Reduced Expression of Nuclear-Encoded Genes Involved in Mitochondrial Oxidative Metabolism in Skeletal Muscle of Insulin-Resistant Women With Polycystic Ovary Syndrome Free

Sixteen obese women of fertile age with PCOS and 13 healthy women, matched according to age and BMI, participated in the study (Table 1). These subjects were selected from a larger cohort who participated in a study reported recently (6). The subpopulation of PCOS patients represents the most insulin-resistant subjects from whom a basal muscle biopsy were obtained. Inclusion and exclusion criteria for PCOS and control subjects are as described previously (6). In brief, criteria for PCOS included oligoovulation (defined as irregular periods during <1 year in combination with a cycle length of >35 days), elevated free testosterone levels (>0.035 nmol/l), and/or hirsutism (total Ferriman-Gallwey score >7) and the absence of diabetes, hypertension, hyperprolactinemia, hypothyroidism, and adrenal enzyme defects. All PCOS subjects had A1C within the normal range (4.9–6.1%). Control subjects had regular menses, normal glucose tolerance, and no family history of diabetes. No subjects were taking medicines known to affect hormonal or metabolic parameters. Informed written consent was obtained from all subjects before participation. The study was approved by the local ethics committee and was performed in accordance with the Helsinki Declaration.

The euglycemic-hyperinsulinemic clamp studies were performed after an overnight fast as described in detail previously (6). In brief, a 2-h basal tracer equilibration period was followed by infusion of insulin at a rate of 40 mU · m⁻² · min⁻¹ for 3 h. The studies were combined with indirect calorimetry, and rates of total glucose disposal, glucose and lipid oxidation, and nonoxidative glucose metabolism were calculated as described previously (6). A muscle biopsy from each subject was obtained from the vastus lateralis muscle in the basal state after the 2-h basal tracer equilibration period using a modified Bergström needle with suction under local anesthesia. Muscle samples were immediately frozen in liquid nitrogen within 30 s. Serum levels of insulin, free testosterone, luteinizing hormone (LH), follicle-stimulating hormone (FSH), and plasma glucose, triglyceride, and free fatty acids (FFAs) were assayed as described previously (6).

Frozen muscle tissue was homogenized in Trizol with a Polytron Homogenizer, and total RNA was purified using TRIzol Reagent (Life Technologies, Gaithersburg, MD), including an extra step after phase separation with chloroform. Briefly, 1 vol phenol-chloroform-isoamylalcohol (25-24-1) (Sigma-Aldrich, St. Louis, MO) was added to the aqueous phase and spun at 12,000 rpm for 20 min at 4°C. Then, 1 vol chloroform was added to the aqueous phase and spun at 12,000 rpm for 20 min at 4°C. Quantity of RNA was determined with a spectrophotometer, and RNA integrity was assessed using Agilent 2100 Bioanalyser and degradometer software (24).

One μg purified total RNA was converted to biotin-labeled aRNA using the MessageAmpTM II-Biotin single-round amplified RNA (aRNA) amplification kit according to the manufacturer's instructions (Ambion, Austin, TX). Labeled aRNA was fragmented as described in the Affymetrix manual (Affymetrix, Santa Clara, CA) and hybridized to Affymetrix HG-U133 Plus 2.0 chips. All glyceraldehyde-3-phosphate dehydrogenase (GAPDH) 3′/5′ probe set hybridization ratios range from 0.92 to 1.42.

The R statistical software (25) was applied for initial data processing and statistical analysis. Global background correction of probe intensities was performed using a method implemented in the robust multi-array average method (26), and nonlinear normalization of probe intensities was done using qspline (27). Gene expression index calculation was done using model-based index calculation (28). Only perfect-match probes were included in data analysis. Differences in gene expression between groups were calculated for each gene in the dataset by using Welch two sample t test. An uncorrected P < 0.05 was considered significant.

