Introduction
Infertility, the inability to conceive and have children, is estimated to affect as many as 186 million people worldwide, and, consequently, represents an important social and medical problem (Inhorn and Patrizio, 2015). It can be considered a complex disease: first of all it is related to two different genomes (oocyte and sperm quality represent major determining factors in reproductive success), depends on endometrium receptivity and is influenced by several environmental factors (Koot and Macklon, 2013; Hart, 2016). In the same way as cancer, cardiovascular and neurodegenerative diseases, discovering a contributing factor and characterizing its involvement, is a difficult undertaking because the effect of any single factor may be obscured or confounded by other contributing factors (Manolio et al., 2009). To add to this complexity, male or female gametogenesis, fertilization and implantation, are regulated by different complex biological pathways and involve many molecules as mRNAs, non-coding RNAs, and proteins. In female gametogenesis, oocyte competence develops through protracted and complex processes beginning during embryonic life and ending at the moment that the MII oocyte is ovulated (Hutt and Albertini, 2007). During embryonic life, primordial germ cells (PGCs) migrate to the genital ridge, proliferate by mitosis transforming into primary oocytes. Primary oocytes enter in meiosis and become arrested in prophase I at the diplotene stage (dyctiate). At this time the oocyte is enclosed in a specialized lineage of ovarian somatic cells, pre-granulosa cells, to form a primordial follicle (Zuccotti et al., 2011). The primordial follicle pool, produced during embryonic life, represents the woman’s ovarian reserve (Reddy et al., 2010). After puberty, some primary follicles are cyclically recruited and develop through primary, secondary and antral stages. Inside antral follicles, in response to hormonal signaling, the oocytes are stimulated to resume meiosis. Most of the antral follicles undergo apoptosis, whereas only one, the dominant Graafian follicle, ovulates to release the mature egg, ready for fertilization (Sun et al., 2009; Zuccotti et al., 2011). At this stage, the follicles consist of the germ cell in the metaphase of the second meiotic division (MII oocytes), in a fluid-filled cavity called the antrum and different layers of somatic cells, the cumulus cells (CC) surrounding the oocytes and the granulosa cells (GC) as walls of follicle (Russell and Robker, 2007; Rodgers and Irving-Rodgers, 2010). Follicle development and oocyte maturation are strictly associated, in fact, the proliferation and the differentiation of somatic follicular cells occur in synchrony with the maturing oocyte, mediated by a constant exchange of signals between somatic cells and the germ cell (Russell and Robker, 2007). The cross-talk between the oocyte and somatic follicular cells occurs by gap-junctions established between the oocyte and CCs and through the Follicular Fluid (FF) accumulated inside the antrum (Rodgers and Irving-Rodgers, 2010). FF consists of a complex mixture of nucleic acids, proteins, metabolites, and ions, which are secreted by the oocytes and somatic cells (Revelli et al., 2009). Recently, it has been demonstrated that in human FF, microRNAs (miRNAs), carried by extracellular vesicles (EVs) such as microvesicles and exosomes are present (Santonocito et al., 2014). The discovery of mechanisms of autocrine and paracrine communication mediated by miRNAs, inside the ovarian follicle, has revealed that these non-coding RNAs represent important regulators inside the pathways involved in folliculogenesis and oocyte maturation (Santonocito et al., 2014; Di Pietro, 2016). Consequently, their characterization could improve our knowledge about female gametogenesis, and could pinpoint new molecules involved in reproductive disorders allowing the formulation of new therapeutic approaches.
The aim of this paper was to identify the miRNAs in the follicular microenvironment and position them within the different compartments of the ovarian follicle identifying the pathways regulated by their targets and the long non-coding RNAs (lncRNAs) possibly involved in oocyte growth and maturation.
