Datasets:

haritzpuerto

/

the_pile_00_PubMed_Central

like 1

Background ========== Osteogenesis imperfecta (OI) is a term describing a group of connective tissue disorders, that are characterized by low bone mass and increased bone fragility (brittle bone disease) \[[@b1-amjcaserep-20-1540],[@b2-amjcaserep-20-1540]\]. Orthopedic manifestations include frequent fractures, scoliosis, and long bone deformities that can severely influence the walking ability \[[@b3-amjcaserep-20-1540],[@b4-amjcaserep-20-1540]\]. Extraskeletal manifestations of osteogenesis imperfecta can include blue/gray sclera, hearing loss, aortic root dilatation, dentinogenesis imperfecta, macrocephaly, and basilar invagination \[[@b5-amjcaserep-20-1540],[@b6-amjcaserep-20-1540]\]. The mainstream of treatment in children who present with extreme deformities of their long bones consists of multiple osteotomies and load sharing intramedullary devices in order to re-align the bone and provide adequate stabilization \[[@b3-amjcaserep-20-1540]--[@b6-amjcaserep-20-1540]\]. The aim of our study was to present our surgical strategy, in a 9-year-old male patient with severe, progressive, bilateral tibia deformity suffering from OI, with osteotomies and intramedullary, flexible Titanium Elastic Nail System (TENS) nails, that can be an excellent alternative when doctors or hospitals do not have access to advanced surgical systems, such as telescoping rods. Case Report =========== A 9-year-old refugee male patient presented by an International Charity Organization to our department with progressive bilateral deformity of femur and tibia. The bowing of both tibias was so severe that walking was impossible ([Figure 1A--1D](#f1-amjcaserep-20-1540){ref-type="fig"}). Regarding our patient's medical history, he had not received any appropriate treatment at all (no bisphosphonates treatment, no previous procedures). Over the years, he experienced extreme deformity of both tibias, leading to anteromedial deformity. Clinical examination revealed bilateral genu varum. Range of motion of the ankle, knee, and hip joints was not limited. No tenderness was found in the knee and ankle joints and no limb length discrepancy was present. Pain was localized exactly over the apex of tibia deformity in both legs. Neurovascular examination was normal. The main problem of our patient was the loss of capacity for independent walking and upright standing. There was an absolute need for crutches. Standing anteroposterior (AP) full limb radiographs of both lower extremities were taken. Radiographs of both tibias were also taken in the AP and lateral view. A significant bowing of both tibias and fibulas was revealed with the apex of curve directed anteromedially ([Figure 2A--2D](#f2-amjcaserep-20-1540){ref-type="fig"}) as well as bowing and varus deformity of both femurs ([Figure 2E, 2F](#f2-amjcaserep-20-1540){ref-type="fig"}). Low bone density, thickening of the cortical bone, medullary canal narrowing and synostosis between tibia and fibula were noticed ([Figure 2A--2D](#f2-amjcaserep-20-1540){ref-type="fig"}). A complete blood test and laboratory analysis were performed. Except for a mild elevation of ALP (alkaline phosphatase), the laboratory evaluation was normal. Pre-operative planning ---------------------- After discussion with the parents, operative treatment with simultaneous bilateral correction of both tibias was decided. The pre-operative plan included the diameter of the tibia bone, the narrowing of the medullary canal, the type of bowing-deformity, and the levels of the possible osteotomies. Radiographic evaluation of mechanical axis of the tibia, anatomical axis of the tibia, medial proximal tibial angle (MPTA), lateral distal tibial angle (LDTA), and joint line congruency angle (JLCA) in AP views were also measured. All these angles were found to be within the normal range. The pre-operative plan was based on the principles of deformity correction that have been well described in the literature \[[@b7-amjcaserep-20-1540]--[@b9-amjcaserep-20-1540]\]. The point of intersection of the proximal and distal axis lines of the deformed tibias was evaluated in order to find the center of rotation of angulation (CORA). Because in our patient the CORA was outside the boundaries of the involved bones, it was clear that a multi-apical deformity was present, and more than one osteotomy would be required to achieve an acceptable alignment of the tibias ([Figure 3](#f3-amjcaserep-20-1540){ref-type="fig"}). The final goal was to realign the axis of the legs, provide adequate stability in order to avoid an impending fracture, and to provide the patient with the ability to walk. Taking into account the age of our patient (open physes) and the fact that we had to choose an alternative to advanced surgical systems (because our patient had no health insurance as a refugee child presented to our clinic from a charity organization) we decided to proceed with multiple osteotomies and intramedullary devices: TENS nails, Synthes 3.5 mm diameter. Surgical technique ------------------ General anesthesia was administrated. The patient was placed in the supine position and under fluoroscopy, 2 TENS nails were inserted antegrade with the entry point being just distal to the proximal epiphyseal plate of the tibia, at the medial and the lateral side of the tibia tuberosity, respectively ([Figure 3A, 3B](#f3-amjcaserep-20-1540){ref-type="fig"}). Insertion of these nails was performed until the level of the pre-planned (more proximal) osteotomy. An open approach directly over the curve of the deformity followed the insertion of the nails. Kirschner wires (K-wires) were used to mark the 2 levels of the osteotomies and to use them as joysticks to help reduction ([Figure 4A, 4B](#f4-amjcaserep-20-1540){ref-type="fig"}). The osteotomies were prepared by performing multiple holes with the drill and finished using an osteotome. An osteotome (and not a saw) was used to gently break the cortical bone at the correct level with the aid of the pre-drilled holes. After performing the osteotomy, the bone was realigned ([Figure 5A, 5B](#f5-amjcaserep-20-1540){ref-type="fig"}), however, more multiple small-sized additional osteotomies were needed ([Figure 5C](#f5-amjcaserep-20-1540){ref-type="fig"}). The synostosis between the tibia and fibula was also removed ([Figure 5C](#f5-amjcaserep-20-1540){ref-type="fig"}). Intra-operatively, it was crystal clear that the medullary canal was extremely narrowed at the level of the deformity. We proceeded with multiple drilling of the canal to increase the diameter ([Figure 6A](#f6-amjcaserep-20-1540){ref-type="fig"}). Then, the 2 nails that were already inserted till the level of the proximal osteotomy, were passed through the 2 fragment ends, which were also stabilized with an additional K-wire ([Figure 6B--6D](#f6-amjcaserep-20-1540){ref-type="fig"}). A significant step was the final adaptations and adjustments of the periosteum ([Figure 7A, 7B](#f7-amjcaserep-20-1540){ref-type="fig"}). [Figure 8](#f8-amjcaserep-20-1540){ref-type="fig"} shows the postoperative x-rays of our patient. Follow-up --------- Both legs were immobilized in a below-knee cast for 6 weeks. [Figure 9](#f9-amjcaserep-20-1540){ref-type="fig"} shows our patient's tibia alignment at 15 days of follow-up. Weight-bearing was not allowed for 3 months. The patient used crutches for mobilization. Touch weight bearing was allowed at 3 months and full-weight bearing was not allowed until signs of callus formation were seen on x-ray, 6 months post-operatively ([Figure 10](#f10-amjcaserep-20-1540){ref-type="fig"}). Complete radiographic union was established 9 months post-operatively. At 1-year follow-up, a small degree (10° to 15°) of tibia valgus was present but the patient was walking without pain, and range of motion of both knee and ankle joints was normal. No evidence of radiographic nonunion was observed. Regarding the femur deformity (bowing and genu varum) no surgical intervention was decided at the time, but close follow-up every 6 months was scheduled in order to proceed with surgical intervention at an appropriate time. Discussion ========== Osteogenesis imperfecta (OI) is a connective tissue disease characterized by a wide variety of phenotypic and molecular heterogeneity \[[@b10-amjcaserep-20-1540]\]. It is an unusual heritable disease (1 in 10 000 to 20 000) with 90% of patients having mutations in type I collagen genes (COL1A1 and COL1A2) \[[@b4-amjcaserep-20-1540],[@b10-amjcaserep-20-1540]\]. The genetic defect is inherited either with autosomal dominant transmission or with autosomal recessive transmission \[[@b11-amjcaserep-20-1540]\]. The first classification system of OI into 4 types (I--IV) was made in 1979 by Silence et al. and it was mainly used for the clinical and radiological classification of OI: type I mild nondeforming, type II perinatal lethality, type III severely deforming, and type IV moderate deforming \[[@b10-amjcaserep-20-1540],[@b12-amjcaserep-20-1540]\]. Since then, new genes have been discovered and the classification has been expanded with OI types V--VII mainly based on cases with unknown genetic etiology \[[@b5-amjcaserep-20-1540],[@b13-amjcaserep-20-1540]\]. OI type I is mostly characterized by a 50% reduction of the amount of collagen type I (quantitative disorder in collagen) while OI types II--IV by sufficient but abnormal collagen I production (qualitative disorder in collagen) \[[@b5-amjcaserep-20-1540],[@b10-amjcaserep-20-1540]\]. Low bone mass is one of the main characteristics of OI that leads to structural deficiency, but the mechanical quality of the bone material is also reduced \[[@b14-amjcaserep-20-1540]\]. This is no surprise as one of the main organic components of bone is collagen type I that is affected by the genetic defects. In addition, deformities of long bones (tibia, femur) and progressive scoliosis are common manifestations of OI \[[@b15-amjcaserep-20-1540]\]. Bowing of large bones result in extreme mechanical stresses in the apex of curvature and thus, they are a significant risk factor of bone fracture \[[@b6-amjcaserep-20-1540]\]. Our patient not only was "one step" before sustaining a fracture but he has already lost the walking ability due to progressive tibia deformity. It can be easily understood, the importance of surgical realignment and stabilization of long bone deformities. In the literature, it seems that the use of load-sharing devices such as intramedullary Rush rods. Kuntscher rods, K-wires, Ender nails, elastic nailing, or telescoping rods are preferred over plating \[[@b3-amjcaserep-20-1540],[@b5-amjcaserep-20-1540],[@b10-amjcaserep-20-1540],[@b16-amjcaserep-20-1540]\]. Enright et al. report a high complication rate in children with OI treated with plating (69.2% complication rate) \[[@b16-amjcaserep-20-1540]\]. Regarding the intramedullary rods, these can be fixed or elongated rods with the advantage of elongated rods being that allows the longitudinal bone growth, but the diameter has to be small enough to not affect the physis \[[@b17-amjcaserep-20-1540]\]. Sterian et al. report in their publication that although with telescoping rods a long-lasting osteosynthesis can be obtained, arthrotomies and nail insertion through the joint cartilage is needed, leading in potential joint stiffness \[[@b18-amjcaserep-20-1540]\]. They concluded that Fassier Duval telescoping nailing is a good alternative that avoids these problems \[[@b18-amjcaserep-20-1540]\]. Sangasoongsong et al. support the view of using humeral nails in femoral fixation in patients with OI over Rush nails as they have a smaller diameter and provide the interlocking property which is better for rotational stability \[[@b19-amjcaserep-20-1540]\]. Mulpuri and Joseph report the results of a 10-year period of intramedullary rodding in 16 patients. They found that the post-operative fracture rate in the elongating rod group was 0.04 per person while in the non-elongating implant group the post-operative fracture rate was 0.21 per person \[[@b20-amjcaserep-20-1540]\]. Possible complications with intramedullary nailing can be bending of the rod, migration, disengagement, fracture of the rod or fracture of the bone, nonunion or delayed union, and hardware loosening \[[@b3-amjcaserep-20-1540],[@b16-amjcaserep-20-1540],[@b21-amjcaserep-20-1540]\]. Chiarello et al. analyzed 29 patients (245 procedures) and compared conservative treatment (e.g. casts) and surgical treatment (e.g., pinning, intramedullary nailing, plating) and although they found no significant difference regarding the complications between the 2 groups, they concluded that in type III OI the use of intramedullary devices in association with bisphosphonates appeared to be better \[[@b21-amjcaserep-20-1540]\]. Bisphosphonates have been extensively used for patients, with OI especially in children aged 3 years-old and they can be administrated for up to 2 years, but orthopedic surgery has the primary role in severe cases \[[@b14-amjcaserep-20-1540]\]. Georgescu et al. published the evidence from 32 operated patients with OI (81 surgeries) either with Sheffield telescoping rod, circular external fixator, or bone transplantation and they conclude that surgical treatment of moderate to severe long bone deformities was the only chance for these patients to walk again \[[@b22-amjcaserep-20-1540]\]. Pre-operative planning and selection of the best implant for patients with OI is very important but problematic as well. The age of the patient, the distortion of the anatomy, the advantages and disadvantages of the various surgical instruments, the availability of the different surgical instruments, the surgical experience, and the post-operative complications are all factors that should be carefully examined in order to select the best implant for each case \[[@b18-amjcaserep-20-1540],[@b19-amjcaserep-20-1540]\]. With our case report, we present a relatively simple surgical technique for correction of long bone deformities that can be a good alternative when there is no access to advanced, innovative surgical systems, such as telescoping rods. Conclusions =========== Our opinion is that each patient with OI is a unique case and the corrective surgical treatment should be adapted to each case. In patients with OI diagnosis, a multidisciplinary team approach by orthopedic surgeons, physicians, pediatricians, and endocrinologists should be performed. From the orthopedic point of view, the final goal should be the ability to walk independently with as minimal complications as possible. This goal can be achieved, regardless of which surgical technique is performed. **Conflicts of interest** None. {#f1-amjcaserep-20-1540} {#f2-amjcaserep-20-1540} {#f3-amjcaserep-20-1540} {#f4-amjcaserep-20-1540} {#f5-amjcaserep-20-1540} {#f6-amjcaserep-20-1540} {#f7-amjcaserep-20-1540} {#f8-amjcaserep-20-1540} {#f9-amjcaserep-20-1540} {#f10-amjcaserep-20-1540} {#f11-amjcaserep-20-1540} [^1]: Authors' Contribution: [^2]: Study Design [^3]: Data Collection [^4]: Statistical Analysis [^5]: Data Interpretation [^6]: Manuscript Preparation [^7]: Literature Search [^8]: Funds Collection [^9]: **Conflict of interest:** None declared

Introduction ============ The need for active postoperative movement of the flexor tendon repairs of the fingers in zones II, III, IV and V, to prevent adhesions and obtain proper range of motion, requires suture stitches with high mechanical resistance. [@BR1900057-1] [@JR1900057-2] Among the various qualifications for optimal repair, such as number of passages, thread qualities, suture volume, among others, ease of performance with minimal surgical trauma is fundamental. [@JR1900057-2] The six-passage "figure-of-eight" suture is easy to perform, it can be made with various types of surgical threads, has great mechanical resistance for active postoperative movement, and its efficiency has been proven in clinical and biomechanical studies. [@JR1900057-3] [@JR1900057-4] [@JR1900057-5] [@JR1900057-6] [@JR1900057-7] Although there are no studies on the preference of Brazilian surgeons for the suture technique used in the flexor tendons of the fingers, it is believed, by empirical observation, that the Kessler suture is one of the most widely used. The classic method to study the mechanical properties of intact or sutured tendons is to subject the specimen to strain deformation at constant speed. [@BR1900057-8] The experimental model to biomechanically test the immediate suture of flexor tendons using swine specimens, by mechanical test of longitudinal traction under constant traction speed, finds reference in the literature. [@JR1900057-9] [@JR1900057-10] [@JR1900057-11] The objective of the present study was to biomechanically evaluate, through longitudinal tensile tests at constant speed, the deformation by tension of the "figure-of-eight" and Kessler sutures in swine flexor tendons. Materials and Methods ===================== The upper limbs of 18 pigs were disarticulated at the elbow, packed in plastic bags and kept in a freezer at -20 degrees Celsius. On the day of the experiments, the anatomical parts were thawed at room temperature, and the deep flexor tendons of the fingers were dissected and isolated. The tendons of the right upper limbs were divided into two groups: group F8 (3 "figure-of-8" stitches) and group K (Kessler suture). The tendons of both groups were sectioned in the central region with a scalpel blade number 15 and submitted to sutures: the F8 group with 3 "figure-of-8" stitches with polypropylene monofilament yarn 3--0 (Prolene, Ethicon, São José dos Campos, SP, Brazil), and, in group K, Kessler suture with the same surgical thread; in both groups there were continuous peripheral sutures with polypropylene monofilament thread 4--0 (Prolene) ( [Figure 1](#FI1900057en-1){ref-type="fig"} ). After suturing, the tendons were fixed in aluminum sinusoidal metal claws, compressed by screws with a distance of 20 mm from the suture region in the central part. The claws were mounted axially in a universal mechanical testing machine with a 1,000-N load cell and application speed of 30 mm/min (EMIC DL 10000, Instron, São José dos Pinhais, PR, Brasil). After the test, the computer coupled to the equipment provided the mechanical properties of maximum load (N) and energy at maximum load (N.mm). {#FI1900057en-1} The statistical analyses of the results were performed using the Student *t* test, with values of *p*  \< 0,5 considered significant. Results ======= In groups F8 and K, ruptures always occurred in the suture area, and it was not possible to determine the sequence of the ruptured stitches, since there was no filming of the mechanical tests. [Table 1](#TB1900057en-1){ref-type="table"} presents the results of the mechanical properties in both groups, which indicate higher values in the F8 group ( *p*  \< 0,5). ###### Means, standard deviations, maximum and minimum values and statistical analyses of the mechanical properties of maximum load (N) and energy at maximum load (N.mm) in the experimental groups ---------------------------------------------------------------------------------------------------------------- Groups Maximum load Energy at maximum load ----------------- ----------------------------------------------- ---------------------------------------------- K ( *n*  = 8) 34.19 ± 11.4; maximum: 58.55; minimum: 18.29 100.9 ± 52.48; maximum: 206.5;\ minimum 34.61 F8 ( *n*  = 10) 63.40 ± 20.40; maximum: 86.04; minimum: 23.17 217.3 ± 93.67; maximum: 365.7; minimum 33.39 *p* -value 0.0024\* 0.0064\* ---------------------------------------------------------------------------------------------------------------- Note: \* *p*  \< 0.5. Discussion ========== The present study demonstrated that the triple "figure-of-eight" suture (six passages) presents values for the mechanical properties of maximum load and energy at maximum load that are statistically higher than those of the Kessler suture, which is in line with the results of the study by Al-Qattan and Al-Turaiki. [@JR1900057-3] The maximum load value of a flexor tendon suture of a finger to enable active movement without risk of rupture or suture spacing is at least 40 N, a value higher than that observed in the K group (34.19 N), and lower than that of the F8 group (63.4 N), indicating the safety of the triple "figure-of-eight" suture. [@BR1900057-12] Knowledge of the maximum load mechanical property is fundamental in assessing the strength of a given tendon suture, and is one of the most used parameters in biomechanical studies. [@JR1900057-2] [@BR1900057-8] [@BR1900057-12] On the other hand, the clinical importance of the energy property at full load is not fully understood. [@JR1900057-2] The energy at maximum load represents the impact absorption capacity of a given material; a larger value of this property could, in theory, mitigate the impact of the suture on the pulley system of the osteofibrous canal in the anterior region of the fingers during articular movement, facilitating tendon slippage and hindering the formation of scar adhesions. The present study has methodological limitations: no mensuration of the necessary load to produce suture spacing that can, theoretically, impair healing; the use of continuous and longitudinal mechanical testing instead of cyclic and curvilinear tests; and, finally, the use of isolated swine tendons instead of human hand or finger tendons. However, within these limitations inherent to the methods used, one should keep in mind that the central basis of the present study was the comparison of the immediate mechanical properties of the "figure-of-eight" and Kessler sutures, both performed and tested under the same experimental conditions and, therefore, the results obtained have scientific validity. Conclusions =========== Under the conditions of this experiment and considering the use of flexor tendons of porcine fingers, the Al-Qattan and Al-Turaik [@JR1900057-3] triple "figure-of-eight" suture (six passages) [@JR1900057-3] is more resistant than the Kessler suture (two passages). The "figure-of-eight" suture with six passages enables active movement in the immediate rehabilitation of flexor tendon repair in fingers, with little risk of rupture or spacing of the suture. **Conflito de Interesses** Os autores declaram não haver conflito de interesses. *Trabalho Desenvolvido no Departamento de Cirurgia E Ortopedia, Faculdade de Medicina de Botucatu, Universidade Estadual Paulista (Unesp), Botucatu, SP, Brasil.* *Study Developed at The Department of Surgery and Orthopedics, Faculdade de Medicina de Botucatu, Universidade Estadual Paulista (UNESP), Botucatu, SP, Brazil.* Estudo biomecânico in vitro das suturas em "oito" e de Kessler de tendões flexores de porcos \* **Objetivo**  Avaliar as propriedades biomecânicas dos pontos de sutura tendinosa em "oito" e de Kessler. **Métodos**  Tendões flexores dos dedos de membros superiores de porcos foram divididos em dois grupos com suturas triplas centrais em "oito" (seis passagens) e de Kessler (duas passagens) associadas a suturas periféricas contínuas simples, e submetidos a ensaios mecânicos longitudinais contínuos, obtendo-se as propriedades mecânicas de carga máxima e de energia na carga máxima. **Resultados**  As médias da carga máxima e da energia na carga máxima na sutura em "8" foram de 63,4 N e 217,3 Nmm, respectivamente; na sutura de Kessler, os valores foram de 34,19 N e 100,9 Nmm, respectivamente. A análise estatística indicou que o ponto em "oito" é superior mecanicamente ao ponto de Kessler. **Conclusões**  Nas condições deste experimento e no tendão flexor de dedo de membro superior de porcos, o triplo ponto em "oito" (seis passagens) é mais resistente do que o ponto de Kessler (duas passagens). O triplo ponto em "oito", com seis passagens, permite movimentação ativa na reabilitação imediata de reparo de tendão flexor de dedo de membro superior com pouco risco de ruptura ou espaçamento na sutura. procedimentos ortopédicos fenômeno biomecânico traumatismos dos tendões traumatismos dos dedos técnicas de sutura Introdução ========== A necessidade da movimentação ativa nos pós-operatório dos reparos dos tendões flexores dos dedos da mão nas zonas II, III, IV e V, para evitar as aderências e obter amplitude de movimento adequada, exige pontos de sutura com grande resistência mecânica. [@BR1900057-1] [@JR1900057-2] Entre as várias qualificações do reparo ideal, como número de passadas, qualidades do fio, volume da sutura, entre outras, a facilidade de realização com mínimo trauma cirúrgico é fundamental. [@JR1900057-2] A sutura em "oito" com seis passadas é de fácil realização, pode ser feita com vários tipos de fios cirúrgicos, apresenta grande resistência mecânica para movimentação ativa no pós-operatório, e sua eficiência tem sido comprovada em estudos clínicos e biomecânicos. [@JR1900057-3] [@JR1900057-4] [@JR1900057-5] [@JR1900057-6] [@JR1900057-7] Apesar de não haver estudos sobre a preferência dos cirurgiões brasileiros em relação ao ponto de sutura dos tendões flexores dos dedos da mão, acredita-se, por observação empírica, que o de Kessler seja um dos mais utilizados. O método clássico para estudar as propriedades mecânicas de tendões, íntegros ou suturados, é submeter o corpo de prova a deformações de tensão com velocidade constante. [@BR1900057-8] O modelo experimental para testar biomecanicamente a sutura imediata de tendões flexores utilizando espécimes de porcos, por teste mecânico de tração longitudinal sob velocidade de tração constante, encontra referência na literatura. [@JR1900057-9] [@JR1900057-10] [@JR1900057-11] O objetivo do presente estudo foi avaliar biomecanicamente, por meio de teste de tração longitudinal com velocidade constante, a deformação por tensão dos pontos em "oito" e de Kessler em tendões flexores de porcos. Materiais e Métodos =================== Os membros superiores de 18 porcos foram desarticulados no cotovelo, acondicionados em sacos plásticos, e mantidos em freezer na temperatura de -20 graus Celsius. No dia dos experimentos, as peças anatômicas foram descongeladas à temperatura ambiente, e os tendões flexores profundos dos dedos foram dissecados e isolados. Os tendões dos membros superiores direitos foram divididos em dois grupos: grupo 8 (triplo ponto em "8") e grupo K (ponto Kessler). Os tendões dos grupos 8 e K foram seccionados na região central com lâmina de bisturi número 15 e submetidos a suturas: grupo 8 com triplo ponto em "8" com fio monofilamentar de polipropileno 3--0 (Prolene, Ethicon, São José dos Campos, SP, Brasil), e, no grupo K, ponto de Kessler com o mesmo fio cirúrgico; em ambos os grupos houve sutura periférica contínua com fio monofilamentar de polipropileno 4--0 (Prolene) ( [Figura 1](#FI1900057pt-1){ref-type="fig"} ). Após a sutura, os tendões foram fixados em garras metálicas sinusoidais de alumínio, comprimidas por parafusos com distância de 20 mm da região da sutura na parte central. As garras foram montadas axialmente em máquina universal de ensaio mecânico com célula de carga de 1.000 N e velocidade de aplicação de 30 mm/min (EMIC DL 10000, Instron, São José dos Pinhais, PR, Brasil). Após o ensaio, o computador acoplado à máquina forneceu as propriedades mecânicas de carga máxima (N) e de energia na carga máxima (Nmm). {#FI1900057pt-1} As análises estatísticas dos resultados foram realizadas com o teste *t* de Student, com valores de significância para *p*  \< 0,5. Resultados ========== Nos grupos 8 e K, as rupturas ocorreram sempre na área da sutura, não sendo possível determinar a sequência dos pontos rompidos, uma vez que não houve filmagem dos ensaios mecânicos. A [Tabela 1](#TB1900057pt-1){ref-type="table"} apresenta os resultados das propriedades mecânicas nos dois grupos, os quais indicam valores maiores para o grupo 8 ( *p*  \< 0,5). ###### Médias, desvios padrões, valores máximos e mínimos e analises estatísticas das propriedades mecânicas de carga máxima (N) e energia na carga máxima (Nmm) dos grupos experimentais K e 8 Grupos Carga máxima Energia na carga máxima ---------------- -------------------------------------------- -------------------------------------------- K ( *n*  = 8) 34,19 ± 11,4; máximo: 58,55; mínimo: 18,29 100,9 ± 52,48; máximo: 206,5; mínimo 34,61 8 ( *n*  = 10) 63,40 ± 20,40; máximo: 86,04; mínimo 23,17 217,3 ± 93,67; máximo: 365,7; mínimo 33,39 Valor de *p* 0,0024\* 0,0064\* Nota: \*p \< 0,5. Discussão ========= O presente estudo demonstrou que o ponto em "oito", na configuração de seis passadas, ou seja, feito três vezes, apresenta valores das propriedades mecânicas de carga máxima e energia na carga máxima maiores estatisticamente do que os do ponto de Kessler, resultados semelhantes aos do estudo de de Al-Qattan e Al-Turaiki. [@JR1900057-3] O valor da carga máxima de uma sutura de tendão flexor de dedos de membro superior para permitir movimentação ativa sem risco de ruptura ou formação de espaçamento na sutura é de pelo menos 40 N, valor maior do que o observado no grupo K (34,19 N) e inferior ao do grupo 8 (63,4 N), indicando segurança do triplo ponto em "8". [@BR1900057-12] O conhecimento da propriedade mecânica carga máxima é fundamental na avaliação da resistência de determinada sutura de tendão, sendo um dos parâmetros mais utilizados em estudos biomecânicos. [@JR1900057-2] [@BR1900057-8] [@BR1900057-12] Por outro lado, a importância clínica da propriedade energia na carga máxima não está totalmente esclarecida. [@JR1900057-2] A energia na carga máxima representa a capacidade de absorção de impacto de determinado material; o valor maior dessa propriedade poderia, em tese, amenizar o impacto da sutura no sistema de polias do canal osteofibroso na região anterior dos dedos durante a movimentação articular, facilitando o deslizamento do tendão e dificultando a formação de aderências cicatriciais. O presente estudo apresenta limitações metodológicas: a não realização das mensurações da carga necessária para produzir um espaçamento na sutura que, em tese, pode prejudicar a cicatrização; a utilização de ensaio mecânico contínuo e longitudinal, em vez de testes cíclicos e curvilíneos; e, por fim, o uso de tendões isolados de porco, em vez de tendões humanos de mãos ou dedos. Contudo, apesar das limitações inerentes aos métodos utilizados, deve-se ter em mente que a base central do estudo foi a comparação das propriedades mecânicas imediatas das suturas em "oito" e de Kessler, ambas realizadas e testadas nas mesmas condições experimentais e, portanto, os resultados obtidos apresentam validade científica. Conclusão ========= Nas condições deste experimento e no tendão flexor de dedo de membro superior de porco, o triplo ponto em "oito" de Al-Qattan e Al-Turaik [@JR1900057-3] (seis passagens) é mais resistente do que o ponto de Kessler (duas passagens). O ponto em "oito" com seis passagens permite a movimentação ativa na reabilitação imediata de reparo de tendão flexor de dedo de membro superior com pouco risco de ruptura ou espaçamento na sutura.

**Dear Editor,** We read an interesting article entitled: "Cytokeratin7 expression in gastric and colorectal adenocarcinoma: Correlation with prognostic factors" in the Caspian Journal of Internal Medicine, 2015: 6 ([@B4]). We admire the authors of above paper for their valuable paper. In this article, it has been explained that in Iran, gastric and colorectal adenocarcinoma are the second and the fifth most common cancers respectively. It is then suggested to increase attention to prevention of colorectal cancer which is very important. On the other hand, almost all people consume fruits as the best and safest way to treatment and elimination of diseases. Fruits such as pomegranates and grapes have remedial purposes. It is worth noting that the foods that introduced in the Holy Qur\'an as good and useful foodstuffs, have many beneficial effects on health human and even they have more effective spiritual effects on humans' life. According to the Holy Qur\'an, pomegranate and grape grow in the gardens of paradise ([@B1]-[@B3]). Cancer develops when cells in the human body begin to grow out of control and crowd out normal cells ([@B4]). Certainly cancers are very dangerous, so attention to reduction and treatment is necessary especially colorectal cancer (CRC). Our goal is to evaluate the benefits of ellagic acid from pomegranates and grapes in CRC. Fruits and vegetables have been said to have a strong protective effect against CRC especially those with ellagic acid. The highest levels of ellagic acid are found in pomegranates and grapes. The anticarcinogenic effects of ellagic acid has been detected. Also, ellagic acid has an anti-inflammatory role in the treatment of chronic ulcerative colitis as to prevent the development of colon cancer ([@B5]-[@B8]). Scholars also suggested that ellagic acid is an efficient multiple-function protector against oxidative stress ([@B9]). Kao et al. have reported the anti-proliferative effects of ellagic acid on different CRC cell types ([@B10]). It should be noted that the pathway of PI3K/Akt plays a central role in tumor genesis in colon ([@B11], [@B12]) and ellagic acid can inhibit chemically-induced colorectal tumorigenesis via mechanism involving the inhibition of Akt phosphorylation at Ser473. Similarly, the inhibitory impact of ellagic acid on Akt phosphorylation (at Thr308 and Ser473) has been observed in another study (13, 14). Researchers have stated that ellagic acid and its metabolites can inhibit CaCo-2 (the Caco-2 cell line is a continuous cell of heterogeneous human epithelial colorectal adenocarcinoma cell) proliferation through cell cycle arrest (15). We conclude that grape and pomegranate juice consumption has remedial and preventive impacts in CRC. We recommend that further studies should be conducted to obtain the other benefits of ellagic acid and use the significant advantages in studies of intestinal health promoting properties and decrease rates of CRC.

Introduction ============ Pyometra is collection of purulent material which occurs when there is interference with its normal drainage. It is an uncommon condition with incidence of 0.1 to 0.5% of all gynecological patients and 13.6% of elderly gynecological patients.[@B1] Apart from genital tract malignancy and consequences of its treatment (radiotherapy), other benign conditions like endometrial polyp, fibroid, senile cervicitis puerperal infection and congenital cervical anomaly can lead to pyometra.[@B2] Stenosis of cervical canal leading to accumulation of pus in the uterine cavity, degenerative or necrotic process in the uterine wall can lead to spontaneous perforation of pyometra. Spontaneous rupture of uterus is an extremely rare complication of pyometra.[@B3] This case is being reported to present a spontaneous rupture of pyometra and generalized peritonitis managed by conservative surgery. Case Report =========== A 65-year postmenopausal lady, Para seven presented with foul smelling vaginal discharge associated with fever since fifteen days and pain in lower abdomen with mild distension of abdomen since five days. She was a known case of hypertension. On general examination her blood pressure was 180/120 mmHg. Her abdominal examination revealed distension of abdomen with tenderness in infraumblical region. Per speculum examination revealed normal cervix but frank pus was present which was sent for culture and sensitivity. On her pervaginal examination uterus was normal in size with mild tenderness. Ultrasonography and X ray abdomen in standing position were inconclusive. Magnetic resonance imaging (MRI) was done which revealed pyometra with uterine perforation in anterior wall with multiple loculated collection ([Fig. 1](#F1){ref-type="fig"}), endometrium and cervix were normal. Her Pap smear was normal. Laparotomy was done. About 500 cc pus was drained and sent for culture and sensitivity. About 1 into 1 cm rent was present on anterior wall of uterus ([Fig. 2](#F2){ref-type="fig"}). Both parametriums were thickened and inflammatory changes were present. Both fallopian tubes and ovaries were normal. Necrotic part of perforated area was excised and stitched. Peritoneal toileting was done. One intraabdominal drain was kept. Culture of the pus showed growth of *gram positive cocci*, *sensitive to piperacillin*. Patient was discharged on fifteenth post operative day. Discussion ========== Pyometra usually present in elderly women. Nearly more than 50% of nonperforated pyometra patients are asymptomatic.[@B4] Classical symptoms of these patients are purulent vaginal discharge, lower abdominal pain and postmenopausal bleeding. Non specific symptoms are common including vomiting, fever, nausea which leads to delayed diagnosis and eventually uterus gets perforated. Spontaneously perforated pyometra is difficult to diagnose preoperatively. The most frequent preoperative diagnosis are generalized peritonitis, pneumoperitoneum and perforated gastrointestinal (GI) tract.[@B5] Correct diagnosis can only be made by laparotomy in most of the cases. Ultrasonography is the first investigation that has high sensitivity in assessing pyometra but has limited role in the diagnosis of perforated pyometra. Additional diagnostic radiographic evaluation use for acute abdomen is computed tomography (CT) scan and MRI.[@B3] In our case-preoperative diagnosis of perforated pyometra was made by MRI. The treatment of ruptured pyometra is immediate laparotomy, peritoneal lavage, drainage, and/or simple hysterectomy.[@B2] In most of the cases peritoneal cavity irrigation followed by total hysterectomy and bilateral oophorectomy is done. In present case patient was frail and uncontrolled hypertensive. On MRI endometrium and cervix were normal, Pap smear was also normal. We performed laparotomy followed by peritoneal toileting and repair of perforation. Patient had good recovery in postoperative period and discharged on fifteenth postoperative day. We want to highlight that preoperative diagnosis of perforated pyometra is absolutely essential. Patient care can be individualized and in selective patients of ruptured pyometra, conservative approach at surgery can be opted. Conclusion ========== Although spontaneously perforated pyometra is rare, the condition must be borne in mind with regard to elderly women with acute abdominal pain. Preoperative diagnosis of perforated pyometra is absolutely essential because these patients are elderly, in poor general condition, and require prompt intervention. CT and MRI are diagnostic tools. Patient care can be individualized and in selective patients of ruptured pyometra, conservative approach at surgery can be opted. **Conflict of Interest:** No potential conflict of interest relevant to this article was reported. {#F1} {#F2}

Cancer Sci 107 (2016) 543--550 **Funding Information** This study was supported by a Grant‐in‐Aid for Scientific Research B from the Ministry of Education, Culture, Sports, Science and Technology of Japan and a Health Labour Sciences Research Grant from the Ministry of Health, Labour and Welfare of Japan. Multistep hepatocarcinogenesis is characterized by the following three phases: premalignant dysplastic nodules (DN) (low grade or high grade), early hepatocellular carcinoma (eHCC) and finally advanced HCC.[1](#cas12893-bib-0001){ref-type="ref"}, [2](#cas12893-bib-0002){ref-type="ref"}, [3](#cas12893-bib-0003){ref-type="ref"}, [4](#cas12893-bib-0004){ref-type="ref"} Histopathologic diagnosis of these three types of lesions was established by an international consensus in 2009 and was also described in the 4th edition of the World Health Organization (WHO) classification in 2010.[5](#cas12893-bib-0005){ref-type="ref"}, [6](#cas12893-bib-0006){ref-type="ref"} Imaging diagnosis of these lesions also supports the presence of premalignant or early malignant lesions without definite features of advanced HCC that can transition to advanced HCC during follow up. In particular, hemodynamic changes detected by contrast‐enhanced computed tomography or ultrasonography and hepatocellular changes detected by gadolinium ethoxybenzyl diethylene triamine pentaacetic acid‐enhanced magnetic resonance imaging are useful for evaluating the malignant potential of these lesions.[7](#cas12893-bib-0007){ref-type="ref"}, [8](#cas12893-bib-0008){ref-type="ref"} Clinically, equivocal lesions smaller than 2 cm without definite features of HCC are frequently encountered in high‐risk patients, such as those with chronic hepatitis and liver cirrhosis, and are generally followed up without treatment. Follow up of these nodules has revealed various possible changes, from disappearance or stability without change of nodule size or hemodynamics to increased nodule size and blood supply; that is, hypervascularization of the nodule.[8](#cas12893-bib-0008){ref-type="ref"}, [9](#cas12893-bib-0009){ref-type="ref"}, [10](#cas12893-bib-0010){ref-type="ref"} It is hypothesized that equivocal nodules on imaging diagnosis include regenerative nodules, dysplastic nodules (DN) and eHCC; however, the clinical behavior of these nodules is quite heterogeneous, and, in particular, clinical diagnosis of eHCC is difficult. Early hepatocellular carcinoma (eHCC) is characterized by an increased cell density, cellular and structural atypia, decreased intratumoral portal tracts, and the presence of variable numbers of unpaired arteries. In addition, stromal invasion is frequently observed; indeed, the presence of stromal invasion is one of the pathological features used to make a diagnosis of HCC.[5](#cas12893-bib-0005){ref-type="ref"} Together with these histological features, a panel of immunohistochemical markers is reportedly useful in making a diagnosis of eHCC. These molecular markers include heat shock protein 70 (HSP70)[11](#cas12893-bib-0011){ref-type="ref"} glypican‐3, glutamine synthetase,[12](#cas12893-bib-0012){ref-type="ref"} adenylate cyclase‐associated protein 2 (CAP2)[13](#cas12893-bib-0013){ref-type="ref"} and the polycomb gene product Bmi‐1.[14](#cas12893-bib-0014){ref-type="ref"} Most published studies have focused on the differential diagnosis of eHCC from DN; however, a detailed histopathological study of eHCC itself has not yet been conducted. The aim of the current study was to investigate the detailed histopathological features of eHCC. Consequently, we conducted immunohistochemical analyses of vascular changes and the expression of various molecular markers and conducted a detailed histopathological evaluation of the degree of cellular atypia, structural atypia and stromal invasion in 66 resected samples of small vaguely nodular type lesions. Materials and Methods {#cas12893-sec-0002} ===================== Patients {#cas12893-sec-0003} -------- From 40 patients who underwent hepatectomy at Keio University Hospital from 1990 to 2013, 66 nodules smaller than 25 mm in diameter that showed as small vaguely nodular lesions macroscopically and were pathologically diagnosed at the time as DN or eHCC were selected for this study from pathological records. The nodule sizes ranged from 5 to 25 mm, with a mean value of 11.1 mm and a median value of 10 mm. The patients were 34 men and 6 women aged between 46 and 79 years (mean age: 62.6 years; median age: 60 years.) Operative indications were tumor resection in 21 patients and living donor liver transplantation in 19 patients. The patient backgrounds are summarized in Table S1. This study was approved by the institutional review boards of Keio University School of Medicine. Pathological analysis {#cas12893-sec-0004} --------------------- All histological diagnoses and evaluations of immunohistochemical stainings were conducted by two pathologists (H.O. and Y.M.) without reference to clinicopathological data. If the initial evaluation provided different results, a consensus interpretation was reached after re‐examination. Histological diagnosis {#cas12893-sec-0005} ---------------------- Nodular lesions were re‐diagnosed as low‐grade dysplastic nodules (LGDN), high‐grade dysplastic nodules (HGDN), or eHCC based on the pathological diagnostic criteria of the WHO classification[6](#cas12893-bib-0006){ref-type="ref"} and the international consensus classification.[5](#cas12893-bib-0005){ref-type="ref"} Using H&E‐stained specimens, we observed and evaluated cellular atypia (size of the nucleus), structural atypia and tumor cell density throughout the whole tumor. Nuclear size (equal to/enlarged compared with non‐tumoral liver tissue) and tumor cellularity (more than twofold/less than twofold compared with non‐tumoral liver cellularity) were determined. Structural atypia was evaluated based on the presence of a scirrhous hepatocellular carcinoma‐like pattern composed of fibrous stroma (denoting a "scirrhous component") in the lesion (Fig. [1](#cas12893-fig-0001){ref-type="fig"}a). Stromal (portal area) invasion was evaluated in elastic Van Gieson‐stained sections. The extent of tumor invasion was evaluated according to the following four grades with reference to the elastic fiber seen in the outer circumference around the portal area: None, invasion was not found or was unclear; Suspected invasion, tumor cells were observed on the border part of the elastic fiber; Distinct invasion/mild, despite distinct invasion of the stroma in the portal area, further severe invasion was not found; Distinct invasion/severe, massive invasion (Fig. [1](#cas12893-fig-0001){ref-type="fig"}b,c) or invasion enclosing the portal vein and periphery of bile duct, invasion in an "Indian file pattern," or hepatic vein wall invasion (Fig. [1](#cas12893-fig-0001){ref-type="fig"}d--f). {#cas12893-fig-0001} Immunohistochemical analysis {#cas12893-sec-0006} ---------------------------- Formalin‐fixed, paraffin‐embedded serial tissue sections 4‐μm thick were placed on silane‐coated slides for immunohistochemical analysis. The sections were deparaffinized, rehydrated in xylene and grade‐diluted ethanol (50--100%), and then immersed for 20 min in 0.3% hydrogen peroxide in absolute methanol to block endogenous peroxidase activity. Immunohistochemical staining using mouse anti‐human HSP70 monoclonal antibody (1:500 dilution; SC‐24; SantaCruz Biotechnology, Santa Cruz, CA, USA), mouse anti‐human CD34 monoclonal antibody (1:50 dilution; QBEnd‐10; Dako, Glostrup, Denmark) and mouse anti‐human h‐caldesmon monoclonal antibody (1:100 dilution; hCD; Dako) was performed using a Leica Bond‐Max automated immunostainer (Leica Microsystems, Bannockburn, IL, USA). For CAP2 and Bmi‐1, the tissue sections were heated at 120°C in 0.01 mol/L sodium citrate buffer, pH 7.0, for 10 min before incubation with rabbit anti‐human CAP2 polyclonal antibody (1:4000 dilution) and mouse anti‐human Bmi‐1 polyclonal antibody (1:200 dilution; Upstate Biotechnology, Lake Placid, NY, USA), and immunohistochemical staining was performed using the ImmPRESS system (Vector Laboratories, Burlingame, CA, USA). For CAP2 and HSP70, tumor cells showing strong expression compared with non‐tumoral liver were considered to be positive (Fig. [2](#cas12893-fig-0002){ref-type="fig"}a,b), and for Bmi‐1, tumor cells showing clear dot‐pattern staining were considered to be positive (Fig. [2](#cas12893-fig-0002){ref-type="fig"}c). Subsequently, for each molecular marker, the number of positive tumor cells was counted by random selection of three areas of the tumor, and the average positive rate was calculated for evaluation. For CD34, positive sinusoidal vascular architectural areas in the scirrhous component of the tumor and in the rest of the tumor (Fig. [2](#cas12893-fig-0002){ref-type="fig"}d) were counted separately. We determined each areal rate of CD34 in 10% increments. All values less than 10% were counted as 5%, because there are no completely negative cases in the liver parenchyma in general, and a detailed measurement of positive areas evaluated at 10% or less may be difficult. For h‐caldesmon, the number of intratumoral positive vessels (arterial tumor vessels) in the non‐scirrhous component was counted in 10 visual fields at ×200 magnification (i.e. ×20 objective lens and ×10 ocular lens). In the scirrhous component; the number of h‐caldesmon‐positive vessels was counted in all countable visual ×200 fields. Then, the mean number per field was calculated and this was taken as the arterial vessel density (AVD) in the two regions (Fig. [2](#cas12893-fig-0002){ref-type="fig"}e). {#cas12893-fig-0002} Statistical analysis {#cas12893-sec-0007} -------------------- The Mann--Whitney *U‐*test was used to analyze the immunohistochemical staining results. A *P*‐value of \<0.05 between two groups was judged to be statistically significant. Hierarchical cluster analysis of all analyzed factors was performed using TM4 MeV (<http://www.tm4.org/index.html>). Heat maps were generated using Microsoft Excel 2010. Results {#cas12893-sec-0008} ======= Histological features of small vaguely nodular lesions and subclassification of high‐grade and low‐grade of early hepatocellular carcinoma {#cas12893-sec-0009} ------------------------------------------------------------------------------------------------------------------------------------------ Of the 66 nodules, 10 were diagnosed as LGDN (15.1%), 10 as HGDN (15.1%) and 46 as eHCC (69.8%). In the eHCC nodules, no or unclear stromal invasion, suspected stromal invasion, mild distinct stromal invasion and severe distinct stromal invasion were observed in 5, 12, 15 and 14 nodules, respectively. In addition, distinct structural atypia with the scirrhous component was found in 10 nodules. Some scirrhous components showed one regional feature per nodule, whereas other patterns showed multiple small scirrhous foci in a single nodule. Six nodules exhibited both the scirrhous component and severe distinct stromal invasion (13.0% of eHCC). Accordingly, we divided eHCC into two groups: 18 nodules (39.1% of eHCC) in which marked stromal invasion and/or the scirrhous component foci were found, and 28 nodules (60.9% of eHCC) that did not possess these features. These two types of lesions were provisionally subclassified as high‐grade early hepatocellular carcinoma (HGeHCC) and low‐grade early hepatocellular carcinoma (LGeHCC) (Fig. [3](#cas12893-fig-0003){ref-type="fig"}a,b) according to our proposed diagnostic criteria (Fig. [4](#cas12893-fig-0004){ref-type="fig"}), and a further study was conducted based on this classification. {#cas12893-fig-0003} {#cas12893-fig-0004} The number of nodules with large nuclei showed a tendency to increase from LGDN, to HGDN, to LGeHCC, and eventually to HGeHCC. Tumor cellularity was less than twofold compared with non‐tumoral liver in all LGDN cases and was more than twofold in approximately 80% of each other nodule type. LGeHCC and HGeHCC nodules tended to be larger than DN nodules (Fig. [5](#cas12893-fig-0005){ref-type="fig"}a). {#cas12893-fig-0005} Immunohistochemical features of small vaguely nodular lesions {#cas12893-sec-0010} ------------------------------------------------------------- We further evaluated these lesions immunohistochemically. CAP2 expression‐positive rates tended to increase stepwise from LGDN to HGeHCC. Although the immunohistochemical expression of CAP2 did not show a statistically significant difference between LGDN and HGDN or between HGDN and LGeHCC, there was a statistically significant difference between LGeHCC and HGeHCC (*P* = 0.0145) and between LGDN and LGeHCC (*P* = 0.0034) (Fig. [5](#cas12893-fig-0005){ref-type="fig"}b). In the present study, it was noted that a strong expression of CAP2 was frequently observed in tumors showing stromal invasion, hepatic vein wall invasion or a scirrhous component (Fig. [6](#cas12893-fig-0006){ref-type="fig"}a--c). {ref-type="fig"}b), (b) hepatic vein wall invasion (arrowheads, the same area of Fig. [2](#cas12893-fig-0002){ref-type="fig"}d and e) and (c) scirrhous component. (d) Areas of scirrhous component in HGeHCC were highlighted by very high levels of arterial vessel density (AVD), as indicated by the h‐caldesmon‐positive vessels. Scale bars = 100 μm.](CAS-107-543-g006){#cas12893-fig-0006} Both HSP70 and Bmi‐1 showed statistically significant (or tendencies to have) low positive rates in DN but higher rates in eHCC. The immunohistochemical expression of HSP70 showed statistically significant differences between LGDN and LGeHCC (*P* = 0.0008) and between LGDN and HGeHCC (*P* = 0.0018) (Fig. [5](#cas12893-fig-0005){ref-type="fig"}c). The immunohistochemical expression of Bmi‐1 showed a statistically significant difference between LGDN and LGeHCC (*P* = 0.0193) but no significant differences between other combinations (Fig. [5](#cas12893-fig-0005){ref-type="fig"}d). Vascularization of high‐grade early hepatocellular carcinoma and their scirrhous component {#cas12893-sec-0011} ------------------------------------------------------------------------------------------ Vascular changes were examined in 65 nodules (excluding 1 nodule that became too small to examine during the preparation of serial sections). The area of CD34‐positive sinusoid‐like vascular architecture (sinusoidal capillarization) in lesions tended to gradually increase from LGDN to HGeHCC, but there were no significant differences. The scirrhous component of HGeHCC showed a very high positive rate compared with the non‐scirrhous component of HGeHCC (*P* = 0.0014) (Fig. [5](#cas12893-fig-0005){ref-type="fig"}e). AVD was low in LGDN, HGDN and LGeHCC, with no significant differences between them. However, AVD was high in the non‐scirrhous component of HGeHCC, being significantly different from that of LGeHCC (*P* = 0.0054). Moreover, the scirrhous component of HGeHCC was highlighted by a very high AVD compared with the non‐scirrhous component of HGeHCC (*P* = 0.0047) (Figs [5](#cas12893-fig-0005){ref-type="fig"}f,[6](#cas12893-fig-0006){ref-type="fig"}d). Comparison between pathological diagnoses and clustering of pathological factors in small vaguely nodular lesions {#cas12893-sec-0012} ----------------------------------------------------------------------------------------------------------------- A heat map was created for the four types of lesions (LGDN, HGDN, LGeHCC and HGeHCC) with raw data for each parameter analyzed (lesion size; tumor cellularity; size of nucleus; structure of tumor cells; extent of stromal invasion; immunohistochemical reactivity of CAP2, HSP70 and Bmi‐1; the areal rate of CD34; and AVD) (Fig. S1). In addition to the original criteria defining high grade (i.e. a scirrhous component and severe stromal invasion), HGeHCC showed a tendency toward nuclear swelling and high positive expression of CAP2, CD34 and AVD compared with LGeHCC and DN. In addition, we performed cluster analysis on each pathological factor (Fig. [7](#cas12893-fig-0007){ref-type="fig"}a), and this resulted in lesions being divided into three groups (Fig. [7](#cas12893-fig-0007){ref-type="fig"}b). Group I was composed of HGeHCC only. Group II mainly comprised HGDN and LGeHCC. Group III was mainly composed of LGDN; in fact, all LGDN appeared in group III. {#cas12893-fig-0007} Discussion {#cas12893-sec-0013} ========== In the present study, we divided eHCC into two subclasses, HGeHCC and LGeHCC, and conducted a detailed histological study comparing them with LGDN and HGDN. We found that HGeHCC and LGeHCC accounted for approximately 40% and 60% of eHCC, respectively. Compared with LGeHCC and DN, HGeHCC tended to have large nuclear size, high cellularity, structural atypia (including scirrhous pattern), and large tumor size together with marked stromal invasion. Immunohistochemically, both the expression of CAP2 and the areal ratio of sinusoidal capillarization were upregulated in a stepwise manner from LGDN, to HGDN, to LGeHCC, and eventually to HGeHCC. In contrast, AVD was almost unchanged in LGeHCC compared with DN, whereas AVD in HGeHCC was higher. Statistically significant differences between LGeHCC and HGeHCC were found for CAP2 expression and AVD. Moreover, vascularization in the scirrhous component of HGeHCC was different from that in the non‐scirrhous component of HGeHCC. Based on the above findings, we considered that HGeHCC is histopathologically distinct from LGeHCC. In a previous study using an oligonucleotide array, we found that CAP2 was upregulated in early HCC,[11](#cas12893-bib-0011){ref-type="ref"} and high immunohistochemical expression of CAP2 was shown in HCC with increased malignant potential, such as those with poor tumor differentiation, vascular invasion and intrahepatic metastasis.[13](#cas12893-bib-0013){ref-type="ref"} Therefore, a significant overexpression of CAP2 in HGeHCC likely indicated increased malignant potential of the lesion. Moreover, hypervascularization in small lesions in chronic liver disease is thought as an indicator of increased malignant potential. AVD and the area with sinusoidal capillarization in HGeHCC were high in comparison with LGeHCC and DN. Indeed, during follow up, small vaguely nodular lesions generally show poor arterial blood flow on hemodynamics imaging;[15](#cas12893-bib-0015){ref-type="ref"}, [16](#cas12893-bib-0016){ref-type="ref"} however, approximately 10--40% of such lesions may become hypervascular tumors within a year.[17](#cas12893-bib-0017){ref-type="ref"}, [18](#cas12893-bib-0018){ref-type="ref"}, [19](#cas12893-bib-0019){ref-type="ref"} It has also been reported that tumor cells of nodular lesions larger than 15 mm, which is similar to the 13 mm mean size of HGeHCC, actively proliferated with the emergence of unpaired arteries.[20](#cas12893-bib-0020){ref-type="ref"} We consider that HGeHCC has distinct histological malignant features that may represent a transitional stage to advanced HCC. While HSP70 and Bmi‐1 are not correlated with the malignant potential of HCC, they are valuable for differentiation between DN and eHCC, as reported previously.[11](#cas12893-bib-0011){ref-type="ref"}, [14](#cas12893-bib-0014){ref-type="ref"} We found no significant differences in the expressions of HSP70 and Bmi‐1 between LGeHCC and HGeHCC, although Bmi‐1 expression tended to decrease slightly. This is in agreement with our previous observation of a higher expression of Bmi‐1 in well‐differentiated HCC (including early HCC) than in moderate and poorly differentiated HCC.[14](#cas12893-bib-0014){ref-type="ref"} The scirrhous pattern or marked stromal invasion was frequently observed in eHCC. These are generally characteristics that support a diagnosis of cancer; however, it should be noted that these features are not pronounced in advanced HCC. Because we observed marked vascularization in the scirrhous component, we believe that the scirrhous component may be an indicator of the high neovascularization capabilities of HGeHCC. Marked stromal invasion also may be an indicator of high invasiveness, resulting in portal vein invasion and intrahepatic metastasis. Therefore, we consider that the criteria presented here for classifying HGeHCC (i.e. the presence of foci with a scirrhous component and/or marked stromal invasion) are reasonable. To assess the validity of our pathological diagnosis of nodules, the pathological factors that we examined were analyzed using cluster analysis. The clustering of pathological factors and the pathological diagnosis (LGDN, HGDN, LGeHCC and HGeHCC) were compared. Interestingly, cluster analysis showed that lesions were divided into three groups. Group I was made up entirely of HGeHCC, and 10 of the 17 HGeHCC nodules (59%) were subclassified into this group. Group II mainly comprised HGDN and LGeHCC, and group III was dominated by LGDN (all 10 LGDN were included in group III). These results further confirmed the distinct features of HGeHCC that we classified histopathologically. The scirrhous component is one of the key histopathological factors between group I and group II. Group II may indicate that the pathological features of LGeHCC (i.e. eHCC excluding HGeHCC) may overlap with those of HGDN. The above findings suggest that HGDN and LGeHCC may be considered "borderline" lesions, whereas HGeHCC is a distinct malignant tumor. In addition, immunohistochemical expression of CAP2 and CD34, and AVD may assist diagnosis of HGeHCC, even if marked stromal invasion or a microscopic scirrhous component is not evident in biopsy specimens. However, further analysis is necessary using a large series of cases with the inclusion of biopsy specimens. In conclusion, we investigated detailed histological and immunohistochemical features of small vaguely nodular lesions. Our results indicated that from the perspective of histopathology, immunohistology and changes in vascularization, approximately 40% of eHCC cases belonged to a distinct, highly malignant tumor group that we subclassified as HGeHCC. In addition, we believe that HGeHCC may already be a transitional stage to advanced HCC. We consider that our grading classification system may be valuable for evaluating treatment strategies for eHCC with a size of approximately 2 cm. Disclosure Statement {#cas12893-sec-0015} ==================== The other authors have no conflict of interest to declare. Supporting information ====================== ###### **Fig. S1.** Heat map of the raw analyzed parameters in each lesion. ###### Click here for additional data file. ###### **Table S1.** Patient background summary. ###### Click here for additional data file. We sincerely thank Mr Hiroshi Suzuki, Dr Yutaka Kurebayashi and Dr Ken Yamazaki for their kind advice on immunohistochemistry and statistics.

Introduction {#Sec1} ============ The two species of mosquitoes, *Culex pipiens* and *Aedes albopictus*, are important vectors of *Dirofilaria immitis* and *Dirofilaria repens* worldwide (Licitra et al. [@CR18]; McKay et al. [@CR20]). *D. immitis*, the agent of heartworm disease, causes severe disorders and even death in dogs in many parts of the world (McCall et al. [@CR19]). The prevalence of both dirofilariasis is very high in the USA (Carleton and Tolbert [@CR6]; Bowman et al. [@CR4]; McKay et al. [@CR20]), in Central America (Bolio-Gonzalez et al. [@CR2]; Caro-Gonzalez et al. [@CR7]), Asia (Oi et al. [@CR23]), Russia (Ermakova et al. [@CR10]), and in some European countries (Genchi et al. [@CR13]). Endemic areas are present in the south of France, in Italy (Capelli et al. [@CR5]; Giangaspero et al. [@CR15]), Spain (Montoya-Alonso et al. [@CR22]), Portugal (Santa-Ana et al. [@CR26]), Germany (Sassnau et al. [@CR27]), Poland (Demiaszkiewicz et al. [@CR8]), Hungary (Farkas et al. [@CR12]; Tolnai et al. [@CR30]), and in Romania (Mircean et al. [@CR21]). In addition, the filarial nematodes, *D. immitis* and *D. repens*, are zoonotic agents (Theis, [@CR29]). In humans, ocular, subcutaneous, and pulmonary forms have been reported (McCall et al. [@CR19]; Kalogeropoulos et al. [@CR17]; Otranto et al [@CR24], [@CR25]). An integrated control program against dirofilariasis may be implemented by the association of macrocyclic lactones and the application of insecticides with an antifeeding effect on mosquitoes (Hellmann et al. [@CR16], Snyder et al. [@CR28]; Genchi et al. [@CR14]; Traversa et al. [@CR31]; Di Cesare et al. [@CR9]). Pyrethroids such as permethrin and deltamethrin are known to be effective against sandflies and mosquitoes and are widely used in companion animals (Beugnet and Franc [@CR1]). Other molecules such as fipronil, metaflumizone, and pyriprole are effective against fleas and tick, but could not be used to prevent mosquitoes from biting dogs (Bouhsira et al. [@CR3]). The aim of the study was to evaluate the adulticidal (or mortality) and repellent effects of a spot-on containing fipronil and permethrin (Effitix®, Virbac, Carros, France) on *C. pipiens* in dogs. Materials and methods {#Sec2} ===================== The study was conducted in the National Veterinary School of Toulouse (ENVT) and was a single-center, randomized, blinded, controlled efficacy study on two groups of seven dogs each. Dogs were handled in accordance with the Animal Welfare and Good Clinical Practice, and the study protocol was approved by the Ethics Committee of Midi-Pyrenees. All personnel involved in the collection of efficacy data were blinded to the treatment. Dogs {#Sec3} ---- Five male and seven female Beagle dogs (3 years of age and weighing between 8.02 and 10.94 kg) were included in the study. They had not been exposed to ectoparasiticides for 3 months prior to the inclusion and remained in good health throughout the study. They were housed in cages individually and had a 4-h daily access to a 2 × 4 m concrete run without contact with another dog. To avoid cross-contamination, treated and untreated dogs were placed in two different exercise areas. Each dog was identified with the number of a subcutaneously implanted microchip. They were fed a commercial dry dog food with a ration that maintained the animal in a healthy physical state. Water was available ad libitum through automatic lickers. Dogs were maintained and handled with due regard for their welfare and were acclimatized to the caged environment for 17 days prior to treatment. They were observed daily for their general health conditions throughout the trial. No concurrent medication was needed to be given during the study. On day −7, each dog was challenged with 80 unfed adult females of *C. pipiens*. The number of engorged female mosquitoes was used for ranking and group allocation. Dogs were ranked in descending order according to their individual pre-treatment mosquito's engorgement status. They were then introduced into blocks of two animals each, and within each block, dogs were randomly allocated into two groups: treatment or control group. Mosquito maintenance and supply {#Sec4} ------------------------------- The *C. pipiens* exposure was performed using laboratory-reared adults (females only). This mosquito strain was obtained from the Interdepartmental Agreement for Mosquito Control (EID) and was maintained at ENVT under laboratory conditions since 2001 using a 3-week egg to adult. Treatment {#Sec5} --------- The six dogs from the control group (group A) remained untreated, and the six dogs from the treated group (group B) received on day 0 a spot-on combination of permethrin and fipronil: one pipette of 1.1 ml (593.4 mg of permethrin and 67.7 mg of fipronil) for dogs weighing between 4.1 and 10.0 kg and one pipette of 2.2 ml (137.2 mg of fipronil and 1197.8 mg of permethrin) for dogs weighing between 10.1 and 20 kg. Treatment dosages were within the range of 67.7--137.2 mg kg^−1^ for permethrin and 7--13.5 mg kg^−1^ for fipronil. For all treated animals, the formulation was applied according to the manufacturer's instructions by parting the hair and applying the formulation directly onto the skin in two areas: between the shoulder blades and the lumbar area. All dogs were observed at 2 and 4 h after treatment for any adverse reaction to the product. Experimental procedure {#Sec6} ---------------------- The 12 dogs were infested with 80 (±2) *C. pipiens* for a total of six times. Two days before exposure, mosquitoes were aspirated from their breeding cage with a vacuum pump and then placed in challenge nets (80 ± 2 females per net) with access to water-soaked cotton and honey. The mosquito challenge assessment cages (60 cm × 40 cm × 50 cm) were constructed from mosquito netting mounted on a wooden frame and placed in environmentally controlled rooms. Mosquitoes were fasted 24 h prior to exposure to dogs by removing honey from the cages. Before exposure, dogs were sedated by intramuscular injection of medetomidine (Dexdomitor®, Pfizer Santé animale, Paris, France), ketamine (Clorketam®, Laboratoire Vetoquinol S.A., Lure, France), and diazepam(Valium®, Roche injectable, Neuilly s/ Seine, France) at a dose rate of 4 μg/kg, 9 mg/kg, and 5 mg/dog, respectively, and then placed in individual infestation proof nets containing mosquitoes. The dosage of the anesthetic was calculated so as to immobilize dogs for 90 min. During infestation, treated dogs and control dogs were placed in separated infestation rooms where temperature and relative humidity were maintained between 25 and 26 °C and between 58 and 72 %, respectively. Cages and nets were thoroughly cleaned after each mosquito challenge. After 90 ± 5 min of exposure, the dogs were carefully taken out of the net and examined for any dead mosquito on their body and then placed back in their cage. All live mosquitoes were aspirated from each challenge net using a vacuum pump and were categorized as live engorged or non-engorged. All dead mosquitoes were collected, counted, and categorized as dead non-engorged or dead engorged. On days −7, 1, 7, 14, 21, and 28, live mosquitoes recovered from individual animals at the end of exposure were placed in separate nets and kept in the experimental room. The mosquitoes were fed on sugar--water and checked for mortality after 24 h. Then, all remaining mosquitoes were discarded. Data analysis {#Sec7} ============= Antifeeding effect {#Sec8} ------------------ For each time point after exposure, the antifeeding effect was calculated as described by:$$\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$ \mathrm{Antifeeding}\;\mathrm{e}\mathrm{ffect}=100*\frac{\mathrm{Ce}-\mathrm{T}\mathrm{e}}{\mathrm{Ce}} $$\end{document}$$ where Ce was the arithmetic mean of engorged female mosquitoes (live engorged and dead engorged) for the control group and Te was the arithmetic mean of the engorged female mosquitoes for the treated group. Mortality effect {#Sec9} ---------------- For each time point after exposure, the mortality effect was evaluated for each group as described by:$$\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$ \mathrm{Mortality}\;\mathrm{effect}=100*\frac{\mathrm{Cl} - \mathrm{Tl}}{\mathrm{Cl}} $$\end{document}$$ where Cl was the arithmetic mean of live female mosquitoes (live engorged and live unengorged) for the control group and Tl was the arithmetic mean of the live female mosquitoes for the treated group. The mortality effect was calculated at 90 min and 24 h post-exposure. Statistical analysis {#Sec10} -------------------- Both groups were compared for the number of engorged females and the number of dead females at each challenge point using the non-parametric test of Kruskal--Wallis. The analyses were performed with Systat 9 software; differences were considered significant at *p* \< 0.05. Results {#Sec11} ======= No adverse events relative to treatment were reported. Antifeeding effect on mosquitoes {#Sec12} -------------------------------- The 12 dogs included in the study demonstrated adequate pre-treatment parasite-holding ability (i.e., over 50 % of engorged females per dog; Fig. [1](#Fig1){ref-type="fig"}). On day −7, the percentage of engorged females was 78.1 and 76.6 % for the treated and control group, respectively. All control dogs maintained an adequate number of engorged females (i.e., between 76.6 and 83.7 %) throughout the study (Fig. [1](#Fig1){ref-type="fig"}). The treatment had an anti-feeding effect between 100 and 99.5 % during the first 2 weeks and between 98.3 and 96.7 % until the end of the trial (Table [1](#Tab1){ref-type="table"}).Fig. 1Mean number of engorged *Culex pipiens* females after 1 h exposure to control and treated dogs. Dogs were treated on day 0 with a permethrin and fipronil combination spot-on and then weekly challenged with 80 *Culex pipiens* femalesTable 1Mortality and antifeeding effect of a permethrin and fipronil combination against *Culex pipiens* on dogsDay 1Day 7Day 14Day 21Day 28Mortality effect (%)1.5 h66.655.93817.212.324 h69.158.13817.815.1Anti-feeding effect (%)1 h10099.597.798.396.7 At each challenge point post-treatment, the difference in engorgement status of *C. pipiens* females between treated and controlled group was significant (*p* \< 0.05). Mortality effect on mosquitoes {#Sec13} ------------------------------ The mortality effects of the treatment calculated at 1 and 24 h post-exposure to treated dogs are reported in Table [1](#Tab1){ref-type="table"}. The mortality effect observed at 1 h ranged from 66.6 to 55.9 % in the first 2 weeks and then decreased dramatically to values ranging from 38 to 12.3 % until the end of the study. The mortality effect of the formulation has not increased within the 24 h post-exposure and was close to the one obtained at 1 h. At each challenge point, there was a significant difference (*p* \< 0.05) in the number of dead mosquitoes found at 1 and 24 h of exposure between the treated and control group. Discussion {#Sec14} ========== The aim of the study was to determine the antifeeding (or repellency) and mortality (or insecticidal) efficacies of a new formulation combining fipronil and permethrin (Effitix®, Virbac, Carros, France) against a European strain of *C. pipiens* in dogs. The formulation was administered according to the manufacturer's recommendations and the dogs received, according to their weight, between 67.7 and 137.2 mg.kg^−1^of permethrin and between 7 and 13.5 mg.kg^−1^of fipronil. In these conditions, the treatment provided an immediate efficacy of 100 % at 24 h after the administration. Then, the formulation provided an excellent inhibition of feeding which remained above 96.7 % for the 4 weeks of the study. The repellency of fipronil combined with permethrin obtained in our study was higher than the repellency obtained with the same association against an American strain of *C. pipiens* (Fankhauser et al. [@CR11]). In this study, the authors obtained a repellency rate of 99.4, 98.9, 94.7, 91.7, and 90.4 % on days 1, 7, 14, 21, and 28, respectively. This could be explained by the fact that dogs were treated at the minimal dose of the product, i.e., 50.48 mg kg^−1^ of permethrin and 6.76 mg kg^−1^ of fipronil versus 67.7 mg kg^−1^ of permethrin and 7 mg kg^−1^ of fipronil in our study. A previous study had been carried out in the same laboratory as the current study to assess the repellency and the insecticidal efficacy of fipronil combined with (S) methoprene on the same European strain of *C. pipiens* (Bouhsira et al. [@CR3]). The repellency rate was 40.2 % on day 1 and 45.8 % on day 7 respectively while the insecticidal efficacy was 32.3 and 51.8 % on the same days. This combination had a reduced activity on *C. pipiens* which could not be considered as efficient enough to protect dogs against this mosquito. Therefore, the excellent inhibition of feeding provided by Effitix® is mainly due to permethrin. The product demonstrated mortality or insecticidal efficacy close to 60 % the first week after treatment which then decreased on days 14, 21, and 28. Fankhauser et al. ([@CR11]) obtained an insecticidal efficacy between 92.1 and 26.9 %. This lower insecticidal efficacy was explained by the strong repellent effect on *C. pipiens*, which could prevent them from landing on the treated animals and therefore limiting contact with the insecticide molecules (Fankhauser et al. [@CR11]). In conclusion, the new fipronil and permethrin ectoparasiticide combination offers a protection against *C. pipiens* in dogs for 1 month following a single topical application. This treatment could contribute to the reduction of stress and annoyance caused by the bite of mosquitoes, and more importantly, it may reduce the risk of heartworm transmission in animals living in or travelling to dirofilariasis endemic areas. However, it should not be seen as a substitute for heartworm prevention treatment but should be part of an integrated prevention program. The authors are grateful to Sonia Gounaud and Martine Roques for their help with the mosquito colony maintenance and the handling of dogs. This study was funded by a grant from Virbac.

Introduction {#s1} ============ Increases in muscle strength through resistance exercise are essential for both clinical and athletic populations and consist of shortening and lengthening contractions that possess their own unique properties. Lengthening contractions can increase lengthening strength three-fold and neural output (EMG) seven-fold more compared to shortening contractions following shortening resistance training (Hortobágyi et al., [@B29]). In a recent meta-analysis (Roig et al., [@B43]), it was shown that lengthening contractions are more effective at increasing strength and muscle mass, though the neurological adaptations were not discussed. Despite many studies supporting the notion that lengthening contractions are superior to shortening contractions not all research concurs (Higbie et al., [@B26]; Andersen et al., [@B3]). Lengthening contractions have generally been shown to have a positive effect on clinical populations such as stroke patients (Engardt et al., [@B17]). In athletes, increases in lengthening contraction strength are associated with a reduction in injury prevalence (Jönhagen et al., [@B33]). Surface EMG, a surrogate measure of neural drive detected at the muscle, has shown muscle activity to be reduced following the cessation of resistance training (Häkkinen and Komi, [@B23]; Gondin et al., [@B21]); however, it is not clear if these changes in neural drive are attributable to supraspinal and/or spinal mechanisms. Investigating the unique control strategies (Duchateau and Enoka, [@B12]; Duclay et al., [@B15]) of shortening and lengthening contractions and the neural modifications from resistance training and detraining will aid clinicians and practitioners in training and rehabilitation programme design. Understanding where the adaptations (cortical, spinal or post-synaptic) occur for each muscle contraction will assist practitioners in maximizing neurological and consequently, strength adaptations. To detect and monitor changes in the central nervous system (CNS) from resistance training, transcranial magnetic stimulation (TMS) and peripheral nerve stimulation (PNS) have been used to identify the possible mechanisms of adaptation (Kidgell et al., [@B35]; Tallent et al., [@B46]). However, despite the large body of research focusing on the early neurological adaptations to resistance training, little continuity exists in the literature. An increase (Griffin and Cafarelli, [@B22]; Kidgell et al., [@B35]), decrease (Carroll et al., [@B7]), or no change (Carroll et al., [@B6]) in corticospinal excitability and reduced (Kidgell and Pearce, [@B34]) or no change in inhibition (Kidgell et al., [@B35]; Weier et al., [@B48]) have been reported following a period of resistance training. At a spinal level, an increased excitability has been reported in some studies (Aagaard et al., [@B2]; Gondin et al., [@B21]), but not others (Holtermann et al., [@B27]; Ekblom, [@B16]). However, V-wave has consistently been shown to increase (Aagaard et al., [@B2]; Gondin et al., [@B21]; Ekblom, [@B16]; Vila-Chã et al., [@B47]). Due to reasons such as reproducibility, research has focused predominantly on assessment during isometric contractions. Lengthening contractions show a greater neuromuscular response to training than shortening contractions. It has been shown that shortening muscle contractions coupled with lengthening contractions have a greater preservation of strength, which is potentially due to the greater neurological adaptations from lengthening contractions (Colliander and Tesch, [@B10]). Conversely, adaptations following lengthening training have also been shown to be less susceptible to detraining (Andersen et al., [@B3]). The neurological mechanisms behind these preservations are unclear. A greater understanding of the contributing factors to the decreases in strength in response to detraining has important applications for the design of taper strategies, but more importantly furthers our understanding of how detraining or inactivity affects neuromuscular function (Bosquet et al., [@B4]). The collective use of tools, such as TMS and PNS, to explore the nervous system during shortening and lengthening contraction could further our knowledge of the responses to resistance training and detraining following task specific shortening or lengthening contractions. With little understanding of the nervous system following detraining, TMS and PNS may also further our understanding of periods of inactivity. The study was split into two parts/aims. The initial aim of this study was to investigate the response to 4 weeks shortening or lengthening resistance training; secondly, we aimed to examine the detraining response following 2 weeks of no training. TMS and PNS was employed to investigate corticospinal and spinal task-specific changes following training and detraining. It was hypothesized that corticospinal and spinal adaptations following resistance training would be modulated in a contraction-specific manner. It is also suggested that such changes would be greater following lengthening resistance training because of the potential greater training responses observed following this mode of exercise. Materials and methods {#s2} ===================== Subjects -------- Following institutional ethical approval from the University of Northumbria Ethics Committee and in accordance with the Declaration of Helsinki, 31 male volunteers completed a health-screening questionnaire and provided written, informed consent. Participants had no structured resistance training history in the preceding 2 years and were randomly assigned to either a shortening resistance training (SHO; *n* = 11), lengthening resistance training (LEN; *n* = 11), or a control (CON; *n* = 9) group. Mean ± *SD* age, stature and mass were 24 ± 3, 24 ± 3 and 27 ± 4 yrs, 175.9 ± 10.6, 176.3 ± 9.6, 172.7 ± 8.5 cm, and 77.1 ± 10.2, 75.7 ± 12.3, 74.7 ± 11.1 kg, respectively). Participants were asked to refrain from any form of resistance training throughout the duration of the study. Footedness was assessed using a questionnaire (Hebbal and Mysorekar, [@B24]); of the 31 participants, 29 were right leg dominant. Study design ------------ The training groups conducted an initial 4 weeks of resistance training and then 2 weeks detraining, with the CON group remaining inactive throughout the 6 week period. Participants allocated to the resistance training group reported to the laboratory on 17 occasions; 5 times for assessments and 12 times for training sessions. The CON group took part in the five assessments only. All participants performed a familiarization session 24 h before the pre-testing assessment (Tallent et al., [@B46]). Midpoint assessment was conducted after 2 weeks (six training sessions) and post-training assessment was after 4 weeks (12 training sessions). Final measures were taken after 2 weeks of detraining (weeks 6) where participants were instructed to remain inactive during this time. Preliminary power analysis using GraphPad StatMate (v5.0, San Diego, CA, USA), and previous work (Hortobágyi et al., [@B30], [@B28]), revealed that to achieve a power of 0.80 (α = 0.05), 10 participants per experimental group were needed. However, as no participants withdrew from the study, the experimental groups consisted of 11 subjects in each individual experimental training group with nine in the control group. Resistance training ------------------- All assessment and training sessions were conducted on an isokinetic dynamometer (Cybex Norm, Cybex International, NY). The experimental groups performed 12 sessions of LEN or SHO resistance training of the tibialis anterior (TA) muscle. The TA was used due to its uniquely high corticospinal drive (Capaday et al., [@B5]) and accessibility of the common peroneal nerve through electrical stimulation. Training consisted of 3 sessions per week. Session 1--5 and 7--11 were conducted at 80% of the relative contraction specific MVC. Sessions 6 and 12 were reduced to 50% MVC to minimize any potential fatigue during the assessment sessions. Following a warm-up set of 8 reps at 50% MVC, participants performed 5 sets of 6 repetitions at 80% MVC of contraction specific MVC at a speed of 15°/s with 2 min rest between sets. Supervised training was conducted under identical conditions to the assessment sessions apart from a shorter 2 s rest period between each repetition. All participants performed all training sessions, with training sessions taking maximum of 20 min. MVC was recorded during each assessment session and training torque was adjusted where necessary at the mid assessment point. Experimental set-up ------------------- Participants were set-up on the isokinetic dynamometer in accordance with manufacturer\'s guidelines to examine the dorsiflexor muscle group (hip, knee and ankle of the dominant leg set at joint angles of 90, 120, and 90°, respectively). The knee and foot was strapped into avoid any unwanted movements in the leg. Whilst performing contractions participants were instructed not to grip the handles of the dynamometer. Torque feedback was displayed on the dynamometer\'s monitor which was set 1 m away from the participant. Participants moved the ankle joint through a 30° range (75°--105°, where 90° was anatomical zero) of dorsi- and plantar-flexion at a speed of 15°/s, resisting for lengthening contractions and assisting for shortening contractions. Following MVC assessments in each laboratory visit, TMS and PNS was performed in a randomized order. Maximal voluntary contraction (MVC) ----------------------------------- After a warm-up, consisting of shortening and lengthening (SHO and LEN) contractions at 60, 80, and 90% of maximal effort, three SHO and LEN maximal voluntary contractions (MVC) were performed. Participants had three MVC attempts with 2 min rest between each contraction. If the third MVC attempt recorded the highest value a fourth MVC was performed. The highest torque as the ankle passed 90° for each individual MVC was recorded. From these values, 80, 50, 25, and 15% of shortening and lengthening MVC were calculated. TMS protocol ------------ The repeatability of the TMS and PNS measures used in this study have been previously established during both shortening and lengthening contractions performed with the TA (Tallent et al., [@B46]). Consequently, the same protocol was used to assess corticospinal and spinal changes resulting from the training intervention. MEPs were elicited via stimulation on the hemisphere contralateral to the dominant leg using a magnetic stimulator (Magstim 200^2^, Magstim Company Ltd, Whitland, UK), with a concave double-coned 110 mm coil (maximal output of \~1.4 T). EMG was recorded in a 550 ms window (50 ms before magnetic stimulation and 500 ms after). Positioning for the optimal coil placement began 1 cm posterior to the vertex, along the sagittal plane. The coil was set-up to induce a postero-anterior current in the underlying motor cortex. The optimal site was marked with a permanent marker to ensure consistency in coil placement across the trial. Resting motor threshold (rMT) was defined as the lowest stimulator output to elicit a peak-to-peak MEP amplitude (≥50 μV) in 5 out of 10 consecutive pulses (Rossini et al., [@B44]). All subsequent pulses in the experiments were delivered at 120% of rMT. All TMS pulses were delivered when the ankle joint angle was at 90° (anatomical zero) and the resultant MEP amplitudes were normalized to peak-to-peak M~MAX~ amplitude. The order of contraction (lengthening and shortening) and intensity (80, 50, 25, 15% MVC) were randomized and counterbalanced. MEP\'s were examined at a range of contraction intensities to investigate the notion that amplitude changes could be more prominent at task specific intensities. Furthermore, the order of TMS and PNS was also randomized. Peripheral electrical stimulation --------------------------------- Electrical stimulation (pulse width, 1 ms) was administrated below the head of the fibula, over the peroneal nerve using a 40 mm diameter cathode/anode arrangement (Digitimer DS7AH, Welwyn Garden City, Hertfordshire, UK). We have previous shown both H-reflex and V-wave can be reliably performed in the TA (Tallent et al., [@B45]). A 10--15% isometric MVC was used to stabilize the H-reflex for resting conditions. After the optimal position for the stimulator was established, it was secured in place. The position was marked with semi-permanent ink to ensure consistency in placement across the trials. The site was checked at the beginning of each assessment session to ensure it was still the optimal site for stimulation. Up to 64 pulses from the first appearance of the H-reflex to M~MAX~ were delivered; the H-reflex at rest was defined as the average of the three highest responses. At least 25 s was left between pulses to ensure the H-reflex has return to pre resting values (Howatson et al., [@B31]). After establishing the highest H-reflex, participants performed 12 shortening and 12 lengthening contractions at 25% of MVC, with 60 s between contractions. Stimulator output was adjusted to elicit an M-wave amplitude of 15--25% of M~MAX~. H-reflex was expressed relative to the contraction type and intensity. In the same manner as MEP\'s, H-reflex was also expressed relative to background EMG. Participants were passively moved in to position 10 s before performing a submaximal contraction and performed a 10--15% MVC to prevent any thixotropic effect. V-waves were examined under four shortening and lengthening contractions. A supramaximal stimulus intensity (150% of M~MAX~) was used to induce V-wave during the maximal contraction (Aagaard et al., [@B2]). Surface electromyography (EMG) ------------------------------ EMG sites were shaved, abraded with preparation gel and then wiped clean with an alcohol swab. Surface EMG was recorded over the TA using pairs of electrodes (20 mm diameter, model; Kendall, Tyco Healthcare Group, Mansfield, MA, USA) spaced 2 cm apart. For the TA, electrodes were placed at one-third distance of the line between the tip of the fibula and the tip of the medial malleolus (Hermens et al., [@B25]). The reference electrode was placed over the medial malleolus. EMG was amplified (× 1000), band pass filtered 10--1000 Hz (D360, Digitimer, Hertfordshire, UK) and sampled at 5000 Hz (CED Power 1401, Cambridge Electronics Design, Cambridge, UK). Despite the well-documented potential limitations of EMG (Farina et al., [@B20]), particularly during dynamic contractions where skin movement across the muscle is increased, our laboratory has previously demonstrated that this method is repeatable (Tallent et al., [@B45]), which is probably attributable to the fact the joint angle is identical between conditions when the stimulation is delivered and therefore delimits the potential for this to be a confounding factor. EMG processing -------------- All TMS responses were averaged from eight responses and all contractions were separated by 30 s. The time from the stimulation artifact to the return of pre-stimulus EMG (within 1 SD) was determined as the silent period length. MEP\'s were also expressed relative to background EMG activity. EMG was rectified with the mean muscle activity 25 ms following the stimulator artifact and the average amplitude was recorded. Ultrasound ---------- To detect any change in muscle thickness, a real-time digital ultrasound imager (Technos MP, Esaote, Genoa, Italy) in B-mode was used to collect sonographic images of the resting TA at the beginning of each experimental test session. A 40 mm linear-array transducer (CA621, Esaote, Genoa, Italy) with a variable center frequency (5--13 MHz) was placed over the longitudinal axis of the TA at a standardized position for all participants. The optimal position was 20% of the distance between the head of the fibula to the lateral tip of the lateral malleolus (Martinson and Stokes, [@B38]). The transducer head was placed perpendicular to the skin along the palpable edge of the tibia and adjusted obliquely to optimize visualization of echogenic landmarks, specifically the deep edge of the tibia, muscle fascicles and the central aponeurosis. Water-soluble hypoallergenic ultrasonic transmission gel (Aquasonic 100, Parker Laboratories Inc., Fairfield, New Jersey) was applied to the head of the transducer probe prior to skin placement. All images were taken unilaterally on the dominant limb with the participant lying supine with their ankle held in a neutral position. Images were captured in triplicate and exported for later analysis offline using publicly available software (Image J, US National Institutes of Health, available at <http://rsb.info.nih.gov/ij/>). Linear muscle thickness (LMTh), previously shown to reflect cross-sectional area (Martinson and Stokes, [@B38]), was determined as the distance between the inferior boundaries of the echogenic muscle fascia. Statistical analyses -------------------- All data were screened for normal distribution. Not normally distributed data was log transformed. To ensure resistance training was conducted at the same relative contraction intensity between groups, a time (training session 1--12) × group (SHO, LEN, CON) repeated measure ANOVA was performed. To confirm TMS and PNS variables were assessed under the same relative torque conditions between groups (SHO, LEN, CON) and across time (PRE, MID, POST) a repeated measured ANOVA was performed. Percentage change in MVC and the corticospinal silent period, H-reflex and V-wave were examined by a repeated measures ANOVA (time \[PRE, MID, POST\] × group \[SHO, LEN, CON\] × contraction type \[shortening, lengthening\]). Finally, differences in MEP amplitude were assessed using repeated measures ANOVA: time × group × contraction type × contraction intensity (15, 25, 50, 80% of MVC). Additionally, 95% confidence intervals (CI) were determined to assess the magnitude of change. Statistical analyses were performed using SPSS (Chicago, Illinois, USA). To detect changes in dependent measures from the training and detraining periods, repeated measures ANOVA\'s were performed. Where necessary, bonferroni *post-hoc* tests were performed for pairwise comparisons. Significance was accepted as ≤ 0.05. Results {#s3} ======= Training and assessment ----------------------- All data is presented as mean ± *SD*. For the raw data relative to M~MAX~ please see Supplementary Tables [1](#SM1){ref-type="supplementary-material"}--[3](#SM1){ref-type="supplementary-material"}. There was no difference (*P* \> 0.05) in relative training intensity between the SHO and LEN training groups, for the 12 resistance training sessions. The repeated measured ANOVA showed no differences (*P* \> 0.05) in relative torque during the assessment sessions across time between groups and relative contraction intensity. Similarly, there was no significant difference in torques during PNS (*P* \> 0.05) assessment sessions. There were no changes in muscle thickness across the training period (*P* = 0.76). Maximal voluntary contraction ----------------------------- The percentage increase in MVC pre to post training (Figure [1](#F1){ref-type="fig"}) showed a time effect \[*F*~(2,\ 56)~ = 25.5; *P* \< 0.001\]. In the LEN group, *post-hoc* analysis showed MVC increased (*P* \< 0.05) from pre to mid training for lengthening and shortening contractions. The SHO group showed an increase in MVC from pre to mid in shortening muscle contractions (*P* = 0.04; 95% CI 6.77--18.8%). {#F1} *Post-hoc* analysis also showed MVC increased from pre to post training for shortening and lengthening MVC in both the SHO (Shortening MVC increased 33.2--40.8 Nm; 95% CI 13.1--18.4%; *P* \< 0.001: Lengthening MVC increased (51.8--56.4 Nm; 95% CI 2.60--15.8%; *P* = 0.006) and LEN (Shortening increase MVC; 33.8--38.5 Nm; 95% CI 3.30--14.6%; *P* = 0.009: Lengthening increased MVC; 49.5--58.6 Nm; 95% CI 13.7--24.6%; *P* \< 0.001) training groups with no change in the CON group (*P* \> 0.05). A significant time-by-group-by-contraction type interaction was also seen \[*F*~(4,\ 56)~ = 6.7; *P* \< 0.001\] for MVC; whereby the SHO group showed an increase in shortening MVC across time when compared to the lengthening MVC (*P* \> 0.001). Similarly, the LEN group showed a greater increase in lengthening when compared to shortening MVC post training (19 vs. 9%; *P* = 0.02; CI = 3.1--17.4%). There was no significant difference in the CON group between the contraction types or across time. Corticospinal variables ----------------------- No changes in resting MEP amplitude (Pre--Post: LEN = 1.3%, SHO = −9.4%, CON = −8.5%) or rMT were found across time (Pre--Post: LEN = 4.4%, SHO = 3.4%, CON = 0.7%). However, there was a main effect of time for MEP amplitude during an active muscle contraction \[*F*~(2,\ 27)~ = 4.7; *P* = 0.01\] when expressed relative to M~MAX~ (Figure [2](#F2){ref-type="fig"}). LEN group showed an increase in MEPs when relative to M~MAX~ during lengthening contractions. Differences were found at, 50% (*P* = 0.001; 95% CI 10.6--45.6%) and 80% (*P* \< 0.001; 95% CI 10.3--46.2%) lengthening contraction intensity (Figure [2](#F2){ref-type="fig"}). There were no difference between contraction type and groups (*P* \> 0.05). A representative trace of MEPs pre-post training can be seen in (Figure [3](#F3){ref-type="fig"}). {#F2} {#F3} MEPs expressed relative to background EMG (Figure [4](#F4){ref-type="fig"}) revealed a main effect across time \[*F*~(2,\ 27)~ = 18.4; *P* \< 0.001\]. Significant increases were only seen in the LEN (Figure [4](#F4){ref-type="fig"}). For MEP\'s relative to background EMG and M~MAX~ significant increases were seen across multiple contraction intensities and both contraction types in the LEN group with no (*P* \> 0.05) changes in the control group. The SHO group showed an increase at intensities 50 and 80% of shortening and lengthening MVC. There was also a group × time interaction \[*F*~(4,\ 56)~ = 4.1; *P* = 0.006\] demonstrating both experimental groups increased compared to the control and baseline. However there was no significant difference between groups (*P* \> 0.05). No differences were found between the contraction type and groups (*P* \> 0.05). Additionally, there was no significant change in the corticospinal silent period across time (*P* \> 0.05). {#F4} PNS --- There was no change (*P* \> 0.05) in M~MAX~ across the study. H-reflex showed no difference (*P* \> 0.05) across time or between groups when expressed relative to M~MAX~ and background EMG. ### V-wave pre to post An increase in V-wave amplitude (Figure [5](#F5){ref-type="fig"}) across time was observed \[*F*~(2,\ 52)~ = 6.0; *P* = 0.004\]. Pairwise comparisons showed the LEN groups V-wave significantly increased during lengthening MVC\'s (0.29 ± 0.12--0.46 ± 0.15 V-wave/M~MAX~; *P* \< 0.001; CI = 44.5--88.9%) and during shortening MVC\'s 0.45 ± 0.17--0.53 ± 0.20 V-wave/M~MAX~; *P* = 0.02; CI = 7.3--45.3%). The SHO group, there was only an increase in V-wave amplitude during shortening MVC\'s 0.43 ± 0.18--0.53 ± 0.19 V-wave/M~MAX~; *P* = 0.03; CI = 7.3--45.3%). No change (*P* \> 0.05) was found in the CON group. Additionally, there was a group-by-contraction interaction \[*F*~(2,\ 26)~ = 8.0; *P* = 0.002\]. LEN group V-wave showed a greater increase compared to the shortening groups V-wave during shortening contractions (*P* = 0.003; CI = 29.3--89.4%; Figure [5](#F5){ref-type="fig"}). {#F5} ### Detraining #### Maximal voluntary contraction There were no differences (*P* \> 0.05) in SHO and LEN MVC (Figure [6](#F6){ref-type="fig"}) following 2 weeks of detraining in all groups. {#F6} Corticospinal ------------- rMT, resting MEPs, and MEPs during an active muscle contraction did not change following 2 weeks detraining (*P* \> 0.05). However, relative to background EMG there was a significant decrease \[*F*~(1,\ 27)~ = 10.2; *P* \< 0.01\] in MEPs (Figure [6](#F6){ref-type="fig"}). There was a significant interaction \[*F*~(5,\ 83)~ = 3.1; *P* = 0.01\], the SHO group showed a significant decrease during 50% (−44%; *P* \< 0.001; CI = 19.4--59.1%), 80% (−47%; *P* = 0.001; CI = 18.3--61.8%) shortening MVC and 15% lengthening MVC (−31%; *P* = 0.01; CI = 7.7--45.8%). There was no change in the corticospinal silent period. Peripheral neuromuscular stimulation ------------------------------------ H-reflex or V-wave amplitude (*P* \> 0.05) did not significantly change after the detraining period. Discussion {#s4} ========== Resistance training ------------------- The main findings from the first part of this study were: (1) shortening resistance training improved shortening MVC more than lengthening MVC, and lengthening resistance training improved lengthening MVC more than shortening MVC, respectively; (2) corticospinal excitability increase for 50 and 80% lengthening contraction in the LEN group, however for a relative motorneuron pool (background EMG), corticospinal excitability increased for both contraction types in both groups, and (3) V-wave increased when tested during shortening and lengthening muscle contractions in the LEN group, but only increased during shortening muscle contractions in the SHO group, showing evidence of task specificity. Additionally, for the first time we have demonstrated lengthening resistance training increases V-wave amplitude during the lengthening test contraction to a greater extent than shortening resistance training and increased V-wave amplitude during shortening test contraction. In line with previous research, a recent meta analysis showed (Roig et al., [@B43]), task specific improvements in MVC. More specifically, lengthening training increased lengthening strength more than shortening training and vice-versa. Notwithstanding, the current study increases our understanding of the neurological adaptations arising from resistance training at a corticospinal and spinal level following 4 weeks of resistance training. These data support previous work demonstrating little change in corticospinal excitability at rest (Carroll et al., [@B7], [@B6]); however, during an active lengthening contraction in the LEN group, there was an increase in corticospinal excitability when expressed relative to M~MAX~. The reason for this observation is probably linked to the differences in corticospinal and spinal balance of excitatory and inhibitory circuits during lengthening muscle contractions. Hortobágyi et al. ([@B29]) demonstrated a seven times greater increase in lengthening background EMG from lengthening compared to shortening resistance training. The increases in background EMG were attributed to a greater recruitment of type II fibers during lengthening contractions (Hortobágyi et al., [@B29]), though this is unlikely to explain amplitude changes in MEP\'s. The superiority of lengthening resistance training has previous been thought to be due to the higher absolute loads that are completed during training (Roig et al., [@B43]); although it has also been shown that when lengthening training was matched to the same absolute loads as shortening resistance training there were no differences in contractions specific MVC. Consequently, it appears that lengthening contractions are a more powerful stimulus to modulate changes in cortiospinal excitability. Contrary to previous results (Carroll et al., [@B7]), these data demonstrate that when MEPs were expressed relative to background EMG, there was a significant increase in MEP peak-to-peak amplitude from pre to post resistance training. Expressing MEP\'s relative to background EMG, represents the excitability of corticospinal cells for a particular level of muscle activity (Carroll et al., [@B7]). Representing MEP\'s to background EMG takes into consideration modifications in muscle activity through changes in strength. Carroll et al. ([@B7]) suggested that fewer motorneurons were activated relative to background EMG due to a change in firing rates and/or intrinsic properties of motorneurons. We propose that for a similar or lower level of background EMG, the cortiospinal pathway is more excitable and consequently the threshold of corticospinal neurons is lower for a given force. The type of resistance training protocols used in each study could explain differences in the findings. For example, performing precision tasks that require an element of complexity or skill can increase corticospinal excitability (Perez et al., [@B40]). Carroll et al. ([@B7]) participants\' were required to isometrically resist an external force, the skill element of this arguably, is considerably less. Even though our results show corticospinal excitability increased following strength training, complexity of the movement appears to play a significant role with changes in corticospinal excitability. Net output from motoneurons projecting to the trained muscle has been shown to increase following a single resistance training session (Nuzzo et al., [@B39]). Previous data from our laboratory (Tallent et al., [@B45]) has also shown MEP\'s to increase following a single isokinetic training session. This further indicates that corticospinal excitability may have only a limited role in strength training, and is more associated with the acquiring the motor patterns of movement. Further research should focus on a greater number of assessment time points in the first 4 weeks of resistance training to increase our understanding of the role of corticospinal excitability in skill and resistance training. Maximal lengthening resistance training has been suggested to cause a decrease in inhibition associated with performing lengthening muscle contractions (Aagaard, [@B1]). This study found no evidence of modifications of the silent period following the training period in any group, suggesting corticospinal inhibition, as measured here, was not responsible for modulating the MEP peak-to-peak amplitude and strength improvement. Previous resistance training studies have shown both a decrease (Kidgell and Pearce, [@B34]; Latella et al., [@B36]) and no change in corticospinal inhibition (Kidgell et al., [@B35]). Recent work using paired-pulse TMS has shown lower short-interval intracortical inhibition to occur after a period of resistance training (Weier et al., [@B48]). Future research should include paired pulse TMS paradigms to further understand any modulation in inhibitory and facilitation networks that influence corticospinal excitability following resistance training. The increases in V-waves found in this study suggest that an increase in efferent output supports the increase in strength early in a resistance training programme. The lack of changes in inhibition suggest that the greater efferent cortical output is needed to overcome the cortiospinal inhibition and increase the motor unit activation. Changes in V-wave amplitude seem to reveal a consistent pattern across different studies and training paradigms (Duclay et al., [@B14]; Ekblom, [@B16]; Vila-Chã et al., [@B47]) with increases of up to 80% (Gondin et al., [@B21]). Our findings further support this by showing increased V-wave in both training groups; however, for the first time we have shown a greater response to lengthening resistance training. A meta-analysis (Roig et al., [@B43]) showed greater gains in total strength (lengthening and shortening MVC combined), which may offer some explanation that an enhancement in efferent output during MVC, at least in part, might be responsible for the greater increases in strength associated with lengthening resistance training. Even though lengthening contractions are suggested to have a greater supraspinal output (Fang et al., [@B18], [@B19]) and greater inhibition at a spinal level (Duclay and Martin, [@B13]; Duclay et al., [@B14]), we hypothesized that the greater CNS adaptations are due to a further increase in supraspinal output. However, tension regulating mechanisms (such as the golgi tendon organ) cannot be excluded as adaptations for supporting the increase in strength. However, it should be noted that the starting value of lengthening V-wave in the LEN group was visually lower compared to the other groups, reasons for this are unclear. Previous research suggests a lack of change in resting H-reflex (Aagaard et al., [@B2]; Holtermann et al., [@B27]; Duclay et al., [@B14]; Ekblom, [@B16]), however, during an active contraction, there are reports of increased spinal excitability (Holtermann et al., [@B27]; Duclay et al., [@B14]). This is likely attributable to a reduction in presynaptic inhibition (Aagaard et al., [@B2]). Given our data found no changes in H-reflex, our laboratory supports previous work suggesting changes in strength have tenuous links with the Ia afferent loop, We examined H-reflex during a relatively low intensity contraction (\~25% MVC) in an attempt to detect changes at a spinal level, but did not observe any change. It is possible that any spinal adaptation may occur at higher contraction intensities that are specific to training intensity and we acknowledge this as a limitation. Detraining ---------- The second part of this study revealed that maximal strength and V-wave amplitude were both retained following a 2 week detraining period independent of the contraction type used during the training period and corticospinal excitability decreased following 2 weeks detraining. Studies have shown an increase in strength above baseline from weeks to months following cessation of resistance training in healthy and special population such as the elderly (Carvalho et al., [@B8]; Popadic Gacesa et al., [@B41]; Correa et al., [@B11]). A recent meta-analysis concluded that following the cessation of training, a decrease in maximal force was evident in the third week of inactivity (Bosquet et al., [@B4]). However, given the relatively short duration of our training programme and the notion that longer resistance training programmes lead to longer-lasting adaptations (Bosquet et al., [@B4]), it is perhaps surprising that in the current study, detraining did not cause a reduction in maximal voluntary force. With so many diverse resistance training paradigms and different periods no activity, little definite conclusions can be made from the literature. Despite this, our results found evidence of neurological changes following 2 weeks post the cessation of resistance training. Both groups showed a decrease from post resistance training MEP amplitude when expressed relative to background EMG. There are few studies that have investigated detraining at a cortical or spinal level. Although statistical analyses (*n* = 4) were not performed, Jensen et al. (*2005*) appeared to show a reduction in the MEP recruitment curve toward baseline values following cessation of resistance training in the biceps brachii muscle. Whether this is due to a withdrawal of the resistance training stimulus or a reduction in strength is unclear. Previous research has showed conflicting results following periods of inactivity (with no prior resistance training) in rodents and humans; an increase in corticospinal excitability (Roberts et al., [@B42]), reduced cortical representation/excitability (Leukel et al., [@B37]) or no change has been reported (Zanette et al., [@B50]). The lack of research in this area leaves definitive conclusions difficult. Others have suggested (Clark et al., [@B9]) that neurological factors contribute to \~48% of the loss in strength from inactivity. Our results add to this body of evidence and show that corticospinal excitability rapidly returns toward baseline following the cessation of strength training. Our results add to Jensen et al.\'s ([@B32]) work by demonstrating that the reduction in corticospinal excitability may largely be due to the cessation of the resistance training. Long periods of inactivity have been shown to decrease spinal excitability (Yamanaka et al., [@B49]). Given there was no change in spinal excitability from the resistance-training programme it is unsurprising there was no change following 2 weeks of detraining. Unlike voluntary activation (Gondin et al., [@B21]), which seems to decrease following a period of inactivity, V-wave was maintained following 2 weeks detraining. However, given that there was only 2 weeks detraining it is difficult, to conclude that V-wave is a longer lasting neurological adaptation; nonetheless it does not see reductions in this epoch. Furthermore, these data cannot rule out the notion that the maintenance of MVC may in part be attributable to a preservation of neural drive, and supports the hypothesis that volitional drive is strongly associated with strength loss. Despite this study being the first study to investigate site specific adaptations from shortening and lengthening resistance training, the muscle used during the resistance training could be considered a limitation given its applicability to other muscles that might be targetted during resistance training programs. Further research should focus on changes in corticospinal and spinal excitability the agonist antagonist relationship. In addition, an extended period of detraining to observe strength loss would allow for more definite conclusions on the time course and underlying mechanisms behind detraining-related losses in strength. In conclusion, this study showed task-specific gains in strength that were greater following lengthening training. The increase in maximal muscle force generating capacity was accompanied with changes in corticopsinal excitability and volitional drive. These V-wave changes were also contraction specific and were greatest from lengthening resistance training, suggesting that efferent neural output at a cortical level increases significantly more from lengthening resistance training. Even though MVC was maintained following 2 weeks of detraining, there was a significant reduction in corticospinal excitability (relative to background EMG). As a decrease in corticospinal excitability occurred from lengthening and shortening resistance training, this study showed little evidence of the notion that lengthening action causes longer lasting neurological adaptations following short term resistance training. Author contributions {#s5} ==================== JT, SG, KG, TH, and GH conceived and planned the study. JT, SG, KG, and GH performed the study. JT, SG, KG, TH, and GH analyzed the data and JT, SG, and GH ran statistical analysis. JT wrote the first draft with contributes from SG, KG, TH, and GH. All authors approved the final draft. 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. Supplementary material {#s6} ====================== The Supplementary Material for this article can be found online at: <http://journal.frontiersin.org/article/10.3389/fphys.2017.00057/full#supplementary-material> ###### Click here for additional data file. [^1]: Edited by: Olivier Girard, Qatar Orthopaedic and Sports Medicine Hospital, Qatar [^2]: Reviewed by: Sidney Grosprêtre, University of Franche-Comté, France; Carolina Vila-Chã, Polytechnic Institute of Guarda, Portugal [^3]: This article was submitted to Exercise Physiology, a section of the journal Frontiers in Physiology

{#sp1 .189} {#sp2 .190} {#sp3 .191}

Introduction {#section1-1533317519845725} ============ Care homes provide care and support to up to 38% of people with dementia,^[@bibr1-1533317519845725],[@bibr2-1533317519845725]^ with the majority of people living in care homes having dementia.^[@bibr2-1533317519845725],[@bibr3-1533317519845725]^ Despite dementia being their core business, the quality of care home care for people with dementia is variable.^[@bibr4-1533317519845725],[@bibr5-1533317519845725]^ Poor quality care for people with dementia is associated with an increase in behaviors such as agitation, apathy, and aggression.^[@bibr6-1533317519845725],[@bibr7-1533317519845725]^ Although the need for psychosocial approaches to support good quality care is recognized,^[@bibr8-1533317519845725]^ there are limited evidence-based interventions to support this^[@bibr9-1533317519845725],[@bibr10-1533317519845725]^ and challenges in the widespread implementation of such interventions into everyday practice.^[@bibr10-1533317519845725],[@bibr11-1533317519845725]^ Dementia Care Mapping™ (DCM)^[@bibr12-1533317519845725],[@bibr13-1533317519845725]^ is a manualized, established intervention^[@bibr14-1533317519845725]^ developed by the University of Bradford, United Kingdom, and used internationally in care home settings.^[@bibr15-1533317519845725]^ It is an observational and practice development tool, implemented as quality improvement cycles, which aim to support the delivery of person-centered dementia care.^[@bibr16-1533317519845725]^ Standard implementation is led by care home staff who attend a 4-day training program in use of DCM. The process includes 5 components: briefing staff about DCM, care practice observation using standardized coding frames, data analysis and summary report production, feedback of findings to the staff team, and action planning for practice development at individual resident and care home levels.^[@bibr17-1533317519845725]^ Cycles are repeated every 4 to 6 months as part of a care home's ongoing quality improvement.^[@bibr16-1533317519845725]^ Reported benefits of DCM implementation in care homes include reduced resident agitation, depression, anxiety, and neuropsychiatric symptoms and improved quality of life.^[@bibr18-1533317519845725][@bibr19-1533317519845725][@bibr20-1533317519845725]--[@bibr21-1533317519845725]^ Some DCM trials, however, have not found positive outcomes for residents compared to usual care control.^[@bibr22-1533317519845725],[@bibr23-1533317519845725]^ DCM implementation has also been reported to improve the quality of staff--resident care interactions,^[@bibr19-1533317519845725]^ reduce staff burnout,^[@bibr24-1533317519845725]^ and improve staffs' feelings about^[@bibr23-1533317519845725]^ and their connection with residents.^[@bibr18-1533317519845725]^ DCM trials have faced challenges with implementation at the system level,^[@bibr22-1533317519845725],[@bibr23-1533317519845725]^ a challenge also highlighted in a recent systematic review of DCM implementation.^[@bibr25-1533317519845725]^ Intervention implementation and fidelity (the extent to which core components are delivered as intended in the research protocol) is important to investigate, to support interpretation of trial outcomes.^[@bibr26-1533317519845725]^ Few DCM studies have been conducted as randomized controlled trials, with only 2 reporting full implementation procedures and process evaluation results.^[@bibr27-1533317519845725],[@bibr28-1533317519845725]^ Thus, relatively little is known about the particular DCM implementation strategies that have proved effective.^[@bibr29-1533317519845725]^ A German study reported largely good adherence to delivery of the requisite number of DCM cycles and cycle components.^[@bibr27-1533317519845725]^ However, wider staff engagement in feedback sessions was low in 2 of the 6 intervention homes, and staff were critical of the quality of DCM delivery. In a Dutch study, involving 13 care units across 5 nursing home sites, DCM intervention adherence was variable.^[@bibr28-1533317519845725]^ Two care units undertook no cycles of DCM, and completion of all DCM cycles across the other care units was variable. The limited available evidence and varied DCM implementation conditions and adherence in trials to date make interpretation of results challenging. More widely, it raises questions about whether randomized controlled trials that have found psychosocial interventions for dementia to be ineffective and not cost-effective might be explained by poor implementation adherence.^[@bibr30-1533317519845725]^ The DCM EPIC cluster randomized controlled trial aimed to assess the effectiveness and cost-effectiveness of DCM in care home settings, including a full process evaluation to understand implementation processes and issues. This article reports on the DCM intervention delivery. It aimed to answer the following questions:How was DCM delivered, compared to protocol?What DCM components were delivered (fidelity, dose, adaptions and reach) compared to protocol? Methods {#section2-1533317519845725} ======= Design and Setting {#section3-1533317519845725} ------------------ The full trial details are reported in the protocol.^[@bibr31-1533317519845725],[@bibr32-1533317519845725]^ In summary, the DCM EPIC trial was a pragmatic, multicenter, cluster randomized controlled trial of DCM plus usual care (intervention) versus usual care alone (control). Sites were residential and nursing homes that provided care for people with dementia in 3 regions of the United Kingdom. Fifty care homes were recruited with 31 randomized to intervention and 19 to control. Data were collected by blinded researchers at baseline (prerandomization) and 6- and 16-month follow-up. The Medical Research Council guidance on process evaluations was followed.^[@bibr33-1533317519845725]^ This included assessing the implementation process in terms of how delivery was achieved and what was delivered, as measured by fidelity (delivery as intended), dose (quantity of delivery), adaptions (changes made by individual sites), and reach (whether the intended audience received the intervention). Intervention {#section4-1533317519845725} ------------ DCM implementation was described in the study protocol following standard procedures reported in the DCM manual and guidance.^[@bibr13-1533317519845725],[@bibr34-1533317519845725]^ Two staff members per home were selected by the home manager to train to use DCM (called mappers), using a set of "mapper qualities criteria." Mappers completed a standard 4-day DCM Basic User course that consists of information on person-centered dementia care, use of the 4 DCM coding frames and their application rules, and instruction on completing the DCM cycle components including staff briefing, data analysis, report writing, staff feedback, and action planning. They were then requested to implement 3 DCM cycles, comprised of briefing, observation, data analysis and reporting, feedback, and action planning. Per cycle, mappers were asked to deliver the following standard DCM components: at least 1 formal briefing and 1 formal feedback session for staff (and additional informal sessions as required); observations on up to 5 residents per mapper; use all 4 of the DCM observational coding frames and make qualitative notes; observe for up to 6 hours over 1 or more days during a week; write reports summarizing data for the care home and each resident mapped including specific feedback points; and produce an action plan with at least 1 action point for each resident mapped and the whole home. "Mapper instruction packs" were provided in hard and electronic formats containing fidelity guidelines and standardized templates for data processing, action planning, and reporting. While DCM EPIC was a pragmatic trial, aiming to investigate DCM's effectiveness when implemented in a manner reflective of standard UK DCM use, a number of additional mechanisms were introduced to support consistent implementation, which could be feasibly introduced in usual practice. This included support for the first DCM cycle by an expert mapper from the research team. The expert mapper provided 2 days of desk-based support for preparation, data analysis, and report writing and spent 3 days in the care home to support briefing, mapping, feedback, and action planning. Other support mechanisms included telephone and e-mail support from a member of the trial team, sending SMS reminders, and a "mapping pack" of paperwork by post to mappers ahead of each cycle. Participants and Data Collection {#section5-1533317519845725} -------------------------------- All mappers were asked to return data on DCM component adherence and fidelity at the end of each mapping cycle. This included data collection forms to record the number and dates of briefing and feedback sessions and the number and role of staff members attending and deidentified copies of their DCM observation sheets, feedback report, and action plans. Expert mappers reported after cycle 1 on component completion, mapper skills, and any concerns regarding sustainability of DCM implementation in the home. One author (CS) and the trial manager completed a standard case review form for each home at each cycle that summarized implementation data. Data Analysis {#section6-1533317519845725} ------------- Data were summarized using descriptive statistics in SAS software v9.4 or Stata v14.^[@bibr35-1533317519845725],[@bibr36-1533317519845725]^ Ethical Issues {#section7-1533317519845725} -------------- All study participants gave formal written consent to participate. Ethical approval for the study was granted by NRES Committee Yorkshire and The Humber---Bradford Leeds (REC ref 13/YH/0016). The trial was registered with the International Standard Randomized Controlled Trial Register (ISRCTN) reference ISRCTN82288852. Results {#section8-1533317519845725} ======= The main trial outcomes are published elsewhere.^[@bibr37-1533317519845725]^ Data Return {#section9-1533317519845725} ----------- There was variable compliance in return of DCM implementation documentation, despite a range of approaches by the trial team to increase return rates. These included sending multiple phone and e-mail remainders and unblinded researchers visiting some homes to collect documentation. Where documentation was returned, there were missing data on some intervention components. Where documentation was provided for a later DCM phase only (eg, mapping data or feedback report), we assumed that undocumented earlier phases (eg, briefing session) had occurred. In some cases, mappers verbally reported that a DCM cycle or components of it had been completed but did not provide any supporting documentary evidence. These were recorded as incomplete for trial purposes. Implementation of the DCM intervention was of interest owing to the pragmatic nature of the trial design. Due to satisfactory completion of DCM cycle 1, the difficulties in obtaining documentary evidence of DCM cycles 2 and 3 completion, and the incremental (phased) recruitment of care homes over 14-month period with a 16-month intervention period, early termination of the trial was not considered. Implementation Process: Mapper Training and Retention {#section10-1533317519845725} ----------------------------------------------------- Mapper training was delivered per protocol (within 2 months of randomization) in 21 (67.7%) of 31 homes. In 1 (3.2%) home, no mappers were trained. In 2 (6.5%) homes, only 1 mapper was trained. Retention of mappers was problematic. One or both mappers withdrew in 17 (54.8%) homes, with reasons including resignation, long-term sickness, maternity leave, and lack of management support to undertake mapping. At 16-month follow-up, 10 (32.3%) homes had no mappers in the role, 7 (22.6%) had 1 mapper, and 14 (45.2%) homes retained both mappers. Although there was funding to train additional mappers (eg, due to mapper resignation or sick leave), this was accessed in only 1 home. Reasons for not training additional mappers included insufficient time or a new mapper being unable to attend training before the trial's end and being unable to identify a suitable replacement mapper. What Was Delivered? Mapping Cycles {#section11-1533317519845725} ---------------------------------- Implementation of the 3 DCM cycles across the 31 intervention homes was variable. Adherence is reported by cycle and highest level completed component in [Figure 1](#fig1-1533317519845725){ref-type="fig"}. There was low adherence beyond the first supported cycle, with 16 (51.6%) homes only completing 1 cycle. Seven (22.6%) homes did not complete a full cycle, with 3 (9.7%) not completing any components, 4 (12.9%) homes completed 2 full cycles, and 4 (12.9%) completed 3 full cycles. Thus, the dose and fidelity of mapping cycles across the intervention homes was inconsistent, with only 4 (12.9%) homes achieving the per-protocol dose. The following sections examine fidelity, dose, adaptions, and reach per DCM component. {#fig1-1533317519845725} Briefing {#section12-1533317519845725} -------- There was a substantial amount of missing data on briefing session completion (see [Table 1](#table1-1533317519845725){ref-type="table"}). Where a briefing session was reported, the median number of staff receiving formal briefing increased per cycle from 10 in the first cycle to 20 in the third cycle. The majority of mappers also briefed staff informally. However, the range of numbers of staff briefed formally (3 to 28) and informally (2 to 31) in each home was wide across the 3 cycles. Therefore, in some homes, very few staff may have received DCM briefing. This indicates variable fidelity, dose, and reach of DCM briefing across the care homes. ###### Summary of Briefing Session Fidelity in Homes Where Component Completed.  Summary of Briefing Sessions by Cycle ----------------------------------------- ----------------- ----------------- ------------------ Number of formal briefing sessions held  1 9 (32.1%) 7 (58.3%) 3 (50.0%)  2 4 (14.3%) 1 (8.3%) 1 (16.7%)  3 2 (7.1%) 1 (8.3%) 1 (16.7%)  Missing 13 (46.4%) 3 (25.0%) 1 (16.7%) Total number of staff attended  Mean (SD); missing 10.1 (4.52); 15 15.8 (7.44); 4 18.0 (8.19); 3  Median (range) 10 (3-20) 14.5 (8-28) 20 (9-25) Informal briefing sessions held  Yes 15 (53.6%) 10 (83.3%) 3 (50.0%)  No 1 (3.6%) 1 (8.3%) 2 (33.3%)  Missing 12 (42.9%) 1 (8.3%) 1 (16.7%) Number of staff informally briefed  Mean (SD); missing 10.5 (7.51); 14 13.1 (10.89); 4 19.3 (1.15); 3  Median (range) 8.5 (2.0-30.0) 7.0 (4.0-31.0) 20.0 (18.0-20.0) Abbreviation: SD, standard deviation. Mapping Observations {#section13-1533317519845725} -------------------- Observation was conducted by 2 mappers for most cycles (see [Table 2](#table2-1533317519845725){ref-type="table"}). The mean number of hours (7.8-9.4) and the median total number of residents observed per home (5-6) were reasonably consistent per cycle, although the range was not (4-12.4 hours and 2-10 residents), indicating variable observation dose and reach per cycle. Quality of DCM data coding improved over the cycles, based on rating of mappers' use of all 4 DCM coding frames and making accompanying qualitative notes. While the percentage of mappers consistently achieving this remained similar at \<50% per cycle, showing only moderate fidelity with the manualized DCM method, the proportion not meeting this criteria at all declined considerably over 3 cycles (from 35.7% to 16.7%). ###### Summary of Mapping Observation Fidelity in Homes Where Component Completed.  Observation Adherence by Cycle ---------------------------------------------------------------------- ---------------- ---------------- --------------- Number of mappers conducting observations  1 1 (3.6%) 1 (9.1%) 1 (16.7%)  2 18 (64.3%) 10 (90.9%) 4 (66.7%)  Missing 9 (32.1%) 0 1 (16.7%) Total mapping time, hours  Mean (SD); missing 8.9 (2.76); 13 9.4 (2.30); 3 7.8 (0.43); 3  Median (range) 9.2 (4.0-12.4) 9.9 (6.5-12.3) 8.0 (7.3-8.0) Total residents observed  Mean (SD); missing 5.4 (1.79); 10 5.7 (2.41); 0 5.2 (1.79); 1  Median (range) 5 (2-8) 6 (2-10) 4 (4-8) Used all 4 coding frames and made at least minimal qualitative notes  Yes 9 (32.1%) 5 (45.5%) 2 (33.3%)  Partially 9 (32.1%) 6 (54.5%) 3 (50.0%)  No 10 (35.7%) 0 1 (16.7%) Abbreviation: SD, standard deviation. Feedback {#section14-1533317519845725} -------- There was considerable missing data on delivery of feedback sessions and numbers of staff attending ([Table 3](#table3-1533317519845725){ref-type="table"}). In each cycle, the majority (50%-73%) of homes documented delivery of a formal feedback session. While the median number of staff attending formal feedback increased per cycle, there was considerable variability, particularly during cycle 1 (2-17 staff attending). Likewise, there was sizeable variation across care homes on how many home- and resident-level feedback points were included in the report. This indicates substantial variability in DCM fidelity, dose, and reach of DCM feedback within and across the homes. ###### Summary of Feedback Session Fidelity in Homes Where Component Completed.  Summary of Feedback Sessions by Cycle --------------------------------------------------------- ---------------- --------------- --------------- Number of mappers participating in the feedback process  1 1 (4.2%) 2 (18.2%) 0  2 13 (54.2%) 7 (63.6%) 3 (50.0%)  Missing 10 (41.7%) 2 (18.2%) 3 (50.0%) Formal feedback sessions held N (%) Missing  Yes 12 (50.0%) 8 (72.7%) 3 (50.0%)  No 2 (8.3%) 2 (18.2%) 1 (16.7%)  Missing 10 (41.7%) 1 (9.1%) 2 (33.3%) Total number of formal feedback sessions  Mean (SD); missing 1.8 (0.83); 12 1.4 (0.79); 4 1.0 (0.00); 3  Median (range) 2 (1-3) 1 (1-3) 1 (1-1) Total number of staff attended formal feedback sessions  Mean (SD); missing 9.6 (4.56) 12 12.3 (4.46) 5 12.3 (4.51) 3  Median (range) 9.0 (2-17) 11.5 (7-18) 12.0 (8-17) N of care home feedback points  Mean (SD); missing 5.0 (3.06); 14 3.7 (1.21); 5 6.0 (5.72); 2  Median (range) 4.5 (2-13) 3 (3-6) 5.5 (0-13) Total number of mapped residents with feedback points  Mean (SD); missing 4.4 (1.78); 12 4.2 (2.23); 5 3.5 (1.73); 2  Median (range) 4.5 (1-7) 5 (1-6) 4 (1-5) Mean number of resident feedback points  Mean (SD); missing 3.2 (2.12); 13 2.5 (0.93); 5 2.3 (0.96); 2  Median (range) 2.8 (0.8-7.8) 2.9 (1.0-3.3) 2.4 (1.3-3.3) Abbreviation: SD, standard deviation. Action Planning {#section15-1533317519845725} --------------- Of the homes that provided evidence of action planning, the percentage who produced a care home--level action plan increased per cycle (see [Table 4](#table4-1533317519845725){ref-type="table"}), from just over 50% at cycle 1 to all homes in cycle 3, indicating higher fidelity of care home--level action planning in homes completing multiple DCM cycles. While the average number of care home action points produced per cycle was consistent, the range was wide, demonstrating dose variability across homes. Where homes commenced a DCM cycle but did not complete all components, action planning was the component most likely to be omitted. Resident action plans were received from 42% to 75% of homes per cycle, and 20% to 76% of residents observed had at least 1 action point written to support their care planning at each cycle, indicating poor protocol fidelity and inconsistent reach and dose of DCM action. Homes consistently used the trial's action plan templates, indicating low adaptation of this component where completed. ###### Summary of Action Planning Fidelity in Homes Where Component Completed.  Action Planning by Cycle ------------------------------------------------------------------------- --------------- --------------- --------------- Care home action plan received, n (%)  Yes 13 (54.2%) 6 (75.0%) 4 (100.0%)  No 6 (25.0%) 2 (25.0%) 0  Missing^a^ 5 0 0 Number of care home action points  Mean (SD) 4.9 (3.20) 5.2 (4.83) 5.0 (2.16)  Median (range) 4 (2-14) 3 (3-15) 4.5 (3-8) Standard care home action plan template used  Yes 13 (100.0%) 6 (100.0%) 3 (75.0%) At least one resident action plan received, n (%)  Yes 13 (41.9%) 6 (75.0%) 3 (75.0%)  No 6 (19.4%) 2 (25.0%) 1 (25.0%)  Missing^a^ 5 0 0 Total number of residents with action points  Mean (SD) 5.5 (1.85) 5.8 (2.86) 4.7 (1.15)  Median (range) 5 (3-8) 5.5 (2-10) 4 (4-6) Mean number of action points per resident where plan completed  Mean (SD); missing 2.0 (1.95) 2.0 (1.24) 1.8 (1.77)  Median (range) 1.6 (0.1-7.8) 2.2 (0.1-3.3) 1.3 (0.3-3.8) Standard resident action plan template used where plan completed, n (%)  Yes 12 (92.3%) 6 (75.0%) 2 (75.0%) At least one action point per observed resident where plans completed  Yes 5 (20.1%) 4 (66.7%) 25 1 (33.3%)  No 8 (33.3%) 2 (33.3%) 2 (66.7%) Abbreviation: SD, standard deviation. ^a^Data may be missing in cycle 1 since components completed could be recorded via confirmation through data collection form completed by expert mapper, even if no mapping documentation received from the mappers. Discussion {#section16-1533317519845725} ========== When implementing a multicomponent care improvement intervention such as DCM, it is important to understand whether core components have been implemented as intended^[@bibr26-1533317519845725]^ and to assess any implementation challenges associated with the different components.^[@bibr38-1533317519845725]^ The results of this process evaluation show that across the 31 intervention homes, there was poor consistency in the fidelity, dose, and reach of DCM. This applied to the number of DCM cycles completed and to execution of each DCM intervention component per cycle. Overall, there was low intervention fidelity and dose of DCM compared to trial protocol, and where cycles did occur, implementation quality was often low. These findings mirror those reported in other studies utilizing care home staff-led DCM cycles,^[@bibr22-1533317519845725],[@bibr28-1533317519845725]^ and in other studies of complex interventions in long-term care settings where staff-led intervention implementation has been utilized.^[@bibr39-1533317519845725]^ Reported barriers and facilitators to the trial's DCM implementation, gained through qualitative interviews with mappers, managers, care home staff, and expert mappers, are discussed in full in a separate article.^[@bibr40-1533317519845725]^ However, in summary, barriers and facilitators were evident at 4 levels, the individual mapper level, care home level, intervention level, and trial level. Care home--level barriers included low staffing levels, high staff turnover, lack of time to undertake DCM, and competing priorities. If the care home manager was not fully supportive of DCM, this was an identified barrier to DCM. In addition, external priorities took precedence such as regulatory inspections and any requirements they might place on a home. These contextual factors offer an explanation for why some homes failed to implement any full cycles of DCM and why so few homes did not conduct more than their first supported cycle. Fidelity was weak for some components of DCM observation. While on average the length of observation and numbers of residents observed during each cycle adhered to the trial protocol, this was inconsistent when examining ranges. Of note was the relatively unchanging proportion of mappers who used all 4 DCM coding frames and who made qualitative notes during observations. Failure to consistently record DCM data in line with the manualized method creates doubts about the accuracy of the DCM data informing practice development. Interviews with mappers about the barriers and facilitators to DCM implementation indicated that there were DCM intervention-level barriers that included perceptions of DCM being too complex and time consuming, including the coding frames used during observations.^[@bibr40-1533317519845725]^ Managers in the same study discussed how identifying staff with the requisite academic skills to be able to successfully implement DCM could be challenging within a care home environment. While senior staff might be more experienced and potentially able to use DCM, they were noted to be less likely to be able to be released to undertake DCM training and implementation, particularly in smaller nursing homes and in residential homes where there were fewer/no nurses. The variability seen in availability and numbers of staff attending briefing and feedback sessions indicates low fidelity and reach of these core DCM components. Briefing sessions facilitate engagement of staff in the DCM process, and feedback sessions offer the crucial opportunity for staff to discuss the DCM findings, analyze their meaning and implications, and undertake action planning. These components are fundamental to staff ownership of practice improvement. Low staff engagement with DCM was also reported in a previous process evaluation study.^[@bibr27-1533317519845725]^ When implementing complex interventions such as DCM in care home settings, engagement of the wider staff team^[@bibr41-1533317519845725],[@bibr42-1533317519845725]^ and good communication around implementation^[@bibr43-1533317519845725],[@bibr44-1533317519845725]^ are identified facilitators of adoption in practice. These factors were unlikely in homes where few staff were involved in DCM briefing or feedback. A range of care home--, mapper-, and intervention-level barriers and facilitators to staff engagement with DCM were identified in the process evaluation interview data from this study.^[@bibr40-1533317519845725]^ These included the culture within the care home and whether staff were open to change, whether the care home manager supported the mappers in facilitating staff attendance at briefing and feedback sessions, whether the mappers were respected by and could easily engage their colleagues, and how well mappers were able to explain DCM and its potential benefits to others. This indicates that the readiness of the care home for DCM ahead of implementation as well as choosing individuals with the requisite skills and status within the home to lead DCM are important prerequisites that may serve to support or undermine the likelihood of engaging the full staff team with the DCM process. There was also considerable variability in the execution of DCM action planning across the intervention homes, with reported inconsistency in reach and dose. This included whether care home-- and individual resident-level action plans were produced and how many action points were written per plan. It is possible that action points were identified during DCM feedback sessions but that formal action plans were not produced. However, a lack of formal written records of practice development plans and only small numbers of staff attending formal feedback sessions in some of the care homes mean it is unlikely all staff were exposed to verbally produced action plans. These protocol lapses potentially jeopardized DCM-related practice change and limited opportunities for monitoring development over time. The interview data from this process evaluation indicated that the mappers found the skills required to undertake data analysis, report writing, and the development of action plans consistently challenging.^[@bibr40-1533317519845725]^ They frequently reported not having the IT skills to produce the standardized feedback reports, and due to this, the process took much longer than anticipated. Mappers were also required to use skills they had not previously had to employ, such as engaging colleagues in discussion of practice development issues and accurate written recording of outcomes. This was compounded by low literacy and numeracy skills and use of English as a second or additional language by some mappers. Given action plans require completion of further paperwork, their inconsistent completion may therefore be related to fatigue, competency, and the amount of time required to complete paperwork. The amounts of paperwork required during the DCM process were identified as a major barrier to its use by mappers. In a German process evaluation,^[@bibr27-1533317519845725]^ they concluded that the well-defined, prestructured components of DCM could be easily implemented by care home staff, with biggest barrier to implementation being the translation of action plans into practice change. In contrast to these findings, and in line with those of van der Ven et al,^[@bibr28-1533317519845725]^ our study indicates that most care home staff were unable to deliver the standard, manualized components of DCM with sufficient fidelity or dose. DCM is identified by mappers as a complex tool hampering their ability to use it accurately.^[@bibr25-1533317519845725],[@bibr45-1533317519845725]^ The complexity of an intervention is a potential barrier to implementation identified in other care home trials.^[@bibr44-1533317519845725],[@bibr46-1533317519845725],[@bibr47-1533317519845725]^ The interview data from this process evaluation^[@bibr40-1533317519845725]^ indicated that not only was the complexity of DCM as a tool a barrier to implementation, but mappers often lacked the skills and confidence to lead a process of change within the care home. This was further hampered by setting conditions within the environment which could serve to facilitate or undermine DCM, including the culture, management support for DCM, and staff willingness to engage with practice change. In this study, loss of mappers was a common occurrence, with one or both mappers withdrawing in 54.8% of homes by 16-month follow-up due to a range of staff and contextual issues, including lack of managerial support to map. A further challenge was the limited opportunity to identify and train further mappers with the requisite skills within the trial period. High staff turnover rates within care home settings and the consequent need to facilitate regular staff retraining have been identified as a challenge to implementing and sustaining complex interventions.^[@bibr43-1533317519845725],[@bibr44-1533317519845725],[@bibr48-1533317519845725],[@bibr49-1533317519845725]^ Since the majority of completed DCM cycles were undertaken by 2 mappers, the loss of 1 mapper in a care home is likely to jeopardize continued use of DCM. Thus, DCM implementation is highly vulnerable to staffing-related issues. The process evaluation interview data supported this, with lone mappers indicating that undertaking a DCM cycle alone felt complex, time consuming, and overwhelming.^[@bibr40-1533317519845725]^ Overall, this process evaluation has identified a range of challenges to the accurate and sustained implementation of DCM by care home staff. These findings also have relevance to the use of other complex interventions within care home settings. Challenges with return of data on DCM cycle completion were a major barrier to intervention fidelity monitoring during the trial. The reliance on care home mappers to return these data and the inconsistency between verbal and written reports of cycle completion suggest researchers conducting care home trials need to consider if and how they will collect fidelity data. Mechanisms that do not rely on care home staff to return fidelity data should be considered and may be preferable. While mapper selection criteria were utilized in this trial to assist manager to select appropriate individuals for the role, the fidelity data suggest that this may not have led to selection of the right individuals with the requisite skills to accurately implement and sustain use of DCM. Interviews with managers^[@bibr40-1533317519845725]^ indicated that the pool of suitable staff who met the mapper selection criteria may be limited. Experienced staff with the requisite skill were usually working in senior roles and it was more difficult to release them to attend training and complete mapping. Completion of the standard 4-day DCM Basic User training course did not equip all mappers to use DCM accurately, even with support from an external expert. Replacement of mappers who left their post was therefore challenging, and many mappers lacked the confidence and skills to continue to implement it alone. Given any intervention designed to change care home practice is likely to require an initial high-level of skilled mapper leadership, future use of DCM or other complex interventions should consider if and how care home staff can be equipped to develop such skills and whether alternative models of implementation are required. For example, ongoing support from an external expert mapper who can continue to work alongside care home staff over the long term may be beneficial. A range of care home--level factors such as management support and staff willingness to engage with change also impact DCM fidelity, reach, and dose. Given the pragmatic, explanatory nature of this trial, care homes were selected randomly from 3 geographic regions. This finding indicates that some care homes may not provide the right setting conditions for DCM and potentially for implementation of other complex interventions. Consideration may need to be given within an intervention as to how the care home culture can be made ready for implementation ahead of commencement. Incentives for care homes to undertake intervention implementation is also a consideration. In this trial, care homes were not provided with funding to backfill staff to attend DCM training or to undertake DCM cycles. If this had been provided, more managers may have been supportive of giving mappers enough dedicated time to complete the DCM cycles. While this may have improved the implementation dose and sustainability, it is unlikely to have addressed the accuracy with which DCM was implemented or the challenges managers faced in identifying individuals with the requisite skills to lead DCM. The small number of care homes completing more than their first supported DCM cycle and variability in the fidelity and reach per component per cycle makes it difficult to draw any conclusions about fidelity between cycles. Limitations {#section17-1533317519845725} ----------- The study had a number of limitations. Given the challenges we experienced in the return of implementation data from mappers and the degree of missing implementation data, the recorded compliance data may be subject to inaccuracies of both under- and overreporting. We were also unable to assess the quality of delivery of some components, for example, briefing and feedback sessions, where carried out. Therefore, reported completion of a DCM component is not an indicator of completion quality. Conclusions {#section18-1533317519845725} =========== In this pragmatic trial of DCM using standard care home staff-led cycles, the implementation process, fidelity, dose, and reach were found to vary across cycles and homes. Only 4 homes implemented DCM according to the trial protocol. DCM Basic User training did not prepare all care home mappers to implement DCM accurately and to sustain its use. There was variability in DCM reach related to both mapper implementation and wider care home--level issues. Identifying individuals with the requisite skills and time to implement DCM appear to be challenging in care home settings. Likewise, whether care homes have the right culture and ethos to successfully implement DCM warrants consideration ahead of commencing mapper training. This finding is informative, given the use of well-established DCM implementation procedures. Future complex intervention trials in care home settings will likely benefit from further research on suitable evidence-based implementation strategies in this setting. Consideration may need to be given to complex intervention delivery being conducted wholly through, or with ongoing support of, external translation experts. **Authors' Note:** Data may be made available from the authors on request. The authors would like to thank all the care homes, individuals with dementia, their family members, and care home staff for taking part in this study and giving freely of their time. We would like to thank the following people who contributed to trial design and delivery: Clive Ballard, Jane Fossey, David Meads, Natasha Burnley, Byron Creese, Murna Downs, Lucy Garrod, Elizabeth Graham, Amanda Lilley-Kelley, Vicki McLellan, Holly Millard, Devon Perfect, Louise Robinson, Olivia Robinson, Emily Shoesmith, Najma Siddiqi, Graham Stokes and Daphne Wallace. Also, Chris Albertyn, Marie Crabbe, Cara Gates, Stephanie Jones, Baber Malik, Harriet Maunsell, Kirsty Nash, Sahdia Parveen, Luisa Rabanal, Bina Sharma, Emily Standell, Miguel Vasconcelos Da Silva, and other researchers who collected the data. **Declaration of Conflicting Interests:** The authors declared the following potential conflicts of interest with respect to the research, authorship, and/or publication of this article: C.S. was previously employed by the University of Bradford, who own the IP to the DCM intervention tested in this trial. In this role, she held responsibility for DCM training and method development. She was a technical author on the British Standards Institute PAS 800 guide on implementing DCM in health and social care provider organizations. **Funding:** The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This project was funded by the National Institute for Health Research Health Technology Assessment (NIHR HTA) program (project number 11/15/13). The views and opinions expressed therein are those of the authors and do not necessarily reflect those of the HTA, NIHR, NHS or the Department of Health and Social Care. The funder had no role in study design, methods, recruitment, data acquisition, analysis or preparation of this paper. **ORCID iD:** Claire A. Surr  <https://orcid.org/0000-0002-4312-6661>

1. Introduction {#sec1} =============== Tumor tracking is an essential component of image-guided radiation therapy (IGRT) systems for treating lung and abdominal area tumors, which move significantly due to respiration. With accurate tracking, large margins added to the planning target volumes can be eliminated, and thus, adverse effects of radiation on healthy tissue can be minimized. Duration of treatment sessions can be shortened considerably by tracking tumors and applying radiation continuously. Conventional tracking approaches determine tumor position using internal and/or external surrogates. Often, multiple passive (metallic) \[[@B1]\] and active (electromagnetic) \[[@B2]\] internal fiducials are implanted around tumor to continuously monitor its motion in X-ray videos. However, surgical implantation of such invasive internal fiducials are harmful as they destroy healthy tissues, and patients have greater chance of developing pneumothorax during CT-guided fiducial placement. Besides, internal fiducials slide and relocate during the course of multiple treatment sessions causing uncertainty in their reference positions. It is also possible to apply correspondence models between external markers (chest and abdominal area pointers) and internal fiducials to indirectly estimate tumor position in optical images \[[@B3], [@B4]\]. Yet, the correlation between external markers and tumor position may be violated easily as a result of complex respiratory biomechanics \[[@B5]\]. Alternatively, parametric models of motion patterns are used to track tumors \[[@B6]\]. Such methods, however, require manual labeling of surrogate regions and tumor positions for a long period of time in order to train models. A multiple template matching method for X-ray images is described in \[[@B7]\]. It should be noted that template matching may fail for low-contrast tumor regions where the image quality is low. Here, we present a tumor tracking method that does not require any invasive internal fiducials or external markers. Unlike the traditional methods, we consider the tumor tracking problem as a regression model fitting task between orthogonal X-ray videos and underlying tumor motion as illustrated in [Figure 1](#fig1){ref-type="fig"}. Our intuition is that tumor motion in orthogonal planes can be approximated by affine motion (or a similar parametric motion) and image features can be linearly correlated with these tumor motion parameters. In other words, the feature vector we compute within the tumor region is supposed to be an indicator of the tumor motion. For this, we use an image feature that is sensitive to motion unlike the insensitive features such as intensity histograms that conventional motion estimation methods often require. We learn an online regression model, which is a single matrix that maps image features to motion parameters using the initial pair of the orthogonal X-ray images. We randomly generate a set of 3D motion hypotheses (affine motion matrices) around the initial tumor location. To efficiently cover the parameter space, we generate motion hypotheses conditioned on the respiratory biomechanics (e.g., depending on the position of the tumor, the motion can be constrained to be translational only). These hypotheses map an initial support window volume tumor in both orthogonal X-ray image pairs by simple 3D-2D projection. In addition, we determine an optimal tracking window size in each orthogonal view. We then compute two image feature vectors (e.g., histogram of oriented gradients) in the corresponding image windows, and we concatenate these vectors. To learn the regression model, we solve an overcomplete least-squares fitting problem between the motion hypotheses and concatenated feature vectors using the geodesic distances. In \[[@B8], [@B9]\], a linear regression function is considered. Since affine motion matrices constitute a Riemmanian manifold, motion hypotheses distances should be computed through geodesics. The benefits of using Riemmanian manifold over previously used motion models in Euclidean space (e.g., just vectorizing the motion hypothesis matrix) can be found in \[[@B10]\] for nonmedical examples. We are inspired by \[[@B10]\], yet we significantly extend its basic idea to 3D tumor motion, multiple orthogonal videos, and joint feature computation while further refining it with an adaptive optimal window selection. To estimate the current tumor position in a new X-ray pair, we simply apply the learned regression model to the concatenated feature vector computed in the new X-ray images within the previous tumor windows. Note that we track the tumor position but not its boundary. Boundary tracking can be done by segmentation, registration, boundary fitting, B-splines, and so forth. However, for a low-contrast and invisible tumor, boundary tracking may not be possible since the tumor boundary is not distinct from its surrounding tissues. Our method can track the position of low-contrast tumors and does not require registration. The presented regression method is possibly one of the fastest tumor tracking algorithms (and the fastest that we tested) as it only requires a feature vector and a matrix multiplication without any expensive search operation (template matching, etc.), iterative updates (mean shift), optimization with smoothing or total variation constraints (optical-flow), or testing a large number of hypotheses during the tracking process (particle filtering) as other tracking techniques. It does not require any offline training or patient-specific adaptation. It does not require a tumor detector either. Extensive experiments with challenging real patient datasets demonstrate that the proposed method is robust to different tumor locations and very accurate for low-contrast tumors. 2. Regression Tracking {#sec2} ====================== Regression tracking learns an online regression model **Ω** to correlate the observed image features **X** to underlying tumor motion **Y**. This can be done either learning a single model in 3D Euclidean space (3DR) or learning two separate regression models in 2D planar space then intersecting them in 3D space (2DR). We formulate the model learning as a fitting problem, which finds the regression coefficients **Ω** through minimizing the geodesic distance between the tumor motion matrices **Y** and its estimates **X** **Ω** as $$\begin{matrix} {{\underset{\mathbf{\Omega}}{\min}{\left. ||{\mathbf{X}\mathbf{\Omega} - \mathbf{Y}} \right.||^{2} + \lambda\left. ||\mathbf{\Omega} \right.||^{2}}},} \\ \end{matrix}$$ where *λ* determines the weight of Tikhonov regularization, that is, preference to solutions with smaller smooth norms. We solve ([1](#EEq1){ref-type="disp-formula"}) through ridge regression to determine **Ω**. The solution is given as $$\begin{matrix} {\mathbf{\Omega} = \left( {\mathbf{X}^{\mathbf{\top}}\mathbf{X} + \lambda\mathbf{I}} \right)^{- 1}\mathbf{X}^{\mathbf{\top}}\mathbf{Y}.} \\ \end{matrix}$$ At the first image pair (*t* = 0) of orthogonal X-ray sequences, we learn **Ω** using the initial location *M* ~0~. Then, for any given image pair *t* \> 0, we simply compute a feature vector *h* ~*t*~ and apply **Ω** to find the tumor position: $$\begin{matrix} {M_{t} = M_{t - 1} \cdot \exp\left( {h_{t}^{\mathbf{\top}}\mathbf{\Omega}} \right).} \\ \end{matrix}$$ A flow diagram of our method is given in [Figure 2](#fig2){ref-type="fig"}. Next, we explain how to compute *h* ~*t*~, **X**, and **Y** and details of our 3DR/2DR formulations. 2.1. Training {#sec2.1} ------------- For a single X-ray sequence, the tumor motion between two consecutive frames can be modeled as a 3 × 3 affine matrix *M* (represented by 6 independent motion parameters in 9 coefficients). We use affine motion (translation, rotation, scale change, and skew) to represent the incremental movement of the tumor. Considering the biomechanical tissue models and respiratory mechanics, such an affine model is an adequate model. \[[@B11]\] further states that the tumor motion can be modeled as a simpler periodic function, featuring more time spent at the exhale phrase. When there is no hysteresis with the patient\'s respiratory system, the pathways for inspiration and expiration of one respiratory cycle are approximately the same and almost linear. Even with the presence of hysteresis, the maximum deviation between the two pathways is usually much smaller than the distance along the primary direction of the tumor motion. Thus, a simple translation only motion would be sufficient in most cases; however, we consider full affine motion in our analysis for the completeness of the discussion. Let *M* represent the transformation from a unit square in the object space to the affine region enclosing the target tumor in the image space; that is, (*x* ~img~,*y* ~img~,1)^**⊤**^ = *M*(*x* ~obj~,*y* ~obj~,1)^**⊤**^. The tumor position *M* ~*t*~ at time *t* can be computed efficiently from *M* ~*t*−1~ by $$\begin{matrix} {M_{t} = M_{t - 1} \cdot \Delta M_{t},} \\ \end{matrix}$$ where Δ*M* ~*t*~ is the incremental motion. Affine motion matrices lie on a Riemmanian manifold. This means that we cannot simply vectorize the motion matrices to compute an Euclidean distance between them. Geodesic distances should be measured, thus we apply Lie group exponential map to determine Δ*M* ~*t*~ $$\begin{matrix} {\Delta M_{t} = \exp\left( {h_{t}^{\mathbf{\top}}\mathbf{\Omega}} \right),} \\ \end{matrix}$$ where *h* ~*t*~ ∈ ℝ^*m*^ is a *m* × 1 feature vector corresponding to the unit square in object space warped from *I* ~*t*~ through *M* ~*t*−1~, **Ω** is a *m* × *d* matrix of regression coefficients, and *d* is the number of motion coefficients. For 2D (3D) affine transformation *d* can be set to 6 (12) independent parameters, or all 9 (16) coefficients of the 3 × 3 (4 × 4) affine motion matrix. In either case, each row of **Ω** should be reshuffled to obtain the corresponding motion matrix. To find the optimal solution of ([1](#EEq1){ref-type="disp-formula"}), we first generate a training set of *n* random affine transformation hypotheses Δ*M* ~*i*~ together with their corresponding feature vectors *h* ~0,*i*~ extracted from image *I* ~0~, as shown in [Figure 3](#fig3){ref-type="fig"}. Then, we construct **X** and **Y** as $$\begin{matrix} {\mathbf{X} = \left( {{h_{0,1}}^{\mathbf{\top}};\ldots;{h_{0,n}}^{\mathbf{\top}}} \right)_{n \times m},} \\ {\mathbf{Y} = \left( {\left( {\log\Delta M_{1}} \right)^{\mathbf{\top}};\ldots;\left( {\log\Delta M_{n}} \right)^{\mathbf{\top}}} \right)_{n \times d}.} \\ \end{matrix}$$ Keep in mind that we generate more hypotheses than the feature vector size and significantly more than the number of motion parameters. This means that we obtain an overcomplete system of equations in above minimization. Random sampling serves to achieve an effective and reliable training set and avoid overfitting. One can also quantize the motion space and use those quantized motion parameter values in training. We observed that the value of *n* has little impact on the performance as long as it is larger than the feature vector dimension; that is, *n* \> *m*. In our experiments, we set *n* = 600 for the 2D regression algorithm. Since the incremental motion of the tumor is small \[[@B11]\], we can also limit the sampling bounds using the maximum translational and rotational motion constraints typical for tumor motion. This not only improves the tracking accuracy but also stabilizes the tumor trajectories by removing possible jitters. We use Histograms of Oriented Gradients (HOG) to describe the tumor window features *h* ~*t*~. HOG has 8-bin histograms for each 5 × 5 block within the tracking window, concatenated into a single column vector. During the tracking process, the model **Ω** can be relearned to adapt changes if necessary. In our simulations we have not observed any drift or model distortion issue with only one training at the initialization. Note that the regression model **Ω** built using the first images of the orthogonal sequences where the initial tumor position is given (by table alignment, etc). Then, it automatically tracks the tumor in the newly given images. Our learning method is blind to patient data; it does not make any patient-specific assumption or require any patient-specific information. It does not use any offline training and does not require manual marking of the tumor trajectory either. For 512 × 512 image resolution, automatic generation of motion hypotheses, computing the corresponding concatenated feature vectors and solving ridge regression, takes 0.05 seconds. 2.2. 3D Regression (3DR) {#sec2.2} ------------------------ To learn a joint regression model that correlates the 3D tumor motion directly with the orthogonal X-ray sequences, we group the feature vectors from the two views and estimate a 3D affine motion corresponding to them directly. We construct the combined feature vectors (*h* ~0,*i*~ ^1^; *h* ~0,*i*~ ^2^) from the two X-ray views *I* ~0~ ^1^ and *I* ~0~ ^2^ based on the labeled tumor position at time *t* = 0 and randomly generate *n* 3D affine matrices Δ*M* ~*i*~ ^⋆^ in 3D Euclidean space. Next, we project Δ*M* ~*i*~ ^⋆^ on the orthogonal X-ray image planes. Within the corresponding regions, we extract the features *h* ~0,*i*~ ^1^ and *h* ~0,*i*~ ^2^ for the first and second X-ray views. In this case, **Ω** ^⋆^ maps combined feature vectors (*h* ~*t*,*i*~ ^1^; *h* ~*t*,*i*~ ^2^)  (*t* = 0, *i* = 1,..., *n*) to their corresponding 3D affine motion matrices Δ*M* ~*i*~ ^⋆^, where Δ*M* ~*i*~ ^⋆^ is a 4 × 4 affine matrix. Thus, **X** ^⋆^ = ((*h* ~0,1~ ^1^;*h* ~0,1~ ^2^)^**⊤**^; ...; (*h* ~0,*n*~ ^1^;*h* ~0,*n*~ ^2^)^**⊤**^), **Y** ^⋆^ = ((log⁡Δ*M* ~1~ ^⋆^)^**⊤**^; ...; (log⁡Δ*M* ~*n*~ ^⋆^)^**⊤**^). 2.3. 2D Regression (2DR) {#sec2.3} ------------------------ Instead of learning a 3D regression model, we could simply learn two separate regression matrices **Ω** ~1~ and **Ω** ~2~ for the two X-ray views, and we apply ([4](#EEq4){ref-type="disp-formula"}) and ([5](#EEq5){ref-type="disp-formula"}) iteratively to track the tumor in each X-ray view. Once we have the two individual tracking results *p* ~*t*~ ^1^ and *p* ~*t*~ ^2^ of the two orthogonal X-ray views at time *t*, we can simply compute the tumor position *p* ~*t*~ in 3D through back projection. To do this, we first connect *p* ~*t*~ ^1^ and *c* ~1~ to form lines *l* ~*t*~ ^1^,  *p* ~*t*~ ^2^, and *c* ~2~ to form line *l* ~*t*~ ^2^ and then compute *p* ~*t*~ as the intersection of *l* ~*t*~ ^1^ and *l* ~*t*~ ^2^, while *c* ~1~ and *c* ~2~ are the source points of the two orthogonal X-ray radiation. In practice, *l* ~*t*~ ^1^ and *l* ~*t*~ ^2^ may not necessarily intersect with each other; thus we choose the midpoint of the shortest path connecting these lines to represent the tumor positions. 2.4. Optimal Tracking Window {#sec2.4} ---------------------------- Tracking window size plays an important role in tumor tracking in soft tissues, especially in lung and abdomen areas. To yield valid tracking results for template based methods, the tracking window should not be too big (may underestimate motion) or too small (may lose track). We examine the self-similarity in the local tumor region to find the optimal tracking window size for each patient data. Given a candidate window size *w* ~*x*~ × *w* ~*y*~, we define the local search region of size 2*w* ~*x*~ × 2*w* ~*y*~ with the same center, and *r* = (*w* ~*x*~ × *x* ~*y*~)(*w* ~*x*~ ^⋆^×*w* ~*y*~ ^⋆^)^−1^ ∈ \[0.5^2^, 1.5^2^\], where *w* ~*x*~ ^⋆^ × *w* ~*y*~ ^⋆^ is the bounding size of the tumor. We compute the feature distance for any pair of image patches within the searching region using *ℓ* ^2^ norm and use the mean error as an indicator to describe the discriminatory power of the candidate window size. Intuitively, the larger this mean error the more discriminatory power the candidate size has. However, there may exist different distance distributions, which have the same mean error. To take into account these cases, we give preference to the small (20% of) feature distances and use their mean as the indicator. In our experiments, the ratio *r* of the optimal windows size to the tumor bounding box varies from 0.9 to 1.2 for the coronal view and from 1.1 to 1.3 for the sagittal view for different X-ray data sets. 3. Results {#sec3} ========== For objective performance evaluations, we test our tracking algorithm on digitally reconstructed radiograph (DRR) sequences obtained from real-patient 4DCT data. These DRR sequences have manually labeled ground-truth 3D landmark positions. Using X-ray videos has several issues. To annotate ground-truth motion, X-ray videos should depict tissues with embedded metallic marker. However, it is problematic to make a tracking algorithm to ignore high contrast marker regions, which are often close to tumor and yet compute uncontaminated image features for an unbiased evaluation. Besides, markers themselves introduce uncertainty on the ground-truth data since they may dislocate from the initial calibrated positions or occlude each other in X-ray videos. We use orthogonal DRR sequences obtained from 10 patients\' 4DCT data \[[@B12]\]. This data has different tumor locations, shapes, and internal volume characteristics. Using a state-of-the-art simulator \[[@B13]\], we embed low-contrast tumors in different shapes, sizes, and locations in the original 4DCT data and then generate DRR sequences representing different breathing patterns. Tumor shapes range from spheroids to very intricate 3D polytopes. Each test case is tested with a different regular breathing signal and two irregular breathing patterns. Since we do not impose any temporal smoothing or linear dynamical model (Kalman filter), the performance is not affected by the different breathing patterns. Each DRR sequence we test has around 900 frames. [Figure 4](#fig4){ref-type="fig"} shows sample coronal and sagittal views. Compared with higher resolution flat-panel X-ray digitizer, this data presents considerable challenges. Since the patient CT has limited number of slices, it is low resolution and drastically blurred particularly in the cranial-caudal direction of the coronal view; for example, \~100 × 256 pixels (underlying DRR are of 256 × 256) for Patients 1--5 and \~128 × 512 (DRR 512 × 512) for Patients 6--10. On the other hand, a typical of the flat panel X-ray digitizer has 2048 × 2048 pixels. Limited resolution causes less discriminative features leading potential tracking failures.It has low contrast. The typical dynamic range of a commercial flat-panel digitizer is 16 bits. However, the DRRs we use are 8 bits encoded to push the algorithm to its limit.It contains noise and imaging artifacts. Unlike X-ray videos from digitizer, DRR sequences inherit all 4DCT imaging artifacts due to limited CT scanning speed.We add white random Gaussian noise to the DRR sequences considering that DRR images might not suffer from X-ray image acquisition noise. These issues certainly make tumor tracking more difficult in our dataset. We compare the performance of our 3DR and 2DR methods with the state-of-the-art including the best existing optical flow implementation (OF) \[[@B14]\] and *ℓ* ~1~-based particle filter (P*ℓ* ~1~) \[[@B15]\]. \[[@B14]\] combines the "classical" flow formulation with image boundaries and designs an optimization framework that utilizes median filtering for flow field estimation. \[[@B15]\] is a template-based robust visual tracking method, which enforces sparse representation on the template set and follows a particle filter-based Bayesian state inference. We also implemented a HOG based particle filter (PHOG) algorithm, which uses the same number of 3D particles as the P*ℓ* ~1~ method but computes the observation likelihood from HOG feature matching. The OF algorithm first finds the tumor motions on the two X-ray views and then estimates the 3D tumor position through back projection. We compute the optical flow between two consecutive frames in full resolution since that the magnitude of the tumor motion is unknown and far away feature points can also contribute to the motion estimation of tumor regions. [Table 1](#tab1){ref-type="table"} presents the detailed performance comparison of these algorithms as well as their average processing time per frame. For each test case, we also list the tumor motion magnitude in the last column. We use the Euclidean distance between the estimated and the ground truth (GT) tumor center in 3D as the error measurement (in pixels). The GT tumor center is calculated as the mass center of the GT tumor, while the estimated tumor center is the center of the estimated tracking window. Our results using 10 different patient data show that 2DR gives 1.05 pixel, 3DR 1.16 pixel, OF 3.57 pixel, P*ℓ* ~1~ 5.01, and PHOG 5.68 pixel error on average where the average tumor displacement in the GT is 13.86 pixels. This means that 2DR estimates are 92.5% accurate (1.05 is 7.5% of 13.86). From [Table 1](#tab1){ref-type="table"}, we can see that 3DR and 2DR consistently achieve most accurate tracking results. Other algorithms vary significantly for different test cases and may lose track of the tumor under certain scenarios (e.g., for Patient 10). We do not claim that just because we achieve a 1.05 pixel average error on lower-resolution images we may obtain the same error at higher resolutions. Our experimental results on the 256 × 256 and 512 × 512 datasets clearly show that the estimation errors in terms of *pixels* remains the same (in fact becomes lower: from 1.63 pixel average error for 256 × 256 to 0.69 pixel average error for 512 × 512) when the image resolution increases. In [Table 1](#tab1){ref-type="table"}, 4DCT data for Patients 1--5 have 256 × 256 DRR image size, while Patients 6--10 have 512 × 512. Even though the underlying total displacement doubled from 8.49 pixel to 19.23 pixel on average (implying the tracking problem becomes more challenging), the tracking errors in terms of *pixels* are better for 512 × 512 sequences. In other words, we can confidently expect similar pixel errors when we use even higher resolution images (as a result, get lower mm error). Instead of DRR, if we used a commercial product (e.g., Siemens Axiom Luminos dRF flat detector), 1.05 pixel error would correspond to 0.14 mm to 0.42 mm error. Here, we also like to mention that, for a fair assessment of tracking methods, it is essential to report pixel errors rather than mm errors that is commonly conveyed in medical literature due to bottom-line clinical requirements. However, by measuring mm error, the same algorithm can produce different mm errors using different resolution input data. This does not mean that the algorithm gets any better or worse as nothing algorithmically changes. In [Figure 4](#fig4){ref-type="fig"}, we show the tracking results of different algorithms on data set Patient 7, and our 3DR/2DR tracks the tumor very well for both inhale and exhale phases. In [Figure 5](#fig5){ref-type="fig"}, we draw the estimated tumor positions in the Cranial-Caudal (CC) direction along with the ground truth for Patient 7. It is apparent that the optical flow based tracker tends to underestimate the tumor motion causing significant errors when the tumor motion is large. The particle filter based algorithms (P*ℓ* ~1~ and PHOG), on the other hand, exaggerates the tumor motions at the two extremities and produces jittery results. Another issue with the particle filter based trackers is that the lack of sufficient texture in X-ray images sometimes causes the selection of wrong motion hypothesis as the mode of the approximated pdf after the importance sampling. Computationally optical flow based tracker (OF) is the slowest one among all trackers taking about 2.8 minutes to process a single frame on an Intel 3.4 GHz CPU, which is prohibitive for real-time tasks. There are of course faster OF methods, but their accuracy is worse. Our method can track the tumor in real time: 3DR (2DR) in less than 0.03 (0.06) seconds. 3DR algorithm can be preferred over 2DR for two reasons. First, 3DR generates tracking results twice faster than 2DR. Second, 3DR can avoid the divergence between two orthogonal views by learning a joint 3D regression model and maintaining the tumor positions in 3D space as opposed to two 2D planes. Note that we do not train and test on the same images. We train on the initial pair where the tumor position is known *and then* track on the rest of the sequence automatically. Since the training step takes less than 0.05 seconds, this algorithm can run under any real-time clinical setting. 4. Conclusion {#sec4} ============= We presented a noninvasive tumor tracking method and demonstrated that this tracker outperforms the state-of-the-art both in accuracy (\~1 pixel error) and speed (0.03 sec). This corresponds to 7.5% tumor positioning error with respect to maximum tumor dislocation. Such a small tumor location error significantly reduces the prescribed treatment volume margins from several centimeters to millimeter range, and thus, prevents radiating healthy tissue in IGRT systems \[[@B16]\]. As a future study, we will extend the regression model to incorporate biomechanical tissue constraints for very complex tumor shapes. {#fig1} {#fig2} {#fig3} {#fig4} {#fig5} ###### Performance comparison (in pixel) of different tracking algorithms. 3DR and 2DR are very robust and consistently achieve most accurate tracking results (the best tracking result is in bold).   3DR 2DR OF PHOG P*ℓ* ~1~ Total displacement ------------ ------------- ------------- -------------- --------------- ------------- -------------------- Patient 1 0.78 ± 0.34 0.52 ± 0.29 2.15 ± 0.79 1.63 ± 0.36 3.28 ± 1.30 8.03 Patient 2 2.61 ± 1.62 2.27 ± 1.37 2.12 ± 0.90 1.88 ± 0.96 7.54 ± 3.99 7.76 Patient 3 0.83 ± 0.42 0.70 ± 0.38 1.10 ± 0.43 2.06 ± 2.11 3.39 ± 0.83 10.47 Patient 4 2.86 ± 1.48 2.67 ± 1.41 4.20 ± 1.10 2.54 ± 0.84 4.38 ± 1.59 10.98 Patient 5 1.07 ± 0.54 0.92 ± 0.50 2.16 ± 0.90 2.14 ± 1.28 4.36 ± 1.19 5.23 Patient 6 0.65 ± 0.54 0.67 ± 0.48 10.98 ± 3.40 2.45 ± 1.50 5.76 ± 1.18 18.31 Patient 7 0.99 ± 0.49 0.94 ± 0.45 4.80 ± 1.87 4.71 ± 2.80 5.94 ± 1.49 22.13 Patient 8 0.69 ± 0.37 0.72 ± 0.39 3.15 ± 1.29 7.45 ± 8.30 5.19 ± 1.22 22.36 Patient 9 0.33 ± 0.29 0.35 ± 0.27 1.20 ± 0.60 3.07 ± 3.27 3.09 ± 1.01 7.28 Patient 10 0.82 ± 0.46 0.77 ± 0.45 3.92 ± 1.80 28.94 ± 11.52 7.16 ± 3.14 26.07 [^1]: Academic Editor: Kayvan Najarian

###### Key messages What is already known about this subject? ========================================= - Apparent diffusion coefficient (ADC) is a newly validated method of quantifying spinal disease activity in axial spondyloarthritis (SpA). What does this study add? ========================= - In a group of participants with axial SpA and back pain, we found that the Ankylosing Spondylitis Disease Activity Score (ASDAS) correlated well with the mean and maximum ADC values after adjustment for confounding factors and Spondyloarthritis Research Consortium of Canada spine MRI indices. How might this impact on clinical practice? =========================================== - Our data show evidence that ASDAS is an objective disease assessment tool in patients with axial SpA. - Our study also provides a new method for validation of future disease assessment tools (eg, biomarkers) in axial SpA. Introduction {#s1} ============ In the past decade, MRI has gained prominence in both diagnosis and monitoring of disease activity in axial spondyloarthritis (SpA). MRI was included as an imaging criterion in the 2009 Assessment of Spondyloarthritis International Society (ASAS) classification criteria.[@R1] When available, MRI is also recommended prior to consideration for biological disease-modifying antirheumatic drugs.[@R3] Disease activity is currently monitored using the Ankylosing Spondylitis Disease Activity Score (ASDAS)[@R4] according to established ASAS guidelines. ASDAS is a composite index that combines five disease activity variables into a single score. It has been developed as an improvement on previously self-rated tools, such as the Bath Ankylosing Spondylitis Disease Activity Index (BASDAI),[@R5] by incorporating objective biochemical measures of C reactive protein (CRP) or erythrocyte sedimentation rate (ESR) into the data-driven index.[@R6] The score is highly discriminatory[@R7] and has outperformed other disease assessment tools[@R8] in the assessment of disease activity. ASDAS paralleled the levels of inflammatory markers both elevated during active disease and decreased after treatment with biologics.[@R9] However, the relationship between clinical disease activity and inflammation on MRI is inconclusive and considerably dependent on the imaging sequence used. In general, ASDAS shows better correlation with MRI inflammation than other clinical disease activity parameters.[@R10] Traditional sequences such short tau inversion recovery (STIR) or T2 fat suppression have reasonable spatial resolution, which is advantageous in showing the extent of inflammation but poorly sensitive in measuring intensity. Diffusion-weighted imaging (DWI) is a newer MRI sequence that exploits the impedance of water molecules at the tissue level[@R15] to visualise the bone marrow oedema of spinal inflammation. By removing artefacts, the corresponding computer-generated apparent diffusion coefficient (ADC) maps produce the most objective measures of intensity of inflammation. DWI is shown to outperform STIR imaging in quantifying disease activity,[@R16] and our recent research has proposed ADC as an imaging biomarker of spinal inflammation in SpA.[@R18] We hypothesise that ASDAS is associated with intensity of inflammation on diffusion-weighted MRI as represented by ADC values. Methods {#s2} ======= This is a cross-sectional study using data from a large DWI observational cohort in participants with axial SpA. It has been registered in the clinical trial registry of The University of Hong Kong. The goals of the cohort are to evaluate the use of DWI in the diagnosis and monitoring of disease activity in patients with axial SpA. We included patients older than 18 years old with back pain and an expert diagnosis of axial SpA from eight rheumatology centres (Queen Mary Hospital, Grantham Hospital, Tung Wah Hospital, Pamela Youde Nethersole Eastern Hospital, Caritas Medical Centre, Tseung Kwan O Hospital, Kwong Wah Hospital and Prince of Wales Hospital) from April 2014 to February 2019. Participants who were pregnant, on biological therapy, on prednisolone (or dose equivalent steroid) dosage of 10 mg or more, or contraindicated for MRI were excluded. All participants were required to sign an informed consent. Details of the cohort have also been reported in our previous publications.[@R18] Study design {#s2-1} ------------ All recruited participants were interviewed for demographic and clinical data. These included age, gender, duration of back pain, family history of SpA, history of inflammatory bowel disease, history of psoriasis and history of enthesitis. Duration of back pain was defined as the time between the first onset of back pain and the date of interview. Physical examination was performed to determine tender joint (44 joints) and swollen joint (44 joints) counts. Participants completed self-assessment questionnaires, including BASDAI and Bath Ankylosing Spondylitis Global Score.[@R20] The questionnaires were measured as Numerical Rating Score from 1 to 10. Blood tests, including CRP, erythrocyte sedimentation rate (ESR) and human leucocyte antigen (HLA)-B27, were done. ASDAS based on CRP (ASDAS-CRP) and on ESR (ASDAS-ESR) were calculated. Radiographs of the lumbosacral spine were done to determine the presence of radiological AS. Fulfilment of the ASAS classification criteria for axial SpA[@R1] was determined in all participants. All recruited participants had whole-spine MRI from the cervical (C2) region to the lumbosacral (S1) region and the sacroiliac (SI) joint MRI done on the day of the interview. T1, STIR and diffusion-weighted images were obtained simultaneously. Only STIR and diffusion-weighted images were used in this study. The MRIs were performed using a 3.0 T imaging unit (Achieva; Philips Healthcare, Best, the Netherlands) with participants in supine position. ADC maps were automatically generated by the MRI system. The technical parameters published in our previous publication[@R18] are summarised as follows: repetition times/echo times 5000/80 (STIR) and 4000/90 (DWI); fields of view 150×240 mm^2^ (STIR) and 300×241 mm^2^ (DWI); slice thicknesses 3.5 mm (STIR) and 4 mm (DWI); and multiple b values 0, 100, 600 and 1000 sec/mm^2^. All magnetic resonance images were performed on a single machine. The acquisition time for STIR sequence and DWI were 2.48 and 2.44 min, respectively. Reading of images {#s2-2} ----------------- Anteroposterior view of lumbosacral radiographs were performed and scored for sacroiliitis by a single reader (HHLT). All radiographs were scored according to the modified New York criteria for Ankylosing Spondylitis (AS).[@R21] The gradings were as follows: 0=normal, 1=suspicious, 2=obvious, 3=partial fusion and 4=complete fusion. Bilateral grade 2 or unilateral grade 3 or above were defined as radiographical axial SpA. MRI inflammation was defined as hyperintensity in the vertebral bodies in STIR images. The STIR images of the whole spine were scored independently by a rheumatologist and a rheumatology trainee (HYC and SCWC) according to the Spondyloarthritis Research Consortium of Canada (SPARCC) Spine MRI Index[@R22] and the SPARCC SI MRI Index.[@R23] HYC had 8 years' experience and SCWC had 4 years' experience in SpA MRI interpretation. The averages of SPARCC scores by the two readers were used in our analyses. A musculoskeletal radiologist (KHL), with 4 years' experience in SpA MRI interpretation, identified all inflammatory lesions in the vertebrae from the scored STIR images. Obvious degenerative lesions were also excluded by KHL. With reference to lesions identified in the STIR sequence, two independent readers (HYC and ETFC) placed regions of interest (ROIs) on the ADC maps of DWI accordingly to determine the maximum apparent diffusion coefficient (ADCmax) and the mean apparent diffusion coefficient (ADCmean) values. Background apparent diffusion coefficient (ADCbg) values were determined from the average of at least two ADC values measured at the centres of adjacent normal-appearing vertebral bodies. Normalised maximum apparent diffusion coefficient (nADCmax) and normalised mean apparent diffusion coefficient (nADCmean) values were calculated by dividing ADCmax and ADCmean, respectively, by ADCbg. All ADC values used in the analyses were the average of ADC values by the two readers. Magnetic resonance images and ADC values were visualised and determined using OsiriX MD V.9.5.2. All MRI and radiology readers were blinded to the clinical and biochemical data. Analyses and statistics {#s2-3} ----------------------- Demographic, clinical, biochemical and imaging data were reported as mean±SD. Univariate and multivariate linear regressions were used to determine the associations between ASDAS-CRP or ASDAS-ESR and ADC values or SPARCC spine MRI scores. We included only patients with measurable ADC lesions in the analyses. ASDAS-CRP or ASDAS-ESR was the dependent variable in univariate regression models. In addition to the independent variables nADCmax, nADCmean, SPARCC spine MRI score and SPARCC SI MRI score, other known or expected associated factors, including age, male gender, HLA-B27 positivity, duration of back pain, family history of SpA, radiological AS, tender joint count, swollen joint count, and current or history of enthesitis were also involved in the analyses. Independent variables with p values less than 0.1 in univariate analyses were retested in multivariate regression models using either ASDAS-CRP or ASDAS-ESR as dependent variables. 'Enter' mode was used in the analyses. Two independent multivariate models using nADCmax and nADCmean were built for each multivariate ASDAS model. Results were reported as regression coefficient (β) and standard coefficient with 95% CI in linear regression models. Intraclass correlation coefficient was used to determine the interobserver agreement between the two SPARCC MRI spine scores, SPARCC MRI SI scores and different ADC parameters. The degree of reliability was interpreted as 0.00--0.20, slight; 0.21--0.40, fair; 0.41--0.60, moderate; 0.61--0.80, substantial; and 0.81--1.00, almost perfect. Unless specified, p values less than 0.05 were considered statistically significant. All statistics were performed with commercial software (IBM SPSS Statistics V.25). Listwise deletions were performed for missing data. Results {#s3} ======= Three hundred one participants with axial SpA and back pain were recruited for the study. Most of them had HLA-B27 positivity and established radiographical SpA. Among our participants, 14.8% had psoriasis and 2.7% had inflammatory bowel disease. Our study population was characterised by a slight male predominance, prolonged disease duration and significant back pain. Participants had high clinical disease activity (ASDAS-CRP 2.0±0.9 and ASDAS-ESR 3.1±1.0).[@R6] The average SPARCC spine MRI score was 6.3±8.6 and the average SPARCC SI MRI score was 3.2±6.0. The group with identifiable ADC lesions had more men, fewer peripheral arthritis, fewer tender and swollen joint count, and higher ESR ([table 1](#T1){ref-type="table"}). ###### Comparing demographic, clinical, biochemical and imaging features between participants with and without identifiable ADC lesions With identifiable ADC lesions Without identifiable ADC lesions P value ----------------------------------------------- ------------------------------- ---------------------------------- --------- Age (N=301) (years) 45.7±13.1 42.7±13.1 0.09 Male gender (N=301) 55 (67.1%) 115 (52.5%) 0.02 Duration of back pain (N=299) (years) 13.4±10.7 11.2±11.2 0.13 HLA-B27 positivity (N=290) 69 (85.2%) 164 (78.5%) 0.20 History of inflammatory bowel disease (N=298) 3 (3.8%) 5 (2.3%) 0.49 History of psoriasis (N=298) 8 (10.0%) 36 (16.5%) 0.16 Family history of SpA (N=286) 15 (19.5%) 48 (23.0%) 0.53 Radiographical axial SpA (N=294) 5 (71.4%) 21 (52.5%) 0.35 Fulfilled ASAS axial SpA criteria (N=300) 79 (97.5%) 200 (91.3%) 0.06 Back pain NRS (N=294) 5.8±2.4 5.6±2.4 0.50 Ever peripheral arthritis (N=298) 36 (45.0%) 132 (60.6%) 0.02 Tender joint count (N=292) 0.8±1.5 1.6±3.1 0.02 Swollen joint count (N=293) 0.3±0.9 0.7±1.6 0.02 Ever enthesitis (N=297) 35 (43.8%) 97 (44.7%) 0.88 CRP (N=300) (mg/dL) 1.2±1.4 1.0±1.9 0.48 ESR (N=299) (mm/hour) 37.1±23.3 30.1±24.9 0.03 ASDAS-CRP (N=292) 2.1±0.8 2.0±0.9 0.44 ASDAS-ESR (N=291) 3.3±1.0 3.1±1.1 0.11 ADC background (N=82) (mm^2^/s) -- -- ADCmax (N=82) (mm^2^/s) 1.45±0.31×10^−3^ -- -- ADCmean (N=82) (mm^2^/s) 0.77±0.19×10^−3^ -- -- nADCmax (N=82) 6.5±2.1 -- -- nADCmean (N=82) 3.4±1.0 -- -- SPARCC SI MRI score (N=297) 3.4±5.9 3.1±6.1 0.68 SPARCC spine MRI score (N=294) 13.7±10.2 3.5±6.0 \<0.001 ADC, apparent diffusion coefficient; ADCmax, maximum apparent diffusion coefficient; ADCmean, mean apparent diffusion coefficient; ASAS, Assessment of Spondyloarthritis International Society; CRP, C reactive protein; ESR, erythrocyte sedimentation rate; HLA, human leucocyte antigen; N, number; nADCmax, normalised maximum apparent diffusion coefficient; nADCmean, normalised mean apparent diffusion coefficient; NRS, Numerical Rating Score; SPARCC, Spondyloarthritis Research Consortium of Canada; SpA, spondyloarthritis. Three hundred twenty-five STIR lesions were found in 98 (32.6%) participants. Two hundred seventy-four (84.3%) STIR lesions from 82 (83.7%) participants could be located in ADC maps ([figure 1](#F1){ref-type="fig"}). The technical success rates of the ADC measurement of inflammatory lesions in STIR images at individual spinal levels were 24/38 (63.2%), cervical spine; 188/203 (92.6%), thoracic spine; and 62/84 (73.8%), lumbosacral spine. Intraclass correlation coefficient of ADCmax, ADCmean, nADCmax and nADCmean between the two readers were 0.86, 0.82, 0.75 and 0.63, respectively. When compared with normal vertebrae (ADCbg), the maximum and average ADCs of inflammatory lesions were 6.4 and 3.4 times higher, respectively. Most of the measurable ADC lesions were located in the midthoracic spine. [Figure 2](#F2){ref-type="fig"} shows the details of their distribution. {#F1} {#F2} Inter-reader reliability of SPARCC spine MRI scores and SPARCC SI MRI score by the two readers was almost perfect (intraclass correlation coefficient was 0.92 for SPARCC spine MRI score and 0.95 for SPARCC SI MRI score). Univariate and multivariate regression analyses {#s3-1} ----------------------------------------------- In univariate analyses, ASDAS-CRP was positively associated with swollen joint count, ADCmax, ADCmean and SPARCC spine MRI score. It was negatively associated with HLA-B27 positivity ([table 2](#T2){ref-type="table"}). ASDAS-ESR was also positively associated with swollen joint count, ADCmax, ADCmean and SPARCC spine MRI score. It was negatively associated with male gender and HLA-B27 positivity ([table 3](#T3){ref-type="table"}). ###### Univariate and multivariate linear regression analyses using ASDAS-CRP as dependent variables Univariate analyses using ASDAS-CRP as dependent variable Multivariate analyses using ASDAS-CRP as dependent variable and ADCmax as independent variable (N=70) Multivariate analyses using ASDAS-CRP as dependent variable and ADCmean as independent variable (N=70) ------------------------------- ----------------------------------------------------------- ------------------------------------------------------------------------------------------------------- -------------------------------------------------------------------------------------------------------- ------- ----------------------- ------ ------- ----------------------- -------- HLA-B27 positivity (N=77) −0.19 −0.44 (−0.95 to −0.07) 0.09 −0.21 −0.51 (−1.03 to 0.02) 0.06 −0.21 −0.49 (−1.02 to 0.05) 0.08 Swollen joint count (N=75) 0.19 0.18 (−0.03 to 0.39) 0.10 0.14 0.14 (−0.07 to 0.34) 0.19 0.15 0.14 (−0.06 to 0.35) 0.17 ADCmax (N=78) 0.29 0.001 (0.00 to 0.001) 0.01 0.27 0.001 (0.00 to 0.001) 0.02 -- -- -- ADCmean (N=78) 0.28 0.001 (0.00 to 0.002) 0.01 -- -- -- 0.21 0.001 (0.00 to 0.002) 0.07 SPARCC spine MRI score (N=74) 0.31 0.03 (0.01 to 0.04) 0.01 0.32 0.03 (0.01 to 0.05) 0.01 0.34 0.03 (0.01 to 0.05) \<0.01 Age (N=78) 0.05 0.003 (−0.01 to 0.02) 0.67 -- -- -- -- -- -- Male gender (N=78) −0.09 −0.15 (−0.55 to 0.25) 0.45 -- -- -- -- -- -- Smoker (N=78) 0.17 0.30 (−0.01 to 0.68) 0.13 -- -- -- -- -- -- Drinker (N=78) 0.12 0.31 (−0.27 to 0.89) 0.30 -- -- -- -- -- -- Duration of back pain (N=77) −0.01 −0.001 (−0.02 to 0.02) 0.94 -- -- -- -- -- -- Family history of SpA (N=75) 0.15 0.30 (−0.16 to 0.77) 0.20 -- -- -- -- -- -- Tender joint count (N=75) 0.16 0.09 (−0.04 to 0.21) 0.18 -- -- -- -- -- -- Ever enthesitis (N=78) 0.10 0.17 (−0.20 to 0.55) 0.37 -- -- -- -- -- -- nADCmax (N=78) 0.01 0.10 (−0.09 to 0.10) 0.91 -- -- -- -- -- -- nADCmean (N=78) 0.04 0.04 (−0.16 to 0.23) 0.73 -- -- -- -- -- -- SPARCC SI MRI score (N=67) 0.13 0.02 (−0.02 to 0.06) 0.30 -- -- -- -- -- -- ADC, apparent diffusion coefficient; ADCmax, maximum apparent diffusion coefficient; ASAS, Assessment of Spondyloarthritis International Society; CRP, C reactive protein; HLA, human leucocyte antigen; N, number; NaDCmax, normalised maximum apparent diffusion coefficient; SI, sacroiliac; SPARCC, Spondyloarthritis Research Consortium of Canada; SpA, spondyloarthritis. ###### Univariate and multivariate linear regression analyses using ASDAS-ESR as dependent variables ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------ Univariate analyses using ASDAS-ESR as dependent variable Multivariate analyses using ASDAS-ESR as dependent variable and ADCmax as independent variable (N=70) Multivariate analyses using ASDAS-ESR as dependent variable and ADCmean as independent variable (N=70) ------------------------------- ----------------------------------------------------------- ------------------------------------------------------------------------------------------------------- -------------------------------------------------------------------------------------------------------- ------- ------------------------ -------- ------- ------------------------ -------- Male gender (N=78) −0.30 −0.60 (−1.05 to −0.16) 0.01 −0.21 −0.43 (−0.88 to −0.03) 0.07 −0.18 −0.36 (−0.83 to 0.11) 0.13 HLA-B27 positivity (N=77) −0.31 −0.82 (−1.40 to −0.24) 0.01 −0.23 −0.63 (−1.23 to −0.02) 0.04 −0.23 −0.62 (−1.22 to −0.01) 0.05 Swollen joint count (N=75) 0.27 0.28 (0.04 to 0.52) 0.02 0.17 0.18 (−0.04 to 0.41) 0.11 0.18 0.20 (−0.03 to 0.42) 0.09 ADCmax (N=78) 0.29 0.001 (0.00 to 0.002) 0.01 0.24 0.001 (0.00 to 0.001) 0.03 -- -- -- ADCmean (N=78) 0.36 0.002 (0.00 to 0.003) \<0.01 -- -- -- 0.22 0.001 (0.00 to 0.002) 0.05 SPARCC spine MRI score (N=74) 0.26 0.03\ 0.02 0.36 0.03 (0.01 to 0.06) \<0.01 0.36 0.03 (0.01 to 0.06) \<0.01 (0.003 to 0.05) Age (N=78) 0.12 0.01 (−0.01 to 0.03) 0.29 -- -- -- -- -- -- Smoker (N=78) 0.11 0.23 (−0.23 to 0.68) 0.33 -- -- -- -- -- -- Drinker (N=78) 0.13 0.39 (−0.29 to 1.07) 0.25 -- -- -- -- -- -- Duration of back pain (N=77) −0.08 −0.01 (−0.03 to 0.01) 0.51 -- -- -- -- -- -- Family history of SpA (N=75) 0.13 0.31 (−0.24 to 0.86) 0.26 -- -- -- -- -- -- Tender joint count (N=75) 0.17 0.11 (−0.04 to 0.25) 0.14 -- -- -- -- -- -- Ever enthesitis (N=78) 0.10 0.19 (−0.25 to 0.62) 0.40 -- -- -- -- -- -- nADCmax (N=78) −0.02 −0.01 (−0.12 to 0.10) 0.85 -- -- -- -- -- -- nADCmean (N=78) 0.08 0.08 (−0.15 to 0.31) 0.49 -- -- -- -- -- -- SPARCC SI MRI score (N=67) 0.11 0.02 (−0.03 to 0.07) 0.38 -- -- -- -- -- -- ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------ ADC, mean apparent diffusion coefficient; ADC, apparent diffusion coefficient; ASAS, Assessment of Spondyloarthritis International Society; CRP, C reactive protein; ESR, erythrocyte sedimentation rate; HLA, human leucocyte antigen; N, number; SI, sacroiliac; SPARCC, Spondyloarthritis Research Consortium of Canada; SpA, spondyloarthritis; nADCmax, normalised maximum apparent diffusion coefficient; nADCmean, normalised mean apparent diffusion coefficient. Multivariate regression analyses showed that ADCmax and SPARCC spine MRI score were independently associated with both ASDAS-CRP and ASDAS-ESR. ADCmean was associated with ASDAS-ESR and tended to associate with ASDAS-CRP. HLA-B27 positivity had negative associations with ASDAS-ESR and had a tendency to associate negatively with ASDAS-CRP. Male gender tended to associate negatively with ASDAS-ESR when ADCmax was used as an independent variable. Results are shown in [tables 2 and 3](#T2 T3){ref-type="table"}. [Figure 3](#F3){ref-type="fig"} shows an example of a female participant with moderate clinical disease activity and moderate level of ADC values. {#F3} Discussion {#s4} ========== As ASDAS is a recommended by ASAS guidelines[@R24] for assessment of disease activity, much research has attempted to show its relationship with STIR--MRI inflammation. In this study using DWI--ADC, we have demonstrated its association with the intensity of spinal inflammation in a large cohort of participants with axial SpA. A positive association between ASDAS and spinal inflammation has been demonstrated in some studies using STIR--MRI[@R10] but has failed in others.[@R13] While STIR sequence is the preferred and recommended MRI sequence for the assessment of disease activity in axial SpA, it is limited by its inability to quantify inflammation. STIR sequence is highly effective in demonstrating the extent of inflammation, but DWI-ADC allows measurement of intensity. The ability of DWI-ADC in quantifying inflammation intensity has been demonstrated in various diseases.[@R25] In previous studies, ADC values were found to correlate with CRP in patients with SpA[@R28] and to be increased in active sacroiliitis.[@R29] Our earlier study has also demonstrated that ADC is a potential imaging biomarker of disease activity in axial SpA.[@R18] Our results showed that both ADCmax and ADCmean were positively associated with ASDAS, even after adjustments for potential confounding factors and the SPARCC MRI spine score. Although the SPARCC MRI spine score semiquantitatively grades intensity of inflammation by comparison to that of cerebrospinal fluid, this was allocated a less important weighting in the overall score. Therefore, the SPARCC MRI spine score functions more as a description of the extent rather than the intensity of spinal inflammation. Incorporating both STIR and ADC, axial disease activity in SpA may be more comprehensively described. Participants with and without MRI inflammation had similar back pain and clinical disease activity scores. These findings were consistent with other reports[@R13] highlighting the importance of MRI imaging in addition to clinical disease activity. As a data-driven disease activity index, ASDAS has been repeatedly shown to outperform another widely used clinical disease assessment tool, the BASDAI.[@R7] Very high disease activity measured by ASDAS predicted radiological progression better than by BASDAI.[@R31] In addition, superior correlations between ASDAS and STIR--MRI inflammation have also been demonstrated in various studies.[@R10] Despite an association between ADC and back pain score in our previous study[@R18] and with ASDAS in this study, it had no association with BASDAI.[@R18] Based on the two studies, ASDAS surpasses BASDAI in describing the intensity of spinal inflammation and should be the preferred choice in the assessment of axial SpA. HLA-B27 has been reported to associate with earlier disease onset, severity of spinal and SI joint inflammation, and radiological sacroiliitis.[@R32] The negative associations between HLA-B27 positivity and clinical disease activities (ASDAS-CRP and ASDAS-ESR) in our analyses were not expected. Interestingly, such an association has also been reported previously by another international study.[@R34] Further studies may help to reveal their true relationship. A major challenge of using spinal ADC is the high degree of variability between MRI machines. A proposed solution is to use normalised apparent diffusion coefficient (nADC), which compares ADC values of inflammatory sites to normal tissues (ADCbg). In this study, the mean of two apparently normal regions near the site of inflammation was used to calculate ADCbg. However, this method has not been validated. Factors such as age, osteoporosis[@R35] and skeletal maturity[@R36] may affect the ADC values, and axial SpA patients are more prone to osteoporosis.[@R37] On normalisation, nADC values showed decreased interobserver reliability, and the associations with ASDAS were lost. A possible reason was the lack of standardisation in drawing the ROIs of the normal non-inflamed regions. AS was also known to be associated with other spinal pathologies, such as osteoporosis and vertebral fractures, which could affect the measured ADC values. This further increased the variability of ADCbg measurements. Having said that, the loss of associations between nADC values and ASDAS should not affect our conclusion. Normalisation of ADC values would not be essential since only a single MRI machine and ADC software to acquire the ADC data were used. Future studies should attempt to find out the best way of ADC normalisation to allow accurate comparison of ADC values between different MRI machines. Potential measurement errors of mean ADC values may be another challenge. Inflammation is rarely homogenous, and inadvertent inclusion of normal tissues within the boundaries of ROIs may result in falsely diluted mean ADC values. Poor visuospatial resolution of ADC maps and small inflammatory lesions, such as corner inflammatory lesions,[@R38] also contributed to errors in measurements. As such, ADCmax may be a more objective way to represent spinal inflammation in axial SpA. Despite this, the two readers had good interobserver reliabilities in both ADCmean and ADCmax measurements. ADC technical failure was another limitation.[@R39] Nevertheless, we encountered extensive inflammatory lesions visualised on STIR images with minimal intensity that render them undetectable on ADC. Because each imaging technique characterises unique aspects of spinal inflammation, the addition of ADC to the more traditional STIR imaging has an added value of quantifying inflammation. Our study has other limitations. Application of ADC is limited to patients with identifiable MRI inflammation only. The technique focuses on quantifying intensity of inflammation, and no meaningful values could be drawn from patients without MRI inflammation. This restricted the number of participants included in our analyses. In contrast to other international studies, we included mainly participants with advanced axial SpA with high clinical activity. It is not clear whether the same results could be replicated in the early disease group. Our inclusion of participants with prerequisite back pain may have excluded those with asymptomatic yet active inflammation, hence contributing to selection bias. Despite this, the percentage of participants with active spinal inflammation was compatible with another international study.[@R40] The exclusion of obvious degenerative lesions might also falsely exclude coexisting inflammatory lesions. Finally, we did not include SI joint ADC values into the analyses because ADC acquisition in SI joints has not been validated. Yet, SPARCC SI MRI was not associated with ASDAS in our study. Conclusion and future direction {#s5} =============================== Using DWI-ADC, we demonstrated that ASDAS is associated with both the extent and intensity of spinal inflammation in patients with detectable inflammatory lesions. Data from other ethnic groups and prospective analyses would provide a more complete picture of this relationship. The authors thank Cynthia Yan-yan Chan for editing the manuscript. HYC and ETFC contributed equally. **Contributors:** Study conception and design: HYC, KHL and CSL. Acquisition of data: HYC, HHLT and SCWC. Interpretation of imaging: HYC, ETFC, KHL, HHLT and SCWC. Analysis and interpretation of data: HYC, ETFC and CSL. Drafting of the article: HYC and ETFC. Revision of the article: HYC and ETFC. **Funding:** This work is supported by the Hong Kong Society of Rheumatology, Novartis Research and seed fund from The University of Hong Kong. **Competing interests:** None. **Patient consent for publication:** Not required. **Ethics approval:** The study was approved by the institutional review board of the University of Hong Kong/Hospital Authority Hong Kong West Cluster (reference number UW 14-085) and local ethics committees. It was conducted in accordance with the Declaration of Helsinki and the guidance of Good Clinical Practice on 30 November 2006. All participants gave written informed consent before recruitment. **Provenance and peer review:** Not commissioned; externally peer reviewed. **Data availability statement:** Data are available upon reasonable request.

Varteks schneidert fleißig am Turnaround. Der Textilgigant aus Kroatien mit namhaften Kunden wie Hugo Boss fertigt in Innovation, Schnelligkeit und Qualität, damit die Kehrtwende passt. Trotz Virus-Krise hofft Varteks auf bessere Zeiten. Das Traditionsunternehmen aus Kroatien stand vor zwei Jahre kurz vor dem Aus. Dann aber ist es geglückt, sich in der schwierigen Modebranche mit frischem Kapital und innovativem Konzept doch noch neue Märkte und Kunden zu erschließen. Statt Tweedstoffen läuft jetzt leichte Baumwolle unter tackernden Nadeln: Selbst in lähmenden Corona- Zeiten schnurren Nähmaschinen fleißig beim Textilgiganten Varteks in Varaždin. Die ersten 70.000 Gesichtsmasken im Wert von 100.000 Euro habe man der Polizei und Altersheimen gespendet, berichtet Vorstandschef Tomislav Babić stolz: \"Für eine Firma, die vor Kurzem vor dem Bankrott stand, ist das ein enormer Betrag.\" Trotz Kurzarbeit und Produktionsdrosselung werde Varteks die Covid-19-Pandemie \"mit weniger Verlusten als viele Konkurrenten\" überstehen, ist der 45-Jährige überzeugt: \"Wir werden aus dieser Krise stärker und mit einem größeren Marktanteil herauskommen, als wir in sie hineingeraten sind.\" Die \"Schnelligkeit bei der Anpassung an eine neue Situation\" bezeichnet der ehemalige Investmentbanker als wichtigste Voraussetzung für das Management eines Unternehmens in der Transformation: \"Das Zweite ist die Kreativität, neue Wege zu entwickeln, und den Mut, diese auch umzusetzen. Und natürlich viel Arbeit - ohne die erreicht man nichts.\" Der Mann mit dem offenen Hemdkragen weiß, wovon er spricht. Noch vor zwei Jahren schien das Flaggschiff der kroatischen Textilindustrie gewissermaßen klinisch tot - und unvermeidlich in den Bankrott zu schlittern. Das Unternehmen sei \"praktisch erledigt\" gewesen, sagt Babić im Rückblick. \"Das mutigste, riskanteste Manöver der kroatischen Wirtschaftsgeschichte\" {#Sec1} =========================================================================== Dass sich damals der heutige Aufsichtsratsvorsitzende und Hauptinvestor Nenad Bakić bereitfand, mit einer Kapitalspritze den Neuanfang zu wagen, bezeichnet Babić als \"das mutigste, aber auch riskanteste Manöver der kroatischen Wirtschaftsgeschichte\": \"Er übernahm eine völlig heruntergekommene Firma und setzte auf ein Konzept, von dem alle dachten, dass es angesichts der viel billigeren Arbeitskraft in Asien und der Türkei nicht verwirklichbar sei: die Produktion hochwertiger Mode in Europa.\" Die Anfänge der turbulenten Geschichte von Varteks gehen bis ins Königreich Jugoslawien zurück. Unter dem Namen Tivar wurde die Firma 1918 zunächst als Stoffproduzent gegründet. Schon Ende der 20er Jahre mauserte sich Tivar mit einem Netz von 190 Niederlassungen zu einem der führenden Kleidungshersteller des Königreichs. Nach dem Zweiten Weltkrieg wurde Tivar im sozialistischen Jugoslawien zu Varteks - und weitete seine Produktion aus. Zu den besten Zeiten arbeiteten bei der Modegröße, die sich auf Anzüge, Mäntel, Kleider, Röcke und Stoffe spezialisiert hatte, mehr als 11.000 Mitarbeiter. Kroatien sei in Jugoslawien als \"fortschrittlichere Republik schon früh nach Westen orientiert\" gewesen - auch in der Mode, so Babić. Varteks beschäftigte die besten Mode-Designer und unterhielt mehr als 200 Filialen: \"Es gab in Jugoslawien keine Konkurrenz, die sich mit Varteks hätte messen können.\" Lange einträgliche Kooperation mit US-Hersteller Levi Strauss {#Sec2} ============================================================= Nicht nur im sozialistischen Osteuropa erschloss sich das Unternehmen bald wichtige Exportmärkte: Mit hochwertigen, aber günstigen Anzügen stieß Varteks selbst in Großbritannien auf eine gute Nachfrage. Im Jahr 1983 begann eine für Varteks lange sehr einträgliche Kooperation mit Levi Strauss & Co.: Über ein Vierteljahrhundert ließ der US-Hersteller seine \"Levi\'s\"-Jeans für den osteuropäischen Markt in Varaždin schneidern. Der Kroatienkrieg zwischen 1991 und 1995 sowie der anschließende Zerfall Jugoslawiens zerstörten die angestammten Märkte des Unternehmens. Mit der Herstellung von Uniformen für die Streitkräfte Kroatiens überbrückte Varteks den Krieg: Bis heute kleidet das Unternehmen die Armee und Polizei im Adria-Staat ein. Bis heute Anzüge als Auftragsarbeit fürs deutsche Mode-Label Hugo Boss {#Sec3} ====================================================================== Zudem begann 1995 die Kooperation für das deutsche Mode-Label Hugo Boss, für das Varteks bis heute Herrenanzüge als Auftragsarbeit fertigt. \"Für uns ist diese langjährige Zusammenarbeit eine der größten Anerkennungen unserer Qualität - und ein Beweis, welche hochwertige Ware wir produzieren\", sagt Babić. Nach dem Kroatienkrieg habe sich Varteks auch dank der Auftragsarbeit für andere internationale Modegrößen \"zunächst sehr gut\" gehalten. Die Zahl der Beschäftigten hatte sich zwar auf 3.500 Mitarbeiter reduziert: \"Doch die Firma war intakt, verfügte über einen guten Marktanteil und operierte erfolgreich\", wie Babić beschreibt. Doch die Weltwirtschaftskrise von 2008/2009 traf nicht nur Kroatien, sondern auch Varteks mit voller Wucht. Von 2009 bis 2014 wies der Küstenstaat fünf Jahre lang in Folge ein Minuswachstum auf. Auch in Varaždin brachen bald alle Dämme. Der langjährige Geschäftspartner Levi Strauss kündigte 2009 die Kooperation mit Varteks auf und verlagerte die Produktion aus Kostengründen nach Asien. Zusätzlich zur sinkenden Kaufkraft auf dem Heimatmarkt erschwerte verstärkte Konkurrenz das Geschäft: Mit dem EU-Beitritt von Kroatien 2013 und der Eröffnung neuer Shopping-Zentren drängten westliche Modeketten wie H&M, Zara oder Peek & Cloppenburg ins Land. \"Es war für Varteks unmöglich, preislich mit in Bangladesh gefertigten Billiganzügen zu konkurrieren\", begründet Babić. Varteks trudelte in eine Abwärtsspirale, in der ein freier Fall kaum aufzuhalten schien. Die unrentabel gewordene Stoffproduktion wurde 2010 eingestellt. Die Belegschaft schrumpfte auf noch 1.100 Mitarbeiter. Dennoch häufte das Unternehmen zwischen 2008 und 2018 mehr als 100 Millionen Euro Verluste an. \"Selbst gesunde Unternehmen können derart hohe Verluste nicht aushalten\", sagt Babić. Das Minus glich Varteks mit dem Verkauf der Filialen aus seinem Immobilienimperium zunächst aus. Aber, wie Babić zu bedenken gibt: \"Wenn alles bergab geht, verschwinden irgendwann die besten Leute. Vor allem unser Mode-Design hatte zu leiden.\" Zinsfreie Kredite vom Teilhaber, weil Banken den Dienst versagten {#Sec4} ================================================================= Die Kehrtwende begann 2018, als Investor Nenad Bakić seine Beteiligung an Varteks von sechs auf 17 Prozent erhöhte und mit einer Kapitalspritze von 20 Millionen Kuna (2,6 Millionen Euro) auf 47 Prozent steigerte. Bakić gewährte dem Unternehmen schließlich selbst zinsfreie Kredite von insgesamt 45 Millionen Kuna (sechs Millionen Euro), weil die Banken ihren Firmenkunden Varteks nicht mehr bedienen wollten. Das radikale Umkrempeln der Unternehmensstrategie und des Managements sollten Varteks aber erst den ersehnten Neustart ermöglichen. Investor Bakić übernahm Anfang 2019 befristet auf ein Jahr selbst das Ruder im Vorstand und leitete mit dem Aufbau eines neuen Managements die Wende ein. \"Wir haben das Management von der Spitze bis in die mittlere Führungsebene völlig umgekrempelt\", berichtet Babić, der im vergangenen Jahr als Entwicklungschef ins Unternehmen eintrat und zu Jahresbeginn 2020 den Vorstandsvorsitz übernahm, als Bakić wieder in den Aufsichtsrat wechselte. Dem vorherigen Management habe es \"an Vision und Mut\" für Varteks gefehlt, bemängelt Babić: \"Wir begannen, uns um neue Wege zu bemühen. Wir beschleunigten Entscheidungsprozesse, beseitigten administrative Hürden, stellten praktisch alle Geschäftssegmente völlig neu auf.\" Immer kürzere Modezyklen mit acht statt zwei Kollektionen pro Jahr begünstigen nach Ansicht des neuen Varteks-Managements den Trend, dass die Produktion näher zum Kunden und damit zurück nach Europa verlagert wird. Verstärkt setzt auch Varteks auf Produktion und Vertrieb hochwertiger Eigenprodukte über das eigene Verkaufsnetz. \"Wichtig ist heute vor allem Qualität, und dass neue Mode so schnell wie möglich in die Läden gelangt\", unterstreicht Vorstandschef Babić: \"Unser Vorteil ist, dass wir in unserem eigenen Hinterhof produzieren. Wir können sehr schnell auf neue Trends und Marktbedürfnisse reagieren. Die Zeitspanne von der Idee bis zu Realisierung ist bei uns sehr kurz. Wir erhöhen die Produktion, wenn ein Artikel gut läuft - zwei Tage später ist er im Laden.\" Nicht nur Design und Stoffe haben sich bei Varteks völlig verändert, seit eine neue Kreativdirektorin eingestiegen ist. Auch die erst auf 22 Filialen geschrumpfte Ladenkette ist nun zu einem Netz von 30 Läden ausgebaut. Pfiffige Marketing-Kampagnen und zudem der durchs Land rollende \"Modebus\" haben den Absatz 2019 um 60 Prozent erhöht. In dem Bus können Kunden nicht nur Varteks-Kleidung anprobieren, sondern werden auch mit deren Bestellung über den Webshop vertraut gemacht. Die eigenen Marken gehen gut im Einzelhandel und im neuen Webshop. \"Auftragsarbeit immer weniger profitabel\" {#Sec5} =========================================== In der Produktion müht sich Varteks, den Schwerpunkt auf die profitablere Herstellung eigener Kleidermarken zu verlegen: Seit 2018 ist deren Umsatzanteil von 40 auf 70 Prozent gewachsen. Heimische Textilproduzenten, die nur auf Auftragsarbeit setzten, könnten den allmählichen Anstieg des Lohnniveaus und des Lebensstandards \"nicht überleben\", prognostiziert Babić: \"Mit steigenden Löhnen wird die Auftragsarbeit immer weniger profitabel.\" Auftragsarbeit für Kunden wie Hugo Boss und J. Lindeberg sichern nur bei Varteks zusätzlich den Cashflow und die Arbeitsplätze ab, erklärt er, warum die Firma dennoch weiter auf beide Segmente setzt: \"Wir haben die Auftragsarbeit zwar nicht reduziert, aber den Absatz unserer eigenen Kleidermarken vergrößert.\" Die geplante Expansion nach Serbien und Slowenien hat Varteks noch nicht abgeblasen, aber wegen der Virus-Krise vorläufig auf Eis gelegt. Mit dem Hilfsprogramm der Regierung, die während der Kurzarbeit die Minimallöhne bezahlt und Kreditgarantien übernimmt, bleibt die Unternehmensführung für die Zeit nach der Krise positiv gestimmt. Und die eigene Kundennähe scheint ein zusätzliches Faustpfand, wie Babić betont: \"Wenn die Läden aufgehen, sind wir besser gerüstet als die Konkurrenten, die Importeure ihrer Waren sind: Die werden Schwierigkeiten haben, im Sommer rechtzeitig ihre Lager zu füllen.\" Die Löhne der Branche seien \"generell gering\", räumt Babić ein. Doch vorbei seien die Zeiten, in denen sich Varteks-Beschäftigte mit dem gesetzlichen Mindestlohn von derzeit 3.650 Kuna netto (482 Euro) zufriedengeben: \"Große Gehaltssprünge können wir uns nicht leisten. Aber wir wollen, dass unsere Mitarbeiter zufrieden sind - und sie zu guter Arbeit stimulieren.\" Veränderung bei Varteks wertet Gewerkschaft positiv {#Sec6} =================================================== Auch die Textilgewerkschaft bewertet die Veränderungen bei Varteks positiv: Der Abbau von Arbeitsplätzen sei gestoppt, sogar Stellen seien in den neuen Filialen geschaffen worden, fasst Nenad Leček als Vorsitzender der nationalen Gewerkschaft TOKG zusammen. Als besonders erfreulich wertet er, dass im vergangenen Jahr \"erheblich\" in die Klimatisierung der Werkshalle investiert wurde: \"Wir glauben, dass Varteks mit der neuen Strategie und den eingeleiteten Veränderungen eine Perspektive hat.\" Die Stärkung der eigenen Marken sei Treiber für Veränderungen, glaubt Leček: \"Wir sind uns bewusst, dass der Umschwung angesichts der wirtschaftlichen Folgen der Pandemie nicht leichtfallen wird. Das größte Problem ist die Liquidität - und die könnte sich durch die Corona-Krise noch verschlechtern.\" Mit dem Verkauf von drei Vierteln des Firmengeländes will die Varteks-Unternehmensführung jetzt die Mittel beschaffen, die zum Begleichen der Altschulden und für neue Investitionen fehlen. Babić sieht mit den \"Trends zur Rückkehr der Mode nach Europa und zu lokalen Marken\" auch mehr \"Licht für die Zukunft\", wie er sagt: \"Wir sind Teil einer Erfolgsgeschichte: In Ex-Jugoslawien gibt es nicht viele Fälle, die dafür stehen, dass eine fast bankrotte Großfirma zu neuem Leben erweckt wurde.\" {#Sec7} Herrenanzüge, Mäntel und elegante Geschäftskleidung für Frauen sind das Kerngeschäft von Kroatiens Modegigant Varteks in Varaždin. Das Unternehmen wurde unter dem Namen Tivar 1918 gegründet und im sozialistischen Jugoslawien in Varteks umbenannt. Exportmärkte in Ost und West wurden in den 70er und 80er erschlossen. Jeans für das börsennotierte US-Handelsunternehmen Levi Strauss & Co. fertigte Varteks zwischen 1993 und 2008. Die bis heute erfolgreiche Kooperation mit Hugo Boss startete 1995, als Kroatien unabhängig wurde. Der Niedergang von Varteks begann mit der Weltwirtschaftskrise ab 2008. Doch neu aufgestellt mit frischem Kapital, Management und Konzept gelang dem Konzern vor zwei Jahren der Turnaround: Varteks setzt seitdem verstärkt auf Eigenmarken, die über die eigene Ladenkette vertrieben werden. www.varteks.com {#Sec8} Den Trend zur Rückkehr der Textilproduktion aus asiatischen Billiglohnländern nach Europa beflügelt zusätzlich die Corona-Krise.Die Nachfrage nach einer kunden- und marktnahen Produktion in der Branche verstärken die immer kürzeren Modezyklen.Das damit einhergehende, vor allem schnellere Anpassen an veränderte Situationen, mehr Kreativität und mehr Mut bei der Realisierung von Ideen und Umsetzungswegen sind also für Unternehmen im Umbruch entscheidend.

Introduction {#sec1} ============ Primordial germ cells (PGCs) are the stem cells of the gametes, providing genome transmission to future generations ([@bib15]). During development, PGCs undergo specification, migration, and proliferation. Reciprocal interactions between germ cells and somatic cells are important for gonadal differentiation ([@bib12]). However, little is known about the regulatory role of germ cells during sexual development. In mammals, agametic male gonads develop into a normal testis cord, while loss of germ cells in ovaries at birth disrupts ovarian structures and folliculogenesis ([@bib22]). In teleosts, the requirement of germ cells for gonadal development appears to be variable. Their absence leads to exclusive male development in medaka and zebrafish ([@bib14; @bib37; @bib38]), but not in goldfish or loach ([@bib4; @bib6]). Mammalian sex determination is regulated by antagonistic pathways, which direct the bipotential embryonic gonad toward ovarian or testicular fate ([@bib46]). Moreover, evidence indicates that somatic sex needs to be reinforced throughout adulthood. In mice, loss of FOXL2 in mature ovary or DMRT1 in mature testis causes transdifferentiation of somatic cells ([@bib21; @bib43]). In zebrafish, oocytes appear essential for the development of females in juveniles and for maintenance of the sexual phenotype in adults ([@bib3]). The number of PGCs likely plays an important role in teleost sexual differentiation. For medaka and stickleback, females possess more germ cells than males due to their sexually dimorphic proliferation ([@bib16; @bib33]). Transplantation of a single PGC into a germline-deficient zebrafish embryo generates males exclusively ([@bib34]). *ziwi* mutants with reduced PGC numbers can develop as males or females; however, a greater reduction due to a hypomorphic allele in *trans* to a null allele gives rise to males ([@bib7]). These data argue that the absolute number of germ cells is important in determining the sexual phenotype of zebrafish. Zebrafish are undifferentiated gonochorists since all individuals first initiate oogenesis via forming an immature ovary ([@bib41]). In developing males, but not in females, a gonad transformation arises from apoptosis-driven degeneration of oocytes ([@bib42; @bib44]) about 23--35 days postfertilization (dpf) leading to subsequent testis development ([@bib26; @bib42]). Molecular control of sex determination and gonad differentiation in zebrafish appears to be complex ([@bib18; @bib26]) and variable across domesticated strains versus wild populations ([@bib19; @bib47]). In this study, we analyze the relationship between the number of PGCs and sexual differentiation in zebrafish. By tracking changes in the PGC number during development, we demonstrate that a dimorphic proliferation of PGCs occurs in the early larvae, underlining the beginning of sexual differentiation. By creating zebrafish containing various numbers of PGCs, we show that a threshold number of PGCs is required for stabilizing the ovarian fate and that PGC number may directly regulate the progression of gonadal transformation. Results {#sec2} ======= Dimorphic Proliferation of PGCs Occurred during Early Larval Stages in Zebrafish {#sec2.1} -------------------------------------------------------------------------------- To understand how the PGC count might be involved in sexual differentiation, we examined their numbers at different development stages using the *Tg(vasa:vasa-EGFP)* (i.e., *zf45Tg*) transgenic zebrafish line ([Figure 1](#fig1){ref-type="fig"}A). For clarity, the term "PGC" was used before 2 weeks of zebrafish development. First, we estimated the number of PGCs present in the gonadal region between 1 dpf and 8 dpf with the squash method. The number of PGCs ranged from 25 to 44 at 1 dpf, without obvious fluctuations in the PGC number during the first week of development ([Figure 1](#fig1){ref-type="fig"}B). Next we used optical sectioning to precisely count PGC numbers at 7 and 14 dpf in WT larvae from two different families (FI and FII). The PGC count at 7 dpf appeared to follow a unimodal distribution ([Figure 1](#fig1){ref-type="fig"}C). At 14 dpf, the number of PGCs showed a bimodal distribution between two distinct populations (means = 40.3 ± 7.7 and 87.7 ± 12.1; [Figure 1](#fig1){ref-type="fig"}D), and the difference between the means was significant (Student's t test, p \< 0.05). Morphologically, green fluorescent protein (GFP)-expressing gonadal regions exhibited a smaller and less dense morphology at 7 dpf ([Figure 1](#fig1){ref-type="fig"}E) and became larger as the number of GFP-expressing cells increased at 14 dpf. In 14 dpf individuals with a higher number of PGCs, gonadal regions were broader and denser due to the increased number of cells populating that region ([Figures 1](#fig1){ref-type="fig"}F and 1G). To determine whether this bimodal distribution in PGC number was due to a family-specific effect, we examined the distribution of PGC number by family. Intriguingly, 83.3% of individuals in FI, the family with a slightly female-biased offspring sex ratio (60%), exhibited gonadal regions with a higher number of PGCs, while nearly 60% of individuals in the male-biased FII (84%) contained a lower number of PGCs at 14 dpf ([Figures 1](#fig1){ref-type="fig"}H and 1I). Our data suggest that a dimorphic proliferation of PGCs occurs between 7 and 14 dpf and that progeny sex ratios within families may correlate with the divergent distribution of PGC numbers. Depletion of PGCs in the Embryos Resulted in Masculinization of Zebrafish Gonads {#sec2.2} -------------------------------------------------------------------------------- To further investigate whether there was a relationship between early PGC count and sexual development, we compared unmanipulated *Tg(vasa*:*vasa*-*EGFP)* individuals with low PGC and high PGC counts and found no difference in their trunk-based expression at 22 dpf ([Figure S1](#app3){ref-type="sec"}A available online) or adult sex ratio ([Figure S1](#app3){ref-type="sec"}B). Next, we generated zebrafish morphants with depleted PGCs using two different methods. In the first approach, we microinjected a diluted morpholino oligonucleotide (MO) directed against the *dead end* gene (*dnd*) into *Tg(vasa*:*vasa*-*EGFP)* zebrafish embryos. Injected embryos were categorized based on the PGC number observed at 24--32 hours postfertilization (hpf) and then grown to adulthood together with uninjected embryos (control), and sexual phenotype was assessed. Next, we used optical sectioning to count PGC numbers in *dnd* morphants at 7 and 14 dpf. In the majority of larvae from zero PGC and severely depleted PGC groups (PGC count \<6 at 24--32 hpf), gonadal structures were absent ([Figures 2](#fig2){ref-type="fig"}A and 2B), and these individuals were excluded from further microscopic analysis. In the remaining PGC-depleted group (PGC count 6--9 at 24--32 hpf), the majority of the larvae at 7 and 14 dpf had loose aggregates of PGCs, but were lacking clear gonadal structures. Of the undepleted gonads (\>20 PGCs at 24--32 hpf), more than 60% showed clear gonadal structures at 7 dpf. Among larvae with visible PGCs, the average PGC number was 7.8 ± 4.8 in the PGC-depleted gonads (6--9) and 29.1 ± 9.4 in undepleted gonads (\>20) at 7 dpf. At 14 dpf, the morphology of PGC-depleted gonads was similar to those in 7 dpf; however, all of the larvae with undepleted gonads (\>20) exhibited distinct gonadal structures ([Figure 2](#fig2){ref-type="fig"}B). The average PGC number was 25.6 ± 16.8 in PGC-depleted gonads; in contrast, the PGC count in undepleted gonads displayed a bimodal distribution comparable to that observed in uninjected larvae (means = 47.4 ± 9.8 and 99.3 ± 20.9; [Figure 2](#fig2){ref-type="fig"}C, right). This suggested that morpholino-induced PGC depletion below a threshold effectively prevented subsequent PGC proliferation in most individuals and that the effect was maintained during later stages. After 3 months postfertilization (mpf), the sex ratios of the *dnd* morphants were evaluated. For both pairwise cross (MO1) and mass cross (MO2), the "zero PGC" group produced exclusively males as expected based on published data ([@bib38]). Their gonads were empty testicular shells completely devoid of germ cells confirming earlier data ([@bib37]). A strong male-biased sex ratio (mean = 90% in four batches for MO1 and 76% in six batches for MO2) was identified in PGC-depleted groups (1--10 PGCs for MO1, 1--7 for MO2) ([Figure 2](#fig2){ref-type="fig"}D). These results suggested that---similar to the minimum PGC number required for proliferation---a certain number of PGCs seems to be necessary for unaffected ovarian development in most individuals. Individuals with severely depleted PGCs (1--5), excluded from the morphological analysis, all developed into fertile males with a gonadal histology similar to WT males and were included in the subsequent transcriptomic analysis. To further explore the correlation between PGC number and the phenotypic sex of fish using a different approach, we applied both single PGC transplantation (SPT; [Figure 3](#fig3){ref-type="fig"}A) and blastoderm transplantation (BdT; [Figures 3](#fig3){ref-type="fig"}B--3G and 3I--3K) for generating germline chimeras. Both procedures utilize sterilized, PGC-less host embryos. GFP-labeled PGCs were used for SPT and GFP-tagged blastodermal cells (containing DsRed-expressing PGCs) were applied for BdT. In general, SPT results in a germline chimera possessing a single PGC, whereas BdT yields individuals with variable number of donor-derived PGCs in the early gonad. PGC counts for the latter are typically much lower than those observed in uninjected controls. In SPT chimeras, a single PGC was present at the gonadal regions of 20.7% of the host embryos at the prim-5 stage ([Figure 3](#fig3){ref-type="fig"}A) and all developed as males. For BdT chimeras, transplanted donor-derived cells were mixed with PGC-depleted host cells uniformly at the prim-5 stage, and most donor PGCs were recognized as red fluorescent cells around the gonadal ridge (3--29 cells; [Figures 3](#fig3){ref-type="fig"}B--3D; see also [Movies S1](#mmc6){ref-type="supplementary-material"} and [S2](#mmc7){ref-type="supplementary-material"}). A total of 109 BdT germline chimeras survived to adulthood (see [Table S1](#app3){ref-type="sec"} for details), at which point identification of the two sexes was performed based on phenotype and/or dissection of their gonads; both female and male chimeras showed mosaic GFP fluorescence and regional red fluorescent protein (RFP) fluorescence ([Figures 3](#fig3){ref-type="fig"}F, 3G, 3J, and 3K). Nearly 80% of BdT recipients developed as males in the PGC-depleted group (1--9) ([Figure 2](#fig2){ref-type="fig"}D). Representative gonads from female and male BdT chimeras were subjected to histological analysis and showed completely developed gonads indistinguishable from those of controls ([Figures 3](#fig3){ref-type="fig"}H and 3L). Although examined gonads of SPT chimeras showed an intact testis only on one side, these fish were sexually active and fertile. None of the "zero PGC" morphants possessed fully developed, fertile gonads. It was noted that ∼10%--20% of PGC-depleted individuals (initial PGC count 1--9) still developed as females ([Figure 2](#fig2){ref-type="fig"}D). Taken together, our data show when the PGC number is artificially lowered below a threshold, a substantial portion of genetic females will be forced through masculinization, increasing the proportion of males among the offspring. Zebrafish Gonads Were in a Meiotic Stage at 14 dpf {#sec2.3} -------------------------------------------------- To understand how a reduced PGC number might influence sexual differentiation in zebrafish, we used a microarray-based approach to investigate differences between the transcriptomes of PGC-depleted (1--9 PGCs) versus WT larvae. We collected samples from developing trunk regions for histology and gene expression profiling. At 14 dpf, WT samples clustered into two groups based on principal component analysis (PCA; [Figure 4](#fig4){ref-type="fig"}A), and seven genes were differentially expressed (DE; \>2-fold, p \< 0.05; see [Table S2](#mmc2){ref-type="supplementary-material"} for gene names and [Table S3](#mmc3){ref-type="supplementary-material"} for a selected set of DE genes). Among them were genes involved in cell-cycle regulation (*fbxl18*, *cdkn2aipnl*, and *cbx3*), consistent with our observation that bimodal proliferation of PGCs occurs between 7 and 14 dpf. The comparison of transcriptomes between WT and PGC-depleted samples indicated that the latter (cluster 1) did not group closely with WT samples (clusters 2 and 3; [Figure 4](#fig4){ref-type="fig"}A), suggesting the differences in transcriptome between PGC-depleted individuals and the future WT males. Further gene expression analysis showed 134 DE genes when comparing combined WT clusters (2 and 3) to cluster 1 ([Table S3](#mmc3){ref-type="supplementary-material"}). Among them, 102 genes were upregulated, and 32 genes were downregulated in WT compared with PGC depleted ([Figure 4](#fig4){ref-type="fig"}B; see [Table S4](#app3){ref-type="sec"} for a selected set of DE genes). Gene ontology (GO) analysis identified gene clusters of meiosis, cell-cycle regulators, immune response, and germ/stem cell genes among those in this DE group. In particular, meiotic genes (*dazl*, *hormad1*, *smc1b*, and *sycp2*) were downregulated, whereas most genes with immune related function (*irf8*, *cd209*, and *ccl1*) were upregulated in the PGC-depleted group compared with WT. In addition, potential ovarian (*org*) and testicular (*rnf17*) markers were expressed at higher levels in the WT group. Our data suggested that (1) the trunk-based transcriptome of PGC-depleted individuals shows clear differences from those of their WT siblings, (2) WT zebrafish gonads are in a meiotic stage at 14 dpf, and (3) entry into meiosis might be an important early step in ovarian differentiation. Multiple Sexual States Were Present during Gonadal Transformation at 22 dpf {#sec2.4} --------------------------------------------------------------------------- At 22 dpf, gene expression data indicated that WT samples could be divided into two clusters based on PCA, with 945 DE genes (\>2-fold, p \< 0.05). When WT and PGC-depleted samples were compared through a PCA plot, five distinct clusters were formed, suggesting that different transitional states of gonads might be present. All WT individuals fell into clusters 1--4, whereas PGC-depleted individuals remained in cluster 5. For analytical purposes, cluster 1 was assigned as "immature females," clusters 2--4 as WT transforming (female-to-male), and cluster 5 as PGC-depleted transforming individuals ([Figure 4](#fig4){ref-type="fig"}C). To find genes with a potential function in early gonad differentiation and/or testis formation, we compared "immature females" (cluster 1) with a subset of WT and PGC-depleted transforming males (clusters 4 and 5). A total of 1,329 DE genes was identified (\>2-fold, p \< 0.05; [Table S3](#mmc3){ref-type="supplementary-material"}). Among them, 1,136 genes were upregulated in "immature females," and 193 genes were overexpressed in future males ([Figure 4](#fig4){ref-type="fig"}D). Unlike at 14 dpf, where PGC-depleted samples showed separation from WT individuals, two of the unmanipulated samples (very likely the most advanced transforming males from cluster 4) coclustered with PGC-depleted individuals at 22 dpf ([Figure 4](#fig4){ref-type="fig"}C). This indicated that the expressed gene set of WT males and PGC-depleted males became more similar to each other by 22 dpf. To gain insight into possible biological functions of these genes, they were sorted into signaling pathways based on the KEGG or PANTHER classification system ([@bib9; @bib23]). The Importance of Transforming Growth Factor β Signaling for Meiotic Progression and Folliculogenesis in Developing Zebrafish Gonads {#sec2.5} ------------------------------------------------------------------------------------------------------------------------------------ In "immature females" (cluster 1), cohorts of genes associated with cell cycle, cell death, and metabolism were upregulated, indicating active remodeling of tissue architecture and morphogenesis. Detailed analysis of this cluster identified genes related to germ/stem cell markers (*pou5f1*, *lin28a*, *nanog*, *nfr*, and *piwi*-related machinery) and ovarian markers (*zp* related, *figla*, *org*, *foxl2*, and *cyp19a1a*), demonstrating that these developing gonads were in an ovarian state. In addition, genes pertaining to transforming growth factor β (TGF-β) signaling, including *bmp15*, *dvr1*, *foxh1*, *gdf2*, and *gdf9*, were also overexpressed in these individuals, some of which have been shown to be important for ovarian follicle development in zebrafish and mammals (see, e.g., [@bib2; @bib11; @bib27]). Similarly, 28 DE genes, including translational regulators (*ddx31*, *lsm14b*, *elav2*, *eif4e1b*, *pabpc1l*) and chromatin modifiers (*chtopb*, *h1m*, *setd8b*, *suv39h1a*) were identified in cluster 1. Some of these (e.g., *dazl*, *spo11*, *smc1b*, and *sycp3)* play essential roles in various aspects of meiosis ([@bib1]). In summary, our data indicated that mRNA translation and epigenetic regulation in the oocyte might be important for meiotic progression, suggesting that TGF-β signaling contributed to the development of meiotic oocytes and ovarian follicles. Several transcription factors were identified in cluster 1 as potential activators involved in profemale development. Some of them, including *sox19b* and *zglp1*, are required for ovary or granulosa cell development in various species ([@bib17; @bib24]). Among 193 genes overexpressed in WT and PGC-depleted transforming males (clusters 4 and 5; [Table S3](#mmc3){ref-type="supplementary-material"}), 20 were upregulated over 3-fold, and nine of them were uncharacterized (for a selected set, see [Table S4](#app3){ref-type="sec"}). To further investigate the transforming process of gonads, cluster 2 was compared with clusters 3 and 4. There was a total of 50 DE genes and 30 of them upregulated in cluster 2 (\>2-fold, p \< 0.05). Among the subset of 20 genes upregulated in clusters 3 and 4 were several potential testis markers, including *rnf14-like* (*loc100333926*), *rtn4b*, *cyb5r1*, and *metrnl* ([@bib25; @bib51]) (for a selected set, see [Table S4](#app3){ref-type="sec"}). To validate the microarray data, we examined the expression profiles of a selected set of DE genes and several candidate sex associated genes by a quantitative PCR (qPCR) array. A total of 26 and 107 genes was tested for 14 dpf and 22 dpf, respectively, and nearly 70% of them were verified ([Figures 5](#fig5){ref-type="fig"}A and 5B; see [Table S5](#mmc4){ref-type="supplementary-material"} for primers and [Table S6](#mmc5){ref-type="supplementary-material"} for detailed results). We identified the upregulation of *zp2* in WT samples and *star* (Leydig cell marker) in PGC-depleted gonads at 14 dpf. In contrast, we did not detect differential expression in *amh*, *sox9a* (Sertoli cell marker), and *star* at 22 dpf consistent with microarray data. PGC-Depleted Gonads May Undergo a Less Distinct "Juvenile Ovary to Testis" Transformation {#sec2.6} ----------------------------------------------------------------------------------------- To ask whether PGC-depleted gonads differentiate in the same manner as those of WT, histological analysis was performed at different developmental stages. At 15 dpf, WT gonads contained different types of germ cells, including meiotic germ cells suggestive of the early stage of primary oocytes ([Figure 6](#fig6){ref-type="fig"}A, see inset). However, no mature, differentiated germ cells were observed in PGC-depleted gonads ([Figure 6](#fig6){ref-type="fig"}B), indicating that they were underdeveloped. This observation was consistent with confocal imaging of gonads with severely depleted PGCs, in which no clear structures were observed at 14 dpf ([Figure 2](#fig2){ref-type="fig"}B). At 23 dpf, in sharp contrast to the gonads of WT individuals ([Figure 6](#fig6){ref-type="fig"}C), there were no perinucleolar oocytes identified in PGC-depleted gonads ([Figure 6](#fig6){ref-type="fig"}D). By 28 dpf, WT gonads contained packed perinucleolar oocytes ([Figure 6](#fig6){ref-type="fig"}E). However, PGC-depleted gonads showed a range of morphologies during differentiation; some had developed ovarian structures ([Figure 6](#fig6){ref-type="fig"}F), while others solely contained germ cells with one or multiple nucleoli in the nucleus ([Figure 6](#fig6){ref-type="fig"}G). At 35 dpf, signs of apoptosis were observed, indicative of gonad transformation ([Figure S3](#app3){ref-type="sec"}). No difference was observed between adult PGC-depleted and WT gonads ([Figure S4](#app3){ref-type="sec"}). Taken together, our histological findings suggest that WT gonadal primordium first develops into an immature ovary with different types of germ cells at 2 wpf. The number and size of perinucleolar oocytes may provide good indicators for ovarian maturation at later stages. In contrast, the development of PGC-depleted gonads appeared to be delayed and inadequate compared with WT. Moreover, PGC-depleted individuals may exhibit a less pronounced or shorter juvenile ovary stage during sexual differentiation. Thus, we propose that gonadal transformation may serve as a checkpoint to ensure the developmental stability of ovaries or testes. Discussion {#sec3} ========== In this report, we show that a dimorphic increase of germ cells occurs in the early larvae of zebrafish, similar to Japanese medaka ([@bib35]). Our data show that there is little change in the germ cell number during the first week of development. However, two distinct germ cell populations appear between 7 and 14 dpf, and this is strongly correlated with the resulting sex ratios of progeny. Thus, we propose that a sex-specific proliferation of germ cells marks the beginning of gonadal differentiation in zebrafish and individuals with a high PGC number have an increased propensity for the female fate. We used two complementary methods, MO-based knockdown and cell transplantation, to generate zebrafish containing a spectrum of PGC numbers. Our findings revealed that a threshold number of PGCs is required for the stability of ovarian fate. Furthermore, no compensatory proliferation of germ cells in the PGC-depleted morphants was observed, thus maintaining the morpholino-induced phenotype during development. In organogenesis, the number of tissue-specific stem cells might regulate the final size of the organs ([@bib40]). Our results suggest that the ability of PGCs to proliferate is constrained by their number, which consequently changes the size and identity of the gonad. The development of PGC-depleted gonads appeared to be inadequate and protracted, which seems to delay differentiation further. When PGC number increased, compensatory growth occurred, indicating the presence of potential feedback molecules, possibly growth factors derived from somatic cells. Interestingly, we also observed 10%--20% of zebrafish with a PGC number below the threshold developed as females, suggesting that these individuals might possess a distinct genetic makeup, leading to different response to the size of germ cell pool. We do not know whether there is any connection between the initial PGC count and the response to the depletion. It is possible that the "additional males" appearing in the MO-depleted batch are genetic females with a lower number of initial PGCs (within the normal female range) prior to injection that forced them to develop a testis. Germ cell sex determination and meiotic initiation are tightly coupled events during the early stage of sexual differentiation ([@bib15]). Our histological analysis and gene expression profiles (*zp2*, *org*, *sycp3*, *dazl*) confirmed that zebrafish gonads are in a meiotic ovarian stage at 14 dpf. Concurrently, we also detected putative testis markers expressed in WT gonads, indicating the plastic nature of early immature gonads expressing "lineage priming" genes of both sexes as shown in mammals ([@bib8]). Comparative transcriptome analysis showed that under unbiased selection, WT individuals display similar expression patterns, indicating that zebrafish undergoing either mode of proliferation (and with either sexual fate) will enter meiosis. On the other hand, PGC-depleted gonads were not able to express ovarian or meiotic markers at that stage; thus, the juvenile ovary stage might be accordingly delayed. In mammals, the mutual antagonism between CYP26B1/FGF9 and retinoic acid (RA) regulates the meiotic entry of germ cells ([@bib12]). Our data suggest that zebrafish might use alternative molecules or mechanisms for executing meiosis since we did not detect dimorphic expression of *cyp26a1* either at the beginning of sexual differentiation (14 dpf) or during meiotic progression of ovary (22 dpf). It should be noted that immune response genes were predominantly upregulated in PGC-depleted morphants at 14 dpf, and this may correlate with PGC depletion inducing a regenerative response similar to that of injured tissues ([@bib5]). Moreover, we observed that immune response genes were upregulated in both WT and the PGC-depleted morphants at 22 dpf, suggesting that these genes may serve key regulatory roles during gonadal development, as in rainbow trout ([@bib49]). The comparison of transcriptomes among WT zebrafish identified seven DE genes at 14 dpf and 945 DE genes at 22 dpf, manifesting the increasing divergence of promale versus profemale pathways during development. At 22 dpf, the expression of multiple signaling pathways revealed a complex regulatory network in gonadal differentiation. Several developmental pathways have been shown to be involved in germline development and the juvenile ovary to testis transformation in zebrafish, including the canonical Wnt, Tp53/Fancl, nuclear factor-κB, and Piwi/piRNA pathways ([@bib7; @bib28; @bib31; @bib39]). Our data demonstrate that genes/pathways supporting ovarian follicle development are coupled with genes associated with meiotic progression and mRNA translation, further promoting the female fate. Histological analysis showed the quantitative difference in perinucleolar oocytes between WT and PGC-depleted *dnd* morphants, consistent with the observation that the ratio of perinucleolar oocytes in total germ cells dictates gonadal fate ([@bib42]). Our findings suggest that oocyte meiosis is important for ovarian development. It has been reported that sex reversal in medaka occurs during meiosis ([@bib36]). Similarly, as shown in *fancl* mutants, only oocytes surviving through meiosis support ovarian differentiation in zebrafish ([@bib31]). Additionally, meiotic germ cells strengthen the ovarian fate by antagonizing the testicular pathway or triggering the maturation of somatic cells in mice ([@bib20; @bib50]). Thus, based on our data, a higher number of PGCs would provide more meiotic oocytes or oocyte-derived signals in sexual differentiation ([@bib29]), further promoting or maintaining female fate as evidenced by the upregulation of *cyp19a1a*, *foxl2* and other profemale genes/pathways. Gonadal Transformation May Function as a Buffering System for Developmental Outcome {#sec3.1} ----------------------------------------------------------------------------------- It has been indicated that testicular differentiation exhibits variability in zebrafish and that all males develop "juvenile ovaries" before the initiation of testis formation ([@bib45]). The earliest reported upregulation of known male sex markers, including *amh*, *sox9a*, or *cyp11c1* (earlier *cyp11b* or *cyp11b2*), and downregulation of *cyp19a1a* (a well-known female marker) in the transforming zebrafish gonads was from 30 dpf onward ([@bib30; @bib44]). In line with these findings, we have not observed differences between the WT transforming gonads and PGC-depleted morphants at 22 dpf. On the other hand, several potential testis markers were identified. Reciprocal expression of *cyp19a1a* and *amh* during gonadal transformation might indicate transdifferentiation from granulosa cells to Sertoli cells in zebrafish ([@bib44]). Our data suggest that germ cells in PGC-depleted gonads were not able to activate/maintain *cyp19a1a* expression in somatic cells, and *amh* expression has not yet been upregulated. Further, histological analysis showed no oocyte or ovarian structures in PGC-depleted morphants at 23 dpf, and the gonads have embarked on male differentiation via the expression of some potential testicular markers and/or the downregulation of ovary associated genes. It has been suggested that gonadal development of zebrafish without PGCs progresses with similar timing as WT ([@bib37]). However, we observed different histological patterns in the PGC-depleted gonads at later stages, and PGC-depleted individuals coclustered with a subset of WT transforming males based on their 22 dpf transcriptomes, confirming the variability in testicular differentiation. Thus, we propose that gonadal transformation may function as a buffering process for stabilizing the developmental outcome of an ovary or a testis, complementing the "sexual canalization" concept championed by others ([@bib32]). A Germ Cell Number Model for Sexual Differentiation {#sec3.2} --------------------------------------------------- In this study, we provide evidence that dimorphic germ cell proliferation is the first sign of sexual differentiation, and the number of germ cells is important for stability of ovarian fate. We further demonstrate that the presence of a higher number of oocytes passing through meiosis (which may generate an increased amount of oocyte-derived signals) is important for maintenance of female development ([@bib29]). On the other hand, the development of the PGC-depleted gonads appears to be slower and often incomplete, likely causing further changes in the course of testicular differentiation. Our data suggest that PGC-depleted individuals may undergo a less pronounced "juvenile ovary" stage, differing in timing and length relative to their WT counterparts ([Figure 7](#fig7){ref-type="fig"}). In conclusion, our work suggests that germ cell number plays an active role during sexual differentiation and may directly regulate the progression of gonadal transformation in zebrafish. Experimental Procedures {#sec4} ======================= Ethics and Zebrafish Strains {#sec4.1} ---------------------------- The study was carried out in accordance with the Guide for the Care and Use of Laboratory Animals in Hokkaido University and Institutional Animal Care and Usage Committee at Temasek Life Sciences Laboratory. Zebrafish were kept under standard conditions and staged according to ([@bib10]). The zebrafish strains used for the experiments were AB WT, golden mutant, *Tg*(*vasa:DsRed2-vasa*); *Tg(bactin:EGFP)* double transgenic line ([@bib48]), and *Tg*(*vasa:vasa*-*EGFP*), also known as *zf45Tg* transgenic line ([@bib13]). Generation of Germ Cell-Deficient Zebrafish {#sec4.2} ------------------------------------------- Two different methods, MO injection and cell transplantation, were used to generate PGC-deficient zebrafish. Our preliminary tests with a control MO have shown that the microinjection itself does not exert a systematic effect on gonadal development ([Figure S2](#app3){ref-type="sec"}). Details are in the [Supplemental Experimental Procedures](#app3){ref-type="sec"}. Quantitation of PGC Number in Early Embryos {#sec4.3} ------------------------------------------- The suitability of determining PGC number by the squash preparation method was initially validated by comparing PGC counts obtained by this approach with those obtained by precise quantification of fluorescence intensity. Both techniques yielded numbers that were in good agreement (for details, see the [Supplemental Experimental Procedures](#app3){ref-type="sec"}). Confocal Microscopy {#sec4.4} ------------------- The isolated trunk region of a larva containing the prospective gonad without internal organs was embedded in 1.5% low melting agarose on a 0.17 mm coverslip and imaged on a Leica SP5 inverted confocal microscope (for details, see the [Supplemental Experimental Procedures](#app3){ref-type="sec"}). Quantitative Image Analysis of PGCs {#sec4.5} ----------------------------------- To determine the number of PGCs present in each sample, confocal z stacks were analyzed using Imaris 7.4 Spot Count function (Bitplane, Andor) using a minimum size of 5.0 μm, a positive threshold for channel 1 (excitation 488 nm) and a negative threshold for channel 2 (excitation 561 nm). The presence or absence of loose aggregates of cells or intact gonadal structures was scored visually in 3D using Imaris. Histology {#sec4.6} --------- The histological analysis was performed as described ([@bib45]) with modifications (for details, see the [Supplemental Experimental Procedures](#app3){ref-type="sec"}). Microarray-Based Transcriptome Profiling {#sec4.7} ---------------------------------------- A customized microarray was used for gene expression profiling. The microarray was designed based on various resources and manufactured by Roche NimbleGen on its 12 × 135 array format with each array containing 117,915 probes. Each probe is 60 bp and mapped to Ensembl 70 (released on January 2013). At both developmental stages, 10 samples were from PGC-depleted (1--9) morphants, and 14 samples were from randomly selected WT individuals. Collection and Preparation of RNA Samples {#sec4.8} ----------------------------------------- All samples used for RNA extraction were from the FI family, which produces progeny sex ratio of 40% male. PGC-depleted morphants and uninjected WT were grown to the desired stages (standard length of 4--5 mm and 6--7 mm for 14 dpf and 22 dpf, respectively). The developing trunk regions without internal organs, between the opercula and anal pores, were carefully dissected. The samples were immediately snap frozen in liquid nitrogen and then stored at −80°C. For detailed information on RNA sample preparation and processing, see the [Supplemental Experimental Procedures](#app3){ref-type="sec"}. Microarray Data Analysis {#sec4.9} ------------------------ Detailed information on microarray analysis can be found in the [Supplemental Experimental Procedures](#app3){ref-type="sec"}. Quantitative RT-PCR for Validation of Microarray Data {#sec4.10} ----------------------------------------------------- Quantitative RT-PCR was carried out using the BioMark HD system (Fluidigm). Trunk sections as opposed to isolated gonads were used, as experimental validation has proven that the presence of other tissues did not have a substantial effect on the gonadal expression level of most genes ([Table S7](#app3){ref-type="sec"}). For detailed information on quantitative RT-PCR (qRT-PCR) assay, see the [Supplemental Experimental Procedures](#app3){ref-type="sec"}. Author Contributions {#sec5} ==================== M.S.H., R.S., and L.O. conceived the project. K.-W.T., R.G., R.S., M.S.H., M.C., and L.O. conceived and designed experiments. K.-W.T., R.G., J.M.S., T.S., R.S., M.S.H., and M.C. performed the experiments. K.-W.T., R.G., J.M.S., R.S., M.C., K.A., E.Y., M.S.H., and L.O. analyzed and discussed the data. K.-W.T., R.G., M.C., and L.O. wrote the paper. Accession Numbers {#app1} ================= The microarray data have been deposited in the GEO database under accession number [GSE57046](ncbi-geo:GSE57046){#intref0010}. Supplemental Information {#app3} ======================== Document S1. Supplemental Experimental Procedures, Figures S1--S4, and Tables S1, S4, and S7Table S2. List of Gene Symbols and Gene Names Used in the Text and Figures, Related to Figure 4Table S3. Differentially Expressed Genes between WT and PGC-Depleted Morphants at 14 and 22 dpf, Related to Figure 4Table S5. Primers for qRT-PCR, Related to Figure 5Table S6. Differential Expression Level of a Selected Set of Genes in WT Individuals and PGC-Depleted Morphants Examined by Microarray and qRT-PCR at 14 and 22 dpf, Related to Figure 5Movie S1. The Experimental Procedures of Blastoderm Transplantation in ZebrafishThis movie was taken under a stereomicroscope. It shows the technique for inducing the BdT chimera during the early to midblastula stage. First, the upper half of blastoderm was cut and removed from the host embryo using a glass fiber stuck to the tip of a glass pipette. Then the whole donor blastoderm was cut and transplanted onto the host blastula and gently pushed from the top.Movie S2. Donor and Host Blastoderms Are Mixed during the Development following Bastoderm TransplantationThe movie was taken under a fluorescent microscope. It shows that a donor blastoderm with GFP fluorescence adhered to the cut surface of host blastoderm and healed. Subsequently, they formed an admixture and developed together as a chirmeric blastoderm after a few hours.Document S2. Article plus Supplemental Information We would like to thank Lisbeth Olsen and Paul Collodi for transgenic lines and Fiona Chia, Xianke Shi, and Daniel Koch for experimental support. Special thanks are given to Kellee Siegfried, Gerd Maack, and Helmut Segner for advice on histological sections, the members of the Orban lab for discussions, and Woei Chang Liew for constructive criticisms on the earlier versions of the manuscript. This work was supported by grants from the Temasek Life Sciences Laboratory and the Agri-Food and Veterinary Authority of Singapore (to L.O.), by PROBRAIN (Promotion of Basic Research Activities for Innovative Biosciences; to E.Y.), by a Grant-in-Aid for Young Scientists (B 18780140 to R.G.), and by the F3 Project Support Office for Female Researchers at Hokkaido University (to R.G.). This is an open access article under the CC BY-NC-ND license (<http://creativecommons.org/licenses/by-nc-nd/3.0/>). {#fig1} {ref-type="sec"} and [S2](#app3){ref-type="sec"}.](gr2){#fig2} {ref-type="sec"} for the survival rates following these manipulations and [Movies S1](#mmc6){ref-type="supplementary-material"} and [S2](#mmc7){ref-type="supplementary-material"} for BdT.](gr3){#fig3} {ref-type="supplementary-material"} for a list of gene symbols and names, [Table S3](#mmc3){ref-type="supplementary-material"} for a list of DE genes, and [Table S4](#app3){ref-type="sec"} for functional classification of DE genes.](gr4){#fig4} {ref-type="supplementary-material"} for primer sequences and [Table S6](#mmc5){ref-type="supplementary-material"} for detailed expression data.](gr5){#fig5} {ref-type="sec"} and [S4](#app3){ref-type="sec"}.](gr6){#fig6} {ref-type="fig"} of [@bib26] with permission.](gr7){#fig7} [^1]: Co-first author [^2]: Present address: Nishiura Station, South Ehime Fisheries Research Center, Ehime University, Uchidomari, Ainan, Ehime 798-4206, Japan [^3]: Present address: Molecular Genetics and Development Division, MIMR-PHI Institute of Medical Research, Clayton, VIC 3168, Australia [^4]: Present address: Department of Anatomy and Neuroscience, University of Melbourne, Parkville, VIC 3010, Australia [^5]: Present address: Research Institute of Fish Culture and Hydrobiology, Faculty of Fisheries and Protection of Waters, University of South Bohemia in České Budějovice, 389 25 Vodňany, Czech Republic [^6]: Present address: Biotech Division, Incepta Pharmaceuticals, Ltd., Dhaka 1208, Bangladesh

Introduction ============ Concurrent strength and endurance training is considered an optimal stimulus to promote both neuromuscular and cardiovascular gains ([@b4-jhk-44-171]; Izquierdo et al., 2001; [@b19-jhk-44-171]; [@b24-jhk-44-171]). However, some studies have shown that concurrent training may result in lower strength gains compared with strength training alone, and this phenomenon is called the "interference effect" ([@b3-jhk-44-171]; [@b14-jhk-44-171]; [@b17-jhk-44-171]). Regarding endurance adaptations, several studies have shown that there are no differences in the magnitude of cardiorespiratory adaptations when endurance training is performed alone or combined with strength training ([@b3-jhk-44-171]; [@b17-jhk-44-171]). However, [@b6-jhk-44-171] observed that performance of strength prior to endurance training resulted in lower enhancements in maximal endurance performed compared with the inverse order, which was justified as a consequence of fatigue resulting from strength training. Opposite results were observed by [@b4-jhk-44-171], who observed no influence of intra-session exercise order on maximal aerobic power in the elderly. Therefore, there is controversy regarding the effect of intra-session exercise order on endurance adaptations. Only few studies, however, have investigated the acute effects of strength training on subsequent endurance performance, and these studies have focused on the effects of strength training on oxidative metabolism during endurance exercise ([@b2-jhk-44-171]; [@b12-jhk-44-171]), or on the effects of inducing muscle damage using eccentric protocols on endurance performance ([@b16-jhk-44-171]; [@b22-jhk-44-171], [@b21-jhk-44-171]; [@b26-jhk-44-171]). However, the acute effects of traditional methods of strength training, including training focusing on muscle hypertrophy or muscle power enhancements, on subsequent endurance exercise have been poorly investigated. In addition, in previous studies investigating the physiological effects of strength training on subsequent endurance exercise, the rest intervals between different parts of training (i.e., strength and endurance) lasted, at least, one hour ([@b2-jhk-44-171]; [@b12-jhk-44-171]). Therefore, it would be interesting to assess acute physiological effects of strength training on the subsequent endurance exercise when the last is performed immediately after the former. This investigation could bring relevant information about the physiological responses during endurance exercise when this modality is performed after strength exercises, as well as regarding the possible influence of strength training on the heart rate during endurance exercise, because the heart rate is often used as an intensity parameter during endurance exercise. Therefore, the purpose of this study was to investigate the effects of two types of strength training sessions aimed at developing muscle hypertrophy and muscle power on subsequent neuromuscular and endurance performance. Our hypothesis was that both methods of strength training would impair endurance performance. In order to test our hypothesis, we chose to investigate the effects of two types of strength training on subsequent endurance performance: hypertrophic and plyometric strength training. The rationale for the use of these two training sessions was because first, strength training is often performed simultaneously with endurance training by practitioners looking for muscle hypertrophy and health promotion; and second, plyometric training is commonly performed by endurance athletes who aim to improve endurance performance ([@b20-jhk-44-171]). Material and Methods ==================== Experimental Design ------------------- Participants in the present study attended the laboratory on several occasions. On the first visit, anthropometric measurements were assessed, and a maximal incremental test was performed on a cycle ergometer to determine maximal oxygen uptake. On the second visit, maximal strength was assessed using the one repetition maximum test. In the last visits, different exercise sessions were performed randomly, with one week of rest between sessions. In two of them, a strength training protocol (i.e., hypertrophic or plyometric) was performed before the endurance exercise protocol and in another, the endurance exercise protocol was performed alone. During each exercise session, physiological markers such as oxygen uptake and the heart rate, as well as the time to exhaustion (TTE) were assessed and compared. In addition, maximal isometric voluntary contractions (MIVC), the rate of force development (RFD) and EMG parameters related to neuromuscular fatigue such as EMG amplitude, median frequency, and the Dimitrov's spectral fatigue index, were assessed before and after each strength training session. The ambient conditions were kept constant during all of the tests (temperature: 22--24° C). In addition, a food diary was completed two days before each training session, and the caloric intake on training days was controlled. These training sessions were performed at the same time in the morning (between 9 and 11 a.m.). All of the participants were experienced with strength and endurance exercises as well as with the maximal tests performed. Subjects -------- Thirteen healthy young and physically active men (age: 23.2 ± 1.6 years; body height: 176.4 ± 6.6 cm; body mass: 72.7 ± 7.1 kg; squat 1 RM = 112.2 ± 12.2 kg; VO~2max~ = 32.4 ± 4.7; n =13) volunteered for the study. The participants were used to performing three strength training sessions per week with the objective of promoting muscle hypertrophy, as well as to performing low intensity endurance training twice per week. The volunteers were carefully informed about the design of the study with information given regarding the possible risks and discomfort related to the procedures. After a verbal explanation of procedures, subjects signed an informed consent form. The study was approved by the Ethics Committee of the Federal University of Rio Grande do Sul and was performed in accordance with the declaration of Helsinki. Exclusion criteria included any history of neuromuscular, metabolic, hormonal and cardiovascular diseases. Subjects were not taking any medication with influence on neuromuscular metabolism. Peak oxygen uptake and ventilatory thresholds --------------------------------------------- To determinate the peak oxygen uptake (VO~2peak~), second ventilatory threshold (VT~2~) and workload at VT~2~ (W~VT2~) an incremental test on a cycle ergometer (Cybex®, USA) was performed. The test started at a 50 W load during two minutes and was progressively increased by 25 W each minute, until volitional exhaustion ([@b17-jhk-44-171]). The cycling cadence was maintained between 70--75 rpm. The expired gases were measured using a portable gas analyzer, model VO2000 (Medical Graphics®, Ann Arbor, EUA), and the data were recorded every 10 s. The VT~2~ was determined using the ventilation curve corresponding to the points of the exponential increase in ventilation in relation to the load and confirmed by ventilatory equivalent (VE/VCO~2~) ([@b27-jhk-44-171]). The load attained in the VT~2~ was accepted as W~LV2~. The peak VO~2~ obtained (ml·kg^−1^·min^−1^) - near exhaustion was considered the VO~2peak~. The heart rate (HR) was measured and registered every 10 s using a Polar monitor (Polar® model FS1, Shanghai, China). The test was considered valid if at least 2 of the 3 listed criteria were met: 1) the maximum heart rate predicted by age was reached (220-age); 2) the impossibility of continuing to pedal at a minimum velocity of 70 rpm; and 3) when RER greater than 1.1 was obtained. Maximal dynamic strength (1RM) ------------------------------ The 1RM was performed on the squat exercise using a multiforce machine (WORLD-Esculptor®, Porto Alegre, RS, Brasil) with 1 kg of resolution. On the test day, the subjects warmed up for five minutes on a cycle ergometer and performed specific movements for the exercise test. To avoid the influence of fatigue, each subject's maximal load was determined with no more than five attempts with a four-minute recovery between attempts. The time for each contraction (concentric and eccentric) was 2 s, controlled by an electronic metronome (Quartz®, CA, USA). The test-retest reliability coefficient (ICC) was 0.99 for the 1RM squat test. Maximal isometric strength -------------------------- To quantify maximal isometric strength, a load cell was fixed to multiforce machine (WORLD- Esculptor®, Porto Alegre, RS, Brasil) and connected to an A/D converter (Miotec®, Porto Alegre, Brazil). The subjects were positioned in the squat position at a knee angle of 90° (0° representing the full flexion). The subjects had two attempts at obtaining the maximum voluntary contraction (MIVC), each lasting 5 s, with a 1 min rest period between each attempt. Researchers provided verbal encouragement so that the subjects could exert possible maximum strength as fast as possible. This procedure was performed before and immediately after each strength training session (traditional and plyometric). MIVC was defined as the highest value of the force obtained during the 5 s contraction. The rate of force development (RFD) was determined from the slope of the force-time curve (Δforce/Δtime) of the isometric contractions over 10 ms time intervals. The test-retest reliability coefficient (ICC) was 0.94 for MVIC and 0.88 for RFD. Electromyography surface variables ---------------------------------- Surface electromyography (EMG) data were recorded throughout the experimental protocol from the vastus lateralis and rectus femoris using a four-channel electromyograph (Miotool®, Porto Alegre, Brazil), with a sampling frequency of 2000 Hz per channel, connected to a personal computer (Dell Vostro 1000®, São Paulo, Brazil). Shaving and abrasion with alcohol were carried out on the muscular belly to ensure low impedance (above 2000 Ω) at skin-electrodes interface ([@b9-jhk-44-171]). To ensure the same electrode position in subsequent tests, the right thigh of each subject was mapped for the position of the electrode moles and small angiomas by marking on transparent paper ([@b5-jhk-44-171]). According to SENIAM EMG recommendations, in the vastus lateralis electrodes were placed at 2/3 on the line from the anterior spina iliaca superior to the lateral side of the patella; and, in the rectus femoris the electrodes were placed at 50% on the line from the anterior spina iliaca superior to the superior part of the patella. The ground electrode was fixed on the anterior crest of the tibia. The raw EMG signal was acquired during the MVIC. EMG signals were analyzed using Matlab software® (R2012, MathWorks Inc., USA). The following parameters were obtained from the sEMG signals for evaluation of neuromuscular fatigue: the root mean squared value (RMS), median frequency (MNF), and the Dimitrov spectral parameter (FInsm5) ([@b9-jhk-44-171]). The RMS and median frequency were calculated in this study because these parameters are traditionally used to evaluate neuromuscular fatigue ([@b15-jhk-44-171]; [@b9-jhk-44-171]; [@b8-jhk-44-171]). Additionally, the Dimitrov fatigue index was determined, because it appeared to be more sensitive to changes in the sEMG power spectrum than the mean or median frequencies ([@b7-jhk-44-171]; [@b9-jhk-44-171]). This parameter is obtained as the ratio between different moment orders (−1 and 5) and emphasizes the increases in the low and ultralow frequencies of the sEMG spectrum due to increased negative after-potentials as well as the decreases in high frequencies due to increments in duration of intracellular action potentials and decrements in action potential propagation velocity. The test-retest reliability coefficient (ICC values) of all EMG measurements was over 0.85. Caloric intake and nutritional report ------------------------------------- The subjects were instructed not to change their eating habits and completed a nutritional record two days before each protocol. On test days, the subjects came to the laboratory in a fasting state and were supplemented with 1 g of maltodextrin per kilogram of body weight (g^−1^·kg^−1^) before each exercise session. The data collection procedures started one hour after the maltodextrin intake. In addition, a food diary was completed two days before each training session, and the caloric intake on training days was controlled. We assessed the nutritional status to ensure that the subjects started the different protocols with no differences on nutritional intake in the previous 48 hours to avoid possible influence of nutritional intake on exercise performance. Strength training protocols --------------------------- On the days of exercise protocol, a warm-up of 10 repetitions of the squat exercise with the bar only was performed before both the hypertrophic and plyometric protocols. The hypertrophic strength training protocol (HT) was performed with a squat exercise using a multiforce machine (WORLD- Esculptor®, Porto Alegre, RS, Brasil). The participants performed 6 sets of 8 repetitions at 75% of their 1RM with a 2 min interval between sets. The performance time was 2 s for each phase of motion (i.e., concentric and eccentric). The plyometric training protocol (PT) was composed of a countermovement jump exercise. The subjects were instructed to perform the eccentric and concentric phases at maximum speed possible, flexing the knee at an angle of approximately 90° before starting the concentric phase. Only body weight was used as the workload. The participants performed 6 sets of 8 repetitions of countermovement jumps with a 2 min rest interval between sets. Endurance exercise ------------------ The endurance exercise was performed on a cycle ergometer until volitional exhaustion. Initially, the subjects warmed up for 5 min at 25 W. Afterwards, the resistance was increased until the load corresponded to the second ventilatory threshold (W~VT2~) obtained during testing. During the protocol, a cadence between 70 and 75 rpm was maintained. Volitional exhaustion was considered when subjects were not able to maintain a cadence of 70 rpm or when they manifested the impossibility of continuing the task, so the protocol was suspended and the time of exhaustion recorded. In addition, the subjects were verbally encouraged during the protocol in order to motivate them to exert the maximal time of exercise as possible. During endurance exercise, VO~2~ and the HR were assessed and recorded in each 10 and 30 s of exercise, respectively, for further analysis. We calculated average VO~2~ and the average HR during the total time of exercise. Statistical analysis -------------------- The SPSS statistical software package was used to analyze all data. Normal distribution was checked with the Shapiro-Wilk test. Results are reported as mean ± SD. A comparison of the oxygen uptake, heart rate and time to exhaustion among different exercise protocols was performed using one-way ANOVA with repeated measures followed by Bonferroni post hoc analysis. The analysis of the neuromuscular parameters between strength training protocols (hypertrophy vs. power) in two moments (pre and post exercise) was performed using a two-way ANOVA (group × time). The retrospective statistical power provided by SPSS after analysis was over 0.90 in all variables. The level of significance was set at 0.05. Results ======= Nutritional status ------------------ The comparison of the nutritional status among the training sessions showed no significant differences for total energy (p =0.621), carbohydrates (p =0.869), proteins (p =0.369) and fats (p =0.448) ([Table 1](#t1-jhk-44-171){ref-type="table"}). Cardiorespiratory and endurance performance variables ----------------------------------------------------- There were no differences among the test days in VO~2~ and HR values before the start of endurance exercise. There were no significant differences among the protocols in mean VO~2~ values or mean HRs during the total time of exercise ([Table 2](#t2-jhk-44-171){ref-type="table"}). In addition, the TTE was significantly greater when endurance exercise was performed alone in comparison to both HST and PST protocols, with no differences between HST and PST ([Figure 1](#f1-jhk-44-171){ref-type="fig"} and [Table 2](#t2-jhk-44-171){ref-type="table"}). Neuromuscular variables ----------------------- There were significant decreases in the MIVC after HST and PST, with no difference between the ST protocols ([Table 3](#t3-jhk-44-171){ref-type="table"}). In addition, significant decreases were also observed in the maximum rate of force development after HST and PST, with no difference between the ST protocols. Moreover, there were no significant changes in RMS values, median frequency or the Dimitrov's spectral fatigue index in both the vastus lateralis and rectus femoris muscles after both ST protocols ([Table 3](#t3-jhk-44-171){ref-type="table"}). Discussion ========== The main finding of the present study was a decreased time to exhaustion on the cycle ergometer after both hypertrophic and plyometric training sessions, which may be related to the concomitant maximal strength and RFD decreases after the ST protocols. In addition, there were no changes in any of the neural activity parameters after the ST protocols or in the average oxygen uptake and heart rate during exercise. These findings expand the knowledge regarding the physiological responses and performance of endurance exercise during a concurrent training session and suggest that the HR may be used as an intensity parameter during endurance exercise after hypertrophic and plyometric training sessions. Moreover, hypertrophic ST or plyometric training sessions should be avoided immediately before endurance exercise if the aim of training is to maintain high-intensity endurance exercise for a prolonged time. Regarding the physiological responses, our results corroborate the results found by [@b2-jhk-44-171], who did not observe differences in VO2 or HR values during endurance exercise performed before and after a strength training session. In contrast, [@b12-jhk-44-171] observed that VO2 was greater when endurance exercise was performed after strength training compared with the performance of endurance exercise alone, which was justified as a consequence of exercise post oxygen consumption (EPOC) resulting from ST contributing to the metabolic demand of the endurance exercise. In the present study, it was also possible that EPOC from the ST protocols was not sufficient to result in the average VO2 during endurance exercise. The discrepancies among the present results and those from previous studies may be a consequence of different methods used because in the study by [@b12-jhk-44-171], the endurance exercise was performed only 5 min after strength training, instead of the 15 min interval used in the present study. Indeed, EPOC was greater in the first 5 min after the end of strength training and was thereafter markedly reduced in the following minutes of recovery (Almeida et al., 2001; [@b2-jhk-44-171]). In addition, in the study by [@b12-jhk-44-171], the strength training session was composed of upper- and lower-body exercises instead of only one lower-body exercise as was used in the present study (i.e., squat or countermovement jump), which may have resulted in a lower metabolic demand during strength training and, consequently, lower EPOC. Moreover, in the study by [@b12-jhk-44-171], aerobic intensity was 50% of VO2peak, while in the present study, the intensity corresponded to the second ventilatory threshold. Finally, although in the study by [@b12-jhk-44-171], the ST session resulted in greater VO2 during endurance exercise in women, the same result was not observed in men, which is in accordance with the present results. Thus, along with the strength training protocols used, different populations may also explain different results. Although no physiological differences were observed during different exercise protocols, there was a significant reduction in time to exhaustion when ST preceded endurance exercise (23% after HST and 17% after PST). Thus, at the same intensity, even with no VO2 changes, high-intensity endurance exercise was performed for a longer time (until exhaustion) when performed alone. It has been shown that after eccentric resistance exercise, there is a decrease in endurance performance as a result of muscle damage until 48 hours after the eccentric bout ([@b16-jhk-44-171]; [@b22-jhk-44-171]). Although the present study did not use a resistance exercise purely eccentric, the high workload performed in the HST protocol, and the characteristic of the PST protocol, which include high eccentric overload, may explain the impaired endurance performance when endurance exercise was performed after strength training. However, to the best of the authors' knowledge, this is the first study to compare the acute effects of performing different types of strength training on high-intensity endurance performance. Previously, [@b6-jhk-44-171] showed that strength training performed prior to high-intensity running (i.e., at VO2max velocity during 50% of pre-training time-to-exhaustion at this velocity) resulted in lower endurance performance than the inverse order. Interestingly, after training, the time to exhaustion in the group that performed strength training prior to endurance training was lower than in the inverse order (346 vs. 417 s) ([@b6-jhk-44-171]). Although we did not investigate the chronic effects of performing strength training prior to endurance exercise in the present study, one could suggest that after a training period, the lower time to exhaustion during endurance exercise that was observed after the HST and PST protocols could result in lowered cardiorespiratory fitness gains. However, our results should be interpreted with caution because endurance exercise was performed at high intensity (i.e., at a workload that corresponded to the second ventilatory threshold), and the time of endurance exercise was until failure. Thus, if the time of exercise were lower (i.e., pre-determined), it is possible that no differences would be detected among the protocols because the HRs and VO2 values were similar among the protocols. To identify possible neuromuscular mechanisms related to the impaired endurance performance after ST protocols, we assessed neuromuscular parameters such as MIVC, RFD, EMG amplitude, median frequency, and the Dimitrov's spectral fatigue index. In the EMG parameters, no changes were observed in the RMS, median frequency and Dimitrov's fatigue index values ([@b10-jhk-44-171]). Regarding EMG amplitude, studies have often shown increases in these values ([@b11-jhk-44-171]) under fatigue conditions, whereas decreases ([@b25-jhk-44-171]) and an absence of change ([@b15-jhk-44-171]; [@b18-jhk-44-171]) have also been observed. In the study by [@b15-jhk-44-171] investigating neuromuscular responses to explosive (40% of 1 RM) and heavy (70% of 1 RM) resistance loads, no changes were observed in EMG amplitude values, whereas only slight increases were observed in the median frequency during the explosive type protocol. In addition, it has been shown that 5 sets of 10 RM (repetitions until concentric failure) lead to significant decreases in median frequency and increases in the Dimitrov's fatigue spectral index ([@b9-jhk-44-171]), which was not observed in the present study. A possible explanation of these different results could be that changes in median frequency and the Dimitrov's fatigue index were observed during dynamic rather than isometric contractions in [@b15-jhk-44-171], in contrast to the assessments in the present study. In fact, recently it has been shown that changes in EMG parameters (amplitude, median frequency and Dimitrov's index) are greater during dynamic rather than isometric contractions when the fatigue protocol is performed dynamically ([@b8-jhk-44-171]). In contrast, there were marked decreases in the MIVC (15% and 17% after HST and PST, respectively) and RFD (23% and 18% after HST and PST, respectively), which is in agreement with previous studies investigating fatigue after explosive and hypertrophic strength training protocols ([@b9-jhk-44-171]; [@b11-jhk-44-171]; [@b15-jhk-44-171]; [@b18-jhk-44-171]). The MIVC and RFD are force outcomes strongly associated with endurance performance ([@b4-jhk-44-171]; [@b20-jhk-44-171]), which may help to explain the impaired endurance performance after ST protocols. The acute decreases in the MIVC and RFD could reflect type II fibre fatigue and, consequently, a lower contribution of this type of fibre during high-intensity endurance exercise performed in concurrent training protocols. Although we were not able to detect changes in the neural parameters assessed, other mechanisms related to the neuromuscular system, including changes in the neuromuscular junction properties, muscle conduction velocity, and reduced glycogen stores in type II fibres, may have induced the impairment in endurance performance after HST and PST. It should be noted that the subjects performed the exercise protocols at the same nutritional status, as shown by the dietary record, and with the same caloric intake before the exercise. Thus, the results do not appear to be influenced by nutritional factors. In summary, hypertrophic and plyometric ST sessions performed prior to endurance training may similarly impair the time to exhaustion during high-intensity endurance exercise. In addition, despite the acute endurance performance impairment, ST protocols did not influence average VO2 values or HRs during exercise. From a practical standpoint, the HR may be used as an intensity parameter during endurance exercise even when it is performed after hypertrophic and plyometric training sessions. However, if the purpose of training is to perform high-intensity endurance training for as long as possible (i.e., second ventilatory threshold), then both hypertrophic ST and plyometric training should be avoided immediately before endurance exercise. This work was supported by the CNPQ, CAPES and FAPERGS Government Associations. {#f1-jhk-44-171} ###### Values are mean ± SD of the alimentary register accomplished before protocols END HST PST --------------------- ---------------- ---------------- ---------------- Total Energy (kcal) 2173.4 ± 587.1 2177.6 ± 647.4 2036.4 ± 471.8 Carbohydrates (g) 270.3 ± 104.1 269.0 ± 107.8 276.5 ± 84.2 Proteins (g) 102.3 ± 31.8 101.9 ± 41.5 86.36 ± 21.8 Lipids (g) 74.9 ± 20.7 77.1 ± 27.7 66.77 ± 18.0 No significant differences. ###### Values are mean ± SD of cardiorespiratory responses and performance during endurance exercise END HST PST ----------------------- ------------- --------------------------------------------------------- --------------------------------------------------------- TTE (s) 1491 ± 399 1152 ± 372^[\*](#tfn3-jhk-44-171){ref-type="table-fn"}^ 1244 ± 421^[\*](#tfn3-jhk-44-171){ref-type="table-fn"}^ HR (beats·min^−1^) 164.8 ± 8.6 168.2 ± 5.1 168.1 ± 9.5 VO~2~ (ml·kg·min^−1^) 27.7 ± 3.05 29.8 ± 3.3 26.9 ± 4.2 TTE, time to exhaustion; HR, heart rate; VO~2~, oxygen uptake; END, endurance exercise alone; HST, hypertrophic strength training protocol + endurance exercise; PST, plyometric strength training protocol + endurance exercise. Significant difference from the endurance only protocol (p\<0.05). ###### Values are mean ± SD of neuromuscular responses before and after strength training protocols. HST PST ----------------------------- ----------------- ----------------------------------------------------------- ----------------- ------------------------------------------------------------ MVIC (kg) 98.2 ± 8.9 83.8 ± 8.2^[\*\*](#tfn5-jhk-44-171){ref-type="table-fn"}^ 111.15 ± 11.6 91.7 ± 11.5^[\*\*](#tfn5-jhk-44-171){ref-type="table-fn"}^ RFD (N·s^−1^) 548 ± 72.2 420.6 ± 46.5^[\*](#tfn6-jhk-44-171){ref-type="table-fn"}^ 448.2 ± 58.5 369.0 ± 40.1^[\*](#tfn6-jhk-44-171){ref-type="table-fn"}^ RMS VL (μV) 190.2 ± 81.4 175.9 ± 51.0 204.1 ± 84.6 182 ± 73.1 RMS RF (μV) 154.7 ± 76.0 146.88 ± 57.28 161.4 ± 63.5 149.18 ± 70.1 Fmedian VL (Hz) 85.4 ± 6.9 85 ± 6 88.3 ± 17.5 86.7 ± 20.2 Fmedian RF (Hz) 99.6 ± 14.1 101.7 ± 19 97.1 ± 11.8 98.2 ± 15.5 FInsm5 VL (Hz^−6^)(10^−10^) 0.0027 ± 0.0019 0.0022 ± 0.001 0.0019 ± 0.0007 0.0025 ± 0.0011 FInsm5 RF (Hz^−6^)(10^−10^) 0.0028 ± 0.0021 0.0017 ± 0.0007 0.0023 ± 0.001 0.0023 ± 0.0012 MVIC, maximal voluntary contraction; RFD rate of force development; RMS root mean squared; Fmedian, median frequency; Flnsm5, Dimitrov's spectral fatigue index, VL, vastus lateralis; RF, rectus femoris; HST, hypertrophic strength training protocol + endurance exercise; PST, plyometric strength training protocol + endurance exercise. Significantly different from pre ST protocol values (p\< 0.01). Significantly different from pre ST protocol values (p \<0.05). [^1]: Authors submitted their contribution to the article to the editorial board.

