Role of microRNAs in premature ovarian insufficiency

Menopause is defined as the permanent cessation of menses. The mean (±SD) age for natural menopause is 50 ± 4 years [15]. The diagnosis of POI is confirmed by amenorrhea for a minimum of 4 months, deficiency in sex steroids (estrogen and progesterone), and FSH > 40 IU/L (in normal women FSH is <10 IU/L) in less than 40-year-old women [1‐3]. POI affects 1/1000 women by the age of 30 years and 1/100 by 40 years [16, 17].

The ovary is uniquely crucial for reproductive and endocrine functions [1]. The healthy ovarian function is vital for the production of both functional gametes and sex steroids [1]. Several investigations have addressed the health issues in patients with POI. Firstly, women diagnosed with POI suffer from infertility, and the chance of spontaneous conception has been estimated to be 4–10% [18, 19]. Infertility cannot be resolved by in-vitro fertilization (IVF) except by receiving donation cycles of oocyte/embryo [18, 19]. Secondly, women with POI showed worse sexual performance, which could not be resolved by hormonal therapy that was accompanied by increased pain and poorer lubrication than the women with normal gonadal function in the control group [20]. Thirdly, the females with early onset of menopause and premature ovarian failure present a lower value of bone mineral density (BMD) in both femoral neck and lumbar vertebrae [21‐26], increasing the bone fragility and risk of fracture than the normal women. Fourthly, the risk of mortality from ischemic heart disease increases up to approximately 80% in the POI group experiencing early menopause as compared to women who underwent late menopause at 49–55 years of age [27]. Finally, the risks of cognitive dysfunction, dementia, social anxiety, depression, Parkinsonism, and impairment of memory are notably increased among women suffering from POI [28‐32]. The primary health implications are summarized in Fig. 1.

Although 90% of the POF cases are idiopathic and the etiology in the remaining 10% is largely heterogeneous, major advances have been made in identifying the underlying mechanisms of POI [6]. The causes of POI have been reviewed thoroughly elsewhere including the treatment-related, environmental, metabolic, iatrogenic, autoimmune, and genetic factors [6, 33‐35]. With the advancements in the whole genome sequencing, much progress has been made in the analysis of the candidate genes [34]. However, the protein-coding genes are usually under intensive research, whereas the underlying regulatory roles of non-coding sequences in POI are yet to be elucidated.

The history of the biological study states that the protein-coding genes have always been the main focus and are well-studied. However, the coding sequences account for only 1.5% of the genome [7]. The term non-coding RNA is employed for RNA that does not encode a protein, but such RNAs also contain information and have functions [36]. Thus, the ncRNAs are gaining increased attention. The ncRNAs are functionally important in development, physiology, and diseases, and are categorized as following: miRNAs, PIWI-interacting RNAs (piRNAs), large intergenic non-coding RNAs (lincRNAs), other long non-coding RNAs (lncRNAs), and circular RNAs (circRNAs). The most widely studied class of ncRNAs is miRNAs, which are small (18–22 nt length) ncRNAs that can mediate the post-translational gene silencing, thereby negatively regulating the target genes.

As illustrated in Fig. 2, the biogenesis of miRNAs is a multi-step process [11]. Firstly, the primary precursor molecules (pri-miRNAs), ~500–3000 nt, are transcribed by RNA polymerase II from either independent genes or representative introns of the protein-coding genes. Secondly, the pri-miRNAs are cleaved by the RNase III family member (Drosha) and its cofactor DGCR8 into an ~70 nt pre-miRNA, which then is exported to the cytoplasm by XPO5. Thirdly, the pre-miRNAs are processed by Dicer, which is also a member of the RNase III family, assisted by transactivation-responsive (TAR) RNA-binding protein (TRBP) to form a ~20bpmiRNA/miRNA* duplex. One strand of the duplex is a mature miRNA (guide strand) that can be incorporated into a miRNA-induced silencing complex (miRISC), whereas the other strand, termed as the passenger strand, is released and degraded. Most miRNAs target the mRNA by yielding continuous base-pairing of 2–8 nt of miRNA that can be designated as the seed region with complementary sequences in the 3′untranslated regions (UTR) of the target mRNA transcripts. The base-pairing results in either mRNA degradation and deadenylation or translation repression. These miRNAs are ascribed to regulate the translation of > 60% of the protein-coding genes that participate in proliferation, differentiation, apoptosis, development, tumorigenesis, and hematopoiesis [9, 10].

