ER-α36 Promotes the Malignant Progression of Cervical Cancer Mediated by Estrogen via HMGA2

Objectives Estrogen is proven to promote the malignant behaviors of many cancers via its receptors. Estrogen receptor alfa 36 (ER-α36) is a newly identified isoform of estrogen receptor alfa (ER-α), the role of ER-α36 in regulating the effects of estrogen and its potential impact on human cervical cancer is poorly understood. Methods Immunohistochemistry staining was used to evaluate the expression of ER-α36, estrogen receptor alfa 66 (ER-α66) and their prognostic values in cervical cancer. The effects of ER-α36 and ER-α66 on the proliferation and metastasis of cervical cancer were measured in vitro. A xenograft tumor assay was used to study the tumorigenesis role of ER-α36 in vivo. Furthermore, the functional gene at the downstream of ER-α36 was obtained via next-generation sequencing, and the biological functions of high mobility group A2 (HMGA2) in cervical cancer cells were investigated in vitro. Results ER-α36 was over-expressed in cervical cancer tissues and elevated ER-α36 expression was associated with poor prognosis in cervical cancer patients. High expression of ER-α36 promoted the proliferation, invasion and metastasis of cervical cancer cells mediated by estrogen, while silencing ER-α36 had the opposite effects. Further research showed that HMGA2 was a downstream target of ER-α36 in cervical cancer cells. The oncogenic effect of ER-α36 was attenuated after HMGA2 knockdown. Conclusions High expression of ER-α36 was correlated with a poor prognosis in cervical cancer by regulating HMGA2. ER-α36 could be a prognostic biomarker and a target for cervical cancer treatment.


INTRODUCTION
Cervical cancer (CC) is the fourth leading cause of cancer death in women. According to the latest statistics, there were 604127 incident cases and 341831 deaths due to CC worldwide in 2020 (1). Squamous cell Carcinoma (SCC) and adenocarcinoma (AC) are the most common histological subtypes accounting for about 70% and 25% of all cervical cancers, respectively (2). At present, radical surgery is the firstline treatment for early-stage CC, the combination of girradiation and cisplatin-based chemotherapy is the standard treatment for advanced CC (3). Although great progress has been made in disease prevention with the emergence of HPV vaccination and early screening, the incidence and mortality of CC are still high in low and middle-income countries (4). Therefore, it is important to elucidate the potential oncogenic molecular mechanisms in CC and develop new therapy methods.
Human papillomavirus (HPV) is a necessary but not sufficient cause of CC (5). In human cervical squamous epithelium, most lesions containing high-risk HPVs do not progress to CC, implicating other factors may be involved in the genesis of CC. Long-term use of oral contraceptives and high parity increase the risk of CC in woman with HPV infection (6,7). Exposure to diethylstilbestrol is associated with increased risk of cervical high-grade squamous neoplasia (8). Chronic estrogen treatment induced the genesis of cervical squamous cancer in HPV type 16 transgenic mice (9). These studies suggest that estrogen is an etiological cofactor for CC.
Estrogen influences physiological and pathological processes in various tissues through its receptors. ER-a is the major estrogen receptor expressed in the cervix. Researchers found that ER-a plays a critical role in cervical carcinogenesis in transgenic mice (10). ER-a is the product of ESR1 gene. The ESR1 gene not only serves as a template for a full-length 66 kDa protein(ER-a66), but also for two alternative isoforms(ER-a46 and ER-a36) (11). ER-a36 lacks transcriptional activation domains (AF-1 and AF-2), but retains dimerization, DNAbinding and partial ligand-binding domains. ER-a36 is mainly located in plasma membrane and cytoplasm, and mediates rapid estrogen signaling pathways (11)(12)(13)(14)(15). Previous studies have indicated that high expression of ER-a36 is associated with a more aggressive phenotype in various cancers, including breast cancer (16)(17)(18), endometrial cancer (14), gastric cancer (19), lung adenocarcinoma (20) and laryngeal cancer (21). However, the role of ER-a36 in regulating the effect of estrogen and its potential impact on human CC remains unknown.
In this study, we first verified that ER-a36 was remarkably over-expressed in CC tissues and was associated with poor prognosis of CC. Through in vitro and vivo experiments, we demonstrated that ER-a36 promoted the malignant progression of CC mediated by estrogen, and these effects on biological behavior were achieved by regulating the expression of HMGA2. To reveal the correlations between ER-a36 and HMGA2, we further researched the expression, biological functions and clinical significance of HMGA2 in CC.