Introduction {#sec1} ============ Cancer is one of the most devastating diseases in the developing countries.^[@ref1]^ The epidermal growth factor receptor (EGFR) plays an important role in cell survival, growth, differentiation, and tumorigenesis. Dysregulation of EGFR is a common mechanism in cancer progression especially in nonsmall cell lung cancer (NSCLC). Also, overexpression of EGFR has been observed in different types of cancers such as breast, ovarian, head and neck, colon, and so forth.^[@ref2]^ Some FDA-approved drugs, EGFR inhibitors such as erlotinib^[@ref3]^ (i), gefitinib^[@ref4]^ (ii), icotinib^[@ref5]^ (iii), lapatinib^[@ref6]^ (iv), and afatinib^[@ref7]^ (v), are used for the treatment of the above-mentioned cancers ([Figure [1](#fig1){ref-type="fig"}](#fig1){ref-type="fig"}). The interplay of reactive oxygen species (ROS) and the EGFR plays an important role in cancer progression. Excessive ROS can induce negative responses such as growth inhibition or death of cancer cells. Mitochondrial dysfunction is also the major mechanism inducing oxidative stress. Higher ROS levels can trigger overoxidation of the Met residue of EGFR T790M and shut down the EGFR downstream survival pathway.^[@ref8]^ Therefore, direct EGFR inhibition or inhibition of EGFR function via excessive ROS generation or both may be a feasible therapeutic approach for cancer treatment. {#fig1} Side effects are major problems with the current EGFR inhibiting anticancer drugs. For example, erlotinib significantly reduced the levels of white blood cells, red blood cells (RBCs), and hemoglobin. It increased liver function markers, aspartate aminotransferase and alanine aminotransferase levels, and damaged the internal organs in an experimental rat model.^[@ref9]^ Similarly, unusual hematologic complications were detected after erlotinib was administered in patients with advanced NSCLC.^[@ref10]^ Therefore, it is important to design new EGFR inhibitors as anticancer agents with low toxicity on normal organs and blood cells. Quinazoline is an important heterocyclic moiety used in drug discovery because of its diverse biological activities.^[@ref11]^ Especially, 4-aminoquinazoline moiety showed good efficacy against various cancers. The structure--activity relationship (SAR) of EGFR inhibitors such as erlotinib and lapatinib revealed a quinazoline moiety to play an important role in antitumor activity, especially 4-aminoquinazoline moiety. 4-Aminoquinazoline moiety seemed particularly very important for activity and showed diverse biological activities such as anticancer,^[@ref12]^ antitubercular,^[@ref13]^ antimalarial,^[@ref14]^ antileishmanial,^[@ref15]^ and antibacterial and antifungal activities.^[@ref16]^ 1,2,3-Triazole is another important pharmacophore in medicinal chemistry and it can form hydrogen bonding with biological targets,^[@ref17]^ which will be useful for the activity. Also, triazole moiety-containing molecules (vi--ix, [Figure [1](#fig1){ref-type="fig"}](#fig1){ref-type="fig"}) are known to show various pharmacological activities such as anticancer,^[@ref18]^ anti-human immunodeficiency virus,^[@ref19]^ antitubercular,^[@ref20]^ and anti-inflammatory^[@ref21]^ activities. These structural features and importance in various biological activities have made this moiety very important in drug discovery. The fight against cancer and hence the research to cure the disease are continuing since last many years. Many novel therapeutics were tried, but most of them suffer from severe toxicities. In an ongoing project in our laboratory on the discovery of new anticancer agents,^[@ref22]^ we were interested to make EGFR inhibitors. Recently, molecular hybridization approach has been widely used for the design and synthesis of small hybrid compounds for the treatment of cancer. The approach mainly involves combining two or more different pharmacophore moieties in a single molecule having a common scaffold. These hybrid molecules have many advantages over the conventional drugs such as toxicity,^[@ref23]^ solubility, multidrug resistance, and so forth. In the present study, we are using the molecular hybridization strategy to combine the biologically important two scaffolds, quinazoline and 1,2,3-triazole, to get a small set of new hybrid compounds ([Figure [2](#fig2){ref-type="fig"}](#fig2){ref-type="fig"}). As discussed earlier, both quinazoline and 1,2,3-triazole moieties are very important for the anticancer activity, thus we synthesized 20 triazole-containing quinazoline hybrid compounds and performed cytotoxicity studies and the molecular docking studies thereafter. A lead compound was used to study EGFR inhibition, ROS generation, and toxicity in normal cells as well as in blood cells. {#fig2} Results and Discussion {#sec2} ====================== The synthesis was started by converting 4-nitro-benzylbromide **1** to the corresponding azide **2** in the presence of sodium azide in tetrahydrofuran (THF)--water in 95% yield. Compound **2** was reacted with different mono-substituted alkynes under classical "click" condition to produce different triazole compounds **3** in good yields. Finally, nitro group in **3** was reduced (Fe/NH~4~Cl in ethanol and water) to the corresponding amine **4**, which was further coupled with different 4-chloro quinazolines to give the desired target compounds (**5a--5t**) in 73--88% yield ([Scheme [1](#sch1){ref-type="scheme"}](#sch1){ref-type="scheme"}). All compounds were fully characterized by ^1^H nuclear magnetic resonance (NMR), ^13^C NMR, Fourier transform infrared (FT-IR), and high-resolution mass spectrometry (HRMS) data. {#sch1} The average 50% inhibitory concentration (IC~50~) values (concentration needed to inhibit cancer cell line proliferation by 50%) of the compounds (**5a--5t**) against four human cancer cell lines that include HCT116 (human colorectal cancer cell line), HEPG2 (human liver cancer cell line), MCF-7 (human breast cancer cell line), and PC-3 (human prostate cancer cell line) were determined using the cytotoxicity assay method. The IC~50~ values are listed in [Table [1](#tbl1){ref-type="other"}](#tbl1){ref-type="other"}, and the marketed anticancer drug erlotinib was used as positive control. On the basis of the cytotoxicity assay results, further efforts were made to elucidate the SAR. All of the synthesized compounds were screened against different cancer cell lines (HepG2, MCF-7, HCT116, and PC-3). Results suggested that most of the compounds showed moderate to good efficacy toward MCF-7 compared to other cell lines. Among all, compound **5b** showed the best activity with an IC~50~ value of 20.71 μM in MCF-7 cell lines. Moreover, it was observed that compounds bearing the −Cl atom at the second position of quinazoline (**5b**, **5e**, **5h**, **5k**, and **5n**) were better compared to other substitutions. Surprisingly, all of the above-mentioned active compounds containing −OMe group at the 6, 7 position of the quinazoline moiety are resulted, strongly suggesting that the chlorine atom at the 2nd position and −OMe group at the 6, 7 position of quinazoline are necessary for the activity. ###### Cytotoxicity Activities of Compounds **5a--5t** against HepG2, MCF-7, HCT116, and PC-3 in Micromolar {#gr7} The best compound **5b** has an IC~50~ value of 20.71 μM, which is higher than the standard compound erlotinib (11.57 μM). Interestingly, **5b** also shows lower toxicity than erlotinib on a normal human epithelial kidney cell line (40.32 ± 4.43 and 29.48 ± 3.32 μM), in human blood RBC (45.6 ± 2.65 and 16.23 ± 3.23 μM), and in human peripheral blood mononuclear cell (37.38 ± 3.55 and 20.46 ± 4.1 μM, respectively). Therefore, though compound **5b** shows relatively low toxicity than erlotinib in cancer cells, it has low toxicity in normal cell lines and normal blood cells. Thus, compound **5b** is expected to be a better drug candidate than erlotinib. Gratifyingly, compound **5b** also has 4-amino quinazoline moiety as the pharmacophore similar to that of erlotinib and gefitinib, and we thought that compound **5b** shows anticancer activity through the EGFR-mediated pathway. For this study, we selected the crystal structure (PDB ID: 1M17)^[@ref24]^ of EGFR tyrosine kinase for the molecular modeling study. To study the interaction between EGFR tyrosine kinase and triazole-substituted quinazoline hybrid derivatives, molecular docking of the EGFR protein (PDB ID: 1M17) and compound **5b** using discovery studio software was performed. Discovery studio was used for the visualization of interaction of the target molecule with EGFR tyrosine kinase. Docking results showed that compound **5b** goes and binds in the adenosine 5′-triphosphate (ATP)-binding pocket of 1M17 and showed two hydrogen bonding interactions with N1 of quinazoline with the main-chain NH group of Met 769 at a distance of 3.13 Å and triazole N2 with Lys 721 in the ATP-binding pocket at a distance of 3.04 Å. Both erlotinib and compound **5b** go and bind in the same ATP-binding pocket of EGFR tyrosine kinase (1M17, [Figure [3](#fig3){ref-type="fig"}](#fig3){ref-type="fig"}). {#fig3} Compound **5b** also showed other interactions such as halogen interaction with Gln 767, π--alkyl interaction with Leu 694, Leu 820, and Cys 751, and π--sulfur interaction with Met 742. Erlotinib was used as the reference compound, which also formed a similar H-bond with Met 769 at a distance of 2.85 Å. EGFR is an important signaling network in the case of cell proliferation, adhesion, migration, and survival. In order to study the mechanism of triazole-substituted quinazoline hybrids, we investigated the effect of compound **5b** on the EGFR signaling pathway in MCF-7 cell lines using western blot analysis.^[@ref25]^ After treatment with compound **5b** in MCF-7 cell lines, the level of EGFR and p-EGFR decreases with different time intervals of 12 and 24 h ([Figure [4](#fig4){ref-type="fig"}](#fig4){ref-type="fig"}). In this experiment, erlotinib was used as positive control. These results confirmed that the antiproliferative effect of compound **5b** in MCF-7 cell lines is mainly due to the decrease of EGFR and phosphorylation of EGFR and its downstream process. EGFR inhibition leads to the activation of ROS generation, which leads to DNA damage and cell death. Cellular generation of ROS is an important factor of apoptotic cell death. We have examined here the status of ROS generation by compound **5b**. The maximum ROS generation was observed at 24 h after the treatment of MCF-7 cells with compound **5b**. Also, flow cytometric analysis revealed that the fluorescein isothiocyanate (FITC) mean intensity was 765 in control cells but 1683 in compound **5b** (20.71 μM) treated cells after 24 h, indicating a shift in FITC mean intensity from the control cells to the cells treated with compound **5b**. These results demonstrated that compound **5b** (20.71 μM) inducing apoptosis in MCF-7 cells at 12 and 24 h might proceed via the ROS-mediated pathway.^[@ref26]^ {#fig4} Apoptosis is the desired way of cancer cell death. To determine the effect of compound **5b** on apoptosis, fluorescence-activated cell sorting (FACS) was performed on MCF-7 cell lines treated with compound **5b** at different time intervals.^[@ref27]^ At the initial stage of apoptosis, phosphatidyl serine is exposed from inside cell membrane to outside and this can bind with annexin V. After treatment of MCF-7 cells with compound **5b** (20.71 μM) at time points (0, 12, and 24 h), MCF-7 cells were stained with annexin V-FITC and propidium iodide (PI) and monitored by flow cytometry. It was observed that early apoptosis rates increased from 5.9 to 24.6% and the late apoptosis rates increased from 0.7 to 14.2%. The results showed that compound **5b** increased cellular apoptosis in a time-dependent manner ([Figure [5](#fig5){ref-type="fig"}](#fig5){ref-type="fig"}). Flow cytometric analysis of control cells revealed that 84.4% of the cell population exhibited fluorescence at the PE-Texas Red A channel, indicating a higher level of cells having a healthy ΔΨ~m~, whereas compound **5b** (20.71 μM) treated cells at 24 h revealed that 29.5% of the cell population exhibited fluorescence at the PE-Texas Red A channel, which showed a loss of ΔΨ~m~ in 54.9% of cell population after 24 h.^[@ref28]^ These results indicate that compound **5b** might induce apoptosis by generation of ROS via the mitochondrial pathway as it leads to lowering of mitochondrial membrane potential. {#fig5} Conclusions {#sec3} =========== In conclusion, a series of triazole-substituted quinazoline hybrid molecules were designed and synthesized as anticancer agents. The results showed that most of the compounds had moderate to good antiproliferative effects against four different cell lines HCT116, HepG2, PC-3, and MCF-7. Among them, compound **5b** showed good antiproliferative effects against MCF-7 cell lines. Molecular docking studies showed that compound **5b** formed hydrogen bond with Met 769 of the EGFR protein, which regulates the conformation of EGFR which is responsible for antiproliferative activity. Compound **5b** decreased the expression of EGFR and p-EGFR in MCF-7 cell lines, from which we can conclude that compound **5b** exhibits antiproliferative effects in MCF-7 cell lines through the EGFR signaling pathway. Compound **5b** also caused change in mitochondrial membrane potential and ROS-mediated apoptosis in MCF-7 cell lines. From the above results, it may be concluded that compound **5b** acts as the EGFR inhibitor and can be used in the treatment of cancer. Further optimization of the structure to show improved EGFR inhibitory and antiproliferative activity is ongoing in the laboratory. Experimental Section {#sec4} ==================== General Remarks {#sec4.1} --------------- All chemicals and reagents were purchased from Sigma-Aldrich. Column chromatography purifications were performed using silica gel grade 9385, 100--200 mesh, neutral alumina (Sigma). Thin-layer chromatography (TLC) was performed on silica gel 60 F254 plates with 0.20 mm thickness (Merck, Germany), which was visualized under an ultraviolet light chamber (254 and 365 nm). NMR experiments were run on Bruker Avance III 600 (600 MHz for ^1^H and 150 MHz for ^13^C). ^1^H NMR spectra were recorded for solution in CDCl~3~ or dimethyl sulfoxide (DMSO) with tetramethylsilane as the standard. Chemical shifts for ^1^H and ^13^C spectra were recorded in parts per million (ppm). Data were reported as follows: chemical shift (ppm), integrated intensity, multiplicity (indicated as s, singlet; d, doublet; t, triplet; q, quartet; m, multiplet), and coupling constants (*J*) in hertz (Hz). HRMS spectra were obtained on a JEOL MS station 700 (JEOL Ltd., Akishima-Shi, Japan). The FT-IR spectra of the samples were recorded on a JASCO FT-IR 4200 spectrometer (JASCO, Easton, MD, USA) using a KBr disk technique. The spectra were recorded from 400 to 4000 cm^--1^. JASCO software was used for data processing. Apoptosis, ROS, and JC1 assay were performed by FACSCalibur flow cytometry (Becton Dickinson, San Jose, CA, USA). Synthetic Procedure {#sec4.2} ------------------- ### General Procedure To Synthesize Compound **2** {#sec4.2.1} To the solution of compound **1** (1 equiv) in THF was added sodium azide (1.5 equiv) which was dissolved in water dropwise at 0 °C. The reaction mixture was then stirred for 1 h (0 °C to rt) and was monitored through TLC. After the complete consumption of the starting material, the solvent was evaporated. The crude mixture was then dissolved in EtOAc and washed with water and brine (3 × 10 mL). Thereafter, the organic layer was collected and dried over Na~2~SO~4~ and concentrated under reduced pressure to get yellow liquid in 95% yield. ### General Procedure To Synthesize Compound **3** {#sec4.2.2} To the solution of compound **2** (1 equiv) in THF were added different acetylene (1 equiv) and copper iodide (0.2 equiv), which was heated to reflux temperature (80 °C) for 3 h. After the total consumption of the starting material (monitored through TLC), the solvent was evaporated. The reaction mixture was then diluted with ethyl acetate, filtered through a Celite bed, and washed with water and brine (3 × 10 mL) in a separating funnel. Thereafter, the organic layer was collected and dried over Na~2~SO~4~ and then evaporated under reduced pressure to get a crude solid, which was purified by column chromatography (silica gel 100--200, per ether/ethyl acetate 4:1) to afford the desired product as yellow solid with 80--85% yield. ### General Procedure To Synthesize Compound **4** {#sec4.2.3} To the solution of compound **3** (1 equiv) in ethanol--water (1:1) were added iron powder (4 equiv) and ammonium chloride (10 equiv), which was heated to reflux temperature (80 °C) for 2 h. The reaction was monitored through TLC until total consumption of the starting material. After completion, ethanol was evaporated, and the crude mixture was diluted with ethyl acetate and passed through a Celite bed. Then, it was washed with water and brine (3 × 10 mL) in a separating funnel. Thereafter, the organic layer was collected and dried over Na~2~SO~4~ and evaporated under reduced pressure to get the crude solid. It was then purified by column chromatography (silica gel 100--200, per ether/ethyl acetate 1.5:1) to afford the desired product as yellow solid with 76--80% yield. ### General Procedure To Synthesize Compound **5** {#sec4.2.4} To the solution of compound **4** (1 equiv) in THF/water (4:1), different substituted 4-chloro quinazoline (0.9 equiv) and anhydrous sodium acetate (3 equiv) were added. The reaction mixture was heated to reflux temperature (80 °C) for 6 h and monitored through TLC. After total consumption of the starting material, the solvent was evaporated. The residue was then dissolved in EtOAc and washed with water and brine (3 × 10 mL). The organic layer was thereafter collected and dried over Na~2~SO~4~. It was then purified by column chromatography (neutral aluminum oxide, dichloromethane/methanol 49:1) to afford the desired product in 73--88% yield as a white powder. In similar manner, all other triazole-containing quinazoline hybrid compounds (**5a--5t**) were prepared. Characterization Data {#sec4.3} --------------------- ### 6,7-Dimethoxy-*N*-(4-((4-phenyl-1*H*-1,2,3-triazol-1-yl)methyl)phenyl)quinazolin-4-amine (**5a**) {#sec4.3.1} Yield 87% (32 mg); pale yellow solid; mp: 220--221 °C; ^1^H NMR (600 MHz, DMSO-*d*~6~): δ 9.53 (s, 1H), 8.65 (s, 1H), 8.44 (s, 1H), 7.86 (d, *J* = 7.8 Hz, 2H), 7.82 (m, 3H), 7.44 (t, *J* = 6 Hz, 2H), 7.40 (d, *J* = 8.4 Hz, 2H), 7.33 (t, *J* = 7.2 Hz, 1H), 7.18 (s, 1H), 5.63 (s, 2H), 3.95 (s, 3H), 3.92 (s, 3H). ^13^C NMR (150 MHz, DMSO-*d*~6~): δ (ppm) 156.7, 154.7, 153.2, 149.3, 147.4, 147.1, 139.9, 131.1, 129.3, 128.7, 128.3, 125.6, 122.9, 121.8, 109.3, 107.6, 100.3, 56.6, 56.2, 53.2. IR (KBr) \[cm^--1^\] ν: 3386, 3314, 3201, 3138, 3007, 1624, 1577, 1517, 1467, 1426, 1351, 1245, 1146, 1070.41, 993, 923, 855, 763, 691, 657. HRMS (ESI-*m*/*z*): calcd for C~25~H~22~N~6~O~2~, \[M + H\]^+^ 439.1822; found, 439.1892. ### 2-Chloro-6,7-dimethoxy-*N*-(4-((4-phenyl-1*H*-1,2,3-triazol-1-yl)methyl)phenyl)quinazolin-4-amine (**5b**) {#sec4.3.2} Yield 82% (37 mg); white solid; mp: 292--293 °C; ^1^H NMR (600 MHz, DMSO-*d*~6~): δ 9.90 (s, 1H), 8.68 (s, 1H), 7.86 (d, *J* = 7.2 Hz, 3H), 7.74 (d, *J* = 8.4 Hz, 2H), 7.43 (m, 4H), 7.33 (t, *J* = 7.2 Hz, 1H), 7.17 (s, 1H), 5.66 (s, 2H), 3.94 (s, 3H), 3.92 (s, 3H). ^13^C NMR (150 MHz, DMSO-*d*~6~): δ 157.9, 155.0, 154.2, 149.0, 148.2, 146.6, 138.4, 131.7, 130.7, 128.9, 128.3, 127.9, 125.2, 122.9, 121.5, 107.2, 106.6, 102.2, 56.3, 56.0, 52.7. IR (KBr) \[cm^--1^\] ν: 3361, 3138, 2988, 2948, 1621, 1572, 1515, 1458, 1426, 1240, 1150, 1001, 963, 842, 763, 697. HRMS \[EI-*m*/*z*\]: calcd for C~25~H~21~N~6~O~2~Cl, \[M\]^+^ 472.1415; found, 472.14136. ### *N*-(4-((4-Phenyl-1*H*-1,2,3-triazol-1-yl)methyl)phenyl)quinazolin-4-amine (**5c**) {#sec4.3.3} Yield 85% (30 mg); white solid; mp: 270--271 °C; ^1^H NMR (600 MHz,, DMSO-*d*~6~): δ 9.85 (s, 1H), 8.66 (s, 1H), 8.58 (s, 1H), 8.55 (d, *J* = 8.4 Hz, 1H), 7.87 (m, 5H), 7.79 (d, *J* = 8.4 Hz, 1H), 7.64 (t, *J* = 7.8 Hz, 1H), 7.42 (m, 4H), 7.33 (t, *J* = 7.2 Hz, 1H), 5.64 (s, 2H). ^13^C NMR (150 MHz, DMSO-*d*~6~): δ 158.1, 154.8, 150.1, 147.1, 139.6, 133.5, 131.5, 131.1, 129.3, 128.7, 128.3, 128.2, 126.7, 125.6, 123.4, 123.0, 121.8, 115.6, 53.2. IR (KBr) \[cm^--1^\] ν: 3258, 3135.49, 3087, 1611, 1570, 1524, 1501, 1415, 1357, 1224, 1074, 1043, 922.80, 766, 683, 511. HRMS \[EI-*m*/*z*\]: calcd for C~23~H~18~N~6~, \[M\]^+^ 378.1593; found, 378.15329. ### *N*-(4-((4-(4-Ethylphenyl)-1*H*-1,2,3-triazol-1-yl)methyl)phenyl)-6,7-dimethoxyquinazolin-4-amine (**5d**) {#sec4.3.4} Yield 88% (39 mg); white solid; mp: 230--231 °C; ^1^H NMR (600 MHz, CDCl~3~): δ 8.68 (s, 1H), 8.06 (d, *J* = 12 Hz, 1H) 7.71 (s, 1H), 7.68 (m, 4H), 7.43 (d, *J* = 6 Hz, 1H), 7.27 (s, 1H), 7.22 (m, 4H), 5.55 (s, 2H), 4.03 (s, 3H), 3.92 (d, *J* = 1.8 Hz, 3H), 2.65 (q, *J* = 7.8 Hz, 2H), 1.23 (t, *J* = 7.8 Hz, 3H). ^13^C NMR (150 MHz, CDCl~3~): δ 156.4, 154.7, 153.4, 149.6, 148.5, 147.5, 144.6, 139.3, 138.9, 129.9, 128.5, 128.4, 127.5, 125.6, 122.3, 119.6, 109.3, 107.7, 99.9, 56.3, 56.2, 53.8, 28.6, 15.4. IR (KBr) \[cm^--1^\] ν: 3379, 2956, 1623, 1577, 1516, 1462, 1423, 1394, 1240, 1140, 1061, 996, 840, 798, 540. HRMS (ESI-*m*/*z*): calcd for C~27~H~26~N~6~O~2~, \[M + H\]^+^ 466.2195; found, 467.2189. ### 2-Chloro-*N*-(4-((4-(4-ethylphenyl)-1*H*-1,2,3-triazol-1-yl)methyl)phenyl)-6,7-dimethoxyquinazolin-4-amine (**5e**) {#sec4.3.5} Yield 84% (34 mg); white solid; mp: 277--278 °C; ^1^H NMR (600 MHz, DMSO-*d*~6~): δ 9.89 (s, 1H), 8.61 (s, 1H), 7.85 (s, 1H), 7.76 (d, *J* = 8.4 Hz, 2H), 7.73 (d, *J* = 8.4 Hz, 2H), 7.42 (d, *J* = 8.4 Hz, 2H), 7.29 (d, *J* = 8.4 Hz, 2H), 7.16 (s, 1H), 5.64 (s, 2H), 3.94 (s, 3H), 3.91 (s, 3H), 2.61 (q, *J* = 7.8, 2H), 1.19 (t, *J* = 7.8, 3H). ^13^C NMR (150 MHz, DMSO-*d*~6~): δ 158.3, 155.4, 154.6, 149.5, 148.6, 147.2, 143.9, 138.9, 132.2, 128.7, 125.6, 123.4, 121.6, 107.7, 107.1, 102.6, 56.7, 56.4, 53.1, 28.3, 15.9. IR (KBr) \[cm^--1^\] ν: 3356, 2965, 1621, 1573, 1515, 1455, 1427, 1343, 1294, 1240, 1150, 1002, 962, 839, 800, 580, 530. HRMS \[EI-*m*/*z*\] calcd for C~27~H~25~N~6~O~2~Cl, \[M\]^+^ 500.1728; found, 500.17178. ### *N*-(4-((4-(4-Ethylphenyl)-1*H*-1,2,3-triazol-1-yl)methyl)phenyl)quinazolin-4-amine (**5f**) {#sec4.3.6} Yield 87% (28 mg); white solid; mp: 250--251 °C; ^1^H NMR (600 MHz,, DMSO-*d*~6~): δ 9.86 (s, 1H), 8.59 (d, *J* = 6 Hz, 2H), 8.54 (d, *J* = 12 Hz, 1H), 7.88 (d, *J* = 6 Hz, 2H), 7.85 (d, *J* = 6 Hz, 1H), 7.79 (d, *J* = 12 Hz, 1H), 7.76 (d, *J* = 12 Hz, 2H), 7.64 (t, *J* = 12 Hz, 1H), 7.41 (d, *J* = 6 Hz, 2H), 7.27 (d, *J* = 7.8 Hz, 2H), 5.62 (s, 2H), 2.61 (q, *J* = 7.8 Hz, 2H), 1.19 (t, *J* = 7.8 Hz, 3H). ^13^C NMR (150 MHz, DMSO-*d*~6~): δ 158.2, 154.8, 150.1, 147.2, 143.9, 139.5, 133.5, 131.6, 128.7, 128.7, 128.6, 128.2, 126.7, 125.6, 123.4, 123.0, 121.4, 115.6, 53.2, 28.3, 15.9. IR (KBr) \[cm^--1^\] ν: 3422, 3290, 3103, 2967, 2927, 1616, 1571, 1526, 1498, 1422, 1358, 1319, 1224, 1049, 924, 830, 770, 677. HRMS \[EI-*m*/*z*\] calcd for C~25~H~22~N~6~, \[M\]^+^ 406.1915; found, 406.19156. ### (1-(4-(6,7-Dimethoxyquinazolin-4-ylamino)benzyl)-1*H*-1,2,3-triazol-4-yl)methanol (**5g**) {#sec4.3.7} Yield 77% (29 mg); off-white solid; mp: 295--296 °C; ^1^H NMR (600 MHz,, DMSO-*d*~6~): δ 9.52 (s, 1H), 8.43 (s, 1H), 8.02 (s, 1H), 7.82 (s, 1H), 7.77 (d, *J* = 7.8 Hz, 2H), 7.36 (d, *J* = 8.4 Hz, 2H), 7.18 (s, 1H), 5.55 (s, 2H), 5.18 (t, *J* = 5.4 Hz, 1H), 4.51 (d, *J* = 6 Hz, 2H), 3.95 (s, 3H), 3.92 (s, 3H). ^13^C NMR (150 MHz, DMSO-*d*~6~): δ 156.7, 154.7, 153.2, 149.3, 148.7, 147.4, 131.3, 128.7, 123.1, 122.9, 109.3, 107.6, 102.3, 56.6, 56.2, 55.5, 52.9. IR (KBr) \[cm^--1^\] ν: 3315, 3146, 3004.33, 2928, 2834, 1623, 1579, 1517, 1466, 1425, 1246, 1140, 1050, 994, 924, 854, 779, 657, 557, 515. HRMS (ESI-*m*/*z*) calcd for C~20~H~20~N~6~O~3~, \[M + H\]^+^ 392.1675; found, 393.1675. ### (1-(4-(2-Chloro-6,7-dimethoxyquinazolin-4-ylamino)benzyl)-1*H*-1,2,3-triazol-4-yl)methanol (**5h**) {#sec4.3.8} Yield 74% (24 mg); light yellow solid; mp: 265--266 °C; ^1^H NMR (600 MHz,, DMSO-*d*~6~): δ 9.89 (s, 1H), 8.04 (s, 1H), 7.85 (s, 1H), 7.70 (d, *J* = 8.4 Hz, 2H), 7.38 (d, *J* = 8.4 Hz, 2H), 7.17 (s, 1H), 5.57 (s, 2H), 5.20 (s, 1H), 4.51 (s, 2H), 3.94 (s, 3H), 3.92 (s, 3H). ^13^C NMR (150 MHz, DMSO-*d*~6~): δ 158.4, 155.4, 154.6, 149.5, 148.8, 148.6, 138.7, 132.4, 128.8, 123.4, 107.06, 107.04, 102.6, 56.7, 56.4, 55.5, 52.8. IR (KBr) \[cm^--1^\] ν: 3390, 3282, 3037, 3007, 2930, 1604, 1576, 1511, 1465, 1420, 1309, 1241, 1136, 1057, 992, 849, 781, 751, 726, 695. HRMS \[EI-*m*/*z*\] calcd for C~20~H~19~N~6~O~3~Cl, \[M\]^+^ 426.1207; found, 426.1216. ### (1-(4-(Quinazolin-4-ylamino)benzyl)-1*H*-1,2,3-triazol-4-yl)methanol (**5i**) {#sec4.3.9} Yield 73% (21 mg); white solid; mp: 268--269 °C; ^1^H NMR (600 MHz, DMSO-*d*~6~): δ 9.84 (s, 1H), 8.58 (s, 1H), 8.54 (d, *J* = 8.4 Hz, 1H), 8.02 (s, 1H), 7.85 (d, *J* = 8.4 Hz, 3H), 7.79 (d, *J* = 8.4 Hz, 1H), 7.64 (t, *J* = 8.4 Hz, 1H), 7.37 (d, *J* = 8.4 Hz, 2H), 5.56 (s, 2H), 5.16 (t, *J* = 6 Hz, 1H), 4.51 (d, *J* = 6.6 Hz, 2 H). ^13^C NMR (150 MHz, DMSO-*d*~6~): δ 158.2, 154.8, 150.1, 148.7, 139.4, 133.5, 131.79, 128.7, 128.2, 126.7, 123.4, 123.1, 123, 115.5, 55.5, 52.8. IR (KBr) \[cm^--1^\] ν: 3433, 3308, 3129, 2923, 1604, 1576, 1532, 1426, 1402, 1364, 1325, 1229, 1128, 1049, 93, 770. HRMS \[EI-*m*/*z*\] calcd for C~18~H~16~N~6~O, \[M\]^+^ 332.1386; found, 332.13748. ### 6,7-Dimethoxy-*N*-(4-((4-(phenoxymethyl)-1*H*-1,2,3-triazol-1-yl)methyl)phenyl)quinazolin-4-amine (**5j**) {#sec4.3.10} Yield 87% (33 mg); white solid; mp: 224--225 °C; ^1^H NMR (600 MHz, DMSO-*d*~6~): δ 9.52 (s, 1H), 8.44 (s, 1H), 8.29 (s, 1H), 7.83 (s, 1H), 7.79 (d, *J* = 7.8 Hz, 2H), 7.36 (d, *J* = 7.8 Hz, 2H), 7.29 (t, *J* = 7.8 Hz, 2H), 7.18 (s, 1H), 7.02 (d, *J* = 8.4 Hz, 2H), 6.94 (t, *J* = 7.2 Hz, 1H), 5.59 (s, 2H), 5.13 (s, 2H), 3.95 (s, 3H), 3.92 (s, 3H). ^13^C NMR (150 MHz, DMSO-*d*~6~): δ 158.4, 156.7, 154.7, 153.2, 149.3, 147.4, 143.5, 139.9, 131.1, 129.9, 128.7, 124.9, 122.8, 121.2, 115.1, 109.3, 107.6, 102.2, 61.4, 56.6, 56.2, 53.0. IR (KBr) \[cm^--1^\] ν: 3375, 1621, 1575, 1514, 1461, 1421, 1237, 1137, 997, 853, 758. HRMS \[EI-*m*/*z*\] calcd for C~26~H~24~N~6~O~3~, \[M\]^+^ 468.1910; found, 468.1891. ### 2-Chloro-6,7-dimethoxy-*N*-(4-((4-(phenoxymethyl)-1*H*-1,2,3-triazol-1-yl)methyl)phenyl)quinazolin-4-amine (**5k**) {#sec4.3.11} Yield 84% (29 mg); light yellow solid; mp: 260--261 °C; ^1^H NMR (600 MHz, DMSO-*d*~6~): δ 9.88 (s, 1H), 8.32 (s, 1H), 7.86 (s, 1H), 7.72 (d, *J* = 8.4 Hz, 2H), 7.39 (d, *J* = 8.4 Hz, 2H), 7.29 (t, *J* = 7.2 Hz, 2H), 7.17 (s, 1H), 7.03 (d, *J* = 7.8 Hz, 1H), 6.95 (t, *J* = 7.2 Hz, 2H), 5.62 (s, 2H), 5.14 (s, 2H), 3.95 (s, 3H), 3.92 (s, 3H). ^13^C NMR (150 MHz, DMSO-*d*~6~): δ 158.4, 158.3, 155.4, 154.6, 149.4, 148.6, 143.5, 138.8, 132.1, 129.9, 128.8, 125, 123.3, 121.2, 115.1, 107.6, 107.1, 102.6, 61.4, 56.7, 56.4, 52.9. IR (KBr) \[cm^--1^\] ν: 3364, 2922, 2853, 1601, 1573, 1515, 1458, 1427, 1293, 1240, 1151, 1004, 961, 840, 758. HRMS (ESI-*m*/*z*) calcd for C~26~H~23~N~6~O~3~Cl, \[M + H\]^+^ 503.1598; found, 503.1599. ### *N*-(4-((4-(Phenoxymethyl)-1*H*-1,2,3-triazol-1-yl)methyl)phenyl)quinazolin-4-amine (**5l**) {#sec4.3.12} Yield 85% (32 mg); white solid; mp: 208--210 °C; ^1^H NMR (600 MHz, DMSO-*d*~6~): δ 9.85 (s, 1H), 8.59 (s, 1H), 8.55 (d, *J* = 8.4 Hz, 1H), 8.29 (s, 1H), 7.87 (m, *J* = 6, 3H), 7.79 (d, *J* = 8.4 Hz, 1H), 7.64 (t, *J* = 13.2 Hz, 1H), 7.37 (d, *J* = 8.4 Hz, 2H), 7.28 (t, *J* = 7.2 Hz, 2H), 7.03 (d, *J* = 8.4 Hz, 1H), 6.94 (t, *J* = 7.8 Hz, 2H), 5.60 (s, 2H), 5.13 (s, 2H). ^13^C NMR (150 MHz, DMSO-*d*~6~): δ 158.4, 158.1, 154.8, 150.1, 143.5, 139.5, 133.52, 131.53, 129.9, 128.8, 128.2, 126.7, 124.9, 123.4, 123, 121.2, 115.5, 115.1, 61.2, 53. IR (KBr) \[cm^--1^\] ν: 3355, 3126, 2922, 1606, 1572, 1531, 1498, 1416, 1311, 1218, 1122, 1041, 828, 776, 683, 506. HRMS \[EI-*m*/*z*\] calcd for C~24~H~20~N~6~O, \[M\]^+^ 408.1703; found, 408.1703. ### (6,7-Dimethoxy-quinazolin-4-yl)-\[4-(4-phenylaminomethyl-\[1,2,3\]triazol-1-ylmethyl)-phenyl\]-amine (**5m**) {#sec4.3.13} Yield 85% (31 mg); light yellow solid; mp: 261--262 °C; ^1^H NMR (600 MHz, DMSO-*d*~6~): δ 9.51 (s, 1H), 8.43 (s, 1H), 8.01 (s, 1H), 7.82 (s, 1H), 7.77 (d, *J* = 8.4 Hz, 2H), 7.33 (d, *J* = 8.4 Hz, 2H), 7.18 (s, 1H), 7.06 (t, *J* = 7.8 Hz, 2H), 6.62 (t, *J* = 7.8 Hz, 2H), 6.53 (t, *J* = 12 Hz, 1H), 6.02 (t, *J* = 6 Hz, 1H), 5.54 (s, 2H), 4.27 (d, *J* = 6 Hz, 2H), 3.95 (s, 3H), 3.92 (s, 3H). ^13^C NMR (150 MHz, DMSO-*d*~6~): δ 156.7, 154.7, 153.2, 149.3, 148.8, 147.4, 146.5, 139.7, 131.2, 129.2, 128.6, 123.1, 122.8, 116.4, 112.7, 109.3, 107.6, 102.2, 56.6, 56.2, 55.3, 52.8, 39. IR (KBr) \[cm^--1^\] ν: 3390, 3282, 3037, 3007, 2930, 1604, 1576, 1511, 1465, 1420, 1309, 1241, 1136, 1057, 992, 849, 781, 751, 726, 695. HRMS (ESI-*m*/*z*) calcd for C~26~H~25~N~7~O~2~, \[M + H\]^+^ 468.2148; found, 468.2150. ### (2-Chloro-6,7-dimethoxy-quinazolin-4-yl)-\[4-(4-phenylaminomethyl-\[1,2,3\]triazol-1-ylmethyl)-phenyl\]-amine (**5n**) {#sec4.3.14} Yield 82% (29 mg); light yellow solid; mp: 280--281 °C; ^1^H NMR (600 MHz, DMSO-*d*~6~): δ 9.87 (s, 1H), 8.03 (s, 1H), 7.85 (s, 1H), 7.69 (d, *J* = 6 Hz, 2H), 7.35 (d, *J* = 9 Hz, 2H), 7.17 (s, 1H), 7.06 (t, *J* = 8.4 Hz, 2H), 6.62 (d, *J* = 9.6 Hz, 2H), 6.53 (t, *J* = 7.2 Hz, 1H), 6.03 (t, *J* = 6 Hz, 1H), 5.57 (s, 2 H), 4.28 (d, *J* = 6.6 Hz, 2H), 3.94 (s, 3H), 3.92 (s, 3H). ^13^C NMR (150 MHz, DMSO-*d*~6~): δ 158.3, 155.5, 154.6, 149.4, 148.8, 148.6, 146.5, 138.7, 132.4, 129.2, 128.7, 123.4, 123.2, 116.4, 112.7, 107.6, 107.1, 102.6, 56.7, 56.4, 52.7, 39. IR (KBr) \[cm^--1^\] ν: 3406, 1603, 1572, 1513, 1426, 1241, 1150, 961, 844, 755. HRMS (ESI-*m*/*z*) calcd for C~26~H~24~N~7~O~2~Cl, \[M + Na\]^+^ 524.1578; found, 524.1586. ### \[4-(4-Phenylaminomethyl-\[1,2,3\]triazol-1-ylmethyl)-phenyl\]-quinazolin-4-yl-amine (**5o**) {#sec4.3.15} Yield 81% (29 mg); light yellow solid; mp: 250--251 °C; ^1^H NMR (600 MHz, DMSO-*d*~6~): δ 9.83 (s, 1H), 8.58 (s, 1H), 8.54 (d, *J* = 8.4 Hz, 1H), 8.01 (s, 1H), 7.86 (m, 3H), 7.79 (d, *J* = 8.4 Hz, 1H), 7.64 (t, *J* = 8.4 Hz, 1H), 7.34 (d, *J* = 8.4 Hz, 2H), 7.06 (t, *J* = 8.4 Hz, 2H), 6.62 (d, *J* = 7.8 Hz, 2H), 6.53 (t, *J* = 7.2 Hz, 1H), 6.02 (t, *J* = 6 Hz, 1H), 5.55 (s, 2H), 4.28 (d, *J* = 6 Hz, 2H). ^13^C NMR (150 MHz, DMSO-*d*~6~): δ 158.1, 154.8, 150.1, 148.8, 146.5, 139.4, 133.5131.7, 129.2, 128.7, 128.2, 126.7, 123.4, 123.1, 123, 116.4, 115.5, 112.7, 52.8, 39. IR (KBr) \[cm^--1^\] ν: 3422, 3289, 2923, 1604, 1571, 1524, 1498, 1414, 1357, 1316, 1256, 1123, 1051, 924, 773, 687, 510. HRMS \[EI-*m*/*z*\] calcd for C~24~H~21~N~7~, \[M\]^+^ 407.1858; found, 407.1855. ### (2-Chloro-quinazolin-4-yl)-\[4-(4-phenyl-\[1,2,3\]triazol-1-ylmethyl)-phenyl\]-amine (**5p**) {#sec4.3.16} Yield 86% (39 mg); white solid; mp: 204--206 °C; ^1^H NMR: (600 MHz, DMSO-*d*~6~): δ 10.25 (s, 1H), 8.68 (s, 1H), 8.55 (d, *J* = 8.4 Hz, 1H), 7.86 (m, 3H), 7.79 (d, *J* = 8.4 Hz, 2H), 7.71 (d, *J* = 8.4 Hz, 1H), 7.65 (t, *J* = 7.2 Hz, 1H), 7.43 (m, 4 H), 7.32 (t, *J* = 7.8 Hz, 1H), 5.65 (s, 2H). ^13^C NMR (150 MHz, DMSO-*d*~6~): δ 159.8, 156.6, 151.3, 147.1, 138.6, 134.6, 132.6, 131.1, 129.4, 128.8, 128.4, 127.4, 127.2, 125.6, 123.9, 123.6, 122, 114.2, 53.1. IR (KBr) \[cm^--1^\] ν: 3311, 2922, 2854, 1620, 1564, 1516, 1423, 1342, 1222, 1187, 1077, 1049, 948, 753. HRMS \[ESI-*m*/*z*\] calcd for C~23~H~17~N~6~Cl, \[M + H\]^+^ 413.1281; found, 472.1286. ### (2-Chloro-quinazolin-4-yl)-{4-\[4-(4-ethyl-phenyl)-\[1,2,3\]triazol-1-ylmethyl\]-phenyl}-amine (**5q**) {#sec4.3.17} Yield 85% (36 mg); white solid; mp: 205--207 °C; ^1^H NMR (600 MHz, DMSO-*d*~6~): δ 10.24 (s, 1H), 8.61 (s, 1H), 8.55 (d, *J* = 8.4 Hz, 1H), 7.88 (t, *J* = 8.4 Hz, 1H), 7.79 (d, *J* = 8.4 Hz, 2H), 7.76 (d, *J* = 8.4 Hz, 2H), 7.71 (d, *J* = 7.8 Hz, 1H), 7.66 (t, *J* = 8.4 Hz, 1H), 7.42 (d, *J* = 8.4 Hz, 2H), 7.26 (d, *J* = 7.8 Hz, 2H), 5.64 (s, 2H), 2.61 (q, *J* = 7.8 Hz, 2H), 1.18 (t, *J* = 7.8 Hz, 3H). ^13^C NMR (150 MHz, DMSO-*d*~6~): δ 159.8, 156.6, 151.3, 147.2, 143.9, 138.5, 134.7, 132.7, 128.8, 128.7, 128.6, 127.4, 127.1, 125.7, 123.9, 123.6, 121.6, 114.2, 53.1, 28.4, 16. IR (KBr) \[cm^--1^\] ν: 3379, 2964, 2924, 1623, 1605, 1562, 1528, 1497, 1424, 1345, 1290, 1220, 1195, 1078, 951, 839, 763. HRMS \[ESI-*m*/*z*\] calcd for C~25~H~21~N~6~Cl, \[M + H\]^+^ 441.1594; found, 441.1627. ### {1-\[4-(2-Chloro-quinazolin-4-ylamino)-benzyl\]-1*H*-\[1,2,3\]triazol-4-yl}-methanol (**5r**) {#sec4.3.18} Yield 83% (34 mg); white solid; mp: 224--226 °C; ^1^H NMR (600 MHz, DMSO-*d*~6~): δ 10.24 (s, 1H), 8.54 (d, *J* = 8.4 Hz, 1H), 8.04 (s, 1H), 7.88 (t, *J* = 7.2 Hz, 1H), 7.76 (d, *J* = 8.4 Hz, 2H), 7.71 (d, *J* = 8.4 Hz, 1H) 7.64 (t, *J* = 7.2 Hz, 1H), 7.38 (d, *J* = 8.4 Hz, 2H), 5.57 (s, 2H), 5.17 (t, *J* = 6 Hz, 1H), 4.5 (d, *J* = 4.8 Hz, 2 H). ^13^C NMR (150 MHz, DMSO-*d*~6~): δ 159.8, 156.6, 151.3, 148.8, 138.5, 134.6, 132.9, 128.8, 127.4, 127.2, 123.9, 123.6, 123.3, 114.2, 55.5, 52.8. IR (KBr) \[cm^--1^\] ν: 3334, 2925, 1607, 1566, 1527, 1424, 1410, 1288, 1200, 1126, 1057, 953, 859, 766. HRMS \[ESI-*m*/*z*\] calcd for C~28~H~15~N~6~OCl, \[M + H\]^+^ 367.1074; found, 367.1162. ### (2-Chloro-quinazolin-4-yl)-\[4-(4-phenoxymethyl-\[1,2,3\]triazol-1-ylmethyl)-phenyl\]-amine (**5s**) {#sec4.3.19} Yield 85% (36 mg); white solid; mp: 200--202 °C; ^1^H NMR (600 MHz, DMSO-*d*~6~): δ 10.25 (s, 1H), 8.56 (d, *J* = 8.4 Hz, 1H), 8.31 (s, 1H), 7.88 (t, *J* = 7.2 Hz, 1H), 7.77 (d, *J* = 8.4 Hz, 2H), 7.71 (d, *J* = 8.4 Hz, 1H), 7.65 (t, *J* = 8.4 Hz, 1H), 7.38 (d, *J* = 8.4 Hz, 2H), 7.29 (t, *J* = 8.4 Hz, 2H), 7.02 (d, *J* = 7.8 Hz, 2H), 6.93 (t, *J* = 7.2 Hz, 1H), 5.62 (s, 2H), 5.12 (s, 2 H). ^13^C NMR (150 MHz, DMSO-*d*~6~): δ 159.8, 158.5, 156.6, 151.3, 143.5, 138.5, 134.7, 132.7, 129.98, 128.9, 127.4, 127.2, 125.1, 123.9, 123.6, 121.3, 115.1, 114.2, 61.4, 52.9. IR (KBr) \[cm^--1^\] ν: 3336, 2922, 2853, 1621, 1565, 1529, 1516, 1426, 1410, 1366, 1237, 1216, 1187, 1058, 1006, 853, 760. HRMS \[ESI-*m*/*z*\] calcd for C~24~H~19~N~6~OCl, \[M + H\]^+^ 443.1387; found, 443.1450. ### (2-Chloro-quinazolin-4-yl)-\[4-(4-phenylaminomethyl-\[1,2,3\]triazol-1-ylmethyl)-phenyl\]-amine (**5t**) {#sec4.3.20} Yield 85% (36 mg); white solid; mp: 212--214 °C; ^1^H NMR (300 MHz, DMSO-*d*~6~): δ 10.23 (s, 1H), 8.54 (d, *J* = 8.1 Hz, 1H), 8.03 (s, 1H), 7.89 (t, *J* = 7.2 Hz, 1H), 7.74 (d, *J* = 8.4 Hz, 2H), 7.71 (d, *J* = 8.4 Hz, 1H) 7.64 (t, *J* = 8.4 Hz, 1H), 7.35 (d, *J* = 9 Hz, 2H), 7.05 (t, *J* = 7.2 Hz, 2H), 6.61 (d, *J* = 7.8 Hz, 2H), 6.52 (t, *J* = 7.2 Hz, 1H), 6.04 (t, *J* = 6 Hz, 1H), 5.56 (s, 2H), 4.27 (d, *J* = 6 Hz, 2H). ^13^C NMR (150 MHz, DMSO-*d*~6~): δ 159.8, 156.6, 151.3, 148.8, 146.5, 138.4, 134.7, 132.9, 129.3, 128.8, 127.4, 127.2, 123.9, 123.6, 123.3, 116.5, 114.2, 112.8, 52.8, 31.2. IR (KBr) \[cm^--1^\] ν: 3366, 2922, 2850, 1621, 1604, 1561, 1451, 1409, 1366, 1287, 1195, 1079, 1056, 957, 839, 765. HRMS \[ESI-*m*/*z*\] calcd for C~24~H~20~N~7~Cl, \[M + H\]^+^ 442.1547; found, 441.1564. General Experimental Procedures for Biological Studies {#sec4.4} ------------------------------------------------------ ### Cytotoxicity Assay {#sec4.4.1} Exponentially growing cells were seeded into 96-well plates at a concentration of 1 × 10^5^ cells per well. After 24 h incubation at 37 °C, the cells were treated with all compounds in different concentrations in triplicates. The cells were incubated for another 24 h. A 20 μL of 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide solution (5 mg/mL) was added to all wells and incubated for 4 h at 37 °C. The solution was discarded, and the insoluble dark blue crystals (formazan) were dissolved in 40 μL DMSO. After 15 min, the absorbance was measured using an ELISA reader at a wavelength of 595 nm. The IC~50~ of all compounds is the concentration (μg/mL) of the compound at which there was 50% growth inhibition with respect to the control culture in different cell lines. ### Analysis of Cellular apoptosis {#sec4.4.2} Apoptosis was assayed by using an annexin-V FITC apoptosis detection kit (Calbiochem, CA, USA). Briefly, cells were treated with or without the compounds in a time-dependent manner (0, 12, and 24 h) and then washed and stained with PI and annexin-V-FITC in accordance with the manufacturer's instructions. The percentages of live, apoptotic, and necrotic cells were determined by the flow cytometric method (Becton Dickinson, San Jose, CA, USA). Data from 1 × 10^6^ cells were analyzed for each sample. ### Measurement of ROS {#sec4.4.3} ROS generation was measured by dichlorofluorescein diacetate (DCF-DA). After treatment with the compounds for the time periods (0, 12, and 24 h), the cells were incubated with 10 μM DCF-DA at 37 °C for 20 min. Then cell pellets were suspended in 1 mL phosphate-buffered saline (PBS). Samples were analyzed at an excitation wavelength of 480 nm and an emission wavelength of 525 nm by FACSCalibur flow cytometry (Becton Dickinson, San Jose, CA, USA). ### Measurement of Mitochondrial Membrane Potential (JC1 Assay) {#sec4.4.4} MCF-7 cells were seeded for 24 h and then treated with and without compounds in a time-dependent manner (0, 12, and 24 h). Cells were washed twice with ice-cold PBS and then incubated with JC-1 dye (5 μg/mL) in darkness for 20 min at room temperature (37 °C). Emission was determined by the flow cytometric method (Becton Dickinson, San Jose, CA, USA) at 525 nm. ### Western Blot {#sec4.4.5} Treated or untreated cells were collected and lysed. Lysates (30 μg of protein per well) were separated by electrophoresis in 12% sodium dodecyl sulfate polyacrylamide gel and electrotransferred to poly(vinylidene difluoride) membranes using a Trans-Blot system (Trans Blot wet transfer; Bio-Rad Laboratories, Inc., Hercules, CA, USA). The membranes were blocked with 5% bovine serum albumin in TBST (tris buffered saline containing 0.1% Tween-20, pH 7.6) for 1 h at room temperature and then incubated with primary antibodies overnight at 4 °C. After washing with TBST, membranes were incubated with horseradish peroxidase conjugated secondary antibody for 1 h. Immunoreactive bands were visualized by chemiluminescence using tetramethylbenzidine as the substrate. β-Actin was used as the loading control. The Supporting Information is available free of charge on the [ACS Publications website](http://pubs.acs.org) at DOI: [10.1021/acsomega.8b01960](http://pubs.acs.org/doi/abs/10.1021/acsomega.8b01960).^1^H NMR and ^13^C NMR spectra of all compounds and crystal data of compound **5c** ([PDF](http://pubs.acs.org/doi/suppl/10.1021/acsomega.8b01960/suppl_file/ao8b01960_si_001.pdf))Crystallographic data of compound **5c** ([CIF](http://pubs.acs.org/doi/suppl/10.1021/acsomega.8b01960/suppl_file/ao8b01960_si_002.cif)) Supplementary Material ====================== ###### ao8b01960_si_001.pdf ###### ao8b01960_si_002.cif The authors declare no competing financial interest. The authors want to thank CSIR for funding in this project and for a senior research fellowship to K.C. We also thank Dr. Arindam Talukdar, IICB for allowing us to utilize his molecular docking felicities using Discovery studio. The authors also want to thank Mr. Sunil Kumar Killi for his critical comments during the manuscript preparations. Thanks are due to Mr. Sandip Kundu, CSIR-IICB, for recording X-ray data, Mr. E.K. Padmanaban for recording the NMR spectra, Soumik Laha for recording the IR spectra, and Sandip Chowdhury for recording the EI HRMS spectra.

1. Introduction {#sec1} =============== Family members\' caregiving responsibilities for patients with cancer have increased dramatically over the past decade and will likely continue to rise as care is now routinely administered on an outpatient basis and patients are living longer \[[@B1]\]. Understanding the psychological factors that contribute to morbidity among these "hidden patients" \[[@B2]\] may help improve the clinical assessment process and mitigate some of the adverse consequences of caregiving. Perceived health is an important predictor of morbidity and mortality across various cultural contexts \[[@B3]--[@B7]\]. Subjective rating scales are often used to assess perceived health and are straightforward and quick to administer in busy clinical settings. Given the widespread use of these rating scales and their constituent items, it is important to identify influences on responses. In this study, we sought to examine whether personality is associated with perceived health in spouses of patients with lung cancer. Although the term "caregiving" was initially used to refer to the provision of dementia care by family members, it is applicable to cancer care as well. Family members are not merely bystanders in treatment but also actual or potential cousers of services \[[@B8]--[@B10]\]. Family caregivers frequently accompany the patient to appointments, administer medications, and provide other critical day-to-day functions. For patients with lung cancer, where the 5-year survival rate is only 17% \[[@B11]\], spouses often provide in-home care. Recognizing that many caregivers derive psychological benefit from the provision of care \[[@B10], [@B12], [@B13]\], up to 30% of cancer caregivers experience significant psychological distress \[[@B10], [@B14]--[@B16]\]. Caregivers of patients diagnosed with lung cancer report more depressive symptoms than other caregivers \[[@B17]\]. Research also suggests that caregivers\' physical health diminishes over time \[[@B18]\]. Meta-analyses have found that caregivers have poorer physical health \[[@B18], [@B19]\], which may result from self-neglect due to caregiving demands and the belief that responding to the health needs of one\'s spouse is more important than attending to one\'s own. In particular, caregivers may implicitly compare their health needs with those of their sick spouses \[[@B20]\], leading to a response shift \[[@B21], [@B22]\] in how they evaluate their own health. This shift might lead caregivers to care for their ill family members at the expense of their own health because they are "healthier" \[[@B20]\]. *Personality and Perceived Health*. Individual differences in how spouses respond to caregiving demands are now well documented \[[@B10], [@B20], [@B23], [@B24]\], and several studies have examined the personality correlates of perceived health \[[@B25]--[@B29]\]. To the best of our knowledge, only one study \[[@B2]\] has examined the personality correlates of health-related quality of life in the caregiving context, and that study did not examine specific perceived health items. Moreover, that study did not control for objective illness burden, so it was unclear whether personality was related to biased health perceptions or merely illness burden. In the current study, we controlled for objective illness burden \[[@B30]\] while examining caregivers\' responses to four self-report perceived health items from the SF-36 \[[@B31]\]. These items were as follows: "I am as healthy as anybody I know," "My health is excellent," "I seem to get sick a little easier than other people," and "I expect my health to get worse." Based on previous research from Chapman and colleagues \[[@B26]\], we derived two sets of hypotheses. The first concerns Neuroticism, a personality dimension involving negative affectivity and emotional instability, which has been associated with more health complaints in general samples \[[@B32], [@B33]\]. We hypothesized that caregiver Neuroticism would be associated with reports of poorer perceived health on all four items, above and beyond the effects of covariates. The second hypothesis concerns Extraversion, a personality dimension marked by positive affect and sociability. Individuals high in Extraversion tend to have positive views about the future \[[@B34], [@B35]\] and higher levels of emotional well-being, which may lead to more optimistic health assessments \[[@B33], [@B36]\]. People who are high in Extraversion may underestimate their symptoms and report overly positive health \[[@B37]\]. Consistent with findings from Chapman et al. \[[@B26]\] we hypothesized that Extraversion would be associated with the item tapping future expectations ("I expect my health to get worse") but not with the other three items. 2. Methods {#sec2} ========== 2.1. Participants and Procedures {#sec2.1} -------------------------------- The study was approved by the IRB at the University of Rochester. The sample consisted of spouses of patients with lung cancer who completed cross-sectional measures as part of a larger cohort study conducted in the Rochester, New York, region \[[@B14], [@B38]\]. As a nonintervention study, the parent investigation examined the natural course of adjustment to the diagnosis and treatment of lung cancer. Spouses of patients who had been diagnosed with and treated for lung cancer were eligible to participate. A member of the research team identified eligible spouses, through introduction by the surgeon or oncologist. The research team member explained the nature of the study and invited the spouse to participate. Of the 340 spouses of lung cancer patients who were approached, 159 (47%) provided written consent for participation, and complete data were available for 114 (72% of those who consented). Informed consent was obtained at the time of the research interview; spouse participants also agreed to allow the research team to review their primary care medical chart. Researcher training sessions were conducted throughout the study to monitor rater drift and ensure the methodological integrity of the data collection process. Trained research assistants administered self-report questionnaires and observer-rated assessment instruments. 2.2. Measures {#sec2.2} ------------- Personality was assessed with the NEO-Five Factor Inventory (NEO-FFI) \[[@B39]\]. From the NEO-FFI, we selected two specific indicators of personality: Neuroticism and Extraversion. Because it is consistently associated with both health behaviors and objective health indicators \[[@B40]--[@B42]\] we controlled for the personality dimension conscientiousness. Internal consistency in this study was good for Neuroticism (Cronbach\'s *α* = 0.84) and conscientiousness (*α* = 0.83), while being lower for Extraversion (*α* = 0.63). The NEO inventories have been used in prior research on spouses of chronically ill patients \[[@B23], [@B42], [@B43]\] and their validity has been well documented \[[@B39]\]. The Cumulative Illness Rating Scale (CIRS) \[[@B44]\] was used to ensure that any relationships between personality and subjective health were not confounded by objective illness burden. The CIRS is a validated \[[@B30], [@B45]\] physician-rated objective health index derived by means of patient history as well as physical examination and laboratory findings and quantifies the amount of physical disease in each organ system at the time of study entry. A physician-researcher reviewed the spouses\' medical charts. Higher scores indicate greater disease burden. Depressive symptoms were controlled for by using the 24-item observer-rated Hamilton Depression Rating Scale (HDRS) \[[@B46]\], which assesses the presence and severity of depressive symptoms over the past week. Symptoms measured on the HDRS include depressed mood, feelings of worthlessness, helplessness, hopelessness, loss of interest in pleasurable activities, fatigue, and somatic complaints. Higher scores reflect greater depressive symptoms. Scores were based on self-report and nonverbal presentation. In the present study, the HDRS had excellent internal consistency (Cronbach\'s *α* = 0.81). To measure subjective health, we dichotomized four items from the General Health Perceptions Scale of the SF-36 \[[@B31]\]. Following Chapman and colleagues \[[@B26]\], we used two positively valenced items ("I am as healthy as anybody I know" and "My health is excellent") and two negatively valenced questions ("I seem to get sick a little easier than other people" and "I expect my health to get worse"). Respondents who endorsed "neutral," "definitely true," or "mostly true" in response to the positive items and "definitely false" and "mostly false" to the negative items were categorized as reporting good perceived health. Respondents who endorsed "neutral," "definitely true," or "mostly true" in response to the negative items and "definitely false" and "mostly false" to the positive items were categorized as having poor perceived health. We dichotomized the item scores because binary scores have more practical utility for health care practitioners, but we also conducted sensitivity analyses on ordinal scores. 2.3. Data Analysis {#sec2.3} ------------------ For each of the outcome variables, we conducted a logistic regression model \[[@B47]\]. Predictors included Neuroticism and Extraversion. Given that perceived health is influenced by sociodemographic factors \[[@B48]\], objective health \[[@B33]\], and depressive symptoms \[[@B25], [@B26]\], the analyses controlled for age, gender, conscientiousness, CIRS, and HDRS. We also ran the regression models using ordinal logits for perceived health with the following ordered categories: good, neutral, and poor. 3. Results {#sec3} ========== 3.1. Participant Characteristics {#sec3.1} -------------------------------- Descriptive statistics are summarized in [Table 1](#tab1){ref-type="table"}. Caregivers had a mean age of 63 and were predominantly female (71%), Caucasian (97%), and living with patients (98%). Their mean level of education was 13 years, and the median income bracket was \$25,000--34,999 (USD). Patients were distributed among cancer stages, and most of the patients (79%) had completed some form of treatment at the time of the spouse\'s participation, with surgery being the most common form of treatment, followed by combination therapy (surgery, radiation, and chemotherapy). [Table 2](#tab2){ref-type="table"} shows the mean Neuroticism and Extraversion scores of those who endorsed good versus poor perceived health for each of the four SF-36 questions. The means and standard deviations for Neuroticism and Extraversion were comparable to those observed in large samples of healthy US adults \[[@B49]\]. 3.2. Hypothesis Testing {#sec3.2} ----------------------- [Table 3](#tab3){ref-type="table"} shows the significant predictors in the multiple regression analysis. Higher Neuroticism was significantly associated with poor perceived health measured by the item "I seem to get sick a little easier than other people" (*p* \< 0.05). Higher Neuroticism was also a significant predictor for the endorsement of "I expect my health to get worse" (*p* \< 0.05). Spouses who reported being more extraverted were less likely to endorse the item "I expect my health to get worse" (*p* \< 0.05). Among the control variables, the CIRS objective illness burden scores (*p* \< 0.01) and age (*p* \< 0.05) were associated negatively with "My health is excellent." Older respondents were more likely to endorse "I expect my health to get worse" compared to younger respondents (*p* \< 0.05). Conscientiousness and HDRS (depressive symptoms) were not significantly associated with responses to any of the perceived health items. Results did not significantly change when the analyses were conducted using ordinal logits. 4. Discussion {#sec4} ============= The purpose of this study was to test hypotheses about the relationship between two personality dimensions---Neuroticism and Extraversion---and responses to individual items tapping perceived health. Support for our hypotheses about Neuroticism was mixed. Neuroticism was found to be significantly associated with poor perceived health on the negatively worded items ("I seem to get sick a little easier than other people" and "I expect my health to get worse"), but not on the positively worded items ("My health is excellent" and "I am as healthy as anybody I know"). Findings for the negatively worded items are consistent with previous research \[[@B26], [@B32], [@B50]\]. Whereas Chapman and colleagues \[[@B26]\] reported significant associations between Neuroticism and the positively valenced items, there are at least four possible explanations for our failure to do so \[[@B51]\]. First, responses to the positively valenced items "My health is excellent" and "I am as healthy as anybody I know" may implicitly require caregiver respondents to make a judgment about their health in comparison to their ill spouse. A second and related point is that very few people reported poor perceived health on the item "I am as healthy as anybody I know," reducing statistical power. Third, the Chapman et al. sample was older, and some of Neuroticism\'s effects on perceived health may not become observable until later adulthood \[[@B25], [@B52]\]. Finally, Neuroticism may confer a greater sensitivity to the presence of negative affect and negatively valenced items as opposed to the absence of positive affect \[[@B53]\]. This hypothesis could be examined in future clinical research. If it is supported, it would underscore for clinicians and researchers the importance of language, word choice, and item framing \[[@B54]\]. As hypothesized, people with higher Extraversion scores are less likely to endorse the item "I expect my health to get worse," controlling for physician assessment of actual health status based on medical records. Extraversion may be particularly relevant in the caregiving context because extraverted people tend to be optimistic and feel comfortable in the presence of others. People who are higher in Extraversion tend to have a more positive outlook on life. A factor termed "optimistic control," characterized by optimistic expectation for life outcomes, positive self-esteem, hope, and internal control, is positively correlated with Extraversion \[[@B34]\]. Additionally, positive affect has been shown to underlie thoughts about the future more than negative affect \[[@B53], [@B55], [@B56]\], which may impact how individuals higher in Extraversion judge the possibility of future health declines. Assuming that middle-aged and older adults can expect their health to deteriorate over time and that caregivers are at increased risk for such deterioration, caregivers lower in Extraversion may make more accurate judgments about their health. In contrast, those higher in Extraversion may overlook important signs and symptoms of disease and fail to report these to a physician. Although a positive outlook on life has many physical and mental health benefits \[[@B57]\], some studies suggest that unrealistic optimism about the future in the face of vulnerability to health issues may undermine specific risk reduction behaviors \[[@B58], [@B59]\]. Extraverted individuals may be identifiable in the consulting room \[[@B60]\]. For example, extraverts often dominate conversations, speak loudly, are gregarious, and are physically animated and enthusiastic \[[@B61]\]. Health care providers are advised to attend to more subtle cues for physical illness among extraverted caregivers and regard with empathy and skeptical curiosity the extravert\'s appraisals of their own health. Patient-reported outcomes (PROs) are increasingly being integrated into clinical research, care, accreditation standards, and reimbursement rates \[[@B62]--[@B65]\], so studies evaluating the psychometrics of commonly used measures, such as the SF scales, are timely. We found, unsurprisingly, that control variables of objective illness burden and age were associated with perceived health, which underscores the importance of controlling for these variables in research aiming to understand biased health perceptions. One prior study has linked personality to perceived health in a caregiver sample \[[@B2]\], and ours is the first to do so while controlling for objective illness burden. If personality can shape responses to self-reported health items, independent of objective illness burden, clearly more research is needed aimed at understanding psychosocial factors that may bias scores on self-reports of health status and other PROs, such as patient satisfaction with care. Several study strengths and weaknesses should be noted. This is the first study of which we are aware to examine personality and perceived health in a caregiving context while controlling for objective illness burden. Additional strengths include rigorous controls for other potential confounders like depression and careful analysis of specific SF-36 questions, a level of detail seldom pursued, despite the fact that the wording of the questions varies greatly and that clinical interviews typically assess health through a few such individual questions, rather than a formal composite score. The main limitation is the correlational design. Causal inferences cannot be drawn. Also, the internal consistency reliability of the Extraversion measure was lower than desirable, so the observed effects may have underestimated the association between Extraversion and perceived health. Finally, although levels of Neuroticism and Extraversion in this sample were comparable to those reported in national samples \[[@B49]\] and we have personality data on more than 70 percent of the cohort, generalizability to the entire cohort of consenting participants cannot be guaranteed. The goal of the present investigation was to examine the association between personality and perceived health in caregivers of patients with lung cancer. An important next step would be to extend these findings by examining moderators of the association, such as demographic and health characteristics. For example, we did not examine health behaviors, such as smoking, drinking, and drug use, and it is possible that the association between Extraversion and optimistic reports of health is stronger among individuals who avoid these behaviors and consequently experience fewer daily reminders of ill health (e.g., coughing and hangovers). Additionally, the association between Neuroticism and poorer perceived health could be stronger for caregivers who have fewer economic resources, whose care receiver has a worse prognosis, or who spend more time providing care. In closing, the present findings suggest that personality is associated with how spouses of cancer patients think about their health. Given that perceived health has prognostic implications for a variety of health outcomes and the mounting evidence for the role of personality in health and longevity \[[@B66], [@B67]\], it is important to explore ways to assess personality in clinical settings in order to target and tailor efforts to modify potentially inflated or deflated misperceptions of one\'s own health. Understanding the factors that contribute to perceived health threats among caregivers can help prevent the commonly observed negative effects of caregiving. This information is of value to the health care providers who care for cancer patients or their spouses. This work was supported by the Roadmap Scholars Award from the Louisiana Clinical and Translational Science (LA CaTS) Center (U54GM104940) from the National Institutes of General Medical Sciences, as well as K07MH01135, R25MH074898, and K24MH072712 from the National Institute of Mental Health. Competing Interests =================== None of the authors has any competing interests in conducting or reporting this research. ###### Descriptive statistics. Variable M (SD) or *N* (%) ----------------------- ------------------- Age, years 63.4 (9.9) Gender, female 81 (71.1%) Education, years 13.0 (2.1) Income    \<\$10,000 2 (1.8%)  \$10,000--24,999 24 (21.1%)  \$25,000--34,999 31 (27.2%)  \$35,000--49,999 31 (27.2%)  \$50,000 or higher 26 (22.8%) Race, Caucasian 111 (97.3%) Residing with patient 112 (98.2%) Cancer stage    I 54 (47.4%)  II 14 (12.3%)  III 28 (24.6%)  IV 18 (15.8%) Cancer treatments    Surgery only 50 (43.9%)  Radiation only 5 (4.4%)  Chemotherapy only 3 (2.6%)  Combination therapy 32 (28.1%)  No treatments 24 (21.1%) *Note*. *N* = 114. Percentages may not sum to 100.0% due to rounding. ###### Personality dimensions associated with good and poor perceived health. ---------------------------------------------------------------------------------------------------------- SF-36 item Perceived health Neuroticism\ Extraversion\ M (SD) M (SD) -------------------------------------------------------- ------------------ -------------- --------------- "I seem to get sick a little easier than other people" Good (*n* = 107) 16.8 (5.39) 28.1 (5.34) Poor (*n* = 5) 20.0 (5.57) 27.2 (6.14) "I am as healthy as anybody I know" Good (*n* = 88) 16.9 (5.23) 28.4 (4.73) Poor (*n* = 24) 16.9 (6.19) 27.1 (7.23) "I expect my health to get worse" Good (*n* = 99) 17.1 (5.37) 28.3 (5.44) Poor (*n* = 13) 15.6 (5.80) 26.2 (4.36) "My health is excellent" Good (*n* = 89) 16.3 (5.14) 28.7 (4.76) Poor (*n* = 23) 19.1 (6.00) 25.9 (6.86) ---------------------------------------------------------------------------------------------------------- *Note*. Perceived health responses were dichotomized. ###### Personality dimensions and covariates associated with poor perceived health. SF-36 item Significant predictor OR (95% CI) *p* -------------------------------------------------------- ----------------------- ------------------- ------ "I seem to get sick a little easier than other people" Neuroticism 1.11 (1.00--1.23) 0.05 "I am as healthy as anybody I know" ---     "I expect my health to get worse" Age 1.06 (1.01--1.13) 0.04 Neuroticism 1.12 (1.02--1.23) 0.02 Extraversion 0.90 (0.80--0.99) 0.05 "My health is excellent" Age 0.94 (0.88--0.99) 0.05 CIRS 1.51 (1.17--2.02) 0.003 *Note*. Categorical regression models are reported. Regression model using ordinal logits did not significantly change results. Covariates were age, gender, conscientiousness, Cumulative Illness Rating Scale (CIRS), and Hamilton Depression Rating Scale (HDRS). [^1]: Academic Editor: Elke Bromberg

Introduction ============ Peste des Petits Ruminants (PPR) is a viral disease of small ruminants, which is manifested by fever, ocular-nasal discharges, anorexia, necrotic stomatitis, fetid diarrhea, enteritis, and bronchopneumonia followed by either death or recovery \[[@ref1]\]. The 4 to 24 months aged goats are more susceptible in the endemic areas \[[@ref2]\]. The causal agent is PPR virus (PPRV), an envelope, a pleomorphic particle containing single-stranded RNA, approximately 16-kb long with negative polarity genome \[[@ref3]\]. The genome of the virus codes for six structural (N, P, M, F, H, and L) and two nonstructural (C and V) proteins \[[@ref4],[@ref5]\]. After the first report of PPR from Ivory Coast, the disease has been reported in the Middle East, the Arabian Peninsula, and most parts of Africa and Asia \[[@ref1]\]. Among the South Asian subcontinent, PPRV under lineage IV was first reported in the southern part of India during 1987 \[[@ref6]\]. Subsequently, frequent outbreaks of PPR have been recorded in South Asian countries like Pakistan, Bhutan, Nepal, Afghanistan, India, and Bangladesh \[[@ref7]\]. Total goat population of Bangladesh is approximately 19.43 million \[[@ref8]\]. In Bangladesh, the PPRV was first identified in 1993 during a severe PPR outbreak \[[@ref2]\]. Now a day, the outbreak of PPR in goats occurs every year and considered as an endemic disease in Bangladesh \[[@ref2]\]. The country faces a huge economic loss due to PPR outbreaks. High morbidity (100%) and mortality (50% to 90%) rates in goats caused by PPRV have been reported in Bangladesh \[[@ref9]\]. Das et al. \[[@ref10]\] also reported that PPR outbreaks caused 74.13% morbidity and 54.83% mortality in Black Bengal goats. For the prevention and control of PPR, proper diagnosis is crucial. Agar gel immunodiffusion test, counter immunoelectrophoresis, indirect fluorescent antibody test, virus neutralization test, virus-specific monoclonal antibodies in an immunocapture enzyme-linked immunosorbent assay (ELISA), sandwich ELISA, and PCR techniques are generally used for identifying the PPR \[[@ref4],[@ref11]\]. But more specifically, PPRV antibodies can be identified by competitive ELISA(c-ELISA) and serum neutralization test \[[@ref12]\]. Due to its circulating in nature, the PPRV can easily be spread from one place to another. Sometimes genetic mutation occurs in its genomic structure \[[@ref13]\]. For that reason, a locally produced vaccine can be a good choice to protect the animals susceptible to PPR viral infection. Thus, in Bangladesh, controlling this disease outbreak in small ruminants, regular seromonitoring, and molecular study are needed to evaluate the vaccine efficacy of the locally produced vaccine and genomic changes of PPRV in different regions of Bangladesh. Few studies in Bangladesh were previously reported on seromonitoring of antibodies against PPRV and molecular characterization of PPRV; for example, in the districts of Dhaka, Rajshahi, Gazipur, Netrokona, Narayanganj, and Mymensingh, there have been reported PPR outbreak investigation and seromonitoring activities \[[@ref9],[@ref13]--[@ref16]\]. However, seromonitoring of PPR in char region ([Fig. 1](#figure1){ref-type="fig"}) of Mymensingh Sadar (besides Brahmaputra River) has not done previously. Jessore region, a south-western district ([Fig. 1](#figure1){ref-type="fig"}) of Bangladesh, is yet to take under consideration for the investigation of genetic variability. This study was designed for the detection of PPRV-specific antibodies in serum of goats at the char area of Mymensingh by using monoclonal antibody-based c-ELISA and to characterize the circulating PPRV from Jessore region of Bangladesh. {#figure1} Materials and Methods ===================== Ethical approval ---------------- The experiment was approved by the Animal Welfare and Experimental Ethical Committee (AWEEC) of Bangladesh Agricultural University, Mymensingh \[approval number: AWEEC/BAU/2018(16)\]. Site selection and sample collection and transportation ------------------------------------------------------- For the detection and evaluation of PPRV-specific antibodies, Char Kalibari ([Fig. 1](#figure1){ref-type="fig"}) was selected as this area is an isolated area separated by the Brahmaputra River to the west. Total goat population of the Char Kalibari was 320 during this study. The village was divided into two experimental zones. The first zone populated with 205 goats considered as Group A and second area as Group B which populated with 115 goats. A total of 50 sera samples (25 sera from each group) were collected. The goats of each group were sub-divided into different age groups; (i) 0--6 months (*n* = 5), (ii) 6--12 months (*n* = 5), (iii) 12--24 months (*n* = 10), and (iv) \>24 months (*n* = 5). After that, all the goats of Group A were vaccinated with PPR-Vac^®^vaccine manufactured by LRI (Livestock Research Institute), Dhaka, and the goats of Group B were considered as non-vaccinated control. After 21 days of vaccination, again 25 sera samples from both groups were collected. Blood samples were collected directly from the jugular vein of animals by the venipuncture method using sterile 5 ml syringe. The samples were placed in icebox without agitation until serum collection. The collected blood samples were placed without agitation for about 50--60 min. The serum samples were then collected into Eppendorf tube and marked specifically. These samples were then stored in the refrigerator at the Department of Microbiology and Hygiene, Bangladesh Agricultural University, Mymensingh, until transportation. For the characterization of the PPRV, a total of 10 nasal swab samples PPR-suspected goats were collected from Jessore district ([Fig. 1](#figure1){ref-type="fig"}). Nasal secretions were taken by swab sticks from the PPR suspected goats. Swab sticks were then emerged into 2.5 ml microcentrifuge tube containing 1 ml viral transport medium. Then the upper part of the swab stick was broken up to the 2 marks of the microcentrifuge tube. Then the tubes were capped along with the broken swab sticks. The samples were stored at −70°C for further use. Detection of PPRV-specific antibodies by c-ELISA ------------------------------------------------ The serum samples were transported to the SAARC Regional Leading Diagnostic Laboratory (SAARC-RLDL) at the Animal Health Research Division, Bangladesh Livestock Research Institute (BLRI) Savar, Dhaka-1341, Bangladesh. The sera samples were then analyzed by c-ELISA by BDSL^®^ cELISA kit of The Pirbright Institute (UK) by following the instructions of the manufacturer. The optical density (OD) values were taken by the BioTek^®^ Power Wave^TM^ 340 ELISA plate reader with interference filters of 492 nm machine. The obtained OD value was then calculated and analyzed to interpret the results. As per the instructions of the kit, the percentage inhibition (PI) value of antibody ≥50% indicates seropositive, whereas PI values \<50% indicates the seronegative result. RNA extraction, RT-PCR, and agarose gel electrophoresis ------------------------------------------------------- The viral RNA samples were extracted from nasal swab sample using Ambion^®^ RNA extraction kit (USA) as per the manufacturer's instructions. The one-step conventional reverse transcription polymerase chain reaction (RT-PCR) was conducted by using the Ambion^®^ Kit (AgPath-ID^TM^ One-step RT- PCR kit, USA) as per the manufacturer's instructions. A 25 μl scale of reaction mixture consisted of 2× RT-PCR buffer 13 μl, forward and reverse primers (100 pmole/μl each) 0.5 μl, 2× RT-PCR enzyme mix 1.0 μl, template RNA 5 μl, and the rest 5 μl was nuclease-free water. The primers used for amplification of *N gene* of PPRV were NP3 (5ʹ-TCTCGGAAATCGCCTCACAGACTG-3ʹ) and NP4 (5ʹ-CCTCCTCCTGGTCCTCCAGAATCT-3ʹ) targeting an amplicon of 351-bp, as described by Couacy-Hymann et al. \[[@ref17]\]. The oligonucleotide primers were obtained from Integrated DNA Technologies (Japan). The thermal profile for the RT-PCR reaction was followed by the conditions mentioned by Couacy-Hymann et al. \[[@ref17]\]. Agarose gel electrophoresis was done for the visualization of the RT-PCR products after staining with ethidium bromide (10 mg/ml) (Sigma^®^, USA). Sequencing of PPR viral N gene ------------------------------ RT-PCR products were sequenced directly using PCR primer specific for *N gene*. First, the RT-PCR products were purified with QIAquick PCR purification kit (QIAGEN, Germany) by following the manufacturer's instructions. Then the cycle sequencing reaction was carried out with Big Dye Terminator v3.1 Cycle Sequencing Kit (Applied Biosystems, USA) in a final reaction volume of 20 μl using 200 μl capacity thin walls PCR tube. Unincorporated dye terminators from extension products for sequencing were removed completely by Big Dye^®^Xterminator^TM^ purification kit protocol. Capillary electrophoresis and data analysis were carried out from a commercial source on the ABI PRISM^®^ 310 Genetic Analyzer (Applied Biosystem, USA). The nucleic acid sequences and their deduced amino acid sequences were aligned with other related sequences retrieved from the GenBank. Sequence editing, alignment, and phylogenetic tree construction were carried out with the software (CodonCode Aligner, USA; Mega5.2). Phylogenetic and multiple sequence alignment analyses ----------------------------------------------------- The phylogenetic tree by the neighbor-joining method was constructed using partial sequences (85-aa) of the previously reported PPRV *N gene* sequences available in the GenBank by using the MEGA (Molecular Evolutionary Genetics Analysis) Ver. 6.0 \[[@ref32]\]. The multiple sequence alignment was done using 55 other PPRV strains available in GenBank using MEGA Ver. 6.0. Results ======= Based on the results of c-ELISA, the overall PPRV-specific antibody in the goats before vaccination was 44% (*n* = 22/50). Considering the goats of Group A (vaccinated), on Day-0, the prevalence was 48% (*n* = 12/25). On the other hand, 40% (*n* = 10/25) goats of Group B were seropositive on Day-0 ([Fig. 2](#figure2){ref-type="fig"}). At the pre-vaccination stage (Day-0), the average PI values of Group A of four different ages were 45.29%, 49.48%, 49.88%, and 51.0181%, respectively ([Fig. 2](#figure2){ref-type="fig"}). Similarly, on Day-0, the average PI values in Group B were 41.51%, 50.04%, 48.47%, and 50.55%, respectively ([Fig. 3](#figure3){ref-type="fig"}). After 21 days of post-vaccination (DPV), 96% (*n* = 24/25) goats of vaccinated Group A were found seropositive ([Fig. 2](#figure2){ref-type="fig"}), whereas, only 16% (*n* = 04/25) goats of non-vaccinated Group B were seropositive ([Fig. 3](#figure3){ref-type="fig"}). The average PI values of the previously described four different aged groups (Group A) goats were 83.00%, 81.31%, 79.62%, and 67.58%, respectively ([Fig. 2](#figure2){ref-type="fig"}). Similarly, among the goats of Group B, the average PI value recorded were 56.56%, 48.20%, 45.36%, and 39.12%, respectively ([Fig. 3](#figure3){ref-type="fig"}). {#figure2} {#figure3} {#figure4} Among the nasal swab samples (*n* = 10), four were found RT-PCR positive ([Fig. 4](#figure4){ref-type="fig"}) targeting *N gene* showing an amplicon of 351-bp. Among the four amplified RT-PCR samples, one was sequenced directly using NP3 and NP4 primers (mentioned previously). Sequencing of the *N gene* was done in the Bangladesh Livestock Research Institute (BLRI), Savar, Dhaka. The nucleotide sequences obtained in this study (GenBank accession KY039156) were aligned with sequences of the other related PPRV strains retrieved from GenBank. The sequence revealed that our isolate had 95%--99% similarities with the previously reported PPRV isolates of Bangladesh. Phylogenetic analysis revealed that our isolate formed a sub-cluster with other Bangladesh isolates under lineage IV ([Fig. 5](#figure5){ref-type="fig"}). However, it is indicated that PPRV reported from African countries are mostly placed under lineage I to III. Multiple sequence alignment of the partial amino acid (N protein) sequence (85-aa) revealed that the PPRV isolate of this study changed its genomic structure. One amino acid (glycine; G) substitution has been found in place of arginine (R), which is considered as the usual amino acid at the same position ([Fig. 6](#figure6){ref-type="fig"}). Discussion ========== PPR is now considered as an endemic disease and a threat toward the development of goat farming in Bangladesh \[[@ref2],[@ref10],[@ref19]\]. Several studies had been reported on different aspects of PPRV in goats in Bangladesh; for example, pathological investigation \[[@ref20]\], evaluation of antibiotic combined with hyperimmune serum therapy \[[@ref21]\], and seroprevalence \[[@ref16]\]. The present study investigated the seromonitoring of PPRV antibodies under the natural condition in vaccinated and unvaccinated goats. Birindwa et al. \[[@ref22]\] reported that 64.7% unvaccinated goats were PPRV positive in Congo. It is crucial that effective implementation of control programs for PPR requires regular vaccination with effective vaccine and seromonitoring of immunity against PPRV. The present study was adopted for seromonitoring of PPR specific antibody in the vaccinated goats after field level vaccination in a selected area of Bangladesh and investigation of the genomic changes in field isolates of PPRV by molecular technique. Based on seromonitoring, overall 44% goats were positive for PPR in this study. Recently, Rahman et al. \[[@ref23]\] could identify RNA for PPRV in 38% goats by real-time reverse transcription polymerase chain reaction (rRT-PCR), which validated our study although the approach of identification was different. However, Islam et al. \[[@ref14]\] reported an overall seroprevalence of 8.7% against PPR in Rajshahi, Gazipur and Sirajganj districts. This variation might be due to difference in geographical location. In Mymensingh region, Banik et al. \[[@ref7]\] found 76.60% average antibody titer at Day-21 of post-immunization in goat. Gowane et al. \[[@ref24]\] reported that PI value at Day-0 of vaccination was 22.50%, and at 28 days of post-vaccination, it was 71.8%. They also found that 94.92% of the total animals showed protective titer on 28 days of post-vaccination. In Rajshahi and Sirajganj districts, the age group of 0--6 months, vaccinated samples had the highest seroprevalence (80.25%; *n* = 65/81) as compared to 12--24 (70.83%; *n* = 34/48) and \>24 months (74.60%; *n* = 47/63) age groups of goats, respectively \[[@ref14]\]. Yousuf et al. \[[@ref25]\] reported the prevalence of PPR in different areas of Bangladesh; for example, Bogra (30%), Sirajganj (86.67%), Mymensingh (10%), and Rangpur (11.67%). In our study, we detected 44% prevalence of PPRV in goats in Mymensingh. This variation might be due to the difference in location within the region and season of sample collection. Previously, in India, overall seroprevalence of PPRV antibody in goats was observed as 35% \[[@ref3]\], whereas in Tanzania, Sudan, and the Republic of Congo, it was 49.50% \[[@ref1]\], 59.15% \[[@ref27]\], and 64.7% \[[@ref22]\], respectively. A similar report from Sudan describing a prevalence of 45.6% PPRV antibodies by c-ELISA was reported by Salih et al. \[[@ref28]\]. Thus, the difference in the seropositivity of PPRV antibody among different countries might be attributed to the difference in the agro-climatic condition of different countries. In our experiment, we found 96% seropositive PI values in vaccinated goats through c-ELISA at 21 DPV, whereas Rahman et al. \[[@ref9]\] found that 62% of the goats were seropositive after 21 DPV by using the same vaccine in Bangladesh. Similarly, Anderson and McKay \[[@ref29]\] found that about 60%--70% animals were seropositive for PPRV. From the above findings, it proved that the animal gained high serum antibody level due to vaccination which made them seropositive, and the increasing trend of PI value ([Fig. 2](#figure2){ref-type="fig"}) indicated that PPR vaccine could produce immune response in goats for at least 21 days. {#figure5} Recently, RT-PCR has emerged as a highly specific and sensitive test for molecular characterization of the PPRV. RT-PCR has become the most popular tool for diagnosis as well as molecular epidemiological studies \[[@ref30]\]. RT-PCR targeting *N gene* was used for the confirmation of PPR. Libeau et al. \[[@ref12]\] also described that among the structural proteins of PPRV, N protein is antigenically most conserved among *Morbillivirus*es and is highly immunogenic in spite of its internal location. So, in the present investigation, *Ngene* was targeted for the detection of PPRV. {#figure6} In this study, 10 nasal swab samples were collected and subjected to RT-PCR for molecular characterization. The primers NP3 and NP4 were used previously designed by Couacy-Hymann et al. \[[@ref17]\]. Bhuiyan et al. \[[@ref31]\] reported that the primers (NP3/NP4) against *N gene* were specific and sensitive to ensure efficient amplification and detection of PPRV in field samples. The primers produced an amplicon of 351-bp amplifying a region of *N gene* in four (*n* = 04) PPRV positive samples. These findings were closely related with the recent findings of Chowdhury et al. \[[@ref32]\] in Bangladesh. The deduced amino acid sequences of the present isolate (KY039156/PPRV/BDG/Jes/2016) were used to construct a phylogenetic tree with other 55 (*n* = 55) PPRV strains from 15 different countries (Bangladesh, India, China, Turkey, Kurdistan, Iran, Turkey, Morocco, Nigeria, Burkina-Faso, Cote d'ivoire, Guinea Bissau, Oman, Ethiopia, and Sudan) available in GenBank ([Fig. 5](#figure5){ref-type="fig"}). The tree shows that the strain of Jessore (KY039156/PPRV/BDG/Jes/2016) is grouped along with other reported strains under lineage IV. The phylogeny showed that the present isolate (KY039156/PPRV/BDG/Jes/2016) formed a sub-cluster under lineage IV, as reported by El Arbi et al. \[[@ref33]\]. The PPRV strain isolated from Jessore, Bangladesh, depicts a common phylogenetic lineage IV ([Fig. 5](#figure5){ref-type="fig"}), and possible spread to/from India as a neighboring country through trans-boundary commerce. From this study, it is revealed that the genomic similarity of the present strain (KY039156/PPRV/BDG/Jes/2016) was 95%--99% identical to each other of different PPRV strains of Bangladesh. Moreover, the sequence homology of the identified virus with rest of the viruses of China, India, Turkey, Iran, and Kurdistan was 97%--98%, 97%, 95%--96%, 95%, and 95%--97%, respectively. The strain isolated from Jessore, Bangladesh, revealed a unique amino acid substitution, where glycine (G) was found in place of arginine (R), isoleucine (I), and lysine (K) ([Fig. 6](#figure6){ref-type="fig"}), indicating that the strain is genetically different as compared to other strains reported from Bangladesh or other parts of the world. Rahman et al. \[[@ref23]\] detected amino acid substitution of PPRV in a 59 amino acid window of N-protein (420 and 497). These close similarities of the variable part of the *N gene* suggest that the trans-boundary circulation of PPR virus among the countries might be the cause of the severe outbreak in Bangladesh. Low sample size for genetic characterization was the limitation of this study. Full-length *N-gene* analysis may reveal a more comprehensive understanding about the circulating PPRV in Bangladesh. Conclusion ========== High antibody titer after vaccination of goats by PPR vaccine indicates that the goat responded well against the vaccine. The presence of PPRV is confirmed by RT-PCR in Jessore district of Bangladesh. Sequence homology and phylogenetic analysis reveal that the PPRV of this report is located within the same cluster under lineage IV of other Bangladeshi isolates. However, unique genetic variation has been detected in our strain. Thus, effective controlling of the PPR in goats can be depended on the use of effective vaccine and consideration of genetic variation. The current control strategies for PPRV in Bangladesh can be reevaluated considering regular seromonitoring and outbreak investigation of the disease throughout the country. This study provides baseline data, which may help in the development of an effective vaccine and controlling PPR in Bangladesh. The chemical, reagents, and laboratory supports of the SAARC Regional Leading Diagnostic Laboratory (SAARC-RLDL) at the Animal Health Research Division, Bangladesh Livestock Research Institute (BLRI) Savar, Dhaka-1341, Bangladesh is accordingly acknowledged. Conflict of Interest ==================== The authors declare that they have no conflict of interest. Authors' contributions ====================== SA and MAY designed the research work. SA, MAY, and MMI conducted the actual experiments. MYA, MAI, and MMM helped in conducting the research works and data analysis. MRI and KHMNHN supervised the research work. All the authors read and finally approved the manuscript for submission.

###### Significance of this study What is already known on this subject? ====================================== - Metabolic diseases such as diabetes or non-alcoholic fatty liver disease are associated with macro- and microcirculation and microcirculation damages, initiating an impairment of endothelium-dependent relaxation, a primary driver of cardiovascular diseases. - Dietary n-3 polyunsaturated fatty acid (PUFA) depletion induces hepatic steatosis and accelerates the development of endothelial dysfunction in mesenteric arteries of apolipoprotein E knockout (Apoe^−/−^) mice. - Gut microbiota plays a crucial role in the control of host intestinal functions, through the release and/or transformation of metabolites (eg, bile acids and short chain fatty acids) which regulate gut endocrine function, these pathways being mostly studied in the context of obesity. What are the new findings? ========================== - Inulin-type fructans (ITFs) are able to restore the endothelial dysfunction observed in mesenteric and carotid arteries from n-3 PUFA-depleted Apoe^−/−^ mice without impacting adiposity. - The improvement of vascular dysfunction by ITF is linked to an activation of the NOS/NO pathway pathway, which could be dependent on events occurring at the microbiota level (increase in NO-producing bacteria) and/or host level (changes in bile acid composition, increase in the L cells density and in glucagon-like peptide 1 production, both acting on the NOS/NO pathway). - Our data support, for the first time in the context of vascular dysfunction, the concept that changing the gut microbiota has a profound influence on key intestinal functions involved in host cardiometabolic health. How might it impact on clinical practice in the foreseeable future? =================================================================== - Intrahepatic and mesenteric endothelial dysfunctions in metabolic diseases are known as actor and predictor of deleterious cardiovascular consequences. If the positive impact of ITF on endothelial dysfunction is confirmed in human studies, they could be proposed as a novel approach in the prevention and the management of metabolic disorders-related cardiovascular diseases. Introduction {#s1} ============ Nutritional quality of diets underlies or exacerbates several chronic pathologies, including cardiovascular disease (CVD).[@R1] [@R2] Western diets that are mainly characterised by an imbalance between energy and fat intakes and a lack of key nutrients, like fibres and n-3 polyunsaturated fatty acids (PUFAs), enhance CVD risk.[@R3] An early key marker of CVD is endothelial dysfunction, reflecting the integrated effects of risk factors on the vasculature.[@R4] It originates from the incapacity of endothelial cells to balance synthesis and release of deleterious versus protective mediators, among which nitric oxide (NO) is the most important. Endothelial dysfunction is characterised by reduced vasodilation in response to endothelial stimuli.[@R5] Interestingly, besides inducing metabolic alterations, nutritional depletion in n-3 PUFAs for 12 weeks accelerates the process of endothelial dysfunction in apolipoprotein E knockout (Apoe^−/−^) mice.[@R6] The gut microbiota comprises the trillions of commensal micro-organisms residing within our intestines.[@R7] All the genes of this community -- the gut microbiome -- represent at least 100-fold more genes than the host genome.[@R8] The gut microbiota may be considered as an 'external' organ playing an important role in host physiology and metabolism. Only a few studies focus on the role of gut microbiota in the context of CVDs. Interestingly, Stepankova *et al* [@R9] reported that germ-free Apoe^−/−^ mice developed more aortic atherosclerotic plaques compared with conventionally raised Apoe^−/−^ mice fed the same low standard cholesterol diet. Rault-Nania *et al* [@R10] demonstrated that inulin treatment attenuated atherosclerosis in Apoe^−/−^ mice, however, the mechanism was unexplored. We found that inulin-type fructans (ITFs) feeding increases the abundance of *Akkermansia muciniphila* in obese mice,[@R11] [@R12] a bacterium known to reduce atherosclerotic lesions induced by a Western diet in mice.[@R13] For years, we and others have proven that dietary fibre supplementation can modulate the composition of the gut microbiota, and thus interact with host physiology.[@R14] [@R15] This is particularly the case for ITFs, which are classified as prebiotics. Prebiotics are 'non-digestible compounds that through metabolisation by microorganisms in the gut, modulate the composition and/or activity of the gut microbiota, thereby conferring a beneficial physiological effect on host'.[@R16] We have additionally shown that ITFs are able to lower hepatic steatosis in n-3 PUFA-depleted wild-type (WT) mice, by modulating gene expression in the liver.[@R17] The microcirculation (e.g. the enteric arterial tree) plays key roles in the metabolic diseases progression.[@R18] [@R19] We have evaluated for the first time the influence of ITFs in a model in which endothelial dysfunction appears independently of obesity and inflammation,[@R6] both being affected by ITFs. Our model allows to point out molecular mechanisms by which changes in gut microbiota might remedy the enteric vascular dysfunction. Materials and methods {#s2} ===================== Additional protocols and complete procedures are described in the online [supplementary material and methods](#SM1){ref-type="supplementary-material"} section. 10.1136/gutjnl-2016-313316.supp1 Animals {#s2a} ------- Nine-week-old male C57Bl/6J (WT) and Apoe^−/−^ (KO) mice (Charles River Laboratories, L\'Arbresle, France) were fed an n-3 PUFA-depleted (DEF) diet (D08041806, Research Diets, New Brunswick, New Jersey, USA) for 12 weeks. At 10 weeks of n-3 PUFA depletion, for each genotype, mice were separated in two groups and supplemented or not with ITFs (OraftiP95, Tienen, Belgium) at 250 mg/mouse/day in the drinking water. Measurement of vascular contraction and relaxation {#s2b} -------------------------------------------------- Endothelium-dependent relaxation was evaluated by cumulative addition of acetylcholine (Ach) (from 10^−8^ M to 3.10^−5^ M) on precontracted arteries with a high KCl solution, in presence or absence of Nω-Nitro-L-arginine methyl ester (L-NAME) (100 µM). Microarray analysis {#s2c} ------------------- Equal amounts of RNA from eight mice were pooled within each group. Microarrays were performed as previously described.[@R20] Data are available under GEO accession number GSE87603. Gut microbiota analyses {#s2d} ----------------------- Genomic DNA was extracted from caecal content using a QIAamp DNA Stool Mini Kit (Qiagen, Germany), including a bead-beating step. Illumina sequencing was performed as previously described.[@R21] The V5-V6 region of the 16S rRNA gene was amplified by PCR with modified primers. The amplicons were purified, quantified and sequenced using an Illumina Miseq to produce 2×300-bp sequencing products at the University of Minnesota Genomics Center. Initial quality filtering of the reads was conducted with Illumina Software, yielding an average of 66766 pass-filter reads per sample. Quality scores were visualised, and reads were trimmed to 220 bp (R1) and 200 bp (R2). The reads were merged with the merge-Illumina-pairs application.[@R22] For samples with \>25 000 merged reads, a subset of 25 000 reads was randomly selected using Mothur 1.25.0[@R23] to avoid large disparities in the number of sequences. Subsequently, the UPARSE pipeline implemented in USEARCH V.7.0.1001[@R24] was used to further process the sequences. Putative chimaeras were identified against the Gold reference database and removed. Clustering was performed with a 98% similarity cut-off to designate operational taxonomic units (OTUs). Non-chimerical sequences were also subjected to taxonomic classification using the RDP MultiClassifier 1.1 from the Ribosomal Database Project[@R25] for phylum to genus characterisation of the faecal microbiome. The phylotypes were computed as per cent proportions based on the total number of sequences in each sample. Statistical analysis {#s2e} -------------------- Results are presented as mean±SEM. Statistical differences were assessed by one-way or two-way analysis of variance followed by Tukey\'s or Bonferroni\'s post-tests. Unpaired *t*-test was used for comparison between KO DEF and KO DEF ITF mice. Data with superscript symbol (\* vs WT DEF, ^\$^ vs WT DEF ITF or ^\#^ vs KO DEF) were significantly different (p\<0.05). Statistical analyses were performed using GraphPad Prism V.5.00 for windows. Results {#s3} ======= Endothelial dysfunction is improved by ITF treatment through the activation of the NO synthase/NO pathway {#s3a} --------------------------------------------------------------------------------------------------------- As expected, mesenteric arteries isolated from n-3 PUFA-depleted Apoe^−/−^ mice (KO DEF) developed a significant endothelial dysfunction, attested by the reduced vasodilation to cumulative doses of Ach ([figure 1](#GUTJNL2016313316F1){ref-type="fig"}A). Fifteen days of ITF supplementation significantly improved the relaxation of mesenteric arteries isolated from n-3 PUFA-depleted Apoe^−/−^ mice supplemented with ITFs (KO DEF ITF) as compared with WT mice, treated or not ([figure 1](#GUTJNL2016313316F1){ref-type="fig"}A). The non-selective NO synthase (NOS) inhibitor (L-NAME) completely blunted the relaxation in mesenteric resistance arteries independently of diet or genetic background, suggesting an implication of the NOS/NO pathway ([figure 1](#GUTJNL2016313316F1){ref-type="fig"}B). In line with endothelial dysfunction, KO DEF mice displayed a 40% decrease in haem-nitrosylated haemoglobin (Hb-NO) levels, measured by electron paramagnetic resonance, as compared with WT DEF mice. Interestingly, ITF treatment restored Hb-NO to similar levels as measured in WT mice ([figure 1](#GUTJNL2016313316F1){ref-type="fig"}C). We investigated the relaxation profile of carotid arteries, another type of vessel, conductance versus resistance arteries, and at distance from the gut. The beneficial effects of ITFs on mesenteric arteries relaxation were also observed in carotid arteries isolated from KO DEF ITF mice ([figure 1](#GUTJNL2016313316F1){ref-type="fig"}D). We used western blotting to assess total endothelial NOS (eNOS) as well as its activating serine 1177 phosphorylated form (p-eNOS^ser1177^) along with its allosteric regulator, caveolin-1 ([figure 1](#GUTJNL2016313316F1){ref-type="fig"}E). eNOS phosphorylation was twofold higher in mesenteric arteries from KO DEF ITF mice compared with KO DEF mice, while total eNOS protein was not significantly affected by genotype and/or diet. This resulted in a significantly increased p-eNOS^ser1177^/eNOS ratio in mesenteric arteries isolated from KO DEF ITF mice, compared with samples from KO DEF ITF. In accordance with previous observations,[@R26] we observed a slight increase in caveolin-1 abundance in KO DEF versus WT DEF ([figure 1](#GUTJNL2016313316F1){ref-type="fig"}F). {#GUTJNL2016313316F1} ITF treatment impacts resting parameters and vascular reactivity in mesenteric arteries {#s3b} --------------------------------------------------------------------------------------- Resting tone was significantly higher in mesenteric arteries isolated from KO DEF ITF mice, compared with untreated mice ([figure 2](#GUTJNL2016313316F2){ref-type="fig"}A). This effect was associated with a significantly larger mean arterial diameter, as compared with WT DEF mice ([figure 2](#GUTJNL2016313316F2){ref-type="fig"}B). Mesenteric arteries from KO DEF ITF mice contracted significantly more in response to KCl-enriched solution (50 mM) compared with all other groups ([figure 2](#GUTJNL2016313316F2){ref-type="fig"}C). Intima-media thickness in mesenteric arteries from KO DEF mice was significantly reduced in comparison with either treated or untreated WT DEF mice ([figure 2](#GUTJNL2016313316F2){ref-type="fig"}D,E). Interestingly, media remodelling was completely reversed by 15 days of ITF supplementation. {#GUTJNL2016313316F2} Microarray analysis highlights several pathways affected by ITF treatment {#s3c} ------------------------------------------------------------------------- To identify the mechanisms underlying the improvement in endothelial function by ITF treatment, we analysed we analysed caecal gene expression profiles by microarray ([figure 3](#GUTJNL2016313316F3){ref-type="fig"}). Further comparison focused on KO genotypes in which most effects were observed. ITF treatment led to an increased expression in (1) various antimicrobial peptides (e.g. *Defa*, *Reg3β*, *SAA1*, *SAA3*), (2) genes involved in bile acid (BA) metabolism (*Slc10a2*), (3) aminopeptidases (*Anpep, Enpep*) and (4) (pro)hormones, neurotensin (*Nts*) and proglucagon (*Gcg*). Quantitative PCR analyses confirmed these microarray data (see online [supplementary figure S1](#SM2){ref-type="supplementary-material"}). {#GUTJNL2016313316F3} 10.1136/gutjnl-2016-313316.supp2 ITF treatment profoundly modifies the gut microbiota composition in WT and KO mice {#s3d} ---------------------------------------------------------------------------------- Shannon and Simpson Indexes ([figure 4](#GUTJNL2016313316F4){ref-type="fig"}A), representing richness and evenness, were significantly decreased in KO DEF ITF mice compared with KO DEF mice just as well as Evenness Indexes (Simpson Evenness and Heip Indexes, [figure 4](#GUTJNL2016313316F4){ref-type="fig"}B). However, the richness (measured using Chao1 and Observed Species Indexes, [figure 4](#GUTJNL2016313316F4){ref-type="fig"}C) was not significantly modified in any groups. The β-diversity was also profoundly changed in ITF-treated WT and KO mice, as observed in the principal coordinates analysis (PCoA) of the Morisita-Horn Index ([figure 4](#GUTJNL2016313316F4){ref-type="fig"}D) and the weighted UniFrac Index ([figure 4](#GUTJNL2016313316F4){ref-type="fig"}E) at the OTU level. The three first principal coordinates respectively explained 56% and 66% of the variation observed between all the groups. In the two PCoA, we observed clustering for ITF-treated versus untreated mice, independently of genotype. At the phyla level ([figure 4](#GUTJNL2016313316F4){ref-type="fig"}F), several shifts clearly operated following ITF supplementation in the two genotypes such as a decreased abundance of the Firmicutes phylum and an increased abundance of the Bacteroidetes, Actinobacteria and Proteobacteria phyla. Intriguingly, KO DEF mice presented a lower relative abundance of the Verrucomicrobia phylum compared with WT DEF mice and ITFs were able to restore its abundance ([figure 4](#GUTJNL2016313316F4){ref-type="fig"}F). Parametrical analysis at lower taxonomical levels revealed additional changes. ITF had growth-fostering effects on the Erysipelotrichaceae family and *Bifidobacterium* genus in both genotypes. In line with the observations at the phylum level, abundance of *Akkermansia* genus was drastically decreased in KO DEF mice while the ITF treatment restored its abundance to a similar level as observed in WT mice ([figure 4](#GUTJNL2016313316F4){ref-type="fig"}G). Concomitantly, ITFs also led to a decreased relative abundance of Desulfovibrionales and Clostridiales orders and of the Ruminococcaceae and Lachnospiraceae families (only in the KO genotype for the last family) ([figure 4](#GUTJNL2016313316F4){ref-type="fig"}H). Discriminant analyses using LEfSe confirmed these results ([figure 5](#GUTJNL2016313316F5){ref-type="fig"}). Enrichment in Actinobacteria (Bifidobacteriales and Coriobacteriales orders), Enterobacteriaceae and Erysipelotrichaceae characterised ITF-treated mice whereas enrichment in Deltaproteobacteria, Ruminococcaceae and Clostridiales was observed in untreated mice ([figure 5](#GUTJNL2016313316F5){ref-type="fig"} and see online [supplementary figure S2](#SM2){ref-type="supplementary-material"}). It appears from visual comparison of cladograms that ITFs have similar but not identical effects in both genotypes, with a few changes occurring only in KO DEF groups (such as enrichment in Bacteroidetes and *Akkermansia* upon ITFs). Among the 93 OTUs significantly modified by the treatments at the q-level, OTU 19, identified as *Escherichia coli*, and OTU 6, identified as *Shigella boydii*, were increased by at least 600-fold, in ITF-treated groups (see online [supplementary table S1](#SM3){ref-type="supplementary-material"}). Moreover, OTU 1, identified as *Akkermansia muciniphila*, was increased by fourfold in KO DEF ITF versus KO DEF mice, reaching 16% of the gut microbiota in KO DEF ITF mice. We also observed that six OTUs, all belonging to the Lachnospiraceae family, were drastically decreased by ITFs in both genotypes (between 127-fold and 2198-fold decrease, mostly observed in KO DEF ITF). OTU 141, identified as unclassified Ruminococcaceae, also decreased by 217-fold in KO DEF ITF as compared with KO DEF mice (see online [supplementary table S1](#SM3){ref-type="supplementary-material"}). {#GUTJNL2016313316F4} {#GUTJNL2016313316F5} 10.1136/gutjnl-2016-313316.supp3 ITF treatment quantitatively and qualitatively modifies plasma and caecal BA profiles {#s3e} ------------------------------------------------------------------------------------- We measured the expression of genes involved in the biosynthesis and enterohepatic cycle of BAs focusing on KO genotype, which develop the endothelial dysfunction. ITF supplementation significantly upregulated hepatic mRNA levels of *Cyp7a1*, the rate-limiting enzyme of the classical BA synthesis pathway ([figure 6](#GUTJNL2016313316F6){ref-type="fig"}A). Although *Cyp27a1,* which is mostly involved in the alternative pathway, was not affected by ITF treatment, the expression of *Cyp7b1* was decreased by ITF treatment. *Cyp8b1* was not modified by the ITF supplementation ([figure 6](#GUTJNL2016313316F6){ref-type="fig"}A). ITF treatment significantly enhanced markers of BAs reuptake in the ileum (*Slc10a2*, *Fabp6, Slc51a/b*) ([figure 6](#GUTJNL2016313316F6){ref-type="fig"}B). ITF treatment also increased the BA concentration, which was mainly due to the rise in free BAs, rather than the tauro-conjugated BAs ([figure 6](#GUTJNL2016313316F6){ref-type="fig"}C--F). More precisely, the primary BAs cholic acid (CA) and chenodeoxycholic acid (CDCA) increased strongly in the portal and systemic blood as well as in the caecal content. Muricholic acid (MCA) and ursodeoxycholic acid (UDCA) were also strongly increased in blood and in caecal content. In contrast, the concentrations of the secondary BA deoxycholic acid (DCA) and its tauro-conjugated form were strongly reduced following ITF supplementation. Another secondary BA, lithocholic acid (LCA), was also significantly decreased in the caecal content by the ITF treatment ([figure 6](#GUTJNL2016313316F6){ref-type="fig"}H). Caecal DCA and LCA positively correlated with abundance in unclassified Ruminococcaceae and with abundance in Lachnospiraceae and unclassified Lachnospiraceae ([figure 7](#GUTJNL2016313316F7){ref-type="fig"} and see online [supplementary figure S3 and table S2](#SM4){ref-type="supplementary-material"}). A PICRUSt analysis predicted no significant difference in primary and secondary BA biosynthesis pathways (see online [supplementary table S3](#SM5){ref-type="supplementary-material"}). These results must be taken into account with caution because of the moderate accuracy of the prediction, evidenced by a mean NSTI value of 0.10±0.006.[@R27] Since *Cyp7a1* also regulates cholesterol metabolism and its enteric elimination, we determined blood lipid profiles in KO DEF and KO DEF ITF mice. ITF treatment did not modify the plasma lipids profile (see online [supplementary table S4](#SM6){ref-type="supplementary-material"}). {#GUTJNL2016313316F6} {#GUTJNL2016313316F7} 10.1136/gutjnl-2016-313316.supp4 10.1136/gutjnl-2016-313316.supp5 10.1136/gutjnl-2016-313316.supp6 ITF treatment increases glucagon-like peptide 1 production {#s3f} ---------------------------------------------------------- ITF treatment in KO DEF ITF mice provoked a threefold increase in active portal glucagon-like peptide 1 (GLP-1) concentrations, as compared with KO DEF mice ([figure 8](#GUTJNL2016313316F8){ref-type="fig"}A). Increased expression of *Gcg* and of an enzyme involved in GLP-1 maturation (namely prohormone convertase 1/3, *PC1/3*) were observed in KO DEF ITF mice ([figure 8](#GUTJNL2016313316F8){ref-type="fig"}B). Differentiation factors *NeuroD1* and neurogenin 3 (*Ngn3*) were not modified by ITF treatment, suggesting they are not involved in higher differentiation of enteroendocrine cells producing GLP-1. However, a large number of GLP-1-positive cells was observed in the proximal colon, thereby suggesting that overall pool of L cells is increased whereas the differentiation is not affected ([figure 8](#GUTJNL2016313316F8){ref-type="fig"}C,D).[@R28] Accordingly, correlation analysis revealed that *Gcg* mRNA positively correlates with the abundance of Bifidobacteriaceae and Verrucomicrobiaceae families, both enhanced by ITFs, and negatively correlates with the abundance of Ruminococcaceae and Lachnospiraceae families ([figure 8](#GUTJNL2016313316F8){ref-type="fig"}E). {#GUTJNL2016313316F8} Discussion {#s4} ========== Metabolic diseases such as non-alcoholic fatty liver disease are associated with macro- and microcirculation and microcirculation damage, initiating an impairment of endothelium-dependent relaxation.[@R19] [@R29] Here, we demonstrate that supplementation in ITF for 15 days corrects the endothelial dysfunction -- an early key biomarker of CVDs -- present in n-3 PUFA-depleted Apoe^−/−^ mice. Blood vessels are assumed to respond to varying conditions and functional demands through continuous adaptive structural changes. Accordingly following ITF treatment, we observed an increase in mesenteric diameter and wall thickness associated with increasing contractile reactivity, all needed to ensure stable vascular adaptation. Interestingly, the favourable outward remodelling, observed locally, was also characterised by improved NO-dependent endothelium relaxation. In most cases, endothelial dysfunction is related to a reduced NO availability in mice.[@R30] Inhibiting the NOS/NO pathway with L-NAME totally masked the vasodilation in mice, abolishing the positive effect of ITFs. The increased ratio of p-eNOS^ser1177^ to total eNOS suggests that ITFs activate the NOS/NO pathway in Apoe^−/−^ mice. Of more relevance for human CVD, effects of ITFs were not restricted to the enteric vascular tree, such effects were also observed at a distance from the mesenteric bed as indicated by the recovery of the circulating level of Hb-NO in peripheral blood and the improved relaxation in carotid arteries. Whether the molecular mechanisms underlying the improved NO-dependent relaxation in the carotid artery fully reflect those involved in the mesenteric tree remains elusive at this stage. Weight loss can improve the impairment of endothelium-dependent relaxation;[@R31] however, this hypothesis can be ruled out in our study as no change in body weight was observed in ITF-treated versus untreated mice (see online [supplementary table S5](#SM7){ref-type="supplementary-material"}). 10.1136/gutjnl-2016-313316.supp7 The Erysipelotrichaceae family, belonging to the Firmicutes phylum, was expanded by ITFs. It has been previously reported that dietary intervention such as prebiotics or berberine in high-fat-fed mice concomitantly led to an expanded abundance of *Allobaculum* sp -- an important member of the Erysipelotrichaceae family -- and to an improvement in metabolic parameters, suggesting a potential beneficial contribution of *Allobaculum* sp to host phenotype.[@R32] [@R33] In agreement with these observations, ITFs increased OTU 4, closely related to *Allobaculum* sp (90.4%), by tenfold in ITF-treated Apoe^−/−^ mice. Furthermore, abundance in Erysipelotrichaceae was strongly and positively correlated with portal CDCA, UDCA and MCAs but not to their conjugates. Abundance of *Akkermansia*, a bacterium capable of reducing fat mass development, insulin resistance and dyslipidaemia in obese mice,[@R11] [@R12] was quadrupled by ITFs in Apoe^−/−^ mice, reaching the same level as that observed in WT mice. In addition, it has been reported that *Akkermansia* positively correlates with higher levels of circulating primary BAs;[@R34] it is in accordance with the positive correlation observed between circulating CA and this bacterium. Based on the results of multiple correlation analyses, we propose that the marked decreases in the Lachnospiraceae and Ruminococcaceae families caused by ITFs contribute to the drop in DCA and LCA concentration, since they are among the dominant families able to produce secondary BAs through 7α-dehydroxylation. These observations are consistent with previous studies showing, in humans and mice with liver diseases, an association between abundance of Ruminococcaceae and Lachnospiraceae families and secondary BA concentrations.[@R35] [@R36] The Desulfovibrionales order, known to be increased by high-fat diet and associated with inflammation and/or altered gut barrier in mouse models,[@R33] [@R37] [@R38] was also decreased by at least twofold in ITF-supplemented mice and strongly correlated with DCA concentration in Apoe^−/−^ mice. We can therefore propose that changing the microbial composition using ITFs impacts largely the production of secondary BAs and could contribute to the improvement of the host\'s health. Changes in BA profiles by ITF treatment also result from the modulation of BA metabolism, including hepatic synthesis (in favour of *Cyp7a1*) and intestinal reuptake. In the liver, BA-activated Farnesoid X receptor (*FXR*) induces the expression of small heterodimer partner (*Shp*) which can inhibit *Cyp7a1* expression; both are negatively correlated in our study (Spearman\'s r=−0.62; p=0.0014). Although *FXR* mRNA expression was not affected by ITFs, α and β MCAs (free or tauro-conjugated) induced by ITFs, have FXR antagonistic properties[@R39] [@R40] and may contribute to the *Cyp7a1* upregulation. In addition, ITFs enhance BA reabsorption, which is negatively associated with changes in Ruminococcaceae and Lachnospiraceae families and positively associated with the Bifidobacteriaceae and Erysipelotrichaceae families and the *Akkermansia* genus. Enterobacteriaceae were largely expanded by the ITF supplementation and were dominated by *E coli/Shigella* (identified as OTUs 19 and 6, respectively) in our study. Interestingly, bacteria like *E coli*, can produce NO either enzymatically from L-arginine via bacterial NOS or non-enzymatically through nitrite reduction.[@R41] [@R42] Another bacterial family increased by ITFs that may contribute to NO generation is the Bifidobacteriaceae family. Indeed, by decreasing the pH, bifidobacteria may generate more NO than any other species through the acidic non-enzymatic reduction of nitrite.[@R43] The promotion of GLP-1 release by ITFs appears to be another potential link between events occurring in the bacterial ecosystem and improvement of endothelial function, since *Bifidobacterium* as well as *Akkermansia* are positively correlated with mRNA expression of *Gcg* and *Nts*, being coexpressed and co-released with GLP-1.[@R44] Moreover, GLP-1 infusion or GLP-1 receptor agonists can correct endothelial dysfunction, in humans and rodents, by reducing oxidative stress or by activating the NOS/NO pathway via eNOS phosphorylation.[@R45] BAs produced through metabolic cooperation between hosts and microbes, may improve endothelial dysfunction through the TGR5 receptor pathway[@R48] [@R49] as has been extensively reported in Fiorucci *et al.* [@R49] We confirmed that free BAs such as CA, CDCA and UDCA directly stimulate the NOS/NO pathway by enhancing serine phosphorylation of eNOS in vitro (data not shown). In addition, TGR5 is highly expressed in enteroendocrine L cells in the small intestine and stimulates GLP-1 secretion, thus indirectly improving the endothelial function.[@R49] [@R50] Plasmatic LDLc and HDLc levels were similar in all groups of mice. Although lipid profiles were measured in mice that were not rigorously fasted, and therefore might be not completely accurate, it is very unlikely that different lipoprotein plasma levels contributed to the beneficial effects of ITFs. The improvement of enteric vascular dysfunction by ITFs could relate to events occurring at microbial level (e.g. increase in NO-producing bacteria) and/or at host level (e.g. changes in BA composition, increase in L cells density and GLP-1 production, both acting on the NOS/NO pathway). Our study is the first to point out the effects of ITFs on host cardiometabolic health, including endothelial and endocrine functions, and can be considered as a rationale to evaluate such an effect in a human intervention study. For sure, the dose at which ITF would be tested is below the one given to animals (0.25 g/day). In human intervention studies, doses ranging from 12g/day to 16 g/day are often given when testing for metabolic effects of ITF.[@R51] In addition, there are certainly differences between humans and mice concerning the gut microbiota composition, but it is rather interesting to point out that similar bacterial changes occur in both humans and mice upon prebiotic intervention (e.g. in favour of the *Bifidobacterium* genus).[@R55] [@R56] While effects on gut function (e.g. increase in GLP-1 production by prebiotics) have been largely described in rodent models, several papers also reported the influence of prebiotics in humans.[@R51] In conclusion, our findings pave the way for innovative therapeutic approaches in prevention of human vascular dysfunction, for which no treatment has been successfully proposed so far. The authors thank B Es Saadi and R Selleslagh for technical assistance and V Joris for in vitro analyses. **Twitter:** Follow Patrice Cani [\@MicrObesity](http://twitter.com/MicrObesity) **Contributors:** Conceptualisation: EC, CD and NMD; Methodology: EC, CD and NMD; Investigation: EC, LBB, AT, SL, AMN, IL, CB, PDC and BS; Formal analysis: J-FG, HP, AE and J-BD; Writing-original draft: EC, CD and NMD; Writing review and editing: all the authors; Supervision: CD and NMD. **Funding:** This work is supported by FNRS (Fond National de la Recherche Scientifique, Belgium; CDR J.0122.15). LBB was a postdoctoral researcher from the FRS-FNRS and is the recipient of subsidies from the FSR (Fonds Spéciaux de la Recherche, UCL). HP is a research fellow at the FRS-FNRS. NMD is a recipient of grants from FRS-FNRS, from Wallonia supported by the competitive cluster Wagralim (ADIPOSTOP project, convention 7366; FOOD4GUT project, convention 1318148) and from the European Union\'s Seventh Framework Program (grant agreement no 613979). PDC is a research associate at the FRS-FNRS and the recipient of grants from FNRS (convention J.0084.15, convention 3.4579.11), PDC (Projet de Recherche, convention: T.0138.14), WELBIO-CR-2012S-02R, the Funds Baillet Latour (Grant for Medical Research 2015). PDC is a recipient of an ERC Starting Grant 2013 (starting grant 336452-ENIGMO). CD is a senior research associate at FRS-FNRS. BS is a recipient of the Institut Universitaire de France. This work was partly supported by grants from the European Genomic Institute for Diabetes (ANR-10-LABX-46) and the ERC Grant (Immunobile, contract 694717). **Competing interests:** None declared. **Ethics approval:** The experiments were approved by and performed in accordance with the guidelines of the local ethics committee. Housing conditions were as specified by the Belgian Law of 29 May 2013, regarding the protection of laboratory animals (agreement no LA1230314). **Provenance and peer review:** Not commissioned; externally peer reviewed. [^1]: CD and NMD contributed equally.

{.962} {.963} {.964} {.965} {.966}

The authors confirm that all data underlying the findings are fully available without restriction. All relevant data are within the paper and its Supporting Information files. Introduction {#s1} ============ The world's third most common cancer, with an annual incidence of over one million new cases, is colorectal cancer (CRC) [@pone.0109600-Siegel1]. Its cure is based on early diagnosis, radical surgery, and possible adjuvant therapy. Such therapy is routine for stage III patients. At stage II, the advantage of adjuvant therapy is, however unclear, since about 80% of surgically treated stage II patients survive without chemotherapy. Although we know many high risk factors; T4-stage, low differentiation, vascular invasion, tumor obstruction, bowel perforation, and inadequate lymph node resection, we cannot always identify patients who will benefit from adjuvant therapy. It would be beneficial to find new biomarkers to aid in treatment decisions. Regenerating islet-derived gene (REG) proteins represent a group of small secretory proteins involved in cell proliferation and regeneration, that also participate in formation of the immune system [@pone.0109600-Dusetti1], [@pone.0109600-Broekaert1]. They belong to the calcium-dependent lectin (C-lectin) superfamily and are divided into four families, REG I to IV, based on their primary structure. At variance with the other REG proteins, REG4 binds polysaccharides independently of calcium [@pone.0109600-Ho1]. The genes encoding REG I to III genes are located on chromosome 2p12, while that of REG4 is on 1p12--13. REG4 was first cloned and identified by Hartupee et al [@pone.0109600-Hartupee1] and by Kämäräinen et al [@pone.0109600-Kmrinen1]. Containing 158 aminoacids and with a molecular of weight of 18 kDA, it is physiologically expressed in the colon and the small intestine, with high expression in enteroendocrine cells. [@pone.0109600-Kmrinen1], [@pone.0109600-Oue1]. In the gastrointestinal epithelium, REG4 is activated during specific phases of differentiation and maturation, its expression is spatially specific, and it has been suggested to support mucinous and neuroendocrine differentiation or [@pone.0109600-Schrder1], [@pone.0109600-Heiskala1]. Up-regulated REG4 expression occurs in inflammatory bowel diseases (IBD) [@pone.0109600-Kmrinen1] and also occurs in many malignancies: colorectal, gastric, and pancreatic cancers [@pone.0109600-Oue2]--[@pone.0109600-Takehara1]. REG4 is suggested to participate in carcinogenesis and tissue regeneration, to act as an antiapoptotic factor, and to promote proliferation and invasion [@pone.0109600-Macadam1]. The ultimate physiological and pathological roles of REG4 still remain elusive, however. Expression of REG4 in GI-tract cancers has in several studies appeared to be of predictive and prognostic value [@pone.0109600-Oue2], [@pone.0109600-Mitani1], [@pone.0109600-Ohara1], [@pone.0109600-Sasahira1]. The findings in CRC have, though, been controversial: increased expression is a sign of poor prognosis according to Numata et al [@pone.0109600-Numata1], but no association with prognosis has emerged in other studies [@pone.0109600-Li1], [@pone.0109600-Zheng1]. This study aimed to evaluate in CRC the role of REG4 as a prognostic marker and its association with clinicopathological parameters in a cohort of 840 patients. Furthermore, in a subgroup of 220 patients, we focused on association of REG4 expression with other markers: markers of mucinous differentiation MUC1, MUC2, MUC4, and MUC5AC, and also markers of neuroendocrine phenotype: synaptophysin and chromogranin. Methods {#s2} ======= Patients {#s2a} -------- The study population comprised 840 consecutive colorectal cancer patients undergoing surgery in 1983--2001 at the Department of Surgery, Helsinki University Central Hospital. A subgroup of 240 comprised consecutively operated patients between 1998--2001. REG4 was studied in the whole patient series, and the other markers studied in the subgroup. The Finnish Population Register Centre provided the follow-up vital-status data needed to compute survival statistics, and Statistics Finland provided cause of death for all those deceased. Median age at diagnosis was 66, with a median follow-up of 5.1 years (range 0--25.8). The 5-year disease-specific survival rate was 58.9% (95% Cl 55.0--62.8%). For the subgroup, median age at diagnosis was 67 with a median follow-up of 6.0 years (range 0--13.2). The 5-year disease-specific survival rate was 64.8% (95% Cl 58.1--71.5%). Clinicopathological characteristics of both groups are in S1. This study complies with the Declaration of Helsinki and was approved by the Surgical Ethics Committee of Helsinki University Central Hospital (Dnro HUS 226/E6/06, extension TMK02 §66 17.4.2013) and the National Supervisory Authority of Welfare and Health gave the permission to use tissue samples without individual informed consent in this retrospective study (Valvira Dnro 10041/06.01.03.01/2012). Tissue microarray {#s2b} ----------------- Formalin-fixed and paraffin-embedded tumor samples came from the archives of the Department of Pathology, Helsinki University Central Hospital. Representative areas of tumor samples on hematoxylin- and eosin-stained slides were marked by an experienced pathologist. Three 1.0-mm-diameter punches taken from each sample were mounted on each recipient paraffin block with a semiautomatic tissue microarray instrument (TMA) (Beecher Instruments, Silver Spring, MD) as described [@pone.0109600-Kononen1]. Immunohistochemistry {#s2c} -------------------- TMA- and tissue-blocks were freshly cut into 4-µm sections. After deparaffinization in xylene and rehydration through a gradually decreasing concentration of ethanol to distilled water, the slides were treated in a PreTreatment module (Lab Vision Corp., Fremont, CA) in pre-treatment buffer for 20 minutes at 98°C for antigen retrieval. The staining procedure by the Dako REAL EnVision Detection system, Peroxidase/DAB+, Rabbit/Mouse (Dako, Glostrup, Denmark) used an Autostainer 480 (Lab Vision). Tissues were incubated with primary antibodies for one hour at room temperature. The REG4 antibody is as described in [@pone.0109600-Heiskala2], others were from commercial sources. Antibodies and their dilutions are in [Table 1](#pone-0109600-t001){ref-type="table"}. 10.1371/journal.pone.0109600.t001 ###### Antibodies for immunohistochemistry. {#pone-0109600-t001-1} Antibody Clone Company Pre-treatment Dilution Positive control ------------------- ------------ ---------------- -------------------- ---------- ------------------ **REG4** mAb In-house Tris-HCl (pH 8.5) 1∶50 Colon **MUC1** mAb, Ma552 Novocastra, UK Citrate (pH 6.0) 1∶25 Stomach **MUC2** mAb, Ccp58 Novocastra, UK Citrate (pH 6.0) 1∶100 Colon **MUC4** mAb, 1G8 Invitrogen,USA Tris-EDTA (pH 9.0) 1∶100 Colon **MU5AC** mAb, CLH2 Novocastra, UK Citrate (pH 6) 1∶50 Stomach **Synaptophysin** mAb 27G12 Novocastra, UK Tris-EDTA (pH 9.0) 1∶200 Colon **Chromogranin** mAb, 5H7 Novocastra, UK Tris-EDTA (pH 9.0) 1∶2000 Colon REG4 antibody described in detail in reference [@pone.0109600-Heiskala2] mAb = monoclonal antibody. Antibody host: mouse. Scoring of samples {#s2d} ------------------ REG4 cytoplasmic expression was scored in tumor cells as either negative of positive. MUC1 and MUC2 expressions were cytoplasmic in tumor cells and were scored as negative-low-moderate-high according to intensity. For further statistical analysis they were grouped into: low (negative to low) and high (moderate to high). MUC4 and MUC5AC cytoplasmic expressions were scored either negative or positive. Neuroendocrinic differentiation (negative vs positive) was evaluated by cytoplasmic synapthophysin and chromogranin positivity. Stainings were scored independently by T.K. and J.H., who were blinded to clinical data and outcome. Differences in scoring were discussed until consensus. Representative images of expression are in [Figure 1](#pone-0109600-g001){ref-type="fig"}. REG4 expression was compared with proliferation index by Ki-67 staining, which we published previously [@pone.0109600-Bckelman1]. {#pone-0109600-g001} Statistical analyses {#s2e} -------------------- Evaluation of the association between REG4 expression and clinicopathological parameters or MUC expressions was done by the exact Pearson chi-square test or the exact linear-by-linear association test for ordered parameters. Disease-specific overall survival was counted from date of surgery to date of death from colorectal cancer, or until end of follow-up. Survival analysis by the Kaplan-Meier method was compared by the log rank test. The Cox regression proportional hazard model served for uni- and multivariable survival analysis, adjusted for sex, age, Dukes classification, and differentiation. Testing of the Cox model assumption of constant hazard ratios over time involved the inclusion of a time-dependent covariate separately for each testable variable. The hazard ratio of differentiation was analyzed in two periods (0 to 1.25 and 1.25 to 5 years) in order to meet the assumptions of the Cox model, with the time-dependent Cox model. Interaction terms were considered and an interaction between REG4 expression and age emerged. We therefore calculated the prognostic role of REG4 separately for patients under and over 65. All tests were two-sided. A p-value of 0.05 was considered significant. All statistical analyses were done with SPSS version 20.0 (IBM SPSS Statistics, version 20.0 for Mac; SPSS, Inc., Chicago, IL). Results {#s3} ======= Immunohistochemistry {#s3a} -------------------- REG4 expression in tumor cells was cytoplasmic and slightly granular. When present, expression was evident in the vast majority of tumor cells, but with no nuclear expression. In whole tissue sections, no clear distinction in expression appeared between the invasive front and the rest of the tumor. Moreover, in whole sections, REG4 expression appeared in some cases in normal epithelium, but was down-regulated in tumor cells. Expression of mucins, synapthopysin, and chromogranin was cytoplasmic, with no nuclear expression. Of the 840 tumors represented in the TMA, REG4 staining could be evaluated in 793; 580 (73.1%) were scored as negative and 213 (28.9%) scored as positive. In a subgroup of 220 tumors, MUC 1 expression was evaluated in 206 (low 83.5% and high 17.5%), MUC2 expression in 210 (low 17.1% and high 82.9%), MUC4 expression in 208 (negative 51.0% and positive 49.0%), and MUC5AC expression in 205 (negative 93.2% and positive 6.8%). Neuroendocrinic positivity (either/both synaptohysin- and chromogranin-positive) could be evaluated in 212 (negative 92.9% and positive 7.1%). REG4 in lymph-node metastasis {#s3b} ----------------------------- Based on TMA results, we chose 10 patients with Dukes C disease, 5 with REG4-positive tumors and 5 with negative tumors. None of the patients' REG4 negative tumors showed positivity in their lymph-node metastasis, whereas of the 5 patients with REG4-positive tumor, 2 showed positivity also in their lymph node metastasis ([Figure 2](#pone-0109600-g002){ref-type="fig"}). {#pone-0109600-g002} REG4 and clinicopathological parameters {#s3c} --------------------------------------- Cytoplasmic REG4 expression associated with less-advanced stage (p = 0.014), location in the right hemicolon (p = 0.035), and mucinous histology (p\<0.0001). We saw no association with age, gender, differentiation, or location (colon vs. rectum) ([Table 2](#pone-0109600-t002){ref-type="table"}). As REG4 associated strongly with tumor histology, we analyzed mucinous and non-mucinous CRC separately. In non-mucinous CRC, REG4 expression associated with less advanced stage (p = 0.005) and higher differentiaton (p = 0.023). No association appeared with age, gender, tumor location, nor tumor side. ([Table 3](#pone-0109600-t003){ref-type="table"}) In mucinous CRC, REG4 expression did not associate with any clinocopathological parameter (data not shown). No association between REG4 and Ki-67 was seen in CRC nor in any of the subgroups analyzed (data not shown). 10.1371/journal.pone.0109600.t002 ###### Association between REG4 expression and clinicopathologic parameters in colorectal cancer. {#pone-0109600-t002-2} REG4 expression ----------------- ------------ ------------ ---------- **Age. years** \<65 245 (42.2) 94 (44.1) 0.686 ≥65 335 (57.8) 119 (55.9) **Gender** Male 313 (54.0) 128 (60.1) 0.124 Female 267 (46.0) 85 (39.9) **Dukes** A 75 (12.9) 44 (20.7) 0.014 B 202 (34.8) 73 (34.3) C 164 (28.3) 55 (25.8) D 139 (24.0) 41 (19.2) **Grade (WHO)** 1 14 (2.4) 13 (6.2) 0.064 2 394 (68.3) 150 (71.1) 3 150 (26.0) 38 (18.0) 4 19 (3.3) 10 (4.7) Missing **Location** Colon 294 (50.7) 116 (54.5) 0.346 Rectum 286 (49.3) 97 (45.5) **Side** Right 147 (25.3) 70 (32.9) 0.035 Left 433 (74.7) 143 (67.1) **Histology** Adenomatous 539 (92.9) 173 (81.6) \<0.0001 Mucinous 41 (7.1) 39 (18.4) Exact Pearson chi-square test for 2×2 tables and exact linear-by-linear association test for tables with ordered variables. Missing data excluded from analyses. 10.1371/journal.pone.0109600.t003 ###### Association between REG4 expression and clinicopathologic parameters in non-mucinous colorectal cancer. {#pone-0109600-t003-3} REG4 expression ----------------- ------------ ------------ ------- **Age. years** \<65 228 (42.3) 76 (43.9) 0.706 ≥65 311 (57.7) 97 (56.1) **Gender** Male 289 (53.6) 101 (58.4) 0.273 Female 250 (46.4) 72 (41.6) **Dukes** A 74 (13.7) 39 (22.5) 0.005 B 186 (34.5) 59 (34.1) C 148 (27.5) 46 (26.6) D 131 (24.3) 29 (16.8) **Grade (WHO)** 1 13 (2.4) 11 (6.4) 0.023 2 375 (70.0) 124 (72.5) 3 132 (24.6) 32 (18.7) 4 16 (3.0) 4 (2.3) Missing **Location** Colon 268 (49.7) 87 (50.3) 0.897 Rectum 271 (50.3) 86 (49.7) **Side** Right 126 (23.4) 49 (28.3) 0.189 Left 413 (76.6) 124 (71.7) Exact Pearson chi-square test for 2×2 tables and exact linear-by-linear association test for tables with ordered variables. Missing data excluded from analyses. REG4 and other intestinal markers {#s3d} --------------------------------- In the subgroup of 220 tumors, we found that REG4 expression significantly associated with higher expression of MUC2, MUC4, and MUC5AC, but not with MUC1 expression. Nor did REG4 expression associate with markers of neuroendocrine differentiation. The same remained true when non-mucinous CRC was analyzed alone ([Table 4](#pone-0109600-t004){ref-type="table"}). In mucinous CRC, REG4 expression associated with no other markers studied (data not shown). 10.1371/journal.pone.0109600.t004 ###### Association of REG4 expression with other biomarkers in colorectal cancer and non-mucinous colorectal cancer. {#pone-0109600-t004-4} REG4 expression ------------------------------------ ------------ ----------- ---------- ------------ ----------- ---------- **MUC1 expression** low 131 (84.0) 36 (78.3) 0.368 127 (84.1) 31 (79.5) 0.492 high 25 (16.0) 10 (21.7) 24 (15.9) 8 (20.5) **MUC2 expression** low 147 (93.6) 24 (48.0) \<0.0001 143 (94.1) 22 (51.2) \<0.0001 high 10 (6.4) 26 (52.0) 9 (5.9) 21 (48.8) **MUC4 expression** neg 90 (57.0) 14 (29.2) 0.001 88 (57.5) 13 (31.7) 0.003 pos 68 (43.0) 34 (70.8) 65 (42.5) 28 (68.3) **MUC5AC expression** neg 150 (96.2) 38 (82.6) 0.004 147 (97.4) 31 (79.5) \<0.0001 pos 6 (3.8) 8 (17.4) 4 (2.6) 8 (20.5) **Neuroendocrine differentiation** Negative 152 (93.8) 45 (90.0) 0.354 147 (93.6) 38 (88.4) 0.323 Positive 10 (6.2) 5 (10.0) 10 (6.4) 5 (11.6) Exact pearson chi-square test for 2×2 tables. Missing data excluded from analyses. Survival analysis {#s3e} ----------------- In non-mucinous, CRC REG4 positivity was a sign of favorable prognosis (p = 0.019, log-rank test); 5-year DSS for patients with positive cytoplasmic REG4 tumor expression was 67.9% (95% CI 60.5--75.3) compared to 57.8% (95% CI 53.5--62.1) for those with no cytoplasmic expression ([Figure 3](#pone-0109600-g003){ref-type="fig"}). In mucinous CRC, no difference appeared (data not shown). When we stratified non-mucinous CRC for patients under or over 65, REG4 expression was a sign of favorable prognosis in patients under 65 (p = 0.049, log-rank test.), with no difference for those older (p = 0.195, log-rank-test). (S2 & S3). {#pone-0109600-g003} Cox regression univariable analyses confirmed these results, with REG4 expression being a sign of a reduced risk of death within 5 years for non-mucinous CRC. A significant difference emerged also for patients under 65 (HR 0.57, 95% CI 0.34--0.94, p = 0.029) but not for those older (HR 0.83, 95% CI 0.57--1.22, p = 0.34). Cox regression multivariable analysis for non-mucinous CRC patients under 65, adjusted for gender, stage, and differentiation, showed that REG4 was an independent factor of favorable prognosis (HR 0.55, 95% CI 0.33--0.92, p = 0.022) ([Table 5](#pone-0109600-t005){ref-type="table"}). 10.1371/journal.pone.0109600.t005 ###### Cox uni-and multivariable analysis of relative risk of death from non-mucinous colorectal cancer within 5 years by REG4 expression for patients under 65. {#pone-0109600-t005-5} REG4 expression HR (95% CI) P-value N(events) HR (95% CI) P-value N(events) ----------------- ------------------- --------- ----------- ------------------- --------- ----------- Negative 1.00 228 (83) 1.00 227 (83) Positive 0.57 (0.34--0.94) 0.029 76 (18) 0.55 (0.33--0.92) 0.022 76 (18) Abbreviations: CI = confidence interval, HR = hazard ratio. Multivariable analysis included adjustment for gender, Dukes class, differentiation grade (G1/2 vs G3/4). Discussion {#s4} ========== Here we show by immunohistochemistry that REG4 expression in non-mucinous colorectal cancer associates with favorable clinicopathological parameters and that REG4 is an independent marker of favorable prognosis in patients under 65. REG4 expression associates with expression of other intestinal markers: MUC1, MUC2, and MUC5AC. REG4 expression was higher in low-stage tumors and in those with of mucinous histology. With mucinous tumors excluded, REG4 expression associated significantly with higher differentiation and low stage. REG4 expression also associated with MUC1, MUC2, and MUC5AC, which supports the finding that REG4-positive tumors are more highly differentiated than are those that are REG4 negative. These results are in accordance with findings of Li et al [@pone.0109600-Li1], who showed in that for REG4, immunohistochemical (IHC) expression in CRC associates significantly with higher differentiation and with absence of venous invasion. Moreover, REG4 expression showed a trend like association with low T-stage, absence of lymph node metastasis, and local disease (Dukes A-B vs C-D). Similar results appear for gallbladder cancer, where positive REG4 IHC expression, associates with higher tumor differention [@pone.0109600-Tamura1]. Controversial results for CRC in Numata et al. show that higher REG4 mRNA expression associates with higher differentiation, deeper invasion (T-stage), lymphatic invasion, presence of liver metastasis, and more advanced stage [@pone.0109600-Numata1]. They, however measured mRNA levels by PCR, not as we did actual protein expression. Association between REG4 levels and carcinoma has been under study in both serum and tissues. Elevated serum concentration of REG4 in carcinoma patients compared to those in healthy controls has been a finding in pancreatic cancer [@pone.0109600-Takehara1], gastric cancer [@pone.0109600-Mitani1] and gallbladder cancer [@pone.0109600-Tamura1], indicating that serum REG4 could serve as a diagnostic biomarker. Increased REG4 IHC expression has been suggested as a marker of poor prognosis in gastric cancer [@pone.0109600-Tao1], and elevated tissue levels of REG4 mRNA may be a marker of poor prognosis in CRC [@pone.0109600-Numata1]. REG4 IHC expression has shown, however, no effect on prognosis in CRC [@pone.0109600-Li1], [@pone.0109600-Zheng1]. One reason for this might be that mucinous and non-mucinous cancers were not analyzed separately. Also differences in antibodies, staining procedures, and analysis of stainings might have differed from our study. In gallbladder cancer, REG4 IHC expression has been associated with better prognosis [@pone.0109600-Tamura1]. Our results show that REG4 IHC expression is a marker of favorable prognosis in non-mucinous CRC, whereas no difference emerged in mucinous CRC. REG4 expression is constitutively high in mucinous tumors (i.e. Pseudomyxoma peritonei, and mucinous cystadenomas and mucocellular gastric cancer) This may explain why no clear variation in REG4 expression was found by IHC in the group of mucinous CRC tumours. Differences in REG4 expression was on the other hand apparent in non-mucinous CRC. In our patients under 65, elevated REG4 was an independent factor for better prognosis in non-mucinous CRC. It thus seems plausible that REG4 mRNA levels may be elevated in CRC patients with poor prognosis, but this is not translated to protein. Further studies are thus warranted to compare REG4 mRNA levels with REG4 IHC case by case. Our results regarding five pairs of REG4-positive primary tumors and their corresponding lymph-node metastases showed that of five lymph nodes, only two showed REG4 expression; this may imply that a REG4-negative subpopulation of tumor cells is more likely to metastasize than REG4-positive cells. REG4, is expressed in inflammatory bowel diseases and also in the margins of peptic ulcers, is considered a marker of inflammation [@pone.0109600-Kmrinen1]. In some of our whole-tissue sections tumor tissue stained negative for REG4, but the adjacent benign epithelium expressed REG4 strongly, apparently representing an inflammatory reaction against the tumor. It is noteworthy that REG4 expression confers a more favorable prognosis only in CRC without mucinous differentiation. In fact, high expression of REG4 has appeared in aggressive forms of gastrointestinal cancer that show mucinous phenotype-like mucocellular (signet ring cell) carcinoma of the stomach [@pone.0109600-Heiskala1]. On the other hand, REG4 is also abundantly present in mucin-rich cystadenomas of the appendix and in its malignant, disseminated form, pseudomyxoma peritonei, which is notoriously therapy resistant. A majority of entero-endocrine cells in normal intestinal mucosa display high physiological expression of REG4 with co-expression of synaptophysin and chromogranin. Intestinal neuro-endocrine tumors of both low and high grade also show REG4 positivity, frequently with a peculiar anatomical distribution with the strongest expression in a single layer of cells in the periphery of the tumor that are in intimate contact with the surrounding stroma [@pone.0109600-Kmrinen1] Given this, it was somewhat intriguing that synaptophysin- and chromogranin-positive CRC tumors in this study remained negative for REG4. Several reports suggest the oncogenic role of REG4 in the development of cancer in the gastrointestinal tract. The ultimate molecular mechanisms have, however, remained elusive. Bishnupari et al [@pone.0109600-Bishnupuri1] reported that treatment of cultured colon adenocarcinoma cells with recombinant REG4 protein induced phosphorylation of the EGF receptor and Akt. They suggested that REG4 is a transactivator of the EGFR/Akt signaling pathway. A further elucidation of the role of exogenous REG4 as a regulator of cell growth potential is, however, awaiting identification of the putative REG4 receptor. No evidence shows that increased expression of REG4 by itself induces cancerous growth, however. Our observations on mice transgenic for human REG4 cDNA under the villin promoter that leads to global overexpression of REG4 in the intestinal mucosa did not induce increased tumor formation or mucosal hyperplasia (unpublished data). Regulation of REG4 expression is still poorly understood. We originally reported strongly up-regulated expression of REG4 also in inflamed IBD mucosa of IBD-like foci of gastritis-induced intestinal metaplasia in the stomach [@pone.0109600-Kmrinen1]. This suggests that inflammatory cytokines may influence REG4 expression. Neuroendocrine differentiation, on the other hand, as seen in enteroendocrine cells, appears to confer constitutive expression of REG4. We recently reported co-expression of REG4 with the neuronal transcription factor Hath-1 (atonal, Math-1) in neuroendocrine tumors [@pone.0109600-Heiskala1]. What remains to be established is whether one regulator of REG4 expression is Hath-1. Compared to whole tissue sections, the TMA technique allows analysis of larger patient cohorts, but with a smaller proportion of the tumors evaluated. This posed no problem here, however, as the REG4 staining pattern in our whole section staining was homogenous. For technical reasons, up to 5% of the specimens were lost in the TMA-production and -staining process. The strength of this study is a large, well-characterized CRC-patient cohort with our long follow-up time. Here we show, to our knowledge for the first time, that REG4 IHC expression is an independent marker of favorable prognosis in non-mucinous colorectal cancer in patients under 65. These results are in disagreement with those obtained by evaluating mRNA levels; our discrepancies with others' findings warrant further studies. Supporting Information {#s5} ====================== ###### Clinicopathologic characteristics of the study population and subgroup population. (DOCX) ###### Click here for additional data file. ###### REG4 expression indicating better prognosis in non-mucinous colorectal cancer for younger patients. Disease-specific survival analysis according to the Kaplan-Meier method for REG4 expression in non-mucinous colorectal cancer in patients under 65 by the log-rank test. (TIFF) ###### Click here for additional data file. ###### REG4 expression showing no difference in survival in non-mucinous colorectal cancer for older patients. Disease-specific survival analysis according to the Kaplan-Meier method for REG4 expression in non-mucinous colorectal cancer in patients 65 and older by the log-rank test. (TIFF) ###### Click here for additional data file. We thank Päivi Peltokangas, Gynel Arifdshan, and Elina Aspiala for their excellent technical assistance. [^1]: **Competing Interests:**The authors have declared that no competing interests exist. [^2]: Conceived and designed the experiments: CH LCA. Performed the experiments: TK JH. Analyzed the data: TK HM. Contributed reagents/materials/analysis tools: SK LCH CH TK. Contributed to the writing of the manuscript: TK JH HM SK LCA CH.

Introduction ============ Telecommunications technologies are changing ways of thinking, acting, and communicating throughout the world and within healthcare. As in any technological area, healthcare and telecommunication definitions and language evolves with developing concepts and technological innovation. An example of this is the distinction between telehealth and telemedicine. Telehealth, one of the oldest areas of applied technology in medicine, refers to the use of electronic information and telecommunications technologies to support long-distance clinical health care, patient and professional health-related education, public health, and health administration. Telemedicine, one of the oldest areas of applied technology in medicine, is described as "the use of electronic information and communications technology to provide and support healthcare when distance separates the participants" ([@b38-v1n1-art-10.5195-ijt.2009.6014]). In an effort to clarify terminology and place telerehabilitation within the larger realm of telehealth, [@b92-v1n1-art-10.5195-ijt.2009.6014] provided a conceptual view of emerging models of telehealth with two major subsets: telemedicine (i.e., delivery of clinical services) and telehealthcare (i.e., management of disability and health). When Winters wrote his now classic article, telerehabilitation was an emerging field, positioned within both telemedicine and telehealthcare. There have subsequently been advances in the conduct of practice, particularly in the areas of physical therapy, occupational therapy, audiology, speech-language pathology, and neuropsychology. We thus propose that telerehabilitation warrants a separate and parallel identity under the "telehealth umbrella" alongside both telehealthcare and telemedicine. There is a growing amount of literature on the use of technology for remote assessment and intervention in medicine ([@b7-v1n1-art-10.5195-ijt.2009.6014]) and rehabilitation ([@b49-v1n1-art-10.5195-ijt.2009.6014]; [@b83-v1n1-art-10.5195-ijt.2009.6014]; [@b92-v1n1-art-10.5195-ijt.2009.6014]). Rehabilitation providers may not be aware of all telerehabilitation options available via innovative healthcare technologies. Additionally, they may not be fully aware of potential challenges inherent to technology application. These issues present an obstacle for the agency, individual provider, or consumer who would like to consider implementing telerehabilitation for a particular environment, purpose, or disability group that may not necessarily match available descriptions or what they know/currently access and use. For telerehabilitation to best benefit the end-user (an individual with a disability), all parties involved need to have access to the greatest possible set of available options to choose what will most likely work for the consumer and the environment in which they function. As telerehabilitation services continue to grow as a complement to traditional face-to-face clinical services, there is an increasing need to standardize appropriate clinical uses, reimbursement, and health care policy regarding the use of telerehabilitation. Review of Science ================= Impetus for Telerehabilitation ------------------------------ Remote areas often experience shortages of professionals and technical resources crucial to the delivery of services related to specialized medical fields ([@b16-v1n1-art-10.5195-ijt.2009.6014]). These shortages negatively affect both health care providers and patients. Rural providers are often isolated from medical advancements and technologies used in larger metropolitan centers. As a result, when an individual in a rural area needs an assessment and/or specific treatment, he or she may have to travel long distances to receive the specialized healthcare necessary to address their needs. Studies have reported that 50% of veterans travel more than 25 miles for healthcare ([@b69-v1n1-art-10.5195-ijt.2009.6014]; [@b93-v1n1-art-10.5195-ijt.2009.6014]). [@b37-v1n1-art-10.5195-ijt.2009.6014] conducted a study within the Veterans Health Administration which determined that veterans with multiple sclerosis have significant barriers to care as a result of their disability. Twenty percent of patients surveyed reported that parking, distance, or transportation difficulties significantly interfered with their receiving treatment. Furthermore, for individuals with sensation issues (e.g., spinal cord injury) prolonged sitting during travel can carry the potential risk of worsening a sore or decubitus ulcer ([@b73-v1n1-art-10.5195-ijt.2009.6014]). For these reasons, many individuals delay or avoid necessary treatment. While issues of access are clearly magnified in rural areas, mobility restrictions and accessibility problems also decrease the quality of healthcare for individuals located in urban areas ([@b38-v1n1-art-10.5195-ijt.2009.6014]). Appropriate selection and application of telerehabilitation technology may be conceptualized as a clinical reasoning task, since appropriate use of telerehabilitation requires assessment of individual needs and environmental factors; consideration of diagnostic issues; implementation of an intervention; and follow-up to determine efficacy. (A related white paper on infrastructure and technology provides more detail about the human and economic factors as well as telerehabilitation technology.) Human factors focus on system usability and designing system interfaces to optimize the user's ability to accomplish their tasks error-free in a reasonable time ([@b8-v1n1-art-10.5195-ijt.2009.6014]). Human factors is an applied science that takes research about human abilities, limitations, behaviors, and processes, and uses this knowledge as a basis for the design of tools, products, and systems ([@b13-v1n1-art-10.5195-ijt.2009.6014]). This type of diagnostic reasoning has been conceptualized as a complex, dynamic process based on hypothesis testing ([@b80-v1n1-art-10.5195-ijt.2009.6014]) suggesting that clinical reasoning must match consumer experiences with clinical knowledge. Since the late 1950's technologists and clinicians have explored the use of advanced telecommunications and information technologies as a way of bridging the gap between individuals with specialized medical needs living in remote areas and the source of specialty care ([@b10-v1n1-art-10.5195-ijt.2009.6014]; [@b39-v1n1-art-10.5195-ijt.2009.6014]; [@b46-v1n1-art-10.5195-ijt.2009.6014]). Again, telemedicine is described as "the use of electronic information and communications technology to provide and support healthcare when distance separates the participants" ([@b38-v1n1-art-10.5195-ijt.2009.6014]). Telemedicine has been applied in many areas, often first through smaller-scale feasibility projects, and later in larger-scale clinical deployments such as cardiology ([@b20-v1n1-art-10.5195-ijt.2009.6014]), dermatology ([@b53-v1n1-art-10.5195-ijt.2009.6014]), neurosurgery ([@b65-v1n1-art-10.5195-ijt.2009.6014]), pathology ([@b5-v1n1-art-10.5195-ijt.2009.6014]), radiology ([@b12-v1n1-art-10.5195-ijt.2009.6014]), oncology ([@b1-v1n1-art-10.5195-ijt.2009.6014]), and space exploration ([@b27-v1n1-art-10.5195-ijt.2009.6014]). For further information, [@b48-v1n1-art-10.5195-ijt.2009.6014] summarizes the state-of-the-art for clinical applications in telemedicine and telehealth. The rehabilitation field has been gradually integrating telecommunication tools into clinical practice. The benefits of using telerehabilitation include: 1) decreased travel between rural communities and specialized urban health centers; 2) better clinical support in local communities; 3) improved access to specialized services; 4) delivery of local health care in rural communities; 5) indirect educational benefits for remote clinicians who participate in teleconsultations; 6) reduced feelings of isolation for rural clinicians; 7) improved service stability in regions with high staff turnover; and 8) multimedia communication ([@b49-v1n1-art-10.5195-ijt.2009.6014]). Worldwide, the business of telehealth doubled from a \$6.8 million industry in 1997 to a \$13.8 million industry in 1998 ([@b74-v1n1-art-10.5195-ijt.2009.6014]). Experts have forecasted that by 2010, at least fifteen percent of health care services worldwide will be provided via telehealth ([@b78-v1n1-art-10.5195-ijt.2009.6014]). Summary of Telerehabilitation Literature ======================================== Assistive Technology Applications --------------------------------- Telerehabilitation is the application of telecommunication technology that provides distant support, assessment and intervention to individuals with disabilities ([@b71-v1n1-art-10.5195-ijt.2009.6014]). Telerehabilitation offers many new opportunities to provide rehabilitation services in alternative ways and in different clinical settings ([@b24-v1n1-art-10.5195-ijt.2009.6014]; [@b49-v1n1-art-10.5195-ijt.2009.6014]). [@b24-v1n1-art-10.5195-ijt.2009.6014], from the University of Pittsburgh, discussed the potential of rapid improvement in telecommunications technology to improve access to assistive technology (AT) services for people with disabilities. Assistive technology commonly refers to "...products, devices, or equipment, whether acquired commercially, modified or customized, that are used to maintain, increase or improve the functional capabilities of individuals with disabilities..." according to the definition proposed in the Assistive Technology Act of 1998. Telerehabilitation offers a diversity of clinical applications. These include: 1) consultation by clinical rehabilitation engineers or specialized clinicians for seating and positioning; 2) the provision of assistive technology using simple Plain-old Telephone Service (POTS) videophones; 3) physicians and nurses performing pressure sore management using either higher-quality camera images or lower-quality images from interactive systems; 4) remote therapy using tools such as EMG-controlled games for stroke rehabilitation, or remote interactive story retelling for brain injury rehabilitation; 5) remote rehabilitation management or teleconsultation by physiatrists; and, 6) specialized clinicians for clinics using group videoconferencing systems over established telemedicine networks ([@b92-v1n1-art-10.5195-ijt.2009.6014]). [@b15-v1n1-art-10.5195-ijt.2009.6014], from Shepherd Center in Atlanta, described the experiences of a specialty hospital serving people with disabilities in exploring telerehabilitation to support assistive technology in the home. The article described four specific case studies to illustrate how telerehabilitation was used in relation to seating evaluation, home accessibility, setup of computer access systems, and augmentative communication device training. Each of the case studies used low-cost video telephone such as the AT&T's Picasso Still Image Video Phone, and American Telecare Inc.'s Aviva 1010 and PTS-2 systems. All the devices transmitted simultaneous audio and video over standard telephone lines. The authors limited their choice to systems using the standard telephone lines because nearly everyone in this research study had access to a phone line. The respective objectives of the four case studies were the following: 1) to observe and assess the seating posture and the effectiveness of a patient's weight shift while seated in their recliner; 2) to evaluate an individual's strengths and limitations within the context of their existing bathroom structure and provide recommendations for adaptive equipment; 3) to provide recommendations for a 23 year old man with tetraplegia for a switch mounted system for computer access; and 4) to provide training to a 57 year old man with cerebral palsy on the use of his augmentative communication device. The case studies showed that telerehabilitation has great promise for expanding availability, accessibility, and affordability of services for people with disabilities. Each scenario, however, demonstrated limitations. During the evaluations, both the audio and video were not ideal for the conditions due to insufficient POTS line bandwidth, as well as the dim lighting for video capture. The authors found that the clinical staff members who provided the respective consults needed more time to familiarize themselves with the technology before mailing it to the clients. In recent years, the availability of high-speed Internet in the home has grown tremendously. Internet-based systems designed as plug-play appliances that do not require a computer are becoming more common. However, POTS devices are still the most prevalent resource for providing home tele-services. Therefore, while considering advanced internet-based applications, the developers and manufacturers of telerehabilitation technologies must also consider compatibility with POTS lines. Prevention of Pressure Ulcers ----------------------------- Several studies have examined the feasibility of preventing pressure sores and promoting pressure relieving activities ([@b9-v1n1-art-10.5195-ijt.2009.6014]; [@b57-v1n1-art-10.5195-ijt.2009.6014]; [@b58-v1n1-art-10.5195-ijt.2009.6014]; [@b72-v1n1-art-10.5195-ijt.2009.6014]; [@b79-v1n1-art-10.5195-ijt.2009.6014]; [@b87-v1n1-art-10.5195-ijt.2009.6014]). [@b67-v1n1-art-10.5195-ijt.2009.6014], from Emory University, studied the use of a telerehabilitation intervention to promote skin and other self-care activities for people with spinal cord injuries. This was a non-controlled clinical series where researchers used a videophone that would transmit video and still images over a standard telephone line. The overall impression of the eleven clients was noted to be positive, however, there were some problems reported with equipment, skin inspection, and safety. Diagnosing skin conditions through verbal description alone was not reliable, and the remote clinician wanted to be in the same room face-to-face to examine the skin. In another study, [@b68-v1n1-art-10.5195-ijt.2009.6014], conducted a non-randomized controlled prospective trial with thirty-seven patients. The authors examined injured spinal cord patients after hospital discharge via three different approaches (i.e. telephone, video, and standard care) to determine which approach was associated with the lowest incidence of pressure ulcers and the fewest re-hospitalizations. The results from this pilot study indicated that the video group reported the highest number of ulcers and did so accurately, while the standard care reported the lowest number of ulcers. A larger sample is required for a more comprehensive analysis. The authors advised that prior to discharge, patients will require additional training on how to set-up and use the equipment to minimize the costs of clinician and technician travel to the patients' homes. Virtual Reality Applications ---------------------------- Virtual Reality (VR) is becoming a practical, affordable technology for the practice of clinical medicine. Modern high fidelity VR systems have practical applications in areas ranging from psychiatry to intervention and rehabilitation ([@b11-v1n1-art-10.5195-ijt.2009.6014]). VR's capacity to allow for creation and control of three- dimensional built environments offers clinical assessment and rehabilitation options that are not available with traditional methods ([@b76-v1n1-art-10.5195-ijt.2009.6014]). Virtualized reality and three-dimensional reconstruction technology provided an effective means to investigate the architectural features of a built environment without an expert visiting the site for individuals who use a wheeled mobility device ([@b44-v1n1-art-10.5195-ijt.2009.6014]; [@b45-v1n1-art-10.5195-ijt.2009.6014]). [@b36-v1n1-art-10.5195-ijt.2009.6014] applied two virtual environments to the assessment and training of inexperienced powered wheelchair users and demonstrated that the two virtual environments represent a potentially useful means of assessing and training novice powered wheelchair users. The University of Medicine and Dentistry of New Jersey designed the remote console (ReCon) ([@b50-v1n1-art-10.5195-ijt.2009.6014]), a telerehabilitation system designed to provide therapists, at a remote location, the tools necessary to oversee patient's rehabilitation session in real-time. This system provides the therapist with three-dimensional representations of patients' movements, VR-based exercise progress, and performance updates while the patient is exercising. During the session, the therapist evaluates the patient's performance and either modifies the current exercise, or sets up the next one. The remote therapist is also provided with tools for audio and video communication with the local site and chat communication with the local therapist. Researchers conducted both usability and evaluation studies to refine the system ([@b50-v1n1-art-10.5195-ijt.2009.6014]; [@b51-v1n1-art-10.5195-ijt.2009.6014]). [@b84-v1n1-art-10.5195-ijt.2009.6014] described the value of VR systems for the investigation and rehabilitation of cognitive and perceptual impairments and discussed current and political applications of VR technology. The following were the neurorehabilitation issues: (1) attention and the reduction of distraction; (2) assessment and remediation of executive function deficits; (3) investigation of impairments of coordinated movement; (4) study and rehabilitation of aphasia and other severe disorders of language; (5) task presentation for functional imaging studies of the brain; and (6) the measurement of mental load in the operation of assistive technology. Speech-Language Pathology Applications -------------------------------------- Speech language pathology (SLP) and audiology are clinical services related to the identification, assessment and management of hearing and communication disorders. SLP is ideally suited for telerehabilitation as the client-clinician interaction is primarily visual and verbal. Many of the assessment and treatment materials can be administered via computerized programs. The Mayo Clinic was one of the earliest to incorporate teleconsultations into SLP which provided a viable alternative to the traditional face-to-face assessment ([@b28-v1n1-art-10.5195-ijt.2009.6014]). Researchers at the National Rehabilitation Hospital in Washington, D.C. developed a custom software package called RESPECT (REmote SPEech language and Cognitive Treatment) that investigates the role of interactive data sharing during teleSLP cognitive communicative treatment. RESPECT augments and extends therapeutic interaction with the following capabilities: virtual desktop, real time shared interaction, work processing documents, scanned workbook pages, computer applications, digital drawing whiteboards, and combined audio/video conferencing where the clinician controls clients' system. Research has established the validity of using the system in a story retelling between face-to-face, (i.e., in-person assessment), and remote telerehabilitation sessions, with a sample of 40 subjects with brain injury ([@b14-v1n1-art-10.5195-ijt.2009.6014]). In addition to equivalent performance between settings, a high level of acceptance of telerehabilitation technology was found regardless of subjects' ages, educational levels, and technology backgrounds. The Telerehabilitation Research Unit at the University of Queensland (<http://www.uq.edu.au/telerehabilitation/>) is another pioneering institution. Their research involves the assessment and treatment for individuals with acquired neurological speech and language disorders. One of their many research projects reported significant improvements in speech for individuals diagnosed with Parkinson's Disease using the Lee Silverman Voice Treatment via the internet ([@b82-v1n1-art-10.5195-ijt.2009.6014]). Other examples include voice therapy at Tripler Army Base in Hawaii ([@b56-v1n1-art-10.5195-ijt.2009.6014]); remote dysphagia evaluations ([@b31-v1n1-art-10.5195-ijt.2009.6014]; [@b66-v1n1-art-10.5195-ijt.2009.6014]); and augmentative and alternative communication evaluations ([@b62-v1n1-art-10.5195-ijt.2009.6014]). Seating and Wheeled Mobility Applications ----------------------------------------- Several studies analyzed the use of telerehabilitation in the field of seating and mobility. Related to seating and positioning, telerehabilitation has the potential to provide evaluation, treatment intervention, and follow-up as needed in the client's home or at a local clinic ([@b67-v1n1-art-10.5195-ijt.2009.6014]). Assessment in the client's home is especially important considering that the use of assistive technologies (e.g., wheelchairs) is only as effective as the ability of the individual to use it in their own environment. Research has demonstrated that the use of telerehabilitation can help promote community re-entry and improve the quality of life of the individual ([@b67-v1n1-art-10.5195-ijt.2009.6014]). The Glenrose Rehabilitation Hospital in Alberta Canada presented a study design and protocol to evaluate the effectiveness and efficiency of using telehealth to provide seating assessment and intervention by comparing groups of clients in three conditions: 1) clients residing in Capital Health assessed in-person; 2) clients from out-of-region assessed in-person; and 3) clients from out-of-region assessed by telehealth whereby results have not been published ([@b52-v1n1-art-10.5195-ijt.2009.6014]). A qualitative case study was conducted among rehabilitation professionals for implementing and planning a telehealth seating clinic ([@b43-v1n1-art-10.5195-ijt.2009.6014]). The study showed that when implementing such a clinic, the involvement of multidisciplinary teams and proper visualization and communication between participants is essential. [@b55-v1n1-art-10.5195-ijt.2009.6014] compared videoconferencing using ordinary POTS lines with videoconferencing using Integrated Services Digital Network (ISDN) lines. Over six months, an occupational therapist completed eight seating and wheelchair mobility evaluations. Four clients were evaluated via videoconferencing using the POTS line, and four clients using an ISDN line. Despite challenges presented via the technology used at the time (i.e., lower quality video afforded by POTS lines such that lower data communications rate led to longer still picture transfer times and jerkier video images than achieved with the ISDN connections) the primary condition and major problem were correctly identified in all cases. This work showed that with advancements in telecommunication technology, telerehabilitation systems could have the potential to greatly affect how services are delivered and to carefully determine the best and most appropriate AT device for the client. [@b25-v1n1-art-10.5195-ijt.2009.6014] compared the type of wheelchair the person actually uses to the wheelchair recommended via telerehabilitation and in-person assessments. Clinicians utilizing telerehabilitation demonstrated a high level of agreement in recommending the same basic type of wheelchair that subjects already owned, showing a high level of agreement in the consistency of wheelchair recommendation. Therefore, telerehabilitation is a potentially useful tool for wheelchair recommendation. Vocational Rehabilitation Applications -------------------------------------- Individuals with significant disabilities experience challenges to achieving vocational goals and employment outcomes. With current census data indicating that the employment rate of working age people with disabilities is 37.5 percent in 2004, access to effective rehabilitation resources and services is of paramount importance ([@b70-v1n1-art-10.5195-ijt.2009.6014]). Persons with disabilities may experience a complex array of functional limitations that, along with external barriers (accessibility, attitudinal, lack of resources), impact their ability to perform effectively in vocational training and employment settings, as well as in social, recreational, and independent community living. Vocational and independent living rehabilitation services have evolved and have been demonstrated through research and practice to be more effective in achieving employment and community independence outcomes. However, due to limited availability of services, and cost and accessibility challenges, not everyone with a disability has access to resources and services that have demonstrated efficacy and are specific to their interests and needs. Telerehabilitation may provide a mechanism to enable individuals with disabilities to gain access to effective vocational rehabilitation services, regardless of limitation imposed by geography and local resource capabilities. This section will focus on telerehabilitation applied to the needs of persons with disabilities in the realm of employment. With respect to employment, this involves vocational assessment; preparing for employment (such as education and training, development of pre-vocational skills and competencies); seeking and obtaining employment (developing job seeking skills, participating in peer or group job search clubs); maintaining employment (on and off site job coaching and supported employment); and monitoring and follow-up (ongoing assessment of performance and consultation in meeting new job demands or addressing obstacles). Despite the potential and need for telerehabilitation in vocational rehabilitation, very few applications have been reported. [@b84-v1n1-art-10.5195-ijt.2009.6014] described the results of a series of focus groups that investigated consumer acceptance of remote monitoring identification of roles for video technology in supported employment. The focus groups provided access to technologies and solicited input and feedback. Results indicated that video technologies were thought to have potential as a medium for vocational support. Specific recommended applications included job development, client monitoring and real time support and reassurance, and instruction/cueing. The video technology was thought to be more useful than audio only in working with individuals with cognitive impairments, and was viewed by the respondents as providing greater flexibility in coverage for vocational support providers. Potential drawbacks included employer technophobia and privacy concerns. Several researchers have indicated the value of telemedicine or telerehabilitation in enhancing the delivery of more traditional rehabilitation services. [@b21-v1n1-art-10.5195-ijt.2009.6014] reported that telemedicine enables the delivery of cognitive prosthetic services by a therapist in the home. [@b71-v1n1-art-10.5195-ijt.2009.6014] reported survey results that individuals with acquired brain injury expressed strong interest in a variety of potential Internet-based rehabilitation services such as memory, attention, problem solving and assistance with activities of daily living. A study designed to test how telerehabilitation technology might be accepted and effectively utilized by persons with cognitive disabilities assessed the effect of high and low bandwidth videoconferencing on persons with known attention, perceptual processing, and language comprehension problems, ([@b60-v1n1-art-10.5195-ijt.2009.6014]). This was done through administration of a battery of psychometric instruments via video teleconferencing in face-to-face, low and high bandwidth conditions. Results revealed that in a sample of 15 individuals with cognitive disability, there was a significant difference between face-to-face assessment and low and high bandwidth on only a complex memory task involving list learning. There was no difference on tests of understanding oral directions, repeating number series, and on tasks involving controlled fluency. These findings suggest that despite information processing limitations in this population and apparent fluctuations in audio and video quality, tele-videoconferencing (including low bandwidth applications) is an effective mechanism for communication. Findings supported the use of telerehabilitation as a potential means of support and rehabilitation intervention with individuals with cognitive disabilities. [@b14-v1n1-art-10.5195-ijt.2009.6014] conducted a study to measure performance by 40 brain-injured subjects with medical diagnoses of stroke or traumatic brain injury on a standardized speech-language pathology evaluation conducted in both face-to-face and videoconference-based telerehabilitation settings. The Story Retelling Procedure (SRP), which measured connected language production and comprehension of spoken narratives, was administered to each subject in both settings. The objectives of the study were to: (1) compare communication as measured by the SRP between experimental settings; and (2) determine if variables such as age, education, technology experience or gender had an effect on performance between settings. The results concluded that no significant difference was found between SRP performances measured in the two settings. The variables of interest (i.e., age, education, technology experience, and gender) did not significantly affect the difference between performances in the two settings. Additionally, the subjects reported a high level of acceptance of using the videoconferencing system when asked if they would use it again to talk with their clinician. [@b19-v1n1-art-10.5195-ijt.2009.6014] conducted a pilot study using remote technologies to provide remote coaching aimed at assessing the feasibility of participants' vocational goals. Remote technology used to deliver services included phone calls, teleconferencing, email, and web-based application sharing that enabled the participant to see the coach's computer screen. Coaching focused on helping participants to clarify their employment goals by contacting experts in the community to learn more about the demands of their target job; evaluating their skills against the demands of the job; and mapping out specific action steps to pursue the "best fit" jobs. A concise feasibility plan summarizing this information was created through the coaching sessions and delivered to the participants and their counselors. Participants were able to achieve specific benchmarks (such as interviewing a community expert) toward determining the feasibility of their employment goals and did so within a very tight time frame. The overall ratings of the project by participants and the use of technology to deliver distance services were generally very positive. By contrast, change in pre and post ratings of clarity of employment goals and understanding of job demands via participant and counselor surveys were minimal, or even at times negative. The authors speculated that increased information about the actual demands of the target job gleaned from the remote coaching might have led to greater feelings of uncertainty. [@b61-v1n1-art-10.5195-ijt.2009.6014] conducted a pilot study to evaluate the potential of a number of remotely implemented telecommunication technologies to enable individuals with cognitive disabilities to succeed in the academic or work environment. In this particular study, several telecommunication technologies were implemented through a wireless Phone/PDA by persons with cognitive disabilities, accompanied by face-to-face and remote training interventions, and tested. The protocol evaluated cellular phone technology (wireless phone and text messaging) and Internet technology (instant messaging, file sharing and e-mail). Results revealed modest gains from the technology. Barriers, including technical difficulties and an overestimation of the subjects' level of functioning, appeared to limit overall success. Authors reported that that distance support technologies had a positive effect on everyday functioning. In particular, text and instant messaging showed the greatest potential for providing rehabilitation supports in vivo to persons with cognitive disabilities ([@b61-v1n1-art-10.5195-ijt.2009.6014]). In summary, while there have been few studies published on the use and efficacy of telerehabilitation applied to vocational rehabilitation, the pilot studies that exist demonstrate potential for broader application of remote technologies to meet the vocational rehabilitation and employment support needs of persons with disabilities. Cost-Effectiveness ------------------ The cost effectiveness of "tele" projects is being studied as there is considerable controversy about how to measure the cost of these efforts ([@b47-v1n1-art-10.5195-ijt.2009.6014]; [@b54-v1n1-art-10.5195-ijt.2009.6014]; [@b63-v1n1-art-10.5195-ijt.2009.6014]; [@b91-v1n1-art-10.5195-ijt.2009.6014]). Critical reviews of the cost-effectiveness and cost-benefit literature have been published ([@b29-v1n1-art-10.5195-ijt.2009.6014]; [@b35-v1n1-art-10.5195-ijt.2009.6014]; [@b54-v1n1-art-10.5195-ijt.2009.6014]; [@b90-v1n1-art-10.5195-ijt.2009.6014]). Although these studies provide evidence that telemedicine may be cost-effective, the cost of providing care within a participating facility increases as a result of additional costs of equipment, transmission lines, additional personnel, and program administration. The question of cost-effectiveness remains unanswered for most of the telemedical services that are developed worldwide because of objective and subjective problems ([@b40-v1n1-art-10.5195-ijt.2009.6014]). Telerehabilitation may not only provide cost-effective treatment options to patients, but may also permit more convenient training of healthcare professionals ([@b16-v1n1-art-10.5195-ijt.2009.6014]; [@b26-v1n1-art-10.5195-ijt.2009.6014]; [@b34-v1n1-art-10.5195-ijt.2009.6014]; [@b42-v1n1-art-10.5195-ijt.2009.6014]; [@b49-v1n1-art-10.5195-ijt.2009.6014]; [@b81-v1n1-art-10.5195-ijt.2009.6014]; [@b94-v1n1-art-10.5195-ijt.2009.6014]). Clinical and Policy Issues ========================== Telerehabilitation has and will continue to encounter considerable resistance or barriers as it moves from the perimeter to the mainstream of healthcare over the coming years. The intent of this white paper is not to address in-depth discussion on policy issues but rather to provide an overview of current clinical concerns associated with telerehabilitation and assistive technology. Some of the questions that have arisen include: How and who is going to pay for reimbursement?Who is qualified to perform such assessments, treatment intervention, and routine service delivery?How is quality assessed? Quality must be defined based on the technology (i.e., signals and data) and the clinical quality of the services (i.e., outcomes). Additional concerns are issues of confidentiality; how remote service delivery fits into the rehabilitation professions' respective codes of ethics; the paucity of scientific evidence; the lack of standards and guidelines; and questions regarding credentialing and interstate licensing regulations. Telerehabilitation is advancing at a time when health insurance providers are reimbursing less for most services and products. Currently, telerehabilitation services are not expected to be fully covered because full coverage for most health services is no longer the norm. On May 22, 2008, Representative C. Michael Thompson and fourteen other co-sponsors introduced H.R. 6163, the Medicare Telehealth Enhancement Act of 2008. The proposed bill would have: Expanded Medicare reimbursement for telemedicine into urban areas;Authorized a study on store-and-forward telemedicine;Expanded originating sites for Medicare to include skilled nursing facilities, dialysis centers and community mental health centers;Added PTs, OTs, and SLPs as telehealth providers under Medicare; andEncouraged the adoption of reciprocity agreements for licensure across state lines. Certain parts of this proposed bill were included in a comprehensive Medicare bill, H.R.6331, but the reimbursement for the rehabilitation providers was not. Unfortunately, the legislative process is often an incremental one, and it is very common to get only a little of what was wanted and thus need to be proactive in upcoming years. Medicare payment of telemedicine and telehealth services is divided into three areas: 1) remote patient face-to-face services seen via live videoconferencing; 2) non face-to-face services that can be conducted either through live video-conferencing or via store-and-forward telecommunication service; and 3) home telehealth services. In order for telerehabilitation to be accepted for reimbursement by third party payers, government officials want to see sound scientific research. Research in telerehabilitation is in its infancy, with only a handful of equivalence trials. As of 2006, most peer-reviewed research articles are case reports of pilot programs or of testing new equipment. Researchers need to conduct many more controlled experiments and present evidence to clinicians and payers that telerehabilitation is clinically effective. [@b64-v1n1-art-10.5195-ijt.2009.6014] assessed current payment practice for telerehabilitation in state Medicaid programs which revealed that only four states at that time were providing reimbursement (Hawaii, Louisiana, Minnesota, and Nebraska). For more information, the [@b18-v1n1-art-10.5195-ijt.2009.6014] prepared an in-depth written report on the current state of telemedicine reimbursement for the Office for the Advancement of Telehealth. Another issue of importance is licensure. Telerehabilitation challenges the traditional practice involving a face-to-face encounter between rehabilitation professional and patient. Telerehabilitation breaks the physical link and complicates decisions about where a telerehabilitation clinician should be licensed if the clinician and patient are located in different states ([@b32-v1n1-art-10.5195-ijt.2009.6014]; [@b33-v1n1-art-10.5195-ijt.2009.6014]). Rehabilitation professionals are typically required to be licensed by the state in which they are practicing their profession. This creates a problem and a dilemma when telerehabilitation services are delivered to a patient across state lines. However, telerehabilitation is not unique, as all telemedicine applications face similar challenges. State practice acts have not outlawed telemedicine, as these laws were passed before any state legislator could imagine its development. Obtaining a professional state license requires the completion of paperwork and the payment of upfront costs and renewal fees. Multiple state licenses can thus involve considerable time and paperwork ([@b22-v1n1-art-10.5195-ijt.2009.6014]). Practitioners need to continue conducting and researching the application of telerehabilitation. They also need to present their findings to their respective professional associations and advocate to government officials at both the state and federal levels. For more information, the [@b17-v1n1-art-10.5195-ijt.2009.6014] prepared an in-depth written report on the current state of telemedicine licensure for the Office for the Advancement of Telehealth. Professional Organizations' Views on Telerehabilitation ======================================================= Discipline-specific professional organizations such as the American Speech-Language and Hearing Association (ASHA), American Occupational Therapy Association (AOTA), American Physical Therapy Association (APTA), and the Commission on Rehabilitation Counselor Certification (CRCC) have different views on what constitutes telerehabilitation. Telerehabilitation is a relatively new application, and the degree to which each organization has acknowledged and supported telerehabilitation varies, as does their level of activity and involvement in telehealth activities. There needs to be increased collaboration and cohesiveness among professional rehabilitation associations concerning telerehabilitation. ASHA uses the term "telepractice" to refer to "the application of telecommunications technology to deliver professional services at a distance" ([@b4-v1n1-art-10.5195-ijt.2009.6014]). Since 1998, ASHA has studied the potential impact of telepractice on speech-language pathologists (SLPs) and audiologists and the individuals they serve ([@b3-v1n1-art-10.5195-ijt.2009.6014]). ASHA has published an abundant amount of literature related to the field of telehealth, including a formal position statement, technical reports ([@b4-v1n1-art-10.5195-ijt.2009.6014]) and issue briefs ([@b3-v1n1-art-10.5195-ijt.2009.6014]) that summarize evidence to date about the use of telepractice in SLP and audiology and discuss future directions and research. The first documented use of distance programs in speech-language pathology (SLP) was through a grant program in the mid-1970s at the Birmingham Veterans Administration Hospital to explore "tele-communicology" as a potential solution to serving patients in remote locations ([@b86-v1n1-art-10.5195-ijt.2009.6014]). The National Rehabilitation Hospital in Washington D.C. and the University of Queensland in Australia have emerged as two of the leaders in studying telerehabilitation activity related to speech-language pathology and audiology. APTA has released a Board of Directors Position on telehealth and reference is made in the position on definitions of telehealth and electronic communications (American Physical Therapy Association Government Affairs). There are also articles explaining how physical therapists are using telehealth to overcome barriers of distance and time. Kathy Lewis, Past-President of the APTA Section on Health Policy and Administration's Technology Special Interest Group, observed that many forms of technology use are increasing in physical therapy practice. She also stated, "It (technology) can increase our practice scope, and it is a tool. I hope that more PTs will take an interest in technology and how it can help us improve the care we provide." AOTA has also published a position paper on telerehabilitation which presents both their stance and the literature supporting methods of service delivery for evaluation ([@b77-v1n1-art-10.5195-ijt.2009.6014]); intervention ([@b87-v1n1-art-10.5195-ijt.2009.6014]); consultation ([@b88-v1n1-art-10.5195-ijt.2009.6014]); education; and the supervision of students and other personnel ([@b41-v1n1-art-10.5195-ijt.2009.6014]). Telerehabilitation, as defined by the AOTA position paper, is the clinical application of consultative, preventative, diagnostic, and therapeutic services via two-way interactive telecommunication technology ([@b89-v1n1-art-10.5195-ijt.2009.6014]). The Canadian Association of Occupational Therapists) is another professional association that has a position statement about telehealth and tele-occupational therapy. CAOT recognizes the ongoing development of tele-occupational therapy, which will promote opportunities for effective, efficient, and accessible occupational therapy services, education and resources to all Canadians. The Commission on Rehabilitation Counselor Certification (CRCC) is the independent credentialing body for vocational rehabilitation and rehabilitation counseling. While CRCC has not established a formal position on telerehabilitation, it has recently adopted a significant revision to the ethics code (a joint code developed jointly by CRCC and the two major professional organizations representing rehabilitation counseling, the National Rehabilitation Counseling Association -- NRCA, and the American Rehabilitation Counseling Association -- ARCA). The revision contains considerable new practice guidelines regarding the use of technology in assessment and counseling. An entire section of the new code is devoted to technology and "distance counseling." Among other areas, the new guidelines state that rehabilitation counselors are held to the same level of expected behavior as defined by the ethics code regardless of the technology used (e.g., cellular phones, email, facsimile, video, audio, audio-visual) or its application (e.g., assessment, research, data storage). It provides detailed practice guidelines on issues related to problematic use of the Internet, privacy, confidentiality and security. Contributions from the State-of-the-Science on Telerehabilitation ================================================================= A "virtual" State-of-the-Science (SOS) conference on Telerehabilitation was held November 17--20, 2008. On November 18, 2008 the topic of Clinical and Vocational Applications for Assistive Technology was presented. This manuscript was used as a platform for discussion and to gain input from attendees. The one-day conference session was in the form of author presentations supported by online PowerPoint presentations and video and case materials, and included a response to the manuscript by four invited experts representing clinicians, researchers, and policy strategists in clinical and vocational rehabilitation. During the SOS conference, 110 individuals attended the session on telerehabilitation clinical and vocational applications. Of the 110 participants, 33% identified themselves as rehabilitation vocational counselors, 15% were physical or occupational therapists, and 8% were speech language pathologists. One psychiatrist and two psychologists attended the session. The remaining group (38%) consisted of individuals from other non-clinical fields, such as engineering or policy. A large proportion of participants (39%) reported that they either never heard of telerehabilitation prior to the conference, or had heard the term but were not familiar with it. Better than one-third (34%) reported that they were well versed in or have participated in research and/or clinical telerehabilitation activities. Roughly one-fifth of the participants reported that they had participated in a telerehabilitation activity related to vocational rehabilitation or employment. External Contributions ---------------------- As part of the State of the Science (SOS) Conference on Telerehabilitation sponsored by the Rehabilitation Engineering Research Center on Telerehabilitation (RERC-TR) at the University of Pittsburgh, the original draft of this white paper was reviewed by two external experts. In addition, as a component of the conference proceedings, a panel of four content experts provided commentary and recommendations; their input is highlighted below. Research Recommendations ------------------------ - It is important to operationalize models of telerehabilitation and in particular to use outcomes based clinical research, patient satisfaction, and/or employer based satisfaction whenever possible. - When examining the potential issues of privacy and security, use existing standards such as those available within the financial services arena. - There is a body of literature on "digital inequities" that looks at discrete socioeconomic factors that influence the way individuals use technology; this literature should be considered in clinical research on telerehabilitation. - In order to examine and explore reimbursement strategies, the need to continue evidence-based research is evident; however, the need to identify a political champion is just as important. - With so many clinical problems and/or issues within the vocational rehabilitation area, we should move to a conceptual model that defines and identifies different technological solutions for specific functional problems and categories within service provision. - Potential research collaborators for the future should include the Job Accommodation Network (JAN), state vocational rehabilitation centers, Veterans Affairs, and school systems. Clinical and Technical Recommendations -------------------------------------- - There are certainly more clinical and vocational opportunities and future directions when looking at telerehabilitation as a developing science. - Since there is "no one size fits all" strategy, it is imperative to have a solid background in fundamental principles that keep the end-user in mind. - One of the major clinical challenges is to determine how we can best integrate telerehabilitation technology into the clinical environment and work flow of the particular service. - The integration of clinical components into one portal system for a broad spectrum of support is an ideal and efficient way to provide service. - The nursing licensure compact is a document of particular interest when examining licensure laws across state lines - Keep the technology simple and practical; this will increase user acceptance. Preliminary Findings ==================== The rapid improvements in telecommunications technology have the potential to improve the delivery of services to people with disabilities and those who are elderly. Transmission of voice, image, and data could provide a means for experts in wheeled mobility and other rehabilitation fields to provide consultation to other healthcare professionals and consumers ([@b24-v1n1-art-10.5195-ijt.2009.6014]). Telerehabilitation may prove to be a promising alternative for individuals who otherwise would have no option but to travel long distances to receive services they need. Researchers have shown that telerehabilitation can improve quality of life and lead to more efficient use of health care resources ([@b15-v1n1-art-10.5195-ijt.2009.6014]; [@b48-v1n1-art-10.5195-ijt.2009.6014]). [@b6-v1n1-art-10.5195-ijt.2009.6014] investigated the cost/benefit ratio of using a technology like telemedicine. While technological advances have produced significant improvements in healthcare, ironically, it is technology itself that has played a role in the rising costs of care. However, these authors believe that telerehabilitation technology can, in fact, improve access and enable patients to receive appropriate care in their own homes or at nearby healthcare facilities and result in improved quality of care. This care would be provided at local health facilities by their current health providers, but with the aid of remote telerehabilitation consultant specialists. Given the apparent success of telerehabilitation in this scenario, such an application could reduce the need for specialists to travel to remote locations, without compromising the quality of care that clients receive. This study further explains that the cost of care at local facilities is likely to be less than that at highly specialized care centers ([@b6-v1n1-art-10.5195-ijt.2009.6014]). Despite encouraging studies demonstrating the feasibility of telerehabilitation, their applications are restricted by limited reimbursement for services. Medicare generally has not been receptive towards increasing telerehabilitation reimbursement based on the fact that "there is very little published peer-reviewed scientific data available on when telemedicine use is medically appropriate" and on the effectiveness of telerehabilitation ([@b38-v1n1-art-10.5195-ijt.2009.6014]). In response, University of Pittsburgh researchers and clinicians have collaborated to investigate the effectiveness of telerehabilitation interventions and explore its potential as a clinical tool to address the gaps and improve quality of care. One of the research tasks of the Rehabilitation Engineering Research Center on Telerehabilitation (RERC-TR) has been the evaluation of remote wheelchair prescription. The need for wheeled mobility devices is increasing as our population is aging and surviving trauma and disease. The availability of practitioners with specific expertise in this area is limited, especially in rural areas. People are isolated from rehabilitation services due to geography or physical limitations whereby large distances require long travel times increasing costs and other burdens. The purpose of this project is to determine the effectiveness of using a secure telerehabilitation consultation model for procuring an appropriate wheeled mobility and seating device via the high speed Internet. Outcome studies in the field of telerehabilitation start in the laboratory or university setting, and extend out to the actual settings and to the patients that will ultimately benefit the most (i.e., in rural or underserved communities). Researchers have taken the custom secure Internet Protocol based videoconferencing system developed by the RERC-TR and installed it within five rural hospitals located at least 100 miles away from Pittsburgh. Before launching the system, feasibility studies were conducted with one of the remote sites to test the camera technology associated with the videoconferencing. It was determined that a basic USB web-camera with up to 8 megapixel, autofocus lens system, microphone with Rightsound technology, and up to 30 frames per second video met our needs. A second camera was installed to remotely control the ability to pan, tilt, and optical zoom to assist with the teleconsultation. A comparative approach against a referenced procedure or a predefined standard such as the one implemented by the Center for Assistive Technology at the University of Pittsburgh Medical Center (CAT-UPMC) was conducted ([@b75-v1n1-art-10.5195-ijt.2009.6014]). Many evaluation studies ignore this basic requirement and therefore often end up with inconclusive results. Current accuracy and outcome studies are being conducted and results will be finalized. A second study focuses upon job coaching and involves activity recognition technologies using video and accelerometers to monitor specific work behaviors and, ultimately, to deliver cues and instruction in response to problem behaviors. The system is trained using accelerometer data from a given task being performed. A model is then developed to automatically recognize components of the task (or errors) during real time task performance and provide task guidance feedback to a client as needed via headphone. Pilot usability studies revealed that results of video captured and analyzed by a machine learning technique used to automatically identify features in the person's movement patterns of a routine fast food work task (e.g. flipping, salting, placing on or taking burgers off the grill) indicated that none of the complex movements were mislabeled and a set of common errors were identified and labeled. The activity recognition was then tested using accelerometers (one on each hand, and one on each forearm) with results being essentially equivalent to video data. Thus, initial work has yielded a model that is able to correctly identify component tasks for multiple subjects, using video or accelerometers. A user interface that is able to utilize logged data to recognize and deliver task guidance cues remotely to the user in real time was developed. Additional investigation including broadening the subject base and developing an interface to share a summary of the data (e.g., success rate, types of errors) with a job coach or other support person is being undertaken ([@b59-v1n1-art-10.5195-ijt.2009.6014]). Conclusions =========== As telerehabilitation moves beyond basic POTS and ISDN-based videoconferencing systems, rehabilitation clinicians and engineers have the increasing ability to upload and download data farther and wider. Yet, it is important to recognize that while the technology may be new, the rehabilitation services and care management systems are still the same. Advanced technologies including broad bandwidth Internet connections, web-based videoconferencing, and multimedia databases have potential to further enhance telerehabilitation capabilities. Telerehabilitation has the potential to both increase the availability of specialty clinical services in geographically remote communities and reduce service delivery costs associated with travel and time. Telerehabilitation's full potential will be only realized by researching and developing telecommunication technology that delivers both effective and efficient care to patients in all areas. Ultimately, the widespread adoption of telerehabilitation will require more than just lower cost technology information and access. These innovative strategies will need to be embraced by rehabilitation practitioners in their respective fields and those who pay for their services. This white paper was written as a project of the Rehabilitation Engineering Research Center on Telerehabilitation at the University of Pittsburgh, supported by a grant from the National Institute on Disability and Rehabilitation Research (NIDRR) (H133E040012). Authors would like to acknowledge the valuable contributions of content experts who reviewed our original draft, including David Brennan, MBE, Senior Research Engineer at National Rehabilitation Hospital - Center for Applied Biomechanics and Rehabilitation Research; and Deborah Theodoros, PhD, SLP, Associate Professor and Co-Director of the Telerehabilitation Research Unit at the University of Queensland, Australia. We also acknowledge the important contributions made by the panelists who provided commentary during the State of the Science presentation of this manuscript, including Suzanne Paone, MBA, Associate Director UPMC eHealth Initiatives; University of Pittsburgh Medical Center; Paul Wehman, Ph.D., Director of the Rehabilitation Research and Training Center on Workplace Supports and Chairman of the Division of Rehabilitation Research, Virginia Commonwealth University; Steven M. Dahling, ATS, Assistive Technology Coordinator, Rusk Institute of Rehabilitation Medicine, Barbara Demuth, RN, MSN, Assistant Director for Telehealth, Center for Remote and Medically-Underserved Areas, St. Francis University; and Christine Woo, MS, Cleveland Spinal Cord Injury/Disorders Center within the Veterans Affairs Medical Center.

INTRODUCTION {#s1} ============ Lung cancer is the leading cause of cancer induced mortality worldwide \[[@R1], [@R2]\]. Non-small cell lung cancers (NSCLC), particularly, adenocarcinoma (AC) and squamous cell carcinoma (SCC), are the most frequently diagnosed histotypes \[[@R1], [@R3]\]. New targeted therapies have been developed and some are currently used to treat advanced AC, but are unsuitable for SCC \[[@R4], [@R5]\]. Moreover, not all AC patients benefit from these treatments \[[@R5]\]. Development of effective and well tolerable immunotherapies to replace or be combined with surgery, radiotherapy or personalized treatments may be of great value. In the last few years, Interleukin(IL)-27, a member of the IL-12 family of cytokines, with important roles in both innate and adaptive immunity \[[@R6], [@R7]\], has revealed potent antitumor effects in the form of anti-proliferation, anti-angiogenesis, and immune system stimulation in a variety of tumors \[[@R8]--[@R10]\]. Its over-expression in murine Lewis lung carcinoma line 1 (LLC1) cells induces a specific cytotoxic T cell and antibody response *in vivo*, and also activates non-immunological mechanisms reducing cancer cell motility and migration \[[@R11]\]. Inhibition of AC cell migration, together with down-regulation of pro-angiogenesis genes by IL-27 has also been reported in the human A549 AC cell line \[[@R12]\]. Moreover, murine IL-27 gene-transfected LLC1 cells have been used to generate an autologous cell vaccine boosting an efficient T lymphocyte activation and IFNγ production \[[@R13]\]. However, definitive *in vivo* proof of IL-27\'s efficacy in pre-clinical models of human lung AC and SCC is still lacking. We here investigate the *in vitro* and *in vivo* effects of IL-27 on the regulation of angiogenesis-, stemness- and epitelial-mesenchymal transition (EMT)-related genes in human AC and SCC cell lines and lung tumors grown in B/T cell deficient mice. Furthermore, by means of molecular biology and immunohistochemical studies, we have assessed IL-27Receptor(R) expression in lung cancer samples and analyzed the rationale for a future IL-27 application in the clinical setting of NSCLC. RESULTS {#s2} ======= Human lung AC and SCC cell lines express IL-27R and respond to IL-27 up-regulating CXCL3 expression and down-modulating stemness- and EMT-related genes {#s2_1} ------------------------------------------------------------------------------------------------------------------------------------------------------- Since AC and SCC are the most common histotypes of lung cancers, \~85% of NSCLC \[[@R1], [@R3]\], we looked to see whether IL-27 acts as an antitumor agent in these forms. Expression of both chains of the IL-27R, gp130 and WSX-1 (TCCR, IL-27Rα), was investigated, by flow cytometry, in a series of cell lines derived from human lung AC, namely A549, GLC82, Calu-6 or from SCC, namely Calu-1 and SK-MES. As shown in Figure [1A and 1B](#F1){ref-type="fig"}, Calu-6 and SK-MES lines expressed the highest levels of both chains (gp130: 68% and 65%; WSX-1: 97% and 70% respectively) and were therefore chosen as representative of AC and SCC histotypes for the subsequent experiments. {#F1} We began by determining whether IL-27 affects the *in vitro* proliferation or apoptosis of these lines by culturing them with or without human (h) recombinant (r) hrIL-27 for 120 hours, and harvesting an aliquot every 24 hours to be analyzed for CFSE intracellular staining and for apoptosis. In both lines hrIL-27 was unable to directly modulate proliferation or apoptosis (not shown). We next investigated whether IL-27 regulated, in both lines, sets of genes shaping tumor malignancy and specifically related to angiogenesis, stemness and invasiveness. IL-27\'s ability to modulate angiogenesis-related genes in different cancer cell types, leading to anti-angiogenic effects *in vivo*, has been documented by us and others \[[@R8]--[@R10]\]. Unexpectedly, in both Calu-6 and SK-MES lines, hrIL-27 treatment considerably (*P* \< 0.05) up-regulated, 18.53 and, 13.7 times respectively, the mRNA expression of *CXCL3* (Figure [1C](#F1){ref-type="fig"}), also known as Growth-Related Oncogene *(GRO)3*, GRO protein gamma *(GROγ)* and Macrophage Inflammatory Protein 2 beta *(MIP2β)*, an angiogenic ELR^+^ CXC chemokine \[[@R14]\], identified as a powerful neutrophil chemoattractant \[[@R15], [@R16]\], and also driving migration and adhesion of monocytes and macrophages \[[@R17]\]. Other angiogenesis-related genes were differently regulated by IL-27 in the two lines (Figure [1C](#F1){ref-type="fig"}). In SK-MES cells only, hrIL-27 up-regulated *IFNγ* (6.13 times) and down-regulated *LAMA5*, encoding for Laminin-α5 (4.5 times), and *THBS1*, encoding for Thrombospondin-1 (8.64 times), whereas in Calu-6 cells, hrIL-27 selectively up regulated Cadherin 5 (*CDH5*), i.e.: *Vascular Endothelial (VE)-Cadherin* (4.5 times), and the Tissue Inhibitor of Metalloproteinase-1 (*TYMP-1*) (5.3 times) encoding for Platelet Derived Endothelial Cell Growth Factor 1 (*ECGF1*). Within a range of stemness-related genes (including *NANOG*, *BMI1*, *CD44v6*, *c-MYC*, see [Supplementary information](#SD1){ref-type="supplementary-material"}), in SK-MES cells, hrIL-27 significantly (*P* \< 0.05) down-regulated mRNA expression levels of Octamer-binding Transcription Factor 4A, *OCT4A* (5.7 times), SRY (sex determining region Y)-box 2, *SOX2* (4.0 times), *SOX9* (7.1 times), Notch homolog 1, *NOTCH1* (7.0 times), and Krüppel-like factor 4, *KLF4* (6.1 times) (Figure [1D](#F1){ref-type="fig"}), whereas in Calu-6 cells it only down-modulated mRNA expression of *Sonic Hedgehog*, *SHH* (4.2 times) (Figure [1E](#F1){ref-type="fig"}). Moreover, in SK-MES cells, hrIL-27 down-modulated mRNA expression levels of *Nestin* (6.7 times), associated with cell stemness and EMT \[[@R18], [@R19]\], and within the *ZEB*, *SNAIL* and *TWIST* families of EMT-activating transcription factors \[[@R20], [@R21]\], hrIL-27 down-modulated mRNA expression of Snail family zinc finger 1, *SNAI1/SNAIL* (5.0 times), Snail family zinc finger 2, *SNAI2/SLUG* (4.3 times), and Zinc finger E-box binding homeobox 1, *ZEB1* (5.0 times) (Figure [1F](#F1){ref-type="fig"}), whereas the expression of *c-MET*, also involved in EMT \[[@R22], [@R23]\], and that of *ZEB2*, *TWIST1*, and *TWIST2* remained unaffected. IL-27 hinders tumour growth in pre-clinical xenograft models of lung cancer in association with a remarkable colliquative necrosis and apoptotic events {#s2_2} ------------------------------------------------------------------------------------------------------------------------------------------------------- *In vivo* studies using pre-clinical models of severe combined immunodeficient SCID/NOD and T-cell deficient athymic-nude mice, s.c. injected with Calu-6 and SK-MES cell lines respectively, showed that hrIL-27 considerably reduced tumor growth in both models. In particular, the mean tumor volume (mtv) ± standard error (SE) of Calu-6 tumors grown in hrIL-27-treated mice was 54.22 ± 24.2 mm^3^ *versus* 241 ± 69 mm^3^ of tumors from controls (*P* = 0.0336) (Figure [2A](#F2){ref-type="fig"}), whereas the mtv ± SE of SK-MES tumors was 59.08 ± 13.1 mm^3^ *versus* 123.3 ± 24.5 mm^3^ of tumors from controls (*P* = 0.0469) (Figure [2B](#F2){ref-type="fig"}). {#F2} To get an insight into the mechanisms underlying the *in vivo* antitumor effects of IL-27, tumor growth/suppression areas were histopathologically analyzed. The histologic features of Calu-6 and SK-MES tumors from control mice, recalled human poorly-differentiated AC and SSC, respectively. Tumors harvested from hrIL-27-treated mice displayed wide areas of colliquative necrosis characterized by a prominent reactive cell infiltrate, (Figure [2C and 2D](#F2){ref-type="fig"}), in addition, SK-MES tumors also presented evident alterations in cancer cell morphology ranging from a spindle to a polygonal-round phenotype (Figure [2D](#F2){ref-type="fig"}). Inflammatory infiltrates were wider in both Calu-6 and SK-MES tumors from hrIL-27-treated mice than in control tumors, because of the significant (*P* \< 0.05) increase in their granulocyte and macrophage content, along with cancer cell expression of CXCL3 (Figure [2C and 2D](#F2){ref-type="fig"}) (Table [3](#T3){ref-type="table"}). In addition to the areas of colliquative necrosis, frequent apoptotic features were evidenced, by the TUNEL assay, close to granulocytes identified by their segmented nuclei, (Figure [2C and 2D](#F2){ref-type="fig"}), in both Calu-6 and SK-MES tumors from hrIL-27-treated mice (Figure [2C and 2D](#F2){ref-type="fig"}) (Table [3](#T3){ref-type="table"}). To understand the molecular mechanism underlying apoptotic events *in vivo*, we next assessed in tumors from hrIL-27-treated and control mice the expression of apoptosis-inducing proteins Tumor Necrosis Factor (TNF)α and TNF-Related Apoptosis Inducing Ligand (TRAIL). Immunostainings revealed that TRAIL was almost undetectable, while TNFα expression was distinct to strong, in the foci of reactive infiltrates, in both tumor types harvested from hrIL-27-treated mice (Figure [2C and 2D](#F2){ref-type="fig"}) (Table [3](#T3){ref-type="table"}). Lastly immunohistochemical analyses revealed that the microvascular network of Calu-6 tumors from hrIL-27-treated mice was similar to that of the controls whereas that of SK-MES tumors from hrIL-27-treated mice was clearly impaired (Figure [2C and 2D](#F2){ref-type="fig"}) (Table [3](#T3){ref-type="table"}) in association with a weakened laminin network and a faint, but distinct cancer cell expression of IFNγ (Figure [3A](#F3){ref-type="fig"}). {#F3} IL-27 down-modulates stemness- and EMT-related genes, particularly in SCC tumors {#s2_3} -------------------------------------------------------------------------------- To assess whether IL-27 regulation of pluripotency- and EMT-related genes also occurs *in vivo* at protein level, we carried out immunohistochemical analyses of tumors from hrIL-27- and PBS-treated mice. In Calu-6 tumors from hrIL-27-treated animals, the percentage of cancer cells displaying a distinct to strong SHH staining was decreased compared with control tumors (Figure [3B](#F3){ref-type="fig"}), whereas in SK-MES tumors from treated animals, the percentage of cells endowed with a distinct to strong nuclear staining for OCT4A, NOTCH1, and KLF4 was clearly decreased (Table [3](#T3){ref-type="table"}) as was reduced the percentage of tumor cells displaying both nuclear and cytoplasmic SOX2 positivity and nuclear SOX9 positivity (Figure [3C](#F3){ref-type="fig"}) (Table [3](#T3){ref-type="table"}). Expression of Nestin was weakened in cancer cells forming SK-MES tumors from IL-27-treated mice (Figure [3D](#F3){ref-type="fig"}). Nuclear and cytoplasmic SNAI1 stainings were dramatically and moderately reduced respectively, and the percentage of cancer cells endowed with a distinct to strong nuclear SNAI1, SNAI2, and ZEB1 staining also significantly (*P* \< 0.05) decreased following hrIL-27 treatment (Figure [3D](#F3){ref-type="fig"}), whereas E-Cadherin expression was reinforced in tumors from hrIL27-treated mice (Figure [3D](#F3){ref-type="fig"}) (Table [3](#T3){ref-type="table"}). Myeloablation by treosulfan thwarts the *in vivo* anti-lung cancer effects of IL-27 {#s2_4} ----------------------------------------------------------------------------------- To assess whether granulocytes and macrophages may, as suggested by the morphological data, account for the anti-lung cancer effects of IL-27 *in vivo*, we next repeated tumor growth experiments in mice pre-treated with myeloablative doses of treosulfan to obtain a severe or complete depletion of bone marrow cells \[[@R18]\]. These experiments revealed that after this treatment both Calu-6 and SK-MES tumors grew similarly in hrIL-27-treated and control mice. In particular, the mtv ± SE of Calu-6 tumors harvested from myeloablated and IL-27-treated mice was 153.4 ± 29.9 mm^3^ *versus* 169 ± 33.99 mm^3^ of tumors from myeloablated controls (Figure [4A](#F4){ref-type="fig"}). The mtv ± SE of SK-MES tumors from myeloablated and IL-27-treated mice was 34.6 ± 4 mm^3^ *versus* 33.6 ± 5.65 mm^3^ of tumors from myeloablated controls (Figure [4B](#F4){ref-type="fig"}). Of note, both Calu-6 and, particularly, SK-MES tumors grown in myeloablated control mice were smaller than tumors developed in not-myeloablated control mice (Calu-6: *P* = 0.0317 and SK-MES: *P* = 0.0159). {#F4} Histological features of both Calu-6 and SK-MES tumors developed in myeloablated hrIL-27-treated mice were similar to those from myeloablated controls. However, both histotypes, developed in myeloablated mice, independently from hrIL-27-treatment, presented small ischaemic necrotic foci (Figure [4C](#F4){ref-type="fig"}) in association with an evident (vessel counts: 3 ± 2 in Calu-6, and 5 ± 2 in SK-MES tumors from myeloablated controls, *versus* 9 ± 3 in Calu-6, and 12 ± 4 in SK-MES tumors from non-myeloablated controls; *P* \< 0.05) decrease of the whole vascular supply (Figure [4C](#F4){ref-type="fig"}). These tumours were almost devoid of granulocyte and macrophage content (Figure [4C](#F4){ref-type="fig"}). CXCL3 was still firmly expressed by tumor cells in both Calu-6 and SK-MES tumors from myeloablated and hrIL-27-treated mice (Figure [4C](#F4){ref-type="fig"}), while a faint IFNγ production was only detected in the latter. Human lung AC and SCC, and their precursor lesions express IL-27R in neoplastic and dysplastic cells, microvessels and tumor-associated reactive cells {#s2_5} ------------------------------------------------------------------------------------------------------------------------------------------------------ To determine whether lung cancer patients could benefit from IL-27\'s antitumor effects, we next immunohistochemically evaluated the expression and distribution of IL-27R in AC and SCC tissue sections. Since expression of gp130 has been documented in lung cancer \[[@R24], [@R25]\], we only assessed IL-27Rα expression in both cancerous and normal lung samples (from both cancer and control patients). In the normal tissue, it was basically found in mononuclear/macrophage-like cells fluctuating within alveolar walls (Figure [5D](#F5){ref-type="fig"}). Normal bronchial epithelia firmly expressed *IL-27Rα* mRNA, whereas in neoplastic samples it was expressed by the majority of AC, 90%, and SCC, 84% (Table [2](#T2){ref-type="table"}) (Figure [5A and 5D](#F5){ref-type="fig"}). Furthermore, within AC, metastatic tumors revealed significantly (*P* \< 0.05) higher expression levels of *IL-27Rα* mRNA than normal bronchial epithelia (Figure [5B](#F5){ref-type="fig"}). Notably, *IL-27Rα* expression in microdissected bronchial epithelium from normal samples of patients with lung cancer was analogous to that in control patients. These molecular data showed a good correlation (ρ = 0.82) with the immunohistochemical findings. {#F5} IL-27Rα expression by the primary tumor was significantly associated with lymph node status (*P* = 0,001), and advanced stages of disease (*P* = 0,02) as assessed by Fisher\'s exact test, whereas no significant association was observed with patient age or smoking history. Immunohistochemistry also revealed IL-27Rα expression in AC and SCC precursor lesions, namely atypical adenomatous hyperplasia (AAH) and severe dysplasia, squamous metaplasia (SM), squamous cell carcinoma *in situ* (SCIS) (Figure [5D](#F5){ref-type="fig"}), respectively (Table [1](#T1){ref-type="table"}). IL-27Rα expression may also be found in microvessels and infiltrating immune cells (Figure [5C](#F5){ref-type="fig"}), mostly identifiable as CD68^+^ monocytes/macrophages (Figure [5D](#F5){ref-type="fig"}) and CD11c^+^ myeloid dendritic cells. They were found in the stroma of both AC and SCC, scattered or within the lymph node--like structures, known as tertiary lymphoid structures (TLS) (Figure [5D](#F5){ref-type="fig"}) \[[@R26], [@R27]\]. CD15^+^ granulocytes that may be found in necrotic foci of rapidly growing tumors also expressed IL-27Rα (Figure [5D](#F5){ref-type="fig"}). ###### Clinic-pathological characteristics of patients with premalignant lung lesions and IL-27Rα expression profiles of these lesions Variables IL-27Rα immunostaining[\*](#tfn_001){ref-type="table-fn"} ------------------------------------------- ----------------------------------------------------------- ---- ----- ----- ***Age:*** (range 50--78) ***Gender*** Male 14 10 3 --- Female 6 5 1 --- ***Histological type*** AAH 7 7 --- --- SM and Dysplasia (mild, moderate, severe) 7 4 2 --- SCIS 6 4 2 --- IL-27Rα immunostaining was scored as negative, positive or weakly positive as described in Methods. DISCUSSION {#s3} ========== AC and SCC constitute the commonly diagnosed lung cancer histotypes \[[@R1], [@R2]\], but their management still results in low overall cure rates, suggesting the need for novel therapeutic approaches. The idea of strengthening patient\'s immune system to fight cancer is of growing interest for oncologists \[[@R28]\]. Our data indicate that IL-27, a well-tolerated and toxicity-free cytokine \[[@R9]\] may provide a new therapeutic option in NSCLC. Previous studies assessing IL-27\'s effects in lung cancer have used mouse autograft models \[[@R11], [@R13]\] or, *in vitro* experiments with both murine and human lung AC cell lines \[[@R11], [@R12]\]. Our data confirm, in pre-clinical xenograft models of human lung AC and SCC, the anti-lung cancer effects of IL-27. They have also revealed unforeseen implications for its immunological antitumor capability in the form of I. boosting the potent granulocyte/macrophage chemo-attractant CXCL3, in both AC and SCC cells, leading to intratumoral myeloid cell recruitment and activation, and II. re-education of these immune cells from the status of "cancer feeder", and host-detrimental \[[@R29], [@R30]\] towards that of "cancer killer", thus host-beneficial \[[@R30]--[@R32]\]. Tumor destruction by the prominent neutrophil and macrophage influx mediated by *CXCL3* \[[@R15]--[@R17]\] overcomes, at least in our setting, its well-known pro-angiogenic effects \[[@R14]\], since microvessel density remained unaltered in AC and slightly decreased in SCC tumors from IL-27-treated animals, when compared with controls. Anti-angiogenesis has a marginal or nil role in IL-27\'s anti-lung cancer efficacy, and apart from a slight *IFNγ* induction and *LAMININ-α5* down-regulation observed in SCC cells, IL-27 even down-regulates the angiogenesis inhibitor *THROMBOSPONDIN-1* \[[@R33]\] in SK-MES cells, and up-regulates *VE-CADHERIN* \[[@R34]\] and the pro-angiogenesis gene *TYMP-1*, encoding for *ECGF1* \[[@R35]\], in Calu-6 cells, but without significant *in vivo* consequences for tumor vascularity. Granulocytes and macrophages are endowed with IL-27R, and may respond to IL-27 by increasing their oxidative burst and cytokine production \[[@R36]--[@R38]\], suggesting that the range of action for this cytokine is not restricted to T cells, the key mediators of its antitumor effect in an immune-intact host \[[@R9]\]. Our results provide the first evidence that intratumor recruited and activated myeloid cells may, in a B/T cell-deficient host, take the place of T lymphocytes in mediating IL-27\'s antitumor activity, leading to a dramatic colliquative necrosis, and TNFα-associated apoptotic events in both AC and SCC. While extending to human SCC *in vivo* the finding of IL-27\'s down-modulation of EMT transcription factors such as SNAI1, SNAI2 and ZEB1 \[[@R20], [@R21]\], our discoveries identify a novel role for IL-27 as a negative regulator of pluripotency genes such as SHH in human AC cells, and SOX2, OCT4A, NOTCH1, KLF4, SOX9 and Nestin \[[@R18], [@R19]\] in human SCC cells, both *in vitro* and *in vivo*. EMT, a key event during the early phases of invasion and metastatization, selects for stem cell property \[[@R39]\], which, in turn, may condition the self-renewal capability of a cancer and correlate with its aggressiveness \[[@R40], [@R41]\]. Down-regulation of critical pluripotency genes by IL-27 increases the propensity of cells to differentiate towards a less aggressive phenotype \[[@R42]--[@R44]\] as shown by their transition from a fibroblast-like to a polygonal-round E-Cadherin-positive phenotype. Tumor responsiveness to IL-27, however, is the prerequisite for its entry into clinical trials. Most AC and SCC express IL-27R, and those with N1 involvement are often endowed with higher expression levels. Dibra et al. observed that WSX-1 expression in tumors induces immune tolerance exerting pro-tumorigenic functions, independently from IL-27 \[[@R45]\]. Our data, drawn from patient samples, suggest that IL-27 may be used to overcome this drawback and be of particular benefit in advanced lung cancer stages. The myeloid cell mediated anti-lung cancer effects of IL-27 may be exerted in humans, since in our samples IL-27R is not only expressed by cancer cells, and microvessels, but also by CD15^+^ granulocytes, CD68^+^ monocytes/macrophages and CD11c^+^ myeloid dendritic cells scattered in the stroma or arranged in TLS. TLS have been associated with a favorable clinical outcome in NSCLC \[[@R26], [@R27]\] and IL-27 may promote, in these lymphoid structures, an efficient adaptive antitumor immunity. Altogether, our results highlight novel aspects of IL-27\'s antitumor potential, specifically in NSCLC, such as the ability I. to drive myeloid cells towards antitumor activities, and II. down-regulate stemness genes, particularly in SCC cells, thus suppressing their self-renewal potential. IL-27 may thus be proposed for clinical trials with the prospect of its clinical use in immune-defective or advanced NSCLC patients. MATERIALS AND METHODS {#s4} ===================== Ethics statement {#s4_1} ---------------- Animal experiments were performed at the IRCCS "San Martino" National Institute for Cancer Research, Genoa, in keeping with the National and International regulations (Italian Legislative Order 27/01/1992, n.116, European Economic Community Council Directive 86/609, OJL 358, Dec. 1, 1987). For studies on human tissues, written informed consent was obtained from patients. The study was approved by the Ethical Committee of the "G. d\'Annunzio" University of Chieti (Italy), and Local Health Authority No.2, in the report n.14 of 19/07/2012, and performed in accordance with the principles outlined in the Declaration of Helsinki. Patients and samples {#s4_2} -------------------- Premalignant and malignant lung samples were obtained from 78 untreated patients with operable NSCLC staged IA--IIIA according to the pTNM system \[[@R46]\]. NSCLC consisted of 33 AC (12/33 with N1 involvement) and 25 SCC (10/25 with N1 involvement). Our panel of 20 premalignant and 58 malignant lesions \[[@R47], [@R48]\], and patient\'s clinic-pathological characteristics \[[@R48], [@R49]\] are shown in Tables [1](#T1){ref-type="table"} and [2](#T2){ref-type="table"}. Normal lung tissue was obtained from both lung cancer patients and control patients operated for other reasons. Sample collection and processing are described in the [Supplementary information](#SD1){ref-type="supplementary-material"}. ###### Clinic-pathological characteristics of patients with lung cancer and IL-27Rα expression profiles of these cancers Variables IL-27Rα immunostaining of the primary tumor[\*](#tfn_002){ref-type="table-fn"} -------------------------- -------------------------------------------------------------------------------- ------------ ------------ ----------- ***Age*** (range 38--79) ***Gender*** Male 45 13 (29%) 26 (58%) 6 (13%) Female 13 7 (54%) 5 (38%) 1 (8%) ***Histological type*** *NSCLC* *58* *20 (34%)* *31 (54%)* *7 (12%)* AC 33 15 (45%) 15 (45%) 3 (10%) SCC 25 5 (20%) 16 (64%) 4 (16%) ***Tumor size*** T1 29 8 (28%) 16 (55%) 5 (17%) T2 27 12 (45%) 13 (48%) 2 (7%) T3 2 2 (100%) --- --- ***Lymph node status*** N0 34 6 (18%) 21 (62%) 7 (20%) N1 24 14 (58%) 10 (42%) --- N2 --- --- --- --- ***Stage*** IA 16 2 (13%) 9 (56%) 5 (31%) IB 18 4 (22%) 12 (67%) 2 (11%) IIA 13 8 (62%) 5 (38%) --- IIB 9 4 (44%) 5 (56%) --- IIIA 2 2 (100%) --- --- ***Smoking history*** Smokers 42 16 (38%) 20 (48%) 6 (14%) Non smokers 16 4 (25%) 11 (69%) 1 (6%) IL-27Rα immunostaining was scored as negative, positive or weakly positive as described in Methods. ###### Immunohistochemical analyses of tumors developed after subcutaneous injection of Calu-6 or SK-MES cells in SCID/NOD and athymic NU/NU Mice, respectively, and treated with PBS or hrIL-27 Calu-6 SK-MES -------------------------------------------------------- -------- ---------------------------------------- ------ ------ ------ ------------------------------------------- ------ ---- ------- ------ ------ ------------------------------------------- **Immune Cells** Granulocytes 7.0 ± 3.3 25.0 ± 7.2[^†^](#tfn_004){ref-type="table-fn"} 14.5 ± 5.6 31.4 ± 6.0[^†^](#tfn_004){ref-type="table-fn"} Macrophages 8.5 ± 4.1 21.4 ± 6.3[^†^](#tfn_004){ref-type="table-fn"} 12.0 ± 4.2 27.0 ± 6.5[^†^](#tfn_004){ref-type="table-fn"} **Blood Vessels** 9.8 ± 4.0 7.7 ± 3.0 12.8 ± 4.9 5.0 ± 2.2[^†^](#tfn_004){ref-type="table-fn"} **Apoptotic Index** 2.5 ± 2.0% 8.9 ± 3.2%[^†^](#tfn_004){ref-type="table-fn"} 3.4 ± 2.2% 10.5 ± 3.5%[^†^](#tfn_004){ref-type="table-fn"} **Human Cytokines**[\*](#tfn_003){ref-type="table-fn"} CXCL3 ± \+ + ± \+ + IFNγ ─ ─ ─ \+ TNFα ─ \+ ± \+ + Nestin[\*](#tfn_003){ref-type="table-fn"} ND ND ++ ± E-Cadherin[\*](#tfn_003){ref-type="table-fn"} ND ND \+ \+ + **Stemness genes** SHH 88.0 ± 8.9% 58.5 ± 10.3%[^†^](#tfn_004){ref-type="table-fn"} OCT4A ND[^‡^](#tfn_005){ref-type="table-fn"} ND 85.0 ± 11.5% 59.4 ± 12.2%[^†^](#tfn_004){ref-type="table-fn"} SOX2 ND ND 79.2 ± 9.0% 40.5 ± 15.6%[^†^](#tfn_004){ref-type="table-fn"} SOX9 ND ND 84.5 ± 11.3% 62.5 ± 9.0%[^†^](#tfn_004){ref-type="table-fn"} NOTCH1 ND ND 86.2 ± 8.0% 63.1 ± 10.5%[^†^](#tfn_004){ref-type="table-fn"} KLF4 ND ND 78.3 ± 10.0% 39.0 ± 14.2%[^†^](#tfn_004){ref-type="table-fn"} **EMT-related genes** SNAI1 ND ND 85.3 ± 9.0% 66.4 ± 8.0%[^†^](#tfn_004){ref-type="table-fn"} SNAI2 ND ND 80.5 ± 8.2% 58.5 ± 10.3%[^†^](#tfn_004){ref-type="table-fn"} ZEB1 ND ND 71.8 ± 10.0% 49.7 ± 9.2%[^†^](#tfn_004){ref-type="table-fn"} Assessment of cytokine, Nestin and E-cadherin expression, and counts of microvessels, immune cells, TUNEL positive cells, stemness and EMT-related gene expressing cells were performed at X400 in a 0.180 mm^2^ field. At least 3 samples (three sections/sample), and 6--8 (depending on the tumor width) randomly chosen fields/section were evaluated. Results are expressed as mean ± SD of CD31 positive microvessels per field; or RB6-8C5 (granulocytes) or CD11b/CD18 (macrophages) positive cells per field; or as mean ± SD percentage of TUNEL positive cells, or stemness- or EMT-related gene expressing cells evaluated on paraffin embedded sections by immunohistochemistry. The expression of cytokines, Nestin and E-Cadherin was defined as absent (─); scarce (±); distinct (+) or strong (++) on paraffin embedded (CXCL3, IFNγ, Nestin and E-Cadherin) or frozen (TNFα) sections stained with the corresponding Ab. Values significantly different (*P* \< 0.05) from corresponding values in tumors developed in PBS-treated mice. ND, not detected. Cell culture and flow cytometry {#s4_3} ------------------------------- Human lung SCC cell lines, Calu-1 and SK-MES, and AC cell lines, Calu-6, GLC82 and A549 (ATCC cell bank) were cultured in RPMI 1640 with 10% FCS (Seromed Biochrom KG, Berlin, Germany). HrIL-27 (R&D System, Abingdon, UK) was used at 100 ng/ml following titration experiments. Details on the assessment of IL-27R expression, tumor cell proliferation and apoptosis by flow cytometry are provided in the [Supplementary information](#SD1){ref-type="supplementary-material"}. Assessment of angiogenesis, stemness- and EMT-related gene expression by real-time RT-PCR {#s4_4} ----------------------------------------------------------------------------------------- RNA was extracted from Calu-6 and SK-MES cells cultured for 24 hours in the presence or absence of hrIL-27, using TRIZOL reagent (Invitrogen, Paisley, UK). Contaminant genomic DNA was removed by DNase treatment (Qiagen GmbH, Hilden, D). RNA was reverse-transcribed with the RT^2^ First Strand cDNA Synthesis kit (SABioscience, Frederick, MD, USA). Experiments are described in the [Supplementary information](#SD1){ref-type="supplementary-material"}. Laser capture microdissection (LCM) and real-time RT-PCR {#s4_5} -------------------------------------------------------- We used the P.A.L.M. Micro Beam System (Bernried, Germany) for LCM of frozen sections from normal human lung, AC, and SCC samples. We selected the bronchial epithelium from normal lung sections and tumor cells from AC or SCC sections. Details are provided in the [Supplementary information](#SD1){ref-type="supplementary-material"}. Mouse studies {#s4_6} ------------- Four- to six-week-old athymic-nude and SCID/NOD mice (Harlan Laboratories, Udine, Italy) were housed under specific pathogen-free conditions. Details on Tumor growth experiments in mice pre-treated or not with the myeloablative agent treosulfan (Medac) \[[@R50]\], and schedule of hrIL-27 administration are provided in the [Supplementary information](#SD1){ref-type="supplementary-material"}. Histopathological and morphometric analyses {#s4_7} ------------------------------------------- Methods for histology, immunohistochemistry, TUNEL assay, and the list of Abs used are reported in the [Supplementary information](#SD1){ref-type="supplementary-material"}. IL-27Rα expression by pre-neoplastic and neoplastic lesions was defined as positive, weakly positive, or negative. Details on the evaluation of IL-27Rα expression, in human samples, and of Nestin, E-Cadherin, and cytokine expression; counts of microvessels, and immune cells, percentage of apoptotic cells, and cancer cells expressing stemness or EMT-related genes, in tumor xenografts, are reported in the [Supplementary information](#SD1){ref-type="supplementary-material"}. Statistical analysis {#s4_8} -------------------- Tumor volumes were reported in mm^3^ *versus* time. Differences in tumor volume, microvessel density and counts of immune cells or percentage of apoptotic cells or stemness and EMT-gene expressing cells between tumors from hrIL-27- and PBS-treated mice were assessed by Student\'s *t* test, and data were reported as mean ± standard deviation (SD). Between-group differences in the relative expression of IL-27Rα mRNA, by real-time RT-PCR, were assessed by one-way analysis of variance (ANOVA) and the difference between each pair of means was evaluated with the Tukey\'s HSD test. Fisher\'s exact test was used to examine the association between IL-27Rα protein expression, evaluated by immunohistochemistry, in primary lung tumors, and clinic-pathological characteristics. The Spearman rank correlation coefficient (ρ) was used to examine the correlation between immunohistochemical staining and real-time RT-PCR for IL-27Rα expression. The SPSS software, version 11.0 (IBM, Armonk, NY, USA) was employed, with *P* \< 0.05 as the significance cut-off. SUPPLEMENTARY INFORMATION {#s5} ========================= This work was supported by grants from the Associazione Italiana Ricerca sul Cancro (AIRC, Investigator Grant n. 13134) and the "Umberto Veronesi" Foundation for the Progress of Sciences to E. Di Carlo; and grants from AIRC (Investigator Grant n. 13018), Ricerca Finalizzata Collaboratore Estero, Ministero della Salute (Grant n. RF-2010-2308270), and from Cinque per mille e Ricerca Corrente, Ministero della Salute to I. Airoldi. **Author contributions** EDC planned and designed experiments, collected data and wrote the manuscript; IA designed and performed experiments; MGT, SE, MVR, and GB, performed experiments and collected data; GC provided tissue samples and clinic-pathological data. **Conflicts of interest** The authors disclose no potential conflicts of interest. AAH : Atypical Adenomatous Hyperplasia AC : Adenocarcinoma BMI1 : Polycomb complex protein 1 CDH : Cadherin CXCL3 : Chemokine (CXC motif) ligand 3 ECGF1 : Platelet-Derived Endothelial Cell Growth Factor 1 EMT : Epithelial-Mesenchymal Transition GROγ : Growth Related Oncogene protein gamma IFNγ : Interferon gamma IL : Interleukin KLF4 : Krüppel-Like Factor 4 LCM : Laser Capture Microdissection LLC1 : Lewis Lung Carcinoma line 1 MIP2β : Macrophage Inflammatory Protein 2 beta mtv : mean tumor volume NANOG : Nanog Homeobox NOTCH1 : Notch homolog 1 NSCLC : Non-Small Cell Lung Cancer OCT4A : Octamer-Binding Transcription factor 4 A R : Receptor SCC : Squamous Cell Carcinoma SCIS : squamous cell carcinoma *in situ* SE : standard error SHH : Sonic Hedgehog SM : Squamous Metaplasia SOX : SRY (sex determining region Y)-box TYMP-1 : Tissue Inhibitor of Metalloproteinase-1 TLS : Tertiary Lymphoid Structures TNF : Tumor Necrosis Factor TRAIL : TNF-Related Apoptosis Inducing Ligand TUNEL : Terminal deoxynucleotidyl transferase dUTP nick end labeling TWIST : twist family bHLH transcription factor VE : Vascular Endothelial ZEB : Zinc finger E-box binding homeobox.

Introduction {#Sec1} ============ The morphological variability existing between unpaired branches of the abdominal aorta has attracted the attention of anatomists, surgeons and radiologists due to its significance in a range of clinical procedures, including surgery of aneurysms or radiological transarterial chemoembolization procedures for tumours \[[@CR19], [@CR25], [@CR26], [@CR30]\]. A great deal of variation is seen in the origins of the inferior phrenic arteries (IPA), i.e. the right inferior phrenic artery (RIPA) and the left inferior phrenic artery (LIPA); however, they generally originate from the abdominal aorta or celiac trunk \[[@CR1], [@CR4], [@CR10], [@CR14]--[@CR16], [@CR21], [@CR29], [@CR32], [@CR33]\], and are known to supply the diaphragm, oesophagus, stomach, liver, adrenal glands and retroperitoneum. The RIPA and LIPA may arise separately from the lateral aspect of the abdominal aorta, immediately above the celiac trunk, or by a common trunk. Some other variants may derive from the celiac trunk, the suprarenal, hepatic, left gastric, renal or superior mesenteric artery \[[@CR1], [@CR4], [@CR10], [@CR11], [@CR14]--[@CR16], [@CR21], [@CR25], [@CR29], [@CR32], [@CR33]\]. The occurrence of accessory hepatic arteries (AHA), the left accessory hepatic artery (LAHA) and the right accessory hepatic artery (RAHA) have been described previously \[[@CR12], [@CR22]--[@CR25], [@CR30], [@CR35]\]. Adrenal arteries (AA) supply the adrenal gland, and consist of three arteries: the superior suprarenal artery (SSA), the middle suprarenal artery (MSA) and the inferior suprarenal artery (ISA). The SSA arises from the LIPA, the MSA arises from the aorta between the IPA and the renal artery (RA), while the ISA derives from the RA; however, they may show variations in their origins \[[@CR2], [@CR7]--[@CR9], [@CR20], [@CR21]\]. This case report describes a rare variant origin of a common trunk for the right inferior phrenic artery (which gives the right accessory hepatic artery) and the superior and inferior suprarenal artery from the right renal artery. Our findings highlight the importance of knowledge of the arterial supply in the abdominal cavity, and these findings are significant for radiologists, anatomists and surgeons specializing in the hepato-bilary and pancreatic areas. Case report {#Sec2} =========== The cadaver of a 64-year-old man was subjected to routine anatomical dissection for research and teaching purposes at the Department of Normal and Clinical Anatomy of the Medical University of Lodz. The dissection was performed in the abdominal cavity. A careful resection was performed of the interrupting tissues, where a common trunk of the right inferior phrenic artery, and superior and inferior suprarenal artery originated from the right renal artery, which originated from the anterior side of the abdominal aorta. The further course of the inferior right inferior phrenic artery gave rise to the right accessory hepatic artery (Figs. [1](#Fig1){ref-type="fig"}, [2](#Fig2){ref-type="fig"}). The right middle suprarenal artery was absent (Figs. [1](#Fig1){ref-type="fig"}, [2](#Fig2){ref-type="fig"}). Fig. 1Origin of a common trunk for the right inferior phrenic artery (which gives rise to the right accessory hepatic artery) and inferior and superior suprarenal artery. White arrowheads present the common trunk for the right inferior phrenic artery, and inferior and superior suprarenal artery. *AO* abdominal aorta, *CT* celiac trunk, *SMA* superior mesenteric artery, *RIPA* right inferior phrenic artery, *RAHA* right accessory hepatic artery, *SSA* superior suprarenal artery, *ISA* inferior suprarenal artery, *RA* renal artery Fig. 2Schema of a common trunk for the right inferior phrenic artery (which gives the right accessory hepatic artery) and inferior and superior suprarenal artery. White arrowheads present the common trunk for the right inferior phrenic artery, and inferior and superior suprarenal artery. *AO* abdominal aorta, *SMA* superior mesenteric artery, *RIPA* right inferior phrenic artery, *RAHA* right accessory hepatic artery, *AG* adrenal gland, *SSA* superior suprarenal artery, *ISA* inferior suprarenal artery, *RA* renal artery, *IVC* inferior vena cava, *K* kidney, *L* liver, *D* diaphragm, *CT* celiac trunk, *CHA* common hepatic artery, *SA* splenic artery The arteries were measured using digital photographic documentation processed through MultiScanBase 18.03 (Computer Scanning System II, Warsaw, Poland). The value and precision of this method have been confirmed in a previous study \[[@CR26], [@CR31]\]. The diameter of the renal artery at its origin from the abdominal aorta was 10.03 mm (the diameter of the coeliac trunk was 12.84 mm, while the diameter of the superior mesenteric artery was 10.17 mm). After 12.53 mm, the right renal artery gave rise to a common trunk of the right inferior phrenic artery, and the superior and inferior suprarenal arteries. The diameter of the common trunk was 3.95 mm, the first branch was the inferior suprarenal artery (diameter 1.84 mm); following this, the course was divided into the superior suprarenal artery (diameter 1.36 mm) and the common trunk of the right inferior phrenic artery and the right accessory hepatic artery. The diameter of the right inferior phrenic artery was 2.55 mm. Furthermore, it was observed that the right hepatic artery branched out: where it did so, the diameter of the right hepatic artery was 1.13 mm and that of the right inferior phrenic artery was 2.11 mm. Discussion {#Sec3} ========== Knowledge of the vascularization of hepatic vascular variations is significant for the daily practice for surgeons specializing in the hepato-bilary and pancreatic area, and also for general surgeons and radiologists, mainly those involved in interventional radiology. Significant improvements have been made in the surgical and/or radiological treatment of benign and malignant diseases of the liver, pancreas and bile ducts. In laparoscopic surgery, there is a need for accurate descriptions of liver vascularization to avoid iatrogenic vascular changes \[[@CR25], [@CR26], [@CR30]\]. An understanding of the extent of vascularization by the IPA is significant because, apart from the main supply of the diaphragm, it can form collateral circulation. Liver cancers commonly derive their arterial supply from the hepatic arteries, and the RIPA and the LIPA are pathways extrahepatic collateral arteries; they can also supply hepatic malignancies, because they neighbour hepatic segments as they transverse the bare area of the liver \[[@CR11], [@CR13]\]. Among the arterial pathways that provide liver cancer, both RIPA and LIPA represent nearly half of the collaterals, with RIPA being the most common \[[@CR11], [@CR13]\] and LIPA being the fourth or sixth most common \[[@CR11], [@CR13]\]. As a result, both RIPA and LIPA are used during transcatheter arterial chemoembolization of liver tumours, especially those located in the peripheral segments of the liver (1--4) \[[@CR11], [@CR13]\]. Variations in the origin of the phrenic arteries are numerous and supplementary phrenic vessels are common \[[@CR1], [@CR10], [@CR15], [@CR29], [@CR32], [@CR33]\]. The inferior phrenic arteries arise from a common trunk (55%), from the aorta or the celiac trunk in 18--30% of cases, or as independent branches from these same sources in 62% \[[@CR32]\]. Other sources may be the hepatic, left gastric, renal, suprarenal, or superior mesenteric arteries in about 8% of cases \[[@CR32]\]. When independent, the right and left phrenics usually arise asymmetrically. The study by Grieg et al. \[[@CR10]\] of 848 bodies, the origin of the inferior phrenic is as follows: right and left separately from the celiac trunk, 20.3%; as a common trunk from the aorta, 19.7%; right artery from the aorta, left from the celiac, 14.2%; common trunk from the celiac, 13.6%; separately from the aorta, 13.2%; right from the celiac, left from the aorta, 6.8%; right from the renal, left from the aorta, 3.7%; right from the renal, left from the celiac, 3.5%; right and left from the left gastric, 0.7%; right from left gastric, left from aorta. 0.5%; right from the aorta, left from renal, 0.5%; right from celiac, left from left gastric, 0.5%; right from aorta, left from left gastric, 0.4%; right and left from renal, 0.4%; and all other sources and combinations, 1.9% \[[@CR10]\]. Also, in some cases in which the right inferior gastric arose from a renal artery, it was from a superior polar renal artery \[[@CR10]\]. More recent studies indicate that the RIPA most commonly originates from the abdominal aorta, as noted by Kimura et al. \[[@CR14]\], Basilie et al. \[[@CR5]\] and Gürses et al. \[[@CR11]\]; however, Loukas et al. \[[@CR16]\] and Aslaner et al. \[[@CR4]\] report the most common origin of RIPA to be the celiac trunk. The second most frequent type of RIPA origin was found to be the celiac trunk by Kimura et al. \[[@CR14]\] and Basilie et al. \[[@CR5]\], the abdominal aorta by Loukas et al. \[[@CR16]\] and Aslaner et al. \[[@CR4]\], and the right renal artery by Gürses et al. \[[@CR11]\]. It is worth mentioning that Kimura et al. \[[@CR14]\] report the possibility of RIPA originating from the dorsal pancreatic artery, while Aslaner et al. \[[@CR4]\] note the possibility of it originating from the common hepatic artery. The difference between the types of origin of the right inferior phrenic artery are shown in Table [1](#Tab1){ref-type="table"}. Table 1Comparison types of origin of the right inferior phrenic arteryTypes of origin of the RIPALoukas et al. (%)Kimura et al. (%)Basilie et al. (%)Gürses et al. (%)Aslaner et al. (%)Present caseAbdominal aorta3857495025.2--Celiac trunk4030413.8530.7--Right renal artery17115.57.6910.4--Proper hepatic artery2--0.5------Left gastric artery3243.854.1--Dorsal pancreatic artery--1--------Common hepatic artery--------0.1--The common trunk of the RIPA, SSA and ISA from the RA----------100*RIPA* right inferior phrenic artery, *SSA* superior suprarenal artery, *ISA* inferior suprarenal artery, *RA* renal artery The suprarenal arteries demonstrate great morphological variety \[[@CR2], [@CR7]--[@CR9], [@CR20], [@CR21]\]. Dutta et al. \[[@CR8]\] note that the SSA did not have any variations in origin; this contrasts with our current report, where the SSA origin of the common trunk which originates from the right renal artery. Dutta et al. \[[@CR8]\] report the MSA to be absent from the right side in 29% of cases; however, our present findings do not indicate the presence of MSA. While Dutta et al. \[[@CR8]\] report variability in the origin of the ISA, i.e. 18% originated from the gonadal arteries and 6% from the right lateral margin of the abdominal aorta, our present study found the ISA to originate from the common trunk. In a study of 200 cadaveric dissections, Michels \[[@CR22], [@CR23]\] describes ten types of hepatic arterial variant. The full classification with its frequency of occurrence is as follows: Type I (normal pattern) − 81%; Type II (a replaced LHA from the left gastric artery) 3%; Type III (a replaced RHA from the superior mesenteric artery) 3.7%; Type IV (replaced RHA and LHA) 0.8%; Type V (an accessory LHA) 3.2%; Type VI (an accessory RHA) 1.6%; Type VII (accessory LHA and RHA) 0.2%; Type VIII (a replaced LHA or RHA with other hepatic artery being an accessory one) 0.35%; Type IX (the hepatic trunk as a branch of the superior mesenteric artery) 1.2% and Type X (common hepatic artery branched from the left gastric artery) 0.04%. This classification by Michels was later modified by Hiatt et al. \[[@CR12]\]. In a study of 1000 livers, Hiatt et al. \[[@CR12]\] note the presence of a normal hepatic artery in 757 specimens (75.7%), a LAHA originating from the left gastric artery in 97 specimens (9.7%) and an RAHA originating from the superior mesenteric artery in 106 specimens (10.6%). In 23 cases they also report the presence of a double-replaced pattern, where the right hepatic artery was a branch of the superior mesenteric artery and the left hepatic artery was a branch of the left gastric artery \[[@CR12]\]. They also note two variants of origin of the common hepatic artery: as a branch of the superior mesenteric artery in 15 cases (2.3%) and originating directly from the aorta in two cases (1.5%). López-Andújar et al. \[[@CR17]\] classified 12 types of hepatic arterial variations. The full classification is as follows: Type 1 (normal hepatic arterial) 70%; Type 2 (a replaced left hepatic artery arises from the left gastric artery) 9.7%; Type 3 (a replaced right hepatic artery arises from the superior mesenteric artery) 7.8%; Type 4 (presence both replaced right and left hepatic arteries, and the replaced right hepatic artery originated from the superior mesenteric artery, while the left originates from the left gastric artery) 3.1%; Type 5 (the LAHA arises from the left gastric artery) 3.9%; Type 6 (the RAHA originated from the superior mesenteric artery) 0.6%; Type 7 (the LAHA originated from the left gastric artery and RAHA arises from the superior mesenteric artery) 0.6%; Type 8 (replaced left hepatic artery arises from the left gastric artery and the RAHA originated from the superior mesenteric artery) 0.3%; Type 9 (the common hepatic artery arises from the superior mesenteric artery) 2.5%; Type 10 (the common hepatic artery originates from the left gastric artery) 0%; Type 11 (the common hepatic artery arises from the superior mesenteric artery and LAHA is a branch of the left gastric artery) 0.3%; Type 12 (the common hepatic artery arises directly from the aorta) 0.7%. Noussios et al. \[[@CR24]\] report the presence of a normal hepatic arterial anatomy in 81% of examined cases, and 19% were variations on the hepatic arteries: the LAHA was present in 1.6%, while the RAHA in 3.2%; replaced right hepatic artery originated from the superior mesenteric artery in 3.7%, while a replaced left hepatic artery originated from the left gastric artery in 3% of cases. Both replaced right and left hepatic artery were observed in 0.8% of cases \[[@CR24]\]. Our previous studies also describe the existence of accessory hepatic arteries, namely five RAHA arising from the celiac trunk, one RAHA branching off the proper hepatic artery, and one RAHA originating from the superior mesenteric artery; it also describes a variant where the LAHA arises from the left gastric artery \[[@CR25]\]. Some variants of the AHA were also described as a case report. Bastos-Neves et al. \[[@CR6]\] reported a rare anatomical variation of the hepatic arterial supply: a RAHA arising directly from the celiac trunk. Panagouli et al. \[[@CR27]\] describe a RAHA originating from the left gastric artery, while Polguj et al. \[[@CR30]\] describe a RAHA originating from the common hepatic artery near the celiac trunk, which ran behind the portal vein to the right lobe of the liver. Yamashita et al. \[[@CR35]\] describe a RAHA branching from the gastroduodenal artery. Lurie et al. \[[@CR18]\] and Paraskevas et al. \[[@CR28]\] describe a LAHA originating from the left gastric artery. The identification of variations and anomalies in the right hepatic artery is not only of value to the anatomist, but also to the surgeon. The RAHA may be injured during resection of the pancreatic head, because the artery is located in close proximity to the portal vein \[[@CR34]\]. The presence of replaced right hepatic artery can save live of patients with biliary tract cancer, because they are further away from the bile duct and tendo to save a cancer, allowing the tumour to be excised \[[@CR34]\]. The common trunk and the branches off this trunk described in this case report are clinically important: the right inferior phrenic artery may form collateral pathways for liver cancer, thus better facilitating surgical procedures. The presence of the AHA is an important consideration in surgery, because recent literature suggests livers with an AHA are preferred for donor liver transplantation. In addition, Aramaki et al. \[[@CR3]\] propose the use of right-sided grafts from donors with an LAHA and a left-sided graft from donors with an RAHA. Knowledge of such anatomical variation is clearly significant in these cases. Conclusion {#Sec4} ========== In conclusion, although the common trunk of the inferior phrenic artery, superior and inferior suprarenal artery and presence of the right accessory hepatic artery observed in this case are very rare, it might be a highly significant factor in the arterial supply to this region. Preoperative knowledge of such anatomic variants is essential in planning surgical procedure such as liver transplantation or removal of the liver lobe. The authors wish to express their gratitude to all those who donated their bodies to medical science. ŁO---project development, data collection and management, data analysis and manuscript writing. AW---data collection, data analysis, manuscript editing. MP---data analysis, manuscript editing. MT---data analysis, manuscript editing. All authors have read and approved the manuscript. There is no funding source. Conflict of interest {#FPar1} ==================== The authors declare that they have no conflict of interest. Ethical approval {#FPar2} ================ This article does not contain any studies with human participants or animals performed by any of the authors.

Glucose is the principal source of energy for the brain and its relationship to neuronal activity are poorly understood. The human brain uses 80% of its energy for ongoing neural activity that occurs in isolation from any particular stimulus. A promising tool for the investigation of glucose metabolism and its relationship to neuronal activity is simultaneous trimodal MR-PET-EEG data imaging. We here demonstrate the first in vivo human trimodal data at 3T. In one session MR, FDG-PET and EEG data were recorded simultaneously at a 3T hybrid MR-BrainPET scanner (Siemens, Germany) equipped with a 32 channel MR-compatible EEG system (Brain Products, Germany) in 11 healthy volunteers (11 males, mean age: 25.2 years SD: 1.2). MR and EEG data acquisition MP-RAGE (TR = 2250 ms, TE= 3.03 ms, 176 sagittal slices. 1 mm, GRAPPA factor 2. MR-based attenuation correction of PET data via UTE: flip angle=15. Two different echo times TE1=0.07 and TE2=2.46 ms, TR=200 ms. EPI sequence (TR: 2.2 s, TE: 30 ms, FOV: 200 mm, 165 volumes, The subjects were requested to close their eyes and relax EEG data were recorded using a 32-channel MR compatible EEG system. App. 200 MBq/μmol FDG were injected, data were acquired in list mode and iteratively reconstructed with all necessary corrections into 153 slices with 256 x 256 voxels sized 1.25 mm^3^. The trimodal approach, recording PET data, MR data and EEG data simultaneously was successful. The high neuronal activity of the structures within the default mode network occurs on the basis of a high glucose consumption rate within the default node network. The activity of the default mode is not tied to a special EEG frequency band.