Gene expression changes were integrated with gene sets and biological pathways using gene map annotator and pathway profiler (GenMAPP 2.0) (29) and gene set enrichment analysis (GSEA 1.0) (30). MAPPFinder 2.0 (31), a tool integrated in GenMAPP 2.0, was applied to examine significant pathways and gene ontology terms. In GSEA, a total of 169 gene sets were applied. All genes were ranked according to changes in gene expression between PCOS patients and control subjects using the t test in the GSEA software. Ten thousand gene permutations were used to obtain the nominal P value, and gene sets with a false discovery rate (FDR) <0.05 were suggested to be significantly regulated.

DNase I–treated RNA from 13 PCOS patients and 13 control subjects was reverse transcribed to single-stranded cDNA using TaqMan reverse transcription reagents and random hexamer primers (Applied Biosystems). Three women with PCOS were not included in the analysis because of a lack of RNA. TaqMan gene expression assays (Applied Biosystems) for nine selected genes (Supplemental Table 1, which is detailed in the online appendix [available at http://dx.doi.org/10.2337/db07-0275]) and TaqMan Universal Master Mix (Applied Biosystems) were used for quantification of gene expression using Applied Biosystems Prism 7700. Expression of target genes was normalized to the endogenous controls PPIA (cyclophilin A) and GAPDH. To perform the most appropriate validation of microarray data, bioinformatic approaches, such as NetAffx (http://www.Affymetrix.com), refseq (http://www.ncbi.nlm.nih.gov), and Ensembl (http://www.ensembl.org), were used to identify the Affymetrix probe set for each gene with the highest similarity to the TaqMan probe sequence (Supplemental Table 1).

Clinical and metabolic characteristics of the study subjects are presented in Table 1. Fasting levels of serum insulin, free testosterone, LH-to-FSH ratio, and plasma free triglycerides were elevated in women with PCOS compared with control subjects (P < 0.01). No difference in fasting plasma glucose or FFA levels was observed. In the basal state, rates of glucose and lipid metabolism were similar in the groups (data not shown). During the insulin-stimulated period, total glucose disposal was 58% lower in women with PCOS than in control subjects (P < 0.01), and this was primarily accounted for by a 72% reduction in nonoxidative glucose metabolism (P < 0.01), but also a 39% decrease in glucose oxidation (P < 0.01). Moreover, the ability of insulin to suppress lipid oxidation and plasma FFA levels was impaired in PCOS patients compared with control subjects (P < 0.01).

By applying the R statistical software, we found that of the 54,675 transcripts represented on the array, 3,730 transcripts were downregulated and 2,407 were upregulated in skeletal muscle of PCOS patients compared with control subjects (uncorrected P < 0.05). Only 34 probe sets with a fold change ranging from −1.09 to −1.47 for downregulated genes and from 1.12 to 1.85 for upregulated genes (Supplemental Table 2) remained differentially expressed after controlling for multiple hypothesis testing using the Benjamini-Hochberg method (FDR <0.1) (32). Nearly one-half of these probe sets had unknown function, and the remaining probe sets did not appear to be of interest in the study of the pathogenesis of insulin resistance in PCOS.

We applied two pathway programs, GenMAPP 2.0 and GSEA 1.0, to perform a general search for pathways associated with insulin resistance in skeletal muscle of PCOS patients. Using MAPPFinder 2.0, the significantly downregulated pathways in muscle of PCOS patients were the electron transport chain, transforming growth factor-β signaling pathway, G-protein signaling, calcium regulation in cardiac cells, adipogenesis, and insulin signaling (z > 2.0) (Table 2). Nine pathways were significantly upregulated, and the 10th most upregulated pathway was insulin signaling (z = 1.7) (Table 3). Interestingly, two pathways, insulin signaling and calcium regulation in cardiac cells, were found among the top 10 of both the up- and downregulated pathways. We could not find evidence that a specific subgroup of genes involved in insulin signaling to glucose metabolism, protein synthesis, or mitogenesis was unambiguously regulated in one direction. Similarly, the genes involved in calcium regulation did not point to specific subgroups of down- or upregulated genes. When evaluating the family-wise error rate (FWER) values for up- and downregulated pathways, only the electron transport chain remained significant. In gene ontology, oxidoreductase activity/acting on NADH or NADPH and closely related pathways were the most significantly downregulated molecular function terms. Of interest, calcium ion transporter activity was the fourth most downregulated molecular function term, further indicating a role for calcium regulation in muscle of women with PCOS. Mitochondrion was the most downregulated cellular component, and mitochondrial electron transport/ubiquinol to cytochrome c was the most downregulated biological process (Supplemental Table 3). Upregulated gene ontology terms are depicted in Supplemental Table 4.

