{"paper_id":"0e1a903a-8585-4b19-a672-13c42bca83f6","body_text":"1\nSCiEntiFiC  RepoRts  | 7:44616 | DOI: 10.1038/srep44616\nwww.nature.com/scientificreports\nEffects of matrix stiffness on \nepithelial to mesenchymal \ntransition-like processes of \nendometrial epithelial cells: \nImplications for the pathogenesis \nof endometriosis\nSachiko Matsuzaki1,2,3, Claude Darcha4, Jean-Luc Pouly1 & Michel Canis1,2,3\nEndometriosis is defined as the presence of endometrial glands and stroma within extrauterine \nsites. Our previous study revealed an epithelial to mesenchymal transition (EMT)-like process in red \nperitoneal endometriosis, whereas membrane localization of E-cadherin was well maintained in \nepithelial cells of deep infiltrating endometriosis (DIE). Here we show that endometrial epithelial \ncells (EEE) grown on polyacrylamide gel substrates (PGS) of 2 kilopascal (kPa), a soft matrix, initiate a \npartial EMT-like process with transforming growth factor- β1 (TGF-β1) stimulation. Increasing matrix \nstiffness with TGF-β1 stimulation reduced the number of cell-cell contacts. Cells that retained cell-\ncell contacts showed decreased expression of E-cadherin and zonula occludens 1 (ZO-1) to cell-cell \njunctions. Few deep endometriotic epithelial cells (DEE) grown on 30-kPa PGS, which may mimic  \nin vivo tissue compliance of DIE, retained localization of E-cadherin to cell-cell junctions with TGF-β 1 \ntreatment. Immunohistochemical analysis showed no phosphorylated Smad 2/3 nuclear localization \nin E-cadherin+  epithelial cells of DIE. We hypothesize that EEE may undergo an EMT-like process \nafter attachment of endometrium to peritoneum in a TGF- β1–rich microenvironment. However, \nTGF-β1 signaling may be absent in DIE, resulting in a more epithelial cell-like phenotype in a rigid \nmicroenvironment.\nEndometriosis, a common gynecological disorder responsible for infertility and pelvic pain, is defined as the \npresence of endometrial glands and stroma within extrauterine sites\n1. This condition affects approximately 10% of \nwomen of reproductive age1. Despite extensive studies, the etiology, pathogenesis, and pathophysiology of endo-\nmetriosis are not fully understood 1. However, one of the most supported theories may be implantation theory: \nendometriosis originates from retrograde menstruation of endometrial tissue, which may then implant into the \nperitoneal cavity\n1.\nOur previous study revealed epithelial to mesenchymal transition (EMT)- and mesenchymal to epithelial \ntransition (MET)-like processes in epithelial cells of pelvic endometriosis2. We previously hypothesized that the \norigin of endometriotic epithelial cells might be endometrial epithelial cells2. Endometrial epithelial cells might \nbe adapted to specific microenvironments after implantation, resulting in different types of pelvic endometrio-\nsis, including superficial peritoneal endometriosis, ovarian endometriosis, and deep infiltrating endometriosis \n(DIE)\n2. Endometrial epithelial cells might undergo an EMT-like process after attachment of endometrium to the \nperitoneum or ovary, resulting in red peritoneal endometriosis or ovarian endometriosis, respectively 2. It has \nbeen postulated that red and black peritoneal lesions may represent different stages of the spontaneous evolution \n1CHU Clermont-Ferrand, CHU Estaing, Chirurgie Gynécologique, Clermont-Ferrand, France. 2Clermont Université, \nUniversité d’Auvergne, ISIT UMR6284, Clermont-Ferrand, France. 3CNRS, ISIT UMR6284, Clermont-Ferrand, \nFrance. 4CHU Clermont-Ferrand, Service d’Anatomie et Cytologie Pathologiques, Clermont-Ferrand, France. \nCorrespondence and requests for materials should be addressed to S.M. (email: sachikoma@aol.com)\nreceived: 08 November 2016\nAccepted: 10 February 2017\nPublished: 17 March 2017\nOPEN\n\nwww.nature.com/scientificreports/\n2\nSCiEntiFiC  RepoRts  | 7:44616 | DOI: 10.1038/srep44616\nof endometriotic implants, with the first stage being red lesions3,4. MET-like processes may then occur during the \nevolution of peritoneal endometriotic implants, resulting in black peritoneal endometriosis2. E-cadherin expres-\nsion is significantly higher in epithelial cells of DIE compared to those of menstrual endometrium, suggesting \nthat a MET-like process might occur in DIE2. Black peritoneal lesions are generally much smaller than DIE tissue. \nWe showed that expression levels of dephosphorylated beta-catenin were significantly higher in DIE compared \nto black peritoneal lesions\n2. The dephosphorylated form of beta-catenin is present at the plasma membrane upon \nWnt stimulation5. Our previous in vitro study showed that the Wnt/beta-catenin pathway is involved in cell prolif-\neration, migration, and/or invasion of endometriotic epithelial cells6. A more epithelial cell-like phenotype along \nwith Wnt/beta-catenin pathway activation might facilitate growth and infiltration in DIE2, which is characterized \nhistologically by dense fibrous tissue 3,7. One of the hallmarks of fibrosis is tissue stiffening. The microenviron-\nment of DIE is therefore more stiff than that of endometrium. Recent studies demonstrated that increased matrix \nstiffness could induce EMT\n8,9. A study showed that increasing matrix stiffness directly activated EMT through \nthe EMT-inducing transcription factor TWIST1 in human MCF10A and tumorigenic mouse Eph4Ras mammary \nepithelial cells\n8. EMT can be induced or regulated by various growth and differentiation factors10,11. Among them, \ntransforming growth factor-β 1 (TGF-β 1) may be a major inducer of physiological as well as pathological EMT \nduring embryogenesis, cancer progression, and fibrosis 10,11. TGF-β 1 is also involved in the pathophysiology of \nendometriosis12.