The role of miRNAs in polycystic ovary syndrome with insulin resistance

Abstract

Purpose

This review aims to summarize the key findings of several miRNAs and their roles in polycystic ovary syndrome with insulin resistance, characterize the disease pathogenesis, and establish a new theoretical basis for diagnosing, treating, and preventing polycystic ovary syndrome.

Methods

Relevant scientific literature was covered from 1992 to 2020 by searching the PubMed database with search terms: insulin/insulin resistance, polycystic ovary syndrome, microRNAs, and metabolic diseases. References of relevant studies were cross-checked.

Results

The related miRNAs (including differentially expressed miRNAs) and their roles in pathogenesis, and possible therapeutic targets and pathways, are discussed, highlighting controversies and offering thoughts for future directions.

Conclusion

We found abundant evidence on the role of differentially expressed miRNAs with its related phenotypes in PCOS. Considering the essential role of insulin resistance in the pathogenesis of PCOS, the alterations of associated miRNAs need more research attention. We speculate that race/ethnicity or PCOS phenotype and differences in methodological differences might lead to inconsistencies in research findings; thus, several miRNA profiles need to be investigated further to qualify for the potential therapeutic targets for PCOS-IR.

Introduction

Polycystic ovary syndrome (PCOS) is a reproductive endocrine disorder, which affects 12% of women at childbearing age [1]. The main characteristics include polycystic ovaries, hyperandrogenism, chronic anovulation, and insulin resistance (IR), despite their highly heterogeneous features [2]. PCOS is associated with various diseases, including anovulatory infertility [3] and metabolic disorders such as metabolic syndrome, type 2 diabetes, and cardiovascular disease in women [4, 5]. The etiology of PCOS may include genetic [6], environmental, and epigenetic factors [7]. IR refers to the decreased sensitivity of peripheral tissue to insulin and thus impairing glucose uptake and production. The insufficient glucose consumption results in blood glucose level upregulation and stimulates the pancreatic β-cells of the islet to secret excessive insulin to lower blood glucose [8], which leads to hyperinsulinism. More than 60% of PCOS patients have varying degrees of IR [9], and their insulin intake efficiency is 20%–40% lower than that in healthy people [10]. IR and compensatory hyperinsulinism drive multiple phenotypic characteristics in PCOS [2], mainly because insulin facilitates androgen secretion predominately by increasing the synthesis and release of ovarian androgens [2, 11] and modulating luteinizing hormone secretion from the pituitary [12]. Thus, hyperinsulinism contributes to clinical manifestations, including ovulatory disturbances [13], menstrual irregularity, anovulatory infertility, and immature follicles increase [14] in PCOS. Clinically, IR is mostly detected and assessed by the homeostasis model assessment of insulin resistance (HOMA-IR) in vivo, and the precision is equivalent to the glucose clamp technique. However, there is great variability in the threshold HOMA-IR levels to define IR [15]. Research showed that only 49.6% of PCOS patients were diagnosed as IR; however, 22.6% and 15.8% of patients were diagnosed as IR when the blood glucose cutoff value was set to 3.46 and 3.8 using HOMA-IR [16]. Another study found that the HOMA-IR index increased with age, stabilized between 13 and 15 years old, and then began to decline year by year among Caucasians [17]. Therefore, the diagnosis of PCOS-IR by HOMA-IR is affected by demographics and setting different cutoff values may generate different results. We recognize the potential validity issue due to the judgment criteria. However, for this review, studies have set similar HOMA-IR threshold index values (approximately between 2.2 and 2.75). Thus, we assume that the heterogeneity of the population does not affect the validity of this review.

miRNAs are small non-coding, single-stranded RNAs with 19–25 nucleotides in length. As post-transcriptional regulators of gene expression, miRNA is involved in suppressing mRNA expression and inhibiting protein translation by binding to the 3′ UTR region of the target mRNAs [18]. Also, miRNAs may influence the expression and alter their downstream target gene [19] and affect cell growth, differentiation, proliferation and apoptosis, inflammation, stress response [20], and malignant tumor metastasis. Many studies have shown that alteration of miRNAs may contribute to the onset of cervical cancer [21], endometriosis [22], poor ovarian response, and cardiovascular diseases [23, 24], and thus can serve as potential diagnostic and monitoring biomarkers in various diseases [25], including diabetes [26] and PCOS [27]. Among PCOS patients, several miRNAs are differentially expressed in serum [28], adipose tissue (AT) [29], follicular fluid (FF), granulosa cells (GCs) [30, 31], and cumulus cells (CCs). Certain miRNAs may be new biomarkers for abnormal metabolism, compromised oocyte quality, and low endometrial receptivity in PCOS [32].

The etiologies of miRNAs alteration in PCOS are not well defined yet. Most research has focused on the dysregulation of miRNAs in PCOS and healthy people [33, 34] and has shown that miRNAs might play an important role in the occurrence and development of PCOS-IR and PCOS non-insulin resistant (PCOS-NIR) [35, 36]. Some studies have found the key differences between PCOS-IR and PCOS-NIR in various metabolic indexes, including significant differences in molecules involved in insulin signaling, IGF system, lipid metabolism, and steroidogenic signaling such as PPAR-γ, CYP19A1, and IRS [37]. In addition, assisted reproductive technology (ART) outcomes were compared between PCOS-IR and PCOS-NIR cohort. Studies found that PCOS-IR patients have lower pregnancy rate, number of oocytes retrieved [38], ovulation rate, pregnancy rate, and worse outcome in ovulation induction therapy. Besides, PCOS-IR patients are associated with increased pregnancy complications, including gestational diabetes and pregnancy hypertension, and adverse pregnancy outcomes including miscarriage and premature births [39]. The aim of this paper is to review relevant literature to find the crucial miRNAs in the pathogenesis of PCOS-IR and explore the possible mechanisms in the insulin signaling pathways (Fig. 1). This review highlights the role of miRNAs in PCOS and might provide novel insights on biomarkers and treatments for PCOS-IR.

