circRNA-Mediated Inhibin–Activin Balance Regulation in Ovarian Granulosa Cell Apoptosis and Follicular Atresia

Ovarian granulosa cells (GC) play an essential role in the development and atresia of follicles. Emerging studies suggest that non-coding RNAs are involved in the regulation of GC apoptosis. Here, we aimed to analyze the function of ssc-circINHA-001, coded by the first exon of the inhibin subunit α gene (INHA), in resisting GC apoptosis and follicular atresia by enhancing the expression of the inhibin subunit β A (INHBA) through a cluster of miRNAs. A higher expression of ssc-circINHA-001 in healthy follicles compared to early atretic follicles was detected by qRT-PCR. Its circular structure was confirmed by RNase R treatment and reversed PCR. The function of ssc-circINHA-001 in GC resistance to apoptosis was detected by in vitro transfection of its si-RNA. Furthermore, the dual-luciferase reporter assay suggested that ssc-circINHA-001 adsorbed three miRNAs, termed miR-214-5p, miR-7144-3p, and miR-9830-5p, which share the common target INHBA. A low expression of ssc-circINHA-001 increased the levels of the free miRNAs, inhibited INHBA expression, and thus raised GCs apoptosis through a shift from the secretion of activin to that of inhibin. Our study demonstrated the existence of a circRNA–microRNAs–INHBA regulatory axis in follicular GC apoptosis and provides insight into the relationship between circRNA function and its coding gene in inhibin/activin balance and ovarian physiological functions.


Introduction
The quantity of ovarian follicles determines the reproductive rate of livestock [1]. Ovarian granulosa cell (GC) apoptosis is a physiological phenomenon that occurs during all stages of follicular development. Research has demonstrated that the percentage of apoptotic GCs increases significantly with the progression of follicular atresia [2].
Inhibin and activin belong to the growth factor-beta (TGF-β) family of cytokines and play contrasting roles in regulating follicle-stimulating hormone (FSH) secretion, whereby inhibin inhibits and activin activates FSH secretion. Inhibin A and B are heterodimers comprising an αand a β(A/B)-subunit, respectively. In other words, inhibin A and B arise from three gene products, including the α-subunit gene (INHA), the βA-subunit gene (INHBA), and the βB-subunit gene (INHBB). Activins are homodimers or heterodimers of only the β subunits (βA, βB, βC, βD, and βE). Inhibins and activins counteract each other, and the balance between the amounts of αand β-subunits will determine how much inhibin versus activin is formed, e.g., increased levels of α-subunit will favor the formation of inhibin, while increased levels of β-subunit will favor the formation of activin [3]. As TGF-β family members, the posttranscriptional regulation of αand β-subunits expression is largely unknown.
Among non-coding RNAs (ncRNAs), microRNA (miRNA) is the most studied and has been widely reported to participate in follicle development and atresia regulation [4]. Our previous study identified different miRNA profiles in porcine healthy follicles (HF) compared to early atretic follicles (AF) [5]. Furthermore, circular RNA (circRNA) is a new type of ncRNA produced from a precursor mRNA through back-splicing of exons. circRNAs are located in the nucleus or cytoplasm and regulate gene expression through transcriptional (e.g., interaction with RNA-binding protein) or posttranscriptional pathways (e.g., adsorption of miRNAs) [6]. An increasing number of studies have pointed out that circRNA has a significant role in cell differentiation, development, and dysfunction [7,8]. Our previous study showed an abundant distribution of circRNAs in porcine follicles, and the differential expression in HF and AF suggested a possible regulatory function of circRNAs during atresia [9].
In this study, we reported that a newly identified circRNA, ssc-circINHA-001, coded by the INHA gene, guaranteed INHBA expression by sponging miR-214-5p, miR-7144-3p, and miR-9830-5p, thus further inducing GC resistance to apoptosis and follicular atresia. Our study demonstrated a circINHA-microRNAs-INHBA regulatory pathway in GC apoptosis and increases knowledge of the balanced regulation of inhibin A formation and the relationship between the circRNA coding gene and its downstream target gene in ovarian physiological functions.