Desai, R. T., Cowee, M. M., Wei, H., Fu, X., Gary, S. P., Volwerk, M., & Coates, A. J. (2017). Hybrid simulations of positively and negatively charged pickup ions and cyclotron wave generation at Europa. Journal of Geophysical Research: Space Physics, 122, 10,408--10,420. <https://doi.org/10.1002/2017JA024479> 1. Introduction {#jgra53834-sec-0001} =============== The Alfvén‐cyclotron instability is driven by a *T* ~⊥~/*T* ~∥~\> 0 anisotropy in the distribution function of a given ion species, where ⊥ and ∥ are defined with respect to the ambient magnetic field. Anisotropic ion populations can be created by the ionization of neutral atoms where the newly formed ions are accelerated or "picked up" by an electric field. The ions are thus energized and, for a sufficiently high plasma beta, generate a number of electromagnetic microinstabilities and plasma wave phenomena (Gary & Schriver, [1987](#jgra53834-bib-0026){ref-type="ref"}). Electromagnetic ion cyclotron waves (ICWs) associated with this ion pickup instability have been observed in the solar wind at comets, Mars, and Venus (Barabash et al., [1991](#jgra53834-bib-0003){ref-type="ref"}; Delva et al., [2008](#jgra53834-bib-0018){ref-type="ref"}; Thorne & Tsurutani, [1987](#jgra53834-bib-0056){ref-type="ref"}), in the Earth\'s polar wind (Le et al., [2001](#jgra53834-bib-0042){ref-type="ref"}), and within the Jovian and Kronian magnetospheres (Kivelson et al., [1996](#jgra53834-bib-0039){ref-type="ref"}; Leisner et al., [2006](#jgra53834-bib-0044){ref-type="ref"}). This article reports the first study addressing both positively and negatively charged pickup ions generating the Alfvén‐cyclotron instability, with application to the Galilean moon Europa. Within Jupiter\'s inner magnetosphere (\<15 *R* ~*J*~), the magnetic field is approximately dipolar and nominally orientated at right angles to the sub‐Alfvénic rotating magnetodisc. Ions picked up from satellites and rings in this environment will have low‐velocity components parallel to the magnetic field and form perpendicular rings in velocity space unstable to the generation of left‐hand (LH) polarized ICWs (Wu & Davidson, [1972](#jgra53834-bib-0062){ref-type="ref"}). If, however, the newly ionized material has a significant velocity component parallel to the ambient magnetic field, the picked up ions will form ring‐beam distributions in velocity space which results in a Doppler shift to the observed ICW resonant frequencies. In supra‐Alfvénic plasma flows such as the solar wind, this can cause LH ICWs to be observed as right‐hand (RH) polarized in the spacecraft frame (e.g., Jian et al., [2009](#jgra53834-bib-0033){ref-type="ref"}; Wicks et al., [2016](#jgra53834-bib-0059){ref-type="ref"}). It is also possible for obliquely propagating ICWs to become linearly polarized and to then reverse polarization if they attain and pass through the crossover frequency of the multicomponent plasma in which they are generated (Petkaki & Dougherty, [2001](#jgra53834-bib-0050){ref-type="ref"}; Rauch & Roux, [1982](#jgra53834-bib-0051){ref-type="ref"}). These gyroresonant transverse electromagnetic fluctuations have been observed to scatter pickup ion distributions into a bispherical shell distribution and toward thermal equilibrium (Coates et al., [1990](#jgra53834-bib-0010){ref-type="ref"}; Tokar et al., [2008](#jgra53834-bib-0057){ref-type="ref"}), with energy distributed between wave growth and ion heating (Cowee et al., [2007](#jgra53834-bib-0016){ref-type="ref"}; Huddleston et al., [1998](#jgra53834-bib-0030){ref-type="ref"}). The moon Europa orbits Jupiter at ∼8.8 *R* ~*J*~ and has been identified as the secondary mass loading source in the Jovian magnetosphere, contributing an estimated 1--100 kg/s of neutral material (Kivelson et al., [2009](#jgra53834-bib-0038){ref-type="ref"}; Shemansky et al., [2014](#jgra53834-bib-0054){ref-type="ref"}). Galileo observed a significant amount of plasma wave activity in the vicinity of the moon (Kurth et al., [2001](#jgra53834-bib-0041){ref-type="ref"}; Volwerk et al., [2001](#jgra53834-bib-0058){ref-type="ref"}) and, during one upstream encounter and two through the moon\'s plasma wake, observed bursty ICW characteristics at the gyrofrequencies of a number of species including K^+^, Na^+^, O ${}_{2}^{+}$, and Cl^+^. A notable trend within this data set was the occurrence of both LH and RH polarized wave power at the Cl^+^ gyrofrequency, a phenomenon absent at the gyrofrequency of other minority species picked up locally at the moon. These waves were highly field aligned, and the RH wave power was consequently hypothesized to result from the pickup of both positively and negatively charged chlorine ions, possible due to the high electron affinity (3.61 eV) and stable configuration of the chlorine anion, Cl^−^ (Kivelson et al., [2009](#jgra53834-bib-0038){ref-type="ref"}; Volwerk et al., [2001](#jgra53834-bib-0058){ref-type="ref"}). Negatively charged ions have been observed being picked up from the icy moon Rhea (Teolis et al., [2010](#jgra53834-bib-0055){ref-type="ref"}), from Saturn\'s main rings (Jones & Coates, [2014](#jgra53834-bib-0035){ref-type="ref"}), in the Enceladus plumes (Coates et al., [2010](#jgra53834-bib-0012){ref-type="ref"}), and in Titan\'s ionosphere (Coates et al., [2007](#jgra53834-bib-0009){ref-type="ref"}; Desai et al., [2017](#jgra53834-bib-0019){ref-type="ref"}) where the outflow of these species is also predicted (Ledvina & Brecht, [2012](#jgra53834-bib-0043){ref-type="ref"}). Negatively charged ions have also been observed in abundance at 1P/Halley and 67P/Churymov‐Gerasimenko (Burch et al., [2015](#jgra53834-bib-0007){ref-type="ref"}; Chaizy et al., [1991](#jgra53834-bib-0008){ref-type="ref"}) and likely form significant plasma populations in the outer solar system where the plasma temperatures are typically lower (lower associative detachment rates) and the solar photon flux is considerably reduced (lower photodetachment rates). The physics of ICWs generated by negatively charged pickup ions has, however, not been examined. The presence of chlorine pickup ions at Europa also suggests the moon to be a net source of this species with implications for the abundance of NaCl in the subsurface ocean. This work examines the physics of ICWs generated by both positively and negatively charged pickup ions with application to the Europan plasma environment. Section [2](#jgra53834-sec-0002){ref-type="sec"} describes a self‐consistent hybrid simulation approach which is used to study the growth and nonlinear evolution of the modes generated in the plasmas of interest. The analysis first looks to characterize the behavior of the negative ion ring instability using linear dispersion theory in section [3](#jgra53834-sec-0003){ref-type="sec"} and hybrid simulations in section [4.1](#jgra53834-sec-0005){ref-type="sec"}, which are contrasted to the well‐known characteristics and properties of the positive ion ring instability. Sections [4.2](#jgra53834-sec-0006){ref-type="sec"} and [4.3](#jgra53834-sec-0007){ref-type="sec"} go on to study scenarios of both instabilites generated within the same system to characterize any interaction between the two and their observable signatures. Section [5](#jgra53834-sec-0008){ref-type="sec"} then describes a parametric study of the instability saturation energy with ring properties such as density and anisotropy with close reference to the Galileo magnetometer observations. 2. Methodology {#jgra53834-sec-0002} ============== Warm plasma dispersion theory and hybrid simulations have previously constrained the behavior of the Alfvén‐cyclotron instability generated by heavy SO ${}_{2}^{+}$ ion rings in the Io plasma torus (Cowee et al., [2006](#jgra53834-bib-0015){ref-type="ref"}; Huddleston et al., [1997](#jgra53834-bib-0031){ref-type="ref"}, [1998](#jgra53834-bib-0030){ref-type="ref"}) and also for water group pickup ion rings within Saturn\'s extended neutral cloud (Cowee et al., [2009](#jgra53834-bib-0014){ref-type="ref"}; Leisner et al., [2006](#jgra53834-bib-0044){ref-type="ref"}; Rodríguez‐Martínez et al., [2010](#jgra53834-bib-0052){ref-type="ref"}). This study focuses on the Europan plasma environment and draws comparisons to these studies. The hybrid code has successfully reproduced the Alfvén‐cyclotron instability for a number of plasma environments (e.g., Cowee et al., [2006](#jgra53834-bib-0015){ref-type="ref"}; Gary et al., [1989](#jgra53834-bib-0024){ref-type="ref"}; Omidi et al., [2010](#jgra53834-bib-0049){ref-type="ref"}). The code specifies ions kinetically to capture phenomena occurring at ion spatial and temporal scales and approximates electrons as a massless neutralizing fluid (Winske et al., [1992](#jgra53834-bib-0060){ref-type="ref"}, [2003](#jgra53834-bib-0061){ref-type="ref"}). The simulations consist of one spatial dimension, *x*, aligned with the ambient magnetic field, **B** ~0~, a setup justified as the main wave power is expected at parallel propagation (Wu & Davidson, [1972](#jgra53834-bib-0062){ref-type="ref"}). All three components of electromagnetic fields and velocities are resolved to capture the predominantly transverse electromagnetic fluctuations along the *y* and *z* axes. The simulations are carried out in the plasma frame with zero‐drift velocity, and the particle‐in‐cell simulation technique is employed where all ions are represented by superparticles whose densities and currents are collected on a spatial grid and used as sources to the field equations. Maxwell\'s equation are implemented in the Darwin limit where the displacement current, *∂* **E**/*∂* *t*, is neglected within Ampére\'s law to eliminate high‐frequency perturbations such as light waves (Darwin, [1920](#jgra53834-bib-0017){ref-type="ref"}). In these initial value simulations the number of superparticles are specified at the beginning of the run and held as constant, in order to represent an injection of pickup ions from Europa, all particles are initialized with a quiet start configuration to minimize any net currents at the start of the run. The inputs into the simulation are representative of the general plasma environments around Europa with emphasis on the eleventh (E11) and fifteenth (E15) Galileo encounters. A variety of pickup velocities, from 55 to 100 km/s, are explored to examine the scaling of the instability with respect to inherent variations of this parameter due to effects such as the slowdown of the incident magnetoplasma in front of the moon, the formation of Alfvén wings, pickup ion currents, wake effects, and the flapping of the Jovian magnetodisk (Khurana et al., [1998](#jgra53834-bib-0036){ref-type="ref"}; Neubauer, [1998](#jgra53834-bib-0047){ref-type="ref"}, [1999](#jgra53834-bib-0048){ref-type="ref"}). Galileo observed electron densities from 30/cm^3^ up to several 100/cm^3^ during the various encounters, an average ion charge of 1.5 *q* ~*e*~ and a mean ion mass of 18.5 unified atomic mass units (u) dominated by low‐charge states of iogenic sulfur and oxygen (Bagenal et al., [2015](#jgra53834-bib-0002){ref-type="ref"}; Kivelson et al., [2009](#jgra53834-bib-0038){ref-type="ref"}). A nominal ion density of 100/cm^3^ and magnetic field of **B** ~0~=400 nT are taken as representative of the Galileo observations during E11 and E15. Further, simulation parameters are outlined in Table [1](#jgra53834-tbl-0001){ref-type="table-wrap"} which are used in all runs unless otherwise stated. ###### Nominal Plasma Parameters for All Simulations Runs Unless Otherwise Stated *j* *m* ~*j*~/*m* ~*p*~ *q* ~*j*~/*q* ~*p*~ *T* ~∥~ (eV) *T* ~⊥~ (eV) *v* ~ring~ (km s^−1^) ---------------- --------------------- --------------------- -------------- -------------- ----------------------- Light ion core 8 +1 100 100 ‐ Cl^−^ core 35 −1 100 100 ‐ Cl^+^ core 35 +1 100 100 ‐ Cl^−^ ring 35 −1 ∼0 1830 100 Cl^+^ ring 35 +1 ∼0 100 100 *Note.* *B* ~0~ = 400 nT; *n* ~0~ = 100/cm^3^; *c*/*ω* ~*pi*~ = 134.8 km; Ω~*i*~/2*π* = 0.17 Hz; *c*/*v* ~*A*~ = 344. The plasma is considered to consist of multiple ion species, either in a zero‐drift Maxwellian velocity distribution or a delta function velocity ring distribution with *T* ~∥~∼0. The *δ* **E** perturbations of the transverse waves act to scatter the ring ions in directions perpendicular to *B* ~0~, while the **v** × *δ* **B** forces scatters them in the directions parallel and antiparallel to *B* ~0~. Any other ion population with a similar mass per charge ratio will gyroresonantly interact with these waves and be subject to the same forces and must also be included in the simulations. Ion species with a different mass per charge ratio are not expected to interact with the instability and are represented by a light ion core population to maintain charge neutrality. Each species has mass, *m* ~*j*~, and charge state, *q* ~*j*~, normalized to the proton scale with parallel and perpendicular velocities, *V* ~∥*j*~=(2*k* ~*B*~ *T* ~∥*j*~/*m* ~*j*~)^1/2^ and *V* ~⊥*j*~=(2*k* ~*B*~ *T* ~⊥*j*~/*m* ~*j*~)^1/2^. *V* ~⊥~ is set as the pickup velocity, which is defined as the difference between the near‐corotational background plasma flow at the orbit of Europa of ≤117 km/s and the moons orbital velocity of ∼14 km/s. *V* ~∥~ is set to nearly zero, under the assumption of minimal relative motion between the atmospheric neutrals and the moon. The temperature anisotropy is defined as *A* ~*j*~=*T* ~⊥*j*~/*T* ~∥*j*~, the Alfvén velocity as $v_{A} = B_{0}/\sqrt{\mu_{0}n_{o}m_{j}}$, the ion gyrofrequency as Ω~*j*~=*q* ~*j*~ *B* ~0~/*m* ~*j*~, the plasma frequency as $w_{pi}^{2} = n_{0}q_{i}^{2}/\varepsilon_{0}m_{j}$, and the ion parallel plasma beta as $\beta_{\parallel j} = 2\mu_{0}n_{0}T_{\parallel j}/B_{0}^{2}$ where the fluid electron beta is equal to the sum of the ion components, $\beta_{e} = \sum\beta_{j}$. The subscript *p* denotes protons, *e* denotes electrons, and *i* denotes chlorine ions. The simulation outputs are given in normalized units of the ion Cl^+^/Cl^−^ gyrofrequency Ω~*i*~ where Ω~*i*~/2*π* = 0.17 Hz, the ion inertial length where *c*/*ω* ~*pi*~ = 134.8 km, and the Alfvén velocity where *v* ~*A*~ = 147.5 km/s. The simulation domain is 512 *c*/*ω* ~*pi*~ in length with 512 grid cells with periodic boundary conditions, and between 50 and 100 superparticles per cell are used within the various runs. The runs are then stepped forward in time steps of 0.05 $\omega_{pi}^{- 1}$, until some time after the instability has reached a quasi‐steady saturation energy level. 3. Linear Dispersion Theory {#jgra53834-sec-0003} =========================== Figure [1](#jgra53834-fig-0001){ref-type="fig"} illustrates solutions to the kinetic linear dispersion equation for electromagnetic fluctuations at **k** × **B** ~*o*~= 0 in a homogeneous, magnetized, collisionless plasma with three components: electrons and protons, each with Maxwellian velocity distributions, and a singly charged helium ion or anion component with a cold velocity ring distribution (e.g., see Gary & Madland, [1988](#jgra53834-bib-0025){ref-type="ref"}, with *α* = 90°). The pickup or ring velocity to Alfvén velocity ratio is specified as **v** ~**r**~/**v** ~**A**~=0.5 and the parallel beta of the proton is ***β*** ~∥~= **0.15**. These are in the range of observed values, with variations around these parameters not significantly altering the calculated solutions. Although the simulations described in subsequent sections address much heavier ions, we expect that the qualitative features of the various modes shown here should be similar to those of the simulations. {ref-type="fig"}a and [1](#jgra53834-fig-0001){ref-type="fig"}e show dispersion of the left‐hand and right‐hand helium cyclotron instabilities, respectively; Figures [1](#jgra53834-fig-0001){ref-type="fig"}b and [1](#jgra53834-fig-0001){ref-type="fig"}f illustrate dispersion of the stable right‐hand polarized magnetosonic modes, Figures [1](#jgra53834-fig-0001){ref-type="fig"}c and [1](#jgra53834-fig-0001){ref-type="fig"}g represent the left‐hand polarized stable Alfvén‐cyclotron modes, and Figures [1](#jgra53834-fig-0001){ref-type="fig"}d and [1](#jgra53834-fig-0001){ref-type="fig"}h illustrate the left‐hand and right‐hand polarized stable helium cyclotron modes, respectively.](JGRA-122-10408-g001){#jgra53834-fig-0001} Plasma parameters used in Figure [1](#jgra53834-fig-0001){ref-type="fig"} are given in the figure caption. The complex frequency is *ω* = *ω* ~*r*~+*i* *γ* with positive *γ* corresponding to fluctuation growth. The left‐hand column shows results for a He^+^ ion velocity ring, whereas the right‐hand column illustrates results for a He^−^ anion velocity ring. In both cases the dispersion equation yields four distinct modes near or below the cyclotron frequency with both positive and negative helicities. Magnetic helicity defines the sense of the spatial rotation of the fluctuating field vectors with respect to the wave vector, **k**, at a given instance and is independent of the reference frame of the observer (Gary, [1993](#jgra53834-bib-0023){ref-type="ref"}). A positive helicity corresponds to either a forward propagating RH wave or a backward propagating LH wave and a negative helicity corresponds to either a backward propagating RH or forward propagating LH wave. The fastest growing modes in both cases are the helium cyclotron instabilities shown in Figures [1](#jgra53834-fig-0001){ref-type="fig"}a and [1](#jgra53834-fig-0001){ref-type="fig"}e. The figure suggests that these two modes are unstable to arbitrarily short wavelengths, however, for the more realistic case of a warm ring (*T* ~\|\|~\> 0) ion cyclotron resonances will damp modes at short wavelengths (e.g., Figure 2 of Gary & Madland, [1988](#jgra53834-bib-0025){ref-type="ref"}) so that we expect maximum growth to arise near *k* *c*/*ω* ~*pi*~. In addition to these, Figures [1](#jgra53834-fig-0001){ref-type="fig"}b and [1](#jgra53834-fig-0001){ref-type="fig"}f show the right‐handed magnetosonic whistler mode that is undamped, and Figures [1](#jgra53834-fig-0001){ref-type="fig"}c and [1](#jgra53834-fig-0001){ref-type="fig"}g show the Alfvén‐cyclotron modes which undergo strong proton cyclotron damping at short wavelengths, and Figures [1](#jgra53834-fig-0001){ref-type="fig"}d and [1](#jgra53834-fig-0001){ref-type="fig"}h which illustrate the helium cyclotron waves which also become damped at sufficiently short wavelengths. 4. Results From Hybrid Simulations {#jgra53834-sec-0004} ================================== 4.1. Negative Ring Instability {#jgra53834-sec-0005} ------------------------------ In simulation Run I, the instability is simulated with negatively charged Cl ions. As outlined in Table [2](#jgra53834-tbl-0002){ref-type="table-wrap"}, a cold Cl^−^ ring distribution is initialized with negligible temperature spread parallel to the magnetic field together with a thermalized Cl^−^ core population. A further positively charged Maxwellian light ion core represents the ambient low‐charge state oxygen and sulfur background ions of the Jovian magnetodisk. This background population was initially specified as protons although a higher mass ion was later used so that larger time steps could be taken while not affecting the growth of the instability and still accurately resolving the ion gyromotion. Figure [2](#jgra53834-fig-0002){ref-type="fig"} shows the fluctuating magnetic field energy density, (*δ*B/*B* ~0~)^2^, and temperature evolution of the different species. The instability initially grows rapidly until Ω~*i*~ *t*= 60--70 then grows slowly until Ω~*i*~ *t*= 110--120 where a quasi‐steady level of (*δ*B/*B* ~0~)^2^=∼2.7 × 10^−4^ is reached. ###### Case Study Simulations for Runs I--IV Described in the Article Text Run Components *n* ~*i*~ (1/cm^3^) ----- ---------------- --------------------- --------- --------- I Light ion core 80 Cl^−^ core 15 Cl^−^ ring 5 II Light ion core 60 Cl^+^ core 15 Cl^−^ core 15 Cl^+^ ring 5 Cl^−^ ring 5 (Both) (Cl^+^) (Cl^−^) III Light ion core 98 98.5 99.5 Cl^+^ ring 1.5 1.5 ‐ Cl^−^ ring 0.5 ‐ 0.5 IV Light ion core 98 98.1 99.9 Cl^+^ ring 1.9 1.9 ‐ Cl^−^ ring 0.1 ‐ 0.1 *Note.* Further plasma input parameters are given in Table [1](#jgra53834-tbl-0001){ref-type="table-wrap"}. {ref-type="table-wrap"} and [2](#jgra53834-tbl-0002){ref-type="table-wrap"}.](JGRA-122-10408-g002){#jgra53834-fig-0002} The fluctuations shown in Figure [2](#jgra53834-fig-0002){ref-type="fig"}a act to pitch angle scatter the anisotropic Cl^−^ ring distribution toward isotropy as shown in Figures [2](#jgra53834-fig-0002){ref-type="fig"}c--[2](#jgra53834-fig-0002){ref-type="fig"}e. The Cl^−^ core population acts to damp the growth of the waves and is consequently heated in the perpendicular direction, while the light ion core remains at a constant temperature as anticipated for components not resonantly interacting with the instability. Figure [3](#jgra53834-fig-0003){ref-type="fig"} shows the *ω*‐*k* spectrum during the growth phase of the instability. The wave power grows out of the Alfvén branch and is concentrated near the chlorine cyclotron frequency with the bulk wave power occurring between *k* *c*/*ω* ~*pi*~ = 1--2. This agrees with the dispersion relations given in Figure [1](#jgra53834-fig-0001){ref-type="fig"} where positive growth rates and similar frequencies are predicted near *k* *c*/*ω* ~*pi*~. The dynamic spectrum in Figure [4](#jgra53834-fig-0004){ref-type="fig"} shows the power spectral density (PSD) of Run I and the magnetic ellipticity as calculated by the UCLA X waves analysis tool, which utilizes the inherently antisymmetric quadrature spectral matrix to provide information on the propagation direction and thus polarization of separated circularly polarized wave components (Means, [1972](#jgra53834-bib-0046){ref-type="ref"}). In the case of superimposed waves with different polarizations, this technique will return the net resultant ellipticity. This was generated using a time series determined from one grid cell in the simulation and while other grid cells may yield slightly different results, the simulation conditions are initially uniform and so significant differences between cells are not expected. The wave power occurs at the Chlorine gyrofrequency as in Figure [3](#jgra53834-fig-0003){ref-type="fig"} but decreases slightly in frequency over time as a result of an inverse cascade to longer wavelengths (Gary & Madland, [1988](#jgra53834-bib-0025){ref-type="ref"}). The variability in the wave power is due to the superposition of a range of modes excited in the simulation. The resultant ellipticities appear close to +1, indicating a RH near circularly polarized wave near the chlorine gyrofrequency as predicted by the positive helicity mode in Figure [1](#jgra53834-fig-0001){ref-type="fig"}. {#jgra53834-fig-0003} {ref-type="fig"}b is close to +1, indicating RH polarized wave power.](JGRA-122-10408-g004){#jgra53834-fig-0004} The results of the negative ion ring instability presented in Figures [2](#jgra53834-fig-0002){ref-type="fig"}, [3](#jgra53834-fig-0003){ref-type="fig"}, [4](#jgra53834-fig-0004){ref-type="fig"} show that a negatively charged ring generates a RH instability in‐line with the predictions of linear theory and is analogous to that of the LH instability studied in Huddleston et al. ([1997](#jgra53834-bib-0031){ref-type="ref"}) and Cowee et al. ([2006](#jgra53834-bib-0015){ref-type="ref"}). 4.2. Simultaneous Pickup of Positive and Negative Ions {#jgra53834-sec-0006} ------------------------------------------------------ This section examines the cogeneration and interaction of both positive (LH) and negative (RH) Alfvén‐cyclotron instabilities within the same system. To explore these effects, Run II contains equal quantities of Cl^+^ and Cl^−^ ring and core ions as outlined in Table [2](#jgra53834-tbl-0002){ref-type="table-wrap"}. This results in twice the amount of free energy available to drive wave growth, and, in the absence of interactions between the two rings, the saturation energy should reach double that of Run I (Figure [2](#jgra53834-fig-0002){ref-type="fig"}a) where only the negative ring ions are initialized. Figure [5](#jgra53834-fig-0005){ref-type="fig"} shows the fluctuating magnetic field energy density growing as anticipated although in this instance intense high‐frequency perturbations are apparent. These oscillations cause the magnetic field energy density, (*δ*B/*B* ~0~)^2^, to reach levels of up to 4 times that of Run I, although a running average produces the anticipated saturation energy level of (*δ*B/*B* ~0~)^2^=∼5.4 × 10^−4^, twice that of Run I. These high‐frequency oscillations occur at a frequency of approximately twice the chlorine ion gyrofrequency although the intensity also changes in a further beating pattern along a time scale of Ω~*i*~ *t*= 25--50 indicating the superposition of multiple modes. The high‐frequency oscillations are also apparent in the temperature histories of the Chlorine core components and to a lesser extent in the nonresonant light ion core. To further examine this behavior, the phase space angular distributions of the different components are plotted in Figure [6](#jgra53834-fig-0006){ref-type="fig"}. Here both the Cl^+^ and Cl^−^ core and ring components can be seen to be bunched in gyrophase space with particles clustered around common positions while gyrating about *B* ~0~. {ref-type="table-wrap"} and [2](#jgra53834-tbl-0002){ref-type="table-wrap"}.](JGRA-122-10408-g005){#jgra53834-fig-0005} {#jgra53834-fig-0006} Nongyrotropic ion distributions have been observed at Comet 26P/Grigg‐Skjellerup (Coates et al., [1993](#jgra53834-bib-0011){ref-type="ref"}) and in the distant Earth\'s magnetotail (Saito et al., [1994](#jgra53834-bib-0053){ref-type="ref"}). Subsequent studies on these determined that increasing the nongyrotropy of anisotropic ion distributions resulted in increased instability growth rates and a larger range of excited wave numbers (Brinca et al., [1993](#jgra53834-bib-0005){ref-type="ref"}; Convery et al., [2002](#jgra53834-bib-0013){ref-type="ref"}). These studies also show, using hybrid simulations, how nongyrotropic ring distributions result in large magnetic oscillations as energy is exchanged between the waves and the time‐varying bunched ion distributions. These effects appear analogous to what is seen in Run II although in this study the ions are initialized in a gyrotropic state and develop their nongyrotropy as the simulation progresses. Density oscillations at half the wavelength of the ICW mode have previously been observed in simulations of anisotropic ring current ions in the Earth\'s magnetosphere (Omidi et al., [2010](#jgra53834-bib-0049){ref-type="ref"}). These were attributed to develop from oppositely directed ICWs intersecting one another twice within each full wave rotation and introducing periodic effects at twice the resonant wave frequency. This caused the core ions to become bunched in gyrospace as seen in Run II although an associated electrostatic periodicity was also reported which is not observed in this study. The half‐length oscillatory behavior seen in Run II also appears significantly greater than that reported in Omidi et al. ([2010](#jgra53834-bib-0049){ref-type="ref"}), presumably due to the presence of both LH and RH waves acting to bunch both positively and negatively charged core components. The cogeneration of the LH and RH instabilities could also introduce multiple half‐length modes which will naturally vary over time moving in and out of phase and could explain the large variations in intensity. Figure [7](#jgra53834-fig-0007){ref-type="fig"} displays the *ω*‐*k* spectrum for Run II and shows that the wave power is spread over wave numbers and frequencies similar to those of Run 1 (Figure [2](#jgra53834-fig-0002){ref-type="fig"}a), although in this instance there is increased wave power due to double the density of ring ions included. The dynamic spectra in Figure [8](#jgra53834-fig-0008){ref-type="fig"} also shows the wave power appearing at and just below the chlorine ion gyrofrequency. The wave power, however, displays no clear circular polarization with ellipticity values sometimes displaying positive or negative values. This effective near‐linear polarization agrees with the theoretical expectation of the summation of LH and RH circularly polarized waves at similar frequencies and wave numbers. The variability in the polarization is attributed to the time‐dependent fluctuating wave amplitudes, frequencies, and wave numbers for each instability which allows either the LH or the RH mode to appear dominant for short intervals. This effect could also be amplified as a result of the gyrophase bunching. {#jgra53834-fig-0007} {ref-type="fig"}, and the magnetic polarization is approximately linear as anticipated from the summation of the LH and RH near circularly polarized components although periods of both LH and RH polarizations are present.](JGRA-122-10408-g008){#jgra53834-fig-0008} 4.3. The Magnetic Signature of Negative Ion Pickup {#jgra53834-sec-0007} -------------------------------------------------- This section describes an ensemble of simulations which further characterize the interaction and scaling of the two instabilities and explore how a trace negative ion population is able to generate an observable signal. These runs cover a wide range of Cl^+^ and Cl^−^ densities, and two specific runs are chosen, termed Run III and Run IV in this article, to illustrate the key findings. Run III and Run IV nominally contain a 3:1 and 19:1 Cl^+^:Cl^−^ pickup ion density ratio, respectively, as outlined in Table [2](#jgra53834-tbl-0002){ref-type="table-wrap"}. Figure [9](#jgra53834-fig-0009){ref-type="fig"} shows the fluctuating magnetic field energy density, *ω*‐*k* spectrum, and magnetic polarization for both Run III and Run IV. The Cl^+^ and Cl^−^ instabilities are also simulated individually and the results overlaid. {ref-type="table-wrap"}. (b and c) The frequency‐wave number spectrum near the chlorine ion gyrofrequency (dashed black line) for both the Cl^+^ and Cl^−^ instabilities, calculated using a Fourier window over 187.5 \< Ω~*i*~ *t* \< 281.2. (e and f) The magnetic polarization for both Cl^+^ and Cl^−^ instabilities.](JGRA-122-10408-g009){#jgra53834-fig-0009} In Runs III and IV the Cl^+^ instability has an increased growth rate and wave amplitudes compared with the significantly reduced Cl^−^ instability and the sum of the LH and RH instabilities are in good agreement with the amplitudes produced when both components are included. In *ω*‐*k* space the LH and RH instabilites appear as distinct pockets of wave power at different wave numbers. The wave power still occurs near the chlorine ion gyrofrequency, but the Cl^−^ instabilities are excited at slightly different frequencies to the Cl^+^ instabilities in both instances. During Run III the magnetic polarization near the chlorine ion gyrofrequency is predominantly LH polarized although, at Ω~*i*~ *t* = 200, when the weaker Cl^−^ RH instability amplitudes become comparable and briefly exceed those of the Cl^+^ LH instability, a polarization reversal occurs and the ICW power becomes predominantly RH polarized before returning to a LH polarization. In Run IV, however, despite the Cl^−^ wave amplitudes never exceeding those generated by the Cl^+^ ring, a polarization reversal is surprisingly also present and a brief burst of RH polarized wave power results at Ω~*i*~ *t* = 250. This phenomenon is attributed to the two instabilities having different growth rates and generating wave power at different wavelengths and frequencies. The *ω*‐*k* spectrum indicates that during this brief instance of a RH‐dominated spectrum, the RH wave power is generated very close to the chlorine ion gyrofrequency, whereas the LH instability saturates earlier and although possessing more wave power, this wave power is spread across a wider range of wave numbers and centers at a lower frequency. This effect means that despite the reduced growth rate of the Cl^−^ instability, the wave power is still able to generate an observable RH signal as a result of the natural time‐dependent variability of the wave amplitudes. The high‐frequency oscillatory behavior is also reduced in Run III and even more so in Run IV, presumably due to the LH and RH components occurring at different wave numbers. This suggests that this half‐length mode is more strongly driven when LH and RH ICWs of a similar wave number and frequency are present. The results obtained from Run III and Run IV demonstrate how positive and negative ring instabilities, although similar in nature, can have different spectral properties as a result of differences in the initial ion distribution functions. The simulations thus show how the appearance of RH wave power within a predominantly LH signal, as was observed by Galileo, can be evidence of negatively charged pickup ions generating a RH Alfvén‐cyclotron instability. The simulations also indicate how an inherent polarization variability will be present in systems containing instabilities driven by both positive and negative ion rings. 5. Application to Europa {#jgra53834-sec-0008} ======================== To better understand the processes occurring at Europa, the simulated waves are compared to the ICWs observed by Galileo. The dominant pickup ion at Europa is O ${}_{2}^{+}$ produced from the moon\'s oxygen‐based exosphere (Hall et al., [1995](#jgra53834-bib-0027){ref-type="ref"}; Johnson et al., [2009](#jgra53834-bib-0034){ref-type="ref"}), but large concentrations of thermalized S^+^ ions originating from Io have a similar mass per charge ratio and resonantly interact to damp the growth of waves near the O ${}_{2}^{+}$ gyrofrequency. Consequently, O ${}_{2}^{+}$ ICWs are not observed in significant quantities and during the E11 encounter these waves appear to be generated in a higher‐frequency band to those concentrated near the Cl^+/−^ gyrofrequency. During E15, however, it is difficult to rule out LH wave power from ICWs generated by the pickup of O ${}_{2}^{+}$ (see Volwerk et al., [2001](#jgra53834-bib-0058){ref-type="ref"}, Plates 1 and 2). The E11 and E15 observations of wave power at the chlorine gyrofrequency occurred between 3 and 4 Europa radii (*R* ~Eur~∼1561 km) downstream of the moon (Volwerk et al., [2001](#jgra53834-bib-0058){ref-type="ref"}). The simulation results presented in Runs I--IV show, for the various densities examined, the instability saturates within 200--320 Ω^−1^, which, in the plasma frame, corresponds to a convection distance of \<1 *R* ~Eur~. The simulations therefore suggest that the Cl^+^ and Cl^−^ ring instabilities would likely have reached saturation when observed by Galileo, if the ions were picked up close to the moon. To relate possible source densities to the Galileo observations, a parametric study of the Alfvén‐cyclotron instability is carried out for magnetic field fluctuations in the range observed by Galileo; see Table [3](#jgra53834-tbl-0003){ref-type="table-wrap"}. During the E11 and E15 encounters the wave amplitudes at the Cl^+/−^ gyrofrequency reached absolute values of 2.5 nT, (*δ* *B*/*B* ~0~)^2^=3.9 × 10^−5^, and 3.5 nT, (*δ* *B*/*B* ~0~)^2^=7.7 × 10^−5^, respectively. The wave saturation amplitude is necessarily dependent on the background density of Cl^+/−^ ions which will act to damp the growth of the instability. In Figure [10](#jgra53834-fig-0010){ref-type="fig"} the ring ion densities are plotted against saturated amplitude for the two limiting cases of a varying core population where the chlorine ring and core ion densities sum to 5%, and for zero‐core damping. The amplitude modulations brought upon by gyrophase bunching are not included in the saturation amplitudes as a running average was used. This effect would complicate the direct interpretation of the wave amplitudes and introduce a factor of 2 additional uncertainty. The pickup or ring velocity is also varied across the anticipated range observed near the moon (Kivelson et al., [2009](#jgra53834-bib-0038){ref-type="ref"}), although during E11 and E15 this was in the upper end of this range (Volwerk et al., [2001](#jgra53834-bib-0058){ref-type="ref"}). ###### Parametric Study of the Scaling of the Chlorine Ion Ring Instability in the Low‐Density Regime Applicable to the Galileo E11 and E15 Encounters With Europa Components Density (1/cm^3^) *n* ~*r*~/*n* ~0~ Pickup velocity (km/s) ------------------ ------------------- ------------------- ------------------------ Light ion core 95 − − Cl^+^/Cl^−^ core 0, 1.9--0.1 ‐ ‐ Cl^+^/Cl^−^ ring 0.1--2, 0.1--1.9 0.05, 1 55, 100 *Note.* The parameter *n* ~*r*~ represents the ring ion density, and *n* ~0~ represents the total chlorine ion density. Further plasma parameters are given in Table [1](#jgra53834-tbl-0001){ref-type="table-wrap"}. {ref-type="sec"}. The Galileo maximum wave amplitudes during E11 (dash‐dotted line) and E15 (dotted line) are overlaid. The input parameters are provided in Tables [1](#jgra53834-tbl-0001){ref-type="table-wrap"} and [3](#jgra53834-tbl-0003){ref-type="table-wrap"}.](JGRA-122-10408-g010){#jgra53834-fig-0010} Linear theory predicts the growth rate of the instability is proportional to the ring velocity and increases as the ring‐to‐core relative density increases (Huddleston et al., [1997](#jgra53834-bib-0031){ref-type="ref"}). Larger amplitude waves are therefore generated due to the increased amount of free energy available, although linear theory is unable to predict at what amplitudes the instability will saturate. Previous studies have, however, constrained the relationship between saturation amplitudes and ring density and anisotropy (e.g., Cowee et al., [2006](#jgra53834-bib-0015){ref-type="ref"}; Fu et al., [2016](#jgra53834-bib-0022){ref-type="ref"}; Huddleston & Johnstone [1992](#jgra53834-bib-0029){ref-type="ref"}), which is well reproduced in Figure [10](#jgra53834-fig-0010){ref-type="fig"} for the negative ion ring instability when relating the varying ring energy to the saturation energy levels, (*δ* *B*/*B* ~0~)^2^. From this parameter space the Cl^+^ and Cl^−^ total maximum densities are constrained to a range of 0.1--1.5/cm^3^ required to produce the 2.5 and 3.5 nT peak values observed during the E11 and E15 Galileo flybys through Europa\'s wake. This spread is dependent on the unknown background chlorine ion density and exact pickup velocity. Cl^+^ densities have been placed at between 1 and 2.5% of the Io plasma torus (Feldman et al., [2001](#jgra53834-bib-0021){ref-type="ref"}; Küppers & Schneider, [2000](#jgra53834-bib-0040){ref-type="ref"}) and highlighted as plausibly present in neutral, negatively, and multiply charged states, as well as produced from Io within compounds such as FeCl~2~, CuCl, ZnCl~2~, NaCl, KCl, and MgCl~2~ (Brown & Hand, [2013](#jgra53834-bib-0006){ref-type="ref"}; Fegley & Zolotov, [2000](#jgra53834-bib-0020){ref-type="ref"}). These compounds will diffuse outward toward Europa and likely become implanted within the surface ice before being sputtered into the near‐Europa plasma environment and torus, in addition to any local source. Cl^+^ and Cl^−^ ions produced locally at Europa would also contribute to the Europan torus and may also contribute to wave damping. The chemical pathways describing the Cl^+^ and Cl^−^ production and loss processes are, however, beyond the scope of this analysis, and the results are parametrized with respect to the unknown background chlorine ion densities. It should, however, consequently be noted how important it is to constrain the iogenic material diffusing outward in Jupiter\'s magnetosphere to understand processes occurring locally at Europa. The upper limits on the inferred peak Cl^+^ and Cl^−^ pickup densities appear high for a sputtered minority species ultimately derived from Io. NaCl has, however, been suggested to be abundant within Europa\'s subsurface ocean (Kivelson et al., [2000](#jgra53834-bib-0037){ref-type="ref"}; Zimmer et al., [2000](#jgra53834-bib-0063){ref-type="ref"}) and could therefore feasibly be present in significant quantities within the Europan ice and plumes. A recent study by Hand and Carlson ([2015](#jgra53834-bib-0028){ref-type="ref"}) also suggest that the characteristic brownish streaks on the moon\'s surface are composed of high quantities of heavily irradiated NaCl, discolored as a result of the intense \>500 krad Europan radiation environment. A Na cloud has also been observed to envelope the moon, and, although Na can be produced through other processes as observed at Mercury (McGrath et al., [1986](#jgra53834-bib-0045){ref-type="ref"}), significant quantities of Cl ions in localized regions are not inconsistent with physical processes inferred to be occurring at the moon. The LH and RH Galileo wave observations are too brief to directly establish the respective amplitudes for the positive and negative ion instabilities. To interpret these further, Kivelson et al. ([2009](#jgra53834-bib-0038){ref-type="ref"}) show the magnetic field rotations using 15 s hodograms, indicating that the RH wave power is detected over the order of this time period. For the parameter space explored, the dynamic spectra showed evidence of both LH and RH polarizations for density ratios as low as 1:19 Cl^−^:Cl^+^. Below this lower limit of ∼5% the Cl^−^ instability was not sufficiently strong to produce a polarization reversal. 6. Summary and Conclusions {#jgra53834-sec-0009} ========================== This study used a hybrid simulation technique to characterize the physics of ICWs generated by anisotropic distributions of positive and negative pickup ions. These simulations have verified the predictions of linear dispersion theory and were used to relate wave amplitudes and wave polarizations observed by Galileo to local ion populations. Through the characterization of the negative ion ring driven instability and the simultaneous generation of the positive and negative ring instabilities within the same system, the following were concluded: A negatively charged pickup ion ring will generate near circularly polarized RH waves consistent with theoretical predictions.The RH ring instability behaves analogously to the LH ring instability with respect to the resonant frequencies, wave numbers excited, and interactions with a gyroresonant core population.The presence of both RH and LH instabilities act to bunch both ring and core ion populations which results in an additional and potentially significant factor of ≤2 modulation to the wave amplitudes.The presence of both positive and negative ring ions results in variability in ICW polarization with the dynamic spectra displaying LH, RH, and linear polarization, the latter as anticipated for oppositely handed circularly polarized waves with otherwise equivalent spectral properties.Despite the modulations associated with gyrophase bunching, the positive and negative ring instabilities behaved independently as indicated by the time histories of their average wave amplitudes, component temperatures, and wave spectral properties. After characterizing the behavior of the instability in the Europan plasma environment, the simulations and Galileo wave observations were compared to determine if Cl^−^ pickup ions are able to explain the wave signatures observed by Galileo and if it is possible to estimate the absolute densities of the chlorine ions from the observed waves. The following were concluded: The simulations and linear theory demonstrate that for the Europan pickup geometry, the RH wave power observed by Galileo is consistent with the pickup of Cl^−^.The simulations suggest that variability in the wave polarizations is observed in systems which contain instabilities driven by positive and negative ion rings, as was observed by Galileo.From the overall E11 and E15 wave amplitudes the Cl^+^ or Cl^−^ densities are estimated to peak in the range of 0.1--1.5/cm^3^, this spread dependent on the unknown background Cl^+^ and Cl^−^ densities and exact pickup velocity.The ICWs at the chlorine ion gyrofrequency observed by Galileo were too brief to establish amplitudes for the LH and RH waves individually, and a threshold of ∼5% of the total chlorine pickup ions was established, below which an observable signal of the Cl^−^ instability would not be present.The occurrence of bunched ion populations, caused by the presence of both LH and RH ICWs can, however, complicate interpretation of ICWs driven by positive and negative pickup ions. This study has investigated the behavior of the RH Alfvén‐cyclotron instability and also a method of using the wave polarization to identify the presence of negatively charged pickup ions. This method is envisaged to be relevant to other environments in the outer solar system where negative ions can exist in abundance and also provides context for planned in situ observations to be gained from the upcoming ESA JUpiter ICy moons Explorer (JUICE) mission and the NASA Europa Clipper mission. R. T. D. acknowledges a LANL Vela Fellowship, STFC Studentship 1429777 and an RAS grant. R. T. D. and M. C. acknowledge the Los Alamos Space Weather Summer School funded by the Center for Space and Earth Sciences (CSES) at LANL. X. F. was supported by Energetic Particle, Composition, and Thermal Plasma (RBSP‐ECT) investigation funded under NASA\'s Prime contract NAS5‐01072. The research effort of S. P. G. was supported by the National Aeronautics and Space Administration grant NNX16AM98G. A. J. C. acknowledge support from the 241 STFC consolidated grant to UCL‐MSSL ST/N000722/1. The simulation data in this paper are available by contacting the corresponding author.

Introduction {#S1} ============ Health care workers (HCWs) are defined as persons working in health-care milieu having potential exposure to patients and to infectious materials; they are exposed to infectious diseases more than the general population although many of these infections are preventable by vaccination. When infected, HCWs may transmit the infection to patients or colleagues \[[@R01]\]. Outbreaks due to vaccine-preventable diseases in health care facilities are often associated with serious morbidity and mortality among patients, disruption of healthcare services, and high costs for the National Health Service \[[@R02]\]. Immunization of HCWs is a major infection prevention measure \[[@R03]\] to protect HCWs and, indirectly, the patients who may not be naturally immunized or vaccinated \[[@R04]\]. Vaccination is a very effective tool to promote safety within health care facilities for both HCWs and patients. Nevertheless, inadequate coverage of HCWs against vaccine-preventable diseases is a global problem \[[@R05]\]. In Italy, vaccination policies for HCWs are based -- on one hand -- on the \"National Vaccination Plan\" \[[@R06]\] which reinforces the recommendation to vaccinate HCWs against selected vaccine preventable diseases. On the other hand, the Italian Legislative Decree 81/08 reorganizes and revises all the provisions regarding health and safety at work. In particular, the Decree ratifies the obligation to the employers to provide vaccination to workers exposed to biological risk and emphasizes the key role of the Occupational physician for assessing biological risk, identifying and managing susceptible workers, and organizing and implementing vaccine campaigns. Therefore, employers and Occupational physicians have to manage vaccination activities within hospitals for their workers. In this sense, the following vaccinations are strongly recommended: anti-hepatitis B, anti-influenza, anti-measles-mumps-rubella, anti-varicella and anti-pertussis. The Italian law has drastically limited the use of the anti-TB vaccination to very few categories of HCWs: those who are at high risk of exposure to multi- drug-resistant tubercular bacterial strains, and those who work in high-risk environments and -- in the case of positive Mantoux Test -- they cannot undergo preventive therapy because of clinical contraindications to the use of specific drugs. The aim of our study was to investigate (1) whether, how, and which vaccination underwent the Sardinia HCW\'s; (2) the variability of policies and solutions in different Hospital Health Managements (HHM) of the whole region. Consequently, an indirect aim was to raise awareness about HHMs\' compliance rate to vaccination programs and orient them in a positive approach towards vaccination in Sardinia. Methods {#S2} ======= In March 2013, we enrolled by both phone and a written questionnaire sent by e-mail the 32 HHM of all Sardinia hospitals which have totally 5,650 beds serving the whole Sardinia population (1.600.000 inhabitants). A self-reported structured form was completed by HHMs. In case of lacking or inadequate response HHMs were successively contacted by phone to complete information. According to some critical aspects highlighted in 2010 by a team of occupational physicians of the Italian Society of Occupational Health and Industrial Hygiene (SIMLII), the questionnaire included 10 items based on multiple-choice questions. Each hospital was allowed to mark more than one answer per item. The study investigated about the immunological pattern of HCWs towards Influenza, hepatitis B (HB), measles- mumps-rubella (MMR), varicella and tuberculosis (TBC). Furthermore, information was collected about the performed educational campaign for influenza vaccination. Specifically, five questions concerned flu vaccination and investigated (a) effective accomplishment of the educational campaign on vaccination and its starting, (b) the campaign planners, (c) the different strategies used to inform health professionals about the opportunity to undergo vaccination, (d) the topics used during the campaign, and (e) available data on vaccination coverage. One question regarded HB vaccination and the gathered data about compliance to vaccination. Two questions were addressed to serological surveillance: the first one asked whether serological tests for detection of MMR and Varicella viruses\' antibodies were included in routine surveillance of HCWs by Occupational Health Physicians. The other one asked whether serological results were collected. The last two questions concerned TBC: one investigated whether and in which occasions HCWs underwent Mantoux skin test, the other one detected whether Mantoux was replaced or integrated by \"in vitro\" test. The results were returned to each HHM in order to provide them a comparison with the other hospitals. In addition, we suggested recommendations about vaccination in HCWs according to National laws and National and International scientific communities. Results {#S3} ======= Out of the 32 hospitals enrolled in the study, 30 joined our survey (94% compliance): three were University Hospitals and 27 were General Hospitals. The two nonresponding hospitals were long-term care institutes. The total number of HCWs of the 30 hospitals was 12,977. Among these, 26% were medical staff, 47% nurses, and 27% represented other hospital staff. INFLUENZA {#S3a} --------- Twenty-five out of the 30 hospitals (77%) accomplished the vaccination campaign against flu carried out between October and December 2011, particularly November, in 18 hospitals. All the hospitals answered to the questions concerning the communication methods during the flu campaign: 18 hospitals (72%) informed HCWs about the chance to be vaccinated by newsletter to unit managers, nursing and technicians coordinator, 8 (32%) sending newsletter to unit managers only, 7 (28%) through active call, 2 (8%) using posters and advertising. The coordination of the campaign was performed by HHM in 20 hospitals (80%), by Health visitors in 8 (32%) and by Occupational Health Physicians in 7 hospitals (28%). In 8 (32%) hospitals the Prevention Department managed the campaign. Therefore different professionals were involved in managing and coordinating the campaign in the different hospitals at the same time. This heterogeneity may depend on the hospital characteristics in terms of structure, complexity, size, and (urban or rural) area. Concerning the topics used during the vaccination campaign to inform HCWs all 25 hospitals enlightened about professional and social responsibility in patients\' protection; 20 reported also efficacy and safety of the vaccine and 17 warned about the frequency and severity of influenza. Data of vaccination coverage among HCWs were reported only by 23 hospitals. The coverage rate is shown in [Figure 1](#F1){ref-type="fig"}; 74% of the hospitals indicated a coverage rate lower than 10%. {#F1} HEPATITIS B {#S3b} ----------- Data on immunization coverage were available in 14 of 30 hospitals ([Fig. 2](#F2){ref-type="fig"}). Regional average was 76%; 5 hospitals were below the average and 9 were above, only one hospital had 92% vaccination coverage. Antibody assay (anti-HBs) was carried out in all the 30 respondent hospitals but only 14 had electronically collected data. {#F2} VARICELLA, MEASLES, MUMPS, RUBELLA {#S3c} ---------------------------------- Most hospitals did not perform serological tests to detect antibodies against Varicella, Measles, Mumps and Rubella viruses in HCWs. Three (10%) hospitals applied health surveillance for those viruses and other two hospitals performed serological test for Rubella in higher risk occupational categories. No hospital had available data on immunity to those infections. TUBERCULOSIS {#S3d} ------------ Intradermal Mantoux test was used in 29 of 30 hospitals: in 32% of hospitals HCWs were submitted to the Mantoux test at the hiring time and/or in post exposure situations. In 1 hospital, Mantoux test was replaced by \"in vitro\" test (Quantiferon) and in 20 hospitals Quantiferon integrated Mantoux test. Discussion {#S4} ========== Vaccination is an important prophylactic action to reduce the number of susceptible HCWs and to indirectly protect susceptible patients and colleagues \[[@R07]\]. An additional benefit is a reduction in work time lost due to illness. Despite the persistence of outbreaks of vaccinepreventable diseases in health care facilities, HCWs vaccination rates remain suboptimal globally. Higher vaccination coverages among HCWs than the actual would be useful both to reinforce occupational safety in health care facilities and to prevent nosocomial outbreaks \[[@R01], [@R08], [@R09]\]. The Italian Health Ministry recommends HCWs to undergo vaccinations (i.e. anti-hepatitis B, anti-flu, antimeasles- mumps-rubella, varicella, anti-pertussis and, when indicated, anti-tuberculosis) not only to reduce employees\' absenteeism that may increase workload for staff, but also to prevent the nosocomial infection risk, especially in elderly or immunocompromised patients \[[@R06]\]. The same Institution have also included among the new objectives of the National Prevention Plan 2014-2018 an increasing vaccination coverage for both general and high-risk population including HCWs, by looking for a cooperation between regions in which regions more experienced can support regions with the lowest vaccination coverage \[[@R10]\]. Several outbreaks of health care facility-acquired influenza, including immunosuppressed or old patients, have been documented as a consequence of low vaccine effectiveness, thus requiring consequently an indirect protection. Influenza vaccination of all HCWs is recommended by the World Health Organization (WHO), \[[@R11]\] US Centers for Disease Control and Prevention (CDC) \[[@R01]\], and by the Italian Ministry of Health \[[@R06]\] in order to prevent transmission of influenza from HCWs to patients. Nevertheless, the same countries show a low uptake of influenza vaccine in HCWs \[[@R12]\]. A global literature review on vaccination programs in HCWs reported a percentage of vaccination coverage ranging from 2.1% to 82%, with highest uptake rates occurring in USA \[[@R13]\] (from 40% to 87.4%) \[[@R14], [@R15], [@R16], [@R17]\], followed by Germany (26.9%) \[[@R18]\], and Spain (from 14.7% to 38%) \[[@R19]\]. In Italy, data on vaccination coverage among HCWs are not regularly collected and the few ad-hoc studies have shown low coverage rates of 0-29% \[[@R20]\], and 20.8% \[[@R21]\]. In the 2012-2013 season, the coverage rate for the working-age population was 10% \[[@R22]\]. In our survey most of Sardinia hospitals had an influenza coverage rate lower than 6%. Some studies highlighted how compulsory influenza vaccination with forfeit for non-adhesions was associated with larger compliance to vaccination programs \[[@R23]\]. Other authors asserted that voluntary vaccination of HCWs against influenza would represent the most effective strategy \[[@R24]\]. Hospitals that used personal contact approach had higher vaccination rates \[[@R24], [@R25]\]. Differently, newsletters sent by HHM to unit managers, and both nurse and technician coordinators were the most frequent tool used by Sardinia hospitals. Nevertheless, this tool obtained a vaccination coverage lower than 10% in 74% of the involved hospitals. A recent Italian study implemented different actions such as education, promotion, and easy access to staff vaccination in order to increase HCWs vaccination coverage \[[@R26]\]. Despite these efforts, the results showed that the vaccination coverage in the hospital staff was underperforming. On the light of these results, new techniques of information and empowerment on the importance of flu vaccination for HCWs should be considered and activated. In 1997, CDC recommended HBV vaccination for all HCWs \[[@R27]\]; nevertheless, HBV vaccine coverage rate in USA was 63.4%, well below the 90% goal \[[@R28]\]. Similar percentages were reported in other countries (85% for Belgium, 55.4% for India, and 55% for Morocco) \[[@R29], [@R30], [@R31]\]. In 1991, Italy introduced HBV mandatory vaccination for infants and 12-year-old adolescents until 2003 when the two cohorts joined together \[[@R32]\]. In persons who received primary immunization in the first year of life, immune memory for HBsAg lasted for at least 17 years and additional booster doses were not needed to enhance long-term immunization \[[@R33], [@R34], [@R35], [@R36]\]. In Italian HCWs who did not receive HBV vaccination as adolescents and are older than 35-36 years, testing HCWs at hiring time and administering the vaccine to those who prove susceptible to HBV infection is strongly advised. In Italy, the coverage rate of HBV vaccination ranged from 43% to 87% \[[@R37]\]. Thus, we emphasize that efforts should be made to increase the number of vaccinated HCWs for HBV protection. Measles, mumps, rubella, and varicella (MMRV) are highly contagious diseases, and pose a high risk for both HCWs and patients \[[@R02]\]. In Italy, the percentage of immunization in HCWs ranged from 85.7% to 95.1% for varicella, from 47% to 96.8% for rubella, from 71.4% to 97.8% for measles, and from 52.5% to 87.6% for mumps \[[@R38], [@R39]\]. These results show a significant proportion of unprotected workers. In fact, most of the Sardinia hospitals did not perform serological tests for MMVR on HCWs. Furthermore, tuberculosis still remains a severe worldwide health problem, especially among immunocompromised patients. High percentage of latent Mycobacterium tuberculosis infection among HCWs and the increasing number of Mycobacterial strains resistant to the main drug therapies increases the need for prevention programs health care workers. Tuberculin skin test (TST) is still globally recommended to screen HCWs \[[@R40]\] despite its limitations. In effect, the need for a return visit to assess the test effect, often, results in a low compliance to follow-up \[[@R41]\]. These limits could be overtaken by advancement in the in vitro test, which detects the interferon-g production in response to other Mycobacterium tuberculosis antigens. QuantiFERON- in tube test (QFT-in tube) does not need for a second visit to interpret results and it has a higher specificity than TST \[[@R42]\]. An Australian study compared QFT-in tube test to the TST in HCWs. The results demonstrated that positive QFT-in tube test strongly correlated with risk factors for TB exposure more than positive TST, especially in a low tuberculosis prevalence population. Yet, a positive TST strongly correlated with a prior history of BCG vaccination \[[@R43]\]. The main limitation of the study regards the data collection method. Specifically, we have used a nonvalidated questionnaire, which can result in a low level of reliability of the data. Furthermore, we did not examine the vaccination coverage according to the different occupational categories and healthcare settings. This did not allow us to identify specific professionals or healthcare areas that mainly need for specific intervention strategies to increase vaccination coverages. Nevertheless, there are several advantages from this research. The data from the whole Sardinia district have shown in HHM a low level of awareness about HCWs vaccination. Furthermore, we have highlighted some critical aspects that can be considered by the local policy- makers to organize tailored preventive interventions. We, thus, emphasize the strong need to implement information programs addressed to HCWs about vaccinationpreventable diseases for potentially spreading in hospital settings complying with a patient-centered care system. Another key-point of our study is a pragmatic low-budget method to communicate with health managers; this method has allowed us to get a large amount of data in a small lapse of time, thus reducing costs. Finally, our study provides evidence for stimulating health managers of Sardinia about the need to identify HCWs susceptible to threats of some *infectious* diseases damaging the hospital setting. This step is essential to achieve preventive strategies aimed to guarantee the health of both patients and HCWs. The study was supported by Department of Public Health, Clinical and Molecular Medicine -- University of Cagliari (Italy). The authors declare that they have no competing interests. The authors would like to thank Hospital Health Managements and Occupational Physicians from Ospedale Civile Paolo Merlo, La Maddalena (OT), Ospedale Civile Paolo Dettori, Tempio, Ospedale Giovanni Paolo II, Olbia, Ospedale S.S. Annunziata Azienda Ospedaliero Universitaria, Sassari, Ospedale A. Segni and Ospedale Civile, Alghero, San Camillo, Sorgono (NU) Ospedale San Francesco, Nuoro, Ospedale Cesare Zonchello, Bosa (OR), Ospedale G.A. Mastino, Lanusei (OG), Ospedale Nostra Signora delle Mercede, Oristano, Ospedale San Martino, Ghilarza (OR), Ospedale Delogu, Isili (CA), Ospedale San Giuseppe, San Gavino Monreale (VS), Ospedale Nostra Signora di Bonaria, Ospedale CTO and Ospedale Santa Barbara, Iglesias, Ospedale Sirai, Carbonia, Ospedale San Marcellino, Muravera (CA), Ospedale Roberto Binaghi, Ospedale Marino, Ospedale SS. Trinità, Ospedale Businco, Ospedale Microcitemico, Azienda Ospedaliera Brotzu, Azienda Ospedaliero Universitaria di Cagliari, Cagliari. The authors declared that no competing interests exist. Funding: none. [^1]: Authors\' Contributions MC, FA and RCC conceived, designed and coordinated the research. BS, NMM and AL collected data. BS, NMM, AL and MG performed the statistical analyses. MC, FA and RCC evaluated the results. BS, NMM, AL, and MG wrote the manuscript. All Authors revised the manuscript and gave their contribution to improve the paper. All authors read and approved the final manuscript.

INTRODUCTION ============ Membrane depolarization--elicited outward current through inward rectifier K^+^ channels exhibits profound relaxation. Consequently, with identical K^+^ concentrations on both sides of the membrane the steady-state outward macroscopic current at positive membrane voltages is much smaller than the inward current at the corresponding negative voltages, a feature called initially anomalous rectification and subsequently inward rectification ([@bib15]; [@bib28]). The first clue to a possible mechanism came from the work by [@bib3], who showed that intracellular TEA blocks squid voltage--activated K^+^ channels in a strongly voltage-dependent manner, rendering them inwardly rectifying. 20 y later, Mg^2+^ was identified as an endogenous voltage-dependent channel blocker causing inward rectification ([@bib25]; [@bib36]). However, the voltage dependence of channel block by Mg^2+^ alone is too weak to account for the strong inward rectification observed in intact cells. Furthermore, significant rectification remains in the absence of Mg^2+^. Thus arose the concept of intrinsic (voltage-dependent) channel gating to explain Mg^2+^-independent current relaxation and rectification (e.g., [@bib14]; [@bib33]; [@bib35]). Little progress was made in the search for additional endogenous blockers until certain polyamines were found to block the channels in a strongly voltage-dependent manner ([@bib8]; [@bib19]; [@bib6]). Still, variable residual inward rectification remains after the inside of a membrane patch is perfused with solutions nominally devoid of polyamines and Mg^2+^ ([@bib1]; [@bib32]; [@bib17]). This finding underlies the hypothesis that inward rectification results from intrinsic channel gating modulated by the binding of Mg^2+^ and polyamines to a putative channel-gating machinery, rather than from voltage-dependent channel block by these intracellular cations. We found, however, that the residual rectification independent of Mg^2+^ and polyamines is related to HEPES in recording solutions ([@bib10]). Voltage jump--induced, time-dependent relaxation is not exclusively associated with outward currents, since hyperpolarization-induced inward currents also exhibit relaxation ([@bib16]). [@bib5] showed that hyperpolarization reduces the open probability of IRK1, and argued that the reduction in channel open probability results from both channel block by extracellular divalent cations and channel gating. Similar phenomena occur to a lesser extent in ROMK2 ([@bib4]). Using IRK1-ROMK2 chimeras, [@bib5] found that the protein segments that form the ion conduction pore underlie the putative channel gating. Consistent with this, [@bib20] reported that gating of IRK1 at negative voltages is significantly perturbed when ester carbonyls replace the amide carbonyls of the two glycine residues within the signature sequence that forms the ion selectivity filter. Also, [@bib31] showed that in low K^+^ solutions in the absence of extracellular divalent cations, hyperpolarization induces a significant inward current relaxation which is likened to the C-type inactivation of voltage-activated *Shaker* K^+^ channels ([@bib13]; [@bib18]; [@bib41]). To resolve the fundamental issue whether the macroscopic conductance of IRK1 has any significant intrinsic voltage dependence, we present here a systematic experimental investigation of the causes underlying voltage jump--induced current relaxations. Our study also helps define the optimal experimental conditions for studying IRK1. MATERIALS AND METHODS ===================== Molecular Biology and Oocyte Preparation ---------------------------------------- IRK1 cDNA ([@bib16]) was cloned into the pGEM-Hess plasmid. RNA was synthesized using T7 polymerase (Promega) from Nhe1-linearized cDNAs. Oocytes harvested from *Xenopus laevis* (*Xenopus* One) were incubated in a solution containing NaCl, 82.5 mM; KCl, 2.5 mM; MgCl~2~, 1.0 mM; HEPES (pH 7.6), 5.0 mM; and collagenase, 2--4 mg/ml. The oocyte preparation was agitated at 80 rpm for 60--90 min. It was then rinsed thoroughly and stored in a solution containing NaCl, 96 mM; KCl, 2.5 mM; CaCl~2~, 1.8 mM; MgCl~2~, 1.0 mM; HEPES (pH 7.6), 5 mM; and gentamicin, 50 μg/ml. Defolliculated oocytes were selected and injected with RNA at least 2 and 16 h, respectively, after collagenase treatment. All oocytes were stored at 18°C. Patch Recording --------------- IRK1 currents were recorded from inside-out membrane patches of *Xenopus* oocytes (injected with IRK1 cRNA) with an Axopatch 200B amplifier (Axon Instruments, Inc.), filtered at 5 kHz, and sampled at 25 kHz using an analogue-to-digital converter (DigiData 1200; Axon Instruments, Inc.) interfaced with a personal computer. pClamp6 software (Axon Instruments, Inc.) was used to control the amplifier and acquire the data. During current recording, the voltage across the membrane patch was first hyperpolarized from the 0 mV holding potential to −100 mV and then stepped to various test voltages, or stepped directly from the holding potential. The duration of the voltage test pulse was 100 ms, which is comparable to those used in the studies where the Mg^2+^- and polyamine-independent rectification was initially observed ([@bib1]; [@bib32]). Background leak current correction was performed as described previously ([@bib21]; [@bib10]). To effectively perfuse the patch, the tip of the patch pipette (∼3 MΩ) was immersed in a stream of intracellular solution exiting one of ten glass capillaries (ID = 0.2 mm) mounted in parallel. Recording Solutions ------------------- All recording solutions contained 100 mM K^+^ contributed by KCl, K~2~EDTA, K~2~HPO~4~, KH~2~PO~4~, and KOH. The phosphate-buffered intracellular solution contained (mM): 5 K~2~EDTA (unless specified otherwise), 10 "K~2~HPO~4~ + KH~2~PO~4~" in a ratio yielding the desired pH, and sufficient KCl to bring total K^+^ concentration to 100 mM ([@bib10]). To adjust pH to 8.0 and above a small amount of KOH was used. The HEPES-buffered intracellular solution, used in [Fig. 3](#fig3){ref-type="fig"} F[igure]{.smallcaps} 3.Other causes underlying apparent inward rectification. (A--D) Currents were recorded from the same inside-out patch with the voltage pulse protocol shown in [Fig. 1](#fig1){ref-type="fig"}. The intracellular solution composition (besides KCl) and pH for each panel are indicated. (E) Normalized I-V curves constructed from the currents determined at the end of each test voltage-pulse. The I-V curves a--d correspond to the currents from A--D, respectively. All data points are mean ± SEM (*n* = 5)., contained (mM): 100 K^+^ (Cl^−^ + OH^−^), 5 EGTA and 10 HEPES, pH 7.2 (adjusted with KOH). The HEPES-buffered extracellular solution contained (mM): 100 K^+^ (Cl^−^ + OH^−^), 0.3 CaCl~2~, 1.0 MgCl~2~, and 10 HEPES, pH 7.6 (adjusted with KOH). In the phosphate-buffered extracellular solution, HEPES was replaced by an equal concentration of "K~2~HPO~4~ + KH~2~PO~4~" in a ratio yielding pH 7.6. The divalent cation-free extracellular solution contained 5 mM EDTA. All chemicals were purchased from Fluka Chemical Corp. RESULTS ======= We showed previously that intracellular "HEPES" blocks IRK1 channels with varying potency depending on the commercial sources ([@bib10]). As shown there and in [Fig. 1](#fig1){ref-type="fig"} F[igure]{.smallcaps} 1.Comparisons of the effects of intracellular HEPES, HEP, and piperazine on IRK1 currents (structures shown at top). (A--C) Currents were recorded with 10 mM HEPES, 3 μM HEP and 0.3 μM piperazine, respectively, from the same inside-out patch with the voltage pulse protocol shown below. In all cases the intracellular solutions contained 100 mM K^+^, 5 mM EDTA, and 10 mM phosphate (pH 7.6) besides the tested chemicals, whereas the extracellular solution contained 100 mM K^+^, 0.3 mM Ca^2+^, 1 mM Mg^2+^, 10 mM phosphate (pH 7.6). The currents are corrected for background current. Dotted lines identify the zero current level. (D--F) Normalized I-V curves, corresponding to A--C, which were constructed from the currents determined at the end of each test voltage-pulse. Current at each voltage is normalized (except for its signs) to that at −100 mV. Each data point represents the mean (± SEM) of currents recorded from 5--7 patches. A, the blocking kinetics are slow even with 10 mM HEPES present. These findings indicate that the block is caused mainly by some impurity in HEPES. Usually, HEPES is synthesized by reacting hydroxyethylpiperazine (HEP)[\*](#fn1){ref-type="fn"} with bromoethanesulfonate ([@bib9]). [Fig. 1](#fig1){ref-type="fig"} B shows the current records in the presence of 3 μM HEP; the I-V curves determined at the end of the voltage pulses in the presence of 10 mM HEPES and 3 μM HEP are shown in [Fig. 1, D and E](#fig1){ref-type="fig"}, respectively. HEPES and HEP block the channels in a qualitatively comparable manner, although HEP is much more potent. The blocking activity of HEP must come from the piperazine ring since piperazine itself is even more potent ([Fig. 1, B](#fig1){ref-type="fig"} versus C, and E versus F). [Fig. 2](#fig2){ref-type="fig"} F[igure]{.smallcaps} 2.HEPES, HEP, and piperazine concentration dependence of channel block. The fraction of unblocked currents in the presence of three representative concentrations of HEPES (A), HEP (B), or piperazine (C) is plotted against membrane voltage. The theoretical curves are fits of the Woodhull equation, which give K~d~(0 mV) = 2.47 ± 0.02 M (mean ± SEM, *n* = 6) and Z = 1.02 ± 0.02 for HEPES, K~d~(0 mV) = 2.97 ± 0.16 mM (*n* = 6) and Z = 1.09 ± 0.03 for HEP, and K~d~(0 mV) = 0.27 ± 0.04 mM (*n* = 8) and Z = 1.08 ± 0.02 for piperazine. shows the fraction of unblocked currents in the presence of three representative concentrations of HEPES, HEP, or piperazine. Analyzing the data in [Fig. 2](#fig2){ref-type="fig"} with the [@bib37] equation gives an apparent K~d~(0 mV) = 2.47 M and Z (valence) = 1.02 for HEPES, K~d~(0 mV) = 2.97 mM and Z = 1.09 for HEP, and K~d~(0 mV) = 0.27 mM and Z = 1.08 for piperazine. Therefore, the apparent channel block by HEPES can be accounted for by a trace amount of HEP, which is well below the limit of impurity specified by the supplier (0.5%). To our surprise, we found in the experiment shown below that HEPES is not the sole source of contaminating blockers in commonly used intracellular solutions. [Fig. 3](#fig3){ref-type="fig"} shows that IRK1 channels exhibit inward rectification in the presence of an intracellular solution like that used by [@bib1], which contained KCl, EGTA and HEPES (pH 7.2). The rectification was reduced only modestly when HEPES (10 mM) was replaced by phosphate (10 mM), but reduce significantly further when we also substituted EDTA (5 mM) for EGTA (5 mM), although the I-V curve did not approach linearity until solution pH was raised from 7.2 to 7.6 \[the final composition is the one we used previously ([@bib10])\]. These findings show that the persisting rectification after removal of endogenous blockers ([@bib1]) was caused by the use of HEPES, EGTA, and low pH. Intracellular EGTA, HEPES (each up to 10 mM), and pH \< 7.4 have been widely used in studies of IRK1. [Fig. 4](#fig4){ref-type="fig"} F[igure]{.smallcaps} 4.IRK1 currents in the presence of three metal ion chelators. (A) Current records collected from the same patch in the presence of intracellular EDTA, EGTA, or CDTA, each at 5 mM. Phosphate was used to buffer pH. (B) Normalized I-V curves in the presence of the metal ion chelators. All data points are mean ± SEM (*n* = 5). compares the effects of intracellular metal ion chelators EDTA, CDTA, or EGTA (each at 5 mM; pH 7.6) on the currents and the I-V curves. Inward rectification is more pronounced with either EGTA or CDTA than with EDTA, even though the affinity of CDTA for a given di- or trivalent cation is much higher than that of EDTA. Since EDTA causes the least rectification and also is nonselective among divalent cations, EDTA instead of EGTA should be used to scavenge contaminating blocking metal ions. To find the intracellular EDTA concentration that yields the most linear I-V curve (at pH 7.6, see below), we obtained current records from the same patch in the presence of five different concentrations of EDTA ([Fig. 5](#fig5){ref-type="fig"}) F[igure]{.smallcaps} 5.Effects of EDTA concentration on IRK1 currents. All traces were recorded from the same patch, with intracellular EDTA concentrations as indicated. The corresponding I-V curves are shown in [Fig. 6](#fig6){ref-type="fig"}. Solution pH was buffered with phosphate.. With 0.1 mM EDTA, outward currents displayed modest inward rectification. As the chelator concentration was raised to 5 mM the rectification is nearly vanished, most probably because the concentration of contaminating free blocking metal ions was reduced. However, when EDTA concentration was increased further, rectification became again more pronounced, presumably reflecting the effect of some impurity in EDTA. For clarity, I-V curves for 0.1--5 mM and for 5--30 mM EDTA are plotted separately in [Fig. 6](#fig6){ref-type="fig"} F[igure]{.smallcaps} 6.Effects of EDTA concentration on the I-V curves of IRK1 channels. Normalized I-V curves in the presence of various concentrations of EDTA. For clarity, I-V curves with 0.1--5 mM intracellular EDTA (A) are plotted separately from those with 5--30 mM EDTA (B). All data points are mean ± SEM (*n* = 4--6). (C) Normalized current at 80 mV, taken from the I-V curves in A and B, is plotted against the concentration of EDTA. The data represented by the circles (mean ± SEM) were determined experimentally, whereas those by triangles were calculated using [Eq. 1](#eqn1){ref-type="disp-formula"}, as described in [discussion]{.smallcaps}. A, and B, showing that 5 mM EDTA is optimal for obtaining a linear macroscopic I-V curve. The normalized current (mean ± SEM) at 80 mV, taken from the I-V curves in A and B, is plotted against the concentration of EDTA ([Fig. 6](#fig6){ref-type="fig"} C, circles). [Fig. 7](#fig7){ref-type="fig"} F[igure]{.smallcaps} 7.Comparisons of channel block by "EDTA" with block by Mg^2+^ or ethylenediamine. (A and C) Currents with 0.1 mM and 30 mM EDTA normalized to that with 5 mM EDTA are plotted against membrane voltage, respectively. (B and D) The fraction of unblocked currents in the presence of Mg^2+^ or ethylenediamine (ED), respectively, is plotted against membrane voltage. All data points are mean ± SEM (*n* = 5). All curves are fits of the equation I/I~o~ = 1/(1 + \[blocker\]/K~d~), where K~d~ = K~d~(0 mV)e^−ZFV/RT^. For A and B, the two nearly superimposed curves for each data set are fits either all data points (continuous curves) or all but the rightmost three data points (dashed curves). For C or D, the dashed curve through each dataset is a fit to all but the rightmost three data points. plots the fraction of unblocked currents in the presence of a low (0.1 mM; [Fig. 7](#fig7){ref-type="fig"} A) or a high (30 mM; [Fig. 7](#fig7){ref-type="fig"} C) concentration of EDTA against membrane voltage. The two blocking curves differ in character. At 0.1 mM EDTA, it is well described by the [@bib37] equation ([Fig. 7](#fig7){ref-type="fig"} A), consistent with that the channels were blocked by a "nonpermeant" blocker such as a divalent cation, an example of which (block by Mg^2+^) is shown in [Fig. 7](#fig7){ref-type="fig"} B. In contrast, at 30 mM EDTA the blocking curve deviates from the Woodhull equation ([Fig. 7](#fig7){ref-type="fig"} C), consistent with that the channels were blocked by a permeant blocker such as ethylenediamine (ED, [Fig. 7](#fig7){ref-type="fig"} D; see [@bib11] for details). [Fig. 8, A and B](#fig8){ref-type="fig"} F[igure]{.smallcaps} 8.Comparisons of the effects on IRK1 currents of intracellular EDTA and ethylenediamine (structures shown at top). (A and B) Currents were recorded from the same membrane patch with, respectively, 30 mM EDTA and 0.1 μM ethylenediamine. (C and D) Normalized I-V curves; each data point represents the mean (± SEM; *n* = 5) of currents., shows the current records with 30 mM EDTA and 0.1 μM ethylenediamine, respectively; the corresponding I-V curves are shown in [Fig. 8, C and D](#fig8){ref-type="fig"}. EDTA and ethylenediamine block the channels in a qualitatively comparable manner, although the latter is \>10,000-fold more potent. Therefore, the blocking effect associated with EDTA can be accounted for by trace of contaminating ethylenediamine used to synthesize EDTA. All subsequent data were recorded with intracellular solutions containing 5 mM EDTA and 10 mM phosphate (without HEPES). To determine the optimal intracellular pH we examined how pH affects IRK1 currents ([Fig. 9](#fig9){ref-type="fig"} F[igure]{.smallcaps} 9.Effects of intracellular pH on the currents of wild-type and D172N mutant IRK1 channels. Currents were recorded at various intracellular pH, with extracellular pH = 7.6 throughout. For each channel type, all currents were obtained from the same inside-out patch. A). The average I-V curve for each pH examined is plotted in [Fig. 10](#fig10){ref-type="fig"} F[igure]{.smallcaps} 10.Effects of intracellular pH on the I-V curves of wild-type and mutant IRK1 channels. (A and B) I-V curves of wild-type and D172N mutant IRK1 channels at various intracellular pH, determined from the current records, as shown (and including those) in [Fig. 9](#fig9){ref-type="fig"}, except for those at pH 7.6, which were taken from those as shown in [Fig. 4](#fig4){ref-type="fig"} A. (C and D) Currents through IRK1 and D172N channels, normalized to those at pH 8.5, are plotted against membrane voltage. All data points are mean ± SEM (*n* = 5). A. The I-V curve is linear at pH 8.5. Lowering pH inhibited IRK1 currents minimally between 8.5 and 7.5 but dramatically between 7.5 and 6.5. The fraction of current not inhibited by protons is plotted in [Fig. 10](#fig10){ref-type="fig"} C, which shows that channel inhibition has voltage-dependent and -independent components. Interestingly, much of the voltage-dependent component of proton inhibition vanished when acidic D172 (located in the inner pore) was replaced by neutral asparagine ([Figs. 9](#fig9){ref-type="fig"} B, and 10, B and D). At higher intracellular pH the outward D172N current is slightly larger than the inward current. As a practical matter, very small membrane patches must be used to achieve adequate removal of endogenous blockers by perfusion. To illustrate this, [Fig. 11](#fig11){ref-type="fig"} F[igure]{.smallcaps} 11.Variable degree and rate of removal of endogenous blockers by perfusion. Current traces shown in A and B were recorded from two separate patches excised from two oocytes injected with different amounts of cRNA (higher in A than in B). The recordings were made at the indicated times following the start of perfusion. A shows three consecutive sets of current records from a small membrane patch, taken at 1-min intervals after the start of perfusion. Any endogenous blockers present in that excised patch were apparently effectively removed within the first minute. In contrast, for a much larger patch (from a separate oocyte injected with much less RNA), removal of endogenous blockers was very slow and incomplete even after prolonged perfusion ([Fig. 11](#fig11){ref-type="fig"} B). Under certain commonly used experimental conditions membrane hyperpolarization also induces relaxation of inward IRK1 currents. The currents in [Fig. 12](#fig12){ref-type="fig"} F[igure]{.smallcaps} 12.Relaxation of inward currents caused by extracellular divalent cations. Currents were recorded with the voltage protocol shown at the top. Intra- and extracellular solutions were buffered with 10 mM phosphate. The extracellular solution contained 0.3 mM Ca^2+^ and 1 mM Mg^2+^ in A, but 5 mM EDTA and no added Ca^2+^ or Mg^2+^ in B. (C) The ratio of currents at the end and the beginning of voltage pulses (I~end~/I~bgn~) is plotted against membrane voltage. The data corresponding to A and B are labeled by letters a and b, respectively. All data points are mean ± SEM (*n* = 5). A (and all those above) were recorded with 0.3 mM Ca^2+^ and 1.0 mM Mg^2+^ present in the extracellular solution buffered with phosphate. The inward currents at very negative voltages relax noticeably. This relaxation largely vanished ([Fig. 12](#fig12){ref-type="fig"} B) in the absence of extracellular divalent cations. The voltage dependence of the ratio of currents at the end and the beginning of voltage pulses (I~end~/I~bgn~) is shown in [Fig. 12](#fig12){ref-type="fig"} C. These results are consistent with the previous finding by [@bib5] that extracellular divalent cations block IRK1 in a voltage-dependent manner. Furthermore, [@bib31] observed relaxation of inward current even in the nominal absence of extracellular divalent cations. The relaxation was prominent only at low concentrations of K^+^. Since solutions in the quoted study were HEPES buffered, we wondered whether the current relaxation was due to the use of extracellular HEPES. Confirming the results of [@bib31], we found that the inward current after strong membrane hyperpolarization exhibits little or no relaxation when the extracellular solution contained 100 mM K^+^, 10 mM HEPES, and 5 mM EDTA, but no added divalent cations ([Fig. 13](#fig13){ref-type="fig"} F[igure]{.smallcaps} 13.K^+^ dependence of the inward current relaxation in the presence of extracellular HEPES. All currents were elicited with the voltage protocol shown in [Fig. 12](#fig12){ref-type="fig"}, with 5 mM EDTA and no added divalent cations present in the extracellular solution. Extracellular solutions were buffered with 10 mM HEPES (A and B) or phosphate (C and D), and contained either 100 mM K^+^ (A and C) or 20 mM K^+^ (B and D). (E) The ratio of currents at the end and the beginning of voltage pulses (I~end~/I~bgn~) is plotted against membrane voltage. The data corresponding to A--D are labeled a--d, respectively. All data points are mean ± SEM (*n* = 5). A). Replacing HEPES with phosphate had no noticeable effect ([Fig. 13](#fig13){ref-type="fig"} C). Also confirming Shieh\'s finding, lowering the K^+^ concentration to 20 mM revealed prominent inward current relaxation ([Fig. 13](#fig13){ref-type="fig"} B). However, this relaxation vanished when we replaced HEPES with phosphate ([Fig. 13](#fig13){ref-type="fig"} D). The voltage dependence of the ratio of currents at the end and the beginning of voltage pulses (I~end~/I~bgn~) is shown in [Fig. 13](#fig13){ref-type="fig"} E. Furthermore, we examined whether channel block associated with extracellular HEPES could also be caused by residual contaminants. As shown in [Fig. 14](#fig14){ref-type="fig"} F[igure]{.smallcaps} 14.K^+^ dependence of inward current relaxation in the presence of extracellular HEP or piperazine. Intra- and extracellular solutions were buffered with 10 mM phosphate. The extracellular solution contained 3 μM HEP (A and B) or 0.3 μM piperazine (C and D), and either 100 mM K^+^ (A and C) or 20 mM K^+^ (B and D). (E) The ratio of currents at the end and the beginning of voltage pulses (I~end~/I~bgn~) is plotted against membrane voltage. The data corresponding to A--D are labeled a--d, respectively. All data points are mean ± SEM (*n* = 5)., like "HEPES", HEP and piperazine block the channels in the presence of 20 mM but not 100 mM extracellular K^+^. Therefore, channel block associated with extracellular HEPES can also be accounted for by a trace amount of contaminating HEP. DISCUSSION ========== In the present study, we systematically investigated the causes underlying the reported IRK1 current relaxations after step changes of membrane voltage, and found that none of them is intrinsic to the channels. In other words, within the usual voltage range the macroscopic IRK1 conductance has no significant intrinsic voltage dependence that causes either inward rectification or inactivation. The reported apparent voltage dependence of macroscopic IRK1 currents is caused by traces of contaminants in routinely used chemicals, such as HEPES and metal ion chelators among which (the latter) EDTA has the least effect. Specifically, block associated with both intracellular and extracellular HEPES can be accounted for by residual HEP (∼500 part per million (ppm) in weight), and that associated with intracellular EDTA by residual ethylenediamine (∼1 ppm). These contamination levels, estimated from the relative specific inhibitory activity of compared chemicals ([Figs. 2](#fig2){ref-type="fig"} and [8](#fig8){ref-type="fig"}), are well below the limits (5,000 ppm) specified by the supplier. Intracellular divalent chelators must be used both to suppress endogenous Ca^2+^-activated Cl^−^ currents and to minimize channel block by contaminating divalent cations. However, a divalent cation chelator such as EDTA has two opposing effects: (a) it reduces free divalent cation concentration and thereby relieves channel block by these ions, and (b) it contains residual ethylenediamine that blocks the channels. Consequently, the current is a biphasic function of EDTA concentration with a maximum at an intermediate concentration of EDTA ([Fig. 6](#fig6){ref-type="fig"}), which may vary somewhat depending on the extent of contamination in a given lot of EDTA and on the level of divalent cation concentration in the solution. Under the present solution conditions, the maximal current corresponding to 5 mM EDTA where there is little time-dependent relaxation of outward current during a 100-ms test-pulse to positive voltage (even up to 100 mV; [Fig. 5](#fig5){ref-type="fig"}), which is sufficiently long for many common studies. In the solution with 0.1 mM EDTA the estimated concentration of contaminating ethylenediamine is ∼0.3 nM ([Fig. 8](#fig8){ref-type="fig"}), which practically causes no channel block, given that the ^ED^K~d~(0 mV) = 1.4 mM and ^ED^Z = 2 ([@bib11]). Therefore, at this low concentration of EDTA the channels are blocked primarily by contaminating metal ions ([Fig. 7](#fig7){ref-type="fig"}). The common divalent cations, Ca^2+^ and Mg^2+^, block inward rectifiers with the same mechanism, although the latter binds with somewhat higher affinity ([@bib23]; [@bib24]). Examining block by intracellular Ca^2+^ is technically more difficult due to the presence of Ca^2+^-activated Cl^−^ currents in oocytes; thus, we estimated the total concentration of contaminating divalent cations in the equivalent of Mg^2+^. To do so, we first determined that ^Mg^K~d~ (0 mV) = 17 μM and ^Mg^Z = 1.1 from the fit of the Woodhull equation to the Mg^2+^-inhibition curve ([Fig. 7](#fig7){ref-type="fig"} B). Then, based on these values, we found, from the same analysis of current inhibition in 0.1 mM EDTA, that the extent of channel block is equivalent to that caused by 68 nM free Mg^2+^ ([Fig. 7](#fig7){ref-type="fig"} A). This free Mg^2+^ concentration requires 20 μM total Mg^2+^ in a 0.1 mM EDTA solution while the EDTA-Mg^2+^ stability constant is 3.5 × 10^6^ M^−1^. Theoretically, 20 μM Mg^2+^ in a 30 mM EDTA solution results in 0.2 nM free Mg^2+^, which would practically cause no block. Thus, at this high EDTA concentration the channels are primarily blocked by contaminating ethylenediamine (0.1 μM; [Fig. 8](#fig8){ref-type="fig"}). With these estimates, we calculated for each EDTA concentration the current at 80 mV, using the following equation, $$\documentclass[10pt]{article} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{pmc} \usepackage[Euler]{upgreek} \pagestyle{empty} \oddsidemargin -1.0in \begin{document} \begin{equation*}\frac{I}{I_{O}}=\frac{1}{1+\displaystyle\frac{\left[Mg^{2+}\right]_{free}}{^{Mg}K_{d}\left(0\;mV\right)e^{-\displaystyle\frac{^{Mg}ZFV}{RT}}}+\displaystyle\frac{\left[ED\right]}{^{ED}K_{d}\left(0\;mV\right)e^{-\displaystyle\frac{^{ED}ZFV}{RT}}}}{\mathrm{,}}\end{equation*}\end{document}$$where quantities F, V, R, and T have their usual meaning. Since the current at 80 mV does not significantly deviate from the fit of the Woodhull equation ([Fig. 7, C and D](#fig7){ref-type="fig"}), ethylenediamine is assumed, for simplicity, to be nonpermeant. The calculated values with the equation agree well with those experimentally observed ([Fig. 6](#fig6){ref-type="fig"} C). As in the case of other inward rectifiers, lowering intracellular pH significantly inhibits IRK1 ([@bib32]). [Fig. 10, A and C](#fig10){ref-type="fig"}, shows that inhibition by protons has both voltage-dependent and -independent components. Several residues underlying voltage-independent inhibition by intracellular protons have been identified in various inward rectifiers (e.g., [@bib7]; [@bib38]). On one hand, if D172 in the pore of IRK1 acts as a surface charge and if protonation of D172 is voltage dependent, protonation of D172 (D^−^ to D) may reduce the single channel conductance and thereby render the channels inwardly rectifying. For the following reasons, this may not, however, be the primary cause underlying the observed voltage-dependent channel inhibition at low pH. Replacing D172 with neutral asparagine does not reduce the single channel conductance ([@bib29]). Furthermore, it renders neither the single channel i-V curve nor the macroscopic I-V curve inwardly rectifying ([@bib29]; [@bib10]; [Fig. 10](#fig10){ref-type="fig"} B). On the other hand, voltage-dependent inhibition of IRK1 by intracellular protons probably primarily reflects protonation of amine groups in the residual endogenous and contaminating exogenous organic blockers and/or EDTA. Protonation of these blockers enhances their affinity for IRK1 ([@bib11]), whereas protonation of EDTA reduces its affinity for trace divalent cations, thus causing further channel block. Consistent with this reasoning, mutant D172N channels, whose affinity for intracellular blocking cations is dramatically reduced ([@bib19]; [@bib8]; [@bib6]; [@bib39]), exhibit dramatically reduced voltage-dependent inhibition at low pH ([Fig. 10](#fig10){ref-type="fig"} D). In any case, for practical purposes one can obtain essentially uninhibited IRK1 currents at intracellular pH 7.6 or higher. The affinity of IRK1 channels for some endogenous blockers is exceedingly high, so that even trace amounts of endogenous blocker may cause significant inward rectification. For example, the K~d~(0 mV) for the binding of fully protonated spermine is ∼10^−7^ M, whereas the effective valence of channel block by spermine is ∼5 ([@bib10],[@bib11]). The calculated K~d~(100 mV) for spermine binding is therefore ∼10^−15^ M, which is almost certainly below the concentration of spermine remaining in an exhaustively perfused membrane patch. Therefore, although spermine is a permeant blocker whose effect can be somewhat relieved by membrane depolarization, at 100 mV in the steady-state most channels will be blocked by spermine at concentrations as low as 1 nM (typical oocyte concentrations are submillimolar; [@bib30]). Assuming the rate constant for spermine binding at 100 mV is diffusion limited and as high as estimated for quaternary ammoniums (10^8^--10^9^ M^−1^ s^−1^; [@bib12]), the predicted current reduction caused by 1 nM spermine is 1--10% at the end of a 100-ms voltage pulse to 100 mV (full steady-state inhibition would require many seconds). Consequently, to limit the extent of channel block by spermine to at most a few percent during a 100-ms pulse, spermine concentration may need to be reduced to ≤1 nM. Even if the precise values of K~d~ and k~on~ (at 100 mV) for spermine are unknown, the above exercise illustrates the practical challenge posed by the need to lower spermine concentration to a level that will leave channel currents essentially unaffected. Not surprisingly, in cases where endogenous blockers cannot be adequately removed despite exhaustive perfusion, significant voltage-dependent channel inhibition persists ([Fig. 11](#fig11){ref-type="fig"}). The problem of residual high-affinity inhibitors can be dramatically relieved, or even practically eliminated, by lowering channel affinity for intracellular cations. For example, a linear I-V curve is readily obtained in IRK1 channels containing the D172N mutation ([Figs. 9](#fig9){ref-type="fig"} and [10](#fig10){ref-type="fig"}; [@bib10]), which significantly lowers their affinity for intracellular spermine (e.g., [@bib39]). Also as expected, we obtained satisfactory removal of endogenous blockers only with very small patches ([Fig. 11](#fig11){ref-type="fig"}). As stated in [materials]{.smallcaps} [and]{.smallcaps} [methods]{.smallcaps}, to more effectively perfuse the patch we positioned the tip of the patch pipette in a rapid stream of the intracellular solution instead of perfusing the entire recording chamber. We also kept oocytes away from the recorded patch, since they are known to release substantial amounts of polyamines ([@bib8]; [@bib19]). Relaxation of inward IRK1 current induced by hyperpolarization has also been observed. [@bib5] found, and we confirmed here ([Fig. 12](#fig12){ref-type="fig"}), that some inward current relaxation results from channel block by divalent cations in the extracellular solution. Furthermore, [@bib31] showed that, in the absence of extracellular divalent cations but with HEPES present, lowering K^+^ concentration reveals profound current relaxation after strong membrane hyperpolarization. Based on this finding, the author suggested that the current relaxation resembles C-type inactivation of voltage-activated *Shaker* K^+^ channels, which is similarly "protected" by K^+^. However, we found that this K^+^-sensitive inward current relaxation can be also accounted for by residual HEP in the HEPES used to buffer extracellular pH ([Figs. 13](#fig13){ref-type="fig"} and [14](#fig14){ref-type="fig"}). On the basis of these findings, we argue that the K^+^-sensitive current relaxation is also not an intrinsic gating property of these channels. The dramatic channel block induced by lowering K^+^ on both sides of the membrane probably results from both a reduced competition of extracellular K^+^ with extracellular blocking ions and a reduced "knock off" effect of the blocking ions by intracellular K^+^ ([@bib3]; [@bib2]; [@bib40]; [@bib22]; [@bib26],[@bib27]; [@bib34]). In summary, at the macroscopic level IRK1 channels inherently have practically ohmic characteristics although in principle, the I-V curve may exhibit slight outward rectification in the complete absence of any endogenous or exogenous blockers. In intact cells, the observed inward rectification of the I-V curve results from voltage-dependent block by intracellular cations such as Mg^2+^ and polyamines ([@bib25]; [@bib36]; [@bib8]; [@bib19]; [@bib6]). However, in excised membrane patches, perfused with solutions nominally devoid of Mg^2+^ and polyamines, the relaxation of both inward and outward IRK1 currents induced by voltage jumps and the resulting nonlinearity of the I-V curve is a reflection, not of intrinsic gating properties of IRK1 channels, but of the unusually high affinity of IRK1 for cations. Because of the extraordinarily high affinity for cations, traces of---usually insignificant---contaminants in commonly used organic pH buffers and metal ion chelators become highly significant and problematic in the study of IRK1 channels. Despite this, practically uninhibited (inward and outward) IRK1 currents and therefore linear I-V curves can be obtained, provided that the recorded membrane patch is adequately perfused and that both intracellular and extracellular solutions contain 100 mM KCl, 5 mM EDTA, and 10 mM phosphate at pH 7.6 or above. We thank L.Y. Jan for IRK1 cDNA, J. Yang for the IRK1-pGEMHess construct, and P. De Weer for critical review of our manuscript. This study was supported by National Institutes of Health grant GM55560. Z. Lu is a recipient of an Independent Scientist Award from National Institutes of Health (HL03814). *Abbreviations used in this paper:* HEP, hydroxyethylpiperazine; ppm, part per million.

{.173} {.174} {.175} {.176} {.177}

All relevant data are within the paper and its Supporting Information files. Introduction {#sec001} ============ Antibiotic use practices, which impact antimicrobial resistance (AMR) worldwide, can vary among individuals, populations and regions \[[@pone.0185827.ref001],[@pone.0185827.ref002]\]. This variation is governed by multiple factors including access to- and quality of antibiotics \[[@pone.0185827.ref003]\], prevalence and types of diseases in a population, nature and quality of health services \[[@pone.0185827.ref001]\], level of public health education \[[@pone.0185827.ref004]\] and the political landscape of a region \[[@pone.0185827.ref002]\]. Despite this complexity, unregulated antibiotic use is the most commonly cited factor contributing to AMR, particularly in low- and middle-income countries (LMICs) \[[@pone.0185827.ref005]\]. Successful control of AMR depends, in part, on understanding the human factors that underlie the use of antibiotics. Knowledge, attitude and practice (KAP) surveys commonly aim to identify factors that influence behaviors, and can form the basis for population-specific interventions \[[@pone.0185827.ref006]\]. Unfortunately, few KAP surveys on antibiotic use and related factors have been reported in sub-Saharan Africa where disease burdens can be high and consequently elevate the demand for antibiotics. Even where these surveys are conducted, informal settlements (slums) have rarely been the focus. Informal settlements are complex areas that are characterized by high population densities, poor housing and sanitation infrastructure, high disease burden and limited healthcare facilities \[[@pone.0185827.ref007]--[@pone.0185827.ref009]\]. Often, they present a mix of factors that can perpetuate the generation and transmission of resistant bacteria. With the global population residing in informal settlements projected to reach two billion within the next thirty years \[[@pone.0185827.ref010]\], the importance of AMR intervention efforts in these communities should be emphasized. The goal of this study was to assess knowledge about antibiotics and antibiotic use practices among people living within an urban informal settlement in Kenya. Surveys were carried out at the beginning and end of a five-month longitudinal observational study focused on antimicrobial resistance. As a secondary objective, we evaluated whether "unintended learning" about antibiotics and their use had occurred among the surveyed respondents, owing to their voluntary participation in the longitudinal study. Methods {#sec002} ======= Study population {#sec003} ---------------- This study was based in Kibera; a large, densely-populated informal settlement situated within Nairobi, a city of \>3 million inhabitants \[[@pone.0185827.ref008]\]. Surveys were conducted in two "villages," Soweto and Gatwekera, which have been the site for a population-based infectious disease surveillance (PBIDS) \[[@pone.0185827.ref011]\] carried out by the Kenya Medical Research Institute (KEMRI) and the Centers for Disease Control and Prevention (CDC). PBIDS participants receive free health services for acute illnesses at the Tabitha health clinic, which is located within the study area. Soweto and Gatwekera have a population density of approximately 77,000 people/km^2^ and are characterized by poor sanitation, unregulated water supply systems and outdoor food vending with inconsistent and minimal hygiene \[[@pone.0185827.ref012]\]. Diarrhea, respiratory and febrile illnesses are prevalent in this area \[[@pone.0185827.ref011]--[@pone.0185827.ref014]\]. As with many parts of Nairobi where access to antibiotics is largely uncontrolled \[[@pone.0185827.ref015]\], a variety of pharmacies--both licensed and unlicensed--serve these villages. Survey design {#sec004} ------------- Two surveys were conducted; at the beginning ('entry', September 2015) and end of a longitudinal study ('exit', between Dec 2015 and Jan 2016). From among 5,320 households participating in PBIDS at the time of this study, 200 randomly selected households consented to participate in the longitudinal study on antimicrobial resistance, out of 217 households that were approached to participate. A sample size of 200 provides a 95% confidence interval of 7% for any given parameter estimate. For each selected household, a single adult representative (\>18 years) with knowledge of household healthcare practices was invited to enroll into the study and participated in both surveys. Household demographic data, including the age of the respondent, household size and structure, and education levels of the male and female heads of households, were collected during enrollment. Surveys were administered by community interviewers who were trained to recognize commonly-used antibiotics and to administer the structured questionnaire used for this study. This questionnaire ([S1 File](#pone.0185827.s001){ref-type="supplementary-material"}) addressed three broad topics: a) knowledge of antibiotics, b) sources of antibiotics and reasons for taking antibiotics, and c) sources of information about antibiotics and the impact of this information on antibiotic use attitudes and practices. The absence of a locally-recognizable Swahili term for the word "antibiotic" necessitated a creative approach to conducting the surveys. Interviews were initiated by asking respondents to name up to three medications--excluding pain-relievers, traditional herbs and anti-malarial drugs--that they had used in the past or about which they were conversant. Probing was done until up to three antibiotics were mentioned. Respondents who were unable to recall any antibiotic were not interviewed further. Respondents who mentioned at least one antibiotic were probed with subsequent true-or-false questions ([Table 1](#pone.0185827.t001){ref-type="table"}) that referenced, by name, the first antibiotic mentioned by the respondent (denoted henceforth as "medication X"). Data were checked for accuracy and completeness within a day of collection and compiled in a Microsoft Access database (Microsoft Corp., Redmond, WA). 10.1371/journal.pone.0185827.t001 ###### Proportion of responses for which respondents responded "FALSE" for a set of true-or-false questions regarding the use of antibiotics. {#pone.0185827.t001g} *Do you think these statements are "TRUE" or "FALSE"?* (Expected response: "FALSE") ------------------------------------------------------------------------------------- ------------- ------------- 1\. one should stop taking {medication "X"} when one feels better 112 (83.6%) 113 (93.4%) 2\. {medication "X"} is effective against colds and flu 50 (34.3%) 31 (25.6%) 3\. it is okay to share {medication "X"} with someone else 116 (86.6%) 118 (97.5%) Figures in parenthesis show the percentage of the total household responses. Data analysis {#sec005} ------------- Summaries on antibiotic knowledge and antibiotic use practices were generated in Microsoft Excel (2016). The occurrence of unintended learning about antibiotics and their use was determined by conceptualizing the same individual before and after "exposure" to the longitudinal study as equivalent to a pair of "raters" who are temporally distinct, and by using agreement statistics (Cohen's kappa) in R (version 3.3.1) \[[@pone.0185827.ref016]\] to determine if people retained their original responses (high agreement) or changed them (low agreement). Changes in antibiotic-related knowledge, attitudes and practices were analyzed for households for which data was available for both entry and exit surveys (n = 149). Knowledge changes were assessed based on respondents' ability to name antibiotics and to correctly respond to the true-or-false questions regarding their use and whether they recalled receiving any information about the correct use of antibiotics. Attitude changes were assessed broadly to not only include the effect of antibiotic-related information on household antibiotic use practices but also dependence on healthcare institutions or healthcare workers for information regarding antibiotics and their use. Practice changes focused on reported behaviors relating to the use of antibiotics, and household decisions regarding the acquisition of antibiotics. Discordant pairs of survey responses by the same respondent were quantified as change. The Wilcoxon's rank-sum test was used to determine whether changes, if any, were statistically significant (*P* \< 0.05). We further analyzed participant responses to determine their direction of change. For the true-or-false knowledge questions, households in which respondents gave a "TRUE" response at the entry survey and "FALSE" at the exit survey. Questions that had several response options were analyzed for each option separately allowing them to be analyzed as "Yes-No" questions. We considered changes that led to improved knowledge, attitudes or practices as positive changes, otherwise the changes were considered negative. Thus, although information on quality---for example---of community pharmacists was not collected, we quantified attitudes and practices that favored contact with these healthcare sources as positive while those that did not (e.g. reliance on previous experiences or receiving information from family and friends) as negative. This study was approved by the Kenya Medical Research Institute Scientific and Ethic's Review Committee (SSC Protocol \# 2998), the Centers for Disease Control and Prevention (CDC protocol \# 6761) and the Washington State University Institutional Review Board (IRB Number \#14413--002). Oral and written consent were also obtained from study respondents before enrollment into the longitudinal study. Results {#sec006} ======= Two-hundred households were interviewed during the entry survey out of which 149 were interviewed during the exit survey (\~5 months between entry and exit surveys). The majority (\>93%) of respondents in both surveys were female. The mean age of respondents was 28.7 in the entry survey and 29.0 in the exit survey. The median household size was five people (range: 2--13) and all households had at least one child aged ≤5 years. The median level of education attained by male and female household heads was primary school. Knowledge of antibiotics {#sec007} ------------------------ Of the surveyed respondents, 67% named at least one antibiotic in the entry survey and 82% in the exit survey. All the antibiotics mentioned in the entry survey were mentioned in the exit survey, with the exception of metronidazole (exit only; [Fig 1](#pone.0185827.g001){ref-type="fig"}). Cotrimoxazole and amoxicillin were the most commonly mentioned antibiotics, jointly accounting for 85% and 77% of all responses in the entry and exit surveys, respectively. Overall, there were no differences in the frequency of respondents that mentioned various antibiotics for both surveys (ANOVA, *P* = 0.29). In both surveys, only 10% mentioned a second antibiotic---commonly amoxicillin or cotrimoxazole---while none mentioned a third antibiotic. The majority (≥ 84%) of respondents in both surveys reported that antibiotic use should not be discontinued following the alleviation of symptoms, and that antibiotics should not be shared. Nevertheless, 66% and 74% of respondents considered antibiotics effective for treating colds and flu in the entry and exit surveys, respectively ([Table 1](#pone.0185827.t001){ref-type="table"}). {#pone.0185827.g001} Sources of antibiotics and antibiotic use practices {#sec008} --------------------------------------------------- When asked where they would obtain the self-identified antibiotics when needed, most respondents mentioned health facilities (80%, entry and 69%, exit), and pharmacies (25%, entry and 30%, exit). In both surveys, the few respondents who mentioned penicillin, doxycycline and amoxicillin/clavulanate only named health facilities as potential sources for these antibiotics. Eighty-seven percent of respondents surveyed during the entry survey reported having used an antibiotic within the 12-month period preceding the study whereas 70% reported previous use in the exit survey. More specifically, 89.3% of the 85 respondents that named amoxicillin and 90% of the 30 respondents that named cotrimoxazole in the entry survey reported having used it in the past 12 months. In the exit survey, 66% of the 47 respondents that named amoxicillin (n = 47), and 68% of the 19 respondents that named cotrimoxazole reporting having used it within the same period. Most (13 out of 14) respondents that mentioned ampicillin in the exit survey reported having used it in the past year (only one respondent mentioned ampicillin in the entry survey). When asked to name the illnesses for which they would normally decide to take the antibiotics mentioned, respondents mentioned colds/flu, coughs, diarrhea, headache, fever, pneumonia and malaria ([Table 2](#pone.0185827.t002){ref-type="table"}). These were either mentioned as single ailments or as combinations of ailments. Amoxicillin, ampicillin, and cotrimoxazole use were mentioned against all of these conditions. Penicillin was not mentioned in the case of malaria, while tetracycline was not mentioned in the case of malaria and pneumonia. Amoxicillin/clavulanate and doxycycline were only mentioned in the case of colds and coughs while erythromycin was mentioned for colds, coughs, diarrhea and fever. Metronidazole was the only antibiotic that was mentioned against one condition (diarrhea). 10.1371/journal.pone.0185827.t002 ###### Survey responses to questions regarding antibiotic use, sources of antibiotics and information on antibiotics. {#pone.0185827.t002g} ------------------------------------------------------------------------------------------------- --------------------- -------------------- **Have you taken {medication "X"} within the last 12 months?** **Entry (n = 134)** **Exit (n = 122)** *Yes* 116 (86.6%) 85 (69.7%) **In what ways would you obtain {medication "X"} if you felt you needed to use it?** *Health facility* 107 (79.9%) 84 (68.9%) *Chemist* 34 (25.4%) 37 (30.3%) *Left-over* 1 (0.7%) 7 (5.7%) *Elsewhere* 1 (0.7%) 5 (4.1%) **How would you normally know which medicine to buy when someone falls ill?** *Recommendation by community pharmacist* 93 (69.4%) 65 (53.7%) *Previous prescription by a clinician* 39 (29.1%) 38 (31.4%) *Own experience* 26 (19.4%) 20 (16.5%) *Opinion of family or friends* 5 (3.7%) 9 (7.4%) **In the last year, do you remember getting information about proper use of {medication "X"}?** *Yes* 35 (31.1%) 56 (45.9%) **From where did you get information?** *Clinician* 25 (71.4%) 32 (57.1%) *Community pharmacist* 15 (42.9%) 14 (25.0%) *Other health professionals (e.g. nurse)* 5 (14.3%) 10 (17.9%) *Friend/family member* 3 (8.6%) 4 (7.1%) **In what ways did the information change your views about the use of {medication "X"}?** *Complete dosage* 16 (45.7%) 25 (44.6%) *Always consult a clinician* 15 (42.9%) 32 (57.1%) *Not take this medicine without prescription* 10 (28.6%) 27 (48.2%) **Which sources would you trust if you needed information on {medication "X"}?** *A clinician* 118 (88.1%) 95 (78.5%) *A community pharmacist* 64 (47.8%) 98 (81.0%) *A hospital* 40 (29.9%) 55 (45.5%) *A nurse* 27 (20.1%) 51 (42.1%) **For which illnesses would you normally decide to take {medication "X"}?** *Colds/flu* 81 (60.4%) 57 (46.7%) *Cough* 81 (60.4%) 60 (53.3%) *Fever* 53 (39.6%) 23 (18.9%) *Diarrhea* 21 (15.7%) 6 (4.9%) *Malaria* 13 (9.7%) 7 (5.7%) ------------------------------------------------------------------------------------------------- --------------------- -------------------- Multiple responses were allowed for survey questions. The combination of symptoms for which most respondents reported they would use an antibiotic was colds/flu with an accompanying cough (26%, entry and 30%, exit). The presence of fever as an additional symptom accounted for an additional 15% of responses in the entry survey and 5% in the exit survey. Antibiotic use for coughs accounted for 7% of responses in the entry survey and 23% in the exit survey. Of the respondents who named any antibiotic and said that they would take it for colds/flu + cough (n = 30, entry and n = 25, exit), 80% named amoxicillin in the entry survey and 56% in the exit survey. An additional one-third (32%) named cotrimoxazole in the exit survey. Antibiotic information sources and their impact on antibiotic usage {#sec009} ------------------------------------------------------------------- Approximately 70% of respondents surveyed at entry and 54% of those surveyed at exit reported that they rely on community pharmacists to recommend medication in the event of an illness within the household. One third of respondents also reported relying on previous prescriptions by a clinician and almost 20% reported that their choice would be based on their own experience. Few respondents reported relying on the opinions of family or friends when deciding which medicine to buy ([Table 2](#pone.0185827.t002){ref-type="table"}). Less than half of all surveyed respondents (31.1%, entry and 45.9%, exit) recalled getting information regarding proper use of antibiotics in the 12-month period preceding the two surveys. The primary information sources were clinicians and community pharmacists, who were also considered the most trustworthy sources of information regarding antibiotics. In both surveys, all respondents that recalled getting information reported that it changed their views on the usage of the antibiotic. These changes in views included always consulting a clinician, not taking non-prescribed antibiotics and completing antibiotic doses ([Table 2](#pone.0185827.t002){ref-type="table"}). Changes in knowledge, attitudes and practices related to antibiotics and their use {#sec010} ---------------------------------------------------------------------------------- The agreement between paired household responses for all variables ranged from no agreement (Cohen's kappa, κ = -0.003) to weak agreement (κ = 0.22). For example, 12 of the 102 respondents that mentioned an antibiotic during the entry survey could not mention an antibiotic during the exit survey. Similarly, 44 of the 87 respondents who did not consider antibiotics effective against colds/flu during the entry survey considered them effective during the exit survey. In general, the level of agreement was independent of whether the question was about knowledge, attitudes or practices. Nevertheless, the only variables with significant changes in responses between the two surveys were knowledge of an antibiotic, the effectiveness of antibiotics against colds/flu, reliance on previous prescriptions when deciding which antibiotic to buy, always consulting a clinician before taking an antibiotic and trusting nurses for information on the correct use of antibiotics (Wilcoxon test, *P*\<0.05; [Table 3](#pone.0185827.t003){ref-type="table"}). 10.1371/journal.pone.0185827.t003 ###### Total households (counts and proportions; n = 122) in which responses for knowledge, attitude and practice survey questions changed between entry and exit surveys, and the significance levels of the observed changes (Wilcoxon rank sum test, *P* \< 0.05). {#pone.0185827.t003g} Sub-group Variable HH changed (%) *P*. value ---------------------------------------------------------------------------------- -------------------------------------------------------------------------------- ---------------- ------------ **Knowledge** Know or can mention an antibiotic[^ŧ^](#t003fn001){ref-type="table-fn"} 44 (29.5) 0.003 Antibiotics effective against colds/flu 52 (42.6) 0.011 Stop taking antibiotics if feeling okay 21 (17.2) 0.356 Sharing dose of antibiotics is acceptable 14 (11.5) 0.205 Remember getting information about antibiotics 58 (47.5) 0.966 Got information from clinician[^Ɨ^](#t003fn002){ref-type="table-fn"} 28 (50.0) 0.131 Got information from health professional[^Ɨ^](#t003fn002){ref-type="table-fn"} 11 (19.6) 0.493 Got information from pharmacist[^Ɨ^](#t003fn002){ref-type="table-fn"} 15 (26.8) 0.753 **Attitude** Changed view on antibiotics 58 (47.5) 0.966 Always consult before using antibiotics[^Ɨ^](#t003fn002){ref-type="table-fn"} 36 (64.3) 0.037 No taking antibiotics without prescription[^Ɨ^](#t003fn002){ref-type="table-fn"} 26 (46.4) 0.317 No self-medication with antibiotics[^Ɨ^](#t003fn002){ref-type="table-fn"} 23 (41.4) 0.816 Completing antibiotic doses[^Ɨ^](#t003fn002){ref-type="table-fn"} 24 (42.9) 0.447 Trust community pharmacist for information on antibiotics 70 (57.4) 0.369 Trust clinician for information on antibiotics 46 (37.7) 0.260 Trust hospital for information on antibiotics 49 (40.2) 0.059 Trust nurse for information on antibiotics 44 (36.1) 0.009 Trust health facility for information on antibiotics 10 (8.2) 0.682 **Practice** Use antibiotics for cough 60 (49.2) 0.803 Use antibiotics for cold or flu 54 (44.3) 0.239 Use antibiotics for diarrhea 18 (14.8) 0.692 Use antibiotics for fever 46 (37.7) 0.764 Use antibiotics for headache 23 (18.9) 0.911 Use antibiotic for malaria 15 (12.3) 0.480 Use antibiotics for pneumonia 15 (12.3) 0.480 Rely on recommendation from community pharmacist 58 (47.5) 0.606 Rely on previous prescription by a clinician 31 (25.4) \<0.001 Rely on own experience on antibiotics 28 (23.0) 0.323 Rely on opinion of family or friends 11 (9.0) 0.705 ^ŧ^Based on 149 household responses (main question) ^Ɨ^based on 56 household responses (sub-question). In general, there was an overall net positive change in households reporting improved knowledge, attitudes and practices with respect to antibiotics ([Fig 2](#pone.0185827.g002){ref-type="fig"}). The highest net positive change was for the number of households that considered community pharmacists as trustworthy information sources on antibiotics (32.8%, entry and 80.3%, exit; n = 122). In contrast, the highest net negative change was in the number of respondents who considered antibiotics effective against colds/flu (44.3%, entry and 73.8%, exit, n = 122). Interestingly, changes in knowledge and attitudes (regardless of direction) did not correspond with changes in reported practices. For instance, despite a significant increase in the number of households that considered antibiotics effective against colds/flu (*P* = 0.01), fewer households reported that they would use antibiotics for these conditions ([S1 Table](#pone.0185827.s002){ref-type="supplementary-material"}), suggesting that knowledge was unlikely to affect practice. {ref-type="supplementary-material"}.](pone.0185827.g002){#pone.0185827.g002} Discussion {#sec011} ========== Many studies attribute the increasing prevalence of antimicrobial resistance in sub-Saharan Africa to indiscriminate antibiotic use, exacerbated by uncontrolled access to these therapies \[[@pone.0185827.ref005]\]. Antibiotic use practices are largely context-specific \[[@pone.0185827.ref002]\], and several studies in high-income countries have documented more antibiotic prescriptions and use in populations with lower- compared to higher socio-economic status \[[@pone.0185827.ref017]--[@pone.0185827.ref019]\]. Nevertheless, the patterns of antibiotic resistance at the community level as well as the factors that influence antibiotic use practices in sub-Saharan Africa remain largely undescribed \[[@pone.0185827.ref005]\]; even less is known about informal settlements. Our survey of two villages in Kibera, a large informal settlement in Kenya, found that while up to 82% of surveyed respondents mentioned one antibiotic when prompted, very few mentioned a second, while none mentioned a third antibiotic. This suggests either poor antibiotic recall or that respondents were only familiar with a few antibiotics. The two antibiotics that respondents seemed most familiar with were amoxicillin and cotrimoxazole; two broad-spectrum antibiotics that are likely the most widely used in this area, and ones for which several studies in Africa have reported that access is generally ubiquitous \[[@pone.0185827.ref020],[@pone.0185827.ref021]\]. Most respondents had partial knowledge on the correct use of antibiotics. For example, they reported that antibiotics should not be shared or that their use should not be discontinued upon alleviation of symptoms. Nevertheless, many considered antibiotics to be effective against colds or flu. While this misconception may be considered an unusual finding for a population that has been the subject of an official infectious disease surveillance program for the past 10 years, studies in different regions of the world report that despite the general population awareness of antibiotics, many people consider these drugs effective against colds and flu \[[@pone.0185827.ref022]--[@pone.0185827.ref024]\] regardless of evidence in the contrary \[[@pone.0185827.ref025]\]. Health facilities were the primary sources of antibiotics for the majority of surveyed respondents, with community pharmacists serving as the next most common sources. Several plausible explanations for the higher preference for health facilities exist: (i) this area is served by a study clinic (Tabitha) where PBIDS participants receive care and medications for acute illnesses free-of-charge; (ii) the likelihood of recall bias may have prompted respondents to refer to "Tabitha clinic" as their default source; (iii) the cost implications of privately purchasing antibiotics may have deterred respondents from using alternative sources such as pharmacies; and (iv) misconceptions about the survey objectives (e.g. that surveys would determine whether or not to withdraw the treatment incentive, or whether or not the clinic was an important facility) could have influenced reporting in favor of the study clinic. Other studies have reported important roles for drug dispensers (e.g. community pharmacists, drug hawkers) as antibiotic sources \[[@pone.0185827.ref026]\] indicating the need for their inclusion in education interventions about antibiotics. The prevalence of reported antibiotic use in this study was marginally higher than that reported in other parts of the world. For example, out of 12 middle- and low-income countries, 35% to 76% of respondents reported taking antibiotics during the previous six months \[[@pone.0185827.ref027]\], perhaps consistent with the expected higher burden of disease in an informal settlement. This same report noted that 57% of respondents felt that they could do little to combat antibiotic resistance, which could be a further consequence of a higher burden to disease in these communities. For the current study, colds/flu and coughs were the most common reasons cited by respondents for seeking antibiotics. With the exception of metronidazole, which was only reported for potential use in the treatment of diarrhea, all antibiotics mentioned in the surveys were considered potential treatments for colds/flu and coughs, with or without associated fever. In Kenya, metronidazole is commonly administered as an anti-parasitic drug, particularly for infections caused by *Giardia lamblia* and *Entamoeba histolytica*. It is therefore likely that even when acquired over-the-counter, most households would consider its use primarily for gastrointestinal symptoms that include diarrhea. In general, however, the patterns of reported antibiotic use practices in Kibera differ from those reported in other surveys in Africa. For instance, a study conducted in a Nigerian slum \[[@pone.0185827.ref026]\] reported significantly higher rates of use of Ampiclox i.e. ampicillin + cloxacillin (79%), tetracycline (54%), metronidazole (51%) and ampicillin (44%). Similarly, a study of antibiotic use in five African counties \[[@pone.0185827.ref002]\]--including Kenya--showed patterns of use that differ somewhat from those reported herein. In this study, the prevalence of amoxicillin and cotrimoxazole prescriptions was only 37.4 and 27.8% (n = 292), respectively, among Kenyan patients with an acute illness. Differences in study methodologies may account for some of the observed differences. Additionally, Feikin *et al*. \[[@pone.0185827.ref028]\] found that antibiotic recall declined by 16--23% per week in the two villages we surveyed, and that respondents were less able to recognize antibiotics by name. Both recall and knowledge of antibiotics may have contributed to the variation observed in our study. Our results show that community pharmacists (who may not be qualified pharmacists) play an important role in influencing household medication-use decisions in this community, consistent with other studies in Africa \[[@pone.0185827.ref006],[@pone.0185827.ref021]\]. Close to 70% of respondents reported seeking recommendations from community pharmacists regarding which medicines to buy in the event of an illness. While this contradicts the finding that the majority of respondents cited health facilities as their primary sources of antibiotics, respondents may have been referring more specifically to instances where use of the Tabitha clinic was not feasible or was not deemed necessary. These likely include instances when illness in a household arises beyond the hours of operation of the study clinic, where time constraints limit the use of the clinic or when illness is not considered severe enough to warrant a clinic visit. Nevertheless, the finding that households also relied on previous prescriptions by a clinician and on personal experiences suggests that an unknown degree of self-medication may occur in these communities. More than half of respondents said they had not received information about the correct use of antibiotics within the last 12 months, which is consistent with reports that antibiotic information is poorly communicated by pharmacies and drug shop operators in sub-Saharan Africa \[[@pone.0185827.ref028]\]. Given the precipitous rate at which antibiotic recall is reported to decline in this community \[[@pone.0185827.ref029]\], it is possible that when information is offered, respondents may not accurately recall unless there is significant reinforcement of these messages. Alternately, respondents may have understood the question as referring to information offered through public health education drives, which may be rare in this area. The finding that respondents consider clinicians, community pharmacists and other healthcare personnel as important and trustworthy sources of information about antibiotics presents an opportunity, not only for providing education on the correct use of antibiotics, but also for correcting the existing misconceptions regarding antibiotics. This can only be realized if physician education about the correct use of antimicrobials is increased. A survey that included 98 resident doctors in a national teaching and referral hospital in Kenya found that only 14.1% of those surveyed had received four or more lectures regarding antibiotic use in the past year \[[@pone.0185827.ref030]\]. Participation in an observational study may indirectly result in knowledge transfer from the research team to the participant, influencing the outcome of interest. Our evaluation of the potential effect of participant enrollment (into the larger longitudinal study) on household knowledge, attitudes and practices suggests an overall net-positive correlation. Nevertheless, it remains unclear whether unintentional learning occurred among our study participants despite the positive effect. Our analysis revealed low agreement between the responses provided during the entry and exit surveys. This is exemplified by the true-false-response questions where up to 40% of households changed their responses from true to false or *vice-versa* between the two surveys, suggesting that knowledge loss and gain is quite variable in this population. We surmise that knowledge about antibiotics is this community is generally lacking, or superficial at best, and that in the absence of public health education, antibiotic KAP surveys are likely to yield general ideas based on respondents' value judgements rather than a deeper understanding of the concepts related with the use of antibiotics. This study probably suffered two common limitations inherent in KAP surveys, i.e., participants providing socially desirable responses, and recall bias. A more unique limitation was that respondents were drawn from an area that has been under health surveillance for a decade, although this ensured our ability to randomly select households and to generally encounter cooperative participants. Thus, the prevalence and types of antibiotics, and in turn, antibiotic use knowledge and practices may have been different from those in other populations. Despite these limitations, the findings from our study contribute to the general knowledge of potential drivers of antibiotic use in informal settlements, which are rarely represented in the literature. Conclusion {#sec012} ========== Our study shows that for the Kibera study population, only 2--3 antibiotics (beta-lactams or sulfa-class antibiotics) are commonly recognized and are probably commonly used in the event of illness within a household, particularly for colds/flu and coughs. The general lack of understanding of what antibiotics are and how they should be used correctly presents a challenge for AMR control efforts. Nevertheless, the existing trust that respondents have in healthcare workers presents an opportunity for targeting educational interventions pertaining to antibiotics and their use, which can over time remedy the prevailing situation. Supporting information {#sec013} ====================== ###### Antibiotic use knowledge, attitudes and practices questionnaire. (PDF) ###### Click here for additional data file. ###### Variables analyzed to show household changes in knowledge, attitudes and practices. (PDF) ###### Click here for additional data file. Special thanks go to the data collection team (Daniel Owiti, Samuel Owino, David Murunga, Eric Olendi and Allan Owuor) and their colleagues/supervisors (Alice Ouma, Shadrack Muema, Moses Omolo and Ondari Mogeni), and to all the study participants for their participation in- and dedication to this study. We also thank the KEMRI Director for approving the publication of this work and the Paul G. Allen School of Global Animal Health--Washington State University, (USA), Kenya Medical Research Institute and the Centers for Disease Control and Prevention-Kenya for making this work possible. [^1]: **Competing Interests:**The authors have declared that no competing interests exist.

**Sir**, We thank [@bib10] for their interest in our recent study. The authors applied our revised CIMP classification to their three different clinical data sets, which are composed of patients with early-onset colorectal cancer (EOCRC) (younger than 45 years), patients with late-onset CRC (LOCRC) (older than 70 years), and individuals diagnosed with synchronous CRC (SCRC). They addressed that only LOCRC cases showed similar tendency of increasing *BRAF* mutation, MSI-high and *MLH1* methylation along with the increase in the number of methylated genes. Moreover, they insisted that prognostic results were only partially confirmed in SCRC. Non-linearity of molecular alterations including *BRAF* mutation, MSI-high and *MLH1* methylation in EOCRC and SCRC might originate from two reasons. First, EOCRC and SCRC have strong germline predispositions to CRC, even though they are not either familial adenomatous polyposis or Lynch syndrome ([@bib3]; [@bib4]). Recent studies revealed that germline predisposition in EOCRC is greater than expected ([@bib7]); these germline predispositions are mainly associated with chromosomal instability rather than CIMP ([@bib2]; [@bib5]). Second, CIMP-P2 CRCs usually occur in elderly patients. Two recent studies showed similar tendency in CRCs with *BRAF* mutation and concurrent MSI or MLH1 methylation ([@bib9]; [@bib11]). The authors tried to validate prognostic value of our revised CIMP classification in their EOCRC, LOCRC and SCRC subgroups. However, considering their previous publication, the sample size of each subgroup is too small to get enough statistical power for the subgroup of low prevalence, such as CIMP-P1 and CIMP-P2 ([@bib6]; [@bib8]; [@bib1]). Considering the low prevalence of EOCRC and SCRC, a multi-centre study might be necessary to validate the prognostic value of our revised CIMP classification. Overall, Tapial *et al*'s results emphasise the fact that EOCRC and SCRC have different molecular landscapes compared with sporadic CRCs. Further comprehensive study might shed light on the complex interaction between germline predisposition, accumulation of somatic mutation and epigenetic alteration. This work is published under the BJC\'s standard license to publish agreement. After 12 months the license terms will change to a Creative Commons AttributionNonCommercial-Share Alike 4.0 Unported License. The authors declare no conflict of interest.

Pancreaticoduodenal artery (PDA) aneurysm is a rare type of visceral artery aneurysm, which also includes splenic, renal, hepatic, mesenteric, and aortic arteries aneurysms.[@CIT1] PDA aneurysms have nonspecific clinical presentations,[@CIT2] which explains why they are not usually considered in the differential diagnosis of patients who present with epigastric pain or vomiting. We report a case of a patient with a PDA aneurysm who presented with repeated vomiting and epigastric discomfort. The aneurysm was found incidentally in a CT scan of the abdomen. Visceral angiography was done to visualize the aneurysm and the diagnosis of PDA aneurysm causing a gastric outlet obstruction was confirmed. The patient was referred to a specialized center and was successfully treated. Although the PDA aneurysm was reported in many cases, the presentation as gastric outlet obstruction is rare. Commonly, gastric outlet obstruction is caused by gastric and peripancreatic malignancies in developed and developing countries.[@CIT3] Peptic ulcer disease is the most common benign cause of gastric outlet obstruction.[@CIT4] We present this case to draw attention to the fact that PDA and other visceral aneurysms are rare causes of gastric outlet obstruction, which should be considered if none of the more common causes of gastric outlet obstruction can be identified. CASE {#sec1-1} ==== A 48-year-old Saudi man presented with repeated vomiting and epigastric discomfort for a duration of ten days. Vomiting occurred one hour after eating, and was projectile, yellowish in color, and contained undigested food with no mucus or blood. It was associated with mild epigastric discomfort, which was of gradual onset, intermittent, dull and aching in nature, and occasionally felt in the back and both hypochondria. There was no history of peptic ulcer disease or drug ingestion. The patient denied any history of fever, cough, weight loss, or trauma. He was diagnosed with hypertension 10 months before presentation for which he was taking fosinopril 10 mg once daily. The family history was negative for aneurysms and malignancies. The patient was retired from the military services. He drank alcohol occasionally and had a 10 pack-year smoking history. Clinical examination revealed a middle-aged man who was fully conscious, afebrile, and hemodynamically stable. Systemic examination revealed no abnormality, but abdominal examination revealed mild tenderness in the epigastric area. The rectal examination was unremarkable. Blood tests were unremarkable ([Table 1](#T0001){ref-type="table"}) Esophagogastroduodenoscopy revealed hiatus hernia with mild gastritis not consistent with the clinical condition of the patient. A gastric biopsy was negative for malignancy. A CT scan of the abdomen demonstrated a rounded structure in the posterior aspect of the pancreas, anterior to the third part of duodenum, which measured 2.5 cm and compressed the duodenum, causing dilatation in the stomach, and the first and second part of the duodenum, which suggested an aneurysm in PDA or gastroduodenal (hepatic) artery ([Figure 1](#F0001){ref-type="fig"}). A CT angiography suggested by the radiologist confirmed the diagnosis of PDA aneurysm causing gastric outlet obstruction ([Figure 2](#F0002){ref-type="fig"}). The patient received supportive treatment and was referred to a tertiary center for embolization of the aneurysm. ###### Laboratory investigations. Complete blood count and coagulation profile ---------------------------------------------------- White blood cell count: 17.8×10^3^ cells per mm^3^ Hemoglobin: 11.7 g/dL Hematocrit: 35.2% Platelets: 585 000 per mm^3^ Mean corpuscular volume: 65.8 fL Mean corpuscular hemoglobin: 21.9 pg Prothrombin time: 14.6 seconds International normalized ratio: 1.09 Partial thromboplastin time: 34.4 seconds **Renal profile** Sodium: 142 meq/L Potassium: 3.7 meq/L Chloride: 100 meq/L Bicarbonate: 27 meq/L Urea: 6.8 mmol/L Creatinine: 89 μmol/L Glucose: 7.2 mmol/L Anion gap: 15 **Liver function tests** Lactate dehydrogenase: 404 U/L Alanine aminotransferase: 107 U/L Alkaline phosphatase: 163 U/L Albumin: 44 g/L Total protein: 80 g/L (60-85 g/L) Amylase: 77 U/L {#F0001} {#F0002} DISCUSSION {#sec1-2} ========== PDA aneurysms are very rare. They comprise 2% of all visceral artery aneurysms.[@CIT5] Since Ferguson et al reported the first case of PDA aneurysm in 1895,[@CIT6] 88 cases of PDA aneurysms were reported in the English literature until 1993 and 52 cases of PDA aneurysms were reported between 1973 and 1999.[@CIT7] More reported cases increased awareness as to the importance of early detection before rupture. Men are four times as likely as women to have PDA aneurysms.[@CIT8] The vast majority of patients with these aneurysms experience epigastric pain and discomfort. This may be secondary to underlying pancreatic disease in 30% of PDA aneurysms.[@CIT8] Other clinical presentations include hemosuccus pancreaticus--which is the presence of bleeding into the pancreatic duct--or hemobilia, jaundice, and shock.[@CIT9] Reported cases of PDA aneurysms have also included a case that presented with vaginal bleeding and incidentally was found to have a pulsatile mass on ultrasound examination.[@CIT10] Another case presented with an incarcerated inguinal hernia that subsequently hemorrhaged into the retroperitoneum from a ruptured PDA aneurysm.[@CIT10] Itoh et al reported a case of PDA aneurysm causing pancreatic pseudotumor and duodenal obstruction, which on angiography was an aneurysm of 8 mm in diameter, found in the posterior superior PDA.[@CIT11] Androulakakis et al reported a case of a gastric outlet obstruction caused by giant gastroduodenal artery aneurysm.[@CIT12] Chiou et al reported a case of pancreaticoduodenal artery aneurysm that began with intestinal angina and weight loss.[@CIT13] The most common cause of these aneurysms is pancreatitis-related vascular necrosis or vessel erosion by an adjacent pancreatic pseudocyst.[@CIT8] Othe causes also include atherosclerosis, infection, congenital defects, fibromuscular dysplasia, connective tissue disorders (polyarteritis nodosa or Takayasu arteritis) and trauma.[@CIT1][@CIT8] Our patient had atherosclerosis risk factors which included a past history of hypertension and a 10-pack year smoking history. Alcohol consumption is also a risk factor for pancreatitis; Any of these factors might have been the cause of his aneurysm. Gastric outlet obstruction is commonly caused by gastric malignancy or stenosis of the pylorus as a complication of chronic peptic ulceration.[@CIT14] Other benign differential diagnoses of gastric outlet obstruction include infections such as tuberculosis and infiltrative diseases such as amyloidosis. Gastric outlet obstruction may also be caused by gallstone, a condition termed as Bouveret syndrome. Superior and inferior PDAs communicate anteriorly to the head of the pancreas and medial to the first part of the duodenum ([Figure 3](#F0003){ref-type="fig"}).[@CIT15] Arteriography is necessary to confirm the existence of PDA aneurysms. CT or magnetic resonance angiography (MRA) are also of importance in recognizing these aneurysms and are helpful in detecting the presence of rupture or associated pancreatic disease.[@CIT10] In our case, a contrast-enhanced CT scan of the abdomen revealed the aneurysm from the PDA, which has a blood supply from both the superior mesenteric artery and gastroduodenal artery and measures 2.5 cm, causing narrowing of the transverse duodenum with dilatation of stomach, and the first and second part of the duodenum. {#F0003} We believe that PDA aneurysms and other visceral aneurysms should be considered in any patient who presents with repeated vomiting and gastric outlet obstruction symptoms, especially if no cause can be found.

1. Introduction {#sec1-jcm-08-00456} =============== The use of Home Exercise Programs (HEPs) for children with disabilities, is a widespread resource that often contributes to an increase in their amount of practice and functioning \[[@B1-jcm-08-00456]\]. The HEPs prescribed for children are usually individualised, based on family goals, and generally include different types of exercises and interventions \[[@B2-jcm-08-00456],[@B3-jcm-08-00456]\]. It is well known that adherence is a desirable and essential behaviour in order to achieve the goals of an HEP. However, parental adherence is estimated to be lower than 50% \[[@B4-jcm-08-00456],[@B5-jcm-08-00456]\]. Nevertheless, as studies have typically measured parents' adherence to HEPs as a whole and single construct \[[@B5-jcm-08-00456]\], this percentage could vary for different exercises. According to evidence from qualitative research \[[@B6-jcm-08-00456],[@B7-jcm-08-00456]\], it is suggested that instead of performing the whole home program, caregivers usually choose those activities that are easiest for them. For example, studies on cystic fibrosis have shown that there are different levels of adherence to different parts of a home therapeutic regime, such as pharmacological treatment, nutritional prescriptions and exercise \[[@B8-jcm-08-00456]\]. Given the fact that there is a lack of knowledge regarding adherence according to different types of exercise \[[@B3-jcm-08-00456]\], there is a need to study whether certain factors associated to the exercises prescribed in an HEP may influence adherence. The existing literature provides some insight into the different factors that influence adherence to different kind of exercises in adult populations \[[@B9-jcm-08-00456]\]. Nevertheless, it has been suggested that adherence to children's exercise regimes is influenced by factors that may not be relevant for adults \[[@B10-jcm-08-00456]\], as more complex relations \[[@B11-jcm-08-00456]\] and circumstances appear simultaneously among children, parents and health professionals. This study examined three issues in a population of children with or at risk of developmental disabilities attending paediatric services in early intervention centres: (1) to determine rates of parents' adherence to different types of exercises featured in their child´s HEP; (2) to identify what factors related to the environment, the parents and the child receiving treatment are more likely to be associated to parents' adherence to these types of exercises; (3) to assess the relative influence of the behaviour of health professionals on parents' adherence to each type of exercise after making adjustments for associated environmental, parental and child-related factors. 2. Methods {#sec2-jcm-08-00456} ========== 2.1. Study Design and Participants {#sec2dot1-jcm-08-00456} ---------------------------------- This study was a multicentre survey design using a self-administered questionnaire. The study obtained approval by the Ethics Committee of the University of Murcia (approval No. 129/05). All participants provided written informed consent prior to the data collection. The inclusion criteria consisted of parents of children aged between six months and six years with a prescribed HEP and attending early intervention centres in Murcia (Spain) over a period of six months at least. The exclusion criteria consisted of the inability to read or write in Spanish. 2.2. Data Collection and Measurements {#sec2dot2-jcm-08-00456} ------------------------------------- The analysis developed in this paper is based on a self-report questionnaire ([Table A1](#jcm-08-00456-t0A1){ref-type="table"}). The questionnaire included questions about components of HEP, adherence behaviours and potential associated factors. A previous qualitative study \[[@B7-jcm-08-00456]\] was used to identify dimensions of healthcare professionals' behaviours associated to parents' adherence. On the basis of comments made by parents, several candidate questionnaire items were written for each domain. An associated literature review was also conducted for identifying representative items of other associated potential factors. The entire group of items was reviewed by two measurement and content experts. The scoring system that was applied considered usefulness and non-repetitive content, clarity and appropriateness for parents. Cognitive pre-test interviews were used to test for comprehension and comfort with the response format and instructions. Additionally, for the factors "healthcare professionals' behaviours" and adherence behaviours, we examined the construct validity and internal consistency reliability using exploratory factor analyses and Cronbach's, respectively. We also examined internal consistency multi-item scales used for measuring other factors (e.g., knowledge and ability to carry out the home program). Finally, for those factors measured by scales with single items, the reliability was indirectly studied by analyzing the number of non-specific answers or answers left unanswered when not well understood by the patients, resulting in a problem with reliability. Parents who accepted to participate in this study received the self-administered questionnaire from their attending therapist, who explained how to use the questionnaire and the types of exercises from the checklist, prior to completing the same. The parent more frequently responsible for HEP completed the questionnaire at home and sent it by post to the research team at the University of Murcia. A stamped addressed envelope was provided to encourage response rates, followed by a verbal reminder from the attending therapist one week later. Concurrently, the therapist also collected demographic information regarding the selected parent. 2.3. Characteristics of Home Program {#sec2dot3-jcm-08-00456} ------------------------------------ The early intervention centres in Murcia commonly provide families with am HEP, so that parents can perform some exercises with their children at home or in everyday situations. Due to the children's age and condition, HEPs were usually implemented by parents or caregivers. Both the contents of the HEP and the specific dosage of each exercise were previously agreed on between each family and the physiotherapist from the early intervention centre, as instructed by the physiotherapist. The HEPs varied depending on the children's development, age and needs, including several exercises and instructions. In our study, the questionnaire included a checklist (based on yes/no responses) with three types of therapeutic exercises commonly used in treatments with paediatric populations and explanations about its contents to clarify them to the parents. The three types of therapeutic exercises had been identified from an International Classification of Therapeutic Exercises, as follows \[[@B12-jcm-08-00456]\]: flexibility exercises, neuromotor development training exercises (NDT) (including the training of motor development skills such as head control, crawling, rolling and walking, necessary for improving children's functioning), and body mechanics and postural stabilisation (BMPS) exercises which included positioning the child with or without external postural support in different positions such as lying, sitting and standing). These types of exercises were recommended to ensure an appropriate alignment of the body segments and the development and control of different postures). 2.4. Adherence Behaviours {#sec2dot4-jcm-08-00456} ------------------------- Adherence was measured for each type of therapeutic exercise, in accordance with the prescribed dosage of the HEP. We used a five-point frequency-based response scale (never, rarely, sometimes, very often and always), adapted from the adherence scale by Sluijs et al. \[[@B13-jcm-08-00456]\]. The frequency scale had to be answered according to the parents' adherence to the recommended dosage, in each case. For example, if they had been recommended to perform an exercise once a day, or for 10 min per day, they had to indicate how often they had been able to perform the exercise, compared to the prescription. Other studies \[[@B5-jcm-08-00456],[@B13-jcm-08-00456]\] have suggested that many parents often overestimate their adherence, and we therefore decided to use a highly demanding level in the scale, in order to consider the adherence. Adherence was considered as a dichotomous variable (adherent or not adherent). The categories 'always' and 'very often' were considered adherent. Furthermore, the same cut-off points were based on the available literature \[[@B3-jcm-08-00456],[@B13-jcm-08-00456]\], and were established before the data analysis. A factor analysis confirmed a one-dimensional structure; the factor explained 60.4% of the total variance, the Kaiser-Meyer-Olkin statistic was 0.61, and the Bartlett statistic was 29.53 (*p* \< 0.01). The result of the internal consistency reliability measure was acceptable: α = 0.70. 2.5. Potential Factors Associated with Adherence {#sec2dot5-jcm-08-00456} ------------------------------------------------ After reviewing the relevant literature \[[@B5-jcm-08-00456],[@B11-jcm-08-00456],[@B14-jcm-08-00456],[@B15-jcm-08-00456]\], we considered three areas of potential factors: (1) individual; (2) social support and resources; and (3) illness/treatment. ### 2.5.1. Individual Variables {#sec2dot5dot1-jcm-08-00456} We assessed parents' demographic and cognitive variables, as well as children's demographic variables. Parents' demographic characteristics were: age (years), gender (male/female), education level (without studies/primary/secondary/university), work participation (yes/no), type of family (two parents/one), and number of children in the home. The child's age and gender were also measured. The following cognitive constructs and instruments were used in the case of parents:Perceived barriers to integrate exercises into a daily routine. This was measured using a single item from a validated instrument \[[@B13-jcm-08-00456]\] on a five-point frequency-based response scale (never, rarely, sometimes, very often and always). This scale was coded for statistical analysis as a dichotomous variable, indicating either a low or high perception of barriers, with responses of "never" and "rarely" coded as a low perception.Knowledge and ability to carry out the home program. This was measured using two items from a validated parent home program compliance questionnaire \[[@B16-jcm-08-00456]\] ("I understand my child's home program" and "I am skillful in carrying out\...") on a five-point agreement-based response scale (strongly disagree, disagree, undecided, agree and strongly agree). Low knowledge was defined as a response of "disagree" or less in the two items. We made these transformations before the data analysis and based on previous work \[[@B3-jcm-08-00456]\]. The result of the internal consistency reliability measure was acceptable: α = 0.70.Self-efficacy. Parents were asked about how confident they felt participating in the HEP. Measurements were based on a single item from a valid self-efficacy scale \[[@B17-jcm-08-00456]\], with a response scale ranging from 0 (minimum) to 10 (maximum). This measure was categorized on two levels (low or high), with responses equal or higher than the median coded as high self-efficacy. The non-answer and non-specific answer rate were lower than 5% in all these items (range 1.8% to 4.6%), and therefore they were not excluded. ### 2.5.2. Factors Related to Social Support and Resources {#sec2dot5dot2-jcm-08-00456} Social support was measured by a single item indicator assessing the instrumental support provided to the parents on a five-point scale (never-always) and grouped into two categories (low or high), with responses of "very often" and "always" coded as high support. The availability of home equipment for exercises (e.g., exercise mat, ball or other equipment) was reported (yes/no) as a physical resource factor. The non-answer and non-specific answer rate were also lower than 5% (range 1.8% to 2.7%). ### 2.5.3. Factors Related to Illness/Treatment {#sec2dot5dot3-jcm-08-00456} We considered the kind of health condition (cerebral palsy/congenital disease/other); functioning (able/not able) for several mobility activities such as sitting, crawling and walking without assistance; and time in treatment (more/less than 2 years). 2.6. Healthcare Professionals' Behaviours during Therapeutic Encounters {#sec2dot6-jcm-08-00456} ----------------------------------------------------------------------- Seven relevant questions, based on common paediatric recommendations \[[@B3-jcm-08-00456],[@B11-jcm-08-00456],[@B18-jcm-08-00456]\], were asked concerning three areas: giving general information to parents, instructions and follow-up of exercises. All these questions used a five-point frequency-based response scale (never-always). This measure was categorised on two levels (suitable frequency or not), with the responses 'always' and 'very often' coded as a suitable frequency. A factor analysis confirmed a one-dimensional structure and therefore the items measured the same underlying construct; the factor explained 59.6% of the total variance, the Kaiser-Meyer-Olkin statistic was 0.86, and the Bartlett statistic was 751.5 (*p* \< 0.01). Cronbach's α was high: α = 0.88; and all of the items contributed to the reliability and construct validity of the questionnaire. One further item about the parents' overall satisfaction with the process of care was also included. This was measured using a unidimensional scale ranging from 0 (minimum) to 10 (maximum) and categorised on two levels (low or high), with responses equal or higher than the median coded as a high satisfaction. 2.7. Data Analysis {#sec2dot7-jcm-08-00456} ------------------ Descriptive statistics using proportions and their 95% confidence interval (CI) were calculated for the description of the sample and the adherence rates. Considering a margin of error (degree of accuracy) of 10% on the estimation of adherence, the minimal sample size required was 96. Respondents and non-respondents to the postal questionnaire were compared by baseline information using the chi-squared test. To study the relations between the variables, we first examined the association of three groups of potential factors---individual, social and illness factors---associated with the adherence to each type of exercise using univariate and multivariate logistic regression analyses. In the univariate analysis, associations were tested for a significant relationship (*p* \< 0.05) with adherence. In the multivariate analysis, the factors with a significant univariate contribution (*p* \< 0.10) were combined in a total model for each type of exercise. The parent's age and gender were always included as independent variables due to their social and psychological relevance. These final models were produced by a process of backwards elimination of independent variables. This procedure consists of dropping an independent variable using the likelihood ratio test at a significance level of *p* = 0.05. The goodness-of-fit and regression for the reduced model were assessed using the methods described by Hosmer et al. \[[@B19-jcm-08-00456]\]. Odds ratios (OR) and 95% CI are reported. In a second stage, a multivariate logistic regression analysis of professionals' behaviour during clinical encounters was used to determine its association with adherence to each type of exercise. Each behaviour and significant variables of the final multivariate first-stage models were introduced as independent variables. Odds ratios and 95% CI are reported for each behaviour in its respective model. The minimal sample size required (*n* = 96) was considered enough for a maximum of five factors in the model, using the recommendation of 10 events (adherent subjects) per variable \[[@B20-jcm-08-00456]\] and considering an estimated 50% of adherence. 3. Results {#sec3-jcm-08-00456} ========== 3.1. Response Rate and Sample Characteristics {#sec3dot1-jcm-08-00456} --------------------------------------------- We identified 393 eligible parents from 18 early intervention centres. Of these, 56% returned the questionnaire. Most respondents were women (91%), and most were living with a partner (91%). [Table 1](#jcm-08-00456-t001){ref-type="table"} shows the characteristics of the sample. Respondents and non-respondents to the questionnaire did not significantly differ in gender (*p* = 0.850), age (*p* = 0.625), education level (*p* = 0.785), marital status (*p* = 0.584), work status (*p* = 0.233) and number of children (*p* = 0.326). Most children were boys (64%) aged between 0.5 and 6 years. 47% were two years old or younger. The most common diagnosis was cerebral palsy (24%) and congenital or hereditary diseases (20%). Almost 70% of children had been receiving early intervention care for less than two years. Most children (94%) received a HEP based on more than one exercise included in the checklist, 51% received prescriptions for flexibility exercises, 70% were prescribed NDT, and 72% were prescribed BMPS exercises. Some parents failed to respond to the adherence questions regarding flexibility, NDT, and BMPS exercises. Thus, the adherence rates were calculated on 110, 140 and 144 parents, for each of these exercise types, respectively. 3.2. Adherence Rates and Associated Factors {#sec3dot2-jcm-08-00456} ------------------------------------------- The percentage of parents who adhered to the flexibility exercises was 34%, 50% of parents adhered to the NDT exercises, and 54.2% adhered to BMPS. There was a significant difference between the flexibility and BMPS exercises. The results of the univariate analyses are presented in [Table 2](#jcm-08-00456-t002){ref-type="table"}. Based on these results, eight factors were entered into the multivariate model of the flexibility exercises; three factors were entered into the model of the NDT exercises; and five into the model of the BMPS exercises. The significant factors in the multivariate models are also presented in [Table 3](#jcm-08-00456-t003){ref-type="table"}. The model of the flexibility exercises identified knowledge and ability as independent factors. Based on this model, high knowledge increases adherence to flexibility exercises. According to the model of the NDT exercises, a low perception of barriers and high self-efficacy increase the odds of exercise adherence. Finally, according to the BMPS model, the odds of adherence increase when parents have a low perception of barriers and high self-efficacy, as well as when children are unable to maintain an upright sitting position. Several professional behaviours were identified as significant factors of adherence, although they varied across the different types of exercises. An association was found between the adherence to flexibility exercises and 'giving information about evolution', 'justifying usefulness of treatments', 'using the child as a model during instruction of exercises' and 'high satisfaction'. Adherence to the NDT exercises was associated with only one professional behaviour: 'justifying usefulness of treatment'. Adherence to the BMPS exercises was also associated with only one behaviour: 'asking about adherence at home'. 4. Discussion {#sec4-jcm-08-00456} ============= We examined parents' adherence to three different types of exercises included in HEPs and its relationship with factors relating to the parents, the child, the environment and the professional. The rates of parents' adherence to the NDT and BMPS exercises were similar, although the adherence rates for flexibility exercises was lower. This variability among the types of exercises is consistent with the variability reported among therapeutic modalities of medical treatments for children with chronic illnesses \[[@B8-jcm-08-00456],[@B21-jcm-08-00456]\]. In previous studies developed in paediatric populations \[[@B5-jcm-08-00456],[@B22-jcm-08-00456]\], in which the adherence to HEPs was measured as a whole and not separately by exercises, we found similar rates to those obtained for the NDT and BMPS exercises in our study, but higher than our adherence rates for the flexibility exercises. The perception of the overall adherence to HEP, indeed, could overestimate the actual adherence for any specific part of the HEP \[[@B16-jcm-08-00456]\]. The associated factors of the adherence to the flexibility exercises were different and more numerous than those associated with others. Adherence to the three types was associated to parental cognitive factors. Nevertheless, whereas the perception of barriers to integrate exercises into daily routine and self-efficacy were common to both the NDT and BMPS exercises, parental knowledge and ability was only associated with flexibility exercises. This may be explained by the fact that these exercises require the use of more manual skills from the parents. The relevance of barriers as an obstacle for adherence is consistent with other studies \[[@B6-jcm-08-00456],[@B23-jcm-08-00456]\], which showed that caregivers only performed home programs if these were easy to integrate into daily family functioning. Self-efficacy has also been identified as critical to sustain the effort of parents with regular advice \[[@B24-jcm-08-00456],[@B25-jcm-08-00456]\]. Moreover, it has been related to adherence to different components of HEP in adults with chronic conditions \[[@B26-jcm-08-00456]\]. We found an inverse relation between the adherence to the BMPS exercises and the child's sitting function, whereas previous studies conducted on children with chronic illness identified that adherence may be lower where there are more functional problems \[[@B27-jcm-08-00456],[@B28-jcm-08-00456]\]. This contradictory effect of functional problems may be because the achievement of specific developmental stages (e.g., sitting) may be more relevant for parents of children with developmental disabilities compared to those with chronic illness. The strong relevance of professional behaviours on the flexibility exercises compared to other exercises studied, in our opinion, is consistent with our finding that parents' knowledge and abilities were strongly associated with the behaviours of professionals. Furthermore, parents usually identify the flexibility exercises as a painful \[[@B29-jcm-08-00456],[@B30-jcm-08-00456]\] and complex \[[@B7-jcm-08-00456]\] activity. This can lead to an inadequate performance of the exercise, different from what the therapist recommended \[[@B6-jcm-08-00456],[@B11-jcm-08-00456]\], or to a reduction in their adherence to this kind of exercises, as our study shows. Thus, it is reasonable to think that when therapists focus on improving parents' knowledge and abilities, by providing information \[[@B23-jcm-08-00456]\], justifying the usefulness of exercises \[[@B2-jcm-08-00456],[@B3-jcm-08-00456]\] and using the child as a model \[[@B3-jcm-08-00456],[@B14-jcm-08-00456]\], this has a positive effect on adherence. It would be interesting for therapists to consider their influence on adherence to these types of exercises when they plan their interventions with families. Understandably, as revealed by studies on the overall adherence to HEPs \[[@B11-jcm-08-00456]\], the variables related to the professional's interaction and communication with parents are key determinants. In our study, when the parents received information regarding the usefulness of exercises, this significantly increased their odds of adherence both for the flexibility and for BMPS exercises. This could be due to the understanding of a cause-effect relationship, thus improving the adherence rates, especially for those exercises which parents found more difficult to perform. Moreover, in such conditions, where the results from treatments are not immediately observed, a good relationship and communication between parents and therapists could have an important role in the performance of treatments \[[@B11-jcm-08-00456]\]. In line with the study by Novak \[[@B2-jcm-08-00456]\], the follow-up of treatments and asking about their performance at home appears to be a strong predictor of adherence to the NDT exercises. Thanks to this follow-up, parents can be invited to explain their difficulties and obtain support from the therapists to overcome barriers and be able to integrate these exercises into their daily routine. However, this follow-up can be perceived in some cases as a source of stress, creating feelings of guilt in parents \[[@B18-jcm-08-00456]\]. Therapists must be aware of this situation, in order to be seen as a source of help instead of as a judgmental onlooker. Other factors, such as the parents' satisfaction with therapists, should be an additional objective, as this was strongly associated with adherence to the flexibility exercises. Previous studies have also found that satisfaction is relevant to the overall adherence to HEP \[[@B18-jcm-08-00456],[@B31-jcm-08-00456],[@B32-jcm-08-00456]\]. To summarise, the results suggest that adherence is a multidimensional construct, which could be different for each component of the HEP. In fact, parents appreciate that there are different levels of difficulty to be learned between the exercise types \[[@B7-jcm-08-00456],[@B14-jcm-08-00456],[@B18-jcm-08-00456]\]. Therefore, therapists should not categorize parents as being adherent or non-adherent, but should find out their performance for each part of the HEP and their difficulties and barriers for their implementation \[[@B33-jcm-08-00456]\]. More specifically, therapists should be especially attentive when prescribing exercises that parents consider difficult, such as flexibility exercises, or when encountering parents who perceive themselves as poorly skilled. It may therefore be advisable for therapists to spend the time required to make sure that the parents get the skills needed to perform the exercises. It would be recommended to use the child as a model when parents are learning the exercises, especially when these are perceived as difficult. Therapists are also encouraged to inform parents about the usefulness of exercises, as it may not be clear in many types of exercises, also with regards to the child's evolution. They should also conduct regular follow-ups for the HEP and be aware of parental adherence and difficulties. Our findings should be interpreted in light of our study's methodological limitations. First, the cross-sectional design did not allow us to make conclusions about causal effects. Longitudinal studies are needed to draw definitive causal effects and confirm the presence or absence of associations. Second, parents' behaviours of adherence were self-reported. Although self-reported adherence has demonstrated adequate validity \[[@B34-jcm-08-00456]\], reported levels are the respondents' perception of their adherence, and are not necessarily a true reflection of actual events. It is therefore possible that our results represent an overly optimistic view of the extent of adherence to HEP. This study also relied on self-reported data regarding health professionals' behaviours during therapeutic encounters. Third, our questionnaire of parents' adherence was developed by integrating items from a wide variety of parent-oriented instruments in order to facilitate external validity. The selected items were pilot-tested with parents via cognitive pretesting (unpublished). Parents were asked to report on the relevance and understanding of each item, and all retained items were clearly and well understood by parents. We did not perform an in-depth psychometric evaluation because our study did not use scales from these items; we only examined and verified the internal consistency for the two items used as the scale (knowledge and ability to carry out the home program). Nevertheless, it would be desirable to analyse the factor structure and stability over time in order to ensure the internal validity, especially for questions that are not exhaustive and/or contextual enough (for example, the perceived barriers to integrate exercises into the daily routine, and self-efficacy). Fourth, we initially tested each initial predictor individually with the response and then fit a final multivariate model, only using the variables which were significant for the univariate analysis. This strategy is common and well established, particularly in health sciences; however, it may have statistical consequences, as sometimes insignificant variables in the bi-variate analysis can become significant in a complex multivariate analysis. Fifth, we did not use interaction terms in our multivariate model. Adding interaction terms to a regression model can greatly expand the understanding of the relationships among the variables, and it enables the testing of whether the effect of one predictor variable on the response variable is different at different values of the other predictor variable. Nevertheless, for this study we selected this option because we were only interested in identifying factors associated to adherence. Finally, our sample was recruited from children with or at risk of development disorders attending early intervention centres, who may differ from other children who are in other settings such as hospitals or paediatric clinics. Until further research is conducted, our results should be generalised with caution. 5. Conclusions {#sec5-jcm-08-00456} ============== The rate of adherence between types of exercises is similar, with the exception of the flexibility exercises with lower rates of adherence. Parents' perceptions of barriers to integrating exercises into their daily routine and self-efficacy contribute significantly to adherence for the NDT and BMPS exercises, and parents' knowledge and ability to do exercises for flexibility. Physicians and therapists can influence the adherence for all exercises by providing information and follow-up on adherence during clinical encounters and, specifically for the flexibility exercises, by using good practice during exercise instruction. Our findings reinforce the need to consider components of the HEP in the management and study of adherence to HEP. Further research is required to identify how interventions to promote adherence to each component contribute to improved clinical outcomes. The authors thank the children, parents and physiotherapists of the early intervention centres for their cooperation during the course of this study. Conceptualization, F.M.-M., C.L.-N. and J.M.-H.; methodology, C.L.-N., S.L.O.-S. and P.E.-R.; software, F.M.-M., C.L.-N., S.L.O.-S. and J.A.G.-V.; formal analysis, F.M.-M., C.L.-N., S.L.O.-S., P.E.-R. and J.A.G.-V.; writing---original draft preparation, C.L.-N., F.M.-M. and P.E.-R.; writing---review and editing, F.M.-M., C.L.-N., J.M.-H., P.E.-R., S.L.O.-S. and J.A.G.-V.; project administration, J.M.-H., F.M.-M. and P.E.-R.; funding acquisition, F.M.-M. and J.M.-H. This research was funded by Ministry of Health and Consumer Affairs, Spain, grant number PI052418. The authors declare no conflict of interest. jcm-08-00456-t0A1_Table A1 ###### Questionnaire of Parents' Adherence to Home Exercise Programs Item scoring, and scales for Questionnaire of Parents' Adherence to Home Exercise Programs. † ----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- **REGARDING YOUR EXPERIENCE WITH THE HOME EXERCISE PROGRAM** 1\. Does the home program fit your daily routine? 2\. I understand my child's home program 3\. I am skilful in carrying out the home program 4\. How confident are you about performing the home program? 5\. My partner supports me at home 6\. I have the equipment required to do the exercises at home **REGARDING THE THERAPIST'S INVOLVEMENT** 7\. The physiotherapist gives me information regarding my child's progress 8\. The physiotherapist justifies the usefulness of the exercises 9\. The physiotherapist gives me written instructions explaining the exercises 10\. The physiotherapist explains the exercises to me using the child as a model 11\. The physiotherapist gives me advice on how to include exercises into daily routines 12\. The physiotherapist regularly checks my skill at performing the exercises 13\. The physiotherapist usually asks me about my adherence to the exercises at home 14\. If you had to mark from 1 to 10 your overall satisfaction with the attention from your physiotherapist, what would your score be? **ADHERENCE BEHAVIOURS** 15\. Currently, which of these kinds of exercises have been recommended to you to do at home with your child? □Flexibility exercises (muscle lengthening, range of motion, stretching)□Neuromotor development training (head control, crawling, rolling, walking...)□Posture control training (positioning the child with or without external postural support in different positions, such as lying, sitting, standing...) 16\. Many parents find it difficult to do the exercises at home. In your case, how often do you generally do each of the exercises with your child in accordance with the prescribed dosage?□Flexibility exercises (muscle lengthening, range of motion, stretching)□Neuromotor development training (head control, crawling, rolling, walking...)□Posture control training (positioning the child with or without external postural support in different positions, such as lying, sitting, standing...) ----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- † Item statements are presented in the order in which they appeared in the questionnaire. However, the style of the questionnaire is not reproduced here. Scoring was based on a 5-point Likert scale for items 1, 5, 7--13, 16 (5 always, 4 very often, 3 sometimes, 2 rarely, 1 never), and items 2, 3 (5 strongly agree, 4 agree, 3 undecided, 2 disagree, 1 strongly disagree). Item 4 was scored with a 0--10 point scale (10 very confident--0 not confident at all). Item 6 (1 yes, 2 no). Item 14 was scored with a 0--10 point scale (10 very satisfied--0 very unsatisfied). Item 15 (recommended/ not recommended). jcm-08-00456-t001_Table 1 ###### Characteristics of the respondents (*n* = 219). Variables N \% ----------------------------------------- ----- ------ **Gender** Male 20 9.1 Female 199 90.9 **Age (years)** 20--30 47 21.5 31--40 143 65.3 \>40 29 13.2 **Educational level** Without studies or with primary studies 125 57.1 Secondary studies or university studies 94 42.9 **Marital status** With a partner 199 90.9 Without a partner 20 9.1 **Work status** Homemaker 114 52.1 Employed 82 37.4 Unemployed 20 9.1 Student 3 1.4 **Number of children** 1 86 39.3 2 93 42.5 ≥3 40 18.2 **Child's Gender** Male 140 63.9 Female 79 36.1 **Child's Age** 6 months--2 years 102 46.6 \>2 years 117 53.4 **Health condition** Cerebral Palsy 52 23.7 Congenital illness \* 39 17.9 Other \*\* 128 58.4 **Time in treatment** \<2 years 156 71.2 ≥2 years 63 28.8 \* Includes muscular torticollis, spinal atrophy, congenital arthrogryposis multiplex and chromosome motor disorder. \*\* Includes developmental delay (unspecified), obstetric brachial plexus palsy and encephalopathy in premature babies. jcm-08-00456-t002_Table 2 ###### Odds Ratio (95% Confidence Interval) of predictive factors of adherence to specific types of exercises featured in the home exercise program. Adherence to Flexibility Exercises Adherence to Neuromotor Development Training Exercises (NDT) Adherence to Body Mechanics and Postural Stabilization Exercises (BMPS) ----------------------------------------------------------------------------- ------------------------------------ -------------------------------------------------------------- ------------------------------------------------------------------------- ------------------------ ------------------------- ------------------------- **Parental and Environmental Characteristics** **Sociodemographics** Age (*n1* = 100) (*n2* = 148) (*n3* = 144) 0.96 \[0.90--1.03\] 0.99 \[0.94--1.05\] 0.98 \[0.93--1.04\] Gender Female (*n1* = 100) (*n2* = 148) (*n3* = 144) 1.93 \[0.57--6.51\] 0.55 \[0.15--1.95\] 1.01 \[0.32--3.18\] With secondary or university studies (*n1* = 100) (*n2* = 148) (*n3* = 144) 0.40 \[0.17--0.95\] \* 0.71 \[0.36--1.38\] 0.84 \[0.43--1.62\] With couple (*n1* = 100) (*n2* = 148) (*n3* = 144) 0.76 \[0.22--2.68\] 0.73 \[0.24--2.22\] 0.88 \[0.29--2.66\] Work participation (*n1* = 100) (*n2* = 148) (*n3* = 144) 0.46 \[0.20--1.05\] † 1.14 \[0.56--2.31\] 1.00 \[0.50--2.00\] Children number (*n1* = 100) (*n2* = 148) (*n3* = 144) 1 \- \- \- 2 0.61 \[0.25--1.50\] 0.82 \[0.40--1.69\] 0.80 \[0.39--1.65\] 3 and more 0.90 \[0.28--2.91\] 0.47 \[0.17--1.28\] 0.43 \[0.16--1.16\] **Cognitive** Low perception of barriers (*n1* = 98) (*n2* = 148) (*n3* = 142) 2.87 \[1.20--6.86\] \* 2.77 \[1.38--5.55\] \*\* 2.47 \[1.16--5.23\] \* 3.08 \[1.5--6.15\] \*\* 2.52 \[1.15--5.55\] \* High knowledge and ability (*n1* = 100) (*n2* = 148) (*n3* = 144) 5.68 \[2.2--14.87\] \*\* 3.96 \[1.35--11.6\] \*\* 1.68 \[0.86--3.28\] 2.59 \[1.3--5.11\] \*\* High self-efficacy (*n1* = 98) (*n2* = 146) (*n3* = 142) 3.28 \[1.29--8.32\] \*\* 2.62 \[1.31--5.24\] \*\* 2.21 \[1.05--4.66\] \* 3.98 \[1.9--8.28\] \*\* 3.69 \[1.6--8.34\] \*\* **Environmental factors** Support by partner at home (*n1* = 100) (*n2* = 147) (*n3* = 144) 0.95 \[0.42--2.16\] 1.68 \[0.86--3.27\] 0.88 \[0.46--1.69\] Having home equipment (*n1* = 100) (*n2* = 147) (*n3* = 144) 1.93 \[0.79--4.73\] 1.35 \[0.63--2.89\] 1.45 \[0.69--3.05\] **Child's Characteristics** **Demographics** Gender female (*n1* = 100) (*n2* = 148) (*n3* = 143) 1.01 \[0.44--2.33\] 1.16 \[0.58--2.30\] 1.67 \[0.83--3.36\] Age: \<2 years old (*n1* = 100) (*n2* = 148) (*n3* = 144) 1.16 \[0.51--2.65\] 1.51 \[0.77--2.96\] 1.05 \[0.54--2.05\] **Health Related Factors** \- Health Condition (*n1* = 87) (*n2* = 130) (*n3* = 129) Cerebral Palsy \- \- \- Congenital illness 2.25 \[0.58--8.74\] 0.85 \[0.28--2.52\] 1.17 \[0.41--3.34\] Other 1.55 \[0.58--4.15\] 0.82 \[0.33--2.08\] 1.27 \[0.53--3.00\] \- Functioning Unable to maintain sitting (*n1* = 99) (*n2* = 147) (*n3* = 143) 1.02 \[0.40--2.63\] 1.19 \[0.55--2.58\] 2.12 \[1.00--4.50\] \* 2.65 \[1.09--6.43\] \* Unable to crawl (*n1* = 98) (*n2* = 145) (*n3* = 141) 1.89 \[0.81--4.38\] 0.72 \[0.37--1.42\] 1.06 \[0.55--2.06\] Unable to walk (*n1* = 100) (*n2* = 148) (*n3* = 144) 1.27 \[0.56--2.87\] 1.27 \[0.65--2.48\] 1.02 \[0.52--2.02\] \- In treatment \< 2 years (*n1* = 99) (*n2* = 147) (*n3* = 143) 2.17 \[0.87--5.42\]† 2.20 \[0.94--5.16\]† 2.14 \[0.9--5.12\] † †: *p* \< 0.10; \*: *p* \< 0.05; \*\*: *p* \< 0.01. jcm-08-00456-t003_Table 3 ###### Adjusted Odds Ratio of professionals' behaviours predictive of adherence to specific types of exercises featured in the home exercise program. ----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- Adherence to Flexibility Exercises Adherence to Neuromotor Development Training Exercises (NDT) Adherence to Body Mechanics and Postural Stabilization Exercises (BMPS) ----------------------------------------------------------------------------------------- ------------------------------------ -------------------------------------------------------------- ------------------------------------------------------------------------- **Professionals' behaviours** **Giving information to parents** Giving information about evolution\ 6.27 \[1.26--31.16\] \* 2.62 \[0.59--11.55\] 1.14 \[0.26--5.01\] (*n1* = 99) (*n2* = 139) (*n3* = 135) Justifying usefulness of exercises\ 9.49 \[2.74--32.9\] \*\* 2.29 \[0.95--5.53\] 3.00 \[1.20--7.48\] \* (*n1* = 97) (*n2* = 141) (*n3* = 137) **Instructions for exercises** Giving written instructions\ 1.43 \[0.48--4.25\] 1.82 \[0.82--4.08\] 1.46 \[0.62--3.43\] (*n1* = 96) (*n2* = 136) (*n3* = 134) Using the child as a model\ 2.72 \[1.03--7.15\] \* 2.06 \[0.92--4.59\] 0.99 \[0.42--2.32\] (*n1* = 99) (*n2* = 141) (*n3* = 137) Giving advice to insert into daily routines (*n1* = 97) (*n2* = 140) (*n3* = 136) 1.38 \[0.54--3.56\] 1.22 \[0.60--2.50\] 1.01 \[0.47--2.16\] **Follow-up of treatments** Checking skills (*n1* = 97) (*n2* = 135) (*n3* = 133) 1.55 \[0.58--4.15\] 1.27 \[0.62--2.60\] 1.07 \[0.49--2.33\] Asking about adherence at home\ 1.83 \[0.70--4.77\] 2.98 \[1.38--6.41\] \*\* 1.42 \[0.65--3.13\] (*n1* = 99) (*n2* = 139) (*n3* = 137) Parents' high satisfaction with the care received (*n1* = 99) (*n2* = 143) (*n3* = 139) 1.71 \[1.11--2.64\] \* 1.14 \[0.88--1.48\] 1.18 \[0.87--1.60\] ----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- \*: *p* \< 0.05; \*\*: *p* \< 0.01

Introduction ============ Prostate cancer and male infertility are both very common disorders, affecting approximately 10% and 8%, respectively, of all men in Western societies.[@ref1] [@ref2] As prostate cancer and many forms of infertility are androgen related, the possible link between these disorders has been investigated previously. Three American studies have reported an increased risk of prostate cancer in men with impaired semen quality,[@ref3] [@ref4] [@ref5] whereas three Scandinavian studies and one American study indicated a lower risk of prostate cancer in childless men.[@ref6] [@ref7] [@ref8] [@ref9] This finding was recently confirmed in a meta-analysis summarising 10 individual studies.[@ref10] Others found no relation between fatherhood and the risk of prostate cancer.[@ref11] [@ref12] [@ref13] [@ref14] Neither fatherhood status nor sperm parameters represent ideal markers of male infertility. Whereas childlessness may be related to the female partner's subfertility, the opportunity to start a family, or personal choice, and thus attributable to social rather than biological factors, semen parameters such as sperm concentration, morphology, and motility are subject to both laboratory and intra-individual variation. Nevertheless, for many infertile men, fatherhood is still possible through use of powerful assisted reproductive techniques. For men with very few sperm (oligozoospermia), or spermatozoa with poor progressive motility (asthenozoospermia), the only option for fatherhood is intracytoplasmic sperm injection (ICSI), in which a sperm is injected into an egg and the embryo put back into the uterus. This is also the only possibility for men with azoospermia, from whom gametes can be microsurgically gathered from the epididymis or testis. For men who do not have such severely deficient spermatogenesis, conventional in vitro fertilisation (IVF), in which sperm are allowed to fertilise retrieved oocytes in a laboratory dish, is the standard procedure. In Sweden, ICSI treatment is mainly used in cases with significantly impaired semen quality, with male factor infertility being the major indication for ICSI.[@ref15] [@ref16] The objective of this study was to use compulsory national registries containing information on prostate cancer diagnoses and infertility treatments to investigate whether the risk of prostate cancer in men who became fathers through IVF or ICSI, reflecting the grade of hampered spermatogenesis, differed in terms of incidence, age at onset, and, where applicable, severity from men who achieved fatherhood naturally. Such information could be important for defining clinical routines for follow-up of men undergoing fertility treatment. Methods ======= Registers and study populations ------------------------------- We retrieved data on all children born alive in Sweden during the period 1994-2014 (n=2 108 569), as well as their fathers, from the Swedish Medical Birth Register and the Swedish Multi-generation Register. We excluded children with missing paternal identification numbers. We matched the remaining records with the Swedish National Quality Register for Assisted Reproduction. The Swedish Cancer Registry, the Swedish Register of Education, and the Swedish Cause of Death Register supplied the paternal prostate cancer diagnoses, paternal education data, and date of death, respectively. Reporting of cancer diagnoses to the national Swedish Cancer Registry is mandated by law for all newly diagnosed cancers, with an approximated completeness of 96%,[@ref17] ensuring a complete assessment of prostate cancer diagnoses. Similarly, reporting of fertility treatments is mandatory, in both private and public clinics, with coverage close to 100%. Data on patients undergoing intrauterine insemination, which is an uncommon procedure in Sweden, was not collected. Thus, in this paper, assisted reproductive techniques refers to ICSI and IVF. As ICSI was first used in 1992, virtually all fathers who conceived through ICSI in Sweden are likely to be included in our cohort. To avoid bias introduced by fathers being counted multiple times, we paired the birth record of the first child born within the cohort interval with the father. This resulted in 1 181 490 children born to the same number of fathers ([fig 1](#f1){ref-type="fig"}). Prostate cancer was defined according to ICD-7 (international classification of diseases, 7th revision) diagnosis code (177) and early onset prostate cancer according to European Association of Urology guidelines[@ref18] (that is, diagnosed before the age of 55). {#f1} Using data from the Swedish Prescribed Drug Register, available from July 2005 to November 2016, we identified men who had received androgen deprivation therapy. This is indicated only in cases of locally advanced or metastatic prostate cancer and not in low risk malignancies.[@ref19] Thus, androgen deprivation therapy can act as a proxy for the severity and clinical significance of malignancy. We identified men receiving androgen deprivation therapy, with the date of their first prescription, by parsing for the following drugs: abiraterone acetate, buserelin, cyproterone acetate, degarelix, enzalutamide, flutamide, goserelin, histrelin acetate, leuprorelin, megestrol acetate, nilutamide, and triptorelin. As receiving testosterone replacement therapy has an unknown effect on risk of prostate cancer, we excluded these men identified by data from the Swedish Prescribed Drug Register in a sensitivity analysis. Prescription of testosterone at any time led to exclusion (336 ICSI treated fathers, 212 IVF treated fathers, and 7495 reference fathers). Statistical analysis -------------------- We grouped the fathers according to mode of conception of their child; ICSI, IVF, or natural conception (reference group). We constructed Kaplan-Meier survival curves stratified on the aforementioned groups, with accompanying log rank tests. We used Cox regression to estimate hazard ratios. In the Cox regressions analyses, we corrected for paternal age by adjusting for fathers' age at childbirth (continuous). We followed the fathers from conception of the child until diagnosis of prostate cancer, death, or end of follow-up (31 December 2014). We estimated the date of conception by using gestational length data from the Medical Birth Register. To adjust for socioeconomic status, the Cox model was adjusted for the father's education level (years of formal education, categorical: ≤10, 11-14, ≥15, or missing data). We tested the assumption of proportionality of hazards by the significance level of the interaction between prostate cancer and the natural logarithm of time within the full Cox regression model with all covariates. We further investigated any evidence of non-proportionality (P\<0.05) by estimating hazard ratios for restricted time intervals. To investigate whether men achieving fatherhood by assisted means had an altered risk of early onset prostate cancer, we did an analysis in which we defined an event as prostate cancer diagnosed before age 55. Follow-up was as above, with the fathers being right censored when they reached age 55. This analysis was adjusted for the same covariates as before (paternal age and paternal education level). We also combined the fathers treated with ICSI and those treated with IVF into one group so that a combined risk estimate could be obtained for men becoming fathers through assisted reproduction techniques. As the follow-up for each father started from conception of the child, men with a prostate cancer diagnosis before that point were excluded, leading to nine ICSI treated fathers, one IVF treated father, and 28 naturally conceiving fathers being excluded from the analyses. However, as cancer treatment may cause subsequent fertility problems, we did a separate sensitivity analysis in which fathers who had been diagnosed as having any cancer (ICD-7: 140-207.9) before child conception, not only prostate cancer, were also excluded (ICSI, n=451; IVF, n=171; natural conception, n=5179). As ICSI is indicated in more severe forms of male infertility (azoospermia, severe oligozoospermia) and IVF is used in female infertility, combined with mild or no male infertility,[@ref15] [@ref16] we tested for a trend between level of infertility and prostate cancer. This assumed an equidistant stepwise function for the level of infertility among fathers conceiving naturally, via IVF, or via ICSI (continuous variable coded: natural conception=0, IVF=1, ICSI=2). Among all fathers with prostate cancer, we compared the fathers who conceived through ICSI and IVF with those who conceived naturally, to estimate the risk for receiving androgen deprivation therapy after diagnosis of prostate cancer. As prescriptions for androgen deprivation therapy could not be ascertained before July 2005, only prostate cancer diagnoses after this date were included in this analysis. After these exclusions 52 ICSI treated, 68 IVF treated, and 2967 reference fathers remained. This analysis also serves to detect overdiagnosis of clinically insignificant cases of prostate cancer among men undergoing assisted reproduction, owing to their contacts with the healthcare system. This potential bias would likely lead to these men being diagnosed as having prostate cancer at an earlier age, with lower grade, and therefore generally not being treated with androgen deprivation therapy. Conversely, observing an equal or higher risk of androgen deprivation therapy for men undergoing assisted reproduction would indicate no such bias. For this analysis, we constructed a binary logistic regression model, adjusted for the father's age at prostate cancer diagnosis (continuous) and paternal educational level, yielding odds ratios, with an odds ratio below 1 indicating possible bias resulting in more diagnoses of low grade prostate cancer among the assisted reproduction groups---for example, due to better access to prostate specific antigen screening. Conversely, an odds ratio of 1 or above points to an increase in prevalence of clinically relevant prostate cancer, which would likely be diagnosed regardless of whether those men were in contact with the healthcare system. We also did sensitivity analyses in which men receiving testosterone were excluded. We calculated risk estimates for prostate cancer and for early onset prostate cancer by using the same Cox regression method as above. We used SPSS version 25 and R version 3.5.0 with the ggplot2 package for statistical analyses. All analyses were two sided, and we defined P\<0.05 as statistically significant. Patient and public involvement ------------------------------ No patients were involved in the design of this study, nor were any patients involved in the implementation of it or consulted on the reporting of the results. There are no plans to disseminate the results to the research cohort or to relevant patient communities. Results ======= Of 1 181 490 fathers, 20 618 (1.7%) had undergone IVF during the study period, 14 882 (1.3%) had undergone ICSI, and 1 145 990 (97.0%) had conceived children by natural conception. [Table 1](#tbl1){ref-type="table"} shows characteristics of the fathers. The total follow-up time was 14 389 198 person years. The mean age at childbirth of both IVF and ICSI treated fathers was 37 years, whereas the fathers who conceived naturally were 4 years younger on average. Of the men who sired pregnancies naturally, 3244 (0.28%) were diagnosed as having prostate cancer, compared with 77 (0.37%) in the IVF group and 63 (0.42%) in the ICSI group. After exclusion of cases of prostate cancer before conception, 3216 (0.28%), 76 (0.37%), and 54 (0.36%) had been diagnosed as having prostate cancer in the reference, IVF, and ICSI groups. The ICSI treated fathers had the youngest mean age of onset (55.4 years), on average almost 2 years younger than the reference group. ###### Characteristics of fathers who conceived offspring naturally, through intracytoplasmic sperm injection (ICSI), and through in vitro fertilisation (IVF). Values are numbers (percentages) unless stated otherwise Characteristics Fathers conceiving naturally (n=1 145 990; 97.0%) Fathers conceiving by IVF (n=20 618; 1.7%) Fathers conceiving by ICSI (n=14 882; 1.3%) --------------------------------------------------------------------- --------------------------------------------------- -------------------------------------------- --------------------------------------------- Mean (SD) age at birth of child, years 32.5 (6.2) 36.6 (5.3) 36.9 (6.0) Mean (SD) age at end of follow-up\*, years 44.0 (9.0) 45.9 (8.1) 45.2 (7.9) Years of formal education:  \<10 139 012 (12.1) 1549 (7.5) 1205 (8.1)  10-14 648 789 (56.6) 10 631 (51.6) 7862 (52.8)  ≥15 348 081 (30.4) 8375 (40.6) 5754 (38.7)  Missing 10 108 (0.9) 63 (0.3) 61 (0.4) All prostate cancer:  Fathers with prostate cancer 3244 (0.3) 77 (0.4) 63 (0.4)  Mean (SD) age at diagnosis, years 57.1 (6.9) 55.9 (5.9) 55.1 (7.0)  Early onset prostate cancer (diagnosis age \<55) 1274 (39.3) 39 (51) 29 (46) Prostate cancer, excluding cases occurring before child conception:  Fathers with prostate cancer 3216 (0.3) 76 (0.4) 54 (0.4)  Mean (SD) age at diagnosis, years 57.2 (6.9) 56.1 (5.8) 55.4 (6.7)  Mean (SD) time between child conception and diagnosis, years 14.5 (4.7) 13.3 (5.2) 9.5 (4.6)  Early onset prostate cancer (diagnosis age \<55) 1257 (39.1) 38 (50) 25 (46) Androgen deprivation therapy†:  Fathers treated with androgen deprivation therapy 387§/2967 (13.0) 8/68 (12) 10/52 (19)  Mean (SD) age at start of therapy, years 60.2 (7.6) 57.0 (4.6) 56.4 (8.8)  Early onset prostate cancer (diagnosis age \<55 years) 105 (8.4) 3 (8) 5 (20) End of follow-up in Cancer Registry (31 December 2014). Deaths not accounted for. Patients receiving androgen deprivation therapy within 1 year of prostate cancer diagnosis. Excluding cases occurring before child conception or diagnosed before 1 July 2005. Including 12 cases in which androgen deprivation therapy was prescribed in the week preceding prostate cancer diagnosis date. In the entire cohort, men who had undergone ICSI treatment had a statistically significantly increased risk of prostate cancer (hazard ratio 1.64, 95% confidence interval 1.25 to 2.15; P\<0.001) compared with natural conception ([table 2](#tbl2){ref-type="table"}; [fig 2](#f2){ref-type="fig"}). IVF fathers also had a statistically significantly increased risk, but of a lesser magnitude (hazard ratio 1.33, 1.06 to 1.66; P=0.02) compared with natural conception. We detected evidence of non-proportionality for ICSI, but not IVF, compared with natural conception (interaction with log (time), P=0.01). The adjusted hazard ratios for the first and second halves of follow-up were 2.22 (1.56 to 3.16; P\<0.001) and 1.18 (0.78 to 1.80; P=0.44), respectively. In all other Cox models, the P value for the interaction was greater than or equal to 0.05. ###### Unadjusted and adjusted risk estimates for prostate cancer. Outcome Fathers conceiving by IVF compared with natural conception Fathers conceiving by ICSI compared with natural conception ------------------------------- ------------------------------------------------------------ --------- ------------------------------------------------------------- --------------------- --------- Prostate cancer:  Unadjusted hazard ratio 2.09 (1.66 to 2.62) \<0.001 3.12 (2.38 to 4.08) \<0.001  Adjusted hazard ratio 1.33 (1.06 to 1.66) 0.02 1.64 (1.25 to 2.15) \<0.001 Early onset prostate cancer:  Unadjusted hazard ratio 2.93 (2.12 to 4.05) \<0.001 3.66 (2.46 to 5.44) \<0.001  Adjusted hazard ratio 1.51 (1.09 to 2.08) 0.01 1.86 (1.25 to 2.77) 0.002 Androgen deprivation therapy:  Unadjusted odds ratio 0.89 (0.42 to 1.87) 0.76 1.59 (0.79 to 3.19) 0.19  Adjusted odds ratio 0.99 (0.47 to 2.10) 0.98 1.91 (0.94 to 3.88) 0.07 ICSI=intracytoplasmic sperm injection; IVF=in vitro fertilisation. {#f2} Moreover, both ICSI treated fathers and IVF treated fathers had a statistically significantly increased risk of early onset prostate cancer (hazard ratio 1.86, 1.25 to 2.77 (P=0.002) for ICSI; 1.51, 1.09 to 2.08 (P=0.01) for IVF) ([fig 3](#f3){ref-type="fig"}). [Table 2](#tbl2){ref-type="table"} shows unadjusted risk estimates. {#f3} Men who became fathers through assisted reproduction techniques (combined ICSI and IVF) had a statistically significantly increased risk of prostate cancer (odds ratio 1.44, 1.21 to 1.71; P\<0.001) and early onset prostate cancer (1.63, 1.26 to 1.63, P\<0.001) compared with men achieving fatherhood naturally. When we excluded fathers who were diagnosed as having any cancer before their offspring's conception date, ICSI treated fathers still had a statistically significant increased risk of prostate cancer (hazard ratio 1.70, 1.29 to 2.22; P\<0.001) and of early onset prostate cancer (1.92, 1.29 to 2.86; P=0.001). Similarly, IVF treated fathers retained their increased risk for prostate cancer (hazard ratio 1.30, 1.03 to 1.64; P=0.02) and early onset prostate cancer (1.45, 1.04 to 2.02; P=0.03). We detected a statistically significant trend for the association of level of infertility and prostate cancer (odds ratio~natural,\ IVF,\ ICSI~ 1.29, 1.15 to 1.45; P\<0.001). Similarly, we detected a trend for the association between level of infertility and the risk of early onset prostate cancer (odds ratio~natural,\ IVF,\ ICSI~ 1.40, 1.18 to 1.66; P\<0.001). The ICSI treated fathers with prostate cancer also had the highest rate of receipt of androgen deprivation therapy (19.2%) compared with the reference group (13.0%) and IVF group (11.8%). We saw no indication that men who underwent assisted reproduction were more often diagnosed as having low grade prostate cancer than men who conceived naturally (odds ratio 1.91, 0.94 to 3.88 (P=0.07) for ICSI; 0.99, 0.47 to 2.10 (P=0.98) for IVF). Exclusion of men who received testosterone replacement therapy had a negligible effect on hazard ratios for prostate cancer and for early onset prostate cancer for both the ICSI and IVF groups. Discussion ========== The main conclusion of this study, comprising virtually all men fathering a child in Sweden during two decades, is that men who achieved fatherhood through assisted reproduction had a remarkably high risk of prostate cancer. Fathers who used ICSI had a 60% higher risk and those who used IVF had a 30% higher risk, compared with men who conceived naturally. The increase in risk was most pronounced before the age of 55 years and was still present after exclusion of men who had been treated for any cancer and therefore with potentially gonadotoxic treatments before the child's conception. Strengths and weaknesses of study --------------------------------- The strengths of this study are access to population based datasets and the completeness of the registries. A drawback is the lack of data in the registries on fertility treatment in men who did not succeed in becoming fathers. Hence, the most severe cases may not have been included. However, our results, specifically the trend showing an increasing risk of prostate cancer and of early onset prostate cancer, may indicate that it is the underlying level of infertility that is associated with the man's risk of this malignancy. This might indicate that our results could be generalised to men undergoing assisted reproduction techniques without achieving fatherhood or possibly even to infertile men in general. This would align well with previous studies showing increased risk of prostate cancer for men with poor semen parameters.[@ref3] [@ref4] [@ref5] Furthermore, data on prostate specific antigen were not available. Such data could have provided direct evidence as to whether men undergoing ICSI receive closer follow-up than their counterparts, but the registries do not include this information. Finally, data on men who died from prostate cancer before they had a chance to become fathers are lacking. However, with a mean age of only 37 years for the ICSI/IVF treated fathers and 32 years for those who conceived naturally, the effect of such selection bias can be considered small. Men attaining fatherhood through IVF and ICSI were on average older and also more well educated at the birth of their offspring. However, the presented risk estimates were adjusted for these factors. The evidence of non-proportionality for ICSI treated men indicates a higher risk in the first decade after conception but no elevated risk thereafter. This might be due to an insufficient number of ICSI treated men with long follow-up, because few men were treated with ICSI in the 1990s, yielding uncertain risk estimates. However, as ICSI treated men have an increased risk of early onset prostate cancer, but not of the late onset variety, this is could be reflecting aetiological differences between early and late onset of prostate malignancy.[@ref20] Strengths and weaknesses in relation to other studies ----------------------------------------------------- The reports indicating a lower risk of prostate cancer in childless or infertile men generally included men with a mean age of 60 years or above,[@ref6] [@ref7] [@ref8] [@ref9] [@ref10] [@ref21] meaning that those with early onset and more aggressive prostate cancer may already have died from their disease. Consequently, studies based on follow-up of younger men with fertility problems,[@ref3] [@ref4] [@ref5] including this report, will tend to include cases of early onset prostate cancer. Studies on older men and their risk of prostate cancer may fail to find this risk increase or even come to the opposite conclusion. However, whereas previous studies relied on smaller datasets, sometimes with self reported childlessness and short follow-up, our study is based on more than one million men and up to 20 years of follow-up. Men undergoing assisted reproduction owing to an increased risk of hypogonadism may have been candidates for testosterone replacement therapy and may therefore have been tested for prostate specific antigen more frequently to exclude ongoing prostate cancer.[@ref22] This bias could lead to a detectable increased risk among IVF and ICSI treated fathers. However, excluding men receiving testosterone had a negligible effect on the risk of prostate cancer, indicating no such bias. Furthermore, few men undergoing assisted reproduction are likely to have prostate specific antigen tests, as these are usually administered only to older men. That said, if such testing, being a consequence of these men being in contact with the health service, is the reason why ICSI treated men are diagnosed as having prostate cancer at younger ages, one would expect a decreased administration of androgen deprivation therapy among ICSI treated men, which is the opposite of what is observed. The same would apply to ascertainment bias, to which the outcome of prostate cancer is particularly prone to. However, increased medical attention for men seeking treatment for infertility can explain neither the almost doubled rate of androgen deprivation therapy among ICSI treated fathers nor the trend seen with the level of infertility. Meaning of study: possible explanations and implications for clinicians and policy makers ----------------------------------------------------------------------------------------- This work shows that men fathering children by assisted reproduction are a high risk group for prostate cancer at early ages. As clinicians have for many years noted that early onset prostate cancer is associated with poor prognosis, even before the era of prostate specific antigen screening, men undergoing assisted reproduction may merit further attention and comprise an easily accessible category of patients who may benefit from early screening. Screening by prostate specific antigen testing seems to be the most appropriate, most cost effective, and least invasive first line method for early detection of prostate cancer, and applying it for men treated with assisted reproduction could be beneficial. We would, therefore, welcome studies in other cohorts investigating the risk of prostate cancer in men being treated for infertility and studies on the efficacy and benefits of screening for this risk group, similarly to what is currently offered for other high risk groups. Unanswered questions and future research ---------------------------------------- This study was limited mainly to men with prostate cancer diagnoses early in life. As ICSI has been used only since the 1990s, detecting late onset prostate cancer in these men requires longer follow-up and is a matter for future research. ### What is already known on this topic 1. As prostate cancer and many forms of infertility are androgen related, a possible link between them has been studied, yielding contradictory results 2. Studies in older men using childlessness as a proxy for infertility show that childless men have a lower risk of prostate cancer 3. Studies in younger men, assessing fertility by means of semen parameters, indicate a higher risk of prostate cancer 4. Previous studies have been limited by small numbers of study participants, self reported diagnoses, or short follow-up time ### What this study adds 1. This large register based study show that men fathering children through assisted reproduction have a 30-60% increased risk of prostate cancer compared with men conceiving naturally 2. They have almost twice the risk of developing early onset prostate cancer, before 55 years of age 3. Men fathering children through assisted reproduction seem to be at higher risk for prostate cancer, so the benefits of prostate cancer screening should be considered for this group Contributors: YLG and AG designed and supervised the study. YA, AE, EW, and IS analysed the data. All authors had access to the raw data, interpreted the analysed data, discussed and revised the manuscript, and approved the final version of the manuscript. The corresponding author attests that all listed authors meet authorship criteria and that no others meeting the criteria have been omitted. YLG is the guarantor. Funding: The study was funded by the Swedish Cancer Foundation (CAN 2014/360 and 2017/392), an ALF government grant (F2014/354), the Malmö University Hospital Cancer Research Fund, the Swedish Prostate Cancer Federation, and the European Association of Urology (scholarship EUSP/REPRO-02-2017). Funders had no role in study design, results, or write-up or manuscript submission decisions other than to provide funding. The researchers were independent of the funders. Competing interests: All authors have completed the ICMJE uniform disclosure form at [www.icmje.org/coi_disclosure.pdf](http://www.icmje.org/coi_disclosure.pdf) (available on request from the corresponding author) and declare: support for the submitted work as detailed above; no financial relationships with any organisations that might have an interest in the submitted work in the previous three years; no other relationships or activities that could appear to have influenced the submitted work. Ethical approval: The Regional Ethical Board in Lund, Sweden, approved the study (No 2015/670). Transparency: The lead author (the manuscript's guarantor) affirms that this manuscript is an honest, accurate, and transparent account of the study being reported; that no important aspects of the study have been omitted; and that any discrepancies from the study as planned (and, if relevant, registered) have been explained. Data sharing: Medical researchers can, if conditions are met under Swedish law, obtain de-identified data by contacting the corresponding author.

Background ========== Infections caused by hepatitis B virus (HBV) and hepatitis C virus (HCV) are important health problems worldwide with a high morbidity and mortality \[[@B1]-[@B3]\]. It is known that 400 million people are HBV carriers and that each year 50 million people contact such a disease, resulting in 1--2 million deaths \[[@B3]\]. Infection with HCV occurs worldwide, the prevalence of antibody against HCV (anti-HCV) in serum in most developed countries range between 1% to 2% \[[@B2]\]. It is also reported that currently 170 million people all over the world are infected with HCV infections and that 8000 -- 10000 people die from HCV infection and of complications per year \[[@B4],[@B5]\]. Turkey, located in a region with moderate risk, has a prevalence in a range of 1.0% to 2.4% \[[@B6]\]. The prevalence of HCV infection was reported 1.5% in Turkey \[[@B7]\]. The hepatitis caused by HBV and HCV is extremely obscure. Symptoms appear only in 35.0% of those infected by HBV and in 25.0% of those infected by HCV. However, HBV and HCV are highly transmittable. Therefore, a decision made by World Health Organization (WHO) and National Institute of Health (NIH) suggests that all the patients should be examined with respect to their risk factors for HCV \[[@B8]\]. Turkey falls into the medium endemisity group in terms of HBV and HCV infection prevalence \[[@B3],[@B6]\]. Such infections develop without any clear symptoms, so it is very important to carry good performance by the GPs in terms of diagnosis, treatment and follow up of patients. Therefore we attempted to evaluate the current practices, knowledge, attitudes of GPs regarding viral hepatitis B and C. Methods ======= A cross-sectional study was carried out at all of 32 primary healthcare centers (PHCCs) in the central district of Samsun, Turkey between March 1 and April 31, 2002. The questionnaires were sent to all GPs (n = 160). One hundred and twenty-nine of 160 (83.7%) GPs from different PHCCs completed the survey. The non-respondent GPs randomly distributed in PHCCs. The questionnaire was self-administered anonymously with an answer sheet. Questionnaire \[see [Additional file 1](#S1){ref-type="supplementary-material"}\]. Question format ranged from closed questions with multiple choices to open comment and true/false type. Demographic variables such as gender, age and practice period in PHCCs were assessed. Also the information was collected regarding diagnostic facilities at PHCCs. The survey questionnaire includes a multiple-choice question relevant to sources of medical information. A question in medical knowledge about transmission of hepatitis viruses includes ten statements. GPs were asked to define these statements as correct or incorrect. This question was all marked out of 100. Ten points were given for each correct choice the participants\' marked and incorrect choice they left untouched. Participants were asked: \"What would you suggest to do to protect an infant, born from a HBsAg positive mother?\" The answers were: Injection of hepatitis B vaccine and hepatitis B immunoglobulin (HBIG), offering only hepatitis B vaccine, offering only HBIG, delivery by caesarean section and no knowledge about the problem. The participants were asked what they would suggest to do to protect an infant, born from an anti-HCV and HCV RNA positive mother with multi-choice question. The answers were as follows: There is not any preventive measure, injection of immunoglobulin, delivery by caesarean section, HCV does not transmit from mother to a baby and no knowledge about the problem. Clinical practice was evaluated with the questions in the following: \"What is the approximate number of individuals with acute or chronic viral hepatitis whom you diagnosed or followed up? Approximately how many individuals with acute or chronic viral hepatitis have you seen in the past 12 months?\" Answers were obtained using fill-in-the-blank spaces. Current knowledge about patient management based on four sample case presentations (below) was investigated with multi-choice answers. Sample Cases ------------ ### Case-I A 35 years old woman with chronic hepatitis B admitted to a primary health care center. Her serum alanine aminotransferase (ALT) level was normal. ### Case-II A 40 years old man has been positive for HBsAg (+) for three years and two consecutive ALT results were 125 IU/L and 120 IU/L, respectively. ### Case-III A healthy/asymptomatic 55 years-old man admitted to a primary health care center. His alanine aminotransferase (ALT) level was elevated (150 IU/L) during check-up for life insurance; anti-HCV in ELISA was positive at subsequent work-up. He had a history of intravenous device that he used in 1963. ### Case-IV A healthy/asymtomatic 32 years-old woman admitted to a primary health care center. Her anti-HCV in ELISA was positive blood donation and ALT level was normal at subsequent work-up. She had no risk factors. Shebab et al. \[[@B9]\] originally used the cases that were related to HCV in a survey. GPs were asked: What their next step would be, which additional test they would order, which treatment regimen they would recommend and if they would order a therapy recommended by a specialist, in the view of these cases. In the text the results were given as means ± standard deviations (SD). Results ======= The questionnaire was completed anonymously, of the 160 surveys, 129 (83.7%) were received and investigated using the data analysis. Demographic characteristics --------------------------- Of the study group, 72 (55.8%) GPs were women and 57 (44.2%) were men. The mean age of participants was 31.3 ± 4.8 years. They had been in practice for 7.2 ± 4.6 years. Knowledge about transmission of HBV and HCV ------------------------------------------- Blood transfusion (96.1%, 95.3%), blood and body fluids (95.3%, 94. 6), sexual transmission (89.9%, 89.9%), intravenous drug use (91.5%, 90.7%) and vertical transmission (88.4%, 80.8) were listed as the most responsible risk factors for HCV and HBV, respectively, by the GPs. When the participants were asked about their suggestions of how to protect an infant, born from a HBsAg positive mother, it was stated by 108 (83.7%) of the participants that Hepatitis B vaccine should be injected to the infant during its delivery in addition to the immunoglobulin therapy. Twenty (15.5%), 12 (9.3%), 7 (5.4%) and 7 (5.4%) are the number of the given answers in 129 GPs for the following questions respectively; baby should be taken out by caesarean section, only immunoglobulin should be given at birth, only Hepatitis B vaccine should be applied at birth and did not actually know what to do. When the participants were asked about what they would suggest do to protect an infant, born from an anti-HCV and HCV RNA positive mother, it was stated by 67 (51.9%) of them that there were no preventive measures to be taken during the birth and the baby should be followed with respect to HCV infection. Thirty-four (26.4%), 33 (25.6%) and 7 (5.4%) of them suggested that immunoglobulin should be applied to the baby at birth, the birth should be carried out by caesarean section, and they aware not actually well informed about the case, respectively. The participants could choose more than one solution regarding protection of baby from viral hepatitis. Experience with viral hepatitis ------------------------------- GPs reported that they had no diagnostic laboratory tests for HBV (108; 83.7%) and HCV (126; 97.7%). Because of the lack of diagnostic laboratory tests, these diseases could not be diagnosed at their PHCCs. One hundred and ten (85.3%) of the participants expressed that no patient with acute HBV (AHB) infection admitted to their health care centers and 84 (65.1%) of them stated that they did not encounter any patient with chronic HBV (CHB) infection during the last 12 months. Rest of 45 GPs encountered total 52 patients at previous year; only 6 GPs examined more than one patient. One hundred and twenty-six (97.7%) of them reported that they did not encounter any patient with HCV infection. Approach to sample cases ------------------------ Table [1](#T1){ref-type="table"} shows the criteria used by the participants in the management of the two cases with HBV infection. ###### Management of Patients with HBV ---------------------------------------------------------------------- ------------ ------------- -------------------- ------- ------- -------------------- **Case I** **Case II** **n** **%** **Current Advice** **n** **%** **Current Advice** **At this point your next step would be to:**  Follow in clinic: level of ALT 53 41.1 Yes 19 14.7 No  Follow in clinic: refer if symptoms develop 43 33.3 Yes \- \- No  Refer to a specialist 29 22.5 No 10 7.6 Yes  Don\'t know 4 3.1 100 77.5 **Which additional tests would you order (check all that apply)\*:**  HBV-DNA by PCR\*\* 79 61.2 No 62 48.1 Yes  HBV genotyping 2 1.6 No 16 12.4 No  Liver biopsy 93 72.1 No 44 34.1 Yes  Don\'t know \- \- 2 1.6 **If you order therapy, which regimen would you recommend:**  Lamivudine 40 31.0 No 26 20.2 Yes  IFN-α + Lamivudine 3 2.3 No \- \- Yes  Interferon (IFN)-α 11 8.5 No 2 1.6 Yes  Don\'t know 75 58.2 101 78.2 ---------------------------------------------------------------------- ------------ ------------- -------------------- ------- ------- -------------------- \* The sum of the answer is higher than the number of the participants for the participants checked more than one item. \*\* Polymerase Chain Reaction (PCR) Table [2](#T2){ref-type="table"} shows the criteria used by the participants in the management of the two cases with HCV infection. ###### Management of Patients with HCV ---------------------------------------------------------------------- -------------- ------------- -------------------- ------- ------- -------------------- **Case III** **Case IV** **n** **%** **Current Advice** **n** **%** **Current Advice** **At this point your next step would be to:**  Repeat the measurement of anti-HCV 16 12.4 No 49 38 No  Follow in clinic: refer if symptoms develop 5 3.9 No 30 23.2 Yes  Refer to a specialist 108 83.7 Yes 50 38.8 Yes  Don\'t know \- \- \- \- **Which additional tests would you order (check all that apply)\*:**  anti-HCV ELISA 52 40.3 No 50 38.8 No  anti-HCV RIBA 2 1.6 Yes 4 3.1 Yes  HCV-RNA by PCR 105 81.4 Yes 106 82.2 Yes  HCV genotyping 3 2.3 No 9 7.0 No **If you order therapy, which regimen would you recommend:**  IFN-α for 6 months 23 17.8 No 20 15.5 No  IFN-α for 12 months 4 3.1 No 4 3.1 No  Ribavirin alone \- \- No \- \- No  IFN-α and Ribavirin 23 17.8 Yes 24 18.6 No  Don\'t know 79 61.3 81 62.8 ---------------------------------------------------------------------- -------------- ------------- -------------------- ------- ------- -------------------- \* The sum of the answer is higher than the number of the participants for the participants checked more than one item. Sources of education -------------------- The answers given by the participants about the sources of the updated current information for HBV and HCV infections were as follows, respectively: using journals: 63.6%, 52.7%, attending to congress: 14.0%, 11.6%, using books: 79.1%, 76.7% and on-line sites: 3.1%, 3.1% (The sum of the answer is higher than the number of the participants for the participants checked more than one item.) Discussion ========== Our data suggest that, the high response rate to our questionnaire indicated that GPs find the problem as an important area to consider. The sample was representative of GPs in Samsun region. The results could be extrapolated to whole Turkey on the basis of educational levels and working conditions, although the practice years of the respondents seem to be smaller than in practice. HBV and HCV cause chronic viral hepatitis \[[@B10]\]. The frequency of the HBV and HCV infections are high in developing countries, including Turkey, especially young children who become infected with HBV are the most likely to develop chronic infection \[[@B5],[@B6],[@B11]\]. Therefore, GPs who constitute almost half of primary health care and outpatient services are expected to know basic knowledge on such infections. Furthermore, they are also supposed to guide their patients appropriately in terms of analysis of their diseases and treatment. GPs should be able to identify HBV and HCV high-risk groups at an early period because of the high importance of the complications like cirrhosis and hepatic failure \[[@B6],[@B9],[@B12]\]. The majority of GPs were well informed about transmission ways for HBV and HCV. However they identified sexual and vertical transmission as being important risk factors for HCV, the estimated transmission risk is only 5% \[[@B13]\]. This indicates confusion between modes of transmission for HBV and HCV. The relatively poorer levels of knowledge about transmission of HCV obtained in this study may also be found in the general population. Therefore strategic programmes of health education and awareness rising to both professionals and public are recommended. Up to 90% of infants of highly infected with HBsAg and HBeAg positive mothers become HBV carriers compared with 10% to 40% of babies born anti-HBe positive mothers. HBV vaccine and HBIG should be applied at birth to the baby born to a HBsAg positive mother \[[@B14]\]. Due to the data gathered at the end of these studies the transmission is not effected by the way of the baby delivery and breast-feeding. Since the frequency of vertical transmission is less than 5%, there was no special recommendation for anti-HCV positive pregnant women \[[@B13],[@B15],[@B16]\]. This study shows that most GPs are practicing within current recommendations regarding hepatitis B immunization of infants, but the use of vaccine and HBIG at birth may not provide an effective safety in preventing perinatal HBV transmission if screening for HBsAg is not performed universally. However most of them (83.7% and 97.7%, respectively) were unable to diagnose the HBV and HCV infections in PHCCs due to the lack of laboratory facilities. Most of the participants pointed out that they had not determined or followed up of any patient with HBV and HCV infections at their health care centers during the last 12 months. This fact clearly shows that PHCCs in Turkey are saw insufficient in terms of the determination of the patients with HBV and HCV infections and their follow up. These results underscore the need of effective mechanisms in place and on time for immunoprophylaxis to infants born to HBsAg positive women. We identified gaps between the knowledge about appropriate use of diagnostic tests and interventions important to identify patients with chronic hepatitis and to make treatment decision. In case I; she has a normal ALT, HBV-DNA testing and liver biopsy should not be recommended \[[@B17],[@B18]\]. However 61.2% of the respondents ordered HBV DNA test and 72.1% liver biopsy in case I. Forty-eight point one percent of the GPs stated HBV DNA test and 34.1% liver biopsy in case II. The EASL consensus statement recommended liver biopsy and consideration for treatment in case II and regular follow up at six-month interval with determination of ALT level, no further testing to confirm the diagnosis of chronic HBV infection in case I \[[@B17]\], but most of the participants did not differentiate these two entities. In case III and IV, respondents would unnecessarily repeat the anti-HCV test. These finding suggest that inappropriate use of tests for viral hepatitis may lead to misdiagnosis, unnecessary testing, and also over-treatment. As long as ALT levels are normal, the treatment is not required even if PCR HBV DNA or HCV RNA is found to be positive \[[@B17],[@B19]\]. However 41.8% and 37.2% of GPs support to treat chronic hepatitis B and C patients with normal ALT, respectively. When asked about the treatment for HCV, only 17.8% of respondents recommended combined interferon and ribavirin therapy to treat the chronic hepatitis C patients with elevated ALT in case III. Twenty-one point eight percent of them recommended either IFN-α or lamivudine to treat chronic viral hepatitis B in case II. The remainder GPs did not actually know what to do with such patients. Therefore the knowledge of treatment options in chronic hepatitis B and C is poor, this may decrease the ability to accurately assess patients and initiate therapy. This clearly identifies that regular updates on current treatments are needed. Protocols, reviews and consensus reports have been useful in the management of chronic viral hepatitis \[[@B17]-[@B19]\]. General practitioners reported medical books and journals to be the most useful source of information regarding viral hepatitis but were concerned about inadequate timely access the journal and also the low rate of obtaining information from on-line sites. However internet may be a solution for difficulties and delays in obtaining relevant information. Most of the updated guidelines and consensus reports could be easily and freely accessible at on-line sites such as <http://www.jhep-elsevier.com> and <http://consensus.nih.gov/cons/116/116cde_intro.htm> \[[@B17],[@B19]\]. Conclusion ========== As the \"gatekeepers\" to the health care system, GPs carry out significant missions in PHCCs where primary protective and curative health services are held out. It is, therefore necessary to completely inform GPs of common health problems as well as of the ways of their solutions. In order to prevent HBV and HCV infections, GPs drive great responsibility of the diagnosis and treatment of the infected patients as well as their follow up. This study demonstrates high awareness of the transmission of viral hepatitis and high-risk groups among GPs but also considerable knowledge gaps about the timing of ordering diagnostic tests and treatment recommendations. Results suggest several opportunities for improvement. Efforts to educate the GPs about the appropriateness and importance of identification of HBV and HCV infection during post-graduation period should be improved. PHCCs should be technically supported to ease the diagnosis of HBV and HCV infections. Finally further coordination between an advanced health care center and specialists should be established to enable such infections to be followed up in PHCCs. Competing interests =================== None declared. Authors\' contributions ======================= YP participated in the design and coordination of the study, SC drafted the questionnaire and participated in study design and coordination. HL conceived the study, particated in its design and drafted the manuscript. MS and SE provided clinical expertise in interpretation of data and drafting manuscript. ATS provided drafted the manuscript and performed the statistical analysis. All authors read and approved the final manuscript. Pre-publication history ======================= The pre-publication history for this paper can be accessed here: <http://www.biomedcentral.com/1471-230X/4/3/prepub> Supplementary Material ====================== ###### Additional File 1 Viral Hepatitis Questionnaire. A questionnaire to evaluate the current practices, knowledge, attitudes of general practitioners regarding viral hepatitis B and C. ###### Click here for file

Background {#Sec1} ========== The design of the insert of the artificial joint can be roughly divided into two types, fixed-bearing and mobile-bearing. Mobile-bearing design was developed as a rotating platform that allows axial rotation of the insert around the longitudinal axis. It was theorized that mobile-bearing total knee arthroplasty (TKA) should allow self-alignment of the polyethylene insert with the femoral component in order to lessen polyethylene surface stresses, minimize post-cam impingement, and possibly increase polyethylene longevity \[[@CR1]\]. However, many studies failed to prove that mobile-bearing insert contributed to reduce insert wear or risk of loosening, as well as to improve clinical outcomes such as range of motion, functional scores, and pain relief \[[@CR2]--[@CR10]\]. Moreover, none of the studies utilizing an accurate kinematic analysis system have proven that femoral-tibial component kinematics differs between mobile-bearing and fixed-bearing inserts \[[@CR11]--[@CR13]\]. However, precise movement of the mobile-bearing insert has not been clarified and the usefulness of the mobile-bearing insert in vivo remains uncertain. Several studies have used embedded tantalum beads to analyze the behavior of the mobile-bearing insert by X-ray fluoroscopy during in vivo knee motion \[[@CR14]--[@CR17]\]. These studies demonstrated that axial rotation of the femoral component with respect to the mobile insert was small. Moreover, the rotation offset, which refers to the difference in the rotational positioning between two components, during flexion was limited \[[@CR15]\]. However, Dennis et al. \[[@CR18]\] demonstrated that the behavior of insert rotation after mobile-bearing TKA is highly variable and depends on the implant design. In the NexGen LPS mobile-bearing knee prosthesis (Zimmer Inc. Warsaw, IN, USA), there is a limited degree of conformity of the femoral component on the insert surface, allowing the femoral component to slide with respect to the insert (± 12° of rotation) \[[@CR19]\]. Garling et al. \[[@CR19]\] showed that, in patients with rheumatoid arthritis, the conformity of the NexGen LPS mobile bearing knee prosthesis was low enough that the femoral component was allowed to translate with respect to the insert without forcing the insert to rotate. Consequently, we can speculate that, if the conformity of the mobile-bearing insert is low, the femoral component can rotate and slide on the insert more than insert rotation relative to the tibial tray. Therefore, the mobile-bearing insert with low conformity between the femoral component and insert may limit its ability to rotate on the tibial tray, which may reduce longevity. Due to the lack of knowledge on the behavior of the mobile-bearing in patients with knee osteoarthritis (KOA), there has not been any consensus on the behavior of the mobile-bearing insert in patients with KOA. The objective of this study was to clarify the tibiofemoral kinematics and rotation of the polyethylene mobile-bearing insert relative to the femoral and tibial components under weight-bearing conditions in patients implanted with mobile-bearing TKA. The hypotheses of this study were as follows: (1) the mobile-bearing insert can rotate on the tibial tray to reduce the offset at FEM/INS at knee extension position and (2) intra-prosthetic axial rotation between the tibial tray and mobile-bearing insert is greater than that between the femoral component and mobile-bearing insert. Methods {#Sec2} ======= Participants {#Sec3} ------------ This study was a cross-sectional study (Level 4) involving patients with severe knee osteoarthritis scheduled to receive TKA at a single regional hospital. The study protocol was approved by the institutional review board of Hiroshima International University (12-025), and informed consent was obtained from all subjects included in the study prior to the initiation of this study. All subjects who met specific selection criteria were enrolled. Inclusion criteria were (a) Japanese males and females and (b) between 50 and 85 years old, primary medial KOA with radiographic severity of grade III or IV in the Kellgren-Lawrence system, with or without lateral or patellofemoral degradation. Exclusion criteria were (a) lateral compartment knee osteoarthritis; (b) history of knee surgery; (c) any other disease involving the knee joint; and (d) history of rheumatoid arthritis, cerebral disorders, neuropathy, or gout. Surgical technique and procedure to embed tantalum beads {#Sec4} -------------------------------------------------------- All osteoarthritic knees underwent TKA with posterior stabilizing (PS) components and mobile-bearing insert (NexGen LPS-Flex, Mobile Surface, Zimmer Inc., Warsaw, IN, USA). This prosthesis was designed to safely allow 155° flexion and has a mobile bearing that allows up to 25° internal or external rotation on an anteriorly placed rotation center. A senior surgeon (YK) performed arthroplasty using his own technique, which is summarized by a paramedian approach for minimal exposure, achieving 6° valgus for the femoral cut and 90° to the tibial shaft, modified ligament dependent cut for femoral rotational alignment. Because the polyethylene insert is transparent during X-ray fluoroscopy, three-dimensional pose estimation of the mobile-bearing insert was achieved using tantalum beads embedded in the insert. Using a custom-made bead injector, four identical metallic tantalum beads (diameter 0.8 mm) were inserted into predefined non-weight-bearing areas of the polyethylene insert during surgery (Fig. [1](#Fig1){ref-type="fig"}). The four holes in the external surface of the template device were the same size as the head of the bead injector, and the tantalum beads were precisely inserted at a constant height and depth from the lateral surface of the insert. Since the actual positions of the embedded tantalum beads must be confirmed to minimize errors in beads positioning, we performed CT imaging on all subjects on a different day from surgery to determine the exact beads positions, and a subject-specific computer-added design (CAD) model of the four beads (bead model) was created in each case. All subjects received same postoperative treatment and rehabilitation program. Fig. 1A custom-made insert holder has four holes, through which a beed injecter was inserted to shoot beeds at reproducible locations Outcomes {#Sec5} -------- We measured active and passive knee flexion angles, Western Ontario and McMaster Universities Osteoarthritis Index (WOMAC) \[[@CR20]\], and articular component kinematics during a squatting activity at 10 weeks and 1 year postoperatively. We used pain, stiffness, physical function scores and total score of 100 mm visual analogue scale WOMAC. A single examiner obtained all clinical and kinematic measurements. Clinical outcomes {#Sec6} ----------------- Weight bearing and passive knee range of motion (ROM) were measured using sagittal images from a movie taken with a commercial video camera and analyzed using the ImageJ software (ImageJ 1.48v, National Institutes of Health, Bethesda, MD). The video camera was positioned 3 m away from the subject to bilateral sides and adjusted the camera so that the center of the knee joint was at the center of the viewing screen. For the weight bearing ROM, subjects stood sideway to the camera and were instructed to squat down from full extension to maximal flexion with maximal effort. They were allowed to hold handrail for safety. For passive knee flexion measurement, subjects were instructed to lie supine on a bed, and a senior physical therapist bent their knee to maximum flexion. Symptoms and physical function were measured using the WOMAC index at 10 weeks and again 1 year postoperatively. Subjects responded to all sections including subscales of pain, stiffness, and physical function during daily living. All measurements were performed by an 11th year physical therapist with research experience of 4 years as a graduate student. Model registration and data processing {#Sec7} -------------------------------------- We chose wide-based squat for articular kinematic analyses. Each subject was asked to perform the wide-base squat during fluoroscopic surveillance using a 17-in. flat pannel (15 Hz, SONIALVISION Safire 17, Shimazu Corp., Japan). To avoid the influence of shoes and insole, all images were obtained while the subject was barefoot. Subjects were instructed to squat as slowly as possible with the hips externally rotated so that the feet were angled at 90^o^ \[[@CR21]\] (Fig. [2](#Fig2){ref-type="fig"}). Subjects were instructed to bend taking more than 5 s and extent more than 5 s. In this position, overlapping of the contralateral knee on the fluoroscopic view could be avoided. As noted above, subjects were allowed to hold onto handrail for safety. However, they were instructed not to allow forward movement of the knee beyond the toes while squatting in order to avoid knee movement outside of the fluoroscopic view. Subjects practiced the activity until they felt comfortable, and the motion became smooth before recording began. Successful knee motions were recorded as serial digital X-ray images. Fig. 2Patients were instructed to stand with his or her hips externally rotated with the feet angled at 90° to avoid overlapping of the contralateral knee. They were allowed to hold an handrail for safety All analyses were performed by a single experienced examiner. Three-dimensional in vivo images of the prostheses were obtained using a 3D-to-2D registration technique \[[@CR22]\]. Calibrated fluoroscopic images and CAD models of the femoral and tibial components, as well as the bead model composed of the four tantalum beads implanted in the polyethylene insert were utilized to obtain in vivo, six-degrees-of-freedom positions and orientation of the models using the JointTrack program ([sourceforge.net/projects/jointtrack](http://sourceforge.net/projects/jointtrack)) (Fig. [3](#Fig3){ref-type="fig"}). The optical geometry of the fluoroscopic system (principal distance, principal point) was determined based on images of a calibration target. An implant surface model was projected onto the geometry-corrected fluoroscopic images, and its 3D position and rotation was iteratively adjusted so that its silhouette matched with that of the knee prosthesis using the JointTrack. After the tibial component was registered, the bead model was first fitted onto the upper surface of the tibial tray and then fitted with the beads in the fluoroscopic image by rotating around the Y rotation axis (or tibial vertical axis). Fig. 33-dimensional CAD models were laid over the calibrated fluoroscopic images to determine the best-fitted position Following the registration process, relative motions of the three CAD models were computed using the 3D-JointManager software (GLAB Inc., Japan) (Fig. [4](#Fig4){ref-type="fig"}). Rotation of the femoral component relative to the tibial component (FEM/TIB) or polyethylene insert (FEM/INS), and rotation of the polyethylene insert relative to the tibial component (INS/TIB), were computed using a joint coordinate system proposed by Andriacchi et al. \[[@CR23]\]. Rotational offset was measured at maximal extension position at each time point in each subject. External rotation of the superior component relative to the inferior was denoted as positive. Fig. 43D-JointManager software (GLAB Corp.) was used to compute 6 degrees-of-freedom positions and orientations of the CAD models at each frame. This figure showes external rotation offset of the mobile-insert relative to the tibial tray and very small rotation of the femoral component relative to the insert Internal validity {#Sec8} ----------------- To analyze knee kinematics during a squatting activity, we selected 20 frames from maximal knee extension to maximal knee flexion. Kinematics was analyzed at 5° increments of knee flexion after B-spline curve approximation was performed. In the knee joint, this matching method has an estimated accuracy of 0.53 mm for in-plane translation, 1.6 mm for out-plane translation, and 0.54° for rotation \[[@CR22], [@CR24]\]. Statistical analysis {#Sec9} -------------------- Characteristics and kinematic parameters of subjects were summarized using means and 95% confidence intervals \[95% CI\] for continuous values. The assumption of normality and equality of variance was assessed using the Shapiro-Wilk test. If the data met parametric assumptions, we used repeated measure ANOVA and Tukey post hoc test for pairwise comparisons. Paired *t* test was employed to examine differences in kinematics between 10 weeks and 1 year postoperatively. If the data did not meet parametric assumptions, Wilcoxon signed-rank test was employed. Out of plane translation (medial-lateral) was excluded from the analyses in this study due to the limited accuracy. Pearson's correlation coefficient was performed for the relationship between weight and rotation motion of each component. All data were analyzed using SPSS Statistics v21 (SPSS Inc., Chicago, IL) with the significance level set at alpha = .05. Results {#Sec10} ======= Thirteen knees of 11 subjects (3 males and 8 females) were evaluated in this study (mean \[95% CI\] age 71.4 \[67.5, 75.3\] years, height 153.4 \[148.8, 158.1\] cm, weight 67.4 \[58.9, 75.8\] kg, body mass index (BMI) 28.4 \[26.3, 30.5\] kg/m^2^ at 10 weeks postoperatively) (Table [1](#Tab1){ref-type="table"}). Table 1Demographic data of study participants10 weeks postoperatively1 year postoperatively*p* valueAge (years)71.4 \[67.5, 75.3\]Height (cm)153.4 \[148.8, 158.1\]Weight (kg)67.4 \[58.9, 75.8\]BMI (kg/m^2^)28.4 \[26.3, 30.5\]Active-ROM (degrees)99.8 \[92.5, 107.1\]92.7 \[83.5, 101.9\]0.173Passive-ROM (degrees)120.1 \[113.0, 127.2\]123.8 \[111.9, 134.9\]0.196WOMAC Pain41 (0-263)5 (0-99)**0.002** Stiffness44 (5-137)0 (0-92)**0.009** Physical function337 (69-696)167 (0-426)**0.019** Total score468 (108-1096)228 (2-489)**0.005**Mean \[95% CI\] for Age, Height, Weight, BMI and ROM resultsMedian (range) for WOMAC results*95% CI* 95% confidence interval, *BMI* body mass index, *ROM* range of motion, *WOMAC* Western Ontario and McMaster Universities Osteoarthritis Index In the clinical measurements, active and passive ROM of the knee on the operated side were 99.8° \[92.5, 107.1\] and 120.1° \[113.0, 127.2\] at 10 weeks and 90.8° \[83.5, 101.9\] and 123.4° \[111.9, 134.9\] at 1 year, respectively. There were no statistically significant differences between data at the two time-points for either active ROM (*p* = 0.173) or passive ROM (*p* = 0.196). Median WOMAC scores for pain, stiffness, physical function, and total score using 100 mm VAS were 41, 44, 337, and 468 at 10 weeks and those at 1 year were 5, 0, 167, and 228, respectively. All scores at 1 year were significantly less than those at 10 weeks (pain, *p* = 0.002; stiffness, *p* = 0.009; physical function, *p* = 0.019; total score, *p* = 0.005) (Table [1](#Tab1){ref-type="table"}). The average flexion angles of the tibial and femoral components at the start of movement (maximal extension) were − 2.0° \[− 5.5, 1.6\] at 10 weeks and − 3.2° \[− 10.9, 4.4\] at 1 year, respectively. FEM/TIB, FEM/INS, and INS/TIB angles at 10 weeks were 6.0° \[2.0, 9.9\], − 0.3° \[− 1.0, 0.4\], and 6.3° \[2.1, 10.4\], respectively. FEM/INS was significantly smaller than INS/TIB (*p* = 0.013). FEM/TIB, FEM/INS, and INS/TIB angles at 1 year were 4.0° \[0.8, 7.2\], − 0.8° \[− 2.4, 0.7\], and 4.9° \[1.4, 8.3\], respectively. FEM/INS was significantly smaller than INS/TIB (*p* = 0.011). FEM/TIB at 1 year was significantly smaller than that at 10 weeks (*p* = 0.041) (Table [2](#Tab2){ref-type="table"}). Table 2Rotational offset of each component at the start of movement10 weeks postoperatively1 year postoperatively*p* valuesKnee flexion angle (degrees)− 2.0 \[− 5.5, 1.6\]− 3.2 \[− 10.9, 4.4\]0.709FEM/TIB (degrees)6.0 \[2.0, 9.9\]4.0 \[0.8,7.2\]**0.041**FEM/INS (degrees)− 0.3 \[− 1.0, 0.4\] ※− 0.8 \[− 2.4, 0.7\] †0.539INS/TIB (degrees)6.3 \[2.1, 10.4\] ※4.9 \[1.4, 8.3\] †0.136*p*-value (FEM/INS with INS/TIB)**0.0130.011**Mean \[95% CI\]※FEM/INS was significantly smaller than that of INS/TIB at 10 weeks postoperatively†FEM/INS was significantly smaller than that of INS/TIB at 1 year postoperatively*FEM/TIB* rotation of the femoral component relative to the tibial component, *FEM/INS* rotation of the femoral component relative to the polyethylene insert, *INS/TIB* rotation of the polyethylene insert relative to the tibial component, *p value* probability value, *95% CI* 95% confidence interval Rotation motion during a squatting activity from 2.0 \[− 5.5, 1.6\] degrees hyperextension to 55.3 \[44.6, 66.0\] degrees flexion for FEM/TIB, FEM/INS, and INS/TIB at 10 weeks were 5.7° \[4.2, 7.3\], 5.9° \[4.5, 7.2\], and 1.8° \[1.4, 2.2\], respectively. Total rotation motion of FEM/INS was significantly greater than that of INS/TIB (*p* \< 0.001). At 1 year, rotation motion during squatting from − 3.2 \[− 10.9, 4.4\] to 53.9 \[48.9, 58.8\] degrees were 6.3° \[4.3, 7.8\], 5.5° \[3.8, 7.2\], and 1.6° \[1.1, 2.2\], respectively. Total rotation motion of FEM/INS was significantly greater than that of INS/TIB (*p* \< 0.001). There were no statistically significant differences between data obtained at 10 weeks and those at 1 year (*p* = 0.702, *p* = 0.547, and *p* = 0.517, respectively) (Table [3](#Tab3){ref-type="table"}). There was no significant correlation between subject's weight and rotation motion. Table 3Rotation excursions during squatting activity10 weeks postoperatively1 year postoperatively*p* valuesKnee extension angle (degrees)− 2.0 \[− 5.5, 1.6\]− 3.2 \[− 10.9, 4.4\]0.709Knee flexion angle (degrees)55.3 \[44.6, 66.0\]57.1 \[46.7, 67.6\]0.760FEM/TIB (degrees)5.7 \[4.2, 7.3\]6.3 \[4.3, 7.8\]0.702FEM/INS (degrees)5.9 \[4.5, 7.2\] ※5.5 \[3.8, 7.2\] †0.547INS/TIB (degrees)1.8 \[1.4, 2.2\] ※1.6 \[1.1, 2.2\] †0.517*p* value (FEM/INS with INS/TIB)\< **0.001**\< **0.001**Mean \[95% CI\]※FEM/INS was significantly greater than that of INS/TIB at 10 weeks postoperatively† FEM/INS was significantly greater than that of INS/TIB at 1 year postoperatively*FEM/TIB* rotation of the femoral component relative to the tibial component, *FEM/INS* rotation of the femoral component relative to the polyethylene insert, *INS/TIB*: rotation of the polyethylene insert relative to the tibial component, *p value* probability value, *95% CI* 95% confidence interval Discussion {#Sec11} ========== This study was aimed to determine clinical outcomes and rotational kinematics of TKA components and mobile-bearing insert during a squatting activity in vivo at 10 weeks and 1 year postoperatively. The results demonstrated that WOMAC scores at 1 year were significantly smaller than those at 10 weeks, whereas there were no significant differences in active or passive ROM between the two time-points. On kinematic analyses, the mobile-bearing insert played a role in offsetting rotation alignment by around 6° at the starting position with a very small rotation of less than 2° motion during squatting. Contrary to the hypothesis, however, the amount of INS/TIB rotation was significantly smaller than that of FEM/INS rotation during a squatting activity. The WOMAC scores improved significantly between 10 weeks and 1 year. We observed that some subjects had local warmth and swelling at 10 weeks and, therefore, inflammation and pain were considered to be the cause of the lower subjective score at that time point. Bonnefoy-Mazure et al. \[[@CR25]\] showed that WOMAC improved significantly between 3 months and 1 year after TKA. They observed that patients still had pain and motion limitation at 3 months. In our study, WOMAC scores at 1 year were lower than those of the previous study \[[@CR25]\]. However, the TKA procedures in our study were considered successful and knee kinematic data at 10 weeks were considered to stay on the good path to achieve good function at 1 year. Therefore, our kinematic findings are considered applicable to successful TKA procedures using the mobile-bearing insert in general. Mobile-bearing TKA is commonly expected to have a self-alignment mechanism compensating for rotational offset between the tibial and femoral components. An in vivo kinematic study has shown that self-alignment of the polyethylene bearing typically occurs on the tibial tray which should hypothetically lessen polyethylene surface stresses, minimize post impingement, and increase the potential for enhanced polyethylene longevity \[[@CR1]\]. This self-alignment mechanism of the polyethylene bearing on the tibial tray is considered superior to a fixed type in maintaining femoral component-to-insert conformity. However, Okamoto et al. \[[@CR26]\] indicated that, although mobile-bearing TKA may have certain advantages over fixed-bearing TKA, for example correction of rotational offset while standing, the kinematics of mobile- and fixed-bearing TKA were not significantly different. Yamazaki et al. \[[@CR16]\] demonstrated that the femoral component and the mobile-bearing insert were already rotated externally to a similar degree with respect to the tibial component. In this study, we observed a 6° offset in axial rotation at maximal extension, and the advantage of self-alignment with respect to the mobile-bearing TKA was confirmed. This finding is similar to those of previous studies, and rotation offset of the mobile-bearing insert can effectively serve to improve the conformity between the femoral component and mobile bearing insert during squatting, which may improve conformity during gait. Therefore, the mobile-bearing insert may reduce contact pressure and delay the wear of the insert over the long term. The rotation motion of INS/TIB during a squatting activity in this study was significantly smaller than the rotation of FEM/INS. This finding was different from previous studies using other implants that demonstrated greater INS/TIB rotation motion than FEM/INS \[[@CR14], [@CR16], [@CR17]\]. Some studies examining the kinematics of the same TKA implant as that used in this study demonstrated rotation of the femoral component on the insert \[[@CR19], [@CR27], [@CR28]\]. Moreover, Garling et al. \[[@CR19]\] showed variations of rotational kinematics of the insert on the tibial tray including greater rotation and no rotation and indicated that the conformity of this prosthesis was low enough that the femoral component is allowed to translate with respect to the insert without forcing the insert to rotate. In our study, the femoral rotation motion on the insert was clearly greater than the insert rotation motion on tibial tray. Garling et al. \[[@CR19]\] measured using step-up motion, and we used a squat motion. The result of this study was similar to that of Garling et al. \[[@CR19]\]. Both activities were under loading condition, and vertical axial force was applied to the mobile-insert, so that the insert was compressed toward the tibial tray. Since this force acted to increase friction at the contact surface, it might have prevented INS/TIB rotation. Therefore, it remains unclear whether this phenomenon also occurs during a non-loading activity such as leg extension in a seated position. Since the conflicting kinematic findings across studies might be related to variations in the implant design and/or loading conditions, the results have a limited generalizability among implants from various manufacturers. There were several strengths and limitations in this study. Strengths include (1) the use of a 3D-to-2D registration technique with fluoroscopy, which is an accurate and well-established technique to measure in vivo knee kinematics \[[@CR22], [@CR24]\]. Our single plane fluoroscopic method utilized a 17 × 17-in. field of view to allow analysis of the knee during a squatting activity; (2) all analyses were performed by a single investigator to avoid potential inter-investigator errors. Thus, systematic errors between investigators did not influence the results; and (3) because X-rays penetrate the polyethylene insert, we injected metal beads into the polyethylene insert to make it visible. Furthermore, CT images were taken in each subject to confirm the exact locations of the beads. The method using single plane fluoroscopic images is less accurate for out-of-plane kinematics because out-of-plane 3D-pose of models on the screen involved a greater error \[[@CR24]\]. Therefore, kinematic analyses of squatting or stair ascending/descending activity in the literature avoided reporting medial/lateral tibial translation. The 3D-to-2D registration technique using dual plane fluoroscopic images is more accurate than analyses using single plane fluoroscopic images but has a smaller radiographic area, which limits the ability to capture greater and more dynamic movement of the joints. Therefore, there is an advantage in utilizing a single plane technique. The limitations of our study include the following: (1) although the sample size was small, this cohort is similar to that of the previous fluoroscopic studies, comparison between FEM/INS and INS/TIB showed a post hoc power of 0.9 or more at 10 week and 1 year after surgery. Thus, we thought to provide reliable data with sufficient sample size for analyses; (2) the 3D-to-2D registration method using single-plane fluoroscopic images provided limited measurement accuracy for out-of-plane kinematics. Therefore, we did not include medial-lateral translation in this study. As for external validity, the results of this study can be generalized to subjects aged 50 years and older who underwent TKA with the same mobile-bearing insert model, but the current data are limited to a Japanese population. Conclusions {#Sec12} =========== The mobile-bearing insert improves the rotational offset of the femoral and tibial components at the starting position, but the amount of rotation motion during squatting at the INS/TIB interface is very small. This study is thought to provide useful data for improving surgical procedures, developing future implant designs and understanding the implant wear pattern in TKA with a mobile-bearing. In a future study, it will be necessary to analyze whether the amount of mobile-bearing insert movement is affected by the presence or absence of weight-loading. 95% CI : 95% confidence interval BMI : Body mass index CAD : Computer-added design FEM/INS : Rotation of the femoral component relative to the polyethylene insert FEM/TIB : Rotation of the femoral component relative to the tibial component INS/TIB : Rotation of the polyethylene insert relative to the tibial component KOA : Knee osteoarthritis ROM : Range of motion TKA : Total knee arthroplasty WOMAC : Western Ontario and McMaster Universities Osteoarthritis Index **Publisher's Note** Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. We thank Yasuhiko Aratani, Hirofumi Dojo, and Shinji Maeda (Radiologists) for data collection. This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors. KH participated in establishing the research design, data collection, data processing, and data analysis as well as compiling the manuscript. GW participated in data collection, data processing, data interpretation, and revising the manuscript. YK, RT, and JF participated in establishing the research design, subject recruiting, data collection, and surgical procedure. KG participated in the research design and provided guidance on data collection, data processing, and editing the manuscript. All authors have read and approved the final version of this manuscript before submission. There is no funding source. Not applicable. The study protocol was approved by the institutional review board of Hiroshima International University (12-025), and informed consent was obtained from all subjects included in the study prior to the initiation of this study. Not applicable. The authors declare that they have no competing interests.

Introduction ============ Breast cancer is a major global public health problem and effective prevention strategies are needed to combat the disease, most desirably at the primary prevention level ([@bib15], [@bib12]). In addition to family history, other risk factors including hormonal and environmental factors, particularly nutrition, have been identified to be major contributors in the etiology of breast cancer ([@bib45], [@bib25], [@bib1], [@bib59]). For example, exposure to Western lifestyles dramatically increases breast cancer incidence in Asian immigrants in the United States (US) during their lifetime and in their offspring over several generations ([@bib4], [@bib65]). Maternal exposures to endocrine-disrupting chemicals (EDCs) and certain dietary factors during pregnancy have been reported to be associated with increased mammary tumorigenesis among female offspring ([@bib47], [@bib11], [@bib40], [@bib10], [@bib55]). However, cohort and human case--control studies have generated conflicting reports regarding the effect of high-fat intake on adult breast cancer risk ([@bib27], [@bib62], [@bib28], [@bib36]), partly due to differences in fat/dietary composition, and study design, including exposure window. For example, consumption of animal fat, consisting mainly of saturated fatty acids, is associated with increased risk of breast cancer in some studies ([@bib5], [@bib13]) but not in others ([@bib33], [@bib29], [@bib43]). The development of rodent models have proved critical to assess the effects of single and/or multiple exposures on susceptible windows of mammary gland development and in the subsequent identification of factors relevant for breast cancer prevention in humans. The mammary gland has been shown to be particularly sensitive during early development, both to dietary fatty acids and EDCs such as bisphenol A (BPA), perhaps because extensive programming of the mammary gland occurs during fetal life ([@bib18], [@bib19], [@bib21], [@bib22], [@bib11], [@bib40], [@bib32], [@bib10], [@bib34], [@bib55]). High-fat intake during pregnancy has been shown to increase offspring' mammary cancer risk by altering mammary gland development, mainly by increasing the number of terminal end buds (TEBs) ([@bib20], [@bib7]), the purported target structures of malignant transformation ([@bib54], [@bib61], [@bib51],[@bib52]). Similarly in rodent models, BPA has been reported to alter the development of the mammary gland suggesting the possibility of increased susceptibility to mammary tumorigenesis ([@bib35], [@bib41], [@bib11]). Using the 7,12-dimethylbenz(a)anthracene (DMBA)-induced mammary carcinogenesis Sprague--Dawley (SD) rat model, Russo and coworkers ([@bib3]) reported that a prenatal BPA dose of either 25 or 250 μg/kg body weight (BW)/day by itself had no tumorigenic effect after DMBA exposure on postnatal day (PND)50. However, when DMBA was administered at PND100 following prenatal BPA exposure at 250 µg/kg BW/day, the EDC exposure resulted in a significantly increased number of terminal ducts ([@bib40]) as well as a higher incidence of mammary tumors ([@bib3]). As the epigenome is most susceptible to perturbations in early development, the adverse effects of *in utero* exposures on adult health are likely mediated by epigenetic dysregulation of gene expression ([@bib44]). Indeed, a high-fat- or ethinylestradiol-supplemented maternal diet has been shown to increase mammary cancer risk in several generations of offspring and is associated with changes in the DNA methylation machinery and methylation patterns in mammary tissue ([@bib8]). We are not aware of any previous studies that have investigated the impact of maternal consumption of a high-butterfat (HBF) diet together with exposure to low-dose BPA on mammary cancer risk of offspring. The present study aimed to test the hypothesis that maternal exposure to low, environmentally relevant doses (2.5--2500 µg/kg BW/day) of BPA, in addition to HBF intake during pregnancy leads to increased incidence of mammary cancer in offspring. According to our previous study, we chose butter as the source of high fat (39% kcal) to mimic a key fat component in the Western diet, and which does not promote obesity in our animal model ([@bib38]). Our results indicate that concurrent exposure of dams to a HBF diet and BPA at 25 μg/kg BW/day dose level during pregnancy increases mammary tumor incidence in offspring treated with DMBA on PND50, accompanied by alterations in mammary gland development. Furthermore, *in utero* HBF and BPA exposure elicited differential effects at the gene level in prepubertal mammary glands through DNA methylation, compared with HBF in the absence of BPA. A signature of top dysregulated genes was subsequently found to be associated with poor overall survival in populations of breast cancer patients from The Cancer Genomic Atlas (TCGA). These findings highlight the importance of future studies to address maternal diets as modifiers of susceptibility to *in utero* exposure to environmental agents for devising new strategies to reduce breast cancer risk. Materials and methods ===================== Animals ------- Female, virgin SD rats at \~7 weeks of age were obtained from Taconic Farms (Germantown, NY, USA). Animals were housed individually in a temperature- and humidity-controlled environment with a 12-h light--darkness cycle, in the AAALAC-approved University of Cincinnati animal facility. All rats were provided food and filtered water *ad libitum*, and animals were housed on sani-chips bedding and maintained in an environment under controlled endocrine-disrupting chemical (EDC) exposures ([@bib60], [@bib37]). All animal procedures were approved by the University of Cincinnati Institutional Animal Care and Use Committee, and experiments were performed following the guidelines of the National Institutes of Health for the proper and humane use of animals in biomedical research. Experimental design and mammary tumorigenesis --------------------------------------------- Female SD rats (7--9 weeks old) were randomized into 6 groups (*n* = 11). After two weeks (acclimation) of experimental diet group exposure, female rats were bred with male SD rats (\~3 months of age). During mating and throughout gestation, dams were fed either a control AIN-93G diet (10% kcal from butterfat) or a modified AIN-93G high-butterfat diet (39% kcal from butterfat), in the presence or absence of BPA (Sigma-Aldrich) at various environmentally relevant doses: 2.5, 25, 250 or 2500 µg/kg BW/day. Diets were controlled for caloric content, vitamins, salts and protein, but varied in fat and carbohydrate content. After birth, dams and offspring were maintained on control AIN-93G diet for the duration of the experiment. Litters were weighed weekly until killed. The date of vaginal opening was recorded as an indication of reaching sexual maturity. At PND50, one female offspring per litter per group was treated with a single oral dose (20 mg/kg BW) of DMBA (Thermo Fisher Scientific) to induce mammary cancer, and another littermate was given corn oil (Sigma-Aldrich) as a control. Animals were palpated weekly to monitor tumor development. Tumor size was measured using a caliper and tumor volume was calculated. All pups were killed at PND140, or when tumor burden exceeded 10% of total BW. Time to first tumor appearance (latency), the number of animals with palpable tumors (incidence) and the number of tumors per animal (multiplicity) were determined. The origin of the tumor was confirmed by a clinical pathologist (AK). Serum hormone levels of 17β-estradiol (Calbiotech, Spring Valley, CA, USA), progesterone (IBL, Minneapolis, MN, USA), leptin (EMD Millipore) and adiponectin (EMD Millipore) at PND21 and 50 were measured using ELISA-based assays. A schematic diagram summarizing the experimental design of this study is presented in [Fig. 1.](#fig1){ref-type="fig"} Figure 1Schematic diagram of the experimental design. Female SD rats (7--9 weeks old) were randomized into 6 groups (*n* = 11 litter/group, one offspring/litter). During mating and throughout gestation, dams were fed a control AIN-93G diet or a modified AIN-93G high-butterfat (HBF) diet in the presence or absence of bisphenol A (BPA) at various concentrations (µg/kg BW (body weight)/day). After birth, dams and offspring were maintained on an AIN-93G diet for the duration of the experiment. Pups were weaned at PND21 and one female offspring per litter was killed for mammary gland transcriptome analysis. At PND50, one female offspring from each dietary group was treated with a single oral dose (20 mg/kg BW) of DMBA to induce mammary cancer. All pups were killed at PND140 for analysis and determination of mammary tumor incidence. Whole mount and morphometric analysis ------------------------------------- The fourth abdominal mammary gland from PND21 female pups was prepared as whole mount for morphological analysis according to an established protocol ([@bib49]). The total number of terminal end buds (TEBs) was evaluated by counting individual structures under a dissection microscope. Laser capture microdissection ----------------------------- PND21 mammary glands from female offspring (*n* = 5 L/group, one offspring/L) fed a HBF diet and a HBF diet with 25 µg BPA/kg BW/day were chosen for laser capture microdissection (LCM) according to our published protocol ([@bib64]). Briefly, mammary glands were first cryosectioned with 10 micron thick, hematoxylin-stained and dried for microdissection. Multiple sections (\~5) per mammary gland were microdissected using an Arcturus Veritas Laser Capture Microdissection System (Thermo Scientific). RNA sequencing -------------- Total RNA was extracted from LCM samples using a Qiagen RNeasy Lipid kit (Qiagen). The RNA quality and quantity were assessed using Agilent Bioanalyzer (Agilent) and NanoDrop ND-1000 spectrophotometer (Thermo Scientific), respectively. RNA libraries were prepared according to manufacturer's protocol of TruSeq RNA sample preparation kit (Illumina, San Diego, CA, USA) and were sequenced with Genome Analyzer II sequencing system in the Genomics, Epigenomics and Sequencing Core at the University of Cincinnati. Analysis of gene expression was performed with our standard pipeline ([@bib14]). Differentially expressed genes of the HBF + BPA group were selected based on *P* \< 0.05 and a fold-change greater than 2, when compared with the HBF-alone group. Functional enrichment analysis of differentially expressed genes was performed using the knowledge-based Ingenuity Pathways Analysis (IPA) (Qiagen, [www.qiagen.com/ingenuity](http://www.qiagen.com/ingenuity)). RNA-seq data were deposited in the NCBI Gene Expression Omnibus database with accession number GSE73604. Real-time PCR (qPCR) analysis ----------------------------- Total RNA extracted from microdissected samples was amplified using the RiboAmp HS PLUS RNA Amplification kit (Applied Bisosytems) according to the manufacturer's protocol. RNA expression was semi-quantified by SYBR GreenER (Thermo Fisher Scientific) using the 7900HT Fast Real-time PCR System (Thermo Fisher Scientific). Primer sequences are listed in [Supplementary Table 1](http://erc.endocrinology-journals.org/cgi/content/full/ERC-17-0006/DC1) (see section on [supplementary data](#supp1){ref-type="supplementary-material"} given at the end of this article). The target gene expression was normalized with endogenous *Rpl19* level, and the relative change in transcript level was calculated using the delta-delta-CT method ([@bib31]). Bisulfite sequencing analysis ----------------------------- Sections from PND21 mammary glands in female offspring (*n* = 4--6 litter/group, one offspring/litter) were used for genomic DNA (gDNA) extraction and for bisulfite sequencing analysis according to our published protocol ([@bib64]). TCGA survival analysis ---------------------- RNA-seq analysis (RNA-seqV2) of The Cancer Genomic Atlas (TCGA) breast cancer samples as well as patient clinical data were downloaded (<http://cancergenome.nih.gov>) on 3/8/2016. The original TCGA breast cancer data set consists of RNA-seq data from 1215 samples. Data on tumor adjacent normal (113), metastatic tumor (7) and male breast cancer samples (12) were excluded. One sample was discarded due to lack of survival data. In this study, only 1082 female samples (746 Caucasian, 180 African American, 61 Asian, 1 American Indian or Alaska Native and 94 unknown) with ER status (795 ER+, 237 ER− and 50 indeterminate or N/A) were used for survival analyses. Normalized data of seven genes identified and verified in the LCM study (*ALDH1B1*, *ASTL*, *CA7* (official gene symbol for carbonic anhydrase VII in human), *CPLX4*, *KCNV2*, *MAGEE2* and *TUBA3E* (equivalent to *Tuba3A* in rat)) were variance stabilizing-transformed before hierarchical clustering with complete linkage based on the Euclidean distance between genes was performed to dichotomize the cohort into two groups. Survival data including days-to-last follow-up and days-to-death were extracted from TCGA clinical data. Overall survival between two groups was analyzed by Kaplan--Meier plot with log-rank test, and Cox proportional hazard models (with age at initial pathologic diagnosis adjusted) were used to analyze time to death. All analyses were carried out using survival package in R (<http://www.r-project.org>). To further determine the characteristics between the two groups, clinical features including age, estrogen receptor status, progesterone receptor status, cancer stage, cancer recurrence as well as lymph node positivity were compared using Student *t*-test (continuous variables) or Pearson's chi square test with contingency tables (categorical variables). The tests were conducted using GraphPad Prism 5.0. Statistical analysis -------------------- Gene expression and bisulfite sequencing analyses were performed using GraphPad Prism 5.0. For gene expression analyses, data were expressed as mean ± standard error of mean ([s.e.m.]{.smallcaps}) and analyzed by Student *t*-test and *post hoc* Mann--Whitney test. Bisulfite sequencing data were expressed as % methylation and analyzed using two-way ANOVA and Bonferroni post-test. *P* \< 0.05 was considered as statistically significant when compared between and among groups. Results ======= Gestational high-fat intake in addition to bisphenol A exposure increases mammary tumor incidence and effects mammary gland morphology -------------------------------------------------------------------------------------------------------------------------------------- To test our hypothesis that maternal exposure to BPA in addition to HBF intake during pregnancy leads to increased incidence of mammary cancer in offspring, we fed female SD rats (F0) with a modified AIN-93G HBF diet containing 39% kcal from butterfat, during mating and throughout gestation, in the presence and absence of BPA exposure ([Fig. 1](#fig1){ref-type="fig"}). To evaluate dose--response, we treated dams with 0, 2.5, 25, 250 or 2500 µg BPA/kg body weight (BW) everyday starting from 2 weeks before conception until the end of the gestational period. Offspring were maintained on control AIN-93G diet for the duration of the experiment. *In utero* exposure to all BPA doses while fed a HBF diet did not affect litter size of the dams or significantly alter the body weights of 2-, 7-, 14-, 21-, 35- and 50-day-old female offspring when compared with either the HBF-alone or control diet groups (data not shown). Similarly, estrous cyclicity of adult female offspring (data not shown) and serum concentrations of 17β-estradiol, progesterone, leptin and adiponectin in PND21 and PND50 female offspring were not significantly different among control and HBF diet groups ([Supplementary Table 3](http://erc.endocrinology-journals.org/cgi/content/full/ERC-17-0006/DC1)). Interestingly, exposure to HBF plus 2.5 µg/kg BW/day BPA significantly delayed (by \~1.5 days) the onset of puberty in the female offspring when compared with the HBF-alone group, as determined by vaginal opening ([Supplementary Table 3](http://erc.endocrinology-journals.org/cgi/content/full/ERC-17-0006/DC1)). Maternal exposure to a HBF diet alone, however, did not alter the time to vaginal opening in offspring when compared with the control diet group. Using a well-established chemically induced mammary cancer model, we treated female offspring with the carcinogen, DMBA, on PND50 ([@bib53], [@bib16]). Using palpable tumor as the endpoint, we found no significant difference in average time to first tumor appearance (i.e. latency) ([Supplementary Fig. 1](http://erc.endocrinology-journals.org/cgi/content/full/ERC-17-0006/DC1)A), number of palpable tumors (i.e. multiplicity) ([Supplementary Fig. 1](http://erc.endocrinology-journals.org/cgi/content/full/ERC-17-0006/DC1)B) or tumor volume (data not shown) in DMBA-treated rats gestationally exposed to control AIN-93G or HBF diet groups ([Supplementary Fig. 1](http://erc.endocrinology-journals.org/cgi/content/full/ERC-17-0006/DC1)A and B). In contrast, offspring of dams exposed to HBF and 25 µg/kg BW/day BPA during pregnancy had significantly higher mammary tumor incidence (90%) compared with HBF-alone controls (45.5%) (*P* \< 0.043) ([Fig. 2A](#fig2){ref-type="fig"}) as well as a significantly shorter tumor-free survival time (*P* = 0.0422) ([Fig. 2B](#fig2){ref-type="fig"}). Offspring exposed to a control diet with 10% kcal butterfat showed no significant difference in mammary tumor incidence in the absence vs presence of 25 µg/kg BW/day BPA (data not shown). Figure 2Gestational exposure to high butterfat and bisphenol A increases mammary tumor incidence and effects mammary gland morphology. (A) Tumor incidence (percentage of rats that developed at least one tumor) at PND140, in DMBA-treated offspring fed a control (AIN-93G) diet (Ctrl), or a high-butterfat (HBF) diet in the presence or absence of bisphenol A (BPA) at various concentrations. (B) Time (days) post DMBA treatment to first palpable tumor in HBF-alone vs HBF + BPA25 (µg/kg BW/day) groups (log-rank test, *P* = 0.0422). (C) Number of terminal end buds (TEBs) in PND21 mammary glands, in offspring fed a Ctrl, or a HBF diet in the presence or absence of BPA at various concentrations. Data are expressed as mean ± [s.e.m]{.smallcaps}. \*\**P* \< 0.01, \*\*\**P* \< 0.001 vs HBF, two-way ANOVA. (D) Representative whole mount images of PND21 mammary glands showing TEBs in offspring fed a HBF diet (left) and a HBF diet with 25 µg/kg BW/day BPA (right). Scale bar: 50 mm. Corner inset is a high magnification view of boxed area. Arrowheads mark the location of TEBs. High-fat intake during pregnancy has been shown to increase offspring' mammary cancer risk by altering mammary gland development, mainly by increasing the number of terminal end buds (TEBs) ([@bib20]). We examined the number of mammary gland TEBs on PND21. In contrast to published results using a corn oil high-fat diet ([@bib8]), our HBF group did not show any significant effect on TEB number when compared with the control AIN-93G diet group ([Fig. 2C](#fig2){ref-type="fig"}). However, when dams were gestationally exposed to 25 µg or 250 BPA/kg BW/day in addition to a HBF diet, the number of TEBs was significantly increased in the mammary glands of offspring at PND21 ([Fig. 2C](#fig2){ref-type="fig"} and [D](#fig2){ref-type="fig"}). Transcriptome analysis identifies HBF BFBPA-regulated genes associated with cancer-related signaling ---------------------------------------------------------------------------------------------------- To gain insight into how gestational HBF intake in addition to BPA exposure (25 µg/kg BW/day) modulates mammary tumor incidence in offspring, we performed genome-wide transcription analysis on laser capture microdissected (LCM) epithelia of PND21 mammary glands. Using this approach, we identified 504 differentially expressed genes (*P* \< 0.05) between the HBF-alone and HBF + BPA (25 µg/kg BW/day) diet groups. Hierarchical clustering clearly segregated the differentially expressed genes into two groups, HBF and HBF + BPA25 (µg/kg BW/day) ([Fig. 3A](#fig3){ref-type="fig"}). Figure 3Transcriptome analysis identifies high butterfat intake and bisphenol A exposure-regulated genes associated with cancer-related signaling. (A) Hierarchical clustering analysis of 504 genes with *P* \< 0.05 from genome-wide transcription analysis of laser capture microdissected epithelia of PND21 mammary glands from offspring fed a high butterfat (HBF) diet vs a HBF diet with 25 µg/kg BW/day bisphenol A (BPA). Using an unbiased gene clustering method, the heat map shows that differential genes are clearly segregated between samples into two distinct groups. Ingenuity pathway analysis of the 504 genes identified two cancer-related networks associated with the HBF + BPA vs HBF-alone group: (B) extracellular signal-regulated kinase (ERK) rapid signaling and (C) androgen receptor (AR) signaling. (D) Merging of these networks identified AR to be the key node. Green represents low expression; red represents high expression. To identify biological processes related to the identified dietary exposure-associated genes, we performed pathway analysis by interrogating the knowledge-based Ingenuity Pathway Analysis (IPA) database (Qiagen, [www.qiagen.com/ingenuity](http://www.qiagen.com/ingenuity)). Interestingly, the top two networks were found to be cancer related, including 'Cancer, Cellular development, Embryonic development' and 'Gene expression, Cancer, and Organismal injury and Abnormalities'. There were 25 genes involved in each network. One was associated with extracellular signal-regulated kinase (ERK) rapid signaling ([Fig. 3B](#fig3){ref-type="fig"}) and the other was related to androgen receptor (AR, [Fig. 3C](#fig3){ref-type="fig"}). When these networks were merged, AR appears to be the key node for those two cancer-related networks ([Fig. 3D](#fig3){ref-type="fig"}). Follow-up analysis of the gene expression array study was then performed using qRT-PCR for the ten most upregulated (*Calcb*, *Msl3l2*, *Aldh1b1*, *Spert*, *Astl*, *Magee2*, *Tuba3a*, *LObib52614*, *Cplx4* and *Kcnv2*) and ten most downregulated (*Olr1229*, *Olr791*, *Fam46d*, *Olr788*, *Olr984*, *Olr750*, *Olr51*, *Ccr9*, *Car7* and *Olr830*) genes ([Table 1](#tbl1){ref-type="table"}). Notably, there was a panel of seven olfactory receptors (Olr) found to be downregulated by BPA exposure. Of the twenty differentially expressed genes, we found 12 genes (*Aldh1b1*, *Astl*, *Cplx4*, *Kcnv2*, *LOC502684*, *Magee2*, *Tuba3a*, *Car7*, *Olr788*, *Olr830*, *Olr791* and *Olr1229*) were differentially expressed in the amplified LCM samples using qPCR analysis ([Supplementary Fig. 2](http://erc.endocrinology-journals.org/cgi/content/full/ERC-17-0006/DC1)). Table 1Top differentially expressed genes in mammary glands of offspring gestationally exposed to high-butterfat ± bisphenol A.**Putative regulatory CpG island^c^Gene symbolGene nameLog 2 fold change**PresenceLocationUpregulated genes *Calcb*^a^Calcitonin-related polypeptide, beta6.6NoNot applicable *Msl3l2*^a^Male-specific lethal 3-like 25.5YesExon 2 *Aldh1b1*^a^Aldehyde dehydrogenase 1 family, member B15.0Yes5′ promoter; TSS; Exon 1 *Spert*^a^Spermatid associated4.8NoNot applicable *Astl*^a^Astacin-like metallendopeptidase (M12 family)4.6Yes\>5 kb upstream of TSS; 3′ end, \>5 kb downstream of Exon 10 *Magee2*^a^Melanoma antigen, family E, 24.3NoNot applicable *Tuba3a*^a^Tubulin, alpha 3A4.3NoNot applicable *LOC502684*Hypothetical protein LOC5026843.5Yes\>10 kb upstream of TSS *Cplx4*^a^Complexin 43.4Yes3′ end; \>15 kb downstream of Exon 3 *Kcnv2*^b^Potassium channel, subfamily V, member 23.2YesExon 1Downregulated genes *Olr1229*Olfactory receptor 1229−17.6NoNot applicable *Olr791*Olfactory receptor 791−9.8NoNot applicable *Fam46d*^a^Family with sequence similarity 46, member D−8.5NoNot applicable *Olr788*Olfactory receptor 788−7.3NoNot applicable *Olr984*Olfactory receptor 984−7.0NoNot applicable *Olr750*Olfactory receptor 750−6.6NoNot applicable *Olr51*Olfactory receptor 51−6.6NoNot applicable *Ccr9*^a^Chemokines (C--C motif) receptor 9−6.4NoNot applicable *Car7*^b^Carbonic anhydrase 7−6.2Yes5′ promoter; TSS; Exon 1 *Olr830*Olfactory receptor 830−6.0NoNot applicable[^2][^3][^4] *In utero* HBF BFBPA exposure dysregulated gene expression in PND21 mammary glands through DNA methylation ---------------------------------------------------------------------------------------------------------- We next investigated whether the dysregulation of genes in the mammary glands of HBF + BPA offspring was associated with aberrant epigenetic regulation. The most common epigenetic alteration is methylation of cytosine in CpG dinucleotides in a gene's promoter region, resulting in alterations of gene expression. Thus, we first checked whether a putative CpG island could be found near the transcription start site of each gene using the rat genome database from The University of California Santa Cruz (RGSC 5.0/rn5) (<https://genome.ucsc.edu/cgi-bin/hgGateway>). Six of the top upregulated genes (*Msl3l2*, *Aldh1b1*, *Astl*, *LOC502684*, *Cplx4* and *Kcnv2*) showed at least one CpG island within their gene locus ([Table 1](#tbl1){ref-type="table"}). In contrast, only one gene of the top ten downregulated genes, *Car7*, has a putative CpG island next to their 5′ regulatory region ([Table 1](#tbl1){ref-type="table"}). We then selected *Car7* ([Supplementary Fig. 3](http://erc.endocrinology-journals.org/cgi/content/full/ERC-17-0006/DC1)), and *Kcnv2* ([Supplementary Fig. 4](http://erc.endocrinology-journals.org/cgi/content/full/ERC-17-0006/DC1)), the only upregulated gene with a CpG island in close proximity to its regulatory region, for bisulfite sequencing analyses. As shown in [Fig. 4A](#fig4){ref-type="fig"}, significant hypermethylation was observed in the CpG island of *Car7* in the BPA-exposed group when compared with the control group (*P* \< 0.0001, two-way ANOVA). In contrast, the level of methylation in the CpG island of *Kcnv2* was significantly (*P* = 0.0068, two-way ANOVA) reduced after *in utero* exposure to BPA (25 µg/kg BW/day) ([Fig. 4B](#fig4){ref-type="fig"}). These methylation changes were inversely correlated with the changes of gene expression as shown in the qPCR experiments ([Supplementary Fig. 2](http://erc.endocrinology-journals.org/cgi/content/full/ERC-17-0006/DC1)). Figure 4*In utero* high butterfat and BPA exposure alters DNA methylation level of the CpG island in *Kcnv2* and *Car7.* Bisulfite-genomic sequencing was conducted to interrogate differential DNA methylation in (A) *Car7* and (B) *Kcnv2*, in PND21 mammary glands of offspring fed a HBF diet vs a HBF diet with 25 µg/kg BW/day BPA. Percentage of methylation sites in the CpG island of *Car7* (53 CpG sites) and *Kcnv2* (27 CpG sites) were calculated. Each dot represents the average methylation percentage of each site. Mean (middle bar) and standard deviation (upper and lower bars) are represented in each group. Two-way ANOVA was performed to determine the difference between two groups. Top HBF BFBPA-dysregulated genes are associated with poor overall survival in TCGA breast cancer cohort ------------------------------------------------------------------------------------------------------- To gain clinical significance, using The Cancer Genomic Atlas (TCGA) breast cancer cohort, we dichotomized 1082 breast cancer subjects into two groups based on the expression of the seven genes (*ALDH1B1*, *ASTL*, *CA7*, *CPLX4*, *KCNV2*, *MAGEE2* and *TUBA3E*). As it is yet to be determined whether the identified top dysregulated genes are involved in human breast cancer, we selected seven out of the 12 that were clearly defined in humans for survival data analysis in TCGA cohort. Interestingly, we found those seven genes can be used to predict a group of 655 patients (Group 2, [Fig. 5](#fig5){ref-type="fig"} left panel) with poor overall survival in the cohort (*P* = 0.0201). Stratifications on race and patients with estrogen receptor (ER)-positive tumors showed that these seven genes have a better prognostic value in Caucasian patients (*P* = 0.00368, [Fig. 5](#fig5){ref-type="fig"} middle panel) as well as in Caucasian patients with ER-positive breast cancer (*P* = 0.00033, [Fig. 5](#fig5){ref-type="fig"} right panel). Further analysis of Caucasian patients with ER-positive breast cancer suggested that the patients with poor overall survival (Group 2) have significantly less progesterone receptor expression (odds ratio = 3.682, *P* \< 0.0001). All other parameters examined including age, lymph node positivity, cancer stage and cancer recurrence showed no statistical difference between the two groups of patients. Interestingly, expression of four genes (*ALDH1B1*, *ASTL*, *CA7* and *TUBA3E*) of the seven gene panel was significantly different between the two groups of human breast cancer subjects (data not shown). The most significantly differentially expressed human genes (*ASTL* and *TUBA3E*) were also found to be upregulated in the poor overall survival group, similar to that of *Astl* and *Tuba3a* in rat mammary glands of the HBF + BPA group. Figure 5Top high butterfat + bisphenol A-dysregulated genes are associated with poor overall survival in breast cancer patients. Seven differentially expressed BPA genes in TCGA RNA-seq expression data were used to stratify a TCGA breast cancer cohort into two groups using unbiased hierarchical clustering. Survival analyses with log-rank test as well as multivariate survival analyses with Cox's regression model (adjusted with age at pathological analysis) based on overall survival data available in TCGA were performed. Patients in Group 1 show significantly better overall survival in all patients (left panel), Caucasian patients (middle panel) and Caucasian patients with ER-positive tumors (right panel) compared with group 2. Discussion ========== The primary aim of this study was to determine if maternal exposure to low doses of BPA (2.5--2500 µg/kg BW/day) in combination with a HBF diet modifies mammary gland development and increases the risk of mammary tumorigenesis in first-generation offspring. Our major findings were that as a result of gestational intake of a diet containing 39% kcal from butterfat, exposure to BPA at 25 µg/kg BW/day was most effective in increasing mammary tumor incidence to 90% compared with 45.5% when BPA was not present in the diet. The dose--response curve was non-monotonic and the effective dose was 200-fold and 2000-fold lower than the current no-observed adverse effect level (NOAEL; 5 mg/kg BW/day) and the lowest-observed adverse effect level (LOAEL; 50 mg/kg BW/day), respectively, as established by the US Environmental Protection Agency (EPA) ([@bib2]). Mammary tumor incidence data correlated with significant increases in TEBs observed for the two lowest doses (2.5 and 25 μg/kg BW/day). In fact, in the present study, the BPA dose (25 μg/kg BW/day) that elicited the greatest mammary tumor incidence is 2-fold less than the US EPA's daily tolerable oral reference dose of 50 μg/kg BW/day for human exposure ([@bib2]), while the lowest effective dose causing increased mammary TEBs is 20-fold lower than the tolerance dose for humans. In rodent models, BPA has been reported to alter the development of the mammary gland at the biochemical, cellular and tissue levels of organization, in manners suggestive of a heightened risk for mammary carcinogenesis ([@bib35], [@bib41], [@bib11]). Betancourt and coworkers ([@bib3]) reported that a prenatal BPA dose of either 25 or 250 μg/kg BW/day together with a standard diet had no tumorigenic effect after DMBA exposure at PND50. However, DMBA-induced carcinogenesis at PND100 resulted in a significantly increased number of terminal ducts ([@bib40]) as well as a higher incidence of mammary tumors in rats exposed prenatally to 250 μg/kg BW/day BPA ([@bib3]), suggesting that BPA at 250 μg/kg BW/day could potentially shift the window of susceptibility for chemically induced mammary cancer from PND50 to PND100. To the best of our knowledge, no previous studies have investigated the impact of maternal consumption of a HBF diet together with BPA exposure on breast cancer risk in offspring. In our current study, we observed that concurrent BPA (25 μg/kg BW/day) exposure with a HBF diet background significantly increases mammary tumor incidence in female offspring treated with DMBA at PND50. We also observed significant increases in the number of TEBs in the mammary glands of PND21 rats born to dams fed a HBF diet and 2.5 or 25 μg/kg BW/day BPA during pregnancy. These effective BPA doses that alter mammary gland morphology and cancer susceptibility are 100-fold or 10-fold, respectively, lower than the 250 μg/kg BW/day BPA dose in the previously reported study ([@bib3]). Additionally, under the HBF diet, the window of susceptibility to DMBA-induced carcinogenesis was reversed back to PND50 in our study. Perhaps, the HBF diet prevents the window-shifting effect caused by BPA, which may make the mammary glands more vulnerable to lower doses of BPA. In addition, independent studies of prenatal exposure to BPA or high-fat diets have observed alterations in the pace at which mammary gland differentiation occurs at different stages of development ([@bib3], [@bib23], [@bib56]). Generally, perinatal administration of EDCs causes accelerated development of the mammary gland associated with increased proliferation and a higher number of TEBs at PND21 ([@bib26]). Therefore, a greater availability of target structures, in addition to a cellular microenvironment favoring carcinogenesis, could explain the increased tumorigenic response ([@bib50], [@bib17], [@bib9]). Contrary to our expectation, we did not observe a higher mammary tumor risk in female offspring born to dams consuming a HBF diet (39% kcal) alone during pregnancy when compared with those born to dams fed an AIN-93G control diet (10% kcal). Furthermore, a trend of reduced mammary tumor incidence (45%) in the offspring exposed *in utero* to a HBF diet vs an AIN-93G control diet (60%) also correlated with a trend of reduced TEB count. Taken together, these findings suggest that offspring born to dams fed a HBF diet were at a reduced risk of developing mammary tumors following DMBA treatment at PND50. Our results are consistent with a previous study ([@bib9]), which observed that exposure to a lard-based high-fat diet during fetal and lactation periods decreases mammary cancer susceptibility in adulthood in rats. Although the source of high fat differed between studies, the fatty acid profile of the high-fat diet used in both studies consisted of saturated (mainly palmitic (16:0) and stearic acids (C18:0)) and monounsaturated (mainly oleic acid (18:1 n-9)) fatty acids, comprising 37% and 38% total fatty acids, respectively ([@bib9]). This is in contrast to published studies that associate *in utero* exposure to a corn oil-based high-fat diet alone, which contains polyunsaturated fatty acids, with an increased susceptibility to mammary tumorigenesis among female offspring ([@bib20], [@bib7], [@bib8]). These results suggest that future studies focusing on dietary fat composition during pregnancy are warranted. Diet is estimated to contribute to the etiology of 30--50% of all breast cancers; however, the mechanism(s) by which dietary patterns or EDCs modify breast cancer risk is not fully understood ([@bib25], [@bib1], [@bib17]). In this study, we showed that gestational exposure to BPA with HBF diet intake dysregulates early cancer-related gene expression even before cancer development, implying that the EDC predisposes cancer risk by reprogramming gene expression in mammary glands. Some of these dysregulated gene events are mediated through epigenetics, DNA methylation in particular. We and others found that alterations in the fetal environment have caused persistent modification in gene expression and susceptibility of disease ([@bib24], [@bib17]). Maternal exposure to BPA during pregnancy increases offspring's risk of prostate cancer and induces hypomethylation of phosphodiesterase type 4 variant 4 (PDE4D4), an enzyme responsible for cyclic AMP breakdown ([@bib24]). Furthermore, a high-fat- or ethinyl-estradiol-supplemented maternal diet has been shown to increase mammary cancer risk in several generations of offspring and is associated with changes in the DNA methylation machinery and methylation patterns in mammary tissue ([@bib8]). In the present study, we observed significant change in the methylation status of the promoter region of *Car7* and *Kcnv2*, concomitantly with significant change in their gene expression level, although both genes have never been reported to be breast cancer related in humans. *CA7*, carbonic anhydrase VII, catalyzes the hydration of carbon dioxide into bicarbonate and proton. It is one of the active isoforms found in cytosolic compartment. Although the mechanistic role of *CA7* in cancer is not much known, it has been speculated that the absence of the antioxidant property of *CA7* could contribute to disease progression ([@bib39]). Its tumor-suppressive potential was identified in colorectal cancer as reduced *CA7* level was associated with shorter disease-specific survival ([@bib63]). *KCNV2*, voltage-gated K+ channel subunit gene family V member 2, can form functional heterotetramers with Kv2 subunits and influence membrane translocation and channel properties ([@bib42], [@bib6]). Its role in cancer is largely unclear and only its close relative, Kv9.3, was known to support the growth of colon, lung and uterine cancer cells ([@bib58], [@bib57], [@bib30]). We also found that a group of olfactory genes in the mammary gland of PND21 female offspring were downregulated in HBF + BPA group when compared with HBF only. Interestingly, the fetal gene expression profile of olfactory receptors has been shown to be modulated by slight nutrient changes in the maternal diet ([@bib48]). Altered expression of genes involved in promoting cell proliferation in the offspring's mammary gland during development of offspring exposed to maternal dietary fat have been found to increase susceptibility to cancer in adulthood ([@bib17]). Using Ingenuity Pathway Analysis, we identified the top biological processes related to the identified dietary exposure-associated genes to include 'Cancer, Cellular development, Embryonic development' and 'Gene expression, Cancer, and Organismal injury and Abnormalities'. One was associated with ERK rapid signaling and the other was related to AR. When these networks were merged, AR was the key node for the two cancer-related networks, suggesting that aberrant activation of these signaling pathways may play a key role in adult mammary tumor risk. Interestingly, recent investigations have identified the AR signaling pathway as a target for breast cancer treatment, with several clinical trials currently ongoing ([@bib46]). In this study, we identified HBF + BPA-related gene dysregulation in developing mammary glands. Although we cannot rule out in the current study that BPA, in the context of a control diet, may dysregulate some of the same HBF + BPA-related genes, offspring exposed to a control diet showed no significant difference in mammary tumor incidence in the absence vs presence of 25 µg/kg BW/day BPA. Pathway analyses suggested that the majority of the identified HBF + BPA-related genes are classified as 'cancer' associated, but little information has been reported about their roles in cancer development. We took advantage of the publicly available RNA-seq and survival data from TCGA and determined that seven differentially expressed genes (*ALDH1B1*, *ASTL*, *CA7*, *CPLX4*, *KCNV2*, *MAGEE2* and *TUBA3E*) could be involved in breast cancer development, especially in Caucasian female patients with ER positivity. It is intriguing to see that the 'signature', which is composed of 7 genes, could be a significant prognostic indicator for breast cancer patients even though the role of each gene in breast cancer remains largely unknown. Therefore, genome-wide data together with patient-based survival analyses provide an effective way to reveal novel genes/pathways involved in cancer development. However, how those genes are linked to ER signaling and whether those genes functions as oncogenes or tumor suppressor genes in a specific genetic background require more detailed investigation in future. In conclusion, our data reinforce findings that maternal diet during pregnancy can determine the susceptibility of offspring to the development of breast cancer in adult life ([@bib17], [@bib8], [@bib9]). Importantly, we found that concurrent exposure to a HBF diet and BPA below the current NOAEL during pregnancy modulates developmental morphology and gene expression in the prepubertal mammary gland and increases the breast cancer incidence in offspring. It is apparent from our findings that the complex interplay of diet and environmental exposure to an endocrine disrupter can reprogram the developing mammary gland permitting a permissive environment for adult breast cancer risk in the first-generation offspring. Supplementary Material {#supp1} ====================== ###### Supporting Figure 1 ###### Supporting Figure 2 ###### Supporting Figure 3 ###### Supporting Figure 4 ###### Supporting Table 1 ###### Supporting Table 2 ###### Supporting Table 3 Declaration of interest ======================= The authors declare that there is no conflict of interest that could be perceived as prejudicing the impartiality of the research reported. Funding ======= This study was supported in part by grants from the National Institutes of Health: U01ES019480, U01ES020988, U54HL127624, RC2ES018765, P30ES006096 and P30ES025128, and the United States Department of Veterans Affairs: 101BX000675. Author contribution statement ============================= Experimental design: Y K L, G V, S B and S M H; performing experiments: Y K L, G V, A C, D S, X Z, A K, R G; data analyses: Y K L, G V, A C, D S, X Z, J Y, M M, S M H; writing manuscript: Y K L, G V, A C, J V, J Y, A K, M M, S M H. Special thanks to Huan Xu for providing initial technical support for RNA-seq data analyses and Dr Neville Tam for technical assistance in cryosectioning of the samples. [^1]: (S Belcher is now at Department of Biological Sciences, Raleigh, North Carolina, USA) [^2]: Genome-wide transcription analysis on laser capture microdissected epithelia of PND21 mammary glands identified 504 differentially expressed genes (*P* \< 0.05) between high-butterfat (HBF)-alone and HBF + bisphenol A (BPA, 25 µg/kg body weight/day) diet groups. The top ten differentially up- and downregulated genes are listed. [^3]: Rat genes that are homologous to human genes; ^b^genes with a promoter CG-rich region analyzed using bisulfite sequencing analysis; ^c^the presence and location of the putative regulatory CpG island were predicted by the UCSC Genome Browser. [^4]: TSS, transcription start site.

INTRODUCTION {#sec1-1} ============ To date, it is understood that around half of the patients suffering from isolated palatal clefts may exhibit additional clinical signs on examination, thus being regarded as syndromic.\[[@ref1]\] Particular syndromes might be present in Pierre Robin sequence (PRS) children presenting with isolated cleft palates and, thus, the clinical picture of PRS therefore needs to be further subdivided into syndromic and into nonsyndromic. Due to coexisting anomalies of various organs and structures as well as disorders of the metabolism and/or the immune system, syndromic PRS patients present particular problems to their interdisciplinary treatment team, especially with regard to surgical reconstructive procedures and perioperative management. The second part of this series addresses the incidence of syndromic and nonsyndromic cases, feeding, and breathing problems as well as the mortality rate associated with a series of 266 PRS patients and 1518 isolated hard and/or soft palate cleft cases. SUBJECTS AND METHODS {#sec1-2} ==================== A total of 266 PRS cases were identified and their charts were extracted for investigation, from an overall database of 4185 cleft cases at the cleft clinic. In addition to the PRS cases, a total of 1518 isolated cleft palate (ICP) cases were found in the database. The remainder of the ICP did not exhibit any other distinguishing signs, other than a hard and/or soft palate cleft and, therefore, represented a specific cleft division of their own. For further in-depth analysis, the PRS cases were subdivided into 21 Siebold--Robin sequence (SRS) cases and 245 Fairbairn--Robin triad (FRT) cases according to their specific clinical signs.\[[@ref2]\] Both subgroups, together with the ICP cases, were subsequently investigated for being either syndromic or nonsyndromic based on the clinical presence of additional signs on examination. As the mortality rate in PRS patients is considered to be significant due to apnea and/or aspiration of fluid or food, the charts were screened for continuous feeding and breathing problems due to glossoptosis. The family history was carefully examined for commonalities, searching for genetic causality in addition to that of environmental exposure. RESULTS {#sec1-3} ======= A total of 266 PRS and 1518 ICP cases were included in this retrospective analysis. Among the 266 PRS cases, 7.9% presented without a cleft palate on examination, therefore being diagnosed as SRS. As shown in [Table 1](#T1){ref-type="table"}, 90.5% of SRS cases were nonsyndromic and 9.5% were diagnosed to be syndromic SRS revealing additional clinical signs such as congenital heart defects or myopathy, as well as two cases of spondyloepimetaphyseal dysplasia. ###### Syndromic versus nonsyndromic cases and occurrence of genetic findings not allocable to any known syndromes  Of PRS group, 92.1% disclosed a cleft palate on examination, in addition to micrognathia and a compromised airway due to glossoptosis. These cases were allocated to the subgroup of FRT and 20.8% of them presented with a syndrome, which is more than double that of the SRS subgroup. An overall of 21 different syndromes in combination with FRT could be detected. The following syndromes occurred in descending order: Stickler (22); Binder (9), among which Stickler and Binder were combined (4); van der Woude (6); Moebius (3); oromandibular hypogenesis (1); Hirschsprung (1); Miller (1); Catel--Manzke (1); fetal alcohol (1); Klippel--Feil (1); oro-palatal digital (1); otomandibular dysostosis (1); popliteal pterygium (1); Wolf--Hirschhorn (1); chromosome 5 + 11 (1); Richter-Costa-Perreira (1); Spondyloepimetaphyseal dysplasia with joint laxity (SEMDL) (1); atelosteogenesis (1); and congenital myopathy (1). In addition, FRT occurred with other genetic abnormalities not being allocable to known syndromes, mostly presenting with additional symptoms related to congenital heart, gastrointestinal, and urinary tract defects, as well as ear and musculoskeletal system defects such as club feet and various hand deformities, including syndactyly. Therefore, roughly 20% of those initially nonsyndromic FRT-labeled patients disclosed genetic abnormalities with no association to already established syndromes. Of the 1518 ICP cases, 26.9% were syndromic. The most common coexisting syndromes in ICP patients in this database were the Demarquay-van der Woude, Treacher-Collins, Binder, and Sphrintzen (22q11.2-deletion) syndromes. Compared to SRS and FRT, however, slightly less nonsyndromic ICP cases (15.8%) revealed genetic abnormalities not being allocable to known syndromes. [Table 1](#T1){ref-type="table"} summarizes the numbers and percentages of syndromic and nonsyndromic cases of ICP and PRS and includes those nonsyndromic cases where genetic abnormalities occurred that were not connected to specific syndromes. A positive family history for clefts may indicate that an environmental exposure is not the only contributing factor in the development of FRT and SRS. Whereas a positive family history for clefts was detected in over 20% of the total cleft database and within the FRT subgroup, it was not so frequent in ICP cases (18%) and rather rare in SRS cases with only 9.5%. While around 40% of FRT cases suffered from long-term or persistent breathing and feeding difficulties, the figure was nearly doubled in the SRS subgroup. Analysis of the mortality rate detected that no patient of the SRS subgroup demised whereas the mortality rate in the FRT subgroup was even higher than the one in the overall cleft database. [Table 2](#T2){ref-type="table"} provides the summary of the results discussed above. ###### Family history, breathing and feeding problems, mortality rate  DISCUSSION {#sec1-4} ========== A database of 266 cases, initially labeled as PRS, was analyzed in detail, resulting in a subsequent subdivision of a major FRT and minor SRS group where clinically no soft and/or hard palate cleft was found. Due to the amount of cases in which concomitant known syndromes could be detected in this database, a further subclassification of the two groups into syndromic and nonsyndromic has been performed. In 1911, Shukowsky\[[@ref3]\] described a deformational PRS group including isolated cases, without any associations to syndromes. He further pointed out that the deficient mandibular growth in those cases might be attributed to physical or mechanical restraint. He then suggested to denominate PRS cases associated with syndromes, chromosomal abnormalities such as deletions or duplications, teratogens, or neuromuscular diseases, and these accounted for 60% of his cases as deformational. A similar attempt to classify PRS cases into "isolated PRS = IRS" and "PRS as part of a syndrome = RSS" has been proposed in 1984 by Pasyayan and Lewis.\[[@ref4]\] They distinguished between nonsyndromic (probably originating from various etiologies) and syndromic cases that may be inherited by similar ways as the concomitant syndromes. They further suggested the use of two terminologies: (1) Sequence, indicating the sequence of events during the pathogenesis in nonsyndromic cases and (2) syndrome, for cases with an accompanying known syndrome. In a more recent analysis of 117 PRS cases in 2001, Holder-Espinasse *et al*. reported around 35% of their PRS cases as syndromic.\[[@ref5]\] They suggested a future subclassification of syndromic and nonsyndromic PRS cases to be of practical and clinical importance. This, aside from a different etiopathogenesis, might further allude to the prognosis and/or the potential of mandibular catch-up growth.\[[@ref6]\] Compared to those results, the database presented here revealed an average of 80.1% nonsyndromic FRT (79.2%) and SRS (90.5%) patients. Both SRS (23.8%) and FRT (20%) cases showed similar percentages relating to genetic findings that were not allocable to any known syndromes. FRT cases (20.8%) presented with more than double as many known associated syndromes compared to SRS cases (9.5%), among which Stickler as described in other patient series\[[@ref7][@ref8][@ref9]\] and Binder were the most common. Family members of FRT patients revealed nearly three times more clefts than those of SRS patients, keeping in mind the fact that SRS patients themselves lack the clinical sign of a cleft palate on examination. Somewhat more astonishing are the results that nearly double the amount of long-term feeding, and breathing difficulties can be found in SRS patients when compared to FRT patients. However, none of these SRS patients demised whereas the mortality rate in FRT patients (3.3%) is almost twice the amount of that in ICP patients. The latter might be explained by the possibility that in case of glossoptosis in FRT patients, the tongue might become trapped in the cleft palate leading to critical airway obstruction with potential fatality. Of major concern is that most fatal cases related to aspiration which occurred at home. The majority of the fatalities presented in this cleft database was patients suffering from holoprosencephaly (1.3%), Patau syndromes (0.5%), and FRT cases (0.2%). Most of the long-term breathing and feeding difficulties in SRS patients are probably due to an early cleft repair in FRT patients which corrects feeding disorders, installing a separation between the oral and nasal cavity and due to a more severe micrognathia in SRS patients where neuromuscular reasons\[[@ref3]\] eventually play a role. The patient group labeled as "PRS" contains a much wider spectrum of diversity than previously described. Cases with soft and/or hard palate clefts, in addition to micrognathia, glossoptosis, and breathing disorders, should be classified as FRT. Whereas breathing and feeding difficulties prevailed more often in SRS cases, and rather higher mortality rate was demonstrated among FRT cases, compared to no deaths in the SRS subgroup. Furthermore, the previous means of classification of PRS cases into syndromic and nonsyndromic remains accurate and applicable, given that 21 identified syndromes were associated with mainly the FRT subgroup. CONCLUSION {#sec1-5} ========== A future refinement in the classification of PRS cases is important. Both a subdivision into FRT and SRS and into syndromic and nonsyndromic cases will result in an adequate and appropriate early and subsequent definite individual treatment plan, thereby preventing unnecessary surgery and/or adverse comorbidities. Multiple FRT cases presented with various concomitant syndromes and genetic abnormalities, but only one type occurred in two SRS cases. The latter presented a significantly different mortality rate when compared to the FRT subgroup Financial support and sponsorship {#sec2-1} --------------------------------- Nil. Conflicts of interest {#sec2-2} --------------------- There are no conflicts of interest. The authors would like to thank Mrs Jennilee Blom for her invaluable job as research assistant.

