The Expression of the miR-25/93/106b Family of Micro-RNAs in the Adipose Tissue of Women With Polycystic Ovary Syndrome

Abstract

Context:

In adipose tissue (AT) micro-RNA-93 (miR-93) is significantly overexpressed in polycystic ovary syndrome (PCOS) women and non-PCOS women with insulin resistance (IR). Overexpressed miR-93 directly inhibits glucose transporter isoform 4, impairing both glucose metabolism and insulin sensitivity. The mechanisms behind increased miR-93 expression are unclear.

Objective:

Our objective was to determine whether miR-93 expression is concordant with its host gene, MCM7, which contains the miR-25/93/106b gene cluster.

Patients:

AT was excised from 16 women with PCOS (eight with and eight without IR) and 15 non-PCOS (nine with and six without IR).

Main Outcome Measures:

Expression of MCM7 and miR-25/93/106b was measured in AT and 3T3-L1 cells.

Results:

MCM7 expression was lower in both non-PCOS/IR and PCOS women and tended to be lowest in women with PCOS and IR. Overall, the expression of MCM7 in human AT was negatively associated with miR-93 expression and with increased subject IR. Additionally, miR-25 and miR-106b expression is uncoupled from the MCM7 host gene and are positively correlated with IR, although no PCOS-specific difference was observed. MCM7 expression appears to be negatively correlated with increasing fasting glucose. In 3T3-L1 adipocytes, increasing glucose had no effect on miR-93 or miR-25, although it reduced MCM7 and increased miR-106b expression in a dose-dependent fashion. In turn, in 3T3-L1 adipocytes, increasing insulin had no effect on either MCM7 or miR-25/93/106b expression.

Conclusions:

Our data suggest that the expression of MCM7 and miR-93/25 is PCOS and IR related, whereas that of miR-106b is related to IR only. In 3T3-L1 adipocytes, neither hyperglycemia nor hyperinsulinemia altered the expression of miR-93 or miR-25, although increasing glucose levels down-regulated MCM7 and paradoxically increased that of miR-106b expression. The expression of the miR-25/93/106b family may be regulated through mechanisms distinct from its host gene, MCM7. Finally, our studies suggest potential epigenetic mechanisms for both IR and PCOS.

Polycystic ovary syndrome (PCOS) is one of the most common endocrine disorders, affecting 7%–9% or more of reproductive-aged women. About 60%–70% of PCOS patients demonstrate insulin resistance (IR) above and beyond that predicted by body mass, race, or age (1), resulting in compensatory hyperinsulinemia (2) and an increased risk for type 2 diabetes mellitus and cardiovascular disease (3, 4). The underlying cellular mechanisms leading to IR in PCOS remains to be completely elucidated, with no gross defects in the traditional insulin signaling pathways being found, including insulin binding, insulin receptor expression, and the insulin receptor substrate-1/phosphatidylinositol 3-kinase/AKT pathway. In women with PCOS, sc adipocyte functions, including the stimulation of glucose transport (3, 5), glucose transporter isoform 4 (GLUT4) production (6), and lipolysis (7, 8), appear to be defective in the disorder (9–11).

Micro-RNAs are short (20–24 nucleotides) noncoding RNAs involved in posttranscriptional regulation of gene expression. Insulin-resistant adipocytes are known to contain a differentially expressed miRNA profile (12, 13). We previously reported that micro-RNA-93 (miR-93) is up-regulated in adipose tissue (AT) from PCOS and non-PCOS women who display IR (12). Overexpressed miR-93 directly inhibits GLUT4 expression, influencing glucose metabolism, although the precise regulation of miR-93 expression in adipocytes is unknown. miR-93 is an intronic miRNA because its transcript is integrated into the 13th intron of the highly conserved minichromosome maintenance-deficient 7 (MCM7) (14) (Figure 1A). As an intronic miRNA, miR-93 is thought to be transcribed along with its host gene, MCM7 (14), although some reports suggest that miR-93 might also have its own promoter (14). Of note, in mouse podocytes high glucose down-regulates miR-93 expression by the decreased expression of it host gene, MCM7 (14). In the present study, we tested the hypothesis that the expression of miR-93 would parallel that of its host gene MCM7 and the expression of other members of the miR-25/93/106b family to better understand the regulation of relevant miRs in AT of IR and/or PCOS women.

