopsies, interspinous cysts with inflammatory changes, bone erosion, and bone hypertrophy have been frequently described and are significantly more common with advanced age, being found in up to 40% of specimens \[[@REF8]\].
The actual contact of the spinous processes seen on plain radiographs as the basis for diagnosing Baastrup\'s disease may often be a minor part of all the patient\'s radiologic findings and may be indicative of underlying segmental degeneration. Computerized tomography (CT) scans show bony hypertrophy of the touching spinous processes combined with reactive sclerosis as well as facet joint hypertrophy \[[@REF9]\]. As MRI became the standard imaging technique of the spine, intraspinal cysts and other facet pathology began to be 'associated' with interspinous fluid bursas. There is a wide range of different bursas and cysts seen in these patients commonly identified in the facet joints. MRI scans reveal a spectrum of abnormalities in the interspinous ligament from fluid cystic change, thickening of the adjacent ligamentum flavum, and anterolisthesis of the involved vertebrae. There was often multilevel disease on MRI. Bursas and clefts were most commonly seen at L4-5, but also seen at L3-4 and L5-S1. Connections between these bursas and the epidural space have been found in 50% of the cases \[[@REF9]\].
Incorporating the biomechanics together with the radiologic findings, we hypothesized that repetitive shear due to weakening of the interspinous and supraspinous ligaments leads to the development of interspinal adventitious bursas and cysts, and extension of an inflammatory process within the ligamentum flavum. This progressive degenerative process contributes to the frequently observed soft tissue canal stenosis reported in many cases, although many patients present initially with localized positional lumbar pain and not neurogenic claudication (Figure [2](#FIG2){ref-type="fig"}).
{#FIG2}
Originally, due to Baastrup\'s description relating the problem to the spinous processes, treatment was directed to the interspinous abnormality seen on plain radiographs and originally consisted of surgery for removal of the spinous process with poor and inconsistent results. Beks reviewed 64 patients who had spinous process resection with very poor results \[[@REF9]\]. Only 11/64 patients reported pain relief. Radiographic studies in the form of x-rays (i.e., no CT or MRI) were retrospectively reviewed and showed that all 53 of the failed surgical patients had other spinal pathology including lumbar spondylosis in 55%, disc degeneration in 23%, and spinal stenosis in 13%. He correctly attributed the unrecognized pathology to be the reason resection of the spinous process was unsuccessful. The conclusion was Baastrup\'s disease is not a disease entity by itself; rather it is a result of mechanical changes in the interspinous and supraspinous ligaments, degenerative lumbar discs, and facets with interspinous bursas, osteophyte formation, and spondylosis \[[@REF2]\].
Initially, individual case reports were published of finding epidural cysts and masses causing stenosis associated with Baastrup's disease, but it was later recognized that these two problems were much more commonly associated \[[@REF10]-[@REF11]\]. Chen, et al. reviewed 10 cases of posterior dorsal intraspinal cysts and noted concurrent degenerative disc disease and variable degrees of stenosis, spondylolisthesis at or below the level of the cyst with marked facet degeneration, and three patients had facet joint effusions \[[@REF12]\]. They concluded that Baastrup\'s disease is associated with interspinous fluid and if the fluid bursa is large enough it could extend into the posterior epidural space causing canal stenosis. They also noted that not all cysts communicate with the facet joints or dorsal epidural space. MRI also showed severe stenosis due to a posterior non-cystic fibrous mass often with a linear fluid signal from the interspinous space. In surgery, an interspinous cleft was found and easily probed without excising the interspinous ligament. The histology reports showed the composition to be a collagen matrix mass with peripheral inflammation. They suggested that the Baastrup\'s cyst may change over time becoming more fibrous. Kwong, et al. performed another radiologic retrospective review of 1008 CT scans and found that 41% had evidence of Baastrup\'s disease, most commonly at L4-5, and the frequency of this finding increased with age and also became multilevel \[[@REF13]\]. They concluded Baastrup\'s disease was part of the expected spinal degenerative changes with age and urged caution before diagnosing Baastrup\'s disease as the sole cause of back pain. A study with the use of interspinous distraction and stabilizing devices clearly demonstrated that stabilizing the interspinous space leads to cyst resolution and symptomatic relief, suggesting that cystic degenerative changes are a reaction to the instability resulting from deterioration and laxity of the degenerative interspinous ligament \[[@REF14]-[@REF15]\] (Figure [3](#FIG3){ref-type="fig"}).
{#FIG3}
Patients present with symptoms of either localized back pain worse with extension (i.e., typical Baastrup\'s disease), or more generalized back pain secondary to either facet degeneration or associated with symptoms due to spinal canal compression \[[@REF3], [@REF13], [@REF16]\]. If patients are symptomatic in \'classical\' Baastrup\'s disease they only have positional and very localized midline back pain \[[@REF1]-[@REF3]\]. It is important to determine if patients have other signs and symptoms more typical of facet degeneration, degenerative spondylolisthesis, spinal stenosis, or neurogenic claudication. If there is a failure of significant pain reduction after conservative treatments including physical therapy and anti-inflammatory medication, injections are then usually considered. Typical injections are done via fluoroscopically guided spinal needle placement between the two spinous processes and/or within the area of fluid formation seen on MRI scan with aspiration first performed (if there is cystic fluid within the interspinous space on MRI scan) and followed by a steroid injection \[[@REF17]-[@REF20]\]. If pain recurs, another option is placing a radiofrequency electrode in the same manner with thermal coagulation of multiple areas within the cyst and interspinous ligament. If there are symptoms of neurogenic claudication and MRI reveals a dorsal epidural cyst with or without ligamentous stenosis, then radiofrequency electrodes can also be used to extend the lesion into the posterior epidural space. This has been shown to decrease or ablate the posterior cyst and associated canal stenosis \[[@REF21]\]. If the predominant symptoms are related to neurogenic claudication or radiculopathy, then surgical decompression or stabilization may be necessary \[[@REF16]\]. In the evaluation of clinical symptoms relative to radiologic findings on CT and MRI, it is critical to not over diagnose the underlying degenerative findings \[[@REF15]\]. If the patient has only localized postural back pain with extension without any radicular findings or claudication, then initial use of simple interspinous infiltration or radiofrequency ablation within the interspinous ligament should be the appropriate initial treatment after failing conservative measures.
Conclusions
===========
Degeneration within the interspinous ligament leads to loss of resistance to flexion. Better diagnostic technology demonstrates the wide range of associated cystic changes in the facet joints as well as fibrotic changes in the ligamentum flavum, which can cause symptoms of both back pain and lead to spinal stenosis with neurogenic claudication. If there is only localized back pain, treatment can be directed at the interspinous fluid or cyst with drainage and injection with corticosteroids, which is similar to treatment for lumbar facet cysts. Patients with larger cysts and especially stenosis with signs of mechanical instability may need cyst drainage or stabilization. Most importantly, the treatment ultimately needs to be focused on the underlying associated pathology of stenosis, spinal segmental instability, and/or spondylolisthesis if the patient does not benefit from simple interspinous steroid injection or there is cyst recurrence.
The authors have declared that no competing interests exist.
| {
"pile_set_name": "PubMed Central"
} |
1. Introduction {#sec1-nanomaterials-09-01643}
===============
The wide-ranging use of antibiotics, dyes, and heavy metals and their reckless release in water has drawn intensive attention due to their toxicity and non-biodegradability \[[@B1-nanomaterials-09-01643],[@B2-nanomaterials-09-01643],[@B3-nanomaterials-09-01643]\]. In recent years, many strategies, such as electrochemical oxidation \[[@B4-nanomaterials-09-01643]\], coagulation and flocculation \[[@B5-nanomaterials-09-01643]\], adsorption \[[@B6-nanomaterials-09-01643]\], membrane filtration \[[@B7-nanomaterials-09-01643]\], and advanced oxidation \[[@B8-nanomaterials-09-01643]\], have been applied for water purification. Photocatalysis, a typical advanced oxidation technique, has become the route with the most potential to conquer these intensifying environmental problems via the utilization of solar light \[[@B9-nanomaterials-09-01643],[@B10-nanomaterials-09-01643],[@B11-nanomaterials-09-01643]\]. Hence, various semiconductors, such as g-C~3~N~4~, TiO~2~, ZnS, ZnO, CuS, and MoS~2~, have been confirmed as cost-effective photocatalysts for the heterogeneous photocatalytic purification of polluted water \[[@B12-nanomaterials-09-01643],[@B13-nanomaterials-09-01643],[@B14-nanomaterials-09-01643],[@B15-nanomaterials-09-01643],[@B16-nanomaterials-09-01643],[@B17-nanomaterials-09-01643]\]. Due to their wide band gap, nontoxicity, and high stability, cerium dioxide (CeO~2~) and TiO~2~ have been widely applied in photocatalytic reactions \[[@B18-nanomaterials-09-01643],[@B19-nanomaterials-09-01643],[@B20-nanomaterials-09-01643]\]. In addition, CeO~2~ exhibits strong UV-light sorption capacity and a high resistance to photocorrosion \[[@B21-nanomaterials-09-01643]\].
The adsorption and desorption capacities of oxygen ions are critical for the catalytic process of CeO~2~. The oxygen storage capacity of CeO~2~ is greatly affected by the redox activity of the Ce(III)/Ce(IV) coupling, further depending on the type and content of oxygen vacancies in the lattice structure \[[@B22-nanomaterials-09-01643]\]. The defect structure of Ce^3+^--O~v~--Ce^4+^ (O~v~- oxygen vacancy) and the formation of Ce^3+^ ions induce a red shift in the band gap of CeO~2~ \[[@B23-nanomaterials-09-01643]\]. In fact, the crystal defect of CeO~2~ is greatly related to its crystal structure and morphologies. Various microstructures of CeO~2~ have been fabricated for practical applications via different approaches in previous works, including bowknot-like crystallites, nanocubes, X-architecture, nanopolyhedra, square-like nanoparticles, nanosheets, nano-octahedrons, ribbon-like nanofibers, urchin-like hierarchical structures, flower-like microspheres, and well-aligned nanorod arrays \[[@B24-nanomaterials-09-01643],[@B25-nanomaterials-09-01643],[@B26-nanomaterials-09-01643],[@B27-nanomaterials-09-01643],[@B28-nanomaterials-09-01643],[@B29-nanomaterials-09-01643],[@B30-nanomaterials-09-01643],[@B31-nanomaterials-09-01643],[@B32-nanomaterials-09-01643],[@B33-nanomaterials-09-01643]\]. Unfortunately, CeO~2~ with indirect band gap energy (3.2 eV) is strictly limited in the photocatalytic system of the UV-light region. Hence, nonmetal or metal ions such as S, N, P, Er, Fe, Sm, and Y have been doped in CeO~2~ to extend light harvesting to the visible-light region, leading to enhanced photocatalytic activity \[[@B34-nanomaterials-09-01643],[@B35-nanomaterials-09-01643],[@B36-nanomaterials-09-01643],[@B37-nanomaterials-09-01643],[@B38-nanomaterials-09-01643],[@B39-nanomaterials-09-01643],[@B40-nanomaterials-09-01643]\]. Apart from a doping strategy, noble metals such as Au and Ag loaded on CeO~2~ have been confirmed to present much better photocatalytic activity \[[@B41-nanomaterials-09-01643]\]. In addition, the fabrication of heterojunctions coupled with other semiconductors can simultaneously enhance the separation efficiency of charge carriers and restrain the recombination rate of photoexcited electron--hole pairs through the interface structure of different semiconductors, especially core--shell structures \[[@B42-nanomaterials-09-01643],[@B43-nanomaterials-09-01643],[@B44-nanomaterials-09-01643]\]. To further avoid photocorrosion in solar energy-driven reaction systems, carbon has been employed for the synthesis of heterojunction composites due to its unique physicochemical properties and low cost \[[@B45-nanomaterials-09-01643],[@B46-nanomaterials-09-01643],[@B47-nanomaterials-09-01643],[@B48-nanomaterials-09-01643]\]. Carbon with a porosity structure and a high surface area not only exhibits excellent affinity for pollutant molecules, but also efficiently captures and transfers the photoexcited electron, leading to an enhancement in photocatalytic activity \[[@B49-nanomaterials-09-01643],[@B50-nanomaterials-09-01643]\]. However, the combined effect of Ag doping and carbon coating on the enhanced photocatalytic capacity of CeO~2~ nanosheets has been scarcely reported in previous works.
This work focused on the effect of Ag quantum dots (QDs) on the photocatalytic activity of carbon-coated CeO~2~ (CeO~2~\@C) nanosheets in the visible-light region. In this strategy, Ag QDs were anchored in situ on CeO~2~\@C nanosheets to form Ag/CeO~2~\@C. The photocatalytic capacity of Ag/CeO~2~\@C was affected by the carbon dosage, the Ag-doping content, the Cr(VI) concentration, the pH value, and inorganic ions. The combined effects of the surface plasma resonance (SPR) of Ag QDs, an electron trapper of carbon shells, and the redox activity of the Ce(III)/Ce(IV) coupling were responsible for enhanced visible-light harvesting and efficient charge transfer and separation, leading to excellent photocatalytic activity in the CeO~2~ nanosheets \[[@B39-nanomaterials-09-01643],[@B41-nanomaterials-09-01643],[@B45-nanomaterials-09-01643],[@B51-nanomaterials-09-01643],[@B52-nanomaterials-09-01643]\]. The possible photocatalytic mechanism of Ag/CeO~2~\@C is discussed in detail.
2. Materials and Methods {#sec2-nanomaterials-09-01643}
========================
2.1. Preparation of Catalysts {#sec2dot1-nanomaterials-09-01643}
-----------------------------
CeO~2~ nanosheets were prepared via a hydrothermal route. Briefly, 1.0 mmol of cerium nitrate hexahydrate (Ce(NO~3~)~3~·6H~2~O) and 2 mmol of hexamethylenetetramine (C~6~H~12~N~4~) were dissolved in 70 mL of deionized water through vigorous stirring. Then, 3 mL of acetic acid (CH~3~COOH) was added to the above solution and stirred at room temperature for 2 h. This mixture was transferred to a 100 mL Teflon-lined autoclave and treated at 433 K for 9 h. After being cooled down to room temperature, the suspension was filtered, washed with ethanol and deionized water, dried at 333 K for 6 h, and calcined at 773 K for 3.0 h to obtain CeO~2~ nanosheets.
Carbon-coated CeO~2~ (CeO~2~\@C) nanosheets were also synthesized through a hydrothermal route. Here, 0.1 g obtained CeO~2~ bulks, 0.2 g glucose (C~6~H~12~O~6~), and 1.0 g polyvinyl pyrrolidone (PVP, M = 58,000) were dispersed into 30 mL of deionized water through intensive stirring at room temperature for 2.0 h and then treated at 453 K for 15.0 h in a 50 mL Teflon-lined autoclave. After being cooled down to room temperature, the above suspension was centrifuged, washed, dried at 343 K for 5.0 h, and calcined at 773 K for 2.0 h in an N~2~ flow rate of 40 mL·min^−1^ to obtain CeO~2~\@C nanosheets (called CeO~2~\@C-1). With the above process, CeO~2~\@C composites with varying carbon contents were obtained with different mass ratios of CeO~2~/glucose. CeO~2~\@C-2 and CeO~2~\@C-3 were obtained via the addition of glucose contents of 0.4 g and 0.6 g, respectively.
Ag QDs anchored in CeO~2~\@C (Ag/CeO~2~\@C) nanosheets were reduced in situ with the addition of sodium borohydride (NaBH~4~). In a typical process, 0.1 g CeO~2~\@C bulks and 0.02 g AgNO~3~ were dispersed into 50 mL deionized water and then stirred at room temperature for 2.0 h. NaBH~4~ bulks of 0.05 g were added to the above solution through serious stirring at room temperature for 1.0 h. After the suspension was filtered, washed, and dried at 333 K for 6.0 h, Ag/CeO~2~\@C nanosheets were obtained and called 1-Ag/CeO~2~\@C. X-Ag/CeO~2~\@C (X = 2, 3, and 4) nanosheets were respectively obtained via the addition of 0.03 g, 0.04 g, and 0.05 g AgNO~3~ in accordance with the above process.
2.2. Characterization of Catalysts {#sec2dot2-nanomaterials-09-01643}
----------------------------------
CeO~2~-based nanosheets were evaluated by X-ray diffraction (XRD, Bruker D8, Karlsruhe, Germany), inductively coupled plasma optical emission spectrometry (ICP-OES, Varian 710-ES, PaloAlto, CA, USA), X-ray photoelectron spectroscopy (XPS, Escalab 250, Waltham, MA, USA), physical adsorption (Quantochrome NOVA-2020, Boynton Beach, FL, USA), scanning electron microscopy (SEM, Hitachi S-3400, Tokyo, Japan), transmission electron microscopy (TEM, JEM-2010, Tokyo, Japan), high-resolution transmission electron microscopy (HRTEM, JEM-2100, Tokyo, Japan), Fourier-transform infrared spectra (FT-IR, Bruke Tensor 27, Karlsruhe, Germany), and UV--Vis diffuse reflectance spectra (UV-Vis DRS, Hitachi U-4100, Tokyo, Japan) (with BaSO~4~ as the reflectance standard and an integrated sphere attachment, Photoluminescence (PL, FLSP 920, Edinburgh, UK), with an excitation wavelength of 325 nm at room temperature and electron spin resonance (ESR, JES-FA200, Tokyo, Japan) with 5,5-Dimethyl-1-pyrroline N-oxide (DMPO) as spin trapping agent. In addition, photocurrents, Mott--Schottky curves, and the electrochemical impedance spectroscopy (EIS) of CeO~2~-based composites were evaluated in a three-electrode electrochemical workstation, which consisted of Pt film, KCl-saturated calomel, and FTO conductive glass (coated with 1 cm^2^ of CeO~2~-based composite films) serving as electrodes and 0.2 mol L^−1^ Na~2~SO~4~ serving as an electrolyte. The photoelectrochemical properties of CeO~2~-based composites were analyzed on a photoelectric instrument (CEL-PECX2000, Beijing CEL Tech. Co., Ltd., Beijing, China) equipped with a Vertex. C. EIS electrochemistry workstation (Ivium Technologies B.V., Eindhoven, Netherlands) and a visible-light source (an Xe lamp) at room temperature.
2.3. Photocatalytic Activity {#sec2dot3-nanomaterials-09-01643}
----------------------------
CeO~2~-based composites were applied for the visible-light-driven photodegradation of Cr(VI) ions and tetracycline hydrochloride using an Xe lamp as a light source (300 W). In a typical process, 0.1 g CeO~2~-based bulks was dispersed into 100 mL of a potassium dichromate (K~2~CrO~4~) solution of 20 mg·L^−1^ (or a tetracycline hydrochloride solution of 20 mg·L^−1^) and stirred in a dark room to reach an adsorption--desorption equilibrium. After irradiation at certain time intervals, the concentration of Cr(VI) was analyzed using a UNICO UV-4802 UV-Vis spectrophotometer, and the content of tetracycline hydrochloride was obtained by an Agilent 1100 with a 5-μm, 4.6 × 250 mm Venusil HILIC column and an ultraviolet detector of 356 nm. The intermediate products of Cr(VI) ions and tetracycline hydrochloride were detected by an ICP-OES (Varian 710-ES) and a UPLC-MS system (Waters UPLC Acquity, Quattro Premier XE), respectively. The effects of carbon content, Ag content, solution concentration, pH value, and inorganic ions on the photocatalytic activity of CeO~2~-based composites were investigated using the above process. The photocatalytic durability of the obtained samples was also obtained under the same conditions.
3. Results {#sec3-nanomaterials-09-01643}
==========
The typical peaks of CeO~2~ phases appeared at 28.57°, 33.08°, 47.47°, 56.42°, 59.18°, 69.42°, 76.95°, and 79.15° in XRD patterns of CeO~2~, CeO~2~\@C, and Ag/CeO~2~\@C ([Figure S1](#app1-nanomaterials-09-01643){ref-type="app"} and [Figure 1](#nanomaterials-09-01643-f001){ref-type="fig"}). These peaks were assigned to the (111), (200), (220), (311), (222), (400), (311), and (420) facets of cubic CeO~2~ phases (JCPDS No. 34-0394), respectively \[[@B20-nanomaterials-09-01643],[@B23-nanomaterials-09-01643],[@B25-nanomaterials-09-01643]\]. The peaks of Ag/CeO~2~\@C at 38.10°, 44.32°, and 64.49° were ascribed to the (110), (200), and (220) planes of face-centered cubic Ag phases (JCPDS No. 04-0783) \[[@B22-nanomaterials-09-01643],[@B23-nanomaterials-09-01643],[@B41-nanomaterials-09-01643]\]. The diffraction peaks of carbon phases were not detected in the XRD patterns of CeO~2~\@C and Ag/CeO~2~\@C because they had less carbon content (\<5%) and weak amorphous carbon intensity \[[@B6-nanomaterials-09-01643],[@B53-nanomaterials-09-01643],[@B54-nanomaterials-09-01643]\]. With an increase in Ag content, the diffraction peak intensities of Ag (110) and CeO~2~ (111) ([Figure 1](#nanomaterials-09-01643-f001){ref-type="fig"}) respectively increased at around 38.10° and 28.57°, indicating the high crystallinity of Ag/CeO~2~\@C.
XPS was applied to investigate the surface compositions and chemical states of the obtained CeO~2~ composites. The Ce 3d spectrum ([Figure 2](#nanomaterials-09-01643-f002){ref-type="fig"}A, [Figures S2A and S3A](#app1-nanomaterials-09-01643){ref-type="app"}) was split into eight Gaussian peaks. The peaks at 900.7 eV (U) and 882.5 eV (V) were respectively assigned to Ce 3d~3/2~ and Ce 3d~5/2~, indicating the formation of Ce^3+^ and Ce^4+^ \[[@B20-nanomaterials-09-01643],[@B21-nanomaterials-09-01643]\]. The peaks at 902.9 eV (U′) and 884.9 eV (V′) were ascribed to Ce^3+^ \[[@B23-nanomaterials-09-01643]\]. The peaks at 916.6 eV (U‴), 907.7 eV (U″), 898.2 eV (V‴), and 888.6 eV (V″) were indexed to Ce^4+^ \[[@B20-nanomaterials-09-01643]\]. The relative content of Ce^3+^ could be calculated by the following equation (listed in [Table S1](#app1-nanomaterials-09-01643){ref-type="app"}):
The surface atomic compositions of these samples were nonstoichiometric values compared to the theoretical values due to the presence of a Ce^3+^ state ([Table S1](#app1-nanomaterials-09-01643){ref-type="app"}). The Ce^3+^ concentrations of CeO~2~, CeO~2~\@C-2, and 3-Ag/CeO~2~\@C-2 were, respectively, 12.15%, 14.45%, and 16.54%. The charge compensation may have been responsible for the increase in Ce^3+^ content of CeO~2~\@C-2 and 3-Ag/CeO~2~\@C-2. It was noticed that the Ce^3+^ content (5.81%) of used 3-Ag/CeO~2~\@C-2 after five cycles was lower than that of a fresh sample. This was attributed to the reduced charge compensation under long-term irradiation, leading to inferior photocatalytic stability \[[@B15-nanomaterials-09-01643]\]. The divided Gaussian peaks at 531.2 eV and 529.6 eV ([Figure 2](#nanomaterials-09-01643-f002){ref-type="fig"}B, [Figures S2B and S3B](#app1-nanomaterials-09-01643){ref-type="app"}) were attributed to the defect oxygen (or oxygen vacancy) and lattice oxygen, respectively \[[@B15-nanomaterials-09-01643],[@B20-nanomaterials-09-01643],[@B31-nanomaterials-09-01643]\]. The varied ratios of defect oxygen (or oxygen vacancy) to lattice oxygen were greatly related to the Ce^3+^ content in the as-obtained samples. The concentration of the defect oxygen (or oxygen vacancy) in the O 1s XPS spectrum of the used 3-Ag/CeO~2~\@C-2 ([Figure 2](#nanomaterials-09-01643-f002){ref-type="fig"}B) was higher than that of a fresh sample. This was due to the released and adsorbed oxygen of CeO~2~ and the defect structure of Ce^3+^--O~v~--Ce^4+^ (O~v~- oxygen vacancy) under reduction and oxidation conditions \[[@B38-nanomaterials-09-01643],[@B42-nanomaterials-09-01643]\]. The C1 spectra of CeO~2~\@C-2 ([Figure 3](#nanomaterials-09-01643-f003){ref-type="fig"}C) and fresh and used 3-Ag/CeO~2~\@C-2 ([Figure 2](#nanomaterials-09-01643-f002){ref-type="fig"}C) were divided into three Gaussian peaks at 288.4 eV, 285.3 eV, and 284.8 eV, which respectively belonged to C=O, C--OH, and C--C/C=C bonds. The splitting peaks at 374.2 eV and 368.19 eV in the Ag 3d XPS spectrum of fresh and used 3-Ag/CeO~2~\@C-2 ([Figure 2](#nanomaterials-09-01643-f002){ref-type="fig"}D) were assigned to Ag 3d~3/2~ and Ag 3d~5/2~, respectively. Although there was no change in the microstructure of used 3-Ag/CeO~2~\@C-2 ([Figure S4](#app1-nanomaterials-09-01643){ref-type="app"}), long-term irradiation induced a difference in the C 1s and Ag 3d XPS spectra between the fresh and used samples ([Figure 3](#nanomaterials-09-01643-f003){ref-type="fig"}C,D), leading to inferior charge transfer and separation \[[@B45-nanomaterials-09-01643],[@B47-nanomaterials-09-01643]\].
The detailed microstructure and surface morphology of CeO~2~, CeO~2~\@C, and Ag/CeO~2~\@C were obtained by SEM and TEM. All of these samples were irregular nanosheets with thicknesses ranging from 10 nm to 20 nm ([Figures S4 and S5](#app1-nanomaterials-09-01643){ref-type="app"}, and [Figure 3](#nanomaterials-09-01643-f003){ref-type="fig"}A,B). Compared to the CeO~2~ precursor ([Figure S5A,B](#app1-nanomaterials-09-01643){ref-type="app"}), there were many more fragments detected in the CeO~2~ after it was treated at high temperature ([Figure S5C,D](#app1-nanomaterials-09-01643){ref-type="app"}). With the assistance of carbon coating, fewer fragments and agglomerated nanosheets were detected in CeO~2~\@C ([Figure S5E--G](#app1-nanomaterials-09-01643){ref-type="app"}) and Ag/CeO~2~\@C ([Figure S6](#app1-nanomaterials-09-01643){ref-type="app"}), which was further confirmed by the TEM images ([Figure S7](#app1-nanomaterials-09-01643){ref-type="app"}). With an increase in carbon content, the agglomeration of CeO~2~\@C gradually disappeared and even formed single nanosheets, while the specific surface area of CeO~2~\@C decreased ([Table S2](#app1-nanomaterials-09-01643){ref-type="app"}). Although the nanosheet structure of Ag/CeO~2~\@C was not affected by the Ag doping, the fragment content increased with increasing Ag-doped content due to the damage effect of NaBH~4~ during the in situ reduction process ([Figure S6](#app1-nanomaterials-09-01643){ref-type="app"}). In addition, the specific surface area of Ag-doped CeO2\@C-2 deceased with increasing Ag-doped content ([Table S2](#app1-nanomaterials-09-01643){ref-type="app"}). As is shown in [Figure 3](#nanomaterials-09-01643-f003){ref-type="fig"}, 3-Ag/CeO~2~\@C-2 had irregular and fragmented nanosheets ([Figure 3](#nanomaterials-09-01643-f003){ref-type="fig"}A,B), and ultrafine Ag nanoparticles with a diameter of around 3 nm (dark section in the red circle) were anchored on the CeO~2~\@C-2 ([Figure 3](#nanomaterials-09-01643-f003){ref-type="fig"}C,D and [Figure S7E,F](#app1-nanomaterials-09-01643){ref-type="app"}), which was evidenced by the HRTEM images ([Figure 3](#nanomaterials-09-01643-f003){ref-type="fig"}E,F). The spacing distances between neighboring lattice fringes of the (111) plane of cubic CeO~2~ and the (111) facet of cubic Ag ([Figure 3](#nanomaterials-09-01643-f003){ref-type="fig"}E,F) were, respectively, 0.312 nm and 0.24 nm, which agreed with the XRD pattern of 3-Ag/CeO~2~\@C-2. The elemental distribution of 3-Ag/CeO~2~\@C-2 was further obtained by the elemental mapping images. As is shown in [Figure 3](#nanomaterials-09-01643-f003){ref-type="fig"}G--J, there were Ce, O, C, and Ag elements existing in the obtained sample and discontinuous and monodisperse distributions of Ag elements on the bulk surface, indicating the formation of Ag QDs anchored in CeO~2~\@C nanosheets.
The molecular structures of the obtained CeO~2~, CeO~2~\@C, and Ag/CeO~2~\@C were obtained by FT-IR, as shown in [Figures S8 and S9](#app1-nanomaterials-09-01643){ref-type="app"}. The peaks at 3440 and 1640 cm^−1^ were ascribed to the stretching vibration and bending vibration of the O--H group of absorbed water and surface hydroxyl \[[@B9-nanomaterials-09-01643],[@B11-nanomaterials-09-01643]\]. The peak at 1539 cm^−1^ was attributed to the H--O--H bending vibration of water molecules. The band peaks around 2921 cm^−1^, 2847 cm^−1^, and 1377 cm^−1^ were due to the bending vibration of the C--H group. The peaks at 2362 cm^−1^ and 2340 cm^−1^ could be assigned to the stretching vibrations of C=O groups of adsorbed CO~2~ in the air. The peaks at 676 cm^−1^, 567 cm^−1^, and 475 cm^−1^ were attributed to the vibration of metal oxygen bonds \[[@B41-nanomaterials-09-01643]\]. The optical properties of CeO~2~, CeO~2~\@C-2, and 3-Ag/CeO~2~\@C-2 were obtained by UV-Vis DRS, as shown in [Figure 4](#nanomaterials-09-01643-f004){ref-type="fig"}. Compared to CeO~2~ and CeO~2~\@C-2, 3-Ag/CeO~2~\@C-2 exhibited strong visible-light-harvesting capacity due to the SPR effect of Ag QDs \[[@B18-nanomaterials-09-01643],[@B22-nanomaterials-09-01643],[@B23-nanomaterials-09-01643]\]. According to the plot of (*αhv*)^1/2^ versus (*hv*), the calculated band gap energy of 3-Ag/CeO~2~\@C-2 (2.47 eV) was lower than those of CeO~2~ (2.61 eV) and CeO~2~\@C-2 (2.86 eV). The photoluminescence (PL) spectra ([Figure 5](#nanomaterials-09-01643-f005){ref-type="fig"}) indicated that the PL peak intensity of 3-Ag/CeO~2~\@C-2 was also weaker than those of CeO~2~ and CeO~2~\@C-2 after an excitation at a 300-nm wavelength \[[@B25-nanomaterials-09-01643]\]. On the basis of the standard quantum efficiency of 100% formed from the absorbance at the excitation wavelength and the photoluminescence intensity, the estimated fluorescence efficiency of 3-Ag/CeO~2~\@C-2 (13.21%) was higher than those of CeO~2~ (5.72%) and CeO~2~\@C-2 (8.36%), meaning a lower recombination of charge carriers over 3-Ag/CeO~2~\@C-2. The broad emission band around 350--550 nm was responsible for Ce^3+^ ions and oxide defects in CeO~2~. In addition, the combined effect of carbon coating and Ag QD-doping was helpful for efficient charge transfer and high resistance to the recombination of electron--hole pairs \[[@B41-nanomaterials-09-01643],[@B45-nanomaterials-09-01643]\].
The separation efficiency of photocatalytic electron--hole pairs was evaluated by electrochemical impedance spectroscopy (EIS), in which the arc radius represented the transfer rate of the photocatalyst charge. In contrast to CeO~2~ and CeO~2~\@C-2, 3-Ag/CeO~2~\@C-2 exhibited a higher transient photocurrent under visible-light irradiation ([Figure 6](#nanomaterials-09-01643-f006){ref-type="fig"}A) and a smaller arc radius of electrochemical impedance ([Figure 6](#nanomaterials-09-01643-f006){ref-type="fig"}B). The small radius of 3-Ag/CeO~2~\@C-2 suggested that the low resistance was suitable for the efficient separation of charge carriers in the obtained 3-Ag/CeO~2~\@C-2 photocatalyst. Although a large specific surface area is favorable in reducing the diffusion length of charge carriers and further prompting charge transfer, the SPR effect of Ag QDs and a tight interface between carbon and CeO~2~ nanosheets are crucial for the enhanced separation of charge carriers and the restrained recombination of photoexcited electron--hole pairs \[[@B47-nanomaterials-09-01643],[@B48-nanomaterials-09-01643]\]. Mott--Schottky curves of the obtained CeO~2~, CeO~2~\@C-2, and 3-Ag/CeO~2~\@C-2 were performed for an evaluation of the semiconductor type and flat band potentials (*V~fb~*), where the *V~fb~* values of these samples could be obtained from the *x* intercept by prolonging the linear part of the Mott--Schottky curves on the potential axis ([Figure S10A](#app1-nanomaterials-09-01643){ref-type="app"}). The *V~fb~* values of CeO~2~, CeO~2~\@C-2, and 3-Ag/CeO~2~\@C-2 were −0.83 V, −0.65 V, and −0.58 V versus a KCl-saturated calomel electrode, respectively. Hence, the valence band values of CeO~2~, CeO~2~\@C-2, and 3-Ag/CeO~2~\@C-2 were −0.59 V, −0.41 V, and −0.34 V, respectively. According to the calculated band gap energy, the conduction band values of CeO~2~, CeO~2~\@C-2, and 3-Ag/CeO~2~\@C-2 were 2.27 V, 2.20 V, and 2.13 V, respectively. Compared to 3-Ag/CeO~2~\@C-2 alone with a bias voltage of 1.0 V, the addition of visible-light irradiation with a light-power intensity of 240 mW cm^−2^ (λ \> 420 nm) could enhance the current response ([Figure S10B](#app1-nanomaterials-09-01643){ref-type="app"}), indicating efficient photoinduced charge in the visible light region.
CeO~2~-based composites were performed for the visible-light-driven photocatalytic reduction of Cr(VI) ions and the photodegradation of tetracycline hydrochloride (TCH). Compared to CeO~2~ nanosheets, carbon-coated composites exhibited better adsorption--photocatalytic activity under the same conditions, especially CeO~2~\@C-2 ([Figure S11](#app1-nanomaterials-09-01643){ref-type="app"}). The enhanced adsorption--photocatalysis behaviors of CeO~2~\@C were due to sufficient active sites of carbon shells and the oxygen vacancy of CeO~2~ cores. In addition, this was ascribed to the trap effect of carbon shells for enhanced charge transfer and efficient charge carrier separation \[[@B9-nanomaterials-09-01643],[@B49-nanomaterials-09-01643],[@B50-nanomaterials-09-01643]\]. With the assistance of carbon shells, the excited electrons could efficiently escape from the conduction band (CB) of CeO~2~ to amorphous carbon, achieving the separation of electron--hole pairs and restraining their recombination. The SPR effect of Ag QDs could further strengthen the photocatalytic activity of CeO~2~\@C-2 ([Figure 7](#nanomaterials-09-01643-f007){ref-type="fig"}A). The photocatalytic activity of Ag/CeO~2~\@C-2 for Cr(VI) removal increased and then decreased with an increase in Ag-doping content. In addition, Ag/CeO~2~\@C-2 composites also exhibited excellent photocatalytic activity for the visible-light-driven photodegradation of TCH compared to CeO~2~\@C-2, as shown in [Figure S12](#app1-nanomaterials-09-01643){ref-type="app"}. A similar tendency in the effects of Ag-doping content on photocatalytic activity was obtained for the removal of TCH in the visible-light region. Among these Ag QDs doped composites, CeO~2~\@C-2 and 3-Ag/CeO~2~\@C-2 with an Ag-doping content of 5.41% presented the best photocatalytic activity. Due to the limited active sites, an excess of Cr(VI) ions could not efficiently access the active sites, leading to inferior photocatalytic efficiency. Hence, the removal efficiency of 3-Ag/CeO~2~\@C-2 decreased with increasing concentrations of Cr(VI) ions (ranging from 10 mg L^−1^ to 40 mg·L^−1^ ([Figure 7](#nanomaterials-09-01643-f007){ref-type="fig"}B)). The optimum pH value facilitated the reaction between hydroxyl (OH^−^) (or H^+^) ions and radical species (such as e^−^ and h^+^) to generate ·O~2~^−^ and ·OH radicals \[[@B20-nanomaterials-09-01643],[@B47-nanomaterials-09-01643]\]. In a photocatalytic reaction system, H^+^ ions can react with ·O~2~^−^ to form ·OOH radicals, and ·OOH can react with H^+^ ions to generate H~2~O~2~ \[[@B3-nanomaterials-09-01643]\]. Subsequently, H~2~O~2~ is likely to react with e^−^ to form ·OH and OH^−^, of which OH^−^ ions are scavenged by h^+^ to produce ·OH \[[@B9-nanomaterials-09-01643]\]. In addition, ·OH is also generated from the reaction between H~2~O and h^+^ \[[@B15-nanomaterials-09-01643]\]. An excess of OH^−^ ions can quench the above chain reactions, leading to inferior photocatalytic activity ([Figure 7](#nanomaterials-09-01643-f007){ref-type="fig"}C).
Inorganic ions such as chlorine (Cl^−^), sulfite (SO~3~^2−^), sulfate (SO~4~^2−^), and phosphate (H~2~PO~4~^−^) could affect the photocatalytic activity of 3-Ag/CeO~2~\@C-2 in Cr(VI) removal. As is shown in [Figure 7](#nanomaterials-09-01643-f007){ref-type="fig"}D, inorganic ions could restrain photocatalytic activity compared to 3-Ag/CeO~2~\@C-2 alone in Cr(VI) removal under the same conditions. The removal efficiency of Cr(VI) was remarkably inhibited by H~2~PO~4~^−^ in comparison to Cl^−^, SO~4~^2−^, and SO~3~^2−^. The impeding effect of inorganic ions was due to ·OH scavengers and reduced active sites \[[@B9-nanomaterials-09-01643],[@B10-nanomaterials-09-01643]\]. On the one hand, inorganic ions served as ·OH scavengers and competed with Cr(VI) ions for ·OH radicals. Although the ion radicals were suitable for the oxidizing pollutants, their low oxidoreductive potential induced inferior photocatalytic rates compared to ·OH, especially in acidic conditions \[[@B13-nanomaterials-09-01643]\]. On the other hand, inorganic ions could adsorb on the surface of 3-Ag/CeO~2~\@C-2, leading to decreased surface-active sites for Cr(VI) ions. As is shown in [Figure 8](#nanomaterials-09-01643-f008){ref-type="fig"}, the photocatalytic activity of 3-Ag/CeO~2~\@C-2 remained slightly changed after five cycles. However, the mass loss of 3-Ag/CeO~2~\@C-2 in the sedimentation and transferring processes induced a decrease in photocatalytic activity in each cycle test. SEM images ([Figure S4](#app1-nanomaterials-09-01643){ref-type="app"}) confirmed that the structure of the used 3-Ag/CeO~2~\@C-2 remained unchanged after five cycles of photocatalytic reactions. However, the surface compositions and chemical states of the used 3-Ag/CeO~2~\@C-2 ([Figure 2](#nanomaterials-09-01643-f002){ref-type="fig"}) were different from the fresh composites due to long-term photocorrosion.
As is shown in [Figure 9](#nanomaterials-09-01643-f009){ref-type="fig"}, ESR signals of photoinduced radicals such as ·OH and ·O~2~^−^ were clearly obtained, in which the intensities increased with an increase in the irradiation time of visible light. Quenching testing ([Figure S13](#app1-nanomaterials-09-01643){ref-type="app"}) indicated that the photocatalytic reaction was suppressed by tert-butyl alcohol (*t*-BuOH), benzoquinone (BQ), and ethylenediaminetetraacetic acid disodium salt (EDTA-2Na), especially *p*-BQ and *t*-BuOH. It was concluded that the ·OH and ·O~2~^−^ radicals were vital for the photocatalytic activity of 3-Ag/CeO~2~\@C-2. The enhanced photocatalytic capacity was due to the SPR effect of Ag QDs, an electron trapper of carbon shells, and the redox activity of the Ce(III)/Ce(IV) coupling. In other words, the photocatalytic activity was related to the amount of Ag^+^ serving as an electron acceptor (Ag^2+^ ↔ Ag^0^) and/or a hole donor (Ag^2+^ ↔ Ag ^+^) and the concentration of Ce^3+^ and oxygen vacancies, which could promote the localization of charge carriers and prolong the separation of electron--hole pairs via trapping at energy levels close to the valance band or conduction band \[[@B13-nanomaterials-09-01643]\]. Previous work has suggested that the temperature-programmed reduction (TPR) of peaks belonging to the conversion of Ag^2+^ and Ag^+^ into Ag^0^ is obtained at 405 K and 421 K, respectively \[[@B55-nanomaterials-09-01643]\]. Hence, a reaction between e^−^/h^+^ pairs and Ag^0^/Ag^+^/Ag^2+^ ions is likely to promote the photoreduction of Cr(VI) ions. The possible photocatalytic mechanism of the 3-Ag/CeO~2~\@C-2 heterojunction is proposed in [Figure 10](#nanomaterials-09-01643-f010){ref-type="fig"}. Under visible-light irradiation, the photoexcited electrons from the valance band (O~2p~, VB = −0.56 eV) to the conduction band (Ce~4f~, CB = 2.30 eV) of CeO~2~ were trapped by carbon shells and then transferred to Ag QDs (Equation (1)). The SPR effect of Ag QDs could strengthen the amount of photoexcited electron--hole pairs. The photogenerated electrons were scavenged by O~2~ molecules (E~O2/O~·~2~^−^ = −0.33 eV/NHE, normal hydrogen electrode) in the atmosphere and in the water solution to yield ·O~2~^−^ and •OH radicals (Equations (3)--(5)): meanwhile, h^+^ and ·O~2~^−^ could react with H~2~O molecules to form an ·OH radical (E~H2O,\ OH~^−^~/•OH~ = 1.99 eV/NHE) (Equations (6) and (7)) \[[@B56-nanomaterials-09-01643],[@B57-nanomaterials-09-01643]\]. In addition, OH− ions were adsorbed on the h^+^ of the valance band to generate an ·OH radical. These radicals were responsible for the efficient photocatalytic reduction of Cr(VI) into Cr(III) ions (Equations (8)--(13)), which was confirmed by the Cr 2p XPS spectrum of the used 3-Ag/CeO~2~\@C-2 after five cycles of photocatalytic reaction ([Figure S14](#app1-nanomaterials-09-01643){ref-type="app"}). The splitting peaks at 588.2 eV and 579.2 eV were indexed to Cr(VI), and the peaks at 586.2 eV and 576.4 eV were indexed to Cr(III), indicating a conversion of Cr(VI) ions into Cr(III) ions according to the electron transfer process \[[@B56-nanomaterials-09-01643]\]. Under acidic conditions, Cr(VI) ions could react with electrons and ·OH to generate Cr(III) ions (Equations (8)--(11)). Under alkaline conditions, low H^+^ ions could react with Cr(VI) ions to form Cr^3+^ ions (Equation (12)) and further generate Cr(OH)~3~ loaded on the surface of 3-Ag/CeO~2~\@C-2 bulks (Equation (13)), leading to inferior light adsorption capacity \[[@B58-nanomaterials-09-01643],[@B59-nanomaterials-09-01643],[@B60-nanomaterials-09-01643]\]. In addition, TCH could react with ·OH and h^+^ to form small molecules, which was evaluated by UPLC--MS ([Figure S15](#app1-nanomaterials-09-01643){ref-type="app"}). Under visible-light irradiation, TCH molecules were degraded via N--C bond cleavage and hydroxylation, and then these intermediates could react with the ·OH radical to destroy C2--C3 double bonds and eliminate NH~3~: they subsequently disintegrated into small molecules and even H~2~O and CO~2~ through demethylation, deamination, and ·OH attack \[[@B55-nanomaterials-09-01643],[@B56-nanomaterials-09-01643]\]. Ag QDs doping and carbon shells could serve as an interfacial charge transfer medium and a recombination center, accelerating the photocatalytic reaction:$$\left. {Ag}/{CeO}_{2}@C + {hv}\rightarrow h^{+} + e^{-} \right.$$ $$\left. O_{2} + e^{-}\rightarrow\ \cdot O_{2}^{-} \right.$$ $$\left. O_{2} + e^{-} + H^{+}\rightarrow\ \cdot{OOH} \right.$$ $$\left. \cdot O_{2}^{-}\ + \cdot{OOH} + H^{+}\rightarrow 2\ \cdot{OH} + O_{2} \right.$$ $$\left. H_{2}O + h^{+}\rightarrow\ \cdot{OH} + H^{+} \right.$$ $$\left. \cdot O_{2}^{-}\ + 2H_{2}O + h^{+}\rightarrow 4\cdot{OH} \right.$$ $$\left. 14H^{+} + {Cr}_{2}O_{7}^{2 -} + 6e^{-}\rightarrow 2{Cr}^{3 +} + 7H_{2}O \right.$$ $$\left. 7H^{+} + {HCrO}_{4}^{-} + 3e^{-}\rightarrow{Cr}^{3 +} + 4H_{2}O \right.$$ $$\left. \cdot{OH} + {HCrO}_{4}^{-} + 8H^{+} + 4e^{-}\rightarrow{Cr}^{3 +} + 5H_{2}O \right.$$ $$\left. \cdot{OH} + {Cr}_{2}O_{7}^{2 -} + 15H^{+} + 7e^{-}\rightarrow 2{Cr}^{3 +} + 8H_{2}O \right.$$ $$\left. 8H^{+} + {CrO}_{4}^{2 -} + 3e^{-}\rightarrow{Cr}^{3 +} + 4H_{2}O \right.$$ $$\left. 4H_{2}O + 2{CrO}_{4}^{2 -} + H^{+}\rightarrow 2{Cr}\left( {OH} \right)_{3} + 3{OH}^{-} \right.$$
4. Conclusions {#sec4-nanomaterials-09-01643}
==============
Ag/CeO~2~\@C nanosheets presented better photocatalytic activity than did CeO~2~ and CeO~2~\@C for Cr(VI) removal in the visible-light region. The photocatalytic activity of Ag/CeO~2~\@C increased and then decreased with an increase in carbon content, Ag doping content, and pH value. Inorganic ions and Cr(VI) content had a negative effect on the removal efficiency of Cr(VI) under the same conditions. The best removal efficiency and the most superior photocatalytic stability after five cycles were achieved by 3-Ag/CeO~2~\@C-2 in the visible-light-driven removal of Cr(VI) ions as well as TCH. The SPR effect of Ag QDs, an electron trapper of carbon shells, and the redox activity of the Ce(III)/Ce(IV) coupling played a vital role in the transfer and separation of charge carriers. •O~2~^−^ and •OH radicals were the primary active species of 3-Ag/CeO~2~\@C-2 in the photocatalytic system.
The following are available online at <https://www.mdpi.com/2079-4991/9/11/1643/s1>: Figure S1: XRD patterns of CeO~2~ and CeO~2~\@C; Figure S2: XPS spectra of CeO~2~; Figure S3: XPS spectra of CeO~2~\@C-2; Figure S4: SEM images of used 3-Ag/CeO~2~\@C-2; Figure S5: SEM images of CeO~2~ precursor (A,B), CeO~2~ (C,D), CeO~2~\@C-1 (E,F), CeO~2~\@C-2 (G,H), and CeO~2~\@C-3 (I,J); Figure S6: SEM images of CeO~2~\@C-2 (A,B), 1-Ag/CeO~2~\@C-2 (C,D), 2-Ag/CeO~2~\@C-2 (E,F), 3-Ag/CeO~2~\@C-2 (G,H), and 4-Ag/CeO~2~\@C-2 (I,J); Figure S7: TEM images of CeO~2~ (A,B), CeO~2~\@C-2 (C,D), and 3-Ag/CeO~2~\@C-2 (E,F); Figure S8: FT-IR spectra of CeO~2~ and CeO~2~\@C composites; Figure S9: FT-IR spectra of CeO~2~\@C-2 and Ag/CeO~2~\@C-2 composites; Figure S10: Mott--Schottky curves of CeO~2~, CeO~2~\@C-2, and 3-Ag/CeO~2~-2 (A) and U--I curves of 3-Ag/CeO~2~-2 (B); Figure S11: Effect of carbon content on the photocatalytic activity of CeO~2~\@C composites in the removal of Cr(VI) in the visible-light region, (Cr(VI) content of 20 mg·L^−1^, catalyst dosage of 40 mg, solution volume of 100 mL, and optical power density of 600 mW·cm^−2^); Figure S12: Effect of Ag content on the photocatalytic activity of Ag/CeO~2~\@C-2 composites in the removal of tetracycline hydrochloride in the visible-light region (tetracycline hydrochloride content of 30 mg·L^−1^, catalyst dosage of 40 mg, solution volume of 100 mL, and optical power density of 600 mW cm^−2^); Figure S13: Reactive species trapping experiments over Ag/CeO~2~\@C-2; Figure S14: Cr 2p XPS spectrum of used Ag/CeO~2~\@C-2 after five cycles; Figure S15: HPLC/MS spectrum of TCH over Ag/CeO~2~\@C-2 in visible-light region; Table S1: Atomic ratio and Ce^3+^ ratio of CeO~2~-based samples; Table S2: Texture parameters of CeO~2~-based samples.
######
Click here for additional data file.
conceptualization, writing---original draft preparation, X.Z.; validation and project administration, S.L. and X.L.; methodology, X.F.; investigation and data curation, Q.C.; resources and supervision, J.W.; writing---review and editing, J.W. and S.L.
This research was funded by the Sichuan Science and Technology Program, grant number 2018JY0451; the Education Department of Sichuan Province, grant numbers 17CZ0019 and 17ZA0218; and the Youth Innovation Promotion Association CAS, grant number 2019423.
The authors declare no conflict of interest.
{#nanomaterials-09-01643-f001}
{#nanomaterials-09-01643-f002}
{#nanomaterials-09-01643-f003}
{#nanomaterials-09-01643-f004}
{#nanomaterials-09-01643-f005}
{#nanomaterials-09-01643-f006}
{#nanomaterials-09-01643-f007}
{#nanomaterials-09-01643-f008}
{#nanomaterials-09-01643-f009}
{#nanomaterials-09-01643-f010}
| {
"pile_set_name": "PubMed Central"
} |
BoNTs are a family of bacterial toxins, classified as one of the six most dangerous potential bioterrorism agents (Category A and Tier 1 select agent in the United States)[@b1]. They are also widely used to treat a growing list of medical conditions[@b2][@b3], including muscle spasms, chronic pain, overactive bladders, as well as having cosmetic applications. There are seven well-established serotypes of BoNTs (BoNT/A--G), traditionally defined based on a lack of cross-neutralization by different antisera raised against each toxin type. All BoNTs share the same structure and function[@b4][@b5][@b6]. They are composed of a light chain (LC, ∼50 kDa) and a heavy chain (HC, ∼100 kDa) connected by an inter-chain disulfide bond. The HC contains two sub-domains: the C-terminal H~C~ that mediates binding to receptors, and the N-terminal H~N~ that mediates translocation of the LC across endosomal membranes. The LC acts as a protease in neurons to cleave a set of proteins: BoNT/A, C and E cleave at distinct sites on a peripheral membrane protein known as SNAP-25; BoNT/B, D, F and G cleave at different sites on homologous vesicle proteins VAMP1, 2 and 3 (vesicle-associated membrane proteins); and BoNT/C also cleaves the plasma membrane protein syntaxin 1. These proteins are prototypes of the SNARE (soluble NSF attachment protein receptor) protein family, whose members mediate various membrane fusion events in eukaryotic cells[@b7][@b8]. Cleavage of any one of the three neuronal SNARE proteins blocks fusion of synaptic vesicles to plasma membranes, thus preventing neurotransmitter release from neurons.
Recognizing all distinct serotypes of BoNTs is essential for developing effective countermeasures against this family of toxins. BoNT/A and BoNT/B were first identified in 1919 by Georgina Burke[@b9]. The last of the seven serotypes, BoNT/G, was discovered in 1969 (ref. [@b10]), and no new BoNT serotype has been found for the past four decades. Recent progress in genomic sequencing has revealed multiple subtypes (designated with Arabic numbers, for example, BoNT/A1), which can be recognized by the same antiserum but contain substantial sequence variations[@b11][@b12][@b13]. Furthermore, there are also multiple mosaic toxins. For instance, a 'type H' was reported in 2013 but was later designated as a mosaic toxin, as its LC shares ∼80% identity with the LC of a BoNT/F subtype, BoNT/F5, and its H~C~ shares ∼84% identity with the H~C~ of BoNT/A1 (refs [@b14], [@b15], [@b16], [@b17]). Consistently, this toxin can be neutralized by antibodies against BoNT/A (ref. [@b16]).
The genes encoding BoNTs can be on a plasmid, a phage, or the chromosome, indicating that these genes are mobile and capable of horizontal gene transfer[@b18][@b19][@b20][@b21][@b22][@b23]. Some strains contain two or even three different BoNT genes[@b15][@b24][@b25]. These strains are usually designated with a capital letter for the toxin type that is expressed at higher levels than the other one, followed with a lower case letter for the second toxin type (for example, BoNT/Af strain). In addition, it has also been reported that some BoNT/A strains contain a complete BoNT/B gene, but only BoNT/A is expressed[@b26][@b27][@b28]. Thus, the BoNT/B gene is considered a silent gene and the strains are known as BoNT/A(B) strains. A recent survey of infant botulism cases reported that ∼8% isolates are BoNT/A(B) strains[@b29].
Here we searched published genomic sequences and identified a novel BoNT gene encoded on the chromosome of *Clostridium botulinum* strain 111. This strain was originally identified from an infant botulism patient in Japan in 1996 (ref. [@b30]). The initial characterizations indicated that the toxicity of this strain is due to BoNT/B[@b30]. Later studies confirmed that this strain expresses a subtype of BoNT/B (BoNT/B2) encoded on a plasmid[@b31][@b32]. The sequence of this novel BoNT gene was deposited into the GenBank database in February 2015, as a part of the genomic sequence of *C. botulinum* 111. We characterized the protein encoded by this gene at functional levels and established it as a new BoNT serotype with a unique substrate profile.
Results
=======
Searching genomic databases revealed a novel BoNT gene
------------------------------------------------------
In an attempt to survey the evolutionary landscape of BoNTs, we performed iterative Hidden Markov model searches of the Uniprot sequence database. Our search identified all known BoNT subtypes and mosaic toxins, as well as the related tetanus neurotoxin ([Fig. 1a](#f1){ref-type="fig"}; [Supplementary Fig. 1](#S1){ref-type="supplementary-material"}). To our surprise, the search revealed a potentially new BoNT, tentatively designated BoNT/X ([Fig. 1a](#f1){ref-type="fig"}, GenBank no.: BAQ12790.1), from the recently reported genomic sequence of *C. botulinum* strain 111. BoNT/X showed the least protein sequence identity with the other BoNTs in pairwise comparisons ([Fig. 1b](#f1){ref-type="fig"}). Furthermore, the low sequence similarity is evenly distributed along the entire BoNT/X sequence ([Fig. 1c](#f1){ref-type="fig"}), indicating that it is not a mosaic toxin. Despite this low sequence identity, the overall domain arrangement of BoNTs is conserved in BoNT/X ([Fig. 1c](#f1){ref-type="fig"}), including a zinc-dependent protease motif HEXXH (residues 227--231, HELVH) in the LC (ref. [@b33]), and a SXWY motif in the H~C~ (residues 1,274--1,277, SAWY), which recognizes the lipid receptor gangliosides[@b34].
Similar to the other BoNTs, the BoNT/X gene is located in a gene cluster[@b23]. All seven established BoNTs are co-expressed with another 150 kDa protein known as NTNHA (non-toxic non-hemagglutinin protein), which forms a pH-dependent complex with BoNTs and protects them from proteases in the gastrointestinal tract[@b35]. The BoNT/X gene is also preceded by a potential NTNHA gene ([Fig. 1d](#f1){ref-type="fig"}). Besides BoNT and NTNHA, a typical BoNT gene cluster contains genes encoding one of the two types of accessory proteins: (1) the HA cluster encoding three conserved proteins HA17, HA33 and HA70, which form a complex with BoNT/NTNHA and facilitate absorption of toxins across the intestinal epithelial barrier[@b36][@b37][@b38]; or (2) the OrfX cluster encoding conserved OrfX1, OrfX2, OrfX3 and P47 proteins with unknown function[@b23]. The BoNT/X gene is located in an OrfX gene cluster, as are BoNT/E, F and members of BoNT/A. Interestingly, the BoNT/X cluster has two unique features ([Fig. 1d](#f1){ref-type="fig"}): (1) there is an additional OrfX2 gene that does not exist in any other BoNT clusters (we designated it OrfX2b); (2) the reading frame of OrfX genes is usually opposite to BoNT/NTNHA genes, but it has the same direction as the BoNT/X gene in the BoNT/X cluster ([Fig. 1d](#f1){ref-type="fig"}). These findings suggest that BoNT/X is a unique branch of the BoNT family.
The LC of BoNT/X cleaves VAMP2 at a novel site
----------------------------------------------
To characterize BoNT/X, we first focused on its LC (X-LC, residues 1--439) and produced it as a His6-tagged protein in *Escherichia coli.* LCs of BoNT/A (A-LC) and BoNT/B (B-LC) were produced and assayed in parallel as controls. Incubation of X-LC with rat brain detergent extracts (BDE) did not affect syntaxin 1 or SNAP-25, but abolished VAMP2 immunoblot signals ([Fig. 2a](#f2){ref-type="fig"}). LCs of BoNTs are zinc-dependent proteases[@b33]. As expected, EDTA prevented cleavage of SNARE proteins by X-, A- and B-LCs ([Fig. 2a](#f2){ref-type="fig"}). Furthermore, incubation of X-LC with the purified recombinant cytosolic domain of VAMP2 (residues 1--93) converted VAMP2 into two lower-molecular-weight bands ([Fig. 2b](#f2){ref-type="fig"}), confirming that X-LC cleaves VAMP2.
To identify the cleavage site, we analysed the VAMP2 (1--93) protein, with or without pre-incubation with X-LC, by liquid chromatography--tandem mass spectrometry (LC--MS/MS, [Fig. 2c--e](#f2){ref-type="fig"}). A single dominant peptide peak appeared after incubation with X-LC ([Fig. 2c,e](#f2){ref-type="fig"}; [Supplementary Fig. 2](#S1){ref-type="supplementary-material"}). Its molecular weight is 3,081.7, which fits only the peptide sequence of A67-L93 of VAMP2 ([Fig. 2c,e](#f2){ref-type="fig"}). Consistently, another fragment from the beginning of the His6-tag to residue R66 of VAMP2 was also detected ([Fig. 2d](#f2){ref-type="fig"}). To further confirm this finding, we repeated the assay with a different VAMP2 fragment: glutathione *S*-transferase (GST) tagged VAMP2 (33--86) ([Supplementary Fig. 3](#S1){ref-type="supplementary-material"}). Incubation with X-LC generated a single dominant peptide peak with a molecular weight of 2,063.1, which fits only A67-R86 of VAMP2 ([Supplementary Fig. 3](#S1){ref-type="supplementary-material"}). Together, these results demonstrate that X-LC has a single cleavage site on VAMP2 between R66 and A67.
R66-A67 is a novel cleavage site on VAMP2, distinct from all established target sites of BoNTs ([Fig. 2f](#f2){ref-type="fig"}). It is also the only BoNT cleavage site located within a region previously known as the SNARE motif ([Fig. 2f](#f2){ref-type="fig"}, shaded regions)[@b39]. The VAMP protein family includes VAMP1, 2, 3, 4, 5, 7 and 8, as well as related Sec22b and Ykt6. R66-A67 is conserved in VAMP1 and 3, which are highly homologous to VAMP2. To validate the specificity of X-LC, we expressed HA-tagged VAMP1, 3, 7, 8 and Myc-tagged Sec22b and Ykt6 in HEK293 cells via transient transfection. Cell lysates were incubated with X-LC. Both VAMP1 and 3 were cleaved by X-LC, whereas VAMP7, VAMP8 and Sec22b were resistant to X-LC ([Fig. 2g](#f2){ref-type="fig"}).
BoNT/X cleaves VAMP4, VAMP5 and Ykt6
------------------------------------
Unexpectedly, Ykt6 was also cleaved by X-LC ([Fig. 2g](#f2){ref-type="fig"}). This finding was confirmed using a purified GST-tagged Ykt6 fragment, which shifted to a lower-molecular-weight band after incubation with X-LC ([Fig. 2h](#f2){ref-type="fig"}). The cleavage site was determined to be K173-S174 by mass spectrometry analysis of the intact Ykt6 versus Ykt6 cleaved by X-LC ([Supplementary Fig. 4](#S1){ref-type="supplementary-material"}). This site is homologous to the cleavage site of BoNT/X on VAMP2 ([Fig. 2f](#f2){ref-type="fig"}). Among VAMP family of proteins, VAMP4 contains the same pair of residues (K87-S88) at this site as Ykt6. We found that X-LC cleaved both purified GST-tagged cytoplasmic domain of VAMP4 ([Fig. 2i](#f2){ref-type="fig"}), as well as native VAMP4 in BDE ([Fig. 2j](#f2){ref-type="fig"}). As a control, Sec22b was not cleaved by X-LC in BDE. In addition, the GST-tagged cytoplasmic domain of VAMP5 was also cleaved ([Fig. 2i](#f2){ref-type="fig"}). The cleavage sites were determined by mass spectrometry analysis to be K87-S88 in VAMP4 and R40-S41 in VAMP5 ([Supplementary Fig. 5](#S1){ref-type="supplementary-material"}). Both sites are homologous to the cleavage site of BoNT/X on VAMP2 ([Fig. 2f](#f2){ref-type="fig"}), demonstrating that the location of the cleavage site is conserved across different VAMPs. The ability of X-LC to cleave VAMP4, VAMP5 and Ykt6 is highly unusual, as their sequences are substantially different from VAMP1/2/3. BoNT/X is the first and the only BoNT known that can cleave VAMPs beyond the canonical targets VAMP1, 2 and 3 (ref. [@b40]).
Proteolytic activation of BoNT/X
--------------------------------
We next examined the linker region between the LC and the HC, which must be cleaved by bacterial or host proteases to convert the toxin to an 'active' di-chain form. We produced a recombinant X-LC-H~N~ fragment (residues 1--891) in *E. coli* and subjected it to limited proteolysis by endoproteinase Lys-C. Samples were analysed using Tandem Mass Tag (TMT) labelling and tandem mass spectrometry. TMT labels free N-termini (and lysines). Limited proteolysis by Lys-C produced one new free N terminus mapped to residue N439 in the linker region ([Fig. 3a](#f3){ref-type="fig"}; [Supplementary Data 1](#S1){ref-type="supplementary-material"}), confirming that the linker region is susceptible to proteases.
We then examined whether proteolytic activation enhances the potency of BoNT/X. It has been shown that incubation of high concentrations of LC-H~N~ of BoNTs with cultured neurons results in entry of LC-H~N~, likely through non-specific uptake into neurons[@b41]. Similarly, X-LC-H~N~ entered cultured rat cortical neurons and cleaved VAMP2 in a concentration-dependent manner ([Fig. 3b](#f3){ref-type="fig"}). Activation by Lys-C increased the potency of X-LC-H~N~: 10 nM activated X-LC-H~N~ cleaved similar levels of VAMP2 as 150 nM intact X-LC-H~N~ ([Fig. 3b](#f3){ref-type="fig"}). Activated X-LC-H~N~ appears to be more potent than activated LC-H~N~ of BoNT/A (A-LC-H~N~) and BoNT/B (B-LC-H~N~), which did not show any detectable cleavage of their substrates under the same assay conditions ([Fig. 3b](#f3){ref-type="fig"}).
The inter-chain disulfide bond in BoNT/X
----------------------------------------
Like other BoNTs, the linker region of BoNT/X contains two conserved cysteines, but there is also an additional cysteine (C461) unique to BoNT/X ([Fig. 3a](#f3){ref-type="fig"}). To determine the cysteine residues that form the essential inter-chain disulfide bond, we generated three X-LC-H~N~ mutants, each with one of the three cysteine residues mutated (C423S, C461S and C467S). These mutants, as well as wild-type (WT) X-LC-H~N~, were subjected to limited proteolysis with Lys-C and then analysed via SDS--PAGE and Coomassie Blue staining, with or without the reducing agent dithiothreitol (DTT; [Fig. 3c](#f3){ref-type="fig"}). Mutating the only cysteine on the LC (C423S) is expected to abolish the inter-chain disulfide bond. Consistently, C423S mutant separated into two ∼50 kDa bands without DTT. In contrast, both C461S and C467S mutants showed as a single band at 100 kDa in the absence of DTT and separated into two ∼50 kDa bands in the presence of DTT. These results suggested that C423 on the LC can form the inter-chain disulfide bond with either C461 or C467 on the HC. We also found that Lys-C treatment degraded a significant portion of C423S mutant as compared with C461S or C467S mutants ([Fig. 3c](#f3){ref-type="fig"}, +DTT), suggesting that losing the inter-chain disulfide bond makes the molecule more susceptible to proteases. We noticed that a portion of WT X-LC-H~N~ formed aggregates at the top of the SDS--PAGE gel ([Fig. 3c](#f3){ref-type="fig"}, marked by an asterisk). These aggregates disappeared in the presence of DTT. C423/C461/C467 are the only three cysteines in the X-LC-H~N~; mutating any one of them abolished formation of aggregates ([Fig. 3c](#f3){ref-type="fig"}, −DTT), suggesting that these aggregates are formed by inter-molecular disulfide bonds due to the existence of an extra cysteine in the linker region.
Interestingly, the majority of activated WT X-LC-H~N~ separated into two ∼50 kDa bands without DTT ([Fig. 3c](#f3){ref-type="fig"}), which is similar to C423S mutant. On the other hand, WT X-LC-H~N~ did not show increased degradation by Lys-C compared with C423S mutant ([Fig. 3c](#f3){ref-type="fig"}, +DTT). One possible explanation is that WT X-LC-H~N~ contains an inter-chain disulfide bond under native conditions, but this bond can rearrange to intra-chain C461--C467 pair under denaturing conditions in the SDS buffer. This phenomenon is known as disulfide bond shuffling, which often occurs among adjacent cysteines. To test this hypothesis, we utilized an alkylating reagent, *N-*Ethylmaleimide (NEM), which permanently blocks free cysteines and prevents disulfide bond shuffling. As shown in [Fig. 3d](#f3){ref-type="fig"}, WT X-LC-H~N~ pretreated with NEM showed as a single band at 100 kDa in the absence of DTT, and separated into two ∼50 kDa bands in the presence of DTT. These results confirm that WT X-LC-H~N~ contains mainly an inter-chain disulfide bond, but it is susceptible to disulfide bond shuffling due to an extra cysteine in the linker region ([Fig. 3e](#f3){ref-type="fig"}).
We further examined the activity of the three X-LC-H~N~ cysteine mutants on cultured neurons. As expected, C423S mutant was inactive, whereas C461S and C467S mutants both showed similar levels of activity as WT X-LC-H~N~ ([Fig. 3f](#f3){ref-type="fig"}). These results confirm that the inter-chain disulfide bond is critical for the activity of BoNT/X.
Generating full-length BoNT/X via sortase-mediated ligation
-----------------------------------------------------------
We then sought to determine whether full-length BoNT/X is a functional toxin. As no antisera against BoNT/X are available, we decided to avoid generating the full-length active toxin gene. Instead, we developed an approach to generate a limited amount of full-length BoNTs in test tubes by enzymatic ligation of two non-toxic fragments of BoNTs. This method utilizes a transpeptidase known as sortase[@b42][@b43], which recognizes the peptide motif LPXTG, cleaves between T-G, and concurrently forms a new peptide bond with other proteins/peptides containing N-terminal glycine ([Fig. 4a](#f4){ref-type="fig"}). We produced two non-toxic fragments of BoNT/X: (1) LC-H~N~ with a LPETGG motif and a His6-tag fused to the C terminus; and (2) the H~C~ of BoNT/X (X-H~C~) with a GST tag, thrombin cleavage site, and an additional glycine residue at its N terminus. Cutting by thrombin releases X-H~C~ with a free glycine at its N terminus. Incubation of these two fragments with sortase generated a small amount of ∼150 kD full-length BoNT/X (X-FL, [Fig. 4a,b](#f4){ref-type="fig"}). We note that X-H~C~ showed poor solubility and a strong tendency towards aggregation, which might be the reason for the low ligation efficiency ([Fig. 4b](#f4){ref-type="fig"}). In contrast, ligation of X-LC-H~N~ with the H~C~ of BoNT/A (A-H~C~) achieved a better efficiency, with the majority of X-LC-H~N~ ligated into a XA chimeric toxin ([Supplementary Fig. 6a](#S1){ref-type="supplementary-material"}). To ensure biosafety, the amount of precursor fragments in the reaction is strictly limited to generate the minimum amount of ligated toxin necessary for functional assays.
We first analysed the activity of ligated BoNT/X using cultured rat cortical neurons. Neurons were exposed to the sortase ligation mixture and control mixtures in culture medium. As shown in [Fig. 4c](#f4){ref-type="fig"}, X-LC-H~N~ alone cleaved some VAMP2 due to its high concentration in the reaction mixture. Mixing X-H~C~ with X-LC-H~N~ without sortase slightly enhanced cleavage of VAMP2 compared with X-LC-H~N~ alone, suggesting that X-H~C~ might be associated with X-LC-H~N~ via non-covalent interactions. This interaction appears to be specific, as mixing A-H~C~ with X-LC-H~N~ did not enhance cleavage of VAMP2 in neurons ([Supplementary Fig. 6b](#S1){ref-type="supplementary-material"}). Ligating X-LC-H~N~ with X-H~C~ by sortase clearly enhanced cleavage of VAMP2 compared with the mixture of X-LC-H~N~ and X-H~C~ without sortase ([Fig. 4c](#f4){ref-type="fig"}). These results demonstrated that the X-H~C~ is functional for targeting cells and that ligated full-length BoNT/X entered neurons and cleaved VAMP2. Similarly, ligated XA also entered neurons and cleaved VAMP2 ([Supplementary Fig. 6b](#S1){ref-type="supplementary-material"}).
BoNT/X was not recognized by antisera against known BoNTs
---------------------------------------------------------
We next carried out dot blot assays using antisera raised against known BoNTs, including all seven serotypes as well as one mosaic toxin (BoNT/DC), to confirm that BoNT/X is serologically unique. Four horse antisera were utilized (trivalent anti-BoNT/A, B and E, anti-BoNT/C, anti-BoNT/DC, and anti-BoNT/F) as well as two goat antisera (anti-BoNT/G and anti-BoNT/D). The specificity and potency of these antisera were first validated by analysing their ability to neutralize BoNTs on cultured neurons. As expected, all antisera neutralized their target BoNTs, without affecting the activity of a different serotype ([Supplementary Fig. 7](#S1){ref-type="supplementary-material"}). We found that these antisera recognized their corresponding BoNTs in the dot blot assay, yet none recognized BoNT/X ([Fig. 4d](#f4){ref-type="fig"}).
We further analysed whether the toxicity of BoNT/X on neurons can be neutralized by these antisera. X-FL generated by sortase-mediated ligation was first activated with limited proteolysis using trypsin. We used trypsin to activate X-FL instead of Lys-C for functional assays, as trypsin allows us to stop proteolysis using trypsin inhibitors. Activated X-FL entered cultured rat cortical neurons and cleaved both VAMP2 and VAMP4 in a concentration-dependent manner ([Fig. 4e](#f4){ref-type="fig"}). Combinations of antisera against known BoNTs (Ab1 (horse antisera): trivalent anti-BoNT/A, B and E, anti-BoNT/C, and anti-BoNT/F; Ab2 (goat antisera): anti-BoNT/G and anti-BoNT/D) did not affect the activity of ligated X-FL, as evidenced by similar degrees of VAMP2 and VAMP4 cleavage in the presence of these antisera ([Fig. 4e](#f4){ref-type="fig"}). These results confirmed that BoNT/X is a new BoNT serotype.
BoNT/X induced flaccid paralysis *in vivo* in mice
--------------------------------------------------
We next sought to determine whether BoNT/X is active *in vivo* using a well-established non-lethal assay in mice, known as the Digit Abduction Score (DAS) assay, which measures local muscle paralysis following injection of BoNTs into mouse hind limb muscles[@b44]. BoNTs cause flaccid paralysis of limb muscles, which is manifested as the failure to spread the toes in response to a startle stimulus. We injected ligated X-FL (0.5 μg, activated by trypsin treatment) into the gastrocnemius muscles of the right hind limb in mice, which induced typical flaccid paralysis and the failure of toes to spread ([Fig. 4f](#f4){ref-type="fig"}), indicating that BoNT/X is capable of causing flaccid paralysis *in vivo*. We note that the potency of ligated X-FL appears to be much lower than other BoNTs in this assay. To further confirm the low toxicity of ligated X-FL, we injected mice with 1 μg of ligated X-FL intraperitoneally (*n*=3). No mice showed any systemic effects and all survived at this dose. Thus, ligated X-FL has a rather low toxicity *in vivo* in mice compared with other native BoNTs, which usually have lethal doses at low picogram levels per mouse.
Full-length inactive BoNT/X
---------------------------
Finally, we developed an inactive mutant of BoNT/X as a potential reagent for generating neutralizing antibodies. Mutations at two residues (R362A/Y365F) in BoNT/A inactivate the protease activity of the LC and abolish the toxicity of BoNT/A *in vivo*[@b45][@b46]. These two residues are conserved in BoNT/X. We introduced the corresponding mutations (R360A/Y363F) in BoNT/X and generated a full-length inactive form, designated as BoNT/X~RY~. As shown in [Fig. 4g](#f4){ref-type="fig"}, BoNT/X~RY~ was purified as a His6-tagged protein in *E. coli*, and it had no activity on cultured neurons ([Supplementary Fig. 8a](#S1){ref-type="supplementary-material"}). Furthermore, intraperitoneal injection of mice with 30 μg BoNT/X~RY~ (activated by trypsin treatment) did not cause any adverse effects (*n*=5), demonstrating that it is not toxic *in vivo*. A substantial portion of BoNT/X~RY~ formed aggregates at the top of the SDS--PAGE gel ([Fig. 4g](#f4){ref-type="fig"}). Adding DTT reduced these aggregates to monomeric BoNT/X~RY~ ([Fig. 4g](#f4){ref-type="fig"}). Thus, full-length BoNT/X is susceptible to forming inter-molecular disulfide bonds. Nevertheless, the monomeric form of BoNT/X can be purified and is stable in solution ([Fig. 4g](#f4){ref-type="fig"}). Furthermore, we developed a scale-up purification protocol, which generated BoNT/X~RY~ with a yield of ∼3 mg per liter of culture and ∼90% purity ([Supplementary Fig. 8b](#S1){ref-type="supplementary-material"}). Highly purified BoNT/X~RY~ remained stable in solution up to 10 mg ml^−1^ in the presence of reducing agent. This atoxic BoNT/X~RY~ will be a valuable reagent for generating neutralizing antibodies.
Discussion
==========
BoNT/X is the first serotype of BoNTs identified by genomic sequencing and bioinformatics approaches. It remains unknown whether BoNT/X is ever produced in *C. botulinum* strain 111. BoNT/X could be a silent gene, or it may not be expressed at detectable toxicity levels under culture conditions in the lab. Thus, this toxin was revealed only by sequencing *C. botulinum* 111. This illustrates the importance of genomic sequencing and bioinformatics approaches for understanding microbial virulence factors. Whether BoNT/X could be expressed and exhibit toxicity under certain environmental conditions remains an intriguing question.
A remarkable feature of BoNT/X is its unique ability to cleave VAMP4 and Ykt6. VAMP4 is widely expressed and is known to mediate vesicle fusion between the trans-Golgi network (TGN) and endosomes, as well as homotypic fusion of endosomes[@b47][@b48]. Ykt6 is an atypical SNARE without a transmembrane domain[@b49]. It is anchored to membranes via lipidation, which allows dynamic regulation of its membrane association. Ykt6 is an essential protein in yeast and implicated in multiple membrane fusion events including ER-Golgi, intra-Golgi, endosome-Golgi-vacuolar, and autophagosome formation. Its function in mammalian cells remains to be established. BoNT/X is the first and only BoNT to cleave these SNAREs that mediate various intracellular membrane-trafficking events.
Interestingly, both VAMP4 and Ykt6 are enriched in neurons. Recent studies suggested that VAMP4 contributes to asynchronous synaptic vesicle exocytosis, enlargeosome exocytosis and activity-dependent bulk endocytosis (ADBE) in neurons[@b50][@b51][@b52]. The role of Ykt6 in neurons remains to be established, but it has been shown to suppress the toxicity of α-synuclein in Parkinson's disease models[@b53][@b54]. The other substrate of BoNT/X, VAMP5, is mainly expressed in muscle cells and its function remains to be established[@b55]. BoNT/X will be a useful tool to investigate the function of VAMP4, Ykt6 and VAMP5, as well as related membrane trafficking events. In addition, because VAMP4 has been implicated in granule release in immune cells[@b56], BoNT/X may have the potential to modulate inflammatory secretion in immune cells.
The X-LC-H~N~ fragment showed a higher level of activity in neurons than either A-LC-H~N~ or B-LC-H~N~, suggesting that its membrane translocation and/or protease activity might be more efficient than the corresponding fragments in BoNT/A and BoNT/B. X-H~C~ is functional for targeting cells, as its presence enhanced cleavage of VAMP2 in neurons over LC-H~N~ alone ([Fig. 4c](#f4){ref-type="fig"}). When present without the translocation and LC domains, X-H~C~ is prone to aggregation. This solubility issue is likely due to separation of X-H~C~ from X-LC-H~N~, as full-length BoNT/X~RY~ remains stable at high concentrations. The X-FL generated by sortase-mediated ligation has a rather low toxicity *in vivo* in mice. It remains unknown whether this low *in vivo* toxicity was intrinsic to BoNT/X. It is also possible that the sortase linking method resulted in an attenuated toxin, as the H~C~ and LC-H~N~ folded separately and there is an additional linker between the H~N~ and H~C~ in ligated toxins. Nevertheless, X-FL produced by sortase-mediated ligation is active on neurons and induced typical flaccid paralysis when injected locally in mice, demonstrating that BoNT/X is a functional toxin ([Fig. 4f](#f4){ref-type="fig"}). It will be necessary to produce native BoNT/X to characterize its *in vivo* potency and determine its biosafety risk. It will be important to generate neutralizing antisera against BoNT/X before producing any native toxin.
Introducing a full-length active toxin gene into any expression system/organism is always a significant biosafety concern. Sortase-mediated ligation assembles a small quantity of full-length toxin from two complementary and non-toxic fragments expressed and purified individually. The amount of the precursor fragments in the reaction can be strictly controlled, so the amount of ligated toxin is precisely limited to ensure biosafety. The ligated toxin should possess the same mode of action as native toxins, but may exhibit lower toxicity *in vivo*, possibly due to the addition of the sortase linker and/or reduced compactness of the molecule. Thus, the ligated toxin could be used for functional studies, but may not be suitable for estimating the lethal dose of native toxin. On the other hand, this reduced toxicity from the sortase linking method could be an advantage to create attenuated toxins for research use. It might be also possible to deliberately attenuate the toxicity of ligated toxins *in vivo* by including additional linkers, such as peptide sequences that are sensitive to serum proteases, to further mitigate biosafety concerns.
Methods
=======
Materials
---------
Mouse monoclonal antibodies for syntaxin 1 (HPC-1, dilution 1:5,000), SNAP-25 (C171.2, 1:5,000), VAMP2 (C169.1, 1:2,000), Syp (Cl7.2, 1:2,000) and Syt I (mAB48, 1:1,000) were generously provided by E. Chapman (Madison, WI, USA). Most are available from Synaptic Systems (Goettingen, Germany), with the exception of Syt I (mAB48), which is available from the Developmental Studies Hybridoma Bank. Rabbit polyclonal antibodies against VAMP4 (Cat. no. 136002, 1:1,000) and Sec22b (Cat. no. 186003, 1:1,000) were purchased from Synaptic Systems. The following mouse monoclonal antibodies were purchased from the indicated vendors: actin (Sigma, AC-15, 1:2,000); anti-HA (Covance, 16B12, 1:2,000); anti-Myc (Millipore, 9E10, 1:1,000). Equine polyclonal antisera against BoNT/A/B/E, BoNT/C, BoNT/DC, BoNT/F and goat polyclonal antisera against BoNT/G were generously provided by S. Sharma (FDA). Goat polyclonal antibody against BoNT/D was purchased from Fisher Scientific (NB10062469). BoNTs were purchased from Metabiologics (Madison, WI, USA). Antibody validation is available on the manufacturers' websites. 293T (\#CRL-3216) cells were originally obtained from ATCC, which were negative for mycoplasma contamination but have not been authenticated.
cDNA and constructs
-------------------
The cDNAs encoding X-LC (residues 1--439), X-H~C~ (residues 893--1,306), A-LC-H~N~ (residues 1--874, GenBank no. M30196), and B-LC-H~N~ (residues 1--860, GenBank no. AB232927) were synthesized by GenScript (New Brunswick, NJ, USA). The cDNA encoding X-H~N~ was generated in-house using the Gibson assembly method. X-LC, A-LC (residues 1--425) and B-LC (residues 1--439) were cloned into pET28 vectors with His6-tag on their N-termini. X-H~C~ and A-H~C~ (residues 875--1,297, GenBank No. AF488749) were cloned into pGEX4T to express as GST-tagged proteins. One extra glycine was introduced into the N terminus of X-H~C~ to increase the sortase ligation efficiency. X-LC-H~N~, A-LC-H~N~ and B-LC-H~N~ were cloned into pET22b vector, with the peptide sequence LPETGG fused to their C-termini, followed by a His6-tag, and purified as His6-tagged proteins. The cDNA encoding rat VAMP2 was generously provided by E. Chapman (Madison, WI, USA). VAMP2 (1--93) was cloned into pET28 vector with a His6-tag on the N terminus. VAMP2 (33--86) was cloned into pGEX4T vector and expressed as a GST-tagged protein. The cDNAs encoding mouse VAMP1, VAMP3, VAMP4, human VAMP5, rat VAMP7 and VAMP8 were generously provided by C. Hu (Louisville, KY, USA). Full-length VAMP1, 3, 7 and 8 were cloned into modified pcDNA3.1 vectors, with an HA tag fused to their C termini. Constructs expressing full-length rat Ykt6 and mouse Sec22b, both in pcDNA3.1 vector with a Myc tag fused to the N terminus of the protein, were generously provided by J. Hay (Missoula, MT, USA). The cytoplasmic domains of VAMP4 (1--115) and VAMP5 (1--70) were cloned between BamHI/XhoI sites in pGEX4T and expressed as GST-tagged proteins. We note that there are seven extra residues from the pGEX4T vector fused to the C-termini of VAMP4 and VAMP5 fragments. Ykt6 fragment (residues 1--192) was also cloned into pGEX4T and expressed as GST-tagged proteins. The construct encoding His6-tagged sortase (SrtA\*) was generously provided by B. Pentelute (Boston, MA, USA)[@b43].
Bioinformatics
--------------
The Uniprot database was searched with Jackhmmer on the HMMER web server, using a BoNT/A1 sequence as the seed (Uniprot accession number A5HZZ9) until convergence. Returned sequences were aligned with Clustal Omega and a NeighborNet phylogenetic network estimated with SplitsTree.
Protein purification
--------------------
*E. coli* BL21 (DE3) was utilized for protein expression. In general, induction of expression was carried out with 0.1 mM IPTG at 22 °C overnight. Bacterial pellets were disrupted in lysis buffer (50 mM Tris pH 7.5, 150 mM NaCl) by sonication, and supernatants were collected after centrifugation at 20,000 *g* for 30 min at 4 °C. Protein purification was carried out using AKTA Prime FPLC system (GE), and purified proteins were further desalted with a PD-10 column (GE, 17-0851-01).
Large-scale production and purification of BoNT/X~RY~
-----------------------------------------------------
cDNA encoding BoNT/X~RY~ was assembled in-house from mutated X-LC (R360A/Y363F), X-H~N~ and X-H~C~. It was cloned into a pET22b vector, with the His6-tag on its C terminus. The corresponding plasmid was transformed into *E. coli* BL21 (DE3). Cultures for expression were first grown using a LEX Bioreactor (Epiphyte3, Ontario, Canada) at 37 °C in 1.5 l of medium until OD~600~ reached 0.8. The temperature was then reduced to 18 °C for induction of expression with 1 mM IPTG, and grown for 16--17 h. Bacteria were harvested, re-suspended in HEPES buffer (50 mM HEPES pH 7.2, 500 mM NaCl, 25 mM imidazole, 5% glycerol, 2 mM TCEP), and lysed with an Emulsiflex-C3 (Avestin, Mannheim, Germany) at 20,000 p.s.i. Lysates were ultra-centrifuged at 200,000 *g* for 45 min. Supernatant was loaded onto a 15 ml Ni-NTA agarose column and washed with wash buffer (50 mM HEPES pH7.2, 500 mM NaCl, 100 mM imidazole, 5% glycerol, 1 mM TCEP). Proteins were eluted with elution buffer (50 mM HEPES pH 7.2, 500 mM NaCl, 250 mM imidazole, 5% glycerol, 1 mM TCEP) and then dialyzed overnight in 50 mM HEPES, 500 mM NaCl, 5% glycerol, and 0.5 mM TCEP. Dialysate was concentrated using a Vivaspin concentrator before being loaded on a Superdex200-16/60 column pre-equilibrated in the same buffer used for dialysis. Elution peak corresponding to BoNT/X was collected and concentrated to ∼10 mg ml^−1^. Sample was aliquoted and flash-frozen in liquid nitrogen for storage at −80 °C.
Cleavage of SNARE proteins in rat BDE
-------------------------------------
Rat brain was homogenized in 15 ml 320 mM sucrose buffer, followed by centrifugation at 5,000 r.p.m. for 2 min at 4 °C. Supernatants were collected and centrifuged at 11,000 r.p.m. for 12 min. The pellet was collected and solubilized for 30 min in 15 ml Tris-buffered saline (TBS: 20 mM Tris, 150 mM NaCl) plus 2% of Triton X-100 and a cocktail of protease inhibitors (Roche, CA). Samples were subsequently centrifuged at 17,000 r.p.m. for 20 min to remove insoluble materials. The final BDE concentration was ∼2 mg ml^−1^. BDE (60 μl) were incubated with X-LC (0.5 μM), A-LC (1 μM), or B-LC (1 μM), for 1 h at 37 °C, and then analysed by immunoblot using the enhanced chemiluminescence (ECL) method (Pierce). As controls, LCs were pre-incubated with 20 mM EDTA for 20 min at room temperature before adding to BDE. Full-blot scans are shown in [Supplementary Fig. 9](#S1){ref-type="supplementary-material"}.
Cleavage of recombinant VAMPs by X-LC
-------------------------------------
VAMP2 (1--93) was expressed and purified as a His6-tagged protein. VAMP2 (33--86), VAMP4 (1--115), VAMP5 (1--70) and Ykt6 (1--192) were expressed and purified as GST-tagged proteins. These proteins (0.3 mg ml^−1^) were incubated with 0.1 μM X-LC in TBS buffer at 37 °C. Samples were either analysed by SDS--PAGE gels and Coomassie Blue staining, or subjected to mass spectrometry analysis.
Cleavage of VAMPs in cell lysates: Full-length HA-tagged VAMP1, 3, 7, 8 and Myc-tagged Sec22b, and Ykt6 were transfected into 293T cells using PolyJet transfection reagents (SignaGen, MD) following the manufacturer's instructions. Cell lysates were harvested 48 h later in RIPA buffer (50 mM Tris, 1% NP40, 150 mM NaCl, 0.5% sodium deoxycholate, 0.1% SDS, 400 μl per 10-cm dish) plus a protease inhibitor cocktail (Sigma-Aldrich). Cell lysates (250 μl) were incubated with X-LC (0.5 μM) for 1 h at 37 °C. Samples were then analysed by immunoblot.
Identification of cleavage sites in VAMPs by LC--MS/MS
------------------------------------------------------
Samples were analysed at Taplin Biological Mass Spectrometry Core Facility at Harvard Medical School. For VAMP2, whole-protein samples were loaded onto a 100 μm internal diameter C18 reverse-phase HPLC column packed with 3 cm of beads off-line using a pressure cell. The column was re-attached to an Accela 600 Pump (Thermo Fisher Scientific). A rapid gradient of increasing acetonitrile was used to elute the protein/peptide from the HPLC column. As peptides eluted, they were subjected to electrospray ionization and then placed into an LTQ Orbitrap Velos Pro ion-trap mass spectrometer to acquire a high-resolution FTMS scan at 60,000 resolution, a second scan at low resolution in the ion trap, and a final scan to perform data-dependent MS/MS. The charge state envelopes were de-convoluted manually to obtain mono-isotopic masses when possible or average masses for the proteins. Peptide and protein identity were determined by matching protein databases with the acquired fragmentation pattern using the software program Sequest (Thermo Fisher Scientific). All databases include a reversed version of all the sequences, and the data were filtered to 1--2% peptide false-discovery rate.
For Ykt6, VAMP4 and VAMP5, samples were first separated on SDS--PAGE. Protein bands were excised and cut into ∼1 mm^3^ pieces. Gel pieces were incubated with 50 mM ammonium bicarbonate solution containing 12.5 ng μl^−1^ modified sequencing-grade chymotrypsin (Roche Diagnostics). Samples were digested overnight at room temperature. Peptides were then extracted and separated with reverse-phase HPLC. As peptides were eluted, they were subjected to electrospray ionization and transferred into an LTQ Orbitrap Velos Pro ion-trap mass spectrometer (Thermo Fisher Scientific). Eluted peptides were detected, isolated and fragmented to produce a tandem mass spectrum of specific fragment ions for each peptide.
Identification of the protease cleavage site between LC and H~N~
----------------------------------------------------------------
His6-tagged recombinant X-LC-H~N~ fragment (residues 1--891) was purified in *E. coli* and subjected to limited proteolysis by endoproteinase Lys-C (Sigma P2289, 100:1 (toxin:Lys-C) molar ratio, 25 min at room temperature). The cleavage site was determined by TMT labelling and tandem mass spectrometry. Briefly, intact X-LC-H~N~ samples were labelled with the light TMT, and equal amounts of X-LC-H~N~ samples treated with Lys-C were labelled with the heavy TMT. Both samples were then digested with chymotrypsin, combined and subjected to quantitative mass spectrometry analysis.
Cysteine alkylation by NEM
--------------------------
Lys-C-activated X-LC-H~N~ fragment was diluted into sodium phosphate buffer (10 mM, pH 6.5) at a final concentration of 0.3 mg ml^−1^, with or without NEM at indicated concentrations (20, 10 and 5 mM) and incubated for 10 min at room temperature. NEM was freshly prepared in sodium phosphate buffer. Samples were mixed with 3 × neutral loading dye (200 mM Tris pH 6.8, 30% glycerol, 6% Lithium Dodecyl sulfate, 10 mM NEM, and 0.06% BPB) at room temperature for 10 min, heated for 10 min at 55 °C, and then analysed by SDS--PAGE and Coomassie Blue staining.
Neuron culture and immunoblot analysis
--------------------------------------
Primary rat cortical neurons were prepared from E18-19 embryos using a papain dissociation kit (Worthington Biochemical) following the manufacturer's instruction[@b57]. Neurons were exposed to either BoNT/X fragments or sortase ligation mixtures in culture medium for 12 h. Cells were then lysed with RIPA buffer plus a protease inhibitor cocktail (Sigma-Aldrich). Lysates were centrifuged for 10 min at maximum speed using a microcentrifuge at 4 °C. Supernatants were subjected to SDS--PAGE and immunoblot analysis.
Dot blot
--------
BoNTs (0.2 μg in 1 μl) were spotted onto nitrocellulose membranes and dried (10 min at RT). The membranes were blocked with 5% milk in TBST (TBS plus 0.05% Tween20) for 30 min and then incubated with appropriate antisera (1:500 dilution) for 30 min. The membranes were then washed three times with TBST and incubated with HRP (horseradish peroxidase)-conjugated secondary antibodies for 30 min, washed three more times with TBST, and analysed by the ECL method. The BoNT/X sample was composed of X-LC-H~N~ and GST-X-H~C~ at 1:1 molar ratio.
Sortase-mediated ligation
-------------------------
GST-X-H~C~ or GST-A-H~C~ was cleaved overnight at 4 °C by thrombin before being added into the ligation reaction mixture. Ligation reaction was set up in 50 μl TBS buffer with X-LC-H~N~ (8 μM), X-H~C~ (4 μM) or A-H~C~ (25 μM), Ca^2+^ (10 mM) and sortase (10 μM), for 40 min at room temperature.
DAS assay
---------
All procedures were conducted in accordance with the guidelines approved by the Institute Animal Care and Use Committee (IACUC) at Boston Children's Hospital (\#3030). Briefly, X-FL generated by sortase-mediated ligation was first activated with limited proteolysis using trypsin (60:1 (toxin:trypsin) molar ratio, 30 min at room temperature). We chose trypsin instead of Lys-C here, as it allows us to stop proteolysis by adding trypsin inhibitors (Soybean trypsin inhibitor, 1:10 ratio (trypsin:trypsin inhibitor). Mice (CD-1 strain, male, purchased from Charles River, 5--6 weeks old, 21--25 g, *n*=4) were anesthetized with isoflurane (3--4%) and injected with X-FL (0.5 μg) using a 30-gauge needle attached to a sterile Hamilton syringe, into the gastrocnemius muscles of the right hind limb. Muscle paralysis and the spread of hind paw in response to a startle stimulus were observed 12 h after injection as previously described[@b44].
Biosafety and biosecurity
-------------------------
All procedures were approved by the Institute of Biosafety Committees at Boston Children's Hospital and at Stockholm University. To ensure biosafety and biosecurity, no active full-length toxin gene was produced in any form. The amount of sortase linking reaction is strictly controlled to ensure that only a minimal amount of ligated toxins was produced, which was immediately utilized for functional studies. The loss of toxicity of BoNT/X~RY~ and reduced toxicity of sortase-linked X-FL were confirmed using the mouse lethality assay. Inactive BoNT/X~RY~ was prepared and is available for developing neutralizing antibodies.
Data availability
-----------------
The data and materials that support the findings of this study are available from the corresponding authors upon request.
Additional information
======================
**How to cite this article:** Zhang, S. *et al*. Identification and characterization of a novel botulinum neurotoxin. *Nat. Commun.* **8**, 14130 doi: 10.1038/ncomms14130 (2017).
**Publisher's note:** Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Supplementary Material {#S1}
======================
###### Supplementary Information
Supplementary Figures
###### Supplementary Data 1
Peptide fragments of X-LC-HN under limited proteolysis analyzed by TMT labeling and quantitative mass spectrometry. His6-tagged recombinant X-LC-HN was labeled with the light TMT. Equal amounts of X-LC-HN samples were exposed to Lys-C and then labeled with the heavy TMT. Both samples were then digested with chymotrypsin, combined, and subjected to quantitative mass spectrometry. A list of identified peptides is shown in the associated Excel file. The light TMT: heavy TMT ratios are within 2-fold of each other for all except five peptides (marked in red), all of which start with N439, indicating that N439 is a new N-terminus generated by Lys-C cutting.
We thank members of the Dong and Stenmark labs for discussion, suggestions and comments. We would like to acknowledge Daniel Nilsson for his technical assistance. We thank Edwin R. Chapman (Univ. of Wisconsin---Madison), Shashi Sharma (FDA), Chuan Hu (University of Louisville), Bradley L. Pentelute (MIT) and Jesse C. Hay (University of Montana) for providing antibodies, cDNAs and other reagents. This study was supported in part by NIH 1R01NS080833 (to M.D.), the Swedish Research Council (2014-5667), the Wenner-Gren Foundation, and the Swedish Cancer Society (to P.S.). M.D. also holds the Investigator in the Pathogenesis of Infectious Disease award from the Burroughs Wellcome Fund.
Boston Children's Hospital has filed a provisional patent application (U.S. 62/360,239) for medical use of BoNT/X, with P.S., S.Z. and M.D. as inventors.
**Author contributions** P.S. identified BoNT/X and analysed its gene cluster. P.S. and D.L. carried out the bioinformatics analysis. P.S., G.M., M.M.-C. and L.H. purified the full-length BoNT/X~RY~. P.S., S.Z. and M.D. designed and performed all other experiments and analysed data. J.Z. performed the DAS assay. Y.S. and S.-I.M. synthesized the cDNA for X-LC-H~N~ of BoNT/X and helped with protein purification. M.D. and P.S. wrote the manuscript, with contributions from all other authors.
{ref-type="supplementary-material"}, with the sequence access number for each toxin gene noted. (**b**) A phylogenic tree of the protein sequence alignment for BoNT/A-G, TeNT and BoNT/X, analysed by the ClustalW method. The percentages of sequence identity between each toxin and BoNT/X are noted. (**c**) Upper panel: a schematic drawing of the three domains of BoNT/X, with conserved protease motif in the LC and the ganglioside binding motif in the H~C~ noted. Lower panel: analysis using a sliding sequence comparison window demonstrated that the low similarity between BoNT/X and other BoNTs/TeNT is evenly distributed along the entire BoNT/X sequence. The *X* axis represents the query sequence position at the center of a 100-amino-acid moving sequence-comparison window. The *Y* axis shows the percentage of identity between that sequence window and each of the aligned background sequences. The two bars at the top of the graph illustrate the best matching sequence (lower bar) and whether the best match is significantly separated from the second-best match (upper bar). (**d**) A schematic drawing of the *orf* gene cluster that hosts the BoNT/X gene (upper panel), which has two distinct features compared with other known *orfX* clusters (middle and lower panels): (1) there is an additional *orfX2* protein (designated *orfX2b*) located next to the BoNT/X gene; (2) the reading frame of *orfX* genes has the same direction as the BoNT/X gene.](ncomms14130-f1){#f1}
{#f2}
{ref-type="supplementary-material"}). (**b**) Cultured rat cortical neurons were exposed to X-LC-H~N~ for 12 h. Cell lysates were harvested and immunoblot analysis carried out to examine syntaxin 1, SNAP-25 and VAMP2. Actin is a loading control. Trypsin-activated A-LC-H~N~ and B-LC-H~N~ were analysed in parallel. X-LC-H~N~ entered neurons and cleaved VAMP2. X-LC-H~N~ activated by Lys-C showed a greater potency than non-activated X-LC-H~N~. X-LC-H~N~ was more potent than B-LC-H~N~ and A-LC-H~N~, neither of which cleaved their substrates. (**c**) WT and mutant X-LC-H~N~ were activated by Lys-C and analysed by SDS--PAGE and Coomassie Blue staining, with or without DTT. C461S and C467S mutants showed as a single band at ∼100 kDa without DTT, and separated into two ∼50 kDa bands with DTT. A portion of WT X-LC-H~N~ formed aggregates, marked by an asterisk, which disappeared with DTT. The majority of activated WT X-LC-H~N~ separated into two ∼50 kDa bands without DTT. This is due to disulfide bond shuffling as described in the following panel. (**d**) Lys-C-activated WT X-LC-H~N~ was incubated with NEM to block disulfide bond shuffling. Samples were then analysed by SDS--PAGE and Coomassie Blue staining. A majority of WT X-LC-H~N~ exists as a single band at ∼100 kDa without DTT after NEM treatment, indicating that native WT X-LC-H~N~ contains an inter-chain disulfide bond. (**e**) Schematic drawings of the disulfide bond in WT and three cysteine mutants of BoNT/X. (**f**) Experiments were carried out as described in **b**, except that neurons were exposed to WT or X-LC-H~N~ mutants. C423S mutation abolished the activity of X-LC-H~N~, whereas mutating C461 or C467 did not affect the activity of X-LC-H~N~. These results confirmed that the inter-chain disulfide bond is essential for the activity of X-LC-H~N~, and this inter-chain disulfide bond can be formed via either C423-C461 or C423-C467. One of two (**b**) or three (**b**,**c**,**f**) independent experiments is shown.](ncomms14130-f3){#f3}
{ref-type="supplementary-material"}. (**f**) X-FL linked by sortase reaction (0.5 μg) was injected into the gastrocnemius muscles of the right hind limb of mice (*n*=4). The injected limb developed typical flaccid paralysis, and the toes failed to spread within 12 h. The left limb was not injected with toxins, serving as a control. (**g**) Full-length inactive form of BoNT/X (BoNT/X~RY~) was purified as a His6-tagged recombinant protein in *E. coli*. Further purified BoNT/X~RY~ is shown in [Supplementary Fig. 8b](#S1){ref-type="supplementary-material"}. One of two (**e**) or three (**c**,**d**) independent experiments is shown.](ncomms14130-f4){#f4}
| {
"pile_set_name": "PubMed Central"
} |
1. Introduction {#sec1-life-06-00037}
===============
Halophilic microorganisms thrive in the briny waters of salt lakes and evaporative solar salterns \[[@B1-life-06-00037]\]. Saturated salinity, from 25%--35%, presents osmotic obstacles for life, but we see representatives from all three domains thriving in these conditions \[[@B2-life-06-00037]\]*.* Hypersaline habitats select for archaea at a higher abundance than bacteria and eukarya in the community \[[@B3-life-06-00037]\]. The present study focuses on this particular group, which likely share genetic strategies for life at high salinity \[[@B2-life-06-00037]\]. Microbial diversity studies show a wide array of halophilic archaea present in hypersaline ecosystems \[[@B4-life-06-00037]\] that are adapted to thrive in the saltiest places on Earth.
Beyond salt, high solar radiation is a feature of these environments; thus, their microbial inhabitants must have evolved to overcome the challenge of ultraviolet (UV) exposure. Halophilic archaea are highly resistant to UV light, first noted by Dundas and Larsen \[[@B5-life-06-00037]\]. For example, one *Halorubrum* species was previously found to be nearly ten-fold more UV-resistant than *Escherichia coli* \[[@B6-life-06-00037]\]. This is due in part to their robust UV DNA repair systems, including photoreactivation and nucleotide excision repair \[[@B7-life-06-00037],[@B8-life-06-00037],[@B9-life-06-00037],[@B10-life-06-00037],[@B11-life-06-00037]\], but there are clearly additional processes that afford the high level of resistance to solar radiation exhibited by halophilic archaea.
Several studies have proposed other potential photoprotective strategies; for example, carotenoids in the cell membranes of halophilic archaea may provide UV resistance \[[@B6-life-06-00037],[@B12-life-06-00037],[@B13-life-06-00037]\], although the mechanism of which remains unclear. These organisms have a unique composition of carotenoids that set halophilic archaea apart \[[@B14-life-06-00037],[@B15-life-06-00037],[@B16-life-06-00037],[@B17-life-06-00037]\]. In fact, some halophiles may have an array of genes that allow their complex interplay with light \[[@B18-life-06-00037]\], including carotenoid biosynthetic pathways, and also genes encoding retinal-containing proteins like bacteriorhodopsin, or "purple membrane" protein, which pumps protons out of the cell upon exposure to light \[[@B19-life-06-00037]\] and may aid in the generation of ATP. An environmental genomics approach to saltern communities found the presence of a large variation of rhodopsin-like genes \[[@B20-life-06-00037]\], indicating the significance of photobiology for halophiles.
One photoprotective mechanism suggested in the literature for decades \[[@B22-life-06-00037],[@B23-life-06-00037],[@B24-life-06-00037],[@B25-life-06-00037],[@B26-life-06-00037]\] relates to the genomes of organisms with high Guanine+Cytosine (G + C) content, such as halophilic archaea. Their genomes typically exceed 60% G + C ([Table 1](#life-06-00037-t001){ref-type="table"}), which effectively reduces the number of thymines (Ts) present. Thymine dimers (T\^T), which form from the UV-induced cyclization of two adjacent thymines on the same DNA strand ([Figure 1](#life-06-00037-f001){ref-type="fig"}), were long thought to be the primary DNA damage arising from sunlight \[[@B23-life-06-00037]\]. Logically, the limitation of T residues is expected to reduce the incidence of adjacent Ts on the same DNA strand and thus, the possibility of T\^Ts. Reduction of DNA damage would reduce mutagenesis during replication of unrepaired lesions. Kennedy et al. \[[@B26-life-06-00037]\] suggested that this strategy of thymine limitation could explain the UV resistance of halophilic archaea, but the idea remains untested.
The most important oversight in this premise is that T\^T lesions are but a subclass of cyclobutane pyrimidine dimers (CPDs), which include each possible adjacent bipyrimidine pair: (5' to 3') T\^T, C\^C, C\^T, and T\^C. Haynes \[[@B21-life-06-00037]\] noted that the sensitivity of microorganisms to UV light was not mathematically proportional to the thymine frequency in the genome, which pointed to other lethal lesions. Any set of two adjacent pyrimidines may cyclize upon exposure to a photon of UV light. Though CPDs represent the majority of solar-induced DNA damage events \[[@B28-life-06-00037]\], other potential lesions resulting from the UV-irradiation of bipyrimidine sequences are pyrimidine (6-4) pyrimidone photoproducts (6-4PPs). The proportion of CPDs to 6-4PPs is dependent on wavelength of light \[[@B29-life-06-00037]\] and on flanking sequences \[[@B30-life-06-00037],[@B31-life-06-00037]\].
Though it is logical to assume T\^T lesions may be limited in halophilic archaea, other photolesions may form; therefore, the photoreactivities of each bipyrimidine pair and the relationship between G + C content and their incidences must be considered in order to assess any photoprotective benefit \[[@B32-life-06-00037],[@B33-life-06-00037]\]. Therefore, we sought to address the hypothesis that thymine limitation is photoprotective using a genome photoreactivity score that incorporates all bipyrimidine pairs and their respective photoreactivities, while accounting for all potential types of UV damage (e.g. CPDs and 6-4PPs). We developed an equation to quantify the theoretical photoreactivity of a genome based on the frequencies of each bipyrimidine doublet within it, weighted by their intrinsic photoreactivities (as determined by \[[@B33-life-06-00037]\]) and genome size. Comparison with other groups and a robust statistical analysis sheds light on an interesting story in genome evolution.
2. Materials and Methods {#sec2-life-06-00037}
========================
2.1. Comparing G + C Content of Halophilic Archaea Versus Other Prokaryotes {#sec2dot1-life-06-00037}
---------------------------------------------------------------------------
A list of all prokaryotes with full-genome sequences available (*n* = 5074) and their corresponding G + C contents was downloaded from the NCBI database \[[@B21-life-06-00037]\]. One representative genome for each species was selected at random, yielding sample groups of *n* = 29 halophilic archaea and *n* = 2231 other prokaryotes.
2.2. Genome Sampling {#sec2dot2-life-06-00037}
--------------------
Tabulated lists of all species with full-genome sequences available were downloaded from the NCBI database \[[@B21-life-06-00037]\] for the sample groups bacteria (taxid: 2, *n* = 4829), halophilic archaea (taxid: 183963, *n* = 33), archaea excluding halophilic archaea (taxid: 2157, *n* = 209), cyanobacteria (taxid: 1117, *n* = 90) and enterobacteriaceae (taxid: 91347, *n* = 736).
Steps were taken to minimize sample bias: for the halophilic archaea group, one representative strain from each species was selected at random, yielding a sample group of *n* = 29 halophilic archaea. For the archaea, cyanobacteria, and enterobacteriaceae groups, one representative from each genus was selected at random, yielding sample groups of *n* = 68 (non-halophilic) archaea, *n* = 32 cyanobacteria, and *n* = 42 enterobacteriaceae. For the bacteria, the first 101 strains of a unique genus to be randomly selected constituted the final sample group of *n* = 101 bacteria.
The full-genome sequences corresponding to each sampled strain were downloaded as .fasta files from the NCBI database \[[@B21-life-06-00037]\].
2.3. Determining Bipyrimidine Incidences {#sec2dot3-life-06-00037}
----------------------------------------
To determine the incidences (i.e., relative frequencies) of each bipyrimidine in the sampled genomes, a novel word-counting program "DinucleotideCounts" was written in the scripting language R (script available at: \[[@B34-life-06-00037]\]). This program determines the frequency of each dinucleotide, the frequency of each nucleotide, and the size of any .fasta-formatted DNA sequence.
The bipyrimidine frequencies within sampled genomes were determined via the DinicleotideCounts program. Bipyrimidine incidences (TC~i~, TT~i~, CT~i~, CC~i~) were then computed by dividing each bipyrimidine's frequency by the size of the corresponding genome in bases.
2.4. Determining Theoretical Genomic Photoreactivity (P~g~) {#sec2dot4-life-06-00037}
-----------------------------------------------------------
We devised the metric *P~g~* to quantify the theoretical photoreactivity of a genome based on its bipyrimidine signature: *P~g~* corresponds to the weighted sum of a genome's bipyrimidine incidences (TC~i~, TT~i~, CT~i~, CC~i~):
The coefficients represent the intrinsic photoreactivity of each bipyrimidine, as experimentally determined by \[[@B33-life-06-00037]\] via establishing the ratio between the frequency of the photoproducts (6-4PPs and CPDs) and the bipyrimidine incidences in DNA with varying G + C content.
2.5. Statistical Methods {#sec2dot5-life-06-00037}
------------------------
G + C content averages were compared ([Figure 2](#life-06-00037-f002){ref-type="fig"}) via a Welch Two Sample t-test using the "stats" package in R \[[@B35-life-06-00037]\]. Intergroup differences in bipyrimidine incidences ([Figure 3](#life-06-00037-f003){ref-type="fig"}) and *P~g~* ([Figure 4](#life-06-00037-f004){ref-type="fig"}) were assessed via one-way analysis of variance (ANOVA) and post-hoc Tukey contrasts using the "multcomp" package \[[@B36-life-06-00037]\] in R \[[@B35-life-06-00037]\]. The regression analysis of G + C content vs. *P~g~* ([Figure 5](#life-06-00037-f005){ref-type="fig"}) and corresponding Pearson's product-moment correlation test were carried out using the "stats" package in R. All randomization was facilitated by the "RAND" function in Microsoft Excel.
3. Results {#sec3-life-06-00037}
==========
3.1. G + C Content of Halophilic Archaea {#sec3dot1-life-06-00037}
----------------------------------------
Halophilic archaea are known to have high G + C content with one notable exception, *Haloquadratum walsbyi* \[[@B37-life-06-00037],[@B38-life-06-00037],[@B39-life-06-00037]\]. To understand the unique genomic features of these halophiles, we compare them to other prokaryotes. The NCBI Genbank database contains the full genome sequences for 29 species of halophilic archaea ([Table 1](#life-06-00037-t001){ref-type="table"}), which were utilized for this study. 2231 other prokaryotic species (bacteria and non-halophilic archaea) were selected as described in Materials and Methods. This analysis ([Figure 2](#life-06-00037-f002){ref-type="fig"}) showed that halophilic archaea (excluding *H. walsbyi*) have a clustered distribution with a remarkably high G + C content (63.1% ± 1.3%) relative to other microbial life (49.7% ± 0.55%), thereby demonstrating the uniqueness of this group and pointing to the relatedness of halophilic archaea.
3.2. Bipyrimidine Signature of Halophilic Archaea {#sec3dot2-life-06-00037}
-------------------------------------------------
Due to their high G + C content ([Figure 2](#life-06-00037-f002){ref-type="fig"}), we predicted halophilic archaea would have low and high incidences of TT and CC dinucleotides, respectively. However, considering the incidences of all bipyrimidine pairs (CC~i~, CT~i~, TC~i~, and TT~i~) is necessary for understanding the overall photoreactivity of their genomes. Hence, bipyrimidine incidences were determined for and subsequently compared between halophilic archaea and the other taxonomic groups bacteria, non-halophilic archaea, cyanobacteria, and enterobacteriaceae (*as described in Materials and Method*s) for insight into unique genomic signatures. If an evolutionary genome strategy were employed to protect organisms from sunlight, one would expect that a photosynthetic group like cyanobacteria might share such a strategy, and that enterobacteriaceae bacteria, which dwell inside higher eukaryotes, would not. Random samplings of species in these groups are thus included as controls, in addition to the bacteria and non-halophilic archaea ([Figure 3](#life-06-00037-f003){ref-type="fig"}).
As predicted, halophilic archaea are distinctive among these control groups in their low TT~i~ and high CC~i~ ([Figure 3](#life-06-00037-f003){ref-type="fig"}). Further, it was found that halophilic archaeal genomes are also characterized by high TC~i~ relative to the comparative groups. Multiple comparisons of means with respect to each bipyrimidine pair (*as described in Materials and Methods*) point to some interesting observations, indicating that on average:Halophilic archaea have larger CCi than any other group (*p* \< 10^−4^ each). No other significant intergroup differences in CCi were detected.Halophilic archaea have smaller CTi than archaea and cyanobacteria (*p* \< 10^−4^ each). Other significant differences in CTi were found between enterobacteriaceae vs. archaea (*p* \< 10^−4^), cyanobacteria vs. bacteria (*p* = 2.35 × 10^−4^), and enterobacteriaceae vs. cyanobacteria (*p* \< 10^−4^).Halophilic archaea have larger TCi than any other group (*p* \< 10^−4^ each). Archaea have the next highest level of TCi, being larger than bacteria (*p* \< 10^−4^), cyanobacteria (*p* = 3.27 × 10^−3^), and enterobacteriaceae (*p* \< 10^−4^).Halophilic archaea have smaller TTi than any other group (*p* \< 10^−4^ each). No other significant intergroup differences in TTi were detected.
3.3. Intergroup Differences in Theoretical Genomic Photoreactivity (P~g~) {#sec3dot3-life-06-00037}
-------------------------------------------------------------------------
Until recently, TTs were thought to be the most photoreactive sequences \[[@B24-life-06-00037]\]. This idea has been challenged by current data on the photoreactivity of each bipyrimidine pair in both naked and intracellular DNA with a variety of genome G + C contents \[[@B31-life-06-00037],[@B32-life-06-00037]\]. Matallana-Surget et al. \[[@B33-life-06-00037]\] experimentally found the relative photoreactivities of the bipyrimidine sequences to be in the decreasing order of TC \> TT \> CT \> CC. Furthermore, these authors quantified the intrinsic photoreactivity of each bipyrimidine via determining ratio between frequency of photoproducts (CPDs and 6-4PPs) and bipyrimidine incidences in DNA with varying G + C content. These ratios were employed as coefficients in the equation for our theoretical quantification of genomic photoreactivity in terms of bipyrimidine signature, *P~g~*, (Equation (1)). *P~g~* was calculated for each sampled genome and then compared across all tested taxonomic groups ([Figure 4](#life-06-00037-f004){ref-type="fig"}).
Intergroup differences in *P~g~* were examined via a multiple comparison of means (*as described in Materials and Methods*), which indicated that on average:
- Halophilic archaeal genomes have a smaller *P~g~* than do genomes from (non-halophilic) archaea (*p* \< 10^−3^) and cyanobacteria (*p* = 1.01 × 10^−3^).
- Similarly, enterobacteriaceae genomes are less photoreactive than (non-halophilic) archaea (*p* \< 10^−3^) and cyanobacteria (*p* = 1.02 × 10^−3^).
- (Non-halophilic) archaeal genomes have a larger *P~g~* than do genomes from the other sample groups, with the exception of cyanobacteria (*p* \< 10^−3^).
3.4. Genomic Strategy of Photoprotection {#sec3dot4-life-06-00037}
----------------------------------------
In analyzing photoreactivity in the studied genomes, we plotted each computed *P~g~* value against its corresponding genome's G + C content ([Figure 5](#life-06-00037-f005){ref-type="fig"}). These data clearly show a strong, negative correlation between *P~g~* and G + C content (R^2^ = 0.7139, *p* \< 2.2 × 10^−16^). [Figure 5](#life-06-00037-f005){ref-type="fig"} further demonstrates that this relationship holds true not only for the halophilic archaea tested, but also for all other genomes with this bias toward G + C, altogether giving evidence that an organism's genomic G + C content contributes to their relative sensitivity to UV radiation.
4. Discussion {#sec4-life-06-00037}
=============
Solar exposure for all prokaryotic life on Earth results in UV-induced DNA damage, the majority of which is in the form of CPDs, a cycloaddition of two adjacent pyrimidine bases ([Figure 1](#life-06-00037-f001){ref-type="fig"}) \[[@B28-life-06-00037],[@B31-life-06-00037]\]. CPDs are mitigated by two universal DNA repair systems: photolyase \[[@B40-life-06-00037],[@B41-life-06-00037]\] and nucleotide excision repair \[[@B42-life-06-00037]\]. Halophilic archaea are no exception as they have a similar ratio of CPD to 6, 4 photoproducts \[[@B8-life-06-00037]\] and efficient repair \[[@B7-life-06-00037],[@B8-life-06-00037],[@B9-life-06-00037],[@B10-life-06-00037],[@B11-life-06-00037]\].
The UV spectrum is broken into three subdivisions: UV-A (315--400 nm), UV-B (280--315 nm), and UV-C (\<280 nm), with UV-C being the most damaging for DNA \[[@B43-life-06-00037]\] due to the shorter wavelengths. The solar spectrum (UV A/B) peaks at 300 nm and results in CPDs as the principal type of UV-lesions, placing less emphasis on 6-4PPs \[[@B27-life-06-00037]\]. UV-C light is sometimes used in laboratory experiments to amplify DNA damage detection \[[@B6-life-06-00037],[@B24-life-06-00037]\], but it does not have real world consequence for biological systems since it is absorbed by the oxygen and ozone in the Earth's atmosphere \[[@B44-life-06-00037]\].
The photoreactivity of a specific bipyrimidine pair may vary depending on the wavelength of UV light utilized, the flanking sequence conditions, and the cellular environment. For example, in one study the ratio of T\^T to other CPDs was greater when UV-C light was used in experiments than when UV-B was utilized \[[@B30-life-06-00037]\]. This may account for the significance placed on the T\^T lesions over other photoproducts in decades of laboratory experimentation. The bipyrimidine photoreactivity coefficients utilized in our *P~g~* calculations were adapted from a study that measured UV-B induced CPDs and 6-4PPs \[[@B33-life-06-00037]\].
Our data clearly show that T limitation reduces T\^T formation ([Figure 3](#life-06-00037-f003){ref-type="fig"}). However, this is an incomplete picture. We developed the *P~g~* formula (Equation 1) to further assess other bipyrimidine impacts on photoreactivity. For example, the corresponding enrichment in C nucleotides and the higher photoreactivity of the TC bipyrimidine \[[@B33-life-06-00037]\] also impact the *P~g~* score for a halophilic archaeal genome. Nevertheless, the strong, negative correlation observed between *P~g~* and G + C content ([Figure 5](#life-06-00037-f005){ref-type="fig"}) gives evidence that T limitation facilitates a net increase in photoprotection. Note that the three most photoreactive sequences, TC, TT, and TC \[[@B32-life-06-00037],[@B33-life-06-00037]\], are T-containing: this could explain the observed relationship between genomic photoreactivity and G + C content. It should also be noted that C\^C photolesions are the most mutagenic \[[@B32-life-06-00037]\], which suggests that there is more work to be done beyond photoreactivity to explore mutagenesis from solar irradiation. Finally, determination of the role that flanking sequences \[[@B30-life-06-00037]\] play in the formation of photolesions could add another dimension to this work.
From a genome evolutionary perspective, if exposure to UV light was driving an organism's genomic bipyrimidine signature, we would expect to see this adaptation in cyanobacteria, a photosynthetic group of microorganisms, and we do not ([Figure 4](#life-06-00037-f004){ref-type="fig"}). Conversely, we would expect members of the enterobacteriaceae group, which dwell inside higher eukaryotes protected from the sun, to be void of UV adaptations, but in fact, their photoreactivity scores are more similar to halophilic archaea than any other sample group ([Figure 4](#life-06-00037-f004){ref-type="fig"}). The analysis of our controls gives evidence that there is not a clear/predictive relationship between *P*~g~ and UV-resistance alone. The similarity between the enterobacteriaceae and halophilic archaea stems not from lifestyle, but instead from their relativity high G + C contents ([Table 2](#life-06-00037-t002){ref-type="table"}), which have been shown to result in low *P~g~* values ([Figure 5](#life-06-00037-f005){ref-type="fig"}). Hence, we find no evidence in our analysis of genomic photoreactivity that UV exposure is a selective pressure for a photoprotective bipyrimidine signature.
One important outlier is *Haloquadratum walsbyi*, a dominant halophilic archaeon in salt lakes and salterns. This microorganism is an exception to the G + C-rich genomes of the other members of this group, having only 48% G + C content ([Table 1](#life-06-00037-t001){ref-type="table"}) \[[@B37-life-06-00037],[@B38-life-06-00037],[@B39-life-06-00037]\]. This square-shaped organism otherwise shares the same ecosystem niche as the other extreme halophiles, thriving in high UV exposure; however, it lacks the photoprotective genome signature seen in other halophilic archaea, having a *P~g~* value of 0.260, which is over 8 standard errors smaller than the sample group's mean of 0.245 ([Figure 4](#life-06-00037-f004){ref-type="fig"}). Bolhuis et al. \[[@B37-life-06-00037]\] noted that *H. walsbyi* has a relatively higher number (four) photolyase genes, and this could impact its ability to counteract solar DNA damage.
If the overwhelming majority of halophilic archaea have genomes with lower photoreactivity because of their high G + C contents ([Table 2](#life-06-00037-t002){ref-type="table"}), then we must ask how the G + C richness evolved? Litchfield \[[@B45-life-06-00037]\] pointed to the stability of genomes rich in base pairs that have more hydrogen bonds; GC pairs have three and AT pairs have two. Halophilic archaea living in high salinity (and having high intracellular cation concentrations) would benefit from DNA helices that are more tightly paired, adding stability to their molecular structure in a destabilizing environment. High C + C content has also been discussed as a hypersaline adaptation that impacts the proteome composition of these microorganisms \[[@B46-life-06-00037]\]. A + T limitations result in preferences for particular (G + C-rich) amino acids over others. For example, acidic residues such as aspartic acid would be over-represented and cysteine would be under-represented. Paul et al*.* demonstrate that this signature would reduce the likelihood of helices forming in protein tertiary structure but would positively impact coil structures. Both stronger DNA hydrogen bonding and preferential codon usage (and resulting protein structure) could impact the ability of halophilic archaea to manage life in hypersaline waters.
Genomic signatures, such as bipyrimidine limitations, should be examined when painting a picture of genome evolution for any microbial community. Each related group of species in their environment over time results in a unique signature of nucleic acid and protein composition \[[@B47-life-06-00037],[@B48-life-06-00037],[@B49-life-06-00037]\], and G + C content is a marker for this signature. The UV-resistance observed in halophilic archaea can certainly be attributed to a genomic strategy. However, the driver for this is not simply UV light as other evolutionary pressures, including hypersalinity, are also significant. Indeed, high G + C reduces the photoreactivity due to the specific bipyrimidine composition, but it also impacts a combination of factors including DNA, RNA and protein-level features. Life for halophilic archaea must be well-suited for the dual extremes of both UV exposure and hypersalinity.
The authors would like to thank the Utah NASA Space Grant Consortium and the Lawrence T. Dee-Janet T. Dee Foundation for funding D.L.J. and B.K.B. In addition, we express much gratitude to Chrono Nu for early guidance in the Dinucleotide Counts script development and Kendall M. Tate for manuscript editing. Jaimi K. Butler and the staff of Great Salt Lake Institute have been invaluable assets in the development of this project.
The code for the R script, "DinucleotideCounts," can be accessed at: <http://www.mdpi.com/2075-1729/6/3/37/s1> or <https://github.com/jones1040/DinucleotideCounts>.
D.L.J. and B.K.B. conceived and designed the experiments; D.L.J. performed the experiments; D.L.J. and B.K.B. analyzed the data; D.L.J. contributed analysis tools; D.L.J. and B.K.B. wrote the paper.
The authors declare no conflict of interest.
![Thymine-thymine cyclobutane pyrimidine dimer (T\^T CPD) forms via butane ring cyclization between adjacent thymines on the same strand of DNA. Similar chemistry occurs at the other bipyrimidine nucleotides TC, CT, and CC \[[@B27-life-06-00037]\].](life-06-00037-g001){#life-06-00037-f001}
{#life-06-00037-f002}
{#life-06-00037-f003}
{#life-06-00037-f004}
{#life-06-00037-f005}
life-06-00037-t001_Table 1
######
Halophilic archaea sampled for the present study, representative of all species with full genome sequences presently available (taxid: 183963). Data were obtained from the National Center for Biotechnology Information (NCBI) database \[[@B21-life-06-00037]\].
Species Strain G + C (%) Size (Mb) Genes
------------------------------------- ------------- ----------- ----------- -------
*Halalkalicoccus jeotgali* B3 62.56 3.699 3717
*Halanaeroarchaeum sulfurireducens* HSR2 62.86 2.210 2213
*Haloarcula hispanica* N601 62.47 3.902 3825
*Haloarcula marismortui* ATCC 43049 61.14 4.275 4226
*Haloarcula* sp. CBA1115 61.98 4.225 4108
*Halobacterium hubeiense* JI20-1 66.58 3.130 3189
*Halobacterium salinarum* NRC-1 65.92 2.571 2629
*Halobacterium* sp. DL1 66.44 3.163 3237
*Haloferax gibbonsii* ARA6 66.24 3.918 3783
*Haloferax mediterranei* ATCC 33500 60.26 3.905 3800
*Haloferax volcanii* DS2 65.46 4.013 3925
*Halogeometricum borinquense* DSM 11551 59.97 3.944 3838
*Halomicrobium mukohataei* DSM 12286 65.51 3.332 3293
*Halopiger xanaduensis* SH-6 65.20 4.355 4174
*Haloquadratum walsbyi* DSM 16790 47.90 3.179 2827
*Halorhabdus tiamatea* SARL4B 62.77 3.146 3069
*Halorhabdus utahensis* DSM 12940 62.90 3.117 2969
*Halorubrum lacusprofundi* ATCC 49239 63.95 3.693 3523
*Halostagnicola larsenii* XH-48 60.87 4.131 3954
*Haloterrigena turkmenica* DSM 5511 64.25 5.441 5074
*Halovivax ruber* XH-70 64.30 3.224 3187
*Natrialba magadii* ATCC 43099 61.03 4.444 4128
*Natrinema pellirubrum* DSM 15624 63.96 4.354 4178
*Natrinema* sp. J7-2 64.06 3.794 3681
*Natronobacterium gregoryi* SP2 62.20 3.788 3720
*Natronococcus occultus* SP4 64.63 4.314 4162
*Natronomonas moolapensis* 8.8.11 64.50 2.913 2793
*Natronomonas pharaonis* DSM 2160 63.08 2.750 2799
*Salinarchaeum* sp. Harcht-Bsk1 66.60 3.255 3036
life-06-00037-t002_Table 2
######
G + C contents (%) of all sample groups utilized in the present study (archaea taxid: 2157, *n* = 68; bacteria taxid: 2, *n* = 101, cyanobacteria taxid: 1117, *n* = 32; enterobacteriaceae taxid: 91347, *n* = 42; halophilic archaea taxid: 183963, *n* = 29). Sample means are denoted with +/− 1.96 standard errors. Data was obtained from the NCBI database \[[@B21-life-06-00037]\].
Sample Group Mean G + C Content (%) Median (%)
-------------------------- ------------------------ ------------
(Non-Halophilic) Archaea 43.87 +/− 2.30 43.05
Bacteria 49.10 +/− 2.55 47.25
Cyanobacteria 44.21 +/− 3.08 42.65
Enterobacteriaceae 47.83 +/− 3.17 51.73
Halophilic Archaea 63.09 +/− 1.28 63.95
| {
"pile_set_name": "PubMed Central"
} |
Introduction
============
Hematopoietic stem cells (HSC) reside in a special bone marrow (BM) niche, which regulates their localization, self-renewal and differentiation. Studies have identified several major cell types of the niche, including mesenchymal stem cells (MSC), osteolineage cells (OLC), adipocytes and vascular endothelial cells (EC).^[@b1-1031969]--[@b5-1031969]^ Besides the key cellular components, some growth and survival factors are also indispensable components of the niche, including C-X-C motif chemokine 12 ligand (CXCL12), vascular cell adhesion molecule1 (VCAM1),^[@b6-1031969],[@b7-1031969]^ stem cell factor (SCF)^[@b4-1031969]^ and osteopontin.^[@b8-1031969]^ A sophisticated network of interactions between these multiple BM cells, extracellular factors and adhesion molecules is essential to regulate different HSC properties during homeostasis and keep normal hematopoiesis in check.
Acute myeloid leukemia (AML) has been considered for decades to be a disorder intrinsic to hematopoietic cells; however, evidence is accumulating that the microenvironment exerts more than a mere bystander effect. Leukemic cells can remodel the niche into a permissive environment favoring leukemic stem cell (LSC) expansion over normal HSC maintenance.^[@b9-1031969]^ Recently, emerging evidence even points to a role for the BM niche as a driver of disease maintenance/progression. Krause *et al*. showed that osteoblast-specific activation of the parathyroid hormone receptor enhances *MLL-AF9* oncogene-induced AML in mouse transplantation models.^[@b10-1031969]^ To date, there are still few studies concerning the role of the bone marrow niche in initiating and maintaining AML and relevant mechanisms remain elusive.
TWIST1 is a highly conserved transcription factor belonging to the basic helix-loop-helix family and is implicated in diverse developmental systems.^[@b11-1031969]--[@b13-1031969]^ Studies have revealed that TWIST1 is a key regulator of MSC self-renewal, enhances their life-span, inhibits MSC osteo/chondrogenic differentiation and promotes adipogenic differentiation.^[@b14-1031969]--[@b16-1031969]^ *Twist1* haploisufficiency leads to Saethre-Chotzen syndrome, which is characterized by alterations in osteogenic precursor cell proliferation, differentiation and survival.^[@b17-1031969]^ Recent studies have demonstrated that TWIST1 promotes angiogenesis by inducing EC proliferation and migration, and deregulation of this mechanism mediates pathological angiogenesis.^[@b18-1031969],[@b19-1031969]^ Arthur *et al*. showed that overexpression of *Twist1* in MSC enhances the capacity to maintain human CD34^+^ cells in long-term culture-initiating cell assays through increasing *Cxcl12* expression.^[@b20-1031969]^ However, the effects of TWIST1 on multiple niche elements and its modulation of normal HSC maintenance and leukemia progression *in vivo* have not been functionally characterized so far.
To explore this issue, we generated a murine model of a *Twist1*-deficient microenvironment. We showed that the major niche cellular components and factors changed remarkably upon *Twist1* deletion, causing severe dysfunction of normal HSC. Nevertheless, these alterations of the BM microenvironment promoted *MLL-AF9* oncogene-induced AML progression in mouse transplantation models, not only pointing to TWIST1 as an instructive signal modulating the stem cell niche, but also emphasizing the importance of the niche for AML development.
Methods
=======
Mice
----
*Twist1*^flox/flox^ mice were purchased from Mutant Mouse Regional Resource Centers. *ER-Cre* mice were a gift from Professor Weiping Yuan. C57BL/6 and B6.SJL mice were purchased from the animal facility of State Key Laboratory of Experimental Hematology. *Twist1*^flox/flox^ mice were crossed with *ER-Cre* mice to generate *ER-Cre*;*Twist1*^fl/fl^ and *ER-Cre*;*Twist1*^+/+^ mice. Eight- to 12-week old mice were used. Cre expression was induced by daily intraperitoneal injection of tamoxifen (75 mg/kg of total body weight in corn oil; Sigma-Aldrich, St. Louis, MO, USA) for 5 days. All animal procedures complied with the animal care guidelines approved by the Institutional Animal Care and Use Committees of the State Key Laboratory of Experimental Hematology.
Transplantation assays
----------------------
For non-competitive BM transplantation, to create the chimeras described in *Online Supplementary Figure S1A*, 2x10^[@b6-1031969]^ whole BM cells from B6.SJL (CD45.1) mice were transplanted into *ER-Cre*;*Twist1*^+/+^ or *ER-Cre*;*Twist1*^fl/fl^ (CD45.2) recipients that were lethally irradiated (9.5 Gy from a Cesium source, 4-24 h before transplantation). Sixteen weeks later, tamoxifen was injected to induce *Twist1* deletion. For competitive transplantation, 300 BM long-term HSC (CD45.1) from tamoxifen-treated *ER-Cre*;*Twist1*^+/+^ or *ER-Cre*;*Twist1*^fl/fl^ chimeric mice were mixed with 2×10^[@b5-1031969]^ congenic BM support cells and injected into lethally irradiated CD45.2 recipients. For the *MLL-AF9* AML model, 5x10^[@b5-1031969]^ GFP^+^ leukemic cells were transplanted into *Twist1*-deleted or control chimeric recipients.
Flow cytometry analysis and cell sorting
----------------------------------------
The BM cell suspensions were flushed from femora and tibiae. Spleen cells were pestled by the plug of a 10 mL syringe. The cells were then filtered through a 74 mm nylon mesh. For flow cytometric analysis of stromal cells, BM was flushed using phosphate-buffered saline with 2% bovine serum, the bones were minced with scissors, then the plugs were digested in 1 mg/mL collagenase I (OLC) or IV (MSC and EC) (Sigma-Aldrich) dissolved in Hank's balanced salt solution with 10% fetal bovine serum for 90 min (collagenase I) or 30 min (collagenase IV) at 37°C. The dissociated cells were collected and kept on ice. Cells were incubated with conjugated antibodies. Stained cells were analyzed with FACS LSR II or sorted with a FACS Aria II instrument (BD Biosciences, Franklin Lakes, NJ, USA). Data were analyzed by FlowJo software.
Statistical analysis
--------------------
The significance of differences between two groups was determined using unpaired two-tailed Student *t* tests. Data are presented as means ± standard deviations. Overall survival curves were plotted according to the Kaplan-Meier method with the log-rank test applied for comparisons. ^\*^*P*\<0.05, ^\*\*^*P*\<0.01, ^\*\*\*^*P*\<0.001.
Details of other experimental procedures are given in the *Online Supplementary Methods*.
Results
=======
Microenvironmental *Twist1* deficiency leads to decreased numbers of mesenchymal stem cells and mature osteoblasts, an increased proportion of endothelial cells, and altered expression of cell factor genes
-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
To explore the role of TWIST1 in the BM niche and its regulation of HSC, we generated microenvironment *Twist1*-deleted and control chimeric mice according to the method described by Schreck and Saez.^[@b21-1031969],[@b22-1031969]^ In brief, 2×10^[@b6-1031969]^ BM cells from B6.SJL wild-type (WT) mice (CD45.1) were transplanted into *ER-Cre;Twist^[@b1-1031969]^*^fl/fl^ and *ER-Cre*;*Twist1*^+/+^ recipients (CD45.2) (*Online Supplementary Figure S1A*). Sixteen weeks later, about 90% of the cells in the peripheral blood of recipients were donor-derived cells (*Online Supplementary Figure S1B*). Tamoxifen was then injected intraperitoneally for 5 days to induce *Twist1* deletion. Two weeks after the last injection, mRNA detection demonstrated that *Twist1* had been knocked out in all the MSC, OLC, and EC isolated from *Twist1*^Δ/Δ^ mice with similar knockout levels (*Online Supplementary Figure S1C*), while the expression of *Twist2* was almost unchanged (*data not shown*).
To define components of the *Twist1*-deleted BM microenvironment that may be altered, stromal populations and extracellular factors were assessed in *Twist1*-deleted and control chimeric mice. We observed that conditional deletion of *Twist1* led to a significant decrease in the number of MSC (CD140a^+^CD51^+^CD45/Ter119/CD31-)^[@b23-1031969]^ in the BM compared with that in control mice, as determined by flow cytometry ([Figure 1A](#f1-1031969){ref-type="fig"}). The decrease in MSC number was further confirmed by a fibroblastic colony-forming unit assay (*Online Supplementary Figure S2A*). Furthermore, *Twist1*-deleted MSC showed a decrease in proliferative cells and an increase in apoptotic cells (*Online Supplementary Figure S2B,C*), indicating the mechanism underlying the reduced number of MSC.
{#f1-1031969}
*Twist1* deficiency resulted in a marked increase in the frequency of Sca-1^−^/CD166^+^ stromal cells ([Figure 1B](#f1-1031969){ref-type="fig"}), which include immature and mature OLC.^[@b24-1031969]^ Meanwhile, the expression of osteoblastic differentiation genes *Runx2*, *Ogn* and *Gpnmb*^[@b14-1031969],[@b23-1031969]^ was significantly upregulated in *Twist1*-deleted MSC (*Online Supplementary Figure S2D*). To assess the ability of MSC to differentiate into the osteoblastic lineage, we induced osteoblast differentiation in MSC and found that *Twist1* deletion clearly increased alkaline phosphatase activity and matrix mineralization (*Online Supplementary Figure S2E*). These results establish that *Twist1* deficiency enhanced MSC commitment toward osteoblasts. However, expression of the mature osteoblast marker, *Bglap*, was downregulated (*Online Supplementary Figure S2F*). In addition, micro-computed tomography analysis also revealed a significant decrease in mature osteoblasts in *Twist1*-deleted mice, which was reflected by a reduction of trabecular volume ([Figure 1C](#f1-1031969){ref-type="fig"}). Collectively, *Twist1* deletion promotes MSC to differentiate toward the osteoblast lineage with a block of mature osteoblast differentiation.
Emerging data demonstrate the role of vascular EC in HSC maintenance, and arterioles and sinusoids exhibit dif ferent properties in relation to HSC distribution and quiescence.^[@b4-1031969],[@b25-1031969],[@b26-1031969]^ Ciuculescu *et al*. reported that *Rac* deletion in MSC leads to an inverted ratio of marrow arterioles and sinusoid vessels and impaired hematopoiesis.^[@b27-1031969]^ We observed that *Twist1* deletion resulted in increased CD45-Ter119^−^CD31^+^, CD45^−^Ter119^−^CD31^+^Sca1^+^ and CD45^−^Ter119- CD31^+^Sca1^−^ populations enriched for total EC, arteriolar EC and sinusoidal EC,^[@b28-1031969]^ respectively ([Figure 1D](#f1-1031969){ref-type="fig"}). The increase of arteriolar and sinusoidal EC was further confirmed by the observation that arteries (Sca-1-staining) and sinusoids (Endoglin-staining) were both significantly increased in *Twist1*-deleted mice compared with control mice by immunofluorescence of femoral sections ([Figure 1E](#f1-1031969){ref-type="fig"}). The increase of EC may be the result of cell proliferation as determined by increased bromodeoxyuridine incorporation into CD45^−^Ter119^−^CD31^+^ stromal cells in *Twist1*-deleted mice ([Figure 1F](#f1-1031969){ref-type="fig"}). We next performed a tube formation assay to determine the effect of TWIST1 on new blood vessel development. Consistent with increased microvessels in knockout mice *in vivo*, capillary tube formation of *Twist1*-deleted EC was also increased on matrigel ([Figure 1G,H](#f1-1031969){ref-type="fig"}), indicating that *Twist1* deletion promotes angiogenesis.
To further evaluate the impact of *Twist1* deficiency on cell factors, we performed quantitative real-time polymerase chain reaction to analyze the expression of key niche factors. The results showed significant decreases in the expression of *Cxcl12*, *Vcam1*, *Angiopoietin-1* (*Angpt1*) and *Scf*, particularly the membrane-bound isoform *Scf* (*m220 Scf*), which was found to be extremely important for HSC maintenance, in MSC (5×10^[@b4-1031969]^ cells) from *Twist1*-deleted mice as compared to control mice ([Figure 2A](#f2-1031969){ref-type="fig"}). The expression of *Opn*, which negatively regulates the HSC pool, was obviously increased in both MSC and OLC ([Figure 2B](#f2-1031969){ref-type="fig"}). Enzyme-linked immunoassay demonstrated the reduced protein levels of CXCL12, VCAM1, SCF and elevated level of osteopontin ([Figure 2C](#f2-1031969){ref-type="fig"}) in BM supernatants of *Twist1*-deleted mice as compared to those in control mice.
{#f2-1031969}
Collectively, *Twist1* deletion leads to significant alterations in various key niche components, demonstrating its functional importance in the BM microenvironment, and implying its potential regulatory role in HSC maintenance.
Microenvironmental *Twist1* deficiency impairs the homing and retention of hematopoietic stem cells but promotes their mobilization
-----------------------------------------------------------------------------------------------------------------------------------
Most HSC are retained in the BM niche in a quiescent, nonmotile mode by adhesion to stromal cells, which are essential for normal hematopoiesis and for protection from myelotoxic injury. *Twist1* deletion resulted in decreased expression of BM CXCL12 and VCAM1, which are critical for HSC retention, homing and mobilization,^[@b29-1031969],[@b30-1031969]^ implying that TWIST1 may have a functional impact on the migration of HSC. We first evaluated homing of normal HSC to the BM of *Twist1*-deleted mice. Freshly isolated c-Kit^+^ cells from B6.SJL (CD45.1) mice were injected into lethally irradiated *Twist1*-deleted or control mice. Sixteen hours after transplantation, the absolute number of CD45.1^+^ and CD45.1^+^ LSK (Lin^-^Sca-1^+^c-Kit^+^) cells was significantly decreased in *Twist1*-deleted mice compared to the number in controls ([Figure 3A](#f3-1031969){ref-type="fig"}), indicating that *Twist1* deficiency impairs homing of hematopoietic stem/progenitor cells (HSPC) to the BM microenvironment.
{#f3-1031969}
We next determined the contribution of TWIST1 to HSC retention and mobilization by assessing total cells and HSPC in the BM, spleen and peripheral blood of *Twist1*-deleted mice and control mice ([Figure 3B](#f3-1031969){ref-type="fig"}). The results revealed that *Twist1* deletion led to decreases in the numbers of total cells ([Figure 3C](#f3-1031969){ref-type="fig"}), colony-forming units ([Figure 3D](#f3-1031969){ref-type="fig"}), and HSC-enriched SLAM LSK (CD150^+^CD48-LSK) in the BM ([Figure 3E,F](#f3-1031969){ref-type="fig"}) but significant increases in those in the spleen and circulation ([Figure 3G-J](#f3-1031969){ref-type="fig"}).
Granulocyte colony-stimulating factor (G-CSF) is a hematopoietic cytokine known as the prototypic mobilizing agent.^[@b31-1031969]^ To determine whether *Twist1* deletion promot ed secretion of endogenous G-CSF, we examined G-CSF levels in the BM supernatants of *Twist1*-deleted and control mice by enzyme-linked immunosorbent assay. We found that G-CSF protein expression was significantly elevated in *Twist1*-deficient mice as compared to that in controls. Furthermore, quantitative real-time polymerase chain reaction results revealed that among MSC, OLC, EC and macrophagocytes, which are all producers of G-CSF, OLC and macrophagocytes were the major source of G-CSF (*Online Supplementary Figure S3A,B*). We next assessed HSPC mobilization after infusion of exogenous G-CSF and found that *Twist1* deletion significantly increased exogenous G-CSF-induced mobilization of total cells and colony-forming cells to the blood and spleen, and SLAM LSK to the spleen (*Online Supplementary Figure S3C-H*).
Taken together, these results demonstrate a functional role of TWIST1 in HSC homing, retention, baseline mobilization and stress mobilization in response to G-CSF.
*Twist1* deficiency impairs hematopoietic stem cell quiescence and self-renewal, and induces enhanced early myeloid lineage differentiation
-------------------------------------------------------------------------------------------------------------------------------------------
Quiescence, self-renewal and committed differentiation are important properties of HSC, which could be controlled by stromal cells, extracellular matrix, cytokines and chemokines.^[@b32-1031969],[@b33-1031969]^ Our study revealed that *Twist1* deletion altered multiple stromal cells and the level of expression of HSC supportive factors, so we next investigated whether these HSC features were consequently changed. Immunophenotypic analysis demonstrated that *Twist1* deletion resulted in a significant decrease in the number of long-term HSC (CD34^−^Flt3^−^LSK) in the BM (*Online Supplementary Figure S4A*). Ki67 staining revealed a significant decrease in the percentage of HSC (CD34^−^LSK) in G~0~ phase in *Twist1*-deleted mice, together with an increase in G~1~ phase ([Figure 4A](#f4-1031969){ref-type="fig"}). Bromodeoxyuridine labeling further identified a higher frequency of proliferating cells in CD34-LSK cells from *Twist1*-deleted mice compared to control mice ([Figure 4B](#f4-1031969){ref-type="fig"}). These data suggest that TWIST1 in the microenvironment plays an important role in maintaining HSC quiescence, and loss of *Twist1* drives aberrant proliferation of HSC.
{#f4-1031969}
To clarify the role of TWIST1 in HSC self-renewal, we conducted serial transplantation assays ([Figure 4C](#f4-1031969){ref-type="fig"}). Three hundred long-term HSC (CD34^−^Flt3^−^LSK, CD45.1) from *Twist1*-deleted or control chimeric mice were transplanted into lethally irradiated recipients (CD45.2), together with 2×10^[@b5-1031969]^ CD45.2^+^ support BM cells. Then secondary transplantation assays were performed 16 weeks later. Three hundred long-term HSC (CD45.1^+^) from the primary recipients were transplanted into lethally irradiated secondary recipients, together with 2×10^[@b5-1031969]^ CD45.2^+^ support BM cells. Donor cells from *Twist1*-deleted mice showed lower engraftment capacity than control cells out to 16 weeks in both primary and secondary transplantation ([Figure 4D](#f4-1031969){ref-type="fig"}). From the above, it can be concluded that *Twist1* deficiency in niche cells impairs HSC self-renewal capacity.
We then performed FACS analysis to evaluate the differentiation capacity of HSC, and found increases in the numbers of common myeloid progenitors (CD34^+^CD16/32^−^Lin^−^Sca-1^−^c-Kit^+^, 1.5-fold, *P*=0.047) and granulocyte/macrophage progenitors (GMP: CD34^+^CD16/32^+^Lin^−^Sca-1^−^c-Kit^+^, 1.3-fold, *P*=0.022) in the BM of *Twist1*-deleted mice in comparison with those in controls, accompanied by decreases in the numbers of megakaryocyte/erythroid progenitors (MEP, CD34-CD16/32^−^Lin^−^Sca-1^−^c-Kit^+^, 1.5-fold, *P*=0.028) and common lymphoid progenitors (Lin^−^Sca^−^1^low^c-Kit^low^IL7R^+^, 2.5-fold, *P*=0.046) ([Figure 4E,F](#f4-1031969){ref-type="fig"}). In accordance with the change of progenitors, the proportion of mature myeloid cells (Mac1^+^, 1.2-fold, *P*=0.007; Gr-1^+^, 1.2-fold, *P*=0.008; Mac-1^+^Gr-1^+^, 1.2-fold, *P*=0.011) was also significantly increased, accompanied by reduced proportions of B lymphoid cells (B220^+^, 1.4-fold, *P*=0.004) and erythrocytes (Ter119^+^, 1.3- fold, *P*=0.008) in the BM of *Twist1*-deleted mice ([Figure 4G,H](#f4-1031969){ref-type="fig"}). These results suggest that *Twist1* deletion in niche cells promotes HSC differentiation into the myeloid lineage. We observed that apart from the increase of HSPC in the spleen of *Twist1*-deleted mice, the numbers of mature myeloid cells and erythrocytes were also increased, while the number of lymphoid cells remained unchanged (*Online Supplementary Figure S4B-D*), indicating the occurrence of extramedullary hematopoiesis in the spleen of *Twist1*-deleted mice.
Taken together, these data suggest that TWIST1 in the BM microenvironment plays a critical role in HSC maintenance, and *Twist1* deletion impairs all the fundamental features of HSC.
*Twist1* deficiency promotes progression of *MLL-AF9*-induced acute myeloid leukemia
------------------------------------------------------------------------------------
It has been reported that the *MLL-AF9* AML model exhibited multiple alterations in the niche compartments, including decreased frequencies of MSC and osteoblasts, an increased number of vascular EC, and downregulated expression of *Vcam1*, *Cxcl12*, *Angpt1*, and *Scf*, together with upregulation of *Opn*.^[@b34-1031969]^ Intriguingly, these phenotypes are quite similar to the niche alterations in our *Twist1*-deleted mice. Considering the emerging importance of the BM niche for leukemia maintenance and progression, we were tempted to speculate that the altered niche in *Twist1*-deleted mice may play a role in the development of *MLL-AF9* AML. To validate this hypothesis, we transduced WT BM c-Kit^+^ cells with retrovirus expressing *MLL-AF9*, and injected these cells into lethally irradiated WT recipient mice, in which the disease was rapidly induced with massive BM and spleen infiltration of GFP^+^ leukemic cells. We next injected 5×10^[@b5-1031969]^ GFP^+^ spleen cells from these mice into *Twist1*-deleted and control chimeric recipient mice ([Figure 5A](#f5-1031969){ref-type="fig"}). Notably, the overall survival of *Twist1*-deleted recipients was significantly shorter than that of control recipients ([Figure 5B](#f5-1031969){ref-type="fig"}), and T*wist1*-deleted recipient mice exhibited a greater infiltration of total cells and leukemic cells in the BM, peripheral blood and spleen than that of control mice (*Online Supplementary Figure S5A-G*).
{#f5-1031969}
Previous studies using *MLL-AF9* AML models have established that LSC are enriched in the leukemic GMP populations (IL-7R^−^Lin^−^GFP^+^c-Kit^hi^CD34^+^CD16/32^hi^)^[@b34-1031969],[@b35-1031969]^ or c-Kit^+^Gr-1^−.[@b36-1031969]^ We found that the infiltration of leukemic GMP cells was significantly increased in the BM and periphery of *Twist1*-deleted recipient mice compared to control mice ([Figure 5C,D](#f5-1031969){ref-type="fig"}), and so was that of GFP^+^c-Kit^+^Gr-1^−^ cells (*Online Supplementary Figure S6A-C*). Additionally, cell cycle analysis showed that the proportion of LSC (GFP^+^c-Kit^+^Gr-1^−^) in the G~0~ phase was decreased and that in the G~1~ or S/G~2~/M phases was increased in both BM and spleen from *Twist1*-deleted mice, but rates of LSC apoptosis in BM and spleen did not differ between *Twist1*-deleted and control mice (*Online Supplementary Figure S6D-G*). We then performed secondary transplants using LSC ([Figure 5A](#f5-1031969){ref-type="fig"}), injecting 5000 GFP^+^c-Kit^+^Gr-1^−^ cells isolated from *Twist1*-deleted and control mice into WT recipient mice. We found that the survival of the secondary recipients of LSC grafts from leukemic *Twist1*-deleted donors was significantly reduced compared to that of recipients of control LSC grafts ([Figure 5E](#f5-1031969){ref-type="fig"}).
Taken together, these results demonstrate that the altered niche in *Twist1*-deleted mice accelerates the pro gression of *MLL-AF9*-induced AML by increasing the expansion and leukemogenic capacity of LSC.
Microenvironmental *Twist1* deletion promotes acute myeloid leukemia development partially through the Notch signaling pathway
------------------------------------------------------------------------------------------------------------------------------
To understand the mechanism underlying TWIST1 function in the development of AML, we performed RNA- sequencing analysis on BM stromal cells isolated from *Twist1*-deleted or control chimeric mice. There were 6948 differentially expressed genes (4764 downregulated and 2184 upregulated; fold change of \>2 and *P* value \<0.05). Analysis of leukemia-related signaling revealed a marked increase of *Jagged-2* expression in stromal cells from *Twist1*-deleted mice ([Figure 6A](#f6-1031969){ref-type="fig"}). Previous studies have demonstrated that TWIST1 regulates the Notch pathway in prospective coronal suture mesenchyme and osteoprogenitors.^[@b37-1031969],[@b38-1031969]^ Moreover, aberrant Notch signaling is a common mechanism in niche-induced AML and pre-leukemic conditions.^[@b39-1031969]--[@b42-1031969]^ To investigate whether *Twist1* deficiency promotes the development of *MLL-AF9* AML through Notch signaling, we determined the expression of all Notch ligands (*Dll1*, *Dll3*, *Dll4*, *Jagged-1*, *Jagged-2*) in MSC, OLC and EC of *Twist1*-deleted mice. The results revealed that *Jagged-2* was significantly upregulated in all these cells ([Figure 6B-D](#f6-1031969){ref-type="fig"}). Additionally, the levels of expression of all four Notch receptors (*Notch1-4*), cleaved Notch1 and the Notch targets *Dtx*, *Hes1*, *Hes5*, *Hey1*, and *Hey2* were significantly upregulated in LSC from *Twist1*-deleted mice compared to those of controls ([Figure 6E-G](#f6-1031969){ref-type="fig"}), indicating increased Notch signaling in this population. Furthermore, pharmacological inhibition of Notch signaling with a γ-secretase inhibitor (DBZ) ([Figure 6H-K](#f6-1031969){ref-type="fig"}) or blockade of Notch with dominant-negative MAML1 (DNMAML1) (*Online Supplementary Figure S7*) partially rescued leukemic cell infiltration and LSC engraftment, and prolonged the overall survival of *Twist1*-deleted recipients. These data suggest that a *Twist1*-deleted microenvironment contributes to *MLL-AF9* AML development at least in part via Notch signaling.
{#f6-1031969}
Discussion
==========
In the current study, we demonstrated that excision of the *Twist1* gene from the BM microenvironment resulted in a significant decrease in the numbers of MSC and mature osteoblasts, and an increase in the number of EC. The expression of CXCL12, VCAM1 and SCF was reduced, while that of osteopontin was increased. These changes led to a marked impairment of HSC localization, self-renewal, quiescence and differentiation. By transplanting MLL-AF9 cells into the *Twist1*-deleted and control chimeric mice, we verified that *Twist1* deletion resulted in accelerated development of leukemia, at least partially through Notch signaling ([Figure 7](#f7-1031969){ref-type="fig"}). These results reveal the essential role of TWIST1 in supporting normal hematopoiesis and perturbing AML development.
{#f7-1031969}
In our model, *Twist1* deletion in the BM microenvironment leads to an increased number of EC and microvessel density, suggesting the existence of an indirect and powerful mechanism for promoting angiogenesis *in vivo*. Ohki *et al*. reported that G-CSF can markedly increase vascular endothelial growth factor (VEGF) release from G-CSF-responsive myelomonocytic cells, which promote the co-recruitment of VEGFR1^+^ (VEGF receptor 1) cells contributing to neo-angiogenesis.^[@b43-1031969]^ Since we have found elevated G- CSF secretion in the BM supernatant of *Twist1*-deleted mice, we then determined VEGF mRNA expression in BM cells and its protein concentration in BM supernatant. As expected, the results revealed a marked increase of VEGF production in *Twist1*-deleted mice compared with control mice (*Online Supplementary Figure S8A,B*). These observations provide a possible explanation for the EC alterations, i.e., *Twist1* deficiency in the BM microenvironment leads to increased production of G-CSF, which in turn induces the secretion of VEGF, exerting a promotive effect on the proliferation of EC. This effect overrides the direct inhibitory role of *Twist1* deletion on EC, and results in increased numbers of EC. As VEGF can be produced by various cell types, the specific mechanism needs further investigation.
Our understanding of niche contributions to AML has increased tremendously over the past decade. However, most studies have focused on how the leukemic cells actively shape their microenvironment to reinforce disease progression. There are a limited number of reports showing that certain niche alterations can act as a driver of AML initiation or progression, without having been educated by leukemic cells.^[@b40-1031969],[@b44-1031969],[@b45-1031969]^ Our present study demonstrates that environmental deletion of *Twist1*, a conserved transcriptional factor gene, results in diverse cellular and factor alterations common to the microenvironmental dysregulation exhibited by AML. These alterations appear to be predis posing or initiating factors for the evolution of AML and point to TWIST1 as an instructive signal that alters the function of the niche.
The opposing effects of TWIST1 on normal HSC and LSC found in this study are of value. In an effort to elucidate the underlying mechanism, we performed RNA-sequencing and quantitative real-time polymerase chain reaction analysis. We found that *Twist1* deletion leads to increased expression of the Notch ligand *Jagged-2* in all the OLC, EC and MSC. LSC from *Twist1*-deleted chimeric mice have robust expression of all Notch receptors and canonical downstream Notch target genes, suggesting the aberrant activation of Notch signaling. A previous report showed that Notch activation promotes expansion and self-renewal of LSC,^[@b40-1031969]^ consistent with our results obtained by deletion of *Twist1*. We also found activation of Notch receptors and target genes in normal HSC (CD34^−^LSK) after *Twist1* deletion (*Online Supplementary Figure S9*). In contrast to the promoting role in LSC, Notch activation in HSC has been reported to cause loss of stem cell quiescence,^[@b46-1031969]^ which often correlates with impaired self-renewal capacity of HSC, in line with the observations in our mouse model. Besides the direct impact of activated Notch signaling on LSC and HSC, the augmented proliferation and infiltration of LSC compared to normal HSC could be favorable for their competition for the niche over HSC. Various studies have demonstrated that LSC could positively remodel the BM microenvironment to enhance support of LSC at the expense of HSC,^[@b47-1031969],[@b48-1031969]^ and this remodeling may in turn further promote leukemia progression and impair normal hematopoiesis. In addition, the reduced expression of *Cxcl12*, *Scf* and *Angpt1* in *Twist1*-deleted mice may also account for the opposing impact of *Twist1* deletion on HSC and LSC, since compared with HSC, LSC are less factor-dependent.^[@b34-1031969],[@b45-1031969],[@b49-1031969]^
In consideration of the important role of TWIST1 in reg ulating MSC, osteoblasts and EC, and to exclude the interference of hematopoietic cells, which were found to express *Twist1* in our previous work,^[@b50-1031969]^ we generated the chimeric mouse model, in which *Twist1* was diffusely deleted in the BM microenvironment. The BM niche comprises multiple cell types, which not only closely connect but also communicate with each other via cell factors and adhesion molecules throughout the BM. Due to the complexity of the niche, an overall environmental knockout strategy will facilitate the detection of direct and indirect effects of TWIST1 on the niche components. Utilizing our model, we uncovered extensive cellular and factor alterations in the BM niche and the AML-like microenvironmental phenotype resulting from *Twist1* deficiency, and demonstrated the essential role of TWIST1 in HSC maintenance and suppression of AML evolution. To refine the contribution of different cell populations, studies in which *Twist1* is modified in specific stromal cell subsets are ongoing in our laboratory.
In conclusion, we used a *Twist1*-deficient chimera model to obtain, for the first time *in vivo*, direct evidence that TWIST1 in the microenvironment plays a key role in maintaining the hematopoietic phenotype and hampering leukemia progression. These findings provide new insights into the importance of the BM niche for AML development, and lay the foundation for tackling leukemia from a different angle to improve current treatments.
Supplementary Material
======================
###### Lui et al Graphical Abstract
###### Liu et al. Supplementary Appendix
###### Disclosures and Contributions
This work was supported by grants from The National Key Research and Development Program of China (2016YFA0100603), CAMS Initiative for Innovative Medicine (2016-I2M-1-017), National Natural Science Foundation of China (81470278, 81670158, 81600138, and 81700106), and Tianjin Municipal Science and Technology Commission grants (17JCZDJC35100 and 17JCQNJC10800).
Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: [www.haematologica.org/content/103/12/1969](www.haematologica.org/content/103/12/1969)
| {
"pile_set_name": "PubMed Central"
} |
Sepsis and septic shock are the leading causes of death in intensive care units (ICUs) and have significant mortality rates (as high as 60%) and health-care cost burdens (40% of total ICU expenditures)[@b1]. Recent epidemiological studies have shown that approximately 47% of ICU patients with severe sepsis have positive cultures for Gram-positive bacteria, and the contribution of these bacteria to sepsis has been steadily increasing since 1979[@b2][@b3]. The clinical manifestation of sepsis is highly variable and influenced by several factors, including the health and immune status of the patient and the infectious agent involved. In the ICUs, where sepsis is a common occurrence following surgery, attention needs to be directed to the hospital-specific attributes, including (but not limited to) altered metabolism and immune-modulation due to opioid administration, to gain mechanistic insights into poly-microbial sepsis and its potential remedy[@b4][@b5].
Due to their analgesic and sedative properties, opioids are widely used in ICUs to optimize patient comfort and facilitate mechanical ventilation[@b6]. The immunosuppressive effects of opioids are well documented[@b5] and raise safety issues, especially in ICU patients. In humans, higher circulating morphine levels are observed in patients with sepsis, severe sepsis, and septic shock[@b7], and several murine sepsis models show that morphine treatment induces bacterial translocation from the gut lumen into the peritoneal organs and circulatory system[@b8][@b9]. We have previously shown that morphine accelerates the progression of LPS-induced sepsis by modulating Toll-like receptor (TLR) pathways and altering endotoxin tolerance[@b10]. However, the exact mechanisms by which opioids modulate sepsis progression remain largely elusive. In the present study, we used cecal ligation and puncture (CLP), a well-established model for inducing poly-microbial sepsis, in C57BL/6J mice treated with either opioids or placebo. The survival rates of mice were analyzed to investigate the effects of opioids on sepsis progression. We demonstrated that both morphine and methadone treatment resulted in high mortality following CLP when compared with placebo-treated animals. Furthermore, morphine promoted bacterial dissemination and increased production of the pro-inflammatory cytokine interleukin-17A (IL-17A).
IL-17A is a member of the interleukin-17 (IL-17) family, which consists of a subset of cytokines that participate in both acute and chronic inflammatory responses. In various diseases, IL-17A is involved in host defense and implicated in excessive inflammation and overt tissue damage[@b11]. Although a few reports implicate IL-17A in poly-microbial sepsis[@b12][@b13], its role in the progression of Gram-positive sepsis is still unknown. Our study showed that overexpression of IL-17A following morphine treatment resulted in increased gut permeability, a higher bacterial load, sustained inflammation, and, subsequently, higher mortality. Concomitantly, neutralization of IL-17A protected morphine-treated animals from sepsis-induced mortality.
We (and others) have previously shown that the source of the bacteria that contributes to morphine-induced sepsis is derived from the commensal pool of the gut microbiome[@b8][@b9]. In this study, we further showed that morphine treatment induced enrichment of the Gram-positive bacteria *Staphylococcus* and *Enterococcus* in the gut lumen, the same species that were isolated from various systemic organs following CLP. Activation of TLR2 by the disseminated Gram-positive bacteria led to overexpression of IL-17A, resulting in higher mortality in the morphine-treated animals. These results are consistent with the clinical observation that *Staphylococcus aureus* is one of the most common Gram-positive isolates from patients with sepsis, and infection with *Enterococcus* species is considered as an independent factor associated with a greater risk of hospital death[@b1].
In summary, the current study provides insight into the influence of opioids on sepsis progression, showing that IL-17A may be a potential therapeutic target for the treatment of sepsis caused by Gram-positive infection, especially in ICU patients who are on a moderate to severe pain management regimen.
Results
=======
Opioids increase mortality in a poly-microbial sepsis model of CLP
------------------------------------------------------------------
To determine the effects of opioids on the progression of poly-microbial sepsis, wild-type (WT) mice were subjected to a CLP procedure and implanted with either a placebo or a slow release morphine pellet subcutaneously. As shown in [Fig. 1a](#f1){ref-type="fig"}, the survival rates were significantly reduced in morphine-treated mice following CLP. At 24 hours after CLP, all placebo-treated mice were alive compared with only 66.67% in the morphine-treated group. None of the morphine-treated mice survived beyond 96 hours, whereas 78.57% of the placebo-treated mice survived for the entire period of observation (7 days). All of the sham-operated mice treated with the placebo survived, whereas 80% of the sham-operated mice treated with morphine survived until the end of 7 days. To investigate the influence of other prescription opioids on the outcome of CLP-induced sepsis, mice were injected with methadone or saline following CLP. Methadone showed similar effects to morphine on the survival rates following CLP ([Fig. 1b](#f1){ref-type="fig"}). A total of 66.67% of the saline-treated CLP mice lived for 7 days, whereas no methadone-treated mice survived beyond the fifth day after CLP. All of the sham-operated mice injected with saline or methadone survived for 7 days. The morphine- and methadone-induced mortality following CLP was significantly reduced by the opioid receptor antagonist naltrexone ([Fig. 1c,d](#f1){ref-type="fig"}), indicating that opioid treatment exacerbated the outcome of poly-microbial sepsis in an opioid receptor-dependent manner.
Morphine promotes bacterial dissemination and inhibits bacterial clearance following CLP-induced sepsis
-------------------------------------------------------------------------------------------------------
Next, the bacterial load was determined in the peritoneal lavage fluid, mesenteric lymph node (MLN), liver, spleen, and blood at different time points following CLP in the presence or absence of morphine. As shown in [Fig. 2](#f2){ref-type="fig"}, at 24 hours, morphine alone induced bacterial translocation into the peritoneal cavity, MLN, liver, and spleen, confirming that morphine itself can compromise the gut epithelial barrier, as previously reported[@b8]. In the placebo-treated CLP animals, the bacterial load in the peritoneal lavage fluid, MLN, liver, spleen, and blood reached its highest levels at 24 hours, and subsequently decreased at 72 hours. At 168 hours after the CLP, almost all bacteria that disseminated into the peritoneal organs were cleared in the placebo-treated animals. In the morphine-treated CLP mice, a significant increase in the amount of bacteria was observed in the peritoneal organs and blood compared with the placebo-treated mice at both 24 and 72 hours, indicating that morphine promoted bacterial dissemination and inhibited bacterial clearance. Bacterial translocation in liver homogenates was further evaluated using bacterial 16 s ribosomal DNA (rDNA). Consistent with data obtained using bacterial cultures and colony forming units, the bacterial load as measured by 16 s rDNA was also significantly higher in the morphine treated-CLP animals than all other groups ([Supplementary Fig. 1](#S1){ref-type="supplementary-material"}). Because none of the animals survived beyond 96 hours in the morphine-treated CLP group, data regarding the bacterial load were not available for the last time point.
Morphine treatment promotes Gram-positive bacterial dissemination and modulates the gut microbiome
--------------------------------------------------------------------------------------------------
We next serotyped the bacterial species that disseminated into the MLN, liver, and spleen following CLP. In placebo-treated CLP animals, the most common bacterial species detected in the MLN, spleen, and liver were non-hemolytic Escherichia coli ([Fig. 3a](#f3){ref-type="fig"}), which are common Gram-negative commensal bacteria resident in the gut lumen; very few Enterococcus species were detected in the MLN and spleen. However, all MLN, spleen, and liver isolates from morphine-treated animals with or without CLP procedures revealed a significant prevalence of the Gram-positive families *Staphylococcus* and *Enterococcus* ([Fig. 3a](#f3){ref-type="fig"}). These data were validated using Illumina sequencing of 16 S rDNA from liver samples of morphine-treated CLP mice, which showed a prevalence of both the *Enterobacteriaceae* and *Enterococcus* families ([Supplementary Fig. 2a](#S1){ref-type="supplementary-material"}).
Analysis of the gut microbiome showed that morphine treatment induced enrichment of mostly the Firmicutes phylum and specifically the Gram-positive bacterial species *Staphylococcus sciuri*, *Staphylococcus cohnii*, and *Staphylococcus aureus* as well as *Enterococcus durans*, *Enterococcus casseliflavus*, *Enterococcus faecium*, and *Enterococcus faecalis* in the gut microbiome ([Fig. 3b,c](#f3){ref-type="fig"}). Interestingly, these species belonged to the same families that were observed to translocate to the peritoneal organs following morphine treatment. Morphine-induced alterations of the gut microbiome were antagonized by the opioid receptor antagonist naltrexone ([Fig. 3b,c](#f3){ref-type="fig"}), further validating that morphine treatment modulated the gut microbiome and thereby influenced the outcome of poly-microbial sepsis in an opioid receptor-dependent manner. This was also reflected in the bacterial sequences obtained from liver samples. When the beta-diversity of 16 S rDNA sequences from liver samples were compared between Placebo-treated (P), Placebo-treated CLP (PC), Morphine-treated CLP (MC), and Morphine + Naltrexone with CLP (MN) animals, the analysis showed that the MC animals clustered distinctly from the P and PC animals. However, morphine animals that were treated with naltrexone clustered with the PC group ([Supplementary Fig. 2b](#S1){ref-type="supplementary-material"}). In summary, the data demonstrate a shift in the bacterial community in both the gut and within the translocated population in the liver following morphine treatment, which prompted us to investigate the effects on the subsequent immune responses.
Morphine up-regulates IL-17A production during sepsis
-----------------------------------------------------
The serum, peritoneal lavage fluid and MLN were collected to measure IL-17A production following CLP. The results show that IL-17A levels in the peritoneal lavage and serum increased 6 hours post-CLP in both placebo and morphine-treated mice. In placebo-treated mice, IL-17A concentrations were reduced to baseline levels at 24 hours, whereas the morphine-treated group sustained a high level of IL-17A ([Fig. 4a,b](#f4){ref-type="fig"}). Previous studies have demonstrated two sources of IL-17A in response to infections: the T helper 17 cells and innate lymphoid cells[@b14]. To determine the type of cells that were producing IL-17A following the CLP procedure, we used anti-CD3 and anti-CD4 antibodies to separate MLN cells into two populations: CD3 + CD4 + T helper cells and CD3- non-T Cells ([Supplementary Fig. 3](#S1){ref-type="supplementary-material"}). Flow cytometric analysis showed that placebo-treated animals did not exhibit any significant up-regulation of IL-17A at 24 hours after CLP ([Fig. 4c,e](#f4){ref-type="fig"}). By contrast, a significant increase in IL-17A production was observed in the morphine-treated animals. Further characterization revealed that the major source of IL-17A in the MLN was the CD3 + CD4 + T helper cells ([Fig. 4c,d](#f4){ref-type="fig"}) but not CD3- non-T cells ([Fig. 4e,f](#f4){ref-type="fig"}).
Neutralization of IL-17A improves the survival rate and attenuates sustained inflammation in CLP mice treated with morphine
---------------------------------------------------------------------------------------------------------------------------
To investigate the role of IL-17A in sepsis progression, a neutralizing antibody to IL-17A or its isotype control IgG was administered to morphine-treated CLP mice. As shown in [Fig. 5a](#f5){ref-type="fig"}, the survival rate at 96 hours after CLP was significantly improved by anti-IL-17A, whereas none of the isotype control IgG injected mice survived. On day 7, the survival rate of mice treated with anti-IL-17A was 54.5% ([Fig. 5a](#f5){ref-type="fig"}). Additionally, mice treated with anti-IL-17A showed significantly reduced bacterial disseminations to peritoneal organs at 24 hours following CLP compared with isotype-control treated animals ([Fig. 5b--d](#f5){ref-type="fig"}), implying that neutralization of IL-17A following the CLP procedure improved gut barrier functions. It has been reported that the serum level of IL-6 is a good marker for severity during sepsis[@b15]. Thus, we determined the IL-6 levels in the serum at different time points following CLP. ELISA results showed that in placebo-treated animals, serum IL-6 peaked at 1729.6 pg/ml at 24 hours after CLP and then was reduced to baseline levels at 72 hours, whereas morphine-treated mice showed sustained high levels of IL-6 in the serum, even at 72 hours ([Fig. 5e](#f5){ref-type="fig"}). Moreover, neutralization of IL-17A significantly decreased the IL-6 serum levels in the morphine-treated animals at 72 hours after CLP ([Fig. 5f](#f5){ref-type="fig"}), validating the pro-inflammatory role of IL-17A during sepsis progression.
High levels of IL-17A compromise gut epithelial barrier function and increase gut permeability
----------------------------------------------------------------------------------------------
Our results show that morphine induced significant IL-17A up-regulation in MLN CD4 + T cells, and because 80 ± 17.4% of the IL-17 + cells in the MLN express α4β7 integrin ([Supplementary Fig. 4](#S1){ref-type="supplementary-material"}), a gut homing molecule that mediates lymphocyte recruitment to the intestines, we hypothesize that the IL-17A + T cells in the MLN migrate to intestinal tissues and modulate gut epithelial barrier function in morphine-treated CLP animals. To determine the gut permeability during sepsis progression, we gavaged mice with FITC labeled dextran and tracked the diffusion of the dextran. As shown in [Fig. 6a](#f6){ref-type="fig"}, morphine treatment resulted in an increase in FITC-dextran diffusion across the gut epithelium, indicating that morphine increased gut permeability during sepsis. In morphine-treated CLP animals, FITC intensities were significantly higher in both the peritoneal lavage fluid and blood compared with placebo-treated animals, further validating the effects of morphine on gut permeability ([Fig. 6b](#f6){ref-type="fig"}). In morphine-treated CLP animals that were injected with anti-IL-17A antibodies, we observed a decrease in FITC-dextran diffusion into the peritoneal cavity and blood ([Fig. 6c,d](#f6){ref-type="fig"}), demonstrating that neutralization of IL-17A restored gut barrier function in morphine-treated CLP animals. To investigate the morphology of the intestinal epithelium in CLP animals, the small intestine and colon were excised, fixed and stained with hematoxylin and eosin (H&E). Histological analysis indicated severe epithelial injury in the small intestinal villi of morphine-treated CLP mice compared with the appearance of continuous epithelial cells lining gut mucosal surfaces in placebo-treated mice ([Fig. 6e](#f6){ref-type="fig"}). Interestingly, the morphology of the colonic epithelium was not affected by morphine treatment ([Supplementary Fig. 5](#S1){ref-type="supplementary-material"}), suggesting differential sensitivities of colonic and small intestinal epithelial cells to inflammatory stimulation. Neutralization of IL-17A also protected the epithelial structure of the small intestines in morphine-treated mice following CLP ([Fig. 6f](#f6){ref-type="fig"}). To determine the direct effects of IL-17A on small intestinal epithelial cells, we determined the barrier function of IEC-6 cells by electrical cell impedance sensing (ECIS) arrays. The IEC-6 cell monolayers were treated with different concentrations of IL-17A ([Fig. 6g](#f6){ref-type="fig"}). The TER values were significantly reduced following IL-17A treatment, and the duration of the effects of IL-17A persisted for a longer time with increasing concentrations. We further determined the barrier function of IEC-6 cells using a trans-well system. The tans-well assay showed that both apical and basolateral stimulation with IL-17A increased the permeability of the IEC-6 monolayer ([Fig. 6h](#f6){ref-type="fig"}). We next investigated the organization of tight junction proteins, which play an important role in modulation of the epithelial barrier function between the epithelial cells. Staining of the tight junction protein zona occludens 1 (ZO-1) in IEC-6 cells also indicated the disruptive effects of IL-17A on ZO-1 organization ([Fig. 6i](#f6){ref-type="fig"}). ZO-1 (green) co-localized with F-actin (red) on the apical side of the membrane in vehicle-treated cells, and its organization was disrupted following IL-17A stimulation. Consistent with the animal studies, we showed that the morphology of the small intestinal villi in human patients with opioid addiction was dramatically different than that of healthy controls. The structures of the small intestinal epithelium were even more severely disrupted in opioid-using septic patients, which again was consistent with our observations in mice ([Fig. 6j](#f6){ref-type="fig"}).
Gram-positive bacteria stimulates the MLN to produce IL-17A in a TLR2-dependent manner
--------------------------------------------------------------------------------------
To investigate the mechanism by which morphine modulates IL-17A production in CLP animals, we cultured the immune cells from the MLN *in vitro* and stimulated the cells with either a Gram-negative bacteria species (*E coli*) or the Gram-positive species that translocated into the MLN following morphine treatment, which was a mixture of Enterococcus and Staphylococcus ([Fig. 3a](#f3){ref-type="fig"}). The results showed that *E. coli* induced significant up-regulation of IL-6, whereas the mixture of *Enterococcus* and *Staphylococcus* induced significantly higher levels of IL-17A production ([Fig. 7a,b](#f7){ref-type="fig"}). Interestingly, morphine treatment alone did not show any direct effects on IL-17A production by MLN cells ([Fig. 7c](#f7){ref-type="fig"}). TLR2 has been shown to play an important role in recognizing cell wall components from Gram-positive bacteria and initiating the immune responses to pathogen stimulation. We measured IL-17A production in the serum, peritoneal lavage fluid, and MLN in TLR2 knockout (TLR2KO) mice following a CLP procedure. No IL-17A induction was observed in placebo or morphine-treated CLP TLR2KO animals, strongly implying that TLR2 plays a role in IL-17A responses in CLP-induced sepsis ([Fig. 7d--f](#f7){ref-type="fig"}).
To understand the different roles of specific immune cells in the IL-17A response, we separated the adherent and non-adherent cells, as described previously[@b16], and separately stimulated the cells with a Gram-positive bacterial lysate. No IL-17A induction was observed in the adherent or non-adherent cells ([Fig. 7g](#f7){ref-type="fig"}), suggesting that antigen presenting cells are required for IL-17A responses. Because dendritic cells have been shown to initiate IL-17 responses[@b17], we isolated non-adherent cells from the MLNs of WT or TLR2KO mice and co-cultured them with dendritic cells purified from the blood of WT or TLR2KO mice. Flow cytometric analysis indicated that approximately 80% of non-adherent cells in the MLN were CD3 + T cells ([Supplementary Fig. 6](#S1){ref-type="supplementary-material"}). In response to bacterial stimulation, WT and TLR2KO T cells, which were co-cultured with WT dendritic cells, could produce high levels of IL-17A. Conversely, when TLR2KO T cells and WT T cells were co-cultured with TLR2KO dendritic cells, no significant IL-17A induction was observed ([Fig. 7h](#f7){ref-type="fig"}), indicating that TLR2 expression on dendritic cells was an essential requirement for IL-17A production. Interestingly, the deficiency of TLR2s in T cells significantly attenuated IL-17A production ([Fig. 7h](#f7){ref-type="fig"}), implying that TLR2s on T cells are also involved in the maximal IL-17A response.
IL-1β and IL-23 promote IL-17A production by MLN Cells
------------------------------------------------------
Previous studies have shown that IL-17A production is mediated by IL-6, TGF-β, IL-1β or IL-23 in different cell types[@b18][@b19][@b20][@b21]. To determine the role of each cytokine in IL-17A production in the MLN, we measured the concentrations of IL-6, TGF-β, IL-1β, and IL-23 in MLN adherent cell supernatant at different time points following bacterial stimulation. We observed significant up-regulation of IL-6, IL-1β, and IL-23 at 24 hours, which trended towards reduction after 48 hours. By contrast, no TGF-β up-regulation was observed until 48 hours after bacterial stimulation ([Fig. 8a--d](#f8){ref-type="fig"}). We next determined the role of these cytokines in IL-17A production using cytokine neutralization antibodies (5 μg/ml) ([Fig. 8e--i](#f8){ref-type="fig"}). As shown in [Fig. 8e,f](#f8){ref-type="fig"}, neutralization of IL-6 or TGF-β did not affect IL-17A production. By contrast, neutralizing IL-1β and IL-23 inhibited IL-17A production with IL-1β showing a more dominant role in IL-17A modulation ([Fig. 8g--i](#f8){ref-type="fig"}). Additionally, IL-1β alone was sufficient to induce IL-17A production, and the mixture of IL-1β with IL-23 showed synergistic effects on IL-17A induction in MLN cells ([Fig. 8j](#f8){ref-type="fig"}), further validating the roles of IL-1β and IL-23 in IL-17A responses.
Discussion
==========
Many factors are independently associated with a greater risk of death attributed to sepsis, including co-morbid cancer, heart failure, and immunosuppression[@b1]. Several studies have indicated that morphine induces spontaneous sepsis in mice[@b8] and that the opioid antagonist naltrexone blocks acute septic shock[@b22]. Although opioids are the mainstay for pain management in the ICU, their role in sepsis progression has not been investigated. In the present study, we demonstrated that two commonly prescribed opioids, morphine and methadone, significantly increased mortality rates in a murine model of poly-microbial sepsis, suggesting that the therapeutic window of opioids for pain management might be narrower in septic ICU patients. Clinical and laboratory studies provide strong evidence that opioids modulate immune function[@b5][@b8][@b23][@b24][@b25][@b26], which is consistent with our current observation demonstrating decreased neutrophil recruitment into the peritoneal cavity in the early stages of sepsis ([Supplementary Fig. 7](#S1){ref-type="supplementary-material"}) in morphine-treated animals. This observation implies that the defensive functions of immune cells for pathogen clearance are compromised in septic animals that are treated with morphine, resulting in a sustained higher bacterial load in different organs and the circulatory system. An analysis of gut microbial composition revealed an overgrowth of *Staphylococcus* and *Enterococcus* in the gut lumen following morphine treatment. Interactions between gut microbiota and the intestinal epithelial surface play important roles in the prevention of pathogenic bacterial outgrowth and maintaining gastrointestinal homeostasis[@b27]. Changes in composition or density of the microbiota may lead to higher susceptibility to a variety of pathogens and abnormal mucosal immune responses[@b28]. In our sepsis model, the overgrowth of *Staphylococcus* and *Enterococcus* in the gut lumen following morphine treatment resulted in Gram-positive bacterial dissemination. TLR2 activation by the disseminated Gram-positive bacteria induced the overexpression of the pro-inflammatory cytokine IL-17A. Enhanced IL-17A associated with morphine treatment in our sepsis model prompted us to investigate the role of IL-17A in sepsis progression.
IL-17A is a pro-inflammatory cytokine involved in the initiation and maintenance of several autoimmune disorders such as encephalomyelitis and inflammatory bowel disease[@b17][@b29]. One of the most important mechanistic attributes of IL-17A in this context is its modulation of the tight junction protein organization in endothelial cells, which thereby promotes blood-brain barrier disruption and central nervous system inflammation[@b30][@b31]. Increased IL-17 production has been shown to be associated with compromised gastro-intestinal integrity in pigtail macaques[@b32]. Our studies similarly show that overexpression of IL-17A during sepsis induces gut epithelial barrier dysfunction by disrupting the organization of tight junction proteins in gut epithelial cells. The role of IL-17A in these defects is supported by the observation that neutralization of IL-17A controlled the inflammation and improved survival following the CLP procedure, which was consistent with Flierl's observation that IL-17A promoted inflammation and produced adverse functions in adult mice with severe sepsis[@b33]. However, contrasting studies show that IL-17 receptor signaling is required to control sepsis[@b34][@b35]. This inconsistency may be explained by the differences in the experimental approaches used: Deshmukh and Freitas used IL17R knockout mice or neutralized IL-17A in neonatal mice. The lack of IL-17 signaling in infant mice might result in compromised development of neutrophils and other immune cells, which may have contributed to the increased host susceptibility to infections.
The major source of IL-17A in our study was CD4 + helper T (TH) cells. To date, scientists have identified two different populations of TH~17~ cells: the inducible TH~17~ (iTH~17~) cells that arise from naive CD4 + T cells in response to antigen and cytokine stimulation and the natural TH~17~ (nTH~1~) cells that acquire the capacity to produce IL-17 during development in the thymus[@b18]. The nTH~17~ cells are poised to rapidly produce IL-17 upon stimulation without further differentiation in the peripheral tissues[@b36] and show greater recruitment to Peyer's patches and the lamina propria in the gut[@b37]. The presence of nTH~17~ cells may explain the quick IL-17A responses to bacterial stimulation in our study. Another novel observation in this study was that IL-17A production by the MLN following Gram-positive bacterial stimulation was independent of IL-6 and TGF-β, the key cytokines driving TH~17~ differentiation in previous publications[@b11][@b20][@b38]. In our sepsis model, dendritic cells expressing TLR2 were required for a robust IL-17A response, which was mediated by IL-1β and IL-23. Recent evidence suggests that the CD4 + T cells differentiate into more regulated classical TH~17~ cells in the presence of TGF-β. By contrast, the TGF-β independent pathway will generate an alternative cohort of TH~17~ cells, which are responsible for the excessive inflammation in many autoimmune disorders[@b17]. Our results corroborate this theory, and we show that IL-17A induction is not dependent on IL-6 or TGF-β; it instead requires IL-1β and IL-23 to produce pro-inflammatory IL-17A. The induction of pro-inflammatory IL-17A contributes to gut epithelial damage and compromised barrier function, leading to sustained bacterial dissemination and high levels of inflammation. γ/δT cells are also important sources of IL-17A, and these cells are particularly enriched at epithelial surfaces. Previous studies have shown that both lung γ/δ T cells and gut γ/δ intraepithelial lymphocytes can produce IL-17A, which plays an important role in the modulation of mucosal immunity[@b39][@b40]. In this study, the major cell type that produced IL-17A in the MLN was CD4 + T cells and only few γ/δT cells were detected in the MLN by flow cytometric analysis ([Supplementary Fig. 8](#S1){ref-type="supplementary-material"}). In addition, we also observed that TLR2s on T cells were involved in the IL-17A response, which was consistent with Joseph Reynolds's study: TLR2 agonists activated TLR2 signaling in CD4 + T cells and led to more robust proliferation and TH~17~ cytokine production, resulting in more severe pathology in autoimmune diseases such as EAE[@b19]. The present study lends further support to the pro-inflammatory role of IL-17A in the progression of sepsis. The advent of antibiotic resistance in various pathogens is a growing concern in clinics, and researchers have renewed their focus on developing novel anti-cytokine strategies for controlling infection-mediated hyper-inflammation rather than relying on antibiotics alone. However, no current large-scale trials of anti-cytokine molecules in the treatment of sepsis have achieved satisfying efficacy[@b41]. Therefore, our study may provide novel therapeutic targets to develop anti-cytokine strategies to control specific Gram-positive infections, especially in the context of opioid pain management.
To summarize, we report that opioid treatment in the context of sepsis results in the enrichment of the Gram-positive bacteria *Staphylococcus* and *Enterococcus* in the gut lumen and promotes dissemination. Disseminated Gram-positive bacteria induce IL-17A overexpression in a TLR2-dependent manner. Overexpression of IL-17A disrupts the gut barrier function and contributes to continued bacterial dissemination and sustained systemic inflammation, leading to a higher mortality rate. This study suggests that opioid administration will increase the risk of mortality associated with sepsis in ICU patients, and neutralization of IL-17A might be a novel strategy to control excess inflammation and improve survival in septic ICU patients who are receiving opioids for pain management ([Supplementary Fig. 9](#S1){ref-type="supplementary-material"}).
Methods
=======
Experimental animals and cell lines
-----------------------------------
Male C57BL/6 and TLR2KO mice were purchased from Jackson Laboratories (Bar Harbor, Maine, USA). The animal studies were approved by the Institutional Animal Care and Use Committee at the University of Minnesota (Protocol No. 1203A11091). All procedures were conducted in line with the guidelines set forth by the National Institutes of Health Guide for the Care and Use of Laboratory Animals.
Typically, 8--10 week old animals were used for our studies. IEC-6 cell lines were obtained from American Type Culture Collection and cultured as recommended by the supplier. IEC-6 cells are rodent small intestinal epithelial cell lines that have been used to study the intestinal barrier and integrity in previous publications[@b42].
Induction of sepsis using a CLP model and opioid treatment
----------------------------------------------------------
Poly-microbial sepsis was induced as previously described[@b43]. Mice were anesthetized with 3% isoflurane, and a 1-cm midline incision was made on the anterior abdomen. The cecum was exposed and ligated, and the distance from the distal end of the cecum to the ligation point was approximately 1 cm. A double puncture was made with a 22-gauge needle to induce sepsis. The cecum was squeezed to allow the cecum contents to be expressed through the punctures. The cecum was then placed back in the abdominal cavity, and the peritoneal wall and skin incision were closed. All animals received 1 ml of saline by subcutaneous injection immediately after the surgery. Sham-operated animals (controls) underwent an identical laparotomy but without cecum ligation or puncture. The survival rates of animals were recorded every 12 hours until 7 days after the surgery.
For morphine treatment, a small incision was made at the dorsal torso of the mice. The placebo or a 25 mg slow release morphine pellet was inserted into the small pocket created during the incision, and the wound was closed using stainless steel wound-clips. Blood levels of morphine were approximately 718.42 ng/ml at 72 hours following pellet implantation ([Supplementary Fig. 10](#S1){ref-type="supplementary-material"}), which were within the range observed in patients on opioids for moderate to severe pain (11 ng/ml to 1440 ng/ml)[@b44]. For methadone treatment, saline or 15 mg/kg of methadone was given by intraperitoneal injection. Methadone doses used in patients range from 50 to 150 mg/day[@b45], which equates mouse doses of 10.28 mg/kg to 30.83 mg/kg according to the distinct Km factors of mice and humans[@b46]. Therefore, the dose of methadone used in our study was also within the physiological range observed in clinics.
For IL-17A neutralization, 100 μg of LEAF^TM^ anti-mouse IL-17A antibodies (BioLegend) were administered to mice by intraperitoneal injection every other day.
Bacterial counts in the blood, peritoneal lavage fluid, MLN, liver and spleen
-----------------------------------------------------------------------------
The bacterial counts were determined 6, 24, 72, and 168 hours after CLP surgery in the peritoneal lavage fluid, MLN, liver, spleen and blood. The peritoneal lavage fluid, blood, and homogenates of the liver, MLN, and spleen were cultured on a blood agar plate overnight. Bacterial colonies were quantified and described as colony forming units (CFU).
Cytokine measurements
---------------------
The cytokine levels were detected in the peritoneal lavage fluid, serum, and supernatants of MLN cells. The following mouse ELISA kits were used: IL-1β, IL-6, IL-23, IL-17A, and TGF- β (eBiosciences). All experiments were performed according to the manufacturer's protocol.
Flow cytometry analysis
-----------------------
The MLN was homogenized, and single cell suspensions were incubated with anti-CD3, anti-CD4, anti-IL-17A and anti-α4β7 antibodies following the manufacturer's intracellular staining protocol (eBiosciences).
Intestinal permeability
-----------------------
The animals were gavaged with 4 kD FITC-dextran (500 mg/kg body weight in a 50 mg/ml concentration) using a 4-cm long, curved needle with a plastic ball at the tip. The images of animals were analyzed with the Xenogen Spectrum system. After sacrifice, the blood and peritoneal lavage fluid were collected, and the intensity of FITC was determined with a fluorometer using an excitation wavelength of 488 nm and detecting the emission at 520 nm.
Dendritic cell isolation
------------------------
Dendritic cells from the blood of WT or TLR2KO mice were purified by a magnetic separation kit (Miltenyi Biotec) according to the manufacturer's instructions.
Measurement of trans-epithelial resistance
------------------------------------------
An ECIS 1600R (Applied BioPhysics, Troy, NY) was used to measure the trans-epithelial resistance (TER) of epithelial monolayers as described previously[@b47]. IEC-6 cells were seeded in the wells of the electrode array and grown to confluence. Then, the medium was changed, and the baseline TER was measured for 60 min to equilibrate the monolayers. Afterward, 400 ml of a medium containing different concentrations of IL-17A was applied to each well. The baseline TER of each experiment was normalized to 1.0 to enable comparisons and statistical analysis of TER changes over time following different treatments.
Transwell assays
----------------
A transwell assay was performed as described previously[@b48]. IEC-6 cells were cultured in the top chamber of a transwell system, and 5 mg/ml 4 kD FITC-dextran was added to the top chamber after IL-17A (100 ng/ml) stimulation. The medium in the lower chamber was collected after 6 hours. The amounts of FITC-dextran in the lower chambers were determined with a fluorometer using an excitation wavelength of 488 nm and detecting emission at 520 nm.
Immunofluorescence
------------------
IEC-6 cells in chamber slides were fixed with 1% paraformaldehyde in PBS for 10 min at room temperature. After washing in PBS and blocking nonspecific binding sites with 5% bovine serum albumin (BSA), the slides were incubated with polyclonal rabbit anti-ZO-1 (5 mg/ml, Invitrogen) in PBS with 5% bovine serum albumin (BSA) for 120 min at room temperature. After washing, the slides were incubated with rhodamine phalloidin (Invitrogen) and DyLightTM 488-conjugated AffiniPure Donkey anti-rabbit IgG (0.075 mg/ml, Jackson Lab, WestGrove, PA) for 60 min. The slides were then washed and mounted with coverslips using ProLong Gold anti-fade reagent with DAPI (Invitrogen). The slides were imaged with a confocal microscope (Nikon).
Gut and liver microbiome analysis
---------------------------------
The fecal content was collected from the gut region encompassing the distal cecum and approximately one inch of the colon. The gut tissue was washed and stored separately. The fecal matter was lysed using glass beads in MagnaLyser tissue disruptor (Roche), and the total DNA was isolated using a Power-soil/fecal DNA isolation kit (Mo-Bio) as per the manufacturer's specifications. All samples were quantified via the Qubit® Quant-iT dsDNA Broad-Range Kit (Invitrogen, Life Technologies, Grand Island, NY), to ensure that they met the minimum concentration and mass for DNA, and submitted to Second genome Inc. for microbiome analysis as follows. To enrich the sample for the bacterial 16 S V4 rDNA region, DNA was amplified using fusion primers designed against the surrounding conserved regions, which are tailed with sequences to incorporate Illumina (San Diego, CA) flow cell adapters and indexing barcodes. Each sample was PCR amplified with two differently bar coded V4 fusion primers and was submitted for pooling and sequencing. For each sample, amplified products were concentrated using a solid-phase reversible immobilization method for the purification of PCR products and quantified by electrophoresis using an Agilent 2100 Bioanalyzer®. The pooled 16 S V4 enriched, amplified, barcoded samples were loaded into the MiSeq® reagent cartridge and then onto the instrument along with the flow cell. After cluster formation on the MiSeq instrument, the amplicons were sequenced for 250 cycles with custom primers designed for paired-end sequencing. Using QIIME and custom scripts, the sequences were quality filtered and demultiplexed using exact matches to the supplied DNA barcodes. The resulting sequences were then searched against the Greengenes reference database of 16 S sequences and clustered at 97% by uclust (closed-reference OTU picking). The longest sequence formed from each Operation Taxonomic Unit (OTU) was then used as the OTU representative sequence and assigned a taxonomic classification via the MOTHUR Bayesian classifier, which was trained against the Greengenes database clustered at 98%.
Human samples and histology
---------------------------
Young adult (20--40 years old; de-identified) human tissue slides were commercially sourced from both the National Disease Resource Interchange as well as Bionet histology resources (University of Minnesota). Sections from patients with a history of intravenous use of heroin in addition to cocaine were compared with control sections. In total, 10 human samples were examined, four from healthy control, four from chronic opioid users and 2 from opioid abusers that have a diagnosis of sepsis. H&E staining was performed by the Comparative Pathology Shared Resource (CPSR) at the University of Minnesota, and slides were imaged using a Leica DM5500 B microscope. Representative images are shown. The Institutional Review Board (IRB) of the University of Minnesota determined that "This project does not meet the regulatory definition of human subject research. No further IRB review is required" under exemption 4.
Statistical analysis
--------------------
The data (except for the survival curves and scatter plots) are reported as the means ± SEMs of values of triplicate results. The means of different treatments were compared by Student's t test or ANOVA followed by Bonferroni's t test (GraphPad Prism Software). Bacterial counts were reported as the means of CFUs and were analyzed by the Mann-Whitney U test (GraphPad Prism Software). The survival rate was expressed as the percentage of live animals, and the Mantel-Cox log rank test was used to determine the differences between survival curves (GraphPad Prism Software version 3). A p value of 0.05 or less was considered significant. For the gut microbiome analysis, the Adonis test was used to determine significant whole microbiome differences among discrete categorical or continuous variables. In this randomization/Monte Carlo permutation test, the samples were randomly reassigned to the various sample categories, and the mean normalized cross-category differences from each permutation were compared to the true cross-category differences. The fraction of permutations with greater distinction among categories (larger cross-category differences) than those observed with the non-permuted data were reported as the p-value for the Adonis test.
Additional Information
======================
**How to cite this article**: Meng, J. *et al.* Opioid Exacerbation of Gram-positive sepsis, induced by Gut Microbial Modulation, is Rescued by IL-17A Neutralization. *Sci. Rep.* **5**, 10918; doi: 10.1038/srep10918 (2015).
Supplementary Material {#S1}
======================
###### Supplementary Information
This work was supported in part by the NIH grants RO1 DA 12104, RO1 DA 022935, RO1 DA031202, K05DA033881, P50 DA 011806 and 1R01DA034582 to S. Roy.
The authors declare no competing financial interests.
**Author Contributions** S.R. and J.Meng designed the study. J.Meng, S.B., G.S. and F.W. analyzed and interpret the results. J.Meng, S.B., G.S., D.L. and J.Ma preformed the laboratory experiments. S.R., J.Meng, and S.B. wrote the manuscript. All authors reviewed the manuscript.
{#f1}
{#f2}
{#f3}
{#f4}
{#f5}
{#f6}
{#f7}
{#f8}
| {
"pile_set_name": "PubMed Central"
} |
Introduction {#Sec1}
============
Atherosclerosis is generally considered to be positively associated with hypertension^[@CR1]^. A single nucleotide polymorphism (SNP) (rs3782886) in breast cancer suppressor BRCA1**-**related associated protein (BRAP) has been associated with the risk of developing myocardial infarction^[@CR2]^. Another study reported that BRAP activates inflammatory cascades, increasing the risk of carotid atherosclerosis^[@CR3]^. Even though the minor allele frequency (MAF) of rs3782886, and aldehyde dehydrogenase 2 (ALH2) polymorphism rs671 are significantly inversely associated with both systolic and diastolic blood pressure^[@CR4]^.
Since strong linkage disequilibrium (LD) values between rs3782886 and rs671 have been reported^[@CR5]^, rs671 could have same the characteristics of rs3782886. However, the mechanism underlying the association between hypertension and these genetic factors (rs3782886 and rs671) have not yet been elucidated.
Atherosclerosis is involved in aggressive endothelial repair. Therefore, the favorable association of genetics factor (rs3782886 and rs671) with hypertension may be caused by stimulating endothelial repair.
Additionally, several studies have demonstrated a close association between bone marrow activity (hematopoietic activity) and endothelial maintenance^[@CR6],\ [@CR7]^. We have showed in a previous study that active endothelial repair, represented by high levels of hematopoietic (CD34-positive) stem cells, is positively associated with atherosclerosis but could have beneficial effect on prevention of hypertension among elderly participants^[@CR8]^.
Hematopoietic bone marrow activity declines with age^[@CR9]^ and aging is also a well-known cause of endothelial injury^[@CR10],\ [@CR11]^. Since hematopoietic activity also influences reticulocyte levels, high levels of reticulocytes among older participants may indicate a high capacity of endothelial maintenance. Therefore, the levels of reticulocytes could play a role in the association between genetic factors (rs3782886 and rs671) and hypertension.
Furthermore, platelet activation is involved in the initial mechanism of an endothelial repair^[@CR12],\ [@CR13]^ and a platelet count indicates the activity of an endothelial repair^[@CR1],\ [@CR14]^. Therefore, evaluating the number of platelets may be an efficient tool to determine the mechanism underlying the association between genetic factors (rs3782886 and rs671) and hypertension.
An extensive prevalence of SNPs should have the same beneficial effect on the participants' daily activities, rather than impose a disadvantage. Therefore, determining the mechanism by which genetic characteristics prevent hypertension inducing progression of atherosclerosis could provide novel knowledge to elaborate strategies for risk estimation and prevention of hypertension.
To determine the potential mechanism underlying the association between the genetic factors (rs3782886 and rs671) and hypertension, we conducted a cross-sectional study comprised of 1,313 elderly Japanese individuals (aged 60--98) who had previously participated in a general health check-up, in 2017.
Methods {#Sec2}
=======
Study population {#Sec3}
----------------
The methods that relates to present risk survey including genetic factor also have been described elsewhere^[@CR1],\ [@CR14]--[@CR16]^.
The study population was comprised of 1,401 Japanese residents (504 men and 897 women) aged between 60 and 99 years old from Goto City (western Japan) who had undergone an annual medical check-up in 2017.
Participants without platelet data (n = 2) or SNP data (n = 1) were excluded from the study population. Additionally, to control for the influence of chronic disease, participants with hypo-nutrition (BMI ≤ 18.0 kg/m^2^) (n = 85) were also excluded. The remaining participants, comprising 1,313 (494 men and 819 women) with a mean age of 72.9 (standard deviation (SD): 7.3; range: 60--98), were enrolled in the study.
All of the procedures involving human participants in this study were performed in accordance with the ethical standards of the institution research committee and the 1964 Helsinki Declaration, and its later amendments for comparable ethical standards. Written consent forms in Japanese were made available to ensure the comprehensive understanding of the study objectives, and informed consent was provided by the all participants. Ethics Committee of Nagasaki University Graduate School of Biomedical Sciences (project registration number: 14051404) approved this study.
Data collection and laboratory measurements {#Sec4}
-------------------------------------------
Specially trained interviewers were tasked with obtaining the medical histories and habitual statuses of the participants. The body weight and height were measured using an automatic body composition analyzer (BF-220; Tanita, Tokyo, Japan), from which the body mass index (BMI; kg/m^2^) was calculated. Systolic (SBP) and diastolic blood pressure (DBP) were recorded at rest. Hypertension was defined as systolic blood pressure ≥ 140 mmHg and/or diastolic blood pressure ≥ 90 mmHg and/or taking anti-hypertensive medication.
Fasting blood samples were collected in a heparin sodium tube, an EDTA-2K tube, and a siliconized tube.
The quantities of platelets, red blood cells (RBC), and reticulocytes in the samples from the EDTA-2K tube were measured using an automated procedure at SRL, Inc. The levels of reticulocytes were determined using the following formula: reticulocytes (×10^4^ cells/μL) = (reticulocytes, ‰) × RBC (×10^4^ cells/μL)/1,000 (for men: 5.37 ± 2.01 \[×10^4^ cells/μL\]; for women: 4.78 ± 1.56 \[×10^4^ cells/μL\]). The concentrations of triglyceride (TG), HDL-cholesterol (HDL-C), hemoglobin A1c (HbA1c), and creatinine were measured using the standard laboratory procedures. All measurements were performed at SRL, Inc. (Tokyo, Japan). The glomerular filtration rate (GFR) was estimated using a recently adapted established method introduced by a working group of the Japanese Chronic Kidney Disease Initiative^[@CR17]^, which yielded an estimate of GFR (mL/min/1.73 m^2^) = 194 × (serum creatinine (enzyme method))^−1.094^ × (age)^−0.287^ × (0.739 for women).
The genomic DNA was extracted from 2 mL of whole peripheral blood using the Gene Prep Star NA-480 (Kurabo Industries Ltd., Osaka, Japan), which was genotyped for SNP rs3782886 (BRAP on chromosome 12q24.12) and rs671 (ALDH2 on chromosome 12q24.2) using the TaqMan method with a LightCycler 480 (Roche Diagnostics, Basel, Switzerland). Expected numbers and percentages were calculated. Deviations from the Hardy--Weinberg equilibrium were evaluated via χ^2^ test for each allele (rs3782886 and rs671).
Statistical analysis {#Sec5}
--------------------
The study population was classified by reticulocyte levels (high and low) according to the median values (5.21 × 10^4^ cells/μL for men and 4.65 × 10^4^ cells/μL for women).
The characteristics of the study population in relation to the reticulocyte levels were expressed as the mean ± standard deviation (SD) for the contentious variables except for TG, and as prevalence for the medication status and habitual status. Since TG showed a skewed distribution, the characteristics of the study population were expressed as median \[first quartile and third quartile\], followed by a logarithmic transformation. Differences between mean values or proportional values of monitored characteristics were analyzed in relation to reticulocyte levels. Significant differences were evaluated using t-test for continuous variables and χ^2^ test for proportional data.
Logistic regression models were used to calculate the odds ratios (OR) and 95% confidence intervals (CI) were used to determine associations between hypertension and platelets, as well as between hypertension and reticulocyte levels.
The numbers of platelets and reticulocytes in relation to the genotype (rs3782886 and rs671) by reticulocyte levels were also calculated.
In addition, OR and 95% CI were also calculated using a logistic regression model to determine the influence of SNP (rs3782886 and rs671) on hypertension by the levels of reticulocytes.
Two different approaches were used to make adjustments for confounding factors. First, one model was adjusted only for sex and age (Model 1). For the second model (Model 2), we included several other potential confounding factors, namely BMI (kg/m^2^), alcohol consumption (never drinker, former drinker, current drinker \[23--45 g/week, 46--68 g/week, ≥ 69 g/week\]), smoking status (never, former, current), TG (mg/dL), HDL-C (mg/dL), HbA1c (%), and GFR (mL/min/1.73 m^2^).
As Japanese men are known to have high rates of drinking, while Japanese women have low rates of drinking^[@CR18]^, to evaluate the influence of SNP (rs3782886 and rs671) on the status of individual who had never had a drink (never drinker), sex-specific analyses were performed.
Furthermore, to evaluate the influence of drinking status (never drinker) on platelet and reticulocyte count, sex-adjusted value of platelet and reticulocyte count in relation to drinking status stratified by reticulocyte levels were calculated by using covariance analysis.
All statistical analyses were performed using the SAS system for Windows (version 9.4; SAS Inc., Cary, NC). P-values \< 0.05 were regarded as statistically significant.
Results {#Sec6}
=======
Characteristics of study population by reticulocyte levels {#Sec7}
----------------------------------------------------------
Characteristics of study population by reticulocyte levels are shown in Table [1](#Tab1){ref-type="table"}.Table 1Characteristics of study population by reticulocyte levels.Reticulocyte levels*p*LowHighNo. of participants656657Men, %37.737.60.983Age, years73.7 ± 7.772.2 ± 6.7 \< 0.001SBP, mmHg134 ± 17135 ± 170.075DBP, mmHg73 ± 1075 ± 11 \< 0.001Anti-hypertensive medication, %39.647.50.004BMI, kg/m^2^22.2 ± 2.623.8 ± 3.1 \< 0.001Current drinker, %31.939.60.004Current smoker, %7.08.10.470TG, mg/dL85 \[65, 113\]^a^102 \[74, 139\]^a^ \< 0.001^b^HDL-C, mg/dL62 ± 1561 ± 150.070HbA1c, %5.7 ± 0.55.8 ± 0.60.035GFR, mL/min/1.73m^2^68.5 ± 14.368.7 ± 14.00.831Plt, × 10^4^ /μL22.5 ± 5.122.7 ± 4.90.453Values: mean ± standard deviation.SBP: Systolic blood pressure. DBP: Diastolic blood pressure. BMI: Body mass index. TG: Triglycerides. HDL-C: HDL-cholesterol. GFR: Glomerular filtration rate. Plt: Platelets.^a^Values are median \[first quartile, third quartile\].^b^Logarithmic transformation was used for evaluating p.
Compared to participants with low reticulocyte levels, the participants with high reticulocyte levels showed significantly higher values of DBP, anti-hypertensive medication use, BMI, current drinker, TG, and HbA1C, but a significantly lower age.
Association between hypertension and platelet among total participants {#Sec8}
----------------------------------------------------------------------
Platelets play an important role in endothelial repair^[@CR13]^. Moreover, platelet count indicates endothelial repair activity^[@CR14]^. As such, we evaluated the association between hypertension and platelet count for the total of the participants (Table [2](#Tab2){ref-type="table"}). Independent from known cardiovascular risk factors, platelets were significantly positively associated with hypertension.Table 2Odds ratios (ORs) and 95% confidence intervals (CIs) for hypertension in relation to platelets and reticulocytes among total participants.Platelet tertiles*p* for trend1 SD increment of plateletsT1 (low)T2T3 (high)No. of participants441432440No. of cases (%)244 (55.3)270 (62.5)279 (63.4)Model 11.001.47 (1.11, 1.94)1.61 (1.22, 2.14) \< 0.0011.23 (1.10, 1.39)Model 21.001.38 (1.02, 1.85)1.56 (1.16, 2.09)0.0041.22 (1.08, 1.38)Reticulocyte tertiles*P* for trend1 SD increment of reticulocytesT1 (low)T2T3 (high)No. of participants437438438No. of cases (%)239 (54.7)264 (60.3)290 (66.2)Model 11.001.31 (0.99, 1.73)1.88 (1.42, 2.50) \< 0.0011.44 (1.26, 1.64)Model 21.001.09 (0.81, 1.45)1.21 (0.89, 1.65)0.2221.19 (1.03, 1.36)Model 1: Adjusted only for sex and age. Model 2: Adjusted further for sex and age, alcohol consumption (never drinker, former drinker, current drinker \[23--45 g/week, 46--68 g/week, ≥ 69 g/week\]) , smoking status (never smoker, former smoker, current smoker), body mass index, HDL-cholesterol, triglycerides, HbA1c, and GFR. Platelet tertiles: For men \< 19.6 × 10^4^/μL for T1, 19.6--23.5 × 10^4^/μL for T2, \> 23.5 × 10^4/^μL for T3. For women \< 20.9 × 10^4^/μL for T1, 20.9--24.6 × 10^4^/μL for T2, \> 24.6 × 10^4^/μL for T3. The 1 standard deviation (SD) increment of platelets was 5.2 × 10^4^/μL for men and 4.8 × 10^4^/μL for women. Reticulocyte tertiles: For men, \< 4.464 × 10^4^ cells/μL for T1, 4.464--5.796 × 10^4^/cells/μL for T2, \> 5.796 × 10^4^/μL for T3. For women, \< 4.0 × 10^4^ cells/μL for T1, 4.0--5.29 × 10^4^ cells/μL for T2, \> 5.29 × 10^4^ cells/μL for T3. The increments of 1 standard deviation (SD) in reticulocytes are 2.0 × 10^4^ cells/μL for men and 1.6 × 10^4^ cells/μL for women.
Association between hypertension and reticulocyte levels among total participants {#Sec9}
---------------------------------------------------------------------------------
Table [2](#Tab2){ref-type="table"} shows the association between hypertension and reticulocyte levels for the total participants. Independent from known cardiovascular risk factors, the reticulocyte levels were significantly positively associated with hypertension.
Association between hypertension and genotype of rs3782886 by reticulocyte levels {#Sec10}
---------------------------------------------------------------------------------
The OR and 95% CI of hypertension for the rs3782886 genotype based on the levels of reticulocytes are shown in Table [3](#Tab3){ref-type="table"}. Despite no significant associations being found between hypertension and MAF of rs3782886 with low levels of reticulocytes, a significantly inverse association was observed in participants with high levels of reticulocytes. We also found that the interactions between the reticulocyte levels had a significant effect on the association between rs3782886 and hypertension, with a fully adjusted p-value (Model 2) of 0.0499. Despite no significant associations being found between platelet count and MAF of rs3782886 with low levels of reticulocytes, a significantly positive association was observed with high levels of reticulocytes. In addition, for both low and high reticulocytes, no significant association between reticulocyte levels and the MAF of the rs3782886 genotype was observed.Table 3Odds ratios (OR) and 95% confidence intervals (CI) for hypertension in relation to rs3782886 genotype by reticulocyte levels.Major homo (A/A)Hetero (A/G)Minor homo (G/G)*p* for trendMAF**Low reticulocyte**No. of participants (%)380 (57.9)232 (35.4)44 (6.7) ≥ 0.05\*No. of expectations (%)375.02 (57.2)241.95 (36.9)39.02 (5.9)No. of cases (%)213 (56.1)128 (55.2)24 (54.5)Platelets, ×10^4^ /μL22.3 ± 5.222.8 ± 4.723.0 ± 5.60.480--Reticulocyte, ×10^4^ cells/μL3.7 ± 0.73.8 ± 0.73.7 ± 0.80.790--Model 11.000.98 (0.70, 1.39)0.95 (0.50, 1.81)0.8720.98 (0.76, 1.27)Model 21.001.13 (0.78, 1.65)0.96 (0.48, 1.92)0.7531.05 (0.79, 1.39)**High reticulocyte**No. of participants (%)395 (60.1)208 (31.7)54 (8.2) ≥ 0.05\*No. of expectations (%)379.00 (57.7)240.01 (36.5)38.00 (5.8)No. of cases (%)271 (68.6)129 (62.0)28 (51.9)Platelets, ×10^4^ /μL22.2 ± 4.823.5 ± 4.923.7 ± 5.00.005--Reticulocyte, ×10^4^ cells/μL6.3 ± 1.86.2 ± 1.16.1 ± 1.00.407--Model 11.000.72 (0.50, 1.03)0.49 (0.27, 0.88)0.0070.71 (0.55, 0.91)Model 21.000.71 (0.48, 1.06)0.53 (0.28, 1.02)0.0250.72 (0.55, 0.96)Model 1: adjusted only for sex and age. Model 2: Adjusted further for sex, age, alcohol consumption (never drinker, former drinker, current drinker \[23--45 g/week, 46--68 g/week, ≥ 69 g/week\]) , smoking status (never smoker, former smoker, current smoker), body mass index, HDL-cholesterol, triglycerides, HbA1c, and GFR. The hypertension is defined as systolic blood pressure ≥ 140 mmHg and/or diastolic blood pressure ≥ 90 mmHg and/or taking anti-hypertensive medication. MAF: minor allele frequency.\*Hardy--Weinberg equilibrium p-values.
Association between hypertension and genotype of rs671 by reticulocyte levels {#Sec11}
-----------------------------------------------------------------------------
Those associations for rs3782886 were essentially the same as those for rs671, as shown in Table [4](#Tab4){ref-type="table"}. Despite no significant association being found between hypertension and the MAF of rs671 with low levels of reticulocytes, a significantly inverse association was observed with high levels of reticulocytes. We also found that the interactions between the reticulocyte levels had a significant effect on rs671 and hypertension, with a fully adjusted p-value (Model 2) of 0.035.Table 4Odds ratios (OR) and 95% confidence intervals (CI) for hypertension in relation to rs671 genotype by reticulocyte levels.Major homo (G/G)Hetero (G/A)Minor homo (A/A)*p* for trendMAF**Low reticulocyte**No. of participants (%)394 (60.1)224 (34.1)38 (5.8) ≥ 0.05\*No. of expectations (%)390.30 (59.5)231.40 (35.3)34.30 (5.2)No. of cases (%)219 (55.6)125 (55.8)21 (55.3)Platelets, ×10^4^ /μL22.3 ± 5.322.8 ± 4.523.0 ± 5.90.515--Reticulocyte, ×10^4^ cells/μL3.7 ± 0.73.8 ± 0.73.7 ± 0.80.876--Model 11.001.02 (0.72, 1.44)0.98 (0.49, 1.94)0.9831.00 (0.77, 1.31)Model 21.001.16 (0.80, 1.69)1.04 (0.50, 2.15)0.5881.08 (0.81, 1.45)**High reticulocyte**No. of participants (%)413 (62.9)195 (29.7)49 (7.5) ≥ 0.05\*No. of expectations (%)396.67 (60.4)227.67 (34.7)32.67 (5.0)No. of cases (%)281 (68.0)123 (63.1)24 (49.0)Platelets, ×10^4^ /μL22.2 ± 4.823.5 ± 4.924.0 ± 5.10.002--Reticulocyte, ×10^4^ cells/μL6.3 ± 1.86.1 ± 1.16.1 ± 1.00.249--Model 11.000.79 (0.55, 1.14)0.46 (0.25, 0.85)0.0120.72 (0.56, 0.93)Model 21.000.75 (0.50, 1.12)0.50 (0.25, 0.98)0.0270.72 (0.54, 0.96)Model 1: adjusted only for sex and age. Model 2: Adjusted further for sex, age, alcohol consumption (never drinker, former drinker, current drinker \[23--45 g/week, 46-68 g/week, ≥ 69 g/week\]) , smoking status (never smoker, former smoker, current smoker), body mass index, HDL-cholesterol, triglycerides, HbA1c, and GFR. The hypertension is defined as systolic blood pressure ≥ 140 mmHg and/or diastolic blood pressure ≥ 90 mmHg and/or taking anti-hypertensive medication.MAF: minor allele frequency.\*Hardy--Weinberg equilibrium p-values.
Despite no significant association being found between the platelet count and the MAF of rs671 with low levels of reticulocytes, a significantly positive association was observed for high levels of reticulocytes. In addition, for both low and high levels of reticulocytes, no significant association between reticulocytes and MAF of rs671 was observed.
Sex-specific association between never drinker and rs3782886 and rs671 genotypes {#Sec12}
--------------------------------------------------------------------------------
The sex-specific associations between MAF of SNP (rs3782886 and rs671) and never drinker are shown in Tables [5](#Tab5){ref-type="table"}, [6](#Tab6){ref-type="table"}. Both the MAF of rs3782886 and rs671 showed a significantly positive association in participants classified as never drinkers.Table 5Odds ratios (OR) and 95% confidence intervals (CI) for never drinker in relation to rs3782886 genotype.rs3782886Major homo (A/A)Hetero (A/G)Minor homo (G/G)*p* for trendMAF**Men**No. of participants (%)282 (57.1)172 (34.8)40 (8.1) \< 0.05\*No. of expectations (%)274.14 (55.5)187.70 (38.0)32.14 (6.5)No. of cases (%)34 (12.1)72 (41.9)31 (77.5)Age-adjusted1.005.28 (3.29, 8.46)26.45 (11.52, 60.71) \< 0.0015.20 (3.63, 7.45)**Women**No. of participants (%)493 (60.2)268 (32.7)58 (7.1) ≥ 0.05\*No. of expectations (%)480.01 (58.6)293.98 (35.9)45.01 (5.5)No. of cases (%)363 (73.6)238 (88.8)56 (96.6)Age-adjusted1.002.93 (1.90, 4.52)10.48 (2.51, 43.70) \< 0.0013.01 (2.06, 4.39)MAF: minor allele frequency.\*Hardy--Weinberg equilibrium p-values.Table 6Odds ratios (OR) and 95% confidence intervals (CI) for never drinker in relation to rs671 genotype.rs671Major homo (G/G)Hetero (G/A)Minor homo (A/A)*p* for trendMAF**Men**No. of participants (%)296 (59.9)164 (33.2)34 (6.9) \< 0.05\*No. of expectations (%)289.24 (58.6)177.52 (35.9)27.24 (5.5)No. of cases (%)38 (12.8)71 (43.3)28 (82.4)Age-adjusted1.005.21 (3.28, 8.27)33.09 (12.79, 85.61) \< 0.0015.47 (3.78, 7.92)**Women**No. of participants (%)511 (62.4)255 (31.1)53 (6.5) ≥ 0.05\*No. of expectations (%)497.78 (60.8)281.44 (34.4)39.78 (4.9)No. of cases (%)376 (73.6)229 (89.8)52 (98.1)Age-adjusted1.003.27 (2.07, 5.16)20.05 (2.74, 146.91) \< 0.0013.49 (2.32, 5.25)MAF: minor allele frequency.\*Hardy--Weinberg equilibrium p-values.
Sex-adjusted reticulocyte count in relation to never-drinker status {#Sec13}
-------------------------------------------------------------------
Since gender determined a statistically significant difference (p \< 0.001) in reticulocyte levels between (mean ± SD) were for men (5.37 ± 2.01 × 10^4^ cells/μL, n = 494) and women (4.78 ± 1.56 × 10^4^ cells/μL, n = 819), adjustment of data by sex was performed to evaluate the relationship between reticulocyte levels and never-drinker status. Never drinkers displayed significantly lower reticulocyte levels than non-never drinkers (*p* = 0.006), with a sex-adjusted reticulocyte count (mean ± standard error (SE)) of 4.88 ± 0.07 × 10^4^ cells/μL for never drinkers (n = 794) and 5.19 ± 0.08 × 10^4^ cells/μL for non-never drinkers (n = 519).
Sex-adjusted platelet count in relation to never-drinker status {#Sec14}
---------------------------------------------------------------
Since gender determined a statistically significant difference (p \< 0.001) in platelet levels (mean ± SD) between men (23.1 ± 4.8 × 10^4^ /μL) and women (21.9 ± 5.2 × 10^4^ /μL), adjustment of data by sex was performed to evaluate the relationship between platelet levels and never-drinker status. We did not find any significant difference in platelet levels between never and non-never drinkers (*p* = 0.059); the sex-adjusted platelet count (mean ± SE) was 22.9 ± 0.2 × 10^4^ /μL for never drinkers and 22.3 ± 0.2 × 10^4^ /μL for non-never drinkers.
Sex-adjusted platelet counts in relation to never-drinker status by reticulocyte levels {#Sec15}
---------------------------------------------------------------------------------------
Among the participants with high levels of reticulocytes, never-drinkers showed significantly higher platelet counts than those in the non-never drinkers (*p* = 0.046); the sex-adjusted platelet counts (mean ± SE) were 23.1 ± 0.3 for never drinkers (n = 369) and 22.2 ± 0.3 for non-never drinkers (n = 288). Among the participants with low reticulocyte levels, no significant association was observed (*p* = 0.444); the corresponding values were 22.7 ± 0.3 for never-drinkers (n = 425) and 22.3 ± 0.4 for non-never drinkers (n = 231).
Present study population in relation to the Hardy--Weinberg equilibrium {#Sec16}
-----------------------------------------------------------------------
An established Hardy--Weinberg equilibrium for both genetic factors (rs3782886 and rs671) was observed only in men (Tables [5](#Tab5){ref-type="table"}, [6](#Tab6){ref-type="table"}).
Discussion {#Sec17}
==========
In terms of the major findings presented in this study, the MAF of rs3782886 and rs671 were found to be significantly inversely associated with hypertension only in individuals with high levels of hematopoietic activity (high reticulocyte levels).
In addition to that the MAF of rs3782886 and rs671 were found to be positively associated with the platelet count only in individuals with high reticulocyte levels. Furthermore, the status of never-drinkers showed significantly higher platelet counts than those in non-never drinkers, only in individuals with high reticulocyte levels, whereas the MAF of rs3782886 and rs671 was significantly and positively associated with the status of a never-drinker. Since ethanol directly attenuates the activation of platelets^[@CR19]^, which plays an important role in endothelial repair^[@CR13],\ [@CR14]^, these associations provide an efficient tool to clarify the influence of genetic factors on hypertension in the Japanese population.
A previous study conducted on a Japanese found that the MAF of rs3782886 and rs671 were significantly inversely associated with both systolic and diastolic blood pressure^[@CR4]^. In the present study, we found further evidence that the MAF of these genotypes were significantly inversely associated with hypertension only in participants with high levels reticulocytes (Tables [3](#Tab3){ref-type="table"}, [4](#Tab4){ref-type="table"}, Fig. [1](#Fig1){ref-type="fig"}c). However, the background mechanism responsible for these associations has yet to be fully elucidated.Figure 1Potential mechanism underlying present results. Associations shown in red (**a**--**j**) were observed in the present study. Genetic factor: minor allele frequency of rs671 and rs3782886. \*1: Observed only among participants with high reticulocyte.
A summary of the potential mechanisms underlying our results is provided in Fig. [1](#Fig1){ref-type="fig"}. Genetic factors stimulate platelet production, which contributes to endothelial repair. Moreover, an age-related reduction in bone marrow activity, which is associated with low levels of reticulocytes, may reduce the impact of this genetic influence.
Recently, a close connection between the bone metabolism and the activity of vascular maintenance has been found^[@CR7]^. Furthermore, an age-related reduction in bone marrow activity (hematopoietic activity) is known to cause anemia in the elderly^[@CR20]^. In present study, individuals with high reticulocyte level shows significantly younger than that of individuals with low reticulocyte level (Table [1](#Tab1){ref-type="table"}, Fig. [1](#Fig1){ref-type="fig"}f). And reticulocyte is revealed to be significantly positively associated with hypertension (Table [2](#Tab2){ref-type="table"}, Fig. [1](#Fig1){ref-type="fig"}b) as like previous our study^[@CR15]^. Since hypertension injures endothelium strongly, lower levels of reticulocytes among the elderly indicate a lower capacity for vascular maintenance. Even though genetic factors have a beneficial effect on preventing hypertension by stimulating vascular maintenance activity, in the present study, participants with low levels of hematopoietic activity did not appear to benefit from this (Tables [3](#Tab3){ref-type="table"}, [4](#Tab4){ref-type="table"}, Fig. [1](#Fig1){ref-type="fig"}c). In other words, the hematopoietic activity evaluated by the reticulocyte levels may influence the association between genetic factors and hypertension.
The platelet has been reported to contribute to endothelial repair^[@CR13]^. Our previous study found that platelet levels were positively associated with hypertension^[@CR1]^ and that the platelet count could serve as an indicator of endothelial repair^[@CR14]^. In the present study, platelet count was found to be significantly positively associated with hypertension (Table [2](#Tab2){ref-type="table"}, Fig. [1](#Fig1){ref-type="fig"}a). Therefore, the platelet count may act as an indicator of endothelial repair activity, since endothelial injury caused by hypertension activates endothelial repair.
On the other hand, the results presented here only showed a significantly positive association between rs3782886 and platelet counts in participants with high levels of reticulocytes (Table [3](#Tab3){ref-type="table"}, Fig. [1](#Fig1){ref-type="fig"}d), whereas the MAF of rs3782886 was found to be significantly inversely associated hypertension in these participants (high levels of reticulocytes) (Table [3](#Tab3){ref-type="table"}, Fig. [1](#Fig1){ref-type="fig"}c). These results seem ambivalent associations since we also found significant positive associations between platelet count and hypertension in the general study population (Table [2](#Tab2){ref-type="table"}, Fig. [1](#Fig1){ref-type="fig"}a). This ambivalent association may be the result of different pathways increasing the platelet levels. Endothelial injury caused by hypertension stimulates platelet production, which results in positive association between these two factors. On the other hand, genetic factors stimulate endothelial repair by inducing the production of platelets, which has a beneficial effect on preventing hypertension, since endothelial dysfunction has a bidirectional relationship with hypertension^[@CR1]^. In a previous study, were found that the MAF of rs3782886 (BRAP) was significantly positively associated with a high platelet count, which is associated with hypertension^[@CR16]^, supports this mechanism. In addition, as in our previous study^[@CR15]^, here, we found a significantly positive association between reticulocyte levels and hypertension (Table [2](#Tab2){ref-type="table"}, Fig. [1](#Fig1){ref-type="fig"}b), while no significant association was found between reticulocyte levels and genetic factors (rs3782886 and rs671) (Tables [3](#Tab3){ref-type="table"}, [4](#Tab4){ref-type="table"}, Fig. [1](#Fig1){ref-type="fig"}e), also supporting this mechanism; hypertension related endothelial injury caused by process of aging such as low grade inflammation and increased oxidative stress^[@CR10],\ [@CR11]^ stimulates the production of reticulocytes, while genetic factors do not stimulate reticulocytes production.
A potential biochemical mechanism that may underlie the association between the rs3782886 genotype and platelet count is the activation of the NF-κB pathway. A higher expression of the BRAP minor allele is associated with an increased risk of atherosclerosis, possibly by heightening the degree of inflammation via the activation of the NF-κB pathway^[@CR3],\ [@CR21]^. Since the activation of the NF-κB pathway may also promote platelet activation proteins^[@CR22]^, the NF-κB pathway may influence the association between rs3782886 and platelet count. An increased platelet count stimulates endothelial repair activity^[@CR14]^.
Strong linkage disequilibrium (LD) values have been reported between rs3782886 and rs671^[@CR5]^. In the present study, we found a significantly positive correlation between the MAF of rs671 and rs3782886 (Fig. [1](#Fig1){ref-type="fig"}j). The simple correlation coefficient (r) of these genotypes were r = 0.94 (p \< 0.001) for men and r = 0.94 (p \< 0.001) for women. Therefore, we found essentially the same associations as for the rs671 genotype (Tables [3](#Tab3){ref-type="table"}, [4](#Tab4){ref-type="table"}, [5](#Tab5){ref-type="table"}, [6](#Tab6){ref-type="table"}, Fig. [1](#Fig1){ref-type="fig"}c,d,e,h). Similar to a previous study^[@CR23]^, we found that rs3782886 was strongly associated with the status of the never drinkers (Table [5](#Tab5){ref-type="table"}, Fig. [1](#Fig1){ref-type="fig"}h). From an anthropological point of view, an extensive prevalence of SNPs should have the same beneficial effect on the participants' daily activities, rather than impose a disadvantage. Ethanol induces oxidative stress^[@CR24]^ that might stimulate reticulocyte production. In the present study, we found that non-never drinkers displayed significantly higher reticulocyte levels than never drinkers (Fig. [1](#Fig1){ref-type="fig"}i). In addition, acute ethanol exposure dramatically inhibits NF-κB activation^[@CR25]^. And rs3782886 exists primarily for the activation of endothelial repair via activating the NF-κB pathway. Furthermore, another study reported that ethanol directly attenuates platelet activation and has a significant endothelial cell-mediated effect on selected markers of atherosclerosis *in vitro*^[@CR19]^. Therefore, avoiding exposure to ethanol is highly beneficial for individuals with the non-major genotype of rs3782886. In the present study, never-drinkers showed significantly higher platelet counts than those in the non-never drinkers with high reticulocyte levels, but not for those individuals with low reticulocyte levels (Fig. [1](#Fig1){ref-type="fig"}g).
Hypertension, which is widely found in Japan, is a known risk factor of stroke^[@CR26]^. Japanese individuals are characterized by higher rates of stroke than myocardial infarction^[@CR27]^. Even SNPs in BRAP have been reported to be associated with the risk of myocardial infarction^[@CR2]^, but not with stroke^[@CR28]^. As such, genetic factor has the potential to have a strong beneficial influence on preventing hypertension, especially in the Japanese population, which has a higher risk of stroke. In the present study, among participants carrying the minor rs3782886 allele, never drinkers were more represented among women than among men; indeed, 88.8% of heterozygous and 96.6% of homozygous women for rs3782886 were never drinkers, while for men the corresponding values were 41.9% and 77.5%. Therefore, the beneficial influence of the present genetic factor might be stronger for women than men. Since the present study population is composed of elderly Japanese, such beneficial influence on lifespan might result in a disrupted Hardy--Weinberg equilibrium in women but not in men. Since the present genetic factor might exert its beneficial influence by activating endothelial repair, further investigations with a long-term follow-up study of oxidative stress is necessary to verify this hypothesis.
This study contains several strengths. Firstly, this study is the first to demonstrate the influence of hematopoietic activity on the association between genetic factors and hypertension. Unlike previous epidemiological studies, this study used multi-faceted analyses to determine potential mechanisms underlying our main findings. Furthermore, the Japanese population is known to have a high prevalence for the minor allele of rs671, which is associated with a low tolerance to ethanol exposure^[@CR29]^. This study is also the first to reports the potential mechanism for why the Japanese population expresses genetic factors that are unfavorable to ethanol exposure.
This study has some limitations that warrant consideration. Firstly, the activation of the NF-κB protein may have influenced our present results, as no data concerning NF-κB protein was available. Further studies taking into account data for the activity of NF-κB protein will be necessary. In conjunction with the circulating CD34-posistive cell levels, the platelet count plays an important role in endothelial repair^[@CR1],\ [@CR14]^. Therefore, the levels of circulating CD34-positive cell may also act as a determining factor in the present analysis. However, due to the difficulty of measuring CD34-posistive cells in a general health check-up, no data on the levels of circulating CD34-posistive cells were included in this study.
In conclusion, the MAF of rs3782886 and rs671 are only significantly inversely associated with hypertension and significantly and positively associated with the platelet count in individuals with high levels of hematopoietic activity (high reticulocyte levels). Furthermore, the status of never-drinkers showed significantly higher platelet counts than those in non-never drinkers with high reticulocyte levels, whereas the MAF of rs3782886 and rs671 was significantly and positively associated with the status of a never-drinker. Since ethanol directly attenuates platelet activation^[@CR19]^ these associations represented efficient tool to clarify the genetic factors that influence hypertension in the Japanese population.
**Publisher\'s note**
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
This study was supported by Grants-in-Aids for Scientific Research from Japan Society for the Promotion of Sciences (no. 17H03740 and 18K06448). We are grateful to the staff of Goto city office for their outstanding support. We would like to thank Editage (<https://www.editage.jp>) for English Language editing.
Y.S. designed the study, performed the statistical analyses, interpreted the data, and drafted and revised the manuscript. Y.S., K.A., Y.N., S.Y.K., H.Y., M.T., and Y.N. assisted with the study design, were involved in data collection, and checked the manuscript. K.A., H.Y., S.Y.K, Y.N. and T.M. participated in the study concept and checked the manuscript. T.M. was the general coordinator and designed the study.
The authors declare no competing interests.
| {
"pile_set_name": "PubMed Central"
} |
Background {#Sec1}
==========
Praziquantel (PZQ) is virtually the sole treatment regimen for Schistosomiasis in sub-Saharan Africa \[[@CR1]\]. This oral schistosomicidal agent is constituted of a racemate mixture, with activity both in vivo and in vitro \[[@CR2], [@CR3]\]. Evidence from studies conducted on *Schistosoma* (*S.*) *mansoni* and *S. japonicum* although not very clear, indicate the mode of action of PZQ is the targeting of calcium channels and antigen exposure rendering the worm susceptible to elimination by antibodies \[[@CR1], [@CR4]\].
After oral administration, PZQ is rapidly absorbed, metabolized and excreted by the kidney. Metabolism of PZQ is primarily via the cytochrome P450 system leading to the production of toxic metabolic intermediates, which are potentially harmful to hepatocytes \[[@CR5]\]. Plasma levels of PZQ are also reported to be reduced by inducers but elevated by inhibitors of cytochrome P450 activity \[[@CR6], [@CR7]\].
Several studies, predominantly in Asian populations, where *S. japonicum* infections are endemic, state conflicting findings on hepatotoxicity associated with PZQ treatment against the helminth \[[@CR8], [@CR9]\]. PZQ treatment is reported to be associated with elevated serum concentrations of liver aminotransferase \[[@CR8]\]. However, in a large retrospective study from China, there was insignificant (less than 1%) incidence of hepatotoxicity among populations treated for *S. japonicum* with PZQ \[[@CR9]\].
Therapy for Schistosomiasis in sub-Saharan Africa has mainly been documented based on intestinal *S. mansoni* infections \[[@CR1]\]. As a result, there is paucity of data on urinary *S. haematobium* and its associated drug metabolism effects on organs involved in metabolizing and excretion of PZQ. This leaves a gap in knowledge about the protective or destructive effect of metabolizing the drug in *S. haematobium* infection. It has further been shown that varied degrees of reduction in incidence and infection rates of *S. haematobium* are reported with mostly single PZQ dosage of 40 mg/kg/day in both children and adults \[[@CR1]\]. There are also indications of drug resistance to single doses of PZQ for treating schistosomiasis \[[@CR10]\]. This heightens the need to probe the outcome of repeated PZQ treatment on urinary schistosome counts against its implication on liver and renal function.
The aim of this study was to assess the effect of PZQ on schistosome egg count, liver and renal function after 3 doses of 60 mg/kg/day (PZQ60) in three months for treating urinary *S. haematobium* infection*.*
Methods {#Sec2}
=======
This was a nested-Case Control study conducted among children and adults of ages 6--30 years from the urban Asokwa District in Kumasi, Ghana (see plate 1). This was part of a larger study to assess plasmodium transmission in persons infected with Schistosomiasis ([NCT02769013](https://www.clinicaltrials.gov/ct2/show/NCT02769013)). Ethical approval was obtained from the Committee for Human Research Publication and Ethics, School of Medical Sciences, Kwame Nkrumah University of Science and Technology, Ghana. All participants were required to sign an informed consent. For minors below 16 years, a signed assent form from the participant and an informed consent from a parent or guardian were obtained. Cases were respondents diagnosed to have *S. haematobium* by routine microscopic examination of urine samples. Controls from the same communities, without laboratory or clinical detection of urinary schistosomiasis infection were age and sex matched with cases.
Study area {#Sec3}
----------
Apromase, Deduako, Emena and Kokoben were the study communities in the urban Asokwa District with a population of 140,161 inhabitants in 36, 183 households (Fig. [1](#Fig1){ref-type="fig"}) \[[@CR11]\]. These communities are located between latitude 6°30′ and 7°00′ North and longitude 1°30′ and 2°00 West of Kumasi, the capital city of the Ashanti Region of Ghana. The four communities have Saman (Kokoben and Apromase), Oda (Deduako) and Subin (Emena) as names of three rivers running through it. Fig. 1Map of study communities and sites (Rivers) in Ashanti region, Ghana
Climatic conditions are tropical with temperatures varying from 20.2 °C to 37.1 °C. Rainfall pattern is seasonally bimodal with major rains extending from late April to August with a minor one from September to October \[[@CR12]\]. The average annual rainfall for the area is 6.25 mm with peaks of 214.3 mm and 16.2 mm in June and September respectively. The dry season (harmattan) is from November to March with humidity ranging between 53 and 93%.
Screening and enrolment {#Sec4}
-----------------------
A census of the selected communities was conducted with the ages and number of inhabitants per building collected along with corresponding GPS coordinates using Personal Digital Assistants (PDAs). Households within the buildings were selected and their members asked for written informed consent to be screened in the study. Twenty millilitres (20 ml) of urine samples were collected once, from consenting participant into well-labelled 30 ml urine containers. The urine samples were collected within the hours of 6:00 am and 12 noon. Subsequently, the samples were kept in cold boxes at temperature of 4--6 °C until transported to the laboratory at Kumasi Centre for Collaborative Research in Tropical Medicine (KCCR) about 15--20 min drive from the study sites.
Asymptomatic schistosomiasis positive (SP) cases and schistosomiasis-negative (SN) controls based on screening results were invited to participate in the study.
Sampling procedures {#Sec5}
-------------------
A total of 1258 participants were screened for schistosomiasis out of which 104 were positive. All 104 schistosome positive participants (Fig. [2](#Fig2){ref-type="fig"}) were placed on PZQ60 treatment. Controls were selected from the same communities and matched in a 2: 1 ratio with cases by sex and age. Out of the 104 positive cases, 32 started the treatment phase with 28 successfully completing the course with samples analyzed. On the other hand, 53 controls had all samples analyzed. Fig. 2Sampling of study participants
Design of Experiment {#Sec6}
--------------------
Biochemical parameters and schistosome counts were analysed before and after treatment for both cases and controls (Fig. [3](#Fig3){ref-type="fig"}). In between pre- and post-treatment, the biochemical and schistosome counts were monitored before the 2nd and 3rd dosages of PZQ60 for cases. Fig. 3Flowchart of experimental design
Laboratory processes {#Sec7}
--------------------
### Processing of urine and S. haematobium quantification {#Sec8}
Freshly voided urine was collected between 6:00 am and 12:00 pm in a sterile wide-mouthed screw-top plastic container (30 ml) and transported to the laboratory on ice at 4 °C to 8 °C. Urine processing and quantification were done as described by Cheesebrough \[[@CR13]\]. Briefly, blunt-ended forceps were utilized to place a polycarbonate membrane filter of pore size of 12.0 μm (Whatmann Nuclepore) on the filter-support of a filter-holder (Swinnex 25 mm support chamber). The filter holder was re-assembled and attached to a 10 ml syringe filled with well-mixed urine which was filtered with the aid of the plunger. The filter was carefully removed and transferred with the face upwards to a clean glass-slide. A drop of physiological saline was added, covered with a cover-slip, and examined by two independent expert microscopists using the 10X objective (Carl Zeiss Microscope) with the condenser iris closed sufficiently to give good contrast. The entire filter was examined systematically for the presence of *S. haematobium* eggs. The number of the eggs counted per 10 ml of urine was recorded and the average of the two counts was calculated. A slide was declared negative when no parasites were detected.
### Blood sampling and processing {#Sec9}
Blood samples were collected from the antecubital vein with the aid of a tourniquet and the puncture site cleaned with 70% alcohol prep. Blood drawn into separator gel tubes (5 ml) were centrifuged at 1780 x g for 10 min at 4 °C to obtain the sera which were subsequently stored at -80 °C.
### Biochemical analysis {#Sec10}
Assays for the liver function; albumin, globulin, aspartate aminotransferase (AST), alanine aminotransferase (ALT) and gamma glutamyl transferase (GGT) and renal function (urea and creatinine), were conducted using a chemistry analyzer (HumaStar 200, Human, Germany). Elevated levels of AST, ALT and GGT are known indications of liver damage which may be caused by drug metabolism, infections and alcohol consumption. Increase in the levels of globulin is also found to correlate with infection or an inflammatory state. Aliquots of the serum were dispensed into cuvettes after thawing and placed at pre-programmed positions in the auto-analyzer and analysis done in batches. Due to the limited strength of serum urea and creatinine as markers of renal function, renal insufficiency based on estimated glomerular filtration was used as a more robust indicator. Renal insufficiency was assessed based on the 4-variable Modification of Diet in Renal Disease (4v-MDRD) and the Chronic Kidney Disease Epidemiology Program (CKDEPI) equations \[[@CR14]\]. The 4v-MDRD estimates the glomerular filtration rate based on serum creatinine concentration, age, sex and race whiles the CKDEPI uses variations of an equation based on cut-offs for serum creatinine concentrations and sex.
Statistical analysis {#Sec11}
--------------------
Data were entered into excel and analyzed with Stata V.12 (StataCorp, USA). Continuous variables were reported as median with interquartile ranges and categorical variables as proportions. The Wilcoxon signed rank test was used to assess differences in continuous variables between groups with statistical significance set at *p* \< 0.05.
Results {#Sec12}
=======
Characteristics of study population before treatment {#Sec13}
----------------------------------------------------
Majority of the study population were males \[78.6 (62.4, 94.8) for schistosome positives and 81.1 (70.2, 92.0) for controls\] with no significant statistical difference in age for the two groups (*p* = 0.547). Serum albumin levels were higher (*p* = 0.003) in controls \[55.9 (52.1, 61.5)\] compared with schistosome cases \[51.9 (48.5, 53.0)\]. On the contrary, serum globulin levels were elevated (*p* = 0.005) in cases 36.7 (32.8, 40.1) compared with controls 30.5 (22.4, 33.8). No significant difference was reported between schistosome cases and controls for liver (ALT, AST and GGT) and kidney (serum creatinine and urea) markers. Similarly, no significant difference (*p* = 0.600) was reported for estimated glomerular filtration rate (eGFR) between schistosome positive \[96.0 (80.0, 116.0)\] and negative \[101.0 (88.0, 127.0)\] groups using the CKDEPI equation before treatment (Table [1](#Tab1){ref-type="table"}). Furthermore, the 4v-MDRD renal equation also showed no difference in levels between schistosome positive and negative groups at pre-treatment (*p* = 0.776). Table 1Demographics, schistosome counts and biochemical variables before treatmentVariableNormal RangesSchistosome positive\
***n*** = 28Schistosome negative ***n*** = 53***p*** value**Male%**78.681.1**Age (years)**15.0 (12.5, 21.5)16.0 (13.0, 22.0)0.547**Sh count (/10 mL)**3.0 (2.0, 11.0)00.033**Albumin (g/L)**35--5551.9 (48.5, 53.0)55.9 (52.1, 61.5)0.003**Globulin**20--3536.7 (32.8, 40.1)30.5 (22.4, 33.8)0.005**Bilirubin (Direct) (**μmol**/L)**0--55.7 (3.4, 7.5)5.7 (4.5, 7.1)0.825**Bilirubin (Total) (μmol/L)**3--2214.1 (11.4, 20.2)14.8 (12.4, 19.2)0.809**Creatinine (μmol/L)**50--11099.0 (77.0, 116.0)91.0 (75.0, 114.0)0.587**GGT (U/L)**9--4820.0 (16.0, 26.0)24.0 (16.0, 33.0)0.243**ALT (U/L)**5--3513.0 (9.0, 16.0)13.0 (9.0, 18.0)0.420**AST (U/L)**7--4033.0 (28.0, 36.0)30.0 (25.0, 35.0)0.825**AST/ALT ratio**0.82.5 (1.9, 3.1)2.6 (2.0, 3.1)0.457**Protein (g/L)**60--8089.0 (82.0, 92.0)87.0 (80.0, 91.0)0.840**Urea (mmol/L)**2.9--8.22.7 (2.1, 3.1)2.6 (2.1, 3.0)0.344**eGFR (CKDEPI)**\> 9096.0 (80.0, 116.0)101.0 (88.0, 127.0)0.600**eGFR (4v-MDRD)**\> 90100.5 (82.5, 120.0)95.0 (83.0, 123.0)0.776*Sh S. haematobium* egg. eGFR (ml/min/1.73m^2^), proportion reported as percentage (interquartile range), biochemical variables reported as median (interquartile ranges)
Effect of treatment of cases with 3 doses of PZQ60 on liver, renal function and S. *haematobium* egg count {#Sec14}
----------------------------------------------------------------------------------------------------------
After the first dose, ALT levels increased significantly (*p* = 0.001) and gradually declined after the second dose and then significantly (*p* = 0.001) towards post-treatment stage (Fig. [4](#Fig4){ref-type="fig"}). AST levels were elevated (*p* = 0.006) after the first dose and significantly (*p* = 0.001) declined before the third dose. However, there was no statistically significant difference (*p* = 0.577) in AST levels from before the third dose to post-treatment. GGT levels increased sharply (*p* = 0.001) after the first dose and declined significantly (*p* = 0.001) after the second dose. Schistosome egg count decreased significantly (*p* = 0.001) after the third dose towards post-treatment. Fig. 4Effect of three repeated monthly doses of praziquantel 60 mg/kg on median liver enzymes and *S*. *haematobium* egg count. 1 = Pre-treatment, 2 = Before 2nd dose, 3 = Before 3rd dose, 4 = Post-treatment
Estimated glomerular filtration rates (Fig. [5](#Fig5){ref-type="fig"}) decreased significantly after the first dose using both CKDEPI (*p* = 0.003) and 4v-MDRD (*p* = 0.004). On the contrary, eGFR increased significantly from before second dose towards post-treatment for both CKDEPI (*p* = 0.001) and 4v-MDRD (*p* = 0.002). Fig. 5Effect of three repeated monthly doses of praziquantel 60 mg/kg on median estimated glomerular filtration rate and *S*. *haematobium* egg count. 1 = Pre-treatment, 2 = Before 2nd dose, 3 = Before 3rd dose, 4 = Post-treatment. eGFR reported in mL/min/1.73m^2^
Effect of treatment with 3 doses of PZQ60 on cure rate {#Sec15}
------------------------------------------------------
Estimated cure rate was 42.9% a month after administering the first dose of PZQ60 (Table [2](#Tab2){ref-type="table"}). This increased to 46.4% before administering the third dose and finally to 96.4% post-treatment. Table 2Cure rates of 60 mg/kg praziquantel at stages of repeated dose treatmentBefore 2nd doseBefore 3rd dosePost-treatment**Total Sample**282828**Median eggs/10 mL**12.0 (1.0, 17.0)6.0 (4.0, 38.0)0**Number cured**121327**Cure rate**42.9%46.4%96.4%Median eggs reported with interquartile ranges
Comparing Schistosome count and biochemical parameters between cases and controls after treatment {#Sec16}
-------------------------------------------------------------------------------------------------
At post-treatment, there was no significant difference in mean levels for albumin (*p* = 0.441) between cases \[48.8 (45.4, 52.5)\] and controls \[52.9 (49.6, 55.4)\]. Similarly, mean levels of ALT, AST and GGT did not vary statistically between cases and controls post-treatment (Table [3](#Tab3){ref-type="table"}). Moreover, creatinine and urea levels did not statistically vary between cases and controls. No significant difference (*p* = 0.753) was recorded using CKDEPI equation at post-treatment (Table [3](#Tab3){ref-type="table"}) between the two groups. Likewise, 4v-MDRD reported no significant difference post-treatment (*p* = 0.866), between cases and controls. Table 3Schistosome count and biochemical parameters after treatmentVariableNormal RangeSchistosome positiveSchistosome negative***p*** value**Sh count (/10 mL)**00**Albumin (g/L)**35--5548.8 (45.4, 52.5)52.9 (49.6, 55.4)0.441**Globulin**20--3535.8 (31.2, 39.1)37.4 (29.7, 43.0)0.767**Bilirubin (Direct) (μmol/L)**0--57.4 (5.8, 8.9)7.0 (6.3, 8.0)0.075**Bilirubin (Total) (μmol/L)**3--2213.6 (11.1, 20.6)14.2 (11.6, 16.7)0.260**Creatinine (μmol/L)**50--11091.0 (79.0, 111.0)95.0 (77.0, 112.0)0.441**GGT (U/L)**9--4823.0 (20.0, 31.0)26.0 (20.0, 39.0)0.678**ALT (U/L)**5--3511.0 (8.0, 13.0)10.0 (9.0, 13.0)0.513**AST (U/L)**7--4032.0 (27.0, 35.0)30.0 (26.0, 33.0)0.722**AST/ALT ratio**0.82.8 (2.3, 3.4)2.7 (2.1, 3.3)0.441**Protein (g/L)**60--8085.0 (77.0, 93.0)85.0 (82.0, 96.0)0.813**Urea (mmol/L)**2.9--8.22.5 (2.0, 2.8)2.2 (1.9, 2.5)0.635**eGFR (CKDEPI)**\> 90116.5 (94.0, 128.0)94.5 (79.0, 115.0)0.753**eGFR (4v-MDRD)**\> 90105.5 (87.0, 116.0)90.0 (75.0, 109.0)0.866*Sh S. haematobium* egg, eGFR (ml/min/1.73m^2^), biochemical variables, ratio and eGFR reported as median (interquartile ranges)
Discussion {#Sec17}
==========
Schistosome egg numbers declined after administration of the second dose of 60 mg/kg praziquantel and almost completely eliminated post-treatment with 96.4% cure rate. The estimated glomerular filtration rate dropped significantly after the first dose of praziquantel but resolved after the third dose. Moreover, levels of liver enzymes increased after the first dose and returned to pre-treatment levels in schistosome positive cases post-treatment.
Several studies assessing the efficacy of praziquantel treatment on *S*. *haematobium* has been based on single doses of 40 mg/kg with cure rates ranging from 39.8 to 88.9% in mainly children below 17 years \[[@CR15]--[@CR17]\]. In addition, multiple dose regimens have been reported to clear *S*. *haematobium* with 53.1 to 100.0% cure rates \[[@CR18], [@CR19]\]. We report a final cure rate of 96.4% after a monthly dose of PZQ60 for three months. Contrary to case reports \[[@CR20], [@CR21]\] where multiple courses of 40 mg/kg failed in clearing persistent *S*. *haematobium* infection, this study finding provides evidence of an efficacious treatment regimen with repeated and higher doses of PZQ60 which can clear Schistosome eggs and eliminate the worm stage of the parasite. This assertion is further supported by an initial increase in schistosome egg count before the second dose which could be attributed to a single dose of PZQ60 having little effect on juvenile worms as concluded in a previous study that persist and perpetuate more eggs \[[@CR22]\]. It is worthy of note that although praziquantel is the standard treatment for schistosome infections due to reported efficacy and minimal side effects \[[@CR23]\], there are studies that have found cases of drug failure \[[@CR20], [@CR21]\]. However, these investigations are not definite on the causes of these treatment failures. It is also plausible that the low cure rates after the first and second dose could be attributed to some level of resistance to PZQ previously reported in *S*. *haematobium* infections \[[@CR24]\] in some endemic regions and could be now present in Ghana.
With respect to organ damage, many studies have tried to analyze the efficacy of varying concentrations of PZQ on organ injury \[[@CR25], [@CR26]\]. However, little is known of its effect to the liver and kidney as a result of repeated and higher doses for treating S. *haematobium* among sub-Saharan subjects. It is well established that metabolizing of PZQ analogs by cytochrome P450 can lead to highly oxidative intermediates \[[@CR27]\] which overwhelm the body's defense resulting in varied levels of organ damage. This study found a significant increase in liver enzymes after the first dose, which steadily dropped to pre-treatment levels. The initial upsurge in liver enzymes could be explained as due to highly oxidative intermediates of PZQ than can be quenched by the antioxidant activity of superoxide dismutase, glutathione and glutathione S-transferase. Consequently, hepatic cells are disrupted leading to the release of compartmentalized liver enzymes. However, the steady return after two more doses of PZQ to pre-treatment levels could be proof of the body system's response of mopping up toxic metabolites by inducing higher expression of cells that synthesize these proteins involved in the body's redox systems.
Similarly, the significant drop in estimated glomerular filtration rate for both the CKDEPI and 4v-MDRD equation after the first 60 mg/kg dose shows the body's initial intolerance to oxidative species from the metabolism of the drug which improved even after monthly repeated doses. It is plausible that sustained higher thresholds of these induced protective antioxidants persist after its initial deficit compared with the highly oxidative drug intermediates associated with organ damage. This suggests possible protective effect of monthly repeated doses of PZQ on the renal filtration with a corresponding decline in schistosome egg count.
It is possible that taking other medications with this treatment regimen may further increase secretion of liver enzymes into blood. However, data backing this assertion is lacking from this study. Moreover, the reported levels of liver enzymes and estimated glomerular filtration rates could have been influenced by viral infections and wasting of muscles respectively. Even though the study finding gives a strong indication that PZQ60 given over 3 months is safe and effective, the willingness of only 32/104 (31%) of egg positive cases to participate in the repeated treatment (Fig. [2](#Fig2){ref-type="fig"}) suggests this regimen may have low compliance. It is recommended that further research is done on improving compliance to increase treatment coverage if multiple dosing is to be employed.
Conclusion {#Sec18}
==========
This study suggests that treatment of urinary *S. hematobium* infections with a repeated high monthly dose of 60 mg/kg of praziquantel for 3 months is effective and safe. It provides an option to consider for *S*. *haematobium* infection cases with drug resistance.
ALT
: Alanine aminotransferase
AST
: Aspartate aminotransferase
CKDEPI
: Chronic Kidney Disease Epidemiology Program
eGFR
: Estimated glomerular filtration rate
GGT
: Gamma glutamytransferase
SN
: Schistosomiasis negative
SP
: Schistosomiasis positive
PZQ
: Praziquantel
4v-MDRD
: 4-variable Modification of Diet in Renal Disease
**Publisher's Note**
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Authors are grateful to the study communities and participants for making this research a success. We also thank Dr. Oumou Maiga-Ascofare for her efforts and hard work in coordinating this study. We are further grateful to Tony Sarfo-Acquah and Patrick Obuam for their assistance in data collection.
SND, EOD and DY conceived the idea. PAK, HH and SBA recruited study participants. SND, PAK and HH conducted sample analysis. SND and STA conducted the statistical analysis. SND drafted the manuscript which was reviewed and approved by all authors.
This study was funded by the EOD group, Kumasi Centre for Collaborative Research in Tropical Medicine, Kumasi, Ghana.
The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request.
Ethical approval was obtained from the Committee for Human Research Publication and Ethics, School of Medical Sciences, Kwame Nkrumah University of Science and Technology, Ghana (CHRPE/AP/227/17). All participants were required to sign an informed consent. For minors below 16 years, a signed assent form from the participant and an informed consent from a parent or guardian were requirements for participation.
Not applicable.
The authors declare that they have no competing interests.
| {
"pile_set_name": "PubMed Central"
} |
Main {#Sec1}
====
RNA interference (RNAi) is a natural cellular process that regulates gene expression and provides an innate defence mechanism against invading viruses and transposable elements^[@CR1]^. The finding that dsRNA initiates RNAi was among the most significant recent contributions to cell biology^[@CR2]^, and since the discovery that RNAi can be mediated by 21 nucleotide (nt) duplexes^[@CR3]^, researchers have worked to harness their potential for addressing biological questions and treating human disease. Some reagents, such as small interfering RNAs (siRNAs), are applied directly to cells, tissues and organisms; others are engineered to be expressed in cells, such as hairpin structures that provide siRNAs when processed. The basic premise underlying the broad utility of RNAi is that, in theory, we can design siRNAs (or vectors encoding them) to target virtually any gene of interest. RNAi technologies use a cell\'s natural machinery to move exogenously applied siRNAs to the appropriate cellular compartment, where they encounter the correct mRNA target and induce its degradation.
Initial work on RNAi in flies and worms moved quickly to larger mammals and fuelled excitement for potential clinical applications. However, in a similar way to other developing fields in human therapy, such as gene- and antibody-therapy, early excitement has been tempered as a realistic understanding emerges of the milestones that must be reached before the eventual approval of human therapy. Over recent years various complex barriers to achieving efficient RNAi have become evident. These hurdles include: specificity for the target gene; delivery to the correct cell or tissues; the durability of RNAi activity and the ability to redose (if needed); and considerations of the stability of the target mRNA and encoded protein. We have also become aware of the problems posed by the various platforms used to elicit RNAi. However, with setbacks come opportunities. For example, early work in which siRNAs were applied to mouse airway epithelial cells to reduce viral burdens *in vivo* elicited inhibition of target gene expression^[@CR4]^, but it was later found that the induction of an innate immune response probably contributed to the efficacy in these and other experiments^[@CR5],[@CR6]^. Altering the chemical make-up of the synthetic RNA diminished the immune response, as did avoiding known pro-inflammatory sequences^[@CR7],[@CR8]^. This finding also opened investigators\' eyes to the possibility of purposefully using immunostimulatory RNAi as a direct therapeutic or adjuvant^[@CR9]^.
Although the clinical utility of RNAi has not yet been realized, ongoing patient trials provide opportunities for success. The numbers of RNAi-based preclinical and clinical trials have grown over the past several years and have included studies in retinal degeneration, dominantly inherited brain and skin diseases, viral infections, respiratory disorders, cancer and metabolic diseases ([Table 1](#Tab1){ref-type="table"}).Table 1Clinical trials for RNAi therapy^\*^Clinical settingDrugIndication(s)Target(s)SponsorStatusOcular and retinal disordersTD101Pachyonychia congenitaKeratin 6A N171K mutantPachyonychia Congenita ProjectCompleted, Phase IQPI-1007Non-arteritic anterior ischaemic optic neuropathyCaspase 2Quark Pharm., Inc.Active, Phase IAGN211745Age-related macular degeneration; choroidal neovascularizationVEGFR1Sirna Therapeutics, Inc.Completed, Phase I, IIPF-655Diabetic macular oedema (DME); age-related macular degeneration (AMD)RTP801Quark Pharm., Inc.Active, Phase ISYL040012Glaucomaβ2 adrenergic receptorSylentisActive, Phase I, IIBevasiranibDiabetic macular oedemaVEGFOpko Health, Inc.Completed, Phase IIBevasiranibMacular degenerationVEGFOpko Health, Inc.Completed, Phase IICancerCEQ508Familial adenomatous polyposisβ-cateninMDRNA, Inc.Active, Phase IALN-PLK1Liver tumoursPLK1Alnyam Pharm.Active, Phase IFANGSolid tumoursFurinGradalis, Inc.Active, Phase IICALAA-01Solid tumoursRRM2Calando Pharm.Active, Phase ISPC2996Chronic myeloid leukemiaBCL-2Santaris Pharm.Ongoing, Phase I, IIALN-VSP02Solid tumoursVEGF, kinesin spindle proteinAlnylam Pharm.Active, Phase INCT00672542Metastatic melanomaLMP2, LMP7, and MECL1Duke UniversityActive, Phase IAtu027Advanced, recurrent or metastatic solid malignanciesPKN3Silence TherapeuticsActive, Phase IKidney disordersQPI-1002/I5NPAcute kidney injuryp53Quark Pharm., Inc.Terminated, Phase IQPI-1002/I5NPDelayed graft function kidney transplantp53Quark Pharm., Inc.Active, Phase I, IIQPI-1002/I5NPKidney injury acute renal failurep53Quark Pharm., Inc.Completed, Phase ILDL loweringTKM-ApoBHypercholesterolaemiaAPOBTekmira Pharm. Corp.Terminated, Phase IPRO-040,201HypercholesterolaemiaAPOBTekmira Pharm. Corp.Terminated, Phase IAntiviralSPC3649Hepatitis C virusmiR-122Santaris PharmActive, Phase IIpHIV7-shI-TAR-CCR5RZHIVHIV Tat protein, HIV TAR RNA, human CCR5City of Hope Medical Center/BenitecActive, Phase 0ALN-RSV01RSV in volunteersRSV nucleocapsidAlnylam Pharm.Completed, Phase IIALN-RSV01RSV in lung transplant patientsRSV nucleocapsidAlnylam Pharm.Completed, Phase IALN-RSV01RSV in lung transplant patientsRSV nucleocapsidAlnylam Pharm.Active, Phase IIAPOB, apolipoprotein B; BCL-2, B-cell CLL/lymphoma 2; CCR5, C-C chemokine receptor type 5; LDL, low-density lipoprotein; LMP2, also known as proteasome subunit beta type 9 (PSMB9); LMP7, also known as proteasome subunit beta type 8 (PSMB8); MECL1, also known as proteasome subunit beta type 10 (PSMB10); Pharm., Pharmaceuticals; PKN3, protein kinase N3; PLK1, polo-like kinase 1; RRM2, ribonucleoside-diphosphate reductase subunit M2; RSV, respiratory syncytial virus; RTP801, also known as DNA damage-inducible transcript 4 protein (DDIT4); VEGF, vascular endothelial growth factor.^\*^From [ClinicalTrials.gov](http://clinicaltrials.gov/).
Here, we provide an overview of RNAi and discuss strategies to use the pathway for directed gene silencing for therapy. We describe delivery systems that might be suitable for different circumstances, and bring to the reader\'s attention issues that must be surmounted for widespread use *in vivo*.
**Harnessing small RNA biogenesis**
The development of RNAi for therapy is based on our understanding of small RNA biogenesis pathways. The two main types of small RNAs involved in gene silencing are microRNAs (miRNAs) and siRNAs, and their processing and targeting is summarized in [Fig. 1](#Fig1){ref-type="fig"} (further details can be found in recent reviews^[@CR10],[@CR11],[@CR12]^).Figure 1The miRNA and siRNA pathways of RNAi in mammals.Primary microRNAs (pri-miRNAs) are transcribed by RNA polymerases^[@CR156],[@CR157],[@CR158]^ and are trimmed by the microprocessor complex (comprising Drosha and microprocessor complex subunit DCGR8) into \~70 nucleotide precursors, called pre-miRNAs^[@CR67],[@CR159],[@CR160]^ (left side of the figure). miRNAs can also be processed from spliced short introns (known as mirtrons)^[@CR161]^. pre-miRNAs contain a loop and usually have interspersed mismatches along the duplex. pre-miRNAs associate with exportin 5 and are exported to the cytoplasm^[@CR162],[@CR163]^, where a complex that contains Dicer, TAR RNA-binding protein (TRBP; also known as TARBP2) and PACT (also known as PRKRA) processes the pre-miRNAs into miRNA--miRNA\* duplexes^[@CR116],[@CR164],[@CR165]^. The duplex associates with an Argonaute (AGO) protein within the precursor RNAi-induced silencing complex (pre-RISC). One strand of the duplex (the passenger strand) is removed. The mature RISC contains the guide strand, which directs the complex to the target mRNA for post-transcriptional gene silencing. The \'seed\' region of an miRNA is indicated; in RNAi trigger design, the off-target potential of this sequence needs to be considered. Long dsRNAs (right side of the figure) are processed by Dicer, TRBP and PACT into small interfering RNAs (siRNAs). siRNAs are 20--24-mer RNAs and harbour 3′OH and 5′ phosphate (PO~4~) groups, with 3′ dinucleotide overhangs^[@CR3],[@CR166],[@CR167]^. Within the pre-RISC complex, an AGO protein cleaves the passenger siRNA strand. Then, the mature RISC, containing an AGO protein and the guide strand, associates with the target mRNA for cleavage. The inset shows the properties of siRNAs. The thermodynamic stability of the terminal sequences will direct strand loading. Like naturally occurring or artificially engineered miRNAs, the potential \'seed\' region can be a source for miRNA-like off-target silencing. shRNA, short hairpin RNA.