1. Introduction {#sec1-marinedrugs-16-00376} =============== A growing body of research on drug resistance among pathogens \[[@B1-marinedrugs-16-00376],[@B2-marinedrugs-16-00376],[@B3-marinedrugs-16-00376],[@B4-marinedrugs-16-00376]\] and previously neglected or poorly understood infectious diseases \[[@B5-marinedrugs-16-00376],[@B6-marinedrugs-16-00376],[@B7-marinedrugs-16-00376],[@B8-marinedrugs-16-00376],[@B9-marinedrugs-16-00376]\] highlights the persistent need for efficient drug discovery efforts in nearly all infectious disease areas. Marine natural products have, historically, been important in the development of therapeutics for infectious diseases, and contemporary efforts in these areas continue to prove their value \[[@B10-marinedrugs-16-00376],[@B11-marinedrugs-16-00376],[@B12-marinedrugs-16-00376],[@B13-marinedrugs-16-00376],[@B14-marinedrugs-16-00376],[@B15-marinedrugs-16-00376],[@B16-marinedrugs-16-00376]\]. Herein, we report the combined results of screening a library of marine fungal extracts against: *Enterococcus faecium*, *Staphylococcus aureus, Klebsiella pneumoniae, Acinetobacter baumannii, Pseudomonas aeruginosa, Enterobacter cloacae*, (i.e., the ESKAPE pathogens), *Mycobacterium tuberculosis*, *Leishmania donovani*, and *Naegleria fowleri*. These diverse microbial targets represent some of the most important threats to human health today. While invertebrates remain a major source of new bioactive marine compounds, developments in microbial isolation, culture, and genome-sequencing techniques continue to expand the natural products frontier of marine microbial sources \[[@B13-marinedrugs-16-00376],[@B17-marinedrugs-16-00376],[@B18-marinedrugs-16-00376],[@B19-marinedrugs-16-00376],[@B20-marinedrugs-16-00376],[@B21-marinedrugs-16-00376],[@B22-marinedrugs-16-00376],[@B23-marinedrugs-16-00376],[@B24-marinedrugs-16-00376],[@B25-marinedrugs-16-00376],[@B26-marinedrugs-16-00376]\]. Mangrove forests are marine-margin habitats regarded for their microbial, chemical, and biological diversity \[[@B27-marinedrugs-16-00376],[@B28-marinedrugs-16-00376],[@B29-marinedrugs-16-00376],[@B30-marinedrugs-16-00376],[@B31-marinedrugs-16-00376]\]. While endophytic microorganisms are well studied in South-East Asian mangroves \[[@B27-marinedrugs-16-00376],[@B28-marinedrugs-16-00376],[@B30-marinedrugs-16-00376]\], the microbial inhabitants of the expansive mangrove forests in the Americas and the Caribbean remain largely unexplored. Surrounded by these "new world" mangroves in North America, our microbial library consists primarily of fungal isolates from mangrove and mangrove associated trees including: *Rhizophora mangle*, *Avicennia germinans*, *Laguncularia racemose*, *Conocarpus erectus*, and *Coccoloba uvifera*. Generally, endophyte isolation protocols are designed to target as many genera of fungi as possible. Nevertheless, because fungi are capable of regulating their biosynthetic pathways in response to variations in both natural and artificial growth conditions \[[@B32-marinedrugs-16-00376],[@B33-marinedrugs-16-00376],[@B34-marinedrugs-16-00376]\], many fermentation techniques have been developed to induce the production of previously undetected, bioactive chemistry in fungi \[[@B35-marinedrugs-16-00376],[@B36-marinedrugs-16-00376],[@B37-marinedrugs-16-00376],[@B38-marinedrugs-16-00376],[@B39-marinedrugs-16-00376],[@B40-marinedrugs-16-00376],[@B41-marinedrugs-16-00376],[@B42-marinedrugs-16-00376],[@B43-marinedrugs-16-00376],[@B44-marinedrugs-16-00376]\]. Most strategies, like co-culture or genome mining, focus on a single or small number of isolates or therapeutic targets. While these techniques are quite effective at exploiting the entire biosynthetic potential of an organism, they do not translate particularly well into larger-scale screening campaigns targeting multiple pathogens. To that end, we describe here methodology utilizing small molecule regulation of fungal transcription to enhance extract libraries for the expansion of screening capabilities ([Figure 1](#marinedrugs-16-00376-f001){ref-type="fig"}). In this study, we employed miniaturized culture conditions \[[@B30-marinedrugs-16-00376]\] and the use of histone deacetylase inhibitors (HDACi) and DNA methyltransferase inhibitors (DNMTi) to examine the effect of epigenetic regulators known to modulate secondary metabolite expression \[[@B32-marinedrugs-16-00376],[@B33-marinedrugs-16-00376],[@B34-marinedrugs-16-00376],[@B35-marinedrugs-16-00376],[@B36-marinedrugs-16-00376],[@B37-marinedrugs-16-00376],[@B38-marinedrugs-16-00376],[@B39-marinedrugs-16-00376],[@B40-marinedrugs-16-00376],[@B41-marinedrugs-16-00376],[@B42-marinedrugs-16-00376],[@B43-marinedrugs-16-00376],[@B44-marinedrugs-16-00376],[@B45-marinedrugs-16-00376]\]. From 530 fungal isolates, 1608 extracts were generated and screened against a panel of infectious disease targets ([Figure 1](#marinedrugs-16-00376-f001){ref-type="fig"}). Potency and cytotoxicity data were used to prioritize lead extracts, defined as those extracts of high interest which would be advanced to scale up and chemical analysis. Comparison among screening results and culture treatments has identified noteworthy trends. The data showcases the diversity and specificity of bioactive natural product extracts from North American mangrove endophytes and supports the use of epigenetic modification as a screening tool. 2. Results {#sec2-marinedrugs-16-00376} ========== 2.1. Biological Materials {#sec2dot1-marinedrugs-16-00376} ------------------------- Mangrove tissues (roots, stems, leaves, flowers) were sampled primarily from shoreline communities ([Figure 2](#marinedrugs-16-00376-f002){ref-type="fig"}) in Florida (Courtney Campbell Causeway, Tampa, FL (CC); Coquina Beach, Sarasota, FL (CQ); Everglades City, FL (EG); Howard Franklin Causeway, Tampa, FL (HF); Keys Marine Lab, Layton, FL (KML)) and Mexico (Tapachula, MX (TAP)), with contributions from opportunistic collections in other environments. Surface sterilized plant tissues placed on nutrient agar produced emergent hyphae that were clipped and purified through repeated streaking, yielding approximately 3000 endophytic fungal strains \[[@B45-marinedrugs-16-00376],[@B46-marinedrugs-16-00376],[@B47-marinedrugs-16-00376]\]. A selection of 530 strains were randomly chosen for these screening studies. 2.2. Extract Library {#sec2dot2-marinedrugs-16-00376} -------------------- The selected fungal strains were cultivated with and without modulators of epigenetic regulation, resulting in 530 extracts each for sodium butyrate (HDACi) treated, 5-azacytidine (DNMTi) treated and non-treated cultures (1608 total cultures) that were extracted with ethyl acetate and distributed into 17 96-well plates at 5 mg/mL in DMSO. 2.3. Screening {#sec3dot2-marinedrugs-16-00376} -------------- ### 2.3.1. ESKAPE Pathogens {#sec2dot3dot1-marinedrugs-16-00376} One or more of the six bacterial pathogens that constitute the ESKAPE panel of bacterial pathogens were sensitive to 203 total fungal extracts. Using serial dilutions of extracts of 200, 100, 50, 25, 10 and 5 µg/mL, scaled scoring (SS) \[[@B46-marinedrugs-16-00376]\] of minimal inhibitory concentrations (MICs) identified the most promising lead extracts. Bioactivity levels above SS 9 were chosen as extracts of interest. Filtering to remove cytotoxic extracts (J774 IC~50~ \< 5 µg/mL) produced a collection ([Table S1](#app1-marinedrugs-16-00376){ref-type="app"}) of 46 *lead extracts* (2.9% of all extracts tested), 24% of which had been cultured under control conditions, 37% under DNMTi conditions, and 39% under HDACi conditions. ### 2.3.2. *Mycobacterium tuberculosis* {#sec2dot3dot2-marinedrugs-16-00376} *Mycobacterium tuberculosis* was scored as percent growth inhibition (GI). Sensitivity measured as GI~50~ resulted in 540 fungal extracts with activity at 100 µg/mL. Filtering extracts to select for low cytotoxicity (J774 IC~50~ \> 5 µg/mL) and GI~95~ resulted in 100 (6.2% of all extracts tested) extracts as *lead extracts*, including 31% untreated extracts, 33% treated under DNMTi conditions, and 36% treated under HDACi conditions. ### 2.3.3. *Leishmania donovani* {#sec2dot3dot3-marinedrugs-16-00376} The 50% inhibitory concentration (IC~50~) for J774 macrophages infected with *L. donovani* found 562 extracts active at 10 µg/mL or less. Filtering the data to select for low cytotoxic extracts (J774 IC~50~ \> 5 µg/mL) and high potency (*L. donovani* IC~50~ \< 1.0 µg/mL) reduced the *lead extract* set to 116 extracts (7.2% of all extracts), including 41 extracts with no epigenetic modulators (35%), 40 (34%) and 35 (30%) extracts subject to DNMTi and HDACi, respectively ([Table S3](#app1-marinedrugs-16-00376){ref-type="app"}). ### 2.3.4. *Naegleria fowleri* {#sec2dot3dot4-marinedrugs-16-00376} In the assay for *N. fowleri*, 34 extracts displayed \>33% inhibition of the amoeba ([Table S4](#app1-marinedrugs-16-00376){ref-type="app"}) at, on average, 50 µg/mL, 2 of which were deprioritized due to high cytotoxicity (IC~50~ \< 5 µg/mL), leaving 32 total *lead extracts* (2% of all extracts). Parsing the screening results by activity level reveals one extract with the indicated \>33% inhibition achieved at 5 µg/mL, and two extracts achieving \>67% inhibition at 50 µg/mL ([Table S4](#app1-marinedrugs-16-00376){ref-type="app"}), providing a clear pathway to prioritization of the *N. fowleri* lead extracts. Over half of the extracts active against *N. fowleri* were non-treated (56%), one quarter were HDACi treated, and the remainder were DNMTi treated. 2.4. Overall Data Analysis {#sec2dot4-marinedrugs-16-00376} -------------------------- Of the 1608 extracts screened, 254 (16%) were determined to be active ("lead extracts") in one or more assay ([Table S5](#app1-marinedrugs-16-00376){ref-type="app"}). These 254 active extracts resulted from 162 endophytes ([Table S6](#app1-marinedrugs-16-00376){ref-type="app"}), and 72 of these fungi (44%) produced active extracts only when cultured in the HDACi or DNMTi conditions ([Figure 3](#marinedrugs-16-00376-f003){ref-type="fig"}). Specifically, 29 fungi were only active after HDACi treatment, 24 were only active after DNMTi treatment, 19 were active in both HDACi and DNMTi treatments (but not the control), 40 fungi were active only in the control, and 69 fungi were active in the control and at least 1 treatment condition. Only 36 lead extracts (14% of total active, 2% of total screened) showed activity against multiple targets ([Figure 4](#marinedrugs-16-00376-f004){ref-type="fig"}, [Table S5](#app1-marinedrugs-16-00376){ref-type="app"}). Specific activity against *N. fowleri* was observed in 24 extracts (9%, 1%), 33 extracts (13%, 2%) were active only in the ESKAPE pathogen screen, 71 (28%, 4%) were specific to *M. tuberculosis*, 92 (36%, 6%) were active only against the *L. donovani* infected macrophage. 3. Discussion {#sec3-marinedrugs-16-00376} ============= Endophytic fungi were isolated from mangrove tissues of five species of mangrove or mangrove associated trees from five North American regions ([Figure 2](#marinedrugs-16-00376-f002){ref-type="fig"}), yielding an average of 500 fungal strains from each geographic location. Strains cultured in a manner to enhance epigenetic expression of secondary metabolites were demonstrated to produce broad yet selective bioactivity profiles. Lead extracts, defined as extracts displaying sufficiently high potency (variable among assays) and low mammalian cytotoxicity (\>5 µg/mL) to be considered candidates for chemical analysis, were found in 16% of all tested samples ([Figure 3](#marinedrugs-16-00376-f003){ref-type="fig"}). Among individual screens, the overall lead extract hit rate was found to be 2--7%, reflecting the stringent potency and cytotoxicity criteria selected for each assay as guidance for advancing samples to chemical analysis. Among the bacterial pathogens, *Staphylococcus aureus* was the most sensitive of the ESKAPE panel ([Figure S1](#app1-marinedrugs-16-00376){ref-type="app"}) to the fungal extracts. While nearly 13% of extracts displayed activity in one or more of the ESKAPE pathogens, only 2.9% of extracts were sufficiently activity (scaled score = 9) \[[@B46-marinedrugs-16-00376]\] to advance as lead extracts. *Mycobacterium tuberculosis* on the other hand proved highly sensitive, with 34% of extracts exhibiting GI~50~ ≤ 100 µg/mL. Focusing on the most promising *M. tuberculosis* activities by restricting the activity to the GI~95~ still produces 100 lead extracts (6.2% hit rate), an encouraging result that holds promise for the discovery of new anti-tuberculosis scaffolds. The pressing need to overcome antibiotic resistance will be advanced by the discovery of new antibiotics with new mechanisms of action \[[@B48-marinedrugs-16-00376]\]. The eukaryotic pathogens studied in this project were the protists responsible for leishmaniasis, *Leishmania donovani*, and primary amoebic meningoencephalitis (PAM), *Naegleria fowleri*, two rare and largely neglected diseases which nonetheless carry a significant disease burden, not merely due to morbidity and mortality \[[@B49-marinedrugs-16-00376],[@B50-marinedrugs-16-00376],[@B51-marinedrugs-16-00376]\], but for social and economic \[[@B52-marinedrugs-16-00376]\] impacts. *L. donovani* was the most sensitive test organism in our project, inhibited at low dose (\<1 µg/mL) with low mammalian cytotoxicity ([Table S3](#app1-marinedrugs-16-00376){ref-type="app"}) by 116 extracts (7.2% of all extracts). Employing a typical natural product molecular mass of 500 g/mol indicates that nearly half of those lead extracts will harbor sub-micromolar compound(s). In contrast, *N. fowleri* was the least sensitive pathogen to our extract set, responding to only 2% of 1608 extracts with low sensitivity (generally 33--66% inhibition at 50 µg/mL, see [Table S4](#app1-marinedrugs-16-00376){ref-type="app"}). Using inhibitors of two epigenetic regulatory mechanisms proved profitable in enhancing screening results. Fungi treated with the HDAC inhibitor sodium butyrate and the DNMT inhibitor 5-azacytidine produced extracts that acted as independent screening samples, displaying unique bioactivity profiles from one another and from untreated extracts. This strategy effectively generated an extract library that was functionally three times the size of the microbial isolate library ([Figure 2](#marinedrugs-16-00376-f002){ref-type="fig"}B). Additionally, the data indicates that this methodology successfully accessed otherwise hidden biosynthetic potential, with 44% of active fungi producing activity only in the presence of epigenetic modification. As expected, instances in which the small molecule pressure eliminated or had no effect on activity were also observed (e.g., extracts that were active only in the control, or in both the control and modified conditions), further validating the approach. A high level of selectivity was observed among the active extracts. Large natural product extract libraries---specifically those of fungal origin---are often considered to be plagued with indiscriminately active nuisance compounds. Nevertheless, these results display a high level of extract selectivity and a hit rate of \~5% in each assay. These statistics strengthen the argument for bioprospecting within the microbial world found in the North American mangrove forests. It is notable that these conclusions have been generated without chemical analysis of the extracts (HRMS, NMR). With nothing more than a diverse set of bioassay data, these extracts can now be rationally prioritized according to potency, specificity, or modification efficacy. In certain bioassays, further information may be extrapolated; e.g., in the ESKAPE panel, extracts acting against both Gram positive and Gram negative pathogens as compared to those displaying selective activity towards Gram negatives. In smaller extract subsets like this one, this may be sufficient to transition directly into scale-up and structure isolation schemes. For larger extract libraries, or in the search for new and novel chemistry, this front-end bioassay data can inform a more cost and time efficient transition into extract chemical analysis such as HRMS- or NMR-based networking for dereplication efforts. Whatever the next step, the accumulation of seemingly unrelated biological data on an extract library can direct a more efficient, effective compound discovery pipeline. Scale up and chemical analysis of active extracts identified herein, including the isolation of bioactive compounds, is underway. In the case of rare and neglected disease targets, those with newly developed drug targets, and in the face of growing drug resistance, both new and previously isolated natural products, may be of interest. We believe that this bioassay-driven approach is ideally suited for target specific isolation and investigation of both new and known bioactive natural products for meaningful drug discovery efforts. 4. Materials and Methods {#sec4-marinedrugs-16-00376} ======================== 4.1. Fungal Isolation {#sec4dot1-marinedrugs-16-00376} --------------------- Tissues from mangroves (*Rhizophora mangle, Avicennia germinans*, and *Laguncularia racemosa*), associated trees (*Conocarpus erectus*, and *Coccoloba uvifera*), sediments, and marine invertebrates were collected over the course of several years (2010--2014) at sites around Tampa Bay, the Florida Keys, the Gulf of Mexico and Tapachula, Mexico. As previously reported \[[@B31-marinedrugs-16-00376]\], small pieces of organic material were surfaced sterilized in bleach and/or isopropyl alcohol and pressed against or transferred onto various solid agar media types meant to target a large scope of bacteria and fungi. Media preparation was as follows: a nutrient medium (e.g., potato dextrose, Sabaurad dextrose, actino, malt, tryptic soy, or glycerol) was combined with agar, salt, and a combination of antibacterial or antifungal small molecules (e.g., nystatin, cycloheximide, or chloramphenicol). Six to 10 variations on these solid media types were utilized in each collection. Collected tissues were plated in triplicate on each of the media types. During incubation at room temperature, plates were routinely checked for growth and colonies were transferred to isolation plates of similar media composition. Once a pure colony had been established, small subsamples of mycelium were inoculated into a 20% glycerol/water solution for long-term preservation. 4.2. Miniaturized Culture Conditions and Extraction Procedures {#sec4dot2-marinedrugs-16-00376} -------------------------------------------------------------- Generally following methods previously reported \[[@B30-marinedrugs-16-00376],[@B31-marinedrugs-16-00376]\], for each fungal isolate, two 1 cm^2^ pieces of fungal material on agar were inoculated into 1.25 mL of each: untreated SDB (Sabouraud dextrose broth), 100 µM sodium butyrate in SDB, and 100 µM 5-azacytadine in SDB and agitated. Each of these aliquots was poured over 3 g autoclaved brown rice in a 20 mL scintillation vial. Cultures were fermented at 28 °C for 21 days. Following fermentation, the fungal material was sprayed with approximately 500 µL MeOH and then soaked with 10 mL EtOAc. The cultures were extracted for 24 h, after which time the extract was decanted, dried, and re-solvated to a concentration of 5 mg/mL in DMSO. Samples were transferred into 96-well plates using a TECAN liquid handler. Plates were distributed for bioassay and remaining extracts archived for use in future analysis in deep 96-well plates at −18 °C. 4.3. ESKAPE Bacterial Strains, Growth Conditions, and Microtiter MIC Determination Assays {#sec4dot3-marinedrugs-16-00376} ----------------------------------------------------------------------------------------- The ESKAPE pathogen clinical isolates used in this study were obtained from Moffitt Cancer Center and Tampa General Hospital. Overnight cultures were grown in lysogeny broth (LB) or tryptic soy broth (TSB), as described previously \[[@B46-marinedrugs-16-00376]\]. MIC assays were performed as previously described \[[@B46-marinedrugs-16-00376]\]. Briefly, overnight culture of all strains were grown in tryptic soy broth then diluted 1:1000 in cation adjusted Mueller Hinton broth (CA MHB, Difco laboratories, a subsidiary of Becton, Dickinson and Company, Sparks, MD 21152 USA) for the assay. The initial testing started at 200 µg/mL extract concentration in a 96-well microtiter plate and allowed to grow 20 h at 37 °C. The extracts were screened using a tiered MIC approach, with active samples at 200 µg/mL progressed to testing at 100 µg/mL. This approach was used to continue testing to 50, 25, 10, and 5 µg/mL. 4.4. Analysis of Antimicrobial Activity against Replicating Mycobacterium tuberculosis {#sec4dot4-marinedrugs-16-00376} -------------------------------------------------------------------------------------- Activity of the fungal compounds was examined against *Mycobacterium tuberculosis* using a reporter-based whole cell assay. The strain *Mtb*-RG, which expresses mCherry and GFP constitutively from the *smyc* promoter and *hsp60* promoter, respectively, was grown to a logarithmic phase (OD~600~ of 0.4--0.8) in 7H9 broth supplemented with OADC (oleic acid, albumin, dextrose, catalase) and kanamycin 50 µg/mL. The culture was diluted to an OD~600~ of 0.05 and added to a black solid bottom 384-well plate containing the fungal compounds at a final concentration of 100 µg/mL to a total volume of 30 µL per well. Rifampicin 10 µM (positive) and 2% DMSO (negative, no drug) controls were also included in the screening plate. The plate was incubated at 37 °C, 5% CO~2~ and fluorescence (excitation/emission 575/485 nm) was measured after 72 h using a Biotek Synergy H4 plate reader (Winooski, VT, USA). The activity of the fungal compounds was calculated as percent inhibition which was determined relative to the no drug control and inhibition by rifampicin taken as 100% using the formula {((DMSO signal-sample signal)/DMSO signal × 100) × 100/Rif inhibition} \[[@B15-marinedrugs-16-00376]\]. 4.5. Leishmania donovani-Infected Macrophage (IM) Assay {#sec4dot5-marinedrugs-16-00376} ------------------------------------------------------- Two thousand J774.A1 cells were seeded in each well of Perkin Elmer CellCarrier-384 Black Optically Clear Bottom plates (Perkin Elmer, Waltham, MA, USA). *Leishmania donovani* amastigotes were centrifuged at 3000 RPM for 5 m and brought up in macrophage media and added to macrophage well at a ratio of 10:1. They were then incubated for 4 h at 37 °C and 5% CO~2~. Non-phagocytized amastigotes were washed away and fresh macrophage media was added and allowed to incubate overnight. Media was removed and extracts diluted in media were added to the 384 well plates and incubated for 72 h. The extracts were all in six concentrations diluted 1:2 starting at 10 μM with positive and negative control and miltefosine standard drug control. The media was removed and fixed in 2% paraformaldehyde (Alfa Aesar, Ward Hill, MA, USA) in media for 15 m. The paraformaldehyde was removed and cells were stained with 5 μM Draq5 (Thermo Scientific, Waltham, MA, USA) diluted in PBS for 5 m. The stain was removed and fresh media was added to the plate. The Operetta (Perkin Elmer, Waltham, MA, USA) high content imager was used to capture six images from the middle of each well using a Far Red filter using 100 ms at 100% excitation at the plane of −1.4 µm. An algorithm in Harmony (Version 4.1, Perkin Elmer, Waltham, MA, USA) software in image analysis mode was programed to find the macrophage nuclei, cytoplasm, and amastigotes within the cytoplasm. The software counted the number of amastigotes per 500 macrophages in each well and calculated IC~50~ values based on non-linear regression of dose response curves. 4.6. Cytotoxicity of Mammalian Cell Line {#sec4dot6-marinedrugs-16-00376} ---------------------------------------- The viability of J774.A1 macrophages was determined by the Cell Titer 96 Aqueous Assay (Promega, Madison, WI, USA) that employs a tetrazolium compound \[3-(4,5-dimethylthiazol-2-yl)-5(3-carboxymethoxyphenyl)-2-(4-sulphophenyl)-2H-tetrazolium (MTS)\] and electron-coupling reagent, phenazine methosulphate (PMS). Test extracts were serially diluted in 100 μL of media in 96 cell plates using a Biomeck 3000 (Beckman Coulter, Miami, FL, USA). From each well, 10 μL was transferred to another 96 well plate and then received 90 μL J774.A1 in media. The J774.A1 macrophages were in a concentration to have 50,000 cells per well. After 72 h 20 μL of MTS solution was added to each well in the 96 well plates. The plates were then incubated 37 °C and 5% CO~2~ for 4 h to achieve optimal color development. After 4 h of incubation, the OD (optical density) values were determined at 490 nm using a Spectra Max M2 (Molecular Devices, Sunnyvale, CA, USA). The results were presented as the percentage of survivors (OD value with test compound divided by that of untreated control). Data analysis was completed using DataAspects Plate Manager analysis software (version 2001--2005, DataAspects Corporation, Los Gatos, CA, USA). Non-linear regression was used to obtain IC~50~ values. 4.7. Naegleria fowleri Culture and Cell Viability Assay {#sec4dot7-marinedrugs-16-00376} ------------------------------------------------------- Activity against *Naegleria fowleri* was screened using a patient isolate from 1969 (ATCC 30215). Previously published methods using the AlamarBlue (Bio Rad, Hercules, CA, USA) colorimetric assay with axenically grown trophozoites were followed. A Biomek 3000 automated liquid-handling workstation (Beckman Coulter, Miami, FL, USA) was used to serially dilute the extracts to 50 and 5 µg/mL in Nelson's culture media (Sigma-Aldrich, St. Louis, MO, USA) with a final concentration of 1% DMSO. The Biomek workstation was used to transfer diluted extracts, followed by the addition of 100,000 or 3000 *N. fowleri* trophozoites/well in 96- or 384-well screening plates, respectively \[[@B48-marinedrugs-16-00376]\]. 4.8. Data Analysis {#sec4dot8-marinedrugs-16-00376} ------------------ To facilitate the direct comparison of the results, extracts were given a simple "active/non-active" designation in each of the assays described above. The ESKAPE pathogens' MICs were transformed into a scaled score for each pathogen by dividing and summing the highest-tested concentration (200 µg/mL) by each concentration in which activity was seen. (e.g., a sample active at 100 µg/mL would receive a scaled score of 3; (200/200) + (200/100) = 3.) For simplicity, activity was evaluated as a single, combined scaled score for all 6 pathogens. An extract was considered active if it had a scaled score ≥ 9. Only extracts exhibiting an IC~50~ value \< 1 μg/mL in the infected macrophage model of the *L. donovani* assays were ranked as active. *M. tuberculosis* inhibition was noted as active when an extract inhibited ≥85% of bacterial growth, and activity against *N. fowleri* was defined as inhibition of \>33% at any of two concentrations tested (50 and 5 μg/mL). Any cytotoxicity against the J774 macrophage cells (from the *L. donovani* infected macrophage assay) up to 20 μg/mL was categorized as active. This research was supported by grants AI103673 (to D.E.K. and B.J.B) and AI103715 (to L.N.S. and B.J.B) from the US National Institutes of Health. Facilities in the Center for Drug Discovery and Innovation were supported by a State of Florida Center of Excellence Award. The following are available online at <http://www.mdpi.com/1660-3397/16/10/376/s1>, Table S1: MIC and cytotoxicity for ESKAPE pathogens treated with mangrove fungal extracts, Table S2: Growth inhibition and cytotoxicity of mangrove extracts inhibiting *M. tuberculosis*, Table S3: Inhibitory concentration and cytotoxicity of mangrove extracts toward *Leishmania donovani*, Table S4: Sensitivity (percent inhibition and cytotoxicity) of *Naegleria fowleri* to treatment with mangrove endophytic fungal extracts, Table S5: Screening data from 286 mangrove endophytic fungal extracts active in one or more pathogen screen, Table S6: Strains, treatment and bioactivity of mangrove endophytic fungi. ###### Click here for additional data file. B.J.B., D.E.K., L.N.S., K.H.R. and M.A.R.-P. conceived the screening campaign; D.H.D. and M.A.K. implemented the screening workflow; field work to collect samples was conducted by D.H.D., B.J.B. and M.A.R.-P.; D.H.D. and M.A.K. performed the fungal isolation, culture and extraction; L.N.S. developed the ESKAPE assay; R.F. and R.T. performed ESKAPE pathogen screening; D.E.K. developed the *Leishmania* spp., *N. fowleri* and J774 macrophage assays; A.A., A.S. and B.V. performed *L. donovani* screening; B.C. and C.A.R. performed *N. fowleri* screening; K.H.R. developed the TB assay R.G. and M.N. performed *M. tuberculosis* screening; D.H.D., M.A.K. and B.J.B. analyzed the data and wrote the paper. The authors declare no conflict of interest. {#marinedrugs-16-00376-f001} {#marinedrugs-16-00376-f002} {#marinedrugs-16-00376-f003} {#marinedrugs-16-00376-f004}

INTRODUCTION {#s1} ============ Colorectal cancer (CRC) is one of the most common human malignancies and a leading cause of cancer-related death worldwide in developed countries \[[@R1]\]. The age-standardized incidence of colorectal cancer in China is 16.9 per 100,000 in males and 11.6 per 100,000 in females, and the age-standardized mortality is 9.0 per 100,000 in males and 6.1 per 100,000 in females \[[@R2]\]. Nearly 15% of CRC patients are diagnosed with metastatic disease at the time of diagnosis, and nearly half of these patients develop metastases during the course of their disease \[[@R3]\]. Despite the improvements in diagnosis and treatment, metastatic colorectal cancer (mCRC) remains an incurable disease with a 2-year median overall survival time \[[@R4]\]. Clearly, new treatments for mCRC are necessary to improve the poor clinical outcomes. Over the past decade, the clinical benefits of monoclonal antibodies to epidermal growth factor receptor (EGFR), including cetuximab and panitumumab, combined with chemotherapy or monotherapy in mCRC patients have been shown \[[@R5]\]. Cetuximab was approved by the Chinese Food and Drug Administration in 2006. However, the response to cetuximab is influenced by a number of factors, the best known being KRAS gene status \[[@R6]\]. Previous pivotal studies have indicated that patients with KRAS wild-type mCRC will obtain a significant improvement in overall survival (OS), progression-free survival (PFS) and overall response rate (ORR) by adding cetuximab to standard chemotherapy. However, those with KRAS mutations are more likely to benefit from standard chemotherapy alone \[[@R7]--[@R9]\]. Thus, clinical guidelines have recommended cetuximab for the treatment of patients with EGFR-expressing, KRAS wild-type mCRC \[[@R10]\]. Therefore, KRAS mutation screening is an important component of the diagnostic plan \[[@R11]\]. Because of the resources required for mutation screening and cetuximab treatment, financial concerns might limit this evaluation. Economic analyses have indicated that cetuximab offers good value-for-money in patients with mCRC in developed countries \[[@R11]--[@R20]\]. However, these results might not be applicable for decision making in China because of the limited health resources in China and Western regions. In regard to the issues mentioned above, the current goal was to examine the outcomes of KRAS screening followed by targeted first-line cetuximab treatment for mCRC from the perspective of Chinese payers. RESULTS {#s2} ======= Base-case analysis {#s2_1} ------------------ The results of a base-case analysis with a 10-year time horizon, as well as economic and health outcomes estimated by the model, are shown in Table [1](#T1){ref-type="table"}. For patients with advanced mCRC, the cetuximab regimen yielded an increase of 0.149 progression-free life-years (LYs), 0.73 overall LYs, or 0.63 quality-adjusted life-years (QALYs) in comparison with the chemotherapy regimen. The incremental direct medical cost amounted to \$8,843 and \$17,086 with and without a patient assistance program (PAP) over the 10-year period, respectively. The incremental cost-effectiveness ratio (ICER) for adding cetuximab to irinotecan, fluorouracil, and leucovorin (FOLFIRI) chemotherapy was \$14,049 and \$27,145 per QALY saved with and without PAP, respectively. ###### Summary of cost (\$) and outcome results from a base-case analysis Regimen Cost Progression-free LYs Overall LYs QALYs Incremental cost per QALY\* Incremental cost per LY\* --------------------------- -------- ---------------------- ------------- ------- ----------------------------- --------------------------- FOLFIRI (control regimen) 30,668 0.795 2.066 0.963 Cetuximab with PAP 39,511 0.944 2.796 1.593 14,049 12,107 Cetuximab without PAP 47,754 0.944 2.796 1.593 27,145 23,393 \* Compared to a control regimen. Sensitivity analysis {#s2_2} -------------------- One-way sensitivity analysis revealed the most sensitive model parameters (Figure [1](#F1){ref-type="fig"}). The most sensitive parameters in the cetuximab regimen using PAP compared to the control included median OS time and cost of cetuximab. Other parameters, such as the cost and probability of managing severe adverse events (SAEs), showed moderate or little impact on the model's outcome. {#F1} The results of the probabilistic sensitivity analyses (PSA) are shown via cost-effectiveness acceptability curves (Figure [2](#F2){ref-type="fig"}). With PAP, the proportions of simulations being cost-effective for cetuximab were nearly 90% in comparison with the control regimen at a cost-effectiveness threshold of US \$22,000. When no PAP was available, the control regimen achieved 75% likelihood of cost-effectiveness. {#F2} DISCUSSION {#s3} ========== Using a Markov analysis model to assess wild-type RAS mCRC, we found that the 10-year ICER for adding cetuximab to traditional chemotherapy was generally unfavorable, at \$61,746 per QALY gained. The ratios were largely attributable to the higher cost associated with the acquisition of cetuximab, whereas other costs, such as RAS mutation testing and management of progressed disease, had little impact. This result was robust based on the results of PSA. For cetuximab PAP, cetuximab treatment with RAS testing for patients with wild-type RAS mCRC might be the most cost-effective option because their ICERs are lower than the threshold, and the probability of cost-effectiveness reaches 90% at a threshold of \$22,000 (Figure [3](#F3){ref-type="fig"}). These results suggest that cetuximab might be cost-effective in the PAP setting, which was supported by the sensitivity analysis. Furthermore, the cost of cetuximab is a sensitive parameter, as shown by a one-way sensitivity analysis. Other studies have also found that the ICER of cetuximab compared to that of other treatments for mCRC patients is high and is sensitive to drug costs \[[@R11]--[@R20]\]. {#F3} To our knowledge, the current report is the first economic analysis evaluating cetuximab for the treatment of wild-type RAS mCRC patients in a representative setting with limited health resources. The pharmacoeconomic results indicate that RAS mutation testing and targeted cetuximab treatment for patients with wild-type RAS mCRC yields an ICER of approximately \$650,000 per LY in comparison with anti-EGFR therapy from the perspective of the United States \[[@R16]\]. One possible reason for differences in these estimates is that this study incorporated survival data derived from a different source, which resulted in a reduced survival benefit (0.0026 years) with the cetuximab regimen. In the National Cancer Institute of Canada trial CO.17, the addition of cetuximab produced an ICER and cost--utility ratio of \$199,742 and \$299,613, respectively. When cetuximab therapy was restricted to patients with wild-type RAS mCRC, the ICER was improved to \$120,061 per LY gained and \$186,761 per QALY gained \[[@R17], [@R21]\]. An economic analysis in Switzerland showed that the ICER of cetuximab treatment as last-line therapy for patients with mCRC was €62,653 per QALY gained compared with that of regimens without cetuximab, which indicates that gene-guided cetuximab treatment is economically favorable \[[@R22]\]. The potential of cetuximab, the first therapeutic antibody for mCRC, to improve survival is a major determinant of clinical and economic outcomes. One-way sensitivity analysis found that the median OS time of the cetuximab regimen was the most influential parameter. This result indicates that the selection of a patient subgroup can increase the cost-effectiveness of the addition of cetuximab. Other independent and influential parameters include health insurance coverage and the price of cetuximab. A higher proportion of coverage will lead to a higher ICER for the addition of cetuximab treatment. As a potential option, providing a more favorable discount or PAP plan for cetuximab would significantly decrease the ICER for the addition of cetuximab. Several important limitations in the current study should be considered. First, modeling to extrapolate clinical survival beyond trial observation is an inevitable limitation in this study. The present model showed that PFS and OS time had substantial effects on the model's outcome. The short median follow-up periods of the pivotal cetuximab trials did not provide enough observed survival data to compare with the median survival estimated by the model. Thus, there was much uncertainty in the long-term survival probability. Second, the model did not fully evaluate the outcomes of using cetuximab in other settings, such as extended RAS testing, sensitivity or specificity of the KRAS mutation-screening test, second- or third-line treatment and combination with other chemotherapy regimens, which should be investigated in the future. Third, the present model did not include other biologicals used as first-line chemotherapy drugs, such as panitumumab, for assessing the incremental cost-effectiveness in comparison with cetuximab, as these drugs have not been approved by the Chinese Food and Drug Administration. Fourth, we did not perform a budget impact analysis of the addition of cetuximab. The age-standardized mortality was 16.9 per 100,000 in males and 11.6 per 100,000 in females \[[@R2]\], and cetuximab might be prescribed to more than 10,000 patients each year. Based on the results from our model, the addition of cetuximab to standard chemotherapy will increase expenditures by approximately \$131 million. Fifth, the clinical data were derived from trials from other countries, potentially influencing the results owing to radial differences. However, the Chinese study showed similar efficacy and safety to that found in a Caucasian population \[[@R23]\]. Sixth, the utility values obtained from other regions and the triangular distribution of cost inputs may have biased the model's output. Finally, the current analysis did not assess the impact of different therapies after disease progression. However, the results of the one-way sensitivity analysis indicated that the costs of disease progression had little impact on the final results. Owing to these limitations, the results should be carefully explained when they are referenced by Chinese decision makers. Our analysis indicates that the addition of cetuximab to traditional chemotherapy in patients with wild-type RAS mCRC is likely to be a cost-effective recommendation in China based on its superior efficacy and association with PAP. Although the current analysis focused on the Chinese medical system, the findings may also be helpful to other medium-income regions, such as Brazil, Russia, Taiwan and Thailand. MATERIALS AND METHODS {#s4} ===================== Analytical overview and model structure {#s4_1} --------------------------------------- A mathematical model was established to measure clinical and economic outcomes of additional cetuximab therapy for patients with mCRC. Patients were assumed to either start standard chemotherapy based on irinotecan, fluorouracil, and leucovorin (FOLFIRI, control regimen) or to start targeted treatment with additional cetuximab if the RAS screening was negative (cetuximab regimen), as shown in Figure [3A](#F3){ref-type="fig"}. Because this chemotherapy has been recommended as the first-line standard treatment for newly diagnosed mCRC by clinical guidelines \[[@R11]\] and the aim was to evaluate the economic outcome of adding cetuximab to the standard chemotherapy regimen, a "no treatment" strategy was not evaluated in this study. Health and economic outcomes were predicted using the Markov state transition model (Figure [3B](#F3){ref-type="fig"}) with three exclusive health parameters: PFS, progressed survival and death. A hypothetical cohort with confirmed newly diagnosed mCRC was created for comparing cetuximab therapy with a control regimen. We set the characteristics of the hypothetical cohort to be similar to the phase III ARTIST trial, which showed the age of 214 Chinese patients with newly diagnosed mCRC was 53 years old (range: 23--77), proportion of male was 50.4% and proportions of primary tumor site in colon, rectum and colorectum were 47.5%, 47.5% and 5.0%, respectively \[[@R24]\]. After cancer progression, patients were treated with second-line chemotherapy or supportive care. The duration was ten years because the median OS of patients with mCRC was lower than 3 years and the probability of survival to year 6 was zero in the FIRE-3 trial. The Markov cycle length was 14 days, and the primary evaluation criterion for all patients was PFS. The risk of disease progression or death was determined by the reported literature \[[@R7], [@R25]\]. This economic analysis was based on a literature review and experimental model and did not require approval by the Institutional Review Board/Ethics Committee. The following outcomes were examined: progression-free LYs, overall LYs, QALYs and cost. Cost and QALYs were annually discounted 5% based on the Chinese guidelines for pharmacoeconomic evaluation \[[@R26], [@R27]\]. The costs are shown as 2016 US dollars. ICERs presented as the cost per additional QALY gained were also examined. Clinical data {#s4_2} ------------- We carried out a literature review to identify all randomized controlled trials (RCT) exploring the clinical effectiveness of cetuximab in combination with FOLFIRI chemotherapy in comparison to FOLFIRI chemotherapy alone in patients with previously untreated mCRC. The following databases were used to search for eligible studies (cut-off date of March 26, 2016): PubMed, Web of science, EMBASE, and the Cochrane Library. The systematic searches identified two studies, which were included in the literature review of clinical effectiveness. Table [2](#T2){ref-type="table"} lists the key model parameters. ###### Key model inputs Parameter Values(ranges) Description and Reference ------------------------------------------------------------------- ------------------------------------------ --------------------------- Weibull survival model of PFS of control regimen Scale=0.00267; Shape=1.89552; r^2^=0.992 \[[@R7]\] Weibull survival model of PFS of control regimen Scale=0.00195; Shape=1.52888; r^2^=0.976 \[[@R7]\] Weibull survival model of OS of cetuximab regimen Scale=0.00540; Shape=1.53841; r^2^=0.982 \[[@R25]\] Weibull survival model of OS of cetuximab regimen Scale=0.00324; Shape=1.26410; r^2^=0.978 \[[@R25]\] RAS mutation prevalence 0.41(0.366-0.454) \[[@R28], [@R29]\] Body surface (m^2^) 1.72 (1.5-1.9) \[[@R37]\] Cost of FOLFIRI per cycle (US \$) 2050.5 (1083-3018) \[[@R38], [@R39]\] Cost of cetuximab per 100 mg (US \$) 637.4 (318.7-637.4) \[[@R40]\] Cost of salvage therapy per cycle (US \$) 2411.8 (1891-2739.1) \[[@R38], [@R39]\] Cost of RAS screening pre unit (US \$) 176.9 (132.7-221.2) \[[@R40]\] Cost of terminal care per cycle (US \$) 1980.1 (769.2-5288.3) \[[@R37]\] Cost of vomiting per event (US \$) 175.7 (134-223) \[[@R41]--[@R43]\] Cost of rash and acne per event (US \$) 11.1 (6.2-16) \[[@R41]--[@R43]\] Cost of fatigue per event (US \$) 1524.6 (421.9-3322.6) \[[@R41]--[@R43]\] Cost of neutropenia per event (US \$) 2694.6 (2154.7-3294) \[[@R41]--[@R43]\] Cost of diarrhea per event (US \$) 891.5 (158.6-1104.6) \[[@R41]--[@R43]\] Utility of PFS 0.85 (0.68-1) \[[@R36]\] Utility of progressed survival in chemotherapy or supportive care 0.24 (0.2-0.28) \[[@R35]\] Utility of progressed survival in targeted therapy 0.68 (0.52-0.78) \[[@R11], [@R34]\] FOLFIRI: irinotecan, fluorouracil, and leucovorin; PFS: progression-free survival; OS, overall survival; RAS: rat sarcoma viral oncogene homolog. Kaplan-Meier survival data of PFS and OS for the control regimen were available from the CRYSTAL trial, which evaluated the efficacy of 599 patients receiving FOLFIRI alone \[[@R7]\], and the clinical benefit of the cetuximab regimen was derived from the FIRE-3 study, which evaluated the efficacy of cetuximab plus FOLFIRI treatment in 297 patients with KRAS (exon 2) codon 12/13 wild-type mCRC \[[@R25]\]. The Weibull survival model was fitted to the reported PFS and OS survival data. Estimated scale and shape parameters, standard errors (SEs), adjusted R^2^ and correlation coefficients are presented in Table [2](#T2){ref-type="table"}. The shape parameter (γ) allows the hazard function to increase or decrease with increasing time; if γ \> 1.0, the hazard rate strictly increases in a nonlinear pattern with increasing time. The scale parameter (λ) is related to the measurement unit of time. It is assumed that RAS mutation status has no impact on the efficacy of FOLFIRI therapy \[[@R7]\]. The prevalence of mutations of KRAS was 41% in Chinese CRC patients \[[@R28], [@R29]\]. Tumor mutation status of KRAS was assessed using a pyrosequencing approach, as described in the FIRE-3 trial \[[@R25]\]. The current analysis assumed that there was no statistically significant difference in the sensitivity or specificity of the KRAS mutation-screening test in comparison with the FIRE-3 trial. Cost and utility {#s4_3} ---------------- We used the Chinese medical insurance perspective to estimate the cost of direct health expenditures, including first-line study treatment and second-line chemotherapy due to disease progression, follow-up and other direct medical costs (Table [2](#T2){ref-type="table"}). Treatment of side effects was considered only for SAEs (grade 3-4). All unit costs of health resources were obtained from the local literature, the health system or the National Development and Reform Commission of China. Catastrophic disease insurance would cover 60% of the medical expenditure \[[@R30]--[@R32]\]. Based on a cycle length of 14 days, the treatment scheme was as follows: FOLFIRI comprised a 60- to 90-min infusion of 180 mg/m^2^ irinotecan, a 120-min infusion of 400 mg/m^2^ racemic folinic acid, and 400 mg/m^2^ fluorouracil followed by a continuous 46-h infusion of 2,400 mg/m^2^ fluorouracil. The cetuximab regimen consisted of cetuximab (initial dose 400 mg/m^2^ infused over 120 min, and 250 mg/m^2^ infused over 60 min weekly thereafter) plus FOLFIRI. Treatment was continued until disease progression or unacceptable toxicity. Once the disease progressed, patients were assumed to receive salvage chemotherapy. To estimate the dosages of chemotherapeutic agents, it was assumed that a typical patient had a weight of 65 kg and a height of 1.64 m with a body surface area (BSA) of 1.72 m^2^, unused drugs in opened vials were discarded \[[@R33]\]. Because of the high price of cetuximab, it is not affordable by many in China; as such, the cetuximab PAP was implemented for Chinese patients with mCRC. In this program, cetuximab is paid for by the payer for the first two months, followed by donations for two months by the producer. Subsequently, cetuximab is supplied by the following scheme: pay for 1 month + donation for 3 months. Therefore, the impact of PAP was incorporated into the scenario analyses. The utility scores of PFS and progressed survival were obtained from previously published studies (Table [2](#T2){ref-type="table"}), and their standard errors were estimated at 25% of the mean value in our sensitivity analyses \[[@R11], [@R34]--[@R36]\]. Sensitivity analyses {#s4_4} -------------------- To test the robustness of the model, one-way sensitivity analyses and PSA were used. In the PSA, key model inputs were simultaneously and randomly sampled from the statistical distributions to generate 1,000 estimates of the cost and QALY for both regimens. Triangle distributions were adopted for cost parameters owing to the limited number of samples for generating the cost data, and the beta distribution was used for probability, proportion and utility score parameters. A cost-effectiveness acceptability curve (CEAC) was shown based on the results of the PSA. One-way sensitivity analyses were carried out for all parameters in a predefined range as shown in Table [2](#T2){ref-type="table"}, which were mainly obtained from the reported literature or by assuming a 25% or 50% base-case value. The model was constructed and analyzed in the R statistical environment (version 3.2.3; R Development Core Team, Vienna, Austria). In accordance with the World Health Organization (WHO) recommendation \[[@R44]--[@R46]\], the 3× per capita gross domestic product (GDP) value of China in 2015 (\$22,200) was used as the cost-effectiveness threshold. Sources of financial support: This work was supported by an unrestricted grant from Shanghai Municipal Commission of Health and Family Planning (No. 15GWZK0901 and 2012ZDXK003). The funders had no role in the design of the study, collection and analysis of data, decision to publish, or preparation of the manuscript. **CONFLICTS OF INTEREST** The authors have no conflicts of interest to declare.

Introduction ============ Obesity is the most common nutritional disorder and is associated with important comorbidities such as dyslipidemia, atherosclerosis, type 2 diabetes and insulin resistance \[[@B1]\]. Anti-obesity pharmacotherapy is a potentially important adjunctive treatment to lifestyle modification. Drugs used to induce weigh loss may act through reducing appetite and increase satiety (e.g. sibutramine and rimonabant), reduce the absorption of nutrients (e.g. orlistat) \[[@B2]\]. Amongst the drugs marketed for weight loss there have been many instances of market withdrawl due to adverse events, leaving only orlistat approved for long term use \[[@B3]\]. The selective cannabinoid 1-receptor blocker Rimonabant, once considered as a promising anti-obesity drug which could improve dyslipidemia associated with the metabolic syndrome, including raising HDL and reducing TG, has been withdrawn recently from the market because of psychiatric adverse events \[[@B3]-[@B5]\]. Another problem with anti-obesity drugs is lack of data on major obesity-related morbidity and mortality. Therefore, development of effective and safe drugs is an area of intense clinical interest. The well-established cardioprotective nature of high-density lipoproteins (HDLs) has made them and their main protein apolipoprotein A-I (apoA-I) popular targets for potential cardiovascular therapies. Current HDL-based therapies including direct infusion of rHDL and apoA-I mimetic peptides have exhibted abroad beneficial effects aside from mediating cholesterol efflux such as anti-inflammatory, anti-oxidative actions and shown potential usefulness as therapy for diseases involving chronic inflammation and oxidative stress \[[@B6]\]. One of the proatherogenic effects of obesity is attributable to its accompanying dyslipidemia. The prominent dyslipidemia in obesity is low high density lipoprotein cholesterol (HDL-C) levels and apoA-I. Epidemiological studies have shown a strong inverse correlation between HDL-C, apoA-I and obesity, especially in individuals with visceral obesity \[[@B7]\]. This inverse correlation was originally attributed to the disturbed metabolism of HDL and apoA-I in obesity status. But recent studies suggested that HDL/ApoA-I had reciprocal effect on obesity. This article will focus on the updated understanding of the anti-obesity effect of HDL and apoA-I. Anti-obesity effect of apoA-I ----------------------------- Obesity is defined medically as a state of increased body weight, more specifically adipose tissue, or sufficient magnitude to produce adverse health consequence \[[@B8]\]. As the body's largest energy reservoir, adipose tissue serves the primary function of lipid storage in the fed state. In fasting state, fatty acid is released from the breakdown of triglyceride (TG) into circulation for energy production \[[@B9]\]. Prolonged energy imbalance between energy intake and expenditure leads to an increase in both fat cell size and fat cell number \[[@B10]\]. Large adipocytes especially adipocytes present in visceral fat have a higher rate of lipolysis. It is known that obesity clearly associates with increased circulating free fatty acids (FFAs) and adipocytokines which not only initiate adipose inflammation but also drive all aspects of metabolic syndrome, including insulin resistance, dyslipidemia, and hypertension, eventually leading to increased risk for cardiovascular diseases \[[@B1]\]. HDL is generally considered as a protective factor for cardiovascular disease. In addition to the well known effect of mediating reverse cholesterol transport, it also exerts other beneficial functions such as anti-oxidative, anti-inflammatory and anti-thrombotic actions \[[@B11]\]. It's well known that obese individuals display lower plasma levels of HDL-cholesterol and apoA-I, with HDL-cholesterol levels associated with both degree and distribution of obesity \[[@B7],[@B8],[@B12]\]. Previous findings that several polymorphrism in apolipoprotein A1(ApoA-I) gene have been associated with obesity in Brazilian population and that body fat content was increased in apoA-I null mice model lead to query about the role apoA-I played in obesity development \[[@B13],[@B14]\]. Several recent studies aimed to answer this question studied the role of apoA-I in regulating obesity through two main lines of investigation: (1)overexpression of genes that encode apoA-I in mouse models; (2) administration of apoA-I mimetics D-4F and L-4F, which share structural and biological features of native apolipoprotein A-I, to mimic increased plasma apoA-I level in mouse models. Diet-induced obesity (DIB) was generated in apoA-I transgenic (ApoA-I-Tg) and wild type mice by feeding high fat diet for three months. Although both groups of mice had the same body weight gain and food intake, the body fat content was significantly lower in ApoA-I-Tg mice than the wild-type mice as evidenced by reduced weight of epididymal and retroperitoneal fat pat. Consistent with this result, in other studies, daily administration of D-4F and L-4F in high fat diet fed mice reduced weight gain and decreased obesity associated hyperglycemia when compared with age-matched vehicle-treated obese mice \[[@B15],[@B16]\]. Taken together, these observations implicated a potential anti-obesity effect of apoA-I. Up to date, there are several strategies for pharmacotherapy of obesity, including: (1) appetite suppression through central stimulation of anorexigenic signals or blocking orexigenic signals; (2) inhibition of nutrient digestion and absorption through gastrointestinal mechanism; (3) stimulation of fat mobilization and decreasing triacylglycerol synthesis and deposition in fat depots; (4) increase lipid oxidation or thermogenesis by uncoupling fuel metabolism, thereby enhancing energy expenditure \[[@B17]\]. In the above mentioned studies, the anti-obesity effect of apoA-I was not associated with a reduction of food intake or increased locomotive movement. Investigation into the metabolic profile confirmed enhanced energy expenditure through increased respiratory exchange ratio with consistent increased expression of uncoupled protein 1 (UCP1) in brown adipose tissue \[[@B16]\]. UCP1 expresses specifically in brown adipose tissue and is responsible for detaching fatty acid oxidation from the coupling to respiratory chain, resulting in thermogenesis \[[@B18]\]. It has been demonstrated that increased expression of UCP1 in brown adipocyte or ectopic expression of UCP1 in mouse or human skeletal muscle and white adipocyte promotes fatty acid oxidation and resistance to obesity \[[@B19]\]. Intriguingly, ApoA-I gene overexpression and D-4F treatment lead to significantly increased expression of UCP1 in brown adipose tissue \[[@B16]\]. Therefore, UCP1 is probably one of the target genes regulated by apoA-I in brown fat, and such regulation might contribute to the increase of energy expenditure observed in apoA-I over-expression and mimetic peptide treated mice. However, the mechanism underlying the anti-obesity effect of apoA-I remains to be further elucidated. Autophagy and obesity ===================== Phenotype transition from WAT to BAT ------------------------------------ Adipose tissue contains two distinct types of fat cells, white and brown. The balance between white and brown adipose tissue (BAT) is a factor that determines obesity. White fat cells are specialized for the storage of chemical energy as triglycerides, while brown fat cells have very limited ability to store lipid and serve the primary function of dissipating chemical energy through adaptive thermogenesis \[[@B20]\]. Since brown adipocyte is more oxidative, BAT is generally considered as an organ counteracting obesity. Generally, proneness to obesity and metabolic disease correlates with decreased BAT activity, whereas resistance to obesity correlates with increased BAT function or the induction of brown adipocyte-like gene expression in white adipose tissue \[[@B21]-[@B27]\]. However, BAT was previously considered to be important only in small mammals such as rodent because in human, BAT was previously thought to disappear soon after birth \[[@B20]\]. Not until recent findings of presence of substantial amounts of metabolically active brown adipose in healthy adult human that the importance of BAT in metabolism has been recognized and that the speculation as to whether controlled recruitment of brown adipocyte would be a potential strategy to combat obesity has been raised \[[@B21],[@B28]-[@B31]\]. Ample evidences support some plasticity exists between white and brown adipocyte. In mice and rats, exposure to cold or β-adrenergic agonists induces the appearance of brown adipocyte in traditional white fat pats, which originate differently from the classic brown adipocyte that develop before birth because these newly formed brown adipocyte did not express the YFP reporter gene, which was conspicuously expressed in the interscapular brown fat cells from the same mice \[[@B32]-[@B35]\]. Similarly, in human, working in the cold temperature could induce occurrence of brown adipose tissue in outdoor workers \[[@B36]\]. Some pathological states such as pheochromocytoma, hibernoma, and diseases involving chronic hypoxia, Chagas' disease, Duchenne dystrophy and cancers, are known to stimulate BAT proliferation in white pats \[[@B37]\]. These evidences support the possibility that under certain conditions,cell with the morphology and molecular phenotype of brown fat can be induced in WAT. In addition, given that BAT and WAT are both present in various adipose tissue depots in human, attainment of a brown adipocyte cell phenotype in white adipocyte may be a potential strategy for combating obesity. Autophagy regulate obesity through phenotype transition ------------------------------------------------------- Autophagy, defined as a highly regulated process involving the bulk degradation of cytoplasmic macromolecules and organelles in mammalian cells via the lysosomal system, plays a housekeeping role in removing misfolded or aggregated proteins, clearing damaged organelles, eliminating intracellular pathogens, and balancing sources of energy at critical times in development and in response to nutrient stress \[[@B38],[@B39]\]. As a survival mechanism, dysregulated autophagy has been linked to many human pathophysiologies, such as cancer, myopathies, neurodegeneration, heart and liver diseases \[[@B40]\]. Interestingly, recent studies revealed that autophagy was upregulated in obese individuals, as evidenced by increased expression of autophagy gene Atg5, LC3A, and LC3B as well as elevated autophagic flux in omental and subcutaneous adipose tissue \[[@B41]\]. Consistent with this conclusion, other group reported that autophagy was strongly up-regulated in patient with diagnosis of Type 2 diabetes (T2D) and overweight \[[@B42]\]. This relationship between increased autophagy and obesity lead to a question that what is the role autophagy play in the state of obesity. On the one hand, excess lipid storage in hypertrophied adipocyte leads to increased endocytoplasmic reticulum (ER) activity, which ultimately overwhelms the capacity of ER to properly fold nascent protein, therefore causing ER stress and subsequent oxidative stress in the mitochondria, FFA release and proinflammatory state of adipoctye, which all impinge on mTOR and induce autophagy \[[@B43]\]. In this sense, activated autophagy may serve as a protective mechanism necessary for cell survival in the challenging environment that develops in adipose tissue. On the other hand, increased autophagy possibly signifies a process underlying cell death of hypertrophied adipocyte \[[@B44]\]. To further elucidate the role of autophagy in the state of obesity, recent studies used knockout mice model with adipose tissue specific deletion of autophagy-related gene Atg7 and Atg5 respectively. The Autophagy-related gene knockout mice exhibited a metabolic favorable phenotype. The mutant mice were slimmer with increased insulin sensitivity and showed resistance to high-fat-diet induced obesity compared with wide-type mice. The mutant mice contain only 20% of the mass of WAT found in wild-type mice \[[@B40],[@B45],[@B46]\]. Consistent with this finding, in vitro studies using cell line preadipoctye 3T3-L1 and primary MEFs showed that autophagy inhibition through Atg7 knockout or 3-MA treatment blocked adipocyte differentiation \[[@B9]\]. Thus, it seems that the metabolic favorable phenotype generated in Atg7/Atg5 knockout mice may result from a blocking in white adipocyte differentiaton which leads to a failure of these cells to accumulate lipid. However, the decreased serum free fatty acids and the absence of accumulation of excess lipid in nonadipose organs such as liver and heart in mutant mice did not support this proposition. Because WAT served as body's main reservoir of lipids, if the WAT capacity of lipid storage was simply blocked without any increased energy expenditure, detrimental elevation of serum FFA and ectopic lipid deposition would occur. Actually, what's interesting in those studies was that most of the mutant white adipocyte showed some characteristics of brown adipocytes. They were smaller, multilocular and contain more mitochondria. These cells also exhibited altered fatty acid metabolism with increased rates of β-oxidation and reduced rates of hormone-induced lipolysis. In addition, brown adipocyte specifically expressed UCP1 was found in WAT of Atg7-knockout mice \[[@B45]\]. Former studies have demonstrated that exogenous overexpression of PGC-1α, a critical mediator of brown fat differentiation, induced the development of features of brown fat in WAT \[[@B47]\]. In autophagy-related gene-knockout mice, PGC-1αwas found to be induced in WAT \[[@B9]\]. Loss of autophagy induced brown-like adipocyte in WAT may either through promoting progenitor cell shutting into the pathway of brown adipocyte differentiation or altering adipocyte transdifferentiation by promoting the conversion of WAT into BAT or blocking the reverse process. Since recent evidence support that WAT and BAT have different precursors and that in the knockout studies, younger mice (3 weeks) failed to show the phenotypic change that was observed in older mice (12 weeks), it is most likely that the phenotypic changes occurred after original WAT formation and was attributed to a phenotype transition from "white to brown" \[[@B48]\]. As regard to how loss of autophagy mediated this phenotypic transition process, it has been known that in rodent, BAT and WAT expressed pre- and post-natal respectively \[[@B49]\]. In human, recent studies demonstrated a transient expression of UCP1 during adipogenesis of adipocyte derived stem cells (ADSCs), which indicate the possibility that ADSC pass through a brown adipocyte-like stage while differentiating into adipocyte \[[@B50]\]. Therefore, autophagy may be critical for removal and degradation of excessive mitochondria or proteins which represent features of brown adipocyte during adiopogenesis. Downregulated autophagy in preadipocyte altered the normal process of adipocyte differentiation and resulted in adipocyte resembling brown adipocyte which has increased ability of energy consumption (Figure [1](#F1){ref-type="fig"}). {#F1} Regulation of autophagy ----------------------- One of the major roles of autophagy is to serve as a cellular adaptive reaction in order to sustain the internal organization in the case of insufficient external nutrient supply or augmented energy demands through digesting their own interior. On the one hand, autophagy needs to be upregulated under nutrient restriction. On the other hand, cells have to avoid excessive and enduring self-digestion. In order to keep the delicate balance between external energy and nutrient supply and internal consumption, complex regulatory network that sense the environmental change is required. The nutrient sensing kinase mTORC1 is well known to negatively regulate the autophagic machinery. Diverse external signals such as growth factor, amino acids, normoxia, or high energy levels all activate mTORC1 and result in the inhibition of autophagy \[[@B51]-[@B53]\]. mTORC1 inhibition, as resulting from rapamycin treatment was surposed to resulted in induction of autophagy. In mammalian cells, however, rapamycin treatment failed to induce autophagy as observed in yeast. This discrepancy is attributed to the rapamycin-resistant function of mTORC1. Although rapamycin fully inhibits mTORC1- dependent phosphorylation of S6K1, it only partially inhibits phosphorylation of other known mTORC1 substrate \[[@B54]\]. As a consequence, some mTORC1 dependent function such as autophagy is left unaffected by rapamycin. Instead, complete inhibition of mTORC1 function by another ATP-competitive inhibitor, Torin-1, could induce autophagy in mammalian cells \[[@B55]\]. In the obesity status, whether the activity of mTORC1 is augmented or attenuated is in debate. Previous studies using high fat induce obesity mice model demonstrated increased phosphorylation of mTORC1 substrate S6K1 in adipose tissue, which might indicate hyperactivity of mTORC1 \[[@B56]\]. However, as mentioned above, S6K1 phosphorylation does not represent the full function of mTORC1 activity. In addition, other study showed attenuated mTORC1 signaling along with affected downstream effects of mTORC1 signaling such as upregulated autophagy and impaired mitochondria in human adipocyte from obese patient with Type 2 Diabetes \[[@B42]\]. The reason for these discrepant results is still not clear. It is possible that the degree of obesity or whether insulin resistance exists may account for it. AMP-activated protein kinase (AMPK) is regarded as the main energy-sensing enzyme that promotes all varieties of catabolic pathways and blocks several anabolic pathways \[[@B57]\]. AMPK has been well known to be linked to regulation of autophagy. Under energy-lower conditions, AMPK is activated, leading to autophagy induction \[[@B58]\]. Activated AMPK was thought to induce autophagy both directly through its action on autophagy initiator UIK1 (unc-51-like kinase 1) and indirectly through its action on mTORC1. In case of alarming energy states, AMPK could counteract the mTORC1-mediated repression on autophagy by at least two mechanisms through its action on TSC2 or Raptor \[[@B52],[@B59]\]. In the state of obesity where energy intake exceeds energy consumption, AMPK activity in adipose tissue is decreased. However, whether this change in AMPK signaling alone under the situation of obesity is sufficient to affect autophagy rate is still not clear. Sterol depletion, a selective form of nutrient depletion, has been reported to induce autophagy. Recently, it was suggested that in the situation of sterol limitation, apart from the de novo synthesis and receptor-mediated uptake pathway, cell could recycle cytoplasmic lipid droplet as a source of cholesterol through the process of autophagy \[[@B60]\]. Sterol regulatory element-binding protein 2(SREBP2), which preferentially controls expression of many cholesterogenic genes, is attributable to the sensing of cholesterol requirement and act as a feedback mechanism to maintain sterol homeostasis \[[@B61]\]. Recent study showed that SREBP2 directly activated autophagy genes during sterol depletion. Besides, SREBP-2 knockdown in the condition of cholesterol depletion decreased autophagosome formation and lipid droplet association of autophagosome related protein LC3 \[[@B62]\]. Numerous studies have proposed that the dilution in membrane cholesterol in hypertrophied adipocye may be sensed as true cholesterol depletion. SREBP2 that is sensitive to membrane cholesterol depletion is selectively activated in hypertrophied adipocyte in several obesity mice models \[[@B63]\]. Although SREBP2 may not be a general regulator of autophagy taken that many other situations that trigger autophagy are independent of cell lipid metabolism and that changes in gene expression is not essential for the rapid initiation of autophagy, it is reasonable to presume that genes involved in the induction of autophagy be activated in response to conditions where cell cholesterol is limited through the regulation of SREBP2. HDL and autophagy ================= HDL prevent autophagy --------------------- Taken that HDL and apoA-I are inversely correlated to obesity and that apoA-I exhibits a direct anti-obesity action, it is of great interest to answer the question whether HDL/apoA-I could modulate adipocyte autophagy. Under conditions of excessive nutrition such as obesity, ER becomes stressed and results in activation of the copying system, which is termed unfolding protein response (UPR), in many metabolic tissues such as adipose tissue. Autophagy is activated by UPR so that unfolded and misfolded proteins could be degraded \[[@B64]\]. Recent studies reported that HDLs prevent ER stress in human endothelial cells and pancreatic β cells by reducing ER signaling and restoring protein folding and trafficking \[[@B65]-[@B67]\]. Thus, it is reasonable to hypothesize that HDLs may affect autophagy rate through alleviation of ER stress under the situation when stimuli exists. Interestingly, there is direct evidence showing that HDLs are able to prevent autophagy triggered by ox-LDL in endothelium \[[@B67]\]. Additionally, a large body of evidences support that HDL could activate PI3K-Akt signaling \[[@B68]-[@B71]\], a well identified upstream signaling of mTORC1 activation, which negatively regulate autophagy. HDL activated PI3K signaling ---------------------------- A major upstream signaling of mTORC1 is the phosphatidylinositol 3-kinase (PI3K) pathway. The binding of growth factor or insulin to cell surface receptors activates PI3K, which converts the plasma membrane lipid PIP2 to PIP3, subsequently recruits PDK1 and Akt to the plasma. Following being phosphorylated by PDK1, activated Akt positively regulates mTORC1 through phosphorylation-dependent inhibition of TSC2 with consequent autophagy inhibition \[[@B72]\]. Although numerous evidences showed HDL/apoA-I actived PI3K-Akt pathway in different cell lines, the underlying molecule mechanism still remains unclear. Rising evidences suggested that SR-BI, which mediated multiple antiatherogenic functions of HDL, was involved in PI3K-Akt signaling activation in response to HDL stimuli. Knock-down of SR-B1 by siRNA significantly attenuated HDL induced PI3K-Akt-eNOS signaling and prostacyclin production in endothelial cells \[[@B73]\]. Consistently, HDL stimulated glucose uptake in 3T3-L1 adipocytes involving PI3K-Akt activation via SR-B1. Knocking-down SR-B1 with RNA interference lead to diminish of glucose uptake stimulated by HDL with consistent pronounced inhibition of Akt activation \[[@B74]\]. Apart from SR-B1, sphingosine-1-phosphate (S1P) receptors binding with S1P of HDL has also been demonstrated to mediate PI3K-Akt signaling in mice myocardiocyte. A recent study demonstrated that HDL applied directly to isolated adult mouse cardiomyocytes enhances cell survival during hypoxia-reoxygenaration through stimulating signal PI3K-Akt. The prosurvival signal is mediated by S1P3 (sphingosine 1-phosphate 3) receptors located on the myocyte and are markedly attenuated by inhibitors of these receptors \[[@B71]\]. Taken together, HDL may activate PI3K-Akt signaling through SR-BI and S1P receptor (Figure [2](#F2){ref-type="fig"}). {#F2} Conclusion ========== Obesity, especially central obesity, is risk factor of cardiovascular disease and type 2 diabetes mellitus. Attainment of phenotype resembling brown adipocyte in WAT could reduce adipose deposit through enhanced β-oxidation. Inhibiting autophagy in adipose tissue leads white adipocyte to acquire features of brown adipocyte, therefore exerting an anti-obesity effect and improving obesity associated insulin sensitivity. Taken that HDL and apoA-I has direct anti-obesity effect in rat and that HDL activate PI3K-Akt signaling which negatively regulate autophagy through mTORC1, we propose that HDL/apoA-I inhibit autophagy through PI3K-Akt signaling, leading to phenotype transition of white adipocyte. This new link between HDL/apoA-I and autophagy provide a new understanding of the possible mechanism underline the antiobesity effect of HDL/apoA-I. Abbreviations ============= AMPK: AMP-activated protein kinase; apoA-I: Apolipoprotein A-I; apoA-I-Tg: apoA-I transgenic; BAT: Brown adipose tissue; C/EBP: CCAAT-enhancer-binding proteins; FFAs: Free fatty acid; HDL: High-density lipoprotein; LC3: Microtuble-associated protein light chain 3; mTORC1: Mammalian target of rapamycin complex 1; MEFs: Mouse embryonic fibroblast; oxLDLs: Oxidised low density lipoproteins; PDK: Phosphoinositide-dependent protein kinase; PI3K: Phosphoinositide 3-kinase; PPAR-γ: Peroxisome proliferator-activated receptor gamma; S1P: Sphingosine 1-phosphate; SR-B1: Scavenger receptor class B type I; SREBP: Sterol regulatory element-binding protein; TG: Triglyceride; TSC: Tuberous sclerosis complex; UCP: Uncoupling protein; UIK1 (unc-51-like kinase) WAT: White adipose tissue. Competing interests =================== The authors declare that they have no competing interests. Authors' contributions ====================== SW and DQP conceived the study, its design and drafted the manuscript. All authors read and approved the final manuscript.

All relevant data are within the paper and its Supporting Information files. Introduction {#sec005} ============ Sleep-wake disturbance is one of the nonmotor symptoms in Parkinson's disease (PD), which can present in the early stage of PD, even prior to the appearance of motor symptoms, and it can significantly impact the patients' quality of life (QoL) \[[@pone.0221219.ref001]--[@pone.0221219.ref003]\]. Deep brain stimulation (DBS) of the subthalamic nucleus (STN) has been reported to have a positive effect on sleep-wake disturbance improving the subjective sleep quality on the Parkinson\'s Disease Sleep Scale (PDSS) \[[@pone.0221219.ref004]--[@pone.0221219.ref011]\] and also objective sleep architecture on polysomnography in PD patients \[[@pone.0221219.ref006],[@pone.0221219.ref008],[@pone.0221219.ref011]--[@pone.0221219.ref014]\]. However, little is known about the long-term effects of STN DBS on sleep because most prior studies had a follow-up of no longer than one year after surgery with a lack of detailed outcomes for sleep parameters \[[@pone.0221219.ref005],[@pone.0221219.ref014]--[@pone.0221219.ref018]\]. In addition, results of the DBS effects on excessive daytime sleepiness are conflicting \[[@pone.0221219.ref005],[@pone.0221219.ref014]--[@pone.0221219.ref018]\]. In this study, we aimed to investigate long-term effects of STN DBS on sleep using the PDSS \[[@pone.0221219.ref019]\] and the Epworth sleepiness scale (ESS) \[[@pone.0221219.ref020]\] by following up sixty-one PD patients who were treated with STN DBS. We hypothesized that the benefit of DBS on the subjective sleep quality might not be sustained beyond one year after DBS surgery. It would be more likely to gradually decline considering the progressive effects of PD and aging on sleep. Additionally, if the sleep parameters do change over time after STN DBS, to investigate clinical variables that may affect the patients' sleep, we evaluated the association of the changes in the PDSS and ESS scores with the changes in the PD-associated motor and non-motor scales scores including the Unified Parkinson\'s Disease Rating Scale (UPDRS) score \[[@pone.0221219.ref021]\], Hoehn & Yahr (H&Y) stage \[[@pone.0221219.ref022]\], quality of life measure, and depression scale. We also investigated the predictive baseline clinical variables for changes in sleep parameters after STN DBS. Materials and methods {#sec006} ===================== Participants {#sec007} ------------ Patients with PD who underwent bilateral STN DBS from July 2011 to October 2015 at the Movement Disorders Clinic at Seoul National University Hospital were included in this study. The surgical procedures were previously described \[[@pone.0221219.ref023]\]. Because patients with impaired cognitive function shown by a Mini-Mental State Examination (MMSE) score of less than 25 points, severe psychiatric or behavioral disturbances, and structural brain lesions on neuroimaging prior to DBS are not recommended for DBS \[[@pone.0221219.ref024]\], those patients were excluded in this study, and patients with unilateral STN DBS and severe post-DBS surgical complications were also excluded. Sixty-one patients met the inclusion criteria of this study. Patients were evaluated before DBS surgery at baseline and at 6, 12, 24, and 36 months after bilateral STN DBS during their regular visits for follow-up. Patients who did not follow-up at each visit were interviewed by telephone on their scheduled visit days in order to find out their reasons for missing the visit and to check their post-DBS condition. The study protocol was approved by the Seoul National University Hospital Institutional Review Board and followed the principles of the Declaration of Helsinki. Clinical evaluation {#sec008} ------------------- We assessed the sleep parameters which were measured by the PDSS and ESS during the regular visits. The PDSS questionnaire consists of 15 items with scores from 0 to 10 for each symptom (0, severe symptoms; 10, absence of symptoms; overall scores range from 0 to 150) \[[@pone.0221219.ref019]\], for which the items are classified into eight domains as follows: overall quality of a night's sleep (item 1), sleep onset and maintenance insomnia (items 2 and 3), nocturnal restlessness (items 4 and 5), nocturnal psychosis (items 6 and 7), nocturia (items 8 and 9), nocturnal motor symptoms (items 10--13), sleep refreshment (item 14), and daytime dozing (item 15). The ESS questionnaire consists of 8 items with scores from 0 to 4 for each symptom (0, absence of symptoms; 4, severe symptoms; overall scores range from 0 to 24) \[[@pone.0221219.ref020]\]. In addition, total sleep hours per day including daytime naps were assessed by use of patients' self-reports. We also evaluated the clinical variables at each visit, which could potentially affect the sleep parameters in PD. The variables included the UPDRS part I-III score \[[@pone.0221219.ref021]\] and H&Y stage \[[@pone.0221219.ref022]\] in the on and off anti-parkinsonian medication condition with the DBS turned on condition, total daily levodopa equivalent dose (total LED) calculated as previously described \[[@pone.0221219.ref025]\], quality of life score measured by the Schwab and England Activities of Daily Living Scale (ADL) \[[@pone.0221219.ref026]\] in the on and off medication condition and the 36-Item Short Form Health Survey (SF-36) physical and mental health score \[[@pone.0221219.ref027]\], and depression score measured by the Beck depression inventory (BDI) \[[@pone.0221219.ref028]\]. Statistical analysis {#sec009} -------------------- We used the linear mixed model to examine the changes over time from baseline to three years post-STN DBS (each visit at 6 months, 1, 2, and 3 years post-STN DBS), of which analysis considers individual differences in interval time by adding random effects, after being tested for a normal distribution using the Shapiro-Wilk test. Post-hoc t-tests with Bonferroni correction for multiplicity were used to assess the difference in changes between the visits. Changes in the total PDSS and total ESS scores from baseline to three years post-DBS were correlated with the clinical variables (the UPDRS part I-III, H&Y stage, total LED, ADL, SF-36 physical and mental health, and BDI) at baseline and their changes from baseline to three years post-DBS using partial correlation adjusted for age, gender, and disease duration. We compared the baseline characteristics of the patients who were regularly followed and completed all of visits and the dropped-out patients by Mann--Whitney U test for continuous variables and Chi-squared test for categorical variables. SPSS 21.0 (SPSS Inc., Chicago, IL, USA) was used for the statistical analysis with the significance set at 0.05 (two-tailed). Results {#sec010} ======= Demographics and PD-associated characteristics of the PD patients with STN DBS {#sec011} ------------------------------------------------------------------------------ The baseline demographics of the sixty-one patients before DBS surgery are shown in [Table 1](#pone.0221219.t001){ref-type="table"}. The mean patient age at STN DBS was 61.2±8.2 years, and the mean disease duration at the time of STN DBS was 17.3±5.8 years. The mean UPDRS part III score and H&Y stage in the on and off anti-parkinsonian medication condition were 22.8±11.1 and 44.4±14.1, and 2.5±0.7 and 3.2±0.8, respectively. Among the sixty-one patients at baseline, fifty-eight patients completed a 6-month evaluation, 55 completed a 12-month evaluation, 45 completed a 24-month evaluation, and 46 completed a 36-month evaluation. Thirty-eight patients completed those total five visits from baseline to the last follow-up at 36-month. In comparison of the baseline characteristics between the 38 patients and the remaining patients, there were no differences in demographics and all of PD-associated characteristics except for the UPDRS part II score in the on medication condition shown in [Table 2](#pone.0221219.t002){ref-type="table"}. On the telephone follow-up for the dropped-out patients at each visit, there were no reports associated with post-DBS complications or other significant changes in their ADL after STN DBS. 10.1371/journal.pone.0221219.t001 ###### Demographics and PD-associated characteristics in PD patients with STN DBS. {#pone.0221219.t001g} Variables PD patients, N = 61 ------------------------------------------- --------------------------- Age at DBS, yr.     Mean (SD) 61.2 (8.2) Gender, n (%)     Male 27 (55.7)     Female 34 (44.3) Disease duration, yr.     Mean (SD) 17.3 (5.8) Age of PD onset, yr.     Mean (SD) 43.9 (8.5) UPDRS score & subscores, mean (SD)     UDPRS part I-III total score       On / Off medication 36.0 (16.2) / 74.9 (20.5)     UPDRS part I       On / Off medication 1.9 (2.0) / 4.0 (3.4)     UPDRS part II       On / Off medication 11.5 (8.6) / 26.5 (9.3)     UPDRS part III       On / Off medication 22.8 (11.1) / 44.4 (14.4) H&Y stage, mean (SD)     On / Off medication 2.5 (0.7) / 3.2 (0.8) Total LED, mg/day     Mean (SD) 1662.6 (744.2) ADL (%)     On / Off medication 78.9 (19.6) / 43.4 (23.3) SF-36 score & subscores, mean (SD)     SF-36 total (physical & mental) score 304.9 (145.6)     SF-36 physical health 145.2 (67.5)     SF-36 mental health 159.6 (91.6) BDI     Mean (SD) 21.7 (10.1) MMSE     Mean (SD) 28.0 (1.6) PD, Parkinson's disease; UPDRS, Unified Parkinson's disease rating scale; H&Y, Hoehn & Yahr; LED, Levodopa equivalent dose; ADL, Activities of daily living; SF-36, 36-Item Short Form Health Survey; BDI, Beck depression inventory; MMSE, Mini-mental state examination. 10.1371/journal.pone.0221219.t002 ###### Baseline demographics and PD-associated characteristics of the patients with or without completion of all five visits in this study. {#pone.0221219.t002g} ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- Variables PD patients with completion of all\ PD patients without completion of all\ *p* values five visits, N = 38 five visits, N = 23 ------------------------------------------- ------------------------------------- ---------------------------------------- ---------------------------------------------------- Age at DBS, yr.     Mean (SD) 60.2 (7.4) 62.8 (9.2) 0.161 Gender, n (%) 0.663     Male 16 (42.1) 11 (47.8)     Female 22 (57.9) 12 (52.2) Disease duration, yr.     Mean (SD) 17.2 (6.2) 17.4 (5.2) 0.788 Age of PD onset, yr.     Mean (SD) 43.0 (8.8) 45.4 (8.0) 0.213 UPDRS score & subscores, mean (SD)     UPDRS part I       On / Off medication 1.9 (1.7) / 4.4 (3.5) 1.9 (2.4) / 3.4 (3.3) 0.443 / 0.218     UPDRS part II       On / Off medication 9.5 (6.7) / 25.4 (9.6) 14.7 (10.3) / 28.3 (8.6) 0.040[\*](#t002fn002){ref-type="table-fn"} / 0.204     UPDRS part III       On / Off medication 21.5 (9.4) / 43.2 (13.6) 25.0 (13.3) / 46.4 (15.7) 0.384 / 0.352 H&Y stage, mean (SD)     On / Off medication 2.5 (0.6) / 3.0 (0.7) 2.5 (0.8) / 3.4 (0.9) 0.265 / 0.053 Total LED, mg/day     Mean (SD) 1644.7 (855.1) 1692.1 (528.5) 0.356 ADL (%)     On / Off medication 82.8 (15.3) / 47.4 (23.2) 72.6 (24.3) / 40.0 (22.4) 0.115 / 0.115 SF-36 score & subscores, mean (SD)     SF-36 total (physical & mental) score 327.6 (156.3) 267.3 (120.0) 0.214     SF-36 physical health 156.4 (73.7) 126.8 (52.1) 0.155     SF-36 mental health 171.2 (94.7) 140.5 (84.8) 0.294 BDI     Mean (SD) 19.7 (8.4) 25.3 (12.0) 0.080 MMSE     Mean (SD) 28.3 (1.5) 27.6 (1.8) 0.111 ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- PD, Parkinson's disease; UPDRS, Unified Parkinson's disease rating scale; H&Y, Hoehn & Yahr; LED, Levodopa equivalent dose; ADL, Activities of daily living; SF-36, 36-Item Short Form Health Survey; BDI, Beck depression inventory; MMSE, Mini-mental state examination. \**p* values \< 0.05 by Mann--Whitney U test or Chi-squared test. PDSS and ESS changes after STN DBS {#sec012} ---------------------------------- The total PDSS score before STN DBS was 79.0±30, and it increased after STN DBS to 100.0±23.3, 98.8±23.0, 97.1±29.6, and 93.3±28.0 at 6, 12, 24, and 36 months, respectively, which showed a significant improvement over time (*p* = 0.006 for the change over time), and compared with the baseline, the largest improvement was at 6 months as described in [Table 3](#pone.0221219.t003){ref-type="table"} and [Fig 1A](#pone.0221219.g001){ref-type="fig"}. Post hoc t-tests revealed significant differences between the total PDSS score at baseline and the scores at 6, 12, 24, and 36 months post-STN DBS (*p* \< 0.001, \< 0.001, 0.001, and 0.005, respectively). Among the eight PDSS domains, the domains scores for overall quality of a night's sleep, sleep onset and maintenance insomnia, and nocturnal motor symptoms showed significant increases after STN DBS (*p* = 0.036, 0.029, and \< 0.001, respectively, for the change over time), and compared with the baseline, the largest improvement was at 6 months shown in [Table 3](#pone.0221219.t003){ref-type="table"} and [Fig 1C](#pone.0221219.g001){ref-type="fig"}. In the post hoc t-test evaluation, the overall quality of a night's sleep score at baseline and the score at 6 months post-STN DBS significantly differed (*p* \< 0.001). The sleep onset and maintenance insomnia score showed differences between the baseline and 6, 12, and 24 months post-STN DBS (*p* \< 0.001, 0.001, and \< 0.001, respectively), and the nocturnal motor symptoms score showed differences between the baseline and 6, 12, 24, and 36 months post-STN DBS in the post hoc t-tests (all *p* \< 0.001). The scores of the nocturnal restlessness, nocturnal psychosis, nocturia, sleep refreshment, and daytime dozing tended to increase after STN DBS but not to a statistically significant degree (*p* = 0.284, 0.894, 0.335, 0.321, and 0.809, respectively, for the change over time). The total sleep hours per day on the patients' reports showed an increase after STN DBS (*p* = 0.001 for the change over time) shown in [Table 3](#pone.0221219.t003){ref-type="table"}. In the post hoc t-test, the differences were found between the baseline and 6, 24, and 36 months post-STN DBS (*p* = 0.002, 0.005, and 0.003, respectively). {#pone.0221219.g001} 10.1371/journal.pone.0221219.t003 ###### Sleep parameters as PDSS and ESS scores and sleep hour per day over time in PD patients with STN DBS. {#pone.0221219.t003g} --------------------------------------------------------------------------------------------------------------------------------------------------------------- Variable Baseline 6 mo. 12 mo. 24 mo. 36 mo. *p* values ----------------------------------------- ------------- -------------- ------------- ------------- ------------- ---------------------------------------------- PDSS total score, mean (SD) 79.0 (30.0) 100.0 (23.3) 98.8 (23.0) 97.1 (27.6) 93.3 (28.0) 0.006[\*](#t003fn002){ref-type="table-fn"} 8 PDSS subscore, mean (SD)     1. Overall quality of night's sleep 4.1 (2.6) 6.1 (2.5) 5.4 (2.6) 5.4 (2.8) 5.5 (2.8) 0.036[\*](#t003fn002){ref-type="table-fn"}     2. Sleep onset and maintenance\ 6.8 (4.9) 10.7 (5.4) 10.1 (5.5) 10.6 (5.3) 9.0 (5.7) 0.029[\*](#t003fn002){ref-type="table-fn"} insomnia     3. Nocturnal restlessness 10.4 (5.6) 12.7 (5.6) 12.9 (5.6) 13.1 (5.6) 11.4 (6.0) 0.284     4. Nocturnal psychosis 12.2 (5.4) 15.4 (4.5) 15.4 (4.5) 13.5 (5.2) 12.8 (4.9) 0.894     5. Nocturia 11.7 (3.6) 12.7 (4.0) 12.9 (4.0) 13.1 (4.5) 12.5 (4.9) 0.335     6. Nocturnal motor symptoms 23.0 (9.5) 29.6 (8.1) 30.0 (8.5) 29.9 (9.0) 29.2 (8.9) \<0.001[\*](#t003fn002){ref-type="table-fn"}     7. Sleep refreshment 4.7 (3.8) 6.2 (3.0) 5.4 (2.7) 5.2 (3.1) 5.8 (5.0) 0.321     8. Daytime dozing 5.9 (2.9) 6.9 (3.2) 7.0 (3.1) 6.3 (3.1) 6.4 (3.3) 0.809 ESS total score, mean (SD) 8.0 (4.8) 7.4 (5.1) 7.8 (4.6) 8.8 (4.8) 9.2 (5.0) 0.055 Total sleep/day, h 6.4 (1.9) 7.5 (1.5) 7.2 (1.5) 7.7 (2.7) 7.7 (2.1) 0.001[\*](#t003fn002){ref-type="table-fn"} --------------------------------------------------------------------------------------------------------------------------------------------------------------- PDSS, Parkinson's disease sleep scale; ESS, Epworth sleepiness scale. \**p* values \< 0.05 by linear mixed model. The total ESS score before STN DBS was 8.0±4.8, and after the surgery, it were 7.5±1.5, 7.2±1.5, 7.7±2.7, and 7.7±2.1 at 6, 12, 24, and 36 months, respectively, which were not significant changes over time (*p* = 0.055 for the change over time) as presented in [Table 3](#pone.0221219.t003){ref-type="table"} and [Fig 1B](#pone.0221219.g001){ref-type="fig"}. On the other hand, antiparkinsonian medications showed significant reductions over time (total LED before STN DBS = 1662.6±744.2; total LED after STN DBS = 509.5±381.1, 512.1±361.4, 578.2±389.5 and 670.7±568.7 at 6, 12, 24, and 36 months, respectively; *p* \< 0.001 for the change over time). Association of the changes in PDSS and ESS with clinical variables after STN DBS {#sec013} -------------------------------------------------------------------------------- When we analyzed the clinical variables that could potentially affect the changes in sleep parameters from baseline to three years post-DBS, we found changes in the total PDSS score were significantly associated with the changes in the UPDRS part I score, especially for the scores of the depression and motivation items in the UPDRS part I and the BDI (*r* = -0.411, -0.465, and -0.557, *p* = 0.007, 0.002, and \< 0.001, respectively) shown in [Table 4](#pone.0221219.t004){ref-type="table"}. Changes in the total PDSS score were also related to the QoL showing significant associations with changes in the UPDRS part II score and SF-36 physical and mental health scores (*r* = -0.387, 0.363, and 0.485, *p* = 0.022, 0.032, and 0.003, respectively). Changes in the UPDRS part III score, H&Y stage, and total LED, however, were not associated with the changes in the total PDSS score (*p* = 0.933, 0.231 and 0.270, respectively). Meanwhile, changes in the total ESS score showed no significant associations with the changes in any of the scores for the clinical variables described in [Table 4](#pone.0221219.t004){ref-type="table"}. 10.1371/journal.pone.0221219.t004 ###### Correlation coefficients of the changes in the PDSS and ESS scores with changes in the clinical variables. {#pone.0221219.t004g} Δ PDSS total score Δ ESS total score --------------------------- ------------------------------------------------- ------------------- Δ UPDRS part I score     On medication -0.556[\*\*\*](#t004fn004){ref-type="table-fn"} 0.052     Off medication -0.564[\*\*\*](#t004fn004){ref-type="table-fn"} -0.227 Δ UPDRS part II score     On medication -0.116 0.156     Off medication -0.387[\*](#t004fn002){ref-type="table-fn"} -0.112 Δ UPDRS part III score     On medication 0.045 -0.003     Off medication -0.015 -0.128 Δ H&Y stage     On medication 0.012 -0.017     Off medication -0.208 0.129 Δ Total LED 0.192 0.101 Δ ADL score     On medication 0.252 -0.055     Off medication 0.327 0.109 Δ SF-36 score     Physical health score 0.363[\*](#t004fn002){ref-type="table-fn"} -0.056     Mental health score 0.485[\*\*](#t004fn003){ref-type="table-fn"} -0.040 Δ BDI score -0.557[\*\*\*](#t004fn004){ref-type="table-fn"} 0.147 Δ ESS total score -0.098 \- Δ, changes from baseline to post-DBS year 3; PDSS, Parkinson's disease sleep scale; ESS, Epworth sleepiness scale; UPDRS, Unified Parkinson's disease rating scale; H&Y, Hoehn & Yahr; LED, Levodopa equivalent dose; ADL, Activities of daily living; SF-36, 36-Item Short Form Health Survey; BDI, Beck depression inventory. \**p* values \< 0.05 \*\* \< 0.01, and \*\*\* \< 0.001 by partial correlation adjusted for age, gender, and disease duration. In the analysis for the predictive baseline clinical factor for sleep outcome after STN DBS by the partial correlation adjusting for age, gender, and disease duration, we observed no significant relationship between the changes in the total PDSS score and the baseline scores for the clinical variables. Discussion {#sec014} ========== In this study, we investigated the long-term effects of STN DBS on sleep-wake disturbance by three-year following up the PD patients who underwent bilateral STN DBS. Our major findings were as follows. First, STN DBS led to a sustained improvement in sleep-wake disturbance of PD patients. Specifically, STN DBS contributed to improvement in nocturnal parkinsonian motor symptoms, overall quality of a night's sleep, and sleep onset and maintenance insomnia, and these benefits were maintained even three years after the surgery. Second, the improved sleep-wake disturbance was associated with an better mood and quality of life after STN DBS. Our findings were consistent with previous studies that showed positive impact of DBS surgery on sleep-wake disturbance; however, most of them had a follow-up of no longer than one year, and the improvement had a decreasing tendency over time \[[@pone.0221219.ref004]--[@pone.0221219.ref010]\]. Based on this and the progression of neurodegenerative PD with aging, we hypothesized that the initial improvement in sleep quality was followed by deterioration over one year after STN DBS. Contrary to the hypothesis, our data showed that the total PDSS score at baseline and those scores at all post-STN DBS visits (6 months, 1, 2, and 3 years post-STN DBS) differed, which meant improved sleep quality was prolonged over time. Although the largest improvement in the total PDSS score was at post-DBS 6 months and it seemed to gradually decline afterwards, there was no difference among those scores at 1, 2, and 3 years post-STN DBS. However, excessive daytime sleepiness assessed by the ESS did not show a significant change in our study. Previous studies of DBS effects on excessive daytime sleepiness in PD patients have shown inconsistent results; some studies reported a significant improvement in daytime sleepiness after STN DBS with reduced antiparkinsonian medication \[[@pone.0221219.ref005],[@pone.0221219.ref014],[@pone.0221219.ref018]\] whereas other studies failed to show any changes in daytime sleepiness \[[@pone.0221219.ref015]--[@pone.0221219.ref017]\]. In the present study, despite the significant reduction in antiparkinsonian medications, the total ESS score and the subscore for daytime dozing in the PDSS did not differ after STN DBS. In addition, changes in the total ESS score were not associated with the increased changes in the total PDSS score and the decreased changes in the total LED from baseline to three years post-DBS. However, total sleep hours per day on the patients' reports were increased after STN DBS in our study. This is probably due to increased night-time sleep given that nocturnal sleep quality was significantly improved after STN DBS. It remains unknown which factors contribute to sleep after DBS surgery. In our study, sleep parameter for nocturnal motor symptoms indicated by the PDSS was significantly improved after STN DBS, and the increased PDSS score was correlated with a decreased depressive mood and an increased motivation reflected by the UPDRS part I and BDI. Better motor symptoms after STN DBS may contribute to the improvement in sleep quality, and the improved sleep quality after DBS surgery may lead to improvement in mood disorders in PD patients. Another possibility is that STN DBS has a direct effect on sleep physiology independent of the improvement in the motor and non-motor symptoms. Because the STN has important reciprocal connections with the sleep-wake modulating structures, it may affect sleep \[[@pone.0221219.ref018],[@pone.0221219.ref029],[@pone.0221219.ref030]\]. Otherwise, as multifactorial effects, several factors including the direct effect of STN DBS and motor and non-motor symptoms may affect sleep quality after STN DBS. Our study has some limitations. This is a retrospective observational study performed in a single center. Among sixty-one patients at baseline, 75.4% of the patients were followed from baseline to three years after STN DBS, which raises the possibility of a dropout bias. However, on the telephone follow-up for the dropped-out patients, we could not find substantial changes in their post-DBS daily living. Furthermore, the baseline characteristics between the dropped-out patients and those who completed all visits in our study did not significantly differ. Another limitation is that we did not include a control group and measured the subjective assessment of sleep parameters which lack validation with the objective measures. Lastly, sleep disorders such as obstructive sleep apnea, restless leg syndrome, rapid eye movement sleep behavior disorder, and other than anti-parkinsonian medication-induced insomnia, which could possibly affect sleep, were not evaluated. In conclusion, this is the largest systematic long-term follow-up study of bilateral STN DBS contribution to sleep in PD patients. Our results demonstrated that bilateral STN DBS contributed to sustained improvement in overall nocturnal sleep quality, and getting the better night\'s sleep was correlated with better mood and quality of life after the DBS surgery. Our novel findings provide convincing evidence for the beneficial effect of bilateral STN DBS on sleep and related factors in PD patients. Supporting information {#sec015} ====================== ###### Sleep parameters data, correlation and post hoc t-test analysis. (XLSX) ###### Click here for additional data file. We would like to thank all participants and their families for their support and our team at the Movement Disorders Clinic at Seoul National University Hospital. [^1]: **Competing Interests:**The authors have declared that no competing interests exist.

1. Introduction {#sec1-nutrients-09-01206} =============== Depression is a major mental health problem and is the leading cause of disability worldwide \[[@B1-nutrients-09-01206],[@B2-nutrients-09-01206]\]. According to the World Health Organization (WHO), more than 300 million people suffer from depression worldwide, and it is more common among women than men \[[@B1-nutrients-09-01206]\]. Postpartum depression (PPD), as a common complication of childbearing, is defined as an obvious depressive symptom or a typical depressive episode within 1 to 12 months after delivery \[[@B2-nutrients-09-01206],[@B3-nutrients-09-01206],[@B4-nutrients-09-01206]\]. Due to differences in assessment tools, research methods, survey time points after childbirth, and countries of study samples, the prevalence rate of PPD varies widely worldwide. Previous systemic reviews reported that the prevalence of PPD ranges from 0% to 60% based on 40 countries in 2002 \[[@B5-nutrients-09-01206]\], and from 3.5% to 63% based on 17 Asian countries between 1998 and 2008 \[[@B6-nutrients-09-01206]\]. The symptoms of PPD include reduced interest, poor concentration, sleep deprivation, altered eating pattern, sadness, hopelessness, anxiety, and irritability \[[@B1-nutrients-09-01206],[@B2-nutrients-09-01206],[@B3-nutrients-09-01206],[@B7-nutrients-09-01206]\]. Thus, PPD constitutes an adverse consequence for mothers, children, and families \[[@B2-nutrients-09-01206],[@B3-nutrients-09-01206],[@B7-nutrients-09-01206],[@B8-nutrients-09-01206],[@B9-nutrients-09-01206]\]. For mothers, PPD is associated with adverse health outcomes (e.g., mental impairment, chronicity of the depression, appetite disorder, decline in sleep quality, and even suicide), and adverse behavioural outcomes (e.g., less responsive caregiving, reduced breastfeeding, less safety concern for their children, rejection of their children, abusive behaviour, and infanticide) \[[@B7-nutrients-09-01206],[@B8-nutrients-09-01206],[@B9-nutrients-09-01206]\]. Children of mothers with PPD have higher a risk of negative consequences in social, emotional, cognitive, and developmental areas, such as sleep disturbances, irritability, negative emotionality, poorer physical growth, infant malnutrition, diarrhoea, and respiratory disease \[[@B7-nutrients-09-01206],[@B8-nutrients-09-01206],[@B9-nutrients-09-01206]\]. For families, PPD could cause tension in marital relationships and breakdown of marriages \[[@B7-nutrients-09-01206],[@B8-nutrients-09-01206],[@B9-nutrients-09-01206]\]. PPD also increases the risk for long-term cognitive impairment, emotional difficulties, and behavioural problems for mothers and children \[[@B2-nutrients-09-01206],[@B3-nutrients-09-01206]\]. At present, the aetiology of PPD is unknown, but it is generally believed to be linked to biological, genetic, hormonal, psychosocial, and environmental factors \[[@B1-nutrients-09-01206],[@B2-nutrients-09-01206],[@B3-nutrients-09-01206],[@B4-nutrients-09-01206],[@B7-nutrients-09-01206],[@B8-nutrients-09-01206],[@B9-nutrients-09-01206],[@B10-nutrients-09-01206]\]. In addition, nutritional factors have been identified as determinants of PPD, such as folate/folic acid (FA), vitamin B-12, calcium, iron, selenium, zinc, and polyunsaturated fatty acids \[[@B3-nutrients-09-01206],[@B8-nutrients-09-01206],[@B11-nutrients-09-01206]\]. Although there are numerous studies examining the relationship between nutrients and PPD, a limited number of investigations assess the link between folate and PPD. However, studies have reported that an adequate level of folate may reduce the risk of depression in the general population \[[@B12-nutrients-09-01206],[@B13-nutrients-09-01206],[@B14-nutrients-09-01206],[@B15-nutrients-09-01206]\]. Furthermore, previous studies have investigated the mechanism of folate on depression in the general population. Folate is essential for the biosynthesis of the monoamine neurotransmitters (e.g., serotonin, dopamine, and norepinephrine). Folate participates in the production of *S*-adenosylmethionine (SAM) through homocysteine (Hcy) remethylation, and SAM is essential for the production of these three aforementioned neurotransmitters \[[@B3-nutrients-09-01206],[@B16-nutrients-09-01206]\]. Lack of folate could inhibit the transformation of Hcy to cysteine, thereby increasing the plasma levels of Hcy \[[@B12-nutrients-09-01206],[@B13-nutrients-09-01206]\]. Several studies have shown Hcy levels to be positively correlated with the severity of depression \[[@B12-nutrients-09-01206],[@B13-nutrients-09-01206],[@B14-nutrients-09-01206],[@B15-nutrients-09-01206],[@B16-nutrients-09-01206]\]. Meanwhile, pregnant women require more intake of folate/FA compared to non-pregnant women, and there is a high prevalence of inadequate folate/FA intake among pregnant women \[[@B11-nutrients-09-01206],[@B17-nutrients-09-01206]\]. According to the mechanism of folate on depression, inadequate folate/FA intake during pregnancy may lead to PPD \[[@B3-nutrients-09-01206],[@B11-nutrients-09-01206]\]. Currently, PPD is a common and serious mental disorder, and there is no effective treatment \[[@B2-nutrients-09-01206],[@B18-nutrients-09-01206]\]. Therefore, it is of great practical significance to explore the prevention of PPD. Although previous studies have identified that adequate levels of folate prevent the onset of depression in the adolescent, adult, and elderly population, pregnant women are excluded as a special population \[[@B3-nutrients-09-01206],[@B12-nutrients-09-01206]\]. Moreover, few studies associate folate deficiency with a high risk of PPD \[[@B3-nutrients-09-01206],[@B11-nutrients-09-01206]\], but to the best of our knowledge, none of these focused on the link between FA supplementation during pregnancy and PPD in Chinese women. The FA supplementation is an effective way to protect against folate deficiency, and thus, it is important to explore an appropriate way of FA supplementation, in order to prevent PPD. Furthermore, the onset of PPD between 6 and 12 weeks after giving birth is most often in the first postpartum year \[[@B3-nutrients-09-01206],[@B19-nutrients-09-01206]\], but no study addresses the association between FA supplementation during pregnancy and PPD at 6--12 weeks postpartum. This study is one of few to examine the effect of the duration of FA supplementation during pregnancy on PPD in Chinese women, and thus, may provide new information regarding the potential beneficial effect of long term FA supplementation during pregnancy. The purpose of this investigation is to assess the association between the duration of FA supplementation during pregnancy, and the onset of PPD 6--12 weeks after delivery in Chinese women. 2. Materials and Methods {#sec2-nutrients-09-01206} ======================== 2.1. Population {#sec2dot1-nutrients-09-01206} --------------- This cohort study recruited participants at 6--12 weeks postpartum, who had their first postnatal check-ups at two maternal and child healthcare centres in Tianjin, China, between July 2015 and March 2017. Questionnaires were used to collect participant information regarding socio-demographic and lifestyle characteristics, obstetric history, FA supplementation, and depressive symptoms, by face-to-face interviews. A total of 1914 women, who delivered a singleton foetus, were recruited in this study, while 322 of them were excluded: 58 women had missing data regarding socio-demographic and lifestyle factors, obstetric history, and FA supplementation; 129 women did not use FA during pregnancy; and 135 women used FA in the preconceptional period only. Finally, 1592 participants who took FA supplements during pregnancy were involved in the further analysis ([Figure 1](#nutrients-09-01206-f001){ref-type="fig"}). All participants provided informed consent in writing. The protocol of this study was approved by the Ethics Committee of the Tianjin Medical University. 2.2. Depressive Symptoms {#sec2dot2-nutrients-09-01206} ------------------------ The Chinese version of the Self-Rating Depression Scale (SDS) was used to assess depressive symptoms. There were 20 items in the SDS, including 10 positive items and 10 negative items. Each item was scored on 4-point scale to evaluate the frequency of symptoms: "1" none or a little time; "2" a small part of the time; "3" a lot of time; "4" most of the time. The raw score was a sum of the scores from the 20 items, which was multiplied by 1.25 to yield the standard score \[[@B20-nutrients-09-01206],[@B21-nutrients-09-01206]\]. Based on the standard score, a cut-off point of 50 was used to define depression symptoms among the participants: no depression (\<50), and depression (≥50) \[[@B21-nutrients-09-01206],[@B22-nutrients-09-01206],[@B23-nutrients-09-01206]\]. Furthermore, a higher score reflected more severe symptoms of PPD. 2.3. Folic Acid Supplementation {#sec2dot3-nutrients-09-01206} ------------------------------- A self-reported questionnaire was used to collect the information on FA supplementation during pregnancy, including the supplement brand name, the time of supplement initiation and termination, intake duration, frequency, and dose. Participants were assessed about their FA supplementation status during pregnancy by using their recall response. FA users included those who used FA supplements, as well as those who used FA-containing multivitamins. Among 1592 participants, 803 (50.4%) participants took FA supplements during pregnancy for a duration of 3 months or less, 146 (9.2%) took FA supplements during pregnancy for a duration 4--6 months, and 643 (40.4%) took FA supplements during pregnancy for a duration of more than 6 months. According to the feature of the FA supplementation's duration in this cohort, participants were divided into two groups: FA-users ≤ 6 months and FA-users \> 6 months. 2.4. Covariates {#sec2dot4-nutrients-09-01206} --------------- A general health questionnaire was used to collect information on demographic characteristics, lifestyle, and obstetric history. Education level was classified into three categories: ≤12 years, 13--16 years, and \>16 years. Household income, which was the total monthly income for all family members, was divided into three categories: \<5000 Chinese Yuan (CNY), 5000--10,000 CNY, and \>10,000 CNY. The variable "live with parents" (own parents and/or parents-in-law) was classified into two categories: yes and no. Parity was divided into two categories: primipara (no previous children) and multipara (≥1 previous child). Delivery mode included vaginal birth and caesarean birth, and gestational age at delivery was divided into two categories (\<37 weeks and ≥37 weeks). 2.5. Statistical Analysis {#sec2dot5-nutrients-09-01206} ------------------------- All analyses were conducted using the Statistical Package for the Social Sciences software version 22.0 (SPSS Inc., Chicago, IL, USA). The chi-squared test and independent-sample t test was used to assess differences in participants' characteristics and depressive symptoms stratified between FA-users ≤ 6 months and FA-users \> 6 months. Propensity score matching (PSM) was used to minimise the effects of confounding variables related to differences in the two groups of participants (FA-users ≤ 6 months and FA-users \> 6 months), in order to assess the association between duration of FA supplementation and the onset of PPD. Propensity is the probability of inclusion into the two groups (FA-users ≤ 6 months and FA-users \> 6 months) depending on the respective participants' characteristics. The propensity score is the predicted probability for each participant from a logistic regression model with the duration of FA usage as the dependent variable and all the participant characteristics as independent variables \[[@B24-nutrients-09-01206]\]. Propensity scoring generates a single score based on the observed covariate data to effectively match women from different groups. In this study, propensity scores were calculated by using a logistic regression model and the following covariates: age, education, household income, live with elders, parity, delivery mode, and gestational age at delivery. Thus, the distributions of characteristics were similar between FA-users ≤ 6 months and FA-users \> 6 months. We did a 1:1 nearest neighbour matching with a caliper distance of 0.02. The caliper is a preset range of PSM. When the difference of propensity score between subjects in two groups is within this range, these two subjects could be matched. Common values of caliper are 0.005, 0.01, 0.02, 0.03, and 0.1 \[[@B25-nutrients-09-01206]\]. According to the results of PSM, a total of 1202 women (601 FA-users ≤ 6 months & 601 FA-users \> 6 months) were involved in the further analysis ([Figure 1](#nutrients-09-01206-f001){ref-type="fig"}). The association between the duration of FA supplementation and PPD would be examined by comparing the matched data. Pearson's chi-squared test and independent-sample t test was used to compare PPD prevalence rates between two propensity score-matched groups. Finally, logistic regression analysis was used to estimate the odds ratio (OR) and 95% confidence interval (CI), in order to further confirm the findings of the PSM. Age, education, household income, live with parents, parity, delivery mode, and gestational age at delivery were evaluated for inclusion in the logistic regression model. *P*-values \< 0.05 was considered statistically significant. 3. Results {#sec3-nutrients-09-01206} ========== A total of 1592 women were enrolled in this study, and the mean age of participants was 31.03 years (standard deviation (SD) = 3.81, age range: 21--43 years). Approximately 90% of participants had post-secondary education, and 68% of the women earned a household income of 10,000 CNY and less. Approximately 30% of participants lived with their parents, including their own parents and/or parents-in-law. More than 70% of the women were primiparous, and over half of the infants were delivered via caesarean birth. Only 5.4% of the infants were premature. Among the 1592 FA-users, 643 (40.4%) took FA supplements during pregnancy for a duration of greater than six months. Women with FA supplementation \>6 months were older, had higher education levels, and higher household income (*P* \< 0.05) ([Table 1](#nutrients-09-01206-t001){ref-type="table"}). Among these 1592 women, 29.4% were identified as having PPD. The prevalence of PPD was significantly higher among participants who reported taking FA for a duration of 6 months or less during pregnancy than those who reported taking FA for a duration of more than 6 months during pregnancy (*P* \< 0.05) ([Table 1](#nutrients-09-01206-t001){ref-type="table"}). According to results of PSM, a total of 1202 women were involved in the further analysis. After adjusting for propensity scores in the current analysis, there were no significant differences in the baseline characteristics of the women in both groups (FA-users ≤ 6 months and FA-users \> 6 months) (*P* \> 0.05). The prevalence of PPD decreased from 32.2% to 30.6% among women with a period of FA intake of 6 months or less, and increased from 25.2% to 25.3% among women with a period of FA intake of more than 6 months; these differences in both groups were statistically significant (*P* \< 0.05) ([Table 2](#nutrients-09-01206-t002){ref-type="table"}). From the results of the logistic regression analysis, the intake of FA for more than 6 months was an independent determinant of PPD, with an OR of 0.76 (95% CI: 0.59--0.98, *P* \< 0.05) ([Table 3](#nutrients-09-01206-t003){ref-type="table"}). Thus, FA supplementation for more than 6 months during pregnancy was associated with a decreased risk of PPD. 4. Discussion {#sec4-nutrients-09-01206} ============= Depression is one of the most common and severe mental disorders worldwide, with a prevalence rate in females 2- to 3-times higher than in males \[[@B1-nutrients-09-01206],[@B3-nutrients-09-01206],[@B16-nutrients-09-01206]\]. It has been indicated that the incidence of depression is high during puberty, as well as the premenstrual, periconceptional, and menopausal periods \[[@B1-nutrients-09-01206],[@B3-nutrients-09-01206]\]. Particularly, PPD as a special form of depression has become a significant public concern. Numerous previous studies have reported prevalence rates of PPD in women from different countries. According to study results in Chinese women, the incidence rate ranged from 6.5% to 30% \[[@B7-nutrients-09-01206],[@B9-nutrients-09-01206],[@B26-nutrients-09-01206],[@B27-nutrients-09-01206]\]. Several studies have reported that the prevalence rate of PPD in Canadian, Malaysian, American, Vietnamese, and Korean women is 8.0% \[[@B28-nutrients-09-01206]\], 14.3% \[[@B29-nutrients-09-01206]\], 15.4% \[[@B30-nutrients-09-01206]\], 19.3% \[[@B4-nutrients-09-01206]\], and 36.3--36.7% \[[@B31-nutrients-09-01206]\], respectively. The present findings show that the prevalence rate of PPD is 29.4% in women who took FA supplements during pregnancy. In present study, women, who did not take FA supplements during pregnancy and took FA supplements in the preconceptional period only, were excluded. The reasons are as follows. Firstly, the aims of this study are to assess the association between the duration of FA supplementation during pregnancy and the onset of PPD 6--12 weeks after delivery in Chinese women. Secondly, these are not common phenomenon in China, that women do not take FA supplements or took FA supplements in the preconception only. The Chinese government places a high priority on promoting the prevalence of periconceptional FA supplementation. In China, reproductive women are recommended to take 400 μg/day of FA supplementation from three months preconception until the end of the first trimester of pregnancy by the National Health and Family Planning Commission \[[@B32-nutrients-09-01206]\]. Moreover, free FA supplements have been provided to reproductive women living in rural areas by a national project "folic acid supplementation to prevent neural tube defects" since 2009 \[[@B32-nutrients-09-01206]\]. Therefore, small sample sizes of women who never used FA supplements, or use FA supplements in the preconception only, are generated. If women who never used FA supplements or used FA supplements in the preconception only are included in this study, the unbalanced sample size will influence the statistical power of results. Thus, this study focuses on women who took FA supplements during pregnancy. According to previous studies worldwide, the determinants of PPD have included demographic factors, socio-economic status, family history and social support, environmental and cultural factors, nutrients, inflammatory factors, hormones, and biological factors \[[@B1-nutrients-09-01206],[@B2-nutrients-09-01206],[@B3-nutrients-09-01206],[@B4-nutrients-09-01206],[@B5-nutrients-09-01206],[@B6-nutrients-09-01206]\]. Several investigations reported that folate level was associated with cognitive function and depression in the general population \[[@B12-nutrients-09-01206],[@B13-nutrients-09-01206],[@B14-nutrients-09-01206],[@B15-nutrients-09-01206]\], while few previous studies focused on the relationship between folate and PPD \[[@B11-nutrients-09-01206],[@B16-nutrients-09-01206],[@B33-nutrients-09-01206],[@B34-nutrients-09-01206],[@B35-nutrients-09-01206]\]. However, the results of studies on the relationship between FA supplementation during pregnancy and the onset of PPD are conflicting. FA supplementation during pregnancy has been identified as a protective factor for PPD in some studies \[[@B11-nutrients-09-01206],[@B16-nutrients-09-01206]\], while others failed to establish a link between FA supplementation and PPD \[[@B33-nutrients-09-01206],[@B34-nutrients-09-01206],[@B35-nutrients-09-01206]\]. For example, Lewis et al. \[[@B11-nutrients-09-01206]\] reported that women who took FA supplementation during pregnancy had lower scores for depression 21 months postpartum, compared to women who did not take supplements. Behzadi et al. \[[@B16-nutrients-09-01206]\] also identified FA as preventing the onset of antepartum depression and PPD in a previous study. In contrast, a cohort study in Singapore showed there was no difference in maternal folate levels between participants with and without PPD \[[@B33-nutrients-09-01206]\]. Miyake et al. \[[@B34-nutrients-09-01206]\] found no association between dietary FA intake during pregnancy and PPD in Japanese women with inadequate folate intake. Blunden et al. \[[@B35-nutrients-09-01206]\] reported that red blood cell folate concentrations, which were measured before pregnancy and during the first trimester, were not associated with PPD. In this previous study, the average time interval between the measurement of red-cell folate during pregnancy and screening for PPD was one year; hence, folate levels in the postpartum period were not consistent with the folate level in early pregnancy. Significantly, investigations regarding the association between the duration of FA supplementation during pregnancy and the onset of PPD are limited. According to a previous review involving 36 countries, the recommended duration of FA supplementation during pregnancy varies for different countries; however, no consensus has been reached regarding this issue \[[@B36-nutrients-09-01206]\]. The aim of this study was to examine the association between the duration of FA supplementation during pregnancy and the onset of PPD, in order to provide useful information regarding appropriate duration of FA supplementation during pregnancy. The present findings show that long-term FA supplementation (\>6 months) during pregnancy is significantly associated with a lower prevalence of PPD. This result not only demonstrates the relationship between FA supplementation during pregnancy and PPD, but also emphasises the importance of long-term use in order to ensure that FA reaches significant levels in the body to activate physiological function. Regarding socio-demographic factors linked to PPD, existing studies have assessed predictors of socio-economic status, demographic factors, lifestyle, and obstetric history \[[@B7-nutrients-09-01206],[@B8-nutrients-09-01206],[@B9-nutrients-09-01206],[@B10-nutrients-09-01206]\]. Based on previous literature, populations with higher prevalence of PPD were more likely to be women with lower education levels, lower household income, poor relationship with family members (e.g., husbands, parents or parents-in-law), primiparous, who had undergone caesarean delivery, and delivered premature babies \[[@B7-nutrients-09-01206],[@B8-nutrients-09-01206],[@B9-nutrients-09-01206],[@B10-nutrients-09-01206]\]. Furthermore, previous studies have also reported that FA supplementation during pregnancy could be influenced by childbearing age, maternal education level, household income, and parity \[[@B37-nutrients-09-01206],[@B38-nutrients-09-01206],[@B39-nutrients-09-01206]\]. Women who reported using FA supplements for a longer duration tend to be older, with higher education level, with higher family income, and primiparous \[[@B37-nutrients-09-01206],[@B38-nutrients-09-01206],[@B39-nutrients-09-01206]\]. Therefore, the present study factored in covariates into the analysis, including age, education, household income, living with parents, parity, delivery mode, and gestational age at delivery. The present results showed that women with FA supplementation for more than 6 months during pregnancy tended to be older, had higher education levels, and higher household income (*P* \< 0.001). Thus, PSM was used to minimise the effects of confounding variables related to differences in the two groups of participants (FA-users ≤ 6 months and FA-users \> 6 months), in order to assess the association between the duration of FA supplementation during pregnancy and the onset of PPD. PSM is widely used in statistical analyses, and can effectively match participants from different groups \[[@B24-nutrients-09-01206],[@B40-nutrients-09-01206]\]. To the best of our knowledge, this study is the first to apply PSM in evaluating the association between the duration of FA supplementation during pregnancy and the onset of PPD. After using PSM, the two groups (FA-users ≤ 6 months and FA-users \> 6 months) were comparable after controlling for multiple variables, including age, education, household income, living with parents, parity, delivery mode, and gestational age at delivery. The results after adjusting for propensity scores showed that longer durations of FA supplementation during pregnancy were associated with a lower prevalence of PPD. The logistic regression analysis also yielded similar results. The main strength of this study is that it provided useful information regarding the effect of the duration of FA intake during pregnancy on the onset of PPD. It demonstrated that populations with FA usage more than 6 months during pregnancy had a lower PPD prevalence. Also, this study included a large number of participants, and balanced the confounding factors more efficiently by using PSM, which could reduce the level of bias due to confounding variables \[[@B24-nutrients-09-01206],[@B40-nutrients-09-01206]\]. PSM could include more confounders, be a better solution of eliminating collinearities, and obtain an estimate of treatment effect which is closer to the true average treatment effect, compared with logistic regression model \[[@B41-nutrients-09-01206]\]. Thus, the present findings were more objective and accurate. There is no denying that this study has a few limitations. First, recall bias was possible, because the data on FA supplementation was collected 6--12 weeks after giving birth. The face-to-face interviews were conducted by professional and well-trained investigators, in order to obtain more accurate information. Second, this study investigated FA supplementation, but dietary folate and serum folate levels were not measured. The serum folate levels depend on dietary folate and FA supplementation. Serum levels of folate and Hcy should be assessed in future studies. Third, the SDS was used as a screening tool, but clinical diagnoses were lacking. Although SDS scores are inconsistent with clinical diagnoses, the total scores showed depressive symptoms to be related to clinical outcomes \[[@B22-nutrients-09-01206]\]. Moreover, the SDS was designed for the general population, and may be less sensitive during the screening of pregnant women. However, previous literatures have confirmed that the SDS can be used to evaluate depression among women post-delivery \[[@B42-nutrients-09-01206],[@B43-nutrients-09-01206],[@B44-nutrients-09-01206]\]. Finally, although several covariates were included in the matching process, more confounding factors should be assessed in future research. For example, antenatal depression and PPD might be a continuum reflecting an underlying chronic condition among women during pregnancy and thereafter \[[@B3-nutrients-09-01206],[@B7-nutrients-09-01206],[@B22-nutrients-09-01206]\]. The unmeasured factors may have an impact on the observed association. Therefore, we suggest that future studies could assess these unmeasured confounding factors further. 5. Conclusions {#sec5-nutrients-09-01206} ============== This study showed a high prevalence rate of PPD among Chinese women. FA supplementation for more than 6 months during pregnancy was found to be associated with a decreased risk of PPD 6--12 weeks after delivery in Chinese women. The finding provides new information regarding the potential beneficial effect of long-term FA usage during pregnancy. This research was supported by a grant from the National Natural Science Foundation of China (grant number: 81472967). Guowei Huang contributed to the conception, study design, subsequently manuscript revising and study supervision. Jing Yan participated in planning and designing of the study, data collection, data analysis and manuscript writing. Yuyan Liu participated in data collection, data analysis and drafting manuscript. Lujia Cao and Yuzhi Zheng performed participants' recruitment, data collection, and results interpretation. Wen Li participated in study design and subsequently manuscript revising. All of the authors have read and approved the final manuscript. The authors declare no conflict of interest. The following abbreviations are used in this manuscript: CIconfidence intervalCNYChinese YuanFAfolic acidHcyhomocysteineORodds ratioPPDpostpartum depressionPSMpropensity score matchingSAM*S*-adenosylmethionineSDstandard deviationSDSself-rating depression scaleWHOWorld Health Organization {#nutrients-09-01206-f001} nutrients-09-01206-t001_Table 1 ###### Characteristics of study participants and prevalence of PPD according to the duration of FA supplementation. Characteristics FA-Users ≤ 6 Months FA-Users \> 6 Months *P* ^a^ ------------------------------------------------ --------------------- ---------------------- --------- Age (*years*, *mean* ± *SD*) 30.7 ± 4.0 31.5 ± 3.5 \<0.001 Education (*years*, *n* (%)) \<0.001 ≤12 132 (13.9%) 33 (5.1%) 13--16 721 (76.0%) 486 (75.6%) \>16 96 (10.1%) 124 (19.3%) Household income (*CNY*, *n* (%)) \<0.001 \<5000 263 (27.7%) 97 (15.1%) 5000--10,000 443 (46.7%) 280 (43.5%) \>10,000 243 (25.6%) 266 (41.4%) Live with parents (*n* (%)) 0.554 Yes 266 (28.0%) 189 (29.4%) No 683 (72.0%) 454 (70.6%) Parity (*n* (%)) 0.323 Primipara 657 (69.2%) 460 (71.5%) Multipara 292 (30.8%) 183 (28.5%) Delivery mode (*n* (%)) 0.854 Vaginal birth 462 (48.7%) 310 (48.2%) Caesarean birth 487 (51.3%) 333 (51.8%) Gestational age at delivery (*weeks*, *n* (%)) 0.195 \<37 57 (6.0%) 29 (4.5%) ≥37 892 (94.0%) 614 (95.5%) Prevalence of PPD (*n* (%)) 0.002 No depression ^b^ 643 (67.8%) 481 (74.8%) Depression ^b^ 306 (32.2%) 162 (25.2%) PPD: postpartum depression; FA: folic acid; CNY: Chinese Yuan; ^a^ Analysis using a chi-squared test and independent-sample *t* test; ^b^ Women were divided into two groups according to self-rating depression scale scores, \<50 for no depression, ≥50 for depression. nutrients-09-01206-t002_Table 2 ###### Characteristics of study participants and prevalence of PPD according to the duration of FA supplementation after propensity matching (*n* (%)). Characteristics FA-Users ≤ 6 Months FA-Users \> 6 Months *P* ^a^ ---------------------------------------------- --------------------- ---------------------- --------- Age (*years, mean ± SD*) 31.7 ± 3.8 31.4 ± 3.5 0.107 Education (*years, n (%)*) 0.731 ≤12 37 (6.2%) 33 (5.5%) 13--16 470 (78.2%) 481 (80.0%) \>16 94 (15.6%) 87 (14.5%) Household income (*CNY, n (%)*) 0.279 \<5000 117 (19.5%) 97 (16.1%) 5000--10,000 276 (45.9%) 279 (46.4%) \>10,000 208 (34.6%) 225 (37.5%) Live with parents (*n* (%)) 0.610 Yes 168 (28.0%) 176 (29.3%) No 433 (72.0%) 425 (70.7%) Parity (*n* (%)) 0.344 Primipara 414 (68.9%) 429 (71.4%) Multipara 187 (31.1%) 172 (28.6%) Delivery mode (*n* (%)) 0.526 Vaginal birth 302 (50.2%) 291 (48.4%) Caesarean birth 299 (49.8%) 310 (51.6%) Gestational age at delivery (*weeks, n (%)*) 0.791 \<37 31 (5.2%) 29 (4.8%) ≥37 570 (94.8%) 572 (95.2%) Prevalence of PPD (*n* (%)) 0.040 No depression ^b^ 417 (69.4%) 449 (74.7%) Depression ^b^ 184 (30.6%) 152 (25.3%) PPD: postpartum depression; FA: folic acid; CNY: Chinese Yuan; ^a^ Analysis using a chi-squared test and independent-sample *t* test; ^b^ Women were divided into two groups according to self-rating depression scale scores, \<50 for no depression, ≥50 for depression. nutrients-09-01206-t003_Table 3 ###### The association between the duration of FA supplementation during pregnancy and risk of PPD. Duration of FA Supplementation OR ^a^ 95% CI *P* ^b^ -------------------------------- -------- ------------ --------- ≤6 months 1 \>6 months 0.76 0.59--0.98 0.037 FA: folic acid; PPD: postpartum depression; OR: odds ratio; CI: confidence interval; ^a^ Adjusted for age, education, household income, live with elders, parity, delivery mode, and gestational age at delivery; ^b^ Analysis using a multivariable logistic regression model.

{#sp1 .74} {#sp2 .75} {#sp3 .76} {#sp4 .77} {#sp5 .78} {#sp6 .79} {#sp7 .80} {#sp8 .81} {#sp9 .82} {#sp10 .83} {#sp11 .84} {#sp12 .85} {#sp13 .86} {#sp14 .87} {#sp15 .88} {#sp16 .89} {#sp17 .90} {#sp18 .91} {#sp19 .92} {#sp20 .93} {#sp21 .94} {#sp22 .95} {#sp23 .96} [^1]: 1\. *A System of Medicine*. Edited by J. Russell Reynolds, M.D., F.R.S. Vol. IV, containing *Diseases of the Heart*. 1877. 2\. *Cyclopædia of the Practice of Medicine*. Edited by Dr. H. von Ziemssen, Professor of Clinical Medicine in Munich. Vol. vi, *Diseases of the Circulatory System, &c*. English translation, edited by A. H. Buck, M.D. New York. London, 1876.

Background ========== With the continuous improvement of surgical techniques, partial nephrectomy (PN) is currently applied to patients with clinical stage T1a and T1b (maximum diameter ≤7 cm) renal masses \[[@b1-medscimonit-23-6026]\]. PN is also internationally recommended for those with solitary kidney, contralateral renal insufficiency, or bilateral renal cell carcinoma. However, the risk of perioperative complications is high, especially when tumor complexity exists. Traditional scoring systems such as the RENAL score and PADUA prediction score are used to predict the complexity of PN \[[@b2-medscimonit-23-6026],[@b3-medscimonit-23-6026]\]. However, these scoring systems which focus only on the renal morphometry do not consider the patient-specific factors, one of which is adherent perinephric fat (APF). APF refers to the inflammatory adhesion between perirenal fat and renal parenchyma, which has been shown to be correlated with surgical difficulty and postoperative outcomes \[[@b4-medscimonit-23-6026]--[@b7-medscimonit-23-6026]\]. Davidiuk et al. developed an imaging-based scoring system called the Mayo Adhesive Probability (MAP) score, which is a quantitative indicator of APF used to predict the difficulty of PN. Preliminary studies have shown that the MAP scoring system is simple, objective, and feasible \[[@b8-medscimonit-23-6026]--[@b10-medscimonit-23-6026]\]. In the present study, we sought to explore the influencing factors of MAP score itself on PN by analyzing patient and tumor characteristics. To the best of our knowledge, this is the first study of influencing factors of the MAP score. Material and Methods ==================== Patient population ------------------ This study was approved by the Ethics Committee of Peking University First Hospital. A total of 97 consecutive patients underwent partial nephrectomy in the Urology Department, Peking University First Hospital between September 2015 and June 2016. All patients underwent contrast-enhanced CT scanning preoperatively in or outside our hospital. In this study, we only included those who underwent CT scanning (Philips Brilliance 64 CT Scanner, Amsterdam, the Netherlands) in our hospital, then 4 patients were consequently excluded. Finally, 93 patients were enrolled and their radiological and clinicopathologic data were retrospectively reviewed. Radiological evaluation ----------------------- All images were processed using Carestream Vue PACS (Carestream, Rochester, NY). With blinding of the readers to patients' APF status, preoperative CT images were assessed by 2 senior radiologists with similar seniority. All data were measured at the level of the renal vein as described by Davidiuk et al. \[[@b8-medscimonit-23-6026]\]. As similarly described by Eisner et al. \[[@b11-medscimonit-23-6026]\], some radiological elements which were not involved in MAP score were also collected. The posterior perirenal fat thickness (P), anterior perirenal fat thickness (A), medial perirenal fat thickness (M), lateral perirenal fat thickness (L), anterolateral perirenal fat thickness (AL), and posterolateral perirenal fat thickness (PL) on the same side of the tumor were measured ([Figure 1](#f1-medscimonit-23-6026){ref-type="fig"}). Here, we define the posterior perirenal fat thickness as the length of a direct line posteriorly from the renal capsule to the posterior abdominal wall, with its reverse line intersecting with the point of the end of the renal vein. Posterolateral perirenal fat thickness was measured from the renal capsule to the lateral abdominal wall on the projected line from the renal vein. Similarly, the other 4 measurements (anterior, medial, lateral, and anterolateral) were defined as the distance from the kidney to nearest viscera or muscle. Each corresponding result was averaged and the evaluation of MAP score was calculated as described by Davidiuk et al. \[[@b8-medscimonit-23-6026]\]. Statistical analyses -------------------- For continuous variables, the sample median (minimum and maximum) is listed. Categorical variables were reported as proportions with number of patients (percentage). Ordinal logistic regression analyses (if feasible) were used to compare both continuous and categorical variables. For dependent variables, MAP score=0 was taken as reference, and link function was Logit. If *p*\<0.05 in the test of parallel lines in ordinal logistic regression, multiple logistic regression analyses were applied. All reported *p*-values are 2-sided and statistical significance was set at \<0.05. All statistical analyses were conducted with SPSS version 20.0 (IBM Corp, Armonk, NY). Results ======= Patients' characteristics ------------------------- Patients' radiological and clinicopathologic characteristics are summarized in [Table 1](#t1-medscimonit-23-6026){ref-type="table"}. The median age was 54 years, and the median BMI was 25.1 kg/m^2^. Most patients were men (65.6%). A total of 6 patients underwent open partial nephrectomy, and the other 87 cases underwent laparoscopic PN. For postoperative pathologic data, tumor stage was only available for renal cell carcinoma patients; tumor grade was unavailable for 20 patients, including 15 patients with benign tumor (14 angioleiomyolipoma and 1 neurilemmoma) and 5 patients with chromophobe renal cell carcinoma. RCC comprised 83.9% of the total tumors (without positive surgical margins). Tumor stage of most RCC was 1a (89.7%). MAP scores of 81.7% of tumors were from 0 to 3. Univariate analysis of associations of patients' characteristics with MAP score is shown in [Table 2](#t2-medscimonit-23-6026){ref-type="table"}. *P*\<0.05 was considered statistically significant. On univariate analysis, MAP score was significantly associated with male sex (*p*\<0.001), older age (*p*=0.001), higher body mass index (*p*=0.001), history of hypertension (*p*=0.004) and diabetes mellitus (*p*\<0.001), and greater perirenal fat thickness (posterolateral \[*p*\<0.001\], lateral \[*p*=0.002\], anterior \[*p*=0.002\], anterolateral \[*p*=0.002\], and medial \[*p*\<0.001\]). The groups were not significantly comparable for tumor side, size, or type (malignant *vs.* benign). No significant difference was found in tumor stage or grade in RCC. As displayed in [Table 3](#t3-medscimonit-23-6026){ref-type="table"}, variables for which *p*\<0.1 in univariate analysis were selected for multivariable analysis. Only posterolateral perirenal fat thickness (odds ratio \[OR\]=0.88 \[0.82--0.95\], *p*=0.001), medial perirenal fat thickness (OR=0.90 \[0.83--0.98\], *p*=0.01), and history of diabetes mellitus (OR=5.42 \[1.74--16.86\], *p*=0.004) remained statistically associated with MAP score. Discussion ========== Partial nephrectomy is currently recommended as a standard treatment for patients with stage T1 renal cell carcinoma in both the American Urological Association and the European Association of Urology guidelines \[[@b1-medscimonit-23-6026],[@b12-medscimonit-23-6026]\]. To predict the difficulty of PN, scoring systems focusing on renal morphometry, such as the RENAL score and PADUA prediction score, have been introduced, but these scoring systems do not include the patient-related factors such as adherent perinephric fat. APF can often cause more surgical difficulties in kidney mobilization and separating the renal tumor from renal parenchyma \[[@b5-medscimonit-23-6026]\]. Chang et al. noted in 2015 that the presence of APF significantly increased the operative time and complexity of kidney mobilization and tumor isolation \[[@b13-medscimonit-23-6026]\]. Similarly, Khene et al. analyzed the data of 202 patients who underwent robot-assisted partial nephrectomy (RAPN) and noted that the average operative time for patients with APF was 40 min longer, with 2-fold more blood loss and an increased risk of converting the operation into open surgery or radical nephrectomy \[[@b14-medscimonit-23-6026]\]. Bylund et al. first sought to explore the factors related to APF, and found that there was a significant correlation between APF and male sex, perirenal fat thickness, and stranding \[[@b4-medscimonit-23-6026]\], but the study was limited by sample size and the rather subjective evaluation of APF. Davidiuk et al. in 2014 prospectively analyzed 100 patients with RAPN and created the Mayo Adhesive Probability (MAP) scoring system, which helps to predict the presence of APF and therefore the surgical complexity in PN \[[@b8-medscimonit-23-6026]\]. Chang et al. analyzed the relationship between APF and the patient-specific factors in 43 patients with RAPN, and noted that APF was significantly correlated with type 2 diabetes, perirenal fat stranding, preoperative serum creatinine, medial perirenal fat thickness, and MAP score \[[@b13-medscimonit-23-6026]\]. By studying the clinical data of 245 patients who underwent minimally invasive partial nephrectomy, Kocher et al. showed that the relationship between APF and age, male sex, posterior perirenal fat thickness, perirenal fat stranding, and the MAP score was statistically significant \[[@b7-medscimonit-23-6026]\]. Similarly, in a retrospective case-control study involving 86 patients, Martin et al. confirmed that the MAP score was predictive of APF in open PN \[[@b9-medscimonit-23-6026]\]. The above studies did confirm the ability of MAP score to predict APF. However, the evaluation of APF has always depended on surgeon experience and subjective determination. We noticed that surgeons with different experience had different understanding of adhesion during the operation; therefore, the severity of adhesion may not be objectively reflected merely from the operative notes. Besides, although the MAP score is an easy-to-use scoring system, it does not include all the factors associated with APF. It has been reported that a radioclinical score based on MAP score and other clinical factors was more predictive of APF \[[@b10-medscimonit-23-6026]\]. Our aim in the present study was to find more objective quantitative factors to optimize the MAP score. To avoid the bias caused by subjective conclusions, we directly analyzed the influencing factors of the more objective MAP score itself. Our study is the first to report the influencing factors of the MAP score. In our study, we found the MAP score was associated with diabetes, posterolateral perirenal fat thickness, and medial perirenal fat thickness in multivariate analysis. It is worth noting that previous studies have shown that APF was significantly associated with sex \[[@b4-medscimonit-23-6026],[@b5-medscimonit-23-6026],[@b8-medscimonit-23-6026]\], showing that men are more likely to develop perirenal 'sticky fat', which increases surgical difficulty, perhaps due to differences in fat distribution between men and women. Anderson et al. found that the thickness of subcutaneous fat was greater in women than in men, while men have thicker perirenal fat than women \[[@b15-medscimonit-23-6026]\], which was subsequently confirmed by Eisner et al. \[[@b11-medscimonit-23-6026]\]. However, in our study, the MAP score was not correlated with sex in multivariate analysis. This may be explained by the interaction between sex and perirenal fat thickness. Our study confirmed that the MAP score was not associated with BMI. Previous studies have demonstrated that BMI can reflect the individual total fat content (includes visceral fat). Macleod et al. compared the complexity of RAPN with the measurement of perirenal fat thickness and BMI, noting that it was perirenal fat thickness but not BMI that was associated with increased blood loss and operative time \[[@b16-medscimonit-23-6026]\]. Therefore, compared with BMI, it is more meaningful to focus on the intra-abdominal fat thickness, which may help predict surgical difficulty and serve as a reference for the choice of operation \[[@b14-medscimonit-23-6026]\]. The association between diabetes mellitus and APF has been previously reported \[[@b13-medscimonit-23-6026]\], and our study shows the significant association between diabetes mellitus and MAP score, with the odds ratio of 5.42. Visceral fat accumulation is closely related with posterior perinephric fat thickness, and also has a reciprocal causal relationship with the occurrence and development of type 2 diabetes mellitus. An increase in total and visceral fat can cause insulin resistance and a series of inflammatory responses \[[@b17-medscimonit-23-6026],[@b18-medscimonit-23-6026]\]. Conversely, insulin resistance can make the distribution of adipose tissue change, and visceral fat increases further \[[@b19-medscimonit-23-6026]\]. This change may then have an impact on the body's inflammatory response, since visceral fat can produce higher levels of inflammatory markers compared with subcutaneous fat \[[@b20-medscimonit-23-6026]\]. Thus, type 2 diabetes can increase both the body's visceral fat content and the body's inflammatory response. In addition, the relationship between perirenal fat stranding and APF may also suggest that the inflammatory response plays a role in the development of APF. However, Dariane et al. reported that no inflammatory infiltration was found in perirenal fat \[[@b10-medscimonit-23-6026]\]. The specific mechanism still needs further study. Davidiuk et al. has reported the association between posterolateral perirenal fat thickness and APF \[[@b8-medscimonit-23-6026]\], although the former was not selected to create the MAP score. As for radiological elements not included in the MAP score, our study showed the potential significance of medial perirenal fat thickness, as reported by a previous study \[[@b13-medscimonit-23-6026]\]. Further study needs to be done to evaluate whether an optimized MAP score including more radiological elements can better predict APF. There are some limitations in our study. This was a retrospectively study and results need to be further validated by prospective studies. Imaging evaluation was performed by 2 different radiologists, so the MAP score may be biased. Moreover, the sample size needs to be further enlarged. Further efforts should be focused on creating a new radioclinical score and assessing whether it is more predictive of APF. Conclusions =========== In conclusion, the MAP score is highly correlated with posterolateral perirenal fat thickness, medial perirenal fat thickness, and history of diabetes mellitus, and may be optimized by these influencing factors. It needs to be further explored whether the MAP score can be improved. **Conflicts of Interest** None. **Source of support:** Departmental sources {#f1-medscimonit-23-6026} ###### Patient (n=93) clinicopathologic and radiological characteristics. Variables Summary (n=93) --------------------------------- ------------------- Age (yr) 54 (19--77)  Age \<55 yr 47 (50.5%)  Age ≥55 yr 46 (49.5%) Gender  Male 61 (65.6)  Female 32 (34.4) Body mass index (BMI, kg/m^Ft^) 25.1 (15.6--33.9)  BMI \<25 44 (47.3%)  25≤ BMI ≤30 40 (43.0%)  BMI \>30 9 (9.7%) Hypertension  Yes 44 (47.3%)  No 49 (52.7%) Diabetes  Yes 16 (17.2%)  No 77 (82.8%) Coronary heart disease  Yes 6 (6.5%)  No 87 (93.5%) Side of the tumor  Left 49 (52.7%)  Right 44 (47.3%) Tumor size (cm) 2.5 (0.7--13.0) Tumor type  Renal cell carcinoma 78 (83.9%)  Benign tumor 15 (16.1%) Tumor stage(n=78)  1a 70 (89.7%)  1b 8 (10.3%) Tumor grade(n=73)  G1 49 (67.1%)  G2 and G3 24 (32.9%) Perirenal fat thickness(mm)  Posterior 15.1 (0--39.6)  Posterolateral 17.0 (0--46.5)  Lateral 16.1 (0--53.7)  Anterior 8.6 (0--39.1)  Anterolateral 10.4 (0--32.7)  Medial 7.3 (0--31.2) Stranding score  0 61 (65.6%)  2 19 (20.4%)  3 13 (14.0%) MAP score  0 23 (24.7%)  1 29 (31.2%)  2 12 (12.9%)  3 12 (12.9%)  4 13 (14.0%)  5 4 (4.3%) Continuous variables are listed as the sample median (minimum and maximum) and categorical variables as number of patients (percentage). Tumor stage was only available for renal cell carcinoma patients. Tumor grade was unavailable for 20 patients, including those with benign tumor and some rare types of renal cell carcinoma. ###### Associations of patients' characteristics with MAP score. Variables MAP score *p* Value -------------------------------- ------------------- ------------------- ------------------- ------------------- ------------------- ------------------- ----------------------------------------------------------------- Age (yr) 49 (19--71) 49 (28--73) 59 (29--67) 62.5 (37--77) 54 (42--74) 62.5 (51--72) **0.001** Gender **\<0.001**  Male 8 (34.8%) 20 (69.0%) 9 (75.0%) 9 (75.0%) 11 (84.6%) 4 (100.0%)  Female 15 (65.2%) 9 (31.0%) 3 (25.0%) 3 (25.0%) 2 (15.4%) 0 (0.0%) Body mass index (BMI, kg/m^2^) 22.5 (17.6--30.5) 25.3 (15.6--31.6) 25.6 (20.7--28.7) 26.3 (22.0--28.6) 26.1 (21.3--33.9) 26.8 (22.0--31.8) **0.001** Hypertension **0.004**  Yes 4 (17.4%) 15 (51.7%) 8 (66.7%) 6 (50.0%) 3 (23.1%) 1 (25.0%)  No 19 (82.6%) 14 (48.3%) 4 (33.3%) 6 (50.0%) 10 (76.9%) 3 (75.0%) Diabetes **\<0.001**[\*](#tfn3-medscimonit-23-6026){ref-type="table-fn"}  Yes 2 (8.7%) 2 (6.9%) 0 (0.0%) 2 (16.7%) 9 (69.2%) 1 (25.0%)  No 21 (91.3%) 27 (93.1%) 12 (100.0%) 10 (83.3%) 4 (30.8%) 3 (75.0%) Coronary heart disease 0.416  Yes 0 (0.0%) 3 (10.3%) 1 (8.3%) 0 (0.0%) 2 (15.4%) 0 (0.0%)  No 23 (100.0%) 26 (89.7%) 11 (91.7%) 12 (100.0%) 11 (84.6%) 4 (100.0%) Tumor type 0.086  Renal cell carcinoma 17 (73.9%) 24 (82.8%) 11 (91.7%) 10 (83.3%) 12 (92.3%) 4 (100.0%)  Benign tumor 6 (26.1%) 5 (17.2%) 1 (%8.3) 2 (16.7%) 1 (7.7%) 0 (0.0%) Tumor size (cm) 2.5 (1.0--6.2) 2.7 (1.2--5.0) 2.2 (1.4--13.0) 2.3 (0.7--12.5) 2.5 (1.2--4.2) 2.3 (1.7--3.7) 0.864 Side of the tumor 0.576  Left 12 (52.2%) 17 (58.6%) 7 (58.3%) 5 (41.7%) 6 (46.2%) 2 (50.0%)  Right 11 (47.8%) 12 (41.4%) 5 (41.7%) 7 (58.3%) 7 (53.8%) 2 (50.0%) Perirenal fat thickness (mm)  Posterolateral 7.0 (0--18.7) 17.6 (3.62--35.7) 21.9 (3.7--46.5) 20.1 (9.7--28.6) 19.6 (8.8--24.6) 28.1 (18.1--37.0) **\<0.001**[\*](#tfn3-medscimonit-23-6026){ref-type="table-fn"}  Lateral 5.6 (0--37.7) 18.2 (0--53.68) 24.7 (3.7--52.1) 16.1 (5.52--28.6) 18.6 (12.9--31.9) 18.0 (5.4--40.1) **0.002**[\*](#tfn3-medscimonit-23-6026){ref-type="table-fn"}  Anterior 3.3 (0--23.8) 10.64 (0--21.0) 9.2 (0--37.2) 11.9 (0--31.7) 14.4 (0--39.1) 12.8 (8.0--17.8) **0.002**  Anterolateral 5.0 (0--29.5) 11.3 (3.2--29.0) 15.1 (0--21.5) 8.7 (0--20.7) 10.9 (2.7--32.7) 20.8 (7.7--30.9) **0.002**[\*](#tfn3-medscimonit-23-6026){ref-type="table-fn"}  Medial 3.5 (0--19.8) 9.0 (0--20.0) 7.4 (0--31.2) 8.1 (0--14.3) 9.3 (2.6--22.0) 15.2 (6.5--26.7) **\<0.001** Tumor stage(n=78) 0.153  1a 14 (82.4%) 21 (87.5%) 10 (90.9%) 10 (100.0%) 11 (91.7%) 4 (100.0%)  1b 3 (17.6%) 3 (12.5%) 1 (9.1%) 0 (0.0%) 1 (8.3%) 0 (0.0%) Tumor grade(n=73) 0.777  G1 8 (66.7%) 16 (66.7%) 8 (72.7%) 7 (70.0%) 9 (75.0%) 1 (25.0%)  G2 and G3 4 (33.3%) 8 (33.3%) 3 (27.3%) 3 (30.0%) 3 (25.0%) 3 (75.0%) MAP -- Mayo adhesive probability. Continuous variables are listed as the sample median (minimum and maximum) and categorical variables as number of patients (percentage). Tumor stage was only available for renal cell carcinoma patients. Tumor grade was unavailable for 20 patients, including those with benign tumor and some rare types of renal cell carcinoma. *P* value was based on multiple logistic regression analyses as ordinal logistic regression analyses were unfeasible. ###### Multivariable analysis of influencing factors of MAP score. Variables *p* Value OR 95% CI -------------------------------- ----------- ----------- -------- ------------- Age(yr) 0.001 0.351 Gender \<0.001 0.315 Body mass index (BMI, kg/m^2^) 0.001 0.225 Hypertension 0.004 0.547 Diabetes \<0.001 **0.004** 5.42 1.74--16.86 Tumor type 0.086 0.983 Perirenal fat thickness(mm)  Posterolateral \<0.001 **0.001** 0.88 0.82--0.95  Lateral 0.002 0.083  Anterior 0.002 0.714  Anterolateral 0.002 0.926  Medial \<0.001 **0.010** 0.90 0.83--0.98 CI -- confidence interval; OR -- odds ratio; MAP score -- 0 was taken as reference. [^1]: Study Design [^2]: Data Collection [^3]: Statistical Analysis [^4]: Data Interpretation [^5]: Manuscript Preparation [^6]: Literature Search [^7]: Funds Collection [^8]: These authors contributed equally to this work

1. Introduction {#sec1} =============== Hyperemesis gravidarum (HG) is a severe subtype of nausea and vomiting in pregnancy (NVP) that is estimated to affect 0.3--2% of pregnancies \[[@B1], [@B2]\]. It is a common reason for admission to the hospital in the first trimester of pregnancy \[[@B2]\]. HG typically begins in the first trimester of pregnancy, likely due to increasing amounts of hormones, and usually ends in the second trimester, although some cases have lasted the entire duration of pregnancy \[[@B3]\]. There is an increased incidence in patients with multiparity and elevated hormone levels of pregnancy \[[@B3]\]. Complications of the disorder include electrolyte imbalances and, rarely, fetal demise \[[@B3]\]. Management includes fluid repletion and antiemetic drugs such as first-generation antihistamines, but there is no specific treatment or preventative measure \[[@B1], [@B2]\]. Because HG is a diagnosis of exclusion, organic causes of gastrointestinal, genitourinary, metabolic, neurologic, psychologic, and pharmacologic sources must be considered during work-up \[[@B4]\]. Herein, we describe a case of a pregnant woman with HG who presented with intractable pain mimicking appendicitis. 2. Case Report {#sec2} ============== A 33-year-old gravida 4 para 3 at 14 weeks\' gestation with HG presented to the emergency department with several days of constant, sharp, 10/10 right back pain that radiated to her right flank, with associated nausea and vomiting, but no fevers, dysuria, abdominal pain, or spinal tenderness to palpation. Past medical history was significant for nephrolithiasis, abdominal vein thromboses, preeclampsia, bipolar I disorder, generalized anxiety disorder, and substance abuse. She was diagnosed with a probable nonobstructing 1 mm right kidney stone the day prior via an abdominal computed tomography (CT) scan at an outside hospital that did not visualize the appendix, and she reported similar symptoms with prior episodes of nephrolithiasis. Ultrasound (US) at presentation showed no hydronephrosis. She was admitted due to intractable pain and nausea. On admission, infectious work-up including blood culture and urine was negative. Urology was consulted and concurred that the severity of pain that the patient was experiencing was unlikely to be attributed to a kidney stone of that size. Labs at admission showed a normal white blood cell count of 7.7, but on day two of admission, labs were notable for significant leukocytosis to 20.0 (neutrophilia of 89.3) and mild hyponatremia. The leukocytosis decreased to 12.1 on day three and resolved to 9.7 on day four without intervention. Despite frequent doses of parenteral morphine and adjunctive pain medication, her pain was unremitting. Given pain worsened with vomiting, the suspicion arose that the pain was secondary to muscle spasm from frequent emesis, especially because her pain did not change with positioning, palpation, or movement. Obstetrics was consulted and optimized her antiemetic regimen for HG, while psychiatry managed her bipolar medication. Pain management was consulted to assist given the unrelenting nature of her pain and ever-increasing amounts of opiate requirement. At this time, the etiology appeared to be musculoskeletal given the exacerbation of pain with emesis, absence of leukocytosis, only moderately elevated inflammatory markers (C-reactive protein 55.8, erythrocyte sedimentation rate 31) which can be seen in pregnancy, negative blood cultures, and negative abdominal US to date. Further work-up was performed to rule out other causes. Magnetic resonance imaging (MRI) and magnetic resonance venography (MRV) without contrast of the abdomen and pelvis were performed and were significant for a 12 mm uncomplicated appendicitis with surrounding inflammatory changes but no evidence of perforation ([Figure 1](#fig1){ref-type="fig"}), as well as mild right pelviectasis at 4.9 mm. While leukocytosis remained resolved and blood cultures were negative, procalcitonin was elevated to 0.61 and the decision was made to start patient on IV piperacillin-tazobactam as surgical intervention was not recommended at this point in her pregnancy being borderline second trimester and the uncomplicated nature of her appendicitis. Over the following days, she began to exhibit symptoms of improvement. She completed six days of IV antibiotics and was discharged home with pain controlled and able to tolerate oral intake. 3. Discussion {#sec3} ============= To our knowledge, this is the first reported case of HG presenting as appendicitis. What is notable about our patient is that her intractable pain unresponsive to significant analgesia appeared to resolve with treatment for appendicitis, despite an uncharacteristic presentation with mildly consistent radiographic findings for such. Thereby, the detection and treatment of concurrent organic causes is important as it may provide significant symptomatic relief in cases of concurrent HG. Imaging for appendicitis is typically performed with CT of the abdomen as this imaging modality has the lowest negative appendectomy rate (NAR) and is more diagnostically accurate \[[@B5]\]. However, given that our patient was pregnant, US was performed instead of repeating the CT abdomen as is the standard of care, given the risks of radiation to the fetus with CT. Since the US was not clear, a CT abdomen would have been considered next in work-up had she not been pregnant, as US is not a good modality for ruling out appendicitis \[[@B6]\]; historically, adding on a CT has prevented timely treatment \[[@B7]\]. Furthermore, it is well-supported that appendectomy is the preferred treatment of acute appendicitis in pregnancy \[[@B8]--[@B11]\] given concerns of rupture upon forgoing surgery \[[@B12]\]. A systematic meta-analysis by Lee, et al. showed no concern for fetal loss in laparoscopic appendectomy and no significant difference in surgical outcomes between laparoscopic and open surgery aside from decreased risk of wound infection and length of stay in the former \[[@B11]\]. However, the growing literature has documented the successful treatment of gestational appendicitis with antibiotics \[[@B13], [@B14]\]. While appendectomy remains first-line for appendicitis in pregnancy, this can serve as a useful tool for patients in rural areas who do not have access to immediate surgical treatment and need time to seek appropriate management \[[@B15]\]. Our paper does not aim to focus on appendicitis management but, rather, its varying presentations. Although our patient had a significant psychiatric history, the consulted psychiatrist offered the patient risperidone, which she readily accepted as she was desperate for any adjunct therapies that could provide benefit. Therefore, there were no concerns of opiate-seeking in this case. Organic causes were strongly sought after, but there was no good evidence in her imaging studies for nephrolithiasis or appendicitis. As discussed in the case, the slightly elevated procalcitonin was enough to empirically treat for appendicitis since she was not responding to a maximized pain regimen, and her symptoms improved accordingly, which makes this case interesting. While pain in a patient with HG has a large differential including appendicitis \[[@B16]\], the atypical clinical presentation, mildly consistent radiographic findings, and mildly elevated inflammatory markers which can be normal in a pregnant person delayed early treatment. While it is difficult to rule in a diagnosis when it presents uncharacteristically, this case reiterates the significance of a thorough work-up and differential, as well as a strong clinical, suspicion. Also, in general, it can be difficult to differentiate a patient with appendicitis-like symptoms from one with true appendicitis, with CT/US providing no known increase in the accuracy of diagnosis \[[@B17]\]. Crucially, it is important to remember that HG is a diagnosis of exclusion, and efforts must be made to rule out other diagnoses, especially in cases of intractable pain that could be resolved without reliance on pain control. Disclosure ========== This case report has been presented as an e-poster presentation during Academic Excellence Day for Creighton University, Phoenix Regional Campus in April 2020, Phoenix, AZ. Conflicts of Interest ===================== The authors do not have any conflicts of interest to declare. {#fig1} [^1]: Academic Editor: Timothy J. Craig

Findings ======== Introduction ------------ Health benefits of physical activity (PA) have been demonstrated for many chronic diseases. For example moderate to vigorous intensity activity (MVPA) has been shown to decrease obesity and lower total cholesterol and blood pressure \[[@B1]\]. Accurate measurement of PA is essential in developing intervention strategies. Physical activity questionnaires (PAQ), diaries, observations*,* indirect calorimetry, double-labeled water (DLW), heart rate monitors and accelerometry have been used \[[@B2]-[@B4]\]. Because of the limitation of PAQ methods and the high cost and subject burden associated with direct observation and doubly labeled water, accelerometry has become the method of choice for objective, valid and reliable measurement in adults \[[@B5]\]. The uniaxial ActiGraph accelerometer (ActiGraph^TM^, Pensacola, CA) is widely accepted as valid in assessing PA in laboratory and FLC \[[@B6]-[@B8]\], and has been used in epidemiological studies \[[@B9],[@B10]\]. Even if the triaxial accelerometer measures physical activity during walking with more precision than the uniaxial accelerometer \[[@B11]\], a recent study showed that there is no difference between uniaxial and triaxial accelerometers in the measurement of PA \[[@B12]\]. However, recently, the manufacturer improved the GT1M for a triaxial accelerometer (GT3X). This device may also be used in uniaxial mode (GT1M mode). It is important to determine if there are discrepancies between the two models in assessing time spent in different intensities of PA, using previous thresholds established with old versions of accelerometers, or if the development of new physical activity threshold values is necessary. If the GT3X accelerometer, in uniaxial mode, has different results than the old generation, then studies that use the GT3X cannot be compared with data from previous studies. Additional studies will be necessary to calibrate and validate the new device. To date, there are no published studies comparing the new generation ActiGraph accelerometer (GT3X) and its predecessor (GT1M). The purpose of our study is to compare the time spent at different intensities of PA by simultaneous measurements involving the ActiGraph GT1M and the GT3X accelerometers. Methods ======= Twenty-five healthy sport science students were recruited. Physical characteristics of the subjects are described in Table[1](#T1){ref-type="table"}. Subjects were required to pass a medical examination to exclude contraindications for participating in the study. The purpose and objectives were carefully explained to each subject and written informed consent was obtained. The study was approved by the local Ethics Committee (Comité de Protection des Personnes). ###### **Physical characteristics of subjects (*n*** **= 25)**   **Males** **Females** --------------------------- ------------- ------------- **N** 14 11 **Age (*yr*)** 25.3 ± 4.8 25.5 ± 4.4 **Weight (*Kg*)** 72.1 ± 10.1 61.0 ± 9.8 **Height (*cm*)** 177.2 ± 5.5 170.8 ± 6.6 **BMI (Kg/m**^**2**^**)** 22.9 ± 2.8 20.8 ± 2.1 Weight was measured to the nearest 0.1 kg using an electronic scale (Oregon Scientific®, GA 101, USA). Height was measured without shoes to the nearest 0.1 cm using a stadiometer (Seca®, Hamburg, Germany). Accelerometers were calibrated according to manufacturer specification. The epoch interval used was set at one min and output was expressed as mean counts per minute. All participants wore the two ActiGraph accelerometers (GT1M & GT3X in uniaxial mode) simultaneously, at the level of the back with the same elastic belt and adjustable buckle, during a typical week day. Subjects were instructed to remove the devices during swimming, showering, and bathing. The accelerometers recorded activity during the day, and were removed at night. Data were uploaded from the monitor to a computer after the period test. The following PA thresholds were used: sedentary activity, 0 to 99 counts·min^--1^, light activity 100--1951 counts·min^--1^, moderate activity 1952--5723 counts·min^--1^, and vigorous activity ≥ 5724 counts·min^--1^\[[@B13]\]. The same accelerometers were used for all participants. All analyses were performed using SAS software version 9.2 (SAS Institute Inc., Cary, NC 25513). Physical activity was measured and analyzed in counts per minute. ANOVA compared PA between the two accelerometers. P values \<0.05 were considered statistically significant. Quantitative variables were described by mean and 95% confidence intervals. Reproducibility between GT1M and GT3X accelerometers was assessed with intraclass correlation coefficient (ICC) at each intensity. The scale used for interpretation of concordance was previously described \[[@B14]\]. The Bland and Altman method was used to test agreement of data output between GT1M and GT3X \[[@B15]\]. The analysis measures bias as estimated from mean differences, the 95% confidence interval for bias, and the limits of agreement, ± 2 standard deviations of the difference. The GT1M was used as the reference for analysis because it had been validated previously and calibrated to assess the PA intensity and/or estimate the energy expenditure during normal daily conditions \[[@B6],[@B7],[@B13],[@B16]\]. Results and discussion ---------------------- Participants wore accelerometers an average of 903 ± 137 min. Mean PA during the recording time was 584 ± 205 counts·min^--1^ for the GT1M® and 595 ± 206 counts·min^--1^ for the GT3X®. No significant differences in PA were found between genders (p = 0.22). The concordance correlation coefficient between accelerometers at each intensity was 0.99 (Table[2](#T2){ref-type="table"}). There were no significant differences in intensity between accelerometers at the four intensities (Table[2](#T2){ref-type="table"}). Differences between accelerometers never exceeded 0.56%. ###### Time spent in different intensity of PA expressed in minutes per day for the both accelerometer generations (n = 25) **Intensity** **Mean \[95% IC\]** **Mean difference \[95% IC\]** **ICC** --------------- --------------------------- -------------------------------- ------------------------------ ------ **Sedentary** 683.40 \[625.97; 740.83\] 682.84 \[415.17; 950.51\] 0.56 \[−0.84; 1.96\] ^†^ 0.99 **Light** 132.92 \[80.82; 185.02\] 132.56 \[80.62; 184.52\] 0.36 \[−1.14; 1.86\] ^††^ 0.99 **Moderate** 47.28 \[28.75; 65.81\] 47.80 \[29.06; 66.54\] \- 0.52 \[−1.38; 0.34\] \* 0.99 **Vigorous** 39.44 \[23.98 ; 54.90\] 39.88 \[24.25; 55.51\] \- 0.44 \[−0.96; 1.84\] \*\* 0.99 \* P = 0.25, \*\* P = 0.11, ^†^ P = 0.44, ^††^ P = 0.64. Agreement was assessed at different intensities. Mean differences were within the limits of agreement and most data points were within the limits of agreement of bias (Figure[1](#F1){ref-type="fig"}). {#F1} Agreement between the devices was also compared for their ability to identify participants who met the MVPA activity guideline of 60 min per day. Participants whose PA intensities met PA guidelines were identical for the two accelerometers. Selecting the ActiGraph accelerometer model is an important issue for researchers. The present study showed a concordance and no significance difference between data output obtained by the new vs. the previous generation. Studies have compared different ActiGraph accelerometers in laboratory or FLC \[[@B8],[@B17]-[@B19]\]. Using a motorized table with a wide range of amplitude and frequencies to assess three generations of ActiGraph monitors, significant differences in activity counts between generations of accelerometers were reported \[[@B8]\]. Results of the study showed inter-accelerometer variability consistently better with GT1M compared with the 7164 or 71256 accelerometers for all frequencies, with a mean difference of ± 20%. Therefore, conclusions about differences among three generations in the present study have to be considered with caution because of intermonitor variability \[[@B8]\]. A study comparing adolescents using the GT1M ActiGraph (Version 1) and Model 7164 in FLC found no significant difference in time spent in moderate and vigorous physical activity when using the same epoch length, although differences were observed in sedentary and light-intensity activity \[[@B17]\]. Compared with Model GT1M, Model 7164 exhibited significantly less time as sedentary and more time as light-intensity activity (P \< 0.001). Corder et al (2007) concluded that data from the GT1M can be compared with historical data using average counts per minute, and the two models are comparable when measuring time spent in MVPA in children using the same epoch length \[[@B17]\]. Differences in time spent at different intensities of PA were not significant. Kozey et al (2010) compared the ActiGraph accelerometer model 7164 with the ActiGraph GT1M during self-paced locomotion at three speeds of walking \[[@B19]\] and concluded that the GT1M is comparable to Model 7164 when estimating habitual activity intensities \[[@B19]\]. A study comparing activity counts between the ActiGraph 7164 and the three versions of the GT1M at given walking and running speeds concluded that there were no statistically significant differences between outputs from the accelerometers, suggesting that researchers can select any of the four ActiGraph accelerometers for measuring PA \[[@B18]\]. The present study adds information, comparing the last version of GT1M with the last version of ActiGraph (GT3X), and confirms results previously published with other generations of ActiGraph accelerometers. Findings suggest that the two devices assess PA similarly and that data from the two devices are comparable in studies of PA patterns. The two devices were equivalent in identifying subjects meeting the 60 min of MVPA · day^--1^. A high correlation was reported between the GT1M ActiGraph accelerometer and oxygen consumption \[[@B6],[@B7]\]. Results from the present study suggest that the GT3X accelerometer is a valid instrument for measuring PA. Further studies are suggested for assessing the capacity of the device to measure PA, especially intra and inter instrument reliability. Although results of the study provide important information regarding the use of accelerometers, there are limitations to consider. One limitation relates to the number of accelerometers used in the study. Because of financial and practical constraints only one accelerometer of each model was used. Wearing multiple accelerometers simultaneously is possible but would be difficult for the subject and could influence PA in free living conditions. Perhaps a complementary study using mechanical set-up, e.g., a motion table where several accelerometers are assessed together, would be helpful in controlling for confounding effects of monitor placement and type of activity. Results from the present study, however, show good reliability and a concordance value of 0.99 at each intensity, sedentary, light, moderate and vigorous. A possible second limitation is the time period used to monitor activity. Perhaps the difference between the two accelerometers would be greater if data were collected for a longer period of time. Finally, the thresholds of Freedson et al, were chosen for the present study because of frequency of use found in the literature. We cannot exclude the possibility that PA would have been different had we used other thresholds. In summary, our findings suggest that the GT3X accelerometer in mode GT1M may be used in clinical and epidemiological investigations without additional calibration or validation studies. Moreover, studies using the new generation of accelerometer can be compared to those using the GT1M. Competing interests =================== The authors declare that they have no competing interests. Authors' contributions ====================== JV, PF, OD and TB carried out study and drafted the manuscript. JM, GBX and LB participated in the design of the study, and contributed with critical review of the manuscript. All authors read and approved the final manuscript. Acknowledgements ================ The authors thank the participants for taking part in the study, and the Laboratory of Human Motricity Studies, University of Lille, Ronchin, France" for the lending of ActiGraph accelerometers.

Background ========== The 46, XX disorder of sex development (DSD) is a rare form of sex reversal in infertile men, that was first described by la Chapelle et al. in 1964 and occurred 1:20000 in newborn subjects \[[@B1]\]. By 1996, 150 patients with classical XX male syndrome had been reported, and more than 100 cases of this disorder have been discribed between 1996 and 2006 worldwide \[[@B2]\]. Clinical phenotypes about 46, XX DSD have been identified to three groups, including males with normal phenotype, males with genital ambiguities and males with true hermaphrodites \[[@B3]\]. Ovotesticular DSD, which is characterized by the presence of both testicular and ovarian tissue in the gonads of the same individual, and testicular DSD characterized by a full development of both gonads as testes without any evidence of ovarian tissue \[[@B4]\]. Approximately 80% of individuals with 46,XX testicular DSD present after puberty with normal pubic hair and normal penile size, but small testes, and sterility resulting from azoospermia \[[@B5]\]. The sex-determining region Y gene (*SRY*) locating in Y chromosome, plays a major role in encoding a testis determining factor (TDF) \[[@B6],[@B7]\]. About 90% of these patients have Y chromosomal material including the *SRY* gene, that are usually translocated to the distal tip of the short arm of X chromosome or autosomal chromosomes. About 10% 46, XX males are negative for *SRY* gene, which could carry different degrees of masculinization \[[@B8],[@B9]\]. There are several pathogenic mechanisms explaining 46, XX testicular DSD patients: 1. translocation of Y sequences, including the *SRY* gene, to an X chromosome or to an autosome; 2. a mutation in a gene in the testis-determining pathway triggering testis differentiation in *SRY* negative XX males; and 3. a hidden Y chromosome mosaicism limited to the gonad \[[@B10]\]. This study aimed to describing five 46, XX male DSD with *SRY*-positive, investigating the clinical characteristics and their relationships with chromosomal karyotype and the *SRY* gene. Methods ======= Participant and Clinical data ----------------------------- We collected 5 untreated patients with *SRY*-positive 46, XX, that were referred for infertility. The physical examination included the measurement of height, potential gynecomastia and the inspection of external sex organs. Bilateral volume was calculated as the sum of the volume of both testes. According to guidelines of the World Health Organization, semen analysis was indicated to azoospermia after centrifugation of the ejaculate. Serum levels of follicle-stimulating hormone (FSH), Luteinizing Hormone (LH), Estradiol, Prolactin, testosterone and free testosterone were assessed. All procedures used in the study confirmed to the tenets of the Declaration of Helsinki. The Ethics Committee of Jinling Hospital approved the protocols used. All participants have known to participate in the study. Written informed consents were obtained from all participants. Karyotype analysis of G-banding in lymphocytes and Fluorescence in situ hybridization (FISH) -------------------------------------------------------------------------------------------- Karyotypes were performed on peripheral blood lymphocytes in five patients respectively including 100 metaphase cells by conventional operating techniques. X chromosome, Y chromosome and *SRY* gene was located using FISH with probes of X chromosome centromere, Y chromosome centromere (CEP X with Spectrum Green, CEP Y with Spectrum Orange, Vysis, Downers Grove, IL; item no.32-111051) and *SRY* gene (*SRY* with Orange,Vysis, Downers Grove, IL; item no.30-190079). Molecular analysis ------------------ Genomic DNA from peripheral blood of the patients using QIAamp DNA Blood Kits was extracted. The three discrete regions, AZFa, AZFb and AZFc, located on the long arm of the Y chromosome, were performed by multiplex PCRs(Polymerase Chain Reaction) amplification. The set of PCR primers for the diagnosis of microdeletion of the AZFa, AZFb and AZFc region included: sY84, sY86, sY127, sY134, sY254, sY255, SRY and ZFX/ZFY. Results ======= Our five patients had a lower body height. Physical examination revealed that their testes were small in volume, soft in texture and normal penis. No potential gynecomastia and congenital hypospadias were seen. And they all described that they had normal sexual function. Semen analyses showed azoospermia. Endocrinological data indicated that the patients had a higher FSH, LH level, lower free testosterone, testosterone level and normal Estradiol, Prolactin level. General characteristics and endocrine hormone levels are shown in Table [1](#T1){ref-type="table"}. ###### General characteristics and endocrine hormone levels **Cases** **Body height (cm)** **Age at presentation, development of secondary sex (year)** **Volume of teste (ml)** **Stretched penile length (cm)** **Testosterone (nmol/L)** **Free testosterone (pmol/L)** **FSH (IU/L)** **LH (IU/L)** **Estradiol (pmol/L)** **Prolactin (mIU/L)** ------------------- ---------------------- -------------------------------------------------------------- -------------------------- ---------------------------------- --------------------------- -------------------------------- ---------------- --------------- ------------------------ ----------------------- **1** 165 12 6 9 6.8 27.7 35.5 13.8 112 201 **2** 162 15 3 8 5.4 15.2 29.2 12.9 70 158 **3** 164 14 4 8.5 8.9 29.4 45.9 25.1 98 78 **4** 167 11 9 11 8.4 28.1 33.7 22.3 107 232 **5** 165 12 7 10 7.0 20.5 31.4 19.6 81 167 **Normal ranges** ≥169 12-14 12-20 8-18 9.4-37.0 30.9-147.6 1-7 2-10 0-250 0-400 Karyotype analysis of all patients confirmed 46, XX karyotype, and FISH analysis showed that *SRY* gene were positive and translocated to Xp (Figure [1](#F1){ref-type="fig"}). Molecular analysis revealed that the *SRY* gene was present, and the AZFa, AZFb and AZFc region were absent (Figure [2](#F2){ref-type="fig"}). {#F1} {#F2} Discussion ========== 46, XX male syndrome is a rare sex reversal syndrome characterized by a female karyotype in discordance with a male phenotype. 90% of 46, XX testicular DSD usually have a normal male phenotypic heterogeneity at birth and are diagnosed after puberty on genital ambiguities, or infertility \[[@B8]\]. Our research reported that five patients had a female karyotype but were phenotypically male (46, XX males). They had normal external genitalia and masculinization, but showed azoospermia. That might be that all males were *SRY*-positive, which translocated on the short arm of X chromosome, and absent of the spermatogenic factors encoding gene on Yq, such as AZFa, AZFb and AZFc region in Y chromosome. *SRY* gene is located in the Y chromosome and encodes a high mobility group(HMG) domain, a conserved motif present in many DNA-binding proteins, which could regulate testicular differentiation \[[@B11],[@B12]\]. SRY protein is expressed in the genital ridge before testis formation, and in the testis during the period of testicular formation early in fetal life, until the development of adult testis \[[@B13]\]. Molecular genetics analysis demonstrated that most 46, XX testicular DSD patients carry *SRY* gene which translocated to X chromosome \[[@B14]-[@B16]\]. There was a report that an *SRY* gene fragment translocated from Y chromosome to autosomal chromosome \[[@B17]\]. Some patients showed *SRY* negative, who always had external genital ambiguities and gynecomastia. Despite the fact that *SRY* gene is considered to be the main regulatory factor for testis determination, phenotypic variability showed in 46, XX sex reversed cases cannot be explained only by whether *SRY* gene is presented. And a number of other genes such as SOX9, DAX-1, WT1, WNT4, FGF9 and RSPO1 have been involved in the process of gonadal differentiation \[[@B8]\]. The phenotype of the XX male observed in *SRY* positive 46,XX individuals varies greatly, from normal internal and external male gonads to abnormal secondary sexual characteristics, small testes and hypospadias, to a true hermaphrodite. It has been suggested that the variation in phenotype is primarily dependent on two mechanisms: X chromosome inactivation(XCI) pattern and the amount of Y material including *SRY* gene that has been translocated to the X chromosome \[[@B18]\]. Reviewing the literature, 46,XX males with true hermaphrodites or gonadal ambiguity have a small portion of the Y chromosome material translocated to the X, presumably allowing for XCI spreading and inactivating the *SRY* gene \[[@B19]\]. A normal male phenotype is expected to result from a larger Yp *SRY* bearing fragment being translocated to the X chromosome, where the length of the Yp fragment may protect the *SRY* gene from silencing by the spread of XCI \[[@B19]\]. In our study, all five cases have normal external genitalia and masculinization, which is expected that more Y chromosome material is present on the X, presumably protecting the *SRY* gene from the spread of inactivation. Because of the unavailable in specimens from the five cases to further study, we cannot do more molecular analysis to confirm the above point. Till now, both random and non-random XCI patterns have been reported in 46, XX males with a normal male phenotype \[[@B18],[@B20]\]. It is indicated that the XCI pattern may be not associated with the XX male phenotype. However, another mechaniam, known as the position effect, has been reported to explain the observed phenotypic differences. The phenotypic differences are dependent on the proximity of the breakpoint to the *SRY* gene as well as the presence or absence of cryptic rearrangements affecting the expression of the *SRY* gene \[[@B21]\]. The rearrangements, may result in transcriptional repression, probably by removing essential regulatory elements or alterations of local chromatin structure \[[@B22]\]. Additionly, phenotypic variability might be associated with variations in genetic polymorphisms and copy number variation of specific genes on the X chromosome, such as NROB1 and TAF7L \[[@B23],[@B24]\]. Classical 46, XX male have normal testosterone level and free testosterone level during adolescence, but may decrease in adulthood, leading to hypergonadotropic hypogonadism \[[@B25]\]. Our cases had normal genitalia and were diagnosed for infertility after puberty. The level of testosterone and free testosterone is deficiency in five patients. In addition, high levels of FSH and LH are observed. This might explain that even though the 46, XX male have a normal external genitalia and masculinization, and they are lack of spermatogenesis. The body heights of the patients we reported were all under 169 cm (the average height of Chinese male) and close to that of normal females. There are some phenotypic similarities between 46,XX men and those with Klinefelter syndrome, but 46,XX men tend to be shorter than men with KS \[[@B9]\]. Kirsch *et al.* indicated that the Y chromosome growth-control gene(*GCY*) which next to the centromere had a possible impact on growth \[[@B26]\]. And there were some papers indicating that *SHOX* gene(short stature homeobox) expression and *SHOX* enhancer regions played a role in the growth \[[@B27]\]. It has been suggested that specific growth genes in the Y chromosome cannot switched to the patients, which might make them to show a female stature. GH (growth hormone) therapy may have some statural effects in the SHOX haploinsufficiency and may be insufficient to prevent the development of skeletal lesions after puberty \[[@B28]\]. Conclusions =========== Our reports adds cases on the five new 46, XX male individuals with sex reversal and further verifies the view that the presence of *SRY* gene and the absence of major regions in Y chromosome should lead to the expectance of a completely masculinised phenotype, abnormal hormone levels and infertility. Abbreviations ============= DSD: Disorder of sex development; *SRY*: Sex-determining region Y gene; TDF: testis determining factor; FISH: Fluorescence in situ hybridization; FSH: Follicle-stimulating hormone; LH: Luteinizing Hormone; PCR: Polymerase Chain Reaction; HMG: High mobility group; *SOX9*: SRY (sex determining region Y)-box 9 gene; *WT1*: Wilms tumor 1 gene, *WNT4*, wingless-type MMTV integration site family, member 4 gene, *FGF9*, fibroblast growth factor 9 gene; *RSPO1*: R-spondin 1 gene; *GCY*: Growth-control gene; *SHOX*: Short stature homeobox; GH: Growth hormone; XCI: X chromosome inactivation; KS: Klinefelter syndrome; *NROB1*(*DAX-1*): Nuclear receptor subfamily 0, group B, member 1 gene; *TAF7L*: TAF7-like RNA polymerase II gene. Competing interests =================== The authors declare that they have no competing interests. Authors' contributions ====================== QW carried out the molecular genetic studies and drafted the manuscript. NL, WL, TL and CZ participated in the laboratory work. YC, JZ and XX conceived of the study, and participated in its design and coordination and helped to draft the manuscript. All authors read and approved the final manuscript. Pre-publication history ======================= The pre-publication history for this paper can be accessed here: <http://www.biomedcentral.com/1471-2490/14/70/prepub> Acknowledgements ================ This work was supported by the Natural Science Foundation of Jiangsu province (BK2011660), Key Foundation of Jiangsu Science and Technology Bureau (BM2013058) and the Natural Science Foundation of China (30901652).We thank all members of the family for their cooperation in the study.

Refer to the page 551-557 An excess number of low density lipoprotein-cholesterol (LDL-C) particles is an independent risk for a cardiovascular event in the future. LDL particles diffused into the developing atherosclerotic plaques bind to proteoglycans in the intima, become oxidized and phagocytosed to macrophages. This causes the generation of foam cells, which aggravate the generation of fatty streaks and induce platelet attachment.[@B1] It is highly probable that platelets may have contact with oxidized LDL (oxLDL).[@B2][@B3] Several results describe that oxLDL can be exposed to circulating blood or can even be formed in plasma at sites of oxidative stress. Platelets express a series of receptors that specifically recognize oxLDL, such as CD36 and scavenger receptor A (SRA). oxLDL-bound CD36 and SRA induce phosphorylation of p38 MAP kinase (MAPK), which in turn triggers the production of thromboxane A2 through the activation of a cytoplasmic type of phospholipase A2.[@B4] Thus, the stimulation with oxLDL enables platelets to induce aggregation independently or with other stimulatory signals such as adenosine diphosphate (ADP), collagen or thrombin in a synergistic manner. Unlike oxLDL, native and non-oxLDL particles are functionally inert and little has been described regarding the activity of native LDL on any type of cells that may contribute to the development of atheroma. Interestingly, platelets express the variant form of ApoB/E receptor (ApoB/E-R) and it contains specific amino acid sequences, which recognize and bind to apoB100 apoproteins. Platelet ApoE-R2\' messenger ribonucleic acid encodes a 130 kDa protein with a single ligand-binding domain that comprises four complement repeats (eight in ApoE-R2 complementary deoxyribonucleic acid), an epidermal growth factor precursor homology domain with an YWTB domain for ligand disscociation, a β-propellor, an O-linked sugar domain, and a single transmembrane domain connected to a short cytoplasmic tail. LDL particles can activate the Thr/Tyr protein kinase p38MAPK in platelets, which is expected to trigger signals for the platelet aggregation, too.[@B5] Recent result showed evidence that LDL particles may prime the aggregating activity of platelets in response to ADP and other agents, but the potency remains uncertain.[@B6] An article by Kang et al.[@B7] published here described that there was no significant association between LDL particle size and on-treatment platelet reactivity (OPR) in patients undergoing percutaneous coronary intervention. All subjects had dual anti-platelet therapy with aspirin and clopidogrel and they measured the magnitude of platelet agglutination induced by either arachidonic acid in the aspirin assay and ADP, and prostagladin E1 in the P2Y12 assay. Platelet reactivity was reported as aspirin reaction units (ARU) and P2Y12 reaction units (PRU). A higher reaction unit, which reflected higher OPR was defined as an OPR greater than 454 ARU or 264 PRU, which could be regarded as resistant to antiplatelet agents. There are several possibilities to show the negative relationship between platelet activity and the characteristics of LDL particles. First, the dominance of small-dense LDL-C suggests a greater number of LDL particles at the same level of LDL-C, which may prime circulating platelets to show an exaggerated response upon the secondary stimulation with ADP etc. However, LDL-C levels for all subjects were very low (\<100 mg/dL), which may minimize the functional relevance of LDL particle size on the platelet activities. Second, small-dense LDL particles are relatively labile and vulnerable to oxidation. Although a group of subjects had small-dense LDL dominancy, most isolated platelets from the circulation may not have a chance to be exposed to the modified form of small-dense LDL. Third, the small-dense LDL-dominant phenotype may reflect a population group that has a higher incidence of cardiovascular disease risks, suggesting that small-sense LDL is a marker as well as a direct mediator. Unfortunately, the subjects analyzed in this study did not show any significant difference of major clinical parameters between type A and B phenotypes and only one exception was the lower HDL-cholesterol levels in type B. It is notable that the medication with statin, for example, was not controlled in this study, which may have been medicated to most subjects. Such intensive lipid modification may diminish the deference of A and B phenotypes. Several reports describe the phenotypic change of A and B types after the statin medication especially atorvastatin.[@B8] During the process of platelet aggregation, other signals also play an important role, too. For instance, lysophosphatidic acid (LPA) generated by lipid oxidation activates the Rho-A pathway in platelets through LPA-receptor activation. Guanosine triphosphage (GTP)-protein G13α and Rho kinase are known to be involved in the process and the resulted phosphorylation of myosin light chain (LC) drives the disc to sphere transition known as the shape change of platelets, which is one of the earliest responses induced by platelet agonists. Moreover, LPA at a higher dose activates the GTP-protein Gq inducing Ca^2+^ mobilization and mechanisms that control Ca^2+^ entry. The rise in Ca^2+^ activates calmodulin-mediated phosphorylation of myosin LC facilitating actin binding. The capacity to increase cytosolic Ca^2+^ makes mildly oxLDL an independent activator inducing aggregation and secretion in the stirred suspensions of platelets. [@B9] Therefore, intensive medical treatment with statins as well as with aspirin and clopidogrel may substantially suppress the activity of platelets at various levels and may precipitate the study result to show an insignificant cause-and-effect relationship in general. Further studies about atherogenic and thrombogenic potential are anticipated in order to evaluate the impact of small-dense LDL particles on platelet activities. The author has no financial conflicts of interest.

Introduction {#s1} ============ During the development of most animal species studied, *Hox* transcription factors specify positional information along the anterior - posterior axis [@pone.0018010-Brglin1]--[@pone.0018010-McGinnis1]. *Hox* genes comprise a subfamily of the homeobox containing gene family, and are organised in four clusters, each located on a different chromosome. The homeobox encodes a DNA binding motif called the homeodomain. A strict control of the expression and function of these *Hox* genes is essential. It has been shown that Pbx family members, and their *Drosophila melanogaster* counterpart *Extradenticle* (Exd), function as cofactors for Hox proteins; they can enhance their binding specificity and affinity for specific target sequences in DNA [@pone.0018010-Knoepfler1]--[@pone.0018010-vanDijk1]. Pbx/Exd family members are part of a particular subfamily of the homeodomain containing proteins, namely the TALE-class. This class is characterised by having a [t]{.ul}hree [a]{.ul}mino acid [l]{.ul}oop [e]{.ul}xtension between the first and second helices of the homeodomain [@pone.0018010-Brglin3]. It has been proposed that cooperative binding of Hox and Pbx/Exd proteins can lead to transactivation while binding of the individual factors leads to repression on the same promoter elements [@pone.0018010-Pinsonneault1]. When Hox proteins bind to DNA cooperatively with a Pbx/Exd family member, the main protein-protein interaction consists of binding of the hexapeptide motif of the Hox protein to a pocket formed by the atypical homeodomain of PBC family members [@pone.0018010-Jabet1]--[@pone.0018010-Piper1] This pocket is composed of the three amino acid loop extension of the PBC homeodomain, residues in the third helix of the homeodomain, and a residue in the C-terminal helix of PBC homeodomains [@pone.0018010-Piper1]. The nuclear localisation of Pbx/Exd proteins is controlled by competing nuclear import and export signals [@pone.0018010-AbuShaar1]. When members of the Meis family, or their *Drosophila* counterpart *Homothorax* (Hth), also members of the TALE-class of homeodomain proteins, are present in the cytoplasm they can interact with Pbx/Exd family members in such a way that the nuclear export signal of the Pbx/Exd family member is shielded, resulting in a net influx of Pbx/Exd into the nucleus, sometimes influencing the function of Hox proteins present [@pone.0018010-Ryoo1], [@pone.0018010-Ryoo2], [@pone.0018010-Jaw1]. However, Pbx/Exd and Meis/Hth proteins are not used exclusively as cofactors for Hox proteins. The myogenic bHLH factors [@pone.0018010-Berkes1] and Engrailed [@pone.0018010-Peltenburg1] also depend on the activity of Pbx and Meis members for proper functioning. For Hox paralog group 1 members, autoregulation dependent on Pbx/Exd and Meis/Hth has been shown in the neurectoderm of mouse embryos [@pone.0018010-Ppperl1], [@pone.0018010-Ferretti1], in *C. elegans* [@pone.0018010-Streit1] and in endoderm of *Drosophila* embryos [@pone.0018010-Ryoo2], [@pone.0018010-Marty1]. Binding of Hox and Pbx family members to bipartite Hox-Pbx binding sites is essential for autoregulation [@pone.0018010-Ryoo2], [@pone.0018010-Ppperl1], [@pone.0018010-Marty1], [@pone.0018010-Grieder1]. Meis proteins have been shown to be indispensable as mediators of this process [@pone.0018010-Ryoo2], [@pone.0018010-Marty1], [@pone.0018010-Grieder1]. In *Xenopus*, a member of the Meis family, *XMeis3*, is a posteriorising factor in the neurectoderm of *Xenopus laevis*, and is required for hindbrain patterning [@pone.0018010-Salzberg1], [@pone.0018010-Dibner1]. Recent findings also show that neurectodermal *XMeis3* mediates the posteriorising action of *XWnt3A* in the developing CNS [@pone.0018010-Elkouby1]. In zebrafish embryos, similar functions have been reported for Meis3 and other Meis family members [@pone.0018010-Vlachakis1]--[@pone.0018010-Choe1]. Expression of *XMeis3* is reported as being initiated in a stripe in the neural plate of early-mid neurula embryos. During neurula and early-tailbud stages, expression is mainly localised to rhombomeres (r\'s) 2, 3, and 4, and the anterior spinal cord, while posterior rhombomeres show some ventral expression [@pone.0018010-Salzberg1]. Expression of *XMeis3* overlaps with neurectodermal expression of *Hoxd1* (r4 and r5) [@pone.0018010-Kolm1], *Hoxb4* (r7, r8, and the anterior spinal cord) [@pone.0018010-Harvey1], and *Hoxc6* (anterior spinal cord) [@pone.0018010-Oliver1], [@pone.0018010-DeRobertis1]. These overlaps are consistent with the idea that *XMeis3* is involved in controlling the function of the Hox proteins with which it is co-expressed. These studies do, however, leave many questions unanswered. They pay little attention to when and where Meis cofactors actually interact with Hox proteins at different stages during the early AP patterning process. These details are likely to be crucial for understanding the mechanism at hand. Studies of vertebrate *Hox* expression and function have already delivered strong evidence that AP patterning depends on a specific early spatiotemporal sequence of *Hox* gene expression. Expression of each Hox gene is initiated in a specific mesodermal domain in the gastrula embryo and then undergoes an establishment phase during which this expression domain changes to a gene specific AP zone in axial mesoderm and the neural plate and finally a maintenance phase during which this AP zone is consolidated. This sequence is employed universally in mammals, birds, fish and amphibians and shows generic features in these different species [@pone.0018010-Duboule1]--[@pone.0018010-Wacker1]. A recent study analysed the early *Hox* expression patterns in *Xenopus*, and this revealed temporally colinear initiation of expression of a sequence of *Hox* genes within a horseshoe-shaped domain of ventrolateral marginal zone mesoderm with the tips of the horseshoe facing dorsal at different stages during gastrulation and then sequential dorsalisation of each *Hox* expression zone corresponding with its translation into a stable AP pattern zone in axial mesoderm and the neural plate [@pone.0018010-Wacker1], [@pone.0018010-Durston1]. This sequence reflects timed interactions between an early ventrolateral mesodermal *Hox* cascade and the Spemann organiser that are probably imperative for AP axis formation. We set out to investigate whether early expression of *Hox* genes depends on the activity of XMeis3 and whether XMeis3 is involved in regulation of expression of these *Hox* genes in mesoderm during gastrulation. In order for *XMeis3* to be able to regulate *Hox* expression in mesoderm, it and *Hox* genes need to be co-expressed there. We performed whole mount *in situ* hybridisation to study the detailed early expression of *XMeis3* and compared it to the early expression patterns of *Hoxd1*, *Hoxb4*, and *Hoxc6* and found significant co-expression in lateral regions of marginal zone mesoderm, early during gastrulation. This is the first time that Xmeis3 expression has been reported in gastrula mesoderm. To gain further insight into the early functions of *XMeis3*, we followed a gain- and a loss-of-function strategy. In the gain-of-function strategy synthetic *XMeis3* mRNA was microinjected into early blastomeres and expression of *Hox* genes was studied. These experiments showed that ectopic expression of *XMeis3* during gastrulation is capable of inducing expression of the *Hox* genes assayed in mesoderm as well as in ectoderm. In the loss-of-function strategy we made use of an antisense morpholino oligonucleotide (reviewed in [@pone.0018010-Heasman1] and references therein) to inhibit the translation of *XMeis3* mRNA (MO*^XMeis3^*). Injection of MO*^XMeis3^* leads to a reduction in expression of *Hoxd1*, *Hoxb4*, and *Hoxc6* in mesoderm and ectoderm during gastrulation, and to severe patterning defects. Finally we show synergy between Hoxd1 and XMeis3 and show that the mesodermal expression of *Hoxd1* during early gastrulation is already dependent on XMeis3 mediated autoregulation. Results {#s2} ======= The expression of *XMeis3* overlaps with *Hox* gene expression in mesoderm {#s2a} -------------------------------------------------------------------------- To determine whether *XMeis3* is co-expressed with *Hox* genes in the mesoderm of gastrula embryos, whole mount *in situ* hybridisation was performed for *XMeis3*, *Hoxd1*, *Hoxb4*, and *Hoxc6* ([Fig. 1](#pone-0018010-g001){ref-type="fig"}). Expression of *XMeis3* is initiated in a horseshoe-shaped domain in ventrolateral marginal zone mesoderm of the early gastrula (st. 10.5) (the tips of the horseshoe face dorsal). By stage 11, expression is lost in the ventralmost tissue, resulting in two lateral expression domains, one on either side of the organiser in mesoderm of early gastrula stage embryos ([Fig. 1A](#pone-0018010-g001){ref-type="fig"}). Expression thus becomes localised to mesoderm lateral to the midline and to a very low extent also possibly to the overlying ectoderm ([Fig. 1A](#pone-0018010-g001){ref-type="fig"}). Expression later, at the beginning of neurulation (st.13) is primarily in neurectoderm, as has been reported previously [@pone.0018010-Godsave1] but there is also remaining expression in dorsolateral mesoderm ([Fig. 1B](#pone-0018010-g001){ref-type="fig"}). Early expression of *Hoxd1*, *Hoxb4*, and *Hoxc6* is initiated in ventrolateral mesoderm and each of these genes follows a similar spatiotemporal expression sequence but with specific timing [@pone.0018010-Wacker1]. During early phases of gastrulation mesodermal expression of *Hoxd1* ([Fig. 1C](#pone-0018010-g001){ref-type="fig"}), *Hoxb4* ([Fig. 1E](#pone-0018010-g001){ref-type="fig"}), and *Hoxc6* ([Fig. 1G](#pone-0018010-g001){ref-type="fig"}) overlaps with expression of *XMeis3* in the dorsolateral domains of these *Hox* genes (compare [Fig. 1A to 1C, 1E, and 1G](#pone-0018010-g001){ref-type="fig"}). At the end of gastrulation, the overlap between mesodermal expression of *Hoxd1* ([Fig. 1D](#pone-0018010-g001){ref-type="fig"}) and *XMeis3* ([Fig. 1B](#pone-0018010-g001){ref-type="fig"}) in mesoderm is maintained, and the newly initiated expression of both genes in the neurectoderm also overlaps. At the same time, the more posteriorly expressed *Hoxb4* ([Fig. 1F](#pone-0018010-g001){ref-type="fig"}) and *Hoxc6* ([Fig. 1H](#pone-0018010-g001){ref-type="fig"}) only partially overlap *XMeis3* expression ([Fig. 1B](#pone-0018010-g001){ref-type="fig"}) in involuted mesoderm. *Hoxb4* expression also partially overlaps expression of *XMeis3* in overlying ectoderm (compare [Fig. 1F to 1B](#pone-0018010-g001){ref-type="fig"}). These results show that there is indeed an overlap in expression of *XMeis3* and of early *Hox* genes in mesoderm during gastrulation, and that expression of *XMeis3* also overlaps with *Hoxd1*, and to some degree *Hoxb4*, in neurectoderm. {ref-type="fig"}, as the gap in the Hox or Meis expression domain, facing up in the left hand panels . Embryos shown are at stage 11, vegetal views with dorsal up (**A**, **C**, **E**, and **G**) and at stage 13, dorsal views with anterior up (**B**, **D**, **F**, and **H**). *XMeis3* expression overlaps with dorsolateral expression of *Hoxd1*, *Hoxb4*, and *Hoxc6* in mesoderm at stage 11 (**A**, **C**, **E**, and **G**). *XMeis3* expression in ectoderm at stage 13 overlaps with expression of *Hoxd1* but not with expression of *Hoxb4* and *Hoxc6* (**B**, **D**, **F**, and **H**). At stage 11, Hox and Meis expression is limited by a sharp boundary, running parallel to the outside of the embryo. This boundary is Brachy\'s cleft, the boundary between involuted mesoderm and external ectoderm Brachy\'s cleft runs from the blastopore to the upper limit of the involuted mesoderm (and is actually visible as a cleft in the upper part of the right panel of [Fig 1C](#pone-0018010-g001){ref-type="fig"}). All early Hox expression is known to be inside this cleft at this stage (mesodermal, not ectodermal) and thus marks the position of the cleft. The early XMeis3 expression shows the same pattern. It is mesodermal. At a later stage (st.13, [Fig 1B](#pone-0018010-g001){ref-type="fig"}), [@pone.0018010-Wacker1] XMeis 3 expression is also outside Brachy\'s cleft (ectodermal).](pone.0018010.g001){#pone-0018010-g001} *XMeis3* gain-of-function upregulates *Hox* gene expression in mesoderm and ectoderm {#s2b} ------------------------------------------------------------------------------------ To investigate whether *XMeis3* is capable of contributing to the regulation of *Hox* gene expression, 2 ng of synthetic mRNA containing the full-length coding region of *XMeis3* was injected into the animal pole of embryos at the one-cell stage. The amount of 2 ng was chosen because this was shown to lead to posteriorisation of injected embryos [@pone.0018010-Salzberg1]. The effects on expression of *Hoxd1*, *Hoxb4*, *Hoxc6*, *Xbra*, and the posterior marker *Xcad3* in gastrula stages were assayed by *in situ* hybridisation ([Fig. 2](#pone-0018010-g002){ref-type="fig"}). The ectopic expression of *Hoxd1* ([Fig. 2A](#pone-0018010-g002){ref-type="fig"}) in injected embryos is remarkable because it is found in the region harbouring the Spemann organiser, tissue that normally does not express *Hox* genes. The horseshoe-shaped domain of expression is also expanded and expression levels appear to be enhanced. Furthermore expression can be found in ectoderm of the animal cap and in the mesoderm underlying it, in the form of a streak of expression in contact with the expanded ring of expression around the blastopore ([Fig 2A](#pone-0018010-g002){ref-type="fig"}). *Hoxb4* also shows ectopic expression in animal cap ectoderm and expansion of the endogenous expression domain ([Fig. 2B](#pone-0018010-g002){ref-type="fig"}), but no closure of the dorsal expression gap neither in organiser mesoderm nor in overlying ectoderm can be observed. Interestingly, induced expression of *Hoxc6* can already be found in dorsal mesoderm at stage 10.25 ([Fig. 2C](#pone-0018010-g002){ref-type="fig"}), significantly earlier than its endogenous initiation of expression (st11) and like ectopic *Hoxd1* expression, this occurs in dorsal mesoderm. In later stages an expansion of the endogenous horseshoe-shaped expression domain is also found (data not shown). Expression of the mesodermal marker *Xbra* appears unaltered in injected embryos ([Fig. 2D](#pone-0018010-g002){ref-type="fig"}), suggesting that changes in *Hox* expression domains are not due to changes in induction of mesoderm, but rather to its patterning. The previously described posteriorising effect of *XMeis3* on neurectoderm is confirmed by anterior expansion of expression of the posterior marker *Xcad3* ([Fig. 2E](#pone-0018010-g002){ref-type="fig"}). {ref-type="fig"}.](pone.0018010.g002){#pone-0018010-g002} *XMeis3* loss-of-function downregulates expression of *Hox* genes and arrests gastrulation {#s2c} ------------------------------------------------------------------------------------------ To determine whether *XMeis3* function is necessary for initiation and/or establishment of *Hoxd1*, *Hoxb4*, and *Hoxc6* expression, an antisense morpholino oligonucleotide directed against *XMeis3* mRNA (MO*^XMeis3^*) was injected into the animal hemisphere of embryos at the one-cell stage. *XMeis3* loss-of-function leads to a loss of trunk structures and defects in axis specification, in a concentration dependent manner. When 12 ng MO*^XMeis3^* was injected a loss of trunk structures and defects in head development and tail formation can be observed, while the anteriormost structure, the cement gland, remains present ([Fig. 3B](#pone-0018010-g003){ref-type="fig"}). When 24 ng MO*^XMeis3^* was injected, an enlargement of the cement gland was visible accompanied by a stronger loss of trunk structures ([Fig. 3C](#pone-0018010-g003){ref-type="fig"}) In half the injected embryos *spina bifida\'s* are observed, suggesting that the embryos suffer from gastrulation problems. When 32 ng or more MO*^XMeis3^* were injected, the embryos arrested during gastrulation at stage 11 ([Fig. 3D](#pone-0018010-g003){ref-type="fig"}). Embryos injected with this high dose of MO*^XMeis3^* appear unaffected and posses normal looking blastopores until the moment of arrest. This is unlike what would be expected if the arrest was caused by toxicity of an injected agent, this would generally generate a much larger spread in stages at which embryos die or arrest, accompanied by irregular formation of the blastopore. Removal of the vitelline membrane revealed that cells have lost cell-cell contact, but appear round and intact (not shown). This suggests that the observed effect is the result of a strong knockdown of *XMeis3* function and not an aspecific effect of MO*^XMeis3^*. Injection of the same amount of a control morpholino (MO^contr^), in sequence unrelated to MO*^XMeis3^*, has no outward effects on embryos (data not shown). These findings support the idea that the gastrulation arrest phenotype is a true result of XMeis3 loss-of-function and that XMeis3 is required for patterning (a part of) the primary axis in *Xenopus* embryos. Actually, this result is perhaps not so surprising because: a recent result shows that EMT timing during internalisation of mesoderm into the gastrula is regulated (delayed) by hox genes [@pone.0018010-Iimura1] and because we present evidence (below) that the important function of Meis3 in the gastrula is to mediate mesodermal autoregulation of *Hox* genes. {ref-type="fig"}) and then disintegrates to a mass of dissociated cells contained within the vitelline membrane (not shown). The specificity of MO*XMeis3* is shown by the rescue with *XMeis3* synthetic mRNA. Embryos were injected with 32 ng of MO*XMeis3* and 125 pg synthetic mRNA for *XMeis3* and allowed to develop until the control embryos reached the tad pole stage (**E**), In the majority of the embryos a large part of the axis was rescued (**F**), in a small number of embryos the phenotype could even be reversed, not only is the axis fully rescued but the embryo shown in (**G**) even possesses additional trunk structures as was revealed by the presence of somites in the axis outgrowth (not shown). The most extreme MO treatment thus produced a gastrulation block. Other treated embryos were allowed to develop to comparable stages (¬ 40--45) as shown by development of stage specific structures, for example the cement gland (seen best in [Figs 3A, B, C, F G](#pone-0018010-g003){ref-type="fig"} as the black spot at the lower front end of each embryo. Front ends are left in 3A, B, E, F, G. Various directions in 3C.](pone.0018010.g003){#pone-0018010-g003} To further test the specificity of the MO*^XMeis3^*, 125 pg of synthetic *XMeis3* mRNA, lacking most of the sequence that the MO*^XMeis3^* is complementary to, was co-injected with 32 ng MO*^XMeis3^* into the animal hemisphere of embryos at the one-cell stage ([Fig. 3F](#pone-0018010-g003){ref-type="fig"}). The exogenous *XMeis3* was able to largely rescue the MO*^XMeis3^* phenotype (compare [Fig. 3D to 3E, and 3F](#pone-0018010-g003){ref-type="fig"}). In a small number of the co-injected embryos a full recovery of the axis can be observed, sometimes accompanied by a secondary axial outgrowth out of the primary axis, containing somites ([Fig. 3G](#pone-0018010-g003){ref-type="fig"}). The effect of XMeis3 loss-of-function on *Hox* expression was studied by injecting 16 ng MO*^XMeis3^* into the animal hemisphere of embryos at the one-cell stage followed by *in situ* hybridisation at appropriate stages. To be able to analyse marker expression in late gastrula stage embryos, the arrest in gastrulation, observed after injection of a high amount of MO*^XMeis3^*, was avoided, by the injection of 16 ng. The XMeis3 loss-of-function leads to downregulation of expression of *Hoxd1* ([Fig. 4A](#pone-0018010-g004){ref-type="fig"}), *Hoxb4* ([Fig. 4B](#pone-0018010-g004){ref-type="fig"}), and *Hoxc6* ([Fig. 4C](#pone-0018010-g004){ref-type="fig"}), early in mesoderm and later in neurectoderm. This led to our conclusion that XMeis3 is necessary for *Hox* gene expression in marginal zone mesoderm, and neural plate ectoderm. {#pone-0018010-g004} Synergy between *Hoxd1* and *XMeis3* {#s2d} ------------------------------------ Autoregulation is known to occur among labial type *Hox* genes in murine hindbrain neurectoderm [@pone.0018010-Ppperl1], [@pone.0018010-Gould1], in endoderm of *Drosophila* embryos [@pone.0018010-Ryoo2], [@pone.0018010-Grieder1], and in *C. elegans* [@pone.0018010-Streit1] For a number of these cases it has been shown that this autoregulation is dependent on a Pbx/Hox bipartite binding site in the *Hox* promoters [@pone.0018010-Ryoo2], [@pone.0018010-Ppperl1], [@pone.0018010-Streit1], [@pone.0018010-Grieder1] Because nuclear localisation of Pbx family members is dependent on the action of Meis family members and because XMeis3 loss-of-function led to a significant downregulation of *Hoxd1* expression in mesoderm and ectoderm, we suspected that *XMeis3* might be involved in Hoxd1 autoregulation. To test our idea that XMeis3 may mediate autoregulation of labial type *Hox* genes in *Xenopus* development, we co-injected relatively small amounts of synthetic mRNA for *XMeis3* and *Hoxd1* and also injected them separately using double the amount of mRNA. Small amounts of mRNA were used to be able to observe compound phenotypes in co-injected embryos. If a strong effect was generated in embryos injected with only a single messenger this would not have been possible. The embryos injected with only a single synthetic messenger show little or no phenotypic effect, while co-injected embryos show a significant retardation in head development ([Fig. 5](#pone-0018010-g005){ref-type="fig"}). This points towards a synergistic relation between Hoxd1 and XMeis3. {#pone-0018010-g005} To further test this synergy, and to test whether XMeis3-mediated Hoxd1 autoregulation is involved in the establishment of *Hoxd1* expression, we wished to investigate the necessity of Hoxd1 for maintaining *Hoxd1* expression in mesoderm. If XMeis3 activity is needed in early gastrula mesoderm to enhance or alter the function of Hoxd1, then Hoxd1 loss-of-function should generate the same effect on *Hoxd1* expression as XMeis3 loss-of-function. To test whether this is the case, 32 ng MO*^Hoxd1^* [@pone.0018010-McNulty1] was injected into the equatorial region of the 2 blastomeres making up the presumptive left side of 4-cell stage embryos. The other half of the embryos served as an internal control. This results in a downregulation of expression of *Hoxd1* in mesoderm on the injected side ([Fig. 6A](#pone-0018010-g006){ref-type="fig"}). This finding extends our recent investigation of the effect of MO knockdown of labial Hox genes on neurectodermal Hox gene expression [@pone.0018010-McNulty1]. To further test whether establishment of expression of *Hoxd1* needs both Hoxd1 and XMeis3, sub optimal amounts of morpholinos against both messengers were co-injected and injected separately. Embryos were harvested at stage 11 and assayed for *Hoxd1* expression ([Fig 6B](#pone-0018010-g006){ref-type="fig"}). Sub optimal morpholino amounts were used to allow different levels of reduction in *Hoxd1* expression, thus allowing possible synergistic effects to be observed. A downregulation of *Hoxd1* expression in embryos injected with a single morpholino and a strong additional reduction by injection of both morpholinos is visible ([Fig. 6B](#pone-0018010-g006){ref-type="fig"}). This suggests that there is indeed a synergistic effect of Hoxd1 and XMeis3 on establishment of *Hoxd1* expression in marginal zone mesoderm during gastrulation. {#pone-0018010-g006} Discussion {#s3} ========== *XMeis3* expression overlaps early *Hox* expression {#s3a} --------------------------------------------------- Much effort has been put into finding out details about the relation between *Hox* proteins and their cofactors Pbx/Exd and Meis/Hth. Although much has been accomplished, many questions remain. In *Xenopus* embryos, it has been shown that *XMeis3* has a function in hindbrain patterning [@pone.0018010-Salzberg1]--[@pone.0018010-Elkouby1], these results are corroborated by recent reports concerned with Meis function in hindbrain formation in zebrafish embryos [@pone.0018010-Vlachakis1]--[@pone.0018010-Choe1]. We show here that *XMeis3* is expressed in marginal zone mesoderm significantly earlier than previously described [@pone.0018010-Salzberg1], [@pone.0018010-Elkouby1]. We went on to show that an overlap is found between expression of *XMeis3* and of early *Hox* genes in ventral and lateral and dorsolateral mesoderm during gastrulation. At st. 11, the overlap is restricted to dorsolateral mesoderm. This co-localisation with early *Hox* genes is compatible with a role for *XMeis3* in the regulation of *Hox* gene expression in mesoderm during the early phases of gastrulation. Ectopic *XMeis3* enhances *Hox* expression in mesoderm {#s3b} ------------------------------------------------------ By gain-of-function experiments we showed that ectopic *XMeis3* is capable of inducing expression of *Hoxd1*, *Hoxb4*, and *Hoxc6*, expanding the endogenous expression domains of these genes in early mesoderm, and ectopically initiating expression in dorsal mesoderm. Interestingly, this induction of *Hox* expression by ectopic *XMeis3* can only be found as expansions of the endogenous expression domains or in streaks of expression still in contact with the expanded endogenous domains of expression. This is most obvious for ectopic expression of *Hoxd1* in dorsal mesoderm, expanding into more animally located mesoderm and ectoderm. This suggests that ectopic *XMeis3* only enhances the expression of the assayed *Hox* genes, requiring factors already present in their endogenous *Hox* expression domains rather than inducing expression *de novo*. We suspect that the endogenous factors required are the *Hox* proteins themselves. These patterns are consistent with our idea (below) that *XMeis3* enhances *Hox* autoregulation in mesoderm of *Xenopus* embryos. *XMeis3* is necessary for *Hox* expression in mesoderm and ectoderm {#s3c} ------------------------------------------------------------------- The injection of MO*^XMeis3^* led to a downregulation of mesodermal expression of all three *Hox* genes assayed. For *Hoxd1* and *Hoxb4* this held true for mesoderm and ectoderm, in the case of *Hoxc6*, mesodermal expression partially recovers during later phases of gastrulation, but ectodermal expression could not be observed. This indicates that *XMeis3* protein is necessary, in ventral and lateral mesoderm and in neurectoderm during gastrulation, for proper establishment and maintenance of *Hox* expression. XMeis3 loss-of-function using small amounts of MO*^XMeis3^* already led to a strong phenotype, indicating the necessity of XMeis3 function in anteroposterior patterning. This phenotype corroborates the results of Dibner and co-workers [@pone.0018010-Dibner1]. The sudden arrest in gastrulation at stage 11, caused by injecting a high amount of MO*^XMeis3^* is very striking. We show by coinjecting a limited amount of *XMeis3* mRNA that the observed effect is not aspecific. We note that there is published evidence that Hox genes regulate cell movement and EMT\'s during gastrulation [@pone.0018010-Iimura1] and suspect that this *XMeis3* effect is connected with this. This is possibly due to an effect on Hox/Meis synergy: See below. The phenotype observed after injection of less morpholino, namely loss of trunk structures, head defects, and retarded tail formation described in this report and by Dibner and co-workers [@pone.0018010-Dibner1], is therefore most likely a result of reduced XMeis3 function, not a complete loss of function. We cannot be certain that the phenotype caused by injection of 32 ng MO*^XMeis3^* represents the complete loss-of-function phenotype, but it suggests the need for XMeis3 in two processes during early development: the progression of gastrulation and *Hox* expression and patterning in the early mesoderm and hindbrain. Synergy between *Hoxd1* and *XMeis3* {#s3d} ------------------------------------ We suspected that Meis3 is important for Hox expression because it mediates Hox autoregulation so we tested whether Hoxd1 and Meis3 synergise in early gastrula mesoderm. The synergistic effects we have observed in the gain-of-function experiment by injection of synthetic *XMeis3* and *Hoxd1* mRNA together show that these two factors, when co-expressed can indeed generate a phenotype that cannot be accomplished by injecting double the amount of either factor separately. These results recall the findings of Vlachakis and co-workers [@pone.0018010-Vlachakis1], who have shown that in zebrafish embryos, Meis3, Pbx4, and Hoxb1 synergise to promote hindbrain fate. Combined Hoxd1 and XMeis3 loss-of-function also indicates synergy; while sub optimal amounts of either morpholino against *Hoxd1* or *XMeis3* led to a reduction of *Hoxd1* expression, the combination led to a much stronger reduction. This adds to the evidence for a synergistic relation between Hoxd1 and XMeis3. Taken together our results show that *XMeis3* is necessary in marginal zone mesoderm to establish the early expression of *Hox* genes. This *XMeis3*-mediated mesodermal *Hox* cascade is of vital importance for axis formation and AP patterning. Autoregulation by *Hoxd1* is necessary for establishment of its expression in marginal zone mesoderm {#s3e} ---------------------------------------------------------------------------------------------------- Autoregulation dependent on Pbx has been shown for Hox paralog group 1 and 4 members in neurectoderm [@pone.0018010-Ryoo2], [@pone.0018010-Ppperl1]--[@pone.0018010-Marty1], [@pone.0018010-Gould1]. This suggests that the the regulation of *Hox* expression by XMeis3 that we have demonstrated could take place at the level of Hox autoregulation. Indeed, injection of MO*^Hoxd1^* led to a reduction in *Hoxd1* mRNA expression. The expectation is that this is the result of a reduction in *Hoxd1* translation, leading to a reduced amount of Hoxd1 protein and we suspect that this causes a reduction in availability of *Hoxd1* mRNA because of autoregulation. This suggests that Hoxd1 autoregulation is an essential step in the establishment (but not initiation), and not only the maintenance (as in neurectoderm), of *Hoxd1* expression in mesoderm during gastrulation in *Xenopus* embryos. We do not yet know whether this autoregulation is direct or indirect and have no evidence as to the mechanism. However, the involvement of *Meis3* suggests that it is by the known mechanism [@pone.0018010-Ryoo2], [@pone.0018010-Ppperl1]--[@pone.0018010-Grieder1], [@pone.0018010-Gould1]. The observed reduction of *Hoxd1* expression could also be explained if binding of MO*^Hoxd1^* to mRNA led directly to destabilisation of the *Hoxd1* messenger, however this effect has, to our knowledge, not been reported and our findings (above) of the necessity of *Meis* for mesodermal Hox expression and for synergy between *Hoxd1* and *Meis3* also point strongly to autoregulation via the known *Meis* dependent mechanism [@pone.0018010-Ryoo2], [@pone.0018010-Ppperl1]--[@pone.0018010-Grieder1], [@pone.0018010-Gould1]. The necessity for *Hoxd1* autoregulation in mesoderm is a remarkable discovery considering that vertebrate Hox autoregulation has previously only been shown in the hindbrain We note that *Hoxd1* loss-of-function is clearly not fully, if at all, rescued by the other labial type Hox genes; : *Hoxa1* and *Hoxb1* that are normally co-expressed during gastrulation. Either *Hoxa1* and *Hoxb1* are not capable of inducing the expression of *Hoxd1*, which seems unlikely taking into account the redundant functions of these paralog group members (reviewed in [@pone.0018010-Morrison1] and references therein), or expression of *Hoxa1* and *Hoxb1* is reduced or prevented by *Hoxd1* loss-of-function. This second idea would suggest the necessity of *Hoxd1* to induce the two other labial homologous (which are expressed slightly later) during gastrulation in *Xenopus* embryos. Additional experiments are needed to distinguish between the two possibilities but whatever the outcome, this finding sheds new light on the initiation and establishment of expression of the early gastrula *Hox* cascade. Obviously, auto and cross regulation can not be involved in initiating the very first expression of Hox genes. We conclude that autoregulation is involved only in the establishment and maintenance phases of Hox expression and not initiation. In fact, we have evidence that Hox expression in the Xenopus gastrula is initiated by Wnt8, which directly induces expression of Hoxd1 and of its paralogues but not of other Hox genes [@pone.0018010-InderRieden1]. Concluding Remarks {#s3f} ------------------ Our investigations shed new light on the roles of Meis3 and of Hox genes in early embryonic development and axial patterning. We made four main findings which relate to the role of the early gastrula non organiser mesoderm which has recently been shown to be very important in early embryonic patterning [@pone.0018010-Wacker1], [@pone.0018010-Durston1]. This early mesoderm is important because it is the first embryonic tissue to express *Hox* genes. It has a temporally collinear sequence of *Hox* gene expression that is used to ste up the spatially collinear *Hox* sequence in the later embryo\'s axial pattern by time- space translation [@pone.0018010-Wacker1], [@pone.0018010-Durston1] We show here that *Xmeis3* and *Meis-Hox* synergy are needed for setting up this early mesodermal *Hox* sequence 1. 1/ We showed for the first time that *Meis3* starts to be expressed earlier in the early Xenopus embryo than preiously reported: in the non organiser mesoderm at the early gastrula stage St 10.5 rather than the early neurula stage, after gastrulation. This early mesodermal *Meis* expression overlaps with the early mesodermal expression of the *Hox* genes. 2. 2/ We showed for the first time that artificial ectopic expression of *Meis3* causes ectopic expression of *Hox* genes in the early gastrula non organiser mesoderm as well as in embryonic neurectoderm. This ectopic expression occurs only in tissue that is in contact with non organiser mesoderm expressing the *Hox* gene in question or another *Hox* gene, indicating the need for additional endogenous factors for ectopic expression. We speculate that these may be the Hox proteins themselves. This finding constituted our first piece of evidence suggesting that *Meis3* may be needed for early gastrula *Hox* expression. 3. 3/ We showed for the first time that *Meis3* loss of function via antisense oligonucleotide morpholinos blocks or downregulates *Hox* gene expression in early gastrula non organiser mesoderm. This is evidence that mesodermal *Meis* is indeed needed for mesodermal *Hox* expression. 4. 4/ We showed for the first time that endogenous and ectopic *Meis3* and *Hoxd1* can and do synergise to induce *Hoxd1* expression in early gastrula mesoderm. This is evidence that synergy between Meis and Hox mediates mesodermal expression of at least one Hox gene. We believe that this reveals a detail of how *Meis3* regulates *Hoxd1* expression. Materials and Methods {#s4} ===================== *Xenopus* embryos and microinjections {#s4a} ------------------------------------- Pigmented *Xenopus laevis* embryos were obtained by *in vitro* fertilisation, and after dejelling in a 2% cysteine solution (pH 8.0), cultured in 0.1× Marc\'s Modified Ringers\'s (MMR) [@pone.0018010-Sive1] containing 50 µg/ml gentamycin at 14--21°C. Embryos were injected in 1× MMR+4% ficoll and afterwards transferred to 1× MMR+1% Ficoll, and cultured in this medium for 1 to 7 hours, after which they were transferred and to 0.1× MMR in which they were cultured until harvesting. Staging of the embryos was performed according to Nieuwkoop and Faber [@pone.0018010-Nieuwkoop1]. Embryos at the one-cell stage were injected into the animal pole with synthetic mRNA dissolved in water. The synthetic capped mRNA was made using the Ambion mMessage mMachine Kit with CS2-*XMeis3*, or CS2-*Hoxd1*, linearised with *Not*I, as template. CS2-*XMeis3* was constructed by cloning the full-length coding region of *XMeis3*, obtained by PCR using stage 15 cDNA as template and the following primers: f: 5′-gcgggatccatggcacaaaggtatgatgag, r: 5′--cgcctcgagcatgtagtgccactgcccctcc, containing an *BamH*I or a *Xho*I restriction site respectively, in the CS2+ vector [@pone.0018010-Rupp1] using the restriction sites in the primers. CS2-Hoxd-1 contains the complete coding sequence of *XHoxd1* in CS2+, kindly provided by W. Van den Akker. MO*^XMeis3^* Gene Tools, LLC, (directed against the XMeis3 mRNA\'s 5′ region) has the sequence: 5′-cctttgtgccattccgagttgggtc, and was injected in amounts of 8 to 36 ng. in a concentration of 8 ng/nl. MO^contr^, supplied by Gene Tools, LLC, has the licence: 5′-cctcttacctcagttacaatttata and was injected using the same amounts and concentrations as MO*^XMeis3^*. Whole mount *in situ* hybridisation and antisense probes {#s4b} -------------------------------------------------------- Whole mount *in situ* hybridisations were performed according to Harland (1991), with minor modifications. The antisense RNA probes were generated by run off *in vitro* translation using DIG RNA labelling mix (Roche), and T7 or Sp6 RNA polymerase (Promega). The probes were generated using the following templates: *Hoxd1*: [@pone.0018010-Sive2], *Hoxb4*: a 708 bp fragment containing the complete *Hoxb-4* ORF cloned in pGEMTE, *Hoxc6*: a 998 bp *Hoxc-6* fragment in pGEM1 containing a part of the homeodomain and extending into the 3′ UTR, *Xcad3*: [@pone.0018010-Pownall1]; *Xbra*: pSP73Xbra [@pone.0018010-Smith1]. Ethics {#s4c} ------ Our work uses early Xenopus embryos. It is carried out according to national and EU guidelines and regulations. The animal work is covered by a 'DEC' licence, No. 513, covering ethical aspects, issued by the University of Leiden\'s Animal Experimentation Committee (Dier Experimenten Commissie) to A.J.Durston. The molecular work is covered by an existing COGEM licence No. GGO 02-055, issued to H.P.Spaink. This work requires no other licence. We thank W. van den Akker for kindly providing the CS2-*Hoxd1* plasmid prior to publication and C. McNulty, G. Mainguy, and S. Wacker for critically reading the manuscript. **Competing Interests:**The authors have declared that no competing interests exist. **Funding:**This work was supported by EU grant FP6 No. LSHM-CT-2003-504468. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. [^1]: Conceived and designed the experiments: PR HJ AD. Performed the experiments: PR. Analyzed the data: PR HJ AD. Contributed reagents/materials/analysis tools: PR HJ AD. Wrote the paper: PR HJ AD.

Introduction {#s1} ============ Nowadays, the global prevalence of diabetes in pregnant women continues to increase. Consequently, the relationship between *in utero* exposure to maternal diabetes and the incidence of cardiovascular diseases in offspring, and especially that of hypertension, has been the subject of clinical and experimental studies. Indeed, numerous works show that offspring of diabetic mothers have a higher systolic blood pressure (SBP) than offspring of control mothers (Bunt et al., [@B15]; Rocha et al., [@B47]; Wichi et al., [@B57]; Nehiri et al., [@B39]; Gomes and Gil, [@B24]; Aceti et al., [@B2]). Increasing evidences propose that functional changes in the vasculature could be a major reason for the increase of blood pressure in offspring of diabetic rats (DMO) (Manderson et al., [@B37]; Rocha et al., [@B47]; Simeoni and Barker, [@B51]; Duong Van Huyen et al., [@B20]; Porto et al., [@B44]; Vessières et al., [@B54]). In experimental studies, it is well known that exposure to maternal diabetes induces vascular dysfunction in large arteries of DMO, reflected by a reduced response to endothelium-dependent vasodilators and an increased contractile response. Indeed, Duong *et al* have demonstrated a specific gene expression profile of the thoracic aorta in favor of vasoconstriction with altered prostacyclin-induced vasodilation, linked to a reduction of prostacyclin receptor expression (Duong Van Huyen et al., [@B20]). Endothelial dysfunction of the microcirculation has also been described; functional studies show that endothelium-mediated vasodilation induced by acetylcholine is reduced (Holemans et al., [@B27]; Rocha et al., [@B47]; Ramos-Alves et al., [@B45]). Recently, we have shown that in resistance arteries of old DMO rats, endothelial vasodilator dysfunction was exacerbated and pressure-induced (myogenic) tone was maintained at a high level (Vessières et al., [@B54]). Because of a close relationship between vascular tone and wall structure (Bakker et al., [@B9]), modifications of vascular tone are able to induce structural adaptation of vessels. It is well known that persistent vasoconstriction induces inward remodeling in several types of arteries (Bakker et al., [@B9], [@B6], [@B5]). This type of remodeling could be inhibited or reversed by vasodilator compounds such as calcium channel inhibitors (Bakker et al., [@B9]) or transglutaminase 2 inhibitor like cystamine (Eftekhari et al., [@B21]). Thus, modifications of vascular tone could not only be involved in the development of hypertension but also have an impact on vascular structure. Even if the relation between prenatal conditioning of hypertension and microvascular function is well documented in the case of *in utero* exposure to maternal diabetes, the impact on vascular structure and blood flow variations have not yet been studied. Moreover, alterations of the vascular structure could influence remodeling mechanisms and responses to hemodynamic variations. In addition maladaptive vascular remodeling is currently recognized as an important contributor to the development of cardiovascular pathologies (Pasterkamp et al., [@B41]). Thus, our objective was to study the impact of *in utero* exposure to maternal diabetes on vascular structure on basal conditions (3 months of age) and remodeling induced by chronic hemodynamic changes. Materials and methods {#s2} ===================== Animals ------- Pregnant Sprague--Dawley rats, weighing 250--300 g, were made diabetic on day 0 of gestation by a single intraperitoneal injection of streptozotocin (35 mg/kg, Sigma, St Quentin Fallavier, France) as previously described (Duong Van Huyen et al., [@B20]). The diabetic state was checked in fasted rats by measuring the plasma glucose concentration (AccuChek®, Roche, Boulogne-Billancourt, France). Only pregnant females whose plasma glucose ranged between 300 and 450 mg/dl were included in the study. This diabetic status was confirmed every 2 days until delivery (Figure [S1](#SM1){ref-type="supplementary-material"}). On the day of delivery the newborn rats were weighed. Each litter was then reduced to 10 pups. All animals were kept in a temperature and light controlled room, at 21°C with a 12 h light cycle. They had access to food (diet n°3430, Serlab, Compiègne, France) and tap water *ad libitum*. Control animals were born of non-diabetic females (CMO). We used 30 CMO males and 30 DMO males from at least 6 different litters for each group. One animal per litter is included in each experimental group. Before sacrifice, glycemia of each fed animal was measured. All experiments were conducted in accordance with the institutional guidelines and the recommendations for the care and use of laboratory animals of the French Ministry of National Education, Research and Innovation. The protocol was approved by the Ethics Committee of "Pays de Loire" (permit n° 00960.01). Arterial blood pressure measurements ------------------------------------ SBP was measured in conscious rats by tail-cuff plethysmography (BP-2000, Visitech, Apex, USA). Briefly, this technique uses transmission photoplethysmography in which variations of light transmitted through the tail is the basic signal that is analyzed to determine the blood pressure and pulse rate. For good quality, reliability and reproducibility of the measurements, rats were trained 1 week before experimentation. Then, blood pressure was recorded in quiet animals. For each animal, the mean SBP was averaged from 15 measurements recorded every day for 7 consecutive days. Model of flow-induced remodeling -------------------------------- Three-month old male rats were anesthetized (Isoflurane, 2.5%) and submitted to surgery in order to modify blood flow in the mesenteric circulation as described previously (Bouvet et al., [@B13]). Briefly, consecutive first-order arteries were used. Ligations (7-0 silk surgical thread) were applied to second-order branches of the first artery (low flow artery, LF). Equivalent arteries located at a distance from the ligated artery were used as control arteries (normal flow, NF) (**Figure 3A**). At the time of anesthesia and at the end of surgery, animals were treated with buprenorphine (TEMGESIC, 0.1 mg/kg, s.c.). After 7 or 21 days the rats were sacrificed by CO~2~ inhalation. The gut was excised and the mesenteric arteries were gently dissected. From each rat, LF and NF arteries were isolated and divided in several segments (from proximal to the distal part of the artery) for functional measurements, histological study and biochemical analyses, respectively. Pressure--diameter relationship in isolated mesenteric arteries --------------------------------------------------------------- LF and NF arteries were cannulated at both ends and mounted in a video monitored perfusion system, as previously described (Retailleau et al., [@B46]). Briefly, cannulated arterial segments were placed in a 5 ml organ bath containing a Ca^2+^-free physiological salt solution with EGTA (2 mmol/L) and sodium nitroprusside (SNP, 10 μmol/L). Pressure was progressively increased (10--125 mmHg) in order to determine passive arterial diameter. Pressure and diameter measurements were collected using a data acquisition system (Biopac MP100 and Acqknowledge® software, Biopac). At the end, arterial segments were fixed with formaldehyde under a pressure of 75 mmHg in order to perform histological studies. Western blot analysis in mesenteric arteries -------------------------------------------- Western blot analysis of proteins of interest was performed. In brief mesenteric arteries were rapidly dissected and frozen. Then, proteins were extracted and the total protein content was determined by the Bradford technique in order to file equal amounts (15 μg) of the denatured proteins. Membranes were incubated with polyclonal antibodies directed against transglutaminase 2 (TG2, ab421, 1/1000 TBST-BSA 0.5%, Abcam) or mitochondrial Mn superoxyde dismutase (SOD2, SOD-110, 1/1000 TBST-BSA 0.5%, Enzo Life Sciences). The detection was performed by chemiluminescence emitted from luminol oxidized by peroxidase (ECL system, GE Healthcare, Velizy-Villacoublay, France). Each protein expression was compared to GAPDH (14C10, 1/1000 TBST-BSA 0.5%, Cell Signaling) and expressed as ratio between protein level and GAPDH. Reactive oxygen species (ROS) measurement ----------------------------------------- ROS detection was performed on 3-month old CMO and DMO rats. Transverse cross-sections 7 μm thick of mesenteric arteries were incubated with dihydroethydine (DHE), as previously described (Cousin et al., [@B18]). DHE, in the presence of superoxide, is briefly oxidized to fluorescent ethidium bromide. Ethidium bromide is trapped by intercalation with DNA, and the number of fluorescent nuclei indicates the relative level of superoxide production. Positive staining was visualized using confocal microscopy (Nikon Eclipse TE2000S) and MetaMorph® software (Molecular Devices, Sunnyvale, USA; Duong Van Huyen et al., [@B20]). Transglutaminase activity ------------------------- Mesenteric arteries of 3-month old CMO and DMO rats were split into 4 groups and incubated at 37°C during 24 h in 100 μL buffer containing Leibovitz medium with 10% fetal bovine serum, a mix of antibiotic-antimycotic solution (1%, 15240062, Gibco, Courtaboeuf, France), Dapi (5 μg/mL, Molecular Probes, Carlsab, USA) and either Alexa Fluor594/Cadaverine (10 μmol/L, A-30678, Invitrogen, Carlsab, USA), or (Aceti et al., [@B2]) Alexa Fluor594/Cadaverine + SNP **(**10^−3^ mol/L), or (Bunt et al., [@B15]) Alexa Fluor594/Cadaverine + dithiothreitol (DTT, 2 mmol/L, 43816, Sigma, St Quentin Fallavier, France), or (Gomes and Gil, [@B24]) Alexa Fluor594/Cadaverine + DTT (2 mmol/L) + SNP (10^−3^ mol/L). Vessels were mounted on Mowiol and imaged on a confocal microscope (Nikon Eclipse TE2000S). Transglutaminase activity was quantified by spatial integration of AlexaFluor594 signal with ImageJ software. Data were corrected for vessel size and depicted in arbitrary units. Isolated perfused kidney ------------------------ Three-month old male rats were operated under anesthesia (2.5% isoflurane). After laparotomy, renal arteries, the abdominal aorta, the superior mesenteric artery and the inferior vena cava were cleaned of the surrounding connective tissue. The kidney was released and the crossroads of the abdominal aorta with left renal artery was carefully dissected. A catheter (PE50/PE20, FMD) was inserted from the abdominal aorta to the renal artery (Weiss et al., [@B56]) and was maintained by a ligation on the abdominal aorta. The kidney, perfused continuously at 37°C by Krebs solution (according to El-Mas et al., [@B22]), was then removed carefully and placed in an organ bath (El-Mas et al., [@B22]) connected to a peristaltic pump to infuse kidney at given rates. The renal perfusion pressure was measured in response to stepwise increase in perfusion flow (4--50 ml/min). Pressure measurements were collected using a data acquisition system (Acqknowledge® software, Biopac, Paris, France). Histomorphometry analysis ------------------------- Each segment of mesenteric or renal arteries and thoracic aorta were embedded in Tissue-Tek (Sakura) and frozen in isopentane. sections (7 μm) were stained with orcein in order to measure histomorphometric parameters (internal and external diameters and medial cross-sectional area, MCSA) after image acquisition (Nikon Eclipse E600 microscope, Sony camera) and analyzed using ImageJ software (NIH). Vascular smooth muscle cell attachment to elastic lamellae ---------------------------------------------------------- Thoracic aorta of 3-month old CMO and DMO rats were fixed in 0.1 M phosphate buffer containing 2.5% of glutaraldehyde (LFG) overnight at 4°C. Samples were then post-fixed with 1% osmium tetroxide/1% potassium ferricyanide for 45 min and dehydrated in a graded series of ethanol. Samples were finally embedded into epoxy resin (Epon, LFG) and ultrathin sections (70 nm) were cut, contrasted with 3% uranyle acetate for 15 min and observed with a JEOL 1,400 transmission electron microscope (JEOL) at an accelerating voltage of 120 keV. Image acquisition was made at a magnification of X 12,000. Vascular smooth muscle cell attachments to elastic lamellae correspond to expansions of VSMCs composed by oxytalan fibers that span obliquely from the dense plaques of VSMCs to the extracellular matrix (ECM) (Dingemans et al., [@B19]). Then the number of anchorage sites between VSMCs and ECM was measured as previously described (Bezie et al., [@B11]). Results were obtained by analyzing 50 images per animal. Statistical analysis -------------------- Results were expressed as means ± SEM. Each "n" corresponds to the number of animals per group. Significance of the differences between NF and LF for pressure-diameter curves or between DMO and CMO for perfusion pressure-flow curves was determined by regular 2-way ANOVA followed by Bonferroni\'s multiple comparison test or a non-parametric Mann-Whitney test for histological analysis. Values of *p* \< 0.05 were considered to be significant. All statistical analysis were performed using GraphPad Prism® software. Results {#s3} ======= Physiological vascular structure in rats exposed *in utero* to maternal diabetes -------------------------------------------------------------------------------- At 3 months of age, DMO and CMO rats have similar physiological parameters (i.e., body weight, glycemia and mean blood pressure, MBP, Table [1](#T1){ref-type="table"}). Interestingly, in basal conditions, mesenteric resistance arteries of DMO rats have a smaller diameter than those of CMO rats (Table [1](#T1){ref-type="table"}). This is associated with a reduced MCSA (Table [1](#T1){ref-type="table"}) indicating a basal structural reorganization of the vascular wall in these arteries. For conductance arteries, no difference in diameter and MCSA is observed between CMO and DMO rats at 3 months of age (Table [1](#T1){ref-type="table"}); but we measured an increased number of connections between VSMCs and ECM in these vessels (Figure [1](#F1){ref-type="fig"}). ###### Physiological parameters (body weight, glycemia and mean blood pressure, MBP) and morphology (external diameter and cross-sectional area of the media, MCSA) of resistance mesenteric artery and conductance thoracic aorta of 3-month old control (CMO) and diabetic (DMO) mother offspring. **CMO** **DMO** ---------------------------------- ----------------------------------------- ------------------------------------------------------------------------- **PHYSIOLOGICAL PARAMETERS** Body weight (g) 435.2 ± 11.1 (*n* = 29) 421.8 ± 10.7 (*n* = 30) Glycemia (mg/dl) 155.3 ± 9.4 (*n* = 29) 153.7 ± 11.4 (*n* = 30) SBP (mmHg) 118 ± 3 (*n* = 29) 123 ± 4 (*n* = 30) **RESISTANCE MESENTERIC ARTERY** External diameter (μm) 368.4 ± 9.9 (*n* = 9) 320.2 ± 12.9[^\*\*^](#TN1){ref-type="table-fn"} (*n* = 9) MCSA (μm^2^) 15.9 × 10^3^ ± 1.6 × 10^3^ (*n* = 9) 10.3 × 10^3^ ± 0.8 × 10^3^[^\*\*^](#TN1){ref-type="table-fn"} (*n* = 9) **CONDUCTANCE THORACIC AORTA** External diameter (μm) 1829.6 ± 40.1 (*n* = 12) 1894.6 ± 22.6 (*n* = 13) MCSA (μm^2^) 414.6 × 10^3^ ± 19.5 × 10^3^ (*n* = 12) 432.7 × 10^3^ ± 19.1 × 10^3^ (*n* = 13) *p \< 0.01 DMO vs. CMO*. {#F1} Modification of resistance artery hemodynamic in rats exposed *in utero* to maternal diabetes --------------------------------------------------------------------------------------------- Because small variations of diameter in resistance arteries could modify vascular resistances, we studied the renal arterial system, the most important microcirculatory system sensitive to flow and peripheral resistance. Under basal condition, the histological study revealed a decreased internal diameter of the renal artery associated with an increase of MCSA in 3-month old DMO compared to CMO rats, leading to an increase in media to lumen ratio in DMO rats (Figure [2A](#F2){ref-type="fig"}). In addition, in isolated perfused kidneys stepwise increases in flow induced a progressive rise in perfusion pressure. This flow-pressure relationship is shifted leftward in DMO compared to CMO (Figure [2B](#F2){ref-type="fig"}), suggesting a rise in renal peripheral vascular resistance. {#F2} Impact of decreased flow on resistance artery remodeling -------------------------------------------------------- After 1 week of ligation, a stepwise increase in pressure induced a significantly increased passive diameter in NF artery than in LF artery of CMO rats (Figure [3B](#F3){ref-type="fig"}). This was associated with a decrease of internal diameter and MCSA (Figure [3C](#F3){ref-type="fig"}), with no modification of M/L ratio (Figure [3C](#F3){ref-type="fig"}). These results are representative of the development of an inward remodeling in LF arteries of CMO rats. Interestingly, for DMO rats, we do not observe a different behavior between NF and LF arteries in response to stepwise increase in pressure (Figure [3B](#F3){ref-type="fig"}). In addition, internal diameter and M/L ratio are similar between LF and NF DMO rats as well as MCSA (Figure [3C](#F3){ref-type="fig"}), showing an impairment of arterial response to the chronic decrease in flow. Interestingly, 3 weeks after ligation, histological parameters and passive diameter evolution of LF mesenteric artery in DMO rats were similar to those obtained 1 week post-ligation (Figure [S2](#SM1){ref-type="supplementary-material"}), which demonstrates a total absence of inward remodeling development in the case of low flow perturbations in these animals. {#F3} Permanent oxidative stress and transglutaminase inactivity in resistance arteries of rats exposed *in utero* to maternal diabetes --------------------------------------------------------------------------------------------------------------------------------- At a basal level, we measure a high level of ROS in NF arteries of DMO compared to CMO rats (Figure [4A](#F4){ref-type="fig"}), indicating an activation of the oxidative stress pathway in DMO animals. Moreover, although ROS level increases in CMO in response to decreased-flow, there is no more increase of ROS production in DMO LF arteries; on the contrary, we observe a 50% decrease of ROS level in theses LF arteries (Figure [4A](#F4){ref-type="fig"}). In parallel, western blot analysis demonstrates a high expression of the protective mitochondrial SOD2 in response to increased oxidative stress in DMO rats although SOD2 protein expression does not change between NF and LF in CMO rats (Figure [4A](#F4){ref-type="fig"} and Figure [S3](#SM1){ref-type="supplementary-material"}). {#F4} In order to test if endogenous TG2 could be activated by a cell-permeable reducing agent, mesenteric arteries were incubated with DTT. In 3-month old CMO rats, TG2 activity, as indicated by incorporation of Alexa Fluor-594/Cadaverine, is higher after stimulation by DTT than in non-stimulated vessels. Moreover, TG2 activity was strongly reduced by SNP (NO donor). Interestingly, in DMO mesenteric arteries, after DTT stimulation we detect a very low TG2 activity compared to CMO mesenteric arteries (Figure [4B](#F4){ref-type="fig"}). In parallel, western blot analysis does not show modification of tissue-TG2 expression level in NF and LF arteries either in CMO or in DMO (Figure [4B](#F4){ref-type="fig"} and Figure [S3](#SM1){ref-type="supplementary-material"}). Absence of inward remodeling of thoracic aorta in response to chronic high blood pressure in rats exposed *in utero* to maternal diabetes ----------------------------------------------------------------------------------------------------------------------------------------- At 18 months of age, although we measure a 1.5-fold increase of SBP in CMO rats, the increase observed in DMO rats is greater (a 2-fold increase of SBP, Figure [5A](#F5){ref-type="fig"}). This strong increase in SBP reflects the development of hypertension in DMO animals, as previously described (Nehiri et al., [@B39]). Histological analysis of thoracic aorta sections did not show any difference in MCSA, external and internal diameters and M/L ratio between CMO and DMO rats at 3 months of age (Table [1](#T1){ref-type="table"} and Figure [5B](#F5){ref-type="fig"}). In 18 months-old animals, we measure an equivalent rise of internal diameter both in CMO and DMO despite hypertension (Figure [5B](#F5){ref-type="fig"}). In addition, M/L ratio is enhanced at 18 months of age (Figure [5B](#F5){ref-type="fig"}). Nevertheless, the magnitude of the remodeling response is less important in DMO than in CMO rats while they have the highest level of blood pressure (Figure [5A](#F5){ref-type="fig"}). These results show an inadaptive arterial remodeling in response to high blood pressure in DMO rats. {#F5} Discussion {#s4} ========== Our study demonstrates, for the first time, that *in utero* exposure to maternal diabetes induces permanent structural modifications of the vasculature. Indeed, we found deep vascular modifications of resistance (i.e., renal and mesenteric arteries) and conductance (i.e., thoracic aorta) arteries in basal conditions, but also an absence of vascular adaptation to hemodynamic perturbations (i.e., decreased flow and hypertension) in male DMO rats. Previous studies in male DMO rats highlighted the impact of maternal diabetes on programming of hypertension; in this model arterial blood pressure rises above normal pressure values around 6 months of age (Nehiri et al., [@B39]). The developmental origin of hypertension during adulthood is now well admitted and linked to nutritional insults in early life both in experimental (Liu et al., [@B35]; Tain et al., [@B53]) and clinical studies (Alexander, [@B3]; Aceti et al., [@B2]; Szostak-Wegierek, [@B52]; Levy et al., [@B34]). Even if sexual dimorphism in developmental programming of hypertension is well established (Ojeda et al., [@B40]), a large number of studies, including ours, are particularly interested in male offspring because cardiovascular function seems to be more affected than in female offspring (Cheong et al., [@B17]). Indeed previous studies have highlighted a protective role of sex hormones leading to a less pronounced phenotype in female than in male offspring (Romano et al., [@B48]). In the present study, we demonstrate that in the absence of any hemodynamic perturbation (3 months of age), male DMO rats have already deep changes in vessel structure. First of all, electronic microscopy analysis of thoracic aorta showed an increased number of connections between VSMCs and elastic lamellae at 3 months of age. This structural modification occurred without difference in wall components (e.g., collagen and elastin levels), as shown in our previous work (Duong Van Huyen et al., [@B20]). This vascular wall restructuring is normally associated with sustained hypertension to produce mechanical adaptation of the arterial wall as described in spontaneously hypertensive rats (Bezie et al., [@B11]). In our study, despite the normal SBP observed in 3-month old DMO rats, the *in utero* exposition to maternal diabetes might induce adaptive vascular development in order to better support the programmed high blood pressure later in life. Furthermore, in DMO rats, a narrowing of resistance arteries seems to be present in basal conditions with smaller internal and external diameters. In this study, we show that DMO rats have a high level of ROS; this result is in agreement with the high protein expression level of SOD2, an endogenous mitochondrial antioxidant, in order to counterbalance the negative effects of ROS. Interestingly, we observe the same profile of SOD2 protein expression between CMO and DMO. However, the levels of ROS in NF and LF of CMO and DMO rats are different. In fact, ROS quantification by DHE represents a global analysis of cellular and mitochondrial stress while SOD2 is bound to the protective mitochondrial oxidative stress (Sack et al., [@B49]). Then the difference observed between CMO and DMO reflects a different origin of ROS; in LF CMO rats, the ROS seem to originate from both cellular and mitochondrial oxidative stress although in NF DMO rats, the ROS seem to have a predominantly mitochondrial origin. Nevertheless, our observations, coupled with the fact that DMO develop a vasoconstrictor phenotype (Duong Van Huyen et al., [@B20]; Vessières et al., [@B54]), could explain the narrowing of resistance arteries observed in basal conditions. But, in the current state it is difficult to exclude an impact of *in utero* exposure to maternal diabetes on vessel development. Indeed, a recent study on chicken embryos has shown that high glucose inhibited development of the blood vessel plexus which led to the development of narrower vessels (Jin et al., [@B30]). On the other hand, insulin, by its positive regulation on IGF-1 (insulin-like growth factor-1) level, could influence vessel size (Piecewicz et al., [@B42]). Thus, a decrease in insulin content may lead to smaller vessels. Such a decrease of fasting venous blood insulin in children exposed to maternal diabetes has been recently described (Sauder et al., [@B50]). But in our previous work, we did not observe modification of pancreatic insulin concentration or anatomical structure of the pancreas in DMO rats (Blondeau et al., [@B12]). A small decrease of a vessel diameter can dramatically increase vascular resistance (because vessel resistance is inversely proportional to the vessel radius to the fourth power according to Poiseuille\'s equation). Then, variations in diameter of resistance arteries can impact homeostatic systems and especially kidney hemodynamic. In this study we observe a higher-pressure response to flow in DMO isolated perfused kidney at 3 months of age, suggesting that renal vascular resistance is altered. Some studies suggest that maternal hyperglycemia can promote remarkable changes both in kidney morphology and function. Then, the increase of vascular resistance could be linked to the basal hypertrophic remodeling (increased MCSA) of renal resistance artery that we described, but also to a decrease in nephron number since others studies have shown an impaired nephrogenesis with a reduction of 30% of nephron number in association with a decreased renal function in DMO rats (Amri et al., [@B4]; Nehiri et al., [@B39]). This impaired kidney development would be the consequence of an ureteric branching morphogenesis reduction (Hokke et al., [@B26]). Nevertheless, modifications of kidney hemodynamics are directly implicated in hypertension. Brenner et al. have been the first to establish a relationship between intra-uterine environment, decreased nephron number and the development of hypertension (Brenner et al., [@B14]). Clinical results show, in association with a higher SBP, that adult offspring of type 1 diabetic mothers have a reduced renal functional reserve, which may reflect a decreased number of nephrons (Abi Khalil et al., [@B1]). As in the case for decreased nephron number, resistance artery remodeling could be implicated in the development of hypertension during adulthood. Inward remodeling is defined as an increase of wall-to-lumen ratio, associated with hypertrophy in conductance arteries (Heagerty et al., [@B25]; Intengan and Schiffrin, [@B29]), or without medial enlargement in resistance arteries (Mulvany and Halpern, [@B38]; Gibbons and Dzau, [@B23]). This type of remodeling, although beneficial in the beginning, may subsequently contribute to cardiovascular disorders. However, chronic alterations in the hemodynamic profile (i.e., chronic changes in blood pressure or blood flow) may potentiate arterial remodeling (Langille, [@B31]; Lehoux and Tedgui, [@B33]). In the present study, although DMO rats have an important rise of blood pressure at 18 months of age, we do not detect any modification of internal diameter and MCSA of the thoracic aorta compared to CMO rats. This absence of inward remodeling in case of hypertensive state could be related to the reorganization of the vessel wall (increased connections between VSMCs and ECM), which contributes to the maintenance of a normal level of wall stiffness despite the increase in the pressure wall stress. In a previous study, we also found an absence of inward remodeling of mesenteric resistance arteries in response to this hypertension (Vessières et al., [@B54]). But we were not able to detect modification in the number of connections between elastic lamellae and ECM in this type of vessel. Nevertheless, in our model of rats exposed *in utero* to maternal diabetes, both resistance (mesenteric) (Vessières et al., [@B54]) and conductance (thoracic aorta) arteries (present study) are unable to develop inward remodeling in response to high blood pressure. A chronic decrease in blood flow is also able to induce a reduction in lumen diameter of resistance arteries (inward eutrophic remodeling) (Buus et al., [@B16]; Baron-Menguy et al., [@B10]). The model used allows the study of comparable mesenteric resistance arteries submitted to low or normal flow *in vivo* without changes in physiological hemodynamic conditions (i.e., blood pressure). Our results show that ligation of mesenteric resistance arteries, resulting in a low blood flow, does not induce inward remodeling either. Bakker et al. have demonstrated that flow-mediated remodeling is directed by vascular tone (i.e., vasoconstriction inducing inward remodeling in opposition to vasodilatation which induces outward remodeling) (Bakker et al., [@B8]). Whether it is triggered by high pressure or low blood flow, the resulting inward remodeling requires 3 conditions: first a partial digestion of the ECM, secondly a chronic vasoconstriction and finally a matrix reorganization (Bakker et al., [@B9]; Langille and Dajnowiec, [@B32]; Huelsz-Prince et al., [@B28]). We have previously demonstrated that *in utero* exposure to maternal diabetes resulted in a fetal programming of vascular function in favor of vasoconstrictor tone (Duong Van Huyen et al., [@B20]; Vessières et al., [@B54]). Nevertheless, a chronic decrease of blood flow does not induce the development of an inward remodeling in DMO rats. This absence of vascular wall reorganization in response to decreased flow is probably the result of an inability of the vessel to further decrease its diameter. Changes in blood flow induce an inflammatory response responsible for oxidative stress followed by the activation of metalloproteases (MMPs) causing a partial dissociation of the ECM (Vessières et al., [@B55]). Also, in the case of decreased blood flow, we do not observe a higher increase in ROS production in DMO rats compared to basal conditions or to control animals. This lack of increasing oxidative stress in LF arteries of DMO rats could be responsible for the absence of MMPs activation implicated in a partial dissociation of the ECM and then, the absence of worsening inward remodeling in DMO rats. Furthermore, several studies have shown that vascular remodeling of resistance arteries after reduced blood flow, hypertension, or exposure to vasoconstrictors depends on tissue-TG2 activity to stabilize arterial wall and normalize shear stress (Bakker et al., [@B5], [@B7]; Eftekhari et al., [@B21]; Pistea et al., [@B43]). Indeed, the major function of TG2 is to stabilize ECM proteins through the formation of specific cross-links (Bakker et al., [@B8]). In LF arteries isolated from DMO rats, TG2 activation by DTT is ineffective; by contrast with LF arteries of CMO rats in which DTT increased MMPs activity by 2 times. This absence of TG2 activity is not correlated to a decrease of TG2 protein expression. Although protein cross-linking is the main feature of TG2, providing mechanical strength to tissues (Lorand and Graham, [@B36]), how vascular smooth muscle tone induces TG2 activity remains unknown (Huelsz-Prince et al., [@B28]). Conclusion {#s5} ========== Our study clearly demonstrates that *in utero* exposure to maternal diabetes induces deep architecture vessel wall modifications with matrix reorganization in early stage of life and impacts vascular remodeling mechanisms in the case of chronic hemodynamic changes (i.e., blood pressure or flow). Then, the inability of conductance and resistance arteries to respond to high blood pressure or decreased blood flow could be an adaptive process in order to prevent cardiovascular complications due to programmed hypertension in DMO rats. Author contributions {#s6} ==================== CF: Designed the protocol, obtained grants, researched data, wrote the manuscript, contributed to discussion and edited the manuscript; AD, CP, MM, JB, and LG: Performed the experiments; AD, MM, LL, ZF, and DH: Contributed to discussion and reviewed the manuscript. 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 Guillaume Mabilleau (SCIAM, Service Commun d\'Imagerie et Analyses Microscopiques, Institut de Biologie Santé, Angers, France) who has performed electronic microscopy images acquisition and Tristan Champin (UMRS CNRS 6015, INSERM U1083) for helpful comments on the manuscript. **Funding.** This study was funded by the Région Pays de la Loire (grant n°2011-12354) and University of Angers (IPMEE project). AD received a grant from the University of Angers to support his PhD. Supplementary material {#s7} ====================== The Supplementary Material for this article can be found online at: <https://www.frontiersin.org/articles/10.3389/fphys.2018.00350/full#supplementary-material> ###### Click here for additional data file. [^1]: Edited by: Joon (Kyungjoon) Lim, La Trobe University, Australia [^2]: Reviewed by: Tamara Paravicini, RMIT University, Australia; Marianne Tare, Monash University, Australia [^3]: This article was submitted to Integrative Physiology, a section of the journal Frontiers in Physiology