Figure 1.

The expression of miR-93 and MCM7 in human AT in 4 subgroups is depicted in panels A and B, respectively. The 4 groups include non-PCOS subjects with and without IR (Non-PCOS/IR and Non-PCOS/non-IR, respectively) and PCOS subjects with and without IR (PCOS/IR and PCOS/non-IR, respectively). Panel C depicts the association of MCM7 with miR-93 expression. Panel D depicts the association of MCM7 expression with HOMA-IR. Values are presented as mean ± SEM.

Materials and Methods

Study subjects

Subcutaneous abdominal AT samples from 31 (15 non-PCOS and 16 PCOS) women were studied. Of these 31 subjects, 28 of these subjects have been included in a previous report studying miR-93 expression (12). This study was approved and all subjects gave informed written consent according to the guidelines of the institutional review board.

The diagnosis of PCOS was made according to the National Institutes of Health 1990 criteria (15), including the following: 1) clinical evidence of hyperandrogenism (hirsutism) and/or hyperandrogenemia; 2) oligoovulation; and 3) the exclusion of related disorders. The criteria for defining hirsutism, hyperandrogenemia, ovulatory dysfunction, and exclusion of related disorders have been previously reported (15). Controls were healthy women with no hirsutism, on no medications, and with a history of predictable regular menstrual cycles, a good predictor of normal ovulatory function in nonhirsute women (16). Basal insulin resistance was estimated using the homeostasis model assessment (HOMA-IR); a HOMA-IR value of less than 2.5 considered as non-IR and a HOMA-IR of 2.5 or greater considered IR, as previously reported (17).

Human AT biopsy and real-time PCR

Approximately 5 g of sc AT was excised through a small incision in the lower abdomen (http://www.youtube.com/fatbx) (18). Total RNA was extracted using the miRACLE isolation kit (Jinfiniti Biosciences). First-strand cDNAs of mRNA and miRNA were synthesized using the high-capacity cDNA reverse transcription kit (Applied Biosystems) and first-strand cDNA synthesis kit for miRNA (Origene). Real-time PCR was performed using an iTag Universal SYBR Green Supermix (Bio-Rad Laboratories, Inc). Primers were purchased from Origene. Experiments were performed on an Applied Biosystems 7300 real-time PCR System. ACTB and Actb were used as internal controls. Relative fold change of target genes expression was calculated by using the 2^-δδCt method.

Differentiation of mouse 3T3-L1 preadipocytes to adipocytes and culturing in high glucose medium

3T3-L1 preadipocytes (catalog number SP-L1-F) were purchased from Zenbio Co. 3T3-L1 preadipocytes were maintained in preadipocyte medium (catalog number PM-1-L1; ZenBio Inc) until 100% confluent in a humidified incubator, 37°C, with 5%–10% CO₂. Cells were provided with PM-1-L1 every other day. Once the cells were confluent, they were incubated an additional 48 hours before differentiation to adipocytes was initiated. In brief, 2 days (day 0 of differentiation) after the cells had been confluent, we removed the preadipocyte medium and replaced it with an appropriate volume of 3T3-L1 differentiation medium (catalog number DM-2-L1; ZenBio Inc). After 3 days (day 3 of differentiation), the differentiation medium was removed and replaced with 3T3-L1 adipocyte maintenance medium (catalog number AM-1-L1; ZenBio Inc). The cells were incubated for an additional 3 days in adipocyte maintenance medium and every 2–3 days were provided 3T3-L1 adipocyte maintenance medium until ready for assay. The assays were done on days 7–14 of differentiation. The 3T3-L1 adipocytes were cultured in normal adipocyte maintenance medium (1×, 3.15 g/L D-glucose) and in high-glucose adipocyte maintenance media (5×, 15.75 g/L; or 10×, 31.50 g/L]) (14). For the osmolarity experiment, the 1× incubation medium once again contained 3.15 g/L D-glucose and no mannitol. For the 5× and 10× mannitol experiments, the incubation mediums contained 3.15 g/L of D-glucose and 12.60 g/L (5×) or 28.35 g/L (10×) of mannitol, respectively, the 3T3-L1 adipocytes were treated in these mediums for 24 hours.