Discussion
Regulatory non-coding RNAs (ncRNAs) control different points of gene expression including chromatin architecture, epigenetics, transcription, RNA splicing, editing, translation and turnover. They are involved in every physiological process and consequently their sequences or expression alterations cause or contribute to different human diseases (Bartel, 2004; Croce, 2009; Fu, 2014). RNA regulatory networks include miRNAs, other classes of small regulatory RNAs and thousands of longer transcripts, named long non-coding RNAs that can be categorized in sense, antisense, bidirectional, intronic, and intergenic transcripts. Surely, the miRNA world is becoming increasingly well known to researchers, but for lncRNAs most of their functions are unknown, even if different papers have demonstrated their role in cell physiology and human pathologies. Their presence in biological fluids, as well as their expression profiles associated with specific human phenotypes, open up the possibility of using them, especially miRNAs, as molecular markers of human diseases (Weber et al., 2010; Cortez et al., 2011). Moreover, their presence in EVs, identifying miRNAs as molecular tools of communication among different cells, gives the possibility to plan specific and individualized therapies (Cortez et al., 2011; Kosaka et al., 2013; Matsui and Corey, 2016).
In reproductive biology, throughout the last decade, the role of miRNAs emerged in an important way and different studies attempted to associate specific miRNA expression profiles to oocyte quality, in granulose and cumulus cells and in FF (Li et al., 2015; McGinnis et al., 2015). Hence, great efforts were made to find promising molecular markers, in order to select the best oocytes to use in In Vitro Fertilization protocols and provide possible therapies to improve oocytes quality (Li et al., 2015; McGinnis et al., 2015).
Unfortunately, the ovarian follicle constitutes, to date, a difficult model to study for different reasons. It is made up of different cell types and represents the functional unit that ensures proper oocyte maturation by processes that begin during the embryonic stage and continue during a woman’s life until ovulation (Hutt and Albertini, 2007). The mature ovarian follicle and ovulation represent the results of different molecular processes prolonged in time and influenced by genetic background, environment and a woman’s life style. To analyze the single components of this complex unit, at the final stage, (MII oocyte, granulosa or cumulus cells, FF) does not provide all the information we need to fully understand the oogenesis and associate specific markers to specific phenotypes. Moreover, in humans, it is not possible to perform functional studies because of ethical limits.
The aim of this paper was to investigate the role of miRNAs and lncRNAs in human ovarian follicles, trying to establish, as far as possible, their potential role within the different components of the follicle.
Firstly, we identified 285 miRNAs inside human ovarian follicles (Figure 1). Interestingly, about 39% of their validated targets are expressed in different cellular components of the ovarian follicle; they are abundant in oocytes (54.2%) and predominantly involved in oocyte maturation (29.71%). 118 miRNAs (more than 40%) were shared by FF and the mature MII oocyte, while FF-miRNAs fraction was larger than the miRNAs exclusively found in oocytes. We propose that the 158 miRNAs absent in MII oocytes and exclusively present in FF have been transcribed by somatic follicular cells. Subsequently secreted in FF, these miRNAs could act as paracrine factors for the different somatic cells and regulate follicular growth. As expected, 82% of FF-miRNAs have been described in exosomes, in fact, according to previous studies, specific miRNAs are preferentially sorted into vesicles (Batagov et al., 2011). On the other hand the 9 miRNAs, absent in FF and exclusively present in the oocyte (O-miRNAs), could represent maternal RNAs, that the germ cells accumulate during their differentiation. In fact, during its growth the oocyte transcribes and stores mRNAs, miRNAs and probably long non-coding RNAs to use them during the first phase of development, before the activation of the embryo genome (Schier, 2007). Interestingly, we found miR-515-5p, miR-519c-3p, miR-520d-5p, miR-548a-3p, and miR-548c-3p specifically expressed in oocytes, and according to ExoCarta, these have not been found incorporated in exosome vesicles. Moreover, miR-515-5p, miR-519c-3p and