Introduction {#s1} ============ When we want to study the conformations of plant proteins, their interactions and their functions in their native intracellular localization, we need to rely on a combination of molecular biophysics and cell biology. The conventional structural biology approaches that aim to elucidate the structure of proteins, such as X-ray crystallography and nuclear magnetic resonance (NMR), traditionally rely on samples of isolated, stable and folded proteins. These samples are the product of elaborate and sometimes tedious purification protocols. At the end, a homogeneous and highly concentrated protein sample usually yields a reliable and accurate description of its structural behavior. Solution-state biomolecular NMR offers an orthogonal approach to crystallographic methods, because the final experiment is not done in solid state, but with a protein that freely diffuses in an aqueous environment. Despite sample limitations in terms of the size, solubility and stability of the protein, NMR does not provide a single structural snapshot in the solid state, rather it provides comprehensive insight into the fully dynamic and flexible state of the protein that is much closer to its real functional existence (Dyson and Wright, [@B13]). A more realistic picture about life at the molecular level requires the observation of protein behavior as it happens in the cell. *In-cell* NMR is one of the techniques par excellence for this purpose. It often requires the intracellular delivery of isotopically labeled protein under conditions compatible with life, which can be accomplished with induced expression, microinjection or electroporation. There are well-documented protocols and insightful reports of proteins being studied inside mammalian cells, yeast and bacteria (Bekei et al., [@B4],[@B5],[@B6]). To the best of our knowledge, there is no precedence of *in-cell* NMR experiments in plants. Samples for solution-state NMR (and thus also for *in-cell* NMR) should fit into a quite narrow tube, which is then placed inside a spectrometer were subtle magnetic field perturbations can be recorded. Among all types of plants and tissues with distinct cellular morphologies, only cells in suspension are suitable for scrutiny when studying proteins via *in-cell* NMR. This contrasts with the convenient and open framework that, for instance, microscopy can offer, yet the high resolution information obtained via *in-cell* NMR has a unique value. Since NMR spectroscopy is an inherently low-sensitivity technique, it requires a relatively high protein concentration (in the range of 10^−6^--10^−3^ M) for collecting reliable information. Such high concentrations for a given protein are not always incompatible with normal physiology. Therefore, only proteins that are abundant in cells are eligible for such *in-cell* NMR studies. In addition, only isotopically labeled proteins (^15^N,^13^C) are detected during the NMR experiment. Hence, the protein that will be studied inside cells should be labeled with these magnetically detectable isotopes. Several types of NMR experiments can then be carried out, for example, carbon detection (Hsu et al., [@B20]) provides an approach which is not sensitive to chemical exchange of protons in the amide groups, i.e., internal pH. This imposes a clear set of conditions: (a) the protein of interest must be obtained in a pure and isotopically labeled form and then introduced into host plant cells, or (b) the protein has to be over-expressed in plant cells under labeling conditions (in a growth medium containing isotopes), preferably under the control of a strong promoter (Figure [1](#F1){ref-type="fig"}). Either strategy has advantages and disadvantages. Yet, producing the protein exogenously (e.g., recombinant expression in *Escherichia coli* under isotope-labeling conditions followed by purification) in combination with a controlled delivery into plant cells, offers the most diverse palette of options and techniques for the purpose of *in-cell* NMR. Recombinant production in *E. coli* often constitutes simple and robust protocols that lead to a sample that is devoid of post-translational modifications. This can be a drawback when the post-translational modification is essential for the protein functionality, but also offers the opportunity to study the post-translational modification inside the plant cell. ![**Schematic work flow representation to develop an *in-cell* NMR strategy to study proteins within plant cells**. The first considerations and planning of the *in-cell* NMR experiments should be based on readily available NMR data. It is desirable that the protein of interest has been characterized *in vitro* by NMR and that a chemical shift assignment is available that ultimately can be used to analyze the *in-cell* NMR dataset. Since NMR is a high resolution technique, it can address specific questions with regard to structural and dynamical aspects at the residue-level of the protein intracellular behavior. The major experimental obstacle will be the intracellular delivery of isotope-labeled protein. The success of the delivery strategy should entice monitoring the subcellular localization, intracellular protein quantification and the protein integrity. These are also 3 important parameters to check upon completing the actual *in-cell* NMR data acquisition. Likewise, it is imperative to verify the cell viability and to demonstrate that the protein has not leaked into the extracellular medium so that the NMR signals truly originate from the intracellular protein (Waudby et al., [@B54]). Another aspect to consider is the occurrence of stress to the cells during the data acquisition, which might be unveiled by an in depth proteomics analysis. To maintain the plant cell suspension in a living state, the experiment can be performed in a bioreactor that was developed for *in-cell* protein NMR. The performance and capabilities of such a bioreactor has been demonstrated by the overexpression of the intrinsically disordered human protein α-synuclein in *Escherichia coli* (Sharaf et al., [@B45]). Ultimately the *in vitro* NMR assignment should be transferred to the *in-cell* NMR datasets and employed to answer the relevant questions.](fpls-08-00519-g0001){#F1} Historically, different techniques have been used to deliver macromolecules (DNA and polypeptides) across the membranous boundaries of cells. The cell membrane can be transiently disrupted by either chemical or physical treatment, causing pore formation that allows the diffusion of material from outside the cell to its interior. Chemical manipulation of cells yielding permeabilization and delivery of materials involves the use of Ca^2+^/Mg^2+^, Ca^2+^/PEG, PEG/DMSO, detergents, toxins (e.g., streptolysin) or cell-penetrating peptides (Hanahan, [@B16]). This has proven to be a convenient way for transforming *E. coli* cells with foreign plasmid DNA. In this regard, plant cells pose a special challenge due to the presence of a glycan-rich cell wall. This can be overcome by either adjusting experimental conditions or removing the cell wall (creating protoplasts) by enzymatic degradation (Vanden Bossche et al., [@B52]). Physical manipulation of cells can rely on the use of strong mechanical forces, which is the case of microinjection or cell bombardment, or via the use of electric field pulses in the case of electroporation (Jaffe and Terasaki, [@B23]; Kikkert et al., [@B26]; Chen et al., [@B11]). Micromanipulators and micro-injecting systems, which usually require a dedicated microscope and a cell sorter, have been used with success to deliver and study proteins inside cells (König et al., [@B27]). Cell size plays an important role here, as in most cases micro-injection has succeeded with very large cells, such as *Xenopus laevis* oocytes (Thongwichian and Selenko, [@B51]). Cell bombardment has also been used successfully for DNA delivery into plant cells and tissues. However, success with this technique is limited by the amount of material that can be loaded on a particle and the ratio of surviving cells. For our study, we have opted for electroporation to deliver the protein of interest into plant cells. To develop our methodology (Figure [1](#F1){ref-type="fig"}), we have selected tobacco BY-2 cells (a plant cell line derived of *Nicotiana tabacum* cv Bright Yellow 2) as model system (Nagata et al., [@B35], [@B36]). This frequently used plant cell model grows in suspension, it is easy to maintain and these cells are amenable for *in vivo* imaging by wide-field and confocal laser microscopy. BY-2 cells can also be manipulated to test different protocols for protein delivery among the repertoire described above (Nagata et al., [@B36]). Finally, we used intrinsically disordered proteins (IDPs) that are generally good candidates for *in-cell* NMR studies due to the high peak intensity and good signal dispersion they exhibit at low concentration. Despite the small ranges of chemical shifts attained in spectra of IDPs, the dispersion of the NMR signals in IDPs is favored by their extremely long relaxation times leading to very narrow resonances, which makes the resolution of their peaks highly likely. ERD14 and ERD10 are such IDPs that belong to the class of dehydrins, a subgroup of Late Embryogenesis Abundant (LEA) proteins (Hundertmark and Hincha, [@B21]). LEA proteins are commonly found in higher plants (Battaglia et al., [@B3]), but they also occur in small rotifers (Chakrabortee et al., [@B10]). Originally discovered as cotton-seed abundant proteins, there is increasing evidence that they occur in different plant tissues and developmental stages (Candat et al., [@B8]). Dehydrins are of particular interest as they exhibited chaperone-like behavior while being intrinsically disordered (Kovacs et al., [@B28]). Experimental data about the actual mechanism of their action *in vivo* is missing. In this contribution we explore the suitability of these two proteins as molecular models for *in-cell* NMR in plant cells. Materials and methods {#s2} ===================== Cloning of LEA proteins ERD14 and ERD10 --------------------------------------- ### Cloning of erd14 and erd10 into gateway™ vectors For Red Fluorescence Protein (RFP) fusions cloning was done according to standard procedures with existing *erd14* and *erd10* fusion constructs (Kovacs et al., [@B29]) which were subsequently amplified and recombined with a pDONR221 vector (Invitrogen GatewayTM cloning kit, Cat. No. 12535 029) according to the manufacturer\'s instructions yielding both *erd14* and *erd10* pENTR vectors without stop codon suitable for recombination with pK7WGR2.0 (an in-frame C-terminal RFP GatewayTM vector (Karimi et al., [@B24]). Reaction between attL1/L2 sites on the pENTR*erd14* (or *erd10*) and attR1/R2 sites on pK7WGR2.0 was carried out using LR clonase at 25°C during 1 h in 10 mM Tris.HCl, 1 mM EDTA, pH = 8.0 (TE buffer). Then proteinase K was added and incubated for 10 min at 37°C. Material was transformed into *E. coli* TOP10 chemically competent cells in order to replicate plasmid DNA that was purified for subsequent work using a Wizard Wizard R Plus SV Minipreps DNA Purification System. (Promega Corporation, Cat. No. A7510). Finally, both *erd14-rfp* and *erd10-rfp* fusion constructs were confirmed correctly in frame within pK7WGR2.0 by DNA sequencing. ### Codon insertion for erd14 for fluorescent labeling via a cysteine residue Restriction-free cloning allowed the introduction of cysteine-186 at the end of ERD14 coding region. This mutant was designed using a pET22b-ERD14 template vector employed to over-express and purify ERD14 from *E. coli*. The primers introducing a C186 mutation into ERD14 (ERD14-C186) are: ERD14-C186fw 5′-GAGGAGAAGAAAGATAAAGAA[TGT]{.ul}TAAGCGGCCGCACTCGCGCACCAC-3′ ERD14-C186rv 3′-CCTCCTCTTCTTTCTATTTCTT[ACA]{.ul}ATTCGCCGGCGTGAGCGCGT-5′. The PCR reaction for this cloning resembles a site-directed mutagenesis reaction but in this case a new codon is inserted at position 186. The PCR reaction mixture comprised 125 ηg of ERD14-C186~fw~ primer, 125 ng of ERD14-C186~rv~ primer and 50 ng of pET22b-ERD14wt template DNA, mixed with 12.5 μL of HiFi KAPA Hot Start ReadyMix (KAPA Biosystems) in a final volume of 25 μL. The PCR programme included 3 min of denaturation at 95°C followed by 25 cycles of 20-s denaturation at 98°C, 15-s annealing at 65°C and 80-s extension at 72°C. Finally, another extension of 6 min at 72°C was included. The reaction mixture was cooled down and 1 uL of DpnI was added in order to digest methylated DNA for 90 min at 37°C. The obtained material was transformed into *E. coli* DH5α Ca^2+^-competent cells and selected on LB-agar plates supplemented with carbenicillin. Finally, successful mutation was confirmed by DNA sequencing. Generation of a stable BY-2 cell line expressing ERD14/10-RFP ------------------------------------------------------------- Generation of these cell lines involved several steps as described below. ### Transformation of agrobacterium tumefaciens with the appropriate DNA constructs of ERD14/10 An aliquot of electrocompetent *A. tumefaciens* LBA4404 cells (50 μL) was transferred into a pre-cooled electroporation cuvette (Bio-Rad Laboratories Inc., Cat No. 165-2086) and mixed with 2 μL of DNA (pK7WGR2.0-ERD14-RFP or pK7WGR2.0-ERD10-RFP) ensuring that no air bubbles were present. Cells were shocked on a Gene Pulser XcellTM electroporation apparatus (Bio-Rad Laboratories Inc.) using the following settings: 2.5 kV, 400 MΩ, 25 F. Next, 300 μL of Super Optimal broth Catabolite repression medium (SOC; 20.0 g/L tryptone, 5.0 g/L yeast extract, 0.5 g/L NaCl) was added to the cuvette and the total volume was transferred to a 12 mL culture tube containing additional 1 mL of SOC medium5. This culture was incubated for 2 h at 28°C, while shaking at 180 rpm. Next, cells were spread onto Yeast Extract Broth (YEB) plates containing gentamycin (20 μg/mL), rifampicin (25 μg/mL) and bacterial selection antibiotic for the gene of interest (spectinomycin for K7WGR2.0-ERD14-RFP and K7WGR2.0-ERD10-RFP fusions). Colonies were detected after 48 h at 28°C and finally transferred into YEB liquid medium (5.0 g/L beef extract, 1.0 g/L yeast extract, 5.0 g/L peptone, 5.0 g/L sucrose, 0.5 g/L MgCl2) supplemented with appropriate antibiotics and grown for 48 h at 28°C for further use or storage as glycerol stocks. ### BY-2 infection with positive clones of *A. tumefaciens* A fresh BY-2 culture was prepared after diluting 1 mL of a 1 week old culture into 39 mL of fresh BY-2 medium (4.302 g/L Murashige-Skoog Salts; Murashige and Skoog, [@B34]) (Cat. No. M0301-0050, DUCHEFA Biochemie B.V.), 0.2 g/L KH2PO4, 30.0 g/L sucrose, 0.08 μg/L 2,4-dichlorophenoxyacetic acid (auxin synthetic analog), 1 μg/L Thiamine and 0.1 mg/L myo-inositol, equilibrated until pH = 5.8 using KOH). Upon incubation for 72 h, 3.7 mL of the fresh dilution was transferred into a Petri dish without antibiotics and mixed with 300 μL of transformed *A. tumefaciens*. The mix was placed for incubation for 48 h at 25°C, without shaking. The content of each transfection dish was spread onto BY-2 solid-medium plates supplemented with vancomycin, carbenicillin and the plant selection antibiotic (kanamycin in the case of pK7RWG2.0-ERD14 or pK7RWG2.0-ERD10). Positive calli were selected on these plates after 4 weeks at 25°C by confirming the expression of RFP using fluorescence microscopy. Positive calli were transferred to fresh plates every 3--4 weeks or to liquid BY-2 medium for imaging purposes. Generation of a stable cell line for overexpression of ERD14 and ERD10 in BY-2 and *A. thaliana* ------------------------------------------------------------------------------------------------ Cloning of *erd14/10* with an inducible promoter region (regulated by estrogen) was performed using the same pENTR constructs obtained above. The destination vector pDEST contains a region for promoter binding upon induction and together with the gene of interest it was integrated into BY-2 facilitated by *A. tumefaciens* infection as described above. The expression profile for either ERD14 or ERD10 in BY-2 and also in *Arabidopsis* cells growing in suspension was monitored for a period of 96 h and detected by Western Blotting. The labeling medium for incorporating 15N isotopes into BY-2 was prepared by replacing the corresponding inorganic salts (ammonium sulfate and ammonium nitrate) in normal Murashige-Skoog media (Murashige and Skoog, [@B34]) with the salts incorporating the ^15^N isotope. Transient transformation of ERD10/14 into *Nicotiana benthamiana* leaves ------------------------------------------------------------------------ For localization *in planta*, ERD14/10 was transiently transformed on *Nicotiana benthamiana* leaves. *A. tumefaciens* LBA4404 transformed with the *erd14-rfp* or *erd10-rfp* fusion gene was grown in 5 ml YEB medium, supplemented with the corresponding antibiotics at 28°C for 48 h shaking at 180 rpm. OD~600~ was measured and cultures were diluted to a final volume of 4 mL to an OD~600~ = 1.5 and spun down for 5 min at 4,000 g. The pellet was resuspended in 4 mL of infiltration buffer using conical-bottom 15 mL tubes (10 mM MgCl~2~, 10 mM MES buffer pH = 5.6 and 0.1 mM acetosyringone). Each tube was incubated for 2 h at room temperature in a tube rotator. Infection was performed by injecting 0.5 mL of the previously described mix on the underside of several *N. benthamiana* leaves using a syringe without needle (Li, [@B30]). Upon infection and incubation for 72 h, plants were investigated by confocal microscopy to verify RFP-fused protein production by cutting a section of a leave and placing it directly on a cover slip (section of 1 cm^2^ approx.). Production and purification of ERD14 ------------------------------------ *E. coli* BL21(DE3) cells were transformed with ERD14-C186 pDNA according to standard procedures and a positive colony was selected from an LB-agar plate supplemented with carbenicillin to prepare a pre-culture of 50 mL liquid NZYM medium (10.0 g/L NZ amine, 5.0 g/L NaCl, 5.0 g/L Bacto-yeast extract, 2.0 g/L of MgSO4.7 H2O) also supplemented with antibiotics that was grown overnight at 37°C while shaking at 180 rpm. The pre-culture was used to inoculate 2 L of NZYM contained in 2 independent baffled flasks each supplemented with 50 mg/mL carbenicillin, these culture flasks were incubated in an orbital shaker at 180 rpm and 37°C. After 2 h, the culture reached OD600 = 0.6; at this stage 500 μL of 1 mM isopropyl-β-D-1-thiogalactopyranoside (IPTG) was added and the cells continued shaking at the same speed and temperature for 4 h. The bacterial pellet was collected by centrifugation at 4,000 g and 4°C for 30 min and resuspended in 50 mL ice-cold lysis buffer (50 mM CAPS, 5 mM EDTA, 5 mM tris(2-carboxyethyl)phosphine (TCEP), 1 Roche complete inhibitor cocktail tablet at pH = 10). Cell lysis was performed on an ice-cold bath by sonication using a Sonics Vibra-Cell™ CV18 model ultrasonic processor (applying 60 power of the probe in 6 cycles of pulses, 15 s ON and 30 s OFF). The suspension was then cleared by centrifugation at 20,000 g for 45 min, and the supernatant was placed into a water bath at 100°C for 5 min. After boiling, the solution was cleared by centrifugation at 20,000 g for 45 min and the supernatant was filtered through a 0.5 μm filter before injection on a QFF HiTrap 5 mL column (GE Healthcare) for anionic exchange. Binding buffer (50 mM CAPS, 5 mM TCEP, pH = 10) was used to equilibrate the column and the supernatant was passed through the column at 2 mL/min flow using an AKTA Pure system. While the ultraviolet (UV) absorbance at 215 nm was used to monitor protein elution. Weakly bound proteins were washed out with 5 column volumes (CV) of 2% elution buffer (50 mM CAPS, 250 mM NaCl, 5 mM TCEP, pH = 10) before starting a linear gradient between 2 and 50% (using same elution buffer) without changing the flow. Fractions containing the protein of interest were identified by SDS-PAGE (detection with PageBlueTM protein staining solution, ThermoScientificTM, Cat. No. 24620) and concentrated using 20 mL spinning filters with a 3 kDa cut-off (Vivaspin 20, Sartorius AG). The concentrated sample containing ERD14-C186 and some minor contaminants was injected onto a Superdex75 16/100 HiLoad column (GE Healthcare) and the gel filtration buffer consisted of phosphate buffered saline (PBS) pH = 7.5 supplemented with 5 mM TCEP). Fractions were collected automatically based on the UV absorbance profile and SDS-PAGE allowed the identification of the protein of interest at the corresponding retention time. Selected fractions were concentrated above 150 μM for use in labeling reactions. Fluorescent labeling of ERD14 by Alexa Fluor 488 and Alexa Fluor 647 -------------------------------------------------------------------- Alexa Fluor® 488 C5 Maleimide (green) and Alexa Fluor® 647 C2 Maleimide Alexa Fluor 647 (red) were purchased as lyophilized salts from ThermoScientificTM, Cat.No.A10254 and Cat.No.A20347, respectively. A HiTrap Desalting 5 mL column (GE Healthcare) was used to remove TCEP from the protein sample, to which either Alexa488 maleimide or Alexa Fluor 647 maleimide reagents were immediately added. The reaction mixtures (5-fold excess of fluorescent dye compared to protein concentration) were placed on a thermal bath at 60°C for 1 h. Labeling of ERD14-C186 was confirmed by SDS-PAGE through the presence of a fluorescent band with the molecular size corresponding to an ERD14 monomer. An additional gel filtration step was performed to separate the labeled protein from disulfide-linked dimers and unreacted fluorophores. Electroporation experiments for protein delivery ------------------------------------------------ Electroporation experiments were performed on a Gene Pulser MXCellTM Electroporation System using 2 mm cuvettes (Bio-Rad Laboratories Inc., Cat No. 165-2086). A volume of 40 μL of non-modified BY-2 cells was placed inside the cuvette together with 10 μL of 250 μM ERD14-C186-Alexa488 in PBS. A broad range of conditions were tested in order to assess protein uptake and cell survivability. These conditions are summarized in Table [1](#T1){ref-type="table"} in the Results section. In every case, cells were electroporated in BY-2 medium and cargo protein ERD14-C186-Alexa488 or ERD14-C186-Alexa Fluor 647 was added in PBS. Cells were collected from the cuvette and resuspended in 100 μL of BY-2 medium. At this stage, cells were allowed to recover for different periods of time before imaging with either confocal or epifluorescence microscopy. ###### **Scouting of various electroporation conditions**. **Experimental condition** ------------- ---------------------------- -------- -------- -------- -------- -------- -------- -------- -------- ---------- -------- ---------- Exponential 150 V 200 V 250 V 300 V 350 V 450 V 250 V 250 V 250 V 250 V 250 V 250 V decay 350 μF 350 μF 350 μF 350 μF 350 μF 350 μF 200 μF 250 μF 350 μF 500 μF 750 μF 1,000 μF 400 V 250 V 150 V 100 V 50 V 50 V 150 V 150 V 150 V 150 V 50 V 50 V 200 μF 200 μF 200 μF 200 μF 200 μF 100 μF 250 μF 500 μF 750 μF 1,000 μF 500 μF 750 μF Square 150 V 200 V 250 V 300 V 350 V 450 V 250 V 250 V 250 V 250 V 250 V 250 V wave 20 ms 20 ms 20 ms 20 ms 20 ms 20 ms 5 ms 10 ms 15 ms 20 ms 25 ms 30 ms *The electric pulse is a wave with an associated voltage, capacitance and time. There are two different waveforms, namely square wave or exponential decay. These experimental conditions were tested to load BY-2 cells with externally supplied ERD14-C186-Alexa Fluor 647 as reporter protein. We monitored the cell integrity and cell viability as a biological read-out through microscopic observations*. Confocal and epifluorescence microscopy --------------------------------------- Cell viability was inspected by mixing electroporated cells with either lipophilic styryl dye (FMTM Thermo Fisher Scientific Inc., Cat. No. F35355) or propidium iodide (PI, Thermo Fisher Scientific Inc., Cat.No.P3566) directly on the cover slide. For cell culture monitoring and calli selection, an inverted epifluorescence microscope (ZEISS Axiovert 135M) was used. Excitation is done with an HBO 100 Hg lamp. Images were captured with a ZEISS Axiocam using the software package axiovision 4.8.2 SP2 (06-2012). Using single GFP filter and RsGFP ex 465-490 nm; em 500--520 nm. Lenses used are either plan-apochromat 20x dry na = 0.75 or plan apochromat 40x dry na = 0.95. In confocal imaging a LSM5 Exciter upright microscope using an AxioImager Z1 microscopy stand was used. Different lenses were used and are described as follows: plan-apochromat 20x dry na = 0.8, C-apochromat 63x water corrected na = 1.2, C-apochromat 40x water corrected na = 1.2. In every case imaging using dual GFP/RFP scanning was performed using software package ZEN2009. Solution state NMR experiments ------------------------------ NMR experiments (SOFAST HN-HSQC; heteronuclear single quantum coherence) (Schanda et al., [@B42]) were performed on a 750 MHz Bruker250 Avance spectrometer equipped with a cryogenically cooled triple resonance 1H{13C/15N} TCI probe. Samples were loaded into 3 mm Shigemi tubes and supplemented with 10% D2O. Spectra of samples in buffer were collected using 128 scans and 1,024 increments while samples in extract were collected using 256 scans and 2,048 increments. Samples recorded either for reference (in buffer) or in cell lysate were 25 μM ERD14 or ERD10 in 50 mM MES buffer; pH = 6.5. Crude Extracts (cell lysates) were obtained by sonicating 4 days old BY-2 cells on an ice-cold bath using a Sonics Vibra-CellTM CV18 model ultrasonic processor (applying 60 power of the probe in 2 cycles of pulses, 10 s ON and 20 s OFF). Cells were suspended in 50 mM MES buffer in presence of 10 mM DTT, supplemented by cOmplete protease inhibitor and PhosSTOP phosphatase inhibitor. Extracts were not centrifuged or cleared of debris. The total protein concentration was determined by the adjusted Bradford assay for NanoDropTM 2000c according to the technical note from the manufacturer. Western blotting ---------------- Western blotting was performed using secondary antibodies at two different wavelengths, namely IRDye 680LT Goat anti-Mouse IgG for 680 nm and IRDye 800CW Goat Anti-Rabbit IgG for 800 nm. Imaging of developed membranes was done using dual-wavelength scanning using a LiCor Odessey® CLx. The primary antibody against ERD10, affinity purified anti-K segment antibody (Agrisera, Cat. No. AS07-206A). Results {#s3} ======= Scientific questions to be addressed by *in-cell* NMR ----------------------------------------------------- It is reported that both endogenous ERD14 and ERD10 are over-expressed *in planta* under abiotic stress, like low temperature and drought (Hundertmark and Hincha, [@B21]). The quest for the structural and functional characterization of ERD14 relied on *in vitro* techniques, including NMR, and revealed that the protein is intrinsically disordered (Kovacs et al., [@B29]; Szalainé Ágoston et al., [@B48]). The NMR chemical shift assignment for ERD14 is already reported (BMRB entry 26636) and further calculations supported the idea that stable secondary-structure elements are not present (Close, [@B12]; Szalainé Ágoston et al., [@B48]). The NMR assignment of ERD10 was recently made publicly available by our group (Cedeño et al., [@B9]; BMRB entry 26949). The availability of the chemical shift assignment for both proteins is a prerequisite when evaluating the feasibility to conduct *in-cell* NMR experiments (Figure [1](#F1){ref-type="fig"}). With such an *in-cell* NMR approach, it becomes possible to probe if the *in vitro* determined structural and dynamical properties actually reflect the cellular *in vivo* conformation (Selenko and Wagner, [@B43]). Other interesting questions relate to the identity and structural influence of binding partners, the effect of post-translational modifications on protein conformation and the ways the proteins respond structurally to cellular processes like cell differentiation. To address the high resolution conformational behavior of the ERD proteins in response to external environmental changes (e.g., abiotic stress), the development of an *in-cell* NMR methodology is indispensable. *In vitro* NMR under molecular crowding conditions -------------------------------------------------- Based on the already available NMR data, we opted to use the same experimental set-up, including the slightly acidic pH of 6.5 that limits the amide-proton exchange (Szalainé Ágoston et al., [@B48]). When we studied both the purified ERD14 and ERD10 by *in vitro* NMR, regardless of their size, we observed a relatively low overlap of the cross-peaks in their ^15^N-^1^H-HSQC spectra (see Figures [2A](#F2){ref-type="fig"}, [3](#F3){ref-type="fig"}). Besides these NMR experiments in a defined buffer, we also collected^15^N-^1^H-HSQC spectra of purified recombinant ^15^N-labeled ERD14 and ERD10 that were added to a cell extract of BY-2 to mimic the intracellular environment (Figures [2B](#F2){ref-type="fig"}, [3](#F3){ref-type="fig"}). Although the total protein concentration in cell extracts is far from that existing in the cytoplasm, this approach provides a reasonable and often applied approximation of cellular conditions, interaction partners and metabolites, interfering with the protein. Macromolecular concentrations closer to those prevailing in the cell can be achieved by the application of polymers, such as polethylene-glycol or dextran, which does not capture specific interactions with the protein, though. Overall, we could visualize some effects of crowding, as evidenced by line broadening for both ERD protein samples. These spectra showed little changes due to local changes of pH along the sequence as can be evidenced by the chemical shift changes experienced by histidines, which are particularly sensitive to this. It is conceivable that ERD14 and ERD10 directly interact with other cellular proteins and/or components in the crude lysates, but the data shown in this contribution do not allow us to detect such interactions (Cedeño et al., [@B9]). Additionally we did not observe signs in the HSQC spectra that would indicate degradation of the ^15^N-labeled proteins during the sample preparation and NMR data acquisition. From these experiments we concluded that the NMR data of the ERD proteins under molecular crowding conditions that resemble the plant cell interior contain sufficient information to initiate further *in-cell* NMR studies. {#F2} {#F3} Intracellular localization of dehydrins ERD14 and ERD10 ------------------------------------------------------- To set the stage for monitoring intracellular delivery of ERD proteins, we set-out to explore the localization of ERD proteins in plant cells. Since the exact localization of ERD14 and ERD10 inside the plant cell remains elusive and debated (Close, [@B12]; Puhakainen et al., [@B41]), we created *erd10* and *erd14* gene-fusions with the Red Fluorescent Protein (RFP) that were introduced into BY-2 cells using *A. tumefaciens* transformants. Calli that exhibited the red fluorescence of ERD14-RFP or ERD10-RFP were selected under the epifluorescent microscope and then transferred to BY-2 liquid media in order to generate stable cell lines for each protein. As clearly seen in Figures [4A,D](#F4){ref-type="fig"}, ERD proteins exhibit a cytosolic localization under our experimental conditions. Although our observation contrasts with previous reports (on nuclear or chloroplast association), it is in agreement with a recently published independent study (Candat et al., [@B8]). The same protein fusionconstructs were used for transient expression on *N. benthamiana* leaves together with GFP as control. As can be seen in Figures [4B,E](#F4){ref-type="fig"}, the fusionproteins remain in the cytosolic strands, while co-expressed GFP migrates into the nucleus (Figure [4G](#F4){ref-type="fig"}). ERD10 was also transiently expressed on *N. benthamiana* leaves and showed similar localization (Figures [4C,F](#F4){ref-type="fig"}). GFP co-expression controls for ERD10 are shown in Figure [4H](#F4){ref-type="fig"}. We found that ERD14 does not change its localization upon induced cold stress and osmotic stress (that mimics dehydration). This suggests that the protective effect of this protein (Battaglia et al., [@B3]; Hundertmark and Hincha, [@B21]) is compatible with its cytosolic localization, as simulated on leaves of *N. benthamiana* (Figure [5](#F5){ref-type="fig"}, freezing is shown in Figure [5B](#F5){ref-type="fig"}, osmotic stress in Figure [5C](#F5){ref-type="fig"}). Altogether, these results suggest that ERD14 is a good candidate for *in-cell* NMR and that the protein is located in the cytoplasm. {#F4} {#F5} Electroporation for intracellular delivery of dehydrins ------------------------------------------------------- Because we could localize intracellular ERD14-RFP upon induced overexpression with success, and because of the quality of the NMR data of ERD14 under molecular crowding conditions, we decided to introduce purified and recombinantly produced 15N-labeled ERD14 inside plant cells using electroporation for *in-cell* NMR studies. We opted specifically for electroporation because DNA and small amounts of proteins can be delivered into plant cells by electroporation, it is possible to prepare samples containing 10^6^ cells in a matter of minutes/hours, and commercially available electroporators offer flexibility in varying the voltage and current as well as pulse timing and recovery rates (Yamano et al., [@B56]). These latter physical parameters determine efficient delivery of the material as well as survivability of the cells after the electric shock. Yet, our approach essentially differs from a transformation with DNA by electroporation strategy in some fundamental aspects: for transformation the main goal is to deliver a limited amount of DNA into a number of cells that must survive and get selected prior to further propagation. In contrast, for *in-cell* NMR, most of the cells should survive the procedure and maintain their (sub) cellular integrity, while most of the cells have to take-up a sufficiently large quantity of isotope-labeled protein material. For this purpose, a great variety of electroporation conditions were tested to identify the mildest one with as small as possible effect on cell viability. In the first instance, we used the fluorescent derivative ERD14- C186-Alexa Fluor 647 to use microscopic visualization to verify if the protein was successfully delivered into the cells. The variables that we scouted in order to explore the delivery/survivability space, are: voltage, capacitance and pulse duration (Table [1](#T1){ref-type="table"}). According to the technical notes of the electroporation device, exponential decay represents the best choice in terms of maximizing cell viability during and after the experiment. Only in a few cases we could detect successful delivery of fluorescently labeled ERD14 inside BY-2 cells (Figure [6](#F6){ref-type="fig"}). However, most electroporation conditions compromised cell viability in such a way that it is not compatible with any downstream *in vivo* experiments. As can be seen in Figure [6](#F6){ref-type="fig"}, both protein localization and cell morphology strongly indicate non-physiological phenomena. In general, the osmotic balance of cells appears to be heavily disrupted during electroporation (Figure [6](#F6){ref-type="fig"}). ERD14-C186-Alexa Fluor 647 is distributed homogeneously in the cytoplasm of some cells (Figures [6E--G](#F6){ref-type="fig"}, red color) showing that most of the inner architecture of the cells is disrupted. Strikingly, as also seen in Figures [6F,G](#F6){ref-type="fig"}, ERD14-C186-Alexa Fluor 647 is not only distributed across the cytoplasm but it can also be observed inside nuclei. There is no evidence of the association of ERD14-C186-Alexa Fluor 647 with membranes. {#F6} These adverse effects do not improve during the recovery period when membrane pores opened by the electric field are supposed to close; it seems that ERD14 can diffuse freely inside cells once pores are opened. Apparently, these set of conditions can break the tonoplast (or vacuolar membrane) whereby the cellular osmotic balance is altered in such a way that is not possible to satisfy both efficient protein delivery and cell viability. Indeed, vacuoles occupy 90--95% of the total internal volume (Marty, [@B32]) and the pH in the lumen reaches acidic values as low as 5.5 (Barbier-Brygoo et al., [@B1]). Disruption of the tonoplast is very likely the reason of the uniform distribution of ERD14-C186-Alexa Fluor 647 inside the cells without the preservation of cytosolic strands (Figures [5E--G](#F5){ref-type="fig"}); this is in striking contrast with the images obtained for healthy cells (Figure [4D](#F4){ref-type="fig"}). It must be highlighted that the internal morphology shown in this figure is not physiologically relevant as this is accompanied with FM penetration and nuclear staining (Figures [6B--D](#F6){ref-type="fig"}). FM does not penetrate healthy cells before electroporation (Figure [6A](#F6){ref-type="fig"}), therefore the conclusion is that most of the cells shown in Figures [6B--D](#F6){ref-type="fig"} are highly compromised, either because of disruption of vacuoles, or due to the lack of fast and efficient mechanisms repairing plasma membrane pores through cytoskeleton remodeling (Zhang et al., [@B57]). The observed state as shown in Figure [6](#F6){ref-type="fig"} is representative of all the conditions of electroporation tested (Table [1](#T1){ref-type="table"}) as there are no differences or trends observed that can be attributed to changes in voltage or intensity. This observed vacuole disruption by electroporation also offers another interpretation concerning our *in vitro* crowding experiments (Figures [2](#F2){ref-type="fig"}, [3](#F3){ref-type="fig"}). These experiments did not show substantial structural changes in cell lysates as compared to a simple *in vitro* buffer system. Cytosolic strands are the regions (Figure [4](#F4){ref-type="fig"}) where both ERD14 and ERD10 are localized and where total, non-homogeneous (Luby-Phelps, [@B31]) protein concentration is expected to be at least 200 mg/mL (Theillet et al., [@B50]). The large volume of water contained in vacuoles (Marty, [@B32]) dilutes the cytosolic strands up to 20 times, since cell lysis disrupts the plasma membrane as well as tonoplasts. In Figures [2](#F2){ref-type="fig"}, [3](#F3){ref-type="fig"}, the crude cell lysate contains 5--7 mg/mL of protein, this low total concentration of proteins is a clear evidence of the dilution effect introduced by of vacuoles. Although the application of cell lysates/extracts is a frequently used approach to approximate or mimic intracellular effects, this might not apply for physical/structural studies where an ideal crowding experiment (such as for plant proteins) needs to be performed at high total protein concentrations like those exhibited inside cells. This offers a pivotal argument to develop a true *in-cell* NMR strategy. Induced overexpression of ERD14 in plant cells ---------------------------------------------- While electroporation-mediated introduction of ERD14 did not yield the desired outcome for further *in-cell* NMR studies, we resorted to controlled overexpression of proteins inside plant cells as reported in the past (Hellwig et al., [@B17]; Ohki et al., [@B40]). In order to enable the production of 15N-labeled ERD proteins inside either BY-2 or *A. thaliana* cells in suspension, we generated transgenic lines of A. thaliana and BY-2 in which *erd14* and *erd10* genes were controlled by an inducible estrogen promoter. Unfortunately, gene transfection of *erd14* did not yield any viable culture and therefore positive *erd14* cell line could not be obtained in A. thaliana. In the case of BY-2, calli and cell suspension culture were obtained for transformants of both genes (ERD14 and 10), however, only ERD10 gets overexpressed upon the addition of estradiol, as detected by immunoblotting after 40 h of induction (Figure [7](#F7){ref-type="fig"}). {#F7} An inducible protein expression system can offer a major advantage over protein delivery via electroporation, because cells are constantly growing under normal culture conditions. Therefore, cell viability is much less affected during expression of ERD10 and thus cultures can grow normally during the progression of the experiment (Figure [7](#F7){ref-type="fig"}). However, overexpression of ERD10 in BY-2 cells imposes a set of restrictions, primarily the inherent background generated by other endogenous proteins under isotope-labeling conditions. In this scenario, ERD10 needs to be 15N labeled (at least) in order to generate a useful sample for *in-cell* NMR. As exemplified by the *E. coli* overexpression systems, translation needs to be fast and strong to obtain a good signal-to-noise ratio in an *in-cell* ^1^H-^15^N-HSQC experiment (Serber et al., [@B44]). Liquid media for plant cultures normally contain a set of inorganic salts (micronutrients) in low concentrations and a large supply of carbon in the form of sucrose (besides vitamins and the buffer, known as Murashige-Skoog media). Vitamins like myo-inositol, thiamine and the hormone auxin determines the healthy growth of BY-2 cells, however, the mixture of vitamins and hormones can differ from one species to the other. The source of nitrogen contained in the original Murashige-Skoog medium is changed for *in-cell* NMR samples. In this case they were replaced by 15N-containing salts to make ERD10 visible for NMR upon estradiol-induced protein expression. The first metabolic step once ammonium and nitrate reaches the cytosol is its absorption in the form of glutamine, hence this amino acid is highly abundant in a nitrogen-rich environment (e.g., *in vitro* cultures) (Stitt et al., [@B47]; Masclaux-Daubresse et al., [@B33]). As shown in Figure [3](#F3){ref-type="fig"} the *in vitro* NMR spectra of ERD10 in buffer and in a cellular extract depict the lower limit of detection of a ^1^H-^15^N-HSQC with an acceptable signal-to-noise ratio given the low concentration and pH differences. Only qualitative information about ERD10 (fingerprint of disorder, post-translational modifications, degradation, etc.) can be extracted from these spectra (Figure [3](#F3){ref-type="fig"}) although lower amounts of protein would not yield easily interpretable data. These experiments show that about 12.5 μM is the lower limit of ERD10 at which NMR signals can be generated and collected with acceptable good signal-to-noise ratio, which sets our practical limit for the feasibility of the *in-cell* NMR experiments This was not the case in our overexpression experiment: the levels of ERD10 in BY-2 cell cultures after 48 h of induction can rise up to 2.5 μM as semi-quantitatively measured by Western Blot (Figure [7](#F7){ref-type="fig"}). When we compare the spectrum of a highly concentrated sample of ERD10 overlapped on the spectrum of BY-2 cells containing 2.5 μM ERD10 (Figure [8A](#F8){ref-type="fig"}), it is clear that sharp and intense signals can be collected from a highly concentrated sample, however, the signals coming from labeled proteins inside BY-2 are not detectable. Instead, two very intense and broad lines are detected in the region of glutamine side chains (Figure [8B](#F8){ref-type="fig"}), which suggests that BY-2 cells are actively metabolizing nitrate and incorporating ^15^N, although the translation products of other proteinaceous species including ERD10 are not detectable. This observation is in accordance with quantification of protein levels by Western blot. {#F8} Discussion {#s4} ========== As the *modus operandi* and functional behavior of intrinsically disordered ERD14 and ERD10 in protection against abiotic stress remain elusive, we set out to develop an *in-cell* NMR strategy with plant cells. First, we established by fluorescent fusion proteins that under our experimental conditions ERD14 and ERD10 are cytosolic proteins that do not change their intracellular localization under abiotic stress conditions (cold and osmotic stress). Next, by simulating the cellular crowding using cell extracts we evaluated that ERD14 and ERD10 are compatible with an NMR read-out to study molecular events inside cells (Smith et al., [@B46]). Whereas the use of cellular extracts has been considered as a reasonable mimic for intracellular conditions and protein behavior, we observed that in the case of plant cell extracts the molecular crowding and excluded volume effect are limited and inappropriate, due to a significant dilution of the cytosol by the vacuolar content. Thus, *in-cell* NMR actually creates a particular opportunity to understand protein structure and activity from a unique perspective. There are several considerations and strategies for protein delivery for the purpose of *in-cell* NMR (see Figure [1](#F1){ref-type="fig"}) and the motivations and methodological approaches can be manifold. Further, there are diverse controls and additional experiments to carry out for exploiting the full potential of this technology. However, major barriers imposed by physiology of the plant cell can restrict its applicability in many ways. There are limitations to the scenario in which *in-cell* NMR is feasible: cells must survive the treatment prior to the insertion in an NMR tube, the physiology should not be altered in such a way that cellular morphology is dramatically affected and the NMR spectra of proteins (isotopically labeled) should be collected at intracellular concentrations close to the physiological ones (in un-treated cells). Electroporation was selected as our preferred delivery method during the initial stages of this study, because pore formation evoked by pulsating electric fields yielded good results for the transformation of both prokaryotic and eukaryotic cells (Kato et al., [@B25]; Boukany et al., [@B7]; Wang et al., [@B53]; Theillet et al., [@B49]). Microinjection cannot be applied because the size and morphology of tobacco cells (BY-2) represented a major challenge (due to the size of its vacuoles and also because microinjection is mostly only useful with larger cells (e.g., oocytes). Vacuoles are dynamic and mechanically resistant in order to perform their function without a loss of integrity, as they are responsible for maintaining the osmotic pressure, pH and membrane potential of the cell. Strong depolarizations and dramatic changes in pH or osmolarity beyond the vacuolar capability are life-threatening for the cell. Basically, electroporation is a strong depolarization able to generate a transient permeabilization of the plasma membrane as has been evidenced in gene transfection (Weaver, [@B55]; Ho and Mittal, [@B19]). However, in light of the current evidence, electroporation also generates a transient disruption of tonoplasts (that are enriched in acidic phospholipids and in sterols (Zhang et al., [@B57]), which likely explains why cells become compromised with their internal cell morphology heavily disrupted. Despite varying electroporation conditions (Table [1](#T1){ref-type="table"}), ERD14 could only enter cells in which survivability was compromised, with large amounts of cells not showing signs of either damage or protein delivery. This is likely the consequence of a strong protective effect of the plant cell wall, suggesting that internal membranes can no longer survive the electric shock once the protective cell wall is brought down in this time scale (micro-to-millisecond of electric pulse) (Batista Napotnik et al., [@B2]). Altogether, these results show limitations of electroporating plant cells with the particular aim of performing *in-cell* NMR. To overcome these obstacles, we also approached the problem by induced overexpression under isotope- labeling conditions. Unfortunately, the levels of protein expression are not compatible with the low sensitivity of the state-of-the-art technique. For each protein the optimal expression conditions (i.e., NMR signal intensity as compared to the background signals that arise from non-specific isotopic enrichment during the expression of endogenous proteins) should be screened and quantitative Western Blotting of cell lysates at different expression times or with different promoters is a reliable technique to accomplish this (Figure [1](#F1){ref-type="fig"}). This strategy also opens the possibility to co-express partner proteins. The choice of cell line should also be considered carefully: in this work we used BY-2 cells and *Arabidopsis* cells in suspension. An interesting alternative that would be compatible with *in-cell* NMR would be the unicellular green algae *Chlamydomonas*, but this would not be the primary choice to study phenomena related to dehydration stress (Yamano et al., [@B56]). For the future, other methodologies should be considered for the gentle manipulation of cells in order to introduce isotope-labeled reporter proteins. These methodologies can exploit the machinery of exogenous agents (bacteria and viruses) in order to penetrate and deliver products selectively inside the cytoplasm. For example, bacterial toxins like streptolysin can be used for the purpose of *in-cell* NMR (Ogino et al., [@B39]), however, to the best of our knowledge, there are no reports of its use on plants. Some bacterial pathogens actively translocate their virulence factors using a molecular needle that is able to penetrate membranes, e.g., the type III and type IV secretion system (t3ss, Galan et al., [@B15]; and t4ss, Fronzes et al., [@B14]). Even though this is a widespread mechanism for bacterial protein delivery in nature, there are limitations for its use toward *in-cell* NMR. Basically, proteins delivered using t3ss are typically of low abundance and the pathogens exploiting this invasion mechanism have not evolved for larger loads in terms of concentration. *A. tumefaciens* infects plants using t4ss and this is the key mechanism for gene introduction. In terms of synthetic biology and engineering, t3ss and t4ss are interesting tools but for *in-cell* NMR, it needs to be coupled with an efficient inducible overexpression system, besides the proper translocation signals on bacteria. The whole approach represents a largely unexplored field of research that is expected to bring exciting results in the future. An interesting alternative for translocation and delivery of cargo proteins of even larger size into cells are cell-penetrating peptides (CPPs). These small peptides were already successfully exploited by NMR spectroscopists to explore the interior of mammalian cells (Inomata et al., [@B22]). CPPs, however, are not yet used at large scale for in-cell NMR, most likely due to their tendency to stay trapped in endosomes dragging the cargo protein with them (Nischan et al., [@B38]). Nonetheless, CPPs constitute an active field of research that is of particular interest for the pharmaceutical industry and they are intensely studied from a technological and methodological point-of-view. In plants, they proved to be useful for delivering proteins into the cytosol as shown by Herce et al. ([@B18]) and more recently by Ng et al. ([@B37]). Therefore, these CPPs are a very promising research avenue that hold promise to be exploited for studying ERD14 or ERD10 structure and function inside living plant cells through in-cell NMR. Author contributions {#s5} ==================== CC and PT contributed to the conception and design of the work. CC was responsible for the acquisition of the data presented in this work. CC and KP performed the analysis and interpretation of the data. All authors contributed to the writing of the paper. Funding {#s6} ======= CC was supported by the Marie Curie Initial Training Network project 264257 (IDPbyNMR). PT was supported by the Odysseus grant G.0029.12 from Research Foundation Flanders (FWO). KP is the recipient of a FWO long-term postdoctoral fellowship (1218713). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. 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. CC is grateful to the research groups of Daniel Vandamme (VIB-Ugent) and Philipp Selenko (FMP-Berlin) for their valuable input in this project. Special recognition goes to Andres Binolfi (FMP-Berlin, Max Planck Laboratory of Rosario---Argentina). The authors thank Denes Kovacs for critical discussions. [^1]: Edited by: Dominique Job, Centre National de la Recherche Scientifique (CNRS), France [^2]: Reviewed by: Karine Gallardo, National Institute for Agronomic Research (INRA), France; Alfonso De Simone, Imperial College London, UK [^3]: This article was submitted to Plant Proteomics, a section of the journal Frontiers in Plant Science

1. Introduction {#sec1-ijerph-16-00241} =============== In Germany, coastal areas of the Northern and Baltic Sea offer possibilities for the construction of offshore wind parks. The growth of the German offshore wind industry as part of the green energy revolution is shown by recent figures: in the first six months of 2018, 62 new wind energy convectors were taken into operation \[[@B1-ijerph-16-00241]\]. All in all, there were 20 German offshore wind parks by the end of June 2018, generating about 5.300 megawatt (MW) \[[@B2-ijerph-16-00241]\]. This development goes along with an increase in employment figures in the branch. Recent estimates indicate that approximately 27,200 workers, most of them men, are currently employed in the German offshore wind industry along the value chain \[[@B3-ijerph-16-00241]\]. Offshore wind jobs are associated with many hardships, as well as physical and psychological demands for their employees \[[@B4-ijerph-16-00241],[@B5-ijerph-16-00241],[@B6-ijerph-16-00241],[@B7-ijerph-16-00241]\]. A key element of working in the German offshore wind branch is the 14/14 work schedule that is applied: offshore wind workers usually spend 14 days offshore, where they work in 12-h-shifts and live on platforms, ships, or neighboring islands. These periods of leave alternate with 14 days of spare time the workers spend at home. From a psychosocial viewpoint, this schedule implies that the workers are confronted with a phase of absolute separation from their families and social environment, followed by an intensive free time period spent together. The 14/14 work schedule is unique and qualitatively different from other work patterns, e.g., work schedules of shift workers or employees with on-call-work. While the latter jobs still allow off-duty time at home each day, offshore workers spend entire weeks away \[[@B8-ijerph-16-00241]\]. The 14/14 schedule of workers in the offshore wind industry also deviates considerably from the schedules of other commute workers: the periods of absence of offshore workers are longer than those of commuters coming home for the weekend, but shorter than those of, e.g., seafarers in the maritime sector spending months away from home. It has recently been noted that the recurrent absences from home represent a job demand for many offshore wind workers, leading to difficulties in reconciling offshore work and family life \[[@B5-ijerph-16-00241],[@B6-ijerph-16-00241],[@B9-ijerph-16-00241]\]. However, the specific lifestyle of the offshore personnel does not only influence the workers themselves, but may also affect their partners and families at home who must deal with the repeated partings, reunions, and phases of intermittent absence. For example, female partners of offshore workers may find it challenging to manage all domestic and parenting tasks alone, to deal with the disruption to shared family activities, and to establish routines in their daily life. Likewise, there could be certain features of the 14/14 schedule that may potentially enhance family life, e.g., the chance to continuously spend intensive time together on a regular basis. The notion of women experiencing difficulties in dealing with the intermittent absences of their partners is supported by studies carried out in branches with extended periods away from home, e.g., the offshore oil and gas branch and mining industries \[[@B8-ijerph-16-00241],[@B10-ijerph-16-00241],[@B11-ijerph-16-00241],[@B12-ijerph-16-00241]\]. Research from the offshore oil and gas sector has indicated that partners staying at home experience various demands on their own \[[@B8-ijerph-16-00241]\], being condensed within the statement of "living two lives". Moreover, early research on the intermittent presence and absence of offshore oil and gas workers has differentiated between three perceived phases for the couples: her single life at home, his offshore life, and their common life together \[[@B11-ijerph-16-00241]\]. Specific demands for the women, as revealed in previous studies, consisted of experienced negative emotions, as well as difficulties related to role allocation and family structures \[[@B8-ijerph-16-00241],[@B13-ijerph-16-00241]\]. Similar to offshore branches, mining industries are also characterized by long distance commuting and specific rosters. They are often referred to as fly-in fly-out (FIFO) branches. The presence of the partner at home is also rather frequent and extended in the FIFO branches in comparison to other industries with prolonged absences. It has been proposed that the living situation of FIFO couples may be described as a cycle with the stages reunion, time commonly spend together, parting, and time alone \[[@B12-ijerph-16-00241]\]. In this cycle, every stage was associated with certain demands, adjustments, and negotiations to be made by the couples, and with a range of perceived emotions attached to the transitions. Although the offshore lifestyle was found to be related to specific demands, an early study among so-called offshore wives did not find differences between the mental and physical health of offshore and onshore wives \[[@B14-ijerph-16-00241]\]. Likewise, recent research results suggest a healthy functioning range for psychological wellbeing, relationship satisfaction, and perceptions of family function in FIFO workers and their partners \[[@B15-ijerph-16-00241]\]. Further research on the health of FIFO families showed a generally high level of family cohesion, healthy flexibility, and a general contentment with regard to family satisfaction and communication \[[@B10-ijerph-16-00241]\]. The ability to deal with the challenges encountered by partners of commute workers may partly depend on the coping strategies being applied. According to the transactional approach of Folkman and Lazarus \[[@B16-ijerph-16-00241],[@B17-ijerph-16-00241]\], coping can be defined as the cognitive and behavioral efforts made to master, tolerate, or reduce external and internal demands, as well as conflicts among them. Coping efforts may either concern the management of the stress-inducing problem (problem-focused coping) or the regulation of emotions or distresses (emotion-focused coping) related to the situation encountered \[[@B16-ijerph-16-00241],[@B17-ijerph-16-00241]\]. Research among FIFO families has, for example, indicated the existence of different coping strategies of the couples to deal with the difficulties they have encountered. As described in an interview study, such strategies encompassed maintaining open communication, parting on good terms and without major arguments or conflicts, and generally maintaining a positive attitude towards the situation \[[@B12-ijerph-16-00241]\]. Despite existing research studies in FIFO and offshore oil and gas industries, there is a lack of evidence regarding the situation of offshore wind couples living the specific 14/14 schedule. Findings for couples in the oil and gas and FIFO branches are not directly applicable to the situation of couples in the offshore wind industry, since the work schedules in the branches can differ considerably, and employees in the afore mentioned branches usually do not follow the 14/14 work pattern as applied in the German offshore wind branch. Moreover, there are further differences between the branches, e.g., regarding specific regulations, work areas, and work tasks \[[@B4-ijerph-16-00241],[@B6-ijerph-16-00241],[@B18-ijerph-16-00241]\]. In addition, much of the existing research on offshore couples dealing with intermittent absences was conducted in the 80's, 90's, and early 00's, indicating a need for an updated view. Within the last decades, profound organizational, technological, and sociological changes have taken place which pertain to offshore industries \[[@B8-ijerph-16-00241]\]. For example, the extension of means of communication provides couples with further possibilities to keep in regular contact. Changes in society have also occurred in terms of women's qualifications and employment rates, making it more likely that women of offshore workers are engaged in paid work. Finally, as a corollary of social changes, women's role models and aspirations have evolved: nowadays, women tend to have stronger expectations as to their husband's involvement in childcare and housekeeping \[[@B8-ijerph-16-00241]\]. Such recent changes could have either beneficial or detrimental impacts on the living situation of couples dealing with intermittent absences, which has not yet been empirically explored. Since a further increase in the offshore wind workforce is to be expected, more couples and families will have to deal with the particularities of living the 14/14 schedule. Therefore, it is crucial to conduct up-to-date research and generate new knowledge on the specific features of this lifestyle. This study intended to offer a contemporary view on the psychosocial adaptation of offshore wind families against the background of the modern society and working world. The aim of our study was to examine how women of male offshore wind workers perceive the specific features of living the 14/14 schedule. In adherence to the approach by Parkes and colleagues \[[@B8-ijerph-16-00241]\], we focused on the different phases of women's daily life, including aspects of being home alone and time spent together as a couple. Moreover, our purpose was to investigate the coping strategies of the women and the couples to deal with the specific living situation. Furthermore, we aimed to explore the women's perceptions of the reconciliation of offshore work and partnership/family life. We proposed the following research questions:What are the advantages and disadvantages of living the 14/14 schedule as perceived by the women of offshore wind workers, particularly regarding the following phases:(1)life without offshore partner;(2)life together as a couple/family;(3)transition phases (reunion/parting)?What coping strategies are employed by the women and couples to deal with the specific features of living the 14/14 schedule?How do the women judge the reconciliation of offshore work and partnership/family life, including aspects of organizational support? 2. Materials and Methods {#sec2-ijerph-16-00241} ======================== 2.1. Study Design and Participants {#sec2dot1-ijerph-16-00241} ---------------------------------- We conducted 14 semi-structured telephone interviews with female partners of German offshore wind workers from January to March 2017. The interviews were carried out by two female psychologists working as researchers in occupational health psychology at the time of the study. We applied purposeful sampling and recruited interviewees of different ages, with and without children, and with offshore partners working in different companies. Participants were eligible if they were female, fluent in the German language, and at least 18 years old. Moreover, the females' partners had to have worked for at least six months in the offshore wind industry, and had to have experiences with the 14/14 work schedule. To recruit suitable participants, we sent invitation mails, emails, and leaflets to German offshore wind workers from different companies who had already participated in a prior interview study conducted at the research institute \[[@B5-ijerph-16-00241],[@B6-ijerph-16-00241]\]. Moreover, we posted the study information on online platforms and forums for German offshore wind workers. We asked the workers to inform their female partners, who then contacted us directly via mail or telephone. Study participation was voluntary. Prior to the interviews, participants were asked to sign a declaration of informed consent. All participants were in a position to understand and consent to the study requirements, and provided written informed consent. The interviews were conducted until no new themes were identified, i.e., data saturation was reached. They were conducted in German and were tape recorded. Interview length was from 28 to 54 min. Participants were able to terminate the interviews at any time. No non-participants were present during the interviews. No repeat interviews were carried out. Field notes were made immediately after each interview. 2.2. Interview Guideline {#sec2dot2-ijerph-16-00241} ------------------------ A semi-structured interview guideline was developed within the framework of the empirical and theoretical background. The interview topic list is depicted in [Table 1](#ijerph-16-00241-t001){ref-type="table"}. A pre-test interview was performed in order to receive feedback from research colleagues and improve the interview guideline. 2.3. Analysis {#sec2dot3-ijerph-16-00241} ------------- All audio recordings were transcribed verbatim. The transcripts of the interviews were anonymized and analyzed in a deductive-inductive process according to Mayring's qualitative content analysis \[[@B19-ijerph-16-00241]\] by means of the software MAXQDA Analytics Pro, version 12 (VERBI Software GmbH, Berlin, Germany) \[[@B20-ijerph-16-00241]\]. An iterative process was applied in which the authors identified and refined codes, categories, and sub-categories. The coding was mutually checked for accuracy and was thoroughly discussed until consensus regarding the final coding system was reached. The final coding system was summarized in a separate document in which the material was further reduced and compacted. During the course of analysis, reflexivity and transparency regarding the potential influence of the researchers' objectives and preconceptions on the results and interpretations were encouraged. Transcripts and results were not returned to the interviewees. All quotes that were used for publication purposes were translated into English. 3. Results {#sec3-ijerph-16-00241} ========== 3.1. Sample Characteristics {#sec3dot1-ijerph-16-00241} --------------------------- As illustrated in [Table 2](#ijerph-16-00241-t002){ref-type="table"} and [Table 3](#ijerph-16-00241-t003){ref-type="table"}, seven women were aged between 31 and 40 years old; 10 women were married; seven women reported to work full-time, two worked part-time, and five were currently on maternity or parental leave; and eight women were mothers of children living in the household at the time of the study. The partners of 10 women had worked offshore from the beginning of their partnership. Furthermore, 12 offshore partners were currently working in a regular 14/14 work schedule, and two partners were working in a different schedule at the time of the interview but did have experience with the 14/14 schedule. Additionally, seven partners had at least three years of offshore work experience. 3.2. Single Life without the Offshore Partner {#sec3dot2-ijerph-16-00241} --------------------------------------------- ### 3.2.1. Advantages {#sec3dot2dot1-ijerph-16-00241} As a major advantage of living the single life, many women found themselves to be more independent and self-reliant in their scheduling of activities and appointments. This was also reported regarding duties and routines within the household and parenting: ""*Very simple things, like, for example, when to make purchases, what to buy, when to clean certain things (...).*"*(Interviewee \#8)*" A further advantage of the time spent alone was that the women pursued their own interests to a greater extent and were able to become engaged in several leisure activities: ""*Advantages, perhaps, that one deals with being alone and tries to find strategies to better cope with it. You do not rely on the fact that there is always someone there, but you put more thoughts in your own leisure activities.*"*(Interviewee \#4)*" Meeting friends and family, doing sports, and following diverse hobbies were stressed by the women as important leisure activities in this phase. ### 3.2.2. Disadvantages {#sec3dot2dot2-ijerph-16-00241} An important disadvantage of living the single life was reported to consist of the management and organization of daily life without the partner's general support. Spare time was described to be affected by the partner's absence in such way that the women perceived being forced to spend big parts of their social life alone: ""*A disadvantage is, of course, that you can only maintain your social contacts alone. If there are invitations or birthday parties, or if you just want to spend a nice evening with friends, this doesn't work.*"*(Interviewee \#13)*" During their time alone, there were moments in which the women especially missed their partners. These often concerned moments that both partners usually shared, in particular regarding time during the evenings or weekends. Difficult situations such as sickness, problems at work, or issues with childcare were also mentioned. Moreover, the partner's being missing on special occasions, e.g., weddings or special moments in the children's development, was emphasized. Some interviewees indicated emotional changes, e.g., feeling lonely or more tense than usual: ""*I realize that I'm not feeling well when he's not there. Sometimes I do not sleep for nights because I cannot stand it.*"*(Interviewee \#5)*" Another disadvantage consisted of perceived insecurities related to child care and parenting. Some interviewees described that they considered themselves as single mothers when their partners were offshore: ""*Well, more or less, I am a single parent for two weeks.*"*(Interviewee \#9)*" The perceived unpredictability, e.g., regarding the exact time of the partner's arrival back home, was also seen as disadvantageous. Moreover, the awareness of constraints in reachability when the partner was offshore---especially in cases of emergency---was highlighted as burdening. Many women spoke about changes in their work routines when their partners were away. They described often working overtime during this phase in order to be able to work less or leave on time when the partner was at home: ""*Of course, when my husband is not there, I work more. When he's at home, I scale that back a bit, so that we have more time for each other.*"*(Interviewee \#9)*" Especially women with children found themselves to be less flexible and more constrained in their work times when they were alone, e.g., due to their children's schedules. Some mothers also described that their children's ill-health seemed to be connected to the absence of their father: ""*It also happens that the children get sick if he is not there for a long time.*"*(Interviewee \#6)*" ### 3.2.3. Communication and Contact Styles {#sec3dot2dot3-ijerph-16-00241} Staying in contact with the partner during his offshore assignments was reported to be crucial for the women. Most of them described being able to contact their partners on a daily basis. The scheduling of contact was stated to depend on the shifts and workloads of the partner: ""*During his shift, I can write him and if he reads it in between, he can also answer me. But we rather talk on the phone during his free shifts.*"*(Interviewee \#12)*" Certain difficulties and restrictions to make contact were also described by some women; however, contact that could be initiated by both partners was reported to be the rule. The usage of several media (phone, messenger services, video telephony) was mentioned by the majority of the women. In particular, improved communication technologies were stated to enable more intensive contact and direct exchange. All in all, the majority of the women stated being satisfied with both the frequency and regularity of contact. Reasons for dissatisfaction were related to the time of contact and the general notion that technology-mediated contacts were not comparable to personal contacts. 3.3. Life as a Couple {#sec3dot3-ijerph-16-00241} --------------------- ### 3.3.1. Advantages {#sec3dot3dot1-ijerph-16-00241} Advantages of the 14/14 work schedule were often attributed to the time spent together as a couple during the partners' free turn. This time was stated to provide the couple with much time for joint activities and family life. The family life was perceived as being even more intensive due to the previous phase of separation, providing the couple with an opportunity to miss and look forward to seeing each other again. This was indicated to contribute to the liveliness of the relationship: ""*It's also good for our relationship, it's good to have such a short break (\...). That brings in a certain freshness.*"*(Interviewee \#6)*" The presence of the offshore partner at home was reported to allow for a higher quality of communication through face-to-face talks. A further advantage of the time spent together was that the partner could be involved in the child care and housekeeping. Family fathers were also able to follow their children's development more closely during their free time at home. More generally, some women described time-wise benefits of the daily living together, such as a greater flexibility for the couple in terms of planning short vacations or having breakfast together during the week. Financial benefits of the offshore job (e.g., in the form of good salaries) that contributed to family life were also mentioned. ### 3.3.2. Disadvantages {#sec3dot3dot2-ijerph-16-00241} A main disadvantage of the daily living together in the 14/14 schedule concerned the lack of habitualness and missing daily routine which could not be established in the course of two weeks: ""*We are all habitual people and habits can be very difficult at 2 weeks/2 weeks, I think.*"*(Interviewee \#11)*" A few interviewees described feeling an increased need to talk to their partners and to plan appointments when they were home. Some women even described a perceived pressure to get all everyday things---for which they normally had four weeks of time---done during the two weeks together: ""*You always have the feeling that you must put everything into these two weeks, because afterwards, your time together is over. Need for action, discussion needs... what you just can't always hold on the phone.*"*(Interviewee \#9)*" In terms of their work, many women described their own job as meaningful to them, regardless of the presence or absence of their partners. However, some women found it harder to go to work with their partners being at home and preferred staying at home with them, e.g., because they experienced feelings of guilt for leaving the partner alone. In addition, some women reported that the situation was especially disadvantageous for their partners: their free turns could be rather unsatisfactory, since the time they were able to spend with their families and friends was restricted due to other people's normal work routines: ""*It was not satisfactory for him either. He was not socializing as much as he had wished. The days are long when all people around you work full time.*"*(Interviewee \#11)*" Other interviewees explained that their partners struggled to find a balance in terms of the time spent with the family and with friends outside home. This was due to the fact that the workers were solicited a lot during their onshore turns, in particular on the weekends: ""*The time on the* *two weekends becomes very, very scarce. When there should be time for the partnership, but also for family, friends, and your own interests.*"*(Interviewee \#4)*" Several women described that their partners absolved work tasks and were contacted for professional purposes during their free turns onshore. Talking about offshore work and being contacted by colleagues were associated with greater difficulties for the partners to mentally detach and recover from work. Therefore, many interviewees disapproved of this behavior. ### 3.3.3. Conflicts and Compromises {#sec3dot3dot3-ijerph-16-00241} When asked about conflicts and compromises due to the specific living situation, about half of the women stated not noticing any specific conflicts. The non-existence of conflicts was attributed to the couple's mutual understanding and awareness of their limited time together: ""*Because we are separated again and again, you appreciate it (the time spent together) very much. And that makes us both feel that we are not arguing so fast and so much.*"*(Interviewee \#10)*" The other women reported that minor conflicts or discussions attributable to their specific living situation sometimes occurred. Conflicts, for example, emerged when the partner refused to get involved in housekeeping, or when he dedicated too much time to his work during his free turn. Further discussions were described to relate to planning difficulties of the couple due to the partners' offshore work. Moreover, minor discussions between the offshore partner and the children were reported to occur. In addition, it was stated that 'offshore couples' sometimes had to deal with a lack of understanding from their friends, who did not comprehend the amount of time the couple needed for themselves. The majority of the interviewees thought that they had to make more compromises compared to couples living a 'normal life'. They expressed that more agreements and consultations were necessary to suitably plan living together. Compromises were, for example, described in terms of the parenting, since the children had to live without their father for a while. Moreover, planning difficulties were a central concern, since all appointments had to be made in accordance with the partner's offshore schedule: ""*We have to direct our everyday life according to these offshore trips. He never knows when the trips will be ---they are not set at the beginning of the year---so we just cannot plan at all.*"*(Interviewee \#5)*" 3.4. Transition Phase {#sec3dot4-ijerph-16-00241} --------------------- ### 3.4.1. Reunion with the Partner {#sec3dot4dot1-ijerph-16-00241} The women described varying feelings, e.g., increasing anticipation and excitement, upon their partners' arrival back home. Typical behavior patterns were tidying up the house and avoiding other appointments: ""*Then I just run from A to B and check that everything is neat (...). That the food is ready and that no more laundry is lying around. That there are no disruptive factors in order for us to simply enjoy this moment together.*"*(Interviewee \#5)*" Only a few interviewees stated that they did not perceive a certain transition phase when their partners arrived back home. The majority declared that they needed a familiarization phase in which they had to adapt to their partner and the two adult household again: ""*At the beginning, you often need some time to get close again, because you have not seen the other person for so long.*"*(Interviewee \#7)*" The transition phase was generally described to last between one and four days. Women described that the arrival of the partner could upset the whole household, and that the habits and routines of the women at home were suddenly turned around: ""*You develop different habits---your own habits---when the partner is not there. And as soon as he comes back, it's all jumbled up.*"*(Interviewee \#8)*" In households with children, it was pronounced that the children behaved more actively and turned up during the father's arrival, demanding more attention than usual: ""*When he comes back, the children are usually there, and then the alarm goes from 0 to 100 in the booth.*"*(Interviewee \#13)*" ### 3.4.2. Needs and Expectations upon the Partners' Arrival {#sec3dot4dot2-ijerph-16-00241} Needs and expectations of the women regarding the time as a couple consisted of spending as much time together as possible and following social activities. Some women particularly highlighted their expectation that the partner should get involved in housekeeping and other duties at home: ""*I indeed expect that he will also take care of the household and of the things that happened while he was not there.*"*(Interviewee \#13)*" In contrast, expectations of the offshore partners stated by the women included that the women should await them at home upon their arrival and that the couple should share a good meal together on the first evening. A relevant need of the partner consisted of physical closeness to the women. Some interviewees believed that their needs and expectations corresponded well with those of their partners; for example, when both partners wished for physical closeness, calmness, and time spent together. In contrast, other women perceived discrepancies, which were especially related to the women's "work situation" versus the partners' "free time situation": while the women had to continue their daily work routine, the offshore partners found themselves to be in a holiday mood: ""*I get up at the same time in the morning, go to work, and come back in the evening. And then my partner took the time as a vacation, but I was still in the working cycle.*"*(Interviewee \#4)*" ### 3.4.3. Parting {#sec3dot4dot3-ijerph-16-00241} The interviewees described that the time spent together as a couple usually passed rapidly. Some women described that during the last days before their partners' departure, the workers started to mentally prepare themselves for their offshore assignments. The departure was termed as a difficult situation by some women, provoking feelings of sadness. Some women also reported that their children's behavior changed during their father's departure. For example, they could demonstrate their displeasure by crying or working themselves up. [Figure 1](#ijerph-16-00241-f001){ref-type="fig"} gives an overview of the specific features, advantages, and disadvantages of living the 14/14 schedule as related to the different phases of daily life. 3.5. Coping Strategies {#sec3dot5-ijerph-16-00241} ---------------------- ### 3.5.1. Strategies of the Women {#sec3dot5dot1-ijerph-16-00241} When asked about strategies to cope with the absence of their partners, many women described that actively searching for the support of families and friends played an important role: ""*I think that you rely more on the social network around you. That you particularly promote your network. You simply intensify other social contacts, family, friends.*"*(Interviewee \#4)*" In general, pursuing an active lifestyle was described as a coping strategy by many women. This included, for example, doing sports or meeting friends. Further ways to cope with the situation were stated to consist of adapting oneself to the schedule, focusing on the time spent together as a couple, and staying in regular contact during periods of separation: ""*For me, this is already somewhat normal. And we talk on the phone in the evenings and write each other during the evenings when he has enough time.*"*(Interviewee \#12)*" Some interviewees stated that they coped with the situation by structuring their time in an organized manner, while others reported that they coped by keeping their expectations low regarding the time spent together. Some women reported that they did not apply any coping efforts. When being asked about sustaining exchange with other women living in a similar situation, most interviewees responded that they had not made contact with other women of offshore workers, although some found such an exchange to be desirable. A few others, in contrast, described having irregular contact with other partners of offshore workers. These contacts were reported to be organized autonomously and without the offshore companies' support. ### 3.5.2. Strategies of the Couples {#sec3dot5dot2-ijerph-16-00241} In terms of coping strategies applied by the couples, many women emphasized the meaning of communication and structure for dealing with the phases of separation: ""*This always means a lot of exchange with each other, and a lot of talking and communicating. Then it works. But those who do not have this ability will find it difficult.*"*(Interviewee \#9)*" The importance of adhering to fixed and regular contact times was stressed. Additionally, spending time as a family/couple when the partner was onshore was underlined, explaining that other obligations or appointments were avoided in order to create more family-time: ""*Certain rituals are that, when he's at home, (...) the last weekend before he leaves, or at least 1 or 2 days, that you have these days completely to yourself. And then accept no appointments.*"*(Interviewee \#10)*" Mutual understanding and trust were also highlighted as important. Still, there were couples who did not apply any coping strategies or rituals to deal with the specific situation. [Table 4](#ijerph-16-00241-t004){ref-type="table"} summarizes the coping strategies applied by the women and couples. 3.6. Reconciliation of Offshore Work and Family Life/Partnership {#sec3dot6-ijerph-16-00241} ---------------------------------------------------------------- ### 3.6.1. Opinions on Reconciliation {#sec3dot6dot1-ijerph-16-00241} The women's views on whether or not they considered offshore work and family life to be reconcilable differed. There were some interviewees who described offshore work as being sufficiently family-friendly, especially when the children were already older in age. The primary reason for the family-friendliness was that fathers were able to spend intensive periods of time at home: ""*Which father of a family can say that he is completely at home for 2 weeks, from morning to night?*"*(Interviewee \#10)*" In contrast, other women did not consider offshore work to be family-friendly due to the several named disadvantages implied by the 14/14 schedule. This was particularly pronounced by interviewees with smaller children, emphasizing that the partner would miss out on too much of their development: ""*I just think that the men miss too much. Especially when a child is born. In the first year, our son actually had only me, his dad was always a bit of a rival.*"*(Interviewee \#5)*" A dividedness regarding interviewees' opinions was also apparent with regard to the reconciliation of offshore work and partnership. Some women believed that offshore work was partnership-friendly and that it helped in keeping the partnership alive. Living with the periodical absences of the partner was reported to be practicable without children: ""*Without a child, I'd say, it works. Then you can deal with it, even without noticing a negative impact on the relationship.*"*(Interviewee \#4)*" However, other interviewees stated that the 14/14 schedule imposed heavy demands on the relationship, and that they personally perceived the situation as burdensome: ""*The big disadvantage is that the private life suffers greatly, that one must cut back on the partnership because the contact is missing.*"*(Interviewee \#5)*" The appraisal of whether or not offshore work was partnership-/family-friendly was stated to depend on the couples' expectations (e.g., the amount of time the couple wanted to spend together and the desired frequency of contact). Moreover, the amount of support from external sources (e.g., parents, friends) played a role in the women's judgement: ""*If you do not have family support, then it is not necessarily family-friendly. So I think that the environment still plays a big role.*"*(Interviewee \#7)*" ### 3.6.2. Needs and Wishes for Improving Reconciliation {#sec3dot6dot2-ijerph-16-00241} When asked about wishes for improving the reconciliation of offshore work and family life/partnership, some women stated that they did not have any specific wishes, or explained that they did not think that any measures for improvement could be taken due to the unchangeable 14/14 schedule. Others, however, described wishes regarding their partners' work schedule (e.g., other days of arrival and departure, longer offshore or onshore stays). The wish for greater regularity and predictability of the offshore assignments was also expressed: ""*These are my concerns, reliability and predictability.*"*(Interviewee \#6)*" Furthermore, it was proposed that the workers should get more free time offshore in order to increase chances for communication with the families and to strengthen the workers' recovery from work. A few women also wished to get to know their partners places of work to develop a better understanding of the work situation offshore. ### 3.6.3. Support from Offshore Companies {#sec3dot6dot3-ijerph-16-00241} The women mostly described that they did not know about specific offers provided by their partners' companies to facilitate the reconciliation of offshore work and family life/partnership. However, they described single offers provided by the companies that they considered to be helpful, e.g., flights back home at short-notice in case of emergencies: ""*Of course, if there was a death in the family or something, definitely. Or now with the child's birth, I could call him anytime and would try to get him off the platform.*"*(Interviewee \#10)*" Some women said that company events were organized for the whole family, and that parental leave for fathers was an option. Further offers consisted of the free use of a telephone and internet connection on the offshore platforms, allowing the couples to stay in contact: ""*That's a good option, I think, that companies put a lot of emphasis on enabling the workers to have regular contact with their families at home.*"*(Interviewee \#10)*" 4. Discussion {#sec4-ijerph-16-00241} ============= By conducting our interview study, we were able to gain important insights into the challenges, advantages, and disadvantages, as well as the psychosocial adaptation, associated with living the 14/14 schedule from the perspective of women of offshore wind workers. We generally found the proposed differentiation between the three distinct social realities for offshore couples (her single life at home, a phase of transition, and the couple's common life together), as suggested by Solheim \[[@B11-ijerph-16-00241]\], to be similarly described in our study. Moreover, our results seem to be in line with the FIFO cycle proposed by Gallegos \[[@B12-ijerph-16-00241]\] for both offshore employees and their partners. For example, the occurrence of mixed emotions during transition phases, as identified in previous research \[[@B12-ijerph-16-00241],[@B15-ijerph-16-00241]\], was also prevalent for the women in our sample. Overall, despite some burdens, the women in our sample seemed to have adapted relatively favorably to the challenges and demands of living the 14/14 schedule. Although minor difficulties and problems related to the partnership and family life were stated, most women seemed to be able to cope with the challenges associated with the 14/14 work schedule. When contrasting our findings with those of earlier research studies in the offshore oil and gas branch, we found previous research to illustrate a slightly more negative picture regarding the psychosocial adaptation of offshore families \[[@B21-ijerph-16-00241]\]. However, the situation seems to have improved over the last decades. For example, Parkes and colleagues \[[@B8-ijerph-16-00241]\] noted a positive trend, and our study further supports this development. 4.1. Single Life without the Offshore Partner {#sec4dot1-ijerph-16-00241} --------------------------------------------- As regards the single life of the women, we found the main advantage to consist of the greater self-reliance and independence women perceived in their daily living. This is in line with previous results showing that women were able to enjoy their independence and freedom \[[@B21-ijerph-16-00241],[@B22-ijerph-16-00241]\], and that they could benefit from their partners' absences in developing greater personal confidence \[[@B8-ijerph-16-00241]\]. However, in general, we found that the women in our study reported more disadvantages than advantages of their single life. Negative aspects, such as perceptions of loneliness when the partner was away, have been similarly revealed in previous studies \[[@B8-ijerph-16-00241],[@B12-ijerph-16-00241]\]. For example, in an interview study, two thirds of the spouses of offshore oil and gas workers reported loneliness to be a problem "sometimes" or "often" \[[@B8-ijerph-16-00241]\]. In contrast to earlier studies in which women reported experiencing social isolation due to their partners' absences \[[@B8-ijerph-16-00241],[@B13-ijerph-16-00241],[@B22-ijerph-16-00241]\], our interviewees did not describe difficulties in fully participating in social life during this phase. This difference could be related to the fact that the women in our study did not seem to center their social lives strictly around their partners; in contrast, many of them stated that they actively engaged in social life when their partners were away. In earlier studies, women were found to deliberately restrict their social lives when their partners were away \[[@B8-ijerph-16-00241],[@B22-ijerph-16-00241]\]. An important finding of our study relates to the use and impact of new ways of communication as a result of technological advances. We found the women in our study to positively highlight their chances for communication via the use of diverse social media. While earlier studies declared problems in communication due to the---back then---existing communication systems \[[@B13-ijerph-16-00241]\], today's offshore women may draw on several communication systems which allow them to keep in contact with their partners. The importance of improved telecommunications in facilitating adjustment and maintaining family connectedness has also been noted for offshore and FIFO families \[[@B8-ijerph-16-00241],[@B10-ijerph-16-00241],[@B22-ijerph-16-00241]\]. One study, for example, found that women who could initiate calls to contact their offshore partners had less difficulty in adjusting to the absence compared to women who were unable to do so \[[@B8-ijerph-16-00241]\]. Our results indicate an intrinsic work motivation among the women in our sample: becoming engaged in work was identified to be important, helping them to fill the days when their partner was offshore. In contrast to an earlier study in which women's employment tended to increase family strain \[[@B23-ijerph-16-00241]\], we did not find this to be the case in our study. This discrepancy should be interpreted in view of the specific sample: six out of 14 women did not have children in the household, thereby potentially increasing their chances to engage in paid work. Furthermore, the result may also be attributable to sociocultural changes that have occurred during the last decades: nowadays, it is more common and socially accepted for women to build their own careers. The trend of increased employment rates among women is also reflected in our sample, in which all women were employed (despite five women currently being on parental or maternity leave). In contrast, in earlier samples of offshore wives, only one third \[[@B23-ijerph-16-00241]\] and two thirds of the women \[[@B8-ijerph-16-00241]\] respectively, were engaged in paid work. Similarly, only two women in our study were in part-time employment, whereas more than half of the women in the study of Parkes and colleagues \[[@B8-ijerph-16-00241]\] worked part-time, which was found to be influenced by the demands of childcare. The fact that most women in our study worked full-time may also be related to currently increasing options for child care, e.g., provided by day care centers and kindergartens. 4.2. Life as a Couple {#sec4dot2-ijerph-16-00241} --------------------- With respect to life as a couple, the women in our study had differentiated views on the costs and benefits of the 14/14 schedule for their living situation. A major advantage related to the 14/14 schedule was seen in the workers' rest periods onshore, allowing an increased duration of presence at home and favoring family life. This advantage was also noted in earlier studies \[[@B8-ijerph-16-00241],[@B12-ijerph-16-00241],[@B13-ijerph-16-00241],[@B21-ijerph-16-00241]\]. As previously identified \[[@B11-ijerph-16-00241],[@B12-ijerph-16-00241],[@B24-ijerph-16-00241]\], we found that both partners initially needed a familiarization phase to readjust to having another adult in the household. Similar to our results, it was previously found that reunions and partings are the most difficult times emotionally for couples and families \[[@B10-ijerph-16-00241],[@B12-ijerph-16-00241]\]. Moreover, in accordance with previous findings \[[@B12-ijerph-16-00241]\], women in our study reported that their routines could become less structured when their partners returned home. In terms of the children's behavior, some women described their children's conduct as varying and depending on the phase of absence, presence, or transition. Gallegos \[[@B12-ijerph-16-00241]\] has similarly described that the behavior of children of FIFO workers could become clingy, and that it could take some time for the children to feel comfortable upon their father's return. Notably, there were only a few conflicts described as having occurred during the time spent together, although conflicts seemed to be a stressor for couples in previous studies \[[@B21-ijerph-16-00241],[@B22-ijerph-16-00241],[@B23-ijerph-16-00241]\], especially during the first days spent together \[[@B21-ijerph-16-00241]\]. In contrast to previous research highlighting women's increased responsibilities as a potential source of conflict \[[@B13-ijerph-16-00241],[@B15-ijerph-16-00241]\], conflicts described by our sample seemed to emerge from the partners' behaviors at home, e.g., their reluctance to get involved in housekeeping. In earlier times, housework did not represent a source of conflict for offshore oil and gas couples \[[@B21-ijerph-16-00241]\]; the major responsibility for domestic work remained with the wives, which was attributed to the more traditional views on household division back then \[[@B21-ijerph-16-00241]\]. In contrast, our results support the notion that women's aspirations and role models have changed \[[@B8-ijerph-16-00241]\], since women in our sample reported distinct expectations regarding the workers' engagement in the household. The fact that conflicts among the couples were described to occur rather seldom could also be related to the coping strategies that were applied, which may be effective in reducing potential conflicts. Moreover, the women in our study seemed to be aware of existing differences compared to non-offshore families; for example, they acknowledged that more compromises had to be made in contrast to other couples. It has been noted previously that such an awareness may increase the implementation of strategies to support family functioning \[[@B15-ijerph-16-00241]\]. 4.3. Coping Strategies {#sec4dot3-ijerph-16-00241} ---------------------- We found the women in our study to apply several coping strategies for dealing with their situation, such as seeking support, thinking positively, and regulating adverse emotions. Thereby, our results undermine previous findings indicating that many offshore women actively pursue some form of coping, e.g., engaging in an active lifestyle or utilizing social support \[[@B8-ijerph-16-00241],[@B12-ijerph-16-00241],[@B22-ijerph-16-00241],[@B23-ijerph-16-00241]\]. In earlier studies, women of offshore oil and gas workers were found to use coping strategies to mitigate loneliness when the partner was offshore (e.g., keeping busy, keeping in touch with the family, and taking part in recreational activities \[[@B8-ijerph-16-00241]\]). Similarly, FIFO workers were found to engage in social networks providing them with assistance and companionship while the workers were away \[[@B12-ijerph-16-00241]\]. Still, there is a need for further investigation of the women's coping strategies. Aspects such as work schedules (full-time, part-time) and childcare may require different coping strategies that should be examined by further research. An important strategy of the couples was reported to consist of staying in regular contact, which was also of importance for oil and gas, as well as FIFO, families \[[@B8-ijerph-16-00241],[@B10-ijerph-16-00241],[@B12-ijerph-16-00241]\]. In studies among FIFO families, regular effective communication was the most important strategy to protect family cohesiveness \[[@B12-ijerph-16-00241]\], and was found to be strongly associated with family satisfaction \[[@B10-ijerph-16-00241]\]. 4.4. Reconciliation of Offshore Work and Family Life/Partnership {#sec4dot4-ijerph-16-00241} ---------------------------------------------------------------- Women's views regarding the reconciliation of offshore work and family life/partnership seemed to partly depend on their availability of external sources, such as support from friends and family, as well as on the couples' own expectations. This agrees with the notion that a stronger accordance between perceptions and expectations of partners, e.g., regarding family satisfaction, could lead to less family conflicts \[[@B10-ijerph-16-00241]\] and less critical family structures \[[@B13-ijerph-16-00241]\]. A support offer provided by offshore companies was named in terms of the flexibility of shifts in cases of emergency. This has also been noted by women of offshore oil and gas workers, who were confident that their partners could be flown home in a family emergency \[[@B8-ijerph-16-00241]\]. However, in sum, the women in our study did not report an intensive bandwidth of offers provided by offshore companies to facilitate the reconciliation of offshore work and family life. This finding might either indicate a need for the companies to improve their offers, or to make existing offers more visible for offshore families. In contrast, in FIFO, as well as oil and gas industries, more company support has become evident, e.g., consisting of counseling services, peer-programs, organized family events, or visits to the site for families \[[@B8-ijerph-16-00241],[@B12-ijerph-16-00241]\]. Such offers may also add to the reconciliation for couples and families in the offshore wind industry. 4.5. Strengths and Limitations {#sec4dot5-ijerph-16-00241} ------------------------------ A strength of our study is the fact that we recruited women with varying sociodemographic characteristics, e.g., different ages and family status. This enabled us to establish a more complete picture of the situation of offshore women with varying backgrounds. To increase the trustworthiness of our findings, we employed rich descriptions of our results and displayed many direct quotes from the interviewees \[[@B25-ijerph-16-00241]\]. Moreover, we discussed our results profoundly within the group of researchers, and contrasted them with empirical references. However, it should be noted that our findings are based on a convenience sample which was partly achieved via a snowballing technique, thereby increasing the risk of self-selection among the participants. For example, women with a greater interest in the topic might have been more prone to participate, and may not be representative of other female partners of offshore workers. Moreover, our sample is likely to represent a self-selected group of 'survivors', as it has been noted for other samples of offshore women \[[@B8-ijerph-16-00241],[@B21-ijerph-16-00241]\]. There are several indications for this assumption: for example, interviewees' partners were currently working offshore, and many of them had already worked offshore for several years. It can be assumed that offshore workers are more likely to continue with their work when their women are also able to adjust to the offshore lifestyle \[[@B8-ijerph-16-00241]\]. Our sample, therefore, likely represents a survivor group of women that have responded rather positively to this lifestyle and experienced less difficulties in adjustment. Moreover, it should be kept in mind that 10 out of 14 workers were already involved in offshore work at the beginning of the relationship, meaning that most couples did not have to deal with a disrupt change of their living situation during the course of the relationship. Further methodological limitations concern the fact that we conducted telephone interviews instead of face-to-face interviews, implying an asynchronous communication of place by telephone and a reduction of social clues \[[@B26-ijerph-16-00241],[@B27-ijerph-16-00241]\]. Another limitation of our study may be seen in the relatively small sample size. However, the size of interviews in our study appeared to be sufficient to achieve data saturation. In support of this, it has been concluded that data saturation usually occurs within the first twelve interviews \[[@B28-ijerph-16-00241]\]. In any case, generalizations of our results are impeded by the nature of our qualitative research design. Further research studies with larger sample sizes are needed. In such studies, it would be interesting to conduct interviews with couples in order to incorporate both the views of offshore workers and their female partners. Moreover, quantitative research studies should be conducted to statistically explore the antecedents, moderators, and outcomes of psychosocial adaptation among offshore couples. Previous research has, for example, suggested that role expectations, the presence of dependent children at home, and the quality of communication in the relationship may influence the effects of stressors on psychosocial adaptation \[[@B10-ijerph-16-00241]\]. Moreover, since our sample likely represents a survivor group of couples living the 14/14 schedule, it seems worthwhile to compare their situation with the situation of families where workers decided to leave offshore work. 5. Conclusions {#sec5-ijerph-16-00241} ============== The present study expanded upon the current scientific evidence and provided an up-to-date perspective on the situation of offshore wind couples and families living the 14/14 work schedule. The women in our study reported differentiated views as to the benefits and costs associated with their particular lifestyle. Various coping strategies were stated by the women, which could facilitate psychosocial adaptation. Despite experiencing certain burdens, most of the women in our sample seemed to have adapted relatively favorably to their living situation. Our results suggest that certain sociological and technological advances within the last decades, e.g., changes in women's role models and improved communication technologies, may ease psychosocial adaptation among offshore families. We would like to thank all women who participated in the interview study. We would also like to acknowledge Johanna Bertram for her help in data preparation. Conceptualization, J.M., S.R., M.K., V.H., and S.M.; methodology, J.M., M.K., and S.M.; recruitment of participants and conduct of interviews, J.M. and M.K.; data curation and analysis, J.M. and S.R.; writing---original draft preparation, J.M. and S.R.; writing---review and editing, M.K., V.H., and S.M.; visualization, J.M.; supervision, S.M. and V.H. This research received no external funding. The authors declare no conflict of interest. The data analyzed during the current study are not publicly available due to German national data protection regulations. They are available on individual request from the corresponding author. {#ijerph-16-00241-f001} ijerph-16-00241-t001_Table 1 ###### Interview topic list. --- ------------------------------------------------------------- ------------------------------------------------------------------------------------------------------------------------------------------- 1 Introduction Study information, confidentiality, informed consent 2 Socio-demographics *Interviewee*: age, marital status, occupation, work schedule, duration of partnership with offshore partner, shared household, children\ *Offshore partner*: offshore experience, occupation, work schedule 3 Phase 1: Single life without offshore partner Advantages and disadvantages of the phase,\ contact opportunities while partner is offshore 4 Phase 2: Daily life with offshore partner Advantages and disadvantages of the phase, associated behavior patterns and feelings 5 Phase 3: Transition phase Associated behavior patterns and feelings,\ needs and expectations upon the partner's arrival 6 Coping strategies Strategies of the interviewee and the couple 7 Reconciliation of offshore work and family life/partnership Reconciliation of offshore work with family life/partnership; wishes for a facilitated reconciliation --- ------------------------------------------------------------- ------------------------------------------------------------------------------------------------------------------------------------------- ijerph-16-00241-t002_Table 2 ###### Participant characteristics. -------------------------------------------------------------- --------- **Interviewee** ***n*** **Gender** female 14 **Age** 20--30 years 4 31--40 years 7 41--50 years 3 **Marital status** not married 4 married 10 **Work schedule** full-time 7 part-time 2 currently in maternity or parental leave 5 **Duration of partnership** 1--5 years 6 6--10 years 4 11--20 years 2 \>20 years 2 **Partnership with partner working offshore from the start** yes 4 no 10 **Children** yes and living in household 8 expecting 4 no 2 **Offshore partner** ***n*** **Work schedule** 14 days offshore/14 days onshore 12 other 2 **Offshore experience** \<1 year 1 1--2 years 6 3--4 years 3 \>4 years 4 -------------------------------------------------------------- --------- ijerph-16-00241-t003_Table 3 ###### Participant characteristics (in detail). ID Age \* Marital Status Occupation Work Schedule Duration of Partnership \* Partnership with Partner Offshore \* Shared Household \* Children (in Household) Occupation Offshore Experience \* Work Schedule \*\* ----------------- ---------------------- ---------------- ------------------------------------------------------- -------------------------- ---------------------------- -------------------------------------- --------------------- ------------------------- ------------------------------ ------------------------ -------------------- **Interviewee** **Offshore partner** 1 28 married not specified (currently on parental leave) full-time 10 6 9, 5 1 (1) service technician 6 14/14 2 33 not married not specified full-time 3 3 2, 5 0 (0) technical project management 6 irregular 3 35 married office management full-time apprenticeship 10 2, 5 10 2 (2) quality management 2, 5 14/14 \*\* 4 32 married social worker (currently on maternity leave) full-time 5, 5 1, 5 5 0 (0) health and safety 1, 5 14/14 5 29 married florist full-time 2, 5 2, 5 2, 5 1 (1) service technician 5 8/4 6 39 married administrative official part-time 15 5 12 2 (2) operations manager 5 14/14 \*\* 7 29 married architect (currently on maternity leave) 10 2 9 0 (0) offshore medic 2 14/14 8 25 married maritime sector full-time 8 0, 5 6 0 (0) platform master/ 0, 5 14/14 \*\* 9 46 married teacher full-time 27 3 25 2 (2) service technician 3 14/14 10 31 married tailoress (currently on maternity leave) part-time 5 1, 5 4, 5 0 (0) nautical officer 1, 5 14/14 11 35 married health insurance part-time 13 2 11 1 (1) service technician 2 14/14 12 34 not married occupational therapist (currently on maternity leave) full-time 3, 5 3, 5 0, 5 0 (0) service technician 14/14 13 42 not married tailoress full-time 2, 5 2, 5 1, 5 2 (2) electrician 3 14/14 14 50 married house economics full-time 30 2 28 1 (1) service technician 2 14/14 \* in years \*\* in turns with office weeks for specific trainings. ijerph-16-00241-t004_Table 4 ###### Coping strategies of the women and the couples. Coping Strategies of the Women Coping Strategies of the Couples ------------------------------------------------ ------------------------------------- Seeking support of families and friends Adhering to fixed contact times Pursuing an active lifestyle Mutual understanding and trust Structuring time in an organized manner Spending an intensive time together Adapting oneself to the living situation Focusing on time spent together as a couple Keeping expectations low Contact with other women in the same situation

Introduction ============ Substantial thermopathology may ensue when the human body is exposed to extremely low temperatures \[[@B1]\]. Every unintentional decline in the core temperature below 35°C is considered to be accidental hypothermia, and when the effect is protracted the term of prolonged accidental hypothermia (PAHT) is used. Nuclei in the pre-optic anterior hypothalamus coordinate heat conservation. Activation of these thermostats and the cutaneous cold receptors initiate a cascade of compensatory physiologic events, the failure of which leads to the clinical and ultra-structural manifestation of PAHT. Myocardial damage after exposure to extremely low temperatures is usually described using the general term \"myocarditis\". However, the effects of PAHT on the myocardium remain unclear and are mainly limited to the clinical picture of circulatory collapse and arrhythmogenesis \[[@B2]\]. A common sign in the electrocardiogram (ECG) is the convex elevation at the junction of the ST segment and the QRS complex, or the so-called Osborn wave \[[@B3],[@B4]\]. Imaging modalities for the diagnosis of PAHT-dependent myocardial damage have not been described so far. In our patient, who had PAHT, echocardiography (ECHO) and cardiac magnetic resonance (CMR) imaging shed further light on the progress of myocardial structural damage. Case presentation ================= A previously healthy 42-year-old woman of Caucasian origin attempted to commit suicide on a winter afternoon in a hilly coastal area. Her intention was to fall to the sea from a tall rock. On her way to the rock, however, it became dark and she encountered a snow blizzard that made her disoriented and led to her subsequent fall towards a ravine. She was discovered after 18 hours. She had no history of alcohol use or substance misuse. She was not on any medication. The patient was admitted to hospital in a semi-comatose condition, with a body core temperature of 29°C and no external injuries. She required warmed (43°C) intravenous fluid infusion, inotropic support and mechanical ventilation due to cardiocirculatory collapse (systolic blood pressure of 70 mmHg). Although cardiopulmonary bypass re-warming was proposed, the patient\'s relatives refused the use of any invasive technique. Her core temperature was restored after four hours of external electric warming. An admission ECG showed atrial fibrillation with a mean rate of 85 beats per minute. The patient was administered 1 mg of atropine intravenously in the ambulance prior to admission due to bradyarrhythmia. ST segment elevation and Osborn waves were apparent in leads V4-V6 (Figure [1A](#F1){ref-type="fig"}). Her ECHO revealed global hypokinesia of left ventricular (LV) wall segments with an ejection fraction of 25%. A mild rise in the patient\'s CK (290 U/l), troponin I (4.15 ng/ml) and BNP (330 pg/ml) values was also observed. Eight hours after the patient was warmed up, her sinus rhythm was restored and Osborn waves were replaced by minor ST elevation in leads V4-V6 (Figure [1B](#F1){ref-type="fig"}). {#F1} Over the next few days the patient\'s ECHO showed a progressive circumferential thickening of the LV wall, which is suggestive of interstitial oedema (Figure [2](#F2){ref-type="fig"}). Her LV contractility also showed some improvement and atypical ST changes appeared on the ECG (Figure [1C](#F1){ref-type="fig"}). Extubation and weaning from inotropes became possible on the third day. The patient then regained full consciousness and became asymptomatic apart from mild dyspnoea. Neurological examination was negative for focal neurological damage and a coronary angiography revealed normal coronary arteries. {#F2} Concerning the ECHO findings, prolonged exposure to a cold and wet environment could have led to a viral insult on the patient\'s myocardium. This could account for the mild rise of troponin I and the observed ischaemia-like changes of ECG. For this reason CMR was ordered, which showed hyper-enhancement of LV myocardium on T1-weighted images, before (Figure [3](#F3){ref-type="fig"}) and after (Figure [4](#F4){ref-type="fig"}) gadolinium, compatible with diffuse myocardial inflammation and oedema. Neither focal increases of mid-wall or subendocardial signal nor residual fibrosis were found. This made the diagnosis of acute viral myocarditis, as well as the need for a CMR guided biopsy, less likely. Furthermore, a viability study excluded myocardial infarction. No other causes on which to attribute the ECHO and CMR pattern were detected apart from PAHT. Viral and immunological screenings were also negative for myocarditis. {#F3} {ref-type="fig"}**, after the administration of gadolinium**.](1752-1947-0003-0000008459-4){#F4} The patient constantly improved and was discharged after 10 days. She was given ACE-inhibitors, beta-blockers and a statin. We prescribed this empirical treatment on the basis of known pleiotropic properties of such drug categories. One month later, the patient\'s clinical and biochemical parameters were completely restored and there were no evidence of myocardial oedema or depressed LV functions on her ECHO. At the time of writing this case report, the patient has already completed one year of clinical and ECHO follow up examinations and is doing well without any pharmacological treatment. Discussion ========== In the setting of PAHT, various cardiovascular complications may occur \[[@B2]\]. In cases of mild PAHT (35°C to 32.2°C), tachycardia, bradycardia, vasoconstriction and an increase in cardiac output and blood pressure appear. Moderate PAHT (32.2°C to 28°C) as in this case provokes a progressive decrease in pulse and cardiac output and may increase atrial or ventricular arrhythmias and Osborn waves. The latter are considered to represent an epicardial-endocardial voltage gradient, which is associated with the localized appearance of a prominent epicardial action potential notch. \[[@B4]\] As hypothermia becomes more severe, so the appearance of Osborne waves also becomes more common. Osborn waves in this case, however, were transient and disappeared eight hours after the patient\'s normal temperature was restored. Severe PAHT (\<28°C) is responsible for a progressive decrease in blood pressure, heart rate and cardiac output, a burst of ventricular arrhythmias, which can terminate in asystole, and pulmonary oedema. Pathophysiologic mechanisms involved in PAHT-induced cardiac damage comprise vasoconstriction, ventilation-perfusion mismatch, increased blood viscosity, decreased oxygen release from haemoglobin, and hypercoagulability. Epicardial petechiae, subendocardial haemorrhages and microinfarcts are found in the ventricular myocardium. This is presumably related to abnormalities in the microcirculation \[[@B5]\]. In this case, a mild rise in troponin I and ischaemic-like ECG alterations were observed. The patient\'s normal coronary angiography and CMR viability test, however, excluded any significant coronary disease. Cardiac interstitial oedema in PAHT has not been described so far. However, myofibrillar injury and mitochondrial oedema are common and reversible ultrastructural findings in medically and experimentally induced moderate or deep hypothermic blood cardioplegia \[[@B6]\]-\[[@B8]\]. We do not know if iatrogenic hypothermia and PAHT share a common microscopic pattern of cardiac damage. In this case, ECHO and CMR detected diffuse myocardial oedema with characteristics quite different from those of myocardial infarction and/or infective myocarditis \[[@B9],[@B10]\]. The latter presents a patchy distribution through the LV, which is located frequently in the lateral free wall and originates from the epicardial quartile of that wall. Other commonly seen patterns in infective myocarditis are mid-wall stria pattern in the basal interventricular septum, residual fibrosis and early changes of ventricular remodeling \[[@B11],[@B12]\]. None of these patterns was found in our patient\'s CMR, which increased the chances that this was a PAHT induced myocardial damage. The negative results of the patient\'s viral and immunological tests pointed to the same conclusion. We do not know if myocardial oedema was related to the PAHT per se or to the re-warming techniques we used. Warmed (43°C) fluid infusions and active external trunk re-warming were applied to the patient. More aggressive invasive approaches were refused by the relatives. However, several re-warming techniques are proposed for treating PAHT, including peritoneal lavage with heated crystalloid, direct hepatic rewarming, closed pleural irrigations with large bore thoracostomy tubes, closed circuit oesophageal tubes, extracorporeal arteriovenous and venovenous rewarming and haemodialysis \[[@B2],[@B13]\]. All of these methods are capable of raising core temperature by 2 to 4°C per hour. Cardiopulmonary bypass is described as the most efficient method of re-warming in PAHT. Although a rise of 1 to 2°C every five minutes was observed, when this technique was applied to PAHT victims with a core temperature of 22°C, a period of one to four hours was necessary before re-warming to 36°C. \[[@B14]\] The value of cardiopulmonary bypass re-warming in patients with circulatory arrest but only moderate hypothermia is doubtful. Furthermore, the upper and lower limits of the core temperature at which attempts justify the use of extracorporeal circulation remain to be fully elucidated \[[@B15]\]. Conclusions =========== In this case, PAHT induced a core temperature of 29°C and presented with bradyarrhythmia, Osborn waves, mild troponin I increase and heart failure. Myocardial oedema detected by ECHO and further investigated by CMR, constituted an intriguing structural finding. In case this imaging pattern is validated by other observations, it can endorse a prognostic value and be used as a criterion for determining the efficacy of this re-warming technique. Abbreviations ============= CMR: cardiac magnetic resonance; ECG: electrocardiogram; ECHO: echocardiogram; LV: Left ventricle; PAHT: Prolonged accidental hypothermia. Consent ======= Written informed consent was obtained from the patient for publication of this case report and any accompanying images. A copy of the written consent is available for review by the Editor-in-chief of this journal. Competing interests =================== The authors declare that they have no competing interests. Authors\' contributions ======================= ES drafted the manuscript. SA helped draft and edit the manuscript. ES, GR, PV and CK looked after the patient clinically. All authors read and approved the final manuscript.

1. Introduction {#sec1} =============== *Yokenella regensburgei* belongs to the family*Enterobacteriaceae* and there is no strong evidence to support its clinical importance. There have been no*Y. regensburgei* infection reports in human immunodeficiency virus (HIV) infected patients, although a few case reports have suggested it is an opportunistic pathogen. Herein, we describe a case of septicemia in South West China caused by*Y. regensburgei* in a patient with HIV infection and present a review on*Y. regensburgei*literature. 2. Case Report {#sec2} ============== A 38-year-old male with a 20-year history of injection drug use was admitted to Chongqing Public Health Medical Center (Chongqing, China) in October 2013 due to a fever in absence of chills and cough. Two months prior to his admission, the patient was admitted to another hospital for epistaxis, dyspnea and odynuria with anemia, leucopenia, thrombocytopenia, and urinary tract infection, which improved with blood transfusions and use of antibiotics (levofloxacin injection, 0.2 grams each time, twice a day, for 14 days). His white blood cell counts ranged from 2.0 × 10^9^ to 2.5 × 10^9^ cells/liter during the period of 1 year prior to this admission. He was confirmed to be HIV-positive since 2009 and received one year later an antiretroviral regimen containing stavudine, lamivudine, and efavirenz, which was discontinued in March 2013 due to renal impairment. He had been on methadone maintenance treatment for about 1 year. He worked as a farmer. Family and social history were noncontributory. He smoked 10 cigarettes a day on average but rarely consumed alcohol. On physical examination, the patient appeared pale and uncomfortable. Temperature was 38.7°C, blood pressure was 127/84 mmHg, heart rate was 104/min, and respiratory rate was 20/min. Oxygen saturation was 100% on 3 liters/minute nasal cannula. Cardiac exam was within normal limits and pulmonary and abdominal exams were benign. His skin was intact and no ulceration was found in his mouth. Chest computed tomography was performed and no abnormalities were found. Blood examinations revealed anemia with hemoglobin of 42 g/liter (reference range, 130 to 175 grams/liter). Blood tests also revealed thrombocytopenia with a platelet count of 17 × 10^9^ cells/liter (reference rage, 125 × 10^12^ to 350 × 10^12^ cells/liter). His white blood cell count was 3.79 × 10^9^ cells/liter (reference rage, 3.5 × 10^9^ to 9.5 × 10^9^ cells/liter), differential with 81% neutrophilic granulocyte (reference rage, 40% to 75%) and 16% lymphocytes (reference rage, 20% to 50%). He was positive for HCV-antibody but HCV RNA was undetectable in his blood. Blood biochemistry showed an increased level of serum creatinine (501.5 *μ*mol/liter; reference rage, 40 to 160 *μ*mol/liter) and urea nitrogen (28.4 mmol/liter; reference rage, 2.2 to 8.3 mmol/liter) and a decreased level of albumin (25.4 grams/liter; reference range, 40 to 55 grams/liter). His CD4 cell count was 111 cells/microliter (reference range, 414 to 1123 cells/microliter) and HIV RNA level was 4.23 × 10^5^ copies/milliliter (reference range, \<20 copies/milliliter). Under the impression that he had severe septicemia, we initiated early goal-directed therapy and 4 separate blood specimens were sampled at 1-hour intervals consecutively for bacterial culture before empirical antibiotics were given. The patient was treated with 1 g cefoxitin every 6 hours and 10 mg dexamethasone every 12 hours intravenously, combined with blood transfusions and erythropoietin injections. The patient\'s condition stabilized the next day and his body temperature returned to normal 3 days later. The cefoxitin treatment was given for another 7 days and was discontinued when his white blood cell count returned to 2.51 × 10^9^ cells/liter (reference rage, 3.5 × 10^9^ to 9.5 × 10^9^ cells/liter) and two posttreatment blood cultures yielded negative results. Three weeks after this admission, his blood creatinine and urea nitrogen levels normalized, his anemia and thrombocytopenia improved, and he was discharged in a stable condition. No recurrence was reported on a follow-up 1 year after discharge. All four blood cultures grew Gram-negative rods. In our laboratory, we used the MicroScan Walkaway (Siemens, Memphis, TN) for the identification of the organism and the results of a series of biochemical assays revealed it to be*Y. regensburgei* with 97.8% probability. It was positive for glucose, sorbitol, rhamnose, L-arabinose, melibiose, lysine, ornithine, citrate, galactoside, and cellobiose tests and negative for sucrose, raffinose, inositol, adonitol, urea, hydrogen sulfide, indole, arginine, tryptophan deaminase, aesculin, acetoin, malonate, tartrate, acetamide, cetrimide, and Voges-Proskauer tests. Susceptibility testing was performed using the MicroScan Walkaway and the Enterobacteriaceae criteria of CLSI were used by our laboratory \[[@B1]\]. The MIC results as shown in [Table 1](#tab1){ref-type="table"}. 3. Systematic Review {#sec3} ==================== We searched PubMed (<https://www.pubmed.com/>) with the keyword of*Yokenella regensburgei* on March 31, 2017, and found 10 articles. And then we used*Koserella trabulsii* as the keyword to search on PubMed on the same day and found another 6 articles. We reviewed all the 16 articles and found seven cases of*Y. regensburgei* infection as well as useful information about*Y. regensburgei*. *Y. regensburgei* is one of a number of infrequent members of the family*Enterobacteriaceae* that have only rarely been isolated in humans. It was originally identified by Kosako et al. \[[@B2]\] through DNA hybridization in 1984 and later recognized by Hickman-Brenner et al. \[[@B3]\], working independently, in 1985, under the name of*Koserella trabulsii*. Subsequently, it was found that*Y. regensburgei* and*K. trabulsii* referred to the same*Enterobacteriaceae* and therefore the use of*K. trabulsii* has been dropped since 1991 \[[@B4]\]. *Y. regensburgei* closely resembles*Hafnia alvei* biochemically and has been misidentified as*Hafnia alvei* by automated systems \[[@B4]\]. It, therefore, has been hypothesized that infections due to*Y. regensburgei* have been underestimated due to misidentification of the bacterium. By studying susceptibility patterns and biochemical properties, Stock et al. \[[@B5]\] found that hydroxyproline amidase, maltosidase, tripeptidase, proline deaminase, catalase reaction, Voges-Proskauer test, and fermentation of glycerol, melibiose, and myo-inositol were suitable parameters to separate*Y. regensburgei* from*H. alvei*.*Y. regensburgei*is noted to possess amp C genes and express highly inducible, potent beta-lactamases and is intrinsically resistant to azithromycin and some beta-lactam antibiotics. However, it is weakly catalase positive and unable to produce hydroxyproline amidase, tripeptidase, or proline deaminase. Jachymek et al. \[[@B6]\] discovered novel trisaccharide repeating units of bacterial O antigens that are characteristic and unique to the*Y. regensburgei*species. Niedziela et al. \[[@B7]\] described the structures of the core oligosaccharides representing novel core types of bacterial LPS that are characteristic for*Y. regensburgei*. *Y. regensburgei* appears to primarily belong to the bacterial flora of insects and has been recovered from the intestinal tracts of insects. Also, it has been isolated from the general environment such as in well water. Isolation of*Y. regensburgei* from a human specimen is rare and only seven cases of*Y. regensburgei* infection have been reported worldwide based on the literature we reviewed. Abbott and Janda \[[@B8]\] described two isolations of*Y. regensburgei* associated with extra-intestinal sites in humans immunocompromised due to alcohol abuse. The first isolate was from a left-knee wound of a 74-year-old male with a provisional diagnosis of a septic knee and a history of alcohol abuse. He was treated with amikacin, and no further data were available regarding the patient\'s clinical course. In the second case, a 35-year-old woman who abused alcohol suffered from an upper gastrointestinal bleeding. A blood culture grew*Y. regensburgei* during her hospital course, although she had no overt signs of sepsis such as fever or chills. The patent was subsequently treated with ciprofloxacin and released. Lo et al. \[[@B9]\] described a patient with membranous glomerulonephritis on immunosuppressive therapy with high-dose steroids (prednisolone total of 30 mg per day) and cyclophosphamide. The patient subsequently developed a soft tissue infection after abrasions to his leg, which had been contaminated with soil. He experienced fevers with chills, and his blood cultures grew*Y. regensburgei*. After receiving treatment with ceftriaxone for 3 weeks, he was discharged from hospital in stable conditions. Bhowmick and Weinstein \[[@B10]\] reported a 48-year-old male with multiple myeloma who had undergone an autologous stem cell transplant and had been on corticosteroids for liver disease. Later, the patient had a bulla on his right leg but had no fevers and chills. The bulla aspirate and 2 blood cultures grew*Y. regensburgei.* Fajardo Olivares et al. \[[@B11]\] described an 82-year-old male with chronic renal failure, venous thrombosis, and perimalleolar ulcer.*Y. regensburgei*was isolated from his ulcerous wound and he was cured with ciprofloxacin. Penagos et al. \[[@B12]\] presented a case of postsurgical secondary osteomyelitis due to*Y. regensburgei* in an immunocompetent woman who had undergone a craniotomy. The patient was successfully treated with ciprofloxacin for 42 days and there was no recurrence of infection at the end of 1-year follow-up. Jain et al. \[[@B13]\] reported a 5-year-old male child with continuous high-grade fever and chills for 7 days. Two blood cultures yielded positive results and the identification of both isolates was confirmed as*Y. regensburgei*. The child was treated with ciprofloxacin for 7 days and he responded clinically to the treatment. No recurrence was reported on a follow-up 3 months later. 4. Discussion {#sec4} ============= The patient in our present report, a confirmed HIV-infected individual with pancytopenia, was viremic in absence of ART when he was admitted due to a fever. The fact that*Y. regensburgei*was isolated from all 4 separate blood samples and the patient was successfully cured with cefoxitin demonstrated that the organism was definitely the cause of septicemia. Noticeably, the patient\'s white blood cell count was higher compared with his baseline values and the number decreased to its pretreatment levels after 10 days of antibiotic treatment, suggesting that*Y. regensburgei* infection could result in elevated white blood cell counts. We did not observe any evidence for the association between the patient\'s septicemia and his farming practice, although the organism could have been from the general environment like the soil. This is the first case of infection caused by*Y. regensburgei* in an HIV-infected patient to the best of our knowledge and our case supports the hypothesis that*Y. regensburgei*is an opportunistic pathogen in humans. However, we were not able to confirm the bacterial identification with sequencing, which is the limitation of this article. All the eight reported cases are summarized in [Table 2](#tab2){ref-type="table"}. Some common features can be observed from the seven reported cases and our case. Firstly, almost all cases have underlying diseases or immunocompromising conditions. Secondly, the majority of those cases have no systematic symptoms like fever and chills. Thirdly, the outcome of*Y. regensburgei* infection appears to be not too severe; only one death occurred among the eight cases, while five cases were cured. And lastly but interestingly, almost all the cases are from regions known for hot and humid weather. In summary, we report a case of*Y. regensburgei* septicemia in a patient with HIV infection and pancytopenia. There are some similarities among the eight cases with regard to underlying conditions, clinical presentations, outcome, and geographical regions. From all the eight cases worldwide, it is reasonable to assume that*Y. regensburgei* is an opportunistic pathogen with a predilection to infect hosts severely immunocompromised by underlying diseases or conditions. The authors thank Professor Jean-Pierre Routy of Centre Universitaire de Santé McGill for his suggestions about language improvement and case discussion. This work was funded by Chongqing Municipal Health and Family Planning Committee (2016HBRC008). Ethical Approval ================ The Chongqing Public Health Medical Center Ethics Committee reviewed and approved this case report. Consent ======= Informed consent was obtained from the patient described in the article. Conflicts of Interest ===================== The authors declare that they have no conflicts of interest. ###### Susceptibility of the isolated *Y. regensburgei*stain. Drugs MIC Susceptibility ------------------------- ------- ---------------- Amikacin ≤8 S Ampicillin \>16 R Amoxicillin/clavulanate ≤8/4 I Aztreonam ≤8 S Ceftriaxone ≤8 S Ceftazidime ≤1 S Ceftazidime/clavulanate ≤0.25 S Cefotaxime ≤2 S Cefotaxime/clavulanate ≤0.05 S Cefoxitin ≤8 S Cefazolin ≤8 S Ciprofloxacin ≤1 S Cefepime ≤8 S Piperacillin/tazobactam ≤16 S Ertapenem ≤2 S Gentamicin ≤4 S Imipenem ≤1 S Levofloxacin ≤2 S Meropenem ≤1 S Cefuroxime ≤4 S Piperacillin \>64 R Cotrimoxazole ≤2/38 S Tetracycline ≤4 S Ticarcillin/clavulanate ≤16 S Tobramycin ≤4 S S: susceptible; I: intermediate; R: resistance. ###### Summary of *Y. regensburgei*infection cases reported worldwide. --------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- Reference Patient sex Patient age (years) Geographical region Probable risk factor Clinical specimen Clinical diagnosis Treatment Outcome -------------------------------- ------------- --------------------- --------------------- -------------------------------------------------------------------------------------- ------------------------ ----------------------- ------------------------------------------------- -------------------------- Abbott and Janda (1994) M 74 California, USA Alcohol abuse Left knee wound Septic knee Amikacin Unknown Abbott and Janda (1994) F 35 California, USA Alcohol abuse, liver disease, pancreatitis Blood Transient\ Ciprofloxacin Discharged, no follow-up bacteraemia Fajardo Olivares et al. (2005) M 82  Spain Chronic renal failure, venous thrombosis Wound Perimalleolar\ Ciprofloxacin Cured ulcer Lo et al. (2011) M 42  Taiwan Type 2 DM, renal disease,\ Blood Cellulitis, sepsis Ceftriaxone Cured steroids, immunosuppressants Jain et al. (2013) M 5 New Delhi, India None Blood Enteric fever Ciprofloxacin Cured Bhowmick and Weinstein (2013) M 48 New Jersey, USA Multiple myeloma, autologous stem cell transplant, corticosteroid use, liver disease Blood, bulla aspirates Soft tissue infection Imipenem/cilastatin, clindamycin and gentamicin Expired Penagos et al. (2015) F 70  Medellin, Colombia Invasive pituitary macroadenoma, neurosurgery Bone Osteomyelitis Ciprofloxacin Cured Index case M 38  Chongqing, China Acquired immunodeficiency syndrome, pancytopenia, drug abuse Blood Septicemia Cefoxitin Cured --------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- [^1]: Academic Editor: Oguz R. Sipahi

Nature Communications 4: Article number: 2673 10.1038/ncomms3673 (2013); Published: 10232013; Updated: 01292014. The original version of this Article contained an error in the accession codes section, which specified an accession code for an unrelated nucleotide sequence. The correct GenBank/EMBL/DDBJ nucleotide database accession code for *Puccinia striiformis* f. sp. *tritici* CY32 is [ANHQ00000000](ANHQ00000000). This has now been corrected in both the PDF and HTML versions of the Article.

Background ========== Selection for resistance to malaria among inhabitants of malaria endemic regions may have influenced polymorphisms in genes encoding a variety of proteins involved in immunity \[[@B1]\]. For example, different subclasses of immunoglobulin G (IgG isotypes) have been proposed to play opposing roles in protection against malaria \[[@B2]\]. Cytophilic IgG (IgG1 and IgG3) antibodies were shown to be protective, while non-cytopihlic ones (IgG2 and IgG4) were found to be competing with the former isotypes \[[@B2],[@B3]\]. Thus, not only levels, but also switching between IgG isotypes is believed to play a role in development of protective immunity. Protein polymorphism within the individual IgG subclasses is in part due to GM/KM allotypes, which are genetically determined serologically detectable antigenic determinants. These allotypic determinants are expressed on both the heavy and light chains of IgG1, IgG2, and IgG3. The combination of individual alleles is referred to as a haplotype \[[@B4]\] and GM haplotypes vary among ethnic groups \[[@B5]\]. KM gene frequencies also vary significantly among various ethnic groups. However, the deployment of GM/KM allotyping for human population genetic analysis, mapping global haplotype distributions, indicated that selection on GM haplotypes is low at the human population level \[[@B6]\]. It has also been reported that the levels of the IgG subclasses are influenced by the GM allotypes in adult Caucasian blood donors \[[@B7]\] and in African American populations \[[@B8]\]. The association of GM/KM allotypes with susceptibility to several different diseases has been reported \[[@B9]\] and their involvement in autoimmune disease has also been proposed \[[@B10]\]. Some data have also indicated a possible association of GM/KM allotypes with malaria morbidity and severity \[[@B11]\]. Differences between ethnic groups in the distribution of GM/KM allotypes and a possible association with malaria susceptibility were recently demonstrated in a study carried out in eastern Sudan involving comparison of groups of West African Fulani origin with indigenous sympatric tribes \[[@B12]\]. At present, there is limited evidence for the involvement of human IgG allotypes leading to functional differences in IgG antibodies as compared to the evident differences seen between IgG subclasses in malaria \[[@B2]\]. In the current study, a hypothesis suggesting that the GM/KM make-up of individual immunoglobulin affects IgG isotype levels, depending on target malaria antigen, was examined. Consequently, the GM/KM allotypes might influence the host susceptibility to malaria. Therefore, ten GM (1, 2, 3, 5, 6, 13, 14, 17, 21, 23) and 2 KM (1, 3) allotypes were investigated and combined with nine years of longitudinal malaria incidence data collection. In addition, baseline antibody response to four leading asexual blood-stage malaria vaccine-candidate antigens, comprising the apical membrane antigen-1 (AMA-1), merozoite surface protein-2 (MSP-2; 3D7 and FC27 alleles), and Pf332-C231, was analysed to test this hypothesis. Results revealed that, development of protective immunity is not only attributed to repeated exposure with increasing age \[[@B13]\], but also to genetic polymorphisms of the IgG in terms of GM/KM phenotypes. Methods ======= Study area ---------- The study was carried out in Daraweesh village, eastern Sudan, where malaria is hypoendemic with occasional quite severe \'malaria seasons\'. The malaria transmission in the region is strictly seasonal but markedly unstable; in \'wet years\' it peaks in October/November, after the summer rain, although a few sporadic cases also occur between February and August. In Daraweesh, malaria affects all age groups, although the incidence decreases after twenty years of age. A detailed geographical, demographic and social description of the area has previously been reported \[[@B14],[@B15]\]. Study population ---------------- The inhabitants of Daraweesh are descendents of a founder population of the West African Fulani-speaking group, originally from Burkina Faso. The village founders migrated to the Sudan more than hundred years ago although the ethnic identity (and Fulani language) has been maintained by frequent inter-marriage between descendents of the small founding group and some marriage with other Sudanese Fulani immigrants. In year 2000, the total population was approximately 500 inhabitants and during the 13-year study period, it ranged between, 400 to 600 individuals. The plasma samples used in this study were collected in the dry season in May 2005. All donors were malaria-free, no malaria parasite was detected by microscopy or PCR in any of the blood samples. Longitudinal clinical and parasitological surveillances ------------------------------------------------------- From 1991 to 2004, the inhabitants of Daraweesh were under regular clinical and parasitological malaria surveillance \[[@B14],[@B15]\]; except for four malaria seasons, where funds were not available or follow-up was not complete. Malaria was monitored by active and passive surveillances by a research team over this period. The malaria was diagnosed if a patient had fever (oral temperature \>37.5 °C), or reported fever, and had microscopically-detectable parasitaemia after examination of a maximum of 200 fields under oil immersion on each slide. Malaria patients were treated with chloroquine or with pyrimethamine/sulphadoxine in case of chloroquine treatment failure or by a combination of both as in year 2003 (therapeutic trial). Notably, severe malaria incidence was extremely low during the entire study period. Although prophylaxis was not used, presumptive treatment was sometimes practiced. Blood samples ------------- Peripheral blood samples (3 ml) were collected from each donor into a vacuum EDTA tube. After centrifugation the plasma was collected into cryotubes and stored at -20°C until used. The study had received ethical clearance from the ethical committee at the Faculty of Medicine, University of Khartoum, and the Sudanese Federal Ministry of Health. Consent was obtained from the village members regarding the whole project and from the individual donors, when blood samples were collected. Measurement of total IgG and IgG isotypes by enzyme-linked immunosorbent assay (ELISA) -------------------------------------------------------------------------------------- An indirect enzyme-linked immunosorbent assay, ELISA, \[[@B16]\] was used for measurement of total IgG and the subclasses; IgG1, IgG2, IgG3 and IgG4, against four asexual blood-stage antigens; MSP2-3D7, MSP2-FC27, AMA-1 and Pf332-C231 \[[@B17]\]. Briefly, EIA/RIA plates (Costar, MA, USA) were coated with AMA-1 at 1 μg/ml, 3D7 MSP 2 and FC27 MSP 2 at 1 μg/ml, and Pf332-C231 at 5 μg/ml. The AMA-1 protein, which has an N-terminal hexa-His tag, was expressed in *Escherichia coli*and refolded *in vitro*. Both the 3D7 and FC27 forms of MSP 2 were expressed in *E. coli*with C-terminal hexa-His tags. The expression and purification of these proteins will be described elsewhere (manuscript in preparation). The recombinant Pf332-C231 corresponds to a 231-amino acid fragment in the C-terminal part of Pf332. The plates were incubated overnight at 4°C, and then blocked for 2 hrs with 0.5% bovine serum albumin (BSA) diluted in carbonate buffer (pH 9.6). Plasma samples diluted in incubation buffer (PBS + 0.5% BSA), 1:1,000 (IgG) and 1:400 (IgG1-4), were added in duplicate and incubated for 1 h at 37°C. The plates were then washed four times, and bound IgG antibodies were detected with goat anti-human IgG-ALP (1:2000) (Mabtech, Nacka, Sweden). IgG subclasses were analysed with their respective biotin conjugated mouse anti-human subclass specific monoclonal antibodies: mouse anti-human IgG1 1:1,000 (M15015, Clone NL16, SkyBio, Bedfordshire, UK), mouse anti-human IgG2 1:3,000 (555874, Pharmingen, Erembodegem, Belgium), mouse anti-human IgG3 1:1,000 (MH 1532, Caltag laboratories, Paisley, UK) and mouse anti-human IgG4 1:2000 (B3648, Sigma, St. Louis, USA). Alkaline phosphatase (ALP) conjugated streptavidin (Mabtech) diluted 1:2,000 was added to detect bound antibodies of IgG2-4, while ALP-conjugated to goat anti-mouse Ig (Dakopatts, Glostrup, Denmark; 1:1,000) were used for IgG1 antibodies and the plates were developed with nitrophenyl phosphate (Sigma-Aldrich Chemie GmbH, Steinheim, Germany). The absorbance was read at 405 nm using a Vmax™ Kinetic microplate reader (Molecular devices, Menlo Park, USA) and the antibody concentrations were deduced from the log-log correlative coefficient of each IgG subclass standard curve (six dilutions of myeloma proteins of IgG1-4 subclasses), ranging from 0.01 to 3 μg/ml for IgG1, 0.001 to 0.3 μg/ml for IgG2, 0.001 to 0.1 μg/ml for IgG3 and 0.01 to 1 μg/ml for IgG4 according to the manufacturer\'s recommendation (Biogenesis, Poole, England). In plots, the antibody concentrations were transformed into log 10 values without any changes in the P - values obtained from the analysis of the raw values. Determination of GM and KM allotypes ------------------------------------ Allotyping of the serum samples from the above subjects was carried out for G1M (1,2,3,17), G2M (23), G3M (5,6, 13,14,21), and KM (1,3) determinants by a standard haemagglutination-inhibition method \[[@B18]\]. Human erythrocytes (Group O Rh+) coated with anti-Rhesus antibodies of known GM allotypes and mono-specific anti-allotype sera were employed. Sera containing IgG of particular allotype inhibit haemaggultination, but negative sera do not. Statistical analysis -------------------- Sigma Stat software was used for data analysis. There is almost absolute linkage disequilibrium between individual GM alleles \[[@B19]\]; therefore, data were analysed as haplotypes and also according to presence or absence of individual markers. For comparisons of measurable parameters e.g. age, levels of antibodies and etc, between different categories grouped on basis of GM or KM allotyping, ANOVA or Kruskal-Wallis One Way Analysis of Variance on Ranks was used. For pair wise analysis of the same parameters between the same groups of donors, T-test or Mann-Whitney Rank Sum Test was used. Results ======= Frequency of GM and KM allotypes in Daraweesh --------------------------------------------- The individual GM allelic type (allotype) 5 and 17 were detected in all (100%) donors involved in this study. The vast majority of the donors had the GM 1 (99.5%) and GM 14 (98.6%) alleles. The GM 13 (82.4%) allotype was the next most frequent, while half of the population carried the GM 6 allotype (49.5%). The other allotypes; 3, 21 and 23, were rare (Table [1](#T1){ref-type="table"}). The donors were categorized into five groups based on their GM phenotypes (combination of allotypes), four predominant phenotypes and a fifth group which combines six minor groups taken together and referred to as \"others\". Most of the donors were carrying the four most frequent phenotypes; 1,17 5,13,14 (40.3%) and 1,17 5,13,14,6 (38.4%), followed by 1,17 5,14 (9.3%) and 1,17 5,14,6 (6.9%), and the \"other\" rare phenotypes were altogether accounting for 5.1%, (Table [2](#T2){ref-type="table"}). The frequencies of the KM 1, 3, and 1,3 phenotypes were; 8.3%, 52.8% and 38.9%, respectively. The age distribution of both GM and KM allotypes were comparable between all study groups. The majority (65.7%) of the donors were females (Table [2](#T2){ref-type="table"}). ###### The frequency of the individual GM allotypes of IgG immunoglobulin in Daraweesh village. GM allotypes Proportions (number of individuals) -------------- ------------------------------------- 5 100% (216) 17 100% (216) 1 99.5% (215) 14 98.6% (213) 13 82.4% (178) 6 49.5% (107) 3 1.9% (4) 21 1.4% (3) 23 0.5% (1) ###### The frequency of the GM phenotypes (upper rows) and KM allotypes (bottom rows) of IgG immunoglobulin in Daraweesh village. GM/KM Phenotypes Frequency Age (years) Sex (F/M) ------------------ ---------------- -------------- ----------- **GM** Mean ± SD 1,17 5,14 9.3% (20) 20.2 ± 8.341 18/2 1,17 5,14,6 6.9% (15) 22.9 ± 16.1 8/7 1,17 5,13,14 40.3% (87) 26.2 ± 17.8 61/26 1,17 5,13,14,6 38.4% (83) 26.8 ± 17.0 48/35 Others \[n, 6\] 5.1% (11) 22.9 ± 16.9 7/4 P = 0.324 **KM** 1 8.3% (18) 23.4 ± 13.1 15/3 3 52.8% (114) 24.9 ± 17.2 67/47 1,3 38.9% (84) 26.8 ± 18.4 60/24 Kruskal-Wallis P = 0.872 Total 216 142/74 The age and sex of the donors are shown. The others are: 1,17 5,6 = **3**; 17 5,13,14 = **1**; 1,3,17 5,13,14 = **1**; 1,3,17 5,13,14,6 = **2**; 1,17 5,13,14,6,21 = **3**; 1,3,17 23 5,13,14,6 = **1** Association of IgG allotypes with susceptibility to malaria ----------------------------------------------------------- The average number of previous malaria episodes experienced by each of the donors over nine years was used as an index for differential susceptibility to malaria between the GM or KM allotype groups. The GM 1,17 5,13,14,6 phenotype carriers experienced the highest number of malaria episodes over the nine years of the follow up (mean ± SD, 1.9 ± 1.5, range 0-7), followed by 1,17 5,14,6 (1.5 ± 1.6) and 1,17 5,13,14 (1.5 ± 1.3) phenotype carriers, (Table [3](#T3){ref-type="table"}). The lowest number of malaria episodes was experienced by carriers of phenotype 1,17 5,14 (1.3 ± 1.5) and the carriers of the minor allotypes (1.4 ± 1.6). Taken all together, differences in the number of malaria episodes between all GM phenotypes were not statistically detectable, P = 0.273, Kruskal-Wallis one way analysis of variance on ranks. ###### The mean number and range of previous malaria episodes experienced by donors with different GM/KM allotypes/phenotypes, over 9-years of follow up in Daraweesh ----------------------------------------------------------------------------------------------------------------- Donors grouping Number of malaria episodes P-values ------------------ ---------------------------- ---------- ------------------------------------------------------ **GM allotypes** 1,17 5,14 1.3 ± 1.5 0 - 5 P = 0.273\ Kruskal-Wallis One Way Analysis of Variance on Ranks 1,17 5,14,6 1.5 ± 1.6 0 - 5 1,17 5,13,14 1.5 ± 1.3 0 - 4 1,17 5,13,14,6 1.9 ± 1.5 0 - 7 **KM allotypes** 1 (18) 1.7 ± 1.2 0 - 3 P = 0.342\ Kruskal-Wallis 3 (114) 1.7 ± 1.5 0 - 7 1,3 (84) 1.5 ± 1.4 0 - 6 ----------------------------------------------------------------------------------------------------------------- However, grouping of the donors based on presence and absence of certain set of GM allotypes (phenotypes), showed that carriers of 1,17 5,13,14,6 phenotype as compared to the non-carriers, had experienced statistically discernibly increased number of malaria episodes over the years of the follow up (P = 0.037). The carriers of the other three major GM phenotypes as compared with the non-carriers of the corresponding phenotypes were equally susceptible to malaria (Figure [1](#F1){ref-type="fig"}). To test whether an individual allotype was associated with increased susceptibility, the analysis was repeated comparing the allotype 6 (1.8 ± 1.5, episodes) versus non-6 allotype (1.5 ± 1.4) carriers, P = 0.148; and allotype 13 (1.7 ± 1.4) versus non-13 allotype (1.3 ± 1.5) carriers, (P = 0.123). The other individual allotypes were either present in almost all combinations e.g. allotype 17, 5, and 1 and 14, or were very rare e.g. allotypes 3, 21 and 23. However, the KM allotype carriers experienced a comparable number of malaria episodes during the longitudinal follow up, whether the comparison was made between the three KM allotype carriers; 1, 3 and 1,3, together or in pairs (Table [3](#T3){ref-type="table"}). {#F1} Total IgG and IgG isotype concentrations in relation to GM allotypes -------------------------------------------------------------------- The levels of the total IgG and IgG isotypes against the four tested malaria antigens showed marked variation between the isotypes, and also for the same isotype between the antigens. For a specific IgG isotype, the concentration varies between individuals with different GM allotypes, depending principally on the target antigen. Comparing the carriers of specific phenotype with the non-carriers of that phenotype was sometimes statistically significant. The carriers of the 1,17 5,13 14,6 phenotype had the highest frequency of high levels of IgG and IgG isotypes in their sera, followed by the carriers of the 1,17 5,13,14 and 1,17 5, 14. However, there was no significant difference between the carriers and non-carriers of the 1,17 5, 14,6 phenotype (Figure [2](#F2){ref-type="fig"}). {#F2} Total IgG and IgG isotype concentrations in relation to KM allotypes -------------------------------------------------------------------- The above comparisons of the IgG and its isotype concentrations to the four target antigens between allotypes carriers were repeated, but using KM (1, 3, and 1,3) instead of GM allotypes. Results showed that there was less influence of the KM allotype on the total IgG and IgG isotypes concentration, regardless of the target antigen. The only significant differences were the higher concentration of total IgG to MSP2-FC27 and to Pf332-C231 antigens in the plasma obtained from the KM 3 compared to that obtained from the KM 1,3 phenotype carriers (Figure [3](#F3){ref-type="fig"}). {#F3} Discussion ========== In malaria hyper-endemic settings, protection from malaria mortality and morbidity is achieved with age as a consequence of immunity and this protection is augmented in subpopulations by inherent, genetically controlled host factors. The genetic factors involved in resistance to malaria are clearly numerous, the best known involving various balanced polymorphisms in haemoglobin genes. In this study, the number of malaria episodes over nine years of clinical and parasitological follow-up was recorded for a cohort of around half of the Daraweesh village inhabitants. The data showed who was most susceptible/resistant to malaria infection. Integration of this data together with genotyping of GM/KM allotypes and measurement of total IgG and IgG isotypes to four malaria antigens revealed interesting findings. The study showed that, the GM allelic combinations but not the KM allelic combinations had significant influence in uncomplicated malaria susceptibly and in the type and concentration of IgG isotype. While the implication of GM allotypes in susceptibility to bacterial infections and other diseases have been known for some time \[[@B20],[@B21]\], the role of GM/KM polymorphisms in malaria susceptibility has only recently been investigated \[[@B12],[@B22]\]. In this study, the GM 1,17 5,13,14,6 phenotype was significantly associated with increased susceptibility to malaria infections although paradoxically it was significantly associated with the highest baseline concentration of certain IgG and IgG isotypes. Both associations were age independent, and the latter one was largely dependent on the target antigen. GM allotypes have been assessed in all villagers tested and the frequencies of each GM or KM allotype determined. Nearly all villagers have GM allotypes 5, 17, 1 and 14 in their GM phenotype. This leaves allotypes 13 and 6 as the major GM allotype differences between villagers, both of which are located within the G3 constant region. Allotypes 1 and 17 are believed to be located in the G1 constant region, and do not, therefore, differ in 215 of the 216 villagers. The fact that the GM differences in this cohort are almost entirely in the G3 constant region and the significance of the GM polymorphisms in terms of possible differences in antibody functionality between the susceptible (i.e. 13, 6 containing allotypes) and the less susceptible (i.e. phenotypes not containing 13, 6 allotypes), worth further investigation. In this study, the GM 1,17 5,14 carriers, compared with the carriers of the other phenotypes or the non-carriers of 1,17 5,14 phenotype, had experienced the least number of malaria episodes although the differences were not significant. The significantly raised antibodies in the plasma of the carriers of the GM 1,17 5,13,14,6 phenotype, were mainly the total IgG and IgG2, but not the IgG3 isotype. The levels of IgG4 isotype to all antigens were generally very low, not associated with age and apparently were not increased after acute infection (unpublished data). The allotype-specific variation in the levels of IgG subclasses was antigen-dependent. This finding is supporting and adding to a previous report suggesting inherent ability of individual malaria antigens in antibody class switching \[[@B23]\]. The GM 1,17 5,13,14,6 phenotype was the second most prevalent phenotype in the area. This same phenotype was found to be more dominant in the Massalit, a tribe presumed to be more susceptible to malaria than Fulani \[[@B12]\]. The above-mentioned implication of GM 1,17 5,13,14,6 phenotype, might be due to the GM 6 allotype component. The allotype 6 was detected in half of the Daraweesh inhabitants, while at a global level the GM phenotypes containing GM 6 was recognized only in Africans, but not in other populations \[[@B6]\]. Recently, it was shown that, there was an inverse relationship between carriage of GM 1,17 5,13,14,6 phenotype and occurrence of uncomplicated malaria in Benin \[[@B22]\]. That observation does not contradict this study findings, as in their study all individuals involved in the study were infected with malaria and there were small in age (less than 10 years). Furthermore, in the Benin study, the clinical grouping was based on a single observation, also, the immune profile of the patients was not known. In this study, the above limitations were surmounted. In addition, the diversity that could result from the differences due to ethnicity was overcome by having a single homogenous ethnic study group. The KM allotypes were not statistically associated with malaria susceptibility. The influence of GM/KM allotypes on antibody response is assumed to be due to alteration of antibody specificity through variable regions \[[@B24]\] or by modification of antibody constant region \[[@B25]\]. The data showed limited effects for the GM/KM phenotypes on the concentration of IgG isotypes. However, there was a tendency for an antigen-dependent pattern of responses. Having examined five (IgG and subclasses) antibody titres to four antigens in the same individual, lack of consistency in levels of any subclass to all examined antigens in any of the GM/KM phenotypes carrier groups, was an evidence for lack of an innate effect for allotypes in levels of subclasses without the influence of the antigens. The antigen-dependent IgGs responses in malaria, are shown to have opposing significances i.e. exposure or protection, in a recent study from Kenya \[[@B26]\]. Recently, it has also been found that, there is a higher prevalence than previously expected, of primary partial immune deficiencies including IgG isotypes and GM allotypes, in apparently healthy individuals \[[@B27]\]. Finally, the influence of GM allotypes (immunoglobulin C-region) in immunity is conventionally thought to be associated with the V-region, as a particular allotype might be in linkage disequilibrium with a particular V- region determinant \[[@B28]\]. Alternatively, GM6 and GM13 allotypes, in this study, could contribute to formation of idiotypes associated with lower immunological responsiveness. On the other hand, the C-region (namely CH2 and CH3 domains) could modify the immunity through its affinity to the FcγR, and there might be certain GM allotype interaction with particular FcγR variant leading to either increased or decreased immune response \[[@B28]\]. A work showing the association between GM allotypes and FcγR polymorphisms and another showing the association between the levels of total IgG and IgG subclasses to different antigens and protection from malaria in the same cohort, are in progress. Conclusion ========== This study shows that, the GM allotypes are involved in susceptibility to uncomplicated malaria. Furthermore, there are indications for involvement of GM and to a lesser extent of KM allotypes in control of total IgG and IgG isotype concentrations, depending on target malaria antigen. The GM 1,17 5,13,14,6 phenotype was associated with higher baseline levels of total IgG and non-cytophilic IgG2. While GM 1,17 5,14 phenotype was associated with least susceptibility and highest baseline IgG3 response. Finally, the allotype GM 6, which is believed to be absent in Europeans was detected in 50% of the study population (Fulani). Competing interests =================== The authors declare that they have no competing interests. Authors\' contributions ======================= HG, DA and TT planed, established, maintained and supported the Daraweesh nine-years clinical and parasitological data; HG, AN and GE supervised and collected the field samples; AR, NI, MTB, KB, GH and JP were responsible for the laboratory data, HG, DA, TT, MTB, KB, JP and GH, were responsible for the study design, data analysis, results interpretation and paper drafting, all authors contributed to manuscript writing Acknowledgements ================ We would like to thank all the inhabitants of Daraweesh in eastern Sudan, our field research teams, all researchers and students who established and involved in this project and the previous local health authorities who supported the project over the years. Professor Robin Anders is acknowledged for the supply of AMA-1, MSP-2 FC27 and 3D7 antigens while Halima Balogun prepared the recombinant Pf332-C231 antigen. The Daraweesh Project is built on several short and long term projects received funds from several donors principally DANIDA and TDR/WHO from 1990 to 2004. Riad Bayoumi, Thor Theander, Lars Hviid, David Arnot and Gwiria Satti, were the principal founders of the project. MTB and KB were supported by grants from the Swedish Agency for Research Development with Developing Countries (SIDA/SAREC), the Swedish Medical Research Council (VR) as well as grants within the BioMalPar European Network of Excellence (LSMP-CT-2004-503578), from the Priority I \"Life Sciences, Genomics and Biotechnology of Health\" in the 6th Framework Programme

Background ========== Porcine reproductive and respiratory syndrome (PRRS) is an economically relevant emerging swine viral disease that was first recognized in North America in 1987 \[[@B1]\] and in Europe in 1990 \[[@B2]\]. The causative agent of this disease, PRRS virus (PRRSV), was first isolated in the Netherlands in 1990 \[[@B2]\] and was designated Lelystad virus (LV). Subsequently, the same agent causing PRRS was also identified in the United States \[[@B3]\]. PRRSV is an enveloped single-stranded RNA virus of the *Arteriviridae*family, a member of the order *Nidovirales*\[[@B4]\]. PRRSV can cause severe reproductive failure in sows that is characterized by late-term abortion, stillbirth, and the birth of weak piglets; it is also associated with porcine respiratory disease complex in combination with secondary infections \[[@B5],[@B6]\]. In the US alone, the economic losses caused by PRRS amount to more than US \$560 million annually, and it is the most significant infectious disease currently affecting the swine industry worldwide \[[@B7]\]. The application of vaccines against PRRSV began in 1993 in Europe and 1 year later in North America. Current PRRSV vaccines have 2 forms: modified live and inactivated virus mixed with adjuvant \[[@B8]\]. In breeding swine, post-vaccination viremia can be induced by this live vaccine, and boars can shed live vaccine virus in their semen \[[@B9]\]. A commercial modified live virus (MLV) vaccine is used to control PRRS in many countries, but PRRS outbreaks are quite common in swine farms despite routine vaccination. These findings justify studying the use of inactivated PRRSV vaccines in breeding pigs. Because the protective immune response induced by attenuated vaccines is influenced by the genetic diversity of PRRSV, attenuated vaccines are not always effective against PRRSV that are genetically different from the vaccine virus strains \[[@B10]\]. For this reason, many researchers have tried to develop a killed virus vaccine that reflects the genetic diversity of PRRSV. Swenson et al. \[[@B9]\] show that use of a killed vaccine appears to reduce virus shedding in semen, but the difference in the number of days of shedding is not statistically significant compared to that of live vaccine. In contrast, Nielsen et al. \[[@B11]\] found that killed vaccine treatment has no effect on the level and duration of virus shedding in semen compared to live vaccine. Duran et al. \[[@B12]\] injected inactivated oily vaccine containing about 10^5.5^median tissue culture infectious doses (TCID~50~) per dose of a Spanish strain of PRRSV grown in porcine alveolar macrophages (PAMs) and subsequently challenged the cells with a live homologous strain. They showed that vaccinated animals devoid of antibodies, as determined by an immunoperoxidase monolayer assay (IPMA) at the time of challenge, were still protected from experimental PRRSV infection. On the other hand, despite the fact that Open reading frame 5 (ORF5) correlates well with the neutralizing antibody titer \[[@B13]\], the recombinant PRRSV ORF5 antigen vaccine did not produce serum-neutralizing antibodies and failed to show protection. Joo et al. \[[@B14]\] reported that sows of a PRRSV-positive herd with detectable serum neutralization (SN) antibody levels were not viremic after reexposure to PRRSV. Osorio et al. \[[@B15]\] showed that increased SN antibody titer is important, because the SN antibody response appears to be well correlated with resistance to infection. The SN antibody against PRRSV protects against viremia, virus replication in lungs \[[@B16]\], transplacental spreading of the virus, and reproductive failure \[[@B15]\]. Nilubol et al. \[[@B17]\] compared killed vaccine (KV)-inoculated infected pigs to infected pigs with and without KV inoculation. In this experiment, the SN titer was significantly higher in the KV-vaccinated groups after challenge than in the non-vaccinated groups. However, virus shedding was not affected. Misinzo et al. \[[@B18]\] reported a similar result in that the KV does not always induce neutralizing antibodies, but it enhances neutralizing antibodies upon viral challenge. Although KV induced faster antibody production after challenge, it failed to prevent the clinical signs associated with PRRSV infection, i.e., post-challenge viremia and transplacental infection of the piglets \[[@B19]\]. PRRSV-KVs induce poor immune responses in naïve pigs \[[@B17]\]. Without neutralizing antibody induction, KVs can result in a significant improvement of sow reproductive performance and litter characteristics \[[@B20]\]. When animals were challenged with heterologous PRRSV, the vaccine failed to protect gilts \[[@B19]\]. The efficacy of inactivated PRRSV vaccine has been seriously questioned \[[@B21]\]. An effective KV program is thought to produce variable results, according to the vaccine and vaccination strategy. Misinzo et al. \[[@B18]\] observed differences in the efficacy of inactivated vaccines depending on the virus strain and the cells used to prepare the vaccines. In addition, some types of adjuvants can be used as effective vaccine adjuvants to enhance the humoral and cellular responses of piglets to PRRSV \[[@B22]\]. Despite frequent vaccine use, there is little existing information about the protective efficacy or potency of PRRSV vaccines evaluated through *in vivo*infectious challenge with wild-type PRRSV. This study was undertaken to compare vaccine efficacy according to the virus antigen quantity and inactivation reagent by analyzing the virus titer after challenge, ELISA antibody titer, and VN titer. Methods ======= Virus production and plaque assay --------------------------------- PRRSV isolated from Jinwang farm in Chungcheongnam-do (virus isolation number: 08-296; Genbank accession number: [HM130677](HM130677)) and propagated in MARC-145 cells was used for experimental vaccine preparation. After 96 h culture in 2-L roller bottles, the cell culture supernatants were centrifuged at 6,000 rpm for 20 min. Virus titers were determined by plaque assay using MARC-145 cells. For the assay, the cells were pre-seeded in a 6-well plate at a density of 3 × 10^5^cells/well for 12-18 h and subsequently infected with serial 10-fold dilutions of virus for 1 h at 37°C with frequent agitation. The cell monolayers were then overlaid with minimal essential medium containing 0.5% SeaKem LE agarose (FMC BioProducts, Rockland, Maine) and 5% fetal bovine serum, followed by 4-day incubation at 37°C in air containing 5% CO~2~. The resultant plaques were visualized by fixation with 7% formaldehyde followed by staining with crystal violet (1% \[w/v\] in 5% ethanol). Virus inactivation ------------------ ### 1. BEI BEI was prepared as a 0.1 M stock solution by stirring 0.1 M 2-bromoethylamine hydrobromide (Sigma, USA) in 0.175 M NaOH at 37°C for 1 h, as described previously \[[@B23]\]. BEI was used shortly after its preparation. BEI stock solution was added to the virus suspension at 1% concentration to attain a final BEI concentration of 0.001 M. Virus suspensions were incubated at 37°C for 24 h. The remaining BEI was subsequently neutralized by the addition of 10% of the volume of BEI of 1 M sterile sodium thiosulfate (Sigma) solution for 2 h. ### 2. Formalin Formaldehyde solution (35%) (Duksan, Korea) was added to the viral suspensions to attain a final concentration of 0.3%. The formalin-treated viral suspensions were incubated overnight at 37°C as described previously {Habib, 2006 \#856}. ### 3. β-propiolactone The pH of viral suspensions was adjusted to 7.3 using NaOH. β-propiolactone was then introduced to a final dilution of 1:2000 (v/v) and mixed on a magnetic stirrer for 24 h, as described previously \[[@B24]\]. Following inactivation, the viral suspensions were gradually heated to 37°C at pH 7.0 over a period of 24 h. Preparation of vaccine formulations ----------------------------------- To confirm if the viruses were completely inactivated, 1 mL virus suspension was inoculated on MARC-145 cells in a 175-cm^2^tissue culture flask with 50 mL of Dulbecco\'s modified Eagle\'s medium (DMEM). After the cells were cultured for 5 days at 37°C, the supernatants were replaced with fresh culture media and incubated for another 5 days. Non-inactivated PRRSV was inoculated on MARC-145 cells as positive control. Cells were analyzed for cytopathic effects (CPE), and the supernatants were tested by polymerase chain reaction (PCR) to confirm the presence of the virus. After confirmation of viral inactivation, each viral antigen were mixed with 10% (v/v) aluminum hydroxide adjuvant (Rehydragel, SEPPIC, France) on a magnetic stirrer (at 180 rpm) for overnight. PRRS ELISA ---------- Serum samples were collected from pigs, and aliquots were prepared and stored at -20°C until ELISA for PRRSV antibodies was performed. ELISA was performed using a commercial kit (HerdChek^®^PRRS 2XR, IDEXX), according to the manufacturer\'s instructions. All reagents required for the assay were provided with the kit, and the assay was conducted at room temperature. The optical density of each well was measured at 650 nm using the Bio-Rad 680 microplate reader. The presence or absence of PRRSV antibody was determined by calculating the sample to positive (S/P) ratio. Samples were considered to be positive for PRRSV antibody if the S/P ratio was more than 0.4. PRRSV VN titer assay -------------------- VN titers were determined by SN test on MARC-145 cells. A 2-fold diluted serum sample was prepared, and an equal volume of virus solution with a titer of 100 TCID~50~/mL was added to each dilution and incubated for 1 h at 37°C. The serum-virus mixture was transferred to a 96-well plate containing a MARC-145 cell monolayer. The CPEs on the cells were analyzed for 7 days after inoculation. The VN antibody titer was defined as the reciprocal of the highest dilution that inhibited CPE in 50% of the inoculated wells. Quantitation of PRRSV with real-time PCR ---------------------------------------- Real-time quantitative PCR was performed using Bio-Rad iQ5 Real Time PCR Detection System. The 20-μL PCR mixture comprised 10 μL of the commercially available mastermix (iQ SYBR Green Supermix; Bio-Rad), 1 μL of the cDNA extract from the serum of each pig, 8.5 μL of RNAase-free water, and 0.25 μL of both forward and reverse primers. In addition, each reaction included PRRSV standards with progressive dilutions of 1:10, which generated the standard curve for the reaction. The following protocol was used: 5 min at 94°C for incubation and Taq activation, followed by 40 cycles of 30 s each at 94°C for denaturation, 30 s at 55°C for annealing, and 45°C for extension. PRRSV content in each sample was estimated by converting the value for the cycle threshold (Ct), which was determined with the Bio-Rad iQ5 qPCR software, to virus titer (copies/mL) by using the coefficient of correlation from the standard curve. Experimental design of animal studies ------------------------------------- All animals were serologically tested against PRRSV just before the experiment to confirm naïve herd status. All animal experiments complied with the current laws of South Korea. Animal care and treatment were conducted in accordance with the guidelines established by the Institutional Animal Care and Use Committee of Optifarm Solution. Twenty-eight SPF hairless white Yucatan miniature pigs were used in this experiment. All pigs were kept in a HEPA-filtered barrier facility. The pigs had free access to sterilized water and unmedicated sterilized feed. Twenty-eight SPF miniature pigs were randomly assigned to 7 treatment groups including a negative control group. The mock-vaccinated control group received 2 mL aluminum hydroxide adjuvant. The 10^4^, 10^5^, and 10^6^groups were vaccinated with 10^4^, 10^5^, and 10^6^PFU/mL inactivated PRRSV (by the BEI method), respectively, with 10% (v/v) aluminum hydroxide adjuvant (Rehydragel, SEPPIC, France). The BEI-, formalin-, and β-propiolactone-inactivated groups were vaccinated with PRRSV inactivated by the abovementioned compounds. The virus titer in the vaccine was 10^6^PFU/mL, which was mixed with 10% (v/v) aluminum hydroxide adjuvant. All the 7 groups were inoculated with vaccine or mock-vaccine 3 times 0, 21, and 42 days post inoculation (DPI). At day 70 after initial vaccination, all the animals (the mock, 10^4^, 10^5^, and 10^6^groups) were challenged intranasally with 10^3.4^TCID~50~field-isolated PRRSV (strain 09-1240 KH, ORF5; Genbank accession number: [GQ375443](GQ375443)). Blood was taken on 0, 3, 6, 9, 14, 21, 28, 35, 42, 49, 56, and 70 days after vaccination and on 0, 3, 4, 7, 10, 15, and 22 days after the challenge. Serum samples were collected and stored at -70°C before testing with ELISA, VN, and Q-PCR. Statistical analysis -------------------- ELISA S/P ratio, V/N test results, and viral load were analyzed by Student\'s *t*-test. A *p*value less than 0.05 was considered statistically significant. Results ======= Measurement of PRRSV-specific antibody levels by IDEXX ELISA after vaccination according to the viral titer ----------------------------------------------------------------------------------------------------------- For the experimental efficacy test according to the viral titer, PRRSV suspensions of titers of 10^4^, 10^5^, or 10^6^PFU/mL were inactivated with binary ethylenimine (BEI) and inoculated into pigs. From 21 to 56 days after the first vaccination, the 10^6^PFU/mL PRRSV vaccine-inoculated group showed a significantly higher sample to positive (S/P) ratio (*p*\< 0.01) than the control group (Figure [1](#F1){ref-type="fig"}). However, all groups were serologically negative as determined by IDEXX ELISA until 70 days of the first vaccination. The positive and negative cut-offs of the IDEXX ELISA are at 0.4 S/P ratio. The S/P ratio of the 10^6^PFU/mL PRRSV vaccine-inoculated group had the highest S/P ratio at 28 day post-infection (DPI) 28, 14 days after the second injection. This result indicated that 10^6^PFU/mL killed PRRSV antigen could not induce enough antibody determined by ELISA. {#F1} Variation of PRRSV ELISA S/P ratio after challenge according to the virus titers -------------------------------------------------------------------------------- PRRSV-specific antibodies were found to be absent by ELISA in all the experimental groups before challenge (Figure [1](#F1){ref-type="fig"}). Until day 4 after challenge, all the groups were serologically negative as determined by IDEXX ELISA (Figure [2](#F2){ref-type="fig"}). At day 7 after the challenge, all the vaccinated groups became serologically positive. The average S/P ratios of the 10^4^, 10^5^, and 10^6^PFU/mL PRRSV-vaccinated groups at 7 days were 0.82, 0.72, and 0.67, respectively. The average S/P ratios of the 10^4^and 10^5^PFU/mL vaccine-inoculated groups were not significantly different from the 10^6^PFU/mL vaccine-inoculated groups (p \> 0.05). However, all vaccine groups showed significantly different results from the control group only on the seventh day after challenge (*p*\< 0.01). This result indicates that, all vaccinated groups showed faster antibody production than the non-vaccinated group. {#F2} Viral load after challenge with killed vaccine according to viral titer ----------------------------------------------------------------------- The mean post-challenge serum PRRSV level is shown in Figure [3](#F3){ref-type="fig"}. The sensitivity threshold of this assay is 10 RNA copies/mL measured by quantitative PCR. {#F3} Because day-0 blood was sampled just before inoculation, the virus was not detected. By day 3 after challenge, all the vaccinated groups showed significantly different results from the control group (*p*\< 0.05). The viral load of the control group was lower than those of the vaccinated groups. On day 4, the viral load of the 10^6^PFU PRRSV-vaccinated group was significantly higher than that of the control group; no significant differences were observed compared to that of the other groups (p \< 0.05). At day 10, the viral loads of the 10^5^and 10^6^PFU/mL PRRSV-vaccinated groups were significantly reduced compared with those of the control and 10^4^groups (*p*\< 0.05). There was no difference in viral load between the control and 10^4^, 10^5^, and 10^6^PFU/mL PRRSV-vaccinated groups. By day 15, viruses were detected in the control and 10^6^PFU/mL PRRSV-vaccinated groups, but there was only a significant difference between the control and 10^4^PFU/mL PRRSV-vaccinated groups (*p*\< 0.05). PRRSV-specific neutralization titer according to BEI-inactivated viral titers from the samples on day 22 after challenge ------------------------------------------------------------------------------------------------------------------------ On the last day after challenge, serum samples were analyzed to determine the difference in the VN titer obtained with different values of the inoculated viral titer. Before challenge, all experimental groups were measured by a VN test and all groups were negative. Twenty-two days after challenge, the 10^6^viral antigen-inoculated group exhibited a significantly higher VN titer compared to the 10^4^and 10^5^PFU/mL virus-inoculated groups (p \< 0.01). In this present study, killed vaccine failed to induce antibody determined by ELISA. But BEI-inactivated vaccines showed higher VN titer according to viral titer. This result indicated that 10^6^PFU/mL viral titer of vaccine can induce higher VN titer (Figure [4](#F4){ref-type="fig"}). {#F4} PRRSV-specific ELISA antibody titer with experimental vaccines according to inactivation methods ------------------------------------------------------------------------------------------------ In order to test the efficacy of different virus inactivation methods, PRRSV suspensions were inactivated using BEI, formalin, or β-propiolactone. Until 70 days after the first vaccination, all experimental groups showed serological negativity (Figure [5](#F5){ref-type="fig"}), as determined by the IDEXX ELISA kit. The positive and negative cut-offs of the IDEXX ELISA were at an S/P ratio of 0.4. The BEI-inactivated group showed significantly different S/P ratios compared to the control group from 21 to 56 days after the first vaccine inoculation (*p*\< 0.01). The highest S/P ratio was observed in the BEI group 14 days after the second vaccination and 28 days after the primary vaccination (0.222 (0.1)). In the formalin group, from 28 to 42 days, only 1 pig (H08-010) had a serologically positive S/P ratio above 0.4, and its highest value in that period was 0.68. {#F5} Variation in PRRS ELISA S/P ratios after challenge with vaccines prepared by different virus inactivation methods ----------------------------------------------------------------------------------------------------------------- The PRRSV-specific antibodies were found to be absent by ELISA in all the experimental groups before challenge (Figure [5](#F5){ref-type="fig"}). Until day 4 after challenge, all the groups remained serologically negative, as determined by IDEXX ELISA (Figure [6](#F6){ref-type="fig"}). At day 7 after challenge, the average S/P ratio of all vaccinated groups became serologically positive, except for 1 and 2 pigs in the formalin-and β-propiolactone-inactivated groups, respectively. {#F6} On the seventh day after challenge, all vaccinated groups showed significantly different S/P ratios compared to the control group (*p*\< 0.01). The S/P ratio of the formalin-inactivated group on the tenth day was significantly different from that of the control group (*p*\< 0.005). There were no significant differences 15 days after challenge. Viral load of the KV-vaccinated group studied according to the different method of virus inactivation ----------------------------------------------------------------------------------------------------- The mean PRRSV load in the serum samples collected after the challenge is shown in Figure [7](#F7){ref-type="fig"}. The threshold of sensitivity of this assay was 10 RNA copies/mL measured by quantitative PCR. {#F7} The virus was not detected, because the blood samples were taken just before the challenge. From days 3-4, all the vaccinated groups, except the β-propiolactone-inactivated group at day 3, had significantly higher viral loads than the control group (*p*\< 0.05). Ten days after the challenge, the BEI-inactivated group had a lower viral load than the control group (*p*\< 0.05). Twenty-two days after the challenge, only the control and BEI-inactivated groups had virus-positive blood. PRRSV-specific neutralization titer according to virus inactivation method from the samples collected 22 days after challenge ----------------------------------------------------------------------------------------------------------------------------- On the last day after challenge, the serum samples were analyzed to determine differences in VN titer, according to inactivation method used to prepare vaccines (Figure [8](#F8){ref-type="fig"}). Before challenge, all experimental groups were measured by VN test, and all groups were negative. Twenty-two days after the challenge, the BEI-inactivated group had a significantly higher titer compared to the formalin- and β-propiolactone-inactivated groups (p \< 0.01). {#F8} Discussion ========== The total annual economic impact of PRRS on the US swine producers is estimated at \$66.75 million in breeding herds and \$493.57 million in growing pig populations \[[@B7]\]; the situation in South Korea is the same. Based on a field survey in 2001, 230 of 256 pig farms in Korea tested positive (89.8%) for PRRSV antibodies \[[@B25]\], indicating that PRRSV has spread throughout the country. Thus, for preventive purposes, MLVs costing as much as 2.5 billion KRW per year are used in South Korea. Despite the use of live vaccines, the prevalence of PRRSV has not decreased dramatically. This is because attenuated vaccines are not always effective against PRRSVs that are genetically different from the vaccine virus strain \[[@B10]\]. The PRRSV ORF5 sequence surveillance data since 2007 identified newly emerging PRRSVs (MN184-like) that have 84.9-87.2% nucleotide similarities compared to VR-2332, based on ORF5 sequence \[[@B26]\]. Recently, attenuated and inactivated vaccines have been introduced to control PRRSV in the field. Considering their safety and flexibility towards emerging virus strains, KVs are preferred over attenuated vaccines. However, in spite of these benefits, KVs have failed to evoke any measurable protective immunity \[[@B8]\] or protect against clinical signs such as post-challenge viremia and transplacental infection \[[@B19]\]. The first purpose of this study was to investigate the effects of experimentally KVs of varying virus titers on humoral immunity. In this experiment, PFU/mL was used as the viral particle-measuring method to obtain a more accurate count. The killed PRRSV vaccine was inoculated 3 times into SPF miniature pigs at 3-week intervals. Although all pigs were seronegative according to the criteria provided by the IDEXX corporation (S/P ratio \< 0.4), the KV group inoculated with 10^6^PFU/mL had significantly higher S/P ratios than the control group (p \< 0.05). However, no VN titers were detected in the entire experimental period before challenge (data not shown). Many researchers have tried to establish the effects of KVs on humoral immunity. Misinzo et al. \[[@B19]\] investigated whether neutralizing antibodies can be induced in pigs upon vaccination with an inactivated vaccine. The antibody measuring protocols, which differ from researcher to researcher, include SN titer \[[@B17]\], IPMA \[[@B18]\], serum immunoglobulin G \[[@B22]\], and ELISA antibody titer \[[@B8],[@B19]\]. In these studies, the effects of KVs on humoral immunity were undetectable or were detected at a very low level by using commercial ELISA systems. There might be a difference between the major antigenic epitopes of killed PRRSV vaccines and the ELISA system. However, Scortti et al. \[[@B19]\] reported that the first seroconversion resulted with the inoculation of Suvaxyn (Fort Dodge) KV. By day 21, all vaccinated gilts had seroconverted with a geometric mean titer (GMT) of 7.43 (0.44), and a peak of anti-PRRSV antibodies measured by ELISA was observed on day 70 with a GMT of 8.22 (0.54) \[[@B27]\]. These varying results allow the assumption that there are many obstacles to establishing effective KV combinations and well-matched diagnostic methods. In the challenge experiment of the present study, all vaccinated groups showed faster antibody production than the control group. Scortti et al. \[[@B19]\] reported that vaccinated groups produced detectable antibodies 5 days after challenge and 9 days before the unvaccinated group. In the present study, PRRSV-specific antibodies were detected at 7 days, that is, 3 days before the control group. This may imply that the killed vaccine can induce enough immune memory for PRRSV, but not enough to induce active humoral immunity. These findings are very similar to those of Zuckermann et al. \[[@B8]\]; only the challenge time is different (28 days post-primary vaccination versus 70 days in the present study). The more rapid development of a post-challenge serological response may confer some degree of protective immunity in pigs \[[@B19]\]. The key point of the present study is that all the vaccinated groups, which had different viral antigen concentrations ranging from 10^4^PFU/mL to 10^6^PFU/mL, induced the same pattern of more rapid humoral response. Only with this data, the 10^4^level of viral antigen inoculation appeared to be sufficient for inducing partial immunity. However, the VN titer measured 22 days after challenge showed that only the 10^6^PFU/mL virus-inoculated group had significantly greater neutralizing antibody titer than the other groups (p \< 0.05). Thus, if this viral neutralization response is related to protective immunity, it is also expected to affect the post-challenge serum viral load. In another study, KV (10^5.5^TCID~50~)-inoculated animals showed a shorter period of viremia, indicating that it is possible that the virus concentration in the blood of vaccinated animals is lower than that in the non-vaccinated animals \[[@B12]\]. Misinzo et al. \[[@B18]\] reported that the incomplete protection of KV vaccines against PRRSV might be caused by over-inactivation, resulting in the destruction of neutralizing viral epitopes. Virus inactivation procedures can affect the conservation of inactivated viral epitopes that are important for the induction of protective immunity \[[@B28]\]. In the present study, the ELISA S/P ratios of all vaccinated groups were significantly greater than those of the control group (p \< 0.01). Because BEI has an effect at the genomic level\--specifically for nucleic acids\--preserving viral epitopes \[[@B28]\], the BEI-inactivated group might exhibit a higher S/P ratio than the other groups. However, BEI had very little adverse effects on the epitopes, whereas BPL significantly altered and formalin partially altered the conformation of most epitopes \[[@B29]\]. All the vaccinated groups showed significant rapid elevation of antibody levels at day 7 (Figure [5](#F5){ref-type="fig"}), and the formalin-inactivated group showed significance at day 10 compared to the control group (*p*\< 0.01). However, in the VN test, BEI resulted in a significantly higher VN titer than other inactivation groups. At least in this experiment, the inactivation method is considered a key factor affecting the VN titer of inactivated PRRSV vaccines. Unexpectedly, in the challenge experiment after vaccination according to the viral titer, the vaccinated groups showed higher viral loads than the control group until day 4. This phenomenon was repeated in the challenge experiment after vaccination according to inactivation method. Scortti et al. \[[@B19]\] found higher viral loads than those in a mock-vaccinated group after 3 and 5 days. In contrast, \[[@B30]\] found lower viral loads in the vaccinated group than in the mock-vaccinated group. The reason why vaccinated groups exhibit significantly higher viral loads in blood is unclear. However, it is clear that while vaccination with inactivated viruses only has a small effect on viremia levels, it reduces the duration of viremia as reported by other researchers \[[@B18]\]. Conclusions =========== In conclusion, killed PRRSV vaccines with different concentrations of virus titer or based on different inactivation protocols did not significantly differ in their ability to induce humoral immunity in pigs even after 3 inoculations. However, all the vaccinated groups reached seropositive status in IDEXX ELISA much before the non-vaccinated group after challenge, which suggests that the KV merely potentiates memory response. Although the 10^6^PFU/mL-vaccinated and BEI-inactivated groups showed significantly greater VN titers 22 days after challenge, all the groups were already negative for viremia. Thus, the BEI-inactivated vaccine with a greater virus titer should be considered for testing to evaluate the efficacy of killed PRRSV vaccines. Competing interests =================== The authors declare that they have no competing interests. Authors\' contributions ======================= The formulation of the study was by JJ, YC, and HK. Animal experiments were by JK, CU and SH. Results were analyzed by SS. Revision by BL and GJ. Scripting was by BK, HM, and DS. All authors read and approved the final manuscript. Acknowledgements ================ This work was carried out with the support of \"Cooperative Research Program for Agriculture Science & Technology Development (PJ006777201004)\" Rural Development Administration, Republic of Korea and national project grant from SMBA of the Korean Government (S1036683).

Background {#s1} ========== The *Diagnostic and Statistical Manual of Mental Disorders Fifth Edition* (DSM-5) provides the diagnostic criteria ([Table 1](#i2168-9709-6-5-222-t01){ref-type="table"}) for major depressive disorder (MDD).[@i2168-9709-6-5-222-b01] Risk factors for depression include female gender, comorbid psychiatric disorders, family history of MDD, chronic medical diseases, unemployment, and lower socioeconomic status.[@i2168-9709-6-5-222-b02],[@i2168-9709-6-5-222-b03] The 12-month prevalence of MDD in older adults is lower compared with the general population, approximately 1% to 5% versus 7%, respectively.[@i2168-9709-6-5-222-b02] When providing care to the older adult, it is important to remember that depression occurs across the life-span, and it is not a normal part of aging.[@i2168-9709-6-5-222-b04] Another consideration when treating the older adult is that this age group is more likely to carry out lethal suicidal behavior.[@i2168-9709-6-5-222-b02] It should be a high priority to identify and treat the older adult with MDD, as adequate treatment of depression may help reduce future functional decline.[@i2168-9709-6-5-222-b03] ###### Diagnostic and Statistical Manual of Mental Disorders Fifth Edition^1^ major depressive disorder criteria  Major Depression in the Older Adult {#s1a} ----------------------------------- The evaluation and diagnosis of MDD in the older adult can be complicated by comorbid conditions or medications predisposing the patient toward development of depression, contributing to depression, or masking the symptoms of depression.[@i2168-9709-6-5-222-b05] These comorbid conditions and medications should be screened for initially, as treatment or optimization of these could improve outcomes without contributing to polypharmacy ([Table 2](#i2168-9709-6-5-222-t02){ref-type="table"}).[@i2168-9709-6-5-222-b06]-[@i2168-9709-6-5-222-b08] Treatment of underlying conditions can ensure the best possible response to the treatment of MDD with an antidepressant. ###### Non-depression causes of depression symptoms^a^  Symptoms of depression such as weight loss, appetite change, psychomotor retardation, loss of energy, fatigue, sleep changes, and decreased concentration may be difficult to identify in the older adult owing to comorbid conditions or decreased activity level. The older adult with MDD may have more somatic complaints such as pain or fatigue compared with a younger adult.[@i2168-9709-6-5-222-b02]-[@i2168-9709-6-5-222-b03] These differences in presentation can lead to misdiagnosis or underdiagnosis of MDD in the older adult.[@i2168-9709-6-5-222-b09] The general pharmacotherapeutic approach for older adults with MDD is to start low (50% of the adult starting dose) and go slow, titrating slowly to an effective dose.[@i2168-9709-6-5-222-b03],[@i2168-9709-6-5-222-b10] The goals of treatment for the older adult with depression are the same as for the general population: (1) achieve remission; (2) reduce relapse and recurrence; and (3) improve quality of life and functioning.[@i2168-9709-6-5-222-b03],[@i2168-9709-6-5-222-b05],[@i2168-9709-6-5-222-b06],[@i2168-9709-6-5-222-b11] Treatment of depression is usually divided into 3 treatment phases referred to as acute, continuation, and maintenance. The acute treatment phase occurs during the first 6 to 12 weeks of treatment with the goal of remission.[@i2168-9709-6-5-222-b11] The continuation phase follows the achievement of remission with the goal of preventing relapse typically lasts 4 to 12 months.[@i2168-9709-6-5-222-b05],[@i2168-9709-6-5-222-b11] If a patient requires further treatment after the continuation phase, they progress to the maintenance phase, with the goal of avoiding recurrence of depression.[@i2168-9709-6-5-222-b11] An older adult with a first episode of depression after the age of 60 may be a candidate for maintenance treatment.[@i2168-9709-6-5-222-b11] Other patients that may benefit from maintenance treatment include patients with 3 or more previous episodes of MDD, those with 2 episodes of MDD with rapid recurrence of episodes, and patients with severe episodes of MDD.[@i2168-9709-6-5-222-b11] When considering pharmacologic treatment duration in the older adult, a major consideration is the issue of polypharmacy, as 80% of older adults have at least 1 comorbid condition and 50% have at least 2.[@i2168-9709-6-5-222-b03] After the older adult has been treated for the appropriate duration, treatment discontinuation should be considered to minimize drug-drug interactions, adverse drug events, and contributions to polypharmacy and the prescribing cascade. Screening tools can be utilized to monitor treatment response and continued remission, aiding with determination of appropriate antidepressant dosage and duration of therapy. Screening the Older Adult for Depression {#s1b} ---------------------------------------- The Geriatric Depression Scale (GDS) and the Patient Health Questionnaire (PHQ-9) are 2 of the most commonly used screening tools to measure depressive symptoms in the older adult population ([Table 3](#i2168-9709-6-5-222-t03){ref-type="table"}).[@i2168-9709-6-5-222-b12]-[@i2168-9709-6-5-222-b14] The GDS was created specifically for community-dwelling older adult populations and has been used for older adults hospitalized for MDD. In contrast, the PHQ-9 was created for the general population, aged 13 and older.[@i2168-9709-6-5-222-b13],[@i2168-9709-6-5-222-b14] In comparison with the PHQ-9, the GDS may have added benefit for the older population owing to ease of use. The GDS utilizes a "yes" or "no" rating scale and requires the patient to recall symptoms from the past week. In contrast, the PHQ-9 uses a more complex 4-point scale and requires a 2-week recall period, which may be difficult for a patient with cognitive impairment.[@i2168-9709-6-5-222-b13] The content of the GDS has been modified to more closely capture symptoms of depression common in older adults, such as sadness, apathy, crying, and thoughts of hopelessness, helplessness, guilt, and worthlessness.[@i2168-9709-6-5-222-b13] The PHQ-9 and other depression scales created for the general population may overemphasize vegetative symptoms reflective of the activity level of many older adults.[@i2168-9709-6-5-222-b15] ###### Comparison of Geriatric Depression Scale and Patient Health Questionnaire-9^a^  The American Geriatrics Society (AGS) recommends screening for MDD in the elderly with the PHQ-2, a 2-item questionnaire utilizing the first 2 questions from the PHQ-9; an answer of yes to either question indicates a positive test. If an older adult has a positive PHQ-2, the AGS recommends a follow-up test using either the PHQ-9, found to possess 88% sensitivity and 92% specificity in elderly primary-care patients, or the 15-item GDS, found to have 92% sensitivity and 81% specificity in elderly primary-care patients.[@i2168-9709-6-5-222-b16]-[@i2168-9709-6-5-222-b18] A diagnosis of depression should be confirmed by the DSM-5 criteria.[@i2168-9709-6-5-222-b12] Both the PHQ-9 and GDS can be used during therapy to monitor for continued treatment response. Antidepressants Listed in the AGS 2015 Beers Criteria {#s1c} ----------------------------------------------------- The Beers Criteria highlight potentially inappropriate medications (PIMs) that are best avoided in the general older adult population and in those older adults with certain diseases or syndromes. If PIMs cannot be avoided, the older adult would likely benefit from prescribing with caution by using reduced doses with careful monitoring. For specific patients, a PIM mentioned in the Beers Criteria may be the best treatment option. As all antidepressants are represented within at least one Beers Criteria section, it is important to understand why these agents are considered PIMs and how to use them in the safest manner possible. The AGS 2015 Beers Criteria Update Expert Panel released an update in October. [Table 4](#i2168-9709-6-5-222-t04){ref-type="table"} summarizes the concerns and recommendations from the Panel regarding the use of antidepressants in the general older adult population, and [Table 5](#i2168-9709-6-5-222-t05){ref-type="table"} summarizes these concerns and recommendations for older adult patients with specific disease states or syndromes.[@i2168-9709-6-5-222-b19],[@i2168-9709-6-5-222-b20] Although this update was not as extensive as the previous 2012 update, the AGS 2015 Beers Criteria included 2 new lists. The first is a list of chronic medications to avoid or dose-reduce based on renal function that could be overlooked. This list includes the serotonin norepinephrine reuptake inhibitor duloxetine, recommending avoidance if creatinine clearance is \<30 mL/min because of the potential for increased gastrointestinal adverse effects, such as nausea and diarrhea. ###### Potentially inappropriate antidepressants for older adults listed in the American Geriatrics Society 2015 Beers Criteria^a^  ###### Potentially inappropriate antidepressants for older adults with specific disease states or syndromes as listed in the American Geriatrics Society 2015 Beers Criteria^a^  The second new list is a selection of non--anti-infective drug-drug interactions to be avoided in the older adult owing to a high association with negative outcomes. This list includes antidepressants, antipsychotics, benzodiazepines, and benzodiazepine receptor agonists when used in combination with 2 or more other central nervous system active drugs, because of the increased risk of falls and fractures.[@i2168-9709-6-5-222-b19] The remainder of this article discusses methods for increasing the safety of older adults treated with antidepressants. Monitoring Antidepressant Therapy for Older Adults {#s1d} -------------------------------------------------- Considering the commonality of depression and the need for treatment with antidepressants on the Beers Criteria, it is likely that an older adult patient will be prescribed a PIM that is the best pharmacotherapeutic option.[@i2168-9709-6-5-222-b19] It is important to note that the monitoring of antidepressant therapy is more extensive for older adults than younger adults. Compared with younger adults, older adults on antidepressant therapy have an increased risk of adverse effects such as falls, hyponatremia, and gastrointestinal upset resulting from comorbid conditions and pharmacokinetic and pharmacodynamic changes. The risk for drug-drug interactions is also increased owing to an increased occurrence of polypharmacy.[@i2168-9709-6-5-222-b09] The older adult may be at an increased risk of adverse effects due to pharmacokinetic or pharmacodynamic changes. In the older adult, elimination of the drug may be decreased due to changes in hepatic metabolism and renal elimination, resulting in accumulation of the drug and increasing the potential for adverse effects. The impact of these elimination changes should be carefully considered when determining dose initiation, dose titration interval, and maximum dose. These decisions should be based on the individual patient and specific medication, with consideration of maximum recommended dosages. For example, the maximum recommended dose of citalopram for a patient over the age of 60 is 20 mg daily because of the increased risk of QTc prolongation. Older adults have increased pharmacodynamic sensitivity to centrally acting medications than younger adults, resulting in efficacy at lower doses of centrally acting medications population-wide.[@i2168-9709-6-5-222-b09] The older adult may also be more sensitive to the anticholinergic actions of medications such as tricyclic antidepressants (TCAs) and the selective serotonin reuptake inhibitor (SSRI) paroxetine, increasing the risk for confusion, dry mouth, and constipation.[@i2168-9709-6-5-222-b09],[@i2168-9709-6-5-222-b19] The older adult may be at an increased risk of falls for a variety of reasons, including poor vision, hypotension, and comorbid conditions.[@i2168-9709-6-5-222-b09] Selective serotonin reuptake inhibitors and TCAs further increase the risk of falls, particularly in patients with a history of falls or fractures and should be used cautiously in these patients. Antidepressants increase fall risk owing to sedative effects, anticholinergic properties, and orthostatic hypotension, with TCAs and paroxetine receiving the most attention in the 2015 Beers Criteria.[@i2168-9709-6-5-222-b19] It is well known that initiation of antidepressants can cause dizziness, especially in older adults, but the precise mechanism for increased fall risk associated with long-term SSRI use remains unclear. It is possible that older adults with SSRI-induced hyponatremia could be at an increased risk of falls, as mild chronic hyponatremia has been associated with falls, unsteadiness, and inattentiveness.[@i2168-9709-6-5-222-b21] Depression, anxiety, and the fear of falling have all been linked to an increased risk of falls, which is a major confounder toward analyzing the relationship between taking antidepressants and having an increased risk of falls.[@i2168-9709-6-5-222-b22] Selective serotonin reuptake inhibitors and other serotonergic antidepressants have been theorized to increase fracture risk because of a serotonergic suppression of osteoblast proliferation, and the clinical relevance of this mechanism is a current avenue for research. One study utilized Medicare data and propensity score--matched cohorts to compare fracture rates among patients taking different antidepressants. The study found that in comparison with patients on secondary amine TCAs, the fracture rate was higher among patients taking highly serotonergic antidepressants, including SSRIs or venlafaxine (hazard ration \[HR\], 1.30; 95% confidence interval \[CI\], 1.12-1.52). This was in contrast to the comparatively similar fracture rate seen with atypical antidepressants, including duloxetine, mirtazapine, nefazodone, or trazodone (HR, 1.12; 95% CI, 0.96-1.31), and tertiary amine TCAs (HR, 1.01; 95% CI, 0.87-1.18).[@i2168-9709-6-5-222-b23] When attempting to reduce the risk of falls for patients taking antidepressants, the provider should review the medication list and limit the use of other medications that can also contribute to falls.[@i2168-9709-6-5-222-b19] A number of fall-risk assessments are available online and include lists of medications besides antidepressants that contribute to falls, such as analgesics, antipsychotics, anticonvulsants, benzodiazepines, antihypertensives, cardiac drugs, antiarrhythmics, and diuretics.[@i2168-9709-6-5-222-b24] The risk of hyponatremia, syndrome of inappropriate antidiuretic hormone, or both is increased in the older adult patient on antidepressant therapy. The risk is further increased if the patient is on another medication that can cause hyponatremia, such as a diuretic.[@i2168-9709-6-5-222-b09] Routine monitoring of serum sodium is not current standard practice for antidepressant therapy. In the older adult, the provider should consider obtaining a baseline serum sodium, repeating 1 month after initiation of therapy, or anytime during therapy if a patient presents with symptoms suggestive of low sodium (fatigue, dizziness, confusion).[@i2168-9709-6-5-222-b19],[@i2168-9709-6-5-222-b25] If hyponatremia is discovered, the offending agent should be discontinued. In the case of SSRI-induced hyponatremia, the hyponatremia is usually reversible, with most cases resolving 2 weeks after discontination.[@i2168-9709-6-5-222-b26] Escitalopram, citalopram, and sertraline are often considered first-line agents for treatment of depression in the elderly. However, diarrhea is a commonly reported side effect of these medications and can lead to therapy discontinuation.[@i2168-9709-6-5-222-b10],[@i2168-9709-6-5-222-b27],[@i2168-9709-6-5-222-b28] Diarrhea is a population-wide adverse effect of these agents, likely resulting from serotonergic actions on the gastrointestinal system.[@i2168-9709-6-5-222-b27] The elderly patient could be particularly affected by this adverse effect owing to decreased mobility, increased rates of incontinence, decreased ability to compensate for dehydration, and taking additional medications that cause diarrhea. Although not mentioned in the Beers Criteria, the older adult treated with SSRI or serotonin norepinephrine reuptake inhibitor therapy may also be at an increased risk of gastrointestinal bleeding as a result of decreased platelet aggregation, especially if taken in combination with nonsteroidal anti-inflammatory drugs, corticosteroids, anticoagulants, or antiplatelet agents such as aspirin or clopidogrel.[@i2168-9709-6-5-222-b09],[@i2168-9709-6-5-222-b10] Similar to hyponatremia monitoring, it is not standard practice to monitor a complete blood count in patients initiating SSRI therapy. The older adult should be monitored for any signs or symptoms of bleeding throughout the duration of therapy. Conclusion {#s2} ========== Major depressive disorder is not a normal part of aging and can occur across the life-span. Older adults should be screened for depression with the PHQ-2; if positive, either the PHQ-9 or GDS can be used for further screening. Comorbid medical conditions, other medications, and recommendations included in the AGS 2015 Beers Criteria should be considered when making pharmacotherapy treatment decisions. Although antidepressants are included in the 2015 Beers Criteria as PIMs, an antidepressant agent may be the most effective pharmacotherapeutic option for the patient. When the benefits of therapy outweigh the risks, these agents can be used safely as long as the patient is monitored appropriately. **Disclosures:** The authors have nothing to disclose.

Introduction ============ The intervertebral disc (IVD) is the largest avascular connective tissue in the human body. It is composed largely of a collagenous extracellular matrix that is sparsely interspersed by IVD cells (\~6,000 mm^-3^) \[[@B1]\]. The activity of IVD cells is important to tissue function, and IVD cell death, along with decreased IVD cellularity, is associated with ageing and degenerative disc pathology \[[@B2],[@B3]\]. Abnormal cell behaviour seen in pathological human IVDs also includes increased cell proliferation and cluster formation \[[@B4]\], the appearance of stellate cells \[[@B5]\], and cell senescence \[[@B6]-[@B8]\]. The avascularity of the disc has long been considered to influence IVD cell function \[[@B1]\], a view supported by *in vitro*studies. Hence, Urban and colleagues \[[@B9]\] demonstrated that reduced levels of glucose and oxygen, combined with other environmental conditions present in the inner parts of human IVDs (that is, decreased pH and increased osmolarity), decrease the anabolic activity of IVD cells. Glucose deprivation, in particular, also results in increased IVD cell death, an effect that increases with increased cell density (see reviews \[[@B9],[@B10]\]). There are also clinical and experimental indications that factors affecting vascular function or glucose supply are associated with an increased risk of disc disease. For example, atherosclerosis \[[@B11]\] and smoking \[[@B12]\], along with diabetes \[[@B13],[@B14]\], have been linked to IVD degeneration, although the influence of insulin-dependent diabetes on susceptibility to human disc disease remains unproven \[[@B15]\]. Few studies have examined the influence of serum-derived factors on the growth and survival of IVD cells, even though these factors generally are considered to be essential for the survival of most normal cell types. This may be because workers in the field have considered it unlikely that growth factors enter the IVD\'s inner regions via diffusion from the peripheral vasculature. Indeed, one early study \[[@B16]\] demonstrated that the diffusion rate of molecules through the cartilage endplate, the major route of glucose and oxygen to the central nucleus pulposus (NP) \[[@B9],[@B10]\], was dependent on their size, charge, and shape, with larger, more globular proteins diffusing more slowly than smaller, more linear molecules. However, there is no direct evidence that serum-derived factors are restricted from entering the IVD. Furthermore, the IVD becomes increasingly vascularised as it degenerates or is damaged, with blood vessels particularly entering disrupted regions of the tissue \[[@B17]-[@B19]\]. Herniated discs are also frequently vascularised \[[@B20],[@B21]\]. Hence, the provision of serum-derived factors to IVD cells may increase in pathological IVDs and consequently play an important role in regulating IVD cell behaviour. In this study, we have examined the relative influence of serum, glucose, and oxygen supply, singly or combined, on the growth pattern of bovine IVD cells. Since the nutrient supply affects the NP more than the anulus fibrosus (AF), the response of NP disc cells was investigated. Materials and methods ===================== Intervertebral disc cell culture -------------------------------- IVD cells were isolated from the NP of adult bovine caudal IVDs by enzymatic digestion and expanded in monolayer culture in standard Dulbecco\'s modified Eagle\'s medium (DMEM)/F-12 supplemented with 10% foetal calf serum (FCS), penicillin, and streptomycin, as described \[[@B22]\] (all reagents from Invitrogen Ltd., Paisley, UK). Freshly isolated NP cells are chondrocytic in phenotype, but during monolayer culture, they become adherent and fibroblast-like in morphology and enter the cell cycle to proliferate, a process that has been termed chondrocyte dedifferentiation \[[@B23]\]. At passage II, these fibroblastic cells were seeded into culture plates at a density of 2 × 10^3^cells per square centimetre in DMEM containing 10% FCS and left to adhere overnight. Following washes with phosphate-buffered saline, seeded cells were fed with DMEM (a) supplemented with or without 20% FCS, (b) supplemented with or without 320 mg/dL glucose, and (c) maintained in a humidified atmosphere containing atmospheric levels (\~21%) of oxygen or 1% oxygen. Various combinations of serum, glucose, and oxygen supplementation were also examined. Alternatively, passage II IVD cells were seeded into alginate beads at a low cell density of 10^6^cells per millilitre and a final alginate concentration of 1.2%. The alginate beads were then incubated in combinations of medium supplemented with or without serum and glucose and in 21% or 1% oxygen, as described above. All culture medium was at pH 6.9 to 7.1 and approximately 312 to 321 mOsm/kg water. Cultures were fed at day 4 after the initial feeding and harvested on day 7. For all analyses, control cultures were designated as those maintained at atmospheric levels of oxygen and in medium supplemented with 20% FCS and 320 mg/dL glucose. Cell proliferation and cell viability ------------------------------------- Monolayers were harvested by trypsinisation and viable cell counts were performed using trypan blue. The proliferation status of cells was also examined by immunocytochemistry for the proliferation-associated Ki-67 antigen, as described \[[@B24]\]. Cell viability in alginate was assessed by \'live/dead\' scoring, in which live cells fluoresce green and dead cells fluoresce red (Live/Dead; Invitrogen Ltd.) \[[@B25]\]. Cell senescence --------------- Senescent cells express much greater levels of beta-galactosidase activity than nonsenescent cells such that its enzymic activity can be detected at the suboptimal pH of 6.0 \[[@B26]\]. Hence, senescence-associated beta-galactosidase (SA-β-gal) activity was examined, as described \[[@B6]\], and the proportion of SA-β-gal-positive cells was quantitated by scoring a minimum of 200 cells for each condition and each sample. Collagen immunolocalisation --------------------------- The production of type I and II collagens was examined by immunolocalisation of cytospins, as described \[[@B24]\]. In brief, formalin-fixed cytospins were incubated with antibodies specific for collagen type I (clone I-8H5; ICN Biomedical Ltd., now part of MP Biomedicals, Irvine, CA, USA) or collagen type II (clone C11C1; Developmental Studies Hybridoma Bank, Iowa City, IA, USA). Immunoreactivity was revealed using a commercial kit (Vectastain ABC Elite; Vecta Labs Ltd., Peterborough, UK) combined with streptavidin-FITC (fluorescein isothiocyanate) (Vecta Labs Ltd.) and counterstained with DAPI (4\'-6-diamidino-2-phenylindole). Immunolocalisation was also performed using isotype-matched irrelevant antibodies (Dako, Ely, UK); this staining was negative. The presence of filamentous actin (F-actin) was assessed using fluorescently tagged phalloidin (FITC-phalloidin; Molecular Probes Inc., now part of Invitrogen Corporation, Carlsbad, CA, USA), as described \[[@B24]\]. Microscopy and image capture ---------------------------- Cultures were viewed using phase-contrast and fluorescence microscopy (Nikon Eclipse TS100; Nikon, Kingston-upon-Thames, UK). Digitized images were captured with a Hamamatsu (C4742-95; Hamamatsu Corporation, Bridgewater, NJ, USA) or Nikon digital camera and examined using IP Lab software (version 3.6; Nikon). Statistical analysis -------------------- The nonparametric Mann-Whitney *U*test was used to assess significant differences between control cultures and cultures deprived of nutrient factors for the following parameters: (a) the harvested cell number, (b) the proportion of Ki-67-immunopositive cells, (c) cell viability, and (d) the proportion of SA-β-gal-positive cells. All data shown are presented as mean ± standard error of the mean, where NP cell cultures were established from caudal IVD obtained from four bovine tails (that is, n = 4). Significance differences were accepted at a *P*value of less than 0.05. Results ======= Growth kinetics of monolayer cultures ------------------------------------- From a seeding population of 20 × 10^3^cells per well (in six-well plates), control monolayer cultures expanded to 166 ± 37 × 10^3^cells per well by day 7. In contrast, monolayer cultures in serum-free medium increased only to 37 ± 6 × 10^3^cells per well. Monolayers in glucose-deprived medium expanded to 403 ± 60 × 10^3^cells per well over the same time period (Figure [1a](#F1){ref-type="fig"}). These significant differences in cell harvests were reflected in the proportions of Ki-67-immunopositive cells (Figure [1b](#F1){ref-type="fig"}). The proliferative response of IVD cells to glucose deprivation was abrogated when cultures were further deprived either of serum or oxygen such that only 60 ± 9 × 10^3^cells per well or 198 ± 47 × 10^3^cells per well were harvested, respectively. Oxygen deprivation alone did not significantly alter the harvested cell number or the Ki-67 index. Day 7 cell harvests of serum-deprived cultures which were also further deprived of oxygen (31 ± 8 × 10^3^cells per well) or of oxygen and glucose (53 ± 21 × 10^3^cells per well) were not significantly different from those cultures deprived of serum alone. In all of these experiments, there was no evidence of monolayered cells becoming nonadherent and cell viability remained at 98% to 99%, except for glucose-deprived cultures in which viability by day 7 had decreased slightly, but significantly, to 94.8% ± 0.6% (Mann-Whitney *U*test; *P*= 0.0286). Hence, serum deprivation was associated with IVD cell growth arrest, whereas glucose deprivation (in the presence of serum and oxygen) resulted in IVD cell proliferation. {#F1} Morphology and senescence ------------------------- Serum-deprived IVD cells in monolayer, but not control or glucose- or oxygen-deprived cells, exhibited a stellate morphology, with many cells extending several branching cell processes (Figure [2a](#F2){ref-type="fig"}). Furthermore, the proportion of SA-β-gal-positive cells was significantly greater in serum-deprived cultures (46% ± 8%) compared with control or glucose- or oxygen-deprived cultures (\~0.5%) (Figure [2b, c](#F2){ref-type="fig"}). However, there was no clear relationship between cell morphology and cell senescence. Increased SA-β-gal positivity was not seen in cultures deprived of glucose and/or oxygen, nor was it seen to any greater extent in serum-deprived cultures that were further deprived of glucose and/or oxygen (data not shown). {#F2} Alginate cultures ----------------- IVD cells cultured in alginate were markedly less proliferative than those in monolayer. Hence, whereas 24% ± 2% of cells in control monolayers were Ki-67-immunopositive, only 3% ± 0.8% of cells were positive in control alginate cultures. None of the cells in serum-deprived alginate cultures was Ki-67-immunopositive. The proportion of Ki-67-immunopositive cells was significantly greater in glucose-deprived alginate cultures (8% ± 1.3%) compared with controls, whereas oxygen deprivation had no significant effect (Figure [3a](#F3){ref-type="fig"}). Therefore, and similar to our observations in monolayer, serum deprivation of alginate cultures induced IVD cell growth arrest whereas glucose deprivation was associated with increased proliferation. This enhanced cell proliferation was also evidenced by the formation of viable IVD cell clusters (Figure [3a](#F3){ref-type="fig"}, inset). Cytochemical positivity for SA-β-gal was significantly increased in serum-deprived alginate cultures (24% ± 3%) in comparison with all other conditions (Figure [3b](#F3){ref-type="fig"}). However, there was no evidence of serum-deprived cells in alginate becoming stellate, with all cells appearing spherical or ovoid. There was a significant decrease in cell viability in serum-deprived alginate cultures, in which only 63% ± 6% of cells remained alive at day 7 (see Additional file [1](#S1){ref-type="supplementary-material"}). Cell viability remained at approximately 95% in all other conditions (that is, in control or glucose- or oxygen-deprived cultures). {#F3} Collagen production ------------------- Collagen type I immunopositivity in IVD cells was more prevalent in control monolayer cultures compared with control alginate cultures. Conversely, collagen type II was largely absent in monolayered IVD cells but was more prevalent in cells in alginate (Figure [4a](#F4){ref-type="fig"}). Serum or oxygen deprivation appeared to have little effect on these patterns of immunopositivity (data not shown). However, glucose deprivation was associated with a drastic reduction in the immunopositivity of both collagen type I and II, whether in monolayer or alginate cultures (Figure [4b](#F4){ref-type="fig"}). Staining for F-actin was used as an indicator of protein production. Hence, F-actin positivity was seen to similar extents in control and glucose-deprived IVD cells (Figure [4c](#F4){ref-type="fig"}). ![Collagen production by intervertebral disc (IVD) cells in monolayer and alginate cultures. Immunocytochemistry of cytospins was used to examine collagen production at day 7; representative images are shown. **(a)**In monolayer control cultures (left panels), collagen type I-immunopositive IVD cells were markedly more prevalent than collagen type II-immunopositive cells, indicative of a dedifferentiated phenotype \[22\]. This pattern of immunopositivity was reversed in alginate control cultures (right panels), suggesting that the cells in alginate had redifferentiated toward a chondrocytic phenotype \[22\]. **(b)**Immunopositivity for both types of collagen was markedly decreased in glucose-deprived cultures, both in monolayer (left panels) and alginate (right panels) culture. **(c)**Phalloidin-fluorescein isothiocyanate staining of IVD cells following monolayer (left panel) and alginate (right panel) culture in conditions of glucose-deprivation at day 7, demonstrating that the presence of filamentous actin (F-actin) was not markedly different in either condition (original magnification ×200).](ar2405-4){#F4} The predominant phenotypic changes seen in IVD cells when cultured in the presence or absence of serum or glucose, both in monolayer and alginate culture systems, are summarised in Table [1](#T1){ref-type="table"}. The minimal influence of oxygen levels on IVD cell growth and behaviour is omitted from this summary. ###### The major phenotypes of bovine intervertebral disc cells in response to altered culture conditions Serum Glucose -------------------- ---------------------- -------------------- -------------------- ---------------------- Cell proliferation ↑↑↑ ↓↓ ↔ ↑↑↑↑ Cell senescence ↔ ↑↑↑ ↔ ↔ Cell death ↔ ↔ ↔ ↔ Collagen synthesis ↑↑↑ Type I ↓ Type II ↔ Type I ↔ Type II ↔ Type I ↔ Type II ↓↓↓ Type I ↓ Type II Serum Glucose Alginate With Without With Without Cell proliferation ↓↓ ↓↓↓ ↔ ↑ Cell senescence ↔ ↑↑ ↔ ↔ Cell death ↔ ↓↓ ↔ ↔ Collagen synthesis ↓ Type I ↑↑ Type II ↔ Type I ↔ Type II ↔ Type I ↔ Type II ↔ Type I ↓↓ Type II ↑, increased; ↓, decreased; ↔, no change (from control conditions). Discussion ========== Pathological conditions of the human IVD are associated with changes in IVD cell behaviour, including increased cell death \[[@B2],[@B3]\], cell division \[[@B4]\], the appearance of stellate cells \[[@B5]\], and cell senescence \[[@B6]-[@B8]\]. What leads to these phenotypes is currently unclear; however, it is generally thought that nutritional deprivation may limit the capacity of IVD cells to function. Here, we have demonstrated that depriving IVD cells of serum, glucose, or oxygen has a variable influence on their growth and survival. Serum supplementation was required for continued IVD cell proliferation in monolayer cultures but was insufficient to drive proliferation to the same extent (as delineated by the Ki-67 index) in alginate cultures. Conversely, serum deprivation of monolayer cultures had no significant effect on cell viability but was associated with increased cell death in alginate cultures. In both culture systems, serum deprivation led to IVD cell growth arrest. These findings fit the general observation of various cell types that serum withdrawal induces cell growth arrest and/or apoptosis \[[@B27],[@B28]\] whereas cell adhesion promotes cell survival \[[@B29]\]. We also found that serum deprivation was associated with increased SA-β-gal positivity in both culture systems and with the appearance of stellate cells in monolayer cultures. Both of these phenotypes have been linked with cell senescence or cellular adaptations to extracellular stresses \[[@B26],[@B30]\]. As SA-β-gal staining was seen in cell populations that had undergone growth arrest, it can be concluded that this may be indicative of premature cell senescence. It remains to be determined how these findings relate to pathological human discs, where SA-β-gal-positive cells were seen most frequently in cell clusters \[[@B6]\] and have been associated with an increased catabolic activity \[[@B8]\]. In addition, Gruber and colleagues \[[@B31]\] have demonstrated increased levels of apoptosis in cells isolated from pathological human AF tissue in response to serum deprivation over a 10-day culture period of monolayer culture (from 0.1% to approximately 1%); this induction of cell death was inhibited by insulin-like growth factor-1 or platelet-derived growth factor. Zhao and colleagues \[[@B32]\] recently found that serum deprivation also induces apoptosis in rat AF cells, an effect that was exacerbated by interleukin-1-β. Hence, it is possible that the initial growth arrest and senescence that we have documented in response to serum deprivation may be followed by apoptotic cell death. Alternatively, NP cells and AF cells may require serum-derived survival factors to different extents. These areas warrant further investigation, as do the actions of individual growth and survival factors that may enter the disc from the vascular supply or that may be synthesised by IVD cells themselves \[[@B33]\]. The availability of oxygen and glucose to IVD cells is dependent on their transport from blood vessels and therefore on the degree of disc vascularisation (reviewed in \[[@B9],[@B10]\]). As this varies according to age and pathology \[[@B17]-[@B21]\], the influence of oxygen and glucose on IVD cell growth may also be expected to vary. However, depriving IVD cells of oxygen alone had no marked effect on their growth or survival *in vitro*, either in monolayer or alginate culture. Risbud and colleagues \[[@B34]\] have shown that rat, sheep, and human IVD cells express hypoxia inducible factor-1-alpha (HIF-1-α), a transcription factor that is responsive to oxygen availability, even in normoxia. HIF-1-α expression was also only minimally induced and activated following hypoxia in culture. Furthermore, the same authors demonstrated that NP cells in rat discs, which are notochordally derived, do not appear to rely upon oxygen for their metabolism *in vivo*\[[@B35]\]. Although the central regions of the human IVD are more hypoxic than its periphery \[[@B36]\] (which may not be the case in smaller rat discs), one interpretation of the studies of Risbud and colleagues, along with our present findings, is that NP cells are adapted not to respond to hypoxia as readily as other cell types. Depriving IVD cells of glucose resulted in increased cell proliferation, so long as serum (and to a lesser extent oxygen) was present, but a clear decrease in collagen production. The reasons for these responses are currently unclear; however, a similar relationship between glucose levels and proliferation was reported recently in Chinese hamster ovary (CHO) cells, which expanded in a low-glucose environment (6 mM) at a rate 1.7 times that seen in high-glucose conditions (33 mM) \[[@B37]\]. Similarly, exposure to higher levels of glucose (greater than 15 mM) was shown to inhibit cell proliferation in fibroblasts \[[@B38]\]. In our experiments, the proliferative response to glucose deprivation was dependent on the presence of serum, which in itself will contain glucose. As normal blood glucose levels vary between 4 and 8 mM (equating to 70 to 150 mg/dL), it may be assumed that the glucose levels in culture medium supplemented with 20% FCS will contain approximately 14 to 30 mg/dL. Hence, the precise relationships between levels of glucose and of other serum-derived factors in regulating IVD cell proliferation and collagen synthesis remain to be fully elucidated. It is noteworthy, however, that the increased cell proliferation seen in glucose-deprived alginate cultures resulted in the formation of NP cell clusters, which also form through increased cell proliferation in the NP *in vivo*and have been suggested to form part of an ineffective response to tissue damage \[[@B4]\]. Conclusion ========== This study demonstrates that factors present in serum interact with other nutrients, notably glucose, to play a major role in regulating the behaviour of IVD cells. Furthermore, by manipulating the nutrient and serum supply to IVD cells *in vitro*, we have observed a number of cellular phenotypes akin to those seen in pathological human IVDs \[[@B2]-[@B7]\]. Hence, this study supports the conclusion that many of the cellular changes associated with IVD degeneration are likely to arise through the cells\' response to altered vascularisation and nutrient supply. Abbreviations ============= AF = anulus fibrosus; DMEM = Dulbecco\'s modified Eagle\'s medium; F-actin = filamentous actin; FCS = foetal calf serum; FITC = fluorescein isothiocyanate; HIF-1-α = hypoxia inducible factor-1-alpha; IVD = intervertebral disc; NP = nucleus pulposus; SA-β-gal = senescence-associated beta-galactosidase. Competing interests =================== The authors declare that they have no competing interests. Authors\' contributions ======================= WEBJ helped to perform the experimental work, helped to conceive of the study, participated in its design and coordination, helped to perform the statistical analysis, and helped to draft the manuscript. SS helped to perform the experimental work and the statistical analysis and helped to draft the manuscript. SR helped to conceive of the study, participated in its design and coordination, and helped to draft the manuscript. All authors read and approved the final manuscript. Supplementary Material ====================== ###### Additional file 1 \'Live/dead\' staining of intervertebral disc (IVD) cells in alginate cultures. A representative image is shown of IVD cells in alginate cultured in serum-deprived conditions for 7 days; the pycnotic nucleus of a nonviable cell appears bright red, with viable cells appearing green (original magnification ×400). ###### Click here for file Acknowledgements ================ This work was funded by the Biotechnology and Biological Sciences Research Council, UK.

HAM/TSP is an inflammatory manifestation of central nervous system caused by HTLV-1 and the mechanism of HAM/TSP development is no well elucidated. Currently, a promising approach on the physiopathogenesis of viral infections has been the evaluation of microRNAs (miRNAs) role. There are few data involving CD4+ T cells miRNA expression in HTLV-1 infection as well as in HAM/TSP establishment. To identify miRNAs differentially expressed in CD4+ T cells among non-infected individuals (CT), asymptomatic (HAC) and HAM/TSP patients we applied quantitative real time PCR. The analysis of miRNA expression profile in these cells showed 56 and 10 miRNAs upregulated 1.5 times in HAM/TSP and HAC groups, respectively. miR-125b-1-1 was upregulated in HAC group and miR-146a in HAM/TSP. Target prediction by in silico analysis showed that IFNG was a potentially miR-125b-1-1 target and IRAK1 and TRAF6 were miR-146a targets. IFNG expression was 1.3 higher in HAC than CT group and 1.8 higher in HAM/TSP than CT group. It was observed that TRAF6 expression was 15.7 and 1.5 times higher in HAM/TSP and HAC groups, respectively. There was no difference of IRAK1 expression among the three groups. Overexpression assays of miR-125b-1-1 altered IFNG expression and overexpression of miR-146a altered IRAK1 gene and protein expression. The results revealed that miRNAs could modulate genes and proteins during HTLV-1 infection. miR-125b and IFNG gene correlation suggests that miR-125b seems to contribute to HAM/TSP development. Besides, interaction between miR-146a and IRAK1/TRAF6 suggests that miR-146a seems to contribute to HTLV-1 establishment in CD4+ T cells.

All relevant measurements and numerical code (Matlab and COMSOL) are available in Supporting Information [S2 File](#pone.0202416.s002){ref-type="supplementary-material"}. Introduction {#sec001} ============ Owing to recent analytical method developments \[[@pone.0202416.ref001]--[@pone.0202416.ref005]\] and to numerous studies demonstrating the potential of applications of compound specific isotopic analysis (CSIA) in estimating the origin and fate of chlorinated solvents in the subsurface \[[@pone.0202416.ref006]--[@pone.0202416.ref008]\], environmental samples are increasingly being submitted for isotopic analysis. Previous studies showed that such analysis can facilitate the determination of the extent of biodegradation in groundwater based on isotopic enrichment factors determined via the Rayleigh equation. This is possible due to the fact that molecules containing different proportions of isotopes of one element (i.e., isotopologues) follow different reaction kinetics. This is caused by the different energy required to break the chemical bonds binding elements with heavy or light isotopes which eventually induces the observed variation in isotopic composition during biodegradation. An accurate simulation of the isotopic ratio evolution during biodegradation may allow a better integration of isotopic data when estimating the fate of a plume undergoing natural attenuation. Several studies have been carried out to address this challenge, some of which additionally successfully applied models to evaluate the fate of chloroethenes plumes \[[@pone.0202416.ref009]--[@pone.0202416.ref013]\]. However, for simplification purposes, most models considered one element only \[[@pone.0202416.ref009], [@pone.0202416.ref011], [@pone.0202416.ref014], [@pone.0202416.ref015]\], and some models simulated heavy and light isotopes of one element of each compound as separate species \[[@pone.0202416.ref014], [@pone.0202416.ref015]\] while others considered isotopologues with regards to one \[[@pone.0202416.ref011], [@pone.0202416.ref016]\] or two elements \[[@pone.0202416.ref013]\]. Such a variety of simplification types reflects the need for a trade-off between including all isotopologues or isotopocules (i.e., isotopomers of all isotopologues of one compound; isotopomers being isomers of isotopologues) and avoiding an excessive complexity leading to computationally demanding models. More specifically, Jin et al. \[[@pone.0202416.ref013]\] demonstrated that considering simultaneously combined C-Cl isotopologues could change simulated isotopic behaviours at late reaction times and for large differences between the C and Cl enrichment factors compared to the simplified method developed by Hunkeler et al. \[[@pone.0202416.ref011]\] where Cl isotopologues/isotopes only were considered. However, secondary isotope effects, which affect elements located in non-reacting bonds, were neglected in this work whichalso ignored PCE. Recent studies have shown that secondary Cl isotopic effects are measurable during chloroethene reductive dechlorination and should therefore not be discounted \[[@pone.0202416.ref016]--[@pone.0202416.ref018]\]. However, few models so far have considered secondary isotopic effects since they were formerly generally assumed to be negligible \[[@pone.0202416.ref013], [@pone.0202416.ref019], [@pone.0202416.ref020]\]. To address this need, Höhener \[[@pone.0202416.ref018]\] and van Breukelen et al. \[[@pone.0202416.ref016]\] recently incorporarted secondary isotopic effects in isotope fractionation models though both studies considered isotopes of all elements independantly. Van Breukelen et al. \[[@pone.0202416.ref016]\] applied a correction term to the reaction rate of each isotope/isotopologue so that the independent isotope networks correspond to the overall reaction progress. On the other hand, Höhener et al. applied a "Cretnik correction" which involved the introduction of an offset. Both successfully tested their models against TCE reductive dechlorination data \[[@pone.0202416.ref017], [@pone.0202416.ref021]\]. To the best of our knowledge, no model without simplification that considers whole molecules as they are in reality (i.e. not considering isotopes of elements separately) has thus far been developed. Such a model might (i) enable determination of the limits of simplified models by comparing its outcome with that of a simplified model and (ii) allow for the development of models for other compounds or even for the incorporation of clumped isotope effects. Additionally, former work has focused on the reductive dechlorination of chloroethenes starting with TCE. Concurrently considering (i) Monod kinetics, (ii) secondary isotopic effects, and (iii) several elements simultaneously and the corresponding isotopocules to which they belong is a challenging and important task in the simulation of the evolution of chlorinated ethenes isotopic composition during reductive dechlorination. Addressing this knowledge gap may help with the integration of isotopic data from field sample measurements in reactive transport models for improved plume fate prediction. To this end, the present contribution proposes a generic isotope modelling framework which is more comprehensive than the current state of the art. With a focus on reductive dechlorination of chloroethenes, this study is focused on: (i) developing and implementing a novel comprehensive and generic general model (GM) which allows simulation of the simultaneous evolution of isotopic composition in chloroethenes during sequential reductive dechlorination considering Monod kinetics and secondary isotopic effects, (ii) setting up a computationally less demanding simplified model (SM) matching the general model which meets the same requirements with regards to Monod kinetics and secondary isotopic effects, (iii) verifying whether this model can accurately reproduce C and Cl isotope data obtained during reductive dechlorination of chloroethenes by two different bacterial consortia, and (iv) identify where the two models diverge in predicting isotopic enrichment/depletion. Finally, this work also seeks to provide the community with a relevant software repository to encourage the integration of primary and seconday isotope effects into future modeling scenarios. Details of the mathematical framework of the models are first presented. This is followed by a description of the implementation, evaluation, and comparison of the GM and SM; an assessment of the feasibility of fitting these models to experimental data; and finally, a discussion of the significance for the assessment of sites contaminated by chloroethenes. Mathematical model development {#sec002} ============================== The general (GM) and simplified (SM) models developed here to simulate isotope trends take into account primary and secondary isotope effects as well as Monod kinetics. Isotopocules designate all isotopomers of all isotopologues, i.e. the entire set of molecules of the same compound differing both in the number and position of light and heavy isotopes. General model (GM): Simultaneous consideration of C and Cl isotopes {#sec003} ------------------------------------------------------------------- ### General expression {#sec004} Traditionally, fractionation between isotopes rather than between isotopocules has been considered \[[@pone.0202416.ref022]\]. Such fractionation is described by the kinetic isotope fractionation factor α~k~ relating the isotope ratio of the instantaneous product to the isotope ratio of the substrate. The difference in rate transfer from reactant to product pool associated with heavy and light isotopes, ^*H*^*k*~bulk~ and ^*L*^*k*~bulk~, respectively, can also be described by α~k~ which equals the ratio ^*H*^*k*~bulk~/^*L*^*k*~bulk~ \[[@pone.0202416.ref022]\]. In order to incorporate isotope fractionation during sequential dechlorination in a reactive transport model, Van Breukelen et al. \[[@pone.0202416.ref014]\] previously established the following expression relating the ^13^C isotope reaction rate ^13^*r*~*S*~ of a consumed substrate (S) to overall reaction rate *r*~*S*~ of S: $${}^{13}r_{S} = r_{S}\frac{\lbrack^{13}S\rbrack}{\lbrack S\rbrack}\alpha_{k}$$ This equation is valid under the assumption that the degradation rate of the predominant ^12^C isotope corresponds to the overall degradation rate of S corrected for the proportion of ^12^C to total carbon \[[@pone.0202416.ref014]\]. The terms \[^13^*S*\] and \[*S*\] correspond to the ^13^C concentration in S and the total S concentration, respectively. A similar approach is applied in the present work to simulate the evolution of a compound's isotopic composition considering simultaneously all elements affected by isotopic fractionation and including secondary isotope effects. In this case, instead of two isotopes, each isotopocule is considered as an individual entity undergoing reaction at a specific rate related to its isotopic composition and therefore to the "isotopocule fractionation" it undergoes. We thus expand the expression suggested by Van Breukelen et al. \[[@pone.0202416.ref014]\] for isotopocules of both produced and degraded compounds. More particularly, we ensurethat a given isotopocule of a produced compound is yielded by specific isotopocules of its precursor compound and thus appears in the expanded expression. The reaction rate of each isotopocule *i* of intermediate (i.e. both produced and degraded) compound γ resulting from the degradation of isotopocules *h* of compound γ-1 and being further degraded to isotopocules *j* of compound γ+1 can then be described by the following general equation (note that the right-most term describes the rate of degradation to γ+1 and the preceeding term describes the rate of production from γ-1): $$\frac{\partial C_{i}^{\gamma}}{\partial t} = r_{i}^{\gamma} = \left( {\sum_{h = 1}^{n_{\text{Iso}~}^{\gamma - 1}}\mathbf{\kappa}_{h,i}^{\gamma - 1\rightarrow\gamma} \cdot v^{\gamma - 1}\frac{C_{h}^{\gamma - 1}}{C_{\text{tot}}^{\gamma - 1}}} \right) - \left( {\sum_{j = 1}^{n_{\text{Iso}~}^{\gamma + 1}}\mathbf{\kappa}_{i,j}^{\gamma\rightarrow\gamma + 1}} \right) \cdot v^{\gamma}\frac{C_{i}^{\gamma}}{C_{\text{tot}}^{\gamma}}$$ $$\forall i \in \left\lbrack 1,\ldots,n_{Iso}^{\gamma} \right\rbrack$$ where *C*~*i*~ and *C*~*tot*~ are, respectively, the isotopocule *i* and total compound concentrations (i.e. $C_{tot}^{\gamma} = {\sum\limits_{i = 1}^{n_{Iso}^{\gamma}}C_{i}^{\gamma}}$ and $C_{tot}^{\gamma - 1} = {\sum\limits_{h = 1}^{n_{Iso}^{\gamma - 1}}C_{h}^{\gamma - 1}}$ where $n_{Iso}^{\gamma}$ and $n_{Iso}^{\gamma - 1}$ are the number of isotopocules of compound γ and γ-1), *r* is the isotopocule reaction rate, *v* is the total compound reaction rate. The matrix ***κ*** describes the difference in isotopocule reaction rates due to the presence of light or heavy atoms in reacting (primary isotope effect) and/or non-reacting positions (secondary isotope effect). It is hence analogous to *α*~*k*~, except that it applies to isotopocules instead of isotopes. It theoretically allows any transition from one isotopocule of a degraded compound (i.e., breaking any reactive bond) to any isotopocule of the compound it produces and, as such, is fully general. For chloroethenes in particular, there are only a few possible transitions from isotopocules of the degraded compound to isotopocules of the produced compound (e.g., maximum 4 transitions for 1 PCE isotopocule degraded to TCE), thus the matrix is sparse, i.e., mostly filled with zero elements which exclude non-existing transitions. The non-zero elements of the matrix represent the isotopocule fractionation factors. Here $\mathbf{\kappa}_{h,i}^{\gamma - 1\rightarrow\gamma}$ thus represents an element of the isotopocule fractionation matrix $\mathbf{\kappa}_{}^{\gamma - 1\rightarrow\gamma}\left( \in \mathbb{R}^{n_{Iso}^{\gamma - 1} \times n_{Iso}^{\gamma}} \right.$) containing kinetic isotopocule fractionation factors associated with isotopocules h of compound γ-1 leading to isotopocule i of compound γ. Analogously, the matrix $\mathbf{\kappa}_{\mspace{180mu}}^{\gamma\rightarrow\gamma + 1}\left( \in \mathbb{R}^{n_{Iso}^{\gamma} \times n_{Iso}^{\gamma + 1}} \right.$) describes the transitions from compound γ leading to compound γ+1. ***κ*** is more specifically defined as: $$$$ where γ and γ-1 designate the produced and degraded compounds, respectively, and h and i the isotopocules of γ and γ-1, respectively. *n*~*RB*~ is the number of reactive bonds, AKIE~*p*~ the position-specific apparent kinetic isotopic effect associated with the atom at relative position p when removing the atom at absolute position X. The weight coefficient matrix ***W***^*γ*-1^ ($\in \mathbb{R}^{n^{\gamma - 1} \times n_{Atoms}^{\gamma - 1}}$) accounts for all possible combinations of heavy (weight = 1) and light (weight = 0) isotopes in the molecule and thus describes all possible isotopocules. For each of the possible bond breakage positions (removal of atom X) a transformation matrix ^***X***^***T***^*γ*-1^ ($\in \mathbb{R}^{n_{Atoms}^{\gamma} \times n_{Atoms}^{\gamma}}$) exists, which transforms W from an absolute reference system (topologically fixed) to a relative reference system (relative to the bond breakage position) as illustrated for PCE in [Fig 1](#pone.0202416.g001){ref-type="fig"}. This transformation step is necessary as the vector AKIE is arranged in a relative reference system. To clarify these concepts, two examples of the ***κ*** matrix associated with PCE reductive dechlorination to *c*DCE (***κ***^PCE→TCE^ and ***κ***^TCE→*c*DCE^) are illustrated in [Fig 2](#pone.0202416.g002){ref-type="fig"}. {#pone.0202416.g001} {#pone.0202416.g002} The product in each non-zero element of ***κ*** takes into account the position-specific apparent kinetic isotopic effects (AKIEs) associated with each atom of the isotopocule of the degraded compound. The latter accounts for primary and secondary isotopic effects in isotopocule fractionation factors and is defined as follows: $$\text{AKIE}^{\gamma} = \left\lbrack {\frac{1}{1 + \frac{\varepsilon_{1}}{1000}},\mspace{180mu}\ldots,\frac{1}{1 + \frac{\varepsilon_{n_{\text{Atoms}}^{c}}}{1000}}} \right\rbrack \in {\mathbb{R}}^{n_{\text{Atoms}}^{\gamma}}$$ where *ε*~*p*~ (with $p \in \left\lbrack 1,\ldots,n_{Atoms}^{\gamma} \right\rbrack$) corresponds to the degradation-related position-specific isotopic effect (either primary or secondary) of the atom at relative position p. Additionally, clumped isotope effects, where the isotope effect associated with the presence of a heavy atom at a certain position depends on the presence of heavy atoms at other positions, could be included in order to have a more comprehensive model. They were however not included when determining ***κ*** due to the lack of experimental evidence for the occurrence of such effects in the case of chloroethenes reductive dechlorination and the challenges of their potential measurement. Yet, as each isotopocule is considered separately, the approach could easily be adapted to include clumped isotope effects should future advances enable their measurement. The only modification necessary would consist of adapting the non-zero elements of ***κ***. ### Application of GM to PCE reductive dechlorination {#sec005} First, the general concept of GM applied to the reductive dechlorination of PCE to *c*DCE is presented followed by a more detailed explanation of how the matrix ***κ*** is generated for the PCE to TCE transformation step. [Fig 2B](#pone.0202416.g002){ref-type="fig"} illustrates the sequential reductive dechlorination of PCE isotopocules to TCE and consecutively to *c*DCE isotopocules. [Fig 2A](#pone.0202416.g002){ref-type="fig"} illustrates the corresponding isotopocule fractionation matrices $\mathbf{\kappa}_{\mspace{180mu}}^{PCE\rightarrow TCE} \in \mathbb{R}^{64 \times 32}$ and $\mathbf{\kappa}_{\mspace{180mu}}^{TCE\rightarrow cDCE} \in \mathbb{R}^{32 \times 16}$. The number of lines in each matrix corresponds to the number of isotopocules of the degraded compound. Similarly, the number of columns in each matrix corresponds to the number of isotopocules of produced compound. In each line corresponding to the degradation of one isotopocule, each non-zero element corresponds to the isotopocule fractionation factor associated with each breakable bond position. In the case of PCE to *c*DCE reductive dechlorination, a total number of $n_{Iso}^{PCE} = 2^{6} = 64$ (6 atoms, each having two possible states, i.e. either a heavy or a light isotope) isotopocules of PCE each yield 4 possible TCE isotopocules (= 4 breakable bonds) among a total number of $n_{Iso}^{TCE} = 2^{5} = 32$ isotopocules of TCE. In ***κ***^PCE→TCE^, each line will thus contain 4 non-zero elements corresponding to the different possible bond breakage positions. Each TCE isotopocules will in turn yield 1 *c*DCE isotopocule (= 1 breakable bond) among a total number of $n_{Iso}^{cDCE} = 2^{4} = 16$ isotopocules of *c*DCE. Similarly, in ***κ***^TCE→*c*DCE^, each line will thus contain 1 non-zero element corresponding to the different possible bond breakage positions. It should be noted that different isotopocules of degraded compound may yield the same isotopocule of produced compound. Furthermore, for symmetric molecules such as PCE and *c*DCE, the total number of isotopocules may be reduced due to symmetries. PCE reductive dechlorination to TCE (ignoring subsequent degradation to *c*DCE), denoted as PT, is considered here to illustrate the application of [Eq (2)](#pone.0202416.e002){ref-type="disp-formula"}. As PCE is being degraded, the concentration of its isotopocules h will thus follow: $$\frac{\partial C_{h}^{\text{PCE}}}{\partial t} = - \left( {\sum_{i = 1}^{n_{\text{Iso}}^{\text{TCE}}}\mathbf{\kappa}_{h,i}^{\text{PCE}\rightarrow\text{TCE}}} \right) \cdot v^{\text{PCE}}\frac{C_{h}^{\text{PCE}}}{C_{\text{tot}}^{\text{PCE}}},\mspace{180mu}\forall h \in \left\lbrack {1,\ldots,n_{\text{Iso}}^{\text{PCE}}} \right\rbrack$$ Conversely, the concentration of isotopocules i of produced TCE will follow: $$\frac{\partial C_{i}^{\text{TCE}}}{\partial t} = \sum_{h = 1}^{n_{\text{Iso}}^{\text{PCE}}}\left( {\mathbf{\kappa}_{h,i}^{\text{PCE}\rightarrow\text{TCE}} \cdot v^{\text{PCE}}\frac{C_{h}^{\text{PCE}}}{C_{\text{tot}}^{\text{PCE}}}} \right),\mspace{180mu}\forall i \in \left\lbrack {1,\ldots,n_{\text{Iso}}^{\text{TCE}}} \right\rbrack$$ These equations are based on the general [Eq (2)](#pone.0202416.e002){ref-type="disp-formula"}. Equations (S1), (S2) and (S3) describing PCE, TCE and *c*DCE isotopocules during sequential reductive dechlorination are included in [S1 File](#pone.0202416.s001){ref-type="supplementary-material"} (Supporting Information). In order to generate ***κ***^PCE→TCE^, the weight coefficient matrix ***W***^PCE^ ($\in \mathbb{R}^{64 \times 6}$) is created to describe the position of heavy (weight = 1) and light (weight = 0) isotopes in all degraded PCE isotopocules by means of an absolute reference system where the position of all atoms in the molecule are "topologically" fixed ([Fig 1](#pone.0202416.g001){ref-type="fig"}, molecule 1). Each of the 64 lines in ***W***^PCE^ represents an isotopocule whereas each of the 6 columns corresponds to one atom at absolute position X being light or heavy. This matrix is then transformed into the aforementioned relative reference system by using one of the four transformation matrices ^***C***^***T***^PCE^, ^***D***^***T***^PCE^, ^***E***^***T***^PCE^, or ^***F***^***T***^PCE^ enabling the correct treatment of a broken bond at absolute position X = C, D, E, or F, respectively ([Fig 1](#pone.0202416.g001){ref-type="fig"}, molecules 3 to 6). The AKIEs themselves are given in the relative system ([Fig 1](#pone.0202416.g001){ref-type="fig"}, molecule 2) and correspond to the kinetic isotopic effects associated with each atom at relative position p ∈ \[1,...,6\]. Typically, AKIE = \[AKIE~Cα~, AKIE~Cβ~, AKIE~C1α1~, AKIE~C1α2~, AKIE~C1β1~, AKIE~C1β2~\], where C~α~ undergoes a primary C isotopic effect and C~β~ a secondary C isotopic effect; Cl~α1~ undergoes a primary Cl isotopic effect while Cl~α2~, Cl~β1~ and Cl~β2~ undergo secondary Cl isotopic effects. Thanks to the transformed weight coefficient matrix ***W***^PCE^·^***X***^***T***^PCE^, these AKIEs will be accounted for (weight = 1) or not (weight = 0) in ***κ***^PCE→TCE^ depending on whether a heavy or a light isotope is present in the absolute position X as described earlier in [Eq (3)](#pone.0202416.e011){ref-type="disp-formula"}. Separate consideration of C and Cl isotopes (SM) {#sec006} ------------------------------------------------ In view of incorporating the change of isotopic ratios during degradation in a reactive transport model, for example, the general model (GM) was simplified to a less computationally expensive simplified model (SM). This simplification was achieved by simulating fewer species simultaneously. This model as applied to chloroethenes is explained here and is illustrated in Fig A in [S1 File](#pone.0202416.s001){ref-type="supplementary-material"}. First, C and Cl atoms are considered separately. Second, as the bond cleavage can take place at any position involving a Cl atom (non-regioselective reaction) for symmetric chloroethenes (e.g. PCE), the positions of heavy and light Cl and C isotopes in the molecule are no longer relevant and we thus consider isotopologues relative to C and Cl instead of isotopocules. Since no differentiation is made between the secondary positions in this model, a single secondary isotopic effect ($\text{AKIE}_{\text{ClSec}}^{\gamma}$) is considered which reflects all position-specific secondary isotopic effects associated with all Cl atoms located in remaining positions (e.g. for PCE: $\text{AKIE}_{\text{Cl}\alpha 2}^{\text{PCE}} = \text{AKIE}_{\text{Cl}\beta 1}^{\text{PCE}} = \text{AKIE}_{\text{Cl}\beta 2}^{\text{PCE}} = \text{AKIE}_{\text{ClSec}}^{\text{PCE}}$). This assumption was made according to the explanations of Cretnik et al. \[[@pone.0202416.ref017]\] who showed that the overall secondary isotopic effect corresponds to the average between all secondary isotopic effects. Finally, a distinction in treatment is made between symmetric and asymmetric molecules. Based on these considerations, the corresponding fractionation factor associated with Cl thus reflects (i) both the isotopic effects induced when cleaving a bond involving a heavy Cl isotope (primary isotopic effect) and the isotopic effect induced by the presence of heavy Cl isotopes in the remaining positions which do not react (secondary isotopic effect) (first condition of ***κ***^*γ*→*γ*+1^ described in [Eq (7)](#pone.0202416.e030){ref-type="disp-formula"}) or (ii) the isotopic effects induced only by the presence of heavy Cl isotopes in the positions which do not react when cleaving a bond involving a light Cl isotope (second condition of ***κ***^*γ*→*γ*+1^ described in [Eq (7)](#pone.0202416.e030){ref-type="disp-formula"}). Isotopologue fractionation factors relative to Cl are hence calculated as follows: $$$$ Contrary to Cl, C atoms are not removed during the reaction, isotopologue fractionation factors associated with C were hence determined as follows: $$\mathbf{\kappa}_{i,j}^{\gamma\rightarrow\gamma + 1} = \left\{ \begin{array}{rc} {\frac{1}{\text{AKIE}_{\text{CPrim}}^{\gamma}} \cdot \left( \frac{1}{\text{AKIE}_{\text{CSec}}^{\gamma}} \right)^{n_{13\text{C}_{i}}^{\gamma} - 1} \cdot \left( \frac{n_{13\text{C}_{i}}^{\gamma}}{n_{\text{C}}^{\gamma}} \right)} & \\ {+ \left( \frac{1}{\text{AKIE}_{\text{CSec}}^{\gamma}} \right)^{n_{13\text{C}_{i}}^{\gamma}} \cdot \left( {1 - \frac{n_{13\text{C}_{i}}^{\gamma}}{n_{\text{C}}^{\gamma}}} \right),} & \begin{array}{r} {\text{if}\mspace{180mu}\text{isotopologue}\mspace{180mu}\text{i}} \\ {\text{leads}\mspace{180mu}\text{to}\mspace{180mu}\text{isotopologue}\mspace{180mu}\text{j}} \\ \end{array} \\ {0,} & \text{else} \\ \end{array} \right.$$ where i and j correspond to isotopologues of compound γ and γ+1. The specific matrices ***κ*** containing isotopologue fractionation factors associated with sequential reductive dechlorination of PCE to *c*DCE are given in Fig B in [S1 File](#pone.0202416.s001){ref-type="supplementary-material"}. These specific matrices ***κ*** differ slightly from their definitions in Eqs ([7](#pone.0202416.e030){ref-type="disp-formula"}) and ([8](#pone.0202416.e031){ref-type="disp-formula"}) as they are further modified to meet the requirements for asymmetric molecules as explained in the following. Contrary to symmetric molecules, the bond cleavage is regioselective for asymmetric chloroethenes (e.g. TCE). For isotopologues containing both heavy and light Cl isotopes, the positions thereof should therefore be taken into account so that the fact that the bond breakage takes place only where the isotope located in the only reactive position may be considered. "Isotopocules" relative to Cl were hence considered in the case of asymmetric chloroethenes instead of isotopologues. The fractionation factors were determined similarly as for symmetric molecules with the exception that instead of considering any Cl position, the presence of heavy or light Cl isotope in the only possible cleaved position (α1) was taken into account separately from heavy or light Cl isotopes located in non-reacting positions (Figs A and B in [S1 File](#pone.0202416.s001){ref-type="supplementary-material"}). In the case of C which constitutes the backbone during sequential dechlorination, the probability that a heavy or a light atom is involved in the bond breakage is considered equal when both a light and a heavy isotope are present in the molecule for symmetric molecules (e.g. PCE). Conversely for asymmetric molecules (e.g. TCE), since only one C is involved in the bond-breakage in one isotopocule, the occurrence of primary and secondary isotopic effect is not distributed between the two positions as illustrated for the case of TCE ($\mathbf{\kappa}^{{TCE}_{C}\rightarrow{cDCE}_{C}}$) (Fig B in [S1 File](#pone.0202416.s001){ref-type="supplementary-material"}). The case of Monod kinetics {#sec007} -------------------------- The bacterial growth on sequential or simultaneous substrates following Monod kinetics can be described as by Kompala et al. \[[@pone.0202416.ref023]\]: $$\frac{\partial X}{\partial t} = \sum_{\gamma = 1}^{n_{\text{Comps}} - 1}\left( {\mu_{\text{MAX}}^{\gamma}\frac{X \cdot C_{\text{tot}}^{\gamma}}{K_{m}^{\gamma} + C_{\text{tot}}^{\gamma}}} \right) - \mu_{\text{DEC}} \cdot X$$ where *n*~Comps~ is the total number of compounds used for growth, X \[g of protein·L^-1^\] is the biomass concentration growing on all compounds, *C*~tot~ \[μmol·L^-1^\] is the concentration, μ~MAX~ \[μmol·g of protein^-1^·s^-1^\] is the maximum growth rate, *K*~*m*~ \[μmol·L^-1^\] is the half saturation constant, and μ~DEC~ \[s^-1^\] is the biomass decay rate constant associated with growth on chloroethenes. As the isotopic shifts triggered by compound consumption underlie the primary interests of this study, the biomass growth phase is our focus. The biomass decay rate constant (μ~DEC~) associated with growth is thus assumed to be zero here. If Monod kinetics are assumed, the following expression for isotopocules (respectively isotopologues for SM) rates may be expressed based on [Eq (2)](#pone.0202416.e002){ref-type="disp-formula"}, considering growth on sequential or simultaneous substrates where substrates correspond to isotopocules: $$\frac{\partial C_{i}^{\gamma}}{\partial t} = \left( {\sum_{h = 1}^{n_{\text{Iso}}^{\gamma - 1}}\mathbf{\kappa}_{h,i}^{\gamma - 1\rightarrow\gamma} \cdot \frac{\mu_{\text{MAX}}^{\gamma - 1} \cdot X \cdot C_{h}^{\gamma - 1}}{Y^{\gamma - 1} \cdot \left( {K_{m}^{\gamma - 1} + C_{\text{tot}}^{\gamma - 1}} \right)}} \right) - \left( {\sum_{j = 1}^{n_{\text{Iso}}^{\gamma + 1}}\mathbf{\kappa}_{i,j}^{\gamma\rightarrow\gamma + 1}} \right) \cdot \frac{\mu_{\text{MAX}}^{\gamma} \cdot X \cdot C_{i}^{\gamma}}{Y^{\gamma} \cdot \left( {K_{m}^{\gamma} + C_{\text{tot}}^{\gamma}} \right)}$$ $$\forall i \in \left\lbrack 1,\ldots,n_{Iso}^{\gamma} \right\rbrack$$ where Y \[g of protein·μmol of released chloride^-1^\] represents the biomass yield. The first term corresponds to the production of compound γ from γ-1 and the second term to its degradation to γ+1. Equations developed for PCE reductive dechlorination to *c*DCE are given in [S1 File](#pone.0202416.s001){ref-type="supplementary-material"}. Initial isotopocules/isotopologues concentrations and final isotopic compositions {#sec008} --------------------------------------------------------------------------------- The initial isotopocules (GM) and isotopologues (SM) concentrations as well as the final C and Cl isotopic compositions were determined as suggested by Jin et al. \[[@pone.0202416.ref013]\] and Hunkeler et al. \[[@pone.0202416.ref011]\] and are described in [S1 File](#pone.0202416.s001){ref-type="supplementary-material"}. Model implementation, evaluation and comparison {#sec009} =============================================== Implementation method {#sec010} --------------------- Both GM and SM models were applied to simulate dechlorination of PCE to TCE (PT), PCE to *c*DCE with TCE accumulation (PTD), and TCE to *c*DCE (TD). Simulations with GM were performed with Matlab while simulations with SM were performed both with Matlab and COMSOL Multiphysics (refer to [S1 File](#pone.0202416.s001){ref-type="supplementary-material"} for the differential equations used in COMSOL and to [S2 File](#pone.0202416.s002){ref-type="supplementary-material"} for the documented modelling files used in this study). Considering TCE dechlorination to *c*DCE (i.e. in TD and PTD simulations), we assume that the Cl positions in TCE are strictly distinguished between reactive and non-reactive positions. The *c*DCE Cl isotopic composition thus reflects the secondary isotopic effect only \[[@pone.0202416.ref017]\]. For simplification purposes, it is also assumed that no selective interconversion of *cis*/*trans*-DCE intermediates occur. The kinetic parameters (i.e., μ~MAX~ and K~m~) chosen for all simulations were comparable to those shared by Yu and Semprini \[[@pone.0202416.ref024]\] and Maymo-Gatell et al. \[[@pone.0202416.ref025]\] while the selected C and Cl enrichment factors are in the generally observed experimental range \[[@pone.0202416.ref022], [@pone.0202416.ref026]\] ([Fig 3](#pone.0202416.g003){ref-type="fig"} and Table F in [S1 File](#pone.0202416.s001){ref-type="supplementary-material"}). Simulations were performed until the concentration of the degraded compound reached 1% of the initial concentration, thus remaining in an experimentally representative context. ![Simulation results, experimental results (one replicate per experiment), and corresponding optimized parameters.\ NSE: Nash Sutcliff Coefficient. Blue, green and red lines correspond to simulated PCE, TCE, and *c*DCE, respectively. Red circles, blue triangles and green triangles correspond to experimental PCE, TCE and *c*DCE. μ~MAX~ is given in μmol·g of protein^-1^·s^-1^; Y is given in g of protein·(μmol of released chloride)^-1^; K~m~ is given in μmol·L^-1^; isotopic effects e are given in ‰; t~lag~ is given in h; P, T, D stand for PCE, TCE and *c*DCE; p and s stand for primary and secondary. ε~C~ and ε~Cl~ are the overall C and Cl enrichment factors in ‰ determined by application of Rayleigh to the simulated data. Corresponding ε~C~ and ε~Cl~ experimentally determined are given in brackets. ^a^: Badin et al. \[[@pone.0202416.ref027]\].](pone.0202416.g003){#pone.0202416.g003} Comparison method {#sec011} ----------------- In order to compare both models and to determine in which cases the detail of the GM is necessary, 31 different sets of isotopic parameters were defined: some which differed in C and Cl enrichment factors, some with no secondary isotopic effect, and some with normal/inverse secondary isotopic effects. Simulations were then performed for each set of parameters with both models (GM and SM). The average isotopic effect of the three Cl atoms located in non-reacting positions was used as Cl secondary isotopic effect in SM. These isotopic parameters are summarised in Table A in [S1 File](#pone.0202416.s001){ref-type="supplementary-material"}. In order to evaluate the goodness of fit between SM and GM with regards to chloroethenes concentration and isotopic behaviour along degradation, the Nash-Sutcliff efficiency coefficient (NSE) \[[@pone.0202416.ref028]\] and a normalised maximal absolute error coefficient (NME) were determined. NSE varies between---∝ and 1, a value of 1 corresponding to a perfect fit and is given by: $$NSE = 1 - \frac{\sum_{i = 1}^{N}\left( {S_{i} - Q_{i}} \right)^{2}}{\sum_{i = 1}^{N}\left( {Q_{i} - Q} \right)^{2}}$$ The NME is given in percent and corresponds to the maximal absolute difference between GM and SM, divided by the maximal amplitude of GM. Compared to the established NSE coefficient, the NME allows a more conservative comparison between the two models and represents the worst case error. The discrepancy ratio η previously defined by Jin et al. \[[@pone.0202416.ref013]\] was additionally used to evaluate the bias in SM due to the fact that we are solving the differential equations for C and Cl isotopologues of a compound independently and thus twice simulating this compound. The more the value of η deviates from 1, the higher is the discrepancy between compound concentrations simulated with the C and Cl systems. NSE, NME and η determined for all parameter sets are given in Tables B and C in [S1 File](#pone.0202416.s001){ref-type="supplementary-material"}. Results of GM vs. SM comparison {#sec012} ------------------------------- NSEs of 0.995 to 1 were obtained for all simulated species (i.e. biomass, chloroethenes concentrations and isotopic ratios) when comparing GM and SM simulated data using the same set of parameters. This indicates that both models simulate chloroethenes concentrations and isotopic behaviours almost identically. An exception is observed in the case of *c*DCE Cl isotopic composition where some NSEs ranging from -3 to -16 are obtained. Such deviation however occurs only when non-reactive Cl atoms in PCE show different secondary isotope effects depending on their position relative to the reacting bond (PTD_diff_sec, Table B in [S1 File](#pone.0202416.s001){ref-type="supplementary-material"}). The observed discrepancy results from the unequal Cl isotope distribution in TCE due to the occurrence of different secondary Cl isotopic effects during transformation of PCE to TCE. The fact that this degradation step further affects the *c*DCE Cl isotopic composition is taken into account when applying the GM but is lost when applying the SM, hence explaining the discrepancy. This *c*DCE specific discrepancy is reflected by NME ranging from 47 to 94% as well (Table C in [S1 File](#pone.0202416.s001){ref-type="supplementary-material"}). However, the absolute maximum difference in *c*DCE Cl isotopic composition between GM and SM is actually of 2 ‰ which is only slightly higher than the analytical uncertainty. Such discrepancy thus poses a problem primarily when little overall shifts are observed in *c*DCE Cl isotopic composition. Apart from this special case, a maximum NME of 2% was found associated with *c*DCE C isotopic composition during TCE degradation to *c*DCE where inverse secondary isotopic effects were considered. Among the 31 simulations with different parameter sets, NME was \< 1% in 96% of the cases (Table C in [S1 File](#pone.0202416.s001){ref-type="supplementary-material"}). These results confirm that GM and SM both simulate chloroethenes concentrations and isotopic behaviours almost identically except for *c*DCE Cl isotopic composition when different secondary isotopic effects occur during the transformation step of PCE to TCE in the overall degradation of PCE to *c*DCE. Finally, acceptable discrepancy values η related to the simultaneous resolution of differential equation systems for compounds relatively to C and Cl ranging from 0.979 to 1.008 were observed. As the discrepancy between SM and GM is negligible in most cases and as η remains in a reasonable range for the 31 sets of simulations, SM can be considered as a sufficiently representative model for most cases. *c*DCE Cl isotopic composition from PCE to *c*DCE degradation where different secondary isotopic effects are considered during the transformation step of PCE to TCE constitutes the only exception. While previous studies have calculated different position-specific isotope effects for some compounds \[[@pone.0202416.ref029]\], these effects have not been experimentally observed for chlorinated ethenes. Thus, it is not presently clear whether the phenomenon investigated in this exceptional case, which would necessitate the GM, occurs in reality. Finally, depending on the required simulation accuracy, either GM or SM may be chosen. Model responses {#sec013} --------------- Graphs representing the chloroethenes concentrations, C and Cl evolution as a function of time as well as dual C-Cl isotope plots obtained for a selection of the 31 simulation sets are given in Table D in [S1 File](#pone.0202416.s001){ref-type="supplementary-material"}. All simulations responded as expected for the different sets of isotopic effects ([Fig 3](#pone.0202416.g003){ref-type="fig"} and Table D in [S1 File](#pone.0202416.s001){ref-type="supplementary-material"}). For example, for one element, no enrichment was observed when all enrichments associated with this element were set to 0 ‰. When secondary effects were set to 0 for Cl, the initial Cl isotopic composition of TCE equalled that of PCE for PT (e.g. PT_no_sec, Table D in [S1 File](#pone.0202416.s001){ref-type="supplementary-material"}). On the contrary, when a normal secondary isotopic effect (i.e., ε~ClSec~ \< 0 ‰) was set for Cl, the initial Cl isotopic composition of TCE was lighter than that of PCE for PT (e.g. PT_normal_sec, Table D in [S1 File](#pone.0202416.s001){ref-type="supplementary-material"}). Experimental data simulation by model SM {#sec014} ======================================== Method {#sec015} ------ We have shown that when evaluating and comparing GM and SM both models almost identically simulate chloroethenes concentrations and the evolution of their relative C and Cl isotopic compositions along reductive dechlorination of PCE to *c*DCE within an experimentally plausible frame. More particularly, the goodness of fit between SM and GM is confirmed by NME \< 1% and the applicability of SM is supported by η \> 0.989 for the sets of parameters used to fit the experimental data. Simulations performed to assess to what extent the developed model can truly reproduce experimental data were hence carried out with SM. Results from a former study, as well as additional experiments whose methods and results are described in [S1 File](#pone.0202416.s001){ref-type="supplementary-material"}, were simulated. Simulations were fit on one replicate of each set of experiments, i.e. for one replicate of reductive dechlorination of PCE to TCE (PT), PCE to *c*DCE (PTD), and TCE to *c*DCE (TD), respectively. Experimentally observed lag phases *t*~lag~ of 45 h, 29 h and 205 h for PT, PTD, and TD, respectively, were taken into account when simulating the experimental data. Simulations were performed for periods corresponding to the experimentally observed degradation time. The kinetic parameters (μ~MAX~, *K*~*m*~) and *t*~lag~ were optimized first, followed by optimization of isotopic parameters (*ε*). The resulting kinetic parameters are in the same range as previously reported values \[[@pone.0202416.ref022], [@pone.0202416.ref023]\]. Isotopic parameters were optimized based on a range varying around experimentally determined C enrichment factor and Cl primary and secondary isotopic effects. The goodness of fit between simulated and experimental data was evaluated by NSE. Results {#sec016} ------- Chloroethene concentrations and isotopic composition plots of both simulated and experimental data are shown in [Fig 3](#pone.0202416.g003){ref-type="fig"}, as are NSEs and optimized model parameters. Concentrations simulated with Monod kinetics are in strong agreement with experimentally measured concentrations (NSEs from 0.66 to 1.00). Isotopic compositions also show generally good agreement with NSEs \> 0.91 in 67% of cases. Notably, the unusual inverse secondary Cl isotopic effect observed for PT and PTD when assuming a one-step scenario could be simulated. This indicates that the developed modelling approach can reliably predict experimental data ([Fig 3](#pone.0202416.g003){ref-type="fig"}). One clear outlier is the C isotopic composition of TCE associated with PT where a NSE of -0.14 was determined where a poorer agreement between simulated and experimental data is observed. This is consistent with the large variability of observed dual C-Cl isotope slopes associated with TCE from PT between experimental replicates which hindered determination of a unique dual C-Cl isotope slope associated with TCE from PT (Fig C and Table E in [S1 File](#pone.0202416.s001){ref-type="supplementary-material"}). Conclusions {#sec017} =========== The dynamics of C and Cl isotopes of PCE, TCE and *c*DCE during sequential reductive dechlorination when taking into account secondary isotope effects and Monod kinetics were successfully simulated when C and Cl isotopes were considered both simultaneously (GM) and separately (SM). Except for a specific case, NSEs \> 0.995 and NMEs \< 2% were obtained when comparing GM and SM simulated data. This indicates that both models almost identically simulate chloroethenes reductive dechlorination for a large set of isotopic effect combinations. The only exception applies for the *c*DCE Cl isotopic composition when differential Cl secondary isotopic effects are considered for the PCE to TCE dechlorination step during PCE to *c*DCE degradation. Here, only GM is able to produce reliable results. However, this type of phenomenon has not yet been documented experimentally for chlorinated hydrocarbons. While GM represents the most accurate and detailed way to simulate the evolution of isotopic composition over time during degradation, the less complex and computationally demanding model SM has been shown to be applicable for a majority of cases. Simulation of experimental data was performed with the models for PCE dechlorination to TCE (PT) and *c*DCE (PTD) as well as for TCE dechlorination to *c*DCE (TD). All models faithfully reproduced the experimental data with NSE \> 0.66 except for TCE C isotopic composition where NSE = -0.14, reflecting the inconsistency between experimental replicates. These results underscore the potential for further incorporation of isotope data into reactive transport models simulating processes occurring at multiple scales. Finally, as highlighted by Meckenstock et al. \[[@pone.0202416.ref030]\], there is a need for understanding processes affecting contaminant degradation at various scales (from conceptual model of aquifers to mass transfer through cell membranes and biochemical enzymatic reaction) in order to increase the accuracy of plume fate prediction where biodegradation occurs. These results offer the possibility to integrate information on processes occurring at the organism scale into models addressing contaminant degradation at the aquifer scale. Supporting information {#sec018} ====================== ###### Supporting information to the article. Additional schematic explanations, initial isotopologue/isotopocule concentration calculation, final isotopic composition calculation, tables summarising parameters used for the simulations as well as a selection of simulation results and corresponding NSE and NME, experimental methods and results, and differential equations used in COMSOL. (PDF) ###### Click here for additional data file. ###### Zip file containing documented Matlab and COMSOL files. The COMSOL and Matlab files used for the modelling aspects of the paper. A detailed "read me" document describing the individual Matlab files is also supplied. (ZIP) ###### Click here for additional data file. Christof Holliger (EPFL) is acknowledged for providing the bacterial consortia and Géraldine Buttet (EPFL) is thanked for her help with setting up microcosms. Alice Badin was supported by the Marie Curie Initial Training Network \"ADVOCATE---Advancing sustainable in situ remediation for contaminated land and groundwater,\" funded by the European Commission, Marie Curie Actions Project No. 265063. Landon Halloran was supported in part by Swiss National Science Foundation (FNS/SNF) project number 166233. [^1]: **Competing Interests:**The authors have declared that no competing interests exist.

The Contribution will be made Open Access under the terms of the Creative Commons Attribution License which permits use, distribution and reproduction in any medium, provided that the Contribution is properly cited. The copyright line for this article was changed on 27 June 2015 after original online publication. **Funding agencies:** The study was funded by Ataxia UK and the Medical Research Council (G0501740). P.G. works at University College London Hospitals/University College London which receives a proportion of its funding from the Department of Health\'s National Institute for Health Research Biomedical Research Centres funding scheme. **Relevant conflicts of interest/financial disclosures:** The authors received the following support for employment during the research: L.B.: Ataxia UK (studentship to B.D., J.M., P.G.). J.M.: Plymouth University. D.V.: Medical Research Council (grant to B.D.). P.G.: University College London. B.D.: Medical Research Council; University College London. Full financial disclosures and author roles may be found in the online version of this article. Balance disorders are commonly observed after cerebellar lesions arising from genetic causes, ischemia, tumors, alcoholism, and trauma.[1](#mds26227-bib-0001){ref-type="ref"}, [2](#mds26227-bib-0002){ref-type="ref"}, [3](#mds26227-bib-0003){ref-type="ref"}, [4](#mds26227-bib-0004){ref-type="ref"} However, we have no clear understanding of the cerebellum\'s role in balance control or the range of fundamental deficits that might be caused by different cerebellar lesions. Balance control involves acting on information about the body\'s current state of stability signaled by multiple sensory modalities. The cerebellum has the potential to participate in this process, because it either directly or indirectly receives considerable multisensory information known to be important for balance, including that from vestibular,[5](#mds26227-bib-0005){ref-type="ref"} proprioceptive,[6](#mds26227-bib-0006){ref-type="ref"} somatosensory,[7](#mds26227-bib-0007){ref-type="ref"} and visual[8](#mds26227-bib-0008){ref-type="ref"} sources. Here we pursue this idea by asking whether cerebellar disease is accompanied by a specific deficiency of sensorimotor processing for balance, and if so, whether the deficiency generalizes across all sensory modalities. To examine these questions, we studied a cohort of patients with spinocerebellar ataxia type 6 (SCA6). SCA6 causes death of Purkinje cells in the superior and anterior parts of the cerebellum and gliosis in the flocculo‐nodular lobe.[9](#mds26227-bib-0009){ref-type="ref"}, [10](#mds26227-bib-0010){ref-type="ref"} but with little or no extracerebellar involvement.[10](#mds26227-bib-0010){ref-type="ref"} Thus, it is a rare but well‐defined and relatively pure form of cerebellar degeneration, which during quiet stance causes clear balance impairments that scale with disease severity.[4](#mds26227-bib-0004){ref-type="ref"} The classical approach for studying balance is to perturb the body and measure the ensuing response. However, natural perturbations of the body inevitably stimulate multiple sensory systems simultaneously, making it difficult to analyze the processing of information from each sensory channel. The approach we have adopted, therefore, is to stimulate each of the three main sensory channels (visual, vestibular, and proprioceptive) in isolation, using stimuli that do not directly perturb the body but that nonetheless produce well‐defined postural responses. The three modes of stimulation were chosen to produce similar postural responses in the same directions so that any differences in response behavior could be attributed to the sensory channel rather than to the motor system generating the response. We consider two fundamental sensorimotor functions that have been proposed for the cerebellum, namely, control of response scaling[11](#mds26227-bib-0011){ref-type="ref"}, [12](#mds26227-bib-0012){ref-type="ref"}, [13](#mds26227-bib-0013){ref-type="ref"} and coordinate transformation.[14](#mds26227-bib-0014){ref-type="ref"}, [15](#mds26227-bib-0015){ref-type="ref"}, [16](#mds26227-bib-0016){ref-type="ref"} If response scaling of a sensorimotor loop is deficient, the amplitude of the balance response will be either too small or too large. With a deficiency in the coordinate transformation of information from a sensory to an action coordinate frame, such as from head coordinates to leg coordinates, the direction of the balance response may be incorrect or excessively variable. Methods {#mds26227-sec-0002} ======= Procedures were approved by the University College London Hospitals NHS Trust ethics committee, and consent was obtained from participants in accordance with the declaration of Helsinki (2004). Subjects {#mds26227-sec-0003} -------- Sixteen subjects with SCA6 from different families were recruited from the Ataxia Centre at the National Hospital of Neurology and Neurosurgery. Sixteen healthy control subjects (HC) were recruited from a local advertisement and acted as controls matched to patients by age, height, and weight. Subjects with SCA6 were included if they 1) were 18 y of age or older, 2) had a confirmed genetic diagnosis of SCA6; 3) had a score greater than zero on the Scale for Assessment and Rating of Ataxia[17](#mds26227-bib-0017){ref-type="ref"} (SARA) or nystagmus. Note that subject 3 scored zero on the SARA but had nystagmus and a subjective feeling of unsteadiness. Subjects in either group were excluded if they 1) were unable to walk 10 m unaided; 2) were unable to stand independently for 10 s with their eyes closed; 3) were taking drugs (medication or alcohol) with side effects of dizziness, drowsiness, or muscle weakness; or 4) had current or past medical conditions, other than SCA6, that could affect balance. No subjects reported headaches or migraines within the week before testing. Clinical Rating of Disease Severity and Sensory Function {#mds26227-sec-0004} -------------------------------------------------------- The SCA6 subjects were assessed using SARA to provide a measure of disease severity (score: 0 = no ataxia, 40 = most severe ataxia). The Inventory of Non‐Ataxia Symptoms was used to screen for non‐ataxia signs.[18](#mds26227-bib-0018){ref-type="ref"} Sensory examination was carried out in all SCA6 subjects. Magnetic resonance imaging reports were reviewed to ensure that only those with restricted cerebellar atrophy were included. Ocular examinations were undertaken and the presence of clinically detectable abnormal features recorded, such as nystagmus, oscillopsia, broken smooth pursuit, and ophthalmoplegia. Biosthesiometer ascending and descending threshold measures of vibration sensitivity were collected over the central tibialis anterior (TA) and medial gastrocnemius (mGAS) muscle bellies, and monofilament tests of sensitivity to light pressure (10 g) were collected using the standardized procedures outlined previously.[19](#mds26227-bib-0019){ref-type="ref"} Measures of near visual acuity and nature of spectacle use was documented (near/distance correction, uni/bifocal). Subjects were asked whether they had ever experienced or were currently experiencing any vertigo symptoms (dizziness, spinning, nausea, migraines). Instrumentation {#mds26227-sec-0005} --------------- Subjects stood on a force plate (model 9286AA, Kistler, Winterthur, Switzerland) that recorded ground reaction forces. Whole‐body motion was recorded using a three‐dimensional motion‐capture system (CODA, Charnwood Dynamics, Rothley, UK). Rigid clusters of four infrared emitting diodes were fixed using non‐slip elastic straps to the head, the torso (level of C7 vertebrae), and the back of the pelvis, and three diodes were attached to each shank and each foot. All signals were synchronized and sampled at 200 Hz. Procedure {#mds26227-sec-0006} --------- Unperturbed body sway was initially recorded for 40 s while subjects stood with a 4‐cm stance width (distance between medial borders of the feet), facing a wall at a distance of 2 m with eyes open. This provided a measure of baseline instability under conditions that were shown previously to give the best correlation with clinical disease severity.[4](#mds26227-bib-0004){ref-type="ref"} For perturbation trials, the stance width was increased to 8 cm, because patients were more stable and therefore found it less tiring when standing for prolonged periods. However, as shown previously,[4](#mds26227-bib-0004){ref-type="ref"} this increase had no effect on baseline body sway in the anteroposterior direction. The three main sensory modalities were investigated using sensory perturbation techniques: 1) visual perturbations in the form of visual motion stimuli (MVS; cw, clockwise; ccw, counter‐clockwise), which leads to a postural response in the same direction as the scene movement[20](#mds26227-bib-0020){ref-type="ref"}; 2) vestibular perturbations using galvanic vestibular stimulation (GVS; r+, anode right cathode left; l+, anode left cathode right)), which evokes a postural response in the direction of the anodal ear[21](#mds26227-bib-0021){ref-type="ref"}; 3) proprioceptive perturbations using muscle vibration (VIB; ts, triceps surae; ta, tibialis anterior) of lower leg muscles, which leads to a postural response in a direction that shortens the vibrated muscle.[22](#mds26227-bib-0022){ref-type="ref"} To compare across sensory modalities, the postural responses were designed to be similar in form, magnitude, and direction for a healthy standing subject. The response directions that could be studied were constrained by the vibratory stimuli, which were applied to ankle flexors and extensors to produce postural responses in the anteroposterior direction. For vestibular and visual stimuli to evoke responses also in the anteroposterior direction, the head was rotated in yaw through 90 degrees (GVS response is directed approximately along the interaural line), and the visual scene movement was limited to rotation in the sagittal plane about the ankle axis. Technical details of the various stimuli employed are given in Supplemental Data. For the perturbation trials, subjects stood with their feet 8 cm apart and head rotated to the right through 90 degrees to face the visual scene 0.4 m away in the sagittal plane. This scene remained stationary in all conditions expect for the moving visual stimulus condition. Subjects wore spectacles or contact lenses if required and a visual field restrictor that limited vision to a 74‐degree horizontal viewing angle and 32‐degree vertical viewing angle (approximating a 60 × 25‐cm visible screen area). Earplugs (32 dB) and background white noise masked equipment‐related noise. The subject wore a safety harness that prevented vertical drops of 5 cm or more. After a random baseline period of 3 to 4 s, a 2‐s sensory stimulus was given followed by a 5‐s post‐stimulus period. Twenty trials of each stimulus (10 per direction) were randomly intermixed with 20 no‐stimulation trials. Trials were randomized according to stimulus type and its direction. Audible tones signaled the start and the end of each trial. Sufficient time was provided between stimuli to allow subjects to adopt the standardized starting position. Rests were included as required during the tests. Measurement {#mds26227-sec-0007} ----------- Body motion was measured from body displacement approximately at the level of the C7 vertebra. This was converted to an angular measure, using the height of the marker‐cluster above ground level. Stimulus‐evoked response mean magnitude and direction were measured from each subject\'s mean traces between 0.2 s and 1.0 s (responses to the moving visual scene \[MVS\] were also measured at 2 s). Direction variability and habituation were measured from single‐trial responses. Baseline sway speed was calculated from the 40‐s period of quiet stance as total horizontal‐plane path/duration as described previously.[7](#mds26227-bib-0007){ref-type="ref"} See Supplemental Data for measurement details. Statistical Analysis {#mds26227-sec-0008} -------------------- Between‐group comparisons of response magnitudes were carried out using two‐tailed Student\'s *t* tests for independent samples (PASW Statistics 18, IBM, Armonk, NY, USA). Equal variances were not assumed if Levene\'s test of equality of variances yielded *P* values less than 0.05. Differences between groups were tested separately for each sensory stimulation mode (GVS, MVS, VIB) and direction (forward, backward), yielding six comparisons for each measure. To account for multiple comparisons, the significance level was set at *P* \< 0.01. Associations between response magnitude and disease severity (SARA score) were determined by using Pearson\'s correlation coefficient. Analyses of response direction were performed using circular‐data statistical procedures described in Supplemental Data. Results {#mds26227-sec-0009} ======= Anthropometric data, clinical assessments, and baseline sway speed are detailed in Table [1](#mds26227-tbl-0001){ref-type="table-wrap"}. All SCA6 subjects scored zero on the Inventory of Non‐Ataxia Symptoms scale, indicating no clinically detectable non‐ataxia symptoms, and all displayed horizontal gaze‐evoked nystagmus with saccadic pursuit. ###### Subject anthropometrics and baseline clinical measures related to disease severity of SCA6 and sensory function Group Subject Sex Age Height (m) Weight (kg) SARA SARA Change Bio_TA Bio_mGAS Monofil Vision Vertigo Sway Speed (deg/s) ---------- -------------------- ------------- ---------------- ---------------- ---------------- --------------- -------------- ---------------- ---------------- --------------- -------------- ---------- -------------------- SCA6 1 M 65 1.78 76.5 9.5 NA 22.7 16.8 9.5 1.8 N ‐ PM 0.25 2 F 71 1.64 67.8 11 3 20 24.3 10 1.25 N ‐ PMHT 0.28 3 F 43 1.57 52.9 0 0 12 13.3 10 0.45 N 0.17 4 F 70 1.65 65.5 17 1.6 23.3 41.3 9.5 1.25 N ‐ PMHT 0.56 5 F 70 1.65 65.5 17 5.1 23.3 41.3 9.5 1.25 N ‐ PMHT 0.32 6 F 67 1.6 63.5 14 0.8 12.5 19.5 10 3.2 N 0.66 7 M 62 1.6 76.6 6 NA 10 25.8 10 3.6 N 0.22 8 M 40 1.83 63.5 17 1.5 14.3 14.3 10 0.4 N 1.10 9 F 66 1.61 54.9 13 1.8 17.7 33.8 9.5 2.6 N ‐ PMHT 0.39 10 F 68 1.6 79.9 13 1.5 10.2 30.9 10 1.5 N 0.30 11 M 60 1.72 74.5 7 NA 30.8 31 10 6.3 N ‐ PMHT 0.21 12 M 65 1.83 81.5 9.5 2.1 10.7 20.5 10 1.6 N 0.44 13 F 73 1.55 78.9 22 2.8 12 33.1 10 3.2 N ‐ PH 1.00 14 M 62 1.8 86 12 NA 28.7 19.3 10 2.25 N ‐ PH 0.51 15 M 46 1.69 77.6 20.5 2.2 11.3 14 10 1.125 N 2.21 16 F 68 1.52 79.4 3 0.7 21 35 9.75 0 N ‐ PMHT 0.42 **SCA6** **Mean** **9F : 7M** **62.3** **1.67** **71.5** **12.0** **1.9** **17.5** **25.9** **9.9** **2.0** **0.56** **CI (low, high)** **57.2, 67.3** **1.62, 1.71** **66.7, 76.3** **9.0, 14.9** **1.3, 2.6** **14.2, 20.9** **21.2, 30.5** **9.8, 10.0** **1.2, 2.7** **0.31, 0.82** **HC** **Mean** **8F : 8M** **60.3** **1.69** **75.3** **0.0** **0.0** **18.5** **21.6** **9.8** **1.7** **0.23** **CI (low, high)** **55.1, 65.5** **1.64, 1.75** **69.4, 81.2** **NA** **NA** **15.1, 21.9** **16.6, 26.6** **9.5, 10.0** **0.8, 2.7** **0.16, 0.30** KEY: SARA, Scale for assessment and rating of ataxia (0‐40); SARA change, mean change in SARA per year to date, positive change, worsening disease severity, NA, not available; Bio_TA, mean of bilateral, 3 trial repeats, ascending and descending threshold scores tested over central tibialis anterior muscle bellies; Bio_mGAS, tested over midpoint of gastocnemius heads at the lower border of the Achilles insertion point and over soleus bellies; Monofil, mean monofilament score /10 of bilateral testing; Vision, near visual acuity score held at 40‐cm distance, mean of left and right eyes; Vertigo, current vertigo questioning; N, No current reports of vertigo signs within 1 week; P, Past reports; M, Migraine including dizziness; H, Headaches; T, Travel sickness; Sway speed, trunk sway speed collected over 40 s; CI, 95% confidence intervals. No group differences in vibration thresholds (TA: *P* = 0.689, mGAS: *P* = 0.225), monofilament testing (*P* = 0.657), or near visual acuity (*P* = 0.704). Mean sway speeds during quiet stance were significantly higher in the SCA6 group (*P* = 0.009). Sway speed correlated with disease severity assessed by SARA (*r* = 0.705, *P* \< 0.001). Figure [1](#mds26227-fig-0001){ref-type="fig"}A shows the time‐course of the sagittal‐plane component of the group mean responses evoked by the three sensory stimuli. The time‐course, magnitude, and direction of responses were deemed sufficiently similar to compare the three sensory modalities. {#mds26227-fig-0001} Response Magnitude {#mds26227-sec-0010} ------------------ The group mean response magnitudes are shown for each stimulus modality and polarity in Figure [1](#mds26227-fig-0001){ref-type="fig"}B. In general, the SCA6 group tended to show larger responses than controls. The magnitude difference was highly significant for MVS (cw: t\[17.55\] = 5.67, *P* \< 0.001; ccw: t\[16.75\] = 3.25, *P* = 0.005) but only showed trends for one of the two polarities for VIB (ts: t\[22.91\] = 2.09, *P* = 0.048; ta: t\[16.78\] = 1.81, *P* = 0.88) and for GVS (r+: t\[29\] = 2.51, *P* = 0.018; l+: t\[29\] = 0.62, *P* = 0.540). The MVS response was measured over a different period compared with the VIB and GVS responses (see Methods). However, when measured over the same time period (0.2‐1 s), the MVS response magnitude combined for the two directions remained highly significantly larger for SCA6 than HC (SCA6, 0.47 ± 0.06; HC, 0.21 ± 0.02; t\[20.10\] = 4.23, *P* \< 0.001). The magnitude of each single‐trial response was measured to investigate habituation to repeated presentation of the same stimulus. Plots of response magnitude versus stimulus presentation order (shown in Supplemental Data) indicated a uniform lack of habituation. Thus, the response magnitudes to the first and the ninth presentations were not significantly different from each other for all types of stimulus in both groups of subjects (*P* \> 0.05 in all cases). Response Direction {#mds26227-sec-0011} ------------------ Statistical analyses of the group mean response directions are shown in Table [2](#mds26227-tbl-0002){ref-type="table-wrap"}. The response directions were significantly concentrated around a mean direction for all stimulus conditions in both groups. No significant differences were seen in mean response direction for the two groups. The dispersion of response directions around the mean, measured by angular deviation reflecting response direction variability within a group, was not significantly different between groups for any stimulus condition, although a trend was seen for a greater dispersion in SCA6 for the VIB‐ts condition (*P* = 0.023). ###### Response directions measured from mean traces of upper trunk displacements **GVS** **MVS** **VIB** -------------------- ---------------- --------- --------- ------- -------- ------- ------ ------ **SCA6** N 15 15 16 16 16 16 Mean direction (°) −82.95 80.72 −96.30 71.46 −72.59 70.02 Concentration *r* 0.933 0.944 0.911 0.793 0.877 0.920 Angular deviation 21.05 19.11 24.15 36.83 28.44 22.92 **HC** N 16 16 16 16 16 16 Mean direction (°) −88.08 72.49 −79.87 71.42 −78.89 82.18 Concentration *r* 0.937 0.968 0.859 0.764 0.977 0.977 Angular deviation 20.39 14.44 30.45 39.33 12.22 12.16 **SCA6 vs HC** Mean direction *F* 0.43 1.70 2.56 0.00 0.60 3.24 *P* 0.515 0.203 0.120 0.998 0.443 0.082 Angular deviation *U* 141 142 164 147 188 155 *P* 0.423 0.401 0.184 0.491 0.023 0.323 Notes: One SCA6 subject did not contribute GVS responses because of technical failure. Response direction is reported relative to the visual screen, with 0° indicating motion directly toward the screen, 90° to the left parallel to the plane of the screen, and −90° to the right. All mean directions were highly significantly concentrated (*P* \<\< 0.001). Mean directions compared using Watson‐Williams test. Angular dispersion compared using Wallraff procedure and tested with two‐tailed Mann‐Whitney test. *P* denotes probability, with significance set at *P* \< 0.01. GVS, galvanic vestibular stimulation; MVS, moving visual scene; VIB, muscle vibration; R+, anode right; L+, anode left; CW, clockwise; CCW, counter‐clockwise; TS, triceps surae; TA, tibialis anterior; SCA6, spinocerebellar ataxia type 6; HC, healthy control. Within‐Subject Response Direction Variability {#mds26227-sec-0012} --------------------------------------------- Although the directions of the each subject\'s mean responses were not different for the two groups, possibly the SCA6 subjects were abnormally variable from trial to trial in their response directions. This was quantified by calculating each subject\'s angular deviation of single‐trial responses. Table [3](#mds26227-tbl-0003){ref-type="table-wrap"} gives the group mean and variability of this measure and shows that the within‐subject response direction variability was not significantly different between groups. ###### Within‐subject response direction variability (angular deviation, degrees) measured from single trials of upper trunk displacement **GVS** **MVS** **VIB** ---------------- --------- --------- --------- ------- ------- ------- ------- **SCA6** n 15 15 16 16 16 16 Mean (°) 31.38 42.31 32.42 45.33 35.02 40.57 SD 16.84 20.69 15.46 18.49 21.00 19.93 Median 24.75 49.64 30.69 48.40 36.01 33.13 Interquartile 28.90 33.40 23.12 17.48 31.43 27.79 **HC** N 16 16 16 16 16 16 Mean (°) 26.63 26.48 42.89 42.60 24.92 35.49 SD 17.34 15.08 16.49 18.24 15.57 16.68 Median 22.53 22.11 43.13 37.11 22.48 35.95 Interquartile 16.07 19.60 18.91 22.36 18.42 21.05 **SCA6 vs HC** *P* 0.446 0.030 0.073 0.616 0.171 0.564 Notes: One SCA6 subject did not contribute GVS responses because of technical failure. Angular deviations compared using two‐tailed Mann‐Whitney test. *P* denotes probability, with significance set at *P* \< 0.01. GVS, galvanic vestibular stimulation; MVS, moving visual scene; VIB, muscle vibration; R+, anode right; L+, anode left; CW, clockwise; CCW, counter‐clockwise; TS, triceps surae; TA, tibialis anterior; SCA6, spinocerebellar ataxia type 6; HC, healthy control. Correlation of Response Magnitude With Disease Severity {#mds26227-sec-0013} ------------------------------------------------------- Response magnitudes were averaged for the two polarities of each stimulus modality and correlated with SARA scores. As shown in Figure [1](#mds26227-fig-0001){ref-type="fig"}C, a significant positive correlation was found between SARA and MVS response magnitude (*r* = 0.543, *P* = 0.030), but not for GVS (*r* = 0.108, *P* = 0.702) or VIB (*r* = 0.387, *P* = 0.138). Discussion {#mds26227-sec-0014} ========== We have asked whether the balance instability of SCA6 subjects is associated with a deficiency of sensorimotor processing and, if so, whether the deficiency generalizes across all sensory modalities. As reported previously,[4](#mds26227-bib-0004){ref-type="ref"} balance control of this SCA6 group was abnormal. Even without sensory perturbations, these patients were more unstable than the control group, showing greater body sway during quiet stance, which scaled with disease severity. Despite this instability, many aspects of the responses to single‐channel sensory perturbations were largely unaffected. The notable exception was the response magnitude to the visual perturbation, which was considerably larger than control by a factor of 3 on average. If this excessively large response were attributable simply to the underlying enhanced body sway, then all sensory stimuli should have produced similarly large responses. However, the exaggerated response was reasonably specific to the visual modality and was disease‐related because it correlated with disease severity measured by SARA. This large mean response could not be explained by differences in the rate of habituation of single‐trial responses to repeated presentation of stimuli, because neither group showed significant habituation. Our results do not replicate those of an earlier study on the visual control of balance in cerebellar patients,[23](#mds26227-bib-0023){ref-type="ref"} which employed a "moving room" stimulus not dissimilar from our moving visual scene. However, many differences exist between the two studies, including predictability of the stimulus, response direction, measurement period, and, most importantly, the clinical cohorts investigated (SCA6 vs. heterogeneous etiology). Specificity and Mechanism of Sensorimotor Disruption {#mds26227-sec-0015} ---------------------------------------------------- The two processes under consideration were response scaling and coordinate transformation. The positive finding of enlarged balance responses supports the concept of disrupted response scaling. An abnormality of response scaling is not unlike the exaggerated whole‐body response to support‐surface perturbation observed in cerebellar patients[12](#mds26227-bib-0012){ref-type="ref"} and may represent another expression of typical cerebellar dysmetria reported for limb movements[1](#mds26227-bib-0001){ref-type="ref"} and eye movements.[24](#mds26227-bib-0024){ref-type="ref"} A possible explanation for over‐scaling is that it arises from cerebellar disinhibition of sensorimotor centers outside the cerebellum caused by a loss of Purkinje cells, which exert tonic inhibitory influence on deep cerebellar nuclei.[25](#mds26227-bib-0025){ref-type="ref"} Why the brunt of the abnormality should fall on the visual channel is not clear. It may have something to do with the fact that visual flow is inherently ambiguous in that it signals motion of the environment as well as self and therefore requires a mechanism to extract the self‐motion component for balance control; vestibular and proprioceptive inputs do not require this because they directly signal changes in body state. An alternative explanation is that the enlarged visuomotor response is a direct result of abnormal cerebellar processing of visual input. The cerebellum receives retinal information indirectly from the accessory optic system and cortically processed visual information via the pons.[26](#mds26227-bib-0026){ref-type="ref"} Disturbed visual processing has been implicated in other aspects of cerebellar function. Stein[27](#mds26227-bib-0027){ref-type="ref"} suggested that the cerebellum may play an important role in the visual guidance of movement, whereas some abnormal aspects of limb‐movement trajectories in cerebellar disease have been attributed to aberrant motor responses to visual information.[28](#mds26227-bib-0028){ref-type="ref"}, [29](#mds26227-bib-0029){ref-type="ref"} A third possibility is that the over‐scaling results from an indirect visual disruption caused by poor oculomotor control. This could occur if retinal signals are distorted by the abnormal eye movements that were clinically detectable in all of the SCA6 patients studied here. Some caution is required when interpreting the lack of disruption to coordinate transformation processes. A deficit in spatial transformation of information from a sense‐organ\'s coordinate frame to an effector\'s frame would have resulted in detectable direction errors or increased direction variability. However, in this experiment, the spatial relationship between the stimulated sense organs and the body remained fixed throughout testing. A more rigorous test of this process would involve a greater variety of postural changes, for example, by studying a range of head and trunk positions with respect to the feet rather than just the one. Can the Visuomotor Disturbance Cause Balance Instability? {#mds26227-sec-0016} --------------------------------------------------------- One hypothesis for SCA6 balance impairment is that it results from a pure motor disruption, for example, dyssynergia or muscle activation timing problems.[2](#mds26227-bib-0002){ref-type="ref"}, [19](#mds26227-bib-0019){ref-type="ref"}, [30](#mds26227-bib-0030){ref-type="ref"}, [31](#mds26227-bib-0031){ref-type="ref"} Can an exaggerated visually evoked balance response provide the basis for an alternative sensory hypothesis for balance instability? The neuro‐mechanical system controlling upright stance is often modeled as a mechanical inverted pendulum under sensory feedback control.[32](#mds26227-bib-0032){ref-type="ref"} One way such a system could go unstable is if the gain of the feedback loops were set too high. The current results could be interpreted as reflecting an excessively high gain of the visual channel without a compensatory decrease in gain of the vestibular and proprioceptive channels, and so is compatible with this hypothesis. However, a simple objection is that cerebellar patients typically become even more unstable when deprived of vision.[23](#mds26227-bib-0023){ref-type="ref"} Nonetheless, the concept of instability through high feedback gains remains a possibility. This could occur because the relative gains of the different sensory channels are not fixed.[33](#mds26227-bib-0033){ref-type="ref"} If a sensory channel becomes unavailable, then the gains of the remaining sensory sources may be automatically increased.[34](#mds26227-bib-0034){ref-type="ref"} The cerebellum has been proposed to play a key role in adaptive gain control, at least for the vestibuloocular reflex,[11](#mds26227-bib-0011){ref-type="ref"}, [35](#mds26227-bib-0035){ref-type="ref"} so one could speculate that a loss of the visuomotor loop with eye closure might cause an abnormally large increase in gain of the other sensorimotor loops. More work is required to examine this hypothesis. Clinical Implications {#mds26227-sec-0017} --------------------- The increased gain of the visuomotor feedback loop for balance shown here may be related to the problems encountered by SCA6 patients in daily life. They often report that balance difficulties are particularly severe in busy visual environments, for instance, when walking alongside a busy road or in a crowd. Whatever the cause of the visuomotor disruption in SCA6, it opens an opportunity for targeted rehabilitation of their balance impairment. This could involve training of the oculomotor response[36](#mds26227-bib-0036){ref-type="ref"} or desensitisation training to respond more appropriately to potentially destabilizing moving visual cues in the environment.[37](#mds26227-bib-0037){ref-type="ref"} Author Roles {#mds26227-sec-0019} ============ 1\. Research Project: A. Conception, B. Organization, C. Execution; 2. Statistical Analysis: A. Design, B. Execution, C. Review and Critique; 3. Manuscript Preparation: A. Writing the First Draft, B. Review and Critique. L.B.: 1B, 1C, 2B, 3A, 3B J.M.: 1A, 2C, 3B D.V.: 1C, 2C, 3B P.G.: 1A, 1C, 2C, 3B B.D.: 1A, 1B, 2A, 2B, 3A, 3B Financial Disclosures {#mds26227-sec-0020} ===================== L.B.: Employment: Plymouth University J.M.: Employment: Plymouth University Grants: Progressive MS Alliance; Knowledge Transfer Partnership; VC Community Research Awards; Physiotherapy Research Foundation; Dr William M. Scholl Podiatric R&D Fund; CUC‐ESF; Royal Devon & Exeter NHS Foundation Trust; Plymouth Hospitals D.V.: Employment: University College London; Organox Ltd. P.G.: Employment: University College London Grants: European Community Grant FP7‐HEALTH‐F2‐2010‐242193 (EFACTS) Honoraria: TAKEDA Cambridge LTD B.D.: Employment: University College London Grants: Medical Research Council; Brain Research Trust; UCL Grand Challenge Honoraria: International Society for Posture and Gait Research Supporting information ====================== Additional Supporting Information may be found in the online version of this article at the publisher\'s web‐site. ###### Supplementary Information ###### Click here for additional data file. We thank all of the SCA6 patients and healthy volunteers who participated in this study. The study was funded by Ataxia UK and the Medical Research Council (G0501740). P.G. works at University College London Hospitals/University College London, which receives a proportion of its funding from the Department of Health\'s National Institute for Health Research Biomedical Research Centres funding scheme.

