Epiregulin promotes trophoblast epithelial–mesenchymal transition through poFUT1 and O‐fucosylation by poFUT1 on uPA

Abstract Objectives The transformation of cytotrophoblasts into mesenchymal‐like extravillous trophoblasts is necessary for successful embryo implantation, and the inadequate transformation may cause abortion. Epiregulin, which is a new growth factor, plays important roles in the reproductive processes. The glycosylation of many proteins in reproduction processes is critical. Protein O‐fucosyltransferase 1 (poFUT1) is the key enzyme for the biosynthesis of O‐fucosylation on the specific glycoproteins. Urokinase‐type plasminogen activator (uPA) contains O‐fucosylated domain on Thr18. However, the functions of epiregulin and poFUT1 in the trophoblast epithelial–mesenchymal transition (EMT) process, the regulatory mechanism of epiregulin on poFUT1 and the resulting O‐fucosylated uPA remain unclear. Materials and methods We employed ELISA and Western blot to detect serum levels of epiregulin and poFUT1 from non‐pregnancy women, pregnancy women and abortion patients. Using two trophoblast cell lines and a mouse pregnancy model, we investigated the underlying mechanisms of epiregulin and poFUT1 in trophoblast EMT process. Results Serum levels of epiregulin and poFUT1 were higher in pregnant women compared with non‐pregnant women, and their levels were significantly decreased in abortion patients compared with pregnant women. The results showed that epiregulin upregulated poFUT1 expression and increased O‐fucosylation on uPA, which further activated the PI3K/Akt signalling pathway, facilitating EMT behaviour of trophoblast cells and embryo implantation in the mouse pregnant model. Conclusions Level of epiregulin and poFUT1 is lower in abortion patients than early pregnancy women. Epiregulin promotes trophoblast EMT through O‐fucosylation on uPA catalysed by poFUT1. Epiregulin and poFUT1 may be suggested as the potential diagnostic biomarkers and useful treatment targets for abortion.


