Histone deacetylase HDAC2 silencing prevents endometriosis by activating the HNF4A/ARID1A axis

Abstract Endometriosis is the most major cause of chronic pelvic pain in women of reproductive age. Moreover, the involvement of histone deacetylase 2 (HDAC2) has been identified in endometriosis. However, the specific mechanism of HDAC2 remains to be further elusive. Therefore, this study was designed to explore the mechanism of HDAC2 orchestrating hepatocyte nuclear factor 4α/AT‐rich interactive domain 1A (HNF4A/ARID1A) axis in endometriosis. Endometriosis cell line hEM15A and clinical endometriosis tissues were obtained, followed by gain‐ and loss‐of‐function assays in hEM15A cells. HDAC2, HNF4A and ARID1A expression was detected by immunohistochemistry and Western blot analysis. Cell viability was determined by Cell Counting Kit‐8 Assay, invasion by Transwell assay and apoptosis by flow cytometry. HDAC2 enrichment in HNF4A promoter region and HNF4A enrichment in ARID1A promoter region was detected through chromatin immunoprecipitation. Mouse models of endometriosis were established, followed by immunohistochemistry of Ki‐67 expression and TUNEL staining of apoptosis in ectopic tissues. HDAC2 was upregulated but HNF4A and ARID1A were downregulated in endometriosis tissues. HDAC2 inhibited HNF4A expression by deacetylation, and HNF4A was enriched in ARID1A promoter region to activate ARID1A. Silencing HDAC2 or overexpressing HNF4A or ARID1A diminished the viability and invasion and augmented the apoptosis of hEM15A cells. HDAC2 silencing reduced the area and weight of endometriosis tissues, suppressed endometriosis cell proliferation and accelerated endometriosis cell apoptosis. The inhibitory action of silencing HDAC2 via HNF4A/ARID1A axis was reproduced in mouse models. Collectively, HDAC2 silencing might upregulate HNF4A via repression of deacetylation to activate ARID1A, thus preventing the occurrence of endometriosis.


| INTRODUC TI ON
Endometriosis is defined as the presence of endometrial glands and stroma outside the normal location of the endometrium. 1 As an oestrogen-dependent gynaecological disease, endometriosis possesses the feature of the existence and growth of ectopic endometrial tissues and usually correlates to inflammation, severe and chronic pain, and infertility. 2 Additionally, endometriosis is the most principled cause of chronic pelvic pain in women of reproductive age and is closely associated with sustained episodes of ovulation, menstruation and circulating steroid hormones. 3 Unfortunately, current treatments, depending on the primary indication (infertility or pelvic pain), are limited to surgical and hormonal treatments and analgesics with many adverse effects that rarely provide long-term relief. 4 Hence, it is necessary to explore the molecular mechanism underlying endometriosis for better and safer therapies.
Histone deacetylase 2 (HDAC2) is one of the class I histone deacetylase, which manipulates epigenetic landscape through histone modification. 5 It has been reported that HDAC2 is capable of orchestrating human smooth muscle cells of the uterus. 6 Furthermore, a prior study has elucidated the upregulation of HDAC2 in endometriosis. 7 According to a literature, HDAC2 overexpression contributes to the deacetylation of the hepatocyte nuclear factor 4α (HNF4A) transcription factor during Alzheimer's disease. 8 Importantly, the involvement of HNF4A has been noted in the pathogenesis of endometriosis. 9 It has been documented that HNF4A can act as a transcription factor to promote the expression of downstream genes. 10 However, the relationship between HNF4A and AT-rich interactive domain 1A (ARID1A) has been rarely investigated, which warranted our research to study their relationship. AT-rich interaction domain 1A (ARID1A) is a 250 kD switch/sucrose non-fermentable chromatin remodelling complex subunit with known tumour suppressor function and is associated with both endometriosis and orchestration of endometrial receptivity. 11 Intriguingly, ARID1A downregulation has been elaborated to be involved in the pathogenesis of endometriosis. 12 In this context, we speculated that the network of HDAC2, HNF4A and ARID1A might be correlated with the pathogenesis of endometriosis. Therefore, tissues, cells and animal experiments were implemented in this research to verify this speculation, thus providing a novel candidate target for endometriosis treatment.

| Study subjects
A total of 40 patients with ovarian endometriosis were selected in The Second Affiliated Hospital of Guangxi Medical University from January 2017 to December 2019. All patients were diagnosed as endometriosis by pathology. Among patients, their age ranged from 27 to 54 years, with a mean age of 38.88 ± 7.26 years. In addition, 30 patients who underwent hysterectomy due to hysteromyoma were enrolled as the positive control group, with a mean age of 39.80 ± 6.96 years. All patients were not treated with sex hormone nor anti-endometriosis drugs in the first three months.

