Foxf2 and Smad6 co‐regulation of collagen 5A2 transcription is involved in the pathogenesis of intrauterine adhesion

Abstract The replacement of normal endometrial epithelium by fibrotic tissue is the pathological feature of intrauterine adhesion (IUA), which is caused by trauma to the basal layer of the endometrium. COL5A2 is a molecular subtype of collagen V that regulates collagen production in fibrotic tissue. Here, we investigated the roles of Foxf2 and Smad6 in regulating the transcription of COL5A2 and their involvement in the pathogenesis of IUA. Small interference‐mediated Foxf2 (si‐Foxf2) silencing and pcDNA3.1‐mediated Smad6 (pcDNA3.1‐Smad6) up‐regulation were performed in a TGF‐β1‐induced human endometrial stromal cell line (HESC) fibrosis model. Assessment of collagen expression by Western blotting, immunofluorescence and qRT‐PCR showed that COL5A2, COL1A1 and FN were significantly down‐regulated in response to si‐Foxf2 and pcDNA3.1‐Smad6. Transfection of lentivirus vector‐Foxf2 (LV‐Foxf2) and pcDNA3.1‐Smad6 into HESCs and qRT‐PCR showed that Foxf2 promoted COL5A2 expression and Smad6 inhibited Foxf2‐induced COL5A2 expression. Co‐immunoprecipitation, chromatin immunoprecipitation and dual‐luciferase reporter assays to detect the interaction between Foxf2 and Smad6 and their role in COL5A2 transcription showed that Foxf2 interacted with Smad6 and bond the same promoter region of COL5A2. In a rat IUA model, injection of ADV2‐Foxf2‐1810 and ADV4‐Smad6 into the uterine wall showed that Foxf2 down‐regulation and Smad6 up‐regulation decreased fibrosis and the expression of COL5A2 and COL1A1, as detected by haematoxylin/eosin, Masson trichrome staining and immunohistochemistry. Taken together, these results suggested that Foxf2 interacted with Smad6 and co‐regulated COL5A2 transcription in the pathogenesis of IUA, whereas they played opposite roles in fibrosis.