Human CC Specimens
A total of 117 cases of paraffin-embedded primary CC samples, 30 cases of paraffin-embedded cervical intraepithelial neoplasia (CIN) and 60 cases of paraffin-embedded normal cervix samples were collected from the Department of Pathology, Qilu Hospital, from January 2012 to December 2014. In addition, 12 cases of CC tissues were collected from patients with CC who underwent primary surgery in Qilu Hospital, while the 8 normal cervix tissues were from patients who received hysterectomy due to benign gynecologic tumors. All the paraffin-embedded samples possessed intact follow-up information. Written informed consents were obtained from all the patients. Ethical approval was issued by the Ethics Committee of Shandong University Qilu Hospital and the approval number is KYLL-2018-174.

Immunohistochemistry (IHC)
The 4um-thick paraffin-embedded tissue sections were dewaxed with xylene and rehydrated in an ethanol gradient. After antigen retrieval, these sections were incubated with a primary antibody at 4°C overnight, then were incubated with secondary antibodies for 30 minutes at room temperature. Subsequently, these slides were stained with 3,3'-diaminobenzidine detection system and haematoxylin.

Quantitative Reverse-Transcription Polymerase Chain Reaction (qRT-PCR)
Total RNA was extracted by TRIzol reagent (15596018; Invitrogen). Complementary DNA (cDNA) was synthesized by the PrimeScript RT reagent Kit (FSQ-301; Toyobo Biotech Co Ltd, OSAKA, Japan). Real-time PCR was performed using the SYBR Green Real-time PCR Master Mix (QPK-201; Toyobo Biotech Co Ltd, OSAKA, Japan). The message RNA(mRNA) level of specific genes was normalized against glyceraldehyde 3phosphate dehydrogenase (GAPDH) using the comparative Ct method (2 −DDCt ). The primers used are shown in the Supplementary Table S1. were purchased from Genechem (Shanghai, China). Caski and hela cells were infected with lentivirus (MOI: 20) and selected with 1.5 mg/ml puromycin for about 1 week. The efficiency of knockdown or overexpression was detected by qRT-PCR and western blotting. The siRNA sequences are shown in the supplementary material.

Western Blotting
Cells and tissues were lysed in RIPA (Beyotime, Shanghai, China) supplemented with PMSF according to the manufacturer's recommendations. The protein concentration was calculated by the BCA Kit (Beyotime, Shanghai, China). Proteins were separated by SDS-PAGE electrophoresis, then were transferred to PVDF membrane. After blocking in the 5% milk for 1 hour at room temperature, these membranes were incubated with primary antibodies overnight at 4°C. Subsequently, these membranes were labeled with the corresponding HRPconjugated secondary antibodies (Cell Signaling Technology, Danvers) for 1 hour at room temperature. Protein bands were detected by enhanced chemiluminescence detection kit (ECL ORT2655, PerkinElmer, Waltham, MA, USA). ImageJ 1.47 was used to analyze the relative protein level. GAPDH was used as an endogenous control.

Cell Viability Assay
Cells (10 3 per well) were seeded into 96-well plates and cultured for 1-6 days, Cell Counting Kit-8 (CCK8) (APExBIO, #K1018) was used to detect cell viability following the manufacturer's protocol. The absorbance at 450 nm was detected by a microtiter plate reader (Thermo Scientific).

Colony Formation Assay
Cells (10 3 per well) were seeded into 6-well plates and cultured for 10-14 days. The colonies were fixed by methanol and stained by 0.5% crystal violet. Colonies more than 50 cells were counted.