| Immunohistochemical staining
The tissue specimens were embedded in paraffin, sectioned, dewaxed and hydrated. The sections were washed in 3% methanol H 2 O 2 for 20 min, followed by antigen retrieval with citrate buffer in pressure cooker (2 min at 100℃ and 5 min at room temperature). Co., Ltd.) for 1 min and blued with 1% ammonia water. The sections were observed and photographed under a microscope after sealing.
Five visual fields were randomly selected from each section to observe and analyse the statistics.

| Reverse transcription-quantitative polymerase chain reaction (RT-qPCR)
TRIzol (Invitrogen) was applied to extract total RNA from tissues and cells. A NanoDrop2000 micro ultraviolet spectrophotometer (1011U, NanoDrop Technologies) was adopted to detect the concentration and purity of total RNA. As per the manuals of a PrimeScript RT reagent Kit (RR047A, Takara), cDNA was generated from RNA.
Primers for HNF4A and ARID1 were designed and synthesized by Takara Company (Table S1). Real-time fluorescent qPCR was implemented in ABI7500 qPCR instrument (7500, ABI). The relative transcription level of target gene was calculated by 2 −ΔΔCT method with glyceraldehyde-3-phosphate dehydrogenase (GAPDH) as a normalizer.

| Western blot analysis
The cultured cells were collected and lysed with enhanced Radio-Immunoprecipitation assay cell lysis buffer containing protease inhibitor (Boster Biological Technology Co., Ltd.). Then, the bicinchoninic acid protein quantitative kit (Boster Biological Technology Co., Ltd.) was adopted to estimate the protein concentration. The protein was transferred to a polyvinylidene fluoride membrane after 10% sodium dodecyl sulphate polyacrylamide gel electrophoresis.

| Cell counting kit (CCK)-8 experiment
A CCK-8 assay (CK04, Dojindo Laboratories) was employed to detect cell viability. Cells were seeded into a 96-well plate at 3 × 10 3 cells/ well. After treatment, 10 μl CCK-8 solution and 100 μl fresh medium were added into each well at 0, 24, 48 and 72 h, and incubated at 37℃ for 2 h. The optical density value was detected at 450 nm using a microplate reader (Bio-Rad 680, Bio-Rad Laboratories). Each sample had five duplicated wells.

| Transwell invasion experiment
Before the experiment, the filter membrane was coated with 50 μl

| Flow cytometry
After 48 h of transfection, hEM15A cells were centrifuged at 500-1000 g for 5 min, and the medium was discarded. hEM15A cells were washed once with 3 mL phosphate buffer saline (PBS). After PBS was removed by centrifugation, cells were fixed at 4℃ with 70% ethanol precooled by ice for 1-2 h. Centrifugation was implemented to discard the fixative, and cells were resuspended with 3 ml PBS for 5 min. The suspension was filtered once with a 400-mesh sieve and centrifuged at 500-1000 g for 5 min, followed by discarding of PBS.
The apoptotic cells were double-stained with propidium iodide and Annexin V-Fluorescein Isothiocyanate Apoptosis Detection Kit I (BD Biosciences). The apoptosis was detected by a flow cytometer (Bio-Rad ZE5, Bio-Rad Laboratories).

| Chromatin immunoprecipitation (ChIP) assay
A ChIP kit (EMD Millipore) was conducted to study the protein enrichment of HNF4A and ARID1A promoters. The cells in the logarithmic growth phase were added with 1% formaldehyde and fixed at room temperature for 10 min to cross-link DNA and protein. After crosslinking, the samples were randomly broken by ultrasonic treatment for 15 cycles at the interval of 10 s with 10 s for each time. After centrifugation at 4℃ and 34017g (some DNA fragments as input), the supernatant was collected and aliquoted into two tubes which were respectively supplemented with IgG (ab18413, 1:50) of normal mice as NC antibody and the specific antibody of target protein (all antibodies from Abcam). After overnight incubation at 4℃, the endogenous DNA protein complex was precipitated by protein agarose/Sepharose. After centrifugation, the supernatant was removed and the non-specific complex was washed and decrosslinked at 65℃ overnight. Then, DNA fragments were extracted and purified with phenol/chloroform. The aggregation level of HNF4A and ARID1A was detected by RT-qPCR by using IgG as internal reference.