\nThese findings may not support our previous hypothesis 2. However, to date, no study has investigated the \neffects of extracellular matrix (ECM) matrix stiffness on EMT-like processes in endometrial epithelial cells. The \naim of the present in vitro study was to investigate the effects of ECM stiffness on EMT-like morphological and \nphenotypic changes of endometrial epithelial cells. Herein we used polyacrylamide gel substrates (PGS) of differ-\nent stiffness (2-, 4-, 8-, 16-, and/or 30-kilopascal [kPa]) to evaluate the effects of substrate rigidity on expression \nof E-cadherin, zonula occludens 1 (ZO-1), N-cadherin, and F-actin in endometrial epithelial cells. We elected \nto use PGS of different degrees of stiffness based on the results of our previous study\n13. Jiang et al. showed that \nthe elasticity of the endometrium was 3.34 ±  0.42 kPa during the proliferative phase and 1.97 ±  0.34 kPa during \nthe secretory phase using three-dimensional multifrequency magnetic resonance elastography 14. Currently no \ndata are available regarding the tissue stiffness of DIE in vivo . However, several studies that evaluated fibrotic \nlungs, fibrotic intestines, and fibrotic livers have shown that the elastic modulus value ranged from approximately \n1–3 kPa for normal tissue to approximately 17–22 kPa for fibrotic tissue\n15–19. The soft substrates (2- or 4-kPa PGS) \nand the rigid substrates (16- or 30-kPa PGS) may mimic in vivo tissue compliance of the endometrium or DIE, \nrespectively13.\nDuring EMT, epithelial cell-cell junctions, which are essential for epithelial integrity, are deconstructed, and \nthe junction proteins are relocalized and/or degraded11,20,21. In cuboidal-shaped and epithelial cells organized in \ncompact islets, F-actin is organized in cortical bundles tightly associated with cell-cell adhesions11,20,21. Cell mor-\nphology is changed to a spindle-shaped morphology and F-actin is assembled into contractile actin stress fibers \nacross the ventral surface of the cells\n11,20,21. The dissolution of tight junctions during EMT is accompanied by the \ndiffusion of ZO-1 from cell-cell contacts11,20,21. E-cadherin, a key component of adherens junctions, is cleaved at \nthe plasma membrane and subsequently degraded11,20,21. Cadherin switching—the loss of E-cadherin and the gain \nof N-cadherin expression—is a major hallmark of EMT 11,20,21.\nResults\nEffects of matrix stiffness with or without TGF-β1 treatment on E-cadherin, N-cadherin, ZO-1, \nand F-actin expression in endometrial epithelial cells of patients with endometriosis (EEE).   \nWe first evaluated E cadherin expression in EEE (n =  5) and endometrial epithelial cells of patients without endo-\nmetriosis (NEEE) (n =  5) grown on 2- or 30-kPa PGS, or plastic with or without TGF-β 1 treatment. We observed \nno difference in E-cadherin expression between EEE and NEEE, when compared to cells grown on a substrate of \nthe same stiffness (2- or 30-kPa PGS, or plastic). Thus, we further analyzed of E-cadherin, N-cadherin, ZO-1, and \nF-actin expression only in EEE.\nWithout TGF-β1 treatment. EEE grown on 2-kPa PGS retained a rounded morphology, cortical actin, and \nE-cadherin and ZO-1 localization to cell-cell junctions (Figs 1A and 2A). In EEE grown on 2-, 4-, 8-, 16-, or \n30-kPa PGS, cells retained cell-cell contacts. E-cadherin was localized to cell-cell junctions (Fig. 1A), but no \nN-cadherin localization to cell-cell junctions was observed (Fig. 2C). When EEE were grown on 30-kPa PGS, \ncells became elongated and F-actin+  stress fiber-like structures were observed (Fig. 2E). When EEE were grown \non plastic, only cells located in the center retained retained localization of E-cadherin and ZO-1 to cell-cell junc-\ntions (Figs 1A and 2A). F-actin+  stress fiber-like structures were also observed (Fig. 2E). Cells were more loosely \narranged than those grown on 30-kPa PGS (Figs 1A and 2A,C,E).\nWith TGF-β1 treatment. EEE grown on 2-kPa PGS largely retained an epithelial cobblestone morphology and \ncell-cell contact (Figs 1B and 2B). However, E-cadherin (Fig. 1B) and ZO-1 (Fig. 2B) localization to cell-cell junc-\ntions was decreased (Figs 1B and 2B), and localization of N-cadherin to cell-cell junctions, albeit in few cells, was \nobserved (Fig. 2D). Peripheral actin filaments were also found around the cell body and protrusions (Fig. 2F). \nHowever, stress fibers were not found in the protrusions (Fig. 2F). Increasing matrix stiffness reduced the num-\nber of cell-cell contacts, and became elongated (Fig. 1B). Cells retained cell-cell contacts that were more loosely \narranged, and decreased E-cadherin (Fig. 1B) and ZO-1 (Fig. 2B) and increased N-cadherin (Fig. 2D) localization \nto cell-cell junctions were observed. F-actin+  stress fiber-like structures were observed in EEE grown on 16- and \n30-kPa PGS (Fig. 2F). When EEE were grown on plastic, morphological changes to fibroblast-like cells were \nobserved in CK+   cells (Figs 1B and 2B,D,F). Few cells retained localization of E-cadherin and ZO-1 to cell-cell \njunctions (Figs 1B and 2B). The majority of cells were CK+  fibroblast-like single cells (Fig. 1B).