| INTRODUC TI ON
Intrauterine adhesion (IUA) is a disease caused by injury to the basal layer of the endometrium resulting in partial or complete obliteration of the uterine cavity and/or the cervical canal. IUA is a major health problem involving the female reproductive system for women of childbearing age. It can lead to menstrual abnormalities, periodic abdominal pain, recurrent abortion, infertility and pregnancy-related complications, such as placenta adhesion and placenta accrete. 1 Most cases of IUA occur after dilation and curettage for missed abortion, selective termination of pregnancy and postpartum placental residual. 2,3 The pathogenesis of IUA involves decreased or absent endometrial glands, and the endometrial stroma is mostly replaced by fibrous tissue, leading to uterine cavity deformation and endometrial fibrosis. 4 Biopsy samples from the uterine wall of patients with IUA contain 50%-80% of fibrous tissue, compared with 13%-20% in patients without IUA. 5 Excessive deposition of extracellular matrix (ECM) substituting the normal endometrium is the characteristic feature of endometrial fibrosis. 6 Collagen is the major component of the ECM and plays a vital role in wound healing; however, excessive collagen production leads to organ fibrosis. 7,8 More than 20 types of collagen have been found, and the most abundant subtypes are types I, III and V, which expressed extensively in fibrous tissue. 9 In our unpublished study, we collected 15 endometrial specimens including five normal, five moderate IUA and five severe IUA samples, which were used for microarray sequencing for gene expression profiles. The results showed that COL5A1, COL5A2 and COL1A1 were expressed at higher levels in the IUA group than in the normal group, especially COL5A2, which was correlated with the degree of IUA ( Figure 1A,B).
Type V collagen (COLV) is a regulatory fibril-forming collagen that plays an important role in the formation of fibrils; it acts in combination with collagens I and III in the formation of fibrils and their deposition in the ECM. 10 COLV includes three different isoforms, COL5A1, COL5A2 and COL5A3. 11 Abnormal expression of COL5A2 is associated with many fibrous diseases, and COL5A2 expression is increased in tubulointerstitial fibrosis and systemic sclerosis. 12,13 Based on these findings together with the results of our previous study, we have been suggested that COL5A2 might play a vital role in the pathogenesis of IUA. Therefore, we examined the mechanism underlying the regulation of COL5A2 and its role in the pathogenesis of IUA. Forkhead box F2 (Foxf2) is a transcription factor that is widely expressed in mesenchymal tissues and plays an important role in organ development and ECM formation. 14,15 Collagens within the ECM are dramatically decreased in the intestines of Foxf2 knockout mice, resulting in tissue disintegration. 16 This suggests that Foxf2 is important for collagen production.
The TGF-β/Smad signalling pathway is involved in the pathogenesis of IUA, as shown in vivo and in vitro. 17,18 In our previous study, we demonstrated the involvement of the TGF-β/Smad signalling pathway in the fibrosis of primary human endometrial stromal cells. 19 Smad6 is the downstream mediator of the TGF-β superfamily and can negatively regulate the TGF-β signalling pathway. 20 Overexpression of Smad6 inhibits collagen production by suppressing epithelial-mesenchymal transition (EMT). 21 In conjunctival fibrosis, valproic acid decreases collagen expression by up-regulating the expression of Smad6. 22 Both Foxf2 and Smad6 are related to the production of collagen, and Foxf2 and Smad6 expression were correlated with the degree of IUA ( Figure 1B). Therefore, we speculated that Foxf2 and Smad6 are important for the pathogenesis of IUA. Both Foxf2 and Smad6 are transcription factors; however, whether they are correlated with the expression of COL5A2 has not been reported to date. We predicted the presence of transcription factor binding sites for Foxf2 or Smad6 in the promoter region of COL5A2 that could be involved in the regulation of its transcription (http://jaspar.gener eg.net/, http://genome.ucsc.edu/).
In the present study, small interference-mediated Foxf2 silencing (si-Foxf2) and pcDNA3.1-mediated Smad6 up-regulation (pcD-NA3.1-Smad6) were transfected into a TGF-β1-induced HESC fibrosis model in vitro. Western blotting, immunofluorescence (IF) staining and qRT-PCR were used to measure the expression of COL5A2, COL1A1 and FN. The EdU assay and flow cytometry were performed to examine cell proliferation and cycle progression. Co-immunoprecipitation (Co-IP), chromatin immunoprecipitation (ChiP) and dual-luciferase reporter assays were performed to confirm the interaction between Foxf2 and Smad6 and their regulation of COL5A2 transcription.
Foxf2 was down-regulated using ADV2-Foxf2-1810, and Smad6 was up-regulated using ADV4-Smad6 in a female SD rat IUA model in vivo. Haematoxylin and eosin (HE) and Masson trichrome immunostaining were used to examine the glands and the fibrosis area in the endometrium. Immunohistochemistry (IHC) was performed to detect the expression of COL5A2 and COL1A1. We showed that

| si-Foxf2 and pcDNA3.1-Smad6 and recombinant adenovirus constructs and transfection
Small interference-mediated Foxf2 (si-Foxf2) silencing and its negative control were purchased from GenePharma (Shanghai, China). Because the efficiency of si-Foxf2 in vivo was low in our pre-experiment, we chose ADV2-Foxf2-1810 to down-regulate Foxf2 expression and ADV4-Smad6 to up-regulate Smad6 expression in rat IUA model. ADV2-Foxf2-1810 and ADV4-Smad6 were directly injected into rat uterine wall. All the transfection procedures were according to the protocol of the manufacturer.

| RNA extraction, reverse transcription and quantitative real-time PCR (qRT-PCR)
TRIzol reagent (Takara, Japan) was used to extract total RNAs from HESCs or tissues. PrimeScript™RT Reagent Kit with gDNA Eraser (Takara) was used to perform reverse transcription. Quantitative realtime PCR was performed by using the CFX Connect Real-Time System (Bio-Rad, USA) with the SYBR Green Kit (Takara) according to the manufacturer's instruction. The primer sequences were synthesized by Sangon Biotech (Shanghai, China) as listed in Table 2. Calculation of the targeted mRNAs was based on the Cq results and normalization to GAPDH expression. All of the reactions were performed in triplicate.