Fig. 1.

The roles of miRNAs in the insulin signaling pathways and insulin resistance in PCOS

Alterations of miRNAs expression of PCOS-IR in different tissues

miRNA is highly complex and is associated with insulin resistance in PCOS patients. Therefore, we looked into the differences in the expression of miRNAs between PCOS-IR and PCOS-NIR, especially in different tissues. With the development of gene sequencing technology, many researchers have carried out miRNA sequencing studies on PCOS with various samples such as follicular fluid, plasma, granulosa cells, and cumulus cells. Since the miRNAs produced at a peripheral tissue are not all be released into the blood, determining the tissue difference of miRNA expression level will indicate the tissue-specific accumulation and reflect the reliability sample collection to get accurate results.

Differential expression of miRNAs in serum

miRNAs are abundant and stable in human serum [40]. Since serum miRNAs are easy to be detected, potentially disease-specific, and have little individual expression differences, they attracted most research attention and became ideal biomarkers [41]. A study found that miR-222, miR-146a, and miR-30c were differentially expressed in the serum of five women with PCOS and five healthy people [28]. Among them, miR-222 was positively associated with serum insulin. Of note, miR-222 was significantly increased in type 2 diabetes and may be related to insulin sensitivity [42]. miR-146a was negatively associated with serum testosterone, and it was also reported to inhibit the release of progesterone, androgens, and estrogens. It is speculated that miR-146a may act as a physiological inhibitor of general secretory activity [43]. In addition, bioinformatics analyses indicate that genes targeted by miR-222, miR-146a, and miR-30 are involved in the cell cycle, apoptosis, and endocrine pathways that participate in the pathogenesis of PCOS via insulin-related signaling pathways, including Wnt signaling pathway and Jak-STAT signaling pathway [28]. A study revealed that the serum levels of miR-24 and miR-29a were lowered in PCOS subjects and could be of diagnostic value [44]. Another study [45] found that miR-24 level in the plasma of type 2 diabetic patients was significantly lower than that in healthy controls and the insulin promoter activity was found to be downregulated by miR-24, thereby resulting in the reduction in insulin mRNA levels. Androgen receptor (AR), StAR-related lipid transfer protein3 (STARD3), and insulin receptor substrate-1 (IRS-1) are reported as targets of miR-29a. Among these, AR and STARD3 are involved in ovarian steroidogenesis, while IRS-1 is involved in insulin signaling.

Jiang et al. found 748 miRNAs expression profiles of three pooled samples (10 PCOS-NGT samples, 10 PCOS-IGT samples, and 10 controls) from the serum of 175 women, among which three miRNAs (miR-122, miR-193b, miR-194) were differentially expressed between PCOS-IGT (impaired glucose tolerance) and PCOS-NGT (normal glucose tolerance) samples [46]. Wang et al. demonstrated that miR-122 was a potential biomarker of obesity and IR [47]. miR-122 and miR-193b expression levels were significantly higher in diabetic cats compared with those of healthy cats [46]. Overexpressed miR-194 in the muscle and liver elevated fasting glucose levels in rats by targeting the IGF-1 receptor [31], suggesting its functions in the regulation of glucose metabolism. Another study [48] detected the expression of miR-204 and analyzed the diagnostic efficacy in the serum of between PCOS-IR patients and PCOS-NIR patients and found that both in vitro and in vivo experiments supported the capabilities of miR-204 and HMGB1 in proliferation and apoptosis of PCOS-IR GCs. In addition, it was revealed that overexpression of miR-204 and downregulation of HMGB1 inhibited TLR4/NF-kB pathway activation, decrease IR and T, and increase ovarian coefficient. It has been suggested previously that overexpression of miR-204 restrains the JAK2/STAT3 pathway [49].

Differential expression of miRNAs in follicular fluid

Follicular fluid provides an appropriate intrafollicular environment for oocytes maturation and development, which comprises the secretory and metabolic activity product of the oocyte, GCs, theca cells (TCs), and the transfer of blood plasma components. Therefore, the analysis of FF ingredients may also reflect metabolic and functional changes in the oocyte and GCs [50]. Oocytes are deemed as high quality when they undergo meiosis and achieve fertilization, followed by embryogenesis and development. Various pathological conditions that disrupt the intrafollicular environment may affect the quality of oocytes, the development of the early embryo, and the outcome of the pregnancy [50, 51]. A study has shown that there are abnormalities in the components of the FF in PCOS patients. Furthermore, alterations in blood circulation biochemical components may suggest changes in the composition of FF.