\n\nwww.nature.com/scientificreports/\n3\nSCiEntiFiC  RepoRts  | 7:44616 | DOI: 10.1038/srep44616\nEffects of matrix stiffness with or without TGF-β1 treatment on E-cadherin expression in endo-\nmetriotic epithelial cells derived from DIE (DEE). Without TGF-β1 treatment. DEE grown on 2-kPa \nPGS retained a rounded morphology and E-cadherin localization to cell-cell junctions (Fig. 3A). When DEE \nwere grown on 30-kPa PGS, cells became elongated, but retained E-cadherin localization to cell-cell junctions \n(Fig. 3A). DEE grown on plastic showed decreased E-cadherin localization to cell-cell junctions (Fig. 3A).\nWith TGF-β1 treatment. DEE grown on 2-kPa PGS largely retained an epithelial cobblestone morphology, but \nfew cells retained localization of E-cadherin to cell-cell junctions (Fig. 3B). DEE grown on 30-kPa PGS had fewer \ncell-cell contacts and became elongated (Fig. 3B). Cells that retained cell-cell contacts were more loosely arranged \nas observed in EEE grown on 30-kPa PGS (Fig. 3B). Few cells retained localization of E-cadherin to cell-cell \njunctions (Fig. 3B).\nWhen DEE was grown on plastic with TGF-β 1 treatment, morphological changes to fibroblast-like cells were \nobserved with TGF-β 1 treatment (Fig. 3B). These fibroblast-like cells were CK+ , but dissociated into single cells \n(Fig. 3B). Few cells retained localization of E-cadherin to cell-cell junctions (Fig. 3B).\nE-cadherin and phosphorylated Smad 2/3 (p-Smad 2/3) expression in DIE and red peritoneal \nendometriotic lesions.  We selected five DIE samples with very high E-cadherin expression and five red \nperitoneal lesions with very low E-cadherin expression from samples analyzed in our previous study 2. In these \nDIE tissues, no p-Smad 2/3 nuclear expression was observed (Fig. 3C,D). In contrast, nuclear p-Smad 2/3 expres-\nsion was observed in both epithelial and stromal cells in red peritoneal lesions (Fig. 3C,D).\nEffects of matrix stiffness with or without TGF-β1 treatment on cell proliferation and Annexin \nV expression in EEE. No significant difference in the cellular proliferation index (CPI) was observed among \ncells grown on 2-kPa or 30-kPa PGS, or plastic, in cells treated with or without TGF-β 1 (Fig. 4A,B). The CPI of \nEEE treated with TGF-β 1 treatment was significantly lower than that without TGF-β 1 treatment, when compared \nto cells grown on a substrate of the same stiffness (2- or 30-kPa, or plastic) (Fig. 4A,B). When cells were grown on \nplastic with TGF-β 1 treatment, the majority of Ki-67+  cells (> 90%) were CK- single cells (Fig. 4B).\nNo Annexin V+  cells were observed in cells grown on various substrates of stiffness (2- or 30-kPa, or plastic) \nwith or without TGF-β 1 treatment.\nFigure 1. Representative photomicrographs of double-labeled immunofluorescence staining for CK and \nE-cadherin in EEE. (A) and (B): CK and E-cadherin in EEE grown on 2-, 4-, 8-, or 30-kPa polyacrylamide gel \nsubstrates (PGS), or on plastic without (A) or with (B) TGF-β 1 (5 ng/mL) stimulation. (A): In EEE grown on 2-, \n4-, 8-, or 30-kPa PGS, E-cadherin was localized to cell-cell junctions. In EEE grown on plastic, only cells located \nin the center retained localization of E-cadherin to cell-cell junctions (arrowheads). (B): Increasing matrix \nstiffness decreased E-cadherin localization to cell-cell junctions (arrowheads) Scale bar: 50 μ m.\n\nwww.nature.com/scientificreports/\n4\nSCiEntiFiC  RepoRts  | 7:44616 | DOI: 10.1038/srep44616\nFigure 2. Representative photomicrographs of double-labeled immunofluorescence staining for CK and \nZO-1, CK and N-cadherin and for CK and F-actin in EEE. (A,B) CK and ZO-1 in EEE grown on 2- or 30-kPa \nPGS, or on plastic without (A) or with (B) TGF-β 1 (5 ng/mL) stimulation. (C,D) CK and N-cadherin in EEE \ngrown on 2- or 30-kPa PGS, or on plastic without (C) or with (D) TGF-β 1 (5 ng/mL) stimulation.  \n(E,F) CK and F-actin in EEE grown on 2- or 30-kPa PGS or on plastic without (E) or with (F) TGF-β 1 (5 ng/mL) \nstimulation. Scale bar: 50 μ m.\n\nwww.nature.com/scientificreports/\n5\nSCiEntiFiC  RepoRts  | 7:44616 | DOI: 10.1038/srep44616\nEffects of increasing matrix stiffness and duration of cell culture on α-smooth muscle actin \n(αSMA)+ stress fibers and collagen type I protein expression in endometrial stromal cells of \npatients with endometriosis (EES). After 7- or 14-day cell culture, collagen type I protein expression was \nobserved in all of the cells grown on 2-kPa or 30-kPa PGS with or without TGF-β 1 treatment. The percentage of \ncells with α SMA+  stress fibers with or without TGF-β 1 treatment was significantly increased in a time-dependent \nmanner in both cells grown on 2-kPa and 30-kPa PGS (See Supplementary Fig. S1, Fig. 5B,C). TGF-β 1 treatment \nsignificantly increased the percentage of cells with α SMA+  stress fibers in cells grown on 30-kPa PGS compared \nto those grown without TGF-β 1 treatment (Fig. 5A,B,C). The percentage of cells with α SMA+  stress fibers was \nsignificantly higher in cells grown on 30-kPa PGS than those grown on 2-kPa PGS with or without TGF-β 1 treat-\nment after 7 days (Fig. 5A,B). However, after 14 days, no significant difference was observed in the percentage of \ncells with α SMA+  stress fibers grown on 2- and 30-kPa PGS without TGF-β 1 treatment (Fig. 5A,C). With TGF-β \n1 treatment, the percentage of cells with α SMA+  stress fibers was significantly higher in cells grown on 30-kPa \nthan those grown on 2-kPa PGS (Fig. 5A,C).