A rare variant in the MARVELD2 gene is associated with Chinesen samples with ovarian endometriosis

Objectives: Endometriosis is a common gynecological disease affecting up to ~10% of women at reproductive age. Prior combined studies implied that MARVELD2 might be involved in the pathogenesis of certain malignancies. Here, 211 Han Chinese samples with ovarian endometriosis were analyzed for the presence of MARVELD2 mutations. Methods: We analyze the potential presence of MAR- VELD2 mutations by direct DNA sequencing. Results: A total of 7 variants, 5 missense and 2 synonymous variants, were identified in our 211 ovarian endometriosis samples with different frequencies. Among the 5 missense variant, a missense rare variant p.V198M (c.592G>A), was identified in 10 out of our 211 samples (4.74%). This rare variant was identified with extremely low frequency in 766 control samples from 766 Chinese women without endometriosis (0.13%, 1/766) and control samples in the public databases. The evolutionary conservation analysis results suggested that the MARVELD2 rare variant lead to highly conserved amino acid substitutions among 14 vertebrate species from Human to Snake. Furthermore, both the SIFT and Polyphen-2 programs predicted this rare variant to be ‘disease causing’. However, we failed to observe any statistical significance between the MARVELD2 rare variant and the available clinical data. Conclusions: We identified a potential pathogenic rare variant in the MARVELD2 gene in Chinese samples with ovarian endometriosis, indicating the MARVELD2 rare variant might play an active role in the pathogenesis of endometriosis.

MARVELD2; Rare variant; Ovarian endometriosis; Han Chinese 1. Introduction Endometriosis is a common gynecological disease, in spite of the subject of many scientific researches, the de- tailed molecular etiology remains still unclear [ 1]. Among which, the ‘Sampsons hypothesis’ is the most extensively admissive interpretation, which implies that endometriosis takes place owing to retrograde menstruation, where en- dometrial tissue passes through the fallopian tube into the peritoneal and pelvic cavity where it implants [ 2]. The im- plantation theory indicates the formation of endometriosis in the peritoneal cavity and ovary needs endometrial tissue or cells fulfilling a process of adhesion, invasion and pro- liferation [3]. Tricellulin, a major component of tight junctions, en- coded by the MARVELD2 gene [ 4], was the first identi- fied protein to be uniquely localized at the tri-epithelial junctions [5]. Prior studies have revealed that MARVELD2 mutations could cause nonsyndromic deafness, and these MARVELD2 mutations were shown to cause tricellulin de- fects and deletion, thereby affecting the adhesion and the tight junction function between epithelial cells [ 6–8]. Fur- thermore, dysregulated MARVELD2 expression was asso- ciated with patients’ prognosis in certain cancer types [ 9– 11] and could promote cell migration [12]. Endometriosis is similar to cancer and exhibits enhanced invasion and migra- tion [13–15]. This disorder is considered as a precancerous lesion and harbored a variety of mutations in certain onco- gene and tumor suppressor genes [ 16–19]. A large-scale sequencing effort has revealed that epithelial cells harbored cancer-associated mutations promoting the pathogenesis of endometriosis [20]. In addition, increased expression of tri- cellulin could promote the invasions and metastasis of cer- tain human tumor [ 21]. Prior studies have revealed that certain germline mu- tations, including single nucleotide polymorphism (SNP), played important roles in the pathogenesis of endometrio- sis via candidate