Introduction {#s1} ============ In neurons, mRNA transport is widely used to differentially regulate protein content in domains distant from the cell body [@pone.0011350-Bassell1]. Especially, mRNA transport and local translation are known to be involved in neuron development, synaptic functions and plasticity [@pone.0011350-Kiebler1], [@pone.0011350-Sossin1], [@pone.0011350-SanchezCarbente1]. A current model stipulates that, along the way from nuclear export to dendritic anchoring, proteins are added or removed from the mRNP complexes in a dynamic way. It was proposed that these proteins finely control the successive steps that ensure proper expression of mRNA at specific times and space. Several combinations of mRNAs and proteins form a highly heterogeneous population of ribonucleoprotein (RNP) complexes that are linked to different forms of synaptic activity and/or plasticity [@pone.0011350-Lebeau1], [@pone.0011350-Mallardo1], [@pone.0011350-Duchaine1]. In particular, two large families of RNPs have been suggested: mRNA particles and mRNA granules [@pone.0011350-Sossin1], [@pone.0011350-Mallardo1]. mRNA particles are distinguished from mRNA granules by the absence of ribosomes. It was suggested that RNA particles might represent the observed transport mRNPs [@pone.0011350-Kiebler1]. Staufen2 (Stau2), a protein mainly expressed in brain is a well accepted player for mRNA localization [@pone.0011350-Duchaine1], [@pone.0011350-Tang1]. The *Stau2* gene expresses four protein isoforms of 62, 59, 56 and 52 kDa that are generated by differential splicing ([Fig. 1](#pone-0011350-g001){ref-type="fig"}) [@pone.0011350-Duchaine1]. Stau2 binds double-stranded RNAs and is incorporated into mRNPs that move along microtubules in neuronal dendrites [@pone.0011350-Mallardo1], [@pone.0011350-Duchaine1], [@pone.0011350-Tang1]. Interestingly, its level of expression in dendrites regulates the level of transported mRNAs showing the importance of Stau2 for mRNA transport. Likely as a consequence, neurons in which Stau2 has been down-regulated by RNAi show a reduced density of dendritic spines, associated with a change in their morphology. These phenotypes result in reduced amplitude of the miniature excitatory postsynaptic currents, a measure of synaptic transmission [@pone.0011350-Goetze1]. ![Immunoprecipitation of Stau2 isoforms.\ (**A**) Schematic representation of Stau2 isoforms. The *Stau2* gene generates four different isoforms of 62, 59, 56 and 52 kDa through differential splicing. Black, grey and white boxes represent double-stranded RNA-binding (dsRBD) consensus sequence having full, partial or no RNA-binding activity, respectively. Hatched boxes represent the tubulin-binding domain (TBD). (**B**) Immunoprecipitation of Stau2 isoforms from embryonic E17-18 rat brain extracts using two different polyclonal anti-Stau2 antibodies, L1 (St2-L1) and L2 (St2-L2). The specificity of these antibodies was previously reported [@pone.0011350-Duchaine1]. Pre-immune (PI) sera were used as controls. The Stau2^56^ isoform is not visible in these cell extracts. \* represents a non-specific IgG band.](pone.0011350.g001){#pone-0011350-g001} Stau2 was described in both the nucleus and somatodendritic compartment of the cells [@pone.0011350-Duchaine1], [@pone.0011350-Macchi1]. Accordingly, Stau2 was shown to associate with nuclear factors suggesting an early role for Stau2 in mRNP assembly [@pone.0011350-Monshausen1], [@pone.0011350-Elvira1]. In the somatodendritic compartment, Stau2 associates with both mRNA granules and mRNA particles. On a sucrose gradient, Stau2^62^ co-fractionates with ribosome-free particles whereas Stau2^59^ and Stau2^52^ were found in fractions that contained ribosomes [@pone.0011350-Duchaine1]. However, the composition of these complexes is still largely unknown. In the work described in this paper, we immunoprecipitated Stau2-containing mRNPs and used a proteomic approach to identify Stau2-associated proteins. Several RNA-binding proteins and proteins of the cytoskeleton have been identified. Results {#s2} ======= Isolation and characterization of Stau2-containing mRNPs {#s2a} -------------------------------------------------------- In order to identify the protein content of Stau2-containing RNPs, extracts of embryonic rat brains were prepared and endogenous Stau2 was immunoprecipitated using the polyclonal anti-Stau2 antibody L1 and its pre-immune serum as control ([Fig. 1](#pone-0011350-g001){ref-type="fig"}). Following separation of co-immunoprecipitated proteins by SDS-PAGE, the gel was cut into bands of 3 mm and the proteins digested in-gel with trypsin. Resulting peptides were identified by mass spectrometry. In addition to Stau2, seven proteins were present in the Stau2 immunoprecipitate ([Table 1](#pone-0011350-t001){ref-type="table"} and Supplemental [Table S1](#pone.0011350.s001){ref-type="supplementary-material"}). Y box-binding protein 1 (YB1), polyadenylate-binding protein cytoplasmic 1 (PABPC1), heat-shock cognate protein 70 (hsc70) and heterogeneous nuclear ribonucleoprotein H1 (hnRNP H1) are RNA-binding proteins previously associated with other mRNPs, α- and β-tubulin are protein components of the cytoskeleton and RUFY3 (rap2-interacting protein X) is still poorly characterized. 10.1371/journal.pone.0011350.t001 ###### Proteomically identified proteins in Stau2-containing mRNPs. {#pone-0011350-t001-1} Name Peptides Description/function References -------------------- ---------- ----------------------------------------------------------------------------------------------------------------------- ------------------------------------------------------------------------- Staufen 2 12 Double-stranded RNA-binding protein. Mainly involved in mRNA transport [@pone.0011350-Duchaine1] PABPC1 7 Polyadenylate-binding protein cytoplasmic 1. Mainly involved in the regulation of translation [@pone.0011350-Mangus1] YB1 3 Y-box 1 RNA-binding protein. Mainly involved in the regulation of translation [@pone.0011350-Evdokimova1]; [@pone.0011350-Skabkin1] Hsc70 3 heat-shock cognate RNA-binding protein 70. Involved in nuclear trafficking, RNA chaperone, kinesin-mediated transport [@pone.0011350-Matsui1]; [@pone.0011350-Henics1]; [@pone.0011350-Tsai1] hnRNP H1 1 heterogeneous nuclear ribonucleoprotein H1. Poly(rG)-RNA-binding protein [@pone.0011350-Matunis1] α-tubulin (1A, 1B) 5 Component of microtubule [@pone.0011350-Lewis1] β-tubulin (2b) 12 Component of microtubule [@pone.0011350-Lewis1] RUFY3 10 Protein interacting with rap2. Role in axonogenesis [@pone.0011350-Mori1] PABPC1, YB1, and hsc70 co-immunoprecipitate with Stau2^62^ and Stau2^59^ {#s2b} ------------------------------------------------------------------------ It was previously shown that Stau2^62^ co-fractionated with ribosome-free mRNPs whereas Stau2^59^ and Stau2^52^ isoforms were found in heavy fractions that also contained ribosomes [@pone.0011350-Duchaine1]. Therefore, we first tested whether the association between Stau2 and the identified RNA-binding proteins is specific for one Stau2 isoform or present in different complexes. In addition, we determined whether the association is direct or whether it involves an RNA intermediate. To this end, N2A cells were transfected with plasmids coding for Stau2^62^-HA~3~ or Stau2^59^-HA~3~ and tagged proteins as indicated ([Fig. 2](#pone-0011350-g002){ref-type="fig"}). Cell extracts were prepared and Stau2 was immunoprecipitated using anti-HA antibody, in the presence or absence of the micrococcal nuclease. Co-immunoprecipitated proteins were analyzed by western blotting using anti-myc or anti-GFP antibody as indicated. In the absence of micrococcal nuclease, PABPC1-myc, YB1-CFP and hsc70-CFP were found in Stau2^62^-HA~3~ and Stau2^59^-HA~3~ immunoprecipitates indicating that they are present in the same complexes as Stau2-HA~3~ isoforms ([Fig. 2](#pone-0011350-g002){ref-type="fig"} A--C). In contrast, hnRNP H1-myc was not detected ([Fig. 2D](#pone-0011350-g002){ref-type="fig"}). When micrococcal nuclease was added to the cell extracts before immunoprecipitation, Stau2^62^-HA~3~ and Stau2^59^-HA~3~ interaction with hsc70-CFP was still observed ([Fig. 2C](#pone-0011350-g002){ref-type="fig"}) whereas interactions with PABPC1-myc and YB1-CFP were lost ([Fig. 2](#pone-0011350-g002){ref-type="fig"} A,B). These results show that several RNA-binding proteins are present in the same complexes as Stau2 isoforms and suggest that only the Stau2/hsc70 interaction involved direct protein-protein interaction. {#pone-0011350-g002} Direct interaction between hsc70 and Stau2^62^ {#s2c} ---------------------------------------------- To confirm the interaction between hsc70 and Stau2 at the protein level, we performed two in vitro binding assays, GST-pull down and surface plasmon resonance (SPR). To this end, bacterially expressed GST, maltose-binding protein (MBP), GST-hsc70 and MBP-Stau2^62^ fusion proteins were affinity purified ([Fig. 3A](#pone-0011350-g003){ref-type="fig"}). For the pull down assay, GST as a negative control and GST-hsc70 were attached to glutathione columns. Their ability to bind MBP-Stau2^62^ in the presence or absence of RNase A was tested by western blotting. In contrast to GST that failed to pull-down MBP-Stau2^62^, the GST-hsc70 fusion protein was able to bring down MBP-Stau2^62^ even in the presence of RNase A ([Fig. 3B](#pone-0011350-g003){ref-type="fig"}). Similarly, MBP and MBP-Stau2^62^ were fixed to the SPR sensor chip surface. Then GST and GTS-hsc70 were flowed over the SPR chip surface and their interaction with the immobilized proteins was monitored in real time. The resulting sensorgrams indicate that a direct interaction occurred between GST-hsc70 and MBP-Stau2^62^ even in the presence of RNase A ([Fig. 3C](#pone-0011350-g003){ref-type="fig"}). These *in vitro* binding studies confirm the ability of hsc70 to interact with Stau2, independent of RNA. {#pone-0011350-g003} ATP modulates the interaction between hsc70 and Stau2^62^ {#s2d} --------------------------------------------------------- hsc70 contains two functional domains, an N-terminal ATPase domain that contains the ATP/ADP-binding site and a C-terminal peptide-binding domain that contains the substrate-binding pocket [@pone.0011350-Frydman1]. Upon ATP binding and hydrolysis a conformational change is induced in the ATPase domain [@pone.0011350-Jiang1], [@pone.0011350-Mayer1] that modifies the structure of the substrate binding domain and consequently modulates its ability to bind substrates. In its ATP bound form, hsc70 binds and releases substrates quickly, while, in the ADP bound state, substrate binding and release are slow [@pone.0011350-Schmid1], [@pone.0011350-Greene1]. Therefore, we next tested whether the Stau2^62^/hsc70 association is sensitive to the presence of ATP. To this end, the co-immunoprecipitation experiment ([Fig. 4A](#pone-0011350-g004){ref-type="fig"}) and the SPR experiment ([Fig. 4B](#pone-0011350-g004){ref-type="fig"}) were reproduced in the presence of 10 mM ATP. In both cases, the presence of ATP almost completely abolished the Stau2^62^/hsc70 interaction. {ref-type="fig"}. Immunoprecipitation of Stau2-containing RNPs was performed with anti-HA antibody and the proteins detected on western blots using anti-HA or anti-GFP antibodies as needed. The experiment was done in the absence (−) or presence of either Microccocal nuclease (+RNase) or ATP (+ATP). These results are representative of at least three experiments. (**B**) As done for [figure 3C](#pone-0011350-g003){ref-type="fig"}, MBP and MBP-Stau2^62^ were immobilized on different lanes of a SPR sensor chip. GST-hsc70 or GST was injected for 3 minutes over the surfaces in the presence or absence of ATP. The baseline obtained with the MBP-coupled reference surface was subtracted from the sensorgram obtained from the MBP-Stau2^62^-coupled surface and a typical result is shown.](pone.0011350.g004){#pone-0011350-g004} Co-localisation of Stau2 isoforms with YB1 and PABPC1 in dendrites of hippocampal neurons {#s2e} ----------------------------------------------------------------------------------------- Altogether, our data suggest that we have isolated Stau2-containing mRNA particles, a sub-population of all Stau2-associated complexes in brains [@pone.0011350-Mallardo1], and that these mRNPs contain YB1, PABPC1 and/or hsc70. To determine whether these proteins can also be detected in the large granule complexes that are visible in dendrites, embryonic hippocampal neurons were first transfected with plasmids coding for PABPC1-myc, YB1-CFP, hsc70-CFP or hnRNP H1-myc and fixed. Tagged-proteins and endogenous Stau2 were detected with anti-myc or anti-GFP and anti-Stau2 antibodies, respectively ([Fig. 5](#pone-0011350-g005){ref-type="fig"}). In addition to their presence in the cell bodies, PABPC1-myc ([Fig. 5A](#pone-0011350-g005){ref-type="fig"}) and YB1-CFP ([Fig. 5B](#pone-0011350-g005){ref-type="fig"}) can be found as puncta in dendrites ([Fig. 5A](#pone-0011350-g005){ref-type="fig"}). Only a very small fraction of these puncta also stained with antibodies that recognized endogenous Stau2. This suggests that PABPC1 and YB1 are not components of the large Stau2-containing complexes in dendrites. Similarly, hsc70-CFP was homogeneously distributed in the cell body and dendrites and did not form observable puncta ([Fig. 5C](#pone-0011350-g005){ref-type="fig"}). Therefore, it is suggested that its association with Stau2 may occur outside the large dendritic RNP complexes. Finally, hnRNP H1-myc was strictly nuclear ([Fig. 5D](#pone-0011350-g005){ref-type="fig"}), indicating that, if confirmed, Stau2/hnRNP H1 association would be restricted to the nucleus. {#pone-0011350-g005} The reverse experiment was also done. Hippocampal neurons were transfected with Stau2^62^-HA~3~ and its co-localization with endogenous PABPC1 ([Fig. 6A](#pone-0011350-g006){ref-type="fig"}) and YB1 ([Fig. 6B](#pone-0011350-g006){ref-type="fig"}) was analyzed with specific antibodies. In these conditions, PABPC1 partly co-localized with Stau2^62^-HA~3~ whereas YB1 displayed only a weak co-localization. Altogether, our results suggest that PABPC1, YB1, hsc70 and hnRNP H1 may be mostly associated with Stau2 in small mRNPs and mainly absent in the large Stau2-containing mRNA granules in dendrites. {#pone-0011350-g006} Discussion {#s3} ========== Stau2 is an RNA-binding protein mainly expressed in brain [@pone.0011350-Duchaine1] and its importance for mRNA transport in dendrites [@pone.0011350-Tang1], [@pone.0011350-Jeong1] and synaptic functions [@pone.0011350-Goetze1] has been established. As a further step aiming to define its molecular functions in neurons and the mechanisms of regulation of Stau2-mediated mRNA transport and translation, we determined the molecular composition of Stau2-containing mRNPs. The absence of ribosomal proteins in the proteomics suggests that Stau2-containing particles have been characterized. They mainly contain components of the microtubules and RNA-binding proteins. While PABPC1, YB1 and hsc70 are clearly associated with Stau2-mRNPs, the presence of hnRNP H1 in these complexes is still unclear. It is possible that hnRNP H1 only transiently interacts with Stau2 in the nucleus and/or that it specifically associates with Stau2^56^ or Stau2^52^ isoform(s). Alternatively, hnRNP H1 may be a false positive hit in the proteomics. However, hnRNP H1 has previously been identified in polysome-free poly(A)-bound mRNA complexes [@pone.0011350-Angenstein1] and in embryonic RNA granules [@pone.0011350-Elvira2] suggesting that it is a component of at least some mRNPs. Post-transcriptional regulation of gene expression relies on a highly heterogeneous population of mRNP particles that ensure proper mRNA processing during splicing, nuclear trafficking, cytoplasmic localization, translation and/or decay [@pone.0011350-Keene1], [@pone.0011350-Keene2]. The differential presence/absence of RNA-binding proteins and other cofactors in each mRNP determines the roles of each mRNP complex in cellular functions and the fate of associated mRNAs. They may also form organized domains such as nuclear speckles, P-bodies or stress granules that are visible under the microscope when using specific markers [@pone.0011350-Buchan1], [@pone.0011350-Parker1], [@pone.0011350-Misteli1]. In dendrites of neurons, large ribonucleoprotein complexes shown to be associated with membranes and/or ribosomes can also be observed [@pone.0011350-Kiebler1], . Our results at the biochemical and cellular levels are consistent with the possibility that we have isolated and characterized Stau2-containing ribosome-free mRNPs. A minor population of soluble, ribosome-free Stau2-containing complexes was already described in neurons [@pone.0011350-Mallardo1], [@pone.0011350-Duchaine1]. This population contained all differentially spliced Stau2 isoforms and was enriched with RNAs [@pone.0011350-Mallardo1]. Accordingly, in a parallel approach, we also isolated mRNAs from the immunoprecipitated Stau2-containing mRNPs indicating that the proteomically identified proteins are components of complexes that also contain mRNAs [@pone.0011350-1]. The role of Stau2 and of its protein partners on the fate of associated mRNAs is still unclear. The presence in the proteomics of a nuclear protein (hnRNP H1), a protein and mRNA chaperone also involved in nuclear import/export (hsc70) and proteins that regulate translation initiation (PABPC1 and YB1) suggests that the isolated Stau2-containning mRNPs may be those involved in mRNP formation in the nucleus and/or in post-transcriptional regulation of bound mRNAs (see below). Stau2 was shown to be involved in mRNA transport in cellular processes [@pone.0011350-Mallardo1], [@pone.0011350-Tang1], [@pone.0011350-Jeong1], [@pone.0011350-Kim1], and based on known functions of its paralog Stau1, it might also be implicated in the control of mRNA stability or translation [@pone.0011350-Kim2], [@pone.0011350-DugreBrisson1]. Interestingly all Stau2-associated proteins except RUFY3 were previously described in different mRNPs. Indeed, α- and β-tubulin, YB1, PABPC1, hsc70 and hnRNP H1, as well as Stau2, were all present in the heterogeneous populations of mRNA granules isolated from embryonic rat brains [@pone.0011350-Elvira2]. Different combinations of the proteins were also described in other mRNPs [@pone.0011350-Angenstein1], [@pone.0011350-Jonson1], [@pone.0011350-Bannai1]. However, they are not universal components of all mRNPs since they were not identified in RNA granules isolated from post-natal rat brains [@pone.0011350-Kanai1] and only PABPC1 and tubulins were found in Stau1-containing mRNPs [@pone.0011350-Brendel1], [@pone.0011350-Villace1] suggesting that they play specialized roles for the transport and translation of specific mRNAs. In contrast, RUFY3 is specific to Stau2-containing mRNPs. This poorly characterized protein is expressed in brain and peaks around post-natal day 4. It accumulates in growth cones of minor processes and axons. Down-regulation of RUFY3 expression by RNAi leads to an increase in the population of neurons bearing surplus axons [@pone.0011350-Mori1]. Its molecular function and its role within RNPs are completely unknown. One of the fundamental questions in the field is how mRNA translation is repressed during transport and reactivated in response to cell needs. It is believed that mRNA transport particles are translationally repressed at the level of initiation whereas ribosome-associated granules are kept silent during elongation [@pone.0011350-Sossin1]. Accordingly, our proteomic results on Stau2-containing RNPs identified YB1 and PABPC1, two proteins known to modulate translation through interaction with initiation factors, as prominent candidates to fulfill translational regulation. YB1 is known to play key roles in cap-dependent mRNA stabilization [@pone.0011350-Evdokimova1] and translation [@pone.0011350-Nekrasov1], [@pone.0011350-Evdokimova2]. It is viewed as a general translational repressor that maintains mRNPs in a translationally silent state by its ability to bind the 5′cap structure thus displacing the initiation factors eIF4E and eIF4G from these mRNAs [@pone.0011350-Evdokimova1], [@pone.0011350-Nekrasov1]. Interestingly, YB1 can be phosphorylated by the Akt kinase [@pone.0011350-Evdokimova3], [@pone.0011350-Sutherland1], the Rsk1/2 kinase and PKC alpha [@pone.0011350-Stratford1]. Its phosphorylation by the Akt kinase specifically diminishes its interaction with the capped 5′ mRNA terminus reducing its ability to inhibit cap-dependent translation. Therefore, upon activation of a signalling pathway, phosphorylation of YB1 would be an efficient mechanism to modulate local translation of Stau2-bound mRNAs in neurons. Similarly, PABPC1 is involved in both activation and repression of translation at the level of initiation. By binding simultaneously to the poly(A) tail of mRNAs and to the initiation factor eIF4G, PABPC1 facilitates the formation of the closed loop structure that facilitates translation. PABPC1 is also involved in translation inhibition by repressors [@pone.0011350-Mazumder1]. Indeed, poly(A) tail, PABPC1 and eIF4G are all required to allow the translational inhibition of the ceruloplasmin transcript by the IFN-gamma-activated inhibitor of translation (GAIT) repressor. Similarly, PABPC1 is involved in the translational repression of its own transcript by binding an adenine-rich autoregulatory sequence (ARS) in the 5′-untranslated region [@pone.0011350-Patel1]. It also binds with lower affinities to a non-poly(A) cis-acting dendritic localizer sequence in the vasopressin mRNA [@pone.0011350-Patel1], [@pone.0011350-Mangus1], [@pone.0011350-Mohr1]. In addition to their roles in translation, both YB1 [@pone.0011350-Chernov1] and PABPC1 [@pone.0011350-Chernov2] strongly bind tubulins. The association between YB1 and tubulin is believed to interfere with mRNA binding thus reducing the YB1/mRNA ratio and facilitating translation. Considering that Stau2 mRNPs are transported on microtubules in dendrites and that both α- and β-tubulin were found in the proteomics, the presence of YB1 and PABPC1 in Stau2-containing mRNPs may contribute to the control of mRNA translation activation by regulating mRNA accessibility. Another proteomically identified protein is hsc70. Hsc70 has a chaperone activity driven by cycles of ATP/ADP bound states [@pone.0011350-Mayer1]. It could be an important factor for Stau2 folding, thus enabling it to bind mRNAs and/or other protein partners. It may also more directly influence mRNA metabolism and/or translation through its ability to modulate the folding of mRNA [@pone.0011350-Henics1] and to stabilize mRNA via its binding to AU-rich sequences [@pone.0011350-Matsui1]. Interestingly, its mRNA binding activity is inhibited by the binding of ATP molecules that compete for the same protein domain [@pone.0011350-Henics1]. We show in this paper that ATP binding also inhibits its interaction with Stau2 ([Fig. 4](#pone-0011350-g004){ref-type="fig"}) suggesting that hsc70 through ATP binding may regulate the release of mRNAs and/or control the dissociation of Stau2-containing RNPs. Releasing and/or destabilising mRNAs at precise moments during synaptic activity can finely tune protein expression at synapses. Hsc70 can play two additional important roles in mRNP transport. First, hsc70 is known to regulate the nuclear export/import trafficking of several karyopherin family members [@pone.0011350-Kose1], [@pone.0011350-Shi1]. Therefore, it could be an important factor for the nuclear shuttling of Stau2 and/or nuclear exit of Stau2-containing mRNPs. However, its role in the trafficking of exportin-5 and CRM1 which are known to export Stau2 from the nucleus [@pone.0011350-Macchi1], [@pone.0011350-Miki1] has not yet been tested. Finally, hsc70 is important for the release of the molecular motor kinesin from its vesicular cargo permitting its precise localization [@pone.0011350-Tsai1]. Stau2 mRNPs are transported on dendritic microtubules [@pone.0011350-Duchaine1] and co-fractionate with tubulin and kinesin [@pone.0011350-Mallardo1] making hsc70 an important candidate for the regulation of their transport. Altogether, these functions of hsc70, especially those related to protein and mRNA folding and to nuclear export/import, suggested that hsc70 is involved in mRNP formation and/or other early steps in the process of mRNA transport. This may explain why the presence of hsc70 in large Stau2-containing mRNA transport complexes in dendrites may not be required ([Fig. 5](#pone-0011350-g005){ref-type="fig"}). In conclusion, our data suggest that we have characterized a heterogeneous population of Stau2-containing mRNPs from embryonic rat brain. Our study indicates that they are largely composed of proteins that are also components of other types of mRNPs. The biochemical characterization of Stau2-containing particles gives us new tools that will contribute to our understanding of mRNA transport in neurons and of its role in neurons. Materials and Methods {#s4} ===================== Ethics statement {#s4a} ---------------- Pregnant Sprague-Dawley rats were purchased from Charles Rivers Canada. Our research involving animals has been conducted according to the guidelines of the Canadian Council of Animal Care (CCAC). Our project has been approved by the "Comité de déontologie de l′expérimentation sur les animaux" at the Université de Montréal. Immunoprecipitation and immunoblotting {#s4b} -------------------------------------- Rabbit polyclonal anti-Stau2 [@pone.0011350-Duchaine1] and a mouse monoclonal [@pone.0011350-DugreBrisson1] anti-HA antibodies were used for immunoprecipitation. For immunoblotting, mouse monoclonal anti-Stau2 [@pone.0011350-Duchaine1], rabbit polyclonal anti-HA (Sigma), goat anti-myc (Bethyl), goat anti-GST (Amersham Pharmacia Biotech), goat anti-GFP (Rockland), and rabbit anti-MBP (New England BioLabs, Inc) were used. For immunofluorescence in hippocampal neurons, rabbit polyclonal anti-Stau2 (a generous gift from Dr Michael Kiebler), anti-PABPC1 (Abcam) and anti-YB1 (Abcam) were used. Immunoprecipitation of Stau2-containing mRNPs was performed on cell extracts prepared from whole brains of E17--E18 rat embryos. Cells were dissociated with a manual putter and lysed in 50 mM Tris-HCl (pH 7.5), 100 mM NaCl, 0.5% Triton X-100, 15 mM EGTA, 1 mM DTT and complete EDTA-free protease inhibitor cocktail (Roche). Cell lysates were centrifuged at 9300 g for 10 min to remove nuclei and cell debris. After centrifugation, supernatants were incubated with a rabbit polyclonal anti-Stau2 antibody [@pone.0011350-Duchaine1] for 2 hours at 4°C, then with a 50% protein A-sepharose slurry for 2 hours at 4°C. Immune complexes were washed five times with the lysis buffer and eluted from the resin by heating at 95°C for 5 minutes in elution buffer (100 mM Tris-HCl pH 7.4, 200 mM DTT and 4% SDS). RNAs were isolated by Trizol (Invitrogen) extraction. For the co-immunoprecipitation experiments, N2A cells were propagated in DMEM medium supplemented with 10% BSA serum (HyClone). Cells were co-transfected with plasmids coding for one of the following proteins: Stau2^59^-HA~3~ or Stau2^62^-HA~3~ [@pone.0011350-Elvira1] and either hsc70-CFP, YB1-CFP, hnRNP H1-myc or PABPC1-myc using the calcium-phosphate technique. Cells were collected 48 h post-transfection. Immunopurifications were performed as above using a mouse monoclonal anti-HA antibody. Detection of the co-immunoprecipitated proteins was done by western blotting with either rabbit polyclonal anti-HA (Sigma), goat anti-Myc (Bethyl) or goat anti-GFP (Rockland) antibodies. To asses if the interaction was RNA dependent, cell extracts were incubated for 30 min at room temperature with 300 U (300 U/ µl) of the Micrococcal Nuclease (Fermentas) in 1 mM CaCl2 before adding the antibody. Proteomic techniques {#s4c} -------------------- Proteomic techniques were essentially done as before [@pone.0011350-Elvira2]. Briefly, immunoprecipitation eluates were separated by SDS-PAGE, stained with Coomasie Blue, and cut into 26 horizontal gel slices with each slice processed for in-gel trypsin digestion and peptide extraction. The extracted peptide mixtures were separated and analyzed in an automated system by nanoscale LC Q-TOF MS/MS. After fragmentation in the MS/MS mode, the resulting spectra were searched with Mascot (version 1.9.03; Matrix Science, London, UK) against a copy of the National Center for Biotechnology Information (NCBI) non-redundant protein database (June 21^st^, 2004) restricted to the *Mammalia* taxonomy. Protein co-localisation in hippocampal neurons {#s4d} ---------------------------------------------- Primary hippocampal neurons were cultured on \#1,5 coverslips as previously described [@pone.0011350-Elvira1], [@pone.0011350-Banker1]. On day 5, neurons were transfected with 2 µg of Lipofectamine™ 2000 and 1 µg of plasmids coding for Stau2^62^-HA~3~, PABPC1-myc, hsc70-CFP, YB1-CFP, or hnRNP H1-myc as indicated in 100 µl of plain Neurobasal for 10 min. Neurons were fixed 24 hours later with PBS/PFA 4%, PFA was quenched with 1 M glycine in PBS/0.1% Triton X-100 for 10 min, blocked with 0.1% Triton X-100/2% BSA in PBS overnight at 4°C. Neurons were incubated with goat anti-myc (1∶400 Bethyl A190--104A), mouse anti-GFP (1∶500, Roche) or anti-HA (1∶3000, 12CA5) and rabbit anti-Stau2 (1∶600, a generous gift from Dr Michael Kiebler), rabbit anti-PABP or anti-YB-1 (1∶200 and 1∶250, Abcam ab21060 and ab12148, respectively) antibodies for 2 h at room temperature, washed in PBS and stained with Alexa Fluor®647 dyed donkey anti-rabbit immunoglobulin G (IgG) and Alexa Fluor®488 dyed donkey anti-mouse or anti-goat IgG antibodies (1∶400, Invitrogen A31573, A21202 and A11055, respectively) for 1 h. Coverslips were mounted on slides (Fisher) using Dako fluorescent mounting medium (Dako). Neurons were visualized under a Nikon E800 widefield microscope Plan Apo 100× N.A. 1.40 oil-immersion objective lens. Protein expression and purification {#s4e} ----------------------------------- Plasmids coding for GST-hsc70 and MBP-Stau2 were cloned into the pGEX (Amersham Pharmacia Biotech) and pMal-C (New England Biolabs) vectors, respectively. The fusion proteins were expressed in *E. coli* BL21 cells, following induction with IPTG for 2 h. Proteins were collected in PBS/1 mM DTT/1% Triton X-100/5 mM benzamidine at 4°C. GST-hsc70 and MBP-Stau2 proteins were purified on glutathione-Sepharose-4B and amylose affinity columns, respectively and eluted with 10 mM reduced glutathione in 50 mM tris-HCl (pH 8.0) or 10 mM of D-maltose in PBS, respectively. For some experiments, lysates were treated with 50 µg/ml RNAse A for 1 h at 4°C before column purification. Protein purification was monitored by SDS-PAGE. Proteins were detected by zinc staining and/or western blotting experiments. GST-pull down assay {#s4f} ------------------- Bacterially expressed and purified GST and GST-hsc70 proteins were attached to glutathione columns. MBP-Stau2 or MBP were loaded onto the columns, extensively washed and eluted as recommended by the manufacturer. Eluted proteins were analyzed by western blotting using anti-GST and anti-MBP antibodies. For some experiments, protein extracts were treated with 50 µg/ml RNAse A for 1 h at 4°C before loading onto the columns. Surface plasmon resonance (SPR) binding assay {#s4g} --------------------------------------------- Binding interactions between purified MBP-Stau2 and GST-hsc70 were examined in real time using a BIACORE 2000 instrument (GE Healthcare Bio-Sciences AB, Upssala, Sweden). Experiments were performed on research-grade CM5 sensor chip at 25°C using filtered (0.2 µm) and degassed HBS-EP \[10 mM Hepes pH 7.4, 150 mM NaCl, 3 mM EDTA, 0.005% (v/v) Surfactant P20\]. Protein-grade detergents \[10% (v/v) Tween-20, 10% (v/v) DDM\] were from Calbiochem; all other chemicals were reagent grade quality. Immobilized sensor chip surfaces were prepared using the Biacore amine Coupling Kit. Briefly, 35 µl of a freshly mixed solution of 200 mM 1-ethyl-3-(3-dimethylaminopropyl)-carbodiimide and 50 mM N-hydroxysuccinimide was injected (at a flow rate of 5 µl min^−1^) to activate surface-exposed carboxymethyl groups into reactive esters. Next, 120 µl of MBP-Stau2 diluted to 70 µg ml^−1^ in 10 mM sodium acetate pH 4.0 was injected (at a flow rate of 10 µl min^−1^) to generate amine-coupled protein surfaces. Finally, 70 µl of 1 M ethanolamine pH 8.5 was injected to deactivate excess reactive groups and remove any non-specifically bound ligand. A reference surface was prepared in a similar manner with purified MBP. To test binding, purified GST-hsc70 (60 µg ml-1) or GST (negative control) were injected over the coupled surfaces at 10 µl min^−1^ (3 min association time and 2.5 min dissociation time). For all SPR assays, the surfaces were regenerated between sample injections at 50 µl min^−1^ with a 30 s single pulse of 0.4 M NaCl followed by a stabilization time after regeneration of 3 min. The assays were performed in duplicates with different batches of purified proteins. Supporting Information {#s5} ====================== ###### Proteomically identified proteins in Stau2-containning mRNPs. (0.05 MB DOC) ###### Click here for additional data file. We thank Louise Cournoyer for her help with tissue cultures and Génome Québec, Réseau Protéomique de Montréal, Montreal Proteomics Network (RPMPN) for mass spectrometry and analysis. **Competing Interests:**The authors have declared that no competing interests exist. **Funding:**This work was supported by operating \[MOP 62751\] and group \[MGC-57079\] grants from the Canadian Institutes of Health Research (CIHR-[www.cihr-irsc.gc.ca/](http://www.cihr-irsc.gc.ca/)) to LDG. MML and LAJ were supported by a postgraduate scholarship and an undergraduate student research award, respectively, from the Natural Sciences and Engineering Research Council of Canada (NSERC-[www.nserc-crsng.gc.ca/](http://www.nserc-crsng.gc.ca/)). MM held a Canada Research Chair on Bacterial Animal Diseases. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. [^1]: Conceived and designed the experiments: MML GL MM LD. Performed the experiments: MML FB MM LAJ GL MM. Analyzed the data: MML FB MM LAJ GL MM LD. Contributed reagents/materials/analysis tools: MM LD. Wrote the paper: MML LD.

Related literature {#sec1} ================== For general backround to boronic acids, see: Hall (2005[@bb8]); Höpfl (2002[@bb9]); Fujita *et al.* (2008[@bb7]); Soloway *et al.* (1998[@bb15]). For hydrogen-bond motifs, see: Bernstein *et al.* (1995[@bb1]); Desiraju (2002[@bb5]). For related structures, see: Wu *et al.* (2006[@bb19]); Bradley *et al.* (1996[@bb2]); Horton *et al.* (2004[@bb10]). For crystal engineering, see: Fournier *et al.* (2003[@bb6]); Rodríguez-Cuamatzi *et al.* (2004[@bb12], 2005[@bb11]). Experimental {#sec2} ============ {#sec2.1} ### Crystal data {#sec2.1.1} C~6~H~5~BF~2~O~2~*M* *~r~* = 157.91Monoclinic,*a* = 3.7617 (11) Å*b* = 12.347 (4) Å*c* = 14.620 (4) Åβ = 95.450 (5)°*V* = 676.0 (3) Å^3^*Z* = 4Mo *K*α radiationμ = 0.15 mm^−1^*T* = 293 (2) K0.37 × 0.35 × 0.22 mm ### Data collection {#sec2.1.2} Bruker SMART APEX CCD area-detector diffractometerAbsorption correction: multi-scan (*SADABS*; Sheldrick, 1996[@bb13]) *T* ~min~ = 0.947, *T* ~max~ = 0.9683196 measured reflections1190 independent reflections1012 reflections with *I* \> 2σ(*I*)*R* ~int~ = 0.028 ### Refinement {#sec2.1.3} *R*\[*F* ^2^ \> 2σ(*F* ^2^)\] = 0.056*wR*(*F* ^2^) = 0.127*S* = 1.151190 reflections106 parameters2 restraintsH atoms treated by a mixture of independent and constrained refinementΔρ~max~ = 0.14 e Å^−3^Δρ~min~ = −0.18 e Å^−3^ {#d5e471} Data collection: *SMART* (Bruker, 2000[@bb3]); cell refinement: *SAINT-Plus-NT* (Bruker, 2001[@bb4]); data reduction: *SAINT-Plus-NT*; program(s) used to solve structure: *SHELXTL-NT* (Sheldrick, 2008[@bb14]); program(s) used to refine structure: *SHELXTL-NT*; molecular graphics: *CAMERON* (Watkin *et al.*, 1996[@bb17]); software used to prepare material for publication: *PLATON* (Spek, 2003[@bb16]) and *publCIF* (Westrip, 2009[@bb18]). Supplementary Material ====================== Crystal structure: contains datablocks I, global. DOI: [10.1107/S1600536808040646/hb2865sup1.cif](http://dx.doi.org/10.1107/S1600536808040646/hb2865sup1.cif) Structure factors: contains datablocks I. DOI: [10.1107/S1600536808040646/hb2865Isup2.hkl](http://dx.doi.org/10.1107/S1600536808040646/hb2865Isup2.hkl) Additional supplementary materials: [crystallographic information](http://scripts.iucr.org/cgi-bin/sendsupfiles?hb2865&file=hb2865sup0.html&mime=text/html); [3D view](http://scripts.iucr.org/cgi-bin/sendcif?hb2865sup1&Qmime=cif); [checkCIF report](http://scripts.iucr.org/cgi-bin/paper?hb2865&checkcif=yes) Supplementary data and figures for this paper are available from the IUCr electronic archives (Reference: [HB2865](http://scripts.iucr.org/cgi-bin/sendsup?hb2865)). This work was supported by Consejo Nacional de Ciencia y Tecnología (CIAM-59213 for HH). Comment ======= Boronic acids, RB(OH)~2~ with *R* = alkyl and aryl, have applications in organic synthesis (Hall, 2005), host--guest chemistry (Höpfl, 2002), the molecular recognition of biochemically active molecules (Fujita *et al.*, 2008) and in medicine as antibiotics, inhibitors and for the treatment of tumors (Soloway *et al.*, 1998). Similar to carboxylic acids they are capable to form hydrogen-bonded dimeric units and, therefore, boronic acids have been used recently as new building blocks in crystal engineering (Fournier *et al.*, 2003; Rodríguez-Cuamatzi *et al.*, 2004; Rodríguez-Cuamatzi *et al.*, 2005). Previously, the structures of 3-fluorophenylboronic acid (Wu *et al.*, 2006), 2,6-difluoroboronic acid (Bradley *et al.*, 1996) and pentafluoroboronic acid (Horton *et al.*, 2004) had been reported. We now present the crystal structure of (I). The molecular structure is essentially planar, O1---B1---C1---C2 = 4.4 (4)°, indicating that there is a π···π interaction between the dihydroxyboryl group and the aromatic ring, to which it is attached (Fig. 1). This interaction is also evidenced by the B---C bond length of 1.566 (3) Å, which is significantly shorter than that observed in boronates containing tetra-coordinate boron atoms (Höpfl, 2002). The crystal structure is stabilized by strong O2---H2···O1 hydrogen-bonding interactions, forming *R*~2~^2^(8) motifs (Bernstein *et al.*, 1995), as well as, O1---H1···F1 and O1---H1···F2 bifurcated hydrogen bonds (Fig. 2; Table 1) (Desiraju, 2002). Due to these interactions each boronic acid homodimer is linked to two neighboring homodimeric units, thus creating a two-dimensional hydrogen-bonded network, in which fluorine is therefore an essential structural component. Experimental {#experimental} ============ 2,4-Difluorophenylboronic acid was purchased from Aldrich and crystallized from water to yield colourless blocks of (I). Refinement {#refinement} ========== The aromatic H atoms were positioned geometrically (C---H = 0.93Å) and refined as riding with *U*~iso~(H) = 1.2*U*~eq~(C). The O---H hydrogen atoms were localized in a difference map and their coordinates were refined with O---H = 0.84+/0.01Å and *U*~iso~(H) = 1.5 *U*~eq~(O). Figures ======= {#Fap1} {#Fap2} Crystal data {#tablewrapcrystaldatalong} ============ ------------------------- --------------------------------------- C~6~H~5~BF~2~O~2~ *F*(000) = 320 *M~r~* = 157.91 *D*~x~ = 1.552 Mg m^−3^ Monoclinic, *P*2~1~/*n* Melting point = 521--522 K Hall symbol: -P 2yn Mo *K*α radiation, λ = 0.71073 Å *a* = 3.7617 (11) Å Cell parameters from 1052 reflections *b* = 12.347 (4) Å θ = 2.3--26.2° *c* = 14.620 (4) Å µ = 0.15 mm^−1^ β = 95.450 (5)° *T* = 293 K *V* = 676.0 (3) Å^3^ Block, colorless *Z* = 4 0.37 × 0.35 × 0.22 mm ------------------------- --------------------------------------- Data collection {#tablewrapdatacollectionlong} =============== --------------------------------------------------------------- -------------------------------------- Bruker SMART APEX CCD area-detector diffractometer 1190 independent reflections Radiation source: fine-focus sealed tube 1012 reflections with *I* \> 2σ(*I*) graphite *R*~int~ = 0.028 Detector resolution: 8.3 pixels mm^-1^ θ~max~ = 25.0°, θ~min~ = 2.2° φ and ω scans *h* = −3→4 Absorption correction: multi-scan (*SADABS*; Sheldrick, 1996) *k* = −14→12 *T*~min~ = 0.947, *T*~max~ = 0.968 *l* = −17→17 3196 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*\[*F*^2^ \> 2σ(*F*^2^)\] = 0.056 Hydrogen site location: inferred from neighbouring sites *wR*(*F*^2^) = 0.127 H atoms treated by a mixture of independent and constrained refinement *S* = 1.15 *w* = 1/\[σ^2^(*F*~o~^2^) + (0.0442*P*)^2^ + 0.2673*P*\] where *P* = (*F*~o~^2^ + 2*F*~c~^2^)/3 1190 reflections (Δ/σ)~max~ \< 0.001 106 parameters Δρ~max~ = 0.14 e Å^−3^ 2 restraints Δρ~min~ = −0.18 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~ B1 0.7704 (8) 0.4548 (2) 0.62567 (18) 0.0452 (7) O1 0.6825 (6) 0.38623 (15) 0.55419 (13) 0.0695 (6) H1 0.748 (9) 0.3215 (9) 0.562 (2) 0.104\* O2 0.6880 (6) 0.55977 (14) 0.61557 (12) 0.0630 (6) H2 0.593 (8) 0.577 (3) 0.5632 (10) 0.094\* F1 1.0385 (5) 0.23728 (11) 0.67473 (11) 0.0768 (6) F2 1.4329 (5) 0.33016 (14) 0.97634 (10) 0.0822 (6) C1 0.9591 (6) 0.41789 (18) 0.72069 (15) 0.0424 (6) C2 1.0796 (7) 0.31430 (18) 0.74175 (16) 0.0470 (6) C3 1.2380 (7) 0.2819 (2) 0.82553 (17) 0.0539 (7) H3 1.3138 0.2109 0.8363 0.065\* C4 1.2785 (7) 0.3593 (2) 0.89223 (17) 0.0547 (7) C5 1.1696 (8) 0.4640 (2) 0.87828 (17) 0.0586 (7) H5 1.2013 0.5150 0.9251 0.070\* C6 1.0119 (7) 0.49169 (19) 0.79282 (16) 0.0498 (6) H6 0.9371 0.5628 0.7827 0.060\* ---- ------------ -------------- -------------- -------------------- -- Atomic displacement parameters (Å^2^) {#tablewrapadps} ===================================== ---- ------------- ------------- ------------- -------------- -------------- -------------- *U*^11^ *U*^22^ *U*^33^ *U*^12^ *U*^13^ *U*^23^ B1 0.0451 (16) 0.0425 (15) 0.0471 (16) −0.0009 (12) −0.0005 (12) 0.0033 (12) O1 0.1000 (17) 0.0484 (11) 0.0540 (11) 0.0158 (10) −0.0241 (10) −0.0009 (9) O2 0.0850 (15) 0.0441 (10) 0.0552 (11) 0.0087 (9) −0.0172 (10) 0.0048 (8) F1 0.1192 (15) 0.0456 (9) 0.0602 (10) 0.0151 (9) −0.0187 (9) −0.0046 (7) F2 0.1070 (14) 0.0804 (12) 0.0527 (10) −0.0120 (10) −0.0262 (9) 0.0181 (8) C1 0.0383 (13) 0.0412 (13) 0.0473 (13) −0.0039 (10) 0.0019 (10) 0.0048 (10) C2 0.0523 (16) 0.0410 (13) 0.0467 (13) −0.0017 (11) 0.0001 (11) 0.0008 (10) C3 0.0584 (17) 0.0449 (14) 0.0564 (15) 0.0002 (12) −0.0053 (13) 0.0124 (12) C4 0.0579 (17) 0.0613 (17) 0.0426 (13) −0.0099 (13) −0.0075 (12) 0.0135 (12) C5 0.0696 (19) 0.0552 (16) 0.0489 (14) −0.0109 (13) −0.0058 (13) −0.0020 (12) C6 0.0559 (16) 0.0399 (13) 0.0522 (14) −0.0011 (11) −0.0011 (12) 0.0029 (11) ---- ------------- ------------- ------------- -------------- -------------- -------------- Geometric parameters (Å, °) {#tablewrapgeomlong} =========================== ------------------- ------------ ------------------- ------------ B1---O2 1.338 (3) C1---C6 1.394 (3) B1---O1 1.361 (3) C2---C3 1.370 (3) B1---C1 1.566 (3) C3---C4 1.363 (4) O1---H1 0.841 (15) C3---H3 0.93 O2---H2 0.841 (15) C4---C5 1.366 (4) F1---C2 1.364 (3) C5---C6 1.374 (3) F2---C4 1.358 (3) C5---H5 0.93 C1---C2 1.382 (3) C6---H6 0.93 O2---B1---O1 118.7 (2) C4---C3---H3 121.8 O2---B1---C1 117.4 (2) C2---C3---H3 121.8 O1---B1---C1 123.8 (2) F2---C4---C3 118.1 (2) B1---O1---H1 116 (2) F2---C4---C5 118.8 (2) B1---O2---H2 115 (2) C3---C4---C5 123.0 (2) C2---C1---C6 114.6 (2) C4---C5---C6 117.9 (2) C2---C1---B1 125.3 (2) C4---C5---H5 121.0 C6---C1---B1 120.1 (2) C6---C5---H5 121.0 F1---C2---C3 116.7 (2) C5---C6---C1 122.9 (2) F1---C2---C1 118.2 (2) C5---C6---H6 118.5 C3---C2---C1 125.1 (2) C1---C6---H6 118.5 C4---C3---C2 116.4 (2) O2---B1---C1---C2 −176.5 (2) C1---C2---C3---C4 −0.3 (4) O1---B1---C1---C2 4.5 (4) C2---C3---C4---F2 179.7 (2) O2---B1---C1---C6 4.6 (4) C2---C3---C4---C5 0.0 (4) O1---B1---C1---C6 −174.5 (2) F2---C4---C5---C6 −179.6 (2) C6---C1---C2---F1 −179.9 (2) C3---C4---C5---C6 0.1 (4) B1---C1---C2---F1 1.1 (4) C4---C5---C6---C1 0.0 (4) C6---C1---C2---C3 0.4 (4) C2---C1---C6---C5 −0.3 (4) B1---C1---C2---C3 −178.6 (2) B1---C1---C6---C5 178.8 (2) F1---C2---C3---C4 180.0 (2) ------------------- ------------ ------------------- ------------ Hydrogen-bond geometry (Å, °) {#tablewraphbondslong} ============================= ------------------ ---------- ---------- ----------- --------------- *D*---H···*A* *D*---H H···*A* *D*···*A* *D*---H···*A* O1---H1···F1 0.84 (2) 2.16 (3) 2.799 (3) 133 (2) O1---H1···F2^i^ 0.84 (2) 2.39 (2) 3.086 (3) 140 (3) O2---H2···O1^ii^ 0.84 (2) 1.97 (2) 2.809 (3) 174 (3) ------------------ ---------- ---------- ----------- --------------- Symmetry codes: (i) *x*−1/2, −*y*+1/2, *z*−1/2; (ii) −*x*+1, −*y*+1, −*z*+1. ###### Hydrogen-bond geometry (Å, °) *D*---H⋯*A* *D*---H H⋯*A* *D*⋯*A* *D*---H⋯*A* ---------------- ------------ ---------- ----------- ------------- O1---H1⋯F1 0.841 (15) 2.16 (3) 2.799 (3) 133 (2) O1---H1⋯F2^i^ 0.841 (15) 2.39 (2) 3.086 (3) 140 (3) O2---H2⋯O1^ii^ 0.841 (19) 1.97 (2) 2.809 (3) 174 (3) Symmetry codes: (i) ; (ii) .

**Eileen Appelbaum**, Center for Economic and Policy Research, Washington, DC, USA.

*Eleusine* is a small genus of annual and perennial grass species within the Eragrosteae tribe and Chloridoideae subfamily. It includes about 9 to 12 species that can hybridize to form intermediates and they are very similar in morphological features ([@bib39]; [@bib44]; [@bib2]; [@bib25]). It is mainly distributed in the tropical and subtropical parts of Africa, Asia and South America ([@bib44]). *Eleusine* contains diploid and tetraploid species, with chromosome numbers ranging from 2n = 16, 18 or 20 in diploids to 2n = 36 or 38 in tetraploids. All of the species are wild except *E. coracana*, which is cultivated for grain and fodder in Africa and the Indian subcontinent. The center of *Eleusine* diversity is East Africa and there are eight species in this genus occurring in this region, which includes *E. africana*, *E. coracana*, *E. kigeziensis*, *E. indica*, *E. floccifolia*, *E. intermedia*, *E. multiflora*, and *E. jaegeri* ([@bib38]; [@bib44]). The genome size of *Eleusine* species is very small and the 2C DNA amount ranges from 2.50 pg to 3.35 pg for diploid species ([@bib27]). Questions remain regarding the evolutionary origins of the polyploid species and their relationship to wild diploid progenitors. *E. coracana*, commonly referred to as finger millet or African finger millet, is the only domesticated *Eleusine*, which is cultivated as both grain and fodder primarily in semiarid regions of Africa and the Indian subcontinent ([@bib6]). *E. coracana* is an allotetraploid species with a chromosome number of 2n = 4x = 36 that was reportedly domesticated from the wild tetraploid *E. africana* (2n = 4x =36) ([@bib26]; [@bib16]). *E. coracana* is by all definitions an orphan crop, an important regional crop that lacks widespread use ([@bib49]). Orphan crops also have societal benefits of aiding to sustain cultural richness and maintain community identity in rural societies ([@bib41]). Global climate change will have negative effects on the yield of major crops, which will conflict with increasing world population growth ([@bib29]). In undeveloped regions of the world, continued failure to maintain increases in food production will lead to food price increases, as well as social unrest and famine ([@bib1]). Orphan crops such as finger millet could be a beneficial food source to ballooning world populations because they can be grown on more marginal land under harsher environmental conditions ([@bib41]). The major limitation to developing orphan crops is that information on germplasm is not readily accessible and little information is found outside of traditional peer-reviewed academic publishing or written in languages not well-known to the scientific community concerned ([@bib22]). In addition, existing knowledge on the genetic potential of minor crops is limited with few genetic resources, like genomes, transcriptomes and ESTs, available online compared to major or industrial crops ([@bib15]). Lack of information about origin and ancestry also inhibits breeding of minor crops. In plant breeding, paternal and maternal germplasm with desirable traits are collected and desirable traits are introduced to the cultivated species through hybridization and backcrossing ([@bib48]; [@bib13]). For example, knowing the parentage aided the development of peanuts since wild diploid *Arachis* species possess genetic variability in pest and disease resistance traits, which were used to improve cultivated peanuts ([@bib50]; [@bib12]). Assessment of phylogenetic relationships is vital for any successful crop improvement since the wild relatives often have good traits and biodiversity. With respect to the *Eleusine* genre, publicly available transcriptome assemblies have been produced for *E. indica* ([@bib11]) and *E. coracana* ([@bib45]; [@bib32]), and 78 plastid protein coding loci were sequenced for *E. coracana* ([@bib18]). A complete chloroplast genome ([@bib54]) and a draft nuclear genome ([@bib53]) have been reported for *E. indica* and a draft nuclear genome has been reported for *E. coracana* ([@bib23]; [@bib30]). [@bib23] used a novel multiple hybrid assembly workflow which is suitable for the assembly of complex allotetraploid species. Although there are more studies conducted for genomic resources of *E. coracana*, there is still only modest information on its evolution and progenitors. *E. indica*, an annual diploid (2n = 2x = 18), is most commonly mentioned as the maternal genome donor based on genomic *in situ* hybridization ([@bib24]; [@bib28]; [@bib5]) although *E. tristachya*, a diploid (2n = 2x = 18) has not been eliminated as the maternal progenitor while *E. floccifolia*, a diploid (2n = 2x = 18) perennial species or an unknown or extinct ancestor is thought to be the paternal genome donor ([@bib7], [@bib5] 2002; [@bib35]). However, for these studies, the evidence was not enough since they only used one or few chloroplast genes or a single low copy nuclear gene as a marker. Thus, our objective was to provide a broader survey of *Eleusine* species evolutionary relationships based on separate analysis of chloroplast and nuclear transcriptomes and to verify the maternal genome donor of *E. coracana*. Materials and Methods {#s1} ===================== Germplasm was acquired from the U.S. National Plant Germplasm System (<https://npgsweb.ars-grin.gov/gringlobal/search.aspx>) Germplasm Resources Information Network (NPGS GRIN) for analysis. An exhaustive search for all available *Eleusine* species was conducted to identify all possible candidate species within the *Eleusine* genus. Seven of the nine known *Eleusine* species were identified and acquired for analysis ([Table 1](#t1){ref-type="table"}). *E. jaegeri* and *E. kigeziensis* were unavailable from NPGS GRIN. No other sources for these two species could be identified. A previously assembled transcriptome ([@bib11]) and plastid genome ([@bib54]) of *E. indica* were utilized as references. ###### Biological, genomic, and GRIN[*^a^*](#t1n1){ref-type="table-fn"} Accession Number for seven *Eleusine* species utilized. Genomic and biological acquired from the following sources Species 2n chromosome numbers, genome, ploidy Life cycle Type GRIN Accession Number ------------------ --------------------------------------- --------------------- ------------ -------------------------------------------- *E. multiflora* 16, CC, diploid Annual Wild 226067 *E. floccifolia* 18, BB or other, diploid Perennial Wild 196853 *E. tristachya* 18 AA, diploid Annual Wild 331791 *E. intermedia* 18 AB, diploid Perennial Wild 273888 *E. africana* 36 AABB, allotetraploid Annual Wild 226270 *E. coracana* 36 AABB, allotetraploid Annual Cultivated 462949 *E. indica* 18 AA, diploid Annual or Perennial Wild Collect[*^b^*](#t1n2){ref-type="table-fn"} *E. jaegeri* 20 DD, diploid Perennial Wild Unavailable *E. kigeziensis* 38 AADD, allotetraploid Perennial Wild Unavailable GRIN, Germplasm Resources Information Network. *E. indica* was collected locally from a crop field in Tallassee, Alabama. In other published work by J.S. McElroy it is referred to by the acronym PBU referring to its origin at the Alabama Agricultural Experiment Station Plant Breeding Unit. *E. indica* is known to exist as a weedy perennial in managed ecosystems of southern Florida and Hawai'i. *Eleusine* species were germinated and grown from seed in a glasshouse environment at 28 ± 2°, and 70% average relative humidity in Auburn, AL (32.35°N, 85.29°W). Seedlings were grown in a native Wickham sandy loam soil with pH 6.3 and 0.5% organic matter. Four-week old entire seedlings were used for RNA extraction. Total RNA was extracted from individual seedlings of *E. multiflora*, *E. floccifolia*, *E. tristachya*, *E. intermedia*, *E. africana*, and *E. coracana* using RNeasy Plant Mini Kit (Qiagen, CA, USA). The quality and quantity of total RNA were determined with gel electrophoresis and Nanodrop 2000 (Thermo Scientific). High-quality RNA was used for transcriptome sequencing. RNA preparation and sequencing was conducted at the Genomic Service Laboratory at Hudson Alpha Institute for Biotechnology (Cummings Research Park, Huntsville, AL) using standard procedures for the Illumina HiSeq 2000 to produce 100 bp paired-end reads ([@bib11], [@bib10]). One complementary DNA (cDNA) library was constructed for each of the six total RNA samples. All samples were subjected to polyA selection prior to sequencing. *E. indica* transcriptome (NCBI Accession No.: SRR1560465) previously assembled by our lab ([@bib11]) was also sequenced by Hudson Alpha using the Illumina HiSeq 2000 platform and same methodology in the same growth conditions. Sequence data analysis and assembly {#s2} ----------------------------------- Raw reads quality were checked by FastQC v.0.11.1 software ([@bib3]) and then processed by Trimmomatic v.0.33 ([@bib8]) to remove adapters and low quality reads and sequences. The trimmed reads were evaluated with FastQC again and normalized with Trinity's in silico read normalization ([@bib21]), with maximum coverage of 30. Three *de novo* transcriptome assemblers were used: Trinity v.2014-04-13p1 ([@bib21]), Velvet v.1.2.08\_ maxkmer101 ([@bib52]), and SOAPdenovo2 v.2.04 ([@bib37]). Trinity k-mer size was 25. Velvet k-mer size was 21 to 91 with step size of 10 and minimum contig length was 200 bp without scaffolding. SOAPdenovo2 k-mer size was 21 and 31. The three *de novo* assemblers thus yielded 11 total assemblies for each species. The script Select_contigs.pl (<https://pods.iplantcollaborative.org/wiki/display/DEapps/Select+contigs>) was used for Trinity and SOAPdenovo2 to select contigs with minimum length 200 bp. To evaluate the quality of the assembly, N50s and contig length distributions of the assemblies were calculated with the script Count_fasta.pl (<http://wiki.bioinformatics.ucdavis.edu/index.php/Count_fasta.pl>). Before merging, "N"s were removed from the assemblies and contigs shorter than 200 bp were discarded. All assemblies were combined into one merged assembly for each species individually. The merged assembly was processed by EvidentialGene tr2aacds pipeline (<http://arthropods.eugenes.org/EvidentialGene/about/EvidentialGene_trassembly_pipe.html>). The EvidentialGene pipeline takes as input the transcript fasta file produced by any of the transcript assemblers and generates coding DNA sequences (CDSs) and amino acid sequences from each input contig then uses fastanrdb to quickly reduce perfect duplicate sequences, cd-hit and cd-hit-est to cluster protein and nucleotide sequences, and Blastn and makeblastdb to find regions of local similarity between sequences. It outputs transcripts into three classes: Okay (the best transcripts with the unique CDS, which is close to a biologically real set regardless of how many millions of input assemblies), Alternate (possible isoforms), and Drop (the transcripts did not pass the internal filter). The unique CDS (Okay set) and possible isoforms (Alternate set) were used for further evaluation and annotation. The overall workflow was summarized graphically in [Figure 1](#fig1){ref-type="fig"}. {#fig1} Annotation and analysis {#s3} ----------------------- Sequences were annotated using Trinotate v.2.02, which is a comprehensive annotation suite designed for automatic functional annotation of transcriptomes, particularly *de novo* assembled transcriptomes ([@bib34]). This pipeline includes: homology search to known sequence data (BLAST+/SwissProt), protein domain identification (HMMER/PFAM), protein signal peptide and transmembrane domain prediction (signalP/tmHMM), and leveraging various annotation databases (eggNOG/GO/Kegg databases). All functional annotation data derived from the analysis of transcripts are integrated into an SQLite database which allows fast efficient searching for terms with specific qualities related to a desired scientific hypothesis or a means to create a whole annotation report for a transcriptome. Blast2GO v.3.0 ([@bib19]) was used to analyze the unique genes between *E. coracana* and *E. africana*. Variants analysis {#s4} ----------------- Variants are mainly classified into five different types: single nucleotide variants (SNVs), multiple nucleotide variants (MNVs), insertions, deletions, and replacements. SNVs are one base replaced by another base, most commonly referred to as a single nucleotide polymorphism (SNP). MNVs are two or more SNVs in succession. Insertions are events where one or more bases are inserted in the experimental data compared to the reference. Deletions are events where one or more bases are deleted from the experimental data compared to the reference. Replacements are more complex events where one or more bases have been replaced by one or more bases, where the identified allele has a length different from the reference. Read mapping and detection of SNVs, MNVs, replacements, insertions, and deletions were conducted using the tools 'map reads to reference' and 'probabilistic variant detection' separately in CLC Genomics Workbench v.6.5.2 (CLC Bio, Aarhus, Denmark). The mapping parameters were set to 'Mismatch cost = 3, Insertion cost = 3, Deletion cost = 3, Length fraction = 0.95, Similarity fraction = 0.95'. The variants calling parameters were set to 'Minimum coverage = 30, Variant probability = 90'. Chloroplast gene comparison {#s5} --------------------------- Complete *E. indica* chloroplast genome (KU833246) were downloaded from NCBI. The other *Eleusine* species' CDS datasets were aligned to the chloroplast genome using Blastn at the E-value threshold 10^−5^, word size 20, and minimum match size 90. *E. coracana* reads were mapped to the aligned *Eleusine* species' CDSs separately. SNVs, MNVs, replacements, insertions, and deletions were called from each of the mappings in CLC Genomics Workbench v.6.5.2 (CLC Bio, Aarhus, Denmark). Phylogenetic analysis {#s6} --------------------- Two separate analyses were conducted to determine the potential parentage of *E. coracana*. First, chloroplast genome was compared among all *Eleusine* species, and second, transcriptomes of nuclear genes were compared among *Eleusine* species. Chloroplast genes of *E. indica* were downloaded from NCBI (KU833246), which was named *E. indica_cp* in phylogenetic tree. Chloroplast genes from *E. indica* transcriptome using blast method were obtained and named *E. indica_trans* in phylogenetic tree and we used this method to verify our result. TBLASTx was used to extract the best chloroplast genes from each *Eleusine* species separately. The results were checked with alignment viewer Seaview v.4 ([@bib20]) and adjusted to exclude any erroneous hits. A supermatrix of nucleotide sequence alignments was produced using FASconCAT-G_v1.02.pl ([@bib31]). Several steps were employed to extract the nuclear genes for phylogenetic analyses. The contigs were translated to coding protein sequences using Transdecoder v.3.0.1 ([@bib47]). The Python script reduce_protein_redundancy.py (<https://github.com/mcelrjo/blastp_nr>) was used to select the longest ORF to produce a set of unique sequences. Orthogroups were extracted and aligned from the set of unique sequences with Orthofinder v.1.1.8 ([@bib17]). A concatenated supermatix was produced using FASconCAT-G_v.1.02.pl ([@bib31]). A codon by gene partition scheme was used in Partition-Finder v.2.0.0 ([@bib33]) and model selection was limited to GTR-GAMMA and GTR-GAMMA+I with greedy search algorithm, and the best scheme was used for subsequent phylogenetic analysis. Individual nuclear gene alignments were reduced to include only representatives of Poaceae and cleaned with gBlocks v0.19b ([@bib9]) using default settings. Both concatenated and individual nuclear gene trees were created using RAxML-MPI-AVX v.8.2.6 ([@bib51]) with 100 rapid bootstraps, and GTRGAMMA model since RAxML employs only one model across all partitions per analysis. Trees were visualized with Figtree v.1.3.1 ([@bib46]). Comparative transcriptome analysis Between E. africana and E. coracana {#s7} ---------------------------------------------------------------------- Comparative transcriptome analyses were conducted with the following steps: 1) A list of unique protein-coding transcripts from the *E. coracana* transcriptome were compiled and queried against *E. africana* transcriptome; 2) For *E. coracana* contigs with no matches to the *E. africana* transcriptome assembly but with matches to the non-redundant database, the sequences of the top hits were retrieved from the non-redundant database and used to query the *E. africana* transcriptome assembly; 3) Those *E. coracana* transcripts that remained unidentified were identified as genes that were expressed in the *E. coracana*, but not expressed in the *E. africana*. Data availability {#s8} ----------------- The sequencing reads of *E. multiflora*, *E. floccifolia*, *E. tristachya*, *E. intermedia*, *E. africana*, and *E. coracana* were deposited at NCBI Sequence Read Archive (SRA) database under the accessions SRR5467257, SRR5468569, SRR5468570, SRR5468571, SRR5468572, SRR5468573, respectively. Transcriptome Shotgun Assembly projects have been deposited at DDBJ/EMBL/GenBank under the accessions GGLR00000000, GGME00000000, GGMD00000000, GGMC00000000, GGMB00000000, and GGMA00000000, correspondingly. All of the versions described in this paper are the first version, GGLR01000000, GGME01000000, GGMD01000000, GGMC01000000, GGMB01000000, and GGMA01000000. Supplemental material available at FigShare: <https://doi.org/10.25387/g3.7994039>. Results and Discussion {#s9} ====================== Transcriptome sequencing and de novo assemblies {#s10} ----------------------------------------------- Read counts before and after quality checking and trimming are presented in [Table 2](#t2){ref-type="table"}. The summary statistics of the assemblies from EvidentialGene tr2aacds pipeline are shown in [Table 3](#t3){ref-type="table"}. Previous research has demonstrated this pipeline to improve transcript integrity and reduce assembly redundancy in transcriptome assembly ([@bib11]). Average read length after trimming was 99.3 to 99.4 nucleotides. The N50 of the unique CDS set ranged from 1,471 to 1,693; however, when the possible isoform set is added, the N50 ranged from 1,232 to 1,451. ###### The number and average length of *Eleusine* transcriptome sequencing reads before and after trimming Species Number of reads Average length Number of reads after trim \% reads removed Average length after trim ------------------ ----------------- ---------------- ---------------------------- ------------------ --------------------------- *E. multiflora* 61,348,758 100 52,236,532 15% 99.4 *E. floccifolia* 59,140,884 100 50,053,954 15% 99.4 *E. tristachya* 53,661,434 100 45,004,810 16% 99.4 *E. intermedia* 106,867,304 100 84,798,308 21% 99.4 *E. africana* 197,003,984 100 156,392,016 21% 99.3 *E. coracana* 139,928,698 100 111,917,028 20% 99.3 *E. indica* 230,466,942 100 183,323,866 17% 99.4 ###### Summary statistics of transcriptome assemblies following implementation of N50, sequences number, and total length in EvidentialGene tr2aacids pipeline Species Unique CDSs Unique CDSs + Possible isoforms ------------------ ------------- --------------------------------- ------------ ------ --------- ------------- *E. multiflora* 1567 30,394 32,083,609 1357 52,610 50,466,628 *E. floccifolia* 1585 36,364 37,932,847 1361 72,602 69,442,718 *E. tristachya* 1549 35,856 37,243,265 1353 72,764 69,722,866 *E. intermedia* 1693 39,540 43,739,409 1451 87,270 87,954,199 *E. africana* 1516 56,375 54,910,276 1236 144,921 129,354,728 *E. coracana* 1471 59,223 561,062,47 1232 144,460 128,133,958 *E. indica* 1562 25,878 28,239,951 1408 36,959 37,055,659 For annotation, unique CDS assemblies of each transcriptome set were initially assigned with Trinotate v.2.02. GoTermParse.py ([<https://gist.github.com/NDHall/>]{.ul}) was used to retrieve GO Terms and three components (Table S1). GoTermParse.py used regular expressions and a dictionary to sort terms into their major functional groups. The GO classification assigned totals of 516,793; 634,349; 578,631; 803,545; 996,369; 1,039,581; and 276,976 GO terms to *E. multiflora*, *E. floccifolia*, *E. tristachya*, *E. intermedia*, *E. africana*, *E. coracana*, and *E. indica* unique CDS set, respectively. All of the GO terms in *E. coracana* 'unique CDS' set have higher scores than in others. Integral_component_of_membrane, transcription_DNA-templated and ATP_binding are the highest GO terms in each corresponding component (Figure S1). E. coracana maternal genome donor {#s11} --------------------------------- In order to elucidate the maternal genome donor of *E. coracana*, *E. coracana* reads were mapped to the assembled and identified chloroplast genes of *E. multiflora*, *E. floccifolia*, *E. tristachya*, *E. intermedia*, *E. africana*, *E. coracana*, and *E. indica*, respectively. *E. coracana* reads were also mapped to its own assembled and identified chloroplast genes ([Table 4](#t4){ref-type="table"}). Since some chloroplast genes have no hit for some species when they do Blast, the genes shared by all of the species were used. The name and type of chloroplast genes are summarized in [Table 5](#t5){ref-type="table"}. A total of 238,136; 246,733; 234,583; 226,923; 248,315; 225,962; and 249,884 reads were mapped to chloroplast genes of *E. multiflora*, *E. floccifolia*, *E. tristachya*, *E. intermedia*, *E. africana*, *E. coracana*, *and E. indica*, respectively, and covered 37,056; 38,012; 34,937; 36,287; 40,171; 37,969; and 42,162 bp of the references, respectively ([Table 4](#t4){ref-type="table"}). The variants (SNVs, MNVs, replacements, insertions, and deletions) detected from the *E. coracana* reads mapping to the chloroplast genes of *Eleusine* species were calculated. The least total variants across all variant types were mapping of *E. coracana* reads to *E. coracana* chloroplast genes. Excluding *E. coracana* and *E. africana*, *E. indica* had lower variants when *E. coracana* reads mapped to chloroplast genes of all *Eleusine* species, followed by *E. tristachya*. The detection of variants between reads of *E. coracana* and other *Eleusine* species in maternally inherited chloroplast further substantiated *E. indica* as the maternal genome donor. Further, this analysis gave us our first indication of a unique possible relationship between *E. coracana*, *E. africana*, *E. indica*, and *E. tristachya* simply based on the lower number of variants that occurred compared to other species. ###### The mapped reads, covered references, mapped percentage and the length of SNVs, MNVs, replacements, insertions, and deletions detected from the *E. coracana* reads mapped to the chloroplast genes of all *Eleusine* species Assembled species Mapped reads Covered reference[*^a^*](#t4n1){ref-type="table-fn"} Mapped percentage SNVs MNVs Replacements Insertions Deletions ------------------- -------------- ------------------------------------------------------ ------------------- ------ ------ -------------- ------------ ----------- *E. coracana* 225,962 37,969 0.2% 15 0 0 0 0 *E. multiflora* 238,136 37,056 0.2% 106 0 0 0 0 *E. floccifolia* 246,733 38,012 0.2% 80 0 0 0 0 *E. tristachya* 234,583 34,937 0.2% 41 0 0 2 0 *E. intermedia* 226,923 36,287 0.2% 364 0 1 1 1 *E. africana* 248,315 40,171 0.2% 14 1 0 2 2 *E. indica* 249,884 42,162 0.2% 33 0 0 0 3 The length of covered reference is similar but not same, because some chloroplast gene sequences are not exactly same. ###### The summary of chloroplast genes used for determination of maternal genome donor of *E. coracana* Category Group Gene name ---------------- ------------------------------------ -------------------------------------------------------- Photosynthesis Subunits of NADH-dehydrogenase *ndhA*, *ndhB*, *ndhD*, *ndhE*, *ndhF*, *ndhG*, *ndhH* Subunits of photosystem I *psaA*, *psaB* Subunits of photosystem II *psbA*, *psbB*, *psbC*, *psbD* Subunits of cytochrome b/f complex *petA* Subunits of ATP synthase *atpA*, *atpB*, *atpE*, *atpI* Large subunit of rubisco *rbcL* Replication Small subunit of ribosome *rps2*, *rps4*, *rps7*, *rps11*, *rps12*, *rps19* Large subunit of ribosome *rpl2* DNA dependent RNA polymerase *rpoA*, *rpoB*, *rpoC1*, *rpoC2* Other Maturase *matK* Protease *clpP* c-type cytochrome synthesis gene *ccsA* Concatenated phylogenetic trees were rooted using chloroplast and ortholog genes separately ([Figure 2A](#fig2){ref-type="fig"}, [2B](#fig2){ref-type="fig"}). In the chloroplast gene derived tree, *E. coracana*, *E. africana*, and *E. indica* formed a clade that is sister to *E. tristachya*. A close phylogenetic relationship of *E. coracana*, *E. africana*, and *E. indica* further supports the hypothesis of *E. indica* as the maternal genome donor to the crop species *E. coracana*. Nuclear gene tree analyses eliminate *E. floccifolia*, *E. intermedia*, and *E. multiflora* as potential maternal genome donors with high bootstrap support. It does not eliminate *E. indica* or *E. tristachya* as a potential maternal genome donor. Our use of single copy genes from an allotetraploid that may have differences in homeologous gene expression limits the conclusions that can be drawn. To better understand the contributions of each subgenome to the super-matrix, subgenome identity was also predicted from individual gene tree topology (Figure S2). These results support *E. indica* as the maternal genome donor of *E. coracana* and again a close relationship between *E. indica* and *E. tristachya*, and also between *E. floccifolia* and *E. multiflora*. Our maternal genome donor conclusions are consistent with approaches such as genomic *in situ* hybridization (GISH), cytogenetic analysis, and phylogenetic analysis that conclude *E. indica* is the maternal parent of *E. coracana* ([@bib5], [@bib6]). [@bib23] also constructed a molecular phylogenetic analysis using two low-copy-number genes in *E. coracana* and concluded that *E. indica* was close to *E. coracana*, consistent with our phylogenetic analysis. Chloroplast DNA is highly conserved and its potential usefulness in phylogenetic studies has been well documented ([@bib14]; [@bib43]; [@bib24]). Here, we broadened the *E. coracana* maternity analysis to all assembled chloroplast genes in all our *Eleusine* transcriptome profiles. In addition, a close relationship between *E. floccifolia* and *E. multiflora* was supported by both of the phylogenetic trees. This relationship has been reported by [@bib42] using *trnT-trnF* region of plastid DNA, by [@bib36] using nuclear *EF-1a* data and by [@bib23] using phosphoenolpyruvate carboxylase 4 (*Pepc4*) gene. {#fig2} Comparative subtraction of the E. africana transcriptome from the E. coracana transcriptome {#s12} ------------------------------------------------------------------------------------------- *E. africana* is considered to be the wild progenitor of domesticated *E. coracana* ([@bib4]). To provide insights into the genomic causes for the evolution in *E. coracana*, comparative transcriptome analysis (single replication of each species only) between *E. africana* and *E. coracana* was conducted, allowing identification of 2,737 genes that were expressed only in *E. coracana* but not in *E. africana*. Phylogenetic analysis ([Figure 2A](#fig2){ref-type="fig"}) also indicated *E. indica* was the maternal genome donor for *E. africana*. These data indicate that *E. indica* and *E. tristachya* possess a close relationship to *E. africana* and *E. coracana*. As such, *E. africana* might be autotetraploid species from *E. indica* genome doubling or through hybridization between *E. indica* and *E. tristachya*. However, such a conclusion is only based on this research, as more evidence using genomic sequencing would be needed to support such a hypothesis. [@bib40] first reported *E. africana* from Africa as a tetraploid form of *E. indica*. Phylogenetic analyses of *E. coracana* genome ([@bib23]) also indicated that *E. indica* and *E. tristachya* were in the same clade with *E. africana* and *E. coracana*, which is consistent with the results in this research. Conclusions {#s13} ----------- In this study, we constructed optimized transcriptome references for *E. multiflora*, *E. floccifolia*, *E. tristachya*, *E. intermedia*, *E. africana*, and *E. coracana* and the relationships among *Eleusine* species were investigated. By comparing the chloroplast genes among *Eleusine* species, we demonstrated that *E. indica* is the maternal genome donor and a maternal relationship exists between *E. indica* and *E. tristachya*. It is traditionally accepted that *E. coracana* evolved from the *E. africana* ([@bib26]) and is substantiated by more recent research ([@bib16]). Transcriptomes are made publicly available for comparison to other species and to aid in identifying the paternal genome donor. Abundant *Eleusine* genetic resources from this research will be useful for the continued study of *Eleusine* evolution. This project was supported by the Alabama Agricultural Experiment Station and the Hatch program of the National Institute of Food and Agriculture, U.S. Department of Agriculture. The authors would like to thank the Alabama Supercomputer Center and the Auburn Hopper Supercomputer clusters for computational support. Hui Zhang was supported by a scholarship from China Scholarship Council (CSC). Supplemental material available at FigShare: <https://doi.org/10.25387/g3.7994039>. Communicating editor: A. Doust

{.554}

Competing interest statement ============================ Conflict of interest: the authors declare no potential conflict of interest. Introduction {#sec1-1} ============ Paroxysmal nocturnal hemoglobinuria (PNH) is an uncommon acquired, as distinct from inherited, genetic disease characterized by complement-mediated hemolysis and clone expansion of affected cells of various hematopoietic lineages that are thought to be derived from an abnormal multipotent hematopoietic stem cell. The mutation is somatic and occurs in the X-linked PIG-A gene encoding for glycosylphosphatidylinositol (GPI)-protein, required for the biosynthesis of glycosylphosphatidylinositol anchor, that attaches dozens of proteins to the cell surface.^[@ref1]^ The GPI anchor is deficient in the affected cells from patients with PNH, leading to the lack of surface expression of multiple GPI-anchored proteins, such as decay-accelerated factor CD55 and CD59, both of which play roles in the protection of red cells from the action of complement.^[@ref5],[@ref6]^ Free hemoglobin released from intravascular hemolysis leads to circulating nitrous oxide depletion and contributes to many of the clinical manifestations of PNH, including fatigue, erectile dysfunction, esophageal spasm, and thrombosis.^[@ref7]^ The condition is extremely rare in children, where it is often misdiagnosed, and it is more likely to occur in teenagers than in younger children.^[@ref8]^ Thromboses are very severe complications of the disease, frequently occurring in unusual sites such as the skin, liver, spleen portal circle, and affecting 29 to 44% of patients.^[@ref13]^ However, due to the rarity of PNH in childhood it is difficult to estimate the exact incidence in affected pediatric populations, with estimates ranging from as low as 5 to 50%.^[@ref14]^ We describe a case of a girl with a severe thrombotic complication of the disease. Case Report {#sec1-2} =========== M. is a 14 year-old female, born full-term to unrelated healthy parents after an uneventful pregnancy. Her parents reported regular growth and denied any clinical problems during the childhood. In February 2014 an unexpected finding of slight macrocytic anemia, leucopenia and mild thrombocytopenia (hemoglobin \[Hb\] 10.6 g/dL, mean corpuscular value 100 fL, iron 81 µg/dL, ferritin 14 µg/L, white blood cell \[WBC\] count 3.8×10^9^/L, neutrophils 2.6×10^9^/L, platelets 62×10^9^/L) was identified at routine blood analysis requested of the pediatrician in response to paleness in the child. A carential anemia due to menarche was diagnosed and iron, vitamin B12 and folic acid commenced. After three months of iron and folic therapy there was no improvement, but rather a worsening in the blood count. In June 2014, after a febrile episode, a blood count was performed which continued to show thrombocytopenia (platelets 101×10^9^/L) and a worsened anemia (Hb 8.4 g/dL), WBC 3.8×10^9^/L, neutrophils 1.3×10^9^/L. In August 2014 the girl was referred to a children's hospital after the onset of progressive fatigue, associated with anorexia and paleness. The patient underwent blood routine biochemistry and a complete blood cell count. Laboratory findings showed severe anemia (6 g/dL) with negative Coombs test, mild leucopenia (WBC 4.9×10^9^/L) and thrombocytopenia (platelets 97×10^9^/L) and high values of lactate dehydrogenase (LDH, 2855 IU/L). During the hospitalization the girl was subjected to a packed red cells transfusion, and to a further diagnostic work-up which included: bone-marrow analysis (which showed a normal cellularity without abnormal cells), autoimmunity antibodies complete panel, vitamin B12 and folate, urine analysis, tumoral markers (α-fetoprotein, CA 19-9, carcinoembryonic antigen \[CEA\], vanillylmandelic acid \[VMA\], neuron-specific enolase \[NSE\]), abdominal ultrasound, chest X-ray and total body computed tomography (CT). Clinical, laboratory and instrumental findings were all unremarkable, with the exception of sporadic hemoglobinuria, increase in reticulocyte count (178,145/μL) and persistently elevated levels of LDH, with no other sign of hemolysis. The girl was therefore discharged. On 9 September, in response to the worsening of symptoms and the onset of pallor, the pediatrician ordered blood tests that showed a severe anemia (Hb 6.8 g/dL) and thrombocytopenia (platelets 112×10^9^/L) and high LDH serum levels (3550 IU/L). The patient was hospitalized, transfusions were started immediately and transfer to the Pediatric Hospital was decided upon. During hospitalization, Coombs test was negative, total bilirubin 1.4 mg/dL, direct bilirubin 0.4 mg/dL, haptoglobin 8 mg/dL, autoimmunity negative, vitamin B12 and folates normal. A urine test showed traces of hemoglobinuria. The patient underwent bone marrow biopsy that showed an absence of aberrant phenotypes, depletion of CD34^+^ stem cells, lymphocytic depletion of B cells, and absence of B cell lymphoid progenitors. Throughout this time, LDH was always \>1.5 × upper limit of normal (ULN; 11 September: 1283 IU/L; 12 September: 1469 IU/L; 15 September: 1436 IU/L). The patient was discharged in good condition and did not report significant disorders. Laboratory findings at discharge were: Hb 11.1 g/L, WBC 3.4×10^9^/L, platelets 119×10^9^/L, and LDH 1210 IU/L. During school hours on 17 October, before any further examination could be performed, the patient presented an episode of loss of consciousness preceded by general discomfort, fatigue, blurred vision, vomiting, mental disorientation and speech difficulties. On arrival at the emergency area, the girl presented complete right hemiplegia, aphasia and underwent a deterioration of mental status leading to coma (GCS 7/15). Magnetic resonance imaging (MRI) and MR angiography (MRA) of the brain were performed immediately; MRI revealed a massive ischemic lesion of the territory fed by the middle cerebral artery in relation to a thrombotic event that had occurred both in this vessel and in the left median cerebral artery and MRA showed a significant flow reduction of left intracranial portion of the internal carotid artery and complete absence of ipsilateral middle cerebral artery flow ([Figure 1](#fig001){ref-type="fig"}). Right hemiplegic pyramidal syndrome with cranial-caudal decreasing gradient and motorial aphasia with residual understanding of simple orders was apparent. Right positive Babinski sign was present. Laboratory finding were: Hb 10.0 g/dL; WBC 2.6×10^9^/L, platelets 95×10^9^/L, LDH 1630 IU/L. After a brief period (24 hours) of hospitalization at the emergency department, during which intravenous low molecular weight heparin at 100 units/kg was started, the girl was transferred to pediatric neuropsychiatry where she rapidly recovered from aphasia and the hemiplegic syndrome. After consulting the pediatric onco-hematologist, a hemoglobin study using high-performance liquid chromatography (HPLC), osmotic resistance and ADAM metallopeptidase with thrombospondin type 1 motif 13 (ADAMTS 13) and related antibodies were performed trying to find the pathogenesis of the cerebral thrombosis: all of them were normal. According to the International Clinical Cytometry Society (ICCS) 2010 guidelines,^[@ref15]^ all patients with cytopenia associated with hemolysis and/or thrombosis should be studied for PNH. Therefore we performed high sensitivity flow cytometry on flow cytometer Navios™ (Beckman Coulter, Ca, USA) and the presence of a PNH clone was shown: CD66b negative equal to 93.7% of granulocytes, CD14 negative equal to 95.2% of monocytes and CD59 negative equal to 1.5% of the erythrocytes. The bone marrow biopsy documented erythroid hyperplasia and left shift of the line myeloid maturation. The girl was initially managed with low molecular weight heparin, steroids, vitamin B12 and folic acid, after that, in November, an oral anticoagulant (warfarin) and intravenous therapy with the humanized monoclonal antibody eculizumab were started with good results (no further transfusions were needed). Concomitant treatment with warfarin was continued. In July 2015 a follow up MRI revealed a gliotic and poroencephalic outcome of the previous ischemic lesion in left cortical-subcortical fronto-temporal region, with ex-vacuo dilatation of the ipsilateral ventriculum ([Figure 2](#fig002){ref-type="fig"}). At follow up in March 2016 aphasia was almost regressed and gait normalized. Motility was recovered in the right upper limb, but was still absent at the level of the wrist and right hand. The patient had not required further transfusion. Hematological samples withdrawn just before the infusion with eculizumab showed average values of Hb 8.9-9.4 g/dL, platelets 120-140×10^9^/L, reticulocytes 170-310,000/mL, LDH 700-1,100 IU/L, total bilirubin 1.2-1.6 mg/dL. Written informed consent was obtained from the parents of the patient for publication of this case report and the accompanying images. Approval from the hospital's ethics committee was not applicable, as this was an individual clinical case. Discussion {#sec1-3} ========== Though arterial occlusion is a less frequent complication in paroxysmal nocturnal hemoglobinuria, accounting for about 15% of total thromboembolism events in adults, cerebrovascular complications such as stroke or transient ischemic attack represent almost 90% of arterial thrombotic events in this condition.^[@ref16]^ Audebert *et al.* reported two cases of cerebral ischemic infarction, one of which was a 16-year old girl with known PNH and aplastic anemia, and reviewed 7 previously reported cases in the literature, one on a 11-year old girl with pancytopenia, obtaining follow-up information on 4 of the 5 patients who survived.^[@ref17]^ Both of the two cases had experienced hemoglobinuria and cytopenia prior to the thrombotic events, and in one case diagnosis of PNH was only established after stroke onset. The review of the other 7 cases showed that two patients died during their in-hospital treatment or during a short period after discharge. Out of the nine patients, four died within a few weeks after the initial thrombotic event, and five had the arterial thrombosis as one of the first manifestations of PNH and the underlying disease was diagnosed later during the clinical course.^[@ref17]^ It appears that in the case of thrombotic complications both the prompt administration of eculizumab and thrombolytic therapy, above all with administration of intravenous tissue plasminogen activator, may be decisive in improving outcome.^[@ref18]^ In children, typical PNH is rare, manifests mostly in teenage years and the incidence of thrombosis has an average of 23%, ranging up to 50%.^[@ref14]^ Due to the disease rarity the diagnosis is, as in our case, often delayed.^[@ref8],[@ref17]^ In our patient eculizumab was not started immediately after cerebral stroke (the correct diagnosis was not made until a few days later) whereas thrombolytic therapy was not performed by decision of the emergency physicians, since the girl showed a rapid recovery of general conditions. In fact, in the event of an earlier diagnosis we would have immediately started therapy with eculizumab and warfarin and the girl would likely not have suffered thrombotic complications. Conclusions {#sec1-4} =========== In conclusion, we think that in the presence of cytopenia, regardless of type, a child should immediately be referred to a pediatric onco-hematology unit and that in the case of an association with hemolytic signs, if the most common diseases (*i.e.* autoimmune hemolytic anemia) have been excluded, the child should be immediately evaluated for a PNH clone. This approach, making use of a simple and low-cost test, could ensure appropriate management of the condition and avoid the likelihood of a severe and potentially fatal outcome, linked to the elevated risk in PNH patients of both venous and arterial thrombosis. We thank Ray Hill, an independent medical writer, who provided English language editing assistance and medical writing support on behalf of Health Publishing & Services Srl. The editorial support was funded by Alexion. No other funding was received for this report. The Parents' Association Sicilian Primary Immunodeficiency Association is acknowledged for supporting the activity of the Pediatric Onco-Hematology Unit of A.R.N.A.S. Ospedali Civico, Di Cristina e Benfratelli. No specific funding was received for this study. {#fig001} {#fig002} [^1]: Contributions: the authors contributed equally. [^2]: Conference presentation: Presented in part as an abstract at the XL Congresso Nazionale AIEOP, Lecce, Italy, May 2015. Abstract P078, Emoglobinuria parossistica notturna in età pediatrica: evenienza rara ma può essere importante pensarci.

1. Introduction {#sec1} =============== Reversible protein tyrosine phosphorylation provides a potent and versatile mechanism for regulating various physiological processes in eukaryotic cells. Protein tyrosine kinases (PTKs), which catalyze the transfer of phosphoryl group to tyrosine residues, and protein tyrosine phosphatases (PTPs), which remove phosphate moiety from phosphotyrosine residues, work together to modulate signaling pathways. In some cases, PTPs play a dominate role in shaping the spatio-temporal dynamics of protein tyrosine phosphorylation network, making it a pivotal target for functional manipulation ([@bib27]). Although the study of PTPs lagged behind of PTKs for historical reasons and technical difficulties, a large number of PTPs have been discovered in recent years, paralleling PTKs in sequence diversity and functional complexity ([@bib1]). Despite a minority reported to be Asp-based PTPs, all others possess at least one signature motif, \[I/V\]HCX~5~R\[S/T\] and share a Cys-based catalytic mechanism. Some PTPs serve as pTyr-specific enzymes, others can also dephosphorylate Ser/Thr residues, thus they are operationally defined as dual specificity phosphatase (DUSP). And the ability of removing phosphate group from phospholipid or RNA is also reported in some DUSPs ([@bib35]). Intriguingly, numerous host--virus interactions are based on the interplay of PTKs and PTPs as revealed by studies on vertebrates and invertebrates ([@bib5], [@bib19]). Mounting a potent immune response to invading virus relies on proper phosphorylation or dephosphorylation of signaling molecules in host cells. Reciprocally, virus develops strategies to utilize or disrupt this regulatory mechanism for survival and transmission. For instance, *in vitro* studies using mammalian cell lines demonstrated that DUSP1, which dephosphorylates MAPKs, facilitates the infection of hepatitis C virus (HCV) and coronavirus infectious bronchitis virus (IBV) ([@bib2], [@bib23]). And congenital human cytomegalovirus (HCMV) can take advantage of CDC25, a cell cycle regulator to promote viral replication ([@bib7]). However, due to the lack of systematic analysis of insect PTPs and distinct antiviral mechanism that insect may take, few surveys reported the relevance of insect PTPs to antiviral immune response. Some virus also encode PTPs, which have been hypothesized or demonstrated to function as virulence factors that impair host immune response or other functions ([@bib11], [@bib28], [@bib29]). The best-known example in insect is the expression of several bracovirus-encoded PTPs in lepidopteran hosts promoted the apoptosis of granulocytes and inhibited spreading and phagocytosis of plasmatocytes, therefore facilitating parasitoid development ([@bib12], [@bib13], [@bib30]). Interestingly, phylogenic analysis showed that those PTP genes were acquired from braconid genome and experienced complex evolutionary change ([@bib6], [@bib32]). Recent studies on another subset of insect virus identified a nucleopolyhedrovirus (NPV)-encoded PTP, which was presumably acquired from a lepidopteran host by horizontal gene transfer (HGT) as an essential factor to induce an enhanced locomotory activity (ELA) in host caterpillars ([@bib15], [@bib36]). And NPV *ptp* deletion mutant lost its ability to manipulate its host behavior. Surprisingly, later study found BmNPV PTP functions as a virus-associated structural protein rather than as an enzyme in regard to the induction of ELA, although it still maintains the catalytic activity ([@bib17]). Moreover, BmNPV *ptp* deletion mutant had reduced progeny production in silkworm larvae and showed a delay in late gene expression in cell lines. An earlier study identified a BmNPV *ptp* homolog encoded by *Autographa californica* multicapsid NPV (AcMNPV) as a dual phosphatase containing RNA 5'-triphosphatase activity which is implicated in processing of viral late mRNAs ([@bib34]). In contrast to BmNPV *ptp* mutant, AcMNPV *ptp* mutant seems to have no defect in replication in insect larvae, but partially defect in occluded virus production in a cell-specific manner ([@bib22]). Given that HGT between host and pathogen has been proposed to be an important mechanism to increase pathogen survival and propagation ([@bib3]), it is interesting to trace back the function of the host copies to better understand the interaction between the host and pathogen. In the present study, we first analyzed BmPTP-h based on its functional categorization, then we investigated the immune-regulatory role it would play under the challenge of BmNPV *in vitro* and *in vivo*. We found that it increased expression level after viral infection and benefited viral replication. Our result provides complementary insights into the molecular machinery exploited by insect virus which could be pursued for antiviral strategy. 2. Experimental procedures {#sec2} ========================== 2.1. PTP identification and classification {#sec2.1} ------------------------------------------ Based on the annotated information of *Bombyx mori* comprehensive gene sets (<http://sgp.dna.affrc.go.jp/ComprehensiveGeneSet/>), sequence of protein predicted to possess tyrosine phosphatase domain was extracted and analyzed by online software SMART (<http://smart.embl-heidelberg.de/>). Protein that contains the PTP signature motif (CX~5~R) was then used as the query in BLASTP search against NCBI sequence repository to identify their paralogs. *B. mori* PTPs were classified into subfamilies according to their orthologous PTPs in human and fruit fly ([@bib1], [@bib10]). 2.2. Insect, cells and virus {#sec2.2} ---------------------------- Silkworm larvae (DaZao P50 strain) were reared on fresh mulberry leaves at 25 °C and relative humidity of 80%. BmE cells were maintained at 27 °C in Grace medium (Gibco, UK) supplemented with 10% FBS (Hyclone, USA). Wide-type BmNPV (Guangdong strain, China) and recombinant BmNPV-GFP (courtesy of Dr. Xiaofeng Wu, Zhejiang University, China) were propagated in BmE cells and silkworm larvae following the standard procedure. Viral titer were determined by plaque assay in BmE cells ([@bib18]). 2.3. *Bmptp-h* over-expression and RNAi {#sec2.3} --------------------------------------- BmPTP-h-coding sequence (NCBI Gene ID: 692515) was cloned into the pSL1180 expression vector with a Flag tag at the N-terminus using primer set 1 in [Table 1](#tbl1){ref-type="table"} dsRNA to *Bmptp-h* or red fluorescent protein gene (*RFP*) (dsBmPTP-h or dsRFP) were *in vitro* synthesized using T7 RiboMAX Large Scale RNA Production System (Promega, USA) with the primer sets 2 and 4. Plasmids or dsRNA were transfected into BmE cells using X-treme GENE transfection reagent (Roche, Switzerland) following the manufacturer\'s instruction. dsRNA was injected into silkworm larvae (20 μg/larva) at second day of fifth instar through the second last stoma in abdomen with a fine needle.Table 1Sequence of primers used in this work.GenePrimer set no.Sequence of primers (5′-3′)Bmptp-h1F: CGGGATCCATG[GATTACAAGGATGACGACGATAAG]{.ul}CCTAAACTTCCCGATAGATGG\ R: ATTTGCGGCCGCTTTAACGATACCTTCTTCTAGTTGTTTCAG2F: TAATACGACTCACTATAGGGAGAGACTCGGTGCAGTCATAGAT\ R: TAATACGACTCACTATAGGGAGACTGCTGTTAGTCCGCTCTT3F: ACAATGCTTGCGGACGGGTAAT\ R: CCTCTAGAAGCGCCGGAATGTCRFP4F: TAATACGACTCACTATAGGG GTACGGCTCCAAGGTGTACG\ R: TAATACGACTCACTATAGGGGGTGTAGTCCTCGTTGTGGGgp645F: CCATCGTGGAGACGGACTA\ R: CTCGCACTGCTGCCTGAsw229346F: TTCGTACTGGCTCTTCTCGT\ R: CAAAGTTGATAGCAATTCCCTBmGAPDH7F: CATTCCGCGTCCCTGTTGCTAAT\ R: GCTGCCTCCTTGACCTTTTGC[^1] 2.4. Viral infection {#sec2.4} -------------------- For infection of cells, 5000 virions were incubated with 1×10^6^ BmE cells 48 h after transfection. For infection of silkworm larvae, 10 000 virions were injected into silkworm larvae at Day 3 of 5th instar. At different time points after infection as indicated in graph, the expression level of *Bmptp-h* or BmNPV *gp64* was subjected to quantitative PCR analysis. 2.5. Quantitative PCR (qPCR) analysis {#sec2.5} ------------------------------------- Total RNA was extracted from cells or tissue samples using Total RNA Kit (Omega, USA) at different time points post viral infection and qPCR was performed using SYBR Premix Ex Taq II (TaKaRa, Japan) on a StepOne Plus Real-Time PCR System (Applied Biosystems, USA) with a program consisting of an initial denaturing step of 30 s at 95 °C and 40 amplification cycles consisting of 5 s at 95 °C followed by 30 s at 60 °C. *Bmptp-h* and *sw22934* were amplified with primer sets 3 and 6 accordingly. For analysis of BmNPV replication, total DNA was extracted from viral infected BmE cells using the Universal Genomic DNA Extraction Kit (Takara) as described. BmNPV *gp64* and *BmGAPDH* were amplified with primer sets 5 and 7 respectively. The expression level of *Bmptp-h* or BmNPV *gp64* genes were normalized to the control *sw22934* or *BmGAPDH*. The relative viral abundance was determined as described previously ([@bib14]). 2.6. Western blotting {#sec2.6} --------------------- Cells were solubilized in lysis buffer (100 mM NaCl, 50 mM Tris--HCl, 0.1% SDS, 1% NP-40 pH 7.5) containing protease inhibitor cocktail (Roche). After centrifugation at 17 000 g for 15 min at 4 °C, the supernatant was collected on ice for western blot analysis. Concentration of cell lysate was determined by BCA assay. A total of 20 μg protein per sample was resolved on 12% SDS-PAGE. After transfer to PVDF membrane (GE Health Care, USA), samples were immuno-blotted with anti-FLAG mAb or anti-Tubulin mAb (Sigma, USA) following the standard procedure. 2.7. Tyrosine phosphatase assay {#sec2.7} ------------------------------- 1×10^6^ BmE cells transfected with BmPTP-h expression vector were solubilized in 200 μl lysis buffer, then pre-cleared and incubated with casein (0.2 mM) for 10 min at 30 °C. The reaction was terminated by addition of TCA buffer. After centrifugation at 17 000 g for 10 min, 200 μl supernatant was mixed with 20 μl molydbate dye and liberated phosphate was visualized at 750 nm using a microplate reader (Promega). 2.8. Statistical analysis {#sec2.8} ------------------------- Data were presented as the mean ± S.D. (n = 3). Statistical significant differences were determined by Student\'s *t*-test (for comparison of two means), or one-way analysis of variance (ANOVA) for multiple comparison test (\**p* \< 0.05, \*\**p* \< 0.01). 3. Results {#sec3} ========== 3.1. Identification and classification of BmPTPs {#sec3.1} ------------------------------------------------ [Table 2](#tbl2){ref-type="table"} summarizes the 36 putative Cys-dependent tyrosine phosphatases identified in silkworm. All of them have full-length cDNA or matching EST sequence which indicates that they are not pseudogenes, except three genes (NCBI Gene ID: 101747207, 101735618, 101737822) which were predicted based on genome information. These tyrosine phosphatases were classified into four subfamilies based on their phosphatase catalytic domain sequences, including classical pTyr-specific PTPs, VH1-like dual-specific PTPs (DUSPs), low molecular weight PTP (LMWP) and CDC25. The classical PTPs were then divided in two subgroups, receptor- or non-receptor type PTPs based on whether they contained transmembrane domain or not. And DUSPs were annotated as MKPs (mitogen-activated protein kinases phosphatase), PTEN (phosphatase and tensin homologue deleted on chromosome 10), PRL (phosphatase of regenerating liver), CDC14, Myotubularins and atypical DUSPs based on their human and fruitfly orthologs, respectively. There are 14 classical pTyr-specific PTPs, 20 DUSPs, 1 LMWP and 1 CDC25 (and 1 Asp-dependent tyrosine phosphatase, EYA, NCBI GI: 827538455). Compared with the 43 Cys-dependent tyrosine phosphatases identified in *Drosophila melanogaster*, silkworm PTP gene set is smaller. Particularly, silkworm has less members in receptor-type, Myotubularin and LMWP subfamilies. Although no clear orthology relationship can be ascribed to some silkworm atypical DUSPs, they bear considerable similarity to human and *Drosophila* DUSPs, especially in PTP domains. Most of silkworm tyrosine phosphatases contain the conserved signature motif sequence \[I/V\]HCX~5~R\[S/T\], a few display variations at the first two amino acid residues and one has Gln instead of Ser/Thr at last position which is predicted to be catalytic inactive. Whereas 79 out of the 107 human PTPs are multidomain proteins, less than a half of silkworm PTPs harbor at least one characteristic domain, which may contribute to the specific interaction with regulatory or targeting molecules.Table 2Categories of silkworm PTPs.PTP subfamily/Gene IDSignature motifCharacteristic domainHuman orthologFruitfly orthologSequence source**Classical pTyr-specific PTPs**Receptor101746939VHCSAGVGRT\ VHCQTGCERSIG, FN3RPTPαPTP69DEST,CDS101736187VHCSAGVGRSFN3RPTPβPTP10DFL cDNA101737535VTCASGAGRSPTPRγCG42327FL cDNA101738718VHCSDGAGRSRPTPN2RPTPN2IA-2FL cDNA101736712IHCSAGIGRTPTPRRPTP-ERFL cDNA101747207VHCSAGVGRTPTPRSPTP99ACDSNon-receptor100579140VHCSAGIGRSPTP1BPTP61FFL cDNA101737573VHCSAGVVRTERM,FERM_C,PDZPTPN4PTPmegEST,CDS101744807VHCSAGIGRTCRAL_TRIO_N,Sec14PTPN9l(1)G0232FL cDNA101735618YCCGEGAGRSERM,FERM_CPTPN14PezCDS101736631IQCGAGAGRSBRO1-likePTPN23MOPEST,CDS101736863VHCSAGVGRTSH2SHP2CswEST,CDS101737822VHCSAGIGRTSH2SHP2CswCDS101735492VHCNDGGGRS VHCNDGGGRSRPTPκPTP36EEST,CDS**VH1-like dual-specificity PTPs**MKP101744546VHCHFGVSRSMKP-4MKP-4FL cDNA101740325VHCVAGVSRSDUSP7MKP-3EST,CDSPTEN101745952VHCKAGKGRTPTEN_C2PTENPTENEST,CDSPRL101743511VHCVAGLGRAPRL-1PRL-1FL cDNASlingshot101737293VHCKMGISRSSSH-N, DEK_CSSH1SSHFL cDNACDC14101740096VHCKAGLGRTCDC14ACDC14EST,CDS101744892IHCHAGLGRTPTP9Q22CDC14EST,CDSMyotubularin101740354VHCSDGWDRTFYVEMTMR3CG3632EST,CDS101736509VHCSDGWDRTFYVEMTMR6CG3530FL cDNA101739438FLCDTDQERQMTMR9[a](#tbl2fna){ref-type="table-fn"}CG5026EST,CDS101736147VHCSDGWDRTGRAMMTMR2MTMFL cDNAAtypical**692515VHCTHGLNRTDUSP11CG13197**FL cDNA101738891VHCYFGVSRSDUSP12MKP-4FL cDNA101738231VHCMIGVSRSDUSP13CG7378FL cDNA101743164VHCVAGVSRSDUSP14CG15528EST,CDS101735761IHCLAGMSRSDUSP22CG10089FL cDNA101744063IHCRQGRSRSDUSP23[b](#tbl2fnb){ref-type="table-fn"}CDC14[b](#tbl2fnb){ref-type="table-fn"}FL cDNA101742480VHCRHGRGRTDUSP23[b](#tbl2fnb){ref-type="table-fn"}CDC14[b](#tbl2fnb){ref-type="table-fn"}EST,CDS101737334VHCKAGRTRSPTPMT1PlipFL cDNA101740551VHCTHGFNRTNADH_4Fe--4SmRNA-capmRNA-capFL cDNA**Low molecular weight PTP**692857FICLGNICRSLMWPPrimo-1FL cDNA**CDC25**100216491FHCEFSLERGRHODCDC25CStgEST,CDS[^2][^3][^4] *Bmptp-h* (NCBI Gene ID: 692515), which was proposed to be the origin of BmNPV *ptp*, encodes an atypical VH1-like DUSP. It is homologous to *Drosophila* CG13197 and distantly related to human DUSP11, and the latter binds splicing complex and has been implicated in inflammatory bowel disease ([@bib9], [@bib37]). BmPTP-h may also have 5'-triphosphatase activity using RNA as a substrate like DUSP11 ([@bib4]). Clear function of these molecules is unknown. 3.2. Expression profile of *Bmptp-h* during the development and in different tissues {#sec3.2} ------------------------------------------------------------------------------------ To evaluate the physiological functions of BmPTP-h, we first analyzed its expression profiles at different developmental stages or in different tissues by qPCR. We found the expression level of *Bmptp-h* increased gradually at larval stage, and it reached the highest level on the 3rd day of fifth instar (I5D3), which was more than 15-times higher than in 1st instar (I1), then decreased afterwards until the end of pupal stage ([Fig. 1](#fig1){ref-type="fig"} A). Next, we examined *Bmptp-h* transcriptional levels in different tissues dissected from I5D3 larvae ([Fig. 1](#fig1){ref-type="fig"}B). A relative high level of *Bmptp-h* was detected in the fat body, followed by testes.Fig. 1Expression profile of *Bmptp-h* during the development and in different tissues. (A) Expression profile at different developmental stages. (B) Expression profile in different tissues of larvae at third day of fifth instar. The *Bmptp-h* mRNA level measured by qPCR was normalized to the internal control (*sw22934*) and represented as mean ± S.D. (n = 3). The letters marked at the top of columns indicate statistical significance of means detected by ANOVA analysis. Means with the same letter are not significantly different from each other. I, instar; D, day; FB, fat body; HE, head; HC, haemocyte; MG, midgut; SG, silk gland; ID, imaginal disc; OV, ovary; TE, testis. 3.3. Expression profile of *Bmptp-h* under NPV infection {#sec3.3} -------------------------------------------------------- Since previous studies have shown BmNPV PTP is important for viral productive infection in silkworm larvae and cell line, and BmPTP-h shares 48.2% identity (80.7% similarity) with BmNPV PTP on amino acid sequence ([@bib34]), we speculated that BmPTP-h also plays a role in regulating interaction between virus and host. For that reason, we injected NPV virions into I5D3 larvae and investigated the expression profile of *Bmptp-h* during infection. Although the expression level of *Bmptp-h* changed over time during the progression, viral infection induced higher level of *Bmptp-h* transcription compared to the uninfected larvae at the same time point ([Fig. 2](#fig2){ref-type="fig"} A). The expression level of *Bmptp-h* in uninfected I5D7 larvae (96 h) was lower than that at I5D3 (0 h), which is in accordance with the result shown in [Fig. 1](#fig1){ref-type="fig"}A. However, in viral-infected larvae *Bmptp-h* transcript level was much higher at 96 h. We also found expression of *Bmptp-h* was potently induced by BmNPV infection in BmE cells after 48 h compared with uninfected cells and increased sharply afterwards ([Fig. 2](#fig2){ref-type="fig"}B).Fig. 2Expression profile of *Bmptp-h* in silkworm larvae and BmE cells under BmNPV infection. (A) Expression of *Bmptp-h* in silkworm larvae infected with BmNPV. (B) Expression of *Bmptp-h* in BmE cells infected with BmNPV. The *Bmptp-h* mRNA level measured by qPCR was normalized to the internal control (*sw22934*) and annotated as fold increase (±S.D., n = 3) over the normalized value at 0 min. Statistical significance between the expression level of *Bmptp-h* in uninfected and BmNPV-infected larvae or BmE cells at the same time point was assessed using Student\'s *t*-test. \**p* \< 0.05, \*\**p* \< 0.01. 3.4. Effects of *Bmptp-h* knock-down and over-expression to NPV infection in BmE cells {#sec3.4} -------------------------------------------------------------------------------------- To test the tyrosine phosphatase activity of BmPTP-h, we expressed a Flag-tagged BmPTP-h in BmE cells ([Fig. 3](#fig3){ref-type="fig"} A), and measured the enzymatic level of cell lysates ([Fig. 3](#fig3){ref-type="fig"}B). Cells expressing Flag-BmPTP-h showed a significant increase in the overall PTP activity level compared to the control, indicating that BmPTP-h is catalytic active and able to regulate the cellular functions via tyrosine dephosphorylation. To investigate whether BmPTP-h is involved in anti-viral response, we performed dsRNA-mediated RNAi to deplete the expression of *Bmptp-h* in BmE cells ([Fig. 4](#fig4){ref-type="fig"} A). About 80% decrease in transcript level was obtained by dsBmPTP-h compared with dsRFP. 48 h after dsRNA transfection, we challenged cells with BmNPV or BmNPV-GFP, and analyzed the viral abundance by qPCR or fluorescence microscopy ([Fig. 4](#fig4){ref-type="fig"}B and C). Replication of viral DNA or the amount of GFP-positive cells was remarkably reduced when *Bmptp-h* was depleted. We then over-expressed *Bmptp-h* in BmE cells and measured the susceptibility of these cells to BmNPV or BmNPV-GFP infection ([Fig. 5](#fig5){ref-type="fig"} ). Viral replication or GFP-positive cells significantly increased in BmPTP-h-overexpressing cells than in control. These results suggest that BmPTP-h promotes viral productive infection in cells.Fig. 3Phosphatase activity of BmPTP-h. (A) Lysates of BmE cells transfected with pSL1180 empty vector or Flag-BmPTP-h expressing plasmid were subjected to immunoblotting with anti-Flag antibody. (B) Cell lysates were precleared of free phosphate and incubated with casein, then mixed with molydbate dye and visualized at 750 nm. The results are given as the mean ± S.D. (n = 3). Statistical significance was assessed using Student\'s *t*-test. \**p* \< 0.05, \*\**p* \< 0.01.Fig. 4Knock-down of *Bmptp-h* reduces BmNPV replication in BmE cells. (A) BmE cells were transfected with dsRFP or dsBmPTP-h, and knock-down efficiency of *Bmptp-h* was measured by qPCR analysis. (B) 48 h after dsRNA transfection, cells were infected with BmNPV. Relative abundance of virus determined by BmNPV *gp64* expression level was measured by qPCR analysis. (C) Fluorescence (GFP) and bright-field (BF) images of dsRFP- or dsBmPTP-h-treated cells at 72 h post infection of BmNPV-GFP. The mRNA level of target genes was normalized to the internal control (*sw22934* or *BmGAPDH*) and represented as mean ± S.D. (n = 3). Statistical significance was assessed using Student\'s *t*-test. \**p* \< 0.05. Scale: 50 μm.Fig. 5Over-expression of *Bmptp-h* enhances BmNPV replication in BmE cells. (A) BmE cells were transfected with vector or Flag-BmPTP-h expressing plasmid 48 h before BmNPV infection. Relative abundance of virus determined by BmNPV *gp64* expression level was measured by qPCR analysis. (B) Fluorescence (GFP) and bright-field (BF) images of vector- or BmPTP-h-transfected cells at 72 h post infection of BmNPV-GFP. The mRNA level of target genes was normalized to the internal control (*sw22934* or *BmGAPDH*) and represented as mean ± S.D. (n = 3). Statistical significance was assessed using Student\'s *t*-test. \**p* \< 0.05, \*\**p* \< 0.01. Scale: 50 μm. 3.5. Effects of *Bmptp-h* knock-down to NPV infection *in vivo* {#sec3.5} --------------------------------------------------------------- In order to investigate whether depletion of *Bmptp-h* has any effect to viral infection *in vivo*, we injected dsRNA into silkworm larvae 12 h before BmNPV infection and measured viral abundance. Since RNAi efficiency might not be the same in different tissues, we dissected several tissues and examined the expression level of *Bmptp-h* in them separately ([Fig. 6](#fig6){ref-type="fig"} A). With the exception of midgut (data not shown), the relative expression levels of *Bmptp-h* decreased to 25%--40% after injection of dsRNA targeting *Bmptp-h* (dsBmPTP-h) compared to the control dsRFP in fat body, trachea and head. Furthermore, in those tissues viral abundance was all significantly decreased ([Fig. 6](#fig6){ref-type="fig"}B), and the reduction was most dramatic at 96 h post infection, suggesting that *Bmptp-h* silencing might have adverse effect to viral replication *in vivo*.Fig. 6Knock-down of *Bmptp-h* reduced BmNPV replication *in vivo*. (A) dsRFP or dsBmPTP-h was injected into silkworm larvae at second day of fifth instar and knock-down efficiency of *Bmptp-h* was measured in different tissues. (B) 24 h after dsRNA treatment, silkworm larvae were injected with BmNPV. Relative abundance of virus post infection determined by BmNPV *gp64* expression level was measured in corresponding tissues. The mRNA level of target genes was normalized to the internal control (*sw22934* or *BmGAPDH*) and represented as mean ± S.D. (n = 3). Statistical significance was assessed using Student\'s *t*-test. \**p* \< 0.05, \*\**p* \< 0.01. 4. Discussion {#sec4} ============= In this study, we found BmNPV infection induced the expression of BmPTP-h, a dual-specific phosphatase in silkworm larvae and BmE cells. Partial depletion of *Bmptp-h* by dsRNA inhibited BmNPV replication, and over-expression of *Bmptp-h* resulted in enhanced viral replication. The results suggest that BmPTP-h might be one of the host factors that are beneficial to the replication of BmNPV. Although RNA interference is considered to be the major antiviral mechanism in insect, induction of immune signaling pathways, including Toll, IMD and JAK-STAT pathways have been observed upon viral infection ([@bib26], [@bib31], [@bib38]). Since the signal transduction largely relies on protein phosphorylation, as a dual-specific phosphatase, BmPTP might be exploited by virus to disrupt the signaling cascades through targeting the phosphorylated molecules, resulting in immunosuppression. Furthermore, since double mutation of NPV *lef-4* and NPV *ptp*, which are the only two viral genes encoding proteins with RNA triphosphatase activity did not alter viral RNA processing, it was proposed that host mRNA-capping enzymes might be involved in viral RNA maturation ([@bib21]). Considering the RNA triphosphatase activity that BmPTP-h possibly possesses, it could also be hijacked by virus for efficient viral RNA synthesis. Identification of host factors important for viral infection, especially for viral replication have been carried out using *Drosophila* and human cells by genome-wide RNAi screening ([@bib8], [@bib20]). Those studies reported that molecules involved in kinase-regulated signaling, ubiquitination and phosphatase activity are enriched in host--pathogen interaction network. However, the mechanism on how virus induces the expression of those "benefit" factors in host is unclear. One hypothesis is that production of viral proteins down-regulates the repressors of those factors in host, such as microRNA, resulting in increased expression of molecules favoring viral replication ([@bib7]). A few studies suggested that baculovirus infection leads to an overall suppression of host cellular microRNA ([@bib25], [@bib33]), whether there is a microRNA targeting *Bmptp-h* needs further identification. It is interesting to note that besides BmNPV *ptp*, several other baculoviral auxiliary genes are also acquired from host insect genome via horizontal gene transfer (HGT), such as ecdysteroid UDP-glucosyltransferase (*egt*) and fibroblast growth factor (*vfgf*) ([@bib16]). Although they are dispensable for virus production in cell line, they are reported to be involved in controlling host physiology. No direct evidence is available showing that BmPTP-h is also playing roles in regulating silkworm wandering behavior, however our study demonstrated that it promotes viral replication, facilitating rapid spread throughout the whole body of the silkworm to overwhelm the immune system. In contrast to this "collaboration" strategy employed by baculovirus in the case of *BmPTP-h*, HGT of host genes usually allows pathogen to compete with hosts for the molecules necessary to maintain physiological integrity or dampen host response by acting as dysfunctional mimics ([@bib24], [@bib39]). Kamita et al. showed inserting *Bmptp-h* into the BmNPV *ptp*-deleted virus rescued the inability to induce ELA, but to a much lesser extent than WT virus, suggesting that the modification on BmNVP *ptp* sequence during evolution may have provided selective advantages and drive their fixation in the viral genome to counteract the host defense. In conclusion, we did a whole genome survey for silkworm PTP gene sets and characterized their homologous relationship with human and fruit fly PTPs. We also found BmPTP-h acts as a host factor facilitating baculovirus infection via promoting its DNA replication. Further understanding of the pathways, in which BmPTP-h participates will provide new insight into the host--virus interactions that control the viral replication. This research was supported by grants of "National Basic Research Program of China" (No. 2012CB114600) and "Natural Science Foundation of China" (No. 31201854). [^1]: F: forward, R: reverse, Flag tag sequence is underlined. [^2]: Information of BmPTP-h is highlighted in bold. [^3]: Human MTMR9 has been proved to be a catalytic inactive phosphatase. [^4]: No clear orthology relationship could be ascribed to two silkworm atypical PTPs (101744063, 101742480).

Toward precise genomic therapies {#Sec2} ================================ From both public health and economic perspectives, lack of treatment response is one of the most pressing health care concerns that exist today. According to a US Food and Drug Administration (FDA) report from 2013, an astounding 38--75% of patients who have a common disease do not respond to treatment. Addressing this problem will be particularly important in increasing quality of life, decreasing mortality, and in turn, reducing health care costs globally. As an example, a 10% reduction in the number of patients who are not responding adequately to treatments for diabetes and heart disease in the USA would save approximately \$200 billion annually \[[@CR1]\]. Fundamental to tackling the issue of differential treatment response is the molecular complexity of common diseases, the scale of which can be vast; a disease may involve thousands of genes across multiple cell types in different parts of the body. In general, tools for clinical diagnostics do not allow for the assessment of such broad-scale changes. Thus, there is a wide gap between the molecular complexity and much of the translational research and clinical practice. Genomic medicine is emerging as a potential solution to bridge this gap \[[@CR2]\]. Fundamental to this concept is the use of genome-wide analses to match precisely an individual patient's molecular alterations with a drug targeting those alterations. This strategy is already being clinically implemented to treat rare monogenic diseases \[[@CR3]\]. In cancer, the focus has mainly been on somatic genetic alterations. The development of increasingly sophisticated analyses of different molecular layers has, however, led to interest in other forms of omics analyses, such as analyses of the transcriptome, epigenome, or metabolome. These promising developments have led to several efforts aimed at implementing genomic medicine in clinical practice for early disease detection or molecularly tailored treatments. Examples include Obama's Precision Medicine Initiative and several other international projects \[[@CR2]\]. Nevertheless, concerns about the effectiveness of genomic medicine have led to questions about what genomic medicine has achieved to date, unresolved problems, and potential solutions. Here, we discuss some clinically relevant examples and what can be learned from early attempts to implement genomic medicine in clinical practice. In summary, we propose that the clinical translation of genomic medicine will require the integration of multi-omics and clinical data and that emerging technologies such as single-cell approaches may facilitate the construction of high-resolution disease models. It is also possible that the complexity of diseases and technologies, coupled with the rapid speed of developments, will require the re-organization of health care structures in order to enable joint efforts to identify, evaluate, and implement new solutions for genomic medicine quickly. Promise and pitfalls of genomic medicine, focusing on genetic alterations {#Sec3} ========================================================================= An early, successful example of a drug that targets somatic genetic alterations was imatinib, a tyrosine kinase inhibitor that is used for the treatment of chronic myeloid leukemia patients who have a translocation that results in the formation of a constitutively active BCR-ABL fusion kinase \[[@CR2]\]. Other successful examples include the treatment of lung cancers harboring mutated epidermal growth factor receptor (EGFR) with EGFR inhibitors, and the treatment of melanomas bearing mutated BRAF with BRAF inhibitors \[[@CR2]\]. These successful examples have spurred concerted efforts to generalize genomic approaches to the treatment of disease, including those based at the National Cancer Institute (NCI), the National Institutes of Health (NIH), and several European and Asian centers. An ideal outcome could be routine sequencing of all tumors so that actionable genetic alterations could be identified and prioritized for treatment, but the clinical relevance of this paradigm has been questioned. Currently, most cancer patients do not have actionable genetic alterations \[[@CR2]\]. The SHIVA trial found no significant improvement in survival for cancer patients treated on the basis of actionable genetic alterations compared with those treated according to physician's choice \[[@CR4]\]. Possible explanations could be that those alterations did not have driving roles or that the patients also had other, untargeted alterations. Thus, from a gene-centered therapeutic perspective, the improved selection of targets or the use of combinatorial treatments is needed. It has also been argued, however, that the most effective current cancer treatments do not target mechanisms that are related to genetic alterations \[[@CR2], [@CR5]\]. This has led to increasing interest in including sources of information beyond genetic alterations. Multi-omics and multi-cellular approaches in the clinical translation of genomic medicine {#Sec4} ========================================================================================= In a recent study of brain tumors, genome-wide DNA methylation patterns were shown to agree with histopathological classification and led to the reclassification of some patients \[[@CR6]\]. There are also examples of diagnostic transcriptomics for cancer and complex diseases \[[@CR7]\]. It is reasonable to hypothesize that the integration of the most predictive markers from different omics layers will have greater potential than the assessment of genetic alterations alone \[[@CR8]\]. In this context, we should mention that there are vast resources of multi-omics data in the public domain that can be mined for translational purposes \[[@CR9]\]. Another important aspect is that most diseases involve multiple pathogenic mechanisms that are dispersed in multiple cell types, which may require combinatorial treatments. Several such examples are well established today. Asthma patients are commonly treated with inhalants that combine glucocorticoids and bronchodilators. In cancer, checkpoint inhibitors that increase immune responses complement drugs that directly target tumor cells. These examples highlight a potential limitation of many genomic analyses in finding diagnostic and therapeutic targets: the analyses are performed on bulk samples. Instead, we should ideally perform simultaneous genome-wide analyses of each individual disease-associated cell type, so that key regulatory cell types and mechanisms can be prioritized and targeted. The emergence of genome-wide methods for single-cell analyses has been proposed as a potential solution \[[@CR7]\]. A large number of studies on cancer and complex diseases support this potential, and although there are currently multiple technical limitations, it is possible that rapid developments in single-cell analyses will result in high-resolution multicellular disease models for drug development and even diagnostics \[[@CR7]\]. Integration of omics and clinical data {#Sec5} ====================================== Recent genomic and epigenomic studies of cancer have supported the diagnostic potential of integrating omics and histopathological data \[[@CR6], [@CR10]\]. There are now several examples of public and private efforts with similar aims that have acquired omics, clinical, and imaging data from hundreds of thousands of patients \[[@CR2]\]. The DNA methylation study also presented a web-based diagnostic tool that can be used for standardized diagnostics in clinics across the globe \[[@CR7]\]. Importantly, the use of this tool in clinical practice will result in the continuous improvement of the tool and potentially in an improved understanding of pathogenic mechanisms that result in disease, as well as of the diagnostics and therapeutics. The study may serve as a model for how different combinations of local or centralized wet lab and computational analyses may contribute significantly to the clinical implementation of genomic medicine. Challenges and opportunities for research and clinical practice {#Sec6} =============================================================== The diagnostic and therapeutic success stories mentioned above support the potential for translating genomic medicine to the clinic. Nevertheless, there is a wide gap between the complexity of the diseases and current clinical practice. Bridging that gap involves addressing several different challenges: from a research perspective, the clinical translation of genomic medicine requires the integration of multiple disciplines, such as omics, epidemiology, bioinformatics, and experimental and clinical research. Such integration is complicated by the compartmentalization of academic and industrial research. A related problem for funders and medical journals is the evaluation of multi-disciplinary research when most reviewers have expertise only in some of the disciplines involved. From a health care perspective, the rapid development of multiple solutions for genomic medicine is likely to make the identification, evaluation, and prioritization of "actionable" diagnostic and therapeutic solutions important challenges to be addressed urgently. Next, the clinical implementation of successful solutions may require new technologies as well as the training of health professionals. This could result in the development of novel clinical disciplines, derived from integrating genomics, systems medicine, and bioinformatics. Taken together, these challenges and opportunities may require new organizational solutions in academia and health care. As an example, the web-based analyses of DNA methylation data from brain cancers suggest that novel combinations of local and centralized solutions for diagnostics may become optimal. This could lead to new roles for health care units that might serve increasingly as dynamic integrators and providers of local and global services---as opposed to focusing on the local accumulation of expertise and technologies. Such a development would require new solutions for the continuous education of health professionals, as well as organizational flexibility to optimize the integration of resources. Conclusions {#Sec7} =========== The huge medical, economic, and societal problems arising from the large number of patients who do not respond to therapy emphasize the need for novel diagnostic and therapeutic solutions for the management of patients. Genomic medicine has considerable potential, but also presents significant challenges that are likely to be addressed by joint international initiatives. This work was supported by the Swedish Research Council, The Cancer foundation, IMI project RTCure, Swedish Rheumatisms Foundation, German Childhood Cancer Foundation (Deutsche Kinderkrebsstiftung), Brain Tumor Charity (BTC), Federal Ministry for Education and Research (BMBF), and the German Cancer Consortium (DKTK). MB and HZ wrote the manuscript draft, while SMP focused on malignant diseases and LK on complex diseases. AM focused on laboratory aspects of genomic medicine. All authors read and approved the final manuscript. Competing interests {#FPar1} =================== The authors declare that they have no competing interests. Publisher's Note {#FPar2} ================ Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Introduction ============ Osteoporosis is a relevant age related disorder, characterized by low bone mass and increased bone fragility putting the patient at risk for fractures \[[@B1]\]. Currently osteoporosis is viewed as a heterogeneous condition which can occur in any age of life and its etiology is attributed to various endocrine, metabolic and mechanical factors (abnormalities of parathyroid hormone and calcitonin secretion, insufficient vitamin D and calcium intake, postmenopausal hormonal condition, pregnancy, nutritional disorders, immobility and consumption of drugs such as cortisone, among others) \[[@B2]\]. Recently, growing understanding of the bone remodelling process suggests that factors involved in inflammation are linked with those critical for bone physiology and remodelling, supporting the theory that inflammation significantly contributes to the aetiopathogenesis of osteoporosis \[[@B3],[@B4]\]. Is osteoporosis an inflammatory process? ======================================== Clinical observations reveal coincidence of systemic osteoporosis with period of systemic inflammation as well as co-localization of regional osteoporosis with areas of regional inflammation \[[@B2]\]. Different epidemiologic studies report an increase in the risk of developing osteoporosis in various inflammatory conditions \[[@B5]-[@B8]\]. Immunological dysfunctions, autoimmune and chronic inflammatory diseases \[[@B9]\], HIV infection \[[@B10]\], hyper-IgE syndrome \[[@B11]\], rheumatoid arthritis \[[@B12]\], haematological diseases, particularly myeloma \[[@B13]\], and inflammatory bowel diseases \[[@B14]\], are associated with osteoporosis. Erosions seen in conditions such as gout, osteomyelitis, rheumatoid arthritis, ankylosing spondylitis, and psoriatic arthritis, are typically associated with inflammation in the joints. Pro-osteoclastic cytokines, such as tumour necrosis factor (TNF)-α and interleukin (IL)-6, are elevated in these conditions and local cytokine profile is consistent with the cytokines that modulate bone resorption \[[@B15],[@B16]\]. C-reactive protein (CRP) production in the liver is upregulated by IL-1, IL-6 and TNF-α, and is regarded as a sensitive marker of systemic inflammation \[[@B17],[@B18]\]. An association between circulating high sensitive (hs)CRP level and bone mineral density has been observed in several immune and inflammatory diseases, as well as in healthy individuals, suggesting a relationship between subclinical systemic inflammation and osteoporosis \[[@B19]\]. Rheumatoid arthritis (RA) is a typical example of the link between inflammation and osteoporosis. Bone loss in RA occurs both in the joints and throughout the skeleton as a result of the release of proteinases (metalloproteinases) and proinflammatory cytokines (IL-1, TNF-α), which are responsible for cartilage and bone destruction. As a result, disease activity is an independent risk factor for osteoporosis in RA \[[@B20]\]. A temporal link between inflammation and osteoporosis also emerges in conditions such as ageing, menopause, pregnancy, transplantation and steroid administration. While nutritional, mechanical and hormonal factors clearly play a role in many of these situations, the concordance of osteoporosis and inflammation is buttressed by emerging molecular evidence of mediating immunological factors \[[@B2]\]. On the other hand, one intriguing aspect of immunosenescence is the increased production of pro-inflammatory cytokines with age and a close link between age-related systemic inflammatory process (inflamm-ageing)\[[@B21]\] and osteoporosis is well documented \[[@B22]\]. A molecular scenario of immune regulated bone loss ================================================== Although osteoporosis is not typically considered an immunological disorder, recent data have indicated overlapping pathways between bone biology and biology of inflammation \[[@B23]-[@B25]\]. Certain pro-inflammatory cytokines play potential critical roles both in the normal bone remodelling process and in the pathogenesis of perimenopausal and late-life osteoporosis \[[@B26]\]. For example, interleukin (IL)-6 promotes osteoclast differentiation and activation \[[@B27]\]. This cytokine is involved in the pathogenesis of various metabolic bone diseases, including postmenopausal osteoporosis, Paget\'s disease and osteoporosis associated with heamtologic malignancies \[[@B28]\]. IL-1 is another potent stimulator of bone resorption \[[@B29]\] that has been linked to the accelerated bone loss seen in idiopathic and postmenopausal osteoporosis \[[@B30]\]. TNF-α is implicated in tumour-induced bone resorption and non-tumour-induced osteopenia \[[@B31]\]. Anti-TNF drugs, currently used in the therapy of several immunological disorders, are also useful in preventing and/or reversing systemic bone loss associated to the disease, as they target both the bone and the inflammatory process \[[@B20]\]. The production of IL-1, IL-6, and/or TNF-α by peripheral blood monocytes has been positively correlated with bone resorption or spinal bone loss in healthy pre- and postmenopausal women \[[@B32]\]. The inflammatory mediator nitric oxide (NO) is also involved in the pathogenesis of osteoporosis. The activation of the inducible NO synthesis (iNOS) pathway by cytokines, such as IL-1 and TNF-α, inhibits osteoblast function in vitro and stimulates osteoblast apoptosis \[[@B33]\]. The TNF-family molecule RANKL and its receptor RANK (receptor activator of NFkB ligand) have been specifically implicated in the bone loss in rheumatoid arthritis \[[@B34]\]. They are key regulators of bone remodelling and are essential for the development and activation of osteoclasts. Intriguingly, RANKL/RANK interactions also regulate T cell/dendritic cell communication, dendritic cell survival and lymphnode formation \[[@B35]\]. Calciotropic factors such as vitamin D3, PGE2, IL-1, IL-11, TNF-α and glucocorticoid induce RANKL expression on osteoblasts \[[@B36]\]. RANKL binding to the RANK expressed on haematopoietic progenitors activates a signal transduction cascade that leads to osteoclast differentiation. Moreover, RANKL stimulates bone resorbing activity in mature osteoclasts via RANK. When RANK is activated, it sends signals into the cells through tumor necrosis factor receptor-associated factors (TRAFs), mainly TRAF6. These RANK associated molecules, through downstream pathways such as NF-kB, JNK/SAPK, p38 and Akt/PKB, regulate bone resorption, activation, survival, and differentiation of osteoclasts and dendritic cells. Osteoprotegerin (OPG), also known as osteoclastogenesis inhibitory factor, functions as a soluble decoy receptor to RANKL and competes with RANK for RANKL binding. TGFβ released from bone during active bone resorption has been suggested as a feedback mechanism by upregulating OPG level. Estrogen can enhance OPG production on osteoblasts which is a possible explanation of postmenopausal osteoporosis following estrogen withdrawal \[[@B35]\]. Since the expression of RANKL/RANK can be controlled by sex hormones, it is possible to speculate that this system may control gender specific differences in immunity and could be involved in the higher incidence of autoimmune diseases and osteoporosis in women. RANK-L, RANK and OPG therefore configure interesting molecular links between bone remodelling, immunity and inflammation. The emergence of osteoimmunology ================================ There exists an intimate interplay between the bone and the immune system which has been termed osteoimmunology \[[@B3]\]. The activity of immune cells affects the balance of bone mineralization and resorption carried out by the opposing actions of osteoblasts and osteoclasts \[[@B37]\]. Dendritic cells, specialized to present antigens, and osteoclasts, specialized to resorb bone, share the same bone marrow precursors of the monocyte lineage and exhibit parallel lifecycles, regulated by a variety of cytokines, transcription factors and inflammatory mediators. Molecules that regulate osteoclastogenesis are key factors in many immunological functions. For example, TRAF6 functions as a molecular bridge spanning adaptive immunity, innate immunity and osteoimmunology \[[@B38]\]. TRAF6-deficient mice lack osteoclasts and concomitantly have defects in cytokine production and T cell stimulation \[[@B39]\]. Activated T cells affect bone physiology producing cytokines that lead to RANKL expression on osteoblasts. Moreover, activated T cells directly express and produce RANKL that induces osteoclast formation and activation through its specific receptor \[[@B28],[@B35]\]. There are multiple mechanisms and interactions by which cytokines regulate bone resorption. IL-6 contributes to RANKL upregulation in osteoblastic cells. IL-1 and TNF-α may not only promote osteoclast generation, but they also appear to stimulate mature osteoclasts to perform more resorption cycles through modulation of RANKL activity. IL-1 is further involved in bone metabolism as an osteoblast activator: osteoblasts secrete RANKL which promotes survival and differentiation of the osteoclast precursors to mature osteoclasts through RANK. IL-1 and IL-6 also directly enhance osteoclast activity by RANKL-independent mechanisms. They may directly extend the lifespan of the osteoclasts by inhibiting osteoclast apoptosis. Both TNF-α and IL-1 inhibit collagen synthesis in osteoblasts and enhance degradation of the extracellular matrix \[[@B18]\]. In inflammatory or autoimmune disease states, activated T cells produce RANKL and pro-inflammatory cytokines, all of which can induce RANKL expression in osteoblasts \[[@B40]\]. However, the constant activity of T cells fighting the universe of antigens to which we are exposed does not usually cause extensive bone loss. Multiple T cell-derived cytokines, such as IL-12 and IL-18, might be able to interfere with RANK signalling and therefore with osteoclastogenesis and osteoclast functions. A crucial counter-regulatory mechanism whereby activated T cells can inhibit osteoclast development and activation is through the action of the antiviral cytokine interferon (IFN)-γ. Mechanistically, IFN-γ activates the ubiquitin-proteasome pathway within the osteoclasts, resulting in the degradation of TRAF6 \[[@B38]\]. Ageing and osteoporosis: immunological links ============================================ Inflamm-ageing may, at least partially, be a common mechanism for the development of lower bone mass and other age-related disorders \[[@B41],[@B42]\]. Senility is in fact notable for acceleration of diseases that are increasingly attributed to inflammation, such as atherosclerosis, Alzheimer\'s disease and asthma \[[@B21],[@B43],[@B44]\]. Many cytokines, including IL-6, TNF-α, IL-1, are elevated during senescence and play direct roles in the pathogenesis of these diseases \[[@B45]\]. All these cytokines are stimulators of osteoclast activity as well. Associations between atherosclerosis and osteopenia have been well documented \[[@B46],[@B47]\]. These findings suggest a potential causal relationship between systemic inflammation that is observed in the elderly and the prevalence of generalized age-related osteoporosis \[[@B2]\]. Moreover, the increased catabolic signal, driven by inflammation also in the absence of clinically diagnosable inflammatory diseases \[[@B44]\], could be able to induce osteoblast apoptosis \[[@B48]\], as well as apoptosis of muscle cells, leading to age-related osteoporosis and sarcopenia respectively. During ageing, under the influence of the lifelong exposure to chronic antigenic load and oxidative stress, the physiological counter-regulatory process which inhibits bone resorption following T cell activation is likely impaired. This would contribute, together with the age-related systemic low-grade inflammation, to the increasing incidence of osteoporosis during senescence. Osteoporosis, as well as other age-related disorders, has a strong genetic component and the rate extent of bone loss during senescence varies widely between individuals. It is tempting to postulate that this may be in part due to individual differences in cytokine activity. In support of this hypothesis it has been demonstrated that certain IL-6 polymorphisms are able to influence the risk of osteoporosis in postmenopausal women \[[@B49]\]. Similarly, IL-1β and IL-1 receptor antagonist (IL-1Ra) gene polymorphisms are associated with reduced bone mineral density and predispose women to osteoporosis at the lumbar spine \[[@B50]\]. Moreover, also the sex hormonal decline which accompanies ageing contributes to the pathogenesis of senile osteoporosis through immunologically mediated mechanisms. It has been postulated that estrogens exert their effect on bone not only by direct action per se, but also by inhibiting IL-6 gene expression. A similar relationship between androgen and IL-6 gene expression also exists \[[@B51]\]. The decline in ovarian function is associated with decreased OPG production and spontaneous increases in proinflammatory and pro-osteoclastic cytokines such as IL-6, TNF-α, and IL-1 \[[@B32],[@B52],[@B53]\]. Osteoporosis within an evolutionary perspective =============================================== During evolution of multi-cellular organisms, the increasing complexity of emerging systemic functions, such as inflammation, may have led to a widening range of demand for key minerals such as calcium and phosphate and storage functions may have evolved to provide mineral reservoirs \[[@B2]\]. Moreover, calcium is an important component of milk and its transport from mother to the fetus and neonate is a vital process to preserve species. Intriguingly, RANKL and RANK also play essential roles in the formation of a lactating mammary gland in pregnancy. This could partly contribute to both the immune remodelling and the accelerated bone loss during pregnancy and lactating \[[@B35]\]. Bone can thus be viewed as a mobilizable reservoir of calcium and phosphate as salts. Over time, the regulation of bone turnover was probably optimized evolutionarily to address the combined metabolic and structural demands of the host. In this perspective osteoporosis may reflect a state of disequilibrium between structural demand for calcium and phosphate and their biologic demand during metabolically active states such as inflammation. Inflammation is the leading force driving immunosenescence and the chronic low-grade inflammatory state characteristic of ageing represents the predisposing substrate on which osteoporosis, as well as other age-related diseases, might emerge \[[@B21],[@B54]\]. The lack of evolutionary selective pressure post-reproduction may have further exacerbated this disequilibrium in modern times and many biologic functions acquired during evolution have become maladaptive. Presently, in most developed countries, the human lifespan is greatly increased, and many individuals are living into post-reproductive senescence, an evolutionarily naive life epoch. Obesity, diabetes, Alzheimer\'s disease and atherosclerosis are examples of diseases of modernity that are attributable to modern living circumstances or that are unmasked during senility, and also the emergence of osteoporosis as a modern disease may be an example of this phenomenon \[[@B2],[@B21],[@B44]\].

Occupational health physicians (OHPs) have faced difficulty in projecting their performance in a comprehensive manner that is easily understood by the management. OHPs have been performing but stored the data in different silos that are not interlinked in a dashboard. For a young OHP, this an area of concern, and the occupational health index (OHI) can serve as the best option to showcase the occupational health performance. At present, OHI is being used as a reflection of quality of deliverance of occupational health in many multinational organizations across the globe. The need is to customize the OHI content to meet the organizational and regional requirements. Through this editorial I am trying to bring the concept to the working table, and each OHP can then work towards a suitable OHI matrix for his/her organization. The genesis of OHI is based on some of the topics covered in OHSAS: 18001:2007Competence, Training, and Awareness (4.4.2)Emergency Preparedness, and Response (4.4.7)Performance Measurement, and Monitoring (4.5.1)Evaluation of Compliance (4.5.2)ISO 14001:2004Environmental Aspects (4.3.1)Company EHS & S standards and guidelines. The OHSAS: 18001:2007, ISO14001:2004, and Company EHS & S standards cover macrolevel aspects of occupational health components, and hence, the need of OHI to elaborate on specific components in greater details. How one defines OHI? OHI is a composite matrix that reflects occupational health performance in an organization. It is the logical extension of OHSAS: 18001 and ISO 14001 with detailing of key performance indicators. OHI is strictly in alignment with local legislations and organizational EHS standards. When one wants to design the OHI matrix, one should consider the major quality components of occupational health services such as compliance, quality, satisfaction score, efficiency score, and culture of health (policies and programs). Some combine wellness with occupation health and many measure them separately; we advocate the combined approach for OHI. In this manner, one can decide on subcomponents within each of the above headings that are critical to the organization. Just to name some such as management review, occupational health surveillance, biomedical waste management, medical records, occupational health center inspection (supported by a relevant checklist), international medical care and workability management, emergency medical response, project site and field employees, employee and stakeholders survey score, occupational health professional development, travel health, water and food safety. In the culture health basket, one can use all the defined policies and programs as subcomponents of measurement such as "EAP and mental health, health protection and promotion, healthy diet, and physical movement." The most challenging task is designing the scoring framework that has group and subgroup components and scoring checklist with weightage. The scoring framework gives a consolidated score for each subgroup and main group key performance indicators that one can color band as "red (bellow 50), orange (51--70), yellow (71--89), green (above 90), and gray when not applicable." Detailing of subgroup components is beyond the scope of this editorial, however, an attempt is made to provide a platform for OHPs, especially the younger lot who are keen to showcase their occupational health performance to their top management through OHI dashboard. Thus, the OHI can be a resource in: OHI is a quality tool for measuring occupational health performanceOHI matrix can be designed to suit organizational needsOHI supports business continuity planRegional benchmarking in occupational healthOHI can be integrated with EHS & S dashboard. The editor expects that some of you will work toward creating the OHI dashboard for your organization and share your experience with us. There can also be diverse ways of creating a performance dashboard for occupational health and expect innovations storming the knowledge market.

During July 2015--May 2016, a mumps outbreak occurred at the University of Iowa, which is located in Johnson County ([@R1]). A total of 301 cases of mumps were diagnosed among students. To characterize the outbreak, the Johnson County Public Health Department, the Iowa Department of Public Health, and the University of Iowa, with assistance from CDC, conducted an investigation through telephone interviews, medical chart abstractions, and review of immunization records. Among 287 (95%) students with mumps for whom clinical information was available, 20 (7%) patients with complications were identified (16 self-reported and four clinician-diagnosed). The 20 cases included 15 (5%) cases of orchitis, three (1%) of transient hearing loss, two of mastitis, and one of meningitis (one patient had both orchitis and transient hearing loss). All 20 patients had documentation of receipt of at least 2 doses of measles-mumps-rubella vaccine. Because data are limited regarding the presentation and clinical course of mumps complications in persons who have received 2 doses of mumps-containing vaccine, three illustrative cases of complications (orchitis, transient hearing loss, and meningitis) in students with mumps are presented. Patient A ========= On November 17, 2015, a man aged 21 years developed right jaw pain and swelling and received a clinical diagnosis of mumps parotitis; the diagnosis occurred 2 weeks after his roommate had received a mumps diagnosis. By the ninth day after symptom onset, the patient's parotitis had resolved, but he reported a fever of 101.0°F (38.8°C), and 2 days later, he developed left testicular pain and swelling. Orchitis was diagnosed and he was prescribed nonsteroidal anti-inflammatory drugs and ice packs and had no further follow-up care. Patient B ========= On October 13, 2015, a woman aged 21 years developed progressive right ear pain, cough, and shortness of breath. Two days later, she was treated in a hospital emergency department where she received a diagnosis of right otitis externa and left otitis media, for which she was prescribed amoxicillin and analgesics. Later that day, she went to the University of Iowa Student Health Center because of worsening respiratory symptoms. During that encounter, she was also noted to have right bullous myringitis (purulent inflammation of the tympanic membrane), right parotitis suspected to be mumps, and suspected pneumonia. Azithromycin was prescribed empirically to treat both the bullous myringitis and atypical pulmonary pathogens. A polymerase chain reaction (PCR) test for mumps was performed on a buccal swab specimen and was negative. However, her symptoms and epidemiologic link to the outbreak met the Council of State and Territorial Epidemiologists case definition for a probable case of mumps. One day later, she noticed tinnitus and diminished hearing in her right ear; on day 8, she had audiology testing and was evaluated by an otolaryngologist, at which time she received a diagnosis of moderate right sensorineural hearing loss, attributed to mumps, and conductive hearing loss, attributed to otitis media and myringitis. She was treated for 1 week with prednisone, and all her symptoms resolved by the thirteenth day after onset of parotitis. No repeat audiology testing was performed. Patient C ========= On November 2, 2015, a man aged 21 years developed left facial pain and swelling and tested positive for mumps by PCR on a buccal swab specimen. Twenty-two days after onset of symptoms, he was treated at an emergency department for neck stiffness, fever, and tachycardia. A lumbar puncture was performed, and he was empirically treated for meningitis with acyclovir and ceftriaxone. Volume of cerebrospinal fluid was inadequate for performing PCR testing for mumps, but Gram stain and bacterial culture were negative, and analysis was consistent with viral meningitis (40 lymphocytes/mm^3^, 60 mg/dL of protein, and 67 mg/dL of glucose). Because the onset of mumps-related meningitis has been described as ranging from 4 days before the onset of parotitis until 3 weeks after ([@R2]), the patient's viral meningitis diagnosis was attributed to mumps. He was discharged with recommendations for symptomatic care, and meningeal symptoms resolved within 1 week. Complications of mumps have been reported less frequently since licensure and widespread use of mumps-containing vaccines. However, this case series demonstrates that complications still occur, even in persons who have received the recommended 2 doses of measles-mumps-rubella vaccine. In addition, complications can occur at varying times throughout the course of the illness and in the absence of parotitis ([@R2],[@R3]). Health officials should remain vigilant for these complications and their relation to mumps, and when mumps is suspected, conduct PCR testing on a buccal swab specimen and serology on a serum specimen ([@R4],[@R5]).

Introduction {#s1} ============ Trenbolone (Tren) belongs to the class of synthetic anabolic-androgenic steroids (AAS) and is structurally characterized by a 4,9,11-triene-3-one structure composing a highly conjugated π-electron system ([Figure 1](#F1){ref-type="fig"}). The significant anabolic properties of Tren resulting in increased muscle size and strength have generated an incentive for illicit applications including doping and livestock breeding. In sports, the use of trenbolone has been prohibited by the World Anti-Doping Agency (WADA) at all times, categorized under S1 1. (anabolic androgenic steroids) in the Prohibited List (WADA, [@B41]). According to WADA\'s annual statistics, anabolic agents are the most frequently misused substance group in sports with a total of 1,823 adverse analytical findings (AAFs) in 2018. Within this group, Tren occurrences account for 6% (WADA, [@B40]). The statistics however can only reflect cases of detectable Tren and does not conclusively address the question whether Tren is less favored by users of AAS or if available detection strategies do not offer the required analytical retrospectivity. {#F1} For that reason the objective of this project was to re-investigate the metabolism of Tren in order to probe for metabolic products potentially supporting the extension of the detection window. The number of previously reported Tren metabolites is scarce, and for doping control purpose the analysis focuses at present on the main human urinary metabolites epitrenbolone (EpiTREN), epitrenbolone glucuronide (EpiTREN Glu), and trenbolone glucuronide (TREN Glu) (De Boer et al., [@B6]; Schänzer, [@B29]). Regarding the detection windows, two data sets have been published spanning from approximately 3 days (Spranger and Metzler, [@B31]) to 32 days (Sobolevsky and Rodchenkov, [@B30]). Besides glucuronides, sulfates (Rzeppa et al., [@B27]), and cysteine conjugates (Sobolevsky and Rodchenkov, [@B30]) of Tren and Epitren were reported. Cysteine conjugates are produced by phase-II metabolism, where the tripeptide glutathione is covalently bound via the sulfur atom by glutathione transferase and, subsequently, glutamate and glycine are eliminated. In general, cysteine conjugates are hydrolyzed employing alkaline conditions (Blair, [@B2]; Fabregat et al., [@B9], [@B10], [@B8]; Pozo et al., [@B26]), but the cysteine conjugate of trenbolone described by Sobolevsky and Rodchenkov ([@B30]). was found to be stable during alkaline hydrolysis and was analyzed by high performance liquid chromatography electrospray ionization tandem mass spectrometry (HPLC-ESI-MS/MS) as in-source fragment in ESI negative mode and as intact conjugate in ESI positive mode (Sobolevsky and Rodchenkov, [@B30]). During *in-vitro* studies, several monohydroxylated metabolites, and trenbolone-diketone were generated (Metzler and Pfeiffer, [@B18]; Kuuranne et al., [@B14]). Nowadays, LC-MS-based methods are commonly used for the analysis of Tren and its metabolites (Thevis et al., [@B34], [@B37]; Tudela et al., [@B39]) as GC-MS-based methods were found to be of limited utility due to derivatization artifacts and low thermal stability of the target analytes (De Boer et al., [@B6]; Ayotte et al., [@B1]; Casademont et al., [@B4]; Marques et al., [@B17]; Brun et al., [@B3]). For systematic metabolism studies, a method for metabolite identification using hydrogen isotope ratio mass spectrometry was developed and successfully applied for the first time in 2013 (Thevis et al., [@B38]). The fundamental principle is analogous to metabolism studies using radioactively labeled compounds (Sano et al., [@B28]). The compounds can be detected selectively because of their isotopic labeling by measuring the radioactivity in case of tritium or ^14^C labeled compounds or by measuring the hydrogen isotope ratios in case of deuterium labeled compounds. Hydrogen isotope ratios are determined by gas chromatography/thermal conversion/isotope ratio mass spectrometry (GC-TC-IRMS). The organic compounds are converted under reducing conditions to CO and N~2~ as well as molecular hydrogen (H~2~), respectively the deuterated isotopologe HD. After ionization, detection is accomplished using *m*/*z* 2 for $\text{H}_{2}^{+}$ and *m*/*z* 3 for HD^+^ by Faraday cups with different amplification factors (factor 1000 difference). Since the natural hydrogen abundance amounts on average to 99.985% for H and 0.015% for D (Dunn and Carter, [@B7]), comparable signals for $\text{H}_{2}^{+}$ and HD^+^ are obtained for samples at natural abundance, while deuterated compounds lead to a significant increase of signals at *m*/*z* 3. Compounds resulting in diagnostic HD^+^ signals are subsequently comprehensively characterized by gas chromatography/electron ionization/high accuracy/high resolution mass spectrometry (GC-EI-HRMS). This concept has been proven in several studies (Thevis et al., [@B38]; Piper et al., [@B24],[@B25], [@B20], [@B21]). Within this project, liquid chromatography/electrospray ionization/high accuracy/high resolution mass spectrometry (LC-ESI-HRMS) was applied for further characterization of trenbolone and its metabolites after GC-TC-IRMS analysis. Twenty metabolites were identified with a detectability of up to 6 days. Four metabolites exhibiting the longest detection windows were characterized by parallel reaction monitoring (PRM) experiments and comparison to reference material. Materials and Methods {#s2} ===================== Chemicals and Steroids ---------------------- Trenbolone reference material and the internal standard 2,2,4,6,6-d~5~-trenbolone were purchased from Toronto Research chemicals (Toronto, Canada), and epitrenbolone from the National Measurement Institute (Sydney, Australia). Steroid reference material for HPLC separation including ETIO (etiocholanolone), A (androsterone), T (testosterone), and PD (pregnanediol) was supplied by Sigma-Aldrich, and 11 K (11-ketoetiocholanolone), 5a (5α-androstanediol), and 5b (5β-androstanediol) were obtained from Steraloids (Newport, RI). Chromabond C18 solid-phase extraction (SPE) cartridges (500 mg, 6 mL) were purchased from Macherey & Nagel (Düren, Germany) and β-glucuronidase from Escherichia coli (140 U/mL) from Roche Diagnostics (Mannheim, Germany). Ultrapure water was prepared by a Barnstead™ GenPure™ xCAD Plus system (Thermo, Germany). All reagents and solvents were of analytical grade. Acetonitrile (ACN), formic acid (FA), methanol (MeOH), tert-butyl methyl ether (TBME), cyclohexane, pyridine, sodium hydroxide (NaOH), sulfuric acid (H~2~SO~4~), glacial acetic acid, and potassium tri-sec-butylborohydride (1 M in THF) were provided by Merck (Darmstadt, Germany). Acetic anhydride was supplied by Sigma Aldrich (Taufkirchen, Germany) and tetrahydrofuran (THF) by VWR (Darmstadt, Germany). *N*-methyl-*N*-trimethylsilyltrifluoroacetamide (MSTFA) was purchased from Chemische Fabrik Karl Bucher (Waldstetten, Germany). Excretion Study --------------- An excretion study was conducted in order to re-investigate the trenbolone metabolism. Following written informed consent, 10 mg of 5-fold deuterated trenbolone ([Figure 1](#F1){ref-type="fig"}) dissolved in ultra-pure water/EtOH (80:20, *v*/*v*) was orally administered to one healthy male volunteer (43 years, 84 kg) who declared not to have used any medication or nutritional supplements during this study and for a wash-out period of at minimum 3 month (any compounds), respectively 6 month (deuterated compounds) before the study. Three blank samples were collected pre-administration, and post administration samples were collected for up to 30 days. During the first 48 h after trenbolone intake, every urine was collected. From day three until day five, two to three urine samples per day were collected, and afterwards only the first morning urine was sampled until the end of the study. The administration study was approved by the Ethics Committee of the National Institute of Sports of Romania (Bucharest, Romania, \#2283, 2016). Analysis of Hydrolyzed Steroids ------------------------------- ### Sample Preparation An extensive sample preparation was required in order to reach adequate purity and undecomposed volatility of the metabolites for GC-TC-IMRS analysis. Urine samples were prepared according to established protocols for isotope ratio analysis of steroids. Here, every sample is separated into four main fractions: unconjugated steroids, glucuronic acid conjugates (Piper et al., [@B22]), sulfo-conjugates (Piper et al., [@B23]), and cysteine conjugates (Fabregat et al., [@B9], [@B10], [@B8]; Pozo et al., [@B26]). Subsequently, the fractions of glucuronides and sulfates are further divided by HPLC into seven sub-fractions (Thevis et al., [@B38]). Sample preparation and analysis is summarized in [Figure 2](#F2){ref-type="fig"}. {#F2} For each sample, a volume of 20 mL urine was required. First, two C-18 SPE cartridges per sample were pre-conditioned with 2 mL of MeOH and subsequently washed with 2 mL of water. Then, every sample was split and 10 mL of urine were applied to each cartridge, which was subsequently washed with 2 mL of water, and finally eluted with 2 mL of MeOH. Both eluates of each sample were combined and evaporated to dryness under a gentle stream of compressed air at 50°C. Following evaporation, samples were reconstituted with 1 mL of an aqueous 0.2 M sodium phosphate buffer at pH 7, and a liquid-liquid-extraction (LLE) step with 5 mL of TBME was performed. Therefore, the mixture was shaken for 5 min and subsequently centrifuged for 5 min at 600 g before separating both layers. The organic layer yielded the fraction of the unconjugated steroids (fraction f; free). The remaining aqueous layer was incubated with 100 μL of β-glucuronidase at 50°C for 1 h. To terminate hydrolysis, 500 μL of an aqueous 20% potassium carbonate buffer (pH 10) were added. In order to extract the liberated steroids, a second LLE step with TBME was performed and the resulting organic layer contained the former glucuronic acid conjugates (fraction Gluc). Then, the pH of the aqueous layer was adjusted to 5 with glacial acetic acid, and purified by SPE as described above. After evaporation, samples were incubated with 2.5 mL of EtOAc/MeOH (70/30, *v/v*) and 1 mL of EtOAc/H~2~SO~4~ (100 mL/200 ng, *v/w*) at 50°C for 1 h. Subsequently, 0.5 mL of methanolic NaOH (1 M) were added, the mixture was evaporated as described above, and reconstituted with 5 mL of water. A third LLE with TBME was conducted to extract the formerly sulfo-conjugated steroids (fraction Sulf). This was followed by alkalization of the aqueous layer with 300 μL of 6 M KOH and an incubation step at 60°C for 15 min. Finally, alkaline-labile steroids comprising potential cysteine conjugates were extracted by the last LLE (fraction Cys). The four resulting TBME extracts per sample were evaporated to dryness and the fractions of glucuronides and sulfates further purified by HPLC. For that purpose, samples were reconstituted in 100 μL ACN/H~2~O (50/50, *v*:*v*), and the entire volume was injected into an Agilent 1100 HPLC-UV system (Waldbronn, Germany) equipped with a X Bridge Shield RP18 column (4.6 × 250 mm) with 5 μm particle size (Waters, Eschborn, Germany). UV signals were acquired at 195 and 360 nm. Gradient elution was conducted as follows with a flow rate of 1 mL/min: Starting at 20% ACN/80% water, the gradient increased to 100% ACN within 25 min, was held for 10 min, and re-equilibrated for 5 min. With support of a Foxy R1 automatic fraction collector (Axel Semrau, Sprockhövel, Germany), the following HPLC sub-fractions were produced using the retention time markers shown in brackets. I: 3.00--10.00 min, II: 10.01--13.50 min (Tren, Epitren), III: 13.51--14.80 min (T), IV: 14.81--17.00 min (EpiT, DHEA, 5b, 5a, ETIO), V: 17.01--19.50 min (PD), VI: 19.51--24.50, and VII 24.51--33.00 min (16 EN). The eluted HPLC sub-fractions were evaporated to dryness. Derivatization of all fractions derived from HPLC clean-up, as well as the fractions containing the cysteine adducts and free steroids was performed in accordance with the applied chromatographic system as described below. ### Derivatization Techniques #### Acetylation For acetylation, samples were reconstituted in 75 μL of pyridine and 75 μL of acetic anhydride and derivatized for 1 h at 70°C. Subsequently, the derivatization mixture was evaporated. Samples were subjected to GC-TC-IRMS/MS (section GC-TC-IRMS/MS Setup), GC-EI-HRMS (QTOF) (section GC-EI-HRMS (QTOF) Setup) and LC-ESI-HRMS (section LC-ESI-HRMS (LC Orbitrap) Setup) analysis. #### Trimethylsilylation For trimethylsilylation, samples were reconstituted in 80 μL of MSTFA:NH~4~:ethanethiol (1000:2:3, *v*:*w*:*v*), incubated at 60°C for 45 min (Mareck et al., [@B16], [@B15]), and measured as described in section GC-EI-HRMS (Orbitrap) Setup. ### Instrument Methods #### GC-TC-IRMS/MS setup After evaporation of the derivatization mixture, the acetylated samples were reconstituted in an appropriate volume of cyclohexane (typically 20 μL) for GC-TC-IRMS/MS analysis. Analysis was performed on a Delta V Plus IRMS coupled via a GC Isolink CNH for thermal conversion at 1,450°C with a ceramic reduction reactor and ConFlow IV to a Trace 1310 GC (Thermo, Bremen, Germany). Chromatographic separation was accomplished on a DB-17 MS column (30 m × 0.25 mm) with a film thickness of 0.25 mm. The temperature gradient was as follows: The temperature remained constant at 100°C for 1.5 min and increased with 40°C/min to 240°C and subsequently with 5°C/min to 320°C with a hold time of 2 min. Samples were injected in splitless mode at 300°C with an injection volume of 5 μL. A single taper inlet liner (900 μL volume, 4 mm inner diameter, 6.47 mm outer diameter, 78.5 mm length) with glass wool from Agilent (part number: 5190-2293) was used. After passing the GC column, the flow was split by a ratio of approximately 1:10 to an ISQ single quadrupole mass spectrometer (Thermo, Bremen, Germany). Data acquisition and processing was accomplished using Isodat 3.0 and Xcalibur 2.2 software (Thermo, Bremen, Germany). #### GC-EI-HRMS (QTOF) setup Following GC-TC-IRMS/MS analysis, the acetylated samples were diluted to a final volume of 200 μL cyclohexane and subjected to GC-EI-HRMS measurements on an Agilent 7200 QTOF system hyphenated to an Agilent 7890A gas chromatograph (Santa Clara, CA). The chromatography setup was equivalent to GC-TC-IRMS/MS described above (section GC-TC-IRMS/MS Setup), including the same analytical column and the same temperature program. The injection volume was reduced to 4 μL. Data were acquired within a range of *m*/*z* 50--800 with an acquisition rate of 5 spectra/s and evaluated with MassHunter software (version B.06, Agilent). Mass calibration was performed before and during each analytical batch. #### GC-EI-HRMS (Orbitrap) setup The trimethylsilylated samples were injected (2 μL injection volume) into a Q Exactive GC Orbitrap (Thermo, Bremen, Germany). Due to the different derivatization technique, the system was operated with modified chromatographic conditions adapted from routine protocols (Thevis et al., [@B35]). The GC was equipped with a HP-Ultra 1 column (17 m × 0.2 mm) of 0.11 mm film thickness. The temperature program started at 180°C and raised with 3°C/min to 240°C and with 40°C/min to 320°C, where it remained constant for 2 min. Samples were injected in split mode with a split flow of 5 mL/min. The mass range for full MS experiments was *m*/*z* 50--700 and a resolution of 60,000 FWHM was applied. Data were evaluated with Xcalibur software. #### LC-ESI-HRMS (LC Orbitrap) setup For LC-ESI-HRMS measurements, the acetylated samples were evaporated and reconstituted with 100 μL of ACN/H~2~O (50:50, *v*/*v*) acidified with 0.1% FA. A Vanquish UHPLC (Thermo, Bremen, Germany) equipped with a Poroshell 120 EC-C8 (2.7 μm, 3 × 50 mm) (Agilent, Santa Clara, CA) was hyphenated to a Q Exactive HF-X (Thermo, Bremen, Germany). ACN and ultrapure water both containing 0.1% FA were employed as solvents, and the flow rate was set to 400 μL/min. A volume of 5 μL was injected per sample. The LC gradient run was as follows: Starting at 40% ACN, it was increased to 99% within 9 min, and hold for further 3 min until re-equilibration, yielding a total analysis time of 15 min. MS experiments comprised a full scan (*m*/*z* 200--800), AIF (all ions fragmentation), and PRM in positive ionization mode at a resolution of 60,000 FWHM. For PRM experiments, the isolation window was set to *m*/*z* 0.4 and in the higher-energy collisional dissociation (HCD) cell, collision energies of 20, 30, 35 or 40 eV were applied. For pseudo MS^3^ experiments, the source induced dissociation (SID) energy was set to 20 eV. Analysis of Conjugated Steroids ------------------------------- For the analysis of intact conjugated steroids, samples were diluted 1:1 with ultrapure water. In order to perform PRM experiments, selected samples were 5-fold pre-concentrated by SPE as described in section 2.3.1, and reconstituted in 50 μL 10% aqueous ACN. The setup for the LC-ESI-HRMS (Orbitrap) system was similar to the settings described in chapter LC-ESI-HRMS (LC Orbitrap) Setup, but employing a modified gradient. Starting at 1% ACN (containing 0.1% FA), the gradient increased to 40% ACN within 9 min, to 99% until 10.9 min, and to 1% until 11 min. The system was re-equilibrated for 3 min. Full MS, AIF, and PRM experiments were conducted and ESI with positive polarity was applied. In selected experiments, also negative ionization was used, which is explicitly indicated in the corresponding data sets. Synthesis of Trenbolone-diol ---------------------------- Trenbolone-diol was synthesized by reduction of Tren under argon atmosphere. To 10 mg of Tren dissolved in 10 mL of anhydrous THF, 100 μL of potassium tri-sec-butylborohydride (1 M in THF) were added under constant stirring. After 15 min, the reaction was stopped with 10 mL of ultrapure water, and subsequently, a LLE with 20 mL of TBME was performed. An aliquot of the yielded products was acetylated as described in section Acetylation. Results and Discussion {#s3} ====================== Metabolite Identification With GC-TC-IRMS ----------------------------------------- The use of GC-TC-IRMS allows for comprehensive metabolite studies, particularly when investigating biotransformation products of deuterated compounds. A significant increase of the ratio *m*/*z* 3 to *m*/*z* 2 indicates the presence metabolites of the administered substance, in this case trenbolone. Exemplary GC-TC-IRMS chromatograms of both a pre- and post-administration sample are displayed in [Figure 3](#F3){ref-type="fig"}. The upper part (A) shows the chromatogram of the fraction containing the hydrolyzed glucuronides (fraction Gluc) of a pre-administration sample with deuterium levels at natural abundance. By contrast, the lower chromatogram (B) shows a sample collected 45 h following drug administration-, and several peaks of deuterated molecules corresponding to metabolites of Tren are visible. In the inset, details of the metabolites eluting at retention times between 1108 and 1110 s are shown. {#F3} Trenbolone Metabolites Identified by LC-ESI-HRMS ------------------------------------------------ Selected HPLC fractions exhibiting signals of deuterated compounds of considerable abundance were additionally analyzed by LC-ESI-HRMS(/MS). For metabolite identification, mass-to-charge-ratios were computed by *in silico* prediction and their presence or absence was assessed in the post-administration samples. Promising metabolites, their corresponding HPLC (sub-)fractions, retention times for LC-ESI-HRMS, exact masses, calculated elemental compositions, excretion forms, and detection windows are summarized in [Table 1](#T1){ref-type="table"}. Obtained elemental compositions were calculated within a maximum permitted mass error of ±4 ppm. Metabolites were determined in HPLC sub-fractions I, II, IV, and V. The number sequence corresponds to increasing retention times and decreasing polarity of the metabolites after hydrolysis, but before acetylation. ###### Deuterated TREN metabolites and diagnostic product ions (CE = 20 and CE = 30 eV) after hydrolysis of phase-II metabolites, HPLC fractionation and acetylation identified by LC-ESI-HRMS. **\#** **HPLC fraction** **Retention time HFX \[min\]** **Exact mass \[*m*/*z*\]** **Elemental composition** **Error \[ppm\]** **Excretion form[^a^](#TN1){ref-type="table-fn"}** **Detection window \[h\]** **Diagnostic production \[*m*/*z*\]** **Elemental composition** **Error \[ppm\]** **Diagnostic production \[*m*/*z*\]** **Elemental composition** **Error \[ppm\]** **Diagnostic production \[*m*/*z*\]** **Elemental composition** **Error \[ppm\]** **Diagnostic production \[*m*/*z*\]** **Elemental composition** **Error \[ppm\]** -------- ------------------- -------------------------------- ---------------------------- ----------------------------- ------------------- ---------------------------------------------------- ---------------------------- --------------------------------------- ----------------------------- ------------------- --------------------------------------- ----------------------------- ------------------- --------------------------------------- ----------------------------- ------------------- --------------------------------------- ----------------------------- ------------------- 1 l 4.39 301.2095 C~20~H~21~D~4~O${}_{2}^{+}$ −1.7 1_Sulf 45 259.2000 C~18~H~19~D~4~O^+^ 2.1 241.1890 C~18~H~17~D${}_{4}^{+}$ 0.5 1_Cys 45 2 I 5.25 301.2091 C~20~H~21~D~4~O${}_{2}^{+}$ −3.0 2_Sulf 45 259.1999 C~18~H~19~D~4~O^+^ 1.7 241.1890 C~18~H~17~D${}_{4}^{+}$ 0.5 2_Cys 118 3 II 3.33 318.2112 C~20~H~20~D~5~O${}_{3}^{+}$ −0.0 3_Gluc 21 276.2003 C~18~H~18~D~5~O${}_{2}^{+}$ −1.2 258.1896 C~18~H~16~D~5~O^+^ −1.8 4 II 3.68 318.2112 C~20~H~20~D~5~O${}_{3}^{+}$ 4_Gluc 45 276.2004 C~18~H~18~D~5~O${}_{2}^{+}$ −0.9 258.1897 C~18~H~16~D~5~O^+^ −1.5 5 II 3.47 373.1944 C~22~H~21~D~4~O${}_{5}^{+}$ −1.0 5_Gluc 94 331.1837 C~20~H~19~D~4~O${}_{4}^{+}$ −1.8 313.1731 C~20~H~17~D~4~O${}_{3}^{+}$ −2.0 289.1731 C~18~H~17~D~4~O${}_{3}^{+}$ −1.8 271.1625 C~18~H~15~D~4~O${}_{2}^{+}$ −1.7 5_Sulf 70 6 II 3.56 313.1731 C~20~H~17~D~4~O${}_{3}^{+}$ −1.7 6_Gluc 94 271.1628 C~18~H~15~D~4~O${}_{2}^{+}$ −0.1 253.1523 C~18~H~13~D~4~O^+^ −0.8 227.1366 C~16~H~11~D~4~O^+^ −1.1 6_Sulf 70 7 II 4.95 357.1995 C~22~H~21~D~4~O${}_{4}^{+}$ −1.0 7_Gluc 118 315.1888 C~20~H~19~D~4~O${}_{3}^{+}$ −1.5 297.1783 C~20~H~17~D~4~O${}_{2}^{+}$ −1.4 273.1784 C~18~H~17~D~4~O${}_{2}^{+}$ −1.2 255.1676 C~18~H~15~D~4~O^+^ −2.2 8 II 5.03 357.1996 C~22~H~21~D~4~O${}_{4}^{+}$ −0.7 8_Sulf 118 315.1888 C~20~H~19~D~4~O${}_{3}^{+}$ −1.5 297.1784 C~20~H~17~D~4~O${}_{2}^{+}$ −1.1 273.178 C~18~H~17~D~4~O${}_{2}^{+}$ −2.6 255.1678 C~18~H~15~D~4~O^+^ −1.4 8_Cys 45 9 IV 1.87 273.1787 C~18~H~17~D~4~O${}_{2}^{+}$ −0.1 9_Gluc 142 9_Sulf 94 10 IV 3.35 313.1736 C~20~H~17~D~4~O${}_{3}^{+}$ −0.1 10_Sulf 94 271.1631 C~18~H~15~D~4~O${}_{2}^{+}$ 0.1 253.1528 C~18~H~13~D~4~O^+^ 1.2 227.1370 C~16~H~11~D~4~O^+^ 0.7 10_Cys 45 11 V 1.58 313.1736 C~20~H~17~D~4~O${}_{3}^{+}$ −0.1 11_Sulf 45 271.1635 C~18~H~15~D~4~O${}_{2}^{+}$ 0.1 253.153 C~18~H~13~D~4~O^+^ 2.0 227.1374 C~16~H~11~D~4~O^+^ 2.4 12 \- 2.55 273.1788 C~18~H~17~D~4~O${}_{2}^{+}$ 0.3 12_Cys 70 255.1685 C~18~H~15~D~4~O^+^ 1.4 229.1531 C~16~H~13~D~4~O^+^ 2.6 13 \- 4.83 359.2152 C~22~H~23~D~4~O${}_{4}^{+}$ −0.8 13_Cys 118 317.2054 C~20~H~21~D~4~O${}_{3}^{+}$ 1.2 299.1946 C~20~H~19~D~4~O${}_{2}^{+}$ 0.8 257.1840 C~18~H~17~D~4~O^+^ 0.8 239.1737 C~18~H~15~D${}_{4}^{+}$ 1.9 *De-conjugation under alkaline conditions expecting potential cysteine conjugates*. Altogether, a total of 20 phase-II metabolites of relevant traceability was identified. Almost all metabolites were found to be eliminated as differently conjugated products including the well-characterized glucuronic acid and sulfate conjugates, as well as alkaline-labile phase-II metabolites. De-conjugation under alkaline conditions is less established for human metabolism, but has already been assessed in preventive doping research to generate cysteine conjugates (Fabregat et al., [@B9], [@B8]; Pozo et al., [@B26]; Gomez et al., [@B11]). Especially for Tren, cysteine conjugates are supposed to be of relevance (Sobolevsky and Rodchenkov, [@B30]). Presumed phase-I reactions comprised hydroxylation, dehydrogenation, dehydration, and reduction of a keto to a hydroxyl moiety or *vice versa*. Most metabolites were identified as the 4-fold deuterated isomers of the metabolites although 5-fold deuterated Tren was administered. This suggests that a metabolic conversion occurs predominantly within the steroidal A or B ring. The 5-fold deuterated isomers of the well-known glucuronic acid-conjugated metabolites Tren (metabolite 4) and EpiTren (metabolite 3) were confirmed in the respective Gluc fraction and were unambiguously identified by retention time and product ion mass spectra in comparison to reference material. For sports drug testing, especially long term metabolites with extended detection windows are of great interest. Within this study, metabolites 7, 8, 9, and 11 were found to have detection times between 118 h (5 days) and 142 h (6 days). Extracted ion chromatograms (EIC) displaying metabolite 9_Gluc in pre- and post-administration samples are depicted in [Figure 4](#F4){ref-type="fig"}. The hydrolyzed metabolites 7 (excreted as glucuronide) and 8 (excreted as sulfate) are considered as isomers due to their identical elemental composition and their minor difference in retention time. Since metabolite 7/8 and metabolite 9 are excreted in two conjugation forms, in particular as glucuronides and sulfates, they were chosen for further characterization (see section Production Ion Mass Spectra). {#F4} Metabolite Characterization --------------------------- ### Product Ion Mass Spectra Besides accurate mass measurements verifying the deuterium content, all identified metabolites were characterized by product ion mass spectra (MS^2^) obtained from PRM experiments. Analysis was performed in ESI positive mode, as it was found to produce more characteristic product ion mass spectra compared to those obtained following negative ionization (Rzeppa et al., [@B27]). Collision induced dissociation (CID) was accomplished at different collision energies ranging from 20 to 40 eV. The PRM mass spectrum of metabolite 9_G after hydrolysis and acetylation is illustrated in [Figure 5](#F5){ref-type="fig"} as an example, the diagnostic product ions, elemental compositions, and mass errors of the other metabolites are listed in [Table 1](#T1){ref-type="table"}. From the elemental composition and the MS/MS-experiments, metabolite structures were postulated. Metabolite 5 is tentatively assigned to a hydroxyl-metabolite of Tren, which was substantiated by the 2-fold loss of an acetyl moiety \[neutral loss of *m*/*z* 42 (Ac) and *m*/*z* 60 (AcOH)\]. For this metabolite, the generation of a fourth double bond by oxidation of a tertiary carbon atom is additionally required. {#F5} The metabolites 7 and 8 are tentatively assigned to derivatives of Tren that result from the reduction of the 3-oxo functionality of the anabolic steroid as supported by the characteristic and repeatedly occurring losses of acetyl moieties. Further, the presence of two additional double bonds (of unknown position) is postulated. The elemental composition of metabolite 9 indicates a potential metabolic conversion to trenbolone-diketone. The metabolite was not derivatized by pyridine/acetic anhydride, which corroborates the absence of hydroxyl functions. In accordance with the current state of knowledge, free hydroxyl groups are acetylated under the applied conditions, but -oxo functions are not affected; however, failed derivatization due to steric hindrance cannot be excluded (Piper et al., [@B22]). The detection of trenbolone-diketone, also known as 17-keto-trenbolone or trendione, was already reported in 1991 (Spranger and Metzler, [@B31]) for *in vivo* samples, and the compound could be generated *in vitro* by the same group by using human liver microsomes (Metzler and Pfeiffer, [@B18]). Noteworthy, an additional abundant signal of *m/z* 97.0652 corresponding to C~6~H~9~O as elemental composition (±4.2 ppm) was observed in the PRM spectra. Since trenbolone-diketone is deuterated in the A- and B-ring, the signal has to be derived from the steroidal C- or D-ring. Structural elucidation of the equivalent diagnostic ion has been accomplished for androst-4-en-3-one-based steroids, but this structure cannot be immediately transferred to the herein investigated molecule as it was shown to originate from the steroidal A-ring (Thevis et al., [@B33]). ### Synthesis of Trenbolone-diol For the metabolites 7_Gluc and 8_Sulf, which were proposed to represent trenbolone-diol derivatives, further in-depth studies were conducted. In order to obtain a reference mass spectrum of trenbolone-diol, an in-house synthesis was accomplished by reducing Tren with potassium tri-sec butylborohydride. Two isomers of trenbolone-diol were successfully synthesized and PRM spectra of the free and acetylated forms were acquired. Remarkably, also 1-fold and 2-fold dehydrogenated trenbolone-diol derivatives were obtained as byproducts in two isomeric forms despite the employed reducing conditions. The 2-fold dehydrogenated trenbolone-diol derivative was found to be analog to the predicted metabolite 7/8 after hydrolysis. In [Figure 6](#F6){ref-type="fig"}, the EIC of the synthesized 2-fold dehydrogenated trenbolone-diol derivative (A) is compared to the metabolites identified in the post-administration urine samples (B). The retention times (4.85 and 4.95 min) of the synthesized products are in accordance with the postulated metabolites. Both synthesized isomers were present in the glucuronide and the sulfate fraction. While the metabolite at 4.95 min is the most prominent isomer eliminated as glucuronide, the sulfated analog with the longest detection window at 5.03 min was not synthesized. The potential cysteine conjugates are less valuable due to their short detection window. {#F6} Product ion mass spectra of the acetylated 2-fold dehydrogenated trenbolone-diol derivative isomer at 4.95 min and the acetylated metabolite 7_Gluc after hydrolysis are in good agreement as demonstrated in [Figure 7](#F7){ref-type="fig"}. The observed mass shift of four Da is caused by the 4-fold deuteration of the metabolite. The PRM mass spectra of the three hydrolyzed metabolites were highly comparable in all excretion forms. As a consequence, they are assumed to be different isomeric forms of the 2-fold dehydrogenated trenbolone-diol derivative. Isomers can be attributed to both stereo centers at C3 and C17, as well as various positions of the additional double bonds. {#F7} ### Phase-II Conjugates The sample preparation comprising SPE, LLE, hydrolysis, HPLC fractionation, and derivatization is very elaborate and may be accompanied by analyte losses and/or the formation of artifacts during derivatization (Piper et al., [@B21]). Moreover, the herein employed approach of analyzing acetylated steroids by LC-ESI-HRMS is certainly unconventional. In sports drug testing, the implementation of new analytes/metabolites into existing methods is of great importance to enable a specific and sensitive detection of the target molecules as well as adequate detection windows. Currently, three different methods are routinely used for the analysis of doping control samples, where an implementation of novel Tren metabolites appears feasible: (a) measurement of analytes hydrolyzed and derivatized as TMS derivatives with GC-EI-MS/MS, (b) measurement of hydrolyzed, but not acetylated analytes, (c) measurement of intact phase-II metabolites with LC-ESI-HRMS. Since the ionization efficacy of steroid diols with low proton affinity such as metabolites 7/8 by ESI is usually limited, the existing LC-ESI-HRMS approach was not considered as the preferred option. When using LC-ESI-HRMS, it is generally advisable to measure the intact phase-II conjugates, which also results in a reduced workload (Gomez et al., [@B11]). In order to investigate the new metabolites regarding their applicability to doping control routine testing, the intact phase-II metabolites were analyzed by LC-ESI-HRMS. For the potential Tren-diol derivatives (metabolite 7_Gluc), a peak corresponding to the singly glucuronidated metabolite was identified. The pseudo MS^3^ mass spectra of the intact glucuronide and the acetylated and hydrolyzed metabolite are compared in [Figure 8](#F8){ref-type="fig"}. Both product ion mass spectra derived from *m*/*z* 273.1787 are in good agreement. Observed variations in the relative intensities are potentially caused by differences in the energetic status of both molecules resulting from the in-source dissociation process. {#F8} ### Comparison to Reference Material As outlined in above, the characterization of metabolite 9 yielded a potential diketone derivative of trenbolone. A commercially available reference standard of trenbolone-diketone, 4,9,11-estratriene-3,17-dione, was analyzed by LC-ESI-HRMS. The pseudo MS^3^ spectrum of the sulfated diketone metabolite in urine and the product ion mass spectrum of the reference standard are displayed in [Figure 9](#F9){ref-type="fig"}. Again, the mass shift of four Da is caused by the 4-fold deuteration of the metabolite. The good agreement of both product ion mass spectra supports the assignment of metabolite 9_Sulf to a potential diketone derivative. It is noticeable that phase-II conjugation appears to occur via keto-/enol-tautomerism as described earlier also for androstenedione (Tajic and Kovacic, [@B32]; Goodall and James, [@B12], [@B13]). {#F9} Determination of Detection Windows for Published Long-Term Metabolites ---------------------------------------------------------------------- State-of-the-art routine doping control methods are primarily based on epitrenbolone, trenbolone glucuronide and epitrenbolone glucuronide (De Boer et al., [@B6]; Schänzer, [@B29]; Brun et al., [@B3]). In addition, trenbolone sulfate and the trenbolone cysteine adduct have been published (Rzeppa et al., [@B27]; Sobolevsky and Rodchenkov, [@B30]). While no detection windows for the sulfate conjugate have been described yet, limited data sets are available for the glucuronide conjugate. An early Tren metabolism study used radioactive labeling and investigated the urinary excretion. Spranger and Metzler detected 54% of the radioactivity within 26 h and 63% within 72 h post Tren administration, indicating a fast elimination (Spranger and Metzler, [@B31]). A publication from the former Russian anti-doping laboratory published the traceability of trenbolone administrations of 32 days when employing epitrenbolone glucuronide as target analyte; similarly the trenbolone cysteine adduct was detected for 32 days. The deuterated analogs of the described metabolites were also detected in this elimination study as illustrated in [Figure 10](#F10){ref-type="fig"} by means of respective EIC. {#F10} Epitrenbolone was detectable for 45 h and Tren for 21 h after oral administration of 10 mg of trenbolone. For the sulfate conjugate, three potential isomers were identified, two of which are obviously induced by epimerization at position 17. The third one remained unclear but might be attributed to a conjugate directed to position 3. This can be derived from the occurrence of the above described assumed conjugation of the diketone derivatives metabolite 9_Gluc and 9_Sulf, which appear to be conjugated via a steroidal keto group. All three isomers were detectable for 45 h following administration. The Tren cysteine conjugate was observed for 45 h on *m*/*z* 276.2006 referring to the steroid structure which is generated by in source dissociation. The intact phase-II metabolite at *m*/*z* 397.2204 was also visible, but for [Figure 10](#F10){ref-type="fig"}, the mass-to-charge ratio of the variant with the longest detectability was chosen. The results generated in this study corroborate the primary data of Spranger and Metzler, where Tren was identified as substance with a fast elimination. Inter-individual variations in the metabolism of the volunteers of the different studies are a conceivable explanation for deviating results, but investigation of a larger population appear necessary and warranted to further substantiate the observations. Limitations of Combining Data Obtained From GC-TC-IRMS/MS, LC-ESI-HRMS, and GC-EI-HRMS -------------------------------------------------------------------------------------- In previous publications, GC-TC-IRMS/MS has been successfully employed in support of steroidal metabolite identification, especially when combined with GC-EI-HRMS (Thevis et al., [@B38]; Piper et al., [@B24],[@B25], [@B20], [@B21]). This concept was adapted to this study, and referring to [Figure 3](#F3){ref-type="fig"}, two metabolites were identified by using the conventional approach. The first peak at 1108 s was identified as the glucuronic acid conjugate of the well-characterized metabolite EpiTren by comparing the retention time and mass spectrum to reference material. The metabolite at 1110 s was formerly unknown and further characterized by GC-EI-HRMS (TOF) analysis. The resulting mass spectrum is shown in [Figure 11](#F11){ref-type="fig"}. Under consideration of the different ionization and dissociation mechanisms, this mass spectrum was found to correspond to that of the deuterated 2-fold dehydrogenated trenbolone-diole derivative characterized by LC-ESI-HRMS ([Figure 7](#F7){ref-type="fig"}). {#F11} Unfortunately, the described approach could not be applied to all metabolites. The quality of analytical data generated by the GC-TOF system suffered from an intense fragmentation, which led to a reduced abundance or absence of the molecular ion, complicating structural elucidation of new metabolites (Thevis et al., [@B38]; Piper et al., [@B24],[@B25], [@B20], [@B21]). Therefore, a GC Orbitrap system with an alternative ion source design favoring higher abundances of ions with larger *m/z* values was used, and the measured TMS derivatives further exhibited a reduced in-source fragmentation compared to acetates (Piper et al., [@B20], [@B21]). In addition, trimethylsilylation of steroidal analytes is routinely used, thus offering a platform for implementing potential new target analytes into sports drug testing methods. However, the analysis of TMS derivatives with the GC Orbitrap remained challenging as the analysis of reference material of trenbolone, epitrenbolone and d~5~-trenbolone introduced as TMS-derivatives resulted in a variety of signals presumably caused by derivatization and thermal degradation artifacts. Consequently, no unreported metabolites could be identified. For the TMS derivatives of trenbolone and epitrenbolone, a signal at *m*/*z* 414.2405 representing the molecular ion is expected. Moreover, peaks at *m*/*z* 412.2248 and *m*/*z* 410.2092 have been reported, which belong to products eliminating two or four hydrogens (De Boer et al., [@B6]; Ayotte et al., [@B1]). As shown in [Figure 12](#F12){ref-type="fig"}, several additional peaks of unknown origin were observed in the reference material, which suggested a significantly limited utility of this setup for identifying additional metabolites in urine. {#F12} It has to be taken into account that the herein investigated molecule possesses poor gas chromatographic properties, and the highly conjugated 4,9,11-triene-3-one structure results in derivatization artifacts with low thermal stability. Other structurally related molecules such as the designer steroid tetrahydrogestrione are not detectable in urine as TMS-derivatives as well, and remained hidden until its first identification as performance-enhancing drug in 2003 by means of LC-MS/MS (Catlin et al., [@B5]; Thevis et al., [@B34]). In other words, Tren has a documented history of challenging analytics as corroborated by a variety of assessed derivatization techniques (De Boer et al., [@B6]; Ayotte et al., [@B1]; Casademont et al., [@B4]; Marques et al., [@B17]; Brun et al., [@B3]; Parker et al., [@B19]). LC-ESI-MS was found to be suitable for the quantification of Tren (Thevis et al., [@B34],[@B36], [@B37]) in the past, and common routine doping control procedures nowadays utilize LC-ESI-MS/MS for analytes of this and related structure. As a consequence, also here a LC-ESI-HRMS system was used to overcome the limitations regarding the mass resolution and the presence of the molecular ion. Furthermore, a derivatization by acetylation improved the proton affinity for ESI positive ionization mode, and especially the ionization of the trenbolone-diol derivatives metabolite was found to be strongly increased. The advantage of the untargeted GC-TC-IRMS approach is unfortunately not as pronounced as in previous studies as aligning LC and GC chromatograms and (presumed) molecular ions for target analyte characterization is difficult. Conclusion {#s4} ========== In order to enhance the retrospectivity and sensitivity of analytical approaches targeting trenbolone misuse in sport, a comprehensive *in vivo* metabolism study was performed. An approach utilizing stable isotope-labeled substrates facilitating the investigation of biotransformations by GC-TC-IRMS was employed. While the strategy proved straightforward in earlier studies, trenbolone and its metabolic products presented comparably challenging target analytes due to their limited compatibility with gas chromatography. Nevertheless, by employing miscellaneous techniques of derivatization and chromatography, a total of 20 metabolites excreted as glucuronides, sulfates and potential cysteine conjugates were identified. Four metabolites, tentatively attributed to trenbolone-diketone and a 2-fold dehydrogenation product of trenbolone-diol, eliminated both as glucuronide and sulfate, were found to complement the existing urinary trenbolone metabolic pattern, offering detection windows of 6, respectively 5 days. Further characterization of these metabolites was conducted by pseudo-MS^3^ experiments and comparison to commercially available or in-house synthesized reference material. To verify or falsify the true added value of the herein identified trenbolone metabolites for routine doping controls, those samples that return suspicious or even adverse analytical findings for trenbolone using established approaches will be further investigated regarding the new potential target analytes. If a positive contribution will be observed, future studies to confirm tentatively assigned structures e.g., by nuclear magnetic resonance analysis after upscaling of the synthesis and an administration study of unlabeled trenbolone could be warranted. Moreover, it might be of interest to administer other doses of trenbolone and to investigate a larger population for examination of inter-individual variations. Data Availability Statement {#s5} =========================== The datasets generated for this study are available on request to the corresponding author. Ethics Statement {#s6} ================ The administration study involving a human participant was approved by the Ethics Committee of the National Institute of Sports of Romania (Bucharest, Romania, \#2283, 2016) according to the declaration of Helsinki. Informed consent was obtained from the volunteer. Author Contributions {#s7} ==================== MT and TP conceived conception and study design, contributed to data interpretation and discussion, and revised and edited the manuscript. MP conducted the sample preparation, measurements, data evaluation, and wrote the draft of the manuscript. 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. The authors like to acknowledge the contribution of Dr. Andreas Thomas, Dr. Josef Dib, and the Romanian Doping Control Laboratory for analytical and administrative support. **Funding.** This project was financially supported by the Manfred-Donike Institute for Dope Analysis (Cologne, Germany) and the World Anti-Doping Agency (grant \#17A31MT). [^1]: Edited by: Alberto Salomone, University of Turin, Italy [^2]: Reviewed by: Benjamin L. Oyler, Vaccine Research Center (NIAID), United States; Roberta Risoluti, Sapienza University of Rome, Italy [^3]: This article was submitted to Analytical Chemistry, a section of the journal Frontiers in Chemistry

1. Introduction {#sec1-nutrients-11-00864} =============== Fishery products provide many beneficial nutritional components such as long-chain polyunsaturated fatty acids (PUFAs), high-quality-proteins, essential elements, and vitamins \[[@B1-nutrients-11-00864],[@B2-nutrients-11-00864]\]. Among them, mussels are considered an excellent source of proteins, and it is estimated that a 100 g portion of mussel meat provides a quarter of an adult's daily protein need. It is also considered that the consumption of this amount of mussels provides the recommended daily intake of vitamin B~12~ \[[@B3-nutrients-11-00864]\]. It has also been reported that mussels are a relevant source of some essential trace elements, such as Se, Fe, and Zn \[[@B3-nutrients-11-00864],[@B4-nutrients-11-00864]\]. Taking into account all these nutritional facts and summing up that mussels have low cholesterol levels and a low glycemic index, they should be considered a balanced, healthy, and dietary food choice based on its contribution of nutrients. However, this mollusk is also a concern because, in addition to beneficial elements, it also has the capacity to accumulate many other toxic or potentially toxic elements \[[@B5-nutrients-11-00864],[@B6-nutrients-11-00864],[@B7-nutrients-11-00864],[@B8-nutrients-11-00864],[@B9-nutrients-11-00864]\]. In fact, mussels have been widely employed as sentinel organisms in coastal pollution monitoring, in particular in regard to heavy metal contamination \[[@B10-nutrients-11-00864]\]. This is because the gill tissue of mussels is particularly rich in metallothionein, and this attribute of mussels therefore constitutes a key interface for the uptake of dissolved metals and their further incorporation into lysosomes and their transport in blood plasma and circulating hemocytes \[[@B10-nutrients-11-00864]\]. In the case of mussels, apart from the abovementioned characteristics, it is also important to consider that a good part of them is consumed in the form of canned preserves, as mussels (similar to the other seafoods) are easily spoiled and very prone to oxidation and to developing off-flavors due to wrong handling or incorrect storage. However, very often, canned foods in general are perceived by consumers as low-quality products, which are thought to be possibly produced using lesser quality raw ingredients, and fresh seafood is in general perceived as the healthier alternative to frozen and processed products \[[@B11-nutrients-11-00864]\]. One of the consumers' suspicions of worse food quality has to do with the packaging material and with the possibility that the canned seafood, apart from their own content in heavy metals from the marine environment, may also be contaminated by heavy metals during the canning process \[[@B12-nutrients-11-00864]\]. The mussel canning industry is the recipient of two-thirds of the annual Spanish production of mussels \[[@B13-nutrients-11-00864]\]. In general, Spanish mussel production is the second largest in the world, after China's, with an annual production of around 250,000 tonnes/year (14% of the world's production) \[[@B14-nutrients-11-00864]\]. Mussels are also the type of seafood most consumed by the Spanish population, which has about 15% of regular consumers with an average consumption of around 20.5 kg/person/year (about half in children). This amount is divided between the consumption of canned (40%), fresh (50%), and frozen (10%) mussels \[[@B13-nutrients-11-00864]\]. This data is very relevant since, when determining the nutrient concentrations in foods, it is important to consider the different ways of preserving them because they can modify the nutritional composition in a very significant way. Therefore, if one wants to be precise in the estimation of the intake of nutrients and/or contaminants through a certain type of food, all possible ways in which said food is consumed should be considered. In the case of mussels, this would represent the estimation of the intake through one of modes of consumption---canned, fresh, and deep frozen. However, although there is abundant scientific literature documenting the levels of essential and toxic elements in mussels, to the best of our knowledge, very few studies have made an exhaustive comparison of their levels according to their mode of conservation \[[@B15-nutrients-11-00864],[@B16-nutrients-11-00864]\], and none has taken into account this differential content in the dietary intake estimation. Consequently, this study was conducted to determine the content of forty-three elements (essential and toxic) in all the forms in which mussels are acquired and consumed in Spain (preserved, fresh, and deep-frozen) with the aim of performing an accurate estimation of the contribution of this food to the daily intake of these elements and of performing a risk--benefit evaluation by comparing the estimated daily intake with dietary and toxic reference values. 2. Materials and Methods {#sec2-nutrients-11-00864} ======================== 2.1. Sampling and Collection {#sec2dot1-nutrients-11-00864} ---------------------------- In this research, we studied a total of 208 pooled mussel samples. Mussels were randomly purchased between July and August of 2018 from supermarkets and fish markets of the Canary Islands (Spain). We intended to cover the main forms of presentation of this food in the market, but avoided purchasing preparations and processed mussels (with pickles, sauces, etc.). Thus, we analyzed 88 samples of canned mussels (only steamed and preserved in salted water), 80 samples of deep-frozen mussels, and 40 samples of fresh mussels. Each sample for analysis consisted of 4--5 individual mussels of each brand that were homogenized together. Regarding the origin, the samples were from Galicia (Spain), Chile, and New Zealand, according to the following distribution ([Table 1](#nutrients-11-00864-t001){ref-type="table"}): i) the canned samples were from Galicia (*n* = 72, 38 name brands, and 34 store brands) and from Chile (*n* = 16, all name brands); ii) the frozen samples were from Galicia (*n* = 36, 8 of them were certified as organic production), from Chile (*n* = 28), and from New Zealand (*n* = 12); and iii) the fresh samples were all from Galicia (*n* = 40). In the sampling design, we tried to represent all the possible national and international brands available throughout the Spanish territory. Thus, all the samples came from large suppliers that serve the entire nation, and we consider that our results could be extrapolated and made representative of the Spanish market. After the purchase, all the frozen and fresh mussel samples were kept on ice to maintain the cold chain until their arrival to the Laboratory of Toxicology of the University of Las Palmas de Gran Canaria (ULPGC), where they were processed immediately. 2.2. Standards and Elements {#sec2dot2-nutrients-11-00864} --------------------------- We determined the concentration levels of 43 elements in mussels, including the essential elements and those elements more classically studied because of their high toxicity. Additionally, we included a suite of other elements, including 1) the elements in the ATSDR's priority list and 2) the rare earth elements (REEs) and other minority elements (ME) that are of increasing concern because of their massive employment in the manufacturing of electric and electronic consumer products and therefore are increasingly appearing as emerging environmental pollutants \[[@B17-nutrients-11-00864],[@B18-nutrients-11-00864]\]. The complete list of elements comprises the following: Ag (silver); Al (Aluminum); As (arsenic); Ba (barium); Be (beryllium); Cd (cadmium); Ce (cerium); Co (cobalt); Cr (chromium); Cu (copper); Dy (dysprosium); Eu (europium); Er (erbium); Fe (iron); Ga (gallium); Gd (gadolinium); Hg (mercury); Ho (holmium); In (indium); La (lanthanum); Lu (lutetium); Mn (manganese); Mo (molybdenum); Nb (niobium); Nd (neodymium); Ni (nickel); Pb (lead); Pd (palladium); Pr (praseodymium); Sb (antimony); Se (selenium); Sm (samarium); Sn (tin); Sr (strontium); Ta (tantalum); Tb (terbium); Th (thorium); Tl (thallium); Tm (thulium); U (uranium); Y (yttrium); Yb (ytterbium); and Zn (zinc). Pure standards of elements in acid solution (5% HNO~3~, 100 mg/L) were purchased from CPA Chem (Stara Zagora, Bulgaria). Two standard curves (twelve points, 100--0.005 ng/mL) were made to avoid interferences between elements: a) one using a commercial multi-element mixture (CPA Chem Catalog number E5B8.K1.5N.L1, 21 elements, 100 mg/L, 5% HNO~3~) containing all the essential elements and main heavy metals; and b) other multi-element mixture tailor-made in our laboratory from individual elements (CPA Chem), which contained the REEs and MEs most frequently employed in the high-tech industry \[[@B18-nutrients-11-00864]\]. 2.3. Analytical Procedure {#sec2dot3-nutrients-11-00864} ------------------------- All the ready-to-eat canned samples were manually pooled using a metal-free Teflon mortar until forming a homogeneous mass. The fresh and frozen samples required a shell opening and cooking in their own juice. This was done by steaming them using a domestic food processor (Thermomix^®^, Vorwerk, Wuppertal, Germany) for a period of 10 min, and after this, the samples were processed in the same way as the canned samples. For the analysis of elements, mussel samples were acid-digested with the aid of a microwave digester (Ethos Up, Milestone SRL, Italy). Briefly, 500 mg of mussel homogenate were weighed into the digestion vessels, and 50 μL of the internal standard solution (Sc (scandium), Ge (germanium), Rh (rhodium), and Ir (iridium) at a stock concentration of 20 mg/mL each) were added. Next, 2.5 mL of concentrated sub-boiling HNO~3~ (65%) and 7.5 mL of Milli-Q water were added to each sample. All samples were digested according to the following program: Step 1: a power (W), temperature (°C), and time (min) of 1800, 100, and 5, respectively; Step 2: 1800, 150, and 5; Step 3: 1800, 200, and 8; Step 4: 1800, 200, and 7. After cooling, the digests were transferred into conic bottom polypropylene tubes and diluted up to 15 mL with Mili-Q water. Finally, an aliquot of each sample was taken for the analysis. Reagent blanks were prepared similarly to the samples, and a reagent blank was included every 14 samples in the analytical batch. For the instrumental analyses, we employed an Agilent 7900 ICP-MS (Agilent Technologies, Tokyo, Japan) equipped with standard nickel cones and a cross-flow nebulizer with a Make Up Gas Port (×400 Nebulizer, Savillex Corporation, MN, USA) for all measurements. All the data were acquired and processed with Agilent MassHunter Data Analysis software (version 4.2, Agilent Technologies, Palo Alto, CA, USA). On a daily basis, the ICP-MS was optimized using a tuning solution consisting of a mixture of Cs (cesium), Co (cobalt), Li (lithium), Mg (magnesium), Tl (thallium), and Y (yttrium) (Agilent Technologies, Palo Alto, CA, USA). All measurements were performed in triplicate from each vial. The entire/complete procedure was validated prior to its use in the analyses of samples. Recoveries obtained ranged from 87 to 118% for toxic and essential elements. Linear calibration curves were found for all elements (regression coefficients ≥0.998). Instrumental LODs and LOQs were calculated as the concentration of the element that produced a signal that was three and ten times higher than that of the averaged blanks, respectively. The sample LOQs were calculated by multiplying the instrumental LOQ by the dilution factor suffered by the sample during the digestion procedure (1:10 v:v). 2.4. Dietary Intake Estimates, Nutritional and Health Risk Assessment {#sec2dot4-nutrients-11-00864} --------------------------------------------------------------------- For the estimation of the intake of elements, the total consumption of mussels was taken into account. That is, the consumption of each mussel type (g/day) \[[@B19-nutrients-11-00864]\] was multiplied by the median values of each element (ng/g fresh weight) in that type of mussel. The total consumption of each element (ng/kg body weight/day) was obtained by adding the individual consumptions obtained for canned, fresh, and frozen mussels. Both average consumers and high consumers (those in the 97.5th percentile (P~97.5~)) were considered, and the estimations were done for two age groups: adults (\>17 years) and children (7 to 12 years). For the estimation of the risk--benefit ratio, the values of estimated daily intake (EDI) of elements for each scenario (average and high consumers) and age group were compared with the reference values. As dietary reference values (in the case of the essential elements, DRVs), the population reference intake (PRI) values as reported by the European Food Safety Authority (EFSA) \[[@B20-nutrients-11-00864]\] were used. According to the European standard, the PRI is the equivalent of the recommended dietary allowances (RDAs) in the USA, that is, the daily dietary intake level of a nutrient considered sufficient to meet the requirements of 97.5% of healthy individuals in each life stage and sex group. In those cases in which the EFSA has not reported the PRI, the adequate intake (AI) was employed as the reference value. AI is the average nutrient level consumed daily by a typical healthy population that is assumed to be adequate for the population's needs. For those estimates of essential elements that exceeded the PRI or AI, the tolerable upper daily intake level (UL) was considered as well. The UL is the maximum level of total chronic intake of a nutrient from all sources judged to be unlikely to pose a risk of adverse health effects in humans \[[@B21-nutrients-11-00864],[@B22-nutrients-11-00864]\]. As toxic reference values (TRVs), the non-carcinogen tolerable daily intake (TDI) values from the US EPA \[[@B23-nutrients-11-00864]\] were employed. No TRV has been established for Pd and Th, so these two elements were excluded from the risk analysis. No official TRV has been established either for the REEs or the other MEs included in this research. However, some authors have proposed a daily allowable intake of 61 µg/kg body weight (bw) for rare earth oxides \[[@B24-nutrients-11-00864],[@B25-nutrients-11-00864]\], which was certificated from human health surveys in REE mining areas and animal experimental results. We employed this value as the TRV for these elements, considered as a group (sum REEs). 2.5. Statistical Analysis {#sec2dot5-nutrients-11-00864} ------------------------- Descriptive analyses were conducted for all variables. Arithmetic means, standard deviation (SD), medians, and ranges were calculated for continuous variables. To those data below the LOQ but above the LOD, a random value between those two limits was assigned. Those data below LOD were considered as non-detected. The normality of the data was tested using both the Kolgomorov--Smirnov test (with Dallal--Wilkinson--Lilie for *p* values), and the D'Agostin---Pearson omnibus test. As expected, most of the data series did not follow a normal distribution. Consequently, we chose not to assume a normal distribution in any case, and comparisons between the groups were performed using non-parametric tests (Kruskal--Wallis test or Mann--Whitney *U* test). We used PASW Statistics v 25.0 (SPSS Inc., Chicago, IL, USA) to manage the database of the study and to perform statistical analyses. Probability levels of \<0.05 (two-tailed) were considered statistically significant. 3. Results {#sec3-nutrients-11-00864} ========== 3.1. Occurrence of Essential, Toxic, and Potentially Toxic Elements in Mussels {#sec3dot1-nutrients-11-00864} ------------------------------------------------------------------------------ In [Table 2](#nutrients-11-00864-t002){ref-type="table"} (essential elements) and [Table 3](#nutrients-11-00864-t003){ref-type="table"} (toxic and potentially toxic elements), we show the descriptive study of the concentrations found in the three types of mussels considered: preserved, fresh, and frozen. For the great majority of elements, there are significant differences in the levels depending on the mode of conservation of the mussels. The exceptions were Mo among the essential elements and Ag, Be, Pd, Tl, and U among the rest of the elements considered in this research. Zn, Al, and Fe were the most abundant elements in all the types of mussels, although fresh mussels had the lowest levels of Fe and Al and yet the highest levels of Zn. The fact that one of the most abundant elements is As is surprising, since it can be considered that this element is a contaminant and not a constituent of the biology of the mussels. Additionally, in this case, we found important differences, depending on the mode of conservation, with the highest levels being those in non-preserved mussels (fresh and frozen samples). The mussels also contain relevant amounts of Sr, Cu, and Mn. Additionally, in these cases, we found concentration differences depending on the mode of conservation of mussels, although these are of lesser importance ([Table 2](#nutrients-11-00864-t002){ref-type="table"}). Cobalt is the essential element present at the lowest concentration by far in all the sampled mussels. With respect to Se, although the concentrations seem not to be high, they can be considered very relevant in nutritional terms, as detailed in the following section. Regarding the Pb, Cd, and Hg, they were detected in practically all the analyzed samples. In particular, the concentrations of Cd and Pb may be considered relatively high ([Table 3](#nutrients-11-00864-t003){ref-type="table"}) in all the three types of mussels, but in general, we found that canned mussels were the ones that presented the lowest concentrations of these toxic elements. With respect to the rest of the toxic and potentially toxic elements studied, the concentrations can be considered very low in all cases. As also observed for the other elements, the levels tended to be somewhat lower in canned mussels than in the other two types of mussels considered. Two striking exceptions are Al and Sn, since they are two of the metals used in the manufacture of cans. In fact, Sn concentrations, although low in all cases, are of the order of 3-4 times higher in canned mussels than in fresh or frozen mussels ([Table 3](#nutrients-11-00864-t003){ref-type="table"}). In the case of REEs considered individually, the levels are also very low ([Table S1](#app1-nutrients-11-00864){ref-type="app"}). However, it is very striking that when considered as a sum, the levels in fresh mussels are practically seven times higher than those found in the fresh and frozen mussel samples ([Table 3](#nutrients-11-00864-t003){ref-type="table"}). In the main body of the article, we decided to present the results obtained according to the type of conservation of the mussels, since we consider that it is one of the factors that most influences the way in which the final consumer will be exposed. However, we have made other types of analysis, considering the origin of production of the mussels, the type of brand (store brands vs. name brands), and the type of production (conventional vs. organic). We present the results of these secondary analyses as [supplementary material (Tables S2--S4)](#app1-nutrients-11-00864){ref-type="app"}. Thus, one of the factors that seems to have a decisive influence on the content of both trace elements and toxic elements of the mussels is their geographical origin. In this research, we sampled mussels from three different geographical regions: Galicia (Spain), Chile, and New Zealand, and statistically significant differences were found. Thus, the mussels of New Zealand presented the highest levels not only of Fe, Mn, Mo, Cr, Co, and Ni but also of As, Cd, Hg, Al, Ba, Sr, Th, Tl, and the sum of REEs. This latter case is particularly striking since the levels of these elements in New Zealand mussels is around six times higher than those from the other origins. On the other hand, Galician mussels presented much higher levels of Pb and of Zn, Cu, Ag, Be, and Sn, although in this latter case it should be taken into account that canned mussels from Galician origin are over-represented in this study, given their high presence in the Spanish market ([Table S2](#app1-nutrients-11-00864){ref-type="app"}), and, as said above, Sn content is higher in canned mussels. With mussels produced in Galicia, we could also make a comparison between those that were used for canning in store brands and those canned under name brands, and we also found several statistically significant differences between the two types of branding. Thus, store brands presented significantly higher levels of not only 4 trace elements (Fe, Zn, Ni, and Cr) but also of two toxic elements (As and Hg) compared with those of the mussels of name brands ([Table S3](#app1-nutrients-11-00864){ref-type="app"}). Pb levels were also slightly higher in store brands as well, although this difference did not reach statistical significance. On the contrary, name brands presented significantly higher levels of Ag, Ba, Be, Th, and the sum of REEs compared with those of the store brands. Finally, it is very interesting to note that we also observed that the concentrations of most of the elements (including the toxic As, Hg, Pb, Ag, Ba, and U) were significantly higher in the mussels of conventional production than in those of organic production ([Table S4](#app1-nutrients-11-00864){ref-type="app"}). The only exception was Al, which was significantly more concentrated in organically produced mussels than in conventionally produced ones. Additionally, in this case, we made the comparison between mussels from Galicia (in this case, all of them were frozen mussels), given that among those of other origins or modes of conservation we did not find organic brands in the Spanish market. According to the legal limits, the only element for which the maximum residue limits were exceeded was Cd (EU-MRL = 1 mg/kg ww \[[@B26-nutrients-11-00864]\]). This limit was exceeded in four samples of frozen mussels from Chile, whereas the other four samples of frozen mussels, also from Chile, reached around 90% of this MRL. In relation to Hg, none of the samples exceeded the MRL established in the EU (0.5 mg/kg ww \[[@B27-nutrients-11-00864]\]), and all the samples analyzed were well below this value. The most contaminated sample in the whole series barely reached 10% of this legal limit. Regarding the Pb content, all the concentrations were below the established limit (1.5 mg/kg ww \[[@B28-nutrients-11-00864]\]), and the most contaminated sample did not even reach the ½ EU-LMR. No legal limits of As nor the rest of the elements studied have been established in the EU for mussels, so we cannot put the levels found for these elements in a legal context. 3.2. Estimated Daily Intake of Essential and Toxic Elements and Risk Assessment {#sec3dot2-nutrients-11-00864} ------------------------------------------------------------------------------- The results of the estimation of the daily intake (EDI) of essential elements are presented in [Table 4](#nutrients-11-00864-t004){ref-type="table"}, and those of toxic elements in [Table 5](#nutrients-11-00864-t005){ref-type="table"} and [Table 6](#nutrients-11-00864-t006){ref-type="table"}. The estimation can be considered as quite accurate, since the partial contributions of the different types of mussels to the Spanish diet have been taken into account. This is very important, since, as we have seen in the previous section, there are notable differences in the concentrations of elements, depending on how they are conserved, so the exposure should be calculated on the basis of the real consumption of each type of mussels. Given that the differences in the consumption data between men and women (or boys and girls) are minimal \[[@B19-nutrients-11-00864]\], we considered it unnecessary to include the comparison between sexes and have presented the results for adults and children, in general, without the consideration of sex. It is noteworthy that the normalized daily intakes per kilo of body weight are practically identical in both adults and children, because the annual consumption of mussels is also practically half in this age group (10.4 kg/year in children vs. 20.5 kg/year in adults), but the average weight of children is half as well. Even so, small differences are observed since the proportions of consumption of the different types of mussels vary between children and adults (adults consume more canned mussels than fresh mussels; in children, however, it is the reverse). As summarized in [Table 4](#nutrients-11-00864-t004){ref-type="table"}, [Table 5](#nutrients-11-00864-t005){ref-type="table"}, and [Table 6](#nutrients-11-00864-t006){ref-type="table"}, we have considered two groups of consumers: those in the average consumption and large consumption groups, considering as such those in P~97.5~. It can be observed that in the latter the exposure to all the elements is approximately three times that in the average consumer, as the mussel consumption is almost triple as well. As is logical, regardless of the level of consumption of mussels, the exposure to elements in quantitative terms faithfully follows the concentrations found in the mussels, except in the case of arsenic. That is to say, the highest EDIs are those of Zn, Fe, and Al, followed at a distance by Sr, Cu, and Mo. As EDI does not follow the order of concentrations, although in the previous section we saw that As is one of the most abundant elements in mussels, due to the fact that for the EDI we only considered the fraction of inorganic As (estimated at 3.5% as indicated by the EFSA \[[@B29-nutrients-11-00864]\]). However, the most relevant findings arise when the EDIs are compared to dietary or toxic reference levels, which are shown in [Figure 1](#nutrients-11-00864-f001){ref-type="fig"} and [Figure 2](#nutrients-11-00864-f002){ref-type="fig"} (essential and toxic elements, respectively). The average adult consumer of mussels obtains between 2 (Mn) and 70% (Se) of the daily requirements of essential elements through the consumption of mussels. In the case of children, this contribution is superior in all cases. In neither of the two age groups does the average consumption of this food seem to represent a very high risk due to exposure to As, Cd, Hg, or Pb, nor any of the rest of the toxic or potentially toxic elements studied. However, it is important to note that, in children, the levels of exposure to inorganic As and Cd would reach around 26% and 22% of the reference levels, respectively ([Figure 1](#nutrients-11-00864-f001){ref-type="fig"}A and [Figure 2](#nutrients-11-00864-f002){ref-type="fig"}A). It is also noteworthy that, in those with high mussel consumption, these levels of exposure multiply, reaching almost 77% of the reference values for As in children and 62% of that of Cd ([Figure 1](#nutrients-11-00864-f001){ref-type="fig"}B and [Figure 2](#nutrients-11-00864-f002){ref-type="fig"}B). 4. Discussion {#sec4-nutrients-11-00864} ============= This study represents a very accurate approximation of the risk--benefit relationship of exposure to trace elements through the consumption of mussels in the Spanish population. An exhaustive sampling of mussels has been performed, attempting to reflect all the possible varieties that are acquired by consumers, and a comprehensive determination of the content of elements has been made, covering all the essential elements, a very large group of well-known toxic elements \[[@B30-nutrients-11-00864]\], and even those that are currently being considered as potentially toxic or at least emerging contaminants of concern (the REEs and other MEs related to the high technology industry) \[[@B18-nutrients-11-00864]\]. Firstly, regarding sampling, our main interest has been to evaluate as closely as possible the exposure of Spanish consumers to elements. Numerous studies have described differences in the levels of elements in mussels due to a multitude of variables. It should be considered that elements in a marine environment have a more complex distribution than organic pollutants and reflect more faithfully local anthropogenic inputs, natural sources, and hydrological conditions \[[@B31-nutrients-11-00864]\]. Therefore, one of the most important of these variables is the geographical origin of the said mussels, which in turn is related to the water quality of these regions \[[@B32-nutrients-11-00864],[@B33-nutrients-11-00864],[@B34-nutrients-11-00864],[@B35-nutrients-11-00864],[@B36-nutrients-11-00864],[@B37-nutrients-11-00864],[@B38-nutrients-11-00864]\]. However, according to food consumption surveys, geographic origin is not the most important criterion that guides the purchase of mussels by Spanish consumers, but their type of conservation. For this reason, in the design of the sampling, we tried to reflect the different available varieties of the three types of mussels that are chosen by the consumers of this country: fresh, canned, and frozen. Very few studies have compared the differences in the element content according to the mode of conservation of mussels \[[@B15-nutrients-11-00864],[@B16-nutrients-11-00864]\], and as far as we know, none has included these three types of mussels. In the mentioned papers, the authors investigated the differences in the content of some essential (Cu, Mn, Se, and Zn) and toxic (Hg, Cd, Pb, Ag, and As) elements in a wide range of fresh, preserved, and frozen fishery products. However, these authors did not include frozen mussels, and the number of fresh and canned mussel samples studied was very small (*n* = 11 and 12, respectively) \[[@B15-nutrients-11-00864],[@B16-nutrients-11-00864]\]. The results of these studies showed that there are significant differences in the concentrations of elements of fishery products depending on their mode of conservation, including mussels. Our results confirm that the differences according to the mode of conservation are remarkable in the case of this mollusk, preserved mussels being those that in general have lower concentrations of elements. This is possibly related to a smaller size and age of the mussels that are used in the bulk of the canning industry (12 to 16 medium-size mussels per can is the most usual form of marketing) since most metals bioaccumulate throughout life and lower concentrations are expected in the earlier stages of life (medium-small size mussels). However, regardless of the size, canned mussels have the highest levels of Al and Sn, probably as a result of contamination from the packaging in metal cans containing these elements. Nevertheless, although we have focused mainly on the conservation mode of the mussels when building our exposure model, other variables such as geographical origin, mode of production, and type of brand were recorded as well. With these variables, we could not force theoretical models of consumption (for example, consumers of only organic products or consumers of only store brands) because not all varieties are available, and therefore they could not all be sampled. However, we would like to highlight some of the results obtained for some of these variables, since such results have never been reported. As far as we know, this study is the first to compare canned mussels of name brands with those of store brands. Surprisingly, although all mussels compared had the same origin (Galicia, Spain), we found significant differences. Mussels of store brands contain slightly higher levels of all the elements (both essential and toxic). This could indicate that mussels of slightly lower quality are used to make these cheaper brands, although the differences are scarcely relevant. The opposite occurs when we compare mussels of conventional production with those of organic production (all of them frozen and of name brands), with the mussels of organic production being those that present lower contents of elements (either essential or toxic). This could be linked to the conditions of the controlled production that organic mussel farming must meet for its certification requirements, which are, for example, a controlled source of the seed, a production unit below 500 rafts and always at a depth of less than 20 m, and the prohibition of some paints \[[@B39-nutrients-11-00864]\]. Thereby, more studies should be carried out to understand and confirm those differences since this type of production is in continuous development and economic growth. According to the exposure estimates for our typical consumer (a Spanish consumer that eats 50% fresh mussels, 40% canned mussels, and 10% frozen mussels), we can summarize that for the average consumption there is a good balance between a moderate to high contribution of trace elements and a moderately low contribution of toxic elements. However, the estimates for toxic elements, in particular As and Cd, can be worrisome, especially in the high percentile of consumption. Thus, with respect to essential elements, the contribution of mussels to the intake of Se is particularly noteworthy. A regular consumption of this food would contribute almost 70% of the requirements of Se of adults and 150% of those of children. The contribution of Se from mussels is so striking that children who are large consumers would intake up to 500% of their daily nutritional requirements. This contribution would be even higher (more than double) if only fresh mussels were consumed, surpassing around 3--10 times the DRV for this element. However, this contribution is still far from reaching the UL for this element, so it would not pose a real problem of toxicity. This finding is important as Se is a trace element required for different biological functions and is increasingly considered to be a key nutraceutical component. Thus, selenoproteins play a variety of functions, including antioxidant effects, T-cell immunity and implications in the thyroid hormone, and skeletal and cardiac metabolism \[[@B40-nutrients-11-00864]\]. In terms of percentage of DRVs, the second element would be Co. The nutritional requirements of Co are not high (of the order of 0.1 µg/kg/day); it is even considered a toxic heavy metal which can cause toxic cardiomyopathy or polycythemia when the exposure to it is excessive \[[@B41-nutrients-11-00864]\]. The nutritional contribution of Co is fundamentally associated with the vitamin B~12~ content of the food. Therefore, given that it has been widely described that mussels are rich in vitamin B~12~ \[[@B42-nutrients-11-00864]\], it is not surprising that the average consumption of these molluscs represents a contribution of around 40% of the daily requirements of Co. With regard to the rest of the essential elements, mussel consumption is also a high source of Zn, Mo, and Fe. For Zn, a moderate consumption of mussels contributes 20% of the nutritional requirements in adults and up to 35% in children. Zn is a key component of cells and plays a role in the mechanism of action of several crucial enzymes, some of them implicated in the binding of RNA molecules and protein--protein interactions \[[@B43-nutrients-11-00864]\]. In the case of Mo, mussel consumption contributes around 15% of the daily requirement in adults and up to 40% in children. This contribution would be even higher (almost double) if only preserved mussels were consumed. Mo is an essential component of certain enzymes that catalyze redox reactions and is also required for enzymes involved in the metabolism of aromatic aldehydes and the catabolism of amino acids \[[@B44-nutrients-11-00864]\]. With regard to the contribution of mussels to Fe intake, these account for around 15% of the nutritional requirements in both adults and children. The wide range of roles played by this metal is well known, as it is essential for the maintenance of basic life functions in the form of hemoglobin and is also necessary for electron transfer, oxidase activities, and energy metabolism \[[@B45-nutrients-11-00864]\]. On the contrary, mussels represent only a discrete source of Cu or Mn (less than 7% and 4% of their respective DRVs). Besides these seven essential elements, the exposure to 36 other toxic or potentially toxic elements through the consumption of mussels was assessed. In general terms, it can be considered that, for the average consumer, mussels pose a very moderate risk of exposure to toxic elements, As, Cd, and Hg being those that reach higher levels. One of the most worrying values is that of the Cd, since, although an average consumer would be exposed to 20% of Cd-TDI, by virtue of the concentrations found, a child consumer who eats only frozen mussels could be exposed to this element so that it reaches up to 40% or even 90%, whether in an average or a high consumer of mussels, respectively. Obviously, this would imply that this type of consumer would greatly exceed the TDI of Cd if the total diet is considered. Something similar happens with As. Our model assumed that 3.5% of the As measured in the mussels is in the most toxic form, the inorganic As \[[@B29-nutrients-11-00864]\]. With these estimated values, an average consumer of mussels (whether a child or an adult) would be exposed to approximately 23--25% of the TDI of this element, which, although being relatively high value, is far from representing a real problem of public health. However, it should be noted that mussel consumers in the high percentile would be exposed to up to 75% of the toxicity reference value for this element, and could even surpass it if they only consumed fresh or frozen mussels. The other toxic element that can be worrisome is Hg, as it has also been described in fish \[[@B46-nutrients-11-00864],[@B47-nutrients-11-00864]\]. Furthermore, in this case, we have made an assumption, which is that 80% of the total measured is in the most toxic form (methylmercury) \[[@B48-nutrients-11-00864]\]. In that case, an average consumer would reach around 10% of the TDI, but this value would increase more than three times (\>33% TDI) in individuals with high mussel consumption, which is especially worrisome in children, given the special vulnerability of their developing nervous system to the toxic effects of this heavy metal \[[@B48-nutrients-11-00864]\]. Of the rest of the toxic elements investigated, none exceeds 3% of their respective TDI. Among them, there are elements whose toxicity has been demonstrated, and in fact, they have been included in the priority list published every two years by the ATSDR \[[@B30-nutrients-11-00864]\] as elements that are also of increasing concern because of their growing appearance as emerging environmental pollutants, as is the case of the REEs \[[@B18-nutrients-11-00864]\]. For many of these, evidence of their toxicity is being provided by many researchers around the world \[[@B49-nutrients-11-00864],[@B50-nutrients-11-00864],[@B51-nutrients-11-00864]\]. Recent studies have shown that mussels can accumulate these kinds of elements as well \[[@B8-nutrients-11-00864],[@B9-nutrients-11-00864]\] and might even be adversely affected by their presence \[[@B52-nutrients-11-00864]\]. In any case, according to the results of this study, neither the REEs nor the toxic elements included in the ATSDR list pose a relevant risk associated with the consumption of mussels. Finally, it is important to note that, although the risks estimated in this study are not high, in this model we have only considered mussel intake, and this should be taken into account in the global risk assessment. Obviously, mussels are not the only pathway to the exposure to harmful elements in humans, but other relevant foods and other sources, such as soil ingestion, dust inhalation, and dermal contact, should be considered as well. The actual risk is the sum of all the exposure pathways and can be much higher than the ones obtained in this study. 5. Conclusions {#sec5-nutrients-11-00864} ============== For the great majority of elements, there are significant differences in the levels of elements depending on the mode of conservation of the mussels. In general, we found that canned mussels were the ones that presented the lowest concentrations of essential and toxic elements, except Al and Sn, which are two elements employed in the manufacture of cans. We also found significant differences according to the geographic origin, the mode of production, and the type of brand. These are interesting variables, but perhaps they condition less the consumer's decision when buying mussels (and consequently also their exposure to essential and toxic elements) than they do the mode of preservation of this mollusk. Our exposure model indicates that the average consumption of mussels by the Spanish population (around 20 kg/year in adults and 10 kg/year in children) represents a valuable contribution of essential elements, particularly Se, Co, Zn, Fe, and Mo, without supposing a high risk derived from the exposure to toxic elements. However, there are some worrying aspects, such as the relatively high levels of Cd and As, which might cause high percentages in the reference values for toxicity that could be reached in the case of high consumption of these mollusks. This suggests that a moderate consumption of mussels should be recommended. The following are available online at <https://www.mdpi.com/2072-6643/11/4/864/s1>. Table S1: Concentrations of rare earth elements and other minority elements in preserved, fresh, and frozen mussels. Median values are provided and expressed in ng/g~fresh\ weight~. Table S2: Concentrations of essential and toxic elements in frozen mussels from three different production areas. Median values are provided and expressed in ng/g~fresh\ weight~. Table S3: Concentrations of essential and toxic elements in preserved mussels from name and store brands. Median values are provided and expressed in ng/g~fresh\ weight~. Table S4: Concentrations of essential and toxic elements in frozen mussels produced under two different production methods. Median values are provided and expressed in ng/g~fresh\ weight~. ###### Click here for additional data file. conceptualization: O.P.L. and M.Z.; methodology: O.P.L., Á.R.-H., and M.Z.; validation: O.P.L., Á.R.-H., and M.Z.; formal analysis: M.Z. and Á.R.H.; investigation: L.D.B. and L.A.H.-H.; data curation: O.P.L. and L.A.H.-H.; writing---original draft preparation: Á.R.-H.; writing---review and editing: O.P.L.; visualization: L.D.B.; supervision: O.P.L.; project administration: L.D.B. and O.P.L. This research received no external funding. The authors declare no conflict of interest. {#nutrients-11-00864-f001} {#nutrients-11-00864-f002} nutrients-11-00864-t001_Table 1 ###### Sampling of the mussels according to the type of conservation, the production area, the type of brand, and the method of production. Galicia Chile New Zealand ---------------------- --------- ------- ------------- ---- ---- ---- ---- ---- ---- ---- ---- ---- Preserved (*n* = 88) 38 34 72 \- 16 \- 16 \- \- \- \- Frozen (*n* = 80) 36 \- 28 8 34 \- 32 \- 12 \- 12 Fresh (*n* = 20) 40 \- 40 \- \- \- \- \- \- \- \- \- nutrients-11-00864-t002_Table 2 ###### Concentrations of essential in mussels from different types of conservation in the Spanish market. Results are expressed in ng/g. --------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- Preserved Mussels\ Frozen Mussels\ Fresh Mussels\ *p* Value ^a^ (*n* = 88) (*n* = 80) (*n* = 20) -------- --------------------- ----------------- -------------------- --------------------- ---------- -------------------- --------------------- ---------- -------------------- --------- --------- --------- **Fe** 34,945.1 ± 11,496.8 33,376.0 25,848.3--41,727.6 45,965.2 ± 44,720.5 32,310.0 26,713.1--50,700.1 24,427.6 ± 5553.9 22,924.9 20,137.1--26,850.9 n.s. \<0.01 \<0.01 **Zn** 46,011.4 ± 24,034.8 38,961.6 33,105.8--48,439.9 47,141.4 ± 17,788.6 46,231.2 37,356.5--58,914.3 56,624.1 ± 14,019.9 54,712.7 45,823.5--70,423.9 \<0.01 \<0.005 \<0.05 **Cu** 1332.5 ± 383.1 1268.8 1028.4--1492.1 1564.6 ± 340.9 1529.7 1349.5--1831.3 1235.9 ± 195.5 1313.7 1035.5--1405.7 \<0.05 n.s. \<0.01 **Se** 646.0 ± 200.8 599.6 538.3--683.3 1075.9 ± 187.9 1085.6 949.0--1229.2 1223.8 ± 125.1 1274.6 1129.5--1315.2 \<0.005 \<0.005 n.s. **Mn** 1251.0 ± 261.8 1227.7 1057.9--1407.0 1726.8 ± 812.5 1424.4 1170.0--2079.5 1479.8 ± 298.8 1569.1 1249.4--1702.4 \<0.05 \<0.01 \<0.05 **Mo** 271.4 ± 279.9 151.3 120.3--288.1 187.6 ± 113.6 142.4 106.7--208.5 187.4 ± 45.7 201.9 183.4--216.0 n.s. n.s. n.s. **Co** 40.8 ± 11.0 40.6 33.6--46.5 66.4 ± 67.2 50.2 42.2--75.1 74.2 ± 16.5 74.1 58.9--91.0 \<0.01 \<0.005 \<0.005 --------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- ^a^ Mann--Whitney *U* test; n.s. non-significant. nutrients-11-00864-t003_Table 3 ###### Concentrations of toxic elements in mussels from different types of conservation in the Spanish market. Results are expressed in ng/g. ----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- Preserved Mussels\ Frozen Mussels\ Fresh Mussels\ *p* Value ^a^ (*n* = 88) (*n* = 80) (*n* = 20) ----------------------------------------------- --------------------- ----------------- -------------------- ---------------------- ---------- -------------------- --------------------- ---------- ------------------ --------- --------- --------- **Major Toxic Elements** **As** 1869.8 ± 634.8 1641.9 1522.5--1973.1 4008.1 ± 59.2 3377.3 3017.5--4992.0 3696.5 ± 432.8 3785.0 3395.5--3994.6 \<0.005 \<0.005 \<0.05 **Cd** 241.9 ± 92.2 232.1 169.1--312.8 463.3 ± 306.4 353.3 252.1--606.6 281.0 ± 89.4 275.3 251.5--335.7 \<0.01 \<0.05 \<0.005 **Hg** 12.4 ± 9.2 11.2 6.4--14.4 14.9 ± 8.6 16.2 6.3--20.6 23.2 ± 3.6 22.5 20.2--26.6 \<0.01 \<0.001 \<0.01 **Pb** 167.5 ± 87.0 184.5 137.3--221.8 176.8 ± 181.0 111.5 21.1--271.9 249.4 ± 64.8 255.9 190.7--296.2 \<0.01 \<0.001 \<0.001 **Other Toxic or Potentially Toxic Elements** **Ag** 5.9 ± 2.9 5.4 3.8--7.4 7.6 ± 3.6 7.1 5.1--8.6 6.2 ± 2.3 5.8 4.6--7.5 n.s. n.s. n.s. **Al** 49,278.1 ± 36,202.8 38,878.3 22,886.8--70,388.9 76,417.8 ± 11,1251.9 35,160.6 24,504.6--87,869.5 30,808.6 ± 31,245.7 21,122.7 7570.1--37,324.1 \<0.01 \<0.001 \<0.001 **Ba** 430.0 ± 303.2 318.4 202.8--618.4 708.2 ± 1259.2 353.9 190.0--659.4 238.6 ± 169.6 194.8 138.2--281.2 \<0.01 \<0.001 \<0.001 **Be** 3.0 ± 2.2 2.7 1.1--4.2 3.3 ± 3.9 2.4 0.8--4.3 3.1 ± 1.7 2.8 1.7--4.1 n.s. n.s. n.s. **Cr** 110.4 ± 47.5 101.3 78.0--135.6 109.4 ± 113.6 91.0 61.7--136.4 123.7 ± 28.9 115.5 108.9--129.5 \<0.05 \<0.05 \<0.05 **Ni** 106.2 ± 29.4 103.2 86.4--114.6 140.1 ± 37.5 125.7 95.6--171.9 196.2 ± 59.8 181.1 164.2--235.8 \<0.01 \<0.001 \<0.005 **Pd** 0.07 ± 0.04 0.06 0.04--0.08 0.08 ± 0.02 0.08 0.07--0.1 0.1 ± 0.04 1.1 0.8--4.3 n.s. n.s. n.s. **Sb** 7.3 ± 26.3 1.0 0.05--1.7 1.2 ± 0.5 1.1 0.08--2.1 3.8 ± 1.3 3.5 2.8--4.6 n.s. \<0.01 \<0.005 **Sn** 26.2 ± 13.2 24.0 16.7--33.1 9.6 ± 9.3 6.0 1.5--18.0 11.7 ± 6.0 9.6 7.1--15.5 \<0.001 \<0.001 \<0.05 **Sr** 6403.4 ± 3618.4 5764.6 5006.4--6607.25 7169.8 ± 1352.8 6871.1 6293.6--7666.4 7903.9 ± 2478.6 8837.4 6395.1--9575.5 \<0.01 \<0.001 \<0.005 **Th** 7.5 ± 5.9 6.3 2.5--10.8 13.5 ± 24.2 5.5 2.0--13.7 8.3 ± 5.4 6.9 4.6--9.7 \<0.05 n.s. \<0.01 **Tl** 1.4 ± 0.6 1.3 0.9--1.8 2.6 ± 2.4 1.8 1.2--2.5 1.3 ± 0.7 1.1 0.8--1.6 n.s. n.s. n.s. **U** 34.5 ± 8.9 33.3 28.5--39.9 36.7 ± 12.3 35.4 29.2--41.8 34.3 ± 14.3 31.8 22.3--48.1 n.s. n.s. n.s. **Sum REE ^b^** 184.8 ± 133.9 144.9 102.1--247.1 330.3 ± 412.1 198.7 155.5--276.2 1913.2 ± 958.3 1570.6 1248.2--2485.42 \<0.01 \<0.001 \<0.001 ----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- ^a^ Mann--Whitney *U* test; ^b^ This is the sum of the individual concentrations of Ce, Dy, Er, Eu, Ga, Gd, Ho, In, La, Lu, Nb, Nd, Pr, Sm, Ta, Tb, Tm, Y, and Yb; ; n.s. non-significant. nutrients-11-00864-t004_Table 4 ###### Estimated daily intake of essential elements through the consumption of mussels by adults and children. ----------------------- -------------------------------------------------------- --------------------------- ---------------------------------- **Adults (\>17 y.o.)---68.48 kg/bw---Both Genders** **Essential Element** **Dietary Reference Value ^a^** **EDI Average Consumer**\ **EDI High Consumer (P~97.5~)**\ **(µg/kg bw/day) ^b^** **(µg/kg bw/day) ^c^** **Fe** 160.63 ^d^ 24.89 74.41 **Zn** 175.23 ^d^ 34.45 105.54 **Cu** 18.98 ^e^ 1.09 3.33 **Se** 1.02 ^e^ 0.69 2.18 **Mn** 43.81 ^e^ 1.06 3.25 **Mo** 0.95 ^e^ 0.14 0.42 **Co** 0.12 ^d^ 0.04 0.12 **Children (7--12 y.o.)---34.48 kg/bw---Both Genders** **Essential Element** **Dietary Reference Value ^a^** **EDI Average Consumer**\ **EDI High Consumer (P~97.5~)**\ **(µg/kg bw/day) ^b^** **(µg/kg bw/day) ^c^** **Fe** 160.63 ^d^ 23.57 62.72 **Zn** 108.06 ^d^ 35.34 99.31 **Cu** 14.60 ^e^ 1.11 3.10 **Se** 0.51 ^e^ 0.77 2.29 **Mn** 21.90 ^e^ 1.09 3.05 **Mo** 0.37 ^e^ 0.14 0.40 **Co** 0.12 ^d^ 0.04 0.12 ----------------------- -------------------------------------------------------- --------------------------- ---------------------------------- ^a^ For comparison purposes, the DRVs have been expressed in µg/kg·bw/day. ^b^ A consumption of 31.94 g/day of preserved mussels, 22.44 g/day of fresh mussels, and 1.87 g/day of frozen mussels in adults or 11.17 g/day of preserved mussels, 15.96 g/day of fresh mussels, and 1.32 g/day of frozen mussels in children. ^c^ A consumption of 87.98 g/day of preserved mussels, 76.8 g/day of fresh mussels, and 6.4 g/day of frozen mussels in adults or 21.86 g/day of preserved mussels, 52.6 g/day of fresh mussels, and 4.4 g/day of frozen mussels in children. ^d^ Population reference intake is the term employed by the European Food Safety Authority (EFSA) instead of the recommended dietary allowance (RDA) and represents the daily dietary intake level of a nutrient considered sufficient to meet the requirements of 97.5% of healthy individuals in each life-stage and sex group. ^e^ Adequate intake as defined by the EFSA. It represents the amount established that is firmly believed to be adequate for everyone in the demographic group where no recommended dietary allowance (RDA) has been established. When a range of different values for males and females was found, we considered the average value. y.o. years old. nutrients-11-00864-t005_Table 5 ###### Estimated daily intake of major toxic elements through the consumption of mussels in adults and children. ------------------- -------------------------------------------------------- --------------------------- ---------------------------------- **Adults (\>17 y.o.)---68.48 kg/bw---Both Genders** **Toxic Element** **Risk Reference Value ^a^** **EDI Average Consumer**\ **EDI High Consumer (P~97.5~)**\ **(µg/kg bw/day) ^b^** **(µg/kg bw/day) ^c^** **iAs ^d^** 0.30 0.07 0.22 **Cd** 1.00 0.21 0.65 **MeHg ^e^** 0.10 0.01 0.03 **Pb** 6.00 0.16 0.49 **Children (7--12 y.o.)---34.48 kg/bw---Both Genders** **Toxic Element** **Risk Reference Value ^a^** **EDI Average Consumer**\ **EDI High Consumer (P~97.5~)**\ **(µg/kg bw/day) ^b^** **(µg/kg bw/day) ^c^** **iAs ^d^** 0.30 0.08 0.23 **Cd** 1.00 0.21 0.62 **MeHg ^e^** 0.10 0.01 0.03 **Pb** 6.00 0.16 0.46 ------------------- -------------------------------------------------------- --------------------------- ---------------------------------- ^a^ We have employed as risk reference values (RRVs) the non-carcinogen tolerable daily intake (TDI) values from the US EPA. For comparison purposes, the RRVs have been expressed in µg/kg bw/day. ^b^ A consumption of 31.94 g/day of preserved mussels, 22.44 g/day of fresh mussels, and 1.87 g/day of frozen mussels in adults or 11.17 g/day of preserved mussels, 15.96 g/day of fresh mussels, and 1.32 g/day of frozen mussels in children. ^c^ A consumption of 87.98 g/day of preserved mussels, 76.8 g/day of fresh mussels, and 6.4 g/day of frozen mussels in adults or 21.86 g/day of preserved mussels, 52.6 g/day of fresh mussels, and 4.4 g/day of frozen mussels in children. ^d^ Although in this work we have not made arsenic speciation, we have taken the scientific consensus with respect to arsenic in seafood as a reference, and considered that the inorganic arsenic content in mussels is 3.5% of the As measured. ^e^ We were not able to perform mercury speciation, but it has been established that in fish and seafood the percentage of the methyl form normally seems to vary between 50 and 80%. We were conservatively estimated that 80% of the total mercury in the mussels is in the form of methylmercury. y.o. years old. nutrients-11-00864-t006_Table 6 ###### Estimated daily intake of other toxic or potentially toxic elements through the consumption of mussels in adults and children. ----------------- -------------------------------------------------------- --------------------------- ---------------------------------- **Adults (\>17 y.o.)---68.48 kg bw---Both Genders** **Element** **Risk Reference Value ^a^** **EDI Average Consumer**\ **EDI High Consumer (P~97.5~)**\ **(µg/kg bw/day) ^b^** **(µg/kg bw/day) ^c^** **Ag** 5 0.00 0.01 **Al** 1000 26.48 78.21 **Ba** 200 0.22 0.66 **Be** 2 0.00 0.01 **Cd** 3,00 0,09 0,27 **Ni** 20,00 0,10 0,32 **Pd** N.A. 0.00 0.00 **Sb** 0.4 0.00 0.01 **Sn** 600 0.01 0.04 **Sr** 600 5.77 17.96 **Th** NA 0.01 0.02 **Tl** 0.07 0.00 0.00 **U** 3 0.03 0.08 **Sum REE** 61 0.59 1.97 **Children (7--12 y.o.)---34.48 kg bw---Both Genders** **Element** **Risk Reference Value ^a^** **EDI Average Consumer**\ **EDI High Consumer (P~97.5~)**\ **(µg/kg bw/day) ^b^** **(µg/kg bw/day) ^c^** **Ag** 5 0.00 0.01 **Al** 1000 24.04 61.99 **Ba** 200 0.21 0.54 **Be** 2 0.00 0.01 **Cd** 3.0 0.09 0.25 **Ni** 20.00 0.11 0.31 **Pd** NA 0.00 0.00 **Sb** 0.4 0.00 0.01 **Sn** 600 0.01 0.03 **Sr** 600 6.22 18.01 **Th** NA 0.01 0.02 **Tl** 0.07 0.00 0.00 **U** 3 0.03 0.07 **Sum REE ^d^** 61 0.78 2.51 ----------------- -------------------------------------------------------- --------------------------- ---------------------------------- ^a^ We have employed as RRVs the non-carcinogen TDI values from US EPA. For comparison purposes, the RRVs have been expressed in µg/kg bw/day. The REEs have been considered as a sum and compared to the TRV proposed by other authors \[[@B24-nutrients-11-00864],[@B25-nutrients-11-00864]\]. ^b^ A consumption of 31.94 g/day of preserved mussels, 22.44 g/day of fresh mussels, and 1.87 g/day of frozen mussels in adults or 11.17 g/day of preserved mussels, 15.96 g/day of fresh mussels, and 1.32 g/day of frozen mussels in children. ^c^ A consumption of 87.98 g/day of preserved mussels, 76.8 g/day of fresh mussels, and 6.4 g/day of frozen mussels in adults or 21.86 g/day of preserved mussels, 52.6 g/day of fresh mussels, and 4.4 g/day of frozen mussels in children. ^d^ This is the sum of the individual concentrations of Ce, Dy, Er, Eu, Ga, Gd, Ho, In, La, Lu, Nb, Nd, Pr, Sm, Ta, Tb, Tm, Y, and Yb.

1. Introduction {#sec1-molecules-24-03622} =============== Doxorubicin (DOX), as an anti-tumor drug widely used in clinical treatment, has an obvious curative effect on various tumors, including leukemia, malignant lymphoma, and various solid tumors \[[@B1-molecules-24-03622],[@B2-molecules-24-03622],[@B3-molecules-24-03622]\]. However, its long-term or high-dose clinical use often leads to irreversible congestive heart failure, which limits its application \[[@B4-molecules-24-03622],[@B5-molecules-24-03622]\]. In previous studies, the cardiotoxicity may be related to the formation of free radicals, lipid peroxidation, Ca^2+^ overloading, and the activation of apoptotic factors \[[@B6-molecules-24-03622],[@B7-molecules-24-03622]\]. In order to find an effective way to solve this problem, many research studies have been carried out, but the results were not satisfactory \[[@B8-molecules-24-03622]\]. The multidrug resistance (MDR) of DOX is a major obstacle to its application in tumor chemotherapy. Although various mechanisms are known to be involved in MDR phenotypes, the overexpression of some members of the ATP-binding cassette (ABC) protein family is considered to be a major contributor to MDR development in tumor cells \[[@B9-molecules-24-03622]\]. Hernandezine (HER), a dibenzyl isoquinoline alkaloid isolated from traditional Chinese medicine, has long been used in the treatment of hypertension \[[@B10-molecules-24-03622],[@B11-molecules-24-03622]\]. HER has been proved to be able to prevent hair cell aminoglycoside-induced injury \[[@B12-molecules-24-03622]\], inhibit protein kinase C signal events in human peripheral blood T cells \[[@B13-molecules-24-03622]\] and neuronal nicotinic acetylcholine receptors (nAChRs), \[[@B14-molecules-24-03622]\], and block non-voltage-operated Ca^2+^ entry activated by intracellular Ca^2+^ store depletion induced by thapsigargin in rat glioma C6 cells \[[@B15-molecules-24-03622]\] and in human leukemic HL-60 cells \[[@B16-molecules-24-03622]\]. In addition, HER was found to be an effective MDR modulator. Recent studies have shown that HER, as a new AMPK activator, could induce autophagy death in drug-resistant cancers \[[@B17-molecules-24-03622]\]. Furthermore, HER could effectively inhibit the transport function of ABCB1 relative to MDR-linked ABC drug transporters ABCC1 and ABCG2 to enhance the drug-induced apoptosis of tumor cells \[[@B18-molecules-24-03622]\]. In summary, HER has a selective inhibitory effect on the MDR of DOX, which could make DOX better in the treatment of cancer \[[@B17-molecules-24-03622],[@B18-molecules-24-03622]\]. Therefore, it is of great significance to study the pharmacokinetic characteristics of the two drugs in combination. The purpose of this study was to investigate the interaction between HER and DOX on the pharmacokinetics: whether HER could improve the absorption of DOX or the effect in turn, and whether HER could reduce the accumulation of DOX in myocardial tissue. At present, there are several analytical methods for determining DOX \[[@B19-molecules-24-03622],[@B20-molecules-24-03622]\], but only one article \[[@B21-molecules-24-03622]\] for HER for the pharmacokinetics of rats by LC-MS. According to the previous investigations, the sample preparation methods adopted are mainly focused on a single protein precipitation step. However, there have been no reports of the simultaneous determination of DOX and HER in rat plasma. Therefore, we developed and verified a simple, specific, and sensitive LC-MS/MS method for the simultaneous determination of DOX and HER in rat plasma, and applied it to the pharmacokinetic study of rats to evaluate the effect of HER and DOX interaction on pharmacokinetics. 2. Results and Discussion {#sec2-molecules-24-03622} ========================= 2.1. Method Development {#sec2dot1-molecules-24-03622} ----------------------- To optimize the MS conditions for detecting DOX, HER, and tetrandrine (IS), all the operational parameters were carefully optimized. The analysis showed that positive ion detection has a stronger response than negative ion detection. The MS/MS ion transition was monitored in the MRM mode to improve the specificity and sensitivity of the detection method considering the complexity of biological samples. DOX has the strongest peak at *m*/*z* 544.2→379.1, HER has the strongest peak at *m*/*z* 653.4→411.2, and IS (tetrandrine) has the strongest peak at *m*/*z* 623.3→381.3. The structures of the proposed daughter ions were given by referring to the related articles \[[@B19-molecules-24-03622],[@B20-molecules-24-03622],[@B22-molecules-24-03622]\]. The ion spectra and chemical structures of DOX, HER, and IS are shown in [Figure 1](#molecules-24-03622-f001){ref-type="fig"}. The chromatographic conditions were optimized, and a good separation effect was obtained with a sharp peak shape, high response, and short run time. The stationary phase and the composition of the mobile phase was studied. An ACQUITY UPLC BEH C~18~ Column (100 mm × 2.1 mm, 1.7 μm) was chosen in this study with good peak symmetry. Different mobile phases (acetonitrile--water and methanol--water or with different concentrations of formic acid or ammonium acetate) were investigated. The results showed that the peak symmetry and response of the acetonitrile--water system were better than those of the methanol--water system. Meanwhile, both gradient elution and isocratic elution were tested, and the result showed that isocratic elution way was more simple, fast, and did not sacrifice any sensitivity and specificity. The retention times of DOX, HER, and IS were 1.46 min, 4.37 min, and 3.65 min, respectively ([Figure 2](#molecules-24-03622-f002){ref-type="fig"}), and the total chromatographic run time was 5.0 min. In this study, a protein precipitation method was firstly considered to prepare samples, which was simple, accurate, and efficient. The extraction recovery and matrix effect were tested. Different precipitation reagents, such as acetonitrile, methanol, and acetonitrile with 0.1% formic acid, were investigated. The results showed that acetonitrile was the best choice, with a higher extraction rate and lower background interference. 2.2. Method Validation {#sec2dot2-molecules-24-03622} ---------------------- The method validation was conducted in strict accordance with US Food and Drug Administration (FDA) guidelines \[[@B23-molecules-24-03622]\], the content of which consists of selectivity and specificity, linearity, limit of quantification (LOQ), limit of detection (LOD), accuracy, precision, recovery, matrix effect, and stability. In the process of method development, a selectivity and specificity test was used to verify that the measured substance is the intended analyte to minimize or avoid interference. The selectivity and specificity test of this experiment was demonstrated by the analysis of blank plasma from six individual rats, which was examined by comparing the retention times of DOX, HER, and IS in blank plasma, the addition of DOX and HER at LOQ and IS to blank plasma, and a plasma sample 2 h after an intravenous administration of the mixture of DOX and HER. The blank plasma should be free of interference at the retention times of the analytes and the IS, and which in spiked samples and actual samples should be consistent, respectively. A typical MRM chromatogram of mixed blank plasma in rats, spiked plasma samples with DOX and HER at LOQ and the IS, and plasma samples of rats after an intravenous injection of a mixture of DOX (5 mg/kg) and HER (5 mg/kg) for 2 h is shown in [Figure 2](#molecules-24-03622-f002){ref-type="fig"}. The results showed that there was no significant endogenous interference in the retention time of the analyte under the established chromatographic conditions. Calibration curves were established by plotting the peak-area ratio (y) between analytes (DOX or HER) and IS against the nominal concentrations. Linearity was evaluated by weighted (1/*x*^2^) least squares linear regression analysis. The correlation coefficient (*r*) should be greater than 0.99, indicating a good linearity. The limit of detection (LOD) is defined as the lowest detectable concentration, judged by the signal-to-noise ratio (SNR) \>10. The limit of quantification (LOQ) was defined as the lowest concentration on the calibration curve, which represents the sensitivity of the method and should be lower than the minimum concentration in all the samples. The linear calibration curve was obtained by plotting the peak area ratio (analytes/IS) versus DOX and HER concentration. A weighted (1/*x*) quadratic least-square regression analysis gave typical regression curves. The calibration curves, correlation coefficients, detection ranges, and LOQ of DOX and HER in plasma and myocardial tissues are shown in [Table 1](#molecules-24-03622-t001){ref-type="table"}. The calibration curves had good linearity with the corresponding range of DOX and HER (r \> 0.99). Under the optimized conditions, DOX LOQ was \<4.0 ng/mL, and HER LOQ was \<2.0 ng/mL in rat plasma, judging from the signal-to-noise ratios (SNR) of \>10. Accuracy and precision tests are critical in determining whether the method is ready for validation, and involve analyzing replicate quality controls (QCs) at different concentrations throughout the assay range. Specifically, the intraday and interday precisions and accuracies were obtained by analyzing five replicates of QC samples at three levels for three consecutive days. Precision, defined as the relative standard deviation (RSD), should be within 15% at each QC level. Accuracy expressed as relative error (RE) must be within ± 15%. Except for the LOQ level, the RSD value of precision should be within 20%, and the RE value of accuracy should be within ± 20%. The intraday and interday precision of the QC samples of DOX and HER were lower than 9.3% and 5.6%, respectively. The accuracy of DOX was −14.0% to 5.5%, and the accuracy of HER was −9.0% to −0.8% (see [Table 2](#molecules-24-03622-t002){ref-type="table"}). All the assay values were within the range of acceptable variables, indicating that the established method was precise and accurate. Recovery of the analytes should be optimized to ensure that the extraction is efficient and reproducible. Recoveries of the analytes at three QC levels (*n* = 5) were determined by comparing the peak area ratios of the analytes to IS from QC samples with those of analyte solutions spiked with post-extracted matrix at equivalent concentrations. The matrix effect was examined to assess the possibility of ion suppression or enhancement. The matrix effect was measured by comparing the peak area ratios of the analytes to IS in solutions spiked with the blank processed matrix with the solutions at three QC levels. In common, it was considered that the matrix effect was obvious if the ratio was less than 85% or more than 115%. The recovery and matrix effect data of DOX and HER in rat plasma were shown in [Table 3](#molecules-24-03622-t003){ref-type="table"}. The matrix effect range of all analytes was 92.9 ± 4.3% to 112.8 ± 1.8%, and the RSD value was lower than 11.6%. The average recovery of DOX and HER at three QC levels was 88.7 ± 6.2% to 108.4 ± 4.9%, and the RSD value was lower than 7.0%. The results showed that this method had no matrix effect, and could be used for biological analysis. Stability was conducted by analyzing three replicates of the samples at three QC levels under the following conditions, including bench top stability after 4 h of exposure at room temperature, auto-sampler stability after 24 h of storage in the auto-sampler at 4 °C, freeze/thaw stability evaluated for three freeze--thaw cycles after freezing at −80 °C and thawing at room temperature, and long-term stability storage at −80 °C for 30 days. The samples were considered stable if the average percentage concentration deviation (expressed as RSD) was within 15% of the actual value. The stability results are shown in [Table 4](#molecules-24-03622-t004){ref-type="table"}. The variation of all the stability studies was less than 15.0%, which met the standard of stability measurement. Therefore, this method could be used for routine analysis. 2.3. Pharmacokinetics --------------------- This validated method has been successfully applied to the determination of plasma concentration of DOX and HER in rats. In this study, we compared the pharmacokinetic parameters of DOX in the combined treatment group with those in the single treatment group. The pharmacokinetic profiles of HER were also compared in the same way. The mean plasma concentration--time profiles of DOX and HER for the three groups were shown in [Figure 3](#molecules-24-03622-f003){ref-type="fig"}. The pharmacokinetic parameters of DOX and HER in rats following the intravenous administration of single DOX (5 mg/kg), single HER (5 mg/kg), and a combination of DOX and HER (5 mg/kg, respectively) were shown in [Table 5](#molecules-24-03622-t005){ref-type="table"}. The C~max~ of DOX in the single group and combined group was 2647 ± 650 ng/mL and 5703 ± 2980 ng/mL, respectively. Meanwhile, the AUC~0--∞~ was 1412 ± 114 ng/mL and 2453 ± 218 ng/mL, respectively. Meanwhile, the t~1/2~ was 4.6 ± 0.8 h and 4.2 ± 0.6 h, and the MRT~0--∞~ was 4.9 ± 0.9 h and 4.5 ± 0.5 h, respectively. Significant differences of C~max~ and AUC~0--∞~ of DOX were observed between the single and combined groups with equivalent doses of DOX administration, which indicated that HER could increase the absorption of DOX. However, there was no significant difference between the t~1/2~ and MRT~0--∞~ of DOX, which indicated that HER had no effect on DOX's elimination and excretion. In turn, we could see from the plasma concentration--time curves of HER in two treatment groups in [Figure 3](#molecules-24-03622-f003){ref-type="fig"} that the combination use of DOX made the pharmacokinetic behavior of HER no longer fitted to a non-compartmental model that was used to calculate the pharmacokinetic characteristics in this study. However, we were still able to reach a conclusion from the plasma concentration--time curve and the pharmacokinetic characteristic of HER that the free drug concentration of HER was reduced by the combination use of DOX. The possible reason might be the enhancement of DOX on the drug--protein binding of HER. The comparison of the accumulated concentrations of DOX in myocardial tissue 8 h after intravenous administration of single DOX and combination of DOX and HER was investigated as shown in [Figure 4](#molecules-24-03622-f004){ref-type="fig"}. A significant difference between the two groups could be observed (*p* \< 0.05), indicating that HER was able to reduce the accumulation of DOX in myocardial tissue. Meanwhile, recent studies demonstrated that doxorubicinol (DOX-ol), a secondary alcohol metabolite of DOX \[[@B24-molecules-24-03622],[@B25-molecules-24-03622]\], which may have caused cardiac toxicity by being poorly cleared from the heart and accumulating there to form a long-lived toxicant to heart \[[@B26-molecules-24-03622]\], was to blame. Therefore, the next step is to study whether HER could inhibit the conversion of DOX into DOX-ol, which might be considered as a therapeutic target for DOX-induced cardiac toxicity. 3. Experimental {#sec3-molecules-24-03622} =============== 3.1. Chemicals and Reagents {#sec3dot1-molecules-24-03622} --------------------------- DOX (purity over 99%) was obtained from Dalian Meilun Biotech Co., Ltd. (Dalian, China). HER and tetrandrine (purity over 99%) were purchased from Chengdu Biopurity Phytochemicals Ltd. (Chengdu, China). Ammonium acetate, HPLC-grade, was purchased from Dikma Company (Lake Forest, CA 92630, USA). Acetonitrile and methanol, LC-MS-grade, were purchased from Merck KGaA Company (Darmstadt, Germany). Ultra-pure water was provied using a Millipore Milli-Q system (Millipore, Bedford, MA, USA). Other chemical reagents were of analytical grade. 3.2. Animals {#sec3dot2-molecules-24-03622} ------------ Sprague--Dawley rats (male, 250 ± 20 g) were supplied by the Experimental Animal Research Center, China Medical University, China. The rats were raised in a temperature-controlled room at 24 ± 2 °C for free feeding and water intake, and the light/dark cycle was 12 h. The rats were fed for 2 weeks to adapt to the laboratory environment. All the rats fasted for 12 h before the experiment, but with water supplied freely. The protocol for animal care and use in our study (protocol number \# CMU2019194) was approved by the Institutional Animal Care and Use Committee at China Medical University. Eighteen rats were randomly divided into three groups (six rats in each group) and given intravenous treatment with different drugs: group A, DOX (5 mg/kg); group B, HER (5 mg/kg); group C, DOX + HER (5.0 mg/kg, respectively). The injection was prepared in normal saline with 0.5% *v/v* DMSO. 3.3. Instrumentation and Conditions {#sec3dot3-molecules-24-03622} ----------------------------------- The biological samples were analyzed with an Agilent series 1290 UHPLC system (Agilent Technologies, Santa Clara, CA, USA), which was coupled to an AB 3500 triple quadrupole mass spectrometer (AB Sciex, Ontario, ON, Canada) with an electrospray ionization (ESI) source. Data acquisition and instrument control were performed using the 1.6.3 version Analyst software package (AB Sciex, ON, Canada). The separation process was performed on an ACQUITY UPLC BEH C~18~ Column (100 mm × 2.1 mm, 1.7 μm, Agilent Technologies, Santa Clara, CA, USA). The column temperature was set at 40 °C. The mobile phase was composed of acetonitrile and 10 mM ammonium acetate aqueous solution (70:30, *v*/*v*) at the flow rate of 0.3 mL/min in an isocratic elution manner. The injection volume was set at 10 μL. DOX and HER were quantitatively determined with MRM in the positive ion mode. The MS condition was as follows: the ion spray voltage (IS) was set at 5500 V, the turbo spray temperature (TEM) was set at 500 °C, and the nebulizer gas and heater gas were set at 50 and 50 arbitrary units, respectively. The curtain gas (CUR) was kept at 40 arbitrary units, and the interface heater was on. The collision cell exit potential (CXP) and entrance potential (EP) were set at 7.0 V and 10.0 V, respectively. The declustering potentials (DPs) of DOX, HER, and IS were set at 160 V, 218 V, and 87 V; the collision energies (CEs) were 60 eV, 60 eV, and 27 eV, respectively. Nitrogen was used in all cases. The optimization of the MS transitions for quantification were accomplished as DOX *m*/*z* 544.2→379.1, HER *m*/*z* 653.4→411.2, and IS (tetrandrine) *m*/*z* 623.3→381.3, respectively. Moreover, the qualifier ions for DOX, HER, and IS were set at *m*/*z* 321.1, *m*/*z* 191.1, and *m*/*z* 174.1, respectively. 3.4. Preparation of Stock Solutions, Working Solutions, Calibration Standards, and Quality Control Samples {#sec3dot4-molecules-24-03622} ---------------------------------------------------------------------------------------------------------- The standard substances of the analytes were accurately weighed and dissolved in methanol to prepare the DOX and HER stock solution with the concentration of 1.0 mg/mL, respectively. The working solution for preparing calibration standards and QC samples was obtained by diluting the stock solution with acetonitrile--water (50:50, *v*/*v*). The IS working solution with a concentration of 200 ng/mL was also prepared. The stock solution and working solution were placed under 4 °C dark condition and brought to room temperature before use. The calibration standards were prepared by spiking 50 μL of rat blank plasma (or blank myocardial tissue homogenate) with 20 μL of the working solution. The concentration of DOX in rat plasma and myocardial tissue homogenate ranged from 32 to 8000 ng/mL, and HER ranged from 20 to 4000 ng/mL. Low, medium, and high quality control (QC) samples were prepared in the same way as above (40.0, 400, and 3200 ng/mL for DOX; 80.0, 800, and 4000 ng/mL for HER) in both rat plasma and myocardial tissue homogenate. Each concentration needed three replicates. 3.5. Sample Preparation {#sec3dot5-molecules-24-03622} ----------------------- In this study, DOX, HER, and IS were extracted from the biological matrix (plasma and myocardial tissue homogenate) by routine step protein precipitation. Detailed steps were as follows: take 50 µL of biological matrix, add 20 µL of acetonitrile--water (50:50, *v*/*v*), and 20 µL of the IS solution, and add 200 µL of precipitation reagent acetonitrile, placed in a 1.5-mL EP tube. Vortex for 1 min, followed by centrifuging at 14,000 rpm for 10 min. Transfer 200 µL of supernatant to another clean 1.5-mL EP tube and centrifuge at 14,000 rpm for another 3 min. Then, an aliquot of 10 µL of the supernatant was injected into the LC-MS system for analysis. 3.6. Pharmacokinetic Study {#sec3dot6-molecules-24-03622} -------------------------- The method was used to determine the concentration--time profiles of DOX and HER in the plasma of rats after the intravenous administration of DOX (5.0 mg/kg), HER (5.0 mg/kg), and the mixture of DOX and HER (5.0 mg/kg, respectively). Blood samples (250 μL) were taken from the orbital vein at 5 min, 10 min, 15 min, 30 min, 45 min, 1 h, 1.5 h, 2 h, 2.5 h, 3 h, 4 h, 6 h, and 8 h, respectively, and were injected into heparinized 1.5-mL EP tubes. Heparin (2 mg/mL blood volume) was used as an anticoagulant for this study, and blood samples were immediately centrifuged at 14,000 rpm for 10 min at room temperature, followed by a supernatant plasma layer collected and stored at −80 °C for analysis. After the last blood sample was taken, the rats were sacrificed for cervical dislocation. The heart was removed and rinsed with cold saline to remove the superficial blood. Then, it was blotted dry with filter paper and weighed accurately. After that, the heart was homogenized with normal saline to prepare a homogenate (0.2 g/mL). All samples were stored at −80 °C for analysis. Plasma concentration--time plots were plotted, and the PK parameters were evaluated by means of non-compartmental pharmacokinetic analysis using DAS 3.2.8 pharmacokinetic program \[[@B27-molecules-24-03622]\]. The PK parameters concerned include half-life (t~1/2~), mean residence time (MRT), area under the plasma concentration--time curve (AUC), clearance (CL), etc. Data was expressed as mean ± SD. The pharmacokinetic parameters were compared using Student's t-test. Differences were considered to be significant at a level of *p* \< 0.05. 4. Conclusions {#sec4-molecules-24-03622} ============== An LC-MS method for the simultaneous determination of DOX and HER in rat plasma was established. The method is sensitive, accurate, easy to follow, and suitable for the pharmacokinetic study. This analytical method has been successfully applied to the pharmacokinetic study of DOX and HER in rats. The results of this study showed that there were significant differences in the pharmacokinetic parameters of DOX and HER after the intravenous administration of a single dose of DOX, single dose of HER, and a combination of the two. This result might help to explain the influence of DOX and HER interaction on pharmacokinetics and provide a basis for guiding clinical medication. We are very grateful to Q.C. and J.C. for making a major contribution to the revision of the article. The authors are grateful to every person who helped in this work. **Sample Availability:** Samples of the compounds doxorubicin, hernandezine and tetrandrine are available from the authors. The experiments were conceived and designed by X.-S.F. The experiments were equally performed by Y.S. and Y.Z., Chromatographic analysis of plasma and biosamples were performed by W.-P.Z., B.-Z.Z., Analysis of pharmacokinetic behavior of alnustone was finished by K.-F.W. All authors read and approved the final manuscript. This research was funded by Key Program of the Natural Science Foundation of Liaoning Province of China (No. 20170541027); Liaoning planning Program of philosophy and social science (No. L17BGL034); and Major subject in education research founded by Chinese Medical Association Medical Education Branch and China Association of Higher Education Medical Education Professional Committee (2018A-N19012). The authors declare no conflicts of interest. {#molecules-24-03622-f001} {#molecules-24-03622-f002} {#molecules-24-03622-f003} {#molecules-24-03622-f004} molecules-24-03622-t001_Table 1 ###### Calibration curves of doxorubicin and hernandezine in plasma and myocardial tissue homogenate of rats. Analytes Samples Calibration Curves Correlation Coefficients (r) Linear Ranges (ng/mL) LOQs (ng/mL) ---------- ------------------------ ------------------------ ------------------------------ ----------------------- -------------- DOX Plasma Y = 0.0047 + 0.00015x 0.994 32--8000 32 Heart Y = −0.0028 + 0.00019x 0.992 32--8000 32 HER Plasma Y = −0.0038 + 0.00298x 0.998 20--4000 20 Heart Y = −0.0197 + 0.00373x 0.994 20--4000 20 molecules-24-03622-t002_Table 2 ###### Precision and accuracy of doxorubicin and hernandezine in plasma of rats (n = 5). RSD: relative standard deviation. ---------------------------------------------------------- Analytes QC Conc.\ Intraday Interday (ng/mL) ---------- ----------- ---------- ---------- ------ ------ DOX 80 1.6 −5.7 4.3 −6.0 800 6.4 5.5 6.8 5.3 4000 4.0 −14.0 9.3 −3.6 HER 40 5.6 −5.2 4.6 −3.3 400 5.6 −0.8 3.1 −1.4 3200 1.4 −7.8 1.9 −9.0 ---------------------------------------------------------- molecules-24-03622-t003_Table 3 ###### Matrix effect and recovery of doxorubicin and hernandezine in plasma of rats (n = 5). Analytes QC Conc. (ng/mL) Matrix Effect Recovery ---------- ------------------ --------------- ------------- ------------ ----- DOX 80 112.8 ± 1.8 1.6 88.7 ± 6.2 7.0 800 95.3 ± 11.1 11.6 103.2 ± 2.6 2.5 4000 92.9 ± 4.3 4.6 95.1 ± 2.1 2.2 HER 40 104.0 ± 1.7 1.7 91.7 ± 4.3 4.7 400 94.2 ± 1.5 1.6 108.4 ± 4.9 4.6 3200 94.5 ± 1.7 1.8 93.2 ± 0.6 0.6 molecules-24-03622-t004_Table 4 ###### Stability results for doxorubicin and hernandezine in plasma of rats under different storage conditions (n = 3). ------------------------------------------------------------------------------------------------------------------------------------------ Analytes QC Conc. (ng/mL) Bench Top Stability\ Auto-Sampler Stability\ Freeze/Thraw Stability Long Term Stability\ (at Room Temperature for 4 h) (at 4 °C for 24 h) (at −80 °C for 30 days) ---------- ------------------ ------------------------------- ------------------------- ------------------------ ------------------------- DOX 80 1.6 4.3 3.4 3.1 800 6.4 6.8 6.1 9.6 4000 4.0 0.2 8.2 5.0 HER 40 5.5 4.7 4.1 1.7 400 5.5 3.1 1.6 0.8 3200 1.3 1.9 1.9 1.6 ------------------------------------------------------------------------------------------------------------------------------------------ molecules-24-03622-t005_Table 5 ###### Non-compartmental pharmacokinetic parameters of hernandezine and doxorubicin in a single doxorubicin group, single hernandezine group, and combination group (n = 6). ---------------------------------------------------------------------------------------------------- Pharmacokinetic\ Single DOX Group Single HER Group Combination Group Parameters --------------------------- ------------------ ------------------ ------------------- -------------- ***C~max~* (ng/mL)** 2647 ± 650 433.6 ± 85.2 5703 ± 2980 116.6 ± 74.0 ***T~max~* (h)** 0.083 0.083 0.083 0.083 ***Ke* (1/min)** 0.150 ± 0.03 0.090 ± 0.01 0.164 ± 0.02 \- ***t*~1/2~ (h)** 4.6 ± 0.8 7.7 ± 1.2 4.2 ± 0.6 \- **AUC~0--*t*~ (ng h/mL)** 1109 ± 102 647.2 ± 54.9 1965 ± 142.5 49.9 ± 12.5 **AUC~0--∞~ (ng h/mL)** 1412 ± 114 1154 ± 85 2453 ± 218 \- **MRT~0--∞~ (h)** 4.9 ± 0.9 10.1 ± 1.4 4.5 ± 0.5 \- ***CL*/*F* (L/kg/h)** 3.5 ± 0.5 4.3 ± 0.5 2.0 ± 0.2 \- ***Vd*/*F* (L/kg)** 23.6 ± 7.1 48.3 ± 5.6 12.4 ± 2.1 81.8 ± 2.3 ---------------------------------------------------------------------------------------------------- [^1]: These authors contributed equally to this work.