Identification and Validation of ssc-circINHA-001
Our previous deep circRNA sequencing analysis showed that the expression level of ssc-circINHA-001, a candidate circRNA produced from the first exon (5 UTR) region of the INHA gene, decreased during follicular atresia [9]. To further elucidate the role of this circRNA in follicular atresia, the expression levels of ssc-circINHA-001 in HF and AF were measured by qRT-PCR, and a significant reduction was observed in AF ( Figure 1A). The circular structure of ssc-circINHA-001 was verified by amplification with divergent primers, followed by Sanger sequencing. The result confirmed that ssc-circINHA-001 was formed by reverse splicing of the first exon region of INHA mRNA precursor ( Figure 1B). As RNase R resistance is one of the key characteristics of circRNAs, the stability of ssc-circINHA-001 was verified by qPCR after RNase R digestion. A dramatically reduced GAPDH expression level proved the efficiency of digestion. However, an increased ssc-circINHA-001 level was detected, demonstrating that ssc-circINHA-001 was resistant to RNase R digestion and structurally stable ( Figure 1C). Furthermore, the distribution of ssc-circINHA-001 in GCs was detected by FISH, and the results indicated that ssc-circINHA-001 was mainly distributed in the cytoplasm of GCs ( Figure 1D). These results revealed the existence and circular structure of ssc-circINHA-001, which might play a role in porcine follicular atresia.

INHBA Regulates Follicular Atresia and GC Apoptosis and Is Targeted by
Three bioinformatic software, PITA [10], Miranda [11], TargetSpy [12] were adopted to predict the potential target genes of miR-214-5p, miR-7144-3p, and miR-9830-5p, and INHBA was highlighted. To explore the possible function of INHBA, the mRNA expression levels of INHBA in healthy and atretic follicles were detected by qRT-PCR. The results showed that the INHBA level was significantly lower in AF than in HF ( Figure 6A). We then synthesized the siRNA of INHBA (si-INHBA) to identify the role of INHBA in GC apoptosis. The results showed that si-INHBA significantly reduced INHBA mRNA levels ( Figure 6B), and FACS showed that si-INHBA significantly increased GC apoptosis rate ( Figure 6C), which indicated that INHBA played a role in GC survival.
To prove the binding of three miRNAs to the target INHBA, dual-luciferase activity assays were applied to verify the binding sites of miR-214-5p, miR-7144-3p, and miR-9830-5p at positions 731~737 nt, 239~250 nt, and 605~613 nt, respectively, on INHBA 3 UTR ( Figure 6D-F). According to the results, luciferase activity decreased significantly after transfection of the miRNA mimics in wild-type vectors. In contrast, no significant changes were observed with mutant vectors. These observations indicated that INHBA is the direct target of miR-214-5p, miR-7144-3p, and miR-9830-5. Subsequently, INHBA mRNA levels and secreted INHBA in the culture of GCs were detected by qRT-PCR ( Figure 6G) and ELISA ( Figure 6H), respectively, after the treatment with the three miRNA mimics. The results showed that both mRNA levels and secreted protein levels of INHBA were significantly decreased after miR-214-5p, miR-7144-3p, and miR-9830-5p overexpression.