Roth et al. found that 5 miRNAs (32, 34c, 135a, 18b, and 9) showed significantly increased expression in the PCOS group, and pathway analysis revealed 3 potential genes (IL-8, SYT1, and IRS-2) involved in insulin regulation and inflammation [52]. IRS-2 mediates the effects of insulin and IGF-1 and is critical for peripheral carbohydrate metabolism and beta-cell function. Mice lacking IRS-2 developed estrous cycle disorders, anovulation, infertility, and IR, similar to women with PCOS [53]. SYT1 is a gene encoding synaptogamin1 and a calcium-binding synaptic vesicle protein [54], and it plays a critical role in calcium-triggered exocytosis throughout the body and is required for exocytosis of insulin in the pancreatic beta-cells. Compared to the healthy group, miR-1290 was one of the highest expressed miRNAs in FF of PCOS [55] and also highly expressed in gestational diabetes [56]. Sorensen et al. also found that the increased miR-1290 is associated with either aberrant glucose homeostasis or dyslipidemia [57], indicating that high levels of miR-1290 may impact glucose and insulin metabolism and act on PCOS pathophysiology. In other studies, the expression of miR-320 is different in the FF with PCOS women [55, 58], indicating that miR-320 might serve as a potential target to improve insulin sensitivity and IR. FF is easily obtained from in vitro fertilization and oocyte harvesting procedures, which becomes an ideal setting to investigate the oocyte quality, maturation, and fertility outcomes; thus, all the above miRNAs have the potential to be developed as new biomarkers for PCOS-IR.

Differential expression of miRNAs in other tissues

IR presents ubiquitously throughout the body and is associated with not only PCOS but also other health issues, including obesity and cardiovascular problems. Several studies revealed the differential expression of miRNAs in other tissues related to IR. In adipocytes of PCOS, the upregulation of miRNA-93 and miRNA-223 downregulated GLUT4 gene expression, which showed that miRNA-93 and miRNA-223 might be associated with IR in PCOS [59]. Subsequently, miRNA-93 in AT correlated with PCOS-IR though its host gene MCM7 was discordantly downregulated [29]. In plasma of PCOS, miRNA-93 and miRNA-223 expressions were also increased, but miRNA levels were independent of insulin and IR [60]. In ovarian theca tissues, the downregulation of miR-92a and miR-92b might lead to augmentation of signal transduction in the insulin pathway between PCOS patients and controls [36]. In the theca cells, Lin et al. reported that the expression of miR-92a and miR-92b was significantly downregulated with PCOS. In another study, GATA6 and IRS-2 exhibited significantly higher PCOS TCs, which is in agreement with Lin et al.’s findings. Also, GATA6 was predicted as a target gene of miR-92a and IRS-2 was regulated by both miR-92a and miR-92b [61]. In ovaries from PCOS-IR rats [62], miR-30 was significantly upregulated, which is consistent with previous studies [28, 35]. However, miR-146 was significantly upregulated in serum from PCOS patients compared with that of healthy women, which is inconsistent with the results in rats [63]. This difference may be explained by the differences of species, natural heterogeneity of PCOS, and/or the different sequencing method employed.

Common miRNAs in different tissues in PCOS-IR

Although the miRNAs in different tissues are diverse in PCOS-IR, there are some commonly expressed miRNAs in different tissues. miR-320a is an inhibitor related to various kinds of tumors, as its abnormal expression is mostly associated with malignant tumors [64]. In the female reproductive system, the expression level of miR-320a is closely related to the development of oocytes, and maintaining a normal expression level is necessary for oocyte development. The research reported that mature oocytes were associated with a higher level of miR-320a [65]. However, miR-320a was expressed at significantly lower levels in serum and FF of PCOS-IR patients than in healthy controls [55]. Similarly, the serum level of miRNA-320 was also reduced in the diabetic population [66] and adipocytes [67], and studies have speculated that miR-320a may reduce insulin sensitivity through the insulin PI3K signaling pathway [68, 69]; therefore, miRNA-320 plays a key role in regulating IR. RAB5B, identified as a PCOS candidate gene in a genome-wide association study, is predicted to be the target gene of miRNA-320. We may postulate that lower miRNA-320 expression levels in the FF of PCOS patients may affect RAB5B gene expression. miR-320 can also reduce the expression of IRS and inhibit the phosphorylation of extracellular-regulated protein kinases (ERK1/2) pathway which may lead to IR [70]. However, the expression of miR-320 was unregulated in AT of PCOS-IR [71]. Some studies have found different miRNA expression results. The research reported the upregulation of miRNA-320 in FF, potentially explaining the hyperandrogenemia found in women with PCOS through downregulation of E2F1/SF-1 proteins, which in turn cause the inhibition of estradiol releasing into FF [58]. Another study showed that there is no differential expression of either miR-132 or 320 in FF of PCOS patients in the USA [52]. Of note, serum miRNA-132 levels are significantly decreased in gestational diabetes mellitus suggesting that it may regulate insulin secretion [72]. This may be related to race/ethnicity or PCOS phenotype differences as Sang et al.’s study [55] population was all Asian, and they used genome-wide analysis in addition to microarray analysis to identify miRNAs, so methodological differences can also explain the differences.