***miRNAs and siRNAs***. miRNAs mediate post-transcriptional gene silencing and are processed from endogenously expressed transcripts ([Fig. 1](#Fig1){ref-type="fig"}). Either processed strand can mediate post-transcriptional gene silencing, but many miRNAs show asymmetry, primarily loading one strand into the [RNA-induced silencing complex](#Glos1){ref-type="list"} (RISC). The small RNA guides RISC to the mRNA target, where the miRNA typically binds to the 3′ UTR. Watson--Crick base pairing between miRNAs and their targets is usually partial, but with high complementarity from bases 2--8 of the miRNA, which is known as the \'[seed](#Glos2){ref-type="list"}\' region^[@CR13]^. Recent data suggest that base pairing can also occur between central miRNA nucleotides and target mRNAs^[@CR14]^. Data from several laboratories showed that miRNAs repress the initiation of translation^[@CR15],[@CR16],[@CR17]^, although more recent work indicates that miRNA--mRNA complexes can be transported to cytoplasmic processing bodies^[@CR18]^, after which deadenylation and mRNA degradation occurs^[@CR14],[@CR19],[@CR20]^. Interestingly, some miRNA-mediated translational repression is reversible^[@CR21]^.
siRNAs are small dsRNAs, 20--24 nt in length, that are processed from longer dsRNAs ([Fig. 1](#Fig1){ref-type="fig"}). One strand is the \'guide\' strand and directs silencing, with the other strand --- the \'passenger\' --- being degraded^[@CR22],[@CR23]^. Which strand becomes which is determined by the thermodynamic properties of the duplex^[@CR24],[@CR25],[@CR26],[@CR27]^. siRNAs generally show full complementarity to their target mRNA, and cleavage occurs 10--12 bases from the 5′ end of the guide strand binding site^[@CR3],[@CR28]^.
***Exogenous inhibitory RNAs***. Our understanding of small RNA biogenesis has enabled the development of several strategies for harnessing RNAi pathways for therapy. Recombinant inhibitory RNAs are designed to mimic [primary miRNAs](#Glos3){ref-type="list"} (pri-miRNAs) (in the case of artificial miRNAs or exogenous miRNAs) or [precursor miRNAs](#Glos4){ref-type="list"} (pre-miRNAs) (in the case of short hairpin RNAs (shRNAs)), whereas chemically synthesized RNA oligonucleotides are designed to mimic [Dicer](#Glos5){ref-type="list"} products or substrates. Each class mediates gene silencing but enters the pathway at a different step ([Fig. 1](#Fig1){ref-type="fig"}). The main differences between exogenously applied oligonucleotide siRNAs and hairpin-based species (shRNA or miRNA shuttles) are the mode of delivery and the duration of gene silencing ([Table 2](#Tab2){ref-type="table"}). However, recent advances in non-viral and viral systems are blurring this distinction. In the following sections we describe the main strategies for the design and delivery of inhibitory RNAs.Table 2Methods for the delivery of RNAi triggers to cells and tissuesSpecies/formulationPackaging capacityApplications and considerationsRefs^\*^***Viral vector***AdenovirusUp to \~35 kb, usually \<10 kbdsDNA vector with large packaging capacity, transient expression, highly immunogenic[@CR76],[@CR77]Adeno-associated virus (AAV)\~4.5 kbssDNA vector, small packaging capacity, mildly immunogenic, lasting expression in nondividing cells, capsid pseudotyping/engineering facilitates specific cell-targeting[@CR82],[@CR91],[@CR103],[@CR108]LentivirusUp to 13.5 kb (larger inserts will decrease titre)RNA vector, integration competent and incompetent forms available, less immunogenic than adenovirus or AAV, envelope pseudotyping facilitates cell targeting, clinical production more difficult than for adenovirus or AAV[@CR83],[@CR84],[@CR85],[@CR86],[@CR87],[@CR88],[@CR140],[@CR155]Herpes simplex virus150kbDNA vector, episomal, lasting expression, immunogenic[@CR119]***Bacterial vector species*** ^***‡***^*Escherichia coli*, S. Typhymurium^§^Delivery of short hairpin RNA or small interfering RNA to gut tissue[@CR73],[@CR74],[@CR75]***Non-viral formulations*** ^\|\|^NanoparticleSelf-assembling, may target specific receptors, requires technical expertise to prepare[@CR59]Stable nucleic acid lipid particle (SNALP)Stable for systemic delivery, broad cell-type delivery[@CR51]AptamerTargeting of specific receptors, requires sophisticated screening to develop[@CR53]CholesterolStable for systemic delivery, broad cell-type delivery[@CR46]^\*^Representative references.^‡^Bacterial minicells can carry plasmids, short interfering RNAs or drugs.^§^*Salmonella enterica* subsp. *enterica* serovar Typhimurium.^\|\|^The nucleic acids in non-viral carriers can be any size from small oligonucleotides to large artificial chromosomes.
**siRNA approaches**
The most common method used to harness the RNAi pathway for targeted gene silencing is to transfect 21--22 nt siRNAs into cells. Another option is to use longer, 25--27 nt duplexes that can be processed by Dicer into siRNAs; these are called \'Dicer-ready siRNAs\'. In some cases, the silencing potency of Dicer-ready siRNAs can be greater than for siRNAs^[@CR29],[@CR30]^. For both synthetic triggers, transfection is generally accomplished to high efficiency in cell lines using commercially available transfection reagents. However, as discussed below, alternative packaging is often required for delivery to primary cells and for *in vivo* applications.
In rational siRNA design it is important to consider the siRNA sequence, the chemical nature of the silencing moiety (for example, RNA with or without modified bases and sugars), the length of the RNA and the nature of the 5′ and 3′ ends. *In vitro* synthesis of siRNAs using T7 polymerase creates 5′ triphosphates, which can induce [type I interferon responses](#Glos6){ref-type="list"} (type I IFN responses)^[@CR1]^. Similarly, blunt-ended siRNAs induce cytoplasmic retinoic acid inducible gene 1 protein (RIG1; also known as DDX58) and IFN production. Chemically synthesized siRNAs lacking 5′ triphosphates and containing appropriate 3′ overhangs alleviate these issues.
Many siRNAs, although able to reduce expression of the target gene, are immunostimulatory in a sequence-independent manner because they are recognized by the pattern recognition [Toll-like receptors](#Glos7){ref-type="list"} (TLRs)^[@CR31]^. TLR3, which can be endosomal or on the cell surface, recognizes dsRNAs and can be activated by uncomplexed 21-mer siRNAs. TLR3 activation inhibits blood and lymphatic vessel growth^[@CR32],[@CR33]^, which can be advantageous in the setting of corneal vascularization, where inhibition of angiogenesis is desired^[@CR34]^. TLR activation can also be advantageous in cancer therapies by stimulating dendritic cells to respond immunologically to cancer cells^[@CR35]^. Recently, the generally unwanted stimulation of TLRs by oligonucleotides was used cleverly to achieve gene silencing and immune stimulation for cancer therapy: well-characterized [CpG oligonucleotide](#Glos8){ref-type="list"} agonists of TLR9 fused to siRNAs targeting an immune suppressor promoted antitumour immune responses in mice^[@CR36]^. This interesting combination of tumour targeting and siRNA immunostimulatory therapy may substantially augment the promising clinical results from the use of TLR9 agonists alone^[@CR37]^.
TLR activation would be contraindicated in other settings, such as in attempts to revert or inhibit ischaemia. In addition, endosomal TLR7 and TLR8 recognize ssRNAs and can be activated when siRNAs complexed with carriers are internalized or taken up through receptor targeting. Altering the chemical nature of the siRNA dramatically reduces TLR responses elicited by exogenously applied siRNAs. For example, using 2′-O-methyl-modified purine nucleosides in the passenger strand reduces IFN induction but retains targeting specificity^[@CR38],[@CR39],[@CR40]^. This modification also improves serum stability by reducing susceptibility to RNases^[@CR41]^.
***siRNA delivery options***. Chemically modified siRNAs are most often packaged into carriers for systemic delivery as their negative charge and size prevent cellular penetration. Uncomplexed siRNAs that are delivered systemically are also readily cleared by the kidney and excreted. The array of carriers is vast, and excellent summaries of their chemical make-up and biological properties can be found elsewhere^[@CR42],[@CR43],[@CR44],[@CR45]^. Among the most common are lipid-based carriers or cholesterol conjugates to the sense strand of the duplex. Cholesterol-conjugated siRNAs, which are commercially available, enable improved uptake to the liver as they are bound by low-density lipoprotein (LDL) in serum and LDL uptake in the liver is robust^[@CR46]^. Lipophilic siRNAs can also bind high-density lipoprotein (HDL); this can target siRNAs to tissues with HDL receptors, such as gut^[@CR47]^, kidney^[@CR47]^ and vaginal^[@CR48]^ epithelial cells and oligodendrocytes in the brain^[@CR49]^.
Exciting data in nonhuman primates showed that a single delivery of siRNAs complexed into [stable nucleic acid lipid particles](#Glos9){ref-type="list"} (SNALPs) reduced target gene expression for almost 2 weeks^[@CR50]^. Recently the same delivery strategy was used successfully to protect nonhuman primates from a lethal challenge of Ebola virus, using siRNAs targeting the expression of three Ebola virus proteins^[@CR51]^. Improvements in SNALPs that reduce the doses required for effective silencing in nonhuman primates by \~tenfold will augment the clinical utility of these reagents^[@CR40]^.
Complexing siRNAs with carriers also provides opportunities for targeting specific cells or, in the case of cancer, tumour beds. In the first in-human study, nanoparticles designed for enhanced uptake to cancer cells by using transferrin-receptor-targeting ligands showed reduced levels of the target mRNA and evidence for siRNA-mediated cleavage of that target^[@CR52]^. As the study continues, it will be interesting to learn the pharmacokinetics of the siRNAs in additional patients and to see the clinical effects of the therapy. Other carriers for tissue-specific targeting include aptamers^[@CR53],[@CR54]^, antibodies^[@CR55],[@CR56],[@CR57],[@CR58]^, peptides and proteins^[@CR52],[@CR59]^, and oligonucleotide agonists^[@CR36]^. Some are synthetically linked to the siRNAs, as in the case of oligonucleotides, peptides and aptamers, whereas others are part of more complex carrier systems (reviewed in Refs [@CR42],[@CR60]).
**Expression systems for RNAi triggers**
***shRNAs and artificial miRNAs***. Expression of shRNAs or artificial miRNAs is achieved by delivering plasmids or by using bacterial or viral vectors. These RNAi triggers are transcribed as sense and antisense sequences connected by a loop of unpaired nucleotides to mimic pre-miRNAs (for shRNAs) or pri-miRNAs (for artificial miRNAs). Following transcription, artificial miRNAs are processed by the Drosha--DGCR8 complex and exported to the cytoplasm, where they engage the pre-RISC complex via the normal miRNA biogenesis pathway ([Fig. 1](#Fig1){ref-type="fig"}). After cleavage of the passenger (non-guide) strand, the guide strand directs RISC to the mRNA target.
Although the goal for shRNAs is to make a pre-miRNA mimic, shRNA transcripts often do not reflect Drosha cleavage products^[@CR61],[@CR62]^. If they do not contain the typical 3′ dinucleotide overhang, transport by exportin 5 will be impaired. Reduced shRNA export could cause nuclear accumulation and toxicity^[@CR63],[@CR64]^. Alternatively, if exportin 5 recognizes the duplexes but shRNA expression is exceptionally high relative to endogenous miRNAs, exportin 5 can become saturated. Overexpression of exportin 5 can alleviate the blockage, but saturation of downstream processing pathways can occur. One method to alleviate saturation of exportin 5 and Argonaute proteins is to overexpress them concomitantly with shRNAs^[@CR65]^. However, for therapeutic purposes it might be more appropriate to use weaker promoters^[@CR61],[@CR66]^ or artificial miRNAs^[@CR64],[@CR67],[@CR68],[@CR69],[@CR70]^. As miRNA hairpins can be embedded in larger transcripts, the artificial miRNA approach is more suited than shRNAs to RNA polymerase II-based expression systems that provide tissue-specific and inducible gene silencing. The products of RNAi expression systems need to be assessed carefully to ascertain whether the desired strand is loaded into RISC and how well the RNAs are expressed and processed ([Fig. 2](#Fig2){ref-type="fig"}).Figure 2Workflow for testing therapeutic RNAi triggers.**a** \| Cartoon depicting a luciferase reporter system that is used to confirm that the appropriate strand of small interfering RNAs (siRNAs) or stem--loop platforms from RNAi expression systems is loaded into the RNA-induced silencing complex (RISC). A plasmid with a luciferase reporter that harbours sequences complementary to the guide strand in the 3′ UTR is cotransfected with the RNAi system, and if the appropriate guide strand is loaded, luciferase activity will diminish. When a reporter that contains sequences complementary to the passenger strand is cotransfected, luciferase activity should not be reduced. Because silencing is based on a microRNA (miRNA)-like mechanism, inhibition of luciferase activity will indicate RISC loading, independent of the sequence\'s ability to induce target cleavage. **b** \| Northern blot analysis can be used to evaluate RNAi triggers expressed from vectors. If the short hairpin RNA (shRNA) or primary miRNA (pri-miRNA) mimics are poorly processed but expressed efficiently, build-up of shRNAs may occur (lane 1). Appropriate processing should yield readily detectable mature, processed siRNAs with minimal levels of unprocessed material (lane 2). Northern blots with probes for the passenger strand can also be used to assess RISC loading of the unintended strand (not shown). **c** \| Small RNA quantitative PCR to quantify the mature product will yield information about overall levels of mature product, which is important to know to understand dosing. The figure shows an example of results obtained from effectively or poorly processed RNA precursors. Cloning and sequencing of the mature small RNAs can be used to assess the silencing RNAs in more detail (not shown). Ct, threshold cycle.
***Delivery systems: non-viral vectors***. RNAi trigger-expressing plasmids can be packaged into many of the same carriers that can be used for siRNA delivery, although the nature of the particle will change with the different cargo (large DNAs versus small RNAs). Several non-viral platforms for gene delivery are being investigated (reviewed in Ref. [@CR71]) and, as for siRNAs, commercial reagents are available for plasmid transfection of cell lines and some primary cells *in vitro*.
Bacteria can be used as an innovative platform for RNAi delivery; this approach is built on earlier work showing that therapeutic bacteria can enter tumours in patients with cancer^[@CR72]^, presumably via a permissive vasculature. The basic premise is that recombinantly engineered *Escherichia coli* can enter mammalian cells after *in vivo* delivery and transfer shRNAs. When bacteria containing plasmids that expressed shRNAs targeting β-catenin were fed to mice, β-catenin expression was reduced in the intestinal epithelium^[@CR73]^. This platform is now in clinical testing for familial adenomatous polyposis, an inherited form of colon cancer^[@CR74]^. A related approach in mice used [minicells](#Glos10){ref-type="list"} derived from *Salmonella enterica* subsp. *enterica* serovar Typhimurium and targeted them to tumour-cell-surface receptors to deliver siRNAs or shRNAs; this method reduced tumour burden and improved mouse survival^[@CR75]^.
***Delivery systems: viral vectors***. The viral vectors used to deliver shRNAs or artificial miRNAs include murine oncoretroviruses, lentiviruses, adenoviruses, [adeno-associated viruses](#Glos11){ref-type="list"} (AAVs), and herpesviruses, among others ([Table 2](#Tab2){ref-type="table"}). We refer the reader to recent reviews describing the biology and production of these vector systems^[@CR76],[@CR77],[@CR78],[@CR79],[@CR80]^. Here we highlight major differences among some of the tools that are relevant to their use for RNAi-based applications, with respect to their tissue tropisms and the fate of their recombinant genomes in host cells.
The genomes of recombinant adenoviruses and AAVs generally remain [episomal](#Glos12){ref-type="list"} after the virus has entered the host cell, the viral [capsid](#Glos13){ref-type="list"} has been uncoated and the nucleocapsid has been transported to the cell nucleus. An advantage of vector systems with genomes that remain episomal is that insertional mutagenesis is avoided. However, unless only transient expression is desired, their use is limited to cells that divide slowly (for example, some epithelial cells) or not at all (for example, neurons) because the episomal genome will be lost following cell divisions. Naturally occurring differences in capsid structures among adenoviruses or among AAVs affect the ability of the viruses to infect diverse cell types *in vitro* or specific cells within tissues. Capsid genes can also be manipulated to artificially alter tropism to a cell surface molecule or tissue of choice^[@CR81],[@CR82]^. Such capsid retargeting takes advantage of the fact that once binding is established, viruses can use secondary receptors for internalization.
A major difference between adenoviruses and AAVs is their packaging capacity (\~4.5 kb for AAVs compared with up to 35 kb (although usually less than 10 kb) for adenoviruses)). Another distinguishing point is that all viral genes are removed from AAV vector genomes, whereas recombinant adenoviruses often express many viral genes that may induce immune responses to transduced cells and cause their subsequent elimination. For these reasons, AAVs are generally useful for achieving RNAi in cells that one wants to save, whereas adenoviruses are useful tools for either transient expression or when immune induction is desired. Notably, both adenoviruses and AAVs infect cells at multiple copies per cell, which can be problematic with shRNAs (instead of artificial miRNAs), as dosing may be amplified. Dosing is a function of the copy number and how efficiently the hairpin is expressed and processed^[@CR62]^ and, in many cases, higher expression is not necessarily beneficial. If hairpins are inappropriately processed, or expressed at very high levels, toxicity ([Box 1](#Sec2){ref-type="sec"}) and/or saturation of the RNAi machinery can occur^[@CR63],[@CR64]^ (see above).
Lentiviruses are another delivery option. If their RNA genomes contain hairpins, they can be cleaved by RNA processing enzymes, which can be a problem during vector production. However, the negative effect this has on vector titres can be rescued by inhibiting the RNAi pathway^[@CR83]^. Expression cassette placement is also important for lentiviruses expressing shRNAs or artificial miRNAs, as some configurations negatively affect vector production or shRNA expression^[@CR84],[@CR85],[@CR86]^. Recombinant lentiviruses can transduce dividing and non-dividing cells and generally integrate into transcriptionally active chromatin. A potential problem is that insertion could activate an oncogene or inactivate a tumour suppressor gene. Insertional promiscuity can be dramatically reduced by using integrase-defective lentiviruses^[@CR87]^. Like adenoviruses and AAVs, the vector tropism of lentiviruses can be altered, in this case through manipulation of the viral envelope used in vector production^[@CR88]^.
The most obvious application for recombinant lentiviruses is for transduction *in vitro* or, in the case of clinical applications, *ex vivo* gene transfer to haematopoietic progenitor cells or peripheral blood lymphocytes for treatment of viral infections^[@CR89]^. In these settings, recombinant lentiviruses are applied to cells at 1 to 5 vectors per cell, because higher concentrations are often toxic. The final number of integrants per cell is approximately 0.5 to 1 copy per cell. This fact is important when considering lentiviruses for RNAi delivery; if the copy number is low, shRNA expression from strong promoters may be preferred.
**Therapeutic applications of RNAi** ***in vivo***
Important considerations for therapeutic RNAi are that gene silencing approaches rarely remove 100% of a transcript, that [off-target](#Glos14){ref-type="list"} silencing can occur ([Box 1](#Sec2){ref-type="sec"}) and that each target organ, cell type and target transcript presents unique challenges. In some cases, the goal is to target every cell in an organ, in other instances promiscuous cell tropism is disadvantageous. For example, one might wish to target cancer cells for gene silencing but avoid normal surrounding tissue, or to express the therapeutic RNA in hepatocytes but not [Küppfer cells](#Glos15){ref-type="list"} after delivery to the liver. Tissue specificity can be achieved in some cases by incorporating ligands on the carrier that direct transfection or infection to the desired cell, as described above. Alternatively, researchers have taken advantage of the natural tropism^[@CR90],[@CR91]^ or modified tropism of viral vectors for cell and tissue targeting^[@CR82]^, or have used cell-specific promoters to express the silencing RNAs.
***Delivery to the respiratory tract***. Alterations in gene expression in epithelial cells of the respiratory tract contribute to disease pathogenesis in many disorders including asthma, chronic obstructive airway disease and cystic fibrosis. In addition, these cells are a key site of interaction between the host and the environment and many common viral pathogens replicate in these cells as the initial step in their life cycle, providing opportunities to silence viral gene products or host genes that modify the viral life cycle or the host\'s response to that virus. These reasons, and the fact that the respiratory tract is an accessible tissue, make the airway epithelium an attractive tissue for exploring RNAi therapies.
The potential of RNAi-based treatments to modify the host response to respiratory virus infections has been extensively studied. In 2004 and 2005 several high-profile papers were published in which synthesized siRNAs or expressed shRNAs were used to inhibit the influenza A virus^[@CR92],[@CR93]^, severe acute respiratory syndrome (SARS) coronavirus^[@CR94]^, respiratory syncytial virus^[@CR4],[@CR94]^ and parainfluenza virus in animal models^[@CR4]^. These studies showed both the promise and potential pitfalls of RNAi as a therapeutic strategy. Although the respiratory tract is readily accessible using topical or aerosol delivery techniques available in the clinic, several years of basic and clinical studies in the field of gene therapy have humbled investigators. These studies used a number of well-conceived viral and non-viral delivery techniques to treat monogenetic disorders such as cystic fibrosis, but the treatments were hindered by the physical barriers posed by the epithelial cells themselves, in addition to their secretions and host defence mechanisms^[@CR95],[@CR96]^. In principle, delivery of RNAi oligonucleotides might pose a less significant challenge than expression plasmids or other genetic payloads as oligonucleotides need only enter the cytoplasm of surface cells to function. In addition, the mucosal surface of the airways and alveoli are active sites of innate and adaptive immunity, and RNAi delivery vectors or RNAi oligonucleotides may elicit immunologic responses. In the context of anti-infection RNAi strategies, immunostimulation confounded the early preclinical results^[@CR5],[@CR6],[@CR32]^, and possibly the ongoing clinical data^[@CR97]^.
In addition to targeting the gene products of respiratory viruses, several groups have used siRNA technology to knock down expression of host gene products or reporter genes in the respiratory tract in animal models. To date, the animal studies have shown little efficacy^[@CR95]^ and poor delivery has also been demonstrated *in vitro* when fully differentiated cell models were studied. In short, effective RNAi activity requires the application of siRNAs before the development of a well-differentiated epithelial barrier^[@CR96]^. Thus, efficient delivery remains an important hurdle to overcome as clinical studies are developed.
***Antiviral strategies in other tissues***. HIV remains an attractive target for drug development, including for therapies based on RNAi^[@CR89]^. Examples of strategies include targeting the receptor for the virus and the virus itself^[@CR55],[@CR98]^. However, the error-prone replication cycles of HIV can be problematic. Combinatorial approaches that include RNAi and other gene silencing approaches have therefore been developed for HIV. An approach undergoing clinical testing uses lentiviral vectors expressing an shRNA targeting an exon shared by HIV *tat* and *rev* genes (the *tat*/*rev* common exon), combined with two HIV-specific RNA-based inhibitors (a nucleolar-localizing TAR RNA decoy and a C-C chemokine receptor type 5 (CCR5)-targeting hammerhead ribozyme)^[@CR99]^. The strategy is to transduce haematopoietic progenitor cells *ex vivo* and then reinfuse them into patients^[@CR99]^. Early data from this Phase I trial show that transduced cells successfully engrafted within 11 days in all four patients treated. Importantly, there were no treatment-related toxicities. Vector expression was documented for up to 24 months in multiple cell lineages, as was expression of the introduced ribozyme and shRNA.
RNAi was recently used to inhibit lethal infection by the filovirus Ebola in a primate model^[@CR51]^. A combination of modified siRNAs targeting Ebola L polymerase, viral protein 24 (VP24) and VP35 were SNALP-formulated and delivered intravenously. This strategy protected animals from death, including those that received siRNAs only after the onset of the infection. RNAi-based therapies are also under development for hepatitis B virus (HBV)^[@CR39],[@CR100]^ and hepatitis C virus (HCV)^[@CR101]^. As chronic hepatitis contributes significantly to hepatocellular carcinoma pathogenesis, this further drives interest in new HBV and HCV therapies as a means to reduce disease burden. As with HIV, the hepatitis viruses have a high mutation frequency during viral replication. Therefore, most current antiviral strategies focus on the delivery or expression of more than one RNAi construct to achieve success against chronic hepatitis infection^[@CR101],[@CR102]^. A current focus of several laboratories is to use miRNA or shRNA expression methods to target more than one viral transcript^[@CR103],[@CR104]^. Host proteins can also be targeted, for example siRNAs directed to diacylglycerol acyltransferase 1 (DGAT1) can reduce HCV virion production^[@CR105]^. Additionally, inhibiting the expression of the host gene product polo-like kinase 1 (PLK1) using siRNA reduces HCV replication^[@CR106]^. PLK1 inhibition is also being used in anticancer studies (see below).
Promising results have also been reported using RNAi to modify virus and host gene expression in a mouse model of genital tract herpes simplex virus 2 (HSV-2) infections^[@CR48],[@CR107]^. Lipid-complexed RNA oligonucleotides inhibited expression of the HSV-2 *UL27* and *UL29* genes and the host receptor nectin 1 (also known as PVRL1). These approaches showed efficacy in both the prevention and treatment of infection^[@CR48],[@CR107]^. The manipulation of host miRNAs to inhibit viral expression is also being explored (see further discussion below).
***RNAi for neurological disorders***. The blood--brain barrier limits access to the central nervous system (CNS) and thus the most practical manner to silence targets in neural cells is through direct injection of the RNAi trigger. As siRNAs have a short half-life, redosing using indwelling catheters would be required for chronic diseases. However, for acute illnesses or delivery to brain tumours, the short half-life of siRNAs may be desirable. By contrast, viral platforms provide lasting expression and may be ideal for chronic disorders. For example, vectors expressing therapeutic RNAi improved disease phenotypes for many months in preclinical studies in rodent models of polyglutamine repeat diseases^[@CR108],[@CR109],[@CR110],[@CR111],[@CR112],[@CR113]^, amyotrophic lateral sclerosis^[@CR114],[@CR115],[@CR116]^, Parkinson\'s disease^[@CR117]^ and Alzheimer\'s disease^[@CR118],[@CR119]^. In nonhuman primate brains, viral-vector-based systems are safe^[@CR120]^ and, given the encouraging results of AAVs in the human brain and eye^[@CR121]^, clinical trials for neurodegenerative diseases with AAVs expressing RNAi triggers are anticipated.
An issue to consider is the delivery to the correct cell type in the brain. Specificity for neurons versus glia has not been achieved for uncomplexed siRNAs, and in fact it has been reported that siRNAs delivered into nonhuman primate brains enter oligodendrocytes^[@CR49]^. This might be a suitable method for therapies aimed at treating multiple sclerosis or other white matter diseases. For encapsidated viral vectors, the nature of the protein coat imparts a natural tropism for neurons (in the case of AAV2, for example) or for other cell types^[@CR91]^. For enveloped viral vectors, such as those lentiviral systems derived from HIV, the tropism is imparted by the envelope used in vector production in a process known as pseudotyping^[@CR122]^. Also, some vectors traffic from the site of injection to other regions of the brain via neuronal connections, but others remain localized. Thus targeting is achieved collectively by the site of injection, the propensity of the virus to infect certain cell types and traffic along or within neuronal axons and neurites, and the promoter used to drive expression of the RNAi trigger.
For chronic, dominantly inherited disorders, it may be preferable to silence only the mutant allele. For some brain diseases there are highly prevalent disease-linked polymorphisms that provide opportunities for allele-specific silencing. Primary dystonia, which is caused by a common GAG deletion in torsin A (*TOR1A*), is one example^[@CR123]^. Other disorders have several disease-linked SNPs present in most patients and these SNPs provide an opportunity for disease allele silencing. However, when targeting a SNP using RNAi, it is critical to consider the potential for off-target binding of the seed sequence in the small RNA ([Box 1](#Sec2){ref-type="sec"}), as unintended off-target silencing could abrogate any beneficial effects from leaving the wild-type allele intact. Huntington\'s disease is a case in point. Preclinical work in animal models shows that partial knockdown of both alleles of huntingtin is tolerated and provides clinical benefit^[@CR124]^, yet genotyping shows that 4 to 5 SNPs in huntingtin may be present in the majority of patients with Huntington\'s disease^[@CR125],[@CR126]^. Small RNAs that are specific to the SNPs may have moderately high off-target potential and therefore should be tested for their long-term tolerability *in vivo*, as should the safety of partial reductions in expression from both mutant and wild-type alleles.
***Targeting metabolic disease and hepatic cancers***. One of the first organs tested for the effectiveness of RNAi *in vivo* was the liver^[@CR127],[@CR128]^, and RNAi-based treatments for metabolic diseases (such as hypercholesterolaemia), viral infections, cancer^[@CR6]^ and liver fibrosis (reviewed in Ref. [@CR129]) are in progress. For metabolic diseases, there are preclinical and clinical trials underway for lowering plasma LDLs using siRNAs that target the expression of apolipoprotein B (APOB) and proprotein convertase subtilisin/kexin type 9 (PCSK9). In this work, the siRNAs are complexed to carriers or embedded in liposomal particles (for example, SNALPs). Data from rodents and nonhuman primates^[@CR50],[@CR122]^ show significant LDL-lowering properties and, in one of the first trials in humans, corporate news releases stated that the SNALP-formulated APOB siRNA was well tolerated at all but the highest dose. Newer formulations that show improved potency in nonhuman primates are under development.
The liver was also one of the first organs targeted in the development of RNAi-based therapies for cancer. One study used SNALPs targeting PLK1, a cell cycle protein that is crucial for the activating phosphorylation of many cell cycle proteins; inhibition of PLK1 induces cell cycle arrest and tumour cell apoptosis^[@CR130]^. Mice with hepatic tumours treated with SNALP-formulated PLK1 siRNA showed significant improvements in survival^[@CR6]^. In December 2010, this technology advanced to a Phase I trial in humans with liver cancer ([Table 1](#Tab1){ref-type="table"}).
Another hepatic cancer application is SNALPs simultaneously delivering siRNAs to kinesin spindle protein (KSP) and vascular endothelial growth factor (VEGF). KSP is required for cell division, and VEGF is required for tumour cell growth. In early 2011, sponsors of a Phase I trial using this approach reported evidence of RNAi activity in biopsied tissue^[@CR131]^. These preliminary reports, along with the first report of RNAi activity from exogenously applied siRNA complexes^[@CR52]^, are important milestones in the development of RNAi delivery systems as cancer therapeutics.
***miRNAs as therapeutic targets***. The identification of misregulated miRNAs in cellular transformation and maintenance of the malignant state has profound implications for cancer therapy. As with other misregulated genes, miRNAs can be targets for gene silencing approaches, whether the miRNAs are encoded in the host genome or expressed from oncogenic viruses (reviewed in Ref. [@CR132]). Inhibition of oncogenic miRNAs that regulate multiple targets might switch off dozens of cancer-promoting signals. Rather than devising siRNAs to target the misregulated miRNA, researchers have developed miRNA sponges^[@CR133]^; these provide alternative binding platforms for the miRNAs and so inhibit their ability to bind and suppress their natural targets ([Fig. 1](#Fig1){ref-type="fig"}). An early example was the intravenous delivery of [antagomirs](#Glos16){ref-type="list"}, which are chemically modified RNA oligonucleotides antisense to the miRNAs^[@CR134],[@CR135]^. In a primate model of HCV infection, oligonucleotides that sequestered miR-122 inhibited virus replication^[@CR136]^. Plasmid- and virus-based approaches are also being used for reducing endogenous miRNA levels^[@CR133],[@CR137]^. Typically for this approach, strong promoters drive expression of a sequence encoding several miRNA target sites downstream of a reporter. The multiple copies expressed become targets for binding of miRNAs, which are sequestered from targeting their endogenous mRNAs.
**Summary and future considerations**
In addition to the developments described above, there has been substantial progress in using gene silencing approaches for treating skin^[@CR138]^ and retinal diseases^[@CR139],[@CR140]^. Like the liver and airway, these accessible tissues were early targets for preclinical testing. Exploiting the small RNA biogenesis and gene silencing pathways for heart diseases, either using siRNAs against single targets^[@CR141]^ or inhibiting the action of misregulated miRNAs^[@CR142]^, has also yielded promising results that are approaching clinical trials.
In addition to its utility as a stand-alone strategy, RNAi may have expanded applications as an adjuvant in multipronged treatment settings. For example, targeting multidrug resistance protein 1 (MDR1; also known as ABCB1) in cancer cells may enhance the activity of chemotherapeutics^[@CR143]^, and other host genes have been targeted for similar ends for cancer therapeutics^[@CR144],[@CR145],[@CR146]^. Another RNAi adjuvant strategy is the use of dsRNA oligonucleotides as immunostimulatory agonists alongside vaccines, as in the case of a RIG1 agonist to enhance the activity of a DNA vaccine against influenza^[@CR9]^.
RNAi therapy development should consider whether regional delivery and partial knockdown, or global delivery and complete knockdown, is required for a therapeutic result. An example of the former is directed delivery to specific regions of the brain for Parkinson\'s disease. By contrast, RNAi therapy for cancer may require delivery to all cancer cells. Another important issue that is yet to be resolved is dosing of the therapeutic RNAi. In the case of cancer, pharmacologists must balance the target cells\' ability to recover from the exogenously applied siRNAs or anti-miRNA treatments with the practical considerations of patient compliance for repetitive dosing. Will clinical success and eventual cure require treatment for weeks, months or years? Viral vector expression systems for RNAi can overcome this problem by providing sustained expression, but this strategy requires genomic integration of the vector if the target cells are dividing. In order to overcome the potential dangers of genomic integration, methods for integration into genomic \'safe harbours\' will be important. With respect to long-term RNAi from viral vectors as therapies for genetic diseases, the question is whether regulated expression is required. To help to answer this question, it is expected that long-term studies in large animals (for example, non-human primates) will yield valuable information regarding chronic application of inhibitory RNAs from various platforms. A further consideration is that although the end points for some trials are obvious, for example lowering blood cholesterol or reducing tumour burden, sensitive and specific end points are not always clear for chronic disorders in which the tissue cannot be easily biopsied or in which biomarkers are not validated.
In slightly more than a decade, we have advanced rapidly from RNAi discovery, to understanding the molecular processes driving small RNA biogenesis and function, to developing reagents that harness the power of the RNAi pathway. Although many hurdles remain for using these technologies for therapy, exciting early clinical results show how far we have come.
{#Sec2}
When designing therapeutic strategies involving small interfering RNAs (siRNAs) or expression-based systems, it is important to know details about the RNA strand that is incorporated into the RNA-induced silencing complex (RISC) and mediates silencing. Genomically encoded microRNAs (miRNAs) in mammalian cells possess the necessary information within the miRNA duplex for appropriate loading of the miRNA or the [miRNA\*](#Glos14){ref-type="list"} strand. So, when using chemically synthesized siRNAs or systems that express short hairpin RNAs or artificial miRNAs, the design must take into account what nature has taught us. For example, the designer should consider whether the artificial miRNA has the 5′ and 3′ ends that are necessary for appropriate processing and export, and whether subsequent biochemical processing results in loading of the correct strand into RISC.
The are several methods to determine which strand is loaded into RISC, including northern blots, small RNA PCRs and luciferase-based plasmid systems with targets for the guide strand or the passenger strand placed in the 3′ UTR of the reporter ([Fig. 2](#Fig2){ref-type="fig"}). Although there are publicly available algorithms that use thermodynamic rules in silencing RNA, it is important to sequence expressed constructs to characterize their termini. Sequencing is necessary to ascertain the relative proportions of the intended mature silencing RNA and RNAs that result when cleavage sites are shifted by one or several nucleotides; such alterations to the RNA could substantially increase off-target silencing. Off-target silencing occurs through the interaction of a seed sequence with transcripts harbouring complementary sequences^[@CR147],[@CR148],[@CR149],[@CR150],[@CR151]^. siRNAs with high off-target potentials, based on seed complement frequencies in the transcriptome, result in increased silencing of unintended mRNAs and toxicity^[@CR152]^. This problem can be reduced by designs that bias RISC loading towards the correct RNA strand and that ensure that processing of silencing RNA precursors only produces the desired small RNA duplex. Incorporating chemical modifications, such as 2′-O-methyl groups, into the guide and passenger strands of chemically synthesized siRNAs reduces indiscriminate effects of ssRNAs and dsRNAs^[@CR153]^. Additionally, incorporation of unlocked nucleic acids (acyclic RNA mimics)^[@CR154]^ makes the RNAs poor ligands for Toll-like receptors.
Varying amounts of off-target silencing can occur with different silencing platforms because the levels of the silencing RNAs will vary. For example, for expression systems, transfection with plasmids or transduction with adenoviruses or adeno-associated viruses will yield more copies per cell than transduction with lentiviruses. siRNA transfection can also result in abundant off-target silencing if the siRNA has low on-target potency and a moderate or high off-target potential (see above). Less off-target silencing with a lentivirus-based system compared with siRNAs was reported recently, but that study used high doses of siRNAs with high off-target potential^[@CR155]^.
Related links {#Sec3}
=============
FURTHER INFORMATION {#Sec4}
-------------------
[Beverly L. Davidson\'s homepage](http://www.medicine.uiowa.edu/labs/davidson)
[Paul B. McCray\'s homepage](http://mccraylab.genetics.uiowa.edu)
[ClinicalTrials.gov](http://clinicaltrials.gov)
The authors thank members of the Davidson and McCray laboratories for critical feedback. Work in the authors\' laboratories is supported by the Roy J. Carver Trust and the US National Institutes of Health grants HL51670, DK54759 and NS50210.
B.L.D. serves on the Scientific Advisory Board of Marina Biotech. P.B.M. declares no competing financial interests.
RNA-induced silencing complex
: (RISC). RISC is a group of proteins, including one of the Argonaute proteins, that induces target mRNA cleavage based on loaded small interfering RNA or microRNA guide strands.
Seed
: A sequence of seven bases in a microRNA that is complementary to the mRNA target. This sequence is essential for the initial binding of the microRNA to most targets. Seeds can also exist arbitrarily in small interfering RNAs and processed short hairpin RNAs, causing microRNA-like silencing.
Primary miRNAs
: (pri-miRNAs). The initial transcriptional products of microRNA genes. They are generally \>100 nucleotides long and may contain one or more microRNA stem loops that are processed by the microRNA biogenesis pathway.
Precursor miRNAs
: (pre-miRNAs). Hairpin precursors of microRNAs formed by the cleavage of primary microRNAs by DCGR8 and Drosha.
Dicer
: A member of the RNase III family of ribonucleases that cleaves dsRNAs into small interfering RNAs, and precursor microRNAs and mirtrons into microRNAs.
Type I interferon response
: An innate immune response to dsRNA, ssRNA, CpG DNA and other stimuli that triggers a protective antiviral response in host cells. Signalling elicits α- and β-interferon release, which activate multiple components of innate and adaptive immunity.
Toll-like receptors
: (TLRs). A family of receptors that recognize pathogen-associated molecular patterns (PAMPs), including some DNA and RNA molecules.
CpG oligonucleotides
: Immunostimulatory dinucleotide motifs that interact with Toll-like receptor 9.
Stable nucleic acid lipid particle
: (SNALP). A lipid nanoparticle formulation for the systemic delivery of small interfering RNAs to tissues.
Minicells
: Bacteria-derived cells that have no chromosomes and are non-living. They can be loaded with drugs, plasmids or small interfering RNAs.
Adeno-associated virus
: (AAV). A member of the genus *Dependovirus*. These viruses have small, ssDNA genomes and are not known to cause disease in humans. AAVs are commonly used as recombinant vectors in gene therapy applications.
Episome
: A dsDNA segment that can persist independently of chromosomal DNA.
Capsid
: The outer protein coat of a virus, such as adeno-associated virus or adenovirus.
miRNA\*
: The precursor microRNA (pre-miRNA) processed by Dicer generates an miRNA duplex containing the miRNA strand and the miRNA\* strand, one of which is loaded into the RNA-induced silencing complex (RISC). The ratio of one strand to the other being loaded into RISC to mediate silencing activity can vary among species, tissues, and disease or developmental settings.
Off-target effects
: Any detectible phenotypic change that is triggered by the RNAi treatment, other than those that are derived directly or indirectly from silencing the targeted mRNA.
Küppfer cell
: A macrophage-like cell that is resident in the liver and is involved in antigen presentation.
Antagomirs
: RNA oligonucleotides that are antisense to endogenous microRNAs. They are used for inhibiting microRNA--mRNA interactions.
| {
"pile_set_name": "PubMed Central"
} |
Background
==========
Accurate and timely diagnosis of malaria infections is a critical part of case management in malaria control programmes aiming to reduce malaria morbidity and mortality. Sensitive and effective detection of infected individuals also plays a vital role in areas where transmission has been reduced markedly due to intensified control measures and elimination is being considered.
Malaria diagnosis is traditionally achieved by microscopic examination of blood smears. Microscopy is able to detect parasite species and determine parasite densities. The limit of detection by thick film microscopy is in the range of 5 to 100 parasites/μL of blood \[[@B1]-[@B3]\]. However, the quality of microscopy can vary significantly \[[@B1],[@B4]\] because its accuracy largely relies on the experience and training of the microscopists to make and stain a blood slide correctly and read it accurately.
Malaria rapid diagnostic tests (RDTs) have been developed and tested over the past 2 decades as an alternative to microscopy, particularly for areas where quality microscopy is absent or hard to maintain. RDTs are lateral flow devices that detect parasite proteins using antibodies. The tests are easy to perform and provide rapid results in 15 to 20 minutes without the need for electricity, expensive equipment or extensive training. The limit of detection, by good quality RDTs, is similar to that of thick film microscopy for *P. falciparum*but often poorer for *P. vivax*\[[@B1]\]. Today, over 50 brands of malaria RDTs are manufactured, and over 150 individual products commercially available.
Polymerase chain reaction (PCR) is a DNA-based molecular detection method that is more sensitive than microscopy and RDT, and has been widely used for diagnosis, confirmation of diagnosis, epidemiology studies and drug efficacy assessment. In theory, PCR is capable of detecting a single parasite in a blood sample, and its sensitivity often is only limited by the volume of the blood. PCR provides accurate determination of parasite species, better sensitivity in detecting low density of parasites and better detection of mixed species/strain infections \[[@B2],[@B5],[@B6]\]. In the field, PCR has been mostly used in epidemiological studies instead of point of care diagnosis due to its requirement for sophisticated equipment, reagents and several hours of turn over time.
In most malaria control settings, where the goal is to significantly reduce malaria morbidity and mortality, quality assured microscopy and RDTs have been shown to effectively detect parasites in the majority of symptomatic patients and thus guide treatment. In contrast, for malaria elimination settings it is critical to detect all infections, including those with low and sub-microscopic parasite densities in asymptomatic carriers as they represent a parasite reservoir in the community capable of effectively transmitting infections to mosquitoes \[[@B7],[@B8]\] and seeding transmission foci \[[@B9]\].
It is well known that malaria epidemiology varies between country and region, and particularly between islands, because the dominant vector species, the characteristics of human populations and factors that influence transmission such as rainfall, temperature, housing conditions and population movement differ. Therefore, the challenges to malaria elimination in different settings will vary. Each area needs to investigate the malaria epidemiology and carefully tailor its diagnosis strategy to the local context.
Temotu Province in Solomon Islands is preparing for malaria elimination. An assessment of the epidemiological characteristics of malaria infections in the island population, particularly the prevalence and distribution of asymptomatic infections, and infections with low or sub-microscopic parasite densities will contribute to the understanding of the requirements of diagnostics in malaria elimination. A baseline malaria parasitological survey was conducted in Temotu Province in October-November 2008 as the first step in a provincial malaria elimination programme. This baseline survey provided opportunities to obtain point prevalence and epidemiological characteristics of malaria infections, as well as to assess how well different diagnostic methods performed in this particular setting. During the survey, blood samples collected from the population were examined by microscopy, with a subsample examined by PCR and RDTs. The overall malaria point prevalence was estimated at 2-3% with predominant *P. vivax*infections \[[@B10]\]. Of concern for diagnosis was the high prevalence of asymptomatic infections (86%) reported in this relatively low transmission setting.
To better understand malaria epidemiology in the Province, the basis of the observed low disease rate and to better evaluate the performance of different diagnostic methods, a detailed comparison of the microscopy, RDT and PCR was undertaken, including re-examination of the blood smears from discrepant samples and the determination of parasite densities. The results revealed that a large proportion of malaria infections in the province were at low and sub-microscopic parasite densities. The finding has an important practical implication to the malaria elimination strategy in the province.
Methods
=======
Survey site and sample collection
---------------------------------
The survey was conducted by the Pacific Malaria Initiative Survey team in the Temotu Province, a remote island group in the Solomon Islands, approximately 166°E, 11°S with a population of approximately 17,000. The survey site, consent process and sample collection methods are as previously described \[[@B10]\].
Preparation of blood films, microscopic examination and quality assurance
-------------------------------------------------------------------------
Thick and thin blood films from finger prick blood of 9,491 survey participants were prepared, air dried, stained and examined as reported earlier \[[@B10]\]. All slides which were considered positive on the initial examination, negative on the initial examination but positive by PCR and a 10% sample of initial microscopy negative samples, were subjected to a second quality assurance (QA) examination (of 300 fields in the thick film). All slides which recorded discrepant results between the initial examination and the QA examination were then subjected to a third \"referee\" examination (of 500 fields in the thick film), which was recorded as the final microscopy result. Identification of parasite species was performed by reference to the thin smears where parasites densities were sufficiently high enough. Otherwise, parasite species identification was made by reference to the thick smear only.
Malaria microscopists who participated in this survey were all certified to WHO malaria microscopy standards by prior participation and assessment in WHO accredited malaria microscopy competency assessment courses in Brisbane or Honiara. Microscopists who performed the initial slide examinations were all certified to competency levels 1, 2 or 3. Microscopists who performed the QA examination were all certified to 1 or 2, and microscopists who performed the third \"referee\" examinations were certified to competency level 1.
Parasite density counts
-----------------------
Parasite densities were determined by counting the number of parasites against 500 white blood cells (WBC) and calculated assuming 8,000 WBC per μL of blood.
PCR assays
----------
Genomic DNA was extracted from a subset of 1,784 filter paper samples. These included samples that were microscopy positive, samples from all febrile patients and 10% of the microscopy negative samples from each village. Following microscopy QA, some of the samples initially included as microscopy positive were reclassified as microscopy negative. This increased the percentage of negative samples examined to 15.9%. The DNA was extracted using QIAamp DNA Mini Kits and a QIAcube robot (QIAGEN, Crawley, UK), and eluted to a 100 μl volume. A multiplex PCR was performed to determine parasite species using published primers and PCR conditions \[[@B11]\] with the number of cycles increased from 43 to 45. The positive results were confirmed using a second round *P. falciparum*and *P. vivax*specific PCR.
RDTs
----
ICT Malaria Combo Cassette Test ML02 (ICT diagnostics) was used as per the manufacturer\'s instructions when a participant was febrile and on a subset of 400 children from villages around Lata, the Provincial capital. For QA, a post purchasing sample of these RDTs were tested at the Research Institute for Tropical Medicine, Manila, Philippines, a WHO-FIND lot testing laboratory and were found suitable. All RDTs were used within one month of the commencement of the survey.
Results
=======
Malaria prevalence determined by microscopy
-------------------------------------------
A total of 9,491 blood samples were collected and examined by microscopy. After the initial microscopy and two rounds of microscopy QA, 256 samples were determined positive giving an overall point prevalence of 2.7%. *P. falciparum*and *P. vivax*made up 17.5% (45/256) and 82.4% (211/256) of positive samples respectively. Two of the 256 positive samples were excluded from density related analyses because parasite densities were not determined due to poor quality of the slides.
A low percentage of microscopy positive subjects were symptomatic
-----------------------------------------------------------------
The proportion of microscopy positive individuals having an aural temperature ≥38°C at the time of examination was 17.8% (8/45) for *P. falciparum*and 2.9% (6/209) for *P. vivax*, respectively, giving an overall fever rate of 5.5% (14/254). The majority of fevers, 50.0% (4/8) and 66.7% (4/6) for *P. falciparum*and *P. vivax*, were observed in the 5-14 year age group. Parasite densities in six of the eight *P. falciparum*-infected fever subjects were \>500/μL. In contrast, 66.7% of *P. vivax*infected fever subjects had parasite densities \<100/ μL.
A large proportion of microscopy positive subjects had low parasite densities
-----------------------------------------------------------------------------
Parasite densities were determined for 254 of the 256 microscopy positive subjects and divided into three groups: \<100, 100-500 and \>500/μL. Overall, 61.0% (155/254) of these subjects, including 40.0% (18/45) of the *P. falciparum*and 65.6% (137/209) of the *P. vivax*infected subjects, had parasite densities \< 100/μL (Figure [1](#F1){ref-type="fig"}). The proportion of subjects with parasite densities \< 100/μL increased with age for *P. vivax*infections: 55.6% in the \< 5 year-old group and reaching 72.6% in the ≥15 year-old age group (Figure [1B](#F1){ref-type="fig"}). No age correlation was observed for the proportion of *P. falciparum*infections having low parasite densities (Figure [1A](#F1){ref-type="fig"}).
{#F1}
A large number of sub-microscopic infections were detected by PCR
-----------------------------------------------------------------
PCR was performed on a subset of 1,784 samples including microscopy positive (n = 256), febrile but microscopy negative (n = 59) and microscopy negative (n = 1469). Of the 256 microscopy positive samples, PCR detected *Plasmodium*DNA in 223 samples, including 55 (24.7%) *P. falciparum*, 139 (62.3%) *P. vivax*and 29 (13.0%) mixed *P. falciparum-P. vivax*. In the subset of 1,469 microscopy negative samples that were randomly selected from different villages, PCR detected 63 *P. falciparum*, 23 *P. vivax*and 10 mixed infections. These sub-microscopic infections, representing 18.6% of all observed infections, were distributed between 33 of the 43 villages where malaria infections were identified.
Malaria prevalence estimated by PCR
-----------------------------------
Assuming that the PCR positive rate (96/1,469) applies to the entire microscopy negative group, 603.5 samples would be expected to be PCR positive in 9,235 samples. Adding the 223 samples positive by both microscopy and PCR, the total number of PCR positive samples in 9,491 samples would be 826.5, giving an overall parasite prevalence of 8.7%, 3.3 fold higher than that determined by microscopy (Figure [2](#F2){ref-type="fig"}). The expected species distribution for *P. falciparum*, *P. vivax*and mixed infections would be 4.8%, 3.0% and 1.0%, respectively (Figure [2](#F2){ref-type="fig"}).
{#F2}
Comparison of microscopy and PCR
--------------------------------
A total of 1,784 samples were examined by both microscopy and PCR producing 352 samples that were positive by either or both methods (Table [1](#T1){ref-type="table"}). In this subset, there was an 85.9% concordance in speciation between methods. For samples that were PCR positive for a single species (*P. falciparum*or *P. vivax*) there was greater concordance with microscopy results for *P. vivax*(135/162) compared to *P. falciparum*(36/118) (Table [1](#T1){ref-type="table"}). Ignoring samples identified as mixed infections by PCR, 85.7% of the discordant samples had parasite densities below 100/μL (Figure [3](#F3){ref-type="fig"}). PCR detected 63 *P. falciparum*, 23 *P. vivax*and 10 mixed infections in 1,469 microscopy negative samples, suggesting that parasite densities in these samples are at sub-microscopy level.
######
Comparison of a subset of samples examined by both microscopy and PCR (n = 1784).
Microscopy
----- ------------ -------- -------- ------------- --------- ------
**Pf** **Pv** **Pf + Pv** **Neg**
PCR Pf 36 19 0 63 118
Pv 4 135 0 23 162
Pf + Pv 5 24 0 10 39
Neg 0 33 0 1432 1465
Total 45 211 0 1528 1784
{#F3}
The 33 microscopy positive but PCR negative samples were examined by two level 1 (expert) microscopists and determined as *P. vivax*. Of these, 31 (94.0%) had parasite densities \<100/μL, of which 18 samples were counted as 8-10/μL, which is equivalent to one parasite in 100-300 fields on a thick film.
Performance of RDT
------------------
ICT malaria combo tests were used to test 414 subjects. The RDT detected 25 *P. falciparum*and one *P. vivax*infection. The ICT combo test detected all 12 microscopy and PCR confirmed *P. falciparum*infections, including two samples with parasite densities \<100/μL (Table [2](#T2){ref-type="table"}). The tests also returned positive *P. falciparum*results on 10 microscopy and PCR negative samples. In contrast, it only detected one of the 13 microscopy and PCR confirmed *P. vivax*samples; this sample had a parasite density \>500/μL (Table [2](#T2){ref-type="table"}). The 12 *P. vivax*samples that the ICT combo test failed to detect had parasite densities either \<100/μL (n = 6) or 100-500/μL (n = 6) (Table [2](#T2){ref-type="table"}).
######
Comparison of detection by ICT malaria combo tests and microscopy, PCR
RDT Parasite density (P/μL) Microscopy results/PCR results
--------- ------------------------- -------------------------------- --------- ------- ------- -------- ------- ------- --------- ---------
RDT Pos **Total** **10** **2\*** **2** **1** **1** **0** **0** **10** **26**
\< 100 0 0 1 1 \- \- \- \- 2
100-500 2 0 0 0 \- \- \- \- 2
\> 500 8 2 1 0 \- \- \- \- 11
RDT Neg **Total** **1** **12** **0** **0** **12** **5** **1** **357** **388**
\< 100 1 6 \- \- \- \- \- \- 7
100-500 0 6 \- \- \- \- \- \- 6
\> 500 0 0 \- \- \- \- \- \- 0
Note:\* RDT determined 1 positive for *P. falciparum*and 1 positive for *P. vivax*.
Discussion
==========
Temotu Province has been selected as a possible malaria elimination site based on the reduction of reported malaria cases due to past malaria control measures. The mass blood survey conducted in November 2008 showed that the overall malaria prevalence at the time of survey, based on microscopy was 2.7%, indicating the Province as a low transmission area.
Based on the low transmission rate determined by microscopy, it would be expected that the level of acquired immunity in the community would be relatively low, and consequently the proportion of infected subjects with symptoms and high parasite densities would be high. The survey findings were contrary to this expectation. Firstly, only 5.5% of the microscopy positive individuals had fever at the time of survey. Secondly, 61.0% of microscopy positive individuals had parasite densities below 100/μL. Such a high proportion of subjects having low parasite densities was reported recently in an area of Cambodia where *P. falciparum*prevalence was \>40% \[[@B12]\], reflecting a high level of acquired immunity in the community. Observations from the current study show the proportion of *P. vivax*infected subjects with parasite densities \<100/μL increased with age, suggesting a correlation with exposure and acquired immunity. However, how immunity is maintained in an area with such low transmission is unclear. Other factors such as host genetics and parasite genetics may also be involved in determining the epidemiological situation observed.
The microscopy results suggest that in Temotu Province a large proportion of individuals infected with *Plasmodium*have parasite densities at the limit of microscopy and RDT detection. The addition of PCR testing of samples enhances the detection capability. Compared to PCR, microscopy performed better in detecting *P. vivax*than detecting *P. falciparum*infected subjects as the *P. vivax*prevalence determined by microscopy was 73% of that estimated by PCR, higher than the average of 50% reported from a meta-analysis of 72 pairs of prevalence measurements \[[@B13]\], while the microscopy determined *P. falciparum*prevalence was only 9.7% of that estimated by PCR. It should be noted that the microscopy results obtained here were by WHO certified level 1 microscopists. This level of detection and accuracy would not have been achieved in many aid posts and health centres which often have less experienced microscopists and thus the typical error in microscopy results could be greater than found here \[[@B1],[@B14],[@B15]\].
The PCR-estimated overall malaria prevalence was 8.7%, 3.3 fold higher than that estimated by microscopy alone. This suggests that there is a substantial proportion of the *Plasmodium*infected subjects infected with sub-microscopic parasite densities, in addition to the 61% of subjects with low parasite densities detected by microscopy. The number of sub-microscopic infections is likely higher than reported since PCR itself has a limit of detection. This limit may be decreased if a PCR targets a gene with higher copy numbers compared to the ssrRNA gene used in this study. Interestingly, the sub-microscopic infections were distributed across the island in 32 of the 43 villages where malaria infections were identified. The findings are in agreement with the result of a meta-analysis demonstrating the proportion of sub-microscopic infection of *P. falciparum*is significantly higher in areas of low transmission \[[@B13]\]. These sub-microscopic and low density infections may persist for a period of time without causing any symptoms and, therefore, contribute to the maintenance of acquired immunity.
PCR is expected to generally have a higher sensitivity than microscopy in detecting very low parasite density infections as well as detecting mixed infections. Interestingly, the majority of samples that were PCR positive and microscopy negative were identified by PCR as *P. falciparum*. An almost identical finding was reported recently from Cambodia \[[@B4]\] where microscopy detected a total of 350 *P. falciparum*infections while PCR detected a further 331 *P. falciparum*single infections from microscopy negative samples. A possible explanation for this discrepancy is that in *P. falciparum*usually only ring-stage parasites are present in blood smears and they are smaller than the ring-stage and mature trophozoites of *P. vivax*and may be easier to miss by microscopy particularly when staining is less than perfect.
There were also discrepancies between microscopy and PCR in determining the parasite species. All four microscopy *P. falciparum*/PCR *P. vivax*samples and 14 of the 19 microscopy *P. vivax*/PCR *P. falciparum*samples had parasite densities \<100/μL. In these low parasite density infections microscopic species identification is based on the appearance of only one or two parasites in a thick smear, and therefore their accuracy may be less than optimal. Additionally, some of these discrepancies may be caused by mixed species infections that had low density parasites of both species and were detected by microscopy as one and PCR as another species.
PCR failed to detect 33 microscopy positive samples all of which were determined as *P. vivax*by microscopy. Of these, 94% had parasite densities \<100/μL, and 18 samples were counted as 8-10/μL. At such a low counts, which is equivalent to one parasite in 100-300 fields on a thick film, the microscopy determination of *P. vivax*was made based on just one or a few mature trophozoites found on the thick smear. The reliability of microscopy results based on one or two parasites may be questionable. Alternatively, it is possible that this level of parasite density is also at the detection limit of PCR. Theoretically, the threshold of detection for PCR is limited to the presence of at least one parasite in a sample. At low parasite densities, the presence of a parasite in a small volume of blood is subject to stochastic variation. This, compounded with possible DNA degradation during storage and transportation of the samples, could result in a false negative PCR although this has been reported to be an unlikely cause \[[@B16]\]. Nevertheless, PCR returned positive results for 82/132 of the subjects with *P. vivax*densities \<100/μL.
RDTs are playing an increasing role in malaria control and elimination. The ICT combo kit detects *P. falciparum*histidine rich protein 2(HRP2) and a *Plasmodium*pan-species antigen, aldolase. The performance of the *P. falciparum*detection component of the ICT combo kit was comparable to our level1 microscopists in detecting *P. falciparum*infections. The kit detected all but one microscopy positive *P. falciparum*sample, including two samples with parasite densities \<100/μL. It failed to detect one *P. falciparum*sample with parasite density of 64/μL and 12 microscopy negative but PCR positive samples. ICT combo kit also determined 10 samples which were negative by both microscopy and PCR to be *P. falciparum*positive. This could be due to its detection of circulating PfHRP2 accumulated from recent infections \[[@B17]-[@B20]\]. In contrast, the *P. vivax*detection component of the ICT combo kit failed to detect 12 of the 13 microscopy positive *P. vivax*samples which had parasite densities varying from 10 to 280 parasites/μL. The only RDT positive *P. vivax*sample had a parasite density \>500/μL. The results suggest that the detection limit of the ICT combo kit is \~100/μL for *P. falciparum*and \>300/μL for *P. vivax*. This is in good agreement with the WHO RDT Product Testing (Round 1) results where the ICT Malaria Combo Cassette Test (ML02) achieved detection rates at 200/μL of 86.08% and 0% for *P. falciparum*and *P. vivax*, respectively \[[@B21]\]. In the epidemiological setting of Temotu where 65.6% of *P. vivax*infected subjects have parasite densities \<100/μL plus a number of sub-microscopic infections, the ICT combo kit is clearly not suitable for detecting *P. vivax*infections.
The Temotu Province is preparing for malaria elimination. The epidemiology data presented here point out clearly that passive and active case detection will not be very useful in identifying infections as majority of them were asymptomatic. Malaria elimination strategies being considered for this island are either some form of active parasite detection programme or mass drug administration (MDA). The first option may take the form of mass screening and treatment of residents of sentinel villages known to have habitual high parasite rates on a regular basis to monitor parasite prevalence. As the majority of infected subjects were observed to carry low and sub-microscopic parasite densities the diagnostic method used to screen will have to be particularly suited to detecting low parasite densities. This is critical for the implementation of screening and treatment interventions, and the monitoring and evaluation of the success of the interventions. The MDA option is controversial and not currently recommended by the WHO.
For case management of *P. falciparum*in this setting, ICT combo kit provided promising results. Several other RDTs that showed comparable panel detection scores at 200/μL in the WHO Product Testing Rounds 1 and 2 \[[@B21]\]\[[@B22]\] would be expected to perform at a similar level. However, this study also highlights the high prevalence of infections at \<200/μL, indicating an urgent need to test how these top performing RDTs in the WHO Product Testing Programme perform at low parasite densities. This type of information is important for malaria elimination countries considering mass screening and treatment campaigns. It is likely that to detect the large number of sub-microscopic *P. falciparum*infections, a molecular-based assay will be required.
Elimination of *P. vivax*will eventually become the problem due to relapsing cases caused by the activation of hypnozoites in the liver 2-3 years after the initial infection. Level 1 microscopy seems to offer an advantage in detecting this species at very low parasite densities. Again, the incorporation of molecular based PCR detection method was found to detect more *P. vivax*infected subjects. However, the conventional PCR is difficult to perform in the field due to requirements for DNA extraction, sophisticated equipment and reagents. Therefore, a simple, cheap and easy-to-perform molecular assay is required. The deployment of such an assay combined with microscopy or RDT will provide the best chance of detecting the majority of infected individuals. The purpose of using this molecular assay would not be for treating clinical patients, but for finding asymptomatic infections where the individuals can be followed up for treatment on the next day or even longer.
Conclusion
==========
Results of a mass blood survey conducted in Temotu Province in Solomon Islands showed that a large proportion of *Plasmodium*infected individuals were asymptomatic with low and sub-microscopic parasite densities. With the proposed elimination of malaria from this Province these findings present a challenge for malaria diagnostics in mass screening and treatment programmes and thus for elimination. A combination of methods, or new diagnostics, may be required to detect infections in these asymptomatic parasite reservoirs.
Competing interests
===================
The authors declare that they have no competing interests.
Authors\' contributions
=======================
IH, DS and QC conceived the study. IH carried out microscopy examination, QA, data analysis and drafted the manuscript. WWS performed microscopy QA, PCR and data analysis. LMB, KG and DK performed PCR and data analysis. AB, LB, AV and DS coordinated and planned the survey. KL performed microscopy and QA. MJ coordinated data collating. MLG contributed to data analysis and the writing of manuscript. QC supervised the PCR, analysed data and drafted the manuscript. All authors read and approved the final manuscript.
Acknowledgements
================
The authors would like to thank the people of the islands of Temotu Province, Solomon Islands for participating in the survey. We would also like to thank the team of Pacific Malaria Initiative Support Centre, University of Queensland conducting the survey which included the Australian Army Malaria Institute, the Vector Borne Disease Control Programme, Ministry of Health Solomon Islands and the University of Queensland staff. We thank Dr Robert Cooper for his critical reading of the manuscript and good comments. The opinions expressed are those of the authors and do not necessarily reflect those of the Australian Defence Force. Funds for the reported survey and diagnosis were partially provided by the Australian Government though AusAID.
| {
"pile_set_name": "PubMed Central"
} |
This article was republished on November 20, 2013, to replace an incorrectly published version. Please download this article again to view the correct version.
**Competing Interests:**No competing interests declared.
| {
"pile_set_name": "PubMed Central"
} |
Sakamoto S, Inoue H, Kaneko MK, et al. Generation and evaluation of a chimeric antibody against coxsackievirus and adenovirus receptor for cancer therapy. Cancer Sci. 2019;110:3595--3602. 10.1111/cas.14196
Sakamoto and Inoue contributed equally to this work.
ADCC
: antibody‐dependent cellular cytotoxicity
CAR
: coxsackievirus and adenovirus receptor
CDC
: complement‐dependent cytotoxicity
CTA
: cancer tissue array
FcγRIIIA
: Fcγ receptor IIIA
NK
: natural killer
SCLC
: small cell lung cancer
1. INTRODUCTION {#cas14196-sec-0001}
===============
Coxsackievirus and adenovirus receptor (CAR) protein that encoded by CXADR gene, is a single‐pass transmembrane protein and is involved in the formation and/or maintenance of epithelial tight junctions.[1](#cas14196-bib-0001){ref-type="ref"} CAR has an essential role in the development of the heart and lymphatic system in mice.[2](#cas14196-bib-0002){ref-type="ref"}, [3](#cas14196-bib-0003){ref-type="ref"} Because CAR is the primary cellular receptor for adenoviruses,[1](#cas14196-bib-0001){ref-type="ref"} its expression level is considered an important factor for adenovirus infection and adenoviral vector‐mediated treatment.
It has been reported that high CAR expression levels occur in various human cancers.[4](#cas14196-bib-0004){ref-type="ref"}, [5](#cas14196-bib-0005){ref-type="ref"}, [6](#cas14196-bib-0006){ref-type="ref"}, [7](#cas14196-bib-0007){ref-type="ref"}, [8](#cas14196-bib-0008){ref-type="ref"}, [9](#cas14196-bib-0009){ref-type="ref"}, [10](#cas14196-bib-0010){ref-type="ref"} Moreover, CAR promotes tumor growth in some cancer types. *CXADR* silencing in lung cancer with high CAR expression suppressed tumor formation ability in xenografts,[11](#cas14196-bib-0011){ref-type="ref"} whereas *CXADR* knockdown in oral squamous cell carcinoma resulted in the inhibition of both anchorage‐independent growth and metastatic tumor formation in vivo.[12](#cas14196-bib-0012){ref-type="ref"} These previous studies suggest that CAR might be an appropriate target molecule for cancer therapy.
Previously, we found that LNCaP‐CR cells, a highly tumorigenic subline of the LNCaP human prostate cancer cell line,[13](#cas14196-bib-0013){ref-type="ref"} express higher levels of CAR than their parental cells.[14](#cas14196-bib-0014){ref-type="ref"} We also developed mouse mAbs against human CAR and found that one of these antibodies (clone 6G10A) significantly inhibited tumor growth in xenografts of human prostate, pancreatic, and colorectal cancer.[14](#cas14196-bib-0014){ref-type="ref"} Based on these findings, we proposed that an anti‐CAR antibody might be a feasible candidate for cancer immunotherapy.[14](#cas14196-bib-0014){ref-type="ref"}
Because the immunogenicity of antibodies needs to be reduced for their therapeutic use in humans, mouse‐human chimerization of mouse mAbs is an important step in the development of therapeutic antibodies.[15](#cas14196-bib-0015){ref-type="ref"} In the current study, we generated a mouse‐human chimeric anti‐CAR antibody (ch6G10A) from 6G10A mouse anti‐CAR antibody (mu6G10A), and characterized ch6G10A by flow cytometry, western blotting, ADCC/CDC analyses, and in vivo anti‐tumor activity against a prostate cancer cell line. In addition, we carried out a CTA analysis to investigate CAR expression levels in lung, prostate, and brain tumors, and examined the anti‐tumor activities of anti‐CAR antibodies against SCLC using mouse xenograft models.
2. MATERIALS AND METHODS {#cas14196-sec-0002}
========================
2.1. Cell lines and reagents {#cas14196-sec-0003}
----------------------------
KHYG‐1/FcγRIIIA cells, a human NK cell line stably expressing Fcγ receptor IIIA (FcγRIIIA), were previously established[16](#cas14196-bib-0016){ref-type="ref"} and were maintained in RPMI‐1640 medium (Nissui Pharmaceutical) supplemented with 10% FBS (PAN Biotech GmbH) and 100 units of human interleukin (IL)‐2 (FUJIFILM Wako Pure Chemical Corporation). Human SCLC cell lines NCI‐H69, DMS53, and DMS114 were maintained in RPMI‐1640 medium containing 10% FBS. Human prostate cancer cell line DU‐145, human SCLC cell line DMS273, GFP‐labeled subline DMS273‐GFP, highly metastatic subline G3H,[17](#cas14196-bib-0017){ref-type="ref"} and C5B (S. Sakamoto, H. Inoue & M. Kawada, unpubl. data, manuscript in preparation) were maintained in DMEM (Nissui) containing 10% FBS.
2.2. Generation of mouse‐human chimeric anti‐human CAR {#cas14196-sec-0004}
------------------------------------------------------
A mouse anti‐human CAR mAb, mu6G10A, was developed as described previously.[14](#cas14196-bib-0014){ref-type="ref"} Generation of mouse‐human chimeric anti‐human CAR (ch6G10A) was previously described.[18](#cas14196-bib-0018){ref-type="ref"} Briefly, the appropriate *V* ~H~ and *V* ~L~ cDNAs of mu6G10A and *C* ~H~ and *C* ~L~ of human IgG~1~ were subcloned into pcDNA3.3/Neo or pcDNA3.1/Zeo vectors (Thermo Fisher Scientific Inc.), respectively. Antibody expression vectors were transfected into CHO‐S cells (Thermo Fisher Scientific Inc.) using Lipofectamine 2000 reagent (Thermo Fisher Scientific Inc.). Stable transfectants of CHO‐S/ch6G10A were selected by culturing the transfectants in medium containing 1 mg/mL G418 (Nacalai Tesque, Inc.) and 0.5 mg/mL zeocin (InvivoGen). Stable transfectants were cultured for 14 days in CHO‐S‐SFM II medium (Thermo Fisher Scientific Inc.), and then ch6G10A immunoglobulin was purified from the culture supernatant using Protein G Sepharose (GE Healthcare United Kingdom, Ltd). Purity of ch6G10A was evaluated by SDS‐PAGE and Coomassie Brilliant Blue staining.
2.3. Flow cytometry {#cas14196-sec-0005}
-------------------
Cells were fixed in 4% neutral buffered formalin for 10 minutes and suspended in PBS containing 0.1% NaN~3~. Cells were resuspended in PBS containing 0.5% BSA and incubated with anti‐CAR antibodies for 1 hour at 4°C and then with PE‐labeled secondary antibodies for 1 hour at room temperature. After washing with PBS containing 0.1% NaN~3~, cells were analyzed by FACSCalibur (BD Biosciences).
2.4. Western blot analysis {#cas14196-sec-0006}
--------------------------
Western blot analysis was carried out as described previously.[19](#cas14196-bib-0019){ref-type="ref"} Briefly, cultured cell pellets were lysed with lysis buffer (20 mmol/L HEPES (pH 7.5), 150 mmol/L NaCl, 1% Triton X‐100, 10% glycerol, 1 mmol/L EDTA, 50 mmol/L NaF, 50 mmol/L β‐glycerophosphate, 1 mmol/L Na~3~VO~4~, 1× cOmplete Protease Inhibitor Cocktail; Roche). Lysates were centrifuged at 13 000 *g* for 15 minutes at 4°C, and the supernatants were boiled in SDS sample buffer containing 0.5 mol/L β‐mercaptoethanol. These samples were separated by 12.5% SDS‐PAGE and transferred to PVDF membranes (Merck KGaA). Protein levels were detected using the following antibodies: rabbit polyclonal antibodies specific for CAR (H‐300/sc‐15405; Santa Cruz Biochemical) and rabbit polyclonal antibodies specific for HSP90 (H‐114/sc‐7947; Santa Cruz Biochemical).
2.5. Immunohistochemical analysis of cancer tissue arrays {#cas14196-sec-0007}
---------------------------------------------------------
Cancer tissue arrays (BC041115c for lung cancer, BC19013 for prostate cancer, and GL803b for brain cancer) were purchased from US Biomax. The arrays were deparaffinized and rehydrated. Then, antigens were unmasked using citrate buffer for 10 minutes at 96°C. Prior to antibody reactions, endogenous peroxidase was inactivated with 3% hydrogen peroxidase for 10 minutes at room temperature followed by rinsing with distilled water and TBST. The arrays were blocked with 10% horse serum for 1 hour at 37°C and incubated with primary antibody against CAR (H‐300/sc‐15405, 1:750; Santa Cruz Biochemical) overnight at 4°C. After washing three times with TBST, the arrays were incubated with secondary antibody peroxidase‐conjugated anti‐rabbit IgG (ImmPRESS Reagent kit; Vector Laboratories) for 30 minutes at room temperature. The arrays were then washed three times with TBST and developed using ImmPACT DAB (Vector Laboratories) for 1 minute, and then briefly immersed in hematoxylin for counterstaining and evaluation under light microscopy. Estimated visual intensity of CAR immunostaining was graded on an arbitrary three‐point scale: negative (0), positive (1), strongly positive (2).
2.6. Antibody‐dependent cellular cytotoxicity {#cas14196-sec-0008}
---------------------------------------------
Antibody‐dependent cellular cytotoxicity was measured with a calcein AM release assay as described previously.[14](#cas14196-bib-0014){ref-type="ref"} Briefly, for ADCC activity, target cells were labeled with 10 μg/mL calcein AM (Dojindo) for 30 minutes at 37°C, washed three times with complete medium, and further incubated for 1 hour at 37°C. Labeled target cells (5 × 10^4^ cells) and effector cells were then cocultured in 96‐well plates and incubated for 4 hour at 37°C. After centrifugation, fluorescence intensity (F) of cell‐free supernatants was measured (excitation 485 nm, emission 538 nm). Cytolytic activity was calculated using the following formula:
$$\begin{array}{r}
{\%\;\text{of\ specific\ lysis} = (\text{F\ experiment} - \text{F\ spontaneous})} \\
{/(\text{F\ maximal} - \text{F\ spontaneous}) \times 100.} \\
\end{array}$$
Assays were carried out in triplicate at least.
2.7. Complement‐dependent cytotoxicity {#cas14196-sec-0009}
--------------------------------------
For CDC activity, target cells (2 × 10^4^ cells) were inoculated into 96‐well plates and incubated with 5% rabbit complement (Cedarlane) and ch6G10A, mu6G10A, control human IgG, or control mouse IgG for 4 hours at 37°C. Cell viability was determined using MTT assay. Assays were done in triplicate at least.
2.8. In vivo anti‐tumor activity {#cas14196-sec-0010}
--------------------------------
Animal experiments were approved by the Institute Committee for Animal Experiments at the Institute of Microbial Chemistry and carried out according to the ethics guidelines of our institute. Female BALB/c nude mice and female SCID‐beige mice were obtained from Charles River Japan (Kanagawa, Japan) and were maintained in a specific pathogen‐free barrier facility inhouse. Mice aged 8 weeks were used for the in vivo anti‐tumor assay. To prepare the subcutaneous xenograft model, 1 × 10^7^ DU‐145 cells or 1 × 10^6^ NCI‐H69 cells in 0.1 mL of 62.5% growth factor‐reduced Matrigel (BD Biosciences) were injected s.c. into the left flanks of the SCID‐beige mice. When the tumor size had reached approximately 100 mm^3^, i.v. injection of the antibodies was started. To prepare the orthotopic transplanted xenograft model of the highly metastatic subline of DMS273 human SCLC cells, 2.5 × 10^5^ C5B cells were injected into the left lungs of nude mice as described previously.[17](#cas14196-bib-0017){ref-type="ref"} An Olympus OV110 Small Animal Imaging System (Olympus Corp.) was used for imaging orthotopic and metastatic tumor formation in the orthotopic transplanted xenograft model. Length (*L*) and width (*W*) of the subcutaneous and orthotopic tumors were measured by calipers and tumor volume (TV) was calculated using the formula: TV = (*L* × *W* ^2^)/2. All the in vivo experiments were carried out at least twice with similar results.
2.9. Statistical analysis {#cas14196-sec-0011}
-------------------------
Results are expressed as mean ± SD. Statistical analysis was carried out by Student\'s *t* test or Mann‐Whitney *U* test as described in the figure legends. *P* \< .05 was considered statistically significant.
3. RESULTS {#cas14196-sec-0012}
==========
3.1. Generation and characterization of a mouse‐human chimeric anti‐CAR antibody (ch6G10A) {#cas14196-sec-0013}
------------------------------------------------------------------------------------------
Previously, we developed a mouse mAb against human CAR (mu6G10A) that recognized an extra‐membranous region of CAR and showed anti‐tumor activities against CAR‐expressing cancer cells in xenograft models.[14](#cas14196-bib-0014){ref-type="ref"} To reduce the immunogenicity of antibodies in humans, we attempted to generate a mouse‐human chimeric antibody (ch6G10A) by fusing the *V* ~H~ and *V* ~L~ regions of mu6G10A with the *C* ~H~ and *C* ~L~ regions of human IgG~1~, respectively. As shown in Figure [1](#cas14196-fig-0001){ref-type="fig"}A, ch6G10A reacted with the human prostate cancer cell line DU‐145, which highly expresses human CAR, as strongly as mu6G10A by flow cytometry. Because mu6G10A can induce both ADCC and CDC,[14](#cas14196-bib-0014){ref-type="ref"} we next examined whether ch6G10A can also induce both ADCC and CDC. When a human NK cell line, KHYG‐1/FcγRIIIA, which stably expresses FcγRIIIA was used as effector cells,[16](#cas14196-bib-0016){ref-type="ref"} ADCC by ch6G10A was observed against DU‐145 cells (Figure [1](#cas14196-fig-0001){ref-type="fig"}B). ch6G10A also significantly induced CDC activity against DU‐145 cells when using rabbit complement (Figure [1](#cas14196-fig-0001){ref-type="fig"}C). Then, we evaluated the anti‐tumor activity of ch6G10A against DU‐145 cells in vivo using a subcutaneous xenograft model in nude mice. ch6G10A, control human IgG (500 μg/mice) or saline was injected i.v. at days 0 and 7, and human NK cells KHYG‐1/FcγRIIIA were also injected around the tumors at days 0 and 7. ch6G10A significantly inhibited the growth of xenograft DU‐145 tumors in vivo (43% inhibition compared with control human IgG, *P *\< .01) without significant body weight loss in the host mice (Figure [2](#cas14196-fig-0002){ref-type="fig"}, data not shown). These results showed that ch6G10A had ADCC‐/CDC‐inducing activities and in vivo anti‐tumor activity against CAR‐expressing prostate cancer cells and that we successfully generated a mouse‐human chimeric anti‐CAR antibody with potent anti‐tumor activity.
{#cas14196-fig-0001}
{#cas14196-fig-0002}
3.2. Expression of CAR in human lung, prostate, and brain tumors {#cas14196-sec-0014}
----------------------------------------------------------------
Our previous results suggest that CAR is highly expressed in tumor tissues such as prostate, lung, and brain.[14](#cas14196-bib-0014){ref-type="ref"} Because we used only small numbers of tissue sections in the previous study, we further assessed the expression of CAR in these tumors through immunohistochemical staining of CTA using commercially available antibody against CAR (H‐300) (Figure [3](#cas14196-fig-0003){ref-type="fig"}). CTA analysis showed that CAR was considerably expressed in lung squamous cell carcinoma (19/40; 48%), lung adenocarcinoma (14/45; 31%), lung neuroendocrine cancer (including small cell lung cancer) (7/12; 58%), prostate adenocarcinoma (45/56; 80%), and glioblastoma (23/39; 59%). Importantly, expression of CAR was significantly higher in all lung cancer types than in normal lung tissue (lung squamous cell carcinoma and lung neuroendocrine cancer, *P *\< .01; lung adenocarcinoma, *P *\< .05), and was significantly higher in glioblastoma than in normal brain tissue (*P *\< .05). Thus, our results show that CAR was highly expressed in these human tumor tissues.
{#cas14196-fig-0003}
3.3. In vivo anti‐tumor activity of ch6G10A antibody against NCI‐H69 small cell lung cancer cells {#cas14196-sec-0015}
-------------------------------------------------------------------------------------------------
In addition to our CTA results, other studies also showed that CAR was highly expressed in SCLC patients.[5](#cas14196-bib-0005){ref-type="ref"}, [8](#cas14196-bib-0008){ref-type="ref"}, [10](#cas14196-bib-0010){ref-type="ref"} Therefore, we investigated CAR protein levels in four human SCLC cell lines (DMS114, DMS53, DMS273, and NCI‐H69) by western blot analysis, and found that CAR was expressed in three of the four lines and was particularly overexpressed in NCI‐H69 cells (Figure [4](#cas14196-fig-0004){ref-type="fig"}A). These expression data of CAR prompted us to evaluate the in vivo anti‐tumor activity of ch6G10A against NCI‐H69 using a subcutaneous xenograft model in nude mice. ch6G10A or control human IgG (500 μg/mice) was injected i.v. at days 0 and 7, and human NK cells KHYG‐1/FcγRIIIA were injected around the tumors at days 0 and 7. As shown in Figure [4](#cas14196-fig-0004){ref-type="fig"}B,C, ch6G10A significantly inhibited the growth of xenograft NCI‐H69 tumors in vivo (37% inhibition compared with control human IgG, *P *\< .01) without significant body weight loss in the host mice (data not shown). These results showed that ch6G10A had anti‐tumor activity against CAR‐expressing SCLC cells and suggested that SCLC might be a candidate disease for targeting therapy to CAR with therapeutic antibody.
{#cas14196-fig-0004}
3.4. In vivo anti‐tumor activity of anti‐CAR antibody against an orthotopic transplantation model of SCLC {#cas14196-sec-0016}
---------------------------------------------------------------------------------------------------------
We previously developed an orthotopic transplantation model of SCLC using GFP‐labeled human SCLC cell line DMS273 (DMS273‐GFP) and two highly metastatic sublines (G3H, C5B); this model has significant metastatic activity and comparable metastatic tropism with that of SCLC patients.[17](#cas14196-bib-0017){ref-type="ref"} Because CAR expression was clearly observed in these cells (Figure [4](#cas14196-fig-0004){ref-type="fig"}A), we attempted to elucidate whether anti‐CAR antibody could suppress orthotopic tumor growth and distant metastasis formation in the model using C5B cells. Owing to the difficulty of giving systemic human NK cells to xenograft mice, which is essential for the anti‐tumor activity of ch6G10A, we used mouse monoclonal CAR antibody mu6G10A in this experiment. Intravenous injection with mu6G10A or control mouse IgG (250 μg/mice) was started a day after orthotopic transplantation of C5B cells. The injection was carried out once a week for 5 weeks. Orthotopic tumor growth was prominently inhibited by mu6G10A (78% inhibition compared with control mouse IgG, *P *\< .005) without significant body weight loss in the host mice (Figure [5](#cas14196-fig-0005){ref-type="fig"}, data not shown). Incidence of distant metastases formation was 67% in the control mouse IgG‐treated group and 33% in the mu6G10A‐treated group, and significantly fewer metastasis‐positive organs were observed in the mu6G10A‐treated group compared with the control mouse IgG‐treated group (*P *\< .05; Figure [5](#cas14196-fig-0005){ref-type="fig"}B). Thus, treatment with anti‐CAR antibody efficiently inhibited tumor growth in the lung as well as metastasis of SCLC. These results support the notion that SCLC might be a potential disease for targeting therapy to CAR with therapeutic antibody.
{#cas14196-fig-0005}
4. DISCUSSION {#cas14196-sec-0017}
=============
In the present study, we successfully generated a mouse‐human chimeric anti‐CAR monoclonal antibody, ch6G10A, which has ADCC‐ and CDC‐inducing activities and in vivo anti‐tumor activity against CAR‐expressing cancer cells. In addition, CTA analysis showed that CAR is highly expressed in neuroendocrine lung cancer including SCLC, and treatment with anti‐CAR antibody inhibited both subcutaneous and orthotopic tumor growth in mouse xenograft models of human SCLC cell lines. These results suggest that targeting therapy to CAR with therapeutic antibodies might be effective against several cancers including SCLC.
Our CTA analysis showed significantly higher expression of CAR in neuroendocrine lung cancers including SCLC compared with normal lung tissues (Figure [3](#cas14196-fig-0003){ref-type="fig"}), consistent with the findings of previous studies.[5](#cas14196-bib-0005){ref-type="ref"}, [8](#cas14196-bib-0008){ref-type="ref"}, [10](#cas14196-bib-0010){ref-type="ref"} Furthermore, treatment with anti‐CAR antibody effectively inhibited both in vivo tumor growth (subcutaneous and orthotopic) and distant metastasis formation in mouse xenograft models of human SCLC cell lines (Figures [4](#cas14196-fig-0004){ref-type="fig"} and [5](#cas14196-fig-0005){ref-type="fig"}). SCLC accounts for 10%‐20% of all cases of lung cancer worldwide and shows aggressive growth and early metastatic spread. The current main treatment against SCLC is chemotherapy but the high chance of disease relapse after chemotherapy results in poor median survival.[20](#cas14196-bib-0020){ref-type="ref"} Therefore, novel therapeutic strategies against SCLC are urgently needed. CAR expression data in SCLC and our current results with anti‐CAR antibody collectively suggested that therapeutic antibodies against CAR might be an alternative strategy for SCLC treatment. Our CTA analysis also showed significantly higher expression of CAR in lung squamous cell carcinoma, lung adenocarcinoma, and glioblastoma than in matched adjacent normal tissues. Because there are many cases of these diseases, validation of the anti‐tumor activity of our anti‐CAR antibody will be meaningful for future investigations.
Our previous study showed that mu6G10A exerts anti‐tumor activity primarily through both ADCC and CDC activities.[14](#cas14196-bib-0014){ref-type="ref"} In this study, we generated ch6G10A with the human IgG~1~ constant region because human IgG~1~ exerts higher ADCC/CDC activities among the IgG~1~ subclass. Therefore, it is presumed that ch6G10A also exerts its anti‐tumor activity primarily through both ADCC and CDC. If this is the case, the anti‐tumor activity of ch6G10A could be enhanced by ADCC‐enhancing technologies such as afucosylation of the Fc domain.[21](#cas14196-bib-0021){ref-type="ref"} Thus, ch6G10A could be a potent lead antibody for cancer immunotherapy.
We showed that the treatment with mu6G10A efficiently inhibited tumor growth in the lung as well as metastasis in the SCLC metastatic model (Figure [5](#cas14196-fig-0005){ref-type="fig"}). Although it is very important as to whether ch6G10A also inhibits SCLC metastasis, there are no suitable metastatic models for the evaluation of chimeric mAb at this point. However, there is an interesting humanized mouse strain, NOG‐IL‐2 Tg, which can induce human NK cells from hematopoietic stem cells (HSC).[22](#cas14196-bib-0022){ref-type="ref"} The report suggested that chimeric anti‐human CCR4 mAb (Poteligeo, Kyowa‐Kirin, Tokyo, Japa) suppressed tumor growth of human lymphoma cells in human HSC‐transferred NOG‐IL‐2 Tg mice through ADCC. If SCLC metastatic models can be generated using mice, evaluation of the effects of ch6G10A in SCLC metastasis might be possible.
Another important point about the result of Figure [5](#cas14196-fig-0005){ref-type="fig"} is the mechanism by which anti‐CAR mAb inhibited SCLC metastasis. It is still unclear whether anti‐CAR mAb directly reacts with tumor cells in blood/lymph vessels and metastatic sites. However, as a result of technical limitations of our model, it is very difficult to answer this question. Because we use fluorescence imaging to chase the metastatic tumor in our model, killing and dissection of mice are needed for quantitative detection of metastatic tumors. This makes it difficult to start treatment with antibody after confirming formation of metastatic tumors in the model. In addition, enucleation of the orthotopic tumor in lung will result in death of the model mouse. This also makes it difficult to examine the direct effect of anti‐CAR mAb on tumor cells in metastatic sites of the model.
One of the major points for the clinical safety of therapeutic antibodies is reactivity with normal tissues. We previously showed that mu6G10A reacts with normal skin, prostate, and kidney tissue.[14](#cas14196-bib-0014){ref-type="ref"} At this point, we could not avoid the possibility that the reactivity of 6G10A to normal tissues could cause some adverse events when the 6G10A‐based antibody is given to humans. Thus, further safety evaluations of the 6G10A‐based antibody using animal models such as primates or human CAR knock‐in mice are required. If the reactivity of 6G10A to normal tissues causes severe problems, cancer specificity of the antibody should be improved.
In conclusion, our studies and other reports collectively suggest that CAR should be considered a candidate target for cancer immune therapy against several cancers including SCLC.
DISCLOSURE {#cas14196-sec-0019}
==========
Authors declare no conflicts of interest for this article.
We thank T. Ohishi, A. Harakawa, and Y. Kohda (BIKAKEN) for helpful discussions and their technical assistance. This work was supported by grants from the Japan Society for the Promotion of Science (26460481 and 17K08777) (S.S.) and from the Kobayashi Foundation for Cancer Research (M.K.). This work was also supported by the Platform Project for Supporting Drug Discovery and Life Science Research (Basis for Supporting Innovative Drug Discovery and Life Science Research \[BINDS\]) from AMED under Grant Number JP18am0101078 (Y.K.). We also thank H. Nikki March, PhD, from Edanz Group ([www.edanzediting.com/ac](http://www.edanzediting.com/ac)) for editing a draft of this manuscript.
| {
"pile_set_name": "PubMed Central"
} |
Acute ischemic stroke patients with large established infarction treated with reperfusion therapy are more likely to have a poor functional outcome and reperfusion edema or haemorrhage.^[@R1]^ The Alberta Stroke Program Early Computed Tomography Score (ASPECTS) score is a 10-item score assessing brain parenchyma hypodensity in predefined anterior circulation regions on brain noncontrast CT as a marker of early ischemic changes.^[@R2]^ Use of the score was part of the imaging inclusion criteria in several large randomized clinical trials on endovascular clot retrieval in ischemic stroke.^[@R3][@R4][@R5]^ The ASPECTS score is widely used to identify acute ischemic stroke patients with established infarction who may not benefit or may be harmed by reperfusion therapy. Although the ASPECTS score correlates well with functional outcome in ischemic stroke patients evaluated for thrombolysis, correlation with functional outcome after thrombectomy is less robust, especially when a cutoff value of the score is used to exclude patients from treatment.^[@R2],[@R6][@R7][@R8]^ Benefit of thrombectomy is preserved in patients with ASPECTS scores lower than the cutoff value proposed in the ASPECTS validation study (i.e., 8) and in patients with scores lower than the cutoff value commonly used for selection of patients for reperfusion therapy (i.e., 6 or 7).^[@R2],[@R8]^
CT perfusion to evaluate ischemic core and penumbra volume of ischemic stroke has shown to correlate well with acute MRI diffusion-weighted imaging (DWI) lesion volume.^[@R9],[@R10]^ Compared to acute brain MRI, CT perfusion is more readily available and rapidly acquired.^[@R9]^ Software processing of CT perfusion source images can generate ischemic core and penumbra maps that can accurately identify the acute infarct and tissue at risk for infarction.^[@R10]^ The addition of CT perfusion imaging, however, is more time-consuming compared to noncontrast CT and CT angiography imaging alone.
We therefore aimed to compare the accuracy of brain noncontrast CT ASPECTS and CT perfusion core volume to identify patients with established infarction.
METHODS {#s1}
=======
Patient population. {#s1-1}
-------------------
We retrospectively analyzed a convenience series of 59 confirmed acute anterior circulation ischemic stroke patients between May 2003 and December 2011 who presented within 6 hours from stroke onset and underwent acute multimodal CT as well as MRI. Each patient underwent a noncontrast CT, CT angiography, and CT perfusion as part of the standard acute stroke assessment. All patients then went on to have MRI brain within 100 minutes from CT perfusion. Thrombolytic treatment was initiated between CT and MRI in eligible patients. As this study was performed before validation of endovascular clot retrieval therapy in acute ischemic stroke, no patient received a thrombectomy.
Imaging protocol. {#s1-2}
-----------------
CT imaging was performed on a 16-slice Philips (Best, the Netherlands) Mx8000 or a 64-slice Philips Brilliance. A 40 mL bolus of contrast agent (Ultravist 370; Bayer HealthCare, Whippany, NJ) at a rate of 6 mL/s was used to acquire CT perfusion images at 45 time points of 1.33 seconds. Four to eight adjacent 5- to 6-mm slices covered 24--40 mm sections, using a 16- or 64-slice scanner, respectively. We used commercially available MIStar software (Apollo Medical Imaging Technology, Melbourne, Australia) to process CT perfusion images and generate cerebral blood volume, cerebral blood flow, mean transit time, and delay time, as well as ischemic core and penumbra maps.^[@R10][@R11][@R12]^ Core and penumbra maps are graphically represented as red and green areas, respectively, superimposed on the noncontrast CT ([figure 1](#F1){ref-type="fig"}). The software automatically selects global arterial input function from a large intracranial artery and a venous outflow function from a large draining cerebral venous sinus. A model-free singular value decomposition is used to deconvolve the tissue enhancement curve and the arterial input function with automated delay and dispersion correction.^[@R13],[@R14]^ Core infarct is defined as the brain region with a relative cerebral blood flow of less than 30% (compared to the contralateral hemisphere) within a region with a delay time of more than 3 seconds and a minimum cluster size of 5 mm. The perfusion lesion is defined as the region with a delay time of more than 3 seconds and the penumbra as perfusion lesion after exclusion of the ischemic core. Areas of no blood flow, chronic infarction, or CSF regions are automatically masked from the perfusion maps: no blood flow pixels are removed by eliminating areas where cerebral blood flow = 0 and a Hounsfield threshold and geometrical analysis is used to remove skull and CSF or ventricle pixels.
![Noncontrast CT, core (red) and penumbra (green) CT perfusion maps, and acute diffusion-weighted imaging (DWI) MRI in 3 different patients\
(A) Agreement among all imaging modalities (Alberta Stroke Program Early Computed Tomography Score \[ASPECTS\] score 5, CT perfusion core volume 58 mL, MRI DWI volume 106 mL). (B) Established large infarct with high ASPECTS score (ASPECTS score 9, CT perfusion core volume 52 mL, MRI DWI volume 90 mL).](NEUROLOGY2016763458FF1){#F1}
MRI brain was performed on a 1.5 T MRI (Siemens \[Munich, Germany\] Avanto) and included axial isotropic DWI, echoplanar spin-echo sequence, time of flight magnetic resonance angiography, and bolus-tracking perfusion-weighted imaging.^[@R15]^
Data collection. {#s1-3}
----------------
Prior to data collection, patients with missing imaging data were excluded (n = 5). ASPECTS was performed by 4 independent vascular neurologists at 2 centers. Assessors were unblinded to the affected hemisphere. CT images were assessed using nonstandard, variable soft copy, narrow window and level settings centered between 35--45 HU width and 35--45 HU level ([aspectsinstroke.com](http://www.aspectsinstroke.com/)). After collection of all imaging data, a consensus meeting was organized to solve disagreement between raters and to obtain a single ASPECTS score per patient. We used commercial MIStar software to calculate the ischemic core volume on CT perfusion, applying the abovementioned thresholds. CT perfusion imaging analysis was performed after ASPECTS rating. We used an area of interest tool for semiautomated selection of lesion outline to delineate acute MRI DWI lesions based on signal intensity (contrast). The tool automatically measured the volume of the DWI lesion by adding the highlighted area in each slice the lesion was visualized. A DWI lesion volume of at least 70 mL was chosen to represent a large infarct volume beyond which the benefit of IV reperfusion therapy may be limited.^[@R10],[@R16],[@R17]^ Assessment of MRI DWI lesions was performed after acute CT perfusion imaging analysis. Before perfusion postprocessing, CT perfusion source images were coregistered to the corresponding MRI DWI images. Next, the MRI DWI areas of interest were transferred to the coregistered acute CT perfusion core--penumbra maps for statistical volume analysis. Volume coverage of the acute MRI DWI lesion by the CT perfusion core lesion was then calculated.
Statistical analysis. {#s1-4}
---------------------
All statistical analyses were performed with SPSS version 20 (IBM Corp., Armonk, NY). The area under the receiver operating characteristic curve (AUC) was calculated to estimate the accuracy of noncontrast CT ASPECTS and CT perfusion core volume to predict a large acute MRI DWI lesion volume (\>70 mL) and the method of Delong et al. was applied to compare the AUC between ASPECTS and CT perfusion.^[@R18]^ Sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), and correct classification rate were then calculated for different ASPECTS cutoff scores and CT perfusion core volume cutoffs to predict a DWI lesion of at least 70 mL using standard 2 × 2 contingency table methodology. Next, the Youden index was used to determine the optimum ASPECTS cutoff score and CT perfusion core volume. Intraclass correlation coefficient (ICC) was used to estimate interrater agreement for ASPECTS scoring (2-way mixed effects model with measures of absolute agreement). AUC was used to determine accuracy for ASPECTS, CT perfusion core volume, and MRI DWI lesion volume to predict poor functional outcome at 90 days after the stroke, defined as a modified Rankin Scale score of 3--6.
The correlation between CT perfusion core volume and MRI DWI lesion volume was measured by calculating Pearson correlation coefficient (Pearson *R*) and the coefficient of determination (*R*^2^). ICC was calculated as a measure for agreement between CT perfusion core volume and DWI lesion volume. A scatterplot and Bland-Altman plot was used to further examine the correlation between CT perfusion and MRI DWI lesion volume.
A Wilcoxon rank sum test was used to assess statistical differences between onset and imaging time in patients with high vs low ASPECTS scores.
Standard protocol approvals, registrations, and patient consents. {#s1-5}
-----------------------------------------------------------------
The study was approved by the Hunter New England Human Research Ethics Committee and all patients gave informed consent.
RESULTS {#s2}
=======
Patient characteristics. {#s2-1}
------------------------
The median patient age was 76 years (interquartile range \[IQR\] 68--84). Forty-nine percent of the population was male. Median NIH Stroke Scale score at admission was 15 (IQR 12--17). Median onset to admission time was 157 minutes (IQR 113--166). Median time from symptom onset to start of imaging was 200 minutes (IQR 177--249). The acute MRI brain was performed at a median of 60 minutes after CT imaging (IQR 34--98 minutes). Time from onset to start of imaging did not differ between high (i.e., 7--10) and low (i.e., 0--6) ASPECTS score subgroups (193 minutes \[IQR 168--221\] vs 228 minutes \[IQR 187--276\]; *p* = 0.08).
Twenty-nine patients (49%) were thrombolyzed. The other patients were either outside of the thrombolysis time window or were excluded from thrombolysis because of large established infarction on CT imaging or comorbidity.
Imaging characteristics. {#s2-2}
------------------------
ASPECTS scores ranged from 1 to 10. Median consensus ASPECTS score was 8 (IQR 5--9). Sixty-three percent of patients had an ASPECTS score of 7 or higher. Eighty-eight percent of patients had an ASPECTS score of 4 or higher (table e-1 and figure e-1 at [Neurology.org](http://neurology.org/lookup/doi/10.1212/WNL.0000000000004028)). ASPECTS interrater agreement was good (ICC 0.71). More details about the interrater reliability can be found in the supplemental data.
CT perfusion ischemic core volumes ranged from 0 to 188 mL. Median lesion volume was 22 mL (IQR 10.4--71.9). Twenty patients (34%) had a CT perfusion core volume of at least 50 mL and 16 patients (27%) had a core volume of 70 mL or more (table e-2). Median acute MRI DWI lesion volume was 24.5 mL (IQR 10--63.9 mL). DWI lesion volume ranged from 1.3 to 215 mL.
Forty-nine patients (83%) had a documented large vessel occlusion. In 27 patients (55% of large vessel occlusions), the M1 segment of the middle cerebral artery was occluded, and in 15 patients (31% of large vessel occlusions), the carotid was occluded or a carotid--middle cerebral artery tandem lesion was present. At the time of MRI, complete vessel recanalization had occurred in 10 patients (20%). Twenty-one patients (43%) showed a persistent occlusion. The other 18 patients (41%) had partial recanalization.
Correlation of noncontrast CT ASPECTS score with acute MRI DWI lesion volume. {#s2-3}
-----------------------------------------------------------------------------
Conventional ASPECTS had a good overall accuracy to predict an acute MRI DWI lesion volume of at least 70 mL (AUC 0.87 \[95% CI 0.78--0.96\]; [figure 2](#F2){ref-type="fig"}, table e-1). The ASPECTS cutoff value of \<7 (i.e., 0--6 vs 7--10) had an optimum Youden index of 0.60. At this cutoff value, sensitivity to detect an acute DWI lesion ≥70 mL was 0.74, specificity was 0.86, PPV was 0.77, and NPV was 0.84 ([figure 2](#F2){ref-type="fig"}, table e-1). Eighty-one percent of patients were correctly classified as either small or large volume lesion. Using a cutoff score of \<7, 5 patients (8.5%) with small DWI lesions would be excluded from reperfusion therapy and 6 patients (10%) with large DWI lesions would undergo reperfusion treatment. The other ASPECTS cutoff score used in several thrombectomy trials (i.e., \< 6) had a sensitivity of 0.65 and a specificity of 0.92 ([figure 2](#F2){ref-type="fig"}, table e-1).
{#F2}
Correlation of CT perfusion volume with acute MRI DWI lesion volume. {#s2-4}
--------------------------------------------------------------------
CT perfusion core volume had excellent accuracy to predict an acute DWI lesion volume of ≥70 mL (AUC 0.95 \[95% CI 0.89--1.00\]; [figure 2](#F2){ref-type="fig"}, table e-2). The cutoff CT perfusion core volume of at least 50 mL most accurately matched a DWI lesion of ≥70 mL (sensitivity 0.86, specificity 0.97, PPV 0.95, NPV 0.92, Youden index 0.84; [figure 2](#F2){ref-type="fig"}, table e-2). Using this volume cutoff, 1 patient (2%) would erroneously be refused therapy and 3 patients (5%) with large DWI lesions would be missed.
There was no significant difference between the accuracy of CT perfusion and ASPECTS to predict an acute DWI lesion of at least 70 mL (*p* = 0.17).
The CT perfusion core volume covered on average 90.3% of the MRI DWI lesion (range 32%--100%; median 100%; IQR 86%--100%; Pearson *R* = 0.88; *R*^2^ = 0.77; figure e-2). ICC between CT perfusion core volume and DWI lesion volume was 0.77 (95% CI 0.22--0.90). Larger DWI lesions in particular were significantly underestimated by CT perfusion (bias −39 mL; median volume difference −30 mL; IQR −4 to −57 mL; *p* \< 0.001; Bland-Altman plot, see figure e-3).
Correlation of ASPECTS, CT perfusion, and MRI with functional outcome. {#s2-5}
----------------------------------------------------------------------
MRI DWI lesion volume most accurately predicted a poor clinical outcome (AUC 0.95 \[95% CI 0.90--1.00\]). The accuracy of CT perfusion core volume and ASPECTS was comparable (AUC 0.79 \[95% CI 0.67--0.90\] vs AUC 0.75 \[95% CI 0.62--0.87\]).
DISCUSSION {#s3}
==========
In this study, we compared the accuracy of ASPECTS with automated CT perfusion core volume estimation to identify patients with established large infarcts on hyperacute MRI, defined as an MRI DWI lesion volume of at least 70 mL. We found no significant difference between the accuracy of ASPECTS and CT perfusion to identify patients with a large DWI lesion (AUC 0.87 vs 0.95, respectively, *p* = 0.17).
The optimum cutoff values for ASPECTS and CT perfusion core volume, however, did result in a higher number of patients incorrectly classified as large or small established infarcts with use of ASPECTS. The optimum ASPECTS cutoff score of \<7 resulted in a higher number of false-positive cases and a lower number of correctly classified patients when compared to CT perfusion (8.5% vs 2% and 81% vs 93%, respectively). The lower specificity of ASPECTS to identify patients with large established brain infarction may in part explain the findings of the retrospective analysis of the unselected endovascular clot retrieval trial Multicenter Randomized Clinical Trial of Endovascular Treatment for Acute Ischemic Stroke in the Netherlands (MR CLEAN), in which thrombectomy increased the chance of independence by 19% in patients with an ASPECTS score of 5--7.^[@R8]^ Conversely, our data indicate that many patients with a normal or near normal brain noncontrast CT (i.e., ASPECTS 8--10) might already have a large area of established infarction. Compared to the optimum CT perfusion cutoff value of ≥50 mL, use of ASPECTS doubled the number of patients erroneously categorized as small volume established lesions. These findings could not be explained by the relative insensitivity of ASPECTS in the very early phase of ischemic stroke.
Although the overall agreement on the ASPECTS score was good, absolute agreement between raters was low and we found significant heterogeneity between individual raters (see supplemental data). These findings are consistent with a recent large meta-analysis and may further complicate the use of the ASPECTS score.^[@R19]^ Where interrater variability limits the use of ASPECTS scoring for selection of patients for reperfusion therapy, CT perfusion software automatically creates core and penumbra maps and calculates core and penumbra volumes that do not need further interpretation.
The accuracy of CT perfusion core volume to predict a large MRI DWI lesion was excellent, and the optimum cutoff volume of ≥50 mL was highly specific and resulted in 93% of patients correctly classified. We did, however, see a tendency to underestimate the volume of large DWI lesions, resulting in a lower optimum volume threshold. The stronger correlation between the CT perfusion and DWI lesion volume may in part explain the enhanced absolute and relative benefit seen in the randomized trials on endovascular clot retrieval where CT perfusion was used in patient selection as compared to CT angiography alone or CT angiography and ASPECTS: CT perfusion patient selection yielded the highest number of patients without disability 3 months postintervention (71% vs 32.6%--60%).^[@R3][@R4][@R5],[@R16],[@R20],[@R21]^ In our population, accuracy to predict clinical outcome was only slightly higher for CT perfusion when compared to ASPECTS. As opposed to the abovementioned trials on endovascular treatment, a large proportion of patients in this study did not achieve reperfusion.
Our study has some limitations and caveats. First, the study sample size is low, which may render us underpowered to find a significant difference between CT perfusion and ASPECTS. Furthermore, the limited number of patients with low ASPECTS scores (15% of patients had an ASPECTS score lower than 5) does not allow us to comment on the accuracy of ASPECTS scores 0--4 to detect very large volumes of established infarct. Second, ASPECTS training differed significantly between individual raters, which is reflected in the interrater reliability results. However, significant heterogeneity in ASPECTS experience is to be expected in daily practice and has been demonstrated to further limit the usefulness of the ASPECTS score.^[@R18]^ Third, we were not able to evaluate the accuracy of multiphase CT angiography. Finally, imaging techniques have evolved during the study period, which may affect the generalizability of our findings. Subanalysis of our results, although limited by the small sample size, does not show a significant difference in patients imaged with a third- vs fourth-generation helical CT scanner (figure e-4).
Although the accuracy of ASPECTS and CT perfusion to evaluate a large area of established infarction was similar, the lower false-positive and false-negative rate seen with CT perfusion makes it a more suitable tool for patient selection for IV therapy. Furthermore, CT perfusion has the additional benefit of providing information about the volume and location of the tissue at risk. Although there currently is insufficient evidence to exclude a subgroup of acute ischemic stroke patients from reperfusion treatment, our findings suggest CT perfusion may be more reliable and accurate than ASPECTS to further evaluate this query.
Supplementary Material
======================
###### Data Supplement
###### Accompanying Editorial
Supplemental data at [Neurology.org](http://neurology.org/lookup/doi/10.1212/WNL.0000000000004028)
Editorial, page 2242
AUTHOR CONTRIBUTIONS
====================
Jelle Demeestere: study concept and design, data acquisition, statistical analysis, interpretation of data. Carlos Garcia-Esperon: study concept and design, data acquisition, interpretation of data. Pablo Garcia-Bermejo: data acquisition, critical revision of manuscript for intellectual content. Fouke Ombelet: data acquisition. Pactrick McElduff: statistical analysis. Andrew Bivard: study concept and design, interpretation of data, critical revision of manuscript for intellectual content. Mark Parsons: data acquisition, interpretation of data, critical revision of manuscript for intellectual content. Christopher Levi: study concept and design, interpretation of data, critical revision of manuscript for intellectual content.
STUDY FUNDING
=============
No targeted funding reported.
DISCLOSURE
==========
The authors report no disclosures relevant to the manuscript. Go to [Neurology.org](http://neurology.org/lookup/doi/10.1212/WNL.0000000000004028) for full disclosures.
ASPECTS
: Alberta Stroke Program Early Computed Tomography Score
AUC
: area under the receiver operating characteristic curve
DWI
: diffusion-weighted imaging
ICC
: intraclass correlation coefficient
IQR
: interquartile range
NPV
: negative predictive value
PPV
: positive predictive value
[^1]: These authors contributed equally to this work.
[^2]: Go to [Neurology.org](http://neurology.org/lookup/doi/10.1212/WNL.0000000000004028) for full disclosures. Funding information and disclosures deemed relevant by the authors, if any, are provided at the end of the article. The Article Processing Charge was funded by The University of Newcastle, Australia.
| {
"pile_set_name": "PubMed Central"
} |
INTRODUCTION
============
Protein interactions take place physically between interface residues of two complementary proteins. Studies focusing on protein interfaces have revealed that binding energies are not uniformly distributed along the protein interfaces. Instead, there are certain critical residues called 'hot spots'. These residues comprise only a small fraction of interfaces yet account for the majority of the binding energy ([@B1; @B2; @B3]). These residues are observed to be critical for function and stability of the protein association ([@B1]). There are several sites collecting the experimental hot spots. Thorn and Bogan ([@B4]) deposited hot spots from alanine scanning mutagenesis experiments, in a database called ASEdb. BID is an effort to organize protein interaction data compiled from the literature and presents amino acids at the protein--protein binding interfaces ([@B5]). Yet, these servers provide hot spots for only a limited number of proteins.
Computational methods can introduce alternative approaches to experimental techniques to detect and catalog hot spots ([@B6]). Several groups have developed energy-based methods to predict hot spots ([@B7; @B8; @B9]). Molecular dynamics studies can also be used to investigate the energetic contributions of interface residues ([@B10; @B11; @B12]). While both energy and MD-based methods are very efficient, they are at the same time costly and not applicable in large-scale hot spot prediction.
Residues in protein interfaces ([@B13]) and functional sites ([@B14]) were observed to be mutating at a slower pace compared to the rest of the protein surface. There are several studies focusing on the detection of hot spots based on conservation. A very recent study based on sequence environment and evolutionary profile of residues predicts computational hot spots ([@B15]). Correlation between hot spot residues and structurally conserved residues were found to be remarkable ([@B16; @B17; @B18; @B19]). These hot spots are also found to be buried and tightly packed with other residues ([@B18]) resulting in densely packed clusters of networked hot spots, called '*hot regions*'.
Here, we present HotSprint, a database documenting computational hot spots in the protein interfaces combining conservation and solvent accessibility of residues in the protein interfaces. HotSprint contains protein interfaces extracted from the structures in Protein Data Bank (PDB) and is the first database, to our knowledge, which exploits sequence conservation to detect hot spots on a large scale. Total 49 512 interfaces are extracted from 34 817 PDB entries as of February 2006. Conserved residues of 35 776 protein interfaces are found using Rate4Site algorithm ([@B20]). NACCESS is used to obtain the solvent accessibility of residues ([@B21]). In summary, HotSprint marks residues that are highly conserved and tightly packed in protein interfaces as hot spots.
METHODOLOGY AND RESULTS
=======================
Interface datasets
------------------
The interfaces, used for the identification of the computational hot spots in the HotSprint, are taken from the updated version of interface dataset generated by Keskin *et al.* ([@B22]). Interfaces were generated by the atomic distance criteria: if the distance between any atoms of two residues, one from each chain, is less than the summation of their van der Waals radii plus a tolerance 0.5 Å, these residues are named as interface residues. If the distance between non-interacting and interacting residues in the same chain is smaller than 6 Å, the non-interacting residue is named a 'nearby' (neighboring) residue. Nearby residues are important for the information about the architecture of the interface and provided in our database. All 15 268 multi-chain PDB structures are used to extract two chain interfaces and then interfaces having less than 10 residues are eliminated. The resulting dataset contains 49 512 two-chained interfaces that are denoted by six-letter nomenclature where the first four letters denote the PDB ID, and the last two letters are the chain identifier.
Detection of computational hot spots in protein interfaces
----------------------------------------------------------
HotSprint database can be accessed through a web interface where users can search for computational hot spots in protein interfaces. The evolutionarily conserved residues are found by Rate4Site algorithm ([@B20]). Rate4Site makes use of topology and branch lengths of the phylogenetic trees constructed from multiple sequence alignments (MSA) of proteins and estimates conservation rates of amino acids based on the empirical Bayesian rule. MSAs of proteins constituting interfaces are taken from HSSP (Homology-Derived Secondary Structure of Proteins) ([@B23]) database as of 14 January 2006. All MSAs obtained from HSSP are converted to FASTA format to be used in Rate4Site step. In addition, some residues are more frequently observed to be hot spots, so each of the 20 amino acids has a different propensity to be a hot spot. Hot spot propensities are used to rescale the conservation scores. Further, hot spots prefer to reside in protein cavities ([@B24]), therefore surface area accessibility of interface residues are incorporated into our hot spot scoring formula.
The computational hot spot score of *i*th residue in a chain is defined as *pScore~i~* = *score~i~ x P~k~*, where *score~i~* is the conservation score from Rate4Site ([@B25]), *P~k~* is the propensity of residue type *k* (i.e, *k* = ALA, VAL, etc.) to be conserved in the interface (details are given in the Supplementary Data). For an amino acid in a protein interface to be considered as a computational hot spot, we propose that following formulation should be satisfied:
*pScore*~i~ \> *t* and ΔASA \> *t*~ASA~ and ASA~complex~ \< *t*~ASA*x*~
where *t, t*~ASA~ and *t*~ASAx~ are user-defined thresholds, the default values are set to 6.2, and 49 and 12 Å^2^, respectively. ΔASA is the ASA change of the residue upon complexation, ΔASA = ASA~monomer~ − ASA~complex~, ASA of the residue in the monomer and complex form, respectively. In ASA calculations, NACCESS ([@B21]) is used and buried ASAs of interface are calculated for each interface. Thus, this formulation combines amino acid conservation scores obtained from Rate4Site \[scaled with amino acid conservation propensities (e.g. aromatic residues are observed to be hot spots independent of their sequence position)\] and ASA of the residue. [Figure 1](#F1){ref-type="fig"} summarizes the flowchart to detect computational hot spots in interfaces. Figure 1.The flowchart of the procedure to predict hot spots and deposit them in the HotSPrint.
We have evaluated prediction performance of our formulation by comparing the results with the experimental hot spot data extracted from ASEdb ([@B4]). We assessed success of the formulations using the statistical analysis using 'Accuracy' and 'f-measure'. Our formulation yields 76.83%, 60.1%, 86.56%, 63.06% and 65.69% for accuracy (percentage of correctly predicted hot spot and non-hot spot residues over all interface residues), sensitivity (ratio of correctly predicted hot spots to all hot spots residues on the interface), specificity (ratio of correctly predicted non-hot spots to all non-hot spot interface residues), positive predictive value (number of correctly predicted hot spots divided by number of interface residues predicted as hot spot) and f-measure \[2 × sensitivity × ppv/(sensitivity + ppv) where ppv is the positive predictive value\], respectively. Ofran and Rost recently developed a sequence environment and evolutionary profile-based method to predict computational hot spots ([@B15]). They considered residues contributing ≥2.5 kcal/mol as hot spots. When we adopt the same convention, their positive predictive value (referred as positive accuracy in their text) of ∼60%, outperforms ours (∼46%). However, our sensitivity (57%, coverage in their text) is remarkably higher than theirs (15%).
Web interface and querying the HotSprint database
-------------------------------------------------
HotSprint provides an easy query screen with three distinct query boxes: (i) hot spot search in protein interfaces for a given PDB ID, (ii) advanced search box and (iii) conservation and ASA querying of the complete protein (including non-interface residues). The computational hot spots in the interfaces can be identified based on one of the three options mentioned in Supplementary Data. One may either choose (i) the default hot spot criterion as defined in the Methods section (*pScore* + ASA, conservation score rescaled with conservation propensity + contribution of ASA), (ii) only conservation criterion (score) or (iii) conservation score rescaled with conservation propensity (*pScore*) in the query page.
The first query box allows the user to fetch associated interfaces of a given protein using its PDB identifier. The default thresholds in these expressions can also be modified by the user. If there exists only a single interface associated with the input PDB identifier (e.g. for PDB ID: 1axd), then information for that interface (1axdAB) is displayed. However, there may be more than one interface extracted from that protein. In this case, interface identifiers of interfaces associated with that PDB are displayed (e.g. for the PDB ID 1yp2, four interfaces are available 1yp2AB, 1yp2AD, 1yp2BC and 1yp2CD). When one selects one of the interface identifiers listed, information for that interface is presented. [Figure 2](#F2){ref-type="fig"} demonstrates the result page yielded after querying the interface 1yp2AB among the associated interfaces of 1yp2. Figure 2.Interface information page for 1yp2AB Interface. Overall properties (number of computational hot spots, number of conserved residues, average conservation score, buried ASA and a link to interface information in the original dataset), individual residues and graphical representation of the interface are all displayed in this page. Using the link to the original dataset, users can get detailed information about interfaces: whether it is a biological or crystal interface, and interface amino acid composition. The graphical representation part contains snapshots of the interface and its hot spots from four different perspectives and a Jmol plugin is loaded in a new window when these images are clicked.
The page presenting interface information consists of three main sections. In the first section, overall properties of the interface such as number of computational hot spots on the interface, number of conserved residues on the interface, average conservation score of interface residues and buried ASA of the interface are presented. The next section lists residues of the interface along with their position, name, conservation score, ASA in monomer, ASA in complex, type (contacting interface residue, neighboring interface residue or none). A residue is highlighted with a red background if it is a computational hot spot. Static snapshots of the interface from four different perspectives are shown using Rasmol ([@B26]) at the bottom of the page ([Figure 3](#F3){ref-type="fig"}). It is possible to include only contacting residues in the presented results using the check box at the bottom of the query box. Figure 3.One of the four snapshots displayed in HotSprint generated by Rasmol for interface 1yp2AB. An interface is composed of two sides (chain A and chain B of potato tuber ADP-glucose phyrophosphorylase with PDB ID 1yp2) from two interacting proteins. Interface residues are shown as balls whereas the rest of the protein is shown as the trace. The purple and red residues represent interface residues of the A and B chains of the interface, respectively. The yellow and green residues are predicted hot spots on the chains A and B, respectively.
The second query box allows advanced search with different options. One can find structures satisfying given criteria among all the structures stored in the database. Interfaces with certain number of computational hot spots, number of conserved residues and average conservation score can be fetched. Furthermore, one may also be interested in finding interfaces with specified conserved propensities or buried accessible surface areas (ASA) in a given range. For example, if interfaces with more than seven hot spots and which have 1000 Å^2^ ≤ ASA ≤ 2000 Å^2^ are queried, a table listing the interface IDs with respective properties is provided.
At the bottom resides the final query box that can be used to access residue information (position, name, conservation score, monomer ASA) of the whole protein including both the interface and non-interface residues. The results for the given structure identifier will be output by the server.
As a case study, we compare the experimental hot spots of the numb PTB domain with HotSprint predictions. [Figure 4](#F4){ref-type="fig"} displays the ribbon diagram of the numb PTB domain that is in complex with numb-associated kinase (NAK)-C (PDB ID: 1ddm) ([@B27]). Numb PTB domain is known to interact with a diverse set of peptides through a large hydrophobic cavity on its surface ([@B28]). The left figure presents the predicted hot spots by using *pScore* only, whereas the right panel illustrates the results when the *pScore* + ASA is used. Red and yellow residues are the identified as hot spots by alanine scanning substitutions on the protein complex. Considering only propensity scaled conservation scores of the residues (left figure) in the interface of 1ddmAB, 8 of the 10 experimentally identified hot spots (red residues) are predicted computationally. Including ASA further filters some of the hot spot predictions (5 of the 10 hot spots are predicted). Figure 4.View of numb protein phosphotyrosine binding (PTB) domain. Red and yellow residues are experimental hot spots. Red residues are correctly predicted by HotSprint. Left and right figures present the results for the prediction of hot spots using *pScore* and *pScore* + ASA, respectively. VMD ([@B29]) is used to graphically represent the protein.
CONCLUSION
==========
In this article, a database of computational hot spots in protein interfaces (HotSprint) is introduced. 49 512 protein interfaces are extracted from the 34 817 structures in Protein Data Bank (PDB) as of February 2006. Conserved residues are mapped to the interfaces. We defined a hot spot as an interface residue that is conserved and buried in the complex form. Conserved residues of 35 776 protein interfaces deposited in the HotSprint. It is the first database, to our knowledge, which exploits sequence conservation to detect hot spots on a large scale. HotSprint highlights the residues that are highly conserved and tightly packed in protein interfaces. We believe study and characterization of hot spots will help to unravel insights of protein associations and will constitute an important step in understanding recognition and binding processes.
AVAILABILITY
============
HotSprint is available at <http://prism.ccbb.ku.edu.tr/hotsprint>. The dataset can be downloaded as a single SQL file from the website. A non-redundant subset of the database (40% homology with respect to BLAST) is also provided for retrieval.
This project has been funded in part with TUBITAK (Research Grant No 104T504) and O.K. has been granted with Turkish Academy of Sciences Young Investigator Programme (TUBA-GEBIP). The Open Access publication charges for this article were waived by the Oxford University Press.
*Conflict of interest statement*. None declared.
| {
"pile_set_name": "PubMed Central"
} |
Introduction {#s1}
============
*Reputations*, information that signals the potential cooperativeness of others (Tennie et al., [@B62]), are known to facilitate cooperation (Milinski et al., [@B41]; Feinberg et al., [@B22]). Due to their impact, however, reputations are subject to distortion (e.g., spreading lies or gossip) by others seeking to flatter allies and tarnish rivals (Mayzlin, [@B39]). Although this fact underscores the importance of disregarding unfounded reputations, this ability appears limited. Bad reputations are difficult to ignore and can continue to affect one's judgment even after they are shown to be false (Suzuki et al., [@B60]). The persistent effects of bad reputations are problematic given their power to cause avoidance (Chevalier and Mayzlin, [@B14]) and ostracism (Feinberg et al., [@B22]) of the target individuals. Elucidating the neural mechanisms underlying this persistence is therefore of theoretical and practical interest.
There has been growing interest in the neural basis of reputation processing (Frith and Frith, [@B26]; Izuma, [@B30]), and a few studies have investigated brain mechanisms underlying the difficulty of "unlearning" reputations through reinforcement learning based on social interactions. For example, Delgado et al. ([@B18]) measured brain activity while participants were playing an iterated trust game and examined how it was modulated by the presence or absence of prior reputation about trading partners. When no reputation was available, the striatum showed greater responses to the partner's cooperation than to cheating; when good or bad reputation was provided in advance, however, such differential striatal activity depending on the partner's behavior diminished. These findings have been elaborated by Fouragnan et al. ([@B24]) who conducted a similar experiment and analyzed the data using a computational model of reinforcement learning. They demonstrated that prior reputation about partners attenuated striatal activity in response to the trial-by-trial prediction error---the difference between the expected value of trusting a partner and the actual outcome from having trusted (i.e., cooperation \[reward\] or cheating \[loss\])---during a trust game.
In contrast, the neural underpinnings of the failure to intentionally disregard false reputations after being given verbal instructions that undermine their credibility (Suzuki et al., [@B60]) remain poorly studied. The present study approached this issue from the perspective of *evaluation transfer* in evaluative conditioning (Martin and Levey, [@B38]; Jones et al., [@B31]; Gawronski and Bodenhausen, [@B27]; but see also Hofmann et al., [@B28]). This refers to the transfer of the evaluation from an unconditioned stimulus (US) to a conditioned stimulus (CS).[^1^](#fn0001){ref-type="fn"}
For instance, suppose you are told that Ken, a bank employee, embezzled money from client accounts. A subsequent encounter with Ken will remind you of his embezzlement, and you are likely to conclude that Ken is untrustworthy because of his cheating behavior. In addition, reputation learning may also cause the transfer of the evaluation from the reputation (US) to the target individual (CS) such that the person acquires an ability to activate a positive or negative evaluation on their own. That is, the negative evaluation made about embezzlement becomes associated with Ken himself and consequently Ken alone generates a negative evaluation.
In neural terms, evaluation transfer would be operationalized as the CS alone activating evaluation-related brain structures. In general, item evaluation is considered to be an essential component of emotional processing (Russell, [@B53]). Thus, many cortical and subcortical structures linked with emotions are assumed to be involved in the evaluation of stimuli, including ventral portions of the prefrontal cortex (Kringelbach and Rolls, [@B35]), anterior parts of the insular (Singer et al., [@B58]) and cingulate cortices (Rushworth and Behrens, [@B52]), the amygdala (Morrison and Salzman, [@B42]), and the striatum (O'Doherty, [@B44]). Of particular relevance to reputation learning, these neural structures have been implicated in the evaluation of others' behavior in previous studies (Sanfey et al., [@B56]; Delgado et al., [@B18]; Buckholtz et al., [@B10]; Rilling et al., [@B50]; Schiller et al., [@B57]; Mende-Siedlecki et al., [@B40]). Thus, after learning one's reputation, the target individual may activate the evaluation-related brain region on his/her own, constituting a neural correlate of evaluation transfer. More specifically, with regard to bad reputations, the involvement of the lateral and ventral portions of the prefrontal cortex and the anterior insula might be expected since these regions have been implicated in anger and disgust (Murphy et al., [@B43]; Vytal and Hamann, [@B65]; but see also Lindquist and Barrett, [@B37]), which are negative emotions closely related to appraisals of harmfulness and immorality (Hutcherson and Gross, [@B29]).
A hallmark of evaluation transfer is the difficulty with which one can intentionally negate its effects (Gawronski and Bodenhausen, [@B27]). Suppose that Ken's reputation as an embezzler is later found to be invalid. Then, you will not reason that Ken is untrustworthy from the inaccurate reputation that he embezzled money. However, if evaluation transfer has occurred, Ken will still elicit negative evaluations independently of the validity of his reputation, and therefore, will continue to be distrusted. It has indeed been reported that the effect of evaluation transfer cannot be neutralized voluntarily (Sweldens et al., [@B61]; Balas and Gawronski, [@B4]), and that negative evaluations are especially transferable (Rydell and Jones, [@B54]; Bell et al., [@B8]; Campbell and Warren, [@B11]). It can therefore be hypothesized that the persisting effects of a bad reputation are related to a transfer of the negative evaluation about the bad reputation to the target person. Here, we report a functional magnetic resonance imaging (fMRI) experiment testing this hypothesis.
In this fMRI study, participants memorized faces paired with either a good or a bad reputation. Next, they viewed the faces alone and inferred whether each person would be likely to cooperate, first while retrieving the memorized reputations and then while trying to disregard them as false. If reputation learning transfers the negative evaluation of bad reputations to target persons, and if the transfer is related to the persisting distrust, the following two predictions would be made. **Prediction 1**: Face stimuli that are paired with bad reputations during a learning task will activate an evaluation-related brain region during an inference task, irrespective of whether participants attempt to retrieve or disregard the reputations. **Prediction 2**: Participants showing higher activity of the region described in Prediction 1 will infer that the persons with bad reputations are less likely to cooperate, irrespective of whether participants attempt to retrieve or disregard the reputations.
Materials and Methods {#s2}
=====================
Participants {#s2-1}
------------
Thirty-two undergraduate and graduate students (18 males and 14 females; age 20--31 years) gave informed consent to participate in this study, which was approved by the Ethics Committees of the National Center for Geriatrics and Gerontology, Japan, and the Graduate School of Environmental Studies, Nagoya University, Japan. Four participants (2 males and 2 females) were excluded from analyses because one withdrew due to fatigue, two expressed suspicion about a cover story for the experiment, and one showed perfect performance in the good-reputation condition of the baseline session of action inference.[^2^](#fn0002){ref-type="fn"}^,^[^3^](#fn0003){ref-type="fn"}
Stimuli {#s2-2}
-------
Twenty-four neutral faces of Japanese individuals (12 males and 12 females) from the Facial Information Norm Database (Watanabe et al., [@B66]) were used as stimuli. They were divided into three groups and each was assigned to one of the three conditions: *good-*, *bad-*, and *no-reputation*. The assignment of reputations to faces was counterbalanced across participants. The three groups of faces were matched for number of males and females and mean trustworthiness rating (1, *very untrustworthy*, to 5, *very trustworthy*) assessed in a preliminary survey with 102 participants (*M* = 2.96, 2.96, 2.97; unpublished data). All face stimuli were presented in gray scale and were cropped into square shapes (270 × 270 pixels) so that only the central facial features (eyes, eyebrows, nose, and mouth) were visible.
Experimental Procedure {#s2-3}
----------------------
The experiment consisted of four tasks in the following order: baseline session of action inference, reputation learning, retrieval, and disregard sessions of action inference (Figure [1](#F1){ref-type="fig"}). The baseline session of action inference was performed on a laptop computer (HP ProBook 4740s, Hewlett-Packard Japan, Ltd., Tokyo, Japan) outside an MRI scanner. The other tasks were administered inside the scanner, with a short rest outside of the scanner after completion of the reputation-learning task.[^4^](#fn0004){ref-type="fn"} Inside the scanner, stimuli were presented with VisuaStim digital goggles (Resonance Technology, Inc., Northridge, CA, USA), and responses were collected via bimanual response pads (Current Designs, Inc., Philadelphia, PA, USA). Throughout the experiment, E-Prime 2.0 (Psychology Software Tools, Inc., Pittsburgh, PA, USA) was used to run the task. The display resolution was set to 800 × 600 pixels.
{#F1}
### Baseline Session of Action Inference {#s2-3-1}
Participants were presented with a cover story about the experiment. They were instructed that they would see the faces of unfamiliar people who had taken part in a two-player "investment game" in a previous (fictitious) study. It was explained that in this game, one player ("lender") is provided with 500 yen (about 5 USD) and decides whether to invest the money in the other player ("borrower"). When 500 yen is invested, the money is quadrupled to 2000 yen, and the borrower decides whether to return a 1000 yen dividend to the lender or to embezzle the whole amount. Participants in the current study were informed that the persons whose faces they would see during the present experiment had played the investment game as a borrower previously, and had received a 500 yen investment from their partners. The participants' task was to infer intuitively, based on their impressions of the faces, whether each person had returned the dividend (pressing the "F" key with their left index finger) or embezzled the investment (pressing the "J" key with their right index finger). The faces were presented until participants responded with no time limit. The task was repeated in two blocks, and each of the 24 faces was presented once per block.
### Reputation Learning {#s2-3-2}
The time course of this task is schematized in Figure [2A](#F2){ref-type="fig"}. In each trial, a face was presented for 5 s along with one of the following labels: "Returned," "Embezzled," and "\# \# \# \#." The words "Returned" and "Embezzled" ostensibly indicated that the person had returned (*good-reputation condition*) and embezzled (*bad-reputation condition*) the investment in the previous experiment. The symbolic label of "\# \# \# \#" indicated that whether the person had returned or embezzled was being kept confidential (*no-reputation condition*). Below each face were also displayed the words "Male" and "Female." Participants were instructed to memorize whether the displayed person had returned or embezzled the money, while concurrently indicating whether they were male (pressing a left-hand pad with their index finger) or female (pressing a right-hand pad with their index finger) as soon as possible. Sex identification was imposed in order to maintain participant attention.
{#g7}; Embezzled = {#g8}; Male = {#g9}; Female = {#g10}; Ret = {#g11}; Emb = {#g12}.](fnhum-10-00028-g0002){#F2}
The task was comprised of three fMRI runs that lasted for 305 s each. Each run started and ended with fixation periods that lasted for 20 and 15 s, respectively, and included 24 trials in between. Each of the 24 faces appeared once per run, and their presentation order was randomized. Between any two trials was a fixation interval the duration of which was randomly set to either 5 or 10 s, with a mean of 6.25 s.
### Retrieval Session of Action Inference {#s2-3-3}
The time course of this task is schematized in Figure [2B](#F2){ref-type="fig"}. In each trial, a face was presented for 5 s along with the abbreviated labels for "Returned" and "Embezzled" at the bottom. The participants' task was to answer as quickly and as accurately as possible whether the person had returned (pressing a left-hand pad with their index finger) or embezzled (pressing a right-hand pad with their index finger) the money by recalling the reputations memorized during the previous task. For individuals with no reputation, participants were asked to make intuitive, face-based judgments, just as during the baseline session.
The task was comprised of two fMRI runs that lasted for 305 s each. Each of the 24 faces appeared once per run, and their presentation order was randomized. The durations of the fixation periods at the start and end of each run and between trials were the same as those in the reputation-learning task.
### Disregard Session of Action Inference {#s2-3-4}
The time course of this task was the same as that of the retrieval session of action inference. Prior to the task, an apology was given to participants, explaining that although the stimulus persons had previously taken part in the investment game as a borrower, the previously presented reputations about them were completely unrelated to their actions. Then, participants were told to disregard these invalid reputations and perform the action inference task again on the basis of their impressions of the faces. The task was comprised of two fMRI runs, and the timeline of each run was the same as that of the retrieval session.
After the completion of this task, participants were asked if they had any doubts regarding the experimental procedure. Two participants spontaneously mentioned their suspicions about the cover story that the stimulus individuals had previously taken part in the investment game, and thus their data were excluded from analyses.
Imaging Protocol {#s2-4}
----------------
MR images were acquired on a 3T scanner (Siemens MAGNETOM Trio, Erlangen, Germany) with a 12-channel head coil. Functional images were acquired using a T2\*-weighted gradient echo planar imaging sequence with the following parameters: repetition time (TR) = 2500 ms, echo time (TE) = 30 ms, flip angle (FA) = 90°, matrix 64 × 64, field of view (FOV) = 192 mm, 39 slices, slice thickness = 3 mm, distance factor = 17%, and slice acquisition order = ascending. Following the completion of the disregard session of action inference, a high-resolution, magnetization-prepared, rapid-acquisition gradient echo (MPRAGE) image was also acquired for anatomical details (TR = 2500 ms, TE = 2.63 ms, FA = 7°, matrix 256 × 256, FOV = 256 mm, 208 slices per slab, slice thickness = 1 mm, and distance factor = 50%).
Image Preprocessing {#s2-5}
-------------------
Reputation learning, retrieval, and disregard sessions of action inference involved three, two, and two functional runs, respectively. In each run, 122 functional images were acquired, of which the first two images were discarded to allow for T1 equilibrium. The remaining functional images were preprocessed with Statistical Parametric Mapping 8 (SPM8, Wellcome Trust Centre for Neuroimaging, London, UK) implemented in MATLAB R2013a (The Mathworks, Inc., Natick, MA, USA). For each participant, the images from all tasks and runs were realigned and resliced to the mean image to correct for head movement. Slice-timing correction was also performed using the twentieth slice as a reference. All functional images and the MPRAGE anatomical image were then co-registered to the mean image of the retrieval session of action inference. The co-registered anatomical image was normalized to a standard T1 template image (ICBM 152), which defined the Montréal Neurological Institute (MNI) space. The parameters from this normalization process were then applied to all functional images. Finally, the normalized functional images were spatially smoothed with an isotopic Gaussian kernel of 6 mm full-width at half-maximum.
fMRI Data Analysis {#s2-6}
------------------
Data from the fMRI were analyzed using SPM8. To depict the neural substrates of the tasks, we employed a summary statistics approach. In the individual-level analysis, a voxel-by-voxel general linear model (GLM) was applied to preprocessed functional images for each of the three in-scanner tasks separately. The design matrix of the model contained three regressors of interest (good-, bad-, and no-reputation conditions) to obtain parameter estimates for each reputation condition. The regressors of interest were created by convolving a delta function (0 s duration), representing trial onset times of each reputation condition, with a canonical hemodynamic response function (Friston et al., [@B25]). Moreover, when analyzing the reputation-learning data, parametric modulation regressors were included in the good- and bad-reputation conditions to remove the effect of the unexpectedness of the reputation. For example, for each reputation-learning trial of the good-reputation condition, the unexpectedness of the good reputation was computed as follows: Unexpectedness = Number of "Embezzled" responses for the displayed person in the baseline session of action inference (i.e., 0, 1, or 2) × 0.5^Run\ number\ −\ 1^. That is, we assumed a decrease of the unexpectedness across runs.[^5^](#fn0005){ref-type="fn"} The vector of the computed values was entered as parametric modulation regressors for the good-reputation condition. The unexpectedness of the bad reputation was computed in the same way by using the number of "Returned" responses.
In the group-level analyses, we first explored brain regions that were activated in response to good and bad reputations during the reputation-learning task. The individual-level analysis of the reputation-learning data produced the contrast images from each of the good- and bad-reputation conditions, where the effect of the unexpectedness of the reputation was controlled for, as well as the contrast images from the no-reputation condition. These individual-level contrast images for each reputation condition during reputation learning were submitted to random-effects GLM analysis. The design matrix contained three regressors of interest (good-, bad-, and no-reputation conditions) to obtain parameter estimates for each reputation condition, as well as each participant's mean response times (RTs) in each condition as covariates to accommodate RT differences between conditions (see "Behavioral Data" Section).
Then, clusters of voxels that were significantly active in the contrast images of good vs. no reputation and bad vs. no reputation were identified as the regions of interest (ROIs) likely related to positive and negative evaluation, respectively. The statistical threshold was set at *p* \< 0.05 (family-wise error \[FWE\] corrected) at the voxel level with no less than 20 contiguous voxels.
In order to test Prediction 1, we examined whether the negative-evaluation-related ROIs were significantly activated by the faces with bad reputations during the retrieval and disregard inference sessions.[^6^](#fn0006){ref-type="fn"} For this, ROI analysis was performed on the data from each session using MarsBaR toolbox for SPM (Brett et al., [@B9]). Specifically, parameter estimates were extracted and averaged across voxels in each ROI. The design matrix was the same as the one used in the analysis of the reputation-learning data. We tested whether each ROI activity, defined as the mean of the voxel values within it, was significantly greater in the bad-reputation cases as compared to the no-reputation cases. Similar analyses were also conducted for positive-evaluation-related ROIs.
In order to test Prediction 2, we examined the relationship between the activity of each negative-evaluation-related ROI and behavioral inferences during the retrieval and disregard sessions. Linear mixed-model analysis (West et al., [@B67]) was performed to achieve this goal. We examined whether post-learning distrust toward the people with bad reputations (DISTRUST_POST) was statistically explained by pre-learning distrust (DISTRUST_PRE) and the activation of the ROI toward those people (ROI_ACT). To allow for different effects of pre-learning distrust and ROI activity across sessions, we fit two models, one without and the other with interaction terms involving the session, and then selected the best-fit model. Specifically, the following two nested models were compared:
Model 1 (without interaction terms): $$\begin{array}{l}
{\text{DISTRUST}\_\text{POST}_{si}} \\
{\text{\quad } = \gamma_{00} + \gamma_{0\text{1}} \times \text{SESS}_{s} + \gamma_{\text{1}0} \times \text{DISTRUST}\_\text{PRE}_{i}} \\
{\text{\quad } + \gamma_{\text{2}0} \times \text{ROI}\_\text{ACT}_{si} + u_{i} + \varepsilon_{si}.} \\
\end{array}$$
Model 2 (with interaction terms): $$\begin{array}{l}
{\text{DISTRUST}\_\text{POST}_{si}} \\
{\text{\quad } = \gamma_{00} + \gamma_{0\text{1}} \times \text{SESS}_{s} + \gamma_{\text{1}0} \times \text{DISTRUST}\_\text{PRE}_{i}} \\
{\text{\quad } + \gamma_{\text{2}0} \times \text{ROI}\_\text{ACT}_{si}} \\
{\text{\quad } + \gamma_{\text{11}} \times \text{SESS}_{s} \times \text{DISTRUST}\_\text{PRE}_{i}} \\
{\text{\quad } + \gamma_{\text{21}} \times \text{SESS}_{s} \times \text{ROI}\_\text{ACT}_{si} + u_{i} + \varepsilon_{si}.} \\
\end{array}$$
The subscript *s* refers to the session of action inference (*s* = 1: retrieval session, *s* = 2: disregard session), and the subscript *i* denotes the *i*-th participant (*i* = 1, ..., 28). The dependent variable DISTRUST_POST~*si*~ indicates each participant's post-learning distrust toward the people that had been paired with a bad reputation, calculated as the difference in rate of judging that people had returned the investment between the no-reputation and bad-reputation conditions. A positive value of DISTRUST_POST~*si*~ means that the people with bad reputations were judged as *unlikely* to have returned the money. The same difference in the baseline session of action inference, labeled as DISTRUST_PRE~*i*~, was entered as a regressor to account for pre-learning distrust toward the people with bad reputation. ROI_ACT~*si*~ was the main regressor, representing each participant's activation of the ROI in response to the people with bad reputations (i.e., contrast estimate of bad- vs. no-reputation conditions) in each session. The other regressor, SESS~*s*~, was a dummy variable indicating the session of action inference, with 0 and 1 for the retrieval and disregard sessions, respectively (i.e., SESS~1~ = 0 and SESS~2~ = 1). *γ*'s were fixed-effect parameters to be estimated. *u~i~* and *ε~si~* indicate random effects associated with each participant and each observation, respectively. They were assumed to be independently and normally distributed with a mean of 0 and variance of $\sigma_{u}^{2}$ or *σ^2^*. The random effects of *u~i~* were included to account for the dependency between sessions due to repeated measurements of the same participants (Aarts et al., [@B1]).
With regard to the differences between the two models, Model 1 assumes that while the intercept of the regression model may vary between sessions (i.e., *γ*~00~ + *γ*~01~ × SESS~*s*~),[^7^](#fn0007){ref-type="fn"} the fixed effects of pre-learning distrust (*γ*~10~) and ROI activity (*γ*~20~) on post-learning distrust are common across sessions. On the other hand, Model 2 allows for different fixed effects of pre-learning distrust and ROI activity across sessions (*γ*~11~ and *γ*~21~, respectively) by including interaction terms, SESS × DISTRUST_PRE and SESS × ROI_ACT.
The fits of the two models to the data were compared using Akaike and Bayesian information criteria (AIC and BIC, respectively) as well as a deviance test (West et al., [@B67]; Snijders and Bosker, [@B59]). AIC and BIC are measures of the fit of data to a model (smaller values indicate a better fit) with a penalty for increased model complexity. Deviance is also a goodness-of-fit index but without a penalty for complexity; therefore, its value is *always* smaller (indicating a better fit) for a more complex model. The difference between the deviance scores of two nested models can be statistically tested because it is distributed asymptotically as chi-squared under the null hypothesis of no difference. A significant difference supports the more complex model, while nonsignificance favors the simpler, more parsimonious model. Thus the information criteria and deviance tests compensate for each other's weaknesses (i.e., lack of a statistical test and insensitivity to model complexity).
Model estimation was conducted using the lme4 (Bates et al., [@B5]) and lmerTest (Kuznetsova et al., [@B36]) R packages (R version 3.0.2). All continuous variables (i.e., DISTRUST_POST, DISTRUST_PRE, ROI_ACT) were centered and scaled before analysis so that standardized fixed-effect parameters were obtained.
Results {#s3}
=======
Behavioral Data {#s3-1}
---------------
### Sex Identification During Reputation Learning {#s3-1-1}
The accuracy of sex identification during the reputation learning task (*M* ± *SD* = 0.894 ± 0.048) did not significantly differ between the reputation trials, *F*~(2,54)~ = 0.003, *MSE* = 0.013, *p* = 0.997, $\eta_{p}^{2}$ \< 0.001. The RT of sex identification significantly varied based on reputation, *F*~(2,54)~ = 9.573, *MSE* = 0.115, *p* = 0.003, $\eta_{p}^{2}$ = 0.262, and was shorter in the no-reputation trials, *M* ± *SD* = 1.245 ± 0.346 s, than in both the good-, *M* ± *SD* = 1.518 ± 0.621 s, *t*~(27)~ = 3.213, *p* = 0.003, and bad-reputation trials, *M* ± *SD* = 1.495 ± 0.616 s, *t*~(27)~ = 3.136, *p* = 0.004.
### Action Inference {#s3-1-2}
Figure [3A](#F3){ref-type="fig"} shows the mean rate of "return" response, the rate at which the stimulus individuals were judged as having returned the investment, as a function of reputation condition and task session. The Reputation × Session interaction was significant, *F*~(4,108)~ = 42.470, *MSE* = 0.016, *p* \< 0.001, est $\eta_{p}^{2}$ = 0.611. *Post hoc* analyses showed that the main effect of Reputation was not significant in the baseline session, *F*~(2,54)~ = 0.248, *MSE* = 0.013, *p* = 0.777, est $\eta_{p}^{2}$ = 0.009, whereas it was significant in the retrieval session, *F*~(2,54)~ = 94.539, *MSE* = 0.026, *p* \< 0.001, est $\eta_{p}^{2}$ = 0.778, and in the disregard session, *F*~(2,54)~ = 3.855, *MSE* = 0.024, *p* = 0.027, est $\eta_{p}^{2}$ = 0.125. In the retrieval session, the positive difference of rate of "return" response in the good-reputation minus no-reputation trials was significant, *t*~(27)~ = 6.211, *p* \< 0.001, as well as the negative difference in the bad- minus no-reputation trials, *t*~(27)~ = 9.221, *p* \< 0.001, indicating the overall success of reputation learning. In the disregard session, the negative difference in the bad- minus no-reputation trials remained marginally significant, *t*~(27)~ = 2.013, *p* = 0.054, whereas the positive difference in the good-reputation minus no-reputation trials did not, *t*~(27)~ = 0.606, *p* = 0.549, replicating the persisting effect of bad reputations (Suzuki et al., [@B60]).
{#F3}
Figure [3B](#F3){ref-type="fig"} shows the mean RT in the action-inference task as a function of reputation condition and task session. A Reputation × Session interaction was significant, *F*~(4,108)~ = 5.380, *MSE* = 0.047, *p* = 0.002, est $\eta_{p}^{2}$ = 0.166. *Post hoc* analyses showed that the main effect of Reputation was not significant in either the baseline, *F*~(2,54)~ = 1.708, *MSE* = 0.037, *p* = 0.192, est $\eta_{p}^{2}$ = 0.060, or disregard sessions, *F*~(2,54)~ = 1.560, *MSE* = 0.027, *p* = 0.220, est $\eta_{p}^{2}$ = 0.055, whereas it was significant in the retrieval session, *F*~(2,54)~ = 8.075, *MSE* = 0.047, *p* = 0.001, est $\eta_{p}^{2}$ = 0.230. In the retrieval session, RT was significantly longer in the no-reputation trials than in both the good-reputation, *t*~(27)~ = 3.422, *p* = 0.002, and the bad-reputation trials, *t*~(27)~ = 3.248, *p* = 0.003. In addition, across reputation conditions the retrieval-session RT was longer than the baseline- and disregard-session RTs (all *p*'s \< 0.10).
fMRI Data {#s3-2}
---------
### Reputation Learning {#s3-2-1}
Brain regions showing significantly greater activation in the bad- as compared to the no-reputation conditions during reputation learning (*p* \< 0.05, FWE corrected at the voxel level; minimum cluster size 20 voxels) were the left ventrolateral prefrontal cortex (vlPFC; cluster size = 47 voxels, peak coordinates = \[−48, 24, 4\], *Z* = 5.258) and the left thalamus (cluster size = 31 voxels, peak coordinates = \[−4, −16, 8\], *Z* = 5.445). As described in the "Introduction" Section, ventral portions of the prefrontal cortex are implicated in the evaluation of stimuli, and, especially, the lateral regions are responsive to negative stimuli (Murphy et al., [@B43]; Kringelbach and Rolls, [@B35]; Elliott et al., [@B19]; Vytal and Hamann, [@B65]). Thus, the vlPFC activity likely reflects the negative evaluation of bad reputations. In contrast, the activity in the thalamus might reflect enhancement of perceptual processing by negative stimuli (Vuilleumier, [@B63]). We thus defined the left vlPFC region as the negative-evaluation-related ROI (Figure [4](#F4){ref-type="fig"}). With regard to the good-reputation trials, no brain region showed significantly greater activation as compared to the no-reputation trials. In addition, no brain region showed significantly greater activation for bad than for good reputations or*vice versa*.
{#F4}
Contrary to our expectation, the anterior insula did not show greater activity in response to bad as compared to no reputation. Instead, this region was activated across reputation trials (Figure [5](#F5){ref-type="fig"}). The anterior insular activity in the no-reputation trials might make sense considering that participants were not told whether the stimulus individuals had returned or embezzled the money in this condition. Thus, although speculative, participants might have perceived these trials as risky (e.g., the odds of having returned or embezzled were 50/50) or ambiguous (i.e., the odds were completely unknown), thereby activating the anterior insula (Singer et al., [@B58]).
{#F5}
We also explored brain regions whose activity increased with the degree of unexpectedness of the reputation. Although with a less stringent threshold compared to the analysis above (*p* = 0.001 at the voxel level, uncorrected; minimum cluster size = 20 voxels), such a trend was detected in the bilateral middle cingulate cortices (left: cluster size = 20 voxels, peak coordinates = \[−10, 8, 46\], *Z* = 3.920; right: 21 voxels, \[14, 4, 48\], *Z* = 4.154) and right middle temporal region (71 voxels, \[56, −34, −2\], *Z* = 3.963), which were shown to be sensitive to the need for performance adjustment (Ridderinkhof et al., [@B49]) and the error in predicting others' behaviors (Behrens et al., [@B6]), respectively. The bilateral fusiform gyri (left: 21 voxels, \[−28, −54, −14\], *Z* = 3.526; right: 33 voxels, \[30, −36, −10\], *Z* = 3.574) showed the same trend as well.
### Action Inference {#s3-2-2}
The mean activity of the voxels in the vlPFC ROI (Figure [4](#F4){ref-type="fig"}) was greater for the faces that had been paired with a bad reputation than it was for those with no reputation. This was marginally significant in the retrieval session (contrast value = 0.414, *Z* = 1.736, *p* = 0.087) and significant in the disregard session of action inference (contrast value = 0.707, *Z* = 2.717, *p* = 0.007). The mean vlPFC activity in the bad- vs. no-reputation conditions did not differ significantly between the retrieval and disregard sessions (*p* = 0.338), and its average across the two sessions was significantly greater than zero (*p* = 0.006).[^8^](#fn0008){ref-type="fn"} In addition to the ROI analysis, we conducted whole-brain conjunction analysis to explore regions showing greater activation in the bad- than no-reputation conditions both during reputation learning and during the two sessions of action inference. With a lenient statistical threshold (*p* = 0.05 at the voxel level, uncorrected; minimum cluster size = 20 voxels), this analysis also identified the vlPFC (cluster size = 100 voxels, peak coordinates = \[−48, 24, 4\], *Z* = 2.390).[^9^](#fn0009){ref-type="fn"} Overall, these results support Prediction 1. The activity in the left thalamus showing significantly greater responses in the bad- vs. no-reputation conditions during reputation learning was also subjected to the same ROI analysis, yielding non-significant results.
As no reputation was displayed during action inference, the abovementioned vlPFC activity in the bad-reputation condition suggests that the stimulus faces themselves might have acquired the capacity to elicit negative evaluation directly. In order to verify this possibility, linear mixed-model analysis was performed to examine whether vlPFC activity in response to the faces paired with bad reputations could explain distrust toward them (i.e., the lower rate of "return" responses). Table [1](#T1){ref-type="table"} compares the goodness of fit of the two regression models, Model 1 and Model 2, without and with an interaction term allowing for different relationships between vlPFC activity and distrust in the retrieval and disregard sessions (see "fMRI Data Analysis" Section for details of the models). Model 1 was selected because of its superiority in terms of both information criteria and model parsimony (West et al., [@B67]). Table [2](#T2){ref-type="table"} summarizes the parameter estimates for Model 1, demonstrating that vlPFC activity was a significant predictor of distrust. That is, during both the retrieval and disregard sessions of action inference, participants who showed larger vlPFC activity in response to faces previously paired with bad reputations inferred that these individuals would be less likely to cooperate (Figure [6](#F6){ref-type="fig"}). These results support Prediction 2 and are consistent with our interpretation that the vlPFC may be involved in negative evaluation.
######
**Fit indices for Models 1 and 2**.
AIC BIC Deviance
--------- --------- --------- ---------- ----------------------------------------------------------
Model 1 −83.011 −70.859 −95.011
Model 2 −80.202 −63.999 −96.202 $\text{χ}_{\text{(2)}}^{\text{2}}$ = 1.191, *p* = 0.5513
######
**Fixed- and random-effects estimates for Model 1**.
Fixed-Effects Estimates
------------------------------ --------------------------
*γ*~00~ (Intercept) 0.071, (0.029, 0.113)
*γ*~01~ (SESS) −0.143, (−0.187, −0.099)
*γ*~10~ (DISTRUST_PRE) 0.004, (−0.268, 0.277)
*γ*~20~ (ROI_ACT) 0.214, (0.014, 0.419)
**Random-Effects Estimates**
*σ~u~* 0.080, (0.040, 0.114)
*σ* 0.082, (0.063, 0.107)
*Note: The values are point estimates with 95% confidence intervals (in brackets)*.
![**Partial regression plot between vlPFC activity in response to the people with bad reputation (ROI_ACT) and distrust toward them (DISTRUST_POST) in the retrieval (Ret) and disregard (Dis) sessions of action inference**. The abscissa represents the residuals from the regression of ROI_ACT on DISTRUST_PRE and SESS, whereas the ordinate represents the residuals from the regression of DISTRUST_POST on the same two variables. The plot thus illustrates the marginal relationship between ROI_ACT and DISTRUST_POST after the effect of the other variables has been removed (Faraway, [@B20]). For reference, the dashed line shows DISTRUST_POST Residuals = 0.214 (point estimate of *γ*~20~ from Table [2](#T2){ref-type="table"}) × ROI_ACT Residuals. See "fMRI Data Analysis" Section for more detailed descriptions of each variable.](fnhum-10-00028-g0006){#F6}
We also performed linear model analysis to examine whether vlPFC activity *during reputation learning* in response to the faces paired with bad reputations could explain distrust toward them in the subsequent action inference tasks. Results showed that vlPFC activity during reputation learning was not significantly related to distrust in either the retrieval (*β* = 0.143, *p* = 0.356) or disregard (*β* = −0.055, *p* = 0.770) sessions. Although descriptive, vlPFC activity had less individual variability during reputation learning (coefficient of variability = 0.922) than it did during action inference (3.086 and 1.910 in the retrieval and disregard sessions, respectively), which might obscure its relationship with the behavioral measure.
Discussion {#s4}
==========
The present study yielded two main findings. First, a region of the left vlPFC was activated when participants were informed of the bad reputations of stimulus people, and more importantly, this same region was also activated by the subsequent encounters with those people, irrespective of whether participants attempted to retrieve or disregard those bad reputations. Second, in the both retrieval and disregard sessions of action inference, participants who showed greater activity of the vlPFC in response to faces that had been paired with bad reputations were more likely to infer that those people would not cooperate. These results overall are consistent with the idea that negative evaluations, which are here assumed to be related to vlPFC activity, are transferred from a bad reputation to the target person, and consequently, the person may continue to be distrusted irrespective of the validity of their prior reputation.
The vlPFC receives multimodal sensory information about external stimuli including visual, auditory, and somatosensory inputs (Price, [@B48]; Romanski, [@B51]). In addition, this region is contiguous with and interconnected to the orbitolateral prefrontal cortex representing internal signals about visceral reactions (Öngür and Price, [@B45]; Price, [@B48]). Visceral reactions can function as a signal of the emotional meaning or value of a certain stimulus for the organism, as they convey information regarding the demands of the organism to maintain homeostasis and satisfy basic needs (Damasio et al., [@B17]; Craig, [@B16]). The vlPFC may thus be able to associate the sensory representations of an external stimulus with its value. Activation in this region upon perceiving an external stimulus may reflect the decoding of the stimulus' associated value (i.e., evaluation of the stimulus). In fact, the vlPFC shows greater activation to emotional than neutral stimuli regardless of the stimulus type (i.e., faces and scenes; Sabatinelli et al., [@B55]) and has been reported to be particularly responsive to negative stimuli (Vytal and Hamann, [@B65]; Mende-Siedlecki et al., [@B40]). These findings support our interpretation that vlPFC activity in the present study is related to negative evaluation. Our arguments, which are subject to the problem of reverse inference (Poldrack, [@B46]), are corroborated by the data showing that vlPFC activity explained the negative evaluations of people with bad reputations.
One may argue that activity in the vlPFC during our action-inference tasks reflects participants' attempts to recall the reputations that had been paired with stimulus faces, rather than the activation of negative evaluations associated with the stimuli. In fact, it has been proposed that the vlPFC contributes to cognitive control of memory, the volitional retrieval and selection of task-relevant knowledge (Badre and Wagner, [@B3]). This interpretation, however, does not easily account for some aspects of the present data. First, vlPFC activity was not greater in the retrieval than in the disregard sessions. Behavioral results indicate that participants actually attempted not to retrieve the memorized reputations in the disregard session. Therefore, if the vlPFC was involved in the recall of reputations, it should have been less active during the disregard than retrieval sessions. Second, in both sessions, vlPFC activity was positively related to inferences congruent with the memorized reputations. If vlPFC activity reflected the recall of the memorized reputations, it should have been positively and *negatively* related to the inferences congruent with the reputations in the retrieval and disregard sessions, respectively, since the recall of reputations enables conscious correction of their biasing influences on judgments in the disregard session (Sweldens et al., [@B61]; Balas and Gawronski, [@B4]). These arguments against the alternative account, however, are admittedly not decisive. The action inference task is limited in that faces are always presented in the context of a judgment. It is important to demonstrate that the mere presentation of faces can activate the vlPFC after reputation learning to gain further support for the evaluation transfer view. Thus, a promising procedure for the future experiment might be to measure vlPFC activity while participants are not engaging in any evaluative task and relate it to subsequent evaluations.
Our claim for the involvement of the vlPFC in evaluation transfer does not preclude the possible role of conscious recall of reputations in performing the action inference task. Although speculative, longer RTs during the retrieval session as compared to the other sessions suggests that the performance was at least partly based on time-consuming, effortful retrieval processes. This might explain why in the conjunction analysis, vlPFC activity in the bad-reputation condition was detected only when a lenient statistical threshold was used. That is, in our data, learned associations between faces and bad reputations might not have been strong enough to robustly activate negative evaluation upon viewing faces alone, thereby resulting in only moderate vlPFC activity during action inference. Although we demonstrated significant vlPFC activity in ROI analysis, it would be important to see whether the results could be replicated with more conservative conjunction analysis when using a more intensive reputation learning task.
In addition to conjunction analysis, we were unable to obtain significant results from the analysis on the relationship between vlPFC activity during reputation learning and behavioral performance in action inference tasks. It is possible that the relationship might be masked by low individual variability in vlPFC activity during reputation learning. The low variability might make sense considering that the vlPFC ROI was selected using the data on reputation learning (i.e., the ROI contained only those voxels showing large signals during reputation learning), and that a bad reputation on embezzlement would be evaluated negatively by everyone. Thus, the use of a variety of (bad) reputations as in previous studies (Schiller et al., [@B57]; Bell et al., [@B8]; Mende-Siedlecki et al., [@B40]) could introduce more variability in vlPFC activity, which might enable detection of the relationship between vlPFC activity during reputation learning and subsequent distrust.
Another major concern regarding the evaluation transfer model is that the activity of evaluation-related limbic structures other than the vlPFC was not detected in this study. In particular, we had expected greater activity of the anterior insula in response to bad reputations considering its possible involvement with anger and disgust (Murphy et al., [@B43]; Sanfey et al., [@B56]; Rilling et al., [@B50]). The role of the amygdala in the evaluation of others' behaviors has also been highlighted (Buckholtz et al., [@B10]; Schiller et al., [@B57]). As mentioned in the "Results" Section, participants might have perceived the no-reputation condition as risky or ambiguous because they were unsure about whether the stimulus individuals in this condition had returned or embezzled the money. The evaluation-related regions listed above are known to be sensitive to risk and ambiguity (Adams et al., [@B2]; Singer et al., [@B58]). In fact, conjunction analysis revealed that these regions were activated not only in the good- and bad-reputation conditions but also in the no-reputation condition. Thus, the subtraction between the good-/bad- and no-reputation conditions could have cancelled out their activity. The use of a more neutral baseline condition in future research may elucidate the roles of those structures in evaluation transfer.
The role of valence in evaluation transfer also remains in need of being carefully examined. Our findings that bad reputations persisted while good ones did not might reflect a human sensitivity to cheaters that has evolved as an adaptation to secure reciprocity in social exchange (Cosmides and Tooby, [@B15]). However, any brain region including the vlPFC did not show greater activation for bad than for good reputations. In addition, although speculative, cheating behaviors would be perceived as not only negative but also uncommon (Mende-Siedlecki et al., [@B40]). It is therefore possible that the bad reputations for having embezzled the money were so unexpected that they captured the attention of participants (Bell and Buchner, [@B7]), facilitating evaluation transfer and the formation of persistent memory. In this study, we controlled for the effect of unexpectedness by means of *post hoc* parametric modulation, and one may doubt the validity of the specific functional assumption in the analysis. We would like to note that the model without the parametric modulation regressors reproduced the vlPFC responses to the faces paired with bad reputations and their relation to distrust, and thus, these main findings should not be affected by the arbitrariness of the presented model. Nevertheless, future research should experimentally manipulate the unexpectedness of good and bad reputations in order to clarify the effect of valence on evaluation transfer. In addition, reputation learning is known to vary across the lifespan (Fett et al., [@B23]), and therefore, our results from a young-adult population might not generalize to other ages. It would be especially important to determine the changes in later life given the possible vulnerability of older adults to cheating (Castle et al., [@B12]).
Finally, the formation of stimulus-value associations has been extensively studied using computational models of reinforcement learning (O'Doherty, [@B44]; Behrens et al., [@B6]). As our experimental tasks did not clearly involve reinforcement (i.e., participants did not receive either reward or punishment), this paper is written within a descriptive framework of evaluative conditioning, the role of which in attitude formation has been highlighted in the social psychology literature (Fazio, [@B21]; Jones et al., [@B32]). Nevertheless, considering that reinforcement values could be defined for any type of stimuli (e.g., pictures; Katahira et al., [@B33]), it would be interesting to apply computational modeling to reputation learning. To achieve this goal, a reputation learning task might need to include fewer stimulus persons and more learning opportunities (e.g., Chang et al., [@B13]) so that reliable estimation is possible.
In conclusion, this study demonstrated that once participants had learned a stimulus face's bad reputation, they came to activate the vlPFC when judging that face, independent of the validity of the bad reputation. In addition, vlPFC activity explained the participants' distrust toward such people. Taken together with the implicated role of the vlPFC in negative evaluation, the findings are interpreted as reflecting evaluation transfer that directly associates the negative evaluation of a bad reputation with a target person. Our results advance a possible neurocognitive explanation as to why bad reputations continue to affect judgments even after they have been shown to be invalid.
Author Contributions {#s5}
====================
AS designed and performed experiments, analyzed data and wrote the article. YI, SK, and MK performed experiments. HO, JK, and HCT gave technical support and conceptual advice. TN performed experiments and gave technical support and conceptual advice. All authors discussed the results and implications and commented on the manuscript at all stages.
Conflict of Interest Statement {#s6}
==============================
The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
This research was supported by the Japan Society for the Promotion of Science Grants-in-Aid for Scientific Research (KAKENHI) awarded to TN (24300186), JK (25285200), and AS (23683019). We thank Professor Hiroshi Yamada at Nihon University, Japan, for allowing us to use the Facial Information Norm Database (Watanabe et al., [@B66]).
^1^This does not mean that the US becomes detached and free from the evaluation. Thus, although the term "transfer" is commonly used, "generalization" or "contagion" of evaluation might be less confusing wording. Jones et al. ([@B31]) have argued that the transfer arises without awareness and so coined the term "implicit misattribution." The current study did not disentangle conscious and unconscious learning, and so we use the more general term "evaluation transfer."
^2^Perfect performance in the baseline session of action inference precluded parametric modulation analysis of the fMRI data described later (see "fMRI Data Analysis" Section).
^3^The remaining 28 participants did not suffer from mental health problems (self-report). Six of them majored in psychology, and may have had some insight into the experimental procedure involving deception. However, exclusion of their data did not change the main results.
^4^The rest was included to minimize fatigue during the long experiment. In order to assist the alignment of head positions before and after the rest, three colored marks were placed on the face of the participant (between the eyes, the left temple, and the middle of the forehead), and the locations of these marks relative to the head coil were kept as consistent as possible.
^5^It would be reasonable to assume that the unexpectedness decreases as face-reputation pairs are repeated. In this study, however, the unexpectedness did not seem to reach zero even in the last (third) run, given that participants' performance on the retrieval session of the action inference task was not perfect (accuracy: *M* ± *SD* = 0.771 ± 0.125). We thus chose the specific functional form described above to model the decrease of unexpectedness (i.e., it was assumed to decrease by 75% from the first to the last run). This assumption may appear arbitrary, but the following results were essentially unchanged even when analyzing the data assuming that the unexpectedness was constant across runs or when removing the parametric modulation regressors. Thus, our findings were not affected by the specific assumption.
^6^Thus, ROI selection and analysis were performed using different datasets of different tasks, mitigating problems caused by non-independent data (Kriegeskorte et al., [@B34]). It is still possible, however, the two datasets have some dependency due to, e.g., common between-participant variance (Poldrack and Mumford, [@B47]).
^7^Session-specific intercepts were assumed for both models because DISTRUST_POST was expectedly smaller in the disregard than retrieval sessions (Suzuki et al., [@B60]).
^8^More specifically, linear mixed model analysis was performed using the following model: vlPFC activity~si~ = *β*~0~ + *β*~1~ × Session~s~ + *u~i~* + *ε~si~*, with *s* denoting the session of action inference (1 = retrieval session, 2 = disregard session), *i* denoting the participant (*i* = 1, ..., 28), Session being a dummy variable (Session~1~ = −1, Session~2~ = 1), *β*~0~ representing the average of the vlPFC activity across sessions, *β*~1~ representing the difference in the vlPFC activity between the two sessions, and *u~i~* and *ε~si~*, respectively, indicating random effects associated with each participant and each observation. The estimates of the fixed effects were *β*~0~ = 0.562 (*p* = 0.006, 95% CI \[0.186, 0.937\]) and *β*~1~ = 0.163 (*p* = 0.338, 95% CI \[−0.170, 0.496\]).
^9^The other regions that were identified by the conjunction analysis were the left superior prefrontal cortex (20 voxels, \[−10, 44, 50\], *Z* = 2.740), the right superior temporal sulcus (54 voxels, \[56, −24, −4\], *Z* = 2.219), and the left medial prefrontal cortex (31 voxels, \[−4, 52, 16\], *Z* = 2.093). In this article, the ROI specification was performed with the data during reputation learning (i.e., Figure [4](#F4){ref-type="fig"}) rather than with this conjunction analysis in order to avoid the criticism of "double dipping" (Kriegeskorte et al., [@B34]; Vul et al., [@B64]) in the following ROI analysis (see also footnote 6).
[^1]: Edited by: Tetsuo Kida, National Institute for Physiological Sciences, Japan
[^2]: Reviewed by: G. Elliott Wimmer, University Medical Center Hamburg-Eppendorf, Germany; Christoph W. Korn, University of Zurich, Switzerland
| {
"pile_set_name": "PubMed Central"
} |
1. Introduction {#sec1}
===============
*Fritillaria cirrhosa*, which belongs to the family Lilliaceae, is primarily distributed in the southwestern China. Bulbs of*Fritillaria cirrhosa* (BFC) have long been used both for food and folk medicine in many Asian countries \[[@B1]\]. The materials from wild collections can be harvested at least for five years \[[@B2]\]. In the past, the original plants of BFC were mainly obtained from their wild species and thus were unable to meet the demands of the industry \[[@B3]\]. In recent years, there is a significant breakthrough in artificial plantation technologies. The bulbs of cultivated*Fritillaria cirrhosa* (BCFC) have become the mainstream of BFC in market, which provide the adequate resources for further exploitation and utilization of BFC.
Studies on phytochemistry of BFC indicated that chemical components of BFC included steroidal alkaloids, saponins, terpenoids, glycosides, and many other compounds \[[@B4]\]. As revealed by modern studies, steroidal alkaloids were the major biological active compositions of BFC \[[@B5]\]. Pharmacological studies found that alkaloids from BFC exhibited remarkable antitussive, expectorant, antiasthmatic properties \[[@B6], [@B7]\], hypotensive effects \[[@B8]\], antibacterial activity, antitumor effects \[[@B9]--[@B11]\], anti-inflammatory effects \[[@B12], [@B13]\], and so on. However, the content of total alkaloids of BFC is quite low and varies wildly, approximately ranging from 0.02% to 0.3%, which is regarded as major quality constraint for its applications. To the best of our knowledge, there are few systematic publication focusing on optimization of the extraction and enrichment for large-scale production of alkaloids from BCFC; thus, there is an urgent need to optimize the extracting and enriching conditions.
Orthogonal array design (OAD) has been widely used to optimize the factors which influenced the extraction yield or extract profiles of bioactive components from natural materials \[[@B14]\]. It has been used to optimize experimental conditions with fewer numbers of experiments \[[@B15], [@B16]\]. In this study, OAD was also used to optimize the extraction conditions.
There are several conventional methods, such as polyamide chromatography, gel chromatography, and silica gel column, available for the enrichment of active constituents \[[@B14]\]. However, these methods have several disadvantages, including long time consuming poisonous residual solvents and low recoveries \[[@B17]\]. Recently, growing attention has been taken to enrich and purify targeted components from crude biological samples using macroporous resins for their convenience, low operating costs, low solvent consumption, high chemically stability, and easy regeneration \[[@B14], [@B17]\]. In this study, the absorption and desorption on macroporous resins were utilized for the enrichment of alkaloids from BCFC.
The present study focused on the optimization of extraction and enrichment of alkaloids from BCFC to develop a cost-effective method for large-scale extraction and enrichment of alkaloids. In extraction process, we tried to optimize four independent variables by using OAD. For enrichment of alkaloids, the optimum macroporous resin was screened and the major process parameters were determined. Meanwhile, the experimental isotherm data were analyzed using pseudo-first- and second-order models, the Langmuir and Freundlich equations. Finally, macroporous resin chromatography was used for the enrichment of alkaloids from BCFC.
2. Materials and Methods {#sec2}
========================
2.1. Plant Materials and Reagents {#sec2.1}
---------------------------------
BCFC were purchased from the dealers who cultivate the*Fritillaria cirrhosa* at Chengdu International Trade City Hehuachi Chinese Medicinal Herbal Market (Chengdu, Sichuan province, China) and identified by Professor Shu Wang (Department of Pharmacognosy, West China College of Pharmacy, Sichuan University, Chengdu, China). The sample (Wang 20120510) has been deposited in the pharmacognosy lab of West China College of Pharmacy, Sichuan University. Anhydrous ethanol, ammonia water, ethylenediamine, and chloroform were analytical grade obtained from Tianjing Kemiou Chemical Reagents Co. (Tianjing, China). Acetonitrile of chromatographic grade was purchased from Sigma Aldrich, Inc. (Steinheim, Germany). Imperialine and peimisine were isolated from BFC in our laboratory \[[@B12], [@B13]\]. The ultrapure water was prepared by a Milli-Q water system (Millipore, USA).
2.2. Selection of Extraction Parameters {#sec2.2}
---------------------------------------
The preprepared powder of BCFC (50 g) was soaked in 70 mL ammonia for different times (0.5 h, 1.0 h, 1.5 h, 2.0 h, 3.0 h, and 4.0 h) and then extracted under reflux with different concentrations of ethanol (50%, 60%, 70%, 80%, 90%, and 100%) with different ratios of liquid/solid (mL : g) (5 : 1, 10 : 1, 15 : 1, 20 : 1, 25 : 1, and 30 : 1) for different given times (60 min, 90 min, 120 min, 150 min, 180 min and 240 min), while the temperature of the water bath was set as 50°C, 60°C, 70°C, 80°C, 90°C, and 95°C, respectively, and kept steady (within ±2.0°C), for different extraction times (once, twice, thrice, and four times) \[[@B18]\].
2.3. Orthogonal Experimental Design of Extraction Procedure {#sec2.3}
-----------------------------------------------------------
Based on single-factor experimental results, four independent parameters (ethanol concentration (v/v, *C*), solid-liquid ratio (mL/g, *R*), extraction time (min, *t*), and temperature (°C, *T*) were confirmed as the major influencing factors; then a L~9~(3^4^) orthogonal experimental design was conducted to evaluate the effects of four independent variables on the extraction efficiency of total alkaloids, which was reflected by extraction yield of total alkaloids of BCFC. The range and levels of individual variables are shown in [Table 1](#tab1){ref-type="table"}. The experiment design is shown in [Table 2](#tab2){ref-type="table"}, along with experimental data. All experiments were run in twice.
2.4. Preparation of Sample Solution and Absorbents {#sec2.4}
--------------------------------------------------
The extract process was done according to optimal extraction conditions obtained in [Section 2.3](#sec2.3){ref-type="sec"}. The extracting solution was filtered, the filtrates were combined and evaporated using a vacuum rotary evaporator at 45°C and dried*in vacuo* for 72 h to yield ethanol extracts. The sample solution was prepared by dissolving the dried ethanol extracts in distilled water. The physical properties of the 16 kinds of adsorbents are summarized in [Table 3](#tab3){ref-type="table"}. For prior use, the adsorbents needed to be soaked for 24 h with 100% ethanol and then washed with 1 mol/L HCl and NaOH solution successively and finally washed with distilled water to remove monomers and porogenic agents trapped inside the pores during the synthesis process \[[@B17]\].
2.5. Adsorption Resins Screening {#sec2.5}
--------------------------------
### 2.5.1. Adsorption Resins Screening {#sec2.5.1}
The optimum resin was screened by static adsorption and desorption tests. The adsorption tests on different resins were performed as follows: preweighed amounts of hydrated resins (equal to 0.5 g dry resin) and 100 mL sample solution (0.3271 mg/mL) were added into 250 mL flasks which were continually shaken for 24 h at 30°C with 120 rpm. The solutions after adsorption were separated from the resins and analyzed by Alpha-1900PC UV-Vis spectrophotometer and Shimadzu-10AT HPLC coupled to Sedex-75 evaporative light-scattering detector (HPLC-ELSD) according to the method described previously \[[@B19], [@B20]\]. The static desorption tests were conducted as follows: after reaching adsorption equilibrium, the resins were washed with 100 mL distilled water and then desorbed with 50 mL 80% ethanol solution. The flasks were continually shaken for 12 h at 30°C with 120 rpm. The adsorption and desorption properties of different resins for alkaloids including adsorption capacity, desorption capacity, and desorption ratio of each resin were quantified with the following equations \[[@B14], [@B21], [@B22]\]: $$\begin{matrix}
{Q_{e} = \left( {C_{0} - C_{e}} \right) \times \frac{V_{i}}{W},} \\
{Q_{d} = \frac{C_{d} \times V_{d}}{W},} \\
{D = \frac{C_{d} \times V_{d}}{\left\lbrack {\left( {C_{0} - C_{e}} \right) \times V_{i}} \right\rbrack} \times 100,} \\
\end{matrix}$$ where *Q* ~*e*~ is the adsorption capacity at adsorption equilibrium (mg/g dry resin); *Q* ~*d*~ is the desorption capacity after desorption equilibrium (mg/g dry resin); *D* is the desorption ratio (%); *C* ~0~, *C* ~*e*~, and *C* ~*d*~ are the initial, absorption equilibrium, and desorption concentrations of alkaloids in the solutions, respectively (mg/mL); *V* ~*i*~ and *V* ~*d*~ are the volume of the initial sample and desorption solution (mL), respectively; and *W* is the dry weight of tested resins (g).
To study the adsorption and desorption properties of the selected resin under different conditions, a series of sample solutions with pH values ranged from 5 to 10 for adsorption were tested, and a series of concentration of ethanol ranged from 10% to 100% for desorption were also investigated. All the solutions after adsorption and desorption at different conditions were analyzed by UV-Vis spectrophotometer \[[@B23]\].
### 2.5.2. Adsorption Kinetics {#sec2.5.2}
Adsorption kinetics tests were conducted by mixing 0.5 g (dry weight) of hydrated H-103 resin with 100 mL of sample solutions (0.3078 mg/mL) in flasks. The concentrations of target compounds in the adsorption solution were monitored at different time intervals until equilibrium. Two most widely used adsorption models, pseudo-first-order and pseudo-second-order model, were used in determining the rate of the adsorption process to investigate the adsorption process of alkaloids of BCFC on H-103 resin \[[@B14], [@B21], [@B24]\].
The pseudo-first-order equation is given as follows: $$\begin{matrix}
{\log\left( {Q_{e} - Q_{t}} \right) = \log\left( Q_{e} \right) - \left( \frac{k_{1} \times t}{2.303} \right),} \\
\end{matrix}$$ where *Q* ~*e*~ and *Q* ~*t*~ are the amounts of analyte (mg/g) adsorbed on the resin at equilibrium and at time *t*, respectively, and *k* ~1~ is the rate constant of pseudo-first-order sorption.
The pseudo-second-order equation is expressed as follows: $$\begin{matrix}
{\frac{t}{Q_{t}} = \frac{1}{k_{2} \times Q_{e}^{2}} + \frac{t}{Q_{e}},} \\
\end{matrix}$$ where *k* ~2~ is the rate constant of pseudo-second-order adsorption; *Q* ~*e*~ and *Q* ~*t*~ are the same as described above.
The linear intraparticle diffusion equation was further fitted using the adsorption kinetic data to determine whether intraparticle diffusion is the rate-limiting step \[[@B24]\]. The intraparticle diffusion equation is given as follows: $$\begin{matrix}
{Q_{t} = k_{i} \times t^{1/2} + C,} \\
\end{matrix}$$ where *k* ~*i*~ and *C* are the intraparticle diffusion rate constants; *Q* ~*t*~ is the same as described above.
### 2.5.3. Adsorption Isotherms {#sec2.5.3}
Afterwards, the tests for equilibrium adsorption isotherms on the selected resin were performed. 100 mL of sample solutions at different concentrations and preweighed amounts of hydrated resins (equal to 0.1 g dry resin) were added into 250 mL flasks and shaken for 24 h at different temperatures (20, 30, and 40°C). The effects of the initial concentration of sample solution and temperature on the adsorption characteristics were studied with the thermodynamic parameters of adsorption determined. Two standard theoretical models, well-known Langmuir model and Freundlich model, were frequently used to describe the adsorption behavior \[[@B21], [@B24]\].
The Langmuir isotherm equation is expressed as follows: $$\begin{matrix}
{\frac{C_{e}}{Q_{e}} = \frac{C_{e}}{Q_{0}} + \frac{1}{K \times Q_{0}},} \\
\end{matrix}$$ where *Q* ~0~ is an empirical constant and *K* is the Langmuir adsorption equilibrium constant. *C* ~*e*~ and *Q* ~*e*~ are the same as described above.
The Freundlich isotherm equation is given as follows: $$\begin{matrix}
{\ln Q_{e} = \ln k_{f} + \frac{1}{n} \times \ln C_{e},} \\
\end{matrix}$$ where 1/*n* is an empirical constant and *k* ~*f*~ is the Freundlich adsorption equilibrium constant \[[@B24]\]. *C* ~*e*~ and *Q* ~*e*~ are the same as described above.
The Gibbs free energy (Δ*G* ^0^) change indicates the degree of the spontaneity of the adsorption process \[[@B24]\]. The equation for Δ*G* ^0^ is defined as follows: $$\begin{matrix}
{\Delta G^{0} = - R \times T \times \ln k,} \\
{\ln k = \frac{- \Delta H^{0}}{R \times T} + \frac{{\Delta S}^{0}}{R},} \\
\end{matrix}$$ where *T* is temperature (*K*), Δ*H* ^0^ is the enthalpy of adsorption (kJ/mol), Δ*S* ^0^ is the entropy of adsorption (J/mol*·*K), and *k* is the equilibrium constant obtained from the best fitted model at 293, 303, or 313 K.
2.6. Dynamic Adsorption and Desorption Tests {#sec2.6}
--------------------------------------------
### 2.6.1. Effect of Concentration of Alkaloids in Sample Solution on Dynamic Adsorption {#sec2.6.1}
Dynamic adsorption tests were first carried out on glass columns (250 mm × 15 mm i.d.) wet-packed H-103 resin with the bed volume (BV) of 8 mL. The sample solution was loaded onto the macroporous resin columns at different concentrations of the loading sample solution (0.10, 0.20, and 0.40 mg/mL). The sample solution flowed through the resin column at a constant flow rate of 1.0 BV/h. The concentration of alkaloids in the aliquots of 1 mL effluents collected at 3.0 BV interval was monitored by UV-Vis spectrophotometer to get the dynamic breakthrough curve and select the most suitable concentration of the loading sample solution.
### 2.6.2. Effect of Diameter to Height Ratio on Dynamic Adsorption {#sec2.6.2}
Besides, the effect of diameter-to-height ratio on breakthrough curve was examined in a similar way. The sample solution flowed through the columns (diameter-to-height ratio 1/3, 1/10, or 1/30) at a constant flow rate of 1.0 BV/h and the concentration of alkaloids in sample solution was approximate 0.20 mg/mL.
### 2.6.3. Effect of Loading Flow Rate on Dynamic Adsorption {#sec2.6.3}
In adsorption process, in addition to concentration of the loading sample solution and diameter-to-height ratio, the effect of loading flow rate on breakthrough curve was also investigated. The sample solution was loaded onto the macroporous resin column (diameter-to-height ratio 1/10) at different flow rates of 2.0, 4.0, and 6.0 BV/h and the concentration of total alkaloids was approximately 0.20 mg/mL.
### 2.6.4. Effect of Ethanol-Water Concentrations on Dynamic Desorption {#sec2.6.4}
After adsorption equilibrium under the optimum condition, the adsorbate-laden column was first washed with 8 BV of distilled water and 4 BV of 10% aqueous ethanol solution at a constant flow rate of 2 BV/h to elute impurities and then eluted with different concentrations of ethanol (70%, 80%, 90%, 95%) at the flow rate of 2 BV/h, respectively.
2.7. Laboratory Preparative-Scale Purification {#sec2.7}
----------------------------------------------
The crude extracts from BFC (400.0 g) were dissolved in water. The sample solution (pH 7.0) was applied to a glass column (120.0 cm × 7.5 cm i.d.) containing 2.0 kg of wet H-103 macroporous resin with a bed volume (BV) of 2.5 L. Initially, distilled water was used to wash the elution solution until almost no color was noted, then 4 BV of 10% ethanol was used to remove the high polar components, and the adsorbent was finally rinsed with 6 BV of 90% ethanol to obtain the alkaloids-rich fraction. The flow rate of each gradient was set at about 2 BV/h, and the elution of 90% ethanol was collected, concentrated, and dried.
2.8. Statistical Analysis {#sec2.8}
-------------------------
The data were presented as means ± standard deviation (S. D.) and evaluated by one-way analysis of variance (ANOVA) using the Statistical Package for Social Sciences (SPSS) computer software program. *P* values \<0.05 were regarded as significant.
3. Results and Discussion {#sec3}
=========================
3.1. Extraction Parameters {#sec3.1}
--------------------------
The extraction rate of constituents from crude resources was affected by many factors, such factors as extraction solvent, extraction time, temperature, and the ratio of liquid-solid, which were usually considered to have significant effect on compounds extraction rate \[[@B18], [@B25]\].
All parameters were tested in a wider range prior to OAD optimization which help narrow down the ranges of the parameters tested. The effect of different parameters (the time of BCFC soaked in ammonia, concentrations of alcohol, extraction time, liquid-solid ratio, temperature, and times of extraction) on the extraction yield of alkaloids was shown in [Figure 1](#fig1){ref-type="fig"}. The results indicated that each selected parameter, except the time of BCFC soaked in ammonia and times of extraction, had a suitable level, at which the total alkaloids extraction rate reached the peak. It can be easily concluded from the corresponding figures that the extraction time 150 min, liquid-solid ratio 10 : 1, and concentrations of alcohol 80% were of suitable levels. So, concentrations of alcohol, liquid-solid ratio, extraction time, and temperature were selected as the four extraction parameters for the following OAD \[[@B18]\].
3.2. Optimization of Extraction Conditions {#sec3.2}
------------------------------------------
To verify whether the effect of individual factors on total alkaloids extraction efficiency is statistically significant, an ANOVA test was used to analyze the experimental data \[[@B25], [@B26]\]. The significance of each factor was evaluated by calculating the *R* value of extreme difference ([Table 2](#tab2){ref-type="table"}) and the *F* value ([Table 4](#tab4){ref-type="table"}). As seen in Tables [2](#tab2){ref-type="table"} and [4](#tab4){ref-type="table"}, we concluded that the order of parameters influencing on extraction rate of total alkaloids was as follows: ethanol concentration \> temperature \> liquid-solid ratio \> extraction time, and the four parameters were all statistically significant. Based on the analysis, the optimum conditions of extraction were therefore determined as follows: ethanol concentration (%, v/v) 90%, temperature 80°C, liquid-solid ratio 15 : 1, and extraction time 120 min. In order to validate the optimum conditions, a verification experiment was carried out. Under these conditions an extraction yield of total alkaloids from BCFC of 97.84% was obtained.
3.3. Screening of Optimum Resin {#sec3.3}
-------------------------------
To obtain the most appropriate resins, absorption and desorption properties of sixteen macroporous resins towards total alkaloids, imperialine, and peimisine were assessed and the results were shown in [Table 5](#tab5){ref-type="table"}. HPD-100, H-103, and HPD-722 resins exhibited notably higher adsorption capacity towards total alkaloids, imperialine, and peimisine than that of other resins. Moreover, the adsorption capacity of alkaloids on H-103 resin was the highest among the three resins. Besides, the desorption capacity and desorption ratio of HPD-722 resin for alkaloids were lower than that of HPD-100 and H-103 and desorption ratio of H-103 resin was close to that of HPD-100. In view of these results, H-103 resin was considered as the optimum one. H-103 resin exhibited better adsorption capacity and desorption ratio not only because of similar polarity with the nonpolar alkaloids from BCFC, but also because of its higher surface area which may correlate with the capability of the resin and the chemical features of the adsorbed substance \[[@B21]\].
3.4. Effects of pH on the Absorption Capacity and Ethanol Concentration on the Desorption Capacity {#sec3.4}
--------------------------------------------------------------------------------------------------
The pH plays an important role in absorption processes because the ionization of solutes can be influenced by pH, thus affecting the absorption affinity between solutes and the adsorbents \[[@B14], [@B27]\]. The effect of pH of initial sample solution on the absorption capacity was shown in [Figure 2(a)](#fig2){ref-type="fig"}. It shows that for total alkaloids the highest adsorption capacity appeared at the pH value of 7.0. The reason for the sharp decrease when the pH value was below 7.0 is that the alkaloids were ionized with decreasing of pH. In addition, increase of pH produces a lot of precipitation, which exerted negative influence on the absorption capacity. Based on these results, pH 7.0 of sample solutions was selected for absorption in the following tests.
Different concentrations of ethanol solutions were used to perform desorption experiments in order to choose proper desorption solution. As shown in [Figure 2(b)](#fig2){ref-type="fig"}, the desorption ratios of alkaloids increased with the increase of ethanol concentration and reached their peak value at the concentration of 80%, then decreased with the increase of ethanol concentration \[[@B23]\].
3.5. Adsorption Kinetics {#sec3.5}
------------------------
Adsorption kinetics of total alkaloids, imperialine, and peimisine on H-103 resin were investigated at 30°C and the adsorption kinetics curves were obtained. As shown in [Figure 3](#fig3){ref-type="fig"}, the adsorption of total alkaloids, imperialine, and peimisine increased with adsorption time before reaching equilibrium. In the first 240 min, the adsorption of total alkaloids rapidly increased and then increased slowly until adsorption equilibrium was reached at 600 min. The adsorption kinetics for imperialine and peimisine showed similar tendency with total alkaloids.
The kinetic parameters and correlation coefficients of the kinetic models were listed in [Table 6](#tab6){ref-type="table"}. As shown in [Table 6](#tab6){ref-type="table"}, the calculated *Q* ~*e*~ of the pseudo-first-order equation values was closer to the experimental ones, and the correlation coefficients were almost higher than 0.95. Therefore, the first-order equation described the experimental data more accurately for total alkaloids, imperialine, and peimisine. On the other hand, the correlation coefficients for the pseudo-second-order kinetic model obtained were quite high, but the calculated *Q* ~*e*~ values deviated significantly from the experimental ones. Thus, the adsorption observed did not fit the pseudo-second-order model.
The intraparticle diffusion model was used to evaluate the possibility of intraparticle diffusion. As shown in [Table 6](#tab6){ref-type="table"}, based on the correlation coefficient of intraparticle diffusion model, *R* ^2^ indicated that pore diffusion was an important rate-limiting step.
3.6. Adsorption Isotherms {#sec3.6}
-------------------------
The adsorption isotherms were shown in [Figure 4](#fig4){ref-type="fig"}. From the adsorption isotherm, the adsorption capacity increased with the initial concentration and reached the saturation plateau when the initial concentrations of total alkaloids, imperialine, and peimisine were 0.204 ([Figure 4(a)](#fig4){ref-type="fig"}), 0.074 ([Figure 4(b)](#fig4){ref-type="fig"}), and 0.024 mg/mL ([Figure 4(c)](#fig4){ref-type="fig"}), respectively. The adsorption isotherms for imperialine and peimisine showed similar tendency with that of total alkaloids.
The Langmuir and Freundlich equations are used to describe how solutes interact with the resins. The Langmuir model describes the monomolecular layer adsorption without mutual interaction between adsorbed molecules, while the Freundlich model is used to describe the adsorption behavior of monomolecular layer as well as that of the multimolecular layer \[[@B28]\]. As shown in [Table 7](#tab7){ref-type="table"}, the correlation coefficients of both Langmuir and Freundlich equations for total alkaloids, imperialine, and peimisine were rather high. The experimental adsorption data were better fitted to Freundlich equation than Langmuir equation. In the Freundlich equation, values 1/*n* \< 1.0 represent favorable adsorption conditions and values 1/*n* \> 1.0 indicates difficulty in adsorption \[[@B24]\]. In [Table 7](#tab7){ref-type="table"}, the 1/*n* values are between 0.1 and 1.0, which indicated that the adsorption of total alkaloids, imperialine, and peimisine on H-103 resin can take place easily. We also could see from [Figure 4](#fig4){ref-type="fig"} that at a constant initial concentration, the adsorption capacities increased with temperature increase from 20 to 40°C, which indicated that the adsorption was an endothermic process.
The thermodynamic parameters of the adsorption process were listed in [Table 8](#tab8){ref-type="table"}. Δ*G* ^0^ values were negative, indicating that the feasibility and spontaneity of the adsorption process. Its absolute values were less than the critical value (20 kJ/mol), showing that the adsorption process was physical \[[@B24], [@B29]\].
3.7. Dynamic Adsorption and Desorption on H-103 Resin {#sec3.7}
-----------------------------------------------------
### 3.7.1. Dynamic Leakage Curves on H-103 Resin {#sec3.7.1}
Adsorption presumably reached equilibrium when the leakage solution concentration was 5% or 10% of the initial concentration, which defined the leakage point \[[@B24], [@B30]\]. The breakthrough point in this study was set as 5% of the initial concentration.
As shown in [Figure 5(a)](#fig5){ref-type="fig"}, the adsorption capacities at the leakage point were 118.24, 108.99, and 92.16 mg/8 mL resin at the concentrations of 0.10, 0.20, and 0.40 mg/mL, respectively. The adsorption capacities at the leakage point increased a little as the concentration decreased from 0.20 to 0.10 mg/mL, but increased significantly as the concentration decreased from 0.40 to 0.20 mg/mL. So the appropriate concentration of total alkaloids during the loading of the sample on the column was set as 0.20 mg/mL.
As shown in [Figure 5(b)](#fig5){ref-type="fig"}, the adsorption capacities at the leakage point were 108.99, 143.76, and 162.95 mg/8 mL resin at the diameter to height ratio of 1/3, 1/10, and 1/30, respectively. The adsorption capacities increased significantly when the diameter-to-height ratio increased from 1/3 to 1/30. The significant increase in adsorption capacities was not observed when the diameter-to-height ratio increased from 1/10 to 1/30. Moreover, excessively more ratio of diameter-to-height was not conducive to large-scale application. So, a moderate diameter to height ratio of 1/10 was used in the following tests.
As shown in [Figure 5(c)](#fig5){ref-type="fig"}, the adsorption capacities at the leakage point were 140.88, 110.81, and 83.42 mg/8 mL resin at the flow rate of 2, 4, and 6 BV/h, respectively. With the increase of the sample flow rate, the adsorption capacities at the leakage point reduced because the interaction time of alkaloids with active sites of the resin surface is decreased with the increase of sample flow rate \[[@B31]\]. The adsorption ratio was highest at a flow rate of 2.0 BV/h. Considering that the lower flow rate gave a long run time and was not conducive to industrial production, the flow rate for loading sample was maintained constantly at 4.0 BV/h. According to the above experiments, the corresponding breakthrough volumes of alkaloids on H-103 resin column of diameter-to-height ratio 1/10 were 88 BV at the concentrations of 0.2 mg/mL, at the flow rate of 4.0 BV/h ([Figure 5(c)](#fig5){ref-type="fig"}).
### 3.7.2. Dynamic Desorption Curves on H-103 Resin {#sec3.7.2}
As shown in [Section 3.4](#sec3.4){ref-type="sec"}, at the 10% ethanol, alkaloids were hardly desorbed. The desorption ratios of total alkaloids reached their peak value at the concentration of 80% in static desorption. Therefore, we used different concentrations of ethanol (70%, 80%, 90%, and 95%) to establish the proper desorption condition on dynamic desorption. The results were shown in [Figure 6](#fig6){ref-type="fig"}. Samples after purification by H-103 resin were combined, concentrated in a rotary evaporator, and dried under vacuum. After dynamic desorption on H-103 resin column chromatography, the contents of total alkaloids in the products which were eluted by 70%, 80%, 90%, and 95% ethanol were 486.14, 550.89, 638.48, and 643.20 mg/g, respectively, which were 17.23-, 19.53-, 22.63-, and 22.80-fold increased. Their recoveries which were eluted by 70%, 80%, 90%, and 95% ethanol were 95.75%, 96.64%, 95.87%, and 95.46%, respectively. Considering efficiency and economy of the process, 90% ethanol was selected as the desorption solution in the dynamic desorption experiment. Thus, the optimal enrichment conditions for alkaloids on H-103 resin were as follows: for adsorption: initial concentrations of total alkaloids in sample solution: 0.20 mg/mL, diameter-to-height ratio of resin column: 1/10, loading flow rate: 2 BV/h, feed volume: 88 BV, and pH 7.0; temperature: 40°C and for desorption: a successive gradient elution: 8 BV distilled water, 4 BV 10% ethanol, 6 BV 90% ethanol, and flow rate: 2 BV/h.
3.8. Laboratory Preparative-Scale Purification {#sec3.8}
----------------------------------------------
All methods used for purifying alkaloids from BCFC were investigated in a small-scale preparation. This was followed by a large-scale preparation of pure alkaloids using the above determined conditions. The elution of 6 BV 90% ethanol gave the alkaloids-rich fraction, in which the contents of total alkaloids, imperialine, and peimisine were 21.40-, 18.31-, and 22.88-fold increased with recovery yields of 94.43%, 90.57%, and 96.16%, respectively. The HPLC chromatograms of the experimental samples before and after purification with H-103 resin were shown in [Figure 7](#fig7){ref-type="fig"}.
4. Conclusions {#sec4}
==============
The present study established an effective extraction and enrichment procedure for alkaloids from BCFC. On the one hand, the extraction of alkaloids from BCFC was investigated with a four-variable, three-level experiment orthogonal experimental design in enhancing the alkaloids extraction yield. On the other hand, this study reports resin adsorption as a means to enrich alkaloids from BCFC extracts was successfully developed. Among 16 tested resins, H-103 resin presented higher adsorption capacity and desorption ratio. The equilibrium experimental data of the adsorption of total alkaloids, imperialine, and peimisine on H-103 resin at different temperatures were well-fitted to the pseudo-first-order kinetics model, Langmuir and Freundlich isotherms models. The processes of dynamic adsorption and desorption were conducted to ensure the optimal purification parameters of the H-103 resin. To the best of our knowledge, this is the first systematic publication focusing on optimization of the extraction and enrichment for large-scale production of total alkaloids from BCFC. In conclusion, the developed methodology can also be referenced for the extraction and purification of other active compounds from herbal materials and large-scale manufacture of alkaloids of BCFC in food and pharmaceutical industry.
The authors are grateful to the Science and Technology Support Program of Sichuan Province (2011SZ0274) from the Department of Science and Technology of Sichuan Province for financial support.
Conflict of Interests
=====================
The authors declare that there is no conflict of interests regarding the publication of this paper.
{#fig1}
{#fig2}
{#fig3}
{#fig4}
{#fig5}
{#fig6}
{#fig7}
######
Factors and levels of orthogonal experimental design.
Levels Ethanol concentration (*C*, v/v%) Solid-liquid ratio (*R*, mL/g) Extraction time (*t*, min) Temperature (*T*, °C)
-------- ----------------------------------- -------------------------------- ---------------------------- -----------------------
1 80 5 : 1 60 70
2 90 10 : 1 90 80
3 100 15 : 1 120 90
######
Orthogonal experimental design for extraction of total alkaloids from BCFC.
Test no. *t* (min) *C* (v/v%) *T* (°C) *R* (mL/g) *Y* (%)
---------- ----------- ------------ ---------- ------------ ---------------
1 1 1 1 1 81.42 ± 2.78
2 1 2 2 2 96.63 ± 0.98
3 1 3 3 3 63.81 ± 4.65
4 2 1 2 3 107.41 ± 3.28
5 2 2 3 1 80.95 ± 6.62
6 2 3 1 2 81.73 ± 3.78
7 3 1 3 2 91.99 ± 4.80
8 3 2 1 3 106.73 ± 3.64
9 3 3 2 1 78.84 ± 2.15
*k* ~1~ 80.62 93.61 89.96 80.41
*k* ~2~ 90.03 94.77 94.29 90.11
*k* ~3~ 92.52 74.80 78.92 92.65
*R* 11.90 19.98 15.38 12.25
*Y*: mean value of extraction yield of total alkaloids; *R* is the difference between the maximum value and the minimum value of *k* ~*i*~ of any columns. The reported values are expressed as mean ± S.D. (*n* = 2).
######
Physical properties of the macroporous resins.
Name Polarity Particle diameter (mm) Surface area (m^2^/g) Average pore diameter (nm)
---------- -------------- ------------------------ ----------------------- ----------------------------
HPD-100 Nonpolar 0.3--1.20 650--700 8.5--9.0
HPD-200A Nonpolar 0.3--1.25 700--750 8.5--9.0
HPD-300 Nonpolar 0.3--1.20 800--870 5.0--5.5
HPD-700 Nonpolar 0.3--1.20 650--700 8.5--9.0
D-101 Nonpolar 0.2--0.60 400--600 10.0--12.0
D-3520 Nonpolar 0.3--1.25 480--520 8.5--9.0
D-4006 Nonpolar 0.3--1.25 400--440 6.5--7.5
H-103 Nonpolar 0.3--1.25 900--1100 8.4--9.4
HPD-450A Middle-polar 0.3--1.25 500--550 9.0--10.0
HPD-750 Middle-polar 0.3--1.20 650--700 8.5--9.0
ADS-17 Middle-polar 0.3--1.25 90--150 20.0--30.0
DM-130 Middle-polar 0.3--1.25 500--550 9.0--10.0
HPD-722 Weak-polar 0.3--1.25 485--530 13.0--14.0
AB-8 Weak-polar 0.3--1.25 480--520 13.0--14.0
HPD-600 Polar 0.3--1.20 550--600 8.0
HPD-826 Polar 0.3--1.25 500--600 9.0--10.0
######
Variance analysis of orthogonal experimental data.
Source of variance Sum of square Degree of freedom Mean square *F* value
----------------------- --------------- ------------------- ------------- --------------
Extraction time 472.780 2 236.390 15.176\*\*
Ethanol concentration 1508.467 2 754.233 48.422\*\*\*
Temperature 754.461 2 377.231 24.218\*\*\*
Solid-liquid ratio 501.241 2 250.620 16.090\*\*
Error 140.187 9 15.576
Total 141896.623 18
*F* critical value (95%) = 4.25; \*\**P* \< 0.01; \*\*\**P* \< 0.001.
######
Adsorption capacity, desorption capacity, and desorption ratio of different resins for total alkaloids, imperialine, and peimisine at 30°C.
Resin Adsorption capacity (mg/g dry resin) Desorption capacity (mg/g dry resin) Desorption ratio (%)
---------- -------------------------------------- -------------------------------------- ---------------------- -------------- -------------- ------------- --------------- --------------- ---------------
HPD-100 57.76 ± 3.01 20.58 ± 3.56 5.71 ± 0.76 53.01 ± 5.23 18.79 ± 2.13 5.09 ± 0.29 91.67 ± 4.28 91.80 ± 5.54 89.46 ± 6.87
HPD-200A 46.45 ± 2.53 16.75 ± 2.33 5.05 ± 0.32 43.94 ± 0.54 15.73 ± 1.97 4.56 ± 0.07 94.71 ± 4.01 94.00 ± 1.28 90.42 ± 4.39
HPD-300 56.09 ± 5.24 19.83 ± 2.46 5.20 ± 0.44 44.92 ± 1.90 16.46 ± 2.86 4.20 ± 0.64 80.27 ± 4.12 82.72 ± 4.18 80.52 ± 5.53
HPD-700 54.85 ± 0.75 19.33 ± 1.05 5.54 ± 0.40 47.77 ± 1.54 16.49 ± 1.71 4.69 ± 0.03 87.10 ± 3.99 85.17 ± 4.23 84.92 ± 5.58
D-101 50.62 ± 1.84 18.01 ± 1.44 4.62 ± 0.48 39.67 ± 2.86 13.96 ± 0.70 3.76 ± 0.17 78.52 ± 8.52 77.59 ± 2.34 81.46 ± 4.76
D-3520 51.16 ± 4.37 18.45 ± 2.00 4.92 ± 0.62 42.96 ± 7.57 14.74 ± 2.53 4.09 ± 0.30 83.66 ± 7.65 79.59 ± 5.08 83.51 ± 4.40
D-4006 54.55 ± 1.94 19.37 ± 2.54 5.60 ± 0.59 48.31 ± 3.15 16.95 ± 0.85 4.94 ± 0.99 88.72 ± 8.93 87.98 ± 7.11 87.69 ± 8.42
H-103 59.40 ± 5.17 21.07 ± 2.78 5.96 ± 0.35 53.54 ± 2.03 19.06 ± 1.71 5.30 ± 0.49 90.33 ± 4.45 90.71 ± 3.86 88.89 ± 3.10
HPD-450A 46.63 ± 1.67 16.82 ± 0.81 4.62 ± 0.37 39.64 ± 1.92 14.16 ± 0.81 3.81 ± 0.64 84.99 ± 1.06 84.41 ± 8.86 82.18 ± 7.32
HPD-750 52.23 ± 2.92 18.74 ± 3.17 5.31 ± 0.88 45.43 ± 6.22 15.95 ± 1.73 4.43 ± 0.43 86.78 ± 7.06 85.55 ± 5.25 83.89 ± 5.77
ADS-17 49.38 ± 0.71 17.06 ± 1.32 4.94 ± 0.37 42.29 ± 1.95 14.22 ± 0.18 4.08 ± 0.02 85.63 ± 2.72 83.64 ± 7.51 82.92 ± 6.70
DM-130 11.05 ± 1.10 3.99 ± 1.39 1.56 ± 0.24 13.13 ± 2.04 4.79 ± 1.93 1.83 ± 0.41 118.43 ± 6.63 116.01 ± 7.13 109.28 ± 8.43
HPD-722 57.64 ± 2.61 20.39 ± 1.51 5.74 ± 0.23 48.65 ± 1.36 15.95 ± 2.56 4.44 ± 0.16 84.54 ± 6.18 77.97 ± 6.78 77.34 ± 5.97
AB-8 56.33 ± 1.25 19.51 ± 1.58 6.16 ± 0.57 52.10 ± 5.46 17.44 ± 3.06 5.48 ± 0.94 92.41 ± 7.64 89.01 ± 8.47 88.71 ± 6.93
HPD-600 43.26 ± 5.71 15.30 ± 2.48 4.46 ± 0.51 37.90 ± 3.71 13.13 ± 2.82 3.79 ± 0.67 87.82 ± 3.02 85.42 ± 4.58 84.68 ± 5.22
HPD-826 41.83 ± 5.53 14.59 ± 2.70 4.87 ± 0.81 42.12 ± 8.56 13.89 ± 2.91 4.51 ± 0.95 100.20 ± 7.22 95.04 ± 2.32 92.24 ± 4.29
######
Kinetics parameters and correlation coefficients for the adsorption of total alkaloids, imperialine, and peimisine on H-103 resin.
Compound *Q* ~*e*~, exp^a^ (mg/g) Pseudo-first-order Pseudo-second-order Intraparticle diffusion
----------------- -------------------------- -------------------- --------------------- ------------------------- ------- ---------------- ------- -------- ------- -------
Total alkaloids 55.745 49.627 3.661 × 10^−3^ 0.971 1.057 2.085 × 10^−3^ 0.997 10.568 1.492 0.846
Imperialine 18.106 16.482 3.651 × 10^−3^ 0.990 1.178 0.287 × 10^−3^ 0.958 1.271 0.540 0.878
Peimisine 5.656 4.608 2.730 × 10^−3^ 0.967 1.092 1.172 × 10^−3^ 0.990 0.457 0.161 0.904
^a^Exp is expressed as experimental value; ^b^Cal is expressed as calculated value.
######
Langmuir and Freundlich adsorption isotherm parameters of total alkaloids, imperialine and peimisine on H-103 resin at different temperatures.
Compound Temperature (°C) Langmuir equation *Q* ~0~ (mg/g) *K* *R* ^2^ Freundlich equation *k* ~*f*~ 1/*n* *R* ^2^
----------------- -------------------------------------------------- -------------------------------------------------- ---------------- --------- --------------------------------------------- --------------------------------------------- ----------- ------- ---------
Total alkaloids 20 *C* ~*e*~/*Q* ~*e*~ = 0.00841*C* ~*e*~ + 0.00065 114.340 13.923 0.952 ln*Q* ~*e*~ = 6.062 + 0.764 × ln*C* ~*e*~ 429.196 0.764 0.944
30 *C* ~*e*~/*Q* ~*e*~ = 0.00823*C* ~*e*~ + 0.00047 121.557 17.385 0.954 ln*Q* ~*e*~ = 6.010 + 0.670 × ln*C* ~*e*~ 407.666 0.670 0.993
40 *C* ~*e*~/*Q* ~*e*~ = 0.01246*C* ~*e*~ + 0.00011 80.267 111.555 0.990 ln*Q* ~*e*~ = 5.324 + 0.378 × ln*C* ~*e*~ 205.232 0.378 0.953
Imperialine 20 *C* ~*e*~/*Q* ~*e*~ = 0.02505*C* ~*e*~ + 0.00063 39.912 39.317 0.939 ln*Q* ~*e*~ = 5.756 + 0.742 × ln*C* ~*e*~ 315.965 0.742 0.978
30 *C* ~*e*~/*Q* ~*e*~ = 0.02178*C* ~*e*~ + 0.00051 45.901 38.903 0.976 ln*Q* ~*e*~ = 6.048 + 0.762 × ln*C* ~*e*~ 423.471 0.762 0.989
40 *C* ~*e*~/*Q* ~*e*~ = 0.03637*C* ~*e*~ + 0.00011 27.493 325.897 0.992 ln*Q* ~*e*~ = 4.710 + 0.388 × ln*C* ~*e*~ 111.089 0.388 0.946
Peimisine 20 *C* ~*e*~/*Q* ~*e*~ = 0.07734*C* ~*e*~ + 0.00014 12.930 121.359 0.976 ln*Q* ~*e*~ = 5.537 + 0.756 × ln*C* ~*e*~ 253.900 0.756 0.984
30 *C* ~*e*~/*Q* ~*e*~ = 0.07734*C* ~*e*~ + 0.00014 16.454 108.524 0.926 ln*Q* ~*e*~ = 5.968 + 0.799 × ln*C* ~*e*~ 390.781 0.799 0.986
40 *C* ~*e*~/*Q* ~*e*~ = 0.11229*C* ~*e*~ + 0.00011 8.905 1006.130 0.992 ln*Q* ~*e*~ = 4.080 + 0.399 × ln*C* ~*e*~ 59.119 0.399 0.955
######
The thermodynamic parameters of adsorption of total alkaloids, imperialine, and peimisine on H-103 resin at different temperatures.
Compound *k* (L/mg) Δ*G* ^0^ (kJ/mol)
----------------- ------------ ------------------- ---------- --------- --------- ---------
Total alkaloids 13.923 39.317 121.359 −6.415 −7.194 −12.268
Imperialine 17.385 38.903 108.524 −8.944 −9.223 −15.058
Peimisine 111.555 325.897 1006.130 −11.690 −11.807 −17.992
[^1]: Academic Editor: John J. Gildea
| {
"pile_set_name": "PubMed Central"
} |
Introduction {#s1}
============
Inbreeding, defined as the mating between two related individuals, increases the occurrence of homozygous deleterious alleles. The loss of heterozygosity leads to a decrease in the fitness of offspring, known as inbreeding depression [@pone.0095963-Charlesworth1] although inbreeding may also be associated with benefits [@pone.0095963-Szulkin1]. Inbreeding depression has been reported for most taxa and has led to a variety of inbreeding-avoidance mechanisms [@pone.0095963-Pusey1] e.g. by sex differences in dispersal [@pone.0095963-Pusey2], [@pone.0095963-Perrin1], [@pone.0095963-Bilde1], in life-history [@pone.0095963-Bukowski1], [@pone.0095963-Clarke1], or by mate choice [@pone.0095963-Tregenza1]. The latter requires a kin-recognition mechanism (but see [@pone.0095963-Yasui1]) and can occur before but also after mating [@pone.0095963-Lihoreau1], [@pone.0095963-Penn1]. Post-mating sexual selection requires multiple mating by females which increases copulation costs that should be offset, at least in part, by benefits [@pone.0095963-Zeh1]. Such benefits are particularly enigmatic if they are only of an indirect nature [@pone.0095963-Jennions1]. Indeed, avoidance or reduction of inbreeding costs through post-copulatory mate choice have been identified as a major benefit of female multiple mating in several taxa, such as house mice [@pone.0095963-Firman1], birds [@pone.0095963-Birkhead1], field crickets [@pone.0095963-Tregenza2], [@pone.0095963-Bretman1], [@pone.0095963-Bretman2], and spiders [@pone.0095963-Welke1].
In mating systems with classical sex roles (unselective males maximise fitness by increasing mating rates while reproductive success of females does not increase linearly with the number of mates [@pone.0095963-Bateman1]), females show a larger investment per offspring [@pone.0095963-Smith1], [@pone.0095963-Parker1] and suffer more from inbreeding through the loss of their individual fitness than males that only invested some sperm. Thus, selection to avoid inbreeding in the context of individual fitness costs should act particularly strong on females. However, in mating systems characterised by low male mating rates, males suffer similar fitness costs from inbreeding as females and avoidance of inbreeding should also be favoured in males. Indeed, when male mating rates are lower than female mating rates, selection should act more strongly on males than on females, particularly when polyandrous females possess means of cryptic female choice. These conditions are met in monogynous or bigynous mating systems, which are especially common in spiders [@pone.0095963-Welke2], [@pone.0095963-Fromhage1], [@pone.0095963-Schneider1], [@pone.0095963-Herberstein1]. Males in such mating systems restrict themselves to mating with a single or maximally two females while females appear to favour multiple mating [@pone.0095963-Schneider1], [@pone.0095963-Miller1]. It has been suggested that females oppose monopolisation by a single male through post-copulatory discrimination against less compatible males and there is some evidence that females cryptically discriminate against the sperm of related males [@pone.0095963-Welke1]. However, to date no study has directly measured natural risks and costs of inbreeding for an individual in such mating systems.
Inbreeding is particularly likely if a small number of individuals split off from the original population and establish a new population representing only a fraction of the gene pool of the source population [@pone.0095963-Mayr1]. Furthermore, the co-settlement of siblings may promote the risk of inbreeding in the newly founded population. Analogous to the classical scenarios of founding populations and bottlenecks, although only short-term, species that actively expand their range will likely experience a decrease in genetic diversity at the forefront of range expansion in comparison to populations in the centre of a species\' range [@pone.0095963-Eckert1]. This may result in an increased risk of inbreeding at least in the short term. Individuals that reproduce in a new patch may be faced with a reduced choice of mating partners that are perhaps even siblings. The lack of compatible mating partners can entail fitness costs as even one generation of inbreeding can lead to drastic fitness losses of the offspring, e.g. in terms of reduced competitive fertilisation success as reported for male *Telegryllus oceanicus* [@pone.0095963-Simmons1] or reduced adult lifespan in the spider *Argiope australis* (Welke & Schneider unpublished data). However, some species are tolerant to short-term inbreeding as for example *Stegodyphus lineatus* [@pone.0095963-Bilde1], *Oedothorax apicatus* [@pone.0095963-Bilde2] and *Anelosimus* cf. *jucundus* [@pone.0095963-Aviles1]. The degree of inbreeding depression can vary depending on the size and age of the mating population [@pone.0095963-Jamieson1], as well as the potentially involved purging of deleterious recessive alleles [@pone.0095963-Barrett1], [@pone.0095963-Crnokrak1]. Generally, species that are successful colonisers are expected to show some tolerance towards the negative effects of inbreeding [@pone.0095963-Purcell1] or a dispersal mode that does ensure genetic diversity even in newly founded sites.
Here, we use the spider *Argiope bruennichi* (Araneae) that unites a mono- and bigynous mating system and has recently extended its range from southern Europe and Asia to Northern Europe [@pone.0095963-Kumschick1]. The rapid colonisation implies that *A. bruennichi* can be considered a successful disperser. In combination with the observation that the species has started its range expansion from a large source population, it is likely that newly established populations even by small numbers of individuals encompass some genetic variation. *A. bruennichi* disperses aerially by ballooning and bridging to move within habitats. This passive mode of dispersal, particularly ballooning, entails a large component of chance as individuals can only influence direction by selecting certain wind conditions to fly [@pone.0095963-Suter1], [@pone.0095963-Bonte1]. The expansion would likely occur through small numbers of individuals establishing new populations and as new meadows are colonised, individual females can expect high reproductive success. *A. bruennichi* spiderlings hatch simultaneously from large clutches after winter and likely disperse in groups from the same brood when conditions are favourable. This will lead to a situation in which many siblings from a single female are likely present in a meadow that also contains other families. Spiderlings may disperse short or long distances. This scenario creates both, inbreeding risk as siblings encounter one another and costs of inbreeding (note that costs of inbreeding require a population that is not inbred). Such a scenario match data derived from mating experiments and field observations that demonstrated selection to avoid the costs of inbreeding [@pone.0095963-Welke1]. Hence, we predict that genetic diversity is present in small recently colonised meadows but that sibling matings will occur. As a consequence, we predict the presence of inbreeding depression from within-population matings, which should be absent in among-population matings. Hence, we expect a larger variation in hatching success resulting from the former matches in comparison from the latter ones and we expect this to be matched by the occurrence of sibling matches within populations.
We collected *A. bruennichi* egg-sacs and juveniles from four similar sized populations located near the northern edges of the species range. We assessed genetic diversity by analysing 16 microsatellite loci and a part of the mitochondrial COI gene. Furthermore, we assembled mating pairs that stemmed from the same egg-sac, from different egg-sacs of the same population or from two different populations and correlated the genetic distance of the mating partners with mating behaviour and hatching success. We predicted differences in mating behaviour with increasing relatedness of the mating partners and expected genetic distance to be positively correlated with hatching success.
While the sampled populations are all located within the recently colonised range of the species, they likely differ in their short-term settlement history in that they may have been populated early in the invasion process or in recent years.
Material and Methods {#s2}
====================
Study species {#s2a}
-------------
*Argiope bruennichi* [@pone.0095963-Scopoli1] did not occur in Northern Europe until the beginning of the 20^th^ century with the exception of an isolated group around Berlin [@pone.0095963-Krehenwinkel1]. It expanded its range since around 1930 [@pone.0095963-Kumschick1], [@pone.0095963-Guttmann1] and colonised Northern Germany including the region around Hamburg since 1975 [@pone.0095963-Guttmann1]. Today, these spiders are very common on meadows all over Northern Europe and can occur in densities of about 3 webs/m^2^ (Zimmer SM, personal observation).
As typical of entelegyne spiders, *A. bruennichi* possess paired mating organs. Females have two copulatory openings that are connected by two ducts to independent sperm storage organs (spermathecae) [@pone.0095963-Foelix1]. The two spermathecae can be filled separately by the same or two and rarely three males [@pone.0095963-Uhl1], [@pone.0095963-Nessler1]. Males have two secondary copulatory organs, the pedipalps, which they use to transfer their sperm. Because males damage their pedipalps during copulation, they can use both of them only once. The damaged genital part acts as a plug in the female\'s genital opening and is very effective in preventing rivals to mate into the same opening. This mechanism limits a female\'s mating rate [@pone.0095963-Nessler1].
Females show a highly aggressive mating behaviour. All females attack males during copulation and 80% of males are cannibalised by the female after mating [@pone.0095963-Schneider2]. These males have used only one of their paired pedipalps. Males that survive their first copulation may return and inseminate the second spermatheca of the same female or they may leave and search for a second mating partner [@pone.0095963-Welke2]. All males inevitably die during their second copulation which can be found in other *Argiope* species as well [@pone.0095963-Foellmer1], [@pone.0095963-Sasaki1].
Study Animals {#s2b}
-------------
We collected egg-sacs and juveniles from four geographically separated populations in the northern part of Germany (distance between population locations range between 42 and 148 km; Pevestorf (53°03′40.69″ N, 11°28′24.59″ E), Quarrendorf (53°15′51.81″ N, 10°01′30.74″ E), Buxtehude (53°27′10.37″ N, 9°40′23.67″ E), and Hamburg-Moorfleet (HH-Moorfleet; 53°30′37.30″ N, 10°6′1.60″ E)) between the end of April and the beginning of June 2010. There were no specific permissions required for these locations and the sampling did not involve endangered or protected species.
The collected egg-sacs were produced in 2009 and had overwintered. Several hundred spiderlings hatch out of the same egg-sac [@pone.0095963-Crome1] and can hence be unambiguously labelled as siblings, although females may mate with two different males that share paternity so that spiderlings from the same egg-sac could be full or half-siblings [@pone.0095963-Zimmer1]. The relatedness of juveniles could not be determined, so that these animals could not be used for sibling matings (see below).
387 individuals were raised from eggs in the laboratory until they reached adulthood. Each spider was individually labelled so that it was known from which population and from which egg-sac it derived. Males were kept in individual 250 ml plastic cups, whereas subadult females were housed in 330 ml plastic cups and were transferred in individual Perspex frames (36 ~\*~ 36 ~\*~ 6 cm) after they moulted to maturity. Mating trials were conducted in the frames, where females built their typical orb-webs. All spiders were sprayed with water five days a week. Males were fed with approx. 15 *Drosophila* spec., subadult and adult females with three *Calliphora* spec. on two days a week. After individuals\' final moult, both females and males were weighed on an electronic balance (Mettler Toledo AB54-S) to the nearest 0.001 mg. All males and females used in the mating experiments were frozen at −80°C and preserved for genetic analyses (see below). Males were preserved after a single copulation and females were kept in the laboratory to produce egg-sacs until they died a natural death.
Mating experiments {#s2c}
------------------
We experimentally staged and closely observed matings between siblings (N = 32), between non-siblings from the same population (N = 45) and between non-siblings from different populations (N = 62). Egg-sacs were collected from two populations (Buxtehude and HH-Moorfleet) so that we derived 15 maternal lines (8 from the population Buxtehude, 7 from the population HH-Moorfleet). Females and males from these matrilines were randomly assigned to one of the three mating trials. Spiders that were collected as juveniles were only used in the treatment where we arranged matings between different populations. 32 females from population HH-Moorfleet and Buxtehude were paired with males from the same family (sibling pairs; hatched from the same egg-sac); 45 females from population HH-Moorfleet and Buxtehude were paired with males from the same population that hatched out of a different egg-sac and 62 were mated to males that originated from different populations (HH-Moorfleet, Buxtehude, Quarrendorf, Pevestorf). Each mating pair was allowed a single copulation. Mating trials began by introducing the adult male into the frame threads of the female\'s web. Trials were terminated after the first copulation. If no mating occurred until one hour had passed, a new male was introduced to the female. A female was presented with a maximum of three males. It never happened that a female was not mated after introducing the third male. During every mating trial, we noted the times of male\'s first contact with the web and the female, the beginning and duration of courtship and copulation, the insemination duct the male copulated into and the occurrence of sexual cannibalism or male escape from a female attack.
Hatching success {#s2d}
----------------
Mated females were transferred from the frame into 500 ml plastic cups where they built their egg-sacs. We obtained egg-sacs from 95 females, each of which produced 3.37±0.18 egg-sacs on average. All egg-sacs were weighed on the day of their construction and were visually inspected. Some egg-sacs were damaged or not completed. We selected all intact egg-sacs and left them to hatch. After the young had hatched from the eggs, egg-sacs were preserved and all eggs and spiderlings were counted under the microscope. Hatching success of all intact egg-sacs was determined by the following calculation: total number of spiderlings/((total number of eggs + spiderlings)/100).
Microsatellite analysis and mitochondrial sequencing {#s2e}
----------------------------------------------------
We used microsatellite typing to estimate genetic distance (measured as the individual proportion of shared alleles; POSA) between individuals within and among the four populations of *A. bruennichi*. We were able to determine genetic distances in seven sibling pairs, 11 within-population pairs and in 31 among-population pairs.
For this, we extracted DNA with the 5 PRIME ArchivePure DNA Kit according to the manufacturer\'s protocol (5 PRIME, Hamburg, Germany).
We genotyped our specimens for a set of 16 previously developed microsatellite loci for *A. bruennichi* [@pone.0095963-Krehenwinkel1]. PCR amplification was performed according to the Qiagen Multiplex PCR Kit Protocol (see Qiagen, Hilden, Germany). We used ABI ROX size standard as size standard. Genotyping was performed on an Applied Biosystems 3730 DNA Analyzer. Microsatellite alleles were then called using GeneMapper 4.0 (Applied Biosystems). Genetic distances (POSA) within and among populations, as well as the overall F~ST~ value and pair-wise F~ST~ values among populations were calculated using Microsatellite Analyser (MSA) 4.05 [@pone.0095963-Dieringer1]. Furthermore, we calculated heterozygosity of each individual and the allelic richness per population across the 16 microsatellite loci using MSA 4.05.
Due to the presence of null alleles (one or more alleles fail to amplify during PCR) for the microsatellite screened, detected with the software Microchecker 2.2.3 [@pone.0095963-VanOosterhout1], we sequenced also a 1200 bp fragment of the mitochondrial COI gene as an additional marker. PCR and sequencing conditions are described in [@pone.0095963-Krehenwinkel1]. Sequences were edited using CodonCode Aligner (CodonCode Corperation, Centerville, USA) and aligned using ClustalW with default settings implemented in MEGA 4.0 [@pone.0095963-Tamura1]. Genetic Diversity (nucleotide and haplotype diversity) of the four populations was then calculated using DnaSP 5.10.1 [@pone.0095963-Librado1].
Statistics {#s2f}
----------
Most data were analysed with the statistical program JMP 7.0.2. Non-normally distributed data (and residuals) were analysed with the non-parametric Kruskal-Wallis test. Significant differences between groups were specified with the Dunn test [@pone.0095963-Zar1]. Tests of equal variances were performed with the Brown-Forsythe test. Linear or logistic regressions were used to test the influence of genetic distances on mating behaviour and hatching success. A multiple regression was used to test the influence of female\'s and male\'s heterozygosity on the hatching success of their offspring. All tests are indicated with the results. Descriptive statistics are given as mean ± standard error (SE). Sample sizes may differ between analyses due to missing data. Data are archived in Dryad: doi:10.5061/dryad.1np06.
Results {#s3}
=======
Hatching success {#s3a}
----------------
After a period of incubation, egg-sacs were opened and unhatched eggs and spiderlings were counted to determine hatching success. Hatching success was highly variable in all three treatments. As expected, the average hatching rate was lowest for sibling matings (28.18%±6.9; median = 3.95, N = 21), followed by within-population matings (40.63%±5.27; median = 46.84, N = 31) and was highest when the pair originated from different populations (57.0%±4.46; median = 67.15, N = 43; Kruskal-Wallis test: χ^2^ = 13.12, p = 0.0014; [Figure 1](#pone-0095963-g001){ref-type="fig"} and [Table 1](#pone-0095963-t001){ref-type="table"}). Multiple comparisons showed a significant difference of among-population and sibling groups (Dunn test; p = \<0.01) as well as the among-population and within-population groups (Dunn test; p = \<0.05); but comparisons between sibling groups and within-population groups were not statistically significant (Dunn test; p = \>0.5; [Figure 1](#pone-0095963-g001){ref-type="fig"}). Variances in hatching success did not differ significantly between within-population and among-population matings (Brown-Forsythe test: F = 0.16, p = 0.69).
{#pone-0095963-g001}
10.1371/journal.pone.0095963.t001
###### Summarised results of the three mating treatments in *A. bruennichi* including averaged observed Heterozygosity (Ho) per female and male, averaged proportion of shared alleles (POSA) and averaged hatching success per mating pairs.
{#pone-0095963-t001-1}
Treatment Female H~o~ Male H~o~ POSA Hatching success
------------------------- ----------------------------------------------------- ----------------------------------------------------- ----------------------------------------------------- ------------------------------------------------------
sibling pairs 0.35±0.04 [\*](#nt101){ref-type="table-fn"}(N = 8) 0.34±0.04 [\*](#nt101){ref-type="table-fn"}(N = 9) 0.37±0.05 [\*](#nt101){ref-type="table-fn"}(N = 7) 28.18±6.9 [\*](#nt101){ref-type="table-fn"}(N = 21)
within-population pairs 0.36±0.05 [\*](#nt101){ref-type="table-fn"}(N = 11) 0.33±0.03 [\*](#nt101){ref-type="table-fn"}(N = 19) 0.61±0.03 [\*](#nt101){ref-type="table-fn"}(N = 11) 40.63±5.27 [\*](#nt101){ref-type="table-fn"}(N = 31)
among-population pairs 0.36±0.02 [\*](#nt101){ref-type="table-fn"}(N = 33) 0.34±0.02 [\*](#nt101){ref-type="table-fn"}(N = 36) 0.62±0.02 [\*](#nt101){ref-type="table-fn"}(N = 31) 56.99±4.46 [\*](#nt101){ref-type="table-fn"}(N = 43)
\* Sample sizes may differ between results due to missing data.
Genetic differences within and between source populations {#s3b}
---------------------------------------------------------
The variability of all 16 microsatellite loci was high in all four source populations with a range from 5.6 to 6.4 numbers of alleles per locus referred to as allelic richness ([Table 2](#pone-0095963-t002){ref-type="table"}).
10.1371/journal.pone.0095963.t002
###### Summarised results for 16 microsatellite loci of four geographic different *A. bruennichi* populations including number of individuals per population (N), observed (Ho) and expected (He) heterozygosity per population, averaged proportion of shared alleles (POSA) per population as well as allelic richness (number of alleles per locus).
{#pone-0095963-t002-2}
Population N Ho/He POSA Allelic richness COI Nucleotide diversity[\*](#nt102){ref-type="table-fn"} COI Haplotype diversity[\*](#nt102){ref-type="table-fn"} No. of Haplotypes[\*](#nt102){ref-type="table-fn"}
------------------- ---- ----------- ------------ ------------------ ----------------------------------------------------------- ---------------------------------------------------------- ----------------------------------------------------
Hamburg-Moorfleet 65 0.38/0.62 0.58±0.003 6.1 0.00082 0.61299 5
Buxtehude 67 0.31/0.61 0.57±0.003 5.9 0.0015 0.70227 6
Quarrendorf 29 0.33/0.6 0.59±0.005 6.4 0.00113 0.68923 4
Pevestorf 31 0.36/0.56 0.53±0.005 5.6 0.00099 0.71077 5
\* Nucleotide diversity and Haplotype diversity, as well as the number of Haplotypes of the four populations were calculated by the mitochondrial COI gene using DnaSP 5.10.1.
On average, all populations had a genetic distance between 0.5 and 0.6 (see [Table 2](#pone-0095963-t002){ref-type="table"}). Comparison of the allelic richness among the four different populations across the 16 microsatellite loci did not reveal significant differences either (Kruskal-Wallis test: χ^2^ = 0.65, p = 0.89, N = 64). The observed heterozygosity of the four populations ranged from 0.31 to 0.38 and was much lower than the expected heterozygosity (range from 0.59 to 0.62; [Table 2](#pone-0095963-t002){ref-type="table"}). Furthermore, the nucleotide diversity (range from 0.0008 to 0.0015) as well as the haplotype diversity (range from 0.61 to 0.71) of the four populations calculated by the mitochondrial COI data set showed similar genetic diversities within the four populations ([Table 2](#pone-0095963-t002){ref-type="table"}). The number of haplotypes of the four populations ranged from 4 to 6 ([Table 2](#pone-0095963-t002){ref-type="table"}). The overall F~ST~ value showed a moderate, but significant differentiation (0.052; p = 0.0001) and differentiation between all population pairs were significant ([Table 3](#pone-0095963-t003){ref-type="table"}).
10.1371/journal.pone.0095963.t003
###### Pair-wise F~ST~ -values (below diagonal) and the p-values (determined by permutation; above diagonal) for the four *A. bruennichi* populations based on 16 microsatellite loci.
{#pone-0095963-t003-3}
HH-Moorfleet Buxtehude Quarrendorf Pevestorf
------------------ -------------- ----------- ------------- -----------
**HH-Moorfleet** 0.0001 0.0028 0.0001
**Buxtehude** 0.060975 0.0001 0.0001
**Quarrendorf** 0.025936 0.0401 0.0017
**Pevestorf** 0.056521 0.069954 0.021097
Genetic composition of pairs and mating behaviour {#s3c}
-------------------------------------------------
We pooled all mating pairs regardless of their origin and tested whether the number of shared alleles between female and male of a mating pair correlated with components of their mating behaviour. The duration of copulation (linear regression: F~1,47~ = 0.02, r^2^ = 0.0004, p = 0.9) and the frequency of cannibalism (logistic regression: χ^2^ = 1.85, p = 0.17, N = 49) were independent of the genetic distance between the mating partners. Furthermore, the genetic distance between a male and a female did not affect the time required until copulation occurred (linear regression: F~1,47~ = 0.02, r^2^ = 0.0004, p = 0.89).
Genetic composition of pairs and hatching success {#s3d}
-------------------------------------------------
Comparing genetic distances among the three mating treatments, we expected to find the lowest genetic distance in sibling pairs, closely followed by a part of the within-population pairs while we expected the largest genetic distance in among-population pairs. A Kruskal-Wallis test showed significant differences of genetic distances between the three treatments (χ^2^ = 15.0, p = 0.0006, N = 49). However, we did not detect significant differences between unrelated pairs derived from the same (0.61±0.03) or from different populations (0.62±0.02; Dunn test; p = \>0.2; [Figure 2](#pone-0095963-g002){ref-type="fig"}) while as expected, siblings had the lowest genetic distance (0.37±0.05; see [Table 1](#pone-0095963-t001){ref-type="table"}) and significantly differed from the other two groups (Dunn test; p = \<0.001). Variances in genetic distance did not differ significantly between the within-population and among-population groups (Brown-Forsythe test: F = 0.02, p = 0.88).
{#pone-0095963-g002}
Using all mating pairs, the genetic distance between female and male of each pair did not significantly predict hatching success of their egg-sacs (linear regression: F~1,32~ = 1.28, r^2^ = 0.04, p = 0.27; [Table 1](#pone-0095963-t001){ref-type="table"}). A multiple regression with the observed heterozygosity of female (F~1,33~ = −0.55, p = 0.58) and male of each pair (F~1,33~ = 0.79, p = 0.44) revealed no significant association with hatching success either (multiple regression: F~2,31~ = 0.37, r^2^ = 0.02, p = 0.69).
Discussion {#s4}
==========
In experimental mating trials, we found the lowest hatching success in egg-sacs from pairs derived from the same brood and the highest hatching success when members of a pair came from different populations, while hatching success was intermediate for pairs of the same population. We genotyped each individual using 16 polymorphic microsatellite markers and expected that the presumed increase of genetic distance between the above groups of mating pairs would be mirrored in estimated proportion of shared alleles. However, while our measures of genetic distance provided expected estimates for siblings, we neither detected differences in genetic diversity between our study populations nor could we relate reduced hatching success in clutches derived from within-population matings to genetic distance between mating pairs. There are two possible explanations for the inconclusive mismatch between genetic and reproductive data. Either reduced hatching success in within-population matings was not caused by inbreeding depression or our genetic markers alone were not appropriate to detect relatedness between pairs.
The low hatching success of egg-sacs from sibling pairs strongly suggests that the species would suffer from inbreeding depression if sibling matings did occur. Studies of other species found much lower costs of sibling matings in terms of fitness traits such as hatching success, fecundity and survival [@pone.0095963-Bilde1], [@pone.0095963-Bilde2] and significant negative effects were apparent after only three generations of inbreeding. It was suggested that spiders might show a high tolerance towards inbreeding, perhaps as an adaptation to cope with a relatively high incidence of sibling matings [@pone.0095963-Bilde1]. Our data imply a comparatively low tolerance to inbreeding in *A. bruennichi*, but also a low risk of inbreeding even in small, recently founded populations. Genetic diversity was high in all populations and was probably even underestimated as the sampling mostly occurred before a possible ballooning event.
Within-population matings resulted in an intermediate hatching success with a very high variation and the variation in genetic distance was also highest in this group. This corroborates our predicted scenario and may suggest that some pairings were distinctly less profitable than others while the majority matched well. By coincidence, the majority of pairings in this treatment may have used offspring from unrelated females. The high variance may tentatively suggest that there is a possibility of less compatible matings if spiders stayed close to their birth site. The design of this study may have not been sufficient to detect the actual probability of sibling matings. It is possible that such matings can only be estimated by investigating small-scale spatial patterning of individuals as it has been measured in e.g. insects [@pone.0095963-Bretman3], [@pone.0095963-Hardy1]. To date, we have no data on the within-population sub-structuring on a scale relevant for mating and distance covered by males during mate search in *A. bruennichi*. Therefore, we cannot accurately estimate the probability for individuals of encountering a sibling.
Generally, a loss in genetic variation would be expected in any species that colonises new habitats as most dispersal mechanisms will result in a small number of individuals that found new populations and hence only represent a subset of the genetic variation of the source population [@pone.0095963-Mayr1]. Spiders lay their eggs in large clutches and egg-sacs of *A. bruennichi* contain several hundreds of eggs [@pone.0095963-Crome1]. In species with an overwintering period such as *A. bruennichi*, all egg-sacs in a population hatch very synchronously regardless of when they were produced [@pone.0095963-Schaefer1]. The common dispersal mode after hatching in spiders is ballooning, which means that the animal releases a thread of silk until it is uplifted by thermic or wind [@pone.0095963-Bell1]. This mode of travelling is generally restricted to very small spiders and is risky since the spider has very limited options to control where it will be going [@pone.0095963-Suter1], [@pone.0095963-Bonte1]. Hence one might expect that at least a proportion of hatchlings remain at their natal site, which has proven to be of sufficient quality. These spiders may disperse by walking or bridging and settle nearby, causing a population substructure with patches of individuals that are closely related. Such a pattern has been found in the eresid spiders *Stegodyphus lineatus* [@pone.0095963-Bilde1] and *S. tentoriicola* [@pone.0095963-Ruch1], in which newly established nests are clustered around maternal sites. Unless there is sex-specific early dispersal, males may mature in the proximity of their sisters promoting inbreeding. In *S. lineatus*, males initially mate close to their birth site accepting a risk of inbreeding and then adopt a long distance mate search of higher risk [@pone.0095963-Bilde1]. Furthermore, a few spider females can produce a lot of offspring and quickly fill suitable web-sites at a location with her offspring. Depending on the degree of substructure and the probability of mating with a sibling, selection should favour kin-recognition mechanisms during mate choice if inbreeding is associated with more costs than benefits. However, generally rejecting related individuals as mating partners can be disadvantageous if the probability of finding a different mating partner is unpredictable. Female web-building spiders do not actively search for mates and face a risk of remaining unmated, hence they may benefit from accepting any male initially to secure fertilisation of her eggs leaving options for further copulations with preferred sires. Polyandry will then be in the female\'s interest because paternity could be biased towards the best mate [@pone.0095963-Jennions1], [@pone.0095963-Zeh2]. Post-copulatory choice has been demonstrated in several *Argiope* species [@pone.0095963-Schneider3], [@pone.0095963-Elgar1] and it was shown to be based on relatedness in *A. lobata* in which females cryptically favour sperm from unrelated males [@pone.0095963-Welke1]. Pre-copulatory recognition seems to be present as well, since siblings mate for shorter and have a lower rate of sexual cannibalism [@pone.0095963-Welke3]. Such a strategy enables males that survive their first copulation to leave and search for a better second mating opportunity [@pone.0095963-Fromhage2]. A trading-up mechanism, in which both, females and males, first mate indiscriminately to secure a sperm supply and then try to re-mate with a higher quality mate, appears to be relatively common in spiders [@pone.0095963-Welke1].
While the above conditions largely apply for *A. bruennichi*, the high variation in all our samples strongly suggests that dispersal is very efficient in this species so that each patch of suitable habitat will soon be inhabited by a relatively large number of individuals from several origins [@pone.0095963-Krehenwinkel1].
Even though our treatment of mating individuals that originated from the same population showed a reduced hatching success, this effect was not apparent in the genetic distance of the experimental pairs. Several authors suggest that a sufficient number of markers are required to detect inbreeding depression in natural populations [@pone.0095963-Pemberton1], [@pone.0095963-Grueber1]. Even studies with a relatively large number of microsatellite loci (\>20) gave poor evidence for inbreeding depression [@pone.0095963-Grueber1]. By using 16 polymorphic microsatellite loci we clearly detected the difference between siblings and non-siblings, but no differences within the latter group added to the notion that such measurements alone are not always appropriate to predict risks and costs of inbreeding. The reduced hatching success of within-population matches might have resulted from incompatibilities that are not detected using microsatellites.
Indeed, we found an amino acid change between Alanine and Threonine in the mitochondrial genome of several individuals. It seems that pairs in which females carry this mutation and mated with males from a different population exhibit a higher hatching success (unpublished data). The interaction between the mutation and genetic composition of mating pairs suggests more complex genetic interactions and might be one possible explanation of the higher reproductive success of among-population pairs compared to the within-population pairs.
It remains an open question how relevant incompatible matings are in natural populations that may show a much larger intermixture of genotypes through long distance dispersal. The rapid range expansion of *A. bruennichi* suggests that they are potent ballooners although it is unclear whether all hatchlings of an egg-sac balloon or whether a proportion stays. Published accounts are inconsistent in this respect [@pone.0095963-Walter1], [@pone.0095963-Follner1]. One would expect that an obligate high-risk dispersal phase should be opposed by selection just as much as the opposite of no dispersal, which would facilitate inbreeding as well as kin competition.
As the calculation of heterozygosity based on the microsatellite data set revealed a conspicuous difference between observed and expected heterozygosity of the four populations that did not relate to the genetic distance data, the presence of null alleles was tested for each locus and was confirmed in some loci. Null alleles occur through a failure of amplification during PCR leading to an over-estimation of homozygotes. Therefore we chose the mitochondrial COI gene as an additional marker to better trace the genetic diversity of the four populations. However, a DnaSP analysis of the COI gene data confirmed the similar genetic diversity within the four populations. A comparison with other studies showed that null alleles seem to be widespread in spiders [@pone.0095963-Rutten1], [@pone.0095963-Bilde3]. This might be explained by enormous population sizes of spiders providing increased mutation opportunities that lead to changes in primer binding sites and consequently inaccurate sequencing with the designed microsatellite primers. Moreover, in *A. bruennichi* an admixture of different lineages occurs resulting in the introgression of Asian alleles in populations of Northern Europe [@pone.0095963-Krehenwinkel1] that might lead to an excess of non-amplifying loci. Future studies on spiders that involve usage of microsatellite markers should be aware of a potentially high risk of null alleles.
In conclusion, our results show that sibling matings lead to severe inbreeding depression in *A. bruennichi* spiders and that there should be strong selection for inbreeding avoidance. The genetic data suggest that active partner choice would be beneficial even in small and recently founded populations as the genetic diversity is high and consequently the probability of finding a compatible partner is generally high. However, reduced hatching success in pairings of spiders derived from egg-sacs of the same population was not mirrored in the genetic distance data. Incompatibilities other than those caused by inbreeding may be responsible for the reduced hatching success.
Due to the experimental exclusion of ballooning and missing data on small scale population sub-structuring in *A. bruennichi*, the probability for individuals of encountering siblings cannot yet be estimated accurately. Studies are under way to close this gap by identifying the genetic population structure of natural populations close to and during the mating season on a small spatial scale. Furthermore, future experiments are of interest to test whether *A. bruennichi* has evolved pre-copulatory avoidance mechanisms to prevent or at least reduce costs of inbreeding depression in the field. During field studies, we commonly observed that males reject virgin females in the field without any obvious reasons [@pone.0095963-Zimmer1], [@pone.0095963-Schulte1]. Incompatibilities that result in reduced hatching success might be a reason.
We thank Tomma Dirks and Angelika Taebel-Hellwig for their energetic support. Great thanks to Claudia Drees, Jes Johannesen, Matjaz Kuntner, Jasmin Ruch, and Tom Tregenza for discussion and fruitful comments on the manuscript. Special thanks to Prof. Dr. D. Tautz for giving access to the genetic laboratory and associated facilities of the Max-Planck Institute for Evolutionary Biology.
[^1]: **Competing Interests:**The authors have declared that no competing interests exist.
[^2]: Conceived and designed the experiments: JMS. Performed the experiments: SMZ. Analyzed the data: SMZ. Contributed reagents/materials/analysis tools: SMZ HK. Wrote the paper: SMZ JMS.
| {
"pile_set_name": "PubMed Central"
} |
Introduction {#sec1}
============
Silicosis is considered as one of the serious types of pneumoconiosis and a potentially fatal occupational fibrotic lung disease \[[@B1]\], which is highly prevalent in developing countries, especially in China, South Africa, and Brazil \[[@B2]\]. Accumulating evidence shows that silicosis is caused by long-term occupational or environmental exposure to free crystalline silica (CS) particles \[[@B3]\]. There are estimated to be tens of millions of workers exposed to CS worldwide and industrialization processes have made this situation even worse \[[@B4]\]. To date, the increasing mortality of silicosis has made silica exposure a high-priority public health concern in countries worldwide \[[@B5]\].
It has been accepted that the hallmarks of lung silicosis consists of massive inflammation and significant lung fibrosis \[[@B6],[@B7]\]. When respirable CS particles are inhaled, their entry into alveoli induces oxidative stress through the formation of reactive oxygen (ROS) and nitrogen species due to the generation of siloxil radicals after crystalline-silica fracturing \[[@B8]\]. Therefore, it leads to extensive fibroblast proliferation and deposition of extracellular matrix (ECM) with the lungs and further triggers cytotoxicity, oxidative stress, pulmonary inflammation, and eventually silicosis \[[@B9]\]. In the meantime, the progression of lung fibrosis is accompanied with infiltration of inflammatory cells including eosinophils, neutrophils, and macrophages. Macrophages, for instance, can phagocyte silica particles and become activated to release a multitude of mediators, such as histamine and serotonin, which are retained in lung throughout inflammation and mediate cascade reaction leading to fibrotic response \[[@B10]\]. In addition, alveolar macrophages, as well as their produced TNF-α, interleukin (IL)-1β and IL-6 has been suggested to play a crucial role in inflammation, as a hallmark of exposure to silica \[[@B11]\]. Besides the aberrant induction of pulmonary inflammation, the intratracheal silica instillation leads to the enlargement of thoracic lymph nodes. The mitogenic responses to T cell receptor stimulation are apparently reduced in lymphocytes from silica-exposed lymph nodes with markedly increased activation-induced cell death, suggesting that silicotic lung apoptosis has been also implicated in the development of the initial inflammation \[[@B12]\]. During the process, Fas-L expression was inversely associated with mast cells, collagen/elastic deposition. As the progression of silicosis involves a fibrotic phase in which the ECM is deposited and lung parenchyma is remodeled, researches have been focussing on the intervening the inflammation processes and CS-induced fibrosis of silicosis \[[@B13]\]. Therefore, there is an urgent need to identify antioxidative and anti-inflammatory agents that might be promising for the prevention and treatment of silicosis.
N-acetylcysteine (NAC), a semi-essential amino acid with a thiol side chain, is the acetylated precursor of cysteine. It is clinically used to treat a wide variety of medical issues and conditions \[[@B14]\], such as acute respiratory distress syndrome, heavy metal-induced toxicity, and ameliorate certain psychiatric disorders \[[@B15]\] for its multiple beneficial properties, including enhancing bone regeneration, reducing post-surgical complications \[[@B16]\], and preventing mutagenic irradiation \[[@B17]\]. Furthermore, NAC has been reported to possess antioxidant properties by scavenging reactive oxygen species (ROS) and also inhibit the activity of cyclooxygenase-2 and membrane lipid peroxidation induced by inflammation \[[@B18],[@B19]\]. Coombes et al. suggest that NAC has reno-protective activities in chronic kidney disease \[[@B20]\] and neuro-protective effect on spinal cord-injury \[[@B21]\]. Preliminary data in rat model revealed that oral treatment with high-dose NAC during early silica exposure can ameliorate the activity of proinflammatory cytokines, down-regulating ROS, and mitochondrial apoptosis signaling, thus attenuating subsequent lung fibrosis, suggesting the potential of NAC in the treatment for silica-induced lung fibrosis \[[@B22],[@B23]\]. In the present study, we employed a mouse model of silicosis and different doses of NAC, in order to determine the potential pulmonary protective effects of NAC and the underlying mechanism.
Methods {#sec2}
=======
Animal groups and treatments {#sec2-1}
----------------------------
Total 138 female wild-type C57BL/6J mice aged 6--8 weeks were purchased from the Experimental Animal Center of Hainan Medical College and randomly divided into the following groups: (1) blank control group (*n*=18)-animals without any treatment; (2) CS-induced model group (*n=*30); and (3) NAC-treated groups-animals were underwent CS exposed, and then administrated with NAC via gavage every day at a dose of 1.73 mg/20g (low-dose, *n=*30, NAC-L group), 3.46 mg/20g (moderate-dose, *n=*30, NAC-M group), or 5.19 mg/20g (high-dose, *n=*30, NAC-H group), respectively. For CS-induced model group, suspensions of CS (CAS: 7631-86-9, 2 μm, U.S. Silica Co., WV, Sigma--Aldrich) were prepared in normal saline (0.9% \[w/v\] NaCl). Each animal was received CS administration (2.5 mg suspended in 60 μl of saline) through intratracheal instillation and then administrated with saline via gavage every day, as previously described \[[@B24]\]. On months 0 (24 h), 1, 2, 3, 4, and 5 after treatment, animals were killed and harvested. For blank control group, three mice were used at every time point. Total five mice were used for analysis for the other experimental groups at every time point. The present study was performed in strict accordance with the recommendations from the Guide for the Care and Use of Laboratory Animals of the National Institutes of Health. All mouse experiments were approved by the Ethics Committee on Animal Research of the Hainan Medical College.
Preparation of bronchoalveolar lavage fluid {#sec2-2}
-------------------------------------------
Bronchoalveolar lavage fluid (BALF) was processed and collected as previously described \[[@B25]\]. Briefly, mice were intraperitoneally injected with 10% chloral hydrate for euthanasia. Immediately after euthanasia, the trachea of each mouse was exposed through a midline incision and cannulated with a sterile 22-gauge Abbocath-T catheter. Approximately 1.0 ml of BALF was retrieved per mouse, and BALF samples were kept on ice to avoid cell lysis.
Serum and tissue sample collection {#sec2-3}
----------------------------------
Mice in the above groups were killed and blood samples were obtained from abdominal aorta. After centrifuged at 1500 rpm for 10 min, serum was collected and stored at −80°C. The lung tissues were removed and immediately frozen in liquid nitrogen. Some of the lung tissues were fixed with 4% paraformaldehyde in phosphate-buffered saline for 24 h. The remaining lung tissues were frozen in liquid nitrogen for RNA detection or ELISA assay.
ELISA assay {#sec2-4}
-----------
The mouse TNF-α ELISA Kit (CSB-E04741m, CUSABIO), mouse IL-1β (CSB-E08054m, CUSABIO), mouse IL-4 (CSB-E04634m, CUSABIO), and mouse IL-6 (CSB-E04639m) were used to determine the concentrations of TNF-α, IL-1β, IL-6, and IL-4 in plasma or BALF, respectively following the manufacturer's instructions.
Analysis of mRNA expression by RT-PCR {#sec2-5}
-------------------------------------
Total RNA was extracted from mice lung tissues with TRIzol reagent (TaKaRa Biotech), and cDNA was synthesized from 1 μg of total RNA using the Bestar qPCR RT Kit (DBI). RT-PCR was performed using SYBR^®^ Premix Ex Taq™ (TaKaRa) with Roche 480 using the following primers: inducible nitric oxide synthase (iNOS) (F: 5′-GTTCTCAGCCCAACAATACAAGA-3′ and R: 5′-GTGGACGGGTCGATGTCAC-3′); NADPH oxidase 1(NOX1) (F: 5′-CCTGATTCCTGTGTGTCGAAA-3′ and R: 5′-TTGGCTTCTTCTGTAGCGTTC-3′); NOX2 (F: 5′-GTATTGTGGGAGACTGGACG-3′ and R: 5′-ACAGACTTGAGAATGGAGGC-3′); NOX4 (F: 5′-GGTGTCTGCATGGTGGTGGTATT-3′ and R: 5′-CAGCCAGGAGGGTGAGTGTCTA-3′); SOD2 (F: 5′-CAGACCTGCCTTACGACTATGG-3′ and R: 5′-CTCGGTGGCGTTGAGATTGTT-3′); xanthine oxidase (XO) (F: 5′-ATGACGAGGACAACGGTAGAT-3′ and R: 5′-TCATACTTGGAGATCATCACGGT-3′); internal standard β-actin (F: 5′-CATTGCTGACAGGATGCAGA-3′ and R: 5′-CTGCTGGAAGGTGGACAGTGA-3′). The expression data were normalized to actin and quantitated using the Stratagene Mx3000P Real time PCR (Agilent, U.S.A.). Relative expression was calculated using 2^−ΔΔ*C*^~T~ method.
H&E staining, IHC, and Masson's trichrome staining {#sec2-6}
--------------------------------------------------
Morphological changes in the lung tissue were investigated using H&E staining. The tissues were embedded in paraffin and cut into 4-μm thick sections with a microtome. H&E was used to perform a H&E staining. As for IHC, antibodies against E-cadherin (Cell Signaling), Vimentin (Cell Signaling), and Cytochrome C (Abcam) were used to incubate slices (4 μm) overnight after deparaffinized. Afterward, the slices were incubated with HRP-conjugated secondary antibody followed by diaminobenzidine and hematoxylin staining orderly. As for Masson staining, slices (5 μm) were stained with Masson trichrome solutions. All the stained slices were then observed under a light microscope (BX51; Olympus Corporation).
Oxidative stress analysis {#sec2-7}
-------------------------
Lung tissues of mice were homogenized with saline (1: 9) on ice, followed by centrifugation at 2500 rpm for 10 min at 4°C. The supernatant was used to determine the antioxidant activity including total antioxidative (T-AOC), glutathione peroxidase (GSH-Px), superoxide dismutase (SOD) and oxidative stress marker malondialdehyde (MDA) using commercial ELISA kits from Jiancheng Institute of Biotechnology (Nanjing, China) according to the manufacturer's instruction.
Flow cytometric ROS analysis {#sec2-8}
----------------------------
Production of ROS in blood was measured using fluorescent dye 2′, 7′-dichlorodihydrofluorescein diacetate (DCFH-DA, Beyotime Institute of Biotechnology, China) according to the manufacturer's instruction. About 1 ml of blood was drawn from the main abdominal vein and anticoagulated with heparin. In brief, two test tubes were prepared, one without PMA (Phorbol 12-myristate 13-acetate, ROS trap phorbol), and the other with PMA, PMA (−) tube and PMA (+) tube. Anticoagulation with heparin 0.1 ml was added to each test tube. DCFH-DA 2 ml was added into the two tubes respectively and then placed in a 37°C water bath for 15 min. To prevent neutrophil agglutination, EDTA 500 μl was added into two tubes and then two tubes were placed in a 37°C water bath for 20 min. Supernatant was removed by centrifugation at 1500 rpm for 5 min. After being washed with PBS twice, cells were resuspended and fluorescence intensity of positive PMN (neutrophil) was measured by using flow cytometry (FACSCalibur, Becton--Dickinson).
TUNEL assay {#sec2-9}
-----------
Fixed lungs were embedded in paraffin, and sequential 5-μm thick sections were processed for TUNEL staining using the ApopTag Peroxidase in Situ Apoptosis Detection Kit (Millipore, Billerica, MA, U.S.A.). Briefly, the lung tissue sections were stained and TUNEL-positive cells were observed under the microscope following the previous report \[[@B26]\].
Statistical analysis {#sec2-10}
--------------------
The presented data in the study were obtained in at least three independent experiments. These quantitative data were expressed as the mean ± S.D. with statistical analysis performed using the SPASS.19.0 statistics package (SPSS Inc., Chicago, U.S.A.). Differences between the groups were assessed by one-way ANOVA test. Statistical significance was set at *P* values \< 0.05.
Results {#sec3}
=======
Morphological findings of lung tissues in mice silicosis model {#sec3-1}
--------------------------------------------------------------
The morphometric assessment was performed in order to evaluate lung damage and inflammatory response in lungs from mice treated with CS. As shown in [Figure 1](#F1){ref-type="fig"}, we found lung chronic inflammation, enlargement of the alveolar air spaces and destruction of the lung parenchyma in all CS-treated model mice, characterized by epithelial hyperplasia, granulomas and inflammatory cell infiltrates, compared with the normal alveolar architecture in blank control group. Moreover, time-dependent progression of the phenomenon was clearly observed. In particular, 2 months after silica administration, nodular silicotic granulomas accompanied by marked epithelial hyperplasia were found in central areas. After 4 months treatment, granulomas were confluent and included large acellular centers with peripheral infiltration of lymphocytes. To determine the effects of NAC on established mice silicosis model, NAC was administered after CS treatment. As expected, we observed an alleviation of emphysema in all NAC groups compared with the CS-induced model group. Moreover, moderate- and high-NAC treatment markedly reduced CS-induced pulmonary injury relative to the low-NAC group.
{#F1}
The effects of NAC on oxidative stress in mice silicosis model {#sec3-2}
--------------------------------------------------------------
To determine the CS-induced early-stage oxidative stress in mice, production of ROS in blood was measured with DCFH-DA assay. FACS analysis was performed. As shown in [Figure 2](#F2){ref-type="fig"}, intracellular ROS nearly kept stable levels in blank control group. In model group, intracellular ROS significantly increased from 1 to 3 months, but slightly decreased at month 4 and 5. After pretreatment with NAC at low, moderate, and high concentration, respectively, intracellular ROS was significantly reduced compared with that in model groups, especially in 3, 4, and 5 months. Next, the levels of oxidative stress marker and antioxidants were determined using ELISA assay.
{#F2}
NAC alleviated inflammatory response and oxidative stress through mediating expression of NOX1, NOX2, iNOS et al. {#sec3-3}
-----------------------------------------------------------------------------------------------------------------
As shown in [Figure 3](#F3){ref-type="fig"}A, the oxidative stress marker, MDA was distinctly increased, but the levels of GSH-PX, SOD and T-AOC, the antioxidant markers to increase clarity, were reduced in the model group as compared with those in the control group. In contrast, mice from the NAC-treated groups recovered. Meanwhile, expression of NOX2, iNOS, SOD2, and XO mRNA increased in lung tissues after induced by CS compared with the control group in time-dependent manner. However, the tendency was reversed significantly by NAC administration compared with Model mice in a dose-dependent manner ([Figure 3](#F3){ref-type="fig"}B).
{#F3}
To investigate the possible effect of NAC on inflammation, we measured the concentrations of TNF-α, IL-1β, IL-4, and IL-6 in plasma and BALF. As shown in [Figure 3](#F3){ref-type="fig"}C, TNF-α was significantly elevated in plasma sample in model group compared with control mice but decreased by NAC administration. No significant differences in IL-1β, IL-6, IL-4, and IL-6 plasma levels were found between groups ([Figure 3](#F3){ref-type="fig"}C upper panels). However, we found a significant increase of TNF-α, IL-1β, IL-4, and IL-6 in BALF from model group when compared with control group or NAC administration group ([Figure 3](#F3){ref-type="fig"}C, lower panels). These results revealed that the administration of NAC alleviated the injury of airway inflammation and repressed pro-inflammatory cytokines in mice silicosis model.
NAC administration alleviated the CS-induced tissue damages and pulmonary fibrogenesis in mice lungs {#sec3-4}
----------------------------------------------------------------------------------------------------
To further evaluate the effects of NAC on CS-induced chronic pulmonary inflammation, we selected lung tissue samples of mice at the 5 months to perform pathological examination. Masson staining was first used to demonstrate silica-induced pulmonary fibrosis. As shown in [Figure 4](#F4){ref-type="fig"}A, there were apparent increasing fibroblast and considerable fibroplasias (Blue), as substantial amount of collagen deposition (Red) was stained in model group compared with control group. These CS-induced symptoms were then obviously reversed by NAC administration. Moreover, TUNEL analysis showed that NAC can attenuate cell apoptosis induced by CS ([Figure 4](#F4){ref-type="fig"}B). Collectively, NAC administration could partially alleviate the tissue damage and fibrogenic response in lungs from mice silicosis model.
{#F4}
Epithelial-mesenchymal transition (EMT) has been reported to play a key role in the pathologic process of lung injury and fibrotic lung diseases \[[@B27]\]. As shown in [Figure 4](#F4){ref-type="fig"}C-D, E-cadherin was suppressed whereas vimentin was increased in lungs from mice following CS exposure. In contrary, NAC administration notably elevated E-cadherin expression but decreased vimentin and Cytochrome C expressions in CS-induced mice silicosis model. These results further indicate that CS exposure induces EMT characteristics in lungs from mice *in vivo*, which could be partially reversed by NAC administration.
Discussion {#sec4}
==========
Silicosis, as a type of chronic pulmonary diseases, is a progressive occupational lung disease caused by the long-term inhalation, deposition, and retention of CS particles \[[@B28]\]. In the present study, we established a mouse silicosis model by chronic exposure to CS in order to investigate the possible effects of NAC on the CS-induced chronic pulmonary disease. Interestingly, previous study with rat model adopted non-tracheal exposure method of disposable intrapulmonary injection of 50 g/l \[[@B22],[@B23]\] while we ulitilized 42 g/l of CS for current study with mice model according to describe previously \[[@B24]\]. We found exacerbated lung injury, increased pulmonary inflammation, and severe fibrogenic response induced by CS exposure in model mice, which was in line with previous finding \[[@B29],[@B30]\]. It suggests that chronic respiratory exposure to CS would increase lung tissue damages and fibrosis, even leading to the occurrence of lung cancer development. As silicosis is caused by long-term occupational or environmental exposure to free CS particles, we thus administered the highest dose of 5.19 mg/20 g (about 260 mg/kg) of NAC, which is much lower than 500 and 600 mg/kg, respectively \[[@B22],[@B23]\], while we extended the observation duration from 28 days in earlier studies to over 5 months in order to simulate the pathogenesis of the disease with mice model. Notably, we observed that oral treatment of NAC in mice ameliorated CS-induced pulmonary fibrosis and inflammation. Consistent with our findings, increasing clinical and animal studies have reported that NAC could reduce acute inflammatory response and tissue damage to lung injury and sepsis, as well as improve impaired function \[[@B31]\].
Lung inflammation and fibrosis are the most common pathological changes resulting from inhalation of airborne particles. As the integral part of the pathogenesis of silicosis, inflammation was induced in the lungs by CS exposure, as demonstrated by elevating cytokines such as TNF-α, IL-1β, IL-4, and IL-6, which has been shown in the study of Li et al. \[[@B36]\]. Among these cytokines, TNF-α and IL-1β are recognized as the earliest factors in lung injury \[[@B37],[@B38]\]. Consistent with the *in vivo* experiments, CS-injured mice given NAC had decreased levels of TNF-α, IL-1β, IL-4, and IL-6 in BALF.