\nDiscussion\nThe present study showed that endometrial and endometriotic epithelial cells can sense changes in ECM stiffness \nand respond to them, resulting in morphological and phenotypic changes in vitro. We observed that endometrial \nepithelial cells as well as endometriotic epithelial cells derived from DIE, grown on plastic with or without TGF-β \n1 treatment, underwent a partial EMT-like process without full acquisition of mesenchymal characteristics. In \nthe present study, we observed that the majority of cells grown on plastic with TGF-β 1 treatment were single CK- \ncells. To date, no marker exists to distinguish CK- cells that are differentiated from CK+  epithelial cells through \nEMT from CK- fibroblasts. Thus, we cannot completely exclude the possibility that these CK- cells were derived \nfrom CK+  epithelial cells through complete EMT. However, endometrial stromal cells proliferated more on plas-\ntic than that on soft substrates13. In contrast, the present study showed no significant effect of matrix stiffness on \ncell proliferation of EEE. In the present study, we observed that the majority of Ki-67+  cells (> 90%) was single \nFigure 3. Representative photomicrographs of double immunofluorescence staining for CK and \nE-cadherin in endometriotic epithelial cells (DEE) derived from deep infiltrating endometriosis (DIE) \nand for E-cadherin and phosphorylated Smad 2/3 (p-Smad 2/3) in DIE and red peritoneal endometriotic \nlesions. (A,B) Double-labeled immunofluorescence staining for CK and E-cadherin in DEE grown on 2- or \n30-kPa PGS or on plastic without (A) or with (B) TGF-β 1 (5 ng/mL) stimulation. (A) In DEE grown on 2- or \n30-kPa PGS, E-cadherin was localized to cell-cell junctions. In DEE grown on plastic, only cells located in the \ncenter retained localization of E-cadherin to cell-cell junctions (arrows). (B) Few cells retained localization \nof E-cadherin to cell-cell junctions (arrows). (C) Representative photomicrographs of double-labeled \nimmunofluorescence staining for E-cadherin and p-Smad 2/3 in DIE (a) and red peritoneal endometriotic \nlesions (b). Scale bar: 50 μ m. (D) p-Smad2/3 expression (percentage of p-Smad2/3+  nuclei among the total \nnumber of 4, 6-diamidino-2-phenylindole [DAPI]-stained nuclei) in DIE and red peritoneal endometriotic \nlesions (R-PE). E: epithelial cells. S: adjacent stromal cells to epithelial cells.\n\nwww.nature.com/scientificreports/\n6\nSCiEntiFiC  RepoRts  | 7:44616 | DOI: 10.1038/srep44616\nFigure 4. Effects of increasing matrix stiffness with or without transforming growth factor (TGF-β1) \n(5 ng/mL) stimulation on Ki-67 expression in endometrial epithelial cells of patients with endometriosis \n(EEE). (A) Cellular proliferation index (CPI) (percentage of Ki-67+  cells among the total number of \ncytokeratin [CK]+  cells) of proliferative EEE (n =  8) on PGS of varying stiffness (2- or 30-kPa) or on plastic. \nNumerical values are presented as box and whisker plots showing medians and the smallest and largest data \npoint ≤ 1.5 ×  interquartile range (IQR) from the 25th and 75th percentiles, respectively. (B) Representative \nphotomicrographs of Ki-67 and CK expression in EEE grown on 2- or 30-kPa substrates or plastic with or \nwithout TGF-β 1 (5 ng/mL) stimulation. Arrowheads indicate of Ki-67+  /CK- cells. Scale bar: 50 μ m.\nFigure 5. Effects of increasing matrix stiffness and the duration of cell culture on αSMA+ stress fibers and \ncollagen type I protein expression in endometrial stromal cells of patients with endometriosis (EES). \n (A) The percentage of cells with α SMA+  stress fibers with or without TGF-β 1 (5 ng/mL) stimulation in \nEES grown on 2- or 30-kPa PGS for 7 days or 14 days with or without TGF-β 1 (5 ng/mL) stimulation. (B,C) \nRepresentative photomicrographs of α SMA+  stress fibers and collagen type I protein expression in EES grown \non 2- or 30-kPa substrates for 7 days (B) or 14 days (C) with or without TGF-β 1 (5 ng/mL) stimulation. Scale \nbar: 50 μ m.\n\nwww.nature.com/scientificreports/\n7\nSCiEntiFiC  RepoRts  | 7:44616 | DOI: 10.1038/srep44616\nCK- cells when they were grown on plastic with TGF-β 1 treatment. Thus, we speculated that CK- cells may not \nbe derived from CK+   endometrial epithelial cells through EMT, but rather may be endometrial stromal cells \nthat were contaminated during isolation of endometrial epithelial cells. Findings from cells grown on plastic \nwith a stiffness in the gigapascal (gPa) range, which is much stiffer than that occurring in vivo (kPa range), may \nnot reflect the behavior of their in vivo counterparts. The present findings clearly showed that caution should be \ntaken when interpreting an EMT-like process in primary endometrial or endometriotic epithelial cells grown on \nplastic or glass.\nThe present study showed that endometrial epithelial cells might begin to undergo a partial EMT-like pro-\ncess even on 2-kPa PGS, a soft matrix, when cells are stimulated with TGF-β  1. A previous study showed that \ndecreasing matrix stiffness increased TGF-β 1–induced apoptosis in normal murine mammary gland epithelial \ncells (NMuMG) and Madin-Darby canine kidney epithelial cells (MDCK)\n9. However, in the present study, we \nobserved no Annexin V+  EEEs grown on a soft matrix with TGF-β 1 treatment and showed that cell proliferation \nwas similar to that of cells grown on a rigid matrix.