| Proteins extraction and Western blotting
Total proteins were extracted from HESCs with RIPA lysis buffer  tor for 2 hours, and then fixed with 4% paraformaldehyde. All the experimental procedure was referred to as protocol. 23 The nucleus was stained with Hoechst 33342 for 30 minutes. Cell images were captured with an inverted fluorescence microscope (Leica, Germany). The proliferation rate was defined as the percentage of EdU-positive cells.

| Flow cytometric analysis
For cell cycle analysis, forty-eight hours after transfection with si-

| Immunofluorescent staining (IF)
Seventy-two hours after transfection, the cells were fixed in 4%

| Chromatin immunoprecipitation (ChIP)
We predicted three binding sites at the promoter regions of COL5A2 The sequence of primers for predicted binding sites is listed in Table 3.

| Dual-luciferase reporter assay
To detect which binding site where Foxf2 and Smad6 may bind at the promoter region of COL5A2, we synthesized three binding site sequences by chemosynthesis and amplified them by PCR. The PCR products were inserted into PGL3-basic vector, respectively, PGL3-COL5A2-1, PGL3-COL5A2-2 and PGL3-COL5A2-3 (GenePharma).
Wild-type and mutant Foxf2 and Smad6 plasmids were synthesized  to the endometrium. 24 In this study, we still exerted dual damage to female rat endometrium to create rat IUA model.

| Evaluation of Adhesion Severity
In our previous study, we demonstrated that the time for the formation of IUA model was two weeks optimally. 24 Thus, in this study, the rats were killed by injection of urethane two weeks after creation of IUA model. The uteri were immediately excised, fixed in 4% paraformaldehyde, embedded in paraffin and sliced. Haematoxylineosin and Masson trichrome staining were performed. The number of endometrial glands and the percentage of endometrial fibrosis area were evaluated in a microscope as described previously. 24 The observer was blind to each group.

| Immunohistochemistry
The above paraffin sections were deparaffinized, rehydrated, antigen-retrieved by microwave heat for 20 minutes, and then immersed in 3% hydrogen peroxide for 15 minutes to block endogenous peroxidase activity and blocked in 5% bovine serum albumin for 10 min- and Smad6 were captured by microphotography. The staining was evaluated by semi-quantitatively scored using the modified histochemical score (H-score). Staining intensity was categorized four degrees: 0 for negative, 1+ for weak staining, 2+ for moderate staining and 3+ for strong staining. The percentage of different staining intensity cells was determined by visual assessment. And then, the score was calculated using the formula 1 × (% of 1+ cells) + 2 × (% of 2+ cells) + 3 × (% of 3+ cells) and produced a final score ranging from 0 to 300. 26 All cases were scored without prior knowledge of each group.

| Statistical analysis
Statistical analysis was performed using the SPSS 20.0 soft-

| Establishment of a HESC fibrosis model
The TGF-β signalling pathway plays a pivotal role in the development of fibrosis, and TGF-β1 is an important pro-fibrotic cytokine. 27 In our previous study, we used TGF-β1 to stimulate primary human endometrial stromal cells to generate a cell fibrosis model. The results showed that 10 ng/mL TGF-β1 was the optimal concentration to stimulate fibrosis in primary human endometrial stromal cells. 21 In the present study, we used TGF-β1 to create a HESC fibrosis model. Cells were incubated with different concentrations of TGF-β1 (0, 1, 5, 10 or 15 ng/mL) to select the optimal dosage to stimulate HESCs. COL1A1, COL5A2 and FN expression in the different concentration groups is shown in Figure 2A.
The expression of these proteins was TGF-β1 concentration-dependent, showing a peak of expression at 10 ng/mL. The mRNA expression also increased according to the TGF-β1 concentration, as shown in Figure 2B. COL1A1, COL5A2 and FN mRNA expression increased in a dose-dependent manner in response to TGF-β1 treatment at 1-10 ng/ mL, and the differences between groups were statistically significant (P < .05). However, COL1A1, COL5A2 and FN mRNA expression did not differ between the 10 ng/mL TGF -β1 group and the 15 ng/mL TGF -β1 group. The optimal concentration of TGF -β1 for stimulating HESCs to fibrosis was 10 ng/mL, which was in accordance with our previous study. 19 Thus, we chose 10 ng/mL TGF-β1 to establish a cell fibrosis model for subsequent experiments.