Applying GSEA on the same dataset, we found that only the gene sets VOXPHOS and electron transport chain were significantly downregulated (FDR <0.05). Moreover, the gene sets termed mitochondria and human mitoDB 6 2002 were among the 10 most downregulated pathways. Fatty acid metabolism was the third most downregulated pathway, and another fatty acid metabolism gene set was the sixth most downregulated pathway (Table 4). No gene sets were significantly upregulated (FDR <0.05) (Supplemental Table 5). Taking the FWER P value into account, only the VOXPHOS and electron transport chain gene sets remained significant. When evaluating the results from GenMAPP and GSEA, pathways representing OXPHOS genes were consistently downregulated, even when using the very stringent FWER P value.

Earlier studies have implicated a role for alterations in OXPHOS in insulin resistance. Using two approaches for global pathway analysis of muscle transcripts, we consistently observed a downregulation of nuclear-encoded OXPHOS genes in muscle of PCOS patients. We therefore focused on this pathway in further analyses. To validate our microarray data, we used quantitative real-time PCR to examine gene expression levels of one gene from each of the five respiratory complexes (I–V) (NDUFA3, SDHD, UCRC, COX7C, and ATP5H) and one uncoupling protein (UCP2). In the microarray experiment, these genes were all downregulated at the single-gene level (uncorrected P < 0.05) in muscle from PCOS patients. Moreover, we studied gene expression of PGC-1α, PGC-1β, and NRF-1, which are known to be involved in the transcriptional control of mitochondrial biogenesis. In accordance with the results obtained from microarray analysis, the expression of four of five respiratory genes together with the expression of UCP2 was significantly downregulated in PCOS patients compared with control subjects (P < 0.05). Moreover, PGC1-α expression was significantly reduced in PCOS patients compared with control subjects (P < 0.01) (Fig. 1). No differences with respect to expression of PGC-1β and NRF-1 were found.

To explore the potential relationship between mRNA levels of PGC-1α and downregulation of OXPHOS genes, we performed simple correlation analysis. In the total population, PGC-1α mRNA levels correlated strongly with the expression of each of the five OXPHOS genes studied (r = 0.59–0.89; all P < 0.001) but not with UCP2 mRNA levels. In control subjects, the relationship between PGC-1α mRNA levels and expression of ATP5H, NDUFA3, and UCRC was preserved (all P < 0.05); whereas in PCOS patients, only NDUFA3 expression (P < 0.01) showed a significant association with PGC-1α levels.

Previous studies of insulin resistance in skeletal muscle of women with PCOS have focused on individual proteins and genes involved in insulin signaling (9–13). Using global approaches, such as transcriptomics and proteomics, it is possible to study the profile of a large number of distinct genes and proteins simultaneously. In the present study, we used DNA microarrays to compare skeletal muscle transcripts between insulin-resistant PCOS patients and matched control subjects. We demonstrate a significant downregulation of nuclear-encoded genes involved in OXPHOS using two different approaches for global pathway analysis. Quantitative real-time PCR analysis validated our findings and showed that downregulation of PGC-1α is strongly associated with reduced expression of OXPHOS genes. These findings provide evidence for an association between insulin resistance and impaired mitochondrial oxidative metabolism in skeletal muscle of women with PCOS.