\nA recent study showed that the peritoneal mesothelium may be responsible for the increased TGF-β 1 levels \nin women with endometriosis22. The study investigators speculated that the development of peritoneal endome-\ntriosis and the increase in TGF-β 1 are likely to go hand-in-hand, because retrograde menstruation and the pres-\nence of endometrial cells within the peritoneal cavity can induce inflammation22. If the origins of endometriotic \nepithelial cells are endometrial epithelial cells, endometrial epithelial cells undergo an EMT-like process after \nimplantation into the peritoneum, resulting in red peritoneal lesions in a TGF-β 1 rich microenvironment. A lim-\nitation of the present study, however, is that endometrial epithelial cells derived from the menstrual phase were \ncollected from only a limited number of patients. Most endometrial epithelial cell samples were derived from \nthe proliferative phase. If the origins of endometriotic epithelial cells are endometrial epithelial cells as stated by \nimplantation theory, endometrial epithelial cells derived from the menstrual phase would be more appropriate \nfor investigation.\nThe present study showed that few endometriotic epithelial cells derived from DIE retained localization of \nE-cadherin to cell-cell junctions, when cells were grown on 30-kPa PGS, a rigid substrate, with TGF-β 1 treatment. \nIn view of the present results and our previous findings\n2, we speculated that TGF-β 1 signaling may be absent in \nDIE that maintains E-cadherin expression in vivo2. Two studies have used intravital imaging to show that TGF-β 1 \nsignaling is transiently and locally activated in disseminating single cells, whereas cancer cells migrate collectively \nin the absence of TGF-β 1 signaling in vivo\n23,24. Collectively migrating cells overexpressed epithelial biomarkers, \nand cadherin-mediated cell-cell adhesions were crucial for cell-cell coordination during collective migration25,26. \nIncreasing substrate stiffness increased collective cell migration speed 27. Membrane localization of E-cadherin \nas well as ZO-1 was maintained in clusters of TbRII KO tumors 24, whereas neither was maintained in TbRIIfl/\nfl tumors at the tumor-stromal interface24. A previous experimental study in a baboon model of endometriosis \nsuggested that collective migration may be involved in pathophysiological processes of DIE28. In addition, cancer \ncells with high levels of TGF-β 1 signaling failed to promote lung metastasis, caused by failure of cells to prolifer-\nate in the lungs23. The downregulation of TGF-β 1 signaling at metastatic sites then permits growth of metastatic \ntumors23. The loss of TGF-β 1 signaling was significantly correlated with increased tumor size and enhanced \ncarcinoma cell survival23. DIE tissue is generally much bigger in size than red peritoneal endometriotic lesions. \nA previous immunohistochemical study showed that expression of a marker of active TGF-β 1 signaling, p-Smad \n2, was most pronounced in endometriotic epithelial cells of DIE29. However, it was not clear whether p-Smad 2 \nwas localized in E-cadherin+  endometriotic epithelial cells of DIE. Thus, we further performed double immuno-\nfluorescence staining for E-cadherin and p-Smad 2/3 in endometriotic tissues of DIE and red peritoneal lesions. \nThese findings may support our speculation that TGF-β 1 signaling may be absent in DIE. However, a limitation \nof the present study is a tremendous gap between in vivo tissue findings by immunohistochemical analysis and \nthe present in vitro experiments. In addition, the absence of TGF-β 1 signaling in already developed surgically \nexcised DIE does not indicate that it had been absent during development of DIE. As speculated for cancer\n23,24, it \nis more likely that TGF-β 1 signaling is transiently and locally activated during the development of DIE. Further \nexperiments are required to confirm our speculation.\nHowever, if TGF-β 1 signaling is absent throughout DIE development, how do fibrotic microenvironments \nin DIE develop? TGF-β 1 is the most potent key mediator of fibrosis30. Our previous studies also supported the \nimportance of the TGF-β 1 signaling in fibrosis of endometriosis13,31,32 and suggested that TGF-β 1 induces ECM \nsynthesis and remodeling, as well as myofibroblast differentiation13,31,32. Moreover, evidence suggests that TGF-β \n1 is involved in the pathophysiology of endometriosis12. Our previous study revealed that α SMA+  stress fibers \nin very few or no EES grown on 30-kPa PGS with 72-h TGF-β 1 stimulation13. However, in the present study, we \nshowed that a longer culture duration promoted EES to differentiate into myofibroblasts without TGF-β 1 treat-\nment. In addition, not only myofibroblast cells produce collagen type I: after implantation, EES may differentiate \ninto myofibroblasts and produce collagen type I; increased stiffness through increased myofibroblast collagen \nproduction may then further increase matrix stiffness, resulting in a fibrotic microenvironment in DIE over time. \nThus, the present findings suggest that TGF-β 1 signaling may not be indispensable to the development of the \nfibrotic microenvironment in DIE.\nIn the present study, only epithelial or stromal cells were cultured, but endometriotic tissue and endometrium \nare composed of multiple cell types and extracellular matrix. Cell culture systems that more closely mimic the cel-\nlular complexity typical of in vivo tissues are required to investigate whether and how TGF-β 1 signaling pathway \nis involved in the pathophysiology of different types of pelvic endometriosisin. Such investigations could provide \nimportant information to support the development of novel therapeutic strategies for endometriosis.