Targeting INHBA in Ovarian Cancer Cells Suppresses Cancer Xenograft Growth by Attenuating Stromal Fibroblast Activation

INHBA-encoded inhibin β A is a member of the transforming growth factor-β (TGF-β) superfamily. INHBA has been reported to be implicated in the progression of multiple types of cancer including ovarian cancer (OC). However, the mechanisms by which INHBA affects OC progression are not well-characterized. The aim of our study was to explore the prognostic value of INHBA for different stages and grades of OC and to identify the possible mechanisms by which INHBA promotes OC progression. Our results demonstrated that INHBA was specifically expressed in OC epithelium, and higher expression was associated with higher risk of mortality in patients with advanced and higher-grade serous OC (SOC). In addition, knockdown of INHBA in cancer cells impaired cancer xenograft growth through reducing OC stromal fibroblast activation in vivo. Further results confirmed that Smad2 signaling pathway was involved in INHBA-induced stromal fibroblast activation, and inhibiting this pathway could effectively reverse activation of stromal fibroblasts. In summary, our results showed that blocking INHBA in cancer cells may be a potential therapeutic strategy to inhibit SOC progression.


Introduction
Ovarian cancer (OC) is the seventh leading cancer diagnosis and eighth leading cause of cancer death among women [1]. OC is highly curable if found early and intervened actively, but OC at early stage usually lacks obvious clinical symptoms. Around 60-70% of women are diagnosed with latestage disease that has already spread within the abdomen [2,3]. Despite numerous targeted drugs have been developed to treat OC, patients' overall survival (OS) is still very dismal [4]. Therefore, it is urgent and significant to identify novel molecules involved in the OC progression and further develop some other effective treatments for OC patients.
INHBA-encoded inhibin β A is a member of the transforming growth factor-β (TGF-β) superfamily [5]. Inhibin β A could further form activin A by homodimerization or be linked to inhibin β B to produce inhibin by heterodimerization [6]. Activin A has been reported to be involved in a variety of biological processes, such as immune response, stem cell differentiation, and glucose metabolism [7]. Recent studies have shown that overexpression of INHBA occurs in multiple types of cancers, including colorectal cancer, breast cancer, lung cancer, esophageal squamous cell carcinoma, and bladder cancer [6,[8][9][10][11]. For example, activin A signaling promotes breast cancer metastasis by regulating IL13Rα2 expression [8]. In esophageal squamous cell carcinoma, high INHBA level predicts poorer prognosis [10]. Upregulation of INHBA expression promotes cell proliferation and predicts poor survival in patients with lung adenocarcinoma [6]. In OC, although some studies have reported that patients with high expression of INHBA have shorter survival times than those with low expression of INHBA [12], the relationship of INHBA with clinical features and its contribution to OC progression have not been fully elucidated.
Overwhelming studies have demonstrated the important contribution of stromal involvement to the pathogenesis of OC [13,14]. As the predominant cell type in the cancer stromal compartment, cancer-associated fibroblasts (CAFs) are reported to actively promote migration and invasion of tumor cells and impede drug delivery through generating extracellular matrix (ECM) components [15][16][17][18]. Currently, activated CAFs were considered to mainly originate from the surrounding fibroblasts under education of cancer cells [19,20]. The ability of tumor cells to direct normal fibroblasts to differentiate into CAFs is dependent on cytokines such as PDGF, TGF β1, and FGF2 [21]. Recent studies have reported that INHBA is also a driver of the CAF phenotype in OC [22,23]. However, our knowledge regarding activation effect of INHBA on fibroblasts is remained insufficient.
Therefore, the purposes of this study were to evaluate the expression of INHBA in OC tissues and to characterize the pivotal role of cancer cell-derived INHBA in stromal fibroblast activation and SOC progression. This study could help us better understand INHBA-mediated interaction between cancer cells and stromal fibroblasts, providing evidence to support that targeting INHBA in cancer cells to inactivate stromal fibroblasts could be a promising SOC therapeutic strategy.