Silicosis is also considered an oxidative stress related disease that can lead to the development of lung cancer \[[@B39]\]. Related studies have showed that phagocytes could produce oxidising enzymes leading to the generation of high levels of ROS during inflammation, which further promote inflammation in a feed forward loop manner \[[@B40],[@B41]\]. Furthermore, oxidative damage caused by ROS may lead to various human diseases, such as cancer and inflammation \[[@B42]\]. As a major source of ROS, NOXs have a critical role in inflammatory response, contributing to ROS production during oxidative damage \[[@B43]\]. These facts further indicated the involvement of oxidative stress is related to chronic inflammation during exacerbation. To clarify the oxidative stress of NAC in mice, we investigated the levels of antioxidant enzymes, oxidative stress marker, and oxdizing enzymes. MDA is an indicator of oxidative stress that is inversely correlated with pulmonary function \[[@B44]\] and could result in the irreversible lung damage \[[@B45]\]. Related study has indicated that antioxidants, such as GSH-PX, T-AOC not only protect against the direct injurious effects of oxidants, but also change the inflammatory events involved in the chronic pulmonary diseases \[[@B45]\]. In our results, CS exposure distinctly increased the oxidative stress marker, MDA, but decreased the antioxidants (GSH-PX, SOD, and T-AOC). In addition, some oxidizing enzymes (iNOS, SOD2, and XO) were also significantly elevated in mice silicosis model. Correspondingly, we found that the antioxidants were increased and oxidizing enzymes were obviously decreased after NAC treatment. To confirm whether NOXs are related with silicosis, several NOXs, including NOX1, NOX2, and NOX4 were determined. The results demonstrated that elevated NOX2 levels in mice silicosis model were significantly reduced after administration with NAC, which suggested that NOXs might play an important role in lung inflammation and fibrosis.
In addition, we investigated the EMT characteristics in lungs from mice silicosis model *in vivo*. EMT is a process that polar adjacent epithelial cells transform to non-polar mesenchymal cells and increase cell mobility, which plays an important role in the development of pulmonary fibrosis \[[@B46]\]. According to the study of Liu et al., EMT was one of the key events in silica-induced pulmonary fibrosis \[[@B47]\]. Interestingly, EMT is closely related with the oxidative stress and its formation could be prevented by agents that have antioxidant properties, thus reducing pulmonary fibrosis \[[@B48]\]. Therefore, we were prompted to investigate whether NAC has a role in EMT characteristics *in vivo* by examining EMT-related markers in mice silicosis model using IHC staining. In line with expectations, we found a significant decrease of E-cadherin and increase of vimentin as well as Cytochrome C in mice silicosis model. On contrary, the expression of E-cadherin was strongly promoted, but vimentin expression was remarkably suppressed by NAC compared with those in the model group. The limitation in our study still exists that although we have tested various indicators, including T-AOC, GSH-Px, SOD, and oxidative stress marker MDA to determine antioxidant activity, hydroxyproline levels in lung tissue, as another indicator of oxidative stress, serum and BALF levels of IL-8 and high-sensitivity C-reactive protein ought to be assessed in further investigation,
N-Acetylcysteine (NAC) is a thiol-containing compound that has been used in clinical practice since the mid-1950s. It acts as a donor of cysteine, leading to replenishment of glutathione and also contributes to antioxidative and anti-inflammatory effects. NAC was originally introduced for the treatment of congestive and obstructive lung diseases. Accumulative data suggest that the drug itself does not accumulate in the body, but rather its oxidized forms and reduced and oxidized metabolites \[[@B49],[@B50]\]. Recent evidence also indicated that, in adults with kidney impairment, NAC can be safely given both intravenous and intra-arterial at a dose of 450 mg/kg \[[@B51]\]. It has been demonstrated the effect of oral treatment with tetrandrine (TD) and NAC jointly on silica-exposed rats, which was better than single use of TD or NAC in alleviating SiO~2~-induced pulmonary fibrosis in rats \[[@B52]\]. Our further study may focus on the sustained or combinated use of NAC in the treatment of silicosis based on evaluation of toxicity and pharmacokinetics profile, which seeks to determine its feasibility in clinical practice.
In summary, our study demonstrated that NAC has protective effects against CS-induced pulmonary fibrosis partially via decreasing CS-induced pulmonary inflammation, ameliorating pulmonary fibrosis, and EMT-like characteristics ([Figure 5](#F5){ref-type="fig"}). Based on these encouraging findings, NAC might become a promising agent for the treatment of silicosis in the future after more clinical investigations have been performed to determine whether NAC has a clinical application in CS-induced pulmonary fibrosis.
{#F5}
Data Availability {#sec5}
=================
The data used to support the findings of the present study are included within the article.
Competing Interests {#sec6}
===================
The authors declare that there are no competing interests associated with the manuscript.
Funding {#sec7}
=======
This work was supported by National Natural Science Foundation of China \[grant number 81360001\] and Foundation of Hainan Province Hygience ministry of China \[grant number 2012PT-24\].
Author Contribution {#sec8}
===================
H.H., M.C., and X.L. made substantial contributions to conception and design, or acquisition of data, or analysis and interpretation of data. F.L., H.W., J.W., and J.C. involved in drafting the manuscript or revising it critically for important intellectual content. H.H., M.L., and X.L. provided final approval of the version to be published.
BALF
: bronchoalveolar lavage fluid
CS
: crystalline silica
ECM
: extracellular matrix
EMT
: epithelial-mesenchymal transition
GSH-Px
: glutathione peroxidase
H&E
: Hematoxylin & eosin
IHC
: immunohistochemistry
MDA
: malondialdehyde
NAC
: N-acetylcysteine
iNOS
: nitric oxide synthase
ROS
: reactive oxygen species
SOD
: superoxide dismutase
T-AOC
: total antioxidative
TD
: tetrandrine
[^1]: These contributed equally to this work.
| {
"pile_set_name": "PubMed Central"
} |
Introduction
============
More than two-thirds of pregnant women exceed the Institute of Medicine (IOM) 2009 gestational weight gain (GWG) recommendations \[[@ref1]\]. According to weight management guidelines for obesity treatment \[[@ref2]\], prenatal care should provide an ideal clinical framework for treatment delivery with frequent visits, weight recording, an established definition of acceptable weight gain, and opportunities for in-person counseling. However, despite several efforts to prevent excessive GWG in clinical trials, it remains unclear if lifestyle interventions can be efficacious, particularly in women with overweight or obesity \[[@ref3],[@ref4]\]. Poor effectiveness in these trials is explained by intervention designs that fail to take advantage of the entire prenatal care continuum because program initiation is often delayed until mid or late gestation and weight management counseling and intervention are limited to one or two in-person sessions \[[@ref3]\]. As more patients have access to mobile phones and 67% of pregnant women subscribe to electronic health information delivery during pregnancy \[[@ref5]\], eHealth interventions designed to target healthy weight gain provide an opportunity for high-intensity and cost-effective interventions to be delivered to all patients throughout prenatal care. The aim of this study was to test whether a personalized gestational weight management program (SmartMoms) delivered in-person or via an intensity-matched mobile phone app could decrease the proportion of women with overweight and obesity that exceed the IOM 2009 guidelines for GWG by 25%.
Methods
=======
This study targeted overweight and obese (body mass index \[BMI\] 25.0-39.9 kg/m^2^) women aged 18 to 40 years expecting a singleton pregnancy in their first trimester. Women with a known fetal anomaly, hypertension (systolic \>160 mm Hg or diastolic \>90 mm Hg), history of or current psychotic or eating disorder, human immunodeficiency virus, preexisting diabetes (self-report or determined by glycated hemoglobin A~1c~ and/or 75 g oral glucose tolerance test in the first trimester), or with contraindications to exercise (by PARmed-X and American College of Obstetricians and Gynecologists committee opinion \#67 \[[@ref6]\]) were excluded. With support of local obstetricians, participants were recruited from brochures placed in various clinics and by study staff during the patients' first prenatal appointment \[[@ref7]\]. Participants were randomized by unblinded intervention staff equally to one of three groups between 10.4 to 13.6 weeks of gestation: (1) no intervention (usual care group), (2) receipt of the SmartMoms intervention in-person (in-person group), or (3) receipt of the SmartMoms intervention via mobile phone (remote group), with randomization stratified by pregravid BMI. The block randomization schedule and sealed numbered randomization envelopes were prepared by the biostatistician. Usual care (control) participants were under the usual care of their obstetrician and did not receive weight management services from the intervention team. The SmartMoms intervention was designed to assist an expectant mother in gaining weight within the recommended 2009 IOM guidelines for her respective BMI class. SmartMoms is grounded in the ability to objectively quantify dietary adherence to an energy intake prescription based on measured body weight and to provide patients with data-driven feedback about their energy intake \[[@ref8]-[@ref12]\]. SmartMoms participants received dietary intake advice, exercise advice, and a weight graph created from the dynamic GWG models to determine the trimester-specific increase in energy intake required by each participant to adhere to the IOM GWG recommendations \[[@ref13]\]. To promote these lifestyle changes, participants received a structured intervention that consisted of 18 lessons and behavior modification strategies. SmartMoms participants received behavior modification counseling weekly between weeks 13 and 24 of gestation and biweekly from week 25 until delivery. Importantly, the content of the lesson materials were identical and only the mode of delivery differed between the two intervention groups. Participants in both the intervention groups were provided with a wireless Internet-connected bathroom scale and a pedometer (in-person group: Omron Healthcare, Lake Forest, IL, USA; remote group: Fitbit Zip, FitBit, San Francisco, CA, USA) to self-monitor body weight and step counts daily. The mobile phone app is similar to a virtual weight management system described for weight loss in which body weight and daily steps are automatically transmitted in real time to personalized charts \[[@ref14]\]. The SmartMoms intervention includes an IOM 2009 GWG weight graph personalized for each patient and behavioral modification tools including daily self-monitoring of weight, dietary intake, and physical activity \[[@ref14]\]. SmartMoms participants were provided with an individualized calorie intake above their estimated prepregnancy energy requirement \[[@ref13]\] or energy gap represented by an ideal weight gain zone \[[@ref15]\], and were coached how to adjust energy intake and/or physical activity to adhere to the IOM 2009 GWG guidelines. The in-person group tracked step counts with pen and paper, and the IOM weight graph was reviewed in hard copy during counseling sessions with interventionists.
Clinic assessments were performed by certified staff who were blinded to group assignment. Maternal weight was measured fasting and in a hospital gown. Total GWG and GWG per week were calculated between the initial (10-13 weeks) and final (35-36 weeks) study visits. GWG per week was used to calculate the proportion of women based on prepregnancy BMI with recommended or excessive GWG per the 2009 IOM GWG guidelines \[[@ref16]\].
Adherence to the SmartMoms intervention was defined as the percentage of days participants weighed and recorded step counts in comparison to the expected number of days. Study economics, including costs incurred for travel to and from treatment sessions and time spent with the counselor while accounting for session attendance and intervention adherence, were calculated for each participant. The clinic economics included cost of interventionist time (training, session preparation, participant contacts, routine staff meetings) and equipment (scale, pedometer) cost.
Statistical analyses were completed using SAS/STAT version 9.4 software of the SAS System for Windows (Cary, NC, USA). Sample size was based on the hypothesis that the proportion of overweight and obese pregnant women in the usual care group exceeding IOM guidelines for GWG would be 58% and that lowering this proportion by at least 25% would be clinically significant. Intention-to-treat comparison of continuous variables (eg, GWG, birth weight) between the three treatment groups used one-way analysis of variance with post hoc pairwise intervention group comparisons. Comparisons of categorical variables (eg, prevalence of excess GWG) between the three treatment groups used Fisher exact test. Equality of adherence to IOM GWG guidelines was tested through one-sided z tests for proportions. Finally, differences in study costs and intervention adherence were assessed through two-sample *t* tests. All tests were performed with significance level alpha=.05, and findings were considered significant when *P*\<alpha. Data are reported as least square (LS) mean and standard error (SE) unless otherwise noted.
Results
=======
Recruitment of participants from community clinics from February 1, 2013 to April 14, 2014, yielded three groups of pregnant women who were similar ([Table 1](#table1){ref-type="table"}). The majority of participants were white and nulliparous or primiparous. No study-related serious adverse events were reported.
Gestational Weight Gain and Guideline Adherence
-----------------------------------------------
The SmartMoms intervention (in-person and remote groups combined) was effective at reducing GWG in overweight and obese pregnant women (usual care: LS mean 12.8, SE 1.5 kg; SmartMoms: LS mean 9.2, SE 0.9 kg; *P*=.04). The in-person group gained significantly less total weight ([Figure 1](#figure1){ref-type="fig"}) during pregnancy than the usual care group (LS mean 8.0, SE 1.3 kg vs LS mean 12.8, SE 1.5 kg; *P*=.04) and weight gain in the remote group was equivalent to the in-person group (LS mean 10.0, SE 1.3 kg; *P*=.04 equivalence) and modestly lower than weight gain with usual care (LS mean 10.0, SE 1.2 kg vs LS mean 12.8, SE 1.5 kg; *P*=.07). Compared to usual care, the rate of GWG was significantly lower in the in-person group (LS mean 0.49, SE 0.06 kg/week vs LS mean 0.31, SE 0.05 kg/week; *P*=.01) and the rate of GWG in the in-person group was equivalent to the remote group (LS mean 0.39, SE 0.05 kg/week; *P*=.04) within 200 grams of weight gained per week. The proportion of women with excess GWG ([Figure 2](#figure2){ref-type="fig"}) was significantly lower in the in-person (56%, 10/18; *P*=.03) and remote groups (58%, 11/19; *P*=.04) compared to usual care (84.6%, 11/13).
######
Baseline characteristics by treatment group (N=54).
-------------------------------------------------------------------------------------------------------------------------------------------------------------------------
Characteristic Usual care (n=17) In-person (n=18) Remote (n=19) *P*^a^
------------------------------------------ ----------------------------------------------------------------------------- ------------------ --------------- --------- ---
Age (years), mean (SD) 29.5 (5.1) 29.2 (4.8) 29.0 (4.2) .96
**Race, n (%)** \ \ \ .24
\ Black 6 (35) 5 (28) 2 (11) \
\ White 11 (65) 11 (61) 16 (84) \
\ Other 0 (0) 2 (11) 1 (5) \
Weight (kg), mean (SD) 86.2 (12) 83.0 (12) 81.1 (13) .45
Parity, n (%) 0.6 (1) 0.7 (1) 0.9 (1) .54
**Pregravid BMI group, n (%)** \ \ \ .84
\ Overweight (25.0-29.9 kg/m^2^) 9 (53) 8 (44) 8 (42) \
\ Obese (30.0-40.0 kg/m^2^) 8 (47) 10 (56) 11 (58) \
**Family income per year (US\$), n (%)** \ \ \ .50
\ \<\$5000-\$39,999 7 (41) 9 (50) 7 (37) \
\ \$40,000-\$99,999 5 (29) 7 (39) 5 (26) \
\ ≥\$100,000 5 (29) 2 (11) 7 (37) \
**Education, n (%)** \ \ \ .24
\ Some high school 0 (0) 1 (6) 0 (0) \
\ High school diploma/GED/1-3 years of college, business, or technical school 7 (41) 5 (28) 6 (32) \
\ College degree 4 (24) 7 (39) 11 (58) \
\ Postgraduate work 6 (35) 5 (28) 2 (11) \
-------------------------------------------------------------------------------------------------------------------------------------------------------------------------
^a^*P* values were derived from ANOVA for continuous variables and Fisher exact test for categorical variables.
{#figure1}
{#figure2}
Intervention Adherence and Study Economics
------------------------------------------
The in-person group recorded weight and step data (weight: mean 57.2%, SD 33.8%; step: mean 44.5%, SD 33.3%) less often than the remote group (weight: mean 71.2%, SD 24.1%; step: mean 72.5, SD 29.0%) and the in-person group attended mean 78% (SD 39%) of planned behavioral sessions. Therefore, overall intervention adherence ([Figure 3](#figure3){ref-type="fig"}) was greater in the remote group than the in-person group (76.5% vs 60.8%; *P*=.049). The intervention cost ([Figure 4](#figure4){ref-type="fig"}) to a participant in the remote group was 3.5 times less than the cost for a participant in the in-person group (mean US \$97, SD \$6 vs mean US \$347, SD \$40 per participant; *P*\<.001). Similarly, the clinic cost to deploy the remote intervention was 50% less than the cost to deploy the in-person intervention (US \$215 vs US \$419 per participant).
{#figure3}
{#figure4}
Discussion
==========
Lifestyle interventions to improve adherence to the IOM GWG recommendations have had modest success in reducing GWG \[[@ref4],[@ref17]\] and little impact reducing the incidence of excess GWG \[[@ref18]\]. The greatest success has been with a recommendation of caloric restriction \[[@ref19]\]; however, due to popular beliefs such as the need to "eat for two" \[[@ref20]\], caloric restriction is not widely accepted among patients, practitioners, or their support systems. Albeit in a small sample, we attribute the success of the SmartMoms intervention to its early initiation (13 weeks gestation) and intervention intensiveness being commensurate with weight management treatment for nonpregnant individuals including self-monitoring with timely feedback, a dietary prescription to foster optimal weight change, and receipt of structured behavior change intervention through delivery of 18 lessons over a 24-week interval beginning at the second trimester.
When deployed remotely through a mobile phone, the SmartMoms intervention was just as effective at reducing the proportion of excess GWG when delivered in-person; however, it was found to be at least 50% more economical for patients and providers with a higher level of patient engagement or adherence. This eHealth intervention, including the provision of a personalized IOM GWG weight graph through Internet-connected devices, easily disseminates supportive health information to patients, and remote patient communication provides an ideal framework for integration into an electronic health record system. Using estimates of interventionist time recorded throughout the study, it is estimated that approximately 30 to 50 new patients per clinician per month could be monitored simultaneously through the remote program by a single health care provider, such as a dietician or lifestyle coach, for universal delivery to all patients within a clinical practice. Similar telehealth services are covered by health care insurance companies, including Medicaid, and are already used by health care facilities across the United States such as in the Veterans Affairs Hospital System for management of chronic health conditions \[[@ref21]\]. The SmartMoms mobile phone intervention tested on community-based obstetrical patients could easily be integrated into standard clinical practice, thereby improving access to effective and efficient health care for millions of American women throughout the entire prenatal care continuum.
We thank the LIFE-Moms consortium members for their contributions to the development and oversight of the common measures and procedures shared across the trials. We also thank numerous local obstetrician gynecologists who allowed us to recruit patients in their clinics and the study participants who dedicated their time to this research.
Funding was received through National Institutes of Health grants U01 DK094418 (LMR), 1 U54 GM104940 (LAG, JHB, JWA), P30 DK072476, and F31 HD084199 (EFS). This study (U01 DK094418) is part of the National Institutes of Health Lifestyle Interventions in Expectant Moms (LIFE-Moms) Consortium. LIFE-Moms is supported by the National Institutes of Health through the National Institute of Diabetes and Digestive and Kidney Diseases (U01 DK094418, U01 DK094463, U01 DK094416, 5U01 DK094466 \[RCU\]), the National Heart, Lung, and Blood Institute (U01 HL114344, U01 HL114377), the Eunice Kennedy Shriver National Institute of Child Health and Human Development (U01 HD072834), the National Center for Complementary and Integrative Health (NCCIH), the NIH Office of Research in Women's Health (ORWH), the Office of Behavioral and Social Science Research (OBSSR), the Indian Health Service, and the Intramural Research Program of the NIDDK. This study was also supported in part by U54 GM104940 from the National Institute of General Medical Sciences of the National Institutes of Health, which funds the Louisiana Clinical and Translational Science Center. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.
Authors\' Contributions: LMR, CKM, and KEH designed the study. EFS, LEC, ADA, JB, and DH collected outcome data. LAG, JWA, SR, HB, AD, TS, and CKM conducted the interventions. JHB and JMT conducted analysis. LMR and LAG wrote the manuscript. All authors reviewed and approved the final draft.
Conflicts of Interest: Drs Redman, Martin, and Thomas developed the trademarked approach of SmartMoms (a registered trademark of the Louisiana State University System). There are no direct benefits to these authors for publication of this manuscript and they have no financial affiliations with the companies who conducted the work to develop the SmartMoms Virtual Weight Management Suite, although they could financially benefit from any licensing of SmartMoms along with LSU-Pennington Biomedical Research Center and Montclair State University. All other authors have no conflicts of interest to declare.
BMI
: body mass index
GWG
: gestational weight gain
IOM
: Institute of Medicine
CONSORT-EHEALTH checklist (v1.6.1).
| {
"pile_set_name": "PubMed Central"
} |
Ghazvini K, Karbalaei M, Kianifar H, Keikha M. The first report of *Streptococcus pluranimalium* infection from Iran: A case report and literature review. Clin Case Rep. 2019;7:1858--1862. 10.1002/ccr3.2374
1. INTRODUCTION {#ccr32374-sec-0001}
===============
*Streptococcus pluranimalium* is unusual streptococcal species, which is rarely isolated from human infection. Limited information is available about the pathogenicity of this species; the present study is the first report from Iran indicating infection with *S pluranimalium*; this study can be a novel insight in pathogenicity of this species.
The members of genus streptococcus are the gram‐positive bacteria, which naturally live in the skin, mucosa membrane, respiratory tract, gastrointestinal tract, and urinary tract. Streptococcal infections were primarily described by Billroth in 1874.[1](#ccr32374-bib-0001){ref-type="ref"} So far, 163 species of this genus have identified, about half of which are reported from human infections.[1](#ccr32374-bib-0001){ref-type="ref"}, [2](#ccr32374-bib-0002){ref-type="ref"} *Streptococcus pluranimalium* was first isolated and reported by Devriese et al (1999) from domestic animal infections.[3](#ccr32374-bib-0003){ref-type="ref"} *S pluranimalium* can cause a wide range of infections, including mastitis, tonsillitis, genital tract infection, and brain abscesses in cattle, respiratory tract infection in canary, septicemia, and endocarditis in chicken, as well as tonsillitis in cat and goat.[4](#ccr32374-bib-0004){ref-type="ref"}, [5](#ccr32374-bib-0005){ref-type="ref"}, [6](#ccr32374-bib-0006){ref-type="ref"} Moreover, there are several reports nowadays about *S pluranimalium* from human infections, containing subdural empyema, endocarditis, brain abscesses, and septicemia.[7](#ccr32374-bib-0007){ref-type="ref"} Since the source of this bacterium is blood, milk, and other infectious secretions of animals, it seems that the microorganism has some animal reservoirs, and transmitted to human in the form of zoonosis.[8](#ccr32374-bib-0008){ref-type="ref"}, [9](#ccr32374-bib-0009){ref-type="ref"} According to review of the literatures, vancomycin, aminoglycosides, and cephalosporins are known as the first choice of therapeutic agents for *S pluranimalium* infection.[5](#ccr32374-bib-0005){ref-type="ref"} Despite many studies regarding the pathogenic or opportunistic nature of this bacterium, the fact is still not clearly understood.[7](#ccr32374-bib-0007){ref-type="ref"}, [10](#ccr32374-bib-0010){ref-type="ref"}
The aim of present study was to report the first case of human septicemia due to *S pluranimalium* in Iran.
2. CASE PRESENTATION {#ccr32374-sec-0002}
====================
An Iraqi 2.5‐month‐old infant with clinical manifestations such as lethargy, vomiting, and anorexia was brought to the emergency department of the pediatric hospital in Mashhad, Iran, in November 2018. The initial examinations were done, which included pupil dilation, temperature: 37.2°C, and blood pressure: 75/55. The laboratory results are listed in Tables [1](#ccr32374-tbl-0001){ref-type="table"}, [2](#ccr32374-tbl-0002){ref-type="table"}, [3](#ccr32374-tbl-0003){ref-type="table"}, and [4](#ccr32374-tbl-0004){ref-type="table"}.
######
Arterial blood values
Index pH pCO~2~ (mm Hg) pO~2~ (mm Hg) HCO~3~ ^−^ (mmol/L)
-------------- ----------- ---------------- --------------- ---------------------
Patient case 7.55 25.6 116.4 26.4
Normal range 7.35‐7.45 35‐45 80‐100 22‐28
John Wiley & Sons, Ltd
######
Analysis of complete blood count (CBC)
Index WBC RBC Hb HCT Platelets
-------------- ------------------------------------------------ ----------- --------- ------- -------------
Patient case 22.760 (PMN: 87%, Lymph: 13%) 2.86 7.7 24.2 1284
Normal 5000‐19500 (PMN: 1000‐9000, Lymph: 2500‐16500) 2.70‐4.50 11‐17.1 33‐55 10000‐45000
John Wiley & Sons, Ltd
######
Analysis of urine
Index WBC RBC Epithelial cells pH SG (specific gravity)
-------------- ----- ------ ------------------ ------- -----------------------
Patient case 20 Many 1‐5 5 1.010
Normal 0 0 1‐3 4.6‐6 1.010‐1.020
John Wiley & Sons, Ltd
######
Analysis of biochemical factors in blood serum
Index Urea Creatinine Na K AST ALT ALP CRP
-------------- --------- ------------ --------- --------- ------ ------- --------- ------
Patient case 23 0.6 137 4.72 33 51 465 78.5
Normal 2.0‐7.0 0.2‐0.4 130‐140 3.5‐6.0 9‐80 13‐45 113‐360 10
John Wiley & Sons, Ltd
However, the sonography results of lungs, kidneys, and urinary tract were normal, and also the result of patient\'s UC (urine culture) was negative. Nonetheless, patient immediately transported to the PICU ware, and pulse oximetry was done for patient. A chronic intraventricular hemorrhage (IVH) grade 1 was observed in the sonography of the right caudothalamic groove of brain (Figure [1](#ccr32374-fig-0001){ref-type="fig"}A). Also, an obvious hypoechoic mass containing internal cystic region lacking vascularity was observed with 31x51 dimensions in the left temporal lobe. LP (lumbar puncture) was taken twice from the patient, and the results included albumin: 3.0‐3.4, CSF culture (twice): negative. In another step, the blood culture was done three times for the patient by BACTEC method; the blood culture results were positive. After three days, the greenish‐pinpoint (Colony size less than 1 mm) colonies were appeared on chocolate agar medium (containing 5% CO~2~) (Figure [1](#ccr32374-fig-0001){ref-type="fig"}B). The characterized properties of bacteria included gram‐positive cocci, α‐hemolytic, and negative catalase. However, the phenotypic tests were nonconclusive; for example, the laboratory results for this bacterium were negative for hippurate hydrolysis, inulin, and VP, as well as optochin‐sensitive (Figure [1](#ccr32374-fig-0001){ref-type="fig"}C). Finally, the desired bacterium was identified as *S pluranimalium* by using the VITEK 2 system, automated instrument for rapid and accurate microbial identification (ID) and antibiotic susceptibility testing (AST) (bioMérieux). Furthermore, the species identification was confirmed using 16S rRNA sequencing method (99% similarity with *S pluranimalium* strain DSM 15636). The phylogenetic relationship of our isolate and the closely related *S pluranimalium* were investigated using 16S rRNA gene sequence by MEGA 5 software, the Neighbor‐Joining (NJ) method and Kimura\'s two parameter (K2P) distance correction model with 1000 bootstrap replications; the phylogenic analysis confirmed the high homogeneity of *S pluranimalium* strains (Figure [2](#ccr32374-fig-0002){ref-type="fig"}).
{#ccr32374-fig-0001}
{#ccr32374-fig-0002}
Based on the CLSI instruction, the antibiogram test was done by disk diffusion method in Mueller‐Hinton agar medium supplemented with 5% sheep blood. In this method, the sensitivity of *S pluranimalium* was investigated against several disks such as ampicillin, cefepime, cefotaxime, ceftriaxone, clarithromycin, clindamycin, levofloxacin, linezolid, penicillin, vancomycin, and trimetoprim‐sulfametoxazol. Nonetheless, due to strict‐growth nature of this bacterium, and lack of primary growth, the results were not reliable. Nevertheless, according to the previous reports, the patient was empirically initiated by using the daily administration of vancomycin (15 mg/kg) and ceftriaxone (50 mg/kg). Fortunately, the symptoms of disease improved and discharged from the hospital after 3 weeks, with proper satisfaction of the infant\'s parents, before they returned to their country.
3. DISCUSSION {#ccr32374-sec-0003}
=============
The *S pluranimalium* is an uncommon streptococcus species, which has been isolated and reported from human and animal infections. The term of pluranimalium indicates that bacterium is able to cause infections in different animals.[3](#ccr32374-bib-0003){ref-type="ref"}, [4](#ccr32374-bib-0004){ref-type="ref"} Based on phylogenic studies of 16S rRNA, it is revealed that this bacterium has close relationship with some streptococcus species such as *S hyovaginalis*, *S thoraltensis*, *S halotolerans*, and salivarius group. In general, it seems that the *Streptococcus sobrinus*, *Streptococcus salivarius, and S pluranimalium* have been derived from a common ancestor.[6](#ccr32374-bib-0006){ref-type="ref"}, [11](#ccr32374-bib-0011){ref-type="ref"} Pan et al (2018) succeeded in sequencing the complete genome of *S pluranimalium*. The sequencing information showed that this bacterium has various virulence factors such fibronectin binding protein (FBP), hemolysin, sortase, IgA1 protease, type IV secretion system, and one series of antibiotic‐resistance genes including *mef* (A), *msr* (D), and *lnu* (C), which cause resistance to Erythromycin and Lincomycin antibiotics.[6](#ccr32374-bib-0006){ref-type="ref"} In terms of phenotypic properties, it seems that *S pluranimalium* is quite similar to *Streptococcus suis*, *Streptococcus acidominimus*, *S hyovaginalis*, and *S thoraltensis*. However, sometimes, this bacterium may not be identified as compared to other closely related *Streptococci* species.[5](#ccr32374-bib-0005){ref-type="ref"}, [11](#ccr32374-bib-0011){ref-type="ref"} The most significant phenotypic features of *S pluranimalium* include white small needle‐like colonies, α‐hemolysis, hippurate hydrolysis, as well production of pyrrolidonyl arylamidase (PYR), β‐galactosidase, alkaline phosphatase (ALP), and arginine dihydrolase enzymes.[5](#ccr32374-bib-0005){ref-type="ref"}, [10](#ccr32374-bib-0010){ref-type="ref"}, [11](#ccr32374-bib-0011){ref-type="ref"} Several methods such MALDI‐TOFMS, sequencing of 16S rRNA, and commercial diagnostic kits such as API 20 Strep, Rapid ID32 Strep, and Vitek 2 are considered as accurate options for rapid and reliable identification of uncommon streptococcus species, including *S pluranimalium* of the clinical samples.[7](#ccr32374-bib-0007){ref-type="ref"} Similar to the present study, four reports of human infections by *S pluranimalium* were diagnosed by the use of Vitek 2.[5](#ccr32374-bib-0005){ref-type="ref"}, [10](#ccr32374-bib-0010){ref-type="ref"}, [12](#ccr32374-bib-0012){ref-type="ref"}, [13](#ccr32374-bib-0013){ref-type="ref"} Despite the reports about various infections in animals by *S pluranimalium*, the reports about human infections are limited, and it is not clear that this microorganism is either the initial pathogen or not.[7](#ccr32374-bib-0007){ref-type="ref"} The present study is so far the seventh report about human infection by this bacterium. Based on the issued reports, this bacterium was isolated from endocarditis, septicemia, and brain abscesses.[12](#ccr32374-bib-0012){ref-type="ref"}, [13](#ccr32374-bib-0013){ref-type="ref"}, [14](#ccr32374-bib-0014){ref-type="ref"} In this study, *S pluranimalium* was isolated from the disseminated infection occurred to a 2.5‐month‐old infant. According to the study by Jayavardhana and Maher, both cases with brain abscesses caused by *S pluranimalium* had predisposing risk factors for the infection.[5](#ccr32374-bib-0005){ref-type="ref"}, [12](#ccr32374-bib-0012){ref-type="ref"} Fotoglidis et al isolated this bacterium from the infective endocarditis in patient, who had previous contact with the infected animals.[13](#ccr32374-bib-0013){ref-type="ref"} Regarding the inefficiency of immune system in infants on the one hand, and also considering the previous studies on the other hand, it seems the infection with *S pluranimalium* can occur following the deficiency of immune system, and/or due to the risk factors with respect to the previous infections (Table [5](#ccr32374-tbl-0005){ref-type="table"}).
######
Information about six previous case reports regarding the *S pluranimalium* infections
Sex Age Infection Previous disease Diagnostic method Ref
--------- --------- --------------- ----------------------------------- ------------------- ------------------------------------------
Male 44 Brain abscess Tuberculosis Vitek 2 [5](#ccr32374-bib-0005){ref-type="ref"}
Male 17 Brain abscess Unknown Vitek 2 [10](#ccr32374-bib-0010){ref-type="ref"}
Male 3 Brain abscess Congenital cyanotic heart disease Vitek 2 [12](#ccr32374-bib-0012){ref-type="ref"}
Male 37 Endocarditis Sinusitis Vitek 2 [13](#ccr32374-bib-0013){ref-type="ref"}
Female 53 Septicemia Health Vitek 2 [14](#ccr32374-bib-0014){ref-type="ref"}
Unknown Unknown Septicemia Unknown PCR [15](#ccr32374-bib-0015){ref-type="ref"}
John Wiley & Sons, Ltd
The studies show that a combination therapy with vancomycin, tetracycline, and the third‐generation cephalosporins is a reliable therapeutic regimen for endocarditis and septicemia infections.[13](#ccr32374-bib-0013){ref-type="ref"} Meanwhile, antibiotics such as vancomycin, the third‐generation cephalosporins, imipenem, meropenem, amikacin are very effective on brain abscesses and CNS infections caused by this bacterium.[5](#ccr32374-bib-0005){ref-type="ref"}, [10](#ccr32374-bib-0010){ref-type="ref"}, [12](#ccr32374-bib-0012){ref-type="ref"}
4. CONCLUSION {#ccr32374-sec-0004}
=============
The current study demonstrates the seventh clinical case report about *S pluranimalium* infection isolated from the neonatal septicemia. Based on the results of present study and the previous studies, it seems that the *S pluranimalium* included in vancomycin, carbapenems, aminoglycosides, and 3rd generation cephalosporins performs as an opportunistic pathogen, and also the appropriate therapeutic regimen for the infections caused by this bacterium.
CONFLICT OF INTEREST {#ccr32374-sec-0005}
====================
None to declare.
AUTHOR CONTRIBUTIONS {#ccr32374-sec-0006}
====================
KG: performed clinical diagnosis and drafted the manuscript. MK: drafted the manuscript. HK: performed clinical diagnosis and contributed to the final revision of the manuscript. MK: drafted, revised the manuscript, and reviewed the manuscript.
| {
"pile_set_name": "PubMed Central"
} |
Background {#Sec1}
==========
Sepsis remains a major cause of mortality, and there is recent data to suggest that the incidence of sepsis is increasing \[[@CR1]\]. Treatment options remain limited to early antibiotic therapy, resuscitation and aggressive support of failing organ systems. Multiorgan dysfunction syndrome (MODS), the most severe end of the spectrum of sepsis syndrome, may involve the myocardium \[[@CR2],[@CR3]\], and this has clear prognostic implications, with some reports pointing to a direct correlation between the onset of myocardial dysfunction and mortality in septic shock \[[@CR4]\].
The pathophysiology of septic cardiomyopathy has not been fully elucidated. Although apoptosis is postulated to play an important role in the pathophysiology of septic cardiomyopathy \[[@CR5],[@CR6]\], *post*-*mortem* studies have demonstrated relatively little cell death, even in cases of undoubted severe septic cardiomyopathy \[[@CR7],[@CR8]\]. This has led to the suggestion that activation of pro-survival pathways may occur in association with pro-apoptotic pathway activation and that septic cardiomyopathy, although ultimately dysfunctional for the organism, is a reactive process of metabolic downregulation, serving a cellular preservation function broadly similar to preconditioning \[[@CR6]\].
Preconditioning is the phenomenon whereby prior exposure to a stimulus, often one normally held to be injurious, triggers phenotypic changes that confer resistance to subsequent insults. It was first described by Murry et al. in 1986 \[[@CR9]\], following the discovery that brief episodes of non-lethal myocardial ischaemia, followed by brief periods of reperfusion, provided protection against a subsequent more prolonged period of ischaemia/reperfusion. The concept of endotoxin-induced increased tissue tolerance first emerged in the 1960s \[[@CR10]\]. These early descriptions of 'endotoxin tolerance' in animal models were followed by discoveries that a similar phenomenon occurs in humans \[[@CR11]\]. Sepsis preconditioning of cardiac tissue may also occur in response to brief, non-fatal episodes of sepsis and bears undoubted observational similarities to ischaemic preconditioning, being inhibited, for example, by cycloheximide, a well-known blocker of '*late ischaemic preconditioning*' \[[@CR12]\]. The mechanisms whereby sepsis confers protection to the heart against subsequent ischaemia/reperfusion injury, however, are unknown, and whether this phenomenon most closely approximates myocardial stunning, preconditioning or other unrelated effects remains to be firmly established.
In the current study, we hypothesised that sepsis would protect the myocardium from subsequent ischaemia/reperfusion (IR) injury. We further hypothesised that the mitogen-activated protein kinases (MAPK) and nuclear factor-kappaB (NF-κB) pathways mediated this protective effect.
Methods {#Sec2}
=======
*Ex vivo* isolated heart model {#Sec3}
------------------------------
All animal studies were carried out with the approval of the National University of Ireland Galway Animal Care Research Ethics Committee and under licence issued by the Public Health Division, Department of Health, Ireland. Specific pathogen-free adult male Sprague Dawley rats (Charles River Laboratories, Kent, UK), 250 to 300 g in body weight, were used for caecal ligation and perforation (CLP)-isolated heart experiments. Animals were randomised to control (*n* = 7) or CLP (*n* = 7) groups. The CLP procedure was carried out under general anaesthesia. Briefly, following midline laparotomy, the caecum was identified and ligated, following which needle perforation was carried out, and wound layers were closed. Animals were returned to housing in individually ventilated cages. At 48 h following operation, animals were harvested for isolated heart protocol using the Langendorff suspension. Animals were sacrificed by means of decapitation. The heart was then removed and suspended from a canula inserted into the aortic root facilitating perfusion with prewarmed Krebs Ringer at 37°C, oxygenated by means of a gas mixture containing 5% CO~2~ and 95% O~2~, maintaining pH of solution at 7.4. Episodes of ventricular fibrillation were promptly treated with lignocaine bolus 200 mg, administered via a t-piece connected to the aortic canula such that it passes directly through the coronary arteries. Measurements of systolic and diastolic left ventricular (LV) functional characteristics were made with a saline-filled balloon placed in the left ventricle. This balloon was connected to a pressure transducer facilitating provision of a digital readout as a continuous waveform. Balloon volume was adjusted to achieve LVEDP 4 to 8 mmHg, following which the volume of the balloon was maintained constant, thus providing a pre-load-independent measure of contractility. Parameters measured were systolic and diastolic left ventricular pressure (LVP), developed (sys-dias) LVP, maximal and minimal derivatives of LVP (dp/dt LVP~max~, dp/dt LVP~min~) and heart rate. Global myocardial ischaemia was induced by means of clamping the inflow line for a period of 30 min, followed by 90 min reperfusion. Upon completion of the protocol, the hearts were stained with 1% 2,3,5-triphenyltetrazolium chloride (TTC), a dye taken up only by viable cells. Myocardial infarct size was determined by direct visualisation, dissection and separation of viable and infarcted heart tissue, which were then weighed and expressed as percentage of total heart weight.
*In vitro* hypoxic/ischaemic studies {#Sec4}
------------------------------------
Preparation of primary cultures of neonatal cardiomyocytes was achieved by trypsin/collagenase digestion of ventricles harvested from 3- to 5-day-old neonatal Sprague Dawley rats following decapitation. Cells were cultured in a 1:1 mixture of Dulbecco\'s modified Eagle\'s medium (DMEM) and Ham\'s F12 (D8437 Sigma-Aldrich Ireland Limited, Wicklow, Ireland) containing 17 mM glucose, supplemented with 1 mM sodium pyruvate (Gibco, Life Technologies, Grand Island, NY, USA), 10% newborn foetal calf serum, 5 μg/ml bovine insulin, 5 ng/ml human transferrin, 5 μg/ml sodium selenite, 100 U/ml penicillin and 100 μg/ml streptomycin. All plates and flasks used to culture cardiomyocytes were previously coated with 2% gelatin solution for 24 h and aspirated prior to plating of cells, to ensure adherence of cardiomyocytes. To prevent fibroblast proliferation, 100 mM 5-bromo-2-deoyuridine (BrdU) was added to the cardiomyocyte media for 48 h following plating, at which point media were replaced. Each well was then re-fed with prewarmed media prior to commencing experimentation. This is required to provide sufficient metabolic substrate for viability and normal function throughout the experimental period.
Human alveolar epithelial cells (A549) were purchased from ATCC, Middlesex, UK. HK2 human adult kidney cells were a kind gift from Prof. Michael Ryan (University College Dublin, Ireland). Cells were cultured in DMEM medium supplemented with heat-inactivated 10% foetal bovine serum, 2 mmol/l glutamine, 100 U/ml penicillin and 100 μg/ml streptomycin.
To investigate putative signaling pathways involved in the protective effect of cellular exposure to endotoxin prior to incubation under hypoxic conditions, inhibitors of MAP kinases extracellular-regulated protein kinase (ERK) 1/2, p38 MAPK (p38), and c-Jun NH2-terminal protein kinase (JNK) in addition to pyrrolidinedithiocarbamate (PDTC) (an NF-κB inhibitor) and anti-HMGB1 antibody were added to cell culture at standard concentrations.
For cell viability experiments, A549 and HK2 cells were grown to confluence in 96-well flat-bottom tissue culture plates (Sarstedt, Nümbrecht, Germany) and cultured in pre-equilibrated and prewarmed RPMI, DMEM and a 1:1 mixture of DMEM:Ham\'s F12 media, respectively. The cells were allowed 24 h to polarise prior to experimentation. The medium was then re-fed, supplemented with LPS, LPS with MAPK inhibitors, PDTC, anti-HMGB1 antibody or vehicle under conditions of normoxia (humidified 5% CO~2~, 95% air) or hypoxia (humidified O~2~/N~2~/CO~2~ in the following ratio: 2:93:5), following randomisation.
Viability was measured by cellular 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyl tetrazolium bromide (MTT) assay, under conditions of normoxia (humidified 95% air, 5% CO~2~) and hypoxia (humidified O~2~/N~2~/CO~2~ in the following ratio: 2:93:5), for an experimental duration of 48 h, on all four cell types. The MTT assay provides an estimate of mitochondrial activity from which cell viability is deduced.
Data analysis {#Sec5}
-------------
Results are expressed as mean (SD) for normally distributed data. Data were analysed using *t* test comparison. A *p* value of 0.05 was considered statistically significant.
Results {#Sec6}
=======
Sepsis preconditioning protects the heart {#Sec7}
-----------------------------------------
Seven animals underwent the CLP procedure to provide a model of polymicrobial sepsis. These septic animals were then sacrificed and hearts suspended in Langendorff preparation. The hearts were also isolated from seven non-septic animals which provided the control group in this series of experiments.
### Indices of myocardial function {#Sec8}
Hearts from septic animals had reduced baseline developed LVP ((d)LVP) compared to those from control animals, mean (SE) 101.11 ± 6.06 mmHg vs. 47.15 ± 6.88 mmHg, *p* = 0.0001 (Figure [1](#Fig1){ref-type="fig"}).Figure 1**Sepsis protects the isolated heart against ischaemia/reperfusion injury - (d)LVP.** Line graph depicting developed LVP in isolated hearts, control and animals subjected to the CLP procedure (*n* = 7 per group), demonstrating that sepsis reduces myocardial function with lower (d)LVP at baseline, but protects the heart against ischaemia/reperfusion injury with more rapid functional recovery at reperfusion.
LVP~max~ was reduced in CLP animals compared to controls at baseline, 109.08 ± 4.65 mmHg vs. 57.49 ± 7.22 mmHg, *p* = 0.00003 (Figure [2](#Fig2){ref-type="fig"}).Figure 2**Sepsis protects the isolated heart against ischaemia/reperfusion injury - LVP** ~**max**~ **.** Line graph depicting LVP~max~ in isolated hearts, control and animals subjected to the CLP procedure (*n* = 7 per group), demonstrating that sepsis reduces myocardial function with lower LVP~max~ at baseline, but protects the heart against ischaemia/reperfusion injury with more rapid functional recovery at reperfusion.
During the early reperfusion phase, (d)LVP was greater for CLP hearts than for control, pointing to more rapid functional recovery at reperfusion. This effect was not lasting, however, as at 90-min reperfusion, (d)LVP was similar for CLP and control animals, 58.51 ± 14.59 mmHg vs. 54.27 ± 5.75 mmHg, as was LVP~max~, 67.16 ± 11.05 mmHg vs. 78.41 ± 10.49 mmHg, *p* = 0.77.
Episodes of fibrillation upon reperfusion occurred more frequently in the control group (2/7, 28%) compared to the sepsis group (1/7, 14%). Episodes were promptly terminated by administration of lignocaine.
### Infarct size {#Sec9}
Infarct sizes after ischaemia/reperfusion were significantly reduced in CLP hearts compared to controls (27% ± 8.8 vs. 44% ± 6.2, *p* \< 0.05) (Figure [3](#Fig3){ref-type="fig"}).Figure 3**Sepsis protects the isolated heart against ischaemia/reperfusion injury - infarct size.** Bar chart depicting infarct size in isolated hearts, control and animals subjected to CLP procedure (*n* = 7 per group), demonstrating that sepsis reduces myocardial function with reduced infarct size in the sepsis group. \*Significantly different (*p* \< 0.05, Students *t* test).
Mechanism of action of septic preconditioning {#Sec10}
---------------------------------------------
### Cardiomyocytes {#Sec11}
Endotoxin preconditioning protected against hypoxia-induced cardiomyocyte cell death vs. hypoxic controls \[cell viability 81.55% ± 8.2% vs. 49.74% ± 1.1%, respectively, *p* = 0.000125\] (Figure [4](#Fig4){ref-type="fig"}). Prior blockade of p38, JNK and NF-κB pathways all caused significant reversal of endotoxin preconditioning with viabilities decreased to 43.13% ± 6.6%, *p* = 0.000004, 47.83% ± 20.2%, *p* = 0.01, 38.67% ± 5.6%, *p* = 0.000001, respectively (Figure [4](#Fig4){ref-type="fig"}). Addition of inhibitors of ERK 1/2 and anti-HMGB1 antibody demonstrated no significant effects.Figure 4**Endotoxin exposure protects isolated cardiomyocytes from hypoxia-induced cell death.** Bar chart depicting cell viability as measured by MTT assay of cardiomyocytes expressed as % of normoxia controls. Groups were normoxia and hypoxia controls and cells pretreated with LPS and inhibitors prior to incubation under conditions of hypoxia (*n* = 6 per group). The figure demonstrates the protective effect of endotoxin exposure against hypoxia-induced cell death, an effect reversed by exposure to inhibitors of p38 and JNK MAPK and NF-κB inhibition. \*Significantly different (*p* \< 0.02, Students *t* test).
### Alveolar epithelial cells {#Sec12}
Endotoxin preconditioning was protective against hypoxia-induced cell death, with viability of 69.7% ± 7.62% vs. hypoxic controls 49.01% ± 7.95%, *p* = 0.003 (Figure [5](#Fig5){ref-type="fig"}).Figure 5**Endotoxin exposure protects pulmonary alveolar cells from hypoxia-induced cell death.** Bar chart depicting cell viability as measured by MTT assay of A549 cells expressed as % of normoxia controls. Groups were normoxia and hypoxia controls and cells pretreated with LPS prior to incubation under conditions of hypoxia (*n* = 5 per group). The figure demonstrates the protective effect of endotoxin exposure against hypoxia-induced cell death. \*Significantly different (*p* \< 0.01, Students *t* test).
### Kidney cells {#Sec13}
Endotoxin preconditioning was protective against hypoxia-induced cell death, with viability of 85.51% ± 3.27% vs. 73.88% ± 4.04%, *p* = 0.0004 (Figure [6](#Fig6){ref-type="fig"}). Addition of anti-HMGB1 antibody and ERK 1/2 inhibitors to LPS-treated cells did not reverse these protective effects, with viabilities of 97.3% ± 6.24% and 96.15% ± 11.7%, respectively, *p* = 0.002 and *p* = 0.007 (Figure [6](#Fig6){ref-type="fig"}).Figure 6**Endotoxin exposure protects human kidney cells from hypoxia-induced cell death.** Bar chart depicting cell viability as measured by MTT assay of HK2 cells expressed as % of normoxia controls. Groups were normoxia and hypoxia controls and cells pretreated with LPS prior to incubation under conditions of hypoxia (*n* = 6 per group). The figure demonstrates the protective effect of endotoxin exposure against hypoxia-induced cell death. \*Significantly different (*p* \< 0.02, Students *t* test).
Discussion {#Sec14}
==========
Previous investigators have reported that exposure of hearts to sublethal dosing of endotoxin confers protective effects against myocardial ischaemia/reperfusion injury in a variety of experimental models \[[@CR13]-[@CR15]\], including the Langendorff mode \[[@CR16]-[@CR19]\]. However, data in the literature examining the effects of polymicrobial sepsis, such as that induced by caecal ligation and operation, is less frequently encountered. It should be noted that inherent difficulties exist with experimental models utilising exposure of animal species to endotoxin. Although a reproducible experimental model that is easy to establish, endotoxaemia is not a perfect model of sepsis, with differing cytokine kinetic responses, duration of sepsis and time to death \[[@CR20]\]. Making a conceptual advance on LPS exposure as a model of sepsis, we utilised the CLP procedure to induce polymicrobial sepsis. CLP is an *in vivo* model of sepsis that has been shown to be valid and reproducible \[[@CR21]\] and is a more accurate replication of systemic inflammatory response to an infectious stimulus \[[@CR22]\], more loosely mimicking an important clinically encountered scenario and is considered by some to be an essential preclinical model for sepsis studies \[[@CR23]\].
The results of our experiments demonstrate that septic animals sustained marked sepsis-induced myocardial dysfunction, as evidenced by significantly reduced baseline (d)LVP and LVP~max~ compared to control animals, *p* = 0.0001 and *p* = 0.00003, respectively (Figures [1](#Fig1){ref-type="fig"} and [2](#Fig2){ref-type="fig"}). Following ischaemia and reperfusion, CLP hearts showed more rapid recovery of functional measures, despite significantly reduced baseline function, and infarct size was significantly reduced in septic hearts (Figure [3](#Fig3){ref-type="fig"}). These results strongly support a preconditioning effect of polymicrobial sepsis.
Episodes of fibrillation were seen more frequently in the control group compared to the sepsis group (28% vs. 14%). This likely reflects the severity of IR injury, with protection conveyed in the CLP group conveyed by sepsis preconditioning. The use of lignocaine to prevent or treat reperfusion arrhythmias in the Langendorff model of the isolated heart has been well described \[[@CR16]\]. Episodes of fibrillation were promptly treated and therefore unlikely to affect functional outcomes or infarct size.
Studies have shown sepsis preconditioning to demonstrate putative similarities to ischaemic preconditioning, particularly so-called '*late preconditioning*' \[[@CR12],[@CR24]\]. For example, with induction of gram-negative sepsis, myocardial protective effects are demonstrable at 6 h \[[@CR18],[@CR25]\], and the myocardial protective effects of LPS exposure were reported to appear after 12 to 24 h \[[@CR12]\]. As mentioned previously, the administration of the drug cycloheximide was found to inhibit protection, an effect demonstrable with late ischaemic preconditioning \[[@CR26]\]. Additionally, there is strong evidence from multiple sources that the protective effects of LPS utilise signaling pathways known to be of central importance in ischaemic preconditioning, namely NOS2 and Akt \[[@CR24]\]. Despite these similarities and intense research in the area of ischaemic preconditioning, the signaling pathway through which sepsis protects the heart from IR injury remains to be fully elucidated.
To investigate the *in vitro* effects of endotoxin exposure, we examined cell viability as measured by the MTT assay. The MTT assay provides an estimate of mitochondrial activity from which cell viability is deduced. In the presence of functioning mitochondria, MTT is reduced from a colourless compound to a blue product (MTT-formazan). The assay involves spectrophotometrical measurement of the concentration of the blue product. This is then calibrated *post hoc* to provide an indirect measure of mitochondrial function.
Lipopolysaccharide exerts its effects through TLR4-dependent NF-κB activation, leading ultimately to pro-inflammatory cytokine release. LPS is known to induce both pro-apoptotic and anti-apoptotic pathways, inducing apoptosis in endothelial cells \[[@CR26]\] and hepatocytes \[[@CR27]\] while promoting survival in monocytes \[[@CR28]\], neutrophils \[[@CR29]\], macrophages \[[@CR30]\] and cardiomyocytes \[[@CR4],[@CR12]\]. Although there is evidence to support a protective role played by LPS against myocardial ischaemia and reperfusion injury \[[@CR13],[@CR16],[@CR19]\], less is known about the effects of LPS on apoptosis of cardiomyocytes. The current study in which LPS pretreatment of cardiomyocytes conferred protection against hypoxia-induced cell death (Figure [4](#Fig4){ref-type="fig"}) supports the findings of Chao et al. \[[@CR4]\] who found reduced apoptosis in cardiomyocytes treated with LPS and subjected to hypoxia and serum deprivation, supporting a preconditioning effect of LPS in *in vitro* models of hypoxia. Limited studies exist outlining the effect of endotoxin preconditioning in other organ systems, and our findings in the current study speak to the fact that these findings can be generalised to other tissues and are not specific to cardiomyocytes through our demonstration of endotoxin preconditioning in alveolar epithelial (Figure [5](#Fig5){ref-type="fig"}) and human kidney (Figure [6](#Fig6){ref-type="fig"}) cells.
Due to the known similarities between ischaemic and sepsis preconditioning, and the important roles that stress kinases and NF-κB activation are known to play in the signaling of ischaemic preconditioning \[[@CR31],[@CR32]\], we utilised blockers of the stress kinases, and ammonium PDTC, a direct inhibitor of the NF-κB pathway, to further examine potential signaling pathways involved in sepsis preconditioning.
The MAPK are a family of kinases that have been shown to be activated following multiple extracellular stimuli \[[@CR33],[@CR34]\], with effects controlling cellular growth and differentiation \[[@CR35]-[@CR37]\]. Three MAPK have been identified in the cardiomyocyte: ERK, p38 and JNK. These are known to play key roles in cellular signal transmission between the cell surface and nuclei \[[@CR38]\]. Although the signaling pathways through which sepsis confers myocardial protective effects have as yet to be elucidated, an important role has been proposed for the MAPK JNK, with additional reports of increased TNF-α-induced apoptosis associated with JNK inhibition \[[@CR39]\]. This is supported by the findings of the current study, which demonstrates that the addition of JNKi to cardiomyocytes incubated under hypoxic conditions attenuated the protective effects against cell death associated with LPS (Figure [4](#Fig4){ref-type="fig"}). p38 MAPK is undoubtedly activated by LPS, although studies have revealed the conflicting effects of attenuated \[[@CR39]\] and enhanced \[[@CR40]\] apoptosis consequent to this. The results of the current study support a role of increased cell survival, as evidenced by attenuation of LPS-induced protective effects against hypoxia in cardiomyocytes (Figure [4](#Fig4){ref-type="fig"}), a finding that warrants further investigation. Although a role has been suggested by ERK 1/2 MAPK in the sepsis preconditioning signaling pathway \[[@CR4]\], incubation of cardiomyocytes with inhibitors of ERK 1/2 had no demonstrable effect on endotoxin protection.
Studies have demonstrated an important role of NF-κB activation in delayed ischaemic preconditioning *in vivo* \[[@CR41]\], and there is some evidence of a role played in signaling pathways of sepsis preconditioning \[[@CR42]\]. In the current study, inhibition of NF-κB through the use of cellular incubation with PDTC attenuated endotoxin preconditioning, providing support to our hypothesis that this mediator may play an important role in sepsis preconditioning.
HMGB1, a so-called endogenous danger signaling molecule, is actively released by activated monocytes and macrophages and is a key late mediator of the inflammatory response to sepsis \[[@CR43]-[@CR45]\], with levels corresponding to the onset of mortality \[[@CR43],[@CR46]\]. There is evidence that HMGB1 may also have preconditioning effects \[[@CR47]\], and for this reason, we utilised anti-HMGB1 antibodies to ascertain whether HMGB1 may play a role in the signaling pathway of endotoxin preconditioning. Our results determined no demonstrable effects with HMGB1 inhibition.
Conclusions {#Sec15}
===========
These studies provide further confirmation of a preconditioning effect of sepsis in the intact heart subjected to global ischaemia/reperfusion. In addition, we have provided evidence of endotoxin preconditioning in cultured cardiomycocytes and further expanded on this to other organ systems through demonstration of effects in pulmonary alveolar cells and human kidney cells. Our investigation of the signaling pathway involved in these effects were reversed by NF-κB, p38 and JNK inhibition, highlighting potential important roles for these mediators in sepsis preconditioning. Taken together, these findings suggest a phenomenon with a high degree of relative organ specificity and requiring a complex signaling pathway.
**Competing interests**
The authors declare that they have no competing interests.
**Authors\' contributions**
CW carried out the laboratory experiments as outlined and drafted the manuscript. JL was instrumental in the initiation of the project and was involved in the drafting and preparation of the manuscript for publication. LK was responsible for the inception of the initial hypotheses, in addition to supervision of laboratory work and manuscript preparation. DOT was responsible for the inception of experiment design, in addition to supervision of laboratory work and manuscript preparation. All authors read and approved the final manuscript.
| {
"pile_set_name": "PubMed Central"
} |
Introduction
============
Chronic hypoxia is an important factor contributing to the near-irreversibility of hypoxia-induced pulmonary hypertension (HPH), a condition characterized by structural remodeling of small pulmonary arteries (PAs). These important morphological changes are characterized by an abnormal increase in numbers of smooth muscle (SM)-like cells \[cells expressing α-SM actin (α-SMA)\] in the PA wall ([@b1-ijmm-42-01-0270],[@b2-ijmm-42-01-0270]), which results in thickening of muscular and elastic vessels, muscularization of distal vessels, and eventual development of pulmonary hypertension. The small pulmonary arteries and veins may be distinguished based on the relative abundance of SM cells or SM-like cells in vessel walls. However, the origin of SM-like cells has remained elusive.
Epithelial-mesenchymal transition (EMT) is a process wherein epithelial cells undergo phenotypic changes and develop into mesenchymal/SM-like cells ([@b3-ijmm-42-01-0270]). Similarly, endothelial cells may acquire a mesenchymal or SM-like phenotype, a process referred to as endothelial-mesenchymal transition (EndMT). Evidence of EndMT has been reported in the context of cardiac and vascular development, wound healing and various diseases, including fibrosis, diabetic nephropathy, heterotopic ossification and cancer ([@b4-ijmm-42-01-0270]--[@b10-ijmm-42-01-0270]). Transdifferentiated cells co-express the endothelial marker CD31 and the SM-like cell type marker α-SMA. Thus, EndMT is regarded as another important mechanism for the generation of SM-like cells. The endothelial cell appears to be one of the targets in hypoxia, and endothelial cell dysfunction has a direct and indirect role in the process of pulmonary vascular remodelling ([@b11-ijmm-42-01-0270]). Ranchoux *et al* ([@b12-ijmm-42-01-0270]) reported that EndMT may be a source of α-SMA-expressing cells. The EndMT participates in vascular remodeling as a characteristic of pulmonary hypertension; however, the underlying mechanism has remained to be fully elucidated.
Hypoxia-inducible transcription factor-1α (HIF-1α) is a critical regulator of the cellular response to hypoxia. HIF-1α activity in the endothelium of PAs has been observed to significantly increase in pulmonary hypertension ([@b13-ijmm-42-01-0270]). The HIF-1α-regulated gene Twist ([@b14-ijmm-42-01-0270],[@b15-ijmm-42-01-0270]) has an important role in EMT, cell movement and proliferation ([@b16-ijmm-42-01-0270]--[@b18-ijmm-42-01-0270]). However, the role of HIF-1α/Twist in EndMT has remained to be fully characterized. In the present study, a significant upregulation of HIF-1α in hypoxic pulmonary microvascular endothelial cells (PMVECs) was demonstrated, and knockdown of HIF-1α and Twist1 effectively blocked hypoxia-induced EndMT. It appeared that HIF-1α regulated the EndMT through binding to the promoter of the Twist1 gene and subsequently activating Twist1 transcription and expression.
Materials and methods
=====================
Animal model
------------
A total of 36 male Sprague Dawley rats (age, 6--8 weeks; weight, 200--250 g) were obtained from the Fourth Military Medical University. All experimental procedures were approved by the Animal Use and Care Committee for Research and Education of The Fourth Military Medical University (Xi\'an, China). Control rats were housed under a 12-h light/dark cycle at room temperature (humidity: 50--60%) and were provided water and a standard laboratory diet *ad libitum*. Rats were housed for 28 days in a chamber containing 10% oxygen for exposure to continuous hypobaric hypoxia. A gradual decrease in oxygen concentration was employed in order to acclimatize the rats to the hypoxic conditions ([@b19-ijmm-42-01-0270]). Over 30 min, the pressure was slowly increased at the beginning of hypoxia by 0.5 atm, and the oxygen concentration was reduced to 10%. The gradual decrease in the oxygen concentration had no observable negative effect on the rats. After 4 weeks, the HPH models were successfully replicated. Rats in the normoxic group (control group) were housed continuously in room air (n=6 per group). At the end of hypoxia exposure, the right ventricular systolic pressure (RVSP) and the ratio of ventricular weight \[right ventricle/(left ventricle+septum), denoted as RV/(LV+S)\] were measured. Increases in RVSP and RV/\[LV+S\] were considered as being indicative of HPH.
Cell culture and chemical reagents
----------------------------------
The lung was isolated from healthy adult rats under sterile conditions. After washing repeatedly with sterile D-Hank\'s buffer, tissues at the edge and surface of the lung (thickness, \~1 mm) were removed, dissected into small pieces and cultured in M200 medium (Gibco; Thermo Fisher Scientific, Inc., Waltham, MA, USA), supplemented with a growth factor cocktail \[fibroblast growth factor, heparin, hydrocortisone and epidermal growth factor (Gibco; Thermo Fisher Scientific, Inc.)\] and 20% fetal bovine serum FBS (HyClone; GE Healthcare, Little Chalfont, UK). Primary cultures used were confirmed to be uncontaminated with SM cells by immunostaining for α-SMA and CD31 or VIII factor. In the normoxic group, PMVECs were cultured at 37°C in a humidified atmosphere of 5% CO~2~ in air. In the hypoxic groups, PMVECs were incubated in a hypoxic chamber containing 1% O~2~, 5% CO~2~ and 94% N~2~ for 7 days, and SM-like cells were then obtained. In another experiment, PMVECs were also stimulated with transforming growth factor (TGF)-β (10 ng/ml) in normoxic conditions for 7 days. Antibodies against α-SMA (MA1-744) were obtained from Thermo Fischer Scientific, Inc. (Waltham, MA, USA). Monoclonal antibodies against CD31 (ab64543) and Twist1 (ab50581) were purchased from Abcam (Cambridge, MA, USA). collagen (Col) 1A1 (sc-293182) and Col3A1 (sc-271249) were purchased from Santa Cruz Biotechnology, Inc. (Dallas, TX, USA). Anti-HIF-1α monoclonal antibody (MAB5382) was purchased from EMD Millipore (Billerica, MA, USA). Secondary antibodies (anti-rabbit or anti-mouse IgG antibody conjugated to horseradish peroxidase; 7074 and 7076; 1:2,000) were purchased from Cell Signaling Technology (Danvers, MA, USA). Recombinant human TGF-β was purchased from R&D Systems (Minneapolis, MN, USA).
Immunofluorescence
------------------
Frozen tissue sections or cultured cells grown on coverslips were washed with PBS and fixed in 1% paraformaldehyde. Cells were subsequently permeabilized with 0.2% Triton X-100 in PBS at room temperature for 5 min. The cells and tissue sections were blocked by 2% goat serum (OriGene Technologies, Beijing, China) at 37°C for 1 h. Then they were incubated with primary antibodies (CD31, 1:500; α-SMA, 1:1,000) overnight at 4°C. Subsequently, samples were incubated with secondary antibody (Alexa Fluor^®^ 594 donkey anti-mouse IgG; R37115; 1:500; Molecular Probes; Thermo Fischer Scientific, Inc.) for 2 h at room temperature. Fluorescence was examined by confocal laser scanning microscopy.
Immunohistochemistry
--------------------
Sagittal sections of the right lung were placed in 4% paraformaldehyde and processed for paraffin embedding. Sections (5 *μ*m) were prepared and mounted on glass slides prior to overnight incubation at 4°C with anti-Twist1 antibody (1:500). Slides were washed and incubated with corresponding secondary antibodies conjugated with alkaline phosphatase (Expose Rabbit specific HRP/DAB detection IHC kit; ab80437; Abcam). Sections were evaluated using an Olympus BX50 optical microscope (Olympus Corporation, Tokyo, Japan) equipped with an image analysis program (Image Pro Plus version 6.0; Media Cybernetics, Inc., Rockville, MD, USA).
Electron microscopy
-------------------
Transmission electron microscopy (TEM) analysis was performed using glutaraldehyde-fixed PMVECs embedded in resin (Epon 812 epoxy resin). The clean slide was placed in 0.25% Formvar solution and then placed vertically on filter paper to air dry. Then the slide was exposed vertically to the surface of the water. The supporting membrane on the surface of the glass slide was removed and the 75-mesh copper grids placed on the membrane, removed from water and allowed to dry. Thin sections (50--80 nm) were collected on the formvar-coated copper grids and stained with 3% uranyl acetate and lead citrate solutions for 20--30 min respectively at room temperature. Following air drying at room temperature, the sections were examined under an electron microscope (JEM1200EX II; JEOL, Tokyo, Japan); images were captured.
Reverse transcription-quantitative polymerase chain reaction (RT-qPCR)
----------------------------------------------------------------------
Total RNA was extracted from lung tissues and PMVECs with TRIzol (Invitrogen; Thermo Fisher Scientific, Inc.) according to the manufacturer\'s instructions. Briefly, template RNA (1 *μ*g/*μ*l) was added to the individual tubes containing the master mix, using the Omniscript RT kit (Qiagen, Inc.). These were centrifuged briefly to collect residual liquid from the walls of the tubes, which were then incubated for 2 h at 37°C. Then 2 *μ*l RNase A (10 mg/ml) was added and the reaction mix incubated for 10 min at 65°C, 5 min at 93°C and then cooled immediately on ice. RT-qPCR was performed with a MyiQ (Bio-Rad Laboratories, Inc., Hercules, CA, USA) using SYBR-Green I dye (Applied Biosystems; Thermo Fisher Scientific, Inc.). Primer pairs used in the present study are listed in [Table I](#tI-ijmm-42-01-0270){ref-type="table"}. Each PCR was performed in a 50 *μ*l mixture containing 1 *μ*l cDNA, 5 *μ*l 10X Qiagen PCR buffer (Qiagen, Inc.), 10 *μ*l 5X Q-Solution (Qiagen, Inc.), 1 *μ*l each of deoxynucleotide triphosphate mix (10 mM), 0.1 *μ*M of each sense and antisense primer and 0.5 *μ*l Taq DNA polymerase (Qiagen, Inc.), by using the Taq DNA Polymerase kit (Qiagen, Inc.). The amplification reaction consisted of initial denaturation at 95°C for 2 min, followed by three-step cycling that consisted of denaturation at 95°C for 10 sec, annealing at 60°C for 30 sec and extension at 70°C for 30 sec for 30 cycles. The comparative threshold cycle (Cq) method was employed for quantification of transcripts ([@b20-ijmm-42-01-0270]).
Western blot assay
------------------
Lung tissues and cultured PMVECs were placed in lysis buffer (50 mM NaCl, 50 mM NaF, 50 mM sodium pyrophosphate, 5 mM EDTA, 5 mM EGTA, 2 mM Na~3~VO~4~, 0.5 mM phenylmethylsulfonyl fluoride and 10 mM 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid at pH 7.4, together with 0.1% Triton X-100, 10 mg/l leupeptin and 5% aprotinin) respectively. Protein concentrations were determined with the bicinchoninic acid protein assay. Equivalent amounts of protein (50 *μ*g) were subjected to 10% SDS-PAGE and transferred onto a nitrocellulose membrane (Bio-Rad Laboratories). Membranes were blocked overnight at 4°C with PBS containing 5% non-fat dry milk and 0.1% Tween-20. Following blocking, membranes were incubated overnight at 4°C with primary antibodies, including anti-HIF-1α (1:500), anti-CD31 (1:500), anti-Twist1 (1:500), anti-α-SMA (1:1,000), anti-vimentin (ab92547; 1:200) and anti-β-actin (A2228; 1:4,000; Sigma-Aldrich; Merck KGaA). Proteins were incubated with secondary antibodies (anti-rabbit or anti-mouse IgG antibody; Cat. 7074 and Cat. 7076; 1:2,000; Cell Signaling Technology) for 1 h at 37°C in PBS containing 1% non-fat dry milk and 0.1% Tween-20. The chemiluminescence signal was detected using enhanced chemiluminescent substrate (SuperSignal™ West Femto Trial Kit, Thermo Fisher Scientific, Inc.).
Cell transfection
-----------------
Primary PMVECs were transduced with lentiviral vector containing small interfering (si)RNA targeting either rat HIF-1α or Twist1 (Genechemgene, Shanghai, China). For transduction, cells were seeded in 6-well plates at a density of 5×10^5^ cells per well. Virus particles were added at a multiplicity of infection of 40, followed by incubation for 36 h. Cells were then washed and incubated at 37°C for 60 h prior to PCR and western blot analysis. The sequences of Twist1 siRNA and HIF-1α siRNA were 5′-GGCGGCCAGGTACATCGACTT-3′ and 5′-AACCAGTTGAATCTTCAGATA-3′, respectively. The sequences of negative control was TTCTCCGAACGTGTCACGT.
Cell proliferation
------------------
PMVECs or hypoxia-induced SM-like cells were seeded on to 96-well plates and incubated for 24 and 48 h under normoxic or hypoxic conditions. Subsequently, 10 *μ*l MTT solution (5 mg/ml) was added into each well and following 4 h of incubation the supernatant was gently removed by a pipette and discarded, and 150 *μ*l dimethyl sulfoxide added. The optical density values were detected at a wavelength of 490 nm using a spectrophotometer (Bio-Tek Power Wave XS, Bio-Tek, Winooski, VT, USA).
Wound healing assay
-------------------
PMVECs and SM-like cells were grown to confluence. A linear wound was made by scraping the cell layer with a 0.5 ml plastic Pasteur pipette. Following two washes with culture medium to remove the detached cells and debris, the cells were exposed to normoxic or hypoxic conditions for 24 and 48 h, and the size of the wounds was measured.
Chromatin immunoprecipitation (ChIP)
------------------------------------
The ChIP assay was performed using the EZ-Zyme™ Chromatin Prep Kit (EMD Millipore) according to the manufacturer\'s instructions. In brief, cell lysates were incubated with 1 *μ*g RNA polymerase II, IgG (1 *μ*g/ml; Wanleibio Co., Ltd., Shenyang, China) or anti-HIF-1α antibody (1:500). The immunoprecipitated DNA was amplified using Twist1 promoter-specific primers ([@b21-ijmm-42-01-0270]). PCR generated a 196-bp product from the regulatory region of the Twist1 promoter containing CCACGTGG. Primers for Twist1 were as follows: Forward 5′-TCGGATGAAAGCACAGTCG-3′ and reverse 5′-GCCACCGACTTCCTGAGA-3′.
Statistical analysis
--------------------
All values are expressed as the mean ± standard error of the mean. Statistical analysis was performed using analysis of variance and multiple comparisons were made by Tukey\'s method. Statistical analysis was processed by SPSS version 16.0 (SPSS, Inc., Chicago, IL, USA). Differences were considered to be statistically significant at P\<0.05.
Results
=======
Hypoxia induces transdifferentiation of endothelial cells into SM-like cells in small, but not in large Pas
-----------------------------------------------------------------------------------------------------------
A rat model of chronic HPH was used to examine the effect of hypoxia on EndMT *in vivo*. In large proximal PAs, hypoxia did not appear to alter the expression of endothelial marker CD31 and mesenchymal marker α-SMA ([Fig. 1A](#f1-ijmm-42-01-0270){ref-type="fig"}). By contrast, a considerable downregulation of CD31 along with significant upregulation of α-SMA was observed in small distal PAs ([Fig. 1B](#f1-ijmm-42-01-0270){ref-type="fig"}). In addition, hypoxia led to PA remodeling, as evidenced by the appreciable intimal thickening ([Fig. 1B](#f1-ijmm-42-01-0270){ref-type="fig"}). Of note, CD31 was co-localized with α-SMA ([Fig. 1B](#f1-ijmm-42-01-0270){ref-type="fig"}), which suggested trans-differentiation of endothelial cells into α-SMA-expressing mesenchymal-like cells (SM-like cells). These results indicate that HPH is associated with remodeling of small PAs and that hypoxia may induce EndMT *in vivo*.
Hypoxia induces transdifferentiation of endothelial cells into SM-like cells in vitro
-------------------------------------------------------------------------------------
Next, the present study sought to determine the role of hypoxia in EndMT *in vitro*. As presented in [Fig. 2](#f2-ijmm-42-01-0270){ref-type="fig"}, Weibel-Palade bodies, as well as abundant pinocytotic vesicles near the plasma membrane, were observed in the cytoplasm of normal PMVECs, while α-SMA stress fibers were undetectable, which is a typical phenotype of endothelial cells. However, following hypoxic exposure, mixed cell populations were observed in PMVECs. Certain cells exhibited well-developed endoplasmic reticulum whose cisternae were expanded, and increased protein secretion and abundant VEC-specific pinocytotic vesicles were present in the cytoplasm ([Fig. 2B](#f2-ijmm-42-01-0270){ref-type="fig"}), whereas other cells exhibited SM-like features, including aggregation of α-SMA filaments in the cytoplasm and presence of dense bodies abutting the cell membrane ([Fig. 2C](#f2-ijmm-42-01-0270){ref-type="fig"}). Pinocytotic vesicles were not observed in these cells ([Fig. 2C](#f2-ijmm-42-01-0270){ref-type="fig"}). These results suggest that hypoxic PMVECs undergo a transition from endothelial cells to SM-like cells.
To further confirm the role of hypoxia in EndMT *in vitro*, endothelial and mesenchymal markers were examined in hypoxic PMVECs. The results indicated that 7 days of hypoxia exposure attenuated the protein expression of CD31, but markedly enhanced the expression of α-SMA and the mesenchymal markers Col1A1 and Col3A1 ([Fig. 3A](#f3-ijmm-42-01-0270){ref-type="fig"}). This effect was comparable to that of TGF-β, a known potent inducer of EndMT. The mRNA expression of these markers exhibited a similar trend ([Fig. 3B](#f3-ijmm-42-01-0270){ref-type="fig"}). These results suggest that the effect of hypoxia in inducing EndMT was akin to that of TGF-β.
Increased cell proliferation and migration are important features associated with EndMT. Therefore, the degree of proliferation and migration of PMVECs and SM-like cells after EndMT under hypoxic conditions was detected. As presented in [Fig. 4A and B](#f4-ijmm-42-01-0270){ref-type="fig"}, PMVECs did not exhibit any proliferation and migration potential under hypoxia over 48 h. Conversely, SM-like cells exhibited time-dependent proliferation and migration under hypoxia. As the SM-like cells were derived from the PMVECs incubated in hypoxia over 7 days, these results demonstrate that hypoxia induces EndMT, resulting in a phenotype with increased cell proliferation and migration.
HIF-1α is essential for hypoxia-induced EndMT
---------------------------------------------
HIF-1α is known to be a master regulator of the hypoxic response. Therefore, the present study examined the role of HIF-1α in hypoxia-induced EndMT. The present results indicated that hypoxia induced the expression of HIF-1α ([Fig. 5A](#f5-ijmm-42-01-0270){ref-type="fig"}). Upregulation of Twist1 expression was also detected in the small PA specimens from HPH rats ([Fig. 6](#f6-ijmm-42-01-0270){ref-type="fig"}). Of note, knockdown of HIF-1α effectively inhibited 7 days of hypoxia exposure-induced expression of Twist1, α-SMA and Col1A1, while reversing hypoxia-induced suppression of CD31 expression ([Fig. 5](#f5-ijmm-42-01-0270){ref-type="fig"}). These results indicate that HIF-1α has an essential role in hypoxia-induced EndMT and that it may act as an upstream regulator of Twist1.
Twist1 is important in hypoxia-induced EndMT
--------------------------------------------
Since hypoxia also upregulates Twist1, the present study next sought to determine the function of Twist1 in hypoxia-induced EndMT. Similar to HIF-1α, knockdown of Twist1 also altered the expression of CD31, α-SMA and Col1A1 in hypoxic PMVECs ([Fig. 7](#f7-ijmm-42-01-0270){ref-type="fig"}). The knockdown efficiency of the siRNAs was 80--85%. It appeared that knockdown of HIF-1α exerted a more obvious effect on hypoxia-induced PMVEC transdifferentiation to mesenchymal-like cells than that of Twist1. This was evidenced in the obviously greater inhibition of hypoxia-induced α-SMA expression by knockdown of HIF-1α, as compared with the effect of Twist1 knockdown ([Fig. 8](#f8-ijmm-42-01-0270){ref-type="fig"}). This observation is consistent with the hypothesis that HIF-1α is an upstream regulator of Twist1. These results suggest that Twist1 may be a mediator of hypoxia/HIF-1α-induced EndMT *in vitro*.
HIF-1α binds with the promoter of the Twist1 gene
-------------------------------------------------
The present study then investigated the mechanism by which HIF-1α regulates Twist1.An 8-bp HIF-1α-binding sequence (CCA CGT GG) located between 321 and 328 bp upstream of the first exon of the Twist1 transcript was identified using JASPAR database (<http://jaspar2016.genereg.net/>). A ChIP assay was performed to determine whether HIF-1α indeed binds to this site. As presented in [Fig. 9](#f9-ijmm-42-01-0270){ref-type="fig"}, HIF-1α physically interacted with the binding sequence of the Twist1 promoter in PMVECs, which suggests that HIF-1α regulates Twist1 expression by binding to the Twist1 promoter, and that the HIF-1α-Twist1 axis is an important pathway in hypoxia-induced EndMT.
Discussion
==========
The present study indicated that endothelial cells in the pulmonary vasculature differentiate towards an SM phenotype via EndM transdifferentiation upon exposure to hypoxia. The endothelium may be a potential source of SM cells in the vascular system, and the HIF-1α-Twist1 intracellular signaling pathway is involved in this process.
Increased expression of α-SMA was observed in the walls of small PAs in response to chronic exposure to hypoxia. These SM-like cells not only lead to abnormal vasoconstriction, but also have the ability to proliferate, migrate and produce extracellular matrix, which contributes to structural remodeling and results in pulmonary hypertension. These SM-like cells are thought to be derived from the proliferation of resident vascular SM cells, an SM-like subpopulation residing within the media ([@b22-ijmm-42-01-0270]) or from adventitial fibroblasts ([@b23-ijmm-42-01-0270]). Several studies have indicated that circulating progenitor cells and resident vascular progenitor cells may acquire an SM-like phenotype at the site of vascular injury ([@b19-ijmm-42-01-0270],[@b24-ijmm-42-01-0270]). However, endothelial cells were observed to retain the capacity to transition into mesenchymal or SM-like cells *in vitro*. With TGF-β treatment, isolated endothelial cells from the bovine aorta were previously reported to convert to spindle-shaped α-SMA-expressing cells ([@b25-ijmm-42-01-0270]). Zhu *et al* ([@b26-ijmm-42-01-0270]) reported that hypoxia induced transdifferentiation of endothelial cells into SM-like cells and that this process was regulated by myocardin. Similarly, Ranchoux *et al* ([@b12-ijmm-42-01-0270]) reported that EndMT may be a source of those α-SMA-expressing cells. EndMT participates in vascular remodeling associated with pulmonary hypertension. In the present study, cells with expression of α-SMA were not observed in the intimal layer of large proximal PA specimens from HPH rats, although certain α-SMA-expressing cells were observed in distal PA specimens. Pulmonary hypertension is characterized by cellular changes in the walls of PAs. These changes include the appearance of SM-like cells in previously non-muscularized vessels, and medial and adventitial thickening of the muscular and elastic vessels. In addition, intimal changes have been consistently observed in rat models of HPH; however, these intimal changes are typically minimal, at least from a morphological point of view ([@b27-ijmm-42-01-0270]). The magnitude of the changes depends on the species studied for exposure to hypoxia. Since it is hard to morphologically distinguish between small pulmonary arteries and veins, small circular vessels that had a greater expression of SM cells than pulmonary arteries were selected for analysis in the present study. *In vitro*, it was observed that hypoxia induced transdifferentiation of PMVECs into SM-like cells. Therefore, endothelium appears to be a source of SM-like cells.
Next, the present study explored whether HIF-1α may modulate the transdifferentiation of endothelial cells. HIF-1α is a transcriptional regulator that has an important role in the cellular response to hypoxia. In addition, animals deficient in HIF-1α have attenuated HPH ([@b28-ijmm-42-01-0270]). HIF-1α transcriptionally controls a diverse number of genes, including those involved in vascular remodeling. Higgins *et al* ([@b29-ijmm-42-01-0270]) reported that hypoxia promotes fibrogenesis through HIF-1α-mediated stimulation of EMT. As an HIF-1α-regulated gene, Twist, a basic helix-loop-helix transcription factor, was identified as a crucial factor in the EMT ([@b30-ijmm-42-01-0270],[@b31-ijmm-42-01-0270]). Ranchoux *et al* ([@b12-ijmm-42-01-0270]) reported that EndMT involves neoexpression of Twist1. Sun *et al* ([@b32-ijmm-42-01-0270]) reported that Twist promoters contain HIF-1α-binding sites, that HIF-1α induces Twist expression in hypoxic tubular cells, and that this has a role in the EMT during renal fibrogenesis. In the present study, when siRNA was used to block the expression of HIF-1α, hypoxia-induced α-SMA expression was markedly suppressed in PMVECs. Knockdown of HIF-1α also resulted in a marked inhibition of Twist1 expression induced by hypoxia. When Twist1 was knocked down by siRNA, the expression of CD31, α-SMA and Col1A1 was also altered in hypoxic PMVECs, and hypoxia-induced EndMT was effectively inhibited. This indicates that Twist1 also mediates hypoxia-induced EndMT in PMVECs, and that HIF-1α is an upstream regulator of Twist1. The process of EndMT is therefore likely to be in part regulated by HIF-1α-Twist1. In addition, the Twist1 gene promoter was identified to contain HIF-1α-binding sites, and that HIF-1α induces Twist1 expression in PMVECs, further confirming this interaction.
Although HPH is an important subtype of pulmonary hypertension and has the features of pulmonary hypertension, hypoxia is not defined as the major etiological factor. Thus, the HPH model used in the present study may not entirely replicate the actual pathogenesis of pulmonary hypertension. To better understand the precise molecular mechanisms involved in the pathogenesis of pulmonary hypertension, other factors, including inflammation, should also be taken into consideration.
In conclusion, the present study provides evidence of the EndMT process taking place in pulmonary vascular remodeling under chronic hypoxic conditions. The HIF-1α-Twist1 pathway was also suggested to be involved in EndMT. These observations may help explain the appearance of SM-like cells in the intima of PAs in HPH. Identification of EndMT in PAs and elucidation of associated signaling pathways will confer novel insight into the pathogenesis of pulmonary hypertension.
Not applicable.
Funding
=======
This study was supported by the National Natural Science Foundation of China (nos. 81571839, 81000017 and 81471816).
Availability of data and materials
==================================
The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request.
Authors\' contributions
=======================
BZ was mainly responsible for the design of the experiment, establishing animal model, statistical analysis and the writing of the article. WN was responsible for immunofluorescence, immunohistochemistry and electron microscopy assay. H-YD was responsible for RT-qPCR and western blot assay. M-LL was responsible for cell transfection, proliferation and wound healing assay. YL was responsible for ChIP assay. Z-CL was responsible for the design of the experiment.
Ethical approval and consent to participate
===========================================
All experimental procedures on animals were approved by the Animal Use and Care Committee for Research and Education of The Fourth Military Medical University (Xi\'an, China).
Consent for publication
=======================
Not applicable.
Competing interests
===================
The authors declare that they have no competing interests.
{#f1-ijmm-42-01-0270}
{#f2-ijmm-42-01-0270}
{#f3-ijmm-42-01-0270}
{#f4-ijmm-42-01-0270}
{#f5-ijmm-42-01-0270}
{#f6-ijmm-42-01-0270}
{#f7-ijmm-42-01-0270}
{#f8-ijmm-42-01-0270}
{#f9-ijmm-42-01-0270}
######
Primer sequences for real-time polymerase chain reaction.
Gene Forward (5′-3′) Reverse (5′-3′)
---------- --------------------------- ----------------------
α-SMA TTGTGGATCAGCGCCTTCAGTT GACAGGCCAGGGCTAGAAGG
vWF CCAACAGCCAGTCTCCCGTT TGACGTAGGAGCAGTTGCCG
CD31 TTCCGAGGGTGGCTTGAGTG GAGGGAAGGCAATGGGGGTT
Vimentin TGCAGTCACTCACCTGCGAA CATTTCACGCATCTGGCGCT
Col1A1 ATCACCAGACGCAGAAGTCATA ACCAGGAGGACCAGGAAGTC
Col3A1 GGAACAACTGATGGTGCTACTG CC AAAATAAGAGGGGTGAAG
β-actin ATCATGTTTGAGACCTTCAACA C ATCTCTTGCTCGAAGTCCA
SMA, smooth muscle actin; vWF, von Willebrand factor; Col, collagen.
[^1]: Contributed equally
| {
"pile_set_name": "PubMed Central"
} |
Passive smoking is the inhalation of second-hand smoke (SHS) by persons other than the "active" tobacco smoker. This exposure to SHS may cause disease, disability, and death \[[@b1-ophrp-10-049]\]. Passive smoking has been shown to be associated with the same diseases as active smoking (such as cardiovascular diseases, lung cancer, and respiratory diseases) \[[@b2-ophrp-10-049]\]. Analysis of the effects of SHS has come from studies of nonsmokers who are married to a smoker, with similar findings also reported in studies of workplace exposure to tobacco smoke \[[@b3-ophrp-10-049]\]. SHS consists of mainstream smoke (the smoke exhaled by a smoker) and sidestream smoke (the smoke released from a cigarette into the surrounding air) \[[@b4-ophrp-10-049]\]. The main component of SHS is sidestream smoke, which is about 4 times more toxic than mainstream smoke \[[@b5-ophrp-10-049]\], and has the most impact on the health of passive smokers \[[@b6-ophrp-10-049]\].
Looking forward, an objective of the Health Plan 2020 (HP2020) in Korea is to lower the exposure to SHS for those individuals who do not smoke tobacco by assigning designated non-smoking areas in public places in Korea such as squares and city parks \[[@b7-ophrp-10-049]\]. This supports the National Health Promotion Act (2017) which prohibits smoking in all middle and high school facilities, and smoking bans in college classrooms \[[@b8-ophrp-10-049]\].
Previous studies on the exposure to SHS have generally focused on the subjective reports of study participants regarding their exposure to SHS \[[@b9-ophrp-10-049]\]. Most studies lack the quantitative measurement of biomarkers of exposure to SHS such as cotinine, and nicotine-derived nitrosamine ketone in urine. The measurement of cotinine in hair and urine is accurate, and non-invasive, and is an effective indicator of exposure to SHS \[[@b1-ophrp-10-049]\].
In the current issue of Osong Public Health and Research Perspectives, a study by Park et al. measured the exposure to SHS amongst non-smoking, nursing college students, by measuring urinary cotinine, and analyzed the demographics, health, and smoking-related factors that may influence exposure to SHS \[[@b10-ophrp-10-049]\]. This cross-sectional study examined nursing college students (*N* = 196) who had not smoked tobacco in the previous year. To measure the exposure to SHS in students, the authors examined urinary cotinine levels, and social factors that may be influential, such as asking a smoker to put out their cigarette. The study observed that 32 students (16.3%) had urinary cotinine levels suggestive of exposure to SHS. There were 80 students (40.8%) affected by risk factors that increased exposure to SHS. In addition, students who were exposed to SHS were 4.45 times more likely to have increased urinary cotinine levels. It was reported that students who asked smokers to extinguish their cigarette, were 0.34 times less likely to test positive for urinary cotinine.
The authors concluded that non-smoking, nursing college students, that displayed self-assertive behavior by requesting smokers to extinguish their cigarette, could avoid exposure to SHS. It was also suggested that systematic support for the enforcement, and expansion of no-smoking zones, should be a requirement in all college facilities.
**Conflicts of Interest**
The author declares no conflicts of interest.
| {
"pile_set_name": "PubMed Central"
} |
Introduction {#s1}
============
Among trace elements, zinc (Zn) is a micronutrient influencing growth and affecting the development and integrity of the immune system \[[@R1]\]. Also, it has important roles in physiological functions of the human body. It has a critical role for the function of over 300 enzymes \[[@R2]\]. Furthermore, it plays an important role in function of the liver. There are hepatic and extrahepatic actions for Zn in the prevention of alcoholic liver injury \[[@R3]\].
Zinc deficiency has been involved in the pathogenesis of a number of clinical findings in chronic liver disease. These include the possible role of Zn deficiency in the pathogenesis of hepatic encephalopathy, by inducing alterations in urea metabolism \[[@R4]\]. In a study, Gur et al. reported decreased level of plasma zinc in cirrhotic patients due to hepatitis B \[[@R5]\]. Hepatitis B and hepatitis C are the most common causes of liver cirrhosis in many countries like Iran \[[@R6]\]. As nutritional habits are different among various populations, it is logical to determine Zn level in Iranian cirrhotic patients as an important trace element. There are some differences in the etiology, disease progress, prognosis and treatment plan of hepatitis B and C, and it is possible that plasma zinc concentration differs in two groups. The aims of the present study were 1) determination of plasma zinc level in a sample of Iranian cirrhotic patients due to hepatitis B or C, 2) to determine if there is any difference between plasma zinc levels between the 2 groups.
Materials and Methods {#s2}
=====================
In a cross sectional study, adult cirrhotic inpatients due to hepatitis B or hepatitis C referred to gastrointestinal and liver disease center of Taleghani Hospital, Tehran, Iran were enrolled for the study. This place is one of the important referral centers of liver disease in Iran. Diagnosis of cirrhosis and hepatitis were determined by clinical, laboratory and liver biopsy from the patients. Exclusion criteria were patients who had been co-infected with hepatitis B and C and those who had taken a zinc tablet or any complementary medicine with zinc in its content. Because of the difference in the zinc amount of foods, we preferred to select patients nourished with a similar regimen prepared by the kitchen of the hospital during the study. After informed consent, 5 cc blood samples were kept from the forearm of each patient in fasting state in the morning. Blood samples were centrifuged in 5000 rpm for 5 minutes and plasma was separated. Then plasma samples were kept in the Frazer at the -25 degree centigra . Because the usual method for assessment of zinc in plasma or serum is the atomic absorption \[[@R7]\], we selected the same method for zinc assessment. Concentrations of 0.1, 0.3, 0.5 and 0.7 ppm of zinc sulfate were prepared as standard samples. Atomic absorptions of them were determined for drawing standard curve. After completion of 60 samples of patients, plasma zinc levels were evaluated by using atomic absorption (Perkin Elmer 1100B).
Results {#s3}
=======
60 cirrhotic patients due to hepatitis B or Hepatitis C were included in the study during 7 months (April 2008 to November 2008). Of them 36 were hepatitis B patients and 24 were hepatitis C patients. Demographic data of patients including age distribution, sex and smoking habits were shown in [Table 1](#s3tbl1){ref-type="table"} Analysis of chi-square showed that there is not any difference between 2 groups regarding to sex, age distribution and smoking habits (P = 0.82, 0.53 and 0.8 respectively).
Mean ± standard deviation (SD) plasma zinc levels determined 0.34±0.22 mg/L and 0.37±0.22 mg/L in hepatitis B and hepatitis C patients respectively. Analysis of t-test showed that there is not any significant difference between 2 groups regarding to plasma zinc level (P = 0.745).
###### Demographic of patients.
Parameter Total Parameter N= 60 HCV Patients (N= 24) HBV Patients (N=36) P-value
---------------- ----------------------- ---------------------- --------------------- --------- ------
Age \< 40 12 6 6 0.53
41-60 40 14 26
61\> 8 4 4
Sex Female 21 8 13 0.82
Male 39 16 23
Smoking Habits No 22 10 12 0.80
Yes 38 14 24
Discussion {#s4}
==========
Plasma trace elements concentrations are frequently reported to be a good indicator for diagnosis and prognosis of some diseases \[[@R8]\]. Previous studies showed a decrease in zinc level in cirrhotic patients. Pramoolsinsap et al. have stated that serum zinc levels were significantly decreased in patients with chronic active hepatitis, cirrhosis, and hepatocellular carcinoma \[[@R9]\]. Lin et al. announced that the zinc concentration in the serum of Chinese patients with hepatic cirrhosis was significantly less than a control group \[[@R10]\]. It should be noted that zinc level is usually related to the nutritional pattern of each population. It has been shown that zinc deficiency is widespread in people living in developing countries like Iranian populations who consume rice-based diets \[[@R11]\]. The phytate and fiber present in cereal diets can form insoluble complexes with zinc leading to its decreased bioavailability \[[@R12]\].
Since nutritional impairment is common in cirrhotic patients \[[@R13]\], it seems that determination of zinc level in Iranian cirrhotic patients due to hepatitis B and C and comparison with a normal amount of healthy people is necessary as an indicator of nutritional status.
Some demographic data may alter zinc plasma concentration in human. Lopez et al. reported that Serum Zn concentrations were slightly higher in men than in women and also there is some elevated serum Zn levels in smoking men rather than non smokers \[[@R14]\]. In the present study, since there is no significant difference between the 2 groups regarding gender, age distribution and smoking habits, these parameters could not lead to biases in interpretation of zinc level in both of the groups.
The results showed that plasma zinc levels of both groups were below the normal range as mentioned by the similar investigations \[[@R9]\]\[[@R10]\]. There are some differences in the normal range of zinc in various populations, but a recent study reported the normal range of 0.89 ± 0.16 mg/L for plasma zinc in healthy volunteers in Tehran \[[@R15]\]. With a comparison result of the present study (0.34±0.22 mg/L and 0.37±0.22 mg/L) with a range of the latter study, it seems that plasma zinc level of the cirrhotic patients are less than half of normal values. As nutritional parameters, this study significantly indicates a zinc level deficiency in Iranian cirrhotic patients due to hepatitis B or C in comparison with healthy volunteers.
The results of the study are more considerable because of the effective role of the zinc supplement in pharmacotherapy of viral hepatitis. Yuasa et al. have shown that zinc may play an important role as a negative regulator of hepatitis C virus (HCV) replication in genome-length HCV RNA-replicating cells. They mentioned that zinc appears to offer a novel approach to the development of future plans for the treatment of intractable chronic hepatitis C \[[@R16]\]. Himoto et al. examined the effects of polaprezinc, a complex of zinc and L-carnosine, on inflammatory activity and fibrosis in the HCV infected patients. They reported that polaprezinc exerts an anti-inflammatory effect on the liver in patients with HCV-related Chronic liver disease by reducing iron overload \[[@R17]\].
Based on the result of the study, administration of zinc may be recommended for Iranian cirrhotic patients due to hepatitis B or C.
In future more studies recommend for the role of zinc administration on clinical, pathological status and pharmacotherapy response of Iranian cirrhotic patients due to hepatitis B or C.
| {
"pile_set_name": "PubMed Central"
} |
Background {#Sec1}
==========
School safety is fundamental to fostering positive short and long-term outcomes for children, including positive mental health, school connectedness, student retention and academic success (Battistich, Schaps, & Wilson, [@CR6]; Horner et al., [@CR37]). We define a safe school as one that minimizes violence, promotes student mental health, and fosters a social climate that promotes positive development. Unfortunately, concerning rates of school violence persist in the US: in 2017, 19.0% of students were bullied, 15.7% carried a weapon at least once in a month (3.8% on school property), and 6% had been threatened or injured with a weapon (Kann et al., [@CR44]). Rates of violent, aggressive, and bullying behaviors are similarly concerning among younger students with 8.1% of elementary and 21.8% of middle school students reporting daily or weekly bullying in 2015--16 (Diliberti, Jackson, & Kemp, [@CR22]). Violence victimization is associated with distress, adjustment difficulties, and mental health problems. Exposure to violence, including direct victimization and well as exposure in the school environment, is a potent risk factor for poor mental health outcomes including depression and anxiety (Kennedy, Bybee, Sullivan, & Greeson, [@CR47]).
The CDC (Perou et al., [@CR76]) identifies mental health problems, including depression and anxiety, as a critical public health issue among youth with significant impact on the individual, family, and community. Mental health issues that go untreated early in life are associated with further problems, including increased likelihood of academic failure, dropout, substance use, relationship conflicts, violence, and suicide (World Health Organization, [@CR94]). In the short term, mental health problems evident in middle school predicts school absences a year later (Suldo, Thalji, & Ferron, [@CR91]). In addition, many children, particularly those living in low resource communities, experience disproportionate risk of violence and subsequent mental health consequences without sufficient treatment and prevention services needed to reduce risk of poor outcomes (O'Connell, Boat, & Warner, [@CR73]). Childhood vulnerability is exacerbated in high stress environments when children and youth receive limited support from adults (O'Connell et al., [@CR73]). Yearly, an estimated 13--20% of children aged 3--17 years experience a mental disorder, and more than half of lifetime psychiatric diagnoses have an initial age of onset before age 14 (Kessler et al., [@CR48]).
School climate plays a major role in shaping the lives of students, affecting violence (Brookmeyer, Fanti, & Henrich, [@CR14]), mental health and wellness (Jacobson & Rowe, [@CR39]), truancy and achievement (Astor, Guerra, & Van Acker, [@CR2]). The National School Climate Council recommends an encompassing definition of school climate that includes experiences of school life, that reflect the norms, goals, values, teaching, organization structure, and relationships. Relationships include connections among students, teachers, and staff; feelings of commitment to the institution; and connection to a community (Pittman & Richmond, [@CR78]).
The relationship between student outcomes and school climate are evident longitudinally (Anderman, [@CR1]; Goodman, [@CR30]). Researchers have found that poor school-based relationships were associated with initiation of deviant behavior (Dornbusch, Erickson, Laird, & Wong, [@CR23]; McNeely & Falci, [@CR65]). Consequently, promoting a positive school climate is an important mechanism by which interventions can reduce risk of poor health outcomes including violence and mental distress. Researchers suggest that schools, particularly those with concentrated poverty, may benefit from multicomponent prevention approaches that improve positive discipline management and support positive psychosocial climates, effectively identify youth experiencing mental distress and improves the physical as well as social environment (Gottfredson, Gottfredson, Payne, & Gottfredson, [@CR31]).
In order to effectively address challenging issues such as school safety, communities need to deliver multicomponent interventions targeting prevention efforts across levels of social ecology (Komro, Flay, Biglan, & Wagenaar, [@CR55]; PriCowan, Vaillancourt, Rossen, & Pollitt, [@CR80]). Even the best individual interventions have relatively limited scope in terms of outcomes when offered alone, and, consequently, small effects when taken to scale; therefore, multicomponent interventions have greater potential to achieve positive outcomes at the school or community level than a single intervention alone (Komro et al., [@CR55]). Evidence-based interventions (EBIs) that address the multifaceted nature of school safety such as Restorative Justice (RJ), Mental Health First Aid (MHFA) and Crime Prevention Through Environmental Design (CPTED) when deloyed as a single, coordinated, multicomponent intervention approach, are promising approaches to improving the school environment.
Yet, more complex, multicomponent interventions are also more challenging to implement. Such interventions require effective implementation strategies to adopt the constellation of EBIs and adapt them to suit the needs of the context, providers and target population. Interventions will fail to achieve their desired effects if not implemented well (Durlak, [@CR24]). Researchers have acknowledged that evidence-based interventions must be complemented by implementation strategies to achieve desired public health outcomes (Kirchner, Waltz, Powell, Smith, & Proctor, [@CR51]). Implementation strategies are highly specified, theory-based methods to enhance EBI delivery in community settings and are key to bridging the research-to-practice gap (Kilbourne et al., [@CR50]). Study designs that assess implementation strategy utility *and* evaluate EBI effectiveness, such as with hybrid designs, provide vital information for stakeholders about using implementation strategies with new innovations to maximize public health impact.
To inform the optimal implementation strategies for these effective interventions, we develop and test an intervention focused on promoting a positive school climate guided by a school-based 3-person leadership team (3-PLT) using a hybrid Type II design. A hybrid Type II design tests the effectiveness of the intervention and determines the feasibility and potential utility of an implementation strategy (Curran, Bauer, Mittman, Pyne, & Stetler, [@CR21]). The 3-PLT includes representatives from the police (School Resource Officer, SRO), school (e.g., administration), and mental health services (i.e., social work), the latter of whom leads the team as a newly appointed climate specialist (CS). The team, led by the CS, work together to support the integration of the key intervention components: (1) RJ practices, (2) MHFA training, and (3) CPTED (see Fig. [1](#Fig1){ref-type="fig"}). The CS coordinates these efforts as a staff member within the school through a process of interactive problem solving and supports, consistent with the implementation facilitation strategy (Ritchie, Dollar, Kearney, & Kirchner, [@CR83]). We will focus on change among students in an early developmental period---elementary school students aged 8--12 years---in a county with significant social and economic challenges. The purpose of this research is to study the effectiveness and implementation of three complementary interventions delivered concurrently to enhance school safety through improved school climate. School climate represents a critical mechanism by which interventions, including multicomponent school safety and mental health interventions, can reduce risk of violence and mental distress among youth. Fig. 1Proposed Conceptual Relationship Between Evidence-Based Interventions, Implementation Strategies and Study Outcomes. Adapted from Proctor et al. ([@CR77]) and Lyon (Lyon, [@CR59])
Methods/design {#Sec2}
==============
Aims and objectives {#Sec3}
-------------------
The overarching objective of this study is to provide a safe school environment to positively impact health, wellbeing, social, educational, violence and delinquency outcomes among youth. This is accomplished through the concurrent delivery of a multicomponent approach consisting of 3 integrated interventions: Restorative Justice (RJ), Mental Health First Aid (MHFA) and Crime Prevention Through Environmental Design (CPTED). The implementation strategy used is a facilitation approach based on the Implementation Facilitation, Enhanced REP, and the iPARiHS framework (Kilbourne et al., [@CR49]; Ritchie et al., [@CR83]) as part of the 3-PLT (see Fig. [1](#Fig1){ref-type="fig"}). Elementary school safety is understudied yet represents a critical period in which to develop positive mental health, build constructive and trusting relationships with adults, and prevent early experiences of violence.
Primary study aim {#Sec4}
-----------------
The primary study aim is to examine the overall effectiveness of the intervention, including change in violence (e.g., fights, bullying, victimization), over time compared with a control group of students who receive school practice as usual.
Secondary study aim 1 {#Sec5}
---------------------
Employ implementation facilitation from an appointed school climate specialist (CS) to support delivery of a multicomponent, integrated intervention and evaluate feasibility and potential utility to support sustainment.
Secondary study aim 2 {#Sec6}
---------------------
Examine specific mechanisms associated with change in mental health (e.g., anxiety, depression, well-being), including perceptions of school climate as a moderator.
Secondary study aim 3 {#Sec7}
---------------------
Estimate the costs of the intervention and its implementation and conduct a cost-benefit analysis for positive outcomes, such as improve school climate perceptions, associated with participation in the interventions.
Methods {#Sec8}
-------
The cluster randomized trial evaluates a school safety intervention in a community with significant need, Genesee County, Michigan (see Fig. [2](#Fig2){ref-type="fig"}). This study was reviewed and approved by the Michigan State University Institutional Review Board (IRB\# × 15-1129e). This study takes advantage of ongoing partnerships with the Genesee Intermediate School District in Michigan. We do not intend to collect data from participants who discontinue or deviate from protocols. Among the primary data sources to be collected (e.g., focus groups, interviews, teacher surveys), the data will be kept on a password protected server and de-identified. The study does not employ a data monitoring committee, but this study is structured such that the intervention team (i.e., employees of the GISD), is separate from the training team (i.e., employees of international and regional organizations, local universities, and the district wide \[GISD\] training team offering training in facets of the intervention), which is separate from the data collection team (i.e., employees of the two universities charged with the process and outcome evaluation, led by a co-PI), which is separate from the data analysis team (i.e., employees of the two universities charged with the process and outcome evaluation, led by a separate co-PI), providing sufficient independence and protecting against potential conflicts of interest. Thus, those tasked with training and technical assistance associated with components of the intervention are not involved in the analysis of data or reporting of outcomes given the potential bias this might engender as a result of the incentives associated with reports of a positive programmatic impact, such as profits associated with increased demand for training and technical assistance in the components of the program. There is no similar financial incentive for university-based researchers charged with the analysis of data and reporting of results, as they are not employed by any of the organizations charged with program implementation (i.e., GISD) or future dissemination of intervention components (i.e., training and technical assistance providers). Fig. 2Cluster RCT of the Comprehensive School Safety Intervention delivered using the implementation facilitation strategy
Study setting {#Sec9}
-------------
The current study is in Genesee County, Michigan. Widespread losses in industrial jobs in many US cities have led to unemployment, population loss and changes in land use (including high rates of vacant buildings); these demographic and economic changes have contributed to neighborhood instability, disadvantage and elevated rates of violent crime (Glaeser & Gyourko, [@CR29]; Sampson, Raudenbush, & Earls, [@CR86]). Genesee County, especially around the city of Flint, has lost almost 90% of the automotive industry jobs that bolstered its economy in the 1960s. The economic hub of Genesee county, Flint's current challenges include high rates of poverty (60% of children live in poverty) (U.S. Census Bureau, 2015) and an unemployment rate that is 50% higher than the state average (Bureau of Labor Statistics, [@CR16]). Across Flint community elementary schools, 92% of students are eligible for free or reduced lunches (Michigan Department of Education, [@CR67]). Surrounding Genesee communities face similar disadvantage with poverty (see Table [1](#Tab1){ref-type="table"}) and violence rates exceeding state averages. This high level of community violence has significant implications for levels of trauma and corresponding mental health issues for youth residing in Flint and surrounding Genesee County. Table 1Study Sites by demographic variablesSchoolFree or Reduced Lunch ^a^Grade 3--8 Students proficient in Math and English (2017--2018)^b^Racial/ethnic minority composition^a^178%7.6%34%286%38.6%31%355%15.0%29%466%26.3%12%591%3.4%73%676%9.9%16%784%10.4%24%884%2.5%86%978%12.9%69%1055%29.4%31%1197%14.6%100%1275%32.0%72%1391%6.4%72%1468%34.1%40%1584%15.0%74%1666%29.5%21%1756%16.9%10%1852%27.1%96%1995%3.2%9%2090%11.1%93%State of Michigan45.9%30.7%67.0%^a^Data provided by the partner intermediate school district, internal data^b^Parent Dashboard for School Transparency, Center for Educational Performance and Information, [https://www.mischooldata.org](https://www.mischooldata.org/) Accessed 9/13/2019
School safety challenges {#Sec10}
------------------------
School safety is a major concern in Flint and Genesee Co. as a whole. In a survey of Genesee Co. 4th through 6th grade students, 42% reported seeing violence in the school once per month or more, 13% did not feel safe while at school, and 3% skipped school because they were afraid of being hurt. There were also widespread mental health challenges: 44% of students worried a lot, 43% felt nervous, 30% could not stop being sad, and more than half felt they were too tired to do things (Prevention Research Center of Michigan, [@CR79]).
Study design {#Sec11}
------------
See Fig. [2](#Fig2){ref-type="fig"} for study flow diagram. In the pre-implementation phase, the project team will meet with intermediate school partners and individual schools, finalize processes and measures, secure formal agreements of study participation with schools, hire staff for the project and schedule all trainings. The current study is a cluster randomized trial in which each school will participate in the study over two years. The interventions will be staggered (5 intervention + 5 control schools starting in Year 1; 5 additional intervention and 5 additional control schools in Year 2) in order to maximize the likelihood of intensive implementation. It focuses on three key areas: (i) at a universal prevention level, focusing on changes to the physical school environment (CPTED) and promoting consistent and fair discipline and reward practices (RJ); (ii) a more targeted approach, to help those with early signs of mental health problems (MHFA); and (iii) engaging those involved in violent or aggressive situations at school through RJ practices. As noted above, the interventions are guided by a 3-PLT who seek to utilize the interventions to promote positive school climate.
Sample {#Sec12}
------
### Schools {#Sec13}
We will recruit 20 schools and implement the intervention across three school years (see Table [1](#Tab1){ref-type="table"}). Co-educational elementary schools in Genesee County Intermediate School District (GISD), Michigan will be invited to participate. Schools participating will be similar in terms of proportion of students receiving free/reduced school lunches and approximate size of the school. Prior to the intervention roll-out, all 20 schools will be randomized to intervention or control. The intervention will be implemented in stages. Cohort 1 schools will include 5 intervention and 5 control schools. Cohort 2 will include10 additional schools (5 intervention and 5 control) for a total of 10 intervention and 10 control schools; each school will implement the intervention for two years.
### Student eligibility and recruitment {#Sec14}
There are 21 public school districts and 13 academies (charter schools) serving over 67,000 students in the GISD. Recruitment efforts will begin with presentations to the superintendents of the schools and academies. This will be followed up by similar presentations and briefings to school building principals and/or their designees (dean of students, academic services, teachers). The briefings explain the goals of the research project, the benefits to participating schools, anticipated intervention and research activities, eligibility, and similar issues. Eligibility includes having at least two grades between grades 4--6 and having a student population with 50% or higher free or reduced school lunch eligibility, being involved in MIBLISI (Michigan's Integrated Behavior and Learning Support Initiative) and PBIS (Positive Behavioral Interventions & Supports) and being committed to the project. Across Genesee County, there are currently an average of 180 4th--6th graders in each school resulting in a maximum sample size of 3600 students over the course of the study (using a 66% consent/retention rate).
Evidence-based interventions to be implemented {#Sec15}
----------------------------------------------
### Universal and targeted prevention {#Sec16}
*Universal strategies* address an entire population regardless of risk level or current behavior. This approach aims to reach a large number of individuals at once; it develops strategies and a supportive environment that enables all children to achieve sufficient competence to thrive, and thus prevent or reduce engagement in violence (Leshner, [@CR57]). In contrast, *targeted strategies* are designed for individuals who meet specified risk criteria, for example, showing signs of mental illness (Leshner, [@CR57]). Simultaneous deployment, particularly one tailored to a school's culture, is commonly considered more likely to effect change (Bonell et al., [@CR11]; Maddox & Prinz, [@CR61]). The proposed intervention incorporates universal and targeted approaches facilitated by a school-based 3-PLT. The team will participate in ongoing activities and engagement with school staff and students to support and reinforce the interventions, foster school-community relationships and support sustainability. These activities will be led by the CS who will apply principles of implementation facilitation to promote effective EBI delivery and tailor the interventions to the context.
The universal and targeted prevention design reflects the integrated and comprehensive intervention and suggested best practice, i.e. is comprehensive, improves access to mental health, balances physical and psychological safety, employs a positive school discipline approach, considers and responds to each schools' culture and context, and acknowledges that change takes time (PriCowan et al., [@CR80]). Prior school safety interventions have tended to implement these strategies in isolation rather than integrating in a coordinated manner. The proposed research will advance comprehensive school safety and using of implementation science to enhance delivery of such programs.
### Restorative justice (RJ) {#Sec17}
Restorative justice is a philosophy and process that defines crime as doing harm to people and relationships, rather than simply violating the law (Zehr, [@CR98]). Thus, it requires consideration of the victim and community during the justice process. Although the definition continues to evolve, it has most commonly been described as "a process whereby all the parties with a stake in a particular offense come together to resolve collectively how to deal with the aftermath of the offense and its implications for the future" (Marshall, [@CR62]). Restorative practices focus on repairing the harm caused, engaging victims and relevant school community members in the decision making process, holding offenders accountable, and preventing similar actions in the future. By focusing on accountability, fairness, and situational responses to unique events, restorative justice provides a useful framework for alternatives to zero tolerance or traditional disciplinary actions for school violence. Overall, researchers have found robust support for RJ practices, including enhanced school safety, reduced discipline problems and other behavioral referrals (Karp & Breslin, [@CR45]; Mirsky, [@CR69]; Stinchcomb, Bazemore, & Riestenberg, [@CR89]).
Prior research suggests that RJ practices may have particular efficacy for youth in the age ranges targeted in this project. Researchers found, among students 14 years and younger, that youths participating in family group conferences were less likely to re-offend than youths participating in other court-ordered programs (McGarrell & Hipple, [@CR64]). Further, offending youths, parents, and victims all expressed much more favorable perceptions of the fairness, respect, and value of the conference experience (McGarrell, [@CR63]). Long-term follow-up of this research demonstrated that conferences following principles of "restorativeness" and "procedural fairness" were associated with reduced re-offending at 24 months and 10 years following the original conference (Hipple, Gruenewald, & McGarrell, [@CR35], [@CR36]).
Restorative justice programs can be implemented in a variety of ways to meet the needs of individual schools, but retain a set of common principles: (1) repair harm, (2) reduce risk, and (3) empower community (O'Brien, [@CR72]). When applied to school environments, the restorative approach comprises an overarching philosophy and processes that build community in classrooms and entire schools and include proactive processes that aim to prevent wrongdoing (Mirsky, [@CR69]). This may be accomplished through school-wide discipline practices promoting fairness in rules and enforcement, as well as consistency of reinforcing positive behavior, and may include more targeted approaches like peer mediation (Bazemore & Umbreit, [@CR7]; Reimer, [@CR81]). Peer mediation involves a facilitated discussion with multiple stakeholders including the offender, victims, family members, friends, school personnel, and community members. It is typically used in response to a specific offense, to explore what happened, address and repair the harm done, and determine strategies for preventing it in the future. The 3-PLT and particularly the mental health expert will lead many of the proposed peer mediation and restorative justice components of the program and it aligns closely with the support in the MHFA and climate change associated with CPTED. Within this framework, individual schools can tailor the approach as needed for their school and more specifically for each mediation need.
### Mental health first aid (MHFA) {#Sec18}
MHFA is defined as help provided in the context of a mental health problem or mental health crisis; importantly, the help is provided until appropriate professional help is received or the crisis is resolved (Yap & Jorm, [@CR95]). The Substance Abuse and Mental Health Services Administration (SAMSHA) (SAMHSA National Registry of Evidence-based Programs and Practices, [@CR85]) lists MHFA as part of its National Registry of Evidence-based Programs and Practices. Trainers deliver an 8-h session not intended to teach therapeutic skills but rather to raise awareness of mental illness symptoms and build skills in providing initial help and guiding someone toward treatment (Kitchener & Jorm, [@CR52]). It includes highlighting evidence-supported treatment and engaging individual resources (e.g., family).
Qualitative studies and randomized controlled trials of MHFA indicate greater confidence to help, less stigma around mental illness, better recognition of symptoms, and improved perceived value (Jorm, Blewitt, Griffiths, Kitchener, & Parslow, [@CR42]; Jorm, Morgan, & Wright, [@CR43]; Kelly et al., [@CR46]; Yap & Jorm, [@CR95], [@CR96]; Yap, Wright, & Jorm, [@CR97]). Research examining MHFA has focused on adults and adolescents, yet preadolescents are also in need of timely and appropriate mental health services, suggesting value in evaluating and validating an extension of the approach for preadolescents.
Teachers are typically not trained in mental health (Reinke, Stormont, Herman, Puri, & Goel, [@CR82]) but are in a position to identify early symptoms, which if managed promptly can substantially reduce negative outcomes and healthcare costs (Knapp, McDaid, & Parsonage, [@CR53]). MHFA training may support teacher needs by providing resources and skills as well as debunking misconceptions and myths which may otherwise reduce helping behavior (Yap & Jorm, [@CR95]). Teachers typically do not feel prepared to identify or manage mental health concerns (Koller, Osterlind, Paris, & Weston, [@CR54]), yet mental health issues among children and youth is on the rise (Fombonne, [@CR26]). Among elementary school teachers, 91% indicated they were concerned about a student's family stressors, 76% about anxiety, and 54% about depression; 94% also agreed that schools should be involved in addressing mental health issues. Further, teachers' beliefs that they could help students depended on their own positive psychological wellbeing, satisfaction with school climate, and confidence (Sisask et al., [@CR88]), highlighting further potential benefits of MHFA through promoting close relationships with community mental health and a positive school climate. The current study seeks to train all school staff to provide MHFA; the climate specialists will conduct periodic MHFA 'boosters,' and serve as a resource to teachers and students.
### Environmental design {#Sec19}
The CDC (Centers for Disease Control and Prevention (CDC), [@CR17]) identifies environmental design as an area of interest for school violence prevention, with an ongoing study of Crime Prevention Through Environmental Design (CPTED). There are typically six components: surveillance, territoriality, image/management, access control, activity support, and target hardening (Cozens, Saville, & Hillier, [@CR20]). *Surveillance* may be natural (e.g., teachers' office overlooking the playground), formal (e.g., playground duty or police patrols), or mechanical (e.g., cameras). *Image/management* refers to maintaining or improving physical spaces (e.g., graffiti cleanup). *Territoriality* refers to promoting a "sense of ownership" by legitimate users, thus reducing the likelihood of "illegitimate" use. *Access control* limits the accessibility of potential targets. *Activity support* promotes "intended patterns" of use for public spaces (e.g., increasing pedestrian traffic). Finally, *target hardening* increases offender effort (e.g., by installing fences). Such elements may be differently considered as they best relate to promoting positive school climate in each school. In consultation with environmental design experts, school staff, and student input, the 3-PLT will work to implement actionable design changes to improve feelings of safety, beautification, ownership, and youth empowerment.
### Surveillance {#Sec20}
Cameras have become increasingly prevalent^3^ and are the second most common security measure (77% of schools), after locking/monitoring doors during school hours (93%) (Gray & Lewis, [@CR32]). While many school administrators believe cameras to be effective (Garcia, [@CR28]), there is little evidence to suggest students share their confidence (Bracy, [@CR12]; Brown, [@CR15]). Cameras do not significantly reduce students' self-reported victimization (Blosnich & Bossarte, [@CR10]); rather, they may be associated with increased likelihood of physical victimization (Jeong, Kwak, Moon, & San Miguel, [@CR40]) and fear of harm (Bachman, Randolph, & Brown, [@CR4]). Other security measures (e.g., guards, metal detectors, locker checks) may increase fear of crime (Schreck & Miller, [@CR87]). These highly visible efforts may increase fear by signaling the school must be unsafe (Schreck & Miller, [@CR87]) or, "coercive" measures may contribute to an "atmosphere of mistrust (Brown, [@CR15])." As a whole, evidence suggests that such measures may not be particularly effective in increasing school safety; by contrast, other CPTED components such as territoriality or image/management remain more promising.
### Image/management {#Sec21}
Though the impact of school physical environment on safety remains relatively understudied (Johnson, [@CR41]), evidence suggests it has merit. Wilcox et al. (Wilcox, Augustine, & Clayton, [@CR93]) found that school disorder (e.g., presence of graffiti, litter) had a significant positive association with teachers' perceptions of student misconduct. Notably, there was a significant negative association between teacher perceptions of school crime and hallway territoriality (i.e. teachers perceived less crime in areas with signs of ownership such as trophy cases or murals).
### Territoriality and undefined space {#Sec22}
CPTED strategies often address undefined space^33^ that include semi-public areas without clear ownership, "that may not be seen as anyone's responsibility to monitor or maintain (Astor, Meyer, & Pitner, [@CR3])." Astor et al. ([@CR3]) used mapping and individual interviews to identify areas that elementary and middle school students perceived as unsafe. Often, these areas had characteristics of undefined space, with a lack of adult supervision and overcrowding. Astor et al. ([@CR3]) recommended additional monitoring strategies: for example, having teachers stand in their classroom doorways and greet students in the hall during transition times. Hallway supervision by staff (other than security guards) was the only security measure that effectively reduced any form of peer victimization (compared to student ID badge requirements, cameras, security guards, or a formal student code of conduct) (Blosnich & Bossarte, [@CR10]).
### Implementation facilitation {#Sec23}
Facilitation is an implementation strategy based on the Integrated-Promoting Action on Research Implementation in Health Services Framework (iPARiHS) (Harvey & Kitson, [@CR33]; Kilbourne et al., [@CR49]; Ritchie et al., [@CR83]) that promotes provider self-efficacy (Bandura, [@CR5]) in mitigating organizational barriers to EBI adoption (see Fig. [3](#Fig3){ref-type="fig"}). Facilitators are individuals who are familiar with the EBIs and organization's procedures, climate, and processes with devoted time to support implementation activities; Facilitation includes diverse, implementation-science informed, tailored activities that enhance EBI delivery (e.g., stakeholder engagement) and identify and solve implementation challenges (Ritchie et al., [@CR83]). Facilitation will be delivered via regular contact with the school staff and other 3-PLT members by the CS trained in program implementation and use of RJ, MHFA and CPTED in schools. The CS will support the school staff and 3-PLT in strategic thinking and program specific skills to address barriers related to the context, innovation, provider and recipients (see Fig. [3](#Fig3){ref-type="fig"}). Fig. 3Implementation facilitation guided by the iPARHIS framework (Harvey & Kitson, [@CR33]), adapted
Although the 3-component intervention approach is consistent across schools, approaches to integrating the EBIs will be tailored, making complete standardization neither feasible nor desirable. The proposed study will include development of an implementation guide as part of the Facilitation implementation strategy. The implementation guide will provide scaffolding for the CS to work with the other 3-PLT members and school staff to integrate the interventions into each school setting. The guide will provide specific steps in delivering the EBIs across sites, but will allow for tailoring of the interventions (innovations) to meet the needs of the schools (context), their staff (providers) and their students (recipients) (Bonell et al., [@CR11]; Patton, Bond, Butler, & Glover, [@CR75]; Toumbourou et al., [@CR92]); we therefore seek to evaluate district-wide (yet individualized) implementation of the three school safety components.
The CS, in collaboration with the 3-PLT will be provided with a comprehensive set of resources and training for components of MHFA, RJ, and CPTED. The 3-PLT will be trained to deliver MHFA training to all school staff, deliver restorative justice efforts (e.g., be trained in peer mediation and similar restorative processes and facilitate this within schools), and liaise with expert consultants of environmental design. Environmental design components will be tailored for each school's unique physical and social structures; in general, aspects of the school's image/management, and increasing ownership of undefined spaces will be emphasized. The CS will coordinate activities and efforts between the 3-PLT, school staff, parents and youth. The CS will also develop processes based on local evidence. These processes will aid in informing refinements to the implementation guide that will support program fidelity while guiding customization of adaptable elements, and long-term sustainability.
Measures {#Sec24}
--------
A summary of the measures is provided in Table [2](#Tab2){ref-type="table"}. Table 2Data sources and measuresAimMeasuresMeasure frequencyData SourcesOriginal Data SourcesPrimary: Intervention effectiveness on violence outcomesAggression, violence victimization, bullyingFall Year 1, Spring Year 1, Fall Year 2, Spring Year 2Michigan Profile for Healthy Youth Survey (MiPHY)^a^Youth Risk Behavior Surveillance Survey (YRBSS)^b^Bully-Free Schools Survey (BFSS)^c^Truancy, behavioral referralsSpring Year 1, Fall Year 2, Spring Year 2School Information System (SIS)School Student DataSecondary: feasibility and potential utility of implementation facilitationMeeting agenda, interview promptsSpring Year 1, Spring Year 2Staff, student and 3-Person Leadership Team (3-PLT) focus groupsQuestions developed by research teamSchool climate, perceptions of program effectiveness (e.g., violence)Spring Year 1, Spring Year 2Delaware School Climate Teacher/Staff Survey^d^Delaware School Survey - Teacher/Staff (DSS-T/S)^d^Secondary: mechanisms and mental healthSchool climate scaleFall Year 1, Spring Year 1, Fall Year 2, Spring Year 2Delaware School Climate Student Survey^d^Delaware School Survey - Student (DSS-S)^d^Anxiety, depressive symptoms, wellbeingFall Year 1, Spring Year 1, Fall Year 2, Spring Year 2Michigan Profile for Healthy Youth Survey (MiPHY)^a^Youth Risk Behavior Surveillance Survey (YRBSS)^b^Secondary: Cost-benefitIntervention (e.g.,materials) and imple-mentation costs (e.g., training, support)MonthlyTime tracking by research and school staffData collected by research team^a^(Michigan Department of Education, [@CR66]), ^b^ (Kann et al., [@CR44]), ^c^ (Lindenberger, [@CR58]), ^d^ (Bear et al., [@CR8])
Primary outcomes {#Sec25}
----------------
### Violence {#Sec26}
We will use the Michigan Profile for Healthy Youth (MiPHY) to evaluate student outcomes related to aggression and violence, and the intermediate school district across Genesee Co. The MiPHY represents the Michigan Youth Risk Behavior Survey (YRBS), part of a nationwide surveying effort led by the Centers for Disease Control and Prevention (CDC) to monitor students' health risks and behaviors (Michigan Department of Education, [@CR66]). YRBS has been conducted annually by the CDC since 1993 and the psychometric properties have been evaluated and published (Brener et al., [@CR13]; Kann et al., [@CR44]). The MiPHY was designed to track students over time and those that move across districts. As a coordinated survey, MiPHY assesses a broad range of content related to students' physical and mental health as well as attitudes, beliefs, and behaviors in their school, home, and neighborhood environments. Related to this project, it will include measurement of aggressive behavior, violence victimization, and bullying. The data will be maintained by GISD and provided in a de-identified format to the evaluation teams.
### School-level discipline {#Sec27}
The School Information System (SIS) is a web-based information system, implemented in GISD schools along with PBIS. It is designed to be an efficient, reliable, and confidential tool for collecting, summarizing, and using student discipline data, enabling school personnel to be more successful decision makers. Presently, schools in Genesee Co. report student violations and truancy via SIS, creating a standardized and comparable metric across schools in the proposed study area. We will compare treatment versus control school outcomes utilizing these data. Additional time will be allocated for more intensive analyses and linkage with MiPHY and focus group results.
Secondary outcomes {#Sec28}
------------------
### Feasibility and utility of implementation facilitation {#Sec29}
#### Focus groups {#FPar1}
We will invite all teachers of senior elementary students to participate in a focus group at each intervention school at the end of the school year, for each year of the intervention. A single discussion will be held at each school at a convenient time (e.g., lunch break). We estimate 85% of teachers will participate, based on previous school research. We will ask teachers about their experiences with the intervention and the implementation supports, including their perceptions of effectiveness. We will also ask about any feedback they have on the interventions and their delivery. Discussions will be approximately 30 min in duration and will be conducted by research staff trained in facilitating focus groups. Teachers will provide informed consent prior to participation. Students from four randomly selected intervention classes will be recruited to participate each year of intervention implementation (approx. *n* = 160, 6--9 students per group). All students with parental consent/student assent who are present on the day will be invited to participate for approximately 30 min. We will ask students about their perceptions of the interventions, including satisfaction. To maintain confidentiality, no identifying details will be collected. Focus groups will be conducted at the end of each the school years by research staff trained in facilitating focus groups (including previous work with children). Finally, we will invite members of the 3-PLT to participate in focus groups and provide feedback regarding barriers and facilitators to implementation, and suggestions for how the implementation process can be improved.
#### Teacher surveys {#FPar2}
Each spring (April--May), we will open an online survey to examine school climate from teachers' perspectives. The survey will consist of items drawn from the teacher and staff version of the Delaware School Climate Survey (Bear, Gaskins, Blank, & Chen, [@CR9]), which covers perceptions of safety, student behavior, and overall climate. In the original student school climate survey, researchers conducted a multigroup confirmatory factor analysis and identified multiple specific factors associated with school climate, including teacher-student relationships, school safety and liking school, using this brief, psychometrically sound scale (Bear et al., [@CR9]). In addition, we will adapt violence and mental health items from the MiPHY to reflect teacher perceptions of the prevalence and severity of violent behaviors and mental health problems among their student populations. All teachers will be notified of the survey during a staff meeting. Reminders will be sent by the 3-PLT until a 75% response rate has been achieved.
#### Mechanisms {#FPar3}
We will assess potential mechanisms by which participation in the intervention influences secondary student outcomes including emotional functioning (anxiety and depression symptoms, wellbeing) using items from the MiPHY survey (Michigan Department of Education, [@CR66]). Specifically, we will examine if school climate mediates the relationship between intervention participation and school climate; school climate will be assessed using items from the Delaware survey (Bear et al., [@CR9]). School climate serves as a critical intermediary between the intervention components and student mental health and safety experience. Collectively, the interventions are thought to improve positive perceptions of school climate across constituents -- students, teachers, school staff and administrators. Students and teachers often have different perceptions of the same objective experiences. Although there is a clear link between student climate perceptions and positive student outcomes (Esposito, [@CR25]; Haynes, Emmons, & Ben-Avie, [@CR34]), there may be indirect influences on student outcomes through *teacher* perceptions of climate. Teachers who perceive poorer school climate report higher job-related stress and poorer teaching self-efficacy (Collie, Shapka, & Perry, [@CR19]), which in turn can influence student outcomes. We assess climate holistically to understand how stakeholders across the school may influence ultimate student outcomes.
#### Cost estimates {#FPar4}
We will estimate the resources required for the intervention and its implementation and estimate the net costs using a resource-based costing approach. We will track activities relevant to site replication or implementation costs, versus intervention development and research costs because only the former would be required of other sites who might adopt the intervention program (Hurley & Matthews, [@CR38]; Neumann, Sanders, Russell, Siegel, & Ganiats, [@CR71]). Specifically, we consider implementation costs to be those costs associated with replication that would be required by those adopting the program (Neumann et al., [@CR71]). This would include components of the intervention that require tailoring for replication, recruitment or promotion costs to engage practitioners and participants, and training, supplies and labor costs for evaluation as indicated (Ritzwoller, Sukhanova, Gaglio, & Glasgow, [@CR84]).
Analysis {#Sec30}
--------
Our primary hypothesis is that students in schools receiving the intervention will report less violence (over time) compared to students in the control group. We also hypothesize that the intervention schools have an overall lower rate of violence over the school year compared with control schools. We expect that using implementation facilitation led by a climate specialist as part of a 3-person leadership team is a feasible and sustainable approach to integrating and effectively delivering the three intervention components. Finally, we expect that students in the intervention schools will report more positive school climate compared with those in control schools and that this, in turn, will result in lower improved mental health outcomes.
A cluster randomization design is proposed with random assignment by school to one of two arms: intervention or control. Evaluation of effectiveness outcomes will involve testing at four time points: a fall of Year 1, spring of Year 1, fall of Year 2, spring of Year 2. Measures will include school records and self-report data from students and teachers. The survey of students makes use of currently collected MiPHY data. Ten schools will be recruited in the first year and 10 in the following year; staggered recruitment will allow some initial evaluation testing of the program and allow the earliest possible start time with a number of schools. Including 20 schools in the research provides an opportunity to examine school level differences. To adjust for bias introduced by multiple comparisons across our outcome variables, we will use a pair of conservative (Tukey and Bonferonni) *p*-value adjustments in addition to reporting direct effect sizes (e.g., group differences in means) to allow for assessment not dependent on standard null hypothesis testing.
Power analyses {#Sec31}
--------------
### Between group analyses {#Sec32}
Since the proposed intervention will be delivered within group (school) settings, data analyses examining treatment effects at the school level must account for the correlation among observations within school (i.e., intraclass correlations; ICC) due to shared factors within a school setting such as school selection factors (e.g., school attendance boundaries, family choice), similarities in treatment experiences (e.g., staff factors, resource factors), and within-school influences (e.g., common experiences with other students, cultural norms). Previous analyses of measures to be used in this study indicate small ICCs within schools for our outcome variables of interest (range: .00 to .03). If an ICC greater than zero is ignored, the outcome variance due to between-school differences (which can be large) is mixed with outcome variance due to between-participants variability within schools (which can be small because of the correlation). This can lead to large standard deviations, larger *p* values, and false-negative results (Chuang, Hripcsak, & Heitjan, [@CR18]; Murray, Phillips, Bimbaum, & Lytle, [@CR70]). Thus, our analytic strategy described below takes the ICC into account. Assuming 20 schools, 60 students per school, and a modest within-cluster correlation coefficient (ρ = .10), we will be able to detect standardized effect sizes at the school level of *d* = .27, representing a medium effect size, with probability (β \> .80). Thus, our design will include 20 schools over the course of 2 years of the study period. The lagged implementation, however, limits across school comparisons (i.e., treatment vs control schools) until year three.
### Individual analyses {#Sec33}
At the individual level, we will examine the hypothesized pathways using structural equation models while controlling for dependency due to school attended as noted above. Across Genesee County, there are an average of 180 4th--6th Graders in Flint schools resulting in a maximum sample size of 3600 students over the course of the study (we use a 33% consent/retention rate for calculations). Using the SEM power calculations suggested by MacCallum, Browne, and Sugawara ([@CR60]), we will have ample statistical power in SEM analyses for even our most complex models with 100 degrees of freedom (*df*). Power is based on the probability of rejecting the hypothesis of not a close fit with the data when the true model fit is excellent (the most stringent test). Our power with the 1200 (600 per condition) students is over .80. Notably, MacCallum et al. ([@CR60])^.^ show that power of .80 is achievable with alpha = .05 and *df* of 100 with a sample as small as 178 for test of not-close fit.
### Behavioral outcome analysis {#Sec34}
We will undertake comparability of the two arms (intervention and control) will be undertaken, including school size and socio-demographic indicators (e.g., proportion of free/reduced school lunches, proportion of students from various ethnic backgrounds, aggregate reports of income). Our primary analytic strategy uses mixed-effects models with full information maximum likelihood, which uses all available data for point estimation and sandwich estimator for the standard errors.
#### Data linking {#FPar5}
Survey data will be linked across intervention Years 1 and 2 for each Cohort. In order to maintain anonymous management of data, GISD will generate individual codes in the de-identified data. The impact of each year of the intervention will be assessed as well as change after 2 years of implementation. Rarely is the stability of an intervention examined, particularly one that is designed to affect change in school climate. The two-year implementation of the intervention provides a unique opportunity to assess the intervention over a longer period.
#### Dichotomous outcome variates {#FPar6}
The outcome measure required is the net average change, i.e. the proportion changing from negative at baseline to positive at follow-up minus the proportion changing from positive at baseline to negative at follow-up. This can be modeled in multinomial form with categories---upward change (from negative for the behavior to positive for the behavior), downward change (the reverse), with no change as reference category---and analyzed using multiple logistic regression for a cluster design by e.g., the GENMOD procedure in SAS.
As a test of the conceptual framework (secondary aim), mediating variables would be included in these analyses (e.g., of school climate on mental health). Change in certain behavioral variables, including composite variables, will be correlated by a multivariate extension of the above techniques to examine the relationship between school climate, mental health, and violence.
#### Quantitative variates (e.g., violence; climate; mental health) {#FPar7}
Essentially the same analytic structure as described above for dichotomous variates will be employed, but without the logit link for linear variates (it may be necessary to transform these quantitative variates, depending on their distributions.)
### Implementation outcomes {#Sec35}
We will assess feasibility and potential utility of using implementation facilitation through focus groups. A minimum of two study team members will take detailed field notes during focus group sessions. We will use the field notes to develop a broad understanding of content as it relates to the project's specific aims and to identify topics of discussion and observation. During this and subsequent steps, we will document initial impressions of topics and themes and their relationships to each other to define the boundaries of specific codes (e.g., the inclusion or exclusion criteria for assigning a specific code; (Miles, Huberman, & Saldaña, [@CR68]) The empirical material from the focus groups will be coded by project team members to condense the data into analyzable units. Segments of text will be assigned codes based on a priori or emergent themes, also known as open coding (Strauss & Corbin, [@CR90]). Codes will also be assigned to describe connections between categories (e.g., barriers and feasibility). Lists of codes developed by each investigator will be matched and integrated into a single codebook. We will use these codes to examine the association between different a priori and emergent categories. Through the process of continually comparing these categories, the different categories will be further condensed into broad themes. We will also evaluate outcomes using descriptive analyses from teacher surveys and a mixed methods approach to develop a comprehensive understand of using Facilitation to deliver the interventions. Results from each data set will be examined side-by-side to explore convergence (i.e., comparing analysis conclusions) to investigate if qualitative and quantitative results concur. We will also investigate how focus group results elaborate on quantitative results (expansion) to deepen our understanding of why and how Facilitation may or feasible for comprehensive school safety in low resource communities (Palinkas et al., [@CR74]).
### Process evaluation {#Sec36}
An intervention of this nature and size requires a comprehensive process evaluation to monitor implementation and sustainability. Specifically, we will focus on barriers and facilitators to implementation. We will investigate barriers and facilitators associated with the context (e.g., implementation climate), innovation (e.g., complexity), providers (e.g., confidence, competence) and recipients (i.e., student needs). We will examine key factors through focus groups with school staff, observations and review of 3-PLT meetings and activities.
### Cost-benefit analysis {#Sec37}
Following findings of reduced violence, we will examine the costs and benefits of the intervention. The program will be compared with the alternative intervention of no program (control). Although the intervention is targeted to reducing violence, it will likely have additional effects; this adds complexity to the analysis. We will focus on benefits associated with reduced violence, based on previously used methodology to obtain costs (Foster & Jones, [@CR27]; Kuklinski, Briney, Hawkins, & Catalano, [@CR56]). This strategy considers the proportion of school violence from national costs estimated for youth violence and related early school separation. School absence costs will also be considered, relative to costs associated with processing school truancy and absences associated with fear of attending school. We will compare these benefits with costs associated with implementing the full program.
Planned dissemination {#Sec38}
---------------------
We have existing mechanisms for dissemination of study results to both research and practice communities. This includes a Center that is part of the CDC-funded network of National Centers of Excellence in Youth Violence Prevention, which may serve as an additional venue for disseminating results to researchers and practice professionals. Study team members also house a Bureau of Justice Assistance training and technical assistance programs for Project Safe Neighborhoods, the Violence Reduction Assessment Tool, and the Innovations Suite Research Practitioner Fellows Academy that provide outstanding mechanisms for the dissemination of findings to teams of practitioners and researchers across the US. Finally, we will disseminate results and lessons learned to practitioners (including school administrators and resource officers) at conferences such as Michigan's SEPLA conference (<http://www.seplainstitute.org/index.htm>) and the conferences of the Michigan Chiefs of Police and Sheriffs Associations.
In addition, we will make information about this research available to a broader audience including the GISD community (students, families, staff, and local residents); local, state, and national policy makers; and the general public. Upon study completion, we will feed back results to the GISD community via newsletters, town halls, and presentations to school administrators, district officials, and the Genesee Intermediate Board of Education. We will make this information available for a common NIJ-sponsored school safety website.
Discussion {#Sec39}
==========
This study will make important contributions in the areas of implementation science, prevention research, violence prevention and the health and well-being of youth in disadvantaged communities. Although implementation science has made notable gains in the translation of EBIs many areas of public health and clinical science, including mental health treatment and cancer prevention, implementation is understudied in violence prevention and school-based EBIs. In addition, this study makes an important contribution to the application of implementation strategies to reduce health disparities through the effective delivery of EBIs in disadvantaged communities. This study is one of the first of its kind to investigate the feasibility of applying an implementation facilitation approach to deliver a comprehensive, multicomponent violence prevention intervention in schools.
The current study will evaluate the effectiveness of a comprehensive intervention to improve three key areas that affect school safety: violence, student mental health, and climate. Results from both the outcomes and process evaluations are expected to have significant implications for criminal justice policy and practice. In particular, this study emphasizes a proactive approach with early intervention, intended to prevent students from negative outcomes such as delinquency, violence, or psychiatric disorders later in life. It also emphasizes a community-based approach (including links with local health services) and a proactive and positive supportive approach (e.g., team-based supervision, restorative justice) over exclusionary discipline, in order to avoid a school to prison pipeline. This study will add to the literature by assessing both Mental Health First Aid and Restorative Justice among relatively young students, including new information about barriers to implementation for approaches that are not well studied in terms of their fidelity. The proposed study will also address less commonly considered components of CPTED, which may provide schools with a lower cost alternative to security cameras. It expands the role of school safety officers, who will serve not just as a visible police presence at school, but as key members of a leadership team that is integrated into the school community. This leadership team in particular is intended as a model for sustainable interventions that focus on building school capacity to improve safety. Finally, the study sample involves high need youths and schools reflective of disadvantaged communities subject to significant safety concerns and at-risk for the school to prison pipeline, absent interventions such as those proposed. This study aims to create a comprehensive, well-integrated model intervention that may be implemented in similar high-risk settings.
The current study will also add to our understanding of implementing multicomponent interventions for school safety in terms of logistical, social and material resource challenges. When delivering multiple interventions simultaneously, some may be more (or less) straightforward to integrate, and/or feasible and acceptable. This is especially critical for schools located in communities serving youth at high risk of poor health outcomes with significant constraints on time and resources. Consequently, this research can provide valuable information about options for schools to integrate these interventions (e.g., simultaneously, staggered), and secure needed social and material resources (e.g., substitute teachers for trainings) for successful implementation. This project will also advance community-research partnerships for school safety and in implementation science. The selection of interventions and implementation strategy represent a collaboration between the research partners, the intermediate school districts and the individual schools. Finally, this project seeks to support sustainability through involving school and district partners in each step of the process, adding personnel in the schools to coordinate implementation efforts and building capacity of school staff to conduct training and support intervention delivery over the long-term.
3-PLT
: 3-Person Leadership Team
CDC
: Center for Disease Control and Prevention
CPTED
: Crime Prevention Through Environmental Design
CS
: Climate Specialist
EBI
: Evidence-Based Interventions
Enhanced REP
: Enhanced Replicating Effective Programs
GENMOD
: Statistical procedure used for solving generalized linear equations
GISD
: Genesee County Intermediate School District
iPARiHS
: Integrated-Promoting Action on Research Implementation in Health Services
MHFA
: Mental Health First Aid
MIBLISI
: Michigan's Integrated Behavior and Learning Support Initiative
MiPHY
: Michigan Profile for Health Youth
NIJ
: National Institute of Justice
PBIS
: Positive Behavioral Interventions & Support
RJ
: Restorative Justice
SAMSHA
: Substance Abuse and Mental Health Services Administration
SAS
: Statistical Analysis System
SEPLA
: Schools, Educators, Police, Liaison Association
SIS
: School Information System
SRO
: School Resource Officer
**Publisher's Note**
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
We thank James Yake from the Genesee Intermediate School District for his assistance with clarifying community-focused study components.
EM, CM and JH were responsible for the study conception and design. AE was responsible for the drafting of the manuscript. AE, AK, JH, SF, EM and CM participated in developing the final text. All authors read and approved the final manuscript.
This project is funded by the National Institute of Justice, grant 2015-CK-BX-0017.
De-identified quantitative survey data, notes from qualitative interviews and transcripts from focus-group data will be provided to the National Archive of Criminal Justice Data (NACJD) at ICPSR at the University of Michigan within 2 years after the completion of the project. The NACJD will make the final determination of whether or not the data are suitable for public release, as well as the conditions associated with the release of the data if it is deemed appropriate for public use. Requests for the data can be made through their online data portal found at: <https://www.icpsr.umich.edu/icpsrweb/content/NACJD/index.html>.
This study was submitted to the Michigan State University Institutional Review Board and approved IRB\# × 15-1129e Category: EXEMPT 1. All adult participants provided informed consent, while youth provided their assent.
Not applicable.
The authors declare that they have no competing interests.
| {
"pile_set_name": "PubMed Central"
} |
1.. Introduction
================
In Sri Lanka, alcohol use is a growing public health issue \[[@b1-ijerph-06-02408]--[@b3-ijerph-06-02408]\], and, because it is a predominantly male habit there, alcohol use disproportionately harms Sri Lankan males. Adolescents and young adults in the country are highly vulnerable to the onset and continuation of the habit \[[@b4-ijerph-06-02408]\]. Beer is probably the most popular alcoholic beverage among Sri Lankan youth, and a significant surge in beer sales has been observed over the last few years \[[@b5-ijerph-06-02408]--[@b7-ijerph-06-02408]\]. In 2003, 19.3 million liters of beer were produced in the country and the corresponding figure for the year 2007 was 29.5 million liters, a 53 percent increase. Socio-cultural factors, such as urbanization and westernization and environmental factors, such as availability and affordability, may have contributed to this upward trend in the sale of alcohol.
The consumption patterns of alcohol by different age groups and the social-cognitive determinants of alcohol use, such as drinking motives, have not been clearly understood in the context of Sri Lanka. This is partly due to the widespread production of illegal liquor in a country where no records of the quantity and sale of illegal liquor are kept \[[@b8-ijerph-06-02408]\]. Another partial reason is the shortage of epidemiological and psycho-social data pertaining to the problem. Preventive programs targeted at young people should be based on a scientific knowledge of the distribution and determinants of alcohol use if they are to be effective in curbing the problem. Thus, there is a significant need to understand how and why young people in Sri Lanka use alcohol.
The reasons and motives for drinking alcohol are closely associated to the drinking patterns and consequences of alcohol use \[[@b9-ijerph-06-02408],[@b10-ijerph-06-02408]\]. Different age groups may develop different motivations towards alcohol use that are shaped by many factors, including culture and environment \[[@b11-ijerph-06-02408],[@b12-ijerph-06-02408]\]. Thus, motivation is considered a key concept in behavioral and psycho-social models of alcohol use. Since different drinking motives are associated with different types of drinking behavior \[[@b9-ijerph-06-02408],[@b10-ijerph-06-02408]\] and culture plays a significant role in motivating or de-motivating people toward various behaviors, proper understanding of motives that direct young people to drink would help public health and education authorities formulate effective public health policies and develop cost-effective measures to curb the alcohol problem.
Researchers have identified a number of domains of drinking motives \[[@b9-ijerph-06-02408],[@b10-ijerph-06-02408]\], and among them, personal enjoyment, social pressure, and tension reduction have been identified as prominent. Further, each domain is shown to be related to different aspects of alcohol use. For example, the personal enjoyment (enhancement) motive was found to be associated with heavy drinking, whereas social motives were identified to be associated with lighter drinking patterns and were more prevalent among young alcohol users \[[@b13-ijerph-06-02408],[@b14-ijerph-06-02408]\]. Tension-reduction (anxiety reduction) motives were found to be related to solitary drinking and to problem drinking \[[@b13-ijerph-06-02408],[@b15-ijerph-06-02408]\]. However, most of these studies have been conducted in the West, limiting the understanding of how these relationships work in non-western cultures. In the Sri Lankan context, there are certain culture-specific motivational factors of drinking behavior. Young males in Sri Lanka drink to become more prominent among peers and to attract females in social gatherings. Since the prevalence of alcohol use in the country is very low among women compared to men (5% verses 53%), alcohol use often symbolizes manhood, and thus, drinking behaviors are occasionally used by males to dominate family members and neighbors. A new scale was developed based on the existing drinking motives questionnaires \[[@b9-ijerph-06-02408],[@b10-ijerph-06-02408],[@b13-ijerph-06-02408]\] that incorporates some of these culture-specific motivational factors.
To date and to our knowledge, no studies have been conducted on the motivation toward alcohol use among people in Sri Lanka. This study's purpose was to test the 3-factor structure of a scale developed to measure drinking motives and to examine the relationships between drinking motives and drinking patterns in a sample of males aged 16--30 years in Sri Lanka. It was hypothesized that the drinking motives scale consist of three factors: personal enjoyment, social pressure, and tension reduction. It was also hypothesized that drinking motives were related to drinking frequency.
2.. Methods
===========
A cross-sectional survey design was used. An anonymous, self-administered questionnaire was employed to collect data. Boards for the protection of human subjects from universities in both the United States and Sri Lanka approved the study protocol.
2.1.. Sampling
--------------
Purposive sampling method was used to select sample subjects. Efforts were made to obtain an equal number of adolescents (16--19 years of age) and young adults (20--30 years of age) for the sample, so that the sample would represent the young males in the study population. Men aged 16--30 years, who were in the streets, universities and other educational institutes, in work places, or at homes during the survey period and who identified themselves as either occasional or regular drinkers, were invited to participate in the survey after explaining to them briefly the purpose of the survey and the anonymous nature of the data collection method. Those who agreed to participate in the study were surveyed using the questionnaire.
The survey was conducted in two settings: urban and semi-urban. Three medical undergraduate students were recruited and trained as data collectors. The people in the area respect and have a high confidence in those working or studying in the medical field, thus recruitment of medical students as data collectors increased the response rate. Only about 5% of the subjects who were asked to participate refused to do so. Data were collected for a period of two weeks in January 2007 and a total of 448 subjects participated in the study.
2.2.. The Instrument
--------------------
The survey instrument consisted of several variables: age, drinking frequency, and 20 questions on drinking motives. Drinking frequency of the participants was obtained using the item *I use alcohol* followed by four response categories: *daily*, *2--3 times a week*, *2--3 times a month*, and *2--3 times a year.* A literature survey was done and expert opinions were sought to identify motives towards alcohol use in young people in Sri Lanka \[[@b10-ijerph-06-02408],[@b16-ijerph-06-02408]--[@b19-ijerph-06-02408]\]. The 3-dimensional drinking motives scale developed by Cooper, Russell, Skinner, and Windle (1992) was used as the base in the development of our new scale. In addition, ideas and opinions expressed on drinking motives by more than 100 young males, by whom the first author came into contact during his health education and prevention activities in the community, were also considered when selecting culturally-appropriate items for the scale. The scale developed by Cooper and colleagues (1992) has a total of 15 items and the new scale has 20 items. Some items from the scale developed by Cooper and colleagues, such as *I use alcohol because it helps me to forget my worries*, were included in the new scale. New culturally-appropriate items, such as *I drink alcohol because it will enhance my creative ability* and *I drink alcohol because it helps me to control others*, were added to the new scale. The new scale varies from the contemporary drinking motive scales that have been developed and used in the West.
A jury of experts consisting of a psychiatrist, two social scientists, and a psychologist examined the items for content validity. Based on their evaluation, the final scale of motive towards alcohol use was constructed using 20 rating items (see [Table 1](#t1-ijerph-06-02408){ref-type="table"} for scale items). Four Personal Enjoyment (PE) items, 10 Tension-Reduction (TR) items, and six Social Pressure (SP) items were included in the scale. For example, *I drink alcohol because I like the taste* was a PE item, and *I drink alcohol because it is customary for men on special occasions* was a SP item. There were four response categories for each of these 20 items which were scored as follows: *to a greater extent* equals 3*; to some extent* equals 2; *to little extent* equals 1; and *not relevant* equals 0. The internal consistencies of the three factors using Cronbach's alpha methods were as follows: alpha~PE~ = 0.48; alpha~TR~ = 0.74; and alpha~SP~ = 0.62. To assess the impact of each factor on drinking habits, each item in each of the three subscales were added up to obtain 3 subscale scores.
2.3.. Data Analysis
-------------------
The collected data were checked for consistency. Descriptive and bivariate analyses of the data set were done using SPSS 15.0 \[[@b20-ijerph-06-02408]\]. Confirmatory Factor Analysis (CFA) of the 20 item motives scale toward alcohol use was conducted using Lisrel 8.80 \[[@b21-ijerph-06-02408]\].
3.. Results
===========
After cleaning and consistency checking, analysis was done using 412 sample subjects. The age of the participants ranged from 16--30 years (*M = 20.27*, *SD = 3.63*). Forty-eight participants (11.7%) were married. Of the total, 102 (24.8%) had a job at the time of the survey and the others were either students or unemployed. Nine participants (2.2%) were daily users, 33 (8%) use alcohol 2--3 times a week, 76 (18.4%) 2--3 times a month and the remaining 294 (71.4%) 2--3 times a year. Means, standard deviations, and correlations among subscales and total scale are provided in [Table 1](#t1-ijerph-06-02408){ref-type="table"}.
Correlations among subscales and total scale ranged from 0.28 to 0.90 (*p \< 0.01*). The strongest association was found between tension reduction and total score, and a relatively weaker association was found between personal enjoyment and social pressure.
3.1.. Confirmatory Factor Analysis (CFA)
----------------------------------------
The 3-factor structure of the scale *motivations towards alcohol use* was tested using Lisrel 8.80. All indicators in the model had loadings above 0.40 and were significant at the *p \< 0.01* level ([Table 2](#t2-ijerph-06-02408){ref-type="table"}). Loadings greater than 0.40 are generally considered as acceptable \[[@b22-ijerph-06-02408]\]. The overall fit of the model was acceptable: χ^2^ (167, *N* = 412) = 323.16; χ^2^/*df* ratio = 1.93:1; CFI = 0.98; and the RMSEA = 0.048 \[[@b23-ijerph-06-02408]\]. Thus, in this sample of alcohol users, motivations towards alcohol use can be divided into 3 factors: personal enjoyment motives, social pressure motives, and tension-reduction motives.
3.2.. Relationships between Motivations and Drinking Habits
-----------------------------------------------------------
To test the second hypothesis, associations between the 3 motivational factors and drinking frequency were examined using multiple regression. This analytical method was chosen so that the effects of the drinking motive dimensions could be mutually adjusted. The results are presented in [Table 3](#t3-ijerph-06-02408){ref-type="table"}.
As can be seen from the individual beta weights, only tension reduction predicts drinking frequency. Thus, it is reasonable to infer that young males in Sri Lanka are less likely to be motivated to use alcohol because of its enhancement effect or because of the social pressure exerted by peer or social groups to use alcohol.
4.. Discussion
==============
The present study aimed to develop and test a 3-factor measurement of drinking motives. It also explored whether drinking motives were related to young males' drinking frequency and, if so, which drinking motives were. As expected, according to CFA, the 3-factor model on drinking motives (i.e., based upon personal enjoyment, tension-reduction, and social pressure) complemented the data on this sample of male alcohol users aged 16--30 in southern Sri Lanka. Although similar instruments used in western countries \[[@b10-ijerph-06-02408],[@b16-ijerph-06-02408],[@b18-ijerph-06-02408],[@b19-ijerph-06-02408]\] were employed in the development of the scale, the importance of the cultural embedding of drinking motives were also considered. Thus, it is safe to recommend that the scale can be used, with modifications if necessary, to examine drinking motives in other population groups across the country. However, future research should also consider use of drinking motives scales that are cross-culturally validated to compare results across countries. To our understanding, this study is the first to report on drinking motives of young males in Sri Lanka, a middle-income country in South Asia. Thus, undoubtedly, further research is needed to refine the scale and confirm the results.
Results of the study suggest that drinking to reduce tension seemed to be the most emphatic motive of alcohol use in this sample of young males. Stress, suicidal ideation, and suicide rates among young Sri Lankan people are ranked among the highest in the World \[[@b24-ijerph-06-02408],[@b25-ijerph-06-02408]\], and the general belief is that alcohol helps to relieve distress. That climate may have motivated tension reduction via alcohol to become more prevalent in this young population. This warrants further research in the field.
In studies conducted in other countries, social motives were found to be associated with moderate alcohol use, and personal enjoyment (enhancement) motives with heavy drinking \[[@b15-ijerph-06-02408],[@b16-ijerph-06-02408],[@b26-ijerph-06-02408]\]. In this study, the quantity of alcohol used by the participants was not determined, so it was not possible to discriminate between moderate and heavy alcohol users in this sample. However, as has been demonstrated in other studies \[[@b3-ijerph-06-02408],[@b27-ijerph-06-02408]\], most of the respondents in our sample should be either occasional or moderate drinkers. Heavy drinkers are generally found in the middle and older age groups in the context of people in Sri Lanka. Personal enjoyment and social pressure motives were found to be weaker motives of alcohol use compared to the tension-reduction motive in this study population. This may suggest that in countries like Sri Lanka, where the norm is abstinence from any type of alcoholic beverage, a relatively lower number of users consider alcohol as a pleasant drink. In fact, alcohol does not have a so called "rich, good taste," and it is only a myth constructed and propagated by the alcohol industry \[[@b1-ijerph-06-02408]\]. Studies have shown that those who use alcohol to enhance their positive affect status often use alcohol alone and in settings such as bars \[[@b13-ijerph-06-02408],[@b16-ijerph-06-02408]\]. Results of this study are consistent with this finding because drinking at bars or in college settings by young men is very rare behavior in Sri Lanka.
The cross-sectional design does not allow for any inference of a cause-effect relationship. Thus, relationships between motives and drinking patterns are better tested with longitudinal data. The use of purposive, non-probability sampling procedure may have limited the generalizability of the findings. However, it should be noted that in countries like Sri Lanka, where alcohol use is considered as "anti-social" behavior, especially by young people, refusing to participate and underreporting drinking behaviors would become major threats to selecting a representative sample if we were to select a random sample from households. Although self-report measures would raise questions of reliability and validity, the use of medical undergraduates in data collection and collecting data in non-threatening atmospheres, such as on the streets in most occasions, reduced this problem tremendously. It is also likely that some items in the scale may have related to more than one motivational factor, and items not included in this scale do have an impact on drinking motives. Further, alcohol habits should ideally be investigated using the type, quantity, and frequency of use. Future research needs to address these methodological and measurement issues.
5.. Conclusions
===============
The study advocates a 3-factor structure for a scale of motivations towards alcohol use, the scale developed and used for this study. However, the scale needs to be validated using a larger representative sample. Tension-reduction motivations seemed to be the most important social-cognitive factor in young Sri Lankan males' drinking behavior. These findings have several implications for public health research and interventions. There is need for a continued focus on individual tension-reduction reasons for drinking in adolescents and young adults in substance use prevention programs. Alcohol motives are likely to have been shaped by other indirect and distal forces such as availability and the media. For example, television scenes that glamorize its use or incorporate strong symbolic meaning, such as rebellion against prejudice, while featuring alcohol may cause young people to initiate and continue drinking. Research that offers a better understanding of psycho-social and environmental factors associated with alcohol use behavior among the younger population in Sri Lanka is urgently needed.
######
Subscales and total scale of motivations towards alcohol use: Correlations, means and standard deviations.
**Motives** **1** **2** **3** ***M*** ***SD***
--------------------------- ------- ------- ------- --------- ----------
**1. Personal Enjoyment** -- -- -- 2.10 1.99
**2. Tension Reduction** 0.43 -- -- 4.26 4.59
**3. Social Pressure** 0.28 0.45 -- 2.54 2.78
**Total Score** 0.64 0.90 0.72 8.91 7.43
######
Scale items and factor loading of the 3-factor model of motivations towards drinking.
**Item** **Factor Loading[\*](#tfn1-ijerph-06-02408){ref-type="table-fn"}**
--------------------------------------------------- --------------------------------------------------------------------
I use alcohol because,
*Social Pressure*
my friends drink 0.61
it is difficult to refuse 0.46
other people are drinking 0.77
it will enhance my creative ability 0.51
it is customary for men on special occasions I 0.59
want to be prominent 0.67
*Personal Enjoyment*
I like the taste 0.62
it makes me feel good 0.71
I get thirsty 0.66
it goes well with the meals 0.41
*Tension Reduction*
it helps me to relax 0.58
it would ease me when I get blamed 0.60
it helps me to sleep 0.61
it helps me to forget my worries 0.70
it helps me to get rid of restlessness and tense 0.61
it helps me to cheer up when I get dull or boring 0.67
it gives me energy 0.65
it is a habit 0.45
it helps me to face difficulties with confidence 0.72
it helps me to control others 0.44
All items were loaded significantly on their respective factors, *p \< 0 .01.*
######
Multiple regression analysis predicting drinking frequency from 3 motives towards alcohol use.
**Drinking Motives** **B** ***SE*B** **β**
------------------------ ------- ----------- -----------------------------------------------------------
**Constant** 1.12 0.06 --
**Personal Enjoyment** 0.04 0.02 0.097
**Tension-reduction** 0.05 0.01 0.311**[\*](#tfn3-ijerph-06-02408){ref-type="table-fn"}**
**Social Pressure** 0.01 0.01 0.001
R^2^ = 0.13 (*ps* \< 0.001),
*p* \< 0.001.
| {
"pile_set_name": "PubMed Central"
} |
1. Introduction {#sec1}
===============
Lung cancer remains a major health problem worldwide. In 2012 lung cancer was the most commonly diagnosed cancer worldwide making up 13.0% of the total incidence of cancer. It was also the most common cause of death from cancer worldwide, accounting for nearly one in five cancer deaths (19.4% of the total) \[[@B1]\]. Lung cancer is clinically divided into two main groups based upon the size and appearance of malignant cells: small cell lung cancer (SCLC) (16.8%) and non-small cell lung cancer (NSCLC) (80.4%) \[[@B2]\]. The most effective option for treatment of lung cancer is surgical resection, when feasible \[[@B3]\]. However, majority of patients are diagnosed at an advanced or metastatic stage of disease in which case chemotherapy and/or concurrent administration of chemotherapy and radiation is the most beneficial form of treatment \[[@B4]\]. Nevertheless, even with treatment, the 5-year survival rate in patients is only 16.6% \[[@B5]\], with poor survival rates mainly being attributed to late stage diagnosis and high frequency of drug resistance. Obtaining a better understanding regarding the molecular mechanisms involved in lung carcinogenesis is of utmost importance in the aim to identify the diagnostic and prognostic markers for early detection and targeted treatment of lung cancer.
Apoptosis plays an important role during development and in the maintenance of multicellular organisms through the removal of damaged, aged, or autoimmune cells \[[@B6]\]. The apoptotic process can be divided into the extrinsic and intrinsic pathway. Each pathway will ultimately result in the activation of cell death proteases, which in turn initiates a cascade of proteolysis involving effector caspases that carries out the completion of the apoptotic process \[[@B7]\]. In contrast to normal cells, cancer cells have the ability to evade apoptosis to promote cell survival under the conditions of environmental stress. There are a number of mechanisms by which cancer cells are able to suppress apoptosis. For example, the tumor suppressor gene*p53* is a widely mutated gene in human tumorigenesis \[[@B8]\].*p53* mutation will inhibit the activation of DNA repair proteins leading to a decrease in the initiation of apoptosis \[[@B7]\], allowing for cells to divide and grow uncontrollably, forming malignant tumors. Furthermore, cancer cells are able to disrupt the balance between pro- (*BCL-2*,*BCL-XL*) and antiapoptotic factors (*BAX*,*BIM*, and*PUMA*) \[[@B9]\]. Increased expression of proapoptotic Bcl-2 protein contributes not only to the development of cancer but also to resistance against a wide variety of anticancer agents, such as cisplatin (DDP) and paclitaxel \[[@B10]--[@B12]\].
MicroRNAs (miRNAs) are a subset of noncoding RNAs of about 20 to 25 nucleotides long which posttranscriptionally regulate gene expression via inhibition of mRNA translation, by binding to specific target sites in their 3′-untranslated region (3′UTR), or inducing degradation of target mRNA through cleavage \[[@B13]\]. An individual miRNA is able to modulate the expression of multiple genes; correspondingly, a single target can be modulated by many miRNAs \[[@B14]\]. MiRNAs were reported to be involved in a vast range of biological processes, including apoptosis (see [Figure 1](#fig1){ref-type="fig"}) \[[@B15]--[@B22]\]. As miRNAs play a key role in an assortment of biological processes, an altered miRNA expression is likely to contribute to human diseases including cancer \[[@B23]\]. Previous studies have shown that compared to normal tissues, malignant tumors and tumor cell lines were found to have widespread deregulated miRNA expression \[[@B24]--[@B28]\]. MiRNAs are critical apoptosis regulators in tumorigenesis and cancer cells are able to manipulate miRNAs to regulate cell survival in oncogenesis. Many studies carried out in the past several years are aimed at elucidating the specific miRNAs associated with apoptosis in cancer and their related target genes. In this review we will examine the recent progress of research on miRNA-mediated regulation of apoptosis in lung cancer and its future therapeutic applications.
2. Antiapoptotic miRNAs {#sec2}
=======================
Evasion of apoptosis is a significant hallmark of tumor progression, and one mechanism by which miRNAs influence development of cancer is through the regulation of the apoptotic process as shown in various studies \[[@B29]--[@B32]\]. miRNA expression can be either upregulated or downregulated and evidence has shown that dysregulated miRNAs can behave as oncogenes or tumor suppressor genes in lung cancers \[[@B18], [@B28], [@B33]\]. Amplification of miRNAs can lead to the downregulation of tumor suppressors or other genes that are involved in apoptosis \[[@B34]\].
*miR-197*. For example, the expression of miR-197 is increased in cancer tissues in comparison to normal specimens. Fiori et al. (2014) demonstrated that knockdown of miR-197 in NIH-H460 and A549 cells promoted induction of apoptosis, evident by the observation of caspases 3--7 activation and increased apoptotic population by Annexin staining. Furthermore, the direct interaction of miR-197 with the 3′UTR of*BMF* and*NOXA*was demonstrated by the luciferase reporter assay \[[@B35]\]. When activated by intra- or extracellular stimuli, proapoptotic Bmf binds to and neutralizes antiapoptotic Bcl-2 family members on the mitochondrial membrane, thus allowing proapoptotic proteins Bak and Bax to dimerize and promote the release of cytochrome c, ultimately leading to cell death \[[@B36]\]. Noxa is a BH-3 only proapoptotic protein transcriptionally activated by*p53*. Collectively, miR-197 is able to act upon different levels of the*p53* pathway to counteract the induction of apoptosis, thus allowing cells to proliferate uncontrollably \[[@B35]\].
*miR-21*. miR-21 is found to be frequently upregulated in a number of cancers; however its potential role in tumorigenesis*in vivo* is not fully explored. Using transgenic mice with loss-of-function and gain-of-function miR-21 alleles, Hatley and colleagues elucidated the role of miR-21 in NSCLC pathogenesis*in vivo*\[[@B37]\]. It was determined that miR-21 regulates tumor proliferation and survival, which are two integral components of NSCLC pathogenesis, by targeting negative regulators of the RAS pathway as well as by targeting proapoptotic genes \[[@B37]\]. In regards to the apoptotic pathway, overexpression of miR-21*in vivo* leads to decreased protein levels of Apaf-1, an important component of the intrinsic mitochondrial apoptotic pathway, as well as decreased expression of FasL, a key initiator of the extrinsic apoptotic pathway. Furthermore,*RHOB*, with a tumor suppressor role, is a target of miR-21 and its dysregulation leads to an increase in cell growth and inhibition of apoptosis \[[@B38]\]. Together these results suggest that relieving miR-21 downregulation of these proapoptotic and tumor suppressor genes could provide a means to enhance the effect of current chemotherapy.
*miR-212*. Acetylcholinesterase (AChE), a component of the cholinergic system, has the ability to influence apoptotic sensitivity both*in vitro* and*in vivo*\[[@B39]--[@B41]\]. In NSCLC tissues AChE levels are low and are associated with tumor aggressiveness, increase risk of postoperative recurrence, and low survival rate \[[@B42]\]. Lu et al. (2013) determined that*AChE* expression in NSCLC was posttranscriptionally modulated by miR-212 binding to its 3′UTR. Interestingly, alterations in neither AChE nor miR-212 expression significantly affected cell survival; however it was observed that during DDP-induced apoptosis miR-212 levels were reduced with a concurrent increase in AChE protein levels. This suggests that miR-212 plays a role in DDP resistance by directly inhibiting AChE and preventing apoptosis. Therefore, interference against miR-212 may potentially be a means to improve the pharmacotoxicological profile of DDP in NSCLC \[[@B43]\].
*miR-17-5p and miR-20a.* The miR-17-92 cluster, which is composed of seven miRNAs and resides in intron 3 of the*C13orf25* gene at 13q31.3, is frequently overexpressed in lung cancers \[[@B44]\]. Matsubara et al. (2007) demonstrated that inhibition of two components of the miR-17-92 cluster, miR-17-5p, and miR-20a, with antisense oligonucleotides can induce apoptosis selectively in lung cancer cells that overexpress miR-17-92 \[[@B45]\]. Previously, miR-17-5p and miR-20a have been shown to directly target*E2F1* \[[@B46]\]; thus inhibition of these miRNAs may cause the induction of apoptosis in part through the induction of*E2F1* and subsequent cell cycle progression into*S* phase \[[@B47]\]. However additional studies would have to be carried out to determine the actual targets for the miR-17-92 cluster to gain a better understanding of the development of this cancer.
3. Proapoptotic miRNAs {#sec3}
======================
MiRNAs that are downregulated are considered tumor suppressor genes. Tumor suppressor miRNAs usually prevent tumor development by negatively regulating oncogenes and/or genes that control cell differentiation or apoptosis \[[@B48]\]. MiRNAs that act as tumor suppressors can be downregulated as a result of deletions, epigenetic silencing, or loss of expression of transcription factors (see [Table 2](#tab2){ref-type="table"}) \[[@B49]\].
3.1. B-Cell Lymphocyte 2 (*BCL-2*) Family Related miRNAs {#sec3.1}
--------------------------------------------------------
Members of the evolutionarily conserved*BCL-2* family are thought to be the central regulators of apoptosis. The expression level of*BCL-2* differs for different cell types; however high levels and aberrant patterns of*BCL-2* expression were reported in a wide variety of human cancers, including lung cancer \[[@B50]\]. Elevation of Bcl-2 protein expression contributes not only to the development of cancer but also to resistance against a wide variety of anticancer agents \[[@B10]--[@B12]\].
*miR-7*. Xiong et al. showed that miR-7 was downregulated in NSCLC cells and*BCL-2* was identified as a direct target \[[@B51]\]. Transfection of miR-7 in A549 cells led to a significant reduction in endogenous*BCL-2* mRNA and protein levels and correspondingly led to increase in the activities of caspase-3 and caspase-7 in cells with apoptotic nuclei \[[@B51]\]. These results thus provide evidence that*BCL-2* may be involved in miR-7 mediated apoptosis induction in A549 cells.
*miR-335*.*BCL-W*, another antiapoptotic member of the*BCL-2* family, was found to be a direct target of miR-335 \[[@B52]\]. miR-335 was downregulated in A549 and NCI-H1299 cells, and upregulation of this miRNA via transfection of miR-335 mimics led to a suppression of cell invasiveness and promotion of apoptosis. Furthermore Dyanan and Tjian (1983) discovered that miR-335 directly targeted*SP1* gene, a member of the family of Sp/Kruppel-like transfection factors \[[@B53]\], which can enhance the activity of promoters of numerous genes involved in cell proliferation, apoptosis, differentiation, cell cycle, progression, and oncogenesis thus regulating these genes\' expression \[[@B54]\].
*miR-608*. Studies in our lab identified a*BCL-XL*-induced miRNA, miR-608, involved in the regulation of cell death in A549 and SK-LU-1 cells \[[@B55]\].*BCL-XL*, a major prototype of the antiapoptotic*BCL-2* gene was found to be overexpressed in NSCLCs \[[@B56]\]. Silencing of*BCL-XL* in A549 and SK-LU-1 led to the significant dysregulation of a number of miRNAs, as determined through miRNA microarray, with miR-608 being the most upregulated miRNA. Upregulation of miR-608 in A549 and SK-LU-1 via miR-608 mimics led to an increase in apoptotic population, as determined by Annexin-V FITC apoptotic assay, in comparison to NP-69 cells (normal human nasopharyngeal epithelial cell line) (see [Table 1](#tab1){ref-type="table"}) \[[@B55]\]. Bioinformatics analysis determined that miR-608 may be associated with various signaling pathways, primarily the phosphatidylinositol 3-kinase/protein kinase B (PI3K/AKT), wingless-type MMTV integration site family (WNT), transforming growth factor (TGF-*β*), mitogen activated protein kinase (MAPK), and the intrinsic pathway. However the true targets of miR-608 and its direct effects on the apoptotic process is yet to be determined.
3.2. Protein Kinase C (PKC) Family Related miRNAs {#sec3.2}
-------------------------------------------------
PKC is a serine/threonine kinase that is involved in various signal transduction pathways including those related to cellular proliferation, differentiation, and apoptosis \[[@B57]--[@B59]\]. PKC plays a role in lung cancer and levels of PKC proteins were found to be increased in various cell lines (A549, NCI-H1355, NCI-H1703, NCI-H157, and NCI-H1155) in comparison to primary normal human bronchial epithelial cells (NHBE) \[[@B60]\].
*miR-203*. To determine the role that miR-203 can play in the influence of cellular function, putative target prediction was carried out and PKC-**α** was determined to be a target \[[@B61]\]. Luciferase reporter assay further revealed miRNA-203 direct binding of the 3′UTR of*PKC-*α** mRNA transcript. miR-203 negatively regulated proliferation and migration through the repression of*PKC-*α**, and miR-203 was also able to modulate cell apoptosis. However, siRNA silencing of*PKC-*α** resulted in a less significant apoptotic phenotype in comparison to that observed by miR-203 overexpression, thus suggesting that miR-203 may modulate multiple apoptotic genes that work together to regulate cell apoptosis \[[@B61]\]. Further studies must therefore be carried out to determine the additional apoptosis related targets of miR-203.
*miR-143*. miR-143 expression was reported to be downregulated in cancer tissues and inhibition of miR-143 promotes cell proliferation but hinders cell apoptosis. To determine the role that miR-143 plays in the apoptotic process, Akita (2002) investigated the possible targets of miR-143 and found that PKC-**ε**, a crucial enzyme in various cellular signaling pathways \[[@B62]\], was a putative target. Using the luciferase reporter assay it was determined that miR-143 specifically targets*PKC-*ε**, and overexpression of miR-143 increases the cell apoptosis in A549 cells \[[@B63]\]. PKC-*ε* was suggested to play a role in regulating the antiapoptotic signaling pathway through the upregulation of Bcl-2 with a concurrent suppression of proapoptotic Bid \[[@B64]--[@B66]\]. Furthermore, PKC-*ε* is able to activate Akt to apply its prosurvival effects \[[@B67], [@B68]\]. Therefore, the targeting of*PKC-*ε** could potentially be a valuable therapeutic strategy for lung cancer.
3.3. Other miRNAs {#sec3.3}
-----------------
*miR-198*. miR-198 is downregulated in NSCLC cell lines and overexpression of this miRNA inhibits cell viability and enhances apoptosis in A549 cells. Overexpression of miR-198 induces the expression of poly(ADP-ribose) polymerase (PARP) and of cleaved caspase-3. miR-198 was also able to inhibit growth of tumor grafts in nude mouse.*FGFR1*, a lung cancer oncogene, which is a membrane-bound receptor tyrosine kinase that regulates proliferation via the MAPK and PI3K pathway, much like EGFR, was found to be a direct target of miR-198 \[[@B69]\].
*miR-146a*. Expression of miR-146a is low in malignant tissues in comparison to corresponding adjacent normal lung tissues. Functionally, miR-146a suppresses cell growth, inhibits cell migration and increases cellular apoptosis \[[@B70]\]. Upregulation of miR-146a expression via miR-146a mimic transfection resulted in the downregulation of EGFR as well as phosphorylated EGFR, both at the mRNA and at protein levels. Furthermore, downstream pathways (ERK-1/2, AKT, and STAT) were also downregulated in response to miR-146a mimic transfection, albeit with a weaker effect as that seen by cells transfected with*EGFR* specific siRNA. miR-146a mimic also led to the decrease of phosphorylation of the NF-*κ*B inhibitor I*κ*B*α*, but not total I*κ*B*α*. Levels of phospho-NF*κ*B, total NF-*κ*B, and the total immune-modulating kinase, IRAK-1, were also found to be decreased following miR-146a mimic transfection, suggesting that miR-146a regulates NF-*κ*B and IRAK-1 signaling \[[@B70]\].
*miR-26a*. miR-26a expression is downregulated in lung cancer tissues relative to normal tissues. Transfection of miR-26a into A549 cells was able to decrease cell proliferation, block the G1/S phase transition of cell cycle, and induce apoptosis \[[@B71]\]. The chromatin regulator enzyme EZH2, which regulates survival and metastasis of cancer cells \[[@B72]\], was found to be a direct target of miR-26a. Downregulation of*EZH2* expression, caused by overexpression of miR-26a will transactivate downstream tumor suppressor genes*DAB21P* and*RUNX3*. DAB21P is a potent growth inhibitor that induces G0/G1 phase cell cycle arrest and could lead to apoptosis \[[@B73]\], while RUNX3 leads to cell cycle arrest, apoptosis, and significant decrease of tumor growth and abrogation of metastasis \[[@B74]\].
*miR-451*. Poor tumor differentiation, advance pathological state, lymph node metastasis, and poor prognosis are associated with downregulation of miR-451, which occurs in lung cancer \[[@B75]\]. To observe the functions of miR-451, Wang et al. (2011) upregulated miR-451 expression via mimics and observed suppressed*in vitro* proliferation, chromatin condensation and nuclear fragmentation upon 4′,6-diamidino-2-phenylindole (DAPI) staining, and significant caspase-3 activity. These results suggested that ectopic expression of miR-451 was able to induce an increase in apoptosis in a caspase-3 dependent manner. In addition, the*RAB14*gene was identified as a direct target of miR-451. Inhibition of*RAB14* led to a decrease in phosphorylation of Akt, which subsequently decreased levels of Bcl-2 protein expression and increased proapoptotic Bax or Bad protein expression. As the expression levels of RAB14 protein were inversely correlated with the expression levels of miR-451 in NSCLC tissues it was concluded that downregulation of RAB14 may be the mechanism by which miR-451 carries out its tumor suppressor functions \[[@B75]\].
*miR-192*. miR-192 was found to be downregulated in A549, NCI-H460, and 95D cell lines \[[@B76]\]. Cell viability was greatly decreased following miR-192 upregulation, while levels of apoptosis were elevated with induced expression of PARP protein and cleaved caspase-7, thus suggesting that miR-192 induces apoptosis through the caspase pathway. Using bioinformatics analysis,*RB1* gene was determined to be a putative target of miR-192 and luciferase reporter assays confirmed direct binding of miR-192 to the 3′-UTR of this gene \[[@B76]\]. Since RB1 plays a vital role in regulating cell apoptosis, its downregulation was shown to induce *γ*-H2AX foci formation, a marker of DNA damage, and to promote apoptosis in A549 cells \[[@B77]\].
4. miRNA and Response to Cancer Therapy {#sec4}
=======================================
Many cancer therapies available today aim to induce tumor-selective cell death; however resistance to chemotherapeutics is a significant obstacle to the long-term treatment and survival of NSCLC patients \[[@B78]\]. Presently, there are various chemotherapeutics that are being utilized in the treatment of lung cancer, including FDA approved drugs (DDP, paclitaxel, docetaxel, gemcitabine, and EGFR-TKIs), natural compounds (curcumin), and small organic compounds (PRIMA-1) (see [Table 3](#tab3){ref-type="table"}). The association of miRNAs as regulators of malignancy and apoptosis has been widely reported; thus it is reasonable to assume that miRNAs play significant roles in sensitivity/resistance to common cancer treatments (see [Figure 2](#fig2){ref-type="fig"}) \[[@B79]\]. Indeed, recent studies have demonstrated miRNAs as potential agents involved in the sensitivity of lung cancer cells to cytotoxic therapy.
4.1. Cisplatin- (DDP-) Related miRNAs {#sec4.1}
-------------------------------------
DDP is a platinum-coordinated complex that is the most widely used chemotherapy for human NSCLC in the past two decades \[[@B80]--[@B82]\]. However, multiple administration of DDP results in the development of drug resistance leading to failure of treatment, as demonstrated by tumor growth or tumor relapse \[[@B78], [@B83]\]. Therefore, to overcome the treatment plateau of DDP on NSCLC, the biological mechanisms by which DDP action is enforced must be further elucidated. As miRNAs act as critical regulators in the development of drug resistance, it would be interesting to research the mechanism through which oncogenic miRNAs modulates DDP-induced apoptosis in NSCLC.
*miR-451.* miR-451 was downregulated in NSCLC tissues in comparison to normal lung tissues, and upregulation of miR-451 enhances DDP chemosensitivity in A549 cells by inhibiting cell growth and inducing apoptosis enhancement \[[@B84]\]. Bian et al. (2011) demonstrated in their study that upregulation of miR-451 enhanced caspase-3-dependent apoptosis through the inactivation of the Akt signaling pathway, which in turn decreased Bcl-2 while increasing expression of Bax protein levels. Furthermore, results of Annexin V-FITC apoptosis assay indicated that in miR-451 transfected A549 cells (A549/miR-451) a higher percentage of apoptosis was observed in comparison to mock A549 cells. Caspase-3 activity in A549/miR-451 treated with DDP was significantly increased against the control, thus suggesting that miR-451 upregulation increases chemosensitivity of A549 cells by enhancing DDP-induced apoptosis. Together these results suggest a possible strategy for treatment of human NSCLC through the combined application of DDP treatment with miR-451 upregulation \[[@B84]\].
*miR-31.* On the other hand, miR-31 is upregulated in NSCLC cell lines and was demonstrated to induce DDP resistance. To demonstrate this, Glavinas et al. (2004) transfected miR-31 mimics into DDP-sensitive SPC-A-1 cells which led to a marked increase in the resistance of SPC-A-1 cells, while transfection of miR-31 inhibitors increased sensitivity of resistant NCI-H1299 to DDP treatment. To elucidate the mechanism by which DDP resistance is induced by miR-31, bioinformatics analysis was carried out and*ABCB9*, a membrane transporter involved in drug uptake \[[@B85]\], was predicted to be a target gene. The luciferase reporter assay then confirmed direct miR-31 regulation of*ABCB9* by binding to its 3′UTR \[[@B86]\]. Overexpression/knockdown studies indicated a significant decrease in the percentage of DDP-induced apoptotic cells when miR-31 was increased via mimics and a marked increase in DDP-induced apoptotic cells when miR-31 inhibitors were introduced, thus suggesting that miR-31 exerts an antiapoptotic effect in DDP-induced apoptosis through the inhibition of*ABCB9*.
4.2. Paclitaxel-Related miRNAs {#sec4.2}
------------------------------
Paclitaxel was the first identified member of taxanes in the list of FDA-approved anticancer drugs. This compound has been shown to have significant single-agent activity against various solid tumors \[[@B87], [@B88]\] including NSCLC \[[@B89]\]. However, combination of this compound with DDP or carboplatin showed superior response and improved survival rates \[[@B90]\].
*miR-133a/b and miR-361-3p*. High-throughput screening (HTS) approach was performed by Du and colleagues in 2013 to identify miRNAs that modulate lung cancer cell survival and response to paclitaxel treatment \[[@B91]\]. Using three NSCLC cell lines that have distinct genetic backgrounds (NCI-H1155, NCI-H1993, and NCI-H358), inhibition of two miRNAs (miR-133a/b and miR-361-3p) was found to potently decrease cell viability, although cytotoxicity of the two miRNAs vary greatly, which may be due to different endogenous expression levels of the miRNAs in each cell line. Interestingly, the inhibitors of miR-133a/b and miR-361-3p were found to reduce cell survival through different mechanisms. miR-133a/b inhibitor was able to dramatically increase apoptotic events as seen by increased percentage of cells undergoing apoptosis and increased levels of activated caspase-3. However miR-361-3p only showed a modest effect on caspase-3 activation thus suggesting that additional mechanisms are involved in the cytotoxicity of this miRNA. The effect of miRNA inhibitors on cell cycle distribution was then evaluated and results indicated that S phase arrest contributes to cytotoxicity induced by miR-133a/b and miR-361-3p inhibitors. Together these results suggest that miR-133a/b and miR-361-3p may function as oncogenes in cancer cells by regulating tumor suppressor genes.
*miR-101*. Increasing evidence has revealed that EZH2 has oncogenic properties, as an increased expression of EZH2 augments proliferation and invasion of cancer cells \[[@B92]--[@B94]\], while depletion leads to a decline in cell proliferation, increased apoptosis, and inhibition of metastatic tumor growth*in vivo*\[[@B95], [@B96]\]. Overexpression of EZH2 has been associated with tumor progression and cancer aggressiveness in NSCLC \[[@B97]\]. In a study by Zhang and colleagues (2011), it was discovered that a decreased expression of miR-101 was associated with EZH2 overexpression in NSCLC tissues \[[@B98]\]. Luciferase reporter assay revealed that miR-101 regulates EZH2 expression through the binding of its 3′UTR mRNA. Overexpression of miR-101 led to a decrease in EZH2 protein levels with subsequent decrease in the proliferation and invasive ability of NSCLC cells. Furthermore, overexpression of miR-101 led to a sensitization of NSCLC cells to paclitaxel.
4.3. Docetaxel-Related miRNAs {#sec4.3}
-----------------------------
Docetaxel, a semisynthetic analog of paclitaxel, is one of the first-line chemotherapy regimens for advanced NSCLC, with genotoxic effects caused by microtubule stabilizing, apoptotic induction through microtubule bundling, and Bcl-2 blocking \[[@B99], [@B100]\].
*miR-100*. In a miRNA microarray profiling carried out by Rui and colleagues in 2010, miR-100 was significantly downregulated in docetaxel-resistant SPC-A1/DTX cells relative to SPC-A1 parental cells \[[@B101]\]. To elucidate the role that miR-100 plays in the formation of docetaxel resistance, the authors\' transfected miR-100 mimics SPC-A1/DTX cells \[[@B102]\]. Results suggested that restoration of miR-100 expression chemosensitizes cells to docetaxel*in vitro*, complemented with a suppression of cell proliferation, enhancement of apoptosis, and cell cycle arrest in the G2/M phase of cell cycle. Ectopic miR-100 expression was also able to downregulate*in vivo*cell proliferating ability. Moreover,*PLK1* gene was identified to be a direct target of miR-100.*PLK1* plays a role in promotion of cell proliferation and overexpression of this gene has been observed in various human cancers \[[@B103]\] including NSCLC \[[@B104]\]. Knockdown of Plk1 protein expression by miR-100 led to a significant suppression of cell proliferation of SPC-A1/DTX, dramatic increase of early apoptosis rate, G2/M arresting population, and an increase in the response of SPC-A1/DTX cells to docetaxel both*in vitro* and*in vivo*. miR-100 was therefore concluded to function as a chemosensitizer restorer to docetaxel by targeting*PLK1* and inducing the suppression of cell proliferation, enhancement of apoptosis, and mitotic arrest.
*miR-650*. High expression of miR-650 can be found in lung cancer tissues, and its dysregulation is correlated with advance clinical stage as a poor prognostic factor for these patients \[[@B105]\]. Furthermore, Huang et al. (2013) determined that the expression of miR-650 is negatively correlated with patients\' response to docetaxel. Using two docetaxel-resistant cell lines (SPC-A1/DTX and H1299/DTX), the authors demonstrated that downregulation of miR-650 was able to reverse the resistance.*ING4*, a novel tumor suppressor gene, was then identified as the functional target of miR-650 and results from flow cytometry and Hoechst staining assays indicated that miR-650 inhibitor was able to induce an increase in caspase-3-dependent apoptosis. Cells transfected miR-650 inhibitors exhibited decreased expression of Bcl-2 protein, with an increased expression of Bax protein, led to the progression of apoptosis. The findings of this study confirmed that miR-650 was able to confer docetaxel chemoresistance through the regulation of Bcl-2/Bax expression by targeting of*ING4* \[[@B105]\].
4.4. Gemcitabine-Related miRNA {#sec4.4}
------------------------------
Gemcitabine, a pyrimidine nucleoside antimetabolite, has been shown to be an effective agent most particularly when administered in combination regimes \[[@B106]\]. Due to its theoretical ability of interfering with the inhibition of repair of platinum-induced DNA damage, gemcitabine is the perfect partner for platinum compounds. Gemcitabine in combination with DDP represents a common first-line treatment for patients with advanced NSCLC, especially in Europe \[[@B80], [@B107]--[@B110]\].
*miR-133b*. miR-133b is greatly reduced in cancer tissue in comparison to adjacent normal lung tissue \[[@B111]\]. Prediction programs identified two common predicted targets of miR-133b, the antiapoptotic*MCL-1* and*BCL-W*, both of which are members of the antiapoptotic*BCL-2* family \[[@B112]\] and have previously been reported to be increased in both solid and hematological malignancies including lung cancer \[[@B113], [@B114]\]. Transfection of miR-133b using pre-miR-133b resulted in a decrease in Bcl-W and Mcl-1 protein expression with a moderate increase of apoptosis. However combination treatment of miR-133b overexpression with 24 hours treatment of gemcitabine resulted in a greater degree of cleaved PARP expression as well as apoptosis. This concludes that miR-133b is able to target prosurvival molecules and induce apoptosis in the setting of chemotherapeutic agents \[[@B111]\].
4.5. Epidermal Growth Factor Receptor Tyrosine Kinase Inhibitors- (EGFR-TKIs-) Related miRNAs {#sec4.5}
---------------------------------------------------------------------------------------------
EGFR is a plasma membrane glycoprotein that belongs to a family of four different tyrosine kinase receptors (EGFR (ErbB1), HER2/neu (ErbB2), HER3 (ErbB3), and HER4 (ErbB4)) \[[@B115]\]. Dimerization of EGFR may result in cancer cell proliferation, inhibition of apoptosis, invasion, metastasis, and tumor induced neovascularization \[[@B116]\]. Mutations and subsequent overexpression of EGFR can be found in all histologic subtypes of NSCLC \[[@B117]\]. Deletion in exon 19, which removes the conserved sequence LREA, and a single point mutation in exon 21, which leads to the substitution of arginine for leucine at position 858 (L858R), are the most clinically relevant and extensively studied drug-sensitive mutations \[[@B118]\]. Studies have shown that these mutations preferentially bind to first generation EGFR-TKIs, gefitinib and erlotinib \[[@B119], [@B120]\]. First generation EGFR-TKIs function by selectively targeting the receptor via a competitive, reversible binding at the tyrosine kinase domain, thus leading to the inhibition of ATP binding and subsequent signal transduction and downstream functions \[[@B121]\]. However, acquired resistance to EGFR-TKIs in the metastatic setting is unavoidable. While the average progression-free survival (PFS) is between 10 and 16 months, treatment duration can last as short as 1 month \[[@B122]\]. Drug resistance therefore still remains a problem and new therapies and strategies must be developed to overcome such resistance.
*miR-30b/c and miR-221/222*. EGF and MET receptors control gefitinib-induced apoptosis and NSCLC tumorigenesis through the downregulation of specific oncogenic miRNAs, miR-30b/c, and miR-221/222 \[[@B123]\]. Using bioinformatics analysis and luciferase assays,*APAF-1* and*BIM*(previously found to play a role in TKI sensitivity \[[@B124], [@B125]\]) were determined to be direct targets of miR-221/222 and -30b/c. To investigate the roles these miRNAs play in gefitinib-induced apoptosis, wild-type EGFR expressing NSCLC cells (Calu-1 and A549) and cells with EGFR exon-19 deletions (PC9 and HCC827) were utilized. Upon gefitinib treatment, significant downregulation of miR-30b/c and miR-221/222 with an increased BIM and APAF-1 protein levels were observed only in PC9 and HCC827 sensitive cells. To further determine the contribution of miR-30b/c and miR-221/222-mediated APAF-1 and BIM downregulation to cellular TKI response, Garofalo et al. (2012) overexpressed APAF-1 and BIM in A549 resistant cells, which consequently led to gefitinib-induced PARP cleavage. Furthermore, as miR-30b/c and miR-221/222 are regulated by MET, a strong downregulation was observed of these miRNAs when Calu-1- and A549-MET overexpressing cells were treated with MET inhibitors SU11274. Furthermore an increase in caspase-3/7 activity and decreased cell viability was observed in SU11274-treated Calu-1 cells following exposure to varying gefitinib concentrations. Together, these results suggest that MET inhibition restores gefitinib sensitivity in TKI-resistant Calu-1 through downregulation of miR-30b/c and miR-221/222 \[[@B123]\].
*miR-214*. miR-214 is significantly upregulated in gefitinib resistant lung adenocarcinoma cell line, HCC827/GR, in comparison to parental HCC827 lung adenocarcinoma cells. HCC827/GR was obtained by exposing HCC827 cells to increasing concentrations of gefitinib over six months \[[@B126]\]. Using dual-luciferase reporter assay, Wang et al. (2012) confirmed PTEN as a direct functional target of miR-214. PTEN encodes a 403 amino acid dual-specificity lipid and protein phosphatase which functions as a tumor suppressor in many tumors \[[@B127], [@B128]\]. Knockdown of miR-214 expression resulted in the upregulation of PTEN protein and inactivation of AKT, which is largely linked to antiapoptotic function \[[@B129], [@B130]\]. Furthermore, knockdown of miR-214 resensitized HCC827/GR to gefitinib, as demonstrated through MTS assay. miR-214 was thus concluded to potentially serve as a therapeutic target to reverse the acquired resistance of gefitinib in lung adenocarcinoma cells.
*miR-133b*. Expression of miR-133b is significantly downregulated in NSCLC tissues in comparison to nonneoplastic lung tissues \[[@B131]\], and the 3′UTR of*EGFR* was found to be a direct target of this miRNA thus inhibiting its expression. Treatment of EGFR-addicted lung cancer cells, PC-9 and A549 with miR-133b mimic inhibited phosphorylation of EGFR, AKT, and extracellular signal-related kinase (ERK)1/2, thus inhibiting their growth and invasion abilities. However in non-EGFR-addicted NSCLC cells NCI-H1650 and NCI-H1975, no significant changes in the expression of phosphorylated EGFR, AKT, and ERK1/2 were found. Furthermore, miR-133b was able to restore or enhance EGFR-TKI sensitivity in NSCLC cells, especially in EGFR-addicted cells. These findings reveal that transfection of miR-133b in EGFR-addicted NSCLC has the therapeutic potential for overcoming EGFR-TKI resistance \[[@B131]\].
4.6. PRIMA-1-Related miRNA {#sec4.6}
--------------------------
The tumor suppressor*p53* gene regulates cell growth through the activation of the transcription of numerous genes specifically those involved in cell cycle regulation, apoptosis, and genomic stability \[[@B132]--[@B134]\] andhas also been implicated in the response to anticancer therapies \[[@B133]\].*p53* has been reported to be frequently mutated in humans cancers with mutations occurring in greater than 50% of lung tumors \[[@B135], [@B136]\]. Restoration of wild-type p53 function has led to regression of cancers in mice \[[@B137], [@B138]\], and thus efforts to treat cancers through the reactivation of*p53* with a low-molecular-weight compound such as PRIMA-1 (*p53*-dependent reactivation and induction of massive apoptosis) \[[@B139], [@B140]\] are widely supported.
*miR-34a*. In a study conducted by Duan et al. (2010), the role of miR-34 family members in regulating PRIMA-1 induced apoptosis was investigated. The authors discovered that PRIMA-1 was able to upregulate miR-34a in*p53* mutant cells. Previous studies have shown evidence that the miR-34 family plays a role in the regulation of cell proliferation and apoptosis \[[@B19], [@B142]--[@B148]\]. The results of this study suggest that PRIMA-1 is able to restore wild-type function to mutant*p53*, which will upregulate miR-34a to induce apoptosis in lung cancer cells \[[@B149]\].
4.7. Curcumin Related miRNA {#sec4.7}
---------------------------
Curcumin is a compound extracted from the rhizomes of*Curcuma longa*L. and studies carried out exhibited its diverse pharmacological effects which include anti-inflammatory, antioxidant, and antitumor activities \[[@B150]\]. Previous studies have also shown that curcumin can induce apoptosis in many types of cancer cells \[[@B151], [@B152]\], through the inhibition of*NF-*κ*B*,*survivin/BIRC5*, and*BCL-2* \[[@B153], [@B154]\]. However few studies have been carried out to report the importance of miRNA expression modulation in mediating the biological effects of curcumin.
*miRNA-186*\*. In a study conducted by Zhang et al.(2010), curcumin was shown to have the ability to inhibit cell proliferation and induce apoptosis in A549 cells. The authors performed a cluster analysis on the expression profiles on curcumin-treated and dimethyl sulfoxide (DMSO) control-treated samples and found that miR-186\* was shown to be significantly downregulated in response to curcumin treatment, thus suggesting that miR-186\* may play an oncogenic role in human lung cancer cells. Inhibition of miR-186\* was shown to greatly decrease cell proliferation in A549 cells and increase the induction of apoptosis. Furthermore, caspase-10 was revealed to be a direct target of miR-186 \[[@B155]\]. This study thus provided the first evidence that miR-186\* is essential for the anticancer effects of curcumin in A549 cells and that caspase-10 may be an important target of miR-186\* in preventing apoptosis.
However, even though curcumin has exhibited antitumor activity, there has been concern regarding the effects of curcumin on multidrug resistant cells \[[@B156], [@B157]\]. To analyze such effects, A549/DDP, the DDP-resistant derivative of parental A549 cells generated by coculturing parental A549 cells with 6 nm DDP to maintain the drug resistance phenotype, was utilized. In a study conducted by Zhang et al. (2010), a comprehensive miRNA profiling of untreated multidrug-resistant cell line (A549/DDP) was performed and compared against results obtained for A549/DDP cells treated with curcumin. Results showed that miR-186\* was downregulated more than 2.5-fold compared to levels in control cells. The antiapoptotic effects of miR-186\* in A549/DDP cells were investigated and it was found that transfection of miR-186\* mimics led to an inhibition of apoptosis in comparison to that in the control, thus suggesting that miR-186\* plays an oncogenic role in this cell line. To confirm the role miR-186\* plays in curcumin-induced A549/DDP apoptosis, flow cytometry was used to detect the rate of apoptosis in A549/DDP cells treated with curcumin, control cells, or curcumin combined with miR-186\* mimic cells. Results indicated that apoptosis in the combination group was significantly decreased in comparison to cells treated with curcumin \[[@B158]\]. These findings reveal that curcumin is able to induce apoptosis in the multidrug resistant cell line by downregulating miR-186\*.
4.8. Multidrug Resistance {#sec4.8}
-------------------------
*miR-200bc/429*. In 2012, Zhu et al. reported that the miR-200bc/429 cluster was downregulated in multidrug-resistant A549/DDP cells, in comparison to parental A549 cell \[[@B159]\]. Recent studies have suggested that aberrant DNA methylation of the promoter region of the miR-200bc/429 cluster may be a critical mechanism leading to dysregulated expression level of the miR-200 family \[[@B160], [@B161]\]. While the roles of the two sequence clusters of miR-200 family on the epithelial-to-mesenchymal transition of tumor cells are well studied, the role that this miRNA family plays on apoptosis has been minimally studied. Zhu et al. demonstrated using MTT that transfection of miR-200bc/429 cluster mimics into A549/DDP greatly enhanced sensitivity of this cell line to various anticancer drugs including vincristine (VCR), etoposide (VP-16), adriamycin (ADR), and DDP. It was found that miR-200bc/429 cluster was able to modulate multidrug resistance (MDR) in lung cancer cell lines, at least in part by inhibiting the antiapoptotic Bcl-2 and XIAP protein expression, thus affecting the mitochondrial release of cytochrome c. Therapeutic methods that target the miR-200bc/429 clusters thus provides a promising method to enhance treatment effect of NSCLC.
*miR-181b*. In another study by Zhu et al. (2010), miR-181b was also found to be downregulated in multidrug-resistant A549/DDP cells, in comparison to parental A549 cell line \[[@B162]\]. To determine whether miR-181b has a direct role in MDR development, MTT assay was performed revealing that all A549/DDP cells transfected with miR-181b mimic exhibited a significant increase in sensitivity to a number of anticancer drugs including 5-fluorouracil (5-Fu), VCR, DDP, VP-16, and ADR. Bioinformatics analysis predicted antiapoptotic*BCL-2* as a potential target of miR-181, with two conserved target sites in the 3′UT region. Transfection of miR-181a in A549/DDP cells led to a significant decrease in Bcl-2 protein levels, as demonstrated by Western blot. Furthermore, A549/DDP miR-181b transfected cells also led to an increase in apoptosis as detected by flow cytometry. Together these results demonstrate miR-181b\'s ability to modulate the development of MDR in lung cancer cell lines, at least in part, by modulation of apoptosis through the targeting of the antiapoptotic*BCL-2*.
4.9. TRAIL-Related miRNAs {#sec4.9}
-------------------------
The Apo2L/tumor necrosis factor- (TNF-) *α*-related apoptosis inducing ligand (TRAIL) is a member of the TNF family that is known to induce apoptosis in various cancers \[[@B163]\]. Treatment of transformed cells with TRAIL has been shown to successfully induce apoptosis both*in vitro* and*in vivo*\[[@B163], [@B164]\]; however a wide range of human cancer cells are resistant to TRAIL-induced apoptosis \[[@B165]\].
*miR-221 and -222*. To identify the mechanisms by which miRNAs may play a role in TRAIL resistance, Garofalo et al. (2008) carried out a genome wide profiling of miRNAs in three different lung cancer cell lines (A459, Calu-1, and NCI-H460) and found that miR-221 and miR-222 were markedly upregulated in TRAIL-resistant cells. In TRAIL sensitive cells NCI-H460, TRAIL was able to induce the activation of the caspase cascade, evaluated by the appearance of cleaved fragments. However, transfection of NCI-H460 cells with pre-miRs-221 and -222 caused a significant reduction of TRAIL-mediated cell death machinery activation. Further experiments deduced that miR-221 and-222 directly targeted *p*27^Kip1^, and inhibition of *p*27^Kip1^ via pre-miR-221 and -222 transfection led to an increase in cell resistance to TRAIL as assessed by Annexin V staining, and PARP and caspase-8 activation. Taken together, the authors\' results demonstrate that increased levels of miR-221 and -222 may modulate sensitivity of NSCLC cells to TRAIL with important implications in the design of new therapeutic agents.
*miR-34a and miR-34c*. In another study, miR-34a and miR-34c expression were found to be significantly downregulated in NSCLC cells and lung tumors in comparison to normal lung tissues. Performing a bioinformatics search, Garofalo et al. (2013) determined that*PDGFR-*α** and*PDGFR-*β** were targets of these miRNAs; both of which have been reported to be overexpressed and associated with poor outcome in lung cancer \[[@B166]\]. Through targeting*PDGFR-*α** and*PDGFR-*β**, miR-34a/c were able to decrease invasiveness as well as increase TRAIL-induced apoptosis. TRAIL resistance is common in lung tumors and it has been reported that PDGFR-*α* and PDGFR-*β* regulate the PI3K/Akt and ERK1/2 pathways \[[@B167], [@B168]\], which play a role in TRAIL-induced apoptosis \[[@B169]\]. Phosphorylation levels of ERKs were found to be decreased following ectopic expression of miR-34a/c; additionally caspase-3/7 assay revealed an increase in TRAIL sensitivity. This study demonstrates that inhibition of*PDGFR-*α** and*PDGFR-*β** by miR-34a/c is able to antagonize tumorigenicity and increase sensitivity to TRAIL-induced cell death \[[@B170]\].
*miR-212*. PED/PEA-15 is a death effector domain (DED) family member, which has been implicated in the processes of cell growth and metabolism \[[@B171]--[@B173]\]. Furthermore, PED/PEA-15 has a broad range of antiapoptotic ability, being able to inhibit both the intrinsic and extrinsic apoptotic pathways \[[@B172], [@B174]\]. Zanca and colleagues (2008) reported that PED/PEA-15 overexpression plays a role in TRAIL resistance in NSCLC \[[@B175]\]; however the mechanism that regulates its expression is not well known. In further studies, Incoronato et al. (2010) reported that NSCLC-affected lung tissue has an increased expression of PED/PEA-15 with a concurrent downregulation of miR-212 and decreased response to TRAIL treatment \[[@B176]\]. miR-212 negatively regulates*PED/PEA-15* by directly binding to its 3′UTR. miR-212 downregulation has previously been reported to be involved in lung cancer response to chemotherapy, in particular to docetaxel \[[@B101]\]. In this study, transfection of NSCLC Calu-1 cells with pre-miR-212 led to a decrease in PED/PEA-15 expression with increased caspase-8 activation following treatment with TRAIL, indicating increased sensitivity of Calu-1 cells to TRAIL-mediated cell death. Therefore, the expression of miR-212 could be used to predict therapeutic response to TRAIL in lung cancer.
5. Conclusions {#sec5}
==============
In terms of molecular events occurring in tumors, evasion of apoptosis is an important hallmark of tumor progression. Recent evidence has exhibited deregulated miRNAs to play a role in the apoptotic process. In lung cancer, upregulated miRNAs have been shown to serve as oncogenes, targeting tumor suppressor, and/or proapoptotic genes, while downregulated miRNAs can function as tumor suppressors, targeting oncogenic and/or antiapoptotic genes. Additionally, studies have also indicated that miRNAs play a significant role in altering sensitivity and resistance to cytotoxic treatment. Targeting of specific miRNAs could therefore potentially be used as valuable therapeutics for lung cancer. Together, these studies have illustrated the importance for further studies and validation of miRNAs and their targets. Furthermore, there is a serious shortage in research being carried out in miRNA-regulated apoptosis in SCLC. As SCLC accounts for 16.8% of lung cancer incidence and is a highly aggressive form of lung cancer it would be of great interest to determine the functions of miRNAs in regulation of apoptosis in this lung cancer subtype.
This work was supported by the High Impact Research Grant (HIR) (H-21001-F000036), the University of Malaya Postgraduate Research Fund (PPP) (PV058-2011B), and the Centre for Research in Biotechnology for Agriculture (CEBAR) RU Operation Grant (RU005C-2014). All funders had no role in decision to publish or preparation of the paper.
*γ*-H2AX:
: Gamma, H2A histone family, member X
3′UTR:
: 3′-Untranslated region
5-Fu:
: 5-Fluorouracil
ABCB9:
: ATP-binding cassette, subfamily B (MDR/TAP), member 9
AChE:
: Acetylcholinesterase
ADR:
: Adriamycin
Akt:
: Protein kinase B
Apaf-1:
: Apoptotic peptidase activating factor 1
BAK:
: BCL2-antagonist/killer 1
BAX:
: BCL2-associated X protein
BCL-2:
: B-cell CLL/Lymphoma
BCL-W:
: BCL2-like 2
BCL-XL:
: BCL2-like 1
BIM:
: BCL2-like 11 (apoptosis facilitator)
BIRC5:
: Baculoviral IAP repeat containing 5
BMF:
: Bcl-2-modifying factor
BTG2:
: BTG family, member 2
DAB2IP:
: Disabled homolog 2-interacting protein
DDP:
: Cisplatin
DMSO:
: Dimethyl sulfoxide
E2F1:
: E2F transcription factor 1
EGFR:
: Epidermal growth factor receptor
ERK1:
: Mitogen-activated protein kinase 3
ERK2:
: Mitogen-activated protein kinase 1
EZH2:
: Histone-lysine N-methyltransferase
FasL:
: Fas ligand
FDA:
: Food and drug administration
FGFR1:
: Fibroblast growth factor receptor 1
FITC:
: Fluorescein isothiocyanate
HTS:
: High-throughput screening
I*κ*B*α*:
: Nuclear factor of kappa light polypeptide gene enhancer in B cells inhibitor alpha
ING4:
: Inhibitor of growth family, member 4
IRAK-1:
: Interleukin-1 receptor-associated kinase 1
MAPK:
: Mitogen activated protein kinase
MCL1:
: Molecule myeloid leukemia 1
MDR:
: Multidrug resistance
miRNA:
: MicroRNA
NF-*κ*B:
: Nuclear factor kappa-light-chain-enhancer of activated B cells
NSCLC:
: Non-small cell lung cancer
NOXA:
: Phorbol-12-myristate-13-acetate-induced protein 1
*p*27^Kip1^:
: Cyclin-dependent kinase inhibitor 1B
p53:
: Tumor protein p53
PDCD4:
: Programmed cell death 4 (neoplastic transformation inhibitor)
PDGFR-*α*:
: Platelet-derived growth factor receptor-alpha polypeptide
PDGFR-*β*:
: Platelet-derived growth factor receptor-beta polypeptide
PEA-15:
: Astrocytic phosphoprotein PEA-15
PED:
: Preimplantation embryonic development
PI3K/AKT:
: Phosphatidylinositol 3-kinase/protein kinase B
PKC:
: Protein kinase C
PKC-*α*:
: Protein kinase C-alpha
PKC-*ε*:
: Protein kinase C-epsilon
PLK1:
: Polo-like kinase 1
PRIMA-1:
: *p53*-dependent reactivation and induction of massive apoptosis
PFS:
: progression-free survival
PUMA:
: BCL2 binding component 3
RAB14:
: RAB14, member RAS oncogene family
RB1:
: Retinoblastoma 1
RHOB:
: Ras homolog family member B
RUNX3:
: Runt-related transcription factor 3
SCLC:
: Small cell lung cancer
SP1:
: Specificity Protein 1
SPRY1:
: Sprouty homolog 1, antagonist of FGF signaling (*Drosophila*)
SPRY2:
: Sprouty homolog 2 (*Drosophila*)
STAT:
: Signal transducer and activator of transcription
TRAIL:
: Apo2L/tumor necrosis factor- (TNF-) *α*-related apoptosis inducing ligand
TGF-*β*:
: Transforming growth factor-beta
VCR:
: Vincristine
VP-16:
: Etoposide
WNT:
: Wingless-type MMTV integration site family
XIAP:
: X-linked inhibitor of apoptosis.
Conflict of Interests
=====================
The authors declare that there is no conflict of interests regarding the publication of this paper.
{#fig1}
{#fig2}
######
Upregulated apoptosis-associated miRNAs in lung cancer.
------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
MicroRNA Target genes Function Cell lines *In vivo* models Citation
-------------------- ---------------------------------------------- ----------------------------------------------------------------------------------- ----------------------------------- --------------------------------------------- ------------
miR-197 BMF, NOXA Repress *p53*-dependent apoptotic cascade\ A549, Calu-1, NIH-H460, NCI-H1299 Nude mice \[[@B35]\]
miR inhibition decreases cell viability\
miR inhibition impairs cell growth and anchorage-independent colony formation
miR-21 SPRY1, SPRY2, BTG2, PDCD4, APAF1, FasL, RHOB Enhance tumor proliferation and survival\ N/T K-ras^LA2^ mice\ \[[@B37]\]
Inhibit apoptosis\ CAG-miR-21 transgenic mice\
miR deletion suppresses Ras-driven transformation CAG-miR-21; K-ras^LA2^ compound mutant mice
miR-212 AChE Prevent apoptosis\ HEK-293T, NCI-H520, NCI-H460,\ Nude mice \[[@B43]\]
Maintain cell proliferation capacity\ SK-MES-1,\
Modulate CDDP-induced NSCLC cell apoptosis BEAS-2B
miR-17-5p, miR-20a E2F1 miR inhibition reduces of cell growth.\ Calu-6, A549,\ N/T \[[@B45]\]
miR inhibition induces apoptosis and increases proportions of sub-G1 populations. ACC-LC-172
------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
N/D: not determined; N/T: not tested.
######
Downregulated apoptosis-associated miRNAs in lung cancer.
-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
MicroRNA Target genes Function Cell lines *In vivo* models Citation
---------- -------------- ----------------------------------------------------------------------------------------------- ------------------------------ --------------------------- ------------
miR-7 BCL-2 Suppress cell proliferation and induce cell apoptosis\ A549, NCI-H1299, NCI-H1355,\ Nude mice \[[@B51]\]
Inhibit cancer cell migration *in vitro* \ NCI-H460, MRC-5, HEK-293T
Reduce tumorigenicity *in vivo*
miR-198 FGFR1 Inhibit lung cancer cells proliferation\ A549,\ Athymic BALB/c nude mice. \[[@B69]\]
Enhance cell apoptosis\ NCI-H460
Inhibits growth of tumor graft in nude mouse
miR-451 RAB14 Inhibit *in vitro* proliferation and enhance apoptosis\ A549, SPC-A1,\ Athymic BALB/c nude mice. \[[@B75]\]
Decrease phosphorylation of AKT and increased BAX or Bad protein level\ NCI-H520
Associated with *in vivo* proliferation capacity
miR-192 RB1 Inhibit cell proliferation and promotes cell apoptosis\ A549,\ Athymic BALB/c nude mice. \[[@B76]\]
Arrest cell in G1 phase\ NCI-H460,\
Inhibit tumorigenesis *in vivo* 95D
miR-335 BCL-W, SP1 Suppress proliferation and invasion ability of cells\ A549, NCI-H1299 N/T \[[@B52]\]
Induce apoptosis\
Suppress metastasis and invasiveness of cells
miR-608 N/D Increase cell death in *Bcl-xL* silenced cells A549, SK-LU-1 N/T \[[@B55]\]
miR-203 PKC*α* Decrease cell proliferation\ A549 N/T \[[@B61]\]
Promote cell apoptosis, but this effect only partially relies on its downregulation of PKC*α*
miR-413 PKC*ε* Inhibit cell proliferation and enhance apoptosis A549, Calu-1 N/T \[[@B63]\]
miR-146a EGFR Inhibit cell growth and induces cell apoptosis\ NCI-H358,\ N/T \[[@B70]\]
Suppress motility\ NCI-H1650,\
Enhance cell proliferation inhibitory effect of TKIs and cetuximab NCI-H1975,\
NCI-H292, HCC827
miR-26a EZH2 Inhibit cell proliferation *in vitro*Block G1/S phase transition and induced apoptosis\ SPC-A1, A549,\ N/T \[[@B71]\]
Decreased metastasis capacity and invasion SK-MES-1
-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
N/D: not determined; N/T: not tested.
######
Drug-associated miRNAs in lung cancer.
------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
Drug MicroRNA Target genes Function Cell Lines *In vivo* models Citation
---------------------- ------------------------ ------------------------------------------------------------------------------------------------- ------------------------------------------------------------------------------------------------------- -------------------------------------------- ------------------ ----------------------
Docetaxel miR-100 Plk1 Chemosensitize lung adenocarcinoma cells to docetaxel\ SPC-A1, A549,\ Nude mice \[[@B102]\]
Suppress cell proliferation, enhance apoptosis and cell cycle arrest in the G2/M phase of cell cycle\ NCI-H1299,\
Down-regulate *in vivo* cell proliferating ability SPC-A1/DTX
miR-650 ING4 Confer docetaxel chemoresistance both *in vitro* and *in vivo* SPC-A1, NCI-H1299 Athymic BALB/c nude mice \[[@B105]\]
Cisplatin miR-451 N/D Enhance DDP chemosensitivity\ A549 BALB/c nude mice \[[@B84]\]
Inhibit growth and enhance apoptosis
miR-31 ABCB9 Induced DDP resistance\ SPC-A1, LTEP-A2,\ N/T \[[@B86]\]
Antiapoptotic effect NCI-H460, NCI-H1299
Paclitaxel miR-133a/b, miR-361-3p N/D Oncogenic\ NCI-H1155, NCI-H1993,\ N/T \[[@B91]\]
miR inhibition reduces cell survival NCI-H358
miR-101 EZH2 Decrease proliferation and invasive ability of cells\ NCI-H226, A549,\ N/T \[[@B98]\]
Sensitize cells to paclitaxel NCI-H358, 801D
Gemcitabine miR-133b MCL-1,\ Increase apoptosis\ A549, NCI-H23,\ N/T \[[@B111]\]
BCL-W Gemcitabine sensitivity NCI-H2172, NCI-H226, NCI-H522, NCI-H2009,\
NCI-H1703
EGFR-TKI miR-30b/c miR-221/222 APAF-1,\ Gefitinib-induced PARP cleavage\ Calu-1, A549, PC-9, HCC827 N/T \[[@B123]\]
BIM Decrease cell viability\
Restore gefitinib sensitivity
miR-214 PTEN Oncogenic\ HCC827, HCC827/GR N/T \[[@B126]\]
Knockdown sensitizes cells to gefitinib
miR-133b EGFR Inhibit cell\'s growth and invasion abilities\ PC-9, A549, NCI-H1650, NCI-H1975 N/T \[[@B131]\]
Enhance EGFR-TKI sensitivity
Curcumin miRNA-186\* Caspase-10 Inhibits cell apoptosis\ A549, A549/DDP N/T \[[@B155], [@B158]\]
Down-regulation of miR-186 by curcumin induces apoptosis
TRAIL miR-221, miR-222 N/D Impair TRAIL-dependent apoptosis\ Calu-1, A549,\ N/T \[[@B169]\]
Induce TRAIL-resistance NCI-H460
miR-34a, miR-34c PDGFR-*α*, PDGFR-*β* Augment TRAIL response\ NCI-H460, A549,\ N/T \[[@B170]\]
Reduce migratory and invasive capacity of cells NCI-H1299, Calu-6\
NCI-H1703,
miR-212 PED Increase sensitivity to TRAIL\ Calu-1, NCI-H460 N/T \[[@B176]\]
Tumor suppressor
PRIMA-1 miR-34a N/D Induce apoptosis in the lung cancer cells containing mutant *p53* A549, NCI-H211,\ N/T \[[@B149]\]
NCI-H1155, NCI-H1299
Multidrug Resistance miR-200bc/429 BCL2, XIAP Enhanced sensitivity to various anticancer drugs including VCR, CDDP, VP-16, and ADR A549, A549/CDDP N/T \[[@B159]\]
miR-181b BCL2 Increased sensitivity to a number of anticancer drugs including VCR, 5-Fu, CDDP, VP-16, and ADR A549, A549/CDDP N/T \[[@B162]\]
------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
N/D: Not determined, N/T: Not tested.
[^1]: Academic Editor: Dong Wang
| {
"pile_set_name": "PubMed Central"
} |
Related literature {#sec1}
====================
For the isostructural compound 5-chloro-2-hydroxybenzaldehyde 4-ethylthiosemicarbazone, see: Lo *et al.* (2011[@bb1])
Experimental {#sec2}
==============
{#sec2.1}
### Crystal data {#sec2.1.1}
C~10~H~12~BrN~3~OS*M* *~r~* = 302.20Monoclinic,*a* = 22.040 (4) Å*b* = 11.844 (2) Å*c* = 9.5102 (19) Åβ = 101.69 (3)°*V* = 2431.1 (8) Å^3^*Z* = 8Mo *K*α radiationμ = 3.54 mm^−1^*T* = 123 K0.20 × 0.10 × 0.05 mm
### Data collection {#sec2.1.2}
Rigaku Saturn70 diffractometerAbsorption correction: multi-scan (*CrystalClear*; Rigaku, 2008[@bb2]) *T* ~min~ = 0.661, *T* ~max~ = 0.8384201 measured reflections2331 independent reflections1760 reflections with *I* \> 2σ(*I*)*R* ~int~ = 0.032
### Refinement {#sec2.1.3}
*R*\[*F* ^2^ \> 2σ(*F* ^2^)\] = 0.042*wR*(*F* ^2^) = 0.114*S* = 0.952331 reflections155 parameters3 restraintsH atoms treated by a mixture of independent and constrained refinementΔρ~max~ = 0.70 e Å^−3^Δρ~min~ = −1.01 e Å^−3^
{#d5e409}
Data collection: *CrystalClear* (Rigaku, 2008[@bb2]); cell refinement: *CrystalClear*; data reduction: *CrystalClear*; program(s) used to solve structure: *SHELXS97* (Sheldrick, 2008[@bb3]); program(s) used to refine structure: *SHELXL97* (Sheldrick, 2008[@bb3]); molecular graphics: *SHELXTL* (Sheldrick, 2008[@bb3]); software used to prepare material for publication: *publCIF* (Westrip, 2010[@bb4]).
Supplementary Material
======================
######
Click here for additional data file.
Crystal structure: contains datablock(s) I, global. DOI: [10.1107/S1600536813008787/ng5322sup1.cif](http://dx.doi.org/10.1107/S1600536813008787/ng5322sup1.cif)
######
Click here for additional data file.
Structure factors: contains datablock(s) I. DOI: [10.1107/S1600536813008787/ng5322Isup2.hkl](http://dx.doi.org/10.1107/S1600536813008787/ng5322Isup2.hkl)
######
Click here for additional data file.
Supplementary material file. DOI: [10.1107/S1600536813008787/ng5322Isup3.cml](http://dx.doi.org/10.1107/S1600536813008787/ng5322Isup3.cml)
Additional supplementary materials: [crystallographic information](http://scripts.iucr.org/cgi-bin/sendsupfiles?ng5322&file=ng5322sup0.html&mime=text/html); [3D view](http://scripts.iucr.org/cgi-bin/sendcif?ng5322sup1&Qmime=cif); [checkCIF report](http://scripts.iucr.org/cgi-bin/paper?ng5322&checkcif=yes)
Supplementary data and figures for this paper are available from the IUCr electronic archives (Reference: [NG5322](http://scripts.iucr.org/cgi-bin/sendsup?ng5322)).
The authors would like to thank the China Scholarship Council (CSC).
Comment
=======
A Schiff ligand was synthesized through one-pot reaction with high yield using 5-bromo-2-hydroxybenzaldehyde and 4-ethyl-3-thiosemicarbazide (Fig. 1). The title compound can be used as tridentate chelating ligand to construct spin-crossover complexes. Isostructural 5-chloro-2-hydroxybenzaldehyde-4-ethylthiosemicarbazone was reported previously (Lo *et al.*, 2011).
In the title compound, a strong intramolecular hydrogen bond O---H···N is observed. An intermolecular N---H···S hydrogen bond connects two molecules into a supramolecular dimer as shown in Figure 2.
Experimental {#experimental}
============
5-Bromo-2-hydroxybenzaldehyde (4.02 g, 20 mmol) in 50 ml ethanol and 4-ethyl-3-thiosemicarbazide (2.38 g, 20 mmol) were reacted for 6 h at 350 K. Slow evaporation of the yellow solution gave large colorless crystals.
Refinement {#refinement}
==========
Carbon-bound H-atoms were placed in calculated positions (C---H 0.95, 0.98 and 0.99 Å) and were included in the refinement in the riding model approximation, with *U*~iso~(H) =1.5U~eq~(C) for methyl H atoms and 1.2U~eq~(C) for the others. The hydroxy and amino H atoms were located in a difference Fourier map, and were refined with distance restraints of O---H 0.85±0.01 and N---H 0.88±0.01 Å; with *U*~iso~(H) =1.2U~eq~(N and O).
Figures
=======
{#Fap1}
{#Fap2}
Crystal data {#tablewrapcrystaldatalong}
============
----------------------- ---------------------------------------
C~10~H~12~BrN~3~OS *F*(000) = 1216
*M~r~* = 302.20 *D*~x~ = 1.651 Mg m^−3^
Monoclinic, *C*2/*c* Mo *K*α radiation, λ = 0.710747 Å
Hall symbol: -C 2yc Cell parameters from 3650 reflections
*a* = 22.040 (4) Å θ = 3.1--27.5°
*b* = 11.844 (2) Å µ = 3.54 mm^−1^
*c* = 9.5102 (19) Å *T* = 123 K
β = 101.69 (3)° Block, colourless
*V* = 2431.1 (8) Å^3^ 0.20 × 0.10 × 0.05 mm
*Z* = 8
----------------------- ---------------------------------------
Data collection {#tablewrapdatacollectionlong}
===============
------------------------------------------------------------------ --------------------------------------
Rigaku Saturn70 diffractometer 2331 independent reflections
Radiation source: Rotating Anode 1760 reflections with *I* \> 2σ(*I*)
Confocal monochromator *R*~int~ = 0.032
Detector resolution: 28.5714 pixels mm^-1^ θ~max~ = 26.0°, θ~min~ = 3.1°
dtprofit.ref scans *h* = −27→20
Absorption correction: multi-scan (*CrystalClear*; Rigaku, 2008) *k* = −9→14
*T*~min~ = 0.661, *T*~max~ = 0.838 *l* = −11→10
4201 measured reflections
------------------------------------------------------------------ --------------------------------------
Refinement {#tablewraprefinementdatalong}
==========
------------------------------------- -------------------------------------------------------------------------------------
Refinement on *F*^2^ Primary atom site location: structure-invariant direct methods
Least-squares matrix: full Secondary atom site location: difference Fourier map
*R*\[*F*^2^ \> 2σ(*F*^2^)\] = 0.042 Hydrogen site location: inferred from neighbouring sites
*wR*(*F*^2^) = 0.114 H atoms treated by a mixture of independent and constrained refinement
*S* = 0.95 *w* = 1/\[σ^2^(*F*~o~^2^) + (0.0752*P*)^2^\] where *P* = (*F*~o~^2^ + 2*F*~c~^2^)/3
2331 reflections (Δ/σ)~max~ = 0.001
155 parameters Δρ~max~ = 0.70 e Å^−3^
3 restraints Δρ~min~ = −1.01 e Å^−3^
------------------------------------- -------------------------------------------------------------------------------------
Special details {#specialdetails}
===============
-----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
Geometry. All e.s.d.\'s (except the e.s.d. in the dihedral angle between two l.s. planes) are estimated using the full covariance matrix. The cell e.s.d.\'s are taken into account individually in the estimation of e.s.d.\'s in distances, angles and torsion angles; correlations between e.s.d.\'s in cell parameters are only used when they are defined by crystal symmetry. An approximate (isotropic) treatment of cell e.s.d.\'s is used for estimating e.s.d.\'s involving l.s. planes.
Refinement. Refinement of *F*^2^ against ALL reflections. The weighted *R*-factor *wR* and goodness of fit *S* are based on *F*^2^, conventional *R*-factors *R* are based on *F*, with *F* set to zero for negative *F*^2^. The threshold expression of *F*^2^ \> σ(*F*^2^) is used only for calculating *R*-factors(gt) *etc*. and is not relevant to the choice of reflections for refinement. *R*-factors based on *F*^2^ are statistically about twice as large as those based on *F*, and *R*- factors based on ALL data will be even larger.
-----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
Fractional atomic coordinates and isotropic or equivalent isotropic displacement parameters (Å^2^) {#tablewrapcoords}
==================================================================================================
------ --------------- -------------- ------------- -------------------- --
*x* *y* *z* *U*~iso~\*/*U*~eq~
Br1 0.484097 (17) 0.67444 (4) 1.01744 (4) 0.04059 (19)
C1 0.69767 (15) 0.6276 (2) 1.0597 (3) 0.0180 (7)
C2 0.67216 (16) 0.6364 (3) 1.1818 (3) 0.0197 (7)
H2 0.6985 0.6352 1.2742 0.024\*
C3 0.60893 (17) 0.6470 (3) 1.1699 (4) 0.0227 (7)
H3 0.5919 0.6537 1.2536 0.027\*
C4 0.57021 (16) 0.6476 (3) 1.0343 (4) 0.0222 (7)
C5 0.59471 (16) 0.6361 (3) 0.9125 (3) 0.0198 (7)
H5 0.5678 0.6343 0.8208 0.024\*
C6 0.65841 (15) 0.6271 (3) 0.9228 (3) 0.0164 (7)
C7 0.68243 (15) 0.6269 (3) 0.7906 (3) 0.0179 (7)
H7 0.6542 0.6340 0.7012 0.021\*
C8 0.81461 (14) 0.6124 (2) 0.6421 (3) 0.0150 (6)
C9 0.91521 (15) 0.5238 (3) 0.7381 (3) 0.0216 (7)
H9A 0.9408 0.5182 0.8363 0.026\*
H9B 0.9339 0.5821 0.6855 0.026\*
C10 0.91576 (17) 0.4111 (3) 0.6625 (4) 0.0270 (8)
H10A 0.9008 0.3519 0.7190 0.040\*
H10B 0.9581 0.3935 0.6524 0.040\*
H10C 0.8887 0.4152 0.5672 0.040\*
H1A 0.7730 (17) 0.634 (3) 1.002 (2) 0.032\*
H2A 0.7325 (16) 0.680 (2) 0.600 (3) 0.032\*
H3A 0.8346 (17) 0.525 (3) 0.810 (3) 0.032\*
N1 0.74029 (12) 0.6174 (2) 0.7914 (3) 0.0169 (6)
N2 0.75628 (13) 0.6333 (2) 0.6594 (3) 0.0177 (6)
N3 0.85244 (13) 0.5580 (2) 0.7468 (3) 0.0172 (6)
O1 0.75977 (11) 0.62315 (19) 1.0781 (2) 0.0207 (5)
S1 0.83506 (4) 0.65794 (7) 0.48834 (9) 0.0201 (2)
------ --------------- -------------- ------------- -------------------- --
Atomic displacement parameters (Å^2^) {#tablewrapadps}
=====================================
----- ------------- ------------- ------------- -------------- -------------- --------------
*U*^11^ *U*^22^ *U*^33^ *U*^12^ *U*^13^ *U*^23^
Br1 0.0165 (2) 0.0815 (4) 0.0261 (2) 0.00156 (19) 0.00976 (16) 0.00158 (19)
C1 0.0186 (18) 0.0130 (14) 0.0228 (17) 0.0006 (13) 0.0052 (14) −0.0009 (13)
C2 0.0227 (19) 0.0187 (15) 0.0172 (16) 0.0008 (13) 0.0031 (14) 0.0002 (13)
C3 0.026 (2) 0.0234 (16) 0.0223 (16) −0.0019 (14) 0.0134 (15) 0.0023 (14)
C4 0.0145 (18) 0.0314 (18) 0.0221 (17) −0.0023 (14) 0.0066 (14) −0.0001 (14)
C5 0.0163 (17) 0.0239 (16) 0.0180 (16) −0.0019 (13) 0.0010 (13) 0.0013 (13)
C6 0.0173 (17) 0.0151 (14) 0.0180 (16) 0.0018 (13) 0.0065 (13) 0.0018 (13)
C7 0.0176 (17) 0.0187 (15) 0.0172 (15) 0.0003 (13) 0.0032 (13) 0.0011 (13)
C8 0.0166 (17) 0.0131 (14) 0.0162 (15) 0.0007 (12) 0.0056 (13) −0.0023 (13)
C9 0.0152 (17) 0.0291 (17) 0.0198 (16) 0.0029 (14) 0.0017 (13) 0.0026 (14)
C10 0.021 (2) 0.033 (2) 0.0274 (18) 0.0060 (15) 0.0064 (15) −0.0024 (15)
N1 0.0195 (15) 0.0169 (12) 0.0158 (13) −0.0005 (11) 0.0073 (11) 0.0006 (11)
N2 0.0166 (15) 0.0213 (13) 0.0164 (13) 0.0046 (11) 0.0060 (11) 0.0034 (11)
N3 0.0145 (14) 0.0227 (14) 0.0145 (13) 0.0024 (11) 0.0036 (11) 0.0030 (11)
O1 0.0153 (13) 0.0259 (12) 0.0205 (12) 0.0017 (10) 0.0031 (10) 0.0040 (10)
S1 0.0192 (5) 0.0253 (4) 0.0175 (4) 0.0038 (3) 0.0079 (3) 0.0032 (3)
----- ------------- ------------- ------------- -------------- -------------- --------------
Geometric parameters (Å, º) {#tablewrapgeomlong}
===========================
--------------- ----------- ------------------- ------------
Br1---C4 1.899 (4) C8---N3 1.329 (4)
C1---O1 1.345 (4) C8---N2 1.351 (4)
C1---C2 1.393 (5) C8---S1 1.703 (3)
C1---C6 1.410 (5) C9---N3 1.460 (4)
C2---C3 1.381 (5) C9---C10 1.517 (5)
C2---H2 0.9500 C9---H9A 0.9900
C3---C4 1.395 (5) C9---H9B 0.9900
C3---H3 0.9500 C10---H10A 0.9800
C4---C5 1.380 (5) C10---H10B 0.9800
C5---C6 1.391 (4) C10---H10C 0.9800
C5---H5 0.9500 N1---N2 1.384 (3)
C6---C7 1.460 (4) N2---H2A 0.879 (10)
C7---N1 1.278 (4) N3---H3A 0.876 (10)
C7---H7 0.9500 O1---H1A 0.846 (10)
O1---C1---C2 117.8 (3) N3---C8---S1 124.2 (2)
O1---C1---C6 122.5 (3) N2---C8---S1 118.0 (2)
C2---C1---C6 119.6 (3) N3---C9---C10 111.7 (3)
C3---C2---C1 120.6 (3) N3---C9---H9A 109.3
C3---C2---H2 119.7 C10---C9---H9A 109.3
C1---C2---H2 119.7 N3---C9---H9B 109.3
C2---C3---C4 119.7 (3) C10---C9---H9B 109.3
C2---C3---H3 120.2 H9A---C9---H9B 107.9
C4---C3---H3 120.2 C9---C10---H10A 109.5
C5---C4---C3 120.4 (3) C9---C10---H10B 109.5
C5---C4---Br1 120.0 (3) H10A---C10---H10B 109.5
C3---C4---Br1 119.5 (3) C9---C10---H10C 109.5
C4---C5---C6 120.6 (3) H10A---C10---H10C 109.5
C4---C5---H5 119.7 H10B---C10---H10C 109.5
C6---C5---H5 119.7 C7---N1---N2 114.8 (3)
C5---C6---C1 119.1 (3) C8---N2---N1 120.6 (3)
C5---C6---C7 118.4 (3) C8---N2---H2A 120 (3)
C1---C6---C7 122.2 (3) N1---N2---H2A 116 (3)
N1---C7---C6 122.0 (3) C8---N3---C9 123.4 (3)
N1---C7---H7 119.0 C8---N3---H3A 115 (3)
C6---C7---H7 119.0 C9---N3---H3A 119 (3)
N3---C8---N2 117.8 (3) C1---O1---H1A 114 (3)
--------------- ----------- ------------------- ------------
Hydrogen-bond geometry (Å, º) {#tablewraphbondslong}
=============================
--------------------- ---------- ---------- ----------- ---------------
*D*---H···*A* *D*---H H···*A* *D*···*A* *D*---H···*A*
O1---H1*A*···N1 0.84 (3) 2.00 (2) 2.674 (3) 137 (3)
N2---H2*A*···S1^i^ 0.88 (3) 2.47 (3) 3.316 (3) 161 (2)
N3---H3*A*···S1^ii^ 0.87 (3) 2.75 (3) 3.510 (3) 146 (3)
--------------------- ---------- ---------- ----------- ---------------
Symmetry codes: (i) −*x*+3/2, −*y*+3/2, −*z*+1; (ii) *x*, −*y*+1, *z*+1/2.
###### Hydrogen-bond geometry (Å, °)
*D*---H⋯*A* *D*---H H⋯*A* *D*⋯*A* *D*---H⋯*A*
------------------- ---------- ---------- ----------- -------------
O1---H1*A*⋯N1 0.84 (3) 2.00 (2) 2.674 (3) 137 (3)
N2---H2*A*⋯S1^i^ 0.88 (3) 2.47 (3) 3.316 (3) 161 (2)
N3---H3*A*⋯S1^ii^ 0.87 (3) 2.75 (3) 3.510 (3) 146 (3)
Symmetry codes: (i) ; (ii) .
| {
"pile_set_name": "PubMed Central"
} |
{
"pile_set_name": "PubMed Central"
} |
|
INTRODUCTION {#s1}
============
Epithelial ovarian cancer (EOC) represents the leading cause of death in gynaecological malignancies \[[@R1]\]. EOC progresses mainly through extensive intraperitoneal and/or retroperitoneal (lymphatic) tumour spread, while distant metastases are rarely seen \[[@R2]\]. So far, the biologic background of these two different modes of progression is poorly understood. Recent investigations have shown that the tumour cell differentiation status (epithelial-mesenchymal-transition) may influence the route of metastases in ovarian cancer \[[@R3]\]. Accordingly, we could previously show that strong expression of an 85 kDa fragment of the epithelial marker E-Cadherin in ovarian cancer cells appears to be associated with intraperitoneal metastases \[[@R4]\]. Contrarily, other analyses revealed that mesenchymal tumour cells exhibited locally restricted tumour growth \[[@R3]\]. Highly metabolic tumour cells in locally growing tumour masses are dependent on vascular endothelial growth factor (VEGF) mediated angiogenesis \[[@R5]\]. The VEGF family consists of VEGF-A,-C,-D and induces different cascades via their receptor-tyrosine-kinases VEGFR-1, VEGFR-2, VEGFR-3 to exhibit their various biological effects. In this context, VEGF-C and -D bind VEGFR-3 and promote additional lymphangiogenesis \[[@R6]\].
The aim of the present study was to identify the diagnostic potential of the VEGF family for identification of ovarian cancer patients being at high risk for retroperitoneal metastases. Identifying these patients may enable tailored therapeutic strategies to improve their prognosis and reduce morbidity.
RESULTS {#s2}
=======
Tumours with lymph node metastases exhibit high levels of VEGF-C expression {#s2_1}
---------------------------------------------------------------------------
Tumour samples of all included 100 patients were analysed for VEGF-A, VEGF-C and VEGF-D expression by Western Blot, including 80 (80%) with 'extensive intraperitoneal' tumour growth and 20 patients (20%) with 'predominant retroperitoneal' tumour involvement. Patient characteristics were balanced between both groups except for an expected higher rate of large bowel resections (69% vs 40% p=0.005) and lower number of resected lymph nodes (median 40 vs 49 lymph nodes p=0.043) in patients with 'extensive intraperitoneal' tumour growth (Table [1](#T1){ref-type="table"}).
###### Patient characteristics according to the pattern of tumour dissemination
Overall cohort Extensive intraperitoneal group Predominant retroperitoneal group p-value
------------------------------------------------------- ---------------- --------------------------------- ----------------------------------- ---------------
**No. of patients** 100 (100%) 80 (80.0%) 20 (20.0%)
**Age at diagnosis (years)** 0.84b
**Median** 62 62 62
**Range** 33-84 33-84 47-75
**Tumour stage** **\< 0.001a**
**pT1c** 3 (3%) 0 (0%) 3 (15%)
**pT2b** 1 (1%) 0 (0%) 1 (5%)
**pT2c** 2 (2%) 0 (0%) 2 (10%)
**pT3a** 2 (2%) 0 (0%) 2 (10%)
**pT3b** 13 (13%) 1 (1.25%) 12 (60%)
**pT3c** 79 (79%) 79 (98.75%) 0 (0%)
**Lymph node status** **\< 0.001a**
**N0** 40 (40%) 40 (50%) 0 (0%)
**N1** 60 (60%) 40 (50%) 20 (100%)
**Number of resected lymph nodes** **0.043b**
**Median** 41 40 49
**Range** 1-97 1-94 1-97
**Mode of tumour dissemination** **\< 0.001a**
**Solely intraperitoneal group (pT2b-pT3c, pN0)** 40 (40%) 40 (50%) 0 (0%)
**Predominant retroperitoneal group (pT1a-3b, pN1)** 20 (20%) 0 (0%) 20(100%)
**Two-sided group (pT3c, pN1)** 40 (40%) 40 (50%) 0 (0%)
**Grading** 0.18c
**G1** 4 (4%) 3 (3.75%) 1 (5%)
**G2** 24 (24%) 22 (27.5%) 2 (10%)
**G3** 72 (72%) 55 (68.75%) 17 (85%)
**Histologic subtype** 0.09a
**Serous** 85 (85%) 71 (88.75%) 14 (70%)
**Endometrioid** 4 (4%) 2 (2.5%) 2(10%)
**Clear cell** 3 (3%) 1 (1.25%) 2(10%)
**Mixed differentiation** 6 (6%) 5 (6.25%) 1 (5%)
**Not determined/unknown** 2 (2%) 1 (1.25%) 1 (5%)
**Surgical procedures**
**Large bowel resection** 63 (63%) 55 (68.75%) 8 (40%) **0.002a**
**Small bowel resection** 8 (8%) 7 (9.6%) 1 (5%) 0.35a
**Upper abdomen interventions**
**Liver** 29 (29%) 25 (31,25%) 4 (20%) 0.23a
**Spleen** 19 (19%) 17 (21.25%) 2 (10%) 0.20a
**Pancreas** 3 (3%) 2 (2.5%) 1 (5%) 0.50a
**Postoperative residual tumour** 0.58a
**Microscopic** 78 (78%) 62 (77.1%) 16 (80%)
**Macroscopic** 21 (21%) 17 (21.6%) 4 (20%)
**Preoperative CA 125 level (kU/l)** 0.32b
**Median** 386 401 307
**Range** 14-13089 14-13089 34-2802
**Postoperative CA 125 level (kU/l)** 0.79b
**Median** 103 101 121
**Range** 11-1267 11-1267 28-896
Overview of the clinical characteristics of all included patients with EOC (n = 100). A total of 20 patients with predominant retroperitoneal tumour dissemination are opposed to 80 patients with extensive intraperitoneal tumour involvement. Both groups were tested with Pearson-Chi-Quadrat test (a), ANOVA analysis (b), as well as with the Kruskal Wallis (c) test upon differences. Statistically significant P values are printed in bold.
FIGO: International Federation of Gynecology and Obstetrics; kU/l: kilo Units per liter.
For VEGF-A no significant difference in expression levels between the different groups was found ('predominant retroperitoneal' vs. 'extensive intraperitoneal' metastases: median 1.18 vs 1.09, p=0.50, Figures [1](#F1){ref-type="fig"}, [2A](#F2){ref-type="fig"}).
{#F1}
{#F2}
A significantly higher expression of VEGF-C could be detected in the 'predominant retroperitoneal' group compared to the 'extensive intraperitoneal' group (median 1.14 vs 0.42, p=0.001, Figures [1](#F1){ref-type="fig"}, [2B](#F2){ref-type="fig"}).
Expression of VEGF-C in tumours with solely intraperitoneal metastases within the 'extensive intraperitoneal' group significantly differed compared to tumours with retroperitoneal involvement. Median expression levels of these tumours were significantly lower compared to the 'predominant retroperitoneal' group (median 0.60 vs 1.14, p=0.03, Figure [2B](#F2){ref-type="fig"}) and higher compared to tumours with both types of metastases (median 0.60 vs 0.35, p=0.007, Figure [2B](#F2){ref-type="fig"}).
In accordance with VEGF-C, a semi-quantitative analysis on VEGF-D expression reveals similar trends among the different groups of metastases, although without statistical significance ('predominant retroperitoneal' vs extensive 'intraperitoneal metastases': mean 1.33 vs 0.96, p=0.09, Figures [1](#F1){ref-type="fig"}, [2C](#F2){ref-type="fig"}).
VEGF-C correlates with postoperative residual tumour {#s2_2}
----------------------------------------------------
No significant correlations between VEGF-C and VEGF-A,-D expression as well as clinical characteristics like prae-/post CA125 values, age and grading were noted. However, VEGF-C exhibits a significant correlation to postoperative residual tumour (Pearson: r=0.225, p= 0.025, Table [2](#T2){ref-type="table"}).
###### Correlation of VEGF-A,-C,-D expression and clinical/pathological patient characteristics
VEGF-C expression VEGF-D expression VEGF-A expression
----------------------------------- ------------------------ ------------------------- ------------------------
**VEGF-A expression** r = 0.062; p=0.538 r = -0.004; p=0.969
**VEGF-D expression** r = 0.026 ; p=0.802
**Pre-CA125 value** r = 0.024; p=0.821 r = 0.022 ; p=0.839 r = 0.154; p=0.150
**Post-CA125 value** r = 0.056; p=0.623 r = 0.134; p=0.256 **r = 0.242; p=0.033**
**Age** r = - 0.058 ; p=0.569 **r = -0.205; p=0.045** r = 0.123; p=0.223
**Grading** r = 0.018 ; p=0.861 r = - 0.045; p=0.663 r = 0.061; p=0.544
**Postoperative residual tumour** **r = 0.225; p=0.025** r = 0.033; p=0.753 r = 0.0001; p=0.999
Postoperative residual tumour was categorized in 'postoperative no tumour residual', 'microscopic tumour residual', 'tumour residual \<1cm' and 'tumour residual \> 1 cm'. p = p-value; correlation coefficient: r = Pearson. Bold values indicate statistically significant correlations.
In addition, decreasing VEGF-D values were noted with rising age of patients (Pearson: r= -0.205, p=0.045, Table [2](#T2){ref-type="table"}), while VEGF-A is correlated with post-CA125 values (Pearson: r= 0.242, p=0.033, Table [2](#T2){ref-type="table"}).
Patients with high VEGF-C expression had a significantly worse overall survival {#s2_3}
-------------------------------------------------------------------------------
In survival analyses, no prognostic differences with regard to modes of progression were noted in terms of progression-free survival (PFS, log Rank p=0.80, [Supplementary Figure 1A](#SD1){ref-type="supplementary-material"}. and overall survival (OS), log Rank p=0.79, [Supplementary Figure 1B](#SD1){ref-type="supplementary-material"}).
Subsequently, the overall patient cohort was divided into two groups according to the median VEGF-C expression values detected by Western Blot analysis. The group of patients with low VEGF-C expression (n=50) consisted of 5 patients with 'predominant retroperitoneal' (10%), 17 patients with 'solely intraperitoneal' (34%) and 28 patients with 'both types' of metastases (56%). In contrast, the group with high VEGF-C expression is composed of 15 patients with 'predominant retroperitoneal' (30%), 23 patients with 'solely intraperitoneal' (46%) and 12 patients with 'both types' of metastases (24%, [Supplementary Figure 2](#SD1){ref-type="supplementary-material"}).
Regarding PFS there was no statistically significant association with high VEGF-C expression compared to low expression (median 23 vs 23 months; HR1.44, 95%-CI 0.89-2.31, p=0.13; Log Rank p=0.13, Figure [3A](#F3){ref-type="fig"}). However, patients with high VEGF-C expression values had a significantly impaired OS with 41 versus 56 months (HR 2.02, 95%-CI 1.12-3.63, p=0.019; Log Rank p=0.016, Figure [3B](#F3){ref-type="fig"}).
{#F3}
DISCUSSION {#s3}
==========
Following the results of the present explorative study including tumour samples from 100 patients with epithelial ovarian cancer, we could identify VEGF-C as a potential molecular marker attributing the risk for retroperitoneal tumour progression to ovarian cancer patients.
VEGF-C is significantly higher expressed in the group of patients with 'predominant retroperitoneal' metastases compared to the group of 'extensive intraperitoneal' tumour involvement. Although the different modes of tumour progression were not directly correlated with differences in survival, a significantly shorter overall survival for patients with high VEGF-C expression was noted, which supports a possible prognostic impact of VEGF-C mediated retroperitoneal metastases. Nevertheless, as we know that patients with extensive intraperitoneal tumor dissemination usually have an impaired prognosis, it underlines that prognostic differences are not regulated by one single factor and that multiple other factors influence the prognosis of patients. Accordingly but not-significant, VEGF-D shows similar trends among the different groups of metastases, while expression levels for VEGF-A did not differ in our cohort.
As previously described, the VEGF-C/VEGFR-3 pathway acts as an enhancer of ovarian cancer progression through autocrine and paracrine mechanisms, hence offering a potential target for therapeutic interventions \[[@R9]\]. Higher intratumour VEGF-C expression and worse clinical outcome of ovarian cancer patients is associated with the presence of lymphatic space invasion which may act as an indicator of lymph node metastases \[[@R10]\]. According to this observation, we could demonstrate that VEGF-C expression in tumours with solely intraperitoneal metastases significantly differed compared to tumours with retroperitoneal involvement.
In line with previous investigations \[[@R3]\], we hypothesise that tumours with predominant retroperitoneal metastases seem to have a more mesenchymal tumour cell biology with high VEGF-C levels and low E-Cadherin expression in contrast to tumours with 'extensive intraperitoneal' tumour distribution \[[@R4]\] (Figure [4](#F4){ref-type="fig"}). Gene signatures in high--grade serous ovarian cancer have been identified with high VEGF-C mRNA levels in the 'mesenchymal' molecular subtype, which shows significantly shorter survival rates compared to 'immunoreactive', 'differentiated' or 'proliferative' subtypes \[[@R11]\]. In contrast, tumours in the subgroup with 'solely intraperitoneal' metastases are characterised by an epithelial phenotype with high E-Cadherin \[[@R4]\] as well as low VEGF-C and --D expression, which might promote intraperitoneal tumour dissemination (Figure [4](#F4){ref-type="fig"}). VEGF-C expression levels in tumours exhibiting intraperitoneal as well as retroperitoneal tumour involvement were lower compared to patients with solely intraperitoneal metastases which may indicate a dedifferentiation of tumour cells in tumours with both modes of progression compared to tumours with solely intraperitoneal or only predominant retroperitoneal metastases. As tumours in more advanced tumour stage with both types of metastases are characterised by a significantly lower VEGF-C expression compared to patients with predominant retroperitoneal metastases, this may indicate a potential of VEGF-C to estimate the risk for early retroperitoneal metastases. Reliable identification of patients in early stage at high risk for retroperitoneal tumour progression would be highly interesting to tailor therapeutic strategies regarding surgery and systemic therapy.
{#F4}
Monoclonal antiangiogenic therapy with bevacizumab, an antibody that targets VEGF-A, has been incorporated into guidelines for firstline therapy of EOC in Europe with a proven effect on progression-free survival in FIGO stages IIIB, IIIC und IV (FIGO classification before 2014, UICC 7^th^ edition) \[[@R12]\]. Prognostic markers and strategies for patient selection regarding antiangiogenic therapy have not been established so far. Therefore, it would be of special interest to be able to identify patients in early tumour stages, who might also benefit from additional anti-angiogenic therapy. In contrast to bevacizumab, multikinase inhibitors address several intracellular tyrosine kinases and initiate additional therapeutic effects within the antiangiogenic treatment concept \[[@R13]\]. In this context, cediranib serves as an example by targeting VEGFR-3 in addition to VEGFR-1, and -2. As VEGF-C and --D exert their biological effects mainly via the VEGFR-3-pathway in addition to VEGF-A, multikinase inhibitors may have the potential to be used as specific antiangiogenic drug for tumours with high VEGF-C- levels at high risk for retroperitoneal metastases. Furthermore, as a future perspective its expression might be used to characterise patients in early cancers who do not require radical systematic lymphadenectomy in staging laparotomies due to their low biological probability of lymph node metastases.
The identification of molecular markers for individual therapy is urgently required to tailor therapy of patients with EOC more individually. Even though an intraoperative decision for lymphadenectomy depending on molecular VEGF-expression is not yet possible, VEGF-C could represent a molecular marker to identify patients with high risk of lymphatic metastases who benefit from retroperitoneal surgery and specific antiangiogenic therapy (e.g. following diagnostic laparoscopy). However, further experimental and clinical investigations will be necessary to verify the benefit of an antiangiogenic therapy in form of individual VEGF-C and -D pathway inhibitors.
MATERIALS AND METHODS {#s4}
=====================
Patients {#s4_1}
--------
A total of 100 patients with EOC treated at the University Medical Centre Hamburg-Eppendorf between 1995 and 2012 were included in the study. All patients underwent primary debulking surgery according to current German guidelines \[[@R7]\]. Patient cohorts were classified according their mode of tumour progression with two different types of metastatic spread. While a total of 20 patients had predominant retroperitoneal tumour involvement ('predominant retroperitoneal', pT1a-pT3b, N1), the other 80 patients exhibited extensive intraperitoneal metastases in tumour stage pT3b-pT3c, pN0 and pT3c pN1 ('extensive intraperitoneal'). This cohort was further subdivided into a subgroup of 40 patients with extensive intraperitoneal tumour dissemination without lymph node involvement (pT3b-pT3c pN0, solely intraperitoneal metastases) and a subgroup of 40 patients with extensive intraperitoneal and retroperitoneal metastases (pT3c, N1, both types of metastases).
Detailed patient characteristics are presented in Table [1](#T1){ref-type="table"}. All patients gave written informed consent to access their tissue and review their clinical records according to our investigational review board and ethics committee guidelines (\#200814). Clinical data was retrieved from a detailed database providing information on clinicopathologic factors, histologic classifications and therapeutic procedures. Clinical outcome of all patients was followed from date of first diagnosis until the end of 2014.
Tissue samples {#s4_2}
--------------
As described previously \[[@R4]\] tissue samples were obtained intraoperatively and immediately stored in liquid nitrogen as fresh frozen samples. The histological characteristics of each sample were evaluated on cryo-cut and haematoxylin-eosin-stained sections. The tissue was tailored if necessary to obtain at least 70% tumour cells in the sample used for protein extraction.
Protein extractions {#s4_3}
-------------------
As described previously \[[@R4]\] samples of approximately 100 mg were cut from the tissue and pulverised using a micro-dismembrator (Braun-Melsungen, Melsungen, Germany) for 2 minutes and 45 seconds at 200 r.p.m. Proteins were lysed in ice-cold sample buffer (50mM Tris pH 6.8, 1% sodium dodecyl sulphate (SDS), 10% sucrose and 10 μl/ml protease inhibitor cocktail (Sigma, Taufkirchen, Germany)). The protein concentration was determined following standard protocols and using bovine serum albumin protein standards.
Western blot analysis {#s4_4}
---------------------
For Western blot analysis equal amounts of protein lysate (20 μg) were analysed per well. The proteins were separated on a 10 percent polyacrylamide gel (11% Glycerin, aqua dest., 10% SDS, Tris and 10% Acrylamid) and transferred to polyvinylidene difluoride membranes (Merck Millipore KGaA Darmstadt, Germany). After that, the membranes were blocked with 5% nonfat dry milk (Bio-Rad Laboratories, USA) in TBST (TBS: 50mM Tris, 150mM NaCl, pH 7.4 with 10% TWEEN 20), washed in TBST and incubated with the primary antibody. The monoclonal antibody for VEGF-A (Abcam ab46154), 1:8000, Cambridge United Kingdom) was diluted in 5% nonfat dry milk in TBST, whereas the monoclonal antibodies for VEGF-C (sc374628, 1:200, Santa Cruz Biotechnology, Heidelberg, Germany) and VEGF-D (H144 sc13085, 1:1000, santa cruz) were diluted in 5% BSA in TBST. All primary antibodies were incubated at 4°C overnight. After incubation, membranes were washed in TBST and incubated with the corresponding secondary antibody (anti rabbit sc 2054, for VEGF-A and --D; anti mouse sc 2005 for VEGF-C, all purchased from santa cruz) in 1.5% nonfat dry milk in TBST for one hour at room temperature. Finally, after washing in TBST, the detection was carried out with the SuperSignal West Pico chemoluminescent kit (Thermo Scientific, Rockford, USA) to visualise the protein expression on FUJI-super RX medical x-ray films (Tokyo, Japan).
Band intensities were quantified by densitometry (GS-700 Imaging Densitometer, BioRad, Munich, Germany). Protein lysates from the breast cancer cell line MCF7 were used as positive controls for VEGF-A, --C and -D. All detected bands of the investigated proteins except for VEGF-D were standardised with the positive controls, which were defined as 100 percent. Expression values were normalised to GAPDH (Santa Cruz FL 335; 1:5000), as loading control. VEGF-D was analysed in a semi-quantitative way because of unmanageable background signals. For VEGF-C, bands at 60 kDa were detected in our collective. As previously reported we assessed the 60 kDa band as heterodimer of the active VEGF-C monomers of 29 kDa and 31 kDa \[[@R8]\].
Statistical analysis {#s4_5}
--------------------
Values representing VEGF-A, VEGF-C and VEGF-D expression levels were logarithmised and tested among the different cohorts by ANOVA analysis, as well as post hoc with the Least Significant Difference Test (LSD). VEGF-C expression was correlated by Pearson with the VEGF-A and -D as well as with clinical- and pathological parameters. P-values \<0.05 were considered to be statistically significant. Boxplots were diagrammed on the basis of absolute VEGF expression levels. Univariate survival analyses were made by Kaplan Meier method and cox regression and are not independent findings. All statistical analyses were carried out with SPSS (IBM SPSS Statistics version 22 for windows).
SUPPLEMENTARY FIGURES {#s5}
=====================
**CONFLICTS OF INTEREST**
The authors declare no conflicts of interest.
| {
"pile_set_name": "PubMed Central"
} |
1. Introduction {#sec1}
===============
Radical cystectomy (RC) remains the standard of care for local treatment of non metastatic muscle invasive bladder cancer (MIBC). However, cancer specific survival is approximately 50% depending on the presence of extravesical extension and/or lymph nodes metastases \[[@bib1]\]. In daily practice, more than 50% of patients die of distant metastases within two years after cystectomy, suggesting the presence of micro-metastases at time of surgery \[[@bib2]\]. Therefore, peri-operative chemotherapy (adjuvant or neoadjuvant) has been developed to increase overall survival, with an absolute benefit of 5% reported for neoadjuvant chemotherapy (NAC) and international guidelines recommend NAC based on the available level I evidence \[[@bib3],[@bib4]\]. The chemotherapy administration time and the optimal chemotherapy regimen to be delivered remain open to discussion. As dose-dense methotrexate, vinblastine, doxorubicin and cisplatin (dd-MVAC) has been shown to be associated with higher response rates in bladder metastatic disease \[[@bib5]\], a better efficacy can also be suspected in the peri-operative setting.
Recently, Choueiri et al. and Plimack et al. reported interesting results from two phase II trials using dd-MVAC as neoadjuvant chemotherapy in MIBC. After three to four cycles, the pathologic downstaging (pT1 N0M0) was quite similar (49% and 53%), the pathologic complete response rates (pT0) were 26% and 38%, respectively \[[@bib6],[@bib7]\]. Our objective was to design a randomized phase III controlled study comparing the efficacy of GC and dd-MVAC in term of progression-free survival in patients for whom chemotherapy has been decided, before or after radical cystectomy.
2. Material and methods {#sec2}
=======================
2.1. - Study design {#sec2.1}
-------------------
This randomized phase III study assesses the efficacy of dd-MVAC and GC peri-operative chemotherapy (adjuvant or neoadjuvant) in patients with bladder cancer disease defined by a T2, T3 or T4a N0 (≤10 mm on CT scan) M0 staging for patients receiving neoadjuvant chemotherapy or pT3 or pT4 or pN+ and M0 for patients receiving adjuvant chemotherapy. Secondary endpoints include overall survival, safety, response rate in the neoadjuvant setting. From February 2013 to March 2018, a total of 500 patients have been randomized in the French GETUG/AFU V05, controlled phase III trial, including 28 participating centers with referent urologist and oncologist investigators ([Fig. 1](#fig1){ref-type="fig"}).Fig. 1Participating centers of the GETUG/AFU V05 multicentre, randomised phase III trial.Fig. 1
As previously mentioned, the peri-operative chemotherapy schedule proposed was:
**[Standard Arm A: GC]{.ul}**.-GEMCITABINE 1250 mg/m2: Day 1 and Day 8-CISPLATIN 70 mg/m2: Day 1**Every 3 weeks, for a total of 4 cycles**
**[Experimental Arm B: dd-MVAC]{.ul}**.-METHOTREXATE 30 mg/m2: Day 1-VINBLASTINE 3 mg/m2: Day 2-DOXORUBICIN 30 mg/m2: Day 2 - CISPLATIN 70 mg/m2: Day 2 - G-CSF: Day 3 to Day 9**Every 2 weeks, for a total of 6 cycles**
The chemotherapy response was evaluated according to RECIST 1.1 criteria.
The treatment toxicity was evaluated according to NCI CTCAE (v 4.0).
The progression-free survival estimation rate of this trial was determined at 3 years.
In our prospective randomized study all the patients were followed for 5 years.
2.2. - Study Procedures {#sec2.2}
-----------------------
At baseline before screening, a CT with contrast of the chest, abdomen and pelvis was performed for all patients, in association with a systematic bone scan and a complete biological evaluation. Follow up visits and their schedules and measurements are clearly reported in [Fig. 2](#fig2){ref-type="fig"}.Fig. 2Study procedures, schedule and parameters of the patients follow-up.Fig. 2
In our clinical trial, a dose reduction of chemotherapy in case of toxicity was allowed. Considering the GC group (standard arm), the cisplatin dose was adapted to renal function (creatinine clearance \> 60 ml/mn: 70 mg/m2; between 50 and 60 ml: 50 mg/m2; between 40 and: 35 ml mg/m2; creatinine clearance \< 40 ml/mn: end of the chemotherapy). As regards haematologic toxicity (neutropenic fever), a dose reduction of 15% was recommended for the two molecules. Considering the dd-MVAC group (experimental arm), the cisplatin dose was adapted to renal function as previously described. A dose reduction of 15% was also recommended for the four molecules in case of grade 4 toxicity and the chemotherapy stopped in the absence of recovery within 14 days.
The study protocol was approved by the Ethics Committee CPP ROUEN NO on 19 April, 2012 and the competent authority on 27 February, 2012. All patients signed the informed consent form to be enrolled in this randomized phase III study ***(Clinical trial registry: clinicaltrials.gov - NCT 018 12369).***
After the exclusion of any analysis of 7 patients who did not meet the inclusion/exclusion criteria, it remained 493 patients for the primary analysis (intent-to-treat population). Baseline characteristics by chemotherapy arm are reported in [Table 1](#tbl1){ref-type="table"}, whereas tumour staging at randomization by type of peri-operative chemotherapy are detailed in [Table 2](#tbl2){ref-type="table"}.Table 1**Baseline characteristics by chemotherapy arm**.Mean (standard deviation) for quantitative data. Frequency (percentage) for qualitative data. Comparisons between GC and dd-MVAC groups are performed with a Student T-test or Chi-2 test. P-value \< 0.05 would assume a statistical difference between GC and dd-MVAC groups.Table 1GCdd-MVACp-valuen = 245n = 248**[Demography]{.ul}**Age63 (7.6)62.6 (7.9)0.62SexMale206 (84%)202 (81%)0.51Female39 (16%)46 (19%)**[Physical examination]{.ul}**Body Mass Index26.6 (4.7)26 (4.4)0.16Body Surface Area1.9 (0.2)1.9 (0.2)0.52WHO status0171 (70%)165 (67%)0.59172 (29%)82 (33%)Not done2 (1%)1 (0%)**[Medical History]{.ul}**No10 (4%)6 (2%)0.43Yes235 (96%)242 (98%)whoseneuropathy3 (1%)1 (0%)0.60hearing disorder35 (15%)46 (19%)0.28high blood pressure100 (43%)89 (37%)0.23infarc9 (4%)11 (5%)0.88coronary insuff.5 (2%)9 (4%)0.45diabetes14 (6%)4 (2%)**0.03**tobacco198 (84%)197 (81%)0.48aromatic amines14 (6%)7 (3%)0.16[Biology and renal function]{.ul}Hemoglobin (g/100 mL)14.3 (6.8)13.9 (1.5)0.33Neutrophil polynuclear cells (1000/mm^3^)7.5 (34.7)6.9 (22.3)0.83Platelets (1000/mm^3^)272.6 (78.2)274.3 (85.4)0.81Total bilirubin (mg/L)5.0 (2.6)5.2 (2.5)0.34ALT (UI/L)23.6 (14.6)22.8 (10.5)0.49AST (UI/L)20.4 (7.2)20.6 (7.2)0.75Alkaline phosphatase (UI/L)77.9 (26.2)77.9 (31.5)0.97Creatinine (mg/L)9.6 (3.3)9.1 (2.1)0.05Clearance of creatinine (mL/min)89.3 (27.6)90.2 (25)0.69Table 2**Staging at randomization by type of peri-operative chemotherapy and arm**Frequency (percentage). For adjuvant chemotherapy, pTNM staging is performed on cystectomy. For neoadjuvant chemotherapy, TNM staging is performed on TURBT. Staging according to 2009 TNMclassification.Table 2Adjuvant chemotherapyNeoadjuvant chemotherapyGCdd-MVACGCdd-MVACn = 26n = 30n = 219n = 218TumourT11 (4%)1 (3%)00T2a01 (3%)141 (64%)138 (63%)T2b3 (12%)2 (7%)66 (30%)59 (27%)T3a8 (31%)12 (40%)4 (2%)7 (3%)T3b3 (12%)6 (20%)4 (2%)5 (2%)T4a11 (42%)8 (27%)4 (2%)9 (4%)NodesN07 (27%)12 (40%)219 (100%)218 (100%)N112 (46%)7 (23%)00N27 (27%)11 (37%)00MetastasisM026 (100%)30 (100%)219 (100%)218 (100%)
2.3. - Objectives {#sec2.3}
-----------------
The primary objective of our study was the evaluation of efficacy in terms of progression-free survival at three years of the combination of dose dense methotrexate, vinblastine, doxorubicin and cisplatin (dd-MVAC) versus gemcitabine and cisplatin (GC) as peri-operative chemotherapy for locally advanced -transitional cell carcinoma of the bladder. Final results for primary endpoint will also be available in mid-2021.