\nIn conclusion, the present studies showed that cells retain the epithelial-related phenotype in EEE grown \non substrates of various stiffness (2-, 4-, 8-, 16-, and/or 30-kPa), when cells are not stimulated with TGF- β 1. \nHowever, EEE might begin to undergo a partial EMT-like process even on a soft matrix (2-kPa), when cells are \n\nwww.nature.com/scientificreports/\n8\nSCiEntiFiC  RepoRts  | 7:44616 | DOI: 10.1038/srep44616\nstimulated with TGF-β 1. We hypothesize that EEE may undergo an EMT-like process after attachment of endo-\nmetrium to peritoneum in a TGF-β 1–rich microenvironment. However, TGF-β 1 signaling may be absent in DIE, \nresulting in a more epithelial cell-like phenotype in a rigid microenvironment.\nMaterials and Methods\nPatients.  Patients age 20–37 years undergoing laparoscopy for endometriosis were recruited at CHU \nClermont-Ferrand, France for the present study. None of the women had received hormonal therapy and \nnone used intrauterine contraception for at least 6 months prior to surgery. Recruited patients had regu-\nlar menstrual cycles (26–32 days) with confirmation of their menstrual history. Endometrial samples from \n40 patients who had histological evidence of DIE and DIE samples from 5 patients were used for the present \nin vitro analysis. In addition, endometrial tissues from 5 patients without endometriosis (patients with uter -\nine fibroma: n =  2, patients with tubal infertility: n =  3) were obtained. The clinical characteristics of patients \nare shown in Supplementary Table S1. The numbers of samples used for each experiment are summarized in \nSupplementary Table S2.\nThe research protocol was approved by the Consultative Committee for Protection of Persons in Biomedical \nResearch (CPP) of the Auvergne (France) region. All experiments were performed in accordance with the \napproved guidelines and regulations. Informed written consent was obtained from each patient prior to tissue \ncollection.\nCell culture. Endometrial and endometriotic epithelial cells and endometrial stromal cells were isolated as \npreviously described6. Isolated cells were plated onto Primaria flasks (BD, Le Pont-De-Claix, France) in phenol \nred-free Dulbecco’s modified Eagle medium (DMEM)/F-12 (Life Technologies, Cergy Pontoise, France) con-\ntaining 10% charcoal-stripped fetal bovine serum (FBS) (Sigma-Aldrich, Lyon, France), 100  U/mL penicillin \n(Sigma-Aldrich), 0.1 mg/mL streptomycin (Sigma-Aldrich), and 0.25 μ g/mL amphotericin B (Sigma-Aldrich) \nand incubated at 37 °C in 95% air/5% CO\n2. Epithelial cells were incubated at 37 °C in 95% air/5% CO2 for 60 min \nto allow contaminated stromal cells to attach to the flask wall. The nonattached endometrial or endometriotic \nepithelial cells were recovered and used, whereas endometrial stromal cells at passage 1 were used for the present \nexperiments. Immunofluorescence staining was performed to determine the purity of the isolated epithelial and \nstromal cells as previously described\n6,13. The results indicated that the purity of epithelial and stromal cells was  \n> 98% and > 99%, respectively.\nPreparation of polyacrylamide gel supports. Polyacrylamide gels of variable stiffness were pre-\npared on glass coverslips using modifications to the protocol of Fischer et al. 33 as previously described 13. The \npolyacrylamide gel can be maintained for several days. For a longer cell culture of endometrial stromal cells, \nstiffness-controlled 96-well plates were prepared using modifications to the protocol of Syed et al.\n34.\nCells were seeded onto coated gels or plastic in full growth medium (phenol red-free DMEM/F-12 containing \n10% charcoal-stripped FBS, 100 U/mL penicillin, 0.1 mg/mL streptomycin, and 0.25 μ g/mL amphotericin B) (Life \nTechnologies) and incubated for 2–3 h at 37 °C in 95% air/5% CO2 to allow adherence. The supernatant medium \nwas removed and the desired cell culture medium (2% charcoal-stripped FBS with or without TGF-β 1 [5 ng/mL]) \n(R&D Systems, Lille, France) was overlaid.\nThen, endometrial epithelial cells were cultured for 120 h with or without TGF-β 1 on 2-, 4-, 8-, 16-, or 30-kPa \nPGS, or plastic. Endometriotic epithelial cells derived from DIE were cultured for 120 h with or without TGF-β \n1 on 2- or 30-kPa PGS, or plastic. Endometrial stromal cells were cultured for 7 days or 14 days with or without \nTGF-β 1 on 2- or 30-kPa PGS.\nA major disadvantage of polyacrylamide is its cytotoxicity. However, this is not a concern for the present 2D \nculture models in which cells are seeded on top of polyacrylamide gels\n35.\nImmunofluorescence staining. Double immunofluorescence staining of endometrial and endometriotic \nepithelial cells and endometrial stromal cells was performed according to the protocol published by Lee et al.36. \nIn endometrial epithelial cells, double immunofluorescence staining for Ki-67 (D2H10, 1;200, Cell Signaling, \nDanvers, MA, USA) and cytokeratin (CK) (MNF116, 1:100, DAKO, Glostrup, Denmark), F-actin (Alexa Fluor \n594-phalloidin, Life Technologies)/CK, E-cadherin (4A2, 1:50, Cell Signaling)/CK ZO-1 (D6L1E, 1:400, Cell \nSignaling)/CK, and N-cadherin (D4R1H, 1:200, Cell Signaling)/CK was performed. In endometriotic epithelial \ncells, double immunofluorescence staining for E-cadherin/CK was performed. In endometrial stromal cells, dou-\nble immunofluorescence staining for collagen I (rabbit polyclonal, 1:500, Abcam, Cambridge, UK) and α  SMA \n(1A4, 1:100, Merck Millipore) was performed.