Introduction {#s0001} ============ With the advancement of personalized cancer treatment, the demand for specimen collection for genomic profiling is rapidly increasing ([@CIT0001]). In clinical practice, individualized treatments, such as molecular targeted therapies, that are selected based on the results of genomic profiling have been indicated for many types of cancer. Also, new clinical study designs, such as umbrella trials and basket trials, in which drugs are selected according to a patient's genomic profile are increasingly conducted ([@CIT0004]), and the demand for next-generation sequencing (NGS) analysis, which is capable of performing multiple genomic analyses at once, is increasing. Percutaneous needle biopsy is recognized as minimally invasive, but the tissue obtained has been considered unfit for genomic profiling because of insufficient specimen quantity compared with surgical specimens. Currently, the rate of successful genomic profiling with percutaneous needle biopsy varies from 47% to 100% ([@CIT0007]). Additionally, there are only a few reports with NGS analyses using specimens obtained by percutaneous needle biopsy ([@CIT0010],[@CIT0011]). As the role of image-guided percutaneous needle biopsy for NGS genomic profiling has not been established, we aimed to evaluate the feasibility of genomic analysis using comprehensive NGS with specimens obtained by image-guided percutaneous needle biopsy with 18-G needles. Materials and methods {#s0002} ===================== Patients {#s0003} -------- This study was a retrospective observational study employing a medical record survey. Selection criteria of the patients are the following: patients who enrolled in a clinical study of genomic profiling using a dedicated cancer gene panel for NGS (Trial of Onco-Panel for Gene-profiling to Estimate both Adverse events and Response by cancer treatment \[TOP-GEAR\] study: Clinical Study Registration No. UMIN000011141) between April 2014 and February 2017. We included patients undergoing genomic profiling with the specimen obtained with image-guided needle biopsy using interventional radiology technique. Patients who were evaluated with out-of-hospital/referred specimens, surgical specimens, and endoscopy specimens were excluded ([Figure 1](#F0001){ref-type="fig"}). This study is Health Insurance Portability and Accountability Act compliant, and the institutional review board at our institution waived the approval. All patients provided written informed consent for biopsy procedures. {#F0001} Biopsy procedure {#s0004} ---------------- Biopsy sites were selected based on the contrast enhancement on computed tomography (CT) and magnetic resonance imaging (MRI) images, MRI diffusion restriction, and increased FDG uptake on positron-emission tomography (PET) that were most likely to correspond with high viability of tumor cells. To minimize risk of complications, a puncture route that would not pass through large vessels or organs was planned ([Figure 2](#F0002){ref-type="fig"}). All biopsies were performed by board-certified interventional radiology (IR) specialists or IR fellows under supervision. Puncture was performed under local anesthesia with the guidance of angio-CT (INFX-8000C/Aquilion 16; Canon Medical Systems, Ohtawara, Japan), which is a combined CT and fluoroscopic apparatus, or ultrasonography (US) (TA 510; Canon Medical Systems, Ohtawara, Japan, or FAZONE CB; FUJIFILM, Tokyo, Japan). Images, in which the needle reached the inside of a lesion, were recorded ([Figure 2](#F0002){ref-type="fig"}). The number of specimens taken was determined according to the size of the specimens obtained and the requirement for the research. The obtained specimens were immediately fixed in 10% neutral-buffered formalin fixative and transported that day to the pathology laboratory. On-site, rapid cytology was not performed. {#F0002} Specimen processing and histopathological diagnosis {#s0005} --------------------------------------------------- Specimens were fixed in neutral-buffered formalin for 24--72 hours and embedded in paraffin. Tissue sections of 2--3-µm thickness were prepared for hematoxylin and eosin (HE) staining, and 5--10-µm-thick sections were prepared for genomic profiling. In addition to HE staining, special stainings, immunohistochemistry, and genomic tests as insurance-approved companion diagnostics were performed as necessary, and pathological diagnosis was rendered by at least two certified pathologists. For patients who consented to the TOP-GEAR study before biopsies were taken, DNA extraction was performed immediately. In contrast, for patients who consented to and participated in the TOP-GEAR study after biopsies were taken, excess specimens harvested during a previous biopsy were used for DNA extraction. Genomic analysis {#s0006} ---------------- An NGS apparatus capable of detecting 114 cancer-related gene mutations/amplifications, 12 fusion genes, and 1 gene deletion/polymorphism in a single assay was used for NGS analysis (NCC-Oncopanel ver. 4.0) NCC-Oncopanel ver. 4.0 was developed at the authors\' institution and is dedicated to the specific research of TOPGEAR described in subsection 'Patients'. ([@CIT0012]). Sequencing libraries were prepared using SureSelect XT reagent (Agilent Technologies, Santa Clara, CA, USA) and a KAPA Hyper Prep kit (KAPA Biosystems, Boston, MA, USA) and were analyzed on a MiSeq sequencer (Illumina, San Diego, CA, USA). Bioinformatics analysis was performed, and final decisions for the report were made in conferences by the multidisciplinary team ([@CIT0012]). Outcomes ascertainment {#s0007} ---------------------- The primary outcome was the rate of successful genomic analysis with specimens obtained by percutaneous needle biopsy. The secondary outcomes were profiling of genetic alterations, technical success rate of biopsy procedures, adverse events evaluated using the Common Terminology Criteria for Adverse Events v. 4.0, rate of success in pathological diagnosis, and cause of failed genomic analysis. Technical success of the biopsy procedure was defined as obtaining tissue sections with imaging confirmation of the biopsy needle within the target. Successful NGS analysis was defined as the ability to perform genomic analysis by NGS using DNA extracted from the specimen. The causes of failed NGS analysis were categorized as: (i) failure of the puncture of the target site (sampling error); (ii) unprocessed for DNA extraction due to insufficient specimen volume; (iii) insufficient DNA volume; and (iv) deteriorated DNA quality. We also calculated the rate of successful genomic analysis excluding NGS analysis that failed due to reasons unrelated to the biopsy procedures, i.e. reasons (ii) and (iv). Statistical analysis {#s0008} -------------------- Categorical variables were estimated as percentages and median values. In order to study whether risk factors existed for the inability to perform genomic analysis, patients were divided into two groups (patients in whom genomic analysis was possible, and patients in whom analysis was not possible), and univariate analysis was performed. Patient factors (age, sex, and primary tumor type), tumor factors (site and size), and biopsy procedure factors (needle type, number of specimens collected, and time until DNA extraction) were included as variables in the analysis. Categorical variables were analyzed using Pearson's chi-square test, continuous variables that showed a normal distribution were analyzed by *t* test, and continuous variables that did not show a normal distribution were analyzed with the Mann--Whitney *U* test. A *P* value less than 0.05 was considered to be statistically significant. Results {#s0009} ======= Patient demographics {#s0010} -------------------- Of the 490 patients enrolled in the TOP-GEAR study during the study period, 48 were included in this study ([Figure 1](#F0001){ref-type="fig"}). Patient demographics and tumor characteristics are given in [Tables 1](#t0001){ref-type="table"} and [2](#t0002){ref-type="table"}. ###### Patient demographics.   *n* \% ---------------------------- ------------- ------ Gender      Male 23 48.0  Female 25 52.0 Age, years, median (range) 54 (23--77)   Primary site      Lung 6 12.5  Breast 6 12.5  Unknown 6 12.5  Colon 3 6.3  Thymus 3 6.3  Bile duct 2 4.2  Pancreas 2 4.2  Others 20 41.7 ###### Characteristics of tumors.   *n* \% -------------------------------- ----- ----------- Tumor location      Liver 20 41.7  Lung 5 10.4  Mediastinum 5 10.4  Pelvis 4 8.3  Peritoneum 3 6.3  Soft part 3 6.3  Retroperitoneum 2 4.2  Pleura 2 4.2  Superficial lymph nodes 2 4.2  Diaphragm 1 2.1  Bone 1 2.1 Tumor size, mm, median (range) 35 (11--180)  \<30 mm 19 39.6  ≥30 mm 29 60.4 Feasibility of genomic analysis {#s0011} ------------------------------- The success rate of genomic analysis using NGS with biopsy specimens was 79.2% (38/48). A total of 52 mutations, 5 amplifications, and 2 homozygous deletions were identified. Twenty-six patients had at least one genetic alteration. The causes of failure of analysis (*n* = 10) were: (i) failure of the puncture of the target site (sampling error) (0/10); (ii) unprocessed for DNA extraction due to insufficient specimen volume (6/10); (iii) insufficient DNA volume (2/10); and (iv) deteriorated DNA quality (2/10). The rate of successful genomic analysis excluding NGS analysis that failed due to reasons unrelated to the biopsy procedures (i.e. reasons \[ii\] and \[iv\]) was 95.2% (40/42). Of the 6 specimens unprocessed for DNA extraction, 3 were due to insufficient volume of the excess specimen, and 3 were due to shortage of the specimen due to requirement of multiple immunohistochemistry tests. The imaging finding of the insufficient DNA volume of 2 patients with liver tumor was hypovascularity in the target region. In one patient, who underwent biopsy after medical treatment for liver metastases from bile duct cancer, there was a marked decrease in the enhancement and the size of the tumor on contrast-enhanced CT ([Figure 3](#F0003){ref-type="fig"}). Ages of the specimens of two patients with deteriorated DNA quality were 1001 and 1611 days, respectively. {#F0003} When comparing the patients with successful and failed genomic analysis, no statistically significant differences were observed in evaluated variables ([Table 3](#t0003){ref-type="table"}). There was a trend correlating failed genomic analysis with tumor size, although it was not statistically significant (*P* = 0.088). ###### Comparison of patient and tumor characteristics between success and failure in gene profiling. Age, years Success (*n* = 38) Failure (*n* = 10) *P* value ----------------------- -------------------- -------------------- -----------  Median 52 61 0.425 Gender        Male 19 5 0.616  Female 19 6 Tumor location     0.199  Liver 17 3 0.684  Non-liver 21 7 Tumor size, mm        Median 36 23 0.088  \<30 mm 14 5 0.270  ≥30 mm 24 5 Type of needle        Automatic 23 6 0.804  Semi-automatic 15 4 Number of cores        Median 3 3 0.171 Age of specimen, days        Median 16 249 0.357 Technical results of biopsies {#s0012} ----------------------------- The characteristics of the target lesions are shown in [Table 3](#t0003){ref-type="table"}. The liver was the most common site (41.7%). The median diameter (largest length in the axial sections) of the target lesion was 35 mm, and 19 lesions (39.6%) were smaller than 30 mm. For the guiding image, US was used in 26 patients (54.2%) and angio-CT in 22 patients (45.8%). All biopsies were performed with 18-G needles with a throw length of 2 cm. Automatic biopsy needles (MUGNUM, Bard Biopsy Systems, Tempe, AZ, USA; Pro-Mag Ultra, ARGON Medical Devices, Plano, TX, USA) were used in 29 patients (60.4%) and semiautomatic biopsy needles (Temno Evolution, BD, Franklin Lakes, NY, USA; Bard Mission, Bard Biopsy Systems, Tempe, AZ, USA) in 19 patients (39.6%). The co-axial technique was used in 25 biopsies (52.1%). The median number of cores was 3 (range, 1--5). Biopsies were technically successful in all patients (100%), and the median procedure time was 20 min (range, 10--50 min). Adverse events associated with the procedure included pneumothorax (grade 1) in 1 patient and bleeding (grade 1) in 4. Two of the bleeding patients underwent biopsy of the liver and were controlled with a needle-tract embolization. There were no severe adverse events or biopsy-related deaths. Pathological diagnostic yield {#s0013} ----------------------------- A histopathological diagnosis was established in 47 patients (97.9%). The diagnoses were adenocarcinoma, 21 patients; epithelioid hemangioendothelioma, 4; thymic cancer, 4; malignant mesothelioma, 2; and other, 17. The tumor cell percentage was measured in 27 patients, and the median was 60% (range, 10%--100%). Median DNA yield was 0.67 μg. The specimens were fresh (within 7 days from biopsy) in 21 patients (43.8%) and archived (8 days or more after biopsy) in 27 patients (56.2%). The median interval from biopsy to DNA analysis was 169 days (range, 10--2068 days). Discussion {#s0014} ========== Previous studies on genomic profiling of biopsy specimens have provided widely varying results ([@CIT0007],[@CIT0013]). Lung cancer was the most commonly reported neoplasm, and *EGFR*, *KRAS*, and *ALK* analysis using polymerase chain reaction or fluorescence *in situ* hybridization was achieved in 67%--100% of the specimens ([@CIT0007],[@CIT0013]). Few reports exist on the use of NGS in percutaneous biopsy specimens. In a retrospective study, Young et al. performed NGS analysis on formalin-fixed, paraffin-embedded (FFPE) specimens from fine-needle aspiration in patients with lung (*n* = 16) or pancreatic (*n* = 23) tumors, and genomic analysis was successful in all specimens (100%) ([@CIT0010]). Zheng et al. performed NGS analysis on 1152 FFPE specimens from surgical, FNA, and percutaneous biopsy samples, with success rates of 99.3%, 96.9%, and 94.4%, respectively. ([@CIT0011]). These prior retrospective studies of NGS reported higher success rates than the present study and identical to the rate without biopsy-unrelated reasons. However, details of the biopsy procedure were not described in these studies. For genomic profiling, which requires the extraction and analysis of nucleic acids, the most important principle is to select a viable, cell-rich area on the imaging. Moreover, sites that appear enlarged in comparison to past images are considered to have higher viability ([@CIT0014]). Intra- and inter-tumoral (with multiple organ metastases) genomic heterogeneity as well as temporary changes may affect the results of genomic analysis and the determination of a treatment plan ([@CIT0016],[@CIT0017]). However, there is currently no diagnostic imaging method that indicates whether a needle biopsy specimen has been collected from a site having genetic mutations representative of the patient's status. In the future, as radiogenomics comparing imaging data and the genomic profile of a tumor ([@CIT0018]) develops, site selection may become more systematic and sophisticated. The required specimen quantity for NGS analysis has previously been determined as a tumor cell percentage of ≥10% ([@CIT0021],[@CIT0022]). If the tumor cell percentage is low, copy number validation detection is difficult, and the effects of artifacts increase ([@CIT0022]). In our study, the median tumor cell percentage measured in 27 patients was 60%. This suggests that specimens with a high tumor cell percentage can be obtained by percutaneous needle biopsy planned and navigated by images. In a study of 1564 patients analyzed using NGS, Cho et al. reported that analysis was possible in 95.9% (1503 patients) ([@CIT0021]), and the recommended parameters for NGS analysis were \>1 mm in size and \>1 unstained slide in the case of FFPE specimens. Regarding differences in the quantity of collected nucleic acid caused by needle diameter, Jamshidi et al. reported that the diameter of the needle contributes more to the quantity of collected nucleic acid than does the number of cores sampled; use of an 18-G needle yielded a 4.8--5.7-fold greater quantity of nucleic acid than use of a 20-G needle ([@CIT0023]). In our study, use of an 18-G needle with a median of 3 cores yielded satisfactory results of genomic analysis and pathological diagnosis. Thus, if the target site is safely accessed, an 18-G needle would be suitable for acquiring samples for genomic analysis using NGS. There are some limitations of this study. First, it was a retrospective study. Second, this study involved a small number of patients at a single facility. Third, all biopsies were obtained with an 18-G needle, so we did not compare nucleic acid yields with needle size. Fourth, surgical specimens were not examined. However, given the standardized technique of biopsy, our results at least have valuable information for the size and the number of cores of specimen. In conclusion, taking image-guided needle biopsy specimens using an 18-G cutting needle yielded a success rate of 79.2% on genomic analysis using NGS; the rate excluding the NGS analysis that failed due to reasons unrelated to the biopsy procedures was 95.2%. Since there were no severe adverse events, we propose that image-guided needle biopsy might become an adoptable method for tissue sampling for NGS. By developing imaging-based selection methods of the target biopsy site and image-guided technology for effectively sampling the target site, the percentage of specimens that can undergo genomic analysis is expected to increase, but evaluations based on prospective studies are warranted. Disclosure statement {#s0016} ==================== The authors declare that they have no conflict of interest. Funding {#s0015} ======= This work was supported by the Japan Agency for Medical Research and Development (AMED) under the Practical Research for Innovative Cancer Control Grant (16ck0106058h0003); the Ministry of Health, Labour and Welfare of Japan under Health and Labor Sciences Research Grant (H26-055); and National Cancer Center under the National Cancer Center Research and Development Fund (29-A-11). ***Miyuki Sone*** is consultant radiologist and the head of Interventional Radiology Center at the National Cancer Center Hospital (NCCH) in Tokyo, Japan. ***Yasuaki Arai*** is the executive advisor to the president at the National Cancer Center (NCC) in Tokyo, Japan, and a staff radiologist at NCCH. ***Shunsuke Sugawara*** is a consultant radiologist at NCCH. ***Takatoshi Kubo*** was a fellow at NCCH and currently is a staff radiologist at the University of Tokyo. ***Chihiro Itou***, is staff radiologists at NCCH. ***Tetsuya Hasegawa*** is staff radiologist at NCCH. ***Noriyuki Umakoshi*** is staff radiologist at NCCH. ***Noboru  Yamamoto*** is the head of Department of Experimental Therapeutics, NCCH and group leader at Exploratory Oncology Research & Clinical Trial Center, NCC. ***Kumiko Sunami*** is a staff physician of Department of Pathology and Clinical Laboratories at NCCH. ***Nobuyoshi Hiraoka*** is the chief of Department of Pathology and Clinical Laboratories at NCCH. ***Takashi Kubo*** is a staff member of the Division of Translational Genomics, Exploratory Oncology Research & Clinical Trial Center, NCC.

1. Introduction {#sec0005} =============== Adolescence is a period characterized by heightened risk-taking behaviors with dire health consequences related to morbidity and mortality, such as substance use, unprotected-sex, and reckless driving ([@bib0085]). Notable increases in risk-taking behaviors are seen between childhood and adolescence, with adolescents engaging in riskier activities than younger or older individuals ([@bib0145]; [@bib0110]; [@bib0280]). These risk-taking behaviors not only have long-lasting effects on individuals' health, social, and career development, but also constitute a public health issue that threatens the overall well-being of young people ([@bib0090]; [@bib0280]). Thus, it is critical to identify the factors and mechanisms that underlie heightened risk-taking behaviors in adolescence. Prior research has emphasized the importance of the motivational system in risky decision-making processes ([@bib0115]; [@bib0175]; [@bib0150]; [@bib0310]; [@bib0330]), yet the mechanisms through which individual differences in motivation may influence adolescents' risk-taking behaviors are not well documented. Accumulating evidence suggests that neural representations of risk (i.e., the variance of potential outcomes) play an important role in risky decision-making behaviors ([@bib0200]). Compared to children and adults, adolescents show greater inter-individual variability in behavioral risk preference as well as a peak in risk-related neural activation during adolescence ([@bib0320]). In the current longitudinal study, we investigate whether adolescents' behavioral risk preference and neural risk processing under uncertainty may partially explain the association between approach and avoidance motivations and laboratory-based risk-taking behaviors in adolescents. We note that the definition of risk in risk preference and risk processing in the current study is consistent with the behavioral economics views on risk, namely variance of potential outcomes. As reviewed by [@bib0255], this definition of risk is different from the clinical definition of risk (i.e., potential for negative outcomes) that is implied by risky behaviors (i.e., behaviors that harm oneself or others). A number of personality theories have suggested the existence of two main motivational systems, approach and avoidance, that influence decision-making processes and goal-directed behaviors ([@bib0035]; [@bib0050]; [@bib0095]; [@bib0130]). These theories posit that approach motivation is a general neurobiological sensitivity to positive/desirable stimuli (i.e., reward) that is accompanied by a behavioral predisposition toward such stimuli, whereas avoidance motivation represents a general neurobiological sensitivity to negative/undesirable stimuli (i.e., punishment) accompanied by a behavioral predisposition away from such stimuli ([@bib0095]). Adolescence is an important developmental period to understand relative contributions of approach and avoidance motivations related to risk-taking behaviors, because it is a critical period of brain development related to risky decision making, and also has observed distinctive developmental patterns between approach and avoidance. Notably, greater approach and less avoidance have been repeatedly found among adolescents compared to adults ([@bib0105]; [@bib0125]; [@bib0305]). For instance, [@bib0060] found that approach motivation toward potential reward displayed an inverted U-shape relation with age, such that the maximal sensitivity to positive feedback occurred during mid-to late adolescence. In contrast, tendencies to avoid negative outcomes strengthen with age in a linear function, not showing full maturity until the adulthood. The different developmental patterns of approach and avoidance motivations lend support to the theoretical perspective that heightened risk-taking behaviors observed in adolescence may be derived not only from increased motivation to rewards and new experiences, but also from the insensitivity to avoid undesirable punishment ([@bib0275]). Empirical studies reveal consistent evidence for heightened approach (reward-related) motivation among adolescents compared to adults ([@bib0120]). However, evidence for reduced avoidance (punishment-related) motivation is less consistent, indicating that adolescents show comparable or even higher levels of risk aversion than adults ([@bib0215]; [@bib0300]). A growing body of evidence points to approach motivation as a key risk factor for risk-taking behaviors, with higher approach motivation being linked to an earlier onset of substance use, higher levels of substance use, and increased risky sexual behaviors ([@bib0150]; [@bib0310]; [@bib0330]). In contrast, findings on the relation between avoidance motivation and risk-taking behaviors are less consistent. While some studies have reported no link between avoidance motivation and substance use ([@bib0075]; [@bib0150]), one study revealed that avoidance motivation was negatively associated substance use in college students ([@bib0115]), and another suggested that low avoidance was linked to a progression into regular substance use in adolescents ([@bib0330]). Such mixed findings call for further investigation to evaluate the joint effects of approach and avoidance motivations on other types of risk-taking behaviors and explore the underlying mechanisms of the above associations. In behavioral economics, expected utility models provide a specific conceptual framework to understand the process of decision making. These models allow for the decomposition of decision-making components, which allows for increased precision in understanding which elements (e.g., expected value, risk) contribute to risky decision making ([@bib0255]). Extant neuroeconomics literature has implicated anterior insular cortex as a key region in both the processing and learning of risk information ([@bib0200]; [@bib0220]; [@bib0225]; [@bib0270]; [@bib0320], [@bib0325]). Specifically, the insular cortex has been found to respond to increasing levels of risk when evaluating decision options in uncertain environments ([@bib0025]; [@bib0200]; [@bib0340]). Greater activation in the anterior insular cortex is associated with greater risk avoidance ([@bib0220]) and subsequent switching from a risky to a safe option ([@bib0170]). Moreover, patients with insular cortex lesions are insensitive to the risk associated with decision options, indicating that the insular cortex may play a key role in signaling the likelihood of aversive outcomes ([@bib0070]). Developmental neuroscience work has emphasized neurobiological pathways underlying adolescent risk-taking behaviors, suggesting that individual differences in the subcortical, motivational systems involved in the processing of reward are linked to risk-taking development ([@bib0055]; [@bib0180]; [@bib0285]). Research on adolescent reward processing showed that heightened activation in the regions such as the ventral and dorsal striatum is linked to greater risky decision making ([@bib0315]). However, there is emerging evidence suggesting that risk processing (i.e., heightened sensitivity to risky options shown by insular cortex activation) is related to decreased risk taking in adolescents in both laboratory-based and real-world behaviors. A study by [@bib0320] revealed that adolescents displayed greater insula activation in response to increasing risk compared to children and adults. That is, adolescents may have greater emotional responses to risks, as increased engagement of the insular cortex may represent a signal leading decision makers to exhibit caution and thoroughly evaluate risky options. Behaviorally, adolescents displayed substantial individual differences in risk sensitivity, however, ranging from risk seeking to risk averse, whereas adults were uniformly risk averse. Further, [@bib0155] demonstrated that low levels of risk-related activation in the insular cortex during anticipation of uncertain outcomes predicted high levels of health risk behaviors among adolescents, particularly for those who exhibited neural patterns indicating poor cognitive control. To date, however, it remains unclear as to what contributes to substantial individual differences in risk sensitivity and risk-related neural processing observed among adolescents. Given the critical role of the insular cortex in the integration of cognitive, emotional, and motivational information, it is likely that risk processing may be influenced by internal states such as emotions and motivations ([@bib0080]; [@bib0270]; [@bib0320]). It follows that individual differences in approach and avoidance motivations may be linked to individual differences in insular risk processing during decision making, which in turn are linked to individual differences in risk-taking behaviors. In the current study, we examined the associations among approach and avoidance motivations, insular risk processing, and laboratory-based risk-taking behaviors in adolescents using longitudinal data repeatedly measured annually over three years. We used a longitudinal design for testing a developmental cascade model, focusing on individual differences in neural risk processing, being influenced by trait motivations and predicting changes in risk-taking behaviors during adolescence. Specifically, our hypothesized *longitudinal progression* model tested whether approach and avoidance motivations measured at Time 1 were related to risk-taking behaviors measured at Time 3 indirectly through parallel mediators of behavioral risk preference and neural risk processing measured at Time 2. Furthermore, our hypothesized *longitudinal change* model tested whether approach and avoidance motivations measured at Time 1 were statistically predictive of behavioral risk preference and neural risk processing measured at Time 2 (controlling for baseline at Time 1), which in turn were related to risk-taking behaviors measured at Time 3 (controlling for baseline at Time 1). 2. Methods {#sec0010} ========== 2.1. Participants {#sec0015} ----------------- The sample included 167 adolescents (47% females, 53% males), 13 to 14 years of age at Time 1 (*M* = 14.13, *SD* = 0.54), 14 to 15 years of age at Time 2 (*M* = 15.05, *SD* = 0.54) and 15 to 16 years old at Time 3 (*M* = 16.07, *SD* = 0.56). The current sample was representative of rural southwest Virginia for household income and race/ethnicity. At all three times, median household income in the sample was \$35,000 - \$49,999, which is close to the median annual household income range (\$36,000 - \$59,000) of the area. Adolescent participants are primarily Caucasian (82%), African-American (12%), and other (6%). At Time 1, a total of 157 adolescents were recruited to participate in a longitudinal study. At Time 2, 10 additional adolescents were recruited, leading to a final sample of 167 dyads. Between Time 1 and Time 3, 19 adolescents did not complete the study for reasons such as: ineligibility for neuroimaging tasks (*n* = 1), moved away (*n* = 1), extenuating circumstances (*n* = 1), lost contact (*n* = 8), and declined participation (*n* = 8). Attrition analyses indicated that the 19 adolescents who did not return for Time 2 or Time 3 were not significantly different on demographic (age, income, race, sex) or main study variables (approach, avoidance, risk processing, and risk-taking behaviors at Time 1) from the adolescents who did return (all *ps* \> .09). Exclusion criteria included claustrophobia, history of head injury resulting in loss of consciousness for more than 10 min, orthodontia impairing image acquisition, and contraindications to magnetic resonance imaging. 2.2. Procedure {#sec0020} -------------- Adolescents and their families were recruited via flyers and emails that were distributed in schools and other community locations. Research assistants described the nature of the study to interested individuals over the telephone and invited them to participate. Data collection took place at the university's offices, where adolescents and their primary caregivers were interviewed by trained research assistants and received monetary compensation for participation. All participants provided written consent for a protocol approved by the institutional review board of the university. 2.3. Measures {#sec0025} ------------- ### 2.3.1. Approach and avoidance {#sec0030} At Time 1, trait approach was measured using the Sensation Seeking Scale ([@bib0345]) and the Behavioral Activation Scale ([@bib0050]), which included three subscales of drive, fun-seeking, and reward-responsiveness. A principal component analysis was conducted on the three subscales of the Behavioral Activation Scale and the Sensation Seeking Scale, and the result indicated that the first component explained a large portion of the total variance (56.46%), with all factor loadings above .65. These scores were standardized, averaged and standardized again to generate the approach scores. Trait avoidance was measured using the Behavioral Inhibition Scale ([@bib0050]) and Shyness and Fear subscales from the Early Adolescents Temperament Questionnaire--Revised Short Form (EATQ--R; [@bib0045]). A principal component analysis was conducted on these three scales and the result indicated that the first component explained a large portion of the total variance (64.48%), with all factor loadings above .74. These three scores were standardized and averaged and standardized again to generate the avoidance scores. Both approach and avoidance variables are highly stable across three assessments. The zero-order correlations of approach variables across three waves were .65--.77 (*ps \<* .001). The correlations of avoidance variables across three waves were .69--.84, (*ps \<* .001). ### 2.3.2. Risk-taking behaviors {#sec0035} At Time 1 and Time 3, the Stoplight task ([@bib0065]; [@bib0290]) was used to measure laboratory-based risk-taking behaviors. Stoplight is a computerized first-person driving task in which participants control the progression of a vehicle along a straight track. The goal is to advance through a series of intersections to reach a finish line as quickly as possible and receive a monetary reward. At each intersection, as the vehicle approaches a changing traffic signal cycling from green to yellow to red, participants must make a decision about whether to brake and lose time by waiting for the light to return to green or run through the light and chance a crash. Successfully crossing an intersection without braking saves time, whereas braking and waiting for the signal to turn green again results in a time delay. However, if participants do not brake and a crash ensues, the loss of time is even greater than if they were to brake and wait for the light. Adolescent participants completed one round involving 32 intersections which were treated as separate trials. The degree of risk taking was indicated by the number of intersections the participant went through without braking divided by the total number of intersections traversed. Prior research indicated that this laboratory-based measure of risk-taking behaviors is significantly related to real-world health risk behaviors among adolescents ([@bib0160]). ### 2.3.3. Imaging acquisition and analysis {#sec0040} Functional neuroimaging data were acquired on a 3 T Siemens Tim Trio MRI scanner with a standard 12-channel head matrix coil. Structural images were acquired using a high-resolution magnetization prepared rapid acquisition gradient echo sequence with the following parameters: TR = 1200 ms, TE = 2.66 ms, FoV = 245 x 245 mm, and 192 slices with the spatial resolution of 1 × 1 x 1 mm. Echo-planar images (EPIs) were collected using the following parameters: slice thickness = 4 mm, 34 axial slices, field of view (FoV) = 220 x 220 mm, repetition time (TR) = 2 s, echo time (TE) = 30 ms, flip angel = 90 degrees, voxel size = 3.4 × 3.4 x 4 mm, 64 × 64 grid, and slices were hyperangulated at 30 degrees from anterior-posterior commissure. Imaging data were preprocessed and analyzed using Statistical Parametric Mapping (SPM8: Wellcome Trust Neuroimaging Center, London). For each scan, data were corrected for head motion using a six-parameter rigid body transformation and realigned. Functional volumes were normalized using parameters from a segmented anatomical image coregistered to the average EPI and smoothed using a 6 mm full-width-half-maximum Gaussian filter. Reasons for excluding scans included not meeting MRI safety criteria (*n* = 3--6), excessive head motion (\>3 mm; *n* = 3--10), and technical errors (*n* = 3--5). ### 2.3.4. Economic lottery choice task {#sec0045} At Time 1 and Time 2, adolescents participated in a modified economic lottery choice task ([@bib0140]), during which they chose between pairs of uncertain gambles while blood-oxygen-level-dependent (BOLD) response was monitored using functional magnetic resonance imaging (fMRI; [Fig. 1](#fig0005){ref-type="fig"}). Each gamble consisted of a high and low monetary outcome associated with different probabilities. To facilitate comprehension of likelihood information for adolescents, colorful pie charts were used to present probabilities for different outcomes. Each pie was separated into 10 equal slices, with each slice representing 10%. Monetary outcomes and probabilities varied across 72 trials, and it took approximately 30 min to complete all the trials. Risk for each gamble was calculated using coefficient of variation (CV), a scale-free metric calculated by dividing the standard deviation by expected value. Prior research has shown that CV is a better metric for explaining choice behavior compared to standard deviation or variance, because outcome variability is often encoded relative to the average outcome rather than in an absolute manner ([@bib0015]; [@bib0335]). For each pair of gambles, one option was always riskier (higher CV) than the other (lower CV). To incentivize performance, participants were informed that they would receive compensations based on their actual winnings from four randomly selected trials.Fig. 1A) In the economic lottery choice task, adolescents chose between pairs of uncertain gambles. For each trial, there was a high and low monetary outcome, each associated with a specific probability. The monetary outcomes and probabilities were presented using different colors. B) Each trial included a decision phase, a jittered fixation phase, an outcome phase (where the result of the participant's choice was shown), and a jittered intertrial interval (ITI). C) During the decision process, adolescents exhibited greater BOLD responses in the bilateral anterior insular cortex to higher, relative to lower, levels of risk at both Time 1, *t*(145) = 7.22, *p* (FWE correction) \< .05), and Time 2, *t*(135) = 7.91, *p* (FWE correction) \< .05. FWE = family-wise error. Adapted from "Neural cognitive control moderates the association between insular risk processing and risk-taking behaviors via perceived stress in adolescents," by [@bib0185], *Developmental Cognitive Neuroscience, 30*. p. 153. Copyright 2018 by the Elsevier Ltd.Fig. 1 ### 2.3.5. Behavioral risk preference {#sec0050} Risk preference was estimated from each participant's 72 decisions in the economic lottery choice task using a standard power utility function ([@bib0010]; [@bib0140]; [@bib0230]), in which the utility for money *X*, where *X* ≥ 0, is described as,where α represents risk preference, such that α = 1 indicates risk neutrality, α \< 1 indicates risk aversion, and α \> 1 indicates risk seeking. Expected utilities (EU) for each option were then computed by multiplying utilities by associated probabilities, where *P~high~* and *P~low~* represent the probabilities of the high and low outcome, *X~high~* and *X~low~* represent the monetary values of the high and low outcomes within each gamble, respectively.$$EU = P_{high}\text{*}X_{high}^{\alpha} + \, P_{low}\text{*}\, X_{low}^{\alpha}$$ Using maximum likelihood estimation, behavioral choices from the modified lottery choice task for each adolescent were fit to a logistic function,$$P\left( {chosen} \right) = \,\frac{1}{1 + e^{\gamma({EU}_{riskier} - {EU}_{safer})}}$$Where γ ≥ 0 represents the inverse temperature, a metric of relative consistency across choice behavior, in which greater values indicated greater consistency across decisions. ### 2.3.6. Neural risk processing {#sec0055} Using general linear model (GLM), the decision and outcome events of the task were analyzed with a duration of 4 and 2 s, respectively at the subject level. The model included a parametric regressor of the decision event representing the CV for chosen gambles. An additional parametric regressor of the outcome event was also included in the model to represent whether, during the outcome phase, subjects received high or low monetary outcomes. At the group level of the GLM, whole brain analysis was conducted to examine how CV for chosen gambles was related to BOLD responses during decision making. Based on prior literature suggesting the critical role of insular cortex in risk processing ([@bib0200]), we hypothesized that BOLD responses in the bilateral insular cortex would be modulated by the level of CV. Through region of interest (ROI) analyses in SPM8, eigenvariate values were extracted for the left and right insular cortex using a 6 mm sphere around the peak voxel coordinates for each region (left: *x* = -30, *y* = 17, *z* = -14; right: *x* = 30, *y*  = 20, *z* = -11). Activation in the bilateral insular cortex during the lottery choice task was illustrated in [Fig. 1](#fig0005){ref-type="fig"}. For all regions associated with increasing CV during the decision phase at Time 1 and Time 2, see Appendix A. A confirmatory factor analysis, with standardized left and right anterior insula activation scores loaded on an overall insula factor score, were conducted. Factor loadings were constrained to be equal for model identification purposes. In the three fully saturated models (χ^2^ = 0, *df* = 0), factor loadings were all significant (.86 for Time 1 and .93 for Time 2, *p*s \< .001). The bilateral insula factor scores were used in the analyses as the neural risk processing variable, with higher scores indicating higher BOLD responses in the insula. 2.4. Statistical analyses {#sec0060} ------------------------- Descriptive analyses were performed to examine the normality of distributions and outliers for all study variables. For skewness and kurtosis, the acceptable levels were less than 3 and less than 10, respectively ([@bib0165]). Outliers (*n* = 8) were identified as values more than 3 *SD* from the mean and were winsorized to retain statistical power and attenuate bias resulting from elimination ([@bib0135]). Multivariate GLM analyses exhibited that demographic variables (adolescent age, gender, race, and family income) at Time 1 did not significantly predict behavioral and neural risk processing at Time 2 and risk-taking behaviors at Time 3 (all *p*s *\>.*10), thus, they were not included as covariates in the main analyses. The hypothesized mediation models were examined using Structural Equation Modeling (SEM) in M*plus* 7.4 (Muthén & Muthén, 1998--2012). In the first model, we examined a longitudinal progression model by testing indirect effects of approach and avoidance at Time 1 on risk-taking behaviors at Time 3 via neural risk processing at Time 2 (see [Fig. 2](#fig0010){ref-type="fig"}). Next, we examined a longitudinal change model by testing indirect effects of approach and avoidance at Time 1 on risk-taking behaviors at Time 3 via neural risk processing at Time 2 while controlling for initial levels of the mediator and outcome variables (see [Fig. 3](#fig0015){ref-type="fig"}). Overall model fit indices were determined by χ^2^ value, degrees of freedom, corresponding *p*-value, Root Mean Square Error of Approximation (RMSEA), and Confirmatory Fit Index (CFI). RMSEA values of less than .05 were considered a close fit while values less than .08 were considered a reasonable fit ([@bib0030]), and CFI values of greater than .90 were considered an acceptable fit while values greater than .95 were considered an excellent fit ([@bib0020]). Indirect effects were calculated using the IND command in M*plus.* Bias-corrected bootstrap confidence intervals (CIs) for these indirect effects were calculated using 10,000 bootstrapping samples ([@bib0235]). These CIs take non-normality of the estimates into account and are therefore not necessarily symmetric ([@bib0205]). Given that full information maximum likelihood (FIML) estimates are superior to those obtained with listwise deletion or other ad hoc methods ([@bib0250]), FIML estimation procedure was performed to deal with missing data in our model ([@bib0005]).Fig. 2Standardized parameter estimates for the associations among approach and avoidance motivation at Time 1, neural risk processing at Time 2 and risk-taking behaviors at Time 3.\**p* \<.05, \*\* *p \<* .01, \*\*\* *p* \< .001.Fig. 2Fig. 3Standardized parameter estimates for the associations among approach and avoidance motivation at Time 1, neural risk processing at Time 2 and risk-taking behaviors at Time 3 while controlling for neural risk processing and risk-taking behaviors at Time 1.^+^*p =* .06, \**p* \<.05, \*\* *p \<* .01, \*\*\* *p* \< .001.Fig. 3 3. Results {#sec0065} ========== Descriptive statistics and correlations for all study variables are presented in [Table 1](#tbl0005){ref-type="table"}. Neural risk processing and risk-taking behaviors all showed moderate stability from Time 1 to Time 2 and from Time 1 to Time 3, respectively. Meanwhile, adolescents showed more insula activation \[*t*(116) = -6.52, *p* \< .001 for left insula and *t*(116) = -6.93, *p* \< .001 for right insula\] and slightly lower risk-taking behaviors \[*t*(130) = 1.75, *p* = .08\] at Time 3 compared to Time 1.Table 1Descriptive statistics of and correlations among study variables.Table 11234567*MSD*1. Approach at Time 1.00.752. Avoidance at Time 1−.05.00.803. Beh. risk preference at Time 1−.02.15.76.564. Beh. risk preference at Time 2.08−.05.43[\*\*\*](#tblfn0015){ref-type="table-fn"}.56.535. Insular risk processing at Time 1−.03.00−.40[\*\*\*](#tblfn0015){ref-type="table-fn"}−.13.00.936. Insular risk processing at Time 2−.30[\*\*](#tblfn0010){ref-type="table-fn"}−.02−.20[\*](#tblfn0005){ref-type="table-fn"}−.33[\*\*](#tblfn0010){ref-type="table-fn"}.34[\*\*\*](#tblfn0015){ref-type="table-fn"}.00.977. Risk-taking behaviors at Time 1.15−.09.22[\*\*](#tblfn0010){ref-type="table-fn"}.26[\*\*](#tblfn0010){ref-type="table-fn"}−.16−.11.32.158. Risk-taking behaviors at Time 3.18[\*](#tblfn0005){ref-type="table-fn"}−.13.17.28[\*\*](#tblfn0010){ref-type="table-fn"}−.19[\*](#tblfn0005){ref-type="table-fn"}−.20[\*](#tblfn0005){ref-type="table-fn"}.40[\*\*\*](#tblfn0015){ref-type="table-fn"}.29.17[^1][^2][^3][^4] We first fit the *longitudinal progression model* to examine whether approach and avoidance motivations at Time 1 predicted risk-taking behaviors at Time 3 via neural risk processing at Time 2 (see [Fig. 2](#fig0010){ref-type="fig"}). This model estimated all possible paths among study variables, thus was a fully saturated model with χ^2^ = 0, *df =* 0, CFI = 1.00, RMSEA = .00. High approach motivation at Time 1 predicted low neural risk processing at Time 2 (*b* = -0.39, *SE* = 0.11, *p* \< .001), which in turn predicted high risk-taking behaviors at Time 3 (*b* = -0.03, *SE* = 0.01, *p* = .037). Approach at Time 1 did not directly predict risk-taking behaviors at Time 3 (*b* = 0.03, *SE* = 0.02, *p* = .19) and approach and avoidance motivations at Time 1 did not covary (σ = -.03, *SE* = 0.05, *p* = .58). The bias corrected bootstrap test for mediation revealed that the indirect effect from high approach at Time 1 to high risk-taking behaviors at Time 3 via low neural risk processing at Time 2 was significant (*b* = 0.013, SE = 0.01, 95% CI \[0.002; 0.032\], *b*\* = .055). In contrast, avoidance motivation at Time 1 was not associated with neural risk processing at Time 2 (*b* = -0.02, *SE* = 0.10, *p* = .81) or risk-taking behaviors at Time 3 (*b* = -0.03, *SE* = 0.02, *p* = .13). The indirect effect from avoidance at Time 1 to risk-taking behaviors at Time 3 via neural risk processing at Time 2 was not significant (*b* = 0.001, SE = 0.003, 95% CI \[-0.007; 0.10\], *b*\* = .004). Next, we fit the *longitudinal change model* by controlling the levels of neural risk processing and risk-taking behaviors at Time 1. This model allowed us to examine whether approach and avoidance motivations at Time 1 predicted changes in risk-taking behaviors from Time 1 to Time 3 via changes in neural risk processing from Time 1 to Time 2 (see [Fig. 3](#fig0015){ref-type="fig"}). This model fit the data well, with χ^2^ = 6.11, *df =* 6, *p* = .41, CFI = .99, RMSEA = .01. High approach motivation at Time 1 significantly predicted decreases in neural risk processing (*b* = -0.45, *SE* = 0.11, *p* \< .001), which in turn predicted increases in risk-taking behaviors (*b* = -0.04, *SE* = 0.01, *p* = .019). Though approach at Time 1 was not directly associated with changes in risk-taking behaviors (*b* = 0.01, *SE* = 0.02, *p* = .638), the indirect effect from high approach to increases in risk-taking behaviors via decreases in neural risk processing was significant (*b* = 0.016, *SE* = 0.008, 95% CI \[0.003; 0.037\], *b*\* = .066). In turn, high avoidance motivation at Time 1 was marginally associated with decreases in risk-taking behaviors (*b* = -0.15, *SE* = 0.08, *p* = .065), however, the indirect effect of avoidance on changes in risk-taking behaviors via changes in neural risk processing was not significant (*b* = 0.003, *SE* = 0.004, 95% CI \[-0.006; 0.014\], *b*\* = .012). We also tested the *longitudinal progression* model using behavioral risk preference to examine whether the effects of approach and avoidance motivations at Time 1 on risk-taking behaviors at Time 3 were mediated by behavioral risk preference at Time 2. Results indicated no significant indirect effects of approach or avoidance motivations at Time 1 on risk-taking behaviors at Time 3 via behavioral risk preference at Time 2 (*b* = 0.005, *SE* = 0.005, 95% CI \[-0.006; 0.022\], *b*\* = .021 for Approach; *b* = -0.002, *SE* = 0.005, 95% CI \[-0.016; 0.011\], *b*\* = -0.01 for Avoidance). We then ran the *longitudinal change* model using behavioral risk preference to examine whether approach and avoidance motivations predicted changes in risk-taking behaviors via changes in behavioral risk preference(by controlling for the levels of behavioral risk preference and risk-taking behaviors at Time 1). Similarly, the effects of approach and avoidance motivations on changes in risk-taking behaviors were not mediated by changes in behavioral risk preference (*b* = 0.005, *SE* = 0.004, 95% CI \[-0.002; 0.019\], *b*\* = .021 for Approach; and *b* = -0.005, *SE* = 0.004, 95% CI \[-0.021; 0.002\], *b*\* = -.025 for Avoidance). Detailed results are reported in Appendix B. Finally, as supplemental analyses, we further tested the specificity of our results to the bilateral insula by exploring the extent to which additional brain regions may play a mediating role between trait motivations and adolescent risk taking. We selected brain regions that were consistently activated in both Time 1 and Time 2: the dorsal anterior cingulate cortex (dACC), left middle occipital cortex, right middle occipital cortex, right precentral gyrus, and right superior orbitofrontal cortex (see Supplemental Tables in Appendix A). We chose this conservative approach to focus on reliable brain activations across both waves and to limit the number of tests performed. We re-ran the longitudinal progression model and longitudinal change model on each of the five ROIs and found that only the longitudinal analyses on dACC were significant. Specifically, in the longitudinal progression model, the indirect effect from high approach at Time 1 to high risk-taking behaviors at Time 3 via low dACC activation at Time 2 was significant (b = 0.011, SE = 0.006, 95% CI \[0.002; 0.027\], b\* = .05). In the longitudinal change model, the indirect effect from high approach to increases in risk-taking behaviors (from Time 1 to Time 3) via decreases in dACC (from Time 1 to Time 2) was significant (b = 0.013, SE = 0.007, 95% CI \[0.003; 0.030\], b\* = .057). The results are not surprising because dACC is known for its involvement in decision making such as integrating information about the reward and costs to estimate expected value associated allocating control ([@bib0260]), whereas the other regions (middle occipital cortex, precentral gyrus, and superior orbitofrontal cortex) are not. 4. Discussion {#sec0070} ============= Prior research has emphasized the importance of the motivational system in risky decision-making processes ([@bib0105]; [@bib0115]; [@bib0150]; [@bib0310]; [@bib0330]), yet the developmental pathways from approach and avoidance motivation to adolescents' risky behaviors are not clearly understood in the extant literature. Given the crucial role of the insula in evaluating risk during decision-making process ([@bib0200]), and the substantial individual differences in risk preference among adolescents ([@bib0215]; [@bib0320]), we tested the potential meditating effects of both behavioral risk preference and neural risk processing in the association between approach versus avoidance motivations and laboratory-based risk-taking behaviors. Our longitudinal data indicated that high approach motivation was related to lower bilateral insular cortex activation, which in turn was linked to higher risk-taking behaviors measured by the Stoplight task, a computerized measure of risk-taking propensity. This indirect path was also obtained when controlling for baseline levels of neural risk processing and risk-taking behaviors at Time 1. Specifically, high approach motivation was related to decreases in bilateral insular cortex activation which in turn were associated with increases in risk-taking behaviors. In contrast, we found no clear evidence that avoidance motivation was directly or indirectly related to risk-taking behaviors over time. Our results demonstrating the crucial role of the insula in the association between approach-oriented motivation and risky behaviors largely corresponds with prior neuroscience literature implying that insular processing may be influenced by motivations and emotions ([@bib0080]; [@bib0210]), and that lower insula activation during risk processing is associated with greater health risk behaviors among adolescents ([@bib0155]). Anatomically, the insular cortex is often thought of as an integration center, which receives inputs from the sensory cortices, while also projecting to brain areas involved in action-oriented functions, such as the prefrontal cortices and anterior cingulate cortex ([@bib0270]). The insula is well positioned to receive affective and cognitive information and has been implicated as a key neural substrate in linking motivation, affective processing, decision making, and behaviors ([@bib0210]; [@bib0240]). In particular, [@bib0265] proposed an integrative model for the function of the insular cortex, suggesting that the insular cortex integrates information from internal bodily states with individual preferences for risk and contextual information to produce a global feeling state to regulate subsequent behaviors. Following this model, increased engagement of the insula may reflect a thorough evaluation of both internal states and contextual factors. Mismatch between anticipated outcomes and negative consequences associated with risky options may evoke anxiety and fear, which in turn can facilitate behaviors that avoid risks ([@bib0320]). In contrast, decreased engagement of the insula may represent difficulties fully integrating information from different inputs, and decisions may be influenced by individual risk preference. For adolescents who have a strong approach motivation, the failure to engage the insula to process risk by evaluating potential negative outcomes may contribute to heightened risky behaviors, as seen in the present study. The longitudinal effect of avoidance motivation at Time 1 indicated no significant direct or indirect effects on risk-taking behaviors at Time 3 in general. However, the direct association between avoidance motivation at Time 1 and *changes* in risk-taking behaviors from Time 1 to Time 3 approached significance. Specifically, low avoidance motivation was marginally related to increases in risk-taking behaviors. This finding is consistent with prior research which found that low avoidance predicted progression into regular substance use ([@bib0330]). It is likely that adolescents with low avoidance motivation are less sensitive to aversive outcomes associated with risky decision making, making them less prone to engage in risk-taking behaviors over time. Taken together, these findings highlight the importance of utilizing a longitudinal design to fully capture the temporal relations between avoidance motivation and risk-taking behaviors. Our data further clarified that the association between avoidance motivation and changes in risk-taking behaviors is not mediated by variations in neural risk processing. The null results may be partly due to the fact that the economic lottery choice employed in this study included gambles in which potential outcomes could only be gained; the task did not include potential loss trials. It has been shown that adolescents show less harm-avoidance brain responses to reward omission than adults do ([@bib0100]). In our study, the lack of gain (i.e., reward omission) may be perceived as loss by the participants, although it may not elicit negative emotions as intense as loss does. Prior work has indicated that risk within a loss context might be processed by both distinct and overlapping neural substrates when potential losses are possible. Specifically, in addition to the insula, the thalamus and the dorso-medial prefrontal cortex were also likely to be activated to process the negative emotion associated with potential losses and adjust strategies for better outcomes ([@bib0200]). It is plausible that loss-related brain areas may be related to avoidance motivation. Future studies would benefit from utilizing tasks that include both explicit gain and loss conditions to better understand the link between avoidance motivation and risk-related processing. We also tested the mediating role of behavioral risk preference in the link between approach and avoidance motivations and laboratory-based risk-taking behaviors. Despite the moderate phenotypic correlations between behavioral risk preference and neural risk processing, the results showed that it was insula activation, rather than behavioral risk preference, that significantly mediated the effects of approach motivation on adolescents' risk-taking behaviors. Our results suggest that risk-related neural response was more sensitively affected by approach motivation than behavioral risk preference. One interpretation of this finding is that behavioral risk preference and risk-related neural response reflect slightly distinct processes of risky decision making, with behavioral risk preference indicating estimates based on decisions made in the task, but risk-related neural responses indicating the neural responses immediately before a decision is made. An alternative explanation could be that behavioral risk preferences may be limited in capturing real-world behavioral responses, while risk-related neural processing is able to more accurately represent individual differences in neurobiological processes ([@bib0245]). Overall, our findings highlight the critical importance of conducting analyses at both behavioral and neural levels in order to better understand the multi-faceted and complex nature of risk processing during decision making under uncertainty that plays a role in determining risk-taking behaviors. Findings from the current study should be interpreted in light of limitations. First, the economic lottery choice task used in the current study includes only gain conditions, not loss conditions. Although the lack of (anticipated) gain in the task may have the effects of loss, future research is warranted to test possible distinctive effects of explicit gain versus loss conditions. Second, while our primary interest was to examine how insular risk-related processing prior to decision making played a mediating role between trait motivations and adolescent risk taking, future studies should examine the extent to which brain activations in other phases of the decision making process such as the outcome phase may play similar or dissimilar roles. Third, approach motivation did not directly predict laboratory-based risk-taking behaviors, but only indirectly through its influences on neural insula activation. Given the significant predictive effects of approach motivation on substance use and risky sexual behaviors ([@bib0150]; [@bib0310]; [@bib0330]), it could be that approach motivation is more sensitively related to real-world risk-taking behaviors than laboratory-based risk-taking behaviors. Fourth, while the current study focused on the insular cortex, which was grounded in theoretical and empirical work, future research should consider investigating how other brain regions interface with the insula in risk-related decision making. Such future investigations may provide important insight into the neural mechanisms through which normal and atypical risk processing occurs. Relatedly, prior research indicates that the insula and inferior frontal gyrus are key players in cognitive control (e.g., [@bib0190]; [@bib0295]), with the insula being important for detecting behaviorally salient events and the inferior frontal gyrus being important for implementing inhibitory control ([@bib0040]). Lastly, given the association between mathematical cognition and cognitive control systems including insula (see [@bib0195] for review), we suggest that future research should consider mathematical and cognitive abilities that may influence insular risk processing. In conclusion, the current findings provide a window into understanding how trait motivation may influence adolescents' risk taking. Notably, our data provide insight into aspects of brain functioning that mediate trait-level individual differences and risk-taking behavioral outcomes over time throughout adolescence. Adolescents high in approach motivation seem to become less thorough in the neural processing of risk information over time (as reflected by decreased insula activation), which in turn, seems to make them vulnerable to engaging in risky behaviors. Intervention work could benefit from the current work, as we provide new insights into identifying adolescents who are vulnerable to develop risk-taking behaviors. Disclaimer {#sec0075} ========== The views and opinions expressed in this article are those of the authors and do not necessarily reflect the official policy or position of any agency of the U.S. government. Declaration of Competing Interest ================================= The authors declare no conflict of interest. Appendix A. Supplementary data {#sec0085} ============================== The following is Supplementary data to this article: This work was supported by grants from the National Institute on Drug Abuse (R01 DA036017 to Jungmeen Kim-Spoon and Brooks King-Casas and F31 DA042594 to Nina Lauharatanahirun). We thank current and former JK Lifespan Development lab members for their help with data collection. We are grateful to the adolescents and parents who participated in our study. Supplementary material related to this article can be found, in the online version, at doi:<https://doi.org/10.1016/j.dcn.2019.100725>. [^1]: *Note.* Beh. *=* Behavioral. [^2]: p \<.05. [^3]: p \< .01. [^4]: p \< .001.

Introduction {#Sec1} ============ Balanced constitutional reciprocal translocations are the most commonly identified structural chromosomal rearrangements in humans and pigs (frequencies of 0.1 and 0.4 %, respectively---Benet et al. [@CR5]; Ducos et al. [@CR14]). Although carriers are generally phenotypically normal, such rearrangements frequently lead to reproductive disorders. Indeed, the prevalence of reciprocal translocations is about 10 times higher in infertile men and some carriers present spermatogenesis disturbances leading to oligospermia or azoospermia, sometimes associated with teratosermia and/or asthenospermia (Dong et al. [@CR12]; Mau-Holzmann [@CR39]; Van Assche et al. [@CR66]). Reciprocal translocations are also responsible for recurrent miscarriages and congenital defects in the offspring of carrier parents (Benet et al. [@CR5]). In reciprocal translocation heterozygotes, homologous regions of the normal and derivative chromosomes involved in the rearrangement pair during the prophase of the first meiotic division, due to the synaptonemal complex (SC), and form a particular structure called a multivalent (Oliver-Bonet et al. [@CR44]; Villagomez and Pinton [@CR70]). In some cases, chromosomal regions within the multivalent, especially around the breakpoints, remain unsynapsed which may trigger meiosis checkpoints leading to spermatogenesis arrest at the pachytene stage (MacQueen and Hochwagen [@CR34]; Roeder and Bailis [@CR56]). Several hypotheses have been proposed to explain such effects of pairing failure on gametogenesis. The first concerns altered transcription of the genes located on the unpaired segments. Indeed, studies in mice revealed transcriptional repression of unpaired regions by a specific mechanism called "meiotic silencing of unsynapsed chromatin" (MSUC) in individuals with partial or total spermatogenesis arrest (Turner et al. [@CR65]). If some genes essential to the correct course of meiosis are located in these unsynapsed regions of the genome, MSUC may lead to the halting of meiotic division. Secondly, associations between the quadrivalent and the XY bivalent, which is transcriptionally silenced by a phenomenon known as meiotic sex chromosome inactivation (MSCI, Turner [@CR64]), were also observed in individuals with altered semen parameters (azoospermic or oligospermic) (Oliver-Bonet et al. [@CR44]; Sciurano et al. [@CR58], [@CR59]). Such an association could result in partial reactivation of the XY body, leading to the expression of some genes located on the X chromosome (Lifschytz and Lindsley [@CR32]) or spreading of XY body inactivation towards the autosomal segments attached to the XY body, without reactivation of this latter (Jaafar et al. [@CR25]). Both mechanisms induce abnormal genetic expression, possibly responsible for gametogenesis failure. Both hypotheses were confirmed by Homolka et al. ([@CR22]) who simultaneously observed the reactivation of some X chromosome genes and the repression of autosomal genes located on unpaired segments, in a case of autosomal reciprocal translocation in mice. Apart from the above-mentioned potential spermatogenesis failure, reciprocal translocations are systematically responsible for the production of genetically unbalanced gametes (Benet et al. [@CR5]), which can subsequently be responsible for the reported miscarriages (embryonic mortality) or newborn defects. Indeed, chromosomes in the quadrivalent may segregate via different mechanisms (i.e., alternate, adjacent I, adjacent II, 3:1 and 4:0), resulting in the possible production of 18 different types of gametes. Only some types of segregation mechanisms produce genetically balanced gametes which contain either the two normal or the two derivative chromosomes (i.e., alternate segregation, or adjacent I segregation with a crossing-over (CO) in the interstitial regions of the quadrivalent---Benet et al. [@CR5]; Sybenga [@CR62]). The frequencies of the different modes of segregation can vary from one translocation to another and depend on the number and distribution of meiotic recombination sites on the quadrivalents, which, in turn, determine their configurations between prophase I and anaphase (chain or ring) (Faraut et al. [@CR17]). Reciprocal translocations can also lead to disturbances in meiotic recombination. These disturbances may only concern the chromosomes involved in the rearrangement. A decreased number of mutL homolog 1 (MLH1) foci on the quadrivalent, as compared to the corresponding bivalents, has been reported for different translocations (Ferguson et al. [@CR18]; Leng et al. [@CR30]; Oliver-Bonet et al. [@CR44]; Pigozzi et al. [@CR46]). Such disturbances may also be more general and affect the segregation of other chromosomes, leading to aneuploid gametes. On some occasions, the percentage of aneuploidy is inversely correlated to the ejaculated sperm concentration (Douet-Guilbert et al. [@CR13]; Rives et al. [@CR54]; Vegetti et al. [@CR69]). In the cases of autosome-autosome translocations in pigs, the semen parameters of carriers were generally normal, and SC analyses did not reveal any particular chromosome pairing behavior which could explain meiotic cell death (Villagomez and Pinton [@CR70]). In fact, reciprocal chromosome translocations in the pig species tend to undergo heterosynapsis during the early pachytene stages, a germ cell mechanism avoiding apoptosis and subsequent meiotic arrest (Gabriel-Robez et al. [@CR19]; Jaafar et al. [@CR24]; Villagomez et al. [@CR71]). Here, we report for the first time in pigs a case of autosome-autosome translocation associated with oligoasthenoteratospermia (OAT) and characterized by the presence of a tiny derivative chromosome prone to disrupted meiotic pairing. Materials and methods {#Sec2} ===================== Animals {#Sec3} ------- The reciprocal translocation t(1;14) was initially identified during national systematic controls of young boars to be used in artificial insemination centers. The carrier individual (Landrace × Duroc crossbred boar) was phenotypically normal, but presented abnormal semen parameters: 19 × 10^6^ spz/mL (instead of 250 × 10^6^, on average, in normal individuals), with only 5 % of live spermatozoa with good motility and 78 and 66 % with acrosome and distal droplets abnormalities, respectively. Plasmatic levels of luteinizing hormone (LH), progesterone, testosterone, and androstenedione were within normal ranges (0.1 ng/mL, 1 nmol/L, 22 nmol/L, and 9 nmol/L, respectively). Five sows were inseminated with sperm from this boar, but no pregnancy could be obtained. Testicular samples were collected by surgical hemi-castration. Pre-anesthesia (intramuscular injection of ketamine, 10 mg/kg; Virbac, Carros, France) was followed by inhalation anesthesia (isoflurane; Virbac, Carros, France). Post-operative follow-up was carried out in a recovery room adjacent to the operating facility. The animal was monitored until he recovered approximately 15 min after the operation. Castration was minimally invasive. Pain was relieved by an intramuscular injection of Finadyne (2 mg/50 kg; MSD Sante animale, Beaucouze, France) which completed the pre-anesthetic analgesia. The times of first getting up and first meal were monitored, and the healing process was controlled daily for 2 weeks. Cytogenetic and molecular characterization of breakpoints {#Sec4} --------------------------------------------------------- Classical cytogenetic analysis (GTG banding) resulted in the identification of a highly asymmetrical reciprocal translocation involving the chromosomes *Sus Scrofa* chromosome 1 (SSC1) and 14 (SSC14). The chromosomal breakpoints were finely localized by an array painting technique as described by Gribble et al. ([@CR20]). The two derivative chromosomes (1^14^ and 14^1^) (10 copies of each) were isolated from the GTG-banded metaphases by mechanical microdissection and their DNA amplified using the PicoPLEX WGA Kit (New Egland Biolabs, Ipswich, MA, USA). The specificity of this material was controlled on normal metaphases by fluorescence in situ hybridization (FISH) (Pinton et al. [@CR48]). The amplified DNA samples were differentially labeled with dUTP cyanine 3 (1^14^ chromosome) and dUTP cyanine 5 (14^1^ chromosome) and co-hybridized onto a pan-genomic custom chip of 4.2-M probes (Roche NimbleGen, Madison, WI, USA), designed from the porcine reference genome Sscrofa10.2. The microarray was then scanned, and the log~2~ ratios calculated from the intensities were plotted against chromosome position. The translocation breakpoints were defined as the positions where the log~2~ ratios changed from high to low ratios (or vice versa). The data were then confirmed by FISH using bacterial artificial chromosome (BAC) clones selected in the two breakpoint regions. The BACs CH242-298E5 (SSC1) and SBAB-234D5 (SSC14) were labeled with biotin using the BioPrime DNA Labeling System kit (Life Technologies, Carlsbad, CA, USA) and revealed by Alexa 594 conjugated to streptavidin (Molecular Probes, Eugene, OR, USA). Histopathological analysis of the testis sample {#Sec5} ----------------------------------------------- Testis samples were fixed in 10 % buffered formalin for 48 h before routine processing. Four-micrometer-thick paraffin sections were stained with hematoxylin and eosin. Anti-caspase-3a immunohistochemistry was performed as already described (Cheat et al. [@CR9]) with an active caspase-3 antibody at 1:300 dilution (R&D Systems, Minneapolis, MN, USA). Briefly, 4-μm paraffin-embedded sections from a testis were dewaxed in toluene and rehydrated first in an acetone bath then in deionized water. Antigen retrieval was carried out in 10 mM citrate buffer at pH 6.0 for 30 min in a water bath at 95 °C. The cooled sections were then incubated in Dako peroxidase blocking solution (Dako, Glostrup, Denmark) to quench any endogenous peroxidase activity. Non-specific binding was blocked by incubation in normal goat serum at 1:10 dilution (Dako, Glostrup, Denmark) for 20 min at room temperature (RT). The primary antibody was anti-active caspase-3 (dilution 1:300) (R&D Systems, Minneapolis, MN, USA). Sections were incubated with primary antibodies for 50 min at RT. Bound primary antibodies were detected with EnVision™ + Horseradish Peroxidase (HRP) Systems (Dako, Glostrup, Denmark) for 30 min at RT. Peroxidase activity was revealed by 3,3′-diaminobenzidine tetrahydrochloride substrate (Dako, Glostrup, Denmark). Finally, the sections were counterstained with Harris hematoxylin, dehydrated, and coverslipped. Analysis of meiotic pairing {#Sec6} --------------------------- ### Immunocytology {#Sec7} Meiotic cells were prepared as described by Pinton et al. ([@CR50]) with some modifications. Detection of the synaptonemal complex proteins 3 (SCP3) and 1 (SCP1) and centromeres was carried out before immunostaining the γH2AX protein. The meiotic proteins were immunolocalized using antibodies at 1:100 dilution in PBT (1× phosphate-buffered saline (PBS), 0.15 % bovine serum albumin (BSA), and 0.1 % Tween 20) as follows. First, the SCP1 and centromeres were detected using the following primary antibodies: rabbit anti-SCP1 (Abcan, Cambridge, UK) and human anti-centromere (Antibodies Incorporated, Davis, CA, USA). Secondary antibodies consisted of DyLight 488 conjugated goat anti-rabbit (KPL, Gaithersburg, MD, USA) and 1-amino-4-methylcoumarin-3-acetic acid (AMCA) conjugated donkey anti-human (Jackson ImmunoResearch Laboratories, Grove, PA, USA). Secondly, SCP3 was detected using rabbit anti-SCP3 (Abcam, Cambridge, UK) and then revealed with secondary antibody Alexa 594 conjugated donkey anti-rabbit (Molecular Probes, Eugene, OR, USA). Spermatocytes were captured using a Zeiss Imager Z2 microscope with CytoVision imaging system (Leica Microsystemes, Nanterre, France). Finally, the γH2AX protein was detected using mouse anti-γH2AX (Abcam, Cambridge, UK) and Alexa 488 conjugated goat anti-mouse (Molecular Probes, Eugene, OR, USA) antibodies. ### Fluorescence in situ hybridization {#Sec8} After SC analysis, the same cells were subjected to FISH with BAC clones (SBAB-428G6, SBAB-413G8, SBAB-498D8 from the National Institute for Agricultural Research (INRA) BAC library) (Rogel-Gaillard et al. [@CR57]): one was located in the telomeric region of the SSC1 p arm (labeled with biotin), and two were located on SSC14, one in the telomeric region (labeled with digoxigenin) and the other in the centromeric region (containing porcine endogenous retrovirus (PERV) sequences, differentially labeled with biotin and digoxigenin). These BAC clones were labeled using the BioPrime DNA Labeling System kit (Life Technologies, Carlsbad, CA, USA), and revealed by Alexa 594 conjugated to streptavidin (Molecular Probes, Eugene, OR, USA) and fluorescein isothiocyanate (FITC) conjugated mouse anti-digoxygenin antibodies (Sigma-Aldrich, Saint Louis, MO, USA). FISH signals from the same cells for which SCs had previously been analyzed were captured and evaluated using an Imager Z2 microscope with CytoVision imaging system (Leica Microsystemes, Nanterre, France). Recombination analysis {#Sec9} ---------------------- Analysis of meiotic recombination was carried out as described by (Mary et al. [@CR36]). The quadrivalent was identified using the BAC clones previously mentioned in the "[Fluorescence in situ hybridization](#Sec8){ref-type="sec"}" section. The images were analyzed using MicroMeasure 3.3 software (Reeves [@CR52]) to determine the length of the SCs, as well as the positions of the centromeres and recombination sites (MLH1 foci). RNA-DNA FISH and fluorescence immunostaining of the γH2AX protein {#Sec10} ----------------------------------------------------------------- PERV RNA-DNA FISH, coupled with immunolocalization of the γH2AX protein, was performed as described by Barasc et al. ([@CR3]). RNA integrity was preserved by carrying out the PERV RNA-FISH experiment before the immunostaining and PERV DNA-FISH. Microarrays for gene expression analysis {#Sec11} ---------------------------------------- Total RNAs were extracted from testis samples from three fertile control boars with normal semen parameters (Supplementary Table [1](#MOESM1){ref-type="media"}) and from the t(1;14) boar using the method described in Congras et al. ([@CR10]). Three independent extractions from three samples were performed for each animal. RNA quantity and quality were measured by NanoDrop dosage and Bioanalyzer analysis (Agilent Technologies, Santa Clara, CA, USA). Samples were then labeled with cyanine 3 fluorochrome and converted into complementary DNA (cDNA) before hybridization on the 60K customized porcine microarray from Agilent (Congras et al. [@CR11]). Only spots of sufficient intensity and quality for at least two out of three technical replicates were retained. Intensity values were log-transformed, normalized by quantile normalization method, then subjected to differential analysis using the Limma package (Bioconductor). The selected threshold parameters were 0.01 for the maximal adjusted *p* value (Benjamini and Hochberg [@CR6]) and 2 for the minimal log-fold change (lfc). Probes were mapped along the chromosomes by aligning all the microarray probe sequences on the pig reference genome (SGSC Sscrofa10.2/susScr3) using Blat software (Kent [@CR27]) and then plotting the differentially expressed (DE) probes along the chromosome coordinates together with their Log~2~ ratio (microarray GEO accession codes: GSE80693) Analysis of gene expression in the testis by real-time PCR {#Sec12} ---------------------------------------------------------- Total RNAs used for the microarray study were converted to cDNA with Superscript II Reverse Transcriptase (Life Technologies, Carlsbad, CA, USA). Expression analysis of six genes located on the X chromosome and up-regulated in the microarray analysis was carried out by real-time qPCR using the primers listed in Supplementary Table [2](#MOESM2){ref-type="media"} and the Mx3005P Real-Time PCR System from Agilent Technologies. Results were analyzed by the 2ΔΔCt method and statistical *t* test. RPL4 was used as the reference gene for relative quantification (Congras et al. [@CR11]). Sperm FISH analyses {#Sec13} ------------------- Sperm was prepared as reported by Bonnet-Garnier et al. ([@CR7]) ### Meiotic segregation profiles {#Sec14} Sperm FISH analyses were carried out using the above-mentioned BAC clones as described in the [Fluorescence in situ hybridization](#Sec8){ref-type="sec"}" section. The signals from 1582 spermatozoa heads were analyzed. ### Interchromosomal effects {#Sec15} Painting probes were used for chromosomes SSCX and SSCY and for the autosomal controls SSC13 and SSC18. These probes were produced by degenerate oligonucleotide-primed polymerase chain reaction from flow sorted (SSC13, SSC18, and SSCY) or microdissected (SSCX) chromosomes. The SSCX probe was labeled with biotin, the SSCY probe with biotin and digoxigenin, and chromosomes 13 and 18 with digoxigenin. Biotin-labeled probes were revealed in red using Alexa 594 conjugated to streptavidin and amplified with rabbit anti-streptavidin coupled to Texas Red (Abcam, Cambridge, UK) and Alexa 594 conjugated donkey anti-rabbit antibodies. Digoxigenin-labeled probes were revealed in green by FITC conjugated mouse anti-digoxigenin and amplified with Alexa 488 conjugated goat anti-mouse and donkey anti-goat coupled to Alexa 488 (Molecular Probes, Eugene, OR, USA) antibodies. Slides were analyzed under a Zeiss microscope fitted with a triple bandpass filter. More than 10,000 sperm nuclei were scored for each experiment. Only spermatozoa with signals of equivalent intensity and separated by a distance at least the size of one signal were counted. Statistical analysis (R software, Ihaka and Gentleman [@CR23]) {#Sec16} -------------------------------------------------------------- Differences in the relative SC lengths in SSC1 and SSC14 between the translocated boar and controls were examined by applying the Wilcoxon test. The same test was used to compare the number of MLH1 foci (per cell, as well as on chromosomes 1 and 14) between the translocated boar and controls. The MLH1 distributions on SSC1 and SSC14 were compared using the Kolmogorov-Smirnov test. For the sperm FISH results, a classical 2 × 2 chi-square test with Yates continuity correction was used to compare the proportions of each sperm category in the t(1;14) boar and a control boar. *P* values \<0.05 were considered statistically significant. Results {#Sec17} ======= Molecular characterization of the breakpoints {#Sec18} --------------------------------------------- Analysis of the GTG-banded karyotypes revealed a balanced reciprocal translocation between the q arm of one SSC1 chromosome and the q arm of one SSC14 chromosome (Fig. [1a](#Fig1){ref-type="fig"}). The breakpoint on SSC1 was very distal (q2.11 band), whereas the breakpoint on the acrocentric chromosome was located close to the centromere (q1.2 band). The array painting experiment accurately located the breakpoints at positions 293 924 898 (SSC1) and 148 080 (SSC14) (Fig. [1b](#Fig1){ref-type="fig"}) of the pig sequence assembly (Sscrofa10.2). These array painting results were confirmed by selecting BAC clones in the regions overlapping the breakpoints on both chromosomes and hybridizing them on to metaphases of the (1;14) boar (Fig. [1c](#Fig1){ref-type="fig"}). The hybridization signals for the two BAC probes were indeed observed to split on the two derivative chromosomes, thereby confirming the breakpoint positions.Fig. 1Characterization of the chromosomal rearrangement. **a** GTG-banded karyotype of the boar carrying a t(1;14)(q2.11;q1.2) translocation. *Arrows* indicate the locations of the breakpoints on the derivative chromosomes. **b** Array painting of derivative chromosomes 1 and 14 on a custom 4.2-M whole genome array designed from the porcine sequence Sscrofa10.2. Transitions in the hybridization profiles indicate the chromosome breakpoints: positions 293 924 898 (SSC1) (B1) and 148 080 (SSC14) (B2). **c** Confirmation of the array painting results by FISH of BAC clones overlapping the breakpoints: clone CH242-298E5 on SSC1 (BAC position: start 293811882--end 293992669) (*C1*) and BAC SBAB-234D5 on SSC14 (BAC position: start 134867--end 279323) (*C2*). These array painting results were confirmed by splitting of the BAC hybridization signals on the derivative chromosomes. *Scale bars* are all equal to 10 μm Histopathology {#Sec19} -------------- Histopathological evaluation of the testis revealed diffuse moderate seminiferous atrophy associated with diffuse severe interstitial cell hyperplasia. Seminiferous tubules showed altered spermatogenesis with few spermatozoa produced in accordance with the oligospermia detected during sperm analysis (Supplementary Fig. [1](#MOESM3){ref-type="media"}). In addition, a significant increase in the number of apoptotic cells, identified as spermatocytes due to their nuclear size and structure (Barasc et al. [@CR3]; Koykul et al. [@CR29]; Pinton et al. [@CR50]), was demonstrated by anti-caspase-3a immunohistochemistry (Fig. [2](#Fig2){ref-type="fig"}). Apoptotic cells were observed as small multifocal clusters of contiguous positive cells (probably due to the fact that they undergo apoptosis at the same developmental stage of meiotic prophase), often in the middle or luminal portion of the seminiferous epithelium (spermatocyte stage). Binucleated positive cells or both daughter cells positive were also regularly observed, showing that apoptosis probably occurred in cells derived from the same spermatogonia lineage with impaired meiosis (Fig. [2](#Fig2){ref-type="fig"}).Fig. 2Histopathological analysis of testicular tissue. Anti-caspase-3 immunohistochemistry on paraffin sections from t(1;14) (**a**, **b**) and control (**c**) testis samples. Clusters of positive cells (dark brown staining of the nuclei) in moderately atrophic seminiferous tubules. (**a** ×100 magnification. **b**, **c** ×400 magnification) (anti-caspase-3a antibody). *Scale bars* are all equal to 100 μm and 50 μm respectively Meiotic pairing {#Sec20} --------------- SC analysis of 284 pachytene nuclei revealed two pairing configurations: formation of a quadrivalent in 70 % of the cells (Fig. [3a--c](#Fig3){ref-type="fig"}) and a "trivalent plus univalent" configuration in 30 % of the others (Fig. [3d](#Fig3){ref-type="fig"}). The univalents always corresponded to the small 14^1^ chromosome. Fifty-two percent of the 284 cells showed meiotic pairing abnormalities (γH2AX-positive unpaired autosomal regions associated or not with the XY body) (Fig. [3b--d](#Fig3){ref-type="fig"}).Fig. 3Analysis of meiotic pairing in pachytene spermatocytes using FISH and immunostaining of the synaptonenal complex proteins SCP1 (*green*) and SCP3 (*red*), as well as γH2AX (*red*) and the centromeres (*blue*). **a**--**c** Spermatocytes with a quadrivalent configuration. **a** Complete synapsis, no γH2AX signal on the quadrivalent, and no association with the XY body. **b** Unsynapsed segment with γH2AX signal on the derivative 14 chromosome but no association with the XY body. **c** Colocalization of the univalent (derivative 14 chromosome) with the XY body in a γH2AX-positive region. **d** Spermatocyte with a "trivalent plus univalent" configuration. The univalent is characterized by an absence of SCP1 signal (*red* synaptonemal complex instead of *orange*) and is included in a γH2AX-positive region with the XY body. *Scale bars* are all equal to 10 μm Recombination analysis {#Sec21} ---------------------- ### Overall recombination rate per cell {#Sec22} Autosomal MLH1 signals, markers of CO events during meiosis I, were counted in 80 pachytene spermatocytes from the boar carrying t(1;14)(q2.11;q1.2) (Fig. [4a](#Fig4){ref-type="fig"}). The results were compared with those obtained for a pool of three normal boars (two Large White and one Meishan) in the study by Mary et al. [@CR37] (Table [1](#Tab1){ref-type="table"}). The average number of recombination events per cell was 33.42 in the t(1;14) carrier (range 25--42), i.e., significantly higher than the value of 31.86 (range 21--40) obtained for the control boars (*p* \< 0.05).Fig. 4Analysis of meiotic recombination in the t(1;14)(q2.11;q1.2) boar. **a** Immunostaining of pachytene spermatocytes. Pachytene spermatocyte after immunostaining of centromeres in *blue*, MLH1 signals in *green*, SCs in *red*, and FISH (*A1*). Schematic representation of the quadrivalent observed in *A1* (*A2*). *Scale bar* represents 10 μm. Distributions of MLH1 foci on chromosomes 1 and 14 (*B1*, *B2*). Comparison between rearranged chromosomes and controls (*red curve*: t(1;14); *blue curve*: control boars); *red bar*: centromere position on SSC1 for translocated boar; *blue bar*: centromere position on SSC1 for control boars; *purple bar*: centromere position on SSC14 for all individuals; *dotted bar*: position of the breakpoints. The *x-axis* represents the percent of SC length, and the *y-axis* indicates the frequency of MLH1 signals Table 1Average number of MLH1 foci per cell for chromosomes 1 and 14IndividualsNo. of cells analyzedAverage no. of recombination foci per cell (without sex chromosomes) and SERangeReferenceControls26431.86 ± 0.1821; 40Mary et al. ([@CR37])t(1;14)(q2.11;q1.2)8033.42 ± 0.39\*25; 42Present studyWilcoxon test\**p* \< 0.05 ### Recombination rates on the chromosomes involved in the translocation {#Sec23} Recombination was only studied in the spermatocytes with a quadrivalent configuration. The relative lengths of the SCs, expressed as a percentage of the total length of all the SCs within a cell, and the number of recombination events (MLH1 foci per chromosome) were measured for chromosomes 1 and 14 (part of the chromatin corresponding to SSC1 and SSC14 in the quadrivalent of the boar carrying the translocation) and compared to the values obtained for normal boars (Table [2](#Tab2){ref-type="table"}). Both the relative SC length and the number of MLH1 foci were significantly higher in the boar carrying the translocation than in the controls for chromosome 1 (*p* \< 0.05). In contrast, no significant difference was observed for chromosome 14.Table 2Comparisons of the relative lengths of SCs and of the number of recombination foci for chromosomes 1 and 14, between the translocated boar and a controlIndividualsNo. of cells analyzedChromosomeRelative length of SCs in % (1 SD)Number of recombination foci (ISE)ReferenceControls264110.64 10.962.74 10.04Mary et al. ([@CR37])148.15 11.281.94 10.03t(1;14)(q2.11;q1.2)80111.6511.39\*3.1210.09\*Present study148.15 11.241.91 10.07Wilcoxon test\**p* \< 0.05 ### Distributions of the MLH1 foci {#Sec24} The distributions of MLH1 foci observed along chromosomes 1 and 14 in spermatocytes from the boar carrying the translocation are presented in Fig. [4b](#Fig4){ref-type="fig"}. These distributions are clearly different (*p* \< 0.05) from those obtained in the normal boars. The most important modifications were observed in the breakpoint regions, where a decrease in the number of MLH1 signals was noted. Analysis of transcriptional activity in the spermatocytes {#Sec25} --------------------------------------------------------- Spermatocytes from the translocated boar were analyzed by RNA-DNA FISH to detect PERV gene expression. Eighteen of the 35 spermatocytes analyzed showed co-localization of the PERV-DNA signal and γH2AX protein. No PERV-RNA signal was observed in these cells (Fig. [5a](#Fig5){ref-type="fig"}). The PERV-DNA signals in the 17 remaining cells were located outside the γH2AX-positive regions. In these cells, the PERV gene was expressed (co-localization of the DNA and RNA signals; Fig. [5b](#Fig5){ref-type="fig"}).Fig. 5Analysis of PERV expression by sequential RNA- and DNA-FISH. **a** Spermatocyte showing a "trivalent plus univalent" configuration with two DNA PERV signals located in down-regulated γH2AX regions (no RNA signals). One of the PERV signals is co-localized with the large γH2AX region (probable XY body). **b** Spermatocyte showing a transcription of the PERV gene, located outside the γH2AX region. *Scale bars* are all equal to 10 μm We also measured the effect of t(1;14) on gene expression in the testis using an Agilent 60K customized porcine microarray. This enabled us to compare the transcriptome of the t(1;14) boar with the transcriptome obtained from a pool of control boars with normal semen parameters. Principal component analysis (PCA) highlighted the differences between the two groups, with all biological replicates clustering together (Fig. [6a](#Fig6){ref-type="fig"}). Interestingly, the first axis (68.4 % of the variance) was explained by the variation between t(1;14) and control boars, while the second axis (10.2 % of the variance) was explained by the heterogeneity between experimental replicates. We then compared probe intensity between t(1;14) and controls. We observed 2278 differential probes for t(1;14). Within the differentially expressed probes, 1707 (75 %) corresponded to a decrease in expression while 571 (25 %) corresponded to an increase in expression. We then mapped the differential probes on the porcine chromosomes and observed that the two chromosomes with the largest number of differential probes were chromosome 1 and chromosome 14, mostly for down-regulated probes (Fig. [6b](#Fig6){ref-type="fig"}). The X chromosome was the only one with more up-regulated than down-regulated probes (Fig. [6b](#Fig6){ref-type="fig"}). Nine genes on the X chromosome, identified through the microarray study, were up-regulated (LAMP2, TMEM47, SAT1, RGN, L1CAM, ZIC3, TSC22D3, PLP2, ARAF1). Real-time PCR was performed on six of them (ARAF1, LAMP2, RGN, SAT1, TSC22D3 and ZIC3). All six tested genes were more expressed in the testis from t(1;14), and the increase in ARAF1, LAMP2, RGN, SAT1, and TSC22D3 was significant and more than twofold (or higher) in the testis from the t(1;14) boar than in control boars (Supplementary Fig. [2](#MOESM4){ref-type="media"}).Fig. 6Gene expression profile from fertile and t(1;14) boar testis samples. **a** Principal component analysis of microarray transcriptomic data from testis RNA extracted from controls (*red*) and t(1;14) (*black*) boars. Gene expression profile from t(1;14) segregates differently from the control group and is reflected on axis 1 that explains 68.4 % of observed variance. The variance represented on axis 2 (10.2 % of variance) is mostly explained by the heterogeneity between experimental replicates (*circle*, *triangle*, and *cross*). **b** Number of differentially up-regulated (*blue bars*) and down-regulated (*red bars*) probes per chromosome in t(1;14) versus controls. Except for the X chromosome, down-regulated probes are always more abundant than up-regulated ones. **c**--**e** Mapping of differentially expressed probes (*blue dots*) on chromosome 1, chromosome 14, and chromosome X in t(1;14) versus controls. The *y-axis* corresponds to the log2 ratio where negative values correspond to down-regulated probes and positive values to up-regulated probes. Only probes with log2 values over 2 or under −2 were kept. The breakpoint is highlighted by a *vertical red line* We then determined the precise locations of the differential probes on chromosome 1, chromosome 14, and chromosome X. We first used array CGH to localize the DNA breakpoints at positions 293 924 898 on chromosome 1 and 148 080 on chromosome 14 (red lines on Fig. [6c, d](#Fig6){ref-type="fig"}). We found that the density of differential probes around the breakpoint was increased on chromosome 1 but not on chromosome 14 (Fig. [6c, d](#Fig6){ref-type="fig"}). Similarly, the probes were homogeneously distributed on chromosome X, except for a big gap of 50 Mb around the centromeric region (Fig. [6e](#Fig6){ref-type="fig"}). Analysis of meiotic segregation profiles {#Sec26} ---------------------------------------- The meiotic segregation profiles were studied using FISH of BAC clones on decondensed sperm heads (sperm FISH). These meiotic products originated from two different multivalent structures (quadrivalent or trivalent plus univalent configurations) in the presence (or not) of interstitial recombination. It was therefore impossible to determine the segregation mode at the origin of the different gametes unequivocally. We therefore employed the term "-like" for all the segregation products to indicate that their exact origin (segregation mode) could not be determined. The data are presented in Table [3](#Tab3){ref-type="table"}.Table 3Segregation patterns for the boar carrying the (1;14) translocationSegregationAlternateAdjacent IAdjacent II3:1Chromosomal constitution1/14[1]{.ul} ^[14]{.ul}^ [/14]{.ul} ^[1]{.ul}^[1/14]{.ul} ^[1]{.ul}^[1]{.ul} ^[14]{.ul}^ [/141]{.ul} ^[14]{.ul}^ [/114]{.ul} ^[1]{.ul}^ [/141]{.ul} ^[14]{.ul}^[1/14/14]{.ul} ^[1]{.ul}^[1]{.ul} ^[14]{.ul}^ [/14/14]{.ul} ^[1]{.ul}^1[1]{.ul} ^[14]{.ul}^ [/1/14]{.ul} ^[1]{.ul}^14[1]{.ul} ^[14]{.ul}^ [/1/1414]{.ul} ^[1]{.ul}^Number of gametes investigated[43036621380501234803754199]{.ul}% By combinations2723644 The most frequently observed segregation products were "3:1-like" (44 %), followed by "alternate-like" (27 %), "adjacent I-like" (23 %), and "adjacent II-like" (6 %). The overall proportions of balanced and unbalanced spermatozoa were 27 and 73 %, respectively. In the "2:2-like" segregation products, a strong disequilibrium was observed between the proportions of reciprocal products: 99 %/1 % for the adjacent I segregation (instead of the theoretical 50 %/50 % distribution) and 86 %/14 % for the adjacent II segregation. Strong disequilibria were also noted for the 3:1-like segregation products. Two types of gametes in that category were highly prominent: sperm containing SSC14 only (54 % of all 3:1-like segregation products) and sperm containing the derivative chromosome 14 only (29 % of all 3:1-like segregation products). Analysis of interchromosomal effects {#Sec27} ------------------------------------ Aneuploidy rates for the SSC13, SSC18, SSCX, and SSCY chromosomes were estimated in the boar carrying the (1;14) translocation and in a karyotypically normal control boar (Table [4](#Tab4){ref-type="table"}). No significant difference was observed between the (1;14) boar and the control (*p* ˃ 0.05), demonstrating the lack of interchromosomal effect (ICE) for this particular rearrangement (at least in the chromosomes studied).Table 4Analysis of interchromosomal effects: disomy, nullisomy, and diploidy rates for chromosomes 13, 18, X, and Y in sperm from control and t(1;14) boarsChromosome numberFrequency of disomy % (n)Frequency of nullisomy % (n)Frequency of diploidy % (n)Controlt(1;14)(q2.11;q1.2)Controlt(1;14)(q2.11;q1.2)Controlt(1;14)(q2.11;q1.2)130.009 (2)00.02 (4)0.01 (1)0.005 (1)0180.05 (11)0.03 (3)0.04 (9)0.02 (2)0.02 (5)0XY0.04 (9)0.03 (3)00.01 (1)XX or YY0.05 (11)0.06 (6)Total no. of spermatozoa20,22610,12720,22610,12720,22610,127*p* \< 0.05 when compared with the value of the control for the same chromosome (chi-square test) Discussion {#Sec28} ========== Meiotic pairing and transcriptional abnormalities {#Sec29} ------------------------------------------------- As suspected, our analysis of meiosis I (prophase) spermatocytes from the t(1;14) boar revealed meiotic pairing abnormalities, characterized by the presence of γH2AX-modified histone on the quadrivalents or univalents and some association with the XY body. Such phenomena have been already described in humans (Ferguson et al. [@CR18]; Leng et al. [@CR30]; Sciurano et al. [@CR58], [@CR59]). Sciurano et al. ([@CR59]) hypothesized that the movements of chromatin loops from autosomes which are actively transcribed, but not involved in the rearrangement, may randomly displace the inactive regions of the multivalent until they associate with the silenced chromatin domain of the XY body. Complementary RNA-DNA FISH experiments revealed a lack of transcriptional activity in the γH2AX-positive regions. Microarray gene expression analysis revealed 2278 differential probes on t(1;14) of which 75 % corresponded to down-regulated genes. The two chromosomes with rearrangements (i.e., chromosomes 1 and 14) also had the largest number of down-regulated probes. Transcriptional activation was consistently detected on the X chromosome with 73 % of the differential probes being up-regulated and confirmed by real-time PCR for six genes distributed all along the X chromosome and corresponding to up-regulated probes. These data for gene expression concord with our observation that MSUC occurred in meiocytes from this boar and previous observations reported in mice (Homolka et al. [@CR22]). Indeed, the autosome-autosome translocation studied by these latter authors was characterized by transcriptional down-regulation of the genes inside the unsynapsed region of the rearranged mouse autosome and an association between the quadrivalent and XY body, leading to X inactivation failure and incomplete silencing of the X chromosome genes in mid-late pachytene. Despite the fact that we were working on frozen testis biopsies rather than purified germ cell populations, we were able to observe significant effects on gene expression that seemed to be related to MSUC (X reactivation and silencing of the rearranged chromosomes). This would explain the meiotic arrest of some spermatocytes and the observed oligospermia in this boar. Impact of the translocation on meiotic recombination {#Sec30} ---------------------------------------------------- Eight human cases have been analyzed in the past using a comparable methodological approach involving immunolocalization of the MLH1 protein (Ferguson et al. [@CR18]; Jiang et al. [@CR26]; Leng et al. [@CR30]; Oliver-Bonet et al. [@CR44]; Pigozzi et al. [@CR47]; Sun et al. [@CR61]; Wang et al. [@CR72]). In four of these cases (t(Y;1), t(1;21), t(11;14), and t(5;7;9;13)), an effect of the rearrangement on the overall rate of recombination (increase or decrease) was observed (Leng et al. [@CR30]; Pigozzi et al. [@CR47]; Sun et al. [@CR61]; Wang et al. [@CR72]). The other four cases (autosome-autosome reciprocal translocations) showed normal rates of recombination (Oliver-Bonet et al. [@CR44]; Ferguson et al. [@CR18]; Jiang et al. [@CR26]) as did the only other reciprocal translocation (t(3;4)) studied in the pig species (Mary et al. [@CR37]). In our study, a significant increase of the (average) overall recombination rate was observed in the t(1;14) boar, as compared with a pool of normal boars (controls). Different factors (genetic background, age, environment, etc.) might explain the observed differences. Furthermore, intraindividual and interindividual variability of the recombination rate has been demonstrated in humans and pigs with normal karyotypes (Hassold et al. [@CR21]; Mary et al. [@CR36]). The specific analyses of the translocated chromosomes showed an increase in the MLH1 foci number and in the SC relative length for SSC1. This is consistent with a positive correlation between the number of MLH1 foci and the SC length observed in different species including humans (Pan et al. [@CR45]), pigs (Mary et al. [@CR36]), and mice with widely different rates of meiotic recombination (Baier et al. [@CR2]). Concerning the CO distribution on the paired SSC1 and SSC14 segments of the quadrivalents, significant differences were observed compared to their normal counterparts in control boars. Indeed, the recombination frequency was decreased in the vicinity of the breakpoint but appeared to be offset by an increase along the translocated chromosomes (Fig. [4](#Fig4){ref-type="fig"}). Similar results were obtained by Mary et al. [@CR37] for the t(3;4) translocation in pigs. This might indicate that as proposed by Mary et al. ([@CR37]), steric hindrance could occur in the breakpoint regions preventing the access of proteins involved in the recombination process. Abnormal meiotic pairing, not visible by immunolocalization (see Libuda et al. [@CR31]) occurring in these regions, could also prevent the formation of CO around the breakpoint. Moreover, we can suspect that, as reported by Libuda et al. [@CR31] in *C. elegans*, partial depletion of the synaptonemal complex central region proteins at the breakpoint attenuates crossover interference, thereby increasing crossovers in the flanking regions. Finally, this decrease might also be offset in other autosomal pairs by an unknown factor. Genetically unbalanced gametes production {#Sec31} ----------------------------------------- Five different segregation modes can occur in reciprocal translocation heterozygotes: alternate, adjacent I, adjacent II, 3:1, and 4:0 segregations, leading to the production of 18 possible types of gametes (Benet et al. [@CR5]; Sybenga [@CR62]). The relative frequencies of each segregation mode (as well as the relative frequencies of balanced and unbalanced segregation products) may vary from one translocation to the other, depending notably on the chromosomes involved, the location of the breakpoints, and the chiasma frequency (Rickards [@CR53]). Sperm meiotic segregation studies performed in human carriers of balanced reciprocal translocations showed that 18.6--80.7 % of the spermatozoa were chromosomally unbalanced (Benet et al. [@CR5]; Morel et al. [@CR41]). Until now, sperm FISH meiotic segregation analyses for three reciprocal translocations have only been carried out in pigs (Kociucka et al. [@CR28]; Massip et al. [@CR38]; Pinton et al. [@CR49]). One of these translocations (t(3;15)) was characterized by a small derivative chromosome (comparable in size to the der(14) chromosome observed in the present study) and a high proportion of 3:1 meiotic products (Pinton et al. [@CR49]). Similar results, i.e., a high rate of genetically unbalanced gametes (73 %), mostly of the 3:1 type (44 %), were observed in the present case. Meiotic pairing analysis of 284 spermatocytes revealed that 70 % formed a quadrivalent and 30 % of the cells exhibited a trivalent plus univalent configuration. CO distribution analysis (Fig. [5c](#Fig5){ref-type="fig"}) revealed a very low recombination frequency in the interstitial region (i.e., between the centromere and breakpoint) of chromosome 14 (red curve). Furthermore, no MLH1 signal was observed on the der(14) chromosome of 27.5 % of cells exhibiting a quadrivalent configuration. This suggests that der14 could segregate randomly to one of the gametes leading to a 3:1 segregation at the end of meiosis I. This is generally the case for the trivalent plus univalent configuration and would explain the relatively high rate of 3:1-like segregation meiotic products observed in our study. Four different types of 3:1 segregations were observed (Table [4](#Tab4){ref-type="table"}). Equivalent proportions of reciprocal segregation products would theoretically be expected for each category (e.g., similar proportions of "14" and "1^14^/1/14^1^" gametes). This was clearly not the case in our study (100 % of "14" gametes and 0 % of "1^14^/1/14^1^" gametes). Various authors (Estop et al. [@CR16]; Rives et al. [@CR55]; Van Assche et al. [@CR67]) suggest that spermatocytes with fewer chromosomes would be better able to survive a 3:1 segregation. Such a phenomenon could partly explain our results. As in earlier studies carried out in humans (Benet et al. [@CR4]; Brandriff et al. [@CR8]; Estop et al. [@CR15]; Martin and Spriggs [@CR35]; Shi and Martin [@CR60]; Templado et al. [@CR63]; Van Hummelen et al. [@CR68]), we observed a deviation from the 1:1 ratio of reciprocal segregation products for the adjacent I and adjacent II segregations. These results might be explained by the presence of unresolved chiasmata at meiosis I (Nicklas [@CR42]; Van Hummelen et al. [@CR68]). Unresolved chiasmata would affect the translocated segments in adjacent I segregation and non-translocated segments in adjacent II segregation, resulting in partial bivalents in meiosis II, which could lead to the segregation of homologous chromosomes, or to meiosis II arrest (Mckim and Hawley [@CR40]; Nicklas [@CR43]). Thus, a higher frequency of products carrying the short translocated segment would be expected in adjacent I segregation and a higher frequency of products carrying the shorter non-translocated segment in adjacent II segregation. This is consistent with our observations (Fig. [7](#Fig7){ref-type="fig"}) that a large excess of sperm carried the short translocated segment (1/14^1^) among the adjacent I products and a smaller amount of sperm carried the shorter non-translocated segment (14^1^/14) among the adjacent II products. Moreover, the low frequencies (example 1^14^/14) or absence (example 1^14^/1^14^, 14/14, 14/14^1^) of some meiotic products can be expected to result from meiosis II arrest.Fig. 7Illustration of sperm products from adjacent I segregation with unresolved chiasmata between translocated segments and from adjacent II segregation with unresolved chiasmata between non-translocated segments. Partial bivalents in meiosis II may then result in segregation of the homologous chromosomes instead of the chromatids which would produce different fluorescent sperm types or an arrest in meiosis II Interchromosomal effect {#Sec32} ----------------------- As classically carried out in human studies (Anton et al. [@CR1]; Douet-Guilbert et al. [@CR13]; Machev et al. [@CR33]; Piomboni et al. [@CR51]; Shi and Martin [@CR60]), ICE occurrence was investigated in chromosomes of different size and structure (SSC13 and SSC18, respectively, the largest and smallest acrocentrics, as well as the sex chromosomes). As in the study by Bonnet-Garnier et al. ([@CR7]) (investigation of ICE for two reciprocal translocations in the pig species), no ICE could be detected for any of the chromosomes tested. To our knowledge, the only publication reporting the occurrence of an ICE in the pig species is that of Pinton et al. ([@CR49]), which involved a boar mosaic for SSC18 trisomy. Unlike the situation in humans, very few cases have been investigated in the pig species until now, and complementary studies are needed to further document this point. Conclusion {#Sec33} ========== The main goal of this study was to investigate the meiotic disturbances that resulted in the infertility of a young OAT boar carrying an autosome-autosome asymmetrical reciprocal translocation. Our results showed that the translocated autosomes exhibiting pairing defects sometimes associated with the XY body (even if the synapsis of these autosomal regions with sex chromosomal axes has not been proved by our approach) led to a reduction in the expression level of some autosomal genes, as well as to an up-regulation of some X-chromosome genes. We hypothesized that the main cause of the meiotic arrest might be the gene silencing of asynapsed autosomal regions and/or an up-regulation of some X chromosome genes. Further studies will be necessary to see if these phenomena (disturbance of autosomal and X gene expression) are dependent on the nature of the chromosomes involved in the translocation or on the lengths of the translocated chromosomal fragments. Electronic supplementary material ================================= {#Sec34} Below is the link to the electronic supplementary material.Supplementary Table 1Semen parameters of the control group and t(1;14) boar used for gene expression analysis (PDF 6 kb) Supplementary Table 2Primers used for qPCR analysis (PDF 6 kb) Supplementary Fig. 1Histopathological analysis of the testes sample (Hematoxylin and eosin stain) Histological analysis of the testes shows altered spermatogenesis with few spermatozoa in the lumen and diffuse hyperplasia of the interstitial cells. (x40 magnification) *Scale bar* is equal to 200 μm (JPG 41 kb) Supplementary Fig. 2Real-time PCR quantification of 6 genes located on the X chromosome and described as up-regulated by microarray analysis. A: Localization of the 6 genes on the X chromosome is shown by red dots. B: Fold change of expression in testis from t(1;14) boar compared to control boars. Results are the mean of three independent experiments performed with biological replicates. (*t-test*, \*\*\*p˂0.001; n.s. not significant) (JPG 63 kb) AMCA : 1-Amino-4-methylcoumarin-3-acetic acid BAC : Bacterial artificial chromosome CO : Crossing-over DE : Differentially expressed FISH : Fluorescence in situ hybridization FITC : Fluorescein isothiocyanate HRP : Horse radish peroxidase ICE : Interchromosomal effect INRA : National Institute for Agricultural Research lfc : Log-fold change LH : Luteinizing hormone MLH1 : mutL homolog 1 MSCI : Meiotic sex chromosome inactivation MSUC : Meiotic silencing of unsynapsed chromatin OAT : Oligoasthenoteratospermia PCA : Principal component analysis PER V : Porcine endogenous retrovirus SC : Synaptonemal complex SCP1 : Synaptonemal complex proteins 1 SCP3 : Synaptonemal complex proteins 3 An erratum to this article is available at <http://dx.doi.org/10.1007/s10577-016-9534-8>. The authors would like to thank all the technical staff of the INRA GenESI experimental unit, and particularly Stéphane Moreau, Franck Guiraud, and Tony Terrasson for breeding the animals as well as Christian Audoux and Sylvain Michel for their technical assistance. We thank our colleague Valérie Fillon for her relevant comments on the manuscript. Ethical standards {#FPar1} ================= According to the European Directive 2010/63/EU on the protection of animals used for scientific purposes, the procedure for testes collection was accepted by the Ethics Committee for Animal Experimentation of the Poitou Charentes region (France) (CE2012-2), under the agreement number A-17-661. Conflict of interest {#FPar2} ==================== The authors declare that they have no conflict of interest. [^1]: Responsible Editor: Fengtang Yang

INTRODUCTION ============ Ambulatory care settings worldwide have dramatically shifted the inpatient surgical services to outpatient settings. Laparoscopic cholecystectomy has been the procedure of choice for symptomatic cholelithiasis around the world. Postoperative recovery time and the length of hospitalization have decreased significantly since routine cholecystectomy changed from an open to a laparoscopic procedure.^[@B1]^ Early positive results of ambulatory laparoscopic cholecystectomy, by Reddicke and Olsen in 1990,^[@B1]^ fueled its further growth, and it is now well accepted as a safe, cost-effective procedure for symptomatic gallstone disease. Various studies have documented the safety, feasibility, cost-effectiveness, and patient acceptability of this operation as an out patient procedure.^[@B1]--[@B9]^ Despite these results, it has only been practiced sporadically at centers in the UK and is not well established. Laparoscopic cholecystectomy has been routinely performed at this hospital, and patients have traditionally been admitted and discharged after an overnight stay. With the creation of a dedicated outpatient unit, ambulatory laparoscopic cholecystectomy (ALC) has been practiced since January 2002. The objective of this study was to evaluate postoperative morbidity and unplanned admissions, as well as readmissions following ambulatory laparoscopic cholecystectomy. We also tried to evaluate the cost savings of this procedure. METHODS ======= From January 2002 to April 2005 (40 months), 253 patients underwent laparoscopic cholecystectomy in the Department of General Surgery. Fifty-five patients had their gall-bladder removed as an inpatient, and 13 patients underwent bile duct exploration. ALC was offered to 198 of 253 well-motivated patients (79% day cases). The study was split into 2 phases **([Figure 1](#F1){ref-type="fig"})**. The first phase was a retrospective analysis of 112 patients from January 2002 to July 2003 (19 months). All medical records were reviewed to document patient characteristics, perioperative details, unplanned admissions, and readmission rates. The second phase was a prospective study involving 86 patients from August 2003 to April 2005 (21 months). Data were collected prospectively for these patients. {#F1} All patients with symptomatic gallstone disease, with no evidence of CBD calculi and who met the selection criteria were offered ALC. Patients who had a common bile duct stone were initially offered an endoscopic retrograde cholangiopancreatography and booked for ambulatory laparoscopic cholecystectomy if considered suitable. Systematic preoperative liver function tests and hepatic ultrasonography were performed. All patients were assessed at a preoperative assessment clinic before the operation. A fully trained surgeon was responsible for confirming the indications and eligibility for outpatient surgery after discussion with the patient. Only patients belonging to ASA grade 1 & 2 were included in the initial study, and a few dedicated patients with ASA grade 3 (12 in all) were considered at a later stage of the study. Another criterion for inclusion was that a responsible adult would be present with the patient for a 24-hour period postoperatively. Patients who presented as an emergency with acute cholecystitis and underwent cholecystectomy on their initial admission were excluded from the study. Patients at significant risk of requiring conversion to an open operation, such as those with previous upper abdominal surgery, were also excluded. All patients were scheduled for outpatient laparoscopic cholecystectomy in this hospital\'s purpose-built outpatient unit. Patients were admitted to the hospital on the morning of the operation, and every effort was made to accommodate them that morning, with the intention of discharging them in the evening. Consultants, associate specialists, and specialist registrars under supervision performed all surgeries. Preoperative cholangiography was not required in any of the patients. Surgery was performed with the patient under general anaesthesia and intubated. Standard 4-port video-laparoscopic cholecystectomy was performed. Hasson\'s method of access was used for CO~2~insufflation. All patients received preoperatively a single dose of broad-spectrum antibiotic and infiltration of local anesthetic to the wound. The anesthetic technique used for these procedures depended on the anesthetist responsible for each surgical session. Induction was with propofol, and intubation was facilitated with rocuronium. Maintenance included N~2~O/O~2~and an inhalational agent. Opiate and anti-emetic usage varied. All patients received either 8mg of ondansetron or 1mg of granisetron. Cycli-morphine was the most common opiate used, although pethidine was utilized in a significant number of cases. All patients received either diclofenac or parecoxib unless there was a contraindication to nonsteroidal anti-inflammatory drug use. At the conclusion of surgery, muscle relaxation was reversed using a neostigmine and glycopyrrolate combination. In recovery, IV analgesic continued with the intraoperative opiate as required. The patients were discharged before 8 p.m., with a responsible adult who could look after them for the first 24 hours, along with leaflets explaining the relevant postoperative advice and encouraging the patients to visit their own physician if they felt it were necessary. All patients were given a supply of a combination of codeine and paracetamol plus a nonsteroidal antiinflammatory drug for 48 hours. Patients who did not meet the discharge criteria, and those whose operation was converted to an open procedure, were admitted. The study ended 6 weeks after the surgery, with follow-up at the routine surgical clinic. RESULTS ======= Of 253 patients, 198 (79% day cases) underwent ambulatory laparoscopic cholecystectomy during the 40-month study period. All of the 112 patients in the first phase of the study were either ASA grade I or II. There were 90 women (80%) and 22 men (20%) with a mean age of 45 years (range, 21 to 78). Thirty-six (32%) patients were over 55 years of age. Surgery was successfully performed in all the patients without any open conversions. However, 23 patients required unplanned admission for different reasons **([Table 1](#T1){ref-type="table"})**. Six patients insisted on an overnight stay and were discharged the next day. Persistent nausea and vomiting was the cause of admission in 5 patients. Other causes included wound pain,^[@B5]^ urinary retention,^[@B2]^ operation in the afternoon,^[@B2]^ severe shoulder pain,^[@B1]^ and 2 patients needed admission after the placement of a suction drain. Twelve (50%) of the 23 patients admitted were more than 55 years of age. ###### Unplanned Admission and Readmissions (January 2002--July 2003) Number ------------------------------------------- -------- Reason for Admission (median=1 d; N=23)     Simple observation 6     Wound Pain 5     Nausea/Vomiting 5     Suction Drain 2     Urinary Retention 2     Operation in the afternoon 2     Severe shoulder pain 1 Reason for Readmission (N=4)     Wound related 3     Leaking cystic artery pseudo-aneurysm 1 In total, 4 (3.4%) patients were readmitted after discharge. Three of these, with wound-related complaints, either hematoma, minor wound infection or wound pain, were treated conservatively. One patient, admitted 10 days after discharge with a massive lower GI bleed, was found at laparotomy, to have a cystic artery pseudoaneurysm eroding into the transverse colon. He recovered well after undergoing surgery. Of the 86 patients in the second phase of the study, 72 (84%) were women and 14 (16%) were men 16 to 78 years of age (median, 48). Forty-three were \>55 years of age. Twelve well-motivated patients with ASA class III were also considered in this phase of the study in addition to classes II and I. In 3 patients, the laparoscopic procedure was converted to open cholecystectomy due to difficult dissection, not being able to identify the proper anatomy, or abnormal anatomy. An unexpected admission was required for 6 (7%) patients, including 3 who had undergone conversion to an open procedure **([Table 2](#T2){ref-type="table"})**. One patient required admission for analgesia and another for continuous nausea. One of the patients had a history of sleep apnea due to obesity; it was thought it would be prudent to observe him as an inpatient. ###### Unplanned Admission and Readmissions (August 2003--April 2005) Number --------------------------------------------------- -------- Reason for Admission (median=1 d; N=6)     Open conversion 3     Simple observation (obesity with sleep apnea) 1     Wound Pain 1     Nausea/Vomiting 1 Reason for Readmission (N=3)     Wound related 2     Bile leak 1 Of the 3 patients readmitted after discharge, 2 were treated conservatively for wound-related problems. One patient developed a biliary leak from CBD injury and was admitted 7 days after discharge with biliary peritonitis. Laparotomy revealed a lateral laceration to the common bile duct, which was repaired with t-tube drainage. DISCUSSION ========== Laparoscopic cholecystectomy has undergone a revolution since the advent of its being performed as an outpatient procedure. With continuing pressure on health service resources, there has also been a drive to reduce in-hospital stays and to increase the efficiency of procedures. The Audit Commission report^[@B10]^ of 1990 encouraged the expansion of outpatient procedures, and laparoscopic cholecystectomy fulfills this niche and has been performed in several centers with success. With an increase in outpatient procedures, it is necessary to evaluate the conditions in which admission for overnight stays could be kept to a minimum, although realizing that the "holy grail" of no admissions is, realistically, unobtainable. It is important to recognize the difference between studies that have evaluated outpatient cases, which relates to discharge on the same day of the procedure without requiring an inpatient bed, and other studies that include patients admitted overnight but discharged within 24 hours. In our study, we analyzed only those who were discharged on the same day of admission (before 8 pm). Whilst discharge the next day (within 24 hours) is admirable, and suggests good early mobilization, it still fails to satisfy the Audit Office criteria of true outpatient procedures.^[@B10]^ Unplanned admission after outpatient surgery is an indicator of quality assurance.^[@B11]^ All discharged patients in our study were reviewed at 6 weeks. The unplanned admission rate, whilst initially high at 21%, fell to a much more respectable 7% (overall 15%) in comparison with that of other centers, which varied from 3% to 39%.^[@B1],[@B12]--[@B17]^ The causes of postoperative morbidity were similar in both phases **([Table 3](#T3){ref-type="table"})** except that 3 patients (1.5%) had to have their laparoscopic procedure converted to an open procedure in the second phase, and this did not occur in the first phase. Conversion rate is comparable to reported rates of 1.8% to 6.7%.^[@B12]--[@B15]^ ###### Results From Phases 1 and 2 January 2002--April 2005 (N-198) Number --------------------------------------------------------------- -------- Reason for Admission (Median=1 d; range 1--3 d; N=29 \[15%\])     Open conversion 3     Simple observation 7     Wound Pain 6     Nausea/Vomiting 6     Suction Drain 2     Operation in the afternoon 2     Urinary retention 2     Severe shoulder pain 1 Reason for Readmission (N=7 \[3.5%\])     Wound related 5     Leaking cystic artery pseudo aneurysm 1     Bile leak 1 A drop in admission, from 21% to 7%, in the second phase is significant and needs to be analyzed further. Six patients were admitted for simple observation in the first phase. This was purely at the discretion of the patient; either they felt they were not fit enough to go home or there was low confidence amongst the nursing staff. This was evident in the second phase of the study when only one patient was admitted for observation as he had sleep apnea syndrome. Patients admitted for pain, nausea, or vomiting were also significantly reduced. Whilst there was not a universal anesthetic protocol, each patient received a preoperative opiate, NSAID, and antiemetic. We could not correlate the significant number of patients admitted with pain, nausea, and vomiting with any of the anesthetics or antiemetics used. Patients who were over the age of 55 years did not have a higher incidence of admission than those of a younger age group, contrary to the perception from the first phase of the study. Only 2 unplanned admission patients were aged above 55 years in the second phase, and both of them were ASA grade III. Previous reports emphasized the duration of procedure as one of the predictors of unplanned admission. In our study, the total operative time ranged from 16 minutes to 89 minutes (median, 35). The readmission rate of 3.5% compares well with a range of 0% to 8% reported by other authors.^[@B12]--[@B17]^ Admission would not have greatly changed the course of these patients, nor would it have prevented these complications from happening. However, biliary leak (7 days postop) would have been picked up earlier if the patient had been admitted. This patient and the patient with a pseudoaneurysm of the cystic artery (10 days postop) were read-mitted a week after initial discharge. Even if they had been operated on as an inpatient, they could have been discharged before the complication became evident. A patient with massive gastrointestinal bleeding deteriorated fairly rapidly and collapsed after admission. There was no clinical evidence of an aortic aneurysm, the possibility of angio-enteric fistula having been considered. Esophagogastroscopy performed with the patient under anesthesia did not reveal any active upper gastrointestinal bleed. Emergency laparotomy was performed. At operation, the large clotted blood was noted at the gallbladder fossa, and some blood-stained fluid was present in the abdomen. The proximal transverse colon was adherent to a large mass of clotted blood in the gallbladder fossa. Following evacuation of the blood clot, there was brisk bleeding from the cystic artery stump proximal to the clips. The end of the vessel was very necrotic. The vessel was under-sewn. A hole was identified in the antimesenteric border of the colon where it had been adherent to the organized blood clot. There was no true pseudocapsule around the blood clot to indicate clearly the presence of an organized pseudoaneurysm, and the exact cause of the fistulation into the colon was unclear. Electrosurgical injury to the cystic artery stump was possible during surgery as it was a difficult laparoscopic cholecystectomy. It appeared that the clotted blood mass had eroded into the colon and was responsible for the gastrointestinal hemorrhage. The small defect was oversewn and recovery was uneventful. It is a known fact that most of the early complications after laparoscopic cholecystectomy occur within a week after surgery. We felt that early review, either by nurse-lead telephonic review or review in a surgical clinic, would pick up the complications earlier. A further change that occurred between these 2 periods was the introduction of a checklist for use by the nursing staff. It was observed that, in the first period, the nursing staff were being asked to assess patients' fitness for discharge, having received no formal training, and fulfilling a role which, in this hospital, had been reserved for medically qualified staff. During the change, a major investment was made in educating nurses about their new role, and a checklist was drawn up to facilitate the nurses in this decision-making. Patients were discharged from the out-patient unit if they were tolerating oral fluids or a light diet, or both, with minimal nausea or vomiting, had passed urine, had adequate pain control and were ambulatory. A discharge letter was faxed to a referring general practitioner with operative details and recommended postoperative care. Consequently, this led to a marked reduction in the number of admissions for nausea and simple observations. Other studies have highlighted the effectiveness of a preoperative visit,^[@B5]^ and our study again shows that, with stringent preoperative assessment, low numbers of unplanned admissions can be obtained. The empowerment of the nurses yielded further rewards as the nurses decided to set up a team to allow follow-up of the patients. Up to August 2005, all the patients discharged were cared for by their own physician until their review in the routine general surgical clinic, 6 weeks after the operation. In September 2005, Telephone Nurse Interview Care Service (TONICS) was set up to review each case on Day 1 and then at 6 weeks following discharge. This proved to be an unqualified success, with only one person requesting a formal outpatient appointment, thereby freeing more of these appointments for new referrals or necessary reviews. It suggests that these patients do not require aggressive postoperative nursing care, after discharge, and that the availability of general practice or accident and emergency service may suffice instead of the costlier district nurse visit. Studies have shown that this is the case as long as a coherent and coordinated system of care is in place.^[@B9]^ Indeed, it may even be that patients prefer a telephone call to a home visit.^[@B18]^ Training has become an important issue as the government strives to ensure that the National Health Service fulfils its service commitments, often to the detriment of training the next generation of medical staff. It is vital that trainees are exposed to all aspects of patient care so as to be fully aware of ambulatory surgery and its place in the surgeon\'s armory. It would seem wise, though, to limit involvement to more experienced trainees so as to have minimal impact on the service commitment and the admission rate. It is also possible that the collection of certain cases in one fixed service may, in fact, be beneficial to the trainee, as it would provide a definite area in which the trainee could focus and develop the practice, especially in the climate of the New Deal and European Working Time Directive.^[@B3]^ Much has been debated about the financial impetus in the move to further outpatient procedures. A cost analysis was undertaken in our trust, which showed that, while the actual operative costs were similar, the real saving came because it was cheaper to carry out the outpatient procedures in their totality compared with elective admissions, and both these mechanisms of laparoscopic cholecystectomy were markedly cheaper than emergency admission. The average cost of the elective inpatient laparoscopic cholecystectomy was £1793 compared with £1174 for out-patient cases (Finance Dept., Causeway Hospital-Year 2002/2003). This would be in keeping with other studies that showed that there was a potential reduction in costs of 11% to 25% per patient.^[@B3]^ CONCLUSION ========== Outpatient laparoscopic cholecystectomy is safe, feasible, and desirable in the majority of patients, with few changes to current practice, and has become established practice at our institution. For the admissions to be kept to a minimum, the procedures should be performed by experienced staff, patients should be given pre-emptive anti-emetics, and analgesics, and experienced staff should be given the task of evaluating the discharge criteria. If this were established nationally, it would impact not only patient waiting times but also would result in significant cost savings. The authors acknowledge the assistance of Gore MG, Mansha M, Ervine E, Knox C, McGowan W. [^1]: Presented as a poster at The Association of Surgeons in Training (ASIT) Conference 2006, Edinburgh, UK, March 17--19, 2006. [^2]: Presented in part at Northern Ireland Surgical Trainees Prize Day, November 2004.

Introduction {#section1-1536012119871455} ============ Although the molecular pathways that lead to the pathogenesis and process of nonalcoholic fatty liver disease (NAFLD) and nonalcoholic steatohepatitis (NASH; a progressive form of NAFLD) remain poorly understood, it is accepted that inflammation, accumulation of extracellular matrix proteins, and proliferation of myofibroblasts are significant risk factors in hepatic injury.^[@bibr1-1536012119871455]^ In the past decade, research has focused on the molecular mechanisms involved in the development from hepatic steatosis to more advanced hepatic inflammation and fibrosis. A variety of different signaling pathways and specific biomarker phenotypes provide solid molecular biology basis for the development of hepatic molecular imaging tools. At present, liver biopsy is the gold standard for diagnosing liver disease and assessing the stage of fibrosis, but several limitations and/or adverse effects are associated with this invasive procedure, including pain, severe complications, and sampling error due to heterogeneous lesion distribution.^[@bibr2-1536012119871455]^ All these challenges present a unique opportunity for noninvasive methods for diagnosis and staging of NAFLD using translational molecular probes. Unlike ex vivo biopsy tests and histological testing, positron emission tomography (PET) / single-photon emission computed tomography (SPECT) imaging biomarkers enable direct characterization, quantification, and multispatial visualization of biological and cellular processes at different stages and the whole organism level. This approach would allow us to study a broad spectrum of physiological processes with the signal pathway, including proliferation, inflammation, and apoptosis, and evaluate various disease stages that represent the hallmarks of NAFLD.^[@bibr3-1536012119871455]^ In addition, these different biomarkers can be further used in more precise functional evaluation complementary to the current metabolic imaging. In this work, we give a brief review of PET/SPECT imaging biomarkers and their applications in the monitoring and staging of NAFLD and future perspectives in the radioligand development. Prevalence and Progression of NAFLD {#section2-1536012119871455} ----------------------------------- Nonalcoholic fatty liver disease is a prevalent type of chronic liver disease in both developed and underdeveloped countries. There is a high prevalence of NAFLD among those who have obesity, insulin resistance, cardiometabolic alterations, pattern, and metabolic syndrome. In developed countries, estimates of NAFLD prevalence vary between 20% and 30%, rising to 90% in morbidly obese populations. The more advanced form of NAFLD, NASH, carries a high risk of progressive fibrosis and cirrhosis and eventually developed to hepatocellular carcinoma (HCC).^[@bibr4-1536012119871455]^ Type 2 diabetes is closely associated with NAFLD---70% of patients having steatosis with type 2 diabetes---and thus it is now recognized to represent the hepatic indication of metabolic syndrome.^[@bibr5-1536012119871455]^ As shown in [Figure 1](#fig1-1536012119871455){ref-type="fig"}, NAFLD encompasses a broad range of hepatic pathology from simple fat accumulation (steatosis) to hepatic inflammation or fibrosis (NASH) and finally cirrhosis and even HCC. Although there is no inflammation and other symptoms in the steatosis stage, approximately 20% of patients with steatosis will continue to develop NASH. Nonalcoholic steatohepatitis occurs when there is persistent scar tissue in the liver, when the scar tissue starts to replace normal tissue, leading to cirrhosis. At the cirrhosis stage, the majority of liver function is significantly impaired, causing a high risk of HCC.^[@bibr5-1536012119871455]^ {#fig1-1536012119871455} Pathogenesis of NAFLD {#section3-1536012119871455} --------------------- Although the pathogenesis of NAFLD/NASH is still poorly understood, the most accepted concept about the pathogenesis of NAFLD involves multiple "hits." These hits might promote isolated steatosis, innate immune activation, inflammation, cell death, or progressive liver damage. ### Inflammation {#section4-1536012119871455} An accepted concept of NAFLD pathogenesis involved a "2-hit" process, in which the abnormal metabolic environment is causing lipid accumulation comprised of the "first hit," and this hit increases the susceptibility of the liver to secondary injuries ("second hit") in inflammation.^[@bibr6-1536012119871455]^ The severe consequences include mitochondrial dysfunction, overproduction, and the release of proinflammatory cytokines and chemokines, which notably include macrophage chemotactic protein 1, tumor necrosis factor α (TNF-α), interleukin (IL) 1β, and IL-1.^[@bibr7-1536012119871455]^ ### Hepatocellular injury and cell death {#section5-1536012119871455} Fibrogenesis occurs through phagocytosis when clearing apoptotic debris. During apoptosis, cells are divided into several small fragments, namely, apoptotic bodies.^[@bibr8-1536012119871455]^ The "professional" hepatic phagocytes, such as Kupffer cells (stellate macrophages) and hepatic stellate cells (HSCs), perform an important role in the clearing of apoptotic bodies.^[@bibr9-1536012119871455]^ The phagocytosis of apoptotic bodies is not merely a "clean-up" process to clear cellular corpses. Instead, phagocytosis initiates intracellular signaling transduction in the phagocyte, which leads to discrete immune responses including cytokines/chemokines generation and induces collagen type I, a biomarker of the cirrhotic scar. Especially, transforming growth factor β (TGF-β) is activated when cells engulf apoptotic bodies. The TGF-β is also a robust fibrogenic signal marker in the liver.^[@bibr10-1536012119871455]^ ### Necrosis {#section6-1536012119871455} Necrosis is a severe inflammatory mode of cell death compared to apoptosis. Necrosis occurs under several different conditions in the progression of hepatic diseases: adenosine triphosphate inhibition as a consequence of mitochondrial dysfunction; drug- or toxin-induced liver injury by xenobiotics; overgeneration of reactive oxygen species as it occurs during ischemia/reperfusion injury; and continual tissue injury as it occurs in chronic liver failure. The cell death model by noxious stimuli usually depends on the concentration, with low levels likely to lead to apoptosis and high levels inducing to necrosis. Therefore, the abovementioned stimuli might induce both modes of cell death at different time points, which depends on the severity of the injury.^[@bibr11-1536012119871455]^ Traditional Diagnostic and Staging Methods for NAFLD {#section7-1536012119871455} ---------------------------------------------------- ### Biopsy {#section8-1536012119871455} At present, liver biopsy is the gold standard for diagnosing liver disease and assessing the stage of fibrosis. In the absence of efficacious interventions for the chronic liver disease, liver biopsy allows rapidly to evaluate the histological components and identify their relationship with disease. Indeed, biopsy provides essential information for the diagnosis of NASH based on the presence of steatohepatitis. The histopathological range of NASH involves 4 major factors: steatosis, inflammation, hepatocellular injury, and fibrosis. Another significant advantage of liver biopsy is to provide an accurate semiquantitative evaluation of the severity of damage.^[@bibr12-1536012119871455]^ However, several limitations and drawbacks are present for this invasive procedure, such as pain, the reluctance of patients, and the risk of severe complications, and is subject to sampling error. Importantly, the person who performs it substantially determines the quality of the procedure. A successful procedure highly depends upon a well-trained hepatologist with sufficient experience. ### Noninvasive assessment {#section9-1536012119871455} Recently, several clinical markers have been utilized for predicting NASH and fibrosis. For example, the NAFLD fibrosis score is a scoring system to predict the advanced fibrosis based on several clinical parameters, such as age, body mass index, hyperglycemia, platelet count, albumin, and Aspartate aminotransferase (AST) - Alanine aminotransferase (ALT) ratio. When most of the biomarkers and scoring systems show similar accuracy for the detection of severity fibrosis, it is difficult to achieve high sensitivity for the diagnosis of early/mild fibrosis. False-positive results may also be caused by upregulation in bilirubin, decrease in haptoglobin, Gilbert syndrome, and cholestasis. In this context, noninvasive imaging techniques could be advocated as alternatives and diagnostic tests for NAFLD. Ultrasound-based transient elastography or FibroScan has shown prospective results for determination of the severity of liver fibrosis and degree of steatosis. However, it is also challenging in obese patients or those with ascites due to limited signal penetration as well as inadequate quantification of monitoring reversion in fibrosis after treatment.^[@bibr2-1536012119871455]^ In this context, nuclear imaging techniques including SPECT and PET targeted with novel signaling pathways may show target-specific and biological characterization in viability and metabolic activity in the progression of NAFLD.^[@bibr13-1536012119871455]^ The Application of PET and SPECT Molecular Probes in the Diagnosis and Staging of NAFLD {#section10-1536012119871455} ======================================================================================= Preclinical Models for Imaging NAFLD {#section11-1536012119871455} ------------------------------------ Animal models of NAFLD are essential in investigating the pathophysiological mechanisms of liver dysfunctions and diseases. To better our understanding of molecular targets that are associated with NAFLD, continuous efforts are contributed to the development of a variety of animal models to mimic the onset and progression of this process. There are currently several representative animal models of NAFLD, including NASH that develops due to high-fat dietary and essential nutrition factors defects and liver fibrosis induced by chemicals and based on surgery. The methionine and choline-deficient (MCD) diet is one of the frequently used and best described dietary models for NASH.^[@bibr14-1536012119871455]^ The MCD diet usually contains a high sucrose content (eg, 40%) and moderate amounts of fat content (10%) but is short of methionine and choline. Attributed to the elevated intake of fatty acids, rodents with an MCD diet usually develop hepatic steatosis, followed by necrosis and inflammation, eventually to pericellular and pericentral fibrosis.^[@bibr15-1536012119871455]^ Nuclear imaging of liver fibrosis has so far relied upon 2 models of chemically induced liver fibrosis and 1 surgical model, namely, carbon tetrachloride (CCl4) model, thioacetamide (TAA) model, and bile duct ligation (BDL) model. The CCl4 induces oxidative stress and necrotic response in the liver, generating toxic lipid and protein peroxidation products.^[@bibr16-1536012119871455]^ Chronic CCl4 administration produces extensive liver damage with necrotic hepatocytes, degenerated and ballooned, as well as features of macro- and microsteatosis and mild mononuclear cell infiltration in the affected areas. The TAA also induces oxidative stress by the oxidation of its sulfur species to the corresponding sulfur oxides and dioxides, leading to hepatic cytochrome P450 enzyme-linked hepatoxicity.^[@bibr17-1536012119871455]^ Bile duct ligation is the most common model used to induce obstructive cholestatic liver injury by surgical manipulation of bile acid circulation, which generates rapid-onset experimental hepatic fibrosis. Bile acids lead to dysfunction of farnesoid X receptor, liver X receptor, pregnane X receptors, and/or G-protein-coupled receptor TGR5, which are involved in a variety of metabolic and hepatic functions.^[@bibr18-1536012119871455]^ Excess bile acids accumulation leads to, in order of increasing severity in liver dysfunction, acute oxidative stress, necroinflammation, fibrosis, cirrhosis, and end-stage liver failure. Translocator Protein 18 kDa as a Molecular PET Imaging Biomarker for Liver Fibrosis {#section12-1536012119871455} ----------------------------------------------------------------------------------- Translocator protein 18 kDa (TSPO), a nucleus-encoded mitochondrial target transmembrane protein, has been indicated to play an essential role in the regulation of mitochondrial function and is increased in the inflammatory cells.^[@bibr19-1536012119871455]^ Overexpression of TSPO is a hallmark of inflammation. Elevated expression of this protein is reported not only during NAFLD but also in reactive retinal microglia and in a rat model of focal cerebral ischemia, which was imaged using (18)F-DPA-714 PET tracer.^[@bibr20-1536012119871455]^ Inflammation occurs during hepatic fibrogenesis. During the progression of liver fibrosis, activated HSCs (aHSCs) generated high TSPO expression, which may accelerate hepatic fibrogenesis. Moreover, TSPO expression has been shown in transformed HSC both in vitro and in vivo, and TSPO is also an inflammation biomarker in PET imaging.^[@bibr21-1536012119871455],[@bibr22-1536012119871455]^ Thus, TSPO is a useful biomarker to monitor the progression of liver fibrosis using specific PET radiotracers. From histopathology and autoradiography studies, Xie et al recently demonstrated that \[^18^F\]FEDAC ([Figure 2A](#fig2-1536012119871455){ref-type="fig"}) uptake was significant from injured hepatocytes and the necroinflammatory loci of CD11b+ macrophages in a mouse model of MCD diet-induced NAFLD^[@bibr23-1536012119871455]^ ([Figure 2B](#fig2-1536012119871455){ref-type="fig"}). These results suggested that inflammation may also be involved in NAFLD process, and \[^18^F\]FEDAC may be a potential imaging tracer for NAFLD. In another mouse model of hepatic fibrosis, Hatori et al demonstrated that TSPO-specific radioligand \[^18^F\]FEDAC provided noninvasive visualization of the progression from fibrosis to cirrhosis.^[@bibr24-1536012119871455]^ The experimental model of this study involved the induction of hepatic fibrosis by CCl~4~ exposure. This study showed significant uptake was mainly from HSCs and TSPO-expressing macrophage in 8-week CCl~4~ group (44.2 ± 0.7, standardized uptake value \[SUV\] × minute) compared to the control group (29.0 ± 0.3, SUV × minute; [Figure 3](#fig3-1536012119871455){ref-type="fig"}). Their results also confirmed the distribution of bound radioactive signals was associated with TSPO binding in the fibrotic liver. The messenger RNA expression level of liver TSPO and associated TGF-β1, PDGF-β, and TNF-α were upregulated by 6 weeks of CCl~4~ treatment. Level of TSPO was correlated positively with the expression of proinflammatory cytokine factors. In summary, these studies demonstrated that TSPO is highly expressed and accurately reflects the histological figure of NAFLD/NASH in murine models. \[^18^F\]FEDAC showed a sensitive and specific visualization and quantification during liver steatosis and fibrosis progression; it may be useful to help build a reliable, noninvasive method for imaging NAFLD. ![(A) Structure of \[18F\]FEDAC and (B) representative positron emission tomography (PET)/computed tomography (CT) images of the livers in methionine and choline-deficient (MCD) and control mice. (Courtesy of Dr Ming-Rong Zhang).](10.1177_1536012119871455-fig2){#fig2-1536012119871455} {#fig3-1536012119871455} Integrin α~v~β~3~ Targeted Imaging Studies {#section13-1536012119871455} ------------------------------------------ Integrins comprise many cell surface receptors by α and β subunits, and each αβ combination has its binding specificity and signal transduction pathway.^[@bibr25-1536012119871455]^ The integrin α~v~β~3~ is the major adhesion receptor that reacts to the ECM and thus plays an important role in control cell migration, proliferation, differentiation, and apoptosis, and dysregulated integrin αvβ3 receptor is another exciting target for NAFLD diagnosis/staging and cancer theranostics.^[@bibr26-1536012119871455]^ A common characteristic of integrins family, including α~v~β~3~, binds to ECM proteins by way of targeting arginine--glycine--aspartate (RGD) tripeptide sequence, a ligand that was previously used in cancer imaging and drug delivery.^[@bibr27-1536012119871455]^ Therefore, several cyclic RGD peptides (cRGD) are radiolabeled and developed as PET/SPECT tracers for binding integrin α~v~β~3~-positive targets.^[@bibr28-1536012119871455][@bibr29-1536012119871455]-[@bibr30-1536012119871455]^ Liver fibrosis studies have shown that integrin α~v~β~3~ exhibits high expression of aHSCs and promotes HSC survival and proliferation.^[@bibr31-1536012119871455]^ Li et al systemically detected the applicability of ^99m^Tc-labeled cRGDfK for SPECT imaging of HSC activity in fibrotic livers of TAA-treated rodent models.^[@bibr32-1536012119871455]^ The normal, moderate fibrotic (TAA treatment for 3 weeks) or severe fibrotic livers (9 weeks TAA treatment) could be distinguished by the mean radioactivity ratio of the liver to heart (MRAR) under SPECT imaging using \[^99m^Tc\]cRGDfK. Coadministration of cold cRGDfK can successfully block \[^99m^Tc\]cRGDfK uptake in the fibrotic liver; this confirmed the specificity of cRGDfk for liver uptake. Expression levels of integrin α~v~ and β~3~ subunits were enhanced with the progression of liver fibrosis and decreased with its regression. These results demonstrated that \[^99m^Tc\]cRGDfK was associated with α~v~β~3~ binding during the fibrotic liver disease. The binding affinity of integrin α~v~β~3~ can be further improved through using dimeric or multimeric cRGD peptides. Zhang et al further used \[^99m^Tc\]3PRGD~2~ ([Figure 4](#fig4-1536012119871455){ref-type="fig"}) in the TAA-induced liver fibrosis model of rats.^[@bibr33-1536012119871455]^ The radiotracer was bound specifically with the integrin α~v~β~3~ mainly expressed on the aHSCs. The MRAR was increased in the fibrotic livers compared to that of controls (TAA, 1.98 vs control, 1.50) at 30 minutes postinjection. The liver t~1/2~ was longer than in the controls (TAA, 27.07 ± 10.69 minutes vs control, 12.67 ± 4.10 minutes). Another work by Yu et al demonstrated \[^99m^Tc\]3PRGD~2~ was not only used to monitor the progression of liver fibrosis but also to measure the decrease in \[^99m^Tc\]3PRGD~2~ uptake in the fibrotic liver after antifibrotic therapy with drug interferon α2b.^[@bibr34-1536012119871455]^ ![Structure of \[^99m^Tc\]3PRGD~2~.](10.1177_1536012119871455-fig4){#fig4-1536012119871455} Asialoglycoprotein Receptors Imaging for Hepatic Fibrosis {#section14-1536012119871455} --------------------------------------------------------- Asialoglycoprotein receptors (ASGPRs) are well known to colocalize with the mammalian liver, of which 90% exists on sinusoidal faces of hepatocytes and only 10% on lateral faces.^[@bibr35-1536012119871455],[@bibr36-1536012119871455]^ The expression of ASGPR on the hepatocytes of patients with liver disease is reduced and recognized as a diagnostic biomarker for the evaluation of liver function. The previous study demonstrated ASGPR activity in the patients with cirrhotic liver was decreased to about 28% compared to the healthy controls.^[@bibr37-1536012119871455],[@bibr38-1536012119871455]^ \[^18^F\]FBHGal is a monovalent galactose derivative that was radiolabeled with fluorine-18 as ASGPRs-specific PET probe. Recent studies also reported several galactosides and *N*-acetyl galactosamine derivatives, including mono- and multivalent ligands, as potential substrates for ASGPRs in vitro. Kao et al performed biodistribution studies of \[^18^F\]FBHGal in both normal and dimethylnitrosamine-induced hepatic fibrosis mouse models.^[@bibr39-1536012119871455]^ In both studies, the receptor indexes (liver/liver plus heart ratio at 30 minutes postinjection) of hepatic fibrosis mice were significantly lower (*P* \< .01) compared to those of normal mice, and the accumulation of \[^18^F\]FBHGal in fibrosis liver (∼15%ID/g) was decreased compared to normal liver (∼21%ID/g) at 30 minutes postinjection. The protein expression level of hepatic ASGPRs in the liver fibrosis mouse was significantly decreased compared to that of normal mice. The results indicate that \[^18^F\]FBHGal ([Figure 5A](#fig5-1536012119871455){ref-type="fig"}) is a feasible agent for PET imaging of liver fibrosis. At the same time, Chang et al also have demonstrated that \[^99m^Tc\]MAMA-DGal is an ASGPRs-specific SPECT imaging probe for monitoring hepatic fibrosis.^[@bibr40-1536012119871455]^ The authors synthesized ^99m^Tc-labeled divalent galactosides, \[^99m^Tc\]MAMA-DGal ([Figure 5B](#fig5-1536012119871455){ref-type="fig"}), and performed SPECT imaging and biological characterization in normal and liver fibrosis mouse model. \[^99m^Tc\]MAMA-DGal provides significant specific binding to ASGPRs in normal liver than fibrosis and then rapidly excreted through both renal clearance and hepatobiliary system. Therefore, \[^99m^Tc\]MAMA-DGal could be used to reveal liver images and provide quantitative results for ASGPRs-related liver dysfunction. Recently, a new synthetic copolymer \[^99m^Tc\]*p*(VLA-co-VNI) was ([Figure 5C](#fig5-1536012119871455){ref-type="fig"}) reported by Zhang et al and imaging studies of \[^99m^Tc\]p(VLA-co-VNI) can identify different stages of liver fibrosis, which is CCl~4~-induced liver fibrosis in mouse models.^[@bibr41-1536012119871455]^ The authors first demonstrated ASGPR expression correlated with liver fibrosis progression. The liver uptake value (LUV) decreased along with the disease progression (control: 25.5 ± 1.58, 4 weeks of CCl~4~: 19.0 ± 2.12, 8-12 weeks of CCl~4~: 14.3 ± 2.41; [Figure 5D](#fig5-1536012119871455){ref-type="fig"}). After the antifibrotic Tan IIA treatment, LUV increased nearly 200% than the control group, which was then confirmed by the Sirius Red staining and hydroxyproline analysis. ![Structure of (A) \[18F\]FBHGal, (B) \[99mTc\]MAMADGal, (C) \[99mTc\]p(VLA-co VNI) and single-photon emission computed tomography (SPECT)/computed tomography (CT) images of control and carbon tetrachloride (CCl4)-induced fibrotic mice show the liver. (Courtesy of Dr Xianzhong Zhang).](10.1177_1536012119871455-fig5){#fig5-1536012119871455} Galactosyl Human Serum Albumin for Staging Fibrosis in NASH {#section15-1536012119871455} ----------------------------------------------------------- Galactosyl human serum albumin (GSA) is a synthetic analog ligand of ASGPRs. ^99m^Tc-labeled galactosyl human serum albumin (\[^99m^Tc\]GSA) has shown that it binds specifically to the ASGPR and allows estimation of regional hepatic function and the progression of chronic viral hepatitis in preclinical and clinical studies.^[@bibr42-1536012119871455],[@bibr43-1536012119871455]^ Haubner et al developed a ^68^Ga-labeled analog^[@bibr44-1536012119871455]^ for PET imaging studies ([Figure 6A](#fig6-1536012119871455){ref-type="fig"}). \[^68^Ga\]NOTA-GSA showed a significant increase in the metabolic stability in the liver and had a lower background activity in other organs. Schnabl et al performed ^68^Ga-labeled DTPA-conjugated neogalactosyl human serum albumin (\[^68^Ga\]DTPA-GSA; [Figure 6B](#fig6-1536012119871455){ref-type="fig"}), using T~90~ values to characterize \[^68^Ga\]GSA uptake in monitoring hepatic fibrosis and progression of NASH in rats.^[@bibr45-1536012119871455]^ In their PET imaging studies, animals with dominant pattern F0 (a rat with a healthy liver) to F1 (a rat with early or mild fibrosis) demonstrated significantly faster accumulation of \[^68^Ga\]GSA (*T* ~90~: 2.40 ± 0.52 minutes) than those with moderate to advanced dominant pattern fibrosis F2 (moderate fibrosis) and F4 (cirrhotic liver; *T* ~90~: 3.48 ± 1.01 minutes). These results demonstrated \[^68^Ga\]GSA accurately distinguishes early from mild experimental fibrosis independent of steatosis grade. ![Structure of \[^68^ Ga\]NOTA-GSA and \[^68^ Ga\]DTPA-GSA.](10.1177_1536012119871455-fig6){#fig6-1536012119871455} Targeting Desmin and Vimentin for aHSCs {#section16-1536012119871455} --------------------------------------- Desmin and vimentin are members of type III intermediate filament protein family and present in both muscle and nonmuscle cells. The expression of both vimentin and desmin is substantially increased during HSCs activation, and its protein level is higher than quiescent HSCs.^[@bibr46-1536012119871455],[@bibr47-1536012119871455]^ GlcNAc has a high affinity for desmin and vimentin; therefore, radiolabeled polyethylenimine-1800 (PEI-1800) modified GlcNAc can be used for targeting the HSCs. Zhang et al utilized \[^99m^Tc\]GlcNAc-PEI ([Figure 7A](#fig7-1536012119871455){ref-type="fig"}) to assess liver fibrosis in a CCl~4~-induced liver fibrosis mouse model.^[@bibr48-1536012119871455]^ The \[^99m^Tc\]GlcNAc-PEI imaging study showed the LUV in 8-week CCl~4~ treatment group was higher than that of 4-week group (4.7%/cc vs 3.3%/cc) and significantly increased compared to the control group (4.68%/cc vs. 2.34%/cc; [Figure 7B](#fig7-1536012119871455){ref-type="fig"}). In vivo imaging results were confirmed by ex vivo biodistribution studies. In addition, \[^99m^Tc\]GlcNAc-PEI was used to detect the therapeutic efficacy of liver fibrosis progression. After clodronate liposomes treatment, \[^99m^Tc\]GlcNAc-PEI uptake was reduced in fibrotic mice (control vs clodronate: 4.62%/cc vs 2.13%/cc). These results demonstrated \[^99m^Tc\]GlcNAc-PEI is a potential tracer to detect fibrosis progression and monitor the treatment of anti-fibrotic drugs. ![(A) Structure of \[99mTc\]GlcNAc-PEI and (B) single-photon emission computed tomography (SPECT)/computed tomography (CT) imaging of carbon tetrachloride (CCl4)-induced fibrotic mice with \[99mTc\]GlcNAc-PEI. (Courtesy of Dr Xianzhong Zhang).](10.1177_1536012119871455-fig7){#fig7-1536012119871455} Conclusion and Future Perspectives {#section17-1536012119871455} ================================== Over the past decade, there has been considerable progress focusing on the cellular and molecular mechanisms involved in the progression of NASH to fibrosis. The research progress of different signaling pathways and specific biomarker expression promote the development of hepatic molecular imaging ([Table 1](#table1-1536012119871455){ref-type="table"}). Although several PET/SPECT-specific molecular probes, particularly, radioactive molecules targeting TSPO, α~v~β~3~, ASGPR, GSA, and desmin/vimentin, have been used in the PET studies of liver fibrosis in animal experiments, there is no PET/SPECT tracer available for human use. Detection of the activation, transformation, and proliferation of myofibroblasts in the early stage of liver fibrosis is still challenging, which represents an unmet and urgent clinical need for imaging studies. Several factors should be considered when applying a PET/SPECT tracer for clinical translation. First, the liver is recognized as one of the most complex organs and contributes most metabolic activities in the human body. The liver consists of different cell types, including HSCs, Kupffer cells, sinusoidal endothelial cells, and hepatocytes, and the activation of HSC is the primary effector cell during liver fibrosis progress. Although TSPO, integrin αvβ3, desmin, and vimentin have high expression in aHSCs, these targets probes also showed nonspecific uptake by other cells, which contributed low signal to background ratio particularly in the early phase of PET/SPECT scans. Therefore, radiotracer labeled with relatively long half-life isotopes, such as ^18^F and ^64^Cu, and/or excretion mainly via a nonhepatic pathway to reduce background uptake and dosimetry, for example, renal clearance, may facilitate the development of new imaging probes for liver dysfunctions. It is also essential to design clinical trials and human translation research combined with emerging targeted therapy drugs for liver diseases and use PET/SPECT as companion readouts for target engagement and assessment for treatment efficacy. It will also be advantageous to identify individuals at high risk of disease progression and stratify patients who would likely benefit from specific targeted therapy (patient selection). Recently, there are several promising therapeutic/diagnostic targets that may provide an alternative approach for PET/SPECT probe development. To name a few, Caravan et al developed a type I collagen--targeted PET probe for detecting and staging pulmonary fibrosis.^[@bibr49-1536012119871455]^ The same probe may provide a translational tool for patients with liver fibrosis and other fibrotic diseases. The endocannabinoid system plays a crucial role in acute and chronic liver injury. Numerous studies in animal models of NAFLD^[@bibr50-1536012119871455][@bibr51-1536012119871455][@bibr52-1536012119871455][@bibr53-1536012119871455][@bibr54-1536012119871455][@bibr55-1536012119871455]-[@bibr56-1536012119871455]^ have implied that CB1 and CB2 receptors, as well as 2 degrading enzymes, namely, fatty acid amide hydrolase and monoacylglycerol lipase, may correlate liver disease states with dysfunction of the endocannabinoid system. In all, the current stage of liver imaging using radioactive probes is still at its infant step; therefore, a new generation of target-specific imaging probes with a novel mechanism of action, proper uptake and washout kinetic, and reasonable metabolic profile, is still urgently needed in the molecular imaging field of liver diseases. We anticipate that there will be more efforts and advances in the development of novel imaging probes to focus on liver diseases, which ultimately would provide better diagnosis and prognosis in personalized medicine. ###### Specific Molecular Tracers for NAFLD.  Molecular Probes Molecular Targets Liver Disease SPECT/PET Reference -------------------------- --------------------- ---------------- ----------- ------------------------------ \[^18^F\]FEDAC TSPO NASH PET ^[@bibr23-1536012119871455]^ Liver fibrosis PET ^[@bibr24-1536012119871455]^ \[^99m^Tc\]cRGDfK Integrin α~v~β~3~ Liver fibrosis SPECT ^[@bibr32-1536012119871455]^ \[^99m^Tc\]3PRGD~2~ ^[@bibr33-1536012119871455]^ \[^18^F\]FBHGal ASGPRs Liver fibrosis SPECT ^[@bibr39-1536012119871455]^ \[^99m^Tc\]MAMA-DGal ^[@bibr40-1536012119871455]^ \[^99m^Tc\]p(VLA-co-VNI) ^[@bibr41-1536012119871455]^ \[^68^ Ga\]NOTA-GSA GSA Liver fibrosis PET ^[@bibr44-1536012119871455]^ (\[^68^ Ga\]DTPA-GSA ^[@bibr45-1536012119871455]^ \[^99m^Tc\]GlcNAc-PEI Desmin and vimentin Liver fibrosis SPECT ^[@bibr48-1536012119871455]^ Abbreviations: ASGPR, asialoglycoprotein receptors; CT, computed tomography; GSA, galactosyl human serum albumin; NASH, nonalcoholic steatohepatitis; NAFLD, nonalcoholic fatty liver disease; PET, positron emission tomography; SPECT, single-photon emission computed tomography; TSPO, translocator protein 18 kDa. The authors thank Professor Thomas J. Brady (Nuclear Medicine and Molecular Imaging, Radiology, MGH and Harvard Medical School), Professor Ming-Rong Zhang (National Institute of Radiological Sciences, National Institutes for Quantum and Radiological Science and Technology, Japan), and Professor Xianzhong Zhang (Center for Molecular Imaging and Translational Medicine, Xiamen University) for helpful discussion. **Declaration of Conflicting Interests:** The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article. **Funding:** The author(s) received no financial support for the research, authorship, and/or publication of this article.

{#sp1 .427}

Introduction {#s1} ============ In their natural environment, plants are exposed to many kinds of stress, such as heat, drought and high light (Suzuki et al., [@B41]) especially during summer or due to anthropogenic activities including herbicides, air pollutants and acid rain (Lichtenthaler, [@B22]). For plants growing under field conditions, occasional exposure to high or even excessive light intensities is normal but has the potential to damage the photosynthetic apparatus. Exposure to high light (HL) generates reactive oxygen species (ROS) in chloroplasts, principally singlet oxygen in PSII (Krieger-Liszkay, [@B18]) and superoxide in PSI (Asada, [@B1]). To protect membrane lipids from photooxidation and PSII from photoinhibition, higher plants have developed a variety of adaptive strategies. In addition to modifications of pigment composition (Lichtenthaler et al., [@B23]) and chloroplast ultrastructure (Austin et al., [@B2]; Brehelin et al., [@B4]), plants synthetize various lipid and water soluble antioxidants such as tocopherol (vitamin E) and ascorbate, respectively (Delong and Steffen, [@B10]; Noctor and Foyer, [@B32]; Smirnoff, [@B39]; Smirnoff and Wheeler, [@B40]; Sattler et al., [@B36]; Havaux et al., [@B14]; Van Breusegem et al., [@B44]). Storage and metabolism of tocopherol but also phylloquinone (Vit K), plastoquinone (PQ) and its derivative plastochromanol (PC8) (Lohmann et al., [@B24]; Vidi et al., [@B45]; Szymanska and Kruk, [@B43]; Zbierzak et al., [@B48]; Eugeni Piller et al., [@B11]) in part take place at chloroplast lipid droplets (plastoglobules; PG) implicating them in light stress responses. PG are attached to the thylakoid membrane by the shared outer lipid leaflet. This arrangement leads to a conduit that may allow the diffusion of lipid molecules between the two compartments (Austin et al., [@B2]). At least, two metabolic enzymes involved in prenylquinone pathways are located at PG: the tocopherol cyclase VTE1 and NDC1 (Vidi et al., [@B45]; Ytterberg et al., [@B47]; Eugeni Piller et al., [@B12]; Lundquist et al., [@B25]). In addition, two kinases ABC1K3 and ABC1K1/PGR6 have been implicated in the regulation of PC8 production as well as α-tocopherol overaccumulation under HL. They may function via phosphorylation of VTE1 (Lundquist et al., [@B25]; Martinis et al., [@B27], [@B28]). Under HL stress, the synthesis of tocopherol is enhanced suggesting that this molecule exerts an essential role as lipid antioxidant (Munne-Bosch, [@B31]; Eugeni Piller et al., [@B11]). It has been shown that tocopherol is important in the maintenance of PSII function (Porfirova et al., [@B34]; Havaux et al., [@B14]). A large proportion of total plastid tocopherol is accumulated in the PG core which enlarges during oxidative stress (Vidi et al., [@B45]; Brehelin et al., [@B4]). The tocopherol head group is derived from homogentisic acid that is converted to 2-methyl-6-phytyl-1,4-benzoquinone (MPBQ) by the activity of homogentisate phytyltransferase VTE2 (Figure [1A](#F1){ref-type="fig"}) (Collakova and Dellapenna, [@B7]). Then MPBQ is methylated by VTE3 to form 2-3-dimethyl-6-phytyl-1,4-benzoquinone (DMPBQ) (Shintani et al., [@B38]; Cheng et al., [@B5]). VTE1 introduces the chromanol ring in MPBQ and DMPBQ leading to δ- and γ-tocopherols, respectively (Porfirova et al., [@B34]; Sattler et al., [@B36]). The last methylation step of tocopherol biosynthesis is catalyzed by VTE4 converting the δ- and γ-tocopherol into β- and α-tocopherol, respectively (Shintani and Dellapenna, [@B37]; Cheng et al., [@B5]). It has been demonstrated that mutations in Arabidopsis affecting steps of the tocopherol pathway (*vte1* and *vte4* mutants), strongly reduces the tolerance of photosynthetic organisms to HL stress (Maeda et al., [@B26]; Dellapenna and Pogson, [@B9]). ![**Tocopherols and tocochromanol biosynthesis and α-tocopherol oxidation pathway in Arabidopsis. (A)** Summary of tocopherols, plastoquinol, and plastochromanol pathways in Arabidopsis. The enzyme abbreviations were shown in red. HST, homogentisic acid solanesyl transferase; VTE, vitamin E synthesis; -PP, pyrophosphate; MPBQ, 2-methyl-6-phytyl-1,4-benzoquinone; DMPBQ, 2,3-dimethyl-6-phytyl-1,4-benzoquinone; MSBQ, 2-methyl-6-solanesyl-1,4-benzoquinol; PQH~2~, plastoquinol; PC8, plastochromanol. Adapted from Dellapenna and Kobayashi ([@B8]), Eugeni Piller et al. ([@B11]). **(B)** α-tocopherol recycling pathway in plants. The enzyme abbreviations were shown in red. LOO-, lipid peroxy radical; TC, tocopherol cyclase. Adapted from Dellapenna and Kobayashi ([@B8]), Mene-Saffrane and Dellapenna ([@B30]), Eugeni Piller et al. ([@B11]).](fpls-05-00298-g0001){#F1} In response to HL stress, plants accumulate tocopherol oxidation products. In contrast to animal membranes, only one such product has been reported to accumulate in plants, namely α-tocopherol quinol (α-TQH~2~) (Figure [1B](#F1){ref-type="fig"}) (Dellapenna and Kobayashi, [@B8]; Mene-Saffrane and Dellapenna, [@B30]). A part from being a product of tocopherol oxidation, several functions for α-TQH~2~ have been proposed: dissipation of excess energy, protection of PSII against photoinhibition (Kruk et al., [@B20], [@B19]; Munne-Bosch, [@B31]) as well as a strong antioxidant activity (Kruk and Trebst, [@B21]; Nowicka and Kruk, [@B33]). Recent studies demonstrate the existence of a plastid-based mechanism for a tocopherol redox cycle (Kobayashi and Dellapenna, [@B17]; Mene-Saffrane and Dellapenna, [@B30]; Eugeni Piller et al., [@B11]). In the first step of this recycling pathway, an α-tocopherol radical is formed by α-tocopherol oxidation via a lipid peroxy radical (LOO^−^). This compound is then oxidized by a second lipid peroxy radical to form α-tocopherol quinone (α-TQ) that is successively reduced to give α-TQH~2~. A yet unidentified plastid-dehydratase activity converts α-TQH~2~ to trimethylbenzoquinone (TMPBQ) (or TMPBQH~2~), which is then cyclized by VTE1 leading to the regeneration of α-tocopherol and completion of the cycle. In the present study, we used ultra-high pressure liquid chromatography-mass spectrometry to analyse the composition of prenylquinones that play a fundamental role in light stress response in a variety of genetic backgrounds. We report that the PG-localized NAD(P)H-dependent quinone oxidoreductase NDC1 participates in the tocopherol redox cycle. NDC1 most likely functions by reducing α-TQ to α-TQH~2~. This hypothesis is supported by the implication of NDC1 in the analogous reduction of PQ to PQH~2~ in PG under HL stress (Eugeni Piller et al., [@B12]). Materials and methods {#s2} ===================== Plant material and growth conditions ------------------------------------ *Arabidopsis thaliana* wild-type plants (WT) refers to var Columbia-2 (Col2). In this work, the *ndc1* mutant always corresponds to the T-DNA insertion line SALK_024063 from the Nottingham Arabidopsis Stock Center (<http://arabidopsis.info>; Alonso et al., [@B1a]). The mutant line *vte1*, obtained by EMS mutagenesis (Porfirova et al., [@B34]), and the overexpressing 35S:VTE1-YFP plants (Kanwischer et al., [@B15]) are a gift from Dr. P. Dörmann (Max Planck Institute, Golm, Germany). The 35S:NDC1-YFP plants were obtained as described below. Plants were grown on soil (Jiffy) under moderate light conditions (150 μmol m^−2^ s^−1^, 22°C, 8/16 h light/dark period) in a controlled environment room. For HL stress, 5 weeks old plants were exposed to 500 μmol m^−2^ s^−1^ (25°C, 8/16 h light/dark period). Overexpression of NDC1 in *A. thaliana* leaves ---------------------------------------------- Plants overexpressing NDC1-YFP under the 35S promoter were obtained using the Gateway recombination technology (Invitrogen): The NDC1 coding sequence was introduced into a donor vector pDONR™221, and subsequently transferred into an appropriate destination vector, the pEarlyGate101-YFP binary vector, resulting in pEarlyGate101-NDC1-YFP. pEarlyGate101-NDC1-YFP was transferred into Arabidopsis WT plants using the floral dip method (Clough and Bent, [@B6]). Transformed plants were selected for BASTA resistance and confirmed by segregation analysis. Western blot analysis --------------------- Total protein was isolated from Arabidopsis leaves according to Rensink et al. ([@B35]) and concentrated by chloroform-methanol precipitation (Wessel and Flugge, [@B46]). Twenty μg of protein were separated by SDS-PAGE and blotted onto nitrocellulose membrane for immunodetection. Immunodetection was carried out using anti-NDC1 serum at 1/1000 dilution in 5% fat free milk powder/TBS (Eugeni Piller et al., [@B12]). Confocal microscopy ------------------- Protoplasts were released from plants overexpressing NDC1-YFP by overnight digestion with macerozyme (0.25%, Serva) and reduced cellulase (1%, Serva) in a solution containing 400 mM Mannitol, 5 mM MES and 8 mM CaCl~2~. Protoplasts were filtered and loaded on a sucrose gradient (21 and 42%) and centrifuged for 10 min at 50 × g. Intact protoplasts were resuspended and fluorescence was monitored with a Leica TCS SP5 confocal microscope using the appropriate parameters for YFP (514-nm laser lines, 520--588-nm detection windows). Prenyllipid extraction from whole plants or PG fractions and lipidomics profiling --------------------------------------------------------------------------------- Prenylquinones were extracted from whole plants using an established method (Martinis et al., [@B29]). Leaves were ground in liquid nitrogen in a mortar. 100 mg were re-suspended in 500 μl of tetrahydrofuran (THF, analytical grade, Normapur). Glass beads (1 mm) were added and samples homogenized at 30 Hz, 3 min (Retsch MM 300). After centrifugation, 200 μl were transferred to a suitable HPLC vial. To measure prenylquinones contained in PG, intact chloroplasts were isolated by centrifugation on a Percoll gradient. Subsequently, PG were separated from thylakoid membranes by flotation on a sucrose gradient as described in Besagni et al. ([@B3]). Four hundred μl of PG and thylakoid fractions were added to 600 μl of water and extracted three times with an equal volume of ethylacetate. Organic phases were pooled, evaporated and pellets were dissolved in 100 μl of THF/water (85/15 v/v) and the solution transferred to an appropriate HPLC vial (Kessler and Glauser, [@B16]). The quantification of prenyllipids was performed using reverse-phase ultra-high pressure liquid chromatography (Acquity UPLC™, Waters) coupled to quadrupole-time-of-flight mass spectrometry (Synapt G2, Waters) (UHPLC-QTOFMS). Absolute concentrations of α-tocopherols, α-tocopherol quinone, plastochromanol and plastoquinone were calculated based on calibration curves obtained from pure standards. The method was also used for untargeted lipid profiling (Eugeni Piller et al., [@B12]; Martinis et al., [@B29]; Kessler and Glauser, [@B16]). Data pre-processing and statistical analysis -------------------------------------------- For comparison of metabolic profiles, raw spectrometric data were processed using Markerlynx XS™ (Waters) which performs automatic peak detection and deconvolution in each chromatogram. The parameters were as follows: initial and final retention times 0.5--3.0 min, mass range *m/z* 300--1200 Da, mass tolerance 0.02 Da, retention time window 0.10 min, automatic peak with detection, automatic measurement of peak-to-peak baseline noise, intensity threshold 400 counts, no smoothing, noise elimination level disabled, deisotope filtering function applied. Peak areas for individual variables were normalized to the total integrated area per sample. Variables were Pareto-scaled before applying principal component analysis (PCA). Statistical tests ----------------- Multivariate statistical analysis of lipid profiles was performed using EZinfo (Umetrics). Univariate Analyses were performed using the software SigmaPlot version 12.0. Data were first analyzed using Shapiro-Wilk to determine whether data were normally distributed. When data passed the test, the Student's test (*t*-test) was applied to evaluate statistically significant difference between values (*p* \< 0.05). For non-normally distributed data, the Mann-Whitney *U*-test was used. Results {#s3} ======= Characterization of NDC1 overexpressing plants ---------------------------------------------- To improve our understanding of prenylquinone biosynthesis and metabolic regulation, we used Arabidopsis wild type (WT) plants, *vte1* and *ndc1* mutants as well as 35S:VTE1-YFP and 35S:NDC1-YFP overexpressing lines exposed to either moderate light conditions or HL stress in lipidomics studies. With the exception of 35S:NDC1-YFP the different lines were characterized previously, including two independent homozygous *ndc1* mutant lines SALK_024063 and GABI_614F03 (Eugeni Piller et al., [@B12]) that showed no differences in phenotype or prenylquinone composition. The transgenic plants overexpressing NDC1 fused to yellow fluorescent protein (YFP) under the 35S promoter (35S:NDC1-YFP) were engineered for this study. The expression of NDC1-YFP was verified by Western blotting using an antibody against NDC1 (Figure [2A](#F2){ref-type="fig"}). 35S:NDC1-YFP plants but neither wild type nor *ndc1* gave a band at around 80 kDa corresponding to the predicted mass of the fusion protein. Endogenous NDC1 protein (57 kDa) was not detected in WT and overexpressing plants due to the low amount of total protein (20 μg) loaded (100 μg are necessary; Eugeni Piller et al., [@B12]). The 35S:NDC1-YFP plants had no apparent phenotype. {#F2} Protoplasts isolated from 35S:NDC1-YFP plants were analyzed by confocal microscopy and gave punctate fluorescence inside the chloroplasts which is in agreement with the PG localization of NDC1 (Vidi et al., [@B45]; Ytterberg et al., [@B47]; Eugeni Piller et al., [@B12]; Lundquist et al., [@B25]) (Figure [2B](#F2){ref-type="fig"}). Moreover, the 35S:NDC1-YFP construct was previously tested by transient expression in Arabidopsis WT protoplasts in the presence of the neutral lipid dye Nile Red which stains PG. The NDC1-YFP and Nile Red signals colocalized by confocal microscopy (Eugeni Piller et al., [@B12]). NDC1 and VTE1, two plastoglobule enzymes involved in prenylquinone metabolism ----------------------------------------------------------------------------- To understand the dynamics of prenylquinone synthesis under changing light conditions, we first analyzed the global lipid composition in *ndc1*, *vte1*, 35S:NDC1-YFP, 35S:VTE1-YFP, and WT genetic backgrounds. The data obtained, using the UHPLC-QTOFMS-based method, were subjected to multivariate analysis. Using this method more than 500 different compounds were detected, not all of which could be identified (Supplementary Table [1](#SM2){ref-type="supplementary-material"}). To investigate the difference in lipid contents, a principal component analysis (PCA) model was established from the data sets. PCA identifies and ranks major sources of variance and allows clustering of samples based on similarities and differences in the measured parameters. Under moderate light conditions, PCA showed the separation of five distinct groups characteristic for each of the genotypes tested in triplicate (Figure [3A](#F3){ref-type="fig"}). PCA loadings revealed that prenylquinones mostly contributed to the separation of these groups (Figure [3B](#F3){ref-type="fig"}). Most of the other lipids extracted were near the origin, suggesting that their contribution to metabolic difference was negligible. {#F3} As expected, *vte1* plants accumulated DMPBQ lacking the chromanol ring of tocopherols. In the wild type loadings, α-tocopherol appeared instead of DMPBQ. The 35S:VTE1-YFP plants accumulated additional VTE1 products: PC8 and γ-tocopherol. The separation of *ndc1* from the other genotypes was mainly based on the presence of the 2-phytyl-1,4-naphthoquinone, the de-methyl precursor of phylloquinone (Eugeni Piller et al., [@B12]). Interestingly, an accumulation of PQ was also observed in *ndc1*. The PCA score plot indicated similar prenylquinone-lipid compositions for a representative 35S:NDC1-YFP line and the WT under moderate light conditions. This finding together with the absence of a visible phenotype suggests that the insertion of the 35S:NDC1-YFP encoding T-DNA construct was without positional effects. To investigate the implication of VTE1 and NDC1 plants in prenylquinone metabolism under HL stress, we compared the lipid profiles of wild type plants with mutants after 4 and 8 days of HL exposure (Figure [4](#F4){ref-type="fig"}). Whereas *vte1* was blocked at the DMPBQ stage, WT plants strongly accumulated antioxidant lipids: γ-, α-tocopherol, PQ, PQH2, and PC8. (Figures [4A,B](#F4){ref-type="fig"}). *ndc1* differed from WT by the presence of the phylloquinone precursor and PQ but interestingly also by the accumulation of α-TQ (Figures [4C,D](#F4){ref-type="fig"}). {#F4} α-tocopherol quinone, an intermediate of the tocopherol redox cycle accumulates under HL stress in *ndc1* --------------------------------------------------------------------------------------------------------- To assess the quantitative impact of NDC1 and VTE1 on prenylquinones after HL stress, we quantified, using pure standards, the principal compounds that were distinguished by PCA: α-tocopherol and α-TQ. As expected under HL conditions, the levels of α-tocopherol (Figure [5A](#F5){ref-type="fig"}) and oxidized α-TQ (Figure [5B](#F5){ref-type="fig"}) increased in WT, about 3 (Student's *t*-test *p*~D0-D8~ = 0.0014) to 6-fold (*p*~D0-D8~ = 0.009) respectively after 8 days. ![**α-tocopherol and α-tocopherol quinone quantification in leaf and after HL treatment. (A)** α-tocopherol. **(B)** α-tocopherol quinone. Lipids were extracted from \[1\] WT, \[2\] *ndc1*, \[3\] 35S:NDC1-YFP, and \[4\] 35S:VTE1-YFP plants and quantified using purified standards. Plants grown under moderate light conditions (D0) were exposed to continuous HL (500 μE m-2 s-1) for 4 (D4) and 8 days (D8). Data are means of three experiments (*n* = 3).](fpls-05-00298-g0005){#F5} The level of α-tocopherol increased about 3-fold in *ndc1*, 35S:NDC1-YFP and 35S:VTE1-YFP after 8 days of HL, in the same manner as in WT (*P*~D8~WT-*ndc1* = 0.7, WT-35S:NDC1-YFP = 0.063, WT-35S:VTE1-YFP = 0.105). With regard to the concentration of oxidized α-TQ, pronounced differences were observed between WT and mutant plants. In WT, the concentration of α-TQ increased under HL stress. However, *ndc1* accumulated at least 3 times as much α-TQ as WT under moderate and HL conditions. The finding for α-TQ was confirmed for two *ndc1* T-DNA insertion alleles (Supplementary Figure [1](#SM1){ref-type="supplementary-material"}). In contrast, in 35S:NDC1-YFP and 35S:VTE1-YFP the level of α-TQ remained unchanged after HL stress and was about 8 times lower than in the WT at D8. NDC1 is implicated in the regeneration of reduced PQ and the formation of PC8 ----------------------------------------------------------------------------- The concentrations of total PQ including the proportion of the oxidized and reduced forms (Figure [6A](#F6){ref-type="fig"}) and of PC8 (Figures [6B,C](#F6){ref-type="fig"}) were measured. ![**Plastoquinone and plastochromanol quantification in leaf and after HL treatment**. Lipids were extracted from \[1\] WT, \[2\] *ndc1*, \[3\] 35S:NDC1-YFP, and \[4\] 35S:VTE1-YFP grown under moderate light conditions (D0) and after 4 (D4) and 8 (D8) days of continuous HL exposition (500μE m-2 s-1). **(A)** Total PQ was quantified using purified PQ as a standard. The white and gray bars indicating respectively the fraction of oxidized (ox) and reduced (red) PQ. **(B,C)** Quantification of PC8 using purified PC8 as a standard. Histogram presented panel **(C)** was a magnification of panels \[1, 2, 3\] panel **(B)**. Data are means of three experiments (*n* = 3).](fpls-05-00298-g0006){#F6} In WT plants, the concentration of total PQ was unchanged after 8 days of HL (*p*~D0-D8~ = 0.471) while a slight decrease of PC8 was observed (*p*~D0-D8~ = 0.013). In 35S:VTE1-YFP, the concentration of total PQ was generally lower than in the WT and this difference increased under HL (*p*~D0~ = 0.068, *p*~D4~ = 0.03, *p*~D8~ = 0.012). Concomitantly, PC8 concentration in 35S:VTE1-YFP was 10 times higher than WT under moderate light conditions and 27 times higher after 8 days of HL. As expected, PC8 was not at all detectable in *vte1* (data not shown). *ndc1* and 35S:NDC1-YFP, accumulated the highest concentrations of total PQ under HL. In *ndc1*, the concentration of oxidized PQ was significantly higher than in the WT after 8 days of HL (PQox *p*~D8~*ndc1*-WT = 0.0078). In contrast, in 35S:NDC1-YFP, the accumulation of the reduced form PQH2 made the difference (PQred *p*~D8~35S:NDC1-YFP − WT = 8.50e^−5^). As expected, the concentration of PC8 detected in *ndc1* mutant plants was significantly lower than in the WT, but no difference was observed between WT and 35S:NDC1-YFP (*p*~D8~WT-35S:NDC1-YFP = 0.1). Plastoglobules, a major compartment of prenylquinone metabolism and repair -------------------------------------------------------------------------- To analyze the distribution of the prenylquinones, we isolated chloroplasts from leafs of WT and *ndc1* after 7 days of HL and separated the thylakoid membranes and the plastoglobules. We then carried out prenyllipid profiling on whole leafs, isolated chloroplast, thylakoids and plastoglobules (Figures [7A,B](#F7){ref-type="fig"}). The galactolipids, MG 1; monogalactosyldiglyceride (18:3/16:3) and MG 2; monogalactosyldiglyceride (18:3/18:3), abundant chloroplast membrane lipids contributing to the thylakoids as well as the lipid monolayer of plastoglobules, were used as an internal reference to assess the enrichment of prenylquinones. Clearly, the prenylquinone compounds were enriched in plastoglobules, i.e., small peaks for MG1 and −2, large peaks for α-tocopherol and PQ/PQH~2~ when compared to the thylakoid membranes in both WT and *ndc1* (approximately 30- and 40-fold, respectively). Using γ-tocopherol as an internal reference (Figure [7C](#F7){ref-type="fig"}), increase of α-TQ and decrease of PC8 were observed in *ndc1* PG. For α-TQ, the enrichment in PG compared to thylakoid membranes was 3- and 5-fold, respectively in WT and *ndc1*. Note that α-TQ cannot be seen as peak in the chromatograms (Figures [7A,B](#F7){ref-type="fig"}) due to its relatively low abundance. As expected phylloquinone (Vit K) was detectable only in WT but not in *ndc1* PG. ![**Amount of prenylquinones in chloroplast fractions isolated from WT and *ndc1* mutant plants under HL**. UHPLC-QTOFMS chromatograms of chloroplast fractions showing prenylquinone enrichment in plastoglobules compared to other compartments **(A)** in WT and **(B)** in *ndc1* mutants. MG 1, monogalactosyldiglyceride (18:3/16:3), MG 2, monogalactosyldiglyceride (18:3/18:3), α-T (alpha-tocopherol), Vit K (phylloquinone), PQH2 (plastoquinol), PQ (plastoquinone), PC8 (plastochromanol), K-CH3 (demethylphylloquinone) after 7 days of HL. Data are means of four experiments (*n* = 4). **(C)** Quantity of α-TQ (α-tocopherol quinone), Vit K, PC8, PQ/PQH2, γ-T (gamma-tocopherol) was estimated from \[1\] WT and \[2\] *ndc1* plastoglobule fractions. Data are means of four experiments (*n* = 4).](fpls-05-00298-g0007){#F7} Discussion {#s4} ========== In this study we analyzed the dynamics of prenyl lipid metabolites during the change from moderate light to HL conditions using a non-targeted lipidomics approach. It is known that during acclimation to HL conditions, several prenylquinones accumulate in Arabidopsis leaves (Kobayashi and Dellapenna, [@B17]; Szymanska and Kruk, [@B43]; Eugeni Piller et al., [@B12]). The four Arabidopsis genotypes used in these experiments resulted in distinct prenylquinone signatures. The most typical compounds that accumulated in each of the respective lines were: DMPBQ in *vte1*, 2-phytyl-1,4-naphthoquinone in *ndc1*, δ-, γ-tocopherol and PC8 in 35S:VTE1-YFP and finally PQH~2~ in 35S:NDC1-YFP overexpressing plants (Figure [3](#F3){ref-type="fig"}). Overall, α-tocopherol was the prenylquinone that increased most during the course of light stress among the 500 compounds analyzed, except in the *vte1* mutant that lacks the tocopherol cyclase (Figure [4](#F4){ref-type="fig"}). This is testimony to the importance of α-tocopherol as a lipid antioxidant at the thylakoid membrane, which is subject to photooxidation and photoinhibition at PSII due to increased ROS production under HL stress (Kobayashi and Dellapenna, [@B17]). We previously showed that the α-tocopherol accumulation coincides with an increase in size and number of PG under HL stress (Martinis et al., [@B28]). This work demonstrates the implication of NDC1 in the tocopherol redox cycle. During HL stress, *ndc1* mutant plants accumulate the α-tocopherol oxidation product, α-TQ (Figure [4D](#F4){ref-type="fig"}). It has already been demonstrated that NDC1 is an enzyme with a wide specificity and able to reduce a range of quinolic substrates *in vitro* such as decyl-PQ, decyl-ubiquinone as well as in purified plastoglobules due to their contents of prenylquinones (Eugeni Piller et al., [@B12]). For α-tocopherol recycling to proceed efficiently it is likely that α-TQ must be present in the reduced form (α-TQH~2~). However, our current methodology does not allow the detection of α-TQH~2~. Nevertheless it is highly probable that NDC1 functions in the reduction of α-TQ to α-TQH~2~ to regenerate α-tocopherol. An analogous reaction mechanism has been demonstrated for the formation of γ-tocopherol, in which VTE1 closes the chromanol ring preferentially in the reduced form of DMPBQ (Grutter et al., [@B13]). NDC1 is also implicated in the reduction of PQ to PQH~2~ as 35S:NDC1-YFP plants exhibited higher PQH~2~/PQ ratios (Figure [6A](#F6){ref-type="fig"}). Thus, NDC1 may directly influence the redox state of the PQ reservoir. Most likely, this increase in PQH~2~ concerns primarily the proportion of the plastoquinone pool present in plastoglobules that is not directly implicated in photosynthesis (Eugeni Piller et al., [@B12]). The observed increase of total plastoquinone in 35S:NDC1-YFP plants (Figure [6A](#F6){ref-type="fig"}) may be necessary to maintain sufficient oxidized PQ to allow electron transport to proceed efficiently at the thylakoid membranes. 35S:VTE1-YFP plants showed a decrease of total PQ after HL stress (Figure [6A](#F6){ref-type="fig"}). It is important to note that VTE1 catalyses the production of PC8 from PQH~2~ (Zbierzak et al., [@B48]). Therefore, the decrease of total PQ is readily explained by the pronounced accumulation of PC8 in this genotype (Figure [6B](#F6){ref-type="fig"}). PQH~2~ may continuously be siphoned off as a substrate of VTE1 to form PC8 explaining the decrease of PQ and the increase of PC8 in 35S:VTE1-YFP. However, in agreement with Szymanska and Kruk ([@B42]), our results show that the production of PC8 is not influenced by HL in WT, *ndc1* and 35S:NDC1-YFP plants but an increase is apparent in 35S:VTE1-YFP plants. Potentially, this could be explained by an increased flux of PQH~2~to PG under HL in the presence of elevated concentrations of VTE1-YFP. In conclusion, we identify NDC1 as a novel enzyme that participates in the α-tocopherol redox cycle probably by reducing α-TQ to α-TQH~2~. In this recycling pathway, VTE1 was already known to convert TMPBQH~2~ to α-tocopherol. By hosting both NDC1 and VTE1, PG appear to play a role as metabolic repair site in the tocopherol redox cycle. This hypothesis is supported by the enrichment of α-TQ in the PG of *ndc1* compared to the WT (Figure [7C](#F7){ref-type="fig"}). Also, the overexpression of both NDC1-YFP and VTE1-YFP suppressed the increase of α-TQ observed in the wild type under HL (Figure [5B](#F5){ref-type="fig"}). This may be explained by higher levels of NDC1 or VTE1 activity that may affect the reaction kinetics of the tocopherol redox cycle and accelerate the reduction of α-TQ. Beyond its role in tocopherol recycling and plastoquinone reduction, *ndc1* lacks phylloquinone and accumulates its de-methyl precursor instead (Figure [4D](#F4){ref-type="fig"}) (Eugeni Piller et al., [@B12]). This indicates that NDC1 has "moonlighting" role in the final methylation step of phylloquinone biosynthesis that is catalyzed by AtMENG. In the future, it will be of great interest to investigate the mechanisms of NDC1 in more detail and to determine the role of this unusual enzyme in other species. Author contributions ==================== Felix Kessler, Gaétan Glauser, and Céline Besagni designed the research. Lucia Eugeni Piller, Céline Besagni, and Gaétan Glauser carried out the experimental work and statistical analyses. Felix Kessler and Céline Besagni wrote the manuscript. 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 work was supported by UniNE, SystemsX PGCE, NCCR Plant Survival, SNF 31003A_127380 and SNF 31003A_144156. Supplementary material {#s5} ====================== The Supplementary Material for this article can be found online at: <http://www.frontiersin.org/journal/10.3389/fpls.2014.00298/abstract> ###### **Amount of α-tocopherol quinone in leaf and after HL treatment**. Lipids were extracted from \[1\] WT and two different *ndc1* mutant lines: \[2′\] *ndc1* SALK_024063 and \[2′\] *ndc1* GABI_614F03. Plants grown under moderate light conditions (D0) were exposed to continuous HL (500 μE m-2 s-1) for 4 (D4) and 7 days (D7). Data are means of three experiments (*n* = 3). ###### Click here for additional data file. ###### **Full data set of lipid contents in Arabidopsis leaf under moderate light conditions**. Lipids were extracted by THF method from WT, *ndc1*, *vte1*, 35S:NDC1-YFP, 35S:VTE1-YFP plants and analyzed by UHPLC-QTOFMS. ###### Click here for additional data file. [^1]: Edited by: Gustavo Bonaventure, BASF Cropdesign, Belgium [^2]: Reviewed by: Marina Gavilanes-Ruiz, Universidad Nacional Autonoma de Mexico, Mexico; Claire Brehelin, Centre National de la Recherche Scientifique, France [^3]: This article was submitted to Plant Metabolism and Chemodiversity, a section of the journal Frontiers in Plant Science.

Introduction {#Sec1} ============ CT pulmonary angiography (CTPA), which requires the injection of iodinated contrast material (CM), is considered the gold-standard diagnostic modality in patients with suspected acute pulmonary embolism (PE)^[@CR1]^. However, caution is warranted in patients with renal risk factors such as acute and chronic nephropathy. Guidelines of the European Society of Urogenital Radiology recommend to use the lowest amount of contrast medium necessary^[@CR2]^. This may be subject to change since administration of intravenous CM in patients with renal risk factors is part of an ongoing controversial discussion and several recent studies have uniformly reported promising results regarding adverse renal reactions such as contrast induced nephropathy (CIN) or acute kidney injury (AKI). Aforementioned studies have indicated that CIN and AKI may be independent from the administration of intravenous CM among different risk groups^[@CR3]--[@CR6]^. Although those retrospective high quality studies may impact future guidelines, some limitations remain, including selection biases, limited information for patients undergoing several CT examinations within a few days and the impact of intravenous CM's amount on acute kidney injury. Thus, if intravenous CM is essential for CT examinations, a reduction to a minimum amount of CM remains recommended, similar to the "*as low as reasonably achievable*" principle for radiation dose reduction^[@CR2]^. Over the last decade, low peak-kilovoltage (kVp) imaging between 70--80 kVp in combination with high-pitch acquisitions has allowed a significant reduction of CM for CTPA protocols with a total dose amount down to 9--21 g^[@CR7]--[@CR9]^. Low kVp imaging is currently limited to 70--80 kVp across all CT manufacturers due to a reduced efficiency of x-ray tubes with insufficient photon emission, increasing image noise at lower kVp levels. Dual-energy (DE) CTPA, which allows the calculation of virtual monoenergetic spectral (VMS) images at various kilo-electron volt (keV) levels with a minimum of 40 keV, can be used to promote iodine based contrast enhancement as an alternative CTPA imaging technique to further reduce CM. Studies have evaluated the feasibility of DE CTPA using a standard CM-, as well as a reduced CM-protocol^[@CR10]--[@CR12]^. However, these studies have used a first generation VMS software technique, which lacks noise compensation at low keV levels, thus limiting diagnostic image quality below 60 keV. Recently, a noise-optimized VMS algorithm was introduced to improve contrast-to-noise ratio at energy levels below 60 keV. Therefore, the aim of this study was to investigate whether DE CTPA in combination with a recently introduced noise optimized VMS algorithm allows for a further reduction of CM in CTPA studies of patients with renal risk factors. Methods {#Sec2} ======= Study design {#Sec3} ------------ This prospective single-center study was approved by the institutional review board (Medical Ethics Committee II, Mannheim, Germany) and complies with both the Declaration of Helsinki and the Health Insurance Portability and Accountability Act (HIPAA). Informed consent was obtained from all participating patients. In total, 150 patients were enrolled in this two-phase study. During the first part of the study 100 patients with suspected PE were equally (1:1 ratio) randomized either to a standard CTPA examination or a DE CTPA examination. Independently of a patient's renal function a standard CM injection protocol was performed to evaluate non-inferiority of the DE CTPA protocol. The main objective of this first phase was to assess image quality (contrast-to-noise-ratio \[CNR\] and image noise) of VMS datasets between 40--70 keV compared to images of a standard non-DE CTPA protocol. Subjective and objective image quality were analyzed one month after the inclusion of the last patient. Subjective image quality was evaluated by two radiologists in a consensus reading (eight and four years of experience in chest CT) using a 5 point Likert scale as previously described^[@CR13]^. Non-inferiority of the DE CTPA protocol was defined as a decrease in CNR of less than 5% of the pulmonary arteries and overall subjective image quality. In this first phase, the secondary objective was to evaluate diagnostic accuracy of VMS datasets between 40--70 keV of the standard DE CTPA in comparison to the 120 kVp virtual polyenergetic spectral (VPS) images. The diagnosis of PE and its severity were classified by two radiologists (T.H.; M.M.) according to recent guidelines^[@CR1],[@CR14]^. Non-inferiority of the DE CTPA protocol was defined as *no false positive or negative PE finding on 40*--70 *keV VMS datasets*. In a second phase, after non-inferiority of the DE CTPA protocol for VMS datasets between 40--70 keV was proven, 50 consecutive patients with renal risk factors underwent DE CTPA using the low CM-protocol. The main objective of this second study phase was to evaluate the objective image quality (CNR and image noise) of the low CM DE CTPA in comparison to the standard CTPA protocol. The secondary purpose of this second study phase was to evaluate the diagnostic accuracy of the low CM DE CTPA as a prerequisite to be implemented into clinical routine. The diagnosis of PE and its severity were classified as described above. One week after each scan, the same radiologists (T.H.; M.M.) analyzed the DE CTPA data in consensus regarding the presence or absence of PE (blinded to the previous rating). The standard of reference was established in consensus by both radiologists after each patient's reading based on the clinical course, including response to therapy and diagnostic follow up, in compliance with the STARD criteria^[@CR15]^. An exploratory objective of this study was to monitor serum creatinine levels before and 24 hours, 72 hours and 3 months after CTPA related CM application. Patients were questioned for re-administration of iodine CM during the follow-up period of 3 months. CIN was defined as an absolute increase in serum creatinine from baseline by ≥ 0.5 mg/dl, indicating acute impairment in renal function^[@CR16]^. Examination Technique {#Sec4} --------------------- The low CM DE CTPA protocol was performed on a 3^rd^ generation dual-source CT (DSCT) (SOMATOM Force, Siemens Healthineers, Forchheim, Germany) with the following scan parameters: tube A: 90 kVp tube voltage, no tin filter, 100 mAs reference tube current; tube B: 150 kVp tube voltage, 0.6 mm tin filter, 90 mAs reference tube current. Rotation time was 0.25 s, pitch 1.0 and the chosen detector collimation 192 × 0.6 mm for both tube systems. The contrast agent administered was iomeprol 400 mg Iodine/mL (mg I/mL, Imeron 400, Bracco Imaging S.p.A., Milan, Italy), diluted via an injector to a concentration of 30% to 70% 0.9% NaCl totaling 45 ml and 5.4 g iodine content. This was followed by a saline chaser of 30 ml. Flow rate was 3 ml/s through an antecubital vein. The following scanning parameters for the standard CTPA protocol (Siemens SOMATOM Definition Flash, Healthineers, Forchheim, Germany) were used: 100 kV tube voltage, no tin filter, 140 mAs reference tube current, 0.28 s rotation time, 1.2 pitch, 128 × 0.6 mm detector collimation. The following scan parameters for the DE CTPA on the same scanner were used: tube A: 100 kV tube voltage, no tin filter, 90 mAs reference tube current; tube B: 140 kV tube voltage with tin filter, 80 mAs reference tube current. For both tube systems the rotation time was 0.28 s, pitch 0.55 and the chosen detector collimation 128 × 0.6 mm. Vessel attenuation was achieved for both standard protocols by injecting 80 mL iomeprol 400 mg I/mL (32 g iodine content) followed by a saline chaser of 40 ml through an antecubital vein at a flow rate of 4 mL/s^[@CR17]^. Both CTPA protocols utilized an automatic tube current modulation (CARE Dose4D, Healthineers, Forchheim, Germany). The CT acquisition was timed by using bolus tracking for both CTPA protocols with a region-of-interest (ROI) placement in the pulmonary trunk. Once a threshold of 100 Hounsfield Units (HU) was reached (standard CTPA protocol triggering at 100 kVp and DE CTPA protocol triggering at 70 kVp), the scan automatically started after a 5 second delay, allowing for a breathing command ("gently breath in"). Table [1](#Tab1){ref-type="table"} summarizes scan acquisition parameters, contrast injection protocols and image reconstruction techniques for the three different CTPA protocols.Table 1Patient demographics, iodine load and dosimetric parameters.Standard CTPA protocol n = 50 \[A\]Low CM DE CTPA protocol n = 50 \[B\]Standard CM DE CTPA Protocol n = 50 \[C\]p-value A vs Bp-value A vs Cp-value B vs CAge \[years\]64 ± 2267 ± 1067 ± 150.30130.38760.8412Male2627250.77\>0.990.8415Body-mass-index (kg/m^2^)26.1 ± 4.227.1 ± 3.127.9 ± 4.50.42320.38760.8764Pulmonary embolism776---\>0.99\>0.99Central pulmonary embolism343\>0.99\>0.99\>0.99Right ventricular dysfunction222\>0.99\>0.99\>0.99Normal creatinine clearance \[\>60 ml/min\]40046\<0.00010.9796\<0.0001Acute kidney failure4292\<0.0001\>0.99\<0.0001Baseline creatinine level \[mg/dl\]0.98 ± 0.21.37 ± 0.30.95 ± 0.4\<0.00010.3623\<0.0001Creatinine level after 24 h \[mg/dl\]1.03 ± 0.21.38 ± 0.40.99 ± 0.4\<0.00010.2221\<0.0001Creatinine level after 72 h \[mg/dl\]1.04 ± 0.21.35 ± 0.30.93 ± 0.5\<0.00010.0925\<0.0001Creatinine levels after 3 Months \[mg/dl\]1.01 ± 0.21.21 ± 0.30.93 ± 0.4\<0.00010.0667\<0.0001Total amount of iodine \[g\]32.05.432.0\<0.0001---\<0.0001CTDIvol \[mGy\]4.65 ± 2.66.83 ± 3.47.91 ± 3.90.0003\<0.00010.0321DLP \[mGy\*cm\]169.2 ± 91.4243.3 ± 117.3285.5 ± 123.10.0012\<0.00010.0215Effective dose \[mSv\]2.4 ± 1.33.4 ± 1.63.9 ± 1.7Note -- CTPA = CT pulmonary angiography; DECT = dual energy CT; DLP = Dose length product; CTDI~vol~ = volume CT dose index. Image Reconstruction {#Sec5} -------------------- For the DECT protocol 120 kVp VPS images were reconstructed at a weighing factor of 0.8 and 0.4 for the low CM and the standard DE CTPA protocol, respectively, and as recommended by the manufacturer. All DE images were reconstructed with a slice thickness of 1.5 mm in axial and coronal planes using a corresponding quantitative DE kernel (2^nd^ generation DSCT Q30f and 3^rd^ generation DSCT Qr40) and a soft tissue kernel for the standard CTPA protocol (I30f). All images were reconstructed using an iterative reconstruction (IR) algorithm. In detail, the 3^rd^ generation DSCT utilized a model based iterative reconstruction method (Advanced Model Based Iterative Reconstruction \[ADMIRE\] Siemens Healthineers, Forchheim, Germany) and the standard CTPA protocol a sinogram affirmed iterative reconstruction algorithm (Sinogram Affirmed Iterative Reconstruction \[SAFIRE\], Siemens Healthineers, Forchheim, Germany). The iterative reconstruction algorithm was set to a strength level of three for all three CTPA protocols, as recommended by the manufacturer and previously validated for low kV imaging of vascular structures^[@CR18]^. The images were exported to a server based workstation (Syngo.via, Version VA30, Siemens Healthineers, Forchheim, Germany) for further post-processing and measurements in a timely manner. VMS images were calculated for DE datasets at energies between 40 and 100 keV with increments of 5 keV (Monoenergetic Plus, Siemens Healthineers, Forchheim, Germany). (Fig. [1](#Fig1){ref-type="fig"})^[@CR19]^.Figure 1Simplified image reconstruction work flow of the monoenergetic reconstruction algorithm. A frequency-split technique is used which decomposes both the low keV images (in which iodine pixels have a high contrast to the surrounding tissue, typically at 40 keV) and images at higher keV (in which surrounding tissue has low image noise, typically at approximately 70 keV) into two sets of sub-images. In a next step, the lower spatial frequency stack at low keV is combined with the high spatial frequency stack at optimal keV from a noise perspective to combine the benefits of both images stacks. Objective Image Quality Assessment {#Sec6} ---------------------------------- Attenuation (HU) and image noise (standard deviation of attenuation) were measured for identical sized ROIs in three regions of the central pulmonary arteries (main, left and right pulmonary artery) and two peripheral pulmonary arteries (one lower and one upper lobe segmental artery) by a radiology resident (M.M), as previously described^[@CR13]^. Further identical sized ROIs were placed in background air (for DE datasets in the VPS images) in order to assess background noise. These parameters were used to determine the SNR and CNR with the erector spinae muscle being the surrounding tissue^[@CR20]^. For further analysis, the averages of the three central pulmonary arteries and the two peripheral pulmonary arteries were calculated. Subjective Image Quality Assessment {#Sec7} ----------------------------------- All technical and personal data were removed from the images prior to the subjective image quality assessment. Subjective image quality was analyzed by a board certified radiologist and a radiology resident (T.H., M.M.) both independently and in a consensus reading. This two-sided approach was chosen to evaluate inter-reader variability of the new contrast media protocol. Both radiologists rated the overall image quality (1 = poor, 2 = fair, 3 = moderate, 4 = good, 5 = excellent) and image noise (1 = unacceptable image noise, 2 = above average noise, 3 = average image noise, 4 = less than average noise, 5 = minimal image noise) on a 5 point Likert scale as previously described^[@CR13]^ and in accordance with the criteria described for chest CT examinations in the European Guidelines on Quality Criteria for CT^[@CR21]^. Radiation dose assessment {#Sec8} ------------------------- The volume CT dose index (CTDI~vol~) and dose length product (DLP) in the patient protocol were used to estimate radiation exposure. The conversion factor to effective organ dose was chosen according to the 'European Guidelines for Multislice Computed Tomography (dose conversion coefficients of thoracic region: 0.0014 mSv Gy^−1^ cm^−1^;^[@CR22]^). Statistical analysis {#Sec9} -------------------- A two-sided binominal test with a significance level of 0.05 and a power of 0.8 were used for calculating the required sample size using the statistical software PASS 11, (LLC. Kaysville, Utah, USA. [www.ncss.com](http://www.ncss.com)). Statistical analysis was performed using JMP 10.0 (SAS Institute Inc., Cary, NC, USA). Normally distributed data were identified using the Shapiro-Wilk W test. Continuous variables are presented as mean ± standard deviation and ordinal variables as median with a 25% to 75% interquartile range. Comparisons between CTPA groups and serum creatinine time points were either using either a two-way analysis-of-variance (ANOVA), if data were normally distributed, or a Kruskal-Wallis two-way analysis-of-variance, if data were not normally distributed. Subsequent Bonferroni correction was performed in order to account for multiple testing influences and Tamhane's-T2 post-hoc testing, respectively, when variances were not equal in Levene's statistics. For intra-patient comparison between VMS datasets repeated ANOVA measures using Dunnett's test were performed. The consensus reading was used for statistical significance assessment between the three DSCT protocols. Sensitivity, specificity and accuracy were calculated on a per-patient base for the 40 keV and 50 keV VMS datasets of both DE CTPA protocols, as well as for the standard CTPA protocol. P-values \< 0.05 were considered statistically significant throughout the entire study. A cohen's kappa statistic was performed in order to assess inter-reader variability of subjective image quality. For all measurements indicating a significant difference a power analysis was performed, indicating a power \>0.64. Results {#Sec10} ======= Patient demographics and dosimetric parameters {#Sec11} ---------------------------------------------- All scans have shown sufficient image quality and were rated as diagnostic. A total of 20 patients showed findings of PE, seven patients in the standard CTPA protocol group, six in the standard DE CTPA protocol group and seven in the low CM DE CTPA protocol group. In each of the CTPA groups, the PE was rated as severe in two patients (both 4%), as these patients showed right ventricular dysfunction on echocardiography. There was no significant difference in sex, age and body-mass-index between all three groups (all p \> 0.05; compare Table [1](#Tab1){ref-type="table"}). As expected by the study assignment, the number of patients with renal impairment (Modification of Diet in Renal Disease (MDRD) estimated glomerular filtration rate (GFR) \>60 ml/min) was lower for both standard CTPA groups compared to the CM DE CTPA group (n = 40/46 and 0, respectively). The radiation dose was significantly higher for both DE CTPA protocols compared to the standard CTPA protocol (p \< 0.05). Patient demographics, serum creatinine levels and radiation dose values are summarized in Table [1](#Tab1){ref-type="table"}. Comparison of objective image quality of the standard CM DE CTPA images, the low CM DE CTPA images and standard CTPA images {#Sec12} --------------------------------------------------------------------------------------------------------------------------- No differences regarding background noise could be observed between the three protocols (standard CTPA 12.1 ± 2.1 HU vs. standard DE CTPA 10.4 ± 1.8 HU vs. low CM DE CTPA 14.3 ± 2.3 HU; for all p \> 0.05). Vessel attenuation (enhancement) and image noise increased with decreasing energy levels for the 40--100 keV VMS datasets (compare Figs [2](#Fig2){ref-type="fig"}--[4](#Fig4){ref-type="fig"}).Figure 273-year-old woman with a peripheral pulmonary embolism (white arrows). Axial slices of main pulmonary arteries of a low contrast media dual-energy CTPA: (**A**) mixed 0.8-weighted virtual polyenergetic spectral (VPS) image, and virtual monoenergetic spectral (VMS) images at a level of 40 keV (**B**), 50 keV (**C**), 60 keV (**D**), 70 keV (**E**), 80 keV (**F**), 90 keV (**G**) and 100 keV (**H**).Figure 3Box- and Whisker-Plots with values for attenuation (**A**,**B**) and contrast-to-noise-ratio (CNR; **C**,**D**) in the main pulmonary arteries (PA, **A**,**C**) and the peripheral PA (**B**,**D**) of a standard CTPA protocol, virtual polyenergetic spectral datasets (VPS) and virtual monochromatic spectral (VMS) datasets of a low contrast media dual-energy CTPA at nine-teen different energy levels.Figure 462-year-old woman with suspected pulmonary embolism. Axial slices of peripheral pulmonary arteries of a low contrast media dual-energy CTPA: (**A**) mixed 0.8-weighted virtual polyenergetic spectral (VPS) image, and virtual monoenergetic spectral (VMS) images at a level of 40 keV (**B**), 50 keV (**C**), 60 keV (**D**), 70 keV (**E**), 80 keV (**F**), 90 keV (**G**) and 100 keV (**H**). The VMS image at 50 keV displays superior subjective image quality when compared to VPS image. The 70 keV VMS dataset for the main pulmonary arteries showed the highest SNR (standard DE CTPA: 23.0 ± 9.1 vs. low CM DE CTPA 17.9 ± 2.8; p \< 0.0001). The 50 keV dataset showed the highest CNR for both DE CTPA protocols, with significantly higher values for the standard DE CTPA protocol (p \< 0.0001). When compared to the standard CTPA protocol datasets, CNR was significantly higher for both DE CTPA protocol VMS datasets at 50 keV (both p \< 0.0073). For the peripheral pulmonary arteries, the 50 keV VMS dataset of the low CM DE CTPA protocol showed the highest SNR and the 70 keV dataset of the standard DE CTPA protocol (standard DE CTPA: 15.6 ± 6.2 and low CM DE CTPA 14.0 ± 6.1). The 40 keV dataset revealed the highest CNR for the low DE CTPA protocol (p \> 0.05) and for the standard DE CTPA protocol both 40 and 50 keV revealed the highest CNR. Similarly, when comparing the standard protocol to the 40 and 50 keV VMS datasets, there was a significant increase in CNR for both DE CTPA protocols (both p \< 0.0064). Table [2](#Tab2){ref-type="table"} summarizes the objective image quality assessment for the standard CTPA protocol, the standard DE CTPA and the low CM DE CTPA protocol datasets.Table 2Objective and subjective image quality between DE CTPA VMS and the standard CTPA datasets.Low CM DECTStandard CM DECTStandard CTPAp-value40 *keV\[A\]*50 *keV\[B\]*40 *keV\[C\]*50 *keV\[D\]\[E\][Main pulmonary arteries]{.ul}[A vs C]{.ul}[B vs D]{.ul}[B vs E]{.ul}[D vs E]{.ul}* Attenuation649.2 ± 59.7437.4 ± 39.5850.2 ± 242.5575.3 ± 161.3375.7 ± 120.3\<0.0001\<0.00010.0022\<0.0001 Signal-to-noise-ratio16.1 ± 2.816.8 ± 2.821.5 ± 9.122.1 ± 9.314.0 ± 3.7\<0.0001\<0.00010.0002\<0.0001 Contrast-to-noise-ratio14.4 ± 2.414.5 ± 2.420.7 ± 9.120.8 ± 9.112.7 ± 3.5\<0.0001\<0.00010.0073\<0.0001 Subjective image quality4\[4--5\]5\[4--5\]4\[4--5\]5\[4--5\]5\[4--5\]0.87620.73350.65810.6632 Subjective image noise4\[3--4\]4\[3--5\]4\[3--4\]4\[4--5\]4\[3--5\]0.85710.76840.80390.1344*[Peripheral pulmonary arteries]{.ul}[A vs C]{.ul}[B vs D]{.ul}[A vs E]{.ul}[C vs E]{.ul}* Attenuation545.4 ± 74.2358.5 ± 48.0857.4 ± 246.0580.1 ± 163.5358.8 ± 110.4\<0.0001\<0.0001\<0.0001\<0.0001 Signal-to-noise-ratio13.8 ± 6.214.0 ± 5.914.6 ± 6.315.1 ± 6.49.7 ± 4.50.40450.3017\<0.0001\<0.0001 Contrast-to-noise-ratio12.1 ± 5.511.7 ± 5.114.1 ± 6.314.1 ± 6.38.8 ± 4.20.14660.05380.0064\<0.0001 Subjective image quality5\[4--5\]5\[4--5\]5\[4--5\]5\[4--5\]5\[4--5\]0.79760.81120.34750.7214 Subjective image noise4\[3--4\]4\[3--5\]4\[3--4\]4\[3--5\]4\[3--4\]0.91170.85490.12240.1463Note VMS = virtual monoenergetic spectral; CTPA = CT pulmonary angiography; keV = kiloelectron Volt; values in brackets represent the 25--75% interquartile. Comparison of subjective image quality {#Sec13} -------------------------------------- There was a substantial agreement between both radiologists for image quality rating (κ = 0.71). No significant difference could be detected in the median image quality or the median image noise for both the 40 keV and 50 keV VMS dataset (all p \> 0.05) when comparing the subjective image quality between both DE CTPA protocol VMS datasets for the main pulmonary arteries and the peripheral pulmonary arteries. Likewise, there was no significant difference in the median image quality or median image noise for both DE CTPA 40 keV and the 50 keV protocols compared to the standard CTPA protocol (all p \> 0.05). Diagnostic accuracy {#Sec14} ------------------- No false positive or false negative findings were revealed considering the 40 keV and 50 keV VMS datasets of the standard DE CTPA, resulting in a sensitivity, specificity and accuracy of 100%. Analogically, considering the 40 keV and 50 keV VMS datasets of the low CM DE CTPA, no false positive or false negative findings were observed, resulting in a sensitivity, specificity and accuracy of 100%. Patient serum creatinine levels before and after the CTPA protocols {#Sec15} ------------------------------------------------------------------- On the 24 hours and 72 hours follow-up, none of the 150 patients suffered from CIN, as serum creatinine levels were similar for the low CM DE CTPA group (mean difference after 72 h −0.02 mg/dl; p \> 0.05). Serum creatinine levels slightly decreased/increased after 72 h for both standard CTPA groups compared to the baseline evaluation (DE standard CTPA: mean difference after 72 h −0.02 mg/dl and standard CTPA mean difference after 72 h + 0.06 mg/dl; both p \> 0.05). In a follow-up serum creatinine evaluation after 3 months, serum creatinine levels again showed no differences to the baseline evaluation for the standard CM dose CTPA protocols (all p \> 0.05) and significantly lower values on 3 month follow-up for the low CM DE CTPA group, mainly because of the normalization of serum creatinine levels of those patients with acute kidney failure due to other reasons. However, due to the small sample size, it has to be noted that the statistical power is not sufficient with π = 0.72. Discussion {#Sec16} ========== The results of this study demonstrate that DE CTPA studies are feasible using only 5.4 g of iodine with a non-inferior image quality when compared to both a standard CTPA and a DE CTPA protocol with a standard CM dose. In addition, no significant increase in serum creatinine levels between baseline and 24--72 hours follow-up within the low CM DE CTPA group was observed, whereas both other groups showed slightly increased serum creatinine levels. However, it is important to point out that none of our patients required temporary hemodialysis during three months of follow-up, which supports the results of the current literature stating that AKI is most likely independent from intravenous CM administration amongst all risk groups^[@CR3]--[@CR6]^. Vessel opacification is mandatory for diagnostic image quality, allowing for diagnostic confidence. In general, an attenuation of 180 HU and higher is regarded as diagnostically sufficient for pulmonary arteries^[@CR23],[@CR24]^. By using a DECT protocol, it is possible to extrapolate images close to a monochromatic x-ray beam of various energy levels. This allows the selection of ideal photon energy levels with regard to the evaluation of vascular compartments^[@CR25]^. Similarly, recent studies have shown improved attenuation and visualization of vascular structures at low kVp levels in a chest CTA^[@CR26]--[@CR29]^. Depending on the DECT technique used, recent studies advocated 60--70 keV and 50--70 keV for an optimal CNR of the pulmonary arteries for dual source and fast kVp switching DE techniques, respectively^[@CR10],[@CR11],[@CR13],[@CR30]^. So far however, studies performed on a DSCT have used a first generation monoenergetic algorithm, which causes a significant increase in image noise for low VMS datasets. This limits diagnostic image quality and confidence, especially in the 40--60 keV VMS datasets^[@CR10],[@CR11],[@CR30]^. In contrast to the above mentioned studies, our study demonstrates the best CNR at 50 keV for the main pulmonary arteries and at 40 keV for the peripheral pulmonary arteries, regardless of the contrast media applied or the generation of DSCT. This is mainly linked to the second generation monoenergetic algorithm with a re-developed noise reduction filter utilizing a frequency split technique, which reduces image noise for the lower VMS datasets^[@CR19]^. VMS datasets at an optimal keV level improve vessel attenuation of pulmonary arteries, allowing for lower volumes of contrast media. Administration of intravenous CM in patients with renal risk factors remains subject to debate in everyday clinical routine, although several recent studies reported that CIN might be independent from the administration of intravenous CM amongst different risk groups^[@CR3]--[@CR6]^. However, CIN has accounted for up to 12% of all hospital-acquired acute renal failures in the past and has been associated with a considerable mortality and morbidity rate^[@CR31]^. So far, data and knowledge for patients undergoing several CT examinations within several days, as well as a potential dose dependency between intravenous CM and AKI remains insufficient. Therefore, lowering the amount of contrast media is recommended by current guidelines^[@CR2]^ and one of the most effective methods of reducing potential contrast media induced cytotoxic effects. Using the introduced novel protocol in this study might be a valid option for high-risk patients, for whom clinicians would order a contrast-enhanced CT more conservatively^[@CR31]^. Several studies have evaluated low contrast media CTPA protocols with iodine loads ranging between 9--30 g^[@CR9],[@CR11],[@CR13],[@CR30],[@CR32]^. However, most of these studies have used a test bolus in order to achieve a low contrast media protocol. This is an unnecessary contrast media application without diagnostic information, thus exposing patients with renal risk factors to additional contrast dose. Furthermore, the above mentioned studies are not entirely explicit whether the iodine amounts included the iodine load of the test bolus, which ranges between 3.7 g and 7 g. In contrast, the low CM DE CTPA protocol used in this study, utilizing a bolus tracking protocol, has only required a total iodine load of 5.4 g to achieve sufficient vessel opacification. This amount is substantially lower compared to previous investigations. None of our patients suffered from a CIN 24 hours and 72 hours after CT examination, although 60% of our patient cohort had known acute or chronic renal insufficiency. The effective dose of 3.4 mSv for the low CM DECT CTPA protocol in this study was significantly higher than the 2.5 mSv of the standard CTPA. However, patients with acute or chronic renal failure may benefit from this protocol, as long as CIN due to intravenously administered CM is not completely ruled out. Therefore, the benefits of a reduced CM amount have to be outweighed against the risk for radiation-induced cancer. The attributed 5-year survival rate of patients with arteriovenous fistulas/ arteriovenous grafts due to end stage renal impairment is 36%^[@CR33]^. The relative cancer risk lies between 1.001 and 1.04 for radiation doses of \<100 mSv with an expected onset 15 to 20 years after exposure^[@CR34]^. The effective dose of 3.4 mSv for the low CM DE CTPA protocol was lower than that reported in previous DECT CTPA protocol investigations, which ranged between 4.6--7.0 mSv^[@CR11],[@CR13],[@CR30]^, and the standard DE CTPA protocol used in this study, which had an effective dose of 3.9 mSv. This is mainly linked to using a 3^rd^ generation DSCT system. This system combines a more efficient detector design with automated tube modulation selection, as well as a 3^rd^ generation IR technique to reduce the applied radiation dose. In the near future, when photon-counting detector CT systems become clinically available, both a reduced radiation exposure and a reduced amount of CM may become feasible by k-edge subtraction imaging while maintaining high image quality. As all patients of the DE CTPA protocol had acute or chronic renal failure, the trade-off between radiation and contrast media amount shifts toward the lowest possible total iodine dose, rather than the lowest possible radiation dose. Nonetheless, each DECT examination should comply with the as low as reasonably achievable principle with reasonable radiation dose. Several study limitations have to be noted. First of all, our primary focus in this study was to evaluate objective and subjective image criteria of a low iodine CM CTPA protocol. Therefore, a control scan with a standard CM amount in order to determine false positive or false negative findings was not performed due to ethical considerations regarding total amounts of CM and additional radiation dose. We counterbalanced this with a strict standard of reference. Furthermore, the sufficient vessel attenuation of 180 HU and above was achieved, as well as increased CNR values of the 40 and 50 keV VMS datasets compared to the standard CTPA protocol. This allowed a decision-making with sufficient diagnostic confidence for confirming or ruling out a PE. Secondly, we did not evaluate or document any specific cardiac output parameters, which may have a considerable effect on vascular enhancement in CTPA^[@CR35]^. Thirdly, there is a selection bias in our study, as we primarily included patients with acute or chronic renal failure in the low CM DE CTPA group, as there was only a low number of patients with renal risk factors in the other two groups. This might confound our exploratory assessment of CIN. Furthermore, it was reported that patients with acute and chronic renal failure have a higher risk of venous thromboembolism^[@CR36]^, which may be one reason for the relative high incidence of PE in this study group. Finally, the CTPA scans were performed on different CT scanners with different scan settings, which might influence image quality. However, the purpose of this study was to demonstrate the implementation of a low CM protocol, which results in similar CNR as a standard CTPA with a standard CM protocol. In conclusion, we have demonstrated that a DE CTPA protocol performs equally to a standard CTPA protocol and can be used to save contrast media when using VMS reconstructions of low energy datasets without a reduction in diagnostic accuracy, vessel opacification, CNR or perceived image quality. VMS datasets of 40--50 keV are ideal when reducing the total iodine amount down to 5.4 g (reduction of 83%). However, the acquisition using a DE CTPA protocol comes at the disadvantage of an increased radiation exposure of 26% compared to a reduced tube voltage standard CTPA protocol of 100 kVp. Therefore, the benefits of a reduced CM amount have to be outweighed against the risk for radiation-induced cancer. **Publisher\'s note:** Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. This research project is part of the Research Campus M²OLIE and funded by the German Federal Ministry of Education and Research (BMBF) within the Framework "Forschungscampus: public-private partnership for Innovations" under the funding code 13GW00926. T.H. and M.M. received funding from the German Federal Ministry of Education and Research (BMBF) within the Framework "Forschungscampus: public-private partnership for Innovations" under the funding code 13GW00926. All other authors have nothing to disclose. M.M.; T.H. Study design: M.M.; T.H.; B.S.; H.H.; C.L. Data acquisition: M.M.; T.H.; H.H. Quality control of data and algorithms: M.M.; T.H. Data analysis and interpretation:M.M.; T.H.; B.S.; S.O.S.; C.L.; C.S. M.M.; T.H. Manuscript preparation: M.M.; T.H.; B.S.; S.O.S.; C.L. H.H. Manuscript editing: M.M.; T.H.; B.S.; S.O.S.; C.L. H.H.; C.S. Manuscript review: M.M.; T.H.; B.S.; S.O.S.; C.L. H.H.; C.S. Ethical Approval for Research: Yes. Competing Interests {#FPar1} =================== C.L. and B.S. are employees of Siemens Healthcare, Forchheim, Germany. All other authors have no conflict of interest to disclose. The authors not employed for Siemens Healthcare were in control of this study.

1. Introduction {#sec1-ijms-20-05323} =============== Ischemic heart disease remains a major worldwide health concern. The identification of ischemic preconditioning (IPC) as a cardioprotective mechanism has raised the possibility that stimulating the endogenous pathways may protect the heart from ischemic injury \[[@B1-ijms-20-05323]\]. It has been suggested that the maintenance of mitochondrial integrity is crucial in the preconditioning of heart \[[@B2-ijms-20-05323]\] and brain tissue \[[@B3-ijms-20-05323]\]. A large body of evidence has implicated mitochondrial ion channels as significant players in these phenomena \[[@B4-ijms-20-05323],[@B5-ijms-20-05323],[@B6-ijms-20-05323]\]; however the precise role of these channels is still under active investigation \[[@B7-ijms-20-05323]\]. The ATP-sensitive potassium channel (mitoK~ATP~) opening protects cardiac and neuronal tissue against ischemia-reperfusion injury \[[@B8-ijms-20-05323],[@B9-ijms-20-05323]\]. Considerable progress has been made in understanding the signaling mechanisms of cardioprotection \[[@B10-ijms-20-05323],[@B11-ijms-20-05323]\], but understanding the role of mitoK~ATP~ in cardioprotection is still incomplete. In 1991, mitoK~ATP~ was first identified using an electrophysiological approach in rat liver mitochondria \[[@B12-ijms-20-05323]\]. mitoK~ATP~ was later found in other mammalian tissues including the heart \[[@B13-ijms-20-05323],[@B14-ijms-20-05323],[@B15-ijms-20-05323]\], brain \[[@B16-ijms-20-05323],[@B17-ijms-20-05323]\], skeletal muscle \[[@B18-ijms-20-05323]\], skin fibroblasts \[[@B19-ijms-20-05323]\] and renal tissue \[[@B20-ijms-20-05323]\]. The presence of the channel has also been reported in non-mammalians including *Crithidia fasciculate* euglenozoa \[[@B21-ijms-20-05323]\], *Caenorhabditis elegans* nematode \[[@B22-ijms-20-05323]\], *Acanthamoeba castellanii* amoebae \[[@B23-ijms-20-05323]\], and in plants, including *Triticum turgidum* wheat and *Pisum sativum* peas \[[@B24-ijms-20-05323]\]. Until recently, progress in identifying its molecular composition has been limited, despite efforts by many laboratories \[[@B7-ijms-20-05323]\]. MitoK~ATP~ has been determined to be functionally and molecularly distinct from sarcolemmal K~ATP~ channels (sarcK~ATP~), based on the finding that diazoxide was 1800 times more potent in opening mitoK~ATP~ than sarcK~ATP~ \[[@B15-ijms-20-05323]\]. This finding was confirmed by Sato et al. \[[@B5-ijms-20-05323]\], who also described the distinguishing effects of HMR1098 and P1075 on the two channels. Initial studies suggested that the inward rectifying K^+^ channel subunits Kir6.1 or Kir6.2 are pore-forming units of mitoK~ATP~, but later studies did not confirm this hypothesis \[[@B25-ijms-20-05323],[@B26-ijms-20-05323],[@B27-ijms-20-05323]\]. Recently, a proteomic screen, together with pharmacological and genetic manipulation, led Foster et al. \[[@B28-ijms-20-05323]\] to conclude that a splice variant of the renal outer medullary potassium channel (ROMK) is a pore-forming unit of cardiac mitoK~ATP~. In this study, we used mitoplasts from rat heart-derived H9c2 cell lines (WT and overexpressing human ROMK2 isoform of the channel) to study the electrophysiological and pharmacological properties of the mitoK~ATP~. Additionally, we performed experiments in order to clarify the structural topology of the channel formed by the ROMK2 in the inner mitochondrial membrane. 2. Results {#sec2-ijms-20-05323} ========== 2.1. Ion Channels Observed in H9c2 Mitochondria {#sec2dot1-ijms-20-05323} ----------------------------------------------- The main aim of our study was to describe the mitoK~ATP~ channel present in heart mitochondria and verify our hypothesis suggesting the ROMK protein as a pore forming subunit of this channel. Therefore, we performed a series of electrophysiological experiments with use of isolated mitoplasts ([Figure 1](#ijms-20-05323-f001){ref-type="fig"}A). In our study, we used mitoplasts isolated from both the wild type H9c2 and H9c2 cells overexpressing human ROMK2 (H9c2 OE ROMK2) \[[@B28-ijms-20-05323]\] (for a detailed description of the cell lines, see the Materials and Methods section). The procedure of mitoplast selection was verified by the preparation of a PCR reaction with use of primers which could recognize mitochondrial and nuclear DNA ([Figure S1, Tables S1 and S2](#app1-ijms-20-05323){ref-type="app"}). We found that all tested vesicles contained mitochondrial but not nuclear DNA. The channel activity from inner mitochondrial membrane of the cell lines was measured in patch-clamp experiments described in the Materials and Methods section. Performed experiments revealed the existence of few ion channel types in the mitochondria of both cell lines. Based on conductance and kinetics, we classified the observed channels into three groups: (a) large-conductance, calcium-regulated potassium (BK~Ca~-type) channel with a conductance between 250 and 300 pS and brief openings at negative voltage; (b) permeability transition pore (PTP)-type channel with a conductance between 500 and 1000 pS and openings with many sub-states (probably mitochondrial permeability transition pore) ([Figure S2](#app1-ijms-20-05323){ref-type="app"}); (c) K~ATP~-type with conductance of around 90--100 pS and typical middle openings and closings of the channel ([Figure 1](#ijms-20-05323-f001){ref-type="fig"}B). Based on this classification we found that the number of observed K~ATP~-type channels in mitochondria from H9c2 OE ROMK2 was significantly higher (*p* \< 0.05) than in the WT cells. The chi-squared statistic was 3.8431 (based on 50 recordings of all channel types for each cell line). A direct comparison of mitoK~ATP~ recordings from the H9c2 WT and H9c2 OE ROMK2 cells revealed the existence of channels with very similar kinetics in the mitochondria of both cell lines ([Figure 1](#ijms-20-05323-f001){ref-type="fig"}C). 2.2. Identification of the mitoK~ATP~ Channel in Mitoplasts of ROMK Overexpressing H9c2 {#sec2dot2-ijms-20-05323} --------------------------------------------------------------------------------------- A very low frequency of mitoK~ATP~ observations in the H9c2 WT cells stopped us from performing full characteristics of the channel. Initially, we analyzed the current--voltage dependence of both channels. [Figure 2](#ijms-20-05323-f002){ref-type="fig"}A shows the single-channel recordings at different voltages of the channel from the H9c2 OE ROMK2 cells. Based on the current--voltage relationship, the calculated conductance of the channel from the H9c2 OE ROMK2 cells was equal to 94 ± 3 pS in symmetric 150/150 mM KCl ([Figure 2](#ijms-20-05323-f002){ref-type="fig"}B). The conductance of the channel present in mitochondria of the H9c2 WT cells was equal to 97 ± 2 in the same experimental condition. In both cases, rectification was not observed. What is important is that both channels showed an almost identical voltage--current correlation, directly suggesting the same properties and origin of both activities. [Figure 2](#ijms-20-05323-f002){ref-type="fig"}C--E show the biophysical parameters for the channel activity from the H9c2 OE ROMK2 cells. The open probability (P~o~) increased from 0.21 ± 0.11 at 50 mV to 0.57 ± 0.14 at −50 mV. We observed a voltage dependence on the mean channel opening time (τ~open~), which decreased from 19.2 ± 6.3 ms at −50 mV to 5.1 ± 1.3 ms at 50 mV. The mean mitoK~ATP~ channel closed time (τ~closed~) increased from 38.9 ± 10.2 ms at --50 mV to 45.3 ± 6.2 ms at 50 mV. 2.3. Specificity of the mitoK~ATP~ Channel Using the Patch-Clamp Technique {#sec2dot3-ijms-20-05323} -------------------------------------------------------------------------- To characterize the specificity of the channel, the ionic current was measured in an asymmetric 50/150 mM KCl solution. In these experiments, the activity of the mitoK~ATP~ channel from the H9c2 OE ROMK2 cells was measured. [Figure 2](#ijms-20-05323-f002){ref-type="fig"}B shows the current-voltage relationship in the asymmetric 50/150 mM KCl solution. The permeability ratios for ions were calculated according to the Goldman--Hodgkin--Katz voltage equation:$$E_{rev} = \frac{RT}{zF}\ln\left( \frac{p_{K}\left\lbrack K \right\rbrack_{o} + p_{Cl}\left\lbrack {Cl} \right\rbrack_{in}}{p_{K}\left\lbrack K \right\rbrack_{in} + p_{Cl}\left\lbrack {Cl} \right\rbrack_{o}} \right)$$ where E*~rev~* is the reverse potential, R is the gas constant, T is the temperature in Kelvin, z is equal to 1, F is the Faraday constant, p is the permeability of a given ion, and \[Cl\] and \[K\] are the respective concentrations of the ion outside (o) (i.e., in the pipette) and inside (in) (i.e., in the chamber). All measurements were carried out at room temperature (25 °C). This result showed that the observed channel was highly specific for cations. The calculated permeability ratio for K^+^ and Cl^−^ was equal to 46.4 according to the Goldman--Hodgkin--Katz voltage equation. The calculated reverse potential of the channel was --26.8 ± 2.4 mV. The asymmetric solution did not significantly affect channel parameters such as conductance ([Figure 2](#ijms-20-05323-f002){ref-type="fig"}B). 2.4. Pharmacological Properties of mitoK~ATP~ Using the Patch-Clamp Technique {#sec2dot4-ijms-20-05323} ----------------------------------------------------------------------------- Next, we tested the typical modulators of the mitoK~ATP~ and ROMK channels. [Figure 3](#ijms-20-05323-f003){ref-type="fig"} illustrates the activity of the channels detected in the inner mitochondrial membrane of the H9c2 OE ROMK2 and H9c2 WT cells under the control conditions and after the application of modulators. All modulators were applied to the matrix site. First, we tested 5-hydroxydecanoic acid (5-HD), a known mitoK~ATP~ inhibitor ([Figure 3](#ijms-20-05323-f003){ref-type="fig"}A). In both cases, the application of 100 µM of 5-HD resulted in the channel inhibition, as is shown in the original recordings ([Figure 3](#ijms-20-05323-f003){ref-type="fig"}A, left panel). In case of the channel from the H9c2 OE ROMK2 cells, an open probability was calculated from three independent experiments. The application of the inhibitor resulted in a decrease of P~o~ from 0.57 ± 0.14 to 0.040 ± 0.014 at --50 mV (*n* = 3). Similarly, the application of Tertiapin Q (TPNQ), which blocks ROMK channels, also resulted in the inhibition of both channels ([Figure 3](#ijms-20-05323-f003){ref-type="fig"}B, left panel). Similarly to that of 5-HD, the application of TPNQ resulted in a statistically significant decrease of P~o~ (to 0.05 ± 0.02 after inhibitor addition, *n* = 3). Apart from the above, we measured the activity of mitoK~ATP~ from the H9c2 cells overexpressing ROMK2 after the application of 1 mM of Mg^2+^ and 500 µM of ATP ([Figure 3](#ijms-20-05323-f003){ref-type="fig"}C). We found that ATP/Mg^2+^ inhibited the channel activity, and P~o~ decreased to 0.1 ± 0.07 at --50 mV. The addition of 30 µM of diazoxide recovered channel activity, and P~o~ increased to 0.48 ± 0.12 at --50 mV (*n* = 3). The channel was also partially inhibited by 50 µM of glibenclamide ([Figure 3](#ijms-20-05323-f003){ref-type="fig"}D). The channel P~o~ decreased from 0.57 ± 0.14 to 0.27 ± 0.09 at --50 mV (*n* = 3). Taken together, the measured single-channel conductance, voltage--dependent regulation, and sensitivity to selective agonists and antagonists strongly indicate that our recordings corresponded to a mitochondrial ATP-sensitive channel as well as a renal outer medullary potassium channel. Based on available data, we conclude that channels from both cells lines classified as mitoK~ATP~ showed very similar biophysical and pharmacological properties. Moreover, we were able to observe mitoK~ATP~ more frequently in the cells overexpressing the ROMK2 protein. These results strongly support the hypothesis that the ROMK protein is the pore-forming subunit of mitoK~ATP~. 2.5. Properties of the mitoBK~Ca~ Channel from H9c2 OE ROMK2 Cells {#sec2dot5-ijms-20-05323} ------------------------------------------------------------------ Apart from the mitoK~ATP~ in both cell lines, we were able to detect large-conductance Ca^2+^-regulated channel with properties corresponding to well-known mitoBK~Ca~ channels ([Figure 1](#ijms-20-05323-f001){ref-type="fig"}B and [Figure 4](#ijms-20-05323-f004){ref-type="fig"}). The single-channel recordings at different voltages revealed the voltage sensitivity of the channel from the H9c2 OE ROMK2 cells ([Figure 4](#ijms-20-05323-f004){ref-type="fig"}A,C). The activity of the channel manifested by P~o~ was significantly reduced in the negative voltages. Based on the mean of the current-voltage relationship, the calculated conductance of the channel was equal to 256 pS in symmetric 150/150 mM KCl ([Figure 4](#ijms-20-05323-f004){ref-type="fig"}B). Additionally, the channel showed a clear sensitivity to Ca^2+^ ions ([Figure 4](#ijms-20-05323-f004){ref-type="fig"}D). The channel open probability was significantly lower after perfusion with a medium containing a low concentration of Ca^2+^. Based on the presented data, we conclude that properties of this channel correspond to previously described mitoBK~Ca~ channels. Our electrophysiological data are in line with a biochemical analysis that showed the presence of the BK~Ca~ channel subunits in mitochondria of the H9c2 cells \[[@B29-ijms-20-05323]\]. 2.6. Localization of the ROMK protein after Transient Expression in H9c2 Cells {#sec2dot6-ijms-20-05323} ------------------------------------------------------------------------------ The above electrophysiological data suggest presence of the same mitoK~ATP~ in the mitochondria of both the WT and ROMK2 OE H9c2 cells. To verify our electrophysiological data, we checked the localization of the ROMK2 after the transient transfection of the wild type H9c2 cells. Indeed, the transient transfection of ROMK2 fused with GFP resulted in a clear GFP signal overlap with mitochondrial staining ([Figure 5](#ijms-20-05323-f005){ref-type="fig"}). Furthermore, a part of the ROMK2-GFP fraction was located outside the mitochondria in other cellular fractions most likely as a result of overexpression. The above data confirm the targeting of the ROMK protein to the mitochondria of H9c2 WT and support functional studies previously performed in these cells \[[@B28-ijms-20-05323]\]. 2.7. Localization in the Inner Mitochondrial Membrane and Possible Mitochondrial Topology of ROMK Protein in H9c2 OE ROMK2 Cells {#sec2dot7-ijms-20-05323} -------------------------------------------------------------------------------------------------------------------------------- Finally, we decided to confirm the presence of the ROMK protein in the inner mitochondrial membrane using a Proteinase K (PK) treatment assay ([Figure 6](#ijms-20-05323-f006){ref-type="fig"}). In these experiments, the mitochondrial fraction from the H9c2 OE ROMK2 cells was used. A western blot analysis showed that ROMK was protected against PK digestion in the mitochondrial samples incubated in an isotonic solution ("--swelling", "+PK"). In these conditions, the mitochondrial outer membrane protects the proteins of the intermembrane space and inner mitochondrial membrane against proteinase K. The parallel staining of the translocase of outer membrane (TOM) complex receptor (Tom20), which is localized in the outer mitochondrial membrane, showed a complete degradation of the protein. In the same fraction, both Tim23 (a control protein with known localization in the mitochondrial inner membrane) and Hsp75 (matrix) were also protected. The mitochondrial swelling and subsequent PK treatment ("+swelling", "+PK") resulted in the degradation of both Tom20 and Tim23. On the other hand, both ROMK and Hsp75 remained untreated after the addition of proteinase. This result suggested the limited access of PK to the epitope of ROMK localized in the C-terminal part of the protein even after mitochondrial outer membrane disruption. However, we could observe the partial degradation of ROMK after the solubilization of mitochondria with 0.1% TritonX-100 and subsequent PK treatment ([Figure 6](#ijms-20-05323-f006){ref-type="fig"}B). This result suggests that the observed signal came from a protein localized to the inner mitochondrial membrane. Moreover, since ROMK2 contains two transmembrane segments, our data might suggest that both the N- and C-termini of the protein might be localized in the mitochondrial matrix. 3. Discussion {#sec3-ijms-20-05323} ============= In our study, we used mitoplast patch-clamping to detect mitoK~ATP~ in H9C2 cells. Previous observations concerning mitoK~ATP~ were based on measurements of K^+^ ion flux across the inner mitochondrial membrane or across the liposomal membrane following the partial purification and reconstitution of mitoK~ATP~. These studies clearly revealed that the mitoK~ATP~ opening leads to mitochondrial depolarization, an increase of oxygen consumption and mitochondrial swelling \[[@B8-ijms-20-05323]\]. The single channel properties of the mitoK~ATP~ of liver, lymphocytes, heart or brain mitochondria have been successfully studied by reconstituting the mitochondrial inner membrane into a planar lipid bilayer \[[@B13-ijms-20-05323],[@B14-ijms-20-05323],[@B30-ijms-20-05323],[@B31-ijms-20-05323],[@B32-ijms-20-05323]\] or with a patch-clamp technique \[[@B12-ijms-20-05323],[@B19-ijms-20-05323],[@B33-ijms-20-05323]\]. Various proteins, including Kir6.1 and Kir6.2, have been proposed to form the channel \[[@B26-ijms-20-05323],[@B27-ijms-20-05323]\]; however, a recent study provided strong evidence that the mitoK~ATP~ in cardiac cells is formed by the ROMK protein \[[@B28-ijms-20-05323]\]. This was supported by a report showing that ROMK-forming mitoK~ATP~ also exists in skin fibroblasts \[[@B19-ijms-20-05323]\]. In the present work, the measurements of ion flows across the inner membrane of mitochondria isolated from the cardiac H9c2 cells revealed the presence of mitoK~ATP~ formed by the ROMK protein. Reported channels have a conductance less than 100 pS, which is close to previous observations \[[@B19-ijms-20-05323],[@B30-ijms-20-05323],[@B33-ijms-20-05323],[@B34-ijms-20-05323]\] but much higher than the values in the 10--30 pS range observed by Mironova and coworkers \[[@B14-ijms-20-05323],[@B32-ijms-20-05323]\]. However, mitoK~ATP~ conductance has often been found to vary. For example, a study describing the electrophysiological properties of the reconstituted human cardiac mitoK~ATP~ reported maximum conductances from \~20 to \~60 pS and \>80 pS \[[@B35-ijms-20-05323]\]. The observed conductance of mitoK~ATP~ in human lymphocytes did not exceed 82 pS \[[@B33-ijms-20-05323]\]. ROMK channels expressed in plasma membrane exhibited a conductance of 30--35 pS \[[@B36-ijms-20-05323]\]. However, the ROMK protein is also part of the thick ascending limb potassium secretory channel, which exhibits ROMK-like characteristics and has a conductance of 70 pS \[[@B37-ijms-20-05323]\]. Differences of conductance between plasma membrane and mitochondrial channels can be result of various factors, including the presence of protein partners, posttranslational modifications of the pore forming unit, differences in the lipid composition of mitochondrial inner membrane, and experimental protocol. A similar difference of conductance has also been observed in the case of the BK~Ca~ and mitoBK~Ca~ channels; moreover, the conductance of mitoBK~Ca~ from the same tissue can vary \[[@B38-ijms-20-05323]\]. The mitoK~ATP~ channels in this study did not show rectification under all conditions, which is similar to what was found with the channel from skin fibroblasts \[[@B19-ijms-20-05323]\]. MitoK~ATP~ from bovine heart reconstituted into black lipid membrane (BLM) showed weak rectification but only in a non-symmetrical solution (50/150 mM of KCl) \[[@B31-ijms-20-05323]\]. By contrast, mitoK~ATP~ from lymphocytes recorded with patch-clamp showed rectification in negative voltages \[[@B33-ijms-20-05323]\]. The other biophysical properties of the observed channels match data described in the literature: the decreased opening probability when positive voltage is applied \[[@B19-ijms-20-05323],[@B31-ijms-20-05323]\] and open/close dwell time \[[@B35-ijms-20-05323]\]. The identification of the observed channel as mitoK~ATP~ was also based on its pharmacological properties. The channel recorded in both cell lines was inhibited by 5-HD, which is a key mitoK~ATP~ inhibitor \[[@B8-ijms-20-05323]\]. This assertion is based on the functional measurements of the cardiac channel in cardiomyocytes \[[@B39-ijms-20-05323]\], in a perfused heart \[[@B9-ijms-20-05323]\] and isolated mitochondria \[[@B40-ijms-20-05323]\]. Tertiapin-Q is a known inhibitor of ROMK-type channels \[[@B41-ijms-20-05323],[@B42-ijms-20-05323],[@B43-ijms-20-05323]\] and was shown to inhibit mitoK~ATP~ by Foster et al. \[[@B28-ijms-20-05323]\]. TPNQ was also shown to inhibit the BK~Ca~-type of channels \[[@B43-ijms-20-05323],[@B44-ijms-20-05323]\]; however, the mitoBK~Ca~ channels have significantly different biophysical and pharmacological properties than mitoK~ATP~, as was also reported in this study. What is important is that the mitoBK~Ca~ channel has previously been found in cardiac mitochondria. It has also been shown that activation of the channel protects the heart muscle against injury induced by ischemia/reperfusion \[[@B45-ijms-20-05323],[@B46-ijms-20-05323],[@B47-ijms-20-05323]\]. The presence of both channels in one cell type raises questions about the interplay of both channels in the cytoprotection mechanism and should be studied in future. A third channel inhibited by TPNQ is the G protein-coupled, inwardly-rectifying potassium channel (GIRK-type, Kir3.x) \[[@B43-ijms-20-05323],[@B48-ijms-20-05323]\]. However, Mg^2+^/ATP did not inhibit the GIRK channels, and ATP increased their activity \[[@B49-ijms-20-05323],[@B50-ijms-20-05323]\]. Because the ROMKs belong to the ATP-inhibited channels \[[@B51-ijms-20-05323]\], we conclude that TPNQ inhibited the ROMK channel. Additionally, we show that the overexpression of ROMK2 in the H9c2 cells resulted in increase of the mitoK~ATP~ channel recordings with the same kinetics and basic biophysical properties as the H9c2 WT cells. Moreover, the transient transfection of the H9c2 cells with ROMK2-GFP revealed that the channel localized to mitochondria. This observation agrees with previous study showing the localization of the ROMK protein to the cristae in mitochondria from both the heart and the liver \[[@B52-ijms-20-05323]\]. Silencing the ROMK protein resulted in a decrease of ATP-, 5-HD-, and TPNQ-sensitive mitochondrial swelling and thallium uptake in the H9c2 cells \[[@B28-ijms-20-05323]\]. All these findings supports the ROMK protein as a potential component of mitoK~ATP~ in cardiac cells. However, it has been suggested that besides ROMK, there may be another protein that can form this type of channel in mitochondria \[[@B53-ijms-20-05323]\]. This idea is supported by different sensitivity to glibenclamide in both cases. The channel described here was fully inhibited by 5-HD and Mg^2+^/ATP, but it was only partially inhibited by glibenclamide. Nevertheless, the partial but statistically significant inhibition of observed mitoK~ATP~ activity by the antidiabetic sulfonylurea glibenclamide may indeed suggest the presence of a mitochondrial inner membrane sulfonylurea receptor (mitoSUR) in addition to the ROMK2 protein. Previous equilibrium binding studies have revealed a single class of low affinity binding sites for sulfonylurea in the inner mitochondrial membrane from beef hearts \[[@B54-ijms-20-05323]\]. Bajgar et al. \[[@B16-ijms-20-05323]\] identified a 63 kD protein labeled with high affinity by BPDIPY-FL-gliburide and proposed that this may be mitoSUR. Several studies have demonstrated that ROMK2 forms a glibenclamide-sensitive K^+^ channel when co-expressed with SUR2B. Moreover, when ROMK2 is expressed alone in *Xenopus laevis* oocytes, it is not sensitive to glibenclamide \[[@B55-ijms-20-05323]\]. Sulfonylurea receptor (SUR2A) subunits have been demonstrated in ventricular myocyte mitochondria \[[@B56-ijms-20-05323]\], but other studies have not supported a role for SUR2B as a component of ROMK2 and have indicated that there is no requirement for an additional subunit to confer glibenclamide sensitivity to ROMK2 \[[@B57-ijms-20-05323]\]. Other studies have suggested that the cystic fibrosis transmembrane conductance regulator (CFTR) may confer ROMK sensitivity to glibenclamide; however, the presence of CFTR in the inner mitochondrial membrane remains unclear \[[@B58-ijms-20-05323]\]. The presence of mitoSUR in the H9c2 cell line has not been clearly demonstrated, and further studies are required to determine the protein partners of ROMK2. The topology of mitoK~ATP~ in the inner membrane of mitochondria is an important issue. Since the ROMK protein contains two transmembrane domains, the C- and N-termini must be located on the same side of the inner mitochondrial membrane. Inoue et al., \[[@B12-ijms-20-05323]\] concluded that Mg^2+^/ATP inhibits from the matrix. Yarov-Yarovoy et al. \[[@B59-ijms-20-05323]\] concluded that Mg^2+^/ATP inhibits from the cytosolic side of the inner membrane. The measurements of K^+^ fluxes across the inner mitochondrial membrane supported this idea and suggested that the ATP binding domain faces the cytosol (intermembrane space), because the channel inhibited by palmitoyl coenzyme A can be restored by GTP added to an external medium \[[@B59-ijms-20-05323]\]. However, a GDP/GTP transport system in the inner mitochondria membrane has been identified \[[@B60-ijms-20-05323],[@B61-ijms-20-05323]\]. Our mitoplast patch-clamp recordings were performed in an inside--out configuration. This implies that during the channel recording, Mg^2+^/ATP had access to the matrix side of the channel. On the other hand, our recordings revealed the inhibition of the channel by Tertiapin Q. It is known that TPNQ binds to the vestibule of the plasma membrane ROMK channel, which faces the extracellular environment and, thus, the opposite site of the membrane to the nucleotide binding domain \[[@B41-ijms-20-05323],[@B42-ijms-20-05323]\]. Our single channel recording with the application of Tertiapin Q could suggest that this peptide also inhibits the channel from the matrix site. Alternatively, it is possible that the toxin could be partially permeable to the mitochondrial membrane, which would allow it to block the pore in a more conventional way. A similar effect of the unexpected inhibition of the mitoBK~Ca~ channel by iberiotoxin (IbTx) was previously observed \[[@B62-ijms-20-05323],[@B63-ijms-20-05323]\]. In case of mitoBK~Ca~ from glial cells, the dual orientation of the channels in the patch-clamp pipette was observed \[[@B64-ijms-20-05323]\]. It is also important that the previous study showed that Tertiapin application from the matrix site did not inhibit the pH-sensitive channel in brain mitochondria. Moreover, this channel was also insensitive to ATP/Mg^2+^ and 5-HD \[[@B65-ijms-20-05323]\]. This suggests that the TPNQ effect observed here is not coincidental. Because the patch-clamp data were unclear, we decided to examine the topology of the channel with a biochemical approach. Our results based on PK accessibility to the ROMK protein suggest that the both N- and C-terminus are located in the mitochondrial matrix ([Figure 5](#ijms-20-05323-f005){ref-type="fig"}). In previous study published by Foster et al., Tertiapin Q added to the mitochondrial suspension was able to inhibit thallium uptake, thus suggesting the location of the inhibitor binding site facing the mitochondrial intermembrane space (IMS) \[[@B28-ijms-20-05323]\]. The presence of the targeting sequence at the N-terminus of the protein \[[@B28-ijms-20-05323]\] suggests that ROMK might be imported to the inner mitochondrial membrane via TOM and then TIM23 (translocase of the inner membrane) \[[@B66-ijms-20-05323],[@B67-ijms-20-05323]\], which might support the topology of the channel with the C- and N-terminus on the matrix side. Interestingly, in the study describing the small conductance calcium-activated potassium channels (SK2) channel in mitochondria of HT22 cells, the PK treatment of swollen mitochondria gave the same result---the channel protein was digested only after mitochondria solubilization with Triton X-100 \[[@B68-ijms-20-05323]\]. However, we propose that the topology of the channel is the opposite (with the vestibule facing mitochondrial matrix), and this conclusion is based on an electrophysiological experiment with whole-mitoplast patch-clamp. It is clear that the important question of mitoK~ATP~ topology remains controversial, and additional experiments are required to resolve this issue. Though our study was limited and based on the cardiomyoblast cell line application of a unique electrophysiological technique, it provides good evidence for ROMK as a possible component of the mitoK~ATP~. However, cardioprotective mechanisms activated by K~ATP~ channel openers are complex phenomena. Previous data have suggested an interplay between the plasma membrane and the sarcoplasmic and mitochondrial K~ATP~ channels \[[@B69-ijms-20-05323]\]. Moreover, the effects of K~ATP~ channel openers seem to be tissue dependent due to the differential expression of the components of the K~ATP~ channels \[[@B70-ijms-20-05323]\]. Therefore, additional studies based on more complex models, including primary cardiomycytes are required in order to achieve further benefits from our discovery. 4. Materials and Methods {#sec4-ijms-20-05323} ======================== 4.1. H9c2 WT and H9c2 OE ROMK2 Cells and Isolation of Mitochondria {#sec4dot1-ijms-20-05323} ------------------------------------------------------------------ The heart-derived H9c2 WT cell line was purchased from ATCC. The H9c2 cell line which stably overexpressed the human ROMK2 protein with the V5 tag on the C-terminus (H9c2 OE ROMK2) was the cell line previously described by Foster and colleagues \[[@B28-ijms-20-05323]\]. The H9c2 OE ROMK2 cell line was developed using gene encoding for human KCNJ1 (I.M.A.G.E. consortium, clone ID 30915211) \[[@B28-ijms-20-05323]\]. Cells were cultured in standard 75 cm^2^ cell culture flasks (Thermo Fisher Scientific, Waltham, MA, USA) in a DMEM medium (Lonza) supplemented with 10% FBS, 2 mM of L-glutamine, 100 U/mL of penicillin, and 100 mg/mL of streptomycin at 37 °C in a cell incubator with a humidified atmosphere with 5% CO~2~. The medium for the ROMK OE cells was additionally supplemented with blasticidin. The cells were fed and re-seeded twice a week. For our study, we used cells from passages 10 to 30. Mitochondria were prepared as previously described \[[@B63-ijms-20-05323]\]. In brief, cells were collected in a PBS medium and centrifuged at 800× *g* for 10 min. The cell pellet was resuspended and homogenized in a preparation solution (250 mM of sucrose, 5 mM of HEPES, pH = 7.2). To isolate the mitochondria, the homogenate was centrifuged at 9200× *g* (10 min). The pellet was then suspended and centrifuged at 790× *g* (10 min). The supernatant was transferred to a new tube and centrifuged at 9200× *g* for 10 min. The pelleted mitochondria were then resuspended in an isotonic solution (150 mM of KCl, 10 mM of HEPES, pH = 7.2). All procedures were performed at 4 °C in a cooling centrifuge, and the tubes and homogenizer were cooled on ice. 4.2. Patch-Clamp Experiments {#sec4dot2-ijms-20-05323} ---------------------------- Patch-clamp experiments using mitoplasts were performed as described previously \[[@B34-ijms-20-05323],[@B63-ijms-20-05323]\]. [Figure 1](#ijms-20-05323-f001){ref-type="fig"} shows a scheme of the patch-clamp experiments. Mitochondria isolated from cells were placed in a hypotonic solution (5 mM of HEPES, 200 μM of CaCl~2~, pH = 7.2) for approximately 1 min to induce swelling and breakage of the outer membrane. Then, a hypertonic solution (750 mM of KCl, 30 mM of HEPES, 200 µM of CaCl~2~, pH = 7.2) was added to restore the isotonicity of the medium. The patch-clamp pipette was filled with an isotonic solution containing 150 mM of KCl, 10 mM of HEPES, and 200 µM of CaCl~2~ at pH = 7.2. The electrical connection was made using Ag/AgCl electrodes and an agar salt bridge (3 M KCl) as the ground electrode. The current was recorded using a patch-clamp amplifier (Axopatch 200B, Molecular Devices Corporation, Sunnyvale, CA, USA). The pipettes, made of borosilicate glass, had a resistance of 10--20 MΩ and were pulled using a Flaming/Brown puller. The currents were low-pass filtered at 1 kHz and sampled at a frequency of 100 kHz. The traces of the experiments were recorded in the single-channel mode. The ionic current was measured, if not otherwise mentioned, in a symmetric 150/150 mM KCl isotonic solution with a 200 µM CaCl~2~ concentration. Mitoplasts are easily recognizable due to their size, round shape, and transparency, which distinguish these structures from cellular debris also present in the preparation. To apply the modulators of the channels, we used a perfusion system containing a holder with a glass tube connected to a peristaltic pump via tubing. The mitoplasts at the tip of the measuring pipette were transferred into the openings of a larger pipe system, and their outer faces were rinsed with the solutions of the channel modulators. The experiments were carried out in patch-clamp inside--out mode. The illustrated channel recordings are representative of the most frequently observed conductance for the given condition. The conductance of the channel was calculated from the current--voltage relationship. The channel open probability (P~o~, open probability) was determined using the single-channel search mode of the Clampfit 10.2 software (Molecular Devices, San Jose, CA, USA). The results are presented as the means ± SD obtained from at least three independent experiments. An unpaired two-tailed Student's t-test was used to identify significant differences; in particular, differences were considered to be statistically significant if \* *p* \< 0.05, \*\* *p* \< 0.01, or \*\*\* *p* \< 0.001. 4.3. Transient Transfection and Fluorescence Staining of H9c2 Cells {#sec4dot3-ijms-20-05323} ------------------------------------------------------------------- The H9c2 WT cells were seeded on a 4-chamber glass bottom dish (Cellvis). After overnight incubation, the cells were transiently transfected with plasmid encoding ROMK2 fused with GFP. Transfection was performed using a Fugene reagent (Promega, Madison, WI, USA). Forty-eight hours after transfection, the mitochondria were additionally stained using 100 nM Mitored (Sigma, St. Louis, MO, USA). Confocal images of living cells were acquired using an Olympus FV 1200. 4.4. Proteinase K and Western Blot Analysis {#sec4dot4-ijms-20-05323} ------------------------------------------- Mitochondria were isolated as described above. Samples were separated into two equal parts and stored on ice. One part was incubated for 15 min in an ice-cold hypotonic buffer to induce swelling and outer membrane breakage, while the second sample was stored in an ice-cold isotonic solution. Additionally, a separate set of mitochondria was solubilized with 0.1% Triton-X 100. Next, half of all samples were treated with Proteinase K (15 µg) for 10 min. Digestion was stopped with 2 mM of phenylmethylsulfonyl fluoride (PMSF) (an inhibitor of the Proteinase K). Next, all samples (around 50 µg of protein each) were separated by 10% sodium Tris-Tricine gel electrophoresis (SDS-PAGE) and transferred onto Polyvinylidene Difluoride (PVDF) membranes (BioRad, Hercules, CA, USA). The membranes were exposed to an anti-ROMK antibody (Sigma-Aldrich, diluted 1:200). To ensure the integrity of mitoplast and the Proteinase K activity, the sample was exposed to anti-Tim23 (BD Biosciences, Franklin Lakes, NJ, USA, diluted 1:2000), anti-Tom20 (Sigma, diluted 1:500) and anti-mHSP75 (Abcam, 1:1000) antibodies. The blots were developed using a secondary anti-rabbit (GE Healthcare UK, Buckinghamshire, UK) or anti-mouse antibody (GE Healthcare UK,) coupled to horseradish peroxidase in conjunction with an enhanced chemiluminescence solution (GE Healthcare UK,). Experiments were performed from independent mitochondrial isolations (*n* = 5). Each mitochondrial isolation was performed from 25--30 mln cells. 5. Conclusions {#sec5-ijms-20-05323} ============== In summary, in the present work, we provide the first direct single-channel patch clamp evidence that the ROMK2 channel is the pore-forming unit of mitoK~ATP~. However, our data do not exclude the possibility that other proteins apart from ROMK might be responsible for ion channel activity with mitoK~ATP~ properties. Our data also suggest the C- and N-terminus matrix localization of the ROMK2 channel in the inner mitochondrial membrane. ###### Click here for additional data file. Supplementary materials can be found at <https://www.mdpi.com/1422-0067/20/21/5323/s1>. Conceptualization, M.L., P.B., B.O., A.S. and B.K.; data curation, M.L., B.A., P.B., J.K. and B.K.; formal analysis, M.L., P.B., A.S. and B.K.; funding acquisition, B.O., A.S. and B.K.; investigation, M.L., B.A., P.B., M.Ż., J.K. and B.K.; methodology, B.A., P.B., A.S. and B.K.; project administration, A.S. and B.K.; resources, P.B., B.O. and A.S.; supervision, P.B., A.S. and B.K.; validation, M.L., B.A., P.B., A.S. and B.K.; writing---original draft, M.L., P.B., B.O., A.S. and B.K.; writing---review and editing, P.B., B.O., A.S. and B.K. This work was funded by Polish National Science Centre (2015/18/E/NZ1/00737 to B.K., 2015/17/B/NZ1/02496 to A.S.) and National Institutes of Health (R01HL108917 to B.O.). The authors declare no conflict of interest. 5-HD 5-hydroxydecanoic acid IMS Mitochondrial intermembrane space IPC Ischemic preconditioning MIM Mitochondrial inner membrane MOM Mitochondrial outer membrane. mitoK ATP Mitochondrial ATP-sensitive potassium channel mitoBK Ca Mitochondrial large-conductance, calcium-regulated potassium channel PK Proteinase K PTP Permeability transition pore ROMK Renal outer medullary potassium channel SUR Sulfonylurea receptor {#ijms-20-05323-f001} {#ijms-20-05323-f002} {#ijms-20-05323-f003} {#ijms-20-05323-f004} {#ijms-20-05323-f005} {#ijms-20-05323-f006} [^1]: Present address: Department of Biosciences, University of Milan, 20133 Milan, Italy. [^2]: Present address: Department of BioMedical Research, University of Bern/Department of Nephrology and Hypertension, Inselspital Bern, Bühlstrasse 28, CH-3010 Bern, Switzerland.