gene or large-scale sequencing strategies [22,23]. In addition, somatic mutations in certain genes, were also shown to confer the risk of endometriosis [ 16– 18,24]. Prior studies have found that dysregulated expres- sion of MARVELD2 was associated with patients’ prognosis in certain cancer types [9–11] and could promote cell migra- Table 1. The potential association of MARVELD2 rare variant with clinical data in 211 Chinese samples with ovarian endometriosis. Features Total sample Wild type (n = 201) Mutation (n = 10) p value Age (years) 211 29.56 ± 7.33 31.5 ± 4.72 0.42 Age of menarche (years) 211 12.53 ± 1.49 13.2 ± 2.14 0.37 Hemoglobin (g/L) 211 122.35 ± 11.34 130.1 ± 4.23 0.16 TSH (mIU/mL) 211 2.12 ± 1.23 1.8 ± 1.18 0.12 FT3 (pg/mL) 211 3.15 ± 0.18 3.06 ± 0.26 0.09 FT4 (ng/dL) 211 1.31 ± 0.13 1.26 ± 0.19 0.08 AFP (ng/mL) 211 2.62 ± 1.36 3.11 ± 0.93 0.35 CEA (ng/mL) 211 1.03 ± 0.42 0.96 ± 0.26 0.22 CA125 (U/mL) 211 115.33 ± 166.23 103.25 ± 78.39 0.56 SCCA (ng/mL) 211 1.47 ± 0.56 1.52 ± 0.86 0.37 Whitebloodcellcount (×109) 211 6.26 ± 2.12 6.12 ± 0.58 0.39 Lymphocyte cell count (×10 9) 211 1.88 ± 0.63 1.46 ± 0.39 0.34 Eosinophil granulocyte (×10 9) 211 0.15 ± 0.07 0.08 ± 0.06 0.55 Mononuclear cell count (×10 9) 211 0.47 ± 0.07 0.42 ± 0.09 0.11 Neutrophil cell count (×10 9) 211 3.88 ± 1.62 3.96 ± 1.22 0.77 Platelet (×109) 211 201.38 ± 52.63 213.66 ± 45.38 0.45 Neutrophil cell proportion (%) 211 58.37 ± 7.66 65.38 ± 3.38 0.27 TSH, thyroid stimulating hormone; FT3, free triiodothyronine; FT4, free thyroxine; AFP , α-fetoprotein; CEA, carcinoembryonic antigen; CA125, cancer antigen 125; SCCA, squamous cell carcinoma antigen. Table 2. The primer information for MARVELD2 gene (NM_001038603) amplification. Gene Exon Annealing PCR amplicon Forward primers (5′-3′) Reverse primers (5′-3′) MARVELD2 2 56 ◦C 1260 bp atcagcatcattgagagga acatacacacacaaatgag MARVELD2 3 52 ◦C 335 bp atcaacctcttaaaattgag ggtcttgaaattctggtctc MARVELD2 4, 5 55 ◦C 706 bp ccacctgatcttcttcctc ctctagatttcagtgctcct MARVELD2 6 58 ◦C 339 bp tctcagtgtgctttgagata aatgctcatttctctaggtt MARVELD2 7 53 ◦C 286 bp tgtagagagcttaactgttac tggtttcaaataagcgtta tion [12]. Considering the important role of tight junction played in cell invasion and migration [24], and endometrio- sis was a premalignant disorder and harbored a number of mutations [25,26], we thus hypothesized that MARVELD2 mutations might also existed in endometriosis samples. Here, we recruited and analyzed a total of 211 Chi- nese patients with ovarian endometriosis for the presence of MARVELD2 mutations. A rare variant in the MARVELD2 gene was identified in 10 out of 211 samples (4.74%). 2. Materials and methods 2.1 Samples A total of 211 Chinese patients with ovarian en- dometriosis, as well as control samples from 766 Chi- nese women without endometriosis were also collected from Jiangxi Provincial Maternal and Child Health Hospi- tal (Nanchang, China). Written informed consent was ob- tained from each sample prior to this study, and the present study was performed according to the tenets of the Helsinki Declaration and was approved by the Institutional Review Board of Jiangxi Provincial Maternal and Child Health Hos- pital. 2.2 Clinical data The clinical data for the participating women with ovarian endometriosis was collected at the time of sam- pling, including age, age at menarche, the laboratorial data included serum hemoglobin, free triiodothyronine (FT3), free thyroxine (FT4), thyroid stimulating hormone (TSH), carcinoembryonic antigen (CEA), cancer antigen 125 (CA125), squamous cell carcinoma antigen (SCCA) and α-fetoprotein (AFP) were determined on day 3 of the menstrual cycle by radioimmunoassay, as described previ- ously [17,18]. In addition, the number of white blood cells, lymphocytes, eosinophil granulocytes, neutrophil granulo- cytes, mononuclear cells, platelet, and neutrophil granulo- cyte proportion was analyzed by an automated hematology analyzer XN-3000 (Sysmex Corporation, Kobe, Japan) (Ta- ble 1), as described previously [ 17,18]. 