\nIn addition, double immunofluorescence staining on paraffin sections for E-cadherin (NCH38, 1:50, DAKO) \nand p-Smad 2/3 (rabbit polyclonal, 1:200, Santa Cruz Biotechnology, Santa Cruz, CA, USA) was performed in \ndeep endometriotic and red peritoneal endometriotic tissues. Sections were deparaffinized, and antigen retrieval \nwas performed; sections were then treated with 3% H\n2O2 solution as described previously6. Sections were incu-\nbated overnight at 4 °C with primary antibodies against E-cadherin and p-Smad 2/3.\nAlexa Fluor 488 (green) goat anti-mouse IgG and Alexa Fluor 594 (red) goat anti-rabbit IgG conjugated antibodies \n(Life Technologies) were used as secondary antibodies. Cell nuclei were stained with 4, 6-diamidino-2-phenylindole \n(DAPI) (Life Technologies). Slides were analyzed with a Leica TCS SPE confocal laser-scanning microscope (Leica \nMicrosystems, Nanterre, France). The cellular proliferation index (CPI) (percentage of Ki-67+  cells among the total \nnumber of CK+  cells), the percentage of cells with α SMA+  stress fibers, and the percentage of collagen I+  cells \namong the total number of DAPI-stained nuclei, were calculated from 10 random high-power (x400) fields through \neach section. For p-Smad2/3 nuclear expression, the percentage of p-Smad2/3+  nuclei among the total number of \nDAPI-stained nuclei was calculated from the entire field of each section.\n\nwww.nature.com/scientificreports/\n9\nSCiEntiFiC  RepoRts  | 7:44616 | DOI: 10.1038/srep44616\nAnalysis of apoptosis by Annexin V staining. Cells grown on 2-kPa or 30-kPa PGS or plastic for 120 h \nwith or without TGF- β 1 were stained with Annexin V-FITC (Annexin V kit, Beckman Coulter, Villepinte, \nFrance) according to the manufacturer’s protocol. Cell nuclei were stained with DAPI (Life Technologies). Slides \nwere analyzed with a Leica TCS SPE confocal laser-scanning microscope (Leica Microsystems).\nStatistical analysis.  The STATA program version 12 (StataCorp, College Station, TX, USA) was used \nfor statistical analysis. Comparisons between different groups were made using the Wilcoxon matched pairs \nsigned-ranks test or the Mann-Whitney U test. Statistical significance was defined as p <  0.05.\nReferences\n1. Giudice, L. C. & Kao, L. C. Endometriosis. Lancet. 364, 1789–1799 (2004).\n2. Matsuzaki, S. & Darcha, C. Epithelial to mesenchymal transition-like and mesenchymal to epithelial transition-like processes might \nbe involved in the pathogenesis of pelvic endometriosis. Hum. Reprod. 27, 712–721 (2012).\n3. Nisolle, M. & Donnez, J. Peritoneal endometriosis, ovarian endometriosis, and adenomyotic nodules of the rectovaginal septum are \nthree different entities. Fertil. Steril. 68, 585–596 (1997).\n4. Fazleabas, A. T. et al. Modified baboon model for endometriosis. Ann. N. Y. Acad. Sci. 955, 308–317 (2002).\n5. Hendriksen, J. et al. Plasma membrane recruitment of dephosphorylated beta-catenin upon activation of the Wnt pathway. J. Cell \nSci. 121, 1793–1802 (2008).\n6. Matsuzaki, S. & Darcha, C. In vitro  effects of a small-molecule antagonist of the Tcf/ß-catenin complex on endometrial and \nendometriotic cells of patients with endometriosis. PLOS One. 8, e61690 (2013).\n7. Matsuzaki, S. et al. Fibrogenesis in peritoneal endometriosis. A semi-quantitative analysis of type-I collagen. Gynecol. Obstet. Invest. \n47, 197–199 (1999).\n8. Wei, S. C. et al. Matrix stiffness drives epithelial-mesenchymal transition and tumour metastasis through a TWIST1-G3BP2 \nmechanotransduction pathway. Nat. Cell Biol. 17, 678–688 (2015).\n9. Leight, J. L., Wozniak, M. A., Chen, S., Lynch, M. L. & Chen, C. S. Matrix rigidity regulates a switch between TGF-β  1-induced \napoptosis and epithelial-mesenchymal transition. Mol. Biol. Cell. 23, 781–791 (2012).\n10. Kalluri, R. & Weinberg, R. A. The basics of epithelial-mesenchymal transition. J. Clin. Invest. 119, 1420–1428 (2009).\n11. Lamouille, S., Xu, J. & Derynck, R. Molecular mechanisms of epithelial-mesenchymal transition. Nat. Rev. Mol. Cell. Biol. 15, \n178–196 (2014). Review.\n12. Omwandho, C. O., Konrad, L., Halis, G., Oehmke, F . & Tinneberg, H. R. Role of TGF-betas in normal human endometrium and \nendometriosis. Hum. Reprod. 25, 101–109 (2010).\n13. Matsuzaki, S., Canis, M., Pouly, J. L. & Darcha, C. Soft matrix inhibit cell proliferation and inactivate the fibrotic phenotype of deep \nendometriotic stromal cells in vitro. Hum. Reprod. 31, 541–543 (2016).\n14. Jiang, X. et al. In vivo high-resolution magnetic resonance elastography of the uterine corpus and cervix. Eur. Radiol. 24, 3025–3033 \n(2014).\n15. Georges, P . C. et al. Increased stiffness of the rat liver precedes matrix deposition: implications for fibrosis. Am. J. Physiol. \nGastrointest. Liver. Physiol. 293, G1147–1154 (2007).\n16. Liu, F . et al. Feedback amplification of fibrosis through matrix stiffening and COX-2 suppression. J. Cell. Biol. 190, 693–706 (2010).\n17. Booth, A. J. et al. Acellular normal and fibrotic human lung matrices as a culture system for in vitro investigation. Am. J. Respir. Crit. \nCare. Med. 186, 866–876 (2012).\n18. Brown, A. C., Fiore, V . F ., Sulchek, T. A. & Barker, T. H. Physical and chemical microenvironmental cues orthogonally control the \ndegree and duration of fibrosis-associated epithelial-to-mesenchymal transitions. J. Pathol. 