The global knockout of Dicer and Drosha led to early embryonic lethality in mice (i.e., embryos die at embryonic day 6.5) [37, 38], which emphasizes the developmental role of miRNAs. The loss of function of Dicer or Drosha in prophase I oocytes, developing oocytes, and follicular granulosa cells have demonstrated that miRNAs play critical roles in both the development and functions of the ovary. Selective inactivation of Dicer in prophase I oocytes of the fetal ovary led to compromised folliculogenesis, premature ovarian failure, as well as, infertility in the adult ovary [38]. Dicer1 is enriched in developing oocytes; the conditional knockout of the gene in growing oocytes results in defective spindle organization and chromosome cohesion that in turn, inhibits meiosis I [39], thereby suggesting that Dicer 1 is indispensable for oocyte maturation. However, its function is dispensable for oocyte growth, development, and response to hormonal signals. The inactivation of Dicer1 in the follicular granulosa cells increased the primordial follicle pool endowment, accelerated the early follicle recruitment, and led to increased degeneration of the follicles in the ovaries [40]. Nevertheless, the selective inactivation of Drosha in the oocytes of either the fetal or the developing ovary did not exert any effect on normal folliculogenesis and female fertility in adulthood [38].

Furthermore, the number of studies with respect to the above phenomena is limited in humans. Only one group reported that XPO5 rs2257082 T variant allele frequently occurs in patients with POI than the controls and this allele may be associated with an increased risk of POI [41]. The molecular machinery associated with POI are illustrated in Fig. 2.

Surprisingly, neither the knockout study of XPO5 in mice ovary has been conducted nor the single nucleotide polymorphisms (SNPs) of the other genes involved in the molecular machinery have been reported in patients with POI. Hence, loss function of XPO5 in either oocytes or granulosa cells will shed light on the correlation between miRNA and ovary development. In addition, the analysis of Dicer and Drosha polymorphisms in POF women would also provide further clues about the genetic etiology underlying POF.

The miRNA expression profiles in the plasma of three idiopathic patients with POI and three control group women with normal cycle were analyzed. Yang et al. identified 10 miRNAs that were significantly up-regulated and 2 miRNAs that were down-regulated in patients with POI [42]. According to another study in Han Chinese patients with POI, a total of 51 differentially expressed miRNAs were identified by miRNA microarray analysis from10 POF and ten control subjects. Of these, 22 miRNAs were significantly up-regulated and 29 were significantly down-regulated in POI group as compared to the control subjects [43], suggesting the essential roles of these miRNAs in POI. Surprisingly, the two groups reported different miRNAs except for miR-23a and let-7c [42, 43]. These different results might at least partially be attributed to the different sample sizes, indicating that the clinical sample size should be expanded in order to obtain persuasive and repeatable results in the future.

Several miRNAs such as miR-23a, miR-27a, miR-22-3p, miR-146a, miR-196a, miR-290-295, miR-423, and miR-608 have been suggested to be associated with POF. Herein, we addressed the implicated role of these miRNAs in the development of POF.

Both miR-23a and miR-27a are significantly up-regulated in the plasma of women with POI [42, 43], implying the regulatory role of these two miRNAs in the development of POI. Reportedly, miR-23a-27a-24-2 cluster plays critical roles in normal and pathological processes such as cell cycle, proliferation, differentiation, apoptosis, hematopoiesis, and cardiac hypertrophy [44, 45]. MiR-23a induced the apoptosis of granulosa cells by down-regulating the X-linked inhibitor of apoptosis protein (XIAP) with a subsequent increase in caspase-3 cleavage [42]. Both miR-23a and miR-27a mediated the apoptosis of granulosa cells by targeting SMAD5 that in turn, directly activated the Fas ligand (FasL)-Fas pathway in vitro [46]. In addition, Kim et al. found that transfecting a miR-27a mimic sequence into granulosa cells decreased the rate of oocyte maturation in mouse follicles [47]. The insulin growth factor-binding protein 2 (IGFBP-2) produced by granulosa cells regulates the bioavailability of the intra follicular ovarian insulin growth factor (IGF) during follicular development [48]. The IGFBP-2 expression is altered as a result of miRNA-27a transfection; however, the downstream target modulated by this pathway has not yet been explored. Previous studies reported that the expression of IGFBP-2 increased in gonadotrophin-stimulated murine granulosa cells [49], thereby indicating the unbound fraction of IGF as the candidate target. In conclusion, miR-23a and miR-27a are critical for folliculogenesis, putatively mediating the apoptosis of granulosa cells and oocyte maturation via regulating the growth factors during follicular development, and thus, playing a role in the development of POI.