Discussion
With the development of high-throughput sequencing technology, our knowledge of circRNA has rapidly improved. Compared with the other two popular ncRNA groups, miRNA and lncRNA, circRNA is produced by reverse splicing and characterized by covalently closed-loop structures, and thus is more stable and is widely distributed in animal tissues [13,14]. In recent years, circRNAs have been reported as essential regulators and valuable diagnostic markers in various diseases, such as tumors [15], kidney diseases [16], neurological disorders [17], cardiovascular diseases [18], and diabetes [19]. In reproductive medicine, specific circRNA profiles were identified in ovarian ageing [20], polycystic ovary syndrome (POS) [21,22], and ovarian carcinoma [23,24] and were proved to be aberrantly expressed during pathological changes. These studies have suggested that circRNA plays a role in gene expression and may serve as a potential biomarker and therapeutic target. In animal reproduction, especially in pigs, although it started relatively late, circRNA profiles were identified and compared between different pig breeds [25], during the onset of puberty [26] and during follicular atresia, in our previous studies [27].
The gradually increasing understanding of the regulatory mechanisms of circRNA has revealed that it often acts as a miRNA sponge to regulate target genes and constructs a circRNA-miRNA-gene network. In our study, three rather than one miRNA, termed miR-214-5p/miR-7144-3p/miR-9830-5p, were proved to bind to ssc-circINHA-001. It is reasonable, and has been bioinformatically predicted, that one circRNA may contain more than one miRNA binding site. Recent studies proved multiple sponging functions; for example, circPUM1 promotes ovarian tumorigenesis by sponging miR-615-5p and miR-6753-5p [28]. However, a thorough identification of such miRNA clusters and their downstream targets requires omics data integration. Thus, we made full use of our previous RNA-seq, miRNA-seq, and circRNA-seq results and analyzed possible circRNA-miRNA-mRNA networks in advance. The identification of circINHA-miR-214-5p/miR-7144-3p/miR-9830-5p-INHBA suggested that integrated omics data facilitate the study of circRNA function.
It is worth noticing that, although it is accepted that circRNA functions are generally independent of the circRNA coding gene, our study supports a different conclusion. ssc-circINHA-001 is back-spliced from the inhibin α-subunit coding gene INHA and holds binding sites for miR-214-5p/miR-7144-3p/miR-9830-5p. Interestingly, these miRNAs also share one target gene, INHBA, which codes for the βA-subunit of inhibin. Theoretically, the product of ssc-circINHA-001 somehow reduced the expression of liner mRNA and protein of the INHA subunit. In addition, increasing circINHA-001 enhanced INHBA subunit expression (Figure 7). In this case, the balance of hormone production would be tilted toward activin, promoting FSH secretion and finally inducing GC resistance to apoptosis and follicular atresia, which nicely agrees with the physiological changes during GC apoptosis and follicular atresia. This inference is also supported by a recent study in bovine, suggesting that INHBA could promote the proliferation of GCs, inhibit GC apoptosis, promote the secretion of estrogen and progesterone, and regulate the development of bovine [29] and chicken follicles [30]. In humans, the role of activins and inhibins was valued in pathological research into, for example, cancers and polycystic ovarian syndrome (PCOS). Despite some contradictions in the different cohort studies, it is accepted that the increased expression of activin A contributes to the development of ovarian cancers, breast cancers, and uterine leiomyoma, while inhibin A was reported to be a suppressor of ovarian stromal cell tumors. In PCOS, lower activin A levels and increased inhibin A levels were observed in patients. According to our study, circINHA accelerates activin A formation while inhibiting ncRNAs such as circINHA, providing new insights for therapeutic strategies.
In conclusion, we first identified a new INHA gene-coded circRNA, ssc-circINHA-001, which was more highly expressed in healthy than in atretic porcine follicles. ssc-circINHA-001 served as a ceRNA to guarantee the expression of INHBA by sponging miR-214-5p, miR-7144-3p, and miR-9830-5p simultaneously, and thus increased the production of activin, with an inhibitory effect on GC apoptosis and follicular atresia. This study demonstrated the existence of a new circINHA-microRNAs-INHBA regulatory pathway, and provides new insights into the relationship between a circRNA coding gene and its downstream target gene in ovarian physiological functions, as well as suggestions for the development of new therapeutic methods.

Follicle Collection and Isolation
Porcine ovaries were obtained from healthy, unstimulated 7-month-old commercial gilts from Su Shi slaughterhouse in Huai'an, Jiangsu. The collected porcine ovaries were transferred to the laboratory as soon as possible in physiological saline containing gentamicin sulfate (80 mg/L) at 37 • C. The ovarian tissue was washed and cut with a scalpel in a Petri dish containing PBS. Antral follicles with a diameter of 3-5 mm were isolated using small scissors and forceps [31]. Follicles were divided into HF and AF according to uniform criteria based on follicle appearance, GC density, and progesterone (P4)/estradiol (E2) ratio [32].