229, 25–35 (2013).\n19. Johnson, L. A. et al. Matrix stiffness corresponding to strictured bowel induces a fibrogenic response in human colonic fibroblasts. \nInflamm. Bowel. Dis. 19, 891–903 (2013).\n20. Thiery, J. P . & Sleeman, J. P . Complex networks orchestrate epithelial-mesenchymal transitions. Nat. Rev. Mol. Cell. Biol. 7, 131–142 \n(2006). Review.\n21. Thiery, J. P ., Acloque, H., Huang, R. Y . & Nieto, M. A. Epithelial-mesenchymal transitions in development and disease. Cell. 139, \n871–890 (2009). Review.\n22. Y oung, V . J., Brown, J. K., Saunders, P . T., Duncan, W . C. & Horne, A. W . The peritoneum is both a source and target of TGF-β  in \nwomen with endometriosis. PLOS One. 9, e106773 (2014).\n23. Giampieri, S. et al. Localized and reversible TGFbeta signalling switches breast cancer cells from cohesive to single cell motility. Nat. \nCell. Biol. 11, 1287–1296 (2009).\n24. Matise, L. A. et al. Lack of transforming growth factor-β  signaling promotes collective cancer cell invasion through tumor-stromal \ncrosstalk. Breast Cancer Res. 14, R98 (2012).\n25. Friedl, P . & Gilmour, D. Collective cell migration in morphogenesis, regeneration and cancer. Nat. Rev. Mol. Cell. Biol. 10, 445–457 \n(2009). Review.\n26. Mayor, R. & Etienne-Manneville, S. The front and rear of collective cell migration. Nat. Rev. Mol. Cell. Biol. 17, 97–109 (2016). \nReview.\n27. Ng, M. R., Besser, A., Danuser, G. & Brugge, J. S. Substrate stiffness regulates cadherin-dependent collective migration through \nmyosin-II contractility. J. Cell Biol. 199, 545–563 (2012).\n28. Donnez, O. et al. Invasion process of induced deep nodular endometriosis in an experimental baboon model: similarities with \ncollective cell migration? Fertil. Steril. 104, 491–497 (2015).\n29. van Kaam, K. J., Schouten, J. P ., Nap, A. W ., Dunselman, G. A. & Groothuis, P . G. Fibromuscular differentiation in deeply infiltrating \nendometriosis is a reaction of resident fibroblasts to the presence of ectopic endometrium. Hum. Reprod. 23, 2692–2700 (2008).\n30. Leask, A. & Abraham, D. J. TGF-beta signaling and the fibrotic response. FASEB. J. 18, 816–827 (2004). Review.\n31. Matsuzaki, S. & Darcha, C. Involvement of the Wnt/β  -catenin signaling pathway in the cellular and molecular mechanisms of \nfibrosis in endometriosis. PLOS ONE. 8, e76808 (2013).\n32. Matsuzaki, S. & Darcha, C. Antifibrotic properties of epigallocatechin-3-gallate in endometriosis. Hum. Reprod. 29, 1677–1687 \n(2013).\n33. Fischer, R. S., Myers, K. A., Gardel, M. L. & Waterman, C. M. Stiffness-controlled three-dimensional extracellular matrices for high-\nresolution imaging of cell behavior. Nat. Protoc. 7, 2056–2066 (2012).\n34. Syed, S., Karadaghy, A. & Zustiak, S. Simple Polyacrylamide-based Multiwell Stiffness Assay for the Study of Stiffness-dependent \nCell Responses. J. Vis. Exp. 97, e52643 (2015).\n35. Smithmyer, M. E., Sawicki, L. A. & Kloxin, A. M. Hydrogel scaffolds as in vitro  models to study fibroblast activation in wound \nhealing and disease. Biomater. Sci. 2, 634–650 (2014).\n36. Lee, G. Y ., Kenny, P . A., Lee, E. H. & Bissell, M. J. Three-dimensional culture models of normal and malignant breast epithelial cells. \nNat. Methods. 4, 359–365 (2007).\n\nwww.nature.com/scientificreports/\n10\nSCiEntiFiC  RepoRts  | 7:44616 | DOI: 10.1038/srep44616\nAcknowledgements\nWe are most grateful to all of the patients who participated in the present study. We acknowledge use of the \nconfocal imaging platform (ICCF , https://www.gred-clermont.fr/directory/platform/) at the GReD laboratory, \nFaculty of Medicine, University of Auvergne. We thank Caroline Vachias, Ph.D. (GReD laboratory, Faculty of \nMedicine, University of Auvergne) for excellent technical assistance in confocal microscopy. This study was \nsupported in part by Karl Storz Endoscopy & GmbH (Tuttlingen, Germany)\nAuthor Contributions\nS.M. was involved in concept and design, sample collection, experiments, acquisition of data, analysis \nand, drafting the article and critical revision of the article. C.D. was involved in interpretation of data. J.L.P . \nwas involved in sample collection and critical revision of the article. M.C. was involved in sample collection, \ninterpretation of data and critical revision of the article. All authors read and approved the final version of the \npaper.\nAdditional Information\nSupplementary information accompanies this paper at http://www.nature.com/srep\nCompeting Interests: The authors declare no competing financial interests.\nHow to cite this article: Matsuzaki, S. et al. Effects of matrix stiffness on epithelial to mesenchymal transition-\nlike processes of endometrial epithelial cells: Implications for the pathogenesis of endometriosis. Sci. Rep. 7, \n44616; doi: 10.1038/srep44616 (2017).\nPublisher's note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and \ninstitutional affiliations.\nThis work is licensed under a Creative Commons Attribution 4.0 International License. The images \nor other third party material in this article are included in the article’s Creative Commons license, \nunless indicated otherwise in the credit line; if the material is not included under the Creative Commons license, \nusers will need to obtain permission from the license holder to reproduce the material. To view a copy of this \nlicense, visit http://creativecommons.org/licenses/by/4.0/\n \n© The Author(s) 2017","source_license":"CC0","license_restricted":false}