Differentiation of mouse 3T3-L1 preadipocytes to adipocytes and culturing in medium contain different doses of insulin

3T3-L1 preadipocytes were differentiated into adipocytes using the method as described by Student et al (19). Briefly, 2 days after cell confluence, the medium was replaced with fresh medium containing 10% fetal bovine serum (FBS), 0.5 mM 3-isobutyl-1-methylxanthine, 1 μM dexamethasone, and 10 μg/ml insulin. Forty-eight hours later, the medium was replaced with fresh medium containing 10% FBS and 10 μg/ml insulin. After another 48 hours, the cells were maintained in DMEM containing 10% FBS. The 3T3-L1 adipocytes were then treated with different doses of insulin (0, 1, 5, 10, 100, and 500 nM) for 24 hours.

Statistical analysis

Comparisons of multiple groups were carried out by an ANOVA followed by a posttest using the Tukey (among groups) and Dunnett (compared with control group) tests (XLSTAT Software). Subjects were subdivided into 4 groups for analysis: PCOS women with insulin resistance as defined (PCOS/IR), PCOS women without IR (PCOS/non-IR), apparently healthy control women with IR (Non-PCOS/IR), and healthy women without IR (Non-PCOS/non-IR). Significant differences were defined as P < .05. For the outlier test, we compared the Pearson's correlation coefficient before and after exclusion of the point that is suspected to be an outlier in a scatter plot. We also ran the Grubbs and Dixon Q tests (data not shown) but believed that these tests were inappropriate in the setting of these experiments because the data contain two variables, and these tests could possibly underestimates a distinct direction of the outlier from the rest of the sample population.

Results

In human AT of PCOS or IR subjects, the expression of MCM7 does not mirror that of miR-93

We examined sc abdominal AT samples from 15 Non-PCOS and 16 PCOS subjects. The clinical characteristics of our subjects are presented in Table 1. As we previously reported (12), miR-93 was overexpressed in AT from non-PCOS/IR and all PCOS women (Figure 1A). However, and contrary to our hypothesis, the expression of MCM7 was significantly lower in the AT of PCOS women vs Non-PCOs women (control vs PCOS: 0.41 ± 0.053 vs 0.24 ± 0.06 relative values, P < .05). When the subjects were assessed according to the presence or absence of both IR and PCOS, we observed that the expression of MCM7 was reduced in the AT of Non-PCOS/IR women and in both PCOS subgroups, compared with women without PCOS or IR (Figure 1B). Although the expression of MCM7 trended lower in PCOS/IR than in all other subjects, the difference did not reach significance (PCOS/IR vs PCOS/non-IR, P = .051; and PCOS/IR vs Non-PCOS/IR, P = .074). Overall, the expression of MCM7 in human AT was negatively associated with the expression of miR-93 (Figure 1C) and increased IR (Figure 1D).

Table 1.