| Establishment of mouse model of endometriosis
A total of 24 NOD/SCID mice aged 5-6 weeks (8 mice were used for uterus donation modelling) were purchased from Model Animal Research Center of Nanjing University and injected subcutaneously with a micro-injector. The mice were assigned into two groups including sh-NC group and sh-HADC2 group (8 mice in each group). Both the donor and recipient mice were injected with 100 mg/kg oestrogen (Santa Cruz Biotechnology Inc.) once a week 14 days before the establishment of the model. One donor mouse uterus was used by two recipient mice. After the mice were euthanized, the uterus was found, and the mesometrium was separated and put into the sterile plate containing normal saline. The tissues were cut into 1 × 1 mm pieces, the tissue fragments were injected into the peritoneal cavity of the recipient mice (100 μl/ mouse), and the culture was continued. After the model establishment was started, sh-NC vector and sh-HADC2 vector were injected intraperitoneally into mice from day 0 once every three days (100 μl/mice) for four weeks. Four weeks later, the mice were euthanized and mice were dissected to attain the ectopic uterine tissues. The weight and area of ectopic uterine tissues were measured and photographed.

| Terminal deoxyribonucleotidyl transferase (TdT)-mediated 2′-deoxyuridine 5′-triphosphatebiotin nick end-labelling (TUNEL) staining
After dewaxing, the paraffin-embedded sections were immersed in 0.85% NaCl for 5 min, in PBS for 5 min and in 4% paraformaldehyde for 15 min in turn. 100 μl of 20 μg/ml proteinase K was added to completely cover the tissues for 20-min incubation at room temperature. The sections were immersed in PBS for 5 min, in 4% paraformaldehyde for 5 min and in PBS for 5 min, which was repeated once.
The excess liquid was removed, and 100 μl equilibrium solution was added to sections for 5-min incubation at room temperature. The equilibrium solution was removed, and the rTdT reaction solution was added. The samples were transferred into a wet box and incubated at 37℃ for 1 h. After the plastic cover glass was removed, the samples were stained for 15 min in a dyeing tank containing 2 × sodium chloride-sodium citrate. After completion, the samples were put into 0.3% hydrogen peroxide water prepared by PBS for 2 min, coloured with DAB, and sealed with neutral resin. SPSS 21.0 (IBM Corp.) was adopted to analyse the data of this study.

| Statistical analysis
The measurement data were expressed as mean ± standard deviation. Unpaired t test was applied for comparison between the two groups, one-way analysis of variance (anova) for comparison among multiple groups and two-way anova for comparison between groups at different time, followed by Tukey's post hoc test. p < 0.05 meant that the difference was statistically significant.

| HDAC2 was highly expressed in endometriosis
In order to deeply understand the molecular mechanism of endometriosis, bioinformatics analysis was conducted to screen the differentially expressed genes in endometriosis. A total of 468 upregulated genes in endometriosis were obtained by differential analysis of GSE37837 ( Figure 1A). The top 1000 genes related to 'endometriosis' were obtained from GeneCards and then were intersected with the upregulated genes in GSE37837, which generated 40 important genes of endometriosis ( Figure 1B). The results of phenolyzer analysis showed that ITGB1, HDAC2 and LAMB1 were the top three important genes in endometriosis ( Figure 1C).
In order to explore the effect of HDAC2 on endometriosis, HDAC2 expression data were extracted from GSE37837 to draw the box plot, which revealed HDAC2 high expression in endometriosis ( Figure 1D). GEPIA also displayed that HDAC2 was highly expressed in UCEC ( Figure 1E). Western blot analysis and immunohistochemistry described that the protein expression of HDAC2 in endometriosis tissues was significantly higher than that in normal tissues ( Figure 1F-G). These results indicated that HDAC2 was upregulated in endometriosis.

| Silencing HDAC2 inhibited proliferation and migration and promoted apoptosis of endometriosis cells
In order to explore the effect of HDAC2 on endometriosis cells,  The data were expressed as median ± interquartile range. *p < 0.05. (F) The expression of HDAC2 protein in normal tissues (n = 30) and endometriosis tissues (n = 40) detected by immunohistochemistry. The data were expressed as median ± interquartile range. (G) Western blot analysis of the expression of HDAC2 protein in normal tissues (n = 30) and endometriosis tissues (n = 40). The data were expressed as median ± interquartile range. *p < 0.05 vs. the control group. The experiment was repeated three times independently

Expression of HDAC2 in GSE37837
Expression P= 0.0372

| HDAC2 inhibited HNF4A expression by deacetylation
It has been shown that HDAC2 can affect the expression of HNF4A through deacetylation. 8 In order to explore the effect of HDAC2 on the expression of HNF4A during endometriosis, we first found a significant negative correlation between HDAC2 and HNF4A expression in endometriosis by GSE37837 ( Figure 3A). Moreover, Western blot analysis and immunohistochemistry depicted that the expression of HNF4A protein in endometriosis tissues was significantly lower than that in normal tissues ( Figure  when HDAC2 was overexpressed ( Figure 3G). In summary, HDAC2 caused HNF4A downregulation through deacetylation.