Cell Migration and Invasion Assay
Migration and invasion assays were performed in transwell chambers (8-mm pores, BD Biosciences, USA) inserted in 24well plates without or with Matrigel (356234, Corning Incorporate, NY). Cells (10 5 per well) were seeded into the upper chambers in serum-free medium, medium supplemented with 20% FBS were added to the lower chambers. After cultured for an appropriate time in an incubator, cells that have migrated through the membrane were fixed with methanol, stained with 0.5% crystal violet and observed under a light microscope.

Cell Cycle Assessment
Cells were collected and stained with PI/RNase Staining Buffer (550825, BD Bioscience, Franklin Lakes, NJ, USA) following the manufacturer's protocol. Flow cytometry was used to evaluate the distribution of cell cycle.

High-Throughput Differential Gene Expression Analysis
The high-throughput RNA-seq experiments were conducted by Annoroad (Beijing, China). In brief, caski cells were transfected with si-ER-a36 or NC (n = 3) for 48 h and treated with 1nM E 2 for 24h. The RNA preparation and library preparation for transcriptome sequencing were performed according to the manufacturer's instructions. Fragments per kilobase of transcript per million fragments mapped (FPKM) was used to estimate gene expression levels. The differentially expressed mRNAs were selected with fold change > 2 or fold change < 0.5 and p value < 0.05 by R package edgeR or DESeq2.

Animal Experiment
The animal study was reviewed and approved by the Ethics Committee of Shandong University Cheeloo College of Medicine, the approval number is 21002. Female BALB/c nude mice (4-6 weeks old; NBRI of Nanjing University, Nanjing, China) were implanted subcutaneously with 0.36 mg of 60-day release 17b-estradiol pellets (Innovative Research, TX). Caski cells transfected with shNC, shER-a36, NC-LV and NC-ER-a36 were collected, 5 × 10 6 cells were subcutaneously injected into the axilla of each mouse. The tumor size was measured by Vernier calliper once every 2 days, and tumor volumes were calculated using the equation: length × width 2 × 0.5. Three weeks postinjection, these mice were sacrificed and tumors were removed, photographed and weighed.

Statistical Analysis
Each assay was repeated at least 3 times independently. The data was expressed as the means ± SEM. Student's t test was applied to compare two independent groups and One-way ANOVA was used to compare multiple groups. The correlation among ER-a36, ER-a66 and clinicopathological characteristics were assessed by chi-square (c2) or Fisher test. Overall survival analysis was performed by Kaplan-Meier method with logrank test. Survival data was evaluated by univariate and multivariate Cox regression analyses to evaluate the independent factors of patients' outcomes. All the analyses above were performed by SPSS v22.0 (SPSS, Inc., Chicago, IL, USA). Images were processed by GraphPad Prism 8.00 (GraphPad Software, La Jolla, CA, USA) and Adobe Photoshop CC 2019 (Adobe, San Jose, CA, USA). Differences were considered statistically significant when P < 0.05(*p < 0.05, **p < 0.01, ***p < 0.001).