2.3 DNA extraction and mutation analysis The genomic DNA was extracted from the peripheral blood of our samples. Omega Blood DNA kit (OMEGA Bio-tek Inc., Doraville, GA) was used to isolate the ge- nomic DNA for our samples, as described previously [17]. The entire coding regions of the MARVELD2 gene 43 Table 3. The identified variants of the MARVELD2 gene (NM_001038603) in our patients. SNP Amino acid /Nucleotide change Frequency SIFT prediction Polyphen-2 prediction rs1185246 p.T33I/c.98C>T 43.60% (92/211) Tolerated Benign rs111458976 p.Y62Y/c.186C>T 55.45% (117/211) - - rs181575833 p.Pro102Pro/c.306G>A 0.48% (1/211) - - rs575942430 p.Y159C/c.476A>G 0.48% (1/211) Tolerated Benign rs201914751 p.V198M/c.592G>A 4.74% (10/211) Damaging Damaging rs771384203 p.G237A/c.710G>C 0.48% (1/211) Tolerated Benign rs1198930354 p.V489M/c.1465G>A 1.42% (3/211) Tolerated Benign Table 4. The allele frequency comparison of MARVELD2 rare variant among our patients and control samples in the present study and public databases. SNP ID Case (N = 211) Normal control (N = 766) p value a Allele frequency in EXAC p value a Allele frequency in 1000 genome p value a rs201914751 10/422 1/1532 2.36 × 10 −6 22/121412 2.2 × 10 −16 3/5008 2.31 × 10 −9 a p value, Fisher’s exact test. (NM_001038603) were amplified with certain PCR primer pairs (Table 2), and subjected to direct DNA sequencing. The PCR reactions were performed as follows, 30 ng DNA, 1.5 µL 10 × PCR buffer, 250 µM dNTPs mix, 0.20 µM each primer, 2.5 mM MgCl 2, 0.5 U Taq DNA polymerase (Takara, Dalian, China). The PCR reaction was performed as follows: 95 ◦C premature for 5 min, 35 cycles in- cluding 94 ◦C for 30 sec, 52–58 ◦C for 45 sec and 72 ◦C for 30 sec, ended with a 10 min extension at 72 ◦C. The obtained PCR products were sequenced on an ABI 3730XL DNASequencer (Applied Biosystems, Waltham, MA, USA). The potential MARVELD2 mutations were an- alyzed and aligned with standard DNA sequences of the Human MARVELD2 gene via DNAStar Lasergene software (Madison, WI, USA). 2.4 Evolutionary conservation analysis of the rare variant of MARVELD2 We downloaded the MARVELD2 proteins from 14 vertebrate species from GenBank database (www.ncbi.nlm.nih.gov/gene), including Human (NP_001033692), Chimpanzee (XP_003310745), Monkey (XP_001094419), Mouse (NP_001033691), Rat (NP_001102406), Cattle (XP_002696327), Dog (XP_019690741), Horse (XP_023474013), Pig (NP_001230877), Cat (XP_019690741), Tree shrew (XP_006169576), Chicken (XP_424965), Bat (XP_006762445) and Snake (XP_039176681). The con- servation of the mutated MARVELD2 residue was analyzed with MEGA4 software (Tokyo, Japan) [ 27]. 2.5 In silico analysis of the MARVELD2 rare variant We used SIFT [ 28] and Polyphen-2 [ 29] online pro- grams to predict the potential pathogenicity of these mis- sense variants of MARVELD2. These programs could auto- matically assess this rare variant to be damaging or benign. 2.6 Statistical analysis We used two-sided Student’s t-test and Mann- Whitney’s method to analyze the potential association of numerical and continuous variables between ovarian en- dometriosis samples with and without MARVELD2 muta- tions, respectively; Fisher’s exact test was used to analyze the allele frequency of MARVELD2 rare variant between the cases and controls; p value less than 0.05 was consid- ered statistically significant. All statistical analyses were performed by the software SPSS 19.0 (SPSS, Inc., Chicago, IL, USA). 