| Foxf2 down-regulation or (and) Smad6 up-regulation have anti-fibrotic effects in TGF-β1stimulated HESCs
Foxf2 and Smad6 may be important for the pathogenesis of IUA, as suggested by our previous study. Here, we showed that Foxf2 was up-regulated and Smad6 was down-regulated in the endometrium of patients with IUA ( Figure 1B)  si-Foxf2-650 groups was lower than that in the pcDNA3.1-Smad6 group, but the difference did not reach statistical significance (P > .05) ( Figure 3A).

| Protein expression of Foxf2, Smad6, COL5A2, COL1A1, FN and α-SMA in each group
The results of Western blot analysis of the protein expression of

| Foxf2 down-regulation or/and Smad6 upregulation affects cell proliferation and cell cycle distribution induced by TGF-β1
We investigated the effect of Foxf2 down-regulation or/and Smad6 up-regulation on the proliferation of TGF-β1-stimulated HESCs using EdU essay ( Figure 4A). The results showed that TGF-β1 promoted cell proliferation; the EdU-positive cell rate in TGF-β1, si-negative control and pcDNA3.1 groups was significantly higher than that in the normal group (P < .05). Down-regulation of Foxf2 or up-regulation of Smad6 expression significantly decreased the cell proliferation induced by TGF-β1 compared with that in the TGF-β1, si-negative control and pcDNA3.1 groups (P < .05).
Flow cytometry assessment of cell cycle distribution showed that stimulation with TGF-β1 affected cell cycle distribution. TGF- group, but the difference was not significant.

| Foxf2 interacts with Smad6 and they coregulate COL5A2 transcription in the pathogenesis of HESC fibrosis
To determine whether These results indicated that Foxf2 interact with Smad6 and bound to the same promoter region of COL5A2 to co-regulate COL5A2 transcription in the pathogenesis of HESC fibrosis, whereas they had opposite effects on COL5A2 transcription.

| Foxf2 down-regulation or/and Smad6 upregulation decrease fibrosis in the pathogenesis of rat IUA
Foxf2 was down-regulated using the ADV2 (CMV/IRES-RFP) vector, and Smad6 was up-regulated using the ADV4 (CMV/IRES-RFP) vector. The efficiencies of transfection were verified by assessing fluorescence expression and qRT-PCR as shown in Figure 6.

| The number of endometrial glands in each group
After transfection of the ADV vector in the IUA model, the uteri were

| The degree of rat endometrial fibrosis in each group
Masson staining was used to evaluate the degree of fibrosis ( Figure 7B).
In the sham group, the fibrosis area was barely detectable, and the endometrial stroma rarely appeared blue, whereas the endometrial stroma in the IUA model and ADV-negative control groups appeared dark blue. The percentage of fibrosis area in the IUA model and ADVnegative control groups was 64.95% ± 4.76% and 62.37% ± 6.67%, respectively, which was significantly higher than that in the sham group (27.35% ± 1.88%) (P < .05). However, the percentage of fibrosis area in the Foxf2 down-regulation group (40.75% ± 4.53%) and the Smad6 up-regulation group (43.77% ± 3.01%) was significantly lower than that in the IUA model and ADV-negative control groups (P < .05). The percentage of the fibrosis area in the combination therapy group was 33.77% ± 2.43%, which was lower than that in the Foxf2 down-regulation group and Smad6 up-regulation group, although the difference was only significant in the Smad6 up-regulation group.
These results indicated that Foxf2 down-regulation or Smad6 up-regulation decreased fibrosis in association with the pathogenesis of rat IUA, and both together were more effective than each alone. The immunostaining was evaluated by semi-quantitatively scored using the modified histochemical score (H-score), and data were showed as mean ± SD. The result showed that down-regulation of Foxf2 or (and) up-regulation of Smad6 inhibited COL1A1 and COL5A2 expression in rat IUA model. # P < .05, compared with sham operation group. *P < .05, compared IUA model and ADV-negative control group. **P < .05, compared with Smad6 up-regulation therapy group groups was significantly higher than in the sham operation group