Expression and Prognostic Significance of ER-a36 and ER-a66 in CC
We detected the expressions of ER-a36 and ER-a66 in CC tissues and cervix tissues with qRT-PCR and western blotting, the results showed that the mRNA and protein levels of ER-a36 were higher in cancer tissues, while the mRNA and protein levels of ER-a66 were higher in cervix tissues ( Figures 1A-C).Then we detected their expressions in H8,siha,hela,caski and C33a cell lines, western blotting demonstrated that ER-a36 was significantly over-expressed in four CC cell lines (siha,hela, caski and C33a) compared with H8, the normal cervical epithelial cells, while ER-a66 was over-expressed in H8 cells ( Figure 1D).These results indicated the possible roles of ER-a36 and ER-a66 in CC.
For further investigation, we carried out IHC staining to analyze the expression of ER-a36 and ER-a66 in 60 cases of normal cervix tissues and 117 cases of CC tissues including 99 cases of cervical SCC and 18 cases of cervical AC. Positive ER-a36 staining was mainly detected in cellular membrane and cytoplasm in CC tissues, while positive ER-a66 staining was mainly detected in the nucleus. These patients were divided into high expression groups and low expression groups based on the IHC scores ( Figures 1G-I). The expression of ER-a36 was significantly higher in cervical SCC (48.5%, 48/99 cases) and AC (55.6%, 10/18 cases) tissues than in CIN tissues (30%, 9/30 cases) and cervix tissues (13.3%, 8/60 cases) ( Figure 1E). However, the expression of ER-a66 was lower in cervical SCC (10%, 9/90 cases) and AC (11.1%, 2/18 cases) tissues than in cervix tissues (35%, 21/60 cases) ( Figure 1F). Kaplan-Meier survival curves showed that high ER-a36 expression was remarkably associated with unfavorable prognosis in cervical SCC and AC patients ( Figure 1J), while the expression of ER-a66 displayed no prognostic significance in CC patients ( Figure 1K). Collectively, ER-a36 was up-regulated and correlated with the poor prognosis in CC.

Clinical Significance of ER-a36 and ER-a66 in CC
To assess the clinical significance of ER-a36 and ER-a66, we analyzed the correlation among ER-a36, ER-a66 and the clinicopathological features of cervical SCC and AC patients. In patients with cervical SCC, high expression of ER-a36 was significantly associated with advanced International Federation of Gynecology and Obstetrics (FIGO) stage (p=0.026), deeper stromal invasion (p=0.032), positive lymph node metastasis (p=0.005) and high expression of Ki67 (p<0.001), whereas there is no correlation between ER-a66 and these features ( Table 1).In cervical AC patients, elevated ER-a36 was associated with positive lymph node metastasis (p=0.036), however no correlation was observed between ER-a66 and this feature ( Table 2).
In addition, we performed univariate and multivariate analyses to evaluate the prognostic values of ER-a36, ER-a66 and other clinicopathological factors. In patients with cervical SCC, univariate Cox regression analysis demonstrated advanced FIGO stage (p=0.042), positive lymph node metastasis (p=0.003), and high ER-a36 expression (p=0.005) were all significantly associated with a low overall survival rate. Multivariate Cox regression analysis confirmed that high expression of ER-a36 could be an independent factor to predict poor survival of cervical SCC (p=0.028) together with positive lymph node metastasis (p=0.025, Table 3). In cervical AC patients, univariate Cox regression analyses suggested that FIGO stage (p=0.023), lymph node metastasis (p=0.038) and ER-a36 expression (p =0.042) had strong correlations with prognosis ( Table 4).

ER-a36 Silence Inhibits Proliferation and Metastasis of CC Cells Mediated by E 2 In Vitro
Caski and hela cells were stably transfected with shER-a36 and negative control (shNC) (Figures 2A, B). Cells were divided into 4 groups; the shNC groups (cells transfected with shNC and treated with ethanol), the shNC+E 2 groups (cells transfected with shNC and treated with 1nM E 2 ), the shER-a36 groups (cells transfected with shER-a36 and treated with ethanol) and shER-a36+E 2 groups (cells transfected with shER-a36 and treated with 1nM E 2 ). The CCK8 assay showed that cells transfected with ShER-a36 (shER-a36 groups vs. ShER-a36+E 2 groups) exhibited lower sensitivity to E 2 stimulation than the control groups (shNC groups vs. shNC+E 2 groups). In the shNC+E 2 groups, E 2 promoted the proliferation of CC cells compared to shNC groups. However, there is no significant difference in proliferation between shER-a36 groups and shER-a36+E 2 groups. In the ShER-a36+E 2 groups, the proliferation induced by E 2 was suppressed in comparison to the shNC+E 2 group. Nevertheless, there was no remarkable difference in cell viability between shNC and shER-a36 groups ( Figures 2C, D). The clonogenic assay also confirmed these results ( Figure 2E). In addition, we transfected si-ER-a66 and negative control (si-NC) into caski and hela cells (Supplementary Figures 1A, B), and there was no significant difference in the capability of proliferation between si-NC+E 2 groups (cells transfected with si-NC and treated with 1nM E 2 ) and si-ER-a66+E 2 groups (cells transfected with si-ER-a66 and treated with 1nM E 2 ) (Supplementaryl Figures 1C-E).
Considering that the expression of ER-a36 was associated with lymph node metastasis in CC patients, we hypothesized ER-a36 participated in the metastasis of CC cells. The transwell assay exhibited that in the shNC+E 2 groups, E 2 enhanced the abilities of migration and invasion of CC cells compared with the shNC groups, but no obvious difference was found between shER-a36+E 2 groups and shER-a36 groups. Comparison showed that ER-a36 knockdown inhibited the ability of metastasis of CC cells mediated by E 2 in the shNC+E 2 and shER-a36+E 2 groups, while there is no statistical difference between shNC and shER-a36 groups ( Figure 2F). There was no remarkable difference between si-NC+E 2 groups and si-ER-a66 +E 2 groups (Supplementary Figure 1F). In conclusion, ER-a36 silence inhibited proliferation and metastasis of CC cells induced by E 2 in vitro.