3. Results 3.1 Sample characteristics The median age of the samples was 32 years (range, 20–52) and the median age at menarche was 14 years (range, 10–19). The detailed clinical data, including hemoglobin, FT3, FT4, TSH, CEA, CA125, SCCA, AFP , and blood cell counts, are summarized in Table 1. 3.2 MARVELD2 rare variant in ovarian endometriosis A total of 7 variants, 5 missense and 2 synony- mous variants, were identified in our 211 samples with ovarian endometriosis with different frequencies (Table 3). Among the 5 missense variant, a rare variant (rs201914751) p.V198M (c.592G>A), was identified with high frequency in our samples with ovarian endometriosis (10/211, 4.74%) (Fig. 1) (Table 3). This rare variant was found in 766 con- trol samples from 766 Chinese women without endometrio- sis with extremely low frequency (0.13%, 1/766) ( p < 0.01); furthermore, this rare variant existed in extremely low frequencies in control samples in the 1000 genome (www.ncbi.nlm.nih.gov/variation/tools/1000genomes) and EXAC (www.exac.broadinstitute.org) databases, which in- cluded 2504 and 60706 samples, respectively ( p < 0.01) (Table 4). The average age and age of menarche of the 10 sample with MARVELD2 rare variant (p.V198M) was 31.5 44 and 13.2 years old, respectively. Fig. 1. The representative sequencing electropherograms of MARVELD2 p.V198M (c.592G>A) mutation, the arrow refers to locations of the mutations. “OCB-176” was an endometriosis samples with MARVELD2 mutation, while “Control” was a con- trol sample without MARVELD2 mutation. 3.3 In silico and Evolutionary conservation analyses of the MARVELD2 missensevariants Both the SIFT and Polyphen-2 programs were used to predict the potential pathogenicities of these variants, the predicted results showed that the p.V198M variant was ‘dis- ease causing’; while other four variants were benign. The evolutionary conservation analysis result suggested that the MARVELD2 rare variant lead to highly conserved amino acid substitution from valine to methionine at the 198th codon (V198M), among the 14 vertebrate species from Hu- man to Snake (Fig. 2). 3.4 Association between MARVELD2 rare variant and clinical data The clinical data between the ovarian endometrio- sis samples with and without MARVELD2 rare variant (p.V198M) was analyzed, including patients’ age, age of menarche, FT3, FT4, TSH, hemoglobin, CA125, SCCA, AFP , CEA, white blood cell count, eosinophil granulocyte, lymphocyte cell count, neutrophil cell count, mononuclear cell count, platelet and neutrophil cell proportion. How- ever, we failed to get any significant association between MARVELD2 rare variant and these clinical parameters (Ta- ble 1). Fig. 2. The evolutionary conservation analysis result of MAR- VELD2 p.V198Mmutation from Human to Snake. 4. Discussion MARVELD2, formerly referred to as TRIC-a, encodes tricellulin, a tricellular tight junction protein [4]. Tricellulin is mainly localized in tricellular cell contacts, while in bi- cellular tight junctions to a lesser extent [ 30]. To date, it’s deemed that tricellulin expression ubiquitously exists in ep- ithelial junctions of tissues and organs throughout the body [6], as well as in ovarian epithelia [ 31]. Prior large-scale sequencing studies have revealed that common variants in certain genes played important role in the pathogenesis of endometriosis, including vascular endothelial growth factor receptor 2 (VEGFR2), mitogen- activated protein kinase kinase kinase 4 (MAP3K4), and Wnt family member 4 (WNT4) [ 32–34]. Recently, in- creasing evidences have suggested that rare variants