| INTRODUC TI ON
Embryo implantation is the process by which the mature blastocyst successfully attaches to the receptive endometrium, followed by the establishment of the placenta, which plays a crucial role in reproduction. 1,2 The trophoblasts are derived from trophectoderm cells in the blastocyst, and differentiate into syncytiotrophoblasts and cytotrophoblasts, which participate in the nutrients and gas exchange for the foetus. The extravillous trophoblast (EVT) cells originated from the cell column of villi act as a diver to lead embryo to invade into the maternal decidua, followed by the placentation and remodelling of the uterine spiral arteries. [3][4][5] During the differentiation process of cytotrophoblasts into EVT cells, cytotrophoblasts undergo epithelial-mesenchymal transition (EMT) and secret extracellular matrix (ECM) degradation-related proteins (MMP-2/MMP- 9) to facilitate embryo to seeding, further implantation. Inappropriate or shallow invasion of trophoblast cells is the major reasons for pregnancy-related complications, such as miscarriage, intrauterine growth restriction and preeclampsia. 6,7 Despite a large number of molecules involved in the embryo-maternal communication have been identified, the detail mechanism related to how the molecules regulate trophoblasts cell migration and invasion capacity requires further study.
Epiregulin belongs to the epidermal growth factor (EGF) family.
It contains 46 amino acids and is secreted in soluble form. Epiregulin usually binds to the EGF receptor (ErbB1 and ErbB4). 7,8 The increasing evidence indicates that epiregulin plays important roles in the reproductive processes. The human epiregulin is mainly found in the placenta and uterus at the site of blastocyst implantation, 9 and Haengseok Song also reported that epiregulin expressed in the luminal epithelium and the underlying stroma surrounding the blastocysts during the attachment reaction. 10 These studies indicate that epiregulin is involved in embryo implantation. Dysregulation of the EGF family in the uterus induces implantation failure. HB-EGF affects blastocyst activities related to implantation. 11 Furthermore, the EGF signalling cascades are necessary in regulating trophoblast differentiation, and its disruption could cause perinatal diseases, such as preeclampsia and intrauterine growth restriction. However, whether abnormal epiregulin can influence blastocyst implantation remains unclear.
Glycosylation is a post-translational modification of proteins, which plays a crucial role in the reproduction. Fucosylation, a typical type of glycosylation, is classified into two forms, N-fucosylation and O-fucosylation. Fucosyltransferases, including N-fucosyltransferases (FUTs) and protein O-fucosyltransferases (poFUT1/poFUT2), transfer fucose residue to the acceptor by N-linkage or O-linkage, respectively. 12-14 poFUT1 adds O-linked fucose to the folding EGF repeats containing the consensus sequence C 2 -X-X-X-X-(S/T)-C 3 . 15 Fucosylation and fucosyltransferases are critical for reproduction, and aberrant fucosylation of proteins is associated with the reproduction disorders. N-glycosylation of embryo proteins is required to facilitate embryo attachment and invasion into the epithelium. 16,17 Overexpression of fucosyltransferase 7 stimulates embryo adhesion and implantation. 18 Leukaemia inhibitory factor promotes embryo adhesion through unregulated expression of FUT1 and Lewis Y. 19 We have previously found that poFUT1 significantly promotes the proliferation, migration and invasion of trophoblast cells in vitro. However, the role of epiregulin on poFUT1/O-fucosylation in the process of trophoblast invasion remains unknown.
Urokinase-type plasminogen activator (uPA) is a single-polypeptide-chain glycosylated zymogen that consists of three do- regulating cell proliferation, migration, invasion and angiogenesis, etc For example, uPA/uPAR system is involved in the male reproductive. uPA can stimulate sperm mobility and promote fertilization. 25,26 uPA is also secreted by trophoblast and foetal membrane cells and endovascular cells, which facilitates trophoblast invasion into the uterine endometrium. 26 However, lower uPA may prevent trophoblast invasion, and possibly as a preeclampsia marker. 27,28 These findings suggest that trophoblast uPA/uPAR is a critical determinant of embryo implantation. uPA contains the fucosylated EGF domain in human SaOS-2 osteosarcoma cells and U-937 lymphoma cells. However, the relationship between uPA, especially the O-fucosylation of uPA, and miscarriage is still unclear.
In the present study, the serum level of epiregulin and poFUT1 in pregnancy women and abortion patients were examined, and the decreased levels of epiregulin and poFUT1 were associated with abortion. The upregulation of poFUT1 by epiregulin increases the fucosylation of uPA, which activates uPA/uPAR-mediated PI3K/Akt signalling pathway, and facilitated EMT of trophoblast cells.

| Cell culture
The human trophoblastic HTR-8/SVneo and JAR cell lines were obtained from the American Type Culture Collection. Cells were cultured in DMEM/F12 (HyClone) conditional medium supplemented with 10% FBS (Gibco) and 1% penicillin-streptomycin. The medium was renewed every 2-3 days. Cells were maintained in a humidified atmosphere containing 5% CO 2 at 37°C.

| Real-time PCR
Cells were treated with RNAiso Plus reagent (Takara) for RNA extraction, and the PrimeScript RT Reagent Kit with a gDNA Eraser kit (Takara) was used to synthesize cDNA. SYBR Premix Ex Taq  Technologies). Quantified data were normalized to those of GAPDH, and the relative quantity was calculated using the 2 −ΔΔCT method.

| Western blot
To prepare whole-cell protein lysates, cells at 90% confluence were washed in PBS before incubation with RIPA lysis buffer. Equal protein was loaded onto 10% SDS-PAGE gels, transferred onto nitrocellulose membranes and blocked with TTBS containing 5% fat-free

| Immunofluorescent staining
Cover slips (cells) or frozen slices (tissues) were fixed in 4% paraformaldehyde or cold acetone for 30 minutes, followed by blocking with 1% goat serum (Beyotime) for 2 hours. Next, the cover slips or slices were incubated with different antibodies at proper dilutions: poFUT1 was applied at 4°C overnight followed by incubation with the FITC or TRITC-conjugated second antibody for 1 hour. After counterstaining with DAPI (blue) for 5 minutes, anti-fade solution (Beyotime) was added to mount the coverslips or slices before imaging under the fluorescent microscope (Olympus).