Overexpression of ER-a36 Promotes E 2 -Mediated Proliferation and Metastasis of CC Cells In Vitro
Caski and hela cells were successfully transfected with ER-a36-LV and NC-LV ( Figures 3A, B). Cells were divided into 4 groups; the NC-LV groups (cells transfected with NC-LV and treated with ethanol), the NC-LV+E 2 groups (cells transfected with NC-LV and treated with 1nM E 2 ), the ER-a36-LV groups (cells transfected with ER-a36-LV and treated with ethanol) and ER-a36-LV+E 2 groups (cells transfected with ER-a36-LV and treated with 1nM E 2 ). The CCK8 assay showed that cells transfected with ER-a36-LV (ER-a36-LV groups vs. ER-a36-LV+E 2 groups) displayed higher sensitivity to E 2 stimulation than the control groups (NC-LV groups vs. NC-LV+E 2 groups).
In the ER-a36-LV+E 2 groups, the proliferation induced by E 2 was enhanced in comparison to the NC-LV+E 2 groups. Nevertheless, there was no remarkable difference in cell viability between NC-LV and ER-a36-LV groups ( Figures 3C, D). The clonogenic assay obtained the similar results ( Figure 3E). Transwell assay showed that E 2 enhanced the abilities of migration and invasion of CC cells in both groups (ER-a36-LV +E 2 vs.ER-a36-LV group and NC-LV+E 2 vs. NC-LV group). Comparison in the NC-LV+E 2 and ER-a36-LV+E 2 groups showed that overexpression of ER-a36 promoted the ability of metastasis of CC cells induced by E 2 , while there was no statistical difference between NC-LV and ER-a36-LV groups ( Figure 3F). In a word, overexpression of ER-a36 promoted E 2mediated proliferation and metastasis of CC cells.

ER-a36 Is Involved in E 2-Stimulated Cell Cycle Progression
Caski and hela cells transfected with shNC, shER-a36, NC-LV, ER-a36-LV, si-NC and si-ER-a66 were incubated with ethanol or 1nM E 2 for 24 hours. Cell cycle analysis revealed that E 2 decreased the percentage of cells in G1 phase and increased the percentage of cells in S phase. Comparing shNC+E 2 groups and shER-a36+E 2 groups, we found that ER-a36 knockdown increased the percentage of cells in G1 phase and decreased the percentage of cells in S phase, while in ER-a36-LV+E 2 groups, up-regulated ER-a36 decreased the percentage of cells in G1 phase and increased the percentage of cells in S phase in comparison to the NC-LV+E 2 groups (Figures 4A, B). There was no remarkable difference between si-NC+E 2 and si-ER-a66 +E 2 groups in cell cycles ( Supplementary Figures 2A, B). Therefore, these results indicated that ERa36 was involved in the cell cycle progression of CC cells induced by E 2 .

