MicroRNAs in ovarian follicular atresia and granulosa cell apoptosis

Female fertility potential is based on the development and growth of ovarian follicles, and animal breeds with higher fecundity often show larger numbers of mature follicles in their ovaries and a higher ovulation rate than those with lower fecundity. However, the mammalian follicle utilization rate is extremely low because most follicles are removed from the ovaries before ovulation via a degenerative process known as atresia [1, 2]. Among domesticated animals, cows and sheep have approximately 1 million follicles, and sows have approximately 5 million primordial follicles in their ovaries at birth. However, more than 95% of these follicles undergo atresia [3‐5]. Of the approximately 2 million primordial follicles that exist in human ovaries at birth, only up to 400 are ovulated throughout a female’s lifetime, which means that more than 99% of the follicles are lost. Therefore, methods to develop and utilize the abundant resources of ovarian follicles are consistently sought in animal reproduction research. Follicular atresia is a complex process that involves multiple regulatory factors, and studies on atresia mechanisms can be summarized in 3 aspects: 1) studies on follicular morphology and biochemistry, 2) studies on regulatory signalling pathways in granulosa cell (GC) apoptosis, and 3) studies on transcriptional/posttranscriptional mechanisms in GC apoptosis.

In 1987, Moor R et al. were the first to identify morphological changes in the atretic follicles of sheep [6], including the degree of translucency, vascularization of the follicle membrane and integrity of the membrane GC layers. Similar morphological features of isolated follicles were also observed in sows and cows [7, 8]. These features have been continually applied as important classification standards for atresia identification.

Endocrinological quantifications in follicular fluid are also standards for identifying antral follicle atresia. The concentration of oestrogen (E2) and the ratio between E2 and progesterone (P4) indicate a high correlation with the macroscopic classification of atresia. The steroid hormone concentrations detected by the enzyme immune assay (EIA) in medium-sized follicles (8–10 mm) in bovine follicular fluid suggest a P4/E2 < 1 in healthy follicles and a P4/E2 > 1 in atretic follicles. The ratio further drops with the degree of atresia [9]. A study on small (3–6 mm) and large (> 6 mm) ewe follicles using a nonextraction, solid-phase radioimmunoassay (RIA) suggested the same criteria [10]. In addition, steroid hormone concentrations detected in medium-sized (3–5 mm) pig follicles using RIA suggest a P4/E2 < 5 in healthy follicles and a P4/E2 > 5 in atretic follicles [11]. Based on our experience, a competitive ELISA method can also be used for hormone detection in pigs, and the classification standard is the same as that for RIA. Although the specific ratio varies depending on the species and detection method, all studies agree that higher E2 concentrations and lower P4/E2 levels indicate healthier follicular conditions.

With the development of ovarian studies, researchers have gradually realized that the basic physical mechanism of follicular atresia is GC apoptosis. When atresia occurs, pyknotic nuclei are first observed in GCs, which is followed by detachment of the GC layer and fragmentation of the basal membrane, ultimately resulting in hypertrophied thecal cells and disruption of thecal integration and thecal vessels. Degeneration of oocytes, however, may occur at any stage of atresia. The above phenomenon was first observed in bovine [12] and subsequently reported in chickens [13], rats [14], cows [15], sheep [16] and sows [17]. Therefore, follicular atresia studies have become increasingly focused on the molecular regulation of GC apoptosis. Further studies have shown that GC apoptosis may occur much earlier than the morphological changes in follicular atresia, which can be observed only when GC apoptosis reaches a certain degree [18, 19]. Generally, proliferation and differentiation of GCs leads to follicular maturation and ovulation, whereas apoptosis and degeneration of GCs results in follicular atresia [13, 20].

A considerable number of known molecular factors are involved in the regulation of GC apoptosis, including proapoptotic molecules, such as FAS (Fas cell surface death receptor), CASPs (caspases), MYC (MYC proto-oncogene BHLH transcription factor), TNF (tumour necrosis factor), F2RL3 (F2R-like thrombin or trypsin receptor 3, Par4) and PHB (prohibitin), as well as molecules that promote cell survival, such as activin, KITLG (KIT ligand, SCF), IFN (interferon), EDNs (endothelins), TP53 (tumour protein P53)b and NOBOX (newborn ovary homeobox-encoding gene) [21‐24]. Although new regulatory factors are continuously being identified, comprehensive knowledge of the signalling networks that function during GC apoptosis remains limited.

In recent decades, the rapid rise of posttranscriptional mechanistic studies, especially miRNA studies, has brought a new perspective to reproductive system research. miRNAs comprise a broad class of endogenous, short, noncoding single-stranded RNA molecules that are normally 18–24 nt in length. Like mRNAs, miRNAs are originally transcribed from coding genes that are hundreds of kilobases long and occupy 1–3% of the genome [25]. miRNA-coding genes are distributed across chromosomes either individually or in clusters in which two or more miRNA genes are located within a short distance on the same segment of a chromosome. Thus, miRNA clusters on a given segment usually show a synergetic transcription pattern and related regulatory functions [26]. Based on gene structure, pri-miRNAs can be classified into three broad categories: Class I pri-miRNAs are transcribed independently of other genes, Class II pri-miRNAs are transcribed as an extension of a protein-coding gene, and Class III pri-miRNAs are transcribed as an extension of a noncoding RNA [27]. However, pri-miRNAs are poorly detected by standard methods, such as RT-PCR and RNA sequencing, due to rapid recognition and processing by the microprocessor complex, an enzyme arrangement made up of one Drosha protein and two DGCR8 proteins [28], resulting in 70–90 nt precursor miRNAs (pre-miRNAs) with hairpin structures. The pre-miRNAs are then transferred into the cytoplasm, and after further processing by Dicer, a cytosolic RNase III-type endonuclease, 18–24 nt double-stranded miRNAs are generated. With the elimination of one strand, the final single-stranded mature miRNA is loaded into an Argonaute (AGO) protein to form the miRNA-induced silencing complex (miRISC, RISC), which binds to specific sequences on target mRNAs, specifically to their 3′-untranslated regions (3′-UTRs), and results in either target mRNA degradation [29] or target mRNA transcriptional repression [30]. The estimation that nearly all mature sequences of coding mRNA transcripts contain miRNA response elements (MREs) implies crucial roles of miRNAs in almost all biological contexts during developmental, physiological and pathological processes. It is notable that one miRNA can regulate more than one gene, and conversely, a single gene can be regulated by multiple miRNAs [31].