| D ISCUSS I ON
Intrauterine adhesion is a major reproductive problem for childbearing age women, and it can lead to menstrual abnormalities, pelvic pain, recurrent abortion, infertility and pregnancy-related complications. 1 The incidence of IUA following early pregnancy loss is 6.3%. 28 Moreover, 36-53 million pregnancies are terminated every year worldwide, of which approximately 90% are terminated in the first trimester. 29 In the United States, approximately 1.2 million abortions were performed in 2008. 30 Therefore, IUA remains a public health problem for premenopausal women that needs attention. Clarifying the mechanism underlying the pathogenesis of IUA is important. Dilatation and curettage (D&C), a primary risk factor for IUA, causes damage to the basal layer of the endometrium leading to endometrial fibrosis, in which fibrous tissue replaces stromal tissue accompanied with a decrease or disappearance of glands. As a result, the uterine cavity and/ or the cervical canal become partially or completely obliterated. 31 IUA is characterized by the replacement of the normal endometrium by fibrous tissue as a result of injury to the basal layer; therefore, uncovering the mechanism underlying fibrous tissue formation is important.
Collagen, which is the main constituent of fibrous tissue, is widely and abundantly expressed in fibrous tissue. 10 In our previous study, we showed that the expression of COL5A1 and COL1A1 is higher in the IUA endometrium than in the normal endometrium, and COL5A2 expression is positively associated with the degree of IUA ( Figure 1B).
COL5A2 is a subtype of collagen V that is defined as regulatory fibril-forming collagen. Collagen fibrils are often heterogeneous and contain more than one collagen type 32 . Collagen V plays an important role in the formation of fibrils and combined with collagens I, III and XII; it forms fibrils that are deposited in the ECM, with collagen V localizing to the core of fibrils. 11 COL5A2 is involved in the pathogenesis of fibrosis in vivo and in vitro. In COL5A2 −/− mice, lack of expression of COL5A2 leads to disorganized type I collagen fibrils, and mice exhibit eye and skin abnormalities. 30 The expression of COL5A2 is significantly up-regulated in association with fibrosis in renal epithelial cell lines and rat liver fibrosis. 13,33 In our previous study, we demonstrated that COL5A2 expression is up-regulated and correlated with the degree of IUA (Figure 1). In the present study, we confirmed that COL5A2 is overexpressed in HESCs with fibrosis and in a rat IUA model (Figures 3 and 8B). This suggested that COL5A2 plays a vital role in the pathogenesis of IUA. It is important to explore the mechanism regulating COL5A2 expression to improve our understanding of the development of IUA. However, the mechanism underlying the regulation of COL5A2 production and its association with the pathogenesis of IUA remain unclear.
Foxf2 is a member of the Fox family of transcription factors that play a vital role in cell growth and tissue development and is ubiquitously expressed in mesenchymal cells. 34 It is located on chromosome 6p25.3 and has a "winged helix" DNA binding domain characterized by a highly conserved sequence that mediates its interaction with target genes such TATA-box binding protein (TBP) and transcription factor TFIIB (TFIIB) to promote or inhibit transcription. 35 Foxf2 is important for ECM formation in intestinal fibrosis. In Foxf2 −/− mice, the collagen in the intestine is strikingly deficient, and cell adhesion is defective. 16 In the present study, Foxf2 was significantly overexpressed in vivo and in vitro in fibrotic cells (Figures 3B and 8C Foxf2 and Smad6 play crucial roles in the production of collagen. Based on our previous study showing that Foxf2 is up-regulated and Smad6 is down-regulated in IUA patients compared with healthy controls (Figure 1 We then examined the effect of Foxf2 and Smad6 on cell proliferation and cell cycle distribution. TGF-β controls many fundamental cell behaviours, stimulating proliferation and altering cell cycle distribution. 