| Labelling and detection of glycoproteins in cell extracts
Cells seeded on culture dish were treated with/without test sugars (200 μmol/L 6-alkynyl fucose) in growth medium at 37°C. After 3 days, cell extracts were prepared by resuspending the cells in 200 μL lysis buffer. Protein extract (1 mg/mL) was labelled for 1 hour at room temperature (azido-probe). The Click-iT ® Protein Reaction Buffer Kit includes the reagents required to perform the click reaction on proteins labelled with an azide-tagged biomolecule. Labelled protein lysate was resolved by SDS/PAGE.

| Microscopic analysis of fluorescent labelling O-glycosylation in cells
HTR-8/SVneo and JAR cells were seeded onto six-well plates containing glass coverslips. Growth medium was supplemented with 200 μmol/L 6-alkynyl fucose. After growing for 3 days, cells on coverslips were fixed and permeabilized with acetone for 10 minutes then subjected to the probe labelling reaction. Subsequently, the fixed and labelled cells were stained with TRITC-conjugated streptavidin for 1 hour. DAPI was used to stain nuclei. Fluorescent images were captured by fluorescent microscope.

| Immunoprecipitation
Immunoprecipitation was performed with the Dynabeads ® Protein G Kit (Life Technologies) by following the standard procedures. Briefly, total protein lysates were added to the Dynabeads-uPA or uPAR antibody (Ab) complex for 30 minutes at room temperate, followed by washing 3 times. Subsequently, the Dynabeads-Ab-protein complex was mixed with elution and lysis buffer, and incubated 15 minutes at 70°C to denature the proteins.

| Mouse embryos collection
All animal experiments from this paper were approved by the Animal Ethics Committee of Dalian Medical University. The species of Kunming mice (7-12 weeks) were from the Laboratory Animal Center of Dalian Medical University, China. The mice were feeding in stable environmental conditions (temperature 20-25°C; humidity: 60%).
After mating, when the female mice were found a vaginal plug in the next morning, it was defined as pregnant day 1. On the day 3.5, the pregnant mice were euthanized by cervical dislocation and cut out the uteri. In order to get embryos, the uterine cavity was flushed with warm PBS (without Ca 2+ and Mg 2+ ). Next, embryos were placed in 96-well plates and cultured at 37°C under 5% CO 2 in humidified air according to the standard procedures.

| Mouse embryo transfer
The pseudopregnant recipient females used for embryo transfer were obtained by natural mating with vasectomized males. The seminal secretions produced by a sterile male were required for the uterus to become receptive to the transferred embryos. To obtain a recipient, 2 females of 7-12 weeks of age were placed with a vasectomized male in the afternoon. The following morning, females were checked for the presence of a vaginal copulation plug, a clump of coagulated proteins from the male seminal fluid.
The day of vaginal plug detection was considered to be day 0.5. Different treatments of mouse morulae or blastocysts were transferred to the bilateral uterus of pseudopregnant recipient female at days 2.5-3. The uterus was removed from the recipient 8 days after embryo transfer.

| Statistical analysis
GraphPad Prism ® (GraphPad Software Inc) was used for statistical analysis. All experiments were performed at least 3 independent times, and the data were shown as means ± SEM. For the analysis of difference between groups, independent samples t test was performed. The Spearman correlation analysis was used to analyse the relationship between epiregulin and poFUT1 in the serum. The receiver operating characteristic curve was applied to evaluate the diagnostic value of the epiregulin and poFUT1. The statistical significance was indicated as the follows: *P < .05, **P < .01 and ***P < .001.

| Low epiregulin, poFUT1 and uPA levels in abortion patients
We first performed ELISA and Western blot to determine the levels of epiregulin, poFUT1 and uPA in the serum from non-pregnant, pregnant and abortive women. As shown by ELISA (Figure 1), the levels of epiregulin, poFUT1 and uPA were decreased in the serum of abortion patients compared with pregnant women ( Figure 1A). Western blot also showed that epiregulin and poFUT1 levels were higher in the pregnant women than in the abortion patients ( Figure 1B). It was worth noting that there was a positive correlation between the epiregulin and poFUT1 expression levels in the serum of pregnant women (r = .9710) and abortion patients (r = .9237) ( Figure 1C). We further evaluated the diagnostic value of poFUT1 and epiregulin by receiver operating characteristic curve (ROC) analysis and determined that epiregulin (area under the curve: 1) and poFUT1 (area under the curve: 0.796) were potential diagnostic markers for abortion patients ( Figure 1D). We compared the expression of epiregulin and poFUT1 in the villi of pregnant women and abortion patients ( Figure 1E). Immunofluorescence staining showed that epiregulin and poFUT1 were localized in villous trophoblast cells, and both stainings were stronger in normal pregnant women than in abortion patients ( Figure 1F,G). We further explored the expression of EMT markers (N-cadherin and E-cadherin) in the villi of pregnant women and abortion patients by immunoblotting and immunostaining ( Figure 1H). The staining showed higher N-cadherin and lower E-cadherin levels in pregnant women than in abortion patients ( Figure 1D-F). These results indicate that low level of epiregulin, poFUT1 and uPA is related to women abortion.