| Silencing HDAC2 depressed the proliferation and invasion and accelerated the apoptosis of endometriosis cells by activating HNF4A
In order to explore the effect of the HDAC2/HNF4A axis on en-  genes were predicted through GeneCards. Then, these predicted results were intersected, which found that the most critical genes were ARID1A and F2 ( Figure 5A).

| Proliferation and invasion of endometriosis cells were inhibited, and cell apoptosis was promoted
A previous study has reported that ARID1A is poorly expressed in endometriosis. 12 The promoter sequence of ARID1A was obtained through UCSC to verify the relationship between HNF4A and ARID1A. Several binding sites of HNF4A to ARID1A promoter were predicted by hTFtarget (Table S2). The co-expression graph of HNF4A and ARID1A was obtained from MEM ( Figure 5B). The co-expression diagram exhibited that HNF4A was positively correlated with ARID1A. Correlation map drawn based on analysis in microarray dataset GSE37837 revealed a positive correlation between HNF4A and ARID1A expression in endometriosis tissues ( Figure 5C).
Next, Western blot analysis and immunohistochemistry showed that the protein expression of ARID1A in endometriosis tissues was strikingly lower than that in normal tissues ( Figure 5D,E). After sh-HNF4A group but was markedly elevated in the oe-HNF4A group compared with oe-NC group ( Figure 5G).

| Silencing HDAC2 inhibited endometriosis in mice
In order to explore the effect of HDAC2 on endometriosis in vivo, the mouse model of endometriosis was established. As shown in Figure S1, a representative mouse showed 4 stitched fragments of tissues, resulting in 3 lesion sites filled with liquid on the peritoneum wall. Then, HDAC2 was silenced. The results showed that compared with the sh-NC group, endometriosis mice in the sh-HDAC2 group had distinct reduction of the area and weight of endometriosis tissues ( Figure 6A,B). Based on immunohistochemistry results, in contrast to the sh-NC group, HNF4A and ARID1A protein expression in the sh-HDAC2 group was substantially augmented whereas HDAC2 expression was diminished ( Figure 6C).
Immunohistochemistry and TUNEL staining manifested that compared with the sh-NC group, Ki-67 protein expression in the sh-HDAC2 group was conspicuously declined in endometriosis tissues, and the apoptosis rate was evidently elevated ( Figure 6D). To sum up, silencing HDAC2 inhibited endometriosis by influencing the HNF4A/ARID1A axis in vivo.

| DISCUSS ION
As a frequent and chronic illness in young women, patients with endometriosis have painful symptoms, and the advanced endometriosis may result in gynaecological malignancies like ovarian cancer and other complications, including infertility. 13 Furthermore, the gold standard for endometriosis treatment is surgery, but surgery contributes to a great morbidity and cost burden for patients. 14 Based on this, there is an ongoing need for understanding molecular mechanism underlying endometriosis to provide a novel target for endometriosis treatment. Thus, our study was designed to explore the role of HDAC2 in endometriosis and the related specific mechanism.
Consequently, our data unravelled that silencing HDAC2 increased HNF4A expression through inhibition of deacetylation to upregulate ARID1A, thus preventing endometriosis.
The microarray-based analysis in our study indicated that upregulated HDAC2 was a key gene for endometriosis, and tissue experiments confirmed the upregulation of HDAC2 in endometriosis tissues. In consistent with this finding, existing research has displayed that HDAC2 expression was higher in endometriotic cells than in endometrial stromal cells. 7 Notably, the data in our research also exhibited that HDAC2 silencing repressed cell proliferation and Furthermore, cell experiments in this research displayed elevation of proliferation and invasion but decrease of apoptosis of hEM15A cells following ARID1A silencing. In line with our finding, a prior study discovered that ARID1A overexpression diminished ovarian granulosa cell proliferation but augmented cell apoptosis in polycystic ovary syndrome. 22 Another existing study also revealed that silencing of ARID1A by small interfering RNA induced proliferation, migration and invasion in CNE1 and HNE1 cells. 23

| CON CLUS ION
Collectively, the current findings elicited that the silencing of HDAC2 might reduce the proliferation and invasion but enhance apoptosis of endometriosis cells to prevent endometriosis by activating the HNF4A/ARID1A axis (Figure 7), which might present a promising therapeutic strategy for the treatment of endometriosis. However, considering the limitations of our study, more extensive research is required to investigate the specific mechanism of HDAC2, with eventual translation to clinical trials.

CO N FLI C T S O F I NTE R E S T
The authors declare no conflicts of interest.

DATA AVA I L A B I L I T Y S TAT E M E N T
The datasets generated and/or analysed during the current study are available from the corresponding author on reasonable request.