Secondary objectives of the trial were to assess toxicity NCI CAE (v 4.0), to assess response rate (RR) in patients treated in the neoadjuvant setting, to assess overall survival (OS), to assess time to progression (TTP) and to study the correlation between response rate, time to progression, overall survival and biological parameters.
2.4. - Patients selection {#sec2.4}
-------------------------
Key inclusion criteria were primary tumour of the bladder; histologically confirmed infiltrating urothelial carcinoma (epidermoid and/or glandular variants are accepted if combined with TCC); disease defined by a T2, T3 or T4a N0 (lymph node ≤ 10 mm on CT scan) M0 staging for patients receiving neoadjuvant chemotherapy or pT3 or pT4 or pN + whatever pT and M0 for patients receiving adjuvant chemotherapy; 18 ≤ age ≤ 80 years; general condition 0 or 1 as per the WHO scale; absence of previous chemotherapy for muscle-invasive disease; haematological function: haemoglobin \>11 g/dl, neutrophils ≥ 1500/mm3, platelets ≥ 100,000/mm3; liver function: grade 0 ASAT and ALAT, grade 0 alkaline phosphatases, normal bilirubin; renal function: calculated (or measured) creatinine clearance ≥ 40 ml/min; patients covered by a social security scheme and having read the information sheet and signed the informed consent form.
Key exclusion criteria were pure adenocarcinoma or pure epidermoid carcinoma or mixed or pure small-cell neuroendocrine carcinoma; ventricular ejection fraction \<50%; history of cancer in the 5 years prior to entry in the trial other than basal cell skin cancer or in situ carcinoma of the cervix; male or female patients not agreeing to use an effective method of contraception throughout the duration of treatment and for 6 months after treatment discontinuation; pregnant women, or female subjects liable to become pregnant or currently breast-feeding; patient already included in another therapeutic trial on an investigational medicinal product; persons deprived of their freedom or under judicial protection (including guardianship); unable to receive medical follow-up during the trial owing to geographical, social or psychological reasons.
2.5. - Statistical analyses {#sec2.5}
---------------------------
The total number of patients expected for this randomized multicenter phase III study was 500, with 250 on each arm. This sample size is sufficient to demonstrate that combination dd-MVAC improves progression-free survival compared to GC with a maximum hazard ratio hypothesis of 0.74 and an alpha risk of 5% and a power of 80%. This sample size allows an interim analysis of the primary endpoint is planned after the occurrence of 174 events. The primary endpoint progression-free survival will be estimated using the Kaplan-Meier method. In order to take into account the mode of administration of the chemotherapy (neo-adjuvant or adjuvant) and the involvement or not of the lymph nodes, the efficacity of the chemotherapy dd-MVAC vs GC will be evaluated by a stratified log-rank test. The adjusted hazard ratio (dd-MVAC/GC) and its 95% confidence interval will be estimated by a proportional hazard Cox model adjusted for the therapeutic option (neoadjuvant vs adjuvant) and the lymph nodes involvement. The hypothesis of proportional hazard of the model will be tested by the Lin method.
2.6. - Ancillary study {#sec2.6}
----------------------
The identification of predictive biomarkers may help urologists in the selection of patients to benefit most likely from NAC. Different studies have concentrated on the assessment of one genetic marker, nevertheless none of them has allowed the validation of a reliable marker in clinical practice. Therefore, ERCC1, telomere length, BRCA1 mRNA expression and p53 mutation did not demonstrate any correlation with response, toxicity or survival \[[@bib7],[@bib8]\].
Concomitant ancillary study has started, focusing on DNA repair genes (ERCC2, ATM, FANCC, RB1) and molecular subtypes determined by gene expression profiling to compare and potentially validate these biomarkers previously proposed for muscle invasive bladder tumour sensitivity to neoadjuvant chemotherapy \[[@bib9], [@bib10], [@bib11]\].
3. Discussion {#sec3}
=============
Neoadjuvant chemotherapy is typically validated in patients with cT2-cT4 N0 muscle invasive bladder cancer, with a normal renal function allowing the use of standard doses of cisplatin \[[@bib3],[@bib4]\]. Patients with locally advanced disease (N1) do require upfront chemotherapy rather than neoadjuvant treatment. Despite the pivotal randomized studies and meta-analyses demonstrating the survival benefit of cisplatin-based combination chemotherapy before cystectomy \[[@bib12],[@bib13]\], there was a low implementation of this approach in daily practice, even if the use of NAC has been increasing during the 2010s, suggesting that continuous efforts are required to convince more urologists and oncologists to use it \[[@bib14], [@bib15], [@bib16]\].
The lack of referral to medical oncologist may be an important factor for under-utilization of neoadjuvant schedule \[[@bib15]\], nevertheless there is always a need of clarification for optimal chemotherapy regimen. Level I evidence regarding the NAC overall-survival benefit was reported with cisplatin-based regimens that are no longer currently used \[[@bib12],[@bib13]\], in fact standard MVAC has been supplanted in the metastatic setting by less toxic regimens as gemcitabin-cisplatin (GC) and dose-dense MVAC (dd-MVAC). Zargar et al. have reviewed the clinical data of 319 patients with cT3-4a N0M0 bladder cancer who underwent NAC before cystectomy from 2000 to 2015 in 20 institutions. One hundred patients received dd-MVAC, whereas 219 patients were treated with GC \[[@bib17]\]. Baseline characteristics were similar between the two groups except for age (patients who received dd-MVAC were younger) and the proportion of variant histology features (higher in the dd-MVAC group). A significantly lower rate of pathological complete response (ypT0N0) was observed in the GC group (14.6% vs 28.0%, p = 0.005). Similarly, the pathological partial response rate (ypT1N0 or less) was 30.1% for GC arm compared to 41.0% for dd-MVAC arm (p = 0.07). These results suggest that dd-MVAC could be the optimal regimen for NACT with a translated longer overall survival for patients (7 years vs 4.6 years, p = 0.001).
The primary endpoint of the French GETUG/AFU Vesper trial is the disease-free survival at 3 years. As the last patient was randomized in March 2018, the primary endpoint results are attempted for mid-2021. Nevertheless, the authors underline the interest of secondary objectives: chemotherapy safety with side effects analysis, response rate in patients treated in the neoadjuvant setting and time to MIBC progression. The dose-dense MVAC schedule is associated with higher response rates in bladder metastatic disease, the real question today is to confirm such benefit in the peri-operative setting.
A substantial high rate of morbidity (66%) or risk of perioperative mortality (4%) and a negative impact on quality of live have been reported after cystectomy. The subgroup of patients downstaged to pT0 after neoadjuvant chemotherapy usually achieve a survival benefit beyond non complete responders, however the real benefit of radical surgery remains today unclear \[[@bib18]\]. Robins et al. reported an interesting series of 48 patients at the Colombia University Irving Medical Center with muscle invasive bladder cancer that were cT0 after neoadjuvant chemotherapy and also refused radical cystectomy \[[@bib19]\]. Five-year cancer specific survival was 87%, disease free survival was 58% and cystectomy-free survival was 79%. A total of 19 patients (46%) relapsed with 5.4 month median recurrence time. Finally, the bladder preservation for patients with complete clinical response after NAC is not recommended currently, but the pathologic downstaging (pT1 N0M0) and the rate of pathologic complete response (pT0) may be significant prognostic factors in NAC response. Vesper Trial results has to confirm the interesting data from two phase II trials using dd-MVAC as neoadjuvant chemotherapy recently reported by Choueiri et al. and Plimack et al. \[[@bib6],[@bib7]\].
Today, immunotherapy is becoming an interesting option in the second line treatment of metastatic urothelial carcinoma and PD-L1 expression by IHC could be correlated to therapeutic response in metastatic or locally advanced urothelial carcinoma treated with PD-L1 antibody suggesting personalized medicine \[[@bib20],[@bib21]\]. Powles et al. obtained a 43% (95% CI \[26%, 63%\]) response rate to atezolimumab in patients with positive PD-L1 tumour (IC or TC or both, at 5% positivity threshold) and a 11% (95% CI \[4%, 26%\]) response rate in patients with negative PD-L1 tumours \[[@bib22]\]. Tomorrow, the oncological challenge will be to determine the best therapeutic option in the peri-operative setting: chemotherapy combined with immunotherapy or immunotherapy in the whole population or personalized treatment with chemotherapy or immunotherapy according to the molecular profiling of tumours. Before the validation of such therapeutic approaches, it is very important to determine the gold standard chemotherapy between gemcitabine and cisplatin (GC) and dose dense methotrexate, vinblastine, doxorubicin and cisplatin (dd-MVAC). In conclusion, the authors hope that final Vesper Trial results will be strongly significant to permit guidelines updating.
Declaration of competing interest
=================================
The authors declare no competing interests with the manuscript.
The authors strongly thank Cecile Pourcher, Frederique Caillot and Assia Ferrane, Rouen University Hospital Clinical Research Assistants, for their active participation in the multicenter monitoring of this phase III trial.
The GETUG/AFU V05 VESPER randomized phase III study was supported by a grant from the French Ministry of Health (PHRC 2011-037). The Northwest Data Center (CTD-CNO) is acknowledged for managing the data. It is supported by grants from the French National League Against Cancer (LNC) and the French National Cancer Institute (INCa).
| {
"pile_set_name": "PubMed Central"
} |
CASE REPORT
===========
A 36-year-old man visited the emergency department. He had suffered from a sudden onset of chest pain, headache, palpitation, and dizziness. His pulse was regular, bounding, and at a rate of 119 per minute, and his blood pressure was at 190/120 mmHg. Electrocardiography and echocardiography were normal. Chest X-ray showed an oval-shaped mass lesion on the left paravertebral area ([Fig. 1A](#F1){ref-type="fig"}). Chest computed tomography (CT) and magnetic resonance imaging (MRI) showed a left paravertebral mass in the region of T4-T6, measuring at 3×6 cm, which was not invasive to the adjacent organs ([Fig. 1B](#F1){ref-type="fig"}). There was no specific finding on abdomen CT. The biochemical studies showed elevated levels of serum norepinephrine (12.905 ng/mL; normal limit \[nl\], 0 to 0.8 ng/mL), urine norepinephrine (4,946 µg/day; nl, 15 to 80 µg/day), urine metanephrine (4.2 mg/day; nl\<0.8 mg/day), and urine vanillylmandelic acid (26.6 mg/day; nl, 0 to 8 mg/day). The serum and urine epinephrine levels were normal. The presence of the tumor was confirmed using an I-123 metaiodobenzylguanidine (MIBG) scan, along with its isolated nature ([Fig. 2](#F2){ref-type="fig"}). In order to control the blood pressure and prepare for surgery, we administered α and β blocking agents to the patient. During the first three weeks, Concor (5 mg), Norvasc (5 mg), Atacand (8 mg), and Xyrem (0.25 mg) were administered. When visited the outpatient department, his blood pressure was near normotensive, but tachycardia and chest discomfort were still present. Therefore, we changed the medication to Cardura-XL (4 mg) for five weeks. The patient no longer felt chest discomfort, but tachycardia was still present. The heart rate was controlled (100 beat/min) after we administered Concor (2.5 mg). Two months after his first visit, we performed the operation. The tumor was found lying on the left paravertebral sulcus, which was hypervascular and well demarcated ([Fig. 3A](#F3){ref-type="fig"}). We performed thoracoscopic resection, but there were transient changes in the blood pressure, which ranged from 130/90 to 200/120 mmHg; therefore, we converted to thoracotomy. In the intraoperative period, we controlled the blood pressure with Brevibloc and Perdipine. After the tumor was resected, the blood pressure dropped sharply. Therefore, we stopped administering antihypertensive drugs, and then the blood pressure remained stable. The resected tumor measured 6.5×4.0×3.0 cm and weighed 47.5 g. Histopathological examination confirmed the tumor to be pheochromocytoma ([Fig. 3B](#F3){ref-type="fig"}). During the postoperative period, the patient remained normotensive without the need of antihypertensive drugs. He was discharged on the postoperative day 8. Two months later, the follow-up biochemical studies showed decreased levels of serum norepinephrine (1.108 ng/mL), urine norepinephrine (149.3 µg/day), urine metanephrine (1 mg/day), and urine vanillylmandelic acid (7.1 mg/day). The patient was asymptomatic with normal blood pressure and without taking antihypertensive drugs for more than 1 year.
DISCUSSION
==========
Pheochromocytoma develops from the chromaffin cells of the sympathetic nervous system \[[@B1]-[@B3]\]. In most cases, pheochromocytoma originates in the adrenal gland, but it can arise anywhere on the paraganglion tissues that contain chromaffin \[[@B1]\]. Extra-adrenal pheochromocytoma, often known as paraganglioma, is rare in adults (approximately 10%) and it occurs in thorax in 1% of the cases \[[@B1],[@B4],[@B5]\]. Pheochromocytomas and paragangliomas are associated with several well-known inherited syndromes such as multiple endocrine neoplasia type IIA, IIB, or III, von Hippel-Lindau disease, von Recklinghausen\'s disease, and Sturge-Weber syndrome. Patients with any of these syndromes have familial histories \[[@B4],[@B6],[@B7]\]. Most pheochromocytomas are catecholamine-secreting, hyperfunctioning tumors, which cause symptoms such as hypertension, headache, palpitations, sweating, and weight loss resulting from the excess production of catecholamines. Such excess production leads to elevated levels of catecholamines in the blood, or increases the level of metabolites in the urine \[[@B6]\]. Thus, blood and urine examinations are useful for diagnosing functional pheochromocytoma. Paragangliomas are easily located using a CT scan, a MRI, or an I-131 MIBG \[[@B1],[@B6],[@B8]\]. Surgical resection is the preferred treatment for paraganglioma. Before the operation, a combination of α and β adrenergic blockades are required to control blood pressure and to prevent intra-operative hypertensive crises \[[@B2],[@B3]\]. Preoperative adrenergic blockades are used until the patient\'s symptoms subside. The end point is normotension or near normotension, with slight postural hypotension and elimination or reduction of paroxysmal spells \[[@B3],[@B6]\]. We also administered α and β blocking agents to the patient for two months until the operation date. Intraoperative monitoring by an experienced anesthesia team is very important because the patient\'s blood pressure and heart rate fluctuate very severely \[[@B2],[@B6]\]. When the operators touch the tumor directly, the blood pressure and heart rate increase rapidly; therefore, blunt dissection through thoracotomy should be excluded and video-assisted thoracic surgery is performed rarely on thoracic lesions \[[@B1]\]. In our case, we attempted to perform thoracoscopic resection, but we experienced severe fluctuation of the vital signs; therefore, we converted to thoracotomy for resection of the tumor. After the tumor was resected, the blood pressure dropped sharply. Thus, the central venous line should be prepared in order to facilitate rapid volume resuscitation and an appropriate α agonist injection \[[@B6]\]. In our case, the blood pressure also dropped sharply after resection of the tumor. Because paragangliomas have a higher recurrence rate than pheochromocytomas, long-term postoperative follow-up is necessary \[[@B1],[@B6]\]. When previous symptoms reoccur, it indicates recurrence of a tumor. Annual measurement of blood pressure and the urinary catecholamine metabolites is necessary. Performing CT or MRI may be helpful, and I-123 MIBG scanning is often used for detecting metastasis and recurrence of the disease \[[@B6]\].
No potential conflict of interest relevant to this article was reported.
{#F1}
{#F2}
{#F3}
| {
"pile_set_name": "PubMed Central"
} |
{#sp1 .556}
{#sp2 .557}
{#sp3 .558}
{#sp4 .559}
{#sp5 .560}
{#sp6 .561}
{#sp7 .562}
{#sp8 .563}
{#sp9 .564}
{#sp10 .565}
{#sp11 .566}
{#sp12 .567}
{#sp13 .568}
{#sp14 .569}
{#sp15 .570}
| {
"pile_set_name": "PubMed Central"
} |
Introduction {#s1}
============
The nucleus accumbens (NAc) is a forebrain structure that regulates the vigor of reward-seeking. Its excitatory inputs likely encode motivational states and the presence of reward-associated cues (Mannella et al., [@B33]). For example, paraventricular thalamic (PVT) input regulates food-seeking behavior under conditions of hunger and threat (Labouèbe et al., [@B27]; Choi and McNally, [@B12]; Do-Monte et al., [@B15]; Cheng et al., [@B11]; Choi et al., [@B13]), while basolateral amygdala (BLA) input encodes the motivational value of reward-associated cues (Ambroggi et al., [@B2]; Stuber et al., [@B47]; Esber and Holland, [@B18]). Few studies, however, have directly compared the behavioral consequences of pathway-specific manipulations, so it remains unclear how each input distinctly contributes to effective reward-seeking. Maladaptive alterations in the strength of NAc inputs are thought to underlie discrete aspects of psychopathologies, including the aversive symptoms of drug withdrawal (Neumann et al., [@B36]; Zhu et al., [@B52]) and stress susceptibility in animal models of depression (Bagot et al., [@B4]). Thus, pathway-specific inactivation of these inputs is critical to gaining insight into how this circuitry contributes to healthy and unhealthy behavior alike.
Archaerhodopsin (ArchT)-mediated photoinhibition of axon terminals is commonly used to test the involvement of specific long-range neural projections in behavior. Sustained activation of this outward proton pump in axon terminals effectively decreases evoked transmitter release but alkalizes affected axon terminals, which has the unintended consequence of increasing spontaneous vesicle release (El-Gaby et al., [@B17]; Mahn et al., [@B30]). It is unclear whether ArchT's off-target effects undermine the interpretation of its behavioral effects or still permit assessment of pathway-specific function. If the aberrant spontaneous vesicle release recruits local circuit feedforward inhibition, as previously suggested (Mahn et al., [@B30]), the intended pathway-specific nature of the manipulation may be compromised.
The shortcomings of ArchT terminal inhibition have been well characterized in acute slice preparations (El-Gaby et al., [@B17]; Mahn et al., [@B30]), but *in vivo* applications of this technique are still widely used to study the circuit-level basis of specific behaviors (Herrera et al., [@B22]; Yamamoto and Tonegawa, [@B51]; Mangieri et al., [@B32]), particularly concerning NAc inputs (Stefanik et al., [@B45]; Zhu et al., [@B52]; Reed et al., [@B41]; Trouche et al., [@B49]). Thus, there remains a need to validate the pathway-specific nature of this manipulation in behaving animals.
Here, we compare photoinhibition targeted to the axon terminals or cell bodies of NAc inputs. We test the efficacy of these two approaches for uncovering pathway-specific contributions of the PVT-NAc and BLA-NAc pathways to behavior. We first demonstrate in brain slice recordings that ArchT photoinhibition of glutamatergic fibers effectively reduced evoked excitatory synaptic currents. We also report that it increased asynchronous transmitter release and consequently interneuron spiking, which broadly suppressed glutamate release *via* presynaptic GABA~B~ receptors. *In vivo*, excitatory axon terminal photoinhibition increased feeding and effortful reward-seeking irrespective of the pathway targeted. These effects are comparable to those obtained with broad inhibition (O'Connor et al., [@B37]) or lesions (Bowman and Brown, [@B9]) of NAc projection neurons. In contrast, cell body inhibition of NAc afferents from the PVT and BLA revealed pathway-specific contributions to distinct aspects of reward-seeking when food was available and during extinction, respectively. These data underscore the off-target behavioral consequences of ArchT-mediated terminal inhibition while highlighting cell body inhibition as a valuable alternative for pathway-specific optogenetic silencing.
Materials and Methods {#s2}
=====================
Experimental Model and Subject Details {#s2-1}
--------------------------------------
Adult wild-type and transgenic C57BL/6J mice were used, including tdTomato Cre-reporter mice (JAX\#007914) and parvalbumin-Cre mice (JAX\#008069, Jackson Laboratory, Sacramento, CA, USA). All animals were bred in-house and kept on a reverse light cycle with a 12 h photoperiod. Animals underwent surgery at approximately 3 months of age (25--30 g). Six weeks later they were placed on a restricted feeding schedule and maintained at 85--90% of their pre-surgery body weight. The number of male and female mice were balanced within groups. All experiments were conducted following the Canadian Council of Animal Care and the McGill Animal Care Committee.
Method Details {#s2-2}
--------------
### Viral Constructs and Surgery {#s2-2-1}
Before surgery, animals were anesthetized using a ketamine (Ventoquinol, 100 mg/kg) and xylazine (Bayer, 10 mg/kg) cocktail. The skull of the animal was then secured to a stereotaxic frame (Kopf Instruments) and prepared for intracranial virus injections according to the standard stereotaxic procedure. Seven-hundred nanoliter of virus (5.0 × 10^12^ GC/ml) was injected bilaterally over 10 min using a Nanoject II Injector with an oil-filled glass micropipette pulled to a tip diameter of 10 μm (Drummond Scientific, 3-000-203-G/X).
For axonal photoinhibition experiments, rAAV5-CaMKIIα-eArchT3.0-eYFP (UNC Vector Core) was delivered into the BLA (AP −1.8 mm, ML ±2.8 mm, DV −5.15 mm) and PVT (AP −1.1 mm, ML ±0.35 mm, DV −3.3 mm) of different cohorts of wildtype mice. Optical fibers with a 200 μm core were implanted in the NAc 10 min later (10° angle, AP 1.5 mm, ML ±1.35 mm, DV −4.6 mm). Animals used for brain slice electrophysiology experiments included PV reporter mice and were prepared in the same manner, but optical fibers were not implanted.
For afferent-specific cell body photoinhibition experiments, retroAAV2-CAG-ArchT-GFP (Neurophotonics Centre at Université Laval) was delivered to the NAc (AP 1.5 mm, ML ±0.62 mm, DV −4.7 to −4.2) and an optical fiber was placed above the BLA (AP −2.06 mm, ML ±3 mm, DV −4.02 mm) or PVT (10° angle, AP −1.20 or −0.95 mm, ML ±0.56 mm, DV −2.82 mm) of different cohorts of wildtype mice.
### Behavioral Testing {#s2-2-2}
Mice were trained in sound-attenuating chambers (Med Associates), in which levers were extended on either side of a centrally located food receptacle. A house light and speaker were located on the opposite side of the chamber. All behavioral data were collected using the Med Associates software.
Food-restricted mice were tethered to optical fiber and placed in these operant chambers daily for 40-min sessions. One lever was randomly designated the active lever. Initially, each press on this lever was reinforced with the delivery of 30 μl of a 15% sucrose solution (m/v) to the food receptacle and a tone presentation (4.8 kHz, 80 dB, 5 s duration). After mice earned 40 rewards in a single session, we switched them to a variable ratio (VR3) reinforcement schedule. The number of active lever presses required for reward delivery and tone presentation then varied randomly between 1 and 5. Inactive lever presses were always inconsequential. Both levers remained extended throughout each session.
Photoinhibition experiments were carried out after animals consistently attained 20 rewards per daily session. Photoinhibition involved bilateral intracranial light delivery (532 nm, 10 mW) for two 8-min periods within the 40-min session. The timing of the photoinhibition periods was counterbalanced across 2 days of testing. Animals subsequently experienced two sessions under extinction conditions, in which presses on the active lever were no longer reinforced. Both extinction sessions were preceded by three daily sessions on the VR3 reinforcement schedule.
### Brain Slice Electrophysiology {#s2-2-3}
Mice were anesthetized and perfused with a modified artificial cerebrospinal fluid that contained, in mM, 92 NMDG, 20 HEPES, 25 glucose, 30 NaHCO~3~, 1.25 NaH~2~PO~4~, 2.5 KCL, 5 sodium ascorbate, 3 sodium pyruvate, 2 thiourea, 10 MgSO~4~, 0.5 CaCl~2~ (pH 7.35). Two-hundred micrometer thick coronal slices containing the NAc were prepared using a VT-1200 vibratome (Leica) and held at 34°C for 10 min in this same solution. Slices were then transferred to a "holding ACSF" at room temperature, which was identical except that NaCl (92 mM) was included instead of NMDG and the MgSO~4~ and CaCl~2~ concentrations were 1 and 2 mM, respectively. The ACSF used on the patch rig was maintained at 31°C and contained, in mM, 119 NaCl, 2.5 KCL, 1.25 NaH~2~PO~4~, 2 MgSO~4~, 2 CaCl~2~, 24 NaHCO~3~, and 12.5 glucose. All ACSF preparations were saturated with 95% O~2~ and 5% CO~2~. Cells were visualized on an upright microscope with infrared differential interference contrast and fluorescence microscopy. Whole-cell patch-clamp recordings were made using a MultiClamp 700B amplifier using 2 kHz lowpass Bessel filter and 10 kHz digitization with pClamp 11 software (Molecular Devices). Recordings were made using glass pipets with resistance 4.0--6.0 MΩ, filled with an internal solution containing, in mM, 117 cesium methanesulfonate, 20 HEPES, 0.4 EGTA, 2.8 NaCl, 5 TEA-Cl, 4 Mg-ATP and 0.4 Na-GTP (pH 7.3). Projection neurons identified by morphology, membrane resistance, and hyperpolarized resting membrane potential were patched in the NAc shell in areas with bright eYFP fluorescence. Patched cells were held at −70 mV to record evoked and spontaneous EPSCs or at 0 mV to record spontaneous IPSCs. sEPSCs were recorded over 10 min (5 min per condition). sIPSCs were recorded over 30 min. EPSCs were evoked in pairs (50 ms interval) once per minute for 30 min with a single stimulating electrode positioned 100--200 mm dorsal to the recorded neuron. Recordings that included photoinhibition and CGP55984 lasted 15 min (5 min per condition). Series resistance (10--25 MΩ) remained stable throughout data collection. To record interneuron spiking, cell-attached recordings were carried out on tdTomato^+^ neurons of PV-reporter mice in the NAc shell in areas with bright eYFP fluorescence. Recordings of spiking activity took place over 10 min (5 min per condition). Opsin activation was achieved with 590 nm light (2 mw, ThorLabs, DC4104) directed through the microscope objective.
### Histology {#s2-2-4}
At the end of each behavioral experiment, animals were anesthetized with 270 mg/kg Euthansol (Merck) and transcardially perfused with 4% paraformaldehyde (PFA, Sigma-Aldrich). Brains were removed, post-fixed in PFA for 24 h, and then transferred to PBS for 48 h. Tissue was then sliced into 60 μm sections on a vibratome (Leica VT1000s) and mounted on microscope slides with a MOWIOL plus DAPI (Sigma-Adrich, St. Louis, MO, USA) solution.
Quantification and Statistical Analysis {#s2-3}
---------------------------------------
For electrophysiological recordings, all data were normalized to the last 3 min of the baseline condition of a given recording. The frequency of sEPSCs, sIPSCs, and spiking was calculated by counting the number of events that occurred in 1-min bins across each recording. Once normalized, these data were used for all time-course graphs. For summary graphs, the mean normalized frequency was calculated for each experimental condition.
Two-tailed paired *t*-tests and two-way ANOVAs were used for statistical comparisons of behavior across photoinhibition conditions and across pathways.
Sidak's multiple comparisons tests were conducted for all ANOVA *post hoc* tests. The significance of all statistical tests was determined using *α* = 0.05. All data are reported as the mean ± SEM.
Results {#s3}
=======
ArchT-Mediated Axonal Inhibition of NAc Afferents Increases Reward-Seeking Behavior {#s3-1}
-----------------------------------------------------------------------------------
The PVT-NAc pathway is thought to integrate hunger signals (Kelley et al., [@B23]; Kirouac, [@B25]; Labouèbe et al., [@B27]; Meffre et al., [@B34]) while the BLA-NAc pathway processes reward-predictive stimuli (Ambroggi et al., [@B2]; Esber and Holland, [@B18]; Beyeler et al., [@B8]), so we hypothesized that inhibition of these pathways would differentially influence the vigor of food-seeking and responsivity to reward-associated cues, respectively. We used an axon terminal photoinhibition strategy to test this idea, bilaterally targeting light to the NAc of mice expressing ArchT in their PVT or BLA ([Figure 1A](#F1){ref-type="fig"}). We trained the mice to lever press for food reward on a variable ratio schedule of reinforcement (VR3) and compared their lever press and food port responses within behavioral sessions, across periods with and without intracranial light delivery. We also assessed the behavioral impact of photoinhibition during interspersed extinction sessions when lever presses were not reinforced.
{#F1}
Photoinhibition of PVT and BLA axons in the NAc increased the frequency of active lever pressing ([Figures 1B,C](#F1){ref-type="fig"}), whereas intracranial light delivery in GFP-only control mice did not affect reward-seeking behavior. Photoinhibition of PVT and BLA axons also increased inactive lever responding during extinction sessions ([Figure 1D](#F1){ref-type="fig"}), consistent with general disinhibition of behavior. These behavioral effects are comparable to those obtained with direct NAc neuron photoinhibition (O'Connor et al., [@B37]) and NAc lesions (Bowman and Brown, [@B9]), which suggests that any inhibitory influence on NAc physiology may similarly disinhibit reward-seeking. Alternatively, given the known off-target effects of ArchT-mediated axonal inhibition, this approach to inhibiting specific pathways may generate broad, unintended disturbances in NAc physiology.
Axon Terminal Inhibition of Excitatory NAc Afferents Increases Local Inhibitory Signaling {#s3-2}
-----------------------------------------------------------------------------------------
We examined the off-target consequences of the ArchT photoinhibition of PVT and BLA axons in NAc neuron brain slice recordings ([Figure 2A](#F2){ref-type="fig"}). We first confirmed that opsin activation in PVT and BLA axons reduced the amplitude of electrically-evoked excitatory postsynaptic currents (eEPSCs) in NAc neurons ([Figures 2B,C](#F2){ref-type="fig"}), consistent with the intended purpose of the manipulation. We next monitored the frequency of spontaneous EPSCs (sEPSCs) and found it to be elevated during periods of ArchT activation ([Figures 2D--F](#F2){ref-type="fig"}), consistent with previous results (Mahn et al., [@B30]). There was no change in the amplitude or decay time of spontaneous synaptic currents in response to photoinhibition ([Supplementary Figures S1A--D](#SM1){ref-type="supplementary-material"}), nor was there any effect of light on eEPSCs or sEPSCs in recordings from wildtype animals ([Supplementary Figures S2A,B](#SM2){ref-type="supplementary-material"}).
![ArchT-mediated inhibition of excitatory axons in the NAc increases spontaneous excitatory postsynaptic currents (sEPSC) frequency. **(A)** Schematic of brain slice recording conditions where electrically-evoked-EPSCs were recorded from NAc spiny neurons before, during, and after Arch-mediated photoinhibition of excitatory axon terminals. **(B)** Example recordings from NAc neurons showing changes in the amplitude of electrically-evoked EPSCs during photoinhibition of PVT (*top left*) and BLA (*bottom left*) axons in the NAc. Summary of relative change in EPSC amplitudes (*right*) over time in response to photoinhibition during minutes 10--20. **(C)** Summary of effect of photoinhibition on EPSC amplitudes. Inset shows data collapsed across pathways, highlighting main effect of photoinhibition \[*n*~PVT~ = 7(3 animals); *n*~BLA~ = 7(3); *F*~(2,24)~ = 10.40, *p* \< 0.001; t~baseline\ vs.\ inhibition\ (24)~ = 4.18\*; t~baseline\ vs.\ inhibition\ (24)~ = 3.68\*\]. **(D)** Schematic of brain slice recording conditions where spontaneous EPSCs were recorded from NAc spiny neurons during the Arch-mediated photoinhibition of excitatory axon terminals. **(E)** Example NAc neuron recordings highlighting the increase in spontaneous EPSCs that accompanied Arch-mediated inhibition of PVT and BLA axons. **(F)** Summary of effect of photoinhibition on normalized sEPSC frequency \[*n* = 9(3); *t*~(8)~ = 3.48, *p* \< 0.0\]. Error bars represent SEM. \*Signifies *p* \< 0.01. MSN, the medium spiny neuron.](fncir-14-00010-g0002){#F2}
Possibly the asynchronous glutamate release may not directly alter the firing rate of NAc projection neurons, since they have a resting membrane potential close to −80 mV and are thought to fire action potentials upon concerted excitatory input (Goto and Grace, [@B20]). However, certain interneuron populations in the NAc are highly excitable and may be responsive to small changes in excitatory signaling. To evaluate whether interneuron activity is affected by ArchT-mediated increases in asynchronous glutamate release, we monitored spontaneous inhibitory postsynaptic currents (sIPSCs) in NAc neurons. The frequency but not the amplitude of sIPSCs was elevated during PVT and BLA axon photoinhibition ([Figures 3A--C](#F3){ref-type="fig"}, [Supplementary Figure S1B](#SM1){ref-type="supplementary-material"}), consistent with the hypothesis that local inhibitory signaling is upregulated by unintended glutamate release from ArchT-expressing fibers.
![ArchT-mediated inhibition of excitatory axons in the NAc increases spiny neuron spontaneous inhibitory postsynaptic currents (sIPSC) frequency and PV+ interneuron spiking. **(A)** Schematic of brain slice recording conditions where spontaneous IPSCs were recorded from NAc spiny neurons before, during, and after Arch-mediated photoinhibition of excitatory axon terminals. **(B)** Example NAc neuron recordings showing changes in sIPSC frequency during photoinhibition of PVT (*top* *left*) and BLA (*bottom left*) axons in the NAc. Summary of relative change in sIPSC frequency over time in response to photoinhibition (*right*). **(C)** Summary of effect of photoinhibition on sIPSC frequency. Inset shows data collapsed across pathways, highlighting main effect of photoinhibition \[*n*~PVT~ = 7(3 animals); *n*~BLA~ = 7(3); *F*~(2,24)~ = 7.26, *p* \< 0.01; t~baseline\ vs.\ inhibition\ (24)~ = 3.79\*\]. **(D)** Schematic of brain slice recording conditions where spiking activity was recorded in tdTomato-labeled PV+ fast-spiking interneurons (FSIs) during Arch-mediated photoinhibition of excitatory afferent inputs. **(E)** Example recording from a PV+ interneuron in the NAc showing elevated spiking activity coincident with Arch-mediated axon terminal photoinhibition (*left*). Summary of relative change in FSI spiking frequency over time in response to photoinhibition (*right*). **(F)** Summary of effect of photoinhibition on normalized interneuron spiking \[*n* = 7(2); *t*~(6)~ = 3.64, *p* \< 0.05\]. **(G)** Schematic of brain slice recording conditions, where electrically-evoked EPSCs were recorded from NAc spiny neurons during Arch-mediated photoinhibition of excitatory axon terminals in the presence of a GABA~B~ antagonist (CGP55984). **(H)** Example NAc neuron recordings showing the effects of a GABA~B~ antagonist on evoked EPSC amplitudes during photoinhibition of PVT (*top*) and BLA axons (*bottom*). **(I,J)** Summary of effect of GABA~B~ antagonist on evoked EPSC amplitude \[*n* = 3(2); *t*~(2)~ = 5.51, *p* \< 0.05\] and normalized pulse-paired ratio (PPR; *t*~(2)~ = 2.90, *p* = 0.10) during photoinhibition of excitatory axon terminals. Baseline data not shown. Error bars represent SEM. \*Signifies *p* \< 0.05. FSI, fast-spiking interneuron; MSN, the medium spiny neuron.](fncir-14-00010-g0003){#F3}
To more directly evaluate this hypothesis, we recorded the spiking activity of parvalbumin-positive (PV) interneurons in the NAc of PV-tdTomato mice in response to ArchT photoinhibition of PVT and BLA axons ([Figure 3D](#F3){ref-type="fig"}). We found a sharp increase in PV interneuron spiking following opsin activation, irrespective of its localization to PVT or BLA axons ([Figures 3E,F](#F3){ref-type="fig"}). While it is unclear if this increase in spiking is a direct consequence of asynchronous glutamate release, it suggests that activation of ArchT in any collection of excitatory axons in the NAc may cause common disruptions in NAc physiology.
While excess PV interneuron activity directly inhibits NAc projection neurons, it may also broadly suppress glutamate release *via* presynaptic GABA~B~ receptors located on excitatory afferents (Kupferschmidt and Lovinger, [@B26]). Indeed, bath application of the GABA~B~ receptor antagonist CGP55984 at 2 μM partially reversed the effects of axonal ArchT activation on eEPSC amplitude and pulse paired ratios ([Figures 3G--J](#F3){ref-type="fig"}) without affecting baseline measures of eEPSC amplitude or sEPSC frequency ([Supplementary Figure S3](#SM3){ref-type="supplementary-material"}). This result raises further doubt about the pathway-specificity of the axonal inhibition approach.
Cell Body Inhibition of NAc Afferents Reveals Pathway-Specific Contributions of PVT and BLA Inputs to Reward-Seeking {#s3-3}
--------------------------------------------------------------------------------------------------------------------
To re-evaluate the validity of the behavioral findings we obtained with axonal ArchT activation, we repeated the experiments described above using an alternative approach to disrupting pathway function. Capitalizing on the efficient axon terminal-infecting virus retroAAV, we drove ArchT expression in all NAc-projecting neurons and targeted light to the PVT or BLA ([Figure 4A](#F4){ref-type="fig"}). This method provided pathway specificity, but the manipulation occurred upstream of the NAc. These animals were trained and tested in the same manner as before, and we evaluated the differential influence of PVT and BLA pathways on operant reward-seeking behavior.
{#F4}
With light directed to the PVT and BLA in different cohorts of mice, we found that photoinhibition of the PVT-NAc pathway but not the BLA-NAc pathway increased lever press and food port responding when food was available ([Figures 4B,C](#F4){ref-type="fig"}). This result was consistent with previous findings that implicate PVT input in the integration of hunger signals that regulate the vigor of food-seeking. In contrast, photoinhibition of the BLA-NAc pathway but not the PVT-NAc pathway increased inactive lever responding during extinction when food was not available ([Figure 4D](#F4){ref-type="fig"}). This finding is consistent with the role of BLA input in regulating cue-reward associations. Together, these results suggest that soma-targeted photoinhibition is a valuable alternative to axon-targeted photoinhibition for assessing pathway-specific contributions to behavior, as only the former revealed dissociable influences of PVT and BLA afferents to the NAc on reward-seeking.
Discussion {#s4}
==========
In trying to identify dissociable influences of PVT and BLA input to the NAc on reward-seeking behavior, we compared axon-targeted and soma-targeted photoinhibition strategies using the outward proton pump ArchT. We found that axon terminal inhibition increased reward-seeking behavior similarly when targeted to either pathway, whereas pathway-specific inhibition of upstream cell bodies produced dissociable behavioral effects that were consistent with previous literature. NAc neuron brain slice recordings confirmed that activation of ArchT in excitatory axon terminals reduces eEPSCs yet increases sEPSCs. The pathway-specific nature of this manipulation was undermined by a concomitant increase in GABAergic interneuron activity, which was associated with a broad GABA~B~ receptor-dependent reduction in eEPSC amplitudes. These results suggest that the ArchT photoinhibition of excitatory axons has off-target consequences that generally disrupt NAc physiology, which may explain why similar changes in behavior result from the inhibition of distinct axonal inputs. Soma-targeted photoinhibition appears to be a valuable alternative for pathway-specific optogenetic silencing.
The NAc integrates excitatory input from several limbic structures (Mannella et al., [@B33]) and is a convergent site of dysregulation in many psychiatric disorders (Ahmari et al., [@B1]; Bagot et al., [@B4]; Francis et al., [@B19]; Creed et al., [@B14]; Neumann et al., [@B36]). Identifying the distinct behavioral contributions of these excitatory inputs is challenging because they similarly engage NAc physiology (Britt et al., [@B10]) and originate in regions that are themselves highly interconnected (Pitkänen et al., [@B39]; Li and Kirouac, [@B28]; Do-Monte et al., [@B16]). The study of these circuit elements thus heavily relies on our ability to selectively disrupt pathway-specific function. Unfortunately, many of the optogenetic tools commonly used to silence long-range neural projections have unclear limitations *in vivo*, and their off-target effects may mask pathway-specific differences, particularly in highly integrative structures such as the NAc. Accordingly, there is considerable value in comparing different pathway-specific inhibition strategies for manipulating multiple, parallel long-range projections.
Alternatives to Targeting ArchT Photoinhibition to Axon Terminals {#s4-1}
-----------------------------------------------------------------
Optogenetic approaches for silencing neural activity have repeatedly been found to produce off-target effects and unpredictable changes in network activity that preclude straightforward interpretations of experimental outcomes. For example, tissue heating in response to prolonged light delivery is sufficient to alter potassium conductance in striatal neurons (Owen et al., [@B38]). Additionally, ion transporters such as ArchT and halorhodopsin can dramatically alter the intracellular ionic environment affecting the pH level and chloride reversal potential, respectively (Raimondo et al., [@B40]; Mahn et al., [@B30]). These disruptions are most pronounced in axons on account of their relatively small intracellular volume. A separate issue is the rebound excitation that often follows any acute hyperpolarization (Arrenberg et al., [@B3]).
Fortunately, new tools and experimental approaches have mitigated some of these unintended effects. The development of highly effective retrograde viral vectors (Tervo et al., [@B48]) has facilitated projection-specific cell body photoinhibition strategies, as demonstrated here. This approach benefits from the capacity of the soma to buffer against significant changes in pH and ionic composition (Wiegert et al., [@B50]). Opsin expression can be further restricted with intersectional strategies involving recombinase proteins such as Cre or FLP and retrograde viral vectors. This approach may be necessary when targeting specific pathways within reciprocally connected brain regions.
Anion-conducting channelrhodopsins (ACRs) may be the best option currently available for photoinhibition experiments. Since their conductance is dependent on the membrane potential, their activation can shunt voltage fluctuations of the cell without inducing strong hyperpolarization or significantly altering the ionic environment (Berndt et al., [@B7]). The light-driven chloride channels that have been engineered (eACRs; Berndt et al., [@B6]) or found in nature (GtACRs; Govorunova et al., [@B21]) are also more light-sensitive than any presently used ion transporter opsins. Unfortunately, chloride channels appear to be excitatory in many axons, due to locally elevated chloride concentrations (Khirug et al., [@B24]; Malyshev et al., [@B31]). Soma-restricted GtACR variants have been developed to circumvent this issue, but their expression level has to be optimized to minimize their excitatory influence in the axon hillock (Mahn et al., [@B29]).
These new tools have fewer drawbacks than their predecessors, but it remains difficult to interpret the effects of inhibiting specific projections that are embedded in recurrent circuitry (Spellman et al., [@B44]; Do-Monte et al., [@B15]). It is unclear how silencing a pathway will ultimately affect downstream neurons, let alone the network as a whole. Limiting the duration of photoinhibition may mitigate some concerns, but the organization of the affected circuitry and behavioral state of the animal should be carefully considered.
PVT and BLA Inputs to the NAc Influence Discrete Aspects of Reward-Seeking {#s4-2}
--------------------------------------------------------------------------
PVT and BLA projections to the NAc have been found to promote and discourage reward-seeking in different contexts (Stuber et al., [@B47]; Millan et al., [@B35]; Zhu et al., [@B52]; Do-Monte et al., [@B15]; Bercovici et al., [@B5]; Reed et al., [@B41]; Shen et al., [@B42]). For instance, photostimulation of PVT input to the NAc can reduce or increase sucrose-seeking behavior (Labouèbe et al., [@B27]; Do-Monte et al., [@B15]; Cheng et al., [@B11]), while photostimulation of different fibers in the BLA-NAc pathway can generate place preference or aversion (Shen et al., [@B42]). These mixed effects likely reflect the heterogeneity of the PVT and BLA. By studying coarse disruptions of pathway activity alongside manipulations of genetically and spatially defined subpopulations in varied contexts (Labouèbe et al., [@B27]; Shen et al., [@B42]), we can begin to understand the net contributions of these pathways to behavior.
Here, we identified a unique role of PVT inputs in modulating the vigor of food-seeking behavior, which is consistent with a wealth of literature implicating this structure in integrating hunger signals and regulating feeding (Kelley et al., [@B23]; Stratford and Wirtshafter, [@B46]; Kirouac, [@B25]; Labouèbe et al., [@B27]; Choi and McNally, [@B12]; Do-Monte et al., [@B15]; Cheng et al., [@B11]; Meffre et al., [@B34]). While stimulation of a smaller glucose-sensing subset of PVT-NAc projectors can promote consumptive behavior (Labouèbe et al., [@B27]), we find that the net effect of bulk PVT-NAc inhibition is an increase in food-seeking, consistent with the net reductions in PVT-NAc pathway activity that have been observed in the rostral NAc during feeding (Reed et al., [@B41]). Gross excitatory drive from this input may, therefore, gate food-seeking behaviors by downregulating NAc activity overall, while smaller, genetically defined populations---like those that sense low levels of interstitial glucose (Labouèbe et al., [@B27])---may target relatively circumscribed areas of the NAc which have opposing influences on feeding behavior.
Our soma-targeted photoinhibition of the BLA-NAc pathway did not affect lever-pressing behavior when food was available but increased it under extinction conditions, consistent with the role of this pathway in regulating operant responding following changes in outcome (Shiflett and Balleine, [@B43]). BLA-NAc activity may thus have a pronounced role in behavioral suppression under varied conditions. Overall, the use of increasingly refined tools to inhibit projection-specific activity will aid efforts to dissect neural circuit function concerning behavior.
Data Availability Statement {#s5}
===========================
The datasets generated for this study are available on request to the corresponding author.
Ethics Statement {#s6}
================
The animal study was reviewed and approved by Canadian Council of Animal Care and the McGill Animal Care Committee.
Author Contributions {#s7}
====================
CL and JB conceived the study, designed the experiments, and wrote the manuscript. CL conducted the experiments. JB supervised the research.
Conflict of Interest {#s8}
====================
The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
**Funding.** This work was supported by the Natural Sciences and Engineering Research Council (05069-2014 to JB and CGSD3-534197-2019 to CL), the Canadian Institutes of Health Research (PT-74038 to JB), and McGill's Healthy Brains for Healthy Lives CFREF (to CL).
Supplementary Material {#s9}
======================
The Supplementary Material for this article can be found online at: <https://www.frontiersin.org/articles/10.3389/fncir.2020.00010/full#supplementary-material>.
######
ArchT-mediated photoinhibition of excitatory axons in the NAc does not affect the amplitude or the decay rate of spontaneous synaptic currents. Related to [Figures 2](#F2){ref-type="fig"}, [3](#F3){ref-type="fig"}. Summary of effect of ArchT axon terminal inhibition on **(A)** the amplitude of spontaneous EPSCs \[*n* = 9(3 animals); *t*~(8)~ = 0.13, *p* = 0.90\], **(B)** the decay rate of spontaneous EPSCs (*n* = 8(3); *t*~(7)~ = 1.61, *p* = 0.15\], **(C)** the amplitude of spontaneous IPSCs \[*n* = 14(6); *t*~(13)~ = 1.22, *p* = 0.25\], and **(D)** the decay rate of spontaneous IPSCs \[*n* = 14(6); *t*~(13)~ = 0.46, *p* = 0.65\].
######
Click here for additional data file.
######
Light alone does not affect the amplitude of eEPSCs or the frequency of sEPSCs in NAc cell recordings from wildtype animals. Related to [Figure 2](#F2){ref-type="fig"}. Summary of the effect of light on **(A)** the amplitude of evoked EPSCs \[*n* = 8(3 animals); *t*~(7)~ = 2.11, *p* = 0.07\] and **(B)** the frequency of spontaneous EPSCs (*t*~(7)~ = 0.63, *p* = 0.55).
######
Click here for additional data file.
######
In the absence of photoinhibition, a GABAB antagonist does not affect the amplitude of eEPSCs or the frequency of sEPSCs recorded from NAc neurons of wildtype animals. Related to [Figure 3](#F3){ref-type="fig"}. Summary of the effect of light on **(A)** the amplitude of evoked EPSCs \[*n* = 5(3 animals); *t*~(4)~ = 0.80, *p* = 0.47\] and **(B)** the frequency of spontaneous EPSCs (*t*~(4)~ = 0.08, *p* = 0.94).
######
Click here for additional data file.
[^1]: Edited by: Yves De Koninck, Laval University, Canada
[^2]: Reviewed by: Naofumi Uesaka, The University of Tokyo, Japan; Catherine J. C. Weisz, National Institutes of Health (NIH), United States
| {
"pile_set_name": "PubMed Central"
} |
**Correction to: BMC Biol**
**https://doi.org/10.1186/s12915-018-0606-4**
Upon publication of the original article, \[[@CR1]\], the authors noticed that the first authors' affiliation contained an error. In the first affiliation it reads "Queensland Brian Institute", whereas it should in fact be "Queensland Brain Institute".
| {
"pile_set_name": "PubMed Central"
} |
Several recent reports point to enterovirus (EV) infections as key environmental triggers of type 1 diabetes (T1D)[@b1][@b2][@b3][@b4][@b5][@b6][@b7]. Conclusions are based on multiple proofs that include the histopathologic detection of EV antigens/genome in the islets of Langerhans of diabetics at different stages of the disease[@b8][@b9][@b10][@b11][@b12]. Recent findings suggest that EVs are causing chronic low-level infection in the islet cells of newly diagnosed T1D patients[@b12].
EVs are small, non-enveloped, single-strand positive-sense RNA viruses belonging to the family *Picornaviridae*. The EV capsid is composed of four structural proteins named VP1, VP2, VP3, and VP4. The main structural differences among VP1, VP2, and VP3 lie in the loops that connect the beta-strands with the N- and C-terminal sequences extending from the beta-barrel domain[@b13]. These amino acid (AA) sequences give each EV its distinct morphology and antigenicity. The VP4 component lies on the inner surface of the shell and is essential for virion stability. Evolution has resulted in a large number of antigenically distinguishable members that have been categorized as EV "serotypes"[@b13]. Not considering human rhinoviruses, the EV genus contains human agents of the A, B, C, D, and unnamed species that, together, comprise 109 different types.
Each of the serotypes correlates with a specific immune response of the host and protection from disease. The serotype-specific protective immune response is directed to the capsid proteins VP1, VP2, and VP3, as the VP4 has no role in the interaction with neutralizing antibodies (i.e., those directed to variable regions of surface capsid proteins). Non-neutralizing antibodies recognizing VP regions that are conserved among different EV types are also produced. The significance of the non-neutralizing antibody response is currently under investigation[@b14].
Immunization with different EV types allowed to produce a variety of monoclonal antibodies (MAbs) that are either type-specific (i.e., responsible of virus neutralization) or directed to conserved regions of capsid proteins[@b15][@b16][@b17][@b18]. Reactivity of the latter antibodies may be limited to sets of different EV types or may be directed to a wider range of EV types. Though molecular methods are held to be more informative than classical serologic methods for virus identification[@b19], "pan-EV antibodies" capable of reacting with all or with the majority of EV types remain desirable reagents for detecting these agents both in the diagnosis of infectious diseases[@b20][@b21][@b22][@b23] and in immunohistochemistry[@b24].
The greater part of immunohistochemical studies in which a conserved region of enteroviral VP1 has been detected in the islets of Langerhans of T1D cases employed the pan-EV MAb 5D-8.1[@b10][@b25]. This MAb has been produced in mice immunized with inactivated coxsackievirus B5 (CV-B5) and has been shown, by immunostaining, to react with multiple EV types[@b18][@b20][@b23]. The 5D-8.1 epitope in VP1 has been partially characterized using competition assays with synthetic peptides and demonstrated to be EV-specific[@b26]. However, subsequent studies suggested that MAb 5D-8.1 may also recognize human proteins, including an isoform of creatine kinase and a mitochondrial ATP synthase[@b27]. Comprehensibly, these results casted doubts on the conclusions of previous immunohistochemical studies of pancreatic tissue in T1D cases[@b25][@b28].
Due to the relevance of MAb 5D-8.1 in diabetes research, we re-investigated this antibody in parallel with the pan-EV MAb 9D5 that is used for diagnosing EV infections in virology laboratories[@b20][@b21]. MAb 9D5 has been obtained from mice immunized with inactivated CV-B3 and shown to react with multiple EV types[@b20].
Reactivity of the two MAbs was defined with the help of innovative microarray technology, substitution scan of peptide epitopes, immunostaining of acutely and persistently infected cell lines, neutralization assays. The MAbs' epitopes were then examined versus the human proteome and versus proteins of diverse viral agents in order to delineate their specificity and define possible cross-reactivities.
Results
=======
Epitopes of Mab 5D-8.1 and MAb 9D5
----------------------------------
Secondary goat anti-mouse IgG Ab did not show background interactions with antigen-derived peptides. As shown in [Fig. 1a,c](#f1){ref-type="fig"}, MAb 5D-8.1 gave defined spots for the VP1 sequence of CV-B1, CV-B4, E-30, and reduced reactivity with CV-A1. The peptide scan indicated IPALTAVETGHT as the consensus sequence containing the epitope of MAb 5D-8.1. Comparable signals were produced by MAb 9D5 ([Fig. 1b,d](#f1){ref-type="fig"}) and attributed to the consensus motif SIGNAYSMFYDG. Thus, relative to the VP1 sequence of CV-B4 (GI: 61031)[@b29], the sequence containing the epitope of 5D-8.1 was close to the N-terminus at AA residues 28--39, whereas that of 5D9 was located towards the C-terminus at residues 187--198. Substitution scan of peptide IPALTAVETGHT against MAb 5D-8.1 allowed to delimit the antibody binding site to a conserved core motif 4-IPALTAAET-12. AA positions 4I, 7L and 8T showed the highest degree of sequence conservation with a nearly complete loss in binding of MAb 5D-8.1 upon exchange by other amino acids. AA positions 5P and 11E were well-conserved, exchange by Q and by H, respectively, caused a 60--80% loss in antibody binding. Compared to this, AA positions 6A and 10A exhibited a slightly higher susceptibility for substitution by selected amino acids with a maximal decrease of 50% of spot intensities. 9A and 12T showed the highest tolerance for exchange by other amino acids. Replacement by F (9A) and A (12T) was accepted without loss in antibody binding.
Three dimensional analysis of the VP1 protein and the two MAb epitopes
----------------------------------------------------------------------
For each viral capsid organization level, the Solvent-Accessible Surface Area (SASA) was calculated in order to estimate the degree-of-burial of antibody epitopes within the protein. The resolved structures of six reference enterovirus strains were obtained from the RCSB database and a 1.4 Å sphere probe was used to represent a water molecule. The exposed surface area was first calculated, and then normalized with the maximum allowed solvent-accessible area[@b30]. Normalized SASA values take the form of Relative Solvent Accessibility (RSA), a quantity which varies between 0 (for completely buried residues) and 100 (for maximally exposed residues). Results are summarized in [Fig. 2](#f2){ref-type="fig"}. The alignment of the VP1 regions recognized by the two MAbs is shown in [Fig. 3a](#f3){ref-type="fig"}. The target residues are mainly accessible in the monomeric form for the two epitopes (N-terminal 5D-8.1, yellow; C- terminal 9D5, green). The exposed residues are highly conserved among different EV types, evidencing their importance in the capsid assembly process. Localization of the VP1 protein within the capsomere ([Fig. 3b](#f3){ref-type="fig"}) shows that the 5D-8.1 epitope is located in a domain where exposed residues are stabilized by a beta sheet structure. [Figure 3c](#f3){ref-type="fig"} shows the two epitopes in the VP1 protein assembled into the capsid.
Detection of MAb epitope sequences among human and viral proteins
-----------------------------------------------------------------
The predicted reactivity of antibodies 5D-8.1 and 9D5 with human proteins and viral agents was explored using BLASTp queries vs. the human proteome and viruses. Results are summarized in [Table 1](#t1){ref-type="table"}. It is deduced that the two MAbs may recognize linear targets producing significant alignments with sufficiently low E values. Interestingly, MAb 5D-8.1 produced significant alignments for creatine kinase U-type and a mitochondrial ATP synthase, among other proteins. These targets may represent the antigens indicated as cross-reactive by Korsgren and collaborators in 2013[@b27]. Similarly, MAb 9D5 may be expected to bind to a variety of human proteins, including the leucine-rich repeat-containing protein 66. However, much better alignments (i.e., much lower E values) have been obtained for enteroviruses, rhinoviruses, and agents of the Rabovirus and Sapelovirus genera that represent the most recent members of the *Picornaviridae* family[@b31][@b32].
The expected reactivity of MAbs 5D-8.1 and 9D5 with different EV types of the A, B, C, D species is shown in [Table 2](#t2){ref-type="table"}. Whereas reactivity of the two MAbs is largely equivalent, 5D-8.1 has a wider coverage of the A species as compared to MAb 9D5. Scattered cases of complimentary specificity also occur, indicating that the combined use of both antibodies could widen the detection range in diagnostic/immunohistochemical procedures.
Immunostaining of enterovirus-infected cell cultures
----------------------------------------------------
In uninfected cells (AV3 and LLC-MK~2~ lines) MAb 9D5 did not produce fluorescence even at the concentration of 5 μg/ml ([Fig. 4a](#f4){ref-type="fig"}), while 5D-8.1 yielded fine perinuclear and cytoplasmic fluorescence when used at the concentration of 1 μg/ml ([Fig. 4b](#f4){ref-type="fig"}), but not at concentrations ≤1 μg/ml ([Fig. 4c](#f4){ref-type="fig"}). The two investigated MAbs produced dotted cytoplasmic fluorescence in human and monkey cells acutely infected with CV-B4 ([Fig. 4d--f](#f4){ref-type="fig"}). Fine dotted fluorescence was also seen in AV3 cells undergoing persistent infection by the CV-B1pc strain isolated from a case of pancreatic carcinoma ([Fig. 4g--i](#f4){ref-type="fig"}). In persistent infection, VP1 was expressed frequently in cells showing mitotic bars or dividing ([Fig. 4h,i](#f4){ref-type="fig"}). The slow infectious process was not accompanied by manifest CPE.
IIF was also used for investigating the inhibitory effects of peptides containing the epitopes of MAbs 5D-8.1 and 9D5 in the acute model of CV-B4 infection. Fluorescent staining by 5D-8.1 was totally inhibited by pre-incubation with the peptide SESIPALTAAETGHT (8 μg/ml), but not with peptide SIGNAYSNFYDG. The reverse was true for MAb 9D5: pre-incubation with SIGNAYSNFYDG (8 μg/ml), but not with SESIPALTAAETGHT, inhibited cytoplasmic fluorescence (data not shown). Thus, IIF confirmed that the two AA sequences encompass the relevant epitopes.
Enterovirus neutralization assays with MAbs 5D-8.1 and 9D5
----------------------------------------------------------
The neutralizing activity of 5D-8.1 and 9D5 was explored against CV-B1 and CV-B4. As controls, horse antiserum against CV-B1 and the CV-B4-neutralizing MAb 356 were used. As shown in [Table 3](#t3){ref-type="table"}, the two pan-EV antibodies did not neutralize CV-B1 and CV-B4 (titer \<1:8). As expected, control antibodies had high homotypic, but not heterotypic, neutralizing titers (anti CV-B1: 1:4096 vs. CV-B1; \<1:8 vs. CV-B4 - anti CV-B4: 1:512 vs. CV-B4; \<1:8 vs. CV-B1). Thus, the investigated monoclonals are devoid of neutralizing activity.
Discussion
==========
Validation of antibodies used to identify specific biomolecules is a critical issue in medicine[@b33]. To this end, a variety of methods can be used, but it is recommended that rather than relying on a single antibody, researchers should have the possibility of using pairs of antibodies designed to bind different parts of the same target protein. The case of MAb 5D-8.1 is remarkable in the context of diabetes research. In fact, numerous immunohistochemical studies with 5D-8.1 documented the presence of EV VP1 within the islets of Langerhans in a large proportion of T1D cases, but not in the pancreas of non-diabetic subjects[@b28]. These studies suggested that viral infection played a pathogenic role in T1D. In 2013, it was shown that MAb 5D-8.1 could bind human islet proteins, specifically the mitochondrial proteins creatine kinase B-type and ATP synthase beta subunit[@b27]. The finding triggered a reassessment of EV infection of pancreatic islets in T1D cases[@b34]. Two subsequent publications settled the issue in part, convening that - under appropriate staining conditions - MAb 5D-8.1 was an adequate probe for EV infection[@b25][@b28].
As virologists, we set out to validate the binding of 5D-8.1 and 9D5 to the EV VP1 capsid component and to identify the possible cross reactivity of these antibodies both with human proteins and viral agents. Binding results and bioinformatics analyses confirmed that the epitopes of 5D-8.1 and 9D5 are distinct and located at the N- and C-terminal domains of VP1. Both antibodies are directed to conserved domains of a capsid protein of picornaviruses, and recognize the majority of EV types. However, they are not neutralizing, as expected for antibodies targeting conserved regions of the viral shell. Our data delineate the spectrum of EV types that each antibody binds to, thus confirming partial published results on the specificity of the two MAbs[@b18][@b20][@b24][@b26]. Bioinformatics analysis indicated that the two antibodies cover EV types of the A species less well than those of the B, C, D species ([Table 2](#t2){ref-type="table"}). Since the binding spectra are not identical, the combined use of MAbs 5D-8.1 and 9D5 should allow to cover almost all EV types. Notably, both antibodies are also predicted to cover Rabovirus and Sapeloviruses, animal viruses of the most recent genera within the *Picornaviridae* family. Immunofluorescence results confirmed that MAb 5D-8.1 (but not 9D5) may produce fine granular fluorescence in the cytoplasm of uninfected human and monkey cells. This, however, occurred only at elevated antibody concentrations (i.e., \>1 μg/ml). The observation is in line with the lack of absolute specificity of EVs reported by Korsgren and collaborators[@b27]. We could however confirm that, when adequately diluted, the antibody produces specific staining of different EV types in cultured cells[@b25] without fluorescent signals in uninfected cells. Of interest to diabetes research, the linear epitopes of both MAbs bear only marginal similarity with the human proteome, with a few possible exceptions. In particular, the 5D-8.1 epitope bears similarity with creatine kinase U-type (*E value*, 3.0), ATP synthase mitochondrial F1 complex assembly factor (*E value*, 12), creatine kinase B-type (*E value*, 66).
Notably, 9D5 - that has been used for a long time in diagnostic virology[@b18][@b20] --does not produce background staining in uninfected cultured cells and delineates clearly the expression of VP1 in acutely- and persistently-infected cells. Immunofluorescent staining of cultured cells cannot be compared directly with that of immunocytochemistry. In formalin-fixed, paraffin-embedded samples, antigen retrieval is mandatory for VP1 detection[@b25] (suggesting that protein denaturation may favor antibody binding). In contrast, staining of cultured cells with either of these MAbs gives excellent results upon acetone or paraformaldehyde fixation (i.e., procedures not causing protein unfolding). Whether the two MAbs can bind to VP1 in its native configuration[@b35] within live virus particles is currently under investigation.
Finally, in persistently-infected cells, expression of VP1 was frequently seen in proliferating cells with both 5D-8.1 and 9D5. The finding may not be accidental. In fact, the cellular factor 68-kDa Src-associated protein in mitosis (Sam68) has been recently shown to interact with the EV IRES during infection, thus enhancing translation of virus proteins[@b36]. This aspect merits further attention due to its possible impact on EV pathogenesis[@b36]
In conclusion, both MAbs bind to the VP1 capsid protein of EVs and of phylogenetically-related picornaviruses. Epitopes are located in distinct stretches of the VP1 protein. The MAbs recognize both acute and persistent infection in cultured cells, and are devoid of EV-neutralizing activity. Thus, these distinct and independent probes will be useful for confirming histopathologic and virology data that indicate EV infection of islet cells as having a pathogenic role in diabetes. Further virus searches in diabetes and other endocrine diseases are expected.
Methods
=======
Identification of linear epitopes in VP1 sequences of four enterovirus types
----------------------------------------------------------------------------
The linear epitopes of MAbs 5D-8.1 and 9D5 in the VP1 capsid protein of enteroviruses were mapped using peptide microarray technology[@b37]. The N- and C-termini of VP1 sequences of four enterovirus types \[coxsackievirus B1 (CV-B1; K4N918), coxsackievirus-B4 (CV-B4; S5PU54), echovirus-30 (E-30; Q9YLK0), and coxsackievirus-A1 (CV-A1; Q9YLP4)\] were elongated by neutral GSGSGSG linkers to avoid truncated peptides. Elongated sequences were linked to a single artificial sequence. The elongated artificial sequence was translated into 15 AA peptides with a peptide-peptide overlap of 14 amino acids and bound to duplicate spots of silica microarrays. The resulting microarrays covered peptides of all four sequences (1,181 different peptides in duplicate). For control, microarrays were framed by HA (YPYDVPDYAG) and FLAG (DYKDDDDKGG) peptides (104 spots). Rockland blocking buffer MB-070 (VWR International Frankfurt, DE), PBS plus 0.05% Tween 20, PBS-Tween plus 10% Rockland blocking buffer were used for blocking, washing, and incubation procedures, respectively. Mouse mAb 5D-8.1 (IgG2a; Dako, Cernusco sul Naviglio, Italy) and 9D5 (IgG3; Millipore, Livingstone, UK) were incubated at 10 μg/ml in the microarray for 16 h at 4 °C with shaking. Goat anti-mouse IgG-DyLight680 (New England Biolabs, Frankfurt, DE) served as secondary antibody and was incubated for 30 min at room temperature. HA and FLAG control peptides framing the microarray were subsequently stained with MAb anti-HA-DyLight680 (red) and MAb anti-FLAG-DyLight800 (green). Light emission was read with a LI-COR Odyssey Infrared Imaging System (resolution 21 μm; 700 nm; LI-COR Biosciences, Bad Homburg, DE). Staining with MAbs anti-HA and anti-FLAG confirmed the assay quality and microarray integrity (scanning intensities red/green 7/7).
Substitution scan of the peptide SESIPALTAAETGHT against MAb 5D-81
------------------------------------------------------------------
Peptide array synthesis and binding detection were performed by PEPperPRINT GmbH (PEPperPRINT, Heidelberg, Germany) as reported[@b38][@b39]. Permutation scans were carried out on the peptide SESIPALTAAETGHT (containing the MAb 5D-8.1 epitope) and on its variants. In the permutation scan, the effect on binding of replacing each of the 15 peptide positions by standard L-amino-acids was analyzed. Each microarray contained 286 peptides printed in duplicate and was framed by HA and FLAG control peptides. One peptide microarray copy was pre-stained with the secondary goat anti-mouse IgG (H+L) DyLight680 antibody (red) in the presence of the monoclonal anti-HA (12CA5)-DyLight800 antibody (green) followed by read-out at scanning intensities of 7/7 (red/green). Incubation of a second peptide microarray with MAb 5D-8.1 (1 μg/ml) was followed by staining with the secondary antibody in the presence of the monoclonal anti-HA (12CA5)-DyLight800 control antibody and then read-out at scanning intensities of 7/7 (red/green). Finally, the FLAG control peptides framing the peptide arrays were stained as additional internal quality control to confirm the assay quality and the peptide microarray integrity (scanning intensities red/green: 7/7).
Quantification of spot intensities and peptide annotation were based on 16-bit gray scale tiff files at scanning intensities of 7/7. Microarray image analysis was done with PepSlide Analyzer. A software algorithm broke down fluorescence intensities of each channel and spot into raw, foreground, background signal and calculated the standard deviation of median foreground intensities. Based on averaged median foreground intensities, an intensity map was generated and interactions in the peptide map were highlighted by an intensity color code with red for high and white for low spot intensities. To provide an in-depth view on the substitution scan, a heat map of the peptide microarray was generated as well as a substitution matrix and an AA plot reflecting the AA preferences at a given position. Data sets were correlated with peptide and intensity maps to analyze the substitution pattern in consideration of conserved/variable amino acids and possible AA exchanges.
The substitution matrix highlighted the preference for a given AA by color codes (red: preferred AA; light blue: less preferred AAs) and was calculated by dividing the spot intensity of a given peptide (e.g., 1-YPYDVQDYA-9) by the averaged spot intensities of all 20 peptides that were substituted at the same position (1-YPYDVQDYA-9). The AA plot was calculated by dividing the spot intensity of a given peptide (1-YPYDVQDYA-9) by the spot intensity of the wild type epitope (1-YPYDVQDYA-9). The position of an AA at a given position thus reflected the intensity ratio compared to the AA of the native epitope at the same position.
Cell lines, viruses, indirect immunofluorescence (IIF) and neutralization assays
--------------------------------------------------------------------------------
The human cell lines AV3, HeLa, RD and the monkey line LLC-MK~2~ (ECAAC, Salisbury, UK) were cultured in DMEM/F12 medium supplemented with 10% FCS plus penicillin/streptomycin. EV types CV-B1 (Conn-5), CV-B4 (J.V.B.), E-30 (Bastianni), CV-A1 (Tompkins) obtained from ATCC (LGC Standards, Sesto San Giovanni, Italy) were amplified *in vitro*, titrated and stored at −70 °C. For acute infection, subconfluent cultures were infected at a multiplicity of infection of 0.5 and incubated 3--6 hours. Persistent infection was investigated in the AV3 cell line that had been chronically infected with CV-B1pc (a virus strain isolated from tissue collateral to pancreatic carcinoma). AV3 cells carrying CV-B1pc do not show evident CPE, maintain apparently normal replication rate, and release minimal amounts of virus in the medium as seen by RT-PCR and virus titration (≤100 TCID~50~/ml; manuscript in preparation). Persistently infected cells were at the 10^th^--20^th^ passage.
For IIF, cells were cultured in 4-well Millicell EZ Slides (Merck, Vimodrone, Italy). Cell monolayers were fixed in 4% paraformaldehyde in PBS (r.t., 30 min), washed 3× in PBS-1% FCS, permeabilized with Triton X100 (0.05% in PBS; 10 min), washed 3× in PBS-1% FCS, briefly immersed in distilled water and dried out. Fixed slides were incubated overnight at 4 °C with 0.5, 1, or 5 μg/ml of mouse pan-EV MAb 5D-8.1 (IgG2a) or MAb 9D5 (IgG3). Staining was achieved with the secondary antibodies FITC-labeled goat anti-mouse IgG H+L (Merck), Alexa Fluor 488 goat anti-mouse IgG2a, or Alexa Fluor 594 goat anti-mouse IgG3 (ThermoFisher, Monza, Italy). ProLong antifade (ThermoFisher) was used as mounting medium. Pictures were taken with either a Nikon E800 Eclipse microscope (Nikon, Firenze, Italy) or a Leica TCS SP8 confocal microscope (Leica, Milano, Italy).
Indirect immunofluorescence (IIF) inhibition assays of MAbs 5D-8.1 and 9D5 have been performed using different concentrations of the synthetic peptides SESIPALTAAETGHT (PEPperPRINT) and SIGNAYSCFYDG (Sigma-Aldrich, Milano, Italy) that encompass the epitopes of 5D-8.1 and 9D5, respectively. Each MAb (1 μg/ml) was pre-incubated with increasing concentrations (0, 2, 8, 32 μg/ml) of either the SESIPALTAAETGHT or the SIGNAYSCFYDG peptide (2 hrs at r.t.). Then, IIF staining was performed as above.
Neutralization assays were performed as reported[@b40]. The following MAbs/antisera were used: MAb 5D-8.1, MAb 9D5, anti-CV-B4 MAb 356[@b41], horse anti-CV-B1 serum (ATCC). Briefly, antibody dilutions in complete DMEM/F12 medium were made in triplicate in 96-well flat bottom plates. Then, 100 tissue culture infectious doses~50~ (TCID~50~) of CV-B1 or CV-B4 were dispensed into each well and mixed with antibody dilutions or control medium. After 1 hr incubation at r.t., 10^4^ LLC-MK~2~ cells were added to each well. Cytopathic effect (CPE) was read at 6--7 days with an inverted microscope. Antibody titer is defined as the highest antibody dilution capable of preventing CPE.
Bioinformatics analysis
-----------------------
A list of complete EV capsid structures were obtained from the *advanced search* method of the RSCB server[@b42], and a selection of the resulting data were downloaded as monomer PDB files (1COV, 1H8T, 1D4M, 1EV1, 4GB3, 4Q4V). For each monomer structure, the complete capsid was assembled based on its BIOMT REMARK included in the PDB file. The Visual Molecular Dynamics software (VMD)[@b43] has been used for visualization, computation and analysis of structural data. The Solvent-Accessible Surface area (SASA) was calculated for the VP1 monomer, the capsomer, the capsid. For each epitope, BLASTp queries were performed in public databases (*Homo sapiens*, Viruses, *Picornavirales,* ssRNA viruses, enteroviruses, rhinoviruses). A BLAST program employing the SEG algorithm[@b44] was used to filter low complexity regions from proteins before executing a database search. The BLASTp results that produced significant alignments of the two MAb epitopes with human proteins or viral agents are reported along with the *Expected*(*E*) *value. E value* is used as a measure of epitope specificity. The lower the *E-value*, or the closer it is to zero, the more "significant" the match is. Analysis of the predicted reactivity of the two MAbs with the different EV types is based on the alignments of epitope sequences with the VP1 sequence of EVs. A cut off value of 70% AA identity has been used.
Additional Information
======================
**How to cite this article**: Maccari, G. *et al.* Properties of Two Enterovirus Antibodies that are Utilized in Diabetes Research. *Sci. Rep.* **6**, 24757; doi: 10.1038/srep24757 (2016).
The generous support of the Juvenile Diabetes Research Foundation and the Network of Pancreatic Donors with Diabetes (JDRF-nPOD-V grant 25-2012-770 to AT) is gratefully acknowledged. Gratitude goes to Volker Stadler and Lisa Steinbrück (PEPperPRINT, Heidelberg) for their professionalism and kind support. The authors recognize the excellent technical assistance of Alessandro Bassani, thank Noel Morgan for critically reviewing the manuscript and Alberto Pugliese for scientific support.
**Author Contributions** Conceived and designed the experiments: A.T. and G.M. Performed the experiments: G.M., A.T., A.G. and S.S. Analyzed the data: A.T., G.M., A.G. and S.S. Wrote the paper: A.T. and G.M.
{#f1}
{#f2}
{#f3}
{#f4}
###### Significant alignments of human proteins and viral agents with the epitopes of monoclonal antibodies 5D-8.1 and 9D5[1](#t1-fn1){ref-type="fn"}.
Target proteins MAb 5D-8.1 MAb 9D5
----------------- -------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- ----------------- ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------ -----------------
*Homo sapiens* Creatine kinase U-type; protein sel-1 homolog 1 isoform 2 precursor; ATP synthase mitochondrial F1 complex assembly factor; cAMP-specific 3′,5′-cyclic phosphodiesterase 4D; olfactory receptor 52I1; DDB1- and CUL4-associated factor 4-like protein 1; interferon regulatory factor 2-binding protein 2 isoform B; receptor-interacting serine/threonine-protein kinase 4; ATP synthase subunit gamma, mitochondrial; serine/threonine-protein kinase mTOR; Creatine kinase B-type 3.0--66 Leucine-rich repeat-containing protein 66; sodium/potassium/calcium exchanger 6, mitochondrial precursor; fin bud initiation factor homolog precursor; cullin-7 isoforms; inter-alpha-trypsin inhibitor heavy chain H3 preproprotein; FK506-binding protein 15; regenerating islet-derived protein 3-alpha precursor; Synaptotagmin 12; leucine-rich repeat-containing protein 36; heterogeneous nuclear ribonucleoprotein U-like protein 2; IQ and AAA domain-containing protein 1-like 1.1--35
Animal viruses Enterovirus species A, B, C, D, E, F, G, H, J; Rhinovirus A, B, C; Rabovirus; Sapelovirus 8 × 10^−5^--1.7 Enterovirus species A, B, C, D, E, F, G, H, J; Rhinovirus A, B, C; Rabovirus; Sapelovirus; Saffold virus; Coronavirus 9 × 10^−6^--1.5
Other viruses Bacterial phages 36--70 Bacterial phages 11--41
^1^BLASTp queries performed in public databases. Taxid: *Homo sapiens*, Viruses, Picornavirales, ssRNA viruses, Enteroviruses, Rhinoviruses.
###### Epitopes of MAbs 5D-8.1 and 9D5: significant alignments with the VP1 protein of enterovirus types belonging to the A, B, C, D species.
^1^Enterovirus types showing less than 70% sequence identity with the epitope sequence are highlighted in gray.
###### Enterovirus neutralization assays. Results for monoclonal antibodies 5D-8.1 and 9D5.
Antibody/antiserum Neutralizing titer[1](#t3-fn1){ref-type="fn"}
-------------------------------------- ----------------------------------------------- -----
Pan-enterovirus MAb 5D-8.1 \<8 \<8
Pan-enterovirus MAb 59D5 \<8 \<8
Anti-CV-B4 MAb 356 (control) \<8 512
Anti-CV-B1 horse antiserum (control) 4096 \<8
^1^Neutralizing titer expressed as the inverse of the greatest dilution giving a positive result.
| {
"pile_set_name": "PubMed Central"
} |
Introduction {#Sec1}
============
The detection and classification of voltage deflections in extracellular recordings caused by action potentials of neurons - called spikes - is known as spike sorting. It is necessary if single neuronal activities must be resolved from multi-neuronal firing activity. The assignment of spikes to neurons is only possible because the spike shapes of neurons differ due to their morphology, their spatial position with respect to the recording electrode(s), the intrinsic membrane properties of the neuron and the surrounding medium (Camuñas-Mesa and Quiroga [@CR6]; Gold et al. [@CR18]). Furthermore, at least to a first approximation, spikes from the same neuron have similar waveforms. It is, therefore, possible to group extracellular potentials based on their waveform assuming that spikes within one group were actually emitted by the same neuron.
In principle, extracellular recordings can be seen as having two different linearly added components, background activity (noise and small action potentials from far away neurons) and spikes of close-by cells (Buzsáki [@CR4]; Camuñas-Mesa and Quiroga [@CR6]). Therefore, for any given piece of data, spike sorting typically solves three problems (Einevoll et al. [@CR9]; Lewicki [@CR27]): First, spikes are detected in the noisy recording. Second, spikes are extracted from the data, aligned and their dimensionality is reduced using feature extraction. Third, spikes, which are now represented by a small number of features, are grouped into clusters of similar spike shapes that putatively originate from the same neuron. However, there is another option: Spikes can be assigned to neurons by using knowledge from a preceeding clustering step, which we refer to as spike classification. At first the classification problem seems to be redundant because spikes are assigned to putative neurons already during the clustering step. But there are several reasons why it is important to treat the classification problem separately from the clustering step. Among these, classification is usually much faster than clustering, an important advantage for online applications, where it might be desirable to use a fast classifier that was derived from an initial offline clustering for real-time spike sorting. Additionally, many clustering procedures scale poorly with the number of spikes and their application becomes infeasible for very long recordings. Then, only a subset of spikes can be clustered and the rest simply classified.
One way to build such a classifier is to calculate the average of all elements for each cluster. This cluster center is called the template. Each unclassified spike is then compared to each template and is subsequently assigned to the template that was most similar to it according to some appropriate similarity measure. This procedure is often referred to as template matching.
Here, we focus exclusively on the detection and the classification problem: If the number of neurons and their templates are known, *e.g.*, as the result of an offline spike sorting procedure, what is the best way to perform template matching? Different approaches to solve this problem were proposed (see *e.g.,* (Abeles and Goldstein [@CR1]; Friedman [@CR15]; Gerstein and Clark [@CR16]; Keehn [@CR24]; Salganicoff et al. [@CR41]) but also recent approaches (Vargas-Irwin and Donoghue [@CR47]; Zhang et al. [@CR50])) including filter based methods (Roberts and Hartline [@CR39]) (see Fig. [1](#Fig1){ref-type="fig"} for an illustration). Although template matching is of great importance for the analysis of extracellular recordings, it was not thoroughly investigated yet what the best strategy is. Even current commercial products rely on very simple strategies like Euclidean distance (Cambridge Electronic Design Limited [@CR5]; Plexon Inc [@CR34]) with manual threshold selection.Fig. 1Illustration of different template matching techniques for a toy example (artificial data; y-axis arbitrary units). A short piece of a 4 electrode recording was simulated by copying two templates and a simulated ripple into Gaussian white noise. The first column of panels shows the data, the second column the templates (and derived filters), the third column the respective single electrode template matching outputs, and the last column the final multichannel template matching output. For all three methods spikes would have to be detected in the template matching outputs by thresholding (in **a** with a threshold on minima, for **b** and **c** on maxima of the output). The second to last column shows the contribution of individual electrodes to the final template matching output. **a** Euclidean distance of the data to each template. Known templates are subtracted at each possible position from the data (for each electrode individually) and the norm of the residual (over all electrodes) is computed. If this residual is close to zero, the template is assumed to be present in the data at the respective temporal position. **b** Convolution of the data with the multi-electrode templates. For each multi-electrode template the data from every electrode is convolved with the respective single-electrode template individually and the results are summed. **c** Convolution of the data with matched filters. Matched filters are computed from the template and the noise covariance matrix (in this example including the statistics of the "ripple")Fig. 2Flowchart of the BOTM method (*left*) and illustration of the most important steps (*right*) on an example overlapping spike from Benchmark 1 (Q) (*top*, *right*). BOTM outputs are computed for each template (box labeled "BOTM") and thresholded to detect and classify spikes. To resolve overlapping spikes two alternative approaches are possible. Box labeled "Option 1": from the individual template BOTM outputs more BOTM outputs are constructed that reflect the presence of overlapping spikes (light grey traces on the right, here shown only for overlaps of maximal 2 spikes and 3 samples temporal difference); the one with the highest peak is marked in dark grey). Box labeled "Option 2 (SIC)": Once a spike is detected by threshold crossing on the BOTM outputs, it is subtracted from each BOTM output and the threshold is reapplied iterativelyFig. 3Example of different template matching outputs on a public benchmark data set with a single electrode simulation. **a** Raw simulated data of two different data files from the benchmark. The true spike times (shown as ticks below the data) are known since the data was simulated. At the position of the true spike times the respective templates are superimposed over the data. **b** Output of Euclidean distance template matching. Colored traces correspond to the template matching output of the respective template in **a**. Even on the high signal-to-noise-ratio data file (*left*) not all spikes can be reliably detected by thresholding the output: the first spike of the orange neuron is obscured by the partial overlap. In the low SNR case (*right*) the output cannot be thresholded in a useful way to detect spikes. **c** Convolution-based template matching gives in both cases clear peaks at the positions of the spikes. But not always is the template matching output with the highest peak also associated with the correct template. **d** The matched filter is in this case very close to the convolution-based template matching. But for the difficult data set (*right*) all spikes would be classified as the blue neuron by a simple threshold. **e** Thresholding the BOTM output recovers spike times and their correct identities in both examples (with the exception of the overlapping spike in the difficult data set). This overlap would have to be resolved by a post-processing step. Since the color of the template matching outputs with the highest peak is not always clearly visible, a line with the same color is placed at the respective peaks (maxima in **c**-**e**, minima in **b**) and can be compared to the true spike identities in **a**Fig. 4Illustration of the BOTM responses on a 160 ms piece of benchmark 2 (H). **a** Intracellular recording of a single neuron. **b** Extracellular recording using a tetrode located nearby the neuron recorded in **a**. At the time points at which a spike was detected in the intracellular recording the spike-triggered average (template) of the target neuron superimposes the recordings (*cyan traces*) for each electrode. **c** BOTM output and sorted spikes. In this recording, 10 putative neurons were found by the initial spike sorting, one of which was assigned to the target neuron (*magenta*). Each putative neuron has an associated BOTM output. For sake of clarity, neurons for which no ground truth was available are depicted in grey. Small ticks over the traces show the sorting output of the BOTM + SIC procedure. The three spikes of the target neuron were correctly detected and classified. The spike in the middle was classified as part of an overlapping spike. The dotted line represents the detection threshold
In the toy example in Fig. [1](#Fig1){ref-type="fig"} three template matching procedures are illustrated and it can be seen that they respond differently to noise, artifacts (in this case a narrow band oscillation or "ripple") and overlapping spikes. Convolution (Fig. [1](#Fig1){ref-type="fig"}b) has the advantage that noise on electrodes which are too far away to measure the activity of a neuron, and, thus, do not carry information about the presence of the neuron's template, are suppressed (topmost electrode). Additionally, overlapping spikes produce detectable peaks in the output. However, an artifact or "ripple" may also lead to strong responses. If the noise and ripple characteristics are known, matched filters can be used instead, which have the potential to suppress unwanted signal components. The implementation in this case can be done by using the ripple on the seemingly useless electrode 2 - on which the templates have nearly no energy - to cancel the ripple on electrodes 3 and 4 (marked with "in phase" and "out of phase" respectively). All three methods have the problem that setting the detection threshold is not straight-forward.
Here, we show that template matching can be seen as a filtering operation and propose a finite impulse response (FIR) filter based online template matching procedure that is - under the additional assumption of Gaussian noise - optimal in a Bayesian sense. We derive its optimality (to discriminate spikes from different neurons and to detect the spikes in noisy recordings) from Fisher's linear discriminant analysis (LDA) and show that this way the outputs of matched filters can be interpreted in a Bayesian sense. The proposed Bayes optimal template matching (BOTM) computes the linear discriminant functions by a convolution of FIR filters with extracellular data and solves both the detection and the classification problem. In contrast to the other methods, BOTM provides an optimal detection threshold analytically and does not require manual intervention. Using BOTM to simultaneously detect and classify spikes removes the need for spike alignment since the peak of the detector output is a robust estimate of the true position of the spike.
We evaluate our method on two previously published benchmark data sets and show that we can significantly improve template matching performance in both cases. Additionally, using BOTM as a post-processing step after clustering can reduce the number of errors which occurred during the initial spike sorting step.
This framework is then extended to overlapping spikes. Detecting and resolving overlapping spikes is an important problem for spike sorting: waveforms of near coincident action potentials will interfere with each other, severely altering the individual waveforms and effecting their detection and classification. Approaches to solve this problem were suggested (Atiya [@CR2]; Ekanadham et al. [@CR10]; Franke et al. [@CR14]; Lewicki [@CR26]; Marre et al. [@CR29]; Pillow et al. [@CR33]; Prentice et al. [@CR36]; Segev et al. [@CR44]; Vargas-Irwin and Donoghue [@CR47]; Wang et al. [@CR49]; Zhang et al. [@CR50]) but are in general computationally expensive. Here, we show that resolving overlapping spikes in the linear discriminant function space could provide a computationally efficient alternative.
Due to its computational simplicity BOTM is especially useful for online and real-time implementation in the context of closed-loop experiments after the initial templates have been found (Einevoll et al. [@CR9]; Franke [@CR13]; Franke et al. [@CR14]; Obeid et al. [@CR32]; Rutishauser et al. [@CR40]).
Method {#Sec2}
======
Definitions {#Sec3}
-----------
Spike sorting relies on the assumption that the action potentials of a single neuron lead to extracellular spikes with similar waveforms (Lewicki [@CR27]). This is generally not true, since spikes of the very same neuron are known to vary, *e.g.*, dependent on the recent firing history of the neuron (Fee et al. [@CR11]) and slowly over time (Franke et al. [@CR14]). Here we will assume non-varying waveforms but, in principle, the filters could be adapted over time. We define the template of neuron *i* as ***ξ***~*i*~ = \[*ξ*~*i*,1~^*T*^ , *ξ*~*i*,2~^*T*^, ..., *ξ*~*i*,*N*~^*T*^\]^*T*^ where *ξ*~*i*,*c*~ is the waveform for neuron *i* on electrode *c* and (·)^*T*^ is the vector transpose. The single electrode waveforms are *L* samples long. Thus ***ξ***~*i*~ is a column vector with *LN* dimensions, where *N* is the number of recording electrodes, *i.e.*, the single electrode waveforms are concatenated. We use an analogous definition for any piece of multi-electrode data starting at time *t*: *X*(*t*) = \[***x***(*t*)~1~^*T*^, ***x***(*t*)~2~^*T*^, ..., ***x***(*t*)~*N*~^*T*^\]^*T*^, where ***x***(*t*)~*k*~ = \[*x*(*t*)~*k*~, *x*(*t* + 1)~*k*~, ..., *x*(*t* + *L* − 1)~*k*~\]^*T*^ represents *L* samples of recorded data on electrode *k* starting at time *t*. The output of a multielectrode FIR filter ***f*** = \[*f*~1~^*T*^, *f*~2~^*T*^, ..., *f*~*N*~^*T*^\]^*T*^ applied to the recordings (which is usually denoted as $\documentclass[12pt]{minimal}
\usepackage{amsmath}
\usepackage{wasysym}
\usepackage{amsfonts}
\usepackage{amssymb}
\usepackage{amsbsy}
\usepackage{mathrsfs}
\usepackage{upgreek}
\setlength{\oddsidemargin}{-69pt}
\begin{document}$$ y={\displaystyle \sum_k^N}\left({x}_k\star {f}_k\right) $$\end{document}$ where ⋆ is the cross-correlation between data of electrode *k* and the filter for electrode *k*) can now be expressed in terms of a vector multiplication: *y*(*t*) = *X*(*t*)^*T*^ ***f***. Note that ***ξ***~*i*~, ***f*** and *X*(*t*) are all column vectors in the same *LN* dimensional vector space.
The noise covariance matrix (Pouzat et al. [@CR35]) is given by: ***C*** = *E*\[*η*(*t*)*η*(*t*)^*T*^\] with *η*(*t*) being a piece of data where no spikes were detected. It characterizes noise from various sources, including small amplitude spikes from far away neurons. Thus ***C*** is of dimensions *LN* × *LN* and captures the complete spatio-temporal covariance (between electrodes and over time) that is induced by both, noise and undetected small amplitude spikes (Pouzat et al. [@CR35]). Since we assume that the data was high pass filtered *E*\[*η*(*t*)\] = 0 holds.
Classical template matching {#Sec4}
---------------------------
Two different similarity measures between a piece of data and a template are commonly employed for template matching: Euclidean distance template matching (*e.g.*, (Cambridge Electronic Design Limited [@CR5]; Plexon Inc [@CR34]; Sato et al. [@CR42])) and convolution or cross-correlation template matching (Friedman [@CR15]; Kim and McNames [@CR25]). Note that convolution and cross-correlation are identical, apart from the time reversal of the filter by the convolution. Although we use the cross-correlation throughout this work, we still refer to template matching based on the cross-correlation as convolutive template matching. The Euclidean distance at time *t* between data *X*(*t*) and template *ξ*~*i*~ is defined as$$\documentclass[12pt]{minimal}
\usepackage{amsmath}
\usepackage{wasysym}
\usepackage{amsfonts}
\usepackage{amssymb}
\usepackage{amsbsy}
\usepackage{mathrsfs}
\usepackage{upgreek}
\setlength{\oddsidemargin}{-69pt}
\begin{document}$$ {D}_i^{Euc}(t)={\left|\left|X(t)-{\xi}_i\right|\right|}^2={\left(X(t)-{\xi}_i\right)}^T\left(X(t)-{\xi}_i\right) = X{(t)}^TX(t)-2X{(t)}^T{\xi}_i+{\xi_i}^T{\xi}_i $$\end{document}$$and the cross-correlation as$$\documentclass[12pt]{minimal}
\usepackage{amsmath}
\usepackage{wasysym}
\usepackage{amsfonts}
\usepackage{amssymb}
\usepackage{amsbsy}
\usepackage{mathrsfs}
\usepackage{upgreek}
\setlength{\oddsidemargin}{-69pt}
\begin{document}$$ {D}_i^{XC}(t)=X{(t)}^T{\xi}_i. $$\end{document}$$
Template matching using the noise covariance matrix {#Sec5}
---------------------------------------------------
If the noise covariance matrix is known, the recorded data can be prewhitened (Pouzat et al. [@CR35]; Rutishauser et al. [@CR40]) before matching the templates. Prewhitening is a linear operation that transforms the cluster of "noise" waveforms in a way that it will be roughly spherical (standard normal distributed) after prewhitening. With respect to template matching, prewhitening is equivalent to using the squared Mahalanobis distance instead of Euclidean distance and the matched filtering procedure instead of the convolution with the template:$$\documentclass[12pt]{minimal}
\usepackage{amsmath}
\usepackage{wasysym}
\usepackage{amsfonts}
\usepackage{amssymb}
\usepackage{amsbsy}
\usepackage{mathrsfs}
\usepackage{upgreek}
\setlength{\oddsidemargin}{-69pt}
\begin{document}$$ {D}_i^{Maha}(t)={\left(X(t)-{\xi}_i\right)}^T{\boldsymbol{C}}^{-1}\left(X(t)-{\xi}_i\right) = X{(t)}^T{\boldsymbol{C}}^{-1}X(t)-2X{(t)}^T{\boldsymbol{C}}^{-1}{\xi}_i+{\xi_i}^T{\boldsymbol{C}}^{-1}{\xi}_i $$\end{document}$$$$\documentclass[12pt]{minimal}
\usepackage{amsmath}
\usepackage{wasysym}
\usepackage{amsfonts}
\usepackage{amssymb}
\usepackage{amsbsy}
\usepackage{mathrsfs}
\usepackage{upgreek}
\setlength{\oddsidemargin}{-69pt}
\begin{document}$$ {D}_i^{Match}(t)=X{(t)}^T{\boldsymbol{C}}^{-1}{\xi}_i=X{(t)}^T{\boldsymbol{f}}_i $$\end{document}$$
where ***f***~*i*~ = ***C***^− 1^*ξ*~*i*~ is the matched filter (see the [appendix](#Sec20){ref-type="sec"} for a more detailed explanation) for template *ξ*~*i*~. Figure [1](#Fig1){ref-type="fig"} illustrates different template matching techniques applied to a toy example. We will refer to *D*~*i*~^*Euc*^ or *D*~*i*~^*Maha*^ as subtractive template matching since templates are subtracted from the data and the energy of the residual is used for spike detection and classification. Accordingly, *D*~*i*~^*XC*^ and *D*~*i*~^*Match*^ are referred to as convolutive template matching.
Bayes optimal template matching (BOTM) {#Sec6}
--------------------------------------
Template matching has to solve two tasks: the detection of a known signal in a noisy recording and the discrimination of spikes originating from different neurons. In the Gaussian noise case, these two problems have known solutions (see [appendix](#Sec20){ref-type="sec"}): the optimal linear detector (*i.e.*, the linear filter that maximizes the signal-to-noise ratio after filtering) is given by matched filtering (Van Trees [@CR46]), while the optimal discrimination between several clusters that share the same covariance matrix is given by linear discriminant analysis (LDA) (Fisher [@CR12]). These two solutions are interrelated, and, by using the probabilistic framework of LDA, it can be shown that a template matching exists that is optimal with respect to both tasks, *i.e.*, it maximizes the signal-to-noise ratio for each neuron while also maximizing their discriminability. This Bayes optimal template matcher (BOTM) is given by:$$\documentclass[12pt]{minimal}
\usepackage{amsmath}
\usepackage{wasysym}
\usepackage{amsfonts}
\usepackage{amssymb}
\usepackage{amsbsy}
\usepackage{mathrsfs}
\usepackage{upgreek}
\setlength{\oddsidemargin}{-69pt}
\begin{document}$$ {D}_i^{BOTM}(t)=X{(t)}^T{\boldsymbol{C}}^{-1}{\xi}_i-\frac{1}{2}{\xi_i}^T{\boldsymbol{C}}^{-1}{\xi}_i+ ln\left(p(i)\right)=X{(t)}^T{\boldsymbol{f}}_i-\frac{1}{2}{\xi_i}^T{\boldsymbol{f}}_i+ ln\left(p(i)\right) $$\end{document}$$
where *p*(*i*) is the prior probability to observe a spike of neuron *i*. For a derivation see eq. ([7](#Equ7){ref-type=""}) in the appendix. In this work we set the prior probabilities equal to a constant (see section"[Evaluation metrics](#Sec9){ref-type="sec"}"and Fig. [5](#Fig5){ref-type="fig"}), although, in principle, the estimated firing rates of the individual neurons could be used. The detection threshold for a spike in the BOTM filter outputs *D*~*i*~^*BOTM*^(*t*) (see Fig. [5](#Fig5){ref-type="fig"}) is then given by (see eq. ([9](#Equ9){ref-type=""}) in the appendix)$$\documentclass[12pt]{minimal}
\usepackage{amsmath}
\usepackage{wasysym}
\usepackage{amsfonts}
\usepackage{amssymb}
\usepackage{amsbsy}
\usepackage{mathrsfs}
\usepackage{upgreek}
\setlength{\oddsidemargin}{-69pt}
\begin{document}$$ thr= \ln \left(1-{\displaystyle \sum_i}p(i)\right) $$\end{document}$$which is usually close to 0 for realistic spike priors (see Fig. [3](#Fig3){ref-type="fig"} bottom row).Fig. 5Distributions of local maxima (in a window of 15 samples) in BOTM filter outputs for single spikes, overlaps (within 15 samples), and noise for the three neurons in benchmark 1 (Q) Easy1 N.10 (**a**) and Difficult 2 N.20 (**b**--**c**). The optimal threshold *thr* for p(n) = 0.99 is shown as a vertical line (magenta). For each neuron, spikes from other neurons (second row from bottom, grey) can cause filter responses to cross the threshold. Thus, spike classification can only be done reliably by combining the information from all three BOTM outputs. **a**-**b** The majority of overlapping spikes (second row from top, dark green) causes the BOTM outputs of all participating neurons to cross threshold. Please note that with *thr* = ln(*p*(*n*)), for this example, the exact choice of p(n) does not strongly influence sensitivity: for a wide range of choices for p(n) the threshold will be close to zero which will separate the noise from the spike distribution. **a** For all three neurons the response to noise (bottom, red) is well separated from the response to the target spikes (top, green) by a large margin. Spike detection and classification of single spikes can be done without error (except for overlapping spikes) since the green, grey and red distributions do not overlap. **b** In the low SNR case the green and red distribution are close to each other and start overlapping. Note that the overlap between the "single spike" distribution and the "other spikes" distribution does not directly imply any classification errors, since the classification depends on the maximal response over all three neurons. **c** Projection of the spikes from **b** in the response space spanned by two of the three BOTM outputs (all three combinations are shown). In this space, the discrimination boundary given by a max-operation on the BOTM outputs is the identity line (magenta). Spikes are colored according to their ground truth identity: spikes that should elicit the strongest response for the BOTM output on the x-axis are shown in green; spikes that should elicit the strongest response for the BOTM output shown on the y-axis, and, therefore, should lie above the identity line, are shown in cyan. Spikes of the neuron for which the BOTM output is not shown are grey. If all BOTM outputs are below the threshold (0) they are not detected ('noise region'). For comparison, the respective decision boundary of *D* ~*i*~^*Match*^ is shown as a dashed line
BOTM computes matched filter outputs for each template (see Fig. [4](#Fig4){ref-type="fig"} for an example with 10 templates). For each template a constant, which depends on the energy (*ξ*~*i*~^*T*^***C***^− 1^*ξ*~*i*~) of the respective template and on the probability *p*(*i*) that the template occurs in the data, is added to the filter outputs to compute the final discriminant functions. Spikes are detected and classified by thresholding those discriminant functions (the BOTM output). For each detected peak, the template that has the maximal BOTM output in a small temporal window around the peak is assigned to the spike. The individual processing steps of the method are summarized in Fig. [2](#Fig2){ref-type="fig"}. Figure [3](#Fig3){ref-type="fig"} shows a visual comparison of the results of BOTM and other template matching procedures applied to a spike sorting benchmark (Quiroga et al. [@CR38]).
In summary, the method works as follows (Fig. [2](#Fig2){ref-type="fig"}): 1. Precomputed matched filters are convolved with the incoming data. 2. Discriminant functions are computed from the matched filter outputs by adding the respective constants. 3. The detection threshold is applied to all discriminant functions. 4. For each threshold crossing, the local maximum of all discriminant functions after the threshold crossing is detected (and before all discriminant functions fall again below the threshold). 5. The temporal position of the peak and the identity of the discriminant function define the identity and time point of the detected spike.
Resolution of overlapping spikes {#Sec7}
--------------------------------
If two neurons fire an action potential very close in time, their spike waveforms will overlap in the recording (Fig. [2](#Fig2){ref-type="fig"}, right). Overlapping spikes are difficult to detect and classify, since the overlap waveform might be very different from the individual spike waveforms. However, as long as the individual waveforms do not cancel each other in a way that the overlap waveform has virtually zero energy, the individual matched filters do still respond to the overlap. Thus, BOTM will assign the single template with the highest peak to an overlap (which could also be a template not participating in the overlap if the overlap waveform is coincidentally similar to that template) missing at least one of the spikes (Fig. [2](#Fig2){ref-type="fig"}, right, 'BOTM'). To also detect and resolve overlapping spikes one could, in principle, construct all possible overlaps between all available templates with different temporal shifts and add those to the template set. This would, however, dramatically increase the number of templates and introduce two problems: First, with increasing number of templates, each waveform, including noise, can be described by a certain combination of templates. Second, this approach, in a naïve implementation, would be computationally prohibitive. Both problems will be addressed in the following.
For independent spike trains the prior probability to observe an overlap is equal to the product of the individual single-spike prior probabilities (the spike prior). Thus, the more templates are involved in an overlap, the lower is the prior probability for the resulting waveform. Therefore, solutions that feature less templates are naturally favored in our probabilistic framework. With increasing number of templates in an overlap the prior probability to observe such an event decreases, providing a natural cutoff to how many spikes per overlap have to be considered.: At some point the discriminant function for the overlap, *i.e.*,$$\documentclass[12pt]{minimal}
\usepackage{amsmath}
\usepackage{wasysym}
\usepackage{amsfonts}
\usepackage{amssymb}
\usepackage{amsbsy}
\usepackage{mathrsfs}
\usepackage{upgreek}
\setlength{\oddsidemargin}{-69pt}
\begin{document}$$ {d}_{overlap}\left({\xi}_{overlap}\right)={\xi_{overlap}}^T{\boldsymbol{C}}^{-1}{\xi}_{overlap}-\frac{1}{2}{\xi_{overlap}}^T{\boldsymbol{C}}^{-1}{\xi}_{overlap}+ ln\left(p(overlap)\right)<thr $$\end{document}$$
will never cross the detection threshold, even for the actual overlap waveform *ξ*~*overlap*~ (see eq. ([10](#Equ10){ref-type=""}) in the appendix). Still, it would be computationally very expensive to compute all convolutions between all (overlap-) templates and the data. However, this is not necessary, since the overlap BOTM discriminant functions (*i.e.,* the BOTM output *d*~*i* + *j*~^*τ*^ for an overlap between template *i* and *j* with temporal shift *τ*) can be directly computed from the individual spike discriminant functions by$$\documentclass[12pt]{minimal}
\usepackage{amsmath}
\usepackage{wasysym}
\usepackage{amsfonts}
\usepackage{amssymb}
\usepackage{amsbsy}
\usepackage{mathrsfs}
\usepackage{upgreek}
\setlength{\oddsidemargin}{-69pt}
\begin{document}$$ {d}_{i+j}^{\tau }(t)={d}_i(t)+{d}_j^{\tau }(t)-{\xi_i}^T{\boldsymbol{C}}^{-1}{\xi}_j^{\tau } $$\end{document}$$
where *d*~*j*~^*τ*^(*t*) and *ξ*~*j*~^*τ*^ are the temporally shifted discriminant function and template of neuron *j*, respectively (see Fig. [2](#Fig2){ref-type="fig"}, right, 'Option 1' and eq. ([10](#Equ10){ref-type=""}) in the appendix for the derivation). Depending on the number of spikes per overlap and their maximal temporal shift considered, this approach can still be computationally expensive.
Therefore, here, we employ a greedy approach. We assume that at least one of the single-spike BOTM outputs will cross the threshold for each overlap and that its peak is giving approximately the correct position of the template in the data. Then, we can subtract the expected influence of this spike from the other discriminant functions, a process which is also referred to as subtractive interference cancellation (SIC, see Fig. [2](#Fig2){ref-type="fig"}, right, 'Option 2' and eq. ([12](#Equ12){ref-type=""}) in the appendix for the derivation). Fortunately, under the assumption that at least one single-spike discriminant function crosses the threshold for each overlapping spike, we only need to compute the overlap discriminant functions in temporal periods around threshold crossings of single-spike discriminant functions.
Noise Covariance Matrix {#Sec8}
-----------------------
The noise covariance matrix ***C*** plays a crucial role for the Mahalanobis distance, the matched filter and BOTM as well as for some spike sorting procedures (Franke et al. [@CR14]; Marre et al. [@CR29]; Pillow et al. [@CR33]; Pouzat et al. [@CR35]; Shoham et al. [@CR45]). In all cases its inverse is needed. However, the noise covariance matrix can be badly conditioned, *i.e.*, it might have eigenvalues that are close to zero, which makes its inversion unstable. A standard procedure to invert ill conditioned covariance matrices is diagonal loading or shrinkage (Hiemstra [@CR22]; Van Trees [@CR46]): A target covariance matrix ***C***~*T*~, often the identity, is added or merged to the original covariance matrix. We decided to chose the diagonal of ***C***, ***C***~*D*~ = *diag*(***C***) as the target and blended it with the original matrix according to ***C***~*L*~ = *α****C*** + (1 − *α*)***C***~*T*~. We noticed that the choice of *α* influenced performance mainly through the temporal length of the resulting filter responses to spikes. With decreasing *α* the filters become more similar to a narrow bandpass filter which could lead to oscillating filter responses. Oscillations in the filter outputs were not a problem for benchmark 3 BOTM + SIC (see below, SIC iteratively subtracts the filter responses to spikes from the filter outputs, therefore removing all oscillations). We did not try to optimize diagonal loading but simply chose *α* = 0.5. For benchmark 3, BOTM + SIC we chose an *α* so that the maximal eigenvalue of ***C***~*L*~ divided by its minimal eigenvalue (*i.e.,* the condition number of ***C***~*L*~) was not larger than 10,000 to ensure that ***C***~*L*~ was invertible.
The noise covariance matrix can be either computed on pieces of noise, that is, periods of data where no spikes were detected (*e.g.*, as in (Marre et al. [@CR29]; Pouzat et al. [@CR35]; Rutishauser et al. [@CR40])), or on residuals of the spikes after subtracting the templates (*e.g.*, (Pillow et al. [@CR33]; Shoham et al. [@CR45])). Since the residuals may contain additional variability, *e.g.*, from mis-alignment of the templates (the correct template was subtracted at a wrong position), clustering errors (the wrong template was subtracted), or variability in the neuronal signal (the correct template does not fit perfectly with the occurring waveform), we chose to compute ***C*** on stretches of noise. This has the side-effect that we do not run the risk of overfitting ***C*** to the spike waveforms.
If *T* is the length of the templates (measured in samples) and *N* is the number of recording electrodes, the dimensionality of ***C*** is *T* ⋅ *N* times *T* ⋅ *N*, which can be very large and thus difficult to estimate. However, ***C*** has a special block structure (Pouzat et al. [@CR35]):$$\documentclass[12pt]{minimal}
\usepackage{amsmath}
\usepackage{wasysym}
\usepackage{amsfonts}
\usepackage{amssymb}
\usepackage{amsbsy}
\usepackage{mathrsfs}
\usepackage{upgreek}
\setlength{\oddsidemargin}{-69pt}
\begin{document}$$ \boldsymbol{C}=\left(\begin{array}{ccc}\hfill {C}_{1,1}\hfill & \hfill \cdots \hfill & \hfill {C}_{N,1}\hfill \\ {}\hfill \vdots \hfill & \hfill \ddots \hfill & \hfill \hfill \\ {}\hfill {C}_{1,N}\hfill & \hfill \hfill & \hfill {C}_{N,N}\hfill \end{array}\right) $$\end{document}$$
where the blocks *C*~*i*,*k*~ are the covariance matrices between electrode *i* and *k* with *C*~*i*,*i*~ = *C*~*i*,*i*~^*T*^ and *C*~*i*,*k*~ = *C*~*k*,*i*~^*T*^. Each block is a Toeplitz matrix of the cross-correlation function of electrode *i* and *k*. Therefore, we did not estimate ***C*** by directly computing the covariance matrix from *T* ⋅ *N* dimensional snippets of noise, where it would not be guaranteed that (*C*~*i*,*k*~)~*m*,*n*~ = (*C*~*k*,*i*~)~*n*,*m*~ and (*C*~*i*,*k*~)~*m*,*n*~ = (*C*~*i*,*k*~)~*m* + 1,*n* + 1~, but by constructing ***C*** from the respective auto- and cross-correlation functions. This procedure reduces the number of free parameters that need to be estimated to *T* ⋅ *N*, which again reduces the risk of overfitting.
Evaluation metrics {#Sec9}
------------------
We evaluated the spike detection and classification performance of the template matching procedures on two different data sets described below. Both were preprocessed in a similar way: The available ground truth information was used to cut true spikes from the recording and to construct the templates. This avoided the problem of aligning the spikes to compute the templates. The template matchers were evaluated for detection and classification performance separately.
For the detection task the template matching output distribution was computed for all spikes as well as for pieces of noise. For each detector the detection error was defined as$$\documentclass[12pt]{minimal}
\usepackage{amsmath}
\usepackage{wasysym}
\usepackage{amsfonts}
\usepackage{amssymb}
\usepackage{amsbsy}
\usepackage{mathrsfs}
\usepackage{upgreek}
\setlength{\oddsidemargin}{-69pt}
\begin{document}$$ \mathrm{Detection}\ \mathrm{Error}=\mathrm{F}\mathrm{P}+\mathrm{F}\mathrm{N} $$\end{document}$$
where FP is the number of false positive noise detections and FN is the number of false negative misses. Then, the overlap of the template matching response distributions to noise and spikes was computed and, except for BOTM, the optimal threshold (which minimized the detection error) was estimated numerically. The alternative would have been to compare receiver operator characteristic curves, but BOTM directly provides a threshold. Therefore, we decided to compare its performance to the optimal performance of the other methods.[1](#Fn1){ref-type="fn"} To detect spikes in the BOTM output a prior probability to observe a spike needs to be chosen. Optimally, this probability is different for each neuron and depends on its firing rate. However, for the benchmarks used in this study, the performance was insensitive to the exact choice of the priors (see Fig. [5](#Fig5){ref-type="fig"}). Therefore, instead of choosing a prior for each neuron independently, we assumed them to be equal and set the prior for noise to 0.99 (*i.e.*, we assumed it 100 times more likely to observe noise than a spike).
The sensitivity, *i.e.*, the detection performance at the optimal threshold was defined as$$\documentclass[12pt]{minimal}
\usepackage{amsmath}
\usepackage{wasysym}
\usepackage{amsfonts}
\usepackage{amssymb}
\usepackage{amsbsy}
\usepackage{mathrsfs}
\usepackage{upgreek}
\setlength{\oddsidemargin}{-69pt}
\begin{document}$$ \mathrm{Sensitivity} = 100\frac{\mathrm{TP}}{\mathrm{TP}+\mathrm{F}\mathrm{N}}, $$\end{document}$$
which denotes the percentage of correct spike detections (TP) divided by the total number of spikes. Note that this quantity indirectly includes the number of FPs since we minimized FP + FN to find the optimal threshold. In this study we use two different data sets (see below) that differ in the availability of ground truth data: for one data set we know the spike times of all neurons (and, therefore, also the number of neurons), while, in the other data set, spike times are known only for one neuron and it is unknown how many more neurons have been recorded from. For this reason, it is not possible to use the same evaluation metric for both data sets. We, therefore, describe the data-set-dependent evaluation metrics in the following section, together with the applied benchmarks.Benchmark 1 (Q): Evaluation on simulated data with full ground truthThe proposed template matching was evaluated on the publicly available spike sorting benchmark data set described in (Quiroga et al. [@CR38]) which we will refer to as benchmark 1 (Q). This data set was already used by other researchers for spike sorting evaluation (see, *e.g.*, (Bestel et al. [@CR3]; Ghanbari et al. [@CR17]; Herbst et al. [@CR21])). The data set consists of 4 sub benchmarks labeled "Easy1" to "Diffcult2". Every sub benchmark consists of 4 different data files (with the exception of "Easy1" which has 8) with decreasing signal-to-noise ratios. All files contain 60s of a simulated single electrode extracellular recording with 24 kHz sampling rate and 3 simulated neurons. Templates and the noise covariance matrix were calculated using the available ground truth information. Short periods of simulated data of length *L* = 61 samples starting 15 samples before the given spike time points were cut and averaged to create the templates. In this benchmark, the ground-truth spike times do not indicate the position of the peak of the spike waveforms in the data but rather their onset. Therefore, we corrected the original spike times by shifting the complete spike train of each neuron by a constant number of samples. This shift yielded a ground truth that reflected the peak positions of the spikes in the data (a feature also a spike sorter without any knowledge of the ground truth could estimate), not their onset. The template matching outputs for the different template matching procedures were computed on the noisy spike waveforms present in the data that were aligned with the corrected ground truth information. Each spike was assigned to the template with the maximal response. Spikes that were correctly assigned were counted as TP while spikes assigned to the wrong template were counted as classification errors CN. The quantity CN relies on knowing the full ground truth, that is, the exact spike times for all neurons. We used the following quantities for the performance comparison:$$\documentclass[12pt]{minimal}
\usepackage{amsmath}
\usepackage{wasysym}
\usepackage{amsfonts}
\usepackage{amssymb}
\usepackage{amsbsy}
\usepackage{mathrsfs}
\usepackage{upgreek}
\setlength{\oddsidemargin}{-69pt}
\begin{document}$$ \mathrm{Classification}\ \mathrm{Performance}=100\frac{\mathrm{TP}}{\mathrm{TP}+\mathrm{C}\mathrm{N}} $$\end{document}$$We chose to weight detection and classification equally, combining them in a final performance score of:$$\documentclass[12pt]{minimal}
\usepackage{amsmath}
\usepackage{wasysym}
\usepackage{amsfonts}
\usepackage{amssymb}
\usepackage{amsbsy}
\usepackage{mathrsfs}
\usepackage{upgreek}
\setlength{\oddsidemargin}{-69pt}
\begin{document}$$ \mathrm{Total}\ \mathrm{Performance}=\frac{\left(\mathrm{Sensitivity}+\mathrm{Classification}\right)}{2}. $$\end{document}$$Benchmark 2 (H): Evaluation on real recording with partial ground truthThis data set is the part of the hc-1 data set described in (Henze et al. [@CR20]) and publicly available under <http://crcns.org/> and was already used for evaluation of spike sorting algorithms (see, *e.g.*, (Ekanadham et al. [@CR10]; Harris et al. [@CR19]; Schmitzer-Torbert et al. [@CR43])). The following files were used: d11221.002, d11222.001, d12821.001 and d14521.001. We chose the files depending on the quality of the intracellular recording, *i.e.*, the ones where the intracellular recording showed clearly visible and easily detectable spikes during the whole recording. Each data file consists of several minutes of simultaneous intra- and extracellular recordings in rat hippocampus with a sampling rate of 20 kHz. Ground truth information was available for only one single neuron, extracted from the respective intracellular recording. The extracellular recordings were high pass filtered at 300Hz.Spike sorting using mean-shift clustering (Marre et al. [@CR29]) was performed in the space of the first 6 principal components after prewhitening to estimate the templates. The sorting was not optimized manually. The template of the cluster whose spikes best matched the ground truth was used to estimate the performance of the template matching procedures and will be referred to as the "target template". For all data files the target template was very similar to the template that can be obtained by using only the spikes given by the ground truth, *i.e.*, in all cases spikes from one of the clusters matched the ground truth well enough to get good template estimation.For the classification task, the template matching output was computed for all templates and all spikes of the ground truth neuron, as well as for all other spikes detected during the spike sorting. Spikes of the ground truth that were correctly matched to the target template were counted as true positives (TP). Since we do not have the full ground truth for this data set, the quantity CN from the previous benchmark cannot be computed for all neurons. We therefore counted spikes of the target neuron that were falsely assigned to a putative other neuron as CNt. Spikes which were found by the automatic spike sorting procedure which belong to putative other neurons and which were correctly not assigned to the target neuron were counted as true negatives (TN). Spikes not included in the ground truth that were assigned to the target template were counted as false positives (FP). How many of the spikes that actually belong to the target neuron were also classified correctly? This is the classification performance.$$\documentclass[12pt]{minimal}
\usepackage{amsmath}
\usepackage{wasysym}
\usepackage{amsfonts}
\usepackage{amssymb}
\usepackage{amsbsy}
\usepackage{mathrsfs}
\usepackage{upgreek}
\setlength{\oddsidemargin}{-69pt}
\begin{document}$$ \mathrm{Classification}\ \mathrm{Performance}=100\frac{\mathrm{TP}}{\mathrm{TP}+\mathrm{C}\mathrm{N}\mathrm{t}} $$\end{document}$$And of all putative spikes detected in the recording that do not belong to the target neuron, how many were correctly classified as noise or spikes from other neurons? This is the specificity (the performance in rejecting spikes that do not belong to the target neuron).$$\documentclass[12pt]{minimal}
\usepackage{amsmath}
\usepackage{wasysym}
\usepackage{amsfonts}
\usepackage{amssymb}
\usepackage{amsbsy}
\usepackage{mathrsfs}
\usepackage{upgreek}
\setlength{\oddsidemargin}{-69pt}
\begin{document}$$ \mathrm{Specificity}=100\frac{\mathrm{TN}}{\mathrm{FP}+\mathrm{T}\mathrm{N}} $$\end{document}$$Many different ways of combining the different performance measures into a final score are possible; however, here, we decided to combine them with equal weight into a final score, the total performance.$$\documentclass[12pt]{minimal}
\usepackage{amsmath}
\usepackage{wasysym}
\usepackage{amsfonts}
\usepackage{amssymb}
\usepackage{amsbsy}
\usepackage{mathrsfs}
\usepackage{upgreek}
\setlength{\oddsidemargin}{-69pt}
\begin{document}$$ \mathrm{Total}\ \mathrm{Performance}=\frac{\left(\mathrm{Sensitivity}+\mathrm{Classification}+\mathrm{Specificity}\right)}{3} $$\end{document}$$Benchmark 3 (Q): Online template matching of full recordingTo compare the performance of BOTM to the spike sorting performance reported in (Quiroga et al. [@CR38]), we did not use the ground truth to cut perfectly aligned spikes (but the ground truth was used to compute the correct templates). Instead, BOTM was run on the whole data and spikes were detected in the template matching output.We used the same data as in benchmark 1 (Q) to evaluate the performance of BOTM as a refinement tool for spike sorting procedures. For this, the performance of BOTM using the correct templates was compared to the performance of a clustering based spike sorting. We chose the spike sorter "wave_clus" described in (Quiroga et al. [@CR38]). "wave_clus" detects spikes using a voltage threshold. The threshold is computed *via* the median absolute deviation (MAD) of the data which is, in the presence of spikes, a more robust way to estimate the standard deviation of the noise. Then, wavelet coefficients are computed for each spike. A subset of the wavelet coefficients is chosen which, putatively, carries most information about the identity of the spikes. Those coefficients are then clustered by using superparamagnetic clustering, a clustering method developed in the context of statistical mechanics and based on simulated interactions between each data point and its K-nearest neighbors.For BOTM, the correct templates and the noise covariance matrix were computed from the data using the available ground truth information. To be more specific, ground-truth spike trains were used to cut short pieces out of the data around the true locations of the spikes. These pieces were averaged to calculate the templates. Therefore, the ('correct') templates contain a realistic amount of noise due to the averaging process. Pieces of data in which no spikes were present were used to compute the noise auto-covariance function from which the noise covariance matrix was estimated. As a consequence, also the noise covariance matrix was noisy due to the estimation process on a limited amount of noisy data. The optimal filters were computed and convolved with the whole extracellular recordings. The discriminant functions were then thresholded with the optimal threshold (see eq. ([9](#Equ9){ref-type=""}) in the appendix). For each period during which at least one discriminant function was above the threshold, the peak of the maximal discriminant function was detected. The identity of the discriminant function was used as the neurons' identity and the position of the peak as the time of the spike. This way, detection, alignment and classification were implemented in a single operation. To avoid a second detection of the same peak due to multiple peaks of the filter response, peaks within 8 samples (.33 ms) were compared and only the maximum was kept. Thus, overlapping spikes within this time window could not be resolved. The BOTM + SIC method was evaluated in the same way but, additionally, once a spike was found in the data and assigned to a neuron, the expected filter response to the neuron's template was subtracted from the discriminant functions (eq. ([10](#Equ10){ref-type=""}) in the appendix) and the detection step was repeated.
Results {#Sec10}
=======
Performance for individual spikes {#Sec11}
---------------------------------
Figure [6](#Fig6){ref-type="fig"} shows a performance comparison of the presented template matching procedures (Benchmark 1 and 2). While *D*~*i*~^*XC*^(*t*) and *D*~*i*~^*Match*^(*t*) give nearly 100% sensitivity in all cases, sensitivity of *D*~*i*~^*Euc*^(*t*) and *D*~*i*~^*Maha*^(*t*) is close to zero. For these two methods, the respective detector output distributions for spikes and noise are strongly overlapping. Lowering the number of misses (FN) thus strongly increases the number of false positive detections. The optimal threshold is therefore so high, that nearly no spike is detected at all. For classification *D*~*i*~^*Euc*^(*t*) and *D*~*i*~^*Maha*^(*t*) have a consistently high performance while *D*~*i*~^*XC*^(*t*) and *D*~*i*~^*Match*^(*t*) perform poorly. The classification performance of *D*~*i*~^*XC*^(*t*) and *D*~*i*~^*Match*^(*t*) has a high variance since for some files, the optimization procedure to find the optimal threshold was able to set it so high that only spikes of a single neuron were found. For data files in which the target template is indeed the template with the highest amplitude, these procedures have thus near perfect performance. The performance of *D*~*i*~^*XC*^(*t*) and *D*~*i*~^*Match*^(*t*) for data files in which the target template had a small or intermediate amplitude is significantly worse.Fig. 6Performance evaluation benchmark 1 (Q) (*left column*) and benchmark 2 (H) (*right column*). *Error bars* indicate standard deviation among different data files of the same benchmark. **a** Total performance of different template matching procedures and **b** Performance for different error categories defined in section "[Evaluation metrics](#Sec9){ref-type="sec"}"
The results suggest that subtractive template matching, in the way we used it here, is basically useless to detect spikes. The bad performance can be remedied to a certain degree by limiting the procedure to only templates with large amplitude and by reducing the dimensionality of the templates through either restricting the number of channels, *e.g.*, to one, or shortening the templates. Reducing the dimensionality helps because the main reason for the low sensitivity is the influence of the high dimensional noise on the waveform which is over-proportionally diminished by removing dimensions. However, there is an optimal way to reduce the influence of high dimensional noise and increase the signal-to-noise ratio: the matched filter.
BOTM increases the average performance from below 60% to over 95% with respect to Euclidean distance template matching. If only the classification performance is considered, using Mahalanobis distance or BOTM increased the average performance with respect to Euclidean distance from 91.8% to 97.2% on benchmark 1 and from 91.8% to 93.9% on benchmark 2. We want to point out that these error numbers are not meant to reflect actual template matching performance and likely overestimate performance. In this benchmark only correctly aligned single spikes were used. A situation hardly realistic for real recordings, since noise and the presence of other spikes make a correct alignment difficult. Furthermore, except for BOTM, detection thresholds were set to optimize performance.
The difference between BOTM and *D*~*i*~^*Match*^(*t*) is a template-dependent additive constant, which depends on the energy of the respective template and its prior probability. Adding this constant to the output of the matched filters allows for distinguishing between neurons by using a simple maximum operation (see Fig. [5c](#Fig5){ref-type="fig"}). While the decision boundary of a maximum operation on the matched filter output will cut through the clusters (Fig. [5c](#Fig5){ref-type="fig"}, dashed line), BOTM moves the coordinate system so that the identity becomes the optimal decision boundary, similar to what LDA would provide. This explains the increased specificity of BOTM as compared to *D*~*i*~^*Match*^(*t*). However, to also provide high specificity, the matched filters need to distinguish signal from noise. Since all templates usually share similar characteristics, such as frequency content, the outputs between the different matched filters are correlated (elongation of clouds in Fig. [5c](#Fig5){ref-type="fig"}), which would not be necessarily expected from clusters in the subspace given by pure LDA.
We conclude that while convolution-based template matching provides high sensitivity, distance-based template matching on well aligned spikes provides good classification performance. BOTM combines both advantages in a single operation, thus avoiding the problem of spike alignment. Furthermore, the performance of its analytical threshold is comparable to the optimal threshold of the other methods.
Effects of errors in the template set on BOTM performance {#Sec12}
---------------------------------------------------------
The performance values reported so far have been established under the condition of prior knowledge of the template set. But in how far do errors in the estimation of the templates, *e.g.*, by an initial spike sorting, affect the template matching? A possible error in the creation of template set could be introduced if all spikes of one neuron are missed or falsely classified as noise. In this case, the corresponding template would be missing in the template set. We studied the effect of missing templates on benchmark 2 (H) (the data set also shown in Fig. [4](#Fig4){ref-type="fig"} where 10 templates were found by the initial spike sorting). For the analysis we simply removed a random selection of templates from the template set while always keeping the template assigned to the corresponding target neuron. Figure [7a](#Fig7){ref-type="fig"} shows the impact of missing templates on classification performance and specificity. While specificity (*i.e.*, the ability of the template matcher to reject spikes from other neurons) strongly decreases with increasing number of missing templates, classification performance (*i.e.*, the number of spikes of the target neuron actually assigned to the correct unit) actually increases. Sensitivity (not shown in Fig. [7](#Fig7){ref-type="fig"}) is not affected by missing templates. The three observations can be easily understood: If a template is missing, the spikes that belong to this template will be either assigned to the closest matching template or will be discarded as noise. If the spikes are assigned to the target template, the specificity will decrease. Since this behavior depends on which template is missing, the variability of the specificity is large (blue band in Fig. [7a](#Fig7){ref-type="fig"}). In contrast, spikes of the target neuron which were wrongly assigned using the full template set might be correctly assigned using a restricted template set if the corresponding template is missing. In the extreme, all templates expect the target template are missing and thus spikes can only be assigned to the target template increasing classification performance. The fact that sensitivity (*i.e.*, the ability to differentiate between spikes and noise) is not affected by missing templates is due to the fact that the detection threshold for a given unit does not depend on the templates of the other units.Fig. 7Performance of BOTM with respect to errors in the template set on benchmark 2 (H). **a** The performance of BOTM as a function of the number of templates used for classification. A random subset of templates, excluding the target template was deleted from the set of all templates generated by the initial spike sorting before the BOTM was run. Colored bands indicate one standard deviation. **b** Influence of noise in the templates on BOTM performance. Left panel: For this analysis a random subset of all available spikes for each neuron was used to compute the template. Right panel: A template is shown on all four electrodes, once for using all available spikes (*blue line*) to compute the template and once for using only two spikes (*red line*). For this analysis the same data set as in Fig. [4](#Fig4){ref-type="fig"} was used. Sensitivity is not shown since it was not affected by the errors on the template set
Another possible problem in the generation of the template set can be noise on the waveforms of the templates. Templates are averages of noisy single spike waveforms, and if not enough single spikes are available to build the template, the resulting waveform may be noisy. We therefore decreased the number of spikes used to compute each template. As Fig. [7b](#Fig7){ref-type="fig"} shows, the number of spikes used to compute a template has only a minor effect on classification performance and specificity as long as at least 30 spikes are averaged. Using less than 30 spikes will cause the resulting template to be noisy. The noise on the template will be similar only to the subset of spikes that were used to compute the template. Thus, template may be very well suited to detect exactly those spikes but will be bad in generalizing to other spike waveforms (with different noise) of the same neuron.
Interestingly, sensitivity is not affected by noise on the template. The spike detection threshold depends on the templates energy, the noise covariance matrix and the prior probability to observe a spike. Noise on the template increases the template's energy while the noise covariance matrix and the spike prior probability remain unchanged. Therefore, the template matching will become more conservative: Noisy templates do not become more sensitive to noise but rather less sensitive to spikes.
Influence of number of templates on BOTM performance {#Sec13}
----------------------------------------------------
How sensitive is BOTM to the number of neurons present in the data? Fig. [8](#Fig8){ref-type="fig"} shows the BOTM performance on benchmark 1 (Q) for the original three templates upon adding more templates to the template set. To make the added templates similar to the original three templates in the benchmark, we created them from the original templates: Each new template was either an original template multiplied by a random amplitude *a*~*i*~ or the sum of two such waveforms. This way we created new templates, which were similar to the original 3 templates. However, to avoid that a new template was too smililar to one of the original ones, *a*~*i*~ was chosen from a bimodal probability density function with peaks at 0.5 and 2 respectively:Fig. 8Performance of BOTM with respect to the number of templates on benchmark 1 (**Q**). To the original template set (**b**), a variable number of newly generated templates was added. An example template set with 16 added templates is shown in **d**. New templates are in cyan, the original ones in magenta. New templates were created in a randomized fashion as a linear superposition of existing templates. The added templates were generated in a way to be similar (*e.g.*, with respect to frequency content, peak position *etc.*) but not identical to any of the original templates. **a** Total performance of BOTM for the original 3 templates as a function of the number of added templates, averaged over all noise levels of each data set and obtained with 5 repetitions of template set generation. The star marks the approximate position of the data point given by the template set shown in **d**. **c** Same as **a** but averaged over all benchmarks for each noise level$$\documentclass[12pt]{minimal}
\usepackage{amsmath}
\usepackage{wasysym}
\usepackage{amsfonts}
\usepackage{amssymb}
\usepackage{amsbsy}
\usepackage{mathrsfs}
\usepackage{upgreek}
\setlength{\oddsidemargin}{-69pt}
\begin{document}$$ {a}_i\sim \Big\{\kern1em \begin{array}{c}N\left(2,0.25\right)\kern.2em \mathrm{with}\kern.3em \mathrm{probability}\kern.3em .2\\ {}\kern2em N\left(.5,0.1\right)\kern.2em \mathrm{with}\kern.3em \mathrm{probability}\kern.3em .8\kern1em \end{array} $$\end{document}$$
As can be seen from Fig. [8](#Fig8){ref-type="fig"}, BOTM performance is negatively affected by adding templates, which reflects the fact that the spike sorting problem becomes more difficult when more templates have to be discriminated. However, the performance does not deteriorate strongly despite the fact, that the templates have similar waveforms. Additionally, the performance saturates for larger number of templates since each additional template is more likely to compete with one of the previously added templates rather than with one of the original three templates.
Influence of noise covariance matrix {#Sec14}
------------------------------------
We compared the performance using different noise covariance matrices, namely the original matrix ***C***, only its diagonal and the diagonally loaded matrix. Performance was decreased on average if only the diagonal of ***C*** was used for BOTM, matched filtering and Mahalanobis distance. This is equivalent to assuming the noise to be uncorrelated. Thus, even though the noise in the benchmarks was not necessarily Gaussian, prewhitening did increase classification performance. In all cases *D*~*i*~^*Maha*^(*t*) and BOTM had the highest classification (and rejection) performance while using *D*~*i*~^*Match*^(*t*) and BOTM had the best sensitivity. We did not try to optimize the computation of the noise covariance matrix (*e.g.*, assuming all electrodes to have a similar auto-covariance like done in (Prentice et al. [@CR36]) could be beneficial for multielectrode arrays) and it might be possible to reduce our error rates by a better choice. However, investigating the effect of ***C*** should be preferentially done on real recordings (and potentially for each recording modality individually).
The lower performance of the methods that ignore noise correlations shows that at least for the benchmark data considered here, modeling noise as correlated by taking the covariance structure into account is beneficial for template matching for both detection and classification. However, it should be noted that this is not necessarily true in general, particularly with data sets having strongly non-Gaussian variability due to, *e.g.*, high level multiunit activity, *etc.*
The threshold derived for BOTM is only optimal if the Gaussian noise assumption is valid, if there is no variation of the neurons' spike shapes, and if there are no changes in its firing rate. To check if these assumptions are violated in a way which would influence the best detection threshold, we also computed the best threshold using the available ground truth in the same way as for the other methods as described in section "[Evaluation metrics](#Sec9){ref-type="sec"}". However, we did not find any significant improvement of performance.
We furthermore investigated, how far the assumption of Gaussian noise is fulfilled by the data. Since Benchmark 1 (Q) is simulated data (and thus the noise is artificial), we used the real data from benchmark 2 (H) to estimate in how far pieces of noise follow a Gaussian statistic. Short pieces of 55 samples length were cut out on all 4 channels (thus forming 220 dimensional vectors) in periods of the recordings when the automatic spike detection did not detect any spikes. We then followed the steps of the noise analysis in (Pouzat et al. [@CR35]): First, we estimated the noise covariance matrix. To avoid over fitting we computed the covariance on one portion of the noise pieces and used it to pre-whiten the other portion. Second, we computed the statistics of third moments for random triplets of the 220 dimensions and tested if they were different from 0. We found, in agreement with (Pouzat et al. [@CR35]), that the distribution of energies of the pre-whitened noise pieces follow a *χ*^2^ -distribution with 219 degrees of freedom, and that the third moment was not significantly different from zero. This indicates that the noise distribution is indeed adequately captured by the noise covariance matrix and the pre-whitened noise pieces follow a multivariate normal distribution.
Resolution of overlapping spikes and iterative refinement of initial spike sorting {#Sec15}
----------------------------------------------------------------------------------
Table [1](#Tab1){ref-type="table"} shows the performance of BOTM on benchmark 3 (Q). We want to emphasize that the direct comparison of the performance of a blind-clustering-based spike sorting with that of a template-matching procedure that uses the correct templates is difficult. Rather, the results show that using an initial spike sorting first to estimate the templates and then BOTM to (re-)detect and classify spikes could strongly decrease the number of errors, especially for overlapping spikes. The decrease in the error rate is achieved by two mechanisms. First (center-left columns in Table [1](#Tab1){ref-type="table"}), spikes are detected using the optimal filter outputs which solves the detection and avoids the alignment problem. The filter output maximizes the SNR yielding a better sensitivity, and its peak is a good feature to align waveforms. Taking its position and template identity to classify the spikes is equivalent to aligning all waveforms on the respective peaks and performing standard template matching like in benchmarks 1 and 2. Second, BOTM is able to resolve overlapping spikes. According to the criterion in (Quiroga [@CR37]), *i.e.*, a maximal time difference of 64 samples, there are 790 overlapping spikes per data set on average in benchmark 3 (Q). The center-left columns in Table [1](#Tab1){ref-type="table"} show that most of those overlaps are correctly resolved by BOTM since there are significantly less errors than overlapping spikes. This is due to the fact that, using *α* = 0.5 for diagonal loading, the peaks in the filter outputs are narrower than the original spikes were and, thus, spike waveforms that were overlapping in the original recording might not lead to overlapping peaks in the filter outputs (the effect of *α*, especially on noise robustness, in a similar context is discussed in (Vollgraf et al. [@CR48])). Those overlaps that were close enough in time to cause simultaneous peaks in the filter outputs could not be resolved by a simple threshold operation on the BOTM filter outputs. Using SIC, however, most spikes were correctly detected and classified (Table [1](#Tab1){ref-type="table"}, center-right columns) significantly reducing the error rate. Therefore, BOTM is a useful method for refining an initial sorting and in the context of online applications where only the beginning of the experiment is sorted offline, *e.g.*, using a clustering procedure. The potential in refining an initial clustering-based spike sorting by BOTM can be seen in the rightmost columns in Table [1](#Tab1){ref-type="table"}. Here, we used the output of the automatic "Wave_clus" sorting (left-most columns) to initialize BOTM. The resulting templates include errors introduced by this spike sorting. For example, in the last data file "Difficult 2, 0.20", "Wave_clus" found only two of the three clusters, thus missing out on one unit. The resulting template matching with only two templates detected most of the spikes of the missing third unit and assigned them mainly to one of the other units. Overall, BOTM was able to reduce the error rates on all data files and led to an increase in the over-all performance from 83.9% to 97.5%.Table 1Performance increase by using BOTM after an initial spike sorting on benchmark 3 (Q)Method:Wave_clusCorrect templates + BOTMCorrect templates + BOTM + SICWave_clus + BOTM + SICError type:DetClassTotalDetClassTotalDetClassTotalDetClassTotalNameNoiseEasy 10.05921 (73.8)1 (100.0)922 (86.9)76 (97.8)2 (99.9)78 (98.9)13 (99.6)4 (99.9)17 (99.8)11 (99.7)2 (99.9)13 (99.8)0.10236 (93.3)5 (99.9)241 (96.6)45 (98.7)3 (99.9)48 (99.3)5 (99.9)2 (99.9)7 (99.9)4 (99.9)2 (99.9)6 (99.9)0.15374 (89.2)5 (99.9)379 (94.5)59 (98.3)5 (99.9)64 (99.1)5 (99.9)0 (100.0)5 (99.9)8 (99.8)0 (100.0)8 (99.9)0.20999 (71.2)12 (99.7)1011 (85.4)61 (98.2)9 (99.7)70 (99.0)9 (99.7)3 (99.9)12 (99.8)9 (99.7)3 (99.9)12 (99.8)Easy 20.05174 (94.9)3 (99.9)177 (97.4)51 (98.5)19 (99.4)70 (99.0)2 (99.9)1 (100.0)3 (100.0)2 (99.9)2 (99.9)4 (99.9)0.10193 (94.5)10 (99.7)203 (97.1)39 (98.9)21 (99.4)60 (99.1)7 (99.8)1 (100.0)8 (99.9)6 (99.8)2 (99.9)8 (99.9)0.15184 (94.6)45 (98.7)229 (96.6)43 (98.7)36 (98.9)79 (98.8)4 (99.9)2 (99.9)6 (99.9)4 (99.9)6 (99.8)10 (99.9)0.20637 (81.9)306 (91.3)943 (86.6)85 (97.6)93 (97.4)178 (97.5)6 (99.8)2 (99.9)8 (99.9)6 (99.8)7 (99.8)13 (99.8)Difficult 10.05274 (91.9)0 (100.0)274 (96.0)37 (98.9)17 (99.5)54 (99.2)3 (99.9)17 (99.5)20 (99.7)2 (99.9)18 (99.5)20 (99.7)0.10201 (94.2)41 (98.8)242 (96.5)39 (98.9)50 (98.5)89 (98.7)18 (99.5)11 (99.7)29 (99.6)18 (99.5)12 (99.7)30 (99.6)0.15217 (93.8)81 (97.7)298 (95.7)46 (98.7)132 (96.2)178 (97.4)11 (99.7)17 (99.5)28 (99.6)9 (99.7)17 (99.5)26 (99.6)0.20405 (88.1)651 (80.9)1056 (84.5)45 (98.7)278 (91.9)323 (95.3)19 (99.4)10 (99.7)29 (99.6)20 (99.4)10 (99.7)30 (99.6)Difficult 20.05183 (94.6)1 (100.0)184 (97.3)45 (98.7)34 (99.0)79 (98.8)9 (99.7)8 (99.8)17 (99.7)8 (99.8)9 (99.7)17 (99.7)0.10157 (95.5)8 (99.8)165 (97.6)33 (99.0)38 (98.9)71 (99.0)4 (99.9)7 (99.8)11 (99.8)5 (99.9)9 (99.7)14 (99.8)0.15193 (94.4)443 (87.1)636 (90.8)51 (98.5)153 (95.6)204 (97.0)5 (99.9)8 (99.8)13 (99.8)8 (99.8)9 (99.7)17 (99.8)0.20492 (85.9)1462 (58.1)1954 (72.0)104 (97.0)386 (88.9)490 (93.0)7 (99.8)14 (99.6)21 (99.7)36 (99.0)1124 (67.8)1160 (83.4)Total5840 (89.4)3074 (94.4)8914 (83.9)859 (98.4)1276 (97.7)2135 (96.1)127 (99.8)107 (99.8)234 (99.6)156 (99.7)1232 (97.8)1388 (97.5)Numbers indicate absolute numbers of respective errors, numbers in brackets indicate performance as defined for benchmark 1 (Q) in percent, rounded to one decimal digit. Spike sorting errors for "Wave_clus" as reported in reported in (Quiroga et al. [@CR38]) (leftmost columns), the proposed methods BOTM (center-left columns), BOTM with subtractive overlap resolution (SIC) (center-right columns, and BOTM + SIC on using the result of "Wave_clus" as an initialization (rightmost columns), are given. "Det" are detection errors, "Class" classification errors and "Total" their sum. Values in column "Noise" indicate the standard deviation of the noise relative to the peak of the templates. For "Wave_clus" the errors were estimated independently for detection and classification. Classification errors were computed using all spikes in the data. Thus, spikes that were not detected were still used for the classification task. For BOTM the spike detection and classification were done in the same step without using the ground truth to align the spikes. Spikes that were not detected were also not classified. On the dataset "Difficult 2, 0.20" "Wave_clus"found only 2 of the three units. This is reflected in higher error rates for "Wave_clus + BOTM + SIC" since one template was missing for the template matching
Discussion {#Sec16}
==========
We analyzed the performance of different template matching procedures. We showed that distance based template matching is not suitable to detect spikes while the performance of convolution-based template matching for classifying spikes may be low depending on the templates in the data. For the case of colored Gaussian noise we use a Bayesian approach to derive an optimal template matching. The proposed BOTM algorithm we show to outperform the other methods on a number of benchmark data sets. The probabilistic framework provides a robust way to resolve overlapping spikes, even in the presence of a relatively large number of templates (see Fig. [4](#Fig4){ref-type="fig"} where 10 templates are present). Since the BOTM procedure is fast and computationally simple it is also suitable for hardware implementation and potential real-time applications. BOTM can in principle be applied to all detection and classification problems which include linear separation of multiple transient signals. Furthermore, there is no need to align the spike waveforms before classification, a step that is usually error-prone.
The probabilistic approach and the solution derived here is related to earlier work (Pillow et al. [@CR33]) but exhibits some important differences. Our approach shows the connection between LDA and matched filtering, and can be implemented by linear filters. This can be an important advantage for hardware implementations and closed-loop experiments, since our method is online-capable. This allows for applying adaptive template matching strategies, where the templates and filters are gradually adapted over time depending on the previously found spikes. We did not attempt using an adaptive strategy in this study (see [discussion](#Sec16){ref-type="sec"} below). The method presented in (Pillow et al. [@CR33]) solves a global optimization problem, while we solve the overlap problem locally in time. Our approach could entail a slight increase in sorting errors in those cases where future spike classifications are important to make the correct decision (resolving an overlapping spike from the future backwards rather than from the past onwards), but it offers the advantage of having an online algorithm and faster classification. Furthermore, the locality reduces the dimensionality of the noise covariance matrix in the range of the template length, not the length of the recordings, which makes it feasible for us to estimate the noise covariance matrix on short periods of noise. The weights of the pros and cons of the methods largely depend, however, on the experimental context.
Non-stationary templates {#Sec17}
------------------------
Spike waveforms from neurons are known to vary on two different time scales: In the range of milliseconds (Fee et al. [@CR11]), depending on the time between to spikes of the same neuron, and on a larger time scale owing to a movement of the neurons with respect to the electrodes (Franke et al. [@CR14]). Both sources of variability, but prevailingly the movement-induced variability, can be addressed by using adaptive filters. The benchmarks used in this study are, due to a lack of strong waveform variability, however, not well suited to test an adaptive approach, so that further work will be necessary to investigate the potential of adaptive strategies.
Noise covariance matrix and gaussian noise assumption {#Sec18}
-----------------------------------------------------
BOTM assumes noise to be multivariate colored Gaussian. This was found to be a good description of real noise (Pouzat et al. [@CR35]), but, other studies claim that the distributions of spike waveforms are better explained by t-distributions (Shoham et al. [@CR45]). Neither of the data sets used in this study were constructed to follow this assumption. In fact, the noise in the benchmark data set from (Quiroga et al. [@CR38]) was created by copying many templates with small amplitudes into the data, but it should be also noted that this dataset did not include multiunit activity, which is one of the main factors introducing deviations from Gaussian distributions. The data set from (Henze et al. [@CR20]) are real recordings where noise is likely to contain small amplitude spikes from neurons that are further away from the electrodes but our analysis showed that the Gaussian assumption might be well justified. For both data sets using the colored Gaussian noise assumption significantly increased spike detection and classification performance. This is consistent with the observation that although in many classification problems the assumption of normality and of a common covariance matrix among clusters is often violated, linear classifiers assuming colored Gaussian noise still perform surprisingly well (Duda et al. [@CR8]; Li et al. [@CR28]).
Should the main variability of clusters be caused by the neurons, *e.g.,* while bursting, and not by noise, the decision boundaries derived *via* the noise covariance matrix (and not the cluster covariance matrix) might be suboptimal. This could be remedied by using multiple templates per neuron, an approach especially promising for bursting neurons which can produce several distinct, sometimes even non-overlapping clusters.
A question that remains open is in how far the noise statistics (and thus the noise covariance matrix) are stable during an experiment and how well they can be estimated, *e.g.*, for artifacts and ripples (Fig. [1](#Fig1){ref-type="fig"}). This will depend strongly on the recording conditions and might vary from setup to setup. However, similar problems were already faced in radar applications (Melvin [@CR30]) and it might be beneficial to determine if those solutions are applicable also to the analysis of extracellular recordings.
Resolution of overlapping spikes {#Sec19}
--------------------------------
Several approaches were recently developed to resolve overlapping spikes (Atiya [@CR2]; Ekanadham et al. [@CR10]; Franke et al. [@CR14]; Lewicki [@CR26]; Marre et al. [@CR29]; Pillow et al. [@CR33]; Prentice et al. [@CR36]; Segev et al. [@CR44]; Vargas-Irwin and Donoghue [@CR47]; Wang et al. [@CR49]; Zhang et al. [@CR50]). Most of them are based either on a greedy iterative subtraction scheme to remove spikes and detect overlaps (*e.g.*, (Marre et al. [@CR29]; Segev et al. [@CR44])) or on searching the best fit in the space of all possible overlaps (*e.g.*, (Pillow et al. [@CR33]; Prentice et al. [@CR36])). Ekanadham and co-authors ([@CR10]) suggest to resolve overlapping spikes while performing clustering. The method we propose here is similar to several of the abovementioned approaches: the brute-force (Option 1 in Fig. [2](#Fig2){ref-type="fig"}) and the iterative subtraction scheme (Option 2, SIC). In contrast to the other methods, however, BOTM performs the overlap resolution using the filter outputs of the matched filters instead of the original recorded data. This increases the discriminability of spike waveforms from different neurons and reduces the influence of noise. Although our method is computationally less expensive it still yields comparable results to (Ekanadham et al. [@CR10]) on Benchmark 3.
Appendix {#Sec20}
========
Here we give the derivation of the BOTM procedure. The derivation shows that applying linear discriminant analysis (LDA) to each sample of the data is equivalent to computing the matched filter output for each template and adding a template-dependent constant. First, we review the matched filter (Choi [@CR7]; Kay [@CR23]; Van Trees 2002). Then, we introduce LDA (Fisher [@CR12]) from the perspective of a classification problem. We continue by recognizing that if the vectors, classified by LDA, are thought to be short pieces of data beginning at each possible temporal position, the linear discriminant functions for each class (*i.e.*, the BOTM filter outputs for each template) can elegantly be computed from the matched filter outputs. From the probabilistic generative model underlying LDA we can also derive the theoretically optimal threshold for spike detection in the BOTM outputs. Finally, we note that to detect and classify an overlapping spike, the discriminant function for the overlap can be computed from the individual discriminant functions by summation with a temporal shift. This offers two possibilities to resolve overlapping spikes, namely the computation of all discriminant functions for overlapping spikes with different shifts or a greedy subtractive procedure.
Matched filtering {#Sec21}
=================
If the waveform *ξ* of a transient signal embedded in a noisy recording is known, the FIR filter which maximizes the filter response $\documentclass[12pt]{minimal}
\usepackage{amsmath}
\usepackage{wasysym}
\usepackage{amsfonts}
\usepackage{amssymb}
\usepackage{amsbsy}
\usepackage{mathrsfs}
\usepackage{upgreek}
\setlength{\oddsidemargin}{-69pt}
\begin{document}$$ y(t)={\displaystyle \sum_i{f}_i*}x(t) $$\end{document}$ to *ξ* while minimizing the response to noise is called the matched filter. The response of the filter to the template is given by *y* = *f*^*T*^*ξ* while its expected response to Gaussian noise *η* ∼ *N*(0, *C*) is given by *y*(*t*) = *E*\[*f*^*T*^*η*\]. Thus, the signal-to-noise ratio can be defined as$$\documentclass[12pt]{minimal}
\usepackage{amsmath}
\usepackage{wasysym}
\usepackage{amsfonts}
\usepackage{amssymb}
\usepackage{amsbsy}
\usepackage{mathrsfs}
\usepackage{upgreek}
\setlength{\oddsidemargin}{-69pt}
\begin{document}$$ SNR=\frac{{\left|{f}^T\xi \right|}^2}{E\left[{\left|{f}^T\eta \right|}^2\right]} $$\end{document}$$
The filter that maximizes the SNR with is then given by (for a derivation see, *e.g.*, (Van Trees [@CR46]) and in the context of spike sorting (Franke [@CR13]))$$\documentclass[12pt]{minimal}
\usepackage{amsmath}
\usepackage{wasysym}
\usepackage{amsfonts}
\usepackage{amssymb}
\usepackage{amsbsy}
\usepackage{mathrsfs}
\usepackage{upgreek}
\setlength{\oddsidemargin}{-69pt}
\begin{document}$$ f=\frac{C^{-1}\xi }{\beta } $$\end{document}$$
where *β* is a normalization constant.
Linear discriminant analysis {#Sec22}
============================
Consider a multi-class classification problem and a new unlabeled data point *x*. To which class should we assign *x*? If we want to minimize the classification error, we will assign *x* to the class with maximal conditional probability *p*(*i*\|*x*), *i.e.*, assign *x* to class *i*~*opt*~ if$$\documentclass[12pt]{minimal}
\usepackage{amsmath}
\usepackage{wasysym}
\usepackage{amsfonts}
\usepackage{amssymb}
\usepackage{amsbsy}
\usepackage{mathrsfs}
\usepackage{upgreek}
\setlength{\oddsidemargin}{-69pt}
\begin{document}$$ {i}_{opt}=\begin{array}{c}\hfill \arg\;\max\;p\left(i\left|x\right.\right)\hfill \\ {}\hfill i\hfill \end{array} $$\end{document}$$
Using Bayes rule one obtains:$$\documentclass[12pt]{minimal}
\usepackage{amsmath}
\usepackage{wasysym}
\usepackage{amsfonts}
\usepackage{amssymb}
\usepackage{amsbsy}
\usepackage{mathrsfs}
\usepackage{upgreek}
\setlength{\oddsidemargin}{-69pt}
\begin{document}$$ {i}_{opt}=\begin{array}{c}\hfill \arg\;\max \hfill \\ {}\hfill i\hfill \end{array}\frac{p\left(x\left|i\right.\right)p(i)}{{\displaystyle {\sum}_kp\left(x\left|k\right.\right)p(k)}} $$\end{document}$$
where *p*(*x*\|*i*) is the data likelihood given class *i*. In the context of the classification problem, we are not interested in the exact probabilities but only in the relations between them. We are searching for a discriminant function $\documentclass[12pt]{minimal}
\usepackage{amsmath}
\usepackage{wasysym}
\usepackage{amsfonts}
\usepackage{amssymb}
\usepackage{amsbsy}
\usepackage{mathrsfs}
\usepackage{upgreek}
\setlength{\oddsidemargin}{-69pt}
\begin{document}$$ \widehat{d} $$\end{document}$ that will give us the same classification function as (eq. [1](#Equ1){ref-type=""})$$\documentclass[12pt]{minimal}
\usepackage{amsmath}
\usepackage{wasysym}
\usepackage{amsfonts}
\usepackage{amssymb}
\usepackage{amsbsy}
\usepackage{mathrsfs}
\usepackage{upgreek}
\setlength{\oddsidemargin}{-69pt}
\begin{document}$$ {\widehat{d}}_i(x)>{\widehat{d}}_j(x)\iff p\left(i\left|x\right.\right)>p\left(j\left|x\right.\right) $$\end{document}$$
but omits all unnecessary computations; *e.g.,* the denominator in ([2](#Equ2){ref-type=""}) appears on both sides of the inequality and can thus be ignored. We obtain$$\documentclass[12pt]{minimal}
\usepackage{amsmath}
\usepackage{wasysym}
\usepackage{amsfonts}
\usepackage{amssymb}
\usepackage{amsbsy}
\usepackage{mathrsfs}
\usepackage{upgreek}
\setlength{\oddsidemargin}{-69pt}
\begin{document}$$ {\widehat{d}}_i(x)=p\left(x\left|i\right.\right)p(i) $$\end{document}$$
If *p*(*x*\|*i*) is a multivariate normal distribution $\documentclass[12pt]{minimal}
\usepackage{amsmath}
\usepackage{wasysym}
\usepackage{amsfonts}
\usepackage{amssymb}
\usepackage{amsbsy}
\usepackage{mathrsfs}
\usepackage{upgreek}
\setlength{\oddsidemargin}{-69pt}
\begin{document}$$ p\left(x\Big|i\right)=\frac{1}{{\left(2\pi \right)}^{\frac{k}{2}}{\left|{\boldsymbol{C}}_i\right|}^{\frac{1}{2}}}{e}^{-\frac{1}{2}{\left(x-{\xi}_i\right)}^T{{\boldsymbol{C}}_i}^{-1}\left(x-{\xi}_i\right)} $$\end{document}$ with mean *ξ*~*i*~ and covariance matrix *C*~*i*~, we obtain$$\documentclass[12pt]{minimal}
\usepackage{amsmath}
\usepackage{wasysym}
\usepackage{amsfonts}
\usepackage{amssymb}
\usepackage{amsbsy}
\usepackage{mathrsfs}
\usepackage{upgreek}
\setlength{\oddsidemargin}{-69pt}
\begin{document}$$ {\widehat{d}}_i(x)=\frac{1}{{\left|{\boldsymbol{C}}_i\right|}^{\frac{1}{2}}}{e}^{-\frac{1}{2}{\left(x-{\xi}_i\right)}^T{{\boldsymbol{C}}_i}^{-1}\left(x-{\xi}_i\right)}p(i) $$\end{document}$$
where *k* is the dimension of the data *x* and the term $\documentclass[12pt]{minimal}
\usepackage{amsmath}
\usepackage{wasysym}
\usepackage{amsfonts}
\usepackage{amssymb}
\usepackage{amsbsy}
\usepackage{mathrsfs}
\usepackage{upgreek}
\setlength{\oddsidemargin}{-69pt}
\begin{document}$$ {\left(2\pi \right)}^{\frac{k}{2}} $$\end{document}$ has been dropped because it is the same for all classes. Note that in the general case each class can have its own covariance matrix *C*~*i*~.
Taking the logarithm we formally obtain, the discriminant function *d*$$\documentclass[12pt]{minimal}
\usepackage{amsmath}
\usepackage{wasysym}
\usepackage{amsfonts}
\usepackage{amssymb}
\usepackage{amsbsy}
\usepackage{mathrsfs}
\usepackage{upgreek}
\setlength{\oddsidemargin}{-69pt}
\begin{document}$$ {d}_i(x)= \ln {\widehat{d}}_i(x)=-\frac{1}{2} \ln \left(\left|{\boldsymbol{C}}_i\right|\right)-\frac{1}{2}{\left(x-{\xi}_i\right)}^T{{\boldsymbol{C}}_i}^{-1}\left(x-{\xi}_i\right)+ ln\left(p(i)\right) $$\end{document}$$
which is called Fisher's quadratic discriminant function. If we can constrain all classes to share the same covariance matrix *C* = *C*~*i*~, ∀ *i* (homoscedasticity) then the term $\documentclass[12pt]{minimal}
\usepackage{amsmath}
\usepackage{wasysym}
\usepackage{amsfonts}
\usepackage{amssymb}
\usepackage{amsbsy}
\usepackage{mathrsfs}
\usepackage{upgreek}
\setlength{\oddsidemargin}{-69pt}
\begin{document}$$ -\frac{1}{2} \ln \left(\left|{C}_i\right|\right) $$\end{document}$ is again shared by all discriminant functions and can be dropped. Expanding the squared form then yields$$\documentclass[12pt]{minimal}
\usepackage{amsmath}
\usepackage{wasysym}
\usepackage{amsfonts}
\usepackage{amssymb}
\usepackage{amsbsy}
\usepackage{mathrsfs}
\usepackage{upgreek}
\setlength{\oddsidemargin}{-69pt}
\begin{document}$$ {d}_i(x)={x}^T{\boldsymbol{C}}^{-1}{\xi}_i-\frac{1}{2}{\xi_i}^T{\boldsymbol{C}}^{-1}{\xi}_i+ ln\left(p(i)\right) $$\end{document}$$
which is the function used in linear discriminant analysis (LDA) (Fisher [@CR12]). The result is similar to the M-ary detection case in (Choi [@CR7]), chapter 2.5.
In the context of spike sorting the classes represent different neurons and all vectors are assumed to be perfectly aligned spike waveforms. Then the assumption that all clusters have the same covariance matrix is partly justified: noise in the spike waveforms can be modeled as being independent of the identity of the spike.
Derivation of Bayes optimal template matching {#Sec23}
=============================================
We now combine the LDA results with the matched filtering procedure. According to the model assumptions, spike templates are centers of multivariate normal distributions with a shared (noise) covariance matrix. If we now interpret each temporal position *t* in a continuous recording *X*(*t*) as a data point and thus as a potential spike, we need to compute the discriminant function of the full continuous recording to all templates.
Linear discriminant functions *d* can be calculated by a filtering the continuous recording with the matched filter *f*~*i*~ = *C*^− 1^*ξ*~*i*~ (of eq. ([6](#Equ6){ref-type=""}))$$\documentclass[12pt]{minimal}
\usepackage{amsmath}
\usepackage{wasysym}
\usepackage{amsfonts}
\usepackage{amssymb}
\usepackage{amsbsy}
\usepackage{mathrsfs}
\usepackage{upgreek}
\setlength{\oddsidemargin}{-69pt}
\begin{document}$$ {d}_i(t)=X{(t)}^T{f}_i-\frac{1}{2}{\xi}_{{}_i}^T{C}^{-1}{\xi}_i+ ln\left(p(i)\right) $$\end{document}$$
where *X*(*t*) is a vector of concatenated multi-electrode data starting at time *t*. The term $\documentclass[12pt]{minimal}
\usepackage{amsmath}
\usepackage{wasysym}
\usepackage{amsfonts}
\usepackage{amssymb}
\usepackage{amsbsy}
\usepackage{mathrsfs}
\usepackage{upgreek}
\setlength{\oddsidemargin}{-69pt}
\begin{document}$$ {c}_i=-\frac{1}{2}{\xi}_i^T{C}^{-1}{\xi}_i ln\left(p(i)\right) $$\end{document}$ does not depend on *X*(*t*), hence$$\documentclass[12pt]{minimal}
\usepackage{amsmath}
\usepackage{wasysym}
\usepackage{amsfonts}
\usepackage{amssymb}
\usepackage{amsbsy}
\usepackage{mathrsfs}
\usepackage{upgreek}
\setlength{\oddsidemargin}{-69pt}
\begin{document}$$ {d}_i(t)=X{(t)}^T{f}_i+{c}_i $$\end{document}$$
The output of the matched filters (apart from an additive constant) can thus directly be interpreted as the Fisher discriminant function.
Discriminant function for noise {#Sec24}
===============================
Obviously, not every sample of the data corresponds to a spike. The problem to decide whether a given sample is noise instead of a spike can be solved by introducing a "template" (and thus a discriminant function) for noise. For zero mean noise we introduce the null vector 0 as the noise template n and obtain the corresponding discriminant function$$\documentclass[12pt]{minimal}
\usepackage{amsmath}
\usepackage{wasysym}
\usepackage{amsfonts}
\usepackage{amssymb}
\usepackage{amsbsy}
\usepackage{mathrsfs}
\usepackage{upgreek}
\setlength{\oddsidemargin}{-69pt}
\begin{document}$$ {d}_n(t)\Big\rangle X{(t)}^T{C}^{-1}0-\frac{1}{2}0{C}^{-1}0+ ln\left(p(n)\right) $$\end{document}$$
where *p*(*n*) is the prior probability for a given sample of data to be noise (potentially including spikes from neurons for which no template is given, *e.g.*, multi-unit activity). With $\documentclass[12pt]{minimal}
\usepackage{amsmath}
\usepackage{wasysym}
\usepackage{amsfonts}
\usepackage{amssymb}
\usepackage{amsbsy}
\usepackage{mathrsfs}
\usepackage{upgreek}
\setlength{\oddsidemargin}{-69pt}
\begin{document}$$ p(n)=1-{\displaystyle \sum_ip(i)} $$\end{document}$, we obtain$$\documentclass[12pt]{minimal}
\usepackage{amsmath}
\usepackage{wasysym}
\usepackage{amsfonts}
\usepackage{amssymb}
\usepackage{amsbsy}
\usepackage{mathrsfs}
\usepackage{upgreek}
\setlength{\oddsidemargin}{-69pt}
\begin{document}$$ {d}_n(t)\Big\rangle thr={d}_n= \ln \left(1-{\displaystyle \sum_ip(i)}\right) $$\end{document}$$
Equation ([9](#Equ9){ref-type=""}) provides a natural threshold for spike detection which will usually be very close to 0.
Discriminant function for overlapping spikes {#Sec25}
============================================
The discriminant function for an overlap between spikes from neuron *i* and *j* with time shift *τ* can be defined analogously by introducing a new template *d*~*i* + *j*~^*τ*^(*t*) = *ξ*~*i*~ + *ξ*~*j*~^*τ*^ for the overlap.
*ξ*~*j*~^*τ*^ denotes the template of neuron *j* shifted by time *τ*. We obtain$$\documentclass[12pt]{minimal}
\usepackage{amsmath}
\usepackage{wasysym}
\usepackage{amsfonts}
\usepackage{amssymb}
\usepackage{amsbsy}
\usepackage{mathrsfs}
\usepackage{upgreek}
\setlength{\oddsidemargin}{-69pt}
\begin{document}$$ {d}_{i+j}^{\tau }(t)=X{(t)}^T{C}^{-1}\left({\xi}_i+{\xi}_j^{\tau}\right)-\frac{1}{2}{\left({\xi}_i+{\xi}_j^{\tau}\right)}^T{C}^{-1}\left({\xi}_i+{\xi}_j^{\tau}\right)+ ln\left(p(i)p(j)\right) $$\end{document}$$
Equation ([10](#Equ10){ref-type=""}) can be rearranged to$$\documentclass[12pt]{minimal}
\usepackage{amsmath}
\usepackage{wasysym}
\usepackage{amsfonts}
\usepackage{amssymb}
\usepackage{amsbsy}
\usepackage{mathrsfs}
\usepackage{upgreek}
\setlength{\oddsidemargin}{-69pt}
\begin{document}$$ {d}_{i+j}^{\tau }(t)={d}_i(t)+{d}_j^{\tau }(t)-{\xi_i}^T{\boldsymbol{C}}^{-1}{\xi}_j^{\tau } $$\end{document}$$
where *d*~*j*~^*τ*^ is a time shifted version of *d*~*j*~. Equation ([11](#Equ11){ref-type=""}) allows computing the overlap discriminant function from the individual spike discriminant functions without actually constructing the overlap template or computing its filter. This method can be extended to overlaps with more than 2 participating spikes.
Subtractive interference cancellation {#Sec26}
=====================================
Since the computation of the overlap discriminant functions can still be very expensive, we can save computation time with a greedy subtraction procedure: Once a single spike is detected in the single spike discriminant functions, we can use eq. ([11](#Equ11){ref-type=""}) to update the other single spike discriminant functions to reflect the presence of the already found spike. This is done by subtracting the expected filter response of each filter to the detected spike from the respective discriminant functions, an approach also referred to as subtractive interference cancellation (SIC) (Moshavi [@CR31]). For this subtraction the precise position of the waveform in the data is needed, otherwise, a shifted version might be subtracted, causing a residual error. The original sampling rate of the data might not be fine enough; therefore, it can be advantageous to upsample the data before subtraction.
To update the discriminant function for template *i* if a spike of template *j* was found at time *t*~0~ we set$$\documentclass[12pt]{minimal}
\usepackage{amsmath}
\usepackage{wasysym}
\usepackage{amsfonts}
\usepackage{amssymb}
\usepackage{amsbsy}
\usepackage{mathrsfs}
\usepackage{upgreek}
\setlength{\oddsidemargin}{-69pt}
\begin{document}$$ {d}_i\left({t}_0-\tau \right)\leftarrow {d}_i\left({t}_0-\tau \right)-{\xi}_i^T{C}^{-1}{\xi}_i^T+ ln\left(p(j)\right) $$\end{document}$$
which follows directly from eq.([11](#Equ11){ref-type=""}). It changes the value of *d*~*i*~(*t*) to the correct value of *d*~*i* + *j*~^*τ*^(*t*) after subtraction of *ξ*~*j*~ from the data at time *t*~0~. This is the update used in this study, but a possible alternative heuristic would be$$\documentclass[12pt]{minimal}
\usepackage{amsmath}
\usepackage{wasysym}
\usepackage{amsfonts}
\usepackage{amssymb}
\usepackage{amsbsy}
\usepackage{mathrsfs}
\usepackage{upgreek}
\setlength{\oddsidemargin}{-69pt}
\begin{document}$$ {d}_i\left({t}_0-\tau \right)={d}_i\left({t}_0-\tau \right)-{\xi_i}^T{\boldsymbol{C}}^{-1}{\xi}_j^{\tau } $$\end{document}$$
by ignoring the prior probability *p*(*j*) for spike *j*. This approach effectively removes the influence of spike *j* and repeats the spike detection process without taking into account that a spike was already found, which is identical to subtracting the template from the raw data and re-computing the complete template matching.
Note that this overestimates the performance of the other methods, since in a real situation their optimal threshold is not known.
We thank Robin Ince, Robert Pröpper, Philipp Meier and Christophe Pouzat for helpful discussions and Wendelin Böhmer and Espen Hagen for feedback on a draft of this paper. This work was supported by DFG GRK 1589/1, the German Federal Ministry of Education and Research (BMBF) with the grants 01GQ0743 and 01GQ0410, by the European Community through the ERC Advanced Grant 267351, "NeuroCMOS", and by the Engineering and Physical Sciences Research Council (EPSRC).
Conflict of interest {#FPar1}
====================
The authors declare that they have no conflict of interest
[^1]: Action Editor: Lance M. Optican
| {
"pile_set_name": "PubMed Central"
} |
Introduction {#S1}
============
A feature of almost all cancers is their ability to escape from the immune system. This process is called immuno-editing and is composed of three different steps: elimination, equilibrium, and escape. During the elimination phase, the immune system recognizes the antigens expressed by tumor cells and eliminates them. If any cells are able to escape from the process of eradication they pass into the second phase in which they modify their antigens to render them unrecognizable by the immune system. In this stage, tumor cells start growing until the mass reaches a considerable volume. This is the third phase of escape, in which the immune system loses control of the tumor which can then spread and become detectable and thus clinically relevant ([@B1]). In this landscape, conventional cancer therapies show some limitations. The inability of chemotherapy and radiotherapy to selectively target cancer cells leads to a very high toxicity. Also, the development of chemo-resistance leaves then surgery as the last chance, if available. Another important aspect not to be underestimated, is the lack of conventional therapies able to create long-lasting immunity preventing metastasis and the relapse of cancer.
Cancer immuno-therapy is a promising new strategy to fight cancer and it consists of the activation and arming of the immune system against tumors. There are many different approaches among which, oncolytic virus therapy (OVT) is one of the most encouraging.
As stated in the name, OVT takes advantage of the oncolytic nature of some viruses \[oncolytic viruses (OVs)\] in order to kill tumor cells. The advantage of these viruses is their ability to infect and replicate in tumor cells without harming normal tissues.
Tumor cells are indeed a good target for OVs. They show a reduction in many of the specific mechanisms used by host cells to respond to viral infection (such as the type I IFN pathway) allowing viruses to replicate successfully in these cells ([@B2]). Moreover, advances in genetic engineering have led to the production of viruses lacking the thymidine kinase gene forcing the virus to replicate only in those cells that have an up-regulation of the RAS pathway like cancer cells ([@B3]).
Oncolytic therapy is not just a dream. Several viruses have already reached the clinical stages. The best example is given by Talimogene Laherparepvec, known as T-Vec. This is a modified herpex simplex virus (HSV) that has two viral gene deletions and is armed with the human GM-CSF gene. T-Vec has been shown---in a phase II study---to increase the number of tumor-specific CD8^+^ T cells and to reduce the number of regulatory and suppressor T cells ([@B4]). Moreover, T-Vec has been tested in a phase III trial in patients with melanoma ([@B5], [@B6]) resulting, in 2015, in the FDA approval for the treatment of melanoma patients with injectable but non-resectable lesions in the skin and lymph nodes ([@B6]). Pushed from the good results obtained with T-Vec, over the last few years a variety of OVs have been tested in clinical trials. Safety profiles have reached an excellent standard through modification of OVs to increase specificity and reduce side effects. Despite these promising results, anti-tumor efficacy is still limited (especially when the viruses are used alone) and as a result, new strategies are needed for further improvement of OVT.
In this review, we would like to highlight the promising therapeutic effect of OVT mostly focusing on the ability of OVT to activate the immune system, and how to further improve the anti-tumor efficacy of current OVT by modulating the host immune responses to the viruses and tumor cells.
Anti-Tumor Effects by OVT {#S2}
=========================
The inhibition of the IFN pathway, the major anti-viral response of the cells, is frequently disfunctioned in cancer cells. As a result, OVs can easily infect the transformed cells and fulfill their function. However, the purpose of oncolytic viral therapy is not only to kill cancer cells but also to activate the immune system, silenced by the tumor microenvironment. In order to achieve this, OVT can act in different ways. OVs are able to create long-lasting memory. As discussed above, transformed cells have the ability to escape from the immune system by mutating their antigens and becoming invisible to leukocytes in a process called immuno-editing. When OVs infect tumor cells, an inflammatory reaction is triggered. This is due to the fact that viruses are able to induce immunogenic cell death (ICD). This process is a particular form of apoptosis in which the death of cancer cells is able to induce an effective anti-tumor response *via* the recruitment and activation of dendritic cells (DCs) and the consequent stimulation of specific T lymphocytes. In ICD, the process of apoptosis is not "sterile" but it triggers the endoplasmic reticulum with the consequent release of some dangerous metabolites called damage-associated molecular patterns (DAMPs). ICD is characterized by the release of three particular molecules that can be classed as DAMPs: calreticulin, ATP, and HMGB1 ([@B7]). APCs in the tumor microenvironment recognize these key metabolites and they are able to generate an immune response. Moreover, when OVs infect and destroy cancer cells, tumor associated or/and specific antigens are released into the microenvironment allowing the immune system to recognize them and to generate a response, breaking down the immuno-editing process. As reported by Breitbach et al., this local stimulation of the immune system is able to create a systemic and long-lasting anti-cancer response from immune cells, which also occurs in advanced stage patients ([@B8]). In many tumors, the tumor microenvironment is "a cold place" in which the process of immuno-editing has created an immuno-suppressive environment. As reported by Bell's group, tumor cell infection with OV creates an inflammatory site with the consequent release of cytokines that activate the immune system, making the "cold" tumor "warm" ([@B8]). In this way, the primary immune response which can be seen as a negative response triggered against OVs (the activation of the immune system against the virus itself) can create anti-tumor immunological memory with very long-term benefits to protect the host against relapse (Figure [1](#F1){ref-type="fig"}).
{#F1}
The anti-tumor responses that OVTs can generate are very promising and give hope for new anti-cancer therapies. However, the true potential of OVs cannot be realized until some natural barriers (intrinsic of many tumors) are overcome. Among these, the hardest problems to solve is the large size of the tumor that can deny OVs access to the tumor core along with other physical barriers such as the elevated interstitial pressure and, in case of intravenous delivery, the poor tumor vasculature.
Anti-Viral Effects by OVT {#S3}
=========================
The battle between anti-tumor and anti-viral immunity exists because triggering the immune system will clear virus and along with it the lytic effect of viral infection, however, anti-tumor immunity will then also be diminished. We need to discover whether or not to focus on engineering a virus that very efficiently kills tumor cells or a virus that stimulates the most robust anti-tumor response.
However, OV infection is the reason why anti-tumor immunity is generated. Upon targeted infection of tumor cells, the immune system detects viral infection and as well as innate immunity, adaptive responses are also triggered. These involve T cells which are primed to lyse cells containing foreign antigens (which usually derive from virus but can also be derived from the cancerous cell itself). This ultimately leads to targeted lysis of not only infected cells but also, the tumor cells themselves. It is therefore important to design OVs that can replicate and spread within tumors quickly to induce the maximal anti-tumor effect before viral clearance ([@B9]).
In light of the fact that anti-viral responses may dampen the effect of OVs by clearing virus prematurely, manipulating the anti-viral immune response by blocking antibody responses to the virus means that the virus has extra time to take effect and kill tumor cells through anti-tumoral T-cell responses generated by the presentation of tumor antigens by infected cells ([@B10]). Recently, we demonstrated an innovative way to bypass anti-viral immunity, showing that repeated local administration of adenovirus can improve the efficacy of anti-cancer therapy in a Syrian hamster model ([@B9]). Li et al. demonstrate that if the virus is injected intratumorally, humoral immunity has no effect on the clearance of the virus which suggests innate immunity and cytotoxic T cells plays a central role. Moreover, they demonstrate that pre-existing immunity against the virus does not affect therapeutic efficacy. As a consequence of this, repeated administration of the virus to the tumor site can trigger a robust immune response against the virus. In this way, a great number of tumor cells are destroyed due to viral infection and the release of several antigens into the tumor microenvironment triggers a huge immune response against the tumor ([@B9]).
Anti-viral immunity occurs as the immune system responds to the presence of virus in tumor cells within the body. This attracts various types of immune cell to the site of infection including innate cells (e.g., NK cells) and adaptive cells (e.g., CTL). These immune cells will lead to the destruction of infected cells (i.e., tumor cells) which augments the direct lysis of tumor cells by viral infection itself. This effect can be improved by arming viruses with immune modulatory proteins like cytokines which aid in the attraction of immune components to the tumor site ([@B11]).
How Innate Immune Cells Affect Anti-Viral Therapy {#S4}
=================================================
Intravenous injection with OVs can have many positive effects, such as the ability to reach metastatic lesions, but it also exposes the virus to the action of the immune system. Innate immunity is very active in this way. Macrophages are the first line of defense in the host and they naturally work as scavenger cells. They are part of the innate immune system and can recognize pathogens quickly and in a non-specific way in order to initiate an adequate inflammatory response. However, in oncolytic therapy this presents a significant hurdle. Systemic delivery of virus can be affected by macrophage phagocytosis ([@B12]) and various studies have also found that after OVT, large amounts of virus are often found in the spleen and liver following capture by mononuclear phagocytes ([@B13], [@B14]). Researchers have developed some strategies in order to overcome this problem. Chemical modification of viral coat proteins is a strategy used to mask the virus and allow it to reach the tumor tissue. To address this topic, some groups have taken advantage of polyethylene glycol (PEG), a hydrophilic, non-immunogenic, and uncharged compound that is able to interact with the biological material to avoid---or at least limit---protein--protein interactions ([@B15]). Eto et al. used PEG to coat the adenovirus surface, avoiding its identification by macrophages and prompting the use of this strategy for all applications that require systemic administration ([@B16]). Moreover, macrophages recognize antigens *via* scavenger receptors which can be saturated and therefore pre-conditioning macrophages before viral administration is a valid method of temporarily inhibiting macrophage function. Another way to overcome the challenge is to deplete macrophages before administration of the virus ([@B16]--[@B18]). The best way to achieve this is to use clodronate liposomes. The mechanism of action is very simple: macrophages phagocytose the liposomes and as a result of phospholipase-mediated disruption of the liposomal structure and consequent release of the clodronate, cell death is triggered. This approach can work in order to deliver the virus to its correct site, but can also reduce the anti-tumor efficacy of the therapy. Indeed, as mentioned above, it is important for the anti-tumor therapy that the neoplastic mass is transformed from "cold to warm" in order to re-activate the immune system, kept silent by the tumor itself.
Tumor-associated macrophages (TAMs) are a key component of this process. Macrophages can polarize into pro-inflammatory cells (M1 classical activation) that sustain the Th1 response, or into anti-inflammatory, tissue repairing cells (M2 alternative activation) that sustain the Th2 response and create an immuno-suppressive environment ([@B19]). TAMs are mostly M2-like and produce cytokines, such as TGF-β and IL-10, in order to make the microenvironment non-reactive against tumor antigens. Despite this, they can twist their phenotype according to the microenvironment ([@B20]). OVT is able to skew the M2-paradigm, activating the immune response against tumor cells. Therefore, it is important not to destroy macrophages but to re-educate them to become powerful weapons. Tan et al. show in their work that TAMs can help the oncolysis of attenuated measles and mumps virus *in vitro* ([@B21]). Despite the limitations of *in vitro* work, it opens the field to a new concept. They sustain that viral administration can change the tumor microenvironment making it "warmer," so making it a site of inflammation. In this context, TAMs (as plastic cells) can change their phenotype from M2-like to M1-like, upregulating their anti-tumor properties. *In vivo*, when the therapeutic virus reaches the neoplastic border and undertakes its lytic effect, it allows the release of many tumor antigens into the tumor environment. In this context, macrophages and DCs recognize the antigens, migrate to the lymph node where they activate T cells that are recruited into the tumor site and can exploit their anti-tumoral functions. In this way, manipulating macrophages to carry vehicle viruses into the tumor and using them as APC, appears to be a very promising strategy to improve the efficacy of OVT.
For these reasons, macrophages are the true double-edged sword of the immune system. It is very important to keep them under control, in order to avoid the clearance of the virus, but at the same time, they represent a potentially useful weapon against cancer due to their ability to "wake up" the immune response.
How Adaptive Immune Cells are Involved in OVT {#S5}
=============================================
Macrophages, DCs and other APCs have the crucial role of activating the adaptive immune response. Lymphocyte activation is slower than the innate immune response but it is more powerful and specific. Lymphocytes recognize specific antigens and start a reaction against them in a pre-determined way. The final aim of immune stimulation during OVT is to activate T lymphocytes against tumor antigens. T cells can recognize tumor peptides and provide long-term immunity. This means that not only will the primary tumor be affected but metastatic sites will as well. Moreover, due to the ability of T cells to generate memory, the patient will be protected even in the case of relapse. For this reason, tumor cells often attempt to become invisible to the immune system in order to avoid activation of lymphocytes. OVT is able to generate a cytokine storm that results in the recruitment of lymphocytes, breaking down the immuno-suppressive environment. Unfortunately, T lymphocytes do not only react against tumor cells but they are also able to initiate a strong and rapid anti-viral response. Managing the balance between anti-viral and anti-tumor activity is a big issue currently under investigation.
Over the last few years, checkpoint pathways have generated a lot of interest. T lymphocytes, as soon as they are activated, start to express checkpoint molecules, such as PD-1 and CTLA/4. During "physiological inflammation" these proteins have the key role of switching off T cells in order to block the aberrant inflammatory response and to avoid the rise of autoimmune disease. Unfortunately, tumors express PD-L1, the ligand of PD-1, in an attempt to switch off the powerful arm of immunity against cancer through the mechanism of immune evasion. For that reason, a lot of effort has been put into developing a way to block these pathways and allow T effector cells to perform their role against the tumor. Ilett et al. demonstrate that the sequential administration of different viruses in combination with an immune checkpoint inhibitor can generate a fully systemic anti-tumor immunity in a model of melanoma ([@B22]). They set up a model in which pre-conditioning with GM-CSF prior to administration of Reovirus, allows the OV to initiate tumor killing by potentiating innate immune activation. The result of this treatment is a low and indirect induction of a Th1 response against tumor antigens. The second part of the treatment consists of the administration of VSV-ASMEL, a VSV-c-DNA library of human melanoma cell lines, encoding for a huge number of tumor antigens. This results in the spread of tumor antigens into the tumor. Previous delivery of the GM-CSF/Reovirus had generated a weak T-cell response. CD4^+^ cells recruited into the microenvironment can now come in contact with a huge quantity of tumor antigens and, despite the weak primary response they can now expand and give rise to a strong cytotoxic response. Finally, the administration of anti-PD-1 antibody facilitates a strong and sustained response ([@B22]). Obviously, this approach has some limitations such as the fact that, as highlighted by the authors themselves, it is impossible to reach the clinical level with the huge library of tumor antigens that were used. Despite this, they proposed a new model to improve tumor killing by avoiding the initiation of the anti-viral response. But the most significant result has been reached by Ribas et al. The authors performed a phase Ib clinical trials using T-Vec in combination with the anti-PD-1 antibody pembrolizumab. The treatment with anti-PD-1 is able to prolong the effect of cytotoxic T lymphocytes and the main cause of its failure is due to the lack of CD8^+^ cells inside the tumor lesion. Ribas et al. demonstrated that the intratumor injection of T-Vec is able to recruit CD8^+^ lymphocytes into the tumor making them responsive to pembrolizumab. Moreover, the combined treatment is able to increase the circulation of both CD4^+^ and CD8^+^ T cells, proving a systemic and durable response ([@B23]). This approach seems to be very effective. Indeed, many new different works are trying to demonstrate the efficacy of the combination of OVs and anti-checkpoint inhibitors. Bourgeois-Daigneault et al. have shown the positive effect of the combined therapy of Maraba rhabdovirus and immune checkpoint blockade in triple-negative breast cancer ([@B24]). Analogous to this Samson et al., demonstrated that this approach is worth also in brain cancer. They showed that the intravenous injection of orthoreovirus in high-grade glioma and in brain metastases is able to increase the T-cell infiltration, improving mice survival after later anti-PD-1 treatment ([@B25]). Similar results were recently obtained in our group using a different approach. Chard et al. demonstrated the efficacy of VV armed with IL-10. IL-10 is an immuno-suppressive cytokine and they proved that administration simultaneously with the virus inhibits early immune response to infection resulting in a dampening of anti-viral but not anti-tumor immunity ([@B26]).
Strategies to Improve the Induction of Tumor-Specific Immunity by OVT {#S6}
=====================================================================
Oncolytic viral therapy has beneficial effects on the immune system. As described before, it is able to create long-lasting immunological memory thereby avoiding relapse and metastatic spread. It has been identified in certain cases that the anti-tumor immune response is much more important for clearing tumor than is the direct oncolytic effect of the virus.
Recently, mainly thanks to the advances made in genetic engineering, new OVs have been made by the insertion of genes encoding for proteins able to stimulate the immune system. The most well-known virus made in this way is T-Vec---already approved by FDA---an HSV-1 virus modified to carry the human GM-CSF gene that has been demonstrated to increase the number of monocyte-derived DCs and, as a consequence, to increase the activity of cytotoxic CD8^+^ lymphocytes by promoting antigen presentation. Considering these promising therapeutic effects, other viruses such as Vaccinia virus, have been modified to carry this gene.
Based on these encouraging results, other strategies have been developed to include interleukins in order to potentiate the adaptive immune response. Among these, IL-2 has been used to activate T lymphocytes and IL-12, IL-15, and IL-18 have been used to activate both T and NK cells ([@B27]--[@B31]). Also, other interleukins such as IL-4 have been tested (in adenovirus) but with unsatisfactory results from a safety point of view ([@B32]). In addition to cytokines, chemokines have also been used to engineer OVs. The most promising results have been obtained from the insertion of CCL5 and CCL3 in adenovirus. CCL5 can attract T lymphocytes to the tumor, while CCL3 is able to recruit neutrophils into the tumor ([@B33], [@B34]). As discussed, the immune system is activated by OVTs and new strategies are now rising in order to improve the immune response. Insertion of a particular cytokine into the virus can attract a specific immune population into the tumor site, favoring the killing of cancer cells and the breakdown of the immuno-suppressive microenvironment.
One of the promising avenues of OV enhancement is to combine them with the use of T-cell engager molecules in order to stimulate existing T cells to lyse tumor cells by creating a bridge between the two, thus allowing the activation of the cytolytic properties of the T cells present in the tumor. It has been shown that the use of bi-specific T-cell engager molecules can induce T-cell activation even when MHC-I is absent from the surface of the target cell. This feature of T-cell engagers would help to overcome one of the major problems of cancer cells which is immune evasion through downregulation of MHC-I molecules displayed on the surface ([@B35]). It has also been shown that the use of bi-specific antibodies that target both cancer cell markers and T cells can lead to long-term anti-tumor immunity which is an important aspect of cancer therapy ([@B36]). Using this therapy in combination with OVs has been seen to improve the T-cell killing of tumor cells in various studies ([@B37], [@B38]). This is because in addition to the direct oncolytic effect of the parental virus, the T-cell engager expressing virus can also recruit the existing T cells to fight cancer cells thereby increasing the number of methods in use to eliminate tumor cells.
However, a potential limitation of this therapy is that high levels of Tregs in individual patients can result in lower efficacy as Tregs reduce the level of T-cell killing activity ([@B39]). This can be overcome, however, by the removal of Tregs before treatment to eliminate the dampening of the T-cell response by these cells.
Another positive effect of arming OVs with T-cell engagers is the bystander effect demonstrated in specific cases. It has been shown that tumor cells negative for the target molecule of the engager (in this case EGFR) were lysed when in proximity to tumor cells that display the target antigen through T-cell induced bystander cell lysis ([@B40]). A phase II clinical trial has been reported for T-cell engager molecules which showed promising initial results but leaves room for improvement in terms of delivery regime in order to reduce the number of patients who need to drop out of the treatment early due to side effects. It would also be worth initiating a study to analyze the effect of T-cell engager delivery by OVs to target tumor cells and potentially reduce side effects.
In parallel to this, our group studied a new elegant strategy to boost the immune response against tumor cells. When a virus is injected, the immune system reacts against it generating an anti-viral response. In some cases, this reaction is too strong and it results in the complete clearance of the virus. In order to overcome this problem, we set up a therapeutic regime in a Syrian Hamster model whereby they were challenged with two different OVs: adenovirus first followed by Vaccinia virus administration ([@B41]). In this way, we demonstrated that the immune system reacts against adenovirus leaving the VV able to exploit its anti-tumor function. It was also found that the addition of the second virus can boost the anti-tumor immune response. This occurs as the immune system is focused on developing a specific response against the first virus and is not able to simultaneously inhibit the therapeutic functions of the second. Nevertheless, further studies are needed in order to demonstrate the possible role of innate immunity in this process.
In addition, neutralizing antibodies (NAb) represent another relevant face of immunity that is worth considering. Normally, they recognize the virus, coat it and lead it to be phagocyted by competent cells. However, Adair et al. show that Reovirus administered i.v. can be rapidly cleared in the plasma but not in the cellular fraction of the blood because it is able to form complexes with blood cells. Further work allowed them to discover that the access of Reovirus to tumors can be paradoxically enhanced by the presence of anti-viral NAb which form complexes with Reovirus for uptake, carriage, and delivery to tumors by monocytes in the blood ([@B42], [@B43]). However, normally NAb block the systemic delivery of almost all OVs therefore a deeper analysis and characterization of this process is necessary.
Finally, another therapy used to boost the immune system is named "trojan horse" therapy. In this approach, immune cells are directly infected with the virus *ex vivo*. Studies have revealed increased efficacy with this methodology by which the immune cells are able to directly load the virus into the tumor site ([@B44], [@B45]) retaining the positive effect that OVT has on the immune system. Despite the promising results, there are some limitations to this technique considering that some tissues are not suitable for this therapy. For example, it is not possible to infiltrate the brain with high numbers of molecules and cells.
Intratumoral delivery of the virus can affect the local tumor microenvironment. However, it is not always feasible because of the inaccessibility of the tumor itself. Pancreatic tumors, for example, are very difficult to directly reach. Moreover, local injection of the virus does not take into account potential metastatic lesions. To overcome all of these issues, an alternative to *in situ* delivery is represented by intravenous injection. Intravenous delivery has the potential to reach metastatic lesions as well as the primary tumor even if they are not clinically detectable. Russell et al. ([@B46]) demonstrated in a phase I study that a complete and durable response was generated in myeloma patients treated intravenously with an oncolytic Measles virus. Of note, this treatment worked only in those patients without NAb. NAb are one of the major obstacles to intravenous delivery of OVs. However, it has been shown that some viruses, such as Reolysin, are not affected by the presence of nAbs ([@B42]). Even if relatively few results have been achieved so far in this direction, a new prospective in the field is now open.
Potential Risk for Modulating OVT-Mediated Host Immune Response {#S7}
===============================================================
Whilst oncolytic therapy can be modified to produce the desired effect in a multitude of ways, the outcome of treatment is not always positive. One of the major obstacles to successful oncolytic therapy is the host immune response to viral infection, especially when the virus is administered IV. Indeed, after IV injection, monocytes and macrophages phagocyte the virus, preventing it to reach the tumor site. This response is highly tuned and extremely effective in most cases and therefore, overcoming this hurdle requires careful design of optimal viral gene deletion and insertion combinations. A good study in this direction in represented by Enadenotucirev, a tumor-selective chimeric adenovirus that has been recently used in a phase I clinical trials. Garcia-Carbonero et al. made a study---in collaboration with PsiOxus Therapeutics---injecting the virus both IV and IT in patients with different epithelial tumors. They found no toxicity caused by IV injection. They showed increased levels of some inflammatory cytokines after the infection which rapidly return to the pre-injection levels in 48 h. The authors claim that this is due to a low-dose infection. Their treatment seems able to decrease the cytokines toxicity, allowing the IV treatment ([@B47]).
The main side effect of the host immune system is efficient clearance of virus and therefore diminished effect of treatment. In this case, the host immune response to viral infection is such that the virus is cleared from the body before it can have an impact on its target tissue (mostly affects viral treatments which are delivered systemically). In order to combat this, viruses can be delivered within host cells which are extracted, infected, and subsequently replaced. This provides a route into the body which is sheltered from the host immune response.
Another problem posed by the host immune response is hyper activation in response to viral treatment. This problem can be combated by attenuation of the virus in order to allow delivery of a higher dose with fewer side effects than a more virulent strain. Subversion of certain pathways involved in the host immune response can also be used in order to dampen the immuno-pathology whilst still allowing generation of anti-tumor immunity.
We also need to consider the feasibility of delivering oncolytic therapy to vulnerable patients such as the elderly and immuno-compromised. In these cases, it is imperative to ensure that the virus does not infect and cause chronic infection due to lowered immune responses as this could lead to negative side effects and resistance to that particular viral treatment option. On this note, it is also important to consider the suitability of each patient for oncolytic therapy. It has been reported that levels of chronic innate immune stimulation detected in patients correlates with the outcome of oncolytic viral therapy. It was seen in this study that higher levels of chronic immune stimulation in patients before treatment with oncolytic adenovirus correlate with poor prognosis, suggesting that patients with low levels of chronic immune stimulation are more likely to respond positively to oncolytic viral therapy ([@B48]).
It is also important to keep in mind that manipulation of viruses by adding cytokines also manipulates the microenvironment. Cytokine treatment requires careful monitoring of concentration. It is important to find a balance between the desired effect and the unwanted side effects that the treatment can create. For example, using PD-1/PD-L1 inhibitor poses the risk of a persistent activation of the immune system and a consequent initiation of autoimmune disease.
Finally, to make sure that research efforts are well spent, we should take in consideration the animal model used. Mice are widely used in research laboratories but sometimes they are not the best model, especially for assessment of OVs. Many OVs are species-specific and cannot efficiently replicate in murine cells without any effects on human cells. This is a particular issue for human-specific adenoviruses that have been reported to have a very poor replication in a great variety of murine tumor cell lines. To avoid this issue, sometimes xenograft models are used in which human tumor cells are implanted into an immuno-compromised recipient. However, this approach does not take in consideration of the complexity of the tumor microenvironment and it is a simplistic solution that cannot be validated for all studies. Tysome et al. propose a different approach, using Syrian hamster as an immuno-competent model. They showed that this model is able to support Human-Adenovirus and Vaccinia virus replication, providing a valid alternative to murine models ([@B41]).
Future Remarks {#S8}
==============
Oncolytic virus therapy is a very promising new treatment for many different types of cancers. Theoretically, every kind of cancer could be treated by viruses. The main problem of this approach is that the pre-clinical animal models could unfortunately have a very different anti-viral response with respect to humans. Considering this, animal studies should be carefully designed and performed. Another limitation in OVT is represented by the immune system itself in that, as mentioned above, it can start a reaction against the virus. It is important to find a balance between the immune anti-viral response and the immune anti-tumor response in order to find the perfect equilibrium, windows to give OVs and consequently the best way to fight cancer. To achieve this goal, new investigations are ongoing in order to design viruses that could stimulate specific immune populations thereby exploiting the true potential of the innate weapon. The final aim of all these studies is to stimulate the immune system to give rise to the correct response, avoiding aberrant inflammation that could result in a risk for patients. Moreover, recent studies have demonstrated that the true potential of OVT is in combination with classical treatments and other new therapies. Next generation studies should be focused on achieving these two aspects: making the viruses safer and more powerful and allowing them to work in synergy with other compounds.
Author Contributions {#S9}
====================
YW provided the conception of the article; GM, AH, and YW designed the article; GM and AH drafted the article while YW made critical revisions related to important intellectual content of the manuscript and approved the final version of the article to be published with Nick Lemoine.
Conflict of Interest Statement {#S10}
==============================
The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
This research was supported by The MRC (MR/M015696/1 and MR/N027655/1), Pancreatic Cancer Research Fund, and Breast Cancer Now, Pancreatic Cancer UK, Ministry of Science and Technology, 13th five year plans for Strategic International Collaboration Programme Grant (2016YFE0200800), and Natural Science Foundation of China (81771776).
[^1]: Edited by: Vincenzo Bronte, University of Verona, Italy
[^2]: Reviewed by: Cristina Fillat, Consorci Institut D'Investigacions Biomediques August Pi I Sunyer, Spain; Pedro Berraondo, Centro de Investigación Médica Aplicada (CIMA), Spain; Carlos Alfaro, Universidad de Navarra, Spain
[^3]: ^†^These authors have contributed equally to this work.
[^4]: Specialty section: This article was submitted to Cancer Immunity and Immunotherapy, a section of the journal Frontiers in Immunology
| {
"pile_set_name": "PubMed Central"
} |
Introduction
============
Viruses are obligatory parasites that critically rely on their ability to transmit their genome from infected to uninfected host organisms. Being an inert particle, viruses have successfully evolved to exploit the behavior and physiology of their host. Viral infection induces the activation of various endogenous responses that enable it to permeate through cell membranes and other barriers to reach the cytoplasm or nucleus.
Herpesviruses have a large double stranded DNA genome enclosed in the viral protein shell (capsid) surrounded by a tegument layer which is enclosed in a lipid envelope with at least eight distinct viral envelope glycoproteins. A characteristic property of these viruses is that after primary infection they establish lifelong latent infection in the infected host with periodic reactivation and re-infection. KSHV or human herpesvirus-8 (HHV-8) is a member of the γ2-lymphotropic-oncogenic herpesviruses. KSHV is etiologically associated with Kaposi's sarcoma (KS) and with at least two lymphoproliferative malignancies, primary effusion lymphoma (PEL) and multicentric Castleman's disease (MCD). It is the newest member of the human herpesvirus family and is closely related to γ-1 Epstein--Barr virus (EBV), γ-2 herpesvirus saimiri (HVS) and Rhesus monkey rhadinovirus (RRV; Ganem, [@B17],[@B18]). KSHV has a double stranded DNA genome of about ∼160-kb encoding more than 90 ORFs designated 4--75 by their homology to HVS ORFs. The genome contains gene blocks conserved with other herpesviruses as well as divergent regions encoding more than 20 KSHV unique genes (K genes). KSHV encodes several proteins that are homologs of host proteins with immunomodulatory, anti-apoptotic, signal induction, transcriptional regulation, and other functions (Cesarman et al., [@B11]; Neipel et al., [@B48]).
The first step of any viral infection is governed by binding and entry into target cells. Therefore, to control KSHV infection, a detailed understanding of how KSHV infects its target cells utilizing the varied set of cellular receptors, envelope glycoproteins, signaling, and modes of entry is essential. Recent advances indicate that KSHV interacts with multiple host cell surface receptors of adherent target cells and these interactions induce a network of rapid intracellular signaling pathways, which facilitate the various steps of successful infection. Here, we review the important steps involved in KSHV entry into target cells utilizing viral envelope--cellular receptor interactions, and signal cascades inducing dynamic cell membrane changes leading to a productive latent infection.
KSHV Tropism
============
To better understand the underlying multistep complex entry mechanism(s) initiated by KSHV, one must appreciate the broad variety of cell types infected both *in vivo* and *in vitro*.
*In vivo* KSHV has a broad tropism as suggested by the detection of its genome and transcripts in a variety of *in vivo* cell types such as CD19+ peripheral blood B cells, endothelial cells, monocytes, keratinocytes, and epithelial cells (Ganem, [@B18]). Latent KSHV DNA is present in vascular endothelial and spindle cells of KS lesions, associated with expression of latency-associated ORF73 (LANA-1), ORF 72 (v-cyclin D), K13 (v-FLIP), and K12 (Kaposin) genes and microRNAs (Boshoff et al., [@B9]; Dupin et al., [@B14]; Ganem, [@B18]). Lytic infection is also detected in \<1% of infiltrating inflammatory monocytic cells of KS lesions (Dourmishev et al., [@B13]; Ganem, [@B18]). Available evidences suggest that B cells and monocytes are the major reservoir of *in vivo* latent infection. Cell lines with B cell characteristics, such as BC-1, BC-3, BCBL-1, HBL-6, and JSC have been established from PEL tumors (Dourmishev et al., [@B13]; Ganem, [@B18]). In PEL cells, in addition to the above set of latent genes, K10.5 (LANA-2) gene is also expressed (Parravicini et al., [@B50]; Ganem, [@B18]). About 1--3% of PEL cells spontaneously enter lytic cycle and virus induced from these cells by chemicals serve as the source of virus. Multiple genome copies of both KSHV and EBV exist in latent form in BC-1, HBL-6, and JSC cells while BCBL-1 and BC-3 cells carry only the KSHV genome (Ganem, [@B18]). An endothelial cell line carrying KSHV has not been established from KS lesions since KS cells grow poorly in cell culture and viral DNA is lost within a few passages (Ganem, [@B18]).
Kaposi's sarcoma associated herpesvirus has been shown *in vitro* to infect several types of human cells such as B, endothelial, epithelial, fibroblast cells, CD34+ stem cell precursors of dendritic cells (DCs), and monocytes (Ganem, [@B18]). KSHV also infects owl monkey kidney cells, baby hamster kidney (BHK-21) cells, Chinese hamster ovary (CHO) cells, and mouse fibroblasts cells (Parravicini et al., [@B50]; Akula et al., [@B2],[@B3], [@B4]; Birkmann et al., [@B8]; Bechtel et al., [@B6]; Inoue et al., [@B27]; Garrigues et al., [@B19]; Jarousse et al., [@B29]).
Infection of primary B cells by KSHV does not result in immortalization and a lytic KSHV replication is seen in activated B cells. Another characteristic feature of *in vitro* infection of human microvascular dermal endothelial cells (HMVEC-d), human umbilical vein endothelial cells (HUVEC), human foreskin fibroblasts (HFF), human endothelial cells immortalized by telomerase (TIME), and human endothelial cells (HEK-293), monkey kidney cells (VERO, CV-1), and mouse fibroblasts (Bechtel et al., [@B6]) by KSHV is the expression of latency-associated genes and the absence of productive lytic replication and thus providing a reasonable model for studying *in vitro* latency. However, latent infection of KSHV *in vitro* is not persistent and leads to the loss of viral genome over time (Grundhoff and Ganem, [@B24]).
Analysis of *in vitro* KSHV interaction with adherent target cells and quantitation of infection has been hampered by the absence of a lytic replication cycle and hence a plaque assay. Since *in vitro* KSHV infection results in the expression of latency-associated genes, various methods have been devised to assess the different phase(s) of KSHV infection (Parravicini et al., [@B50]; Table [1](#T1){ref-type="table"}).
######
**Methods employed to study the various stages of *in vitro* KSHV infection**.
Stage of infection Detection methods
--------------------------------------------------- -----------------------------------------------------------------------------------------------------------------------------------------------------------------------
1\. Binding \[H3\] Thymidine labeled virus and FITC labeled (FACS) binding assay; viral DNA (ORF73 gene) DNA quantitation by real-time DNA PCR; electron and confocal microscopy.
2\. Signal induction Quantitation of signal molecules induction by Western blots and ELISA; use of chemical inhibitors or dominant-negative signal molecules.
3\. Viral DNA internalization (entry) Real-time DNA PCR for KSHV ORF73 gene after removal of unbound/partially bound virus by trypsin--EDTA; electron and confocal microscopy.
4\. Cytoplasmic trafficking of KSHV Confocal microscopy by colocalizing virus with microtubules and endosomal vesicles; physiological ligand uptake assays.
5\. Nuclear delivery of KSHV DNA Real-time DNA PCR for ORF73 gene in the isolated nuclei of infected cells.
6\. Viral gene expression, host gene manipulation Real-time RNA PCR for KSHV and host gene expression; confocal microscopy and FACS for KSHV ORF73 gene and GFP expression.
KSHV Binding and Entry into Target Cells
========================================
Kaposi's sarcoma associated herpesvirus uses multiple envelope glycoproteins to complete the binding and entry processes. KSHV binding to the target cells and identity of the receptors involved in binding and entry were elucidated by using labeled virus binding to the target cells at 4°C as well as other methods (Table [1](#T1){ref-type="table"}). These studies have demonstrated that KSHV binds and enters a variety of target cells which include human (293, HFF, HeLa, HMVEC-d, HUVEC, TIME, BCBL-1, BJAB, Raji), monkey (Vero, CV-1), hamster (BHK-21, CHO), and mouse (Du17) cells. This is demonstrated by the detection of viral DNA, limited viral gene expression, and GFP expression (Table [1](#T1){ref-type="table"}). Real-time DNA PCR of internalized KSHV DNA demonstrates a rapid internalization of viral DNA in the infected endothelial and HFF cells (Krishnan et al., [@B34]).
KSHV Envelope Glycoproteins-Mediators of Binding and Entry
==========================================================
Kaposi's sarcoma associated herpesvirus envelope glycoproteins play critical roles in mediating virus attachment, entry, assembly, and egress of virus. Like other herpesviruses, KSHV encodes five conserved glycoproteins gB (ORF 8), gH (ORF 22), gL (ORF 47), gM (ORF 39), and gN (ORF 53; Cesarman et al., [@B11]; Neipel et al., [@B48]; Ganem, [@B17]). In addition to these glycoproteins, KSHV also encodes the unique lytic cycle associated glycoproteins ORF 4, gpK8.1A, gpK8.1B, K1, K14, and K15 (Cesarman et al., [@B11]; Neipel et al., [@B48]; Ganem, [@B17]). Among these, ORF 4, gB, gH/gL, gM/gN, and gpK8.1A are associated with KSHV envelopes (Baghian et al., [@B5]; Parravicini et al., [@B50]; Akula et al., [@B2],[@B3]; Birkmann et al., [@B8]; Wang et al., [@B66]; Naranatt et al., [@B44]; Koyano et al., [@B33]).
Kaposi's sarcoma associated herpesvirus gB is a major envelope glycoprotein. It is synthesized as a 110-kDa precursor protein which undergoes cleavage and processing to yield envelope associated disulfide linked 75 and 54-kDa polypeptides with high mannose and complex sugars (Baghian et al., [@B5]; Akula et al., [@B2]; Wang et al., [@B67]). KSHV-gB mediates viral binding and entry by interacting with cell surface heparan sulfate and integrins α3β1, αVβ3, and αVβ5 (Akula et al., [@B2],[@B3], [@B4]; Veettil et al., [@B64]). KSHV-gB interaction with target cells modulates the host cell signaling pathways by inducing integrin associated focal adhesion kinases (FAK), Src, PI3-K, and Rho-GTPase activities (Sharma-Walia et al., [@B60]). Studies have demonstrated that besides playing a role in KSHV binding and entry of target cells, gB is critical for virus maturation and egress (Krishnan et al., [@B36]).
Distinct from other herpesviruses, the KSHV gpK8.1 gene encodes two alternatively spliced messages yielding glycoprotein gpK8.1A and gpK8.1B. Both gpK8.1A and gpK8.1B contain N- and O-linked sugars, and gpK8.1A is the predominant form detected within infected cells and in the virion envelopes (Neipel et al., [@B48]; Zhu et al., [@B69],[@B70]). Like gB, gpK8.1A also possesses the heparan sulfate binding motif and interacts with cell surface heparan sulfate molecules (Wang et al., [@B66]). Similar to herpes simplex type 1 (HSV-1) gB glycoprotein, recent studies have reported that KSHV gB and gpK8.1A are enriched in membrane microdomains, lipid rafts (LRs), during early infection in endothelial cells (Bender et al., [@B7]; Chakraborty et al., [@B12]).
Like in other α, β, and γ-herpesviruses, KSHV gH and gL form a non-covalently linked complex consisting of 120-kDa gH and 42-kDa gL proteins. KSHV gL is required for processing and intracellular transport of gH and this complex is critical for KSHV entry (Naranatt et al., [@B44]). Recent studies show that KSHV gH, as well as complement binding KSHV ORF4, also interacts with cell surface heparan sulfate (Mark et al., [@B41]; Hahn et al., [@B25]). Studies have demonstrated that anti-gH and anti-gL antibodies inhibit KSHV entry without affecting binding to the target cells (Naranatt et al., [@B44]).
Kaposi's sarcoma associated herpesvirus gM and gN also form a glycosylated heterodimeric complex and are involved in virus penetration and egress. gN has been shown to be essential for proper post translational modification and transport of gM to the cell surface (Koyano et al., [@B33]). KSHV gM and gN forms a heterodimeric complex and were shown to inhibit cell fusion in an *in vitro* cell fusion assay (Koyano et al., [@B33]).
Cellular Receptors Recognized by KSHV
=====================================
Most cell surface molecules that herpesviruses bind fall into two main categories depending on the functional consequences of the interaction. The major groups include the attachment or binding factors and entry receptors. Attachment factors promote binding and concentration of viruses on target cells but may not be very specific. Often, these involve charge interactions involving heparan sulfate or other carbohydrate moiety bearing molecules.
The other group comprises entry receptors, including a broad variety of cell surface molecules capable of either viral fusion to cell membranes or initiating signaling to promote endocytosis. Entry receptors are highly specific and vary by cell types. Often, these molecules are internalized along with the virus, hence are actively utilized by viruses.
Heparan Sulfate as KSHV Attachment Factor in Target Cells
=========================================================
Like many herpesviruses, KSHV utilizes ubiquitous heparan sulfate (HS) molecule for binding to most of its target cells. HS bears carbohydrate residues that facilitate concentration of virus particles on target cells owing to charge interactions. KSHV infection can be inhibited by soluble heparin but not by chondroitin sulfates A and C (Akula et al., [@B2]). Pre-treatment of KSHV with soluble heparin prevents virus binding and subsequent signal induction. This is a specificity control for KSHV-induced phenomena. Several B cell lines and primary B cells lack the Ext1 enzyme, which promotes glycosylation in HS biosynthesis. Due to this, B cells have low HS expression, and thereby refractory to KSHV infection. However, expression of HS in BJAB (EBV negative cell line) results in greater susceptibility to KSHV infection (Jarousse et al., [@B29]). These studies suggest that KSHV infection of BJAB cells depends on the level of expression of HS; BJAB cells expressing HS were readily infected while cells lacking HS were not infected (Kabir-Salmani et al., [@B30]).
Kaposi's sarcoma associated herpesvirus envelope glycoproteins gB, gpK8.1A, ORF4, and gH bind to cell surface HS molecules, thereby emphasizing the importance of cell surface HS for KSHV attachment to target cells (Birkmann et al., [@B8]; Akula et al., [@B4]; Wang et al., [@B67]; Hahn et al., [@B25]). Incubation of KSHV with soluble heparin and enzymatic removal of cell surface HS by heparinase I and III inhibits KSHV infectivity. Binding of the soluble forms of gB and gpK8.1A to target cells is saturable and can be blocked by soluble heparin (Birkmann et al., [@B8]; Wang et al., [@B66], [@B67]).
Many proteins possess more than one of the two heparin binding domains (HBD), containing the XBBXBX and XBBBXXBX sequences, where B is a positively charged basic amino acid (lysine, arginine, or histidine) flanked by an additional positively charged residue separated by hydrophobic amino acids "X." The extracellular domain of KSHV gB possesses 108HIFKVRRYRK117, which is a BXXBXBBXBB type HBD, and is conserved throughout the γ2 herpesviruses. KSHV gpK8.1A also possesses two possible atypical heparin-binding motifs, ^150^SRTTRIRV^157^ (XBXXBXBX) and ^182^TRGRDAHY^189^ (XBXBXXBX) whereas KSHV gH lacks the typical HBD.
Several lines of evidence indicate that KSHV-gB and gpK8.1A bind to cell surface HS molecules (Akula et al., [@B3]; Birkmann et al., [@B8]; Wang et al., [@B67]). Binding of soluble forms of the proteins made in baculovirus is saturable and can be blocked by soluble heparin (Wang et al., [@B66], [@B67]). Full length gB and gpK8.1A in the virion envelope specifically bind heparin--agarose, and can be eluted by high concentrations of soluble heparin, but not by chondroitin sulfates (Wang et al., [@B66]; Akula et al., [@B4]). KSHV-gpK8.1A binds to heparin with an affinity comparable to that of glycoproteins B and C of herpes simplex virus and gpK8.1A binds more strongly than gB (Wang et al., [@B67]).
DC-SIGN as KSHV Entry Receptors
===============================
Dendritic cell specific intercellular adhesion molecule-3 (ICAM-3) grabbing non-integrin (DC-SIGN; CD209) is a C-type lectin present on the dermal DC surface. It has been shown to be used by many viruses as a receptor including human immunodeficiency virus (HIV) and more recently by Bunyaviruses (Geijtenbeek et al., [@B20]; Lozach et al., [@B39]). Similarly, KSHV also appears to utilize DC-SIGN during infection of human myeloid DCs, macrophages, and activated B cells (Rappocciolo et al., [@B54], [@B53]). DC-SIGN blocking monoclonal antibodies and mannan efficiently inhibited KSHV binding and infection. However, the pre-treatment of cells with anti-DC-SIGN antibodies did not completely block KSHV binding and infection possibly due to binding to HS and/or other receptors. KSHV envelope glycoprotein(s) interacting with DC-SIGN is not known yet and KSHV gB with its high mannose sugar is a potential candidate. Recent studies have shown that activated B cells are infected more efficiently due to increased expression of DC-SIGN. However, whether HS and other KSHV receptors are also expressed at higher levels possibly contributing to the increased infectivity in activated B cells need to be studied further.
Recent studies have also elucidated that KSHV effectively binds and enters THP-1, a human acute monocytic leukemic cell line, using heparan sulfate and integrins (Kerur et al., [@B32]). Blocking DC-SIGN in these cells did not affect KSHV binding; however KSHV entry was reduced, suggesting that KSHV utilized DC-SIGN as part of the entry receptor(s) in addition to the previously identified integrins (α3β1, αVβ3, and αVβ5) in THP-1 cells (Kerur et al., [@B32]).
KSHV is the First Herpesvirus Shown to Utilize Integrins as Entry Receptors in Adherent Cells
=============================================================================================
Several viruses utilize multiple integrins for target cell infection and engaging integrin receptors leading to induction of potent signaling responses critical for virus infectivity. KSHV-gB possesses an integrin-binding RGD (Arg-Gly-Asp) motif at amino acids 27--29. The RGD motif is the minimal peptide sequence of many integrin ligands known to interact with subsets of cellular integrins. Several studies have demonstrated that α3β1, αVβ3, and αVβ5 integrins play roles in KSHV infection. Using RGD peptides, antibodies against RGD-gB (RGDTFQTSSSPTPPGSSS), and the extracellular matrix (ECM) protein fibronectin studies have shown the role of integrins in KSHV infection of HMVEC-d and HFF cells (Akula et al., [@B4]). In HMVEC-d and HFF cells, KSHV interacts with integrin α3β1 as demonstrated by a 30--50% reduction in infection by pre-treating cells with function blocking anti-α3 and β1 antibodies and by mixing virus with soluble α3β1 integrin before infection, as well as the immunoprecipitation of virus-α3 and β1 complexes by anti-KSHV-gB antibodies (Akula et al., [@B4]). Though expression of α3 integrin in CHO cells increases infectivity, the levels of infection do not reach that observed in the HMVEC-d and HFF cells strongly suggesting that KSHV uses multiple receptors (Akula et al., [@B4]; Naranatt et al., [@B45]; Krishnan et al., [@B34]).
Virus binding and DNA internalization studies suggest that integrins αVβ3 and αVβ5 also play roles in KSHV entry (Veettil et al., [@B64]). Variable levels of inhibition of virus entry into adherent HMVEC-d, 293 and Vero cells, and HFF was observed by pre-incubating virus with soluble αVβ3, αVβ5, and α3β1 integrins, and cumulative inhibition was observed with a combination of integrins. Confocal microscopy studies confirmed the association of KSHV with α3β1 integrins (Veettil et al., [@B64]; Chakraborty et al., [@B12]). A study also revealed the roles of integrins (α3β1, αVβ3, and αVβ5) in the entry of KSHV in THP-1 and primary monocyte cells (Kerur et al., [@B32]).
Discrepancies Regarding the Role of Integrins in KSHV Biology
=============================================================
Two studies did not detect a role for α3β1 in KSHV infection. However, methodological differences could explain the discrepancies. For example, Inoue et al. ([@B27]) reported the inability of soluble α3β1 integrin and RGD peptides to block KSHV infectivity in the 293-T cell line. However, the validity of this observation is questionable since in this study, cells were pre-treated with RGD peptide and infected with KSHV by using centrifugation and polybrene. Both centrifugation and polybrene are known to enhance virus infection without the need for the virus to interact with specific receptors. It is well known that polybrene forming a complex with viral envelope is used for gene delivery into various target cells bypassing the need of specific receptors. Moreover, Inoue et al. ([@B27]) studies also pre-incubated the cells with integrins, washed, and infected with KSHV. This is not a correct design of experiment since to demonstrate the role of integrin in any viral infection cells need to be incubated first with anti-integrins antibodies, but not soluble integrins, prior to viral infections. Similarly, virus needs to be pre-incubated with soluble integrins before addition to cells to demonstrate the role of respective integrin in viral infection.
Another study utilized a 15-mer-AHSRGDTFQTSSGCG peptide of KSHV-gB and demonstrated that this peptide mediated the human fibrosarcoma HT1080 cell adhesion which was blocked by αVβ3 and αVβ5 antibodies while peptide bound beads detected only αVβ3 integrin (Garrigues et al., [@B19]). The GCG amino acids in the peptide used in this study are not present in the KSHV-gB sequence and may potentially give rise to dimers and multimers due to the cystine residue (Garrigues et al., [@B19]). Though HT1080 cell infection was inhibited by anti-αVβ3 antibodies, the ability of anti-α3β1 and αVβ5 antibodies to block infection was not examined (Garrigues et al., [@B19]). Moreover, co-immunoprecipitation and colocalization studies to demonstrate the direct association of KSHV with integrins were not done in these studies. Hence, to disproof that KSHV infection depends upon α3β1, studies should be carried out with the same target cells that have been used to show the role of α3β1 integrin in KSHV target cell infection. Moreover, the differential ability of integrins to block infection remains questionable in HT1080 cells since another study could not infect these cells (Veettil et al., [@B64]). Mouse keratinocytes lacking α3β1 were infectable with KSHV and expression of human α3 resulted in only 55% of infection in these cells. Even though the level of αVβ3 in these cells and the ability of anti-αVβ3 to block KSHV infection were not tested, it was concluded that α3β1 expression must have a dominant-negative effect on αVβ3 integrin. Later studies clearly demonstrating the association of KSHV with α3β1 integrin and LR-KSHV and α3β1 integrins (Veettil et al., [@B64]; Chakraborty et al., [@B12]), it is clear the conclusions from the above two studies is not correct and should not be considered further.
The discrepancies in the usage of different integrins were expected since it is common for different cells to express different combinations of integrins and a specific integrin could be one of the receptors in some but not in all target cells of KSHV. Since herpesvirus--cell receptor interactions are temporarily coordinated events mediated by interactions of viral glycoproteins with one receptor leading to conformational changes in the viral glycoproteins allowing interaction with the next receptor(s), detection of α3 and β1 in HFF or HMVEC-d cells incubated with virus or purified gB (Akula et al., [@B4]) could be representing the event occurring during virus--host cell interactions under physiological conditions. Further studies are required to determine the role of additional receptor(s) in different target cells of KSHV.
Later studies demonstrated that αVβ3 and αVβ5 integrins also play roles in KSHV entry of adherent target cells (Veettil et al., [@B64]; Chakraborty et al., [@B12]; Table [2](#T2){ref-type="table"}; Figure [1](#F1){ref-type="fig"}). Studies have also demonstrated the neutralization of KSHV infection in HMVEC-d and HFF cells upon blocking with anti-αVβ3 and αVβ5 antibodies. Supporting these findings, KSHV entry was inhibited by soluble α3β1, αVβ3, and αVβ5 integrins in HMVEC-d, HFF, 293, and Vero cells (Veettil et al., [@B64]). Though a cumulative inhibition was observed with combinations of integrins, complete block in infection was not achieved which may be due to KSHV's first interaction with HS. The role of integrin in KSHV infection of HUVEC cells, B cells, monocytes, keratinocytes, and other cells has not been studied (Table [2](#T2){ref-type="table"}).
![**Schematic model depicting the different overlapping phases of KSHV entry and infection in endothelial (HMVEC-d) target cell**. KSHV infection is initiated by binding to the cell surface proteoglycan heparan sulfate \[HS; Stage 1: binding; inset electron microscopic, (EM) picture\], followed by subsequent temporal associations with integrins (α3β1, αVβ3, αVβ5) and xCT molecules in the non-lipid raft (NLR) parts of the membranes. KSHV's interactions with integrins activate FAK at tyrosine 397, which creates a binding site for the SH2 domain containing Src family kinases, subsequently leading to the activation of PI3-K and Rho-GTPases (Stage 2: signal induction). These rapid overlapping host cell signal induction play roles in actin modulation, formation of endocytic vesicles, and virus entry and trafficking through the cytosol leading into a productive infection (Blue arrows). KSHV infection induces the phosphorylation of c-Cbl and the phosphorylated c-Cbl forms a complex with p85-PI3-K, leading into the interaction of c-Cbl with downstream molecules. c-Cbl mediates a rapid selective translocations of KSHV into the lipid rafts (LRs) along with the α3β1, αVβ3, and xCT receptors. KSHV-associated αVβ5 remains in the NLR parts of the membranes. Activated c-Cbl localizes with LRs, associates with myosin IIA and actin, and is rapidly recruited to membrane blebs. This also leads into c-Cbl mediated ubiquitination of actin and myosin. c-Cbl mediated monoubiquitination of translocated receptors is followed by productive macropinocytic entry. Myosin IIA interactions with actin may be providing the ATP-dependent force to generate actomyosin contraction, bleb retraction to form macropinosomes along with KSHV (Stage 3: internalization; inset EM picture). NLR associated KSHV bound αVβ5 and other receptors are polyubiquitinated and directed to a clathrin-dependant lysosomal non-productive pathway. RhoA activates Dia-2 aiding in formation and movement of endosomes through the cytoplasm. KSHV capsid is released from the endocytic vesicles by fusion of viral envelope with endosomal vesicles (Stage 3: internalization; inset EM picture). Released capsids are transported toward the nucleus utilizing a Rho-GTPase dependant pathway that involves acetylation of microtubules (MT; Stage 4: movement in cytoplasm). KSHV capsid disassembly at or near the nuclear pore results in the delivery of KSHV DNA into the infected cell nucleus (Stage 5: nuclear delivery) followed by viral and host gene expression (Stage 6) initiated by KSHV binding and entry induced ERK and NF-κB pathways. All these events demonstrate that KSHV has evolved to utilize its interactions with cellular receptors to manipulate host cell signaling and to induce an environment that is conducive for a productive infection.](fmicb-03-00006-g001){#F1}
######
**Binding and entry receptors and entry pathways of KSHV in various *in vitro* human target cells**.
Target cells Receptors recognized Entry pathway
---------------------------------------------------------------------------------------------------------------------------- ------------------------------------------------------- ---------------------------------------
Human foreskin fibroblast cells (HFF) Heparan sulfate (binding), α3β1, αVβ3, αVβ5, xCT/CD98 Clathrin-mediated endocytosis
Human microvascular dermal endothelial cells (HMVEC-d) Heparan sulfate (binding), α3β1, αVβ3, αVβ5, xCT/CD98 Macropinocytosis
Human umbilical vein endothelial cells (HUVEC) Heparan sulfate (binding) Macropinocytosis mediated endocytosis
Human embryonic kidney epithelial cells (HEK-293 with adenovirus 5 DNA nts 1--4344 integrated into chromosome 19 (19q13.2) Heparan sulfate (binding) Endocytosis
Monocytes HS, DC-SIGN Endocytosis
B cells HS, DC-SIGN Endocytosis
Role of xCT as a KSHV Entry Receptor
====================================
Kaleeba and Berger ([@B31]) identified the 12-transmembrane glutamate/cysteine exchange transporter protein xCT as a fusion-entry receptor in adherent cells. Ectopic expression of xCT rendered the non-susceptible adherent target cells to become susceptible to KSHV infection (Kaleeba and Berger, [@B31]). xCT is part of the cell surface 125-kDa disulfide linked heterodimeric membrane glycoprotein CD98 (4F2 antigen) complex containing a common glycosylated heavy chain (80-kDa) and a group of 45 kDa light chains. The xCT molecule is one of the light chains (Fenczik et al., [@B15]; Feral et al., [@B16]; Kabir-Salmani et al., [@B30]; Veettil et al., [@B64]). CD98 was initially identified as a molecule associated with integrin α3 and plays multiple roles including amino acid transport, cell adhesion, fusion, proliferation, and integrin activation. It is interesting to note that CD98 and integrin α3 were identified as fusion regulation protein 1 (FRP-1) and FRP-2, respectively, as this interaction was shown to play crucial roles in cell--cell fusion and virus-induced cell fusion (Fenczik et al., [@B15]; Feral et al., [@B16]; Kabir-Salmani et al., [@B30]; Veettil et al., [@B64]). Studies show that xCT is a component of a multimolecular signaling complex formed during KSHV macropinocytosis in HMVEC-d cells (Veettil et al., [@B64]; Chakraborty et al., [@B12]). This suggests that xCT plays a role within a multimolecular complex that may help to regulate the signaling pathways associated with the endocytic pathway of the virus. It is possible that the xCT present in the multimolecular complex may initiate a distinct cellular signaling pathway to control the various events associated with KSHV infection. Since a direct interaction of KSHV envelope glycoproteins with xCT has not been shown as yet, the exact role of xCT in the multimolecular complex and its role in KSHV infection need to be investigated further.
Multimolecular Integrin Complex Formation during Early KSHV Infection
=====================================================================
CD98 also mediates membrane clustering, β1 integrin-mediated signaling events, and stimulation of α3β1-dependent adhesion of cells and signal transduction cascade of αVβ3 integrin (Fenczik et al., [@B15]; Feral et al., [@B16]; Kabir-Salmani et al., [@B30]; Veettil et al., [@B64]). Co-immunoprecipitation and immunofluorescence studies in KSHV infected HMVEC-d cells have shown a time-dependent interaction of CD98/xCT with integrins α3β1, αVβ5, and αVβ3 (Veettil et al., [@B64]). Three different time-dependent temporal patterns of association and dissociation of KSHV interactions with cell surface molecules were observed. Integrin αVβ5 interaction with CD98/xCT predominantly occurred by 1 min post-infection (p.i.) and dissociated at 10 min p.i., whereas α3β1--CD98/xCT interaction was maximal at 10 min p.i. and dissociated at 30 min p.i., and αVβ3--CD98/xCT interaction was maximal at 10 min p.i. and remained at the observed 30 min p.i. Confocal microscopy studies confirmed the association of CD98/xCT with α3β1 and KSHV. Studies also showed that α3β1--CD98/xCT interactions could be inhibited by pre-incubating KSHV with soluble heparin and α3β1 strongly suggesting that KSHV's first contact occurs with HS, and integrins are essential elements in subsequent CD98--xCT interactions (Veettil et al., [@B64]). These studies demonstrated temporal interactions of KSHV with a family of functionally related proteins such as HS, integrins, and CD98--xCT molecules in endothelial, epithelial, and fibroblast cells (Wang et al., [@B66], [@B67]; Akula et al., [@B4]; Veettil et al., [@B64]).
Biological Relevance of Utilizing Multiple Entry Receptors by KSHV
==================================================================
It is fascinating to note that KSHV uses multiple molecules to enter target cells. The presence of multiple receptors is evolutionarily advantageous for KSHV and one of the major reasons for the broad tropism of the virus. Although virus--receptor interactions are highly specific, the affinity of interactions are low (Marsh and Helenius, [@B42]). Hence, KSHV binding to multiple receptors (mainly integrins) possibly increases binding avidity that leads to receptor clustering, a key step to activate signaling pathways. It is widely known that integrins are a signaling hub for a variety of processes including adhesion, motility, and endocytosis.
Deciphering the interactions between entry receptors and signaling leading to active endocytosis of viruses has been an active area of research for the past decade. However, in KSHV biology, recent advancements have revealed that receptor recruitment by KSHV is tightly succeeded by an active modulation of host cell membranes involving actin--myosin and other cellular signaling (Raghu et al., [@B52]; Valiya Veettil et al., [@B63]; Chakraborty et al., [@B12]). Detailed understanding of how KSHV induced such changes in the cell membrane very early during infection comes from its ability to manipulate a variety of host cell molecules to regulate signaling. The usage of multiple receptors possibly adds greater gradient to the succeeding signaling events required for efficient entry into target cells.
KSHV-Induced Signaling during Early Infection
=============================================
Many viruses, including KSHV make use of host cell signaling to enter target cells and establish productive infection. Binding and interaction of KSHV glycoproteins with integrins and other cellular receptors initiates intracellular signaling cascades to induce internalization. This internalization appears to be very rapid, reaching a peak by 60-min p.i. (Krishnan et al., [@B34]) and therefore signaling induced by the virus is very active within this time frame. Integrins are closely associated with FAK and Src kinases and is known to regulate them through a variety of other pathways. Since KSHV utilized a variety of integrins as its cellular receptor, it is evident that the virus has evolved ways to induce the integrin associated signaling.
Kaposi's sarcoma associated herpesvirus induces the phosphorylation of FAK and the subsequent phosphorylation of a variety of focal adhesion associated signal molecules such as Src, PI3-K, Rho-GTPases (RhoA, Rac, and Cdc42), and diaphanous-2, as well as several other downstream effector molecules, producing actin rearrangements that eventually lead to the internalization of KSHV. Induction of these signaling pathways is critical for active internalization of KSHV in target cells. The fact that a low multiplicity of infection is able to induce a sequential cascade of signaling events is a unique feature of KSHV infection. Several studies have elucidated that a low dose of KSHV (5--10 DNA copies/cell) is able to induce signaling events critical for entry, expression of viral genes, and latent infection (Sharma-Walia et al., [@B59]; Veettil et al., [@B64]; Valiya Veettil et al., [@B63]; Chakraborty et al., [@B12]). KSHV induces macropinocytic blebs as early as 1 min p.i., engage clustering of integrins, activate adaptors like c-Cbl, mediate receptor translocations, and induce ERK1/2 for establishing latent infection (Sharma-Walia et al., [@B59]; Veettil et al., [@B64]; Valiya Veettil et al., [@B63]; Chakraborty et al., [@B12]).
Integrin Associated Signaling: Focal Adhesion Kinases and Their Relevance in KSHV Entry
=======================================================================================
Focal adhesions are multifunctional organelles that mediate cell--ECM adhesion, force transmission, cytoskeletal regulation, and signaling. Focal adhesions consist of a complex network of trans-plasma-membrane integrins and cytoplasmic linking of the ECM to the actin cytoskeleton (Riveline et al., [@B55]). FAK, a major component of focal adhesions, is a multidomain non-receptor tyrosine kinase involved in signaling downstream of integrins. Ligand interaction with integrins activates FAK by autophosphorylation at tyrosine 397, a key step initiating integrin outside-in signaling (Calderwood et al., [@B10]; Giancotti, [@B21]). Activated FAK associates with a lot of other signal molecules to regulate a broad range of functions including cell growth, endocytosis, and apoptosis (Giancotti, [@B21], [@B22]).
Kaposi's sarcoma associated herpesvirus infection induced tyrosine phosphorylation of FAK within minutes of infection in HMVEC-d, HFF, 293, and FAK +/+ mouse Du17 fibroblasts (Akula et al., [@B4]; Naranatt et al., [@B45]; Wang et al., [@B67]; Sharma-Walia et al., [@B60], [@B59]; Veettil et al., [@B65]; Raghu et al., [@B51]). Soluble glycoprotein gB also induced FAK autophosphorylation (Wang et al., [@B67]; Sharma-Walia et al., [@B60]) and the phosphorylated FAK colocalizes with Src, RhoA, and cytoskeletal proteins like vinculin and paxillin in the infected cells (Akula et al., [@B4]; Wang et al., [@B67]; Sharma-Walia et al., [@B60]; Veettil et al., [@B65]). Additionally, virus or gB pre-incubated with soluble α3β1 integrin or a soluble form of gB in which the RGD sequence had been mutated inhibited the activation of FAK (Akula et al., [@B4]; Wang et al., [@B67]; Sharma-Walia et al., [@B60]). Since FAK activation is a hallmark of integrin-mediated signaling, FAK knockout cells or cells transfected with integrin have been used to study the role of FAK in integrin-mediated signaling. KSHV infection studies with FAK negative (Du3) and FAK positive (Du17) mouse fibroblasts and CHO cells transfected with human α3 integrin demonstrated a significant role of FAK in KSHV infection (Naranatt et al., [@B45]; Krishnan et al., [@B35]). Heparin inhibited the binding of KSHV in both DU3 and DU17 cell types. FAK negative Du3 cells showed an approximately 70% reduction in KSHV DNA internalization and over-expression of FAK (Du17) increased viral DNA internalization, thus suggesting that FAK plays a significant role in signaling and KSHV entry. In addition to this, over-expression of FAK-related non-kinase (FRNK), a FAK dominant-negative inhibitor, significantly decreased KSHV entry in DU17 cells. Decreased viral entry, nuclear delivery, and viral gene expression in Du3 cells suggest that another protein may be able to compensate for FAK's function early during infection. KSHV infection in Du3 cells induced the phosphorylation of the FAK-related proline-rich tyrosine kinase (Pyk2) molecule, which demonstrated that the Pyk2 molecule compensate for the loss of some of the functions of FAK in FAK negative cells during KSHV infection. Moreover, inhibition of Pyk2 by an autophosphorylation mutant of Pyk2 also significantly reduced viral entry in DU3 cells (Krishnan et al., [@B35]). Since activation of FAK plays a central role in integrin-mediated signaling, rearrangement of actin and endocytosis, KSHV must have evolved to take advantage of these signaling pathways both to promote entry and the subsequent steps of infection (Figure [1](#F1){ref-type="fig"}).
Role of Src and PI3-Kinases in KSHV Entry and Infection
=======================================================
The autophosphorylation site of FAK (Tyr397) creates a binding site for the SH2 domain of Src kinases and the p85 subunit of PI3-K. KSHV infection induced a strong phosphorylation of Src within minutes of infection, and the phosphorylated Src colocalized with FAK (Veettil et al., [@B65]). KSHV-gB also induced the FAK dependent Src phosphorylation in adherent target cells (Sharma-Walia et al., [@B60]) and the activated Src kinases then phosphorylate a number of FA components. Furthermore, Src is required for the activation of PI3-K and other downstream targets such as Rho-GTPases. The critical role of Src in the KSHV entry process has been validated by multiple lines of evidences such as the failure of KSHV to enter Src negative mouse fibroblast cells (unpublished observation), increase in Src activity by LR disruption resulting in enhanced virus entry (Raghu et al., [@B51]). Another observation showed that RhoA-GTPase facilitated KSHV entry into adherent target cells in a Src-dependent manner. This study suggests that KSHV-induced Src is involved in RhoA activation, which in turn results in positive feedback activation of Src to increase viral entry.
PI3-Kinases (PI3-K) are heterodimeric proteins, consisting of a p85 regulatory subunit and a catalytic p110 subunit while phosphorylation of specific tyrosine residues on the p85 subunit is an indication of PI3-K activation. PI3-kinases play crucial roles in several signaling pathways and regulate multiple functions such as Rho-GTPase activation, apoptosis, survival, and migration (Giancotti, [@B21]; Sastry and Burridge, [@B57]). KSHV induces PI3-K within 5 min p.i. which decreased after 15 min and this response can be inhibited either by pre-incubating virus with integrin or by the PI3-K inhibitors wortmannin and LY294002 (Naranatt et al., [@B45]). LY294002 and wortmannin did not affect KSHV-gB induced Src phosphorylation whereas the Src kinase inhibitor SU6656 completely blocked KSHV-gB induced p85-PI3-K phosphorylation suggesting that PI3-K is downstream of Src (Sharma-Walia et al., [@B60]). In FAK positive Du17 cells, KSHV-gB induced PI3-K p85 phosphorylation whereas in FAK-null Du3 cells there was no significant induction of PI3-K p85 phosphorylation. This study suggests that FAK is an essential molecule for the induction of PI3-K during KSHV infection. Treatment of cells with PI3-K inhibitors wortmannin and LY294002 reduced PI3-K activation and viral entry in a concentration dependent manner suggesting the role of PI3-K activation in the entry of KSHV in target cells. Induction of PI3-K eventually leads to the induction of Rho-GTPases and their effectors (Sharma-Walia et al., [@B60]; Veettil et al., [@B65]) which in turn regulates the remodeling of actin, endosome formation, and the movement of endocytic vesicles. Therefore, these studies using chemical inhibitors, dominant-negative proteins, or cells lacking these molecules have demonstrated that FAK, Src, and PI3-K activation were necessary for KSHV entry.
Activation of RhoA and Actin Dynamics during KSHV Infection
===========================================================
RhoA, Rac, and Cdc42-Rho-GTPases are master regulators of a diverse set of signaling pathways, including cytoskeleton rearrangement and morphological changes (Giancotti, [@B21]; Hall and Nobes, [@B26]; Ishizaki et al., [@B28]; Palazzo et al., [@B49]). Immediately following infection, KSHV induces PI3-K Rho-GTPase-dependent cytoskeletal rearrangements and the formation of structures such as filopodia (Cdc42), lamellipodia (Rac), and stress fibers (RhoA) in the target cells (Naranatt et al., [@B45]; Veettil et al., [@B65]; Greene and Gao, [@B23]; Raghu et al., [@B52]). Soluble gB induced the FAK-Src-PI3-K Rho-GTPase signaling pathway and extensive cytoskeletal rearrangement in target cells (Sharma-Walia et al., [@B60]). KSHV-induced RhoA colocalized with Src in the infected cells (Sharma-Walia et al., [@B60]). Ezrin, an actin cross linking protein with the plasma membrane, was also induced by KSHV via Rho-GTPases, thereby modulating membrane changes (Sharma-Walia et al., [@B60]). Treatment of target cells with a potent RhoA inhibitor, *Clostridium difficile* toxin B (CdTxB), or transfecting dominant-negative constructs of RhoA resulted in significant inhibition of KSHV entry by modulation of Src activity (Veettil et al., [@B65]).
Several studies have demonstrated that RhoA-GTPases mediate rearrangement of cytoskeleton through the activation of its downstream effector molecules, formin family diaphanous 1 and 2 (Hall and Nobes, [@B26]; Ishizaki et al., [@B28]; Palazzo et al., [@B49]). KSHV infection induced diaphenous-2 without any significant activation of Rac-1 and Cdc42-dependent PAK1/2 and stathmin molecules. Dia-2 co-immunoprecipitated and colocalized with activated Src in the infected cells which were inhibited by Src inhibitors (Veettil et al., [@B65]). Together with the reduced virus entry in RhoA dominant-negative cells, these results suggest that activated RhoA-dependent Dia-2 probably functions as a link between RhoA and Src in KSHV infected cells, mediates the sustained Src activation, and that KSHV-induced Src and RhoA play roles in facilitating entry and nuclear delivery of viral DNA.
KSHV-Induced Adaptor Molecule c-Cbl and Associated Membrane Dynamics
====================================================================
The Cbl family of adaptor proteins plays important roles in signal transduction as negative regulators by mediating the ubiquitination and down-regulation of proteins while it acting as a positive regulator through their scaffold function in assembling signaling complexes (Thien and Langdon, [@B62]; Schmidt and Dikic, [@B58]). Recent evidences have elucidated that KSHV-induced c-Cbl tyrosine phosphorylation is required for membrane ruffling known as blebs (Valiya Veettil et al., [@B63]). Blebs are protrusions from the plasma membrane that have often been characterized as a preceding step in macropinocytosis and are known to be induced by many viruses including vaccinia virus and KSHV early during infection (Mercer and Helenius, [@B43]; Raghu et al., [@B52]; Valiya Veettil et al., [@B63]). The role of c-Cbl in the entry and signaling of any viral infection has not been reported. Hence it opens up new directions of studies on the broad cellular requirements required by viruses for infecting target cells.
Kaposi's sarcoma associated herpesvirus is known to induce a variety of endothelial cell membrane alterations including filopodia and bleb formation (Raghu et al., [@B52]; Valiya Veettil et al., [@B63]). These studies elegantly deciphered the mechanism behind KSHV-induced membrane bleb formation. KSHV infection increased c-Cbl interaction with PI3-K in a time dependant manner. In KSHV infected cells, activated c-Cbl is recruited to the macropinocytic blebs and associated with its novel interacting partner, myosin IIA inside the membrane blebs very early by 5 min p.i. Studies using shRNA against c-Cbl not only reduced the macropinocytic blebs induced by KSHV, but also significantly reduced viral entry by \>70% and hence KSHV gene expression by 90% (Valiya Veettil et al., [@B63]). This study demonstrates that recruitment of c-Cbl--myosin in macropinocytic blebs very early during KSHV infection is essential for its entry via macropinocytosis in endothelial cells. Although many viruses utilize macropinocytosis to enter target cells, the mechanisms behind it were unclear (Mercer and Helenius, [@B43]); however, the role c-Cbl and myosin IIA in macropinocytic modulation of KSHV potentiate new avenues of interest in the phenomena of membrane dynamicity induced by viruses.
Role of Lipid Rafts: A Signaling Platform Utilized by KSHV in Endothelial Cells
===============================================================================
Lipid rafts, the detergent resistant microdomains in the exoplasmic leaflet of plasma membranes, are made up of cholesterol and sphingolipids (sphingomyelin and glyco-sphingolipids) and play roles in clustering cell surface receptors and signal molecules (Simmons, [@B61]). Lipid microdomains favor specific protein--protein interactions including ligand receptor interactions activating signal cascades. KSHV binding was not affected upon disruption of LRs by methyl beta cyclodextrin or nystatin, but pre-treatment of HMVEC-d with these LR disrupting drugs significantly reduced the expression of ORF73 and ORF50 (Raghu et al., [@B51]). Although internalization of viral DNA was increased, the association of internalized viral capsids with microtubules was reduced upon LR disruption, microtubules disorganized, and nuclei associated viral DNA decreased (Raghu et al., [@B51]).
Mechanistic studies on the role of LRs in KSHV infection revealed more intricate details with relation to entry receptors and signal molecules. A recent study reported that very early during infection (1 min p.i.), c-Cbl induced the selective translocation of KSHV into LRs along with α3β1, αVβ3, and xCT receptors, but not αVβ5 (Chakraborty et al., [@B12]). Evidence also suggests that activated c-Cbl localized to LRs at the junctional base of macropinocytic blebs, thereby aiding macropinocytosis (Chakraborty et al., [@B12]). Such partitioning of entry receptors in LRs by specific signaling adaptors was not reported before and evokes new insights to the molecular mechanisms of viral--host cell interactions, often favoring viral internalization.
The role of LRs is complex and affects KSHV-induced signaling. For instance Raghu et al. ([@B51]) reported that LR disruption affects signal pathways induced by KSHV such that phospho-Src levels had increased without affecting FAK or ERK1/2. However, KSHV-induced PI3-K, Rho-GTPases, and NF-κB activation were significantly reduced. The fact that p-Src increased with LR disruption is indicative of a strong regulation of Src by LRs. LR disruption also affects PI3-K and RhoA with subsequent reduction in KSHV-induced RhoA mediated acetylation and aggregation of MTs (Raghu et al., [@B51]).
Cellular signaling can be generated in many ways. Viruses most often activate cellular signaling directly by using receptors or induce signaling by clustering specific cell-surface proteins or lipids (Marsh and Helenius, [@B42]). KSHV serves an excellent model system in this regard as it utilizes both pathways; firstly by activating associated integrin signaling and secondly by clustering activated integrins and signal molecules in LRs, generating significant amplification of the signaling response; however, the molecular partners behind such signaling amplification are yet to be studied.
Receptor Ubiquitination: A Critical Step Required for Productive KSHV Internalization
=====================================================================================
Ubiquitination of receptors has been recognized as an internalization signal based on the nature and type of ubiquitin modifications (Levkowitz et al., [@B37]; Dupin et al., [@B14]). Studies have revealed an important role of c-Cbl, an E3-ubiquitin ligase, in differential ubiquitination of KSHV integrin receptors (Chakraborty et al., [@B12]). Essentially, LR translocated integrins (α3β1 and αVβ3) were monoubiquitinated leading to productive macropinocytic entry, whereas non-LR associated αVβ5 was polyubiquitinated leading to clathrin mediated entry that was targeted to lysosomes, the non-infectious pathway (Chakraborty et al., [@B12]). This elucidates the complexities of viral endocytic mechanisms and the ability of KSHV to utilize E3-ubiquitin ligases to regulate and sort out productive pathways.
Diverse Internalization Pathways of KSHV in Target Cells
========================================================
Current evidences show that KSHV enters human B cells (Rappocciolo et al., [@B53]), fibroblast (Akula et al., [@B1]), epithelial (Inoue et al., [@B27]; Liao et al., [@B38]), and endothelial cells (Raghu et al., [@B52]) by endocytosis. KSHV was detected by electron microscopy in large endocytic vesicles within 5 min of HMVEC-d and HFF cell infection while fusion of virion envelope with the endocytic vesicles was also observed (Akula et al., [@B2], [@B1]). Viral capsids were detected in the vicinity of the nuclear membrane by 15 min p.i. and anti-KSHV-gB and gpK8.1A antibodies colocalized with virus-containing endocytic vesicles (Akula et al., [@B2]; Greene and Gao, [@B23]; Raghu et al., [@B52]; Table [2](#T2){ref-type="table"}).
The roles of different endocytic pathways have been studied using specific inhibitors of each type. Clathrin-mediated endocytosis is the predominant pathway of entry in HFF cells. Chlorpromazine, an inhibitor of clathrin-mediated endocytosis, significantly inhibited entry in HFF cells, whereas nystatin, an inhibitor of caveolae and cholera toxin B, a LR inhibiting agent did not have any effect on entry. A significant inhibition of gene expression was also observed after blocking endosomal acidification by NH~4~Cl and bafilomycin A in HFF cells (Akula et al., [@B1]; Raghu et al., [@B52]) suggesting that post internalization endosomal acidification was required for KSHV trafficking. In addition to this, electron microscopy revealed KSHV virions in large endocytic vesicles within 5-min of HFF cell infection and fusion of virion envelope with endocytic vesicles was also observed (Akula et al., [@B1]; Raghu et al., [@B52]). Evidence for clathrin-mediated endocytosis has also been observed in BJAB and 293 cells (Akula et al., [@B2]; Inoue et al., [@B27]). KSHV enters THP-1 cells and primary monocytes by clathrin and caveolin dependant endocytosis, which required endosomal acidification (Kerur et al., [@B32]).
In HMVEC-d and HUVEC cells, entry and gene expression of KSHV were significantly blocked by macropinocytosis inhibitors EIPA and rottlerin. Macropinocytosis of KSHV is an actin dependant endocytic pathway and it was also inhibited by cytochalasin D (Raghu et al., [@B52]). Cytochalasin D inhibited actin polymerization and formation of lamellipodial extensions significantly inhibited the entry and expression of KSHV (Naranatt et al., [@B45]; Greene and Gao, [@B23]; Raghu et al., [@B52]). The LR inhibiting agents reduced viral gene expression in HMVEC-d cells but not in HUVEC or HFF cells, indicating the role of LRs in KSHV infection (Greene and Gao, [@B23]; Raghu et al., [@B52]). Colocalization studies using clathrin-mediated endocytosis marker, transferrin and macropinocytosis marker, dextran showed significant association of KSHV with dextran but not with transferrin or the caveolar marker caveolin. The dynamin inhibitor, dynasore, did not affect viral entry into endothelial cells while inhibiting entry into HFF cells (Raghu et al., [@B52]). The small GTPase Rab34, a key regulator of macropinocytosis, associated with KSHV and Rab34-siRNA considerably decreased KSHV gene expression (Raghu et al., [@B52]). These studies suggested that KSHV utilizes the actin polymerization-dependent, dynamin-independent macropinocytic pathway involving a Rab34 GTPase-dependent late endosome and low-pH environment for its infectious entry into HMVEC-d and HUVEC cells (Table [2](#T2){ref-type="table"}).
Recent studies have confirmed the role of LRs in aiding the macropinocytosis of KSHV bound receptors in endothelial cells, whereas non-LR bound receptors were targeted toward a clathrin mediated non-infectious lysosomal pathways (Chakraborty et al., [@B12]). This study suggests that both macropinocytosis and clathrin-mediated endocytosis are occurring in endothelial cells, however the former being productive while the latter is degradative. KSHV internalized by clathrin-mediated endocytosis clearly associated with lysosomal compartments (Chakraborty et al., [@B12]). Another report suggested that clathrin-mediated endocytosis is the predominant pathway of entry in endothelial cells. This discrepancy could be due to optimal concentration of the inhibitor used and the method of quantification used to analyze the entry of KSHV (Greene and Gao, [@B23]). The detection of KSHV by ORF65 (KSHV capsid protein) in perinuclear regions of HUVECs does not necessarily correlate with a productive endocytosis pathway and hence needs further clarification (Table [2](#T2){ref-type="table"}).
Some evidences have also revealed the role of fusion of KSHV envelope glycoproteins with target cell membranes (Akula et al., [@B2]; Wang et al., [@B66], [@B67]; Naranatt et al., [@B44]). The minimal fusion machinery of KSHV probably comprises of gB, gH, and gL since anti-gB, gH, gL, and gpK8.1A antibodies neutralize KSHV infection without affecting virus binding to the target cells (Akula et al., [@B2]; Wang et al., [@B66], [@B67]; Naranatt et al., [@B44]). However the mechanism of neutralization is not known. Multiple mechanisms could be possible depending upon the glycoprotein and the targeted region of the glycoproteins. For example, neutralization could be due to: (a) interference in the interaction of KSHV glycoproteins with integrins and other receptors; (b) interference at the activation of a subset of signal molecules that are essential for the c-Cbl mediated receptor translocation, ubiquitination, bleb formation, and macropinocytosis; and (c) interference of fusion of viral envelope with the endosomal membrane. Further studies need to be carried out to precisely define the mechanism behind the neutralization by these antibodies. Nevertheless, these studies suggested that these glycoproteins play critical roles in the entry process after attachment has occurred, possibly as a result of interaction with additional cell surface molecules.
Post Entry Steps: KSHV Trafficking and Productive Infection of KSHV
===================================================================
Like most herpesviruses, KSHV replicates in the nucleus of infected cells. Therefore, to reach the nucleus KSHV capsids traffic through the crowded cytosol before releasing viral DNA into the nucleus (Lyman and Enquist, [@B40]). Delivery of KSHV DNA into the infected cell nucleus reached a peak by 90 min p.i. suggesting the rapidness of the host cellular trafficking utilized by the virus (Naranatt et al., [@B45], [@B46]; Krishnan et al., [@B34]; Veettil et al., [@B65]; Raghu et al., [@B52]).
Productive infection comprised of KSHV internalized by macropinocytosis in endothelial cells or by clathrin-mediated endocytosis in fibroblast cells utilize the extensive microtubule (MT) network to traffic through the cytosol, tightly regulated by Rho-GTPases. Microtubule bundles colocalized with KSHV capsids and this colocalization was abolished by the microtubule destabilizing agent, nocodazole, and PI3-K inhibitor affecting the Rho-GTPases (Naranatt et al., [@B45], [@B47]). Depolymerization of microtubules did not affect KSHV binding and internalization, but nuclear delivery of viral DNA and infection in HFF, HMVEC-d, and HUVEC cells was inhibited (Naranatt et al., [@B46]; Veettil et al., [@B65]; Raghu et al., [@B52]). Inhibition of Rho-GTPase activities by CdTxB abolished microtubular acetylation and subsequently the delivery of viral DNA to the nucleus. Conversely, activation of Rho-GTPases by *Escherichia coli* cytotoxic necrotizing factor significantly increased the intracellular trafficking and delivery of viral capsids to the nucleus. Similarly, nuclear delivery of viral DNA was increased in cells expressing a constitutively active RhoA mutant and decreased in cells expressing a dominant-negative mutant of RhoA (Naranatt et al., [@B46]). Taken together, these studies indicate that KSHV induces Rho-GTPases, modulates stabilization of microtubules and promotes the rapid trafficking of viral capsids toward the nucleus (Table [3](#T3){ref-type="table"}).
######
**Host molecules and their roles in KSHV entry/infection in HMVEC-d and HFF cells**.
Molecules Function in KSHV biology
------------------------------ ----------------------------------------------------------------------------------------
Heparan sulfate (HS) Binding receptor
Integrins (α3β1, αVβ3, αVβ5) Entry receptors
xCT/CD98 Entry (fusion) receptor
FAK Initiate endocytosis
Src Initiate endocytosis
PI3-K, RhoA, Dia-2, Ezrin Entry, actin modulation, endocytosis, and nuclear trafficking
c-Cbl KSHV-receptor translocations into LR and adaptor for macropinocytosis in HMVEC-d cells
ERK and NF-κB Viral and host gene expression
Kaposi's sarcoma associated herpesvirus utilizes ATP dependant directional transport along microtubules governed by motor proteins. Dynein motor proteins responsible for minus-end transport from the periphery to the cell center actively carry KSHV to the vicinity of the nuclear membrane. Inhibition of dynein activity by sodium orthovanadate significantly reduced the infection and delivery of KSHV DNA into the nucleus (Naranatt et al., [@B46]). These studies demonstrate that KSHV movement across the cytoplasm to reach the nucleus is a series of well-orchestrated phenomenon probably involving viral proteins and host proteins. Further studies are essential to define this process and the mediating viral and host proteins.
Cross talk between ECM and integrins activates FAK, which initiates a cascade of intracellular signals that eventually activate the mitogen activated protein kinase (MAPK) pathways. As early as 5 min p.i., KSHV activates MEK (MAPK/ERK kinase) and extracellular-signal-regulated kinase (ERK; Naranatt et al., [@B46]). Soluble KSHV gpK8.1A, but not gB, induced MEK-mediated ERK1/2 phosphorylation as early as 5 min p.i., and ERK1/2 phosphorylation facilitated the establishment of KSHV infection in HFF and HMVEC-d cells (Sharma-Walia et al., [@B59]). PI3-K and protein kinase C-ζ (PKC-zeta) are recruited as upstream mediators of the KSHV ERK pathway and inhibitors specific for PI3-K, PKC-ζ, MEK, and ERK significantly reduce virus infectivity without affecting virus binding to the target cells (Table [3](#T3){ref-type="table"}).
Rapid activation of NF-κB as early as 5--15 min p.i. led to the translocation of p65-NF-κB into the nucleus (Sadagopan et al., [@B56]). KSHV incubated with heparin significantly reduced NF-κB activation. During the observed 72 h periods of *in vitro* KSHV latency, a sustained moderate level of NF-κB induction was observed, and inhibition of IκB phosphorylation by Bay11-7082 drastically reduced this activation (Sadagopan et al., [@B56]). In contrast, high levels of ERK1/2 activation during the earlier time points and a moderate level of activation at later time points were observed. The p38-MAP kinase was activated only during the later time points, and AKT was activated at lower levels in a cyclic manner. Studies with UV-KSHV suggested a role for virus entry stages in NF-κB induction and requirement of KSHV viral gene expression for sustained induction. Though inhibition of NF-κB did not have any effect on KSHV entry into cells, expression of viral latent ORF 73 and lytic ORF 50 genes was significantly reduced. Several transcription factors were activated during KSHV infection, and inhibition of NF-κB significantly affected the activation of Jun D, Jun B, phospho-c-Jun, cFos, and FosB factors. These results suggested that during *in vitro* infection, KSHV induces sustained levels of NF-κB to regulate viral genes thus possibly regulating the establishment of latent infection.
Productive KSHV primary infection in adherent cells involves an initial lytic phase (2 h p.i.) succeeded by a latent phase (24 h p.i.) with a decline of lytic phase (Krishnan et al., [@B34]). A report by Yoo et al. ([@B68]) suggests that KSHV infection is permissive in HUVECs at early time points with an initial production of infectious virus particles (lytic cycle), while the surviving cells later enter a latent phase with spontaneous lytic replication. However, the percentage of cells that were infected, cells that entered into lytic cycle and cells that entered latency were not determined in these studies. Hence, these studies need to be reexamined.
Perspectives and Future Directions
==================================
Kaposi's sarcoma associated herpesvirus exemplifies an excellent model for viruses requiring multiple cellular molecules to enter target cells. Although, integrins and associated signaling, adaptor molecules and a host of pre-existing signaling molecules have been identified to play a crucial role in KSHV entry and infection, the nature and mechanism by which the virus is able to successfully utilize the sequential series of host signaling still remains a mystery. Moreover, there is a difference in the entry pathways of KSHV in infectable cell types, making this interesting story more complicated. What is clear from studies so far is that KSHV overcomes cellular barriers making conditions conducive to infection by utilizing more than one endocytic mechanism. However, further studies need to be performed to ensure whether differential activation of signal responses by KSHV coupled to different endosomal vesicles accounts for the varied modes of entry. Moreover, whether KSHV utilizes more than one E3-ubiquitin ligase and their modulation by viral induced cellular targets remains to be explored. Another interesting future study involves the recognition of KSHV in the endosomal vesicles that direct productive trafficking of virus away from the lysosomal compartments. Cellular signaling targets of KSHV are broad and hence critical molecules regulating such multiple pathways need to be targeted for therapeutics. These studies will shed valuable insights into the molecular mechanisms of cytosolic entry of KSHV.
Conflict of Interest Statement
==============================
The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
This study was supported in part by Public Health Service grants (NIH) CA 075911 and CA 168472, and the Rosalind Franklin University of Medicine and Science H. M. Bligh Cancer Research Fund to Bala Chandran. We thank Keith Philibert for critically reading the manuscript.
[^1]: Edited by: Keiji Ueda, Osaka University Graduate School of Medicine, Japan
[^2]: Reviewed by: Keiji Ueda, Osaka University Graduate School of Medicine, Japan; Frank Neipel, Universitaet Erlangen, Germany
[^3]: This article was submitted to Frontiers in Virology, a specialty of Frontiers in Microbiology.
| {
"pile_set_name": "PubMed Central"
} |
###### Summary box
What is already known about this subject?
=========================================
1. Current treatment of acute gastroenteritis as recommended by the WHO includes the use of oral rehydration solution and zinc.
2. Studies carried out on racecadotril in children have proved its efficacy in reducing duration and severity of diarrhoea.
3. The European Society of Paediatric Gastroenterology, Hepatology and Nutrition has stated that racecadotril may be used in the management of acute gastroenteritis in children.
What are the new findings?
==========================
1. Racecadotril did not significantly impact the duration or severity of diarrhoeal disease in children with severe gastroenteritis.
2. One of the first studies using racecadotril to treat children in East Africa.
How might it impact on clinical practice in the foreseeable future?
===================================================================
1. Defining which populations of children with acute gastroenteritis would have a maximum benefit from racecadotril. This could allow recommendations on its use to be refined.
Introduction {#s1}
============
Diarrhoea is one of the most common causes of morbidity and mortality in children under 5 years in the developing world. Kenya is ranked in the top 10 countries bearing the burden of high mortality rates in children suffering from diarrhoeal disease with a prevalence of 16.7%.[@R1] [@R2] In the 1970s--1980s, trends in mortality showed a significant decrease with the advent of oral rehydration solution (ORS). Since then this decline has slowed with current data at 530 000 deaths per year.[@R3]
Racecadotril is a diesterified derivative of thiorphan.[@R4] Once absorbed after oral administration, it is converted to its parent compound (thiorphan) which acts to increase the half-life of enterocyte methionine-enkephalin (a potent antisecretory agent).[@R5] Racecadotril has been in use for the past 20 years in Europe and has been consistently useful in the management of diarrhoeal disease with an acceptable adverse effect profile.[@R6]
Although zinc has been shown to significantly affect the duration and severity of diarrhoea and forms part of the core treatment recommendations by the WHO, none of the trials studying efficacy of racecadotril have included it in their protocols.[@R7] [@R11]
Despite a number of studies performed in different parts of the world, there has been no published data involving racecadotril from the African continent. Of the few studies carried out in the developing world, majority have been inpatient studies with methodological flaws noted in many.[@R7]
This study was carried out to bridge gaps in knowledge about the use of racecadotril and its usefulness when used with zinc in African children with severe gastroenteritis.
Methodology {#s2}
===========
Study design {#s2a}
------------
This was a prospective randomised, double-blind, placebo-controlled, parallel-group study conducted at the Kenyatta National Hospital. This is a tertiary level, national referral hospital and has four dedicated paediatric wards. Participants were assigned to the control arm or treatment arm with a 1:1 ratio.
Eligibility criteria {#s2b}
--------------------
The inclusion criterion was: children aged 3--60 months requiring admission for severe acute gastroenteritis, as evidenced by a Vesikari score of \>11 ([figures 1](#BMJGAST2016000124F1){ref-type="fig"} and [2](#BMJGAST2016000124F2){ref-type="fig"}) with written parental consent.[@R12] [@R13]
![The Vesikari clinical severity scoring system parameters and scores. Table adapted from Vesikari Clinical Severity Scoring System Manual.[@R13]](bmjgast2016000124f01){#BMJGAST2016000124F1}
![Grading of severity of diarrhoea according to the Vesikari score. Table adapted from Vesikari Clinical Severity Scoring System Manual.[@R13]](bmjgast2016000124f02){#BMJGAST2016000124F2}
Exclusions were: children who had severe vomiting (scoring 3 on the Vesikari score for maximum number of vomiting episodes per day); those with a clinical diagnosis of dysentery or a known diagnosis of liver or renal failure; children who had prescriptions of probiotics or any other antidiarrhoeal medication (loperamide, attapulgite, activated charcoal, diphenoxylate and kaolin).
Interventions {#s2c}
-------------
All children were admitted and started on WHO recommended treatment by the attending physician. They received either intravenous fluids as per WHO plan C (30 mL/kg followed by 70 mL/kg over 1 and 5 hours, respectively, in infants and over 30 min and 2.5 hours for those over 12 months of age) or low osmolality ORS as per WHO plan B (75 mL/kg over 4 hours). Zinc was prescribed at 10--20 mg/day.
Participants were recruited within 24 hours of admission after initial hydration had been completed. Eligibility was determined by calculating the Vesikari score based on the information recorded by the primary physician in the patients\' chart. This included the patients\' history up to and the examination at the time of admission.
A computer program was used to generate random numbers in blocks with varying sizes. These were used to assign patients randomly to either the test or control arm of the study.
Both the drug and placebo were supplied by the manufacturer, Racedot---Macleods (India) and distributed by Sai Pharmaceuticals (Kenya). These were packed in tamper proof brown bags, sealed and labelled by a study pharmacist based off site.
The test arm received racecadotril at a dose recommended by the manufacturer: 10 mg per dose for children below 12 months of age and 30 mg for those over 12 months of age. This was administered as granules dissolved in 10 mL of water and taken three times a day for a maximum duration of 3 days. The control arm had a placebo preparation administered in the same way.
An initial dose of the drug or placebo was given at the time of enrolment. The parents were taught how to administer the drug at this time and this was confirmed by return demonstration.
The participants were followed up using daily interviews asking about: the number of stools, the consistency using a Bristol stool chart,[@R14] the presence of blood in stool, any new symptoms and the introduction of an antidiarrhoeal or other medication.
Treatment was given either until the stools were formed or for a total of 3 days, whichever came first.
Children who were discharged before complete resolution of symptoms were followed up by phone interviews until recovery.
Outcome measures {#s2d}
----------------
The primary outcome measure was the number of stools in the first 48 hours after introduction of the drug. The secondary outcomes included the duration of inpatient stay, the duration of illness and the number of adverse events associated with racecadotril.
Sample size calculation {#s2e}
-----------------------
Estimation of sample size was carried out using the formula for comparison of two sample means. The study was powered at 90% with a significance level of 95%. Assumptions for the formula were based on the study by Cojocaru *et al*[@R8] using his outcome of total number of stools at 48 hours. A provision of 10% was made for attrition and a final sample size of 60 participants in each arm was derived. The attached online [supplementary file](http://bmjopengastro.bmj.com/doi/suppl/10.1136/bmjgast-2016-000124) contains the complete sample size calculation and references.
Statistical methods {#s2f}
-------------------
Baseline data for the two groups were collected and analysed by comparing proportions to determine the presence of any significant differences. Once the data were collected and analysed, it was noted to be positively skewed. Medians with IQRs were calculated for the outcome measures and the two groups were compared using the Mann-Whitney U test according to intention-to-treat analysis. STATA V.11.0 software was used for the above.