miR-145 is elevated in FF, while their equivalents in GCs showed diminished expression in recent research [73]. In GCs of mice, miR-145 was found to be in the suppression of proliferation [74]. The decreased level of miR-145 in GCs of PCOS patients and its contribution to negative regulation of proliferation via targeting IRS1 have been studied, and miR-145 overexpression was found to suppress cell survival rate and DNA synthesis levels in human GCs [75]. In the ovarian cortex, miR-483 was downregulated in PCOS patients and dual-luciferase reporter assay showed IGF1 was a direct target for miR-483. Insulin treatment in KGN cells inhibits miR-483, promotes IGF1, and induces cell proliferation. miR-483-3p was confirmed to inhibit KGN cell proliferation and arrest cell cycle, possibly via suppressing IGF1, suggesting the potential to use miR-483 as a biomarker and therapeutic alternative for diagnosing and treating PCOS [76]. In CCs, qRT-PCR (quantitative real-time PCR) validated that miR-483–5p was downregulated in patients with PCOS [77]. miR-483–5p is a conserved sequence encoded in the second intron of the IGF2 gene [78]. IGF2, as a miR-483–5p co-expressed gene [79], was significantly downregulated in CCs with PCOS, which may be also one of the pathogenic factors for PCOS. In another study, miR-483-5p was identified to regulate the Notch and MAPK signaling pathway. miR-483-5p expression was increased in PCOS CCs, whereas the expression of MAPK3 mRNA and protein was decreased as shown in deep sequencing and qRT-PCR. Moreover, miR-483-5p regulated MAPK3 expression by directly binding 3′ UTR of mRNA based on the bioinformatics prediction and experimental validation [30]. A study has found that miR-483-5p can promote the proliferation of GCs in PCOS and play an important role in reducing IR [77].

miRNAs and glucose metabolism in PCOS-IR

Many differentially expressed miRNAs in PCOS women are involved in pathological insulin signaling processes, the disorder secretion of inflammatory factors, and hormones [80]. Hyperinsulinism may further contribute to the PCOS pathology. Nowadays, the relationships among miRNAs, PCOS, and IR have been assessed at transcript, genomic, and metabolic pathway levels. miRNAs participate in the insulin signaling process through activation of the insulin receptor which increases phosphorylation of intracellular substrates (IRS), principally related IRS proteins (IRS1, IRS2). For example, miR-145 directly inhibited IRS1 expression by binding to its 3′ UTR and the IRS1 overexpression abrogates the suppressive effect of miR-145 mimics. The overexpression of miR-145 inhibited MAKP/ERK signaling pathways in GCs. A high concentration of insulin can reduce miR-145 expression, upregulate IRS1 protein expression, and promote cell proliferation [75]. These studies indicate that the expression of miR-145 may not only inhibit GC proliferation but also improve IR in PCOS patients. In addition, miR-370 is a scaffold protein involved in the insulin pathway modulating IRS1 expression [81] and the downregulation of miR-370 may target IRS1 and disturb insulin metabolism of GCs and affect oocyte quality [82, 83]. Table 1 shows the literature that summarized the expression and function of miRNAs among PCOS-IR patients.

Table 1.

Related microRNAs observed in polycystic ovary syndrome (PCOS) with insulin resistance

miRNA	Species	Samples	Criterions of IR	Expression	Targets and pathways	Proposed functions	Ref.
miR-204	Human	Serum	HOMA-IR > 2.57	Decreased	HMGB1; TLR4/NF-KB pathway	Reduces IR, boosts proliferation, restrains apoptosis	[48]
miR-612	Human	Cumulus cells	HOMA-IR > 2.2	Decreased	Rap1b; MAPK pathway	Involved in pathophysiology	[84]
miR-93	Human	Adipose tissue	HOMA-IR ≥ 2.5	Increased	Glut4; IRS1/PI3K/AKT pathway	Related to AT metabolism	[59]
	Human	Serum	-	Increased	—	Can be a biomarker potentially	[60]
miR-223	Human	Adipose tissue	HOMA-IR ≥ 2.5	Increased	Glut4	Inhibits Glut4 expression and induces IR	[85]
miR-33b-5p	Rat	Ovarian tissue	HOMA-IR ≥ 2.5	Increased	Hmga2; PI3K/Akt	Inhibits GLUT4 production by targeting HMGA2	[86]
miR-92a	Human	Ovarian tissues	HOMA-IR > 2.14	Decreased	GATA6, IRS-2	Related to the pathogenesis of PCOS	[36]
miR25/93/106b	Human	Adipose tissue	HOMA-IR ≥ 2.5	Increased	MCM7	May not be regulated by target gene MCM7	[29]
miR-485-3p	Human	Skeletal muscle	Hyperinsulinemic-euglycemic clamp	Increased	PGC-1α	Inhibits PGC-1α and impart IR in PCOS	[87, 88]
miR-93	Human	Ovary cortex	HOMA-IR ≥ 2.14	Increased	CDKN1A	Promotes cell proliferation	[89]
miR-451a	Human	Serum	—	Decreased	PI3K/AKT pathway	A novel candidate biomarker for PCOS diagnosis	[90]
miR-483-5p	Human	Cumulus cells	—	Decreased	IGF2	Reduces IR in PCOS	[77]
miR-486-5p	Human	Cumulus cells	—	Decreased	PTEN, DOCK3; PI3K/Akt pathway	Promotes the proliferation of CCs	[77]
miR-323-3p	Human	Cumulus cells	—	Decreased	IGF1; Wnt and TGF-β pathway	Inhibits steroidogenesis and CCs apoptosis	[91]
	Human	Follicular fluid		Not detected			[55]
miR-222	Human	Serum	—	Increased	—	Correlated to serum insulin in PCOS	[28]
miR-378a-5p	Human	Serum	—	Increased	PGC-1β, PI3K pathway	Acts between PCOS-HA and IR	[57, 92]
miR-15b	Human	Adipose tissue	—	Decreased	Dicer	Be important in adipose differentiation	[93]
miR-320	Human	Follicular fluid	—	Decreased	RAB5B	Regulates IR in PCOS women	[55]
miR-320a	Human	Ovarian tissues	—	Increased	PCGF1	Reduces KGN cell proliferation and apoptosis	[94]
	Human	Follicular fluid		Increased	E2F1/SF-1	Causes inhibition of estradiol release into FF	[58]
miR-193b	Human	Serum	—	Increased	—	Correlated with HOMA-IR	[46]
miR-122	Human	Serum	—	Increased	—	A new biomarker of obesity and IR possibly	[46, 47]
miR-194	Human	Serum	—	Increased	IGF-1	Regulates glucose metabolism	[46, 95]
miR-145	Human	Granulosa cells	—	Decreased	IRS-1; MAPK/ERK pathway	Improve IR in PCOS	[75]
	Human	Follicular fluid		Increased	—	Involved in pathogenesis of PCOS	[31]
miR-27a-3p	Human	Granulosa cells	—	Decreased	STAT1, STAT3; SMAD5	Decreases proliferation and promotes apoptosis	[96]
miR-140	Human	Granulosa cells	—	Increased	RAP2A; RAP2A/AKT pathway	Increases proliferation and reduces apoptosis	[97]
miR-483	Human	Ovary cortex	—	Decreased	IGF1	Inhibits KGN cell proliferation	[76]
miR-9119	Rat	Granulosa cells	—	Decreased	p65, DICER; NF-κB/p65 pathway	Inhibits cell proliferation and promotes apoptosis	[98]