Clinical Characteristics of PCOS and Non-PCOS Subjects With and Without IR

	Non-PCOS/Non-IR (n = 6)	Non-PCOS/IR (n = 9)	PCOS/Non-IR (n = 8)	PCOS/IR (n = 8)
BMI, kg/m2	22.80 ± 1.32	35.78 ± 7.68a	25.49 ± 5.56	31.92 ± 4.95a,a
Age, y	32.33 ± 5.03	33.66 ± 6.37	30.00 ± 5.57	27.75 ± 4.98
mFG score	0.33 ± 0.57	0.83 ± 0.98	11.00 ± 4.98a	6.63 ± 3.88a
Free T, pg/mL	2.20 ± 0.66	3.00 ± 0.43	5.26 ± 2.05a	5.30 ± 1.30a
Total T, ng/mL	27.00 ± 8.89	26.33 ± 0.58	49.33 ± 22.48	30.86 ± 5.24
Fasting glucose, μg/dL	67.50 ± 24.75	82.13 ± 5.87	81.00 ± 8.18	95.81 ± 18.07c
Fasting insulin, mIU/mL	4.00 ± 1.41	14.60 ± 1.95a	4.671 ± 0.57	21.91 ± 7.27a,d
HOMA-IR	0.71 ± 0.48	2.94 ± 0.31a	0.94 ± 0.20	5.07 ± 1.50a,d

Abbreviations: BMI, body mass index; mFG score, modified Ferriman-Gallwey hirsutism score. Data are expressed as mean ± SD. Subjects were subdivided into 4 groups for analysis: PCOS women with insulin resistance as defined (PCOS/IR), PCOS women without IR (PCOS/non-IR), apparently healthy women with IR (Non-PCOS/IR), and healthy women without IR (Non-PCOS/non-IR).

^aP < 0.01 vs Non-PCOS/non-IR group.

^bP < 0.05 vs PCOS/non-IR group.

^cP < 0.05 vs Non-PCOS/non-IR group.

^dP < 0.01 vs PCOS/non-IR group.

In human AT of PCOS or IR subjects, the expression of miR-25 and miR-106b does not mirror that of miR-93 or MCM7

Contrary to that of their host gene MCM7 and different from that of miR-93, we observed that the expression of miR-25 and miR-106b was higher in IR subjects than in non-IR subjects, regardless of the presence or absence of PCOS (Figure 2, A and B). Although the levels of miR-25 were higher in the AT of PCOS/IR than all other subjects, the differences did not reach significance in comparison with Non-PCOS/IR subjects (P = .061; Figure 2A). Both miR-25 and miR-106b expressions in AT were positively associated with IR (Figure 2, C and D) and miR-93 expression (Figure 2, E and F).

Figure 2.

The expression of miR-25 and miR-106b expression in human AT of individuals with IR and/or PCOS is depicted in panels A and B, respectively. Panels C and D depict the association of miR-25 and miR-106b with HOMA-IR, respectively. Panels E and F depict the association of miR-25 and miR-106b with miR-93 expression, respectively. Values are presented as mean ± SEM.

In 3T3-L1 adipocytes, higher extracellular glucose concentrations reduce the expression of MCM7 but do not alter the expression of miR-93

Our data indicated that the expression of MCM7 was negatively correlated with fasting glucose levels in the women studied in vivo (Figure 3A). To investigate whether excess glucose regulates MCM7 and miR-93 expression in adipocytes, we treated 3T3-L1 adipocytes with glucose and found that 5× and 10× (15.75 g/L and 31.50 g/L glucose concentrations, respectively, relative to a normal culture medium level of 3.15 g/L) significantly reduced MCM7 expression in a dose-dependent fashion (Figure 3B). Mannitol was used as an osmolarity control, and our data indicate that medium osmolarity, within the range tested, did not affect the expression of MCM7 (Figure 3C). We found that miR-25 and miR-93 expressions were unresponsive to changes in medium glucose concentration (Figure 3D). However, the expression of miR-106b was increased in the highest glucose (10×) treatment.

Figure 3.