Ethical consideration {#s2g}
---------------------
The study was conducted after the approval of the Ethics and Research Committees of the Aga Khan University Hospital and the Kenyatta National Hospital/University of Nairobi. The development of known side effects, suspected adverse effects or new symptoms were communicated to a Drug Safety Monitoring Board (DSMB). Adverse events were graded based on the Division of Allergy and Infectious Diseases (DAIDS) table for grading the severity of adult and paediatric adverse events summary sheet.[@R15]
Results {#s3}
=======
A total of 154 children with severe gastroenteritis were screened for the study during the data collection period from 3 February to 26 March 2014. The details of recruitment and allocation are shown in [figure 3](#BMJGAST2016000124F3){ref-type="fig"}. Analysis included all 60 participants in each group. Three participants from the drug group and two from the placebo group that did not complete the protocol were included in the analysis. All results for the outcome measures (except adverse events) are given as medians with IQRs.
{#BMJGAST2016000124F3}
Baseline characteristics {#s3a}
------------------------
There was a statistically significant predominance of girls in the drug group (p=0.04) despite an overall male prevalence in the study. This difference was believed to arise due to chance as randomisation was carried out according to protocol.
The difference in dehydration levels between the groups was not statistically significant (p=0.25). It was noted that most children in the study were mild to moderately dehydrated with severe dehydration present in only 15% of the drug group and 27% of the placebo group. This finding was attributed to use of the Vesikari score as an inclusion criteria. Similar results were found in the study carried out by Schnadower *et al*.[@R16] Baseline data are summarised in [table 1](#BMJGAST2016000124TB1){ref-type="table"}.
######
Table showing baseline characteristics of the participants enrolled in the trial and the p value between the drug and placebo groups to show presence of statistical significance
----------------------------------------------------------------------
Variable Racecadotril\ Placebo\ p Value\*
n (%) n (%)
------------------------------- --------------- ---------- -----------
Sex: male 27 (45) 38 (63)
Female 33 (55) 22 (37) 0.04†
Malnutrition
None 34 (57) 28 (47) 0.27
Mild to moderate 17 (28) 16 (27)
Severe 09 (15) 16 (27)
Duration of diarrhoea‡ (days)
≤4 31 (51) 32 (53) 0.97
5 15 (23) 14 (23)
≥6 14 (23) 14 (23)
Dehydration
No 3 (5) 4(6) 0.43
Some 26 (43) 32 (53)
Severe 31 (51) 24 (40)
----------------------------------------------------------------------
\*Pearson constant on χ^2^ test.
†Significant difference in sex ratio between the placebo and drug groups, suspected to be due to chance.
‡Duration of diarrhoea given as the number of patients (and %) who fell into each of the three groups as given by the Vesikari score.
Number of stools at 48 hours {#s3b}
----------------------------
The difference in the median number of stools on intention-to-treat analysis revealed no statistically significant difference. Median (IQR) 5[@R3] stools for the drug and 5 (2.5--7.5) for the placebo (p=0.63).
Duration of inpatient stay {#s3c}
--------------------------
This was measured as the number of days from the start of the medication to the day of discharge as determined by the attending physician. The duration was shorter in children in the drug group: median (IQR) of 4 (1.5--6) days as compared with placebo: 4.5 (1.8--6.3) days. The results were not statistically significant (p=0.71).
Duration of illness {#s3d}
-------------------
This was defined as the duration from the time of introduction of the drug to the appearance of ≤3 formed stools in 24 hours. This was noted to be lower in the placebo group: median (IQR) of 2[@R1] days and 3 (2.5--3.5) days for the drug. The results were not statistically significant (p=0.77).
Adverse events {#s3e}
--------------
The proportion of patients experiencing any adverse event was similar for the two groups (23%). There were four mortalities overall, two in each group. Convulsions were noted in three children who were on the drug (5%) but none in children on the placebo. All three children were investigated for concurrent meningitis and this diagnosis was confirmed in one. Overall, the proportions of children who had serious adverse reactions were higher in the racecadotril group, although this was not statistically significant: relative risk 3.0 (CI 0.63 to 14.27); p=0.16. Non-serious adverse events: relative risk 0.81 (CI 0.36 to 1.83); p=0.62. The adverse events are listed in [table 2](#BMJGAST2016000124TB2){ref-type="table"}.
######
Table showing the number of adverse events in the drug and placebo groups, respectively
Event Drug group (frequency) Placebo group (frequency)
------------------------ ------------------------ ---------------------------
Skin rashes 6 8
Pruritus 0 0
Angiooedema 2 1
Tonsilitis 1 3
Mortality\* 2 2
Others (convulsions\*) 3 0
Total 14 14
\*Severe adverse events as given by the DAIDS table. No statistical difference between the drug and placebo group: relative risk 3.0 (CI 0.63 to 14.27); p=0.16.
DAIDS, Division of Allergy and Infectious Diseases.
Analysis of the data using means and SD, with comparison using the Student's t-test, was also carried out. The results showed no significant difference. Linear regression analysis revealed malnutrition was a significant factor for duration of in-hospital stay. This analysis is presented in the online [supplementary file](http://bmjopengastro.bmj.com/doi/suppl/10.1136/bmjgast-2016-000124).
Discussion {#s4}
==========
This study, carried out in children with severe acute gastroenteritis who were admitted to hospital and on treatment with zinc and ORS, compared the impact of racecadotril to a placebo. There was no difference noted between the group on racecadotril and that on placebo for the outcomes measured. While this result is consistent with that seen by Santos *et al*[@R17] in their study carried out in 2009, it differs from most other studies carried out on the subject.[@R7] [@R8] [@R18]
The Vesikari score made up part of the inclusion criteria for our study, thus ensuring all children recruited had severe gastroenteritis. This is a composite score that was originally developed by Ruuska and Vesikari.[@R12] [@R13] It incorporates the patient\'s temperature, level of dehydration, need for hospitalisation as well as the duration and frequency of vomiting and diarrhoea. Although the Vesikari score was originally used to determine severity of gastroenteritis at the conclusion of illness, it was shown by Schnadower *et al* to be useful for grading episodes at the point of first contact and during follow-up. They recommended its use in clinical trials involving acute gastroenteritis disease.[@R16] This recommendation was later endorsed by the European Society of Pediatric Gastroenterology, Hepatology and Nutrition (ESPGHAN) in their 2014 update on the guidelines for management of acute gastroenteritis in children.[@R21] To the best of our knowledge, the incorporation of the Vesikari score in the inclusion criteria has never been performed before in any studies on racecadotril and this may have led to the lack of effect seen with the medication in this study. In their review carried out in 2011, Lehert *et al*[@R7] alluded to the same noting that baseline level of dehydration was a significant negative predictor for duration of diarrhoea after introduction of racecadotril.
Previous studies on racecadotril have included children with diarrhoea lasting \<5 days.[@R7] The WHO defines persistent diarrhoea as that which lasts 14 days or longer and children with disease duration less than this are treated for acute gastroenteritis.[@R11] Almost half of the participants included in this study had diarrhoea of 5 days or more (46% in both groups). Continued damage to the intestinal mucosa results in an osmotic effect which contributes to the mechanism of diarrhoeal causation. Racecadotril has a minimal effect on this mechanism of diarrhoea and this may have contributed to the results in this study.[@R5]
Another difference in the population studied was the inclusion of children with malnutrition. In a trial carried out by Jean *et al*,[@R19] children with weight for age \<80% were excluded. In this study, children with severe undernutrition made up 15% of the drug group and 26% of the placebo group. However, the mean weight of participants in this study was 7.4 kg which was similar to a study carried out by Savitha[@R20] (as quoted in the Lehert *et al*\'s[@R7] review). In both the Jean *et al* and Savitha studies, there was a significant improvement in the stool output for children on racecadotril.
The difference in the length of inpatient stay between the two groups constituted about half a day. This was not statistically significant but was comparable with other studies carried out at the same site. Osano *et al*[@R22] studied admissions of children with acute gastroenteritis at the Kenyatta National Hospital and noted that the average duration of stay was 4.2 days. None of the other inpatient studies carried out in children on racecadotril have described differences in duration of stays for their participants. Cojocaru *et al* stated that the effect of racecadotril on duration of hospitalisation could not be commented on due to the short patient stay and follow-up in his study.[@R8]
The use of zinc in this study may have contributed to similar durations of illness between the two groups. Previous studies have shown a link between the use of zinc and the duration of illness in developing countries.[@R11] [@R23] In most studies performed, racecadotril has been used alongside ORS although none have incorporated zinc into the treatment protocol.[@R7]
Clinically significant adverse events were equally distributed between the two groups. The results are similar to those in other studies including the review carried out by Lehert *et al*.[@R7] Adverse events in this study involved mainly the skin and upper respiratory tract (13%), which are comparable to those found in the literature.[@R6] [@R7] Convulsions were noted in the drug group of this study, a finding that has not been reported in studies with racecadotril thus far. The significance of this finding cannot be commented on due to the sample size.
Limitations {#s4a}
-----------
Although the number of stools was recorded on a daily basis, measurement of stool weights may have been a better measure to determine stool output as has been done in similar studies on the subject. The recording of number of stools as reported by the parents/guardians may have introduced a recall bias that affected the results. This may have been avoided by investigators observing the number of stools rather than parents providing information. The presence of comorbid conditions was not considered when the participants were recruited to the trial. Some of these conditions (especially HIV) may have had an impact on the results.
Conclusion {#s5}
==========
This study does not demonstrate a significant reduction in the number of stools at 48 hours for children admitted with severe gastroenteritis after introduction of racecadotril. Furthermore, it does not reduce the duration of inpatient stay or the duration of diarrhoea when compared with placebo.
Supplementary Material
======================
###### Supplementary Materials
The authors thank the Aga Khan University Hospital Nairobi---financial and logistical support to carry out the study; University of Nairobi, Kenyatta National Hospital---permission to use facility as study site; Sai Pharmaceuticals, Nairobi, Kenya---provision of drug (Racedot) and placebo for the trial. The authors also thank research assistants: Gurpreet Rooprai, Benjamin Matei and Shimona Menezes; and Drug Safety Monitoring Board: Dalton Wamalwa, Bonface Osano, Peter Ngwatu and Hussein Dossajee.
**Contributors:** JG (primary investigator for the study) took part in design, execution, analysis and presentation of the study results. AL (content expert) and FW (statistics expert) took part in design and analysis of the study with input on manuscript preparation and presentation of results.
**Funding:** Aga Khan University, Nairobi; Sai Pharmaceutical Distributors, Nairobi, Kenya.
**Competing interests:** None declared.
**Patient consent:** Obtained.
**Ethics approval:** Ethics Review Board of the Aga Khan University Hospital, Nairobi, and Scientific and Ethics Committee of the Kenyatta National Hospital/University of Nairobi.
**Provenance and peer review:** Not commissioned; externally peer reviewed.
**Data sharing statement:** No additional data are available.
| {
"pile_set_name": "PubMed Central"
} |
{#sp1 .T21}
{#sp2 .T22}
{#sp3 .T23}
{#sp4 .T24}
{#sp5 .T25}
{#sp6 .T26}
{#sp7 .T27}
{#sp8 .T28}
{#sp9 .T29}
{#sp10 .T30}
{#sp11 .T31}
{#sp12 .T32}
{#sp13 .T33}
{#sp14 .T34}
{#sp15 .T35}
{#sp16 .T36}
{#sp17 .T37}
{#sp18 .T38}
{#sp19 .T39}
{#sp20 .T40}
{#sp21 .T41}
{#sp22 .T42}
[^1]: Read 3rd November 1926.
| {
"pile_set_name": "PubMed Central"
} |
Introduction {#sec1}
============
Mesoporous semiconductor oxide electrodes are used in different applications including electrochemical sensing, electrochromic devices and photoelectrochemical generation of fuels or electrical energy.^[@ref1]−[@ref5]^ All these applications rely on the external manipulation or tracking of the charge transfer between an optically and/or chemically active layer and an external contact. The macroscopic rate of charge transfer between the mesoporous film and a conductive substrate is the result of a sequence of intermingled microscopic processes. These processes, which are associated with carrier generation, recombination, transport and transfer, take place on different time scales and compete kinetically with each other. Due to a high concentration of trap states electron transport in mesoporous semiconductor oxide films is orders of magnitude slower than in single crystals. In photoelectrocatalytic and photovoltaic applications electron collection at the external contact competes with charge recombination in the bulk of the semiconductor and at the interfaces within the porous film, limiting solar conversion efficiency. The use of electrochemical doping (charge transfer reductive doping), where electron/proton or electron/Li^+^ pairs incorporated within the oxide "passivate" the electron traps,^[@ref6]−[@ref8]^ is a very efficient way to temporarily improve photoelectrochemical activity, and as such, constitutes an elegant way to improve charge separation within these materials. Although many efforts are being pursued in this direction, a comprehensive study of these electronic defects is still missing. To reach such a level of understanding, one needs to carefully identify the nature of the recombination centers and transport-limiting traps in mesoporous semiconductor electrodes, which is a significant challenge for both experimentalists and theoreticians.
It is well established that the fundamental processes associated with the transport,^[@ref9]−[@ref12]^ transfer,^[@ref13],[@ref14]^ and recombination^[@ref15]^ of photogenerated charge carriers in mesoporous semiconductor electrodes are determined by both the distribution of band gap states and their population, i.e., the position of the Fermi level within the film. Electrochemical methods such as cyclic voltammetry and electrochemical impedance spectroscopy have proven useful in characterizing electronic states in nanostructured semiconductor oxide electrodes, though the chemical nature of the traps remains controversial.^[@ref5],[@ref16]^ Specifically, charge/discharge measurements provide information on the distribution of electrochemically active states in mesoporous electrodes, where electron accumulation is compensated by the adsorption of ions at the oxide surface. For TiO~2~ electrodes in an aqueous acidic electrolyte the generation of Ti^3+^ centers is compensated by proton uptake ([eq [1](#eq1){ref-type="disp-formula"}](#eq1){ref-type="disp-formula"})Importantly, in the case of small cations (such as H^+^ or Li^+^), charge injection and compensation not only take place at the semiconductor/electrolyte interface, but can become a three-dimensional process via insertion of ions into subsurface regions of the nanocrystals. This process is often referred to as electrochemical or charge transfer reductive doping.^[@ref6],[@ref17]^ The reversibility of this charge accumulation opens up the possibility of driving fast reduction reactions at the semiconductor/electrolyte interface.^[@ref18]^ On the other hand, electrochemical doping was found to modify, at least temporarily, the electrode performance in different applications ranging from dye-sensitized solar cells^[@ref8],[@ref19],[@ref20]^ to photocatalysis^[@ref6]−[@ref8],[@ref21]−[@ref24]^ and supercapacitors.^[@ref25]^ The doping of mesoporous TiO~2~ electrodes with Li (which is isovalent with H) has recently been demonstrated to enhance electron transport and improve efficiency in perovskite solar cells.^[@ref26]^ Whereas different studies discuss an increase of the electrode performance upon electrochemical doping phenomenologically by accelerated charge transport and reduced recombination,^[@ref6]−[@ref8],[@ref19],[@ref22]^ the underlying microscopic details remain to be elucidated.
For TiO~2~, theoretical studies have recently addressed at the single particle level the geometry and energetics of electron trap states in the bulk^[@ref27]^ and at the semiconductor/electrolyte interface.^[@ref28]−[@ref31]^ Furthermore, intrinsic trapping properties of grain boundary interfaces have been studied. Deep electron traps located at the grain boundary are found to slow down charge transport unless high current densities ensure a high average occupation of transport-limiting traps.^[@ref9]^ Such trap filling effects have recently been highlighted for deep traps in oriented TiO~2~ nanotube arrays by dynamic photocurrent measurements.^[@ref10]^ In addition, these states may act as recombination sites exerting a further deleterious impact on the photocurrent.^[@ref32]^
The high structural and electronic complexity of mesoporous semiconductor oxide electrodes makes an investigation of the nature, concentration, and location of electronic trap states and the elucidation of their impact on charge recombination and transport very challenging. Designing appropriate model systems to understand the action of these states is difficult. Indeed, their complexity must be high enough to realistically mimic processes in technologically relevant materials, but low enough to result in clear structure--activity relationships that can be supported by both experiments and theoretical modeling.
In this work, we combine first-principles theoretical calculations with the electrochemical characterization of nanostructured rutile TiO~2~ films to demonstrate that particle/particle interfaces introduce deep traps. These interfaces represent favorable locations for proton segregation, which can be induced by the electrochemical doping of the porous electrodes. A long lasting (hours to days), but reversible accumulation of electrons and protons (i.e., e^--^/H^+^ or H^0^ doping) at the interface is tracked by cyclic voltammetry via the shift toward more positive potentials of a pair of capacitive peaks, which is associated with trap states at the particle/particle interface. The passivation of recombination centers by e^--^/H^+^ doping leads to a transient photocurrent enhancement due to improved electron/hole separation for those electrodes, which are characterized by a high concentration of particle/particle interfaces (i.e., random particle networks). For electrodes lacking a high density of particle/particle interfaces (i.e., arrays of oriented nanocolumns), only a minor improvement of the photocurrent is observed upon doping. This study highlights the importance of particle/particle interfaces in mesoporous films and provides strategies to actively manipulate the density of electronic states and their population by electrochemical methods. The resulting long-lasting (\>15 h) improvement of photoelectrode performance after electrochemical doping is explained by using theoretical calculations that are in qualitative agreement with experiments.
Experimental Section {#sec2}
====================
Thin Film Preparation {#sec2.1.1}
---------------------
Slurries of rutile TiO~2~ nanoparticles (Sachtleben, Nano Rutile) were prepared by grinding 1 g of TiO~2~ powder with 3.2 mL of H~2~O, 60 μL of acetylacetone (99+%, Aldrich), and 60 μL of Triton X (Aldrich) and were spread with a glass rod onto fluorine-doped tin oxide (FTO) conducting glass (Pilkington, TEC 15, resistance 15 Ω/□) using Scotch tape as a spacer. Alternatively, Ti foils (Goodfellow, 99.6+%, 250 μm) were used to investigate a possible impact of the substrate type on the electrochemical and photoelectrochemical properties. However, no such effect was observed for the experiments reported here. The nanoparticle (NP) films were annealed and sintered for 1 h at 450 °C in air. After sintering a film thickness of 3.5 ± 2.0 μm was determined by scanning electron microscopy (SEM). The films are formed by a random network of elongated particles with a length of ∼50 nm and a width of ∼20 nm ([Figure S1a,b](#notes-1){ref-type="notes"} and ref ([@ref33])) and are of pure rutile phase ([Figure S2](#notes-1){ref-type="notes"}). As shown in a previous study, nanoparticles are elongated in the \[001\] direction. Furthermore, it was estimated that a high fraction of the exposed surface is formed by (110) facets.^[@ref33],[@ref34]^
Electrodes formed by a rutile TiO~2~ nanocolumn (NC) array were prepared using a hydrothermal synthesis.^[@ref35]^ Concretely, 21.6 mL of a 6 M HCl solution were mixed with 360 μL of Tetra-*n*-butylorthotitanat (98%, Merck Millipore). The solution was placed in a Teflon-lined steel autoclave (45 mL, Parr Instruments) containing FTO substrates and was heated to 150 °C for 15 h. After synthesis, the electrodes were thoroughly rinsed with water. These films consist of rutile TiO~2~ nanocolumns with a rectangular cross section and with a width of 80--180 nm and a length of ∼1.5 μm ([Figures S1c,d and S2](#notes-1){ref-type="notes"}). As highlighted previously, individual nanocolumns are porous and consist of a bundle of oriented and single crystalline nanowires with a diameter of 10--20 nm.^[@ref36]^ The nanowires are elongated along the \[001\] direction and are expected to expose (110) facets at the surface.^[@ref35]^ The endings of the nanowires can be observed at higher magnifications at the top parts of the nanocolumns (inset in [Figure S1c](#notes-1){ref-type="notes"}).
For both types of electrode a copper wire was attached to the conducting substrates with silver epoxy. The contact area and the uncovered parts of the substrate were finally sealed by epoxy resin.
Theoretical Calculations {#sec2.2}
------------------------
Spin polarized density functional theory (DFT) calculations are performed using the projector augmented wave formalism as implemented in the Vienna ab initio simulation package.^[@ref37],[@ref38]^ The 3d and 4s electrons of Ti, and the 2s and 2p electrons of O are treated as valence electrons and expanded in a plane wave basis with energies up to 500 eV. We use the Perdew--Burke--Ernzerhof exchange correlation functional and correct for the self-interaction error (SIE) for electrons by employing a DFT+*U* approach.^[@ref39]^ The Hubbard *U* parameter for the Ti 3*d*-states is taken from previous work which fitted to spectroscopic properties of surface oxygen vacancies (*U*~Ti~ = 4.2 eV).^[@ref40]^ We also employ a Hubbard *U* term to correct the SIE on O 2*p*-states (*U*~O~ = 7.5 eV) in order to make the results transferable to future calculations which will consider electrons and holes.^[@ref41]^ However, we note that the addition of a Hubbard *U* term on O does not affect the calculated trapping energies reported in the present work. For the conventional cell of rutile a 6 × 6 × 9 Monkhorst--Pack k-point grid is used and structural optimization is performed until forces are less than 0.01 eV/Å. Using the Perdew--Burke--Ernzerhof exchange correlation functional we obtain lattice parameters within 2% of experiment (*a* = 4.67 Å and *c* = 3.03 Å).
To investigate the interaction of electrons with the grain boundary defect, we attempt to localize an electron polaron at all inequivalent Ti sites within the grain boundary supercell. To achieve this, we create a precursor potential well for electron trapping by displacing nearest neighbor anions away from a particular Ti site by 0.1 Å followed by full self-consistent optimization of the structure.^[@ref9],[@ref42]^ In cases where this displacement procedure alone is insufficient to direct the self-consistent optimization into the desired charge localized metastable state, we manually set the orbital occupancy using a modification to the VASP code developed by Allen and Watson.^[@ref43]^ However, we stress that in all cases the resulting metastable states are fully and self-consistently optimized. For calculations involving charged defects (such as electron polarons), overall neutrality is ensured by employing a uniform compensating charge.
To identify prospective proton incorporation sites in the grain boundary supercell, we make use of the fact that protons will form a bond with lattice oxygen ions at a distance of approximately 1.0 Å.^[@ref44]^ We computationally identify the set of positions within 1.0 ± 0.1 Å of each lattice oxygen ion. We further reduce the number of possible proton positions by identifying the proton positions around each oxygen ion that has the lowest electrostatic potential (thereby representing the most favorable position for the proton on electrostatic grounds). In this way, we can readily obtain a large number of prospective proton positions, which provide the initial coordinates for full geometry relaxations. Using this procedure, inequivalent proton sites with the lowest energy can be obtained systematically. This procedure is straightforward to implement and may, with suitable modification, be applicable to modeling protons in low symmetry structures (such as nanoparticles or surfaces) in a wider range of oxide materials.
Results and Discussion {#sec3}
======================
Experimental Results {#sec3.1}
--------------------
Rutile TiO~2~ nanoparticle (NP) electrodes (consisting of a random network of TiO~2~ particles, [Figure S1a,b](#notes-1){ref-type="notes"}) and nanocolumn (NC) electrodes (consisting of an array of oriented nanocolumns, [Figure S1c,d](#notes-1){ref-type="notes"}) have been used in the present study as model systems for investigating the effect of nanocrystal organization and interconnection on electrochemical and photoelectrochemical properties. The voltammetric response of a rutile TiO~2~ NP electrode is characterized, in the absence of significant faradaic currents (i.e., in 1 M methanol/0.1 M HClO~4~ aqueous solution purged from O~2~), by a charge accumulation region at low potentials ([Figure [1](#fig1){ref-type="fig"}](#fig1){ref-type="fig"}a).^[@ref45]^ The photocurrent onset potential, which yields an estimate of the conduction band edge position in the semiconductor,^[@ref5]^ lies for this electrode at *E*~Ag/AgCl~ ∼ −0.5 V ([Figure S3a](#notes-1){ref-type="notes"}).
{#fig1}
Previous analyses of the density of electrochemically active band gap states in mesoporous TiO~2~ films by cyclic voltammetry and electrochemical impedance spectroscopy^[@ref5],[@ref16]^ have demonstrated the presence of a broad exponential distribution of states below the conduction band edge in the accumulation region in the case of anatase electrodes, which is absent in rutile TiO~2~ films.^[@ref46]^ Currents in the accumulation region (*E*~Ag/AgCl~ \< −0.25 V, [Figure [1](#fig1){ref-type="fig"}](#fig1){ref-type="fig"}) of rutile TiO~2~ electrodes have been attributed to the population/depopulation of electronic states in the conduction band compensated by proton adsorption at the oxide surface.^[@ref46]^ However, for both TiO~2~ modifications, a narrow distribution of deep trap states is typically present and gives rise to a pair of capacitive peaks in the cyclic voltammograms (CVs), which is also observed on rutile TiO~2~ NP electrodes ([Figure [1](#fig1){ref-type="fig"}](#fig1){ref-type="fig"}a). For pristine NP electrodes, these peaks appear at *E*~Ag/AgCl~ ∼ −0.13 V and lie thus ∼0.3--0.5 V below the conduction band edge of the semiconductor. In the following, we will focus in detail on the intensity of these signals and on the energetics of the associated trap states in mesoporous films featuring different morphology (i.e., NP versus NC films) and will follow their modification upon electrochemical doping.
The intensity of the capacitive peaks depends significantly on electrode morphology. The corresponding signal is much less pronounced for the rutile TiO~2~ NC electrode ([Figure [1](#fig1){ref-type="fig"}](#fig1){ref-type="fig"}b) as compared to the NP electrode ([Figure [1](#fig1){ref-type="fig"}](#fig1){ref-type="fig"}a). Previous studies reported that for both ordered one-dimensional nanostructures and single crystal electrodes, the peaks are virtually absent, whereas they show a high intensity for thin film electrodes consisting of random nanoparticle networks.^[@ref32],[@ref33],[@ref47]^ In line with previous interpretations,^[@ref32],[@ref33],[@ref47]^ we assign the couple of capacitive peaks observed for the NP electrode to the contribution of electron traps at particle/particle interfaces. Interestingly, the contribution is asymmetric, the anodic peak being much broader than the cathodic peak ([Figures [1](#fig1){ref-type="fig"}](#fig1){ref-type="fig"}a and [S4a](#notes-1){ref-type="notes"}). Slow kinetics for H^+^ extraction (compare [eq [1](#eq1){ref-type="disp-formula"}](#eq1){ref-type="disp-formula"}) or a change in the electrode conductivity may contribute to this effect. The CVs in [Figure [1](#fig1){ref-type="fig"}](#fig1){ref-type="fig"} were obtained by applying to the electrode a linear potential profile with a scan rate of 20 mV·s^--1^. Importantly, at fast scan rates, it is possible that not all of the deep traps in the mesoporous film are equilibrated with the Fermi level of the conducting substrate. This is the reason why even large perturbation techniques such as cyclic voltammetry may yield for deep trap states only apparent chemical capacitances.^[@ref32],[@ref48]^ Therefore, we performed charging and discharging measurements using extremely long lasting perturbations in the potential range featuring the capacitive peaks ([Figure S5](#notes-1){ref-type="notes"}). We measured the capacitive currents upon stepping the electrode potential in potential steps Δ*E*~Ag/AgCl~ = 0.02 V first from 0.2 to −0.2 V (charge accumulation) and then from −0.2 V back to 0.2 V (charge extraction). After every step, the potential was kept constant for 60 s and the accumulated/extracted charge density associated with each potential step (left axis in [Figure S4b](#notes-1){ref-type="notes"}) was determined by integration of the resulting current transient ([Figure S5](#notes-1){ref-type="notes"}). To obtain the chemical capacitance associated with interface traps (right axis in [Figure S4b](#notes-1){ref-type="notes"}) the charge was then referred to Δ*E*~Ag/AgCl~. Such an analysis yields a much higher symmetry of charging and discharging branches, nevertheless, there is still an imbalance of positive and negative charge pointing to a partial irreversibility of charge accumulation. The chemical capacitance extracted from these measurements (right axes in [Figure S4a,b](#notes-1){ref-type="notes"}) has thus to be considered an apparent capacitance. From the total charge accumulated upon stepping the potential from 0.2 to −0.2 V (35 μC·cm^--2^, [Figure S4b](#notes-1){ref-type="notes"}), we estimate (using the average values of film thickness and particle size and assuming a film porosity of 0.5) the number of extracted charges to correspond to ∼25 electrons per TiO~2~ nanoparticle.
As previously reported, TiO~2~ electrodes can be electrochemically doped by cathodic polarization.^[@ref6],[@ref17]^ Following polarization at *E*~Ag/AgCl~ = −0.6 V significant changes are observed in the CV of a rutile TiO~2~ NP electrode ([Figures [1](#fig1){ref-type="fig"}](#fig1){ref-type="fig"}a and [S4a](#notes-1){ref-type="notes"}): the peak corresponding to deep traps is displaced by ∼0.08 V toward more positive potentials, while a slight increase of the peak intensity is observed upon doping. The same observations are made in the absence of methanol ([Figure S6](#notes-1){ref-type="notes"}). Qualitatively the same conclusions can be drawn from the large perturbation charging/discharging experiment ([Figure S4b](#notes-1){ref-type="notes"}). Electrochemical doping induces only minor changes at *E*~Ag/AgCl~ \< −0.2 V, although a slight increase of the capacitive current is observed at −0.45 V \< *E*~Ag/AgCl~ \< −0.25 V. Importantly, we do not observe a shift of the photocurrent onset potential upon electrochemical doping ([Figure S3b](#notes-1){ref-type="notes"}) indicating that the band edges are not displaced significantly. All changes in the CVs are reversible with respect to prolonged polarization at 0.8 V (*t*^dedop^ \> 15 h, [Figure [1](#fig1){ref-type="fig"}](#fig1){ref-type="fig"}a). These observations point to a dynamic and transient change of the density of electrochemically active states upon electrochemical charge accumulation in NP electrodes. Importantly, no significant change of the CV is observed upon doping of a NC electrode ([Figure [1](#fig1){ref-type="fig"}](#fig1){ref-type="fig"}b).
Electrochemical doping has a beneficial effect on the photoelectrochemical performance of rutile TiO~2~ NP electrodes as deduced from photocurrent transients ([Figures [2](#fig2){ref-type="fig"}](#fig2){ref-type="fig"}a and [S7](#notes-1){ref-type="notes"}) and CVs ([Figure S8a](#notes-1){ref-type="notes"}). Concretely, the photocurrent generated by the electrode in a 0.1 M HClO~4~ aqueous solution containing 1 M methanol as a hole scavenger depends significantly on the electrochemical pretreatment of the film as shown in the following. First the photocurrent of a pristine electrode was recorded at *E*~Ag/AgCl~ = 0.8 V. Then the electrode was polarized at progressively more negative potentials in the accumulation region corresponding to the electrochemical doping of the film. After every doping step the photocurrent was again recorded to sample the impact of doping on the photoelectrocatalytic activity of the electrode. Two doping parameters were systematically changed--doping potential ([Figure S7](#notes-1){ref-type="notes"}) and doping time ([Figure [2](#fig2){ref-type="fig"}](#fig2){ref-type="fig"}). Whereas electrode polarization for 20 min at *E*~dop~ = −0.5 V induces only minor changes of the photoelectrocatalytic activity, we observe an up to 3-fold photocurrent increase when doping at *E*~dop~ = −0.6 V ([Figure S7](#notes-1){ref-type="notes"}). The photocurrent enhancement by electrochemical doping is a very slow process. Only after 4 h of polarization at −0.6 V no further changes are observed in the transients ([Figure [2](#fig2){ref-type="fig"}](#fig2){ref-type="fig"}). After such a long doping time the photocurrent has experienced an increase by a factor of ∼7 (photocurrent enhancement factor, PCEF = 7, [Figure [2](#fig2){ref-type="fig"}](#fig2){ref-type="fig"}a).
{#fig2}
Importantly, the photocurrent increase is reversible with respect to prolonged polarization at positive potentials ([Figure S8](#notes-1){ref-type="notes"}). However, even after 15 h of charge extraction (by electrode polarization at 0.8 V) the photocurrent still exceeds its initial value by ∼30%. These results highlight that the beneficial effect of electrochemical doping is transient, but long lasting.
The relative photocurrent enhancement upon doping is much less pronounced for rutile TiO~2~ NC electrodes (*PCEF* = 2, [Figure [2](#fig2){ref-type="fig"}](#fig2){ref-type="fig"}b). Also in this case the beneficial effect is reversible with respect to polarization at 0.8 V (not shown). A comparison of the photocurrent evolution following the progressive electrochemical doping of NP and NC electrodes is shown in the chronoamperometric profiles in [Figure S9](#notes-1){ref-type="notes"}. Importantly, these results confirm that the increased current measured upon UV exposure of doped electrodes corresponds indeed to a faradaic photocurrent and does not simply result from a light-induced extraction of charges accumulated in the doping step. The additional charge transferred (after doping) from the TiO~2~ film to the conducting substrate upon UV exposure exceeds by far the charge injected from the conducting substrate into the TiO~2~ film upon electrochemical doping ([Figure S9a](#notes-1){ref-type="notes"}).
From the electrochemical characterization of NP and NC films we have gained the following pieces of information about the impact of electrode morphology and electrochemical doping on the density of electrochemically active states and on the photoelectrocatalytic activity:
\(i\) For porous films consisting of a random particle network (NP electrodes) a high density of deep electron traps gives rise to a couple of capacitive peaks in the CV. This signal is virtually absent in films consisting of oriented nanocolumn arrays (NC electrodes).
\(ii\) Prolonged polarization of NP electrodes at *E*~Ag/AgCl~ = −0.6 V (electrochemical or charge transfer reductive doping) induces a displacement of these capacitive peaks by ∼0.08 V toward more positive potentials and a minor increase in the chemical capacitance at −0.45 V \< *E*~Ag/AgCl~ \< −0.25 V. These changes are reversible with respect to charge extraction (dedoping) upon prolonged electrode polarization at *E*~Ag/AgCl~ = 0.8 V. Both processes (doping and dedoping) are extremely slow (hours to days).
\(iii\) Electrochemical doping increases the photoelectrocatalytic activity of NP electrodes toward methanol oxidation as sampled by a 7-fold increase of the photocurrent (PCEF = 7). The activity enhancement is transient and the photocurrent relaxes slowly back to its initial value (*t*^dedop^ \> 15 h). The beneficial effect of electrochemical doping is much less pronounced for rutile TiO~2~ NC electrodes (PCEF = 2).
Theoretical Results {#sec3.2}
-------------------
To help interpret the experimental results discussed above and provide deeper atomistic insight into the effect of protons on electron trapping we perform first-principles theoretical calculations for a model interface in nanocrystalline TiO~2~. In particular, we consider the (210)\[001\] rutile TiO~2~ grain boundary, the structure of which has been investigated previously both experimentally and theoretically ([Figure [3](#fig3){ref-type="fig"}](#fig3){ref-type="fig"}a).^[@ref49],[@ref50]^ While this interface possesses a high degree of symmetry it has atomistic features which are expected to be representative of more general interfaces in nanocrystalline TiO~2~, namely, reduced ion coordination and local strain at the interface. In a recent theoretical study, it was demonstrated that this grain boundary is associated with interfacial Ti ions which can trap electrons more strongly than bulk Ti lattice sites.^[@ref9]^ This effect is due to local variations in the electrostatic potential near the grain boundary and changes in ion coordination and bond strain with similar effects found at TiO~2~ surfaces.^[@ref42]^[Figure [4](#fig4){ref-type="fig"}](#fig4){ref-type="fig"} shows the distribution of electron trapping energies (defined with respect to the energy of an electron trapped on a bulk Ti site) for Ti ions within ±6 Å of the grain boundary plane. We note that at finite temperature electrons may hop between Ti sites at the interface. The activation energy for electron hopping between adjacent sites was calculated previously to be about 0.3 eV in the bulk and up to 50% higher at the interface.^[@ref9]^ Owing to their increased stability the equilibrium occupation of interfacial traps will remain higher than that in the bulk. Therefore, the presence of deep traps at this interface provides a semiquantitative model for the voltammetric feature of electron trapping states (pair of capacitive peaks) observed in the pristine TiO~2~ NP electrodes by cyclic voltammetry ([Figure [1](#fig1){ref-type="fig"}](#fig1){ref-type="fig"}a). As such it is a useful reference system on which to explore the interaction of protons with interfaces and the subsequent effect they have on electron trapping. For the following discussion, it is important to keep in mind that the energy scale and the electrochemical potential scale have opposite signs, i.e., a trap state becoming more stable (i.e., deeper) will be characterized by a more negative trapping energy and a more positive electrochemical potential.
![(a) Optimized structure of the pristine (210)\[001\] rutile TiO~2~ grain boundary showing the electron spin density associated with an electron in the most stable site (isosurface shown in purple). The region within ±6 Å of the grain boundary is highlighted. (b) H^+^ decorated (210)\[001\] rutile TiO~2~ grain boundary. (c) (H^+^)(e^--^) decorated (210)\[001\] rutile TiO~2~ grain boundary. (d) (H^+^)(e^--^) decorated (210)\[001\] rutile TiO~2~ grain boundary with an additional electron trapped in the most stable site. Ti sites, O sites, and H^+^ ions are represented by blue, red, and green spheres, respectively.](ja-2016-08636q_0004){#fig3}
{#fig4}
Electrochemical doping of TiO~2~ by prolonged polarization is likely to be associated with the incorporation of H^+^ ions from the aqueous solution to compensate the negative electron charge trapped at interfaces. To assess this possibility we first investigate the interaction of protons with the (210)\[001\] rutile TiO~2~ grain boundary. On introduction into the TiO~2~ lattice protons form bonds with lattice O^2--^ ions resulting in OH^--^ species. While previous theoretical studies have identified the most stable structure of the OH^--^ species in bulk rutile TiO~2~, it is not straightforward to deduce the likely proton configurations in the lower symmetry grain boundary region. To address this problem we identify prospective positions for H^+^ incorporation based on analysis of the three-dimensional electrostatic potential and screen 80 different configurations to identify the most stable structure (see [Experimental Section](#sec2){ref-type="other"}). For these calculations we consider one H^+^ ion in a supercell of dimensions 9.1046 × 10.439 Å corresponding to a density of 1.05 × 10^14^ cm^--2^. The most stable H^+^ incorporation site is found at the grain boundary and is 0.6 eV more stable than in the bulk ([Figure [3](#fig3){ref-type="fig"}](#fig3){ref-type="fig"}b). The presence of H^+^ induces a transformation in structure near the grain boundary as compared to the pristine structure. In particular, one of the Ti ions near the grain boundary relaxes toward the OH^--^ ion.
We next investigate the interaction of electrons with the H^+^ decorated grain boundary structure identified above. By making suitable initial atomic distortions around each Ti site in the supercell followed by full optimization of the total energy with respect to relaxation of all ion coordinates we obtain a series of metastable configurations corresponding to electrons trapped on different Ti ions (see [Experimental Section](#sec2){ref-type="other"}). The most stable electron trapping Ti site is located at the grain boundary directly adjacent to the OH^--^ species ([Figure [3](#fig3){ref-type="fig"}](#fig3){ref-type="fig"}c). This defect can be considered as a H^0^ atom with the proton and electron dissociated onto neighboring sites. Very similar defect centers are found in nanocrystalline MgO where they have been characterized in detail by electron spin resonance and theoretical calculations.^[@ref51],[@ref52]^ Hereafter, we will refer to these proton plus electron defects produced by H^0^ doping as (H^+^)(e^--^) centers, following the nomenclature of previous studies. The (H^+^)(e^--^) center should provide a reasonable model for the electron traps in the NP electrodes following electrochemical doping and polarized for sufficiently short times at a positive potential. We have also computed the Fermi contact hyperfine coupling parameter for (H^+^)(e^--^) in the most stable position segregated at grain boundary −7.5 MHz. This is significantly reduced compared to that calculated for the isolated H atom −1402.6 MHz (close to the experimental value of 1422 MHz). This could provide an experimental signature of (H^+^)(e^--^) centers at the grain boundary.
If each proton at the grain boundary has already trapped an electron forming a (H^+^)(e^--^) center (as shown in [Figure [3](#fig3){ref-type="fig"}](#fig3){ref-type="fig"}c) one may ask how additional electrons added to the system would interact with the interface (for example as realized experimentally by CV measurements on electrochemically doped electrodes). To address this question we obtain fully optimized metastable configurations corresponding to the localization of a second electron on all Ti sites in the supercell. An electron trapped on a bulk-like Ti site has a very similar local geometry and spin density to the bulk-like polaron in the pristine grain boundary. This provides a reference with which to assess the trapping energies of sites in the vicinity of the grain boundary. We find a distribution of trapping energies for Ti ions within ±6 Å of the grain boundary plane spanning a similar range to that found for the pristine interface and the most stable electron trap is again located close to the grain boundary ([Figure [3](#fig3){ref-type="fig"}](#fig3){ref-type="fig"}d). The distribution of electron trapping energies associated with the pristine and (H^+^)(e^--^)-doped grain boundaries are compared in [Figure [4](#fig4){ref-type="fig"}](#fig4){ref-type="fig"}. The pristine grain boundary presents 18 Ti sites per supercell that can trap electrons more strongly than in the bulk (i.e., traps with *E*~t~ \< 0, hereafter referred to as interface traps). This corresponds to an interface trap concentration of 1.89 × 10^15^ cm^--2^ with an average trapping energy of −0.14 eV. Following (H^+^)(e^--^)-doping the number of interface traps is reduced dramatically. Only 9 Ti sites per supercell are found to trap electrons corresponding to an interface trap concentration of 0.95 × 10^15^ cm^--2^. Analysis of atomic structures indicates that a number of effects are responsible for the modification of interfacial traps on (H^+^)(e^--^)-doping. As noted above, the presence of H^+^ induces a localized deformation, which changes the structural and electrostatic environment of Ti sites near the grain boundary. We find a strong correlation between the electron trapping energy of a given Ti site and its corresponding electrostatic potential, as discussed previously for the pristine case.^[@ref9]^ In particular, the presence of the (H^+^)(e^--^) center modifies the electrostatic potential on Ti sites near the grain boundary destabilizing a number of traps. At the same time one of the Ti sites adjacent to H^+^ that was not a trap in the pristine case becomes a trap after doping. The net result is that the number of Ti sites available to trap electrons is reduced by 50%. The optimized atomic structures for the pristine and doped grain boundaries are provided in the [Supporting Information](#notes-1){ref-type="notes"}. In addition to the reduced concentration of interface traps there is also a reduction in the average trapping energy (i.e., from −0.14 to −0.26 eV). The shift in average trapping energy of about 0.12 eV is of the same order as that observed experimentally for doped electrodes by CV (∼0.08 eV, [Figure [1](#fig1){ref-type="fig"}](#fig1){ref-type="fig"}a). Although the average depth of grain boundary traps is increased, the significant decrease in the density of interface traps provides an explanation for the improved photoelectrocatalytic activity of electrochemically doped electrodes observed experimentally.
General Discussion {#sec4}
==================
Due to the high specific surface area of mesoporous semiconductor electrodes, the main contribution to the density of electronic band gap states as sampled by electrochemical methods such as cyclic voltammetry results from processes at the semiconductor/electrolyte interface. Indeed these electrodes typically show a reversible charging/discharging behavior on short time scales. For TiO~2~ electrodes the corresponding accumulated charge was shown to scale linearly with the internal area of the semiconductor/electrolyte interface.^[@ref33]^ Nevertheless, processes with different kinetics contribute with different relative intensities to the overall signal. Consequently, when extracting a chemical capacitance from the measured capacitive current,^[@ref16]^ those electronic states getting populated by very slow charging processes will be underrepresented.^[@ref48]^ This is true for electronic states in subsurface regions of the semiconductor, such as the electron traps at particle/particle interfaces giving rise to the pair of capacitive peaks in the CVs of rutile TiO~2~ NP electrodes ([Figure [1](#fig1){ref-type="fig"}](#fig1){ref-type="fig"}a). Importantly, whereas electrochemical methods based on charge/discharge measurements provide information on the distribution of electrochemically active states, it must not be ignored that a persistent charge accumulation in the mesoporous film may modify both the Fermi level and the density of states itself.^[@ref6],[@ref19],[@ref20]^ In this context, it is well established nowadays that long lasting reductive treatments of mesoporous films may result in a long lasting accumulation of charges (electrochemical or charge transfer reductive doping), thereby significantly influencing the macroscopic electrode behavior in different applications.^[@ref6]−[@ref8],[@ref19]−[@ref25]^ Whereas the technological implications of such an electrochemical manipulation of the electrode properties are clear, the underlying reasons and microscopic details of this phenomenon are unknown.
Our observations from electrochemical measurements and results from first principle theoretical calculations give a consistent picture of electron and proton trapping in nanocrystalline TiO~2~ films of different morphology. The main conclusions of our study are depicted in [Figure [5](#fig5){ref-type="fig"}](#fig5){ref-type="fig"}. Calculations indicate that (H^+^)(e^--^) decoration modifies the distribution of electron traps at particle/particle interfaces ([Figure [5](#fig5){ref-type="fig"}](#fig5){ref-type="fig"}i). An increase of the depth of interface traps goes along with a 50% decrease in their density. Consistent changes of the density of electrochemically active states are tracked by voltammetry ([Figure [5](#fig5){ref-type="fig"}](#fig5){ref-type="fig"}ii). Upon electrochemical doping of rutile TiO~2~ NP electrodes, i.e., upon a long lasting accumulation of electron/proton pairs in the film, we observe a reversible shift toward more positive potentials of capacitive peaks associated with trap states at particle/particle interfaces. In addition an increase of the chemical capacitance is observed at more negative potentials, i.e., at −0.45 V \< *E*~Ag/AgCl~ \< −0.25 V. Such a modification was previously related to the population of subsurface states upon a light-induced insertion of protons and electrons affording faster charge transport in dye-sensitized TiO~2~ films.^[@ref19]^ Here we show that the partial removal of interface traps upon (H^+^)(e^--^) decoration of particle/particle interfaces may contribute to such a capacitance change. The main contribution to currents in the accumulation region of both pristine and doped electrodes (contributions at *E*~Ag/AgCl~ \< −0.25 V highlighted in red in [Figure [5](#fig5){ref-type="fig"}](#fig5){ref-type="fig"}ii), however, are associated with the population/depopulation of electronic states in the rutile TiO~2~ conduction band compensated by proton adsorption at the oxide surface (e^--^~CB~/H^+^~ads~ states, [Figure [5](#fig5){ref-type="fig"}](#fig5){ref-type="fig"}c,d) as discussed in detail in a previous study.^[@ref46]^
{#fig5}
Electrochemical charge accumulation in NP electrodes is reversible on the time scale of a CV measurement, i.e. when recording the electrode's voltammetric response between 0.8 and −0.6 V (and vice versa) at a scan rate of 20 mV·s^--1^ ([Figure [5](#fig5){ref-type="fig"}](#fig5){ref-type="fig"}a--c). In this case charge accumulation takes place mainly at the particle surface and at those grain boundary states located near the oxide/electrolyte interface. Prolonged polarization of NP electrodes at *E*~Ag/AgCl~ = −0.6 V (*t*~--0.6V~^dop^ = 4 h) induces electrochemical doping and thus a population of trap states deep within the particle/particle interfaces ([Figure [5](#fig5){ref-type="fig"}](#fig5){ref-type="fig"} d). This charge accumulation is not reversible on the time scale of a CV measurement ([Figure [5](#fig5){ref-type="fig"}](#fig5){ref-type="fig"}d--f). Rather there is a long lasting (though reversible, vide infra) modification of the density of electrochemically active states. In the voltammetric experiment ([Figure [5](#fig5){ref-type="fig"}](#fig5){ref-type="fig"}d--f) charge accumulation and charge extraction take now place on a film featuring particle/particle interfaces, which are partially decorated by (H^+^)(e^--^). Consequently, band gap states are associated with modified trapping energies resulting in a modified density of electrochemically active states ([Figure [5](#fig5){ref-type="fig"}](#fig5){ref-type="fig"}i and ii.
The concerted uptake of e^--^/H^+^ pairs in particle/particle interface regions upon cathodic polarization (i.e., electron injection from the conducting substrate and insertion of protons from the electrolyte into the oxide) and the reverse process taking place upon anodic polarization (i.e., electron transfer to the substrate and proton diffusion through the solid phase into the electrolyte) are expected to proceed very slowly. Consequently, cyclic voltammetry samples only an apparent density of deep traps. The increased intensity of the pair of capacitive peaks associated with traps at the particle/particle interface can be explained by the enhanced conductivity in the doped film which allows populating and depopulating electron traps faster and deeper within the particle/particle contact area. The persistence of electrochemical doping is associated with the slow kinetics of H^+^ diffusion from the GB core to the oxide/electrolyte interface. The dedoping of the film can thus only be achieved upon prolonged polarization at *E*~Ag/AgCl~ \> 0.2 V (e.g., *t*~0.8V~^dedop^ = 15 h, [Figure [5](#fig5){ref-type="fig"}](#fig5){ref-type="fig"}a,f).
The increased photoelectrocatalytic activity of doped electrodes can be attributed to the deactivation of a major fraction of interface traps and recombination sites by the decoration of particle/particle interfaces with (H^+^)(e^--^) ([Figure [5](#fig5){ref-type="fig"}](#fig5){ref-type="fig"}a,f). Indeed, calculations point to a 50% decrease in density of interface traps following (H^+^)(e^--^) doping ([Figure [4](#fig4){ref-type="fig"}](#fig4){ref-type="fig"}), which is expected to affect charge separation at the particle/particle interface in two ways: by accelerating electron transfer across the grain boundary and by reducing electron/hole recombination. Faster electron transport in electrochemically doped TiO~2~ nanoparticle films has been proven recently.^[@ref8]^ On the other hand, Kamat and co-workers^[@ref7]^ reported on the deactivation of recombination centers due to trap filling and the generation of Ti^3+^/H^+^ centers in TiO~2~ electrodes. Enhancement of the transport properties of mesoporous TiO~2~ electrodes for perovskite solar cells via Li doping has also been demonstrated and is proposed to involve a similar mechanism.^[@ref26]^ While not investigated theoretically in this work, hole trapping may also be modified in a favorable way upon (H^+^)(e^--^) doping. Related studies are underway. Beneficial effects of electrochemical doping have been reported not only for TiO~2~, but also for ZnO, WO~3~, and BiVO~4~ films.^[@ref21],[@ref24]^ We believe that our findings will contribute to a better understanding of interfacial processes at play in different metal oxide-based materials.
Conclusions {#sec5}
===========
We have tracked the long lasting accumulation of electron/proton pairs in rutile TiO~2~ films consisting of a random nanoparticle network (i) via a reversible shift of a capacitive peak in the CV, which we associate with trap states located at particle/particle interfaces, and (ii) via the transient enhancement of the photoelectrocatalytic activity toward methanol photooxidation. Theoretical calculations indicate that interfaces between crystals in TiO~2~ represent favorable locations for the segregation of proton defects, being up to 0.6 eV more stable than in the bulk crystal. Importantly, (H^+^)(e^--^) doping of grain boundaries significantly modifies the electronic properties of the particle/particle interface. For Ti ions within ±6 Å of the interface a shift in the average trapping energy of deep traps of −0.12 eV is predicted with respect to the pristine interface. A 50% reduction of the overall number of deep electron traps at the grain boundary is considered to be the main reason for the beneficial effect of electrochemical doping of rutile TiO~2~ NP electrodes on their photoelectrocatalytic activity. The qualitative agreement between our experimental results and theoretical calculations strongly supports our detailed description of these complex interfacial systems.
The Supporting Information is available free of charge on the [ACS Publications website](http://pubs.acs.org) at DOI: [10.1021/jacs.6b08636](http://pubs.acs.org/doi/abs/10.1021/jacs.6b08636).Further experimental details, SEM images and Raman spectra of rutile TiO~2~ NP and NC films, as well as further electrochemical data; electron trapping energies for interface trapping sites ([PDF](http://pubs.acs.org/doi/suppl/10.1021/jacs.6b08636/suppl_file/ja6b08636_si_001.pdf))Optimized atomic structures of the pristine grain boundary ([CIF](http://pubs.acs.org/doi/suppl/10.1021/jacs.6b08636/suppl_file/ja6b08636_si_002.cif))Optimized atomic structures of the (H^+^)(e^--^)-doped grain boundary ([CIF](http://pubs.acs.org/doi/suppl/10.1021/jacs.6b08636/suppl_file/ja6b08636_si_003.cif))
Supplementary Material
======================
######
ja6b08636_si_001.pdf
######
ja6b08636_si_002.cif
######
ja6b08636_si_003.cif
The authors declare no competing financial interest.
We gratefully acknowledge support from COST Action CM1104. K.P.M. acknowledges financial support from EPSRC (EP/K003151). Via our membership of the UK's HEC Materials Chemistry Consortium, which is funded by EPSRC (EP/L000202), this work used the ARCHER UK National Supercomputing Service (<http://www.archer.ac.uk>). J.M.J. and T.B. acknowledge financial support from the Austrian Science Fund (FWF): \[P28211--N36\]. J. Winkler and B. Proft from Sachtleben Chemie GmbH (Duisburg, Germany) are acknowledged for kindly providing us with Nano-Rutile powder. Data relating to the theoretical calculations performed during this research are available by request from the University of York Research database <http://dx.doi.org/10.15124/fdde4827-db8d-4491-a309-9133e7d93899>.
| {
"pile_set_name": "PubMed Central"
} |
Introduction
============
Most coaches and professionals linked to sports teams agree on the importance of both an adequate climate and good social relations, and the high perception of skill among team members in order to optimize group efficacy. Thus, many expert investigators in group dynamics have identified cohesion and efficacy as two of the most important properties within a sports team ([@b6-jhk-34-129]; [@b8-jhk-34-129]; [@b13-jhk-34-129]; [@b20-jhk-34-129]; [@b22-jhk-34-129]; [@b29-jhk-34-129]). Notwithstanding, there are few studies that have attempted to analyze the evolution of these psychological variables over the season, as well as what kind of variables could influence this process ([@b14-jhk-34-129]; [@b12-jhk-34-129]).
Considering each of these concepts specifically, the term team cohesion is defined by [@b7-jhk-34-129] as "a dynamic process that is reflected in part by the tendency of a group to stick together and remain united in the pursuit of its instrumental objectives and/or for the satisfaction of member affective needs". As expressed in the definition, this concept is changing and variable over time and is affected by a series of environmental, personal, leadership, and team factors that promote the level of cohesion. If it is taken into account that one of team factors refers to the group's desire for success, this indicates that the greater these desires and higher expectations of success, the higher will be the degree of cohesion shown by the players to achieve their goals.
In addition, within the conceptual model of Carron et al., cohesion consists of four dimensions based on two levels of distinction. The first level distinguishes individual attractions to the group and group integration; the second level distinguishes the task and social aspects of group involvement. Thus, four constructs are identified: Group integration-task (GI-T), Group integration-social (GI-S), Individual attractions to the group-task (ATG-T) and Individual attractions to the group-social (ATG-S). In our study we only used one of the levels, because the aim was to examine the evolution of task cohesion aspects, which reflects the degree to which group members work together to achieve common goals, and social cohesion, which reflects the degree to which team members empathize with each other and enjoy the companionship of the group ([@b5-jhk-34-129]; [@b10-jhk-34-129]; [@b9-jhk-34-129]) in different moments of the season. These dimensions change as a function of the diverse above-mentioned factors. Therefore, as seen in previous studies, success expectations can have a close relation with levels of player-perceived cohesion over the season ([@b19-jhk-34-129]).
With regard to perceived efficacy, diverse studies have attempted to appraise this construct through players' and coaches' opinions ([@b1-jhk-34-129]; [@b2-jhk-34-129]; [@b11-jhk-34-129]; [@b17-jhk-34-129]; [@b20-jhk-34-129]). Thus, there are various types of efficacy related to the sport context ([@b4-jhk-34-129]), among which are noteworthy coach-perceived efficacy and teammate-perceived efficacy, self-efficacy, and collective efficacy. [@b2-jhk-34-129] first defined self-efficacy as an individual's belief in his or her capacity to successfully organize and perform a specific task. Coach-perceived efficacy is the coach's assessment of each player's capacities and skills with regard to the good development of the play actions ([@b11-jhk-34-129]), whereas teammate-perceived efficacy is the players' belief in the capacities of their teammates to adequately meet the demands of the sport ([@b17-jhk-34-129]). Lastly, collective efficacy is defined as the group's shared beliefs in their capacities to organize and perform the actions required to achieve certain goals ([@b2-jhk-34-129]).
As with team cohesion, the perception of efficacy will evolve as a function of a series of antecedents. These sources of information may be past performance, success expectations, the group's physiological state, leadership, the motivational climate, team cohesion, etc. ([@b2-jhk-34-129]). Thus, as previously noted by [@b19-jhk-34-129], among the diverse antecedents of efficacy, success expectations, understood as the belief in the attainment of a series of goals or sports success, can play an important role in the perception of efficacy ([@b26-jhk-34-129]).
The relevance to examine these aspects is due to the narrow relationship between cohesion and different types of efficacy ([@b12-jhk-34-129]; [@b16-jhk-34-129]; [@b20-jhk-34-129]; [@b24-jhk-34-129]; [@b25-jhk-34-129]; [@b28-jhk-34-129]) and these variables with respect to performance ([@b8-jhk-34-129]; [@b13-jhk-34-129]; [@b19-jhk-34-129]; [@b22-jhk-34-129]; [@b29-jhk-34-129]). In accordance to this issue, several works have shown that players who perceived greater cohesion level and perceived efficacy achieve higher performance at the end of the league ([@b8-jhk-34-129]; [@b13-jhk-34-129]; [@b19-jhk-34-129]; [@b22-jhk-34-129]; [@b29-jhk-34-129]). Therefore, it might be interesting to know how these variables progress during a season, and how success expectations, which have been shown as cohesion and efficacy antecedent might influence those variables ([@b4-jhk-34-129]; [@b19-jhk-34-129]). Moreover, it is important to note that this line of research has not been pursued much in the sport sphere, although some investigations have indirectly examined cohesion and efficacy in different moments over a season ([@b14-jhk-34-129]; [@b12-jhk-34-129]; [@b21-jhk-34-129]).
The most relevant studies about the evolution of cohesion and efficacy were carried out in diverse sports such as handball ([@b14-jhk-34-129]; [@b12-jhk-34-129]), and basketball ([@b14-jhk-34-129]). These works present a similar line of results: [@b14-jhk-34-129] and [@b12-jhk-34-129] found that all the factors of cohesion decreased their levels as the season advanced and approached the end. Likewise, the levels of players' perceived collective efficacy were higher at the start than at the end of the season ([@b14-jhk-34-129]; [@b12-jhk-34-129]; [@b21-jhk-34-129]). Our research at the same time that shows new information about evolution of those variables in football players, aims to examine whether success expectations perceived at the beginning of the season might influence the perception of cohesion and efficacy at the end of the league. Unlike most studies, this checking was developed through the composition of two groups regarding whether expectations was achieved or was not achieved and how dependent variables changes in both groups during the league.
Thus, the main goal of the study is to examine the evolution of players' perception of cohesion and efficacy over the season and their relation with success expectations. Hence, as first hypothesis, we propose that the diverse factors of cohesion, perceived efficacy, and success expectations will undergo significant changes over the season. The second hypothesis states that success expectations will emerge as an antecedent in the evolution of cohesion and efficacy.
Material and Methods
====================
Participants
------------
The research sample comprised 265 male soccer players, aged between 15 and 19 years (*M* = 16.96, *SD* = 0.76). All the players who made up the sample belonged to the 15 federated teams that played in the XI group of the Sub18 National League, and each participant held a federative card with his personal and sports data. The final sample was formed by 146 players who completed the questionnaires at the start and at the end of the season, and the players who did not complete the two measurements, or who had completed them on different teams (due to a possible change during the season) were eliminated. The team coaches (*N* = 15), aged between 29 and 45 years, with at least 7 years experience, also participated in the study. The study received ethical approval from the University of Extremadura. All participants were treated according to American Psychological Association ethics guidelines regarding consent, confidentiality, and anonymity of responses.
Measures
--------
**Cohesion**. To assess cohesion we used the Spanish version of the Group Environment Questionnaire (GEQ: [@b10-jhk-34-129]), carried out by [@b15-jhk-34-129]. This instrument has 18 items grouped into four factors. Despite this issue, due that our aim was to analyze the evolution of social and task dimensions, we used the two global factors grouped items in *task cohesion* (9 items, i.e., "The team members unite their efforts to achieve the goals during the training sessions and the games") and *social cohesion* (9 items, i.e., "The team members like to go out together"). The items are rated on a 5-point Likert-type scale. In this study, we analyzed internal consistency through Cronbach's alpha coefficient, obtaining in the first measure values of .76 for task cohesion and .73 for social cohesion, and in the second measure values of .68 for task cohesion and .75 social cohesion respectively.
**Efficacy**. To measure self-efficacy, collective efficacy, and teammate- and coach-perceived individual efficacy, we elaborated a questionnaire based on the guidelines of [@b3-jhk-34-129] for all the dimensions, which has been used in other studies ([@b19-jhk-34-129],[@b20-jhk-34-129]). This questionnaire measures *self-efficacy*, in which each player rates himself; *collective efficacy*, in which each player rates the team's capacity; *teammate-perceived efficacy*, where each player rates all the other members' efficacy; and *coach-perceived efficacy*, in which the coach rates each player. All the items were grouped into a single main factor that includes perceived efficacy in all stages of the game. The items are responded on a 5-point Likert-type scale in all cases. The measurement was carried out in diverse phases of the game, valuing technical and tactical aspects in the phase of attack and defense (i.e., "How do you rate yourself in the defense phase at a tactical level?"), the physical and psychological aspects (i.e., "How do you rate yourself in the mental and psychological aspects?"), and a last item of general rating of the player in the game (i.e., "In general, how do you rate yourself as a player?"). In the initial and final measurements, the scale yielded alpha values of ,73 and ,76 for collective efficacy, ,85 and ,80 for self-efficacy, ,80 and ,77 for teammate-perceived efficacy, and ,86 and ,87 for coach-perceived efficacy. All the factors obtained adequate internal consistency ([@b23-jhk-34-129]).
**Success expectations**. To measure success expectations, we asked each player at the beginning and near the end of the season about the position that he believed he would occupy in the classification at the end of the season, with a response range of 1 to 16.
**Classification**. To assess the final performance of the team, we used the final position of each team in the classification table at the end of the regular league.
**Expectations/Classification**. To assess the degree to which each team perceived expectations matched the final position achieved, we decided to use a quotient between each teams' success expectations, using average player's perceived matched at the start of the season and the final team position in the classification. As a result, the closer score was to 1, the more closely the expectations at the beginning matched the final team classification; and, contrariwise, the closer the score was to 0, the farther were the expectations from the final result achieved. But this indicator is greater than 1 if the final ranking of a team is better than success expectations that a player indicated at the beginning of the league.
Procedure
---------
The study was carried out using a correlational methodology, with a longitudinal design of evolutionary analysis which consisted of two measurements at two different points in time, analyzing a subpopulation or specific group across the time interval. The measurements were carried out at the start of the season (Measurement 1) and at the end of the season (Measurement 2) with approximately 20--22 weeks between them. Measurement 1 was carried out during the first third of the competition to ensure that the teams had competed together in several official games and within a three-week interval. Measurement 2 was carried out in the last third of the season, following the same guidelines used for Measurement 1.
With regard to the measurement procedure, first, the main investigator of the study contacted each one of the coaches of the teams that could participate in the investigation to request the inclusion of their teams in the study. They were informed about the goals and the procedure of the study that would be carried out if they agreed to participate in the investigation. In total, 15 out of the 16 teams that make up Group XI of the Sub18 League agreed to participate in the first and second administration of the questionnaires. The players were also informed about the goals of the investigation, emphasizing that their participation was voluntary and that their responses would be confidential.
To collect the data, we developed a protocol to ensure that data obtention would be similar in the two measurements and in all the participants involved in the investigation. The participants completed the questionnaires before the training session to avoid possible alterations due to the interactions that could arise during the sessions and that could affect the measurements. Measurements were conducted in the changing room, without the presence of the coach, individually, in an appropriate climate that allowed the players to concentrate without any kind of distractions. Completing the questionnaires took approximately 20 minutes; the main investigator was present at all times and emphasized that the players could ask for clarification of any doubts that might arise during the process.
Analysis
--------
Data was analyzed using the SPSS 18.0 software for diverse types of analyses to determine the relations among the variables. Firstly, various tests were conducted to determine the nature of the data. We used the K-S test for independent samples to verify the normality of the groups, the runs test for randomness, and Levene's test for the homoscedasticity or equality of variances. As the data were shown to be parametric, we applied parametric tests in the data analysis. The techniques used for the study were factor analysis, reliability analysis, descriptive analysis, *t*-tests for related samples, and discriminant analysis.
Results
=======
Descriptive statistics and comparison of means in two related samples
---------------------------------------------------------------------
[Table 1](#t1-jhk-34-129){ref-type="table"} shows the descriptive statistics of the variables of the study at both measurement times. In general, the means of the components of cohesion and of efficacy are high both at the start and at the end of the season. Only coach-perceived efficacy increased its scores at the end of the season. The participants expressed high success expectations at the start of the season but these scores decreased at Measurement 2.
To determine changes in the diverse variables over the season, we conducted a *t*-test for related samples ([Table 1](#t1-jhk-34-129){ref-type="table"}). In general, significant changes in the factor task cohesion were observed (*p* \> .00), with higher scores at Measurement 1 (at the start of the season) than at Measurement 2 (at the end of the season.
Similarly, both the levels of collective efficacy and of teammate-perceived efficacy decreased as the season advanced (*p* \> .05). In contrast, significant differences were observed in coach-perceived efficacy (*p* \> .03), with higher levels at Measurement 2 than at Measurement 1.
Lastly, there were significant differences between the measurements of success expectations (*p* \> .00). That is, the players had higher expectations at the start of the season, and their levels dropped at the end of the season. Therefore, at the beginning of the league, the players thought they would have a good season because they had high expectations, but at the end of the season, their expectations better matched what actually occurred.
Analysis of differences
-----------------------
In view of the decrease in the levels of the variables analyzed, we decided to establish two groups as a function of the variable expectations/classification, differentiating the players whose expectations matched the performance (EMP), that is, their expectations at the start of the season in the classification table approached the final team classification, and the players whose expectations did not match the performance (ENP), that is, their expectations in the classification table distanced from the final team classification. Thus, by means of the median (*Md* = 0.72), the sample was divided into two groups: the set of data lower or equal to the median represents 50% of the players, and the data higher than the median represents the other 50% of the total sample.
In order to determine the changes in the variables over the season, we carried out a t-test for two related samples as a function of the variable expectations/classification. [Table 2](#t2-jhk-34-129){ref-type="table"} shows the significant changes in the factors of social cohesion (*p* \> .04) and task cohesion (*p* \> .00), in both groups. In both cases, there are significant differences, with higher scores at Measurement 1 (at the start of the season) in comparison to Measurement 2 (at the end). However, there was no significant evolution in the cohesion variables of the players of the EMP group.
A similar effect was noted when analyzing the values of perceived efficacy, where both the levels of collective efficacy (*p* \> .03) and of teammate-perceived efficacy (*p* \> .04) decreased significantly between the start of the season and the end of the season in the ENP group. In contrast, no significant evolution was observed in the players of the EMP group in these variables, which remained stable.
Discriminant analysis
---------------------
To determine the set of variables that maximize the differentiation of the groups and help to predict matching expectations to performance, we used discriminant analysis. For this purpose, the variables that showed significant mean differences as a function of the expectation groups created (EMP and ENP) were entered in the discriminant analysis. The analysis showed that the statistics that analyze the significance of the discriminant function are adequate (Wilks' λ =.839, *χ*^2^ = 24.875, *df* = 5, *p* \< .01)
[Table 3](#t3-jhk-34-129){ref-type="table"} shows the structure matrix created after the discriminant analysis. There are five factors that can discriminate the match between expectations and performance (those with a coefficient higher than .30). The factor with the highest discriminate capacity was social cohesion measured at the end of the season, with a structure coefficient of 0.64. Next was teammate-perceived efficacy, which was a negative predictor, that is, it discriminated the subjects whose expectations did not match the performance. Next came social cohesion measured at the start of the season and collective efficacy measured at the end of the season; both of them were positive discrimination, i.e., they discriminated the players whose expectations matched the performance. Lastly, was self-efficacy measured at the start of the season, which, as with teammate-perceived efficacy, was a negative predictor.
Of all the cases, 71.9% were correctly classified, which means that this function (and the variables that form it) is capable of predicting a very high percentage of cases.
Discussion
==========
The main goal of the study was to examine the evolution of players' perception of cohesion and efficacy over the season and their relation with success expectations, and for this purpose, we examined the three hypotheses previously presented.
The first hypothesis proposed that the diverse factors of cohesion, perceived efficacy, and success expectations would undergo significant changes over the season. In this sense, in general, the scores of all the variables decreased over the season, that is, the levels of cohesion and perceived efficacy dropped as the end of the league drew near.
Significant changes were observed in the main factor of task cohesion, with higher scores at Measurement 1 than at Measurement 2. Taken a priori, these results do not seem logical as the first measurement was carried out at the start of the season when the players barely knew each other, and one would expect an increase of both variables as the season advanced. However, in most of the investigations ([@b14-jhk-34-129]; [@b12-jhk-34-129]), as in this study, the first measurement was taken after starting the league, after three matches, when the levels of cohesion are very high because the players have more hopes at the start of the season and they unite their efforts to achieve greater heights. Therefore, the players have a greater predisposition to persevere to achieve a better performance. The levels of cohesion are lower at the end of the season because at this time, the players perceive the goals to be achieved either as closer or farther away, and the variables that unite the group to seek a better performance no longer seem so important. Similar results were found by other authors in sports such as handball ([@b12-jhk-34-129]) and basketball ([@b14-jhk-34-129]), and they note that all the cohesion factors decrease their levels as the season advances and the end is nearer.
In the same vein, the levels of collective efficacy and of teammate-perceived efficacy decrease significantly from the start to the end of the season. The significant differences in success expectations between both measurements are also noteworthy, with expectations at the start of the season being higher than at the end. The same reasons used to justify the decrease in the levels of cohesion are valid to justify the fact that the scores of collective efficacy, teammate-perceived efficacy, and success expectations also decrease at the end of the season. The simple fact of performing the measurement at the start of the season increases the teams' success expectations. Their desire to achieve the initially proposed goals also makes them perceive higher levels of efficacy in their teammates and in the group. At the end of the season, all the scores decrease, because the players' perceptions of efficacy and expectations are more objective, that is, more realistic. Various authors like [@b14-jhk-34-129], [@b12-jhk-34-129] and [@b21-jhk-34-129] also observed a decrease in the levels of collective efficacy from the start to the end of the season.
In contrast, in coach-perceived efficacy, we observed significant differences, with higher levels at Measurement 2 (at the end of the season). This may be due to the fact that, at the start of the season, coaches are more cautious about their team's efficacy and also, coaches always have a more objective and realistic view of their players' efficacy.
Taking the above comments into account, the first hypothesis is confirmed, verifying that the levels of cohesion, perceived efficacy, and success expectations changed along the season, decreasing as the end of the league approached.
Due to the decrease in the levels of the variables of the study, we differentiated the players whose expectations were met from those whose expectations exceeded their final performance. Thus, we proposed the second hypothesis, which stated that success expectations would be a determinant of the evolution of cohesion and efficacy.
Firstly, we observed significant changes in the two main factors of cohesion, social and task cohesion, in the ENP group. In both cases, there were significant differences, with higher scores at Measurement 1 than at Measurement 2. This may be due to the fact that the goals proposed at the start of the season---such as staying in the category, being in the middle of the table, being at the top, etc.---may be unattainable or can no longer be attained, so the levels of cohesion drop. However, we did not observe any significant evolution of the cohesion variables in the players of the EMP group because, at this time of the season, they may still have been struggling to achieve their goals.
Therefore, when analyzing the values of perceived efficacy, both the levels of collective efficacy and of teammate-perceived efficacy decreased significantly from the start to the end of the season in the ENP players. In contrast, no significant evolution was observed in these variables in the EMP players because, as they were still struggling to attain the proposed goals, their perceptions of team efficacy and teammate efficacy were still high. Hence, we corroborated that the levels of cohesion and perceived efficacy decreased in the ENP players, whereas the EMP players maintained their degree of cohesion and perceived efficacy.
Lastly, discriminant analysis was used to determine which variables best discriminated between the EMP and ENP groups. After analyzing the results, we observe that both social cohesion and collective efficacy at Measurement 2 (end of the season) allow us to classify the players into the EMP and ENP groups. At the start of the season, social cohesion clearly discriminates the players whose expectations match the final performance and the players whose expectations do not match the final classification. In contrast, teammate-perceived efficacy and self-efficacy had a negative discrimination. Therefore, cohesion and collective efficacy, which are collective perceptions, discriminate which players will be classified as the EMP group, whereas individually perceived efficacy (self-efficacy, coach-perceived efficacy and teammate-perceived efficacy) discriminates the players who will be classified as the ENP group. In accordance to this, we can test the second hypothesis, which states that success expectations will emerge as a determinant in the evolution of cohesion and efficacy.
Thus, the main conclusion reached in this study is that the coaches should attempt to clarify the main goals of the season, both personal and collective, for the players ([@b18-jhk-34-129]; [@b27-jhk-34-129]), in order to create expectations that match the team's possibilities. If the players' expectations exceed the team's possibilities, then their levels of cohesion and perceived efficacy might decrease and therefore, their performance should also decrease. If the goals to be attained are clearly defined, each player's expectations will match those of the group, and in this way, the levels of cohesion and efficacy should be more constant and the team's final performance might be better.
In order to reaffirm these conclusions, it might be interesting to present some limitations of the study. In this regard, this work is developed with players at a learning stage and only two measurements throughout the season, so lot of information during the league that might influence these results might have been overlooked. Furthermore, only relationships at inter - individual level have been developed, so the association between variables at intraindividual and inter -- teams level might give us valuable information regarding this research topic.
Therefore, it would be interesting to perform investigations with professional teams, because their orientation towards performance is much higher, and this might prevent the slight decrease in the number of players between the two measurements due to their absence from the training sessions. Moreover, three measurements could be performed throughout the season, which would provide another view of how the variables fluctuate over time. Finally, it might be important to perform works through multilevel analysis, because it would offer deeper relationships between those variables at intraindividual, inter -- individual and overall, inter -- teams level.
Another important prospective to follow in further investigations will be to develop studies under experimental methodology, to test whether working through success expectations might influence cohesion level and perception of efficacy by group, which might lead to an important finding in order to elaborate psychological issues at the highest sports level.
The authors would like to thank the Fundación Fernando Valhondo Calaff for its financial support, as a result of which this investigation could be carried out.
######
Means, standard deviations and analysis of differences at Measurements 1 and 2
Measurement 1 Measurement 2 Differences Measurements 1 and 2
----------------------------- --------------- --------------- ---------------------------------- ------ -----
Social cohesion 4.00 .66 3.91 .67 .14
Task cohesion 3.76 .69 3.58 .68 .01
Collective efficacy 3.80 .56 3.71 .55 .03
Self-efficacy 3.91 .61 3.88 .51 .51
Teammate-perceived efficacy 3.75 .49 3.65 .43 .04
Coach-perceived efficacy 3.61 .68 3.74 .75 .03
Success Expectations 14.47 2.28 12.97 3.13 .00
######
Means, standard deviations, and analysis of differences at Measurements 1 and 2 as a function of high and low group expectations/performance
EMP ENP
----------------------------- ------ ----- ------ ----- ----- ------ ----- ------ ----- -----
Social cohesion 4.08 .67 4.07 .62 .93 3.92 .64 3.75 .69 .04
Task cohesion 3.69 .68 3.62 .78 .43 3.98 .77 3.69 .71 .00
Collective efficacy 3.82 .58 3.77 .56 .38 3.78 .53 3.64 .54 .03
Self-efficacy 3.92 .56 3.81 .55 .12 3.96 .70 3.83 .62 .19
Teammate-perceived efficacy 3.65 .47 3.60 .50 .42 3.86 .48 3.69 .36 .04
Coach-perceived efficacy 3.72 .74 3.73 .78 .94 3.51 .57 3.66 .77 .11
EMP = Expectations match performance; ENP = Expectations do not match performance.
######
Structure matrix of the discriminant analysis of the expectations/performance match
Function 1
------------------------------------- ------------
Final social cohesion .64
Initial teammate-perceived efficacy −.44
Initial social cohesion .39
Final collective efficacy .33
Initial self-efficacy −.32
[^1]: Authors submitted their contribution of the article to the editorial board.
| {
"pile_set_name": "PubMed Central"
} |
Introduction
============
There is an established literature on gender differences in humor perception and humor styles. Men have been noted to prefer humor that has sexual or aggressive themes whereas women appear to prefer neutral or absurd humor ([@B1]). Whereas earlier studies showed that sexist humor (i.e., humor that upholds gender role stereotypes) is preferred over non-sexist humor ([@B4]), other studies report that both men and women prefer humor that has the opposite gender as the butt ([@B33]; [@B23]). Furthermore, men typically rate themselves higher than women in humor initiation whereas women tend to rate themselves higher in humor appreciation, but when humor is studied in actual conversational contexts a more nuanced picture emerges (see [@B15]; [@B13], [@B14]; [@B26]). Similarly, whereas some studies have found that humor produced by men is judged to be more humorous than that produced by women ([@B3]), other studies have not found this effect ([@B12]), and still other work suggests a bias operating, whereby men are perceived to be the "funnier sex" regardless of how their humorous creations are actually judged ([@B20]; [@B10]).
Taken as a whole, the literature on gender and humor eludes easy generalization (see [@B18], for a review). Methodologically, early studies have been criticized for their use of decontextualized or "canned" humor samples instead of spontaneously generated humor that arises naturally in conversation ([@B8]; [@B16]). The literature on gender differences in humor preferences has also been critiqued for its reliance on classifications of humor type (as "hostile" or "sexual" or "sexist") based on experimenter intuitions rather than eliciting participants' own perceptions of humor type, which may not coincide with those of the experimenter (e.g., [@B23]). Moreover, studies that have sought to measure the construct of a sense of humor have led to many promising instruments, such as the three Witz-Dimensionen (3WD) instrument by [@B25], and to new adaptations of established instruments for use with non-English speakers (see [@B22]). At the same time, it is recognized that for a construct as slippery and contextual as humor, it is important to consider multiple, converging measures across different groups and settings.
This recognition of the complexity of studying humor, together with a growing shift in regarding gender as performative, has led to a shift in humor scholarship in the direction of studying humor as it is enacted by men and women in a range of social contexts (e.g., [@B9]; [@B6]), and in a range of laboratory contexts. Our own previous work has explored the relationship between cognitive, neurocognitive, and psycholinguistic aspects of humor detection and comprehension (e.g., [@B28]; [@B34]; [@B32], [@B35]; [@B11]; [@B17]). Our work on humor production has sought to develop controlled ways of eliciting humor to study its cognitive and social underpinnings. For example, we developed a concept comparison task in which participants were asked to produce "catchy" ways in which the concepts were related, which invariably elicited humorous responses, e.g., MONEY and CHOCOLATE: *one swells the wallet, the other, the hips* ([@B12]). Another task involved generating rejoinders to proverbs, e.g., *Absence makes the heart grow fonder, but also makes the eyes wander* ([@B30]). Other prior work in our laboratory has examined the role of culture in judgments about when humor (vs. silence) is an appropriate response to embarrassing situations encountered in daily life ([@B31]). Finally, we have examined how individuals' perceptions of their own humor styles compare with their perceptions of humor styles of members of their gender category and/or same or different cultural group ([@B24]; [@B27], [@B29]).
As an extension of our interest in gender and cultural dimensions of humor, the aim of the present research was to characterize how gender and country of origin (as a proxy for culture) may shape how individuals conceptualize an ideal sense of humor. The motivation for this study was a previous study which examined the role of gender in lay conceptions of an ideal sense of humor ([@B7]) in a large sample of United States-based participants of different ages and backgrounds. Participants in this study were asked to provide a brief narrative describing the humor characteristics of a person they knew who embodied an outstanding sense of humor. [@B7] reported that a majority of the participants identified a male figure as the person who embodied an outstanding sense of humor. Indeed, of the 141 respondents (49 men, 92 women), nearly 84% of men and 67% of women selected a male figure. The researchers also classified the humor characteristics mentioned into five categories: creativity (witty, clever, quick comeback), caring (humor used to put others at ease), real life (grounding the humor in real life experiences), jokes (having a repertoire of jokes), and hostility/sarcasm (satirical, biting humor) and noted that creativity, caring and real life were mentioned most often, and that there were no discernible differences in the weighting of these characteristics as a function of either participant gender or target gender.
Over 25 years have passed since the [@B7] study. While gender continues to be a salient element structuring society, women have also become more visible in a number of domains of public life, including in the realm of comedy. It is possible that gender stereotypes may have become less entrenched in the present day. We therefore wondered if the preference for a male figure as the embodiment of an outstanding sense of humor noted previously still holds among young adults in the present age. We also wondered whether individuals from other countries would show a similar preference, given that they might be less likely to be influenced by Western gender stereotypes (including stereotypes regarding men as being the canonical humor initiator), but might have their own cultural stereotypes about humor, gender, and the relation between the two. Although there have been a few prior studies of humor stereotypes in different nationalities, the focus of our study was on how individuals from the United States compared to those from two other countries in articulating characteristics of an ideal sense of humor, as embodied in someone they knew. Our interest was to uncover patterns of commonalities as well as differences across groups and across genders.
In searching the literature, we could find only one other empirical study conducted since the study by [@B7] that used their open-ended prompt. This study, by [@B21], was conducted on men and women in Singapore. It, too, found that the embodiment of an outstanding sense of humor was male. Of the 18 men and 46 women in the study, 76% of respondents selected a male target ([@B21]). The researchers further noted that the preference for a male target was more pronounced in men, but no additional analyses were reported in terms of specific humor characteristics mentioned by men and women. Thus, we felt another study was warranted.
The Present Research
--------------------
Our study had two goals. The first was to investigate if the male preference first reported by [@B7] still holds. To examine this, we pooled data from United States-based college students tested from 2004 to the present. The second goal was to investigate if the pattern of a male preference as the embodiment of an ideal sense of humor is restricted to United States participants or is generalizable to other samples. In particular, we considered samples drawn from Iran and from Turkey, as these particular groups have been understudied in the humor literature; where country-based differences have been studied, they have either tended to be within north American/European samples or have compared Western with east Asian samples (e.g., China). Turkey is considered geographically and culturally as a bridge between Asia and Europe. Thus, we aimed to compare participants raised in a Western (American), a Middle Eastern (Iranian), and a blended (Turkish) culture. We did not have *a priori* expectations of how participants across the three groups would respond on the task; our study is exploratory with regard to the cultural dimension, as our sample sizes were limited and varied in other respects (e.g., age) and we recognize that much more follow up investigation would be needed to fully understand the nature of any differential patterns uncovered.
Materials and Methods {#s1}
=====================
Participants
------------
Participants included male and female United States born (American) and international students (born in Iran or in Turkey) recruited from a university town in the southwestern region of the United States, from a university in Istanbul, and from online responses. The American sample consisted of 279 undergraduate students (including 201 women) who ranged in age from 18 to 23 years, with a mean of 21 years. The majority self-identified as white, and the numbers of Latinx, African American, or Asian Americans were too few to permit separate subgroup analyses. The Iranian sample comprised 71 participants (47 women) who ranged in age from 17 to 54 years, with a mean age of 31.32 years, and the Turkish sample consisted of 79 undergraduate students (48 women) ranging in age from 18 to 25 years with a mean of 21.7. The American and Turkish participants completed the task as part of a class activity; the Iranian sample was recruited by placing an announcement in social media and participants completed an online version of the task. All participants received and answered the prompt in their primary language. The Iranian and Turkish data were translated into English by native Farsi- and Turkish speakers who had advanced English proficiency. Most of the data were coded by the same researcher (with gender of participants masked) to provide consistency in coding. A subset of the data were also intercoded to ensure some level of consensus (at least 80%).
Materials and Procedure
-----------------------
Participants were given a response sheet on which they were to write a brief narrative in response to an open-ended prompt adapted from [@B7] study. They were instructed to think of a specific individual they knew who had an outstanding or ideal sense of humor and then to describe the characteristics of that humor, using three to five descriptors. They were then asked to describe the person who embodied that humor (we refer to this person as the *humor target*), indicating, for example, whether it was a family member, a friend, co-worker, or a comedian, and/or noting their gender, age, and ethnicity. There were no time constraints for responding. Since not all respondents stated the humor target's gender, this information sometimes had to be inferred from the stated relation to the target (e.g., brother, sister, girlfriend, particular celebrity, etc.) or from the participants' choice of pronouns in describing the person (however, this approach was helpful only for the English dataset as pronouns in Farsi and Turkish are not marked for gender).
Data Analyses
-------------
Two sets of comparisons were conducted using chi square and regression analyses. The first examined percent mention of the target gender by participant gender and country. The second examined percent mention of each of the five categories of humor descriptors identified by [@B7] in relation to participant gender and country.
The five coding categories were as follows: **Creativity:** This characteristic includes terms referring to creative aspects of humor, like *witty, quick comeback, playing with language, clever*, as well as being spontaneous or natural. An example of this characteristic from our sample is "very quick in answering with a witty comment." **Caring:** This characteristic indicates the kind of humor that makes people laugh and helps to change their mood when they are upset or in a tough situation. An example of this characteristic is "their humor helps relieve the tension." **Real Life:** This characteristic shows the ability of the humorous person to tell stories and recount real life events in a humorous way. An example of this dimension is "a great story-teller to bring out humor." **Jokes:** This characteristic refers to the use of actual jokes. An example of this dimension is "holds the crowd's attention with a simple joke." **Hostility/Sarcasm:** This category consists of attacking, insulting, and destructive humor as well as sarcasm. An indication of this characteristic is "can come up with the worst sexist insult."
Results
=======
A summary of the relative distribution of target gender of the ideal humor person is provided in **Table [1](#T1){ref-type="table"}** by participant gender and country of origin. Also included in the table are the number of participants per group for whom humor target gender was not specified. The latter comprised 21.81% of the American sample, none of the Turkish sample, and 59.15% of the Iranian sample.
######
Gender distribution of humor target per participant gender and country.
Country of origin Participant gender Target gender
------------------- -------------------- --------------- ----- ----
American Female 45 118 38
Male 5 50 23
Total 50 168 61
Turkish Female 15 33 0
Male 2 29 0
Total 17 62 0
Iranian Female 6 13 28
Male 3 7 14
Total 9 20 42
Identified Target Gender by Country of Origin
---------------------------------------------
A chi square analysis was done excluding those whose target gender was unspecified to compare the relative percent mention of a male vs. female humor target, collapsed across participant gender. The analysis showed no significant effect of country of origin, χ^2^= 0.33, *p* = 0.85, *N* = 331. That is, regardless of their country of origin, participants showed a consistent tendency to select a male figure as their humor ideal: 77.1% of Americans, 78.5% of Turkish, and 73.5% of Iranian participants identified a male.
Identified Target Gender by Country of Origin (American vs. Turkish vs. Iranian) and Participant Gender
-------------------------------------------------------------------------------------------------------
A logistic regression was conducted to see if the gender of the ideal humor target person could be predicted based on the participants' gender or the participant's country of origin (American, Turkish, Iranian). Again, only participants whose responses indicated the gender of their humor ideal were included in the analysis. Dummy coding was applied for the analysis. The model was significant, χ^2^= 13.78, *p* = 0.003, *df* = 3 and explained 6.2% of the variance. Gender of participant was a significant predictor of gender of humor target (χ^2^= 11.08, *p* = 0.001, *odds ratio* = 0.29): male participants were more likely than female participants to select a male target as the embodiment of an ideal sense of humor (89.6% vs. 71.9%, respectively). Country of origin, on the other hand, was not a significant predictor (χ^2^= 0.32, *p* = 0.85) (Turkish vs. Iranian: χ^2^= 0.13, *p* = 0.71, *odds ratio* = 1.19; American vs. Iranian: χ^2^= 0.31, *p* = 0.57; *odds ratio* = 0.94; American vs. Turkish: χ^2^= 0.04, *p* = 0.84, *odds* ratio = 0.79). See **Figure [1](#F1){ref-type="fig"}** for a depiction of the percent mention of male targets per participant gender and group.
{#F1}
Identified Target Gender by Time Period and Participant Gender -- American Sample Only
--------------------------------------------------------------------------------------
A logistic regression was conducted on the American sample to see if there was a difference related to time at testing in the percent mention of a male target by men and women. Here, [@B7] were compared with data from the American sample (which was collected over two different time periods, 2004 and 2014).
The model was significant, χ^2^= 14.97, *p* = 0.002, *df* = 3 and explained 6.9% of variance. There was not a difference between the American 2014 and the 1991 data. However, the American 2004 data showed a difference than both the 1991 data, χ^2^= 6.33, *p* = 0.012, *B* = -1.12 and the 2014 data, χ^2^= 4.88, *p* = 0.027, *B* = -1.01. Participants from the 2004 sample (89.6%) revealed more male favored results than the 2014 sample (73%) and than the original study sample (73%).
Participants' gender was also a significant predictor, χ^2^= 5.29, *p* = 0.021, *B* = -0.767. That is, the selection of a male humor ideal was significantly higher when the participant was a male than when the participant was a female. In the original study male participants' preference for a male target was 83.7% and female participants' preference for a male target was 67.4%. In our study, male preference for a male target was 90.9% while female preference for a male target was 72.4%.
### Analyses of Ascribed Humor Characteristics
An additional set of analyses was conducted on the influence of participant gender on relative mention of each of five characteristics of an ideal sense of humor. (A preliminary analysis that included target gender as an additional predictor yielded no effect of this variable and so we do not report it here.) **Table [2](#T2){ref-type="table"}** provides a summary of the relative mention of each characteristic by male and female participants in each of the three groups. Note that these values represent all of the data per group, including those for whom target gender was not specified.
######
Relative mention of each characteristic by each gender participant in each of the three groups.
Characteristic Country Male Female
---------------- ----------- ------ --------
Creativity Americans 56.4 66.7
Iranians 45.8 44.7
Turkish 45.2 50
Caring Americans 32.1 38.3
Iranians 12.5 29.8
Turkish 3.2 2.1
Real life Americans 21.8 32.3
Iranians 12.5 27.7
Turkish 9.7 2.1
Jokes Americans 15.4 17.9
Iranians 8.3 10.6
Turkish 29 20.8
Hostility Americans 35.9 42.8
Iranians 0 0
Turkish 12.9 22.9
Numbers show percent mention
.
### Humor Characteristics by Participant Gender and Country of Origin
Inspection of the relative percent mention of the five humor characteristics shows an overall predominance of mention of the creativity characteristic by men and women and across all groups. For Americans, the next most mentioned characteristic was hostility/sarcasm, followed by caring. For Iranians, by contrast, the order of mention of the five characteristics was: creativity, real life and caring, and for the Turkish sample, the order was creativity, joke and hostility/sarcasm (see **Figure [2](#F2){ref-type="fig"}**).
{#F2}
A multivariate regression analysis was conducted to jointly examine the effect of participant gender and country of origin (American, Turkish, and Iranian) with each of the five humor characteristics (creativity, caring, real life, jokes, and hostility/sarcasm) considered as separate dependent variables. The results demonstrated that overall, both participant gender (χ^2^= 2.67, *p* = 0.022, *df* = 5) and country of origin (χ^2^= 22.15, *p* \< 0.001, *df* = 5) were significant predictors. However, no gender-specific effect was observed in any of the five humor characteristics. The results demonstrated that gender was a multivariate phenomenon, but gender did not specifically predict any of the five characteristics. Country of origin was a significant predictor for creativity (χ^2^= 10.30, *p* = 0.001, *R^2^* = 0.03, *odds ratio* = 0.905), caring (χ^2^= 13.39, *p* \< 0.001, *R^2^* = 0.04, *odds ratio* = 0.902), and hostility (χ^2^= 54.85, *p* \< 0.001, *R^2^* = 0.12, *odds ratio* = 0.816). Creativity was mentioned significantly more by American participants (64%) than by Turkish (48%) or Iranian participants (45%). Moreover, American (37%) and Iranian (34%) participants used caring to describe their ideal humor significantly more than did Turkish participants (2.5%). Further, hostility was mentioned significantly more by American participants (41%) followed by Turkish participants (19%) and it was mentioned least by Iranians (nearly 0%).
Discussion
==========
The aim of this study was to examine how men and women describe a specific person who embodies their ideal sense of humor. The study provided an opportunity to test whether the finding of a male target preference first noted by [@B7] for United States based participants and by [@B21] for Singaporean participants persists for Americans in the present period and is evident to the same extent among two other groups whose cultures are considered to be somewhat more traditional in terms of gender role stereotypes than American culture. Our findings show that the selection of a male as the embodiment of an ideal sense of humor was a pervasive and robust finding across the three samples we tested. Moreover, the size of this effect did not vary across the three groups. Of course, it is possible that the three samples we selected are on the gender inegalitarian end of the continuum and that had we selected a more egalitarian country we might not have found the effect. That remains for future work to test.
Our analysis of the United States samples tested at different periods of time further revealed that a male preference was actually somewhat stronger in the 2004 sample than it was for either the 1991 sample or a more recent 2014 sample. Perhaps the stronger male bias exhibited in the 2004 sample is a reflection of a public discourse in the country around that time regarding whether women can ever be as good at comedy as men. Nevertheless, it is interesting to note that, across all time periods sampled, the selection of a male target was significantly more likely when the participant was himself male. Thus, despite changes in societal consciousness about gender and humor that may have occurred (to differing degrees) over the past 25 years, there is a consistent preference for men to consider men as the embodiment of an ideal sense of humor. Moreover, this effect was found in the analysis by country of origin as well.
The finding that men are perceived as the embodiment of an ideal sense of humor may in part reflect an availability bias arising from the fact that male comedians and comedy writers still greatly outnumber female comedians and comedy writers. This difference in base rate may thus perpetuate a gender stereotype of men as the funnier sex and therefore prime people to think of men (rather than women) among their own acquaintances who exemplify an ideal sense of humor. Incidentally, among the American participants who provided information on their relationship to the gender target, a sizeable number (males and females) mentioned that the ideal humor person was their father. Further work should examine target gender demographic characteristics to provide insights into their relationship, if any, to the humor characteristics they embody.
Is there a difference in the types of characteristics used by men and women for male vs. female humor targets? Our analysis of the five dimensions noted by [@B7] to describe an ideal humor showed that the most frequently mentioned attribute by Americans in our sample is creativity, defined here as being witty, clever, and quick in coming up with a response. Creativity was mentioned by the majority of participants of both genders. The next most frequently mentioned dimensions for the American sample were hostility/sarcasm and caring. This may seem like an odd juxtaposition at first sight but it may not be that surprising given that the characteristic of "sarcasm" was coded under "hostility" and sarcasm (in American culture) is a way of interacting with one's friends. Although the Turkish and Iranian samples also chose creativity most often as a defining characteristic of an ideal sense of humor, they differed from each other and from the American sample in other characteristics: caring was mentioned by the Iranian and the American samples to the same extent but was mentioned hardly at all by the Turkish participants. By contrast, hostility/sarcasm was mentioned hardly at all by the Iranian sample. We do not wish to over-interpret the particular group differences obtained, as we did not have *a priori* expectations. We present them here as descriptive data, in need of further exploration.
Our findings corroborate the overall pattern noted in the previous study by [@B7] on which the present research was based -- namely, a preference for a male figure as the embodiment of an ideal sense of humor. However, the pattern of mention of the five different humor characteristics of the embodiment of an ideal sense of humor does not entirely concur with the pattern noted by [@B7]. As already noted, there were some clear differences across the three cultural groups in the relative frequency of mention of some of the five humor characteristics.
Moreover, we recognize that our analysis of humor characteristics may also have been influenced in a substantive way by whether the target they were thinking of was male or female. As brought up by one of the reviewers of our article, it is possible that people tend to interpret or remember a given behavior differently depending on whether it comes from a man or a woman. Because of stereotypes or violation of expectations, a joke made by a woman could be interpreted as mean whereas the same joke made by a man could be interpreted as funny. Alternatively, participants could be selectively remembering humorous statements that conform to their gendered expectations of women as being caring, and thus describe the ideal humor for a female target as more caring. Unfortunately, given the paucity of female targets in our findings, we could not analyze humor characteristics as a function of target gender, but this is an important issue for further work.
Furthermore, we confined our analysis to only five humor categories based on those used in the previous study, a number of responses mentioned by participants in our study were not easily classifiable within that coding scheme. For example, a number of participants referred to "being able to laugh at themselves/take a joke" as being a valued characteristic in the person with an ideal sense of humor. Similarly, a number of respondents emphasized that the humor of this person was "inappropriate" but "not mean-spirited." Or that it involved impersonation, funny facial expressions, etc. Based on these types of responses, it would be important in further work to do a more detailed content analysis than was possible in the present study. A need for additional investigation of this issue is underscored by our finding that a third of the Iranian sample's narratives and from 5 to 10% of the American and Turkish narratives do not refer to any of the five characteristics of humor analyzed in this paper, and instead mention other characteristics of ideal humor that were not captured by these five categories. The use of political humor in everyday discourse, humor addressing tensions between the traditional vs. the modern (e.g., [@B2]), or individual differences in the relationship between humor and psychological well-being (e.g., [@B19]) would be interesting to explore in further work. Other approaches to measuring everyday humor (e.g., [@B5]) are also worth incorporating in future investigations of the role of gender and culture in the perception of an ideal sense of humor (see also [@B36]).
Moreover, in addition to looking at humor type, it would be important to consider individuals' age as another factor that will likely influence what is considered desirable in a sense of humor. As people grow older, they might prefer caring or wise humor over hostile humor. In this regard, it may be relevant to point out that the Iranian sample -- which showed practically no mention of hostility/sarcasm -- had a broader age range among participants than did the other two samples.
A further limitation of our study is that while the prompt was intentionally made open-ended, this made for some messiness in coding, as there were some terms, e.g, *inappropriate, crude or vulgar*, that we treated as interchangeable for the purpose of our coding, but which may not have been perceived by participants as synonyms. Similarly, a number of participants used *sarcastic* as an attribute, and in the present coding scheme it was coded as hostility, but it might not have been considered by participants to be a negative attribute. Of course, it is possible that an ideal sense of humor is perceived to include darker elements in addition to positive elements.
Another limitation of this study is that we classified participants and targets solely on the basis of their assigned gender, and thus cannot say anything about perceptions of humor by and about individuals whose gender assigned at birth does not coincide with their gender at the time of testing, or who consider themselves non-binary with regard to gender. Relatedly, since we did not administer any measures to assess participants' gender identity or gender role attitudes, our data do not allow us to say anything about how participants' attitudes toward feminism or traditional gender roles may have informed their responses. Similarly, a gendered humorous persona might not be considered appealing to some participants in this "gender fluid" era.
These limitations notwithstanding, our study, using an implicit, elicited measure of lay conceptions of an ideal sense of humor, allows us to conclude that, in highly gender inegalitarian societies, the ideal sense of humor is strongly gendered in favor of male targets, especially among men. Further, this male preference is as firmly entrenched in contemporary American culture (at least for the young adult age range sampled in the present research) now as it was nearly 25 years ago. Third, a bias for thinking of men as the embodiment of an ideal sense of humor is not restricted to those in an American cultural context but is also found among members of two other nationalities. What is important to note is that these groups, despite being considered more "traditional," nevertheless did not show a stronger gender effect than that observed in the American sample. Finally, our results indicate that across all groups the most salient dimension of an ideal sense of humor is the ability to be witty, creative, and quick; this dimension is most pronounced for the American sample, who also appeared to value hostility/sarcasm and caring. The Iranian sample in turn appeared to value real life humor and caring, whereas the Turkish group placed least value on caring but instead emphasized jokes and hostility/sarcasm.
In further work it will be important to probe deeper into the social context of humor use in everyday life to determine not just what the characteristics are of an ideal sense of humor in an abstract sense, but how those characteristics are brought to life in different kinds of interactions.
Ethics Statement
================
This study was carried out in accordance with the recommendations of the Institutional Review Board guidelines with written informed consent from all subjects. All subjects gave written informed consent in accordance with the Declaration of Helsinki. The protocol was approved by the Texas A&M University Institutional Review Board Committee.
Author Contributions
====================
JV and ST contributed to the design of experimental stimuli and procedure. All authors contributed to the manuscript in terms of conception, implementation of experimental protocols, data collection and analysis, and editing of drafts.
Conflict of Interest Statement
==============================
The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
We thank Erica Dittmer, Paige Dusthimer, and Omar Garcia for assistance in data collection.
[^1]: Edited by: *Tracey Platt, University of Wolverhampton, United Kingdom*
[^2]: Reviewed by: *Jill Ann Jacobson, Queen's University, Canada; Kuba Krys, Institute of Psychology (PAN), Poland*
[^3]: This article was submitted to Personality and Social Psychology, a section of the journal Frontiers in Psychology
| {
"pile_set_name": "PubMed Central"
} |
Introduction {#s1}
============
Rabies virus (RABV) is a prototypical member of the *Lyssavirus* genus of the *Rhabdoviridae* family, and the causative agent of rabies. Rabies is associated with severe neurological symptoms and high mortality rate, causing more than 50,000 human deaths annually, mainly in Asia and Africa (Sudarshan et al., [@B49]). Although rabies has been studied for over 100 years, it remains incurable and a serious threat to human health. To develop effective antiviral drugs, a deeper understanding of the pathogenic mechanisms of RABV infection is required.
The life cycle of RABV occurs exclusively in the cytoplasm with the transcription of five viral genes that encode the viral proteins: nucleoprotein (N), phosphoprotein (P), matrix protein (M), glycoprotein (G), and the viral RNA polymerase (L), respectively. The N, P, and L proteins, together with the viral RNA genome, form a helical ribonucleoprotein (RNP) complex that is responsible for viral RNA transcription and replication (Albertini et al., [@B2]). The M protein recruits RNPs to the cell membranes and interacts with the G protein to facilitate virion budding and release (Mebatsion et al., [@B33]). Furthermore, the M protein has been reported to regulate the balance of viral RNA synthesis and be related to the pathogenicity of RABV (Finke and Conzelmann, [@B11]; Pulmanausahakul et al., [@B42]; Wirblich et al., [@B51]).
The degeneration of neuronal processes is considered to play an essential role in the fatal neurological symptoms of rabies (Baloul and Lafon, [@B5]; Scott et al., [@B45]; Kojima et al., [@B24]). An increasing number of studies indicate that microtubules are vital to the development and maintenance of axons and dendrites throughout the life of the neuron, and microtubule abnormalities are closely associated with neurodegenerative diseases (Matamoros and Baas, [@B30]; Stevenson et al., [@B48]). Thus, we hypothesized that the neuronal degeneration that occurs during RABV infection might be linked to microtubule abnormalities.
Microtubules are hollow, cylindrical structures composed of associated protofilaments of α and β-tubulin dimers and play important roles in multiple cellular processes, such as intracellular transport, signaling pathways, and cellular division (Downing, [@B9]; Hammond et al., [@B18]; Akhmanova and Steinmetz, [@B1]). The functional versatility of microtubules is dependent on microtubule dynamics, which are regulated by microtubule-associated proteins (MAPs; Akhmanova and Steinmetz, [@B1]; Alfaro-Aco and Petry, [@B3]). Among these, MAP4 binds to the microtubule lattice to stabilize microtubules, and thereby prevent depolymerization (Kadavath et al., [@B23]). Alternatively, stathmin depolymerizes microtubules by sequestering tubulin subunits and promoting microtubules shrinkage, so as that the growth of microtubules was blocked. This procedure was termed as "microtubule catastrophe" (Howell et al., [@B20]). In addition, microtubule dynamics have been found to be regulated by post-translational modifications, such as acetylation, phosphorylation, or palmitoylation of α-tubulin (Hammond et al., [@B18]; Song and Brady, [@B47]). Among these modifications, the acetylation which occurs on lysine 40 of α-tubulin is fairly common and considered to be a well-known marker of stable microtubules (L\'Hernault and Rosenbaum, [@B27]; Perdiz et al., [@B41]). Furthermore, α-tubulin can be deacetylated by histone deacetylase 6 (HDAC6), a primary α-tubulin deacetylase (Matsuyama et al., [@B31]; Yang et al., [@B54]).
Numerous viruses have been reported to utilize microtubules to facilitate the intracellular transport of virions or subviral particles, including human immunodeficiency virus type 1 (HIV-1), vaccinia virus, herpesvirus, and circovirus (Sanderson et al., [@B44]; Nishi et al., [@B37]; Pasdeloup et al., [@B40]; Cao et al., [@B8]; Fernandez et al., [@B10]). In addition, several viruses have been reported to take advantage of microtubules to regulate the formation of viral inclusion bodies, such as reovirus and orthopoxvirus (Parker et al., [@B39]; Howard and Moss, [@B19]). Conversely, some viruses induce microtubule depolymerization to facilitate viral infection, \[e.g., rotavirus, Epstein-Barr virus, and Sendai virus (SeV)\] (Ogino et al., [@B38]; Martin et al., [@B29]; Liu et al., [@B28]). Although a recent study demonstrates that α-tubulin is incorporated into RABV particles (Tu et al., [@B50]), the role of the microtubule cytoskeleton in the process of RABV infection, especially regarding viral transcription and replication remains poorly understood. Therefore, the aim of this study was to investigate: (1) the effects of RABV infection on the microtubule cytoskeleton; and (2) the role of microtubule dynamics in viral transcription and replication.
Materials and methods {#s2}
=====================
Cells, virus, and reagents
--------------------------
Mouse neuroblastoma N2a cells, baby hamster kidney (BHK-21) cells, and human embryonic kidney epithelial (HEK) 293T cells were cultured in DMEM supplemented with 10% fetal bovine serum (Gibco/Invitrogen, USA) at 37°C with 5% CO~2~. The challenge virus standard-11 strain of fixed rabies virus (CVS) was propagated in N2a cells. All the cells and virus used in this study were stored in our laboratory. Cells were infected with CVS at a multiplicity of infection (MOI) of 1 and the supernatant was harvested after 72 h post-infection (hpi). Virus preparations were titrated on N2a cells, and then stored at −80°C.
Drug treatment
--------------
The chemicals used to treat the RABV-infected N2a cells included the microtubule-depolymerizing drug nocodazole (Noco) (S1765; Beyotime, China), the microtubule-polymerizing drug paclitaxel (Taxol) (S1150; Selleckchem, USA), the histone deacetylase (HDAC) inhibitor trichostatin A (TSA) (T1952; Sigma) and the HDAC inhibitor sodium butyrate (NaBut) (S1539; Beyotime). The N2a cells were infected with RABV for 4 h and then treated with the various drugs. The solvent dimethyl sulfoxide (DMSO) was used as a control.
Cell viability assay
--------------------
N2a cells cultured in 96-well-plates were incubated with each drug or DMSO for 20 h. Cell viability was measured using a CCK-8 Cell Counting Kit (A311-02; Vazyme, China) and expressed as the percent of the control culture as described previously (Zan et al., [@B55]).
Transient transfection
----------------------
N2a cells were seeded into 6-well-plates or 35 mm glass bottom dishes (Shengyou Biotechnology Co. Ltd., Hangzhou, China). On the following day, the N2a cells were transfected with p-CMV-Myc-M (Zan et al., [@B56]) or the pCMV-myc empty vector using Exfect Transfection Reagent (Vazyme Biotech Co. Ltd., Nanjing, China), and the transfection procedure was carried out according to the manufacturer\'s instructions.
Confocal microscopy
-------------------
N2a, BHK-21, or 293T cells grown overnight into 35-mm glass bottom dishes were infected with RABV (MOI = 1) or not. At the indicated hours post-infection (hpi), the cells were fixed with 4% paraformaldehyde, permeabilized with 0.5% Triton X-100 in PBS, and incubated at 4°C with primary antibodies overnight. Cells were then incubated with FITC-conjugated-secondary antibodies (KPL) and/or Alexa Fluor 546-conjugated secondary antibodies (Invitrogen) at 37°C for 1 h. Cellular nuclei were stained with 10 μg/ml DAPI (Roche) for 5 min, and viewed by an LSM780 laser scanning confocal microscopy (Carl Zeiss). All images were taken using multitrack scanning for each fluorophore to prevent bleed-through. FITC fluorescence was detected after excitation at 488 nm with an emission long-band filter at 505--530 nm (green). Alexa Fluor 546 fluorescence was detected after excitation at 561 nm with an emission long-pass filter at 550--600 nm (red). DAPI was detected after excitation at 405 nm with an emission long-pass filter at 445--450 nm (blue). The pinholes were set to an Airy unit of 1 (equal in size to an Airy disk). Images were acquired with the identical capture settings and analyzed using Zen 2012 software (Zeiss). Rabbit anti-α-tubulin antibody (11224-1-AP; Proteintech, China), rabbit anti-acetylated α-tubulin antibody (5335S, Cell Signaling Technology), mouse anti-RABV-N antibody, and mouse anti-RABV-M antibody (Zan et al., [@B55]) were used as primary antibodies.
RNA preparation and quantitative real-time PCR
----------------------------------------------
N2a cells were infected with RABV (MOI = 1) for 4 h and then treated with either the drugs or DMSO for another 20 h. The total RNA was extracted from the cells using an RNA isolater Total RNA Extraction Reagent (R401-01, Vazyme). RNA (500 ng) from each sample was reverse-transcribed into cDNA using a PrimeScript® RT reagent kit (Takara). Quantitative real-time PCR was performed on the 7500 real-time PCR system (Applied Biosystems) using AceQ qPCR SYBR Green Master Mix (Q111-02, Vazyme). qPCR primers were used as described previously (Zhang et al., [@B57]; Xu et al., [@B53]).
Determination of virus titer
----------------------------
N2a cells were infected with RABV (MOI = 1) for 4 h and then treated with either the drugs or DMSO for another 20 h. The viral titers from culture supernatants were determined by viewing the infected cells under a fluorescent microscopy and calculating the TCID~50~ per 0.1 mL as described previously (Zan et al., [@B55]).
Western blotting
----------------
Western blotting was performed as previously described (Zhang et al., [@B57]). Briefly, cells were lysed with an NP40 lysis buffer (P0013F, Beyotime) after infection or transfection for the indicated times. Lysates were collected, and the protein concentrations were determined with BSA protein assay kit (P0010S, Beyotime). Equivalent amounts of cell lysates (80 μg) were resuspended in 5 × SDS-PAGE loading buffer (P0015, Beyotime) and boiled for 10 min. After centrifugation, the soluble cell lysates were separated on 12% SDS-PAGE gels, electro-transferred onto 0.22-μm-pore-size nitrocellulose membranes (GE Healthcare), and subjected to immunoblot analysis. Membranes were blocked in PBS containing 0.05% Tween 20 (PBST) and 5% skimmed milk for 1 h. Membranes were then incubated with primary antibody overnight at 4°C, washed three times with PBST for 5 min each time, and incubated with an appropriate secondary antibody conjugated to horseradish peroxidase (HRP) (KPL) for 1 h at 37°C. Finally, membranes were washed three more times with PBST before visualization was performed using SuperSignal West Pico chemiluminescent substrate (34079, Thermo) under the conditions recommended by the manufacturer. Images were captured using optimal auto-exposure settings on a chemiluminescent imaging system (Cell Biosciences, USA) and the densities of protein bands were normalized against the GAPDH signal and quantified using ImageJ software (National Institutes of Health, USA). Rabbit anti-HDAC6 antibody (ab1440, Abcam), rabbit anti-acetylated α-tubulin antibody (5335S, Cell Signaling Technology), rabbit anti-α-tubulin antibody (11224-1-AP, Proteintech), mouse anti-RABV-N antibody, rabbit anti-Myc antibody (0912-2, HuaAn Biotechnology, China), rabbit anti-stathmin antibody (11157-1-AP, Proteintech), rabbit anti-MAP4 antibody (11229-1-AP, Proteintech), and anti-glyceraldehyde-3-phosphate dehydrogenase (GAPDH) antibody (R1210-1, HuaAn Biotechnology, China) were used as the primary antibodies.
Statistical analysis
--------------------
The data were statistically analyzed and graphed using GraphPad Prism 5 (GraphPad Software, San Diego, CA). All of the results are presented as the means ± standard deviations. Statistically, significant differences between groups were determined by a Student\'s *t*-test. ^\*^*P* \< 0.05 and ^\*\*^*P* \< 0.01.
Results {#s3}
=======
RABV infection damages the host microtubule cytoskeleton
--------------------------------------------------------
To investigate the effects of RABV infection on the host microtubule cytoskeleton, we performed an immunofluorescence analysis of RABV-infected N2a cells at 0, 8, 16, and 24 hpi with antibodies against the viral M protein and α-tubulin (Figure [1A](#F1){ref-type="fig"}). In the mock-infected cells at 0 hpi, we observed an intense green fluorescence throughout the cytoplasm, reflecting a high concentration of microtubules (Figure [1A](#F1){ref-type="fig"}). In the RABV-infected cells at 8 hpi, the viral M protein had diffused throughout the cytoplasm, and in some of the infected cells, M protein was clustered to form an intensely filamentous structure which appeared predominantly as a network in the cytoplasm, called "filamentous network" (indicated by the white arrows in Figure [1A](#F1){ref-type="fig"} and Supplementary Figure [1](#SM1){ref-type="supplementary-material"}). This result was consistent with our previous studies (Zan et al., [@B55],[@B56]). Surprisingly, the microtubules were virtually undetectable in the cells in which the viral M protein formed a filamentous network (indicated by the white triangles, Figure [1A](#F1){ref-type="fig"}). This phenomenon was also observed in RABV-infected cells at 16 and 24 hpi, and the number of cells with a filamentous network of viral M protein and disrupted microtubule at 8, 16, and 24 hpi was about 7, 17, and 24%, respectively, which was progressively increased in a time-dependent manner (Figures [1A,B](#F1){ref-type="fig"}). This suggests that the disruption of the microtubule cytoskeleton induced by RABV infection is likely associated with the formation of the M protein filamentous network. Moreover, similar results were also obtained with the BHK-21 and 293T cells (Supplementary Figure [2](#SM1){ref-type="supplementary-material"}). These results indicate that RABV infection damages the host microtubule cytoskeleton and is associated with the accumulation of the viral M protein.
{#F1}
Depolymerization and polymerization of microtubules have an inverse effect on the RNA synthesis of RABV
-------------------------------------------------------------------------------------------------------
Since we found that RABV infection damages the microtubule cytoskeleton (Figure [1](#F1){ref-type="fig"}), we next wanted to elucidate the effects of microtubule dynamics on the transcription and replication of RABV. For this purpose, N2a cells were infected with RABV for 4 h to allow sufficient time for viral entry into the cytoplasm and intracellular transport (Xu et al., [@B52]). Subsequently, the cells were then treated with the microtubule-depolymerizing drug Noco, the microtubule-polymerizing drug Taxol, or DMSO for another 20 h, respectively. As shown in Figure [2A](#F2){ref-type="fig"}, treatment with Noco or Taxol did not substantially affect cellular viability. Surprisingly, the levels of the viral N and P mRNA, viral genomic RNA, and infectious RABV progeny were all significantly increased in Noco-treated cells, but were notably reduced in Taxol-treated cells (Figure [2B](#F2){ref-type="fig"}). These findings suggest that RABV-induced microtubule depolymerization promotes viral RNA synthesis.
{#F2}
Furthermore, we investigated whether microtubule dynamics affect the filamentous network formation of the viral M protein. As observed in Figure [2C](#F2){ref-type="fig"}, microtubule depolymerization neither inhibited the filamentous network formation of the viral M protein nor blocked the RABV-induced disruption of the microtubule cytoskeleton in Noco-treated cells (indicated by a white triangle). However, the filamentous network formation of the viral M protein was significantly inhibited by small amounts of M protein clustered only near the nucleus in Taxol-treated cells, where the microtubules polymerized into bundles, compared to that in DMSO-treated cells (Figure [2C](#F2){ref-type="fig"}). Thus, we hypothesized that RABV infection induces microtubule depolymerization. In addition, we also found that microtubule depolymerization or polymerization did not affect the formation of Negri bodies (NBs; Supplementary Figure [3](#SM1){ref-type="supplementary-material"}; the sites of RABV transcription and replication) which accumulated by the viral N protein in the cytoplasm, and was consistent with a previous study (Lahaye et al., [@B26]). Taken together, these results indicate that microtubule depolymerization promotes the RNA synthesis of RABV, but does not affect the formation of the viral M protein filamentous network.
RABV infection reduces the acetylated modification of microtubules through the upregulation of HDAC6
----------------------------------------------------------------------------------------------------
Microtubule dynamics are regulated by MAPs (e.g., MAP4 and stathmin) and are closely associated with post-translational modifications; in particular, α-tubulin acetylation is recognized as a well-known maker of stable microtubules and is governed by HDAC6 (Howell et al., [@B20]; Matsuyama et al., [@B31]; Perdiz et al., [@B41]; Yang et al., [@B54]; Akhmanova and Steinmetz, [@B1]; Alfaro-Aco and Petry, [@B3]; Kadavath et al., [@B23]). To clarify RABV-induced microtubule depolymerization, we next examined the expression levels of MAP4, stathmin, HDAC6 and its substrate, acetylated α-tubulin (ace-tubulin) in RABV-infected or mock-infected N2a cells at 24 and 48 hpi. The results in Figures [3A,B](#F3){ref-type="fig"} revealed that the expression levels of MAP4 and stathmin remain relatively stable, while the HDAC6 expression levels significantly upregulated at 48 hpi. Moreover, there was a decrease in the level of ace-tubulin expression in RABV-infected cells, compared to those in the mock-infected cells. In addition, the acetylated microtubules were weakly stained or nearly undetectable in the RABV-infected cells but not in the mock-infected cells as indicated by confocal microscopy (Figure [3C](#F3){ref-type="fig"}). Collectively, these results indicate that RABV infection induces upregulation of HDAC6 expression to reduce ace-tubulin expression, and results in destabilizing the microtubule cytoskeleton.
{#F3}
Inhibition of the deacetylase activity of HDAC6 decreases RABV RNA synthesis
----------------------------------------------------------------------------
To investigate whether the deacetylase activity of HDAC6 has any effect on the RNA synthesis of RABV, the N2a cells were infected with RABV for 4 h and then treated with TSA, NaBut, or DMSO for another 20 h. The Western blotting analysis and immunofluorescence staining (Supplementary Figures [4A,B](#SM1){ref-type="supplementary-material"}) revealed that the ace-tubulin expression levels were significantly increased in the TSA-treated cells but not in the NaBut or DMSO-treated cells, suggesting that the inhibition of HDAC6 activity by TSA is effective. In addition, the viability of the cells treated with TSA or NaBut were not significantly affected, and were comparable to that of the DMSO-treated cells (Figure [4A](#F4){ref-type="fig"}). As expected, treatment with TSA significantly inhibited viral RNA synthesis and downregulated the viral titer compared to treatment with NaBut or DMSO (Figure [4B](#F4){ref-type="fig"}). This suggests that the deacetylase activity of HDAC6 is important for viral RNA synthesis during RABV infection.
{#F4}
Furthermore, we investigated whether the inhibition of the HDAC6 deacetylase activity affected the filamentous network formation of the viral M protein. As observed in Figure [4C](#F4){ref-type="fig"}, treatment with TSA neither inhibited the filamentous network formation of the viral M protein nor blocked the RABV-induced disruption of the microtubule cytoskeleton (indicated by a white triangle), suggesting that HDAC6 plays a partial, but not a complete, role in RABV-induced microtubule depolymerization. In addition, we also observed that the inhibition of HDAC6 deacetylase activity did not affect the formation of NBs (Supplementary Figure [3](#SM1){ref-type="supplementary-material"}). Taken together, these results indicate that the inhibition of HDAC6 deacetylase activity decreases the RNA synthesis of RABV.
M protein alone induces microtubule depolymerization
----------------------------------------------------
Since the disruption of the microtubule cytoskeleton by RABV infection was closely associated with the filamentous network formation of the viral M protein (Figure [1](#F1){ref-type="fig"}), it was hypothesized that the M protein might be involved in the induction of microtubule depolymerization. To explore this possibility, N2a cells were transfected with p-CMV-Myc-M or the pCMV-myc empty vector (as a control) for 36 h, and then analyzed by Western blotting. As shown in Figures [5A,B](#F5){ref-type="fig"}, the viral M protein alone did not modify the expression of MAP4 and stathmin, but significantly upregulated the expression of HDAC6, resulting in the downregulation of ace-tubulin expression. Furthermore, we analyzed the effects of the M protein alone on the integrity of the microtubule cytoskeleton via confocal microscopy. As shown in Figures [5C,D](#F5){ref-type="fig"}, both the microtubules and acetylated microtubules were depolymerized with weak staining in cells expressing the viral M protein, while microtubules or acetylated microtubules remained intact with bright staining in the non-transfected or empty vector-transfected cells. Taken together, these results indicate that the M protein alone depolymerizes microtubules, and is associated with the upregulation of HDAC6 and downregulation of ace-tubulin expression.
{#F5}
Discussion {#s4}
==========
The microtubule cytoskeleton plays an important role for multiple cellular functions, including cellular division, signaling, and intracellular trafficking (Downing, [@B9]; Hammond et al., [@B18]; Akhmanova and Steinmetz, [@B1]). Currently, numerous viruses have been reported to utilize the microtubule cytoskeleton for intracellular trafficking, including HIV-1, vaccinia virus, herpesvirus, circovirus, and RABV (Sanderson et al., [@B44]; Nishi et al., [@B37]; Pasdeloup et al., [@B40]; Cao et al., [@B8]; Fernandez et al., [@B10]; Xu et al., [@B52]). A recent study reported that α-tubulin was incorporated into RABV particles (Tu et al., [@B50]), suggesting that with the exception of intracellular trafficking, the microtubule cytoskeleton might play other roles during RABV infection. In the present study, we were the first to demonstrate that microtubule depolymerization induced by RABV infection facilitates viral RNA synthesis.
The rhabdoviral M protein is a multifunctional protein which plays important roles in regulating the balance between viral transcription and replication, and is associated with cellular pathogenesis, virion assembly, and budding (Mebatsion et al., [@B33]; Finke and Conzelmann, [@B11]; Pulmanausahakul et al., [@B42]; Wirblich et al., [@B51]). In addition, the M protein of vesicular stomatitis virus (VSV), another member of the *Rhabdoviridae* family, has been reported to self-assemble to form long fibers *in vitro* via the association between the N-terminal portion of the M protein and the globular domain of an adjacent M molecule (McCreedy et al., [@B32]; Gaudin et al., [@B15], [@B16]; Gaudier et al., [@B13]). Based on similar 3-dimensional crystal structures of the M proteins from VSV and the Lagos bat virus, the highly conservative sequence of the interacting N-terminal region, as well as the globular domain of the M proteins of the *Lyssavirus* genus (Gaudier et al., [@B14]; Assenberg et al., [@B4]; Graham et al., [@B17]), rhabdoviral M proteins likely share the ability to self-assemble into homo-oligomers. In this study, the viral M protein of RABV was found to accumulate to form a filamentous network (Figure [1](#F1){ref-type="fig"}) which may indicate the display pattern of the RABV M protein self-assembly. In addition to self-assembly, the N-terminal domain of the VSV M protein has been reported to interact with tubulin, resulting in a disruption of the microtubule cytoskeleton (Melki et al., [@B34]). Strikingly, we observed that microtubules were nearly undetectable in RABV-infected cells where viral M protein accumulates to form the filamentous network (Figure [1](#F1){ref-type="fig"}). Thus, we hypothesize that the disruption of the microtubule cytoskeleton by RABV may be due to the possible interaction of the viral M protein with tubulin or the stretching forces from the filamentous network of viral M protein that physically impair the microtubule cytoskeleton. Moreover, the polymerization of microtubules substantially inhibits the assembly of the viral M protein (Figure [2C](#F2){ref-type="fig"}), which further confirms our above hypothesis. Although the viral M proteins of VSV and chandipura virus have been shown to interact with cellular tubulin (Melki et al., [@B34]; Rajasekharan et al., [@B43]), regrettably, we could not detect the interaction between tubulin and the RABV M protein via a co-immunoprecipitation assay (data not shown). In addition, we also demonstrated that the expression of the viral M protein alone disrupts microtubules (Figure [5](#F5){ref-type="fig"}), suggesting that neuronal degeneration induced by RABV infection is likely related to the disruption of microtubules.
HDAC6 is a predominant deacetylase of microtubules and has been reported to play critical roles in viral infection (L\'Hernault and Rosenbaum, [@B27]; Matsuyama et al., [@B31]; Perdiz et al., [@B41]; Yang et al., [@B54]). Some viruses induce significant acetylation of α-tubulin to stabilize the microtubule cytoskeleton for their intracellular transport by regulating HDAC6, including circovirus, herpes simplex°virus, and influenza A virus (Husain and Harrod, [@B21]; Zhong et al., [@B58]; Cao et al., [@B8]). Recently, HDAC6 has also been reported to be involved in the replication of certain viruses. HDAC6 promotes hepatitis C virus replication and reactivates the lytic phase of HIV-1 and Kaposi\'s sarcoma-associated herpesvirus replication (Banerjee et al., [@B7]; Shin et al., [@B46]; Kozlov et al., [@B25]). Similarly, we found that HDAC6 affects RABV RNA synthesis. Moreover, the inhibition of HDAC6 by TSA significantly suppresses both RABV transcription and replication (Figure [4B](#F4){ref-type="fig"}). Unexpectedly, the inhibition of HDAC6 neither prevents RABV from disrupting microtubules nor blocks the formation of the filamentous network of viral M protein (Figure [4C](#F4){ref-type="fig"}). This finding suggests that HDAC6 may play a partial role in the RABV-induced disruption of microtubules and other unknown mechanisms for impairment of microtubules should exist during RABV infection. Although it remains unknown how RABV regulates HDAC6, the finding that HDAC6 is involved in RABV RNA synthesis is highly important.
Cellular tubulin has been reported to stimulate the viral RNA synthesis of some other negative-stranded RNA viruses, including SeV, measles virus, and VSV (Moyer et al., [@B36], [@B35]; Ogino et al., [@B38]). As a negative-stranded RNA virus, RABV RNA synthesis is also mediated by tubulin. In this study, the depolymerization of microtubules significantly upregulated the transcription and replication levels of RABV (Figure [2B](#F2){ref-type="fig"}), while the polymerization of microtubules or the blockade of HDAC6 deacetylase activity notably inhibited viral RNA synthesis (Figures [2B](#F2){ref-type="fig"}, [4B](#F4){ref-type="fig"}). These findings suggest that RABV infection induces microtubule depolymerization into tubulin monomers, which stimulate viral RNA synthesis. Increasing studies have demonstrated that the RABV M protein is a negative regulator of transcription by condensing and recruiting viral RNP to the budding sites on the cell membrane, resulting in the inhibition of both viral transcription and replication (Finke and Conzelmann, [@B11]; Finke et al., [@B12]; Jayakar et al., [@B22]; Banerjee, [@B6]). Therefore, RABV-induced microtubule depolymerization for the stimulation of viral RNA synthesis is most likely due to the fact that tubulin dissociates the viral M protein from RNP, leading to the release of M protein inhibition on the viral RNP.
In conclusion, our study reports for the first time, that RABV infection induces microtubule depolymerization to facilitate viral RNA synthesis, and that HDAC6 is involved in this process. Although additional studies remain to be explored to reveal the molecular basis for the activation of viral RNA synthesis by microtubule depolymerization, the results of the present study deepen our understanding of RABV pathogenicity and provide novel insight into the relationship between RABV and the microtubule cytoskeleton.
Author contributions {#s5}
====================
Design and supervise the study: JYZ, ML, JYG, and JZ. Perform the experiments: JZ, KKM, SL, DNS, YY, JL, and BLH. Prepare the manuscript: JZ, ML, JYG, and JYZ.
Conflict of interest statement
------------------------------
The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
This study was supported by grants from National Key R & D Program of China (Grant No. 2015BAD12B01) and the National Special Fund for Public Welfare Industry (Project No. 201103032) of China.
Supplementary material {#s6}
======================
The Supplementary Material for this article can be found online at: <http://journal.frontiersin.org/article/10.3389/fcimb.2017.00146/full#supplementary-material>
######
Click here for additional data file.
RABV
: rabies virus
RNP
: ribonucleoprotein
MAPs
: microtubule-associated proteins
HDAC6
: histone deacetylase 6
M protein
: matrix protein
Noco
: nocodazole
Taxol
: paclitaxel
TSA
: trichostatin A
NaBut
: sodium butyrate
DMSO
: dimethyl sulfoxide
CVS
: the challenge virus standard-11 strain of fixed rabies virus
MOI
: multiplicity of infection
Ace-tubulin
: acetylated α-tubulin
GAPDH
: glyceraldehyde-3-phosphate dehydrogenase.
[^1]: Edited by: Lorenza Putignani, Bambino Gesü Ospedale Pediatrico (IRCCS), Italy
[^2]: Reviewed by: Stacey Gilk, Indiana University School of Medicine, USA; Xiaohui Zhou, University of Connecticut, USA
| {
"pile_set_name": "PubMed Central"
} |
1. Introduction
===============
Postpartum depression (PD) is one of the commonest psychological problems to affect postpartum women\'s health. Its prevalence rate worldwide was reported to be 11.9% and a higher rate of developing countries.^\[[@R1],[@R2]\]^ PD could develop symptoms as depressed emotion, fatigue, and even suicidal behavior.^\[[@R1],[@R3]\]^ Mothers with PD were associated with autistic disorder and developmental retardation in their offspring.^\[[@R3],[@R4]\]^ Thus, it is urgent to prevent pregnant women from PD.
Nutrition during pregnancy is a vital factor of PD. Accumulating evidence has proved that deficiencies in unsaturated fatty acids and vitamin D were more commonly in patients with PD.^\[[@R5],[@R6]\]^ The possible mechanism might be involved in regulation of proinflammatory cytokines and serotonergic neurotransmitters.^\[[@R6]\]^ Physiological increases in nutrition related biochemical indicators, such as blood glucose, lipid profile, and uric acid (UA), are reported during pregnancy. Some previous studies have reported an association between serum lipid concentrations and the risk of depression,^\[[@R5],[@R7]\]^ while its mechanisms were not well understood. Some other studies demonstrated an increased risk of anxiety and depression in postpartum women who had a rapid reduction in the concentrations of these biochemical indicators.^\[[@R8]\]^ The physiological increases may vary in pregnant women from different regions.
Chinese economy has enjoyed a rapid development over the past 3 decades and the lifestyle in Chinese people is changing dramatically, especially reflected on food consumption. An increased intake of animal-based foods, but a decreased intake of plant-based foods is characterized as modern Chinese.^\[[@R9]--[@R11]\]^ In this study, we conducted a case-control study to explore the association between nutrition assessments during pregnancy and the risk of PD in Chinese women.
2. Method
=========
2.1. Study design
-----------------
This study was hospital-based case-control designed with a study period from January 2016 to June 2019. Participants with PD were recruited in the case group according to the criteria as follows:
1. first time diagnosis for PD;
2. singleton and term delivery;
3. without other postpartum disease;
4. without severe co-morbidities including neoplasm, cardiovascular disease, mental disorder, or severe infection during pregnancy;
5. no smoking or drinking during pregnancy;
6. age from 20 to 40 years.
The controls were selected from puerpera who had childbirth in our hospital as the same period of the cases and were free of PD. Protocol of this study was in accordance with the Declaration of Helsinki and the research ethics principles of the Committee of our hospital. Eligible participants attended this study with signed informed consent.
Baseline information was extracted from medical records of our hospital information system, including maternal age, maternal weight, neonatal weight, gestational weeks, neonatal gender, type of delivery.
2.2. PD
-------
The Chinese version of Edinburgh postnatal depression scale (EPDS) was employed to identify PD,^\[[@R12]\]^ which has a fine reliability and validity and has been widely applied in China.^\[[@R13]--[@R15]\]^ EPDS contained 10 items including mood, pleasure, remorse, anxiety, fear, coping ability, insomnia, sadness, crying, and NSSI.^\[[@R12]\]^ Score in each item range of 0 to 3 points.^\[[@R12]\]^ Puerpera obtained an EPDS score ≥10 points to suggest the presence of PD, while an EPDS score \< 10 points to suggest free of PD.^\[[@R12]\]^
2.3. Food consumption during pregnancy
--------------------------------------
A food frequency questionnaire (FFQ) was designed to collect food consumption before the childbirth preceding month. Stable food (rice, wheat, other), vegetable (light, dark), fruit, fish, meat (pork, beef, and mutton, other red meat), poultry, and water were list on the FFQ (8, 9, 31). Participants were interviewed by well-trained investigators. Frequency and amount to food consumption before the childbirth preceding month were recorded. Several options were applied for food frequency, including never, times per day, times per week, and times per month.
2.4. Biochemical indicators detection during pregnancy
------------------------------------------------------
A routine fasting blood sample (5--10 mL) was collected from each participant in the third trimester of pregnancy. The samples were stored at −80°C until used. Biochemical indicators including fasting blood-glucose (GLU), total cholesterol (TC), triglyceride (TG), low density lipoprotein (LDL), HDL, and UA were detected by valid methods as described previously.^\[[@R5],[@R7]\]^ Dyslipidemia in pregnant women was diagnosed when TC, TG, LDL concentrations were above the 95th percentile, and HDL concentration was below the 5th percentile for gestational age.^\[[@R16]\]^ Hyperglycemia in pregnant women was defined as GLU ≥7.0 mmol/L.^\[[@R17]\]^ Hyperuricemia was defined as UA ≥ 390 μmol/L.^\[[@R18],[@R19]\]^
2.5. Statistical analysis
-------------------------
R version 3.5.1 (The R Foundation, Vienna, Austria) was applied for statistical analysis. Categorical variables were given as frequency and percentile, while continuous variables were given as mean and standard deviation. Based on variable types, Student *t* test or Chi-squared test was used to evaluate variable difference in the 2 groups. A logistic regression model was applied to evaluate the association between nutrition related biochemical indicators (GLU, TC, TG, LDL, HDL, and UA) and PD. Crude odds ratio (OR) and its corresponding 95% confidence interval (CI) were computed. After then, adjusted ORs were calculated by adjusting for several possible risk factors (maternal age, maternal weight, neonatal weight, gestational weeks, neonatal gender, delivery type, systolic blood pressure, and diastolic blood pressure). A *P*-value \< .05 was set to describe statistical difference.
3. Results
==========
3.1. Characteristics of eligible participants
---------------------------------------------
A total of 565 participants were eligible in this study, which comprised 182 puerpera with PD and 383 puerpera without. Flowchart exhibiting participant selection is showed in Figure [1](#F1){ref-type="fig"}. Average age in case group was 31.9 ± 5.3 and 31.5 ± 5.7 years in control group. Vaginal delivery and caesarean section delivery were 68.1% and 31.9% in case group, and 63.4% and 36.6% in control group. Other baseline information about eligible participants is summarized in Table [1](#T1){ref-type="table"}.
{#F1}
######
Characteristics of eligible participants with postpartum depression and without.
3.2. Association between pregnant biochemical indicators and PD
---------------------------------------------------------------
GLU in case group (5.0 ± 1.0 mmol/L) was significantly higher than the control group (5.3 ± 2.1 mmol/L). Logistics analysis demonstrated that patients with PD have a 2.67-fold (95%CI = 1.67--4.11, *P*-value = .012) higher odds exposed to elevated GLU level. After adjusting for several possible risk factors, the result remained.
Serum lipid indicators including TC, TG as well as LDL in case group were also statistically higher than the control group, but HDL was found to be lower in case group. Compare with the controls, ORs for increased TC, TG, and LDL during pregnancy were 1.73 (95%CI = 1.22--2.46, *P*-value = .023), 2.43 (95%CI = 1.55--3.81, *P*-value \<.001), and 3.41 (95%CI = 2.09--5.57, *P*-value \< .001), respectively in patients with PD. Additionally, an inverse association between decreased HDL and PD (OR = 3.41, 95%CI = 2.09--5.57, *P*-value \< .001) was identified. The positive associations mentioned above remained after matching with some risk factors.
However, with respect to UA, no statistically association was identified (OR = 2.23, 95%CI = 0.82--6.26, *P*-value = .137). Detail information is shown in Figure [2](#F2){ref-type="fig"} and Table [2](#T2){ref-type="table"}.
{#F2}
######
Association between pregnant related biochemical indicators and postpartum depression.
3.3. Food consumption before the childbirth preceding month
-----------------------------------------------------------
According to FFQ, patients with PD were found to have a higher meat intake (142.9 ± 57.7 g/d) before the childbirth preceding month when compared to those free of PD (121.0 ± 65.8 g/d), while lower consumptions of vegetable, fruit, fish, and poultry were characterized in patients with PD. There were lack material difference about stable food, water, and energy intake between the 2 group. Detail information is demonstrated in Table [3](#T3){ref-type="table"}.
######
Various kinds of food consumption (g/d) between participants with postpartum depression and without postpartum depression.
4. Discussion
=============
In this hospital-based case-control study, women with postnatal depression were found to have higher odds of elevated GLU, TC, TG, and LDL, but decreased HDL in later stages of pregnancy when compared with parturient free of PD. These findings were robust as the associations remained significant after adjusting for possible confounders. What is more, diet pattern of women with PD was different from those without. Women with PD were characterized as higher meat consumption, but lower vegetable, fruit, fish, and poultry consumption during later pregnancy.
Gestational diabetes mellitus would receive much attention due to its adverse effects on maternal and offspring. In this study, a higher OR of increased GLU was found in women with postnatal depressive symptoms. Mak et al^\[[@R20]\]^ also proved the positive association between gestational diabetes and the risk of postnatal depression through a prospective cohort study comprising 1449 mothers. Our results were in accordance with possibly biological plausibility that abnormal regulation of blood glucose regulation during pregnancy may increase susceptibility to occurrence of PD through activating and releasing cortisol and inflammatory adipokine. Function of hypothalamic-pituitaryadrenal axis in patients with PD were reported to be irregular and an increased level of cortisol was observed, which could partly attribute to abnormal blood glucose or diabetes.^\[[@R21],[@R22]\]^ Irregular blood glucose was also proved to be involved in activation and secretion of inflammation and adipokine, such as Interleukin-6, which are known to be associated with elevated risk of PD.^\[[@R23],[@R24]\]^ In addition, a previous study indicated that symptoms and adverse birth outcomes associated with abnormal blood glucose might bring additional psychological burden and stress to these mothers.^\[[@R20]\]^
The current study also supported the material associations between dyslipidemia during pregnancy and the risk of postnatal depression. Among several potential mechanisms documented in previous publications, the role of serum lipids on serotonin attracts the most attention. Serotonin is a neurotransmitter and plays a vital role in adjusting controlling appetite, sleep, and mood.^\[[@R25]\]^ The fluctuations in serum lipids, such as reduction of cholesterol, would affect normal structure of brain cell including reduction of cholesterol in cell membranes and lipid micro viscosity (23). Without to integrate structure, function of serotonin receptors on brain cell would be damaged, such as impairing its uptake function of serotonin.^\[[@R26]\]^ Reduced serotonin in brain was proved to be associated with depression.^\[[@R27]\]^
Another possible mechanism regarding the association between serum lipids and PD may be explained by hypothalamic-pituitaryadrenal axis. Neurochemical functions, such as synthesis and used of norepinephrine, dopamine, and serotonin in patients with depressive disorders are abnormal, which may attribute to dysregulation adrenocorticotropin hormone and elevating cortisol and cortisone levels.^\[[@R28],[@R29]\]^ Previous study confirmed a material association between cortisol excretion rate and serum lipids, indicative of an indirect effect of serum lipids on development of depression.^\[[@R8],[@R30]\]^ More effectors are needed to explore the biological mechanisms about this association.
Our findings suggest a higher meat consumption, but lower vegetable, fruit, fish, and poultry consumption in women with PD before the childbirth preceding month. To supported growth of fetal and infant, women during pregnancy, or lactation should intake much more nutrients, in particular the essential nutrients, such as n-3 fatty acids and high-quality protein.^\[[@R7]\]^ An increasing evidence had confirmed a significant association between food consumption and development of depression.^\[[@R31],[@R32]\]^ A higher meat intake may be associated with low-grade inflammation, such as elevated C-protein, which have an involvement in the pathogenesis of depression.^\[[@R33]\]^ Vegetable and fruit contain abundant antioxidants, such as vitamin C and anthocyanin, which exert beneficial protective effects on against depressive disease.^\[[@R34],[@R35]\]^ Besides, vegetable and fruit could also provide dietary fiber and block gut microbiota into blood circulation which may have adverse effect on central nervous system.^\[[@R36]--[@R38]\]^ The protective effect of poultry and fish intake against depression may be owing to their high content of polyunsaturated fatty acids.^\[[@R34]\]^ These fatty acids have important anti-inflammatory properties and exert irreplaceable role in brain function and serotonin neurotransmission, such as the fluidity of neurons cell membrane.^\[[@R31]\]^ Additionally, fatty acids especially n-3 fatty acids, were thought to improve mood and cognition in patients with major depression.^\[[@R34]\]^
Findings from this study should be considered in view of some limitations. To start with, this study was case-control design, which could provide clue of risk factors for PD, but cannot draw a causal association between nutritional assessments in pregnancy and the risk of PD due to its reverse-chronological order inference of disease and cause. Secondly, bias from observational study cannot be eliminated, in particular recall bias. Although we performed a multi-factor logistic regression model to adjust the effects of several potential risk factors, some other possible confounding factors might exist. Thirdly, the sample size is small and participants were from a local place in China, indicating the limitation of generalizability.
5. Conclusion
=============
In this study, women with postnatal depression were found to have higher odds of elevated GLU, TC, TG, and LDL, but decreased HDL in later stages of pregnancy. Diet pattern of women with PD were different from those without. Women with PD were characterized with higher meat consumption, but lower vegetable, fruit, fish, and poultry consumption during later pregnancy. As observational study cannot draw a causal link, further well-designed perspective and experimental studies are warranted.
Author contributions
====================
**Conceptualization:** Wen Feng.
**Data curation:** Dan Shi, Wen Feng.
**Formal analysis:** Guo-hua Wang.
**Funding acquisition:** Wen Feng.
**Investigation:** Dan Shi.
**Methodology:** Guo-hua Wang, Wen Feng.
**Resources:** Wen Feng.
**Software:** Guo-hua Wang.
**Supervision:** Wen Feng.
**Writing -- original draft:** Dan Shi, Wen Feng.
**Writing -- review & editing:** Guo-hua Wang.
Abbreviations: CI = confidence interval, EPDS = the Chinese version of Edinburgh postnatal depression scale, FFQ = food frequency questionnaire, GLU = fasting blood-glucose, HDL = high density lipoprotein, LDL = low density lipoprotein, OR = odds ratio, PD = postpartum depression, TC = total cholesterol, TG = triglyceride, UA = uric acid.
How to cite this article: Shi D, Wang G-h, Feng W. Nutritional assessments in pregnancy and the risk of postpartum depression in Chinese women: a case-control study. *Medicine*. 2020;99:33(e21647).
The authors have no funding and conflicts of interest to disclose.
The datasets generated during and/or analyzed during the current study are available from the corresponding author on reasonable request.
| {
"pile_set_name": "PubMed Central"
} |
Background {#Sec1}
==========
After the completion of the Human Genome Project \[[@CR1]\] and a successful case-control experiment in identifying age-related markers using single-nucleotide polymorphism (SNP) \[[@CR2]\], the number of genome-wide association study (GWAS) has been rising exponentially \[[@CR3]\]. GWAS provides an efficient way to scan the whole genome to locate SNPs associated with the trait of interest which may potentially lead to identification of the susceptibility gene through linkage disequilibrium. Unlike linkage analysis that requires data collection from genetically related subjects, GWAS is applicable to a more general setting involving independent subjects. This makes GWAS highly desirable because for many diseases, it may not be feasible to recruit enough related subjects for linkage analysis. For example, the parents of human subjects with late onset diseases are usually not available. Furthermore, many statistical programs such as PLINK \[[@CR4]\] have been developed to manage and analyze high dimensional GWAS data from *independent subjects*.
Due to reduced costs for SNP arrays, in recent years, many family studies have collected GWAS data \[[@CR5]--[@CR7]\]. If existing methods designed for independent subjects are adopted to analyze these data, the power of association tests will be greatly reduced because only a subset of data can be used. On the other hand, employing all the data in the analysis (i.e. ignoring the correlation between genetically related subjects) may result in false positive findings \[[@CR8]\]. Yu et al. (2006) \[[@CR9]\] proposed a compromise between these two approaches that used all related subjects while adjusting for the relatedness by random effects in a linear mixed model. This approach has been widely studied and the original algorithm has been improved to be applicable to larger scale studies \[[@CR10]\]. However, this approach can only handle a *univariate phenotype* such as a positive or negative diagnosis.
Many complex diseases such as mental health disorders have multiple phenotypic traits that are correlated \[[@CR11]\]. These multivariate phenotypes may point to a shared genetic pathway and underscore the relevance of pleiotropy (i.e., a gene or genetic variant that affects more than one phenotypic trait, Solovieff et al. (2013) \[[@CR12]\]). Furthermore, a statistical model searching for loci that are simultaneously associated with multiple phenotypes has higher power than a model that only considers each phenotype individually \[[@CR13]\]. Our research team recently developed a multivariate association test based on the Fisher combination function that can be applied to analyze GWAS data with multivariate phenotypes \[[@CR14]\]. This method, however, can only handle *independent subjects*. Taken together, advanced methods that can handle multivariate phenotypes and related subjects *simultaneously* are highly desirable.
The crucial problem in GWAS is to deal with confounders such as population structure, family structure, and cryptic relatedness. Astle and Balding \[[@CR15]\] reviewed approaches to correcting association analysis for confounding factors. When family-based samples are collected, analysis based on the transmission disequilibrium test is robust to population structure. Several methods have been developed for multivariate phenotype data collected from family-based samples \[[@CR16]--[@CR20]\]. However, the major challenge of this type of studies is to recruit enough families in order to conduct the analysis with sufficient power. This type of studies also have limited applications to late-onset diseases. Recently, Zhou and Stephens \[[@CR21]\] proposed a multivariate linear mixed model (mvLMMs) for identifying pleiotropic genes. This approach can handle a mixture of unrelated and related individuals and thus has broader applications. However, it was recommended for a modest number of phenotypes (less than 10) due to computational and statistical barriers of the EM algorithm \[[@CR21]\].
For related subjects with multivariate phenotypes, there are two sources of correlations between multivariate phenotypes: one is the correlation arising from genetically related subjects whose phenotypes are more highly correlated because of shared genotypes; and the other is the correlation between multiple phenotypes which exists even when independent subjects are employed. This study proposes a new statistical method that can model both sources of correlations. We also compare the performance of the proposed method with that of the mvLMMs method. The rest of this paper is organized as follows. In the "[Methods](#Sec2){ref-type="sec"}" section, we review our previous work on multivariate phenotypes in independent subjects and also extend the method to handle related subjects. The "[Results](#Sec5){ref-type="sec"}" section summaries the results of simulation studies and statistical analysis on the Study of Addiction: Genetics and Environment (SAGE) data. Future directions and major findings are presented in "[Discussion](#Sec12){ref-type="sec"}" and "[Conclusions](#Sec13){ref-type="sec"}" sections, respectively.
Methods {#Sec2}
=======
Suppose that for each subject, we measure *R* different phenotypes and run an assay with *S* SNPs. The resulting measurements can be organized with two data matrices. The genotype data are stored in a *S*×*N* matrix where *N* is the total number of subjects and each element of the matrix is coded as 0, 1, or 2 copies of the reference allele. The phenotype data are stored in an *N*×*R* matrix where each row records the individual's multivariate phenotypes. Studying the association between genotypes and phenotypes, thus, involves measuring and testing the association between each row of the genotype matrix and the entire phenotype matrix. Since one SNP is consider at a time, the association test is repeated *S* times for all SNPs. Specifically, Let *β* ~1~,...,*β* ~*R*~ be the effect sizes of a candidate SNP on *R* different phenotypes. The null hypothesis of testing the pleiotropic gene is $$\documentclass[12pt]{minimal}
\usepackage{amsmath}
\usepackage{wasysym}
\usepackage{amsfonts}
\usepackage{amssymb}
\usepackage{amsbsy}
\usepackage{mathrsfs}
\usepackage{upgreek}
\setlength{\oddsidemargin}{-69pt}
\begin{document} $$H_{0}: \beta_{1}= \ldots = \beta_{R} = 0. $$ \end{document}$$ If this *H* ~0~ is rejected, we claim that the corresponding SNP is associated with the pre-determined multivariate phenotypes.
When the phenotype is univariate, the association test for GWAS data can be carried out using commonly adopted software such as R \[[@CR22]\] or PLINK \[[@CR4]\]. For multivariate phenotypes in independent subjects, Yang et al. (2016) \[[@CR14]\] has conducted a comprehensive review of various multivariate methods and proposed a method using the Fisher combination function. They further showed that their proposed method is better than other existing methods. The following sections briefly review their method and extend it to handle related subjects by employing a linear mixed model to adjust for relatedness.
Review of previous work on independent subjects with multivariate phenotypes {#Sec3}
----------------------------------------------------------------------------
To illustrate the method proposed by Yang et al. (2016) \[[@CR14]\], we define the notations for genotypes and phenotypes. Let *i*(=1,...,*N*) be the index of individuals. Define $\documentclass[12pt]{minimal}
\usepackage{amsmath}
\usepackage{wasysym}
\usepackage{amsfonts}
\usepackage{amssymb}
\usepackage{amsbsy}
\usepackage{mathrsfs}
\usepackage{upgreek}
\setlength{\oddsidemargin}{-69pt}
\begin{document}$y^{r}_{i}$\end{document}$ as the *r*th phenotype of individual *i* (*r*=1,...,*R*) and $\documentclass[12pt]{minimal}
\usepackage{amsmath}
\usepackage{wasysym}
\usepackage{amsfonts}
\usepackage{amssymb}
\usepackage{amsbsy}
\usepackage{mathrsfs}
\usepackage{upgreek}
\setlength{\oddsidemargin}{-69pt}
\begin{document}$g^{s}_{i}$\end{document}$ as the *s*th genotype of individual *i* (*s*=1,...,*S*). Therefore, the vector $\documentclass[12pt]{minimal}
\usepackage{amsmath}
\usepackage{wasysym}
\usepackage{amsfonts}
\usepackage{amssymb}
\usepackage{amsbsy}
\usepackage{mathrsfs}
\usepackage{upgreek}
\setlength{\oddsidemargin}{-69pt}
\begin{document}$\boldsymbol {y}^{r} = (y^{r}_{1},\ldots,y^{r}_{N})$\end{document}$ represents the *r*th marginal phenotype collected from *N* individuals and the vector $\documentclass[12pt]{minimal}
\usepackage{amsmath}
\usepackage{wasysym}
\usepackage{amsfonts}
\usepackage{amssymb}
\usepackage{amsbsy}
\usepackage{mathrsfs}
\usepackage{upgreek}
\setlength{\oddsidemargin}{-69pt}
\begin{document}$\boldsymbol {g}^{s} = (g^{s}_{1},\ldots,g^{s}_{N})$\end{document}$ represents the genotypes of the *s*th SNP from *N* individuals.
When *R*=1 (i.e., the phenotype is univariate), a regression model is commonly adopted to model ***y*** ^*r*^ as a function of ***g*** ^*s*^ with covariates in the model to adjust for confounding factors or to increase the precision of estimates. When *R*\>1 (i.e., multivariate phenotypes), Yang et al. (2016) \[[@CR14]\] proposed a two-step approach. In the first step, for each phenotype *r*, a marginal *p*-value, *p* ~*rs*~, is derived from a likelihood ratio test under a linear regression of ***y*** ^*r*^ on ***g*** ^*s*^. The next step is to test association between *R* multivariate phenotypes and the *s*th SNP based on these marginal *p*-values of *p* ~1*s*~,...,*p* ~*Rs*~. The Fisher combination function is used to combine them and the test statistic is defined as $$\documentclass[12pt]{minimal}
\usepackage{amsmath}
\usepackage{wasysym}
\usepackage{amsfonts}
\usepackage{amssymb}
\usepackage{amsbsy}
\usepackage{mathrsfs}
\usepackage{upgreek}
\setlength{\oddsidemargin}{-69pt}
\begin{document} $$ \xi_{s} = \sum_{r=1}^{R} -2\log (p_{rs}). $$ \end{document}$$
The SNP *s* is claimed to be associated with the *R* multivariate phenotypes if *ξ* ~*s*~ is statistically significant. Although −2 log(*p* ~*rs*~) follows a chi-square distribution with 2 degrees of freedom, *ξ* ~*s*~, which is a sum of *dependent* chi-square random variables, does not follow a chi-square distribution with 2*R* degrees of freedom. The permutation method may be adopted to calculate the *p*-value of *ξ* ~*s*~ but it is computationally too expensive in the context of GWAS (see Yang et al. (2016) \[[@CR14]\] for details).
Under the the null hypothesis, the statistic *ξ* ~*s*~ is the sum of dependent chi-square statistics. Thus, the null distribution of *ξ* ~*s*~ follows a gamma distribution \[[@CR23], [@CR24]\] with the mean and variance being functions of the shape parameter *k* and the scale parameter *θ*: $$\documentclass[12pt]{minimal}
\usepackage{amsmath}
\usepackage{wasysym}
\usepackage{amsfonts}
\usepackage{amssymb}
\usepackage{amsbsy}
\usepackage{mathrsfs}
\usepackage{upgreek}
\setlength{\oddsidemargin}{-69pt}
\begin{document} $$\begin{array}{*{20}l} E[\xi_{s}] &= k\theta,\\ Var[\xi_{s}] &= k \theta^{2}. \end{array} $$ \end{document}$$
Applying the method of moments, we can derive the following equations based on the first two sample moments: $$\documentclass[12pt]{minimal}
\usepackage{amsmath}
\usepackage{wasysym}
\usepackage{amsfonts}
\usepackage{amssymb}
\usepackage{amsbsy}
\usepackage{mathrsfs}
\usepackage{upgreek}
\setlength{\oddsidemargin}{-69pt}
\begin{document} $$\begin{array}{*{20}l} k\theta &= 2R, \end{array} $$ \end{document}$$
$$\documentclass[12pt]{minimal}
\usepackage{amsmath}
\usepackage{wasysym}
\usepackage{amsfonts}
\usepackage{amssymb}
\usepackage{amsbsy}
\usepackage{mathrsfs}
\usepackage{upgreek}
\setlength{\oddsidemargin}{-69pt}
\begin{document} $$\begin{array}{*{20}l} k \theta^{2} &= 4R + \sum_{r\neq r'} cov(-2\log (p_{rs}), -2\log (p_{r's})). \end{array} $$ \end{document}$$
Yang et al. (2016) \[[@CR14]\] showed that the pairwise sample correlation $\documentclass[12pt]{minimal}
\usepackage{amsmath}
\usepackage{wasysym}
\usepackage{amsfonts}
\usepackage{amssymb}
\usepackage{amsbsy}
\usepackage{mathrsfs}
\usepackage{upgreek}
\setlength{\oddsidemargin}{-69pt}
\begin{document}$\rho _{rr'}=\text {cor}(\boldsymbol {y}^{r},\boldsymbol {y}^{r'})\phantom {\dot {i}\!}$\end{document}$ can be used to accurately estimate $\documentclass[12pt]{minimal}
\usepackage{amsmath}
\usepackage{wasysym}
\usepackage{amsfonts}
\usepackage{amssymb}
\usepackage{amsbsy}
\usepackage{mathrsfs}
\usepackage{upgreek}
\setlength{\oddsidemargin}{-69pt}
\begin{document}$cov(-2\log (p_{rs}), -2\log (p_{r's}))\phantom {\dot {i}\!}$\end{document}$ as follows: $$\documentclass[12pt]{minimal}
\usepackage{amsmath}
\usepackage{wasysym}
\usepackage{amsfonts}
\usepackage{amssymb}
\usepackage{amsbsy}
\usepackage{mathrsfs}
\usepackage{upgreek}
\setlength{\oddsidemargin}{-69pt}
\begin{document} $$ {}cov\left(-2\log (p_{rs}), -2\log (p_{r's})\right) \approx \sum_{l=1}^{5} c_{l} \rho_{rr'}^{2l} - \frac{c_{1}}{N} \left(1-\rho_{rr'}^{2}\right)^{2}\!, $$ \end{document}$$
where *c* ~1~=3.9081, *c* ~2~=0.0313, *c* ~3~=0.1022, *c* ~4~=−0.1378 and *c* ~5~=0.0941. This approximation is very accurate as the maximum difference is less than 0.0001. Thus, we can efficiently estimate *k* and *θ* using Eqs. ([2](#Equ2){ref-type=""}) and ([3](#Equ3){ref-type=""}) with the *cov*(·) in Eq. ([3](#Equ3){ref-type=""}) substituted by the right-hand side of Eq. ([4](#Equ4){ref-type=""}).
The proposed method for related subjects with multivariate phenotypes {#Sec4}
---------------------------------------------------------------------
The multivariate method described in the previous section only applies to independent subjects. When multivariate phenotypes data are collected from genetically related subjects, there are two types of correlations: 1) the correlation between multivariate phenotypes (even when the subjects are independent); and 2) the correlation due to genetically related subjects (even when the phenotype is univariate). The approach described in the previous section only addresses the first type of correlation. To address both correlations in the regression model, the marginal regression model in the first step needs to be modified to account for genetically related subjects. Recall that the original regression model has the form of $$\documentclass[12pt]{minimal}
\usepackage{amsmath}
\usepackage{wasysym}
\usepackage{amsfonts}
\usepackage{amssymb}
\usepackage{amsbsy}
\usepackage{mathrsfs}
\usepackage{upgreek}
\setlength{\oddsidemargin}{-69pt}
\begin{document} $$ \boldsymbol{y}^{r} = \boldsymbol{\alpha}^{r} + \boldsymbol{g}^{s} \boldsymbol{\beta}^{r} + \boldsymbol{\epsilon}^{r}, $$ \end{document}$$ where ***α*** ^*r*^ is the intercept term, ***β*** ^*r*^ is the genetic effect and ***ε*** ^*r*^∼*N*(0,*σ* ^2^ ***I***) is a vector of error terms. When the subjects are genetically related, we modify the model to be a linear mixed model: $$\documentclass[12pt]{minimal}
\usepackage{amsmath}
\usepackage{wasysym}
\usepackage{amsfonts}
\usepackage{amssymb}
\usepackage{amsbsy}
\usepackage{mathrsfs}
\usepackage{upgreek}
\setlength{\oddsidemargin}{-69pt}
\begin{document} $$ \boldsymbol{y}^{r} = \boldsymbol{\alpha}^{r} + \boldsymbol{g}^{s} \boldsymbol{\beta}^{r} + \boldsymbol{z}^{r}+\boldsymbol{\epsilon}^{r}, $$ \end{document}$$
where the added term ***z*** ^*r*^ is a random effect and it follows $\documentclass[12pt]{minimal}
\usepackage{amsmath}
\usepackage{wasysym}
\usepackage{amsfonts}
\usepackage{amssymb}
\usepackage{amsbsy}
\usepackage{mathrsfs}
\usepackage{upgreek}
\setlength{\oddsidemargin}{-69pt}
\begin{document}$N(0,\sigma _{g}^{2} \boldsymbol {K})$\end{document}$ where the matrix ***K*** is called the genetic relationship matrix (GRM) \[[@CR25]\].
Direct calculation of the best linear unbiased estimates of the fixed effects and the best linear unbiased predictors of the random effects for a large sample size is extremely slow and may be beyond the memory capacity of most computers. Many flexible and efficient methods have been developed to carry out GWAS using linear mixed models. For example, the efficient algorithm implemented in GCTA \[[@CR25]\] uses the restricted maximum likelihood (REML) method to estimate $\documentclass[12pt]{minimal}
\usepackage{amsmath}
\usepackage{wasysym}
\usepackage{amsfonts}
\usepackage{amssymb}
\usepackage{amsbsy}
\usepackage{mathrsfs}
\usepackage{upgreek}
\setlength{\oddsidemargin}{-69pt}
\begin{document}$\sigma _{g}^{2}$\end{document}$ and *σ* ^2^ under the null model while the GRM ***K*** was estimated from all the SNPs. To test *H* ~0~:***β*** ^*r*^=0, the estimates of the random effects ($\documentclass[12pt]{minimal}
\usepackage{amsmath}
\usepackage{wasysym}
\usepackage{amsfonts}
\usepackage{amssymb}
\usepackage{amsbsy}
\usepackage{mathrsfs}
\usepackage{upgreek}
\setlength{\oddsidemargin}{-69pt}
\begin{document}$\sigma _{g}^{2}$\end{document}$, *σ* ^2^, and ***K***) under the null model were plugged in for the estimation of the variance of the $\documentclass[12pt]{minimal}
\usepackage{amsmath}
\usepackage{wasysym}
\usepackage{amsfonts}
\usepackage{amssymb}
\usepackage{amsbsy}
\usepackage{mathrsfs}
\usepackage{upgreek}
\setlength{\oddsidemargin}{-69pt}
\begin{document}$\boldsymbol {\hat {\beta }}^{r}$\end{document}$. In this way, the Wald test statistic can be constructed. Under *H* ~0~, this statistic follows an asymptotical chi-squared distribution with 1 degree of freedom; and the corresponding marginal *p*-value indicates the strength of association between the SNP and a marginal phenotype.
The resulting marginal *p*-values, *p* ~1*s*~,...,*p* ~*Rs*~, can then be combined together using the Fisher combination function in Eq. ([1](#Equ1){ref-type=""}) to form the test statistic *ξ* ~*s*~ for the association between the *s*th SNP and *R* multivariate phenotypes. Based on the linear mixed model in Eq. ([5](#Equ5){ref-type=""}), it can be shown that for different traits ***y*** ^*r*^ and $\documentclass[12pt]{minimal}
\usepackage{amsmath}
\usepackage{wasysym}
\usepackage{amsfonts}
\usepackage{amssymb}
\usepackage{amsbsy}
\usepackage{mathrsfs}
\usepackage{upgreek}
\setlength{\oddsidemargin}{-69pt}
\begin{document}$\phantom {\dot {i}\!}\boldsymbol {y}^{r'}$\end{document}$, the covariance between them is $$\documentclass[12pt]{minimal}
\usepackage{amsmath}
\usepackage{wasysym}
\usepackage{amsfonts}
\usepackage{amssymb}
\usepackage{amsbsy}
\usepackage{mathrsfs}
\usepackage{upgreek}
\setlength{\oddsidemargin}{-69pt}
\begin{document} $$\text{cov}\left(\boldsymbol{y}^{r}, \boldsymbol{y}^{r'}\right) =\nu\sigma_{g}^{2} \boldsymbol{K} + \rho_{rr'} \sigma^{2} I, $$ \end{document}$$ where *ν* is the genetic correlation due to related subjects and $\documentclass[12pt]{minimal}
\usepackage{amsmath}
\usepackage{wasysym}
\usepackage{amsfonts}
\usepackage{amssymb}
\usepackage{amsbsy}
\usepackage{mathrsfs}
\usepackage{upgreek}
\setlength{\oddsidemargin}{-69pt}
\begin{document}$\phantom {\dot {i}\!}\rho _{rr'}$\end{document}$ is the correlation between phenotypes (even when only independent subjects are involved). Because the test statistic *ξ* ~*s*~ is a function of *p* ~1*s*~,...,*p* ~*Rs*~ which are derived with the relatedness between subjects being adjusted by the random effect ***z*** ^*r*^ in Model ([5](#Equ5){ref-type=""}), we can use the pairwise correlation between residuals, $\documentclass[12pt]{minimal}
\usepackage{amsmath}
\usepackage{wasysym}
\usepackage{amsfonts}
\usepackage{amssymb}
\usepackage{amsbsy}
\usepackage{mathrsfs}
\usepackage{upgreek}
\setlength{\oddsidemargin}{-69pt}
\begin{document}$\phantom {\dot {i}\!}\text {cor}(\boldsymbol {\epsilon }^{r}, \boldsymbol {\epsilon }^{r'})$\end{document}$, from this model to estimate $\documentclass[12pt]{minimal}
\usepackage{amsmath}
\usepackage{wasysym}
\usepackage{amsfonts}
\usepackage{amssymb}
\usepackage{amsbsy}
\usepackage{mathrsfs}
\usepackage{upgreek}
\setlength{\oddsidemargin}{-69pt}
\begin{document}$\phantom {\dot {i}\!}\rho _{rr'}$\end{document}$ and plug this estimate into Eq. ([4](#Equ4){ref-type=""}). In this way, the null distribution of *ξ* ~*s*~ can be approximated.
Although the GRM associated with the random effect ***z*** ^*r*^, in principle, contains information about the population structure resulting from systematic differences in ancestry, the random effect is not likely to be estimated perfectly in practice. For this reason, we proposed to extend the linear mixed model in Eq. ([5](#Equ5){ref-type=""}) by adding principal components \[[@CR26]\] estimated from genotype data as covariates for the purpose of improving the precision of the estimates for marginal *p*-values. This was based on the results of Astle el al. (2009) \[[@CR15]\] showing that combining GRM with principal components could account for the population structure and relatedness better. Because contemporary American genomes resulted from a sequence of admixture process involving individuals descended from multiple ancestral population groups \[[@CR27]\], this additional adjustment may potentially be a crucial step and its effectiveness was evaluated through simulation studies described in the next section.
Results {#Sec5}
=======
Simulation studies {#Sec6}
------------------
### Generating the genotype data {#Sec7}
We simulated genotype data based on two different population structures (parents from the same population or parents from two different populations) and two different relatedness structures (independent subjects or related subjects) so there were four different types of data sets reflecting all possible combinations. We generated a set of allele frequencies (corresponding to a total of 10,000 SNPs) from uniform random numbers between 0.1 and 0.9 to represent Population I; and another set of allele frequencies to represent Population II. Given a set of population allele frequencies, we can generate the genotypes of parents from the particular population. Through random mapping, we can generate three types of parents (1/3 each): (1) both parents from Population I; (2) both parents from Population II; and (3) one parent from Population I and the other parent from Population II.
Once we had simulated parents' genotypes, the genes were dropped down the pedigree according to Mendel's law to simulate children's genotypes. Our procedure ensured that children from different families represented independent subjects and children within a family represented strongly related subjects. To generate a sample of independent subjects, we simulated 1000 families with one child from each family. To generate a sample of related subjects, we simulated 250 families with four children in each family. Depending on whether parents' genotypes were generated from one population (either Population I or Population II) or from two different populations, we had four scenarios of children genotype samples: 1) independent samples from non-admixed/isolated population (Non-admixed Independent); 2) related samples from non-admixed/isolated population (Non-admixed Related); 3) independent samples from admixed population (Admixed Independent); and 4) related samples from admixed population (Admixed Related).
### Evaluating the phenotype correlation estimates {#Sec8}
To evaluate the methods for estimating the correlation between phenotypes $\documentclass[12pt]{minimal}
\usepackage{amsmath}
\usepackage{wasysym}
\usepackage{amsfonts}
\usepackage{amssymb}
\usepackage{amsbsy}
\usepackage{mathrsfs}
\usepackage{upgreek}
\setlength{\oddsidemargin}{-69pt}
\begin{document}$\phantom {\dot {i}\!}\rho _{rr'}$\end{document}$, we simulated bivariate phenotypes using bivariate normal (BVN) random variables. An additive genetic effect was used to model the relationship between the genotype and bivariate phenotypes. Let *e* be the genetic effect size. The mean value of the marginal phenotype *μ* ~*r*~(*r*=1,2) was −*e* if the genotype was *AA*; 0 if the genotype was *AB*; and *e* if the genotype was *BB*. The specific model to simulate phenotypes is: $$\documentclass[12pt]{minimal}
\usepackage{amsmath}
\usepackage{wasysym}
\usepackage{amsfonts}
\usepackage{amssymb}
\usepackage{amsbsy}
\usepackage{mathrsfs}
\usepackage{upgreek}
\setlength{\oddsidemargin}{-69pt}
\begin{document} $$ \left(\begin{array}{c} Y_{1}\\ Y_{2} \end{array}\right) \sim \text{BVN } \left(\left(\begin{array}{c} \mu_{1}\\ \mu_{2} \end{array}\right), \Sigma \right), $$ \end{document}$$
where *Σ* is a 2×2 symmetric matrix with the diagonal elements being 1 and the off-diagonal element *ρ*. The value of *ρ* determines the correlation between the phenotypes. For each data set, the values of *ρ* ranged from 0 (independent) to 0.9 (highly dependent), and the values of *e* ranged from 0 (no effect) to 1 (large effect).
Each configuration was simulated 1000 times. In each simulated data set, we calculated the estimate for $\documentclass[12pt]{minimal}
\usepackage{amsmath}
\usepackage{wasysym}
\usepackage{amsfonts}
\usepackage{amssymb}
\usepackage{amsbsy}
\usepackage{mathrsfs}
\usepackage{upgreek}
\setlength{\oddsidemargin}{-69pt}
\begin{document}$\phantom {\dot {i}\!}\rho _{rr'}$\end{document}$ based on three methods: Method 1: the residuals from the linear mixed model; Method 2: the residuals from the linear mixed model with the first ten principal components as covariates; Method 3: the correlation between simulated phenotypes.
The third one was a näive method that did not adjust for the correlation due to related subjects and thus was expected to overestimate $\documentclass[12pt]{minimal}
\usepackage{amsmath}
\usepackage{wasysym}
\usepackage{amsfonts}
\usepackage{amssymb}
\usepackage{amsbsy}
\usepackage{mathrsfs}
\usepackage{upgreek}
\setlength{\oddsidemargin}{-69pt}
\begin{document}$\phantom {\dot {i}\!}\rho _{rr'}$\end{document}$. The program GCTA was used to fit linear mixed models and calculate corresponding residuals.
The simulation results based on the three methods of estimating $\documentclass[12pt]{minimal}
\usepackage{amsmath}
\usepackage{wasysym}
\usepackage{amsfonts}
\usepackage{amssymb}
\usepackage{amsbsy}
\usepackage{mathrsfs}
\usepackage{upgreek}
\setlength{\oddsidemargin}{-69pt}
\begin{document}$\phantom {\dot {i}\!}\rho _{rr'}$\end{document}$ were shown in Figs. [1](#Fig1){ref-type="fig"}, [2](#Fig2){ref-type="fig"}, [3](#Fig3){ref-type="fig"} to [4](#Fig4){ref-type="fig"} using boxplots to represent the distribution of $\documentclass[12pt]{minimal}
\usepackage{amsmath}
\usepackage{wasysym}
\usepackage{amsfonts}
\usepackage{amssymb}
\usepackage{amsbsy}
\usepackage{mathrsfs}
\usepackage{upgreek}
\setlength{\oddsidemargin}{-69pt}
\begin{document}$\hat {\rho } - \rho $\end{document}$. A good method was identified by choosing the one with the mean values of $\documentclass[12pt]{minimal}
\usepackage{amsmath}
\usepackage{wasysym}
\usepackage{amsfonts}
\usepackage{amssymb}
\usepackage{amsbsy}
\usepackage{mathrsfs}
\usepackage{upgreek}
\setlength{\oddsidemargin}{-69pt}
\begin{document}$\hat {\rho } - \rho $\end{document}$ close to zero. Comparing the accuracies of these three methods, it shows that the accuracy of estimation depended on the values of the true correlation and effect size. When the effect size *e* was 0 (no effect) or when the phenotype correlation was highly correlated (near 0.9), all three methods performed well. On the other hand, when the effect size was large (*e*=1) and the phenotype correlation was small (near 0), all three methods over estimated the true correlation. However, in this situation, the methods using residuals from linear mixed models performed better than the näive method. To our surprise, adding principal components in the linear mixed model did not substantially improve the accuracy of estimates. Because of the poor performance of the näive method, it was not used for the simulations evaluating the type I error rate and power. Fig. 1The accuracy of correlation estimations based on three methods for data from *non-admixed independent* subjects (Method 1: linear mixed model; Method 2: linear mixed model with principal components; Method 3: correlation without adjusting for relatedness) Fig. 2The accuracy of correlation estimations based on three methods for data from *non-admixed related* subjects (Method 1: linear mixed model; Method 2: linear mixed model with principal components; Method 3: correlation without adjusting for relatedness) Fig. 3The accuracy of correlation estimations based on three methods for data from *admixed independent* subjects (Method 1: linear mixed model; Method 2: linear mixed model with principal components; Method 3: correlation without adjusting for relatedness) Fig. 4The accuracy of correlation estimations based on three methods for data from *admixed related* subjects (Method 1: linear mixed model; Method 2: linear mixed model with principal components; Method 3: correlation without adjusting for relatedness)
### Evaluating the type I error and power of the proposed method {#Sec9}
Because the proposed method was designed to identify pleiotropic genes, evaluating the performance of the multivariate association test in terms of the type I error and power is essential. We simulated four correlated phenotypes using multivariate normal (MVN) random variables. The values of the genetic effect size, *e*, were 0 (no effect), 0.1 (medium effect), and 0.2 (large effect) and the values of the correlation *ρ* were 0 (independent), 0.4 (moderate correlated), and 0.8 (highly correlated). Each configuration was repeated 1000 times.
Figures [5](#Fig5){ref-type="fig"} and [6](#Fig6){ref-type="fig"} shows the distribution of − log10(*p*) for different values of correlation *ρ* and genetic effect size *e*. Large values of − log10(*p*)-value are equivalent to small *p*-values. Thus, when the effect size was large, we expected − log10(*p*) to be large. Based on our configuration to generate phenotypes, there was no difference between the four marginal *p*-values. Hence, we only presented the distribution of marginal *p*-values corresponding to the first marginal phenotype. The multivariate *p*-values were derived using the proposed method with the correlation estimated by the residuals from the linear mixed model with the first ten principal components as covariates. The findings from this simulation study were summarized as follows: When there was no genetic effect (*e*=0), both the marginal and multivariate methods produced uniform *p*-values distributions which reflected the null distribution of *p*-values. When the genetic effect size increased, the value of − log10(*p*) increased. Therefore, the simulation showed that both marginal and multivariate tests were unbiased.When the population structure and relatedness were fixed, increasing the correlation between phenotypes decreased the power of multivariate tests. The negative relationship between the correlation of multivariate phenotypes and power has also been observed in Yang et al. (2016) for various multivariate testing statistics \[[@CR14]\].When the genetic effect was not zero, the proposed multivariate method was more powerful than the marginal test in all situations. The advantages of using the multivariate method was most evident when the correlation between phenotypes was small to moderate. But even when the correlation between phenotypes was as large as 0.8, the multivariate method was still more powerful than the marginal tests. Therefore, combining multivariate phenotypes could increase the power of test.When the sample size was held constant (recall that the sample size was the same across different population structure and relatedness in our simulation), the difference in power between admixed and non-admixed samples or between independent or related samples were very small. Fig. 5The distribution of values of − log10(*p*) using samples with different population structures and relatedness for various correlations *ρ* between phenotypes under the null hypothesis: the effect size *e*=0. The white boxes correspond to the marginal test; and the gray boxes correspond to the multivariate test Fig. 6The distribution of values of − log10(*p*) using samples with different population structures and relatedness for various correlations *ρ* between phenotypes under the alternative hypotheses: the effect size *e*=0.1,0.2. The white boxes correspond to the marginal test; and the gray boxes correspond to the multivariate test
### Comparing the proposed method with the mvLMMs method {#Sec10}
We further evaluated the performance of the proposed method in comparison to a competing method, the multivariate linear mixed model (mvLMMs) method, that has been implemented in the GEMMA \[[@CR28]\] software. Here, we adopted the most complex situation from the previous simulation experiment in which genotypes were simulated based on related people from admixed populations (Admixed Related). Specifically, we simulated genotypes from 250 families each of which had four children and resulted in 1000 related individuals. Next, we simulated the following phenotypes from these genotypes by extending Model ([6](#Equ6){ref-type=""}) to $$\documentclass[12pt]{minimal}
\usepackage{amsmath}
\usepackage{wasysym}
\usepackage{amsfonts}
\usepackage{amssymb}
\usepackage{amsbsy}
\usepackage{mathrsfs}
\usepackage{upgreek}
\setlength{\oddsidemargin}{-69pt}
\begin{document} $$\left(\begin{array}{c} Y_{1}\\ \vdots\\ Y_{4} \end{array}\right) \sim \text{BVN } \left(\left(\begin{array}{c} \mu_{1}\\ \vdots\\ \mu_{4} \end{array}\right), \Sigma \right), $$ \end{document}$$ where *Σ* was a 4×4 symmetric matrix with the diagonal elements being 1 and the off-diagonal element *ρ*. We manipulated the value of *ρ* to be 0.1 (weak correlated) or 0.5 (moderate correlated). Let ***e***=(*e* ~1~,...,*e* ~4~) be the genetic effect sizes corresponding to the phenotypes *Y* ~1~,...,*Y* ~4~. We considered the following combinations: Small effect sizes: ***e***=(0.1,0.1,0.1,0.1);Increasing effect sizes: ***e***=(0.05,0.1,0.15,0.2);Medium effect sizes: ***e***=(0.15,0.15,0.15,0.15).
We did not consider the situation of no effect (i.e., ***e***=(0,0,0,0)) because both methods have been shown to control the type I error.
We simulated each configuration 1000 times. For our proposed method, we estimated pairwise correlation $\documentclass[12pt]{minimal}
\usepackage{amsmath}
\usepackage{wasysym}
\usepackage{amsfonts}
\usepackage{amssymb}
\usepackage{amsbsy}
\usepackage{mathrsfs}
\usepackage{upgreek}
\setlength{\oddsidemargin}{-69pt}
\begin{document}$\phantom {\dot {i}\!}\rho _{rr^{\prime }}$\end{document}$ based on **Method 1** described in the previous section for its good performance.
Figure [7](#Fig7){ref-type="fig"} shows the distribution of − log10(*p*) for different values of the correlation *ρ* and the genetic effect sizes ***e***. A powerful method should result in small *p*-values (or equivalently, large values of − log10(*p*)). The findings form this simulation study were summarized as follows: The power of both methods depended on the effect sizes. When the effect sizes were increased from small to medium, the power of both methods increased. More importantly, the scale of such increase was larger for the proposed method.When the effect sizes were fixed, both methods had higher power when the correlation between phenotypes was weak.When the effect sizes were not equal among marginal phenotypes, the proposed method still maintained its high performance.Overall, the proposed method was more powerful than the mvLMMs method. The proposed method had a larger median value of − log10(*p*) compared to the mvLMMs method in 5 our of the 6 configurations. The mvLMMs only achieved the same level of performance when the phenotypes had a medium correlation and the effect sizes were increasing. Fig. 7Comparing the power of two competing methods: the distribution of values of − log10(*p*) under different correlations *ρ* and effect sizes *e*. The gray boxes correspond to the proposed Fisher method; and the white boxes correspond to the mvLMMs method
In addition to high power, the proposed method has the advantage of being computationally efficient even when the number of phenotypes is large. The mvLMMs method, on the other hand, was recommended for a modest number of phenotypes (less than 10) due to computational and statistical barriers of the EM algorithm \[[@CR21]\].
Real data analysis {#Sec11}
------------------
We demonstrate the application of the proposed method by conducting analysis on the data from the Study of Addiction: Genetics and Environment (SAGE).
The SAGE is a study that collected data from three large scale studies in the substance abuse field: the Collaborative Study on the Genetics of Alcoholism (COGA), the Family Study of Cocaine Dependence (FSCD), and the Collaborative Genetic Study of Nicotine Dependence (COGEND). The total number of subjects in all three studies was 4121. Each subject was genotyped using the Illumina Human 1M-Duo beadchip which contains over 1 million SNP markers. From the original 4121 individuals, some subjects were genotyped twice so we eliminated duplicate samples and the sample size was reduced to be 4112. Although dbGap provided a PED file to show pedigree and relationship among participants, we used the KING program \[[@CR29]\] to verify their relationship. As a result, we confirmed and identified 3921 unrelated individuals and the remaining 191 were family members of these unrelated individuals. Using the chosen 4112 individuals, we restricted SNPs to 22 autosomes and conducted quality control of SNPs based on the minor allele frequency (\>0.01), Hardy-weinberg equilibrium test (*p*-value \>10^−5^), and frequency of missingness per SNP (\<0.05) \[[@CR30]\]. The final total number of SNPs chosen for analysis was 711,038.
Because our research aimed to identify the SNPs associated with the risk for alcohol dependence, four correlated phenotypes were used for the analysis: age_first_drink:the age when the participant had a drink containing alcohol the first time.ons_reg_drink: the onset age of regular drinking (drinking once a month for 6 months or more).age_first_got_drunk:the age when the participant got drunk the first time.alc_sx_tot: the number of alcohol dependence symptoms endorsed.
To deal with missing values in any of these four phenotypes, we imputed them using the mi package \[[@CR31]\] from R software. The sample distributions of phenotypes and their pairwise correlations are shown in Fig. [8](#Fig8){ref-type="fig"} and Table [1](#Tab1){ref-type="table"}, respectively. The first three variables are the onset ages of important "milestone" events of alcoholism. Earlier onset ages are indicators for higher vulnerability and have been shown to predict later progression to alcohol dependence \[[@CR32]\]. Thus, they were expected to be positively correlated with each other and negatively correlated with the number of alcohol dependence symptoms. Fig. 8The distributions of four phenotypes indicating the risk for alcohol dependence using real data from 4121 participants Table 1The correlations among the 4 alcohol dependence phenotypesCorrelationage_first_drinkons_reg_drinkage_first_got_drunkalc_sx_totage_first_drink1.000.470.60-0.47ons_reg_drink0.471.000.55-0.26age_first_got_drunk0.600.551.00-0.30alc_sx_tot-0.47-0.26-0.301.00
We conducted marginal genome-wide association tests on each of these four phenotypes using the GCTA program to account for relatedness among subjects. We also added the first ten principal components to increase the precision of estimates. In addition to these principal components, the participant's gender, age at interview, and self-identified race were included as covariates in the model. The regression model for the marginal phenotype is $$\documentclass[12pt]{minimal}
\usepackage{amsmath}
\usepackage{wasysym}
\usepackage{amsfonts}
\usepackage{amssymb}
\usepackage{amsbsy}
\usepackage{mathrsfs}
\usepackage{upgreek}
\setlength{\oddsidemargin}{-69pt}
\begin{document} $$ \boldsymbol{y}^{r} = \boldsymbol{\alpha}^{r} + \boldsymbol{x} \boldsymbol{\eta}^{r}+ \boldsymbol{g}^{s} \boldsymbol{\beta}^{r} + \boldsymbol{z}^{r}+\boldsymbol{\epsilon}^{r}, $$ \end{document}$$ where ***x*** contains the participant's first ten principal components, gender, age at interview, and self-identified race, and the corresponding regression coefficients ***η*** ^*r*^ are treated as fixed effects. The QQ-plots of the *p*-values for the marginal association tests are shown in Fig. [9](#Fig9){ref-type="fig"}. The QQ-plot of the *p*-values for the proposed multivariate tests is displayed in Fig. [10](#Fig10){ref-type="fig"}. Since a primary assumption in GWAS is that most SNPs are not associated with the phenotype studied, most points in the QQ-plots should not deviate from the diagonal line. Deviations from the diagonal line may indicate that potential confounders such as the population structure or relatedness are not adjusted. Both of Figs. [9](#Fig9){ref-type="fig"} and [10](#Fig10){ref-type="fig"} indicate that potential confounders were well adjusted in our real data analysis. Fig. 9The QQ-plots of − log10(*p*)-values from marginal tests for four alcohol dependence phenotypes using real data from 4121 participants Fig. 10The QQ-plot of − log10(*p*)-values from the multivariate test for four alcohol dependence phenotypes using real data from 4121 individuals
To identify the SNPs associated with the risk for alcohol dependence, we declared significant SNPs if their *p*-values were less than the significance level of 10^−6^. Based on the marginal tests, two SNPs (*rs*9825428, *p*-value =2.3962×10^−7^; *rs*16822575, *p*-values =7.6411×10^−7^) were associated with age_first_drink; one SNP (*rs*11157640, *p*-value =2.2658×10^−7^) was associated with ons_reg_drink; one SNP (*rs*7700665, *p*-value =7.0618×10^−7^) was associated with age_first_got_drunk; and one SNP (*rs*10914375, *p*-value =3.2978×10^−7^) was associated with alc_sx_tot. On the other hand, using the proposed multivariate method, two SNPs (*rs*7523645, *p*-value =1.1872×10^−7^; and *rs*11157640, *p*-value =6.6655*e*×10^−7^) were significantly associated with these four correlated phenotypes for the risk of alcohol dependence.
Based on the findings in the marginal tests, we identified five SNPs each of which was associated with an individual phenotype; none of these five SNPs was associated with two or more phenotypes. Thus, the marginal tests were limited in terms of finding pleiotropic genes associated with multivariate phenotypes. In contrast, using the proposed multivariate method, we identified two SNPs associated with the four phenotypes together and one of these two SNPs (*rs*11157640) was also found to be associated with ons_reg_drink in the marginal test. Hence, combining multiple phenotypes not only can increase the power of identifying SNPs that may not be identified by marginal tests but also can provide insight into the pleiotropic genes contributing to the common risk expressed by multivariate phenotypes.
Discussion {#Sec12}
==========
Although the simulation shows that adding principal components as covariates to the linear mixed model did not substantially improve the accuracy of estimating the correlation between phenotypes, it can adjust for potential population structure and cryptic relatedness in GWAS as well as improve the estimation of marginal genetic effects \[[@CR15]\]. More research is needed to study the optimal number of principal components to be added to the proposed model. Moreover, the simulation study did not consider negative correlations between multivariate phenotypes because the situation is rare in practice. Nevertheless, previous studies have demonstrated that multivariate methods such as MANOVA \[[@CR33]\] and MultiPhen \[[@CR34]\] tend to have higher power to detect a pleiotropic gene in such a situation. Furthermore, in this study, we only considered *continuous* multivariate phenotypes. Future studies may extend the methodology work to the case of correlated *discrete* phenotypes. For example, in the substance abuse field, many outcomes are zero-inflated count data \[[@CR35]\] or ordinal data \[[@CR36]\]. A future direction that is particularly challenging is how to analyze multivariate phenotypes with different measurement scales.
Conclusions {#Sec13}
===========
In this study, we propose a new multivariate method for GWAS when multivariate quantitative phenotypes are used to indicate the risk for a complex disease and the data are collected from related subjects. Our approach is a two-step approach. The first step models the association between the genotype and marginal phenotype using a linear mixed model. The linear mixed model uses a random effect to account for the relatedness between subjects. We also extend the linear mixed model by adding principal components as covariates to adjust for potential population structures. Since the sample size in GWAS generally reaches thousands and a certain population structure exists within the subjects, the benefit from adjusting for population structures out-weights the loss of ten degrees of freedom in the linear mixed model. The linear mixed model in the first step also has the flexibility to add demographic variables or other confounding variables to improve precision of estimation. The second step of the proposed method uses the correlation between residuals of the linear mixed model to estimate the null distribution of the Fisher combination test statistic.
The simulation results show that our proposed method controls the type I error rate and is more powerful than the marginal tests across different population structures (admixed or non-admixed) and relatedness (related or independent). The proposed method is also computationally efficient. The first step takes advantage of the efficient program GCTA to carry out marginal tests under linear mixed models. In practice, a few hours are sufficient to derive all marginal *p*-values. The second step only takes a few minutes to compute the Fisher combination test statistic and its null distribution using R software. Furthermore, the real data analysis on the SAGE database demonstrates that applying the multivariate association test may facilitate identification of the pleiotropic genes contributing to the risk for alcohol dependence commonly expressed by the four correlated phenotypes.
Not
: applicable.
We would like to acknowledge our usage of the data from the Study of Addiction: Genetics and Environment (SAGE), which is part of the Gene Environment Association Studies (GENEVA) initiative supported by the National Human Genome Research Institute (dbGaP study accession phs000092.v1.p1).
Funding {#d29e3252}
=======
This research was supported by National Institutes of Health (NIH) grants: R01 DA035183 (A. Buu), R01 AI079139, and R01 HL118267 (L. K. Williams). The funding agencies had no role in study design, analysis, interpretation of results, decision to publish, or preparation of the manuscript. The content is solely the responsibility of the authors and does not necessarily represent the official views of the NIH.
Availability of data and materials {#d29e3257}
==================================
The simulated datasets are freely available via GitHub at <https://github.com/jyangstat/Fisher_LMM>.
The Study of Addiction: Genetics and Environment (SAGE) is available at <http://www.ncbi.nlm.nih.gov/projects/gap/cgi-bin/study.cgi?study_id=phs000092.v1.p1>.
JJY and AB conducted the literature review and simulation studies; LWK and AB provided funding and obtained the approval of data access; JJY, LKW, and AB performed real data analysis and interpreted the results; JJY and AB proposed the statistical methods and wrote the manuscript with input from LKW. All authors read and approved the final manuscript.
Ethics approval and consent to participate {#d29e3280}
==========================================
Not applicable.
Consent for publication {#d29e3285}
=======================
Not applicable.
Competing interests {#d29e3290}
===================
The authors declare that they have no competing interests.
Publisher's Note {#d29e3295}
================
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
| {
"pile_set_name": "PubMed Central"
} |
INTRODUCTION
============
Pregnancy in a rudimentary noncommunicating uterine horn has already been described in the literature.^[@B1]--[@B8]^ Only 9 cases have been reported of a placenta accreta in a pregnant rudimentary uterine horn, but this situation led to hemorrhagic rupture of the uterus in 8 women.^[@B1],[@B3]^ In the literature, pregnant rudimentary uterus ruptures, with or without placenta accreta, tend to occur during the second or third trimester of pregnancy, and in most cases the consequent hemorrhage threatens women\'s lives.^[@B1],[@B3]--[@B5],[@B7]^ However, early diagnosis may be possible and would allow the resection of the rudimentary uterine horn with scarce maternal morbidity.^[@B6]^ We report the case of a rudimentary uterine horn pregnancy with a placenta percreta revealed and managed laparoscopically at 7-weeks gestation, with favorable outcomes, and we argue for the laparoscopic removal of rudimentary uterine horns when they are revealed in young women who desire further pregnancies.
CASE REPORT
===========
A 28-year-old, healthy, nulliparous woman underwent a routine pelvic ultrasound examination at 7-weeks gestation. As regards her antecedents, she had a right ectopic fallopian pregnancy 4 months previously that had been treated laparoscopically at another establishment. She was not aware of any particular findings during this surgical procedure that led to the removal of the ectopic pregnancy by a right salpingotomy.
The ultrasound examination revealed an image, suggesting a left ectopic pregnancy measuring 3cm with Doppler embryo heart activity. However, a pseudo-pattern of asymmetrical bicornuate uterus was noted as well as a gestational sac surrounded by myometrial tissue and completely separated from the uterine cervix. No communication with the contralateral uterine cavity could be found. These arguments suggested a rudimentary noncommunicating horn pregnancy. The β-human chorionic gonadotropin titer was 10,091 µUI/mL. Even though the woman was free of pain, a surgical procedure was proposed because of the expected risk of rudimentary uterus rupture.
The laparoscopy was carried out using one umbilical ancillary 12-mm trocar and 3 ancillary 5-mm trocars placed downward. The peritoneal cavity was free of any blood or liquid collection. A right unicornuate uterus with a seemingly normal right fallopian tube and a left pregnant rudimentary uterus with a normal fallopian tube were present **([Figure 1](#F1){ref-type="fig"})**. The 2 uteri were linked by a fibrous, less congestive flap. The pelvic peritoneum had several disseminated, red, congestive, nodular lesions that appeared to be endometriotic implants. No apparent signs of uterine rupture were revealed. A left hysterectomy with left salpingectomy using a classical procedure (bipolar coagulation and section by scissors of the mesosalpinx, fibrous flap and board ligament) was carried out. The procedure took 45 minutes, and no incidents were encountered. Biopsies were performed on 2 endometriotic lesions, the other being coagulated. Postoperative outcomes were favorable, and the woman was discharged 3 days later. Abdominal ultrasound examination showed that both kidneys were present.
{#F1}
Histopathologic examination revealed a rudimentary left horn uterus with a 2-cm pregnancy of 7-week\'s gestation. There was no communication with the right unicornuate uterine cavity. The chorioamniotic sac contained the embryo and the placenta infiltrating through the primitive müllerian myometrium. The placental infiltration had completely pierced the myometrium thickness and was found exteriorized on the anterior uterine side. This finding corresponded to a placenta percreta **([Figure 2](#F2){ref-type="fig"})**.
{#F2}
DISCUSSION
==========
This case report is peculiar not only because of the rarity of the pathology, but also due to the onset of circumstances and early management of the pregnancy, avoiding any hemorrhagic incident. To our knowledge, in the literature, few cases have been reported of pregnant rudimentary uterine horn where pregnancy had not led to a symptomatic uterine rupture.^[@B1],[@B2],[@B6]^ In 2 similar cases, the pregnant rudimentary uterine horn diagnosis was carried out early by MRI examination.^[@B6]^ Despite the high accuracy of MRI, we considered that it was not necessary in this case because of both the strong ultrasound presumption of pregnant rudimentary uterus and the decision to carry out a laparoscopic procedure anyway. Otherwise, we agree that MRI is useful in preoperative assessment when the ultrasonography is not able to affirm a rudimentary uterine horn pregnancy and to rule out an ectopic pregnancy.
Early diagnosis allows safe laparoscopic removal of the pregnant rudimentary uterus with rapid favorable outcomes. In the literature, the laparoscopic procedure often had not been done in circumstances of severe hemoperitoneum due to uterine rupture with major threats to the woman\'s life.^[@B2]--[@B5],[@B7]^ Even though in the literature uterine rupture is reported to mainly occur during the second trimester of pregnancy, we consider that no safe bet can be made on the pregnant rudimentary uterine horn capacity to relax during the first trimester. Subsequently, we consider that a prompt surgical procedure ought to be resorted to without delay.
Peritoneal endometriosis is often associated with obstructive malformations of the genital tract,^[@B9]^ and this finding is an argument in favor of the pathogenic menstrual blood reflux theory of endometriosis. It is worth noting that despite the extended peritoneal area involved in active endometriosis and the antecedent of right salpingotomy, pregnancy could occur in the left rudimentary uterine horn. It appears that 2 mechanisms may be involved in the occurrence of this pregnancy.^[@B4],[@B8]^ The first supposes that spermatozoons go up to the peritoneum by the right permeable fallopian tube, transmigrate intraperitoneally and fecundate the ovule that had been released either by the left or right ovary. Nahum et al^[@B8]^ showed that intraperitoneal sperm and ovum transmigration appears to occur respectively in 50% and 40% of all cases of human pregnancy. However, several authors^[@B10]^ suggest that peritoneal endometriosis would be harmful to spermatozoic mobility and survival. An alternative hypothesis is that fertilization might have occurred within the peritoneal cavity with subsequent intraperitoneal transmigration of the resulting fertilized ovum and contralateral tubal pick up.^[@B8]^ Whatever the mechanism, the onset of pregnancy despite these unfavorable circumstances may be considered worthy of speculation.
The placenta accreta is seemingly frequently found in rudimentary uterine horn pregnancies. On the strength of cases previously reported in the literature, Oral et al^[@B3]^ estimated that the prevalence of placenta accreta in rudimentary uterine horn pregnancies may be greater than 10%. The hemorrhagic risk due to placenta accreta and that of spontaneous uterine rupture due to the thinness of the myometrium represent in our opinion 2 sufficient arguments to recommend the immediate surgical removal of a pregnant rudimentary uterine horn as soon as its diagnosis is carried out. This attitude allows a safe laparoscopic procedure with rapid favorable outcomes and avoids hemorrhagic injuries requiring emergency surgical procedures by laparotomy.
On the basis of the high rate of spontaneous sperm transmigration across the peritoneal cavity, Nahum et al^[@B8]^ had recently suggested that salpingectomy should be preferable to the salpingotomy in women with ectopic pregnancy and unilaterally damaged fallopian tube to avoid recurrences. Similarly, on the basis of the major risk related to the pregnancies located in a rudimentary uterine horn, we ought to go forward and suggest laparoscopic removal of rudimentary uterine horns when they are revealed in young women with a desire for further pregnancies. In our opinion, the option of a planned laparoscopic procedure is always less morbid than whatever surgical procedure is carried out in pregnancy circumstances.
| {
"pile_set_name": "PubMed Central"
} |
Introduction {#S1}
============
Disulfide bonds between pairs of cysteines contribute importantly to the folding and stability of many proteins. In bacteria, this advantageous modification is generally limited to proteins that are exported to the cell envelope or beyond. Many proteins involved in bacterial virulence (such as toxins, adhesins, flagella, fimbriae, pili, and type II and III secretion systems) require disulfide bonds for their stability and activity^[@R1]^. Thus, inactivation of enzymes that make protein disulfide bonds interferes with the activity of multiple bacterial virulence factors. Inhibitors of these enzymes could have profound effects on pathogen virulence.
In Gram-negative bacteria, disulfide bonds are introduced into substrate proteins as they cross through the cytoplasmic membrane into the cell envelope^[@R2],[@R3]^. The periplasmic enzyme DsbA, a member of the thioredoxin family, oxidizes pairs of cysteines in substrate proteins through its Cys-X-X-Cys active site^[@R4]^. The resulting reduced DsbA is re-oxidized by the membrane protein DsbB, regenerating DsbA's activity. DsbB itself is reoxidized by membrane-imbedded quinones, from which electrons are transferred to the electron transport chain ([Figure 1](#F1){ref-type="fig"}).
While the DsbB/DsbA system is widespread in bacteria, some bacteria, e.g. the Actinobacteria and Cyanobacteria, use a somewhat different pathway^[@R5]^. This alternate pathway retains a DsbA-like protein, but uses the membrane protein VKOR instead of DsbB to oxidize DsbA^[@R6]-[@R8]^. Bacterial VKOR is a homologue of the vertebrate protein, vitamin K epoxide reductase, an endoplasmic reticular enzyme involved in blood coagulation and the target of the anticoagulant warfarin (Coumadin©). While bacterial VKORs show no homology to DsbB, like DsbB they encode two extra-cytoplasmic soluble domains containing essential pairs of cysteines, one of which is a Cys-X-X-Cys motif. The interactions between the redox-active cysteines of DsbA and VKOR proceed by the same steps seen between DsbA and DsbB^[@R7]^. Therefore, a *vkor* gene from *Mycobacterium tuberculosis* (*Mtb*VKOR) expressed in *E. coli* complements a *dsbB* null mutant; while *E. coli* DsbB (*Ec*DsbB) complements a *vkor* null mutant of *Mycobacterium smegmatis*^[@R9]^.
Although some bacterial VKORs, like their eukaryotic homologues, are sensitive to coumarin-based anticoagulants, *Ec*DsbB is not^[@R6],[@R9]^. Furthermore, VKOR is essential for Mycobacterial growth^[@R9],[@R10]^, while neither DsbB nor DsbA is essential for aerobic *E. coli* growth.
The fact that *Ec*DsbB and *Mtb*VKOR are functionally homologous but do not share amino acid homology suggested a methodology for identifying specific inhibitors of either enzyme. This methodology is greatly facilitated by the ability to detect inhibitory effects on disulfide bond formation in growing *E. coli* cells. The methodology is further enhanced by a sensitive assay for disulfide bond formation in *E. coli*, provided by a version of the enzyme β-galactosidase that is exported to the *E. coli* periplasm where it is inactivated by the introduction of non-native disulfide bonds^[@R11]^. This disulfide-sensitive β-galactosidase (β-Gal^dbs^ ) is the product of a hybrid gene encoding a β-galactosidase fused to a periplasmic domain of the membrane protein MalF^[@R4],[@R12]^. In *E. coli* cells with an intact disulfide bond pathway, the activity of β-Gal^dbs^ is two to three orders of magnitude lower than when disulfide bond-forming enzymes are absent. Thus, wild-type cells expressing the β-Gal^dbs^ form white colonies on agar media that contain the chromogenic β-galactosidase indicator, X-Gal (5-bromo-4-chloro-3-indolyl-β-D-galactopyranoside) while *dsb* mutant colonies expressing the same β-Gal^dbs^ are blue. Importantly for limiting the array of targets in this screen, our previous genetic studies revealed that only null mutations in *dsbA* or *dsbB* restore high levels of β-galactosidase activity. Much weaker restoration of β-galactosidase activity results from certain non-null mutations of the *dsbA* or *dsbB* genes or in genes encoding proteins required for cytoplasmic membrane protein assembly^[@R12],[@R13]^. These latter mutations restore only \~1% of the β-galactosidase activity, presumably because strong mutations in these genes are lethal.
Employing this highly sensitive assay of disulfide bond activity, we carried out a High Throughput Screen (HTS) to identify compounds potentially useful in the development of antibiotics. The rationale follows: 1)Disulfide-bonded proteins are important for bacterial virulence; 2)Detecting inhibition of *Mtb*VKOR or *Ec*DsbB can be achieved in growing *E. coli* cells; 3)Detecting high levels of β-Gal^dbs^ activity requires strong inhibition of either *Mtb*VKOR or *Ec*DsbB; 4) Strong inhibitors of other pathways that restore β-galactosidase activity are not likely to be detected; 5)Screening compounds in parallel for inhibition of the non-homologous *Mtb*VKOR and *Ec*DsbB enzymes provides reciprocal controls, allowing us to tentatively eliminate inhibitors that are influencing β-galactosidase activity by affecting membrane protein assembly or acting directly on DsbA. Such inhibitors would show up as hits in the screen for both strains while specific inhibitors would only register as hits against one strain or the other.
Here, we report the results of a screen of 51,487 compounds that yielded six specific inhibitors of *Ec*DsbB, but none of *Mtb*VKOR. Chemical modifications led to more potent inhibitors that directly inhibit purified *Ec*DsbB activity and inhibit anaerobic growth of *E. coli*. Furthermore, these small molecules inhibit DsbBs of other gram-negative pathogenic bacteria.
Results {#S2}
=======
Target- and cell-based screen using agar-multiwell plates {#S3}
---------------------------------------------------------
We screened inhibitors of disulfide bond formation in two *E. coli* strains in parallel. One expresses chromosomally encoded *Ec*DsbB and the other is deleted for the *dsbB* gene but complemented by the *Mtbvkor* gene expressed from an IPTG-inducible promoter. Both strains express the β-Gal^dbs^ from the chromosome. In these strains, strong inhibition of disulfide bond formation should lead to a substantial increase in β-galactosidase activity^[@R4],[@R12]^. Since the enzyme directly responsible for disulfide bond formation in both strains is DsbA, inhibitors of DsbA or of other processes enhancing the activity of β-galactosidase would raise the activity in both *Mtb*VKOR- and *Ec*DsbB-based strains. In contrast, compounds that specifically inhibit *Ec*DsbB would increase β-galactosidase activity in the *Ec*DsbB-dependent strain but not in the *Mtb*VKOR-dependent strain and vice versa. Assaying the effects of compounds on the two strains in parallel allowed us to pick candidate inhibitors that were specific to either *Ec*DsbB or *Mtb*VKOR. Therefore, we eliminate from further consideration compounds giving positive signals with both strains, even though these might include compounds that inhibit both enzymes directly.
Initially, we compared screens with 384-well plates in which wells contained liquid or agar-based growth media. It appeared that aeration interfered with growth in the liquid media leading us to choose the agar media where growth occurred on the media surface. When these mini-agar wells included X-Gal, high levels of β-galactosidase activity in the presence of an inhibitor were readily seen in the blue color resulting from X-Gal hydrolysis ([Supplementary Results, Supplementary Figure 1](#SD1){ref-type="supplementary-material"}). Using such agar-filled wells required special procedures that quickly became routine (Online Methods).
We observed that while the *Ec*DsbB strain exhibited no visible trace of blue in the wells, the *Mtb*VKOR-complemented strain did show a pale blue color, indicating that the *Mtb*VKOR did not completely restore levels of disulfide bond formation^[@R9]^. This difference did not interfere with the principles of the screen, but did result in a greater number of false positives in the *Mtb*VKOR screen.
The HTS was carried out with 50,374 compounds from the collection of Harvard University's Institute for Chemistry and Chemical Biology (ICCB) and 1,113 compounds from the National Institute of Allergy and Infectious Diseases (NIAID) collection of inhibitors of *M. tuberculosis* H37Rv growth^[@R14]^. Each compound was assayed in duplicate for each strain. Only a small number of *Ec*DsbB hits were observed, seen easily by the clear blue/white difference. Because of the weaker expression of *Mtb*VKOR, real potential hits with the *Mtb*VKOR-based strain were harder to distinguish from the light blue background color. We initially set a very low threshold for assigning hits (i.e. anything bluer than the background) and identified 11 (hit rate of 0.02%) and 150 (hit rate of 0.29%) potential inhibitors of *Ec*DsbB and *Mtb*VKOR, respectively. The much higher rate in the case of *Mtb*VKOR is explained by the lower effectiveness in its oxidation of DsbA mentioned above and in the Discussion.
To verify as potential inhibitors the compounds identified in the screen, we retested them in the same assay. Failure to retest positively reduced the numbers of inhibitors of *Ec*DsbB and *Mtb*VKOR to 8 and 62, respectively. Other candidate compounds were discarded after careful retesting showed inhibition of both *Ec*DsbB and *Mtb*VKOR. Evaluation of the possible chemical reactivity and stability of the hits helped us further reduce our sets of candidates, leaving us with four potential inhibitors of *Mtb*VKOR (\#1-4) and seven of *Ec*DsbB (\#5-11).
Testing for candidate inhibitors of *Mtb*VKOR {#S4}
---------------------------------------------
We examined the potential *Mtb*VKOR inhibitors to determine whether any might be lead compounds for antibiotic development against tuberculosis since *Mtb*VKOR is essential for *M. tuberculosis* growth. We tested these compounds (\#1-4) and the potential *Ec*DsbB inhibitors (\#5-11) for their effects on *M. tuberculosis* growth in three different media ([Supplementary Results, Supplementary Table 1](#SD1){ref-type="supplementary-material"}). Compound 4, obtained from the NIAID collection, strongly inhibited *M. tuberculosis* growth, but was also a weak inhibitor of *Ec*DsbB, eliminating it from further study. Compounds 2 and 3 were eliminated since they did not affect *M. tuberculosis* growth. The *Ec*DsbB inhibitors (\#5-10) showed little to no inhibition of *M.tuberculosis* growth as expected.
For a potential *Mtb*VKOR inhibitor to be a useful antibiotic, it should not have anticoagulant activity (as does warfarin). To test for anticoagulant activity, we assessed the activity of compound 1 in inhibiting the VKOR activity present in mouse liver ([Supplementary Results, Supplementary Figure 2a](#SD1){ref-type="supplementary-material"}). We also assessed the effect of compound 1 on insulin reduction by human Protein Disulfide Isomerase (PDI) which is a thiol-redox active enzyme ([Supplementary Results, Supplementary Figure 2b](#SD1){ref-type="supplementary-material"}). While compound 1 inhibits both enzymes, the strong inhibitory effect of compound 1 on the off-target PDI eliminates it as a specific *Mtb*VKOR inhibitor.
These tests eliminated all potential inhibitors of *Mtb*VKOR from further study. However, in a screen with bioactive compounds, one specific hit against *Mtb*VKOR was brominedione, a known anticoagulant inhibitor of human VKORc1. This positive result indicates that the screen should identify new classes of anticoagulants as well as specific *M. tuberculosis* inhibitors.
Sorting out potential *Ec*DsbB inhibitors {#S5}
-----------------------------------------
Of the seven inhibitors in the *Ec*DsbB screen, compound 11 was eliminated because it showed inhibitory activity in the *Mtb*VKOR re-test and caused growth defects with *M. tuberculosis* ([Supplementary Results, Supplementary Table 1](#SD1){ref-type="supplementary-material"}). Hence, the false discovery rate is 45% (5/11) for *Ec*DsbB screen. The remaining 6 potential *Ec*DsbB inhibitors (\#5-10) did not inhibit the *Mtb*VKOR-dependent strain, making it likely that they directly inhibit *Ec*DsbB. These compounds shared a pyridazinone ring structure ([Supplementary Results, Supplementary Table 1](#SD1){ref-type="supplementary-material"}).
Inhibition of purified *Ec*DsbB {#S6}
-------------------------------
We tested compounds 5-10 for their inhibitory effect on *Ec*DsbB-mediated ubiquinone reduction using purified enzymes. In this assay, reduced DsbA provides the source of electrons that are used by *Ec*DsbB to reduce ubiquinone-5. The compounds show dose-dependent inhibition of *Ec*DsbB with Inhibitory Concentrations 50 (IC50s) in the low μM range ([Supplementary Results, Supplementary Table 2](#SD1){ref-type="supplementary-material"}). The lowest IC50, 1.7 μM, was exhibited by compound 9 ([Figure 2a](#F2){ref-type="fig"}). Enzyme kinetics analysis of compound 9 revealed a K~I~ of 46 ± 20 nM (The K~m~ for ubiquinone-5 is 1.03 ± 0.12 μM, [Supplementary Results, Supplementary Figure 3](#SD1){ref-type="supplementary-material"}).
Inhibition of disulfide bond formation *in vivo* {#S7}
------------------------------------------------
We recognized that the agar-based 384-well plates did not allow calculation of real concentrations of the compounds in the wells since aliquots of the compounds are dropped onto solidified agar and might concentrate on top of the agar and/or diffuse slowly into the agar. Therefore, we attempted to determine in aerated liquid media the minimal inhibitory concentration (MIC) of the compounds necessary to restore detectable β-galactosidase activity. However, we did not observe any increase in β-galactosidase activity during growth in liquid media with the strongest of the inhibitors, compound 9, at concentrations up to between 50-100 μM where the compound begins to precipitate. In the HTS agar assay, the "calculated" (see above) but likely underestimated MIC for compound 9 required to show any blue color in the agar assay was 5.7 μM. Thus, the HTS agar assay is a more sensitive but still specific detector of inhibitors than an HTS liquid assay. If we had used a liquid assay in the HTS, we would never have detected these compounds as *Ec*DsbB inhibitors.
To directly determine compound concentrations inhibitory of *Ec*DsbB in liquid media, we assayed the oxidation of its substrate (DsbA). A standard alkylation assay that distinguishes oxidized (disulfide-bonded) from reduced (free cysteines) DsbA, showed that the IC50 of compound 9, at which oxidation of DsbA was lowered by 50%, is 8.5 ± 0.6 μM ([Figure 2b](#F2){ref-type="fig"}). This disparity between observable effects on β-galactosidase activity and effects on DsbA oxidation in liquid media is probably due to the high sensitivity of β-galactosidase to disulfide bond formation resulting from the presence of 16 cysteines in the protein; oxidation of any pair of these to disulfide bonds is likely to cause misfolding and lack of activity of the enzyme^[@R15]^. Furthermore, the aeration of the liquid cultures could increase the air oxidation of DsbA, increasing the likelihood that β-galactosidase would be inactivated.
Nevertheless, while the agar assay itself cannot be used to obtain an accurate measure of MICs or IC50s for individual compounds, it allows us both to detect inhibitors that would not be found in a liquid assay and to rank a collection of compounds for their inhibitory potency. For instance, from dilutions of compounds and assaying effects in the agar assay, we estimate that compound 10, the weakest found in the HTS, is 6-fold less inhibitory than compound 9. Since compound 9 fully inhibits DsbB at 20-50 μM, our HTS is detecting compounds (e.g. 10) that fully inhibit DsbB only at high micromolar concentrations. This result means that the nature of the HTS is such that we would probably not detect compounds that inhibit fully only in the high micromolar or millimolar range.
In experiments that follow which estimate MICs from the agar-based assay for calculating inhibitor concentrations, we will refer to these presumably overestimated concentrations as "agar-based calculated" MICs.
Approach to obtain more effective *Ec*DsbB inhibitors {#S8}
-----------------------------------------------------
The common structural features of the six *Ec*DsbB inhibitors obtained led us to seek available compounds with similar structures amongst which we might find stronger inhibitors. We first asked whether any of the 51,487 compounds screened that were [not]{.ul} inhibitors of *Ec*DsbB had pyridazinone core structures. We found 46 such compounds, all but four of which failed to inhibit DsbB in the agar assay. These four were weaker inhibitors than any of those detected in the HTS and their inhibitory activity could only be detected by assays in a strain expressing lower levels of *Ec*DsbB than wild-type (Online Methods). We also examined the list of the remaining 549,283 ICCB-compounds from which we found only five with related structures, three of which inhibited *Ec*DsbB in our assay.
From a survey of compounds that were not inhibitors and from the commonalities found among the effective inhibitors, we selected 24 additional compounds that were commercially available and more similar to the active members of the pyridazinone family. Amongst the criteria we used to choose compounds to order were 1) variation in the groups linked to the pyridazinone; 2) introducing halogen or other groups in the benzyl ring; 3) substitution of the benzyl ring of the pyridazinone by phenyl. Amongst these, 6 compounds exhibited *Ec*DsbB inhibitory activity as well as or better than compound 9 ([Table 1](#T1){ref-type="table"}).
Mechanism of *Ec*DsbB inhibition by compound 12 {#S9}
-----------------------------------------------
Among the compounds discussed above, we found two *Ec*DsbB inhibitors (13 and 12), that showed 10- and 23-fold more inhibitory activity than \#9, respectively ([Table 1](#T1){ref-type="table"}). Compound 12 has a Ki of 0.8 ± 0.1 nM (IC50 of 18.85 nM) in the *in vitro* assay ([Figure 2a](#F2){ref-type="fig"} and [Supplementary Results, Supplementary Figure 3](#SD1){ref-type="supplementary-material"}) and an IC50 of 0.9 ± 0.5 μM in inhibiting DsbA oxidation in aerobically growing cells ([Figure 2b](#F2){ref-type="fig"}). Additionally, when the *in vivo* redox state of the cysteines of *Ec*DsbB is probed with the cysteine alkylating agent MaleimidePEG-2k (ME2k) after treatment of cells with compound 12, *Ec*DsbB shows only one ME2k modification ([Supplementary Results, Supplementary Figure 4a](#SD1){ref-type="supplementary-material"}). This indicates that of the four essential *Ec*DsbB cysteines (the two non-essential cysteines are mutated to alanine and valine, respectively), two are in the disulfide state, one in the reduced state (labeled by ME2k) and the forth is unavailable to react with ME2k.
Further, we noticed that the slightly pinkish color of purified DsbB turns yellow when treated with compound 12. The pink color represents a small population of DsbB that is in the charge-transfer complex state with ubiquinone, which absorbs strongly at 500 nm^[@R15]-[@R17]^. When DsbA~C33A~ is used as a substrate for DsbB, it forms a stable mixed disulfide complex in which Cys44 is trapped in the charge-transfer complex state. Likewise, when the compound is added to the DsbB-DsbA~C33A~ dimer, its characteristic color turns yellow ([Supplementary Results, Supplementary Figure 4b](#SD1){ref-type="supplementary-material"}). This is not due to the dissociation of the dimer since non-reducing SDS-PAGE shows the complex intact. Moreover, when compound 12 is added to DsbB before the addition of DsbA~C33A~, the pink color quickly develops but does not persist. These results suggested that the compound influences the interaction between Cys44 of DsbB and ubiquinone.
To determine whether the interaction between compound 12 and the DsbB-DsbA~C33A~ is due to a covalent bond, we performed ion-trap mass spectrometry which revealed an adduct of 253.6 Da with the DsbB-DsbA~C33A~ dimer ([Supplementary Results, Supplementary Figure 4c](#SD1){ref-type="supplementary-material"}). Since the theoretical Mw of compound 12 is 289.54 Da, the 35.9 Da mass loss is possibly representing a leaving chloride ion. This mass adduct was not observed when treating either DsbB (oxidized), DsbA (reduced) or DsbA~C33A~ alone with the compound ([Supplementary Results, Supplementary Figure 4c](#SD1){ref-type="supplementary-material"}). These data indicate a covalent modification of DsbB by compound 12 which is occurring after the formation of the charge-transfer complex during ubiquinone reduction. Consistent with this expectation, high resolution tandem mass spectrometry of chymotrypsin digested DsbB-DsbA~C33A~ complex treated with the compound shows that Cys44 of DsbB has an adduct of 252.995 Da ([Supplementary Results, Supplementary Figure 4d](#SD1){ref-type="supplementary-material"}).
To test for off-target reactivity of compound 12 towards other thiol-containing redox enzymes, we assessed its inhibitory effect on the reduction of insulin by PDI. Compound12 did not show significant inhibitory effect on PDI at concentrations up to 100 µM ([Supplementary Results, Supplementary Figure 2b](#SD1){ref-type="supplementary-material"}) in contrast to compound 1. The possibility that compound 12 might form a covalent adduct with DsbA sufficient to inhibit disulfide bond formation is ruled out by the finding that this compound does not inhibit disulfide bond formation in the *Mtb*VKOR-dependent *E. coli* strain in which DsbA is functional in disulfide bond formation. Moreover, electrospray-ionization mass spectrometry of reduced DsbA shows no difference in mass upon addition of the compound ([Supplementary Results, Supplementary Figure 4c](#SD1){ref-type="supplementary-material"}).
Other properties of DsbB inhibitors {#S10}
-----------------------------------
Since *E. coli* mutants lacking DsbA or DsbB do not grow anaerobically, we tested the effect of compound 12 on anaerobic *E. coli* growth in minimal medium and found that it fully inhibits anaerobic growth of a wild-type strain at a concentration of 10 μM ([Figure 2c](#F2){ref-type="fig"}).
We tested all pyridazinone inhibitors for their effects on *Mtb*VKOR expressed in the *E. coli dsbB* null strain using the agar-filled well plates. We observed no inhibition of *Mtb*VKOR at "agar-based calculated" MICs up to at least 100 μM. An additional method for demonstrating inhibitor specificity for *Ec*DsbB is made possible by our finding that a *vkor*-deletion strain of *M. smegmatis* grows very poorly but that growth is restored when the strain is complemented by a plasmid expressing either *Mtb*VKOR or *Ec*DsbB^[@R9]^. Therefore, we added compound 12 to test the growth of *Ec*DsbB-dependent and *Mtb*VKOR-dependent *M. smegmatis*. Compound 12 completely inhibited growth of the *Ec*DsbB-dependent *M. smegmatis* strain at 400 nM, while the VKOR-dependent strain grew normally at concentrations up to and including 100 μM. Although neither the relative expression levels of *Ec*DsbB and *Mtb*VKOR in these two strains nor their relative efficiencies in oxidizing the *Msmeg*DsbA can be assessed, these vast differences in sensitivity to compound 12 are consistent with the specificity of the compound found in the *E. coli* studies.
Inhibition of DsbBs from other gram-negative pathogens {#S11}
------------------------------------------------------
We further tested the *Ec*DsbB inhibitors for their ability to inhibit DsbBs from other gram-negative pathogens when expressed in *E. coli*. We cloned the *dsbB* genes from *Acinetobacter baumanni*, *Klebsiella pneumonia*, *Vibrio cholerae*, *Hemophilus influenza*, *Francisella tularensis*, two *dsbB* homologues from *Pseudomonas aeruginosa* (*dsbB* and *dsbH*), and two *dsbB* homologues from *Salmonella typhimurium (dsbB* and *dsbI*) under an IPTG-inducible promoter. All DsbB homologues complemented the *dsbB* null strain in maintaining the β-Gal^dbs^ in the disulfide-bonded state as indicated by the absence of blue color in agar growth assay (Online Methods). We then tested the complemented strains for their sensitivity to the collection of *Ec*DsbB inhibitors and to the related non-inhibitory compounds. Several dilutions of these inhibitors were dropped onto the agar media in 384-well plates with the complemented strains, thus allowing us to rank the different inhibitors in terms of their ability to inhibit each DsbB. For the most part, the seven compounds that did not inhibit *Ec*DsbB also did not inhibit the DsbBs of the other gram-negative bacteria, while at least one of those that inhibited *Ec*DsbB also inhibited the other DsbBs. Interestingly, while the most effective inhibitor of several of the organisms was \#12, other DsbBs were more effectively inhibited by other compounds of this group in the agar assay ([Figure 3](#F3){ref-type="fig"}). The only DsbB homologue that was not inhibited by any of these compounds, *St*DsbI, is an unusual homologue that appears to be involved in a specialized pathway of disulfide bond formation^[@R18],[@R19]^. Although we can rank the inhibitors in terms of strong vs weak, it is not possible to compare effectiveness of their action on different DsbBs without knowing for each DsbB the expression level and effectiveness of oxidizing *Ec*DsbA.
In order to begin determining whether those compounds that inhibited other gram-negative DsbBs would act to inhibit disulfide bond formation in the organism from which they were cloned, we tested the effect of compound 12 on the property of twitching motility in *P. aeruginosa*. Twitching motility allows certain pathogens to attach to the cell surface and colonize the host. *P. aeruginosa* motility depends on Type IV pili and on the presence of a disulfide loop at the C-terminus of PilA^[@R20]^. A *dsbA* mutant of *P. aeruginosa* shows decreased virulence and is defective in twitching motility^[@R21],[@R22]^. Similarly, the double mutant *dsbB,dsbH* of *P. aeruginosa* is defective in twitching motility^[@R23]^. Therefore, we tested *P. aeruginosa* motility in the presence of inhibitor and observed that the wild-type strain in the presence of 50 μM of compound 12 is completely defective in motility ([Supplementary Results, Supplementary Figure 5](#SD1){ref-type="supplementary-material"}). Growth itself was not affected as indicated by colony size on agar media. While the phenotype is consistent with an inhibition of disulfide bond formation, we cannot rule out the possibility that the compound inhibits a different process affecting motility.
Discussion {#S12}
==========
The HTS procedure described here allows detection of compounds that inhibit the DsbB protein of the gram-negative bacterium *E. coli* and presumably in further screenings can do the same for the VKOR of *M. tuberculosis*. Since inhibitors of these enzymes cause reduced activity of virulence factors, defects in anaerobic growth and in the integrity of the bacterial outer membrane, they may be models for a new class of antibiotics or antivirulents.
The novel aspects of our approach may be useful in other types of screens. First, employing agar media-filled wells in which compounds are dropped in the well may concentrate the compounds at the already solidified agar surface, making the assay much more sensitive than a liquid assay would have been. Inhibitors were readily detected in the strains carrying β-Gal^dbs^ as causing blue color in the wells. For a cellular process for which β-galactosidase can be used as a reporter for effects on components of the process, the agar well assay could provide a more sensitive approach for detecting inhibitors.
Second, assessing potential inhibition of two proteins in parallel (*Ec*DsbB and *Mtb*VKOR) that perform the same function, but are not homologues, allows detection of inhibitors that are specific to one or the other protein, making it a target-based *in vivo* screen. This approach may be extended to pathways with similar advantages or may be made possible by assaying a native and an alternative laboratory-generated pathway in parallel screens. For example, distinct pathways for disulfide bond formation in either the *E. coli* cytoplasm or cell envelope have been evolved in the laboratory, presenting possible screens like ours that focus on components of those pathways^[@R24],[@R25]^.
Third, the screening assay in *E. coli* which depends on the target protein being an enzyme that can regenerate active DsbA, has allowed screens in which the native DsbB is replaced by DsbBs from numerous gram-negative bacteria or with the enzyme VKOR from *M. tuberculosis*. With such complemented strains one can test existing and developed inhibitors of *Ec*DsbB for their effects on these DsbBs and VKORs or proceed to HTS in *E. coli* with them. Positive results with inhibitors in *E. coli* could then be followed by tests for their effects on the organisms from which the DsbB or VKOR was cloned. Moreover, if human VKOR can be expressed in *E. coli* as a functional replacement for *Ec*DsbB, such a strain could be used to screen for new classes of potential blood anticoagulants.
Since the completion of this first screening, we have constructed a more effective *Mtb*VKOR-complemented strain in which DsbA is efficiently oxidized. We expect that planned further screens for *Mtb*VKOR inhibitors in this improved strain will allow more ready detection of strong inhibitors of *Mtb*VKOR and not cause the detection of so many false positives.
From the *Ec*DsbB screen and subsequent medicinal chemistry analysis, we have obtained a collection of pyridazinone-like compounds that inhibit *Ec*DsbB-based but not *Mtb*VKOR-based disulfide bond formation in *E. coli*. For instance, comparing sensitivity to compound 12 in the agar-filled wells, we found no inhibition of the VKOR-based strain at concentrations of at least two orders of magnitude higher than the minimal amount of \#12 needed to give blue color with the DsbB-based strain. Since our *Mtb*VKOR strain is already weaker in terms of inactivating β-Gal^dbs^, and thus would be more sensitive to inhibitors, the difference is certainly greater than seen from this last result. Further, compound 12 inhibits *M. smegmatis* growth in a strain which we have made dependent on *Ec*DsbB for disulfide bond formation, but it does not inhibit *M. smegmatis* growth when disulfide bond formation is dependent on *Mtb*VKOR. These findings together indicate that the *Ec*DsbB inhibitors obtained are specific to *Ec*DsbB in that they do not inhibit *Mtb*VKOR despite the similar mechanisms of action of the two thiol-redox proteins.
Since disulfide bond-formation in many bacteria is important for their virulence, it might seem that DsbA would also be a useful target for antivirulents. However, DsbA belongs to the widespread thioredoxin family, while DsbB has no known homologues in eukaryotes, raising questions about DsbA as a target for potential antivirulents. Inhibitors of DsbA may also inhibit members of the thioredoxin family that in humans contribute to the reducing environment of the cytoplasm and to protein folding in the endoplasmic reticulum. Admittedly, VKOR [does]{.ul} have two homologues in humans, one (VKORc1) involved in blood coagulation and the other (VKORc1L1) suggested to be involved in redox-maintenance^[@R26],[@R27]^. Thus any inhibitors of *Mtb*VKOR detected in further HTS must be assayed for their effects on the two VKORs found in most vertebrates.
A recent report using a target immobilized NMR screening with DsbB solubilized in micelles and nanodiscs describes inhibitors of DsbB appearing from a collection of 1071 drug fragments^[@R28]^. This study allowed the identification of 93 fragments that bind to DsbB, 8 of which inhibit electron transfer to quinone either by competing with quinone binding or disturbing the affinity of DsbB and quinone. We note that the pyridazinone-like compounds found in our *in vivo* screening appear to interfere with the mode of interaction of quinone with DsbB.
Our primary intention in this screen was to develop DsbB inhibitors as anti-virulents, since inactivation of the disulfide bond formation pathway interferes with the activity of multiple virulence factors. Others have suggested that targeting the overall virulence rather than bacterial viability could provide a successful strategy for developing new antibiotics. Such compounds could result in weaker selective pressure to develop resistance by disarming the pathogen, which may then be cleared by the host immune response^[@R1],[@R29],[@R30]^. However, since compound 12 inhibits anaerobic growth of *E. coli*, the development of these compounds to effectively inhibit disulfide bond formation *in vivo* may directly prevent infections, since many Gram-negative pathogens infect tissues in low oxygen environments^[@R31],[@R32]^. In such pathogens, the inhibitors may function as antibiotics while simultaneously reducing virulence. In addition, since reducing disulfide bond formation has multiple effects on the bacterial cell envelope, these compounds may exhibit synergistic effects with other known antibiotics.
We show that compound 12 binds covalently to the DsbB-DsbA complex *in vitro* via DsbB's Cys44, apparently losing a single chlorine atom in the process. This finding is consistent with our results indicating that the compound acts by interfering with the quinone-Cys44 charge-transfer complex. Compounds that bind covalently to periplasmic substrates, such as β-lactams, have proved quite successful as antibiotics. Recent review articles have promoted the search for compounds that covalently bind to their targets as an important approach to drug development, in general^[@R33],[@R34]^.
Finally, we point out that the HTS for potential new antibiotics or anti-virulents described here would not have been possible without the many years of basic research on disulfide bond formation in bacteria.
Online Methods {#S13}
==============
Bacterial strains and growth conditions {#S14}
---------------------------------------
The strains and plasmids used in this study are listed in [Supplementary Table 3](#SD1){ref-type="supplementary-material"}. The *malF*-*lacZ*102 fusion (referred to as β-Gal^dbs^) with Kanamycin resistance derived from pNG102^[@R11]^ was integrated into the chromosome of HK295 and HK320 strains by the λInCh method^[@R35]^ to generate HK314 and HK325 strains, respectively. MER672 and DHB7658 were constructed inserting pTrc99a at the recombined λatt site by λInCh into the chromosome of HK314 and HK325 strains, respectively. In order to generate the DHB7657 strain, instead pTrc99a*Mtb*VKOR was moved to HK325 by λInCh. In order to stabilize both insertions at the λ attachment site, the *recA*^−^ mutation (BW10724, Keio collection) was moved by P1 transduction into the three strains. Strains DHB7935 and DHB7936 were constructed by introducing into the chromosome plasmids pDSW206*dsbB* and pDSW206 at the ϕ80 attachment site of HK325 as described previously^[@R36]^. *Pseudomonas aeruginosa dsbB* and *Klebsiella pneumoniae dsbB* genes were placed into the chromosome of *E. coli* due to toxicity problems when growing cells in glucose-maltose M63 minimal media. All *dsbB* homologs from other gram-negative bacteria were expressed from plasmids. In order to generate CL315 and CL320, a PCR product that extends from the *lacI* gene to the ampicillin-resistance cassette (primers Cl65 and Cl66, [Supplementary Table 4](#SD1){ref-type="supplementary-material"}) of pCL25 (*PadsbB*) and pCL24 (*KpdsbB*) plasmids were introduced into the Δ*dsbB* loci of HK320 strain using λ−Red proteins expressed from pCL58 plasmid. Then, each insertion was moved to HK325 strain by P1 transduction. The other strains expressing different *dsbB* genes were obtained by transformation of the respective plasmid into HK325. All of the *dsbB*-complemented strains were verified by their motility in 0.3% agar minimal media and adjusted in X-Gal minimal media plates to levels of IPTG that resulted in white colonies, i.e. complementing the *dsbB* mutant phenotype. Thus, the IPTG concentrations used for agar assays were: 50 μM for the *E. coli* strain expressing *PadsbH* and *AbdsbB*, 75 μM for the strain expressing *StdsbI* and 2 mM for the strain expressing *FtdsbB.* For strains expressing *KpdsbB, PadsbB*, *StdsbB, VcdsbB* and *HidsbB* genes the basal levels of expression were enough to complement so, no IPTG was required to add.
All strains were grown in NZ or in M63 broth and agar media at 30ºC when indicated. The antibiotic concentrations used were: ampicillin 25 μg/mL (chromosomal) or 100 μg/mL, kanamycin 40 μg/mL and chloramphenicol 10 μg/mL.
Agar Screening Plate Preparation {#S15}
--------------------------------
A Matrix Wellmate (Thermo Scientific) fitted with a small-bore tubing cartridge was used to dispense 50 μL of hot agar M63 medium containing 0.2% glucose and 0.9% agar, supplemented with kanamycin (40 μg/mL), ampicillin (50 μg/mL), IPTG (1 mM), and X-Gal (120 μg/mL) to 384-well tissue culture-treated plates (BD Falcon \#353961). In order to prevent agar solidification in the Wellmate tubing or inactivation of the antibiotics and X-Gal, the medium was maintained at 57 ºC in a water bath throughout the pouring process. In addition, the Wellmate tubing was pre-warmed by washing with sterile hot water immediately prior to loading the agar medium, and the plates were poured as quickly as possible. We were able to prepare up to 80 uniform 384-well screening plates at a time. After the agar solidified, the plates were stored overnight at 4 °C in which condition the blue color of the wells is enhanced.
High-Throughput Chemical Screen {#S16}
-------------------------------
Most of the compound collections were supplied by the Institute of Chemistry and Cell Biology (ICCB) at Harvard Medical School. The initial screen included 50,374 compounds from several commercial small molecule libraries (Asinex 1, ChemBridge 3, ChemDiv 4, Life Chemicals 1, and Enamine 2) and small libraries of known bioactive molecules and natural products. In addition, a 1113-compound library of *M. tuberculosis* H37Rv growth inhibitors provided by the National Institute of Allergy and Infectious Diseases^[@R14]^ (<http://urlm.co/www.taacf.org>) was assayed. Aliquots (100 nL) of library compounds typically 5 mg/mL in dimethyl sulfoxide (DMSO, Sigma), were transferred to the agar surface in each well of the screening plates by pin transfer (EPSON compound transfer robot). It was important to place the compounds on top of the medium rather than injecting them deep within the agar, where contact with bacteria would be extremely limited. The agar concentration was optimized (0.9%) to balance this requirement with the need for the medium to remain in the liquid state while pouring the plates.
Overnight *E. coli* cultures grown in minimal media were diluted to an OD~600~ of 0.05 with M63 minimal medium containing 0.2% glucose, 40 μg/mL kanamycin, 50 μg/mL ampicillin, and 1 mM IPTG. The diluted bacteria were added to each well in 10 μL aliquots with a Matrix Wellmate dispenser. Positive (a strain lacking DsbB) and negative (DMSO no inhibitor) controls were included on each plate. The plates were sealed with breathable sealing film (Axygen BF-400) and incubated for three days at 30 °C in a humidified box. To enhance the blue color, the plates were incubated for 12 to 24 h at 4°C ([Suplementary Table 5](#SD1){ref-type="supplementary-material"}).
The compounds identified as inhibitors in the first round of screening were retested in a cherry-pick assay. The experiment was identical to the initial screen, except that the compounds were added to aliquots of bacteria with PocketTips (Thermo Scientific) rather than directly to the plates, and then the bacteria-compound mixtures were transferred to 384-well agar plates.
Compound Resupply for Retesting {#S17}
-------------------------------
Compounds 1, 2, 4, 5, 6, 8, 9, 10, 14, 21, 28 and 30 were purchased from ChemBridge (San Diego, CA); 11 from Asinex Ltd. (Moscow, Russia); 3 from Sequoia Research Products Ltd. (Pangbourne, UK); 7 from Key Organics Ltd. (Camelford, UK); 15 and 29 from AK Scientific (Union City, CA); 12, 16, 23, 27, and 31 from Ambinter (Orleans, France); 20 from Ryan Scientific (Mt. Pleasant, SC); and compounds 13, 17, 18, 19, 22, 24, 25, 26, 32, 33, 34 and 35 from Enamine (Ukraine). Larger quantities of compound 12 were obtained by custom synthesis from Enamine (Ukraine, purity over 95%, [Supplementary Information](#SD1){ref-type="supplementary-material"}). All purchased compounds were analyzed by mass spectrometry (LCMS) to verify their molecular weights and to confirm their purity (over 90%).
*M. tuberculosis* H37Rv growth inhibition {#S18}
-----------------------------------------
The bacteria were grown to stationary phase (OD~600~ of 2.0) and diluted to OD~600~ of 0.003. Chemical compounds were dissolved in growth medium and subsequent serial two-fold dilutions, 0.12 μg/mL being the lowest concentration, were performed in 96-well plates. A bacterial inoculum of 50 μL was added to each well, yielding a final volume of 100 μL/well. This was done in triplicate and repeated with bacteria grown in Middlebrook 7H9 medium (Difco), 7H9 medium supplemented with 0.2% glycerol, 5% Tween 80 (Sigma) and OADC (Becton Dickinson), and Sauton's medium (Modified from ^[@R37]^ pH 7.4 and filter sterilized instead of autoclaving). The plates were sealed with breathable film and incubated in a shaker at 37°C. After five days, 10 μL of Alamar Blue reagent (Biosource) was added to each well and the plates were incubated for 24 h at 37°C. The MIC was defined as the lowest drug concentration that prevented a color change from blue to pink.
Preparation of mouse liver microsomes {#S19}
-------------------------------------
Mouse hepatic microsomes were obtained from mouse liver by homogenization in PBS/20% glycerol/protease inhibitor cocktail (Calbiochem, 1x final concentration) using a Potter tissue grinder with an attached power unit (Con-Torque/Eberbach). Mouse liver (10 g) was homogenized with ten strokes of the tissue grinder 4 times with cooling on ice after each set of strokes. The sample was centrifuged at 10,780 × g for 10 min at 4ºC. The supernatant was collected and the remaining pellet was subjected to another cycle of homogenization as before. After two more cycles the four supernatants were pooled and subjected to centrifugation at 38,000 × g for 1 h at 4º C. The pellet was resuspended in PBS/20% glycerol/PIC/0.2% phosphatidycholine/0.5% CHAPS and sonicated twice with a Microson XL sonicator (Misonix) at power level 4 with cooling on ice after each sonication. The sample was centrifuged at 38,000 × g for 1 h at 4ºC. The supernatant from this centrifugation containing the solubilized liver microsomes was stored at −80 C.
Assay for vertebrate VKOR enzymatic activity {#S20}
--------------------------------------------
Solubilized mouse liver microsomes (20 μL) were added to 180 μL of buffer (25 mM N-\[Tris(hydroxymethyl)methyl\]-3-aminopropanesulfonic acid, pH 8.6 in 150 mM NaCl/30% glycerol). When inhibitors were used they were added at the indicated concentrations to the solubilized microsomes in buffer and the mixture incubated for 10 min at 4°C. The substrate, 4 μL of 12 mM vitamin K epoxide (VKO) in isopropanol, was added to the microsomes and 5 μL of 200 mM DTT was added to start the reaction. The reaction was incubated for 24 h at room temperature protected from light and it was stopped by adding 500 μL of a mixture of 0.05 M AgNO~3~ in isopropanol (5:9 v/v). The mixture was vortexed for 1 min and centrifuged to separate the phases. The upper organic phase (400 μL) was transferred to a brown vial and dried with a gentle stream of nitrogen. The dried sample was dissolved in acetonitrile:isopropanol:water (100:7:2 v/v) which also served as the mobile phase for HPLC. The concentration of VKO was determined by HPLC analysis on a C18 column (Vydac) and the amount of VKO converted to vitamin K calculated using a known concentration of VKO as a standard.
Insulin reductase activity assay {#S21}
--------------------------------
The activity of human Protein Disulfide Isomerase was tested using the insulin reduction assay (ProteoStat^™^, Enzo life Sciences) with minor modifications. The catalyzed reduction of insulin was measured in the presence of tris(2-carboxyethyl)phosphine (TCEP) instead of DTT in order to avoid possible reaction of DTT with compounds. The reaction mixture contained 200 µM TCEP and 330 µM insulin in phosphate/EDTA buffer (pH 7.4). PDI was used at a final concentration of 1 μM. Inhibitor compounds at the indicated concentrations were included in the reaction mixture up to 100 µM. The reaction was initiated by the addition of TCEP in 96-well plates (100 µl final volume) for 30 min at 25º C in triplicates. The uncatalyzed reduction of insulin was subtracted from the reaction containing PDI.
*In vitro* assay of DsbB inhibition {#S22}
-----------------------------------
Potential *E. coli* DsbB inhibitors were evaluated *in vitro* by a ubiquinone-5 (UbQ-5) reduction assay at 275 nm. DsbB was produced in *E.coli* using fermentation^[@R15]^ and purified as described (Purity over 90% as determined by SDS-PAGE, [Supplementary Information](#SD1){ref-type="supplementary-material"}). Purified DsbA or DsbA~C33A~ were reduced by 10 mM DTT for 30 min on ice. DTT was subsequently removed by gel filtration using a PD10 column and reduced DsbA was used as a source of electrons for DsbB-mediated reduction of UbQ-5. For determination of inhibition constants (IC50 and K~I~), various amounts of inhibitors were mixed with DsbB (2 nM) in phosphate buffer (pH 6.5) containing 0.1% DDM (Affymetrix Inc.; Santa Clara, CA, USA), 100 mM NaCl and UbQ-5 (1-50 μM). Reactions were started at room temperature by the addition of small amounts of highly concentrated DsbA solution to give a final concentration of 20 μM.
Mass spectrometry analysis {#S23}
--------------------------
DsbB, DsbA~C33A~ and DsbB-DsbA~C33A~ dimer (each at 100 μM) were mixed with compound 12 on ice at 1:2 ratio in 50 mM Tris buffer (pH:8) containing 300 mM NaCl and 0.05% DDM. Compound 12 was added by two increments of 100 μM. For DsbB-DsbA~C33A~ dimer, when absorbance at 500 nm reached a minimum upon addition of compound (around 4 minutes). Samples were directly diluted 5-fold with aqueous solution of 0.1% trifluoroacetic acid and sent for electrospray ionization mass spectrometry analysis using LTQ Velos Pro ion-trap mass spectrometer (ThermoFisher, San Jose, CA). In order to determine the site of modification, DsbB-DsbA~C33A~ dimer treated with compound 12 was reduced on ice with 50 mM DTT for 30 min, loaded on a HPLC column and eluted fractions corresponding to DsbB were digested with chymotrypsin. Chymotryptic peptides were subsequently pressure loaded onto a HPLC column. Peptides were detected, isolated, and fragmented to produce a tandem mass spectrum of specific fragment ions and analyzed using HCD LTQ-Orbitrap at 15,000 resolution (ThermoFisher, San Jose, CA).
*In vivo* chemical alkylation of *E. coli* DsbA and DsbB {#S24}
--------------------------------------------------------
The *in vivo* chemical alkylation procedure was done as described previously^[@R38]^. Precipitated proteins (reduced or not) were solubilized in 50 μL of 100 mM Tris.HCl, pH 6.8 containing 1% SDS and 5 mM 4-acetamido-4′-maleimidylstilbene-2,2′-disulfonic acid (AMS, Life technologies) or 12.5 mM maleimide-PEG2000 (ME2k, NOF Corporation, Japan). Reducing 5X-SDS sample buffer was then added and 2 μL (for DsbA detection) or 5 μL (for DsbB detection) of sample was applied to SDS-PAGE directly. Tris-HCl polyacrylamide (12%) gels were used. The proteins were transferred onto PVDF membranes and immunoblotted with α-DsbA antibody^[@R4]^ or α-DsbB antibody^[@R39]^. The IC50s of compound 9 and 12 were obtained by measuring the relative amount of the reduced and oxidized DsbA-bands by Image J (NIH). The amount of reduced-DsbA and concentration of inhibitor were used to extrapolate the 50% reduced-DsbA with trend function in Excel from at least two independent experiments.
Medicinal chemistry approach of pyridazinone inhibitors {#S25}
-------------------------------------------------------
A substructure search of compounds with a pyridazinone core was performed to detect molecules similar to compound 9 (DsbB inhibitor) among the ICCB-libraries of compounds tested and not tested in the agar screening. The obtained list of similar molecules indicated that compounds with substitutions at position 6 of the pyridazinone were discarded since it was detrimental to the inhibitory activity of the compound. Compounds that did have a single change either at position 2, 4 or 5 were selected as candidates. To determine if those candidates were commercially available, a substructure search for the pyridazinone was done using SciFinder software (American Chemical Society). 24 compounds out of the 57 commercially available candidates were purchased ("Compound resupply" section) and tested in liquid media against *E. coli* DHB7935 strain ([Table 1](#T1){ref-type="table"}), which expresses *dsbB* gene from a weaken Trc promoter^[@R40]^, making it more sensitive to weak inhibitors. This strain allowed us to easily rank the compounds. The Relative Inhibitory Concentration 50 (RIC50) for each compound was determined by quantifying the β-galactosidase activity of DHB7935 strain in the presence of different concentrations of compound. The RIC50 was defined as the concentration of compound in which the strain reaches 50% β-galactosidase activity compared to the 100% obtained in Δ*dsbB* mutant strain (DHB7936). To measure β-galactosidase activity, the velocity of hydrolysis of o-nitrophenyl- β-galactoside (ONPG, Sigma) was determined. The assay was done in a flat bottom 96-well plate (Thermo Scientific) as described previously^[@R41]^. Briefly, DHB7935 cells were inoculated to an OD~600~ of 0.01 in 200 μL of M63 with 0.2% glucose as a carbon source, 0.2% maltose to induce the expression of β-Gal^dbs^ and with serial dilutions of inhibitor. The cells were incubated for 12 hours at 30ºC, 80% humidity and 900 rpms in an orbital shaker (Multitron, ATR). 100 μL of cells were lysed using 10 μL of PopCulture reagent (Novagen) with 400 U/mL lyzosyme and incubated with 90 μL of 4 mg/mL ONPG at 28ºC in a microplate reader (VERSAmax). The OD~420~ was measured every minute during 1 hour to follow the kinetics of ONPG hydrolysis and the velocity of the reaction was calculated by SoftMax®Pro software (Molecular Devices, LLC). Miller Units were determined using 1.81 (CF1), 2.45 (CF2) and 3.05 (CF3) as constants and the relative β-galactosidase activity was calculated normalizing to the full activity obtained for the *dsbB* mutant (100%). Finally, the RIC50 was calculated by GraphPad Prism Software (La Jolla California, USA) with non-linear log dose-response normalized curve using 4 parameters. The RIC50 values and 95% confidence intervals were obtained using data of at least three independent experiments. In order to observe the difference in activity and have more meaningful idea of the inhibitor potency, all the RIC50 values were compared to the RIC50 of compound 9, the stronger compound before this analysis. So, the inhibition ratio was calculated for each dividing the RIC50 mean of compound 9 (5.10 μM) between the RIC50 mean of the compound ([Table 1](#T1){ref-type="table"}).
*E. coli* growth inhibition under anaerobic conditions {#S26}
------------------------------------------------------
To test the sensitivity of *E. coli* to inhibitors under anaerobic conditions, strains were first grown aerobically overnight in M63 glucose medium. These cultures were diluted 1:100 into fresh medium and grown aerobically to mid-log phase. All cultures were adjusted to an OD~600~ of 0.1 and transferred into a Coy anaerobic chamber. Cultures were then diluted 1:1,000 into anaerobically-prepared M63 glucose medium containing 40 mM fumarate (± compound) and grown without shaking at 37°C. Samples were taken over time, serially diluted, and plated aerobically on NZ medium for enumeration.
*M. smegmatis* growth inhibition {#S27}
--------------------------------
To determine the MIC for compound 12 against *M. smegmatis* expressing *E. coli* DsbB or *Mtb*VKOR, strains RD263 and RD265 were grown for two days in 7H9 medium containing 0.05% Tween 80. These dense cultures were diluted 1:200 into fresh medium, grown to mid-log phase, and then diluted to a starting OD~600~ of 0.05. Cultures were aliquoted into 96-well plates containing 2-fold dilutions of compound 12 and then placed in an ATR Multitron low-orbital shaker. Cells were grown at 37°C in 90% humidity and shaken at 900 rpm for 24 h. Growth was considered inhibited if the OD~600~ \<0.1 at this time point.
Testing DsbB inhibitors against other gram-negative DsbB {#S28}
--------------------------------------------------------
In order to test *E. coli* strains expressing the *dsbB* genes from other organisms, the 384-well plates were prepared in the same way as in the HTS only differing in that 0.2% maltose was included in the media to induce the expression of the β-Gal^dbs^ and different concentrations of IPTG indicated in "Bacterial strain and growth conditions" section.
Two compound plates (Corning 384-well storage plates, polypropylene round bottom) were prepared with the entire collection of compounds purchased as a result of the medicinal chemistry approach (Compound resupply section). Dilutions of the compounds were dispensed in the 384-well plate ranging from 30 mM to 0.6 μM. 100 nL aliquot of the compounds were transferred to solidified-agar plates by pin transfer (EPSON compound transfer robot) in order to have a final concentration ranging from 50 μM to 0.001 μM of compound, except compounds 27 and 23 which highest concentration started at 28.9 μM and 26.4 μM, respectively. Then, 10 μL of the bacterial cultures at 0.05 of OD~600~ were added to the agar plates with a Matrix Wellmate (Thermo Scientific). Plates were sealed with breathable sealing film (Axygen BF-400) and incubated at 30ºC for one day in humidified boxes and one day at 4ºC to analyze the results. MER672 was used as *Ec*DsbB expressing strain and CL379 was included as a positive control in each plate. The minimal inhibitory concentration (MIC) of each compound that caused the bacteria to turn light blue was registered for all of the strains expressing *dsbB* genes from pathogenic bacteria. Since the MIC is related to the expression of each DsbB in *E. coli*, we ranked the compounds from strong to weak in each DsbB-expressing strain normalizing the data for each strain. Thus, we divided the MIC observed for each compound between the lowest MIC observed for that particular strain expressing DsbB, so the most potent compounds have a ratio of 1 ([Figure 3](#F3){ref-type="fig"}). We obtained the ranking ratio of three independent experiments and the ratio average was calculated and plotted in a 3-rule color-coded table (beige as 1, olive green as 10 and black as 1500) using Excel. For the compounds that did not inhibit at the maximum concentration tested, an arbitrary ratio was assigned (1500) to have the highest value and darkest color in the table (black, non-inhibitor).
Twitching motility assays in *Pseudomonas aeruginosa* {#S29}
-----------------------------------------------------
Twitching motility of *P. aeruginosa* was assayed on a thin-layer solid surface^[@R42]^. Briefly, 100 × 15 mm plates were filled with 10 mL of M63 minimal media with 1% agarose (instead of 1.5% agar, since agarose is a more purified version). The compound was dissolved in the media before pouring and immediately after solidification one fresh colony was stabbed in the plates without and with 50 μM of compound 12. The plates were incubated for 2 days at 37ºC in humidified boxes.
Statistical Analysis {#S30}
--------------------
All results presented are reported as average ± S.D. and represent data from a minimum of three independent experiments, unless otherwise stated in figure legends. RIC50 values were calculated from a minimum of three independent experiments and their 95% confidence intervals were estimated by non-linear log dose-response normalized curve using 4 parameters (GraphPad Prism, La Jolla). Then, inhibition ratios presented in [Table 1](#T1){ref-type="table"} were calculated for each compound dividing the RIC50 mean of compound 16 between the RIC50 mean of each compound. K~I~ values were estimated from Dixon plots by taking the average values of intersection points.
Supplementary Material {#SM}
======================
We thank the ICCB-Longwood screening facility for access to their compound libraries, equipment, screening supplies and to the staff including Dr. Su Chiang, Dr. Kyungae Lee, Stewart Rudnicki and Doug Flood for helpful advice and data handling. We thank Dr. Robert Goldman of the National Institute of Allergy and Infectious disease for the collection of *M. tuberculosis*-growth inhibitors. We thank Dr. Ross Tomaino of Taplin Mass Spectrometry Core Facility at Harvard Medical School for his assistance in sample analysis. This work was supported by National Institute of General Medical Sciences grants GMO41883 (to J.B. and D.B.), Harvard Catalyst Pilot Grant Harvard \#149734 (to J.B), NIH Grant \#5-UL1RR02568-01 (to J.B), Blavatnik Biomedical Accelerator at Harvard University (to J.B), NIH Grant 3P01-A1AI074805-04S1 Subaward R01638 (to E.J.R), NIH Grant PO1-HL087203 (to B.F), NIH Grant RO1-HL092125 (to B.C.F) and NERCE-BEID Grant U54-AI057159. C.L. was partially supported by CONACYT postdoctoral fellowship. J.L.B. was supported by T32-HL07917 Grant (to B.F). B.M.M. was supported by Ruth L. Kirschstein National Research Service Award. M.E. was supported by New England BioLabs Grant. A.M. was supported by HSPH Yerby postdoctoral fellowship. J.B. is an American Cancer Society Professor.
**Competing Financial Interests Statement**
The authors declare no competing financial interests.
{#F1}
{ref-type="supplementary-material"}. (c) Inhibition of *E. coli* anaerobic growth with compound 12. Growth curve of wild-type *E. coli* (black) and *dsbB* mutant (red) under anaerobic conditions in the absence (solid lines) or presence (dotted lines) of 10 μM compound 12. Results are the average of three independent experiments ± SD.](nihms-650529-f0002){#F2}
{#F3}
######
Inhibition of *Ec*DsbB by pyridazinone compounds.
---------------------------------------------------------------
**ID Number** **Structure** **RIC50 Ratio**\
(RIC50 compound \#9 / RIC 50)
--------------- --------------- -------------------------------
12 23.03
13 10.77
14 3.80
15 2.49
16 1.65
17 1.43
**9** **1.00**
18 0.39
19 0.38
20 0.37
21 0.37
22 0.28
7 0.20
8 0.18
6 0.17
23 0.13
5 0.08
10 0.07
24 0.05
25 0.02
26 0.01
27 \<\<0.01
28 \<\<0.01
29 \<\<0.01
30 \<\<0.01
31 \<\<0.01
32 \<\<0.01
33 \<\<0.01
34 \<\<\<0.01
35 \<\<\<0.01
---------------------------------------------------------------
*In vivo* inhibition of *Ec*DsbB was measured by effects on β-galactosidase activity in liquid cultures of a strain expressing β-Gal^dbs^ and *EcdsbB* under a weak IPTG-inducible promoter. Relative Inhibitory Concentration 50 (RIC50) was calculated as the concentration of compound required to reach 50% of βgalactosidase activity compared to the full β-galactosidase activity of the *dsbB* mutant. RIC50 ratio was calculated to rank the compound potency relative to compound 9. The RIC50 average of at least three independent experiments was used to calculate the RIC50 ratios.
[^1]: Contributed equally to this work.
[^2]: (current address) FAS Center for Systems Biology, Harvard University, Cambridge, MA 02138, USA
[^3]: (current address) New England Biolabs, Ipswich, MA 01938, USA
[^4]: (current address) School of Electrical Engineering and Computer Science, Oregon State University, Corvallis, OR 97331, USA
[^5]: **Authors contributions**
M.E. developed the agar assay. J.L.B., M.E., R.D., H.A., N.K., and D.B. performed the HTS. J.L.B. performed cherry-pick tests. J.L.B. and C.L performed inhibitor retests. C.L. performed substructure analysis, *in vivo* DsbA and DsbB inhibition and other gram-negative bacteria assays. F.H. performed enzyme kinetics and *in vitro* analysis. B.M.M performed anaerobic and *M. smegmatis* growth assays. M.B., B.C.F., and B.F. performed *in vitro* mice VKOR assays. A.M., S.M., and E.J.R. performed *M. tuberculosis* growth assays. C.L., J.L.B., F.H., B.M.M., D.B. and J.B. analyzed and discussed the data. C.L. and J.B. wrote the paper.
| {
"pile_set_name": "PubMed Central"
} |
1. INTRODUCTION {#sec1-1}
===============
Worldwide, cervical cancer is the third most common cancer among women and the second most common cause of death, with 300,000 deaths per year ([@ref1]). PAP test is cheap, simple and most commonly used technique for cervical cancer and preinvasive cervical lesions screening. Atypical squamous cells (ASC) is a term that refers to inflammatory, reactive and reparative processes which are atypical and of higher level and whose quality and quantity is insufficient to be classified as cervical intraepithelial lesions (CIN). ASC was first described in Betheda classification in 1988, and revised in 2001 and 2014 when it was divided into ASCUS (atypical squamous cells of undetermined significance) and ASC-H (atypical squamous cells-high level of lesion cannot be excluded). ASC continues to be defined as general category with subcategorisation as ASCUS and ASC-H ([@ref2], [@ref3]).
Every year, over 2 million women in the USA are found with atypical squamous cells in PAP test ([@ref4]). Frequency of ASCUS lesions is between 1,6% and 9% ([@ref5]), and their frequency shouldn't exceed two to three times frequency of low grade lesions (LSIL) ([@ref6]). Metaanalysis showed that progression of ASCUS can progress into invasive cervical cancer may occur in 0,25% within 2 years, HSIL in 7,13%, while regression occurs in 68,19% ([@ref7]). Out of 1000 women with atypical squamous lesions one already has invasive cervical cancer ([@ref8]).
Aims of this study were to examine: a) Frequency and age of the examinees with ASCUS lesions; b) Frequency of HPV infection in ASCUS lesions; c) Regression, stagnation and progression of ASCUS lesion during six-month period.
2. MATERIALS AND METHODS {#sec1-2}
========================
Prospective study was conducted over a period of 3 years, from January 2012 to December 2014. In private gynecological ambulance „Dr Mahira Jahić" 1784 PAP tests were examined as routine screening in first year. Patients with ASCUS were reevaluated after local treatment with PAP test and have been monitored in period of one year. ASCUS treatment includes repeated cytology, HPV typization and colposcopy. Protocol of monitoring depended on the result of repeated PAP test. PAP test was normal in 1530 patients and they were advised to make control test once a year. ASCUS was found in 133 patients and 50 of them agreed to be monitored. Analysis of vaginal secretion and HPV typization have been done in these 50 patients and PAP test has been repeated after six months. All cytology samples were screened in the same laboratory. PAP tests were taken by conventional method, colored by method of Papanicolaou and examined with Kruss microscope. Analysis of vaginal discharge was performed from a PAP test and HPV typization-In situ by hybridization of HPV. X² test was used for statistical analysis of data.
3. RESULTS {#sec1-3}
==========
Analysis of 1784 PAP smears showed normal results and benign cellular changes in 86,6% (N-1530), and abnormal in 13% (N-254) of women. ASCUS was found in 7,4% (N-133) and ASC-H in 0,5% (N-9), LSIL in 4,4% (N-80), HSIL in 1,3% (N-24), CIN II in 1,2% (N-20), CIN III in 0,2% (N-4) ([Table 1](#T1){ref-type="table"}).
######
Results of 1784 PAP tests
Patients with ASC-H lesion 0,5% (N-9) were transferred to biopsy of cervix and PH results showed 22% (N-2) Carcinoma in situ (CIS), 33% (N-3) CIN II, 22% (N-2) CIN I and 22% (N-2) chronic cervicitis.
25% (N-10) of patients with ASCUS lesion didn't give birth and 70% (N-35) didn't have abortion. 94% (N-47) were married and 6% (N-3) weren't. 14% (N-7) of women used contraception ([Table 2](#T2){ref-type="table"}). Progression occurred in 18% ([@ref9]), persistence in 74% (37) and regression in 8% out of 50 women with ASCUS lesion after treatment and repeated PAP test ([Table 3](#T3){ref-type="table"}). There wasn't progression into high grade lesions (HSIL) in this period of time. Bacterial vaginosis has been found in 10% (N-4) of patients with ASCUS and CIN I in 33% (N-3) while patients with normal PAP results didn't have it. Patients with CIN I in 88% (N-7) were positive on HPV of high risk. Patients with persistent ASCUS result were positive in 51% (N-19). Grade of lesion was higher with the higher incidence of HPV infection ([Table 4](#T4){ref-type="table"}). The number of CIN I lesions found in women with ASCUS is bigger and statistically significant (p\<0,05) in relation to number of CIN I findings found in regular examinations of 1784 women ([Table 5](#T5){ref-type="table"}).
######
Vital characteristcs of patients with ASCUS lesion
######
Results of PAP tests of 50 women with ASCUS lesions after 6 months
######
Distribution of patients (N-50) after repeated PAP test in relation to infection
######
Difference in frequency of CIN I findings in regular screening (N-1784) in relation to repeated ASCUS findings (N-50)
4. DISCUSSION {#sec1-4}
=============
The average age of patients with ASCUS lesion was 37,6 ± 12,6 years, and frequency of 7.4% is in accordance with data cited in literature ([@ref9]), but Jahic in her research found ASCUS in 12.9% and CIN I in 5.4% ([@ref10]). Cytologic diagnosis of ASCUS is 1.7 times more often than frequency of LSIL which was 4.4% of total number of analyzed PAP tests. The category of ASCUS is by far the most commonly reported abnormal cervical cytology interpretation. Lack of specific criteria for the diagnosis of ASCUS results in differences in the incidence of these lesions in works ([@ref11]). One of the obstacles in the interpretation of cytologic finding is presence of inflammation. Miguel and ass. analyzed 50 cervical and vaginal smears in which 15 of them, at first, were interpreted as ASCUS and 34 as normal ([@ref12]). After report using Bethesda criteria for ASCUS, he found changes that associate with the presence of Candida sp., and reclassified 10 out of 16 cases as normal findings ([@ref13]). Candida was present in 24% (N-12) of patients provided that in ASCUS lesion was more often 21% (N-8) and 3 women were reevaluated as normal findings. HPV of high risk was positive in 52% (N-26) and negative in 48% (N-24) of patients. Boardman found presence of high risk HPV in 68% of 527 women with ASCUS ([@ref14]). Other authors state frequency of HPV in ASCUS lesion from 49 to 51% ([@ref15]) and 46% in CIN I ([@ref16]). Our study showed similar representation of HPV infection in 51% (N-19) of women with ASCUS. During monitoring, regression was found in 8% ([@ref4]), persistence in 74% (37) and progression to CIN I in 18% (N-9). Progressions to high grade lesions such as HSIL weren't found in this study, although some authors state it is possible in 7.13%. Frequency of HPV of high risk was bigger with higher grade of lesion. Similar incidence of 20% of CIN I was found in Barcelos' study of analysis of ASCUS and cytologic criteria. Barcelos states that it is necessary to seriously monitor patients with ASCUS lesions. In her monitoring of patients with normal/inflammatory cytology, 85.7% had normal findings after six-months monitoring and one was CIN I, so she recommended that repeated PAP test should be performed in six months time if cytologic diagnosis after ASCUS lesion is normal. Basically, changes that are ASCUS lesion are of low grade but they enable selection of women who should be examined more often because of the possible progression of lesion or possible higher grade of lesion hidden behind this diagnose. Reports say that 10-30% of LSIL and HSIL are diagnosed in women with ASCUS lesion who were monitored ([@ref17]). Rinku finds cervicitis as a result of biopsy after ASCUS in 11 out of 17 cases, but repeated PAP test after 4-8 months shows ASCUS in all patients. After monitoring in next six months 6 was ASCUS, 2 LSIL and 1 HSIL ([@ref18]). Persson warns that there will be approximately 30% of women with ASCUS and HPV positive finding for high risk group who will progress to CIN II in period of 3.8 years ([@ref19]). Castle and ass. advise that patients can get back to regular monitoring procedure if 5-year risk of CIN III is lower than 2%, if the risk is 2-10% they recommend annual monitoring and if it's \>10% than colposcopy is obligatory ([@ref20], [@ref21], [@ref22]). Risk assessment based on genotypization probably won't change clinical management always since the risk of precancerous lesions can be so high that colposcopy should be done immediately ([@ref23]).
5. CONCLUSION {#sec1-5}
=============
This study shows the importance of monitoring women with ASCUS lesion, especially HPV positive to high risk group for early detection of lesion that can progress to higher level of change such as cervical cancer. It is statistically significant frequency of CIN I changes in women with ASCUS than in regular screening. This is also the best way of selection of women who should be treated and monitored in order to prevent cervical cancer.