ER-a36 Enhances the Proliferation of CC Cells Mediated by E 2 In Vivo
To further explore the role of ER-a36 in the oncogenic function induced by E 2 in vivo, caski cells transfected with shNC, shER-a36, NC-LV, NC-ER-a36 were subcutaneously injected into the armpits of female nude mice which were subcutaneously inoculated with 0.36 mg 60-day released E 2 pellets. Consistent with the results in vitro, the volumes of the tumors in the shER-a36 group were remarkably decreased compared with those in the shNC groups, and the tumors in the ER-a36-LV groups were boosted as compared to the control group ( Figures 5A-D). IHC staining was performed to tumor mass, the results showed tumors in shER-a36 groups exhibited significantly lower ER-a36 expression and Ki67 proliferation index vs. the control group, whereas tumors in ER-a36-LV groups exhibited up-regulated ER-a36 expression and high staining intensity of Ki67 ( Figures 5E, F). These results indicated ER-a36 enhanced the proliferation of CC cells mediated by E 2 in vivo.

HMGA2 Is a Downstream Target of ER-a36 in CC, Elevated HMGA2 Expression Is Correlated With the Poor Prognosis of CC
To illuminate the mechanisms by which ER-a36 promotes CC's malignant progression induced by E 2 , next-generation sequencing (NGS) was conducted. A total of 72 differentially expressed genes (DEGs, fold change ≥2, p < 0.05) were identified, including 17 upregulated genes and 55 downregulated genes ( Figure 6A). The top 24 downregulated genes were chosen to validate the DEGs by qRT-PCR assay ( Figure 6B). HMGA2 was shown to be particularly downregulated and has been confirmed to be associated with tumorigenesis. Therefore, we hypothesized that ER-a36 might promote the malignancy of CC by regulating HMGA2 expression. Western blotting assay showed that the protein level of HMGA2 in CC cells was decreased after ER-a36 knockdown and increased after ER-a36 overexpression ( Figure 6C). Moreover, qRT-PCR ( Figure 6D) and IHC ( Figures 6E, F) assays revealed HMGA 2 expression was positively correlated with ER-a36 expression in CC tissues. These results suggested that HMGA2 was a downstream effector of ER-a36 in CC.
To explore the role of HMGA2 in CC, we detected the expression of HMGA2 in CC tissues and cervix tissues with qRT-PCR and western blotting, the results showed that the mRNA and protein levels of HMGA2 were remarkably higher in cancer tissues ( Figures 6G, H). Then we examined HMGA2 expression in H8, siha, hela, caski and C33a cell lines, western blotting revealed HMGA2 was significantly over-expressed in four CC cell lines (siha, hela, caski and C33a) compared with H8 cells ( Figure 6I). To evaluate the expression pattern and clinical significance of HMGA2 in CC, IHC staining assay was applied.
The results verified the expression of HMGA2 was elevated in CC tissues ( Figures 6J, K). In cervical SCC patients, clinicopathological characteristic analysis showed HMGA2 expression was correlated with FIGO stage (p=0.012), DSI (p=0.03), lymph node metastasis (p=0.005) and ER-a36 expression (p<0.001, Table 5). In cervical AC patients, high expression of HMGA2 was associated with positive lymph node metastasis (p=0.036) and elevated ER-a36 expression (p=0.00, Table 6). Kaplan-Meier survival curves showed high HMGA2   expression was associated with low overall survival rate in both cervical SCC and cervical AC patients ( Figure 6L). Furthermore, the expressions of ER-a36 and HMGA2 were combined to evaluate the prognostic value of co-expression of ER-a36 and HMGA2. Kaplan-Meier survival curve revealed patients with high expression of ER-a36 and HMGA2 seemed to have worse prognosis than others ( Figure 6M), indicating that co-expression of ER-a36 and HMGA2 may be a more sensitive factor in CC. In conclusion, HMGA2 was highly expressed in CC tissues and predicted unfavorable prognosis.