Glucose enters the cell membrane mainly through glucose transporter 4 (GLUT4) [99], which is an insulin-stimulated glucose transporter and one of the most important candidate genes for IR. The expression level of GLUT4 is positively correlated with glucose translocation [100] tolerance as well as insulin sensitivity [101] and its lower expression will impair the glucose metabolism process [102]. The expression of GLUT4 is detected in AT from PCOS-IR, PCOS, and PCOS-NIR patients. Compared with the healthy control group, the upregulation of miR-93 decreased the expression of GLUT4 and impaired glucose transmembrane transport in AT of PCOS patients, which suggested that miR-93 plays a major part in PCOS-IR [59]. In addition, miR-93 potentially regulates signaling pathways of hepatic growth factor (HGF) detected in the ovary [103] and increases insulin secretion by increasing the amount of islet beta-cells through compensatory mechanisms and thus leads to IR. HGF is an adipokine and related to obesity and metabolic syndrome [104–106]. Another study found that miR-33b-5p targeted HMGA2 and SREBF1 to inhibit GLUT4 production and resulted in IR of PCOS [86]. We can speculate that certain miRNA might affect glucose metabolism and lead to IR in PCOS by regulating the expression of GLUT4.

Insulin-like growth factor 1 (IGF-1) is a peptide that is closely related to carbohydrate metabolic disturbance, including IR, type 2 diabetes, high blood pressure, obesity, and metabolic syndrome [107]. A study found that miR-122 mimic directly targeted the 3′ UTR of IGF-1 mRNA to significantly downregulate IGF-1 expression levels, suggesting that miR-122 can inhibit IGF-1 expression and decrease insulin sensitivity [108]. Four highly expressed genes (SOCS3 [109], SRF [110], FOXO1 [111], and PTEN [112]) were targeted by miR-483-5p, which is associated with IR and glucose intolerance in CCs of PCOS patients [77]. In addition, miR-485-3p can target peroxisome proliferator-activated receptor γ coactivator 1α (PGC-1α) [87], which is a mediator of insulin signaling and maintains the function and integrity of mitochondria [113] and has lower expression levels in PCOS women [88]. The underlying mechanism is still unclear.

miRNAs and lipid metabolism in PCOS-IR

Besides endogenous IR, about 70% of PCOS patients have weight-related IR with dysfunctional and impaired glucose metabolism of AT. The lipid level can increase with the increase of glucose concentration in PCOS patients [114]. Legro et al. reported that the incidence of IR in PCOS obese patients was significantly higher than that in PCOS non-obese patients [115]. Adipocytes with IR have differentially expressed miRNAs [116], and several miRNAs are related to carbohydrate metabolism in AT. Diaz et al. found that miR-451a was the lowest level of miRNAs in PCOS compared with healthy people [90] and suppressed hepatic gluconeogenesis by downregulating glycerin kinase [117], which is considered a key driver of PCOS. It may target calcium-binding protein 39 (CAB39) to interfere with the PI3K/AKT signaling pathway in vivo [118] and lead to the suppression of the PI3K/AKT/mTOR pathway that is associated with diabetes, obesity, high blood pressure, and cancer [119–121]. The overexpression of miR-451a can also reduce triglyceride accumulation in mouse liver and cultured hepatocytes [122]. However, there is no sufficient evidence illustrating the function of miR-451a in PCOS-IR at present, which indicates a new research direction. Research showed that AT-specific GLUT4 could regulate tissue glucose tolerance, insulin sensitivity, and systemic glucose metabolism [101, 102]. In 3T3-L1 adipocytes, along with the increased expression levels of miR-320, IR was induced with high levels of glucose and insulin, and when adding anti-miR-320 oligos, GLUT4 expression was upregulated, and IR was brought back to normal [67]. Since miR-320 is abundant in FF of PCOS women [55], it can serve as a potential target to improve insulin sensitivity and IR. Although miR-223 can regulate the expression of GLUT4 in cardiomyocytes [93], there is no significant association among miR-223, GLUT4, and the evaluation of homeostasis model assessment of IR (HOMA-IR) in PCOS-AT [59]. When compared to the control group, miR-146a and miR-155 are shown to be differently expressed in GCs from the PCOS women [123]. miR-146a is involved in IR with higher expression under inflammatory stimulation in type 2 diabetic patients [124, 125], indicating the overexpressed miR-146a may contribute to inflammation and aggravate IR in PCOS. Gonzalez et al. found that a pro-oxidative and pro-inflammatory state caused by saturated fat intake may be an important factor for IR in PCOS [126]. miR-155 takes effect in autoimmune diseases and chronic inflammatory diseases [127–129] and regulates insulin sensitivity in both in vivo and in vitro mouse models [130, 131]. In addition, a study showed that the combined application of miR-193b and BMI might potentially be a new indicator to predict the incidence of impaired glucose metabolism in patients with PCOS [46]. More strong evidence demonstrating this relationship is that metformin can normalize the phenotype of PCOS more than oral contraceptive (OC) by reducing hepato-visceral adiposity and inducing IR [132]. However, no research has looked into the change in miRNA expression after metformin treatment among PCOS patients [133].