• Conflict of interest: none declared.
| {
"pile_set_name": "PubMed Central"
} |
Introduction
============
Humans possess robust metacognitive capacities to evaluate their performance on various tasks and make predictions about how such performance might alter in the future ([@niw018-B54]; [@niw018-B50]; [@niw018-B42]). Metacognitive evaluations are often studied by eliciting confidence judgments. For example, a student may predict their success on an upcoming exam by reflecting on their current level of knowledge and preparation (a prospective metacognitive judgment; P-metacognition). After taking the exam, the same student may then estimate his or her grade before receiving feedback (a retrospective metacognitive judgment; R-metacognition). Metacognitive capacity -- the extent to which judgments track performance -- is dissociable from first-order task performance and associated with distinct neural substrates (see [@niw018-B25]; [@niw018-B59], for reviews). However, it is unknown whether prospective and retrospective judgments draw on distinct or common resources.
Behaviorally, few studies have directly compared the accuracy of P- and R-judgments for the same task and stimulus material. [@niw018-B16] compared probability judgments made about future and past events such as "What is the probability that next (last) week IBM stock will finish the week higher than it began the week?" He found that, when the same subjects make both past and future judgments, the Brier score (a measure of probability judgment accuracy) was better for past than future judgments. However, in this case the events to be judged are external to the subject and not evaluations of self-performance. Siedlecka *et al.* (2016) compared prospective and retrospective judgments of performance while participants solved anagrams. Participants rated their confidence that a particular word was the solution, either before or after their binary response of "yes" or "no," and before or after seeing the suggested solution. Confidence ratings given after the decision were more accurate than when ratings were prospective. [@niw018-B52] showed that rhesus macaques were able to make accurate confidence judgments -- bets on performance -- both before and after responding to a delayed-match-to-sample memory task, suggesting temporal flexibility in the use of confidence responses in nonhuman primates. However in this study, first-order performance also differed between the prospective and retrospective judgment tasks, making direct comparison of metacognitive accuracies difficult.
There is evidence for neural dissociations between P- and R-metacognition ([@niw018-B17]; [@niw018-B25]). For example, [@niw018-B66] found that damage to the right ventromedial prefrontal cortex was associated with a decrease in metacognitive accuracy for judgments about future recall (feeling of knowing), but did not affect accuracy for retrospective confidence judgments. In contrast, [@niw018-B60] found that patients with lateral frontal lesions were impaired on retrospective confidence judgments, but not judgments about future task performance. Studies using functional MRI have shown that prospective metacognition activates medial PFC ([@niw018-B66], [@niw018-B67]; [@niw018-B51]), while retrospective metacognitive accuracy in a short-term memory task is correlated with rostrolateral PFC activity ([@niw018-B79]). A related line of research has demonstrated that post-decision confidence judgments also recruit rostrolateral PFC ([@niw018-B24]; [@niw018-B29], [@niw018-B26]).
Together these studies suggest that humans and nonhuman primates have the capacity to make P- and R-metacognitive judgments about the same stimulus material, and that R-metacognition is typically more accurate than P-metacognition. However, it is clear that there are conceptual and methodological differences between different types of prospective metacognitive judgment. For some prospective judgments, such as a feeling-of-knowing, a specific representation of stimulus strength is available (albeit perhaps weakly) to the system on each trial. For other types of judgment, such as predicting one's success at a sporting event, judgments must instead be made based on an aggregate likelihood of success, with little or no information pertaining to individual trials. Finally, compared to P-judgments, R-judgments are able to draw on additional trial-specific cues of response fluency, response identity and stimulus type or difficulty, potentially explaining their increased accuracy (Siedlecka *et al.* 2016). Thus, while previous studies have quantified differences in R- and P-metacognitive accuracy, the influence of different cues and their temporal dynamics (e.g. the recent history of performance and confidence on the task) on each judgment type have received less attention ([@niw018-B63]).
The dissociations between P- and R-metacognition noted above referred to metacognitive accuracy (or discrimination), the extent to which moment-to-moment variations in confidence track task performance. In contrast, bias (or calibration) reflects the tendency to be over- or underconfident ([@niw018-B27]). While metacognitive accuracy has been shown to differ between tasks ([@niw018-B64]; [@niw018-B7]; [@niw018-B49]; [@niw018-B28]; [@niw018-B1]), perhaps reflecting differences in the cues that subjects use to construct their confidence in each domain, bias may be more stable, transcending temporal focus: people have been found to have high or low confidence in their performance, irrespective of task ([@niw018-B64]; [@niw018-B58]; [@niw018-B70]; [@niw018-B1]; [@niw018-B35]). Several studies have found that subjects are overconfident in their judgments ([@niw018-B45]; [@niw018-B8]; [@niw018-B15]; [@niw018-B34]; [@niw018-B5]), but in some experiments underconfidence is found ([@niw018-B22]; [@niw018-B12]; [@niw018-B77]). In particular, while overconfidence may be the default in more difficult tasks, underconfidence may appear for easier tasks ([@niw018-B8], [@niw018-B9]), a phenomenon known as the hard--easy effect ([@niw018-B31]).
In the present study, we set out to quantify influences on prospective and retrospective judgments of self-performance. We employed the same visual discrimination task for both judgment types, thereby matching performance and task characteristics across temporal focus. We elicited numerical probabilities of success, allowing assessment of both overconfidence (bias) and accuracy of confidence ratings. Retrospective ratings were provided on every trial, whereas prospective judgments of upcoming task performance were made every five trials, before seeing the stimulus. By using repeated, similar stimuli of constant difficulty, we allowed subjects to build up knowledge of their own performance over time ([@niw018-B38]). The elicitation of subjective judgments was incentivized to ensure subjects treated both prospective and retrospective judgments with similar importance. To assess metacognitive accuracy, we calculate both the area under the type 2 ROC and measures of probability judgment calibration and discrimination ([@niw018-B27]). We hypothesised that P- and R-metacognitive judgments would draw on separate cues, such as fluency for retrospective judgments, and recent outcome history for prospective judgments, and that metacognitive accuracy and calibration would be greater for retrospective compared to prospective judgments. In contrast, based on evidence that overconfidence is pervasive and domain-general, we hypothesized that overconfidence would be similar across the two judgment types.
Methods and Materials
=====================
Participants
------------
The experiment was conducted in December 2012 at the Laboratory of Experimental Economics in Paris (LEEP) of the University of Paris 1. Subjects were recruited by standard procedure from the LEEP database and gave written informed consent to take part in the experiment. A total of 47 subjects (26 men; age 18--29 years, mean age, 22.1 years) participated in this experiment for pay. The session lasted around 90 min and subjects were paid on average €19.7. We excluded subjects from analysis due to insufficient variation (SD \< 0.02) of R-confidence (4 subjects) or P-confidence (4 subjects) for estimation of metacognitive accuracy (see below). The final sample included 39 subjects for analysis.
Stimuli
-------
The experiment was conducted using Psychophysics Toolbox version 3 ([@niw018-B13]) running in Matlab. The stimuli consisted of two circles with a variable number of dots in each circle. All dots were of the same size and the average distance between dots was kept constant. One of the two circles always contained 50 dots while the other contained 50 + *c* dots. The position of the target circle (on the left or right) was randomly chosen on each trial. Before the experiment, we estimated the value of *c* needed to obtain a success rate of 71% using a psychophysical staircase ([@niw018-B44]; see below). This dot difference (*c*) was kept constant throughout the main experiment, such that all trials were of equal objective difficulty. The position of the circle containing the greater number of dots was randomly assigned to be on the left or right on each trial.
Task and procedure
------------------
### Practice and thresholding
Subjects initially performed practice trials of the dots task without confidence ratings, in which full feedback was given. We used these trials to calibrate task difficulty. The calibration phase used a one-up two-down staircase ([@niw018-B44]): after two consecutive correct answers one dot is removed, and after one failure one dot is added. We stopped the calibration after 30 reversals in the staircase, and the value of *c* was calculated as the mean dot number across the two last reversals of the staircase. Subjects then performed 20 trials of the task with confidence ratings (20 R-confidence and 4 P-confidence ratings) with feedback both on their accuracy and on the results of the confidence elicitation mechanism.
### Experiment phase
The experimental design is summarized in [Fig. 1A](#niw018-F1){ref-type="fig"}. Each trial consisted of the following sequence. First two outline circles (diameter 5.1°) were displayed with fixation crosses at their centers at eccentricities of ± 8.9°. The subject was free to initiate the trial when they wished by pressing the "space" key on a standard computer keyboard. The dot stimuli (diameter 0.4°) then appeared at random positions inside each circle for 700 ms, and subjects were asked to respond as to whether the left or right circle contained a higher number of dots by pressing the "f" or "j" keys, respectively. There was no time limit for responding. After responding subjects were asked to indicate their level of confidence in their choice (R-confidence; 50% to 100% in steps of 10%), using the F5-F10 keys, again with no time limit on the response. On every fifth trial, we asked subjects first to give their level of confidence in getting the upcoming trial correct (P-confidence; same scale as R-confidence). No feedback was given following either choices or confidence ratings. The experimental phase consisted of 200 trials. Each subject provided 200 ratings of R-confidence and 40 ratings of P-confidence.
{#niw018-F1}
### Incentivization
Subjects were paid according to the accuracy of their stated confidence. We incentivized confidence ratings using the probability matching rule ([Fig. 1B;](#niw018-F1){ref-type="fig"} see [@niw018-B47], for details). This rule provides incentives for the subject to truthfully reveal a subjective probability of success, *p*. For each trial, a random number is drawn from 1 to 100 (*l*~1~). If *p * \> * l*~1~, the computer checks to see if the subject is correct. If the judgment is correct, an additional 1 point is won; if incorrect, 1 point is lost. If *p *\< * l*~1~, a new random number is drawn, *l*~2~. If *l*~2 ~ ≤ *l*~1~, 1 point is won; if *l*~2 ~ \> *l*~1~, 1 point is lost. The rule can be intuitively understood as follows. The higher the initial rating of *p*, the more likely the correctness of the decision will determine earnings. The lower the rating, the more likely earnings will be determined by chance (the second lottery). A particular rating value (e.g. 70%) thus reveals how subjects trade off a belief in their decision being correct against a randomly determined reward. Note that this mechanism is a proper scoring rule and provides incentives for a subject to reveal true beliefs regardless of his or her preferences. Specifically, the expected reward for this mechanism with a subjective rating *p* and a probability of success *s* is $p\, \times \,\left\lbrack {\left\lbrack {+ 1} \right\rbrack \times s + \left\lbrack {- 1} \right\rbrack \times \left\lbrack {1 - s} \right\rbrack} \right\rbrack + \left( {1 - p} \right) \times \left\lbrack {\left\lbrack {+ 1} \right\rbrack \times \frac{1 + p}{2} + \left\lbrack {- 1} \right\rbrack \times \frac{1 - p}{2}} \right\rbrack$ which is equal to $2ps - p^{2}$ and achieves its maximum for $p = s$. Prior to the experiment, we explained various possible outcomes to subjects together with their intuitive interpretation until they understood how different rating strategies impacted upon their potential earnings, how over- or underreporting confidence would lead to nonoptimal payoffs, and why it is in their financial interests to report their true beliefs. The final payment comprised €5 for participation and the accumulated points paid at the exchange rate of 1 point = €0.15.
Data Analysis
-------------
### Metacognitive bias and accuracy
We defined R-trials as those followed by a retrospective confidence rating, excluding those immediately preceded by a prospective confidence rating (160 trials per subject). The remaining trials were P-trials (40 trials per subject), which were both preceded and followed by confidence ratings. We did not analyze the retrospective rating given on P-trials (R\* in [Fig. 1A](#niw018-F1){ref-type="fig"}) to ensure that any effects on R-confidence could not be trivially explained by anchoring to the immediately preceding prospective rating given on the same trial. Global overconfidence (bias) was calculated by subtracting the mean accuracy from the average confidence level for each trial type. To estimate metacognitive accuracy (the degree to which participants can discriminate their own correct from incorrect decisions), we calculated the area under the type 2 ROC for each judgment type (AUROC2; [@niw018-B18]; [@niw018-B30]; [@niw018-B27]). We also considered that an optimal strategy for prospective judgments in a stationary environment is to assess the average rate of success, and specify this probability on every trial. Thus prospective judgments may be well calibrated on average, but fail to approach the trial-by-trial accuracy that is typically observed for retrospective judgments. To address these shortcomings of the signal detection approach, we complemented AUROC2 with a well-studied metric of forecasting accuracy, the Brier score, which assesses the squared difference between the confidence rating *c* and decision accuracy *o* (where *o* = 1 or 0 for correct or incorrect decisions): $$BS = {\sum\limits_{i}\left( {o_{i} - c_{i}} \right)^{2}}.$$
As the Brier score is an "error" score, a lower value is better. We can further decompose the Brier score into the following components ([@niw018-B53]): $$BS = O + C - D,$$ where *O* is the "outcome index" and reflects the variance in performance: $O = \overset{-}{o}\left( 1 - \overset{-}{o} \right)$; *C* is "calibration," the goodness of fit between probability assessments and the corresponding proportion of correct responses; and *D* is "discrimination" or "resolution," the variance of probability assessments. Calibration is calculated as follows: $$C = \frac{1}{N}{\sum\limits_{j = 1}^{J}{N_{j}\left( {c_{j} - {\overset{-}{o}}_{j}} \right)^{2}}}.$$ where *j* indicates each confidence-rating bin and *N* is the number of trials. Calibration quantifies the discrepancy between the mean performance level at each scale step (e.g. 60% correct) and its associated confidence level (e.g. 80%), with a lower discrepancy giving a better score. In contrast, discrimination (*D*) is a measure of the variance of probability assessments, and quantifies the extent to which correct and incorrect answers are assigned to different probability categories (equivalent to a probability judgment analog of a gamma correlation, or AUROC2). Here we used the adjusted normalized discrimination index (ANDI) suggested by [@niw018-B78], which provides a proxy for the confidence--accuracy relationship normalized by a participant's performance level and by the range of confidence ratings used. The first step in computing ANDI is to compute the normalized discrimination index, NDI: $$NDI = \frac{\frac{1}{N}{\sum\limits_{j = 1}^{J}{N_{j}\left( {\overset{-}{o}}_{j} - \overset{-}{o} \right)}^{2}}}{var\left( o \right)},$$ where *o* is a vector of success or failure (1 or 0), *J* indicates the number of confidence levels used by the subject, and *N* is the number of trials. The adjusted NDI corrects for the bias introduced by the number of judgment categories used: $$ANDI = \frac{N.NDI - J + 1}{N - J + 1}.$$
We assessed the relationship between our measures of P- and R-metacognition (bias, AUROC2, calibration, and ANDI) using Pearson's product--moment correlations. Mean values of these scores were compared across judgment type using paired *t*-tests.
### Hierarchical mixed-effects models
We examined trial-by-trial influences on R and P-confidence judgments using hierarchical mixed-effects models (using the ME package in STATA). These models allow an estimation of lagged factors with random intercepts and slopes at the individual level. We considered four candidate models of R-confidence and P-confidence.
Observed R-confidence, R(t), and P-confidence, P(t), were assumed to be related to current accuracy, O(t), and reaction time RT(t), past confidence, R(t-i) and P(t-i), and past accuracy, O(t-i). We included lagged factors modeling the influence of the previous trials. The window selected for these predictors followed the frequency of P-confidence judgments (which occurred every five trials); thus we included the previous five outcomes, the previous four R-confidence judgments and the previous P-confidence judgment. We compared the following models: $${~R\left( t \right)~or~P\left( t \right) = ~\beta}_{0} + ~\beta_{1}O\left( t \right) + ~\beta_{2}RT\left( t \right) + \epsilon$$$${R\left( t \right)~or~P\left( t \right) = ~\beta}_{0} + ~\beta_{1}O\left( t \right) + ~\beta_{2}RT\left( t \right) + \beta_{3}P\left( {t - 5} \right) + {\sum_{i = 1}^{4}{\beta_{3 + i}~R\left( t - i \right)}} + \epsilon$$$${R\left( t \right)~or~P\left( t \right) = ~\beta}_{0} + ~\beta_{1}O\left( t \right) + ~\beta_{2}RT\left( t \right) + {\sum_{i = 1}^{5}{\beta_{2 + i}~O\left( t - i \right)}} + \epsilon$$$${~R\left( t \right)~or~P\left( t \right) = ~\beta}_{0} + ~\beta_{1}O\left( t \right) + ~\beta_{2}RT\left( t \right) + \beta_{3}P\left( {t - 5} \right)~{\sum_{i = 1}^{4}{\beta_{3 + i}~R\left( {t - i} \right) +}}{\sum_{i = 1}^{5}{\beta_{8 + i}~O\left( t - i \right)}} + \epsilon$$$$~R\left( t \right)~or~P\left( t \right) = ~\beta_{0} + \beta_{1}O\left( t \right) + ~\beta_{2}RT\left( t \right) + \beta_{3}R\left( {t - 1} \right) + \epsilon.$$
For both R- and P-judgments, our regression models assume that current confidence is related to objective features of the decision (accuracies and reaction times) and/or previous subjective ratings. To identify the best-fitting models we computed information criteria. Bayesian information criterion (BIC; [@niw018-B68]) scores were compared at the group level using Kass and Raftery's grades of evidence ([@niw018-B36]). The difference in BIC provides support for one model against another with the following grades: none for a negative difference; weak for a value between 0 and 2, positive between 2 and 6; strong between 6 and 10; and very strong for a difference greater than 10. We additionally computed the Akaike Information Criterion (AIC, [@niw018-B2]) which penalizes the number of parameters less strongly ([@niw018-B75]).
### Learning models
To complement our regression analyses, we examined whether past successes and/or previous confidence ratings affected subsequent P-confidence within a reinforcement learning framework ([@niw018-B72]). These models are convenient tools to analyze how individuals learn predictions over time through trial and error (see [@niw018-B20]; [@niw018-B55], for reviews). We specified the relationship between reported P-confidence ($\left. P_{obs} \right)$ and predicted P-confidence ($\widehat{P}$) by the following regression equation: $P_{obs} = \,\beta_{0} + \,\beta_{1}\widehat{P} + \epsilon,~$with $\epsilon$ following a Normal distribution. $\widehat{P}$ was generated from different candidate learning models:
\(A\) Objective Model: $$\widehat{P}\left( {t + 1} \right) = ~\widehat{P}\left( t \right) + ~\alpha\left\lbrack {O\left\lbrack t \right\rbrack - ~\widehat{P}\left( t \right)} \right\rbrack.$$
\(B\) Subjective Model: $$\widehat{P}\left( {t + 1} \right) = ~\widehat{P}\left( t \right) + ~\alpha\left\lbrack {R\left\lbrack t \right\rbrack - ~\widehat{P}\left( t \right)} \right\rbrack.$$
Both models assume that P-confidence at *t * + * *1 is related to its value on the previous trial *t*. In addition, both models compute a "prediction error" (in square brackets), which is the difference between either the obtained outcome, $O\left( t \right)$, and previous P-confidence (in Model A), or the current trial's R-confidence and previous P-confidence (in Model B). The prediction error can thus be thought of as driving the update of subsequent P-confidence, with the magnitude of this update being controlled by the free learning rate parameter$~\alpha.$ Model (A) only takes into account objective success and thus the prediction error is affected by the accuracy of previous trials, $O\left( t \right)$, as in standard RL models ([@niw018-B72]). Model (B) instead computes a prediction error based on subjective ratings of performance ([@niw018-B19]; [@niw018-B32]): the difference between previous P- and R-confidence. We additionally compared each model to a null (intercept-only) model in which $\widehat{P}$ remained constant across trials.
In all models $\widehat{P}$ was initialized to 0.75. Best fitting parameters were obtained by minimizing the error term in the regression equation above using a nonlinear optimization routine in Matlab (*fminsearch*). Predicted $\widehat{P}$ values obtained from the best-fitting parameters for each subject could then be compared against the observed P-confidence on a trial-by-trial basis. The log-likelihood of each subject's P-confidence ratings under each candidate model was entered into the following equation to obtain a BIC score (where *L *= * *the log-likelihood, *k *= * *number of free parameters, *n * = * *number of observed P-confidence ratings): $$BIC = - 2 \cdot {\ln{L + k}} \cdot {\ln n}.$$
Results
=======
Subjects completed 200 trials of a visual discrimination task in which task difficulty remained constant across trials ([Fig. 1](#niw018-F1){ref-type="fig"}). Each trial was followed by a retrospective confidence judgment (R-trials), and every 5th trial was preceded by a prospective confidence judgment (P-trials). Task performance on P-trials (mean 66.8%, SD 10.1%) did not significantly differ from performance on R-trials (mean 67.2%, SD 7.0%; *t*(38) = 0.33, *P* = 0.74). The distribution of confidence ratings given for each judgment type is shown in [Fig. 2](#niw018-F2){ref-type="fig"}.
{#niw018-F2}
### Bias and calibration
We first examined subjects' global level of overconfidence for each judgment type by computing the difference between average confidence and average performance. In these analyses, we excluded one subject with an extreme prospective calibration score that can be seen in the scatter plot in [Fig. 3D](#niw018-F3){ref-type="fig"}. Consistent with a large body of literature ([@niw018-B8]; [@niw018-B34]; [@niw018-B5]), subjects were systematically overconfident for both prospective (one-sample *t*-test against zero, *t*(38) = 3.27, *P* \< 0.01) and retrospective (one-sample *t*-test against zero, *t*(38) = 7.13, *P* \< 10 ^−^ ^7^) judgments ([Fig. 3A](#niw018-F3){ref-type="fig"}). Furthermore, this overconfidence was stable across judgment type: there was no significant difference between prospective and retrospective overconfidence ([Fig. 3A;](#niw018-F3){ref-type="fig"}*t*(38) = 1.63, *P* = 0.11), and both measures were correlated across subjects ([Fig. 3B;](#niw018-F3){ref-type="fig"}*r* = 0.43, *P *= 0.007). Together these results indicate that global overconfidence (bias) in decision-making transcends the temporal focus of judgments of performance.
{#niw018-F3}
Within each judgment type, we additionally quantified the discrepancy between mean performance level at each scale step (e.g. 60% correct) and its associated confidence level (e.g. 80%), with a lower discrepancy giving a better calibration score. Retrospective calibration was significantly better (lower) than prospective calibration ([Fig. 3C;](#niw018-F3){ref-type="fig"}*t*(37) = 2.49, *P* = 0.017), although unlike for global overconfidence, calibration was not significantly correlated across subjects ([Fig. 3D](#niw018-F3){ref-type="fig"}; *r = *0.22, *P* = 0.19).
### Metacognitive sensitivity and discrimination
We next considered metacognitive sensitivity -- the ability to track changes in performance with changes in confidence (measured as the area under the type 2 ROC; AUROC2). R-metacognitive sensitivity was systematically higher than P-metacognitive sensitivity ([Fig. 4A;](#niw018-F4){ref-type="fig"}*t*(38) = 5.77, *P* \< 0.001), and these measures were not significantly correlated across subjects ([Fig. 4B;](#niw018-F4){ref-type="fig"}*r* = −0.25, *P *= 0.13). Indeed, prospective judgments did not carry reliable information about subsequent accuracy, with AUROC2 being statistically indistinguishable from 0.5 (*t*(38) = 0.42, *P* = 0.68). The same pattern remained for an alternative measure of metacognitive sensitivity derived from the forecasting literature (ANDI; [@niw018-B78]). P-ANDI was systematically lower than R-ANDI ([Fig. 4C;](#niw018-F4){ref-type="fig"}*t*(38) = 7.03, *P* \< 10 ^−^ ^7^), and these measures were not significantly correlated across subjects ([Fig. 4D;](#niw018-F4){ref-type="fig"}*r* = −0.26, *P *= 0.11).
{#niw018-F4}
### Formation of subjective ratings
The previous analyses indicate that while global confidence levels transcend prospective and retrospective ratings, prospective judgments of performance show markedly lower calibration and sensitivity to local fluctuations in performance. This is consistent with subjects monitoring trial-specific decision accuracy post-decision, while leveraging long-run performance estimates to construct prospective judgments. We next investigated how subjects form prospective and retrospective ratings during the task. For instance, we might expect a prospective judgment of performance to be based on past experience of the task. If you have been successful in 7 out of 10 previous trials, it is sensible to predict a 70% chance of success for the subsequent trial -- and in a stationary environment in which task difficulty remains constant (as in the current experiment), such a strategy will lead to reasonable forecasts. However, previous studies have shown that subjects do not show optimal learning even in stationary environments, and instead are prone to biases such as the hot-hand fallacy (where a win leads to an inflated prediction of success on the next trial; [@niw018-B6]; [@niw018-B57]). We might therefore expect that the value of P-confidence depends on recent performance, and that either objective aspects of previous task performance (such as accuracy) and/or previous confidence ratings will affect subsequent prospective judgments.
We used hierarchical mixed models (see "Materials and Methods" section) to estimate the effects of previous ratings and previous accuracy on the formation of both R- and P-confidence. [Tables 1](#niw018-T1){ref-type="table"} and [2](#niw018-T2){ref-type="table"} show the regression coefficients for different models of R and P-confidence, and [Fig. 5](#niw018-F5){ref-type="fig"} plots the coefficients from the full model. We found significant influences of current-trial accuracy and reaction time on R-confidence, with faster and more accurate decisions being associated with greater confidence. For lagged factors, the previous trial's R-confidence had an effect on current-trial confidence in Model 2, whereas previous accuracy did not have an effect. For R-confidence, both BIC and AIC scores provided very strong support for Model 5, which included only current-trial predictors (RT and accuracy) and the immediately preceding R-confidence judgment. In contrast, in models of P-confidence, we found a significant dependence on the previous level of P-confidence, as well as previous ratings of R-confidence over the previous four trials. Previous accuracy had a weaker effect, especially when controlling for previous R-confidence (Model 4). The BIC scores provided very strong support for Model 2, which included predictors for previous R- and P-confidence, over the next best Model 4 which included all predictors. However, a comparison of AIC scores revealed inconclusive evidence (ΔAIC \< 3) for Model 4 over Model 2, indicating that the difference in BIC is primarily driven by the penalization for model complexity.
{ref-type="table"} for full details.](niw018f5p){#niw018-F5}
######
Hierarchical linear regressions of R-confidence on past and present accuracy, past R-confidence, and reaction time. Lag into the past is indicated by increasing indices (e.g. R-confidence_2 indicates the R-confidence judgment made two trials previously).
R-confidence \(1\) \(2\) \(3\) \(4\) \(5\)
---------------- ------------------------------------------------- ------------------------------------------------- ------------------------------------------------- ------------------------------------------------- -------------------------------------------------
accuracy 0.0394[\*\*](#niw018-TF2){ref-type="table-fn"} 0.0396[\*\*](#niw018-TF2){ref-type="table-fn"} 0.0394[\*\*](#niw018-TF2){ref-type="table-fn"} 0.0395[\*\*](#niw018-TF2){ref-type="table-fn"} 0.0397[\*\*](#niw018-TF2){ref-type="table-fn"}
accuracy_1 0.0024 −0.0022
accuracy_2 0.0027 −0.0004
accuracy_3 −0.0024 −0.0013
accuracy_4 −0.0010 −0.0011
accuracy_5 −0.0018 −0.0025
R-confidence_1 0.1309[\*\*](#niw018-TF2){ref-type="table-fn"} 0.1325[\*\*](#niw018-TF2){ref-type="table-fn"} 0.1428[\*\*](#niw018-TF2){ref-type="table-fn"}
R-confidence_2 0.0099 0.0100
R-confidence_3 0.0004 0.0015
R-confidence_4 −0.0081 −0.0073
P-confidence_1 0.0396 0.0379
RT −0.0588[\*\*](#niw018-TF2){ref-type="table-fn"} −0.0624[\*\*](#niw018-TF2){ref-type="table-fn"} −0.0587[\*\*](#niw018-TF2){ref-type="table-fn"} −0.0624[\*\*](#niw018-TF2){ref-type="table-fn"} −0.0584[\*\*](#niw018-TF2){ref-type="table-fn"}
Intercept 0.8337[\*\*](#niw018-TF2){ref-type="table-fn"} 0.7052[\*\*](#niw018-TF2){ref-type="table-fn"} 0.8336[\*\*](#niw018-TF2){ref-type="table-fn"} 0.7078[\*\*](#niw018-TF2){ref-type="table-fn"} 0.7212[\*\*](#niw018-TF2){ref-type="table-fn"}
AIC −8551 −8545 −8553 −8555 −8697
BIC −8504 −8431 −8499 −8468 −8636
The AIC and BIC score for each model is provided.
\*\**P* \< 0.001.
######
Hierarchical linear regressions of P-confidence on past accuracy, past R-confidence, and past P-confidence. Lag into the past is indicated by increasing indices (e.g. R-confidence_2 indicates the R-confidence judgment made two trials previously).
P-confidence \(1\) \(2\) \(3\) \(4\) \(5\)
---------------- ------------------------------------------------ ------------------------------------------------ ------------------------------------------------ ------------------------------------------------ ------------------------------------------------
accuracy 0.0003 −0.0017 0.0008 −0.0009 −0.0001
accuracy_1 0.0142[\*\*](#niw018-TF4){ref-type="table-fn"} 0.0084
accuracy_2 0.0143[\*\*](#niw018-TF4){ref-type="table-fn"} 0.0075
accuracy_3 0.0030 0.0011
accuracy_4 0.0024 0.0002
accuracy_5 0.0058 0.0073
R-confidence_1 0.1206[\*\*](#niw018-TF4){ref-type="table-fn"} 0.1167[\*\*](#niw018-TF4){ref-type="table-fn"} 0.1564[\*\*](#niw018-TF4){ref-type="table-fn"}
R-confidence_2 0.1041[\*\*](#niw018-TF4){ref-type="table-fn"} 0.1003[\*\*](#niw018-TF4){ref-type="table-fn"}
R-confidence_3 0.0529[\*\*](#niw018-TF4){ref-type="table-fn"} 0.0541[\*\*](#niw018-TF4){ref-type="table-fn"}
R-confidence_4 −0.0375[\*](#niw018-TF5){ref-type="table-fn"} 0.0365[\*](#niw018-TF5){ref-type="table-fn"}
P-confidence_1 0.1820[\*\*](#niw018-TF4){ref-type="table-fn"} 0.1790[\*\*](#niw018-TF4){ref-type="table-fn"}
RT −0.0016 0.0004 −0.0014 0.0004 −0.0007
Intercept 0.7481[\*\*](#niw018-TF4){ref-type="table-fn"} 0.3633[\*\*](#niw018-TF4){ref-type="table-fn"} 0.7208[\*\*](#niw018-TF4){ref-type="table-fn"} 0.3539[\*\*](#niw018-TF4){ref-type="table-fn"} 0.6254[\*\*](#niw018-TF4){ref-type="table-fn"}
AIC −3304 −3585 −3333 −3587 −3411
BIC −3267 −3542 −3290 −3517 −3363
The AIC and BIC score for each model is provided.
\*\**P* \< 0.001;
\**P* \< 0.05.
In summary, when comparing prospective and retrospective judgments, we found that R-confidence is strongly influenced by features of the current and immediately preceding decision, whereas P-confidence showed a dependence on past confidence extending back over a longer time window (the past four R-confidence ratings and previous P-confidence).
To complement our regression analyses, we fit reinforcement learning models to P-confidence judgments that updated predictions of performance using either previous outcomes or subjective ratings ([@niw018-B72]; see "Materials and Methods" section). Model A updated P-confidence based on previous successes and failures, whereas Model B updated P-confidence based on previous subjective ratings. Both models outperformed a null intercept-only model that did not allow P-confidence to adapt as a function of past experience (differences in group BIC score \> 100). The learning rate parameter (alpha) in both models was similar (Model A: mean alpha = 0.20; Model B: mean alpha = 0.23; paired *t*-test, *t*(38) = 0.63, *P* = 0.53). The fits of each candidate model for three example subjects is shown in [Fig. 6](#niw018-F6){ref-type="fig"}. By comparing model fits at the group level using a Bayesian random-effects model selection algorithm ([@niw018-B71]), we found that Model B provided the best account of subjects' data (exceedance probability = 0.98). Together with our regression analyses, these model fits indicate that prospective predictions of performance are themselves influenced by recent retrospective confidence, over and above effects of objective accuracy.
![Models of P-confidence updates. (**A**) Fits of each candidate learning model to data from three example subjects. The blue lines show subject ratings of P-confidence; the red lines show model fits. (**B**) Difference in BIC scores for Models A and B for each subject. A difference in BIC of 6 or more is considered strong evidence in favor of a particular model. By comparing model fits at the group level using a Bayesian random-effects model selection algorithm ([@niw018-B71]), we found that Model B provided the best account of subjects' data in the group as a whole (exceedance probability = 0.98).](niw018f6p){#niw018-F6}
Discussion
==========
Here we directly compared prospective and retrospective metacognitive judgments of performance in the same visual discrimination task in which difficulty remained constant across trials. In line with our hypothesis we found that, despite retrospective judgments having access to additional trial-specific information, participants' global confidence levels generalized across judgment types. This finding is consistent with a global level of confidence being a stable individual difference that may generalize across different task contexts ([@niw018-B1]). We also found that retrospective judgments exhibited greater accuracy and calibration compared to prospective judgments. This increase in accuracy is likely due to retrospective judgments having additional access to the internal state of evidence supporting a particular choice, rather than only the aggregate likelihood of success ([@niw018-B62]). In turn, trial-to-trial stimulus variation (such as changes in dot position) may be a potential source of fluctuation in internal sensory evidence. In line with this interpretation, we found that local variables such as current-trial accuracy and response time predicted retrospective judgments of confidence. This is compatible both with theories of metacognition that emphasize the importance of trial-by-trial inferential cues when judging confidence ([@niw018-B41]), such as response fluency ([@niw018-B3]), and computational perspectives that emphasize a continuous tracking of the evidence in favor of a decision ([@niw018-B74]; [@niw018-B39]). Intriguingly, a recent study found a boost in accuracy for retrospective compared to prospective judgments even when trial-specific stimulus evidence was available in both cases (Siedlecka *et al.* 2016), suggesting that the simple act of making a response may provide a further cue to improve metacognitive accuracy.
In contrast, prospective judgments require learning about the overall task difficulty (which was kept constant in this experiment) in order to predict success on an upcoming trial. Our regression models reveal a differential influence of past confidence on P- and R-judgments: P-confidence was influenced by past R-confidence over a longer time-window, whereas R-confidence exhibited a dependence only on the last rating. Nevertheless, we observed a reasonably high learning rate (alpha ∼ 0.2) for the integration of previous retrospective confidence judgments into a prediction of P-confidence. This recency effect is suboptimal in a stationary environment in which subjects should predict confidence in an upcoming decision based on their long-run experience with the task, rather than immediately preceding outcomes, but is consistent with findings of strong sequential dependencies in confidence reports ([@niw018-B63]). However, we note that some subjects did exhibit above-chance prospective metacognitive sensitivity. One alternative, but not mutually exclusive, explanation for this variability is that some subjects exhibit a "self-fulfilling prophecy." Having rated high confidence, one might then devote greater attention and focus to the subsequent trial such that one's confidence is justified. In line with this interpretation, [@niw018-B80] showed that biasing confidence levels with false feedback had a positive effect on future task performance. Future studies are required to distinguish between these perspectives on the accuracy of prospective judgments, for instance by asking whether the presence or absence of prospective judgments affects features of individual decisions.
There is increasing recognition in the neurosciences that the brain may use a common schema for trial-and-error learning about different aspects of the environment (see [@niw018-B55], for a review). This class of reinforcement learning algorithms has been successfully applied to explain behavior and neural responses when learning about rewards and punishments ([@niw018-B56]; [@niw018-B21]), social reputation ([@niw018-B11]), and during interactive games ([@niw018-B40]; [@niw018-B33]). More recently, these models have also been applied to explain changes in metacognitive variables such as subjective confidence in the absence of explicit feedback ([@niw018-B19]; [@niw018-B32]). Here we provide initial evidence that subjects' prospective judgments of performance can also be modeled as a trial-to-trial update based on previous subjective confidence. Specifically, a model in which prospective judgments of performance are constructed from local fluctuations in recent retrospective confidence provided a better fit to the data than one in which predictions were built from outcomes (objective accuracy) alone. How these simple learning mechanisms may affect metacognitive accuracy remains an important question for future study. (We checked for correlations between individual differences in prospective calibration, ANDI and AUROC2 with the best-fitting parameters of Model B but did not find any significant associations (*P* \> 0.05).)
It is perhaps more striking that bias, or overconfidence, is stable across prospective and retrospective judgments. There are a number of previous accounts of overconfidence. The ecological perspective proposes that overconfidence is due to a biased retrieval of heuristic cues when answering general knowledge questions ([@niw018-B31]). However, this model cannot account for systematic overconfidence when evaluating performance on perceptual discrimination tasks such as the one used here. An alternative proposal is that stochastic sampling of evidence leads to overconfidence ([@niw018-B23]; [@niw018-B48]). However, here we find stable overconfidence not only for post-decision assessments that are naturally accommodated by an evidence accumulation framework ([@niw018-B48]; [@niw018-B61]) but also for prospective assessments of performance that may rely on distinct mechanisms. Our result is instead consistent with previous findings that overconfidence reflects a stable trait that transcends particular judgment types ([@niw018-B76]; [@niw018-B37]; [@niw018-B1]), and that is potentially distinct from variability in metacognitive accuracy ([@niw018-B73]; [@niw018-B27]; [@niw018-B1]). Our finding that sequential dependencies exist between retrospective and prospective judgments of performance provides one potential explanation for why stable overconfidence is maintained across temporal focus.
More broadly, our study raises the question of the appropriate generative model for prospective and retrospective metacognitive judgments. Recent progress has been made in understanding the computational basis for retrospective judgments of decision confidence ([@niw018-B30]; Pleskac and Busemeyer 2010; [@niw018-B46]; [@niw018-B62]). On a signal detection model, confidence is computed by comparing an internal (perceptual, mnemonic) signal to a criterion and then further processed as an explicit metacognitive report. However, this model is limited to designs in which metacognitive judgments are elicited after the first-order task has been completed, and would appear difficult to extend to prospective judgments such as judgments of learning ([@niw018-B4]). In the present study, subjects were asked to make prospective judgments of how likely they were to succeed on a subsequent trial. Information about future success can be garnered from previous experience, and it would be of interest to extend current models of metacognition to encompass learning over one's past performance as a relevant internal signal for confidence. On a practical level, SDT measures of metacognition are still likely to be useful for analyzing prospective judgments, as they naturally separate sensitivity from bias (over- or underconfidence).
We close with some limitations of the present study. Several trials are needed to get robust estimates of AUROC2, and in our dataset the number of P-trials is low. However, we note that the same conclusions hold when using an alternative measure, ANDI, which does not rely on the same parametric assumptions as SDT ([@niw018-B78]). In addition, despite the asymmetry in trial number (40 P-trials and 160 R-trials), due to an incentivized elicitation mechanism each trial contributed equally to subjects' earnings in the task. Thus it is unlikely that motivational differences between conditions can explain the discrepancy in judgments of confidence. In addition, here we only consider prospective judgments made before stimulus material pertaining to the current decision has been experienced. In other domains, subjects are able to form reliable single-trial prospective judgments such as feelings-of-knowing or judgments-of-learning ([@niw018-B16]; [@niw018-B17]; [@niw018-B81]). It may be possible to augment the current task design to more closely mimic those used in metamemory tasks, e.g. by asking subjects to predict how well they will be able to discriminate an identical stimulus in a subsequent testing phase. Conversely, it remains to be seen whether the trial-to-trial dynamics of confidence observed here, such as the influence of previous confidence on future predictions of performance, generalize to other metacognitive domains such as memory and general knowledge.
Summary
=======
To conclude, previous studies have typically focussed on retrospective metacognitive judgments of perceptual decision-making. Here we compare the construction of retrospective and prospective confidence judgments within the same task using repeated stimuli of constant difficulty. We find dissociable influences on each judgment type: retrospective judgments are strongly influenced by current-trial fluency and accuracy and confidence in immediately preceding decisions, whereas prospective judgments are influenced by previous confidence over a longer time window. In contrast, global levels of confidence were correlated across judgments, indicative of a domain-general overconfidence that transcends temporal focus. Our findings extend the study of metacognition of perception to prospective judgments, and lay the groundwork for future studies of the neural basis of prospective confidence.
The authors report no conflict of interest. Raw data and code for reproducing the figures can be obtained from <https://github.com/smfleming/Past-future-metacognition-paper>. Author Contributions: S.F., T.G., S.M., and J.C.V. designed the task; S.M. carried out data collection; S.F. and S.M. carried out data analysis; S.F. and S.M. wrote the article; all authors interpreted the findings and provided critical comments on the article.
Funding
=======
S.M.F. was supported by a Sir Henry Wellcome Fellowship (WT096185). S.M. and J.C.V. were supported by the ANR (Riskemotion-ANR-08-RISKNAT-007-01).
[^1]: †Fleming, Stephen: http://orcid.org/0000-0003-0233-4891
[^2]: ‡These authors contributed equally to this work.
| {
"pile_set_name": "PubMed Central"
} |
Background
==========
It is particularly challenging to provide policy-relevant information on women in the area of substance use and misuse, as the topic covers such broad territory. Problem substance use has been found to arise from a complex interplay of biological, genetic, psychological, social, cultural, relational, environmental and spiritual factors. Because of the interplay of these factors, it remains a challenge for policy makers and program planners to develop and implement the very broad, collaborative, systemic responses required and to do so in a manner that links prevention, enforcement, harm reduction and treatment strategies. In spite of the overall challenges, advances in gender-specific policy and programming are especially worthy of consideration and could be of tremendous benefit to the health of women and their families.
This chapter presents findings from both surveillance sources and select research reports. As there has not been recent or adequate surveillance specific to substance use by Canadians, cross-sectional data from the Canadian Community Health Survey (CCHS) Cycle 1.1 (2000) are highlighted. While trends in women\'s substance use and substance use problems could not be provided, an overview of key research findings is offered that supports a comprehensive understanding of the topic from a gender perspective and places the findings of the CCHS in the appropriate context. Readers are encouraged to consult existing research reports mentioned throughout the chapter to further their understanding of this complex topic.
Physical Health Impact of Problem Substance Use
-----------------------------------------------
Health Canada\'s Straight Facts about Drugs and Drug Abuse \[[@B1]\] outlines the health consequences of high doses of all classes of psychoactive substances, including stimulants, hallucinogens, cannabis and central nervous system depressants (opioid analgesics, alcohol, inhalants/solvents, benzodiazepines, barbiturates).
For some drug classifications, evidence of sex differences in the physical health impact of substance use is available. \[[@B2]\] To take alcohol as an example, women metabolize alcohol and other psychoactive substances more slowly than men, allowing harmful metabolites to remain longer in the body. Women are more likely than men to develop cirrhosis of the liver after consuming lower levels of alcohol over a shorter period of time. \[[@B3],[@B4]\] These findings also apply to brain shrinkage and impairment, \[[@B5]-[@B7]\] breast cancer, \[[@B8]-[@B12]\] gastric ulcers \[[@B13]\] and alcoholic hepatitis. \[[@B14]\]
Although there is not adequate research on sex differences in the impact of illicit drugs, gender differences are beginning to be documented. \[[@B15]\] Injecting drug use (IDU) is a key risk factor among women for the transmission of blood-borne diseases such as HIV/AIDS and hepatitis C. \[[@B16]\] Also linked to illicit drug use are high-risk sexual behaviours (e.g. sex trade work), which, in turn, are associated with a range of negative health impacts. \[[@B17]\]
Mental Health, Trauma and Substance Use
---------------------------------------
Research has shown that as many as two thirds of women with substance misuse problems may have a concurrent mental health problem, such as depression, post-traumatic stress disorder, panic disorder and/or an eating disorder. \[[@B18]\] Research also shows that a large proportion of women with substance use problems are victims of domestic violence, incest, rape, sexual assault and child physical abuse. \[[@B19]-[@B22]\] Victimization has been linked to a variety of negative outcomes, including post-traumatic stress disorder, depression, anxiety, suicidal behaviour and low self-esteem among women in the general population. \[[@B23],[@B24]\] Compared with non-abused clients, women in treatment for problem substance use who have been victimized are more likely to suffer from depression and suicidal ideation, have lower self-esteem, negative psychological adjustment and more post-traumatic stress symptoms. \[[@B25]-[@B30]\]
Pregnancy, Mothering and Substance Use
--------------------------------------
Substance use by pregnant women and mothers has received a great deal of attention, and there is a voluminous literature documenting the adverse effects of alcohol, tobacco and other drugs on the fetus. Alcohol use during pregnancy, particularly in combination with poor nutrition, poor general health, experience of trauma and mental health problems, and lack of prenatal care, has been found to have the most harmful effects. \[[@B31],[@B32]\]
Maternal use of licit and illicit drugs can also result in problems that have short- or long-term consequences for those prenatally exposed. However, study of the impact of these drug categories is hampered by methodological flaws that fail to take into account the use of more than one drug. \[[@B33]\] Rather than the provision of effective outreach, engagement and treatment, the responses in some countries and to some extent in Canada towards women who use illicit substances during pregnancy have been blame, attempts to force women into treatment, and even criminal sanctions. \[[@B34]-[@B37]\]
The Health Canada document entitled Best Practices: Fetal Alcohol Syndrome/Fetal Alcohol Effects and the Effects of Other Substance Use during Pregnancy \[[@B32]\] provides a comprehensive view of the issues and of promising fetal alcohol syndrome (FAS) prevention practices grounded in both the literature and the expertise of Canadians working in the field. Programs such as the Breaking the Cycle Program in Toronto and the Sheway Program in Vancouver are examples of effective programs being developed in Canada to serve women at high risk of having a child affected by FAS.
Stigma and Barriers to Treatment for Women
------------------------------------------
In the forefront of psychosocial influences on women\'s misuse of alcohol and other drugs is the stigma arising from societal attitudes towards women\'s substance use. \[[@B38]\] This stigma causes women to feel considerable guilt and shame as their substance use/misuse continues and creates barriers to their accessing help. \[[@B39]-[@B43]\] The stigma associated with women\'s substance use affects service providers as well. Women often encounter mis-information, denial, inaction and even punitive attitudes towards their substance use by professionals in a position to intervene, and thus they may not be identified as needing help. \[[@B37],[@B44]-[@B46]\]
Stigma intersects with structural and other barriers that arise from experience of victimization and mental health problems, fear of having one\'s children apprehended, minority status, income and rural location, to name but a few. Health Canada\'s report Best Practices: Treatment and Rehabilitation of Women \[[@B40]\] has been developed as a key resource for health care providers in helping to reduce barriers to treatment access for diverse groups of substance-using women.
The barriers to accessing supportive treatment services are even greater for pregnant and parenting women. Programs that accept both mothers and children are very scarce. Finding affordable and safe care for their children as well as residential care is often an overwhelming obstacle for mothers needing treatment. \[[@B47]\] Although outpatient counselling and other supports for pregnant and parenting women have been made available, appalling barriers remain for mothers in need of residential treatment for problem substance use in Canada.
Gender and Harm Reduction Approaches
------------------------------------
Harm reduction is a newer policy approach to substance misuse, which arose initially as a response to the spread of AIDS among injection drug users. \[[@B48]\] The primary goal of harm reduction policy and practice is a reduction in the negative consequences associated with use rather than complete cessation of substance use. Key harm reduction programs and policies in Canada relating to illicit drugs include methadone maintenance; needle/syringe exchange; supervised injection sites; increased attention to the decriminalization of use of small quantities of cannabis; and education, information and diversion programs. Such initiatives can be beneficial to women who are drug users and/or who have partners who are drug users in helping prevent HIV transmission, improve their access to drug treatment and health care services, prevent birth defects and disabilities in their children, make incarceration experiences safer, and stabilize their health overall. Particularly promising is harm reduction programming offered to high-risk, marginalized pregnant women and their support networks, an approach that has been shown to prevent FAS and other alcohol- and drug-related disabilities, support retention of custody, and increase the health and social stability of both mothers and children. \[[@B47]\]
The Current Study
-----------------
We explored the use of alcohol, licit and illicit substances by Canadian women and subgroups of women (including girls, elderly women, Aboriginal women and pregnant women). Additionally, the health consequences of substance use, and the coexistence of mental health, trauma and substance use issues, were explored.
Methods
=======
Data
----
Cross-sectional data from the Canadian Community Health Survey (CCHS) Cycle 1.1 (2000--2001) \[[@B49]\] are presented below. The data are weighted to represent the Canadian population. Age-adjusted data are presented where appropriate. The sample comprised 125,574 individuals, aged 12 years and older, from across all provinces and territories. Data from other sources (including select research reports, such as the Canadian Profile on Alcohol, Tobacco and Other Drugs \[[@B50]\]) are similarly provided to highlight what is known about the use of alcohol, licit and illicit drugs by women, and the effects of this use. Where pertinent, comparisons between females and males are made.
It is important to note that comparability issues arise when different surveys, data sources and research reports are compared. Key issues of concern are related to design (e.g. the jurisdictions included in the survey) and methodology (e.g. the form of the survey -- telephone or face-to-face interviews). For a full understanding of the comparability of the research studies reported on here, the reader is encouraged to consult the original source. Likewise, it is important to point out some of the key limitations of general population surveys. These include the difficulty in obtaining information from and about hard-to-reach populations (heavy substance users living on the street or in an institution), under-reporting because the behaviour is socially sanctioned, recall difficulties, and/or under-reporting of psychoactive medication use among older adults due to their lack of awareness of the reasons for taking multiple medications. Nonetheless, survey instruments are able to provide information regarding the lower-bound estimates of substance use measures important to public health monitoring. \[[@B50]\]
Measures
--------
### Alcohol Use
Data from the Alcohol (and Breast-feeding) modules of the CCHS were extracted in order to determine lifetime use of alcohol, alcohol use and frequency of use in the previous 12 months, heavy drinking (defined here as five or more drinks on at least one occasion at least once per month), and alcohol use and frequency of use in last pregnancy. Differences by sex, age and Aboriginal status were explored. Data examined by Single and colleagues from 1994 to 1995 and 1995 to 1996 from the Canadian Institute for Health Information (CIHI) on hospital separations attributed to alcohol use are strongly endorsed within the field and are also presented. \[[@B51]\]
### Licit Drug Use
Data from the Drug Use module of the CCHS were extracted in order to determine the use of the following licit substances: pain relievers, tranquilizers, diet pills, antidepressants, opioid analgesics (such as codeine, Demerol or morphine), and sleeping pills. Only participants from Ontario (all health regions) were administered the Drug Use module of the CCHS. As a result, findings may not be generalizable to other provinces and territories. Differences by sex, age and Aboriginal status were explored. The literature on the health consequences of long-term licit drug use is summarized.
### Illicit Drug Use
Information about illicit drug use was not collected through the CCHS Cycle 1.1, therefore data from the 1994 Canada\'s Alcohol and Other Drugs Survey (CAODS),\[[@B52]\] the 1998 Canadian Campus Survey \[[@B53]\] and hospital separations attributed to illicit drug use \[[@B51]\] are presented.
### Mental Health Issues, Trauma and Substance Use
Data from the Depression, Contacts with Mental Health Professionals, and Suicidal Thoughts and Attempts modules of the CCHS Cycle 1.1 were extracted in order to explore the relation between mental health issues and substance use in women. Seventy of the 136 health regions in Canada (all in the provinces of Nova Scotia, New Brunswick, Quebec, Manitoba, Alberta, and a few in Ontario) administered the Suicidal Thoughts and Attempts questionnaire. All but two health regions administered the Depression questionnaire. Data concerning rates of trauma among substance-using women in treatment centres in Canada are presented.
Results
=======
Alcohol Use
-----------
### Prevalence
According to the 2000--2001 CCHS, \[[@B49]\] 73.1% of Canadian women aged 12 years and older had used alcohol at least once in the previous year and 13.3% reported using alcohol during their lifetime but not in the previous year (see Figure [1](#F1){ref-type="fig"}). Alcohol use was more prevalent among males than females. The majority of females who reported having had a drink in the previous year tended to drink infrequently (see Figure [2](#F2){ref-type="fig"}).
{#F1}
![Patterns of Alcohol Consumption among Women who Reported Having had a Drink in the Previous 12 Months: Canadian Community Health Survey \[[@B49]\]](1472-6874-4-S1-S8-2){#F2}
The rate of alcohol use in the previous 12 months among adolescent girls aged 15 to 19 years was 71.1% and among Canadian women, on average, was 75.5%. It appears that more Aboriginal than non-Aboriginal women reported ever drinking alcohol and drinking alcohol in the previous 12 months. Roughly 11% of the women reporting alcohol use in the previous 12 months met the criterion for any occasion heavy drinking. Twice as many Aboriginal women as non-Aboriginal women met this criterion.
### Alcohol Use in Pregnancy
Fourteen percent of all women who indicated that they consumed alcohol in their lifetime also reported that they had consumed alcohol during their last pregnancy. The vast majority of women reporting alcohol use during a pregnancy reported drinking infrequently: 75.4% drank less than once per month; 9.7% drank once per month; 6.5% drank two to three times per month; 5.3% drank once per week; and 1.3% drank every day.
### Morbidity
Single and colleagues \[[@B51]\] examined 1994--1995 and 1995--1996 data from CIHI and 1991--1992 data from the then Laboratory Centre for Disease Control, Health Canada, on hospital separations attributable (fully and partially) to alcohol (e.g. alcohol toxicity, alcoholic liver cirrhosis) and estimated that 29,181 women (51,765 men) were hospitalized in 1995--1996 because of alcohol. The greatest number of hospitalizations among women was for accidental falls (7,689), followed by motor vehicle accidents (3,433), alcohol dependence syndrome (3,247) and suicide/self-injury (2,473). Hospital separations per 100,000 have remained fairly stable among women (from 206 in 1991--1992 to 193 in 1995--1996) and have declined among men (from 401 in 1991--1992 to 349 in 1995--1996).
### Mortality
According to Single and colleagues\' analysis of 1995 CIHI data, the estimated number of deaths attributed (fully and partially) to alcohol consumption among hospitalized women was 1,823 or 12.4 per 100,0000 population (among men it was 4,681 or 32.3 per 100,000 population). The greatest number of deaths among women was attributed to motor vehicle accidents (357), alcoholic liver cirrhosis (257), accidental falls (202), breast cancer (192), alcohol dependence syndrome (135) and suicide/self-inflicted injury (134). The potential years of life lost in 1995 were estimated at 48,392 for women or 327.9 per 100,000 population (123,734 for men or 853.7 per 100,000 population).
Licit Drug Use
--------------
**Note:**The licit drug use identified in the CCHS survey should not be seen as indicating misuse or problemuse.
### Prevalence
According to the 2000--2001 CCHS, the proportion of females aged 12 or older reporting use in the previous year of selected prescription and non-prescription drugs was 23.7% for pain relievers, 2.1% for opioid analgesics, 1.7% for sleeping pills, 1.1% for tranquilizers, 2.1% for antidepressants and 0.4% for diet pills. Females consistently reported higher rates of use than males in all categories (see Figure [1](#F1){ref-type="fig"}). Non-Aboriginal women were more likely than Aboriginal women to report the use of pain relievers and sleeping pills.
### Morbidity
Women were twice as likely as men to have benzodiazepines prescribed to them for \"non-clinical\" symptoms, \[[@B54]\] such as stress from work or home life, grief, acute or chronic illness, physical pain or adjustment to a major life change, \[[@B55]-[@B57]\] and to have them prescribed for longer periods. \[[@B58]\] It is becoming clear that women are over-prescribed benzodiazepines to cope with difficult life circumstances rather than to relieve severe clinical symptoms. Long-term benzodiazepine use is associated with several negative health consequences. The sedative/hypnotic effects of these drugs place elderly individuals at increased risk of psychomotor, cognitive and memory impairment; \[[@B59]\] emotional clouding; violent behaviour; depression; \[[@B60]\] and hip and femur fractures. \[[@B61],[@B62]\]
Illicit Drug Use
----------------
### Prevalence
Overall, self-reported illicit drug use by females in Canada is low (see Figure [1](#F1){ref-type="fig"}). Data from the 1994 CAODS \[[@B52]\] show that the proportion of females aged 15 and older reporting use of selected illicit drugs in the previous year was 5.1% for cannabis, 0.7% for LSD, speed or heroin, and 0.5% for cocaine. When these proportions are compared with data on lifetime use of illicit drugs from the same survey, the rates of use in each of the three categories decrease slightly: 18.7% for cannabis, 3.6% for LSD, speed or heroin, and 2.7% for cocaine. More recent data (1998 Canadian Campus Survey \[[@B53]\]) reveal comparatively high rates of illicit drug use among university students (although these figures do not represent the use among Canadians in general): 8.9% of female students reported use of illicit drugs in the previous 12 months (not including cannabis) and 28.0% reported cannabis use. Illicit drug use by males was higher than by females inall categories.
Injecting drug use (IDU) provides a good illustration of the need for a gender-based analysis of illicit drug use. Surveillance data from the Centre for Infectious Disease Prevention and Control, Health Canada, \[[@B16]\] reveal that up to December 31, 2000, IDU was identified as the risk factor in 1,123 reported positive HIV cases in women and in 2,570 cases in men. Figure [3](#F3){ref-type="fig"} shows that, among both females and males, there has been a decrease over time in the number of positive tests attributable to IDU. However, IDU is a greater risk for females, accounting for an average of 45.7% of all female HIV cases and only 25.6% of all male cases from 1995 to 2000. Over this six-year period there was a substantial decrease of IDU as a risk factor for HIV among females (from 53.6% cases in 1995 to 38.5% in 2000), whereas among men the rate remained relatively stable (from 23.6% cases in 1995 to 22.4% in 2000).
![Number of Positive HIV Test Reports Attributable to Injecting Drug Use (IDU) among Adult Females and Males 15 Years of Age and Older, by Year of Test \[[@B16]\]](1472-6874-4-S1-S8-3){#F3}
There are extremely limited Canadian data on solvent use. Information from the CAODS shows that 0.3% of females aged 15 and older reported use of a solvent(s) throughout their lifetime. The Ontario Student Drug Use Survey \[[@B63]\] reported that between 1977 and 2001, the average prevalence of glue sniffing with the intention of getting high among females during the previous 12 months was 2.1%; the percentage was highest in 1999 (3.7%) and lowest in 1991 (1.1%).
### Morbidity
From their analysis of hospital separations attributed (fully and partially) to illicit drugs, Single and colleagues estimated that 2,405 women (4,542 men) were hospitalized in 1995--1996 because of illicit drugs. \[[@B5]\] The hospitalizations for women were a result of drug psychoses (768), followed by cocaine dependence or abuse (375), psychotropic poisoning (323), and opioid dependence/abuse (314). The number of hospital separations per 100,000 population has remained fairly stable between 1991--1992 and 1995--1996 for women and men.
### Mortality
According to Single and colleagues\' analysis of 1995 CIHI data, the number of deaths attributable to illicit drugs was estimated at 108 for hospitalized women or 0.7 per 100,000 population (695 for men or 4.8 per 100,000 population). The greatest number of deaths resulted from opiate poisoning (42), poisoning (intent undetermined) (23), AIDS (18), suicide/self-inflicted injury (15) and cocaine poisoning (11). The potential years of life lost in 1995 were estimated at 4,959 for women or 33.6 per 100,000 population (28,710 for men or 198.1 per 100,000 population).
Mental Health, Trauma and Substance Use
---------------------------------------
### Mental Health and Substance Use
Of the women meeting the criterion for heavy drinking, 25.7% reported feeling sad, blue or depressed in the previous two weeks. Depression can be a precursor or antecedent of problem alcohol use, or both. Roughly 15% of heavy drinking women also reported visiting a mental health professional in the previous 12 months. Further, 18.6% of them reported that they had previously seriously considered committing suicide. Of the women who had considered suicide in the previous 12 months, 37.3% had previously attempted to take their own life.
### Trauma and Substance Use
Research shows high rates of victimization among substance-abusing women. \[[@B15]\] Data collected from substance use treatment centres in Canada corroborate these estimates. In a sample of substance-abusing women from nine treatment centres across Ontario, Cormier \[[@B64]\] found that 85.7% of the 98 women in her sample had been victimized. They were most likely to have reported being a victim of adult physical abuse (56.1%), childhood sexual abuse (56.3%), childhood physical abuse (56.1%) and adult sexual abuse (45.4%). A treatment facility for substance-abusing women in Vancouver (Aurora Centre) reported that 65% of its female clients had been physically assaulted as adults, and 38% had been sexually assaulted; 47% reported physical violence in childhood, and 53% reported sexual abuse at that time. \[[@B47]\]
Discussion
==========
Data Limitations
----------------
The past quarter-century has been characterized by inconsistent survey data collection on substance use and abuse at the national, provincial/territorial and municipal levels. Further, there has been limited recognition in the research that social factors specific to sex and gender (e.g. employment, cultural displacement, abuse) have a significant role in substance use problems. An overarching limitation of most surveys and research studies is the marked discrepancy in definitions of substance use, dependence, or problematic use.
In comparison with current information on alcohol use and abuse in Canada, there is extremely limited knowledge of illicit and licit drug use and, in particular, the associated morbidity and mortality. Gender-specific research on prescription and non-prescription drug use is required, since females consistently report higher rates of use than males, thus increasing the probability of problematic use. Health Canada and pharmaceutical company data (e.g. PURDUE Pharma) are alternative sources of information on the dispensing practices of physicians. In addition, there has been collection of drug plan utilization data at the provincial level (e.g. New Brunswick, Saskatchewan) that could be replicated. Further, while the first cycle of the CCHS Cycle 1.1 did not include questions about illicit drug use, the most recent cycle of the survey (CCHS Cycle 1.2) will be an appropriate source from which to collect this information.
There are also knowledge gaps concerning substance use and misuse, and concerning competent programming that addresses the needs of poor women, rural women, homeless women, women with disabilities, lesbians, elderly women, adolescent girls and women from different ethnic backgrounds, all of whom are often at increased risk for general health problems. The limited data available suggest that these subpopulations of women may be at greater risk of substance use problems than the general population. \[[@B65]-[@B73]\]
Several suggestions to address the current data limitations follow. Each of these requires the collection of sex- and gender-specific data.
Recommendations
===============
Incidence and Prevalence Data
-----------------------------
1\. Surveillance in the form of a national incidence/prevalence survey should be carried out on a regular basis (e.g. every three years).
2\. Canada should provide more support for its substance use monitoring systems, such as the Canadian Community Epidemiology Network on Drug Use, \[[@B74]\] so that data will be readily available and attempts to collate data in reports such as this one will not be such a daunting task. Integral to such a system must be the reporting of data by sex, attention to social and other determinants of women\'s substance use, and the exploration of substance use problems in relation to gender.
Treatment Data
--------------
3\. There is a need to develop a mechanism for capturing evaluation data on substance use interventions (including women-specific interventions at various levels of care).
4\. There is also a need to examine the availability of data collected from special populations (e.g. National Youth Solvent Abuse Program, First Nations and Inuit Health Branch).
Mortality and Morbidity Data
----------------------------
5\. Greater and affordable access to national systems for collecting and reporting information on hospitalizations is necessary (i.e. CIHI); all aspects of the standardization in data collection must be addressed (e.g. ICD-9/10 codes); and hepatitis C in addition to HIV/AIDS should be included.
Other Data Fundamental to Women-Centred Policy and Programming
--------------------------------------------------------------
6\. Central to action on women\'s substance use is addressing the misinformation and prejudice directed at women with this common health-related problem. This is essential in order to help reduce the guilt and shame that prevent women from identifying both that they use substances and that they need help. This approach should underlie all the following recommendations.
Reducing Barriers to Identifying and Preventing FAS
---------------------------------------------------
7\. There is a need to identify viable and sensitive methods for implementing screening for substance use by women on the part of a wide range of professionals in a position to make referrals to treatment and other resources when necessary. It is particularly important for this screening to be in place to reach women of child-bearing years and pregnant women, so that FAS and other alcohol- and drug-related developmental disabilities are prevented. Some surveys on alcohol and drug screening practices of physicians have been done, and such surveys can assist in promoting tailored training and action in this important area.
Understanding and Responding to Diverse Needs
---------------------------------------------
8\. Knowledge gaps need to be addressed concerning the level, type and impact of substance use and the adequacy of programs that should be reaching vulnerable subgroups of women (e.g. Aboriginal women, poor women, homeless women, lesbians, women living in rural areas). For the most part, these would require smaller-scale, local studies.
Understanding the Impact of Illicit Drug Use on Women and Harm Reduction Approaches
-----------------------------------------------------------------------------------
9\. There is a critical need for data on sex and gender differences in the experience of illicit drug use and the potential need for, and impact of, harm-reduction-oriented policy and programming.
Improving Treatment Access, Especially for Mothers
--------------------------------------------------
10\. Data on the impact of barriers to treatment are needed. Treatment programming that addresses sex and gender differences in the experience of addictions needs to be made more accessible to women in Canada. Programming that is accessible and relevant to women who are mothers is of particular priority.
Blending Treatment/Support in Substance Use, Mental Health and Trauma Concerns
------------------------------------------------------------------------------
11\. It is recommended that linkages be enhanced (and in some cases program integration be considered) between mental health treatment, substance misuse treatment and programming for women who have experienced trauma/relationship violence, in order to address the strong interconnections among these three serious health problems for women.
Notes
=====
The authors wish to acknowledge the contributions of Bette Reimer, Canadian Centre on Substance Abuse, who conducted literature searches for this chapter, and Karen Devries, who assisted with the organization of the reference section for this chapter.
| {
"pile_set_name": "PubMed Central"
} |
Background
==========
Atherosclerosis is a highly prevalent chronic inflammatory disorder that causes extensive mortality \[[@b1-medscimonit-25-8820]\], but the details of the underlying mechanism are unknown. In recent decades, accumulating evidence has revealed the essential regulatory role played by inflammation throughout atherosclerotic progression \[[@b2-medscimonit-25-8820]\]. Moreover, proinflammatory tumor necrosis factor-α (TNF-α) promotes atherosclerosis progression and subsequent cardiovascular events via induction of inflammatory responses \[[@b3-medscimonit-25-8820]\]. TNF-α is a typical member of the cytokine family, which promotes inflammation, differentiation, and apoptosis of endothelial cells (ECs). Atherosclerosis is increasingly thought to be an age-associated chronic disorder, and endothelial dysfunction always precedes the development of atherosclerosis. Endothelial senescence is an important mediator of vascular inflammation and dysfunction \[[@b4-medscimonit-25-8820],[@b5-medscimonit-25-8820]\].
Sirtuins, a group of NAD+-dependent class III histone/protein deacetylases (HDACs), are involved in essential processes such as proliferation, apoptosis, senescence, and differentiation \[[@b6-medscimonit-25-8820],[@b7-medscimonit-25-8820]\]. Sirtuin1 (SIRT1) is the most extensively characterized and best-studied member of the sirtuin family. SIRT1 acts as a regulatory switch in vascular endothelial cells homoeostasis \[[@b8-medscimonit-25-8820],[@b9-medscimonit-25-8820]\] and is critically involved in regulating inflammation-related disorders via inhibition of proinflammatory cytokine secretion \[[@b10-medscimonit-25-8820]--[@b12-medscimonit-25-8820]\]. Previous studies have demonstrated that hypoxia, lipopolysaccharide (LPS), TNF-α, and high glucose decreased SIRT1 expression in ECs \[[@b9-medscimonit-25-8820],[@b10-medscimonit-25-8820],[@b13-medscimonit-25-8820]\]. SIRT1 expression is modulated at diverse levels, and previous studies have shown that post-transcriptional regulation is important in modifying SIRT1 expression. Previous reports have shown that the SIRT1 3′-untranslated region (3′ UTR) is critical for its mRNA stability. Additionally, a series of microRNAs (miRNAs), such as miR-34a, miR-23b-3p, and miR-132-3p, modulate SIRT1 expression \[[@b9-medscimonit-25-8820],[@b14-medscimonit-25-8820],[@b15-medscimonit-25-8820]\].
miRNAs are small noncoding RNAs (containing approximately 22 nucleotides) responsible for post-transcriptional modulation of gene expression and suppression of mRNA translation and stability, typically through binding to the 3′ UTR of target mRNAs \[[@b16-medscimonit-25-8820]\]. Several miRNAs have been found to exert a vital role in cardiac and endothelial function, which is associated with cardiovascular disease progression \[[@b17-medscimonit-25-8820],[@b18-medscimonit-25-8820]\]. miR-155, a typical miRNA with multiple functions, was recently identified as an element of the inflammatory signaling pathway in atherosclerosis \[[@b19-medscimonit-25-8820],[@b20-medscimonit-25-8820]\]. miR-155 regulates gene expression associated with inflammation in various cell types *in vitro* and affects atherogenesis *in vivo* \[[@b21-medscimonit-25-8820]--[@b26-medscimonit-25-8820]\]. miR-155 can enhance or suppress lesion formation at a range of atherosclerotic stages. Therefore, miR-155 may be linked to the pathogenesis of atherosclerosis \[[@b27-medscimonit-25-8820]\].
Recent evidence has demonstrated that some clinical drugs can modulate miRNA expression, suggesting that miRNAs may be therapeutic targets of these agents \[[@b9-medscimonit-25-8820]\]. Statins and sartan are widely applied to treat dyslipidemia and related vascular disorders. However, whether valsartan and simvastatin modulate the effects of proinflammatory cytokines on endothelial function is largely unknown.
Despite the acknowledged importance of SIRT1 and miR-155 in the modulation of endothelial function, the relationship between them and the regulatory effects of inflammation on endothelial function during atherosclerosis remain unclear. Therefore, this study was designed to assess the possible association between SIRT1 and miR-155 and to determine how miR-155 regulates cardiovascular homeostasis. We found that simvastatin or valsartan can ameliorate TNF-α-induced HUVEC senescence through inhibition of miR-155.
Material and Methods
====================
Endothelial cell culture
------------------------
Human umbilical vein endothelial cells (HUVECs) (ScienCell, CA, USA) were cultured in endothelial cell medium (ECM) (ScienCell) containing 5% FBS, 1% endothelial cell growth supplement (ECGS) (ScienCell), and penicillin/streptomycin and were cultured at 37˚C with 5% CO~2~ and 21% O~2~. ECs were seeded at a proper density in accordance with each experimental procedure. HUVECs were treated with ±TNF-α (20 ng/ml) (PeproTech, USA).
miR-155 inhibition and overexpression
-------------------------------------
To inhibit miR-155 expression, HUVECs at 50%--70% confluency were transfected using miR-155 inhibitor (GenePharma, Shanghai, China) or miRNA inhibitor negative control (NC) (50 nmol/l) for 36 h using Lipofectamine 3000 (Invitrogen, Life Technologies, NY, USA) in Opti-MEM, following the manufacturer's instructions. After transfection, Opti-MEM was exchanged for medium containing TNF-α, and cells were cultured for another 12 h. HUVECs were transfected using miR-155 mimic (GenePharma) to overexpress miR-155, or with NC mimic (50 nmol/l) for 48 h prior to additional analysis.
Western blotting
----------------
After washing, cold RIPA with protease inhibitors (Roche) was used to lyse cells for 15 min. Then, after centrifugation, the cell supernatant was kept frozen at −80°C until use. Bradford assays (Bio-Rad Laboratories; CA, USA) were employed to assess protein concentrations. Samples (20 μg) were separated via SDS-PAGE, after which dry transfer to nitrocellulose membranes was conducted. After being blocked in 5% BSA for 1 h, primary antibodies were used to probe the membranes overnight at 4°C, including antibodies targeting human SIRT1, acetylated-p53 at K382 (Ac-p53), p21, forkhead box O1 (FoxO-1) (Cell Signaling Technology; USA), Ac-FoxO-1, total p53 (1: 2000 dilution), and senescence marker protein-30 (SMP-30) (all from Santa Cruz Biotechnology). β-actin served as an internal control and was purchased from Sigma Aldrich. After incubation with the appropriate secondary antibodies (1: 10 000 dilution; GE Healthcare, Buckinghamshire, UK), the immunostained protein bands were visualized with an ECL system (ProteinSimple; Santa Clara, USA), followed by quantification using ProteinSimple image software. All samples had 3 biological replicates, and each biological replicate was assayed in duplicate; therefore, the data for each time point are an average of 6 individual replicate runs. Representative images of the immunostained bands are presented in the figures.
qPCR
----
TRIzol (Invitrogen) was used for EC total RNA extraction in accordance with standard protocols. A Nanodrop 2000 spectrophotometer (Invitrogen) was employed to determine RNA concentration and purity. cDNA was produced from 500 ng of total RNA utilizing a qScript microRNA Quantification System kit (microRNAs; Quanta Biosciences) or PrimeScript RT Master Mix (mRNAs; TaKaRa) in line with standard protocols. qPCR was conducted utilizing PerfeCTa SYBR Green SuperMix (microRNAs; Quanta Biosciences) and Power SYBR^®^ Green PCR Master Mix (mRNAs; Applied Biosystems; USA) using an ABI 7500 machine (Applied Biosystems). The primer sequences were commercially obtained from Invitrogen and were as follows (forward and reverse, respectively):
1. SIRT1, 5′-GCT GTG AAA TTA CTG CAA GAG TG-3′ and
2. 5′-AAT ACC ATC CCT TGA CCT GAA G -3′;
3. CYPA, 5′-TTG CTG TTC CTT AGA ATT TTG CC-3′ and
4. 5′-ACC CTG ACA CAT AAA CCC TG -3′;
5. miR-155, 5′-TTA ATG CTA ATC GTG ATA GGG GT-3′;
6. U6, 5′-GAT GAC ACG CAA ATT CGT G-3′.
mRNA and miRNA expression were normalized to CYPA and U6, respectively. The ΔΔCt method was utilized to calculate fold changes in expression. All mRNA and miRNA samples had 3 biological replicates, and each biological replicate was also assayed in triplicate; therefore, the data for each time point are an average of 9 individual replicate runs.
Luciferase reporter assay
-------------------------
Firefly luciferase cDNA fused with the amplified 3′ UTR of human SIRT1 from a genomic DNA sample was cloned into psiCHECK-2 (Promega, USA). Luciferase in this vector was replaced by the produced wild-type (WT) and mutant 3′ UTR of SIRT1. HUVECs were added to 24-well plates, grown to 70--80% confluency, and then transfected using psiCHECK-2 that had been cloned so as to contain the WT or mutant SIRT1 3′ UTR (300 ng/well) using Lipofectamine 3000 in line with standard protocols. Following co-transfection using miR-155 mimic or NC mimic (50 nmol/l) for 48 h, a dual-luciferase assay system (Promega) was utilized, with firefly luciferase normalized to activity from Renilla luciferase for a control to determine transfection efficiency.
Senescence assessment
---------------------
Senescence-associated β-galactosidase (SA β-Gal) activities were quantified in cells that had been fixed using formaldehyde following standard protocols (Sigma Aldrich). In brief, after PBS washing, cells underwent a 6-min fixation step followed by washing again in PBS. Next, staining buffer was added to wells overnight at 37°C. Stained cells were then imaged via microscopy (DM IL LED, Leica). Blue (positive) cells versus total cells in every field were determined by ImageJ software (NIH). The data are fold changes in activity relative to controls.
Cell proliferation assay (BrdU assay)
-------------------------------------
A Cell Proliferation ELISA kit (Roche) was used to facilitate BrdU incorporation during DNA synthesis to measure proliferation. To do so, we added 10 μl of BrdU labeling solution per sample in 100 μl of culture medium in 96-well plates. After 2 h at 37°C, FixDenat (200 uL) was administered per well and then discarded after 30 min at 25°C. Next, 100 μl per well of anti-BrdU-POD was added for 90 min at 25°C. Following 3 washes and the addition of 100 μl of substrate solution, absorbance was assessed at 370 nm (reference: 490 nm) after 20 min at 25°C. The values are presented as the fold absorbance difference relative to that of the control group.
Cell viability assay
--------------------
We used CellTiter 96 AQueous One Solution (Promega) to assess the number of viable cells. This solution consists of MTT together with a reagent for electron coupling (PES). In brief, 20 μl of this compound was added per sample in 100 μl of media in 96-well plates for 1 h at 37°C. Absorbance was then determined at 490 nm (reference wavelength, 630 nm). The results are presented as the fold absorbance difference relative to that of the control group.
Statistical analysis
--------------------
GraphPad Prism6 (CA, USA) was used for statistical testing. Data are means±SEM. Comparison of multiple groups or 2 groups was performed via one-way ANOVA and *t* tests, respectively. *P*\<0.05 was the significance threshold. All experimental procedures were conducted in triplicate or more.
Results
=======
TNF-α increased miR-155 expression and reduced SIRT1 expression in HUVECs
-------------------------------------------------------------------------
miR-155 is a frequently analyzed miRNA in the progression of atherosclerosis. To assess whether miR-155 participates in the effects of specific inflammatory stimuli, HUVECs were exposed to TNF-α (0, 10, 20, or 40 ng/ml) for 8 h, followed by qPCR to measure miRNA expression. Compared with NG, treatment with TNF-α from 10 to 40 ng/ml mediated dose-dependent increases in miR-155 expression by more than 2-fold ([Figure 1A](#f1-medscimonit-25-8820){ref-type="fig"}). We further performed a time-response assay of miR-155 expression in TNF-α-treated HUVECs. Specifically, miR-155 expression in HUVECs was determined after administration with TNF-α (20 ng/mL) for 0, 1, 2, 4, 6, and 8 h, and the results revealed that TNF-α (20 ng/ml) significantly enhanced miR-155 expression. This was apparent at 2 h, and expression peaked at 6 h ([Figure 1B](#f1-medscimonit-25-8820){ref-type="fig"}). Moreover, miR-155 expression was also triggered by LPS, an inflammatory cytokine ([Figure 1C](#f1-medscimonit-25-8820){ref-type="fig"}).
Following the above-described results, we further determined the effects of TNF-α on SIRT1 expression. HUVECs were administered TNF-α (20 ng/ml) for 8 h, followed by WB analysis and qPCR to assess the SIRT1 protein and mRNA levels, respectively. SIRT1 protein expression in HUVECs was reduced by TNF-α at 20 ng/ml ([Figure 1D](#f1-medscimonit-25-8820){ref-type="fig"}).
miR-155 regulated TNF-α-mediated SIRT1 expression in HUVECs
-----------------------------------------------------------
TNF-α bolstered miR-155 expression in addition to having reduced SIRT1 expression, which suggested that miR-155 was likely to contribute to TNF-α-mediated SIRT1 downregulation. To verify this hypothesis, we examined whether miR-155 could decrease SIRT1 expression. Following HUVEC miR-155 mimic transfection for 48 h, the SIRT1 mRNA and protein levels were assessed by qPCR and WB, respectively. miR-155 mimic transfection significantly enhanced miR-155 expression ([Figure 2A](#f2-medscimonit-25-8820){ref-type="fig"}), demonstrating that upregulation of miR-155 in HUVECs reduced the SIRT1 protein level by nearly 50% ([Figure 2B](#f2-medscimonit-25-8820){ref-type="fig"}). Therefore, miR-155 inhibited SIRT1 expression in HUVECs.
To confirm whether miR-155 inhibition attenuated TNF-α-mediated downregulation of SIRT1 expression, we transfected HUVECs with miR-155 inhibitor for 36 h, after which stimulation with TNF-α for 12 h was performed. Then, qPCR and WB were utilized to determine SIRT1 mRNA and protein levels, respectively. Transfection with miR-155 inhibitor induced a significantly decrease in miR-155 expression, showing that inhibition of miR-155 in HUVECs significantly upregulated SIRT1 protein expression by more than 30% ([Figure 2C](#f2-medscimonit-25-8820){ref-type="fig"}). Moreover, we transfected cells with miR-155 inhibitor under basal conditions and observed a significant increase in SIRT1 expression ([Figure 2D](#f2-medscimonit-25-8820){ref-type="fig"}), indicating that miR-155 modulated SIRT1 expression under physiological and pathological conditions. These results confirmed that upregulation of miR-155 in HUVECs suppressed SIRT1 expression, while downregulation of miR-155 enhanced SIRT1 expression, suggesting that SIRT1 may be a target of miR-155.
miR-155 directly targets SIRT1
------------------------------
Our present results provided a strong rationale to test the hypothesis that miR-155 directly targets SIRT1. miRNAs inhibit the translation initiation step via binding to target mRNA 3′ UTR. Therefore, the online prediction tool TargetScan was employed to recognize putative miR-155 targets. Bioinformatics analysis showed that SIRT1 was a miR-155 target, suggesting this sequence (perfect match) might be a molecular target of miR-155 ([Figure 3A](#f3-medscimonit-25-8820){ref-type="fig"}). According to the bioinformatics analysis, 2 PCR product fragments (F1-R1 and mut F1-R1 T1) were cloned into a psiCHECK-2 vector ([Figure 3A](#f3-medscimonit-25-8820){ref-type="fig"}). Unfortunately, based on the results of luciferase reporter assays, co-transfection with miR-155 mimic decreased the luciferase activity in both the FI-R1 and mut F1-R1 T1 groups ([Figure 3B, 3C](#f3-medscimonit-25-8820){ref-type="fig"}). Thus, we failed to confirm our hypothesis. These findings indicated that this sequence (perfect match) is not a molecular target of miR-155 ([Figure 3A](#f3-medscimonit-25-8820){ref-type="fig"}).
Previous work has demonstrated that if miRNA and target mRNA are an exact or nearly exact complementary match, target mRNA can be degraded. However, translational repression occurs when there is partial (imperfectly matched) complementarity between miRNA and target mRNA \[[@b20-medscimonit-25-8820]\]. Therefore, new potential target sequences were predicted that had a mismatch base pairing in the seed sequence ([Figure 3D](#f3-medscimonit-25-8820){ref-type="fig"}). In addition, we identified 3 other possible sites and named these sites with or without a mismatch target sequence 1 to 4 (T1--T4, where T1 is the perfect match mentioned above). Subsequently, we evaluated the sequences T1 to T4 ([Figure 3E](#f3-medscimonit-25-8820){ref-type="fig"}). Based on these potential target sequences, we cloned the F5-R2 PCR product into the psiCHECK-2 vector and detected no difference between the miR-155 mimic and NC group, as indicated by luciferase reporter assays ([Figure 3F](#f3-medscimonit-25-8820){ref-type="fig"}). These findings suggested that the target sequence may be in the F1-R1 sites. Thus, we designed different forward primers, and the amplified sequences (F2-R1, F3-R1, F4-R1, and F5-R1) were cloned into the psiCHECK-2 vector. As shown in [Figure 4A--4D](#f4-medscimonit-25-8820){ref-type="fig"}, as the cloning sequence become closer to R1, the results of the luciferase reporter assay changed from positive to negative, suggesting that the real target sequence was between the F3-R1 and F4-R1 fragment (the T2 sequence). Therefore, we generated a T2 sequence mutant ([Figure 4E](#f4-medscimonit-25-8820){ref-type="fig"}). According to the result of luciferase reporter assays ([Figure 4F](#f4-medscimonit-25-8820){ref-type="fig"}), co-transfection of cells the WT SIRT1 3′ UTR T2 plasmid with the miR-155 mimic decreased relative luciferase activity. Furthermore, miR-155 did not repress the activity of the luciferase reporter with a mutant SIRT1 3′ UTR T2 plasmid, and these differences were all significant (*P*\<0.05). Overall, these data provided experimental evidence that SIRT1 is directly targeted by miR-155.
Effects of miR-155 on HUVEC proliferation and senescence
--------------------------------------------------------
BrdU and MTT assays were used to evaluate proliferation of HUVECs, which demonstrated that proliferation of HUVECs was significantly inhibited by miR155 expression ([Figure 5A, 5B](#f5-medscimonit-25-8820){ref-type="fig"}). Then, HUVECs underwent transfection with miR-155 mimic to observe whether upregulation of miR-155 could reduce cell proliferation. A significant effect on cell proliferation was observed in transfected HUVECs compared to the effect on the control cells after incubation for 48 h ([Figure 5C, 5D](#f5-medscimonit-25-8820){ref-type="fig"}). We confirmed that miR-155 downregulation increased proliferation of HUVECs. As expected, the results showed that suppression of miR-155 significantly promoted TNF-α-induced HUVEC proliferation ([Figure 5E, 5F](#f5-medscimonit-25-8820){ref-type="fig"}).
SIRT1 is an anti-aging gene that is has a vital role in preventing endothelial senescence. The results revealed that SIRT1 was directly targeted by miR-155, and TNF-α could upregulate miR-155 expression, implying that miR-155 might participate in senescence promotion. Thus, determination of cellular senescence of HUVECs was conducted using 2 indicators -- anti-SMP-30 and SA β-Gal staining -- and TNF-α significantly reduced SMP-30 protein expression by almost 50% and enhanced SA β-Gal staining by nearly 2-fold ([Figure 6A](#f6-medscimonit-25-8820){ref-type="fig"}). Then, the role of miR-155 in senescence of HUVECs was assessed, and the results showed that miR-155 mimic transfection increased SA β-Gal staining and downregulated SMP-30 ([Figure 6B](#f6-medscimonit-25-8820){ref-type="fig"}). Moreover, miR-155 was suppressed to verify reversal of the increased senescence. Briefly, after transfection of HUVECs with miR-155 inhibitor, the senescence of TNF-α-stimulated cells was examined as described above. Addition of the miR-155 inhibitor decreased SA β-Gal staining and increased SMP-30 protein levels ([Figure 6C](#f6-medscimonit-25-8820){ref-type="fig"}). Thus, we concluded that miR-155 promoted senescence in TNF-α-induced HUVECs.
miR-155 regulated HUVECs proliferation and senescence through SIRT1/FoxO-1/p53/p21 signaling
--------------------------------------------------------------------------------------------
We next investigated the signaling pathway by which miR-155 regulated HUVECs proliferation and senescence. SIRT1 siRNA transfection decreased the protein level of SIRT1, confirming the silencing effect of SIRT1 siRNA on SIRT1 ([Figure 7A](#f7-medscimonit-25-8820){ref-type="fig"}). SIRT1, a deacetylase for multiple transcription factors, including FoxO-1 and p53, has been demonstrated to modulate critical cellular processes, including proliferation, senescence, differentiation, and apoptosis. Thus, the downstream signaling targets of SIRT1, including the protein levels of acetylated FoxO-1, acetylated p53, and p21, were also determined. Knockdown of miR-155 decreased p21, Ac-p53, and Ac-FoxO-1 protein expression, and SIRT1 silencing reversed this effect ([Figure 7B](#f7-medscimonit-25-8820){ref-type="fig"}).
Then, we assessed whether HUVEC proliferation and senescence were influenced by siSIRT1, and the results showed that siSIRT1 prevented the decrease in TNF-α-induced cell proliferation triggered by miR-155 ([Figure 8A](#f8-medscimonit-25-8820){ref-type="fig"}). Similarly, the effect of miR-155 on the upregulation of senescence and SMP-30 expression were substantially decreased by SIRT1 knockdown ([Figure 8B](#f8-medscimonit-25-8820){ref-type="fig"}). These outcomes indicated that miR-155 regulated HUVEC proliferation and senescence through the SIRT1/FoxO-1/p53/p21 pathway.
Valsartan and simvastatin ameliorated TNF-α-induced HUVEC senescence through inhibition of miR-155
--------------------------------------------------------------------------------------------------
HUVECs were pretreated with or without valsartan (40 μmol/L) or simvastatin (1 μmol/L) for 3 h, followed by subsequent stimulation with TNF-α for 8 h. WB analysis and qPCR were employed to determine expression changes in miR-155 and SIRT1. Our results demonstrated significantly enhanced miR-155 expression in HUVECs exposed to TNF-α, accompanied by a parallel reduction in SIRT1 levels, as previously shown. When HUVECs were treated with TNF-α together with valsartan or simvastatin, miR-155 expression was significantly reduced, accompanied by a parallel upregulation in SIRT1 expression ([Figure 9A, 9B](#f9-medscimonit-25-8820){ref-type="fig"}) and a decrease in TNF-α-induced HUVEC senescence ([Figure 9C](#f9-medscimonit-25-8820){ref-type="fig"}). In addition, we examined the effect of valsartan or simvastatin in HUVECs overexpressing miR-155 to confirm whether the endothelial-protective function of valsartan or simvastatin depended on miR-155. Transfection of HUVECs with miR-155 mimic and exposure to TNF-α with valsartan or simvastatin significantly increased miR-155 expression. However, transfection with miR-155 mimic negated the protective effects of valsartan or simvastatin on SIRT1 and senescence in TNF-α-induced HUVECs ([Figure 10A--10C](#f10-medscimonit-25-8820){ref-type="fig"}). Overall, downregulation of miR-155 expression mediated the protective effects of valsartan and simvastatin on endothelial dysfunction.
Discussion
==========
Here, we found TNF-α significantly enhanced miR-155 expression and decreased SIRT1 expression. Moreover, SIRT1 was directly targeted by miR-155. In addition, miR-155 regulated endothelial cell senescence and proliferation via the SIRT1/FoxO-1/p53/p21 pathway. Overexpression of miR-155 contributed to the TNF-α-triggered reduction in SIRT1 expression and impairment of endothelial senescence. Moreover, anti-miR-155 inhibitor partially reduced TNF-α-induced SIRT1 suppression. Functional j ju showed that miR-155 inhibitor reduced TNF-α-mediated endothelial senescence and bolstered cellular proliferation. It was also verified that miR-155 inhibitor enhanced SIRT1 expression under basal conditions, suggesting that miR-155 can control SIRT1 expression under physiological conditions. Moreover, the presence of simvastatin or valsartan restored SIRT1 expression and inhibited senescence via suppression of miR-155 in TNF-α-treated HUVECs. However, miR-155 mimic prevented simvastatin- or valsartan-mediated protection. These outcomes suggested that miR-155 might be a valuable intervention target to prevent cardiovascular disease ([Figure 11](#f11-medscimonit-25-8820){ref-type="fig"}).
miR-155 directly targets SIRT1
------------------------------
This study is the first to identify that SIRT1 is a miR-155 target, which promotes TNF-α-induced senescence via the SIRT1/FoxO-1/p53/p21 pathway in HUVECs, and is also the first study to demonstrate that valsartan modulates the effects of proinflammatory cytokines on endothelial function.
Accumulating evidence has suggested a vital role of miR-155 in regulating a variety of cellular functions, including inflammation, hematopoiesis, immunological response, and viral infection \[[@b22-medscimonit-25-8820],[@b23-medscimonit-25-8820],[@b25-medscimonit-25-8820]\]. A recent report showed significant downregulation of the circulating level of miR-155 in people with coronary artery disease \[[@b28-medscimonit-25-8820]\]. Previous studies demonstrated that miR-155 can enhance or suppress lesion formation, based on atherosclerosis stage \[[@b27-medscimonit-25-8820]\]. Furthermore, previous data have indicated that miR-155 is critically involved in inflammation and is potentially correlated with atherosclerosis pathogenesis in numerous cell types \[[@b2-medscimonit-25-8820],[@b27-medscimonit-25-8820]\]. Together, these studies demonstrated that miR-155 might be a vital modulator of cardiovascular disease processes.
miR-155, a typical miRNA with multiple functions, simultaneously modulates numerous biological process by targeting diverse genes \[[@b20-medscimonit-25-8820]\]. Thus, luciferase reporter assays have identified several direct miR-155 targets, such as helper T 1 (Th1), Sma- and Mad-related protein 2 (SMAD2), secretogranin II (SCG2), SH2-containing inositol 5′-phosphatase 1 (SHIP1), and SOCS1 \[[@b2-medscimonit-25-8820],[@b29-medscimonit-25-8820]\]. Moreover, other research groups have found that eNOS and NF-κB are direct miR-155 targets in ECs \[[@b30-medscimonit-25-8820]\]. Taken together, the results show that miR-155 might exert a conflicting function in endothelial cells.
This is the first study showing that miR-155 targets SIRT1. miR-155 downregulated SIRT1 primarily via SIRT1 3′ UTR binding. First, the online prediction tool TargetScan indicated that miR-155 may target SIRT1. However, the luciferase reporter assays indicated that this sequence, which was a perfect match, was not a molecular target of miR-155. Previous studies have shown that when a miRNA and target mRNA are exactly or almost exactly complementary to one another, degradation of target mRNA will occur. However, translation repression occurs only in the case of partial (imperfectly matched) complementarity between miRNA and target mRNA \[[@b20-medscimonit-25-8820]\]. Therefore, we predicted new potential target sequences that had a mismatch base pairing in the seed sequence. To determine the general position of the sequence, these mismatch sequences were cloned into cells. The luciferase reporter assays indicated that one of these mismatch sequences is the real target binding sequence of miR-155. Although a previous \[[@b31-medscimonit-25-8820]\] study reported that SIRT1 may be a direct target of miR-155, the authors revealed that miR-155 downregulated SIRT1 expression by binding to 3 sites on the SIRT1 3′ UTR. The sequences in the previous study and our study were different. Moreover, our study was performed in different cells.
SIRT1, referred to as an anti-aging gene, is critical for preventing endothelial senescence. Accumulating evidence indicates that senescence of the endothelium plays an important role in promoting vascular inflammation and dysfunction. Further evidence has suggested that the SIRT1 3′ UTR is critical for SIRT1 mRNA stability. Additionally, a series of miRNAs, including miR-132-3p, miR-23b-3p, and miR-34a, were found to modulate SIRT1 expression \[[@b9-medscimonit-25-8820],[@b14-medscimonit-25-8820],[@b15-medscimonit-25-8820]\]. Our results demonstrated that miR-155 upregulation inhibited SIRT1 expression and HUVEC proliferation and promoted TNF-α-induced senescence in HUVECs. The data showed that miR-155 reduced the protein level of SIRT1 but had no effect on the mRNA level. These findings indicate that SIRT1 expression is reduced by miR-155 via translational repression, not degradation, of the target mRNA. Furthermore, inhibition of miR-155 attenuated TNF-α-triggered downregulation of SIRT1 expression and decreased TNF-α-induced senescence in HUVECs. Interestingly, we also found that transfection with a miR-155 inhibitor enhanced SIRT1 expression under basal conditions, indicating that miR-155 regulates SIRT1 expression under physiological and pathological conditions. Moreover, the data revealed that miR-155 regulates HUVEC proliferation and senescence through the SIRT1/FoxO-1/p53/p21 pathway.
This indicates that miR-155 plays a key role in cardiovascular homeostasis regulation via the inhibition of SIRT1 expression.
Valsartan and simvastatin ameliorated TNF-α-induced HUVEC senescence through inhibition of miR-155 expression
-------------------------------------------------------------------------------------------------------------
Recent growing evidence has demonstrated that some clinical drugs can modulate miRNA expression, suggesting that miRNAs may be therapeutic targets of these agents \[[@b9-medscimonit-25-8820]\]. Statins are widely applied to treat dyslipidemia and relevant vascular disorders. *In vitro* and *in vivo* studies has revealed statins to have pleiotropic actions, including maintenance of plaque stability, induction of increased NO availability, anti-platelet effects, reduction of inflammatory effects, and immune regulation, in addition to their anti-fibrotic and anti-oxidant properties and lipid-lowering effects \[[@b32-medscimonit-25-8820],[@b33-medscimonit-25-8820]\]. Previous studies have reported that simvastatin ameliorated TNF-α-induced endothelial vasorelaxation through inhibition of miR-155 expression via mechanisms involving upregulation of eNOS and its downstream signaling pathway in HUVECs \[[@b29-medscimonit-25-8820]\]. Another previous study reported that SIRT1 expression was reduced by TNF-α and enhanced by simvastatin \[[@b34-medscimonit-25-8820]\]. Nevertheless, whether miR-155 modulates the protective roles of simvastatin on SIRT1 expression and endothelial cell function following treatment with TNF-α remains unknown. Sartan, an orally administered selective antagonist of angiotensin II (Ang II) receptor blockers (ARBs), is approved for treating hypertension. Previous experimental studies have demonstrated that telmisartan, an ARB, modulates endothelial inflammation induced by inflammatory stimuli in HUVECs \[[@b35-medscimonit-25-8820]\]. Moreover, effects similar to those of telmisartan were achieved with 2 lipophilic ARBs -- olmesartan and candesartan. However, whether valsartan modulates the effects of proinflammatory cytokines on endothelial function is largely unknown.
To investigate this potential relationship, we pretreated HUVECs with valsartan or simvastatin. To the best of our knowledge, the present study is the first to demonstrate that valsartan attenuates the inflammatory process induced by TNF-α. Here, our results showed that simvastatin or valsartan decreased miR-155 expression, accompanied by a parallel upregulation in SIRT1 expression and a decrease in TNF-α-induced HUVEC senescence. After transfection of miR-155 mimics, simvastatin or valsartan led to a negation of the protective roles of SIRT1 expression, as well as senescence, in TNF-α-induced HUVECs. The above findings indicate that downregulation of miR-155 expression mediated the protective effects of simvastatin or valsartan on endothelial dysfunction. Thus, simvastatin or valsartan promoted SIRT1 expression via the downregulation of miR-155 expression. Hence, these results might supply novel insights into the potential mechanism of the anti-senescence effect of simvastatin or valsartan in ECs.
Conclusions
===========
We showed that miR-155 mediates TNF-α-induced endothelial senescence via suppressing SIRT1. Simvastatin disrupted this through altering miR-155/SIRT1/FoxO-1/p21 pathway signaling. As such, miR-155 may be a viable therapeutic approach for endothelial senescence during the development of cardiovascular diseases.
**Source of support:** This study was supported by grants from the National Natural Science Foundation of China (NSFC) 81700405 and 51890894 to E. Zhang, and 81970326 and 51890891 to Y. Wu. Q. Guo was supported by Chinese Scholarship Council (CSC) scholarship 201706210415 and China Postdoctoral Science Foundation 2019M650032
{#f1-medscimonit-25-8820}
{#f2-medscimonit-25-8820}
{#f3-medscimonit-25-8820}
{#f4-medscimonit-25-8820}
{#f5-medscimonit-25-8820}
{#f6-medscimonit-25-8820}
{#f7-medscimonit-25-8820}
{#f8-medscimonit-25-8820}
{#f9-medscimonit-25-8820}
{#f10-medscimonit-25-8820}
{#f11-medscimonit-25-8820}
[^1]: Study Design
[^2]: Data Collection
[^3]: Statistical Analysis
[^4]: Data Interpretation
[^5]: Manuscript Preparation
[^6]: Literature Search
[^7]: Funds Collection
[^8]: Erli Zhang and Yongjian Wu contributed equally to this work
| {
"pile_set_name": "PubMed Central"
} |
All relevant data are within the manuscript and its Supporting Information files.
Introduction {#sec001}
============
Swans are among the largest existing flying birds of the waterfowl family Anatidae. They are represented around the world (except in Africa and Antarctica) by six species from the genus *Cygnus*: four in the temperate and artic zones of North America and Eurasia (*Cygnus olor*, *Cygnus buccinator*, *Cygnus cygnus* and *Cygnus columbianus*), one in the temperate zones of South America (*Cygnus melancoryphus*) and another one in the south of Australia and New Zealand (*Cygnus atratus*) \[[@pone.0226331.ref001],[@pone.0226331.ref002]\]. Environmental aspects related to swans have triggered numerous conservation strategies around the world, for example the successful re-introduction of the trumpeter swan *C*. *buccinator* by wildlife American agencies after it was close to extinction during the 1930's \[[@pone.0226331.ref003],[@pone.0226331.ref004]\]. Currently, the presence, population abundances and reproductive success of swans have been used as proxies for environmental changes in threatened wetlands \[[@pone.0226331.ref005],[@pone.0226331.ref006],[@pone.0226331.ref007]\].
The black-necked swan *C*. *melancoryphus* is the only representative of the genus in South America, and nearly 100,000 swans \[[@pone.0226331.ref008]\] inhabit freshwater and coastal wetlands located between ca. 28°-52°S \[[@pone.0226331.ref009]\]. These water birds prefer habitats with abundant subaquatic banks of macrophytes serving as their primary food source. Thus, *C*. *melancoryphus* has been described as an herbivorous species \[[@pone.0226331.ref010],[@pone.0226331.ref011],[@pone.0226331.ref012]\], similar to all the other swans which are primarily vegetarians \[[@pone.0226331.ref001],[@pone.0226331.ref002]\]. Because of their low digestive efficiency, these birds dedicate nearly 50% of their daily activity to foraging \[[@pone.0226331.ref011],[@pone.0226331.ref012]\]. Therefore, the population abundances of herbivorous swans exert a significant foraging pressure over the aquatic macrophyte standing stocks as it has been shown for *C*. *melancoryphus* in Chile and Argentina \[[@pone.0226331.ref010],[@pone.0226331.ref013]\], *C*. *atratus* in eastern Australia \[[@pone.0226331.ref014]\], *C*. *olor* in eastern USA \[[@pone.0226331.ref015]\] and *C*. *columb*i*anus* in Canada \[[@pone.0226331.ref016]\].
The Río Cruces wetland is one of the most well-known estuarine wetlands along the Chilean coast (ca. 39°S; [Fig 1](#pone.0226331.g001){ref-type="fig"}), characterized for its high diversity of aquatic macrophytes and water birds \[[@pone.0226331.ref017]\] and its tectonic origin \[[@pone.0226331.ref018]\]. In 1981, the central area of the wetland was declared the first Ramsar site in Chile (Ramsar site Santuario de la Naturaleza Carlos Anwandter, [https://www.ramsar.org](https://www.ramsar.org/)). Until 2004, the Río Cruces wetland was the main reproductive and nesting area of *C*. *melancoryphus* in South America \[[@pone.0226331.ref019]\], a fact probably related to the abundance of the aquatic macrophyte *Egeria densa* \[[@pone.0226331.ref020]\], which represents the primary food source for *C*. *melancoryphus* and other herbivorous water birds in this region such as coots (*Fulica* spp.) \[[@pone.0226331.ref021]\].
{#pone.0226331.g001}
Although *C*. *melancoryphus* has been described as a migratory species \[[@pone.0226331.ref022],[@pone.0226331.ref023],[@pone.0226331.ref024]\], the observed patterns of movement within the Río Cruces wetland characterize this water bird as rather resident \[[@pone.0226331.ref010],[@pone.0226331.ref019]\], similar to the behavior reported for swans from the northern hemisphere \[[@pone.0226331.ref025]\]. Between 1987 and 2003, the swan population of the Río Cruces wetland showed particularly high emigration rates with inter-annual variations between 2,000 and 12,000 birds \[[@pone.0226331.ref024]\], which were apparently related to large-scale climatic forcing by El Niño Southern Oscillation events (ENSO). However, after reaching an approximate number of 5,400 birds in early 2004, the swan population dropped to less than 600 birds during 2005 \[[@pone.0226331.ref026]\] due to massive migration. It was argued that the putative cause of that migration was an episodic change in water quality leading to the disappearance of *E*. *densa* from vast areas of the wetland, which was concomitant with the onset of production of a new pulp mill plant located nearly 25 km upstream from the wetland \[[@pone.0226331.ref027]\]. Since 2012, a gradual recovery of *E*. *densa* was observed across the wetland, followed by a gradual recovery of the population of swans \[[@pone.0226331.ref026],[@pone.0226331.ref027],[@pone.0226331.ref028]\] reaching numbers as high as 16,000 birds in early 2019 ([https://www.conaf.cl](https://www.conaf.cl/)).
Thus, the variability in population abundance and permanence of black-necked swans within the Río Cruces wetland appears to be strongly coupled to the cover of the aquatic macrophyte *E*. *densa*. Consequently, the dietary composition of these water birds is expected to be a reliable proxy for temporal changes in the cover of the most common aquatic macrophytes in coastal wetlands. To test this hypothesis, we studied the diet of *C*. *melancoryphus* by analyzing their feces and the interannual variability of the most common macrophyte species in the Río Cruces wetland as derived from remote sensing data.
Material and methods {#sec002}
====================
CV thanks the support from the Graduate School of Faculty of Sciences, Vicerrectoría de Investigación y Creación Artística (project I-2015-10) and Graduate Office of Universidad Austral de Chile (Assistance Scholarship, 2015--2016). FAL received funds from CONICYT PIA ANILLOS ACT172037. All the authors acknowledge the support of CONAF- Valdivia; especially the assistance in the field of Luis Miranda and Roberto Rosas and the logistical support of Mario Maturana (the administrator of the Ramsar site). The captures of the swans were authorized by the National Forest Corporation (Corporación Nacional Forestal) (resolutions N°01/2015 PCM/RAA; N°1/2016; N°327840/2016; N°417735/2016) and by the Agricultural and Livestock Service (Servicio Agrícola y Ganadero) (resolutions N°1786/2016; Nº3670/2016; N°255/2017) at the Agricultural Ministry of Chile (Ministerio de Agricultura).
Study area {#sec003}
----------
The Río Cruces wetland is an extensive inundated area formed by co-seismic continental subsidence during the 1960 Valdivia earthquake with the largest magnitude Mw 9.5 ever recorded by seismic instruments \[[@pone.0226331.ref018]\]. During this event, the Río Cruces river banks were flooded to form an extensive area with shallow water levels (less than 2 m depth), which was colonized by subaquatic macrophyte banks dominated by *E*. *densa* \[[@pone.0226331.ref020]\]. Eight tributary rivers join the Río Cruces central axis forming together a wetland area of approximately 6000 ha ([Fig 1](#pone.0226331.g001){ref-type="fig"}). The wetland is an estuarial system characterized by a tidal variability between 40 and 50 cm \[[@pone.0226331.ref029]\]. The climate in this region is temperate and rainy, with precipitations between 1300 and 3500 mm per year and an annual cycle with minimum and maximum rainfall during January-March and May-August, respectively \[[@pone.0226331.ref017]\].
The diet of swans {#sec004}
-----------------
The direct analysis of swan feces is the most common approach to study their dietary habits, probably because is not expensive and does not harm the birds. Other techniques include analyses of gut content \[[@pone.0226331.ref030],[@pone.0226331.ref031]\] and stable isotopes of C and N \[[@pone.0226331.ref032],[@pone.0226331.ref033]\]. Since the early seventies, several studies concerning the diet of *C*. *melancoryphus* have been carried out in wetlands across Chile, Argentina and Uruguay ([S1 Table](#pone.0226331.s001){ref-type="supplementary-material"}). Nearly 47% of these studies used or included analyses of swan feces, including the 7% corresponding to Chilean sites. For the sake of comparison and due to the benefits listed above, we chose the same technique in order to study the dietary habits of *C*. *melancoryphus* in the Río Cruces wetland.
Capture of swans and acquisition of feces {#sec005}
-----------------------------------------
A total of 152 adult swans were captured between 2012 and 2015: 12 in February 2012, 20 in April 2012, and 30 in September 2015, May and July 2016, and April 2017, respectively. The capture was carried out with nets operated from a motorboat in the central area of the Río Cruces wetland (within nearly 15 km along the river axis) ([Fig 1](#pone.0226331.g001){ref-type="fig"}). As revealed by GPS trackers, the capture zone encompasses virtually the whole area usually occupied by swans during their daily activities (see <http://www.birdecologylab.cl/cisnes-con-collares/>). The area was selected according to two criteria: 1) the possibility to safely execute the maneuvers of the motorboat necessary to capture and then liberate the birds, and 2) the major occurrence of swans during the period 1991--2017 according to the census carried out by the Corporación Nacional Forestal (CONAF) of Chile ([http://www.conaf.cl](http://www.conaf.cl/)) in that area. After their capture, the swans were carefully placed into resistant cloth bags leaving the head and neck free to minimize stress. The bags were necessary to ensure that the collected feces were fresh and originated from the same specimen. Subsequently, the feces were stored in hermetic bags and preserved in 70% alcohol solution.
Analysis of feces {#sec006}
-----------------
We applied micro-histological techniques to analyze the taxonomic composition of aquatic macrophytes in the swan feces \[[@pone.0226331.ref034]\]. The macrophytes were identified by examining the size, shape and structure of epidermic cells from intact tissue fragments in the feces, which were compared to a reference histologic catalog of the plant species from the Río Cruces wetland compiled for this study \[[@pone.0226331.ref035]\]. The analysis was based on high-resolution digital photographs obtained with an ACCU-SCOPE camera connected to an optical microscope (10x magnification) and subsequently processed with the Micrometrics Premium software.
The feces collected from each individual swan on a given sampling date were treated as a single replicate, which was subjected to microscopic analysis to obtain a high-resolution record of the presence/absence of plant species. A suite of pseudo-replicates was generated for each replicate according to the following procedure: at each sampling date, two portions of approximately 0.01 cm^3^ were randomly extracted from the feces collected from each individual swan; each portion was then uniformly distributed over a Neubauer counting chamber, where 10 field views (each with an area of 1 mm^2^) were randomly selected and analyzed. In total, 3040 field views were examined (20 field views per individual swan x 152 swans).
We calculated the occurrence frequency of each plant species within the sampled feces collected on each sampling date to evaluate the occurrence and temporal variability of different dietary items. Each plant species was counted only once in the feces of each individual swan (i.e. one record per swan), irrespective of the number of times the plant appeared in pseudo replicates. We also calculated the average occurrence frequency (± 1 standard error) of each plant species across the six sampling dates.
On the other hand, the pseudo-replicates were used to assess the incidence of plant species within the swan feces as a proxy for the consumption intensity on each sampling date (the higher the incidence, the higher the proportion of biomass in the feces). Based on the averaged frequency of plant species for each replicate obtained from the two Neubauer chambers, the general incidence of a plant was calculated as the grand average (± 1 standard error) among the total of collected feces on each sampling date. Only those feces where the plant was present were considered for this calculation. These average incidence values are independent from the occurrence frequency calculated at the replicate level, and provide an indirect quantitative estimate of the relative importance of different macrophytes at the time of their consumption.
Statistical analysis of macrophyte consumption {#sec007}
----------------------------------------------
To evaluate the seasonal variation of the swan's diet, we applied a square root transformation of the occurrence frequencies of dietary items and calculated a similarity matrix using the Bray-Curtis index \[[@pone.0226331.ref036]\]. The matrix was used for: 1) an Analysis of Similarities (ANOSIM, 999 permutations) to test for possible dietary differences between the six sampling dates, which were evaluated by using paired difference tests with the application of a Bonferroni correction, and 2) a Similarity Percentage Analysis (SIMPER) to identify the macrophyte species with major percental contribution and the observed dietary similarity of swans. The analyses were carried out with the PRIMER v. 6.0 software \[[@pone.0226331.ref037]\].
Remote sensing data {#sec008}
-------------------
To describe the availability of the main aquatic macrophytes, we estimated their area of distribution across the Río Cruces wetland. We generated species distribution models (SDMs) using geo-referenced occurrences of the most important macrophytes sampled between 2015 and 2019 throughout the wetland. For the spring-summer season of those years, geographic coordinates for large mono-specific patches of *E*. *densa*, *Potamogeton pusillus* and *Potamogeton lucens* were recorded and used to fit the SDMs. For those five years, SDMs were fit using remote sensing layers extracted from a Landsat 8 Operational Land Imager (OLI) scenes recorded on location 233/88 of the path/row of Worldwide Reference System 2 (WRS-2) \[[@pone.0226331.ref026],[@pone.0226331.ref028],[@pone.0226331.ref038]\] ([S2 Table](#pone.0226331.s002){ref-type="supplementary-material"}). For each OLI scene, bands 2 through 7 were processed as described by \[[@pone.0226331.ref039]\] and yielded top-of-atmosphere reflectance percentage values (RTOA). In addition, four indices were calculated \[[@pone.0226331.ref040], [@pone.0226331.ref041]\]:
\(1\) the blue/green ratio, as a proxy for chlorophylzl content: $$CHL = \frac{Band2}{Band3}$$
\(2\) the normalized difference vegetation index (NDVI) \[[@pone.0226331.ref040]\]: $$NDVI = \frac{\left( {Band5 - Band4} \right)}{\left( {Band5 + Band4} \right)}$$
\(3\) the enhanced vegetation index (EVI): $$EVI = 2.5\frac{\left( {Band5 - Band4} \right)}{\left( {Band5 + 6Band4–7.5Band2 + 1} \right)}$$
\(4\) and the modified normalized water difference index (MNWDI) \[[@pone.0226331.ref041]\]: $$MNDWI = \frac{\left( {Band3 - Band6} \right)}{\left( {Band3 + Band6} \right)}$$
This procedure yielded ten GIS predictive layers that characterize the studied scenes with a spatial resolution of 30 m. Species SDMs were fit using Maximum Entropy Species Distribution Modelling software v.3.3 (MaxEnt) using a 5-fold cross-validation scheme, thus allowing every occurrence data point to be used as part of the training and evaluation data sets \[[@pone.0226331.ref041],[@pone.0226331.ref042],[@pone.0226331.ref043],[@pone.0226331.ref044],[@pone.0226331.ref045]\]. MaxEnt uses information on spatial occurrences or presences and GIS layers or features to estimate the probability of a species being present across the study area \[[@pone.0226331.ref041],[@pone.0226331.ref042],[@pone.0226331.ref043],[@pone.0226331.ref045]\]. Recent work has shown that the Maximum Entropy statistical distribution is equivalent to that obtained from an inhomogeneous Poisson Process (IPP), which allows Maxent's 'raw' output format to be used directly as a model of relative abundance \[[@pone.0226331.ref045]\]. However, in order to obtain probability of presence as an output variable, a Bernoulli generalized linear model whose link function is termed a complementary log-log (cloglog) link is used to transform the raw output to a probability of presence \[[@pone.0226331.ref045]\]. Model performance was assessed using the Area-Under-the-Curve (AUC) statistic for the Receiver Operating Characteristic (ROC) \[[@pone.0226331.ref042]\]. Fitted models were later projected over the Río Cruces wetland, using the same GIS predictive layers, converting probability values across the wetland to binary predictions, (i.e. a prediction of macrophyte presence and absence across the landscape). This was done by applying a threshold to the predicted cloglog presence probabilities, using the probability threshold value that maximizes the sum of sensitivity and specificity (MSS) \[[@pone.0226331.ref046]\]. While it has been shown that Maxent models derived from occurrence records may show correlations with independently measured local abundance values, such independent measures or estimates of total population size are required to estimate absolute abundance. As a result, we do not aim to estimate variations in macrophyte abundance, but rather in available geographical area covered by the most important macrophyte species in the Río Cruces wetland. The estimated distribution map was then used to calculate the area for each of the three macrophytes, yielding a time series of estimated distribution area for *E*. *densa*, *P*. *pusillus* and *P*. *lucens* across the Río Cruces wetland.
To estimate historical variation of the distribution area for the three studied macrophytes, the MaxEnt model fitted on the 2014--2015 occurrences and remote sensing data were transferred or projected using remote sensing scenes from previous spring-summer seasons. This allowed us to hind cast the estimated distribution area for each species. We used the fitted model to predict expected HSI values across the wetland, using Landsat 5 thematic mapper (TM) and Landsat 7 enhanced thematic mapper (ETM) remote sensing scenes for the previous 5 years. Specifically, TM scenes for WRS-2 location 233/88 were downloaded for the spring-summer seasons of 2009--2010 and 2010--2011, while ETM scenes were downloaded for the spring-summer seasons of 2011--2012, 2012--2013 and 2013--2014 ([S3 Table](#pone.0226331.s003){ref-type="supplementary-material"}).
All those scenes were processed in the same manner as described for the 2015--2019 Landsat 8 OLI scenes, in order to obtain the same 10 layers for each of these temporal samples. These sets of GIS layers were used together with the 2014--2015 fitted MaxEnt model to predict the expected HSI values and distribution areas across the wetland for each of the three macrophytes. For ETM scenes, two scenes per year were used to correct the missing data caused by the failure of the Scan Line Corrector. A composite SDM was generated by projecting the 2014--2015 fitted MaxEnt on both ETM scenes per year, and then replacing missing data with information from the second layer. This yielded five retrospective estimates of spring- summer distribution area for *E*. *densa*, *P*. *pusillus* and *P*. *lucens* across the wetland (seasons 2009--2010, 2010--2011, 2011--2012, 2012--2013 and 2013--2014). Temporal variation in geographic distribution for each of these dominant macrophytes was described by their coefficient of variation, while cross-correlation between pairs of macrophytes was assessed using Pearson correlation coefficients and two-sided t-tests. We also tested whether estimated average distribution areas of the three dominant macrophytes was positively correlated with their observed occurrence frequency (OF) and average incidence (AI) values in the swan feces, using ordinary least squares (OLS) linear regression.
Results {#sec009}
=======
Consumption of aquatic macrophytes {#sec010}
----------------------------------
The diet of swans consists of six aquatic macrophyte species, which are characterized by four life habits: submerged, floating, floating freely, and emerged ([Fig 2](#pone.0226331.g002){ref-type="fig"} and [Table 1](#pone.0226331.t001){ref-type="table"}). *Egeria densa*, *Potamogeton pusillus* and *Potamogeton lucens* were the only species that were registered on all six sampling dates. *Myriophyllum aquaticum* and *Schoenoplectus californicus* were registered on four sampling dates, whereas *Limnobium laevigatum* was documented on three sampling dates ([Table 1](#pone.0226331.t001){ref-type="table"}).
![Field photographs of aquatic macrophytes consumed by the swans in the study area.\
The species are classified in groups according to four life habits \[[@pone.0226331.ref047]\]. Submerged (a. *Egeria densa*, b. *Potamogeton pusillus*, c. *Myriophyllum aquaticum*); floating (d. *Potamogeton lucens*); freely floating (e. *Limnobium laevigatum*), and emerged (f. *Schoenoplectus californicus*).](pone.0226331.g002){#pone.0226331.g002}
10.1371/journal.pone.0226331.t001
###### The diet of swans.
Occurrence frequency (OF), average occurrence frequency through the study period (AOF; mean ± 1 standard error) and average incidence (AI; mean ± 1 standard error) of each dietary item from swan feces in the study area during the sampling dates. The (+) symbol indicates single values.
{#pone.0226331.t001g}
Macrophyte species February 2012 April 2012 September 2015 May 2016 July 2016 April 2017 AOF
------------------------------- --------------- ------------ ---------------- ------------ ----------- ------------ ------- ------------ ------- ------------ ------- ------------ -------------
*Egeria densa* 91.7 61.0 ± 7.9 100.0 76.8 ± 4.5 100.0 94.3 ± 2.1 100.0 96.2 ± 1.1 100.0 98.3 ± 0.7 100.0 99.8 ± 0.2 98.6 ± 1.4
*Potamogeton pusillus* 83.3 81.1 ± 5.6 80.0 63.3 ± 4.3 73.3 37.2 ± 4.3 23.3 23.1 ± 1.9 33.3 15.4 ± 1.5 20.0 10.0 ± 0.0 52.2 ± 12.1
*Potamogeton lucens* 33.3 12.9 ± 1.0 50.0 27.7 ± 1.7 40.0 11.1 ± 0.4 30.0 21.5 ± 1.4 53.3 15.8 ± 1.2 30.0 10.8 ± 0.4
*Myriophyllum aquaticum* 50.0 22.0 ± 2.9 75.0 25.3 ± 2.4 13.3 12.0 ± 0.6 20.0 11.1 ± 0.4 39.6 ± 11.6
*Limnobium laevigatum* 3.3 10.0 (+) 3.3 35.0 ± 2.7 13.3 10.0 ± 0.0 6.6 ± 2.4
*Schoenoplectus californicus* * * 3.3 10.0 ± 0.0 10.0 13.3 ± 0.7 16.7 16.7 ± 1.3 6.7 13.3 ± 0.7 9.2 ± 2.3
During the period of evaluation, *E*. *densa* was the most frequently ingested macrophyte, and it was detected in 99.3% of the 152 analyzed feces. Its occurrence frequency on each sampling date was 100%, except in February 2012 when it was 91.7%. The average occurrence frequency of *E*. *densa* among all sampling occasions was found to be 98.6 ± 1.4% ([Table 1](#pone.0226331.t001){ref-type="table"}). Furthermore, the average incidence of *E*. *densa* within the studied feces showed a gradual and consistent increase within the studied period from 61.0 ± 7.9% to 99.8 ± 0.2% ([Table 1](#pone.0226331.t001){ref-type="table"}). In turn, *P*. *pusillus* was detected in 46.7% of all feces, and its occurrence frequency decreased from relatively high initial values of 83.3% to 20% towards the end of the sampling period. The average occurrence frequency of this macrophyte was estimated with 52.2 ± 12.1% ([Table 1](#pone.0226331.t001){ref-type="table"}). Furthermore, the average incidence of *P*. *pusillus* in the feces progressively decreased over time from 81.1% to 10% ([Table 1](#pone.0226331.t001){ref-type="table"}). *E*. *densa* and *P*. *pusillus* both appear as important trophic elements in the studied period, but their respective average incidences in the swan's diet show a significant negative correlation (r ~Spearman~ = -0.61; p \< 0.05).
On the other hand, *P*. *lucens* was detected in 39.5% of the feces and its occurrence frequency over the studied period varied between 30% and 53.3% with an average value of 39.4 ± 4.2%. Similarly, the average incidence of *P*. *lucens* maintained generally low values between 10.8% and 27.7% ([Table 1](#pone.0226331.t001){ref-type="table"}). *M*. *aquaticum* was detected in only 20.4% of the feces, showing important variations in occurrence frequency (13.3% - 75%) and average incidence (11.1% - 25.3%) and a general decreasing tendency over time (average occurrence frequency 39.6 ± 11.6%) ([Table 1](#pone.0226331.t001){ref-type="table"}). Both *L*.*laevigatum* and *S*. *californicus* were detected only since 2015 and the occurrence frequency and the average incidence of both species were relatively low (as well as their average occurrence frequency) ([Table 1](#pone.0226331.t001){ref-type="table"}).
The ANOSIM results indicate that the composition of the swan's diet varied significantly over time (global R = 0.25; p = 0.001). The comparisons between sampling dates (with corrected α = 0.0033) showed significant differences in 13 out of 15 paired tests (0.001 ≤ p ≤ 0.035). The non-significant differences corresponded to the comparisons between May 2016 and July 2016 and between May 2016 and April 2017 (0.124 ≤ p ≤ 0.155, respectively).
According to the SIMPER analysis, the recorded differences are mainly controlled by the dominance of some macrophyte species in the swan's diet. The species with major percental contribution to the contrasting characteristics of sampling dates are (in decreasing order of their importance and with individual percentage of contribution): i) February 2012: *E*. *densa* and *P*. *pusillus* with 91.5%; ii) April 2012: *E*. *densa*, *P*. *pusillus*, and *M*. *aquaticum* with 95.5%; iii) September 2015: *E*. *densa* and *P*. *pusillus* with 96.8%; iv) May 2016: *E*. *densa* with 96.8%; v) July 2016: *E*. *densa* with 91.8%; and vi) April 2017: *E*. *densa* with 97.1%.
Cover of aquatic macrophytes {#sec011}
----------------------------
All MaxEnt species distribution models fitted for data between 2015 and 2019 showed high AUC values, with all species presenting values above 0.92 ([S4 Table](#pone.0226331.s004){ref-type="supplementary-material"}). For all studied years, *E*. *densa* appears as the dominant aquatic macrophyte, followed by *P*. *pusillus* and *P*. *lucens* (observed average areas ± s.e. are 4275 ha ± 772 ha, 2701 ha ± 580 ha and 1623 ha ± 562 ha, respectively) ([Fig 3](#pone.0226331.g003){ref-type="fig"}). The estimated distribution areas show important fluctuations; *P*. *lucens* is characterized by the greatest variability with a coefficient of variation (CV) of 77%. It is followed by *P*. *pusillus* and *E*. *densa* with CV values of 48% and 40%, respectively. The estimated areas of *E*. *densa* and *P*. *pusillus* show a significant negative correlation (two sided correlation t-test: t\[[@pone.0226331.ref002]\] = -5.9942, p \< 0.05, r = -0.97), while the area of *P*. *lucens* is negatively correlated with *E*. *densa* and *P*. *pusillus* (*P*. *lucens*, two sided correlation t-test: t\[[@pone.0226331.ref002]\] = -0.7944, p \> 0.05, r = -0.49 and t\[[@pone.0226331.ref002]\] = 0.4594, p \> 0.05, r = 0.31 for *E*. *densa* and *P*. *pusillus*, respectively).
{#pone.0226331.g003}
The analysis of the relationships between distribution areas of the three dominant macrophytes and their occurrence frequency (OF) and average incidence (AI) values in the swan feces, show that the average OF and AI values are significantly and positively correlated with the average distribution area for *E*. *densa*, *P*. *pusillus* and *P*. *lucens* ([Fig 4](#pone.0226331.g004){ref-type="fig"}). A significant zero-intercept ordinary least squares (OLS) linear regression was found for both variables (OLS F \[[@pone.0226331.ref001],[@pone.0226331.ref001]\] = 1243, p \< 0.001, R^2^ = 0.99 and F \[[@pone.0226331.ref001],[@pone.0226331.ref001]\] = 141.6, p \< 0.001, R^2^ = 0.98 for OF and AI, respectively).
{#pone.0226331.g004}
Discussion {#sec012}
==========
Even though our sampling strategy did not continuously cover the entire period from 2012 to 2017, the results suggest that the composition of the swan's diet and the relative importance of the dietary items remained similar over the entire period and marked by an increasing dominance of *E*. *densa*. This trend is likely controlled by the progressive recuperation of *E*. *densa* macrophyte banks along the shallow areas of the Río Cruces wetland, which started at the beginning of the study period in 2012 and eight years after the abrupt decline of this macrophyte in mid-2004 \[[@pone.0226331.ref026],[@pone.0226331.ref028]\].
Even though the results of this study do not allow for a direct evaluation of the dietary preferences of black-necked swans \[[@pone.0226331.ref012]\], they suggest that *E*. *densa* is not being selectively consumed and its dominance within the diet is likely due to its high occurrence in the shallow areas of the wetland \[[@pone.0226331.ref010],[@pone.0226331.ref028]\]. This is reinforced by the positive correlation between occurrence frequency, average incidence and the average area occupied by *E*. *densa* and *P*. *pusillus*, apart from *P*. *lucens*, which represent the other dominant species in the swan's diet (see [Fig 4](#pone.0226331.g004){ref-type="fig"}). The results thus point to the opportunistic behavior of black-necked swans, similar to the observed in other *Cygnus* species (*C*. *olor*, *C*. *columbianus* and *C*. *cygnus*) in coastal wetlands of the northern hemisphere \[[@pone.0226331.ref048],[@pone.0226331.ref049],[@pone.0226331.ref050]\]. In order to achieve these results, we base our analysis on the assumption that the geographic area occupied by these macrophytes adequately represents their availability for consumption by swans, which in turn requires that our estimates of geographic distribution area within the Río Cruces Ramsar site accurately reflect plant availability for black necked swans. This assumption seems reasonable, given that this species is able to migrate over large geographical areas and its foraging areas span most Río Cruces Wetlands \[[@pone.0226331.ref051],[@pone.0226331.ref052]\]. In this regard, it is relevant to note that the Maximum Entropy statistical distribution has been shown to be is equivalent to the distribution obtained from an inhomogeneous Poisson Process (IPP), which allows Maxent's 'raw' output format to be used directly as a model of relative abundance \[[@pone.0226331.ref045]\]. Hence, the probability of observing a given species in a pixel can be considered as a coarse filter of species abundance, which is consistent with the general relationship between species abundance and distribution \[[@pone.0226331.ref053],[@pone.0226331.ref054]\]. An important caveat to these results is the fact that the Maximum entropy modelling does not consider any potential interactions among species, and thus, estimated distribution areas may be overestimated. This potential bias is likely to be greater for *P*. *pusillus* and *P*. *lucens* than for *E*. *densa*. Hence, modelled plant distributions do not take into consideration the possible effects of competitive interactions either in reducing modelled distribution, or in decreasing plant abundance or quality. Further studies would require concurrent sampling of plant quality, biomass or abundance in order to address this limitation of our data.
Consumption of *E*. *densa* would be comparatively rewarding because its energy content (ca. 16.3 kJ g^-1^) \[[@pone.0226331.ref010],[@pone.0226331.ref055]\] is higher than that of *P*. *pusillus* and *P*. *lucens* (ca. 14.4 and 12.1 kJ g^-1^, respectively) \[[@pone.0226331.ref056]\]. Furthermore, the life habit of macrophytes might also be a key factor in the foraging of *C*. *melancoryphus* \[[@pone.0226331.ref047]\], since its diet is dominated by submerged macrophytes such as *E*. *densa* and *P*. *pusillus*. This could be related to the complexity and presence/absence of structural components that support each type of macrophyte (e.g. sclerenchyma, mesophyll, epidermis and cuticle) and the morphology of its foliar structures (e.g. filiform, broad, thin or thick) \[[@pone.0226331.ref057]\]. In general, the submerged macrophytes have a slender and less developed mesophyll, very thin or absent epidermis and thin or filiform leaves (as in *E*. *densa*). In turn, other macrophytes are characterized by more resistant, complex and thick structures \[[@pone.0226331.ref056],[@pone.0226331.ref057]\], which are more difficult to digest and therefore less consumed by the herbivorous water birds \[[@pone.0226331.ref034],[@pone.0226331.ref035]\]. Interestingly, the index of digestibility (ID) of the submerged *E*. *densa* has been estimated to be 6 times greater than the ID of the free-floating *L*. *laevigatum* (17.9% and 2.8%, respectively) \[[@pone.0226331.ref011]\].
A higher consumption of submerged macrophytes by *C*. *melancoryphus* has been also described in other coastal wetlands in South America, such as Laguna de Rocha (ca. 34°S, Uruguay) and Mar Chiquita (ca. 37°S, Argentina), where they graze mainly on underwater banks of *Zannichelia palustris* and *R*.*maritima* \[[@pone.0226331.ref013],[@pone.0226331.ref058]\], and Lago Budi (ca. 38°S; southern Chile) where their main food is *Stuckenia pectinata* \[[@pone.0226331.ref059]\].
Shallow water areas have great importance as foraging sites not only for swans, but also for other herbivorous water birds such as coots, pochards, ducks and geese \[[@pone.0226331.ref021],[@pone.0226331.ref060]\]. These ecosystems and their dynamics are highly sensitive to variable water levels, which directly impact the foraging behavior of water birds. When water levels in coastal wetlands of southern Chile reach very high levels, *C*. *melancoryphus* is forced to forage on swamp land macrophytes in the periphery of the wetlands \[[@pone.0226331.ref011],[@pone.0226331.ref012],[@pone.0226331.ref059],[@pone.0226331.ref061]\]. Hence, swans are forced to use the riparian zones of the wetlands as foraging sites \[[@pone.0226331.ref006],[@pone.0226331.ref012],[@pone.0226331.ref062],[@pone.0226331.ref063]\], which in turn makes them more vulnerable to attacks by land predators \[[@pone.0226331.ref064],[@pone.0226331.ref065]\].
Our results demonstrate a significant correlation between average macrophyte cover and average frequency and incidence of the dominant macrophyte species in the diet of black necked swans. Hence, the dietary habits of swans might prove as a reliable proxy for the availability of dominant macrophyte species in their habitat, coincident with the large abundance of the macrophyte *E*. *densa* in the Río Cruces wetland. To further characterize the foraging behavior of black-necked swans, future studies should focus on the spatio-temporal variability in plant cover or biomass, as well as on the nutritional properties of dominant dietary items in the Río Cruces wetland. The results of the present study have important implications for the integral conservation of coastal wetlands inhabited by *C*. *melancoryphus*. They highlight the importance of preserving shallow water habitats mainly occupied by macrophytes, which provide the main food source for these iconic water birds. Our results further demonstrate that an evident change in the composition of the bird's diet, might indicate important variations in the patterns of distribution and concentration of the corresponding macrophytes in coastal areas such as the Río Cruces wetland. Recent studies have highlighted that biogeographical distribution of plant communities in coastal wetland ecosystems of central and southern Chile (32-40ºS) cannot be fully explained by climatic conditions, suggesting possible feedbacks between biological and environmental factors \[[@pone.0226331.ref066],[@pone.0226331.ref067]\].
Supporting information {#sec013}
======================
###### Geographic locations in South America, approximate latitudes, types of wetlands, methods used and references related to studies on trophic diets of swans.
(DOCX)
######
Click here for additional data file.
###### List of Landsat satellite images analysed to model distribution of aquatic macrophytes in the study area.
For every spring-summer season, we indicate the Landsat mission as well as the scene acquisition date.
(DOCX)
######
Click here for additional data file.
###### List of Landsat satellite images analysed to project historical distribution of aquatic macrophyte in the study area.
For every spring-summer season, we indicate the Landsat mission and sensor, as well as the identification codes of the Landsat scene and the acquisition date. For data from Landsat 7 satellite, two scenes were downloaded in order to fill the gaps caused by the sensors.
(DOCX)
######
Click here for additional data file.
###### Summary statistics for the fitted MaxEnt ENMs for aquatic macrophytes at the Rio Cruces wetland.
The table shows for each spring-summer period the observed average sample sizes and average AUC values for training and test cross validations sets used for the macrophytes *Egeria densa*, *Potamogeton lucens* and *Potamogeton pusillus*.
(DOCX)
######
Click here for additional data file.
CV thanks the support from the Graduate School of Faculty of Sciences, Vicerrectoría de Investigación y Creación Artística (project I-2015-10) and Graduate Office of Universidad Austral de Chile (Assistance Scholarship, 2015--2016). FAL received funds from CONICYT PIA ANILLOS ACT172037. All the authors acknowledge the support of CONAF---Valdivia; especially the assistance in the field of Luis Miranda and Roberto Rosas and the logistical support of Mario Maturana (the administrator of the Ramsar site). The captures of the swans were authorized by the National Forest Corporation (Corporación Nacional Forestal) (resolutions N°01/2015 PCM/RAA; N°1/2016; N°327840/2016; N°417735/2016) and by the Agricultural and Livestock Service (Servicio Agrícola y Ganadero) (resolutions N°1786/2016; Nº3670/2016; N°255/2017) at the Agricultural Ministry of Chile (Ministerio de Agricultura).
[^1]: **Competing Interests:**There are no competing interests with anybody. The conclusions of this study are just those of the authors and should not be interpreted as representing other points of views or policies from the Arauco Company that provided funding for this study. That funding does not alter authors\' adherence to PLOS ONE policies on sharing data and materials collected during this study.
| {
"pile_set_name": "PubMed Central"
} |
Introduction {#s1}
============
South Africa, a land of stark contrasts, contains a diverse natural beauty that can easily be compared with some of the world's most majestic outdoor scenes. One of the new seven wonders of the natural world, Table Mountain, parades its splendour to the capital of South Africa, Cape Town. Unfortunately, South Africa is also considered by many to be one of the tuberculosis (TB) capitals of the world. The incidence of TB in South Africa is estimated to have increased by over 400% in the past 15 years. This is confounded by a staggering co-infection rate of approximately 73% with the human immunodeficiency virus (HIV).[@R01]
One of the most dreaded complications of TB pericarditis is pericardial scar formation. Due to scarring, the pericardium becomes calcified and contracts over the cardiac chambers, thereby encasing the heart in a fibrocalcific skin that impedes diastolic filling.[@R02] Constrictive pericarditis (CP) is the natural consequence of about 17 to 40% of cases of TB pericardial infection.[@R03] The definitive treatment of CP is surgical removal of the pericardium, a procedure with a significant peri-operative mortality rate of approximately 15%.[@R04]
South Africa is on the forefront of research on TB heart disease and has recently published the large, multi-centre IMPI trial.[@R05] One of the goals of the IMPI trial was to assess the impact of corticosteroids in the management of TB pericarditis. The major findings of the study included (1) corticosteroids had no impact on mortality rates in patients with TB pericarditis, (2) corticosteroids decreased the incidence of pericardial constriction by 46%, and (3) HIV-positive patients who received corticosteroids had a significantly increased risk of developing HIV-associated malignancies.
In established TB, early and effective treatment with shortcourse anti-TB therapy is the mainstay of management. Various strategies have been investigated as adjuncts to anti-TB drugs in the prevention of pericardial constriction. The ongoing discussions and numerous investigations into a wide array of agents as possible 'magic bullets' in the prevention of pericardial constriction (post-TB infection) illustrates both the interest in the field, and also the lack of a satisfying solution to this problem. The following strategies have previously been evaluated: Mycobacterium indicus pranii immunotherapy,[@R05] corticosteroids,[@R05] pericardiocentesis,6 open surgical drainage (pericardial window),[@R07] thalidomide,[@R08] instilling intrapericardial fibrinolytic therapies,[@R09]-[@R11] and a wide array of non-steroidal anti-inflammatory medication. Not one of these therapies has, to date, been internationally recognised as an acceptable standard of therapy, and the choice of adjuvant treatment varies significantly among experts in the field.
Colchicine is an inhibitor of microtubule polymerisation. It acts by binding to tubulin and is registered for the acute treatment of gout crystal arthropathies. The plant source of colchicine, the autumn crocus (Colchicum autumnale), was described as treatment for arthritis in the Ebers Papyrus in 1500 BC.[@R12] In modern medicine, colchicine has however played a wider role in the treatment of pericarditis of various aetiologies, both acute and chronic. This has been investigated in a prospective, randomised trial named COPE (Colchicine for Acute Pericarditis),13 and the major findings concluded that colchicine significantly reduced the recurrence rates and symptom persistence due to pericarditis. To date however, the use of colchicine has, to the best of our knowledge, never been systematically assessed in the context of pericardial TB. The purpose of this research was to assess the merit for the use of colchicine in the context of TB pericarditis.
Methods {#s2a}
=======
This research was conducted in the Northern Cape province of South Africa at a secondary-level hospital in Kimberley between August 2013 and April 2015. The research was approved by the ethics committee of the University of the Free State and the study was registered with the National Health Research Committee. The research was conducted in accordance with the Declaration of Helsinki.
This pilot study was designed as a prospective, doubleblind, randomised, control cohort. All patients presenting to the Kimberley Hospital complex (KHC) with pericardial effusions were assessed for inclusion and exclusion criteria. In the absence of contra-indications, patients underwent therapeutic pericardiocentesis if the procedure was deemed safe and possible. Standard therapy was initiated in accordance with the South African National Tuberculosis Management Guidelines:14 weight-adjusted anti-TB drugs (Rifafourf^®^) and oral corticosteroids. (prednisone: 1.5 mg/kg per day for four weeks; 1.0 mg/kg per day for two weeks; 0.5 mg/kg per day for one week; 0.25 mg/kg per day for one week). HIV co-infected patients not previously on treatment were initiated on fixeddose combination (FDC) antiretroviral treatment six weeks after initiation of TB treatment (FDC: Tenofovir Disoproxil Fumarate 300 mg, Emtricitabine 200 mg and Efavirenz 600 mg).
Patients were randomly assigned to the intervention group with the use of a web-based randomisation system that ensured assignment concealment. The intervention group received colchicine (dose 1.0 mg per day) for a total of six weeks, whereas the control group received a placebo for the same period ([Fig](#F1){ref-type="fig"}. 1).
{#F1}
Patients subsequently underwent serial echocardiographic examinations on an out-patient basis and adherence checks, including pill counts, were done at follow-up visits. The primary outcome assessed was the development of pericardial constriction and this diagnosis was made echocardiographically at four months post initial presentation. Upon completion of the follow-up period of all patients, the blinding was unveiled and data were presented for statistical analysis.
Two groups of patients were included: (1) definite TB pericarditis: the presence of TB bacilli was observed on microscopic examination of pericardial fluid; cultures of pericardial fluid were positive for Rifampicin-sensitive Mycobacterium tuberculosis (MTB); pericardial fluid was positive for MTB on direct polymerase chain reaction (PCR) (Gene Xpert); and (2) probable TB pericarditis: proof of TB was found elsewhere (positive cultures for MTB on sputum or cerebrospinal fluid); pericardial fluid with adenine deaminase (ADA) level \> 40 U/l; a total diagnostic index score \> 6 on using the Tygerberg clinical prediction score ([Table 1](#T1){ref-type="table"}).[@R15]
###### The Tygerberg clinical prediction score for the diagnosis of TB pericarditis. A total diagnostic score \> 6 yields a sensitivity of 82% and a specificity of 76% for the diagnosis of TB pericarditis
*Admission variable* *Diagnostic index*
----------------------------- --------------------
Weight loss 1
Night sweats 1
Fever 2
Serum globulin \> 40 g/l 3
Leukocyte count \< 10 × 109 3
The exclusion criteria were: patients with renal or hepatic impairment (creatinine clearance rate \> 85 ml/min or transaminases \> 1.5 upper limit of normal); and pregnant patients or patients intending to become pregnant within four months.
The gold-standard diagnostic test for the diagnosis of CP is the demonstration of increased interventricular interdependence during cardiac catheterisation. Doppler echocardiography and other novel echocardiographic techniques have provided us with reliable non-invasive alternatives to the diagnosis of CP. In this study, the diagnosis of CP was made by means of echocardiography by adhering to the principles in the article by Dal-Bianco et al. on the echocardiographic diagnosis of CP.[@R16]
Initial echocardiographic assessment ensured that no features of constriction were present at the time of enrolment in the study. Follow-up echocardiograms were performed four months after the initiation of therapy.
The echocardiograms were performed and co-reviewed by two experienced echocardiographers (who had both attended a dedicated workshop at a tertiary-level academic hospital aimed at the echocardiographic diagnosis of CP). A GE Vivid E6® ultrasound machine was used to perform a systematic examination according to the basic minimum standards as stipulated by the British Society of Echocardiography.[@R17] Numerous other echocardiographic parameters were assessed, including the presence of a septal shudder, respiratophasic septal shift, left atrial enlargement and echocardiographic features of pericardial thickening ([Figs 2](#F2){ref-type="fig"}--[4](#F4){ref-type="fig"}).
{#F2}
{#F3}
{#F4}
Statistical analysis {#s2b}
====================
Statistical analysis was performed by the Department of Biostatistics of the University of the Free State, Bloemfontein, South Africa. The SAS Version 8.3 was used. Groups were compared regarding outcomes using frequency tables with appropriate hypothesis testing (chi-squared of Fisher's exact test) and 95% confidence intervals for differences in percentages. The standard deviation value p \< 0.05 was considered significant.
Results {#s3}
=======
Thirty-three patients met the initial inclusion criteria. Three patients passed away while in hospital and an additional three passed away during the follow-up period. In-patient deaths were due to neutropenic sepsis, cerebrovascular incident and nosocomial pneumonia, respectively. In all out-patient deaths, the cause was undetermined. Five patients were lost to follow up and one patient was removed from the study due to presumed drug side effects. A total of 21 patients completed the follow-up period ([Fig. 5](#F5){ref-type="fig"}).
{#F5}
The study population had a female preponderance (66% females) and the mean age of the studied patients was 31 years. Disseminated pericardial tuberculosis was found to be a disease exclusive to the immune-compromised in this cohort; all 21 patients were HIV positive. The median CD4^+^ count was 162 and 346 cells/mm^3^ in the colchicine and placebo groups, respectively.
Of the 21 eligible participants, 12 had been assigned to the treatment group and the remaining nine were in the placebo group. The diagnosis of definite pericardial tuberculosis was made in 23.8% of the patients, while the remaining 76.2% were diagnosed on the basis of suggestive clinical and biochemical features (see inclusion criteria). Of the studied patients, 47.6% underwent pericardiocentesis, whereas the remaining 52.4% could not undergo safe pericardiocentesis.
The average volume of fluid drained via single pericardial aspiration was 622 ml. The macroscopic appearance of the fluid varied from serosanguineous to haemorrhagic, reflecting the different pathological stages of development. Mycobacterium tuberculosis was proven on pericardial aspirates in 50% of cases, either by positive culture (30%) or by direct PCR technique (Gene Xpert) (20%) ([Table 2](#T2){ref-type="table"}).
###### Pericardial fluid biochemistry
*Biochemical parameter* *Average*
------------------------- -----------
Protein (g/l) 62.7
ADA (U/l) 96.6
LDH (U/l) 4494
pH 7.3
Glucose (mmol/l) 2.8
Pericardial constriction is the natural sequela of approximately 17 to 40% of TB pericardial infections.[@R03] In our cohort, the incidence of pericardial constriction (demonstrated by echocardiography) four months after the initial diagnosis was 23.8%. Of the five patients who developed pericardial constriction, two were in the control group and the remaining three were in the group treated with colchicine. Of those who did not develop pericardial constriction, nine were in the colchicine group and seven were in the placebo group.
The data from [Table 3](#T3){ref-type="table"} yields a p-value of 0.88. The relative risk for developing constriction in the colchicine group compared to the intervention group was 1.07 (95% CI: 0.46--2.46). There was therefore no statistically demonstrable correlation between the use of colchicine and pericardial constriction in this study cohort.
###### Two-by-two table demonstrating the primary study outcome
*Colchicine* *Placebo* *Total*
----------------- -------------- ----------- ---------
Constriction 3 2 5
No constriction 9 7 16
Total 12 9 21
The side effects among the patients using colchicine were usually minor; 56% of the initial 19 patients who were in the colchicine group reported self-limiting diarrhoea during their hospital stay. Serious side effects were observed in one patient who developed hepatitis during his course of treatment. The patient was removed from the study and daily liver function testing showed a rapid recovery.
Although the study was neither empowered nor designed to evaluate the effect of pericardiocentesis on the subsequent development of pericardial constriction, a very apparent and interesting finding was observed. We found that, with the exception of one patient, all those who developed pericardial constriction were in the group that did not undergo pericardiocentesis. Conversely, in the group that underwent pericardiocentesis, only one participant developed pericardial constriction. Pericardiocentesis therefore seemed to be very effective in the prevention of pericardial constriction and in this cohort only one patient (10%) who underwent pericardiocentesis developed constriction. These findings are observational and disregard the initial group allocations.
Discussion {#s4a}
==========
The proverbial 'eureka moment' in the management of TB pericarditis seems to be elusive. Numerous interventions have been postulated and investigated in an attempt to prevent the devastating post-inflammatory changes in the pericardium following TB pericarditis. In this pilot study, the merit of adding colchicine to the current management guidelines was investigated in a systematic manner. As all the participants of this study were HIV positive, the findings can only be applied to this subgroup of patients with TB pericarditis.
There was a notable difference in the median CD4^+^ lymphocyte count between the treatment and placebo groups, but when assessed as an independent variable, no correlation could be demonstrated between degree of immunocompetency, as measured by CD4^+^ count, and the risk for development of constriction.
This pilot study could not demonstrate any benefit derived from the addition of colchicine to the routine management of HIV-positive patients with TB pericarditis. The power of this pilot trial was insufficient to detect small differences in outcome; however, it appears that colchicine use has no correlation with the prevention or formation of post-TB CP. This pilot trial could not assess the beneficial effects of colchicine in the HIV-negative patient with TB pericarditis.
After considering the findings of this pilot research, the costs of the drug, the polypharmacy these patients are exposed to, drug--drug interactions and side effects (albeit mild), this study would advise against the use of colchicine in the management of HIV-positive patients with TB pericarditis.
The implementation of a pericardiocentesis until dryness (with or without extended drainage) was up to this point never studied in a controlled or comparative manner. Research conducted by Reuter et al.[@R06] in 2007 found the first evidence to suggest the benefit of a pericardiocentesis until dryness with extended drainage. In their research, 162 patients with TB pericarditis underwent pericardiocentesis, and over a followup period of six years, only two patients (1.23%) developed fibrous pericardial constriction. The research concluded that echocardiographic-guided pericardiocentesis with extended drainage is a safe and effective management option, and when combined with short-course anti-tuberculous therapy, it almost completely prevents the development of CP.
A few leading centres are employing a routine 'pericardiocentesis until dryness' approach based on this literature, whereas most do not. The interesting observation made in our pilot study was that the findings made by Reuters et al. in 2007 were reproducible on a much smaller scale. Pericardial constriction, although having a low incidence, was almost exclusively seen in the group that did not undergo pericardiocentesis (observational -- disregard original group allocation). As suggested by some expert opinion and as supported by the data published by Reuters et al. and observational findings of our pilot trial, the practice of routine pericardiocentesis until dryness in the absence of contraindications appears to be the preferred management option and this might well be the long-awaited 'eureka moment', in an attempt to halt the development of pericardial constriction.
Linitations of the study {#s4b}
========================
The diagnosis of pericardial constriction was made with echocardiography, whereas the gold standard for diagnosing CP is invasive haemodynamic studies. Work done by Oh et al.[@R18] and Boonyaratevej et al.[@R19] demonstrated that one of the most characteristic findings of CP, a respiratory variation in early transmittal inflow velocity, is neither perfect in its sensitivity nor specificity for the diagnosis. In patients with markedly elevated left atrial pressures, the respiratory variation in the inflow velocities may be less than 25%. Furthermore, in patients with chronic obstructive pulmonary disease and severe right ventricular dysfunction, the variation may be elevated in the absence of CP. This research emphasises the importance of using a variety of recognised echocardiographic diagnostic tools to confirm a non-invasive diagnosis of CP.
The duration of follow up was only four months. Some comparative research had follow-up periods of up to six years. Most patients who develop CP, do so in a period of three to four months. There may however be patients who will only develop constriction after four months. Research to address this aspect may be valuable.
Corticosteroids were used as part of the standard therapy in all patients. However, subsequent to the initiation of the research, the IMPI trial brought to light their findings that corticosteroids should not be used in TB pericarditis in HIV-infected patients. The South African National TB guidelines published in 2014 still advised the use of corticosteroids in all patients and the findings of the IMPI trial had not yet been incorporated into current South African National Tuberculosis Management Guidelines.[@R15]
Conclusion {#s4}
==========
Based on current research, the use of colchicine in addition to standard antituberculous therapy cannot be advised in the context of TB pericarditis in the HIV-positive population. The jury is still out on which adjuvant strategies may prove to be beneficial in the prevention of CP, especially in the HIV-coinfected subgroup. Based on observations from this research and some other studies, routine pericardiocentesis until dryness with extended drainage may prove to be the long-awaited solution to the common dilemma of post-TB CP.
Our special thanks go to Ms Kassandra Barnard, Dr Danie Steenkamp and Dr Pieter van Der Bijl for technical assistance and quality control with the echocardiographic assessment of study participants. Our sincere gratitude goes to Mr Shaun Zeelie for supplying and monitoring the treatment of the participants.
| {
"pile_set_name": "PubMed Central"
} |
[^1]: **Session:** 167. Preclinical Study with New Antibiotics and Antifungals
*Friday, October 6, 2017: 12:30 PM*
| {
"pile_set_name": "PubMed Central"
} |
The management of trauma and other orthopedic diseases has progressed by leaps and bounds during the last 40 years. Similarly, the cost of treatment has increased proportionately with the advent of joint replacement and of course costly trauma implants. India is a land of contrasts where there are 5-star hospital treatment facilities as well as quacks treating everything in villages and smaller towns. We do not have enough resources to treat all patients to the Western standards even if it may be desirable. In India, we need more of the options which will restore the function, are less costly and more dependent on biology rather than implants. This book does exactly the same.
Prof. Tuli, in his inimitable way, has described various biological options for these conditions. Book is written in lucid and clear language. He has given several useful tips and practical solutions for complex problems. The basics and fundamental principles have been brought out well. Many useful but forgotten operations such as Huntington\'s procedure, Papineau\'s bone grafting, and pantalar arthrodesis, which give functional results, have been well described.
The book is divided into two sections. Each chapter is divided into headings and subheadings. Each chapter is giving key points of the condition. The chapter on bone and joints infection has been written with an aim to emphasize the importance of early diagnosis and treatment. Some common congenital dysplasias and deformities and common tumors have also been discussed. Chapter 10 discusses poliomyelitis whose new cases are not seen, but the old cases may still be there. Section 2 incorporates regional orthopedic conditions where several common painful but common conditions are discussed. Glenoplasty for recurrent dislocation of shoulder, which was popularized by the author also gets a place in this book.
This book is recommended for young and practicing orthopedic surgeons. It will also be a useful revision book for postgraduate students.
| {
"pile_set_name": "PubMed Central"
} |
1. Introduction {#sec1-metabolites-07-00038}
===============
Duchene muscular dystrophy (DMD) is a neuromuscular disorder characterized by progressive loss of muscle mass that renders the mobility of its victims to wheelchair before the age of 12 \[[@B1-metabolites-07-00038]\]. Genetically, X-linked Duchenne muscular dystrophy is caused by the absence of dystrophin protein and is a lethal inherited disorder that mainly affects male. The disease progresses with a gradual development of respiratory insufficiency, cardiomyopathy, and skeletal muscle weakness, frequently causing death by the late teens or early twenties \[[@B2-metabolites-07-00038],[@B3-metabolites-07-00038]\]. Of the three naturally-occurring mammalian (murine, feline, and canine) DMD models \[[@B4-metabolites-07-00038]\], in addition to the two non-mammalian (zebrafish and *Caenorhabditis elegans*) models \[[@B5-metabolites-07-00038]\], the canine models are thought to be the most suitable ones for studies because of their pathological resemblance and clinical relevance to human patients of DMD \[[@B5-metabolites-07-00038],[@B6-metabolites-07-00038]\]. Like DMD, the genetically homologous Golden Retriever Muscular Dystrophy (GRMD) dog model of DMD is characterized by muscle necrosis, progressive paralysis, differential weakness of flexor and extensor muscles muscles, and mixed atrophy and hypertrophy of specific skeletal muscles \[[@B7-metabolites-07-00038],[@B8-metabolites-07-00038]\]. The GRMD phenotype includes marked atrophy of the long digital extensor (LDE) as compared to sibling unaffected normal dogs, while the biceps femoris (BF), a member of the hamstring muscle group, undergoes less pronounced atrophy \[[@B7-metabolites-07-00038],[@B8-metabolites-07-00038]\].
Recent studies have begun to implicate metabolic defects in muscular dystrophies, including DMD. Qualitative and quantitative lipid analysis of DMD, Becker's Muscular Dystrophy, facioscapulohumeral muscular dystrophy (FSHD), and limb girdle muscular dystrophy-2B (LGMD-2B) patient biopsies found significant increases in glycogen \[[@B9-metabolites-07-00038]\]. Fatty acids were also altered in DMD, BMD, FSHD, and LGMD-2B, although DMD biopsies exhibited unique increases in cholesterol in DMD patients \[[@B9-metabolites-07-00038]\]. A recent microarray analysis of GRMD muscle identified the first evidence of potential alterations in metabolism, specifically alterations in genes associated with lipid metabolism and energy production \[[@B10-metabolites-07-00038]\]. Impairment of bioenergetics has recently been reported in congenital muscle dystrophy Type 1A (MDC1A) and Leigh Syndrome, linking metabolism to disturbances in skeletal muscle cell apoptosis \[[@B11-metabolites-07-00038]\]. In the present study, we undertook a non-targeted metabolomics analysis to determine how GRMD skeletal muscle compared to age-matched control muscle metabolically using gas chromatography-mass spectrometry (GC-MS) to identify underlying metabolic defects specific for GRMD skeletal muscle.
2. Results {#sec2-metabolites-07-00038}
==========
2.1. Determination of Metabolomics Changes in the GRMD Biceps Femoris vs. Control (t-Test) {#sec2dot1-metabolites-07-00038}
------------------------------------------------------------------------------------------
The first statistical analysis we ran was designed to determine if there were significant alterations in the GRMD BF muscle compared to controls. To do this, we analyzed biopsy samples from GRMD-affected (and non-affected sibling) dogs at 6 months of age, at which time phenotypic changes have occurred \[[@B7-metabolites-07-00038],[@B8-metabolites-07-00038]\]. Previous studies identified marked atrophy of the LDE compared to sibling unaffected normal dogs, while the BF, a member of the hamstring muscle group, undergoes less pronounced atrophy \[[@B7-metabolites-07-00038],[@B8-metabolites-07-00038]\]. The limited number of samples run (6 GRMD and 4 controls) are due to the cost of housing and maintaining ill dogs for 6 months and the limited number of animals in the colony. A *t*-test was then performed on the metabolites to determine BF metabolites affected. Analysis of the less atrophied, early stage disease BF yielded 75 metabolites muscle ([Figure S1A](#app1-metabolites-07-00038){ref-type="app"}), including 19 imputed ([Figure S1B](#app1-metabolites-07-00038){ref-type="app"}), which upon partial least squares discriminant analysis (PLS-DA) produced separation of the control and GRMD groups from the first two components ([Figure 1](#metabolites-07-00038-f001){ref-type="fig"}A). PLS-DA VIP analysis identified pyrophosphate (VIP Score 2.65) and stearamide (VIP Score 2.24) with the highest contributions to this signature ([Figure 1](#metabolites-07-00038-f001){ref-type="fig"}B). *t*-Test analysis of the two groups identified eight significant metabolites (carnosine, fumaric acid, steramide, myoinositol-2-phosphate, lactamide, proline, oleic acid, glutamic acid, [Figure 1](#metabolites-07-00038-f001){ref-type="fig"}C).
We next investigated what these eight significant metabolites had in common in terms of metabolic pathways. To do this, we determined their enrichment, or likelihood of being significantly altered by random chance, in metabolic pathways. Pathway analysis of these eight significant metabolites in GRMD BF identified four affected pathways, including: (1) arginine and proline metabolism; (2) alanine, aspartic acid, and glutamic acid metabolism; (3) butanoate metabolism, and (4) histidine metabolism, with false discovery rates (FDR) of 5--9% ([Figure 2](#metabolites-07-00038-f002){ref-type="fig"}A). The FDR estimates the likelihood that a conclusion that a relationship exists, whereas in reality it does not for an individual comparison \[[@B12-metabolites-07-00038]\]. Its use in the the current context represents the likelihood that the pathways identified as significantly enriched may be false and is generally set at 5%. In the current studies, we have presented data with just higher FDR, as we are necessarily analyzing a smaller number of biological biopsies. Enrichment analysis of the BF *t*-test significant metabolites against a disease-associated metabolite set (blood) identified \> 20-fold enrichment in histidine metabolism and alanine metabolism ([Figure 2](#metabolites-07-00038-f002){ref-type="fig"}B). Enrichment of arginine and proline metabolism had the lowest p value (p = 5.88 × 10^−4^, FDR = 4.7 × 10^−2^) enriched \> 15-fold ([Figure 2](#metabolites-07-00038-f002){ref-type="fig"}B). To further identify metabolic pathways to which our significant metabolites may be related, we compared them to changes in other disease sets, including urine, CSF, and location-based sets for further clues to the metabolic pathways affected. Metabolic enrichment BF *t*-test significant metabolites were analyzed against disease-associated metabolite sets (urine, CSF) and location-based metabolite sets and identified nearly 80-fold enrichment for carnosinuria/carnosinemia ([Figure S2A](#app1-metabolites-07-00038){ref-type="app"}), \> 10-fold for spinocerebellar degeneration ([Figure S2B](#app1-metabolites-07-00038){ref-type="app"}), and 5-fold for skeletal muscle ([Figure S2C](#app1-metabolites-07-00038){ref-type="app"}). Of the eight (8) *t*-test significant metabolites, five (5) were significantly decreased (stearamide, carnosine, fumaric acid, lactamide, and myoinositol-2-phosphate), while three (3) were significantly increased (oleic acid, glutamic acid, and proline) ([Figure 2](#metabolites-07-00038-f002){ref-type="fig"}C).
2.2. GRMD vs. Control Long Digital Extensor Muscle vs. Control (t-Test) {#sec2dot2-metabolites-07-00038}
-----------------------------------------------------------------------
The second statistical analysis we ran was designed to determine if there were significant alterations in the more atrophied, later stage disease GRMD LDE compared to controls. The LDE functions to flex the tibiotarsal joint and also serves as a digital extensor \[[@B13-metabolites-07-00038]\] and is significantly atrophied in GRMD to a greater extent than the BF muscle \[[@B7-metabolites-07-00038],[@B8-metabolites-07-00038]\]. In this analysis, we compared biopsy samples from GRMD-affected and non-affected sibling LDE samples at 6 months of age and performed a *t*-test to determine if there were any significant alterations in the GRMD LDE muscle compared to controls ([Figure 3](#metabolites-07-00038-f003){ref-type="fig"}). Of the 75 metabolites identified in LDE muscles ([Figure S3A](#app1-metabolites-07-00038){ref-type="app"}), 14 were imputed ([Figure S3B](#app1-metabolites-07-00038){ref-type="app"}), separation on partial least squares discriminant analysis (PLS-DA) was seen ([Figure 3](#metabolites-07-00038-f003){ref-type="fig"}A). PLS-DA VIP analysis identified inosine-5'-monophosphate (VIP Score 3.91) and 3-phosphoglyceric acid (VIP Score 3.08) with the highest contributions to this signature. *t*-Test analysis of the two groups identified two (2) significant metabolites (phosphoglyceric aid and inosine-5'-monophosphate ([Figure 3](#metabolites-07-00038-f003){ref-type="fig"}C), which were both decreased ([Figure 3](#metabolites-07-00038-f003){ref-type="fig"}D).
2.3. GRMD LDE vs. GRMD BF vs. Control LDE vs. Control BF (ANOVA) {#sec2dot3-metabolites-07-00038}
----------------------------------------------------------------
A third statistical analysis analysis was performed to identify if there were any common changes in the more atrophied later stage disase GRMD LDE and less atrophied, early stage GRMD BF compared to control LDE and BF dogs (four groups total). To do this, an ANOVA analysis was performed to identify any common changes in GRMD muscles, a question that the first two *t*-tests did not address ([Figure 4](#metabolites-07-00038-f004){ref-type="fig"}). By PLS-DA analysis, considerable overlap of the BF metabolic profiles (GRMD and Control) and LDE (GRMD and Control) by both component 1 and component 2 was seen when all four groups were analyzed together ([Figure S5](#app1-metabolites-07-00038){ref-type="app"}). Of the 75 metabolites from each group ([Figure S4](#app1-metabolites-07-00038){ref-type="app"}), imputed as described above ([Figure S1B, Figure S3B](#app1-metabolites-07-00038){ref-type="app"}), eight (8) were significant, including carnosine, fumaric acid, stearamide, myoinositol-2-phosphate, lactamide, and phosphoric acid ([Figure 4](#metabolites-07-00038-f004){ref-type="fig"}A). Pathway analysis identified three pathways as significant: (1) TCA cycle; (2) alanine, aspartic acid, glutamic acid metabolism; and (3) beta-alanine metabolism ([Figure 4](#metabolites-07-00038-f004){ref-type="fig"}B), with \> 6-fold enrichment for protein biosynthesis ([Figure 4](#metabolites-07-00038-f004){ref-type="fig"}C). The unique significant metabolite in this analysis compared to the *t*-test run on the BF group was phosphoric acid, which was significantly less in the GRMD BF as compared to control and GRMD LDE groups ([Figure 5](#metabolites-07-00038-f005){ref-type="fig"}A). While ANOVA analysis of stearamide revealed a single difference between control and GRMD BF muscles only ([Figure 5](#metabolites-07-00038-f005){ref-type="fig"}B), the significant decreases in lactamide and myoinositol-2-phosphate seen in GRMD BF were significantly attenuated in GRMD LDE ([Figure 5](#metabolites-07-00038-f005){ref-type="fig"}C,D).
In the preceeding three analyses, pathway analyses identified that beta-alanine metabolism, arginine/proline metabolism, and the TCA cycle were the pathways most affected based on both the *t*-test and ANOVA significant metabolites identified in the GRMD BF ([Figure 2](#metabolites-07-00038-f002){ref-type="fig"}A, [Figure 4](#metabolites-07-00038-f004){ref-type="fig"}B). To further delineate the context of these alterations in context of their metabolic pathways, we next documented the changes in the pathways themselves ([Figure 6](#metabolites-07-00038-f006){ref-type="fig"}). In the beta-alanine and arginine/proline metabolic pathways, significant decreases in carnosine were identified ([Figure 6](#metabolites-07-00038-f006){ref-type="fig"}A), along with significant increases in glutamic acid and proline ([Figure 6](#metabolites-07-00038-f006){ref-type="fig"}B,C). In BF, significant decreases in fumaric ([Figure 7](#metabolites-07-00038-f007){ref-type="fig"}A) and malic acid ([Figure 7](#metabolites-07-00038-f007){ref-type="fig"}B) were identified, with decreases in citric/isocitric acid ([Figure 7](#metabolites-07-00038-f007){ref-type="fig"}C) and succinic acid also seen (not significant) ([Figure 7](#metabolites-07-00038-f007){ref-type="fig"}D).
2.4. Integrated Metabolomics Analysis {#sec2dot4-metabolites-07-00038}
-------------------------------------
The two primary approaches in functional analysis of metabolomics consist of metabolite set enrichment \[[@B14-metabolites-07-00038]\] and metabolic pathway analysis \[[@B15-metabolites-07-00038]\], as performed in the described in the prior two analyses \[[@B16-metabolites-07-00038]\]. A new approach that has been developed utilizes analysis of metabolomics experiments in combination with transcriptomics studies to exploit the models from KEGG (Kyoto Encyclopedia of Genes and Genomes) pathways using a method called integrated pathway analysis \[[@B15-metabolites-07-00038]\]. This integrated pathway analysis combining evidence from both metabolite concentrations and gene expression was developed to pinpoint pathways involved in the underlying biological processes \[[@B15-metabolites-07-00038]\]. In the present study, we performed an integrated pathway analysis using GRMD BF/LDE (6 months of age) significant metabolites caronosine, oleic acid, pyrophosphate, 3-phosphoglyceric acid, campesterol ([Table S2](#app1-metabolites-07-00038){ref-type="app"}) and 55 significantly altered genes from the GRMD gastrocnemius (6 months of age) ([Table S3](#app1-metabolites-07-00038){ref-type="app"}). This analysis was the first time that both metabolomics and microarray evidence were compared in combination, to determine how these complementary datasets mights point to common alterations in metabolism.
The integrated pathway analysis gives two values for each metabolic pathway: enrichment and topology. Topology uses the structure of a given pathway to identify the relative importance of genes and compounds. In our first analysis, we analyzed the data sets to look at gene-metabolite pathways and found the most enriched/topology pathways to be "Biosynthesis of unsaturated fatty acids", "Histidine metabolism", and "β-Alanine metabolism" ([Figure 8](#metabolites-07-00038-f008){ref-type="fig"}A). When the analysis was run in a gene-centric manner so that the transcriptomics were the main focus of analysis, we identified the most enriched/topology pathways to be "cardiac muscle contraction", "hypertrophic cardiomyopathy", and "tight junction" ([Figure 8](#metabolites-07-00038-f008){ref-type="fig"}B). These findings expand upon the conclusions of the initial published microarray analysis of the GRMD heart and gastrocnemius tissues, finding a common alteration in lipid metabolism, energy production, and inflammatory responses \[[@B10-metabolites-07-00038]\] and the primary finding that the highest induced transcript in the GRMD gastrocnemius muscle was SPP1 \[[@B10-metabolites-07-00038]\], previously shown to be an indicator of muscle injury prior to microarray analyses \[[@B17-metabolites-07-00038]\].
3. Discussion {#sec3-metabolites-07-00038}
=============
The metabolic component of DMD has not been previously analyzed directly. At the transcriptional level, studies of GRMD skeletal, but not cardiac, muscle have been described as consistent with a "metabolic crisis" \[[@B18-metabolites-07-00038]\], with the down-regulation of dozens of energy production-associated mRNAs and multiple electron transport component mRNAs in the TCA cycle \[[@B18-metabolites-07-00038]\]. In the present study, we investigated two GRMD pelvic limb muscles, the more proximal BF muscle which is a member of the hamstring group and is relatively spared and the more distal LDE that undergoes progressive atrophy. We hypothesized that the BF and LDE would have different metabolic profiles consistent with their milder/earlier versus more severe/chronic stages of disease, respectively. There were several differences in metabolism, with nine significantly altered metabolites in the GRMD BF muscle (eight by *t*-test, [Figure 2](#metabolites-07-00038-f002){ref-type="fig"}C, one additional by ANOVA, [Figure 4](#metabolites-07-00038-f004){ref-type="fig"}A) as compared to only 2 in the more chronically affected GRMD muscle (LDE). In addition, we identified for the first time in this system significant increases in metabolites involved in β-alanine metabolism and arginine/proline metabolism ([Figure 6](#metabolites-07-00038-f006){ref-type="fig"}).
There are non-specific conditions that have been linked to muscle catabolism in humans, which may lower muscle glutamic acid levels, including emphysema \[[@B19-metabolites-07-00038]\], COPD \[[@B20-metabolites-07-00038]\], elderly fraility \[[@B21-metabolites-07-00038]\], and in acute sepsis and surgical trauma \[[@B19-metabolites-07-00038]\]. Several studies have been suggestive of these metabolic changes in DMD, using NMR analyses. In the mdx mouse, recent studies using ^1^H-magnetic resonance spectroscopy to study the metabolic profile of quadriceps and diaphragm muscle identified increases in glutamic acid and carnosine \[[@B22-metabolites-07-00038]\]. Other studies using NMR analysis of vastus lateralis muscle biopsies from DMD patients (N = 11 vs. 7 normal controls) identified significant decreases in glutamate/glutamine (not discernable from each other) \[[@B23-metabolites-07-00038]\].
Skeletal muscles most abundantly require glutamic acid \[[@B24-metabolites-07-00038]\]. Glutamic acid is the only amino acid that is actively taken up from the circulation by the muscles in the post-absorptive state \[[@B25-metabolites-07-00038]\]. Intracellular degradation of myofibrillar proteins also release glutamic acid with large amounts contributing to the muscle glutamic acid pool \[[@B26-metabolites-07-00038]\]. The role of glutamic acid in muscular dystrophy was studied in the early 1960s in rabbit models (due to vitamin E deprivation) \[[@B27-metabolites-07-00038]\]. In these studies, body weight loss and the onset of terminal disease was accelerated when animals were given exogenous glutamic acid \[[@B27-metabolites-07-00038]\]. While the reasons for and sources of elevated BF glutamic acid levels (\~2.5-fold controls) are not clear ([Figure 6](#metabolites-07-00038-f006){ref-type="fig"}), their presence could potentially play a role in the eventual worsening of the disease process.
The role of oleic acid in DMD disease has been investigated in the mdx mouse model, where dietary oleic acid was replaced with the n-3 poly-unsaturated fatty acid (PUFA), alpha-linolenic acid (ALA) \[[@B28-metabolites-07-00038]\]. Interestingly, these studies found that shifting to a high PUFA diet resulted in the expected higher n-3 PUFA content in muscle, while reducing arachidonic acid content in skeletal muscle phospholipids, but also made the mdx muscle more susceptible to sarcolemmal leakiness \[[@B28-metabolites-07-00038]\]. When DMD patient erythrocytes have been studied for long chain fatty acid composition, decreases in oleic, linoleic, and arachidonic acids have been identified \[[@B29-metabolites-07-00038]\]. Where these fatty acids are being utilized, however, has not been investigated, nor have their levels in skeletal muscle been determined, to our knowledge. In the present study, the significantly elevated oleic acid (18:0) in GRMD BF muscle ([Figure 2](#metabolites-07-00038-f002){ref-type="fig"}C) may represent alternate energy substrate utilization and be directly related to the decreased oleic acid (18:0) metabolite stearamide ([Figure 2](#metabolites-07-00038-f002){ref-type="fig"}C). Alternatively, the significant increase and GRMD BF oleic acid may be a consequence of a greater accumulation of muscle triacylglyceride (TAG), a storage form of oleic acid. In the GRMD model, fatty infiltration is rare before 4 months of age, but common in those \> 13 months of age \[[@B8-metabolites-07-00038]\], so would be expected to be present in the 6 month old muscle investigated here. If this is the case, its possible that mitochondrial dysfunction may be responsible, as mitochondrial dysfunction has been reported to induce lipid droplet formation (including triacyglycerol and cholesterol esters) in reponse to stress \[[@B30-metabolites-07-00038]\].
The primary amides of oleic (18:0), palmitic, palmitoleic, eladic, and linoleic acid were identified in human plasma in 1989, although their significance was not understood \[[@B31-metabolites-07-00038]\]. Subsequently, other fatty acid amides (including stearamide) have been isolated from human tear gland secretions \[[@B32-metabolites-07-00038]\]. The role of fatty acid amides is not completely clear, but the fact that the arachidonic acid metabolite arachidonamide is the best substrate for the membrane-bound serine hydrolase fatty acid amide hydrolase (FAAH) suggests a role in signaling \[[@B33-metabolites-07-00038]\]. In muscle, primary fatty acid amides reduced sarco/endoplasmic reticulum Ca^2+^-ATPase (SERCA) activity in the micromolar (μM) range, although stearamide inhibition was relatively minimal \[[@B34-metabolites-07-00038]\]. The significance of the reduced GRMD BF stearamide (0.2-fold control muscle) is not clear but may reflect altered signaling pathways within affected skeletal muscle not seen in unaffected GRMD muscle (LDE, [Figure 5](#metabolites-07-00038-f005){ref-type="fig"}B).
Carnosine (β-alanyl-[L]{.smallcaps}-histidine) is a dipeptide found to be highly concentrated in skeletal muscles \[[@B35-metabolites-07-00038]\]. At least three roles have been attributed to carnosine, including: a lactic acid neutralizer under anaerobic conditions \[[@B36-metabolites-07-00038]\], a potent scavenger of singlet oxygen protecting the cells from oxidative damage \[[@B37-metabolites-07-00038]\], and a physiological activator of myofibrillar ATPase activity \[[@B38-metabolites-07-00038]\]. The significant decrease in GRMD BF carnosine (0.4-fold control BF levels, [Figure 2](#metabolites-07-00038-f002){ref-type="fig"}C) could set the stage for eventual muscle damage (e.g. due to lactic acid or oxidative damage). It may also be a cause of limited myosin ATPase activity, which relies on carnosine \[[@B39-metabolites-07-00038]\], and be a contributing factor to muscle fatigue. It has been reported that vitamin B6 plays a pivotal role in maintaining the carnosine concentrations in the skeletal muscles \[[@B40-metabolites-07-00038],[@B41-metabolites-07-00038]\]; however, vitamin B6 (pyridoxine) was not measured in the present study in order to determine if this was an underlying cause. Interestingly, GRMD BF β-alanine levels were significantly increased (2.9-fold of control levels), while carnosine was decreased ([Figure 6](#metabolites-07-00038-f006){ref-type="fig"}). Since carnosine is hydrolyzed to its constituent amino acids (β-alanine and histidine) by the enzyme carnosinase \[[@B42-metabolites-07-00038]\], increased carnosinase may be a contributing cause of the elevated GRMD β-alanine and decreased carnosine.
In the present study, we identified elevated proline and factors associated with the arginine/proline metabolic pathway. Arginine is taken up by cells, metabolized to proline, which can be used as a precursor for collagen \[[@B43-metabolites-07-00038]\]. This is significant because a major finding in chronic myopathies, such as DMD, is collagen deposition/fibrosis. Previous studies of the GRMD cranial sartorius muscle identified patchy increases in endomysial space as early as 6 months of age \[[@B8-metabolites-07-00038]\]. In previous studies, the GRMD dog muscles were found to have higher levels of hydroxyproline, a biomarker of fibrosis and collagen content, across muscles (cranial sartorius, vastus lateralis, long digital extensor, and lateral head of the gastrocnemius) compared to controls at 8--9 months of age \[[@B44-metabolites-07-00038]\]. Similarly, increased fibrosis has been reported in 12 month old GRMD carrier hearts, along with inflammation and/or fatty changes, compared to control hearts \[[@B45-metabolites-07-00038]\]. Since elevated levels of proline are detected in the pathophysiology of fibrosis, including liver fibrosis \[[@B46-metabolites-07-00038]\], sarcoidosis \[[@B47-metabolites-07-00038]\], and rheumatoid arthritis \[[@B47-metabolites-07-00038],[@B48-metabolites-07-00038]\], the increases in the GRMD proline/arginine metabolic pathway could be linked to increases in inflammation-related collagen synthesis/muscle fibrosis known to the be a central part of the pathogensis of DMD.
There are weaknesses and assumptions in the present study to consider. These include the limited number of biological replicates analyzed per group, due to the expense and rarity of the GRMD model itself. In particular, the identification of metabolites that were decreased, but not significant (e.g., citric/isocitric acid, and succinic acid, [Figure 7](#metabolites-07-00038-f007){ref-type="fig"}) may reflect either a lack of samples (i.e., power), the variability of the phenotype, or may actually not be different. Therefore, caution should be exercised in interpreting these data and the findings should not be used as evidence to support a hypothesis. Additionally, the data presented here may be a consequence of the observed muscular dysfunction, and not a cause of the effects on the TCA cycle intermediate pools. For example, brief periods of endurance training can reduce TCA intermediate pools, as does overweight-to-obese sedentary lifestyles \[[@B49-metabolites-07-00038]\], whereas dynamic exercise can increase TCA intermediates \[[@B50-metabolites-07-00038]\] in the absence of disease.
Summary
-------
While transcriptional analysis of GRMD skeletal muscle has suggested a metabolic component by the down-regulation of dozens of energy production-associated mRNAs and multiple electron transport component mRNAs in the TCA cycle, the present study is the first to our knowledge to analyze the metabolic component directly. We utilized untargeted metabolomics analysis to assess two GRMD pelvic limb muscles, the more proximal BF muscle which is a member of the hamstring group and is relatively spared, and the more distal LDE that undergoes progressive atrophy. We hypothesized that the BF and LDE would have diffent metabolic profiles consistent with their milder/earlier versus more severe/chronic stages of disease, respectively. There were eight (8) significantly altered metabolites in the milder/earlier stage disease GRMD BF muscle, including five (5) which were decreased (stearamide, carnosine, fumaric acid, lactamide, and myoinositol-2-phosphate) and three (3) which were increased (oleic acid, glutamic acid, and proline). Pathway analysis of these metabolites identified enrichment for arginine/proline metabolism and β-alanine metabolism. Identification of alterations across all four groups (GRMD BF, Control BF, GRMD LDE, Control BF) additionally identified phosphoric acid. Overall, many Krebs cycle intermediates were significantly decreased (e.g., malic acid, fumaric acid) and other intermediates were decreased but not significant (e.g., citric/isocitric acid, and succinic acid). These findings support the first direct evidence for decreases in TCA cycle intermediates and suggest a mechanism of altered energy metabolism may be underlying the observed dysfunction. We also identified elevated levels of the monounsaturated fatty acid oleic acid (18:0) in GRMD BF muscle, which has been related to increased inflammation in DMD that would be expected in earlier stages of disease, with associated necrosis.
4. Materials and Methods {#sec4-metabolites-07-00038}
========================
4.1. Golden Retriever Muscular Dystrophy Dog Model {#sec4dot1-metabolites-07-00038}
--------------------------------------------------
All GRMD dogs used in this study were from a colony at the University of North Carolina at Chapel Hill (UNC-CH), now located at Texas A&M University. Dogs were produced by mating a heterozygote affected male to a carrier female, such that \~ 25% each of the pups should be either normal males, affected males, heterozygous (carrier) females, or homozygous (affected) females. Normal females are not produced with this mating. Blood creatine kinase levels taken shortly after birth were used to diagnose neonates with GRMD \[[@B51-metabolites-07-00038]\], along with PCR, as previously described \[[@B52-metabolites-07-00038]\]. Samples of the BF and LDE muscles were taken at necropsy from GRMD (n = 6; 4 males and 2 females) and littermate wild type (control) (n = 4; all male) dogs at 6 months of age immediately following barbiturate euthanasia. These dogs were part of a larger cohort that was studied previously with a genome wide association study (GWAS) \[[@B53-metabolites-07-00038]\]. Dogs were used and cared for according to principles outlined in the National Research Council Guide for the Care and Use of Laboratory Animals. The experiments were approved by the Institutional Animal Care and Use Committee (IACUC) review board at UNC-CH and performed in accordance with federal guidelines.
4.2. Metabolomics Determination by GC--MS Instrumentation {#sec4dot2-metabolites-07-00038}
---------------------------------------------------------
Bicep femoris (BF) and long digital extensor (LDE) muscles were flash frozen after biopsy in liquid nitrogen and stored at --80 °C. A fraction of the muscle was then weighed (\~25--50 mg wet weight), finely minced, and quickly added to fresh ice-cold pre-made buffer (50% acetonitrile, 50% water, 0.3% formic acid) at a standard concentration of 25 mg/475 μL, homogenized on ice for 20--25 seconds and stored at −80 °C. The samples were "crash" deproteinized by methanol precipitation and then spiked with D27-deuterated myristic acid (D27-C14:0) as an internal standard for retention-time locking only and dried. The deuterated standard was not used for relative quantification/correction of analyte peak responses due to the limited number of samples processed in single batches for the different tissues. The derivatized trimethylsilyl (TMS)-D27-C14:0 standard retention time (RT) was set at \~16.727 min. Reactive carbonyls were stabilized at 50 °C with methoxyamine hydrochloride in dry pyridine. Metabolites were made volatile with TMS groups using N-methyl-N (trimethylsilyl) trifluoroacetamide or MSTFA with catalytic trimethylchlorosilane at 50 °C. GC/MS methods generally follow those of Roessner et al., \[[@B54-metabolites-07-00038]\], Fiehn et al., \[[@B55-metabolites-07-00038]\], and Kind et al., \[[@B56-metabolites-07-00038]\], and used a 6890N GC connected to a 5975B inert single-quadrupole MS (Agilent Technologies, Santa Clara, CA, USA). The two wall-coated, open-tubular (WCOT) GC columns connected in series were both from J&W/Agilent (part 122--5512), DB5-MS, 15 meters in length, 0.25 mm in diameter, with an 0.25 μm luminal film. Positive ions generated with conventional electron-ionization (EI) at 70 eV were scanned broadly from 600 to 50 m/z in the detector throughout the 45-min cycle time.
Data were acquired using MSD ChemStation (Agilent Technologies) and metabolites annotated based on their mass fragmentation patterns and RT. Raw data formatted files were exported for further analysis in Automatic Mass Spectral Deconvolution and Identification Software (AMDIS version 2.72, build 140.24, freeware developed by Steve Stein, W. Gary Mallard, and their coworkers at National Institute of Standards and Technology or NIST \[[@B57-metabolites-07-00038],[@B58-metabolites-07-00038],[@B59-metabolites-07-00038]\]). Deconvoluted spectra were annotated as metabolites, to the extent possible, using an orthogonal approach that incorporates both RT from GC and the fragmentation pattern observed in EI-MS. Peak annotation is based primarily on our own RT-locked spectral library of metabolites. The library is built upon the Fiehn GC/MS Metabolomics RTL Library (a gift from Agilent, part number G1676-90000), Golm Metabolome Library \[[@B56-metabolites-07-00038]\] (courtesy of Joachim Kopka and coworkers at the Max Planck Institute of Molecular Plant Physiology, Golm, Germany \[[@B60-metabolites-07-00038]\]), the Wiley 9th-NIST 2011 commercial library (Agilent G1730-64000), and other spectral libraries. Once annotation was complete, a cross-tabulated spreadsheet was created, listing the integrated peak area for each metabolite versus sample identity. This was accomplished using a custom Visual Basic program in Microsoft Excel that grouped peaks across samples based on identical metabolite annotation and RT proximity. Peak alignment across samples was further confirmed using SpectConnect \[[@B61-metabolites-07-00038]\] to assess similarity in spectral fragmentation patterns and by manual curation. The raw, transformed, and sorted data used for each of the three comparisons in the metabolomic analyses (next) can be found in [Table S1](#app1-metabolites-07-00038){ref-type="app"}. The data obtained in this study is accessible at the NIH Common Fund's Data Repository and Coordinating Center (supported by NIH grant, U01-DK097430) website, <http://www.metabolomicsworkbench.org>.
4.3. Metabolomic Statistical Analysis {#sec4dot3-metabolites-07-00038}
-------------------------------------
Metaboanalyst (v3.0) run on the statistical package R (v2.14.0) used metabolite peak areas (as representative of concentration) \[[@B16-metabolites-07-00038],[@B62-metabolites-07-00038],[@B63-metabolites-07-00038],[@B64-metabolites-07-00038]\]. The control group contained 4 biological replicates, while the GRMD group contained 6. If more than one individual in the control group did not have a metabolite detected in a group (of 4 total), that metabolite was excluded from further analysis. Likewise, if more than 2 individuals in the GRMD group did not have a metabolite detected in a group (of 6 total), that metabolite was excluded. In groups with values missing, the lowest value of that group was used to impute those values. These data were scaled using the Pareto scaling feature. Initially, a *t*-test comparing control to GRMD muscles was performed on both the BF and LDE muscle. In a second analysis, a One-Way Analysis of Variance (ANOVA) and Fisher LSD post-hoc test across the groups (control BF, GRMD BF, control LDE, GRMD LDE) was performed using Metaboanalyst v3.0. The data were analyzed by both unsupervised principal component analysis (PCA) and supervised partial least squares discriminant analysis (PLS-DA) to further determine the metabolites that separated groups. The specific metabolites contributing most significantly to the differences identified by PLS-DA between control BF and GRMD BF (or control LDE and GRMD LDE) muscles were determined using the variable importance in projection (VIP) analysis in the Metaboanalyst environment. Both *t*-test and ANOVA significant metabolites were matched to metabolomics pathways using the Pathway Analysis feature in Metaboanalyst 3.0. Only metabolites identified in the one-way ANOVA were included. All data from this study are available in [Table S1](#app1-metabolites-07-00038){ref-type="app"}. All heat maps were generated using the heat map analysis feature in Metaboanalyst v3.0. Data are presented as mean +/- SEM, unless otherwise indicated.
4.4. Integrated Microarray and Metabolomics Statistical Analysis {#sec4dot4-metabolites-07-00038}
----------------------------------------------------------------
The metabolites used in the integrated microarray analysis included *t*-test significant and VIP \> 2 metabolites from BF and LDE, limited to those with HMDB designations (carnosine, oleic acid, pyrophosphate, 3-phosphoglyceric acid, and campesterol) ([Table S2](#app1-metabolites-07-00038){ref-type="app"}). The microarry datasets used in parallel were accessed from GEO (GSE68626) and extracted using the GEO2R online tool, which were recently published \[[@B10-metabolites-07-00038]\]. Briefly, datasets from skeletal muscle (gastrocnemius) tissue from normal golden retriever (or golden retriever muscular dystrophy model) as were used in the current analysis. The normal golden retriever medial head of the gastrocnemius muscles (N = 3) were from animals ages 6, 6, and 7.5 months (accession number GSM1677427, GSM1677428, GSM1677429, respectively). The GRMD gastrocnemius medial head of the gastrocnemius muscles (N = 3) were from animals ages 6.75, 6, and 6 months (GSM1677430, GSM1677431, GSM1677432, respsectively). Data included GRMD mRNA expressed \> 1.9 fold (vs. control) and \< −1.3 fold (found in [Table S3](#app1-metabolites-07-00038){ref-type="app"}).
4.5. Other Statistical Analysis {#sec4dot5-metabolites-07-00038}
-------------------------------
Metabolites not found to be significant were analyzed in a post-hoc analysis by a Student's *t*-test in Microsoft Excel (Version 15.34 (170515, Office 365, Seattle, WA, USA) using a one-tailed test assuming the two-samples had heteroscedastic (unequal) variance and plotted in Prism 7.0 (GraphPad Software, Inc., La Jolla, CA, USA).
This work was supported by the National Institutes of Health (R01HL104129 to MW), the Leducq Foundation Transatlantic Networks of Excellence (to MW), the American Heart Association (Post-Doctoral Fellowship to T.P.) and the International Research Support Initiative Program (IRSIP) by HEC Pakistan (to MA).
The following are available online at [www.mdpi.com/2218-1989/7/3/38/s1](www.mdpi.com/2218-1989/7/3/38/s1), Figure S1: Heatmap analysis of all named metabolites from GRMD and control biceps femoris (BF) included in the t-test analysis. N = 4 control BF and N = 6 GRMD BF, Figure S2: Pathway analysis of GRMD biceps femoris t-test significant metabolites, Figure S3: Heatmap analysis of all named metabolites from GRMD and control long digital extensor (LDE) muscle included in the t-test analysis, Figure S4: Heatmap analysis of all named metabolites from GRMD and control long digital extensor (LDE) and biceps femoris (BF) muscle included in the One-Way ANOVA analysis, Figure S5: Untargeted metabolomics analysis of golden retriever muscular dystrophy (GRMD) biceps femoris (BF) and long digital extensor (LDE) muscle, Table S1: Non-targeted metabolomics performed on six month old control and age-matched golden retriever muscular dystrophy (a) biceps femoris used in t-test analysi; (b) long digital extensor used in t-test analysis; (c) long digital extensor or biceps femoris used in ANOVA analysis, Table S2: Significant metabolites included in combined gene and metabolite analysis in MetaboAnalyst3.0, Table S3: T-test Significant mRNA in GRMD by Microarray used in the combined gene and metabolite analysis in MetaboAnalyst 3.0.
######
Click here for additional data file.
J.K., M.W., C.P., M.M. and J.B. conceived and designed the experiments; J.K., A.H., T.P., C.B., S.O., J.B., M.M.and C.N. performed the experiments and wrote the materials and methods; M.A., M.W. J.B., M.M., S.O. C.N. analyzed and interpreted the data; M.A., J.K. and M.W. wrote and edited the manuscript.
The authors declare no conflict of interest.
The following abbreviations are used in this manuscript:
BF
biceps femoris
DMD
Duchenne muscular dystrophy
FDR
false discovery rate
GRMD
golden retriever muscular dystrophy
LDE
long digital extensor
{#metabolites-07-00038-f001}
{#metabolites-07-00038-f002}
{#metabolites-07-00038-f003}
{#metabolites-07-00038-f004}
{#metabolites-07-00038-f005}
{#metabolites-07-00038-f006}
{#metabolites-07-00038-f007}
{ref-type="app"}) and mRNA \>1.9 or \< −1.3 fold from GRMD muscle (downloaded from GEO, as published in *Pediatr Res*. 2016 Apr;79(4):629-36) and listed in [Table S3](#app1-metabolites-07-00038){ref-type="app"} with fold change calculations) were included in the Metaboanalyst integrated analysis.](metabolites-07-00038-g008){#metabolites-07-00038-f008}
| {
"pile_set_name": "PubMed Central"
} |
Background {#Sec1}
==========
According to the provisions of the current European Medical Device Directive 93/42/EEC \[[@CR1], [@CR2]\] and the new European Medical Device Regulation (applicable as from May 2020) \[[@CR3]\], each manufacturer of medical devices has to set up a comprehensive system in order to identify, evaluate and integrate clinical data derived from the field application of a medical device after market access during post-market surveillance (PMS). Small and medium-sized enterprises in the field of medical devices are in need for operable systems for post-market data retrieval in order to enhance their PMS strategies and to be prepared for the growing requirements of the new European Medical Device Regulation.
A wide range of both internal (own quality management and compliant system) and external (scientific databases, medical congresses, internet-based knowledge & experiences, PMS by competent authorities) sources have to be monitored in order to integrate all accessible data about a medical device's safety and performance. Currently, these detailed, continuous searches are still performed manually with a high input of time and personnel resources, making PMS a daunting task.
Literally, an employee has to type all search queries (e.g., "safety AND coronary stent") into the input fields of a large number of different databases, congress sites and public search engines in order to gain a broad, unspecific hit list, interspersed with single, relevant content. This strategy of data retrieval reaches its limits assuming that several search strings have to be applied in order to monitor a whole range of medical devices, each featured by a variety of decisive search questions.
Additionally, each notable medical database uses its own, inherent syntax to specify search queries. This incompatibility between databases and the amount of different search tasks combined with the manual application to a multitude of databases results in low efficiency and a high potential for human error.
Setting up more complex queries in a simple manner by domain experts enables the definition of the topic of interest in a more specific way and circumvents the problem of retrieving irrelevant content.
In the OntoVigilance project (predecessor project of OntoPMS), we developed the Search Ontology (**SON**) v1.0 \[[@CR4]\], a promising approach for an ontology-based specification of complex search queries. The modular architecture of the SON enables the re-use of ontology parts in different use cases as well as a quick and easy adaptation and extension of the ontology according to the specific requirements.
The developed Search Ontology and its application was evaluated in the OntoVigilance project by domain experts. This previous study has proven that the SON is a suitable method for modelling complex queries, but it needed to be optimized to face further requirements of domain experts. On the one hand, the structure of the Search Ontology can be simplified to improve the usability by domain experts. On the other hand, certain extensions are necessary to model all relevant query types. Based on these findings, we further developed and optimized the Search Ontology in the OntoPMS project. The Search Ontology v2.0 is presented in this paper. The SON is generic and can be used in any domain. For the application of the SON in a particular domain, it has to be extended by a domain ontology, such that the classes of the domain ontology are subclasses of the SON classes. We call such domain ontologies Domain-specific Search Ontologies (**dSON**). Whereas SON stands for exactly one ontology (namely the Search Ontology presented in this paper), dSON represents a class or a type of ontologies. Various Domain-specific Search Ontologies can be developed, such as dSON-PMS for the whole PMS domain or dSON-Niti for modelling queries regarding the material Nitinol. In addition, we developed an Excel template to specify the information required to create a dSON, which significantly simplifies the ontology development by domain experts. For the automatic generation of a dSON from the Excel template, we implemented the Search Ontology Generator (SONG). In contrast to the OntoQueryBuilder (OQB) \[[@CR4]\], the SONG generates the complete dSON including all specified queries in the correct query syntax from the Excel template and provides it for external tools (e.g., the search engine). In this way, the search engine can get the complete dSON by accessing the SONG service without any requests or generating queries at search time.
Methods {#Sec2}
=======
In Europe, market access of a medical device based on a CE-mark is granted after a successful so-called "conformity assessment process", which includes passing an extensive series of tests, risks analyses and evaluations of clinical data on the medical device's safety and efficacy. Nevertheless, the behaviour of a medical device over time in broad application can be investigated a priori only in a limited manner. Thus, PMS strategies are set up in order to retrieve and summarize application data of medical devices and to identify residual risks. Expressive search queries are needed to precisely define the topic of interest. The problem is that the manual creation of complex queries requires knowledge of the correct query syntax and is time-consuming and error-prone. This paper focusses on developing the Search Ontology Generator (SONG), a framework for ontology-based specification and generation of powerful search queries by domain experts with less effort and without knowing the query syntax.
Example PMS question {#Sec3}
--------------------
Reports on unfavourable interactions between implant material and patient's tissue have to be identified and evaluated in order to a) control residual and/or unexpected risk, b) determine vulnerable patient subpopulations and c) improve the respective medical implant or material, respectively.
An example PMS question could be to find out the unexpected side effects of the metal alloy Nitinol used for construction of endoscopic clipping systems. A search query has to be constructed covering the different aspects of the PMS question such as "unexpected complication", "type of medical device" (endoscopic clipping system) and "used material" (Nitinol) by suitable search terms (e.g., "Nitinol", "Nickel Titanium" and "NiTi" as synonyms of Nitinol or "unexpected complication", "unforeseeable risk", "adverse event", etc. to describe the complication). Furthermore, it can be necessary to specify terms that should not appear in the text (negation), for instance, to exclude descriptions of preclinical tests or studies (e.g., terms like "animal", "study" and "preclinical"). Finally, the desired terms have to be assembled to a valid search query using supported operators and brackets.
This example is used in the paper for further explanations.
Approach {#Sec4}
--------
Figure [1](#Fig1){ref-type="fig"} illustrates the overall architecture of the SONG environment.Fig. 1SONG and its application. The Domain Expert specifies the dSON in Excel. The SONG Manager uploads/downloads files and tests the service. The SONG Service generates the dSON (including complete search queries) and offers different methods for external applications. The Searcher uses the SONG Search App and can select desired concepts or queries for the execution by a search engine
The Domain Expert specifies information required to generate a dSON (including search queries) using an easily applicable Excel template.
The SONG Manager uploads/downloads files and tests the service using the SONG Config App. The SONG service generates the dSON (in our example the dSON-Niti) in OWL and JSON format out of the Excel template, allows adding new entities to the ontology and provides the generated dSON for external tools, especially for search apps. After each file upload (Excel or OWL) or after an adding of a new entity, the new ontology (OWL and JSON) is generated.
The SONG manages the generation of OWL and JSON from Excel as well as JSON from OWL. The OWL format is used for a possible optimization of the generated ontology with an ontology editor or for an integration of external ontologies. Any ontology that contains concepts and their labels can be considered as a dSON and can be easily integrated with our approach. JSON is utilized for communication with external tools.
The Searcher uses the SONG Search App, which visualizes the dSON, and can select desired concepts or queries for the execution by a search engine.
The next sections introduce the two additional components, search engine and Corpus Builder that were used in the OntoPMS project in combination with SONG.
### Search engine {#Sec5}
The SONG can be used with any Lucene-based search engine for the generation of queries in the Lucene query syntax out of the Excel template. In addition, new expressive query operators were implemented in the OntoPMS project, which significantly extend the Lucene query syntax.
To identify risks or complications (for PMS) in unstructured documents, complex patterns have to be detected. Such patterns go beyond the standard capabilities of state-of-the-art search engines such as Elasticsearch \[[@CR5]\]. Therefore, we extended these capabilities by creating our own search plugin, providing the required functionalities and improving search quality. The extension was realized as an Elasticsearch plugin and contains, among others, the following additional features: improved tokenization, lemmatization and word decomposition; build-in support for several normal forms / term types; improved quality for ambiguous searches; named entity, date and measurement recognition; additional search modes; NEAR operators.
In particular, the search modes and new search operators are extensively used in our dSONs to produce more precise queries. The search modes correspond to the different types of terms (exact \[E\], diacritics-normalized \[D\], lemmatized baseforms \[B\], compounds-parts \[C\]) that we use in our index. For example, with the query "MODE/E (SafeSet)" we only search for SafeSet with two upper case S. Words within a NEAR/n query must not have a token-distance greater than n. With NEAR/S and NEAR/P, words must occur within one sentence or one paragraph. These new NEAR-Operators can be combined and nested with other queries in an arbitrary way.
### Corpus Builder {#Sec6}
PMS requires information from several types of sources including proprietary manufacturer data and information from the web. Getting information from the web requires some knowledge on what to look for, how to look for it, how to access it and which parts to extract. Additionally, following the links found on a page identified by a given URL quickly turned out to be a potential trap, since the referenced pages may be completely out of scope. Therefore, we concluded that an ontology would help to define the scope and thereby control the automated data acquisition process.
For data acquisition, we developed the Corpus Builder \[[@CR6], [@CR7]\]. Its input component, the Prospector, metaphorically speaking, "roams" the internet in order to identify suitable data to "feed" the OntoPMS Corpus. To achieve this, it uses a set of special corpus queries, which are part of our dSON. Currently, the Prospector delivers its documents to a NLP pipeline, which analyses the contents to identify documents that are important to the respective projects and rejects (i.e. blacklist) documents that shall not be included into the OntoPMS corpus. This processing is done by a kind of control circuit. The "plant" of the feedback loop is controlled by a seed list, produced by the prospector and by the feedback component. It does the crawling and gathering of new URLs by following forward and backward links and reading the contents identified by the URLs. Then, the output is checked by the "sensor". The sensor is controlled by the corpus queries and allows a deep analysis of the content. After that, questionable content is fed back to a splitter component which sorts out garbage (blacklist) and boosts domains with a high amount of documents we want to include (whitelist). URLs, based on those white listed domains are then included if not yet part of the seed list. If the output contains unwanted documents, we have to improve the corpus queries. Hence, we have a semi-supervised learning component, with which the manual part of supervision is made on the abstract level of ontologies. This enables us to change the behaviour of the corpus builder by changing the underlying ontology instead of changing the software.
Results {#Sec7}
=======
Search Ontology (SON) v2.0 {#Sec8}
--------------------------
The Search Ontology (SON) provides a general structure for all dSONs (the classes of the dSONs have to be subclasses of the SON classes; only relations and properties defined in the SON are allowed). This section presents the optimized Search Ontology v2.0, which contains some improvements compared to the v1.0. One of the advantages of the new version is that the SON-based dSONs can be specified in a specially developed Excel template rather than with an ontology editor. Furthermore, the keywords for the search (search terms) are directly associated with search concepts as annotations. In the v1.0, the search terms had to be defined as instances of the search term classes (with different labels) and linked to the search concepts using the property restrictions (based on the property *described_by*). Both aspects simplify the structure of the ontology. Other extensions include negated concepts and direct storage of queries in the ontology. The ontology contains 9 classes and 13 properties.
The SON models three types of entities: search concepts, search terms, and search queries (Fig. [2](#Fig2){ref-type="fig"}). The search concepts are concepts (in the sense of General Formal Ontology, GFO \[[@CR8], [@CR9]\]), whose descriptions or designations have to be found in texts. The other two entities are symbolic structures (gfo: Symbolic_Structure) and serve to model single keywords or phrases of the concept description as well as queries.Fig. 2SON. The SON models three types of entities: search concepts, search terms, and search queries as well as several relations between them
### Search terms {#Sec9}
The search terms are descriptions or designations of search concepts. A distinction is made between simple and composite terms. The simple terms are either single words (e.g., "clip", "Nitinol") or fixed (defined by the user) phrases (e.g., "endoscopic clipping system"). The composite terms are combinations (*has_part*) of simple terms of two search concepts (*has_terms_of_concept_as_part*). They are defined by the user by choosing the two concepts (e.g., Unexpected and Complication) and are generated by the generator as an AND-connection of the OR-linked simple terms of the selected concepts (e.g., "(unexpected OR unforeseeable OR unknown) AND (complication OR failure OR incident)").
### Search concepts {#Sec10}
The search concepts are described or designated by search terms (*described_by*, *simple_term*, *composite_term*). We distinguish between standard (e.g., Complication) and negated (e.g., No_Preclinical) concepts. While the terms of the standard concepts (e.g., "complication", "failure", "incident") have to be contained in the resulting documents, the terms of the negated concepts (e.g., "animal", "study", "preclinical") have to be excluded/negated. Each concept is additionally associated with a single concept query (see Search queries), which is used for the search for descriptions of the concept.
### Search queries {#Sec11}
The single concept query is an OR-connection of all terms (simple and composite) for a standard concept or a negated OR-connection of all terms for a negated concept. The query can additionally contain specified search operators and brackets. The multiple concept query is an AND-connection of single concept queries of selected concepts (*has_query_of_concept_as_part*).
Excel template {#Sec12}
--------------
The design of the Excel template was derived from the SON structure. Sheets, tables and fields were created allowing the specification of all the information required for generating a valid dSON (including search queries). Figure [3](#Fig3){ref-type="fig"} illustrates our example by screenshots of the several sheets.Fig. 3Excel template (excerpt). The different aspects of the PMS question such as "unexpected complication", "type of medical device" and "used material" were modelled within the easily applicable Excel template
The Excel template consists, on the one hand, of the predefined data sheets (Negated_Concept, Composite_Term and Multiple_Concept_Query) and, on the other hand, of the user-defined sheets (facet sheets) for the specification and classification of the search concepts and simple terms. For the observation of medical devices during PMS, we introduced among others the following facets: Medical_Device, Medical_Area, Medical_Problem, Incident, Material and Risk. In our example, the different facets of the search are represented by the Excel sheets Material, Medical_Device and Incident.
In every facet sheet, a subclass structure represents a categorization of the knowledge within the appropriate area. For instance, we subdivided medical devices in clip, stent, occluder and implant; moreover, these device categories can be subdivided into special types, e.g., endoscopic clipping system, PFO occluder, PDA occluder, and so on. Next to the nodes of the subclass hierarchy, the domain expert can enter simple terms; two columns ("Simple Terms (en)" and "Simple Terms (de)") are used for enabling the separation of English and German simple terms. Several types of query operators, such as the wildcard or boost (e.g., incident\^5), can be applied to simple terms in order to refine the search query.
For excluding documents that contain descriptions of certain concepts (e.g., complications in preclinical tests), the Negated_Concept sheet is used. The concept is specified (e.g., No_Preclinical) and described by simple terms to be excluded (e.g., "animal", "study", "preclinical") in the column "Excluded Simple Terms".
The Composite_Term sheet is used for the specification of terms based on other concepts. The composite terms for describing the search concept Unexpected_Complication (column: "Concept") are combined from the simple terms of Unexpected (column: "part1") and Complication (column: "part2").
For the creation of complex (multiple concept) queries, which are based on the conjunction of queries of multiple search concepts (e.g., Unexpected_Complication, Nitinol, Endoscopic_Clipping_System and No_Preclinical), the Multiple_Concept_Query sheet is used. The name of the query is specified in the column "Query". The definition of the improved MODE or NEAR (e.g., NEAR/S) operators (columns: "MODE" and "NEAR") is possible for both, composite terms and multiple concept queries. Concepts that should be combined to a multiple concept query are specified in the columns "Concept".
Generating a dSON by SONG {#Sec13}
-------------------------
By the generation of the ontology from the Excel template, the model presented in Fig. [2](#Fig2){ref-type="fig"} is not applied one-to-one, but rather simplified. For example, simple terms and single concept queries are defined as annotations of the search concept classes.
Firstly, the SONG generates the class hierarchy. The search concept trees from the user-defined sheets (facet sheets) are placed under Search_Concept and the negated concept tree under Negated_Concept. Next, the simple terms (from the columns "Simple Terms (en)" and "Simple Terms (de)") are linked to their concepts using the annotation property *simple_term*.
For each row in the Composite_Term table, a composite term class is generated as subclass of Composite_Term. The annotation properties *has_term_of_concept_as_part_1* and *has_term_of_concept_as_part_2* (shortly: *part_1* and *part_2*) are used to specify the two search concepts whose simple terms have to be put together. In addition, the specified MODE or NEAR operators are generated as annotations. Then, for each composite term class, the corresponding term is generated as an AND-connection of the OR-linked simple terms of the selected concepts, and is associated to the composite term class using the annotation property *query*. The possibly specified query operators for composite terms are taken into account in the correct syntax. The composite term classes are referenced in the search concept classes by the annotation property *composite_term*.
After that, the single concept queries of all search concepts are generated as an OR- connection for standard concepts or a negated OR-connection for negated concepts from all their terms, and are associated with the respective concept using the annotation property *query*. For negated concepts, the standard concepts can also be specified, whose terms have to be excluded (*excluded_concept*).
The multiple concept queries specified on the Multiple_Concept_Query sheet are generated as subclasses of Multiple_Concept_Query. The multiple concept query classes are associated with the concepts whose single concept queries are to be combined using the annotation property *has_query_of_concept_as_part* (shortly: *search_concept*). The single concept queries are AND-linked and stored using the annotation property *query*. Similar to composite terms, the possibly specified operators for queries are taken into account during the generation process.
In Fig. [4](#Fig4){ref-type="fig"}, some parts of the generated dSON-Niti are illustrated. The upper part shows the search concept Unexpected_Complication, which is described by the composite term Unexpected\_\_Complication (with two underscore characters), and the search concept Nitinol. In the lower left corner, the search concept Endoscopic_Clipping_System is presented. The lower middle part demonstrates the negated concept No_Preclinical, which is used for the exclusion of several simple terms. In the lower right corner, the complete (multiple concept) query is illustrated, which consists of different search concepts. In the annotation property *query* is the generated query, which can be executed by a search engine. The query parts (single concept queries) of the multiple concept query are highlighted.Fig. 4Parts of the dSON-Niti in Protégé. The upper part shows the search concept Unexpected_Complication, the composite term Unexpected\_\_Complication and the search concept Nitinol. In the lower part, the search concept Endoscopic_Clipping_System, the negated concept No_Preclinical and the complete (multiple concept) query are illustrated
SONG Search App {#Sec14}
---------------
The SONG Search App accesses the SONG service and receives the generated dSON in JSON format (using the getDSON method of the service). The search concepts and multiple concept queries are visualized on the GUI of the app as a tree with checkable nodes (Fig. [5](#Fig5){ref-type="fig"}). The searcher can select desired concepts or multiple concept queries by checking the corresponding checkboxes. The overall query appears in the boxes "English Query" or "German Query" (depending on which language was used for specifying simple terms). The user can enter a search engine URL directly or can select an available search engine via the dropdown list. In Fig. [5](#Fig5){ref-type="fig"}, the OntoPMS search engine was selected. After pressing the buttons "Run English Query" or "Run German Query", the query is forwarded to the search engine URL as GET parameter. On the website of the OntoPMS search engine (Fig. [6](#Fig6){ref-type="fig"}), the search results are displayed. Matching terms are highlighted. When minor changes are needed (e.g., deleting the NEAR or boost operators), the searcher can modify the query in the input field. However, the important changes of the query that should be available for further searches have to be made in Excel (see "Modifying dSON" in Fig. [7](#Fig7){ref-type="fig"}).Fig. 5SONG Search App. The searcher clicks the checkboxes of desired concepts or multiple concept queries, selects a search engine and submits the generated queryFig. 6OntoPMS search results page. The search results of the OntoPMS search engine are displayed. Matching terms are highlighted. The query can be modified in the input fieldFig. 7Ontology-based search pipeline. The figure summarizes the different steps of the ontology-based search. A recursive optimization of the query is possible
Additionally, the app supports the creation of new multiple concept queries. To achieve that, the user has to select desired concepts, to define the query name and to press the "Create Query" button. The app sends this information to the SONG service (using the method addQuery). After that, the SONG adds the new query to the ontology und transmits the updated dSON back to the GUI. The query is then available for further searches.
Figure [7](#Fig7){ref-type="fig"} summarizes the different steps of the ontology-based search.
Evaluation {#Sec15}
----------
Comprehensive testing confirmed the correct execution of all SONG functions. The applicability of our method and developed tools was evaluated by domain experts.
In the OntoVigilance project, development and editing of specific ontologies was conducted in Protégé \[[@CR10]\]. For commercial use (in our case post-market surveillance purposes for medical devices), ontologies are rather dynamic constructs with a constant pace of re-editing to react upon changing search parameters. Designated key operators are Regulatory Affairs and/or Quality managers positioned at the respective departments of medical device manufacturers. Hence, the prototypical domain expert will be rather untrained in deep informatics/ontology processing. It turned out that Protégé's basic functionality can be learned by a domain expert. However, the ontology editor was considered cumbersome and complex to operate. Consequently, in the OntoPMS project, an Editor easy to learn and at once applicable was formulated as an essential requirement.
Due to the broad application of MS office in business application, an Excel-based platform was highly appreciated by domain experts involved in the project. Three naïve domain experts were shortly instructed in the structure and function of the SON as well as the SON-based Excel template and equipped with a manual written by the SON developers.
Then each domain experts had to fulfill seven pre-defined tasks using the Excel template:Specification of a new facet "Equivalent product" (task 1.1)Specification of a new search concept "Clip XY" in the respective facet (task 1.2)Linking the search concept with pre-defined search terms in form of several simple terms (task 1.3)Specification of a new facet "Safety" (task 2.1)Specification of a new search concept "Unexpected Side Effect" in the respective facet (task 2.2)Linking the search concept with pre-defined search terms in form of composite terms (adjective-noun-phrase) (task 2.3)Specification of a negated concept "No Preclinical" with simple terms to be negated in the query (task 3)
Afterwards the test persons were introduced to the SONG and asked to generate a query to search for unexpected side effects of the Clip XY that are not related to preclinical tests (task 4).
Each test was accomplished successfully in a reasonable time. Table [1](#Tab1){ref-type="table"} provides the results of processing the tasks; Table [2](#Tab2){ref-type="table"} shows the time expenditure. Afterwards the test persons were briefly interrogated by preformed questions and asked to provide a general statement (Table [3](#Tab3){ref-type="table"}). The test persons congruently rated the SON-based Excel template as clearly structured, featured by intuitive operator control logic and highly user-friendly. The Excel platform appears familiar. By application of the SONG, also complex queries can be generated quickly, transparent and reproducible. In comparison to a conventional periodic, manual data search, relevant content was identified by SONG-based data retrieval in a more convenient and comprehensive way.Table 1Successful task managementtask 1.1task 1.2task 1.3task 2.1task 2.2task 2.3task 3task 4Proband 122222222Proband 222222212Proband 3222221220 = task not fulfilled1 = task fulfilled; additional support required (Minor support was necessary in form of short explanations of the template structure while processing the tasks 2.3 and 3)2 = task fulfilled without further supportTable 2Expenditure of time in secondstask 1 completetask 2 completetask 3 completetask 4 completeProband 172601320Proband 247602019Proband 361891622Mean60701620SD131742Table 3Interview resultsYesNoPrefer not to sayDo you have any previous experiences with Ontology Editors?030Is Excel a common standard in regular work environment?300Is the structure of the ontology transparent in this set-up?300Do you rate Excel as feasible for ontology editing?300Is this set-up helpful for query generation?300
In conclusion, the SONG framework significantly facilitates the creation of search queries. According to the domain experts in charge of the testing, the system is intuitive. Queries once generated can be saved and reused. This feature fosters transparency of a systematic search and repeatability, which is a legal imperative in the post-market surveillance of medical devices.
Additionally, BfArM's Research group Safety of Medical Devices evaluated the applicability of SONG from a regulatory perspective giving a direct feedback from an assessor's point of view. The evaluation was conducted based on the FDA coding system for medical device problems, which is implemented at BfArM to classify reported incidents. The evaluated tasks included not only adding, editing and deleting facets, concepts, simple and composite terms within an existing template but also generating complex queries, a task to be considered as highly important from a regulatory perspective. Since none of the assessors had been trained on the concept of ontologies before participating, the initial test results showed some minor but expected difficulties at first. Thereafter a fast learning process was observed, leading to desired performances within a very short period of time. Table [4](#Tab4){ref-type="table"} shows, that no help of any kind was needed (intuitional), whenever the complete information required for a given task was presented in one facet. Tasks including working with more than one facet or adding new information sometimes lead to mistakes, which could be corrected quickly and only by use of debug output (easy). Additional information was only required (explanation needed), when new concepts or composite terms had to be added. These kind of mistakes occurred only if it has been necessary to add new information while also consider consequences of this action to other facets. Notably no task needed to be explained more than once. Generating complex queries becomes a very easy task using this tool. One of the advantages of the current implementation is that it allows quick and easy modification of specified queries (e.g., by adding new simple terms) if the ontology needs to be altered due to a better understanding of the subject. The evaluation showed the fast increase in understanding the concept of search ontologies as well as the applicability of SONG to model the risk classification and to generate powerful search queries in a very systematic and efficient way.Table 4BfArM's evaluation resultsAddEditDeleteFacetEasyIntuitionalIntuitionalSimple termsEasyIntuitionalIntuitionalConceptExplanation neededIntuitionalIntuitionalComposite termsExplanation neededEasyIntuitionalMultiple concept queryIntuitionalIntuitionalIntuitionalIntuitional = no help of any kind neededEasy = quick correction of mistakes only by use of debug outputExplanation needed = additional information required (Explanation was needed only if it has been necessary to add new information while also consider consequences of this action to other facets. Notably no task needed to be explained more than once)
Related work {#Sec16}
============
Ontology-based information retrieval {#Sec17}
------------------------------------
Since finding meaningful and intelligent information is difficult, there are different ontology information retrieval techniques and methods available \[[@CR11]\]. In the wide world of semantic searches, the approach of this paper can be classified as Research Search \[[@CR12]\], because we denoted search queries by concepts. Semantic searches are usually executed not on plain documents but on ontologies, which requires expensive manual annotation or natural language processing steps (NLP) for extracting semantic data out of the documents. After that step, the information of the documents is stored in a semantic knowledge base \[[@CR13]\] or in a semantically enriched enhanced document index \[[@CR14]\], on which semantic searches can be applied by using semantic retrieval languages such as SPARQL \[[@CR15]\] or SeRQL \[[@CR16]\]. The early TAMBIS project \[[@CR17]\] provides a foresighted semantic search approach for accessing multiple bioinformatics databases, using a complex biological concept model for query formulation. Despite of semantic knowledge bases or structured data sources, the approach of this paper builds up on indexed documents which can be retrieved by complex Boolean expressions, which are difficult to construct \[[@CR18]\]. Using ontologies as navigation tree structure in form of a Concept-based Information Retrieval Interface (CIRI) seems to be more effective than a direct interface (input field) \[[@CR19]\].
GoPubMed \[[@CR20]\] uses the Gene Ontology for search on PubMed. In contrast to our approach, the user is not able to increase the precision of the search by simply developing and using his own dSON, exactly tailored to his needs.
Textpresso \[[@CR21]\] is a text-mining system for scientific literature. It implements categories of terms (an ontology) which can be used for a search on a database of articles. Regular expressions have to be created for each category to match the corresponding terms in the text and the documents have to be labelled according to the lexicon of the ontology. The documents are then indexed with respect to labels and words. Our solution does not use any in the ontology contained information for pre-processing or indexing of the documents. The ontology is constantly under development and is adapted by the domain experts to meet their current needs. Our approach does not require any additional pre-processing steps (e.g., labelling) as well as re-indexing the document collection when the ontology changes.
Excel-based ontology development {#Sec18}
--------------------------------
Since ontology engineering is difficult for non-ontologists, there is a need for a rapid and collaborative ontology engineering methodology and easy to use tools \[[@CR22]\]. The transformation of spreadsheets in OWL is already used in life science projects \[[@CR23]--[@CR25]\]; tools and plugins enable the population of OWL ontologies out of spreadsheet templates \[[@CR26]\]. The template we developed differs in that way, that it is not intended for the ontology development in general based on modelling certain OWL constructs. Instead, our template is exactly tailored to the SON-based specification of dSONs in order to make its use by domain exerts as intuitive as possible.
Discussion {#Sec19}
==========
The specification of complex search queries is a recurring task in different domains. The search for complications in the usage of medical devices within the post-market surveillance or the classification of the incident reports are only two examples in this area. Other examples are the patent examination or the search for relevant information in medical documents (e.g., discharge letter). Our solution is domain-independent and thus allows advanced search for relevant documents in different domains using suitable dSONs. The SON approach was also already successfully exploited for another use case in the medical domain to generate complex XPath expressions for querying archetype-based EHRs \[[@CR27]\].
On the other hand, our approach is also generic regarding the used search engine. In the OntoPMS project, an Elasticsearch \[[@CR5]\] based search engine was used, which was extended by developed plugins to provide novel expressive query operators. In addition, the document corpus was specially prepared and optimized for the PMS issue. However, other search engines can also be integrated relatively easily with our approach. The corresponding query syntax only needs to be implemented. For instance, scientists could use our solution to search in PubMed \[[@CR28]\]. Nevertheless, not only scientists can benefit from our work. An office employee could model queries, which are relevant in his daily work and use them for the Google or Bing search.
Additionally, our solution supports the classification of documents (search results). When a query of a search concept delivers certain documents, they can be classified in this concept. The whole taxonomy of search concepts (part of the dSON) can then be considered as a classification of the entire result document set.
Classification of search results {#Sec20}
--------------------------------
From the regulatory perspective of the German competent authority for risk assessment for medical devices, the Federal Institute for Drugs and Medical Devices (BfArM), there is an increasing demand to have the risk assessment process of critical incidents supported by intelligent IT solutions. With respect to the exponential increase in reported incidents with medical devices in Germany, specific dSONs for different aspects of the incident must be developed. These specific dSONs will allow the identification of similar incidents and thereby an automatic recommendation of classification of new incidents will become possible. The aspects currently in focus include dSONs for the resulting health problem, device problem, root cause and components, all of which are at present being developed on the basis of the FDA coding system, which is currently revised by the IMDRF working group on Adverse Event Terminology and Coding \[[@CR29]\]. Using the SONG approach allows domain experts to easily create and modify the subsequent search concepts for each of the FDA/IMDRF Codes. Our approach will allow for accelerated risk classification, which in turn allows to create individual views of device-specific problems as well as to monitor the performance of different manufacturers within certain device groups such as hip implants, cardiac pacemakers, instruments for bone surgery or insulin pumps.
Future work {#Sec21}
===========
Future work includes the application of our approach to other domains, especially the search for relevant information in medical documents as well as the integration of further search engines including the implementation of the corresponding query syntax. Since the PubMed search engine can process GET parameters, we can forward the generated query strings directly to PubMed. In future work, we planned to integrate fields such as "author", "journal" and "title" as well as the MeSH terminology \[[@CR30]\], including the subdivision in main and sub headings.
After the definition of a sufficiently extensive knowledge base in form of dSONs, ontology learning can be exploited for supporting a semi-automatic query creation.
Conclusion {#Sec22}
==========
We presented the improved Search Ontology, a promising domain-independent approach to specify complex search queries. Our solution allows advanced search for relevant documents in different domains using suitable dSONs and supports an automatic classification of search results. The second version of the Search Ontology includes enhancements such as the inclusion of new search operators, negated concepts and direct storage of generated queries in the ontology. For easier handling by non-ontologists, we developed an Excel template, which facilitates a SON-based specification of dSONs without the usage of ontology editors or knowing the query syntax. A service-oriented architecture was introduced; in the core of the architecture stands the Search Ontology Generator (SONG), which provides methods for an access by search engines as well as dSON administration methods. By the enhancement of the SON, the Excel template, the SONG Search App and the service-oriented architecture, we improved the access to the Search Ontology for domain experts and external tools.
dSON
: Domain-specific Search Ontology
PMS
: Post-market surveillance
SON
: Search Ontology
SONG
: Search Ontology Generator
**Publisher's Note**
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
An earlier version of the paper has been presented at JOWO 2017 (Joint Ontology Workshops) / ODLS (Ontologies and Data in Life Sciences) in Bozen-Bolzano, Italy.
This work has been funded by the German Federal Ministry of Education and Research (BMBF) in the "KMU-innovativ" funding program as part of the "OntoPMS" project (reference number 01IS15056B).
We acknowledge support from the German Research Foundation (DFG) and Leipzig University within the program of Open Access Publishing.
AU developed the search ontology method for modelling and generation of search queries, designed and implemented SON and SONG. SK made substantial contributions to the design and implementation of SON and SONG, analyzed the recent related works. HH was responsible for project management, conception and semantic foundation of developed ontologies. TW, SM, KG, SR, MOS, acting as domain experts, set up the content design of the SON, created the evaluation plan and conducted the evaluation. The main contribution of WB was the designing and implementing the Corpus Builder with the Prospector component including the integration of the ontology to control developed components. CG and PB created an Elasticsearch plugin providing advanced functionalities such as additional search modes and NEAR operators. RS, CB, MK, KP and WL implemented a risk classification on the basis of the FDA coding system and performed the evaluation of the system from the regulatory perspective of the German Federal Institute for Drugs and Medical Devices (BfArM). JUM and MW contributed to the design of the overall system. All authors read and approved the final manuscript.
This work has been funded by the German Federal Ministry of Education and Research (BMBF) in the "KMU-innovativ" funding program as part of the "OntoPMS" project (reference number 01IS15056B). The aim of the funding initiative is the strengthening of innovation capacity of small and medium sized enterprises in Germany. Nevertheless, the funding body played no role in the design of the study and collection, analysis and interpretation of data and in writing the manuscript.
The datasets generated and/or analyzed during the current study as well as the tools developed and further material used in the OntoPMS project are not publicly available. They contain information that could compromise research participants consent and contrast with the projects funding objective with the aim of strengthening the innovation capacity of small and medium sized enterprises in Germany. However, are available from the corresponding author on reasonable request.
Not applicable.
Not applicable.
The authors declare that they have no competing interests.
| {
"pile_set_name": "PubMed Central"
} |
{
"pile_set_name": "PubMed Central"
} |
|
Introduction {#S1}
============
Fruits and vegetables are known to have high nutrient content, making them the basis of a healthy diet. Many of these foods can be eaten raw, and although this represents a practical advantage, it also makes them notoriously relevant to foodborne illnesses. *Salmonella enterica* is one of the most common human pathogens found in fresh produce ([@B3]; [@B24]). Previously, plants were thought to be passive vectors for human pathogens, but recent studies showed that *S. enterica* can induce plant defense responses ([@B26]; [@B10]; [@B25]; [@B27]). Intriguingly, although the mechanism is not fully understood, this bacterium can overcome plant defense ([@B31]; [@B32]) and survive for weeks inside diverse plants species, including lettuce (*Lactuca sativa* L.) ([@B13]; [@B21], [@B20]; [@B16]; [@B30]). These findings have prompted further research on the interaction between plants and human pathogens.
Artificial inoculation of plants is a common technique used to study plant interaction with phytopathogens ([@B17]; [@B15]). Nevertheless, this approach has some technical limitations when studying plant interaction with enterobacteria, in particular *S. enterica* and enterohemorrhagic *Escherichia coli*, due to the relative low number of these bacteria inside the plant. In fact, recent studies have shown that *S. enterica* population can decrease with time in many plant species in an inoculum concentration-dependent manner ([@B8]; [@B16]). Beyond that, the plant species and the inoculation procedure itself can affect bacterial population dynamics inside plants. For instance, tomato (*Solanum lycopersicum*) seedlings dip-inoculated with *S. enterica* at a concentration of 8 Log CFU ml^--1^ maintains the population size 1 day after inoculation (DAI) followed by a decrease after 14 DAI ([@B2]). Similarly, when adult lettuce leaves were dip-inoculated with 8 Log CFU ml^--1^ of *S. enterica*, the Log CFU cm^--2^ of leaf showed no alteration in bacterial population until three DAI, but a reduction in the population size after 7 DAI ([@B30]). Nonetheless, when lower inoculum concentration of 4.7 Log CFU ml^--1^ of *S. enterica* was used to infiltrate small areas of fully expanded *Nicotiana benthamiana* leaves, a 100-fold increase in bacterial population was observed at three DAI ([@B26]). These findings indicate that the inoculation method and/or the initial concentration of the inoculum can influence the bacterial population dynamic in leaves.
In the field, plants can be exposed to variable amounts of pathogen load depending on the source of the inoculum. In a survey to quantify *Salmonella* in irrigation water, [@B1] found an average of 0.03 MPN (most probable number) of cells per 100 ml of water. Additionally, animals are reservoirs of bacterial pathogens of humans and might shed high level of inoculum in their feces. For instance, cattle feces can shed *E. coli* O157 at concentrations \>4 Log CFU g^--1^ ([@B28]), whereas some animals such as mice are considered super-shedders of *S.* Typhimurium ([@B11]).
Once crops are exposed to these environmental inocula, bacterial cells can internalize into edible leaves through natural openings and wounds ([@B4]; [@B21]; [@B31]). Understanding human pathogen survival inside the leaf apoplast is very important as this niche protects the bacterium from common sanitation procedures of leafy vegetables ([@B29]), posing a risk to reach the human host. Thus, we performed vacuum infiltration procedures using a range of low to high concentrations of bacterial inoculum (3--7 Log CFU ml^--1^) to assess the effect of inoculation dose on bacterial survival and the detection limit of our procedure using contrasting lettuce cultivars over a period of 20 days. The findings of this study will assist with designing of plant phenotypic screening useful for breeding programs.
Materials and Methods {#S2}
=====================
Plant Material and Growth Conditions {#S2.SS1}
------------------------------------
Approximately 15 lettuce seeds of each cultivar (Red Tide, Lollo Rossa, and Salinas) were germinated in water-soaked filter paper for 2 days at room temperature. Each germinated seed was transplanted to a peat pot pre-soaked with distilled water for 10--20 min. Ideally, sprouted seeds with approximately the same root size should be selected for transplanting. Pots were placed in trays covered with plastic dome, leaving a small space (around 4 cm) to avoid water condensation, and kept at 18 ± 2°C, 240 ± 10 μmol m^--2^ s^--1^ with a 12-h photoperiod, and 80 ± 10% of air relative humidity. One week after transplanting, seedlings were fertilized with 0.05 g of fertilizer per plant mixed with 30 ml of distilled water. Three- to 4-week-old plants were used for inoculation ([Figure 1](#F1){ref-type="fig"}).
{#F1}
Bacterial Inoculum Preparation {#S2.SS2}
------------------------------
*S. enterica* subsp. *enterica* serovar Typhimurium strain 14028s was streaked from frozen glycerol culture stock on low-salt Luria Bertani (LSLB) agar plate, supplemented with 60 μg ml^--1^ kanamycin, and incubated overnight at 28°C. Late in the afternoon of the day before the inoculation assay (around 5 pm), one single colony was placed in 100 ml of LSLB medium with 60 μg ml^--1^ kanamycin in a 125-ml Erlenmeyer flask. As a blank control, 5 ml of the LSLB plus antibiotic solution was placed into a clean culture tube. Bacterial and blank solutions were incubated in a rotary shaker at 28°C, 150 rpm, overnight ([Figure 1](#F1){ref-type="fig"}).
In the morning of the next day, bacterial and blank solutions were removed from the incubator and the optical density at 600-nm wavelength (OD~600~) was measured using a spectrophotometer. It is important to shake the culture flask before transferring 1 ml to a sterile cuvette to avoid errors during OD readings due to bacterial settling on the bottom of the flask. The OD~600~ should be between 0.8 and 1.0 to ensure that the bacterial growth is still in the log phase. A two-step bacterial dilution was used to prepare the final inoculum at the desired concentration. Step 1: the volume of the bacterial solution needed to obtain a bacterial OD~600~ of 0.2 was calculated using the formula C1 × V1 = C2 × V2, where C = concentration and V = volume. After transferring the desired bacterial solution volume (V2) to a 50-ml centrifuge tube, bacterial cells were harvested by centrifugation at 1,200 × *g* for 20 min at 22 ± 2°C. The supernatant was discarded, and the pellet was resuspended in sterile distilled water by vortexing. The centrifugation step was essential to remove the excess of LBLS media plus kanamycin, to avoid bacterial growth inhibition within the leaf due to the presence of the antibiotic, as well as to reduce the volume of *Salmonella* solution handled in the lab. Step 2: 0.0001, 0.01, 1, or 100 ml of the final solution from step 1 (OD~600~ = 0.2) was added to a new flask containing 1,000 ml of sterile distilled water to obtain the final inoculum concentration of OD~600~ 0.00000002 (1 Log CFU ml^--1^), 0.000002 (3 Log CFU ml^--1^), 0.0002 (5 Log CFU ml^--1^), or 0.02 (7 Log CFU ml^--1^). Finally, 0.1 ml of Silwet was added to the inoculum to obtain a final concentration of 0.01%. Inoculum was stirred with a magnetic bar.
Vacuum Inoculation of Lettuce Leaves {#S2.SS3}
------------------------------------
Three- to four-week-old lettuce plants (four plants per cultivar) were vacuum-infiltrated with the final bacterial solution of 1.8, 3.5, 5.4, or 7.7 Log CFU ml^--1^. These concentration values were estimated by serial dilution plating of the inoculum. Each potted plant was placed upside-down into a 400-ml beaker containing enough inoculum to immerse the plant shoot completely. Aluminum foil was placed at the base of the plant to avoid the contact of soil with the inoculum. Submerged plants were placed in a vacuum chamber and vacuum was applied for 1 min. To enable a uniform filling of the leaf apoplast with inoculum, the vacuum was released quickly by disconnecting the suction tube to the vacuum chamber, allowing the chamber to depressurize. The leaves should become dark green due to inoculum infiltration ([Figure 1](#F1){ref-type="fig"}). Fresh inoculum was added to the beaker to ensure total immersion of the inoculated leaves and after three plants were inoculated. Inoculated plants were placed back in the trays and partially covered with the plastic dome for the duration of the experiment.
Enumeration of Apoplastic Bacterium Population {#S2.SS4}
----------------------------------------------
Bacterial population size was estimated in the second true leaf of the inoculated plants at 2 h post inoculation (HPI), and 1, 10, and 20 DAI. After 2 HPI when the water soaking disappeared from the leaves, the second true leaf was excised at its base, the fresh weight (FW) was measured using an analytical balance, and the leaf was rinsed by immersion in sterile distilled water for 1 min. Sampling at this time point was crucial to determine the total number of bacteria inoculate on and in the plant. For the following time points, after FW measurement, the leaf was surface sterilized with 70% ethanol for 1 min and rinsed in water for 1 min.
For all time points, the excess water was removed from the leaves by gently blotting them on paper towel. Leaf was placed in a homogenizer blender bag containing 20 ml of phosphate-buffered saline (PBS) solution and macerated with a homogenizer until a green solution with very small leaf debris was obtained. The green solution, but not the leaf debris, was transferred to a 50-ml centrifuge tube and centrifuged at 2,200 × *g* for 15 min at 22 ± 2°C. The supernatant was removed, and fresh 20-ml PBS solution was added to the pellet containing bacterial cells, followed by centrifugation at 2,200 × *g* for 15 min ([Figure 1](#F1){ref-type="fig"}). To confirm that no bacterium was present in the supernatant, this solution was plated on solid LSLB medium.
After the wash-centrifugation steps, 0.2 ml of sterile Milli-Q water was added to recover the bacterial pellet and transferred to a clean 1.7-ml tube. Please note that water was used as further DNA extraction was desired, but PBS buffer could be used if only serial dilution and plating would be performed. Immediately after bacterial recovery, 10 μl of the bacterial solution was added to 90 μl sterile water in another 1.7-ml microfuge tube, making a 1:10 dilution. This solution was diluted to 10^--2^ for the low inoculum dose and down to 10^--8^ for the highest inoculation dose. Including the most concentrated leaf sample, 10 μl of all dilutions was plated on LBLS agar with 60 μg ml^--1^ kanamycin ([Figure 1](#F1){ref-type="fig"}; [@B15]). Dilution plates were air-dried and subsequently incubated at 28°C overnight. The next day, bacterial colonies were counted at the dilution column that allowed for the visualization of individual colonies using a stereoscope.
Data Analysis {#S2.SS5}
-------------
The number of single colony-forming units (CFU) was used to estimate the bacterial population per gram of fresh leaf tissue by multiplying the CFU counts by the dilution factor times 10, to account for the 10 μl out of the 100 μl used for plating. Data points represent the average of three biological replicates (three different plants) and two technical replicates during plating (*n* = 6). Average and the standard error (SE) were calculated using Microsoft Excel. Statistical significance among the different cultivars and time points was estimated by the analysis of variance (ANOVA) followed by Scott--Knott test with a significance threshold of α = 0.05, using the square root of the means. This data transformation method is recommended when the variance is proportional to the mean ([@B23]). The graph was plotted with the Log CFU per gram of leaf FW over time using untransformed data.
Total DNA Extraction and qPCR Analysis {#S2.SS6}
--------------------------------------
After the wash-centrifugation steps described above, 5 μl of the DNA extraction buffer was added to 50 μl of the recovered bacterial solution. After vortexing for 30 s, the solution was centrifuged for 1 min at 13,000 rpm at room temperature, and the supernatant was transferred to a clean tube. DNA in the supernatant was precipitated by adding 0.1 volumes of 5 M ammonium acetate and one volume of isopropanol, followed by vortexing and 1-h incubation at room temperature. Next, two washes were performed to remove excess salt by adding 1 ml of cold 70% ethanol, vortexing for 30 s, and spinning-down for 1 min at 13,000 rpm, room temperature. After each centrifugation, the ethanol solution was discarded. Finally, the DNA pellet was dried out on the bench for 15 min and resuspended in 30 μl of DNase-free water. Quantitative PCR (qPCR) was performed with 3 ng of DNA template, 200 nM of reverse and forward gene-specific primers, and 10 μl of iTaq Fast SYBR Green Supermix in a total reaction volume of 20 μl. Reactions were carried out in an Applied Biosystems 7300 thermocycler, using the following cycling parameter: 1 cycle of 95°C for 5 min, and 40 cycles of 95°C for 10 s and 60°C for 30 s. The dissociation curve was determined for every reaction to confirm the presence of a single amplicon and the lack of primer dimers and non-specific products.
The primer set efficiency was assessed using the standard curve method. The linear regression equation was plotted using the cycle threshold (CT) value and the Log of the DNA concentrations of 10-fold serial dilutions, using the Microsoft Excel software. The slope values were used to calculate the efficiency for each pair of primers tested ([@B18]) and number of DNA copies ([@B5]). *S. enterica*-specific primers (forward---TCGTCATTCCATTACCTACC and reverse---AAACGTTGAAAAACTGAGGA; [@B12]) and the ribosomal 16S primers (forward---CCAGCAGCCGCGGTAAT and reverse---TTTACGCCCAGTAATTCCGATT; [@B6]) were selected for this assay. The number of DNA copies per gram of leaf tissue was calculated using the formula: number of DNA copies = (ng × 6.002 × 10^23^)/(length × 1 × 10^9^ × 650), in which ng is the Log (CT -- standard curve intercept/slope standard curve), 6.002 × 10^23^ is Avogadro's number, length is the size of the *S. enterica* strain 14028s genome (4,964,097 bases), 1 × 10^9^ is used to account for the ng unit conversion, and 650 is the molar mass in grams per mole of one single DNA base pair ([@B5]). The number of DNA copies is equal to the number of cells per reaction for 1 μl of DNA sample, when using *S. enterica*-specific primers.
List of Materials {#S2.SS7}
-----------------
1. Peat Pellets 42 mm (peat moss pellets) (Jiffy 7, catalog number: SO-JFPP).
2. Plastic trays without holes (Hummert International, catalog number: 65-6963-2).
3. Fertilizer (Peters Excel^®^ pHLow^®^ 19-11-21 Multi-Purpose, catalog number: G99001).
4. Plastic domes (Hummert International, catalog number: 65-6964-1).
5. Soil mix (Sun Gro^®^ Sunshine^®^ \#1 Grower Mix with RESILIENCE^TM^).
6. 50-ml centrifuge tubes (Fisher Scientific, catalog number: 553860).
7. 1.7-ml microcentrifuge tubes (VWR, catalog number: 87003-294).
8. Culture Tubes, Plastic, with Dual-Position Caps (VWR, catalog number: 60818-703).
9. 125-ml Erlenmeyer flasks (Pyrex^®^, catalog number: 4980-125).
10. 250-ml and 1,000-ml beakers (VWR, catalog numbers: 10754-952 and 10754-960).
11. Filter Whirl-Pak(R) homogenizer blender filter bag 207 ml (Millipore Sigma, catalog number: WPB01385WA-250EA).
12. Sterile inoculating loops (VWR, catalog number: 82051-146).
13. Magnetic stir bars (VWR, catalog number: 58948-988).
14. Square petri dish with grid (VWR, catalog number: 60872-310).
15. Round petri dishes, medium (100 × 15 mm) (VWR, catalog number: 25384-302).
16. Disposable plastic cuvettes (VWR, catalog number: 97000-586).
17. Micropipettes (Rainin Pipet-Life^TM^).
18. Tweezers (VWR, catalog number: 89259-984).
19. Silwet L-77 (Lehle Seeds, catalog number: VIS-30).
20. Agarose (VWR, catalog number: 97062-250).
21. Tryptone (IBI Scientific, catalog number: 41116105).
22. Yeast extract (US Biotech Sources, catalog number: Y01PD-500).
23. Sodium chloride (Fisher Scientific, catalog number: S271-500).
24. Bacteriological agar (IBI Scientific, catalog number: IB49171).
25. LSLB medium (broth and agar; see Recipes).
26. TRIS---tris(hydroxymethyl)aminomethane (VWR, catalog number: 33621.260).
27. EDTA---ethylenediaminetetraacetic acid (VWR, catalog number: 20294.294).
28. SDS---sodium dodecyl sulfate (VWR, catalog number: 1.13760.0100).
29. Ammonium acetate (VWR, catalog number: 0103-500G).
30. Kanamycin (GoldBio, catalog number: K-120).
31. iTaq Fast SYBR Green Supermix (BioRad, Hercules, CA, United States).
32. Sterile distilled water.
33. Sterile Milli-Q water.
34. Ethanol pure grade (Sigma-Aldrich, catalog number: 459836).
35. PBS buffer (see Recipes).
36. Lettuce cultivars (Red Tide, Lollo Rossa and Salinas, stored at 4°C).
37. *S. enterica* stock cultures (stored in 20% glycerol at −80°C).
Required Equipment {#S2.SS8}
------------------
1. Plant growth chamber (Caron Products & Services, model: 6341-2).
2. Shaker incubator (VWR, catalog number: 12620-946).
3. Spectrophotometer (Thermo Fisher Scientific, model: Spectronic 20D + or equivalent).
4. Centrifuge (Eppendorf, model: 5810).
5. Homogenizer Hand Model (Bioreba, catalog number: 400010).
6. Digital hygrometer (VWR, catalog number: 35519-047).
7. Quantum meter (Apogee, catalog number: BQM).
8. Vortex (BioExpress, GeneMate, catalog number: S-3200-1).
9. Analytical Balance (VWR, catalog number: 10753-570).
10. Magnetic stirrer (VWR, catalog number: 97042-642).
11. Stereoscope (VWR, catalog number: 89404-502).
12. Applied Biosystems 7300 thermocycler (Applied Biosystems, Foster City, CA, United States).
13. 20-, 200-, and 1,000-μl micropipettes and tips.
14. Milli-Q filter (Millipore Sigma, catalog number: C85358).
15. Autoclave.
16. Biological safety cabinet level 2 (Labconco^TM^ Purifier^TM^ Axiom^TM^ Class II, Type C1, Kansas City, MO, United States).
Solution Recipes {#S2.SS9}
----------------
### Low-Sodium Luria Bertani Medium {#S2.SS9.SSS1}
1. 10 g L^--1^ Tryptone
2. 5 g L^--1^ Yeast extract
3. 5 g L^--1^ NaCl
4. 15 g L^--1^ Agar (only for solid medium)
5. Autoclave medium at 15 psi, 120°C for 15 min.
6. Allow medium to cool down to about 55°C and add appropriate antibiotic if needed.
### Phosphate-Buffered Saline Solution {#S2.SS9.SSS2}
1. 8 g L^--1^ NaCl
2. 0.2 g L^--1^ KCl
3. 1.44 g L^--1^ Na~2~HPO~4~
4. 0.24 g L^--1^ KH~2~PO~4~
### DNA Extraction Buffer {#S2.SS9.SSS3}
1. 200 mM Tris (pH 7.5)
2. 250 mM NaCl
3. 25 mM EDTA
4. 0.5% SDS
Results {#S3}
=======
Previously, it was reported that *S. enterica* persistence is dependent on the bacterial inoculum concentration ([@B8]; [@B16]) and on the lettuce cultivar ([@B14]). Therefore, we tested whether our protocol was useful to reliably enumerate bacterial cells in lettuce leaves using four different concentrations of bacterial inoculum (1, 3, 5, or 7 Log CFU ml^--1^) and three commercial cultivars of lettuce with contrasting bacterial growth patterns (Red Tide, Lollo Rossa, and Salinas). The lowest inoculum concentration (1 Log CFU ml^--1^) is impractical to use as no live bacteria could be recovered at 2 HPI, i.e., no colonies grew on the medium after plating.
At the inoculum concentration of 3 Log CFU ml^--1^, *S. enterica* grew in Red Tide leaves with a 2.3-fold increase in CFU observed between 2 HPI and 1 DAI, while in Lollo Rossa and Salinas, the bacterial titer showed a 1.8-fold or no increase in the same time period (*p* \< 0.05), respectively ([Figure 2A](#F2){ref-type="fig"}). From 1 to 20 DAI, the bacterial population decreased for all plant cultivars. However, the extent of bacterial population decrease was smaller in Red Tide (1.6-fold), whereas it decreased 10-fold in Lollo Rossa and 8.9-fold in Salinas (*p* \< 0.05; [Figure 2A](#F2){ref-type="fig"}).
{#F2}
When inoculated with 5 Log CFU ml^--1^, a similar trend of higher bacterial population in Red Tide leaves as compared to Lollo Rosa and Salinas was observed throughout the experiment (*p* \< 0.05). However, the bacterium CFU per gram of leaf remained constant in Red Tide until 20 DAI ([Figure 2B](#F2){ref-type="fig"}). Bacterial population inside Lollo Rossa and Salinas leaves remained constant between 2 HPI and 1 DAI; however, the bacterial titers decreased 2.4- and 3.6-fold between 1 and 10 DAI in Lollo Rossa and Salinas, respectively (*p* \< 0.05). By 20 DAI, *S. enterica* decreased further in Lollo Rossa and Salinas by 5.0- and 8.9-fold, respectively (*p* \< 0.05; [Figure 2B](#F2){ref-type="fig"}).
Remarkably, at 2 HPI with 7 Log CFU ml^--1^ of *S. enterica*, a significantly larger bacterial population in Red Tide leaves was observed in comparison to that in Lollo Rossa and Salinas ([Figure 2C](#F2){ref-type="fig"}). The bacterial population further increased by 4.1-fold at 1 DAI in Red Tide leaves, but subsequently decreased by 56.8- and 40.9-fold at 10 and 20 DAI, respectively (*p* \< 0.05; [Figure 2C](#F2){ref-type="fig"}). Lollo Rossa also supported a higher number of *S. enterica* cells at 1 DAI, a 2.3-fold increase from 2 HPI, followed by a decrease of 26.5- and 64.9-fold between 1 and 10 DAI or 1 and 20 DAI, respectively (*p* \< 0.05; [Figure 2C](#F2){ref-type="fig"}). Bacterial growth inside Salinas leaves had a 2.9-fold decrease in population size between 2 HPI and 1 DAI, also decreasing the number of bacterial populations in its leaves by 2.8- and 6.3-fold at 10 and 20 DAI, respectively (*p* \< 0.05; [Figure 2C](#F2){ref-type="fig"}). These findings suggest that high levels of inoculum concentration lead to a higher *S. enterica* death rate inside the leaf apoplast, independently of the plant cultivar, considering that the second true leaf was fully expanded before inoculation and the role leaf was sampled. Although Red Tide supported larger bacterial populations at all times ([Figure 2C](#F2){ref-type="fig"}).
Overall, our results showed that, independently of the inoculation dose, Red Tide supported higher bacterial population than Lollo Rossa and Salinas, in which the inoculum concentration of 3 Log CFU ml^--1^ enhanced these differences, mainly at later time points ([Figure 2](#F2){ref-type="fig"}). No macroscopic symptoms such as chlorosis or necrosis were observed on mock-inoculated leaves or leaves inoculated with 3 Log CFU ml^--1^ of *S. enterica*, for all three cultivars tested. Red tide showed some chlorosis when inoculated with 5 and 7 Log CFU ml^--1^, while no chlorosis was observed for Lollo Rossa or Salinas when these same concentrations of bacterial inoculum were used ([Figure 2D](#F2){ref-type="fig"}).
To support the results of the serial dilution plating method in estimating the bacterial population size, we used qPCR as a second approach ([@B9]). This is a simple assay widely used in microbial community analysis, as well as it is quick and less labor-intensive than other methods ([@B5]; [@B7]). To this end, we chose to test the sample from Red Tide leaves at 1 day post vacuum inoculation with 3 Log CFU ml^--1^ bacteria. Due to the small number of recovered bacterial cells (∼390 ± 41.6 cells per gram of leaf) ([Figure 2A](#F2){ref-type="fig"}), technical error during plating could have occurred. Using the *S. enterica*-specific primers, we estimated that 612 ± 54.7 bacterial cells were present per gram of leaf, while no amplicon was detected in mock-inoculated leaves, the negative control ([Figure 3A](#F3){ref-type="fig"}). To rule out the possibility that the lack of amplicon was due to the lack of DNA in the PCR reaction, the 16S primer set was used with the same DNA samples from bacterium- and mock-inoculated leaves. This primer set also aligns with the lettuce mitochondrion genome (NCBI reference NC_042756.1, e-value between 1 × 10^--3^ and 6 × 10^--4^); thus, amplification of both plant and bacterium DNA was expected. Similar amount of DNA was recovered from both *S. enterica*- and mock-inoculated plants, indicating a consistent DNA extraction protocol, in which DNA from bacteria was precipitated together with the plant DNA ([Figure 3B](#F3){ref-type="fig"}).
![Number of DNA copies estimated by qPCR with **(A)** *Salmonella enterica*-specific primers ([@B12]) or **(B)** 16S primers ([@B6]) in Red Tide leaves. Bacterial population (*S. enterica* strain 14028s) was determined at 1 day post vacuum inoculation with 3.5 Log CFU ml^--1^ of bacterial cells or water + 0.01% Silwet as a mock control. Results are shown as the average of three biological replicates (*n* = 3 ± SE). Value of the mock treatment in graph **(A)** is zero and the error bars are very small to appear in graph **(B)**. CFU = colony forming unit, g, gram; ns, non-significant statistically.](fmicb-11-01516-g003){#F3}
Discussion {#S4}
==========
Although *S. enterica* induces plant defense responses ([@B26]; [@B10]; [@B25]; [@B27]), it can still persist for long periods in the leaf apoplast depending on the bacterial strain and the plant genotype ([@B33]; [@B14]). Furthermore, recent studies have shown that variations in the *S. enterica* culturing conditions, such as temperature and nutrients in the medium ([@B19]), and environmental conditions for the plant cultivation, such as temperature and humidity ([@B8]; [@B16]; [@B30]), can interfere mainly with the ability of *S. enterica* to internalize plant tissues. Nevertheless, variations in environmental conditions not only affect the bacterial internalization, which indeed is an important step during bacterial colonization of plants, but also interfere with the outcome of the plant--pathogen interactions and the persistence phenotype. These findings highlight the importance of establishing inoculation procedures and bacterial enumeration methods with controlled conditions that enable a realistic understanding of *S. enterica* survival in the plant, isolating the plant phenotype from environmental effects.
The method described here is simple and robust to assess *S. enterica* persistence in plant leaf apoplast and, in addition, to allow for comparisons among different inoculation doses and lettuce cultivars. We have determined that 3 Log CFU ml^--1^ is the minimum concentration of bacterial inoculum in which bacterial cells can be reliably recovered from leaves right after inoculation. This inoculum level also enabled us to follow the drastic decrease of bacterial population size in the cultivars Lollo Rossa and Salinas, when 23 ± 3.2 and 19 ± 3.2 cells per gram of leaf was detected at 20 DAI, respectively ([Figure 2A](#F2){ref-type="fig"}). We used two methods to verify the results, in which the number of bacterial cells per gram of leaf tissue estimated by plating or qPCR was comparable. The plating method estimated the number of live bacterial cells only, whereas qPCR amplifies DNA from all cells present in the tissue, which might explain the lower cell number estimate by plating (390 ± 41.6 cells) as compared to qPCR (612 ± 54.7 cells). Plating also has the advantage of being cheaper and less labor-intensive than qPCR. However, if automation is an option, qPCR might be a better choice for larger screening procedure.
In addition to bacterial plating and qPCR techniques, other methods are known to be used to enumerate bacterial cells. Among these, treatment with propidium monoazide (PMA) alone or PMA + deoxycholate (DC) can be used before DNA extraction to detect bacterial cell death in the qPCR analysis. However, it may kill cells injured from experimental treatments that otherwise could have recovered ([@B22]). Moreover, microscopy techniques, such as fluorescent *in situ* hybridization (FISH), and cell sorting techniques, such as flow cytometry (FC) and the specialized method of fluorescence-activated cell sorting (FACS), can also be used to access bacterial population ([@B7]), but these are expensive and labor-intensive. Hence, frequently conventical plating is qualified as the most robust and reliable method for cell quantification ([@B5]; [@B22]).
Understanding the various aspects of human bacterial pathogen interactions with plants is important to establish successful strategies to prevent, or at least reduce, contamination of fresh produce. We anticipate that this method will enable one to address questions related to the survival of human pathogens in leaves, such as the plant immune responses triggered by them, how human pathogens can affect the plant environment and its microbial community, and the mechanisms involved in the process. It is important to note that we chose vacuum infiltration in order to address bacterial survival in the leaf apoplast, but if internalization processes are the goal of future studies, we would indicate dip or spray inoculation followed by leaf surface sterilizing and print of the leaves in a petri dish with culture media, to confirm that the sterilization procedure was efficient. Ultimately, this procedure can be used to phenotype mapping populations to further identify genomic regions in the plant associated with defense against *S. enterica*, in addition to be useful for bacterial competition assays *in planta* to determine the relative fitness of various strains in this niche.
Data Availability Statement {#S5}
===========================
All datasets generated for this study are included in the article/supplementary material.
Author Contributions {#S6}
====================
PO and MM conceived the research, analyzed the data, and wrote the manuscript. PO performed the experiments. Both authors contributed to the article and approved the submitted version.
Conflict of Interest {#conf1}
====================
The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
**Funding.** This research was supported by grants from the U.S. Department of Agriculture--National Institute of Food and Agriculture (NIFA; 2015-67017-23360 and 2017-67017-26180) and NIFA Hatch grant (CA-D-PLS-2327-H) to MM.
[^1]: Edited by: Dario De Medici, Istituto Superiore di Sanità (ISS), Italy
[^2]: Reviewed by: Barbara Bertasi, Experimental Zooprophylactic Institute of Lombardy and Emilia Romagna (IZSLER), Italy; Ana Allende, Consejo Superior de Investigaciones Científicas (CSIC), Spain
[^3]: ^†^Present address: Paula Rodrigues Oblessuc, Department of Protection of Specific Crops, InnovPlantProtect CoLab, Estrada de Gil Vaz, Elvas, Portugal
[^4]: This article was submitted to Food Microbiology, a section of the journal Frontiers in Microbiology
| {
"pile_set_name": "PubMed Central"
} |
Erratum {#Sec1}
=======
During production of the original article \[[@CR1]\], a document supplied by the author containing a set of further corrections to the manuscript was overlooked. The article therefore published containing a number of errors which are detailed here.
The publisher would like to apologise for these errors, which have all been corrected in the original article.
In the 'Results' section of the abstract, the number of responses received to the questionnaire was incorrectly given as 1698. This should have been 1693, as was correctly reported throughout the article itself.
The first names of all authors except Martin Ho Yin Wong were mistakenly abbreviated to a first initial. The original article has been updated with the correct, full first names, as they appear in this erratum.
The affiliation of Hannah Fosker was incorrect, and has now been updated to: Leicestershire Partnership NHS Trust, Bradgate Mental Health Unit, Glenfield Hospital, Groby Road, Leicester, LE3 9EJ, UK. In the original article, this has been inserted as affiliation 3, and subsequent affiliations have been renumbered to reflect this.
The affiliation of Alec Knight has also been amended to the correct address: King's Improvement Science, Health Service and Population Research Department, Institute of Psychiatry, Psychology & Neuroscience, King's College London, IoPPN Main Building, London SE5 8AF, UK.
The affiliation of Monica Lakhanpaul has been changed from affiliation 7 (Department of Primary Care and Public Health Sciences, King's College London) to affiliation 1 (Population, Policy and Practice, UCL Institute of Child Health).
In the Acknowledgements section, it was stated that: "AK was supported by the National Institute for Health Research (NIHR) Collaboration for Leadership in Applied Health Research and Care South London at King's College Hospital NHS Foundation Trust."
This sentence should have included Peter Littlejohns, to read: "PL and AK were supported by the National Institute for Health Research (NIHR) Collaboration for Leadership in Applied Health Research and Care South London at King's College Hospital NHS Foundation Trust."
Finally, there were minor formatting errors throughout the article, mainly involving the inconsistent use of hyphens, which have now been corrected.
The online version of the original article can be found under doi:10.1186/s12909-015-0510-3.
| {
"pile_set_name": "PubMed Central"
} |
INTRODUCTION
============
The way science is done or everyday practice of science is a major theme of Fred Grinnell\'s most recent work. The author presents a critique of the linear model of science followed by many scientists and its failure to represent how scientists really work. It is not a surprising observation to most of us. What is surprising about this book is the way it provides various avenues to engage in productive discussions about doing research. It is not a perfect book, but it manages to be provocative and an easy read with only six chapters in fewer than 200 pages. The book is divided into two sections: "Science" and "Science and Society." Grinnell presents a candid account of the scientific process to make it relevant to those outside academic science. He also provides a fairly accurate view of scientists and how they work.
THE PRACTICE OF SCIENCE
=======================
In chapter 1, "Practicing Science," Grinnell highlights what he calls the two conversations of science by illustrating the interaction of two processes: the circle of discovery and the circle of credibility (p. 5, Figure 1.1). He is quick to point out that researchers interact with a small part of the world, limiting those interactions to making observations and doing experiments. The interactions within the research community are primarily collaborative and competitive in nature. As Grinnell progresses in his discussion of these interactions, he reveals some of the uncertainties and conflicts that emerge during the practice of science. Chapter 2, "Discovery: Learning New Things about the World," deals with the nonsystematic, nonlinear nature of the scientific process. For instance, Grinnell points out that scientific papers rarely describe scientific failure and rarely communicate to students the notion that "10 research notebooks' worth of experiments might be required to publish a 10-page research paper" (p. 21). This is an interesting assertion that indicates to the nonscientist reader the level of compression that goes on when myriad observations, deductions, and analyses are meshed into a coherent scientific paper. Another interesting idea presented in this chapter deals with the type of question(s) a research group decides to answer. Very often, Grinnell adds, "resource limitations...prevent new initiatives from starting even if they would be worthwhile to carry out. Investing in one project almost always means that something else will not be accomplished" (p. 35). This statement is important, as it calls to the attention of future researchers the various factors that control the nature of everyday practice of science. Chapter 3, "Credibility: Validating Discovery Claims," makes a case for the uncertainty that surrounds achieving credibility. Grinnell shows in Figure 3.1 (p. 64) his version of the credibility process, which involves a rather complex web of researchers interacting with one another, editors and reviewers, and even the public. It seems as if credibility does not necessarily come right away, and a scientist must use her or his intuition and a large dose of optimism in attempting to achieve credibility. Research grants are shown as a major determinant of the credibility process (p. 79, Figure 3.2), provided the research proposal aligns with the priorities of the funding agency (pp. 80--81).
SCIENCE, SOCIETY, ETHICS, AND RELIGION
======================================
The second part of the book presents some interesting notions about research ethics and misconduct with which every graduate student should become familiar. Grinnell provides some interesting notions about the influence interest groups (p. 103) and grant reviewers may have on the integrity of the everyday practice of science. A particular example is the issue of intellectual property (pp. 122--126), which gives an aura of business to the research university. Grinnell takes a neutral posture on these issues, and the reader is allowed to make her/his own inferences about the long-term effects on the integrity of the scientific process. Chapter 5, "Informed Consent and Risk: The Intersection of Human Research and Genetics," deals with various issues, such as principles of human research ethics, ethical challenges, and genetics research and vulnerability. This particular chapter should be most useful to students and teachers discussing recent examples of clinical trials. Chapter 6, "Faith: More Than One Way to Practice the World," makes an intersection between science and religion. Grinnell claims that "science and religion represent distinct human attitudes toward experience based on different types of faith" (p. 161). He compares both categories, and tries to dissect the complementarity between those two. I found this chapter to be the least effective. For instance, when he argues that religion "requires a different kind of faith than science but in no way gives up the demand for reason" (p. 169), he does not provide a convincing argument about the reasoning involved in faith. His suggestion that religion is the source of our values, working in a complementary way with science (p. 181, Figure 6.1), is an extremely controversial assumption with which many of the scientists mentioned in the book may disagree. He concludes the book by saying: "Perhaps solving global problems will require the scientific *and* religious attitudes---both types of faith---rather than one or the other" (p. 185). Perhaps not.
CONCLUDING REMARKS
==================
This book is an interesting addition to other books detailing the realities of science practice. The book appears to be aimed at a broad audience, which may include teachers, students, and those interested in science. However, I am not sure whether the book would be appealing to those outside the scientific community. In addition, many of the research examples are in biology, which may preclude some nonbiologists becoming as engaged as I did. I especially enjoyed those sections dealing with the process of inquiry, which may benefit those of us who are in the classroom presenting scientific ideas and literature. However, I would have liked to read about Grinnell\'s take on other forms of inquiry, such as those in the humanities and social sciences. Moreover, although Grinnell deftly presents the role of the scientist in the complex world of laboratory research, the additional role of the scientist as a citizen is not well developed. Instead, he chose to veer toward the notion of science and religion. The word *passion* is mentioned in the subtitle, but there is very little of it in the book. Perhaps in the near future Grinnell will treat us with an account of the scientist as a socially responsible individual, which leads me to my last remark. This book is a clear testament that we need more books that address issues of how to educate future scientists. While practicing science is most fascinating, understanding different types of inquiry, as well as the process of engaging in productive discussions with our students and colleagues, could make the everyday practice of science more passionate than it already is.
| {
"pile_set_name": "PubMed Central"
} |
Introduction {#Sec1}
============
For people leaving homeless or women's shelters, the transition to community living can be challenging. Because much has to be arranged during this stressful time, people are often in need of practical and emotional support. They can no longer utilize shelter services, which are generally terminated after shelter exit, and most of them have few supports that they can rely on in their new living environment (Herman et al. [@CR19]). Relationships with family members and other potential social supports may need to be repaired first and ties to professional supports in the community may be weak or not yet established due to waiting lists. As a result, people leaving shelters experience a discontinuity of support. Post-shelter services are, therefore, vital in preventing negative outcomes such as recurrent homelessness and re-abuse (Caton et al. [@CR6]; McQuistion et al. [@CR31]; Tan et al. [@CR43]; Tutty [@CR46]).
Critical time intervention (CTI) is a time-limited, strengths-based case management model designed to prevent adverse outcomes in vulnerable people at the time of a critical transition in their lives, such as following discharge from institutional settings (Herman et al. [@CR19]). CTI facilitates community integration and continuity of care by ensuring that a person has enduring ties to their community and support systems during these critical periods. It has been recognized by the Substance Abuse and Mental Health Services Administration (SAMHSA [@CR41]), the Public Health Agency of Canada ([@CR34]), and the Coalition for Evidence-Based Policy ([@CR9]) as an evidence-based practice (EBP). In the United States, this intervention has been found to be effective in preventing recurrent homelessness and re-hospitalization as well as reducing psychiatric symptoms and substance use after the transition from shelters, hospitals, and other institutions to community living in people with severe mental illness (Herman et al. [@CR20]; Kasprow and Rosenheck [@CR26]; Susser et al. [@CR42]). Furthermore, CTI is a cost-effective alternative to usual care for mentally ill men moving from a shelter to the community (Jones et al. [@CR24]).
Few evidence-based interventions for vulnerable people leaving institutional settings have been tested rigorously outside the United States (de Vet et al. [@CR13]; Jonker et al. [@CR25]). Before EBPs are widely implemented internationally, it is necessary to test whether they are effective, because most of these practices have been developed to address place- and time-specific social issues (de Vet et al. [@CR13]). In addition, different nations usually have distinct systems of care, which may influence the effectiveness of interventions (Toro [@CR44]). Differences between systems of care might require adaptations of an intervention during implementation. These adaptations should be consistent with the model, so that its active ingredients are preserved. By evaluating whether CTI is effective outside the United States, we could possibly add to the evidence base supporting that this intervention's mechanisms of effect are not dependent on a particular social context or health care system. We initiated two multi-center randomized controlled trials (RCTs) to test the effectiveness and model fidelity of CTI for homeless people and abused women in the Netherlands.
In modern effectiveness research, the development and use of fidelity criteria is considered obligatory to asses model adherence, that is, the degree to which a given intervention has been implemented in accordance with essential theoretical and procedural aspects of the model (Bond et al. [@CR2]; Hogue et al. [@CR22]). Earlier research shows that faithfully implemented EBPs produce better outcomes. For example, high fidelity to assertive community treatment (ACT) and strengths-based case management has been found to have a positive effect on client-level outcomes (Cuddeback et al. [@CR12]; Fukui et al. [@CR16]; McHugo et al. [@CR30]).