HMGA2 Knockdown Impairs Proliferation and Metastasis of CC Cells
To assess the biological function of HMGA2 in CC cells, si-HMGA2 and its negative control(si-NC) were transfected into caski and hela cells ( Figures 7A, B). CCK8 and colony formation assays showed HMGA2 silence inhibited the proliferation of CC cells ( Figures 7C-E). Transwell assay revealed that HMGA2 knockdown decreased the migration and invasion ability of CC cells ( Figure 7F). Taken together, these data implied the oncogenic effects of HMGA2 in CC.

ER-a36 Promotes E 2 -Induced Malignancy of CC by Regulating HMGA2
To verify whether E 2 and ER-a36 promote aggressive behaviors of CC through HMGA2, a rescue experiment was performed. We introduced si-HMGA2 and si-NC into caski and hela cells that were previously transfected with NC-LV or ER-a36-LV ( Figure 8A). Cells were treated with 1nM E 2 . The results showed that knockdown of HMGA2 obviously attenuated the proliferation and metastasis induced by ER-a36 overexpression ( Figures 8B-E). These results suggested that HMGA2 was involved in E 2 and ER-a36 mediated oncogenic behaviors of CC cells.

DISCUSSION
Development of CC has generally been considered to be unrelated to estrogen, however, a growing number of research has found that estrogen and its receptor ER-a were not only involved but also played important roles in the genesis and development of cervical carcinoma (9,10,(23)(24)(25). ER-a has three isoforms: ER-a36, ER-a46 and ER-a66. ER-a36 is a truncated variant of ER-a66, with a unique 27 amino acid domain at the C-terminal, which may endow it different characteristics (11). In terms of expression, ER-a66 could suppress ER-a36 expression by inhibiting ER-a36 promoter activity through the AF-1 domain, on the contrary, ER-a36 could also suppress ER-a66 expression by negatively regulating the transcription of ER-a66 (26). Previous studies have found ER-a36 was upregulated in various cancer tissues, such as lung (20), gastric (27), primary hepatocellular (28), thyroid (29) and breast (30) cancer, and correlated with poor prognosis. Consistent with these researches, we found ER-a36 was overexpressed in CC cell lines and tissues, and predicted unfavorable prognosis, while ER-a66 was low-expressed in these cell lines and tissues, and had no prognostic significance in CC. Sun et al. (31) reported that ER-a36 was mainly expressed in the plasma membrane and cytoplasm of caski and hela cells. In our study, we got the similar results. The IHC assay showed that ER-a36 was mainly located in the cellular membrane and cytoplasm of cervical cancer tissues, only a small part of it was located in the nucleus. In addition, according to analyze the correlation between clinical characteristics and ER-a36 expression, we discovered in cervical SCC, elevated ER-a36 was associated with advanced FIGO stage, deeper stromal invasion, positive lymph node metastasis and high expression of Ki67 (the proliferation index), in AC, high expression of ER-a36 was associated with positive lymph node metastasis. These results indicated the oncogenic role of ER-a36 in CC. Tong et al. (14) reported that ER-a36 could rapidly activate the PKCd/ERK pathway in response to E 2 , leading to an increase of cyclin D1/cyclin-dependent kinase 4, resulting in the promotion of cell cycle progression and proliferation in endometrial cancer. Chaudhri et al. (17) demonstrated ER-a36 activated MAPK/ERK and PI3K/AKT paths under the stimulation of E 2 , contributing to metastasis of breast cancer. Nofrat Schwartz et al. (21) found that ER-a36 could promote the aggression of laryngeal cancer through PKC pathways induced by E 2 . These studies indicated E 2 and ER-a36 were involved in the progression of certain malignant tumors. Therefore, we hypothesized that E 2 and ER-a36 may be related to the development of CC.
In our study, we found E 2 (1nM, the level equivalent to that in premenopausal women (32) promoted the cell cycle progression, and enhanced the proliferation and metastasis of CC cells. Furthermore, we verified it was ER-a36 not ER-a66 that promoted the malignant behaviors of CC cells induced by E 2 .