Introduction {#S1} ============ The peroxisome proliferator-activated receptor (PPAR) molecular family is widely studied ([@B1]--[@B3]). These nuclear receptor proteins possess transcription factor activities and influence multiple cellular events at the molecular level including adipocyte differentiation and metabolism. Among them, PPARgamma is of particular interest being expressed by all adipose tissue subtypes and being indispensable for adipose tissue development and for the homeostasis of physiological metabolism ([@B4]--[@B7]). As a consequence, in the mouse systemic loss of PPARgamma activity severely impairs glucose and lipid metabolism as characterized by others ([@B8]--[@B10]). In accordance, PPARgamma null mice are only viable if using conditional knockout strategy ([@B11]). Similar to the mouse above, in human PPARgamma haplo-insufficiency leads to the development of a rare metabolic condition known as familial partial lipodystrophy, type 3 (FPLD3, ORPHA 79083) also characterized by diabetes and dyslipidemia ([@B12]--[@B15]). In mammals, systemic PPARgamma activity may be increased at multiple levels. Environmental factors including excessive caloric consumption or corticosteroid exposure increase PPARgamma activity systemically ([@B16]--[@B18]). Pharmacological systemic activation may be achieved through administration of thiazolidinediones previously used as part of oral antidiabetic treatment, but currently neglected due to adverse cardiovascular side effects ([@B19], [@B20]). Genetic engineering-based enhancement of PPARgamma activity in mouse models has also been performed ([@B21]). In every case, increased PPARgamma activity promotes adipose tissue development at multiple sites of the body. Thymic aging is observed as adipose involution during which the functional thymus niche that normally supports T-cell production is gradually lost and replaced by adipose tissue ([@B22]). The process starts focally in childhood then spreads and accelerates with puberty due to hormonal changes ([@B23]). Diminishing T-cell production results in decreased availability of fresh naive T-cells ([@B24]). Consequences include increasing incidence of infection, cancer and autoimmunity observed at senior ages ([@B25], [@B26]). Thymic adipose involution appears to be PPARgamma-dependent: any condition that systemically enhances PPARgamma activity---either environmental, pharmacological, or genetic---accelerates thymic senescence or adipose involution with all its immunological consequences ([@B27]--[@B32]). However, the opposite phenomenon whether systemically decreased PPARgamma activity can ameliorate long-term functional immune parameters has barely been addressed ([@B33], [@B34]). For this reason, we have set out to characterize the effect of systemic genetic PPARgamma loss of function on long-term immune homeostasis in both mouse and human. Materials and Methods {#S2} ===================== Human Thymus Samples {#S2-1} -------------------- Formalin-fixed, paraffin-embedded (FFPE) human thymus samples from age groups 30--40, 50--60, and 70--80 years were obtained from the Department of Pathology (Faculty of Medicine, University of Pecs, Hungary.) Experiments involving human thymus samples were performed with the consent of the Regional and Local Ethics Committee of Clinical Centre, University of Pecs (ref. no.: 6331/2016) according to their guidelines. All subjects gave written informed consent in accordance with the Declaration of Helsinki. Human Immunohistochemistry {#S2-2} -------------------------- Human thymus lobes were fixed in paraformaldehyde (4% PFA in PBS) then paraffin embedded. 5 µm thick sections were stained using immunohistochemistry ([@B35]). First, the slides were rinsed in heated xylene and were washed with a descending series of alcohol to remove paraffin. After deparaffination the slides were rehydrated in distilled water and antigen retrieval was performed by heating the slides in Target Retrieval Solution (pH 6 DAKO) at 97°C for 20--30 min. Subsequently slides were washed in dH~2~O and endogenous peroxidase activity was blocked with 3% H~2~O~2~ containing TBS (pH 7.4) for 15 min. Then slides were washed three times with TBS containing Tween (0.05%, pH 7.4). Pre-blocking was carried out with 3% BSA in TBS for 20 min before overnight incubation with anti-PPARgamma (1:100, rabbit monoclonal antibody clone: C26H12 Cell Signalling Technology) primary antibody at 4°C. Following incubation slides were washed with TBS for three times then incubated with peroxidase conjugated secondary antibody (1:100, Polyclonal Goat Anti-Rabbit IgG, DAKO) for 90 min. Antibody labeling was visualized with the help of liquid DAB Substrate Chromogen System (DAKO). For nuclear counterstaining, hematoxylin staining was performed. Finally, slides were mounted with Faramount Aqueous Mounting Medium (DAKO). Histological evaluation was performed with the help of Panoramic MIDI digital slide scanner (3DHistech). Image analysis was performed using ImageJ software with IHC toolbox plug-in. Mouse Breeding and Maintenance {#S2-3} ------------------------------ For certain experiments, we have used wild-type and PPARgamma heterozygous (haplo-insufficient) or PPARgamma null (KO) mice of C57BL/6J genetic background. The mice were age matched, and both genders were used for the investigation. The design to generate PPARgamma KO mice was described previously ([@B11]). Briefly, PPARgamma^+/−^/Sox2Cre^+^ male mice were crossed with PPARgamma fl/fl female mice to generate heterozygous PPARgamm^afl/−^/Sox2Cre^−^ and homozygous PPAR gammaΔ^fl/−^/Sox2Cre^+^ mice, wherein the floxed allele was recombined resulting a null allele. Mice were housed under minimal disease conditions in the Laboratory Animal Core Facility of University of Debrecen. Animal rooms were ventilated 15 times/h with filtered air, mice received autoclaved pellet diet (Altromin VRF1) and tap water *ad libitum*. The cages contained sterilized bedding. Room lightning was automated with 12 h light and 12 h dark periods. The room temperature was 21 ± 2°C, the relative humidity is between 30 and 60%. Senescent animals developed and aged normally, without any treatment. Permission to perform the described animal experiments was granted to the relevant utilities of the University of Pecs (ref. no.: BA02/2000-46/2016). Permission to generate PPARgamma GM mice was granted to the relevant utilities of the University of Debrecen (ref. no.: TMF/82-10/2015). Permission to perform experimental procedures with PPARgamma GM mice was granted to the relevant utilities of the University of Pecs (ref. no.: TMF/124-11/2017). Mouse Immunofluorescence {#S2-4} ------------------------ Immunofluorescent staining was performed on 8 µm cryosections of mouse thymus lobes as described previously ([@B35]). Briefly, the slides were fixed in cold acetone, then dried and blocked to prevent non-specific staining using 5% BSA in PBS for 20 min before staining with fluorochrome-conjugated or primary antibodies: anti-EpCAM1-FITC (1:100, rat monoclonal antibody clone: G8.8), anti-Ly51-PE (1:100, rat monoclonal antibody clone: 6C3, eBioscience), and anti-PPARgamma (rabbit monoclonal antibody clone: C26H12 Cell Signaling Technology). For secondary antibody, Alexa-555 conjugated a-rabbit goat IgG (1:200, Life Technologies) was used. In certain cases, DAPI (Life Technologies) nuclear counterstain was also applied. Sections were analyzed using a Nikon Eclipse Ti-U microscope equipped with a CCD camera (Andor Zyla 5.5) and NIS-Elements software. The medulla/cortex ratio was calculated using ImageJ software. Mouse Flow Cytometry {#S2-5} -------------------- Thymocyte subsets and T-cell subpopulations in blood were investigated by flow-cytometry as published by others ([@B36], [@B37]). Thymocytes and PBMC were isolated from mice and labeled with fluorophore-conjugated antibodies in PBS-BSA (5% BSA diluted in PBS). In every case, 100,000 cells were stained for measurement. Incubation with antibodies was performed at 4°C for 60 min followed by a washing step. FACSCanto II flow-cytometer and FACSDiva software (Becton Dickinson) were used for analysis. In every case, 10,000 events (parent R1 morphological lymphocyte gate) were recorded by flow-cytometry. For thymocyte subset measurement, Alexa-647 conjugated antimouse CD4 (clone: YTS 191) and FITC-conjugated antimouse CD8 (clone: IBL 3/25) antibodies were used (both produced in the Department of Immunology and Biotechnology, University of Pecs, Hungary). For peripheral blood T-cell subpopulation analysis, Pacific Blue-conjugated antimouse CD3 (clone: 17A2), PerCP-conjugated antimouse CD4 (clone: GK1.5), APC/Cy7-conjugated antimouse CD8 (clone: YTS156.7.7), PE-conjugated antimouse CD44 (clone: IM7), APC-conjugated antimouse CD62L (clone: MEL-14) (all purchased form BioLegend), and FITC-conjugated antimouse CD19 (clone: 1D3, produced by the Department of Immunology and Biotechnology, University of Pecs, Hungary) were used. T-Cell Recombination Excision Circle (TREC) Measurement by Digital qPCR in Mouse and Human {#S2-6} ------------------------------------------------------------------------------------------ T-cell recombination excision circle by-products of gene-rearrangement in fresh naive T-cells were also assessed. We performed mouse Trec (mTrec) digital qPCR using mouse and human Trec (hTrec) digital qPCR using human samples by adapting methods published by others ([@B38]). Briefly, DNA was isolated from mouse thymocytes using the NucleoSpin Tissue kit (Macherey-Nagel) according to the manufacturer's instruction. For human, peripheral-blood samples were processed using the DNA Blood Mini kit (Qiagen) following the manufacturer's guides. Absolute copy numbers were measured by digital PCR on the QuantStudio 3D Digital PCR platform (ThermoFisher) using 30 ng DNA per sample. Taqman primers/probes and digital qPCR reagents were also purchased from ThermoFisher and used as suggested. For age-matched range of healthy human hTrec values, refer to the work of Lynch et al. ([@B38]). Permission to perform the described animal experiments was granted to the relevant utilities of the University of Pecs (ref. no.: BA02/2000-46/2016). Experiments involving human blood samples were performed with the consent of the Regional and Local Ethics Committee of Clinical Centre, University or Pecs (ref. no.: 6439/2016) according to their guidelines. Oral Tolerance Induction in Mouse {#S2-7} --------------------------------- Induction and evaluation of oral tolerance was performed as described by others ([@B39]--[@B41]). Briefly, both wild-type and PPARgamma haplo-insufficient mice received 5 mg/ml ovalbumin (OVA, Sigma-Aldrich) in drinking water for seven days. On day 7, mice were challenged with an intraperitoneal injection of 5 µg ovalbumin in 200 µl of 1:1 of PBS:complete Freund adjuvant. On day 14, mice received an intraperitoneal injection of 5 µg ovalbumin in 200 µl of 1:1 of PBS:incomplete Freund adjuvant. Serum was collected on day 21 and anti-OVA IgG antibodies were measured by ELISA. Briefly, 96-well Microtest Plates (Sarstedt) were coated with OVA and blocked with BSA. Then plates were incubated with serial dilutions of mouse serum samples (1:100--1:3,200). The antibody content was visualized with the help of HRP-conjugated antimouse immunoglobulin antibody (rabbit polyclonal, Dako). Optical density was measured at 492 nm with iEMS Reader MF equipment (Thermo Labsystems). Influenza Vaccination in Mouse {#S2-8} ------------------------------ The efficiency of influenza vaccination was investigated as described elsewhere ([@B42]). Briefly, both wild-type and PPARgamma haplo-insufficient mice were injected intramuscular once with 0.1 ml human seasonal influenza vaccine cocktail (3Fluart) to mimic human vaccination at 9 months of age. In order to imitate human exposure pattern serum antibody IgG titer against H1N1 A/California/7/2009 strain (part of 3Fluart) was measured by ELISA three months after initial single vaccination at 12 months of age. For detection, ELISA plates were coated with 0.05 μg HA protein of influenza strain A (Recombinant subtype H1N1 A/California/7/2009 His Tag, Life Technologies). Then plates were incubated with serial dilutions of mouse serum samples (1:5--1:1,600). The antibody content was visualized with the help of HRP conjugated a-mouse immunoglobulin antibody (rabbit polyclonal, Dako). Optical density was measured at 492 nm with iEMS Reader MF equipment (Thermo Labsystems). Statistical Analysis {#S2-9} -------------------- All experiments were performed at least on three occasions, representative experiments are shown. Measures were obtained in triplicates, data are presented as mean and +SD as error bars. Graphpad Prism software was used for statistical analysis. Two-tailed Student's *t*-test was applied. Significant differences are shown by asterisks (ns for not significant, \* for *p* ≤ 0.05, \*\* for *p* ≤ 0.01, \*\*\* for *p* ≤ 0.001). Results {#S3} ======= PPARgamma Distorts the Ratio of Thymic Epithelial Compartments with Age {#S3-1} ----------------------------------------------------------------------- Previously reported mouse results showed increasing PPARgamma expression with age in the thymic epithelial compartments, accompanied by thymic adipose involution. We have set out to prove human relevance of previous mouse findings and test whether PPARgamma activity influences the ratio of thymic epithelial compartments. ### PPARgamma Expression Increases in the Adult Thymus with Age {#S3-1-1} Human FFPE thymic sections were analyzed for their PPARgamma expression in several adult age groups from young through middle-aged to senior (Figures [1](#F1){ref-type="fig"}A--D). Our results indicate that PPARgamma expression significantly and progressively increases with age (Figures [1](#F1){ref-type="fig"}A--C). Of note, total cellular areas shrink at senior ages in both human (Figure [1](#F1){ref-type="fig"}C) and mouse (Figure [1](#F1){ref-type="fig"}F). As a result the ratio of PPARgamma-expressing cellular areas shows relative increase with age (Figure [1](#F1){ref-type="fig"}D). Immunofluorescent staining of mouse thymic cryosections at 15 months of age (Figure [1](#F1){ref-type="fig"}F) provides visual support for thymic epithelial to adipose transdifferentiation in harmony with the working hypothesis of cellular transdifferentiation. A portion of stromal cells shows dual staining for epithelial identity and adipose differentiation, a hallmark of thymic adipose involution. This phenomenon is not observed at young adult age (Figure [1](#F1){ref-type="fig"}E). {ref-type="fig"}F). For exact numerical data, refer to Supplementary Material. Significant differences are shown by asterisks (ns for not significant, \* for *p* ≤ 0.05, \*\* for *p* ≤ 0.01, \*\*\* for *p* ≤ 0.001).](fimmu-08-01515-g001){#F1} ### PPARgamma Skews the Ratio of Epithelial Compartments with Age {#S3-1-2} Mouse thymic cryosections were differentially stained for medullary and cortical epithelial compartments at several ages and using various genetic backgrounds (Figures [2](#F2){ref-type="fig"}A--D). Our results show that in the wild-type setting the medullary epithelial compartment significantly shrinks with age as reported previously ([@B31]). This, however, is not observed in PPARgamma deficient settings. Loss of PPARgamma activity shows protection in a progressive manner presenting dose--response (Figure [2](#F2){ref-type="fig"}E). PPARgamma deficiency efficiently and significantly prevents the erosion of the medullary epithelial compartment, otherwise prone to shrink with senescence. {#F2} PPARgamma Affects Thymic T-Cell Production and Peripheral Blood T-Cell Distribution with Age {#S3-2} -------------------------------------------------------------------------------------------- We have observed changes in thymus architecture in response to PPARgamma status. Consequently, we were interested in whether morphological changes alter thymus function: naive T-cell production. Going beyond, we were eager to see if sustained influence of PPARgamma status on thymocyte function is also reflected in the peripheral blood. ### PPARgamma Disturbs Thymic T-Cell Output with Age {#S3-2-1} Age-related changes in thymocyte levels of mTrec (DNA loop by-product of mouse T-cell receptor gene rearrangement) were evaluated in wild-type and PPARgamma deficient settings using digital qPCR (Figure [3](#F3){ref-type="fig"}A). Our results indicate slight (though not significant) decrease of mTrec and hence fresh-naive T-cell output with age in thymocytes of wild-type mice. PPARgamma deficiency significantly and progressively counteracts the process also showing dose-responsive increase of thymocyte mTrec levels. In further analyses, the percent distribution of thymocyte subpopulations was assessed using flow-cytometry in wild-type and PPARgamma deficient mice (Figure [3](#F3){ref-type="fig"}B). All thymocyte subpopulations showed near identical distribution pattern with all genetic backgrounds. Taken together, PPARgamma deficiency progressively enhances thymocyte development in adult age, but without skewing the distribution of thymocyte subpopulations or their differentiation preference. {#F3} ### PPARgamma Influences T-Cell Subpopulation Distribution in Adult Peripheral Blood {#S3-2-2} Peripheral blood T-cell subpopulations were evaluated by flow-cytometry at 12 months of age in wild-type and PPARgamma deficient animals. Our results do not show differences in the percent distribution of the major T-cell groups of helper T-cells and cytotoxic T-cells (Figure [4](#F4){ref-type="fig"}A) within the CD3-gate of T-cells. However, the evaluation of naive T-cell and memory T-cell ratio reveals significant effect of PPARgamma deficiency (Figure [4](#F4){ref-type="fig"}B). There is significant increase of naive T-cells in the peripheral blood of PPARgamma deficient animals compared to wild-type animals, conversely and significantly decreasing the memory T-cell pool within the CD3-gate of T-cells. Deeper analysis of the memory T-cell pool reveals it is the mobile effector memory T-cell subpopulation that shows significant decrease and not central memory T-cells (Figure [4](#F4){ref-type="fig"}C) within the CD3-gate of T-cells. Sustained and prolonged naive T-cell production due to PPARgamma deficiency in the thymus as suggested by mTrec values above apparently affects peripheral blood T-cell subpopulations as shown here. {#F4} Functional Immunological Consequence and Human Relevance {#S3-3} -------------------------------------------------------- Having seen the far-reaching influence of PPARgamma status on thymus architecture, thymus function and peripheral blood T-cell composition with age, we have set out to test whether these changes have functional immunological relevance. If so, it would be also of high interest to test if our comprehensive mouse results have human relevance. ### PPARgamma Modulates Immune Regulation and Immune Response {#S3-3-1} We have tested the capacity to mount oral tolerance to the foreign protein OVA in wild-type and PPARgamma deficient aged adult mice by measuring OVA-specific IgG titers following oral and/or intraperitoneal OVA challenge (Figure [5](#F5){ref-type="fig"}A). As reported by others, age impairs oral tolerance in wild-type animals ([@B40], [@B41]). As a consequence, there is only moderate, insufficient decrease of OVA-specific IgG titers in case of parallel oral OVA administration and i.p. OVA-injection in senior animals. However, PPARgamma deficiency rescues oral tolerance in the same experimental setting despite age, profoundly and significantly decreasing OVA-specific IgG titers (Figure [5](#F5){ref-type="fig"}A). Consequently, naive T-cell dependent immune regulation (oral tolerance) remains efficient in PPARgamma heterozygous animals despite their age. {#F5} The capacity to mount immune reaction to foreign influenza antigens was also tested as human seasonal influenza vaccine was injected into aged adult wild-type and PPARgamma deficient animals. Subsequent analysis of serum IgG titers specific to a vaccine component showed elevated protective antibody production (maximal ELISA OD values) in PPARgamma deficient animals, but not in their wild-type littermates (Figure [5](#F5){ref-type="fig"}B). This tendency is not significant because of individual variation observed due to the applied human vaccination protocol being inferior to standard mouse immunization protocol. Nevertheless, naive T-cell dependent immune response proves to be efficient in aged, PPARgamma heterozygous animals. ### Human Evidence of PPARgamma Deficiency Preventing Thymic Senescence {#S3-3-2} Genetic PPARgamma deficiency is a rare, but existing condition in human called FPLD3 ([@B15]). It leads to a metabolic phenotype called lipodystrophy, similar to the mouse ([@B11]--[@B15]). Other rare human conditions not affecting PPARgamma can also lead to lipodystrophy ([@B12]--[@B15]). In case of FPLD2 lamin mutations trigger similar metabolic changes ([@B14]). Peripheral blood hTrec (DNA loop by-product of human T-cell receptor gene rearrangement) levels were measured using digital qPCR in age-matched patients with FPLD2 condition and FPLD3 condition (Figure [6](#F6){ref-type="fig"}). As expected and in perfect harmony with previous mouse thymocyte results elevated mean hTrec levels were detected in FPLD3 samples compared to FPLD2 samples. The tendency is not significant due to individual variation within the patient groups. Unfortunately, current patient sample numbers cannot be increased due to the extremely rare nature of these conditions (FPLD2 or ORPHA 2348 has prevalence of ≤1/1,000,000 and FPLD3 or ORPHA 79083 also has prevalence of ≤1/1,000,000) ([@B14], [@B15]). For age-matched range of healthy human hTrec values, refer to the work of Lynch et al. ([@B38]). Lower limit of healthy human hTrec threshold (approx. 200 copies/μg DNA) is not reached by FPLD2 (lamin) patient samples, but this is rescued in FPLD3 (PPARgamma) patients despite being age and disease matched. {ref-type="fig"}). Patient sample numbers were *n* = 3 for FPLD2 and *n* = 5 for FPLD3. For exact numerical data, refer to Supplementary Material. For age-matched (approx. 50 years of age) range of healthy human hTrec values, refer to the work of Lynch et al. ([@B38]). Accordingly, the lower limit of healthy human hTrec threshold (approximaterly 200 copies/μg DNA) is represented by dotted line.](fimmu-08-01515-g006){#F6} Discussion {#S4} ========== PPARgamma Drives Thymic Epithelial to Adipose Trans-Differentiation with Age {#S4-1} ---------------------------------------------------------------------------- It has been previously suggested based on direct fate-mapping experiments that with senescence thymic adipose tissue develops from the thymic stromal or epithelial compartment ([@B28]). Based on indirect evidence others have also supported this concept ([@B29]). In further support, we here present visual evidence of epithelial to adipose transdifferentiation in the mouse. This is indicated by the presence by EpCAM-1/PPARgamma double-positive cells shown by histology (Figure [1](#F1){ref-type="fig"}D). These cells still express cell surface markers of their fading thymic epithelial identity (EpCAM-1), but already show early signs of the novel adipocyte differentiation program in their nuclei (PPARgamma). The fact that such double positive cells show rather scattered and not uniform staining pattern at a given time point may provide explanation for gradual thymic adipose involution observed during senescence. PPARgamma Impairs Naive T-Cell Production with Age {#S4-2} -------------------------------------------------- Thymus histology data show that the medullary compartment is rescued from age-related shrinking in case of PPARgamma deficiency (Figures [2](#F2){ref-type="fig"}A--D). Extended survival of this stromal niche ensures permissive environment for sustained thymus function: naive T-cell production. This is indicated by elevated mTrec values showing direct correlation with PPARgamma deficiency (Figure [3](#F3){ref-type="fig"}A). Of extreme importance and highlighting human relevance, peripheral blood leukocyte hTrec values from adult FPLD3 patients (with genetic PPARgamma deficiency) also exceed adult FPLD2 patient values (with unrelated genetic background) despite being age-matched and disease-matched (lipodystrophy, diabetes) (Figure [6](#F6){ref-type="fig"}). Of note, such metabolic disorders are known to impair thymus function indicated by decreased hTrec values as reported by others ([@B43], [@B44]). For exactly, this reason have we used disease-matched controls (FPLD2 vs. FPLD3) to show enhanced thymus function with PPARgamma deficiency despite metabolic disorders. Unlike lower than physiological hTrec values measured in FPLD2 (lamin) patients, those measured in FPLD3 (PPARgamma) patients are within healthy human physiological range (Figure [6](#F6){ref-type="fig"}). Since both mTrec and hTrec DNA loops originate from gene rearrangement during thymocte development this is direct evidence of sustained T-cell development indicating intact thymic niche in PPARgamma deficient animal models and human patients ([@B38]). Of note, the distribution of thymocyte subpopulations shows identical pattern irrespective of PPARgamma status proving that sustained, enhanced thymocyte development does not skew differentiation preference, but rather enhances fresh, naive T-cell production of all thymocyte subtypes uniformly (Figure [3](#F3){ref-type="fig"}B). Finally, since sustained thymic naive T-cell production is not restricted to a given time-point, but rather represents a continuous trend, the peripheral blood naive T-cell population shows cumulative differences as it is rescued from age-driven shrinking, against the memory T-cell population---more specifically against the effector memory T-cell pool (Figures [4](#F4){ref-type="fig"}B,C). PPARgamma Hampers T-Dependent Immune Regulation and Immunity with Age {#S4-3} --------------------------------------------------------------------- Oral consumption of foreign T-depended antigen normally initiates immune tolerance inhibiting any eliminative immune response (e.g., serum IgG), despite parallel immunization in young adult individuals with appropriate naive T-cell supply. Unfortunately, the phenomenon is disrupted at senior age due to the lacking naive T-cell pool in the Peyer's patches of the gut ([@B40], [@B41], [@B45]) This loss of oral tolerance (impaired immune regulation) is a possible link to increasing food intolerance prevalence observed in the aging adult population ([@B46]--[@B49]). However, the phenomenon may be rescued by PPARgamma deficiency despite age providing evidence that sustained T-cell production is necessary for efficient oral (immune) tolerance (Figure [5](#F5){ref-type="fig"}A). Senescence-triggered decrease of naive T-cell output also impairs T-dependent immunity. An example in the senior human population is decreased protection from seasonal flu strains despite annual vaccination campaigns ([@B50]--[@B52]). The phenomenon has well established animal models ([@B53]--[@B55]). This is caused by low levels of neutralizing antibody titers due to lacking naive T-cells necessary during T-B cooperation to mount adequate innate immune response against T-dependent antigens of the vaccine. This, however, is not the case with PPARgamma deficiency (Figure [5](#F5){ref-type="fig"}B). Single intramuscular vaccination against seasonal flu (mimicking human vaccination campaign) resulted in higher maximal antibody production three months later (a typical delay in human exposure). This confirms that the cause of decreased vaccination efficiency in the senior population is impaired T-dependent immunity due to thymic senescence. In our experiments, we have focused on the decline of T-dependent immunity since the thymus shows early and dramatic signs of senescence during adipose involution. This, however, is not the case for the B-cell compartment for which aging has been reported to occur later and in a more gradual fashion, lacking such profound histological changes ([@B56]). PPARgamma is an enigmatic transcription factor showing unique expression pattern in both time and space throughout the body ([@B57]). PPARgamma affects both hemopoietic and stromal compartments during development and aging. Further dissection would require to perform, e.g., bone-marrow transplantation experiments between control and PPARgamma deficient animals. However, PPARgamma KO animals develop severe metabolic disorders that hamper such experiments, especially at elevated ages. Limitations and Perspectives {#S4-4} ---------------------------- We here present the long-term thymus- and T-dependent immunity-preserving effect of systemic (genetic) loss of PPARgamma function as observed in PPARgamma deficient mouse models and in a human rare disease (FPLD3). In both cases, there are severe metabolic drawbacks (diabetes, dyslipidemia etc.) due to systemically lacking PPARgamma activity. However, alternative, thymus tissue-restricted suppression of PPARgamma activity would likely solve the issue. Of note, as reported previously, overexpression of Wnt4 glycolipoproteins by thymic epithelial cells can efficiently counteract PPARgamma ([@B31]). Also, Wnt4 was described to travel in extracellular vesicles including exosomes and affect thymocyte differentiation ([@B58], [@B59]). Hence, it is conceivable that thymic epithelium-derived, enriched exosomes would efficiently home to the thymus and deliver their Wnt4 cargo locally even when administered systemically. This would, in theory, allow for the natural, tissue-specific, protein-mediated maintenance of thymic epithelial identity and prevent thymic senescence from developing. Although tissue senescence is ultimately inevitable, there are conditions that accelerate thymic senescence including certain viral infections, intoxications, irradiation, chemotherapy, etc. Outcomes include increased incidence of infection, cancer and autoimmune disorder. In any case, the identification of molecular level targets for potential intervention is highly desired. Therefore, molecular level insight into immune senescence has medical, economical, and personal relevance, all at once. Ethics Statement {#S5} ================ Experiments involving human thymus samples were performed with the consent of the Regional and Local Ethics Committee of Clinical Centre, University or Pecs (ref. no.: 6331/2016) according to their guidelines. Experiments involving human blood samples were performed with the consent of the Regional and Local Ethics Committee of Clinical Centre, University or Pecs (ref. no.: 6439/2016) according to their guidelines. All subjects gave written informed consent in accordance with the Declaration of Helsinki. Permission to perform the described animal experiments was granted to the relevant utilities of the University of Pecs (ref. no.: BA02/2000-46/2016). Permission to generate PPARgamma GM mice was granted to the relevant utilities of the University of Debrecen (ref. no.: TMF/82-10/2015). Permission to perform experimental procedures with PPARgamma GM mice was granted to the relevant utilities of the University of Pecs (ref. no.: TMF/124-11/2017). Author Contributions {#S6} ==================== DE performed most histological, molecular biology, and statistics work in the project and was involved in manuscript preparation. KB performed all human IHC work. ZK performed oral immune tolerance experiments. AP was in charge for the breeding, metabolic, and genetic characterization of PPARgamma haplo-insufficient and null mice. JL was in charge for planning human experiments, involved in manuscript preparation as well as local supervision of respective department. PE was involved in planning mouse experiments, involved in manuscript preparation as well as local supervision of respective department. KK was involved in histological, molecular biology and statistics work, also in planning experiments and manuscript preparation, and supervised the project. Conflict of Interest Statement {#S7} ============================== The authors declare that they have no conflicts of interest with the contents of this article. The research was conducted in the absence of any commercial or financial relationship that could be construed as a potential conflict of interest. The authors wish to thank the PPARgamma^+/−^ and PPARgamma^fl/fl^ mice that were obtained from Yaacov Barak PhD (Salk Institute, La Jolla, CA, USA) and also the Sox2Cre mice obtained from Beatrice Desvergne MD, PhD (University of Lausanne, Switzerland). We are grateful for Gregory D. Sempowski MD PhD (Duke Human Vaccine Institute, Duke University, Durham, NC, USA) for providing the Trec primer and probe sequences along with protocols for TaqMan qPCR, and also Peter Balogh MD PhD (Department of Immunology and Biotechnology, University of Pecs, Hungary) for providing test antibodies for mouse CD3, CD4, CD8. The authors wish to thank David B. Savage, MD, PhD (Metabolic Research Laboratories, School of Clinical Medicine, University of Cambridge, Cambridge, United Kingdom) for providing peripheral blood DNA samples from genetically verified FPLD2 and FPLD3 rare disease patients. **Funding.** Scientific research support was provided by the Hungarian National Science Foundation (no. 78310) and PTE AOK KA-2016-16 to KK. The project was also supported by the University of Pecs in the frame of Pharmaceutical Talent Center program and the Viral Pathogenesis Talent Center program *via* KK. The Janos Bolyai Scholarship of the Hungarian Academy of Sciences also supported KK. JEP was supported by the European Union and he State of Hungary, co-financed by the European Social Fund in the framework of GINOP 2.3.2--15--2016-00022 TAMOP-4.2.2. A-11/1/KON-2012-0024, TAMOP-4.2.4.A/2-11/1-2012-0001 "National Excellence Program," PTE AOK-KA-2013/22 and EFOP-3.6.1-16-2016-00004. The present scientific contribution is also dedicated to the 650th anniversary of the foundation of the University of Pecs, Hungary. Supplementary Material {#S8} ====================== The Supplementary Material for this article can be found online at <http://www.frontiersin.org/article/10.3389/fimmu.2017.01515/full#supplementary-material>. ###### Click here for additional data file. [^1]: Edited by: Duncan Howie, University of Oxford, United Kingdom [^2]: Reviewed by: Susan Mariola Schlenner, KU Leuven, Belgium; Juei-Tang Cheng, Chang Jung Christian University, Taiwan [^3]: Specialty section: This article was submitted to Immunological Tolerance and Regulation, a section of the journal Frontiers in Immunology

Ultrashort pulse lasers have made great progress since its emergence. As an important research field, they have been motivated by the rapid advance of various applications ranging from optical imaging and spectroscopy, to laser processing and optical communication[@b1][@b2]. Their dramatic development and extensive applications urge us to fully and intactly characterize an ultrashort pulse. Among all the techniques, frequency-resolved optical gating (FROG) is a common and effective approach. Conventional FROG system requires a nonlinear optical (NLO) material in bulk in order to gain abundant signal photons for detection, and as a trade-off, the setup is always elaborate and presents a considerable complexity, especially for pulses like super-continuum[@b3][@b4]. Meanwhile, modern spectro/microscopy like coherent Raman spectroscopy, tip-enhanced spectroscopy, and multi-photon microscopy have got emerging interests towards higher resolution in temporal, spectral and spatial domains. Because of the sophisticated system, however, it is impossible to implement an *in-situ* measurement, and thus detailed information of the target site is still missing[@b5][@b6][@b7]. Nano-scale materials have the inherent advantage on fine structure imaging and spectral analysis, since they can be employed as tracking material without perturbing the local environment. The usage of nanomaterial based FROG (nano-FROG) would relieve the phase-matching condition, and provide an avenue to monitor the nano-domain distortions in near field[@b8][@b9][@b10][@b11]. Exsiting nano-FROG methods, either via randomly dispersed nanoparticles or via nanoparticles attached to fiber taper, their FROG signal collections still rely on free-space optics. The loss by a series optical elements largely limits the sensitivity and reliability of the apparatus. Moreover, the NLO signal of these methods is mixed with fundamental wave in collection, so it requires optical filters to differentiate them. The irradiation detection further imposes a restriction in measuring broadband pulse when fundamental wave partly overlaps with signal in spectrum, and thus filters are not able to separate them. Besides, the variation and irregularity of the nanomaterials make them improbable to reproduce a consistent result among different measurements. More recently, transverse frequency conversion in semiconductor nanowires has been attracted lots of attentions because of its high conversion efficiency and low divergence angle[@b8][@b12][@b13][@b14]; although the feasibility of pulse measurement has been discussed, however, using this scheme to perform FROG measurement and retrieval of ultrashort and broadband pulses have not been reported yet. In this paper, we demonstrate a waveguide based FROG method with single nanowire (NW) as NLO material, through which the phase and amplitude of pulses are measured. Results ======= NW synthesis and Characterization --------------------------------- Here we chose the zinc oxide (ZnO) as the NLO material, which is a typical II-VI semiconductor with moderate absorption loss and wide band gap (3.37 eV, 368 nm)[@b15]. ZnO NW used here was synthesized via a vapor-liquid-solid (VLS) process[@b16]. [Figure 1(a)](#f1){ref-type="fig"} gives the scanning electron microscopy (SEM) image of as-fabricated ZnO NW, showing the NWs with good smoothness and uniformity; to obtain high reflection end faces, we use a tungsten probe to fracture the NW by micro-manipulation[@b13][@b14], and the endface image is shown in [Fig. 1(b)](#f1){ref-type="fig"}. Before manipulation, the ZnO NWs were removed from the growth substrate and deposited on a low refractive-index MgF~2~ substrate, and the typical optical micrograph was shown in [Fig. 1(c)](#f1){ref-type="fig"}. The as-fabricated NWs were all around 200 \~ 400 nm in diameter and the lengths can be up to several hundreds of micrometers. By means of butt-coupling technique under a microscope[@b13][@b14], ZnO NW can be manipulated and placed onto a a silica optical fiber taper, so that the light can be efficiently transferred into the NW. [Figure 1(d)](#f1){ref-type="fig"} shows the micrograph of a random fiber taper-NW couple, and its optical guiding behavior can be examined by launching a 1064-nm-wavelength continuous-wave (CW) laser as shown in [Fig. 1(e)](#f1){ref-type="fig"}, from which we can find there is no scattering spots on the NW, and its conducting feature is good. Experiment setup ---------------- The schematic diagram of the measurement setup is illustrated in [Fig. 2](#f2){ref-type="fig"}. A Ti:Sapphire mode-locked laser with central wavelength at 810 nm (80 MHz, \~70 fs, Spectra-Physics) was used here. After going through an isolator and a pinhole, the laser beam was split into two halves by a wave plate and a polarization beam splitter (PBS). The input power of the 810 nm pulsed light was \~10 mW for each arm and the light transfer efficiency was measured to be \~60% for single taper-NW coupling. One beam was used as target pulse, and the other beam as gate pulse was introduced into a retro-reflector, by which the length of optical path can be tuned directly. Two beams were individually coupled into two single-mode fibers (112 cm long each) whose ends were tapered to microfibers. Both tapered microfibers were suspended above a MgF~2~ microchannel with \~200 μm in seperation. A 245-μm-long, 310-nm-diameter ZnO NW was coupled with them via micro-manipulation under optical microscope[@b13][@b14][@b17]. The angle and position of both tapers can be carefully adjusted to maintain an optimal input coupling efficiency. The surface emitted signal from the NW was collected by a objective lens, and then filtered and directed into a CCD (DS Ri1, Nikon); part of the signal was reflected by a neutral beam-splitter (45/55) into a spectrometer (QE65 pro, Ocean Optics). The mechanism of surface emitted sum frequency (SF) signal generated in a NW is sketched in [Fig. 3(a)](#f3){ref-type="fig"}, and we take second harmonic (SH) process as a special case of SF with two identical inputs. Briefly, two series of pulses are injected from each end of the NW and counter propagate in it. By tuning the delay line, pulses from opposite directions collide within the NW[@b8][@b14]. As required by phase-matching conditions, the nonlinear optical emission will be generated transversely as shown in [Fig. 3(a)](#f3){ref-type="fig"} inset. With two series of 810 nm pulses injected from both ends, the spectra of signal can be detected perpendicularly. As shown in [Fig. 3(b)](#f3){ref-type="fig"} that the signal was at the half-wavelength position of fundamental light with several nanometers in width, which confirmed it to be SH signal. Since the optical path interval between two neighboring pulses is in the scale of several meters, only one pair of pulses can be in the NW at the same time. Besides, from our result, the SH signal generated by one pulse beam through the birefringent process was tens of times weaker than the SH signal generated by two pulse correlating, which allowed for the measurement of pulses with very low energy. Auto-correlation ---------------- To verify the feasibility of using single NW for pulses characterization, we first study the pulses colliding process in the NW. [Figure 4(a--e)](#f4){ref-type="fig"} were SH images recorded at around 405 nm during the pulse series optically gated by itself. The target pulses came from right and the gate controlled by retro-reflector came from left. It can be seen that the SH signal was first generated from left end of ZnO NW \[[Fig. 4(a)](#f4){ref-type="fig"}\], become stronger with the pulses moving in \[[Fig. 4(b)](#f4){ref-type="fig"}\], then reached its maximum in [Fig. 4(c)](#f4){ref-type="fig"} and disappeared from right end of NW finally \[[Fig. 4(d,e)](#f4){ref-type="fig"}\]. The corresponding position-dependent SH intensity curves of [Fig. 4(a--e)](#f4){ref-type="fig"} were plotted in [Fig. 4(f--j)](#f4){ref-type="fig"} in black dots by integrating the intensity of image pixels, and the Y-axis of each plot were altered to form a better contrast. The periodical pattern was originated from the guided modes of the counter-propagating waves. As proved in our previous study[@b13][@b14], the guided modes in NW depend on the diameter of NW, the relative position and the angle between the fiber taper and the NW. And the guided modes would further influence the SH emission. The spatial distributions in transverse SH emission are determined by the relationship of *I*^*SH*^ ∝ cos^2^ (Δ*β*z), where Δ*β* represents the propagation constant difference between the encountered guided modes. So the SH images of NW could manifest an oscillation behavior when Δ*β* ≠ 0. Nevertheless, the SH spectrum would not be influenced by the change of patterns. The red line for each panel in [Fig. 4(f--j)](#f4){ref-type="fig"} plotted the Gaussian fitted profiles in order to make the movement of the intensity peak more obvious. FROG measurement of 22-nm-wide pulses ------------------------------------- From the intensity pattern in [Fig. 4](#f4){ref-type="fig"}, we could find that with continuous movement of the delay line, the SH maximum could move in and out of the NW. For a rough estimation of pulse duration, we observed the variation of intensity at an arbitrary position of NW \[i.e. right in the middle of [Fig. 4(f)](#f4){ref-type="fig"}\], and marked the displacement of the delay line between its two half-maximum, which was \~150 μm. So the full width at half maximum (FWHM) of a pulse could be calculated by \~300 μm divided by light speed, which was in the scale of picoseconds in time profile. To operate a more precise pulse characterization, we measured the auto-correlating spectra of pulses counter-propagating through the NW. With a speed of 25 μm/step, the surface emitted SH signal could be scanned out. The pulse energy was about 50 pJ/pulse and a typical spectrum was taken with 1 s exposure time. [Figure 5(a)](#f5){ref-type="fig"} shows the spectra collection in form of a 128 × 128 grid graph, in which 128 spectra were placed in chronological order, and the intensities of spectra were represented by colormap. Notice that the temporal index was transformed from free-space optical-length displacement of delay line. Based on the results, a retrieved FROG trace could be readily obtained by a method given in ref. [@b18] \[[Fig. 5(b)](#f5){ref-type="fig"}\]. Briefly, the method involved a solution to two-dimentional phase-retrieval problem. Two constraint sets, nonlinear-optical set and the experimental data set, are to be satisfied by alternately iteratively projecting from an initial guess. The intersection of the two constraints will lead to a solution of pulse electric field with reliable temporal and spectral information. The measured and retrieved FROG traces in [Fig. 5](#f5){ref-type="fig"} are in good qualitative agreement with the G error of 0.0107, which could be elevated with a higher-resolution spectrometer. The pulse intensity and phase profile are also presented in both temporal and spectral domains, as shown in [Fig. 5(c,d)](#f5){ref-type="fig"}, respectively. The intensity profile in [Fig. 5(c)](#f5){ref-type="fig"} indicates the pulse duration of 2.8 ps, much longer than the source output. The pulse broadening mainly occurred in two parts: the \~1.1 m-length fiber, and the \~10-mm-length tapered microfiber. For a rough estimation, we can expect the dispersion as \~118 ps/nm/km for an 810-nm pulse with 22-nm width in a 1-m-long single-mode fiber (SMF-28, Corning), which leads to pulse broadening of more than 2.8 ps[@b19]. While for the fiber taper, the dispersion in the 1-μm-diameter microfiber end was estimated around 400 ps/nm/km[@b20]. Assuming that the dispersion changes linearly in the taper, the pulse width should increase for another \~100 fs. Since the length of the ZnO NW is relative short (\~200 μm), the pulse broadening imposed was neglected here. Therefore, in the process the pulse change should be \~2.8 ps, almost consistent with our measurement results. Cross-correlation FROG (XFROG) measurement of 300-nm-wide pulses ---------------------------------------------------------------- Since ZnO NW has broadband transmittance in the visible to NIR spectral region, we further operated cross-FROG to measure supercontinuum (SC) pulses on the same system. The SC was generated by adding a photonic crystal fiber (FemtoWHITE, Newport) into the path between M3 and L2 in [Fig. 2](#f2){ref-type="fig"}. A norch filter (NF03--808E, Semrock) was used to block the fundamental wave portion in the SC, otherwise the signal from the broadband portion would be under low contrast. Besides, a long-pass filter (BLP01--664R, Semrock) was inserted to avoid the absorption of NW, which might induce undesired luminescence. After both filters, the SC power output from PCF was left for \~60 mW, which was 2.5 pJ/nm/pulse in average for overall 300-nm-wide band. For the XFROG measurement, the SC was gated by the 810 nm pump pulse. The delay between the SC and gate pulses was scanned in 300-fs steps, and a series of SF signal generated by SC and the gate pulses was resolved at each delay forming a 256 × 256 trace \[[Fig. 6(a)](#f6){ref-type="fig"}\]. Using the same method as refered above[@b18], the measure trace was further interpolated into a 4096 × 4096 trace for retrieval as required by the discrete Fourier transform relation, and the result was presented in [Fig. 6(b)](#f6){ref-type="fig"}. The G error was 0.0123, which was not as good as that in [Fig. 5](#f5){ref-type="fig"}, it is because the complexity of pulses was largely increased in both time- and spectrum-bandwidths. In [Fig. 6(c,d)](#f6){ref-type="fig"}, we have plotted the retrieved pulse intensity and phase profile in temporal and spectral domains after 20-point smoothing, indicating the SC pulses with a \~55-ps time span \[[Fig. 6(c)](#f6){ref-type="fig"}\] and 670 nm \~ 970 nm in wavelength \[[Fig. 6(d)](#f6){ref-type="fig"}\]. It is obvious that the SC pulses were largely broadened temporally even more than the pump pulses measured above. It is reasonable because its wavelength band was expanded to more than 300 nm in the PCF, and most of it stood in the same dispersion regime as it did for pump pulses. Therefore, the chirp became more serious in the conducting fiber compared to the 20-nm pump pulses, and led to higher pulse duration. [Figure 6(e,f)](#f6){ref-type="fig"} present the images of NW without and with pulses imposing from each end of it (810 nm from left and SC from right). It could be noticed that almost no SH emission was generated by gate or target pulses themselves, which ensured a clean spectral background for pulse measurement. Discussion ========== In this work, we have implemented a nano-FROG method for ultrafast pulse characterization. Since the pulse modulation (stretching, compressing, etc.) by all the components have been involved and counted, the measured results of our method reveal the *in-situ* characteristics of the pulse, which means it could be a good solution to the measurement in fiber-based integrated systems. Compared with the previously reported FROG approaches, this fiber taper-NW coupling system exhibits several superiorities: (I) In our system, the target pulses are optically guided along NLO NW rather than spacially focused. Different from the free-space FROG, the majority of energy loss caused by the small duty ratio (the cross section ratio between the NW and focal spot), and by the scattering on the interface could be eliminated, which makes it more sensitive to low-energy pulses. More important, compared with the exsiting free-space nano-FROG methods, the light wave is guided along the NW, so the out-of-focus problem caused by wide-waveband dispersion can be avoided, permitting the measurement of SC pulses with broadband spectrum. While for the pulse disturbance from the waveguide system, it is predictable and thus can be compensated before measurement. (II) Since the signal was emitted from the NW surface, the direction of SF signal and input pulses are perpendicularly to each other, so the remnant fundamental wave will be spatially precluded. In most FROG approaches where SF and fundamental waves are propagating coaxially, optical filters are required to separate them spectrally. Compared with this, spatial separation by transverse emission has good advantage especially when a spectral overlap exists between fundamental and SF wave. Imagine the following situation, 800-nm gate pulses and SC pulses with 400 nm \~ 1000 nm band are cross-correlated. The SF wave will be mixed with the SC in the region of 400 nm \~ 444 nm, which will cause erroneous measurement results. A pre-filter for the blue part of the SC can avoid the mixing, but the SC pulses' profile is broken. In contrast, the spatial separation between signal and inputs makes our method facile to fulfill a complete measurement of broadband pulses. (III) Due to the nanometer scale of the NW, phase matching constraints can be relaxed. Thus the measurable wavelength band can be largely extended since all the frequency components can be resolved within the NLO material. The complexity brought in by an angle-dithered crystal is avoided, and the optical alignment does not have to be very precise in a waveguide based measurement, both of which result in a simplified system. However, if this method is used for free-space pulse characterization, it needs a pre-survey on the dispersion and nonlinearity of delivering fiber's, for instance, numerical simulations or a series of measurements with different fiber taper lengths have to be done. As a whole, this system is of great potential in integrating optics as an ultra-broadband optical device, and it also provides a promissing platform for intact observation of molecules dynamics and local field enhancement in future. Additional Information ====================== **How to cite this article**: Yu, J. *et al.* Frequency-resolved optical gating measurement of ultrashort pulses by using single nanowire. *Sci. Rep.* **6**, 33181; doi: 10.1038/srep33181 (2016). This work was supported in by China Postdoctoral Science Foundation (2015M581636), 973 Program (2015CB352001), National Natural Science Foundation of China (11304202 and 91221304), the Hujiang Foundation of China (D15014) and National Key Scientific Instrument Project (2012YQ150092). **Author Contributions** J.Y. and F.G. conceived and designed this work. F.L. and J.Y. performed experiments and analysed data. J.Y., F.G. and H.Z. co-wrote the manuscript. All authors discussed the results and commented on the manuscript. {#f1} {#f2} {#f3} {#f4} {#f5} {#f6}

1. Introduction {#sec0005} =============== Brown tumors (BT) are non-neoplastic, expansive bone lesions that occur only in the setting of hyperparathyroidism. It has been reported in 4.5% of the patients with primary hyperparathyroidism (PHPT) and in 1.5--1.7% of those with secondary hyperparathyroidism (SHPT) \[[@bib0005]\]. The most usual localization of BT is in mandible, ribs and large bones, are very rare in the spine, and in cervical spine there are few cases reported in literature \[[@bib0010], [@bib0015], [@bib0020]\]. To date there are only 11 cases of cervical BT reported globally \[[@bib0010],[@bib0025]\]. The aim of this work is to report a 25 year old female with end stage renal failure (ESRF) who developed neck pain without trauma history, finally was determinate the presence of BT in C4 vertebra. This work was reported in line with the SCARE criteria \[[@bib0030]\]. 2. Case report {#sec0010} ============== A 25-year-old woman presented with an history of 2 months of worsening cervicalgia without history of trauma. She complained about progressive neck pain with irradiation to both shoulders and right arm paresthesias. She denied gait alterations or other symptoms. Medical history of ESRF secondary to Wegener´s granulomatosis diagnosed 7 years ago. Current treatment with hemodialysis since the last 3 years. Secondary hyperparathyroidism was documented one year ago. Physical examination was normal. Laboratory tests at admission: leucocytes 5.50/mcl; hemoglobin 7.4 mg/dL; platelets 190 000/mcl; creatinine 8.57 mg/dL; ureic nitrogen 61 mg/dL; calcium 8.4 mg/dL; phosphorus 5.6 mg/dL; sodium 138 mmol/L; potassium 5.72 mmol/L; chlorine 91 mmol/L. The CT scan ([Fig. 1](#fig0005){ref-type="fig"}A--C). showed a lytic lesion on the fourth vertebral body of cervical spine (C4) and a displacement of underlying vertebrae and the posterior elements compromising the cervical canal and displacing the spine. In the magnetic resonance imaging (MRI) we could observe bone erosion of C4 and subluxation of underlying vertebrae with hyperintense images in spinal cord. ([Fig. 1](#fig0005){ref-type="fig"}D--F).Fig. 1Sagittal (A) and axial (B,C) computerized tomography scan demonstrating C4 vertebrae erosion in anterior and posterior components and subluxation of the underlying segments compromising the cervical canal. Sagittal (STIR) (D), axial (E) and coronal (F) T2 weighted MRI images showing the displacement of vertebrae and compression of the spinal cord.Fig. 1 It was managed with corpectomy of C4, decompression of cervical canal and the insertion of fixation system and four screws ([Fig. 2](#fig0010){ref-type="fig"}). After surgery the patient only developed transitory radicular type pain in C4,C5 territory of the right arm without another symptom.Fig. 2Transoperative images showing the titanium fixation system at site of C4 corpectomy.Fig. 2 Microscopic examination showed giant multinucleated cells with bone fragments and intertrabecular fibrosis and cartilage. Giant multinucleated osteoclast type cells mixed with fusiform cells and stroma hemosiderin deposits (Pearls). Ki-67 was positive in less than 1% of cells. ([Fig. 3](#fig0015){ref-type="fig"})Fig. 3Optic microscopy of fragments of C4 brown tumor. Hematoxylin and eosin. (A) Multinucleated giant cells with intertrabecular fibrosis and cartilage. (B) Giant multinucleated osteoclast type cell mixed with fusiform cells and fibrocollagenous tissue. (C) Pearls. Hemosiderin deposits in the lesion stroma.Fig. 3 3. Discussion {#sec0015} ============= The BT also known as osteoclastoma \[[@bib0015]\] is one of the multiple musculoskeletal complications of hyperparathyroidism secondary to ESRF and specially in patients who are in hemodyalisis. The most frequent localization is in the mandible, the ribs and finally in the large bones. The BT is due to a disorder in the bone resorption and a disorder in the metabolism of PTH and [d]{.smallcaps}-vitamin \[[@bib0010]\]. Cervical BT is extremely rare, with few reported cases globally, the first report was in 1993 \[[@bib0035]\] and there are only 12 cases reported including this to the date. The importance of the prompt diagnosis of the BT is to establish a multidisciplinary management to prevent progression, neurologic complications and sequelae (radiculopathy, myelopathy) despite its slow progression and a certainly benign behavior. The initial suspicion of this disease must be since the first clinician contact of the patient and with the past medical history of ESRF plus recent neurologic manifestations. The aim of neurosurgical management of these patients is to promote spinal stability and release spinal cord and nerve roots to eliminate risk of neurological deficits \[[@bib0040]\]. The mainstay in the treatment or prevention of occurrence of a BT is the elimination of the underlying metabolic disorder with parathyroidectomy \[[@bib0040]\]. We report the case number 12 of cervical BT in the literature. The knowledge about this rare disease is essential to an adequate diagnostic approach and treatment. Conflict of interest {#sec0020} ==================== We disclose about any financial and personal relationships with other people or organisations that could inappropriately influence (bias) their work. Funding {#sec0025} ======= We declare that we do not received any funding for this research. Ethical approval {#sec0030} ================ We do not require ethical approval to write a case report paper. Consent {#sec0035} ======= Written informed consent was obtained from the patient for publication of this case report and accompanying images. Author contribution {#sec0040} =================== 1\. Mauricio Daniel Sánchez-Calderón. Responsibilities: the conception and design, data acquisition, analysis of data, drafting of the manuscript, critical revision. 2\. Diego Ochoa-Cacique. Responsibilities: data acquisition, analysis of data, critical revision. 3\. Oscar Medina Carrillo. Responsibilities: data acquisition, analysis of data, critical revision. 4\. Ulises García González. Neurosurgeon. Responsibilities: the conception and design, critical revision. 5\. Rosa María Vicuña González. Responsibilities: data acquisition, analysis of data, critical revision. 6\. Carlos Cesar Bravo Reyna. Responsibilities: data acquisition, analysis of data, critical revision. 7\. José Raúl Guerra-Mora. Responsibilities: the conception and design, (data acquisition, (analysis of data, drafting of the manuscript, critical revision. Registration of research studies {#sec0045} ================================ Case reports not need to be registered. Guarantor José Raúl Guerra Mora. Provenance and peer review {#sec0050} ========================== Not commissioned, externally peer-reviewed.

1. Introduction =============== The genus *Valeriana*, from the family *Valerianaceae*, consists of about 250 species widely distributed all over the World. *Valeriana* is a perennial herb native to Europe, Asia and North America \[[@B1-molecules-18-14138],[@B2-molecules-18-14138]\]. Eleven out of 28 *Valeriana* genus species (including one variant) found in China are traditionally used as medicines. Their dried underground parts (roots and rhizomes) exhibit anodyne, antiphlogistic, expectorant and antiasthmatic activities \[[@B3-molecules-18-14138],[@B4-molecules-18-14138],[@B5-molecules-18-14138]\]. Valerians have been used clinically as tranquillizers for the treatment of nervousness, agitation and as a mild sedative to improve sleep \[[@B6-molecules-18-14138],[@B7-molecules-18-14138],[@B8-molecules-18-14138],[@B9-molecules-18-14138],[@B10-molecules-18-14138],[@B11-molecules-18-14138],[@B12-molecules-18-14138],[@B13-molecules-18-14138]\]. Previous studies reported that *Valeriana* contains numerous chemical constituents, including volatile oil, iridoids, flavones, alkaloids, amino acids, and lignans, *etc*. \[[@B14-molecules-18-14138],[@B15-molecules-18-14138],[@B16-molecules-18-14138],[@B17-molecules-18-14138],[@B18-molecules-18-14138],[@B19-molecules-18-14138],[@B20-molecules-18-14138]\]. Valerian (*Valeriana officinalis Linn*.) has been included in the pharmacopoeias in Europe and the United States \[[@B21-molecules-18-14138],[@B22-molecules-18-14138]\], its extracts are sold as dietary supplements and were listed among the top 10 best-selling herbal supplements in the United States in 2002 \[[@B23-molecules-18-14138]\]. Valerian is also widely used as a precious perfume added into food, drink, cosmetics, tobacco, and so on. With significant medicinal and commercial value, it will surely deserve future development. *V. officinalis* var. *latiofolia*, a variant of *V. officinalis Linn*, is produced mainly in the Guizhou, Sichuan and Yunnan provinces in Southwest China. Particularly in Guizhou, *V. officinalis* var. *latiofolia* is widely cultivated and has become one of the local industry pillars \[[@B1-molecules-18-14138],[@B24-molecules-18-14138]\]. This variant shares some common pharmaceutical actions and chemical constituents with *V. officinalis Linn* \[[@B25-molecules-18-14138]\]. Abundant research has been done on *V. officinalis* var. *latiofolia* so far, and it is assumed that the sesquiterpenoids from the volatile oil and iridoids are the main contributors to its antidepressant and antinervousness activities \[[@B24-molecules-18-14138],[@B25-molecules-18-14138]\]. Pairs of active compounds have been isolated and identified from *Valeriana*, including germacrane-type sesquiterpenoids, volvalerenals A--E, volvalerenic acids A--C, valerianin A--B and heishuixiecaoline A--C \[[@B13-molecules-18-14138],[@B16-molecules-18-14138],[@B17-molecules-18-14138],[@B18-molecules-18-14138],[@B19-molecules-18-14138]\]. However, neither the active component(s) responsible for the therapeutic properties of *Valeriana* nor the related molecular mechanisms are clearly understood, which severely hinders the wider application of *Valeriana* products. Therefore, in this paper the chemical constituents of *V. officinalis* var. *latiofolia* have been systematically investigated, and eight germacrane-type sesquiterpenoids (including three new compounds, volvalerenal F (**1**), volvalerenal G (**2**) and volvalerenic acid D (**3**)) were isolated and identified from chloroform extracts of this herb's syrup. Additionally, the NGF-potentiating activities of the obtained products were evaluated. 2. Results and Discussion ========================= Volvalerenal F (**1**) was isolated as a colorless oil. High-resolution mass spectrometry (HR-ESI-MS) (*m/z* 271.1678 \[M+Na\]^+^, calcd. for C~16~H~24~O~2~Na, 271.1669) of **1** indicated that its molecular formula is C~16~H~24~O~2,~ and together with the NMR data ([Table 1](#molecules-18-14138-t001){ref-type="table"}), implied five unsaturated degrees in the molecule. The IR spectrum indicated the presence of carbonyl (1730 cm^−1^) and carbon-carbon double-bond (1627 cm^−1^) absorptions. The ^1^H-NMR spectrum of **1** ([Table 1](#molecules-18-14138-t001){ref-type="table"}) displayed three olefinic protons at δ~H~ 5.27 (1H, dd, *J* = 5.4, 10.2 Hz, H-1), 5.34 (1H, ddd, *J* = 9.0, 9.0, 3.2 Hz, H-4), and 5.21 (1H, dd, *J* = 9.0, 9.0 Hz, H-5) indicating the presence of two double bonds, methylene proton peaks at δ~H~ 4.42 (1H, d, *J* = 12.0 Hz, H-14a) and δ~H~ 4.21 (1H, d, *J* = 12.0 Hz, H-14b), and three methyls at δ~H~ 0.98 (3H, s, 13-CH~3~), 1.04 (3H, s, 12-CH~3~) and 2.01 (3H, s, 16-CH~3~). Considering the DEPT spectra, the ^13^C-NMR spectrum of **1** ([Table 1](#molecules-18-14138-t001){ref-type="table"}) suggested the existence of an acetate carbonyl carbon at δ~C~ 171.3 (C-15), four olefinic carbons at δ~C~ 128.8 (C-1), 133.1 (C-10), 128.3 (C-4) and 129.6 (C-5), as well as five methylenes at δ~C~ 28.2 (C-2), 26.8 (C-3), 23.0 (C-8), 35.3 (C-9) and 61.7 (C-14), two methines at δ~C~ 26.4 (C-6) and 31.9 (C-7) and three methyls at δ~C~ 15.6 (C-13), 21.1 (C-16) and 28.9 (C-12). molecules-18-14138-t001_Table 1 ###### ^1^H and ^13^C-NMR Spectroscopic Data of Compounds **1**--**3**. NO. 1 2 3 ----- ------- ---------------------------- ------- ---------------------------- ------- ---------------------------- 1 128.8 5.27 *dd* (5.4, 10.2) 128.2 5.22 *dd* (5.4, 11.4) 130.7 5.33 *dd* (5.4, 11.4) 2a 28.2 2.23 *m* H-α 28.3 2.22 *m* (H-α) 29.2 2.38 *m* (H-β) 2b 2.11 *m* H-β 2.15 *m* (H-β) 2.26 *m* (H-α) 3a 26.8 2.14 *m* H-β 24.0 2.71 *dt* (12.0, 4.2, H-α) 27.1 2.71 *dt* (12.6, 4.0, H-α) 3b 2.08 *m* H-α 2.05 *td* (12.6, 4.2, H-β) 2.17 *td* (12.6, 4.0, H-β) 4 128.3 5.34 *ddd* (9.0, 9.0, 3.2) 143.9 132.1 5 129.6 5.21 *dd* (9.0, 9.0) 158.9 6.48 *d* (9.0) 144.8 6.72 *d* (9.6) 6 26.4 1.14 *t* (9.0, H-α) 31.2 1.52 *t* (13.2, 9.0, H-α) 30.0 1.27 *dd* (9.6, 4.0, H-α) 7 31.9 0.47 *m* (H-α) 38.8 1.02 *m* (H-α) 36.7 0.82 *m* (H-α) 8a 23.0 1.73 *m* (H-β) 24.6 1.86 *m* (H-β) 24.7 1.81 *m* (H-β) 8b 0.76 *m* (H-α) 0.90 *m* (H-α) 0.89 *m* (H-α) 9a 35.3 2.39 *m* (H-β) 35.5 2.60 *m* (H-β) 36.2 2.36 *m* (H-β) 9b 1.87 *t* (12.6, α) 1.88 *m* (H-α) 2.25 *m* (H-α) 10 133.1 139.1 134.3 11 17.2 22.4 21.5 12 28.9 1.04 *s* 28.7 1.16 *s* 28.9 1.12 *s* 13 15.6 0.98 *s* 16.0 1.20 *s* 16.1 1.14 *s* 14a 61.7 4.42 *d* (12.0) 196.4 9.20 *s* 172.9 14b 4.21 *d* (12.0) 15 171.3 59.1 3.69 *d* (12.0) 62.8 4.32 *d* (12.0) 3.43 *d* (12.0) 4.15 *d* (12.0) 16 21.1 2.01 *s* 172.0 17 20.8 1.97 *s* Compound **1** recorded in CDCl~3**.**~ Compounds **2** and **3** recorded in CD~3~OD. ^1^H-NMR recorded at 600 MHz. ^13^C-NMR recorded at 150 MHz. To confirm the structure of **1**, ^1^H-^1^H COSY and HMBC experiments were conducted ([Figure 1](#molecules-18-14138-f001){ref-type="fig"}), which showed the key correlations such as H-1/H-2, H-2/H-3, H-3 (a,b)/H-4, H-4/H-5, H-5/H-6, H-6/H-7, and H-8/H-9 in its COSY, H-14/C-1, H-14/C-9, H-14/C-10, H-14/C-15, H-3/C-4, H-5/C-4 and H-12/C-6, H-12/C-7, H-12/C-11 and H-12/C-13 in its HMBC. The coupling constant of 9.0 Hz between H-4 and H-5 indicated the *Z-*configuration of the double bond \[[@B26-molecules-18-14138],[@B27-molecules-18-14138]\]. The coupling constant of 9.0 Hz and the NOE correlations between H-6 and H-7 suggested the *syn* and α-oriented of the cyclopropane moiety, and the α-orientation of H-6 and H-7 were assigned by the correlations of H-7/CH~3~-12 and H-6/CH~3~-12 \[[@B28-molecules-18-14138]\]. The correlations of H-2(b)/H-15(a, b) indicated Δ^1,10^ to be *Z* configured. From the above data, the structure of **1** was identified as 14-acetoxy-11,11-dimethylbicyclo\[8.1.0\]undeca-4*Z* (5),10*Z* (1)-diene, and the compound was named volvalerenal F. {#molecules-18-14138-f001} Compound **2** was isolated as a colorless oil. The molecular formula was assigned as C~15~H~22~O~2~ on the basis of HR-ESI-MS from the \[M+H\]^+^ signal at *m/z* 235.1696 (calcd. for C~15~H~23~O~2~, 235.1693), with five degrees of unsaturation. The IR spectra displayed the presence of carbonyl (1726 cm^−1^), α,β-unsaturated aldehyde (1678 cm^−1^), and carbon-carbon double-bond (1626 cm^−1^) absorptions. The ^13^C-NMR and DEPT spectra of **2** ([Table 1](#molecules-18-14138-t001){ref-type="table"}) showed an aldehyde carbon at δ~C~ 196.4 (C-14), four double bond carbons at δ~C~ 128.2 (C-1), 143.9 (C-4), 158.9 (C-5) and 139.1 (C-10), as well as five methylenes (one oxygenated) at δ~C~ 24.0 (C-3), 24.6 (C-8), 28.3 (C-2), 35.5 (C-9) and 59.1 (C-15), two methines at δ~C~ 31.2 (C-6) and 38.8 (C-7), and two methyls at δ~C~ 16.0 (C-13) and 28.7 (C-12). These data suggested that compound **2** was also a germacrane-type sesquiterpenoid. Its ^1^H-NMR spectrum displayed an aldehydic proton obviously at δ~H~ 9.20 (1H, s, H-14), an oxygenated methine proton at δ~H~ 3.69 (1H, d, *J* = 12.0 Hz, H-15a) and 3.43 (1H, d, *J* = 12.0 Hz, H-15b). Above data indicated **2** was structurally similar to heishuixiecaoline B reported in the literature \[[@B13-molecules-18-14138]\]. The proposed structure was further confirmed by HMBC correlations ([Figure 2](#molecules-18-14138-f002){ref-type="fig"}). Key long-range correlations were observed between H-12/C-6, H-12/C-7, H-12/C-11, H-12/C-13, H-15/C-1, H-15/C-10, H-15/C-9, and H-14/C4, H-14/C-5, which suggested the aldehyde group and hydroxyl group were located to C-4 and C-15, respectively. The α-orientation of H-6 and H-7 were assigned as being the same as those of **1** by the correlations of H-6/H-7, H-7/CH~3~-12 and H-6/CH~3~-12 in the NOESY experiment ([Figure 4](#molecules-18-14138-f004){ref-type="fig"}), and the *E*- and *Z*- configurations of Δ^4,5^ and Δ^1,10^ were determined to be the same as in compound **1** by the correlations of H-5/H-14, H-2 (a, b)/H-15 and H-5 with H-1. {#molecules-18-14138-f002} On the basis of the above data, the structure of **2** was identified as 4-formyl-10-hydroxymethyl-11, 11-dimethylbicyclogermacren-4*E* (5), 10*Z* (1)-diene, and the product was named volvalerenal G. Volvalerenic acid D (**3**) was also isolated as a colorless oil. The HR-ESI-MS of **3** indicated that its molecular formula is C~17~H~24~O~2~ (*m/z* 293.1744 \[M+H\]^+^, calcd. for 293.1747). The IR data was completely similar to that of compound **1**, which suggested that **3** was also a germacrane-type sesquiterpenoid. The NMR spectrum of compound **3** ([Table 1](#molecules-18-14138-t001){ref-type="table"}) showed the following signals: an acetate carbonyl carbon, four olefinic carbons, five methylenes (one oxygenated), two methines and three methyls, which were quite similar to those of compound **1**. In addition, it is noteworthy that the NMR data displayed an obvious carboxyl carbon signal at δ~C~ 172.0 (C-16). The ^1^H-^1^H COSY spectrum showed key correlations such as H-6/H-7 and H-8/H-9, and key long-range correlations were observed in the HMBC experiments between H-3/C-14 and H-5/C-14 ([Figure 3](#molecules-18-14138-f003){ref-type="fig"}), which suggested the carboxyl group was located to C-4. {#molecules-18-14138-f003} The NOESY correlations ([Figure 4](#molecules-18-14138-f004){ref-type="fig"}), the coupling constant of 4.0 Hz and the NOE correlations between H-6 and H-7 suggested a *trans* geometry around the cyclopropane ring, and the correlations of H-6/H-3a, H-2 (a, b)/H-15 suggested the *E*- and *Z*- configuration of Δ^4,5^ andΔ^1,10^. {#molecules-18-14138-f004} From the above data, the structure of **3** was identified as 15-acetoxy-4-carboxy-11, 11-dimethyl-bicyclogermacren-4*E* (5), 10*Z* (1)-diene, and it compound was named Volvalerenic acid D. The five known compounds were identified as madolin A (**4**) and B (**7**) \[[@B26-molecules-18-14138]\], vovalerenal A (**6**) and B (**5**) \[[@B18-molecules-18-14138]\] and heishuixiecaoline B (**8**) \[[@B13-molecules-18-14138]\] by comparing their NMR spectroscopic data with the literature values. The structures of compounds **1**--**8** are shown in [Figure 5](#molecules-18-14138-f005){ref-type="fig"}. {#molecules-18-14138-f005} The propensity of compounds **1**--**8** to enhance the activity on nerve growth factor (NGF)-mediated neurite outgrowth in PC 12D cells was assessed as described previously \[[@B25-molecules-18-14138]\]. The neurite-bearing cells accounted for 22.74% and 100% in the control experiments incubated with 2 and 50 ng/mL NGF after 48 h, respectively. Under 2 ng/mL NGF conditions, all eight tested compounds (at 10, 30, 100 µmol) showed NGF-potentiating activities in various levels. Compound **3** (at 100 µmol) reached 50.15%, in particular ([Table 2](#molecules-18-14138-t002){ref-type="table"}). molecules-18-14138-t002_Table 2 ###### Effects of compounds **1**--**8** on the proportion of neurite-bearing PC 12D cells in the presence of NGF. Compound NGF(ng/mL) Cell viability (%) ---------- ------------ -------------------- ------------------ ------------------ **1** 2 24.85 ± 0.98 33.97 ± 1.77 ^b^ 43.61 ± 2.11 ^c^ **2** 2 24.42 ± 1.12 35.34 ± 1.48 ^b^ 44.30 ± 1.85 ^c^ **3** 2 26.48 ± 0.89 ^a^ 37.51 ± 1.66 ^b^ 50.15 ± 2.23 ^c^ **4** 2 23.50 ± 1.26 29.33 ± 0.88 ^b^ 33.87 ± 1.63 ^c^ **5** 2 24.37 ± 1.01 30.12 ± 1.97 ^b^ 40.79 ± 1.17 ^c^ **6** 2 24.74 ± 1.47 34.84 ± 2.35 ^b^ 44.15 ± 2.19 ^c^ **7** 2 24.45 ± 1.01 33.96 ± 1.13 ^b^ 46.69 ± 2.14 ^c^ **8** 2 24.26 ± 0.73 33.79 ± 1.17 ^b^ 49.25 ± 1.25 ^c^ 2 23.08% ± 1.28 50 100% 0 3.12% ± 0.88 Results expressed as mean ± SD (n = 6) of eight independent experiments. The 100% value was obtained from 50 ng/mL NGF in the absence of compounds. A statistically significant difference (a, b and c, *p* \< 0.01) from the control (2 ng/mL NGF) in the absence of compounds was apparent. 3. Experimental =============== 3.1. General ------------ Optical rotations were measured with a Perkin-Elmer 343 polarimeter (Perkin-Elmer, Waltham, MA, USA). IR spectra were recorded on the Bio-Rad FTS-65A spectrometer (Bio-Rad, Richmond, VA, USA). UV spectra were recorded using the UV-2501PC spectromter (Shimadzu, Japan). ^1^H and ^13^C-NMR spectra were obtained on a JNM-ECS400 MHz spectrometer (JEOL, Tokyo, Japan) and a Varian UNITY INOVA 600 spectrometer (Varian, Palo Alto, CA, USA), and the chemical shifts were given on δ (ppm) scale with TMS as an internal standard. The HR-ESI-MS were recorded on a 9.4-TQ-FT-MS Apex Qe (Bruker Co., Billerica, MA, USA). Silica gel (60--120 mesh, 200--300 mesh, Qingdao Marine Chemical Group Co., Qingdao, China), and Sephadex LH-20 (Pharmacia, Uppsala, Sweden) were employed for column chromatography. HPLC was carried out using Waters 600E system (Waters, Milford, MA, USA): an analytical column, ODS (5 µm, 4.6 × 250 mm, Hanbon Science & Technology Co., Ltd, Huaian, China), preparative column, a YMC C18 (5 µm, 20.0 × 250 mm, YMC, Kyoto, Japan), detector, Alltech ELSD (evaporative lightscattering detector, Alltech, Los Angeles, CA, USA) 2000ES. Flash chromatography was carried out on Teledyne ISCO Combi Flash *Rf* with Prepacked 80 g silica gel (200--300 mesh) columns (Teledyne Isco, Lincoln, NE, USA). TLC was carried out using silica gel 60 (\>230 mesh, Qingdao Marine Chemical Group Co.) and GF254 plates precoated with silica gel 60. Spots on TLC were visually observed under UV light and/or by spraying with anisaldehyde-H~2~SO~4~ reagent followed by heating. 3.2. Plant Material ------------------- The dry roots of *V. officinalis* var. *latifolia* were collected from the Jiangkou region of Guizhou Province, China, in April 2012. The plant was identified by Prof. Bin Li (Beijing Institute of Radiation Medicine), and a voucher specimen (KYXC-20120313) is deposited in the herbarium of the Beijing Institute of Radiation Medicine, Beijing, China. 3.3. Extraction and Isolation ----------------------------- The air-dried roots of *V. officinalis* var. *latifolia* (50 kg) were exhaustively refluxed three times with 60% EtOH (400 L) to give a residue (11 kg) after removal of solvent under reduced pressure. The EtOH extract was suspended in H~2~O and then partitioned successively with CHCl~3~ (3 × 10 L). The CHCl~3~ extract (152.2 g) was subjected to silica gel (200--300 mesh) column chromatography, eluted with petroleum ether-acetone (from 100:1, 75:1, 50:1, 30:1, 25:1, 20:1, 15:1, 10:1, 7:1, 5:1, 3:1, 2:1 and 1:1, v/v) to afford thirteen fractions (A--M). Fraction E (4.711 g) was subjected to flash silica gel chromatography column (80 gram flash column, 60 m L/min) with CHCl~3~/CH~3~OH (from 50:1 to 10:1) to yield eight fractions, E1-E8. Fractions E2-E3 (0.225g) was chromatographed by Sephadex LH-20 (2 × 120 cm, CHCl~3~/CH~3~OH, 1:1) to obtain compound **1** (33 mg). Fractions E4-E7 (0.818 g) was subjected to flash silica gel chromatography (40 gram flash column, petroleum ether-acetone, 5:1, 30 mL/min) and purified by Sephadex LH-20 (2 × 120 cm, CHCl~3~/CH~3~OH, 1:1) to afford compound **6** (86 mg). Compound **4** (168 mg) was isolated from fration E8 by a series of repeated Sephadex LH-20 (2 × 150 cm, CHCl~3~-CH~3~OH, 1:1) column chromatography fractionations. Fraction I (3.678 g) was separated chromatographically on flash silica gel column (80 gram flash column, 60 mL/min) with CHCl~3~/CH~3~OH (40:1 to 2:1), and a total of 50 tubes (15 mL each) were collected. Tubes 19--40 (2.245 g) were chromatographed by flash silica gel chromatography (40 gram flash column, petroleum ether-acetone, 4:1, 30 mL/min) to obtain compound **8** (tubes 10-13, 78 mg). Fraction J (6.366 g) was subjected to a series of purification steps using flash silica-gel column chromatography (80 gram flash column, CHCl~3~/CH~3~OH, 20:1 to 1:1, 60 mL/min) to give ten fractions (J1-J8). Fraction J2-J4 (0.167 g) was chromatographed by Sephadex LH-20 column chromatography (2 × 150 cm, CH~3~OH), and purified by preparative HPLC (CH~3~OH/H~2~O, 75:25, flow rate: 2.0 mL・min^−1^) to afford compounds **2** (68 mg) and **5** (35 mg). Compounds **3** (31 mg) and **7** (78 mg) were obtained from fraction J6-J8 (0.167 g) by preparative HPLC (CH~3~OH-H~2~O, 50:50, flow rate: 2.0 mL・min^−1^). 3.4. Compound Characterization ------------------------------ *Volvalerenal F* (**1**): colorless oil.  +20.0 (c 0.8, CHCl~3~); UV (CHCl~3~) λ~max~ 237 nm; IR (film) *ν~max~* 3445, 3171, 2960, 2924, 2852, 1730, 1627, 1261, 1095, 1024 cm^−1^; ^1^H-NMR (CHCl~3~, 600 MHz) data, see [Table 1](#molecules-18-14138-t001){ref-type="table"}; ^13^C-NMR (CHCl~3~, 150 MHz) data, see [Table 1](#molecules-18-14138-t001){ref-type="table"}; HR-ESI-MS *m/z* 271.1678 \[M+Na\]^+^ (calcd. for C~16~H~24~O~2~Na, 271.1669). *Volvalerenal G* (**2**): colorless oil.  +53.9 (c 0.1, MeOH); UV (CHCl~3~) λ~max~ 264 nm; IR (film) *ν~max~* 3382, 2933, 2864, 1618, 1298, 1190, 1072, 921, 793 cm^−1^; ^1^H-NMR (CHCl~3~, 600 MHz) data, see [Table 1](#molecules-18-14138-t001){ref-type="table"}; ^13^C-NMR (CHCl~3~, 150 MHz) data, see [Table 1](#molecules-18-14138-t001){ref-type="table"}; HR-ESI-MS *m/z* 235.1696 \[M+H\]^+^ (calcd. for C~15~H~23~O~2~, 235.1693). *Volvalerenic Acid D* (**3**): colorless oil.  +8.5 (c 0.47, MeOH); UV (CHCl~3~) λ~max~ 240 and 271 nm; IR (film) *ν~max~* 3384, 3245, 2931, 2862, 1238, 1118, 1027, 862, 768 cm^−1^; ^1^H-NMR (CHCl~3~, 600 MHz) data, see [Table 1](#molecules-18-14138-t001){ref-type="table"}; ^13^C-NMR (CHCl~3~, 150 MHz) data, see [Table 1](#molecules-18-14138-t001){ref-type="table"}; HR-ESI-MS *m/z* 293.1744 \[M+H\]^+^ (calcd. for C~17~H~25~O~2~, 293.1747). 3.5. Activity Screening *in Vitro* ---------------------------------- PC 12D cell line was obtained from Insitute of Biochemistry and Cell Biology, CAS. It was cultured in Dulbecco's modified Eagles Medium (DMEM, Gibco, New York, NY, USA) with 10% fetal calf serum (Gibco, New York, NY, USA), and 5% equine serum (Gibco), and then the cells were maintained at 37.0 °C in a humidified atmosphere which contained 6% CO~2~ \[[@B25-molecules-18-14138]\]. The test cell line was seeded in 24-well culture paltes (2 × 10^4^ cells/well) coated with poly-L-lysine (Gibco). After 24 h, the medium was changed to test medium containing 1% fetal calf serum, 2% equine serum and varying concentrations of NGF (50 ng/mL for positive control, 2 ng/mL for test samples and significant difference control, Sigma, St. Louis, MO, USA) and test compounds 1--8 (10, 30, 100 µmol). After incubating for 48 h, the cells were fixed with 1% glutaraldehyde (Sigma) in phosphate buffer, and the cells with neurites outgrowth were counted (with at least 100 cells examined/viewing area, three viewing areas/well, six wells/sample). The ratio of the neurite-bearing cells to total cells was determined and expressed as a percentage. 4. Conclusions ============== Three new germacrane-type sesquiterpenoids, volvalerenal F (**1**), volvalerenal G (**2**) and volvalerenic acid D (**3**), along with five known compounds **4**--**8**, were isolated from the CHCl~3~ soluble partition of the ethanol extract of *Valeriana officinalis* var. *latiofolia*. NGF plays a key role in the functions of the central and peripheral nervous system \[[@B29-molecules-18-14138]\] and all the sesquiterpenoids obtained displayed certain NGF-potentiating activities. From the current investigation it can be predicted that sesquiterpenoids will be promising candidates for dietary supplements and medicines, although further studies are needed to determine the pharmacological activities and the mechanism of these eight compounds in animals. We are grateful to Yan Xue, Mei-feng Xu and Yu-mei Zhao of the National Center of Biomedical Analysis for the measurements of the MS and NMR spectra. *Sample Availability*: Samples of the compounds **1**--**8** are available from the authors. The authors declare no conflict of interest. [^1]: These authors contributed equally to this work.

1. Introduction {#sec1-sensors-20-02063} =============== Digital terrain models (DTMs) are effective and important tools of environmental investigations, engineering, and planning \[[@B1-sensors-20-02063],[@B2-sensors-20-02063],[@B3-sensors-20-02063]\]. DTMs are often used for the management of natural risks, e.g., assessments of inundation exposure or volcanic active areas, especially if these areas are populated and involve infrastructure \[[@B4-sensors-20-02063],[@B5-sensors-20-02063],[@B6-sensors-20-02063]\] There are several ways to produce these models, such as interpolating surfaces from surveyed field data or vectorized contours of maps and using the principles of stereo photogrammetry (airborne and satellite), SfM (structure from motion) technique; the most dynamically developing technique is the application of airborne LiDAR/ALS (LiDAR---light detection and ranging; ALS---airborne laser scanning), which provides a three-dimensional point cloud stored in binary LAS (LiDAR archive standard) format \[[@B7-sensors-20-02063],[@B8-sensors-20-02063],[@B9-sensors-20-02063],[@B10-sensors-20-02063],[@B11-sensors-20-02063],[@B12-sensors-20-02063],[@B13-sensors-20-02063],[@B14-sensors-20-02063],[@B15-sensors-20-02063]\]. The ALS technique has the potential to collect multilayer data including the ground points if laser beams can reach the bare earth: emitted beams have echoes (reflections or discrete returns) from the top of the objects as the "first echo" and from the ground as the "last echo", and of course there are internal echoes as well \[[@B16-sensors-20-02063],[@B17-sensors-20-02063]\]. However, the last echoes do not always reach the ground; consequently, generated terrain models can have a bias \[[@B18-sensors-20-02063]\]. In spite of the difficulties, there could be several ways to filter out the noise and the ground points of a three-dimensional point cloud before DTM generation; therefore, the final results can be improved and can be used for most purposes. Noise is considered to consist of outliers which have different characteristics than the neighboring points, i.e., supposing a locally planar area (i.e., kernel window) defined by the average distance from a center or considering its *k* neighbors, when outlying points fall outside it. Noise filtering can be based on principal component analysis \[[@B19-sensors-20-02063]\], neighborhood distance \[[@B20-sensors-20-02063],[@B21-sensors-20-02063]\], or distance from surface \[[@B22-sensors-20-02063]\]. Ground point filtering is also a crucial point of data preparation \[[@B13-sensors-20-02063]\]. Several types of filtering algorithms have been developed for the extraction of ground points automatically. Some algorithms also apply the kernel approach and compare points to their neighbors (binning) assuming that the minimum values of the kernel window represent the ground \[[@B23-sensors-20-02063]\] or that slopes cannot exceed a given angle value within the kernel \[[@B24-sensors-20-02063]\]. Besides the fixed kernel windows, other robust methods exist, including the iterative multiscale spline, which was developed directly for densely forested areas \[[@B25-sensors-20-02063]\]. Furthermore, \[[@B26-sensors-20-02063]\] used a progressive morphology, \[[@B27-sensors-20-02063]\] used a segmentation-based robust interpolation, and \[[@B28-sensors-20-02063]\] used a combination of a multilevel adaptive filter (MAF) with morphological reconstruction and a thin plate spline (TPS) interpolation algorithm for the classification procedure to extract more precisely the bare earth. In addition, nowadays, there is an increasing number of easy-to-use algorithms which are implemented in freely available software to help for the users with the filtering process. For example, \[[@B29-sensors-20-02063]\] has promoted a new method, the cloth simulation filter (CSF), implemented in many open-source software, e.g., Python \[[@B30-sensors-20-02063]\] and CloudCompare \[[@B31-sensors-20-02063]\], which extracts ground points by simulating a physical process in which a virtual cloth covers the inverted point cloud. Its advantages are that it can be used for various landscapes and the parameters are easy to set. Floodplains represent a complex environment as they are important from the perspective of flood management, agricultural production, and nature conservation \[[@B32-sensors-20-02063],[@B33-sensors-20-02063],[@B34-sensors-20-02063]\]. An accurate DTM can be exploited to delineate trajectories of moving water during high and retreating floods, to find appropriate places for ploughing or animal grazing, to identify environmental conditions which provide good habitats for valuable species, or even to help with the detection of heavy metal hotspots \[[@B2-sensors-20-02063],[@B4-sensors-20-02063],[@B35-sensors-20-02063],[@B36-sensors-20-02063],[@B37-sensors-20-02063],[@B38-sensors-20-02063]\]. However, the most important locations of floodplains are usually impervious places with dense vegetation combined with permanent or periodical water cover (swales, oxbow lakes, peats, or marshes) \[[@B39-sensors-20-02063]\]. Accordingly, field surveys cannot be effective; furthermore, photogrammetry also has problems due to the vegetation cover and its ability to produce only a digital surface model (DSM) \[[@B40-sensors-20-02063],[@B41-sensors-20-02063]\]. The ALS technique is also limited, as beams do not always reach the ground due to dense vegetation, and water cover on the landforms will absorb the emitted light. Accordingly, point clouds always have a bias in a complex environment where the vegetation and/or the topography impede the penetration of laser beams to the ground. However, despite these issues, this is the most effective method of data collection in these areas \[[@B42-sensors-20-02063],[@B43-sensors-20-02063]\], and pre- and postprocessing and different classification procedures can mitigate the errors \[[@B44-sensors-20-02063]\]. The floodplain of a typical meandering river is usually characterized by concave or convex shapes, i.e., deeper or higher terrains with specific characteristics known as swales and point bar series \[[@B45-sensors-20-02063],[@B46-sensors-20-02063]\]. Swales are shallow concave forms often covered with herbaceous (e.g., reed) or aquatic vegetation (e.g., common duckweed) \[[@B47-sensors-20-02063]\]. Point bars are the antonyms of swales; they have a convex shape and are higher. In the floodplain of a medium-sized river, the height differences are relatively small (1--5 m), but a swale--point bar series usually has a relative difference of 0.5--1 m. This small relief highlights the main problem of the surveys: a small error has relevant consequences in the final terrain model and in all the extracted secondary information. Therefore, digital terrain modeling is a great challenge in this flat environment as small errors are conspicuous and fluvial forms become distorted. Common methods which fit for hilly and mountainous areas can fail, and postprocessing (e.g., filtering of the DTM itself, sink fill) can be misleading, diminishing real phenomena (even the forms themselves). In this study, we aimed to reveal the efficiency of different denoising approaches and the fine-tuning of CSF as a ground point classification technique. We intended to conduct LAS data processing in open-source environment and to use the most widespread algorithms. Although there are several research studies presenting details and suggestions concerning the best practice to generate DTMs \[[@B13-sensors-20-02063],[@B48-sensors-20-02063],[@B49-sensors-20-02063]\], there has been no comprehensive analysis of how the multiple factors of noise filtering, ground point classification, interpolation techniques, DTM resolution, and fluvial geomorphology together influence the accuracy of the models. Finding the DTM the most precisely reflects the terrain characteristics is crucial in such a flat environment where even a centimeter-scale error can change the topography and therefore change the determination of the waterflow direction (flood risk management), sediment accumulation (floodplain land use management) or the identification of fluvial forms. In this work, our aim was to evaluate and quantify the differences of point cloud classification algorithms and to compare the resulting DTMs. As ALS provides a high density of information about the terrain, deterministic interpolation methods were used and compared here. Our hypotheses were the following: (1) ground points have a significant effect on the generated DTMs, and filtering methods decrease the errors to a relevant degree; (2) ground point identification is highly dependent on the cloth size and the threshold parameters of CSF; (3) interpolation techniques and the grid size can enhance or smooth the errors of the DTMs; (4) different morphological forms, in our case swales and point bars, have a significant effect on ground point density. 2. Study Area and Topographic Characterization {#sec2-sensors-20-02063} ============================================== The study area is situated near to the town of Rakamaz, in NE Hungary (the coordinates of its corners are as follows: upper left 48°7'6.8226" N 21°26'37.1286" E, upper right 48°7'6.3942" N 21°28'24.8376" E, lower right 48°6'48.0528" N 21°28'24.672" E, lower left 48°6'48.4806" N 21°26' 36.9702" E), in the floodplain of the Tisza River ([Figure 1](#sensors-20-02063-f001){ref-type="fig"}a). It covers approximately 1 km^2^ and is characterized by point bar and swale series ([Figure 1](#sensors-20-02063-f001){ref-type="fig"}b, [Figure 2](#sensors-20-02063-f002){ref-type="fig"}a), which has remained as a consequence of the continuous lateral movement of the former Tisza River bed. It was selected due to its diverse environment. The widths of these landforms are various, mostly ranging between 10 and 30 m, but there are some narrower ones (3--5 m), and some are really well spread, with a width of more than one hundred meters. Differences in terrain height in most of the cases are less than one meter (0.3--1 m) between the point bars and swales situated next to each other. The different morphology of the two landforms---point bars are positive, swales are negative forms---cause essential discrepancies: e.g., the ground water level is closer to the surface in the swales due to their concave shapes, and also precipitation and snowmelt run off from the concave form and gather here ([Figure 1](#sensors-20-02063-f001){ref-type="fig"}b). Besides, they have differences in their granulometric composition, with swales having finer sediments that also slow down the infiltration of the water \[[@B46-sensors-20-02063]\]. All these features provide a higher percentage of moisture in swales, which support denser vegetation (that in some cases becomes impervious) and afford good conditions for aquatic vegetation (reed, sedge, etc.). In contrast, the vegetation density of point bars is relatively sparse compared to swales, except when they lie in a relatively lower part of the floodplain, because in this case their surface can be also covered by dense reeds. In [Figure 2](#sensors-20-02063-f002){ref-type="fig"}b,c, we highlight the pattern of the vegetation as it is shown in a portion of the 3D view of a point bar and swale series. The density of the vegetation points is higher and that of the ground points is lower in the case of the swales. In our previous work \[[@B50-sensors-20-02063]\] we also quantified this fact. The permanent and temporary water surface, the sedge-marsh and reeds, and the pastures and grazing lands make up the landscape mosaic of the study area and provide valuable habitats. All the floodplain here belongs to the Natura 2000 network and Ramsar sites. 3. Materials and Methods {#sec3-sensors-20-02063} ======================== 3.1. Aerial LiDAR Dataset {#sec3dot1-sensors-20-02063} ------------------------- The study area was surveyed by a RIEGL LMS-Q680i ALS LiDAR fixed on a Cessna C-206 Skywagon aircraft on 20 August, 2012 ([Table 1](#sensors-20-02063-t001){ref-type="table"}) in the framework of a project to prevent and manage natural disasters \[[@B51-sensors-20-02063]\]. The georeferenced point cloud of the study area, which was our basic dataset for the input of the analysis ([Figure 3](#sensors-20-02063-f003){ref-type="fig"}), was produced by Envirosense Ltd. and provided by the Trans Tisza Water Directorate. 3.2. Data Preparation {#sec3dot2-sensors-20-02063} --------------------- We applied two types of noise reduction filters for all the points of the point cloud data ([Figure 3](#sensors-20-02063-f003){ref-type="fig"}). Both filters apply a fixed kernel window, which assigns a certain number of points. Points delineate a small local plane area and the algorithm excludes the points exceeding a threshold value based on the distance between the delimited points ([Section 3.2.1](#sec3dot2dot1-sensors-20-02063){ref-type="sec"}) or the distance from the surface ([Section 3.2.2](#sec3dot2dot2-sensors-20-02063){ref-type="sec"}). As these filters use mean and standard deviation, they are called statistical outlier removal (SOR) filters. The analyses were carried out in CloudCompare 2.10.2 software \[[@B31-sensors-20-02063]\]. ### 3.2.1. Neighborhood Distance-Based Filter {#sec3dot2dot1-sensors-20-02063} The neighborhood distance filter is a kernel-based, neighborhood-related approach where the user-defined *k* nearest neighbors are investigated by each point of the dataset to see whether there are points exceeding the average distance plus standard deviation (Equation (1)). $$maximum~distance = {\underline{d}}_{k} + n\sigma\ \left\lbrack - \right\rbrack$$ where ${\underline{d}}_{k}$ denotes the average distance of the *k* neighbors around a given point (center); $\sigma$ is the standard deviation of the distances from the center; and *n* is the user-defined parameter, usually with a value of 1--3. We followed the recommendations and performed the noise filtering with 8 neighboring points (*k*) and with 2*σ*. ### 3.2.2. Surface Distance-Based Filter {#sec3dot2dot2-sensors-20-02063} This filter also uses a kernel window based on the user-defined number of neighboring points on a search radius, but outliers are selected calculating the distance from the local surface within the kernel (Equation (2)). Noise can be identified similarly to Equation (1), but the distance from the surface, or absolute maximum, can also be defined. We can exclude isolated points, where the number of neighbors is less than 3. $$maximum~distance = {\underline{sd}}_{k} + n\sigma\ \left\lbrack - \right\rbrack$$ where ${\underline{sd}}_{k}$ denotes the average distance from a local planar surface defined by k neighboring points around a given point (center); $\sigma$ is the standard deviation of the distances from the planar surface; and *n* is the user-defined parameter. If we choose a relative error, it is usually with a value of 1--3 (to exclude the statistical outliers). We applied the relative error option with 8 neighbors, 2$\sigma$, and excluding island points. ### 3.2.3. Ground Point Classification {#sec3dot2dot3-sensors-20-02063} We applied the CSF developed by \[[@B29-sensors-20-02063]\] to identify the ground points. Cloth refers to the grid size covering the study area, and higher values result in coarse DTMs. The procedure works "by analyzing the interactions between the cloth nodes and the corresponding LiDAR points, the locations of the cloth nodes can be determined to generate an approximation of the ground surface" \[[@B29-sensors-20-02063]\]. There are three types of terrain that can be considered in terms of their surface rigidness and defined by users: mountain areas with steep slopes, hilly areas understood as complex landscapes with trees and houses, and flat areas with high houses. CSF parameters are dependent on the point cloud density (cloth size, i.e., resolution) and the complexity of the terrain (threshold, i.e., the distance between points and the simulated terrain), and although suggestions exist, there is no definite rule for setting these parameters. Rather, a range can be used to find the ideal ones. We conducted the classification with the flat terrain option (as the relief was very low in the study area) and with cloth sizes of 2 and 5 (the larger the cloth, the coarser the DTM), according to \[[@B29-sensors-20-02063]\]. A value of 2 was suggested based on our point density, but we also intended to test the effect of a larger value (i.e., 5) We used classification thresholds of 0.2, 0.5, and 1, with 500 iterations in each case. The CSF was performed in CloudCompare 2.10.2 \[[@B31-sensors-20-02063]\]. 3.3. DTM Generation {#sec3dot3-sensors-20-02063} ------------------- We loaded the LAS files of the ground points into: (1) the LAS dataset, which is most commonly used for storing LAS files; (2) the terrain dataset (TD), which is a multiresolution, TIN-based (triangular irregular network) surface assembled from the ground points; and (3) a single point with z information. The different storage methods offered different opportunities to produce digital terrain models. In the case of the terrain dataset, two building options were used: the first was without thinning (z-tolerance), abbreviated as TD; the second was with the Z minimum point selection method and the moderate secondary thinning method with a threshold of 1 \[[@B52-sensors-20-02063]\], abbreviated as TH. We expected that the secondary thinning could improve the final models' accuracy, as it could thin points which were 'far' from the minimum. For the DTM production three interpolation methods were used: (1) natural neighbor interpolation with the minimum cell assignment type (NA); (2) linear interpolation with the nearest neighbor cell assignment type (LI); and (3) topo to raster (TT). Each of them was generated with two cell sizes, 1 m and 2 m, as these seemed to be the most appropriate pixel sizes due to the LiDAR point density and the width of the landforms. The natural neighbor is a weighted-average method using Thiessen polygons for the analysis of proximity to determine a cell value \[[@B53-sensors-20-02063]\]. The linear interpolation assigns the z values from the plane determined by the surface triangle that contains the x,y coordinates of a given point \[[@B52-sensors-20-02063]\]. The topo to raster creates a hydrologically correct raster from the points using the ANUDEM algorithm (elevation gridding method) developed by \[[@B54-sensors-20-02063],[@B55-sensors-20-02063]\]. Altogether, 180 DTMs were generated. All DTMs were produced in Esri ArcGIS 10.3 software with 3D Analyst and Conversion Tools \[[@B52-sensors-20-02063]\]. 3.4. Validation and Statistics Analyses {#sec3dot4-sensors-20-02063} --------------------------------------- A field survey was carried out by a Stonex S9 RTK (real-time kinematic) GPS (global positioning system) using the "stop-and-go" method with a real-time differential correction of the GNSS (global navigation satellite systems) permanent station system of Geotrade Ltd., Hungary. The accuracy was ±0.01 m both vertically and horizontally. A total of 604 reference points were taken crossing 8--15 swale--point bar series ([Figure 2](#sensors-20-02063-f002){ref-type="fig"}a). The dataset was used to validate the models. We defined the term of accuracy as the difference between the reference measurements and the modeled data. We extracted the values of all models where ground control measurements were available (604 RTK points) and subtracted them from the measured values of the RTK points. Finally, considering all factorial combinations (with 4 factors), 107,937 pieces of data were available in the analysis ([Supplementary material](#app1-sensors-20-02063){ref-type="app"}). We analyzed the dataset in terms of the efficiency of noise reduction, the different settings of CSF, the interpolation method, and the resolution. Finally, we also examined the effects of the different combinations of data preparation and interpolation techniques on the accuracy of the representation of fluvial forms, i.e., how the point bars and swales can be modeled and which approach results in the most accurate model. Spearman correlation (r) was used to analyze the correlation between the number of points and the model accuracies gained; we reported correlation at p (significance) \< 0.05. This type of correlation does not suppose a normal distribution and is not influenced by recurring similar data \[[@B56-sensors-20-02063]\]. We applied the Welch test for one-way comparisons with the Tukey HSD (honestly significant difference) post hoc test. The Tukey test is not sensitive to normal distribution \[[@B57-sensors-20-02063]\]. A robust two-way factorial ANOVA (analysis of variance) using 20% trimmed means and bootstrapping (999 replications) was used to reveal the interactions between the factor variables. As a result of the trimming, the analysis was not sensitive to outlier data. Based on the bootstrap approach---i.e., generating several replications, in our case 999---with random sampling from the original dataset, statistical parameters can be calculated related to the prediction error, variance of mean, etc. Accordingly, this robust approach does not require normal distribution \[[@B58-sensors-20-02063]\]. Swales and point bars were analyzed with the Wilcoxon test with Monte Carlo analysis (with 99,999 repetitions) \[[@B59-sensors-20-02063]\]. The H~0~ was that there was no difference between the number of points per square meter of the fluvial forms. Statistical analyses were conducted in R 3.53 statistical software \[[@B60-sensors-20-02063]\] with the coin \[[@B61-sensors-20-02063]\], the onewaytest \[[@B62-sensors-20-02063]\] and the WR2 \[[@B63-sensors-20-02063]\] packages. List of statistical abbreviations used in the results paragraph:-df: degree of freedom-F: F-statistic-p: significance-pmc: Monte Carlo simulation based p-value (significance)-Q: Q-statistic for 2-way ANOVA (analysis of variance)-r: Spearman correlation-W: Wilcoxon test statistic-z: z-score 4. Results {#sec4-sensors-20-02063} ========== 4.1. Number of Points and Accuracy {#sec4dot1-sensors-20-02063} ---------------------------------- Noise reduction resulted in a smaller number of points, reducing the input data of the ground point classification from 10.1 million to 8.7 million ([Table 2](#sensors-20-02063-t002){ref-type="table"}). The original dataset---without noise filtering---contained several points which caused false terrains in the final models. The different parameters of the CSF provided ground points on a large scale from 8.3 million to 3.7 million. Thus, the difference was large in the models, and the number of points had a moderate correlation with the mean accuracies (r = −0.65, *p* \< 0.05), indicating that smaller points provided better agreement with the field measurements. The smallest differences belonged to the neighborhood-based noise filtering, the larger cloth size (5), and the smaller threshold value (0.2). Generally, the differences were low, between 0.09--0.16 m, but the standard deviations were high (0.13--0.18 m), indicating high relative standard deviation (even more than 100%). 4.2. Effect of Noise Reduction and the CSF Parameters {#sec4dot2-sensors-20-02063} ----------------------------------------------------- Considering the noise reduction itself, according to the Welch test, all models had significant differences (F = 194.1; df = 7.18 × 10^−4^; *p* \< 0.001). In the following step, we analyzed the effects of CSF parameters, the cloth sizes, and thresholds ([Figure 4](#sensors-20-02063-f004){ref-type="fig"}). The smallest differences were obtained using the point clouds with noise reduction and the cloth size of 5, and a threshold of 0.2 occurred in the models (the difference was −0.08 m in relation to the reference). In the case of the original point cloud, these parameters provided the worst model, having the poorest accuracy (−0.12 m). The two-way factorial ANOVA revealed the relevance of the noise filtering and the CSF parameters ([Figure 5](#sensors-20-02063-f005){ref-type="fig"}). The results confirmed the observations from [Figure 4](#sensors-20-02063-f004){ref-type="fig"}, i.e., that both the noise filtering and the specifications of the CSF parameters had a significant effect on the modeled values. There were significant main effects for noise reduction (Q = 147.8; *p* \< 0.001), for cloth size parameters (Q = 19.65; *p* \< 0.001), and for their interaction (Q = 17.82; *p* \< 0.001). We observed a similar case with the model of threshold values: the main effects were significant (for noise Q = 149.1, *p* \< 0.001; for threshold Q = 87.67, *p* \< 0.001), and the interaction of noise reduction and threshold was also significant (Q = 231.3, *p* \< 0.001). 4.3. Consequences of the Interpolation Algorithms {#sec4dot3-sensors-20-02063} ------------------------------------------------- Considering the medians of the accuracies, values ranged between −0.16 and −0.05 m ([Figure 6](#sensors-20-02063-f006){ref-type="fig"}). However, the quartiles and the outliers varied on a larger scale (minimum: −1.67 m; maximum: 0.619 m) indicating that modeled values were influenced by several factors. Subtracting the worst from the best model, the largest difference in modeled height was −1.71 m. The visual interpretation highlighted that the model without filtering, using inappropriate CSF parameters and interpolation techniques, resulted in a DTM with a high number of pixels with noise. The noise is mainly concentrated in the area with higher density of vegetation (we will consider this issue more deeply in [Section 4.5](#sec4dot5-sensors-20-02063){ref-type="sec"}) ([Figure 7](#sensors-20-02063-f007){ref-type="fig"}). Generally, interpolations resulted in differences between the model pairs, except in the cases of TT, TD, and the terrain dataset with thinning TH methods (F = 745.8; df = 5.35 × 10^−5^; *p* \< 0.001; [Figure 8](#sensors-20-02063-f008){ref-type="fig"}). The LI usually resulted in models with the largest differences, and the NA interpolation was the most accurate. Although TT, TD, and TH were not the most accurate, there was another important issue to be considered: the narrower range of data. The most accurate model, based on the median difference, was the one with the NA interpolation with a cloth size of 5 and a threshold of 0.2; however, this model had the highest outliers in a positive direction. The TT, TD, and TH models were the models with minimal differences in medians (and insignificant differences based on the post hoc test). 4.4. Effect of the Resolution on the Accuracy {#sec4dot4-sensors-20-02063} --------------------------------------------- We analyzed the dataset from the perspective of the resolution of the final, interpolated maps ([Figure 9](#sensors-20-02063-f009){ref-type="fig"}). The two-way ANOVA revealed that the coarser resolution (2 m compared to 1 m) resulted in more accurate models: models were 0.009--0.012 m closer to the reference surface, on average. Noise reduction and resolution had a significant main effect (Q = 144.8, *p* \< 0.001 and Q = 120.1, *p* \< 0.001, respectively), but their interaction was not significant (Q = 0.14, p = 929). In the case of interpolation methods, all the main effects were significant (for interpolation Q = 2563.6, *p* \< 0.001, for resolution Q = 79.9, *p* \< 0.001) and their interaction was also significant (Q = 4478, *p* \< 0.001). According to the previous results, LI, TT, TH, and TD interpolations had the same values, and the resolution did not change the differences, but in the case of NA a relevant improvement (0.058 m on average) was observed. 4.5. Effects of the Noise Reduction and the Ground Point Classification on the Fluvial Forms {#sec4dot5-sensors-20-02063} -------------------------------------------------------------------------------------------- Point bars had larger point/m^2^ values in each model except in two cases (both distance- and neighborhood-based noise reduction models with the cloth size of 5 and the threshold of 0.2). Accordingly, the difference was significant (W = 140; z = 2.373; p~MC~ = 0.015). We revealed that the greatest accuracy (i.e., the smaller differences between the reference and measured values) was found in the ground point classifications of smaller points densities: 1.38 points/m^2^ for point bars and 1.90 points/m^2^ for swales ([Figure 10](#sensors-20-02063-f010){ref-type="fig"}). Furthermore, the best CSF parameters were in accordance with the previous results, and the fewest points, the cloth size of 5, and the threshold of 0.2 resulted in the highest accuracy. The noise reduction was efficient, and, although the difference between them was slight (less than 0.01 m), the neighborhood-related filtering was the most effective. Resolution had a significant effect on the accuracy, but the difference was only 0.009--0.011 m between the 1 and 2 m geometric resolution models. Generally, the accuracy of the swales was always below that of the point bars by 0.057--0.069 m. 5. Discussion {#sec5-sensors-20-02063} ============= Aerial LiDAR is a promising technology for collecting large amounts of data even in areas which are hard to access due to dense vegetation or their topography \[[@B64-sensors-20-02063]\]. Surveys result in continuous data from the target areas, and the outcomes are point clouds with several million data points with horizontal and vertical coordinates, the intensity of the returning beams, and, in the case of multiple echoes, the number of returns \[[@B65-sensors-20-02063]\]. Although these datasets represent the fastest and most accurate method of surveying and provide the possibility of object detection beside the elevation data, they have limitations as well \[[@B66-sensors-20-02063]\]. Our study area was a fluvial landscape with swales and point bars, with areas of temporary and locally permanent water cover, and with different vegetation densities (including trees, bushes, grasslands, and aquatic vegetation). These factors bias the final models derived from the point clouds, and we focused on the steps involved in the data preparation and the process of creating a digital terrain model. In previous works, e.g., \[[@B19-sensors-20-02063],[@B28-sensors-20-02063],[@B67-sensors-20-02063],[@B68-sensors-20-02063],[@B69-sensors-20-02063]\], it was found that noise reduction during preprocessing yielded a better digital terrain model, but on the other hand this procedure could sometimes reduce the accuracy of the model, as \[[@B70-sensors-20-02063]\] found in their study. We had a hypothesis that the noise reduction of the point cloud results in more accurate models in our case, and we pointed out that different noise reduction techniques can have a significant effect on the input data which are the basis of the next stage of the process, i.e., ground point classification. Although the distance-based method including island removal (i.e., deleting points clusters without connection to other points) seemed a powerful method, in the end we found that neighborhood-related noise removal provided the best input data for ground point classification in this fluvial area. However, noise removal was an important step, and both methods resulted in better models than the original point cloud without noise filtering. The difference between the two noise filters is that the distance-based algorithm also removes too many ground points and consequently leaves fewer points for the interpolation phase. Nevertheless, the difference between the noise filtering techniques was small, resulting as 0.01 m on average with same standard deviation ([Table 2](#sensors-20-02063-t002){ref-type="table"}). Accordingly, we confirmed that noise removal is a relevant beginning step in DTM generation from LiDAR point clouds. Our results are in accordance with the suggestions of \[[@B13-sensors-20-02063]\], showing that in the case of flat terrain, noise filtering based on a statistical approach can be an effective technique. Regarding all combinations, our final DTMs were more accurate using preliminary noise filtering, with slightly better outcomes using the neighborhood distance-based method. As one of the key steps in the ALS data processing is the point cloud filtering, the number of the different filters for ground point extraction is continuously increasing. In our study we utilized the method of \[[@B29-sensors-20-02063]\]---which applies a cloth simulation filter---where attention needs to be paid to set the most effective parameters, as it requires specific experience. The cloth simulation filter has two important parameters. The first parameter is cloth size, which is in accordance with the point cloud density, i.e., too low and too high values can also result in inappropriate models. We found that models produced from a cloth size of 5 m were more accurate than the finer, 2 m, setting. Thus, larger cloths (larger kernel windows) have more relating points where the algorithm can validate the threshold setting. An important result is that point density was 4 points/m^2^; furthermore, calculating with multiple echoes, it can even reach 10 points/m^2^ \[[@B50-sensors-20-02063]\]. Accordingly, a finer cloth size would have been reasonable, the recommendation of the developer is one-third of the point spacing (<http://ramm.bnu.edu.cn/researchers/wumingzhang/english/default_contributions.htm>), but according to the mean difference between the two settings, the 5 m cloth was 0.012 m better (for the neighborhood-related filter ([Figure 4](#sensors-20-02063-f004){ref-type="fig"})). The second CSF parameter is the threshold. A threshold value of 0.5 was suggested by \[[@B29-sensors-20-02063]\], but this was not the best setting in our case. A smaller value of 0.2 resulted in more accurate models with all interpolation methods. We confirmed that, generally speaking, lesser points provided better input for DTM generation, which agrees with the results of \[[@B68-sensors-20-02063],[@B69-sensors-20-02063],[@B71-sensors-20-02063]\]. However, lesser points did not mean the least points. A distance-based filter resulted in the least points, whereas the dataset of the neighborhood-based filter was the best input for the interpolation ([Table 2](#sensors-20-02063-t002){ref-type="table"}). In studies by \[[@B71-sensors-20-02063],[@B72-sensors-20-02063]\], the NA algorithm was proposed as the most appropriate one for the representation of coastal subdued areas using a LiDAR point cloud. In our work, interpolations had a relevant effect on the DTMs, and there were significant differences between the methods studied. According to the median-based rank of the differences, the NA had the best and the LI had the worst performance; the TT, TD, and TH models were found in the middle range. The TT, TD, and TH interpolations had very similar results without significant differences (*p* \> 0.05). Although NA interpolation (with a cloth size of 5 and a threshold of 0.2 as CSF parameters) provided the lowest differences according to the medians, it had a skewed distribution with a relevant number of outliers in the positive differences. As differences were calculated by subtracting the modeled terrain heights from the reference, this means that the most accurate NA model had several underestimated real heights, while models usually overestimated them. The TT, TD, and TH (without significant differences) showed medians only with slightly lower differences than the best NA, but the ranges were the narrowest. This can be an advantage as the potential error is smaller. However, unlike LI and NA methods, which are embedded in several software packages, including open-source solutions (e.g., ArcGIS \[[@B52-sensors-20-02063]\], GRASS GIS \[[@B73-sensors-20-02063]\], SADA \[[@B74-sensors-20-02063]\], Surfer \[[@B75-sensors-20-02063]\], R \[[@B76-sensors-20-02063]\], and Python \[[@B77-sensors-20-02063]\]) which ensure the widespread usage of the algorithms, TT, TD, and TH interpolations are available only in ArcGIS. Thus, these algorithms can be considered software-specific solutions limited to the users of that software. Differences were not high, but were greater than results of \[[@B48-sensors-20-02063]\], who found differences from the reference data ranging from --0.05 to +0.05 m. In our case, the range varied from --0.10 m to +0.60 m, and we observed large differences in individual points (--1.67 m to +0.62 m, which was relevant in the floodplain, considering that the relative difference between the deepest points of the swales and the highest point of the point bars was only 0.80--1.10 m). Resolution seemed a significant influencing factor in DTM generation, but without interaction with the noise filtering: the 2 m setting was more accurate by 0.007--0.010 m for each combination of the noise filters or the original dataset. This was in accordance with the analysis of cloth size parameter, where the larger (5 m) size resulted in more accurate models. Furthermore, interpolations were usually insensitive to the resolution, but the NA method, which was considered the best performing technique, was accurate to 0.005 m with the 2 m option. Accordingly, our recommendation is to use the coarser resolution of 2 m. The differences were not high but were significant. In floodplains, and especially in our study area, the dominant forms were the swales and the point bars. Point bars are always in a higher terrain position in relation to swales; accordingly, swales' water coverage lasts longer, and the vegetation's water supply is relevantly better. In our previous study \[[@B50-sensors-20-02063]\], we revealed that the vegetation density is significantly higher in swales; therefore, the number of ground points per m^2^ was significantly smaller (5.88 vs. 4.91). NDVI differed significantly (F = 1567, *p* \< 0.001), which was a relevant background factor which influenced the model accuracy when investigating these fluvial forms: removing the dense vegetation from the surface cannot be as accurate for swales as point bars. Generally, we have to note that these differences were not high, and we can come to the conclusion that, on average, all were in a negligible range. However, as reflected in [Figure 7](#sensors-20-02063-f007){ref-type="fig"}, the spatial appearance of the anomalies can relevantly alter the surface. These small differences can generate a different environment and change the characteristics of the fluvial forms, altering the waterflow modeling. The error propagation is unpredictable; thus, our primary task is to provide the best model and geomorphology seemed a good indicator by which to choose them. 6. Conclusions {#sec6-sensors-20-02063} ============== This study provides a brief description of point cloud processing from noise reduction to digital terrain generation. Our hypothesis was that noise filtering, ground point classification, interpolation, and geometric resolution have significant effects on the generated DTMs. We found that all types of preliminary noise filtering had significantly more accurate results related to the processing than when only applying the original database. CSF as a ground classification technique was a powerful tool, and the resulting DTMs had very low errors (from −0.03 to −0.22 m as upper and lower quartiles). CSF parameters had a significant effect on accuracy, where a coarser cloth size (5 m) and a smaller threshold (0.2) resulted in the best model performance. In the case of interpolations, we have drawn two conclusions: the natural neighbor method provided the most accurate model considering the medians; regarding the range of the differences, the topo to raster and terrain dataset approaches with natural neighbor interpolations provided the best DTMs. Although the density of the ALS point cloud made it possible to use a 1 m geometric resolution for the final DTM, the 2 m resolution was more accurate. We also revealed that landform elements, even when the line of sight is not limited by the topography, can decrease the models' accuracy; swales had significantly larger model errors due to denser vegetation and water absorption related to point bars. These results proved our hypotheses and can serve as a guidance for ALS LiDAR point cloud preprocessing, classification, and interpolation and for choosing the right resolution in a fluvial environment. We would like to thank to the Trans Tisza Water Directorate who provided the LiDAR point cloud for the input data of the analysis. The datasets analyzed for this study can be found in the Mendeley Data <https://data.mendeley.com/datasets/publish-confirmation/d93f3vmdm5/1>. Conceptualization, S.S. and Z.S.; Methodology, S.S. and Z.S.; Software, Z.S. and S.S.; Validation, C.A.T.; Formal Analysis, Z.S.; Investigation, I.H.; Resources, Z.S. and C.A.T.; Data Curation, S.S., Z.S. and I.H.; Writing---Original Draft Preparation, S.S., Z.S., I.H., and C.A.T.; Visualization, Z.S., S.S.; Supervision, S.S. All authors have read and agreed to the published version of the manuscript. The project was supported by the KH 130427 NKFI project. S.S. and I.H. were financed by the Thematic Excellence Programme of the Ministry for Innovation and Technology in Hungary (ED_18-1-2019-0028), within the framework of the Space Sciences thematic programme of the University of Debrecen. The authors declare no conflict of interest. {#sensors-20-02063-f001} {#sensors-20-02063-f002} {#sensors-20-02063-f003} {#sensors-20-02063-f004} {#sensors-20-02063-f005} {#sensors-20-02063-f006} {#sensors-20-02063-f007} {#sensors-20-02063-f008} {#sensors-20-02063-f009} {#sensors-20-02063-f010} sensors-20-02063-t001_Table 1 ###### The parameters of the survey. Parameters Value -------------------------------------------- ----------------- Designed point density 4 pts/m^2^ Average accuracy (horizontal and vertical) ±0.15 m Overlap 30--60% Pulse repetition rate 270 kHz Registration discrete return Laser wavelength 1550 nm AGL height 688 m Extent of the surveyed area 126 ha sensors-20-02063-t002_Table 2 ###### Accuracies as reflected in the noise reduction and CSF parameters (o: original LAS dataset; d: distance-based filter with island detection; n: neighborhood-based filter; CS: cloth size; Thd: threshold; SD: standard deviation). ------------------------------------------------------------------------------------------------------------ Filtering Method Noise Reduction CSF Parameters (CS; Thd) Point Number Accuracy (mean ± SD; m) ---------------------- ----------------- -------------------------- -------------- ------------------------- Original point cloud \- \- 10,120,880 −0.15 ± 0.17 Noise filter d \- 8,718,994 −0.13 ± 0.15 n \- 10,073,485 −0.12 ± 0.15 Ground\ d 2; 1 6,943,468 −0.15 ± 0.17 point filter d 2; 0.2 5,199,607 −0.12 ± 0.13 d 2; 0.5 6,375,149 −0.14 ± 0.14 d 5; 1 6,875,994 −0.15 ± 0.17 d 5; 0.2 3,720,552 −0.09 ± 0.16 d 5; 0.5 5,905,299 −0.13 ± 0.14 n 2; 1 8,050,253 −0.14 ± 0.16 n 2; 0.2 5,958,207 −0.12 ± 0.13 n 2; 0.5 7,395,394 −0.14 ± 0.14 n 5; 1 7,971,242 −0.14 ± 0.16 n 5; 0.2 4,246,638 −0.09 ± 0.16 n 5; 0.5 6,842,756 −0.13 ± 0.14 o 2; 1 8,293,970 −0.15 ± 0.18 o 2; 0.2 6,999,426 −0.15 ± 0.16 o 2; 0.5 7,837,259 −0.15 ± 0.16 o 5; 1 8,287,750 −0.15 ± 0.18 o 5; 0.2 6,729,067 −0.16 ± 0.16 o 5; 0.5 7,781,547 −0.15 ± 0.16 ------------------------------------------------------------------------------------------------------------

INTRODUCTION ============ Primary care in England is *'reaching saturation point'*.[@b1] Between 2007 and 2014 the number of consultations with GPs increased by 16%.[@b1] In 2015--2016, 12% of GP training posts were unfilled.[@b2] In the next 5 years, one-third of GPs plan to retire, and 28% plan to become part-time.[@b3] These challenges have prompted calls for alternative models of care, revised skill mix, digital technologies, and increased supported self-management.[@b4]^--^[@b8] Several technologies provide alternatives to face-to-face consultations. Telephone consulting is well established (66% of practices in England and Scotland report using this).[@b9] Although 25% of GPs have exchanged emails with patients, it is not routine practice.[@b10] Only 6% of UK practices report using email regularly. Most have no plans to do so.[@b9] A pan-European study of email consulting found wide variation in use across countries.[@b11] There is concern among GPs that alternatives to face-to-face consultations may increase workload and compromise safety.[@b3]^,^[@b9]^,^[@b12]^,^[@b13] The ESTEEM trial of telephone consulting found a 29% reduction of face-to-face contacts over 28 days, but an overall increase (38%) in all contacts.[@b14] A Cochrane review of email consultations was inconclusive regarding effect on workload.[@b15] Two models of online consultations (also called e-consultations) are currently available.[@b16] One is pharmacy led and explicitly avoids contact with the GP (patients obtain private prescriptions for a limited range of conditions online). The other involves web-based history-taking communicated to the patient's GP surgery, with potential for a face-to-face consultation depending on how the GP interprets the information. Despite equivocal evidence, NHS England plans to offer every practice support to adopt online consultation systems, committing an estimated £45 million investment.[@b5] In this article, the authors report a case study of an online consultation system recently incorporated into an inner-city general practice and consider how the introduction of an online consultation system is changing the work of general practice. METHOD ====== The authors conducted a qualitative case study of the development and implementation of an online consultation system (Tele-Doc) by a large, multi-site NHS GP partnership (Forest Group) and linked practice (Willow Surgery). (Tele-Doc, Forest Group, and Willow Surgery are all pseudonyms; see [Appendix 1](#app1){ref-type="app"} for further details of Tele-Doc.) The authors conceptualised 'the case' as context dependent, and evolving over time.[@b17]^,^[@b18] They collected data from multiple sources, including narrative interviews with a maximum variety sample of seven stakeholders (three development/operational staff at Forest Group, and four end-users, two of whom were GPs and two administrators at Willow Surgery), a purposive sample of six documents (including a pilot report, training presentations, and reports on user demographics), and a review of Tele-Doc. How this fits in ---------------- Online consultation systems are proposed as one policy response to increasing workload in primary care. Little is known about how online consultations play out in practice, or what this new consultation is. Structured online consultations may not reduce (and may even increase) overall workload, and may be ill-suited to consulting about complex problems. Expectations that new technologies will increase efficiency may be effective in attracting funding for technology development, but efficiencies may be difficult to achieve in practice. The authors invited participants to consider an example of using Tele-Doc prior to interview. Interviews lasted up to 50 minutes, used a topic guide (with four narrative-eliciting questions; see [Box 1](#box1){ref-type="boxed-text"}), and enabled insights into how participants construct meaning and identity.[@b19]^,^[@b20] ###### Topic guide and interview questions 1. Technical evolution of Tele-Doc 2. Changes to work patterns (individual, organisational) 3. Training/staff support 4. Barriers and facilitators to using Tele-Doc 5. Doctor--patient relationship 6. Contextual factors influencing Tele-Doc 7. Patient experience and use of Tele-Doc 8. Future development of Tele-Doc ```{=html} <!-- --> ``` 1. Could you talk me through your own story of involvement with Tele-Doc? 2. Could you talk me through the example \[of use of Tele-Doc\] that I asked you to think of before the interview? 3. How, if at all, has Tele-Doc shaped your working practice? --- Practice/organisational work 4. How, if at all, do you think patient experience is altered by Tele-Doc? 5. Is there something else you would like to talk about? 6. Is there someone else you think we should talk to as part of this project? Interviews were transcribed and analysed along with institutional documents using thematic analysis (identifying key themes) and then discourse analysis (to understand how and where meanings are constructed),[@b21]^,^[@b22] 'zooming in' on the nuance of talk and 'zooming out' to broader context in ways that kept interest in actual practice in the foreground.[@b23] The authors imported social theory to extend analysis, adopting a sociotechnical orientation[@b24]^,^[@b25] that conceptualises people, technologies, and material artefacts as interconnected networks,[@b25]^,^[@b26] and drawing the focus on to what happened when Tele-Doc was put to use in general practice. The authors also drew on the sociology of expectations literature.[@b27] RESULTS ======= Findings are organised in three interrelated themes: online consultation systems as innovation, managing the 'messiness' of general practice consultations, and redistribution of the work of general practice. Online consultation systems as innovation ----------------------------------------- > '*The* \[Forest Group\] *have always tried to innovate. Remote consultation is surely an innovation that's coming. Let's try and capture that.'* > > (Developer, Forest Group) > *'The big thing for me at the moment is, you know, there is this whole thing about health is the last industry that needs to move online, OK. Everyone does everything else online --- banking, travel, everything. So health needs to follow, and* \[Tele-Doc\] *is a tool that will facilitate that.'* > > (Developer, Forest Group) These quotes highlight the organisational context in which Tele-Doc was developed, with innovation valued as something to strive towards for its own sake. Sociological theory identifies technological innovation as future oriented, framed as a means to develop previously non-existent opportunities in which abstract expectations about the innovation (in this case Tele-Doc) can be shared among groups, providing direction and attracting investment.[@b27] Participants' accounts revealed a sense of the inevitable (*'an innovation that's coming'*), with use of future tense (*'will facilitate'*) aligning with a modernist perspective, in which it is assumed that technology will provide neat solutions to contemporary problems.[@b28] Use of extreme case formulations ('whole', 'everyone', 'everything')[@b29] and three-part lists ('banking, travel, everything*'*)[@b30] as rhetorical devices inspire confidence that Tele-Doc 'will facilitate' increased online health care. The implication is that health could and should 'move online', presenting technology as the key driver (rather than, for instance, the needs of patients). Significant resources were required to develop Tele-Doc: "'*It was a* \[Forest Group\] *decision to invest some ...* \[long pause\] *because it is money, in the GP time to assign five GPs, kind of 4 weeks where they didn't go and do surgeries. They just sat in a room every day and kicked around until they'd kind of refined these templates ... that was the kind of tipping point for us, I think. So we then had something like a product ... then launched* \[Tele-Doc\] *with the* \[Forest Group\] *patients initially*.'(Developer, Forest Group)" This quote suggests significant organisational slack within Forest Group (essential for development of any innovation),[@b31]^,^[@b32] enabling five GPs to be released from clinical work for 4 weeks to develop Tele-Doc templates, resulting in the emergence of path dependency ('tipping point')[@b27] when they had 'something like a product' to 'launch' and expectations became temporarily stabilised.[@b33] Expectations about an innovation and its use typically change over time.[@b27] Participants described how the implementation of Tele-Doc was an ongoing process, with continuing adjustment of expectations. For example, when Tele-Doc was piloted, GPs were expected to manage three consultations in 10 minutes: "*'When* \[Tele-Doc was\] *first introduced, it was a bit of an ask.* \[Tele-Doc\] *was a pilot scheme in addition to our allocated set number of* \[clinics\]*. It was roughly 20 contacts per* \[clinic\]*. We were having three of these per 10 minutes, which was quite ... Admittedly you're expected to do three consults within 10 minutes. And, clearly, that's not a sustainable ethos, so then you dropped down to two, two consults per 10 minutes, and then, I think, now we are sort of agreed we do one consult per 10 minutes.'*(Clinician, Willow Surgery)" The difficulty of completing three online consultations in 10 minutes (after a full clinic) is implied ('*which was quite* ...'), leading to a gradual process of reducing to one consultation per 10 minutes. There did not appear to be a clear way forward (*'I think, now we are sort of agreed ...'*), suggesting that Tele-Doc consultations will continue to evolve. Hence, while the original aspiration of developers for increased efficiency helped to attract funding (the pilot was funded by the clinical commissioning group (CCG) and a charity), it did not appear that efficiencies had been gained in practice. Tele-Doc was recast as being of *'equivalent standing to face-to-face and phone consults'* (pilot report), and a new narrative of 'respite' for clinicians during long face-to-face clinics emerged. The vision for Tele-Doc had further evolved into an ambition to pioneer online consultation systems widely within primary care, offering what developers described as *'a new channel and a new concept'* that *'time's only going to tell on how this does really pan out'*, speculating that it could *'completely change the model of general practice'.* Managing the 'messiness' of general practice consultations ---------------------------------------------------------- > '*I really like the way that* \[the Tele-Doc document\] *goes through the history and you can scan the yes/no bits quite quickly, and they're well flagged for bits that you should pay more attention to.*' > > (Clinician, Willow Surgery) Tele-Doc is one expression of a range of contemporary policy and professional developments, including standardisation of care (for example, protocols), a 'systems' approach,[@b34] aspirations for a 24/7 'customer' service (three participants drew parallels between Tele-Doc and online banking), and diversification of GP roles to include managerial and commercial ventures.[@b35] Partners at Forest Group showed considerable flexibility to adapt to this context and embrace its potential for doing things differently. A key finding was that Tele-Doc embodied a desire on behalf of the developers to make the consultation less 'messy' -- to *'carve off 10, 20, 30% of the stuff that comes in general practice, that's quite easy'* and make it *'more streamlined'*. These quotes suggest that the 'easy' parts of the consultation are readily identifiable and separable from undifferentiated symptoms. General practice consultations have previously been conceptualised as therapeutic in their own right,[@b36]^--^[@b39] offering patients an opportunity to make sense of their illness by co-constructing narratives in dialogue with their doctor [@b40]^,^[@b41] with a clinician who 'bears witness' to suffering.[@b42]^,^[@b43] Face-to-face consultations were described as being more 'taxing' and involving 'multiple problems': "'Some patients will bring in their child or their partner, or they'll come in with a second or third problem ... that's where it becomes tiring.'(Clinician, Willow Surgery)" It was precisely this 'messiness' that was difficult to remove from online consultations, with their highly structured organisation. Participants explained that when Tele-Doc was first introduced some patients entered free-text comments into questionnaires designed for conditions unrelated to their own. Patients' problems did not always fit neatly into the yes/no boxes provided. Forest Group responded: "'\[Developing\] *something called the "generic template", where you go on if you've got an undifferentiated symptom ... It was dramatic. Within a week twice as many Tele-Docs were coming through*.'(Developer, Forest Group)" The 'generic' template ('general advice' in the patient's view of Tele-Doc) allowed patients to express their concerns. It quickly gained popularity over condition-specific templates. As Morton and Cornwell have argued, the *'irreducible variability'* implicit in health care presents difficulties when attempts are made to rely on standardising approaches.[@b34] They go on to suggest that '*the only way to eliminate variability completely would be to eliminate patients*'.[@b34] Arguably, the online consultation goes some way towards this, albeit it also offers convenience for patients who do not want to travel or wait. Minimising variability may succeed in reducing the emotional labour of consulting (making it less 'tiring'). However, patients appreciate free expression. One clinician speculated upon a possible future with: "'*Tele-Doc automatically recognising that there is nothing here that a GP needs to do.*'" Redistribution of the work of general practice ---------------------------------------------- > *'"I am doing Tele-Doc." They* \[the other administrators\] *all know that means "stay away.*"' > > (Administrator, Willow Surgery) One consequence of Tele-Doc was redistribution of work from GPs to administrators and patients. Patients took on consulting work by completing a 'very thorough history', frequently involving many 'pages' of questions. New organisational routines were worked out by administrators (for example, sorting incoming Tele-Doc templates to identify appropriate recipients, entering Tele-Doc consultations into appointment slots). The assumptions underpinning Tele-Doc emphasised standardised working methods (geared towards achieving manageable workloads and efficiency savings) and created new work for administrators (*'doing Tele-Doc'*) and a shared understanding that staff would not be disturbed when engaged in it. The combination of an extreme case formulation (*'all'*)[@b29] and the voiced imperative (*'stay away'*) convey the intensity of this work. The need for 'more attention and focus' was in part to minimise new errors that were possible since integrating Tele-Doc. For example, if a Tele-Doc consultation was allocated an appointment slot in the clinical system, this would prompt an automated text message to the patient (inviting them to an appointment). To avoid this, administrators developed a 'workaround', booking an 'unregistered' patient instead, then typing the patient's name in small font indicating 'online consultation'. Administrators received Tele-Doc templates by email, uploaded them as attachments into electronic records (*'a lot of clicking'*) and decided how to allocate them. This process was prone to misunderstanding and error. For example, it was possible for unregistered patients to submit templates to Willow Surgery. Patients also used Tele-Doc in unintended ways: "'Some patients use it as a place to complain. They find it is a way to get us to sit up and complain ... not about Tele-Doc ... generally about the actual practice.'(Administrator, Willow Surgery)" "The practice had received minor complaints via Tele-Doc (for example, about waiting too long for an appointment) that, although unlikely to warrant a formal complaint, nonetheless demanded attention(*'they ... get us to sit up'*)." Uptake of Tele-Doc at Willow Surgery had been low, with 0--10 templates submitted daily (one GP said, *'We have so few it hasn't really made a dent'*). However, the administrative burden was substantial, beginning with reading the completed template to decide whether to 'book' an online consultation (meaning allocate it an appointment slot for GP review), and deciding whether work was clinical or administrative. For example, if on reading the Tele-Doc template administrators decided there was *'nothing a doctor really could do over the phone. The doctor has to see it'* (the example was a rash), they would book a face-to-face consultation, taking on some aspects of clinical decision making themselves. GPs had complained that they received requests they deemed administrative (*'Stuff that should be filtered out at reception'*) and had called for careful attention to this classification work. It had not been entirely successful: "*'If I find I don't know* \[if the Tele-Doc is clinical or administrative work\] *then I will approach someone who is higher up than me. I will either talk to the doctor, or I will talk to one of the admin staff.'*(Administrator, Willow Surgery)" The adaptation of working patterns to the technology, and parallel adaptation of the technology to meet different staff groups' needs, was not anticipated. Tele-Doc training focused on technical aspects of using the software and was described as 'superficial' or 'a single training session'. None of the participants referred to training about the impact of Tele-Doc on the non-technical aspects of their roles. Tele-Doc developers were aware that implementing Tele-Doc was not straightforward. Tele-Doc is now accessible to 231 UK practices, although approximately one-sixth of these practices do not use it: "'We've seen practices switch it off ... It's effectively who holds the power in practice, and it's not always the GPs. It's maybe the practice manager, for example ... they said, "It makes more work for us."'(Developer, Forest Group)" The data suggest that Tele-Doc generates substantial work for non-clinical staff who may be important mediators of the success (or not) of technology implementation. DISCUSSION ========== Summary ------- Findings show that Tele-Doc represented a small fraction of the clinical work in the case study. However, it constituted a significant re-thinking of what it means 'to consult', and aligned with the contemporary impetus for marketisation, standardisation, and commercialisation.[@b44] Clinical leads at Forest Group were innovation enthusiasts, viewing technology as a means to improve services, manage demand, and improve efficiency.[@b45]^,^[@b46] However, although the development of Tele-Doc had been successful (in terms of attracting funding, being 'up and running,' and dissemination to 231 practices), implementation appeared less successful (uptake by patients remained low; there was little evidence that efficiency gains were realised). This study suggests that clinicians and administrators worked hard to accommodate Tele-Doc, reshaping their working routines accordingly. This involved new work for administrators, different kinds of work for patients, and a need for clinicians to engage in a process of continual adaptation to embed Tele-Doc into clinical practice. The research suggests that --- at least for GPs and administrators --- the assumed potential of Tele-Doc for increased efficiency is difficult to achieve. As with other innovations,[@b47] further adaptation of Tele-Doc seems likely. Developers and clinicians invested considerable time and resource into Tele-Doc's initial condition-specific templates. However, it was the 'generic' template that proved most popular with patients. This may reflect patients' reluctance or inability to commit to a specific condition at the outset of their consultation, or a poor fit between the nature of patients' problems and the algorithmic logic inscribed into Tele-Doc. Further research is needed to explore patients' experiences of Tele-Doc. Strengths and limitations ------------------------- This study was small, undertaken as an MSc project, and focused on staff not patients. It involved one atypical practice --- an early adopter, closely related to the software developers, and interested in commercial opportunity. Participants were likely to be heavily invested in making Tele-Doc work. As case study researchers, the authors prioritise opportunity to learn over concerns about typicality [@b48] and particularisation over generalisation.[@b49] Single case studies, like this one, can be valuable in shaping future research in emerging areas, illuminating matters that typical situations might not, and providing in-depth analysis of what actually happens when (as in this case) technologies are implemented in practice. Comparison with existing literature ----------------------------------- These findings resonate with existing literature that describes the disappointingly low uptake of many novel technologies in healthcare settings,[@b50]^--^[@b52] concerns that the push for new technologies is driven by the interests of policy and industry rather than clear evidence of patient benefit,[@b28]^,^[@b53] and the importance of studying technology-in-practice[@b54] as transformation, not just implementation.[@b16] Implications for research and practice -------------------------------------- These findings challenge some of the assumptions underlying current digital health policy (for example, that technology will save time and money). For instance, although the aspiration that *'teams* \[in general practice\] *need support and space if they are to adopt new ways of working'* [@b5] is welcome, the findings suggest that staff training and support may be insufficient when attempting to introduce new technologies. Ethnography, preferably involving contrasting sites, has potential for illuminating the complexities of introducing online consultation systems. Empathy, presence, and compassion are traditionally regarded as important hallmarks of good general practice.[@b55] Traditionally, being fully 'present' involves not only physical co-presence in time/space, but may also include emotional, intellectual, and spiritual presence.[@b36] This kind of presence is called into question with the emergence of programmes like Tele-Doc, where patient and clinician are not physically present, rarely in dialogue, and communicating asynchronously. Further work involving both patients and practitioners is needed to fully appreciate the consequences of a shift away from physical presence in consultations, and the implications of online consultation systems for the quality of general practice consultations. For example, how is 'presence' negotiated at distance? How is empathy accomplished in alternative modes of consulting? Likewise, the primary care consultation has traditionally been understood to be exception rich, with the GP managing ambiguous and undifferentiated symptoms, tolerating uncertainty,[@b56] and bearing the emotional burden that this entails. Tele-Doc may offer convenience for patients with clearly defined problems, and respite from busy clinics for clinicians. However, these findings suggest that the emotional work of consulting may be marginalised, and that there may be important limits about what is achievable in this new genre of consulting. The authors would like to thank the participants from the case study sites for their involvement in the study, and the peer reviewers for helpful feedback on the manuscript. Funding ======= The study was carried out as Michael Casey's dissertation project within an MSc programme in International Primary Health Care. Michael Casey was funded by Newham CCG (Newham Education Training Academy) for his MSc studies, but the funder was not involved in selecting the topic of study nor the research sites, and the project was carried out without any funding for research costs. Ethical approval ================ The study was granted ethical approval by Queen Mary Research Ethics Committee (QMREC1608a). Provenance ========== Freely submitted; externally peer reviewed. Competing interests =================== The authors have declared no competing interests. Discuss this article ==================== Contribute and read comments about this article: **[bjgp.org/letters](bjgp.org/letters)** Tele-Doc was developed by the Forest Group in 2012, piloted in 2013, and is used by all Forest Group practices. It was launched as a commercially available product (owned by Forest Group Company, a separate company linked to the Forest Group) in 2014, and is now available for use in 231 UK practices. Forest Group includes 19 primary care centres (13 GP surgeries and six urgent-care centres, approximately 100 000 registered patients), based in Cityham, a large city with a diverse, high-density population. Willow Surgery is an inner-city surgery with two GP clinical leads, and four salaried GPs, serving a population of 10 000 patients, of whom 10% do not speak English, and 30--40% speak English as a second language. Remote consultations in Tele-Doc are accessed through the Willow Surgery website. Patients are offered three options --- search for specific conditions, ask for general advice (if the patient is not sure of their condition), and request administrative help. Patients can search \>100 common conditions. These are organised by groups, such as breathing problems, women's health, and mental health, and can also be viewed alphabetically and pictorially by areas of the body. Once a condition is selected, patients can choose self-help information, pharmacy information on relevant over-the-counter medications, direction to the 111 service (a national 24-hour telephone advice service for non-urgent health problems), and a structured online consultation (Tele-Doc). Tele-Doc is available 24 hours a day. Patients can expect a response by the end of the next working day. A Tele-Doc consultation begins with patients completing a condition-specific online questionnaire (template) that is submitted to the practice as an email. The questionnaire includes five sections, all of which must be completed. About you (for example, age, contact details).Your expectations (free-text boxes, limited to 500 characters).Your condition (systematic questionnaire, with pull-down menus offering tick box answers to multi-level questions on symptoms). Urgent or emergency symptoms prompt the appearance of a red warning box, advising patients to seek urgent/emergency services. If they select 'End my consultation, I will seek urgent care instead', the information they have entered is not submitted to the GP practice.Your health (for example, pre-existing conditions).Review and send to GP. The template is received by the practice as an email attachment and dealt with according to locally specified organisational routines. There is no real-time consultation, no audio or video component, no instant messaging, and no email exchange with the patient.

Introduction {#s1} ============ Benchmarking has gained popularity as a health care quality performance measurement tool with benchmarks used to compare delivery of care across institutions and jurisdictions and to encourage excellent performance by ranking institutions and highlighting top performers \[[@MZT069C1]\]. Establishing realistic performance benchmarks that monitor implementation of evidence-based best practice has been shown to improve performance compared with audit and feedback alone \[[@MZT069C2]--[@MZT069C6]\]. Benchmarks based on subjective or expert panel consensus rather than empirical data may be viewed as invalid. Data-derived benchmarks based on the average or median by definition are unlikely to drive excellence. The Achievable Benchmarks of Care (ABC™) approach is a method to establish 'real world' performance benchmarks by examining performance across all relevant organizations or health care providers and then determining the best care achieved by at least 10% of the eligible patients across organizations to identify 'top' performance levels \[[@MZT069C7], [@MZT069C8]\]. This method produces benchmarks that can be seen as realistic targets, as they have been achieved by at least one provider caring for at least 10% of all eligible patients in the sample. Currently, there is limited information on appropriate benchmarks for acute stroke care delivery \[[@MZT069C9]--[@MZT069C11]\]. In Ontario, a regionally based system of stroke care delivery---the Ontario Stroke System---was established in 2000 \[[@MZT069C12]\]. Within this system, regional stroke centers are accountable for leading the implementation of stroke care best practices across a geographic region, which includes a number of community hospitals, rehabilitation facilities (inpatient and ambulatory), community-based providers, community support agencies, health promotion practitioners, long-term care facilities and pre-hospital care providers \[[@MZT069C12]\]. The regional stroke centers are typically large teaching or academic hospitals with neurology and neurosurgical services, sophisticated diagnostic technologies and annual stroke/transient ischemic attack (TIA) volumes ranging from 400 to \>1100 per year. The established and organized approach to stroke care within Ontario provides a unique opportunity to develop and use stroke benchmarks for quality improvement. Methods {#s2} ======= We used clinical data collected at all acute care institutions in the province of Ontario, Canada, and among eleven regional stroke centers, by the Registry of the Canadian Stroke Network (RCSN). The ABC ™ methodology was used to calculate benchmarks for quality indicators for acute stroke care including: (1) thrombolysis among patients with ischemic stroke arriving within 2.5 h of symptom onset, without contraindications, (2) care on an acute stroke unit, (3) neuroimaging within 24 h of hospital arrival, (4) carotid imaging among ischemic stroke patients without atrial fibrillation, (5) dysphagia screening within 72 h of hospital arrival, (6) antithrombotic therapy, (7) anticoagulation for atrial fibrillation, (8) antihypertensive therapy and (9) lipid-lowering therapy \[[@MZT069C2]\]. Data source {#s2a} ----------- The RCSN includes: (1) a periodic random sample audit (Ontario Stroke Audit, OSA) of all acute care institutions with at least 10 stroke cases annually and (2) data on consecutive stroke/TIA patients seen at eleven regional stroke centers in the province of Ontario, Canada ([www.rcsn.org](www.rcsn.org)). The RCSN is a 'prescribed' registry under provincial privacy legislation, and charts are audited without patient consent. The overall project is approved by the Research Ethics Board of Sunnybrook Health Sciences Centre as well as the Research Ethics Board of each participating stroke center. Data are collected on all aspects of acute stroke management, including demographics, comorbidities, use of the emergency medical services, emergency department and in-hospital processes of care and complications by centrally trained neurology research nurses. Chart validation by duplicate chart abstraction has shown excellent agreement (kappa scores or intra-class correlation coefficients of \>0.8) for key variables in the database including age, sex, thrombolysis administration, stroke unit care and other processes of care \[[@MZT069C13]\]. Data sample {#s2b} ----------- Patients of eighteen years of age or older with stroke or TIA seen in a hospital emergency department or admitted to hospital were identified from administrative databases the National Ambulatory Care Reporting System and the Discharge Abstract Database maintained by the Canadian Institute for Health Information using International Classification of Diseases, Tenth Revision (ICD-10-CA). Those assigned codes I60, I61, I63, I64, H34.1 and G45 (excluding G45.4) were included. To calculate the benchmarks, we used data from the 2008/2009 RCSN Ontario Stroke Audit, which captured patients discharged from 142 acute care hospitals between 1 April 2008 and 31 March 2009. Ninety-nine percent of eligible hospitals participated in this audit, and a simple random sample of 17% of eligible cases was included (*n* = 3931) with over-sampling at low-volume institutions to ensure each institution contributed a minimum of 10 cases and 50 cases at smaller specialized stroke centers \[[@MZT069C14]\]. Because of concern about a small and unequal sample size for the development of some benchmarks, secondary analyses used data from consecutive patients seen at 11 regional stroke centers between 1 April 2006 and 31 March 2008 (*n* = 8 109). Acute Stroke Quality of Care Indicators {#s2c} --------------------------------------- Table [1](#MZT069TB1){ref-type="table"} lists the nine stroke quality indicators evaluated in this study. The indicators of stroke care performance are based on the Canadian Stroke Strategy\'s 2008 Performance Measurement Manual, are used for reporting within the Ontario Stroke System, reflect a subset of indicators identified by Accreditation Canada for hospitals seeking stroke care distinction status and are also recommended or reported by other organizations in other jurisdictions \[[@MZT069C2], [@MZT069C10], [@MZT069C11], [@MZT069C16]--[@MZT069C19]\]. Table 1Acute Stroke Quality of Care IndicatorsPerformance indicators1. Proportion of suspected stroke/TIA patients who receive a brain CT/MRI^a^ within 24 h of hospital arrival to the emergency department.2. Proportion of stroke/TIA patients treated on a stroke unit at any time during their inpatient stay.3. Proportion of ischemic stroke patients who arrive within 2.5 h of symptom onset and receive acute thrombolytic therapy (tPA)^b^ (excluding patients with known contraindications).4. Proportion of ischemic stroke patients without atrial fibrillation who receive carotid imaging prior to inpatient hospital discharge.5. Proportion of stroke (excluding TIA, unconscious patients) in patients with documentation that an initial dysphagia screening was performed within 72 h of hospital arrival.6. Proportion of ischemic stroke/TIA patients who were prescribed antithrombotic^c^ therapy at discharge.7. Proportion of ischemic stroke/TIA patients with atrial fibrillation prescribed anticoagulant therapy^b^ on discharge from acute care (excluding patients with contraindications).8. Proportion of ischemic stroke/TIA patients who were prescribed antihypertensive therapy at discharge.9. Proportion of ischemic stroke/TIA patients who were prescribed lipid-lowering therapy at discharge.[^1] Statistical analysis {#s2d} -------------------- We calculated overall indicator performance as the proportion of eligible patients that received each stroke care indicator rounded to the nearest whole number. We then used the ABC™ methodology to calculate benchmarks for each of the nine quality indicators. The overall indicator performance was assessed using Kiefe *et al*.\'s algorithm to determine the minimum sufficient denominator (MSD) (i.e. eligible patients) and whether a Bayesian adjustment was needed to adjust for a small number of eligible patients \[[@MZT069C7]\]. If a hospital\'s eligible patient sample did not meet the MSD (i.e. number of eligible patients) for each indicator, we applied a Bayesian adjustment to calculate the hospital\'s adjusted performance fraction described by Kiefe *et al*. \[[@MZT069C7], [@MZT069C8]\]. Starting with the highest performing hospital for the particular indicator and continuing through the next highest performing hospital, we cumulatively added each hospital\'s eligible patients until the total number of patients represented in the denominator included at least 15% of the total eligible patients across all hospitals. We based the benchmark on at least 15% of the total eligible patients across hospitals because the calculation of the ABC™ benchmark is based on crude data and we wanted to increase the number of eligible patients and hospitals to include in the benchmark and to reduce the influence of hospitals with small numbers of eligible patients on the benchmark. The benchmark was determined by dividing the total number of patients receiving the best practice care by the total number of patients eligible to have received best practice care in this subset. See [Supplementary material, Appendix A](http://intqhc.oxfordjournals.org/lookup/suppl/doi:10.1093/intqhc/mzt069/-/DC1) for a sample calculation. For each indicator, hospital(s) were considered 'top' performers if their performance rate was at or above the ABC™ benchmark. The ABC™ process was repeated using the validation cohort from 11 regional stroke centers, except that there was no need to apply the Bayesian adjustment as the number of eligible patients exceeded the MSD denominator at all of these hospitals. For each indicator benchmark, we report the total number of hospitals included in our benchmark calculations, as well as the range in the number of eligible patients at hospitals included in the ABC™ benchmark calculation. For each indicator, we report (1) the median, 25th and 75th percentile of eligible patients among the hospitals, (2) categories of the number of eligible patients, \<10, 10--24, 25--75, 75 or more and (3) the percentage of hospitals in the audit with eligible patient samples below the MSD. Results {#s3} ======= Table [2](#MZT069TB2){ref-type="table"} describes the characteristics of participating hospitals and patients in the 2008/09 Ontario stroke audit. Of the 142 hospitals included in the audit sample, 28 (20%) were regional or district stroke centers, 70 (49%) were high volume sites (with more than 100 patients with stroke or TIA annually), 90 (63%) had computed tomography (CT) scanners, 26 (18%) had stroke units, 70 (49%) had interdisciplinary stroke teams and 39 (28%) had a secondary stroke prevention clinic on site. The study sample included 3931 patients with a median age of 75 years. The number of patients sampled across the 142 hospitals ranged from 10 to166. Baseline characteristics of study patients are summarized in Table [2](#MZT069TB2){ref-type="table"}. Table 2Characteristics of stroke/TIA patients (*N* = 3931) and hospitals (*N* = 142) in the 2008/09 Ontario Acute Stroke AuditCharacteristicN (%)Hospital characteristics(N = 142)Designation: regional stroke centers^a^9 (6) District stroke centers^b^19 (13) Non-designated hospitals^c^114 (80) Urban hospitals83(58)Annual stroke/TIA volume ≥10070 (49)Hospitals with stroke units26 (18)Hospitals with designated stroke teams70 (49)Hospitals with CT on site90 (63)Hospitals with secondary stroke prevention clinics on site39 (27)Patient characteristics(N = 3 931) Stroke2 425(62) Transient ischemic attack1 167(30) Unable to determine338 (9) Stroke type(N = 2 370)  Ischemic1896 (80)  Intracerebral hemorrhage308 (13)  Subarachnoid hemorrhage119 (5)  Undetermined24 (1) Male1965 (50) Median age (years, 25th and 75th percentile)75 (60, 81) Rural residence590 (15) CNS scores \>82 948 (75) Transported by ambulance2 240 (57) Independent prior to admission3 302 (84) Diabetes983 (25) Hypertension2 555 (65) Hyperlipidemia1 454 (37) Atrial fibrillation^d^589 (15) Previous TIA/stroke1 336 (34) Previous MI511 (13)[^2] There was a wide range in performance for all nine indicators across acute care hospitals within the OSS. The overall performance ranged from 30% (admission to a stroke unit) to 94% (prescribing of antithrombotic therapy at discharge). Only two of the nine indicators (neuroimaging within 24 h and antithrombotic therapy prescribed on discharge) had overall performance rates of \>85% (see Table [3](#MZT069TB3){ref-type="table"}). Table 3.Achievable Benchmarks of Care in Ontario\'s acute care hospitals (2008/09)Stroke process of care indicatorsNumber of eligible patientsOverall performance (%)BenchmarksOSA--ABC™ (%)Number of hospitals included in OSA benchmark calculation (median, range of eligible patients included in the benchmark calculation)Accreditation Canada targets (10 ref) (%)Stroke Centre ABC™ (%)Neuroimaging \<24 h3176869811 (51, 21--138)≥90%99Admitted to a stroke unit2457307710 (46, 4--101)75%94Arrived within 2.5 h of stroke onset and received tPA^a^4693059^a^5 (11, 7--33)NA42Carotid imaging1186759311 (18, 12--56)NA87Dysphagia screening1924628818 (24, 3--66)≥90%90Antithrombotics on discharge2883949811 (43, 33--100)≥90%97Warfarin among patients with atrial fibrillation on discharge456709413 (5, 3--15)75%88Antihypertensives on discharge2883789231 (16, 6--60)NA79Lipid-lowering agents on discharge2883607722 (27, 3--65)NA68[^3] Five of Ontario\'s nine ABC™ benchmarks for each acute stroke quality indicator were \>90%: neuroimaging within 24 h (either CT or magnetic resonance imaging (MRI) of the brain) and antithrombotic therapy prescribed at discharge, 98%; warfarin prescribed at discharge for patients with atrial fibrillation, 94%; carotid imaging prior to discharge, 93% and antihypertensive therapy prescribed at discharge, 92%. ABC™ benchmarks of \<90% were dysphagia screening, 88%; admission to a stroke unit and lipid-lowering therapy prescribed at discharge 77% and thrombolysis administration, in patients presenting within 2.5 h of stroke symptom onset and without contraindications for thrombolysis, 59%. The five Accreditation Canada indicator non-data-derived targets correspond to two of the ABC™ benchmarks (neuroimaging and antithrombotic prescribing) but are lower for stroke unit admission and warfarin prescribing and higher for dysphagia screening. Benchmarks derived from regional stroke center data were similar to those derived from the entire provincial hospital sample for neuroimaging and antithrombotic therapy, higher for stroke unit admission and dysphagia screening and lower for all other indicators (Table [3](#MZT069TB3){ref-type="table"}). Less than a quarter of all hospitals in the audit (22%) were included in the benchmark calculation. The median number of eligible patients included in the benchmark calculation ranged from 5 for warfarin prescribing among patients with atrial fibrillation to 51 patients for neuroimaging within 24 h. The number of hospitals that met or exceeded the benchmark for any indicator ranged from 2 hospitals for patients presenting within 2.5 h of stroke symptom onset and without contraindications for thrombolysis to 61 hospitals for prescribing antithrombotic therapy on discharge (data not shown). Table [4](#MZT069TB4){ref-type="table"} illustrates the distribution of hospitals by categories of eligible patient sample size for each acute stroke quality of care performance indicator. The number of hospitals contributing to a performance indicator ranged from 93 (for thrombolysis) to 141 (for use of medications for secondary stroke prevention). Among our nine acute stroke quality indicators, the MSD varied from 4 to 36 eligible patients based on overall indicator performance. The percentage of hospitals with eligible patients below the MSD varied from 14% for stroke unit admission to 91% for warfarin prescribing on discharge among patients with atrial fibrillation. For each indicator, at least 65% of hospitals had \<25 eligible patients sampled, and for thrombolysis administration, only 1 hospital had at least 25 eligible patients and for warfarin prescription, no hospitals were sampled with at least 25 eligible patients. With the exception of prescribing antihypertensive therapy on discharge, all of the benchmarks had at least two specialized stroke hospitals (regional or district stroke centers) that achieved or exceeded the benchmark. Table 4.Distribution of Eligible Patient Volume in the 2008/09 Ontario Stroke AuditEligible patients (hospital, N)Neuroimaging within 24 hStroke unit admissiontPA among patients arriving within 2.5 h of symptom onsetCarotid imagingSwallowing assessmentDischarge medicationsAntithromboticsWarfarin for patients with AFAntihypertensivesLipid-lowering0244910713611\<1062727681766894686810--2434261643352512252525--75373518224604646\>75750022022Overall^a^14013893132135141106141141Median eligible patients (25th and 75th percentile)^b^11 (7, 32)9 (5, 27)3 (1, 7)5 (2, 13)6 (3, 21)10 (7, 34)3 (1, 6)10 (7, 34)10 (7, 34)MSD23441383611138% Hospitals below MSD331458745378915333Number of RSC^c^ at or above the ABC™ benchmark2/91/90/90/92/90/90/90/92/9Number of DSC^d^ at or above the ABC™ benchmark0/193/192/193/191/195/193/190/194/19[^4] Discussion {#s4} ========== This study provides benchmarks for nine important acute stroke care indicators using the ABC™ methodology with adjustment made for hospitals with small numbers of eligible patients, detailed indicator definitions to allow our work to be replicated, and the median and range of eligible patients within the hospitals included in the benchmark calculation is provided to assess the precision of the benchmark. These benchmarks are derived from a representative sample of stroke/TIA patients seen at 142 hospitals providing stroke care in a province of \>13 million people within an organized system of stroke care. We also provide the distribution of eligible patients for each indicator across the hospitals to assess overall indicator performance within the Ontario Stroke System. In addition, we also provide benchmarks based on data from regional stroke centers. Most of our benchmarks are similar to those found by Hinchey *et al.* \[[@MZT069C9]\] using the ABC™ method to calculate stroke care benchmarks for ischemic stroke patients from 17 volunteer hospitals (13 were community hospitals) across 9 US states (*N* = 2294). Our neuroimaging, stroke unit admission, prescribing antithrombotics and anticoagulants for patients with atrial fibrillation benchmarks are similar to the overall performance levels reported in other countries from a subset of hospitals that participate in quality improvement initiatives \[[@MZT069C11], [@MZT069C15]--[@MZT069C19]\]. Compared with the consensus-based non-data-derived targets used by Accreditation Canada to designate centers of stroke distinction, our data-derived ABC ™ benchmarks based on a sample of eligible patients across all acute hospitals in Ontario meet or exceed the Accreditation Canada targets with the exception of dysphagia screening \[[@MZT069C10]\]. Deciding what benchmark is appropriate depends on the purpose of that benchmark. Some might argue we only need to have consensus-based targets or performance better than the overall average especially if funding is tied to achieving a benchmark. However, under a quality improvement framework where data are essential for measuring performance, providing the number of eligible patients contributing to the benchmark provides organizations full disclosure in assessing the credibility, acceptability and achievability of the benchmark \[[@MZT069C20]\]. In Ontario, the OSS evaluation program has developed report cards on the quality of stroke care to be used by the OSS regions, hospitals, system planners and provincial agencies to understand the strengths and weaknesses and prioritize quality improvement initiatives in acute stroke care across Ontario hospitals \[[@MZT069C12], [@MZT069C21]\]. The benchmarks included in the report card are based on the provincial audit data (OSA). Benchmarks will need to be recalibrated over time, and the advantage of ABC™ methodology is that as all organizations improve the benchmarks rise as well while remaining achievable. The ABC™ approach to deriving benchmarks is data-driven, represents a level of excellence and is demonstrably attainable \[[@MZT069C7]\]. However, chance fluctuations, small sample sizes or samples of unequal size can produce unstable performance measures \[[@MZT069C7], [@MZT069C22]--[@MZT069C24]\]. Small and unequal sample size is a limitation of our study as it is for most performance measurement studies \[[@MZT069C22]--[@MZT069C24]\]. We evaluated the effect of small sample size on our benchmark calculations three ways. Firstly, we excluded facilities with \<25 eligible patients. This analysis resulted in no change to the neuroimaging benchmark, and minor (1 to 4%) decreases to two benchmarks and increases to four benchmarks (data not shown). In our study, four out of nine benchmarks (neuroimaging, stroke unit admission, antithrombotic and lipid-lowering agents prescribed on discharge) were not dominated by hospitals with numbers of eligible patients; median eligible patients among hospitals contributing to the benchmark were 51, 46, 43 and 27, respectively (Table [3](#MZT069TB3){ref-type="table"}). Five of our benchmarks were driven by small sample sizes therefore may be viewed with caution (median number of eligible patients were \<25); these include warfarin prescribing on discharge among patients with atrial fibrillation, thrombolysis, carotid imaging, dysphagia screening and antihypertensive medication prescribing these benchmarks may be viewed with caution (Table [3](#MZT069TB3){ref-type="table"}). In particular, the warfarin prescribing benchmark did not have any hospital included in the benchmark calculation with \>15 eligible patients. However, we have presented the median and range of eligible patients among the hospitals included in the benchmark calculation for the reader to assess the acceptability the benchmark based on a sample of all hospitals in Ontario. Secondly, we explored whether the benchmarks would change if we excluded hospitals where the MSD was not met for each indicator. We found all benchmarks remained unchanged, with the exception of warfarin prescribing for atrial fibrillation where the benchmark was reduced from 94 to 76% (data not shown). Finally, we repeated the benchmark calculations using data from 11 regional stroke centers where eligible patient sample sizes are larger and demonstrate less variation. We found that six of the nine ABC™ benchmarks were lower (1 to 17%) than those derived from the entire sample of hospitals (Table [3](#MZT069TB3){ref-type="table"}). However, neuroimaging, admission to a stroke unit and dysphagia screening benchmarks were higher (1, 17 and 2%, respectively). This finding may be a statistical artifact; difference in years of data may reflect real differences in processes of care at these regional stroke centers. For example, regional stroke centers are large tertiary centers and may have facility/system issues that prevent them from achieving higher levels of performance. Stroke patients compete with trauma and cancer patients for access to assessment and diagnostic imaging not faced by the smaller district stroke centers. Given the structural issues and more complex patients at regional stroke centers, the benchmarks derived from regional stroke centers may be considered to have better face validity among those centers with similar characteristics (e.g. large academic/teaching hospitals) to be used for peer benchmarking. Additionally, the time-dependent difference between the regional stroke center data (2006 to 2008), and the provincial hospital population-based data (2008/09) may also explain the lower benchmarks obtained using the regional stroke center data. However, when we recalibrated the regional stroke center-derived ABC™ benchmarks using the same fiscal year data as the OSA ABC™ benchmarks, 1 April 2008 to 31 March 2009, we did find a significant change to four of ABC™ benchmarks based on 2006--2008 data. In particular, thrombolysis increased from 42 to 48%, warfarin prescribing decreased from 88 to 76% and antihypertensive and lipid-lowering prescribing increased from 79 to 86% and 68 to 82%, respectively. Hierarchical modeling has been cited as an appropriate statistical method to use for performance measurement and in particular for outcomes of care rather than processes of care to deal with the issue of unstable performance metrics based on small and unequal sample sizes \[[@MZT069C23]--[@MZT069C25]\]. Hierarchical modeling takes the performance estimate of a facility and 'shrinks' it closer to the mean performance among all facilities and, the degree of 'shrinkage' is greater for facilities with smaller sample sizes. A benchmark could be based on the 90th percentile among these 'shrunken' performance estimates derived from hierarchical modeling. O\'Brien *et al.* \[[@MZT069C22]\] generated benchmarks based on the 90th percentile using hierarchical modeling and found this 'shrinkage' resulted in minimally lower benchmarks (\<3%) compared with the benchmarks derived using the ABC™ methodology with the Bayesian estimator to reduce the influence of a small sample for five out of their eight ABC™ benchmarks, but more marked (7 to 32% lowering) for the remaining three benchmarks with the most dramatically lower benchmark for the indicator dominated by hospitals with \<25 eligible patients (\>90% of the hospitals). We chose not to use hierarchical modeling to calculate our first release of our process of care benchmarks as we wanted to provide an easily understood method to stakeholders. The ABC™ methodology uses the actual performance with a Bayesian adjustment made for facilities that are below the MSD to reduce the influence of small sample sizes and has greater transparency compared with a statistical modeling approach to adjust hospital performance rates. Furthermore, O\'Brien *et al.*\'s work demonstrated minimal difference for the majority of their process of AMI care benchmarks generated by hierarchical modeling compared with the ABC™ benchmarks, and Arling *et al.*\'s ischemic stroke quality performance indicators demonstrated modest increases in the 90th percentile performance level when applying multilevel model empirical Bayes estimation methods compared with the unadjusted 90th percentile performance \[[@MZT069C22], [@MZT069C24]\]. Arling *et al.* also observed small denominators (i.e. eligible patients) for many of their quality indicators as did our study and considerable uncertainty remained with the estimated Bayes adjusted performance indicators. We, however, did not produce confidence intervals around our ABC™ benchmark because the purpose of our benchmarks is to drive quality improvement rather than identify good versus poor performance. Finally, our quality indicators and benchmarks reflect the Canadian context, and therefore our benchmarks may not be transferable to other jurisdictions. The data were collected to measure performance based on the Canadian perspective according to the Canadian Stroke Strategy Performance Measurement Handbook, which was designed to measure the implementation of the Canadian Stroke Strategy Best Practice Recommendations (2). Conclusion {#s5} ========== Benchmarks are an important tool for quality improvement initiatives and empirically derived benchmarks including patient sample size offers transparency to allow the credibility of the benchmarks to be assessed. This is the first Canadian study to report benchmarks from all acute care hospitals within a large province. Although the acute hospitals vary widely in size and complexity of service provision, these benchmarks for acute stroke care delivery can be used for reporting and quality improvement to strive for excellence in delivering acute stroke care. Further research is needed to examine what is the optimal benchmark for treatments that have high prevalence of contraindications such as anticoagulation in atrial fibrillation and thrombolysis. Supplementary material {#s6} ====================== [Supplementary material is available at *INTQHC* online.](http://intqhc.oxfordjournals.org/lookup/suppl/doi:10.1093/intqhc/mzt069/-/DC1) Funding {#s7} ======= This work was supported by an operating grant from the Canadian Stroke Network. The Registry of the Canadian Stroke Network is funded by an operating grant from the Ontario Ministry of Health and Long-Term Care. The Institute for Clinical Evaluative Sciences is supported by an operating grant from the Ontario Ministry of Health and Long-Term Care. The results and conclusions are those of the authors and are not attributed to any of the sponsoring or funding agencies. The funding agencies had no role in the design or conduct of the study or the collection, management, analysis or interpretation of the data. The manuscript was reviewed and approved by the publications committee of the Registry of the Canadian Stroke Network. Funding to pay the Open Access publication charges for this article was provided by the Canadian Stroke Network. Supplementary Material ====================== ###### Supplementary Data We thank the Ontario Stroke Network\'s (OSN), Stroke Evaluation and Quality Committee (SEQC) Knowledge Translation Subcommittee members (Cally Martin, Jim Lumsden and Beth Linkewich) for providing input for selecting the ABC™ benchmarking methodology for use within the OSN and for reviewing this paper; Kathryn Hodwitz for manuscript formatting. [^1]: All indicators exclude unable to determine final diagnosis except indicator 1. All indicators exclude stroke type unknown except indicator 1, 2 and 5. ^a^CT and MRI based on records with admission date/time, scan date/time recorded. ^b^tPA, tissue plasminogen activator. At the time the data were collected, the tPA window of time was 3 h. Midway through the time of data collection (September 2008), trial results demonstrated the safe therapeutic window for tPA delivery from stroke symptom onset had increased to 4.5 h. However, we chose to base it on the longstanding therapeutic window of 3 h as practice change was not expected to change immediately. ^c^Antithrombotic therapy includes acetylsalicylic acid (ASA), combination ASA and dipyridamole, clopidogrel and warfarin. [^2]: ^a^Regional stroke center: all the requirements of a district stroke center, plus neurosurgical facilities and interventional radiology. ^b^District stroke center: facilities with written stroke protocols (e.g. transport and triage, thrombolytic therapy and neuroimaging), clinicians with stroke expertise and linkages to rehabilitation and secondary prevention. ^c^Non-designated: acute care hospital that does not fit the definition of district or regional stroke center. ^d^Atrial fibrillation was based documented on past history of OR new onset during hospital stay. [^3]: ^a^only hospitals with capacity to deliver tPA. OSA, Ontario Stroke Audit; NA, not available. [^4]: ^a^Number of hospitals with at least 1 eligible patient. ^b^Median number of eligible patients among hospitals with at least 1 eligible patient. ^c^regional stroke center: all the requirements of a district stroke center, plus neurosurgical facilities and interventional radiology. ^d^District stroke center: facilities with written stroke protocols (e.g. transport and triage, thrombolytic therapy and neuroimaging), clinicians with stroke expertise; and linkages to rehabilitation and secondary prevention. MSD, minimum sufficient denominator (i.e. eligible patients).

{#sp1 .310}

1. Introduction {#sec1} =============== Renal transplantation is the treatment of choice for chronic kidney disease and has been shown to have better outcomes than dialysis in multiple studies.[@bib1] In the standard procedure, the donor renal artery and vein are anastomosed to the external iliac vessels of the recipient. In cases of an unusable vein, such as thrombosis of the iliac vein or inferior vena cava (IVC), renal transplantation becomes extremely difficult. These issues were originally considered contraindications to renal transplant, but several case reports have demonstrated ways to circumvent the obstructed veins by using other systemic or portal veins in the area. We present an unusual case of kidney transplantation on a right external iliac vein (EIV) that contained a chronic thrombus extending to the common iliac vein (CIV) and infrarenal IVC, hence, partially obstructing the right EIV. It was identified intraoperatively after a standard initial anastomosis. The transplanted kidney was salvaged by utilizing the recipient right gonadal vein to bypass the iliac outflow obstruction. 2. Case presentation {#sec2} ==================== The patient is a 54-year-old female with a long history of systemic lupus erythematosus (SLE) that progressed rapidly to end-stage renal disease. She started hemodialysis in 1993 and has had her lupus controlled by low dose Prednisone. She exhausted her right and left upper arm arterio-venous (AV) fistulae and has been using a left thigh graft for vascular access. Her past medical history includes peripheral vascular disease, hypertension, dyslipidemia and coronary angioplasty with two stents placement (on Clopidogrel). As part of the patient\'s pre-transplant workup, ultrasound duplex (USD) of the iliac vessels revealed normal blood flow six months before transplantation. Her hypercoagulable workup was negative for lupus anticoagulant. A right kidney became available from a 33-year-old deceased donor who died from a head injury. 2.1. Description of the procedure {#sec2.1} --------------------------------- The risks and potential complications of the surgery were explained to the patient, and informed consent was obtained. Preparation of the renal allograft was performed on ice. The right renal vein was elongated by reconstructing the attached donor IVC. Through a standard Gibson incision, retroperitoneal space was created in the right iliac fossa. The reconstructed renal vein was connected to the recipient EIV using 5/0 Prolene sutures, followed by the arterial anastomosis between the renal artery and the recipient external iliac artery. After unclamping the anastomosis site, the kidney became extremely congested with microcapillary bleeding and began to ooze from the venous suture line. After realizing that the venous pressure was elevated, the arterial inflow was immediately reclamped. Intraoperative USD revealed chronic clot partially occluding the proximal EIV, extending to the CIV and infrarenal IVC. The ovarian vein was simultaneously dilated, so the gonadal vein was dissected until it became tension-free. The reconstructed renal vein was partially clamped at its side wall using a Satinsky clamp. After flushing with heparinized saline, an end to side anastomosis was performed between the ovarian and renal vein using 6/0 Prolene in running fashion ([Fig. 1](#fig1){ref-type="fig"}). The graft regained its pink color and produced a few drops of urine. A standard ureteroneocystostomy was performed over a double J stent. Warm ischemia time was 90 minutes. This included alternating periods of clamping and unclamping.Fig. 1Schematic drawing depicts the steps of gonadal vein utilization to bypass the outflow obstruction. The patient received induction immunosuppression of thymoglobulin (1 mg/kg) along with methylprednisolone (500 mg intravenous) for three daily doses, and basiliximab (20 mg) for two doses. The patient was maintained on low-dose tacrolimus (target 12 hours trough level: 4***--***6 ng/ml) and enteric-coated mycophenolate sodium (720 mg) twice daily, as well as low-dose Prednisone. The patient was heparinized postoperatively and switched to Warfarin on postoperative day 3. Although she developed a peri-renal hematoma that was drained percutaneously using CT guidance, her recovery was uneventful and was discharged on postoperative day 7. At four years follow-up, she has maintained adequate renal function with a stable serum creatinine (1.2 mg/dl) and patent renal allograft vessels ([Fig. 2](#fig2){ref-type="fig"}).Fig. 2Post-transplant MRI of abdomen demonstrates patent ovarian (blue arrow) and renal vein (yellow arrow). Note the ovarian vein dilation to compensate for the increase in renal blood flow. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.) 3. Discussion {#sec3} ============= Renal transplantation to the iliac veins remains standard under ideal conditions. However, a nonfunctioning iliac vein or IVC is not necessarily a contraindication for successful transplantation. Case reports have shown favorable outcomes with anastomoses to other nearby vessels, including portal and systemic veins. Kumar et al.[@bib2] described a successful transplantation in which his team connected the donor renal vein to the recipient splenic vein (portal system) after a thrombus was discovered in the IVC. Patel et al.[@bib3] have demonstrated successful use of the inferior mesenteric vein in renal transplantation, while Aguirrezabalaga et al.[@bib4] have used the superior mesenteric vein. Although Wong et al.[@bib5] reports a case in which renal transplantation was accomplished using the left ovarian vein in a patient with an iliac vein/IVC thrombosis, we report the first case in which the gonadal vein was utilized in the setting of outflow obstruction after reperfusion to save the allograft. Preoperative USD screening and thorough radiological investigation of all vessels should be performed in all potential renal transplant patients to determine the state of the iliac vein/IVC and to choose alternative anastomoses if needed. Regarding our patient, although pre-transplant workup was negative for lupus anticoagulant and USD didn\'t capture the venous thrombus probably due to its partial obstruction, the clinician should maintain a high level of suspicion particularly in patients who pose a greater risk of thrombosis such as recurrent AV fistulae clotting, peripheral vascular disease, and history of SLE. Unfortunately, there are no current guidelines or consensus on the management of a transplant with thrombosed IVC/iliac veins. While smaller veins may be used during anastomosis, they may predispose to thrombosis and other adverse effects.[@bib3] 4. Conclusion {#sec4} ============= Kidney transplantation on gonadal vein is safe and efficient. Nevertheless, clinicians should use their clinical judgment and radiologic evidence to determine if an alternative vessel can be safely and effectively utilized for anastomoses if the IVC and iliac veins are not usable. Patients should also be closely followed after transplant for evidence of thrombosis or other adverse effects when alternative vessels are used for anastomosis. Conflict of interest {#sec5} ==================== Neither author has funding or conflicts of interest to report. AV : Arterio-venous CIV : Common iliac vein EIV : External iliac vein IVC : Inferior vena cava SLE : Systemic Lupus Erythematosus USD : Ultrasound duplex

Background {#Sec1} ========== Canine leishmaniosis (CanL) caused by *Leishmania infantum* is a zoonotic vector-borne disease with a wide geographical distribution in both the Old and New World. Infected dogs are the main domestic reservoir of the parasite \[[@CR1]\]. Dogs can manifest a chronic subclinical infection, self-limiting disease, or non-self-limiting illness \[[@CR1], [@CR2]\] as previously documented in humans \[[@CR3]\]. Therefore, several degrees of disease severity are found in dogs ranging from mild disease to severe fatal disease. Two clinical staging systems are currently used in the clinical setting \[[@CR2], [@CR4]\]. LeishVet clinical staging system ranges from stage I-mild disease to stage IV-very severe disease with different clinical outcomes, prognosis and treatment options \[[@CR2]\]. Cutaneous lesions are the most common clinical signs in CanL \[[@CR5]\] and they are very pleomorphic from a clinical and histopathological point of view as well \[[@CR6]\]. The most common dermatological signs observed in dogs with leishmaniosis include exfoliative dermatitis, ulcerative dermatitis and onychogryphosis \[[@CR5]\]. However, other less typical manifestations such as papular dermatitis, muco-cutaneous nodular dermatitis or sterile pustular dermatitis are also diagnosed \[[@CR5], [@CR6]\]. This clinical variation is due to a wide variety of pathological mechanisms occurring secondarily to the inflammation, immune complex deposition and/or autoantibody production \[[@CR7]\] and to the genetically determined or acquired inability of the immune system to control parasite multiplication and tissue invasion \[[@CR8]\]. Among the cutaneous manifestations of CanL, papular dermatitis is the only permissible dermatologic manifestation in stage I leishmaniosis \[[@CR2]\]. Dogs with papular dermatitis commonly show no other clinico-pathological abnormalities and anti-*Leishmania* antibodies are negative or weakly positive. This dermatological problem is associated with a good specific cell-mediated immune response as well as the spontaneous resolution of the lesions within 3--5 months in some cases \[[@CR9]--[@CR11]\]. The normal-looking skin has been scarcely studied either in diseased or in infected but clinically healthy dogs \[[@CR12]--[@CR15]\]. However, only one study evaluated both clinically-lesioned and normal-looking skin from the same individuals \[[@CR14]\]. In addition, to the best of our knowledge, comparative studies between paired clinically-lesioned and normal-looking skin samples from dogs with different stages of disease severity are lacking. Normal-looking skin of dogs with leishmaniosis, with or without dermatological manifestations, frequently shows microscopic lesions along with the presence of *Leishmania* amastigotes \[[@CR5]\]. However, this might not apply in less severe clinical cases. The objective of this study was to characterise and compare the inflammatory pattern and the parasite burden by microscopic examination, immunohistochemistry (IHC) and real-time polymerase chain reaction (qPCR) analysis in paired clinically-lesioned and normal-looking skin from the same dogs with dermatological manifestations due to CanL with different stages of disease severity (stage I-mild disease *versus*stage II-III-moderate to severe disease). Methods {#Sec2} ======= Dogs and diagnosis of leishmaniosis {#Sec3} ----------------------------------- Twenty-five dogs with CanL and dermatological manifestation were prospectively enrolled at the time of diagnosis from January 2014 to February 2016. The dogs were from different Catalonian and Balearic veterinary centers from Spain: Fundació Hospital Clínic Veterinari (Bellaterra, Barcelona), Hospital Ars Veterinaria (Barcelona), Hospital Mediterrani Veterinaris (Reus, Tarragona), Consultori Montsant (Falset, Tarragona) and Hospital Mon Veterinari (Manacor, Mallorca)*.* The diagnosis of canine leishmaniosis was made based on the results of the physical examination and cytological or dermatopathological examination of cutaneous lesions. Moreover, a complete blood count using System Siemens Advia 120 (Siemens Healthcare GmbH, Germany), a biochemical profile including creatinine, urea, total proteins, alanine transaminase and total cholesterol by Analyzer Olympus AU 400 (Olympus, Center Valley, USA), serum protein electrophoresis by Hydrasys® (Sebia Electrophoresis, Lisses, France), urinalysis with urinary protein/creatinine ratio and quantitative serology for the detection of *L. infantum* specific antibodies by means of a serial dilution in-house ELISA were performed \[[@CR16], [@CR17]\]. Dogs were classified in four different stages (stage I-mild disease, II-moderate disease, III-severe disease and IV-very severe disease) at the time of diagnosis as previously described \[[@CR2]\]. Collection and processing of skin samples {#Sec4} ----------------------------------------- Two skin fragments from paired clinically-lesioned and normal-looking skin were collected from each dog. Normal-looking skin was obtained whenever possible from the lateral aspect of the neck. In cases where this region was affected, normal-looking skin was collected as far away as possible from the macroscopic lesions. Each skin sample was then immediately cut into two halves. One half was fixed in 10% formalin for routine histological and immunohistochemical examination and the other one submerged in RNA later (RNAlater® Stabilization Solution, Ambion, Inc., Austin, Texas) and kept at -80 °C until used for RNA extraction and consecutively DNA purification for qPCR analysis. Histological examination and *Leishmania* immunohistochemistry {#Sec5} -------------------------------------------------------------- The dermal inflammatory pattern and the cell population were evaluated histologically in haematoxylin and eosin (HE)-stained sections. The distribution pattern of the infiltrate (perivascular to interstitial or nodular to diffuse with or without granuloma formation); the inflammatory cells (macrophages, lymphocytes, plasma cells and neutrophils); the degree (none, mild, moderate and severe) of cellular infiltration in the dermis and the epidermal changes (hyperplasia, spongiosis and exocytosis) were evaluated as previously described \[[@CR18]\]. IHC for the detection of *L. infantum* amastigotes was performed as previously described \[[@CR18]\]. The parasite load in immunolabelled sections was determined as the average number of microorganisms counted in five high power fields of areas with inflammatory infiltrate: 0, no microorganisms; 1, 1--10; 2, 11--30; and 3, \> 30 \[[@CR12]\]. qPCR {#Sec6} ---- RNA was isolated from skin biopsies using the RiboPure Kit (Ambion, Inc., Austin, Texas) and stored at −80 °C until used for future studies. DNA was purified from the interphase and organic phase generated from the RNA purification process by means of QIAamp DNA Mini Kit (Qiagen, Manchester, UK) following the manufacturer\'s instructions with slight modifications. Briefly, 20 μl of proteinase K solution and 200 μl of tissue sample were used in all cases. The other steps were performed as per manufacturer\'s protocol. A fragment of spleen and/or skin from a clinically healthy non-infected dog from a non-endemic area (United Kingdom) was used as a control for DNA contamination during DNA extraction. qPCR was performed with a relative quantification as previously described with minor modifications \[[@CR19]\]. Briefly, PCR mix reaction was prepared with 4 μl of DNA, 10 μl of master mix (TaqMan® Fast Advanced Master Mix, Thermo Fisher Scientific Inc.), 1 μl of *Leishmania* primers and probes (Custom TaqMan® Gene Expression Assay, ThermoFisher Scientific Inc., Waltham, USA) or 1 μl of another type of assay primers and probes \[Eukaryotic 18S rRNA Endogenous Control (VIC™ ⁄ MGB Probe, Primer Limited, ThermoFisher Scientific Inc., Waltham, USA)\] and 5 μl of H~2~O. In order to verify that the PCR was done successfully, a positive control for *Leishmania* and a negative control from a non-infected clinically healthy dog were included in the plate. PCR was carried out in a QuantStudio Flex™ 7 Real-Time PCR system (ThermoFisher Scientific Inc., Waltham, USA). Thermal cycling profile consisted of 50 °C for 2 min in order to activate the enzyme called amperase and afterwards, a total of 40 cycles were carried out. Each cycle comprised 20 s at 95 °C followed by 40 cycles of 1 s at 95 °C and 20 s at 60 °C. To compensate for variations in total DNA input, mean values of cycle threshold (CT) from duplicate determinations from the *Leishmania* and 18S rRNA-PCR were taken for the calculation of the delta CT (difference of expression between *Leishmania* CT-18S rRNA CT). Statistical analysis {#Sec7} -------------------- The statistical analysis was performed using the SPSS 22.0 for Windows software (SPSS Inc., USA). Categorical data were expressed as percentage and statistical analysis was performed using the McNemar\'s test and Fisher's exact test to compare results among related or independent variables, respectively. Quantitative data were expressed as means and standard deviations and a non-parametric Wilcoxon signed-rank test and Mann-Whitney *U*-test were used to compare results among related or independent variables, respectively. Differences were considered significant with a 5% significance level (*P* \< 0.05). Results {#Sec8} ======= Description of clinical data of dogs {#Sec9} ------------------------------------ Both sexes were represented by 11 females and 14 males. The median age was 2.5 years with a range from five months to 10 years. Eleven purebred dogs belonging to ten breeds and 14 mixed-breed dogs were included. Dogs were classified in three clinical stages: stage I-mild disease characterised by persistent papular dermatitis (11 dogs, six females and five males, median age 10 months), II-moderate disease (12 dogs, three females and nine males, median age 54 months) and III-severe disease (two female dogs, median age 54.5 months). For comparative analysis dogs were divided into two groups: group A (11 dogs with stage I) and group B (14 dogs with stage II and III). Age difference was statistically significant among groups (Mann-Whitney *U*-test, *Z* = -2.773, *P* = 0.006). In group A, six dogs were serologically negative, three were low positive and two medium positive, whereas in group B one was low positive, one was medium positive and 12 were high positive. Moreover, dogs from group A had significantly lower levels of *Leishmania* antibodies (136.8 ± 196.1 ELISA units, EU) than dogs from group B (8,892.7 ± 17,807.7 EU; Mann-Whitney *U*-test, *Z* = -3.747, *P* \< 0.0001). Descriptive histopathology {#Sec10} -------------------------- ### Normal-looking skin {#Sec11} The prevalence of microscopic lesions and presence of *Leishmania* by means of HE in normal-looking skin samples are shown in Table [1](#Tab1){ref-type="table"}. The epidermis was normal in all cases but one, with epidermal hyperplasia and ulceration. This case also showed moderate inflammatory infiltrate in the dermis with amastigotes visible with HE-stained sections. The inflammatory pattern observed ranged from perivascular to interstitial mainly in the superficial and mid-dermis in all cases (Fig. [1](#Fig1){ref-type="fig"}). The intensity of the dermatitis was mild to moderate in all cases where inflammation was present. Macrophages with lymphocytes and plasma cells were the predominant cells. In normal-looking skin samples, the detection of intramacrophagic structures compatible with amastigotes was demonstrated in 5/25 (20%) samples, all of them from dogs from group B (Fisher's exact test, *P* = 0.0464) (Fig. [2](#Fig2){ref-type="fig"}).Table 1Frequency of microscopic lesions and detection of *Leishmania* by means of HE, IHC and qPCR on paired skin samples from the dogs studied based on disease stage. Values with the same superscript differ significantlySkin samplesMicroscopic lesionsDetection of *Leishmania*HEIHQqPCRNormal-looking skin (*n* = 25)14/25 (56.0 %)5/25 (20%)8/25 (32.0%)18/25 (72.0%) Stage I (*n* = 11)3/11 (27.3%)^a,b^0/11 (0%)^c^1/11 (9.1%)^d^5/11 (45.5%)^e^ Stage II-III (*n* = 14)11/14 (78.6%)^b^5/14 (35.7%)^c^7/14 (50.0%)^d^13/14 (92.9%)^e^Clinically-lesioned skin (*n* = 25)25/25 (100%)11/25 (44.0%)23/25 (92.0%)25/25 (100%) Stage I (*n* = 11)11/11 (100%)^a^1/11 (9.1%)^f^9/11 (81.8%)11/11 (100%) Stage II-III (*n* = 14)14/14 (100%)10/14 (71.4%)^f^14/14 (100%)14/14 (100%)*Abbreviations*: *HE* haematoxylin and eosin stained sections, *IHC Leishmania* immunohistochemistry, *qPCR* quantitative PCR^a^McNemar\'s test: *P* = 0.008^b^Fisher's exact test: *P* = 0.0172^c^Fisher's exact test: *P* = 0.0464^d^Fisher's exact test: *P* = 0.0421^e^Fisher's exact test: *P* = 0.0068^f^Fisher's exact test: *P* = 0.0037 Fig. 1Superficial and mid perivascular to interstitial dermatitis in normal-looking skin from a dog with stage II leishmaniosis (haematoxylin and eosin staining) Fig. 2Numerous intracellular *Leishmania* amastigotes in macrophages (*arrows*) from the inflammatory infiltrate present in the dermis of normal-looking skin sample from a dog with stage II leishmaniosis (haematoxylin and eosin staining) ### Clinically-lesioned skin {#Sec12} The prevalence of microscopic lesions and detection of *Leishmania* by means of HE in clinically-lesioned samples are shown in Table [1](#Tab1){ref-type="table"}. The most common epidermal changes were hyperplasia (20/25), followed by ulceration (8/25) and hyperkeratosis (7/25). Only two samples had normal epidermis. Moderate to severe lympho-plasmacytic and macrophagic infiltrates were noted in the dermis of all patients together with few neutrophils in some patients. The inflammatory pattern observed was nodular to diffuse in 13 samples (nine from group A and four from group B) and perivascular to interstitial in 12 clinically-lesioned samples (two from group A and ten from group B). Therefore, skin samples from group A were more frequently characterised by a nodular to diffuse pattern than skin samples from group B (Fisher's exact test, *P* = 0.0154). Granulomas were only observed in four samples, all of them from group A (Fisher's exact test, *P* = 0.0166) (Fig. [3](#Fig3){ref-type="fig"}). Amastigotes compatible with *Leishmania* were noted in 11/25 (44%) samples. Most of these (10/11) were samples from group B and this difference was statistically significant (Fisher's exact test, *P* = 0.0037).Fig. 3Nodular to diffuse dermatitis with granuloma formation in clinically-lesioned skin from a dog with stage I leishmaniosis (haematoxylin and eosin staining) ### *Leishmania* immunohistochemistry {#Sec13} The prevalence of positive IHC in clinically-lesioned and normal-looking skin samples are shown in Table [1](#Tab1){ref-type="table"}. Amastigotes were noted in 8/25 (32%) normal-looking skin samples. Seven out eight of these samples were from dogs from group B (Fisher's exact test, *P* = 0.0421; Fig. [4](#Fig4){ref-type="fig"}). The majority of positive samples (6/8) had few amastigotes (1--10 per high power field) with one between 11--30 and another with more than 30 per high power field.Fig. 4Few (1--10 per high power field) intracellular *Leishmania* amastigotes (arrows) are visualized in macrophages from the inflammatory infiltrate present in the dermis of normal-looking skin sample from the same dog as in Fig. [1](#Fig1){ref-type="fig"} (*Leishmania*-specific IHC staining) On the other hand, amastigotes were noted in 23/25 (92%) clinically-lesioned skin samples. Two samples with negative IHC were from dogs from group A. Although marginally statistically significant, there was a trend for a higher parasite load in clinically-lesioned skin from dogs from group B compared with group A (Mann-Whitney *U*-test: *Z* = -1,943, *P* = 0.052; Fig. [5](#Fig5){ref-type="fig"}; Table [2](#Tab2){ref-type="table"}).Fig. 5Note only one intracellular *Leishmania* amastigote (*arrow*) in the center of a granuloma in the inflammatory infiltrate present in the dermis of clinically-lesioned skin from the same dog as in Fig. [3](#Fig3){ref-type="fig"} (*Leishmania*-specific IHC staining) Table 2Parasite load by means of *Leishmania-*specific IHC and qPCR on paired skin samples from the dogs studied based on disease stageSkin samplesIHC^a^\ (mean ± SD)qPCR^g^\ (mean ± SD)Normal-looking skin (*n* = 25)0.4 ± 0.8^b^3.0 ± 4.7^h^ Stage I (*n* = 11)0.1 ± 0.3^c,d^6.1 ± 4.0^i,j^ Stage II-III (*n* = 14)0.5 ± 0.7^d,e^1.7 ± 4.5^j,k^Clinically-lesioned skin (*n* = 25)1.5 ± 0.9^b^1.5 ± 4.9^h^ Stage I (*n* = 11)1.1 ± 0.8^c,f^3.4 ± 4.4^i,l^ Stage II-III (*n* = 14)1.7 ± 0.9^e,f^-0.4 ± 4.7^k,l^*Abbreviations: qPCR*, quantitative PCR *IHC Leishmania* immunohistochemistry, *SD* standard deviation^a^For method of grading, see Methods^b^Wilcoxon Signed-rank test: *Z* = -4.345, *P* \< 0.0001^c^Wilcoxon Signed-rank test: *Z* = -2.887, *P* = 0.004^d^Mann-Whitney *U*-test: *Z* = -2.169, *P* = 0.03^e^Wilcoxon Signed-rank test: *Z* = -3.274, *P* = 0.001^f^Mann-Whitney *U*-test: *Z* = -1.943, *P* = 0.052^g^Delta CT (difference of expression between *Leishmania* CT -18S CT)^h^Wilcoxon Signed-rank test: *Z* = -3.332, *P* = 0.001^i^Wilcoxon Signed-rank test: *Z* = -2.023, *P* = 0.043^j^Mann-Whitney *U*-test: *Z* = -2.021, *P* = 0.043^k^Wilcoxon Signed-rank test: *Z* = -2.691, *P* = 0.007^l^Mann-Whitney *U*-test: *Z* = -2.026, *P* = 0.043 ### qPCR {#Sec14} The normal-looking skin of 18/25 (72%) dogs studied was qPCR positive for *Leishmania* (Table [1](#Tab1){ref-type="table"}). Negative qPCR was almost always associated with a microscopically normal skin. Only one dog presented mild perivascular dermatitis in the deep dermis and qPCR was negative. From 11 samples without histological lesions, five resulted qPCR positive. The prevalence of negative qPCR on normal-looking skin samples from dogs from group A was higher than that detected in normal-looking skin from dogs from group B (Fisher's exact test, *P* = 0.0068). The parasite load studied by means of qPCR in normal-looking skin samples was always lower than in clinically-lesioned skin whatever the stage of disease (Wilcoxon signed-rank test, group A: *Z* = -2.023, *P* = 0.043; group B: *Z* = -2.691, *P* = 0.007; Table [2](#Tab2){ref-type="table"}). The relative amounts of parasites in normal-looking skin from dogs from group A was lower than in normal-looking skin from dogs from group B (Mann-Whitney *U*-test: *Z* = -2.021, *P* = 0.043; Table [2](#Tab2){ref-type="table"}). As expected, 25/25 (100%) of clinically-lesioned skin were qPCR positive and the parasite load was higher in samples from dogs from group B compared with dogs from group A (Mann-Whitney *U*-test: *Z* = -2.026, *P* = 0.043, Table [2](#Tab2){ref-type="table"}). Discussion {#Sec15} ========== In this study, we demonstrated histological and parasite load differences not only among clinically-lesioned and normal-looking skin of the same dogs but also among skin samples of dogs with different clinical stages of leishmaniosis. In agreement with previous studies, we demonstrated that the normal-looking skin of dogs with leishmaniosis frequently shows microscopic lesions (56%) and harbours the parasite, as demonstrated by routine HE staining (20%), *Leishmania*-specific IHC (32%) and, more often, by qPCR (72%). However, there are some differences among our results and those previously reported \[[@CR12]--[@CR15]\]. The prevalence of microscopic lesions and detection of amastigotes either by routine histology or by IHC in our study was at the lower limit of the ranges reported in previous studies. Microscopic lesions have been noticed in 50--100% of the skin samples obtained from the normal-looking skin of dogs with CanL \[[@CR5], [@CR12]--[@CR14]\]. Moreover, amastigotes were seen in up to 100% of the cases, depending on the sensitivity of the method employed \[[@CR5]\]. These findings are probably related to the fact that in the present study about half of the dogs had mild disease, i.e. papular dermatitis. Conversely, previous studies included either dogs with more severe disease, i.e. exfoliative dermatitis \[[@CR14]\] or even stray dogs, which could present co-factors, such as co-infections or malnutrition, affecting the severity of disease \[[@CR12], [@CR13]\]. In the present study, we demonstrated that dogs with different clinical stages of leishmaniosis presented differences in the frequency of microscopic lesions and parasite load in normal-looking skin. The skin biopsies from normal-looking skin from dogs with stage I-mild disease (papular dermatitis) were significantly less frequently inflamed. Furthermore, *Leishmania* was more frequently demonstrated by routine histology, immunohistochemical examination or qPCR in normal-looking skin of dogs with stage II-III-moderate to severe disease. In addition, in the latter group, there was a significantly higher parasite load studied by means of qPCR than in dogs with less severe disease. These results suggest that dermal inflammation and cutaneous parasitism in normal-looking skin were directly related to the severity of clinical disease. Normal-looking skin of dogs with stage I-mild disease may resemble the skin of seronegative infected but clinically healthy dogs that is characterized by no histological lesions and absence of parasites by IHC, although their presence can be demonstrated by PCR \[[@CR12]\]. Microscopic lesions and presence of amastigotes in the inflammatory infiltrate in normal-looking skin of diseased dogs is suggestive of haematogenous dissemination of the parasite and tropism for the skin \[[@CR12]\]. Moreover, it has been demonstrated that dissemination to the skin varies between dogs, being greater in sick and infectious dogs \[[@CR20]\]. Therefore, lack of these changes in the majority of dogs with normal-looking skin with stage I-mild disease would further suggest a protective immune response in these dogs able to control parasite dissemination at the site of parasite inoculation and multiplication as previously proposed \[[@CR11], [@CR18]\]. Histological findings observed in clinically-lesioned skin of dogs included in this study were in accordance with the literature \[[@CR5], [@CR6], [@CR18]\] and amastigotes were variably seen in 44 and 92% of the cases, depending on the method employed. However, the results of this study further confirm that skin biopsies from dogs with papular dermatitis (stage I-mild disease) are characterised by the nodular to diffuse pattern and a significant higher frequency of granuloma formation compared with more severe cutaneous manifestation of CanL (stage II--III- moderate or severe disease) \[[@CR18]\]. It has been proposed previously that there is a trend for a lower parasite burden in skin samples from dogs with stage I-mild disease \[[@CR18]\]. Although amastigotes were more frequently noted in HE stained slides from stage II-III diseased dogs when compared with stage I dogs, there were no statistically significant differences in prevalence between positive IHC or qPCR among both groups studied. Nevertheless, the parasite load studied by means of qPCR was lower in samples from dogs with stage I-mild disease compared with dogs with severe disease. Taken together, these data might reinforce the idea of a protective immune response that these dogs have as described elsewhere \[[@CR10], [@CR11], [@CR18]\]. Several studies have focused on the capacity of dogs to infect phlebotomine sand flies. It has been reported that the proportion of infected sand flies increases with the appearance and severity of the clinical signs and that good predictors of infectiousness are antibody levels and clinical disease, since no dogs have been found to be infectious before the detection of anti-*Leishmania* IgG antibodies \[[@CR21], [@CR22]\]. Moreover, it has been recently suggested that high parasite loads in dog ear skin, rather than the simple presence of parasites, is the most important metric to identify likely infectious individuals and potential reservoir populations \[[@CR20]\]. Therefore, the fact that dogs with stage I-mild disease or papular dermatitis are characterised by reduced parasite load in both normal-looking skin and clinically-lesioned skin, emphasizes the concept that these dogs do not play a significant role in *L. infantum* infection of phlebotomine sand flies as opposed to dogs with stage II-III disease. Conclusions {#Sec16} =========== In conclusion, this study confirms that normal-looking skin from dogs with stage I is less likely to present microscopic lesions as well as harbour the parasite when compared with dogs with moderate to severe CanL. Moreover, clinically-lesioned skin from dogs with stage I shows a lower parasite load than clinically-lesioned skin from more diseased dogs. CanL : Canine leishmaniosis CT : Cycle threshold ELISA : Enzyme-linked immunosorbent assay IHC : Immunohistochemistry QPCR : Quantitative polymerase chain reaction The authors thank all veterinarians and dog owners that contributed to this study. Specially, we are grateful to Dr. Marta Planellas (Hospital Clínic Veterinari of Universitat Autònoma e Barcelona, UAB) and Marta Blanchart (Ars Veterinaria). The authors are also grateful to Dr. Pamela Martínez-Orellana, Dr. Lorena Alborch and Eduardo Dos Santos Silva for their technical help. Publication fees of this manuscript have been sponsored by Bayer HealthCare - Animal Health division (Germany) in the framework of the 12th CVBD World Forum Symposium. Funding {#FPar1} ======= This study was supported by a Spanish ministry grant, Ministerio de Economía y competitividad (AGL2012-32498) and the ESVD Research grant 2012. Dr. Laia Solano-Gallego holds a Ramón y Cajal senior researcher contract awarded by the Ministerio de Ciencia e Innovación (Spain) and the European Social Fund. Availability of data and materials {#FPar2} ================================== The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request. Authors' contributions {#FPar3} ====================== LO and LSG designed the research study. AD, MO, JLL and LO included all the dogs. LO and LSG coordinated the veterinary clinics enrolled. LO performed all the histological and immunohistological work. LO and SM performed the molecular work of this study. LO and LSG contributed with data analysis and interpretation. LO wrote the manuscript. LSG revised the manuscript. All authors read and approved the final manuscript. Competing interests {#FPar4} =================== The authors declare that they have no competing interests. Consent for publication {#FPar5} ======================= Not applicable. Ethics approval and consent to participate {#FPar6} ========================================== A signed informed consent was obtained from all owners of dogs. Ethical approval was obtained by "Comissió d'Ètica en l'Experimentació Animal i Humana de la Universitat Autònoma de Barcelona" (CEEAH 1586, February 2012).

All relevant data are within the paper and its Supporting Information files. Algorithm is available in the protocols.io (dx.doi.org/10.17504/protocols.io.ig7cbzn) and in the GitHub website (<https://github.com/dianapina/Maxillary-Sinus-Quantification>). Introduction {#sec001} ============ The paranasal sinuses are complex anatomical structures with significant interindividual variations \[[@pone.0190770.ref001], [@pone.0190770.ref002]\].The actual function of the maxillary sinus (MS) is largely unknown \[[@pone.0190770.ref003]\]. Previous studies have suggested that the MS lessens the weight of the skull, confers resonance to speech, and warms and moistens inspired air \[[@pone.0190770.ref003], [@pone.0190770.ref004]\]. In this context, volumetric measurements provide important information about the MSs \[[@pone.0190770.ref005], [@pone.0190770.ref006]\]. Comprehensive knowledge of MS development is important for elucidating sinus pathologies and selecting adequate treatments \[[@pone.0190770.ref007]\]. Computed tomography (CT) is the most comprehensive and objective method of measuring disease severity and is very sensitive to changes that are caused by different interventions \[[@pone.0190770.ref007]--[@pone.0190770.ref009]\]. In clinical practice, evaluation based on air-free or modified mucosa volume of the sinuses may help the diagnosis and management of sinusopathies \[[@pone.0190770.ref002], [@pone.0190770.ref010]--[@pone.0190770.ref014]\]. For example, volumetric evaluation contributes to the assessment of patients with chronic rhinosinusitis (CRS) because such a volumetric method better correlates with the severity of CRS compared with most widely used Lund-Mackay CT scoring system \[[@pone.0190770.ref010]\]. Furthermore, MS volume have been evaluated for verifying the response to chemo- and radiotherapy in malignant tumors \[[@pone.0190770.ref011]\] and for planning procedures that involve sinus floor elevation in the placement of implants \[[@pone.0190770.ref012], [@pone.0190770.ref013]\] and endoscopic sinus surgery \[[@pone.0190770.ref002], [@pone.0190770.ref014]\]. The size and shape of the MS has been investigated in many previous studies \[[@pone.0190770.ref001]\]. The literature shows the influence of breathing patterns \[[@pone.0190770.ref004], [@pone.0190770.ref015]\], dental problems \[[@pone.0190770.ref007], [@pone.0190770.ref016]\], anatomical features \[[@pone.0190770.ref002], [@pone.0190770.ref016], [@pone.0190770.ref017]\], gender \[[@pone.0190770.ref001], [@pone.0190770.ref002], [@pone.0190770.ref018]\], age \[[@pone.0190770.ref018], [@pone.0190770.ref019]\], ethnicity \[[@pone.0190770.ref020], [@pone.0190770.ref021]\], and climatic factors \[[@pone.0190770.ref022], [@pone.0190770.ref023]\] on MS volume. Furthermore, many chronological and pathological events can affect MS volume. Studies in literature varied in relation to measurements devices and study objectives \[[@pone.0190770.ref024]\], thereby causing a lack of consensus regarding MS volume results \[[@pone.0190770.ref011], [@pone.0190770.ref025]\]. Among other reasons, this may be attributable to different computed tomography data acquisition techniques, segmentation methods, and focuses of investigation \[[@pone.0190770.ref025]\]. Manual and semiautomated methods are commonly used for the quantification of MS volume for pathological sinuses \[[@pone.0190770.ref005], [@pone.0190770.ref023], [@pone.0190770.ref026], [@pone.0190770.ref027]\], however, such task is labor intensive. Automated techniques may reduce both time and effort and provide relatively quick and easy segmentation, more accuracy, better control, and higher sensitivity \[[@pone.0190770.ref004]\]. The purpose of the present study was to develop an automated tool to quantify the total and air-free volume of the MS using CT images. The tool can quantify sinus volume even under pathological conditions, such as tumors and rhinosinusitis, thus contributing to the clinical management of patients. Therefore, the methodology that was developed herein may potentially replace the current radiologic scores, such as the Lund-Mackay CT scoring system, because of its accuracy in MS volumetry. Furthermore, the present automated tool seeks to standardize MS volume measurements. Methodology {#sec002} =========== Patient selection {#sec003} ----------------- This retrospective study was developed with ethical approval from the authors' institutions and national review panels (protocol no. CAAE 42225115.4.0000.5411). The study involved 30 patients who were randomly selected from a pool of 287 patients who were treated in the Hospital das Clínicas de Botucatu, Brazil, between January 2013 and December 2015. Paranasal sinus CT exams that were evaluated in the present study were indicated for patients with clinically suspected rhinosinusitis and septal deviation. The gender distribution was predominantly male (16 \[53.3%\]). The mean age of the patients was 28.4 ± 5.2 years. The inclusion criteria were patients older than 20 years who underwent a paranasal sinus CT exam. Patients with a history of previous nasal, nasopharyngeal, paranasal sinus, or adenoidectomy surgery, maxillofacial trauma, and congenital nasal abnormalities were excluded from the study. Data acquisition {#sec004} ---------------- Paranasal sinus CT exams without contrast enhancement were evaluated in the study. All the CT scans were acquired using a Toshiba Activion 16 Helicoidal device (Toshiba America Medical Systems, Tustin, CA, USA) with the following parameters: pixel size range of 0.30 mm × 0.30 mm to 0.38 mm × 0.38 mm, 512 × 512 pixel matrix, 3.0 mm increment between slices, 3.0 mm slice width, and 120 kV tube voltage. Volumetric reconstructions were made from raw data using a 0.5 mm slice width and 0.3 mm interval between slices. Automated tool {#sec005} -------------- An automated algorithm was developed using Matlab R2013a (Mathworks, Natick, MA, USA) for the volumetric quantification of the MS based on CT exams. The algorithm is available in the protocols.io repository (dx.doi.org/10.17504/protocols.io.ig7cbzn) and in website GitHub (<https://github.com/dianapina/Maxillary-Sinus-Quantification.git>). It was initially necessary to detect and segment MS areas in the CT exams. [Fig 1](#pone.0190770.g001){ref-type="fig"} shows this process in one CT slice using a hybrid methodology. The algorithm is defined as the following: 1. CT image was read ([Fig 1A](#pone.0190770.g001){ref-type="fig"}). 2. The original image was thresholded to remove soft tissue and mucous membrane thickening, cysts, and/or fluid, highlighting bone regions from the MS ([Fig 1B](#pone.0190770.g001){ref-type="fig"}). In this step, the threshold was 150 Hounsfield Units (HU). 3. The resulting thresholded image was then fine-tuned using morphological image processing operators \[[@pone.0190770.ref028]\]. In this step, an opening operation (i.e., erosion followed by dilation) was applied to fill the boundaries of the MS and remove small-sized areas ([Fig 1C](#pone.0190770.g001){ref-type="fig"}) \[[@pone.0190770.ref029]\], thus automatically reducing sparse voxels that were attributable to imaging noise \[[@pone.0190770.ref028]\]. 4. The watershed technique was applied, which can be classified as a region-based segmentation approach \[[@pone.0190770.ref030], [@pone.0190770.ref031]\]. This step computes a complete partition of the image into basins. The watershed was then determined by boundary detection. [Fig 1D](#pone.0190770.g001){ref-type="fig"} shows segmentation by the watershed technique, in which each gray level represents different basins. 5. A rule-based system was applied to compare the major areas by assessing data on the position, shape, and symmetry of the segmented regions (basins) and selecting only the MS areas ([Fig 1E](#pone.0190770.g001){ref-type="fig"}). 6. The areas of free-air and involvement (mucous membrane thickening, cysts, and/or fluid) were classified ([Fig 1F](#pone.0190770.g001){ref-type="fig"}; represented by blue and red regions, respectively). This process was performed using the threshold technique. The range of attenuation of air in the MS was set between -200 and -1200 HU \[[@pone.0190770.ref001]\]. {#pone.0190770.g001} The methodology described above was initially applied to the middle slice of the CT exam. The immediately lower and upper slices were also evaluated using this methodology. This process was repeated until the rule-based system was no longer satisfied. Thus, the entire CT exam was assessed, resulting in a volumetric region of interest. The total and air-free volumes of the MS were measured by multiplying the number of voxels in the volumetric region by the voxel volume, which could be reconstructed into a three-dimensional (3D) image \[[@pone.0190770.ref016]\]. The volume of both MSs was calculated in all of the CT exams. [Fig 2A](#pone.0190770.g002){ref-type="fig"} illustrates the 3D shaded surface of the MS, highlighting MS involvement. The MS inside the reconstructed head is shown in [Fig 2B](#pone.0190770.g002){ref-type="fig"}. {#pone.0190770.g002} Validation by manual segmentation {#sec006} --------------------------------- To evaluate the accuracy of the automated MS volume quantification based on CT images, the results from the same 30 patients were compared with manual segmentation that was performed by an experienced radiologist (J.M.C.A.) using a standard procedure. Each MS was segmented by carefully tracing the outlines of the MS while following the inner bone surface, proceeding in an axial direction \[[@pone.0190770.ref011]\]. In MSs with involvement, the air-free area was also segmented. The MS areas were segmented in all slices of the CT exam, thus defining a volumetric region of interest. The total and air-free volumes of the MS were measured by multiplying the number of voxels in the volumetric region by the voxel volume, which was obtained from the DICOM header of the CT images. All of the data were measured in cm^3^, and the measurements were performed by the same radiologist to prevent possible interobserver variability \[[@pone.0190770.ref027]\]. An example of the left MS manual segmentation process of one slice is shown in [Fig 3](#pone.0190770.g003){ref-type="fig"}. In this step, the radiologist segmented the MS boundaries (blue line) and free-air areas (green line). {#pone.0190770.g003} Statistical linear regressions and mean percent differences between the automated and manual methods were performed for the same CT exam. Comparisons between volume quantifications were performed using Bland-Altman statistics \[[@pone.0190770.ref032]\] to assess agreement between the presently developed algorithm and the reference standard, with a confidence interval of 95%. Results {#sec007} ======= The automated volume quantification method that was developed herein and manual segmentation method were compared based on the same 30 patient examinations, for a total of 60 MSs. An average total MS volume of 14.7 ± 4.4 cm^3^ was found in the evaluated patients. The raw data for the quantification are presented in [S1 Table](#pone.0190770.s001){ref-type="supplementary-material"}. The mean percent difference between both methods was 7.19% ± 5.83% and 6.93% ± 4.29% for total and air-free MS volumes, respectively. For total MS volume quantification, the linear regression ([Fig 4A](#pone.0190770.g004){ref-type="fig"}) was y = 0.96x -- 0.22, with a correlation coefficient of *R^2^* = 0.96. The statistical Bland-Altman plot is shown in [Fig 4B](#pone.0190770.g004){ref-type="fig"} for the difference and average between the automated and manual methods. {#pone.0190770.g004} For air-free MS quantification, the linear regression ([Fig 5A](#pone.0190770.g005){ref-type="fig"}) was y = 0.95x + 0.088, with a correlation coefficient of *R^2^* = 0.98. The statistical Bland-Altman plot is shown in [Fig 5B](#pone.0190770.g005){ref-type="fig"}. {#pone.0190770.g005} Figs [4](#pone.0190770.g004){ref-type="fig"} and [5](#pone.0190770.g005){ref-type="fig"} show good agreement between the automated and manual methods for both total MS volume and air-free volume within the MS. The Bland-Altman analyses showed no significant differences. For total MS volume ([Fig 4B](#pone.0190770.g004){ref-type="fig"}), the bias was -0.77 cm^3^, with -2.63 cm^3^ and 1.10 cm^3^ as limits of agreement at a 95% confidence interval. For MS air-free volume ([Fig 5B](#pone.0190770.g005){ref-type="fig"}), the bias was -0.68 cm^3^, with -2.34 cm^3^ and 0.99 cm^3^ as limits of agreement at a 95% confidence interval. Discussion {#sec008} ========== The size of the human paranasal sinus was initially determined by taking anatomical measurements, injecting different materials into cadavers, or performing plain radiography \[[@pone.0190770.ref011], [@pone.0190770.ref023], [@pone.0190770.ref033]\]. The introduction of CT and magnetic resonance imaging has allowed more precise assessments of these structures \[[@pone.0190770.ref001], [@pone.0190770.ref011], [@pone.0190770.ref015]\]. CT is widely used for evaluating the volume of the paranasal sinuses \[[@pone.0190770.ref001], [@pone.0190770.ref007], [@pone.0190770.ref009], [@pone.0190770.ref010], [@pone.0190770.ref034]\]. Although magnetic resonance imaging is superior to CT in rendering soft tissue, its use is limited by its relatively high cost and restricted accessibility \[[@pone.0190770.ref015]\]. The major advantage of CT is the excellent osseous anatomic detail that it provides, highlighting the boundaries of the MS. Therefore, CT is the gold-standard imaging modality for inflammatory diseases of the paranasal sinuses \[[@pone.0190770.ref009], [@pone.0190770.ref034]\]. Previous studies have evaluated gender differences in MS volume, in which MS volume is significantly larger in males than in females \[[@pone.0190770.ref001], [@pone.0190770.ref002], [@pone.0190770.ref018]\]. However, Pirner *et al*.reported no significant differences between male and female MS volumes \[[@pone.0190770.ref005]\]. Some studies have shown that MS volume is significantly less in patients with chronic rhinosinusitis compared with controls \[[@pone.0190770.ref016], [@pone.0190770.ref033]\]. However, Fernández *et al*. showed that MS volumes in chronic rhinosinusitis patients were larger than in the control group \[[@pone.0190770.ref011]\]. Other studies have evaluated the influence of dentition status on MS volume. Cho *et al*. found that a group of patients with abnormal teeth presented no difference in MS volume compared with patients with normal teeth \[[@pone.0190770.ref016]\]. In contrast, Möhlhenrich *et al*. found that dentition status influenced the volume of the MS \[[@pone.0190770.ref026]\].Because of the lack of a gold-standard technique for quantifying the volume of the paranasal sinuses, no consensus has been reached regarding such measurements \[[@pone.0190770.ref011], [@pone.0190770.ref024]\]. These studies also did not take into account possible influences of instrumental, physical, and human limitations that may exist, which could cause MS volume measurements to be different from the actual value \[[@pone.0190770.ref011], [@pone.0190770.ref025]\]. In clinical practice, CT-based sinus volumetry has been reported to be a useful tool for objectively evaluating sinus disease \[[@pone.0190770.ref002], [@pone.0190770.ref010]--[@pone.0190770.ref014]\]. Pallanch *et al*. evaluated the total percent volume of sinus disease based on CT and the Lund-Mackay scoring system in patients who were being medically treated for CRS. Volumetric scoring using CT exams had a better correlation with disease severity (i.e., symptoms, endoscopic scoring, and quality of life) compared with Lund-Mackay scoring. These results show that sinus volumetry can contribute to the clinical management of CRS patients. However, as indicated by these authors, a tool is still needed to automatically segment sinus volume to reduce the effort that is required for manual segmentation \[[@pone.0190770.ref010]\]. To our knowledge, only manual or semiautomated methods have been applied for paranasal sinus segmentation. These techniques present inter- and intraobserver variability and are both time-consuming. Deeb *et al*. attempted to perform 3D volumetric measurements based on CT scans of the MS in patients with chronic rhinosinusitis using image analysis software. However, their method was too cumbersome to evaluate the whole sinus in a large number of patients \[[@pone.0190770.ref028]\]. Kirmeier *et al*. tested a semiautomated volumetric analysis technique for MS quantification. They achieved good results with their time-consuming measurement procedure, supporting its applicability for clinical evaluations of small changes in MS volume following sinus augmentation or tooth extraction. However, they concluded that a reasonable goal would be to develop a fully automated sinus volume determination technique with high accuracy \[[@pone.0190770.ref025]\]. Thus, CT-based volume determinations can currently be performed accurately and effectively, but a fully automated method may have better applicability in clinical practice \[[@pone.0190770.ref007]\]. In the present study, we developed an automated method for MS volume measurements based on CT images. The statistical comparisons between the automated and manual quantification methods revealed strong agreement and low dispersion between variables. These promising findings were maintained over the entire range of MS volume evaluation, with no increase in quantification error as the MS volume varied between patients. Furthermore, the mean percent difference between the automated and manual techniques was approximately 7% for both measurements. These differences were sufficiently small to yield the same level of confidence for both quantification methods. The average total MS volume found was within the range of published mean volumetric measurements for adult patients (10.9 ± 2.8 cm^3^ \[[@pone.0190770.ref004]\] to 24.7 ± 9.0 cm^3^ \[[@pone.0190770.ref028]\]). Importantly, our database was composed of relatively young patients (mean age, 28.4 ± 5.2 years). Some studies have reported an age-related reduction of volume \[[@pone.0190770.ref006], [@pone.0190770.ref018], [@pone.0190770.ref019]\]. Our volumetric findings for the MS within the same age range are similar to the results that were reported by Jun *et al*. (18.6 ± 7.8 cm^3^) \[[@pone.0190770.ref019]\] and Park *et al*. (14.8 ± 1.5 cm^3^) \[[@pone.0190770.ref035]\]. However, Kawarai *et al*. evaluated patients with a mean age of 29.5 years and reported volumes of 21.3 ± 6.5 cm^3^ and 23.1 ± 6.7 cm^3^ for the right and left MSs, and these volumes were greater than those found in the present study \[[@pone.0190770.ref006], [@pone.0190770.ref019]\]. The combination of different image processing techniques using a fully automated hybrid method indicates the novelty of this study. Our method was shown to accurately quantify MS volume, including both total and air-free volumes. The tool allows comprehensive assessments of the MS while taking into account MS involvement, which is determined based on the relationship between air-free and total volume. In pathological sinuses, defining pixel intensity is difficult because of their non-homogeneous constitutions of bone, air, and mucosa. This occurs because of the anatomical complexity of the paranasal sinuses \[[@pone.0190770.ref005]\]. However, the tool was shown to be suitable for quantifying the volume of involvement in pathological sinuses and evaluating mucous membrane thickening, cysts, fluid, tumors, and other materials with different densities. This is a powerful advantage because the available methods for MS volume quantification are unsuitable for pathological sinuses \[[@pone.0190770.ref005]\]. Advances in multislice CT equipment have enabled high-resolution scans with consequently high structural definition, mainly due to the ability to produce exams with several thin slices \[[@pone.0190770.ref036]\]. Since the MS is imaged using \~120 slices per exam, manual segmentation requires approximately 2 hours to complete because of the required per-slice user interaction. The present automated tool was able to quantify MS volume in approximately 3 minutes per exam, showing to be less time-consuming. Furthermore, the presently developed tool does not require complex or expensive equipment. The algorithm may be applied using conventional computers, thus allowing better implementation in clinical practice. The present study has some limitations. Our methodology was analyzed using only one protocol with a slice thickness of 0.5 mm. Prionas *et al*.\[[@pone.0190770.ref037]\] reported a greater error of volume quantification for thicker slices. Moreover, our objective was to validate our tool, and the MS volumes that are presented herein should not be considered a cohort study. Further studies are needed to evaluate MS volume in patient groups with different ages, genders, and ethnicities. The study is limited to the MS, and therefore its role in functional endoscopy surgery is yet to be proven. Nevertheless, the automated tool may be adapted to quantify volume in other paranasal sinuses. In conclusion, the present study found a good correlation between the manual and automated MS volume estimation techniques. Our automated measurements of MS volume based on CT exams were reliable, robust, and accurate compared with the manual method. Our findings suggest that this automated tool may be applied in clinical practice. It does not require substantial user expertise, and it is reproducible and fast. Our tool may allow comparisons between different patient groups by standardizing measurements of MS volume while obviating the variability that is inherent in different measurement procedures. Such standardization is extremely important for comparisons between studies. The tool may also be applied to determine the factors that influence MS pneumatization. Furthermore, MS volumetry using our method is feasible even under pathological conditions, which might contribute to a better clinical assessment of the extent of nasal pathology. Supporting information {#sec009} ====================== ###### Raw data for the maxillary sinus volumetry by automatic and manual quantifications. (DOCX) ###### Click here for additional data file. [^1]: **Competing Interests:**The authors have declared that no competing interests exist.

All relevant data are within the paper. Introduction {#sec001} ============ Chronic neck pain and neck shoulder stiffness are common symptoms in the general population \[[@pone.0210802.ref001]\]. One study showed that radiographic degenerative changes in the cervical spine were associated with the severity of neck pain in a general population \[[@pone.0210802.ref002]\]. However, degenerative changes in the cervical spine are also common in asymptomatic individuals, challenging the simple concept of cause and effect \[[@pone.0210802.ref003]\]. The number of subluxations and the incidence and severity of degenerative changes in the cervical spine increase with age \[[@pone.0210802.ref004], [@pone.0210802.ref005]\]. Typical changes include osteoarthritis of the facets with reduced joint space and disc narrowing. Cervical myelopathy, a common degenerative spinal disease that interferes with normal activities \[[@pone.0210802.ref006]\], partly involves compression of the cervical spinal cord by spinal canal stenosis. Cervical spinal canal stenosis can arise from developmental canal stenosis, intervertebral disc protrusion into the spinal canal, or thickening of the ligamentum flavum \[[@pone.0210802.ref007]\]. Degenerative cervical changes can also appear in the intervertebral discs, and these changes progress with age \[[@pone.0210802.ref008]\] \[[@pone.0210802.ref009]\]. Epidemiologic studies have shown that the prevalence of cervical canal stenosis increases with age \[[@pone.0210802.ref010]\] \[[@pone.0210802.ref011]\]. Reported risk factors for cervical disc degeneration (CDD) include age \[[@pone.0210802.ref009]\] \[[@pone.0210802.ref002]\] \[[@pone.0210802.ref012]\], genetics \[[@pone.0210802.ref013]\], bone metabolism \[[@pone.0210802.ref012]\], smoking \[[@pone.0210802.ref014]\], metabolic syndrome \[[@pone.0210802.ref015]\], manual labor \[[@pone.0210802.ref016]\], and lumbar spinal disorders \[[@pone.0210802.ref017]\] \[[@pone.0210802.ref018]\]. Lumbar disc degeneration is associated with atrophy of the paraspinal muscles \[[@pone.0210802.ref019]\] \[[@pone.0210802.ref020]\]. In contrast, a study in Japanese asymptomatic subjects found no association between the cross-sectional area of cervical muscles and degeneration of the cervical discs \[[@pone.0210802.ref017]\]. Thus, the relationship between CDD and the strength of the neck, trunk, and extremities muscles in the general population remains unclear. Here, we hypothesized that muscle weakness is related to CDD. To test this hypothesis, we investigated the associations between CDD and the strength of various muscles in subjects recruited from the general population in a Japanese community. The obtained results may help clinicians determine specific preventive measures for disc degeneration. Methods {#sec002} ======= Participants and study design {#sec003} ----------------------------- The subjects were recruited through a community-based public health project in a small city in northern Japan. For 10 years, this project has provided annual health checkups to the general population and made the services of physicians, surgeons, orthopedists, gynecologists, urologists, psychiatrists, dermatologists, and dentists available to the community. This group is suitable for a cohort study, because it is in a rural area with little change in the population. Each year, the program serves about 1000 city residents \> 19 years of age \[[@pone.0210802.ref021]\]. Our research on neck complaints was conducted as part of this project. In 2015, we recruited volunteers from the 1112 people who participated in the community health project, as described in detail previously \[[@pone.0210802.ref022]\]. All of the participants filled out questionnaires about their past medical history, lifestyle, fitness habits, occupational history, family history, and health-related quality of life. They also provided disease-specific information and described any symptoms related to the cervical spine and extremities. All the participants were examined for neurological status (deep tendon reflex), height (Ht), weight, individual body resistance, bone status including bone mineral density, and biochemical tests. Individuals were excluded if they had a history of stroke, cerebral bleeding, cervical spine trauma or surgery, or any systemic disease involving the cervical spine (such as rheumatoid arthritis), or if they did not complete the questionnaire. Cervical spine MRIs were evaluated randomly with respect to age. All volunteers gave informed, written consent before participating. This cross-sectional survey was approved by the ethics committee of the Hirosaki University Graduate School of Medicine. Measurement of muscle mass {#sec004} -------------------------- Body resistance (R) was measured at 50 kHz using the Tanita MC-190 body composition analyzer (Tanita Co., Tokyo, Japan). Skeletal muscle mass (SM) was calculated using Janssen's regression equation, which is based on the relationship between bioelectrical impedance analysis and SM measured by MRI \[[@pone.0210802.ref023]\], as follows: SM (kg) = \[(Ht^2^/R × 0.401) + (gender × 3.825) + (age × −0.071)\] + 5.102, where Ht is in centimeters, R in ohms, and age in years. Gender value is male = 1 and female = 0. The coefficient of determination (*r*^2^) in this regression equation was 0.86; the standard error of the estimate (SEE) was 2.7 kg or 9%. The skeletal muscle index (SMI) was calculated as SM/Ht^2^ × 10^2^ (kg/m^2^) to standardize differences influenced by height. ### Assessment of muscle strength {#sec005} To assess the associations between disc degeneration and muscle strength, trunk and leg strength were measured as described in detail previously \[[@pone.0210802.ref024]\] and neck and handgrip strength were additionally measured in this study. To measure the strength of the neck muscles, we used a device consisting of a torque machine with a MicroFET2 dynamometer (Nihon MEDIX Inc., Chiba, Japan). Isometric neck-muscle strength was measured with the subject in the prone position for extension and supine position for flexion as described in detail previously \[[@pone.0210802.ref025]\]. Isometric muscle strength was measured as peak torque (N) using the maximum pushing force on the pad of the MicroFET2. This torque value was adjusted by the subject's body weight (N/kg). Strength in the trunk muscles was measured using an iron frame combined with a QTM-06b as described in detail previously \[[@pone.0210802.ref024]\] \[[@pone.0210802.ref026]\]. Isometric trunk muscle strength was measured in both extension and flexion as peak torque (Nm) with the maximum pushing force on the QTM-06b. This torque value was adjusted by the subject's body weight (Nm/kg). Handgrip strength (in kg) was measured with a handheld dynamometer with the subject standing upright. The better of two trials was used for each hand. Isometric muscle strength in the lower extremities was measured with an S-13129 (Takei Scientific Instruments Co., Ltd, Niigata, Japan) with the knee joint stabilized at a 90° angle as described in detail previously \[[@pone.0210802.ref024]\]. According to the determined arm length, 0.175 m, the peak force (kg) was used to calculate the peak torque (Nm/kg), adjusted by the subject's body weight. MRI procedures {#sec006} -------------- All MRI studies used a mobile MRI unit (Intera Achieva 1.5 T; Philips, Amsterdam, Netherlands) with a 1.5-Tesla (T) superconducting imager and phased array coils. Cervical spine MRIs were obtained with the subject in a supine position and the following imaging protocol: sagittal T2-weighted fast-spin echo (FSE): repetition time (TR) 4000 ms/echo, echo time (TE) 200 ms; field of view (FOV) 300 × 320 mm); and axial T2-weighted FSE: TR 4000 ms/echo; TE 120 ms; FOV 180 × 180 mm. Sagittal T2-weighted images were used to assess the intervertebral spaces from C2/3 to C7/T1. ### Evaluation of CDD on cervical MRI {#sec007} Intervertebral disc degeneration was evaluated on MRIs at all cervical levels from C3/4 to C7/T1, using Matsumoto's classification system \[[@pone.0210802.ref008]\]. The signal intensity of the invertebral disc was graded as follows: Grade 0, as bright as or slightly less bright than the cerebrospinal fluid; Grade 1, markedly darker than the cerebrospinal fluid; Grade 2, no signal. Disc space narrowing was graded as a percentage of the height of a healthy upper disc, as follows: Grade 0, 100--75%; Grade 1, 75--50%; and Grade 2, less than 50%. CDD severity was scored by adding the Matsumoto grades for signal intensity and disk space narrowing for all intervertebral sections from C2/3 to C7/T1. The total was defined as the degenerative score, with 0 corresponding to the normal cervical disc condition and 20 to the most severely degenerative disc condition \[[@pone.0210802.ref012]\]. Statistical analysis {#sec008} -------------------- SPSS ver. 12.0J was used for data input and statistical calculations (SPSS Inc., Chicago, IL, USA). Differences in age, BMI, degenerative score, SMI, and muscle strength results between men and women were assessed by the Mann--Whitney *U* test, and differences in exercise habits and the prevalence of smoking were analyzed by the Chi-square test. Correlations between the degenerative score and age, SMI, or muscle strength were analyzed by Spearman's rank correlation coefficient. Stepwise multiple linear regression analyses were conducted using the degenerative score as the dependent variable. The independent variables used were age, BMI, and a muscle-strength parameter, for each gender. For all analyses, a *P* value \< .05 was considered significant. Results {#sec009} ======= Among the volunteers from our public health-project population, 151 men (mean age 54.2) and 193 women (mean age 55.5) were enrolled in the study. The youngest of the 344 participants was 20, and the eldest was 86 years old ([Table 1](#pone.0210802.t001){ref-type="table"}). [Table 1](#pone.0210802.t001){ref-type="table"} summarizes the participants' age, smoking, exercise habits, BMI, SMI, degenerative score, and muscle strength. There was no significant difference in age between the men and women. The men had a significantly higher prevalence of smoking (*P* \< .0001) and a higher BMI (*P* \< .001). Although there was no significant difference in mean degenerative score between the men and women, the men had a significantly higher mean SMI and had significantly more strength in the neck, trunk, hands, and legs ([Table 1](#pone.0210802.t001){ref-type="table"}). 10.1371/journal.pone.0210802.t001 ###### Characteristics, degenerative score, and muscle strength of study participants. {#pone.0210802.t001g}   Men (*n* = 151) Women (*n* = 193) *P* value[^b^](#t001fn003){ref-type="table-fn"} ------------------------------------------------------------- ----------------- ------------------- ------------------------------------------------- ------------ Age[^a^](#t001fn002){ref-type="table-fn"}, y 54.3 ± 15.1 55.6 ± 14.6 .432 20--39 30 38   40--59 60 68   60--79 59 82   80--99 2 5   Current Smoker, *n* (%) 33 (21.9) 18 (9.3) \< .0001^\#^ Exercises regularly, *n* (%) 37 (24.5) 47 (24.4) .916 BMI[^a^](#t001fn002){ref-type="table-fn"}, kg/m^2^ 23.9 ± 3.3 22.3 ± 3.3 \< .001\* SMI[^a^](#t001fn002){ref-type="table-fn"}, kg/m^2^ 18.9 ± 1.6 15.4 ± 1.5 \< .001\* Degenerative score[^a^](#t001fn002){ref-type="table-fn"} 8.4 ± 3.5 8.1 ± 3.4 .509 Neck strength[^a^](#t001fn002){ref-type="table-fn"}, N/kg (*n* = 151) (*n* = 191) *Flexion* 1.2 ± 0.4 0.9 ± 0.5 \< .0001\* *Extension* 1.8 ± 0.6 1.6 ± 0.7 .003\* Trunk strength[^a^](#t001fn002){ref-type="table-fn"}, Nm/kg (*n* = 125) (*n* = 159) *Flexion* 2.1 ± 0.7 1.6 ± 0.6 \< .0001\* *Extension* 5.2 ± 1.3 3.5 ± 1.2 \< .0001\* Handgrip strength[^a^](#t001fn002){ref-type="table-fn"}, kg (*n* = 126) (*n* = 163) *Right side* 40.5 ± 8.0 23.8 ± 4.7 \< .0001\* *Left side* 38.8 ± 7.9 23.0 ± 4.6 \< .0001\* Leg strength[^a^](#t001fn002){ref-type="table-fn"}, Nm/kg (*n* = 119) (*n* = 151) *Flexion* 1.5 ± 0.4 1.1 ± 0.3 \< .0001\* *Extension* 2.8 ± 0.8 2.0 ± 0.5 \< .0001\* BMI, body mass index. SMI, skeletal muscle index. ^a^Mean ± S.D. ^b^Significant differences (*P* \< .05) between values for men and women were calculated by \*Mann-Whitney *U* or ^\#^Chi-square test. Single correlation analyses revealed a significant positive correlation between degenerative score and age for both sexes and SMI in women ([Table 2](#pone.0210802.t002){ref-type="table"}) and significant negative correlations between the degenerative score and the strength of trunk flexion in both sexes, handgrip (both sides) in men, and handgrip (right side) and leg flexion and extension in women ([Table 2](#pone.0210802.t002){ref-type="table"}). 10.1371/journal.pone.0210802.t002 ###### Relationships between degenerative score and muscle strength. {#pone.0210802.t002g} Partial correlation coefficient: Degenerative score ---------------------------------- -------------------- ------------ ------------ ----- **Men** * * * *   Age 0.637 \< .0001\* 151 SMI, kg/m^2^ −0.06 .935 151 Neck strength, kg *Flexion* 0.102 .214 151 *Extension* -0.040 .629 151 Trunk strength, Nm/kg *Flexion* −0.247 .006\* 125 *Extension* −0.011 .903 126 Handgrip strength, kg *Right side* −0.178 .046\* 127 *Left side* −0.254 .004\* 127 Leg strength, Nm/kg *Flexion* −0.177 .054 119 *Extension* −0.158 .087 119 **Women**       Age 0.683 \< .0001\* 193 SMI, kg/m^2^ 0.233 .001\* 193 Neck strength, kg *Flexion* -0.048 .510 190 *Extension* -0.021 .776 190 Trunk strength, Nm/kg *Flexion* −0.272 .001\* 158 *Extension* -0.051 0.523 157 Handgrip strength, kg *Right side* −0.169 .032 162 *Left side* −0.152 .054 162 Leg strength, Nm/kg *Flexion* −0.343 \< .0001\* 148   *Extension* −0.190 .020\* 150 SMI: Skeletal muscle index. r = correlation coefficient. The relationship between degenerative score for Age, SMI and muscle strength were analyzed by Spearman's rank partial correlation analysis. P-Values below 0.05\* indicate significance. Stepwise multiple linear analyses showed a significant correlation between the degenerative score and age in both men (B = 0.179, 95%CI: 0.139--0.218) and women (B = 0.162, 95%CI: 0.128--0.196), but not between the degenerative score and any muscle strength measure ([Table 3](#pone.0210802.t003){ref-type="table"}). In this analysis, the muscle-strength parameters were those found to be associated with CDD in the single correlation analysis (trunk flexion and handgrip in men, and trunk flexion and leg strength in women). 10.1371/journal.pone.0210802.t003 ###### Stepwise multiple regression analysis relative to degenerative score[^a^](#t003fn002){ref-type="table-fn"}. {#pone.0210802.t003g} Logistic regression (degenerative score) B β 95% CI *P* value ------------------------------------------ ----------------------- -------- -------- ----------------- ---------- **Male** Age 0.179 0.802 0.139 to 0.218 \< .0001 BMI, kg/m^2^ 0.106 0.107 -0.065 to 0.276 0.222 Trunk strength, Nm/kg *Flexion* 0.460 0.098 -0.334 to 1.255 0.253 Handgrip strength, kg *Right side* 0.047 0.113 -0.054 to 0.147 0.360 *Left side* -0.082 -0.192 -0.183 to 0.020 0.114 **Female** Age 0.162 0.704 0.128 to 0.196 \< .0001 BMI, kg/m^2^ 0.095 0.089 -0.066 to 0.256 0.247 Trunk strength, Nm/kg *Flexion* 0.012 0.002 -0.763 to 0.786 0.977 Leg strength, Nm/kg *Flexion* -0.033 -0.003 -2.092 to 2.026 0.975 *Extension* 0.292 0.048 -0.762 to 1.345 0.585 B, regression coefficient; β, standardized regression coefficient; *r*^2^, coefficient of determination (adjusted). ^a^Stepwise multiple regression analysis was performed by gender using the degenerative score as the dependent variable and age, BMI, and muscle strength as independent variables. The muscle-strength parameters were those found to be correlated with the degenerative score in the single correlation analysis (trunk flexion and handgrip in men, and trunk flexion and leg strength in women). Discussion {#sec010} ========== To our knowledge, this is the first community-based survey of associations between CDD and muscle strength in a rural Japanese population. We showed that the strength of trunk flexion in men and women, hand grip in men, and leg strength in women were negatively correlated with CDD scores in single-variable correlation analyses. Multiple linear regression analyses that comprehensively included age and limb or trunk muscle strength showed that age was the strongest independent factor associated with CDD in both sexes, while the associations of limb and trunk muscle strength were attenuated. Our findings that age was the strongest determinant of CDD in both sexes is in agreement with an MRI based study in the Japanese population.They found that progression of cervical-spine degeneration. Another study using MRI showed that the progression of cervical-spine degeneration on MRI was frequently observed during a 10-year period, along with symptom development in healthy volunteers \[[@pone.0210802.ref009]\]. In that study, the only factor related to the progression of cervical-spine degeneration was age. In addition, a genetic association study found age to be the most significant determinant of lumbar disc degeneration \[[@pone.0210802.ref027]\]. We previously assessed CDD on X-rays and demonstrated that age was correlated with CDD \[[@pone.0210802.ref002]\]; we also found signficant correlations between CDD and cross-linked N-telopeptide of type 1 collagen as a bone metabolism marker and isoleucine as an amino acid marker in men, and lysine as an amino acid marker in a general community population \[[@pone.0210802.ref012]\]. These studies collectively indicate that in assessing the cervical spine, clinicians should pay particular attention to age-related bone metabolism factors that may be associated with disc degeneration. In the present study, single-variate correlation analyses showed that the muscle strength of trunk flexion in both sexes, hand grip in men, and leg strength in women were negatively correlated with CDD. Okada et al. reported that CDD is associated with changes in the cross-sectional area of the extensor muscles of the cervical spine on MRIs \[[@pone.0210802.ref017]\], but not with changes in the extensor muscle volume. Hakkaku et al. reported that the repeated crash forces experienced by athletes subjected to contact and external stress are risk factors for CDD \[[@pone.0210802.ref025]\]. Other studies have investigated associations between disc degeneration and muscles in the lumbar spine \[[@pone.0210802.ref013]\] \[[@pone.0210802.ref028]\]. Videman et al. reported that lumbar disc degeneration was associated more strongly with body weight, lifting strength, and axial disc area than with a history of physically demanding work or activities \[[@pone.0210802.ref028]\]. Therefore, physical loading and mechanical stress may be crucial steps in the progression of discdegeneration. We recently reported that lumbar spondylosis and the muscle strength of the trunk are associated with locomotive syndrome, in which individuals exhibit deteriorating locomotorium and thus may require nursing or other support \[[@pone.0210802.ref024]\]. Taken together, weaker trunk muscle strength has been considered to be related to the progression of CDD. Notably, in the present cross-sectional study when we considered the association among CDD, age, and muscle strength at the same time, we confirmed that the muscle strength of the limb and trunk decreased linearly with age, and that the correlation between CDD and muscle strength was attenuated in the middle to elderly population. Based on the current results, a longitudinal study of the cervical spine following the same individuals might provide clues to the underlying cause of the disc degeneration, as long as researchers are careful to be aware of the strong linearity between age and muscle strength. This study had several limitations that should be noted. First, we did not evaluate the duration, region, or distribution of neck symptoms. Progressive structural changes over a long period of time are likely to correlate significantly with future clinical symptoms. Second, our study population was geographically limited to a district with many farming villages, and thus may not be representative of Japan as a whole. Therefore, lifestyle factors such as occupation and hobbies should be considered, and caution should be applied when generalizing our results to other populations. Third, we did not investigate the subjects' level of education or other aspects of their medical history, such as bone metabolism \[[@pone.0210802.ref012]\], metabolic syndrome \[[@pone.0210802.ref015]\], or lumbar spinal disorders \[[@pone.0210802.ref017]\] \[[@pone.0210802.ref018]\] in this study. In the present study, the strength of trunk flexion in men and women, hand grip in men, and leg strength in women were negatively correlated with CDD scores in a single-variate correlation analysis. On the other hand, multiple linear regression analyses in which age and muscle strength were included simultaneously, showed that age was the most significant determinant of CDD, and that the effects of muscle strength were attenuated and not significant. Although a longitudinal study is needed to determine the strategies by which enhancing muscle strength will prevent CDD, our current findings suggest that researchers should take the colinearity of age and muscle strength into account when they examine these strategies epidemiologically. [^1]: **Competing Interests:**The authors have declared that no competing interests exist.

Background ========== Eukaryotic genes are often interrupted by intragenic, non-coding sequences called introns \[[@B1]\]. However, prokaryotic genes lack introns. Therefore, \'Intronless\' genes are characteristic features of prokaryotes. Interestingly, many eukaryotic histone \[[@B2],[@B3]\] and GPCR \[[@B4]\] genes are predominantly \'intronless\'. A number of vertebrate \'intronless\' genes have been complied \[[@B5]\]. The human genome report identified 901 Otto predicted single exon genes (The Celera approach to gene prediction is called Otto) \[[@B6]\]. The presence of a sizeable amount of single exon genes (SEG) in eukaryotic genomes is intriguing. The SEGE database contains eukaryotic SEG derived from GenBank \[[@B7]\]. For most genomes, SEGE does not provide representative \'intronless\' gene sets because GenBank often contains redundant sequences from the same species deposited by different authors. It should also be noted that all sequences obtained from genome projects are not available in GenBank. Representative sets of SEG from specific genomes will provide meaningful biological insights to subsequent bio-computational analysis for comparative and evolutionary studies. In order to facilitate such research we developed Genome SEGE, a database containing all putative SEG from completely sequenced eukaryotic genomes. Here, we describe the usefulness and construction of Genome SEGE. Construction and content ======================== Data source and methodology --------------------------- The annotated eukaryotic genome sequence data was downloaded from NCBI \[[@B8]\]. \'Intronless\' genes were identified using the \'CDS\' annotation in the FEATURE as described elsewhere \[[@B7]\]. It should be noted that organellar sequences (annotated as \'chloroplast\', \'plastid\', \'mitochondrial\', \'mitochondrion\') were removed from further analysis. A flowchart describing the construction of the database is shown (Figure [1](#F1){ref-type="fig"}). \'Intronless\' pseudogenes -------------------------- Data processing and cleaning is an essential part of biological knowledge discovery. Hence, we eliminated all identifiable processed pseudogenes by scanning for polyadenylation signal (AATAAA) and polyadenylation tail using a modified procedure of Harrison and colleagues \[[@B9]\]. In this procedure, by definition, we consider a sequence to represent a pseudogene if it contains a polyadenylation tail (\>15A) within 1000 nucleotides from the stop codon with a preceding polyadenylation signal. Prosite motifs and InterPro --------------------------- We characterized \'intronless\' gene products using PROSITE, which is a method of identifying the functions of uncharacterized proteins translated from genomic sequences \[[@B10]\]. We chose PROSITE because it is complete, highly specific, fully documented and regularly updated. The search tool provides pre-computed PROSITE motifs for each sequence in the database with appropriate hyperlinks to InterPro \[[@B11]\]. Content ------- A Database is created to store all eukaryotic \'intronless\' sequences derived from completely sequenced genomes. The database contains three sets of data for each genome: (1) \'intronless\' sequence set, (2) \'intronless\' pseudo-genes sequence set, and (3) \'intronless\' sequence set without pseudo-genes. Caveats ------- Genome annotation is an inherently dynamic process in which it is necessary to use many different sources of data, which are not updated in a rigorous fashion. It should also be noted that annotation is not generally uniform and consistent because various procedures are used by different groups for genome annotation. During genome annotation, a gene may have been annotated with a single exon CDS in the FEATURE for three main reasons: (1) the gene is truly \'intronless\' and functional, (2) the gene is of retroposition origin \[[@B12],[@B13]\], (3) false positive prediction by gene finding algorithms. False positives are not removed from the current dataset due to lack of a methodology. Nevertheless, the gene finding algorithms are reasonably optimized to find SEG. Update ------ The database will be refreshed on a quarterly basis or as and when an update is noticed to genome files in the public domain. Utility and discussion ====================== Genome SEGE is an extension of SEGE \[[@B6]\] and these two databases complement each other in their biological utility and application. SEGE and Genome SEGE differ primarily in their content, as the datasets are created from different sources. The degree and quality of annotation also varies between them. SEGE could be used for general purpose studies involving \'intronless\' genes from different genomes, while Genome SEGE is of particular interest for researchers interested in comparative genomics. A wealth of information can be obtained by comparing \'intronless\' gene sequences between two or more genomes to identify features conserved or diverged during evolution. Comparison of more closely related genomes can reveal similarities in gene order. Such analysis could also shed light on genome architecture and help understand why the genome is arranged the way it is and how its structure affects function. A systematic mapping between functional genes and their \'intronless\' paralogs can provide a matrix for genomic rearrangement and gene duplication. Different \'intronless\' gene sets available in the database will provide an opportunity to perform many-to-many comparison between genomes. Such analysis will provide information on paralogy and orthology at a molecular level. Analysis of the datasets using non-linear probabilistic models may provide acceptable evidence for retro-position events during evolution. The search tool in the database provides options to scan through each dataset using gene name or protein name. The result page produces information on chromosomal location, organism name, gene name, product name, GenBank Index, nucleotide sequence and protein sequence. The result page also shows all PROSITE motifs in the sequence with specific hyperlinks to PROSITE documentation and InterPro (Figure [2](#F2){ref-type="fig"}). Conclusions =========== The biological role of \'intronless\' genes in the genomes of higher organism is perplexing. \'Intronless\' gene sets available in the database will be of use for subsequent bio-computational analysis in comparative genomics and evolutionary studies. Such analysis may help to revisit the original genome data for re-examination and re-annotation. Different eukaryotic genomes have varying proportions of \'intronless\' genes and a sizeable fraction of them are found in many intron-rich multi-cellular genomes. We believe that these estimates will improve our understanding on the differential selection (as a process or force) of \'intronless\' genes in different eukaryotic genomes. The different datasets made available in the database can serve as a data source for evolutionary and functional studies. They will also help to answer questions such as, (1) How many of \'intronless\' genes are expressed in each genome? (2) How many of them are of prokaryotic origin? (3) How many of them have multi-exon correspondence within genome? (4) Do they evolve by retro-position? It is our hope that the database we make available will encourage molecular biologists and computational molecular evolutionist to address this problem. The unique features that distinguish Genome SEGE from SEGE is the service providing representative \'intronless\' datasets for completely sequenced genomes. Such service will persuade researchers to use representative data sets for investigating a number of biologically significant evolutionary phenomena. We also hope to provide this service for other completely sequenced genomes as and when they are available in the public domain after appropriate examination and analysis. It is also our interest to compare the contents of SEGE and Genome SEGE on a genome by genome basis for the examination of data bias in SEGE. Availability and requirements ============================= Database is available freely at <http://sege.ntu.edu.sg/wester/intronless>. List of abbreviations ===================== GPCR G-protein coupled receptors SEGE Single exon genes in eukaryotes CDS Coding sequence Acknowledgements ================ We thank Vincent Chow Tak Kwong, Iti Chaturvedi, Subbiah Subbramanian and Dmitri A. Petrov for their contribution and discussion. This research is supported by A\*STAR-BMRC research grant \#01/1/21/19/191. We thank the anonymous referees for their suggestions, which helped us make the manuscript contextual. Figures and Tables ================== {#F1} {#F2}

Introduction ============ Hepatitis B virus (HBV) infection is one of the most common viral infections in humans. Approximately 2 billion people have been infected with HBV and 350 million of them became chronically infected. Individuals with chronic hepatitis B infection are at an increased risk of developing liver cirrhosis, hepatic decompensation, and hepatocellular carcinoma (HCC); 15%--40% of these individuals will develop these serious sequelae during their lifetime ([@b1]; [@b30]). Drugs that are currently approved by the Federal Drug Administration (FDA) for the treatment of chronic HBV consist of two groups: the immunomodulators such as conventional interferon alpha and pegylated interferon alpha-2a, and nucleoside--nucleotide analogs such as lamivudine, adefovir dipivoxil, and entecavir. However, not all patients with chronic hepatitis B infection respond to these treatments. The currently approved antiviral regimens, especially the nucleoside--nucleotide analogs, have been shown to improve the short-term outcome of disease but lack the ability to provide cure or induce durable remission in most patients with chronic HBV ([@b26]). Lamivudine, the first nucleoside--nucleotide analog to be approved for the treatment of chronic HBV, has a favorable safety profile but long-term therapy with this drug can lead to the selection of drug-resistant mutants ([@b42]; [@b34]). The risk of mutation increases with the duration of treatment. At the end of the first, second, third, and fourth year of treatment, the incidences of resistance are: 15%--32%, 38%, 56%, and 67% respectively ([@b27]). Adefovir dipivoxil is another oral nucleoside--nucleotide analog that requires long-term therapy. Initial reports show that adefovir dipivoxil selects resistant mutants in only a limited proportion of patients, but adefovir-resistant mutants do occur at cumulative rate of 18% by 192 weeks ([@b28]). A phase III study on entecavir showed that entecavir is superior to lamivudine in hepatitis B e antigen (HBeAg) positive and HBeAg negative nucleoside--nucleotide naïve patients ([@b4]; [@b40]). Entecavir, with its profound suppression of serum HBV DNA, has less risk of resistance over time with no resistance at 96 weeks in HBeAg positive lamivudine-naïve patients ([@b10]). Entecavir resistance in lamivudine-resistant patients is 7% genotypically and 1% phenotypically at 1 year. But follow-up studies to determine the resistance rate after 2--5 years of therapy are required, especially in lamivudine-resistant patients. Another major limit is the lack of knowledge on how long nucleoside--nucleotide therapy needs to be continued, especially in HBeAg negative patients. This group of patients has been shown to have persistent benefit with 144 weeks of adefovir dipivoxil therapy ([@b13]). Furthermore, sustained response after withdrawal of entecavir has also been shown to be less than optimal in both HBeAg positive and HBeAg negative patients ([@b3]; [@b20]). Therefore, in view of the limitations of current therapies for chronic HBV, there is a need to develop new agents for the treatment of chronic HBV infection with improved efficacy. Immunomodulators ================ Conventional interferon alpha ----------------------------- Conventional interferon alpha-2b is the first drug to be approved by the FDA for treatment of chronic HBV infection. However, the efficacy of conventional interferon alpha, defined as sustained loss of HBeAg and HBV DNA, is limited. In a meta-analysis of 15 randomized, controlled trials, loss of HBeAg and HBV DNA in HBeAg positive patients is seen in 33% and 37% on conventional interferon alpha-treated patients compared with 12% and 17% of untreated patients, respectively ([@b42]). The studies reviewed in this meta-analysis employed direct spot hybridization for the detection of serum HBV DNA. In Caucasians, the long-term durability of HBeAg is as high as 90% ([@b34]), while around 20%--70% of patients with loss of HBeAg and hepatitis B e antibody (anti-HBe) seroconversion will eventually lose hepatitis B surface antigen (HBsAg) ([@b34]; [@b22]). In those with detectable HBV DNA after HBeAg seroconversion, the HBV DNA will be undetectable in 60%--100% of those who lose HBsAg. In HBeAg negative variants (precore mutants and others), prolonged interferon at a dose of 3--5 MU thrice weekly for at least 12 months results in a sustained biochemical remission in 15%--25% of patients ([@b31]; [@b35]). Factors that predict a favorable response to conventional interferon alpha include low pretreatment level of HBV DNA (\<200 pg/ml), high levels of serum aminotransaminase (\>100 U/L), and evidence of necroinflammatory activities in the liver ([@b2]). In contrast, male sex, length of chronic state, Asian origin, precore mutants, and human immunodeficiency virus coinfection are factors associated with poor response to conventional interferon alpha ([@b38]). In a recent study from Taiwan, patients with genotype B were more likely to respond to conventional interferon alpha than those with genotype C ([@b18]). A small study on the use of conventional interferon beta at 3 MU weekly for 24 weeks also showed a 50% HBeAg seroconversion rate similar to that achieved with conventional interferon alpha ([@b17]). However, conventional interferon beta has never been recommended for use as a first-line agent for the treatment of chronic HBV due to a lack of large-scale randomized studies. However, conventional interferon alpha is somewhat tedious and requires daily or 3 subcutaneous injections a week with a high occurrence of side-effects. Furthermore, the rate of achieving HBeAg seroconversion in Asian patients has been low ([@b6]). This difference between Asian and Caucasian patients is thought to be related to the duration of the chronic state, the difference in genotype, and baseline characteristics such as serum alanine aminotransaminase (ALT) levels. While Asians acquire HBV perinatally, Caucasians acquire HBV predominantly in their adolescence or adulthood. In perinatally acquired infection, infection is followed by a lengthy period of immune tolerance during which the HBV DNA is high while the serum ALT levels are normal or near normal and liver necroinflammation is minimal ([@b21]; [@b29]; [@b6]). In the latter, there is more active host immune response directed towards clearance of the infection with raised ALT levels ([@b36]). These drawbacks with conventional interferon alpha therapy have led to the development of pegylated interferon alpha. Pegylated interferon ==================== Pegylation ---------- Pegylated interferon alpha-2a (40 kDa) joins a number of therapeutic agents that are pegylated by the incorporation of a polyethylene moiety into the active product. Pegylation of the interferon alpha molecule is undertaken mostly to enhance the pharmacokinetic properties of unmodified interferon alpha, which will enable once a week dosing. Although a larger pegylation molecule can result in better stability and longer half-life, it will interfere with the active receptor binding ([@b44]). Pegylation may occur at multiple sites of the interferon molecule, but pegylation at most of them does not retain the biochemical and therapeutic properties. Therefore it is very important to strike a balance between the pegylation site, size of the molecule, steric properties, and biochemical properties ([@b44]). For example, the molecular weight of the pegylation chain must be greater than 4000 to avoid poisoning by ethylene glycol ([@b43]). The sites of attachment may be more than one single site, but multiple chains may lead to steric hindrance and prohibit the binding of the peglyated interferon to its effector ([@b14]). The bonds must be strong and resistant to degradation ([@b41]). The pegylation polymer is usually covalently attached, via an amide or urethane bond, to a lysine or histidine residue or the N-terminus of the protein. By controlling the reagents, condition, and the pH, pegylation of interferon at a specific site with a defined pegylated polymer can be achieved ([@b19]; [@b44]). The first pegylated interferon alpha-2a to be developed was 5 kDa in size. However, this drug has limited overall clinical and laboratory benefits. Since then, pegylated interferon alpha-2a (40 kDa) and pegylated interferon alpha-2b (12 kDa) have been developed. As a result of their difference in size and structure, these two molecules have different in vivo and in vitro characteristics. Pegylated interferon alpha-2a (40 kDa) has a longer half-life and is mainly catabolized in the liver and has active breakdown products. Pegylated interferon alpha-2b (12 kDa) is a smaller molecule, has a shorter half-life, and acts as a pro-drug depot by slowly releasing interferon ([@b41]; [@b19]). Pegylated interferon alpha-2a (40 kDa) -------------------------------------- The pegylation of interferon alpha-2a involves 2 chains of 20 kDa polyethylene glycol conjugated to the lysine residues (position 31, 121, 131, and 134) of the interferon alpha-2a molecule. The plasma level reaches its peak between 72 and 96 hours after administration and the volume of distribution is 8--12 L, suggesting it is highly compartmentalized in the intravascular space. The clearance half-life is 40--80 hours. The serum antiviral activity, as measured by the 2'-5' oligoadenylate synthetase activity, peaks at 24--48 hours after administration and it remains high for 1 week ([@b37]). Because of its highly intravascular compartmentalization, dose adjustment according to body weight is not necessary ([Table 1](#tbl1){ref-type="table"}). ###### Comparison between conventional interferon alpha-2a and pegylated interferon alpha-2a Conventional interferon alpha 2a Pegylated interferon alpha-2a ---------------------------------- ---------------------------------- ------------------------------- Time to peak serum level (hours) 7.3--12 80 Absorption half-life (hour) 2.3 50 Volume of distribution (L) 31--73 8--12 Clearance (L/hour) 6.6--29.2 0.06--0.10 Elimination half-life (hours) 3--8 65 Pegylated interferon alpha-2a (40 kDa) in HBeAg positive patients ----------------------------------------------------------------- When pegylated interferon alpha-2a was tested in a phase II study at 90, 180, and 270 μg/week for 24 weeks against conventional interferon alpha-2a, the HBeAg seroconversions were 37%, 35%, 29%, and 25% respectively ([@b8]). The combined response (HBeAg loss, HBV DNA suppression \< 500 000 copies/mL, ALT normalization) was higher in all pegylated interferon alpha-2a doses combined (24% vs 12%). The response was still higher among patients that were difficult to treat: 27% in patients with \< 2 times upper limit of normal (ULN) of baseline ALT vs 11% of interferon alpha-2a; 20% vs 0% in patients with HBV DNA \>11.0 log copies/mL. The side-effects seemed to be dose-dependent and occurred more often in the 270-μg and 180-μg groups. However, no difference in side-effect could be observed when the 270-μg group was compared with the 180-μg group. The beneficial effect of pegylated interferon alpha-2a was further substantiated in 2 multinational phase III studies ([@b33]; [@b25]). In the phase III HBeAg positive study, 814 HBeAg positive chronic HBV-infected patients were randomized to receive either pegylated interferon alpha-2a 180 μg weekly monotherapy, pegylated interferon alpha-2a 180 μg weekly plus lamivudine 100 mg daily combination therapy, or lamivudine 100 mg daily for a total of 48 weeks and assessed at 24 weeks after the end of therapy ([@b25]). More than 85% of patients in this study were Asians, and the mean HBV DNA was 9.9--10.1 log copies/mL. About 15%--18% of patients had severe fibrosis or cirrhosis by liver biopsy at baseline, and 9%--15% had received lamivudine therapy and 2%--3% of patients had previously been treated with conventional interferon alpha-2a. HBeAg seroconversion and suppression of HBV DNA to less than 100 000 copies/mL were significantly higher with pegylated interferon alpha-2a monotherapy and pegylated interferon alpha-2a plus lamivudine combination therapy when compared with lamivudine monotherapy ([Table 2](#tbl2){ref-type="table"}). More importantly, loss of HBsAg with development of hepatitis B surface antibody (anti-HBs) was achieved in 8 of the 271 patients (3.0%) on pegylated interferon alpha-2a monotherapy, 8 of the 271 patients (3.0%) on pegylated interferon alpha-2a plus lamivudine combination therapy, and none of the 272 patients (0%) on lamivudine monotherapy (p=0.004 for both comparisons) ([@b25]). ###### Efficacy of pegylated interferon alpha-2a on hepatitis B e Antigen positive chronic hepatitis B virus patients ---------------------------------------------------------------------------------------------------------------------------------------------------- Pegylated interferon alpha-2a plus placebo\ Pegylated interferon alpha-2a plus lamivudine\ Lamivudine\ (n = 271) (n = 271) (n = 272) ------------------------------ -------------------------------------------------- ------------------------------------------------- ------------- -- Co-primary endpoints HBeAg 32% (p \< 0.001)[a](#tf2-1){ref-type="table-fn"} 27% (p = 0.023) 19% seroconversion HBV DNA \< 100 000 copies/mL 32% (p = 0.012)[a](#tf2-1){ref-type="table-fn"} 34% (p = 0.003)[a](#tf2-1){ref-type="table-fn"} 22% Secondary endpoints HBeAg loss 34% (p \< 0.001)[a](#tf2-1){ref-type="table-fn"} 28% (p = 0.043)[a](#tf2-1){ref-type="table-fn"} 21% ALT normalization 41% (p = 0.002)[a](#tf2-1){ref-type="table-fn"} 39% (p = 0.006)[a](#tf2-1){ref-type="table-fn"} 28% ---------------------------------------------------------------------------------------------------------------------------------------------------- Adapted from [@b25]. compared with lamivudine therapy. Pegylated interferon alpha-2a (40 kDa) in HBeAg negative patients ----------------------------------------------------------------- In another randomized, partially double-blind phase III controlled study, 537 HBeAg negative chronic HBV patients were randomized to receive either pegylated interferon alpha-2a 180 μg weekly, combination pegylated interferon alpha-2a180 μg weekly plus lamivudine 100 mg daily, or lamivudine 100 mg daily monotherapy for 48 weeks and followed up for another 24 weeks after therapy ([@b33]). Patients were included into this trial if they had been HBeAg negative and anti-HBe positive for at least 6 months, had an HBV DNA of more than 100 000 copies/mL, a serum ALT level greater than 1 but less than 10 times the upper limit of normal, and had findings on liver biopsy within the previous 24 months showing evidence of prominent necroinflammatory activity. The 2 primary end points assessed at 24 weeks after the completion of therapy of this study were normalization of serum ALT and suppression of HBV DNA below 20 000 copies/mL. After 48 weeks of therapy, suppression of serum HBV DNA from baseline was the greatest with combination pegylated interferon alpha-2a plus lamivudine therapy. On the other hand, suppression of HBV DNA from the baseline was similar in patients on pegylated interferon alpha-2a monotherapy and lamivudine monotherapy. At 24 weeks after therapy, normalization of serum ALT was higher in patients receiving pegylated interferon alpha-2a monotherapy (59%) and combination pegylated interferon alpha-2a plus lamivudine therapy (60%) when compared with those receiving lamivudine monotherapy (44%). Virologic response was also higher in patients receiving pegylated interferon alpha-2a monotherapy (43%) and combination pegylated interferon alpha-2a plus lamivudine (44%) than in patients receiving lamivudine monotherapy (29%). Suppression of HBV DNA to below 400 copies/mL at week 72 was also higher in those receiving pegylated interferon alpha-2a monotherapy (19%) and combination pegylated interferon alpha-2a plus lamivudine therapy (20%) when compared with those on lamivudine monotherapy alone (7%) ([@b33]). Most importantly, loss of HBsAg occurred in 7 patients receiving pegylated interferon alpha-2a monotherapy (5 Asians and 2 Caucasians) and in 5 patients receiving combination pegylated interferon alpha-2a plus lamivudine therapy (4 Asians and 1 Caucasian). This was significantly higher compared with lamivudine monotherapy alone (n = 0) (p = 0.007 and p = 0.030 respectively). HBsAg clearance with development of anti-HBs occurred in 8 patients on pegylated interferon alpha-2a (5 on pegylated interferon alpha-2a monotherapy and 3 on combination pegylated interferon alpha-2a plus lamivudine therapy) compared with none in patients receiving lamivudine monotherapy (p = 0.029) ([@b33]). Combination pegylated interferon alpha-2a (40 kDa) plus lamivudine therapy -------------------------------------------------------------------------- Disappointingly, data generated from 2 studies do not support the use of combination therapy with pegylated interferon alpha-2a and lamivudine in terms of achieving a sustained off-treatment response ([@b33]; [@b25]). In both phase III HBeAg positive and HBeAg negative studies, the degree of viral load suppression at the end of treatment was higher in those on a lamivudine-containing regimen than those on pegylated interferon alpha-2a alone (7.2 log vs 4.5 log respectively in the HBeAg positive study and 5.0 log vs 4.1 log respectively in the HBeAg negative study), but the rate of sustained disease remission was higher in the latter ([@b33]; [@b25]). This finding suggests that the mechanism of viral load reduction, in addition to the degree of viral suppression, is an important factor affecting sustained disease remission. However, one benefit of combination pegylated interferon alpha-2a plus lamivudine therapy is a lower YMDD mutation (1%--4%) compared with lamivudine-only therapy (18%--27%) ([@b9]). Predictors of response ---------------------- Patients infected with genotype A had the highest HBeAg seroconversion at 24 weeks after pegylated interferon alpha-2a (± lamivudine) therapy (52%) compared with patients infected with genotype B or C (30%--31%), but the rate of the gentotype C group was still better than with lamivudine-only therapy ([@b5]). Patients with a high baseline ALT level and low HBV DNA are also more likely to achieve sustained response with pegylated interferon alpha-2a therapy. High baselineALT levels (\>5 times ULN) and low HBV DNA (\<9.1 log copies/mL) achieved HBeAg seroconversion rates of 41% and 53% respectively ([@b5]). The authors also found that a more profound HBeAg suppression at week 12 of therapy (less than 10 IU/mL) was associated with a higher HBeAg seroconversion (53%). A prior use of other antiviral therapies (lamivudine, or conventional interferon alpha) does not preclude patients from treatment with pegylated interferon alpha-2a, as the response rates were similar to those without such use before ([@b24]). Durability of off-therapy sustained response -------------------------------------------- In a longer follow-up study on the durability of off-therapy response with pegylated interferon alpha-2a ([@b32]), 177 HBeAg negative patients with biochemical and virological response at 6 months after the completion of 48 weeks of pegylated interferon alpha-2a were rolled over into a long-term observational study. The rates of biochemical and virologic response measured 12 months after the end of treatment with pegylated interferon alpha-2a monotherapy were similar to those reported 6 months after the end of treatment: 59% vs 59% for ALT normalization; 42% vs 43% for HBV DNA \< 20 000 copies/mL; and 17% vs 19% for HBVDNA \< 400 copies/mL. In a subanalysis of those patients who responded to pegylated interferon alpha-2a monotherapy at the end of treatment, more than half (75%) had HBV-DNA levels \< 100 000 copies/mL for most of the 12 month follow-up; 30% had HBV-DNA levels \< 20 000 copies/mL, and 15% had HBV-DNA levels permanently \< 400 copies/mL. Effect on liver histology ------------------------- The effect of pegylated interferon alpha-2a on liver histology was analyzed by Lau et al and Cooksley et al ([@b7]; [@b23]). Both studies found that pegylated interferon alpha-2a therapy can result in histological improvement (defined as drop of 2 points in the Modified Histologic Activity index) ([@b16]). Forty-nine percent of HBeAg positive and 59% of HBeAg negative patients treated with pegylated interferon alpha-2a had histological improvement on second liver biopsy at 24 weeks after the end of therapy ([@b7]). Histologic improvement is more pronounced in patients with virological response. Thus, HBeAg positive patients who had achieved ALT normalization, HBV DNA suppression, and HBeAg seroconversion are more likely to have histologic improvement. Similarly, HBeAg negative patients with HBV DNA suppression or ALT normalization also showed improvement in liver histology. HBeAg negative patients achieving a combined response (normalization of serum ALT and suppression of HBV DNA) had a higher histological response (78% vs 49%) ([@b23]). Adverse effects --------------- The drop-out rates were low in both phase III studies (2%--7%). Most patients (about 80%) finished the prescribed dose ([@b24]). More patients given pegylated interferon alpha-2a suffered from at least 1 adverse effect (88%--89% vs 48%--56%). Most of them had fever, fatigue, headache, myalgia, alopecia, and injection site reaction. About 4%--6% of patients had serious adverse effects as a result of pegylated interferon alpha-2a monotherapy and pegylated interferon alpha-2a plus lamivudine combination therapy, and 2%--3% of lamivudine monotherapy had a serious adverse effect. Four deaths occurred in the combination therapy but 3 of them were not related to the treatment. The only death that was probably related was the development of thrombotic thrombocytopenia purpura. Two patients in the lamivudine monotherapy had liver failure resulting in 1 liver transplantation and 1 death. Optimal duration of therapy with pegylated interferon alpha-2a (40 kDa) ----------------------------------------------------------------------- Two studies have demonstrated the efficacy of 48 weeks of pegylated interferons alpha-2a either as monotherapy or in combination with lamivudine for the treatment of HBeAg positive and negative chronic HBV infection ([@b8]; [@b33]; [@b25]), but it is uncertain if a shorter duration of treatment with pegylated interferons will affect the sustained virological response rate. This is because the current licensed duration of therapy with conventional interferon alpha is 16--24 weeks. At the moment, no direct comparison between 24 weeks and 48 weeks of pegylated interferon alpha for chronic HBV infection has been performed. In a recent review of our experience in treating HBeAg positive Chinese patients in Hong Kong with either 48 weeks of pegylated interferon alpha-2a or 24 weeks of pegylated interferon alpha-2b, we found that those treated with 48 weeks of pegylated interferon alpha-2a had a higher sustained virological response, defined as HBeAg seroconversion with serum HBV DNA less than 10^5^ copies/mL at week 72 (34% vs 8% respectively, p = 0.04) ([@b15]). However, owing to the small sample size, the use of different pegylated interferon alphas and the retrospective nature of this study, the results should be interpreted with caution and show the need for a large-scale randomized prospective study comparing 24 with 48 weeks of pegylated interferon alpha in order to determine its optimal duration of therapy. Conclusions =========== Pegylated interferon alpha-2a will have an important role in the treatment of chronic HBV infection. The choice of pegylated interferon alpha-2a as a first-line therapy for chronic HBV is based mostly on its efficacy in inducing off-therapy sustained disease remission ([@b33]; [@b25]). Newer nucleoside--nucleotide analogs such as entecavir and telbivudine with a more pronounced and rapid suppression of HBV replication are expected to be approved later. Higher or more pronounced suppression of HBV DNA may be achievable with these drugs. However, their rates of HBeAg seroconversion after 48 weeks of therapy do not seem to be more pronounced than that achieved with lamivudine or adefovir dipivoxil. It is also uncertain if these new nucleoside--nucleotide analogs can lead to off-therapy sustained disease remission. A finite course of pegylated interferon alpha-2a with an increased rate of virologic and biochemical response coupled with its improved off-therapy response makes pegylated interferon alpha-2a a first-line therapy for chronic HBV. In those who do not respond to pegylated interferon alpha-2a, long-term maintenance therapy with a nucleoside--nucleotide analog either as monotherapy or combination therapy may have to be considered. However, drug resistance and its avoidance is a major obstacle to maintenance therapy. New nucleoside--nucleotide analogs such as entecavir and telbivudine have a more pronounced viral suppression of HBV replication, but their rates of viral resistance during long-term therapy have yet to be evaluated and may be a frequent problem, making them unsuitable for effective long-term therapy ([@b11]; [@b11]). Hence, control of chronic HBV may require long-term therapy consisting of combination therapy. The challenge is to find the correct combination ([@b39]).

Ye R‐Y, Kuang X‐Y, Zeng H‐J, Shao N, Lin Y, Wang S‐M. KCTD12 promotes G1/S transition of breast cancer cell through activating the AKT/FOXO1 signaling. J Clin Lab Anal. 2020;34:e23315 10.1002/jcla.23315 ATCC : American Type Culture Collection BTB/POZ : Bric‐a‐brac, Tram‐track, Broad complex poxvirus zinc finger CDK : cyclin‐dependent kinase CKIs : CDK inhibitors DMEM : Dulbecco\'s modified eagle medium DMSO : dimethyl sulfoxide FBS : fatal bovine serum GABA~B~R : GABA~B~ receptors KCTD12 : potassium channel tetramerization domain containing 12 p‐AKT : phosphorylation of AKT p‐FOXO1 : phosphorylation of FOXO1 p‐Rb : phosphorylated Rb qPCR : quantitative PCR Rb : phosphorylate retinoblastoma TCGA : The Cancer Genome Atlas 1. INTRODUCTION {#jcla23315-sec-0005} =============== Breast cancer is the most frequently diagnosed cancer and the leading cause related with cancer in females, with about 2.1 million new cased and 0.6 million deaths in 2018.[^1^](#jcla23315-bib-0001){ref-type="ref"} Breast cancer has been become a threat to women\'s health.[^2^](#jcla23315-bib-0002){ref-type="ref"} Over the past decades, exploring the specific etiologic factors related with breast cancer never stop. However, the underlying precise mechanisms of tumorigenesis and progression in breast cancer are still poorly known. The tumorigenesis and progression of breast cancer are composed of multiple steps. Among them, sustaining proliferation is the most fundamental step. Accelerated proliferation is usual linked to the uncontrolled cell cycle.[^3^](#jcla23315-bib-0003){ref-type="ref"} During cell cycle, G1/S transition, known as restriction point, is the crucial point that determines whether cells enter into proliferation. In early G1 cyclin‐dependent kinase (CDK), 4/6 is activated by cyclin D, and in late G1, CDK2 is activated by cyclin E. The cyclin‐dependent kinases in turn phosphorylate retinoblastoma (Rb), a tumor suppressor. Phosphorylated Rb (p‐Rb) leads to the dissociation of Rb from E2F, which further results in transcription of S phase genes and facilitating entry into S phase and DNA synthesis.[^4^](#jcla23315-bib-0004){ref-type="ref"} Furthermore, during this step, cells will integrate multiple signals. If the conditions go against division, the restriction point will be activated and cell cycle is arrested. For example, the CDK inhibitors (CKIs) can combine with CDKs/cyclins complex and prevent Rb phosphorylating to halt cell cycle procession.[^5^](#jcla23315-bib-0005){ref-type="ref"}, [^6^](#jcla23315-bib-0006){ref-type="ref"}, [^7^](#jcla23315-bib-0007){ref-type="ref"} The research on cell cycle is increasing in recent years. However, the internal and external factors linked to the activation of cell cycle are still to be investigated. Potassium channel tetramerization domain containing 12 (KCTD12, also named as Pfetin), belonging to KCTD family, was originally identified in human fetal cochlea.[^8^](#jcla23315-bib-0008){ref-type="ref"} It contains two conserved binding domains, a voltage‐gated potassium (K^+^) channel tetramerization T1 domain and a BTB/POZ (Bric‐a‐brac, Tram‐track, Broad complex poxvirus zinc finger) domain.[^9^](#jcla23315-bib-0009){ref-type="ref"} Recent study demonstrated that KCTD12 is the auxiliary submit of GABA~B~ receptors (GABA~B~R) and can alter emotionality and neuronal excitability through GABA~B~ receptor signaling.[^10^](#jcla23315-bib-0010){ref-type="ref"} Not only that KCTD12 plays an important suppressive function and its high level is related with favorable prognosis in gastrointestinal stromal tumors.[^11^](#jcla23315-bib-0011){ref-type="ref"} Abbaszadegan et al showed that KCTD12 can suppress Wnt/Notch signaling, stem cell factors, and chromatin remodelers in esophageal squamous cell carcinoma.[^12^](#jcla23315-bib-0012){ref-type="ref"} Nonetheless, KCTD12 is dramatically upregulated and closely correlated with larger tumor sizes, higher pathologic stages, and poor survival in cervical and lung cancers.[^13^](#jcla23315-bib-0013){ref-type="ref"} Therefore, the expression and function of KCTD12 in cancers remain controversial. And the roles of KCTD12 in breast cancer are unknown. Herein, we found the level of KCTD12 is significantly decreased in breast cancer and cells, and lower level of KCTD12 predicts poorer survival for patients with breast cancer. Further cell function tests illustrated that downregulation of KCTD12 significantly promotes cell proliferation and in vitro tumorigenesis. Besides, molecular biologic experiments showed that downregulation of KCTD12 can enhance the G1/S transition through activating the AKT/FOXO1 signaling; especially, knockdown of FOXO1 can restore the inhibitory effect of KCTD12 on cell proliferation of breast cancer. Accordingly, we inferred that downregulation of KCTD12 can be a novel factor for poor prognosis in breast cancer. 2. MATERIALS AND METHODS {#jcla23315-sec-0006} ======================== 2.1. Patients and cell line {#jcla23315-sec-0007} --------------------------- The tissues used in the study were gathered from The First Affiliated Hospital of Sun Yat‐sen University. Before collecting, informed consent from patients and approval from the Institutional Ethics Committee were both obtained. The normal breast cell HBL100 and breast cancer cell lines MDA‐MB‐231, MCF‐7, BT‐549, SK‐BR‐3, and T47D in the study were purchased from American Type Culture Collection (ATCC). The breast cancer cells were cultured in Dulbecco\'s modified eagle medium (DMEM; Invitrogen) containing 10% fatal bovine serum (FBS; HyClone). 2.2. Plasmid {#jcla23315-sec-0008} ------------ Full‐length KCTD12 was amplified by PCR using the total cDNA as the template and cloned into pLVX plasmid to obtain the pLVX‐KCTD12 plasmid overexpressed KCTD12. To downregulate KCTD12, two human shRNA fragments were cloned into pLKO.1 plasmid. The fragments were as follows: sh\#1, GCGCTACACCTCGCGCTATTA; sh\#2, CTTCCGCTACATCCTGGATTA. psPAX2 (virus‐packaging plasmid), pMD2G (envelope plasmid), and pLVX‐KCTD12/pLKO.1‐shKCTD12 were transfected into 293T cells to produce pseudotyped lentiviral particles. 48 hours later, we harvested the particles, and then, the particles were used to infect MCF‐7 cells overnight. 48 hours after infection, the normal culture media was replaced. Stable cell lines were screened by 5 μg/mL puromycin for 7 days. 2.3. RNA extraction and quantitative PCR {#jcla23315-sec-0009} ---------------------------------------- Total RNA was extracted using the TRIzol reagent (Invitrogen) according to the guidance method. quantitative PCR (qPCR) was performed using ABI Prism 7500 Sequence Detection System (Applied Biosystems). The primers are as follows: KCTD12, forward, 5′‐CCGGAATTCCACCTCTCTGTCATGGCTCT‐3′ and reverse 5′‐CTGCAGAGAACTCAGCACCAAG‐3′; p21, forward 5′‐AGGTGGACCTGGAGACTCTCAG‐3′ and reverse 5′‐TCCTCTTGGAGAAGATCAGCCG‐3′; p27, forward 5′‐CGTCCTCCATAGCAGCCAAGAT‐3′ and reverse 5′‐ACCCAATGGAGCCCAGGATGAA‐3′; CCND1, forward 5′‐TCTACACCGACAACTCCATCCG‐3′ and reverse 5′‐TCTGGCATTTTGGAGAGGAAGTG‐3′; CCND2, forward 5′‐GAGAAGCTGTCTCTGATCCGCA‐3′ and reverse 5′‐CTTCCAGTTGCGATCATCGACG‐3′; CCND3, forward 5′‐AGATCAAGCCGCACATGCGGAA‐3′ and reverse 5′‐ACGCAAGACAGGTAGCGATCCA‐3′; CCNE1, forward 5′‐TGTGTCCTGGATGTTGACTGCC‐3′ and reverse 5′‐CTCTATGTCGCACCACTGATACC‐3′; CCNE2, forward 5′‐CTTACGTCACTGATGGTGCTTGC‐3′ and reverse 5′‐CTTGGAGAAAGAGATTTAGCCAGG‐3′. 2.4. Western blotting assay {#jcla23315-sec-0010} --------------------------- The total proteins were collected using lysis buffer (Cell Signaling Technology) and then maintained on ice for 30 minutes. Before Western blotting assay, the protein was mixed with loading buffer and then boiled at 95°C for 5 minutes. The proteins were separated using SDS‐PAGE gel. The detailed protocol was in accordance with the introduction previously described.[^13^](#jcla23315-bib-0013){ref-type="ref"} 2.5. MTT assay {#jcla23315-sec-0011} -------------- The cells were seeded into 96‐well plates. At the indicated time, the cells were stained using 0.5 mg/mL MTT reagent (Sigma) for 4 hours. Subsequently, 150 μL dimethyl sulfoxide (DMSO) was added into each well. The absorbance was detected at 570 nm and 655 nm as the reference wavelength. 2.6. Colony formation and anchorage‐independent growth ability assay {#jcla23315-sec-0012} -------------------------------------------------------------------- 5 × 10^2^ cells were seeded into 6‐well plates. 10 days later, the colonies formed were fixed using 10% formaldehyde and then dyed using 0.1% crystal violent. Finally, the colony numbers were counted. The anchorage‐independent growth ability assay was performed according to the protocol previously described.[^13^](#jcla23315-bib-0013){ref-type="ref"} 2.7. Flow cytometry assay {#jcla23315-sec-0013} ------------------------- The cells were treated using 75% ethanol for 1 hour and then incubated in propidium iodide staining buffer for 10 minutes away from light. The flow cytometry assay was performed using FACSCalibur instrument (BD Biosciences). The percentages of cells distributed in different cell cycle were analyzed using CellQuest 3.3 software. 2.8. Statistical analysis {#jcla23315-sec-0014} ------------------------- All experiments were performed three times independently. All values were analyzed using SPSS version 21.0 (SPSS) and presented as mean ± standard deviation. Statistical differences were evaluated by Student\'s *t* test, and *P* \< .05 was considered as statistical significance. 3. RESULTS {#jcla23315-sec-0015} ========== 3.1. KCTD12 is significantly decreased in breast cancer and negative correlated with patients' overall survival {#jcla23315-sec-0016} --------------------------------------------------------------------------------------------------------------- Through analyzing the public data from The Cancer Genome Atlas (TCGA; <https://cancergenome.nih.gov/>), we found the expression of KCTD12 is significantly decreased in primary breast cancer tissues (Tumor) relative to normal breast tissues (Normal) (Figure [1A](#jcla23315-fig-0001){ref-type="fig"}). Meanwhile, we analyzed the KCTD12 in paired tissues. Compared with matched normal breast tissue, KCTD12 is significantly reduced in breast cancer tissues (Figure [1B](#jcla23315-fig-0001){ref-type="fig"}). Furthermore, we analyzed the effect of KCTD12 on overall survival of patients with breast cancer using Kaplan‐Meier plotter datasets ([www.kmplot.com](http://www.kmplot.com)). As demonstrated in Figure [1C](#jcla23315-fig-0001){ref-type="fig"}, lower levels of KCTD12 are closely related with poorer survival. {#jcla23315-fig-0001} In addition, further analysis of TGCA dataset revealed that KCTD12 mRNA expression was much lower in triple‐negative breast cancer (TNBC) tissues and non‐TNBC tissues compared with normal breast tissues, but there isn\'t significant difference between non‐TNBC tissues and TNBC tissues (Figure [2A](#jcla23315-fig-0002){ref-type="fig"}). Similarly, the expression of KCTD12 in all molecular subtype of breast cancer is significantly decreased compared with normal breast tissues. Especially, the levels of KCTD12 in the basal‐like subtype of breast cancer, which present similar characteristics with TNBC, are much lower than Luminal A, but not lower than Luminal B and Her2 (Figure [2B](#jcla23315-fig-0002){ref-type="fig"}). According to the above analysis, KCTD12 does not have a significant potential effect on hormone receptor. {#jcla23315-fig-0002} Subsequently, we examined the KCTD12 mRNA expression in normal breast cell HBL100 and breast cancer cell lines MDA‐MB‐231, MCF‐7, BT‐549, SK‐BR‐3, and T47D and found that KCTD12 mRNA levels is significantly decreased in breast cancer cells compared with normal breast cells. Especially, the mRNA expression of KCTD12 in MCF‐7 is medium (Figure [2C](#jcla23315-fig-0002){ref-type="fig"}), so we choose MCF‐7 as a research subject. Finally, we examined the mRNA and protein levels in paired fresh breast cancer tissues. The qPCR assay showed relative KCTD12 mRNA expression is significantly decreased in breast cancer tissues compared with corresponding normal breast tissues (Figure [3A](#jcla23315-fig-0003){ref-type="fig"}). And the Western blotting assays showed the similar trend (Figure [3B](#jcla23315-fig-0003){ref-type="fig"}). {#jcla23315-fig-0003} Altogether, the expression of KCTD12 is significantly reduced in breast cancer tissues and cells, and its low level predicts poor overall survival for patients with breast cancer. 3.2. Downregulation of KCTD12 promotes breast cancer cell proliferation and in vitro tumorgenesis {#jcla23315-sec-0017} ------------------------------------------------------------------------------------------------- Next, we investigate the function of KCTD12 on breast cancer progression. The correlation analysis between KCTD12 levels and proliferative marker Ki67 in fresh breast tissues were performed. As shown in Figure [4A](#jcla23315-fig-0004){ref-type="fig"}, there is negative correlation between them. Subsequently, we screened the cell lines that stably overexpressed or silenced KCTD12 using breast cancer cells MCF‐7, respectively (Figure [4B](#jcla23315-fig-0004){ref-type="fig"}). Furthermore, MTT assay showed that downregulation of KCTD12 dramatically increases, while upregulation decreases breast cancer cell proliferation (Figure [4C](#jcla23315-fig-0004){ref-type="fig"}), which was confirmed by colony formation assay (Figure [4D](#jcla23315-fig-0004){ref-type="fig"}). To examine the oncogenic function of KCTD12 on in vitro tumorignicity of breast cancer cells, we performed the anchorage‐independent growth ability assay. As demonstrated in Figure [4E](#jcla23315-fig-0004){ref-type="fig"}, the colony numbers and sizes significantly increased in KCTD12‐silenced cells, while decreased in KCTD12‐upregulated cells, suggesting that KCTD12 significantly suppresses anchorage‐independent growth ability of MCF‐7 cells. {#jcla23315-fig-0004} Altogether, downregulation of KCTD12 significantly promotes breast cancer cell proliferation and in vitro tumorigenesis. 3.3. Downregulation of KCTD12 significantly promotes the G1/S transition of breast cancer cells {#jcla23315-sec-0018} ----------------------------------------------------------------------------------------------- We further investigate the molecular mechanisms of KCTD12 involved in breast cancer cell proliferation. Firstly, flow cytometry assay illustrated that downregulation of KCTD12 dramatically increases the cell percentage in S phase, but reduces the percentage in G0/G1 phase. Whereas, upregulation of KCTD12 showed the inverse results (Figure [5A](#jcla23315-fig-0005){ref-type="fig"}). The abovementioned results showed that downregulation of KCTD12 markedly promotes the G1/S transition of MCF‐7 cell cycle. {#jcla23315-fig-0005} Moreover, we performed the genes involved in G1/S transition. As shown in Figure [4B](#jcla23315-fig-0004){ref-type="fig"}, the mRNA expression of p21 and p27 significantly increases in KCTD12‐overexpressed cells, while reduces in KCTD12‐silenced cells. CCND1 mRNA showed the adverse variation trend. Nevertheless, there are nearly no change in the mRNA levels of CCND2, CCND3, CCNE1, and CCNE2 in different stable cell lines (Figure [5B](#jcla23315-fig-0005){ref-type="fig"}). Then, we detected the protein expression of p21, p27, and Cyclin D1 using Western blotting assay. As shown in Figure [5C](#jcla23315-fig-0005){ref-type="fig"}, they showed the same variation trend as their mRNA expression. It has been documented that FOXO1 can regulate p21, p27, and Cyclin D1 at the transcription levels.[^14^](#jcla23315-bib-0014){ref-type="ref"}, [^15^](#jcla23315-bib-0015){ref-type="ref"} Western blotting assay showed that knockdown of KCTD12 increases and its upregulation decreases the phosphorylation of FOXO1 (p‐FOXO1). Since FOXO1 is a downstream of the AKT, the phosphorylation level of AKT was analyzed. As shown in Figure [5C](#jcla23315-fig-0005){ref-type="fig"}, downregulation of KCTD12 increases and its upregulation decreases the phosphorylation of AKT (p‐AKT). These results showed that KCTD12 regulates the cell cycle of breast cancer cells through AKT/FOXO1 pathway. Eventually, we used 0.1 μM FOXO1 inhibitor, AS1842856, to inhibit the expression of FOXO1 in KCTD12‐upregulated cells. MTT and colony formation assay showed that FOXO1 inhibitor can significantly restore breast cancer cell proliferation in KCTD12‐upregulated cells (Figure [5D](#jcla23315-fig-0005){ref-type="fig"} and E), suggesting that FOXO1 plays an essential role in the antiproliferative of KCTD12. Altogether, downregulation of KCTD12 significantly promotes G1/S transition of breast cancer cell cycle through the AKT/FOXO1 signaling. 4. DISCUSSION {#jcla23315-sec-0019} ============= The research presented the important evidence that downregulation of KCTD12 promotes the breast cancer cell proliferation and in vitro tumorigenesis. Our findings also showed that knockdown of KCTD12 inhibits the expression of p21 and p27, while upregulates the expression of Cyclin D1. These results provided evidence that downregulation of KCTD12 plays an indispensable function in enhancing breast cancer cell proliferation. Uncontrolled cell proliferation is the fundamental characteristics for malignancy[^16^](#jcla23315-bib-0016){ref-type="ref"}, [^17^](#jcla23315-bib-0017){ref-type="ref"} and is involved in multiple alterations of genes and proteins correlated with proliferation and cell cycle.[^18^](#jcla23315-bib-0018){ref-type="ref"}, [^19^](#jcla23315-bib-0019){ref-type="ref"} Therefore, identifying the factors leading to promoting cell proliferation is essential to develop effective therapeutic strategy. Herein, we found that mRNA and protein levels of KCTD12 both significantly downregulate in breast cancer tissues and closely related with poor survival of patients with breast cancer. Moreover, downregulation of KCTD12 promotes breast cancer cell proliferation and in vitro tumorigenesis, which implicates that KCTD12 may be served as a suppressive protein in the progression of breast cancer. The uncontrolled cell proliferation usually accompanies the uncontrolled cell cycle progression. One of the crucial regulations of cell cycle is the existence of checkpoints. Rb is one of the notable regulatory proteins for checkpoint and is a potent inhibitor of G1/S transition. Its activation is dependent on the CDKs/cyclins complex.[^4^](#jcla23315-bib-0004){ref-type="ref"}, [^20^](#jcla23315-bib-0020){ref-type="ref"}, [^21^](#jcla23315-bib-0021){ref-type="ref"} In the present study, we found that cyclin D1 is significantly increased in KCTD12‐silenced breast cancer cell to activate the phosphorylation of Rb, which can promote G1/S transition. Moreover, the phosphorylation of Rb is relied on AKT/FOXO1 signaling. FOXO1 plays a vital role in suppressing cell proliferation via transcription regulation of several proteins such as p21, p27, and cyclin D1.[^22^](#jcla23315-bib-0022){ref-type="ref"} It was reported that deregulation of FOXO1 has been found in multiple cancers.[^23^](#jcla23315-bib-0023){ref-type="ref"}, [^24^](#jcla23315-bib-0024){ref-type="ref"}, [^25^](#jcla23315-bib-0025){ref-type="ref"}, [^26^](#jcla23315-bib-0026){ref-type="ref"} Herein, we found that downregulation of KCTD12 can promote the phosphorylation levels of FOXO1. Nevertheless, our result showed that KCTD12 can affect cancer proliferation in breast cancer, but the direct target of KCTD12 needs our future study. The study on the effect of KCTD12 on cancer cell proliferation is little. Zong et al showed that KCTD12 can bind with CDK1 to regulate CDK1 phosphorylation and further to influence Hela cell proliferation by immunoprecipitation and mass spectrometry analysis.[^13^](#jcla23315-bib-0013){ref-type="ref"} And whether the same fact exits in breast cancer needs more evidence. In summary, our research KCTD12 may be a suppressive protein of breast cancer. The therapeutic use of KCTD12 needs more evidences, for example, in vivo experiments, correlation analysis between KCTD12 levels and clinical characteristics. We will perform these tests in future. ETHICS STATEMENT {#jcla23315-sec-0020} ================ The tissues used in the study were gathered from The First Affiliated Hospital of Sun Yat‐sen University. Before collecting, informed consent from patients and approval from the Institutional Ethics Committee were both obtained. Medical research was conducted according to the World Medical Association Declaration of Helsinki.