The most important finding of the study is that expression of nuclear-encoded genes involved in mitochondrial oxidative metabolism is decreased in muscle of women with PCOS. A similar decrease in OXPHOS gene expression has previously been reported in patients with type 2 diabetes and their first-degree relatives (20–22) but is, to our knowledge, a novel finding associated with insulin resistance in PCOS, where it is independent of obesity and type 2 diabetes. The findings of a more pronounced reduction in OXPHOS gene expression in patients with type 2 diabetes compared with their first-degree relatives (22) and that insulin treatment partly normalizes OXPHOS gene expression in poorly controlled patients with type 2 diabetes (20) raise the possibility that impaired oxidative metabolism is, at least in part, secondary to elevated circulating glucose levels. However, a lower mitochondrial ATP flux has been observed in insulin-resistant, glucose-tolerant, first-degree relatives of patients with type 2 diabetes (19). In our study, all PCOS subjects had normal A1C. Thus, although the etiological mechanisms of insulin resistance in PCOS and first-degree relatives of patients with type 2 diabetes may differ, this study provides further support for an association between reduced OXPHOS gene expression and insulin resistance in skeletal muscle at an early stage before hyperglycemia develops. In skeletal muscle of obese subjects and patients with type 2 diabetes, mitochondrial dysfunction seems to be caused by both a lower number of and decreased functional capacity of mitochondria (7,17,18). Further studies are needed to establish whether reduced OXPHOS gene expression is also associated with a lower content and function of muscle mitochondria in PCOS. Transcriptional profiling of ovarian tissues in PCOS did not show decreased expression of OXPHOS genes, suggesting that this may be specific for muscle and not necessarily an etiological factor in PCOS (33–35). Moreover, as in obesity and type 2 diabetes, it remains to be determined whether impaired OXPHOS in muscle of PCOS is a cause or consequence of insulin resistance (7).

In addition to mitochondrial dysfunction, insulin resistance in obese subjects, patients with type 2 diabetes, and their first-degree relatives is characterized by a lower proportion of oxidative, type 1 muscle fibers, reduced maximal oxygen consumption (Vo_2max), and increased intramyocellular lipid content (7,8). Moreover, lipid oxidation is impaired under basal conditions in obese and type 2 diabetic subjects (36,37). Recently, reduced Vo_2max was demonstrated in PCOS (4), and Mootha et al. (21) found that the expression of OXPHOS genes was strongly correlated with Vo_2max. Although we did not measure Vo_2max, these findings indirectly support our hypothesis of an impaired mitochondrial oxidative metabolism in muscle of PCOS. To our knowledge, no data are available concerning muscle fiber type composition, intramuscular lipid content, or leg lipid oxidation in PCOS. Using GSEA, we observed that two sets of genes representing fatty acid metabolism were the third and sixth most downregulated pathways in PCOS. Thus, it is likely that abnormalities in lipid metabolism also exist in muscle of PCOS patients. However, further studies are needed to address whether alterations in fiber type composition, increased lipid content, and impaired lipid oxidation in skeletal muscle are also components of insulin resistance in PCOS.

Expression of nuclear- and mitochondrial-encoded mitochondrial genes is thought to be coordinated by the transcriptional coactivators PGC-1α and -1β through activation of NRF-1 and -2 (38). In skeletal muscle, PGC-1α also stimulates expression of the insulin-sensitive glucose transporter GLUT4 and promotes an increased proportion of oxidative, type 1 muscle fibers (38). Patti et al. (22) provided evidence that downregulation of OXPHOS genes in muscle of patients with type 2 diabetes and their first-degree relatives is likely explained by reduced expression of PGC-1α and -1β, although these observations were made in separate study cohorts. They also found reduced levels of NRF-1 but only in patients with type 2 diabetes, indicating that this is not an early alteration associated with insulin resistance. In the present study, we show that decreased expression of PGC-1α is associated with reduced OXPHOS gene expression within the same cohort of insulin-resistant women with PCOS, whereas the expression of PGC-1β and NRF-1 was unaltered. Moreover, we provide evidence for a strong association between expression of PGC-1α and several subunits of the respiratory complexes. The correlations observed were stronger in control subjects than in PCOS subjects in whom a tight correlation between only PGC-1α and the complex I subunit NDUFA3 was seen. This could explain that, in particular, genes representing subunits of the respiratory complex I were downregulated in PCOS subjects. Our results strongly suggest that downregulation of OXPHOS genes are mediated mainly through reduced expression of PGC-1α and further point to an important role for PGC-1α in impaired oxidative metabolism in skeletal muscle insulin resistance in PCOS and the risk of type 2 diabetes in these patients.