| Epiregulin upregulates poFUT1 expression through AP-1
To explore whether epiregulin promotes EMT of trophoblast cells through poFUT1, we further detected the effect of epiregulin on poFUT1. Using real-time PCR, Western blot and immunofluorescence staining, the effect of epiregulin on poFUT1 expression was investigated. Cells were treated with epiregulin at different F I G U R E 1 Epiregulin and poFUT1 levels in pregnant and abortion women. Serum levels of epiregulin, poFUT1 and uPA in non-pregnancy, pregnancy and abortion patients detected by ELISA (A) and Western blot (B). C, Correlation analysis between Epiregulin and poFUT1 expression in serum of pregnancy and abortion patients. D, ROC curve analysis of the diagnostic between epiregulin and poFUT1. E, poFUT1, epiregulin, E-cadherin and N-cadherin expression in villi tissues were detected by Western blot. F, Immunofluorescent staining of poFUT1 (green) in tissues. CK-7 (red) was stained as the villi marker. Nuclei were stained with the DAPI (blue). G, Immunofluorescent staining of epiregulin (red) in tissues. CK-7 (green) was stained as the villi marker. H, Villi tissues were analysed for N-cadherin (green) and E-cadherin (red) by immunofluorescent staining. Scale bars, 100 μm. CBB, Coomassie Brilliant Blue; poFUT1, protein O-fucosyltransferase 1; ROC, receiver operating characteristic; uPA, urokinase-type plasminogen activator F I G U R E 2 Epiregulin promotes EMT of human trophoblast cells. A, JAR and (B) HTR-8/SVneo cells were treated with epiregulin at different concentrations (0, 10, 50, 100 ng/mL) and with 50 ng/mL epiregulin for different times (0, 24, 48, 72 h). Total RNA was analysed for E-cadherin and N-cadherin expression by RT-PCR, with GAPDH serving as the internal control. C, JAR and (D) HTR-8/SVneo cell lysates were prepared for immunoblotting analysis for E-cadherin, N-cadherin and vimentin, and GAPDH served as the internal control. E and F, Analysis of MMP9 activity by gelatine zymography. G and H, Cell extension, and invasive ability were examined by Transwell invasion assays. I, JAR and (J) HTR-8/SVneo cells were analysed for E-cadherin (green) and N-cadherin (red) by immunofluorescent staining. DAPI was used for nuclear staining. Scale bars: 50 μm. The statistical analysis was shown: *P < .05; **P < .01; ***P < .001. EMT, epithelial-mesenchymal transition concentrations of (0, 10, 50, 100 ng/mL) for 48 hours, and the expression of poFUT1 was significantly increased (Figure 3A-D).
Calnexin was used as a marker for Golgi apartment localization. To determine whether AP-1 (c-Fos and c-Jun) was involved in epiregulin-induced poFUT1 expression, cells were treated with epiregulin (50 ng/mL), epiregulin antibody and EGFR antagonist, the poFUT1 level was detected. Western blot results showed that poFUT1 expression was increased after epiregulin treatment, whereas poFUT1 expression was decreased after incubation with epiregulin antibody or EGFR antagonist. We then measured the c-Jun