also facilitate the initiation and development of endometriosis [35,36]. In the present study, we have screened a total of 211 Han Chinese samples with ovarian endometriosis for the presence of MARVELD2 mutations, via sequencing the whole coding region and the exon-intron boundaries of the MARVELD2 gene. Here, we identified a MARVELD2 mis- sense rare variant, p.V198M (c.592G >A), in 10 out of 211 Han Chinese samples with ovarian endometriosis (4.74%), the frequency of this rare variant is significant higher than that either in the local control women without endometrio- sis or in the control samples in the 1000 genome and EXAC databases (p < 0.01). The evolutionary conservation anal- 45 ysis result showed that the MARVELD2 rare variant caused highly conserved amino acid substitution among 14 ver- tebrate species. Furthermore, the bioinformatic programs prediction result showed this rare variant might be damag- ing. MARVELD2 mutations could cause nonsyndromic deafness [ 6–8] and the molecular mechanism involved in affecting the paracellular permeability, leading to a toxic- ity for cochlear hair cells [ 37]. Subsequent studies found that dysregulated expression of MARVELD2 could change the capacities of cell invasion and migration in diverse cancer types, including colorectal and gastric cancers, and the potential underlying molecular mechanism involved in actin and cytoskeletal reorganization, as well as epithelial- mesenchymal transitions (EMT) process [ 21,38]. As a pre- malignant condition, endometriosis usually exhibited with dysregulation of cell invasion and migration, we thus spec- ulated that the MARVELD2 rare variant identified in the present study might play active role in the pathogenesis of endometriosis. On the other hand, we failed to observe any positive association between the MARVELD2 rare variant and the available clinical data. Of note, our sample size of ovarian endometriosis was relatively small, we will continue to ob- tain additional samples in order to re-analyze the potential association. 5. Conclusions We identified a relatively high frequency of MAR- VELD2 rare variant in our samples with ovarian en- dometriosis, indicating this rare variant might play positive role in the pathogenesis of this disease. Author contributions QW—investigation and manuscript preparation; RL—sample collection; YZ—investigation; YL— investigation; JZ—data analysis; YD—sample collection; XZ—data analysis; GG—sample collection; OH— methodology, manuscript revision. Ethics approval and consent to participate All subjects gave their informed consent for inclu- sion before they participated in the study. The study was conducted in accordance with the Declaration of Helsinki, and the protocol was approved by the Ethics Committee of Jiangxi Provincial Maternal and Child Health Hospital (ap- proval number: JXSFYBJYY2020KJK015). Acknowledgment We thank the sample donors who participated in this study. Funding This project was supported by the National Natural Science Foundation of China [Grant No. 81771559 and 82160291]. Conflict of interest The authors declare no conflict of interest.

[1] Gaetje R, Holtrich U, Engels K, Kissler S, Rody A, Karn T,et al. Differential expression of claudins in human endometrium and endometriosis. Gynecological Endocrinology. 2008; 24: 442– 449. [2] Sampson JA. Metastatic or Embolic Endometriosis, due to the Menstrual Dissemination of Endometrial Tissue into the V enous Circulation. American Journal of Pathology. 1927; 3: 93–110. [3] Y oung VJ, Brown JK, Saunders PTK, Horne AW. The role of the peritoneum in the pathogenesis of endometriosis. Human Repro- duction Update. 