Methods
2.1. Cell Culture. We purchased human ovarian cancer cell lines (SKOV3, CAROV3, OVCAR8, and OV90) and fibroblast cell line MRC-5 from ATCC (Rockville, MD, USA) and the cell bank of the Chinese Academy of Sciences, respectively. All the cell lines were confirmed to be mycoplasma-free by the source organizations prior to purchase. We isolated and purified primary normal ovarian fibroblasts (NOFs) from OC patients normal fresh ovary tissues following procedures as previously described [24]. Briefly, all the patient tissues were obtained under the supervision of the Ethics Committee of Tongji Hospital and confirmed by two senior pathologists. To collect tissue homogenate, fresh normal ovary tissues of 1 mm 3 were digested on a shaker in serum-free DMEM/F-12 containing collagenase and hyaluronidase (Sigma) for 2-3 hours. After the termination of digestion with FBS (Gibco), all the tissue samples were incubated with red blood cell lysate (BioLegend), filtered using a 40 μm cell strainer (BD Biosciences) to obtain single cells, and then subjected to antifibroblast microbeads (Miltenyi Biotec; 130-050-601) for fibroblast isolation. Finally, single cell samples were sorted using an MACS column (Miltenyi Biotec). The quality of the NOFs was confirmed using a PDGFRα antibody (ab203491, Abcam, USA). All OC cell lines were cultured in McCoy's 5A medium, and MRC-5 and the primary NOFs were maintained in DMEM/F-12 medium. All the cells were cultured in an incubator at 37°C, 5% CO 2 , and 80% humidity. All of aforementioned media were mixed with 1% penicillin/streptomycin (Thermo Scientific) and 10% FBS (Gibco).

Public Database Analysis.
We used Oncomine online tool (https://www.oncomine.org) to examine INHBA expression in microdissected ovarian profile GSE26712 and TCGA dataset. Student's t-test was used to calculate statistical significance. Gene expression data (GSE26193, GSE9891, GSE51088 profiling data) were obtained from Gene Expression Omnibus online website (https://www.ncbi.nlm.nih .gov/geo). The TCGA expression dataset and coexpression genes with INHBA were obtained via the cBioPortal tools (http://cbioportal.org). Screening criteria was based upon Spearman's correlation. The David analysis tool (https:// david.ncifcrf.gov) and the Kobas analysis tool (http://kobas .cbi.pku.edu.cn) were used to perform the GO analysis and the KEGG pathway analysis.
2.3. Single-Sample Gene Set Enrichment Analysis (ssGSEA). In order to investigate the relationship between INHBA expression and the 141-stroma signature activation degree in GSE9891, GSE51088, GSE26193, and TCGA dataset, ssGSEA was used to generate geneset activation score as described previously [25].
2.6. Quantitative Real-Time-(RT-) PCR. Total RNA of cells was isolated with TRIzol Reagents (Invitrogen) according to the standard protocol [26]. Reverse transcription of 2 μg RNA was performed using random primers and M-MLV reverse transcriptase (Takara, Japan). RT-PCR reactions were carried out using the Bio-Rad CFX96 system with IQ SYBR Green supermix (Bio-Rad, Hercules, CA). Relative expression levels of interesting genes were analyzed using the ΔΔ Cq method [27]. GAPDH served as the internal control. The primer sequences of INHBA are as follows: forward, 5 ′ -ACACAACAACTTTTGCTGCC-3 ′ , and reverse, 5 ′ -TCGTGTCACCACTGTCTTCTC-3 ′ .

Collagen Gel Contraction
Assay. Briefly, 6 × 10 5 cells were prepared in each group and washed with PBS. Then, the cells were resuspended in 200 μl collagen gel mixture per well in a 96-well ultra-low adhesion plate (Corning Life Sciences, Corning, NY). 200 μl of collagen gel mixture formula was as follows: 31.25 μl Type 1 Rat Tail Collagen (Thermo), 168.75 μl DMEM/F-12 medium, and 0.72 μl 1 N NaOH. After 12 h, the images were taken and contraction area was quantified with ImageJ software. All contraction assays were performed in triplicate.
2.11. Cellular Viability Assay. Cells were seeded in 24-well plates in triplicate with an initial density of 1 × 10 4 cells per well. At 24 h, 48 h, 72 h, and 96 h since planting, the cells were digested with trypsin and counted using an automated cell counter.