So far, only one study has published CTI fidelity scores (Olivet [@CR33]). This study was conducted by the Center for Social Innovation (C4) to assess differences in implementation and client outcomes between face-to-face and online CTI training. Fidelity was measured with the CTI fidelity scale, a quantitative tool developed by Conover and Herman ([@CR11]). The CTI fidelity scale consists of 20 items, which are rated on a five-point scale ranging from *not implemented* to *ideally implemented*. Item-level ratings can be combined to compute an overall fidelity score (Conover [@CR10]). In the C4 study, overall fidelity scores were calculated nine months after training and were based on compliance fidelity, which is the degree to which providers implemented the key elements of the CTI model (eight items), and chart quality, which measures how well the intervention was documented (four items). The 15 North American homeless service agencies that participated in the C4 study obtained an average overall score of three on the five-point scale, which corresponds to *fairly implemented* according to the CTI Fidelity Scale Manual (Conover [@CR10]). The present study was designed to provide insight in the implementation of CTI practice in three different ways. Firstly, we also conducted a fidelity assessment, which would allow us to examine whether a similar fidelity score would be achieved in the Netherlands as was obtained by the C4 study in North America. Secondly, we set out to compare the level of fidelity between two distinct service delivery systems---services for homeless people and services for abused women---which was possible because the two RCTs on CTI employed the same ongoing training and monitoring efforts during the same period in each service delivery system (Lako et al. [@CR28]). Earlier studies of the effectiveness of CTI have already demonstrated that the CTI model can be successfully adapted for several types of populations (Herman and Mandiberg [@CR21]). However, the hypothesis that CTI is suitable for a range of populations would be supported further if similar levels of model fidelity could be obtained in different service delivery contexts with the same implementation approach. Lastly, we aimed to provide insight into facilitators and barriers to CTI practice by conducting focus groups with the case managers trained in CTI (referred to as "CTI workers"). This will provide important information on which key aspects should be paid attention to when implementing CTI.
The present study will answer the following three research questions: What is the fidelity of CTI for homeless people and abused women making the transition from shelters to community living in the Netherlands? Is it possible to obtain similar fidelity ratings in two distinct service delivery systems (i.e., services for homeless people and services for abused women) with the same implementation approach? And which factors may have facilitated or impeded CTI workers to adhere to the CTI model in these service delivery systems?
Method {#Sec2}
======
Procedure and Participants {#Sec3}
--------------------------
This study is part of two RCTs examining the effectiveness of CTI for adult homeless people and abused women who are about to move to housing in the community and are willing to accept case management services during and after shelter exit. The two RCTs were initiated by the Academic Collaborative Center for Shelter and Recovery. The 18 shelter organizations that participated in these trials were members of this platform. In the Netherlands, services for homeless people are operated in a service delivery system that is separate from services provided to abused women; these two distinct service delivery systems will be referred to as *services for homeless people* and *services for abused women* in the remainder of this article.
Participant recruitment began in December 2010 and was completed in December 2012. In total, we recruited 183 clients from 18 homeless shelters, who had been rehoused within 14 months of entering the shelter, and 136 clients from 19 women's shelters, who had been victim to any violence committed by an intimate partner (intimate partner violence) or committed to protect or restore the family honor (honor related violence) and stayed in the shelter for at least 6 weeks before being rehoused. The trials comply with the criteria for approval by an accredited Medical Research Ethics Committee (aMREC). Upon consultation, the aMREC region Arnhem-Nijmegen concluded that these studies were exempt from formal review (registration numbers 2010/038 and 2010/247). The methods of the two RCTs have been reported elsewhere in more detail (Lako et al. [@CR28]).
Written informed consent to share client charts with the research team was obtained before participants were randomly allocated to CTI or care-as-usual. To assess the intervention's fidelity to the CTI model, we randomly selected a sample of 70 charts, stratified by service delivery system, from participants allocated to the experimental condition. \[Socio-demographic characteristics of these 70 participants are presented in the online supplement to this article.\] In the two trials, 164 participants were allocated to CTI. In July 2013, we assessed which client charts were available for the fidelity assessment. Fifteen participants allocated to CTI had never been assigned a CTI worker (*n* = 15) and, as a result, did not have a CTI client chart that could be included in the assessment. Reasons for not assigning a CTI worker were that participants refused to receive services after randomization (*n* = 8), organizations were unable to provide CTI due to full case-loads or participants' place of residence (*n* = 5), or participants were mistakenly assigned to another case manager (*n* = 2). For 17 participants, who had been allocated to CTI in the last 6 months of recruitment, the intervention had not yet ended and their CTI workers were therefore unable to supply these clients' charts. Earlier research has shown that implementation of an EBP with a sufficient level of fidelity takes time (Fukui et al. [@CR16]; Rapp et al. [@CR37]) and, therefore, CTI workers were expected to adhere more closely to the model at the end of the study than at the beginning. Because we aimed to draw a sample of charts representative for the study period as a whole, we decided to create temporal balance by excluding charts from participants who had been allocated to CTI in the first 6 months of recruitment (*n* = 33). To select a sample from the remaining charts available (*n* = 99), a computer-generated list of random numbers was used.
Tailoring the Model {#Sec4}
-------------------
CTI is divided into three phases, of 3 months each, with decreasing intensity of support over time (see Fig. [1](#Fig1){ref-type="fig"}). During the intervention the CTI worker provides practical and emotional support and helps to extend and strengthen the client's social and professional network. Gradually, responsibility for the client's care is transferred from the CTI worker to significant members from the client's social and professional support system (Herman et al. [@CR19]). Timing is crucial: An important principle of the model is that the CTI worker and the client have started building a working relationship before the actual transition begins (Herman and Mandiberg [@CR21]).Fig. 1Intensity and focus of services during the three phases of the CTI model
When CTI was first introduced to the Netherlands, the model was adapted to enhance continuity of services for people with schizophrenia and a history of homelessness (Valencia et al. [@CR48]). A pilot study tested the feasibility of implementing the adapted intervention. Adaptations were informed by data on housing instability among schizophrenia patients, interviews with clinicians and peer-specialists, and the investigators' clinical and research experience with hard-to-engage populations (van Hemert n.d.). One of the adaptations was a more flexible time frame compared to the original model. A cardinal element in the CTI model is that the phase transition is automatically made at the three-month time point rather than driven by readiness criteria. In the adapted intervention, the time frame could be altered depending on the complexity of clients' needs and problems, clients' and case managers' skills, and community factors, such as limited access to services due to waiting lists (Valencia et al. [@CR47]). This adaption fits well with a growing interest in the Netherlands for the concept of providing personalized care (Evers et al. [@CR14]). Decisions to transition to a subsequent phase were made by CTI workers and their supervisors during team meetings, which is an adaptation that was also incorporated in the implementation of CTI in the present study.
In addition to this model adaptation, implementation of CTI was also adapted to include elements from the strengths model (Rapp and Goscha [@CR36]). Since most of the participating organizations had implemented a strengths-based approach to shelter services shortly before the start of the trial, principles from both the CTI and strengths model were integrated to ensure continuity in service approach during the transition from shelter to community. Because the strengths model stimulates clients' capacity for autonomy and self-reliance by focusing on their strengths (Rapp and Goscha [@CR36]), it is very compatible with CTI.
Besides modifications to improve the fit of the CTI model with the health care system and shelter services in the Netherlands, the intervention was also tailored to meet the special needs of women (and their children) who have experienced abuse. Although the idea for CTI was conceived in the mid-1980s when many people with psychiatric disorders were becoming homeless, this model also seems to suit the complex service needs of women who have experienced abuse. Earlier research has shown that when these women successfully obtain desired community resources and increase their social support, this will enhance their overall quality of life. This improvement in well-being appears to serve as a protective factor from subsequent abuse (Bybee and Sullivan [@CR4]). We adapted the CTI model to employ practices familiar to the field, such as motivational interviewing (Millner and Rollnick [@CR32]), and include a number of key components geared toward helping these women address and prevent problems that they and their children face. The original six CTI areas of intervention, which were selected because these had been identified as the most essential for treatment of people with a severe mental illness during a 'critical time' of transition, were adapted in consultation with managers and practitioners from shelter organizations in the Academic Collaborative Center for Shelter and Recovery. The final 10 areas of intervention were based on experiences of practitioners as well as literature on risk and protective factors for re-abuse. These factors were also incorporated in the Risk and Needs Assessment, a tool in the CTI client chart that helps to assess individual risks for recurrent homelessness and/or re-abuse and discontinuity of care.
Training, Monitoring, and Support {#Sec5}
---------------------------------
Two or three case managers were drawn from existing staff of participating organizations to participate in the trials as CTI workers; they were generally part of service teams working in the community with vulnerable clients. Most of these case managers did not have any responsibilities within shelters. In order to qualify, staff members needed to have a bachelor's degree in social work or a related field. In the fall of 2010, potential CTI workers were introduced to CTI by the research team and experienced trainers. The CTI workers completed three one-day training sessions to become familiar with CTI's theoretical and procedural aspects and to acquire essential skills for CTI practice.
In addition to the initial training, CTI workers from all participating organizations attended centralized half-day training sessions---(bi)monthly during the first year and quarterly during the second year of study. With the aim of enhancing CTI practice, the research team and CTI trainer facilitated discussions in which workers from the participating organizations exchanged experiences, offered workshops on how to use CTI chart forms as tools for clients' care, and invited CTI experts to present methods for enhancing CTI model fidelity.
Participating organizations were required to assign an internal coach, who was responsible for ensuring sufficient organizational support for the CTI workers and monitoring the model fidelity of the intervention. To this end, CTI workers had biweekly face-to-face supervision with their internal coach. Coaches received a one-day training session at the start of the trials and four half-day training sessions during the study period.
For the implementation of an EBP to be effective, leaders in an organization need to be committed to the change process (Brownson et al. [@CR3]; McHugh and Barlow [@CR29]). Several steps were taken to secure leadership buy-in. Firstly, the RCTs were initiated by the Academic Collaborative Center for Shelter and Recovery and designed in consultation with this platform's steering committees and working groups, consisting of directors, managers, and practitioners from the member organizations. Secondly, each participating organization was visited at least twice by the research team before the start of the CTI training. During the first site visit, any possible challenges to the implementation of CTI were discussed with directors and managers. The second site visit was aimed at team leaders and practitioners to fill them with enthusiasm for the intervention. Lastly, presentations and workshops were conducted regularly at conferences and meetings to highlight the importance of the intervention, the trials' objectives, and the study progress. The aim of organizing and attending these conferences and meetings was to ensure (continued) leadership buy-in of the participating shelters and policy makers in local government and other funding bodies.
Fidelity Scale and Measures {#Sec6}
---------------------------
Fidelity was measured with the CTI fidelity scale, a quantitative tool developed by two of the authors (Conover and Herman [@CR11]). The CTI fidelity scale has been applied in a number of settings; however, this scale has not been formally validated so far (Herman and Mandiberg [@CR21]). For the purpose of the two RCTs, the fidelity scale was adapted in consultation with the original authors and translated into Dutch. \[The adapted version of the CTI fidelity scale and the rationale for each item in the original scale are available in the online supplement to this article.\] Adaptations to the fidelity scale were for language as well as for elements from the strengths model. Items were not adapted to account for the planned change in the model with regard to flexibility in the time frame. Hence, the fidelity scale provided the opportunity to measure the deviation from the original model that resulted from this more flexible time frame.
Each item of the CTI fidelity scale consists of one to five criteria, which can be rated positively or negatively. In order to obtain fidelity ratings at item-level, the number of positively rated criteria is divided by the total number of criteria to calculate percentages. These percentages are then converted into a five-point scale rating (see Fig. [2](#Fig2){ref-type="fig"}). Finally, all item-level ratings are added up, divided by the number of fidelity items, and rounded to the nearest integer to compute an overall fidelity score (Conover [@CR10]).Fig. 2Conversion of percentages of positively rated criteria into five-point scale ratings
The 20 items of the original CTI fidelity scale belong to one of three sections that each measure a different component of model fidelity: compliance fidelity, competence fidelity, and context fidelity (Conover [@CR10]). The first section, compliance fidelity, is the degree to which workers practiced the key elements of the CTI model and is measured by eight items. Four of these indicate whether the intervention was delivered according to the intended CTI structure (i.e., as a nine-month intervention divided into three equal phases with a focus on up to three intervention areas): *Three Phases*, *Nine*-*Month Follow*-*Up*, *Time*-*Limited*, and *Focused*. The other four items are concerned with developing relationships with clients and their social and professional support systems: *Early Engagement*, *Early Linking*, *Outreach*, and *Monitoring*. The second section, competence fidelity, refers to the extent to which these key elements were delivered to clients with skill and attention to the craft (Fixsen et al. [@CR15]) and is measured by nine items. Four of them rate how well the intervention was documented: *Intake Assessment*, *Phase Planning*, *Progress Notes*, and *Closing Note*. The other five items measure program quality: *Worker's Role With Client*, *Worker's Role With Linkages*, *Clinical Supervision*, *Fieldwork Coordination*, and *Organizational Support* (Olivet [@CR33]). The third section, context fidelity, indicates whether the organizational requirements were met to allow the intervention's practice to operate smoothly (Fixsen et al. [@CR15]). Context fidelity is indicated by three items: *Caseload Size*, *Team Meetings*, and *Case Review*.
The organizations that participated in this study did not implement CTI throughout their organizations, but instead two to three CTI workers, who had other cases besides their CTI caseloads, operated independently within larger service teams for the benefit of the two RCTs. Due to this deviation in team structure and the small number of "active" CTI clients per worker at any time, conducting site visits as outlined in the CTI Fidelity Scale Manual (Conover [@CR10]) was not appropriate. Therefore, the items that measure program quality (five items) and context fidelity (three items) could not be rated and were excluded from the assessment. The remaining 12 items of the CTI fidelity scale, which measure compliance fidelity and chart quality, were retained. The CTI Fidelity Scale Manual prescribes that these 12 items are rated by reviewing client charts.
Fidelity Assessment {#Sec7}
-------------------
For the sample of 70 charts, we collected all the CTI chart forms that the CTI workers had completed. At a minimum, each CTI client chart had to contain an Intake Form, a Strengths Assessment, a Risk and Needs Assessment, an Activity Log, a Personal Recovery Plan for each phase, and a Closing Note. \[The content and function of each CTI chart form are described in the online supplement to this article.\] The Strengths Assessment and Personal Recovery Plan originate from the strengths model (Rapp and Goscha [@CR36]); these chart forms were adapted to include elements that increase their compatibility with the CTI model and that are essential for their use during fidelity assessment (Wolf et al. [@CR49]). CTI workers sent copies of the chart forms to the research team using postage-paid envelopes or e-mail. The research team tracked receipt of all forms in a password protected database. Digital copies of CTI chart forms were stored on a secure server and hard-copies of CTI chart forms were stored in locked cabinets.
Review of CTI chart forms and additional notes was conducted by two fidelity assessors, who were part of the research team and both had extensive knowledge of the CTI model. Agreement between assessors, derived from an independently rated subsample of 17 charts, was very high (Cohen's κ = .80). The fidelity assessors used CTI fidelity worksheets (Conover [@CR10]), which had been modified in line with the adaptation of the CTI fidelity scale, to record and rate the criteria of each item during chart review.
In addition to chart review, we conducted two focus groups with a convenience sample of CTI workers to assess which factors may have helped or hindered them to adhere to the basic components of the CTI model. The first focus group was conducted in February 2013 with CTI workers who supported abused women (*n* = 5) and the second group discussion was carried out in April 2013 with CTI workers who provided services to formerly homeless people (*n* = 6). Before the start of the focus groups, we obtained written informed consent from the participants. The questioning route was determined in advance and each focus group lasted approximately 110 min. During the interview process, the group moderator regularly restated or summarized information and then questioned the participants to determine accuracy. The group discussions were recorded and transcribed verbatim. Six weeks after the focus groups took place, meetings were organized to verify the results with CTI workers and internal coaches. Preliminary codes and themes, and carefully selected fragments from the focus group transcripts to illustrate these, were presented to the attendees, to which they could respond by correcting misinterpretations or adding more information.
Analysis {#Sec8}
--------
For each item of the CTI fidelity scale, percentages of positively rated criteria were calculated at client-level using IBM SPSS Statistics for Windows, Version 20.0. Mean percentages for all client charts together and separately for services for homeless people and services for abused women were subsequently converted into fidelity ratings and an overall fidelity score for competence fidelity and chart quality.
Because fidelity ratings on separate items and the overall fidelity score could not be calculated at client-level, we tested for differences between services for homeless people and services for abused women before converting percentages into the five-point scale ratings. Mann--Whitney *U* tests were conducted to test for differences in percentages of positively rated criteria at item-level. An independent samples *t*-test was employed for the average percentage across all items. Because the group sizes are relatively small, and the analyses may lack statistical power as a result, we also calculated effect sizes.
Transcripts from the group discussions were explored using thematic analysis. The two lead authors (RV and DL) familiarized themselves with the data by listening to the recordings and rereading the transcripts. From one of the transcripts, they independently selected fragments considered to be relevant to the third research question. The supervising author (JW) reconsidered the relevance of extracted fragments and coded them inductively, developing an initial code frame. The lead authors used this frame to code the second transcript, using deductive and inductive analysis. To determine the validity of the information obtained and the code frame, a second data source was consulted, which consisted of questions and concerns about implementation from the CTI workers and internal coaches, and responses to these questions and concerns from the research team, collected during the study period. This document was continuously updated and disseminated during the centralized half-day training sessions. One of the lead authors (RV) combined the final codes into overarching themes, which were reviewed by two other authors (JW and MB). Existing themes were refined and finalized in consensus among the authors.
Results {#Sec9}
=======
Fidelity Ratings {#Sec10}
----------------
Table [1](#Tab1){ref-type="table"} presents the percentages of positively rated criteria and fidelity ratings at item-level as well as the overall fidelity score for all client charts together (*n* = 70) and separately for services for homeless people (*n* = 35) and services for abused women (*n* = 35). Ratings of *Monitoring* (item 8) are based on a subsample of 63 client charts, because for seven clients---four clients from services for homeless people and three from services for abused women---the intervention had ended before phase 3 had begun. For all client charts together, the overall fidelity score for competence fidelity and chart quality is three out of five, which according to the CTI Fidelity Scale Manual indicates that fidelity to the CTI model is fair. On eight of the 12 items, CTI workers adhered fairly or well to the model; the other four items were not or poorly implemented.Table 1Percentages of positively rated criteria, fidelity ratings and overall fidelity score for all client charts together and each service delivery system separatelyFidelity scale itemsAll client charts togetherServices for homeless peopleServices for abused womenPercentageRatingPercentageRatingPercentageRatingCompliance fidelity Item 1: three phases251191311 Item 2: nine-month follow-up854844865 Item 3: time-limited613573663 Item 4: focused623563683 Item 5: early engagement653643653 Item 6: early linking653673633 Item 7: outreach724744703 Item 8: monitoring482452522Chart quality Item 9: intake assessment784724834 Item 10: phase planning492401583 Item 11: progress notes724764673 Item 12: closing note361401321Overall fidelity score603583623Ratings: 1 = not implemented, 2 = poorly implemented, 3 = fairly implemented, 4 = well implemented, 5 = ideally implemented
In relation to the intervention's structure, CTI workers had generally divided the intervention into three phases, but failed most of the time to start and end each phase within a two-week margin of the intended three-phase structure. As a result, *Three Phases* (item 1) received a rating of 1, indicating this aspect of CTI had not been implemented. CTI workers scored well on *Nine*-*Month Follow*-*Up* (item 2), indicating that most of the time they managed to stay in touch with their clients for nine months and there were few major gaps where clients disappeared. They found it more difficult, however, to also end the intervention on time; *Time*-*Limited* (item 3) received a fair rating. A fair rating was also obtained on being *Focused* (item 4), which prescribes that the intervention should be limited to a maximum of three intervention areas.
With regard to relationship development, CTI workers should have met clients several times before shelter exit in order to gain an understanding of their clients' histories; this *Early Engagement* (item 5) received a fair rating. *Early Linking* (item 6), which was also implemented fairly, prescribes that CTI workers maintain a high level of client contact during the first weeks after discharge and convene a joint meeting with family members and service providers to ensure continuity during this critical transition period. An element that was put into practice well is *Outreach* (item 7), which indicates that CTI workers regularly met in the community with clients and people in their support systems during phase 1. The poor rating on *Monitoring* (item 8) shows that, in phase 3, CTI workers had difficulty with adapting to their monitoring role; often, they met with or spoke to clients too frequently in that last phase.
With respect to chart quality, required sections of the Strengths Assessment and the Risk and Needs Assessment, which are both part of the *Intake Assessment* (item 9), and the *Progress Notes* (item 11) in the Activity Log had generally been completed; CTI workers scored well on these items. Unfortunately, *Phase Planning* (item 10) information on the Personal Recovery Plans was often incomplete. In addition, important elements were missing from the *Closing Note* (item 12) most of the time. These items were not or poorly implemented.
Differences Between Service Delivery Systems {#Sec11}
--------------------------------------------
To compare the level of model fidelity between services for homeless people and services for abused women, we tested for differences in percentages of positively rated criteria at item-level and in the average percentage across all items. According to the independent samples *t*-test, the average percentage of positively rated criteria across all items did not differ between the two service delivery systems (*t*(68) = −1.42, *p* \> .05). When percentages of criteria met at item-level were compared, we found a trend for three items (*p* \< .10). CTI workers providing services to homeless people seem to be more careful to complete their *Progress Notes* (item 11; *U* = 461.50, *p* = .07), while CTI workers providing services to abused women seem to adhere better to the criteria regarding the *Intake Assessment* (item 9; *U* = 469.00, *p* = .06) and *Phase Planning* (item 10; *U* = 461.00, *p* = .06). For all three items, the effect size was small (*r* = −.22).
CTI Workers' Perceptions {#Sec12}
------------------------
The eight factors that emerged as prominent themes affecting model adherence are discharge and shelter services, working relationship, clients' needs and attitudes, community support system, perceived effectiveness, model adaptation and trial design, organizational and team support, and tools and training. These themes are described below.
### Discharge and Shelter Services {#Sec13}
During the focus groups, CTI workers confirmed that continuity of care is crucial for a smooth transition from shelter to community living. Filling out an Intake Form together with a client and shelter case manager before discharge resulted in fewer loose ends once the client had moved. Most of the workers agreed that if they had been unable to engage clients before discharge, three months was too short for the first phase (*Three Phases*). Being assigned to clients who had already left the shelter made organizing a meeting with the client and shelter case manager (*Early Engagement*) more difficult, because often shelter case managers would be unavailable or clients too preoccupied, according to the CTI workers."So after a month and a half I got \[the meeting with the shelter case manager\]. That's how it works... Every time it's like: The person that's responsible is never there when you need them and that's why it goes wrong all the time. And so I didn't have any information from the client's chart at all, so I just had to completely rely on the client at that point, and I really felt the lack of that meeting."
With regard to shelter services, working towards similar goals and with similar chart forms during shelter stay facilitated adherence to the CTI model. If clients had already completed, for instance, a Strengths Assessment in the shelter, then this version could be used by the CTI workers as a basis to expand from (*Intake Assessment*). CTI workers indicated that if clients had worked on strengthening their informal network with their shelter case manager, they seemed more willing to accept help in this area after discharge, as the following comment by a CTI worker reflects:"But one thing you can sort out \[in the shelter\], I think, has to do with their social network.... If the network doesn't get mobilized while they're in the shelter, then it's very hard to mobilize it once they get their own place, because I've noticed clients are then like: I don't need that any more.... So I think the time to seek help is in the shelter. If you engage \[the network\] at that point, then you can keep it involved later."
### Working Relationship {#Sec14}
In the CTI workers' view, having a good working relationship with a client was also instrumental in model adherence. Workers indicated that it could take several weeks, or even months, to build a positive working relationship. Being able to engage clients early to start building a positive relationship was an important facilitating factor. CTI workers were very positive about having a meeting with clients together with their shelter case managers before shelter exit:"And the reason why that worked so effectively was... well, it gives the client a sense of safety, like: 'Hey, my \[shelter\] case manager also thinks it's a good thing that I'm going to start working with you.' Quite primal, actually."
During the intervention, a trusting relationship between client and CTI worker appeared to be essential in helping to motivate clients. For example, several workers indicated that, even though some clients were reluctant at first, they had been successful in organizing a joint meeting with social supports (*Early Linking*) by following the client's lead.
### Clients' Needs and Attitudes {#Sec15}
Clients' support needs, as well as their attitudes towards receiving support, also had an influence on model fidelity. Some clients, for instance, were quite hesitant to accept support from other professionals besides their CTI worker. Workers experienced that, even though other supports were available, certain clients would keep appealing to them, resulting in frequent contact during the last phase of the intervention. Workers also felt inclined to increase the intensity of the intervention if a client's situation suddenly deteriorated, for example, due to an emotional or financial crisis. This could help to explain why the fidelity rating for *Monitoring* was poor. A crisis situation, however, could also motivate a client to become more accepting of help from others (*Outreach*), according to some of the CTI workers."Halfway through the second phase, they discovered a spot on my client's lungs. So then everything basically stood still for a while, but because of that, we did get to know his social network and could start drawing on that."
### Community Support System {#Sec16}
Model adherence also depended on workers' success in developing community support. During the focus groups, CTI workers indicated that sufficient community support was necessary to allow them to decrease and eventually terminate contact with a client (*Time*-*Limited*) and that a client's support system could help them gain more insight into a client's situation and restore contact with a client when it had been disrupted due to frequent no-shows (*Nine*-*Month Follow*-*Up*). Several workers experienced difficulty linking clients to professionals due to austerity measures and this lack of access had hindered them in moving from the first to second phase (*Three Phases*):"\[She\] had an intellectual disability -- or at least they \[shelter staff\] said 'suspicions of' -- and as soon as she was home again \[living in the community\], you could tell. I basically ran into a brick wall trying to refer her. From pillar to post: Go there, try this and that. And at some point that frustrated her so much that she started rejecting everything."
Others experienced that, due to time constraints, professionals were often unwilling to attend joint meetings with a client's support system, which according to the CTI model should be organized at the start and end of the intervention (*Early Linking* and *Closing Note*).
### Perceived Effectiveness {#Sec17}
Whether the workers perceived a certain component of the intervention as effective seemed to have had an influence on their willingness to adhere to the model. Several workers mentioned that CTI's three-phase structure fitted well with their clients' process of adaption to community living. For some of the workers, the decreasing intensity of CTI in the second and third phase meant they could spend more time with their clients during that first, crucial phase; they felt that they were able to match service intensity to their clients' needs thanks to the implementation of the CTI model. During the focus groups, workers discussed how the time-limited nature of the CTI model had helped them change their mindset: They would make better use of supports in clients' networks instead of providing support directly, especially in the second and third phase."You're already aware: During the first three months, I too will have to work very hard arranging and setting up practical things. But in the second phase you already start asking the client: 'Okay, how would you do that, what things could you consider? Who can you turn to?' And in the last phase you've resolved all that. You're in a different position then."
The CTI workers expressed that, as a result, they had generally been comfortable with ending the intervention at nine months (*Time*-*Limited*).
CTI workers expressed far less motivation to adhere to model components that they did not regard as beneficial. For example, the model prescribes that CTI workers organize a transfer-of-care meeting at the end of the intervention (which should be documented in the *Closing Note*). The transfer-of-care meeting is a joint meeting during which significant members from the informal and formal support system, along with the client, reach a consensus about the components of such an ongoing support system. In the view of several workers, having such a transfer-of-care meeting was unnecessary, because each member's role in the support system had already been discussed during a joint meeting in the first phase, and the system had been functioning well. Their perception of this element as redundant has most likely contributed to the poor fidelity rating for the *Closing Note*.
### Model Adaptation and Trial Design {#Sec18}
CTI workers mentioned that decisions about whether to move to a next phase with a client were made together with the internal coach and other CTI workers and were often based on a checklist of requirements for each phase---referred to as *anchor points*---(Wolf et al. [@CR49]), which had been provided to them during the training sessions. The workers found this checklist a helpful tool in deciding whether they could move on to subsequent phases:"Something to fall back on \[anchor points\] is great, because that's what you're working to achieve.... But you're also both aware -- the client and the practitioner -- that you've got that set amount of time to sort out those basic things.... So you just start to work. That's great."
The decision to move a client to a subsequent phase was ultimately made by the CTI workers to enable them to provide personalized health care. This represents a considerable deviation from the CTI model, which most likely contributed to the poor degree of fidelity to *Three Phases*.
As mentioned before, CTI workers had other cases besides their CTI caseloads and operated independently within larger service teams. At the beginning of the recruitment period and at certain recruitment sites where few clients were eligible to participate, workers had few active CTI clients, which, according to the CTI workers, made it difficult for them to internalize the CTI model. Some of the CTI workers mentioned they were expected to have full standard caseloads at all times by their team supervisor. If a new participant had been assigned to CTI, they would often have to transfer clients receiving usual services to colleagues, or would sometimes be pressured to work overtime.
### Organizational and Team Support {#Sec19}
CTI workers indicated that generally they felt supported by their organizations, although organizational support was lacking in some organizations with respect to chart documentation. Several workers had to maintain a second client chart that met all of the organization's standards, which may have resulted in less time spent on and lower quality of the CTI client chart. Furthermore, in one organization, standard procedures with regard to ending services after several no-shows were enforced for clients assigned to CTI, which directly contradicts with *Nine*-*Month Follow*-*Up*.
Having team meetings on a regular basis was crucial in adhering to the CTI model. According to the CTI workers, these meetings helped to reflect upon the delivery of the intervention and thereby reinforced activities that were consistent with CTI principles:"Reflection \[during team meetings\]. But also right in the middle of your work when someone suddenly reminds you: 'Why haven't you reached that point yet?'"
Although the CTI model stipulates that team meetings should be organized every 2 weeks, several CTI workers mentioned that they met less often and did not feel properly supported by their internal coach. Reasons for having infrequent team meetings were having to travel large distances to meet, having too little time to meet due to full caseloads, and having little reason to meet due to the small number of active CTI clients.
### Tools and Training {#Sec20}
Having the right tools, and sufficient training to use them to a client's advantage, facilitated adherence to the model as well. For example, CTI workers mentioned that the Personal Recovery Plan helped clients to set attainable short-term goals (*Phase Planning*), because clients had to indicate on a five-point scale how likely they were to achieve each goal in the next three months. Several workers mentioned that the ecogram, a tool to visually map support systems (Hartman [@CR17]), proved to be helpful, especially with clients who relied heavily on their CTI workers. Drawing an ecogram together with the client made clear who else was available for support in their network, which, in turn, made it easier for the CTI worker to "pull back" when the intervention progressed (*Time*-*Limited*)."For example, I'd had a client make an ecogram.... Then I covered up somebody's name with my thumb and said, 'What happens if she's not around?' That was somebody who was to come hang the light fixtures. 'Oh, well then I'll get my uncle to come round.' And then we did a few more, and at some point I put my thumb on my own name, and then she said something like, 'Yeah... well perhaps I could phone my aunt sometime.' And then it dawned on her: Hey, who could I call on then if he's not available anymore?"
Discussion {#Sec21}
==========
The first and second aims of the study were to establish fidelity to the CTI model of an intervention for homeless people and abused women moving from shelters to community living in the Netherlands and to show whether it is possible to obtain similar CTI fidelity ratings in two distinct service delivery systems (i.e., services for homeless people and services for abused women) when the same implementation approach is employed during the same period. With an average of 60 % of positively rated criteria across all items, the intervention received an overall fidelity score for competence fidelity and chart quality of three out of five, which indicates CTI was fairly implemented according to the CTI Fidelity Scale Manual. This finding is similar to the overall fidelity rating in a previous multisite CTI study conducted with 15 service agencies in the United States and Canada (Olivet [@CR33]). In the present study, the degree of fidelity on individual items ranged between not implemented (*Three Phases* and *Closing Note*) and well implemented (*Nine*-*Month Follow*-*Up*, *Outreach*, *Intake Assessment,* and *Progress Notes*). The two service delivery systems did not differ significantly on any of the items, although trends on three items related to chart quality were found. Effect sizes for these trends were small. This finding supports the hypothesis that CTI can be adapted for use with various populations, as suggested by Herman and Mandiberg ([@CR21]). Further research is needed to investigate whether this assertion holds when context fidelity and program quality, which are measured with the eight items from the CTI fidelity scale that were omitted in the present study, are taken into consideration. So far, however, the evidence seems to support that CTI's context-sensitive timing is applicable to a range of service delivery systems that serve vulnerable populations. Perhaps, that is due to the fact that its program components were developed in collaboration with practitioners, which lead to a pragmatic intervention that may be somewhat atheoretical in nature (Jenson [@CR23]).
The third study aim was to report CTI workers' views on factors that may have facilitated or impeded adherence to the CTI model. From these factors, eight overarching themes emerged: discharge and shelter services, working relationship, clients' needs and attitudes, community support system, perceived effectiveness, model adaptation and trial design, organizational and team support, and tools and training. CTI worker's perceptions on factors that influence service delivery have been studied previously in a sample of 12 practitioners using CTI in a community agency or clinical trial setting in New York City (Chen [@CR7]; Chen and Ogden [@CR8]). Four of the themes that emerged in the present study---discharge and shelter services, working relationship, community support system, and organizational and team support---relate to the findings of this earlier study. Similarly to the CTI workers in the present study, practitioners interviewed by Chen ([@CR7]) stressed the importance of establishing contact with a client before the transition to a community residence. Not only have the benefits of early engagement been reported by practitioners, its effects on housing outcomes have also been empirically established (Herman et al. [@CR20]). The present study corroborates the importance of fostering a trusting relationship to enhance client motivation and following clients' leads as a practice strategy, as previously established by Chen and Ogden ([@CR8]). Furthermore, CTI workers at community agencies in New York City revealed making frequent use of their own agencies' existing service programs (Chen [@CR7]), which highlights the importance of easy access to community supports. In addition, they experienced that organizational policy occasionally conflicted with essential elements of the CTI approach, which was also the case in the present study.
Although the other four themes that emerged from the present study---clients' needs and attitudes, perceived effectiveness, tools and training, and model adaptation and trial design---were not corroborated by earlier research on CTI practice, parallels can be drawn with findings from other studies of EBP implementation in mental health services. In a study conducted in child and adolescent mental health settings, clients' concerns (for example, about the fit of an EBP with their own needs) and clients' values were identified as factors affecting implementation (Aarons et al. [@CR1]). In adult mental health services, clients have also expressed concerns that EBPs will result in limited choice in service options and less say in the specifics of their services (Scheyett et al. [@CR40]). Integrating recovery principles with evidence-based interventions could be a good strategy to address concerns about the fit of EBPs with clients' support needs and attitudes towards receiving support (Torrey et al. [@CR45]). Concerning perceived effectiveness, Rapp et al. ([@CR35]) identified practitioners' resistance toward an EBP as a barrier to implementation at several community mental health centers; this initial resistance emanated from the practitioners' assumptions about what works that contradicted with the EBP. Although the CTI workers participating in the present study were generally enthusiastic about the intervention, their assumptions did have a negative influence on their commitment to implement certain model elements. For instance, workers who deemed the transfer-of-care meeting to be unnecessary when the support system was functioning well, were unlikely to organize such a meeting at the end of the intervention. The importance of tools and training is addressed in another paper by Rapp et al. ([@CR37]) which describes strategies for successful implementation of EBPs. The authors emphasize the importance of reinforcing the application of tools to achieve results, for example, by developing training units that focus specifically on the use of certain tools (such as the Strengths Assessment) in practice and by including tools in all aspects of systematic case review during team meetings. Regarding model adaption, CTI workers' views on phase transitions, as well as the fidelity rating for *Three Phases*, pointed towards a deviation from the original model in the present study. This deviation, however, was in line with an a priori decision to adapt the model by focusing on readiness instead of making phase transitions automatically at each three-month time point. Fidelity scales can be a useful instrument in measuring model adaptation of EBPs, as illustrated by a study that focused on transferring clients from an ACT program to a less intensive adaptation of the ACT model (Salyers et al. [@CR39]). Although programs which are more faithful to the original model have demonstrated better client outcomes, the need for adapting EBPs, which are generally developed in a particular socio-cultural and economic context, to local conditions has also been recognized (Bond et al. [@CR2]).
In this article, we have distinguished eight factors that influence model fidelity. Whether other factors that have been identified previously as facilitators or barriers to EBP implementation also apply to CTI practice in Dutch shelter services, warrants further research.
Strengths and Limitations {#Sec22}
-------------------------
Together with an evaluation of a strengths-based intervention for homeless young adults (Krabbenborg et al. [@CR27]), this study is the first to conduct a fidelity assessment of an evidence-based intervention in Dutch shelter services. Generally, few results from assessments of fidelity to the CTI model have been published (Herman [@CR18]) and none of these previous studies have compared levels of model fidelity in two distinct service delivery systems. Moreover, this study contributes to a better understanding of model fidelity and implementation, because it combined quantitative and qualitative data to answer the research questions related to this topic rather than using either approach on its own (Robins et al. [@CR38]). However, several limitations of the study need to recognized as well.
In the CTI Fidelity Scale Manual, cut-off points are provided to convert percentages of positively rated criteria into five-point fidelity ratings. In addition, norms are provided for how to interpret these ratings, ranging from *not implemented* (one out of five) to *ideally implemented* (five out of five). However, the CTI fidelity scale has not been formally validated so far and, as such, norms for good implementation have also not yet been established. Appropriate validation of the CTI fidelity scale is needed to determine whether the existing cut-off points and norms can be upheld.
Another limitation of the present study is that fidelity scores were calculated based on a subset of items from the original CTI fidelity scale. Because CTI was delivered to clients in a research context, participating organizations did not implement CTI throughout their organizations and the number of active CTI clients per worker was generally small. As a result, conducting site visits was not appropriate and eight of the 20 items of the original CTI fidelity scale, which measure program quality and context fidelity, had to be excluded from the fidelity assessment. If the omitted items would have been included, this could have altered the overall fidelity score as well as interpretation of the results. Inferences drawn based on the fidelity assessment are strictly limited to competence fidelity and chart quality and, based on these findings, no assumptions can be made about program quality or context fidelity of the intervention.
Nevertheless, valuable information about the context in which the intervention was delivered was obtained from CTI workers in focus groups. The use of focus groups, however, has certain limitations that should be highlighted, such as the possibility of social desirability and recall bias. Furthermore, data collected as the session progresses may represent opinions that are shaped by the group discussion (Carey [@CR5]). The members of the group should, therefore, feel comfortable with each other. In the present study, focus group participants knew each other and the researchers well through the ongoing training sessions and were assured that the information they provided would be anonymously reported on. Therefore, we expect the data to accurately reflect the opinions of the focus group participants.
Implications for Policy and Practice {#Sec23}
------------------------------------
The CTI fidelity scale and the assessment provide agencies and local policy makers with a framework for the development and quality assurance of EBPs that support vulnerable citizens during transitions in their lives. The identified facilitators and barriers to implementation might be used by policy makers and practitioners to improve fidelity to EBPs in shelter services and to provide the necessary conditions for successful implementation. Several recommendations for successful implementation of CTI can be made based on the study findings. First, staff should be committed to recovery and CTI principles, including the importance of fostering a good working relationship with clients. Important to model adherence is also their perception of the intervention's components as effective. Assessing whether these core principles are part of the organization's culture and the intervention's components are integrated into work processes before implementation, and, if necessary, helping staff to internalize those principles through knowledge transfer (Rapp et al. [@CR35]), would be recommended. Sufficient access to a community support system is also important; CTI programs are unlikely to reach high fidelity in environments where access to informal as well as formal supports is very limited. In addition, CTI workers should be provided with sufficient organizational and team support as well as ongoing coaching. Coaching should foster mutual learning by reflecting together on the CTI model during regular case review and on the use of CTI chart forms as tools to improve clients' care. Furthermore, workers should have full CTI caseloads to gain ample experience. Lastly, fidelity to the CTI model would improve if organizations integrate similar tools and principles in their residential shelter services and CTI workers are assigned at least several weeks before clients exit the shelter, which will enhance continuity of care during the transition from institutional to community living. In addition, training for shelter staff in how to enhance communication and collaboration pre-discharge could maximize the potential benefits from early engagement, as suggested by Chen ([@CR7]).
Conclusions {#Sec24}
===========
This study shows that CTI was fairly implemented in the two multi-center RCTs testing the effectiveness of CTI for homeless people and abused women in the Netherlands. In these distinct service delivery systems---services for homeless people and services for abused women---the same implementation approach, employed during the same time period, resulted in very similar overall and item-level fidelity ratings. These findings are in line with the results from earlier studies that found CTI to be effective in different service delivery contexts: CTI seems to be an intervention suitable for a range of vulnerable groups who are going through a transition in their lives. Analyzing CTI workers' perspectives on factors that may have influenced model fidelity has yielded important recommendations for successful implementation of CTI in other service delivery systems.
Electronic supplementary material
=================================
{#Sec25}
Below is the link to the electronic supplementary material. Supplementary material 1 (DOCX 37 kb)
Renée de Vet and Danielle A. M. Lako contributed equally to the study.
This research was funded by ZonMw, the Netherlands Organization for Health Research and Development, and the Academic Collaborative Center for Shelter and Recovery.
| {
"pile_set_name": "PubMed Central"
} |
{#sp1 .381}
| {
"pile_set_name": "PubMed Central"
} |
The measurement of cognitive functioning via neuropsychological (NP) testing is an important component of assessment after mild traumatic brain injury (mTBI), also known as concussion. A consensus statement on concussion in sport \[[@B1]\] concluded that such testing provides valuable information when evaluating a person with mTBI. The US military also mandates that service members are administered NP assessment to detect cognitive impairment associated with mTBI \[[@B2]\].
Traditional NP assessments are typically comprised of well-established measures with large normative databases and demonstrate evidence of adequate psychometric properties (i.e., reliability and validity). However, these tests are usually administered in a one-on-one format by a trained professional with paper, pencil and stopwatch, and require interpretation by a neuropsychologist. This can make them expensive and time intensive, not feasible for assessing large groups (e.g., athletic teams, service members) or using on the sideline or in combat settings. Over the past few decades, alternatives to traditional NP assessment batteries have emerged in the form of computerized neurocognitive assessment tools (NCATs).
NCATs offer several potential logistical advantages over traditional NP tests. They can be much less time consuming and do not require administration by a testing specialist. Scoring is automated, and test performance can be easily generated into a summary report for interpretation or an electronic spreadsheet for statistical analysis. Furthermore, NCATs allow for cognitive assessment to be obtained in geographic areas where traditional NP services are limited. They are easier to use for obtaining baseline tests (e.g., preseason, predeployment) that can be used for comparison to assessment after concussion, which can be especially advantageous where examinees may have conditions that prevent comparison to normative reference groups (i.e., abnormal cognitive development, ADHD, among others) \[[@B3],[@B4]\]. Also, the computerized nature of NCATs makes it possible to administer alternative forms of a test with numerous combinations of test stimuli, which mitigate practice effects, and allow for multiple administrations in a short amount of time to track recovery after injury \[[@B3]\]. Moreover, being computerized allows for more accurate measurement of reaction time (RT), possibly making NCATs more sensitive to subtle cognitive effects \[[@B5]\].
Despite potential advantages, NCATs are not without limitations, as discussed by Echemendia *et al*. \[[@B4]\]. Specifically, alternate forms might not be equivalent, computers settings can cause erroneous RT measurement, there are differences between administering to groups (as is often done for baseline administrations) and to individuals (as is often done postinjury), and the tests are marketed to professionals (e.g., athletic trainers) who may have little or no training in cognitive testing. Additionally, one of the most important limitations of NCATs is that the psychometric properties are not fully established. Although NCATs have gained momentum as a tool in the management of mTBI, particularly in military and athletic settings, commonly used NCATs have not undergone the same level of validation as many traditional tests. According to a Joint Position Paper of the American Academy of Clinical Neuropsychology and the National Academy of Neuropsychology \[[@B3]\], though NCATs may seem to be analogous to traditional NP tests, there are important differences between them that need to be explored. Specifically, modifications of existing measures warrant investigations of the new tests' psychometric properties, such as validity \[[@B6]\].
This manuscript will summarize and evaluate the existing literature regarding the validity of four NCATs commonly used for both clinical and research purposes: Automated Neuropsychological Assessment Metric (ANAM), CNS-Vital Signs (CNS-VS), Axon/CogState/CogSport (CogState) and Immediate Post-Concussion Assessment and Cognitive Testing (ImPACT; please see [Table 1](#T1){ref-type="table"} for a description of each NCAT). The tests that will be covered are commonly used in research and clinical settings. Specifically, CNS-VS has been used in several clinical trials and can be billed to medical insurance, ANAM is required for US military Service Members, CogState is commonly used in Australian athletic settings and ImPACT has been approved by the US FDA to detect cognitive deficits following mTBI. Interestingly, to date, these measures have not generated adequate evidence of validity, yet they are commonly used for TBI-related assessment in sports and military settings. This summary and review will focus on the comparisons of these tests to traditional NP batteries as well as evaluations of the ability of these tests to provide clinically meaningful information regarding cognitive functioning after concussion. The existing state of the literature will be evaluated based on criteria put forth by Randolph *et al*. in a 2005 \[[@B7]\] review of the literature regarding NP testing after sport-related concussion (SRC). That review and those criteria are discussed below (see [Box 1](#BX1){ref-type="boxed-text"}). This is not a systematic literature review, but is rather meant to serve as a concise summary and reference, with recommendations for future studies and considerations identified.
###### **Descriptions of computerized neurocognitive assessment tools reviewed.**
----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
**Test** **Subtests and cognitive construct intended to measure** **Classification and summary scores**
---------- ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------ ----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
ANAM4 Simple Reaction Time (SRT): visuomotor processing speed, simple motor speed and attentionProcedural Reaction Time (PRO): processing speed, visuomotor reaction time and attentionCode Substitution Learning (CDS): visual scanning, visual perception, attention, associative learning and processing speedCode Substitution Delayed (CDD): learning and delayed visual recognition memoryMathematical Processing (MTH): basic computational skills, concentration and working memoryMatching to Sample (M2S): visual-spatial processing, working memory and visual recognition memorySRT, Second Administration (SRT2): index of attention (i.e., reaction time \[RT\] and vigilance) Summary scores:Throughput (TP; number of correct responses per minute of available response time)Standardized subtest TP -- standardized composite TPComposite score (standardized average TP \[*z*-score\])Classification of impairment: ANAM composite *z* ≤ -1.28
CNS-VS Verbal Memory Test, Immediate (VBM): word recognition and memory, immediate and delayed recallVisual Memory Test, Immediate (VIM): visual recognition and memory, immediate and delayed recallFinger Tapping Test (FTT): motor speed, fine motor controlSymbol Digit Coding (SDC): information processing speed, complex attention, visual perceptual speedStroop Test (ST): SRT, complex reaction time, inhibition, executive skills, processing speedShifting Attention Test (SAT): executive functioning, RTContinuous Performance Test (CPT): sustained attention, choice reaction time (CRT), impulsivityVerbal Memory Test, Delayed (VBM): word recognition, memory and delayed recallVisual Memory Test, Delayed (VIM): visual recognition, memory and delayed recall Summary scores:Neurocognitive Index (NCI)Composite memoryVerbal memoryVisual memoryPsychomotor speedRTComplex attentionCognitive flexibilityProcessing speedExecutive functionSimple attentionMotor speed\
Composite score: IQ scale and percentile\
Classification of impairment: N/A
CogState Detection Task: SRTIdentification Task: processing speedOne Back Task: attention, working memoryOne Card Learning Task: learning and recognition memory Summary scores: score for each subtestComposite score: *z*-scoresClassification of impairment:-1.64 SD on at least two subtests\
CogState composite \< -1.64
ImPACT Word Memory, Immediate: verbal recognition memoryDesign Memory, Immediate: visual recognition memoryX\'s and O\'s: visual working memory and visual processing/visual motor speedSymbol Match: visual processing speed, learning and memoryColor Match: CRT and impulse control/response inhibitionFour Letters: working memory and visual-motor response speedWord Memory, Delayed: verbal recognition memoryDesign Memory, Delayed: visual recognition memory Summary scores:Verbal memoryVisual memoryVisual motor speedRTImpulse control\
Composite score:Test-specificStandardized scores and percentiles\
Classification of impairment: N/A
----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
Note: Each NCAT is online or desktop based and the approximate administration time, inclusive of testing and acquisition of medical history and demographics, is 30 min, with the exception of CogState (20 min). These summaries generally capture the 'standard' battery for each NCAT; however, the batteries used in the reviewed studies may include some variations and different combinations of subtests.
ANAM4: Automated neurocognitive assessment metric, version 4; CNS-VS: CNS-vital sign; ImPACT: Immediate post-concussion assessment and cognitive testing; IQ: Intelligence quotient; N/A: Not applicable; NCAT: Neurocognitive assessment tool.
Adapted with permission from \[[@B8]\].
Validity {#S0001}
========
Prior to discussing the literature on validity, it is important to establish what is meant by the term 'validity' (see [Box 2](#BX2){ref-type="boxed-text"}). Validity is the most important aspect of test construction and thus is a key consideration when evaluating the clinical utility of a test. In psychometric research, validity describes whether a test measures what it claims to measure, by meeting the criteria that have been established to determine its accuracy \[[@B6]\]. Various models of test validity have been proposed \[[@B9],[@B10]\], though in general, there are three ways to describe the validity of a test: by its content, by the construct it is purported to measure or by its ability to measure a certain criterion \[[@B11],[@B12]\].
Content validity describes the relevance of the test items to the construct that is to be measured \[[@B11],[@B12]\]. For example, determining if a test of attention is comprised of test items and stimuli that accurately and adequately measure attention, rather than some other construct, such as RT. Content validity is often assessed by the subjective agreement among subject matter experts, such as neuropsychologists, that the test items are relevant and appropriate for the test purpose. Construct validity describes the extent to which the measure represents the basic theoretical construct, such as cognitive functioning. It is primarily evaluated with correlations, regression or factor analysis between a domain of interest and other well-established, 'gold-standard' measures \[[@B11]\]. It is typically conceptualized as convergent and discriminant validity. Specifically, tests assessing similar constructs should have higher correlations (i.e., convergent validity) than tests of dissimilar constructs (i.e., discriminant validity). Criterion validity describes the relatedness of the measure to a specified criterion, such as a condition of interest or outcome (e.g., concussion), and is often divided into concurrent and predictive types of validity. Concurrent validity is determined by how well a test accurately identifies a diagnosis or condition of interest when that condition is known (e.g., control vs concussion cohorts), as compared with an existing 'gold standard'. Predictive validity is determined by the test\'s ability to inform about some type of future outcome. It is important to note that there can be some overlap in these different types of validity, as they are not meant to conceptually represent mutually exclusive subcategories of validity, but rather describe the various ways in which validity can manifest \[[@B12]\].
Past literature reviews {#S0002}
=======================
There are several existing literature reviews of NCATs, including those focused on one NCAT, such as ANAM or ImPACT, for example, \[[@B14],[@B15]\], as well as those focused on the broader body of NCAT literature, for example, \[[@B7],[@B16]\]. In a comprehensive review of literature on NP testing (traditional and computerized) in SRC published from 1990 to 2004, Randolph *et al*. \[[@B7]\] identified several gaps with regard to the use of both traditional and computerized NP testing after SRC. They proposed five criteria that needed to be satisfied with additional research in order to consider NP testing standard of care after concussion (see [Table 1](#T1){ref-type="table"}). Until these requirements are satisfied, the authors suggest that professionals should use NP tests, including NCATs, with caution and rely more on self-report measures and medical evaluations. In this literature review, we will place a specific focus on the validity-related 'Randolph criteria' (i.e., criteria two through five) in order to establish whether the existing research, including the research that has emerged since their review, sufficiently demonstrates that NCATs have satisfied those criteria and demonstrate adequate clinical utility.
Reviews since the Randolph *et al*. \[[@B7]\] paper seem to indicate that while the Randolph criteria have been partially addressed, there is still insufficient evidence that NCATs adequately satisfy the criteria. Resch *et al*. \[[@B8]\] conducted a similar literature review as Randolph *et al*. \[[@B7]\], though for research completed between 2005 and 2013, and for NCATs used primarily for SRC. The authors reported that the evidence of validity varies between NCATs, suggesting that more research is necessary in order to elucidate the relationship between NCATS and their traditional NP counterparts. Iverson and Schatz \[[@B16]\] conducted a literature review of NP assessment in SRC research and presented some evidence indicating that NCATs may be superior to their traditional counterparts because they can be more precise in the detection of cognitive impairment. However, all subsequent reviews, similar to Randolph *et al*.\'s \[[@B7]\] conclusions, suggest additional research is needed in order to further validate NCATs against their traditional NP counterparts and within mTBI populations.
Summary of literature {#S0003}
---------------------
The sections below provide a review of the literature published to date investigating the validity of the four NCATS: ANAM, CNS-VS, CogState and ImPACT. Their utility as a neurocognitive assessment is presented in two contexts: the extent to which the test measures the same constructs as traditional NP batteries and the extent to which it provides clinically meaningful information about group membership or cognitive impairment. The reader should refer to [Tables 2--5](#T2 T3 T4 T5){ref-type="table"} for details on the specifics of the methodology and findings of each of the studies described (as well as the full definitions of NP test-specific acronyms).
###### **Summary of Automated Neuropsychological Assessment Metric validity studies.**
--------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
**Study (year)** **Study sample** **Methods: tests** **Methods: statistics** **Results** **Ref.**
----------------------------------- ------------------------------------------------------------------------------ ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------ ---------------------------------------------------------------------------------- ------------------------------------------------------------------------------------------------------------------------------------------ ------------
Bleiberg *et al*. (2000) 122 healthy high school and college students Compared ANAM to WAIS-R, Finger Tapping, TMT A and B, CT total, PASAT, HVLT and Stroop Color-Word Test Pearson *r* correlation coefficients\ Correlations ranged from -0.60 to 0.66\ \[[@B17]\]
Stepwise regression PCA Stepwise regression: MTH and STN predict NP test scores\
PCA: identified a four-factor solution accounting for 66% of the variance
Kabat *et al*. (2001) 191 veterans referred for outpatient NP testing Compared ANAM to WAIS-R Digit Symbol Coding, CVLT and TMT A and B Pearson *r* correlation coefficients\ Correlations ranged from -0.64 to 0.66.\ \[[@B18]\]
Stepwise regression PCA CDS best predict TMT B and WAIS-R DS\
PCA identified two three-factor solutions accounting for 60--62% variance
Woodard *et al*. (2002) Uninjured high school athletes Compared ANAM to HVLT, COWAT, WAIS-III DS and SS, BTA Pearson *r* correlation coefficients Correlations ranged from -0.07 to 0.82.\ \[[@B19]\]
MTH most associated with traditional NP test scores\
MTH was significantly correlated with four of six traditional scores
Jones *et al*. (2008) 77 healthy college students Compared ANAM to WJ-III Pearson *r* correlation coefficients\ Correlations ranged from -0.10 to 0.55. Strongest correlation was between LGR TP and GIA scores\ \[[@B20]\]
Stepwise regression PCA LGR and MTH best predictors of WJ-III (Efficiency Index and Visual Matching)\
PCA: MTH identified as a distinct factor, accounting for 19% variance
Woodhouse *et al*. (2013) 143 patients referred for outpatient NP assessments.\ Compared ANAM to RBANS RBANS Total Index score (≤15th percentile) Pearson *r* correlation coefficients\ Correlations ranged from 0.01 to 0.52. Strongest correlation was between MTH and RBANS total scores\ \[[@B21]\]
RBANS used to group participants as impaired (n = 30) and not impaired (113) Logistic regression ANAM Composite TP score accounted for 37% variability in RBNAS Total Index Score\
ANAM subtest scores on predicting RBANS impairment: sensitivity = 81%, specificity = 89.1%, PPV = 56.7%, NPV = 87.9%, odds ratio = 34.65
Bleiberg and Warden (2002) US Military Academy cadets, 68 with mTBI and 16 healthy controls Administered ANAM at baseline and then again four-times over 2-week period (first 2 weeks of recovery for mTBI group) RCI-defined impairment\ mTBI group, RCI-based decline in 19% (2 scores) and 81% (1 score)\ \[[@B22]\]
Fisher\'s exact test Fisher\'s: MTH identified 21% of mTBI group.\
Fisher\'s: mTBI failed to demonstrate practice effect (80%)
Bleiberg *et al*. (1997) Six with mTBI and six healthy controls Administered ANAM and WAIS-R, WRAT-R, FFT, Stroop Color-Word Interference, PASAT, CVLT and COWAT\ MANOVA MANOVA: Group differences in three out of five of the ANAM subtests\ \[[@B23]\]
ANAM administered 30-times in four sessions over a 2-week period Practice effects in control group, while mTBI group\'s improvement was shorter and more variable
Kelly *et al*. (2012) 71 with acute mTBI and 166 controls, deployed in combat environments Administered ANAM, traditional NP battery and questionnaires within 72 h of injury ROC curve, including AUC Difference at enrollment:\ \[[@B24]\]
SRT + PRO, AUC = 0.73, discriminant ability = 71%\<br\> sensitivity = 59%, specificity = 82%\
Change from baseline: SRT + MTH + M2S, AUC = 0.79\
Discriminant ability = 75%, sensitivity = 53%, specificity = 98%
Coldren *et al*. (2012) 47 with mTBI and 108 healthy controls, deployed in combat environments Compared performance at predeployment baseline and ≤3, 5, 10+ days postinjury Mann-Whitney U Significant differences on five of six ANAM subtests at ≤3-day postinjury\ \[[@B25]\]
No differences at 5 and 10+ days intervals.
Norris *et al*. (2013) 165 soldiers with mTBI Correlated performance on ANAM with demographic variables (age, number of blast exposures, % of prior mTBI) at 3- and 5-day postinjury Spearman\'s *ρ* statistics\ Spearman\'s *ρ*: day 5 ANAM performance was associated with Return to Duty (RTD) rather than demographics\ \[[@B26]\]
*z*-scores\ SRT2 at day 3 most correlated with RTD time\
Kaplan--Meier plot Kaplan-Meier: lowest 25% took 19 days RTD, highest 25% took 7 days RTD
Register-Milhalik *et al*. (2013) 38 healthy college players and 132 college athletes with mTBI Compared ANAM and SOT and GSC. Control group was tested two-times (average of 4 days apart). mTBI group was tested two-times (preseason baseline and within 5 days following mTBI) RCI-defined impairment sensitivity and specificity ANAM (80% CI, average of seven subtests): sensitivity = 0.09, specificity = 0.95\ \[[@B27]\]
When combined with SOT and GSC, ANAM (80% CI): sensitivity = 0.50, specificity = 0.96
Nelson *et al*. (2016) 165 mTBI and 166 healthy controls, athletes Administered ANAM at baseline and then at 1, 8, 15 and 45 days (postinjury/baseline) ANOVA\ ANOVA: significant for 7 of 9 scores at day 1 (*d* ranged from -0.58 to -0.84). Limited significant differences at days 8, 15 and 45.\ \[[@B28]\]
Cohen\'s *d* effect sizes\ AUC: significant for 7 of 9 scores at 1-day assessment (0.63--0.73). Limited significant differences at days 8, 15 and 45
RCI-defined impairment\
ROC curve, including AUC
--------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
ANAM: Automated neuropsychological assessment metric; ANOVA: Analysis of variance; AUC: Area under the curve; BTA: Brief Test of Attention; CDS: Code substitution learning; COWAT: Controlled Oral Word Association Test; CT: Consonant Trigrams; CVLT: California Verbal Learning Test; DS: Digit Symbol; FFT: Finger Tapping Test; GIA: General intellectual ability index; GSC: Graded Symptom Checklist; HVLT: Hopkins Verbal Learning Test; LGR: Logical reasoning; M2S: Matching to sample; MANOVA: Multiple analysis of variance; mTBI: Mild traumatic brain injury; MTH: Mathematical processing; NP: Neuropsychological; NPV: Negative predictive value; PASAT: Paced Auditory Serial Addition Test; PCA: Principal component analysis; PPV: Positive predictive value; PRO: Procedural reaction time; RBANS: Repeatable Battery For the Assessment of Neuropsychological Status; RCI: Reliable Change Index; ROC: Receiver operating characteristics; RTD: Return to Duty; SOT: Sensory Organization Test; SRT: Simple reaction time; SS: Symbol Search; STN: Sternberg memory procedure; TMT: Trail Making Test; WAIS-III Wechsler Adult Intelligence Scale-Third Edition; WAIS-R: Wechsler Adult Intelligence Scale-Revised; WJ-III: Woodcock Johnson, Test of Cognitive Abilities-Third Edition; WRAT-R: Wide Range Achievement Test-Revised.
###### **Summary of CNS-Vital Signs validity studies.**
---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
**Study (year)** **Study sample** **Methods: tests** **Methods: statistics** **Results** **Ref.**
------------------------------ ---------------------------------------------------------------------------------------------------------------------------------------------------- ---------------------------------------------------------------------------------------------------- ----------------------------------------------- --------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- ------------
Gualtieri and Johnson (2006) 144 with neuropsychiatric disorders and 36 healthy controls Compared CNS-VS to RAVLT, WMS LM subtests, FTT, the ST, TMT B and the VF test Pearson *r* correlation coefficients Correlations ranged from -0.53 to 0.79\ \[[@B29]\]
Strongest correlation was between CNS-VS SDC and WAIS DS.
Lanting *et al*. (2012) 50 with mTBI Compared CNS-VS to NAB, RST, WTAR. Assessed at 6--8 weeks following injury Pearson *r* correlation coefficients Correlations ranged from 0.28 to 0.58\ \[[@B30]\]
Strongest correlation was between CNS-VS Psychomotor Speed score and NAB memory index standard score (*r* = 0.58)
Gualtieri and Hervey (2015) 179 with psychiatric disorders Compared CNS-VS to WAIS-III Pearson *r* correlation coefficients\ Correlations ranged from 0.33 to 0.59. Strongest correlation was between CNS-VS SAT and FSIQ\ \[[@B31]\]
Exploratory and confirmatory factor analyses\ CNS-VS SAT and VIM scores were the only significant predictors of FSIQ
Stepwise discriminant function analysis\
Logistic regression
Lanting *et al*. (2012) 50 with mTBI and 31 with orthopedic injury Administered CNS-VS at 6--8-week postinjury MANOVA No significant differences between groups \[[@B32]\]
Gualtieri and Johnson (2008) 145 controls and 141 examinees with TBI separated into four groups: PCS = 13; mTBI recovered = 15; recovered from STBI = 85, unrecovered STBI = 28 Administered CNS-VS.\ MANOVA\ MANOVA: 18 of 28 scores were significantly different. *Post hoct*-tests demonstrated significant differences on most CNS-VS scores between healthy controls and the four TBI cohorts, with mTBI recovered performing as well as healthy controls\ \[[@B33]\]
PCS tested within 3-month postinjury\ *Post hoc t*-tests comparing the five groups\ *Post hoct*-tests demonstrated the STBI groups performed significantly worse than the mTBI groups\
mTBI tested within 12-month postinjury\ ROC curve, including AUC AUC, group membership between injured and healthy groups: significant for psychomotor speed (0.75), NCI (0.75) and cognitive flexibility (0.71)
STBI time since injury unspecified
Dresch *et al*. (2015) 458 active-duty soldiers (deemed fit for duty) Administered CNS-VS, demographic questionnaires and biomarkers 30 days before and after deployment Cohen\'s *d* effect sizes Cohen\'s *d* = 0.22 on pre--post-deployment comparisons\ \[[@B34]\]
Cohen\'s *d* = 0.40 on postdeployment comparisons, with the sample divided into 'no traumatic stress' or 'traumatic stress groups'
---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
AUC: Area under the curve; CNS-VS: CNS-vital sign; FSIQ: Full Scale Intelligence Quotient; FTT: Finer Tapping Test; LM: Logical memory; MANOVA: Multivariate analysis of variance; mTBI: Mild traumatic brain injury; NAB: Neuropsychological assessment battery; NCI: Neurocognitive Index; PCS: Postconcussive syndrome; RAVLT: Rey Auditory Verbal Learning Test; ROC: Receiver operating characteristics; RST: Reynolds Intellectual Screening Test; SAT: Shifting attention test; SDC: Symbol digit coding; ST: Stroop Test; STBI: Severe TBI; TMT: Trail Making Test; VIM: Visual memory test; WAIS DS: Wechsler adult intelligence scale digit symbol; VF: Verbal fluency; WAIS-III: Wechsler Adult Intelligence Scale-Third Edition; WMS: Wechsler Memory Test; WTAR: Wechsler Test of Adult Reading.
###### **Summary of Axon/CogState/CogSport validity studies.**
-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
**Study (year)** **Study sample** **Methods: tests** **Methods: statistics** **Results** **Ref.**
-------------------------- ------------------------------------------------------------------------------------------------------------- ------------------------------------------------------------------------------------------------------------------------------ -------------------------------------------------------------------------------- --------------------------------------------------------------------------------------------------------------------------------------------------------- ------------
Makdissi *et al*. (2001) 240 healthy athletes. Six of original cohort sustaining acute mTBI and seven matched controls were retested Compared CogState\'s SRT subtest to DSST and TMT B\ Pearson *r* correlation coefficients for each group\ Correlations:\ \[[@B35]\]
mTBI group tested at baseline and then at 72-h postinjury\ MANOVA Better for controls than mTBI SRT and DSST: *r* = -0.48.x\
Control group retested ∼34 days following baseline SRT and TMT: *r* = 0.42\
MANOVA: mTBI group had significant (36%) decline in performance on CogState SRT\
RT variability was significantly different with the mTBI group
Collie *et al*. (2003) 240 elite athletes and 60 demographically matched controls Compared CogState to DSST and TMT-B Pearson *r* correlation coefficients Correlations ranged from -0.86 to 0.44\ \[[@B36]\]
Strongest correlation was between CogState Decision-Making Speed and DSST
Schatz and Putz (2006) 30 healthy controls Compared CogSport to TMT and WAIS-R DS. Also administered ImPACT\ Pearson *r* correlation coefficients Correlations ranged from -0.28 to 0.54\ \[[@B37]\]
Administered three-times over a 5-day period, at 48-h intervals Strongest correlation was between CRT and TMT A and B
Maruff *et al*. (2009) 215 healthy controls\ Compared CogState to GPB, TMT, SDMT, BVMT, RCFT and WMS-III SS subtest Pearson *r* correlation coefficients\ Correlations ranged from 0.49 to 0.83\ \[[@B38]\]
50 participants with cognitive impairment (mTBI, schizophrenia, AIDS Dementia Complex) ANOVA\ ANOVA: significant differences and medium to large effect sizes for impairment (0.60 to -1.80)\
Cohen\'s *d* effect sizes\ Non-OL%: CogState subtests could identify 53--78% impairment that was unique to mTBI alone
Non-OL%
Collie *et al*. (2006) 615 Australian Rules football players completed baseline assessments\ Compared CogState to TMT-B, DSST and standardized symptom checklist. Tests administered at baseline retested 11 days later.\ ANOVA\ Control group, no significant changes from baseline\ \[[@B39]\]
61 sustained mTBI and 84 healthy controls were retested Symptom assessment used to categorize examinees as symptomatic or asymptomatic *z*-score change Asymptomatic significantly declined on one of seven scores\
Symptomatic group significantly declined on three of seven scores (z-score change ranged from -0.60 to -0.86)
Louey *et al*. (2014) 29 athletes with acute mTBI and 260 healthy control athletes Administered CogState\ Reliable Change Index (RCI)-defined impairment\ ANCOVA: significant differences between groups on all scores\ \[[@B40]\]
Compared normative and baseline comparison methods to classify cognitive impairment (subtest scores ≤1.65 SD below mean) ANCOVA\ Cohen\'s *d* ranged from -0.94 to -2.95\
Cohen\'s *d* effect sizes\ Chi-square: baseline data, sensitivity = 96.6%, specificity = 86.9%, CCR = 87.9%\
Chi-squared analyses to determine diagnostic accuracy of impaired performance\ Normative data, sensitivity = 69%, specificity = 91.5%, CCR = 89.3%
CCR
Gardner *et al*. (2012) 51 rugby players with acute mTBI and 41 controls Compared CogState/CogSport to WAIS-III. Also administered ImPACT *t*-tests\ Significant differences between groups on four out of five CogState tests\ \[[@B41]\]
Effect sizes\ Effect sizes ranged from -0.46 to -0.88\
Logistic regression Logistic regression: CogState improved classification accuracy 3.5% beyond demographics
Nelson *et al*. (2016) 165 athletes with mTBI and 166 healthy controls Administered CogState at baseline and then at 1, 8, 15 and 45 days (postinjury/baseline) ANOVA\ ANOVA: significant differences for four of five scores at day 1 (*d* ranged from -0.51 to -0.72). Limited significant differences at days 8, 15 and 45\ \[[@B28]\]
Cohen\'s *d* effect sizes\ AUC: significant for four of five scores at day 1 (0.64--0.69). No significant differences at days 8, 15 and 45
RCI-defined impairment\
ROC curve, including AUC
-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
AUC: Area under the curve; ANCOVA: Analysis of covariance; ANOVA: Analysis of variance; BVMT: Brief Visual Memory Test; CCR: Correct Classification Rate; CRT: Choice reaction time; DS: Digit Symbol; DSST: Digit Symbol Substitution Test; GPB: Grooved Pegboard; ImPACT: Immediate post-concussion assessment and cognitive testing; MANOVA: Multivariate analysis of variance; mTBI: Mild traumatic brain injury; Non-OL%: Nonoverlap statistics; RCFT: Rey Complex Figure Test; RCI: Reliable Change Index; ROC: Receiver operating characteristics; RT: Reaction time; SDMT: Symbol Digit Modalities Test; SRT: Simple Reaction Time; SS: Spatial span; TMT: Trail Making Test; WAIS-III: Wechsler Adult Intelligence Scale-Third Edition; WAIS-R: Wechsler Adult Intelligence Scale-Revised; WMS-III: Wechsler Memory Scale-Third Edition.
###### **Summary of Immediate Postconcussion Assessment and Cognitive Testing validity studies.**
------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
**Study (year)** **Study sample** **Methods: tests** **Methods: statistics** **Results** **Ref.**
------------------------------ ------------------------------------------------------------------------------------------------------------------------------ ---------------------------------------------------------------------------------------------------------- --------------------------------------------------- ------------------------------------------------------------------------------------------------------------------------------------------------------ ------------
Iverson *et al*. (2005) 72 athletes with acute mTBI. Compared ImPACT to SDMT Pearson *r* correlation coefficients\ Correlations ranged from -0.60 to 0.70\ \[[@B42]\]
EFA EFA: revealed a two-factor solution describing, first, speed/RT and second, memory
Schatz and Putz (2006) 30 healthy controls Compared ImPACT to TMT and WAIS-R DS. Also administered CogState\ Pearson *r* correlation coefficients Correlations ranged from -0.506 to 0.641\ \[[@B37]\]
Administered three-times over a 5-day period, at 48-h intervals Strongest correlation was between CRT and TMT-A (0.54) and TMT-B (0.54)
Maerlender *et al*. (2010) 68 healthy controls Compared ImPACT to CVLT, BVMT-R, DKEFS, CPT, GPB and PASAT Pearson *r* correlation coefficients\ Correlations ranged from -0.39 to 0.59\ \[[@B43]\]
Canonical correlations Canonical correlations: two of the five canonical dimensions were significant (0.80 and 0.73), indicating the batteries measured similar constructs
Maerlender *et al*. (2013) 68 healthy controls Conducted additional analyses of Maerlender *et al*. (2010) data Point-biserial correlations Point-biserial correlation: ImPACT did not significantly discriminate between dissimilar measures \[[@B44]\]
Allen and Gfeller (2011) 100 healthy controls Compared ImPACT to the NFL NP battery: HVLT-R, BVMT-R, TMT, COWAT and WAIS-III Pearson *r* correlation coefficients Correlations ranged from -0.38 to 0.43.\ \[[@B45]\]
Strongest correlation was between WAIS-III DS and ImPACT Visual Motor Speed
Solomon and Kuhn (2014) 226 NFL draft picks, with and without mTBI history Compared ImPACT to Wonderlic Pearson *r* correlation coefficients Correlations ranged from -0.26 to 0.49. Strongest correlation was with Visual Motor Speed \[[@B46]\]
Van Kampen *et al*. (2006) 122 athletes with acute mTBI and 70 healthy controls Administered ImPACT at baseline and then at 2-day postinjury for mTBI group, and postseason for controls RCI-defined impairment\ 93% of the mTBI group performed lower than at their baseline\ \[[@B47]\]
Percent change from baseline 30% of control group performed lower than baseline
Broglio *et al*. (2007) 75 with mTBI high school students Compared ImPACT to HVLT, TMT, SDMT, DS, COWAT Chi-square ImPACT demonstrated better sensitivity to mTBI (62.5%) than traditional battery (43.5%) \[[@B48]\]
Covassin *et al*. (2008) 57 athletes with mTBI (36 with no history of mTBI, 21 with history of 2+ mTBI) Administered ImPACT at baseline and then at 1- and 5-day postinjury Chi-square and univariate *post hoc* tests mTBI group exhibited significantly lower scores on Verbal Memory (p = 0.01) and RT (p = 0.023) at 5-day postinjury \[[@B49]\]
Gardner *et al*. (2012) 51 rugby players with acute mTBI and 41 controls Compared ImPACT to WAIS-III. Also administered CogState ANOVA\ Significant differences between groups on 1/4 of the ImPACT scores. Cohen\'s *d* effect size = -0.57\ \[[@B41]\]
Cohen\'s *d* effect sizes\ Logistic Regression: ImPACT improved classification accuracy 3.5% beyond demographics
Logistic regression to distinguish between groups
Schatz *et al*. (2006) 72 high school athletes with mTBI and 66 healthy controls with no history of mTBI Administered ImPACT at 3-day postinjury MANOVA\ MANOVA: significant differences between groups on four out of five index scores (partial Eta^2^ ranged from 0.19 to 0.31).\ \[[@B50]\]
DFA: classification of group status DFA: sensitivity = 81.9, specificity = 89.4
Schatz and Maerlender (2013) Analyzed pre-existing data of 21,537 athletes' baseline assessments and 560 post-mTBI Performed factor analysis on preexisting ImPACT datasets Factor analysis Identified two primary cognitive factors: first, memory (verbal and visual memory) and second, speed (visual motor speed and RT) \[[@B51]\]
Schatz and Sandel (2013) Analyzed pre-existing data of athletes with mTBI (81 symptomatic and 37 asymptomatic), and matched controls (n\'s = 81 & 37) Administered ImPACT at baseline and then at 3-day postinjury for mTBI group, and postseason for controls DFA DFA: symptomatic vs controls\ \[[@B52]\]
Symptomatic group: sensitivity = 91.4%, specificity = 69.1%\
DFA: asymptomatic vs controls:\
sensitivity = 94.6%, specificity = 97.3%
Nelson *et al*. (2016) 165 athletes with mTBI and 166 healthy controls Administered ImPACT at baseline and then at 1, 8, 15 and 45 days (postinjury/baseline) ANOVA\ ANOVA: significant for four or five scores at day 1 (Cohen\'s *d* ranged from -0.70 to -0.80). Minimal significant differences at days 8, 15, or 45\ \[[@B28]\]
Cohen\'s *d* effect sizes\ AUC: significant for four of five scores at day 1 (0.70--0.71). Minimal significant differences at days 8, 15 and 45
RCI-defined impairment\
ROC curve, including AUC
------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
AUC: Area under the curve; ANOVA: Analysis of variance; BVMT-R: Brief Visual Memory Test-Revised; COWAT: Controlled Oral Word Association Test; CPT: Continuous Performance Test; CRT: Choice Reaction time; CVLT: California Verbal Learning Test; DFA: Discriminant Functional Analysis; DKEFS: Delis Kaplan Executive Functioning System; DS: Digit Symbol; EFA: Exploratory factor analysis; GPB: Grooved Pegboard; HVLT-R: Hopkins Verbal Learning Test-Revised; ImPACT: Immediate postconcussion assessment and cognitive testing; MANOVA: Multivariate analysis of variance; mTBI: Mild traumatic brain injury; NFL: National Football League; NP: Neuropsychological; PASAT: Paced Auditory Serial Addition Test; RCI: Reliable Change Index; ROC: Receiver operating characteristics; RT: Reaction time; SDMT: Symbol Digit Modalities Test; TMT: Trail Making Test; WAIS-III: Wechsler Adult Intelligence Scale-Third Edition; WAIS-R: Wechsler Adult Intelligence Scale-Revised.
Methods {#S0004}
=======
The search for primary literature involved several search engines (e.g., Google Scholar, PubMed, EBSCOhost and ScienceDirect). Articles were chosen based on their relevance to evaluations of the validity of the four above-mentioned NCATs. Specifically, the selection criteria were based on search terms such as ANAM, CNS-VS, CogState/CogSport/Axon and ImPACT in conjunction with any number of the following terms: validity, validation, construct validity, criterion validity, convergent validity, discriminant validity, diagnosis, group differences, sensitivity/specificity, mTBI and concussion. Studies were primarily included if analyses involved either, first, comparison of performance on NCATs and traditional NP tests or second, comparison of group differences in NCAT performance between healthy controls and individuals who sustained an mTBI. Revisions to this methodology (i.e., extending relevant study populations to those with neuropsychiatric disorders) were permitted as alternative ways of capturing measures of NCAT validity when search findings were insufficient. For example, we included several articles that studied adolescent samples as many studies of SRC combined high school and college athletes. In addition, several studies of non-mTBI samples (e.g., psychiatric disorders) were also included as they often compared NCAT scores to traditional NP tests in a group of healthy controls. Since this was not a rigorous and systematic literature review, the conclusions drawn should be considered with caution by the reader. However, we believe consolidating these findings in a single review that is invaluable for those interested in knowing where the literature currently stands in regards to the validity and clinical utility of these NCATs.
Automated neuropsychological assessment metric {#S0005}
==============================================
Comparisons to traditional NP tests {#S0006}
-----------------------------------
Research to date has largely demonstrated that scores on ANAM and traditional NP tests have weak-to-moderate correlations (see [Table 2](#T2){ref-type="table"} for more details on the methodology and findings of these studies). Bleiberg *et al*. \[[@B17]\] concluded that ANAM measures similar cognitive constructs as traditional NP batteries in a group of healthy controls, as correlations were generally moderate. ANAM throughput (TP) scores more strongly correlated with traditional NP test scores than RT and accuracy, and ANAM Mathematical Processing (MTH) and Sternberg Memory Procedure (STN) were most closely associated with scores from the Paced Auditory Serial Addition Test in the traditional NP battery. Kabat *et al*. \[[@B18]\] similarly found moderate correlations in a group of veterans, the strongest of which were between the ANAM Code Substitution Learning median RT and Trail Making Test (TMT) B. However, the median RT score is not a commonly used ANAM score for clinical or research purposes. In another study, with uninjured high school athletes, MTH demonstrated the most statistically significant correlations (i.e., moderate to strong) with a traditional NP score (Digit Symbol Coding) \[[@B19]\]. In a comparison of healthy college students' performance on ANAM and Woodcock Johnson, Test of Cognitive Abilities-Third Edition (WJ-III), Jones *et al*. \[[@B20]\] found some evidence of construct validity, as ANAM moderately correlated with many of the WJ-III subtests and clusters, with the strongest correlation between the WJ-III General Intellectual Ability index (GIA) score and the ANAM Logical Reasoning (LGR) TP score. Woodhouse *et al*. \[[@B21]\] additionally observed several statistically significant correlations between the ANAM and Repeatable Battery for the Assessment of Neuropsychological Status (RBANS), administered to a mixed clinical sample referred for assessments of cognitive functioning. These patients were diagnosed postevaluation with a variety of neurologic and psychiatric disorders. Each of the seven ANAM subtests was correlated with RBANS performance. The strongest correlation existed between ANAM MTH TP and the RBANS Total Index score.
Studies using regression analyses investigate the ability of ANAM to predict scores on traditional NP batteries. Results have generally provided evidence for construct validity, as certain ANAM scores can significantly predict performance on traditional NP batteries \[[@B17],[@B20]\]. MTH, STN and LGR appear to be the ANAM subtests that best predict performance on traditional tests such as Wechsler Adult Intelligence Scale-Revised (WAIS-R), WJ-III, TMT and RBANS. Principal component analysis (PCA) has also been used to further investigate the relationship between scores on ANAM and traditional NP tests. Generally, data from such studies also provide evidence for construct validity, demonstrating that ANAM is assessing underlying cognitive constructs of efficiency, working memory and resistance to interference similarly to traditional NP tests \[[@B17],[@B20]\].
Group differences {#S0007}
-----------------
Results from several studies to date suggest that ANAM may have some diagnostic utility in mTBI cases, particularly in the acute phase of injury (see [Table 2](#T2){ref-type="table"} for accompanying details on the existing literature to supplement this text). Bleiberg and Warden \[[@B22]\] administered baseline ANAM assessments to US Military Academy cadets and made comparisons in performance at four time points over a 2-week period (during the first 2 weeks after injury for those in the mTBI cohort). Using ANAM\'s Reliable Change Index (RCI) as a cutoff for impairment \[[@B53],[@B54]\]), ANAM scores were generally able to differentiate examinees with mTBI from healthy controls, as the mTBI group had two or more RCI-based performance declines, while controls did not. Additionally, significant practice effects were only demonstrated in the control group (53%). The authors suggested that the absence of a practice effect in the mTBI group might be one of the better ways to identify cognitive impairment following concussion.
An earlier study \[[@B23]\] investigating the diagnostic capabilities of ANAM as compared with traditional NP batteries found differences between the mTBI and demographically matched control group on four of the five ANAM subtests. Participants were tested 30-times over the course of 4 days (i.e., six-times on day 1, eight-times per day for the next 3 days) to attempt to replicate previous findings that examinees with mTBI had larger variability on measures of RT and response speed over multiple assessment sessions \[[@B55]\]. Their findings also revealed control participants demonstrated less variability and a practice effect over time, while the mTBI group\'s performance was more variable and actually declined across repeated test sessions.
In their assessment of a mixed clinical sample, Woodhouse *et al*. \[[@B21]\] used logistic regression to determine the classification accuracy of ANAM to predict RBANS scores that were ≤15th percentile (i.e., 'impaired'). All seven subtests generated significant differences among the groups of healthy and impaired individuals. This model indicates that ANAM TP scores can predict impairment status with high sensitivity and specificity, suggesting that ANAM is capable of classifying impairment similarly to the RBANS. However, the RBANS is typically used in the assessment of age-related cognitive decline, and therefore may not be the most suitable assessment for postconcussion evaluation. Kelly *et al*. \[[@B24]\] found that, in baseline and intergroup comparisons among concussed and healthy soldiers deployed to combat environments, the best area under the curve (AUC), which is an indicator of discriminant ability (i.e., differentiate between those with mTBI and controls), came from Simple Reaction Time (SRT) TP scores. The data suggest that this distinction may be the most accurate within 72 h of injury. Some other score combinations improved ANAM\'s discriminant ability (e.g., SRT + Procedural Reaction Time \[PRO\] for normative comparisons, SRT + MTH + Matching to Sample \[M2S\] for baseline comparisons), though not drastically so. These results possibly provide support for using only RT- and PRO-based tasks in a potentially shorter ANAM battery. Coldren *et al*. \[[@B25]\] also sought to evaluate the diagnostic capability of ANAM in the combat environment and found that, in comparison to controls and baseline scores, the mTBI group demonstrated lower scores with statistically significant differences on five of the six ANAM subtests scores at ≤3 days postinjury. However, only minimal differences were found at the 5 and 10+ days intervals.
Norris *et al*. \[[@B26]\] found that ANAM assessments at 3- and 5-day postinjury may demonstrate prognostic utility. In their study, those soldiers who performed at or lower than 25% needed 19 days to recover and be cleared to return to duty (RTD), while those who performed in the top 25% were able to RTD in just 7 days after injury, with the largest effect sizes seen for the SRT2 subtest. Results from another study suggest that the use of ANAM as a diagnostic tool for concussion may be limited. In a sample of college football athletes, few examinees with concussion were consistently classified as impaired across ANAM and a traditional NP battery \[[@B27]\]. In this study, ANAM had high specificity but low sensitivity; however, when combined with the Sensory Organization Test and symptom measures, the sensitivity improved, albeit only slightly. These results indicating low sensitivity raise questions about the isolated use of ANAM or any other concussion test (Sensory Organization Test or Graded Symptom Checklist) for clinical decision-making.
Finally and most recently, Nelson *et al*. \[[@B28]\] prospectively compared three NCATs, including ANAM, in groups of concussed and healthy athletes at 1, 8, 15 and 45 days following injury, with similar intervals for matched controls. At 1-day postinjury, AUC was fair and scores from six of the seven subtests as well as the composite score were significantly different from the control group. At days 8 and 15 postinjury, only one subtest (M2S) showed significant differences, and there were minimal significant differences at 45-day postinjury. The authors concluded that ANAM has limited clinical utility after 8 days following mTBI.
Summary {#S0008}
-------
When evaluating the existing literature on ANAM according to the Randolph criteria \[[@B7]\], it does not appear that these criteria have been sufficiently addressed. Correlations with traditional NP tests are generally moderate at best, though often weaker. Moreover, the stronger correlations are not consistently between tests that purport to measure the same cognitive construct. The scores that seemed to be the most robust from the ANAM were MTH and those that are primarily RT based, often most strongly correlated with traditional tests of motor and processing speed. Therefore, construct validity as measured by correlation analyses is questionable at best. However, there are indications from regression analyses and PCA that similar cognitive constructs are being measured by ANAM and traditional NP tests, though perhaps in a slightly different manner.
The results from the existing literature also suggest that ANAM has questionable sensitivity to the effects of concussion, especially if testing is completed more than a week from injury. While the mTBI groups often displayed worse performance, more variability in performance or a lack of practice effects as compared with controls, the diagnostic utility of these differences is currently unconvincing. Though specificity was often high and approaching clinically meaningful levels, sensitivity was generally lower than desired. However, there are indications that identifying cognitive impairment rather than mTBI status may be more meaningful and yield better diagnostic accuracy. This approach was recommended by Iverson and Schatz \[[@B16]\] and may be the best approach to addressing Randolph *et al*.\'s \[[@B7]\] second criterion evaluating the sensitivity to the effects of concussion.
CNS-Vital Signs {#S0009}
===============
Comparisons to traditional NP tests {#S0010}
-----------------------------------
There is not a large body of published literature regarding the validity of CNS-VS. The correlational studies suggest some degree of relatedness between CNS-VS and traditional NP tests, although no consistently clear patterns have been determined (see [Table 3](#T3){ref-type="table"} for details on the methodology and results of these investigations). Gualtieri and Johnson \[[@B29]\] found significant correlations between CNS-VS and a traditional NP test battery in groups of healthy controls and patients with various neuropsychiatric disorders, including postconcussion syndrome (PCS) and severe brain injury. CNS-VS Symbol Digit Coding and WAIS Digit Symbol Coding subtest scores were identified as the strongest correlated scores, providing some evidence of convergent validity. Another study \[[@B30]\] evaluating scores in a sample of examinees 6--8 weeks removed from concussion generated significant, though modest, correlations between CNS-VS and the traditional NP tests, with the strongest correlation between CNS-VS Psychomotor Speed Standard Score and the Neuropsychological Assessment Battery Memory index Standard Score. Gualtieri and Hervey \[[@B31]\] found that overall, correlations among the CNS-VS and traditional battery were weak to moderate (-0.33 to 0.59) in a sample of psychiatric patients. WAIS-III and -IV Full Scale Intelligence Quotient (FSIQ) and CNS-VS Shifting Attention Test were the most strongly correlated. The authors also conducted multiple regression analyses to further explore the relationship between CNS-VS and traditional NP tests, demonstrating that only two of the CNS-VS scores (Shifting Attention Test and Visual Memory Test) were significant predictors of FSIQ.
Group differences {#S0011}
-----------------
To date, there are three published studies looking at the diagnostic utility of CNS-VS with mTBI (see [Table 3](#T3){ref-type="table"} for study summaries to supplement this text). Lanting *et al*. \[[@B32]\] administered CNS-VS to patients 6--8 weeks after sustaining either an mTBI or orthopedic injury. Though the mTBI group did have a higher proportion of scores at least one standard deviation below the mean, effect sizes were small and multivariate analysis of variance demonstrated no statistically significant differences between the two groups. Gualtieri and Johnson \[[@B33]\] compared healthy controls to a TBI cohort divided into four subgroups: those with PCS, those recovered from mTBI, those recovered from severe TBI and those who had not recovered from a severe TBI. Multivariate analysis of variance demonstrated statistically significant differences between the five groups in 18 of the 28 scores investigated. *Post hoc* *t*-tests clarified significant differences from healthy controls existed on all scores in both severe TBI groups and on five of six CNS-VS scores in the PCS group. There were no differences between the controls and those in the mTBI-recovered group. Receiver operating characteristic (ROC) curve analyses revealed which index scores better identified differences between the groups. A greater AUC identified those tests that could best distinguish between groups as the CNS-VS Psychomotor Speed index score, which had the greatest AUC (0.752), followed by the Neurocognitive Index (NCI; 0.747) and Cognitive Flexibility Score (0.708). Although these results indicate that the NCI may have some diagnostic capabilities, the authors question the clinical use of the NCI score, as it is currently only utilized in research settings and is not common in traditional NP assessment. Lastly, another study compared CNS-VS scores before and after deployment in active-duty service members. Though there were significant differences between examinees on pre- and postdeployment measures, there were no significant differences on CNS-VS performance \[[@B34]\].
Summary {#S0012}
-------
Unfortunately, the existing research suggests somewhat mixed, though largely unfavorable, results for the validity of the CNS-VS battery. Specifically, correlation analyses show no clear pattern of convergent or discriminant validity, and generally CNS-VS is at best moderately correlated with traditional NP tests. Group comparisons suggest no clear or clinically meaningful differences between groups with mTBI and control groups. However, there is a paucity of research for CNS-VS, and further investigation is required to address the Randolph criteria. Additional studies taking different approaches may yield different and more promising results. For example, more clinically meaningful differences may be evident when comparing those with acute mTBI to control groups, as the existing literature was based solely on assessments administered long after the acute postinjury timeframe. Additionally, Gualtieri and Johnson \[[@B33]\] found significant differences between still symptomatic groups and controls, suggesting identification of cognitive impairment or still symptomatic individuals may be more clinically meaningful in identifying someone as recently concussed or not.
CogState {#S0013}
========
Comparisons to traditional NP tests {#S0014}
-----------------------------------
Studies comparing CogState to traditional NP tests have typically focused on traditional tests of processing speed and executive functioning, generally finding some evidence for construct validity (see [Table 4](#T4){ref-type="table"} for more detail associated with the findings). Makdissi *et al*. \[[@B35]\] reported statistically significant, though moderate at best, correlations between the CogState SRT subtest and the traditional NP tests (i.e., Digit Symbol Substitution Test and TMT-B) in samples of healthy controls and patients with acute mTBI. They found increases in variability and latency of responses in the dataset from these injured players. In a similar study \[[@B36]\], correlations were weak between the CogState battery\'s accuracy scores and the DSST and TMT scores; however, when CogState speed scores were used, there were several strong correlations, most notably between the DSST and the decision-making and working memory speed scores (-0.86 and -0.72, respectively). Schatz and Putz \[[@B37]\] reported moderate correlations between CogSport and a traditional NP battery, with SRT being the strongest correlated score with WAIS-R Digit Symbol Coding. In a study where healthy controls' performance on CogState was compared with a larger battery of traditional NP tests, Maruff *et al*. \[[@B38]\] reported moderate-to-strong correlations between the various CogState domains (processing speed, attention, working memory and learning) and the traditional NP tests measuring similar constructs, suggesting support for the construct validity of CogState.
Group differences {#S0015}
-----------------
There have been four studies published to date looking at the difference in performance on CogState between healthy controls and mTBI patients, with mixed results (see [Table 4](#T4){ref-type="table"} for accompanying details on the methodology and results). In one of the earlier studies \[[@B35]\], traditional NP scores did not significantly change from baseline in either the concussed or nonconcussed samples of football players, though the concussed group did demonstrate a significant (36%) decline in performance on the CogState SRT task. *Post hoc* *t*-tests demonstrated that the control and concussed groups' SRT variability were not statistically different at baseline, but the concussed groups had significantly more RT variability at follow-up. Similarly, a prospective study of cognitive functioning following concussion in football players found that the symptomatic group of patients who sustained a concussion demonstrated a significant decline in CogState performance and no change in traditional NP test scores, while the controls and asymptomatic concussion groups mostly improved in their performance in both CogState and the traditional NP battery \[[@B39]\]. Maruff *et al*. \[[@B38]\] found evidence of criterion validity by administering CogState to three groups of examinees with cognitive impairment (mTBI, schizophrenia and AIDS Dementia Complex), and three groups of demographically matched controls. Of interest to this review, the mTBI group was significantly different from the control group, with large effect sizes observed on the OCL/learning task. In addition, the authors used a measurement called the nonoverlap statistic (non-OL%) to identify the percentage of each group\'s data distributions that do not overlap. Using this metric, they found that each of the CogState subtests was able to identify between 53 and 78% of the impairment unique to the mTBI group (p \< 0.0001). Utilizing both baseline and normative reference groups, Louey *et al*. \[[@B40]\] have also provided evidence that CogState can be used to detect concussion-related cognitive impairment. The authors found that the baseline method demonstrated a higher sensitivity and comparable specificity to the normative method, and even after taking into account baseline performances, the concussed group showed performance declines. The baseline and normative methods could be used to correctly classify individuals as cognitively impaired up to an accuracy of 87.9 and 89.3%, respectively.
However, in two of the studies to date, comparisons among healthy and impaired individuals provided less convincing support of CogState\'s criterion validity, as CogState either only moderately improved classification between groups or, similar to ANAM, could do so only at earlier postinjury time points. Gardner *et al*. \[[@B41]\] administered CogState alongside ImPACT and WAIS-III to rugby players with or without acute mTBI. They observed statistically significant differences between the groups on four of the five CogState subtest scores. However, logistic regression demonstrated that CogState scores only minimally improved classification accuracy above what demographics predicted when added to the regression model. In their prospective study (previously mentioned in the ANAM review), Nelson *et al*. \[[@B28]\] found that at 1-day postinjury, all of CogState\'s subtests were significantly different from the control group. There were two subtests (Attention Speed and Learning Speed) that also demonstrated statistically significant differences at 8-day postinjury, though with small effect sizes. The ROC analyses revealed that only the CogState subtests administered at 1-day postinjury demonstrated significant AUC, suggesting that at the later time points, CogState subtests likely do not have diagnostic utility.
Summary {#S0016}
-------
Though correlations between CogState and traditional NP tests have been wide ranging, there is some evidence for convergent validity, with a general pattern of tests supposedly measuring similar cognitive constructs being more strongly correlated than dissimilar measures. Investigations of the clinical utility of CogState with concussion have had mixed results. Some tests have demonstrated the ability for CogState to distinguish between those with concussion and controls, even outside of the 7-day window (e.g., 1 month). In fact, one study suggested CogState may have had more clinical utility in postconcussion assessments than traditional NP tests \[[@B39]\]. Also, CogState was able to correctly classify over 88% of individuals as concussed or not by comparing scores to both normative databases and baseline performance \[[@B40]\].
Similar to research with other NCATs, the clinical utility of CogState may be increased by identifying individuals who are symptomatic after injury, rather than just comparing those with concussion (and possibly asymptomatic) to healthy controls. However, other studies have found that CogState\'s ability to detect postconcussive symptoms may be limited outside of the acute stage of injury (e.g., beyond 7 days), and even in the acute stage it may not provide much information beyond demographics. Overall, though the Randolph criteria have been largely addressed by the existing research, there is inconclusive support for meeting those criteria and additional research with CogState is warranted. The study samples and traditional NP test batteries used in the existing studies have been fairly narrow. Additionally, the wide range of correlations between CogState and traditional NP tests warrants regression and PCA to determine the extent to which CogState is measuring similar cognitive constructs to traditional NP tests. Also, additional studies investigating the clinical utility of CogState to detect cognitive impairment, both in and beyond the acute injury phase, are necessary.
ImPACT {#S0017}
======
Comparisons to traditional NP tests {#S0018}
-----------------------------------
Alsalaheen *et al*. \[[@B15]\] conducted a comprehensive and systematic review of the validity of ImPACT, and the intention is not to repeat their work. The reader is encouraged to consult their review for a comprehensive summary of ImPACT literature to date. Alsalaheen *et al*. \[[@B15]\] concluded that there is strong evidence for convergent validity of ImPACT though weak or inconclusive evidence for discriminant validity, criterion validity or diagnostic accuracy and utility. This would suggest inconclusive support for meeting the Randolph criteria \[[@B7]\]. Below, we highlight several studies investigating convergent and criterion validity, as well as diagnostic utility in mTBI cases (see [Table 5](#T5){ref-type="table"} for study summaries to accompany the findings described below).
Iverson *et al*. \[[@B42]\] compared ImPACT results with those of a paper and pencil test commonly used as a measure of attention and processing speed (Symbol Digit Modalities Test \[SDMT\]) in a cohort of young athletes. The strongest correlations with SDMT were ImPACT\'s Processing Speed and RT composite scores. Exploratory factor analysis (EFA) uncovered a two-factor solution of speed/RT and memory, suggesting ImPACT is measuring similar cognitive constructs as SDMT. Schatz and Putz \[[@B37]\] found moderate correlations among ImPACT and a traditional NP battery in a group of healthy controls, with the strongest correlation being ImPACT Choice RT score and Trails A. Similarly, Maerlender *et al*. \[[@B43]\] found that ImPACT was moderately correlated with tests of similar cognitive domains. Canonical correlation analyses indicated that two of the five canonical dimensions were statistically significant, with coefficients of 0.801 and 0.729, confirming that the two batteries generally measure similar cognitive constructs. However, a follow-up study by the same authors \[[@B44]\] re-analyzed the 2010 dataset to specifically evaluate the discriminant validity of ImPACT as compared with traditional NP tests. The results indicated that while the traditional battery demonstrated evidence of discriminant validity (i.e., all domains' p-values \> 0.05 except RT), ImPACT did not discriminate between measures of different cognitive skills. Specifically, three of the four domain scores were strongly correlated with expectedly different traditional NP measures.
Allen and Gfeller \[[@B45]\] compared performance measures of ImPACT to those of the NFL NP battery, which consists of the Hopkins Verbal Learning Test-Revised, Brief Visual Memory Test-Revised, TMT, Controlled Oral Word Association Test and three subtests from the WAIS-III, in a sample of healthy controls. Correlations were moderate at best, with the strongest correlation between WAIS-III Coding and ImPACT\'s Visual Motor Speed Composite. Solomon and Kuhn \[[@B46]\] examined the relationship between performance on the Wonderlic and ImPACT in 226 NFL draft picks with and without a history of concussion. Concussion history did not have a significant effect on performance on either of the tests. Correlations between the batteries were weak to moderate, with Visual Motor Speed being the most strongly correlated with Wonderlic performance.
Group differences {#S0019}
-----------------
Studies to date comparing ImPACT to a variety of traditional NP tests and among many different patient populations have corroborated that ImPACT may be useful as a diagnostic tool postconcussion, and perhaps even the most sensitive of the four NCATs described in this review (see [Table 5](#T5){ref-type="table"} for supporting summaries of each study). Van Kampen *et al*. \[[@B47]\] compared the ImPACT performance of college athletes with acute concussion to matched controls, also utilizing preseason baseline assessment scores. RCI scores defining abnormal performance indicated that 83% of the participants in the mTBI group performed abnormally lower than their baseline. When cognitive data were combined with symptom questionnaires, 93% were categorized as abnormal. However, 30% of the control group also generated abnormal ImPACT test data or self-reported symptoms. Broglio *et al*. \[[@B48]\] reported that groups of students with and without acute mTBI differed on all indices except Impulse Control. Furthermore, the ImPACT battery demonstrated better sensitivity to mTBI (79.2%) than a traditional NP battery (43.5%). Similarly, in a study of recently concussed college athletes, Covassin *et al*. \[[@B49]\] found that there were significant differences in Verbal Memory and RT based on whether participants had a history of prior concussion (i.e., those with a prior concussion performed worse) at both 1- and 5-day postinjury. Schatz *et al*. \[[@B50]\] also observed that ImPACT classified a group of recently concussed high school athletes with a sensitivity/specificity of 81.9/89.4. Schatz and Maerlender \[[@B51]\] performed factor analyses using existing ImPACT datasets, which included 21,537 baselines and 560 postinjury assessments. They identified two primary cognitive factors, memory (comprised of Verbal and Visual Memory domains) and speed (comprised of Visual Motor Speed and RT domains), that accurately classified individuals as concussed or not concussed, with a sensitivity/specificity of 89/70.
However, there has not been universal evidence that ImPACT adequately differentiates between healthy controls and recently concussed individuals. As previously mentioned, Gardner *et al*. \[[@B41]\] administered ImPACT, CogSport and WAIS-III to professional rugby players with acute mTBI and to matched controls. They found statistically significant differences between the groups on only one of the four ImPACT composite scores (Visual Motor Speed). Logistic regression demonstrated that ImPACT scores were unable to distinguish between the injured and control groups beyond demographic variables, as ImPACT scores only added 3.5% improvement in accuracy to the overall classification model. ROC curve analyses demonstrated modest sensitivity and specificity for the ImPACT composite score.
There is additional support for the clinical utility of ImPACT from studies investigating the test\'s ability to distinguish between symptomatic and asymptomatic mTBI patients. Schatz and Sandel \[[@B52]\] administered ImPACT to groups of high school and college athletes with acute mTBI (symptomatic and asymptomatic) within 72 h of injury. The data were compared with demographically matched controls with pre- and postseason assessments. ImPACT data demonstrated the ability to detect differences between the groups (sensitivity/specificity of 91.4/69.1 and 94.6/97.3 for the symptomatic and asymptomatic groups, respectively). In the prospective NCAT comparison by Nelson *et al*. previously described \[[@B28]\], all ImPACT composite scores were significantly different at 1 day following injury. However, there was only one score that was significantly different, and with a small effect size, after this timeframe (day 8, Verbal Memory, *d* = -0.40). Likewise, AUC for the composite scores were fair at the 1-day postinjury assessment \[0.70,0.71\] though poor (\<0.69) for the other timeframes. However, of the three NCATs evaluated, ImPACT demonstrated the highest percentage of test scores that significantly declined from baseline to 1-day postinjury according to the RCI criteria (67.8% for both symptomatic and asymptomatic concussed populations), although the test also had a slightly higher false-positive rate than ANAM and CogState in the same 24-h period (29.6% compared with 25.0 and 22.0%, respectively). When examinees were dichotomized as symptomatic or not, ImPACT also demonstrated the largest percentage of patients with a significant decline from baseline performance (53.8% at 1-day postinjury).
Summary {#S0020}
-------
ImPACT is the most widely studied of the NCATs, and as such the Randolph criteria \[[@B7]\] have been thoroughly addressed through the existing body of research. Though the Randolph criteria have been satisfied to a degree, as Alsalaheen *et al*. \[[@B15]\] concluded that there are mixed results regarding the overall validity of ImPACT. Specifically, there appears to be solid evidence that ImPACT has adequate relatedness with traditional NP tests, especially those purported to measure similar cognitive constructs. More advanced statistical approaches suggest there is also evidence that ImPACT is measuring similar cognitive constructs to traditional NP testing. However, there is not a clear pattern of weaker relationships between tests of dissimilar cognitive constructs, calling into question the discriminant validity of ImPACT. ImPACT\'s tests of RT and processing speed, especially Visual Motor Speed, seem to have the most robust relationships with traditional NP tests. And with regard to identifying postconcussion issues, ImPACT does show the ability to distinguish between concussed and noninjured individuals during the early stages postinjury. And though sensitivity is generally better than specificity, there were some studies that found comparable sensitivity and specificity, both of which approached desired levels for clinical decision-making. However, after the early postinjury stages, and certainly outside of 7 days, the clinical utility of ImPACT for postconcussion assessments appears limited. Improved clinical utility may be demonstrated if identification of symptomatic individuals postinjury is the focus, rather than identifying individuals as concussed or not.
Discussion {#S0021}
==========
The goal of this review was to provide a summary of literature regarding the validity on four commonly used and studied NCATs: ANAM, CNS-VS, CogState and ImPACT. The literature was viewed through the lens of Randolph *et al*.\'s criteria presented in their 2005 \[[@B7]\] literature review of NP testing after SRC ([Box 1](#BX1){ref-type="boxed-text"}). NCATs are becoming the standard of care for mTBI screening in athletic and military deployment settings given the improvement in efficiency and feasibility of test administration over their traditional NP counterparts. However, it is clear from the above summary of the literature to date that there has yet to be definitive evidence in support of the validity of any of the four NCATs, per Randolph\'s validity-related criteria (i.e., criteria two through five).
Currently, the body of literature suggests mixed results regarding NCATs' validity. Specifically, there is evidence that NCATs measure similar cognitive constructs as traditional NP tests (i.e., Randolph\'s 3rd criterion). And there is some support that NCATs, or at least components of each NCAT, can distinguish between individuals with acute concussion and healthy controls, or between still symptomatic individuals and individuals who are symptom free (i.e., Randolph\'s 2nd criterion and 5th criterion). However, there is little to no evidence for discriminant validity as compared with traditional NP tests, and inconsistent evidence for the clinical utility of NCATs for identifying concussion-related problems, especially beyond the first 7-day postinjury and when the tests are used in isolation. We did not review the literature regarding Randolph\'s 1st criterion, related to test--retest reliability, as this was beyond the scope of the paper. With regard to Randolph\'s 4th criterion, establishing RCIs and probability-based algorithms for clinical use is dependent on well-established test--retest reliability and well-defined constructs of the tests. That is, we need to know what the test is measuring, how it is measuring it and how consistently it does so before we can calculate them. As such, additional research will be needed before any of the NCATs fully satisfy the criteria for validity and ultimately for clinical utility.
Although there is not consistent evidence regarding the validity and clinical utility of NCATs, and the criteria presented by Randolph *et al*. \[[@B7]\] have not been sufficiently addressed, there is evidence suggesting that NCATs are of potential benefit in postconcussion assessments. It may be that the tests are fundamentally different than traditional NP tests, and therefore using traditional NP tests as a point of comparison, or using traditional psychometric approaches to defining validity creates a logical fallacy of false analogy or an 'apples to oranges' comparison. That is, perhaps NCATs should not be faulted for not being a good proxy for traditional NP tests, but rather should be investigated as an altogether different assessment tool. Therefore, we explore future directions for this field of research through the lens of the Randolph criteria.
Studies should seek to address Randolph\'s 2nd and 5th criteria by designing studies that "establish the ability to identify cognitive impairment after concussion and distinguish between individuals who are symptomatic and those who are asymptomatic post-injury". This approach shifts away from a group-based approach (e.g., mTBI vs controls) that has dominated the literature to date, focusing on cognitive impairment and symptom-driven approaches, while allowing for a wider range of methodology in future studies. There were several studies identified in this review that demonstrated NCATs consistently found more clinically meaningful differences between symptomatic versus asymptomatic groups, and that asymptomatic individuals often performed like healthy controls \[[@B28],[@B39],[@B52]\]. Similar future research may prove more valuable in elucidating the clinical utility of these tests.
This impairment and symptom-based approach is also consistent with the recommendation of Iverson and Schatz \[[@B16]\] to specifically investigate cognitive impairment rather than mTBI status. They go further and describe new approaches to identifying cognitive impairment, such as taking a base rate approach and categorizing performance based on the total number of low scores across a battery. Determining clinically meaningful definitions of cognitive impairment, and then establishing the NCATs' sensitivity and specificity in classifying individuals with concussion as cognitively impaired, will be key to further establishing the validity, and ultimately, the clinical utility of NCATs, especially with regard to informing return to play and RTD decisions.
Randolph\'s 3rd criterion may be addressed by studies seeking to "determine what cognitive constructs NCATs are measuring, and if those constructs, and the manner in which they are measured, are clinically meaningful". This direction is suggested in light of the evidence that the standard statistical approaches of assessing validity have yielded at best moderate convergent validity, poor discriminant validity and inconsistent evidence that NCAT scores predict traditional NP scores. However, alternative statistical approaches, such as PCA and EFA, have suggested NCATs are measuring similar cognitive constructs, though perhaps in different ways. Specifically, it may be that the names given to NCAT subtests and index scores may not accurately reflect the actual cognitive construct being measured. Therefore, statistically guided comparisons, rather than those guided by nomenclature, could yield better evidence for convergent and discriminant validity.
We also recommend a shift away from 'standard psychometric procedures' since this often relies on comparisons to a gold standard, such as traditional NP tests. However, there is mixed evidence for the utility of traditional NP tests for use in postconcussion assessments, especially outside of the acute injury phase \[[@B24]\]. NCATs are typically presented as potential proxies for traditional NP tests, and as such, validity is often evaluated by direct comparisons between supposedly comparable tests. However, adapting pencil and paper tests to a technological interface can fundamentally change the test. Some have suggested that an NCAT\'s ability to precisely measure RT may be an advantage over traditional NP tests in detecting subtle cognitive declines after concussion \[[@B5]\]. In fact, RT and processing speed scores are often the most robust in studies predicting concussion status or cognitive impairment. Also, several studies have identified alternative scores or interpretative methods that may provide more clinical utility than the standard scores currently provided. For example, RT variability and lack of practice effects may be more sensitive to concussion-related effects \[[@B22],[@B35],[@B58]\]. Thus, the potential technological advantages provided by NCATs warrant closer investigation. A caveat, however, as others have identified sources of error that are introduced into test scores due to the use of technology. This includes a participant\'s familiarity with using a computer \[[@B61]\] to hardware and software configurations \[[@B4],[@B62]\]. The literature is limited in identifying how technology can affect the measurement of performance, and this will be important to clarify in future studies.
There are several other considerations with regard to the manner in which NCATs assess cognitive functioning, and the subsequent impact on clinical utility. First, though comparisons to baseline assessments have routinely been used with NCATs, and can be helpful in the context of cognitive changes in examinees with pre-existing unique cognitive abilities (i.e., upper or lower 20th percentile, ADHD or LD), the use of baseline testing does not appear to be necessary for determining cognitive deficits following concussion \[[@B64],[@B65]\]. Research should focus on the ability for baseline assessments, normative comparisons or some combination of the two to accurately and adequately identify symptomatic individuals. Also, the use of group versus individual settings during test administration should be considered, as there is mixed evidence regarding the potential impact a group versus individual test setting has on test scores \[[@B66],[@B67]\]. The different administration settings could potentially impact the findings as NCATs are often administered in group settings either preseason in athletics or predeployment in the military, and then individually postinjury. Additionally, the environment in which NCATs are often desired to be administered, such as athletic sidelines or combat zones, is an important consideration as much of the research takes place in highly controlled settings. The clinical utility of NCATs in such austere environments warrants further investigation \[[@B68],[@B69]\].
Conclusion & future perspective {#S0022}
===============================
Though the body of literature regarding the validity of the four NCATs discussed in this review has been steadily growing, there appears to be insufficient evidence suggesting that these tools are adequate proxies for traditional NP tests and have limited clinical utility in postconcussion assessments. However, by investigating NCATs with the same methodology used to investigate traditional NP tests, these tests may have been set up for failure. Using the 2005 Randolph criteria, we have provided additional and alternative ways forward for investigating the validity and utility of NCATs that are better suited for the intended clinical use and design of these tests. Future efforts are encouraged to focus on cognitive impairment (e.g., symptomatic vs asymptomatic) rather than group status (e.g., concussed vs controls), the ability to inform return to play and RTD decisions, and utilization of alternative and novel statistical approaches (e.g., RT variability, base rate analyses to identify impairment, etc.). Additional prospective comparisons of multiple NCATs in differing study samples, similar to the one conducted by Nelson *et al*. \[[@B28]\] are also warranted. NCATs have the potential to fundamentally change the nature of care following mTBI. However, until their clinical utility can be further established and clarified, they should be used with caution and at most as screening tools in combination with multifaceted assessments.
###### 'Randolph criteria' for proposed neuropsychological batteries.
**Criterion & description** {#S0023}
---------------------------
- Establishing test--retest reliability over time intervals that are practical for this clinical purpose
- Demonstrating, through a prospective controlled study, that the battery is sensitive in detecting the effects of concussion
- Establishing validity for any novel test battery, through standard psychometric procedures employed to determine which neurocognitive abilities a new NP test is measuring
- Deriving reliable change scores, with a probability-based classification algorithm for deciding that a decline of a certain magnitude is attributable to the effects of concussion, rather than random test variance
- Demonstrating that the proposed battery is capable do detecting cognitive impairment once subjective symptoms have resolved
Note: These are criteria set forth by Randolph *et al*. \[[@B7]\] for both traditional and computerized NP batteries. Randolph *et al*. proposed that NP tests should first meet these criteria prior to their consideration as part of routine standard of care for sport-related concussion.
NP: Neuropsychological.
###### Evidence of validity.
**Content-related** {#S0024}
-------------------
- The relevance of the test items to the construct that is to be measured
- Evaluated by the subjective agreement among subject matter experts, such as neuropsychologists, that the test items are relevant and appropriate for the test purpose
**Construct-related** {#S0025}
---------------------
- The extent to which the measure represents the basic theoretical construct
- Evaluated with correlations, regression or factor analysis between a domain of interest and other well-established, 'gold-standard' measures
**Criterion-related** {#S0026}
---------------------
- The relatedness of the measure to a specified criterion, such as a condition of interest or outcome (e.g., mTBI)
- Evaluated by the group differences, accuracy of diagnosis or identification of a specific condition of interest
It is important to note that there can be some overlap in these different types of validity, as they are not meant to conceptually represent mutually exclusive subcategories of validity, but rather describe the various ways in which validity can manifest \[[@B12]\].
mTBI: Mild traumatic brain injury.
###### Executive summary
**Background** {#S0027}
--------------
- The assessment of cognitive functioning has been identified as an important aspect in the management of concussion.
- Due to logistical advantages over traditional neuropsychological (NP) tests, computerized Neurocognitive Assessment Tools (NCATs) have gained popularity in athletic and military settings.
- However, the psychometric properties, especially validity, and clinical utility of NCATs have yet to be consistently established.
**Neurocognitive assessment tools** {#S0028}
-----------------------------------
- Automated Neuropsychological Assessment Metric is commonly used to assess cognitive functioning in US Military Service Members.
- CNS-Vital Signs is commonly used in psychiatric and neurological clinical trials.
- CogState/Axon/CogSport is commonly used in Australian athletics.
- Immediate Post-Acute Concussion Test (ImPACT) is the most widely used NCAT in US athletics. It has the US FDA approval for postconcussion assessments.
**Existing evidence for validity & clinical utility** {#S0029}
-----------------------------------------------------
- Automated Neuropsychological Assessment Metric related best with traditional NP tests of processing speed, with evidence of moderate sensitivity/specificity for concussion or postconcussive symptoms during the acute injury period.
- CNS-Vital Signs had the least amount of validity-related research, with findings revealing at best moderate correlations with traditional NP tests and no clear evidence for clinically meaningful differences between concussed and controls, though data were from the postacute injury timeframe.
- CogState demonstrated some evidence of validity with several moderate to strong correlations to traditional NP measures and the ability to detect concussion-related cognitive decline during the acute injury period. However, research has had a narrow focus on primarily reaction-based scores and with Australian athletes.
- ImPACT is the best studied of the NCATs, with research indicating mixed results regarding validity. It does appear ImPACT is measuring similar cognitive constructs as traditional NP tests, with some evidence for detecting concussion-related cognitive decline during the acute injury period at levels approaching those desired for clinical decision-making.
**Future perspective** {#S0030}
----------------------
- Additional investigation of the validity and clinical utility of NCATs is warranted, with future efforts encouraged to focus on cognitive impairment (e.g., symptomatic vs asymptomatic) rather than group status (e.g., concussed vs controls), the ability to inform return to play and return to duty decisions and novel statistical approaches (e.g., reaction time variability and base rate analyses to identify impairment).
- Additional prospective comparisons of multiple NCATs in differing study samples, similar to the one conducted by Nelson *et al*. (2016) are also warranted.
**Disclaimer**
The views expressed herein are those of the author(s) and do not reflect the official policy of the Department of the Army, Department of Defense or the US Government.
**Financial & competing interests disclosure**
This material is published by permission of the Defense and Veterans Brain Injury Center, operated by General Dynamics Information Technology for the US Defense Health Agency under Contract No. W91YTZ-13-C-0015. The authors have no other relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript apart from those disclosed.
No writing assistance was utilized in the production of this manuscript.
**Open access**
This work is licensed under the Creative Commons Attribution 4.0 License. To view a copy of this license, visit <http://creativecommons.org/licenses/by/4.0/>
| {
"pile_set_name": "PubMed Central"
} |
INTRODUCTION
============
Despite advances in care, morbidity following traumatic brain injury (TBI) remains high ( [@nlw045-B1] , [@nlw045-B2] ). Following the primary direct mechanical injury to brain parenchyma, important secondary injuries develop over hours to days from multiple addressable mechanisms including hypoxemia, ischemia, and excitotoxicity ( [@nlw045-B3] ). Clinical observational studies have found a strong association between early hypoxemia (pre-hospital or emergency room setting) following TBI and poor clinical outcomes ( [@nlw045-B4] ). Preclinical investigations of hypoxemia immediately after experimental TBI have demonstrated exacerbations in brain edema and ischemia, hippocampal neuronal cell death, neuroinflammation, axonal injury, and behavioral deficits ( [@nlw045-B8] ).
Patients with TBI in the intensive care setting are also at high risk for delayed secondary hypoxemia due to pulmonary contusions, aspiration, pneumonia, atelectasis, acute respiratory distress syndrome, and as a result of procedural interventions such as endotracheal tube displacement, emergent diagnostic imaging, and surgical interventions. However, little is known about the role that delayed hypoxemia (beyond the pre-hospital and emergency room setting) plays in secondary brain injury following TBI. A small prospective study found that any hypoxemia, defined as O ~2~ saturation \< 90%, within the first 24 hours after TBI was associated with an increase in mortality ( [@nlw045-B13] ). The majority of previous animal studies investigating post-TBI hypoxemia have utilized a 15- to 30-minute exposure of low inspired oxygen (8%--12%) occurring within 1 hour of experimental TBI ( [@nlw045-B8] ). No studies to our knowledge have investigated delayed hypoxemia after TBI in an animal model.
Although traumatic axonal injury (TAI) is considered to be the major contributor to long-term cognitive deficits in TBI patients ( [@nlw045-B14] ), only 1 preclinical study has investigated the contribution of immediate postinjury hypoxemia to axonal injury ( [@nlw045-B11] ). Axonal injury following TBI is most commonly characterized histologically by immunohistochemical staining of amyloid precursor protein (APP) and neurofilament. APP under normal conditions traverses the length of the axon, but in response to injury will accumulate at axonal varicosities and retraction bulbs ( [@nlw045-B19] ). Changes in structural integrity of the axonal cytoskeleton can result in increased neurofilament immunoreactivity ( [@nlw045-B16] , [@nlw045-B23] , [@nlw045-B24] ). These 2 approaches, when combined, can provide a more complete histological characterization of axonal injury following TBI ( [@nlw045-B19] , [@nlw045-B25] , [@nlw045-B26] ). To our knowledge, the impact of delayed hypoxemia on axonal injury has not been investigated in preclinical models of TBI.
Based on retrospective pediatric intensive care data, we developed a clinically relevant mouse model of delayed systemic hypoxemia following TBI. We used this model to test the hypothesis that delayed posttraumatic hypoxemia following controlled cortical impact (CCI) in young mice would exacerbate white matter injury.
MATERIALS AND METHODS
=====================
Patient Data
------------
The Institutional Review Board at Washington University approved the retrospective clinical study and waived the requirement for obtaining written informed consent. Data were abstracted from registry of pediatric patients with severe TBI (postresuscitation Glasgow Coma Score (GCS) of 8 or less who were admitted to the pediatric intensive care unit (PICU) at Saint Louis Children's Hospital between July 1, 1999 and December 31, 2011. Patients were excluded if they had a GCS of 3 with fixed and dilated pupils at time of admission, cardiac arrest prior to arrival in the PICU, abusive head trauma, and gunshot wounds to the head. Age at time of injury, sex, postresuscitation GCS, pediatric risk of mortality score (PRISM III), injury severity score, injury mechanism, PICU length of stay, hospital length of stay, PICU-free days at 28 days of admission, Glasgow outcome scale score at hospital discharge, and destination at hospital discharge were recorded. Initial head computed tomography scan was categorized by using the Marshall classification of initial head computed tomography ( [@nlw045-B27] ). PICU-free days at 28 days of admission was defined as the total numbers of days alive outside the PICU at 28 days after admission. Complete arterial blood gas, hourly pulse oximetry, and end tidal CO ~2~ data were abstracted from the medical record. Hypoxemia was defined as a pulse oximetry reading \< 90% or an arterial PO ~2~ \<60 mm Hg. A discreet episode of hypoxemia was defined as complete with the first subsequent normal blood gas or pulse oximetry reading.
Animal Studies
--------------
### Injury
All procedures were approved by the Washington University Animal Studies Committee, and are consistent with the National Institutes of Health guidelines for the care and use of animals. Five week-old C57BL/6J male mice (Jackson Laboratory, Bar Harbor, ME) weighing 16--20 g were used in these experiments. For studies of axonal injury, mice were killed at 2 different time points: 48 hours and 1 week after injury (n = 20 total for each time point). Two additional cohorts of 20 animals each were utilized for the same 2 time points (48 hours and 1 week after injury) to evaluate the response of microglia (1 week time point only) and astrocytes (48 hours and 1 week). For each time point, 8 mice underwent sham injury and 12 mice underwent CCI ( [@nlw045-B28] ). The mice were anesthetized with 5% isoflurane at induction, followed by maintenance at 2% isoflurane for the duration of the procedure. The head was shaved and head holders were used to stabilize the head within the stereotaxic frame (MyNeurolab, St. Louis, MO). Then, a single 5-mm craniotomy was performed by an electric drill on the left lateral side of the skull centered 2.7 mm lateral from the midline and 3 mm anterior to lambda. Animals were randomized to sham or injury after craniotomy utilizing a computer generated numbers randomization. For injured animals, the 3-mm electromagnetic impactor tip was then aligned with the craniotomy site at 1.2 mm left of midline, 1.5 mm anterior to the lambda suture. The impact was then delivered at 2 mm depth (velocity 5 m/s, dwell time 100 ms). The head holders were released immediately after the injury. All animals then received a loose fitting plastic cap secured over the craniotomy with Vetbond (3M, St. Paul, Minnesota). The skin was closed with interrupted sutures and treated with antibiotic ointment before removing the mouse from anesthesia and allowing recovery on a warming pad.
Delayed Hypoxemia
-----------------
One day after injury or sham surgery, animals were randomized to normoxemia (room air) or hypoxemia (8% O ~2~ , 4% CO ~2~ ) for 30 minutes. During normoxemia or hypoxemia animals were placed in fresh cages with littermates randomized to the same treatment arm and all mice were subjected to identical transport and handling. Animals randomized to hypoxemia were placed in a Coy Labs Hypoxia Chamber (6′ × 3′ × 4′) (Coy Laboratory, Grass Lake, MI), fitted with airlock, oxygen sensor, carbon dioxide sensor, and gas controllers.
Arterial Blood Gas Sampling
---------------------------
A separate cohort of 18 mice was utilized for the arterial blood gas experiments. Animals were randomized to injury or sham surgery (n = 9 for each group). One day after injury or sham surgery, animals were anesthetized with 5% isoflurane at induction, followed by maintenance at 2% isoflurane for placement of right carotid artery tunneled catheters. Once completely recovered and awake, animals underwent 30 minutes of low inspired oxygen (8% O ~2~ ,) and arterial blood samples were drawn at the end of the 30 minutes. Arterial blood gas analysis was performed on a Bayer Rapidlab 845 blood gas analyzer (Siemens Medical Diagnostics, Bayer, Tarrytown, NY). Initial screening was performed on 1 sham and 1 injured mouse at each of the following CO2 concentrations: 0.5%, 2%, 4%, and 6%. The conditions that produced normocarbia were then repeated with the 10 remaining mice (n = 5 for each group).
Immunohistochemistry
--------------------
Mice were killed under isoflurane anesthesia by transcardial perfusion with 0.3% heparin in phosphate-buffered saline. Whole brains were removed and fixed in 4% paraformaldehyde for 48 hours, followed by equilibration in 30% sucrose for at least 48 hours prior to sectioning. Serial 50-μm-thick coronal slices were cut on a freezing microtome starting with the appearance of a complete corpus callosum and caudally to bregma −3.08 mm. Sets of 12 sections spaced every 300 μm were mounted on glass slides and used for immunohistochemical studies.
Staining was performed on free-floating sections washed in Tris-buffered saline (TBS) between applications of primary and secondary antibodies. Endogenous peroxidase was blocked by incubating the tissue in TBS + 3% hydrogen peroxide for 10 minutes. Normal goat serum (3%) in TBS with 0.25% Triton X (Tris-buffered saline (TBS)-X) was used to block nonspecific staining for all antibodies. Slices were then incubated at 4 °C overnight with one of the following primary antibodies: polyclonal rabbit anti-β-APP (Invitrogen, Carlsbad, CA) at a concentration of 1:1000; polyclonal rabbit anti-neurofilament-200 (Sigma, St. Louis, MO) at a concentration of 1:8000; polyclonal rabbit anti-Iba1 (Wako Chemicals USA, Richmond, VA) at a concentration of 1:1000; or polyclonal rabbit anti-glial fibrillary acidic protein (GFAP) (Dako North America, Carpinteria, CA) at a concentration of 1:1000. Biotinylated goat anti-rabbit secondary antibodies (Vector Laboratories, Burlingame, CA) in TBS-X were used at a 1:1000 concentration to detect bound primary antibodies. Colorization was achieved using the Vectastain ABC Elite Kit (Vector Laboratories) followed by the application of 3-3′ diaminobenzidine (DAB).
Detection of Brain Tissue Hypoxia
---------------------------------
To determine if brain tissue hypoxia was occurring during delayed hypoxemia in regions where TAI was occurring, we utilized Hypoxyprobe ^TM^ -omni kit (Hypoxyprobe, Burlington, Massachusetts) (pimonidazole) to quantify white matter tissue hypoxia ( [@nlw045-B31] ). Pimonidazole selectively adducts with thiol groups in proteins, peptides, and amino acids under hypoxic conditions (PO ~2~ \<10 mm Hg) ( [@nlw045-B34] ). Twenty mice (CCI, n = 12, sham, n = 8) were administered 60 mg/kg of pimonidazole intraperitoneally and then immediately randomized to either 30 minutes of hypoxemia or room air. Seventy-five minutes after injection of pimonidazole, mice were killed and brains were prepared for immunohistochemistry as described above. Affinity purified anti-pimonidazole rabbit antiserum (PAb2627AP, Hypoxyprobe) at a concentration of 1:1000 was utilized as the primary antibody.
Quantification of Immunohistochemistry
--------------------------------------
The extent of tissue loss of the ipsilateral hemisphere for each animal was quantified using images of β-APP stained slices acquired using Hamamatsu NanoZoomer 2.0-HT System (Hamamatsu Corporation, Middlesex, New Jersey). Tissue loss in the injured hemisphere was calculated as a percentage of the tissue volume in the contralateral hemisphere as described by others ( [@nlw045-B35] ). Stereological analysis was performed using StereoInvestigator software version 8.2 (MBF Bioscience, Williston, Vermont). Assessments were made by an investigator blinded to group assignment. The optical fractionator function was used to quantify target markers per cubic millimeter of tissue. For quantification of axonal injury, a grid size of 250 × 250 µm, a counting frame of 40 × 40 µm, and a dissector height of 15 µm with a guard zone of 5 µm were used for all quantifications, resulting in 3% of the regions of interest (ROI) being randomly sampled. All ROIs were traced at 4× magnification and markers were counted at 60× magnification. The ipsilateral corpus callosum and external capsule spanning 12 sections starting with the appearance of a complete corpus callosum and caudally to bregma −3.08 mm were used as the ROI for the β-APP and neurofilament-200 (NF200). This region was defined as the white matter area between midline and the lateral edge of the cingulum in rostral sections; in caudal sections, a horizontal line drawn laterally from the end of the fimbria served as the end boundary of the ROI. Injured axons were identified by β-APP-positive varicosities greater than 5 µm. Similarly, NF200-positive axonal varicosities greater than 5 µm in diameter were counted as injured axons during stereological assessment. For stereological quantification of Iba1-positive cells, the optical fractionator function was again used, with a grid size of 180 × 180 µm and a counting frame of 80 × 80 µm. The ROI began with the most anterior slice containing hippocampal dentate gyrus and ended with the most posterior section containing corpus callosum fibers that cross midline which yielded 3--4 sections for analysis per animal. The midline served as the medial boundary for the ROI, whereas the lateral boundary was formed by drawing a horizontal line between the ventral hippocampus and dorsal thalamus in each coronal section. Gunderson's coefficients of error were \<0.1 for all stereological quantifications.
Automated quantification of axonal injury was performed with Visiopharm software version 4.3.1.0 (Visiopharm, Broomfield, Colorado) on images acquired using Hamamatsu NanoZoomer 2.0-HT System. Using the same ROI described above, 3 labels were generated for β-APP stained slices: 2 labels for positively stained axons (APP puncta, for cross sectional and APP axons for longitudinal), and 1 label for background white matter. The classification method used was threshold, and the ranges for the 2 stained axon labels were 0--7, and 7.01--190 for background white matter. The post process parameters were as follows, in order: APP puncta with an area \<12 µm ^2^ and APP axons with an area \<30 µm ^2^ were transformed to background tissue, APP puncta were separated into discreet objects, APP puncta with an area greater than 50 µm ^2^ were transformed into APP axons, APP axons were separated into discreet objects, APP axons label with an area greater than 120 µm ^2^ were transformed to background tissue, and APP puncta labeled areas with an irregularity greater than 3 were transformed to background tissue. For NF200 staining, the classification method used was threshold, and the ranges for NF-200 puncta, NF200 axons, and tissue were 0--30, 0--30, and 30.01--200, respectively. The post processes applied were as follows, in order: NF200 puncta were separated into discreet objects, NF200 puncta with an area \<15 µm ^2^ were transformed into background tissue, NF200 puncta with an irregularity greater than 4 were transformed into NF200 axons, NF200 axons with an area \<25 µm ^2^ were transformed into tissue, NF200 axons were then separated into discrete objects, and NF-200 axons with an area greater than 200 µm ^2^ were transformed into background tissue.
Quantification of brain tissue hypoxia staining was performed in the same ROI for axonal injury. Two labels, 1 for hypoxic cells and 1 for background white matter tissue were utilized. Threshold was again used as the classification method, and the ranges for hypoxic cells and background tissue were 0--90 and 90.01--210, respectively. The post processes were as follows: Separate objects for hypoxic cells and change hypoxic cells with an area of \<6 µm ^2^ to background tissue. Test--retest reliability was evaluated by staining a second set of 12 slices from each brain analyzed and having the same blinded operator repeat the image analysis on these stained slices. Interrater reliability was assessed by having a second blinded operator perform the image analysis on the same set of images.
For GFAP measurements, 5× images of the ipsilateral corpus callosum and external capsule were obtained using the Nanozoomer. Using ImageJ, files were converted to 32-bit images and positive signal on each image was selected using the Max Entropy auto thresholding method. ROIs were selected and drawn as described for stereological assessment of microglia in ipsilateral white matter. Total area of ROIs and area of positive thresholding from each slice were calculated and summed to determine total percent area of positive GFAP staining.
All postprocessing and automated quantification algorithms are freely available upon request.
Statistical Analysis
--------------------
All data were analyzed using Statistica version 12 (Dell Inc., Tulsa, OK). Clinical continuous data is presented as median and interquartile ranges. The association of variables and hypoxemia was performed using Kruskal--Wallis for continuous data and Fisher exact test for categorical data. Animal data results are presented as mean ± SD. For all animal data sets there was no evidence for significant deviations from normal distribution (p \> 0.05 by Shapiro--Wilk tests). Quantitative histologic data were analyzed with 2-way ANOVA, followed by Tukey tests for multiple comparisons with a significance level of p \< 0.05. Spearman correlations were utilized to assess the correlation between stereological and Visiopharm-based quantification of axonal injury.
RESULTS
=======
Delayed Hypoxemia After Severe TBI Is Common
--------------------------------------------
Seventy-four patients met study inclusion criteria and had complete arterial blood gas and hourly pulse oximetry data available. Twenty-eight patients (35%) were found to have experienced hypoxemia (PaO ~2~ \<60 mm Hg or pulse oximetry \<90%) within in the first 7 days of admission. The length of all hypoxemic episodes was observed to be no \>1 hour, but their exact length could not be determined due to the resolution of the physiologic data in the medical record (hourly recordings). The majority of these episodes occurred in the first 48 hours ( [Fig. 1](#nlw045-F1){ref-type="fig"} ). All patients were observed to be normocarbic during episodes of hypoxemia. Patients with hypoxemia were more commonly female; otherwise there were no differences in demographic data or injury severity between the groups ( [Table 1](#nlw045-T1){ref-type="table"} ). Patients who experienced hypoxemia were more likely to have a longer PICU stay. Patients with delayed hypoxemia trended toward poorer outcome (GOS of 3 or less) at hospital discharge (54% vs 29%) but this did not reach statistical significance. However, when GOS at hospital discharge was dichotomized, patients with delayed hypoxemia had poorer outcome (54% vs 29%, p = 0.047).
{#nlw045-F1}
######
Demographic Characteristics, Severity of Injury, Injury Mechanisms, and Raw Outcomes
Hypoxemia Normoxemia p Value
---------------------------------------- --------------------- --------------------- ---------
Age (months) 121 \[55--189\] 144 \[109--179\] 0.31
Sex, male 11 (42%) 28 (58%) 0.14
Post resuscitation GCS 4 \[3--5\] 5 \[3--6\] 0.25
PRISM III 6 \[2--14\] 5 \[2--8\] 0.26
ISS 28 \[16 --41\] 28 \[21 --35\] 0.61
Injury mechanism 0.088
Motor vehicle accident 18 (69%) 34 (71%)
Pedestrian accident 4 (15%) 9 (19%)
Fall 1 (4%) 5 (10%)
Other 3 (12%) 0 (0%)
Marshall CT classification 0.33
Diffuse Injury I 0 1 (2%)
Diffuse Injury II 24 (92%) 38 (79%)
Diffuse Injury III 1 (4%) 6 (13%)
Diffuse Injury IV 0 1 (2%)
Evacuated mass legion 0 2 (4%)
Nonevacuated mass legion 1 (4%) 0
PICU LOS 15.1 \[12.2--23.0\] 12.3 \[5.0--15.5\] 0.014
PICU free days at 28 days 11.6 \[4.0--15.8\] 15.1 \[11.8--22.7\] 0.017
Hospital LOS 45.0 \[15.0--89.0\] 26.5 \[16.5--37.5\] 0.19
GOS at hospital discharge 0.12
1 death 1 (4%) 2 (4%)
2 persistent vegetative state 0 (0%) 0 (0%)
3 severe disability 13 (50%) 12 (25%)
4 moderate disability 11 (42%) 26 (54%)
5 good recovery 1 (4%) 8 (17%)
Dichotomized GOS at hospital discharge 0.047
Poor (GOS 1--3) 14 (54%) 14 (29%)
Good (GOS 4--5) 12 (46%) 34 (71%)
Destination 0.87
Home, no assistance 18 (69%) 30 (63%)
Rehab facility 7 (27%) 15 (31%)
Acute care hospital 0 (0%) 1 (2%)
Death 1 (4%) 2 (4%)
Data are presented as median and interquartile range.
GCS, Glasgow coma score; PRISM III, pediatric risk of mortality score; ISS, injury severity score; LOS, length of stay; GOS, Glasgow outcome scale score; PICU, pediatric intensive care unit, PRISM, pediatric risk of mortality score.
Mouse Model of Clinical Relevant Delayed Hypoxemia After TBI
------------------------------------------------------------
To model delayed normocarbic hypoxemia following TBI explicitly, we exposed awake, spontaneously breathing mice 1 day after injury or sham surgery to 8% oxygen. As part of our experimental design, we avoided the use of inhaled anesthetics during delayed hypoxemia because they are not routinely utilized in the intensive care setting and may in fact be neuroprotective ( [@nlw045-B36] ). However, allowing mice to breathe spontaneously in a low oxygen environment results in tachypnea and hypocarbia, which produce decreases in cerebral blood flow, and would cloud the clinical relevance of our model. Utilizing mice that had undergone tunneled arterial cannulation, ambient carbon dioxide concentration was increased stepwise from 0.5% to 6%. We found that 4% carbon dioxide maintained normocarbia (despite hypoxemia-related hyperventilation) in awake, spontaneously breathing mice during 30 minutes of 8% oxygen ( [Table 2](#nlw045-T2){ref-type="table"} ).
######
Arterial Blood Gas Samples From Awake Spontaneously Breathing Mice After 30 Minutes of 8% O ~2~ and 4% CO ~2~ 24 Hours After Controlled Cortical Impact or Sham Surgery
Group pH PaCO ~2~ (mm Hg) PaO ~2~ (mm Hg)
-------------- ------------- ------------------ -----------------
Sham (n = 5) 7.36 ± 0.02 40.8 ± 2.0 50.1 ± 1.4
CCI (n = 5) 7.35 ± 0.03 40.6 ± 1.9 49.8 ± 1.1
CCI, controlled cortical impact injury.
Pericontusional White Matter Hypoxia Is Worsened by Delayed Hypoxemia
---------------------------------------------------------------------
As expected, hypoxemia 24 hours after CCI resulted in increased white matter immunohistochemical staining in the ipsilateral corpus callosum and external capsule for a marker of tissue hypoxia compared with CCI alone ( [Fig. 2](#nlw045-F2){ref-type="fig"} ). Minimal white matter tissue hypoxia was observed in sham or sham + hypoxemia. Quantification of white matter tissue hypoxia with Visiopharm software was performed in the ipsilateral corpus callosum and external capsule with strong interrater and test--retest reliability ( [Fig. 3](#nlw045-F3){ref-type="fig"} ). Two-way ANOVA revealed a significant injury (F ~1,16 ~ =561.8, p \< 0.0001), hypoxemia (F ~1,16~ =26.8 p \< 0.0001), and injury + hypoxemia interaction effect (F ~1,16 ~ =17.5, p \< 0.0007). *Posthoc* Tukey tests demonstrated significantly more extensive tissue hypoxia in injured mice with hypoxemia compared with injury alone (p \< 0.001) ( [Fig. 2Q](#nlw045-F2){ref-type="fig"} ). Interestingly, there was no difference in Hypoxyprobe staining in the white matter when comparing sham with sham + hypoxemia. Thus, this moderate degree of hypoxemia does not appear to cause white matter brain tissue hypoxia in isolation but substantially worsens tissue hypoxia in peri-contusional white matter.
{#nlw045-F2}
{#nlw045-F3}
Delayed Hypoxemia Following TBI Is Associated With an Increase in Peri-Contusional White Matter TAI
---------------------------------------------------------------------------------------------------
Following CCI in rodent models, extensive TAI has been reported in the peri-contusional white matter ( [@nlw045-B40] , [@nlw045-B41] ). To determine the effects of delayed hypoxemia after TBI on TAI, we assessed TAI in the ipsilateral corpus callosum and external capsule using 2 different markers: β-APP and NF200. As previously reported, we observed increased β-APP-positive varicosities in the peri-contusional corpus callosum and external capsule of injured mice ( [@nlw045-B25] , [@nlw045-B41] ). We also noted a reduction in immunohistochemical staining for both markers over time. Injured mice that experienced hypoxemia 1 day after CCI appeared to have increased β-APP-positive staining compared with injured mice without hypoxemia ( [Fig. 4](#nlw045-F4){ref-type="fig"} ). Hypoxemia did not affect nonspecific staining characteristics of the tissue ( [Fig. 4](#nlw045-F4){ref-type="fig"} ). We did not observe β-APP stained axonal swellings in the contralateral white matter of any animals at either time point ( [Fig. 5](#nlw045-F5){ref-type="fig"} ). Additionally, we observed a small but statistically significant increase in tissue loss in mice that experienced delayed hypoxemia following CCI compared with CCI alone ( [Fig. 6](#nlw045-F6){ref-type="fig"} ).
{#nlw045-F4}
{#nlw045-F5}
{#nlw045-F6}
Our qualitative observations were confirmed by blinded stereological quantification of β-APP staining in the ipsilateral corpus callosum and external capsule ( [Fig. 4M, N](#nlw045-F4){ref-type="fig"} ). At 2 days postinjury, 2-way ANOVA revealed a significant injury (F ~1,16~ =279.7, p \< 0.0001), hypoxemia (F ~1,16~ =22.5 p \<0.0002), and injury + hypoxemia effect (F ~1,16~ =22.6, p \<0.0002). *Posthoc* Tukey tests demonstrated a significantly higher number of β-APP positive varicosities in injured mice with delayed hypoxemia compared with injured mice without hypoxemia (p \< 0.001). Two-way ANOVA at 7 days postinjury revealed similar results with significant injury (F ~1,16~ =104,5, p \< 0.001), hypoxemia (F ~1,16~ =19.1, p \< 0.005), and injury + hypoxemia effect (F ~1,16~ =19.3, p \< 0.005). *Posthoc* Tukey tests also demonstrated a significantly higher number of β-APP-positive varicosities in injured mice with hypoxemia compared with injured mice without hypoxemia at 7 days postinjury (p \< 0.001).
Immunohistochemical staining for NF200 was observed in injured mice in a more rostral aspect of the ipsilateral corpus callosum and external capsule. There was an increase in NF200 axonal swellings noted in injured mice with hypoxemia compared with injured mice without hypoxemia ( [Fig. 7](#nlw045-F7){ref-type="fig"} ). Two-way ANOVA of stereological quantification of NF200-positive axonal swellings in the ipsilateral corpus callosum and external capsule at 2 days postinjury revealed a significant injury (F ~1,16~ =481.3, p \< 0.0001), hypoxemia (F ~1,16~ =16.5, p \< 0.0009), and injury + hypoxemia interaction effect (F ~1,16~ =19.6, p \< 0.0004). *Posthoc* Tukey tests demonstrated a significantly higher number of NF200-positive swellings in injured mice with hypoxemia compared with injured mice alone (p \< 0.001). Similarly, 2-way ANOVA at 7 days postinjury revealed a significant injury (F ~1,16~ =199.4, p \< 0.0001), hypoxemia (F ~1,16~ =26.5, p \< 0.0001), and injury + hypoxemia effect (F ~1,16~ =24.9, p \< 0.0001). *Posthoc* Tukey tests also demonstrated a significantly higher number of NF200 varicosities in injured mice with hypoxemia compared with injured mice alone (p \< 0.0001).
{#nlw045-F7}
Automated Quantification of Axonal Injury
-----------------------------------------
Quantification of white matter injury immunohistochemistry with stereology is a powerful and accurate method to assess treatment effects on axonal injury. However, stereological quantification of axonal injury is laborious, expensive, and temporally limits experimental throughput (eg, for screening of candidate therapeutics for efficacy in axonal injury, assessments of effects of genotype, gender, age, etc.). To improve experimental efficiency, we developed a higher throughput image analysis protocol using the Visiopharm Integrator System software for quantification of axonal injury. Axonal swellings are identified by first thresholding the ROI, and then further distinguished by user developed parameters based on size, shape, and object separation ( [Fig. 8](#nlw045-F8){ref-type="fig"} ). Using these parameters, 2 populations of positively stained axonal swellings can be identified: short axis (green) and long axis (red) ( [Fig. 8B, E](#nlw045-F8){ref-type="fig"} ). The sum of these 2 populations identified by this method correlated very well with traditional stereological quantification for both β-APP and NF200 stained axons ( [Fig. 8C, F](#nlw045-F8){ref-type="fig"} ). Throughput was approximately 7-fold faster than traditional stereology.
{#nlw045-F8}
Delayed Hypoxemia Following CCI Does Not Significantly Alter Microglial Response
--------------------------------------------------------------------------------
To determine if increase microglial activation was associated with the increase in axonal injury following delayed hypoxemia, we examined the peri-contusional white matter with Iba1 immunohistochemical staining at 7 days postinjury. We selected the 7 day time point as it has been previously reported that microglial activation increases over the first 7 days after CCI in the mouse ( [@nlw045-B42] ). Following CCI, we observed increased prominence of Iba1-positive cells in the ipsilateral corpus callosum and external capsule 7 days after initial injury ( [Fig. 9](#nlw045-F9){ref-type="fig"} ). This injury effect was confirmed by stereological analysis ( [Fig. 9I](#nlw045-F9){ref-type="fig"} ) (F ~1,16~ =217, p \< 0.0001). Stereological quantification of Iba1-positive cells did not demonstrate a statistically significant effect of delayed hypoxemia (F ~1,16~ =1.51, p = 0.24) or the interaction between hypoxemia and injury (F ~1,16~ =1.94, p = 0.18). No difference in microglial activation was observed in sham animals with or without hypoxemia.
{#nlw045-F9}
Delayed Hypoxemia Augments Astrocyte Response Following CCI
-----------------------------------------------------------
To assess effects of delayed hypoxemia on astrocytic response to TBI, we performed GFAP immunohistochemical staining. As expected, we observed increased GFAP-positive cells in the ipsilateral corpus callosum and external capsule in injured animals compared with shams at 48 hours and 1 week after CCI ( [Fig. 10](#nlw045-F10){ref-type="fig"} ). Hypoxemia alone did not increase GFAP staining in shams at 48 hours or 1 week postinjury. However, delayed hypoxemia after CCI did increase GFAP staining in the ipsilateral corpus callosum and external capsule compared with CCI alone at 48 hours and 1 week after injury ( [Fig. 10](#nlw045-F10){ref-type="fig"} ). Two-way ANOVA of percent area of GFAP staining in the ipsilateral corpus callosum and external capsule demonstrated a significant injury (F ~1,16~ =359.6, p \< 0.0001), hypoxemia (F ~1,16~ =8.7, p \< 0.01), and injury + hypoxemia interaction effects (F ~1,16~ =4.8, p \< 0.05) at 2 days postinjury. *Posthoc* Tukey test confirmed a higher percent area of GFAP staining in mice with delayed hypoxemia and CCI compared with CCI alone (23% vs 18%, p \< 0.001). Similarly at 1 week postinjury, there were significant injury (F ~1,16~ =181.2, p \< 0.0001), hypoxemia (F ~1,16~ =12.7, p \< 0.003), and injury + hypoxemia interaction effects (F ~1,16~ =12.1, p \< 0.003). *Posthoc* Tukey test demonstrated a higher percent area of GFAP staining with delayed hypoxemia compared with CCI alone (45% vs 24%, p \< 0.01).
{#nlw045-F10}
DISCUSSION
==========
In summary, hypoxemia was common in children with severe head injury during the first 7 days of intensive care. To model this clinical phenomenon, we exposed mice to 30 minutes of systemic hypoxemia 24 hours after moderate CCI and observed increased peri-contusional axonal injury. We detected increased axonal injury in mice experiencing hypoxemia after TBI utilizing 2 different immunohistochemical markers of axonal injury, β-APP and NF200, at 48 hours and 1 week postinjury in the ipsilateral corpus callosum and external capsule. Our findings are further supported by increased white matter hypoxia and astrogliosis in mice undergoing CCI followed by delayed systemic hypoxemia compared with CCI alone. Together, these data support the hypothesis that delayed hypoxemia following CCI in mice exacerbates axonal injury in peri-contusional white matter.
Our model has several strengths. The electromagnetic CCI device utilized produces precise reproducible injury severity reducing injury variability and the sample size of mice required ( [@nlw045-B28] ). It should be noted, however, that modeling human severe TBI in the rodent is technically challenging. Unlike in the clinical setting, the animals in this study were able to feed and care for themselves without medical intervention within 1 hour of injury. We utilized retrospective clinical data as a foundation for our model design of hypoxemia to ensure clinical relevance and maximize the translatability of our findings. As part of our experimental design, we avoided the use of inhaled anesthetics during delayed hypoxemia because they are not routinely utilized in the intensive care setting and may in fact be neuroprotective, thereby confounding results ( [@nlw045-B36] ). By adding CO ~2~ in a controlled manner to the hypoxia chamber, we were able to achieve normocarbia, permitting us to examine the effects of hypoxemia without the confounding influence of hypocarbia. Because children with TBI and hypoxemia were normocarbic, this approach also supported fidelity between clinical data and our preclinical model.
We assessed white matter axonal injury utilizing 2 different immunohistochemical markers of axonal injury, β-APP a marker of impaired axoplasmic transport (IAT) and NF200, a marker of neurofilament compaction (NFC). IAT and NFC are thought to occur in different subset of axon populations following TBI ( [@nlw045-B43] ). IAT detected by anti-β-APP can be found in both larger and smaller axons whereas NFC is typically detected by immunohistochemistry in larger axons ( [@nlw045-B19] ). Both markers of axonal injury had greater densities in the ipsilateral corpus callosum and external capsule at 48 hours and 1 week after CCI in animals that had experienced delayed hypoxemia versus normoxemia, thereby demonstrating that a delayed systemic insult can exacerbate TAI. We also observed variations in IAT and NFC density based on location. The majority of NFC detection in our CCI model was observed in white matter rostral to the impact epicenter with little or no NF200 staining at the center of the impact. β-APP immunostaining was more diffuse with the greatest amount observed in white matter at the impact epicenter.
Our innovative approach to axonal injury quantification with a high throughput image analysis protocol using the Visiopharm software correlated well with traditional stereological quantification ( [Fig. 8C, F](#nlw045-F8){ref-type="fig"} ), and had excellent interrater and test--retest reliability ( [Fig. 3C, D](#nlw045-F3){ref-type="fig"} ). However, unlike stereology, our method does not estimate density of axonal pathology through the entire thickness of the ROI in each slice. Our image analysis is also highly dependent on high quality and consistent immunohistochemistry, DAB colorization, and imaging to achieve accurate quantification. Avenues for future development could include alterations in the histological approach, such as thinner sections spaced closer together. Nevertheless, this quantification technique with the stated caveats provides the opportunity for fast and accurate unbiased quantification of histopathology. This technique is very attractive as an initial screening tool for future investigations assessing the efficacy of therapeutic candidates targeting axonal injury.
We also assessed the microglial response in the ipsilateral corpus callosum and external capsule following delayed hypoxemia after CCI using Iba-1 immunohistochemistry and observed a strong microglial response 1 week after CCI ( [Fig. 9](#nlw045-F9){ref-type="fig"} ), consistent with previous reports ( [@nlw045-B42] , [@nlw045-B46] ). By contrast, the addition of delayed hypoxemia did not result in a significant increase in the microglial response in the ipsilateral corpus callosum and external capsule. It should be noted that our evaluation of the microglia response was limited to Iba1 immunohistochemical staining and we did not investigate any influence of delayed hypoxemia on activations of M1 and M2 phenotypes. Previous investigations in a weight-drop rat model of TBI with immediate postinjury hypoxemia did not demonstrate an increase in microglial activation compared with TBI alone ( [@nlw045-B10] ). Furthermore, when microglia are reduced prior to and after injury in a murine repetitive concussion model, no alterations in axonal injury were observed ( [@nlw045-B47] ). These data are most consistent with the hypothesis that microglia do not play a prominent contributory role to the increase axonal injury observed after delayed hypoxemia following TBI. Conversely, GFAP immunohistochemistry demonstrated more prominent astrogliosis following delayed hypoxemia after CCI versus CCI alone at 48 hours postinjury with further increases observed at the 1 week postinjury interval. Astrocyte proliferation after CCI has been reported to peak at 3--4 days postinjury ( [@nlw045-B46] , [@nlw045-B48] , [@nlw045-B49] ). We observed similar levels of reactive astrocytes at 2 days and 1 week postinjury in mice experiencing CCI alone but the addition of delayed hypoxemia resulted in a marked increase in reactive astrocytes at 1 week post-CCI. It is unclear if this astrocyte response is simply a marker of the increased axonal injury observed following delayed hypoxemia, plays a protective role, or contributes in a negative fashion to injury.
There are many other logical directions for future research on this topic. Our investigations into the effects of delayed hypoxemia after TBI were limited to a focal injury model. It would be of interest to see if similar exacerbations of axonal injury could be observed in a more diffuse mouse model of TBI. We have demonstrated axonal susceptibility to systemic hypoxemia 24 hours after primary TBI but in these experiments we did not determine the limits of this temporal window of vulnerability. Future experiments are planned to investigate how long after the initial TBI can axonal injury be exacerbated by systemic hypoxemia. Immediate systemic hypoxemia following experimental TBI has been observed to exacerbate hippocampal neuronal injury and we plan to investigate the effects of delayed hypoxemia on hippocampal injury following CCI. We did not observe any immunohistochemical evidence of TAI in the contralateral white matter with the markers described above. However, prominent silver staining contralateral to the impact site following CCI has been observed by others ( [@nlw045-B40] , [@nlw045-B50] ). Further investigations evaluating the effects of delayed hypoxemia on white matter regions remote to the impact site are planned. Our evaluation of TAI in white matter also did not include investigations into the effects of delayed hypoxemia on myelination of axons or changes in oligodendrocytes or their progenitor cells. Future investigations are planned including the use of electron microscopy to assess changes in axon myelination and Wallerian degeneration. Additional future directions include investigating the effect of age on the brain's response to delayed hypoxemia. We utilized 5-week-old male mice, the developmental equivalent of a school age/adolescent patient to model the median age of our clinical retrospective data. It is not known whether younger children would be more or less vulnerable to delayed hypoxemia following severe TBI. Likewise, it is not known whether female mice would be affected similarly by delayed hypoxemia, though males and females are affected similarly by CCI in isolation ( [@nlw045-B51] ). Future studies are planned to assess the effects of delayed hypoxemia on long-term behavioral deficits and neuropathology such as hippocampal neurons, white matter atrophy, and microglial activation.
In conclusion, clinically relevant hypoxemia 24 hours after CCI in mice resulted in white matter tissue hypoxia and increased axonal injury as determined by 2 different immunohistochemical markers. Targeting delayed hypoxemia after TBI in the ICU setting provides a unique opportunity for therapeutic interventions, permitting rapid administration of neuroprotective agents with short temporal windows of efficacy and the prospect to administer exceptionally safe neuroprotective agents prior to episodes of secondary brain hypoxia in high risk patients with severe TBI.
The authors thank Ronaldo Perez for technical assistance, T.J. Esparza for methodological assistance, and Tina Day for assistance with the clinical data set.
[^1]: This work was supported by the Hope Center Alafi Neuroimaging Lab and NIH Shared Instrumentation Grant award to Washington University (S10 RR027552), as well as NIH grants K08 NS064051 (SHF), R01 NS065069 (DLB), and St. Louis Children\'s Hospital Foundation (JP).
| {
"pile_set_name": "PubMed Central"
} |
Pediatric intestinal failure (IF), as a result of short bowel syndrome (SBS) or gastrointestinal motility disorders, is a condition characterized by insufficient bowel function to maintain hydration and nutrient absorption for growth and development[@b1][@b2]. IF-associated liver disease (IFALD) is a serious complication and the leading cause of morbidity and mortality in children IF patients[@b3][@b4]. However, the mechanisms underlying the development of IFALD are poorly understood. To unravel the mechanisms of IFALD, we performed population based cross-sectional study on serum fibroblast growth factor 19 (FGF19) and bile acid (BA) homeostasis in relation to histological liver damage in pediatric IF patients.
The human FGF family includes about 23 members with diverse biological functions[@b5]. FGF19 is secreted by ileum in response to activation of farnesoid X receptor (FXR) with bile acids[@b6][@b7]. FGF19 has been shown to regulate BA homeostasis through a negative feedback control of bile salt synthesis in human[@b8]. FGF19 has also been implicated in the regulation of carbohydrate, lipid and energy metabolism in the liver[@b9][@b10]. The studies recently have been demonstrated that the decreased serum concentration of FGF19 associated with increased BA synthesis in patients with Crohn's disease and intestinal failure[@b11][@b12][@b13]. Annika *et al*. reported that total or partial loss of ileum decreased serum FGF19 concentration corresponding to hepatic inflammation and fibrosis[@b12]. However, the roles of FXR/FGF19 signaling in BA metabolism among pediatric IF patients remained unknown. We hypothesized that pediatric IF patients decreased serum concentration of FGF19 in association with ileal inflammation, and corresponding to bile acid dysmetabolism, leading to histological liver injury. To this end, we determined the serum concentrations of FGF19, pro-inflammatory cytokines, histological liver injury, biochemical liver function tests and serum BA levels, and analysis the relationship between them.
Results
=======
IF patients characteristics
---------------------------
A total of twenty-three patients at median age 8.0 months (IQR 3.3--58.7) participated in the study ([Table 1](#t1){ref-type="table"}). Causes of IF included short bowel syndrome (necrotizing enterocolitis (NEC): n = 4, small bowel atresia: n = 5, and mid-gut volvulus: n = 3) and intestinal dysmotility disorders (chronic intestinal pseudo-obstruction (CIPO): n = 7 and extensive aganglionosis of hirschsprung's disease: n = 4). No significant differences in age were observed in controls and IF patients. In total, twenty patients preserved Ileocaecal valve and ileum. Five patients were on parenteral nutrition (PN) and eighteen had weaned off PN 0.6 years (0.5--0.9) earlier, after 3 months (1.7--4) on PN. The PN energy comprised 50.79% (47.14--55.36) of glucose and 31.96% (28.03--36.25) of fat. PN fat was given as soy oil-based emulsion \[1.5 g/kg/day (1.10--1.78)\] and combined with fish oil-based emulsion (0.8 g/kg/day) in two patients.
The IF patients exhibit histological liver injury
-------------------------------------------------
As seen in [Fig. 1A](#f1){ref-type="fig"}, the control liver tissues exhibited normal liver histology. In contrast, the liver sections from pediatric IF patients exhibited liver damages characterized by bile duct proliferation, lymphocytes infiltration, and hepatocyte ballooning ([Fig. 1A](#f1){ref-type="fig"}). Apoptotic hepatocytes were detected by TUNEL staining of liver sections. As expected, few TUNEL-positive cells were observed in the control liver specimens. Conversely, TUNEL-positive hepatocytes increased significantly in liver sections from patients ([Fig. 1A and B](#f1){ref-type="fig"}). When compared with the controls, extensive portal fibrosis was indicated in liver slides of pediatric IF patients ([Fig. 1A and C](#f1){ref-type="fig"}).
The FGF19 is associated with liver injury and inflammation
----------------------------------------------------------
As shown in [Fig. 1D](#f1){ref-type="fig"}, in IF patients, serum FGF19 concentration was significantly lower (n = 23, 51.42 ± 47.03 pg/mL, p \< 0.01) compared to controls (n = 21, 102.31 ± 71.23 pg/mL) ([Fig. 1D](#f1){ref-type="fig"}). As seen in [Table 2](#t2){ref-type="table"}, abnormal values in liver enzymes, and parameters of cholestasis were indicated. The values of liver enzymes including alkaline phosphatase (ALP) (51.76 ± 69.24 U/L vs. 90.34 ± 66.99 U/L), alanine aminotransferase (ALT) (23.23 ± 16.01 U/L vs. 55.77 ± 19.91 U/L, p \< 0.01) and aspartate aminotransferase (AST) (36.35 ± 16.66 U/L vs. 61.67 ± 27.51 U/L, p \< 0.01) were increased in patients' serum compared to the controls ([Table 2](#t2){ref-type="table"}). Plasma total bilirubin (5.33 ± 2.55 μmol/L vs. 6.24 ± 1.35 μmol/L) and conjugated bilirubin (3.16 ± 1.43 μmol/L vs. 3.19 ± 2.26 μmol/L) were unregulated in patients that compared to controls ([Table 2](#t2){ref-type="table"}). Correlated analysis showed that serum FGF19 concentrations were inversely correlated with both the levels of ALP (r = −0.34, p = 0.03) and AST(r = −0.3, p = 0.06) ([Table 2](#t2){ref-type="table"}). As indicated in [Fig. 1D](#f1){ref-type="fig"}, we also found that the pro-inflammatory factors serum interleukin-6 (IL-6) (13.28 ± 21.78 pg/ml vs. 50.63 ± 45.99 pg/ml, p \< 0.001) and tumor necrosis factor-alpha (TNF-α) (0.41 ± 0.20 pg/ml vs. 0.89 ± 0.43 pg/ml, p \< 0.01) concentrations were higher in patients compared to controls ([Fig. 1D](#f1){ref-type="fig"} and [Table 2](#t2){ref-type="table"}). In addition, the serum IL-6 (r = −0.32, p \< 0.05) was inversely associated with serum FGF19 ([Table 2](#t2){ref-type="table"}).
FXR/FGF19 signaling contributes to BA homeostasis
-------------------------------------------------
As shown in [Fig. 2A](#f2){ref-type="fig"}, the total plasma BA significantly elevated in IF patients ([Fig. 2A](#f2){ref-type="fig"}). The BA composition of plasma was markedly altered in pediatric patients with a significant in the proportion of the primary bile acid, including the cholic acid (CA) and α, β, ω-muricholic acid (MCA) ([Fig. 2B--D](#f2){ref-type="fig"}). In contrast, the secondary or tertiary bile acids, including the deoxycholic acid (DCA), taurochenodeoxycholic acid (TCDCA), lithocholic acid (LCA) and hyodeoxycholic acid (HDCA), slightly decreased in IF patients compared to controls ([Fig. 2B--D](#f2){ref-type="fig"}). As a FXR agonist, TCDCA positively related to concentration of serum FGF19 ([Fig. 2E](#f2){ref-type="fig"}). In intestinal, we found that the IF patients without intestinal surgery, suffering from ileal inflammation, had numbers of CD3-positive lymphocytes infiltrating ileal mucosa ([Fig. 3A and B](#f3){ref-type="fig"}). As result of ileal mucosa injury, the FXR expression was reduced significantly in patients when compared to controls ([Fig. 3A and C](#f3){ref-type="fig"}). As shown in [Fig. 3D](#f3){ref-type="fig"}, the CD3-positive lymphocytes in ileal mucosa negatively correlated to the FXR expression ([Fig. 3D](#f3){ref-type="fig"}). In addition, ileal mucosa inflammatory degrees was inversely associated with serum FGF19 levels (r = −0.5, p = 0.04) ([Fig. 3E](#f3){ref-type="fig"}).
As seen in [Fig. 4A](#f4){ref-type="fig"}, the total BA contents in IF patients' liver were higher (442.49 ± 312.93 nmol/mg) compared to controls (199.15 ± 134.59 nmol/mg) ([Fig. 4A](#f4){ref-type="fig"}). The BA composition of the liver tissues of patients was also altered when compared with controls. Both unconjugated and conjugated primary BA, including cholic acid (CA), glycocholic acid (GCA), taurocholic acid (TCA), chenodeoxycholic acid (CDCA) and glycodeoxycholic acid (GDCA), increased in livers of patients related to the controls ([Fig. 4B and C](#f4){ref-type="fig"}). As seen in [Fig. 4D--F](#f4){ref-type="fig"}, contents of CA (r = −0.70, p \< 0.05), CDCA (r = −0.73, p \< 0.05) and GCA (r = −0.71, p \< 0.05) inversely correlated with serum FGF19 ([Fig. 4D--F](#f4){ref-type="fig"}). FXR is an important BA receptor and essential to the BA homeostasis[@b14][@b15][@b16]. We here showed that the hepatic FXR expression was lower in IF patients compared to controls ([Fig. 4G--J](#f4){ref-type="fig"}). In liver, the classic bile acid synthesis is controlled by cholesterol 7a-hydroxylase (CYP7A1)[@b17]. We here found that the hepatic CYP7A1 protein expression increased significantly in patients when compared to the controls ([Fig. 4G--J](#f4){ref-type="fig"}).
Discussion
==========
In this study, we firstly showed that pediatric IF patients exhibited liver injury that characterized by cholestasis, portal inflammation, as well as hepatic apoptosis and fibrosis. Secondly, It demonstrated that the altered FXR/FGF19 signaling was contributed to the cholestasis and liver injury in pediatric IF patients.
As alteration in BA composition can cause hepatotoxicity[@b18], thus analysis of the BA composition is essential to assess the impact of altered BA composition on the development of IFALD. In this study, the IF patients exhibited increased levels of primary BA in blood and liver, including CA and CDCA, concurrent with decreased levels of secondary or tertiary BA, such as DCA. In human, the primary BA, including CA and CDCA, can converted into secondary BA, DCA, LCA, through gut microbial 7-dehydroxylation[@b19]. The primary BA increased in IF patients might be attributed to differences in gut microbiota composition that shifts the microbial modification of the BA. Similar changes in BA composition levels have been reported in various liver diseases, including cirrhosis[@b20], alcoholic liver disease[@b21], cholestasis[@b22], in which it was observed increased CDCA levels. In addition, alterations in BA metabolism are likely to induce the liver damage through affecting the solubilization of phospholipids, cholesterol and other lipids[@b23]. It reported that pro-inflammatory cytokines, including IL-6 and TNF-α, had been shown to be important mediators of cholestatic liver injury[@b24][@b25]. We here found that serum IL-6 and TNF-α concentrations were higher in IF patients compared to controls, which suggest alterations of bile acids metabolism potentiate hepatotoxicity may partly through pro-inflammatory mechanisms.
During bile acids enterohepatic circulation, FGF19 mediates bile acids hemostasis through a negative feedback way[@b9][@b26]. In the enterocyte, bile acids reclaimed by terminal ileum can upregulate FGF19 gene expression via activation of FXR[@b27]. After releasing into circulation, FGF19 reaches the liver and inhibits hepatic bile acids synthesis through suppression of CYP7A1 gene that encodes the rate limiting enzyme in synthesis of bile acids[@b28]. In this study, serum FGF19 concentrations were markedly decreased in IF patients compared to healthy controls. Annika and colleagues recently showed that loss of ileum led to reduced FGF19 production in patients with intestinal failure[@b12]. However, the mechanisms underlying the FGF19 associated with liver injury is still not fully established. Although patients with ileum preserved in this study, the inflammatory infiltrating presented in ileum, resulted in intestinal mucosa injury and further reduced FXR expression. In addition, we also indicated serum FGF19 concentrations were inversely associated with pro-inflammatory cytokine IL-6. IL-6 has been previously reported to involved in cholestasis among infants and children with intrahepatic and extrahepatic cholestasis[@b29]. The intestinal injury in IF patients could cause terminal ileum epithelial cells damaged and reduced the terminal ileum FXR expression. In contrast, the intestinal injury may allow translocation of bacterial antigens in the circulation and cause increased expression of pro-inflammatory including IL-6, leading to disturbed hepatocyte bile acid homeostasis. These findings here support the hypothesis that ileal inflammation leads to impaired intestinal FGF19 expression and altered bile acid homeostasis may through inhibiting ileal FXR activation in patients with IF. Experimental study indicated that administration of FGF19 protected the liver from cholestasis by reducing hepatic synthesis and primary hydrophobic BA[@b30]. In our study, we showed that serum FGF19 levels inversely correlated with hepatic primary bile acids including CA, CDCA and GCA. In addition, the expression of CYP7A1 was markedly increased in liver tissues from patients compared controls, which reinforce the concept that FGF19 can protect liver from cholestatic liver injury via repression of CYP7A1 expression, which suggests that FGF19 may be an important mediator in the pathogenesis of IFALD. Indeed, we showed that FGF19 were conversely related to the markers of liver injury, such as alkaline phosphatase, (ALP). Interestingly, we here did not found that FGF19 was tightly correlated to the bilirubin or lipids in the serum.
As pediatric IF is a rare disease, a relatively limited sample has been one of several limitations in this study. Although these results of our study show association rather than causality, this study adds novel data on FXR/FGF19 pathway in the pathogenesis of IF in pediatric patients. Further studies conducted in more patients with biopsy-proven steatosis, portal inflammation, and histological cholestasis or with different fibrosis stages are warranted to analysis relationship between them and FGF19. Collectively, these results bring new insight to possible approaches for prevention and treatment of IF.
Materials and Methods
=====================
Patients
--------
This study was approved by the Faculty of Medicine's Ethics Committee of Xin Hua hospital (XHEC-C-2016-063), School of Medicine, Shanghai Jiao Tong University, Shanghai, China. A total of 23 serum samples and 7 liver specimens were obtained from pediatric IF patients who underwent surgery. A total of 21 serum specimens were obtained from healthy controls with matched age. 6 normal adjacent non-tumour tissues, taken from hepatoblastoma patients, were used as liver controls. All patients' guardians provided written informed consent. The patients' characteristics are presented in [Table 1](#t1){ref-type="table"}. All methods in this study were carried out in accordance with the relevant guidelines.
Biochemical measurements in serum
---------------------------------
A total of 21 controls' serum samples and 23 patients' serum samples were analyzed in this study. The blood samples were analyzed for alanine aminotransferase (ALT), aspartate aminotransferase (AST), and bilirubin and conjugated bilirubin by using routine hospital laboratory methods. Serum total cholesterol, triglycerides, and high density lipoprotein cholesterol (HDL-C) and low density lipoprotein cholesterol (LDL-C) were determined enzymatically.
Determination of FGF19, IL-6 and TNF-α
--------------------------------------
Serum samples were stored at −80 °C until analyzed after blood collection and centrifugation. FGF19, IL-6 and TNF-α concentrations in serum were measured by ELISA kits (R&D Systems, MN, USA) according to the manufacturer' s instructions (21 controls and 23 patients).
Histological analyses and fibrosis determination
------------------------------------------------
Histological examination was stained with hematoxylin and eosin (H&E). Fibrosis was determined by mason's trichrome stain according to the method described in a previous study[@b31]. Masson's trichrome staining was performed according to the manufacturer's protocol (Genmed Scientifics, Wilmington, DE). The collagen fiber was stained blue, the nuclei were stained black, and the background was stained red. Liver tissues were analyzed for apoptosis using the TUNEL test (TdT-mediated dUTP nick end labeling) was performed using the "*In Situ* Cell Death Detection Kit" from Roche Diagnostics according to the manufacturers' instructions. Sections were analyzed with a fluorescence microscope. Liver biopsies were analyzed by two researchers and a pathologist, blinded to clinical data, until consensus was reached.
Western-blot and Immunohistochemistry (IHC)
-------------------------------------------
Western-blot and Immunohistochemistry (IHC) assays were performed as previously described[@b32]. For western-blot, the protein was extracted from liver tissues of IF patients using RIPA buffer with protease inhibitor cocktail (Pierce). The soluble fraction of the cell lysates was isolated by centrifugation at 12, 000 g for 10 minutes in a microfuge. BCA reagent (Pierce, Rockford, IL, USA) was applied to determine protein concentration. The equal amounts of proteins (150 ug/well) were separated by 4--12% SDS-PAGE, and transferred to nitrocellulose membranes. The membranes were incubated overnight at 4 °C with primary antibodies. Antibodies for GAPDH (Cell signaling technology, Danvers, MA, USA), CYP7A1 (Millipore, Darmstadt, Germany) and FXR (Invitrogen, Camarillo, CA) were used here. After incubation, the membranes were washed with PBS (containing 0.1% Tween) and incubated with horseradish-peroxidase conjugated detected the antigen-antibody complexes using an ECL Plus chemiluminescence reagent kit (Pierce, Rockford, IL, USA). For immunohistochemistry (IHC) analysis, paraffin-embedded tissues were incubated with xylol and descending concentrations of ethanol. Antigen retrieval was performed using citrate buffer, pH 6.0 or PH 8.0. Endogenous peroxidases were removed by incubation with 0.3% H~2~O~2~ for 15 minutes at room temperature (RT) and blocking was performed using 10% bovine serum albumin (BSA) for 1 hour at RT. Primary antibodies were then applied in an optimal concentration overnight in a wet chamber (CYP7A1. Millipore, Darmstadt, Germany; CD3, Abcam, Bristol, UK and FXR, Invitrogen, Camarillo, CA). Following incubation overnight at 4 °C, the slides were rinsed in phosphate-buffered saline (PBS) and incubated with the secondary antibody for 1 hour at RT. Antibody binding was visualized by a liquid DAB Substrate Chromogen System (Dako, Glostrup, Denmark). The slides were rinsed in PBS and counterstained with hematoxylin. IHC images analysis was used software Image Pro Plus (Media Cybernetics) 20 fields/sample.
Measurement of bile acid composition
------------------------------------
Bile acid composition was determined according to the methods that previously reported[@b33][@b34]. A total of 40 bile acids including cholic acid (CA), glycocholic acid (GCA), taurocholic acid (TCA), chenodeoxycholic acid (CDCA), glycochenodeoxycholic acid (GCDCA), taurochenodeoxycholic acid (TCDCA), deoxycholic acid (DCA), glycodeoxycholic acid (GDCA), taurodeoxycholic acid (TDCA), ursodeoxycholic acid (UDCA), glycoursodeoxycholic acid (GUDCA), tauroursodeoxycholic acid (TUDCA), lithocholic acid (LCA), glycolithocholic acid (GLCA), taurolithocholic acid (TLCA), hyocholic acid (HCA), glycohyocholic acid (GHCA), taurohyocholic acid (THCA), α-muricholic acid (αMCA), tauro-α-muricholic acid (TαMCA), β-muricholic acid (βMCA), tauro- β-muricholic acid (TβMCA), ω-muricholic acid (ωMCA), tauro-ω-muricholic acid (TωMCA), hyodeoxycholic acid (HDCA), glycohyodeoxycholic acid (GHDCA), taurohyodeoxycholic acid (THDCA), murocholic acid (MuroCA), dehydrocholic acid (DHCA), glycodehydrocholic acid (GDHCA), taurodehydrocholic acid (TDHCA), 3-dehydrocholic acid (3-DHCA), 7-dehydrocholic acid (7-DHCA), isodeoxycholic acid (isoDCA), apocholic acid (apoCA), 6-ketolithocholic acid (6-KLCA), 7-ketolithocholic acid (7-KLCA), 12-ketolithocholic acid (12-KLCA), 23-nordeoxycholic acid (23norDCA), and dehydrolithocholic acid (DHLCA). Deuterated internal standards (IS) lithocholic acid-2,2,4,4-D4 (LCA-D4) and cholic acid-2,2,4,4-D4 (CA-D4) obtained from Steraloid, Inc. (Newport, RI) as standards. A total 39 plasma samples (18 controls and 21 patients) and 10 liver samples (5 controls and 5 samples) were analyzed in this study. Briefly, about 50 μL of plasma was added 150 μL of methanol with IS (100 nmol/L of CA-D4 and LCA-D4), and then the mixture was incubated for 10 minutes at RT. After centrifugation at 20 000 g for 10 min, 160 μL of supernatant was transferred to a clean tube. The supernatant was dried under vacuum and reconstituted with 40 μL of acetonitrile (with 0.1% formic acid) and 40 μL of water (with 0.1% formic acid). After centrifugation, 5 μL of supernatant was injected for measurement. Liver tissue was weighed and homogenized in 50 μL of icecold 40% methanol, and then centrifuged at 20 000 g for 10 min. The supernatant was transferred to a clean tube. Then, 80 μL of ice-cold methanol/chloroform (3:1, v/v) was added to the remaining pellet and rehomogenized. After centrifugation, the two supernatants were combined, spiked with 10 μL of IS (1200 nmol/L of CA-D4 and LCA-D4), and dried under vacuum. The extracts were reconstituted with 40 μL of acetonitrile (with 0.1% formic acid) and 40 μL of water (with 0.1% formic acid). After centrifugation, 5 μL of supernatant was injected for measurement. The analysis was performed with a Waters ACQUITY ultra performance liquid chromatography (BEHC-18, 1.7 μm 2.1 × 100 mm column) coupled with Waters Xevo TQ-S triple quadrupole mass spectrometry (Waters MS Technologies, Ltd.). Data acquisition and bile acids quantification were performed using the MassLynx 4.1 software (Waters, Ltd.).
Statistical analysis
--------------------
Data statistics are presented as medians IQR or mean ± SD. For comparisons of different groups, statistical significance was determined based on the Student's t test. Correlations were tested with Spearman rank correlation. P values \< 0.05 were considered statistically significant.
Additional Information
======================
**How to cite this article**: Xiao, Y.-T. *et al*. Altered systemic bile acid homeostasis contributes to liver disease in pediatric patients with intestinal failure. *Sci. Rep.* **6**, 39264; doi: 10.1038/srep39264 (2016).
**Publisher's note:** Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
The authors are grateful to Dr Ying Wang and Dr Weihui Yan for their technical assistance. This study is supported by the Shanghai Key Laboratory of Pediatric Gastroenterology and Nutrition (14DZ2272400); Science and Technology Commission of Shanghai Municipality (14411950400/14411950401); Shanghai Municipal Commission of Health and Family Planning (201640264, 2013ZYJB0017); National Natural Science Foundation of China (81400861) and National Natural Science Foundation of China-Key Program (81630039).
**Author Contributions** Y.T.X., W.C., K.J.Z. designed experiments. Y.T.X., W.C., K.J.Z., Y.C., N.L.L. performed experiments. Y.T.X., W.C. analyzed data. Y.C., N.L.L. provided key reagents, protocols and critical reading.
{#f1}
{#f2}
{#f3}
{#f4}
###### Patients characteristics.
Variable Median (IQR)
----------------------------------- -------------------
Patients (n) 23
Male/Female (n) 16/7
Age (months) 8 (3.3--58.7)
Gestation age (weeks) 39 (38--40)
Gestation weight (g) 3200 (2800--3568)
Weight Z-score −4.2 (−7.3\~−3.2)
Height Z-score −3.3(−4.4\~−1.6)
Age at PN start (months) 4.5 (1.6--11.3)
Duration of PN (months) 3.9 (2--6)
Weaned off PN (n) 13
Time after weaning off PN (years) 0.6 (0.5--0.9)
Data are frequency or median (interquartile range, IQR). PN, parenteral nutrition.
###### Liver biochemistry, serum lipids, glucose and inflammatory cytokines in the patients.
Variable Control Patients p value Correlation with FGF19
---------------------------------------------- --------------- --------------- --------- ------------------------ ------
**Liver enzymes**
Plasma alkaline phosphatase, ALP (U/L) 51.76 ± 69.24 90.34 ± 66.99 0.08 −0.34 0.03
Plasma alanine aminotransferase, ALT (U/L) 23.23 ± 16.01 55.77 ± 19.91 \<0.01 −0.19 0.24
Plasma aspartate aminotransferase, AST (U/L) 36.35 ± 16.66 61.67 ± 27.51 \<0.01 −0.3 0.06
**Markers of cholestasis**
Plasma total bilirubin (μmol/L) 5.33 ± 2.55 6.24 ± 1.35 0.15 −0.15 0.37
Plasma conjugated bilirubin (μmol/L) 3.16 ± 1.43 3.19 ± 2.26 0.97 −0.07 0.67
**Serum lipids**
Serum HDL cholesterol (mmol/L) 0.76 ± 0.26 0.63 ± 0.19 0.09 0.07 0.64
Serum LDL cholesterol (mmol/L) 1.50 ± 0.56 1.83 ± 0.45 0.04 0.02 0.92
Serum total cholesterol, TC (mmol/L) 2.00 ± 0.47 2.4 ± 0.61 0.02 −0.16 0.32
Serum triglycerides, TG (mmol/L) 0.79 ± 0.31 1.13 ± 0.60 0.03 0.08 0.6
Plasma glucose (mmol/L) 3.80 ± 1.16 3.21 ± 0.88 0.21 0.18 0.26
**Markers of inflammation**
Serum IL-6 (pg/mL) 13.28 ± 21.78 50.63 ± 45.99 \<0.01 −0.32 0.04
Serum TNF-α (pg/mL) 0.41 ± 0.2 0.89 ± 0.43 \<0.01 −0.33 0.32
| {
"pile_set_name": "PubMed Central"
} |
Background {#Sec1}
==========
*Fasciola gigantica* and *Fasciola hepatica* are omnipresent agents of a zoonotic parasitic disease, fascioliasis, which continues to be a major health burden on animals and humans. Fascioliasis can adversely affect the sustainability of the farm animal industry \[[@CR1]\]. The annual global economic loss due to fascioliasis has been estimated to be in excess of three billion dollars \[[@CR2]\]. Worldwide, at least 2.4 million people have been infected with fascioliasis, with a further 180 million people at risk of being infected \[[@CR3]\]. Despite this high impact and investigations for decades using clinical studies as well as animal models, knowledge about host defense mechanisms against *F. gigantica* is limited. This challenge is partly due to the fact that *Fasciola* spp. are very efficient modulators of the host immune response \[[@CR4]\]. The immunomodulatory capacity of *F. gigantica*, mediated by parasite-derived effector molecules, is believed to play important roles in the establishment of long-lasting infection in the host.
Several studies have investigated various excretory/secretory products (ESPs) and virulence effector molecules employed by *F. gigantica* flukes to ensure their survival and establishment of persistent infection \[[@CR5], [@CR6]\]. Rab proteins are a family of small GTP-binding proteins, part of the Ras superfamily, which regulate intracellular membrane trafficking of several pathogens; including parasites (e.g. *Plasmodium*, *Theileria*, *Cryptosporidium* and *Babesia* \[[@CR7]\] and *Toxoplasma gondii* \[[@CR8]\]), bacteria (e.g. *Mycobacterium* spp. \[[@CR9]\] and *Listeria monocytogenes* \[[@CR10]\]) and fungi \[[@CR11]\]. Despite their crucial role as regulators of vesicular membrane traffic, the roles of Rab proteins in the pathogenesis of *F. gigantica* infection remain largely unknown. Understanding the influence of parasite-secreted proteins on the function of immune cells, such as goat peripheral blood mononuclear cells (PBMCs), is essential due to their important role in the immunopathogenesis of fascioliasis \[[@CR12]\]. In a recent study, we cloned and expressed a recombinant *F. gigantica* 14-3-3 epsilon protein (rFg14-3-3e), and characterized its effects on specific functions of goat PBMCs \[[@CR6]\].
In the present study, we expand our investigation of the effects of *F. gigantica* ESPs on the functions of these immune cells. Specifically, the gene encoding *F. gigantica* Rab10 (FgRab10) was cloned and expressed in *Escherichia coli*. Then, the modulatory effects of the purified recombinant FgRab10 (rFgRab10) protein on the functions of goat PBMCs; including cytokine secretion, proliferation, migration, nitric oxide (NO) production, phagocytosis and apoptosis were investigated. Our results indicate that rFgRab10 protein can significantly influence key functions of goat PBMCs, all are critical facets of the immunopathogenesis of *F. gigantica* infection.
Methods {#Sec2}
=======
Animals {#Sec3}
-------
Three crossbred goats (3--6 months-old) were obtained from the teaching and research flock at Nanjing Agricultural University. Goats were treated with triclabendazole (50 mg/kg body weight) in order to exclude the possibility of any prior infection with liver flukes. Two weeks post-treatment, a faecal specimen from each goat was examined microscopically to exclude the presence of helminth eggs. Female Sprague Dawley (SD) rats (150--200 g) were purchased from the Experimental Animal Center of Jiangsu Province, China (Certificate: SCXK 2008-0004), and used for the production of antibodies. Rats were raised under specific pathogen-free conditions, and fed with sterilized food and water *ad libitum*. All efforts were made to minimize the suffering of animals, and daily health checks were performed throughout the experiment.
Purification and culture of goat PBMCs {#Sec4}
--------------------------------------
Venous blood samples were collected from the jugular vein of goats, with no history of *F. gigantica* infection or other parasitic infections, into Vacutainer tubes coated with ethylenediaminetetraacetic acid (EDTA). PBMCs were isolated from freshly collected blood using a PBMC isolation kit (TBD, Tianjin, China). Culturing was done by incubating the isolated PBMCs in RPMI 1640 medium containing 10% fetal bovine serum (FBS) and 1% penicillin-streptomycin (Gibco, New York, USA). Cultures were maintained in a humidified atmosphere of 5% CO~2~ at 37 °C. The number of PBMCs was adjusted to 10^6^ cells/ml in RPMI 1640 medium and cell viability was assessed using trypan blue dye exclusion method. The number of viable cells was counted in a haemocytometer and only cells with \> 95% viability were used in the experiments. All assays were performed using freshly isolated PBMCs. For the monocyte phagocytosis experiment, adherent monocytes were obtained after incubation of PBMCs for 48 h at 37 °C in 5% CO~2~, followed by removal of the non-adherent cells. In all assays performed to investigate the effects of rFgRab10 protein on certain functions of goat PBMCs and monocytes, cultured cells were exposed to rFgRab10 protein at concentrations of 10 μg/ml, 20 μg/ml, 40 μg/ml or 80 μg/ml. PBMCs treated with SUMO protein expressed in pET-SUMO expression vector and sham-treated with phosphate buffered saline (PBS, pH 7.4) were used as controls in all assays. All experiments were performed in triplicate.
Parasite strain {#Sec5}
---------------
Adult *F. gigantica* flukes were harvested from the gall-bladder of naturally infected buffaloes at local abattoirs in Guangxi Zhuang Autonomous Region, PR China. The collected flukes were washed twice in PBS and used for RNA isolation or stored frozen at -80 °C with RNA stabilizer for later analysis. The identity of the flukes collected was ascertained to be *F. gigantica* based on PCR and sequencing analysis of the second internal transcribed spacer (ITS2) of ribosomal DNA \[[@CR13]\]. This analysis revealed 100% ITS2 sequence homology between the flukes collected and *F. gigantica* samples examined previously from the same region (GenBank: AJ557569).
Cloning and characterization of *FgRab10* gene {#Sec6}
----------------------------------------------
Because *F. gigantica* genome sequences are not available, we searched the NCBI/BLASTx (<https://blast.ncbi.nlm.nih.gov/Blast.cgi>), *F. hepatica* cDNA library and liquid chromatography-tandem mass spectrometry (LC-MS/MS)-based *F. hepatica* ESPs dataset from previously unpublished proteomics studies to identify homologous Ras family protein gene sequences. This analysis identified *F. hepatica Ras* family protein gene (GenBank: PIS87503.1), which was utilized to design primers to amplify *F. gigantica* Ras-related protein *Rab10* (*FgRab10*) gene. Total RNA was isolated from adult *F*. *gigantica* flukes using Trizol reagent (Invitrogen, San Diego, USA), and was used to synthesize first-strand cDNA with reverse transcription polymerase chain reaction (RT-PCR) using the RevertAid First Strand cDNA Synthesis Kit \[Thermo Scientific (EU), Vilnius, Lithuania\]. The cDNA was used for the amplification of *FgRab10* gene using a forward primer (5\'-CCG [GAA TTC]{.ul} ATG GCT AAG AAG TCG TAC GAT G -3\') and a reverse primer (5\'-ATT T[GC GGC CGC]{.ul} TGT AGG ACA CCA GGA GCA-3\'). *Eco*R I and *Not* I restriction sites, which were underlined, were incorporated into the primers. The amplified *FgRab10* gene was digested with *Eco*R I and *Not* I, and ligated into the corresponding cloning sites in the T-vector pMD19 (Simple) (Takara, Dalian, Liaoning, China). The recombinant plasmid was transformed into *Trans5α* chemically competent cells (TransGen Biotech, Beijing, China). Positive clones were sequenced by GenScript (Nanjing, Jiangsu, China) in order to confirm the correct insertion/orientation of *FgRab10* gene in the vector. The *FgRab10* cDNA sequence was translated into amino acid sequences using EditSeq (DNASTAR Lasergene). The signal peptide and conserved domains of FgRab10 were predicted by SignalP 4.1 Server (<http://www.cbs.dtu.dk/services/SignalP/>), and (<https://www.ncbi.nlm.nih.gov/Structure/cdd/wrpsb.cgi>), respectively.
Expression and purification of rFgRab10 protein {#Sec7}
-----------------------------------------------
Positive clones containing the *FgRab10* gene were selected. The *FgRab10* gene fragment was sub-cloned into the pET-SUMO expression vector. The pET-SUMO-*FgRab10* plasmid was transformed into *E. coli* BL2 (DE3), then induced by addition of 1 mM isopropyl-β-d-thiogalactopyranoside (IPTG) (Solarbio, Beijing, China) to a culture with an optical density at 600 nm (OD~600~) of 0.6 at 37 °C. After 5 h, the bacterial cells were collected by centrifugation (10,000× *g*, 10 min, 4 °C), resuspended in 5 ml imidazole buffer (20 mM phosphate, 0.5 M NaCl, 10 mM imidazole), and frozen at -20 °C. The cell preparation was sonicated at 600 W for 20 min and further centrifuged at 10,000× *g* for 10 min at 4 °C, and the recombinant His-tagged protein was isolated from bacterial pellets under denaturing conditions in a solution containing 8 M urea. After a third centrifugation, the supernatant was purified using the His GaviTrap kit (GE Healthcare, Buckinghamshire, UK). The eluted protein was first dialyzed in a solution containing 0.1 M urea to be refolded and renatured in a buffer containing a linear decreased urea gradient (6 M, 4 M, 2 M and 0 M urea), and then dialyzed against PBS to remove imidazole. The non-recombinant pET-SUMO expression vector (without *FgRab10* gene) was used to produce the control His-tagged SUMO protein, which was expressed and purified using the same method used to obtain rFgRab10 protein. The lipopolysaccharide (LPS) was depleted from the rFgRab10 protein using Detoxi-Gel Affinity Pak prepacked columns (Thermo Scientific, Waltham, MA, USA). The concentration of the rFgRab10 protein was adjusted to 1 mg/ml. The endotoxin level of the sample was measured with a Limulus Amebocyte Lysate (LAL) gel clot assay using a Pyrosate® Kit (Cape Cod Inc., East Falmouth, MA, USA).
Preparation of polyclonal antibodies {#Sec8}
------------------------------------
Polyclonal antibodies were produced by subcutaneous immunization of SD rats with purified rFgRab10 protein. Briefly, SD rats were immunized with 300 μg of the purified protein emulsified with complete Freund's adjuvant (1:1). After each 2-week interval, SD rats were boosted four times with the same dose of the rFgRab10 protein mixed with incomplete Freund's adjuvant. One week after the last injection, rat serum was collected and stored frozen at -80 °C.
Western blot analysis of rFgRab10 protein {#Sec9}
-----------------------------------------
The purity of the protein was detected by 12% SDS-PAGE, followed by Coomassie blue staining. Also, 20 μg of the purified rFgRab10 protein were resolved on 12% SDS-PAGE and transferred onto Hybond-C extra nitrocellulose membrane (Amersham, London, UK). The membrane was washed 5 times (5 min each) in TBS-T. The membrane was then incubated with primary antibodies (serum from sheep naturally infected with *Fasciola*) for 1 h at 37 °C (1:100 in TBS-T). After washing 5 times in TBS-T, the membrane was incubated with horseradish peroxidase (HRP)-conjugated rabbit anti-goat immunoglobulin (Ig)-G antibody (Sigma, St. Louis, MO, USA) at 37 °C for 1 h (1:2000 in TBS-T). The immunoreaction was visualized using WesternBright ECL (Advansta, Menlo Park, CA, USA) and LucentBlue X-ray film (Advansta, Menlo Park, CA, USA).
Immunofluorescence staining of rFgRab10 protein bound to goat PBMCs {#Sec10}
-------------------------------------------------------------------
PBMCs were incubated with rFgRab10 protein in a humidified atmosphere of 5% CO~2~ at 37 °C for 1 h. Cells were fixed with 4% paraformaldehyde, washed 3 times in PBS and treated with blocking solution (4% BSA in PBS) for 1 h to minimize background staining. rFgRab10-treated, SUMO protein-treated, or PBS-treated PBMCs were incubated with rat anti-rFgRab10 protein antibody (1:100 in 4% BSA) for 1 h at 37 °C. Cells were stained with Cy3 conjugated goat anti-rat IgG secondary antibody (1:500 in 4% BSA) (Beyotime, Jiangsu, China) for 1 h at 37 °C. Hoechst 33342 (Invitrogen, Oregon, USA) was used to stain the nuclei. Stained cells were imaged using a confocal laser scanning microscope (LSM710, Zeiss, Jena, Germany) and digital images were analyzed using Zen 2012 imaging software.
Quantification of cytokines in goat PBMCs culture supernatant {#Sec11}
-------------------------------------------------------------
Cytokine production was detected in the culture supernatants of PBMCs incubated with rFgRab10 protein. PBMCs were plated at a cell density of 10^6^ cells/well in 24-well tissue culture plates in 1 ml culture medium with various concentrations (10, 20, 40 and 80 μg/ml) of rFgRab10 protein. The culture plates were maintained in 5% CO~2~ at 37 °C for 24 h. The PBMCs supernatant was then collected and assayed for the production of interleukin (IL)-2, IL-4, IL-10, transforming growth factor beta (TGF-β) and interferon gamma (IFN-γ) using goat ELISA kit (Mlbio, Shanghai, China). Limits of detection were between 2 and 1200 pg/ml depending on the cytokine.
Assessment of nitric oxide (NO) production {#Sec12}
------------------------------------------
PBMCs were seeded into 24-well flat-bottomed plates at 10^6^ cells per well in 1 ml RPMI medium. PBMCs were allowed to adhere for 2 h in 5% CO~2~ at 37 °C; cells were then stimulated by rFgRab10 protein using the same concentrations used above. The supernatant was collected after 24 h and analyzed for NO production by Total Nitric Oxide Assay Kit (Beyotime, Haimen, Jiangsu, China). A microplate reader (Bio-Rad, Hercules, CA, USA) was used to measure the optical density at 540 nm (OD~540~). NO levels were calculated against a standard curve generated using 0-80 μM/l sodium nitrites.
Determination of apoptosis assay using Annexin V/PI staining {#Sec13}
------------------------------------------------------------
PBMCs were seeded into 24-well tissue culture plates at 10^6^ cells/well in 1 ml RPMI medium with each of the rFgRab10 protein concentrations used above. The plate was incubated in 5% CO~2~ at 37 °C for 24 h. PBMCs were then stained with Annexin V and PI using the Annexin V-FITC kit (Miltenyi Biotec, Bergisch Gladbach, Nordrhein-Westfalen, Germany). Apoptosis in PBMCs was quantified by flow cytometry. Data were analyzed using FlowJo 10. Annexin V-positive/PI-negative PBMCs were considered to be apoptotic and were analyzed as a percentage of the entire PBMCs population.
*In vitro* goat PBMCs migration assay {#Sec14}
-------------------------------------
To test whether rFgRab10 protein can stimulate the *in vitro* the migration of PBMCs, we assessed the movement of untreated and rFgRab10-tretaed PBMCs from the upper chamber of a Transwell plate (Corning, Kennebunk, USA), through an 8-μm-pore polycarbonate membrane inserts toward the lower chamber. Briefly, \~10^6^ cells in 100 μl RPMI medium without or with rFgRab10 at the concentrations used above were seeded into the upper chamber of the Transwell in 24-well tissue culture plates. The bottom chamber was filled with \~600 μl RPMI medium. The plate was incubated in 5% CO~2~ at 37 °C for 4 h. After incubation, PBMCs were collected from the bottom chamber and the number of cells migrated through the membrane into the lower chamber was determined using an Automated Cell Counter (Countstar, Shanghai, China). The results are presented as percentages of the seeded PBMCs.
Measurement of FITC-dextran uptake {#Sec15}
----------------------------------
Goat monocytes were seeded into 24-well tissue culture plates at 10^6^ cells/well in 1 ml RPMI medium with the various rFgRab10 protein concentrations. The plate was incubated in 5% CO~2~ at 37 °C for 48 h. After centrifugation at 500× *g*, monocytes were resuspended in chilled PBS, and incubated with 1 mg/ml FITC-dextran (average molecular weight of dextran, 4000; Sigma, St. Louis, MO, USA) in RPMI 1640 Medium at 4 °C or 37 °C for 1 h in darkness. To halt phagocytosis, chilled PBS containing 2% FBS was added after incubation and goat monocytes were washed trice with PBS and resuspended in chilled PBS. FITC-dextran uptake of monocytes was analyzed using flow cytometry (BD Biosciences, San Jose, CA, USA). The data were analyzed using FlowJo 10.
Determination of goat PBMCs proliferation {#Sec16}
-----------------------------------------
PBMCs were plated in 96-well tissue culture plates at a cell density of 10^4^ PBMCs/well in triplicate in 100 μl complete RPMI medium with each of the rFgRab10 protein concentrations used above. The plate was incubated in a humidified atmosphere of 5% CO~2~ at 37 °C. After 48 h, 10 μl of Cell Counting Kit-8 (CCK-8) (Beyotime, Jiangsu, China) reagent was added to each well and the plate was further incubated under the same conditions in darkness. A microplate reader (Bio-Rad) was used to measure the optical density at 450 nm (OD~450~).
Bioinformatics {#Sec17}
--------------
We used blastx (<https://blast.ncbi.nlm.nih.gov>/) to obtain the protein sequence of rFgRab10 and the Fragments Library of ROBETTA (<http://robetta.bakerlab.org>/) \[[@CR14]\] to obtain the 3-mer and 9-mer fragments. Using the sequence of rFgRab10, and the 3-mers and 9-mers, the *de novo* 3D protein structure was predicted using ROSETTA's AbinitioRelax function ([https://www.rosettacommons.org/](https://www.rosettacommons.org/))) \[[@CR15]\]. The pdb result was aligned to the human Rab10 chain A protein using the Match-maker and match-align tools of UCSF Chimera (<https://www.cgl.ucsf.edu/chimera/>) \[[@CR16]\]. UCSF Chimera was also used to visualize the *ab initio* structure and the alignment to the human Rab10 chain A protein.
Protein Blast \[[@CR17]\] was used to retrieve the protein sequence of humans (taxid: 9606), most similar to the rFgRab10 protein sequence. The human Rab10 protein (Uniprot ID: P61026) was used to query interactions in the following databases: Intact \[[@CR18]\], String \[[@CR19]\], BioPlex \[[@CR20]\] and BioGrid \[[@CR21]\]. The results were summarized and filtered by P61026 (Rab10) protein using a Python script; this list of interacting pairs was the input used to obtain structure interactions using Interactome3D \[[@CR22]\]. All experimental and modeled interactions were used to build a model using the aligned fasta recombinant Rab10 protein sequence to the P61026 (Rab10) protein. A total energy analysis was performed using FoldX Suite \[[@CR23]\] in order to calculate the structure energy of each protein involved in the interaction. We used Cytoscape (<http://www.cytoscape.org>/) to build a protein interaction network. Finally, Gene Ontology (GO) enrichment analysis was performed, using as input the 13 human proteins that interacted with P61026 (Rab10), using FuncAssociate 3.0 \[[@CR24]\].
Statistical analysis {#Sec18}
--------------------
Data values were expressed as the mean ± standard deviation (SD). Differences between the groups were examined for statistical significance using GraphPad Prism 6.0. Statistical analysis was performed using one-way analysis of variance (ANOVA). *P* values \< 0.05, 0.01, 0.001 and 0.0001 were represented as \*, \*\*, \*\*\* and \*\*\*\*, respectively, in comparison with untreated controls. All experiments were repeated at least three independent times.
Results {#Sec19}
=======
Cloning of *FgRab10* gene {#Sec20}
-------------------------
PCR products (\~618 bp) of *FgRab10* gene were successfully cloned into T-vector pMD19. The identity of the construct (pMD19T-*FgRab10*) was verified by DNA sequence analysis. FgRab10 is a 205 amino-acid protein with a molecular mass of \~23 kDa (GenBank: MH532439). Conserved domain analysis showed that *FgRab10* protein belonged to P-loop NTPase domain superfamily. Amino acid sequence comparison of the *FgRab10* gene with that of *F. hepatica Ras* family protein gene (GenBank: PIS87503.1) showed 100% sequence homology.
SDS-PAGE and Western blot analysis {#Sec21}
----------------------------------
The *FgRab10* gene fragment was successfully sub-cloned into pET-SUMO expression vector and the positive clones, pET-SUMO-*FgRab10*, were induced by IPTG in *E. coli* BL2 (DE3). The recombinant rFgRab10 protein was successfully expressed as a soluble His-tagged fusion protein using the small ubiquitin-like modifier (SUMO). A protein with an apparent molecular mass of \~39 kDa (\~23 kDa rFgRab10 protein + \~16 kDa SUMO tag protein) was detected in SDS-PAGE (Fig. [1a](#Fig1){ref-type="fig"}). The specificity of rFgRab10 protein was confirmed by Western blot analysis, wherein rFgRab10 protein specifically reacted with serum from *Fasciola*-infected sheep as indicated by the presence of a single \~39 kDa band. No bands were observed in negative controls \[empty pET-SUMO vector without *rFgRab10* gene insert or in samples probed with serum from healthy sheep (Fig. [1b](#Fig1){ref-type="fig"})\].Fig. 1SDS-PAGE and Western blot analyses of the purified rFgRab10 protein from the sonicated culture sediment of *E. coli*. **a** rFgRab10 protein was resolved on 12% acrylamide gel and stained with Coomassie blue. Lane M: prestained protein ladder in kilodaltons; Lane P: purified recombinant protein, rFgRab10, which appeared as a single band of \~39 kDa. **b** rFgRab10 protein was run under non-reducing conditions, and the immunoreaction was visualized using chemiluminescent HRP substrate and X-ray film. Lane M: prestained protein ladder; Lane P: loaded with rFgRab10 protein expressed from *E. coli*. Serum from *Fasciola*-infected sheep detected a single band of \~39 kDa; Lane C1: loaded with protein expressed from sonicated culture supernatant of *E. coli* transformed with empty pET-SUMO vector without *rFgRab10* gene insert, which did not react with goat serum containing anti-*F. gigantica* IgG antibodies; Lane C2: loaded with rFgRab-10 protein expressed from *E. coli* that did not react with the serum of healthy uninfected goats
Binding affinity of rFgRab10 protein to goat PBMCs {#Sec22}
--------------------------------------------------
Binding of rFgRab10 protein to PBMCs surface was determined by immunofluorescent staining. By incubating rFgRab10-treated PBMCs with specific anti-rFgRab10 antibodies, we detected an even distribution and localization of the red Cy3 dye on the PBMCs membrane, indicating the successful binding of rFgRab10 protein to PBMCs surface (Fig. [2](#Fig2){ref-type="fig"}). No fluorescence was observed in PBMCs incubated with SUMO protein obtained from sonicated culture supernatant of *E. coli* transformed with empty pET-SUMO vector or in PBS-treated goat PBMCs.Fig. 2Localization of *F. gigantica*-derived rFgRab10 protein to the PBMCs surface. rFgRab10 protein binding to goat PBMCs surface was determined by incubating PBMCs treated with purified rFgRab10 protein, protein from sonicated culture of *E. coli* transformed with empty pET-SUMO vector or sham-treated with PBS (control) with primary anti-rFgRab-10 protein antibody raised in rats. PBMCs' nuclei and rFgRab10 protein were stained by Hoechst 33342 (blue) and Cy3-conjugated secondary antibody (red), respectively. Surface staining was detected in rFgRab10-treated cells, but not in PBMCs incubated with SUMO protein and sham-treated PBMCs. *Scale-bars*: 10 μm
rFgRab10 protein modulates cytokine production by goat PBMCs {#Sec23}
------------------------------------------------------------
rFgRab10 protein caused alteration in the levels of five cytokines, IL-2, IL-4, IL-10, IFN-γ and TGF-β. As shown in Fig. [3](#Fig3){ref-type="fig"}, when PBMCs were treated with different concentrations of rFgRab10 protein, the production of cytokines was significantly increased, when compared with PBMCs treated with SUMO protein obtained from sonicated culture of *E. coli* transformed with empty pET-SUMO vector (ANOVA tests: IL-2 (0 μg/ml: *F*~(5,\ 48)~ = 53.63, *P* = 0.0778; 10 μg/ml: *F*~(5,\ 48)~ = 53.63, *P* \< 0.0001; 20 μg/ml: *F*~(5,\ 48)~ = 53.63, *P* \< 0.0001; 40 μg/ml: *F*~(5,\ 48)~ = 53.63, *P* \< 0.0001; 80 μg/ml: *F*~(5,\ 48)~ = 53.63, *P* \< 0.0001), IL-4 (0 μg/ml: *F*~(5,\ 48)~ = 47.06, *P* = 0.8419; 10 μg/ml: *F*~(5,\ 48)~ = 47.06, *P* = 0.0020; 20 μg/ml: *F*~(5,\ 48)~ = 47.06, *P* \< 0.0001; 40 μg/ml: *F*~(5,\ 48)~ = 47.06, *P* \< 0.0001; 80 μg/ml: *F*~(5,\ 48)~ = 47.06, *P* \< 0.0001), IL-10 (0 μg/ml: *F*~(5,\ 48)~ = 39.92, *P* = 0.9661; 10 μg/ml: *F*~(5,\ 48)~ = 39.92, *P* \< 0.0001; 20 μg/ml: *F*~(5,\ 48)~ = 39.92, *P* \< 0.0001; 40 μg/ml: *F*~(5,\ 48)~ = 39.92, *P* \< 0.0001; 80 μg/ml: *F*~(5,\ 48)~ = 39.92, *P* \< 0.0001), TGF-β (0 μg/ml: *F*~(5,\ 48)~ = 83.54, *P* = 0.0839; 10 μg/ml: *F*~(5,\ 48)~ = 83.54, *P* \< 0.0001; 20 μg/ml: *F*~(5,\ 48)~ = 83.54, *P* \< 0.0001; 40 μg/ml: *F*~(5,\ 48)~ = 83.54, *P* \< 0.0001; 80 μg/ml: *F*~(5,\ 48)~ = 83.54, *P* \< 0.0001), and IFN-γ (0 μg/ml: *F*~(5,\ 48)~ = 83.54, *P* = 0.0839; 10 μg/ml: *F*~(5,\ 48)~ = 83.54, *P* \< 0.0001; 20 μg/ml: *F*~(5,\ 48)~ = 83.54, *P* \< 0.0001; 40 μg/ml: *F*~(5,\ 48)~ = 83.54, *P* \< 0.0001; 80 μg/ml: *F*~(5,\ 48)~ = 83.54, *P* \< 0.0001).Fig. 3rFgRab10 protein induced the cytokine secretion. Goat PBMCs were incubated for 24 h in the presence of serial concentrations of rFgRab10 protein. PBMCs incubated with SUMO protein or PBS-treated PBMCs served as controls. The levels of cytokine concentration in the supernatant of cultured PBMCs was quantified by ELISA. Graphs represent means ± SD of data from three independent biological replicates. Asterisks indicate statistical significance between the different indicated groups (\*\**P* \< 0.01; \*\*\*\**P* \< 0.0001; ns, non-significant)
rFgRab10 protein promotes goat PBMCs NO production {#Sec24}
--------------------------------------------------
As shown in Fig. [4](#Fig4){ref-type="fig"}, NO release from PBMCs was significantly increased in the presence of rFgRab10 protein compared with controls, but not at 0 μg/ml and 10 μg/ml (ANOVA tests: 0 μg/ml: *F*~(5,\ 24)~ = 126.2, *P* = 0.2183; 10 μg/ml: *F*~(5,\ 24)~ = 126.2, *P* = 0.9998; 20 μg/ml: *F*~(5,\ 24)~ = 126.2, *P* \< 0.0001; 40 μg/ml: *F*~(5,\ 24)~ = 126.2, *P* \< 0.0001; 80 μg/ml: *F*~(5,\ 24)~ = 126.2, *P* \< 0.0001).Fig. 4Effects of rFgRab10 protein on the production of intracellular NO. PBMCs were sham-treated with PBS, SUMO protein or serial concentrations of rFgRab10 protein and incubated for 24 h at 37 °C at 5% CO~2~. NO concentration in the PBMCs was measured using Griess assay. Graphs represent means ± SD of data from 3 independent biological replicates. Asterisks indicate statistical significance between the different indicated groups (\*\*\*\**P* \< 0.0001; ns, non-significant)
rFgRab10 protein stimulates goat PBMCs apoptosis {#Sec25}
------------------------------------------------
We monitored the level of apoptosis in rFgRab10-treated PBMCs using the Annexin V-FITC/PI double staining apoptosis assay. rFgRab10 protein significantly induced apoptosis in PBMCs in a dose-dependent manner (ANOVA tests: 0 μg/ml: *F*~(5,\ 48)~ = 117.6, *P* = 0.5249; 10 μg/ml: *F*~(5,\ 48)~ = 117.6, *P* = 0.7685; 20 μg/ml: *F*~(5,\ 48)~ = 117.6, *P* = 0.0531; 40 μg/ml: *F*~(5,\ 48)~ = 117.6, *P* \< 0.0001; 80 μg/ml: *F*~(5,\ 48)~ = 117.6, *P* \< 0.0001) (Fig. [5](#Fig5){ref-type="fig"}).Fig. 5rFgRab10 protein induced apoptosis in PBMCs. Annexin V/PI staining combined with flow cytometry analysis was used to count apoptotic cells. **a** Dot plots showing apoptosis of PBMCs in response to exposure to rFgRab10 protein. **b** Percentage of cell population was compared with plotted apoptotic cells (Annexin V+/PI-). Graphs represent means ± SD of data from three independent biological replicates. Asterisks indicate statistical significance between rFgRab10-treated, SUMO protein and sham-treated PBMCs (\*\*\*\**P* \< 0.0001; ns, non-significant). The statistical significance was only achieved at the two highest concentrations (40 μg/ml and 80 μg/ml)
rFgRab10 protein stimulates goat PBMCs migration {#Sec26}
------------------------------------------------
rFg14-3-3e protein significantly promoted, in a dose-dependent manner, the migration of goat PBMCs at all tested concentrations compared to control PBMCs treated with SUMO protein (10 μg/ml: *F*~(5,\ 12)~ = 176.5, *P* = 0.0230; 20 μg/ml: *F*~(5,\ 12)~ = 176.5, *P* = 0.0015; 40 μg/ml: *F*~(5,\ 12)~ = 176.5, *P* \< 0.0001; 80 μg/ml: *F*~(5,\ 12)~ = 176.5, *P* \< 0.0001; one-way ANOVA) (Fig. [6](#Fig6){ref-type="fig"}).Fig. 6rFgRab10 protein promoted goat PBMCs migration. PBMCs were treated with PBS, SUMO protein or serial concentrations of rFgRab10 protein and then the goat PBMCs migration percentage (%) was determined. Graphs represent means ± SD of data from 3 independent biological replicates. Asterisks indicate statistical significance between the different indicated groups (\**P* \< 0.05; \*\**P* \< 0.01; \*\*\*\**P* \< 0.0001; ns, non-significant)
rFgRab10 protein increases goat monocyte phagocytosis {#Sec27}
-----------------------------------------------------
We determined whether rFgRab10 protein influences the phagocytic ability of monocytes that were cultured with PBS, SUMO protein or serial concentrations of rFgRab10 protein prior to exposure to FITC-dextran. The phagocytic capability of monocytes was examined by assessment of FITC-dextran uptake. rFgRab10 protein at 10 μg/ml, 20 μg/ml and 40 μg/ml (but not at 80 μg/ml) significantly increased the phagocytic ability of monocytes compared with cells incubated with SUMO protein (10 μg/ml: *F*~(5,\ 48)~ = 28.37, *P* = 0.0006; 20 μg/ml: *F*~(5,\ 48~) = 28.37, *P* \< 0.0001; 40 μg/ml: *F*~(5,\ 48)~ = 28.37, *P* \< 0.0001; *F*~(5,\ 48)~ = F7 495 28.37, *P* = 0.4258; oneway ANOVA) (Fig. [7](#Fig7){ref-type="fig"}).Fig. 7rFgRab10 protein promoted phagocytosis of goat monocytes. rFgRab10 protein increased the phagocytic ability of goat monocytes as indicated by the increase in FITC-dextran uptake in goat monocytes treated with serial concentrations of rFgRab10 protein, compared with those treated with PBS or SUMO protein. Graphs represent means ± SD of data from three independent biological replicates. Asterisks indicate statistical significance between the different indicated groups (\*\*\**P* \< 0.001; \*\*\*\**P* \< 0.0001; ns, non-significant)
rFgRab10 protein inhibits goat PBMCs proliferation {#Sec28}
--------------------------------------------------
rFgRab10 protein at 20 μg/ml, 40 μg/ml and 80 μg/ml concentrations (but not at 10 μg/ml) significantly inhibited PBMCs proliferation compared with PBMCs treated with SUMO protein obtained from *E. coli* transformed with empty pET-SUMO vector (10 μg/ml: *F*~(5,\ 30)~ = 140.5, *P* = 0.5720; 20 μg/ml: *F*~(5,\ 30)~ = 140.5, *P* \< 0.0001; 40 μg/ml: *F*~(5,\ 30)~ = 140.5, *P* \< 0.0001; 80 μg/ml: *F*~(5,\ 30)~ = 140.5, *P* \< 0.0001; one-way ANOVA) (Fig. [8](#Fig8){ref-type="fig"}).Fig. 8rFgRab10 protein inhibited goat PBMCs proliferation. Goat PBMCs were treated with PBS, SUMO protein or serial concentrations of rFgRab10 protein and incubated for 48 h at 37 °C at 5% CO~2~. Proliferation of goat PBMCs was determined using CCK-8 assay. Results indicated that rFgRab10 protein significantly inhibited PBMCs proliferation. Mean values and the respective standard deviation of three independent experiments are presented. Asterisks indicate statistical significance between the different indicated groups (\*\*\*\**P* \< 0.0001; ns, non-significant)
3D modeling {#Sec29}
-----------
The Protein Blast result showed that the rFgRab10 protein was very similar to Chain A, structure of Human Rab10, with an identity of 63%, an E-value of 3e-85, a Query cover of 90% and a Max score of 253. The human Rab10 protein is a 200 amino-acid protein, with a molecular mass of \~22.5 KDa and is annotated as Ras-related protein Rab10 (Uniprot ID: P61026). *De novo* rFgRab10 protein structure (Fig. [9](#Fig9){ref-type="fig"}) was similar to human Rab10 protein. Figure [10](#Fig10){ref-type="fig"} shows the structure alignment of rFgRab10 protein to the human Rab10 protein, and Fig. [11](#Fig11){ref-type="fig"} shows a 3D representation of both proteins at the same time: Chain A of Rab10 human protein (red) and rFgRab10 protein (blue), with the aligned region shown in yellow.Fig. 9*De novo* 3D *ab initio* model for rFgRab10 protein. The model was generated using RosettaFig. 10The structure alignment of rFgRab10 to the human Rab10 protein. The predicted rFgRab10 model shared 63% similarity with the human Rab10 protein (Uniprot ID: P61026)Fig. 113D representation of both rFgRab10 protein and human Rab10 protein: Chain A of human Rab10 protein (red) and rFgRab10 protein (blue), with the aligned region shown in yellow
Protein interactions analysis {#Sec30}
-----------------------------
Interactome3D analysis detected 36 interactions between human Rab10 (P61026) and 13 other human proteins. All interactions with their respective total energy for the WildType (P61026 - human Rab10), Total Energy Mutant (rFgRab10 protein sequence aligned to P61026), and the difference of the previous values are provided in Additional file [1](#MOESM1){ref-type="media"}: Table S1. The protein interactions network (Fig. [12](#Fig12){ref-type="fig"}) showed that Rab proteins geranylgeranyltransferase component A 1 (CHM) and Rab proteins geranylgeranyltransferase component A 2 (CHML) had more stable interactions (red edges) with rFgRab10 protein compared to the human ortholog, whereas \[F-actin\]-monooxygenase MICAL3 (MICAL3) and guanine nucleotide exchange factor MSS4 (RABIF) had fewer stable interactions (blue edges) with rFgRab10 protein. These interacting partners were enriched in terms related to RabGTPase signalling, with Rab GDP-dissociation inhibitor activity (GO:0005093), Rab-protein geranylgeranyltransferase complex (GO:0005968) and protein geranylgeranylation (GO:0018344), being the overrepresented GO terms (Additional file [2](#MOESM2){ref-type="media"}: Table S2).Fig. 12The protein interaction network revealed more stable interactions, compared to the human ortholog Rab10, between Rab proteins geranylgeranyltransferase component A 1 (CHM) and Rab proteins geranylgeranyltransferase component A 2 (CHML) and rFgRab10 sequence (red edges); whereas fewer stable interactions were detected between \[F-actin\]-monooxygenase MICAL3 (MICAL3) and Guanine nucleotide exchange factor MSS4 (RABIF) proteins and rFgRab10 (blue edges)
Discussion {#Sec31}
==========
To establish a persistent infection in their mammalian definitive host, liver flukes have developed sophisticated strategies to interfere with the host's immune response. In a recent study, we showed that rFg14-3-3e protein, another *F. gigantica* ESP, modulates several cellular and immunological functions of goat PBMCs \[[@CR6]\]. In the present study, the gene encoding Rab10 of *F. gigantica* was amplified by PCR, cloned, and expressed in a prokaryotic expression system. SDS-PAGE and Western blot analysis of the recombinant protein (rFgRab10) using sera from *Fasciola*-infected sheep revealed a molecular mass of \~23 kDa. rFgRab10 protein cross-reacted with anti-rFgRab10 antibody as indicated by specific binding of rFgRab10 protein to the surface of the PBMCs. Thus, it is reasonable to hypothesize that stimulation of goat PBMCs with rFgRab10 protein may interfere with several functions of PBMCs. Understanding the mechanisms of rFgRab10 interaction with PBMCs has important *in vivo* relevance for the definition of early events of the cellular responses to rFgRab10 released by liver flukes during infection. Therefore, we examined the influences of rFgRab10 protein on various functions of goat PBMCs. When incubated with PBMCs, rFgRab10 protein significantly (i) promoted the secretion of IL-2, IL-4, IL-10, TGF-β, and IFN-γ; (ii) increased the release of total NO from PBMCs; (iii) stimulated goat PBMCs apoptosis and migration; and (iv) enhanced monocyte phagocytosis; but (v) inhibited PBMCs proliferation. Alterations of these functions in PBMCs may contribute to multiple aspects of *F. gigantica* pathogenesis.
Cytokines are produced by a variety of leukocytes and play many roles in maintaining immune homeostasis in the host \[[@CR25], [@CR26]\]. *Fasciola gigantica* infection can induce a mixed Th1/Th2 immune response \[[@CR27], [@CR28]\]. Our evaluation of the cytokine production \[Th1 (IFN-γ and IL-2) and Th2 (IL-4, IL-10 and TGF-β)\] in PBMCs culture supernatants showed that rFgRab10 protein can induce both humoral and cellular immune responses, confirming an earlier observation \[[@CR27], [@CR28]\]. Th1 cells secrete IFN-γ, which activates monocytes, thereby promoting the production of NO and phagocytosis \[[@CR29]\], which may influence the development of the adaptive cellular immune response. Th2 cells secrete IL-4 to mediate humoral immunity and control helminthic infection \[[@CR30]\] *via* the production of IgE by B cells, which can activate antibody-dependent killing of eosinophils, macrophages and mast cells \[[@CR31]--[@CR33]\]. Earlier studies indicated that apoptotic cells may participate in suppressing inflammatory responses through inducing the anti-inflammatory cytokines IL-10 \[[@CR34], [@CR35]\] and TGF-β \[[@CR36], [@CR37]\], which inhibit T-cell activation and differentiation \[[@CR38]\]. Thus, reduced proliferation of rFgRab10-treated PBMCs is likely to be caused by cytokines IL-10 and TGF-β, released from apoptotic PBMCs.
Induction of apoptosis to prevent potentially harmful immune responses has been reported in macrophages and eosinophils exposed to *F. hepatica* ESPs \[[@CR39], [@CR40]\]. The same observation has been reported in peritoneal leucocytes of sheep experimentally infected with *F. hepatica*, and has been assumed to facilitate the juvenile flukes survival during the migration through the peritoneum to the liver \[[@CR41]\]. *Fasciola hepatica* seemed to suppress the inflammatory response by inducing apoptosis in goat PBMCs \[[@CR42]\]. Likewise, *F. gigantica* rFg14-3-3e protein significantly induced apoptosis in goat PBMCs \[[@CR6]\]. In support of these findings, we have demonstrated the induction of apoptosis in goat PBMCs cultures following *in vitro* exposure to rFgRab10 protein. rFgRab10-mediated reduction in cellular proliferation and increased apoptosis of PBMCs seems to be mechanisms employed by *F. gigantica* to evade host immune defenses.
A previous study indicated that the intrinsic mitochondrial pathway and the extrinsic death receptor pathway may impact *F. hepatica*-induced apoptosis in goat PBMCs \[[@CR42]\]. However, apoptotic mechanisms may vary in response to different parasite effectors and the mechanism that underpins rFgRab10 protein-induced apoptosis remains unknown. The host immune system produces substances with oxidizing activities, such as NO, reactive oxygen species (ROS) and reactive nitrogen intermediates (RNI) in order to counter the parasite infection \[[@CR43], [@CR44]\]. NO can be released by a variety of immune cells, especially in response to IFN-γ stimulation \[[@CR29], [@CR45]\]. Our data show that rFgRab10 protein promoted the secretion of IFN-γ from PBMCs \[[@CR46]\]. IFN-γ is a potent inducer of nitric oxide synthase 2 (Nos2) in a range of leukocytes, which converts l-arginine into l-citrulline and NO \[[@CR47]\]. NO may therefore serve as a mediator of IFN-γ-induced cellular damage. Further studies are warranted to investigate the process by which rFgRab10 protein induces apoptosis in goat PBMCs.
Our bioinformatics analysis revealed 63% identity between rFgRab10 and human Rab10 protein (Uniprot ID: P61026). The diverse roles of rFgRab10 protein suggest that this protein functions through coordinated interactions with several regulatory partners. Thus, we used human Rab10 as an input to identify potential interactions of rFgRab10 protein with human proteins in major databases. Interactions identified between rFgRab10 and human proteins were of special interest. The protein interaction network revealed that Rab proteins geranylgeranyltransferase component A 1 (CHM) and Rab proteins geranylgeranyltransferase component A 2 (CHML) had more stable interactions with rFgRab10 protein, whereas \[F-actin\]-monooxygenase MICAL3 (MICAL3) and guanine nucleotide exchange factor MSS4 (RABIF) had fewer stable interactions with rFgRab10 protein. Host-parasite protein interactions are critical determinants in pathogenesis. Our study provides the first identification of host protein partners for rFgRab10 protein. Further understanding of the interactions existing between rFgRab10 and host proteins, and how the involved proteins interact, can advance our understanding of molecular mechanisms involved in *F. gigantica* infection and provide new opportunities for developing immunomodulatory therapies. We also performed GO enrichment analysis, which identified terms related to Rab GTPase signaling, with Rab GDP-dissociation inhibitor activity, Rab-protein geranylgeranyltransferase complex, and protein geranylgeranylation. These findings suggest that rFgRab10 can initiate signaling events to modulate various cellular immune functions that may have implications in *F. gigantica* pathogenesis.
Conclusions {#Sec32}
===========
To our knowledge, this is the first study to characterize the rFgRab10 protein and to illustrate that this protein, when incubated with goat PBMCs, modulates the cellular immune response, and enhances the release of NO, cell apoptosis, cell migration, and monocyte phagocytosis, but inhibits cell proliferation. The demonstration of the cross-reactivity of rFgRab10 protein with serum from infected animals, its binding to and interaction with goat PBMCs, and its influence on various cellular functions suggest that rFgRab10 protein plays various roles in the immunopathogenesis of fascioliasis. The immunomodulatory roles of rFgRab10 protein and its potential to influence the host-parasite equilibrium in the context of *F. gigantica* infection makes rFgRab10 a potentially good candidate for developing an immunomodulatory therapy for *F. gigantica* infection. Our study provided a "proof of concept" for clinical evaluation of the immunomodulatory roles of rFgRab10 protein in infected goats.
Additional files
================
{#Sec33}
Additional file 1:**Table S1.** List of all interactions with its respective total energy for the WildType (P61026 - Rab10), Total Energy Mutant (rFgRab10 sequence aligned to P61026) and the difference of the previous values. (XLSX 10 kb) Additional file 2:**Table S2.** Results of the gene ontology enrichment analysis of the interacting proteins. (XLSX 12 kb)
BSA
: Bovine serum albumin
cDNA
: Complementary deoxyribonucleic acid
ELISA
: Enzyme linked immunosorbent assay
ESPs
: Excretory/secretory proteins
FBS
: Fetal bovine serum
FITC
: Fluorescein isothiocyanate
HRP
: Horseradish peroxidase
IFA
: Immunofluorescence assay
IFN-γ
: Interferon gamma
IgG
: Immunoglobulin G
IL-10
: Interleukin 10
IL-2
: Interleukin 2
IL-4
: Interleukin 4
ITS
: Internal transcribed spacer
LC-MS/MS
: Liquid chromatography-tandem mass spectrometry
LPS
: Lipopolysaccharide
MFI
: Median fluorescence intensity
NO
: Nitric oxide
PBMCs
: Peripheral blood mononuclear cells
PBS
: Phosphate-buffered solution
PI
: Propidium iodide
PS
: Penicillin-streptomycin
RNA
: Ribonucleic acid
RT-PCR
: Reverse transcription polymerase chain reaction
SD
: Sprague Dawley
SD
: Standard deviation
SDS-PAGE
: Sodium dodecyl sulfate polyacrylamide gel electrophoresis
TBS-T
: Tris-buffered saline containing 0.1% tween-20
TGF-β
: Transforming growth factor beta
We thank Ms Rong Li for valuable technical help regarding the experiments.
Funding {#FPar1}
=======
The work was supported by a grant from the National Key Basic Research Program (973 Program) of China (Grant No. 2015CB150300) and by the Elite Program of Chinese Academy of Agricultural Sciences.
Availability of data and materials {#FPar2}
==================================
The data supporting the results of this article are included within the article and its additional files. The nucleotide sequence of the *F. gigantica* Ras-related protein Rab10 (FgRab10) analyzed in this study has been deposited in the GenBank database under the accession number MH532439.
XQZ, XRL and HME conceived the idea, planned the experiments and critically revised the manuscript. ALT performed the experiments, analyzed the data and drafted the manuscript with the help of HME. MML, FKZ, XWT, WJW and SYH helped in the implementation of the study. GCM, EP and HME performed and interpreted the computational analysis. All authors read and approved the final manuscript.
Ethics approval and consent to participate {#FPar3}
==========================================
All experiments in this study were approved by the Science and Technology Agency of Jiangsu Province \[Approval number: SYXK (SU) 2010--0005\].
Consent for publication {#FPar4}
=======================
Not applicable.
Competing interests {#FPar5}
===================
The authors declare that they have no competing interests.
Publisher's Note {#FPar6}
================
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
| {
"pile_set_name": "PubMed Central"
} |
1. Introduction {#sec1}
===============
With the increasing popularity in seeking complementary and alternative medicine (CAM) as a healthcare service, the prevalent use of herbal medicine as part of treatment is inevitable. Along with the rapid growth in consumption comes the pressing question on the safety of herbal medicine. While much research and investigation on the potential uses of herbal medicine has been done widely, evaluation on the safety of herbal medicines is relatively scarce. The lack of knowledge of the nature and mechanism of interaction of herbal medicines in the human body has brought about exaggerated reports and extreme measures to counter the adverse effects reported. This paper aims to review the safety of Rhizoma Coptidis (RC) and berberine, using the prohibition of use and subsequent lifting of ban on RC and berberine in Singapore as an illustration to highlight the importance of evidence-based studies in Traditional Chinese Medicine (TCM).
2. Rhizoma Coptidis and Berberine {#sec2}
=================================
RC is a commonly used TCM herb for "*clearing damp-heat, quenching fire and counteracting poison*" and is found in prescriptions for various ailments including febrile illness, hepatobiliary diseases, and gastroenteritis \[[@B1]\]. The earliest record of RC dates back to*Shen Nong Ben Cao Jing* (*Shen Nong\'s* Herbal Classic) compiled in the Han Dynasty of China, in which it was classified as a top-grade drug \[[@B2]\]. The prevalent use of RC as part of a TCM compound formula (*fufang* in Chinese) was cited in several renowned TCM medical compilations, namely,*Shang Han Za Bing Lun* (Treatise on Febrile and Miscellaneous Diseases),*Wai Tai Mi Yao* (Medical Secrets from Royal Library), and*Ben Cao Gang Mu* (*Compendium of Materia Medica*) \[[@B3]\], some of which are still in use today clinically and in evidence-based studies \[[@B4]--[@B7]\]. TCM formulae containing RC have also been used during pregnancy and in neonates for various conditions including neonatal jaundice (NNJ) \[[@B8]\].
Berberine, an alkaloid isolated from RC, was first isolated in the early nineteenth century \[[@B9]\]. Numerous studies on these two subjects have been conducted. Berberine and RC were also reported to have antimicrobial effects \[[@B10], [@B11]\] and are used to treat bacteria-induced diarrhea \[[@B12]\]. Recent area of interest is the antineoplastic effects of RC and berberine \[[@B13], [@B14]\], specifically in areas of breast cancer \[[@B15], [@B16]\], leukemia \[[@B17], [@B18]\], gastric cancer \[[@B19], [@B20]\], pancreatic cancer \[[@B21]\], and nasopharyngeal cancer \[[@B22]\]. It was also reported that RC and berberine have hepatoprotective \[[@B23]\], nephroprotective \[[@B24]\], neuroprotective \[[@B25]\], and cardioprotective \[[@B26]\] effects. Studies have also explored the effect of berberine and RC in controlling metabolic syndrome \[[@B27]\], hyperlipidemia \[[@B28], [@B29]\], and type II diabetes \[[@B30]\], highlighting berberine and RC as multifaceted drugs with immense therapeutic potential.
3. The Prohibition of Use of RC and Berberine in Singapore {#sec3}
==========================================================
TCM was first introduced in Singapore by the influx of Chinese immigrants in the early days as part of culture heritage and healthcare. It provided an affordable and familiar healthcare service for the new Chinese immigrants before the 1960s who found Western medical care foreign and unaffordable. TCM has since developed to become an integrative part of complementary and alternative medicine (CAM). Currently, Western Medicine serves as the mainstream of the healthcare system in Singapore \[[@B31]\]. Nonetheless, TCM remains the most widely used CAM in Singapore, accounting for 88% of CAM use locally \[[@B32]\]. Prior to its ban in 1978, RC was widely consumed as part of oral administered compound formula in Singapore. Its properties of "*clearing damp-heat, quenching fire and counteracting poison*" as described earlier were suitable to treat diseases caused by tropical weather in Singapore.
Mass screening for G6PD deficiency in newborns has been introduced in Singapore since 1965. This is a vital move as preventive measures and effective counseling needed to be instituted early to prevent exposures to triggers \[[@B33]\]. Under this surveillance program, more than two decades of prevalent kernicterus was observed in Singapore. It was found that half of the babies suffering from kernicterus were suffering from deficiency of glucose-phosphate-6 dehydrogenase (G6PD) deficiency \[[@B34], [@B35]\]. G6PD deficiency is an X-linked disorder of the antioxidant homeostasis that is caused by mutations in the G6PD gene \[[@B36]\]. This condition currently affects about four hundred million people worldwide \[[@B37]\], making it the most common enzymopathy in the world. G6PD is an important enzyme that helps to protect the erythrocytes from oxidative damage. Within the restricted metabolism of erythrocytes, G6PD catalyses the first step in the hexose monophosphate pathway, converting glucose-6-phosphate to 6-phosphogluconolactone and reducing the cofactor nicotinamide-adenine dinucleotide phosphate (NADP) to NADPH. The second enzymatic step in the pathway is also associated with the reduction of NADP to NADPH ([Figure 1](#fig1){ref-type="fig"}). As G6PD is the only source of NADPH, which is essential in the protection of erythrocytes from oxidation, premature lysis of erythrocytes may occur in the absence of the enzyme \[[@B38]\]. In severe cases, this results in NNJ and kernicterus, causing permanent damage to the brain, resulting in mental retardation, convulsion, cerebral palsy, hearing deficit, or even death \[[@B37], [@B39], [@B40]\]. Triggers identified include common drugs like aspirin, methylene blue, primaquine and nitrofuran, and common environmental factors like mothballs (naphthalene), henna, and fava beans \[[@B41]--[@B46]\]. Infections have also been identified as trigger of hemolysis in G6PD deficient individuals \[[@B38], [@B47], [@B48]\]. Chinese herbal medicine containing berberine was also identified as triggers of acute hemolysis in G6PD deficient babies \[[@B49]\].
In the 1980s, a study in Singapore found that a high level of "indirect plasma bilirubin" was observed in local babies in general and nearly 100% were visibly jaundiced in the first week of life, compared to about 30% in Caucasian babies born in Singapore \[[@B50]\]. It also found that a retrospective comparison of a cohort of G6PD deficient neonates yielded a result of 22 out of 102 suffering from severe NNJ after exposure to TCM herbal medicines*in utero,* as compared to 2 out of 34 for those without. The same study also observed that exposure to mothball also triggered NNJ with a prevalence of 29 out of 100 as compared to the 20 out of 113 who were not exposed to mothballs. The authors then concluded that TCM herbal medicines, particularly RC, were the cause of severe NNJ in G6PD deficient neonates in Singapore \[[@B50]\]. Based on this study, the Department of Health (known as Ministry of Health now) announced the prohibition of use RC and items containing berberine in Singapore. These items have since been regulated under the Poisons Act until 2013 \[[@B31]\]. The prohibition had forced local TCM practitioners to need to source for substitute Chinese herbal medicine (CHM) during treatment, which affected the efficacy of the compound formula prescribed for patients.
4. Safety of RC and G6PD Deficiency Studies {#sec4}
===========================================
The implementation of this policy sparked active research and discussion in Hong Kong, Taiwan, and China, where prevalence of G6PD deficiency is pronounced and use of TCM is ubiquitous. More studies observing the safety of RC, as well as its relation to G6PD deficient individuals, were also conducted, bringing greater understanding to the safety of the herb and genetic condition.
4.1. *In Vivo* and*In Vitro* Studies of RC and Berberine {#sec4.1}
--------------------------------------------------------
Several studies have supported the retrospective epidemiology findings described by Wong \[[@B50]\] that identified CHM, particularly RC, as the cause of NNJ in G6PD deficient neonates. A study by Ko et al. on the prooxidative effects of Chinese herbal medicine on G6PD deficient erythrocytes*in vitro* found that RC significantly reduced GSH level and increased the level of methaemoglobin in G6PD deficient blood samples, pointing to the possibility of RC as the cause of neonatal jaundice in G6PD deficient neonates \[[@B51]\]. A study found that chronic intraperitoneal administration of 10 and 20 mg/g of berberine daily for 1 week to adult rats resulted in a significant decrease in mean bilirubin serum protein binding, due to an*in vivo* displacement effect and a persistent elevation in steady-state serum concentrations of unbound and total bilirubin, possibly caused by inhibition of metabolism \[[@B52]\]. Another study by Yeung et al. yielded similar results, discovering that RC had a significant effect in displacing bilirubin from its serum protein binding as assessed by peroxidase oxidation method \[[@B53]\], which results in elevation of free bilirubin that can readily cross the blood brain barrier, resulting in kernicterus in neonates.
Besides the potential damage that RC may cause in G6PD deficient neonates, it was also reported that RC was known to have caused several adverse reactions including respiratory failure, extrapyramidal system reactions, severe arrhythmia, liver function injury, and even death in China \[[@B54]\]. Studies were done to identify the toxic constituents, as well as to evaluate the safety of RC. Ma et al. identified alkaloids to be the cause of toxicity of RC. The same study also identified berberine as the main toxic constituent in RC from both their*in vivo*studies and their*in vitro* studies, stating that the lethal dose of total extract of RC was 2.95 g/kg in mice for oral administration \[[@B55]\]. Another study by Yi et al. evaluated the safety of the main alkaloids from RC, namely, berberine, coptisine, palmatine, and epiberberine. This study which included cytotoxicity, acute toxicity in mice, and subchronic toxicity of RC and alkaloids in RC conducted on rats reported that the oral administration lethal dose of berberine dissolved in water was 713.57 mg/kg. The study also found that subchronic 90-day oral toxicity study of RC alkaloids and RC did not yield significant differences in clinical signs, body weight gaining, organ weight changes, urinalysis and hematological parameters, gross necropsy, and histological alterations compared to the control group. The team concluded that current recommended doses of RC alkaloids and crude RC consumption are relatively safe \[[@B56]\].
It is also important to note that the dosage plays a crucial role in determining the toxicity of RC. In a study evaluating the no-observed-adverse-effect level (NOAEL) and toxicity of RC in rats, urinalysis reflected a significant rise in N-acetyl-beta-glucosaminidase in male rats. However, no mortality or remarkable clinical signs were observed during the study. It was also found that RC had no adverse effects on hematology and serum chemistry. The NOAEL of RC through oral administration was ultimately found to be 667 mg/kg/day for male rats and 2000 mg/kg/day for female rats \[[@B57]\]. In another study by Kheir et al., it was found that median lethal doses for intravenous and intraperitoneal routes of discovery were 9.0386 and 57.6103 mg/kg, inconclusive for oral administration. The team found that an oral administration of berberine at 20.8 g/kg yielded a berberine blood concentration of 0.168 *μ*g/mL, while an increase of dosage to 41.6 g/kg yielded a berberine blood concentration of 0.432 *μ*g/mL, which resulted in mortality rate of 30%. The continual increase in dosage to 83.2 g/kg yielded similar berberine blood concentration and mortality rate as that of 41.6 g/kg, suggesting that the absorption of berberine by animal\'s intestine system has its own limit. The team also found that the bioavailability of berberine varies with different routes of drug administration, with the highest found in intravenous administration and the lowest in oral administration. The team concluded that dosage of berberine for oral administration at 20.8 g/kg (or a berberine blood concentration of 0.168 *μ*g/mL) is safe in mice, and the safety dosage for humans would be 2.97 g/kg of human body weight, which is much higher than the clinically recommended dosage of 15 mg berberine/kg of human body weight \[[@B58]\].
On the other hand, there were also skeptics who questioned the hyperbole of RC as a trigger to hemolytic jaundice in G6PD deficient individuals. A study on the influence of RC and berberine on erythrocyte osmotic fragilities of G6PD deficient rats found that a general dose of RC and berberine through oral administration did not cause hemolysis to erythrocytes of G6PD deficient rats \[[@B59]\]. An animal study concluded that the normal dosage of 2--5 g of RC will neither cause hemolysis of red blood cells (RBC) nor change antioxidant system and functions of RBC \[[@B60]\]. A retrospective Cox model analysis of 412 NNJ cases was done by Wang and Lin, establishing a relationship that points out that instead of causing NNJ, RCs seem to have a preventive effect on NNJ \[[@B61]\].
4.2. Clinical and Epidemiological Observations {#sec4.2}
----------------------------------------------
Clinical and epidemiological studies have also explored the effects of RC on neonates, fetal growth, and G6PD neonates. A study by Fok et al. on 1004 mother-baby pairs found that there was no difference in the degree of jaundice between the infants born to mothers with or without antenatal consumption of herbs \[[@B62]\]. A cohort study by Weng et al. also found that G6PD deficient neonates are at increased risk for hyperbilirubinemia even in the nursery free from agents that can potentially cause hemolysis to G6PD deficient red cells. This study demonstrated that hyperbilirubinemia may occur spontaneously even without exogenous factors for G6PD deficient neonates \[[@B63]\]. Another study by Weng et al. indicated that a combination of CHM with a traditional Chinese maternal diet showed a decrease in the development of jaundice in infants as compared to those mothers who did not consume such medicines and diet \[[@B64]\], showing evidence against the argument of CHM causing NNJ in neonates. Another study by Chuang et al. involving 9895 pregnancies found that ingestion of RC during pregnancy had no significant adverse effect on fetal growth. However, it showed a nonsignificant slight decrease in mean birth weight and increased risk of low birth weight and small for gestational age babies if the frequency of using RC was more than 56 times \[[@B65]\]. Another investigation by Lin et al. found that RC was unable to aggravate jaundice of G6PD in neonates in Guangxi, which is another area with a high prevalence of G6PD deficiency \[[@B66]\]. A review by Valaes found no convincing epidemiological data relating Chinese herbs to hemolysis in G6PD deficient neonates, noting that the only report of NNJ in G6PD neonates occurring after consumption of CHM was that reported by Singapore \[[@B37]\]. Another comprehensive review from China has also pointed out the safety of RC, citing several retrospective studies on the lack of association of RC with NNJ and literature reviews on ancient TCM literatures that found no mention of RC as a prohibited CHM for pregnant woman \[[@B67]\].
A study conducted in Singapore that explored the organ toxicity or electrolyte imbalance in patients with chronic hematological diseases found that oral concoction containing RC was not associated with any aggravation of anemia or liver dysfunction. The same study also reviewed the fact that the composition of Chinese herbs implicated in severe NNJ described by Wong \[[@B50]\] was not specifically analyzed. Along with the literature reviewed, the authors concluded that, based on traditional dosage and indication, the use of RC in oral concoction is safe \[[@B1]\].
5. Lifting of the Prohibition on Berberine and RC {#sec5}
=================================================
After more than three decades of regulation under the Poisons Act, the authorities announced the lifting of the prohibition of RC in Singapore in 2012 as suggested by the Berberine Expert Panel, after taking into account the sufficient safeguards available and the safety of berberine when used appropriately. The ban will be progressively lifted, starting from the use of Chinese propriety medicine containing berberine from January 1, 2013. In the absence of major safety issues, the Health Sciences Authority will review the possible further lifting of prohibition on Chinese herbs containing berberine by 2015 \[[@B68]\].
6. Discussion {#sec6}
=============
With more studies investigating the safety of RC and the effect of RC on G6PD deficient individuals, we can better evaluate the safety of RC on normal individuals and understand the relationship between G6PD deficiency and RC. It is important to note that many factors will affect the toxicity of RC to a great extent. One such factor to be concluded from the review is the route of administration. Several studies have established the toxicity of RC through intraperitoneal and intravenous injections. However, studies exploring the toxicity of RC or berberine through oral administration have found that the recommended dosage of RC and berberine clinically is safe. This was explained by Kheir et al. through their observation that oral administration of berberine actually limited the absorption of berberine as increased dosage of oral administrated berberine yielded similar berberine blood concentration, which interestingly coincides with that of the lethal dose via intravenous administration. Such study is fundamental in evaluating the safety of RC, especially since RC is typically administered orally. Such a study can also alert clinicians that other routes of administration may not be as desired as the traditional oral route when it comes to the use of traditional herbal medicines. Another factor that affects the toxicity of RC would be the dosage. Cytotoxicity,*in vivo,* and*in vitro* tests have been done to evaluate the dosage level of RC that may cause toxicity. The outcomes vary between different tests, as these outcomes may have been affected by the solvent used as well as the line of cells and breed of animals used. Nonetheless, these tests give clinicians a good idea on the subchronic toxicity of RC and its alkaloids. However, it is also crucial for one to note that these toxicity studies of RC*in vivo* and*in vitro* merely provide us with an overview of safety of RC when used individually. One should bear in mind that RC is commonly used as one of the CHM in a compound formula, which results in complicated drug-drug interactions that may alter the toxicity of RC and its alkaloids. More research could be studied to determine the safety of RC in compound formula. Clinical studies involving oral administration of RC to pregnant mothers and infants exhibited the safety of RC in humans within recommended dosage. Although such studies seem to provide a more realistic picture of the toxicity of RC clinically, it involves ethical issues and may risk the safety of the participants, especially since such studies are exploring the safety of traditional herbal medicines. Other studies have also shown that bilirubin levels of G6PD deficient neonates may spontaneously spike even in the absence of environmental triggers, shedding some light on the spontaneity of hyperbilirubinemia in G6PD deficient neonates.
The case of banning and the subsequent lifting of prohibition on the use of RC in Singaporeserved 10 as an epic illustration of how evidence-based studies on safety of herbal medicine can eventually vindicate a useful herbal medicine. The initial prohibition of RC was perhaps a decisive act back then to control and eradicate the occurrence of kernicterus in Singapore. However, the decision was made based solely on a retrospective survey with limited knowledge to the actual mechanism and the composition of the CHM involved. Reviews have commented that there was a lack of direct causal relationship between RC and NNJ in the study that had led to the eventual prohibition of RC in Singapore. This is perhaps due to both the inadequate understanding of the TCM theories that governed the use of CHM at the point of study and the overall lack of evidence-based studies on the safety of RC. After more than three decades, studies have accumulated enough evidence to prove that the use of RC in oral administered compound formula abiding by recommended dosage and indication is relatively safe. This has subsequently led to the reversal of the policy, reflecting the important role of evidence-based studies in policy making.
With increasing acceptance and use of CHM worldwide, safety of CHM is under scrutiny. While there is extensive research investigating the possible therapeutic effects of CHM, more research should be done to evaluate the safety of CHM to safeguard the interest of its users.
Conflict of Interests
=====================
The authors declare that there is no conflict of interests regarding the publication of this paper.
![G6PD generates the NADPH which protects the erythrocytes against peroxides and superoxides generated by oxidative stresses \[[@B43]\].](ECAM2014-871720.001){#fig1}
[^1]: Academic Editor: Rainer W. Bussmann
| {
"pile_set_name": "PubMed Central"
} |
Introduction
============
The extract of *Colchicum autumnale*, which is more commonly known as autumn crocus, wild saffron, naked lady, or any of several other names, has been used in the therapy of gout for more than 15 centuries.[@b1-ijn-6-2697] At present, it is still in clinical use for the treatment of gout, as well as several other inflammatory diseases including familial Mediterranean fever and Behçet's disease.[@b2-ijn-6-2697],[@b3-ijn-6-2697] Colchicine and its colchicinoid derivatives possess the ability to bind irreversibly to tubulin, forming tubulin-colchicine complexes, which hinder microtubule formation and inhibit cell mitosis.[@b2-ijn-6-2697]--[@b4-ijn-6-2697] It has been described that colchicine possesses anti-inflammatory properties, mainly mediated by inhibition of leukocyte adhesion and activity.[@b2-ijn-6-2697],[@b5-ijn-6-2697] At higher doses, tubulin-colchicine complexes induce depolymerization of microtubules, resulting in destabilization of the tubulin cytoskeleton.[@b4-ijn-6-2697],[@b6-ijn-6-2697],[@b7-ijn-6-2697] Whereas most cells rely on actin for their cell morphology, endothelial cells of angiogenic tumor vasculature are more dependent on tubulin to maintain their typically enlongated shape.[@b6-ijn-6-2697],[@b8-ijn-6-2697] Therefore, upon colchicinoid-induced microtubule depolymerization, the tumor endothelial cells lose their shape, thereby exposing the vascular basement membrane, which subsequently leads to coagulation, decreased perfusion, and hemostasis.[@b9-ijn-6-2697],[@b10-ijn-6-2697] This process, known as vascular disruption, deprives the surrounding (tumor) cells of oxygen and nutrients, leading to massive tissue necrosis. Currently, however, there is no use for colchicine and colchicinoids in cancer therapy due to their high systemic toxicity.[@b11-ijn-6-2697] Although in preclinical cancer models doses of colchicine higher than 5 mg/kg induce a significant reduction in the perfusion of tumors, the maximum tolerated dose (MTD) of colchicine is limited to around 1 mg/kg.[@b12-ijn-6-2697],[@b13-ijn-6-2697] Even doses below 0.5 mg/kg, as used in the clinical management of gout and familial Mediterranean fever, are frequently accompanied by gastrointestinal comorbidity (eg, nausea, vomiting, and diarrhea) and hematologic disorders, such as thrombocytopenia. [@b14-ijn-6-2697] Colchicine doses higher than 0.5 mg/kg are generally considered toxic, although lower doses may still cause significant side effects, illustrating its narrow therapeutic index. Overdosing of colchicine may eventually lead to multiple organ failure, including bone marrow suppression, hemolysis, liver failure, renal failure, convulsions, and cardiac arrest, and is often lethal.[@b14-ijn-6-2697],[@b15-ijn-6-2697]
One strategy to limit the side effects caused by colchicinoid therapy is to design colchicinoid prodrugs, which possess pharmacological activity only upon conversion.[@b16-ijn-6-2697] Colchicinoids have a partition coefficient (log P) of around 1 and a relatively high volume of distribution (±2 L/kg), which implies that upon intravenous injection they immediately redistribute into the tissues, explaining the high risk for side effects.[@b17-ijn-6-2697]--[@b19-ijn-6-2697] Therefore, by creating a colchicinoid prodrug with improved aqueous solubility, its volume of distribution is expected to be reduced, confining its distribution to the circulation and extracellular compartment and lowering its off-target toxicity. Additionally, to keep the prodrug in the proximity of its target cells, that is, the angiogenic endothelial cells, the tissue penetration of the prodrug may be reduced by increasing its molecular weight. Previously, colchicinoid prodrugs based on glycopeptide dendrimers and cobalamin (vitamin B12) have been synthesized and characterized in vitro.[@b20-ijn-6-2697],[@b21-ijn-6-2697] However, to be converted to the active colchicinoid, both conjugates required cellular uptake in the tumor tissue. For exploiting the direct cytotoxic activity of colchicinoids -- the inhibition of tumor cell mitosis -- this is a rational approach. For colchicinoid-induced vascular disruption, however, a colchicinoid prodrug that is converted extracellularly, preferably in the proximity of the tumor vascular endothelium, is needed. This may be achieved by utilizing polymer-based colchicinoid prodrugs that are more readily transformed into the active colchicinoid, such as by hydrolysis of an ester bond which allows conversion in aqueous conditions. Previous work reported the synthesis of a hydrophilic colchicinoid prodrug, where colchicine was derivatized and conjugated to poly(ethylene glycol) (PEG) using a linker liable to hydrolysis.[@b22-ijn-6-2697] The synthesis of nanomedicines by conjugating PEG-chains (PEGylation) to low-molecular-weight drugs increases the hydrophilicity and size of the construct, and shields them from interactions with plasma proteins.[@b23-ijn-6-2697]--[@b25-ijn-6-2697] Upon intravenous injection, instantaneous and random diffusion of the colchicinoid prodrug into cells is impeded by the relatively large PEG moiety, thereby preventing the binding to tubulin and limiting its toxicity. However, due to the enhanced permeability of the imperfect angiogenic vasculature, the nanosized colchicinoid prodrug may be passively targeted to the tumor tissue, where, promoted by the reductive microenvironment in the tumor tissue, it hydrolyzes to the active colchicinoid.[@b26-ijn-6-2697] In the present study, a polymeric colchicinoid prodrug containing a hydrolysable linker was studied in vitro and in vivo for its therapeutic potential and toxicity as vascular disrupting agent.
Materials and methods
=====================
Synthesis of polymeric colchicinoid prodrug
-------------------------------------------
Colchicine was derived and conjugated to PEG~5000~ using methodology reported elsewhere ([Figure 1](#f1-ijn-6-2697){ref-type="fig"}).[@b22-ijn-6-2697] In brief, colchicine was hydroxyl-functionalized by substituting the N-acetyl moiety with an N-2-hydroxyacetyl moiety. Subsequently, the hydroxyl group was reacted with methoxy PEG-acetic acid to obtain the hydrolysable polymeric colchicinoid prodrug. The amount of colchicine derivative per milligram of material (ie, colchicine equivalents) was determined by means of ultra performance liquid chromatography (UPLC) using an Acquity UPLC^®^ BEH C18 1.7 μm column (Waters, Milford, MA) and ultraviolet detection at 350 nm (Acquity UPLC^®^ PDA; Waters). The mobile phase consisted of a gradient from 5%--95% methanol in water (v/v) and trifluoroacetic acid as modifier.
In vitro hydrolysis study
-------------------------
The hydrolysis kinetics of the colchicinoid prodrug were determined at 4°C and 37°C in phosphate buffer (20 mM, pH 7.4). During 72 hours, samples were taken at regular time intervals, and stored at −20°C before analysis. For each time point, the concentration of colchicinoid prodrug and hydrolyzed prodrug were determined by UPLC, using the methodology described in the previous section.
In vitro cytotoxicity
---------------------
Human umbilical vein endothelial cells (HUVECs) were grown at 37°C and 5% carbon dioxide in angiogenic growth factor rich EGM^®^-2 medium (Lonza Ltd, Basel, Switzerland). Cells were seeded in 96-well plates (1 × 10^4^ cells/well) for 24 hours before further treatment. Subsequently, the cells were incubated with colchicine and colchicinoid prodrug at concentrations ranging from 0.025--2.5 μM colchicine equivalents. The cytotoxicity of each drug after 6 hours, 24 hours, and 48 hours incubation was determined by colorimetric XTT (2,3-bis-(2-methoxy-4-nitro-5-sulfophenyl)-2H-tetrazolium-5-carboxanilide) cell viability assay.[@b27-ijn-6-2697]
In vivo vascular disrupting efficacy of colchicinoid prodrug
------------------------------------------------------------
All animal experiments were conducted in agreement with the local applicable Dutch law, "Wet op de dierproeven" (1977),[@b28-ijn-6-2697] and the European Convention for the Protection of Vertebrate Animals used for Experimental and Other Scientific Purposes (1986).[@b29-ijn-6-2697] The mice were housed in steel cages, and water and food were provided ad libitum. Female pathogen-free C57BL/6 inbred mice of 21--24 g (Charles River Laboratories International, Inc, Wilmington, MA) were subcutaneously inoculated with 1 × 10^6^ B16F10 cells. Ten days after tumor cell inoculation, when tumor size reached \>100 mm^3^, phosphate buffered saline, colchicine (1 mg/kg), and the colchicinoid prodrug (1 mg/kg and 5 mg/kg, colchicine equivalents) were administered intravenously in the tail vein. The mice were sacrificed at 4 and 24 hours after injection. The tumors were excised, snap frozen in liquid nitrogen, and stored at −80°C upon sectioning.
Histological evaluation
-----------------------
Frozen tumor samples (n = 3 per group) were sectioned (5 μm), acetone fixed, and hematoxylin and eosin stained. Images were taken with an inverted microscope (Nikon Eclipse TE2000U; Nikon Corporation, Tokyo, Japan) using NIS Elements software (Nikon Corporation). Small magnification (10×) overlapping images were taken of the complete tumor area and subsequently stitched together with PhotoFit (v 1.4; Tekmate, Inc, Anchorage, AK) software.
Results and discussion
======================
Although colchicine is widely recognized as a promising VDA for cancer therapy, its dose-limiting toxicity has prevented it from realising this potential.[@b11-ijn-6-2697] Only by dosing colchicine well above its MTD, could significant vascular disruption and subsequent necrosis of tumor tissue be observed.[@b12-ijn-6-2697],[@b13-ijn-6-2697] In the present study, a PEG-based polymeric nanomedicine of colchicine was synthesized to attenuate systemic toxicity and to enhance its therapeutic index by improving its aqueous solubility. To this end, colchicine was derived and conjugated to PEG~5000~ via a hydrolysable linker ([Figure 1](#f1-ijn-6-2697){ref-type="fig"}). The molecular structure of colchicine was modified at the acetamido moiety, which is not part of the pharmacophore, creating a colchicinoid also known as colchifoline, with similar anti-inflammatory and tubulin-binding activity.[@b7-ijn-6-2697],[@b30-ijn-6-2697],[@b31-ijn-6-2697] Hydrolysis studies at physiological conditions (37°C, pH 7.4) showed that the half-life of prodrug conversion was approximately 5 hours, whereas, this was calculated by zero-order extrapolation at approximately 14 days at low temperature ([Figure 1](#f1-ijn-6-2697){ref-type="fig"}). The conversion rate of the prodrug at physiological conditions correlated with its activity in endothelial cell viability experiments. To investigate the antimitotic tubulin-binding capacity as a measure of efficacy, colchicine and the colchicinoid prodrug were incubated at different concentrations (0.025--2.5 μM, colchicine equivalent) with primary HUVECs ([Figure 2](#f2-ijn-6-2697){ref-type="fig"}). After 6 hours of incubation, few or no apparent effects on cell viability were measured for each treatment (two-way analysis of variance, *P* \> 0.05), indicating that several hours of incubation are needed to allow colchicine to interfere with tubulin dynamics. However, after 24 hours and 48 hours of incubation, HUVEC viability was markedly decreased for both colchicine (dose ≥ 0.025 μM, *P* \< 0.001) and the polymeric colchicinoid prodrug (dose ≥ 0.125 μM, *P* \< 0.001). The prodrug, of which \>95% is converted after 24 hours at 37°C, showed at the highest doses a similar cytotoxicity in comparison with colchicine. However, at lower concentrations the prodrug was less potent than colchicine after 24 hours and 48 hours incubation (*P* \< 0.05, 0.125--0.25 μM at 24 hours; 0.025--0.25 μM at 48 hours), despite the fact that practically all prodrug has been converted at these time points. The lower activity of the prodrug can be explained by the delayed availability of the colchicinoid due to the time needed for conversion of the prodrug.
The in vivo efficacy and toxicity of colchicine and the colchicinoid prodrug as VDAs in solid tumors were assessed in mice bearing subcutaneous B16F10 melanoma tumors. To study the systemic toxicity, the weight of the mice was determined before and 24 hours after intravenous treatment with either colchicine or the prodrug. Approximately 8% of total body weight was lost 24 hours after administration of 1 mg/kg colchicine (*P* \< 0.05 one-tailed paired *t*-test) ([Figure 3](#f3-ijn-6-2697){ref-type="fig"}). The high loss of body weight at 24 hours after treatment with 1 mg/kg colchicine illustrates the high toxicity of colchicine, which limits the maximum dose to a level considered insufficient to result in VDA activity.[@b13-ijn-6-2697] However, at 1 mg/kg colchicine equivalent dose, the polymeric prodrug did not induce significant weight loss, and only upon administration of a 5× higher dose (5 mg/kg), did it cause a drop in average body weight similar to that of colchicine at its MTD (12%, *P* \< 0.05). This much higher tolerability of the prodrug compared with free colchicine may therefore allow for colchicinoid doses more likely to result in vascular disrupting activity.
The mice were sacrificed at 4 and 24 hours after treatment and the tumors were excised, sectioned, and stained to examine vascular disruption-induced tissue necrosis. No tumor necrosis was observed at 4 hours ([Figure 4A and B](#f4-ijn-6-2697){ref-type="fig"}) or 24 hours (data not shown) after intravenous injection of phosphate buffered saline or colchicine dosed at its MTD (1 mg/kg). The polymeric colchicinoid prodrug, however, induced tissue necrosis in multiple areas in the tumors 4 hours after administration at colchicine equivalent doses of 1 mg/kg and 5 mg/kg ([Figure 4C and D](#f4-ijn-6-2697){ref-type="fig"}, respectively). A similar extent of necrosis (approximately 50% of total tumor mass) was seen after 24 hours in the tumors of mice treated with colchicine equivalents of 1 mg/kg or 5 mg/kg of colchicinoid prodrug ([Figure 4E and F](#f4-ijn-6-2697){ref-type="fig"}). Although it has been shown previously for colchicine that intravenous doses of 5 mg/kg or higher are required to induce observable vascular disruption and subsequent necrosis in solid tumors,[@b12-ijn-6-2697],[@b13-ijn-6-2697] the polymeric colchicinoid prodrug exhibited vascular disrupting efficacy at a much lower dose (1 mg/kg), despite its reduced potency in vitro.
Polymer conjugation is a successful strategy in prodrug development that has been employed regularly for improving the aqueous solubility of the parent compound.[@b25-ijn-6-2697] An improved aqueous solubility changes the tissue distribution, which might explain the potency of the colchicinoid prodrug in relation to colchicine.[@b16-ijn-6-2697] By employing a PEG-chain larger than 35 kDa, or by utilizing colloidal drug delivery systems such as liposomes, a significant decrease in plasma clearance of the colchicinoid prodrug may be achieved, potentially enhancing its in vivo efficacy even more.[@b25-ijn-6-2697],[@b32-ijn-6-2697],[@b33-ijn-6-2697] Nevertheless, the vascular disrupting efficacy at a low, nontoxic dose makes the polymeric colchicinoid prodrug presented here a promising VDA for cancer therapy. The observed favorable characteristics of the prodrug in vivo, on one hand, may be related to enhanced accumulation of the prodrug in the tumor tissue mediated by its improved aqueous solubility, limiting its distribution into other tissues and allowing it to penetrate via the "leaky" immature tumor vasculature.[@b34-ijn-6-2697] On the other hand, the increased expression and activity of reductive enzymes, such as esterases and carboxylesterases, in tumor and endothelial cells may augment tumor-specific conversion of the prodrug into the active colchicinoid, and thus improve its efficacy at the target site, while the polymer conjugation, as such, limits its toxicity toward other healthy tissues.[@b26-ijn-6-2697],[@b35-ijn-6-2697]--[@b39-ijn-6-2697]
Conclusion
==========
The vascular disrupting efficacy and toxicity of a hydrolysable polymeric colchicinoid prodrug was studied in vitro and in vivo. The presented data convincingly demonstrate that the rate of hydrolysis of the prodrug at physiological conditions correlates with its reduced in vitro efficacy compared with colchicine. In vivo, the colchicinoid prodrug was found to be less toxic, while showing higher VDA efficacy than the parent compound, colchicine. Taken together, this study demonstrates the employment of a promising prodrug strategy using a polymeric nanomedicine for improving the vascular disrupting efficacy of colchicinoids while reducing their systemic toxicity, thereby opening the door for the application of these potent VDAs in cancer therapy.
**Disclosure**
This work was supported by MediTrans, an Integrated Project funded by the European Commission under the Nanotechnologies and Nano-Sciences, Knowledge-based Multifunctional Materials and New Production Processes and Devices (NMP) program, a thematic priority of the European Commission's Sixth Framework Programme.
![Synthesis and hydrolysis kinetics of colchicinoid prodrug. The synthesis of the colchicinoid prodrug is performed in three steps: (1) colchicine is deacetylated to obtain N-deacetylcolchicine; (2) N-deacetylcolchicine is acylated with glycolic acid resulting in a hydroxyl functionalized colchicinoid also known as colchifoline; and (3) the colchicinoid is coupled to methoxy PEG~5000~ to form the colchicinoid prodrug. By using esterification to conjugate PEG to the colchicinoid, a prodrug that is hydrolysable at physiological conditions is created: at 37°C, the prodrug is cleaved within a day (t~1/2~ 5.4 hours), while at 4°C the hydrolysis rate is limited (calculated t~1/2~ 14 days \[zero-order kinetics\]).](ijn-6-2697f1){#f1-ijn-6-2697}
{#f2-ijn-6-2697}
{#f3-ijn-6-2697}
{#f4-ijn-6-2697}
[^1]: The first two authors contributed equally to this work.
| {
"pile_set_name": "PubMed Central"
} |
I[ntroduction]{.smallcaps} {#sec1-1}
==========================
The neonatal intensive care unit (NICU) is a specialized care unit that treats premature and medically fragile newborns and infants. Advancements in medical technology, focused-training of skilled medical personnel, and the establishment of critical care protocols have resulted in greater neonatal survival in the NICU. In the United States, although the number of babies admitted to the NICU has increased over the years to 77.9 per 1000 births,\[[@ref1]\] the infant mortality rate is at an all-time low with only 5.98 infant deaths per 1000 live births.\[[@ref2]\] Because of the growth in infant survival rates and NICU admission rates, a great deal of research has focused on identifying and creating the ideal NICU environment for both the neonate and the staff. Specifically, attention has been dedicated to monitoring and regulating the acoustic environment in the NICU.