In mammalian ovaries, folliculogenesis and atresia processes are tightly regulated by a complex network of genes. Therefore, miRNAs, which can originate from both cellular and extracellular sources, function as mediators of these processes via their extensive involvement in posttranscriptional mRNA regulation. The roles of ovarian miRNAs have been summarized in previous reviews on follicular and luteal development [32, 33], ovarian function [3, 34‐36], and ovarian diseases and disorders [37‐39]. Thus, this review primarily focuses on the latest findings concerning the roles of miRNAs in follicular atresia and GC apoptosis.

Studies on ovarian miRNAs in humans, mice, cows, sows, sheep and goats have gradually assembled specific profiles of miRNA transcription in mammalian ovarian tissues. Ovarian miRNA profiles were also reported in studies on the reproduction of poultry animals, such as chicken [40] and geese [41]. High-throughput miRNA profiling studies have provided valuable references for deeply understanding the molecular mechanisms in ovarian tissue. Herein, we list the mammalian ovarian miRNA profiles that have been identified in whole ovaries, isolated follicles, or certain parts of follicles, including extracellular vesicles (EV) in follicular fluid, during various physiological and pathological processes (Table 1).

These studies applied microRNA sequencing, microarray or PCR array methods and provided useful candidates for further studies, which may lead to the identification of biomarkers for medical diagnosis in ovarian disorders. To gain a general view of miRNA functions in ovaries, we generated 2 cloud charts (Fig. 1) based on previously performed high-throughput studies by simply integrating the results of research on 1) miRNAs differentially expressed during follicular atresia [42, 43] and 2) miRNAs differentially expressed among small, medium-sized and large follicles [43‐48]. Clearly, miRNAs that are involved in different biological processes are remarkably divergent in the ovary. Among the most notable miRNAs that changed during atresia, miR-1275, which is known to regulate insulin-like growth factor-2 mRNA-binding proteins (IGF2BP1, IGF2BP2 and IGF2BP3) in cancer [49], was recently shown to affect E2 synthesis and lead to GC apoptosis by targeting liver receptor homologue (LRH)-1 [50]. In addition, although not yet related to ovarian atresia, miR-1826 was reported to regulate β-catenin (CTNNB1) by the observance of direct binding in cancer studies [51], while miR-190a was shown to mediate VEGF [52] and oestrogen receptor (ER)-related signalling in angiogenesis [53], and miR-210 was proven to be related to hypoxia [54, 55]. It is clear that high-frequency miRNAs are connected to atresia-related pathways and can thus be considered helpful hits in follicular atresia and cell apoptosis-related studies.

To explore the effective functions of miRNAs during atresia, gain- and loss-of-function studies have been applied by researchers in vitro. A list summarizing known studies on miRNAs directly or indirectly involved in GC proliferation and apoptosis is presented in Table 2, which includes combinations of miRNAs, their verified target genes and miRNA functions. The list will certainly continue to grow with contributions from research on this topic. Here, we attempt to summarize several main families and groups of miRNAs that play critical roles in GC apoptosis and the follicular atresia process.

miRNA-mediated regulation is now widely valued in ovarian function studies. Although the mechanisms of follicular atresia have not been precisely defined, recent reports have demonstrated a causal link between the levels of certain miRNAs and the fate of GCs/follicles in several species. Profiling studies have demonstrated that diversified miRNAs play roles in various ovarian physiological and pathological processes. In the specific area of follicular atresia and GC apoptosis, some miRNAs are known to work individually to regulate particular genes, while others function in clusters to finely adjust certain signalling pathways or cellular processes. However, in both cases, the function of miRNAs should not be viewed as a single factor but as part of a regulatory network that involves hormones, growth factors and other regulatory noncoding RNAs from each component of the follicle. The ultimate functions of miRNAs on certain target genes can be affected by multilevel regulatory networks, from the transcriptional regulation of their own coding genes to interference from other competing endogenous RNAs, to achieve timely adjustment in cells. Therefore, the following areas may attract more attention in future studies: 1) competition and cooperation of regulatory noncoding RNAs, such as miRNAs, lncRNAs (long noncoding RNAs) and piRNAs (Piwi-interacting RNAs); 2) communication among miRNAs generated from different parts of the follicle, including oocytes, GCs (from the cumulus or near basement membrane) and thecal cells; 3) influence of single-nucleotide polymorphisms (SNPs) on miRNA genes that are involved in ovarian pathological processes; and 4) functions of small RNAs in EV. In addition, with the development of high-throughput methods, comprehensive studies on noncoding RNA networks and interactions with their targets will lead to a deeper understanding of miRNA functions in ovarian follicular atresia. The advancement of our knowledge of miRNA-mediated networks will not only help elucidate the mechanisms of GC apoptosis, follicular development, atresia and other disorders but also provide potential uses for future clinical diagnosis and treatment.