Xenograft Tumor Model.
All procedures performed in studies involving mice were carried out according to the Regulations for the Administration of Affairs Concerning Experimental Animals of China (2017) and were approved by the Committee on the Ethics of Animal Experiments in the Hubei Province. Ten female BALB/c nude mice (weight 20-23 g, 4-6 weeks of age) were purchased and cultured in laminar flow cabinets under specific pathogen-free conditions. The mice were randomly divided into two groups (n = 5 per group). 2 × 10 6 SKOV3 tumor cells expressing sh-INHBA and sh-Ctrl were subcutaneously implanted in the right backs of the mice in the INHBA knockdown group and in the control group, respectively. All mice were killed humanely on day 28 after transplantation, and tumor nodules were dissected, weighed, and paraffin-embedded for subsequent detection.
2.13. Statistics. All statistical analyses were performed using SPSS 22.0 (IBM, Ehningen, Germany) and GraphPad Prism 5.0 software (GraphPad Inc., San Diego, CA, USA). The Shapiro-Wilk test was used to evaluate whether the data were normally distributed. For normally distributed data, the data were presented as means ± standard error of the mean ðs:e: m:Þ for at least three independent experiments and evaluated using Student's unpaired two-tailed t-test for comparisons between two groups, and one-way ANOVA followed by Tukey's posttest for analyses among multiple groups. For nonnormally distributed data (INHBA IHC scores), the   data were presented as medians ± interquartile range, and the Mann-Whitney test was performed to compare the IHC scores between the normal ovary and OC groups.
The Kruskal-Wallis test and chi-squared test were used to analyze the relationship between INHBA expression and clinical pathological parameters. P < 0:001 was considered very significant ( * * * ), P < 0:01 was considered highly significant ( * * ), and P < 0:05 was considered statistically significant ( * ).  (Figure 1(h)).    (Figures 3(b) and 3(c)). However, cell immunofluorescence for Ki-67 indicated that there were no significant differences in the positive staining rate between the si-INHBA group and control group (Figure 3(d)). Clone formation and cell counting assays showed that knockdown of INHBA did not significantly inhibit SKOV3 cellular growth ability (Figures 3(e) and 3(f )). Collectively, the above results confirmed that knockdown of INHBA has no noteworthy effect on the proliferation of SKOV3 in vitro.

INHBA Blockade in Tumor Cells Causes Xenograft Tumor Growth Inhibition and Reduces Stroma Components
In Vivo. INHBA encodes Inhibin β A which is a member of the TGF-β pathway and has been declared to be a driver of the CAF phenotype in OC [22,23].  The KEGG and GO analyses showed that these genes related to INHBA were highly enriched in pathways that encoded ECM processes, such as extracellular matrix organization, extracellular structure organization, and collagen fibril organization (Figures 5(a) and 5(b)). Next, single-sample GSEA (ssGSEA) showed that INHBA mRNA expression was positively associated with the stromal activation score in the GSE 9891, GSE 26193, GSE 51088, and TCGA datasets ( Figure 5(c)). Western blot data showed that exposure of fibroblasts to activin A obviously increased the expression of FAP, α-SMA, and FSP1 ( Figure 5(d)). Furthermore, an INHBA-neutralizing antibody significantly reversed the α-SMA elevation and cytoskeletal stretch caused by SKOV3 CM in fibroblasts ( Figure 5(e)). In addition, SKOV3 CM-induced upregulated ability of stromal fibroblasts in contracting ECM was also reduced by INHBA-neutralizing antibody ( Figure 5(f)). Meanwhile, CM from SKOV3 CM-treated fibroblasts in turn increased the Ki-67-positive rate and cell growth in SKOV3, and these effects were eliminated by INHBA-neutralizing antibody (Figures 5(g) and 5(h)). These data implied that SKOV3 cell-derived INHBA promoted stromal fibroblast activation, and these activated fibroblasts maintained tumor cell growth.
3.7. The SMAD2 Signaling Pathway Is Involved in INHBA-Induced Fibroblast Activation. It was well documented that INHBA is a member of the TGF-β pathway and can activate Smad signaling by binding to ACVR2A [5]. Similarly, our western blot analysis showed that SKOV3-derived CM and recombinant activin A both induced phosphorylation of Smad2 in NOFs and MRC5 (Figure 6(a)). Furthermore, SB431542, an inhibitor of the Smad signaling cascade, not only suppressed phosphorylation of Smad2 but also decreased stromal fibroblast activation induced by SKOV3derived CM and recombinant activin A (Figures 6(b) and 6(c)). Immunofluorescence assay to evaluate α-SMA, and collagen contraction experiments, demonstrated that SB431542 effectively reversed stromal fibroblast activation caused by SKOV3-derived CM (Figures 6(d) and 6(e)). These data suggested that the Smad2 signaling pathway was involved in INHBA-induced fibroblast activation.