The current noise standards for the NICU environment were established decades ago and have not been modified. In 1974, the U.S. Environmental Protection Agency (EPA) recommended that indoor hospital areas maintain an average sound level of less than or equal to 45 decibels, A-weighted (dBA) during the day, and 35 dBA at night to maximize opportunity for patient recovery.\[[@ref3]\] In a 1997 Position Statement summarizing the effects of noise upon the developing fetus and the infant, the American Academy of Pediatrics (AAP) cited EPA standards and applied this recommendation to the NICU environment, stating that average sound levels in the NICU should not exceed 45 dBA.\[[@ref4]\] Over the following decades, several research groups have made additional or updated recommendations in an effort to re-evaluate the EPA and AAP standards. While the recommendations for the average sound levels in the NICU environment did not drastically change, new recommendations regarding the presence of transient sounds in the environment (i.e., doors slamming) were issued, with the maximum level for transient sounds (*L* ~max~ averaged over 1 s) being either 65 or 70 dBA.\[[@ref5][@ref6]\]
Despite the presence of these recommendations, existing analyses of the acoustic environment in the NICU have indicated that these noise standards are being exceeded regularly. Studies from separate NICU environments have demonstrated that average noise levels range from 48 to 55 dBA\[[@ref7]\] and 53.9 to 60.6 dBA.\[[@ref8]\] An additional study found that sound levels exceeded the recommended standard more than 70% of the time,\[[@ref9]\] while another found that noise levels only met recommendations 5.51% of the time.\[[@ref10]\] Furthermore, it appears that NICU incubators also serve as a potential source of noise. Sound levels measured inside enclosed and activated incubators were never measured below recommended noise standards.\[[@ref7]\]
In an effort to decrease noise levels and comply with recommendations, numerous noise abatement programs have been developed and implemented; however, none have been successful in creating an environment that effectively maintains current standards. Most programs have focused on a combination of staff training programs and minor structural modifications, but results have demonstrated that these adjustments did not result in any significant reduction in noise levels.\[[@ref11][@ref12][@ref13]\] Because acoustic noise cannot be avoided in the NICU, it is important to know the effect of noise on the neonate.
A large contrast exists between the intrauterine and NICU acoustic environments. While the fetal peripheral auditory system is fully developed by approximately the 24^th^ week of gestation,\[[@ref14]\] responses to sound emerge earlier. Fetal responses to low frequency sounds emerge approximately six to eight weeks earlier than responses to high frequency sounds.\[[@ref15]\] This evolution of responses may reflect the natural maturation of the fetal central auditory system and have major implications for auditory brain development in infants born prematurely. Studies have demonstrated that exposure to sounds containing frequencies above 500 Hz is atypical for the developing fetus.\[[@ref16]\] Environmental sound is attenuated 40--50 dB at frequencies above 500 Hz by the body barrier of the womb and the impedance mismatch between air and embryonic fluid. In contrast, in the NICU environment, infants are exposed to acoustic frequencies higher than 500 Hz 57% of the time.\[[@ref17]\] Alarms and sounds emitted from necessary life support equipment such as extracorporeal membrane oxygenation machines may contain frequencies as high as 16,000 Hz.\[[@ref18]\] Therefore, infants who are born prematurely and cared for in the NICU are routinely exposed to sound levels and frequencies they are not developmentally prepared to handle.\[[@ref19]\]
Numerous studies have demonstrated the adverse effects of the acoustic environment in the NICU by examining the relationship between acoustic events and alterations in infant physiologic state. Studies have demonstrated that high intensity, transient noises are associated with behavioral disturbances and increases in infant muscle tension.\[[@ref20]\] Other studies have documented a relationship between acoustic noise and changes in infant vital signs including heart rate, respiratory rate, O~2~ saturation, blood pressure, and intracranial pressure.\[[@ref21][@ref22][@ref23][@ref24][@ref25]\]
Though the presence of a relationship between noise and infant state is well documented, it is important to consider how the effects of noise may be unique to the infant who is born prematurely. The preterm infant is physiologically immature in all of its major systems, including the central nervous system. As a result of global systemic immaturity and the lack of developed homeostatic mechanisms, preterm infants demonstrate distinctive responses to environmental stress in the NICU.\[[@ref26]\] Multiple studies have demonstrated that the effects of noise upon heart rate, respiratory rate, and oxygen saturation were more pronounced in preterm infants than in full-term infants.\[[@ref21][@ref27]\] Moreover, in a study that compared term and preterm infant reactions to acoustic stimuli, researchers found that preterm infants, unlike term infants, did not demonstrate the ability to habituate to the stimuli after repeated exposures.\[[@ref28]\]
In addition to instantaneous changes in infant state, the stressful stimuli present in the NICU environment have also been associated with long-term effects on infant development. Physiologic stress responses to acoustic events such as changes in heart rate, intracranial pressure, and oxygen saturation may have a significant impact on the preterm infant's future neurologic development due to altered perfusion and oxygenation of the brain tissue.\[[@ref26]\] This effect is likely intensified by the preterm infant's inability to regulate these systems.\[[@ref29]\] Studies have also suggested that the stressful stimuli present in the NICU environment may be associated with an increased risk for future attention, language, and hearing disorders, which may be partially attributed to the immature auditory system's exposure to certain types of noise in the NICU and limited exposure to speech and language.\[[@ref30][@ref31]\]
Because of the negative outcomes associated with NICU stay, researchers agree that the primary purpose for establishing and maintaining noise standards in the NICU should be to create the most ideal environment for the fragile neonate to grow, heal, and thrive. The NICU environment is full of stressful events, and premature infants are forced to expend a significant amount of energy mediating these stressful stimuli. The goal should be to remove as many of these stressors as possible, so that neonates can reserve their energy for healing, which may result in a reduction in their NICU stay and earlier release to their families.\[[@ref32]\] To accomplish this, attention must be given to the infant's sensory systems, and recommendations for all exposures in the NICU must yield a supportive, nurturing setting that optimizes the neurodevelopmental outcomes of preterm infants.\[[@ref10][@ref33][@ref34][@ref35]\]
S[tudy]{.smallcaps} A[im]{.smallcaps} {#sec1-2}
=====================================
Given the evidence demonstrating an inability for NICUs to comply with current noise level recommendations, the known effects of noise on the physiologic state of the infant, and the long-term impact of certain types of acoustic stimuli on the developmentally immature preterm infant, the aim of the present study is twofold: first, to identify the type, rate, and levels of acoustic events that occur in the NICU, as well as the signal-to-noise ratios (SNRs) corresponding to those events; and second, to identify how these events may affect infant physiologic state as measured by changes in infant heart rate and respiratory rate. This information will lead to a more precise understanding of the characteristics of acoustic events that may impact infant physiologic state and could lead to the development of more practical NICU noise standards that allow for both protection of the infant and reasonable implementation by hospital staff.
M[aterials and]{.smallcaps} M[ethods]{.smallcaps} {#sec1-3}
=================================================
Sound level recordings {#sec2-1}
----------------------
Larson Davis Spark 706RC, 705+, and 703+ Type II Noise Dosimeters (Larson Davis Laboratories, Provo, Utah) were used to record the acoustic environment in both the open and private room settings in the Level IV NICU at St. Louis Children's Hospital. Recordings took place during the morning and afternoon hours of February 23, 2017. In the private room, dosimeters were placed on a narrow counter behind the infant isolette, and microphones were taped near the level of the infant's head. In the open room, dosimeters were placed on a counter with microphones taped to a post located between two isolettes. [Figure 1](#F1){ref-type="fig"} shows the sound recording diagram for both the open (a) and private (b) room. All dosimeters were set to A-weighted, fast detector setting (0.125-s interval), 1-s sample interval, 30 dB gain, and 3 dB exchange rate. Each dosimeter was calibrated before and after the recording session and was found to be following appropriate American National Standards Institute standards.\[[@ref36]\] Data were downloaded and exported via Blaze v 6.1.1 software (Larson Davis Laboratories, Provo, Utah).
{#F1}
The descriptions of acoustic events were documented by an observer, who was physically present in the room during recording periods. The observer categorized perceived events into four main groups: alarm noise, infant-generated noises such as crying, staff and/or family conversations, and transient events such as a cabinet door closing. A clock synced to Greenwich Mean Time (GMT) was used to record the onset time and duration of acoustic events. In addition to tracking acoustic events, the observer also documented perceived infant state throughout the course of the recordings. If the infant was being handled, was moving or crying, or was receiving any type of intervention, the infant's state was determined to be "active." For all other times, the infant was considered to be in a "passive" state. The data recorded during times when the infant was active were removed prior to analysis to prevent potentially confounding changes in physiologic state that may have been explained by handling of the infant.
Average equivalent continuous sound level (*L* ~eq~) and the A-weighted sound levels exceeding 10, 50, and 90% of the time (*L* ~10~, *L* ~50~, and *L* ~90~, respectively) were calculated for the duration of the recording period in each room. Acoustic events were defined as the occurrences of levels that exceeded the *L* ~10~ for that room plus 2 dB, which was added to eliminate spurious events that resulted from minor perturbations in the sound level. This addition of 2 dB also accounts for the accuracy of the dosimeter microphones, which is ±2 dB, as well as the difference limen for intensity in humans, which is about 1 dB for average healthy adult ears.\[[@ref37]\] To include transient events, time points when the *L* ~10~ of the peak levels for that room plus 2 dB was exceeded were also included. The levels of acoustic events were compared to the acoustic level that was exceeded 50% of the time (*L* ~50~) for each room, and an SNR *L* ~50~ was calculated (level of acoustic event−*L* ~50~ = SNR *L* ~50~). Average *L* ~eq~ and average peak level were also compared between time points categorized as events and those classified as nonevents for each room type.
The number and description of perceived acoustic events documented by the observer was compared to the number of events recorded in the acoustic log. The average *L* ~eq~, average peak, and average SNR *L* ~50~ were calculated and compared between rooms. The samples of documented alarm noises were examined and determined to have peak levels that ranged between 82 and 86 dBA. To efficiently and accurately compare the number of events that occurred in each room, all 1-s sampling intervals containing peaks exceeding 84 dBA (median peak level of alarms) were counted for each room and averaged over the duration of the recording, yielding an average number of seconds per hour during which the chosen peak was exceeded for each room.
Infant physiologic recordings {#sec2-2}
-----------------------------
Physiologic data were collected from three infants, two in the open room (Babies A and B) and one in the private room (Baby C), using the BedMaster system.\[[@ref38]\] BedMaster data were collected for most infants in the NICU at St. Louis Children's Hospital. It is an automated system that records and stores patient vital sign data and alarms processed by patient monitors. The data extracted for comparison in this study included heart rate and respiratory rate, which have been reported to be good physiologic indicators of stress in the premature infant,\[[@ref39]\] and have been associated with infant stress responses to adverse environmental stimuli in the NICU.\[[@ref26]\]
The relationship between acoustic events and alterations in infant physiologic state was examined by comparing the physiologic data measured at time points relative to GMT when there was a marked increase in the sound levels to the physiologic data measured at time points when the acoustic levels were steady. To analyze and account for the latency of physiologic responses relative to acoustic events, infant state was offset from acoustic events by 2, 4, 8, and 16 s. Each infant served as his or her own control for baseline state, as physiologic norms are not well established in this population of critically ill infants.\[[@ref40]\]
R[esults]{.smallcaps} {#sec1-4}
=====================
Sound levels and acoustic events {#sec2-3}
--------------------------------
Acoustic data for the open room and private rooms are shown in Tables [1](#T1){ref-type="table"} and [2](#T2){ref-type="table"}. A summary table with comparisons of sound levels between room types is located in [Table 3](#T3){ref-type="table"}. The *L* ~eq~ for the duration of the 4-h recording period in the open room was 58.1 dBA, while the *L* ~eq~ for private room was 54.7 dBA. The noise levels in both environments exceeded the AAP recommended levels, further confirming that noise standards are not being met in the modern NICU. In fact, during the hours of acoustic recordings collected for this study, the 1-s *L* ~eq~ was never measured at or below 45 dBA in either room. The time waveforms of the *L* ~eq~ and peak levels for open (left) and private rooms (right) are shown in [Figure 2](#F2){ref-type="fig"}. Both the absolute levels and the number of acoustic spikes reveal the differing acoustic signatures of the two environments.
######
Open room acoustic data
------------------- -------------- -----------
Time of recording 10:25--14:30
Overall *L* ~eq~ 58.1 dBA
Events Nonevents
Overall *L* ~eq~ 65.8 dBA 56.2 dBA
SNR *L* ~50~ 11.2 dB 1.6 dB
Average peak 88.5 dBA 82.3 dBA
------------------- -------------- -----------
######
Private room acoustic data
------------------- -------------- -----------
Time of recording 10:15--14:30
Overall *L*~eq~ 54.7 dBA
Events Nonevents
Overall *L* ~eq~ 63.8 dBA 50.5 dBA
SNR *L* ~50~ 14.1 dB 0.8 dB
Average peak 90.0 dBA 76.8 dBA
------------------- -------------- -----------
######
Acoustic data by room type
Open room Private room
------------------------------------ ----------- --------------
Overall *L* ~eq~ 58.1 dBA 54.7 dBA
*L* ~90~ 51.1 dBA 48.3 dBA
*L* ~50~ 54.6 dBA 49.7 dBA
*L* ~10~ 61 dBA 53.9 dBA
Average level of acoustic events 65.8 dBA 63.8 dBA
Average SNR *L* ~50~ of events 11.2 dB 14.1 dB
Occurrences (s) of peaks ≥84 dBA/h 579 s 94 s
{ref-type="table"}--[3](#T3){ref-type="table"}.](NH-20-121-g002){#F2}
The occurrences of peak levels exceeding 84 dBA were more frequent in the open room than in the private room. Peak levels were exceeded an average of 579 s per hour in the open room compared to an average of 94 s per hour in the private room. The average level of acoustic events in the open room was 65.8 dBA with an SNR *L* ~50~ of 11.2 dB, while the average level and SNR *L* ~50~ of nonevents in the open room were 52.6 dBA and 1.6 dB, respectively. In contrast, the average level of acoustic events in the private room was 63.8 dBA with an average SNR *L* ~50~ of 14.1 dB during acoustic events and 50.5 dBA with an SNR *L* ~50~ of 0.8 dB during nonevents.
Infant physiologic responses {#sec2-4}
----------------------------
The infants in this study were extremely variable in their physiologic state, which is likely a direct result of their medically fragile condition. [Figure 3](#F3){ref-type="fig"} displays the intersubject variability of infant heart and respiratory rate for all three infants. Each panel displays 1 h of the total recording of the heart rate (black) and respiratory rate (gray) with a 1-s resolution for each infant. While Babies A and B (left) were next to each other in the open room setting with the same acoustic environment, the heart and respiratory rate measures were more variable in Baby B. Consequently, correlations could not be established between acoustic events and alterations in infant physiologic state. Preliminary data analysis showed no difference between mean heart rate or mean respiratory rate during the periods of acoustic events when compared to periods of nonevents due to extreme infant variability (*P* = 0.42 and 0.43, respectively). Furthermore, as shown in [Figure 4](#F4){ref-type="fig"}, no significant difference was found between mean infant heart rate during the acoustic event (0 s) or at delays of 2, 4, 8, and 16 s from the onset of the acoustic event (*F* = 0.70, *P* = 0.53).
{#F3}
{#F4}
It is worth noting that several instances of change in infant heart rate were identified at specific time points when a documented acoustic event occurred. However, infants were inconsistent in their responses, and they also displayed large disturbances even when the acoustic environment was relatively stable. [Figure 5](#F5){ref-type="fig"} demonstrates an example of multiple acoustic events that occurred in the open room while both Babies A and B were passive. The observer noted that these events were the result of alarms that were unrelated to either infant. Baby B's heart rate appeared to fluctuate in concordance with the onset of these events, yet, the heart rate for Baby A in the same environment showed little to no variation. It is possible that Baby A has some other comorbidity such as hearing impairment that could explain this lack of response to the acoustic events. It is also possible that the changes observed in Baby B are coincidental and have more to do with his or her fragile state than with the occurrence of the alarm peaks. Because of the variability in the physiologic state of infants included in this study, the presence or absence of a relationship could not be determined. Given the extreme physiologic variability of these medically fragile neonates, increasing the number of infants observed would not be feasible or cost-effective in this environment as we would need several hundreds of observations to overcome this variability.
{#F5}
D[iscussion]{.smallcaps} {#sec1-5}
========================
The results of this study regarding overall sound levels support and expand upon the findings of earlier research, confirming the need for an improved method of quantifying and regulating the acoustic environment in the NICU. Levels in both the open pod and private rooms consistently exceeded the 45 dBA standard, with sound levels in the open pod room being 3.4 dB higher than levels recorded in the private room. Given the fact that the combined noise emitted from the necessary life support equipment, heating, ventilation, and air conditioning (HVAC) system, lights, and monitors does not meet standards even without additional noise introduced by staff, visitors, or maintenance, it would be futile to continue making efforts to meet the standards as they are written today. Rather than continuing to focus on the overall noise level, an event-based approach may lead to an attainable standard that minimizes stress on the preterm infant and addresses the most critical needs of neonates.
The results of this study indicate that the frequency of acoustic events as well as the levels of these events relative to the background noise level is dependent on NICU room type (open pod vs. private room). The number of events was far greater in the open room due to the combined activities of multiple infants, families, and staff members. The occurrence of events with peak sound levels exceeding 84 dBA was 600% greater in the open room than in the private room. As a result, the *L* ~50~, or the sound level of the room exceeded 50% of the time, was 4.9 dB lower in the private room (49.7 dBA) than the open room (54.6 dBA). Therefore, the SNR of the documented acoustic events to the estimated background noise level (*L* ~50~) is higher for the private room than the open-pod nursery. This caused acoustic events to be much more salient in the private room.
Despite many documented examples of acoustic events throughout the recording periods, the constant variability in infant physiologic state precluded a qualitative assessment of whether or not correlations between the acoustic environment and physiologic state exist. While visual examination of [Figure 5](#F5){ref-type="fig"} suggests a trend in heart rate variance with the onset of acoustic events for Baby B, the overall variation of Baby B's heart rate as shown in Figures [3](#F3){ref-type="fig"} and [4](#F4){ref-type="fig"} prohibits any meaningful correlation between events and physiologic changes. It may not be helpful or cost-effective to collect more data in this medically fragile population due to the extreme physiologic variability that clouds any relationship between infant state and acoustic events. Future studies should consider analyzing potential correlations between these events and changes in physiologic state for infants who are more stable, such as those in the well-baby nursery. If a relationship between acoustic events and physiologic state is observed in healthier infants, these findings may be used to develop recommendations that can be applied to the NICU environment. Although it is not yet clear which acoustic events in the environment might disturb preterm infants most, the findings in this study suggest a need to decrease both the rate of occurrence and the SNR of events in the NICU. Therefore, rather than continuing to focus efforts solely on gross sound abatement strategies, efforts should be made to reduce the occurrence of unnecessary acoustic events. Many events, such as alarms on feeding pumps or heart rate monitors, cannot be prevented and are essential for the care and safety of infants. However, a portion of the transient acoustic events recorded by the observers in both rooms included supplies being dropped on hard surfaces or cabinet and formula refrigerator doors closing loudly, which could possibly be circumvented with minor modifications in equipment and/or human behavior. It may be feasible to develop a software program that analyzes sound levels in the NICU to track the frequency with which these types of events occur. The digital monitoring of acoustic events will eliminate human observer bias and error. Hospital administrators could then use these data to establish protocols for decreasing the amount of preventable events.
Along with reducing as many unnecessary events as possible, it may be beneficial to decrease the SNR of unavoidable acoustic events, such as alarms. One such method might include the introduction of a developmentally appropriate stimulus into the infant's environment. While this approach entirely contradicts the recommendation for reducing overall noise level set forth by current NICU noise standards, this added stimulus could serve as an effective masker that would elevate the noise floor to decrease the SNR of acoustic events perceived by the infant. Though correlations between acoustic events and changes in physiologic state could not be established in this study, it is worth noting that several examples of acoustic events that occurred in the open room were documented during the periods of relative infant physiologic stability and did not yield any significant change in infant state. This could potentially be explained by the fact that the infants whose data were extracted for this study were housed in covered isolettes with ports closed, which are known to have an interior running noise of 58 dBA.\[[@ref7]\] It is possible that this mechanical noise served as an acoustic masker that elevated the noise floor and decreased the SNR of acoustic events at the infant's ear. If this masking noise did contribute to a lack of response from the infants, it would be beneficial to explore other types of acoustic maskers that could provide a developmentally supportive stimulus while simultaneously shielding the infant from potentially adverse acoustic events. This would generate a more tranquil environment that promotes long-term recovery and development.
One example of an appropriate stimulus may include an acoustic recording of the infant's mother. Numerous studies have demonstrated that exposure to recorded maternal sounds can influence the physiologic state of infants in the NICU. In one study, researchers found that infants exposed to maternal sound stimulation experienced fewer cardiorespiratory events compared to infants who were exposed to routine hospital sounds and concluded that maternal sound stimulation may provide short-term improvements in the physiologic stability of preterm infants.\[[@ref41]\] Similarly, another study monitored the heart rate of preterm infants in the NICU during the times of exposure to maternal sounds and generic hospital background noise and found that all infants displayed lower heart rate when exposed to the maternal recordings. This led the researchers to conclude that preserving the preterm infant's natural environment to the greatest extent possible through exposure to maternal sounds may improve the infant's autonomic stability while creating a more soothing environment, which promotes sleep and recovery.\[[@ref42]\]
In addition to providing a more tranquil setting for the infant, some sounds, specifically spoken language, have been shown to augment the infant's development. The NICU environment has been associated with negative long-term outcomes in language development.\[[@ref43]\] Though previous standards have fixated upon the need to create silence in the NICU, this sensory deprivation and elimination of speech and language exposure has been demonstrated to have negative effects on the infant's developing auditory cortex, which can result in language delays later in life.\[[@ref44]\] Additionally, research has shown that language comprises a very small percentage of the sounds to which infants in the NICU are exposed,\[[@ref45]\] despite the fact that greater adult language exposure is associated with better infant language and cognitive scores later in life.\[[@ref46]\] One study found that preterm infants who were exposed to recorded maternal voice and heartbeat showed significantly larger auditory cortical development compared to preterm infants who experienced routine hospital sounds.\[[@ref47]\] Therefore, one reasonable approach could be to incorporate more meaningful sounds, such as maternal speech, into the NICU environment to serve both as acoustic maskers and stimuli that developmentally benefit the infant.The findings of these studies demonstrate that a balance needs to be established between reducing unnecessary stimuli that could cause the infant stress and depriving the infant of meaningful stimuli that could result in impaired development. Thus, future research should focus on the methods of decreasing alarm noise, and other nonhuman noises that can stress the infant,\[[@ref48]\] but should also seek to identify the types of stimulation that will generate the most tranquil, developmentally supportive acoustic environment possible for the preterm infant to achieve optimal neurodevelopment.
This study proposes a method for quantifying and describing acoustic events in the NICU. The findings indicate that acoustic events in the NICU occur frequently, and though a relationship between events and changes in infant physiologic state could not be established due to infant variability, it is likely that these events do disturb the infants. Given the results of this study regarding the rate of occurrence and high SNR of acoustic events as well as the documented developmental benefits of introducing the infant to meaningful stimuli, it is concluded that rather than continuing to strive for the diminution of all sound, the goal should be to incorporate developmentally appropriate acoustic stimuli into the infant's environment. Noise regulations should focus on reducing the number of preventable acoustic events that occur in the NICU, and future studies should aim to identify the characteristics of acoustic stimulation that would decrease the SNR of unavoidable events and result in the most positive developmental outcomes for preterm infants.
Financial support and sponsorship {#sec2-5}
---------------------------------
Nil.
Conflicts of interest {#sec2-6}
---------------------
There are no conflicts of interest.
| {
"pile_set_name": "PubMed Central"
} |
Introduction {#Sec1}
============
The use of combined modality treatment, including surgery, chemotherapy, and radiation, has resulted in increased disease control in locally-advanced head and neck cancer (HNC) \[[@CR1]\], but improved disease control occurs at the expense of increased acute and late effects from therapy. In the HNC population, acute tumor and treatment effects have garnered tremendous interest and have been extensively investigated. Until recently, late effects and their sequelae have been largely underrecognized and underappreciated. Improved treatment methodologies and the changing epidemiology, most notably the rise in HPV-associated oropharyngeal cancers, have resulted in a rapid increase in the number of HNC survivors. Accordingly, this expanding survivor population has generated a surge of interest in the late effects of HNC therapy. Evolving data demonstrate that acute toxicities may persist long-term and develop into late effects. In addition, late effects may manifest months or years after completion of therapy, persisting for years or even lifelong, far longer than previously believed \[[@CR2], [@CR3]\]. When severe, late effects may profoundly affect function and quality of life \[[@CR4]\].
The most frequently studied late effects of therapy are those that are due to *local* tissue damage from cancer or its therapy. However, late *systemic* symptoms, which may have a more ubiquitous and profound impact on long-term function, have remained elusive from the standpoint of both research and management. Systemic symptoms, also known as sickness behaviors, include fatigue, central pain, neurocognitive dysfunction, mood disorders, thermal discomfort, sweating, gastrointestinal symptoms, and sleep disturbances. Systemic symptoms tend to occur in clusters, whichis felt to be due in part a common underlying pathobiology. While the mechanisms and pathways that contribute to systemic symptoms have yet to be fully elucidated, neuroinflammation is believed to be one of the important connective threads.
During the acute phase of illness or injury, the body must coordinate complex biologic pathways and behaviors in order to optimize the body's response to disease and promote healing \[[@CR5]--[@CR7]\]. The illness response is mediated in part through peripheral pro-inflammatory and immune-activating cytokines which act as peripheral-to-central nervous system signaling molecules. The peripheral cytokines can induce a neuroinflammatory state and its associated systemic symptoms. Acutely, the illness response characterized by systemic symptoms such as fever, lethargy, and anorexia may be adaptive \[[@CR8], [@CR9]\]. However, if the inflammatory signal is overly exuberant or protracted, functional and anatomical central nervous system changes may develop. This may result in protracted or chronic systemic symptoms \[[@CR9]--[@CR11]\].
HNC and its treatment are both associated with elevations in peripheral pro-inflammatory cytokines \[[@CR12], [@CR13]\]. While the level of pro-inflammatory cytokines has been correlated with the grade of acute toxicity \[[@CR14]\], the relationship between pro-inflammatory cytokines and late effects has not been reported. Furthermore, while available data indicate that head and neck cancer patients experience chronic systemic symptoms such as fatigue, anxiety, and depression \[[@CR10], [@CR15], [@CR16]\], data describing the breadth, severity, and impact of late systemic effects are not available. To that end, we conducted a cross-sectional, observational, mixed-methods study in HNC survivors to determine the prevalence and impact of late systemic symptoms. Herein we report the results of the quantitative analysis.
Materials and methods {#Sec2}
=====================
Patients {#Sec3}
--------
All patients were recruited from the Henry Joyce Cancer Clinic in the Vanderbilt-Ingram cancer center between November 6, 2014 and November 21, 2016. Patients included in this analysis were consented to participate in two clinical trials, the first of which included 92 patients and was entitled "Characterization of Chronic and Unremitting Symptoms in Long Term Survivors of Head and Neck Cancer." The second study, entitled "Characterization of Chronic Systemic Symptoms among Participants in the Fibrosis-Lymphedema Continuum Study in Head and Neck Cancer," included 13 patients who participated in an earlier, R01-funded study and had agreed to be contacted for participation in subsequent clinic trials. A convenience sample of 105 patients completed study measures and were included in the analysis.
Study eligibility criteria for both trials included the following: age 21 years or older, the ability to speak English, a history of histologically-proven HNC, completion of treatment a minimum of 12 months prior without evidence of recurrence. All eligible patients were approached and provided with information about the study. Interested patients were contacted by study staff and signed informed consent prior to completing study-related questionnaires.
Methods {#Sec4}
-------
After signing informed consent, the participants completed the study questionnaires on an electronic web-based electronic data capture application (REDCap™).
Questionnaires {#Sec5}
--------------
### Socio-demographic data form (self-report) {#FPar1}
Captured birthdate, gender, race, ethnic category, highest educational level, marital status, employment status, area of residence, insurance coverage, and annual household income.
### Disease and treatment data form (medical record review by study staff) {#FPar2}
Captured data related to the patient's cancer and treatment including diagnosis date, location, stage of disease, surgical treatment, medical oncology treatment, and radiation oncology treatment.
### Patient-reported outcome measures {#FPar3}
Patient-reported outcome (PRO) measures were included to address common local symptoms in the HNC population (Vanderbilt Head and Neck Symptom Survey version 2.0) as well as systemic symptoms (General Symptom Survey, Profile of Mood States-Short Form, Neurotoxicity Rating Scale). In addition, questionnaires were included to address body image and quality of life due to our interest in assessing the relationship between systemic symptoms and these outcomes.
### Vanderbilt Head and Neck Symptom Survey version 2.0 plus General Symptom Survey (VHNSS v2.0 plus GSS) {#FPar4}
The VHNSS v2.0 \[[@CR17]\] assesses the prevalence and severity of treatment-related symptoms and their functional impact in patients with head and neck cancer. The VHNSS v2.0 consists of 50-items within 13 domains including nutrition, swallowing, xerostomia, mucositis, excess mucus, speech, hearing, taste change, smell, dental health, mucosal sensitivity, range of motion, and pain. Items are scored on a numeric scale rating the severity of the symptom from 0 (none) to 10 (severe). The VHNSS v2.0 takes approximately 10 min to complete. Cronbach's alpha are \> 0.9 in six symptom clusters and \> 0.7 in the four remaining clusters \[[@CR17], [@CR18]\]. The GSS includes 11 additional items directed toward the systemic effects of cancer and therapy. Items are scored on a scale of 0 (none) to 10 (severe). The General Symptom Survey was specifically developed to assess the systemic symptoms associated with head and neck cancer and its therapy through review of the systemic symptom literature, patient interviews, and expert panel review. Content validity is being tested in an accompanying qualitative analysis to be published separately. Systemic symptoms investigated in this report include the items in the GSS plus two items from the VHNSS ("weight loss" and "loss of appetite").
### Profile of Mood States-Short Form {#FPar5}
The Profile of Mood States-Short Form (POMS-SF) is a psychological evaluation tool used to assess mood states. This tool contains a 37-item scale consisting of adjectives rated on a 5-point Likert-like scale. It is composed of six subscales: depression (maximum possible score 28), vigor (maximum possible score 20), confusion (maximum possible score 20), esteem-related affect (24), tension (maximum possible score 24), anger (maximum possible score 24), and fatigue (maximum possible score 20). Cronbach's alphas range from 0.78 to 0.91 \[[@CR19]\].
### Neurotoxicity Rating Scale {#FPar6}
The Neurotoxicity Rating Scale (NRS) is a self-report measure examining neurocognitive symptoms associated with neurotoxicity of medical treatment. Its 37 items are symptoms rated in severity using a 5-point Likert-like scale bounded by "not present" and "extremely severe" \[[@CR20]\]. Seven items from the NRS (restlessness, no interest in people, distractibility, irritability, decreased motivation, tension, and slowed movements) were chosen for inclusion in this analysis. Although the NRS has not been validated in the oncologic population, the selected items address unique symptoms that may be related to neuroinflammation and are absent in the other tools.
### Body image quality of life inventory {#FPar7}
The body image quality of life inventory (BIQLI) is a 19-item instrument which was developed to quantify the effects of body image on various experiences and life contexts \[[@CR21]\]. Participants rate the impact of their own body image using a 7-point bipolar scale from − 3 to + 3, thereby permitting reports of negative, positive, or no impact \[[@CR22]\]. Overall impact of body image can be determined by averaging the scores of all items. The Cronbach's alpha of the scores in this study was 0.90 \[[@CR21]\].
### Quality of life {#FPar8}
QOL was measured using two scales: a 5-item domain-specific QOL and a single-item self-anchoring scale \[[@CR23], [@CR24]\].
Statistical analysis {#Sec6}
--------------------
SPSS version 24.0 was used for statistical analysis. Frequency distributions were used to summarize nominal data. Median and inter-quartile range (IQR) were used for continuous data summaries due to skewness of those distributions. Spearman's rank correlation was used for determining the strength of the association between two global indicators of QOL. Two-step log-likelihood clustering using Schwarz's Bayesian Criterion (BIC) was used to create groups of patients with similar levels of systemic symptoms.
Results {#Sec7}
=======
Patient and treatment characteristics {#Sec8}
-------------------------------------
The sample was comprised of 105 patients. Available demographic and treatment characteristics are summarized in Table [1](#Tab1){ref-type="table"}. Characteristics with some missing data are indicated in the table. Median age was 62.8 years. A majority of patients were male (74.3%) and Caucasian (94.2%). The most common primary tumor site was the oropharynx (57.1%) and a majority of patients had locally advanced disease, with 9.5% of patients diagnosed with stage III disease and 63.8% of patients with either stage IVa or IVb. A majority of patients received radiation-based therapy (*n* = 94 of 97 cases with information, 96.0%).Table 1Patient demographic and treatment data (*N* = 105)CharacteristicMean (SD) (min, max)Age (*N* = 104)61.9 (9.6) (37, 85)Years of education (*N* = 104)14.0 (2.3) (9, 20)*N* (%)Gender Male78 (74.3) Female27 (25.7)Race (*N* = 103) White97 (94.2) Black/African-American4 (3.9) Other2 (1.9)Location of HNC Oral cavity13 (12.4) Nasopharynx6 (5.7) Oropharynx60 (57.1) Larynx9 (8.6) Other7 (6.7) Unknown10 (9.5)Stage at diagnosis Stages I and II7 (6.7) Stage III10 (9.5) Stage IV67 (63.8) Unknown21 (20.0)Surgery (*N* = 94) No58 (61.7) Yes36 (38.3)*n*, median (IQR)Time since completion of treatment (months)98, 44.4 (18, 70)
Prevalence and severity of systemic symptoms {#Sec9}
--------------------------------------------
The item pool assessed includes the 11 items from the GSS and 2 items from the VHNSS v2.0. The prevalence of systemic symptoms (score \> 0) was high, with 6 out of the 13 items endorsed by more than half of participants: fatigue (52.4%), fatigue limiting day to day activity (50.5%), joint pain/muscle aches (53.3%), problems staying asleep (52.9%), sensation of cold (53.8%), and neurocognitive symptoms (53.4%). Another 4 out of 13 of the systemic symptoms were reported by between one third and one half of participants: anxiety (43.8%), problems falling asleep (41.0%), sensation of warmth (38.1%), and depression (38.1%). Weight loss was the rarest of the systemic symptoms, reported by 19.2% of participants. From the standpoint of intensity, 48.4% of patients experienced at least one systemic symptom of moderate-to-severe degree, as determined by a score ≥ 4.0 out of 10. Therefore, not only were symptoms frequent but also of clinically meaningful intensity.
Clusters of systemic symptoms {#Sec10}
-----------------------------
A cluster analysis of patients with responses for all systemic symptoms (*n* = 95) was undertaken. This analysis resulted in the identification of two unique groups: a *low systemic symptom group* (*n* = 66 of 95, 69.5%) characterized by patients with none or very few moderate-to-severe systemic symptoms (none = 49 of 66, 74.2%, one = 12 of 66, 18.2%, two or three = 5 of 66, 7.6%) and a *high systemic symptom group* (*n* = 29 of 95, 30.5%) characterized by patients with at least two moderate-to-severe systemic symptoms (Fig. [1](#Fig1){ref-type="fig"}). Of note, 17 patients (58.6%) in the high systemic symptom group had 5 or more such symptoms. In the high systemic symptom group, the most common moderate-to-severe symptoms were fatigue (\> 79.3%), difficulty staying asleep (69.0%), and cold (62.1%). Approximately 44.8% also reported moderate-to-severe issues with memory and joint pain or muscle aches.Fig. 1Vanderbilt Head and Neck Symptom Survey version 2.0 plus General Symptom Subscale by low and high systemic symptom patient clusters
Body image {#Sec11}
----------
The median BIQLI score was 1.7 with an IQR of 0.7 to 2.5 indicating generally positive body image in the group.
Neuropsychiatric dysfunction {#Sec12}
----------------------------
The NRS responses confirmed the high prevalence of neuropsychiatric symptoms in this patient population (score \> 0). The results are similar to those noted with the VHSS responses, providing convergent validity. Among the 7 neuropsychiatric symptoms included in this analysis, 6 were reported by between one-third to one-half of participants: restlessness (47.6%), tension (47.5%), decreased motivation (46.7%), distractibility (38.5%), slowed movements (38.5%), and irritability (38.5%). The remaining item, lack of interest in other people, was reported by 21.9% of participants. Among the positive responders, 45% indicated that at least one symptom was moderate-to-severe intensity (score ≥ 2). The highest prevalence reports at the moderate-to-severe level were for slowed movements (20.3%), lack of interest in people (11.3%), distractibility (16.3%), decreased motivation (18.8%), and tension (15.6%).
Summaries of the NRS severity reports for the two systemic symptom cluster groups are illustrated in Fig. [2](#Fig2){ref-type="fig"}. There were striking differences in the prevalence of both mild and moderate-to-severe neuropsychiatric symptoms between the patient cluster with low systemic symptom burden and high systemic symptom burden. Of particular importance were the differences in the prevalence of both mild and moderate-to-severe "decreased motivation" and "slowed movements" between the two groups.Fig. 2Unique items on Neurotoxicity Rating Scale by low and high systemic symptom patient clusters
Correlations among systemic symptom and neuropsychiatric symptom severity {#Sec13}
-------------------------------------------------------------------------
Correlations among the symptom groups are shown (Fig. [3](#Fig3){ref-type="fig"}). Gradations of shading indicate the strength of the correlations with the darker shades reflecting stronger correlations. Almost all correlations showed at least a minimally statistically significant correlation; those that fell below this threshold are depicted by a lack of shading (*r*~s~ \< 0.20, *p* \> 0.05).Fig. 3Correlogram between GSS items and selected NRS items
Mood {#Sec14}
----
Scores on the POMS-SF were generally within a more positive direction. For the negative emotions (fatigue, depression, confusion, tension, anger), most values were within the lower quarter of the possible range of scores. For vigor, the single positive emotion, the scores were in the mid-range of possible scores, see Fig. [4](#Fig4){ref-type="fig"}.Fig. 4POMS-SF
Quality of life {#Sec15}
---------------
Finally, responses to the quality of life measures are summarized by systemic symptom cluster in Fig. [5](#Fig5){ref-type="fig"}. Patients in the high systemic symptom cluster demonstrated lower-median quality of life on all domains except for quality of life as compared to median quality of life for the low systemic symptom cluster. In addition to lower-median QOL, the high systemic symptom patient cluster was noted to have a higher proportion of patients rating quality of life as poor (scores of 0--4), respectively: physical 27.6% vs. 6.1% for the low systemic symptom patient cluster, emotional 27.6% vs. 6.1%, spiritual 6.9% vs. 6.1%, and intellectual 19.2% vs. 6.1%. Poor global quality of life was reported by 20.7% of those in the high systemic symptom cluster as opposed to 1.5% in the low systemic symptom patient cluster. A strong positive correlation was observed between the "global" domains on the QOL measure with the single QOL ladder (*r*~s~ = 0.65, *p* \< 0.001).Fig. 5Single-item quality of life domains and global quality of life by low and high systemic symptom patient clusters
Discussion {#Sec16}
==========
As the number of HNC survivors increases, there has been growing interest in the quality of their survivorship. This includes a wide array of outcomes ranging from symptom burden and functional deficits to overall quality of life. Despite the rising interest in these topics, research in tumor- and treatment-associated late systemic symptoms has received little attention to date. To address this gap, we evaluated the prevalence of chronic systemic symptoms in HNC survivors as well as the association between chronic systemic symptoms and quality of life. Our findings demonstrated that a surprisingly high percent of long-term survivors experienced late systemic effects, with prevalence rates ranging from 25% to 40%. This is particularly striking when considering the population included in this study: HNC survivors at least one year out from completion of all cancer-directed therapy.
These late systemic symptoms are not only common but also severe. Our data demonstrate that approximately half of those reporting these symptoms rated the intensity in the moderate-to-severe range. It is, therefore, not surprising that chronic systemic symptoms were strongly associated with decreased quality of life, with patients in the high systemic symptom cluster rating median quality of life lower than in the low systemic symptom cluster in all domains except for spiritual quality of life. Interestingly, this finding mirrors the existing literature demonstrating associations between the burden of local symptoms and decreased quality of life \[[@CR4]\].
In addition to lower overall median quality of life in the high systemic symptom patient cluster, our data demonstrate that 15% of HNC survivors reported poor global quality of life (rating quality of life ≤ 4/10). This particular cohort is likely unrecognized because much of the existing quality of life literature reports median quality of life \[[@CR4]\]. This approach by its very nature fails to depict the range of participant responses. Furthermore, it fails to identify subsets of patients for who quality of life, symptom burden, or functional deficits are highly problematic. Given that traditional reporting methods for quality of life are frequently insufficient, more comprehensive statistical reporting methods should be used.
The cluster analysis is an innovative statistical approach that allows for the identification of subpopulations with distinct characteristics, including distinguishing symptomatology. By conducting a cluster analysis, we can evaluate and identify cohorts of patients with distinct features. In this study, we performed a cluster analysis that identified two discrete groups of patients: those with low and high levels of chronic systemic symptoms. Through this analysis, it was made clear that majority of patients who reported systemic symptoms experienced more than one. Although the taxonomy of chronic systemic symptoms continues to evolve, this phenomenon is frequently termed "central sensitization." Multiple systemic syndromes are felt to manifest central sensitization, including fibromyalgia, chronic fatigue syndrome, irritable bowel syndrome, and others \[[@CR25]\]. These conditions are marked by a high burden of systemic symptoms and "sickness behaviors" that cluster and give rise to the aforementioned well-defined syndromes. In recent years, there has been increasing interest in the biologic underpinnings of these conditions. Of note is that endocrinopathies are another proposed mechanism and should be ruled out by clinicians encountering patients with significant systemic symptom burden. However, recent data in the HNC population has demonstrated a relationship between fatigue and acute inflammation \[[@CR26]\], lending further support to the theory of an inflammatory mechanism underlying many systemic symptoms. Furthermore, the non-oncologic literature has increasingly demonstrated that deranged interactions between the endocrine, immune, and neurologic systems result in systemic symptoms and sickness behaviors such as fatigue, depression, anorexia, widespread pain, and others \[[@CR10], [@CR11]\]. Therefore, it has been postulated that these symptoms may have a common underlying mechanism, and that the root of which may lie in neuroinflammation.
Neuroinflammation refers to a "cascade of altered neural activity that includes the induction of pro-inflammatory cytokines within the brain and spinal cord," resulting in changes to cognition, pain, and even global function \[[@CR9]\]. The biologic underpinnings of this state are still under investigation, but it has been proposed that this may indirectly result from elevated levels of peripheral (i.e., bloodstream) pro-inflammatory cytokines \[[@CR9]\]. Even though peripheral cytokines cannot passively cross the blood-brain barrier, they can still communicate with the CNS through multiple mechanisms (e.g., active transport), whereby they can cause elevation of the levels of central pro-inflammatory cytokines. Elevations in central pro-inflammatory cytokines are associated with systemic symptoms, such as depression, anxiety, depression, cognitive changes, pain, and others \[[@CR10], [@CR11]\]. Owing to the inherent plasticity of the central nervous system, central neuroinflammation may result in irreversible changes to neural pathways \[[@CR27]\].
Neuroinflammation and its effects are relevant not only to the phenomena of interest in this study (chronic systemic symptoms) \[[@CR28]\] but also to the HNC patient population itself. There is mounting evidence that patients with HNC are at particular risk for neuroinflammation: HNC cell lines have been noted to release elevated levels of inflammatory mediators, and radiation to HNC in vivo has been also associated with increased levels of pro-inflammatory cytokines \[[@CR12], [@CR29]\]. These data suggest that both HNC tumors and their treatment can both contribute result in peripheral inflammation. Given the evidence that elevations in peripheral pro-inflammatory cytokines can lead to neuroinflammation and systemic symptoms, including in the HNC population \[[@CR9]\], it would be reasonable to deduce that HNC and radiation-based treatment lead to neuroinflammation, but further research is needed to establish the relationship between HNC, radiation-based therapy, and neuroinflammation.
Systemic symptoms have been extensively reported in the general cancer population as well as specific cancer populations. Most reports address patients undergoing active treatment; however, increasing attention is being paid to the survivor population. Much of the literature describing chronic systemic symptoms in cancer survivors focuses on individual symptoms such as fatigue, depression, or neurocognitive changes, all of which have been well documented \[[@CR30]--[@CR32]\]. For example, neurocognitive effects of treatment have been extensively described in the breast cancer patient population \[[@CR30], [@CR33]--[@CR35]\]. In addition to descriptive studies, multiple prospective, randomized studies have been undertaken employing physical \[[@CR36], [@CR37]\] and cognitive interventions \[[@CR35]\], with varying results. Similarly, chronic fatigue has been documented in most cancer populations. For example, in a recent study of 275 rectal cancer survivors, 87% complained of "feeling worn out," 85% complained of feeling "tired," and 66% reported difficulty with sleeping \[[@CR38]\]. To better understand the relationship between individual symptoms and underlying pathophysiology, a smaller but critical literature has emerged, exploring the clustering of systemic symptoms. For example, in a study of 74 newly diagnosed stage I to III breast cancer patients, using Bayesian network methods, investigators were able to identify that cognitive function was significantly influenced by sleep deprivation, thereby identifying a potential interventional target \[[@CR39]\]. Select systemic symptoms, such as temperature dysregulation, remain poorly studied.
There are two unexpected findings from our study. The first of these was that patients expressed a positive body image. At first glance, this finding might seem counterintuitive, as one could expect that disfigurement associated with cancer and its therapy would create a negative body image. However, recent qualitative data indicate that HNC patients and survivors make efforts to disassociate themselves from the physical manifestations of their cancer, concentrating instead on functionality \[[@CR40]\]. In this manner, a neutral or positive body image could be seen as part of a positive coping strategy. An alternative explanation for this finding could be that contemporary surgical and radiation techniques result in a lesser degree of damage to healthy tissues. For example, patients treated with radiation therapy with current techniques often note changes in the contour of their soft tissues without overt disfiguration.
An additional noteworthy finding was that the median score on the POMS-SF Vigor subscale revealed the HNC survivor population to be less energetic than the general population of cancer patients. The median vigor score of 10 in our study population indicates that the HNC survivors experience more lethargy and lack energy. Creating a point of reference for scores on the vigor score can be challenging. In one study in the oncologic population, the median vigor scores for depressed vs. non-depressed cancer patients were 9.98 and 14.37, respectively \[[@CR41]\]. The impact of lethargy and motivation, a distinct but related construct, on function and quality of life warrants more extensive investigation.
While novel, this study is not without limitations. First, the cross-sectional nature of this study prevents study of the trajectories of chronic systemic symptoms. Therefore, further study is warranted to establish the trajectories of chronic systemic symptoms in the HNC population. This study also did not have a control population; however, the broad inclusion criteria effectively allowed for the approximately 60% of participants without chronic systemic symptoms to serve as a control group for those without. Despite the association between chronic systemic symptoms and quality of life, this study was not able to truly bring to light the impact of chronic systemic symptoms on HNC survivors. To that end, this study was paired with a qualitative analysis regarding impact, the data from which will be reported elsewhere the results of which are published separately. Finally, the NRS tool has yet to be validated in the oncologic population, thus further study is needed to validate it in this population; a cluster analysis could be beneficial to establish subscales.
Our study demonstrates the high prevalence and severity of chronic systemic symptoms in HNC survivors and the association between high systemic symptom burden and poor quality of life. Validation of these findings is needed. Optimally, prospective studies should be undertaken using patient-reported outcomes to capture systemic symptoms, objective measures of physical and cognitive function, and inflammatory biomarkers assessed at key time points along the treatment trajectory (baseline, during, and post-treatment). In addition, information regarding the pathobiologic underpinnings of systemic symptoms continues to emerge from the general medical literature. This work may inform ongoing and future work in the oncology population. A better understanding of mechanisms is necessary before interventional trials can be developed. In the meantime, HNC patients and their caregivers would benefit from pre- and post-treatment counseling regarding the chronic systemic symptoms.
HNC
: Head and neck cancer
CSS
: Chronic systemic symptoms
VHNSS-GSS
: Vanderbilt Head and Neck Symptom Survey with General Symptom Subscale
POMS-SF
: Profile of Mood States-Short Form
NRS
: Neurotoxicity Rating Scale
QOL
: Quality of life
**Publisher's note**
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
The authors declare that they have no conflict of interest.
| {
"pile_set_name": "PubMed Central"
} |
Introduction {#sec1-1}
============
^131^I-NaI has been used very effectively over several years for both the diagnosis and therapy of differentiated thyroid carcinoma (DTC). ^131^I whole body scintigraphy has been used for detection of remnant, residual as well as metastatic disease in patients of DTC. Though highly specific, ^131^I whole body scans also show a number of false positive accumulations, while not many. It is essential to correctly identify these false positive lesions at the earliest in order to avoid subsequent unnecessary radioiodine treatment in these cases. Here, we present a case series of three cases of unusual false positive radioiodine uptake on whole body iodine-131 (^131^I scan) in three different organs and three different pathologies that demonstrated this uptake.
Case Reports {#sec1-2}
============
Case 1 {#sec2-1}
------
A 36-year-old female, with diagnosis of multifocal papillary carcinoma thyroid (initial presentation with focal hypoechoic space-occupying lesion (SOL) in the right lobe (1.7 cm × 1.2 cm) with vascularity within, had undergone total thyroidectomy and treatment with 54 μCi of ^131^I subsequently. Posttherapy scan showed multifocal iodine-avid foci in the neck. Large dose scan with ^131^I \[[Figure 1a](#F1){ref-type="fig"}\] after 6 months of therapy showed an iodine focus in the right side of the chest \[thyroglobulin (Tg) and thyroid stimulating hormone (TSH) values mentioned in [Table 1](#T1){ref-type="table"}\]. Chest x-ray was normal and as Tg was low; this iodine-avid focus in the right lung was investigated further to rule out a false positive etiology. Fluorodeoxyglucose-positron emission tomography/computed tomography (FDG-PET/CT) was done at that time, which showed a very low grade FDG uptake in a nonactive fibrotic lesion in the right lung on computed tomography (CT), corresponding to the lesion seen in the iodine scan. The repeat diagnostic radioiodine scan was repeated after 1 year, which again showed persistence of the same lesion in the right lung (Tg and TSH values mentioned in [Table 1](#T1){ref-type="table"}). Correlative SPECT-CT images \[[Figure 1b](#F1){ref-type="fig"} and [c](#F1){ref-type="fig"}\] in the lung and mediastinal window and FDG-PET/CT \[[Figure 1d](#F1){ref-type="fig"}\] showed a spiculated nonehnacing lesion with adjacent area of fibrosis in the posterior segment of the right upper lobe and 1 cm-sized round node in the subcarinal region. CT features of the lung lesion on the right were in favor of benign pathology (infective sequela).
{#F1}
######
Serum Tg and TSH values of Patient 1 at various time course
Case 2 {#sec2-2}
------
A 38-year-old female, with diagnosis of multicentric follicular variant of papillary carcinoma thyroid who had prior treatment with total thyroidectomy and 49 μCi of ^131^I, showed iodine-avid foci in the neck and mediastinum in the posttherapy scan. Large dose scan with ^131^I \[[Figure 2a](#F2){ref-type="fig"}\] after 6 months therapy showed an iodine foci in the neck with neck uptake being 0.27% and iodine focus in the right lower chest region (Tg and TSH values mentioned in [Table 2](#T2){ref-type="table"}). SPECT-CT was done, which showed the focus in segment VIII of the liver that was hypodense on CT \[[Figure 2b](#F2){ref-type="fig"}\]. PET-CT with FDG \[[Figure 2c](#F2){ref-type="fig"}\] was done at that time, which showed non-FDG avid hypodense segment VIII liver lesion. The ultrasonography (USG) showed a well-defined lesion with posterior acoustic enhancement and no intralesional vascular flow on Doppler imaging. The T2-weighted (T2W) images revealed a well-defined hyperintense lesion, which did not show any enhancement on the postcontrast T1-weighted (T1W) magnetic resonance imaging (MRI) \[[Figure 2d](#F2){ref-type="fig"}-[f](#F2){ref-type="fig"}\]. All these imaging features were suggestive of simple cyst of the liver. Thus, with low Tg and other modalities ascertaining benign etiology, the cystic liver lesion was considered as a false positive.
![The diagnostic 131I scan (a) showing iodine-avid focus in segment VIII of liver; Correlative SPECT CT \[a\] and PET CT images (c) in abdomen window shows a well-defined round hypodense lesion in segment VIII of the liver, which showed iodine avidity on SPECT scan. T2W image (d) showing well-defined hyperintense lesion (star) in segment VIII with no significant enhancement in postcontrast fast spoiled gradient recalled (FSPGR)-echo images (e). Diffusion MRI (f) showed facilitated diffusion with high apparent diffusion coefficient (ADC) values. The correlated USG (g) shows anechoic lesion with posterior acoustic enhancement (arrowhead). All these imaging features were suggestive of simple cyst of the liver](WJNM-15-137-g003){#F2}
######
Serum Tg and TSH values of Patient 2 at various time course
Case 3 {#sec2-3}
------
A 36-year-old female with history of follicular variant of papillary carcinoma thyroid for which she was treated with total thyroidectomy and 49 μCi of ^131^I postoperatively, demonstrated an iodine-avid foci in the left side of the chest on large dose scan with ^131^I after 6 months of therapy (Tg and TSH values mentioned in [Table 3](#T3){ref-type="table"}). The chest x-ray was normal and as Tg was low, this iodine-avid focus in the left chest region was investigated further. SPECT-CT showed iodine-avid foci in the left breast \[[Figure 3a](#F3){ref-type="fig"} and [b](#F3){ref-type="fig"}\]. Whole body FDG-PET/CT was normal. Mammogram as well as USG of both the breasts showed no focal abnormality \[[Figure 3c](#F3){ref-type="fig"}\]. The patient retrospectively gave a past history of fibroadenoma excision from the left breast undertaken previously. The findings of low Tg as well as that of other imaging modalities showed normal glandular architecture of the left breast; the left breast lesion was a false positive lesion.
######
Serum Tg and TSH values of Patient 3 at various time course
{#F3}
Discussion {#sec1-3}
==========
^131^I-NaI is considered as a very specific radiopharmaceutical with respect to detecting thyroid pathology, particularly in patients of DTC following thyroidectomy. The major mechanisms considered important for radioiodine uptake are:
1\) Functional sodium/iodide symporter (NIS) expression (in normal tissues, including thymus, breast, salivary glands, and gastrointestinal tract, or various benign and malignant tumors of these organs), 2) Metabolism of radioiodinated thyroid hormone, 3) Retention of radioiodinated body fluids (saliva, tears, blood, urine, exudate, transudate, gastric, and mucosal secretions, etc.) associated with or without structural change, 4) Retention and uptake of radioiodine in inflamed tissue, 5) Contamination by physiologic secretions, and 6) Other unknown factors.\[[@ref1]\]
Various authors have over the years reported the importance of identification of false positives in iodine scans in the form of case reports,\[[@ref2]\] where false positive findings have been reported in various organs of the body. False positive radioiodine uptake in the chest has been reported due to acute respiratory infection,\[[@ref3][@ref4]\] pulmonary tuberculosis,\[[@ref5]\] pulmonary aspergilloma,\[[@ref6]\] rheumatoid lung disease,\[[@ref7]\] and bronchiectasis.\[[@ref8]\] It is essential to identify these false positives as pulmonary metastasis and this can be commonly encountered in cases of DTC. Also, metastasis in the ribs is another such entity in DTC. One uncommon metastasis seen in the thoracic region is cardiac metastasis.\[[@ref9]\] Radioiodine uptake in the breast is also not very uncommon. The important causes are lactating breasts, breast fibroadenoma,\[[@ref10]\] breast carcinoma,\[[@ref11]\] and breast cyst.\[[@ref12]\] No case of metastasis to the breast from DTC has been reported. Thus, the cause of radioiodine uptake in breasts could be either physiological (lactating breasts) or pathological (abovementioned causes). Radioiodine uptake in cystic lesions is reported. It is seen in ovarian dermoid or struma ovarii,\[[@ref13]\] nabothian cysts,\[[@ref14]\] renal cysts,\[[@ref15]\] ovarian cysts,\[[@ref16]\] bronchogenic cysts,\[[@ref17]\] and pleuropericardial cysts.\[[@ref18]\] Till date, there is no literature regarding radioiodine uptake in case of simple cysts in the liver. False positive radioiodine uptake in cystic liver disease is reported in cases of hydatid liver disease\[[@ref19]\] and hemangioma of the liver. Liver metastasis, though uncommon, can be encountered at times in some cases of DTC. So, focal iodine-avid lesions in the liver require careful assessment to distinguish between metastasis and false positives.
Along with whole body ^131^I scan findings, various other factors need to be considered before ascertaining a lesion to be metastasis from DTC. The single most important factor to be considered in this is the Tg level. The Tg is expected to be raised in case of systemic metastatic lesions.\[[@ref20]\] If the serum Tg is very low/undetectable in DTC, the organ lesion in question is metastatic. Along with Tg, it is essential to find out antithyroglobulin antibodies (antiTgAb) to ascertain any interference. In all the three cases that we have reported, Tg was very low with the antiTgAb being negative that ruled out any interference.
In cases of discordant findings of low Tg but positive ^131^I whole body scan, SPECT-CT should be strongly considered where available. SPECT-CT has been documented to be of incremental value in the localization of iodine-avid foci in these cases where Tg is low and there is a high suspicion of false positives. In all the three cases reported, SPECT-CT played a pivotal role for the exact localization of the iodine-avid foci in the three organs. Also, the use of other correlative imaging modalities like MRI, PET-CT, and USG should be considered when in doubt. All these noninvasive modalities are of added importance in determining the exact nature of such lesions when invasive biopsy cannot be done or is refused by the patient. In all the three cases, these ancillary imaging modalities were used and were helpful in arriving at the diagnosis.
Conclusion {#sec1-4}
==========
In the identification of false positives, ^131^I whole body scan is crucial in the management of cases of DTC. It should be considered in discordant cases where the iodine scan is positive but Tg is low. Past history is very important in such cases and can give a clue to the origin of the false positive focus. SPECT-CT should be carried out in all such cases as it provides a definite incremental value. Other imaging modalities should be considered in these cases to ascertain whether the iodine-avid foci are metastases from DTC or not.
**Source of Support:** Nil.
**Conflict of Interest:** None declared.
| {
"pile_set_name": "PubMed Central"
} |
Introduction {#Sec1}
============
NK cells represent a subset of lymphoid cells that are components of innate immunity acting as first line of defense against viral infection and other pathogens, and in the early cellular transformation and tumor surveillance^[@CR1]^. The functions of NK cells are partly regulated by the family of KIR receptors (killer cell immunoglobulin-like receptor) coded by 11 genes (*2DL1*, *2DL2/2DL3*, *2DL4*, *2DL5*, *2DS1*, *2DS2*, *2DS4*, *2DS3/2DS5*, *3DL1/3DS1*, *3DL2 and 3DL3*) and two pseudogenes (*2DP1* and *3DP1*) located on the chromosome 19q13.4^[@CR2]--[@CR4]^. Some of these genes may present sequence variations; thus, it has been reported a 22 bp deletion in the second extracellular domain of *2DS4* that affect substantially the sequence of amino acids, whereas the exon 2 can be absent in *3DP1*^[@CR5]^. Also, it has been found that *2DL5* gene is encoded by different loci (A and B)^[@CR6]^. The KIR family is primarily expressed on NK cells, but they can also be expressed on CD4, CD8 and γδ T cells^[@CR7]--[@CR9]^. There are four promoter types based on intermediate promoters (ProI), which are associated with distinct expression in KIR genes, thus *2DL2*, *2DS2 and 2DL3* are the first to be expressed by NK cells after Hemopoietic Stem Cell Transplantation^[@CR10]^. The *3DL3* is not expressed by circulating CD56 dim NK cells, and *2DL4* is expressed by CD56-bright and dim NK cells in a non-variegated manner; and finally, the remaining KIR genes are expressed by CD56-dim NK cells^[@CR10],[@CR11]^. T cells express *3DL2* more than other KIR genes, probably as a result of ProI activation earlier in the development of T cell^[@CR10]^. In addition, the KIR gene family has bi-directional promoters, which control variegated expression, whereas ProI correlates with protein expression^[@CR10]^.
Composition of KIRs may be complex, thus, two haplotypes (A and B) and genotypes (AA and Bx, where x can be A or B) have been reported for KIR based on gene content^[@CR12]^ (Fig. [1](#Fig1){ref-type="fig"}). Actually, there are over 500 different Bx genotypes (<http://www.allelefrequencies.net>). KIR genotype AA is homozygous for the A haplotype, which is an inhibitory haplotype formed by the loci *3DL3*, *2DL3*, *2DP1*, *2DL1*, *3DP1*, *2DL4*, *3DL1*, *2DS4* and *3DL*2; whereas Bx genotype has either one (AB heterozygous) or two (BB homozygous) B haplotypes, and is an activator haplotype (formed by *3DL3*, *2DS2*, *2DL2*, *2DL5B*, *2DS3/2DS5*, *2DP1*, *2DL1*, *3DP1*, *2DL4*, *3DS1*, *2DL5A*, *2DS3/2DS5*, *2DS1*, *2DS4* and *3DL2* genes)^[@CR13]^. The A haplotype usually has a fixed number of genes, while B has a variable gene content with additional activating KIR genes. KIR haplotypes consists of two regions, the centromeric region from *3DL3* to *3DP1*, and the telomeric region from *2DL4* to *3DL2;* and both regions can be cenA or cenB, and telA or telB depending on the haplotype^[@CR13],[@CR14]^. *2DL5*, *2DS3* and *2DS5* genes have been identified in centromeric and/or telomeric region^[@CR14]^. Based on the gene content 9 centromeric regions (cA01, cA02, cA03, cB01, cB02, cB03, cB04, cB05 and cB06) and 8 telomeric regions (tA01, tB01, tB02, tB03, tB04, tB05, tB06 and tB07) have been described^[@CR14]--[@CR17]^. KIR B haplotype can also be classified according to B content genes, and B content score is calculated by adding the number of cenB and/or telB motifs in each genotype^[@CR18]^.Figure 1Composition of KIR haplotypes (**A** and **B**) based on gene content. KIR genes are tightly organised head-to-tail over approximately 150 kb within the Leukocyte Receptor Complex (LCR). Inhibitory KIR genes are shown in white, activating genes in black and pseudogenes in gray.
*Helicobacter pylori* (*H*. *pylori*) infects the gastric mucosa of over 50% of the world population and represents the main risk factor to develop gastric cancer (GC) and duodenal ulcer^[@CR19],[@CR20]^. Different immune cells are involved in the development of gastric pathologies by causing a chronic, unregulated mucosal inflammation^[@CR21]^. Thus, cells of the innate and adaptive system such as T lymphocytes and natural killer (NK) cells have a critical role in the regulation of the immune response^[@CR21]^. *H*. *pylori* causes an increase of NK cells in the gastric mucosa, where they produce TNF-α and INF-γ^[@CR21]--[@CR23]^, and have an important role in the inflammatory process that drive tissue damage. In this work we aimed to study polymorphisms in KIR receptors genes and identify any possible association with GC.
Results {#Sec2}
=======
The characteristics of the population studied are described in Table [1](#Tab1){ref-type="table"}. We studied two groups of patients, one with a diagnosis of NAG and the other with GC, formed by 130 and 112 patients respectively, and both were compared with an asymptomatic group (n = 146). It can be observed that the NAG group showed significantly higher seroprevalence to *H*. *pylori* and to CagA, with an OR of 3.23 and 2.74, respectively, as compared to the asymptomatic patients.Table 1Characteristics of the groups of patients studied.Diagnosis groupn^a^Age, years, median (IQR)^b^Female/Male*H*. *pylori* +/− (%)OR (95% CI)^c^CagA + /− (%)OR (95% CI)^c^Asymptomatic14641 (36--50)54/92109/37 (74.7)74/72 (50.7)Non-atrophic gastritis13047 (39--57)93/37111/19 (85.4)^d^3.23 (1.39--7.54)89/41 (68.5)^d^2.74 (1.39--5.42)Gastric cancer11259 (49--72)49/6387/25 (77.7)2.56 (0.93--7.02)65/47 (58)2.27 (0.99--5.21)^a^n = number of subjects. ^b^IQR = interquartile range. ^c^OR estimated using the asymptomatic group as the reference and adjusted by age and gender. ^d^*p* \< 0.05.
KIR genes {#Sec3}
---------
In order to characterize the KIR genotype frequencies in the study groups, genomic DNA was isolated from peripheral blood leukocytes, and the KIR genes responsible for the activating signals (*2DS1*, *2DS2*, *2DS3*, *2DS4*, *2DS5*, *3DS1*), the inhibitory signals (*2DL1*, *2DL2*, *2DL3*, *2DL4*, *2DL5*, *3DL1*, *3DL2*, *3DL3*), and the two pseudogenes (*2DP1* and *3DP1*) were genotyped using single specific primer-polymerase chain reaction (SSP-PCR). KIR genotypes were assembled according to the presence or absence of each gene locus.
The frequency data obtained was analyzed between groups to determine differences in KIR genes between asymptomatic and disease groups. The framework genes of centromeric (*3DL3* and *3DP1*) and telomeric (*2DL4* and *3DL2*) regions were present in 100% of the three groups studied. The frequencies of *2DL1* (99.3%, 99.2% and 100%), *2DL3* (97.9%, 96.9% and 92.9%), *2DS4* (93.8%, 100% and 88.4%) and *2DP1* (99.3%, 99.2% and 100%) genes were not statistically different among the groups (Asymptomatic, NAG and GC, respectively). The KIR genes with a significant association with disease are shown in Table [2](#Tab2){ref-type="table"}. When compared with healthy controls, most of the activating and inhibitory genes studied were found significantly associated with either NAG or GC, whereas some showed an increasing tendency of association from NAG to GC, like *2DS1* (OR of 2.56 to OR 5.45), *3DS1* (OR of 3.52 to OR 4.75) and *2DL5* (OR of 3.77 to OR 6.21). In contrast, *2DS3* presented a significantly decreasing tendency of association from NAG to GC, (OR of 187.7 to OR 24.98). *2DS2* and *2DL2* showed significant association with NAG (OR of 3.75 for the two genes), whereas *3DL1* was associated with protection for GC (OR 0.61). The above associations remained significant in a multivariate model of analyses (Table [3](#Tab3){ref-type="table"}), where it can be observed that age was constantly associated with risk for GC, and *H*. *pylori* and gender for NAG. The multivariate analyses showed that *2DS1*, *2DS3*, *2DS5*, *3DS1* and *2DL5* were found as risk for GC, whereas *3DL1* was found as protective (OR 0.21) (Table [3](#Tab3){ref-type="table"}).Table 2Distribution of KIR genes giving significant differences between clinical groups.GenesAsymptomaticsNon-atrophic gastritisGastric cancern = 146^a^ (%)n = 130^a^ (%)OR (95% CI)^b^n = 112^a^ (%)OR (95% CI)^b^Activating 2DS1^d^56 (38.4)72 (55.4)2.56 (1.35--4.85)79 (70.5)5.45 (2.20--13.54) 2DS248 (32.9)70 (53.8)3.75 (1.87--7.52)35 (31.3)NS^c^ 2DS3^d^21 (14.4)113 (86.9)187.78 (43.83--804.46)50 (44.6)24.98 (5.85--106.61) 2DS546 (31.5)18 (13.8)0.20 (0.081--0.48)55 (49.1)3.16 (1.31--7.58) 3DS1^d^55 (37.7)76 (58.5)3.52 (1.80--6.86)76 (67.9)4.75 (1.97--11.50)Inhibitory 2DL248 (32.9)70 (53.8)3.75 (1.87--7.52)36 (32.1)NS^c^ 2DL5^d^62 (42.5)86 (66.2)3.77 (1.92--7.38)81 (72.3)6.21 (2.41--15.98) 3DL1139 (95.2)130 (100.0)NS^c^99 (88.4)0.61 (0.008--0.45)^a^n = number of subjects. ^b^Comparisons were made using the asymptomatic group as the reference group, *pc* \< 0.05 and OR (95% C.I.) were adjusted by age and gender. ^c^NS = not significant. 2DS2 = 1.13 (0.46--2.72), 2DL2 = 1.13 (0.46--2.72. ^d^Linear trend analysis for 2DS1, 3DS1 and 2DL5 genes from asymptomatic to non-atrophic gastritis to gastric cancer (P \< 0.00001).Table 3Multivariate logistic regression analysis of KIR genes associated with gastric disease.GeneVariableNon-atrophic gastritisGastric cancer*p*OR (95% CI)^a^*P*OR (95% CI)^a^2DS12DS1^b^0.0052.14 (1.25--3.65)\<0.00013.41 (1.87--6.22)*H*. *pylori* +0.0382.07 (1.04--4.10)NS ^c^male\<0.00010.23 (0.13--0.39)NS≥50 years0.0132.01 (1.16--3.47)\<0.00018.09 (4.48--14.62)2DS22DS2^b^0.0012.71 (1.57--4.67)NS*H*. *pylori* +0.0162.33 (1.17--4.64)male\<0.00010.22 (0.13--0.38)≥50 years0.0142.01 (1.15--3.51)2DS32DS3^b^\<0.000132.38 (15.9--65.7)\<0.00014.66 (2.34--9.26)*H*. *pylori* +NSNSmale0.0070.37 (0.20--0.80)NS≥50 yearsNS\<0.00018.57 (4.70--15.66)2DS52DS5^b^0.0060.40 (0.21--0.80)0.0082.25 (1.24--4.10)*H*. *pylori* +0.0142.32 (1.18--4.56)NSmale\<0.00010.27 (0.16--0.46)NS≥50 years0.0092.08 (1.20--3.61)\<0.00018.90 (4.97--15.94)3DS13DS1^b^0.0012.49 (1.45--4.27)\<0.00013.35 (1.83--6.13)*H*. *pylori* +NSNSmale\<0.00010.23 (0.13--0.39)NS≥50 years0.0112.04 (1.17--3.56)\<0.00018.37 (4.63--15.13)2DL22DL2^b^\<0.00012.75 (1.60--4.76)NS*H*. *pylori* +0.0132.40 (1.20--4.40)male\<0.00010.22 (0.13--0.38)≥50 years0.0132.02 (1.16--3.53)2DL52DL5^b^\<0.00012.97 (1.72--5.12)\<0.00013.60 (1.94--6.67)*H*. *pylori* +0.0372.10 (1.04--4.18)NSmale\<0.00010.23 (0.13--0.40)NS≥50 years0.0062.21 (1.26--3.87)\<0.00018.75 (4.82--15.91)3DL13DL1^b^NS0.0050.21 (0.07--0.63)*H*. *pylori* +NSmaleNS≥50 years\<0.00019.93 (5.46--18.05)^a^Comparisons were made with the asymptomatic as the reference group. ^b^Adjusted by the other independent variables. ^c^NS = not significant.
KIR haplotypes {#Sec4}
--------------
In order to determine the association of genotypes with the disease groups, the genotypes were grouped as A or B based on gene content. We found that the genotype AA (inhibitory) was highly protective for NAG (OR 0.04, 95% CI 0.01--0.12) and GC (OR 0.22, 95% CI 0.08--0.57), whereas the genotype Bx (activator) was found as risk for both diseases (NAG OR 26.35, 95% CI 8.26--84.11; GC OR 4.57, 95% CI 1.74--12.0). Also, the results show that in our population the frequency of the genotype AA was close to 44%, while genotype Bx was 56%. The multivariate analyses confirmed the association with protection for genotype AA and with risk for genotype Bx (Table [4](#Tab4){ref-type="table"}). As before, in this multivariate analyses gender was associated with NAG and age with GC.Table 4Multivariate logistic regression analysis of distribution of KIR genotype and variables in KIR centromere-telomere and its association with gastric disease.GenotypeVariableNon-atrophic gastritisGastric cancerPOR (95% CI)^a^pOR (95% CI)^a^AAAA^b^\<0.00010.073 (0.03--0.18)\<0.00010.23 (0.12--0.47)*H*. *pylori*NS^c^NSmale\<0.00010.22 (0.12--0.40)NS≥50 years0.0361.90 (1.04--3.45)\<0.00018.18 (4.52--14.91)BxBx^b^\<0.000113.62 (5.68--32.61)\<0.00014.2 (2.11--8.49)*H*. *pylori*NSNSmale\<0.00010.22 (0.12--0.40)NS≥50 years0.0361.90 (1.04--3.45)\<0.00018.18 (4.52--14.91)**Centromere-TelomereVariablePOR (95% CI)** ^**a**^**pOR (95% CI)** ^**a**^cAcA-tAtAcAcA-tAtA^b^\<0.00010.073 (0.03--0.18)\<0.00010.23 (0.12--0.47)*H*. *pylori*NS^e^NSmale\<0.00010.22 (0.12--0.40)NS≥50 years0.0361.90 (1.04--3.45)\<0.00018.18 (4.52--14.91)cBx-tBxcBx-tBx^b^\<0.00017.08 (3.80--13.17)0.0082.55 (1.27--5.10)*H*. *pylori*0.0272.28 (1.10--4.73)NSmale\<0.00010.21 (0.11--0.37)NS≥50 yearsNS\<0.00018.73 (4.88--15.61)^a^Comparisons were made with the asymptomatic as the reference group. ^b^Adjusted by the other independent variables. ^c^NS = not significant. AA = homozygote to A. Bx = homozygote or heterozygote to B. c = centromere. t = telomere.
Because the centromeric and telomeric regions have different gene content, genotypes AA and Bx were analyzed based on their distribution. Four different combinations of centromeric-telomeric (cAcA-tAtA, cAcA-tBx, cBx-tAtA and cBx-tBx) region of the genotypes were found. The cAcA-tAtA distribution was found significantly associated with protection (NAG OR 0.04, 95% CI 0.01--0.12; GC OR 0.22, 95% CI 0.08--0.57) and cBx-tBx with risk (NAG OR 10.65, 95% CI 4.84--23.44; GC OR 11.15, 95% CI 3.33--37.30) for both, NAG and GC. We noted that cAcA-tAtA was constant in asymptomatic population (43.8%), whereas cBx-tBx was more frequent in NAG (56.2%, vs 16.4% in asymptomatic and 33% in GC) and cAcA-tBx in GC (42.9%, vs 22.6% in asymptomatics and 33.1% in NAG). The multivariate analyses confirmed the association with protection for cAcA-tA-tA and with risk for cBx-tBx (Table [4](#Tab4){ref-type="table"}).
KIR B score {#Sec5}
-----------
The B content score analyses showed an association with protection of genotype AA for NAG (OR 0.04) and GC (OR 0.22) whereas the score 2 was associated with NAG (OR 9.7), and score 3 with GC (OR 78.7) (Table [5](#Tab5){ref-type="table"}). The above associations were confirmed in a multivariate analysis (Table [6](#Tab6){ref-type="table"}), and the risk of score 3 with GC remained as high (OR 10.9) as with age (OR 9.9).Table 5KIR B content score and its association with gastric disease.GenotypeB scoreCentromere-TelomereAsymptomaticsNon-atrophic gastritisGastric cancern = 146^a^ (%)n = 130^a^ (%)OR (95% CI) ^b^n = 112^a^ (%)OR (95% CI)^b^AA0cAcA-tAtA64 (43.8)7 (5.4)0.04 (0.01--0.12)^c^16 (14.3)0.22 (0.08--0.57)^c^Bx1cAcA-tAtB54 (37)50 (38.5)NS^d^56 (50)NScAcB-tAtA2cAcA-tBtB25 (17.1)71 (54.6)9.7 (4.4--21.5)^c^26 (23.2)NScAcB-tAtBcBcB-tAtA3cAcB-tBtB3 (2.1)2 (1.5)NS14 (12.5)78.7 (4.9--1246.6)^c^cBcB-tAtB4cBcB-tBtB000^a^n = number of subjects; ^b^Comparisons were done using the asymptomatic group as the reference group and OR (95% C.I.) adjusted by age and gender. ^c^*pc* \< 0.05. ^d^NS = not significant. AA = homozygote to A. Bx = homozygote or heterozygote to B. c = centromere. t = telomere. B score is the number of cB and/or tB motifs in each genotype^[@CR18]^.Table 6Multivariate logistic regression analysis of KIR B content score and its association with gastric diseases.B scoreVariableNon-atrophic gastritisGastric cancerpOR (95% CI)^a^POR (95% CI)^a^00^b^\<0.00010.073 (0.03--0.18)\<0.00010.23 (0.12--0.47)*H*. *pylori*NS^c^NSmale\<0.00010.22 (0.12--0.40)NS≥50 years0.0361.90 (1.04--3.45)\<0.00018.18 (4.52--14.91)22^b^\<0.00016.6 (3.5--12.2)*H*. *pylori*0.0132.5 (1.2--5.1)male\<0.00010.21 (0.12--0.38)≥50 years0.0421.8 (1.02--3.3)33^b^0.00110.9 (2.7--43.8)*H*. *pylori*NSmaleNS≥50 years\<0.00019.9 (5.4--17.9)^a^Comparisons were made with the asymptomatic as the reference group. ^b^Adjusted by the other independent variables. ^c^NS = not significant.
KIR centromeric and telomeric distribucion {#Sec6}
------------------------------------------
To further analyze the association of gene content of centromeric or telomeric regions with NAG and GC we studied the different combinations that have been reported. The multivariate analyses showed a strong association of risk for GC with cA01\|cB03 (OR 36.3) and tB01\|tB01 (OR 7.55), and of protection with tA01\|tA01 (OR 0.23) (Table [7](#Tab7){ref-type="table"}). The analysis also showed a strong risk for NAG with tA01\|tB0X (OR 26.04) and high protection with tA01\|tA01 (OR 0.08).Table 7Multivariate logistic regression analysis of distribution of KIR according to centromeric and telomeric gene content.Gene contentVariableNon-atrophic gastritisGastric cancer*p*OR (95% CI)^a^*p*OR (95% CI)^a^cA01\|cA01cA01\|cA01^b^\<0.00010.28 (0.16--0.48)NS^c^*H*. *pylori*0.022.30 (1.14--4.62)male\<0.00010.21 (0.12--0.38)≥50 years0.0142.04 (1.16--3.59)cA01\|cB02cA01\|cB02^b^\<0.00012.83 (1.59--5.03)NS*H*. *pylori*0.0242.20 (1.11--4.38)male\<0.00010.21 (0.13--0.38)≥50 years0.0161.98 (1.14--3.46)cA01\|cB03cA01\|cB03^b^NS0.00136.39 (4.32--306.85)*H*. *pylori*NSmaleNS≥50 years\<0.000110.65 (5.80--19.55)tA01\|tA01tA01\|tA01 ^b^\<0.00010.08 (0.04--0.16)\<0.00010.23 (0.12--0.43)*H*. *pylori*NSNSmale\<0.00010.23 (0.13--0.43)NS≥50 yearsNS\<0.00017.88 (4.32--14.36)tA01\|tB01tA01\|tB01^b^0.0012.51 (1.45--4.35)0.0162.06 (1.15--3.70)*H*. *pylori*NSNSmale\<0.00010.23 (0.14--0.40)NS≥50 years0.0132.02 (1.16--3.50)\<0.00017.90 (4.44--14.07)tA01\|tB0XtA01\|tB0X^b^\<0.000126.04 (7.45--90.97)NS*H*. *pylori*0.0063.0 (1.38--6.54)male\<0.00010.30 (0.17--0.54)≥50 yearsNStB01\|tB01tB01\|tB01^b^NS0.0017.55 (2.31--24.70)*H*. *pylori*NSmaleNS≥50 years\<0.000110.45 (5.70--19.16)^a^Comparisons were made with the asymptomatic as the reference group. ^b^Adjusted by the other independent variables. ^c^NS = not significant. c = centromere. t = telomere. The number was according gene content^[@CR15]--[@CR17]^. 0X = the number had not been assigned so far.
Discussion {#Sec7}
==========
In the present study, we showed the association of KIR receptors family genes with gastric pathologies. This association was observed with presence of genes, with genotypes, with centromere-telomere regions and with B score. Of particular interest, we observed that both activator and inhibitor genes were associated with GC. A balance between activator and inhibitor genes is necessary during immune surveillance by NK receptors; but when unregulated, their activity may contribute to pathogenesis of diverse diseases including tumor development^[@CR24]^. We found that the presence of *2DS1*, *2DS3* and *3DS1* activating genes was associated with risk for both, NAG and GC. In studies in other populations *2DS1* has also been found associated with cancer, although with breast cancer^[@CR25]^, whereas *2DS1*, *2DS3* and *3DS1* have been associated with more severe pulmonary tuberculosis in Asian population^[@CR26]^; *2DS1* and *3DS1* with ankylosing spondylitis in Spain^[@CR27]^, and *2DS2* with rheumatoid arthritis in Latin America^[@CR28]^. In our population the activating *2DS5* gene was found associated with risk for GC, and of note, a recent meta-analysis concluded that *2DS5* was associated with risk for colorectal cancer^[@CR29]^; these results would suggest that *2DS5* may be associated with cancers in the gastrointestinal tract.
Concerning the inhibitory genes, our results were contrasting; whereas *2DL5* was associated with risk for GC, *3DL1* showed association with protection. Interestingly, a study in Chinese population found that the expression of *3DL1* in NK cells was significantly increased in patients with gastric, pancreatic and colorectal cancer, but was not correlated with disease progression^[@CR30]^. The inhibitory genes have also been reported associated with other immune and infectious diseases; *3DL1* in combination with the HLA-B\*57 allele showed a protective effect against progression to AIDS in Zambian patients^[@CR31]^, and *2DL2* was reported associated with rheumatoid arthritis^[@CR28]^.
In our population Bx genotype was frequent and associated with both gastric diseases, although the risk was stronger for NAG than for GC. In our patients Bx genotypes were a combination of A and B haplotypes, with very few B homozygotes, which is in agreement with studies in other human populations in America, including Amerindian groups^[@CR32]^. This high frequency of Bx genotype may have resulted from the selection by the infectious and chronic diseases that have been prevalent in our population for many generations; although this selection process may have caused an increased risk for GC in the region.
Our work shows eight KIR genes associated with gastric diseases, five of them were associated with risk for GC (*2DS1*, *2DS3*, *2DS5*, *3DS1*, and *2DL5*) and belong to B haplotype, which is an activator haplotype. This unexpected association might partially be explained by the pathogenesis of GC^[@CR19]^, which main risk factor is an infection. *H*. *pylori* infection is strongly pro-inflammatory and invariably causes a chronic, decades-long inflammation of the gastric mucosa^[@CR20]^. In the context of a decades-long mucosal inflammation, NK cells may be constantly and chronically recruited and activated; until in some patients the regulation of this activation might be lost.^[@CR21]^ Unregulated NK cell may help to increase inflammation leading to mucosal damage and development of precancerous lesions and eventually to GC^[@CR1],[@CR9],[@CR23],[@CR30]^. On the other hand, activating KIR haplotypes would have opposite effects on distinct malignancies depending on whether inflammation is or is not a major component of tumor pathogenesis^[@CR33]^. Although, it should be noted that an activator haplotype could also be expected to be associated with increased ability to eliminate tumors^[@CR1],[@CR34]^. In fact, it was reported that patients with metastatic colorectal cancer had complete response to FOLFIRI (5-fluorouracil, leucovorin and irinotecan) treatment when B haplotype was present^[@CR35]^. Interestingly, there was a strong association between the KIR B haplotype and p53 alteration in Basal cell carcinoma tumors, with a higher likelihood that KIR B carriers harbor abnormal p53^[@CR34]^.
Concerning the distribution of A and B, we observed that A was more frequent in centromere and telomere of asymptomatic healthy adults, whereas B was more common in both, NAG and GC. The multivariate analysis confirmed a highly significant association of cA01\|cB03 with GC, in fact this association (OR 36.39) was over 3 times higher than the association with age (OR 10.65 for \>50 years old), which was usually the strongest factor. In addition, whereas tB01\|tB01 was also a significant risk factor for GC, tA01\|tA01 showed a significant association with protection. To our knowledge, there is no report describing the analysis of gene content in centromere and telomere regions and gastric cancer. Our analysis also confirms that the telomeric part of the KIR B genotype may have a role in the development of gastric diseases, particularly the cluster of genes *2DS1*, *2DS3*, *2DS5*, *3DS1* and *2DL5*, which showed a risk association with GC. In contrast, genes *2DS2* and *2DL2* were found as risk for NAG; these genes are located in the centromeric region of B haplotype, and have been reported with high linkage disequilibrium^[@CR16]^. Thus, in our population the telomeric region of KIR B was more associated with GC, and the centromeric region with NAG. Within the B haplotype the telomeric region is more diverse, and probably the observed association with GC is due to an unbalanced response by the NK cell and a reduced ability to kill cancer cells. In contrast, the centromeric region is more conserve and the response of the NK cell is probably more balanced and efficient to fight cancer.
In order to determine the participation of B motifs in GC, we evaluated the B score in centromere and telomere, and the multivariate analysis confirmed its importance in GC, showing a trend following the course of the disease. A B-score of zero was more frequent in asymptomatics and strongly protective for GC (OR 0.23), and a score of 2 was associated with NAG, whereas a score of 3 increased the risk for GC almost 11 times (OR of 10.9). These results suggest that the exacerbated function of the B haplotype contributes to the damage of the gastric mucosa, favoring the development of GC. The B-score was previously evaluated in patients with acute myelogenous leukemia, where patients receiving transplant from donors with a B-score of 2 or greater showed a better protection from relapses, and an increased disease-free survival^[@CR18]^; which would suggest that the exacerbated function of B results in an efficient response against leukemia.
On the other hand, the A genotype was associated with protection for GC, although only the *3DL1* gene showed a significant protection. This gene is within the telomeric region of A genotype and probably the observed association could be due to the strong linkage disequilibrium that *3DL1* has with the other genes of the A genotype.
It was recently reported that tumor-infiltrating NK cells were decreased in human GC; moreover, the production of IFNγ and TNFα by these cells was impaired by tumor-associated monocytes/macrophages via TGFβ1^[@CR36]^. In contrast, patients with GC had a better survival when they presented higher concentrations of NK cells, an effect that was more evident in advanced stage cases^[@CR37]^. Our work associates the risk to develop GC with the B KIR genotypes and the gene cluster included within the telomeric part. There is a need to better understand the functional role of the diversity in KIR genes content in GC, together with the participation of other factors involved in GC development, such as peptides derived from cancer that are presented by HLA class I molecules to KIR receptors^[@CR34]^. Since HLA molecules are ligands of NK cells, they regulate the variation in immune responses to different antigens by selection and suppression/activation of NK cells, and have a relevant role in the combat against GC^[@CR9],[@CR38]^. However, for the eight genes that presented association with NAG and GC in our study, only three ligands are known (*2DS1-C2*, *2DL2-C1* and *C2*, *3DL1-Bw4*)^[@CR38]^. In addition, it is known that B allotypes can influence the binding with *3DL1*; thus, in the *Bw4* dimorphic position 80, isoluecine (HLA-B\*51-\*53, \*57, \*58, HLA-A\*24) generally exhibit stronger inhibition than threonine (HLA-B\*13, B\*27, B\*37, B\*44)^[@CR33],[@CR39]^. Besides, *3DL1* and the other KIR genes associated with GC reported in this study are expressed by CD56-dim NK cells, which migrate to acute inflammatory sites and display a higher cytotoxic activity than CD56-bright cells^[@CR40]^, and the B haplotype could also influence the cytotoxic activity on tumor cells. It is necessary to further study the role of HLA-Cw and KIR gene alleles in gastric cancer surveillance since receptor-ligand combinations are important in the regulation of NK cell responses^[@CR38]^.
Although one limitation of our study is the sample size, we were still able to identify a strongly significant risk association of a gene cluster located in telomeric region of B genotype with GC. We should consider that GC is a multifactorial disease and consequently a multivariate analysis is necessary to better understand the importance of KIR gene variants in GC. We acknowledge that whereas our work present evidences of a significant association of KIR gene variants with gastric pathology, this association is not probe of causality and further studies are now needed to show that unregulated NK cells in the stomach mucosa may lead to gastric pathology. In conclusion, we found that *2DS1*, *2DS3*, *2DS5*, *3DS1*, *2DL5*, Bx genotype, cBx-tBx, cA01\|cB03, tA01\|tB01, tB01\|tB01 and B motifs were risk factors for GC. Mutated NK cells may contribute to GC development by increasing gastric mucosa inflammation, leading to constant tissue damage. The impact of the NK cell response on GC control might be determined in part by the genetic variation in KIR genes.
Materials and Methods {#Sec8}
=====================
Study subjects {#Sec9}
--------------
A total of 388 unrelated adults were recruited in this study, 146 healthy individuals (asymptomatic), 130 with non-atrophic gastritis (NAG) and 112 with GC. Patients with NAG were adults over 30 years old who were attended for symptoms at the gastroenterology service, whereas GC patients attended the oncology service for GC treatment; both groups attended the Instituto Mexicano del Seguro Social (IMSS) Medical Center in Mexico City. We selected NAG and GC patients without treatment of antibiotics, bismuth compounds, proton pump inhibitors and nonsteroidal anti-inflammatory drugs for at least two weeks prior to the study. GC patients without previous treatment for cancer were selected. Diagnosis was based on endoscopic examination and histopathology studies^[@CR41]^. The individuals of the asymptomatic group were selected from healthy blood donors who attended the blood bank of the IMSS Medical Center, with an age over 30 years old and without any symptom or medication. To minimize the genomic diversity in different regions of the country of Mexico^[@CR42]--[@CR45]^, all groups of patients included in this study, patients and controls, received medical coverage from the same institute, IMSS at hospitals in the same city, Mexico City. Patients and controls were informed about the nature of the study and those willing to participate were asked to sign an informed consent letter. The study was approved by the ethics committee from the National Council for Research on Health, IMSS, Mexico and all procedures were performed in accordance with relevant guidelines and regulations.
Collection of samples {#Sec10}
---------------------
For the NAG patients seven gastric biopsies were taken and processed for histology to study the presence of precancerous lesions and *H*. *pylori* infection. Biopsies were collected from both the lesser and the greater curvature, four from antrum and three from corpus. Mucosal inflammation was graded according to the Karttunen classification^[@CR46]^, and only patients without precancerous lesions were included. In the case of GC patients, a tissue sample from the tumor lesion and a sample from adjacent non-cancerous tissue was also obtained and the lesion was classified according to the Karttunen classification^[@CR46]^. A sample of 5 ml of peripheral blood was drawn from each patient, and each healthy volunteer, and mononuclear cells were purified by centrifugation through a Ficoll-Hypaque density gradient. DNA was isolated from these cells using the salting-out microtechnique^[@CR47]^ and frozen at −70 °C until genotyping. The serum fraction was frozen at −20°C until tested.
Definition of *H*. *pylori* infection {#Sec11}
-------------------------------------
Serum samples were tested by ELISA to detect IgG antibodies against *H*. *pylori* whole-cell extract and against recombinant CagA protein, as previously described^[@CR48]^. Infection was also diagnosed by histology in both antrum and corpus. The patient was considered infected with *H*. *pylori* when either tests, ELISA and/or histology, were positive, and non-infected when both tests were negative.
Genotyping of KIR {#Sec12}
-----------------
The presence of each *KIR* gene was used to define the *KIR* gene content of patients. *KIR* genes were tested using a commercial kit (Invitrogene, Brown Deer, Wisconsin, USA) based on the technique of single specific primer-polymerase chain reaction (SSP-PCR), which can identify *2DL1*, *2DL2*, *2DL3*, *2DL4*, *2DL5A*, *2DL5B*, *2DS1*, *2DS2*, *2DS3*, *2DS4*, *2DS5*, *3DL1*, *3DL2*, *3DL3*, *3DS1*, *2DP1* and *3DP1* genes (including the variants of *2DS4* and *3DP1*). PCR reaction and cycling conditions were according the instructions recommended by the manufacturer.
Statistical analysis {#Sec13}
--------------------
The gene, genotype, centromere-telomere gene content and B score frequencies in patients with NAG and GC were compared with the asymptomatic group. Chi-squared or Fisher's exact test were used to test differences among groups, using the Epidat 3.1 Software; p values ≤ 0.05 were considered as significant^[@CR49]^. The significance of association was assessed using odds ratios (OR) with confidence intervals (CI) of 95%^[@CR50]^. OR values were corrected for gender and age using a logistic regression model. The analyses were performed using SPSS Statistics 22.0 (IBM SPSS Data Collection). The role of *H*. *pylori*, gender and age as variables influencing risk factor for GC was estimated in a multivariable logistic regression analysis.
Data Availability {#Sec14}
-----------------
The datasets generated and analyzed during the current study are available from the corresponding authors on reasonable request.
Electronic supplementary material
=================================
{#Sec15}
Distribution of KIR according to centromeric and telomeric gene content Supplementary Dataset
**Electronic supplementary material**
**Supplementary information** accompanies this paper at 10.1038/s41598-018-24464-2.
**Publisher\'s note:** Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
The study was supported by Coordinacion Nacional de Investigacion en Salud, IMSS, México, grants FIS/IMSS/PROT/PRIO/13/027 and FIS/IMSS/PROT/130; and the Consejo Nacional de Ciencia y Tecnologia (Fronteras de la Ciencia grant 773), México.
E.G.H. performed K.I.R. genotyping. O.P.R., M.N.R. and I.R.V. participated in the sample collection. O.P.R. and M.C.P. performed seroprevalence of *H*. *pylori*. J.T. conceived and participated in design and coordination and provided helpful discussions and helped to edit the manuscript. M.P.R. conceived and participated in statistical analysis, and in design and coordination and edited the manuscript.
Competing Interests {#FPar1}
===================
The authors declare no competing interests.
| {
"pile_set_name": "PubMed Central"
} |
{#sp1 .890}
| {
"pile_set_name": "PubMed Central"
} |
Introduction
============
Arising from notochord remnants such as the clivus, retropharyngeal space, and the ventral part of the cranio-vertebral junction (CVJ), chordoma has been considered a clinically malignant tumor because of its active biological nature. Maximal resection of the tumor is required because these tumors have high recurrence rates, and controversially, the surgical treatment should be minimally invasive to maintain the patient\'s activities of daily living (ADL).^[@B1]--[@B5]^ Gross total resection is difficult because of tumor extension and surgeons should select the most suitable approach.Historically, various midline anterior approaches to the CVJ have been employed. The transoral approach (TOA) is one of the midline anterior approaches, and provides a direct extradural route without brain retraction around the CVJ. Surgical exposure via the TOA is limited by the extent of the patient\'s mouth opening, and TOA has been modified in various ways to extend the surgical field for caudal and rostral tumor extension. These modifications include the transpalatal approach (TPA), the transmaxillary approach, and the mandibular swing-transcervical approach.^[@B6]^ TPA can offer a TOA surgical field for rostral tumor extension by removal of the hard palate. However, these approaches are invasive and reconstruction takes many hours. The recently developed endoscopic endonasal approach (EEA) is useful for treating lesions not only in the sella turcica but also around the clivus. This approach is considered minimally invasive to neurovascular structures, changing the history of treatment of skullbase tumors.^[@B7]--[@B16]^ However, removal of lesions around the CVJ is difficult even when using the EEA, because the hard palate is located at the same level as the CVJ, and the surgical view of lesions around the CVJ when using the EEA is obstructed by the hard palate.
The present study is a retrospective analysis of chordomas operated in a single institution and resected by EEA or transoral-transpalatal approach (TO-TPA) during the period 2002--2012. The study aimed to compare the surgical results and find a suitable approach for clival and upper cervical chordomas.
Patients and Methods
====================
Between 2002 and 2012, 18 patients---11 females and 7 males---aged 10--78 years (mean 43.5 years), underwent EEA or TO-TPA. Each approach was performed on nine patients. We started using the EEA in conjunction with the Department of Otolaryngology in 2008. Subsequently, the transnasal approach with a microscope was replaced with endoscopy. TO-TPA was performed in conjunction with the Department of Plastic surgery. The patients\' symptoms, upper and lower limits of surgical location, extent of resection, incidence of cerebrospinal fluid (CSF) leakage and complications were analyzed. The upper and lower margins of the tumor were confirmed by pre-operative magnetic resonance imaging (MRI). The degree of resection was classified as gross total removal (GTR, confirmed by surgical observation and MRI after surgery), subtotal removal (STR, more than 90%), or partial removal (PR, less than 90%). Operative time, pre- and post-operative Karnofsky performance status (KPS) scores, and hospitalization were analyzed in both groups. Post-operative KPS was evaluated at discharge. Hospitalization does not include the stay for post-operative radiation therapy.
Operative Procedures
====================
I.. Transoral-transpalatal approach
-----------------------------------
General anesthesia was performed using standard endotracheal intubation. The patient was positioned supine with the head down using a shoulder pad, without head fixation. A 2-cm incision was made in the mucosa of the midline gingivo-buccal sulcus and the anterior nasal spine was exposed. The submucosal nasal floor was dissected along the palate, and the nasal septum cut and freed from the palatal bone. After this procedure, a Crockard retractor was attached to open the mouth.
A U-shaped mucosal incision was made on the palate, along the maxillary alveolar process to the second/third molars ([Fig. 1](#F1){ref-type="fig"}). A midline palatal incision was added along the palatine raphe, sparing the uvula, to form the muco-periosteal flaps, which were supplied by the palatal artery, and were retracted laterally using strings ([Fig. 2](#F2){ref-type="fig"}). The exposed palatal bone was drilled and cut along the margin of the mucosal incision using a chisel, and the horseshoe-shaped palatal bone was removed. By retraction of the muco-periosteal flaps using the Crockard palatal arms, a long surgical field from the posterior nasal cavity to the pharynx was obtained ([Fig. 3](#F3){ref-type="fig"}). In the middle surgical field, the pharyngeal tonsil was observed. The epidural tumor was exposed using a straight or S-shaped pharyngeal incision. The surgical field covered the total length of the clivus, cranio-cervical junction, and C2 body. Bone resection along the tumor margin is required to achieve total resection of the chordoma. In a tumor with subdural invasion, the dural defect was covered with abdominal fat and coated with fibrin glue, and the mucosa was tightly sutured. Cases in which a mucosal defect was present, especially on the pharyngeal tonsil, a septal mucosal flap was used for closure. After tumor removal, the palatal bone was replaced and fixed by suturing the muco-periosteal flap in two layers ([Fig. 4](#F4){ref-type="fig"}). The tracheal tube was removed the next day, but tube feeding was necessary for 1 week after surgery.
II.. Endoscopic endonasal approach
----------------------------------
The surgical approach and tumor resection were performed using a straight 4-mm endoscope; observations were made using angled scopes. An image guidance technology navigation system was also used. Resection of the right middle turbinate was performed and a nasoseptal flap was created. When the tumor extended into the subdural space, the binostril four-hand technique was applied to obtain a wider surgical space, and the posterior segment of the nasal septum was resected. Drilling of the floor of the sphenoid sinus and clivus were performed towards the CVJ, if necessary ([Fig. 5](#F5){ref-type="fig"}). The vidian canal was a good landmark for the carotid arteries. The subdural component was removed carefully because the tumor was noted to involve the basilar arteries and brainstem. Reconstruction was performed using a multilayer approach: the fascia as a subdural inlay, a pedicled nasoseptal flap to cover the fascia, and oxidized cellulose with fibrin glue at the dural and osseous edges. Pressure was applied using a 14-F Foley catheter to maintain the multilayer reconstruction. Three or four days after the operation, the catheter was removed.
Results
=======
The upper limit of the surgical approach was the frontal skull base in both TO-TPA and EEA. For EEA, the upper limit was the posterior wall of the frontal sinus, and for TO-TPA, the upper limit was the optic chiasma. The lower limit of the surgical approach was the C3 vertebral body in TO-TPA, and the C1 level in EEA.
[Table 1](#T1){ref-type="table"} shows a clinical summary and the surgical results for the TO-TPA group, and [Table 2](#T2){ref-type="table"} shows those for the EEA group. In the TO-TPA group, GTR was limited in a case of upper cervical chordoma (Case 9). However other large tumors extending from the clivus to the upper cervical spine were not completely removed (STR 4, PR 4) because of lateral tumor extension into the parapharyngeal space. However, in the EEA group, all tumors except one were localized only in the clivus. GTR was achieved in 3 cases and STR in 2 cases, and the incidence of subdural invasion was the same as that in the TO-TPA group (5 cases each).
Surgical complications were lower in TO-TPA than EEA. One patient, who had been treated with heavy-ion radiation pre-operatively, complained of pharyngeal fistula after surgery. No post-operative CSF leakage was observed in the TO-TPA group; the subdural part was removed only in the case with a dural defect smaller than 10 mm, with the aim of preventing CSF leakage ([Fig. 6](#F6){ref-type="fig"}). In the EEA group, surgical complications occurred in 4 cases: postoperative meningitis with or without CSF leakage, acute hydrocephalus with brain stem infarction, and diabetes insipidus. They occurred only in tumors with subdural extension, and these EEA complications were caused by active surgical removal in the subdural space ([Fig. 7](#F7){ref-type="fig"}). In the TO-TPA group, mean operative time was 7.52 hours. On the other hand, mean operative time was 5.25 hours in the EEA group.
Oral intake was started an average of 15.5 days after surgery in the TO-TPA group, while all patients in the EEA group started oral intake the day after the operation. The mean hospitalization in the EEA group was almost same as that of TO-TPA group: 41 days in the EEA group and 38.9 days in the TO-TPA group. In patients without surgical complication, average hospitalization was shorter in EEA (13.6 days) than TO-TPA group (29.3 days). However 5 patients having surgical complications needed longer hospitalization more than 1 month. According to the pre- and post-operative KPS scores, no patient deteriorated following surgery, with two exceptions in the EEA group with complications (Cases 4, 5).
Discussion
==========
Both TOA and EEA have the advantage of providing a direct extradural route without brain retraction. EEA is less invasive with minimal displacement or distortion of surrounding structures.^[@B7]^
Upward access of TOA is limited by the degree of mouth opening, the patient\'s tongue size, and the position of the soft palate. TOA provides midline exposure of the inferior one-third of the clivus. Inferior exposure provided by TOA is limited by the degree of mouth opening and the size of the patient\'s oral cavity, and is reported to extend from the C2 vertebra down to the C3 vertebra.^[@B6],[@B17]^ TO-TPA increases the rostral exposure of the TOA by removing the hard palate bone and provides surgical fields superior to the sphenoid sinus. It is useful for removing tumors around the CVJ that extend beyond the hard palate.^[@B6]^ In a cadaveric study using endoscopy, Visocchi et al. reported that TOA provides better exposure of the CVJ, both in the sagittal and coronal plane.^[@B18]^ TOA also has the advantage of easier access between the C1 and C3 regions. However, a disadvantage of TO-TPA is the lateral limitation by microscopic axis toward parapharyngeal tumors due to the presence of the pterygoid process superiorly and atlanto-occipital joint inferiorly. Therefore, tumor invasion into the parapharynx or lateral to the CVJ limits complete removal, and that is the main reason for the lower radicality compared to EEA, as shown in this article.
TPA is currently partially being replaced by EEA. EEA was developed as a new strategy for treating skull base tumors, the surgical field extending upward to the frontal sinus and downward to the CVJ, decreasing the lateral dead angle. Although the lateral limit of the EEA is basically the internal carotid artery (ICA),^[@B19]^ by using transpterygoid approach cavernous sinus lateral to the ICA and furthermore middle cranilal fossa can be reached.^[@B20]^ The lesion in the dorsum sellae or posterior clinoid could be resected by superior transposition of the pituitary gland and infundibulum.^[@B21]^ It was reported that EEA has 100% GTR rates in mid-clival chordomas, but this percentage decreases to 50% for lower clival lesions, even when treated in the most technologically advanced institutions.^[@B19]^ EEA should be preferred for clival lesions,^[@B10]^ but it remains difficult to use this approach for lesions around the CVJ, because the operative tool is hampered by the hard palate. The nasopalatine line (NPL) and the nasoaxial line (NAxL), as defined in sagittal computed tomography (CT) images, predict the inferior limit of usefulness of the EEA.^[@B23],[@B24]^ The NPL connects the inferior point of the nasal bone to the posterior edge of the hard palate in the mid-sagittal plane. The NAxL connects the midpoint of the line from the rhinion to the anterior nasal spine of the maxillary bone, to the tip of the posterior nasal spine of the palatine bone. The NPL predicts the inferior limit to be the C2 body.^[@B23]^ NAxL predicts the inferior limit to range from the dens to the upper half of the C2 body. Aldana et al. reported that the NAxL is more accurate for predicting this lower limit, while the NPL overestimates the inferior surgical dissection limit, based on cadaver dissections.^[@B24]^
In this study, C1 was the most inferior level in the EEA series, while the C3 vertebral body was the most inferior level in the TO-TPA series for surgical resection. [Figure 8](#F8){ref-type="fig"} shows the range of TOA, TPA and EEA in the sagittal plane. EEA can be used if the tumor lies above the C1 and C2 odontoid processes, but one should consider selecting TOA when it extends below the C2 vertebral body. Therefore, a combination of endoscopic TOA and EEA will be a better option for large tumors extending from the upper clivus to the C2 body. Such a combination may increase the radicality by extending the superior and lateral overview through the endoscope, thus avoiding an invasive palatal incision.
Subdural tumor removal has been a risk factor for CSF leakage and infection via the transnasal/oral routes. A comparison of bacterial colonization rates in transoral and transnasal nasopharyngeal samples revealed higher rates in transoral nasopharyngeal swabs.^[@B25]^ Therefore in the current series the subdural lesion was removed by TO-TPA only in a case in which the subdural mass was smaller than 10 mm. The dural defect was closed safely using abdominal fat and a mucosal covering in the pharynx in such cases. This might explain the low incidence of CSF leakage and meningitis in our series of TO-TPA. In the EEA group, however, removal of the subdural part was more aggressive than in the TO-TPA group, resulting in a high rate of meningitis or brain stem infarction with or without post-operative CSF leakage. The mucosal layer of the naso-pharynx around the pharyngeal tonsil is very fragile and it is difficult to achieve primary mucosal closure as in the pharynx; in addition, complete closure by EEA is not easy in the case of a large dural defect, even when using a naso-septal mucosal flap. In case of a large subdural mass, two-stage EEA followed by a transcranial approach, such as the suboccipital condylar approach^[@B26]^ or condylar fossa approach^[@B27]^ may be more effective than EEA alone at preventing CSF leakage and infection.
Finally, it should be noted that most chordomas are likely to regrow, even when surgery seems to suggest GTR. Therefore, selection of the surgical method should be based not only on the degree of removal, but also on how to maintain the patient\'s quality of life (QOL) in combination with additional radiosurgery or a second operation.
Conclusion
==========
EEA is a less invasive and a more effective surgical method for chordomas localized in the clivus, for giving a better lateral overview than the microscope. However, in cases in which the tumor extends down to the C2 body, TOA, or a TOA/EEA combination should be considered. Reducing the surgical complications during or after removal of the subdural part is an important task for EEA.
{#F1}
{#F2}
{#F3}
{#F4}
{#F5}
{#F6}
{#F7}
{#F8}
######
Patient characteristics and surgical results in transoral-transpalatal approach
Case 1 2 3 4 5 6 7 8 9
------------------------------- ----------- --------- ------------- --------- -------- --------- ------- ------- --------------------
Age/Gender 71F 12F 12F 30M 10F 11F 62M 43M 72F
Symptom II III IV V XII VI IXX VI VIII XXIXII None VI VI None
Upper end FB MC LC MC LC LC UC UC C1
Lower end C1 C2 body C2 odontoid CVJ C3 C2 body LC CVJ C3
Subdural extension Yes Yes Yes No Yes No Yes No No
Extent of resection STR PR PR PR PR STR STR STR GTR
Intra/post-operative CSF leak Yes/No No/No No/No No/No Yes/No No/No No/No No/No No/No
Other complication No No No No No No No No Pharyngeal fistula
Operative time (hours) 9.41 6.41 9.91 6.30 7.36 9.80 7.55 7.40 3.50
Oral intake 11 POD 10 POD 10 POD 7 POD 12 POD 5 POD 4 POD 7 POD 80 POD
Hospitalization days 53 35 34 17 29 26 16 25 115
Pre-/post-operative KPS 90/90 70/70 80/80 90/90 50/60 100/100 90/90 90/90 90/90
CSF: cerebrospinal fluid, CVJ: cranio-vertebral junction, FB: frontal base, GTR: gross total removal, KPS: Karnofsky performance status, LC: lower clivus, MC: middle clivus, POD: post-operative day, post-operative KPS: KPS at the time of discharge, PR: partial removal, STR: subtotal removal, UC: upper clivus.
######
Patient characteristics and surgical results in endoscopic endonasal approach
Case 1 2 3 4 5 6 7 8 9
------------------------------- --------- --------- ------- --------- ------------ -------- ------------ ------- ---------
Age/Gender 68M 42M 44M 74F 62F 62F 78F 33M 41F
Symptoms VI VI VI II III III III IV III IV VI IV XII
Upper end UC UC UC UC UC UC FB UC UC
Lower end MC MC LC MC MC MC C1 UC LC
Subdural extension No No No Yes Yes Yes Yes No Yes
Extent of resection GTR GTR STR PR GTR PR STR PR PR
Intra-/post-operative CSFLeak No/No No/No No/No No/No Yes/No Yes/No No/Yes No/No No/No
Other complications No No No AHC BSI Meningitis DI Meningitis No No
Operative time (hours) 4.71 3.91 4.06 6.00 3.18 8.31 10.18 3.03 3.83
Oral intake 1 POD 1 POD 1 POD 1 POD 1 POD 1 POD 1 POD 1 POD 1 POD
Hospitalization days 17 14 12 39 48 32 182 10 15
Pre-/post-operative KPS 100/100 100/100 90/90 80/60 90/70 90/90 50/50 90/90 100/100
AHC: acute hydrocephalus, BSI: brainstem infarction, DI: diabetes insipidus, FB: frontal base, GTR: gross total removal, KPS: Karnofsky performance status, LC: lower clivus, MC: middle clivus, POD: post-operative day, post-operative KPS: KPS at the time of discharge, PR: partial removal, STR: subtotal removal, UC: upper clivus.
[^1]: **Conflicts of Interest Disclosure**
The authors have no conflict of interest directly relevant to the content of this article.
| {
"pile_set_name": "PubMed Central"
} |
Introduction {#s1}
============
Pulmonary fibrosis can result from a variety of causes, including lung injury, environmental particle and toxin inhalation, chemotherapy, systemic autoimmune diseases, or as an idiopathic entity in form of idiopathic interstitial pneumonias (IIP) [@pone.0002142-Kim1]--[@pone.0002142-Martinez1]. Idiopathic pulmonary fibrosis (IPF), the most common form of IIP, represents a progressive and lethal disorder with unresolved pathogenesis and unresponsiveness to currently available therapies [@pone.0002142-Walter1]. Distortion of the normal lung architecture in IPF is evident by temporo-spatially heterogeneous histology, including areas of normal parenchyma, mild interstitial inflammation due to mononuclear infiltrates, septal fibrosis with subepithelial fibroblast foci, and honeycombing [@pone.0002142-Visscher1], [@pone.0002142-AL1].
Fibroblast foci represent the hallmark lesions of IPF, as they constitute aggregates of activated myofibroblasts, which promote excessive ECM deposition [@pone.0002142-AL1]. The occurrence of fibroblast foci represents an important prognostic factor, since their numbers have been correlated with survival in IPF [@pone.0002142-King1]. Fibroblast foci occur in subepithelial layers, close to areas of alveolar epithelial cell injury and repair, suggesting that impaired epithelial-mesenchymal crosstalk contributes to the pathobiology of IPF [@pone.0002142-King1], [@pone.0002142-Noble1]. Indeed, it is well accepted that repetitive injury and subsequent repair of alveolar epithelial type II (ATII) cells, in the presence or absence of local inflammation, represent a key pathogenic mechanism in IPF, which leads to aberrant growth factor activation and perpetuation of fibrotic transformation [@pone.0002142-Selman1]. Although several soluble mediators, such as transforming growth factor (TGF)-β1 or interleukin (IL)-1β, have been assigned a clear pathogenic role in IPF and experimental models thereof (9, 10), therapeutic options neutralizing their activity have not been successful in clinical use as of yet.
The Wnt family constitutes a large family of highly conserved secreted growth factors essential to organ development, a process often recapitulated in organ failure. The best characterized Wnt signaling pathway is the β-catenin-dependent, or canonical, Wnt signaling pathway [@pone.0002142-Moon1]--[@pone.0002142-Johnson1]. Here, in the absence of active Wnt ligands, β-catenin is constitutively phosphorylated by its interaction with axin, adenomatosis polyposis coli (APC), and glycogen synthase kinase (Gsk)-3β, and subsequently degraded. In the presence of Wnt ligands, two distinct membrane receptors, the frizzled (Fzd) or the low density lipoprotein receptor-related proteins (Lrp) 5 and 6, are activated upon ligand binding. In detail, Wnt stimulation leads to phosphorylation of Lrp6 by Gsk-3β and casein kinase γ in its cytoplasmic region, which leads to the recruitment of axin. Subsequently, β-catenin phosphorylation is attenuated, its degradation inhibited, and accumulated β-catenin undergoes nuclear translocation, where it regulates target gene expression through interaction with members of the T-cell-specific transcription factor/lymphoid enhancer-binding factor (Tcf/Lef) family [@pone.0002142-Moon1], [@pone.0002142-Logan1].
Importantly, increased nuclear β-catenin staining was recently reported in IPF tissue sections [@pone.0002142-Chilosi1], indicative of increased Wnt signaling. In addition, unbiased microarray screens have also revealed an increased expression of Wnt target genes, such as matrix metalloproteinase (Mmp) 7, or secreted frizzled-related protein (Sfrp) 2 in IPF [@pone.0002142-Yang1]--[@pone.0002142-Lewis1]. We therefore hypothesized that canonical Wnt signaling is aberrantly activated in IPF, recapitulating developmentally active programs in this chronic disease. To this end, we achieved our aim to elucidate the expression, localization, and activity of the Wnt/β-catenin pathway in IPF.
Results {#s2}
=======
Initially, we sought to quantify the mRNA expression of canonical Wnt/β-catenin signaling components in lung tissue samples of transplant donors and IPF patients using quantitative real-time (q)RT-PCR. As depicted in [Figure 1a](#pone-0002142-g001){ref-type="fig"}, canonical Wnt ligands were variably expressed in the human lung. Wnt1, 2, 3a, and 7b were expressed at similar levels in normal lung tissue, while Wnt10b was only little expressed. In IPF lung specimens, Wnt1, 7b, and 10b mRNA levels were markedly upregulated (log-fold change of 1.19±0.43, 1.05±0.43, and 1.58±0.59, respectively), whereas Wnt3a was significantly downregulated (log-fold change −1.93±0.65) ([Figure 1a](#pone-0002142-g001){ref-type="fig"}).
{#pone-0002142-g001}
Next, we analyzed the expression of common Wnt receptors and co-receptors. As shown in [Figure 1b](#pone-0002142-g001){ref-type="fig"}, the most abundant receptors in the human lung were Fzd1 and 4, and the co-receptors Lrp5 and 6, but their expression was similar in control and IPF lungs. Interestingly, Fzd2 and 3 were expressed at low levels in control as well as IPF lungs, but significantly increased in IPF (log-fold change 1.04±0.24 and 1.41±0.31, respectively) ([Figure 1b](#pone-0002142-g001){ref-type="fig"}).
The main canonical Wnt signal transducers Gsk-3β and β-catenin were both expressed in normal and fibrotic lung tissue, with a significantly increased expression of β-catenin in IPF (log-fold change 0.98±0.28) ([Figure 1c](#pone-0002142-g001){ref-type="fig"}). With the exception of Tcf1, all members of the Tcf/Lef family of transcription factors were expressed in normal and fibrotic lung tissue. Lef1 was significantly upregulated in IPF (log-fold change 0.85±0.34) ([Figure 1c](#pone-0002142-g001){ref-type="fig"}).
We went on to localize cell types capable of Wnt ligand secretion, as assessed by immunohistochemistry of Wnt1 and 3a ligands, and cell types capable of Wnt signaling, as assessed by immunohistochemistry of β-catenin and Gsk-3β, in donor and IPF lung tissue ([Figure 2](#pone-0002142-g002){ref-type="fig"}, [3](#pone-0002142-g003){ref-type="fig"}, [4](#pone-0002142-g004){ref-type="fig"}, [5](#pone-0002142-g005){ref-type="fig"}). Wnt1 was mainly expressed in bronchial and alveolar epithelium, with strong staining of alveolar epithelial type II (ATII) cells ([Figure 2](#pone-0002142-g002){ref-type="fig"}). Additionally, Wnt1 was expressed in vascular smooth muscle cells ([Figure 2a, upper panel](#pone-0002142-g002){ref-type="fig"}). In IPF ([Figure 2b](#pone-0002142-g002){ref-type="fig"}), Wnt1 staining was observed in hyperplastic ATII cells and bronchial epithelial cells. An apical staining pattern of Wnt1 in bronchial epithelial cells was observed in IPF, suggesting increased secretion of Wnt1. Interestingly, Wnt1 was also expressed by endothelial cells in IPF tissues ([Figure 2b, lower panel, arrowhead](#pone-0002142-g002){ref-type="fig"}). Wnt3a protein expression was mainly detected in ATII cells ([Figure 3a, b, lower panels](#pone-0002142-g003){ref-type="fig"}) and selected ciliated bronchial epithelial cells ([Figure 3a, b, upper panels](#pone-0002142-g003){ref-type="fig"}) in donor as well as IPF lung tissue.
{#pone-0002142-g002}
{#pone-0002142-g003}
{#pone-0002142-g004}
{#pone-0002142-g005}
We then analyzed protein expression pattern of β-catenin and Gsk-3β ([Figure 4](#pone-0002142-g004){ref-type="fig"} and [5](#pone-0002142-g005){ref-type="fig"}, respectively). Strong membranous and cytoplasmic β-catenin expression was observed in bronchial epithelial and ATII cells, as well as endothelial cells in donor lung tissue ([Figure 4a, arrowhead](#pone-0002142-g004){ref-type="fig"}). Importantly, in IPF, β-catenin expression was less membranous and enhanced in the cytoplasm, with clear nuclear staining observed in single cells ([Figure 4b, arrow](#pone-0002142-g004){ref-type="fig"}). Strong β-catenin expression was noticed in hyperplastic ATII cells, in particular in areas of bronchiolization ([Figure 4b](#pone-0002142-g004){ref-type="fig"}), as assessed by immunohistochemistry. Gsk-3β showed a very similar expression pattern to β-catenin, with predominant staining in bronchial epithelial and ATII cells, as well as endothelial cells in donor lung tissues ([Figure 5a, arrowhead](#pone-0002142-g005){ref-type="fig"}). In IPF lungs, intense staining of basal bronchial epithelial and hyperplastic ATII cells was observed ([Figure 5b](#pone-0002142-g005){ref-type="fig"}). Taken together, the Wnt/β-catenin system was largely expressed in the bronchial and alveolar epithelium in normal and IPF tissue.
To further corroborate these results, we quantified cell-specific gene expression of Wnt/β-catenin signaling components in primary human ATII cells and fibroblasts derived from IPF patients or transplant donors, using qRT-PCR ([Figure 6](#pone-0002142-g006){ref-type="fig"}). As depicted in [Figure 6a](#pone-0002142-g006){ref-type="fig"}, we observed increased mRNA expression of Wnt7b and 10b in ATII cells from IPF patients. Wnt3a was downregulated in IPF ATII cells, while Wnt1 levels remained unchanged. Furthermore, we observed significantly increased mRNA levels of the receptor Fzd3 and the intracellular signaling molecules Gsk-3β, β-catenin, and Lef1 in IPF ATII cells ([Figure 6b and 6c](#pone-0002142-g006){ref-type="fig"}, respectively). Interestingly, we routinely observed higher expression levels of these components in primary ATII cells compared with fibroblasts (data not shown), thereby confirming our localization analysis. In sum, these data revealed the expression of all required Wnt components in the lung, and significant upregulation thereof in IPF, mainly in the bronchial and alveolar epithelium.
{ref-type="sec"}. The mRNA levels of Wnt1, 3a, 7b, and 10b (a), the receptors Fzd2 and 3, and Lrp5 and 6 (b), and Gsk-3β, β-catenin, and Lef1 (c) in ATII cells were assessed by qRT-PCR. Results are derived from 3 different cell isolations each and presented as mean±s.e.m., \* p\<0.05.](pone.0002142.g006){#pone-0002142-g006}
To accurately assess, whether the Wnt/β-catenin signaling pathway was activated in IPF, we performed Western blot analysis of phospho-Gsk-3β, phospho-Lrp6, and β-catenin. As presented in [Figure 7a](#pone-0002142-g007){ref-type="fig"}, we observed an increased phosphorylation of both, Gsk-3β and Lrp6, in IPF. This is consistent with increased expression of total β-catenin ([Figure 7a](#pone-0002142-g007){ref-type="fig"}), and indicative of activated Wnt/β-catenin signaling. Next, we analyzed the expression of the well-characterized Wnt target genes fibronectin (Fn) 1, matrix metalloproteinase (Mmp) 7, and cyclin D1 in IPF lungs. As depicted in [Figure 7b](#pone-0002142-g007){ref-type="fig"}, all of these Wnt target genes were upregulated in IPF lung samples, as compared with transplant donor samples.
{#pone-0002142-g007}
We then sought to investigate biological effects elicited by Wnt ligands on key cell types involved in IPF pathogenesis. To this end, we assessed proliferation, as well as (myo)fibroblast activation and collagen deposition in A549 lung epithelial cells und NIH-3T3 fibroblasts, respectively. Using a Tcf/Lef-driven reporter gene assay, we first demonstrated that Wnt3a elicited a potent canonical Wnt/β-catenin response, while Wnt7a did not (fold induction of 8.96±1.59 and 0.89±0.09 for Wnt3a and Wnt7a, respectively; [Figure 8a](#pone-0002142-g008){ref-type="fig"}). Furthermore, Wnt3a stimulation led to a strong increase of A549 cell proliferation (268×10^3^±28×10^3^ versus 119×10^3^±16×10^3^ for Wnt3a and control, respectively; [Figure 8b](#pone-0002142-g008){ref-type="fig"}).
{#pone-0002142-g008}
Wnt3a led to a significant induction of the Wnt target gene cyclin D1 and the myofibroblast activation markers smooth muscle actin (Acta2) and fibroblast-specific protein (Fsp) 1, as assessed by qRT-PCR of Wnt3a- or vehicle-treated fibroblasts ([Figure 9a](#pone-0002142-g009){ref-type="fig"}). This coincided with increased collagen production, assessed by Sircol assays, in response to Wnt3a, to levels similar to those due to treatment with TGF-β1 (fold induction of 3.07±0.3 and 2.6±0.2 for Wnt3a and TGF-β1, respectively; [Figure 9b](#pone-0002142-g009){ref-type="fig"}). This was confirmed by immunofluorescence staining of type I collagen in fibroblasts, demonstrating increased collagen staining in response to Wnt3a ([Figure 9c](#pone-0002142-g009){ref-type="fig"}). In contrast, Wnt3a treatment did not affect fibroblast proliferation (data not shown), suggesting that Wnt ligands elicit profibrotic effects in a cell-specific manner on resident lung cells.
{#pone-0002142-g009}
Discussion {#s3}
==========
IPF is the most common form of the idiopathic interstitial pneumonias. IPF exhibits a poor prognosis and unresponsiveness to currently available therapies, reflecting our limited understanding of the basic mechanisms and mediators implicated in the pathogenesis of this progressive and fatal disease [@pone.0002142-Martinez1], [@pone.0002142-Walter1]. Historically, inflammatory processes were thought to represent the main trigger of IPF initiation and progression. This view has recently been questioned, due to the ineffectiveness of anti-inflammatory therapies in IPF [@pone.0002142-Selman1]. Major key pathophysiological events in IPF currently discussed include repetitive alveolar epithelial cell injuries, in the presence or absence of local inflammation, impaired epithelial-mesenchymal crosstalk, and subsequent fibroblast to myofibroblast activation [@pone.0002142-Selman1], [@pone.0002142-White1]--[@pone.0002142-Scotton1]. These mechanisms are mediated by aberrantly activated signaling molecules that drive the fibrotic process, such as TGF-β, IGF, PDGF, or TNF-α [@pone.0002142-Noble1], [@pone.0002142-Scotton1].
In this respect, the Wnt signaling system is of particular interest, as it constitutes a developmentally active pathway, which is reactivated in chronic diseases characterized by pathologic tissue remodeling [@pone.0002142-Moon1]--[@pone.0002142-Johnson1], [@pone.0002142-Nusse1], [@pone.0002142-Selman3]. Unbiased microarray screens have recently revealed the overexpression of Wnt target genes, including Wnt-induced signaling pathway protein (Wisp) 1, matrix metalloproteinase (Mmp) 7, or secreted frizzled-related protein (Sfrp) 2, in IPF lungs [@pone.0002142-Selman2], [@pone.0002142-Lewis1], [@pone.0002142-Selman3]. Furthermore, a recent study localized β-catenin staining to the nuclei of ATII cells and interstitial fibroblasts in IPF lungs [@pone.0002142-Chilosi1], suggestive of activated Wnt signaling [@pone.0002142-Morrisey1]. Thus, we hypothesized in our study that canonical Wnt signaling is reactivated in IPF lungs, in particular in hyperplastic ATII cells, thus contributing to disease development and progression in IPF. While this has not yet been adressed in the adult lung, canonical Wnt/β-catenin signaling is known to play an essential role in lung development, as lung epithelial cell-specific deletion of β-catenin prevents formation of the distal, but not the proximal airways [@pone.0002142-Mucenski1]. Furthermore, epithelial-cell specific expression of constitutively active β-catenin leads to epithelial cell dysplasia and abnormal epithelial differentiation in mice [@pone.0002142-Mucenski2].
To this end, we performed a comprehensive analysis of canonical Wnt signaling components at the mRNA and protein level, assessing expression, localization, as well as activity of Wnt ligands, receptors, and intracellular signaling molecules in IPF. We demonstrate that all essential components were expressed in the human lung, and particularly localized to the alveolar and bronchial epithelium in normal, as well as IPF lungs, as demonstrated by qRT-PCR and immunohistochemistry. Lung homogenate, as well as cell-specific analysis of Wnt ligands and receptors demonstrated increased expression of Wnt ligands (compare [Figures 1](#pone-0002142-g001){ref-type="fig"} and [6](#pone-0002142-g006){ref-type="fig"}), with the exception of Wnt1. While Wnt1 was increased in IPF lung homogenates, it was not regulated in IPF ATII cells, suggesting that other cell types, including bronchial epithelial cells or endothelial cells ([Figure 2](#pone-0002142-g002){ref-type="fig"}), may serve as the primary source for Wnt1 expression in IPF.
Similar to the Wnt ligands, Gsk-3β and β-catenin were highly expressed in hyperplastic ATII cells and bronchiolar epithelial cells at sites of bronchiolization in IPF. This suggests that Wnt signaling, which is known to drive epithelial cell hyperplasia in non-pulmonary epithelia [@pone.0002142-Johnson1], may regulate ATII cell hyperplasia and increased bronchial epithelial cell proliferation in IPF. Together with our observation that Wnt ligands are mainly secreted by lung epithelial cells, this indicates that Wnt signaling in the adult lung initiates from the epithelium and acts in an autocrine and paracrine fashion on epithelial and mesenchymal cells, respectively. While this has been underappreciated in the normal and diseased adult lung, it represents a very well-documented pathway active in the developing lung [@pone.0002142-Morrisey1].
Our current understanding about the functional relevance of Wnt signaling in lung epithelium is largely derived from transgenic animal models. If deleted by homologous recombination in mice, Wnt5a deficiency leads to overexpansion of distal airways and inhibition of lung maturation, accompanied by enlarged intersaccular interstitial compartments [@pone.0002142-Li1]. Similarly, the loss of Wnt7b causes perinatal death due to respiratory failure, subsequent to impaired mesenchymal growth and vascular development that is due to defective autocrine and paracrine Wnt signaling by the airway epithelium [@pone.0002142-Shu1]. Here, we also report that Wnt ligands induce lung epithelial cell proliferation and fibroblast activation and collagen synthesis ([Figures 8](#pone-0002142-g008){ref-type="fig"}, [9](#pone-0002142-g009){ref-type="fig"}). These observations further support the importance of properly regulated Wnt signaling for normal epithelial-mesenchymal interactions, and emphasize the impact of dysregulated Wnt signaling in diseases characterized by an impairment thereof, such as IPF.
In this study, we used recombinant Wnt3a to assess the functional effects of canonical Wnt signaling in epithelial cells and fibroblasts. While we observed decreased expression of Wnt3a in IPF homogenates and ATII cells ([Figures 1](#pone-0002142-g001){ref-type="fig"} and [6](#pone-0002142-g006){ref-type="fig"}), we proceeded with Wnt3a for the following reasons: First, Wnt3a has been repeatedly reported to potently stimulate β-catenin-dependent Wnt signaling in vitro [@pone.0002142-Lu1], and has been recognized as the prototypic Wnt ligand for in vitro stimulations [@pone.0002142-Moon1]. Second, Wnt3a is one of the few Wnt ligands that is available in active and recombinant form. Other Wnt ligands used in *in vitro* studies, such as Wnt1, are commonly either overexpressed by viral transduction, or supplied in the form of conditioned media harvested from Wnt1 overexpressing cell lines [@pone.0002142-Szeto1], [@pone.0002142-Taneyhill1]. As we sought to define time-dependent effects of canonical Wnt signaling in our study, we therefore chose to proceed with Wnt3a stimulation.
One of the key questions arising from this study is whether different Wnt isoforms, such as Wnt1 and Wnt3a, are able to induce different effects, or whether they are merely expressed in a distinct spatio-temporal fashion in the lung to elicit different phenotypes, if deleted in mice. In general, a specific biological effect induced by Wnt ligand exposure depends on the expression of distinct Wnt receptors and signaling molecules, distinct ECM molecules such as glypicans, but also on the presence of Wnt inhibitors, such as Dickkopfs, secreted frizzled receptor proteins, or Wnt inhibitory factor [@pone.0002142-Kikuchi1], [@pone.0002142-Mikels1]. Second, different affinities of Wnt ligands for distinct receptor subtypes have been shown to encode signal specificity [@pone.0002142-Mikels1]--[@pone.0002142-Cadigan1]. Third, crosstalk between Wnt/β-catenin signaling with other pathways, such as the TGF-β pathway, may induce tissue- and cell-type-specific effects of relevance to IPF. Of interest, it has recently been demonstrated that the Wnt/β-catenin signaling compounds axin and GSK-3 impact TGF-β signaling via controling Smad3 protein stability [@pone.0002142-Guo1].
We also demonstrated increased levels of β-catenin in IPF, which were predominantly localized to bronchial epithelial and hyperplastic ATII cells. In contrast to a previous investigation presenting increased nuclear β-catenin staining in bronchial epithelial and ATII cells, as well as interstitial fibroblasts [@pone.0002142-Chilosi1], we did not observe widespread nuclear staining of β-catenin using immunohistochemistry in our study. This may be due to different antibody or tissue preparations in these two studies, but moreover suggests that Wnt activation in IPF is more subtle than previously assumed. This may also indicate that only a minority of the cellular β-catenin content in epithelial cells is responsible for Wnt signaling, while the majority of β-catenin molecules is present within the cytosol and at cell-cell contacts [@pone.0002142-Brembeck1], [@pone.0002142-PerezMoreno1]. Therefore, to generate a more comprehensive yet concise view of Wnt signaling activity in IPF, we sought to combine expression and localization analysis of Wnt signaling components, paired with phosphorylation analysis of Lrp6 and Gsk-3β, together with quantitative expression analysis of Wnt target genes.
In sum, we report increased functional Wnt signaling in IPF, documented by increased phosphorylation of Lrp6 and Gsk-3β, which has recently been demonstrated to present as a most sensitive indicator of Wnt activity in tissue sections [@pone.0002142-Bilic1], [@pone.0002142-Forde1]. We also observed increased expression of the Wnt target genes Cyclin D1, Mmp 7, and Fibronectin 1. In particular, Mmp 7 and Fibronectin 1 have recently been assigned a causative role in pulmonary fibrosis and shown to be expressed in interstitial fibroblasts and ATII cells [@pone.0002142-Zuo1], [@pone.0002142-Muro1]. The Wnt signaling pathway may therefore present as a novel pathogenic system re-activated during chronic tissue remodeling observed in IPF [@pone.0002142-Selman3]. As pointed out in this study, Wnt ligands are secreted in a cell-specific fashion, but act on a multitude of adjacent cell types, thereby modifying Wnt ligand activity by cellular traps. Future *in vitro* and *in vivo* studies will undoubtedly shed more light on the mechanistic principles underlying Wnt activation in IPF, and whether therapeutic modulation thereof will present as a suitable therapeutic tool in this devastating disease.
Materials and Methods {#s4}
=====================
Antibodies and reagents {#s4a}
-----------------------
The following antibodies were used in this study: Total β-catenin (\#9562), phospho-S9- and total Gsk-3β (\#9336 and \#9315, respectively), phospho- and total Lrp6 (\#2568 and \#2560, respectively; all from Cell Signaling Technology, Beverly, MA), Wnt1 (ab15251, Abcam, Cambridge, UK), Wnt3a (38-2700, Zymed Laboratories/Invitrogen, Carlsbad, CA), CyclinD1 (06-137, Upstate, Temecula, CA), α-smooth muscle actin (SMA, A2547, Sigma-Aldrich, Saint Louis, MO), collagen type 1 (T40777R, Biodesign, Saco, ME) and Lamin A/C (sc-20681, Santa Cruz Biotechnology, Santa Cruz, CA). Dulbecco\'s modified Eagle\'s medium (DMEM) and fetal calf serum (FCS) were obtained from Invitrogen. Recombinant Wnt3a and Wnt7a was purchased from R&D Systems.
Human tissues {#s4b}
-------------
Lung tissue biopsies were obtained from 15 IPF patients with histological usual interstitial pneumonia (UIP) pattern (4 females, 11 males; mean age = 58±8 years; mean VC = 48%±7%; mean TLC = 50%±5%; mean DL~CO~/VA = 23%±3%; O~2~ = 2--4 l/min; Pa~O2~ = 49--71 mmHg, Pa~CO2~ = 33--65 mmHg) and 9 control subjects (organ donors; 4 females, 5 males; mean age 42±10 years). Individual patient characteristics are shown in [Table 1](#pone-0002142-t001){ref-type="table"}. Samples were immediately snap frozen or placed in 4% (w/v) paraformaldehyde after explantation. The study protocol was approved by the Ethics Committee of the Justus-Liebig-University School of Medicine (AZ 31/93). Informed consent was obtained in written form from each subject for the study protocol.
10.1371/journal.pone.0002142.t001
###### Characteristics of IPF patients with UIP pattern.
{#pone-0002142-t001-1}
No. Diagnosis Gender Age (yr) VC (%) DL~CO~/VA (%) TLC (%) O~2~ (l/min) Pa~O2~ (mmHg) Pa~CO2~(mmHg)
----- ----------- -------- ---------- -------- --------------- --------- -------------- --------------- ---------------
1 IPF (UIP) male 63 56% 33% 48% 3 52 33
2 IPF (UIP) male 62 50% 26% 52% 3 49 38
3 IPF (UIP) male 58 49% na na na na na
4 IPF (UIP) male 65 59% 20% 42% 3 53 38
5 IPF (UIP) male 65 59% 20% 42% 4 69 41
6 IPF (UIP) male 43 48% 27% 51% na na na
7 IPF (UIP) male 71 40% 24% 46% na na na
8 IPF (UIP) male 64 59% 22% 52% 2 58 38
9 IPF (UIP) male 60 51% 18% 49% 2 59 39
10 IPF (UIP) male 65 51% 20% 66% 2 53 38
11 IPF (UIP) male 44 47% 25% 55% 2 36 35
12 IPF (UIP) female 43 40% na na 2 54 35
13 IPF (UIP) female 42 50% 17% 58% 3 52 36
14 IPF (UIP) female 66 29% 23% 45% 4 56 45
15 IPF (UIP) female 62 27% na 48% 4 71 65
VC = vital capacity, TLC = total lung capacity, DL~CO~/VA = diffusing capacity of the lung for CO per unit of alveolar volume (all in % predicted), Pa~O2/CO2~ = partial pressure of O~2~/CO~2~ in the arterial blood.
Cell culture {#s4c}
------------
Human alveolar epithelial type II (ATII) cells were isolated, as previously described [@pone.0002142-Fang1]. The purity and viability of ATII cell preparations was consistently \>90% and \>95%, respectively. Primary human fibroblasts were generated by explant cultures as previously described. Identification of fibroblasts was based on the expression of vimentin, collagen, and αSMA [@pone.0002142-Konigshoff1]. The NIH-3T3 murine fibroblast cell line \[German Collection of Microorganisms and Cell Cultures (DSMZ, Braunschweig, Germany)\] and the human lung epithelial cell line A549 (ATCC \#CCL-185) were maintained in DMEM containing 10% FCS and cultured in a humidified atmosphere of 5% CO~2~ at 37°C.
Reverse transcription and quantitative real-time PCR {#s4d}
----------------------------------------------------
Total RNA was extracted using Qiagen extraction kits according to the manufacturer\'s protocol, and cDNAs were generated by reverse transcription using SuperScript™ II (Invitrogen) [@pone.0002142-Konigshoff1], [@pone.0002142-Kitowska1]. Quantitative (q)RT-PCR was performed using fluorogenic SYBR Green and the Sequence Detection System Fast 7500 (PE Applied Biosystems), as previously described. *Hprt1* and *Pbgd*, ubiquitously and equally expressed genes free of pseudogenes, were used as a reference gene in all human and mouse qRT-PCR reactions, respectively. PCR was performed using the primers listed in [Table 2](#pone-0002142-t002){ref-type="table"}, at a final concentration of 200 nM. Relative transcript abundance of a gene is expressed in ΔCt values (ΔCt = Ct^reference^ -- Ct^target^). Relative changes in transcript levels compared to controls are ΔΔCt values (ΔΔCt = ΔCt^treated^ -- ΔCt^control^). All ΔΔCt values correspond approximately to the binary logarithm of the fold change as mentioned in the text. When relative transcript abundance is of information, expression levels are given in ΔCt levels.
10.1371/journal.pone.0002142.t002
###### Primer sequences and amplicon sizes.
{#pone-0002142-t002-2}
Gene Accession Sequences (5′→3′) Length Amplicon
----------- ----------- ------------------- ----------------------------- ---------- --------
β-Catenin NM001904 for AAGTGGGTGGTATAGAGGCTCTTG 24 bp 77 bp
rev GATGGCAGGCTCAGTGATGTC 21 bp
Cyclin D1 NM053056 for CCGAGAAGCTGTGCATCTACAC 22 bp 94 bp
rev AGGTTCCACTTGAGCTTGTTCAC 23 bp
Fn1 NM212482 for CCGACCAGAAGTTTGGGTTCT 21 bp 81 bp
rev CAATGCGGTACATGACCCCT 20 bp
Fzd1 NM003505 for AGCGCCGTGGAGTTCGT 17 bp 64 bp
rev CGAAAGAGAGTTGTCTAGTGAGGAAAC 27 bp
Fzd2 NM001466 for CACGCCGCGCATGTC 15 bp 63 bp
rev ACGATGAGCGTCATGAGGTATTT 23 bp
Fzd3 NM017412 for GGTGTTCCTTGGCCTGAAGA 20 bp 72 bp
rev CACAAGTCGAGGATATGGCTCAT 23 bp
Fzd4 NM012193 for GACAACTTTCACACCGCTCATC 22 bp 164 bp
rev CCTTCAGGACGGGTTCACA 19 bp
Hprt1 NM000194 for AAGGACCCCACGAAGTGTTG 20 bp 137 bp
rev GGCTTTGTATTTTGCTTTTCCA 22 bp
Gsk-3β NM002093 for CTCATGCTCGGATTCAAGCA 20 bp 86 bp
rev GGTCTGTCCACGGTCTCCAGTA 22 bp
Lef1 NM016269 for CATCAGGTACAGGTCCAAGAATGA 24 bp 93 bp
rev GTCGCTGCCTTGGCTTTG 18 bp
Lrp5 NM002335 for GACCCAGCCCTTTGTTTTGAC 21 bp 138 bp
rev TGTGGACGTTGATGGTATTGGT 22 bp
Lrp6 NM002336 for GATTCAGATCTCCGGCGAATT 21 bp 83 bp
rev GGCTGCAAGATATTGGAGTCTTCT 24 bp
Mmp7 NM002423 for GAACGCTGGACGGATGGTA 19 bp 64 bp
rev CAAGTTCATGAGTTGCAGCATACA 24 bp
Tcf3 NM031283 for ACCATCTCCAGCACACTTGTCTAATA 26 bp 71 bp
rev GAGTCAGCGGATGCATGTGA 20 bp
Tcf4 NM030756 for GCGCGGGATAACTATGGAAAG 21 bp 89 bp
rev GGATTTAGGAAACATTCGCTGTGT 24 bp
Wnt1 NM005430 for CTCATGAACCTTCACAACAACGA 23 bp 80 bp
rev ATCCCGTGGCACTTGCA 17 bp
Wnt2 NM003391 for CCTGATGAATCTTCACAACAACAGA 25 bp 78 bp
rev CCGTGGCACTTGCACTCTT 19 bp
Wnt3a NM033131 for GCCCCACTCGGATACTTCTTACT 23 bp 98 bp
rev GAGGAATACTGTGGCCCAACA 21 bp
Wnt7b NM058238 for GCAAGTGGATTTTCTACGTGTTTCT 25 bp 65 bp
rev TGACAGTGCTCCGAGCTTCA 20 bp
Wnt10b NM003394 for GCGCCAGGTGGTAACTGAA 19 bp 59 bp
rev TGCCTGATGTGCCATGACA 19 bp
mCyclinD1 NM007631 for ATGCCAGAGGCGGATGAGA 19 bp 98 bp
rev ATGGAGGGTGGGTTGGAAAT 20 bp
mFsp1 NM011311 for AGGAGCTACTGACCAGGGAGCT 22 bp 102 bp
rev TCATTGTCCCTGTTGCTGTCC 20 bp
mActa2 NM007392 for GCTGGTGATGATGCTCCCA 19 bp 80 bp
rev GCCCATTCCAACCATTACTCC 21 bp
mPbgd NM007392 for ATGTCCGGTAACGGCGGC 18 bp 139 bp
rev GGTACAAGGCTTTCAGCATCGC 22 bp
All primer sets worked under identical real-time PCR cycling conditions with similar efficiencies to obtain simultaneous amplification in the same run. Sequences were taken from GeneBank, all accession numbers are denoted.
Immunohistochemistry {#s4e}
--------------------
Human lungs were placed in 4% (w/v) paraformaldehyde after explantation, and processed for paraffin embedding. Sections (3 µm) were cut, mounted on slides, subjected to antigen retrieval, and quenching of endogenous peroxidase activity using 3% (v/v) H~2~O~2~ for 20 min. Immune complexes were visualized using suitable peroxidase-coupled secondary antibodies, according to the manufacturer\'s protocol (Histostain *Plus* Kit; Zymed/Invitrogen) [@pone.0002142-Konigshoff1], [@pone.0002142-Kitowska1].
Immunofluorescence {#s4f}
------------------
NIH-3T3 cells were plated on chamber slides, fixed with acetone/methanol (1∶1), and blocked for unspecific binding sites with 3% (m/vol) BSA. Fixed cells were incubated with the indicated primary antibodies for 60 min in PBS containing 0.1% (m/vol) BSA. Indirect immunofluorescence was performed by incubation with Alexa 555-conjugated secondary antibodies (Molecular Probes, Eugene, Oregon). Nuclei were visualized by 4,6-diamidino-2-phenylindole staining (DAPI; Roche Diagnostics).
Western blot analysis {#s4g}
---------------------
Human lung tissue specimens were homogenized in extraction buffer \[20 mM Tris-Cl, 150 mM NaCl, 1 mM EDTA, 1 mM EGTA, 1% (v/v) Triton X-100, supplemented with Complete™ Proteinase Inhibitor Cocktail (Merck Biosciences)\] and whole proteins were extracted by centrifugation (12.000× g) for 10 min at 4°C, as described previously [@pone.0002142-Konigshoff1]. Samples containing 25 µg of protein were separated by electrophoresis on a 10% SDS-polyacrylamide gels. The separated proteins were transferred to nitrocellulose membranes (Invitrogen), blocked with 5% skim milk, and incubated with the indicated antibodies. Proteins were then visualized by enhanced chemiluminescence detection (ECL, Amersham Biosciences, Uppsala, Sweden), as reported. Prior to reprobing, nitrocellulose membranes were incubated with stripping buffer \[100 mM 2-mercaptoethanol, 2% SDS, and 62.5 mM Tris-HCl (pH 6.7)\] at 50 °C for 30 min.
Reporter gene assay {#s4h}
-------------------
Lung epithelial A549 cells were transiently transfected with the reporter construct TOPflash or FOPflash (kindly provided by R. Moon, University of Washington, Seattle) using Lipofectamine™ 2000 (Invitrogen). The TOPflash construct contains eight Tcf/Lef binding sites upstream of a minimal TA viral promoter and the firefly luciferase cDNA. The FOPflash construct is identical with the exception that it contains mutated copies of TCf/Lef binding sites and is used as a control for measuring nonspecific activation of the reporter construct. Treatment with Wnt3a (100 ng/ml) or Wnt7a (100 ng/ml) for 24 h was performed 4 h after transfection. Luciferase activities were determined using the Dual Luciferase Assay System (Promega) on a Fusion™ luminometer (Packard BioScience).
Collagen assay {#s4i}
--------------
NIH-3T3 fibroblasts were plated at a density of 30.000 cells/well in 6-well plates, synchronized for 24 h in serum-free medium, and treated for 24 h as indicated. Total collagen content was determined using the Sircol Collagen Assay kit (Biocolor, Belfast, Northern Ireland). Equal amounts of protein lysates were added to 1 ml of Sircol dye reagent, followed by 30 min of mixing. After centrifugation at 10.000× g for 10 min, the supernatant was carefully aspirated and 1 ml of Alkali reagent was added. Samples and collagen standards were then read at 540 nm on a spectrophotometer (Bio-Rad). Collagen concentrations were calculated using a standard curve generated by using acid-soluble type 1 collagen.
Statistical analysis {#s4j}
--------------------
All ΔCt values obtained from real-time RT-PCR were analyzed for normal distribution using the Shapiro-Wilk test, using assignment of a normal distribution with p\>0.05. Normality of data was confirmed using quantile-quantile plots. The means of indicated groups were compared using two-tailed Student\'s *t*-test, or a one-way analysis of variance (ANOVA) with Tukey HSD post hoc test for studies with more than 2 groups. All p values obtained from multiple tests were adjusted using the procedure from Hochberg and Benjamini [@pone.0002142-Hochberg1]. Results were considered statistically significant when p\<0.05.
We are indebted to Simone Becker and Andreas Jahn for excellent technical assistance, Walter Klepetko at the University Hospital Vienna for providing lung tissues, and all members of the Eickelberg Lab for stimulating discussions.
**Competing Interests:**The authors have declared that no competing interests exist.
**Funding:**The German Research Foundation\'s (DFG) Excellence Cluster "Cardiopulmonary System (ECCPS)" (W.S., O.E.), Clinical Research Group 118 "Lung Fibrosis" (W.S., O.E.), the International Graduate Program IGRK1062 "Signalling Mechanisms of Lung Physiology and Disease (SMLPD)" (M.K., W.S., O.E.), and the Sofja Kovalevskaja Award by the Alexander von Humboldt-Foundation to O.E.
[^1]: Conceived and designed the experiments: OE MK NB. Performed the experiments: MK NB EP IC MK. Analyzed the data: OE MK NB EP IC MK. Contributed reagents/materials/analysis tools: WS OE MK NB EP IC MK. Wrote the paper: OE MK.
| {
"pile_set_name": "PubMed Central"
} |
Pancreaticoduodenectomy (PD) remains the standard surgical treatment for resectable peri-ampullary tumors. The first PD operation was reported by Codavilli in 1898 in a patient with an epithelioma of the pancreas, but the patient died from cachexia on the 21^st^ post-operative day.\[[@CIT1]\] In 1946, Whipple described a one-stage PD in which the pylorus was resected.\[[@CIT2]\] The first report of pylorus preserving PD (PPPD) was by Watson in 1944 for ampullary carcinoma\[[@CIT3]\] but it did not gain popularity at that time.
In both the classic PD and PPPD, the head of pancreas, duodenum, and distal bile duct are resected. The main difference is that in classic PD, the gastric antrum and pylorus are resected with the creation of a gastro-jejunostomy, while in PPPD, the gastric antrum and pylorus are preserved and the line of resection is through the first part of duodenum and a duodeno-jejunostomy is performed \[Figure [1 a](#F0001){ref-type="fig"} and [b](#F0001){ref-type="fig"}\].
{#F0001}
Traverso and Longmire reintroduced the concept of PPPD for benign peri-ampullary lesions in 1978 as they thought it would decrease the incidence of post-gastrectomy complications.\[[@CIT4]\] In 1980, they published their experience in PPPD for malignant lesions which included 18 patients with peri-ampullary, duodenal, and pancreatic carcinomas with encouraging results of normal gastric emptying and acidity.\[[@CIT5]\] Since, PPPD has been applied widely to patients with peri-ampullary lesions, benign, or malignant.
In spite of the reported good outcomes of PPPD, many surgeons still question the benefit of this procedure especially the reported high incidence of delayed gastric emptying and, more importantly, the negative impact that pylorus preservation has on tumor clearance, recurrence, and long-term survival.
We will try in this review article to answer the question of the safety of PPPD as compared to classic PD (CPD) in terms of operative factors, peri-operative complications, tumor recurrence, survival, and long-term quality of life.
OPERATIVE FACTORS {#sec1-1}
=================
Operating time {#sec2-1}
--------------
Sugiyama in 2000 compared 10 patients with PPPD to 14 patients with CPD.\[[@CIT6]\] Although there was a trend toward a shorter operative time in the PPPD group, it did not reach statistical significance and that was due to a low volume study. A large, multicentre, randomized, controlled trial of 170 patients comparing PPPD with PD also had found no significant difference in the operating time.\[[@CIT7]\] In a meta-analysis, Traverso had confirmed the previous observation where there had been a trend toward a shorter operating time in PPPD but also not statistically significant.\[[@CIT8]\]
Two large volume retrospective studies have looked at the operating time difference between PPPD and CPD and it had been clear that the PPPD operating time was significantly shorter than that of CPD.\[[@CIT9][@CIT10]\] That observation has been further supported by a meta-analysis by Karanicolas in 2006 and has found that PPPD was 72 min shorter than PD.\[[@CIT11]\] A more recent meta-analysis\[[@CIT12]\] has also shown that PPPD was 41.3 min shorter.
Blood loss and a need for blood transfusion {#sec2-2}
-------------------------------------------
Several reports have indicated no significant difference in intra-operative blood loss and blood transfusion between PPPD and PD.\[[@CIT6]--[@CIT8]\] In a meta-analysis, however, although there has been no significant difference in blood loss, more patients in the PD group have required blood transfusions.\[[@CIT12]\]
Other studies with a larger patient volume, on the other hand, have shown significantly less blood loss and blood transfusions in the PPPD group\[[@CIT9]--[@CIT11]\] that could be partly due to the fact that there is less dissection in PPPD. This observation is very important, as blood transfusions in pancreatic cancer have been associated with a decreased survival rate.\[[@CIT13]\] So if an operative procedure results in less blood loss it should translate into a longer survival.
Operative mortality {#sec2-3}
-------------------
In retrospective analyses, peri-operative mortality has been similar in PPPD and PD groups.\[[@CIT9][@CIT10][@CIT14]\] Two meta-analysis studies have shown a trend toward lower peri-operative mortality in the PPPD group.\[[@CIT11][@CIT12]\]
A randomized controlled trial comparing 13 patients with CPD to 14 patients with PPPD has shown no significant difference in mortality (15.4% and 28.6%, respectively, *P*-value 0.65) but these are very high mortality rates for any pancreaticoduodenectomy in comparison to the widely reported 3% in most studies.\[[@CIT15]\] In a multicentre, randomized, controlled trial involving 170 patients, mortality has been 7% in the CPD group vs. 3% in the PPPD group (*P*-value 0.27)\[[@CIT7]\]
POST-OPERATIVE COMPLICATIONS {#sec1-2}
============================
Delayed gastric emptying {#sec2-4}
------------------------
DGE is probably one of the most studied complications following any type of pancreaticoduodenectomy. There has always been the thought that pylorus preservation would increase the chance of DGE. In a large series from Japan including 1066 patients who underwent PPPD, the incidence of DGE was 46%,\[[@CIT16]\] which supported the idea of higher DGE with PPPD. A small volume, randomized controlled trial has shown DGE to be 15% in PD vs. 64% in the PPPD group (*P*-value 0.2).\[[@CIT15]\]
On the other hand, several other studies have not shown the same observation. A retrospective analysis of 113 patients has shown no significant difference in DGE but half of PPPD patients with DGE had co-existing intra-abdominal complications which could have attributed to DGE.\[[@CIT14]\]
Two retrospective studies have shown no significant difference in DGE between the two groups.\[[@CIT6][@CIT9]\] This was also confirmed in a multicentre, randomized, controlled trial.\[[@CIT13]\] A retrospective analysis of 239 patients showed that DGE in the CPD group was double that of the PPPD group (6 vs. 13%), but there was a higher percentage of T4 and more extensive resections in the CPD group.\[[@CIT10]\]
Several meta-analysis studies have also shown that DGE is not higher in the PPPD group.\[[@CIT8][@CIT11][@CIT12][@CIT17]\]
It seems that DGE is not increased by preservation of the pylorus rather, by other factors including postoperative complications especially intra-abdominal collections. The presence of portal venous hypertension and preoperative cholangitis also increases the chance of post-operative DGE.\[[@CIT18][@CIT19]\]
Shan\[[@CIT22]\] has made a distinction between subjective DGE and objective DGE as measured by cholescintography and has concluded that although subjective DGE was higher in the PPPD group, objective DGE was similar between the CPD and PPPD groups. He has proposed that loss of the distal stomach mechanoreceptors in the CPD group reduces the patient\'s sensation of subjective DGE.
Additionally, Kim\[[@CIT23]\] proposed that pylorospasm could be a cause of DGE in PPPD and has shown a decrease incidence of DGE with the addition of pyloromyotomy. On the other hand, other studies have shown that abnormal gastric motility post surgery is the main cause of DGE regardless of the type of reconstruction.\[[@CIT24][@CIT25]\]
Several methods have been tried to further decrease the incidence of DGE in PPPD. The drug erythromycin has been shown to increase contractility of the stomach and decrease the incidence of DGE.\[[@CIT26][@CIT27]\] On the other hand, somatostatin which is sometimes used to decrease the severity of pancreatic anastomosis leak increases the chance of DGE by more than 3-fold.\[[@CIT28]\]
An interesting observation was that the use of ante-colic doudeno-jejunostomy as opposed to a retro-colic reconstruction in PPPD decreased the incidence of DGE.\[[@CIT19]--[@CIT21]\]
Anastomotic leak {#sec2-5}
----------------
Anastomotic leak, especially from pancreatico-jejunostomy (PJ), is the main factor for morbidity post-PD. A review of 1066 PPPDs in Japan has revealed a leak rate of 16%.\[[@CIT16]\] In a randomized, controlled trial and two meta-analyses, there has been no difference between CPD and PPPD in terms of PJ leak rate.\[[@CIT11][@CIT12][@CIT15]\] Tani\[[@CIT29]\] has shown that the Traveso-type construction (Duodeno-jejunostomy (DJ) distal to PJ) has a lower leak rate than the Billroth I type reconstruction (DJ proximal to PJ).
Intestinal acidity and anastomotic ulceration {#sec2-6}
---------------------------------------------
Not performing an antrectomy could, in theory, result in higher intra-gastric and intestinal pH in the PPPD patient in comparison to the CPD patient. Geenen *et al.* \[[@CIT30]\] has found that intra-gastric and intestinal pH was not reduced in the PPPD patient but in fact intestinal pH was increased.
Marginal ulceration in PPPD was increased with the use of Roux-en-Y jejuna loop which is not exposed to the diluted effect of pancreatic juice.\[[@CIT31]\]
Hospital stay {#sec2-7}
-------------
Usually, the reason for a prolonged hospital stay is either anastomotic leak or DGE. As indicated above, there is no significant difference between CPD and PPPD in terms of leak rate or DGE, so hospital stay should be no different.
Several retrospective, prospective and meta-analysis studies failed to indicate that PPPD causes an increase in hospital stay.\[[@CIT8][@CIT12][@CIT14][@CIT15]\] In fact, one meta-analysis and one retrospective study showed a trend toward a shorter hospital stay with PPPD\[[@CIT9][@CIT11]\]
TUMOR RECURRENCE AND LONG-TERM SURVIVAL {#sec1-3}
=======================================
The adequacy of PPPD as a cancer operation has always been questionable especially after Sharp and his colleagues had reported three cases of PPPD where the duodenal resection margin was positive for carcinoma.\[[@CIT32]\] That observation, however, was not supported by other studies. In a retrospective study in 1993, pathological examination of all positive margins in PPPD for peri-ampullary carcinoma has shown that the most common site for a positive margin was peri-pancreatic soft tissue followed by the pancreatic resection line and then the bile duct resection line and no duodenal-positive margin was identified.\[[@CIT33]\]
The adequacy of resection was further supported by a Japanese group who has found that the number of lymph nodes retrieved was equal in PPPD and total pancreatectomy, which is even more radical than the classic CPD. The number of positive lymph nodes was also no different.\[[@CIT8]\]
One paper has shown that diffuse peritoneal seeding recurrence was higher in the PPPD group, while liver metastasis and retroperitoneal recurrence were higher in the CPD group.\[[@CIT34]\] Out of nine patients in the PPPD group who developed retroperitoneal recurrence, six developed obstruction of the first jejuna loop requiring bypass surgery to relieve the obstruction; therefore, retroperitoneal passage of jejuna loop should be avoided. In another study, however, there was no difference in type of recurrence.\[[@CIT35]\] Both studies revealed no difference in rate of recurrence in general between PPPD and CPD.
Long-term survival was studied extensively as it is the main measure of cancer surgery efficacy. Several retrospective studies have shown that the type of resection does not influence survival.\[[@CIT6][@CIT10][@CIT14][@CIT34][@CIT35]\] Two randomized, controlled trials have also shown no difference in survival.\[[@CIT7][@CIT15]\] A recent meta-analysis has found that a 5 year survival rate was higher in the PPPD group when all tumors where included (*P*-value 0.002), but in the peri-ampullary tumor group there has been no difference in the survival rate.\[[@CIT12]\]
QUALITY OF LIFE {#sec1-4}
===============
The main reason for adopting PPPD was to reserve the whole stomach and to improve digestive function. Post-operative weight gain was comparable between the PPPD and CPD groups\[[@CIT12][@CIT13]\] but patients in the PPPD group reported better gastrointestinal quality of life in terms of appetite, nausea, and diarrhea and an earlier return to work.\[[@CIT12]\] Although weight gain was comparable, pre-operative weight was reached faster in PPPD patients and they exhibited a better mixture of food with bile.\[[@CIT10][@CIT36]\]
Hyperalimintation for malnutrition was less and serum albumin was higher 6 months following surgery in the PPPD group.\[[@CIT6]\]
In patients receiving post-operative chemotherapy, a significant increase in body weight was seen with the preservation of the pylorus.\[[@CIT14]\]
CONCLUSION {#sec1-5}
==========
PPPD in comparison to CPD for peri-ampullary carcinoma is at least as effective in terms of peri-operative morbidity or mortality, tumor recurrence, or long-term survival. It may have some advantages in terms of a shorter operating time, less blood loss, fewer blood transfusions, and a better quality of life.
Therefore, PPPD (a well established procedure) remains a good option for any patient with peri-ampullary carcinoma except if the first part of the duodenum or pylorus is grossly involved with tumor.
**Source of Support:** Nil
**Conflict of Interest:** None declared.
| {
"pile_set_name": "PubMed Central"
} |
Introduction {#sec1}
============
Stable integration of retroviral vectors encompassing a therapeutic transgene enables gene correction of severe blood and immune disorders. Over the past 25 years, murine leukemia virus (MLV)-based vectors have shown therapeutic benefit in gene therapy studies for primary immunodeficiencies (PIDs), such as X-linked severe combined immunodeficiency (SCID-X1), adenosine deaminase deficiency-severe combined immunodeficiency (ADA-SCID), and Wiskott-Aldrich syndrome (WAS).[@bib1], [@bib2], [@bib3], [@bib4] MLV-based vectors were successfully used in the first clinical trials for ADA-SCID.[@bib5], [@bib6], [@bib7] This led to the recent European approval of a retrovirus-based gene therapy product (Strimvelis; GSK GlaxoSmithKline Pharmaceuticals) to treat patients that lack a suitable human leukocyte antigen (HLA)-matched related stem cell donor.[@bib8], [@bib9] However, in clinical trials for other PIDs, several patients developed leukemia or myelodysplasia, raising concerns about the safety of gene therapy.[@bib10], [@bib11], [@bib12], [@bib13], [@bib14] These side effects have been directly attributed to the integration pattern and vector design. Insertional mutagenesis occurred as a consequence of vector integration preference in proximity of proto-oncogenes and activation by strong viral promoter and enhancer elements in the long terminal repeat (LTR) of retroviral vectors. To prevent insertional mutagenesis, self-inactivating (SIN) vectors with deleted enhancer sequences were designed. The lack of promoter/enhancer activity is compensated by weak heterologous promoters to drive transgene expression, such as the elongation factor 1 α short (EFS) and phosphoglycerate kinase (PGK) promoters.[@bib15], [@bib16] Additionally, introduction of genetic insulator sequences has improved the safety of viral vectors.[@bib17], [@bib18], [@bib19], [@bib20], [@bib21], [@bib22] The efficacy of these modified vectors was confirmed in pre-clinical studies and they are now in phase I/II clinical trials for several PIDs.[@bib19], [@bib23], [@bib24], [@bib25], [@bib26], [@bib27]
A complementary approach to improve the safety of gene therapy is to alter the integration pattern, directing integration away from potentially unsafe regions. Gammaretroviral integration is not random, but rather is dictated by host cellular cofactors, such as the bromodomain and extraterminal domain (BET)-containing family of proteins (BRD2, BRD3, and BRD4) that serve as anchors on the host chromatin.[@bib28], [@bib29] A motif in the unstructured C-terminal tail of MLV integrase (IN) interacts with the extraterminal (ET) domain of BRDs, where the latter tethers the retroviral pre-integration complex (PIC) to chromatin regions enriched in BET proteins and thereby defines the integration profile.[@bib28], [@bib29], [@bib30] Deletion of the C-terminal domain (Δ23 amino acids \[aa\], IN~1--380~) or a single substitution (IN~W390A~) uncouples the BET interaction, resulting in BET-independent (Bin) MLV vectors that transduce target cells at wild-type (WT) efficiency but with diminished integration in the vicinity of retroviral integration markers.[@bib31]
Here, we developed next-generation BinMLV vectors with a potentially safer integration profile and lower propensity to activate nearby genes in an effort to alleviate the risk of insertional mutagenesis by interfering with the chromatin-tethering process. We linked chromatin binding peptide sequences to the C-terminal end of BinMLV IN and demonstrated that fusion of these peptides to BinMLV IN generates vectors that produce at high titers and transduce cells at wild-type efficiency. The addition of the chromodomain of CBX1 to MLV~IN_W390A~ efficiently retargeted integration away from gene regulatory elements. More importantly, the retargeted vector transduced human CD34^+^ hematopoietic stem cells (HSCs) at wild-type efficiency, while genotoxicity assays revealed reduced transformation potential.
Results {#sec2}
=======
Efficient Transduction and Integration of Next-Generation BinMLV Vectors {#sec2.1}
------------------------------------------------------------------------
To direct BinMLV integration away from potentially unsafe chromosomal regions, we tailored the chromatin-tethering process by fusing tethering peptides (between 16 and 61 aa long) to the C-terminal end of IN~W390A~ in the MLV packaging plasmid. We opted for peptides that bind histone markers that are widely spread across the chromatin ([Figure 1](#fig1){ref-type="fig"}A; [Table 1](#tbl1){ref-type="table"}). On one hand, we used peptides derived from cellular proteins that bind specific epigenetic histone modifications, such as the chromodomain of heterochromatin-binding protein 1β (CBX1, aa 20--73) and the chromodomain of Y-like protein (CDYL; aa 1--60),[@bib32], [@bib33] giving rise to IN~W390A-CBX~ and IN~W390A-CDYL~, respectively. Alternatively, we fused virus-derived peptides, such as the tethering domain of the human papilloma virus (HPV8) E2 protein (aa 240--255)[@bib34] and the N-terminal end of Kaposi sarcoma's latency associated nuclear antigen (LANA; aa 1--31), which bind to core histone 2A and 2B,[@bib35] resulting in IN~W390A-E2~ and IN~W390A-LANA~, respectively.Figure 1Transduction Efficiencies of Next-Generation BinMLV Vectors(A) Schematic representation of MLV-based vector production. Structure of IN~WT~, IN~W390A~, and the peptides fused to IN~W390A~ are highlighted. The N-terminal HHCC zinc binding domain, the catalytic core domain (CCD), and the C-terminal domain (CTD) are indicated. Red arrowheads indicate the position of the W390A point mutation. The size of the fused peptides is proportionally represented. (B) FACS analysis of SupT1 cells transduced with equal RTUs of the indicated vectors at different MOIs. Three days post-transduction, the percentage of EGFP-positive cells was determined. Average values and standard deviations of triplicate measurements are shown. Data represent measurements from a representative experiment. (C) Mean fluorescence intensity of SupT1 cells transduced with the indicated next-generation BinMLV vectors 3 days post-transduction. Average values and standard deviations of triplicate measurements are shown. Data represent measurements from a representative experiment. ψ, packaging signal; CMV, cytomegalovirus promoter; GAG, group-specific antigen; IN, integrase; LTR, long terminal repeat; PolyA, polyadenylation signal; Pol, polymerase; PR, protease; RT, reverse transcriptase; VSV-G, vesicular stomatitis virus glycoprotein G.Table 1Overview of the Tethering Domains Fused to BinMLV VectorsProteinAccession NumberSelected PeptideBinding SiteSequenceCBX1[P83916](ncbi-wgs:P83916){#intref0025}CD~1~ CBXH3K9me2EYVVEKVLDRRVVKGKVEYLLKWKGFSDEDNTWEPEENLDCPDLIAEFLQSQKTH3K9me3CDYL[Q9Y232](ncbi-wgs:Q9Y232){#intref0030}CD~1~ CDYLH3K9me2LMTFQASHRSAWGKSRKKNWQYEGPTQKLFLKRNNVSAPDGPSDPSISVSSEQSGAQQPPAH3K27me2H3K27me3HPV8 E2[P06422](ncbi-wgs:P06422){#intref0035}E2H2A/H2BQTETKGRRYGRRPSSRKSHV LANAE5LC01LANAH2A/H2BMAPPGMRLRSGRSTGAPLTRGSCRKRNRSPE[^3]
The respective packaging plasmids were subsequently used to produce vesicular stomatitis virus glycoprotein G (VSV-G) pseudotyped MLV-based vectors encoding an LTR-driven EGFP reporter (MLV~IN_W390A-CBX~, MLV~IN_W390A-CDYL~, MLV~IN_W390A-E2~, and MLV~IN_W390A-LANA~; [Figure 1](#fig1){ref-type="fig"}A). In line with previous results,[@bib31] transduction efficiencies in SupT1 cells were at similar levels for MLV~IN~\_~W390A~ and MLV~IN~\_~WT~ ([Figure 1](#fig1){ref-type="fig"}B). The addition of peptide sequences to the C-terminal end of MLV~IN_W390A~ resulted in BinMLV vectors that transduced as efficiently as MLV~IN~\_~WT~ at different MOIs ([Figure 1](#fig1){ref-type="fig"}B) and resulted in comparable expression levels (measured as mean fluorescence intensities \[MFIs\] at day 3; [Figure 1](#fig1){ref-type="fig"}C). Transduction efficiencies and MFIs were corroborated at 10 days post-transduction, underscoring stable expression for the respective integrated vectors ([Figures S1](#mmc1){ref-type="supplementary-material"}A and S1B). Similar data were obtained following transduction of HeLa cells (data not shown). Collectively, these results indicate that the addition of peptide sequences at the C-terminal end of MLV~IN_W390A~ does not impair vector integrity nor transduction efficiency compared to MLV~IN~\_~WT~ and MLV~IN_W390A.~
MLV~IN_W390A~-Peptide Fusions Redistribute Bin Vector Integration {#sec2.2}
-----------------------------------------------------------------
Next, we set out to investigate whether the respective IN chimeras redistribute MLV vector integration. Integration sites were amplified and sequenced, yielding a total of 43,676 unique sites and their computationally generated matched random control (MRC) sites. Initially, we evaluated integration frequencies relative to transcription start sites (TSSs), CpG islands (typically enriched in/near housekeeping gene promoters), and DNase hypersensitive sites (DHSs), both surrogate markers for open areas of active chromatin. In accordance with our previous work,[@bib31] MLV~IN_W390A~ integration near TSSs and CpG islands decreased ∼2-fold ([Figure 2](#fig2){ref-type="fig"}A; [Table S1](#mmc1){ref-type="supplementary-material"}A). Whereas integration frequencies for MLV~IN_W390A-CDYL~ and MLV~IN_W390A-E2~ did not differ from MLV~IN_W390A~ near TSSs or CpG islands, the integration frequencies for both MLV~IN_W390A-CBX~ and MLV~IN_W390A-LANA~ were ∼4-fold and ∼2-fold lower when compared to MLV~IN_WT~ and MLV~IN_W390A~, respectively ([Figure 2](#fig2){ref-type="fig"}A; [Table S1](#mmc1){ref-type="supplementary-material"}A; p \< 0.001, compared to MLV~IN_W390A~). The detargeting effect of MLV~IN_W390A-CBX~ and MLV~IN_W390A-LANA~ was also evident near DHSs. 33% of MLV~IN_W390A~ integrations occurred in a 2-kb window around DHSs (∼13% less than MLV~IN_WT~), whereas only 25.66% and 27.84% of MLV~IN_W390A-CBX~ and MLV~IN_W390A-LANA~ integrations occurred in this window ([Figure 2](#fig2){ref-type="fig"}A; [Table S1](#mmc1){ref-type="supplementary-material"}A). For comparison, integration datasets of prototype foamy viral vectors (FVs[@bib36]) and HIV-derived lentiviral vectors (LVs[@bib37]) were juxtaposed ([Figure 2](#fig2){ref-type="fig"}A). FVs are known to have a lower tendency to integrate near promoter regions compared to MLV vectors.[@bib36] MLV~IN_W390A~ integration near TSSs was comparable to that of FVs (10.17% and 10.3%, respectively), whereas MLV~IN_W390A-CBX~ and MLV~IN_W390A-LANA~ integration occurred ∼2-fold less near TSSs ([Figure 2](#fig2){ref-type="fig"}A; [Table S1](#mmc1){ref-type="supplementary-material"}A) yet more frequently than LVs.[@bib37] Similar results were obtained for larger window sizes (4-kb window, data not shown). Together, the data confirm that fusion of the CBX1 chromodomain and LANA peptide to MLV IN shifts vector integration away from the traditional markers associated with MLV integration.Figure 2Integration Site Distribution of Next-Generation BinMLV Vectors(A) Murine leukemia virus (MLV)-based vector integration sites obtained from SupT1 cells and their genomic distribution. Integration percentages in 2-kb windows around TSSs, CpG island midpoints, and DHSs are listed. For comparison, integration of computer-generated match random controls (MRCs) for MLV~IN_W390A~, prototype foamy viral vectors (FVs), and HIV-derived lentiviral vectors (LVs) were included. *P* values show significant departures (\*\*\*p \< 0.001, pairwise Fisher test) from MLV~IN_W390A~. (B and C) Heatmaps summarizing the relation between vector integration site frequency and different genomic (B) or epigenetic features within a10-kb interval (C) in SupT1 cells. Evaluated vectors are indicated above the columns. Features analyzed are shown to the left of the corresponding row of the heatmap. Tile colors indicate whether a particular feature is favored or disfavored for integration of the respective datasets relative to their MRCs, as detailed in the colored ROC area scale at the bottom of the panel. Number of integration sites is indicated below each column for the respective vector. *P* values show the significance of departures from MLV~IN_W390A~ integration sites in SupT1 cells (\*\*\*p \< 0.001, Wald statistics referred to χ^2^ distribution). CpG, CpG-rich island; DHS, DNase I-hypersensitive site; TSS, transcription start site.
In a next step, we analyzed integration preferences relative to a wider set of genomic features to evaluate overall vector integration (represented by heatmaps, [Figure 2](#fig2){ref-type="fig"}B). The tile color depicts the correlation for an integration dataset with the respective genomic feature (left) relative to the MRC, as quantified by the area under the receiver operating characteristic (ROC) curve. Asterisks indicate statistical significance of the integration site distributions of the respective vectors relative to that of MLV~IN_W390A~. MLV~IN_W390A-CDYL~ and MLV~IN_W390A-E2~ integration profiles were similar to that of MLV~IN_W390A~ ([Figure 2](#fig2){ref-type="fig"}B), whereas MLV~IN_W390A-CBX~ and MLV~IN_W390A-LANA~ showed significant differences. The addition of CBX1 chromodomain to MLV ~IN_W390A~ (MLV~IN_W390A-CBX~) significantly shifted integration for most of the genomic features (p \< 0.001, compared to MLV~IN_W390A~), resulting in redistributed integration with a more random pattern ([Figure 2](#fig2){ref-type="fig"}B, compare tile color of MLV~IN_W390A~ and MLV~IN_W390A_CBX~; tile colors shift toward gray). On the other hand, MLV~IN_W390A-LANA~ only showed significant effects in smaller window sizes for the typical determinants of MLV vector integration, such as TSSs, CpG islands, and DHSs.
Since some of the peptides recognize specific chromatin marks, we also analyzed integration preferences near a collection of epigenetic features ([Figure 2](#fig2){ref-type="fig"}C). In line with previous data, uncoupling of BET interaction (MLV~IN_W390A~) yields a more random integration pattern (compare colors between MLV~IN_WT~ and MLV~IN_W390A~; tile colors shift toward black).[@bib31] Nonetheless, MLV~IN_W390A~ integration still correlates with histone marks associated with open and transcriptionally active chromatin, such as H3 acetylation, H3K4 mono-, di-, and trimethylation, and H3K36 mono- and trimethylation albeit to a lesser extent (tile colors shift to darker blue; integration is enriched compared to MRC), while disfavoring transcriptionally silent regions or heterochromatin, such as di- and trimethylation of H3K9, H3K27, and H3K79[@bib38] (tile colors shift to yellow; integration is depleted compared to MRC). Fusion of the CBX1 chromodomain and LANA peptide to IN~W390A~ shifted integration away from markers correlating with transcriptionally active open chromatin (blue tiles overall shift toward darker blue; p \< 0.001, compared to MLV~IN_W390A~), whereas the addition of the other peptides had no effect (MLV~IN_W390A-CDYL~ and MLV~IN_W390A-E2~, [Figure 2](#fig2){ref-type="fig"}C). CBX1 is known to bind H3K9me2/3 epigenetic marks via its chromodomain.[@bib32] Interestingly, MLV~IN_W390A-CBX~ shifts integration more into transcriptionally silent heterochromatin regions, which is generally disfavored for integration, marked by di- and/or tri-methylation of H3K9 and H3K27 (p \< 0.001, compared to MLV~IN_W390A~; yellow tiles are darker). Together, these data show that the fusion of peptide tethers to the C-terminal end of MLV IN~W390A~ effectively retargets integration. As integration is shifted away from traditional MLV integration markers, known to associate with insertional mutagenesis in gene therapeutic trials,[@bib39] a potentially safer integration site profile might be obtained.
Addition of Small Peptides to MLV~IN_1--380~ Rescues Its Transduction Defect {#sec2.3}
----------------------------------------------------------------------------
Apart from W390A substitution, deletion of the C-terminal tail of MLV IN (Δ23 aa, IN~1--380~) similarly detargeted MLV integration, but with a transduction efficiency for MLV~IN_1--380~ that was ∼3-fold lower than for MLV~IN_WT~.[@bib31] Therefore, we fused the respective peptides as an alternative C-terminal tail to IN~1--380~ to assess whether transduction efficiency could be improved (MLV~IN_1--380-CBX~, MLV~IN_1--380-CDYL~, MLV~IN_1--380-E2~, and MLV~IN_1--380-LANA~; [Figure 3](#fig3){ref-type="fig"}A). Indeed, the addition of all peptides improved transduction efficiency to near wild-type levels at different MOIs ([Figures 3](#fig3){ref-type="fig"}B and [S1](#mmc1){ref-type="supplementary-material"}C) and resulted in similar expression levels ([Figures 3](#fig3){ref-type="fig"}C and [S1](#mmc1){ref-type="supplementary-material"}D), underscoring the importance of the C-terminal tail.Figure 3Effects of Fusing the Chromatin Binding Peptides in MLV~IN_1--380~ Backbone(A) Schematic representation of IN~WT~, IN~1--380~, and peptides fused to IN~1--380~. The N-terminal HHCC zinc binding domain, the catalytic core domain (CCD), and the C-terminal domain (CTD) are indicated. The size of the fused peptides is proportionally represented. (B and C) TE (B) and mean fluorescence intensity (C) in SupT1 cells 3 days post-transduction at different MOIs. Average values and standard deviations of triplicate measurements are shown. Data represent measurements from a representative experiment.
Next, we amplified 28,607 unique integration sites ([Table S1](#mmc1){ref-type="supplementary-material"}B) to evaluate whether these IN chimeras also redistributed vector integration. In line with earlier data, MLV~IN_1--380~ integration was decreased near TSSs, CpG islands, and DHSs compared to MLV~IN_WT~[@bib31] ([Figure S2](#mmc1){ref-type="supplementary-material"}A; [Table S1](#mmc1){ref-type="supplementary-material"}B). Fusion of CDYL or E2 peptides to MLV~IN_1--380~ did not alter integration site distribution, whereas fusion of the CBX1 chromodomain and LANA peptide redistributed integration sites in a similar fashion as for the MLV~IN_W390A~ fusions (compare [Figures 2](#fig2){ref-type="fig"}A and [S2](#mmc1){ref-type="supplementary-material"}A; compare [Tables S1](#mmc1){ref-type="supplementary-material"}A and S1B). Additionally, MLV~IN_1--380-CBX~ integration redistributed similar to MLV~IN_W390A-CBX~ for a wide range of genomic features ([Figure S2](#mmc1){ref-type="supplementary-material"}B) and histone modifications ([Figure S2](#mmc1){ref-type="supplementary-material"}C), highlighting the specificity of the detargeting effects achieved by fusion of the CBX1 chromodomain or the LANA peptide to MLV~IN_W390A~ and MLV~IN_1--380~.
Addition of Peptide Tethers Does Not Alter the Local MLV Integration Site Sequence {#sec2.4}
----------------------------------------------------------------------------------
Retroviral INs show weak but discernable target sequence preferences surrounding the site of integration. This local integration site sequence is mainly determined by IN contacts with the (nucleosomal) DNA template.[@bib40], [@bib41] To assess whether the addition of alternative peptide tethers to MLV~IN_W390A~ influenced the local integration site sequence, we constructed sequence logos ([Figure S3](#mmc1){ref-type="supplementary-material"}). Results indicate that the local integration site sequence preferences remained unaffected.[@bib42] Similar results were obtained for MLV~IN_1--380~ peptide fusions (data not shown).
Evidence for Safer Integration of Next-Generation BinMLV~IN_W390A-CBX~ Vector {#sec2.5}
-----------------------------------------------------------------------------
Comparative integrome analysis identified MLV~IN_W390A-CBX~ and MLV~IN_W390A-LANA~ as next-generation BinMLV vectors with substantial detargeting from traditional MLV markers without compromising transduction efficiency. Initially, we sought to assess whether the altered integration profiles of next-generation BinMLV vectors are potentially safer. We determined integration frequencies near a set of previously defined genomic regions[@bib43], [@bib44] that should be avoided to prevent cellular transformation, such as regions proximal to TSSs (\<50 kb TSS), cancer-related oncogenes (\<300 kb AllOnco), and microRNA (miRNA) coding regions (\<300 kb miRNA), or within transcription units (TUs) and ultraconserved regions (UCRs). For each integrome dataset, the percentage of integrations near these features was determined ([Table S2](#mmc1){ref-type="supplementary-material"}). When assessing the individual safe harbor criteria, MLV~IN_W390A-CBX~ integrated less frequently near each of these features when compared to MLV~IN_WT~, MLV~IN_W390A~, and LV integration. Taken together, this computational analysis suggests an overall safer integration profile that is less likely to disturb nearby genes.
Therapeutic Potential of Next-Generation BinMLV Vectors {#sec2.6}
-------------------------------------------------------
An essential factor in engineering new viral vectors includes evaluating their therapeutic potential to transduce clinically relevant cells. Therefore, we challenged primary human CD4^+^ T cells and CD34^+^ HSCs with MLV~IN_WT~, MLV~IN_W390A~, MLV~IN_W390A-CBX~, and MLV~IN_W390A-LANA~ vectors carrying a more relevant SIN gammaretroviral vector genome (MLV.SIN) driving reporter gene expression from a spleen focus-forming virus (SF) enhancer/promoter,[@bib21], [@bib45] referred to as MLV.SIN.SF~IN_WT~, MLV.SIN.SF~IN_W390A~, MLV.SIN.SF~IN_W390A-CBX~, and MLV.SIN.SF~IN_W390A-LANA~, respectively ([Figure S4](#mmc1){ref-type="supplementary-material"}A) and monitored transduction efficiency (percentage of eGFP^+^ cells) over time (MOI of 4.5, [Figures 4](#fig4){ref-type="fig"}A and 4B, respectively). MLV.SIN.SF~IN_W390A-CBX~ and MLV.SIN.SF~IN_W390A-LANA~ transduced CD4^+^ T cells and CD34^+^ HSCs to the same extent as MLV.SIN.SF~IN_WT~ and MLV.SIN.SF~IN_W390A~ and this was sustained over time (fluorescence-activated cell sorting \[FACS\] plots, [Figures 4](#fig4){ref-type="fig"}A and 4B, respectively; compare transduction efficiency (TE) and MFI at day 3 and day 10), indicating the absence of silencing effects due to the retargeted integration preference of next-generation BinMLV vectors. Additionally, we performed colony-forming unit (CFU) assays for CD34^+^ HSCs harvested from three different donors to determine whether the retargeted next-generation BinMLV vectors efficiently transduced hematopoietic progenitor cells ([Figure 4](#fig4){ref-type="fig"}C). After 14 days in culture, the number of colonies derived from CD34^+^ HSCs transduced with MLV.SIN.SF~IN_W390A~, MLV.SIN.SF~IN_W390A-CBX~, and MLV.SIN.SF~IN_W390A-LANA~ was in line with MLV.SIN.SF~IN_WT~, indicating that BinMLV vectors do not affect the functionality of CD34^+^ HSCs ([Figure 4](#fig4){ref-type="fig"}C). FACS analysis of the respective pooled transduced CFUs revealed comparable percentages of EGFP-positive cells ([Figure 4](#fig4){ref-type="fig"}D) and no difference in MFIs could be observed ([Table S3](#mmc1){ref-type="supplementary-material"}) for the different BinMLV vectors, confirming the lack of increased transgene silencing.Figure 4Translational Potential of Next-Generation BinMLV Vectors(A and B) TE of next-generation BinMLV vectors in primary CD4^+^ T cells (A) and CD34^+^ HSCs (B) at different time points. Percentage of EGFP-positive cells and mean fluorescence intensities (MFIs) are indicated 3 days and 10 or 13 days post-transduction for CD4^+^ T cells and CD34^+^ cells, respectively. Data represent measurements from a representative experiment. (C) Colony-forming unit (CFU) assay of human CD34^+^ HSCs harvested from three donors and transduced with the indicated vectors. The number of colonies was scored after 14 days. (D) TE (percentage of EGFP-positive cells) of CFU colonies (described in C) at 17 days post-transduction from three donors.
Reduced Transformation Potential of Next-Generation BinMLV~IN_W390A-CBX~ Vector {#sec2.7}
-------------------------------------------------------------------------------
The intrinsic integration preference of MLV-based vectors has been shown previously to be one of the driving factors of vector-mediated genotoxicity that occurred when the integrated vector dysregulated host genes, leading to oncogenic transformation. Therefore, we evaluated the genotoxic potential of next-generation BinMLV vectors carrying a MLV.SIN-vector genome with an internal SF enhancer/promoter, which is known to trigger insertional transformation events[@bib21], [@bib45] ([Figure S4](#mmc1){ref-type="supplementary-material"}A) in the in vitro immortalization assay (IVIM)[@bib46] and the transformational incidence in a murine CFU assay ([Figure S4](#mmc1){ref-type="supplementary-material"}B).[@bib47], [@bib48] To ensure the chance of immortalization, at least 55% of cells were transduced, corresponding to a mean vector copy number (VCN) of \> 2.[@bib49] Murine hematopoietic lineage-depleted (lin^−^) bone marrow cells were transduced with the indicated vectors at different MOIs in six independent transductions in three IVIM assays ([Figure S4](#mmc1){ref-type="supplementary-material"}C). Transduction efficiencies and integrated VCNs were comparable for the different vectors ([Figures 5](#fig5){ref-type="fig"}A and 5B). The IVIM assay revealed immortalization for all MLV vectors carrying the MLV.SIN.SF vector architecture ([Figures 5](#fig5){ref-type="fig"}C and [S4](#mmc1){ref-type="supplementary-material"}C). The replating frequency is a measure for the fitness of clones, while the number of positive assays reflects the incidence of immortalization events. MLV.SIN.SF~IN_W390A~ and MLV.SIN.SF~IN_W390A-LANA~ resulted in replating clone numbers in line with MLV.SIN.SF~IN_WT~, whereas a reduction in clone numbers was observed for MLV.SIN.SF~IN_W390A-CBX~ (mean number of positive wells; mean replating frequency of 2.7 × 10^−3^, [Figure S4](#mmc1){ref-type="supplementary-material"}C). MLV.SIN.SF~IN_W390A-CBX~ also displayed a reduced replating frequency/copy number (1.4-fold or ∼30% lower) compared to MLV.SIN.SF~IN_WT~, although it was not statistically significant (p \> 0.05, compared to MLV~IN_WT~, Mann-Whitney U test, [Figures 5](#fig5){ref-type="fig"}C and [S4](#mmc1){ref-type="supplementary-material"}C).Figure 5Assessment of Genotoxicity Profile of Next-Generation BinMLV Vectors(A and B) TE at day 4 (A) and mean vector copy number at day 5 post-transduction (B) of next-generation BinMLV vectors in murine hematopoietic lineage marker-negative (lin^−^) bone marrow cells at increasing MOIs. MOI 20 was performed only once. (C) Replating frequencies corrected for the mean vector copy number, measured in DNA of mass cultures at day 5; horizontal lines indicate the median for a given vector design. (D and E) Serial CFUs assay. The number of colonies per 5,000 cells plated in methylcellulose is shown. Each dot represents the number of colonies formed for an independently transduced culture. A negative control represents non-transduced lin^−^ cells, while a positive control represents lin^−^ cells transduced with an LTR-SFFV-driven MLV-based vector. After the first round (D), the number of colonies was scored and cells from positive assays were harvested and re-seeded at 5,000 cells for second round colony formation (E). Colony counts \> 200 colonies are shown as 220. The experimental setup is shown in [Figure S4](#mmc1){ref-type="supplementary-material"}B. VCN, vector copy number.
The serial replating CFU assay is a method used to confirm cellular anchorage-independent growth in vitro. The assay provides a stringent method for the detection of the tumorigenic potential of transformed murine HSCs in semi-solid medium.[@bib47], [@bib48] In a parallel approach, we sought to assess the serial colony-forming capacity of the transduced lineage marker-negative cells 2 weeks after expansion in the IVIM assay. We included cells transduced with an LTR-driven MLV-based vector as an additional positive control.[@bib46] Cells were plated at a density of 5,000 cells/well in semi-solid medium ([Figure S4](#mmc1){ref-type="supplementary-material"}B). After 10 days in culture (first round, [Figures 5](#fig5){ref-type="fig"}D and [S4](#mmc1){ref-type="supplementary-material"}B), the number of colonies obtained for all MLV.SIN.SF vectors was not significantly different from that of non-transduced cells (p \> 0.05, compared to negative control, Mann-Whitney U test, [Figure 5](#fig5){ref-type="fig"}D), whereas the number of colonies for the positive control was significantly higher (p \< 0.05, compared to negative control, Mann-Whitney U test, [Figure 5](#fig5){ref-type="fig"}D). After isolating the cells from the colony assays, 5,000 cells were replated (second round, [Figures 5](#fig5){ref-type="fig"}E and [S4](#mmc1){ref-type="supplementary-material"}B). MLV.SIN.SF~IN_WT~- and MLV.SIN.SF~IN_W390A~-transduced cells formed numbers of colonies comparable to the positive control (p \> 0.05, compared to positive control, Mann-Whitney U test, [Figure 5](#fig5){ref-type="fig"}E), whereas the number of colonies from MLV.SIN.SF~IN_W390A-CBX~ transduced cells was significantly reduced (p \< 0.05 compared to MLV.SIN.SF~IN_WT~, Mann-Whitney U test, [Figure 5](#fig5){ref-type="fig"}E). Taken together, these data indicate that engineering the MLV vector configuration by modifying the IN protein (IN~W390A-CBX~) reduced the outgrowth of replating clones, indicating a potentially safer profile.
Discussion {#sec3}
==========
Retroviral vector technology offers great potential to treat genetic disorders and is a powerful tool for long-term correction of genetic defects in a variety of severe hematological disorders.[@bib1], [@bib50], [@bib51] Despite the initial success, a subset of patients developed serious adverse events, such as leukemia or myelodysplasia, that could be directly related to the design and characteristics of the viral vector used.[@bib10], [@bib11], [@bib12], [@bib13], [@bib14] Understanding the mechanisms of retroviral vector genotoxicity is, therefore, essential to engineer improved viral vectors with reduced genotoxic potential.
Vector-mediated genotoxicity is defined by (1) the specific integration profile and (2) the design of the integrating proviral genome. Each retroviral family displays a specific integration profile. MLV integration is significantly enriched near TSSs and active enhancer regions[@bib38] and thus potentially causes insertional mutagenesis. Significant efforts have been made to develop next generations of retroviral vectors with reduced genotoxic potential, such as SIN vectors and weaker internal promoters.[@bib19], [@bib21], [@bib52], [@bib53], [@bib54], [@bib55], [@bib56], [@bib57], [@bib58] Nonetheless, their integration is still targeted to gene regulatory regions,[@bib38], [@bib59] where they have the potential to disrupt or dysregulate the transcription of nearby genes by other mechanisms.[@bib60] Thus, the development of viral vectors that integrate away from genes may be safer for clinical applications. One approach is the development of other viral vector platforms with more favorable genomic distributions, like lentiviral, foamy, and alpharetroviral vectors.[@bib25], [@bib60], [@bib61], [@bib62] Alternatively, retroviral vectors can be re-engineered to obtain a more desirable integration pattern that is detargeted from its traditional chromosomal locations and ultimately only occurs at the preferred sites of the host-cell chromosome.
Retroviral integration site selection is dictated by the interaction between the viral IN as part of the PIC and cellular cofactors. Previously, we re-engineered the MLV- and HIV-cellular tethering cofactors (BET and LEDGF/p75, respectively) and demonstrated efficient redistribution of retroviral integration without compromising transgene expression.[@bib29], [@bib63], [@bib64], [@bib65] However, this approach requires the introduction (at least transient) of artificial anchors in target cells prior to application of the therapeutic vectors,[@bib63], [@bib64] which is not always desirable in a clinical setting. A more straightforward strategy is to directly engineer vector particles to contain proteins with adapted or unique binding domains to direct integration. Other groups attempted to redistribute MLV integration through modifications of the MLV Gag protein by fusion of tethering peptides to rescue a mutated MLV p12 protein.[@bib66], [@bib67] However, integration site distribution of the engineered MLV p12 chimeras was not altered,[@bib66], [@bib67] suggesting that the primary role of MLV p12 is tethering the virus/vector to the condensed host-cell chromatin rather than targeting the genomic integration (reviewed in Rein[@bib68]). Analysis of MLV IN-BET interaction[@bib29] enabled us to generate Bin MLV vectors.[@bib31] Interaction with BET proteins was uncoupled either by deletion of the C-terminal tail of MLV IN (MLV~IN_1--380~) or by a single point mutation (MLV~IN_W390A~).
Here, we successfully modified the MLV integration profile by fusing alternative chromatin binding peptides to BinMLV IN, resulting in a significantly different integration site pattern ([Figure 2](#fig2){ref-type="fig"}). The addition of the CBX1 chromodomain or LANA peptide redistributed MLV integration to the same extent relative to the typical MLV markers (TSSs, DHSs, and CpG islands), with a more than 3-fold reduction in integration frequency near TSSs and CpG islands compared to wild-type MLV ([Figures 2](#fig2){ref-type="fig"}A and [S2](#mmc1){ref-type="supplementary-material"}A; [Table S1](#mmc1){ref-type="supplementary-material"}). Comparative integrome analysis near a range of genomic and epigenetic features corroborated that IN~LANA~ vectors (MLV~IN_W390A-LANA~ and MLV~IN_1--380-LANA~) and IN~CBX~ vectors (MLV~IN_W390A-CBX~ and MLV~IN_1--380-CBX~) redistribute along the chromatin, although the effect was much more pronounced for IN~CBX~ vectors ([Figures 2](#fig2){ref-type="fig"}B and 2C and [S2](#mmc1){ref-type="supplementary-material"}B and S2C). A similar strategy was employed by Hocum et al.,[@bib69] where the full CBX1 protein was fused to the IN of foamy virus and additional modifications in the FV Gag protein were required to achieve significant effects on retargeting FV integration. The newly generated foamy retroviral vectors (FV) integrated ∼2-fold less frequent near genes and proto-oncogenes.[@bib69] In the case of MLV, fusion of a single CBX1 chromodomain to the C-terminal end of MLV IN~W390A~ was sufficient to shift integration toward epigenetic markers for transcriptionally silent regions (H3K9me2/3 and H3K27me3), known to be bound by CBX1[@bib32] ([Figure 2](#fig2){ref-type="fig"}C; [Table S1](#mmc1){ref-type="supplementary-material"}).
Next to the retargeting effect, next-generation BinMLV vectors can be produced at high titers and efficiently transduce clinically relevant cells such as primary CD4^+^ T cells and CD34^+^ HSCs without any apparent transgene silencing ([Figure 4](#fig4){ref-type="fig"}), highlighting the translational potential of these vectors. Finally, we assessed the newly engineered Bin MLV vectors (MLV~IN_W390A-CBX~ and MLV~IN_W390A-LANA~) in the IVIM assay as well as in a murine serial replating CFU assay to predict the genotoxic profile ([Figure S4](#mmc1){ref-type="supplementary-material"}B).[@bib46] BinMLV and wild-type MLV-based vectors were produced carrying the same SIN gammaretroviral vector genome (MLV.SIN.SF) with an internal SF enhancer/promoter to drive EGFP expression ([Figure S4](#mmc1){ref-type="supplementary-material"}A), a design known to trigger insertional transformation.[@bib21], [@bib45] Whereas MLV.SIN.SF~IN_WT~, MLV.SIN.SF~IN_W390A~, and MLV.SIN.SF~IN_W390A-LANA~ displayed a comparable frequency of IVIM clones, immortalization events were slightly reduced for MLV.SIN.SF~IN_W390A-CBX~ ([Figures 5](#fig5){ref-type="fig"}C and [S4](#mmc1){ref-type="supplementary-material"}C) but were not significantly different. The results of the IVIM assay in our experiments confirmed that the strong spleen focus-forming virus (SFFV) enhancer/promoter present in all vector genomes produced replating clones with similar efficiency in all cultures independent of the IN, while the low replating in two assays of MLV~IN_W390A-CBX~ may be attributed to the detargeted integration pattern. Hence, at least for SFFV, promoter strength has a greater impact on the generation of insertional mutants than integration site preference in the IVIM assay.
To further address whether detargeted BinMLV vectors may be less genotoxic, we employed serial replating CFU assays of expanded murine HSCs[@bib47], [@bib48] to assess vector integration-related genotoxicity. The murine CFU assays demonstrated that MLV.SIN.SF~IN_WT~ and MLV.SIN.SF~IN_W390A~ potently transformed murine HSCs, whereas fusion of the CBX1 chromodomain to MLV IN~W390A~ (MLV.SIN.SF~IN_W390A-CBX~) resulted in a significantly reduced transformational incidence (p \< 0.05, [Figure 5](#fig5){ref-type="fig"}E).
In conclusion, we demonstrate the potential to engineer MLV-based vectors that detarget from unsafe regions by fusing peptide fragments to the C-terminal end of MLV IN. The mere addition of the CBX1 chromodomain (and LANA peptide, to a lesser extent) was sufficient to detarget integration preference away from the traditional markers of MLV integration. In order to translate these findings into suitable vectors for the (pre)clinical field, the performance of the new Bin MLV should be evaluated in more relevant pre-clinical safety assays and bone marrow/HSC transplantation assays to validate their improved safety profile. Together, our findings will help achieve better control of MLV-based vector integration preferences. Combining next-generation BinMLV packaging constructs with next-generation SIN gammaretroviral vector architectures that incorporate a weaker physiological promoter less likely to dysregulate nearby genes[@bib19], [@bib70] will lead to an additional reduction in genotoxicity.
Materials and Methods {#sec4}
=====================
Plasmids {#sec4.1}
--------
BinMLV IN was cloned as previously described.[@bib31] Chromodomians of CBX and CDYL fusions were cloned with gBlocks (IDT) in PacI- and NotI-digested pcDNA3.MLV.gp packaging plasmid, a kind gift from Prof. Axel Schambach.[@bib71] HPV8 E2 and LANA peptide fusions were introduced into the indicated vectors by an oligonucleotide annealing strategy using the same restriction sites. Oligonucleotide sequences are listed in [Table S4](#mmc1){ref-type="supplementary-material"}. All enzymes were purchased from Thermo Fisher Scientific. The integrity of all plasmids was verified by DNA sequencing.
Cell Culture {#sec4.2}
------------
SupT1 cells were cultured in RPMI-1640 medium (Gibco-BRL/Life Technologies) supplemented with 10% heat-inactivated fetal bovine serum (Gibco-BRL) and gentamicin (50 μg/mL; Gibco-BRL). HeLa cells were cultured in DMEM (Gibco-BRL) supplemented with 8% heat-inactivated fetal bovine serum and gentamicin. All cells were grown in a humidified atmosphere with 5% CO~2~ at 37°C.
Primary Cell Purification {#sec4.3}
-------------------------
Human peripheral blood mononuclear cells (PBMCs) were purified from buffy coats of three different donors, obtained from the Red Cross blood transfusion center, using density-gradient centrifugation (Lymphoprep; Axis-Shield). Primary CD4^+^ T cells were selectively enriched using bi-specific monoclonal antibody (mAb) CD3.8 (0.5 μg/mL, NIH AIDS Reagents Program; <https://www.aidsreagent.org>) for 5 days. CD4^+^ T cells were cultured in RPMI medium supplemented with 15% FBS, gentamycin, interleukin (IL)-2 (100 U/mL; Peprotech), and MEM Non-Essential Amino Acids Solution (MEM NEAA) (50 μg/mL; Gibco-BRL), referred to as T-cell medium (TCM). CD34^+^ HSCs were positively selected with anti-CD34-conjugated microbeads according to the manufacturer's instructions (MACS; Miltenyi Biotec) and stimulated for 48 hr in StemSpan SFEMII medium containing CC100 Cytokine Cocktail (STEMCELL Technologies).
Retroviral Vector Production and Transduction {#sec4.4}
---------------------------------------------
Viral vectors were produced as previously described.[@bib72] Briefly, MLV-based vectors were produced by a triple polyethylenimine (PEI)-based or Ca-phosphate transfection of 293T cells with a pVSV-G envelope, pcDNA3.MLV.gp packaging plasmids or their derived fusions (see above), and p450-GFP transfer plasmid (kindly provided by F.D. Bushman) encoding an LTR-driven EGFP reporter. For the IVIM assay, the transfer plasmid pSRS11.SF.GFP.pre (referred to here as MLV.SIN.SF.EGFP.pre) was used, which was kindly provided by Axel Schambach.[@bib71]
Produced vectors were concentrated by tangential flow filtration and normalized based on RT units (RTUs; non-functional titration) by the SYBR Green I product-enhanced reverse transcriptase assay (SG-PERT).[@bib73] Subsequently, functional transducing titers were determined in SupT1 cells reaching titers \> 10^7^ TU/mL. For transduction of laboratory cell lines, SupT1 cells (12 × 10^4^/well) and HeLa cells (2 × 10^4^/well) were seeded in 96-well plates and transduced with a MOI of 1 and 3 of the respective vectors.
Seventy-two hours post-transduction, 50% of the cells were harvested for FACS analysis, while the remaining 50% were cultured for 10 days post-transduction for a second FACS analysis and to perform integration site sequencing. Prior to primary cell transduction, CD4^+^ T cells (25.104/well) and CD34^+^ HSCs (10.104/well) were pre-stimulated for 5 days in TCM and 2 days in StemSpan medium enriched with CC100 Cytokine Cocktail, respectively. An MOI of 4.5 of the different vectors was applied by spinoculation (2 hr, 1,200 g). Cells were analyzed for EGFP expression by flow cytometry at the indicated time points.
Genomic DNA Isolation and qPCR {#sec4.5}
------------------------------
Genomic DNA (gDNA) isolation and qPCR were performed as previously described.[@bib31] Briefly, 2 million cells were pelleted and genomic DNA was extracted using a Mammalian Genomic DNA Miniprep Kit (Sigma-Aldrich). Samples corresponding to 700 ng genomic DNA were used for analysis. Each reaction contained 12.5 μL iQ Supermix (Bio-Rad), 40 nM forward and reverse EGFP primer, and 40 nM EGFP probe in a final volume of 25 μL. RNaseP or β-actin was quantified as the endogenous control (TaqMan RNaseP control reagent; Applied Biosystems). Samples were run in triplicate for 3 min at 95°C followed by 50 cycles of 10 s at 95°C and 30 s at 55°C in a LightCycler 480 (Roche Applied Science). Analysis was performed using the LightCycler 480 software supplied by the manufacturer.
Recovery of Integration Sites and Analysis of Integration Site Distributions {#sec4.6}
----------------------------------------------------------------------------
Recovery of integration sites was performed as previously described.[@bib29] Briefly, linkers were ligated to restriction enzyme-digested (MseI) genomic DNA isolated from transduced cells and virus-host DNA junctions were amplified by nested PCR. Samples were individually barcoded with the second pair of PCR primers to generate 454 libraries. PCR products were purified and sequenced using 454/Roche pyrosequencing (titanium technology). Reads were quality-filtered by requiring perfect matches to the LTR linker, barcode, and flanking LTR and were subsequently mapped to the human/mouse genome. All sites were required to align to the reference genome within 3 bp of the LTR edge. To control for possible biases in the datasets due to the choice of the MseI restriction endonuclease in cloning integration sites, random control sites were generated computationally and matched to experimental sites with respect to the distance to the nearest MseI cleavage site (MRC).[@bib74], [@bib75] To do so, each experimental integration site was paired with three random control sites in the genome with respect to the distance to the nearest MseI cleavage site in the genome. A more detailed explanation can be found in the supplemental guidelines included in Ocwieja et al.[@bib76]
Analyses were carried out as described in Marshall et al.[@bib77] A detailed account of the statistical methods used and the methods for forming and analyzing heatmaps using ROC curves can be found in Brady et al.[@bib78] Consensus sequence analysis at the point of integration was performed using WebLogo3 (<http://140.114.98.75/weblogo/>). For association with specific genomic features, the distance of each integration site (in kilobases) to the respective genomic feature was calculated (midpoint of the CpG island or DHS, and the X5-end of genes as a measure for the TSS). Integration sites upstream of the genomic feature were given negative kilobase values, while downstream integration sites were calculated as positive. For heatmaps, comparisons were carried out over three different interval sizes surrounding each integration site (5 kb, 10 kb, and 50 kb), since previous studies have shown that the interval sizes chosen for comparison can influence the conclusions. In this study, results were in line for each interval size examined (data not shown). Only the data for the 10-kb interval are shown. In the heatmap, the distribution of experimental MLV sites is normalized to that of the MRC sites, as a control for recovery bias due to cleavage by restriction enzymes (in our case, MseI).[@bib74], [@bib75] Results of statistical tests comparing the distributions of integration sites to the reference dataset are summarized as asterisks on each tile of the heatmap. Datasets used in the safe harbor analysis were retrieved from the Ensembl and/or UCSC databases (TxDB knownGenes, miRNA biotype, UCR; hg19) using BioMart.[@bib79] The AllOnco list was used for oncogenes as published in Sadelain Analysis was performed according to the parameters defined in Papapetrou.
CD34^+^ HSC CFU Assay {#sec4.7}
---------------------
For human CD34^+^ HSC CFU assays, freshly purified cells were seeded at 5,000 cells/well in human methylcellulose medium (methocult H4230; STEMCELL Technologies) supplemented with CC100 Cytokine Cocktail. Cells were plated in 35-mm petri dishes and cultured in a fully humidified atmosphere with 5% CO~2~ at 37°C for 14 days. The number of colonies was scored after 14 days. For FACS analysis of the CFUs, colonies were harvested, washed twice with PBS, and evaluated for EGFP expression.
IVIM Assay {#sec4.8}
----------
The IVIM assay was performed as described earlier.[@bib46] Briefly, murine lineage marker-negative bone marrow cells were isolated from the tibias and femurs of C57BL/6 mice and enriched for stem and progenitor cells (mouse lineage cell depletion kit; Miltenyi Biotec) and frozen in aliquots. After thawing and 48 hr of prestimulation, 1 × 10^5^ lin^−^ cells were transduced on 2 consecutive days on RetroNectin-coated (Takara; Clontech) wells with a MOI of 5, 10, or 20. TE was analyzed by flow cytometry 4 days thereafter. Cells were expanded for 2 weeks in IMDM, 10% FCS, 1 mM glutamine, 1% (v/v) penicillin/streptomycin, murine stem cell factor (mSCF) (50 ng/mL), human FMS-like tyrosine kinase 3 ligand (hFlt3L) (100 ng/mL), murine IL (mIL)-3 (20 ng/mL), and human IL (hIL)-11 (100 ng/mL; all cytokines purchased from Peprotech) and diluted to a cell density of 500,000 cells/mL approximately twice a week. Cells were the seeded on 96-well suspension plates at a density of 100 cells per well (48 wells seeded from each culture). Replating clones were detected by microscopic scoring. The replating frequency (according to Poisson distribution) was calculated with L-calc (STEMCELL Technologies) and normalized by VCN as determined 5 days post-transduction.
Murine Serial CFU Assay {#sec4.9}
-----------------------
Murine lineage marker-negative bone marrow (BM) cells were purified and transduced as in the IVIM assay. After 2 weeks of expansion, cells were plated (5,000 cells/well) in methylcellulose (HSC006; R&D Systems) supplemented with 20 ng/mL interleukin-3, and 50 ng/mL murine stem cell factor. The number of colonies was scored after 10 days. For serial replating, colonies were harvested and replated in fresh methylcellulose medium at the same density of 5,000 cells/well for the subsequent round.
Author Contributions {#sec5}
====================
S.E.A., D.V.L., and R.G. designed the experiments. S.E.A., D.V.L., and R.G. wrote the manuscript. S.E.A., D.V.L., J.D.R., F.S., U.M., and R.G. performed experiments and analyzed data. J.D. and L.V. performed bioinformatics for integration site analysis. R.G. and Z.D. supervised the project. All authors read and approved the final manuscript.
Supplemental Information {#app2}
========================
Document S1. Figures S1--S4 and Tables S1--S4Document S2. Article plus Supplemental Information
We thank Paulien Van de Velde and Irina Thiry for excellent technical assistance, as well as Dr. Annelies Michiels for critical reading of the manuscript. Viral vector production was performed at the Leuven Viral Vector Core. This work was supported by grants from the KU Leuven Research Council (OT/13/098-3M130157), the KU Leuven Interdisciplinary Research Programmes (IDO; Interdisciplinaire onderzoeksprogramma's) (IDO/12/008-3E130241), and the Belgian IAP Belvir (P7/45-P-3M120222), IWT (3M140565 SB/151139), and FWO Vlaanderen (3M120412 Asp/12 and G0B3516N). U.M. and F.S. are supported by funding from the LOEWE Centre of Cell and Gene Therapy Frankfurt (HMWK III L 4-518/17.004 \[2013\]). S.EA is a doctoral fellow supported by Interdisciplinary Research Programmes. D.V.L and L.V. are doctoral fellows of the Research Foundation--Flanders (FWO-SB), and J.D. is a postdoctoral fellow of the Research Foundation--Flanders (FWO).
Supplemental Information includes four figures and four tables and can be found with this article online at [http://dx.doi.org/10.1016/j.omtn.2017.04.002](10.1016/j.omtn.2017.04.002){#intref0020}.
[^1]: These authors contributed equally to this work.
[^2]: Present address: The Francis Crick Institute, NW1 1AT London, UK
[^3]: Overview of the aa sequences and binding sites of CBX1 and CDYL chromodomains, HPV8 E2, and KSHV LANA used in this study. The aa position of each peptide in the respective protein is indicated in superscript. Protein sequences were downloaded from the UniProt database.CBX1, chromobox homolog 1; CD, chromodomain; CDYL; chromodomain protein Y-like; HPV, human papilloma virus; KSHV, Kaposi sarcoma human virus; LANA, latency-associated nuclear antigen.
| {
"pile_set_name": "PubMed Central"
} |
1. Introduction {#sec1}
===============
RIS, structurally, a benzisoxazole derivative, is one of the most frequently prescribed atypical antipsychotics in the management of schizophrenia, mood disorders, and autism \[[@B1]\]. Its mechanism of action is not entirely clear. Although the use of atypical antipsychotic drugs has been successful in the treatment of schizophrenia, they can provoke some complications, including weight gain, sedation, movement problems, sleepiness, vision difficulties, constipation, hyperprolactinemia, and extrapyramidal side effects \[[@B2]\]. Patients using these drugs tend to disrupt treatment primarily due to their adverse effects. In addition, RIS is a highly potent drug, extensively metabolized in the liver and excreted by the kidneys \[[@B3]\]. RIS and its primary active metabolite, 9-hydroxyrisperidone, are eliminated by the kidneys. In patients with moderate to severe renal disease, clearance of the sum of the parent drug and metabolite has been demonstrated to decrease by 60% compared to healthy subjects \[[@B4]\].
There are a few cases reporting atypical antipsychotic drug-associated renal damage in patients (including those using RIS), but the mechanism that leads to RIS-related injury is not well understood \[[@B5]\]. Adverse outcomes potentially attributable to these drugs, such as hypotension, acute urinary retention, and rhabdomyolysis, are known to cause this injury \[[@B6]\]. Moreover, pneumonia, acute myocardial infarction, and ventricular arrhythmia have been associated with these drugs in former population-based studies and renal damage may also co-occur with these events \[[@B7]\].
Nowadays, medicinal plants are a considerable source of drug synthesis. RSV, trans-3,5,4′-trihydroxy stilbene, has potent antioxidative and anti-inflammatory properties \[[@B8]\]. It is a polyphenolic phytoalexin present in many edible plants, including mulberries, peanuts, and grapes \[[@B9]\]. RSV has been demonstrated to exhibit a wide range of health-promoting benefits for the coronary, neurological, hepatic, and cardiovascular systems. It has been shown to inhibit inflammation, viral infection, oxidative stress, platelet aggregation, and the growth of a variety of cancer cells \[[@B10]\]. In addition, RSV has been studied in vivo and in vitro \[[@B11]\] and has been shown to possess a series of pharmacological benefits, including nephroprotective effects \[[@B12]\], as a result of its antioxidant and cytoprotective properties \[[@B13]\]. The antioxidant properties of RSV are mainly dependent upon the upregulation of endogenous cellular antioxidant systems, but the compound also displays direct reactive oxygen species (ROS) scavenging properties \[[@B14]\]. The high potency and low systemic toxicity of RSV make it a promising alternative to conventional therapeutic drugs. The fact that RIS exposure induces an excessive increase in metabolic alterations suggests that RSV could be used as an alternative therapy.
To our knowledge, there is no report regarding the protective and therapeutic effects of RSV on RIS-induced renal damage because it has not been studied until now. Therefore, the present study was designed to investigate the possible beneficial impact of oral supplementation with RSV against RIS-induced renal damage in rats for the first time. To achieve our goal, we performed several biochemical and histological analyses in female rats.
2. Materials and Methods {#sec2}
========================
2.1. Chemicals {#sec2.1}
--------------
RIS was bought from Janssen Turkey, Istanbul, and RSV (CAS number 501-36-0) was purchased from Sigma-Aldrich (St. Louis, MO, USA). They were dissolved in distilled water. All other chemicals used were of the best analytical grade.
2.2. Animals {#sec2.2}
------------
In this study, thirty-five female Sprague Dawley rats were used. Healthy adult (twelve weeks old) rats weighing 250--300 g were obtained from the Fırat University Laboratory Animal Production and Research Center. All animal care and follow-up procedures were performed at this center. The experiments were performed in accordance with the protocol approved by Fırat University Faculty of Medicine, Laboratory Animals Ethics Committee (protocol number 2016/25). The rats were kept at 21 ± 1°C for 12 h in a light-dark cycle, were fed standard rat chow, and drank tap water ad libitum. RIS and RSV were administered to the rats for two weeks.
2.3. Experimental Design {#sec2.3}
------------------------
In our study, 35 rats were randomly assigned to five groups with an equal number in each group. We used the simple randomization technique in this experimental study. Thirty-five female Sprague Dawley rats were divided into five groups as follows: group 1, control; group 2, RIS; group 3, RIS+RSV-1; group 4, RIS+RSV-2; and group 5, RIS+RSV-3. The control group was given physiological saline solution by gavage once a day. 2 mg/kg/day RIS was administered by gavage for two weeks to all groups, except the control group. 20 mg/kg/day RSV was given for two weeks to the RIS+RSV-1 group. 40 mg/kg/day RSV was given for two weeks to the RIS+RSV-2 group. 80 mg/kg/day RSV was given for two weeks to the RIS+RSV-3 group. All compounds were suspended in physiological saline solution and administered by gavage once a day. Body weight was recorded at the beginning and end of the study. The treatment course lasted two weeks for all groups. At the end of the second week of the treatment period, the animals were euthanized by exsanguination through cardiac puncture under diethyl ether anesthesia. Before killing, rats were individually weighed, venous blood samples were collected, and serum samples for biochemical analysis were separated by centrifuge at 2800*g* for 15 min and then stored at −80°C. Kidneys were surgically removed, weighed, and stored at −80°C for subsequent biochemical and pathohistological analysis.
The daily dose of RIS varies between 0.25 and 16 mg, and a frequently prescribed dose in previous studies was 2 mg \[[@B15]\]. The maximum recommended human dose (MRHD) of RIS is 0.4--12 mg/day. The experimental dose was calibrated as 2.0 mg/kg (MHRD × 10) on the basis of per kg body weight per day and its suitability to the rat animal model \[[@B16]\]. Therefore, a 2 mg dose was selected for the present study. The assigned dosage of powdered RIS (2 mg/kg once a day for two weeks) was administered by a gastric tube daily between 8:00 and 9:00 a.m. in line with a previous report \[[@B17]\]. RSV (20, 40, and 80 mg/kg body weight/day) was also administered by a gastric tube daily between 8:00 and 9:00 a.m. The dose and duration of RSV were selected according to results from a previous study \[[@B18]\].
2.4. Biochemical Analysis {#sec2.4}
-------------------------
BUN was measured using a commercially available enzymatic colorimetric method and analyzer system (Hitachi 917 modular device; Roche Diagnostics, Basel, Switzerland). Serum Cr was measured by Jaffe\'s method using the same analyzer system \[[@B19]\].
2.5. TAS and TOS Determination {#sec2.5}
------------------------------
The automated calorimetric measurement methods developed by Erel were used to define the TAS (mmol/L) and TOS (*μ*mol/L) (a serum oxidant parameter). The measurement of TAS and TOS in serum samples was determined by a TAS and TOS kit (REL Assay Diagnostics) \[[@B20], [@B21]\].
2.6. OSI Determination {#sec2.6}
----------------------
OSI was determined as TOS to TAS ratio and was calculated as follows: OSI (arbitrary unit) = ((TOS, *μ*mol H2O2 eq/L)/(TAS, *μ*mol Trolox eq/L)) \[[@B22]\].
2.7. Terminal Deoxynucleotidyl Transferase dUTP Nick End Labeling (TUNEL) Assay {#sec2.7}
-------------------------------------------------------------------------------
Apoptotic cells were defined using the ApopTag Plus Peroxidase In Situ Apoptosis Detection Kit (Chemicon, cat. number: S7101, USA) according to the manufacturer\'s instructions. Sections (5 *μ*m) taken from the paraffin blocks were entrenched onto polylysine-coated slides, deparaffinized using xylene, dehydrated with a series of alcohol rinses, and then washed with phosphate-buffered saline (PBS). Then, tissues were incubated with a proteinase K solution (0.05%) and with 3% hydrogen peroxide for 5 min to forestall endogenous peroxidase activity. After washing with PBS, the tissues were incubated with Equilibration Buffer for 6 min and in working solution (70% *μ*L reaction buffer + 30% TdT enzyme) at 37°C under moist conditions for 60 min. Tissues were then incubated in stop/wash buffer for 10 min and incubated in anti-digoxigenin-peroxidase for 30 min. Apoptotic cells were examined using the diaminobenzidine substrate. Cross sections contrast stained with methyl green were sealed using a proper covering solution. Mamma tissue was used as a positive control. PBS was used instead of the TdT enzyme in the negative control tissue. Preparations were examined and assessed using a research microscope (Olympus BH2 light microscope) and then photographed. To assess the TUNEL staining, after staining with methyl, green cells with green nuclei were considered normal whereas cells with brown nuclei were considered apoptotic. Apoptotic (TUNEL positive) cells were counted and statistically assessed. This analysis was made in at least eight areas of each kidney section (two sections/animal), and the sections were analyzed at 400x magnification \[[@B23]\].
The evaluation of TUNEL staining was made based on the extent of the staining of apoptotic cells. The extent of TUNEL staining was scored semiquantitatively as 0 (none), 1 (light), 2 (medium), and 3 (intense) \[[@B24]\].
2.8. Statistical Analyses {#sec2.8}
-------------------------
Statistical analysis was performed using SPSS 16.0 (SPSS, Chicago, IL, USA). All data were expressed as mean values ± their standard errors (SEM). Normality for variables in the groups was determined by the Shapiro-Wilk test. For the comparison of the mean weight of all groups, a paired *t*-test was performed. The groups were compared with the paired-samples *t*-test at the beginning and end of the treatment. One-way analysis of variance (ANOVA) followed by the LSD post hoc test were used for the comparison of biochemical parameters and total oxidant/antioxidant levels. Significance was considered at the *p* \< 0.05 level. For histopathological analysis, results were expressed as the means ± standard deviation (SD). The statistically significant difference was determined by ANOVA followed by Tukey\'s multiple comparison test. Probability values (*p*) less than 0.05 were considered to be statistically significant.
3. Results {#sec3}
==========
3.1. Effects of RIS and RSV on Weight Gain/Loss {#sec3.1}
-----------------------------------------------
Body weight measurements showed that during the two weeks, animals grew from 238.28 g at day one to 252.85 g for the control group, from 234.57 g at day one to 248.00 g for the RIS group, from 225.28 g at day one to 233.71 g for the RIS+RSV-1 group, from 232.40 g at day one to 226.80 g for the RIS+RSV-2 group, and from 244.80 g at day one to 246.80 g for the RIS+RSV-3 group at day 14 ([Table 1](#tab1){ref-type="table"}: paired-samples *t*-test for the body weight at day 14; *p* = 0.000, *p* = 0.005, *p* = 0.005, *p* = 0.071, and *p* = 0.537, resp.).
There was a significantly increased total body weight gain in the control, RIS, and RIS+RSV-1 treatment groups (*p* = 0.000, *p* = 0.005, and *p* = 0.005, resp.). However, the RIS+RSV-2 group was observed to have a decreased weight gain and the RIS+RSV-3 group an increased weight gain had no significant effect on these measurements (*p* = 0.071, *p* = 0.537, resp.) ([Table 1](#tab1){ref-type="table"}; [Figure 1](#fig1){ref-type="fig"}).
3.2. Effects of RIS and RSV on Biochemical and Oxidative Stress Parameters {#sec3.2}
--------------------------------------------------------------------------
We measured levels of biochemical parameters in serum, the results are shown in [Table 2](#tab2){ref-type="table"}, and oxidative stress parameters are shown in [Table 3](#tab3){ref-type="table"}. Serum Cr level was significantly increased in the RIS group compared to the control, RIS+RSV-1, RIS+RSV-2, and RIS+RSV-3 groups (*p* \< 0.00). Serum Cr level was significantly increased in the RIS+RSV-1 group compared to the RIS+RSV-2 and RIS+RSV-3 groups (*p* = 0.01, *p* = 0.00, resp.) ([Table 2](#tab2){ref-type="table"}; [Figure 2](#fig2){ref-type="fig"}).
BUN level was significantly increased in the RIS group compared to the control, RIS+RSV-1, and RIS+RSV-2 groups (*p* = 0.000, *p* = 0.000, and *p* = 0.001, resp.). BUN level was significantly decreased in the RIS+RSV-1 group compared to the RIS, RIS+RSV-2, and RIS+RSV-3 groups (*p* = 0.000, *p* = 0.000, and *p* = 0.000, resp.) ([Table 2](#tab2){ref-type="table"}; [Figure 2](#fig2){ref-type="fig"}).
Ameliorative effects of RSV treatment against RIS administration significantly increased the TAS level and decreased TOS and OSI levels (*p* \< 0.05). The control group had a significantly higher TAS level compared to the RIS group (*p* = 0.014). The RIS+RSV-1 group had a significantly higher TAS level compared to the control, RIS, RIS+RSV-2, and RIS+RSV-3 groups (*p* = 0.037, *p* = 0.000, *p* = 0.002, and *p* = 0.000, resp.). The RIS+RSV-2 group had a significantly higher TAS level compared to the RIS+RSV-3 groups (*p* = 0.011). TOS level was significantly higher in the RIS group compared to the control, RIS+RSV-1, RIS+RSV-2, and RIS+RSV-3 groups (*p* \< 0.001). The RIS+RSV-1 group had a significantly higher TOS level compared to the RIS+RSV-3 group (*p* = 0.003). OSI level was significantly higher in the RIS group compared to the control, RIS+RSV-1, and RIS+RSV-2 groups (*p* = 0.003, *p* = 0.001, and *p* = 0.002, resp.) ([Table 3](#tab3){ref-type="table"}; [Figure 3](#fig3){ref-type="fig"}).
3.3. Evaluation of Apoptosis in Kidney Tissues {#sec3.3}
----------------------------------------------
The results of the apoptotic index are shown in [Table 4](#tab4){ref-type="table"} and [Figure 4](#fig4){ref-type="fig"}. Using TUNEL assay to detect apoptotic renal tubular cells in the kidney sections, the control group ([Figure 4(a)](#fig4){ref-type="fig"}) showed only a few TUNEL-positive cells. The number of TUNEL-positive cells markedly increased in the RIS group ([Figure 4(b)](#fig4){ref-type="fig"}) compared with the control group (*p* \< 0.05). RIS+RSV-1 ([Figure 4(c)](#fig4){ref-type="fig"}), RIS+RSV-2 ([Figure 4(d)](#fig4){ref-type="fig"}), and RIS+RSV-3 ([Figure 4(e)](#fig4){ref-type="fig"}) groups were similar and showed rare TUNEL-positive cells. Treatment with RSV (RIS+RSV-1, RIS+RSV-2, and RIS+RSV-3 groups) (Figures [4(c)](#fig4){ref-type="fig"}, [4(d)](#fig4){ref-type="fig"}, and [4(e)](#fig4){ref-type="fig"}) reduced the number of TUNEL-positive cells as compared with the RIS group (*p* \< 0.05).
4. Discussion {#sec4}
=============
Certain herbal medicines have been declared to be effective in the treatment of adverse outcomes attributed to atypical antipsychotic drugs, including RIS, and combination treatment with drugs and herbal medicines has been reported to be useful in enhancing treatment influence and reducing recovery time and adverse effects. In the present study, RIS (2 mg/kg/day) plus RSV (20, 40 and 80 mg/kg/day) was given by gavage to rats for two weeks. All doses of RSV caused dose-dependent decreases in weight gain and alleviated renal damage of the animals compared to the group that received only RIS. Thus, the present study aimed to investigate whether RSV implemented a protective effect on RIS-induced renal damage and metabolic side effects in an experimental animal model.
Second-generation antipsychotics (SGAs) are extensively used in several psychiatric disease entities and exert to various extents the risk of antipsychotic-induced weight gain. In a study by Domecq et al., weight gain was associated with the use of RIS \[[@B25]\]. A meta-analysis by Allison et al. in adults estimated that the mean weight gain after 10 weeks of treatment was 2.00 kg for RIS \[[@B26]\]. Weight gain can impair both physical health and psychological well-being. Therefore, it is important to define factors that are associated with the risk of weight gain with atypical antipsychotics. These factors may include lifestyle issues, particularly diet. RSV, being a dietary constituent, is an excellent therapy candidate for disorders with a metabolic origin. Since several studies using animal models of diet-induced obesity have displayed the beneficial effects of RSV on reducing obesity, many clinical trials have been effected to assess its effect in humans. In a study by Gómez-Zorita et al., RSV decreased whole body weight in obese rats by reducing oxidative stress \[[@B27]\]. Dal-Pan et al. postulated resveratrol\'s facilitation of weight loss in nonhuman primate models of obesity by increasing the metabolic rate and suppressing torpor expression. In fact, RSV administration in primate models led to a 13% reduction in energy intake while increasing the resting metabolic rate by 29% \[[@B28]\]. In the current study, RSV cotreatment diminished antipsychotic-induced weight gain with a more obvious effect and significantly decreased only with a 20 mg/kg dose. These results imply that the mechanism of action for RSV occurs by increasing energy expenditure, directly reducing energy storage and inhibition of energy intake. This weight decrement effect of RSV is estimated to be in part attributable to its effects on adipocytes \[[@B29], [@B30]\]. Hence, RSV, a dietary supplement, significantly prevents RIS-induced weight gain, which might suggest a potential effectiveness on human subjects. Therefore, RSV is a safe compound for coadministration with RIS for mitigating antipsychotic-induced weight gain/obesity without influencing their therapeutic action. In addition, these results are certain to influence the choice of protective compound when several options exist and to institute preemptive strategies for weight management.
Several adverse outcomes attributed to atypical antipsychotic drugs are known to cause renal damage. RIS was also described as increasing the risk of this damage in a population study \[[@B31]\]. Serum Cr and BUN were the classical standards to evaluate renal damage \[[@B32]\]. Accordingly, a study conducted by Hsu et al. demonstrated that elevated serum Cr was observed in RIS administration \[[@B33]\]. Therefore, the study of Hsu et al. is an indicator to prove renal damage through the use of RIS. RIS may impair tubular function and result in several renal complications. As a polyphenolic phytoalexin, RSV has been reported to be useful in the prevention of numerous types of kidney disease \[[@B34], [@B35]\], drug-induced renal damage \[[@B12]\], and ischemia-reperfusion and sepsis-induced kidney injuries \[[@B36]\]. Similarly, Wu et al. demonstrated that RSV effectively attenuated renal oxidative stress in the diabetic rat kidney \[[@B37]\]. A number of independent studies further explored the molecular mechanisms of RSV-mediated nephroprotection. The results of the present study demonstrate that RIS exposure produced a significant increase in the level of serum Cr and BUN indicating damaged structural and functional kidney integrity. Consistent with these observations, the beneficial effects of RSV were identified in the prevention of renal tubular damage and dysfunction. In addition, oral supplementation of RSV demonstrated the restoration of the elevated serum Cr and BUN to the normal levels \[[@B38], [@B39]\]. RSV cotreatment ameliorated these changes in all doses, with an especially obvious effect in high doses. These results are supported by data in the literature \[[@B40]\].
Oxidative stress is one of the key mechanisms responsible for renal damage and disease progression. Antioxidants try to fight oxidative stress and minimize its damage. TAS measurement has been used to evaluate the overall performance of the antioxidant system. TOS measurement provides a sensitive index of lipid peroxidation and oxidative stress \[[@B41]\]. The TOS/TAS ratio is termed "OSI," which is an indicator of the oxidative stress degree \[[@B42]\]. In the present study, RIS administration resulted in a decrease in TAS level and an increase in TOS and OSI levels as in former studies \[[@B43]\]. Conversely, we observed that RSV could protect kidneys from RIS-induced renal damage and provide useful changes in the basal levels of stress biomarkers. We found that the serum TAS level increased and the TOS and OSI levels prominently decreased with RSV treatment as reported in previous studies \[[@B44]\]. Similarly, Wong et al. demonstrated that RSV reduced oxidative damage biomarkers during aging in F2 hybrid mice \[[@B45]\]. In addition, RSV cotreatment ameliorated these changes with a more obvious effect in 20 mg/kg doses. The beneficial effects of RSV, observed in the current work, are very likely due to its strong antioxidant properties and may be associated with its constituent compounds. This implies that the RSV presented a nephroprotective activity probably due to its antioxidant capacity. The restoration of tissue antioxidant function by RSV may be attributed to its ability to upregulate antioxidant gene expression. Also, RSV may display its antinephrolithic properties by both inhibiting free radical formation and attenuating the expression of a number of inflammatory mediators \[[@B46]\].
Histological observations added more evidence of the protective effect of RSV. Apoptosis involves interactions among several protein families that regulate activation of various apoptotic markers. More recently, Elmorsy et al. demonstrated that some antipsychotics, including RIS, can induce apoptosis \[[@B47]\]. Apoptosis effects triggered by RIS appear to be cell-type dependent \[[@B48]\]. ROS can also induce cell death via apoptosis in many cell types. Such an effect was also blocked by RSV \[[@B49]\]. RSV is accepted as a potent antioxidant and antiapoptotic agent \[[@B50], [@B51]\]. In the present study, histopathological evaluation of the kidneys showed severe damage ensued through loss of normal architecture, which included vacuolar degeneration and fatty changes in RIS-administered rats. These toxic effects were effectively prevented by RSV treatment. Furthermore, RSV cotreatment ameliorated these changes with a more obvious effect in 80 mg/kg doses. The antiapoptotic effect of RSV based on the evaluated reduction in TUNEL-positive kidney cells may also contribute to its therapeutic impact against RIS. Therefore, the antioxidant activity of this compound may be a potential mechanism for antiapoptotic activity. Accordingly, RSV treatment of the cells against RIS exposure prevented apoptotic cell injury and death. The underlying mechanism of protection of RSV may be associated with the suppression of apoptosis via death receptor-mediated pathways. On the other hand, because the kidneys contain more mitochondria than the other organs, the nephrotoxic effects might also be due to the direct action of RIS on renal mitochondria. Therefore, the reducing effects of RIS on mitochondrial functions may also be important in the pathogenesis of nephrotoxicity \[[@B52], [@B53]\]. Hence, it can be assumed that the antioxidant activity of RSV may be due to the effect on the mitochondria-dependent apoptotic pathway, but this has to be supported with further experimentation. Therefore, RSV may be a good option against RIS-induced side effects. The mechanism of action of RIS has not been fully explained, and its exact renal damage mechanism is still being researched because of insufficient experimental study in the literature histopathologically investigating the effects of RIS on renal cells in rats.
RSV was revealed to facilitate a significant reduction in serum Cr and BUN levels and to alleviate renal damage and dysfunction, which was further verified by apoptosis analysis of renal histology. In addition, RSV reduced weight gain, contributed to the oxygen radical scavenging activity, and increased antioxidant activity to accomplish its protective ability against damage caused by the side effects of RIS. Therefore, RSV may act through a range of mechanisms whose effects might have major therapeutic potential in RIS-induced side effects.
In conclusion, our results clearly indicate that RSV oral supplementation, in certain doses, protects against RIS-induced renal damage. Therefore, RSV may be a clinically promising agent in RIS-induced renal damage and its metabolic side effects. Further studies should be undertaken to examine the potential effect of RSV on renal damage in human and animal models.
BUN:
: Blood urea nitrogen
MRHD:
: Maximum recommended human dose
OSI:
: Oxidative stress index
PBS:
: Phosphate-buffered saline
RIS:
: Risperidone
ROS:
: Reactive oxygen species
RSV:
: Resveratrol
Cr:
: Creatinine
SGAs:
: Second-generation antipsychotic drugs
TAS:
: Total antioxidant status
TOS:
: Total oxidant status
TUNEL:
: Terminal deoxynucleotidyl transferase dUTP nick end labeling.
Conflicts of Interest
=====================
The authors have no conflict of interest.
{#fig1}
{#fig2}
{#fig3}
{#fig4}
######
Body weight (g) of animals during treatment.
Design of treatment Control RIS RIS+RSV-1 RIS+RSV-2 RIS+RSV-3
----------------------------------------------------------- --------------- --------------- --------------- --------------- ---------------
Initial study 238.28 ± 7.53 234.57 ± 7.27 225.28 ± 4.16 232.40 ± 3.55 244.80 ± 9.88
Final study 252.85 ± 7.46 248.00 ± 5.86 233.71 ± 3.68 226.80 ± 5.25 246.80 ± 9.42
Statistical comparison (initial study versus final study) (*p*)
Control **0.000**
RIS **0.005**
RIS+RSV-1 **0.005**
RIS+RSV-2 **0.071**
RIS+RSV-3 **0.537**
Changes in the body weight of experimental rats. Values are expressed as mean ± SEM of seven animals. The groups were compared with the paired-samples *t*-test at initial and final treatment. *p* ≤ 0.05. RIS: risperidone; RSV: resveratrol; RIS+RSV-1: 2 mg/kg RIS+20 mg/kg RSV; RIS+RSV-2: 2 mg/kg RIS+40 mg/kg RSV; RIS+RSV-3: 2 mg/kg RIS+80 mg/kg RSV.
######
Levels of serum biochemical parameters for all groups.
Parameters Control RIS RIS+RSV-1 RIS+RSV-2 RIS+RSV-3 *p*
------------- --------------------- ---------------------- --------------------- --------------------- -------------------- -----------
Scr (mg/dL) 0.46 ± 0.00^b,d,e^ 0.53 ± 0.00^a,c,d,e^ 0.45 ± 0.01^b,d,e^ 0.38 ± 0.01^a,b,c^ 0.37 ± 0.01^a,b,c^ **0.000**
BUN (mg/dL) 18.71 ± 0.68^b,d,e^ 25.85 ± 0.45^a,c,d^ 19.85 ± 0.50^b,d,e^ 23.28 ± 0.42^a,b,c^ 24.57 ± 0.48^a,c^ **0.000**
Each group represents the mean ± SEM for seven rats. ^a^*p* \< 0.01 versus the control group; ^b^*p* \< 0.01 versus the RIS group; ^c^*p* \< 0.01 versus the RIS+RSV-1 group; ^d^*p* \< 0.01 versus the RIS+RSV-2 group; and ^e^*p* \< 0.01 versus the RIS+RSV-3 group. RIS: risperidone; RSV: resveratrol; Scr: serum creatinine; BUN: blood urea nitrogen; RIS+RSV-1: 2 mg/kg RIS+20 mg/kg RSV; RIS+RSV-2: 2 mg/kg RIS+40 mg/kg RSV; RIS+RSV-3: 2 mg/kg RIS+80 mg/kg RSV.
######
Comparison of serum oxidative stress parameters among the groups.
Parameters Control RIS RIS+RSV-1 RIS+RSV-2 RIS+RSV-3 *p*
---------------- -------------------- ------------------------- ---------------------- ------------------- --------------------- -----------
TOS (*μ*mol/L) 9.74 ± 0.71^b,e^ 16.04 ± 1.03^a,c,d,e^ 10.45 ± 1.06^b,e^ 8.19 ± 0.56^b^ 6.27 ± 1.09^a,b,c^ **0.000**
TAS (mmol/L) 1.60 ± 0.39^b,c,e^ 0.78 ± 0.14^a,c^ 2.28 ± 0.23^a,b,d,e^ 1.23 ± 0.10^c,e^ 0.38 ± 0.03^a,c,d^ **0.000**
OSI (AU) 791.59 ± 162.39^b^ 2805.78 ± 857.32^a,c,d^ 464.05 ± 25.47^b,e^ 692.31 ± 69.32^b^ 1797.28 ± 452.35^c^ **0.004**
Each group represents the mean ± SEM for seven rats. ^a^*p* \< 0.04 versus the control group; ^b^*p* \< 0.02 versus the RIS group; ^c^*p* \< 0.05 versus the RIS+RSV-1 group; ^d^*p* \< 0.03 versus the RIS+RSV-2 group; and ^e^*p* \< 0.05 versus the RIS+RSV-3 group. RIS: risperidone; RSV: resveratrol; TAS: total antioxidant status; TOS: total oxidant status; OSI: oxidative stress index; RIS+RSV-1: 2 mg/kg RIS+20 mg/kg RSV; RIS+RSV-2: 2 mg/kg RIS+40 mg/kg RSV; RIS+RSV-3: 2 mg/kg RIS+80 mg/kg RSV; AU: arbitrary units.
######
Effects of risperidone and resveratrol on apoptotic index (%) in rat kidneys.
Groups Apoptotic index (%) (AI; mean ± SD)
----------- -------------------------------------
Control 4.83 ± 1.17^b,c,d,e^
RIS 32.50 ± 3.56^a,c,d,e^
RIS+RSV-1 14.16 ± 2.40^a,b^
RIS+RSV-2 14.20 ± 2.86^a,b^
RIS+RSV-3 12.00 ± 2.60^a,b^
The apoptotic index of all the groups. Values are mean ± SD for seven rats in each group. ^a^Significant from control; ^b^Significant from RIS; ^c^Significant from RIS+RSV-1; ^d^Significant from RIS+RSV-2; and ^e^Significant from RIS+RSV-3 (*p* ≤ 0.05). RIS: risperidone; RSV: resveratrol; RIS+RSV-1: 2 mg/kg RIS+20 mg/kg RSV; RIS+RSV-2: 2 mg/kg RIS+40 mg/kg RSV; RIS+RSV-3: 2 mg/kg RIS+80 mg/kg RSV. The extent of TUNEL staining was scored semiquantitatively as 0 (none), 1 (light), 2 (medium), and 3 (intense).
[^1]: Academic Editor: Ersin Fadillioglu
| {
"pile_set_name": "PubMed Central"
} |
Foodborne illness resulting from consumption of food contaminated with pathogenic bacteria has been of vital concern to public health. Consumers today are increasingly concerned about chemical preservatives in food and tend to choose food products that are natural, safe and with multi-health benefits\[[@ref1][@ref2][@ref3]\]. Foodborne illness is a major problem associated with enormous costs. Foodborne pathogens occur widely in nature and it is difficult to prevent them from entering raw foods. *Salmonella* sp., *Listeria monocytogenes*, *Bacillus subtilis* and *Escherichia coli* account for the largest number of outbreaks, cases and deaths, and are capable of attaching to inert surfaces and subsequently forming bio films on food processing equipment and environment\[[@ref4][@ref5]\]. *Staphylococcus aureus* causes a range of illnesses and was found to be the most resistant organism\[[@ref6]\]. *Salmonella* mutants survive and are able to persist in the food chain\[[@ref7]\]. Many *Pseudomonas* spp. can cause food spoilage. Novel antipseudomonal activity is of particular interest as it is the leading cause of nosocomial infections and has developed mechanisms of resistance to common classes of antibiotics\[[@ref8][@ref9]\]. The resistance of bacteria and other microorganisms to antimicrobial agents has become a wide-spread medical problem especially as nosocomial pathogens. To reduce health hazards and economic losses due to foodborne microorganisms, the use of natural products as antibacterial compounds is gaining importance. However, it is necessary to establish the scientific basis for the therapeutic actions of traditional plant medicines. Several plants have been reported to be used in treating and managing the complicated diseases.
The food antimicrobials are classified into natural and synthetic substances depending on their origin. Although, many synthetic antimicrobials are found naturally (benzoic acid in cranberries, sorbic acid in rowanberries, citric acid in lemons, malic acid in apples and tartaric acid in grapes), the perception of natural has become important for many consumers\[[@ref10]\]. The problems mentioned introduced new research directions in the field of bioactive principles from natural sources and their application as food additives or dietary supplements.
*Momordica charantia* L. (Cucurbitaceae) commonly known as 'bitter gourd' and 'bitter melon', '*karela*' is a multipurpose herb widely cultivated in many tropical and subtropical regions of the world. The fruits are used as medicinal vegetable in different parts of the world. Apart from their role in food consumption, a wide array of pharmacological activities such as antidiabetic\[[@ref11]\], antioxidant\[[@ref12]\], anticancer activities\[[@ref13]\] and antiulcer\[[@ref14]\] are reported for this plant.
MATERIALS AND METHODS {#sec1-1}
=====================
Collection of the plant material: {#sec2-1}
---------------------------------
Different parts (aerial, peel, pulp and seed) of *Momordica charantia* L. were collected in September 2011 from Chotila, Surendranagar, Gujarat, India and identified by comparison with specimens (PSN333) available at the Herbarium of the Department of Biosciences, Saurashtra University, Rajkot, Gujarat, India. The parts were separated, washed thoroughly with tap water, shade dried, homogenised to fine powder and stored in airtight bottle.
Hydroalcoholic extraction method: {#sec2-2}
---------------------------------
The dried powders of all the four parts were extracted individually by cold percolation method\[[@ref15][@ref16][@ref17]\]. The hydroalchoholic extraction was done using methanol and water\[[@ref18]\]. The dried powder was first defatted by hexane and then extracted in 100% methanol (MeOH), 75% MeOH, 50% MeOH, 25% MeOH and 100% water (aqueous). Ten grams of dried powder was taken in 100 ml of hexane in a conical flask, plugged with cotton wool and then kept on a rotary shaker at 120 rpm for 24 h. After 24 h, the extract was filtered with eight layers of muslin cloth; centrifuged at 5000 rpm for 10 min. Supernatant was collected and the solvent was evaporated. The residue was then added to 100 ml of each solvent, that is 100% MeOH, 75% MeOH, 50% MeOH, 25% MeOH and water in a conical flask, plugged with cotton wool and then kept on a rotary shaker at 120 rpm for 24 h. After 24 h, the extract was filtered with eight layers of muslin cloth; centrifuged at 5000 rpm for 10 min, the supernatant was collected and the solvents were partially evaporated using rotary vacuum evaporator (Equitron, India) then kept in petri plates to dry. The extract was stored at 4° in air tight bottles. The residues were weighed to obtain the extractive yield.
Antimicrobial activity: {#sec2-3}
-----------------------
The microorganisms used in this investigation were obtained from National Chemical Laboratory, Pune, India. The microorganisms were maintained at 4°. The Gram-positive bacteria studied were *Staphylococcus aureus* ATCC29737 (*SA*), *Staphylococcus albus* NCIM 2178 (*SAL*), *Corynebacterium rubrum* ATCC14898 (*CR*), *Listeria monocytogenes* ATCC19112 (*LM*), *Micrococcus flavus* ATCC10240 (*MF*); Gram-negative bacteria used were *Pseudomonas aeruginosa* ATCC27853, (*PA*) *Pseudomonas stutzeri* NCIM5136 (*PSt*), *Pseudomonas pictorum* NCIB9152 (*PPi*), *Pseudomonas putida* NCIM2872 (*PP*), *Pseudomonas testosteroni* NCIM5098 (*PT*), *Pseudomonas syrigae* NCIM5102 (*PS*); and fungi were *Candida albicans* ATCC2091 (*CA*), *Candida neoformans* NCIM3542 (*CN*), *Candida glabrata* NCIM3448 (*CG*), *Candida epicola* NCIM3367 (*CE*). The organisms were maintained on nutrient agar and MGYP medium (Hi-Media, India) for bacteria and fungi respectively, at 4° and subcultured before use. The microorganisms studied are clinically important ones causing several infections and food spoilage. Ampicillin (AMP 10 μg/disc), chloramphenicol (CH 30 μg/disc), tetracycline (T 30 μg/disc), amphotericin B (AP 100 units/disc) and nystatin (NS 100 units/disc) were used as standard to determine antimicrobial susceptibility. Chloramphenicol and ceftazidime (CF) were used during minimum inhibitory concentration (MIC) and minimum bactericidal concentration (MBC) determination. All antibiotics were purchased from Hi-Media Laboratory Pvt. Ltd., (Mumbai, India).
Agar well diffusion method: {#sec2-4}
---------------------------
*In vitro* antimicrobial activity of the different solvent extracts was studied against pathogenic microbial strains by the agar well diffusion method\[[@ref19][@ref20][@ref21][@ref22]\]. Mueller-Hinton No. 2/Sabouraud dextrose agar (Hi-Media) was used for the antibacterial and antifungal susceptibility test, respectively. The different solvent extracts were diluted in 100% dimethyl sulfoxide (DMSO) to give a concentration of 20 mg/ml. The Mueller-Hinton agar/Sabouraud dextrose agar was melted and cooled to 48--50° and a standardised inoculum (1.5×10^8^ CFU/ml, 0.5 McFarland) was then added aseptically to the molten agar and poured into sterile Petri dishes; wells (8.5 mm) were prepared in the seeded agar plates. The test compound (100 μl) was introduced into the well. The plates were incubated overnight at 37° and 28° for 24 and 48 h, respectively, for bacteria and fungi. DMSO was used as negative control. The microbial growth was determined by measuring the diameter of the zone of inhibition and the mean values are presented with ±SEM (standard error of mean).
Preparation of bacterial inocula and extracts or antibiotics for MIC and MBC study: {#sec2-5}
-----------------------------------------------------------------------------------
The inoculum of the test organisms were prepared using the colony suspension method\[[@ref23]\]. Colonies picked from 24 h old cultures, grown on nutrient agar, were used to make suspension of the test organisms in saline solution to give an optical density of approximately 0.1 at 600 nm. The suspension was then diluted 1:100 by transfer of 0.1 ml of the bacterial suspension to 9.9 ml of sterile nutrient broth before use to yield 6×10^5^ CFU/ml. Twofold serial dilutions using 100% DMSO were carried out from the 1250 μg/ml stock plant extract to make six test concentrations ranging from 39 to 1250 μg/ml for each solvent extracts. Twofold dilutions of chloramphenicol and ceftazidime (1--32 μg/ml) were used as a positive control.
Determination of minimum inhibitory concentration: {#sec2-6}
--------------------------------------------------
The MICs were determined only for the test organisms that had shown \>15 mm zone of inhibition of the crude extracts. Micro broth dilution method performed in sterile flat bottom 96 well micro test plates (Tarsons Products Pvt. Ltd.) was performed to evaluate MIC of the plant extracts\[[@ref24]\]. One hundred and fifty microlitres of Mueller-Hinton broth was introduced into all the 96 wells and 20 μl of varying concentrations of the extract was added in decreasing order along with 30 μl of the test organism suspension. A final volume of 200 μl was achieved in each well (150 μl Mueller-Hinton broth, 30 μl of the test organism suspension and 20 μl plant extract/antibiotic). Three control wells were maintained for each test batch. The positive control (antibiotic, Mueller-Hinton broth and test organism) and sterility control (Mueller-Hinton broth and DMSO) and organism control (Mueller-Hinton broth, test organism and DMSO). Plates were then incubated at 37° for 24 h overnight. Experiments were performed in triplicate. After incubation, 40 μl of 2-(4-iodophenyl)-3-(4-nitrophenyl) 5-phenyltetrazolium chloride (INT, Himedia, India) solution (0.2 mg/ml) dissolved in sterile distilled water was added to each well\[[@ref25]\]. The plates were incubated for further 30 min, and estimated visually for any change in colour to pink indicating reduction of the dye due to bacterial growth. The highest dilution (lowest concentration) that remained clear corresponded to the MIC.
Determination of minimum bactericidal concentration: {#sec2-7}
----------------------------------------------------
MBC was determined from all wells showing no growth as well as from the lowest concentration showing growth in the MIC assay for all the samples. Bacterial cells from the MIC test plate were sub-cultured on freshly prepared solid nutrient agar by making streaks on the surface of the agar. The plates were incubated at 37° for 24 h overnight. Plates that did not show growth were considered to be the MBC for the extract or drug used\[[@ref26]\]. The experiment was carried out in triplicate.
Determination of MIC index and statistical analyses: {#sec2-8}
----------------------------------------------------
The MIC index (MBC/MIC) was calculated for each extract and positive control drug to determine whether an extract had bactericidal (MBC/MIC ≤4) or bacteriostatic (\>4 MBC/MIC \<32) effect on growth of bacteria\[[@ref27]\]. All experiments were repeated at least three times. Results are reported as mean±SEM.
RESULTS {#sec1-2}
=======
The extractive yield varied among different parts of *M. charantia* and also among different hydroalcoholic extracts (hexane, 100, 75 and 25% methanol and water) as shown in ([fig. 1](#F1){ref-type="fig"}). The hexane extract had very negligible yield in all the four parts of *M. charantia*. The areal part aqueous extract had slightly more extractive yield that 100% MeOH. As the concentration of methanol decreased, there was a slight increase in extractive yield ([fig. 1a](#F1){ref-type="fig"}). The peel aqueous extract had considerably more extractive yield than 100% MeOH. As the concentration of methanol decreased, the extractive yield increased almost reaching to that of pure aqueous extract ([fig. 1b](#F1){ref-type="fig"}). The extractive yield of hydroalcoholic extracts of pulp showed a trend similar to that of peel ([fig. 1c](#F1){ref-type="fig"}). Both these parts, that is peel and pulp had maximum extractive yield. In seed also, pure methanol had considerable less extractive yield than aqueous extract; the extractive yield of other hydroalcoholic extracts was similar to that of aerial parts ([fig. 1d](#F1){ref-type="fig"}).
{#F1}
Antimicrobial activity of Momordica charantia aerial part: {#sec2-9}
----------------------------------------------------------
The antimicrobial activity of different hydroalcoholic extracts of aerial part of *M. charantia* is shown in Table [1](#T1){ref-type="table"} and [2](#T2){ref-type="table"}. All the extracts showed activity against *M. flavus* and *S. aureus*. The highest activity was in 50% MeOH followed by hexane extract against *M. flavus*. In Gram-negative bacteria, all the extracts showed activity against *P. syrigae* and *P. testosterone*, except *P. stutzeri*. The hexane extract showed maximum activity against *Pseudomonas* spp. All extracts showed moderate activity against fungi.
######
ANTIBACTERIAL ACTIVITY OF DIFFERENT PARTS OF *M. CHARANTIA*
######
ANTIFUNGAL ACTIVITY OF DIFFERENT PARTS OF *M. CHARANTIA*
Antimicrobial activity of *Momordica charantia* peel: {#sec2-10}
-----------------------------------------------------
The antimicrobial activity of different hydroalcoholic extracts of peel part of *M. charantia* is shown in Table [1](#T1){ref-type="table"} and [2](#T2){ref-type="table"}. In Gram-positive bacteria, all extracts showed activity against *M. flavus* and the highest activity was in 50% MeOH extract. *S. albus* was resistant to all the extracts. In Gram-negative bacteria, all extracts showed activity against *P. syrigae*, *P. testosteroni* and *P. putida*; while remaining extracts showed different levels of activity against *P. aeruginosa*, *P. stutzeri and P. pictorum*. The 100 and 50% MeOH extracts showed activity against all *Pseudomonas* spp. screened. All extracts showed moderate activity against the fungi.
Antimicrobial activity of *Momordica charantia* pulp: {#sec2-11}
-----------------------------------------------------
Antimicrobial activity of different hydroalcoholic extracts of the pulp of *Momordica charantia* is shown in Table [1](#T1){ref-type="table"} and [2](#T2){ref-type="table"}. In Gram-positive bacteria, all extracts showed activity against *M. flavus* and the highest activity was in 75% MeOH extract followed by the hexane extract. *C. rubrum* and *S. albus* were resistant to all extracts. *L. monocytogenes* was slightly susceptible to only 50% MeOH extract. In Gram-negative bacteria, all extracts showed activity against *P. aeruginosa*, *P. testosteroni* and *P. putida*; while remaining extracts showed different levels of activity against *P. syrigae* and *P. pictorum*. The hexane extract showed maximum activity against *P. stutzeri*. All extracts showed moderate activity against the fungi.
Antimicrobial activity of *Momordica charantia* seed: {#sec2-12}
-----------------------------------------------------
Antimicrobial activity of different hydroalcoholic extracts of seed part of *M. charantia* is shown in Table [1](#T1){ref-type="table"} and [2](#T2){ref-type="table"}. In Gram-positive bacteria, all the extracts showed activity against *M. flavus* and *S. aureus* but susceptibility of *M. flavus* was considerably more than that of *S. aureus*. The highest activity was shown by the hexane extract against *M. flavus*. In Gram-negative bacteria, all the extracts showed activity against *P. testosteroni*, *P. putida*, *P. stutzeri* and *P. syrigae*; while *P. aeruginosa* and *P. pictorum* were not susceptible to any of the hydroalcoholic extracts. The highest activity was shown by 100% MeOH extract against *P. aeruginosa*. All extracts showed moderate activity against fungi.
All 24 extracts were compared with 5 standard antibiotics. These antibiotics were tested against 15 medically important microbial strains, the results of which were presented in [Table 3](#T3){ref-type="table"}. The antimicrobial activity of some of the hydroalcoholic extracts was comparable with that of standard antibiotics.
######
ANTIMICROBIAL ACTIVITY USING STANDARD ANTIBIOTICS
Determination of MIC, MBC and MIC index: {#sec2-13}
----------------------------------------
The MIC and MBC values of different hydroalcoholic extracts of different parts of *M. charantia* and standard antibiotics are shown in Tables [4](#T4){ref-type="table"}-[7](#T7){ref-type="table"}. Inhibitory effects of bacterial growth by the extracts from different parts were in the range from \<39 to \>1250 μg/ml expressed as MIC values and in the range from 625 to \>1250 μg/ml expressed as MBC values. Inhibitory effects of bacterial growth by the standard antibiotics were in the range from 4 to \>32 μg/ml expressed as MIC values and in the range from 16 to \>32μg/ml expressed as MBC values.
######
MIC AND MBC OF DIFFERENT SOLVENT EXTRACTS OF *M. CHARANTIA* AERIAL PARTS
######
MIC AND MBC OF DIFFERENT SOLVENT EXTRACTS OF *M. CHARANTIA* PULP
######
MIC AND MBC OF DIFFERENT SOLVENT EXTRACTS OF *M. CHARANTIA* PEEL
######
MIC AND MBC OF DIFFERENT SOLVENT EXTRACTS OF SEEDS OF *M. CHARANTIA*
MIC and MBC values of different hydroalcoholic extracts of aerial part of *M. charantia* and standard antibiotics are shown in [Table 4](#T4){ref-type="table"}. Extract of 75% MeOH showed least MIC value and MBC values, that is 156 and 1250 μg/ml, respectively against *M. flavus*. However 100 and 75% MeOH extracts showed MIC index of 4; therefore both extracts showed bactericidal effect; it is comparable to standard chloramphenicol. Extract of 75% MeOH showed bactericidal effect against *P. pictorum*.
MIC and MBC values of different hydroalcoholic extracts of *M. charantia* pulp and standard antibiotics are shown in [Table 5](#T5){ref-type="table"}. Extracts of 75 and 100% MeOH showed least MIC values 312 and 156 μg/ml, respectively, and MBC was 1250 μg/ml against *M. flavus*. Extracts of 100% and 75% MeOH showed MIC index of 8 and 4, respectively; therefore, 100 and 75% MeOH showed bactericidal and bacteriostatic effect against *M. flavus*. Extracts of 100% MeOH showed MIC value of 625 μg/ml while MBC was \>1250 μg/ml against *P. putida*.
MIC and MBC values of different hydroalcoholic extracts of *M. charantia* peel and standard antibiotics are shown in [Table 6](#T6){ref-type="table"}. Extracts with 50% and 100% MeOH showed least MIC values, 312 and 156 μg/ml, respectively, and MBC was 1250 and 625 μg/ml, respectively, against *M. flavus*. Both extracts showed bactericidal effect.
MIC and MBC values of different hydroalcoholic extracts of *M. charantia* seed and standard antibiotics are shown in [Table 7](#T7){ref-type="table"}. MeOH extracts (50 and 100%) showed least MIC values \<39 μg/ml and MBC was 625 μg/ml against *P. putida*. Remaining extracts showed \>1250 μg/ml MIC and MBC values. In this study, bactericidal effect was shown by 100, 75 and 50% MeOH extracts against *M. flavus* and *P. pictorum* while remaining extracts showed bacteriostatic effects.
DISCUSSION {#sec1-3}
==========
Normally, a high extraction yield is required for an efficient process; although it is not necessary that high concentration of bioactive components are present in them. Since some bioactive components are very sensitive to oxygen and heat\[[@ref28]\], care should be taken to prevent their oxidation and thermal degradation. Therefore, the extraction yield and the bioactive component characteristics should also be considered when an extraction method is selected. The traditional healers or practitioners make use of water primarily as a solvent but there are many reports where organic solvents showed better activity as compared with aqueous extracts\[[@ref29][@ref30][@ref31]\]. In the present study, extractive yield was considerably more in water than in organic solvent methanol and as the concentration of methanol decreased, extractive yield increased clearly indicating that in these plant parts, water soluble phytoconstituents were more.
The results of antimicrobial activity clearly indicated that *M. flavus* was the most susceptible Gram-positive bacteria and *P. testosterone* was the most susceptible Gram-negative bacteria. Aqueous extract showed poor activity as compared with pure methanol or hydroalchoholic extracts; best antibacterial activity was shown by 100% MeOH extract, which is an extract with a pure organic solvent. Almost all the extracts showed antifungal activity against all the four fungi studied, though the activity was moderate.
Gram-negative bacteria were more susceptible towards all the extracts than Gram-positive bacteria. This is very good report since there is a general consensus that plant extracts are more active against Gram-positive bacteria than Gram-negative bacteria\[[@ref32][@ref33][@ref34][@ref35][@ref36][@ref37][@ref38][@ref39][@ref40]\]. Therefore, the search is always to find plant extracts that are capable of inhibiting Gram-negative bacteria, which are equally dangerous in causing infectious diseases like Gram-positive bacteria. The Gram-negative cell wall (made up of lipopolysaccharide) is complex and multilayered structure, which makes access to membrane more restricted and barrier to many environmental substances including synthetic and natural antibiotics. The results of the present study indicate that extracts of *M. charantia* contain some secondary metabolites, which are able to cross this tough barrier.
The MIC is defined as the lowest concentration of the antimicrobial agent that will inhibit the visible growth of a microorganism after overnight incubation\[[@ref41][@ref42]\], whereas the MBC is interpreted as the lowest concentration that can completely remove the microorganisms. A pinkish coloration is indicative of microbial growth because of their ability to convert INT to red formazan\[[@ref43]\]. The concentrations of MIC and MBC for plant extracts and standard antibiotics were 1250--39 μg/ml and 32--1 μg/ml, respectively. MIC and MBC were expressed in terms of μg/ml. Braca *et al*.\[[@ref44]\], Coutinho *et al*.\[[@ref45]\] and Roopashree *et al*.\[[@ref46]\] used in their studies *M. charantia* extracts/essential oils, which possessed potential activity against *Staphylococcus aureus*. Castilho *et al*.\[[@ref47]\] reported that antimicrobial activity of *Origanum* essential oils and all the studied extracts showed MIC values \>200 μg/ml against *P. aerugenosa*. In the present study, better results were found in seed (100% methanol and 50% methanol) and MIC value was \<39 μg/ml, which was near to ceftazidime against *P. pictorum*.
The spread of multidrug-resistant strains of microorganisms and the reduced number of drugs available makes it necessary to discover new classes of antibacterial and antifungal agents that overcome these resistant mechanisms. This led to search for therapeutic alternatives, particularly among medicinal plants and compounds isolated from them used empirically for their antibacterial and antifungal properties.
Foodborne disease is one of the major concerns to food producers and consumers and spoilage of foods is still a major problem in different parts of the world. In an effort to meet this demand, the food industry has a great interest in using natural antimicrobial compounds. The hydroalcoholic extracts of *M. charantia* possessed significant antibacterial activity (MIC ≤39 μg/ml) against *Pseudomonas* spp. Therefore, the use of this plant as antimicrobial agent is validated by the results obtained in this work. Further studies are in progress to identify the chemical compounds present in these extracts with antimicrobial activity as well as to identify synergism between plant extracts and standard antibiotics. The results of the present investigation also provide an approach to develop promising natural antimicrobial agents with potential applications in the food and pharmaceutical industries. This fact is of paramount importance from the point of view of food safety.
Rakholiya, *et al*.: *Momordica charantia* L. Extracts against Foodborne Pathogens
| {
"pile_set_name": "PubMed Central"
} |