miRNAs and hyperandrogenemia in PCOS-IR

Hyperandrogenemia is a prominent pathological feature and clinical manifestation of PCOS and characterized by increased testosterone levels, which is associated with the abnormal distribution of abdominal fat [134], IR [135], and ovulatory dysfunction [136] in PCOS. Insulin stimulates androgen production by activating 17-hydroxylase/C17-20 lyase cytochrome P50 (CYP17) enzyme, decreasing the production of SHBG, increasing cAMP concentrations by working in coordination with LH, and resulting in StAR activation and androgen production [137, 138]. Hyperinsulinemia may lead to the onset of hyperandrogenemia in PCOS. In FF among PCOS patients, a study provided evidence that the miRNA expression profile was altered in three groups (PCOS-normo-androgenic, PCOS-hyperandrogenic, and healthy group) and specific miRNAs were associated with phenotypical traits of PCOS [27]. Further research showed that 750 miRNAs were differentially expressed in the three groups using TaqMan low-density arrays, and hyperandrogenic PCOS women showed more severe IR compared to normo-androgenic PCOS women and had a higher prevalence of metabolic syndrome [57]. In ovarian theca cells, a study on the miRNA profile in PCOS ovarian theca tissue showed that miR-92a might target IRS-2 and GATA6 [36] indicating cross-talk relationships between androgenic and insulin signaling pathways. However, in a case-control study, PCOS patients with elevated insulin and testosterone had higher levels of miR-93 and miR-223 in circulation, which were independent of IR or HA when compared to age- and weight-matched controls [60]. miRNAs in PCOS-IR and hyperandrogenemia might provide a new direction to understand the pathology and thus find potential therapeutic targets but requires further study.

miRNAs and endometrial receptivity in PCOS-IR

Among patients who undertake ART treatment, up to 67% with recurrent implantation failure is associated with poor endometrial receptivity (ER), which is closely regulated by multiple genes, signal transduction pathways, and cell-cell interactions [139]. A study showed that oocytes matured in vitro from PCOS-IR did not affect oocyte and embryo quality but did lower the pregnancy rate; thus, we may infer that IR and hyperinsulinemia can injure endometrial receptivity and disturb the embryo implantation process [140]. Romina et al. found that the expression of IRS-1, IRS-2, and the degree of tyrosine phosphorylation in endometrial tissues of PCOS-IR patients were lower than those of PCOS-NIR patients, suggesting that insulin signaling molecules may alter downstream kinase activity and trigger endometrial IR [141]. Compared with PCOS-NIR patients, cortisol was increased and inhibited the phosphorylation of protein kinase B (AKT) and GLUT-4 translocation by inducing phosphatase and leading to PCOS-IR in the endometrium [142]. Furthermore, ER chips showed that miRNAs could mediate multiple genes to complete the recognition at the maternal-fetal interface [143], for example, miR-323-3p is supposed to be a critical region for placental growth and embryonic development [144]. There are 44 differentially expressed miRNAs of endometrial stromal cells in vitro decidualization [145], and several differentially expressed miRNAs were associated with endometrial status in healthy women in the pre-implantation and implantation phases, late proliferate phase, and mid-secretion phases [146] in a previous study. Specifically, 148 miRNAs were differentially expressed in the uterus from the PCOS-IR rat model compared with the control [147]. Bioinformatic analysis and pathway analysis showed four miRNAs (− 484, − 375-3p, − 324-5p, and − 223-3p) and their target genes were involved in insulin secretion and signaling pathways, such as Wnt, AMPK, PI3K-Akt, and Ras. IGF and its related proteins were confirmed to participate in the physiological process of the endometrium by increasing apical fibronectin on blastocysts to increase attachment and invasion in an in vitro model of implantation [148]. Decidual cells of primates can secret IGF and possibly affect embryo implantation in the invasion stage of trophoblast cells [149] by targeting the androgen or IGF1 signaling pathway [36]. Investigating the alteration of the miRNA profile provides a better understanding of the mechanisms for poor ER in PCOS-IR; however, most of the studies are based on animal models, and few studies on miRNAs were conducted in human subjects, suggesting a potential research direction.