Cell Culture and Transfection
The porcine ovaries were washed with sterile physiological saline containing gentamicin sulfate (80 mg/L) at 37 • C until the saline was clear and transparent. GCs were drawn from clear antral follicles with 3-5 mm diameter using a syringe with a 20-gauge needle. GCs were then cultured at 37 • C and 5% CO 2 in DMEM/F12 medium (Gibco, Carlsbad, CA, USA) containing 10% fetal bovine serum (FBS) (Gibco, Carlsbad, CA, USA), 100 units/mL penicillin, and 100 mg/mL streptomycin. HEK293 cells were maintained under the same conditions in DMEM medium (Sigma, St. Louis, MO, USA). For the FISH assay, the cells were cultured on round coverslips placed in 6-well plates. Inhibitor and mimics of circRNA siRNA and miRNA were synthesized by GenePharma (Shanghai, China) (Supplementary Table S1). Cells were transfected using Lipofectamine TM 3000 (Invitrogen, Carlsbad, CA, USA) and Opti-MEM (Gibco, Carlsbad, CA, USA) according to the manufacturer's instructions. All cell experiments were done in triplicate.

RNA Preparation and qRT-PCR
Total RNA was extracted using Trizol reagent (Invitrogen, Carlsbad, CA, USA) according to the instructions and qualified by Nanodrop and agarose gel electrophoresis. For the quantitative determination of circRNA, the first-strand cDNA was synthesized using a PrimeScript RT Master Mix (Vazyme, Nanjing, China) kit. qPCR was performed by the ABI StepOne system, Applied Biosystems, Carlsbad, CA, USA) using SYBR Premix Ex Taq (Takara, Dalian, China), according to the manufacturer's instructions, with GAPDH as the internal reference. For RNase-R-treated samples, GAPDH without RNase R treatment was considered as an internal reference for all samples. For miRNA quantification, reverse transcription was performed with the BioTek miRNA cDNA first-strand synthesis kit (BioTeke, Wuxi, China). Then, detection was performed with the BioTek miRNA fluorescence quantitative detection kit (BioTeke, Wuxi, China), with U6 as the internal reference. The relevant primer information is listed in Supplementary Table S2.

RNase R Treatment and RT-PCR Product Sequencing
The extracted RNA of each GC sample was divided into two equal aliquots. One aliquot was directly subjected to reverse transcription, while the other was treated with RNase R (Epicentre, Madison, WI, USA) before reverse transcription. The 20 µL digestive system included 5 µg RNA, 2 µL 10× Reaction Buffer, 15 U RNase R(20 U/µL), and RNase-Free water. The reaction was performed at 37 • C for 15 min and then at 70 • C for 10 min, using SYBR Premix Ex Taq (Takara, Dalian, China), according to the manufacturer's instructions. The sequence of the PCR products was verified by agarose gel electrophoresis and Sanger sequencing (Bioengineering, Shanghai, China).

FISH
Labelled probes were specifically synthesized for ssc-circINHA-001, miR-214-5p, miR-7144-3p, and miR-9830-5p by Wuhan Seville Biotech, and DAPI was used to stain the cell nuclei. GCs were cultured on coverslips, fixed in 4% paraformaldehyde (containing DEPC) for 20 min, washed three times while shaking with PBS (pH 7.4), and proteinase K (20 µg/mL) was finally added for 5 min for digestion [27]. All procedures were conducted according to the manufacturer's instructions (Sevicebio, Wuhan, China). Fluorescent images were acquired using a Nikon upright fluorescence microscope (Nikon DS-U3, Tokyo, Japan). Each experiment was performed three times. The probe sequences are shown in Supplementary Table S7.

Apoptosis Assay
GCs apoptosis was assessed by the Annexin V-FITC/PI staining assay (Vazyme, Nanjing, China) according to the manufacturer's protocol and our previous description [27]. The percentage of apoptotic cells was determined by flow cytometry (Becton Dickinson FACS Calibur, Franklin Lakes, NJ, USA). The data were analyzed using FlowJo v7.6 software (Stanford University, Stanford, CA, USA).

ELISA Assay
The expression level of INHBA in GC cultures was detected using the INHBA ELISA assay kit according to the manufacturer's instructions (Jianglaibio, Shanghai, China). The OD value of each well was detected at 450 nm by a plate reader (Multiskan GO, Thermo Scientific) and analyzed by the supplied program.

Statistical Analysis
GraphPad Prism 5 software was used to perform statistical analyses. Two-tailed Student's t-tests were used to evaluate significance when the two groups were compared. All data are presented as the mean ± SEM. p-values of <0.05 (*) and 0.01 (**) were considered to indicate significant and highly significant differences, respectively.