In Chinese patients with idiopathic POF, miR-22-3p was found to be down-regulated and negatively associated with serum FSH [43]. MiR-22-3p has also been reported as one of the most abundant miRNAs in human granulosa cells and is predicted by bioinformatics analysis to negatively regulate the FSH secretion and target estrogen receptor 1 (ESR1) and phosphatase and tensin homolog (PTEN). ESR1 and PTEN have been considered as candidate genes responsible for POF [34], significantly in folliculogenesis [34]. Moreover, miR-22-3p acts as a suppressor of ESR1 by directly reducing the mRNA level of the gene in breast cancer cells in vitro [50]. However, the predicted interaction of miR-22-3p with ESR1 and PTEN has not yet been determined in ovaries. The potential roles of miR-22-3p in the pathogenesis of POI and physiological functions remain to be elucidated.

In a Korean study in women with POI [51], the authors demonstrated that the miR-146aC allele coupled with the miR-196a2C allele and the miR-146aG allele coupled with the miR-196a2T allele reduced the risk of POF. However, the miR-146aC allele in combination with the miR-196a2T allele and the miR-146aG allele in combination with the miR-196a2C allele increased the risk of POF [51]. The expression of miR-146a was found to be up-regulated in the plasma and ovarian granulosa cells obtained from idiopathic patients with POI [42, 52]. MiR-146a is expressed in the oocyte during bovine oocyte maturation and preimplantation embryo development, and one of the putative target genes of miR-146a is Fas that regulates oocyte apoptosis during folliculogenesis via the caspase8 pathway [53‐55]. In addition, miR-146a is also involved in tumor necrosis factor (TNF)-α-dependent regulation of folliculogenesis and atresia by activating a cascade of caspases [56]. Furthermore, miR-146a induced apoptosis in granulosa cells via interleukin-1 receptor-associated kinase (IRAK1) and tumor necrosis factor receptor-associated factor 6 (TRAF6); these molecules regulated the activity of nuclear factor-kappa B(NF-κB) and inhibitory kappa B α(IκBα), thereby modulating the expression of caspase pathway that could be attenuated by caspase inhibitors [52]. The apoptosis of granulosa cells is pivotal in POI. The understanding of the regulatory mechanisms of miRNAs in apoptosis paves the way for illustrating the pathogenesis of POI. MiR-196a inhibit the expression of newborn ovary homeobox gene (NOBOX) during embryogenesis [57]. The mutations in the NOBOX gene have been associated with POF [58], suggesting that miR-196a might increase the risk POF by regulating NOBOX. Nevertheless, whether NOBOX is regulated by miR-196a in ovary remains to be elucidated. The target genes are possibly influenced by miR-146a and miR-196a; subsequently, the aberrant folliculogenesis occurs. However, the underlying mechanisms are yet to be revealed in future studies.

Recently, Medeiros et al. reported that female miR-290-295-deficient mice are rendered sterile by POF due to the mislocalization of the migrating primordial germ cells (PGCs) [59]. The specific duration of the expression of miR-290-295 explains at least a proportion of idiopathic POF. Strikingly, this cluster is not listed in the miRNAs of adult women with POI.

Another Korean study suggested that breast cancer-related miRNA polymorphisms, including miR-27a A > G, miR-423C > A, and miR-608G > C, are associated with increased idiopathic risk of POI via interaction among the miR-27aG, miR-423A, and miR-608G variants [60]. However, the study should be further substantiated owing to the small sample size and low statistical power.

Amniotic fluid stem cells (AFSCs)-derived exosomal miR-10a contributes significantly to prevent POF caused by chemotherapy in mice [61]. Therefore, the study provides new insights into POF, thereby advancing the therapeutics. In addition, the work suggest that loss function of miR-10a in ovary may help to illuminate the connection between miR-10a and POF in the future.