In order to illuminate the mechanism by which ER-a36 promotes CC's malignant progression induced by E 2 , we conducted NGS and found the expression of HMGA2 was particularly down-regulated in the si-ER-a36 groups. Then we detected the protein level of HMGA2 in caski and hela cells transfected with shER-a36 or ER-a36-LV and their negative controls. The results showed the protein level of HMGA2 was decreased after ER-a36 knockdown and increased after ER-a36 overexpression. In addition, qRT-PCR and IHC staining assays confirmed a positive correlation between the expression of ER-a36 and HMGA2 in CC. HMGA2 is a non-histone nuclearbinding oncofetal protein, which modulates transcription through promoting conformational changes (33). Previous research found HMGA2 was over expressed in embryonic tissue and in various malignant tumors such as colorectal (34), breast (35), pulmonary (36) and ovary cancer, but was rare detected in normal adult tissues (37). In our study, we found the expression of HMGA2 was higher in CC tissues than in normal cervical samples.
Moreover, in cervical SCC, high expressed HMGA2 was correlated with advanced FIGO stage, positive lymph node metastasis, high expression of ER-a36 and poor prognosis. In cervical AC, elevated HMGA2 was associated with positive lymph node metastasis, high ER-a36 expression and predicted poor prognosis. These results indicated that HMGA2 might play an oncogenic role in tumorigenesis and progression of CC. Wei et al. (38) reported HMGA2 promoted the proliferation, migration and invasion of endometrial cancer. Here, we demonstrated HMGA2 silencing suppressed the malignancy behaviors of CC. In addition, knockdown of HMGA2 in ER-a36-overexpressing cells reversed the effect of ER-a36. Some limitations of our study must be acknowledged. Firstly, the research sample was relatively small. In the future, large-scale cohort samples are necessary to further evaluate the values of ER-a36 in the diagnosis and prognosis of CC. Secondly, the molecular mechanism of the interaction between ER-a36 and HMGA2 has not been elucidated. Although ER-a36 is a membrane receptor, in our study, the expression of ER-a36 was observed in the nucleus of some cervical cancer cells. Wang et al. (30) reported that in ER-a36+ breast cancer cells, tamoxifen or estrogen could induce the nuclear translocation of ER-a36 to regulate the transcriptional activity of ER-a to increase ALDH1A1 expression. Similarly, we speculate that in cervical cancer cells, estrogen promotes nuclear translocation of ER-a36, thereby directly regulating the transcription of HMGA2. In the future, co-immunoprecipitation, dual-luciferase reporter assay and other experiments are needed to reveal the interaction between ER-a36 and HMGA2. In summary, our study found ER-a36 was highly expressed in CC samples and CC cell lines. Elevated expression of ER-a36 was associated with a poor prognosis in CC. Overexpression of ER-a36 promotes E 2 -mediated proliferation and metastasis. Depletion of ER-a36 caused G0/G1 arrest, decreased the sensitivity of CC cells to E 2 . We further demonstrated that ER-a36 promoted E 2 -induced malignancy of CC by regulating HMGA2 expression. In conclusion, ER-a36 might be a novel therapeutic target for the treatment of CC.

DATA AVAILABILITY STATEMENT
The datasets presented in this study can be found in online repositories. The names of the repository/repositories and accession number(s) can be found below: https://www.ncbi. nlm.nih.gov/, GSE173120.

ETHICS STATEMENT
The studies involving human participants were reviewed and approved by Ethics Committee of Shandong University QIlu Hospital. The patients/participants provided their written informed consent to participate in this study. The animal study was reviewed and approved by Ethics Committee of Shandong University Cheeloo College of Medicine.

AUTHOR CONTRIBUTIONS
XY, QS, and CW designed this study. TZ, SZ, and KW collected samples and information of CC patients. CW performed the experiments, analyzed the data and wrote the manuscript. XY revised the manuscript. All authors contributed to the article and approved the submitted version.

FUNDING
This work was financially supported by National Natural Science Foundation of China (no. 81874105).