miRNAs and oocyte quality in PCOS-IR

Recent studies indicated that the dysfunction and abnormal expression of miRNAs in PCOS could affect the physiological function of the ovary and might lead to ovulation dysfunction and reproductive endocrine abnormalities [52, 96]. High insulin concentration can induce the arrest of follicular growth and development [150, 151]. For example, insulin can stimulate the proliferation of buffalo GCs at concentrations of 50 ng/ml and above [152]. The expression of miR-320a is increased in PCOS tissues in an insulin-dependent manner, and its downregulation decreases cell viability and colony formation and leads the cell apoptosis [94]. When KGN cells were treated with insulin, downregulated miR-483 promoted IGF-1 expression and insulin-induced cell proliferation [76] by targeting the IGF1 3′ UTR directly. After insulin treatment, the expression of miR-27a-3p was decreased in GCs by mediating signal transducers and transcription activators STAT1 and STAT3. It is worth noting that the overexpression of miR-27a-3p in KGN cells can reduce cell proliferation and promote apoptosis [96].

Furthermore, increased apoptosis of GCs and CCs may affect follicular development by preventing the conversion of androgens to estrogen and changing hormone levels in follicles [153, 154]. On the one hand, the apoptosis and dysfunction in CCs may be the important contributors to multiple symptoms in patients with PCOS [155], including irregular menstruation, follicular hypoplasia, anovulation, and infertility [156]. CCs provide nutrition and growth regulators essential for oocytes and secrete ovarian steroids and produce growth hormone [157]. In a case-control study, Hu et al. found that a total of 59 known miRNAs and 617 genes were differentially expressed in CCs of PCOS-IR. The target genes of miRNAs were analyzed by gene ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway annotation and found to contribute to the etiology and pathophysiology of PCOS-IR [158]. Rap1b is a regulator of the MAPK pathway and was suppressed directly by miR-612 in CCs of PCOS-IR [158]. IGF-1 is an effective mitogenic factor involved in the growth of preantral follicles in PCOS [159] and is essential for the development of ovarian follicles [160]. IGF-1 could act upon autocrine to stimulate CCs replication and differentiation [161], and elevated IGF-1 activity is proposed as a mechanism leading to the increased activation and accelerated growth of early-stage follicles within PCOS ovaries [162]. miR-323-3p targets IGF-1 to inhibit steroid production and apoptosis of CCs which leads to PCOS and has been reported to be downregulated in the FF from embryos with low quality [163]. It can also regulate the expression of epidermal growth factor receptor (EGFR), a marker of mammalian oocyte capacity [164], and affect meiosis and maturation of oocytes [91]. Another miRNA, mir-99a, regulates the apoptosis and proliferation of human GCs in PCOS by targeting IGF-1R [165]. On the other hand, studies showed that GC dysfunction with reduced proliferation, increased apoptosis, or imbalanced hormone secretion [166, 167] is one of the causes of abnormal follicle formation in PCOS [168]. Proliferating GCs provide a suitable microenvironment for the initiation of follicular growth, oocyte maturation, and atresia [169], and the communication is essential for the maintenance of normal fertility [170]. Recent studies found the increased GC apoptosis in patients with PCOS and PCOS animal models [171, 172]. There was an increased high-mobility group box 1 (HMGB1) in the serum from PCOS-IR women compared to PCOS-NIR, which promotes apoptosis in primary cultured rat ovarian GCs [173], and miR-204 downregulation can induce HMGB1 upregulation, promote proliferation, and inhibit apoptosis of GCs [174]. A study showed increased miR-93 targeted CDKN1A, which promotes the proliferation and cell cycle progression of GCs in PCOS [175], and it may be explained by the high concentration of insulin in PCOS [89]. Of note, metformin can effectively restore the periodic and natural ovulation of patients with PCOS by increasing the expression of CDKN1A in cultured bovine GCs [176]. miR-370 is a scaffold protein involved in the insulin pathway modulating IRS-1 expression [81] and the downregulation of miR-370 may target IRS-1 leading to disturbed insulin metabolism of GCs and affected oocyte quality [82, 83]. However, there is no relevant research confirming this hypothesis. These related miRNAs may influence the oocyte quality in PCOS-IR and may contribute to the adverse outcomes in PCOS patients undergoing ART, but there have been only a few reports about it.

Common differentially expressed miRNAs in different phenotypes in PCOS-IR

We reviewed and summarized the miRNAs that were altered in PCOS-IR and found that some common miRNAs can target multiple metabolic pathways, including insulin metabolism, androgen metabolism, lipid metabolism, and oogenesis. We explored the possible mechanisms in the insulin signaling pathways of PCOS-IR (Fig. 1). The expression of miR-323-3p in the CCs was downregulated in the PCOS group and was inversely proportional to the expression level of IGF-1, which was the direct target of miR-323-3p demonstrated by the dual-luciferase reporter assay [91]. Qing et al. detected miR-323-3p in the micro-vesicles-free FF from PCOS and control patients [55]. Moreover, IGF-1 may play a significant role in the regulation of steroidogenesis and apoptosis of CCs in the process of follicular development via the PI3K pathway [177]. In addition, Yang et al. [178] reported that a higher expression of IGF-1 could increase the vulnerability of granulosa cells, which eventually led to cell apoptosis. These results provide evidence that miR-323-3p, along with its target IGF-1, regulated steroid metabolism and cell viability.