2013; 19: 558–569. [4] Nayak G, Lee SI, Y ousaf R, Edelmann SE, Trincot C, V an Itallie CM, et al. Tricellulin deficiency affects tight junction architec- ture and cochlear hair cells. Journal of Clinical Investigation. 2013; 123: 4036–4049. [5] Ikenouchi J, Furuse M, Furuse K, Sasaki H, Tsukita S, Tsukita S. Tricellulin constitutes a novel barrier at tricellular contacts of epithelial cells. Journal of Cell Biology. 2005; 171: 939–945. [6] Riazuddin S, Ahmed ZM, Fanning AS, Lagziel A, Kitajiri S, Ramzan K, et al . Tricellulin is a Tight-Junction Protein Nec- essary for Hearing. The American Journal of Human Genetics. 2006; 79: 1040–1051. [7] Chishti MS, Bhatti A, Tamim S, Lee K, McDonald M, Leal SM, et al. Splice-site mutations in the TRIC gene underlie autosomal recessive nonsyndromic hearing impairment in Pakistani fami- lies. Journal of Human Genetics. 2008; 53: 101–105. [8] Zheng J, Meng W, Zhang C, Liu H, Y ao J, Wang H, et al. New SNP variants of MARVELD2 (DFNB49) associated with non- syndromic hearing loss in Chinese population. Journal of Zhe- jiang University: Science B. 2019; 20: 164–169. [9] Somorácz A, Korompay A, Törzsök P , Patonai A, Erdélyi-Belle B, Lotz G, et al. Tricellulin Expression and its Prognostic Signif- icance in Primary Liver Carcinomas. Pathology Oncology Re- search. 2014; 20: 755–764. [10] Kojima T, Sawada N. Regulation of tight junctions in human normal pancreatic duct epithelial cells and cancer cells. Annals of the New Y ork Academy of Sciences. 2012; 1257: 85–92. [11] Korompay A, Borka K, Lotz G, Somorácz A, Törzsök P , Erdélyi- Belle B, et al . Tricellulin expression in normal and neoplastic human pancreas. Histopathology. 2012; 60: E76–E86. [12] Kyuno T, Kyuno D, Kohno T, Konno T, Kikuchi S, Arimoto C, et al . Tricellular tight junction protein LSR/angulin-1 con- tributes to the epithelial barrier and malignancy in human pan- creatic cancer cell line. Histochemistry and Cell Biology. 2020; 153: 5–16. [13] Xu Z, Zhang L, Y u Q, Zhang Y , Y an L, Chen Z. The estrogen- regulated lncRNA H19/miR-216a-5p axis alters stromal cell in- vasion and migration via ACTA2 in endometriosis. Molecular Human Reproduction. 2019; 25: 550–561. [14] Choi YS, Park JH, Y oon JK, Y oon JS, Kim JS, Lee JH,et al. Po- tential roles of aquaporin 9 in the pathogenesis of endometriosis. Molecular Human Reproduction. 2019; 25: 373–384. [15] Liu J, Wang Y , Chen P , Ma Y , Wang S, Tian Y ,et al. AC002454.1 and CDK6 synergistically promote endometrial cell migration and invasion in endometriosis. Reproduction. 2019; 157: 535– 46 543. [16] Lac V , V erhoef L, Aguirre-Hernandez R, Nazeran TM, Tessier- Cloutier B, Praetorius T, et al. Iatrogenic endometriosis harbors somatic cancer-driver mutations. Human Reproduction. 2019; 34: 69–78. [17] Zou Y , Zhou J, Guo J, Wang L, Luo Y , Zhang Z,et al. The pres- ence of KRAS, PPP2R1a and ARID1a mutations in 101 Chinese samples with ovarian endometriosis. Mutation Research. 2018; 809: 1–5. [18] Zou Y , Zhou JY , Wang F, Zhang ZY , Liu FY , Luo Y , et al . Analysis of CARD10 and CARD11 somatic mutations in pa- tients with ovarian endometriosis. Oncology Letters. 2018; 16: 491–496. [19] Cao B, Zeng Y , Wu F, Liu J, Shuang Z, Xu X, et al. Novel TR- ERF1 mutations in Chinese patients with ovarian endometriosis. Molecular Medicine Reports. 2018; 17: 5435–5439. [20] Suda K, Nakaoka H, Y oshihara K, Ishiguro T, Tamura R, Mori Y ,et al. Clonal expansion and diversifification of cancer- associated mutations in endometriosis and normal endometrium. Cell Reports. 2018, 24: 1777–1789. [21] Zhang JX, Qin MB, Y e Z, Peng P , Li SM, Song Q,et al. Associ- ation of tricellulin expression with poor colorectal cancer prog- nosis and metastasis. Oncology Reports. 