miR-520d-5p, members of C19MC, which is a primate-specific cluster, seem to have a role in early embryo development during maternal-zygotic transition, when zygotic transcription starts and maternal mRNAs are degraded (Donker et al., 2012; Battaglia et al., 2016). Go analysis showed that development processes, regulation of cell cycle, signal transduction, and cell death represent BPs shared by oocytes and somatic follicular cells. In the same way, the p53 pathway is significant in both cell types. This confirmed our knowledge about the processes involved in oocyte maturation and carried out by both compartments (Hutt and Albertini, 2007). Apoptosis has been amply described in the mammalian ovary (Tilly, 2001). Cell death by apoptosis affects about 99% of primordial follicles present at birth in mammalian ovaries (Tiwari et al., 2015). The production of the mature oocyte ready for fertilization is a highly selective process: only a few follicles in the human ovary survive to complete their growth, only fully competent oocytes will be ovulated and only an embryo without major genetic alterations will be capable of uterine implantation. Apoptosis of granulosa cells reduces the number of recruited follicles and apoptotic machinery in MII oocytes selects viable embryos (Tilly, 2001; Santonocito et al., 2014). Interestingly, we found different BPs involved in the response to exogenous stimuli, significant only for O-miRNAs (Figure 2C). Among the different cells of the ovarian follicle the germ cell is unique: it must be able to respond and react to external stimuli more efficiently than somatic cells. It has been described that the maturing oocyte and early embryo are quite sensitive to exogenous stresses. Oocytes and early embryos can undergo physiological adaptations to environmental perturbations; these adaptations could influence the embryo genome signature involving epigenetic modification (Latham, 2015).
As concerns the 118 shared miRNAs, not being able to pinpoint the cells producing them, we can assert that these miRNAs could mediate the communication between the oocyte and somatic cells. Eleven of them, more abundant in oocytes, could have been transcribed by germ cells and used as signaling for cumulus and granulose cells. This hypothesis is supported by the finding that miR-518d-3p, miR-518f-3p, and miR-523-3p, up-regulated in oocytes, are members of the C19MC cluster, in the same way of some O-miRNAs. These miRNAs could be transcribed together, some of them stored in oocytes, others packaged in vesicles and secreted in FF. A similar consideration can be made for miR-372 (up-regulated in oocytes) and miR-371 (O-miRNA). On the contrary, the miRNAs up-regulated in FF, especially let-7b-5p, miR-15b-5p, miR-24-3p, miR-130b-3p, miR-146b-5p, miR-212-3p, miR-222-3p, miR-223-3p, miR-339-3p and miR-483-5p, with fold change values higher than 100-fold, could be transcribed in somatic follicular cells and move to oocytes by means of exosomes. Enrichment analysis of validated target genes of Common-miRNAs showed a strong correlation with the maintenance of the primordial follicle quiescent stage, oocyte maturation, oocyte meiosis, development, cancer and stem cell related pathways. Moreover, a high representation of apoptosis signaling pathway, hypoxia response via HIF activation and oxidative stress response have been detected (Figure 3C).
To understand miRNA-mediated gene regulation, we investigated if other classes of non-coding RNAs play a role inside the ovarian follicle. LncRNAs can act as miRNA sponges, reducing their regulatory effect on mRNAs introducing an extra layer of complexity in the miRNA-target interaction network (Paraskevopoulou and Hatzigeorgiou, 2016). Moreover, the important role of lncRNAs in chromatin remodeling is well known, as well as the importance of this process in oocyte maturation, when, before meiosis resumption, transcriptional silencing is mediated, over all, by mechanisms involved in large-scale chromatin structure changes. The cellular and molecular pathways involved in these processes are today poorly understood and lncRNAs could play a major role (De La Fuente et al., 2004). Recently, lncRNAs were identified in granulosa and cumulus cells, oocytes and early embryos, and their role in oocyte and early embryo development has been suggested (Yan et al., 2013; Yerushalmi et al., 2014; Hamazaki et al., 2015; Xu et al., 2015). A comprehensive review on lncRNA functions in mammalian and in different species has been recently published (Taylor et al., 2015).