miR-99a expression was downregulated in PCOS, and insulin could lower miR-99a expression in a dose-dependent manner, especially at a level above 80 ng/ml [165], so we speculate that the hyperphysiological dose of insulin in PCOS patients might be one of the reasons for decreased miR-99a level in GCs. Li et al. found that insulin-induced IGF-1/mTOR pathway activation was accompanied by a significantly reduced miR-99a level, and elevated IGF-1 enhanced IR in patients with PCOS. This indicates that IR, IGF system, and the expression of miR-99a possess some latent mutual effects [179]. Apart from the influence on folliculogenesis, a high concentration of free serum IGF-1 in patients with PCOS can adversely affect embryo implantation, causing higher blastocyst absorption rate and pregnancy miscarriages [180, 181]. Therefore, the upregulation of miR-99a might be an effective method to attenuate the adverse effects of excessive activation of the IGF-1/IGF-1R system in PCOS. At present, many studies have recognized that miRNAs target several genes and metabolic pathways, leading to PCOS-IR. However, the specific mechanism is not well understood. It is our focus to analyze these essential miRNAs related to PCOS-IR.

Conclusion

PCOS is the most common endocrine disease in women at childbearing age [182], and there are 3 different diagnostic criteria currently: NIH criteria [183], Rotterdam criteria [184], and Androgen Excess and PCOS Society [185], which are yet to be unified. Considering that women with PCOS already have an aberrant metabolic condition in their early adulthood, it is necessary to stratify PCOS patients by their metabolic risk, independently from the PCOS phenotypes. IR is one of the major determinants in the pathology of metabolic abnormalities, reproductive disorders, and the occurrence of PCOS. Several studies have shown that treatment targeting IR in PCOS can largely alleviate obesity and hyperandrogenism [186, 187]. Improved insulin sensitivity may facilitate glucose metabolism control, androgen level reduction, ovulation rate improvement, and thus improve fertility. miRNAs participate in hyperinsulinemia, ovarian insulin sensitivity, inflammation, and oocyte quality in PCOS. Due to the specificity of miRNAs, they may become unique and reliable biomarkers for the diagnosis of PCOS and can be used to distinguish the different phenotypes as determined by the Rotterdam criteria.

However, at present, we are still unable to identify the exact targets of dysfunctional miRNAs and their direct downstream processes. For example, ovarian function and regulation are highly species specific, and several human-specific miRNAs have been described [188], which appear to be GC specific. Therefore, maybe miRNAs in specific cell types of the ovary exist but have not been described or studied, which also plays a role in ovarian hormonal metabolism and development of PCOS. We speculate that the reasons may be the following: (1) the expression profiles of miRNAs may be related to race. Ding et al. found that there were no overlaps in changes to miRNAs between previous studies [52, 55, 80] and theirs, which may be due to the race of subjects (Han vs. the Caucasus) [189]. Moreover, studies showed an opposite effect of different miRNAs in FF of PCOS in China and the USA, and this may relate to race/ethnicity or PCOS phenotype and differences methodological [52, 55]. Therefore, it is important to investigate miRNA profiles from different races in a larger study; (2) in the previous studies, the sources of collected samples are diverse and cover a wide spectrum, such as plasma, FF, GCs, and CCs. As a result, the miRNA expression profiles found are different yet overlapping, which may be explained by the diversity of samples, the limitation of sample size, and the complexity and heterogeneity of PCOS; (3) each miRNA can have numerous gene targets and the gene targets may be different based on cell types, systems, and organs including the ovary [190]. miRNAs can also be regulated by other paracrine/autocrine factors, which makes it a highly complex system [191, 192].

Nowadays, several molecular approaches are rapidly developing which facilitate the identification of genomic regions associated with pregnancy establishment and fertility, including functional genomic approaches, structural genomics and integrated analysis of the transcriptome, and epigenome (DNA methylation and histone modification) using high-throughput next-generation sequencing techniques. High-throughput profiling and sequencing of miRNAs can provide an all-around tool to investigate all kinds of signaling pathways, improve clinical management, and determine appropriate infertility treatment based on abnormal follicular development in PCOS. With the help of novel technologies and searchable databases (including ontology and biological information databases), miRNA can provide us with exquisite analysis, prevention, and management for reproductive diseases like PCOS. The top priority is to conduct large-scale replication studies to identify certain miRNAs that have strong modulation effects in PCOS. Considering miRNAs often exist in families, an increasingly important approach is to modulate the levels of entire miRNA families or co-regulated miRNAs together as well as separately to establish roles in the intact tissues or organs. We can also analyze the functional role of single miRNAs in the pathology of PCOS-IR and verify the function of a single miRNA both in vivo and in vitro. It is worth noting that many miRNAs have the strongest effects at the level of translation inhibition, and the study of mRNA expression arrays may have limited use to reveal the real impact of regulated miRNAs on the PCOS phenotype. Recently, many miRNA mimics, anti-miRNA oligonucleotides, and small interfering RNA therapeutics are being evaluated. Currently, there is no treatment to target the miRNAs in PCOS-IR directly, and studying related miRNAs may pose new research directions for the treatment of PCOS-IR in the future to utilize miRNAs as biomarkers.

Authors’ contributions

Yingliu Luo: Writing-original draft preparation and editing

Chenchen Cui: Writing-reviewing and editing

Xiao Han: Writing-reviewing, editing, and supervision

Qian Wang: Supervision

Cuilian Zhang: Supervision, project administration, and funding acquisition

Funding

This work was supported by the National Natural Science Foundation of China (81571407).

Compliance with ethical standards

Conflict of interest

The authors declare that they have no conflicts of interest.