The association between the expression of MCM7 and circulating fasting glucose in vivo is depicted in panel A. We tested for an outlier effect for the data point that is lowest on the x-axis (fasting glucose concentration) and highest on the y-axis (MCM7 expression) by comparing the Pearson's correlation coefficient (r) before and after the exclusion of value. However, removal of the value did not change the correlation coefficient significantly (r = −0.538 before and r = −0.433 after exclusion of the value), suggesting that this data point was not an outlier. Panels B and D depict the effect of varying insulin concentrations in the culture medium on the expression of MCM7 and miR-93 in 3T3-L1 adipocytes. Panel C depicts experiments assessing the effect of mannitol, as an osmolarity control, on MCM7 expression (n = 3). Values are presented as mean ± SEM.

In 3T3-L1 adipocytes higher extracellular insulin concentrations does not alter the expression of MCM7 or miR-25/93/106b

Compensatory hyperinsulinemia is the main characteristic of patients with PCOS (2). To test whether insulin regulates MCM7 and miR-25/93/106b expression in adipocytes, we treated 3T3-L1 adipocytes with different doses of insulin from low to high 0, 1, 5, 10, 100, and 500 nM for 24 hours. Our results indicated that insulin treatment (even in the highest dose, 500 nM) does not alter both MCM7 (Figure 4A) and miR-25/93/106b (Fig. 4B) expression in 3T3-L1 adipocytes. To confirm that the insulin treatment was having its predicted effect in the adipocytes, GLUT4 expression was also assessed. Our data indicated that insulin treatment began reducing GLUT4 the expression at the dose of 5 nM in the 3T3-L1 adipocytes (Figure 4C), similar to that previously reported (20).

Figure 4.

The effect of varying insulin concentrations in the culture medium on the expression of MCM7 in 3T3-L1 adipocytes is depicted in panels A. Panels B and C depict the effect of varying insulin concentrations on the expression of miR-25/93/106b and GLUT4 (n = 4). Values are presented as mean ± SEM.

Discussion

In a previous report (12), we demonstrated that miR-93 was overexpressed in the AT of all women with PCOS and in non-PCOS asymptomatic women with measurable IR, which was confirmed in the present experiments. Overexpressed miR-93 directly inhibits GLUT4 expression, impairing glucose metabolism and insulin sensitivity (12). In this study, we sought to determine whether miR-93 and other members of the same miR family are up-regulated in the women with PCOS and/or IR as a result of the increased expression of its host gene, MCM7. Although the precise mechanisms regulating miRNA expression are largely unknown, approximately half of the conserved miRNA genetic information can be found within a host gene intron (21), and the miR-25/93/106b family is found within the intronic region of the gene MCM7 (22). Several studies report that the expression of miR-93 correlates with that of MCM7. For example, MCM7 and miR-93 were both down-regulated in mouse podocytes in hyperglycemic conditions as well as during human monocytic differentiation (14, 23). In concert, miR-93 and MCM7 transcriptions were both up-regulated in gastric tumors (22).

However, in the present study we found that the expression of miR-93 and its host gene MCM7 appears to be discordant in the AT of women with PCOS women and/or IR. The expression of miR-93 and MCM7 was negatively associated (Figure 1C), and the expression of MCM7 was actually lower in the AT of all women with PCOS and those non-PCOS women with IR, although the expression of MCM7 tended to be lowest in women with both PCOS and IR (Figure 1B). Alternatively, miR-93 is overexpressed in these same groups (Figure 1, A and B). Discordant expression of miR-93 and MCM7 has also been reported in leiomyomas, in which the expression of miR-93 was down-regulated, whereas that of MCM7 was increased (24).

Similar to the expression of miR-93, the expression of the other members of the miR family (miR-25 and 106b) found within the intronic region of MCM7 does not appear to parallel that of MCM7. However, and in contrast to that of miR-93, both miR-25 and miR-106b were overexpressed only in women with IR, regardless of PCOS status (Figure 2, A and B), although miR-25 tended to be highest in women with both PCOS and IR. Importantly, the expression of the miR-25/93/106b family (positively) and host gene MCM7 (negatively) all were associated with measurable IR in vivo (as measured by HOMA-IR) (Figures 1D and 2, C and D). Taken together, these results suggest that the expression of MCM7, miR-93, and miR-25 are PCOS and IR related, whereas that of miR-106b is related to IR only.