| Epiregulin promotes trophoblastic cell invasion by increasing poFUT1 expression and O-fucosylation on uPA, which stimulates the PI3K/Akt signalling pathway and MMP9 activity
The Western blot results showed that epiregulin reduced E-cadherin expression, whereas it increased vimentin and N-cadherin expression. Gelatine zymography data also showed that epiregulin increased MMP9 activity (Figure 6E-H). Silenced poFUT1 was associated with an enhancement of E-cadherin expression, whereas vimentin/N-cadherin expression and MMP9 activity were reduced.
Meanwhile, LY294002 inactivated EMT process. Similarity, TIIA, an inhibitor of AP-1, depressed the EMT. Treatment cells with LY294002 and TIIA also inhibited the activity of MMP9 ( Figure 6E-H). Western blot results showed that poFUT1 siRNA inhibited the activation of PI3K/Akt signalling pathway, and epiregulin activated the signalling pathway ( Figure 6A-D). Transwell assays were performed to further verify that epiregulin could promote trophoblast cell invasion by targeting poFUT1 and O-fucosylation on uPA through PI3K/Akt signalling pathway ( Figure 6I,J). The results show that epiregulin promotes invasion through the activated PI3K/Akt signalling pathway, whereas poFUT1 knockdown and treatment with LY294002 and TIIA suppress invasion through PI3K/Akt signalling pathway ( Figure 6).

| Epiregulin promotes embryo implantation in vivo
To further investigate the effects of epiregulin on embryo implantation, a pregnant mouse model was employed. Mouse embryos were collected at PD3.5 and were treated with epiregulin. After incubated 48 hours, the embryos in each group were observed, and the immunofluorescent staining results showed that epiregulin promoted the expression of poFUT1 and invasion potential of mouse embryos ( Figure 7A). Transfection of mouse embryos with poFUT1 siRNA or poFUT1 cDNA inhibited and promoted the expression of poFUT1 and the capacity for invasion, respectively ( Figure 7B). Most importantly, that epiregulin reduced E-cadherin expression, whereas it increased N-cadherin expression in mouse embryos. The invasion potential cells migrated far from the embryo centre ( Figure 7C).
Then, we observed that epiregulin increased the expression of HLA-G, which is an EVT trophoblast cell marker ( Figure 7D). Then, we further analysis the role of epiregulin and poFUT1 on embryo implantation rate. Mouse morulae or blastocysts were incubated with epiregulin antibody or poFUT1 siRNA for 48 hours; then, the embryos were transferred into pseudocyesis mice uterus, and the mice were sacrificed at PD8 to analyse the embryo implantation rate. The statistical results showed that poFUT1 siRNA suppressed the embryo implantation rate compared with the control group, and epiregulin antibody blockade group decreased the embryo implantation rate compared with control group (Figure 7E,F). These findings suggest that epiregulin and poFUT1 are essential for promoting embryo invasion and successful embryo implantation in mice.

| D ISCUSS I ON
During embryo implantation, the transformation from the trophoblastic epithelial-like cytotrophoblast (CTB) to the mesenchymallike EVT is an essential process that facilitates embryo invasion of the uterine epithelium. The CTB-EVT conversion process has been described as "pseudo-EMT." [29][30][31] Many regulators affect EVT migration and invasion, such as hormones, cytokines and chemokines. For example, gonadotropin induces trophoblast cell migration and invasion by EMT upregulation. 32 MMP14 activation is critical for EMT and migration by regulating the levels of cadherins in the cells. 33 Autocrine TGFβ/TGFβR signalling can stimulate the EMT in the cancer cells. 34 Our results also provided the evidence that poFUT1, epiregulin and uPA were expressed at lower levels in trophoblast from abortion patients ( Figure 1A) than normal pregnant women, and decreased level of these molecules could hamper embryo implantation by interfering the trophoblast EMT process.  46 We previously found that proges-

CO N FLI C T O F I NTE R E S T
The authors declare that they have no competing interests.

FIGURE 7
Epiregulin promotes embryo implantation in vivo. Mouse embryos were collected at PD4 and treated with epiregulin for 48 h, were analysis of poFUT1 (red) and CK-7 (green) by immunofluorescent staining (A). The mouse embryos were transfected with poFUT1 siRNA and poFUT1 cDNA were analysis of poFUT1 (red) and CK-7 (green) by immunofluorescent staining (B). Treated with epiregulin, mouse embryos were analysis of N-cadherin (green) and E-cadherin (red) by immunofluorescent staining (C). And mouse embryos were analysis of poFUT1 (green) and HLA-G (red) by immunofluorescent staining (D). DAPI was used for nuclear staining. Scale bars: 50 μm.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data that support the findings of this study are available from the corresponding author upon reasonable request.