Discussion
In the present study, we demonstrated that INHBA mRNA and protein were overexpressed in ovarian cancer (OC) tissues, and INHBA expression significantly increased with the advance of serous ovarian cancer (SOC) pathological grades and clinical stages. Kaplan-Meier plotter analysis showed that patients with SOC with higher INHBA expres-sion had worse overall survival (OS) outcomes. Through comprehensive in vivo and in vitro experiments, we confirmed that knockdown of INHBA in tumor cells reduced OC stromal fibroblast activation, which turned to inhibit tumor growth. Furthermore, we found that the Smad2 signaling pathway was involved in INHBA-induced stromal fibroblast activation (Figure 7(a)). Previous researches have demonstrated that the expression of INHBA is related with prognosis of different types of cancer, such as lung cancer, colorectal cancer, gastric cancer, urothelial carcinoma, and breast cancer [6,[8][9][10][11]. Okano et al. reported that INHBA was a predictor of poor prognosis in patients with colorectal cancer [9]. Wang et al. showed that INHBA overexpression implied adverse clinical outcomes in patients with gastric cancer [29]. In lung adenocarcinoma, Seder et al. demonstrated that upregulated expression of INHBA promoted cell proliferation and was related with poor survival [6]. In OC, INHBA was reported to cause sex cord-stromal tumors, which was evidenced by the occurrence of these tumors in INHA knockout mice [30]. However, the contribution of INHBA to epithelial OC progression is controversial. Dean   stem-like cells and to promote cancer cell metastasis [32]. Additionally, INHBA gene is found to mediate activation of the TGF-β signaling pathway to promote gastric cancer cell migration and invasion [33]. In breast cancer, INHBA signaling promotes breast cancer metastasis by regulating IL13Rα2 expression [8]. However, in OC, the role of INHBA is mixed and has not yet been fully elucidated. Welt et al. found that the majority of ovarian cancer cells did not exhibit accelerated proliferation in response to activin A treatment [34]. Ramachandran et al. showed that some epithelial OC cell lines did not respond to activin A, while others showed growth inhibition [35]. In our study, we demonstrated that INHBA was abundantly expressed in SKOV3 cells, and knockdown of INHBA did not significantly influence SKOV3 cellular growth ability in vitro. However, OC cells grow in highly complicated stromal microenvironments that nurture them through metabolic remodeling, catabolism, autophagy, and inflammation and are capable of facilitating metastasis and resistance to therapy [36]. The results of in vitro experiments did not adequately represent the real situation in vivo. Our findings confirmed that decreased INHBA expression in SKOV3 had no effect on proliferation of tumor cells in vitro, but could hamper tumor xenografts growth by decreasing activation of stromal fibroblast in vivo. To better elucidate the regulation of stromal activation by INHBA in OC, we defined 23 genes coexpressed with INHBA including FAP, THBS2, COL5A2, VCAN, and COL11A1. Further analyses demonstrated that these 23 genes were mainly related to the stromal ECM and collagen-regulated processes. In addition, the stromal fibroblast activation caused by SKOV3-driven CM was significantly reversed by an INHBA-neutralizing antibody. Furthermore, we found that phosphorylation of Smad2 was involved in INHBA-induced fibroblast activation, and SB431542, an inhibitor of Smad signaling cascade, not only suppressed phosphorylation of Smad2 but also decreased stromal fibroblast activation induced by SKOV3-derived CM and recombinant activin A. This study suffered from several limitations. First, singlesource tumor tissue samples seem to be inadequate to reach greater reliability. Further multicentric and large-scale studies are required to verify our present findings. Second, the specific molecular mechanism by which INHBA activates stromal fibroblasts requires further characterization.

Conclusions
This study revealed the prognostic role of INHBA in SOC at different clinical stages and pathological grades. Additionally, our results shed light on the activation role of OC cell-derived INHBA in stromal fibroblasts, which was via the p-Smad2 pathway and promoted tumor xenograft growth. Our observations suggested that inhibition of INHBA in tumor cells could be a potential therapeutic approach to inhibit tumor progression and improve survival rates.

Data Availability
The data used to support this study are available from the corresponding author on reasonable request.

Ethical Approval
The study was approved by the Institutional Research Ethic Committee of Tongji Hospital, Huazhong University of Science and Technology.