By using bioinformatic prediction we found 41 lncRNAs significantly correlated with 9 oocyte miRNAs (Table 1). Most of them are expressed in embryonic cells, in the ovary or in the placenta and are involved in several human pathologies. NEAT1 was found exclusively localized in paraspeckles and is a core component of these nuclear bodies involved in nuclear retention of mRNAs (Bond and Fox, 2009). Moreover, NEAT1 seems to be essential for the formation of the corpus luteum and the establishment of pregnancy in mice, although its precise molecular mechanism remains to be investigated (Nakagawa et al., 2014). Recently, increased levels of NEAT1 was found associated with placental dysfunction in Idiopathic Intrauterine Growth Restriction (IUGR) fetuses (Gremlich et al., 2014). Another lincRNA that was found to reside predominantly in the nucleus is the Metastasis associated lung adenocarcinoma transcript-1 (MALAT-1) that localizes to nuclear bodies known as nuclear speckles involved in pre-mRNA splicing (Lennox and Behlke, 2016). MALAT-1 down-regulation was described to impair proliferation, cell cycle, apoptosis, and migration of trophoblast cells involved in the preeclampsia (Chen et al., 2015). Another human lincRNA, the non-coding RNA activated by DNA damage (NORAD) is induced after DNA damage in a p53-dependent manner and plays a crucial role in maintaining genomic stability by sequestering PUMILIO proteins, which repress the stability and translation of mRNAs to which they bind (Lee et al., 2016). In view of the location and the function of these 3 lncRNAs in regulation of mRNA stability and translation, we suppose that in oocytes these could stabilize maternal RNAs, allowing their storage and use during early development, before the activation of embryo genome, as it has been described for cytoplasmic polyadenylation (Reyes and Ross, 2016). Not surprisingly, we found the X-inactive specific transcript (XIST) and the growth arrest specific 5 (GAS5), two lncRNAs involved in embryogenesis. It has been demonstrated that many aspects of embryogenesis seem to be controlled by ncRNAs, including the maternal–zygotic transition, the maintenance of pluripotency, the patterning of the body axes, the specification and differentiation of cell types and the morphogenesis of organs (Pauli et al., 2011). XIST, responsible for the mammalian X chromosome inactivation, is the first lncRNA expressed starting at the 4-cell stage of human preimplantation embryos, consistent with embryonic genome activation, and iPSC reprogramming (Briggs et al., 2015).
Encoded within introns GAS5 increases OCT4, NANOG and SOX2 by Nodal regulation and is directly regulated by these stemness factors in hESCs forming a circuit that promotes pluripotency (Xu et al., 2016). OIP5-AS1 is an antisense transcript of the Opa interacting protein 5 (OIP5) gene. The protein encoded by this gene localizes to centromeres, where it is essential for recruitment of CENP-A and it is required for centromeric heterochromatin organization. Expression of this gene is upregulated in several cancers, making it a putative therapeutic target.
Recently, a new regulatory circuitry in which RNAs can crosstalk with each other and modulate the biological function of miRNAs has been proposed (Cesana et al., 2011). LncRNAs that localize primarily in the nucleus (e.g., XIST, NEAT1, MALAT-1) have been described to physically interact with mature miRNAs (Leucci et al., 2013; Gernapudi et al., 2015; Yu et al., 2017). Several observations have shown the presence of mature miRNAs in the nucleus (Hwang et al., 2007; Rasko and Wong, 2017). In facts, miRNAs can be transported from the cytoplasm to the nucleus and act in an unconventional manner to regulate the biogenesis and functions of ncRNAs (Liang et al., 2013).
Among the identified lncRNAs in oocytes, NEAT1, MALAT-1, GAS5, XIST and OIP5-AS1 have been predicted as components of the same network (Figure 4). Even if the mechanisms of most of these lncRNAs remain unknown, and it remains to be seen whether they can function within human ovarian follicle, these putative interactions lead us to hypothesize a possible role inside the female human germ cell.