2020; 44: 2174–2184. [22] Angioni S, D’Alterio MN, Coiana A, Anni F, Gessa S, Deiana D. Genetic Characterization of Endometriosis Patients: Review of the Literature and a Prospective Cohort Study on a Mediter- ranean Population. International Journal of Molecular Sciences. 2020; 21: 1765. [23] Deiana D, Gessa S, Anardu M, Daniilidis A, Nappi L, D’Alterio MN, et al. Genetics of endometriosis: a comprehensive review. Gynecological Endocrinology. 2019; 35: 553–558. [24] Anglesio MS, Papadopoulos N, Ayhan A, Nazeran TM, Noë M, Horlings HM, et al . Cancer-Associated Mutations in En- dometriosis without Cancer. New England Journal of Medicine. 2017; 376: 1835–1848. [25] Li X, Zhang Y , Zhao L, Wang L, Wu Z, Mei Q, et al . Whole-exome sequencing of endometriosis identifies frequent alterations in genes involved in cell adhesion and chromatin- remodeling complexes. Human Molecular Genetics. 2014; 23: 6008–6021. [26] Wang K, Li T, Xu C, Ding Y , Li W, Ding L. Claudin-7 downreg- ulation induces metastasis and invasion in colorectal cancer via the promotion of epithelial-mesenchymal transition. Biochem- ical and Biophysical Research Communications. 2019; 508: 797–804. [27] Tamura K, Dudley J, Nei M, Kumar S. MEGA4: Molecular Evolutionary Genetics Analysis (MEGA) software version 4.0. Molecular Biology and Evolution. 2007; 24: 1596–1599. [28] Choi Y , Chan AP . PROVEAN web server: a tool to predict the functional effect of amino acid substitutions and indels. Bioin- formatics. 2015; 31: 2745–2747. [29] Adzhubei IA, Schmidt S, Peshkin L, Ramensky VE, Gerasimova A, Bork P , et al. A method and server for predicting damaging missense mutations. Nature Methods. 2010; 7: 248–249. [30] Krug SM, Amasheh S, Richter JF, Milatz S, Günzel D, West- phal JK, et al. Tricellulin Forms a Barrier to Macromolecules in Tricellular Tight Junctions without Affecting Ion Permeability. Molecular Biology of the Cell. 2009; 20: 3713–3724. [31] Raleigh DR, Marchiando AM, Zhang Y , Shen L, Sasaki H, Wang Y ,et al. Tight junction-associated MARVEL proteins marveld3, tricellulin, and occludin have distinct but overlapping functions. Molecular Biology of the Cell. 2010; 21: 1200–1213. [32] Steinthorsdottir V , Thorleifsson G, Aradottir K, Feenstra B, Sig- urdsson A, Stefansdottir L, et al . Common variants upstream of KDR encoding VEGFR2 and in TTC39B associate with en- dometriosis. Nature Communications. 2016; 7: 12350. [33] Uimari O, Rahmioglu N, Nyholt DR, Vincent K, Missmer SA, Becker C, et al . Genome-wide genetic analyses high- light mitogen-activated protein kinase (MAPK) signaling in the pathogenesis of endometriosis. Human Reproduction. 2017; 32: 780–793. [34] Nyholt DR, Low S, Anderson CA, Painter JN, Uno S, Morris AP ,et al. Genome-wide association meta-analysis identifies new endometriosis risk loci. Nature Genetics. 2012; 44: 1355–1359. [35] Rahmioglu N, Montgomery GW, Zondervan KT. Genetics of Endometriosis. Women’s Health. 2015; 11: 577–586. [36] Sapkota Y , Vivo ID, Steinthorsdottir V , Fassbender A, Bowdler L, Buring JE, et al. Analysis of potential protein-modifying vari- ants in 9000 endometriosis patients and 150000 controls of Eu- ropean ancestry. Scientific Reports. 2017; 7: 11380. [37] Nayak G, Lee SI, Y ousaf R, Edelmann SE, Trincot C, V an Itallie CM, et al. Tricellulin deficiency affects tight junction architec- ture and cochlear hair cells. Journal of Clinical Investigation. 2013; 123: 4036–4049. [38] Masuda R, Semba S, Mizuuchi E, Y anagihara K, Y okozaki H. Negative regulation of the tight junction protein tricellulin by snail-induced epithelial-mesenchymal transition in gastric car- cinoma cells. Pathobiology. 2010; 77: 106–113. 47