Our finding of the discordant expression of miR-25/93/106b and their host gene MCM7 at baseline and in hyperglycemic conditions (Figures 1, A and B, 2, A and B, and 3, B and D) support the notion that in AT these miRs may have their own regulatory mechanism, independent of MCM7. The possible mechanisms accounting for the disparity in miRNA and its host gene expression include disparate transcript stabilities, posttranscriptional miRNA regulation, or host-gene independent expression of intronic miRNA (25). For example, changes in the intensity of expression from that of family members miR-106b and miR-25, as well as the presence of an intronic region containing CpG islands, have led to the speculation that miR-93 may have its own promoter (14, 26). Indeed, in mouse osteoblasts, an miR-93 individual promoter region has been reported just upstream of the miR-93 transcription start site (TSS) for the specificity protein-7 transcription factor 7 (specificity protein-7; Osterix), a zinc finger transcription factor and a critical regulator of osteoblast mineralization (27). Monteys et al (26) identified an individual TSS and different transcription factor binding sites near the predicted TSS for miR-25, miR-93, and miR-106b, further supporting the hypothesis that each of these miRs may have their own promoters.

In mouse podocytes, high glucose down-regulates miR-93 expression by decreased expression of it host gene, MCM7 (14). In addition, MCM7 expression was negatively associated with circulating fasting glucose in vivo (Figure 3A). Thus, it was possible that circulating extracellular glucose plays a role in the discordant expression the miR-25/93/106b family and host gene MCM7 in AT. We found that increased glucose concentrations in the medium reduced MCM7 expression in a dose-dependent fashion. Alternatively, and contrary to the prior findings in mouse podocytes (14), in 3T3-L1 adipocytes, hyperglycemia did not alter the expression of miR-93 expression or that of miR-25 and paradoxically increased the expression of miR-106b. These data suggest that circulating hyperglycemia appears to not play a significant role in regulating the expression of miR-93 and miR-25, but it may be operant in up-regulating the expression of miR-106b in women with IR.

In addition, compensatory hyperinsulinemia, a main feature of many patients with PCOS (2), could account for the abnormal expression of miR-93 in this disorder. However, in 3T3-L1 adipocytes, we found that increasing insulin did not alter the expression of either MCM7 or the miR-106b/93/25 genes.

Finally, RNA binding proteins (RBPs) could play a role in the discordance of expression. RBPs are often key to determining the function and expression of an miRNA because they control their biogenesis, localization, activity, and degradation (25). Although most miRNA controllers may act globally, RBP regulation is unique in that it is often specific to a single miRNA or its family, making it reasonable to suggest that specific, and as yet to be determined, RBPs may play a role in the discordant expression of miR-106b/93/25 family and its host gene, MCM7 (25).

In conclusion, our results indicate that the increased miR-93 expression observed in the sc AT of PCOS patients and non-PCOS women with IR, and miR-25 and 106b in women with IR, does not mirror that of its host gene MCM7. Furthermore, our results suggest that neither hyperglycemia nor hyperinsulinemia regulate miR-93 or miR-25 expression in adipocytes, although high glucose levels may foster the up-regulation of miR-106b expression in subjects with IR. These data suggest that these miRs likely have their own unique regulatory mechanisms, which is dysfunctional or aberrant in women with PCOS and/or IR. In addition, our data support the possibility that the miR-25/93/106b family plays a role in the development of IR and/or PCOS. Further studies are needed to determine whether there are a specific miR-25/93/106b family promoters or RBPs in action in this setting. Understanding the mechanisms that regulate the overexpression of the miR-25/93/106b family in subjects with PCOS and/or IR, independent of the regulation of their host gene MCM7 could identify potential new therapeutic targets in these women.