43 In the present study, TGF-β1 induced HESC proliferation and promoted cell cycle progression from G0/G1 to S phase ( Figure 4A,B). However, it had been reported that TGF-β1 inhibited cell proliferation in many studies. [44][45][46] TGF-β1 may have a dual role on cell proliferation, and there had also many studies reported that TGF-β1 promotes proliferation. 47,48 The TGF-β signalling pathway is initiated when an activated TGF-β ligand binds to its membrane receptors, TGF-β type II receptor (TβRII) and TGF-β type I receptor (TβRI). This results in the formation of a complex that activates R-Smad phosphorylation and combines with Smad4 to exert its biological activities. 19 Smad6 competes with R-Smads and inhibits R-Smad phosphorylation and activation to change cell behaviours. 20 Other transcription factors can regulate TGF-β1 signalling. It had been reported that forkhead subfamily proteins interact with Smad3-Smad4 complexes in epithelial cells and promote the expression of the CDK inhibitors p21 CIP1 and p15 INK4B , affecting cell cycle progression. 49 In the present study, both down-regulation of Foxf2 and up-regulation of Smad6 inhibited TGF-β1-induced proliferation and cell cycle distribution.
Foxf2 and Smad6 are transcription factors and play important roles in the fibrosis of HESCs, although the underlying mechanism remains unclear. Smad3 and Smad4 interact with the forkhead transcription factor FOXL2 to regulate Fshb, Gnrhr and Fst transcription in vitro. 50 Whether Foxf2 interacts with Smad6 during fibrosis development has not been reported to date. We performed Co-IP assays to detect whether Foxf2 interacted with Smad6 in HESC fibrosis. As shown in Figure 5A, Foxf2 interacted with Smad6 in association with the pathogenesis of HESC fibrosis.
In the pathogenesis of HESC fibrosis, Foxf2 and Smad6 interact with each other and play opposite roles in collagen production, although the underlying mechanism remains unclear. Based on the finding that combined down-regulation of Foxf2 and up-regulation of Smad6 reduced COL5A2 expression more effectively than single treatment ( Figure 3A,B), we predicted that Foxf2 and Smad6 may bind to the promoter region of COL5A2 and designed three binding site sequences to which Foxf2 and Smad6 could bind (http://jaspar. gener eg.net/, http://genome.ucsc.edu/). ChIP and luciferase assays were performed to test the interaction. The results of ChIP assay showed that there are potential binding sites to which Foxf2 and Smad6 may bind at the promoter region of COL5A2 ( Figure 5B,C).
We then constructed a vector and performed luciferase assays to determine which binding site they bind to. The results showed that expression. Therefore, we concluded that Foxf2 and Smad6 had an opposite role in the pathogenesis of fibrosis.
In the present study, we showed that Foxf2 and Smad6 are involved in the pathogenesis of fibrosis in vitro and in vivo. Foxf2 promoted collagen production, whereas Smad6 inhibited it, and both factors bound to the same promoter region of COL5A2 to regulate its transcription. However, the balance between Foxf2 and Smad6 for regulating COL5A2 remains unclear. The regulation of COL5A2 transcription by Foxf2 and Smad6 may involve competition between the two factors for binding to the promoter region of COL5A2, or Smad6 interaction with Foxf2 to change its activity in binding to the promoter region of COL5A2.

| CON CLUS ION
The present results suggested that Foxf2 and Smad6 co-regulated COL5A2 transcription in the pathogenesis of IUA. Foxf2 down-regulation or/and Smad6 up-regulation may inhibit collagen production and cell proliferation during fibrosis, providing a potential new strategy for the prevention and treatment of IUA.

ACK N OWLED G EM ENTS
The authors would like to thank Dr Dongxian Peng, Xuefeng Wang, Xiafei Fu and Ying Ma for valuable discussions of this study. We thank the International Science Editing (http://www.inter natio nalsc ience editi ng.com) for editing this manuscript. The work was supported by a grant from the National Natural Science Foundation of China

CO N FLI C T O F I NTE R E S T
The author(s) declared no potential conflicts of interest concerning the research, authorship and publication of this article.