Studies of cultured fibroblasts and myotubes from PCOS patients have provided evidence that constitutively enhanced serine phosphorylation of the insulin receptor and IRS-1 could play an important role in the pathogenesis of insulin resistance in PCOS (10–13). In cultured myotubes (11) but not in fibroblasts (39), increased serine phosphorylation of IRS-1 was associated with enhanced mitogenic signaling. In skeletal muscle, which is the major site of insulin-mediated glucose uptake, only a few significant abnormalities have been demonstrated in muscle of women with PCOS in vivo. This includes a transient decrease in insulin stimulation of IRS-1–associated PI 3-kinase activity, increased phosphorylation of extracellular signal-regulated kinase 1/2, and increased IRS-2 protein abundance (9,11). No differences in mRNA expression of IRS-1 or IRS-2 were found (9). Although similar abnormalities in proximal insulin signaling have been reported in obesity and type 2 diabetes (7,8), it has been hypothesized that the molecular mechanisms underlying insulin resistance in PCOS differ from those seen in these conditions (12). In the present study, we found that the insulin signaling pathway, as defined by GenMAPP, was among the 10 most up- and downregulated pathways in muscle of women with PCOS. A number of genes mediating metabolic and mitogenic actions of insulin and modulators of insulin action were dysregulated; however, in none of these subpathways, the genes were uniformly up- or downregulated. In recent studies of muscle transcripts, dysregulation of the insulin signaling pathway was not detected in patients with type 2 diabetes (20–22). This indicates that changes in expression of genes or proteins involved in insulin signaling, whether compensatory or not, could play a greater role for insulin resistance in skeletal muscle in PCOS than in type 2 diabetes. It also emphasizes the need for further studies of expression and activity of proteins known to mediate and modulate insulin signaling in muscle of PCOS patients.

An intriguing finding of the study was that a calcium pathway termed calcium regulation in cardiac cells was significantly up- and downregulated (GenMAPP), and calcium ion transporter activity was a significantly downregulated gene ontology term in PCOS subjects. There is increasing evidence supporting a modulating role for Ca²⁺ influx, calmodulin, and Ca²⁺/calmodulin-dependent protein kinase (CaMK) in insulin-stimulated glucose transport in skeletal muscle (40–43). In mitochondria, Ca²⁺ is important for activation of key enzymes to enhance ATP production, and an increase in cytosolic Ca²⁺ and activation of CaMK also induce mitochondrial biogenesis and GLUT4 expression via activation of different transcription factors, including NRF-1 and -2, and the coactivator PGC-1α (44–46). It is easy to infer that dysregulation of Ca²⁺ homeostasis could have a pronounced disturbing effect on insulin-stimulated glucose disposal and mitochondrial function. Abnormal Ca²⁺ homeostasis has been reported in skeletal muscle of patients with type 2 diabetes (47), and more recently, serum calcium was demonstrated to be independently associated with insulin resistance measured with euglycemic-hyperinsulinemic clamp (48). Although similar abnormalities in muscle transcripts of calcium pathway genes were not reported in studies of patients with type 2 diabetes (20–22), it is too early to conclude that the observed changes should play a unique role for insulin resistance in skeletal muscle of PCOS subjects. Further studies will be required to assess the precise implication of Ca²⁺ homeostasis in the pathogenesis of insulin resistance in PCOS.

In summary, using transcriptional profiling, we demonstrate that insulin resistance in skeletal muscle of women with PCOS is associated with reduced expression of genes involved in mitochondrial oxidative metabolism and that reduced expression of PGC-1α could play a role for these abnormalities. Moreover, our data indicate that transcriptional alterations in insulin signaling pathways, fatty acid metabolism, and calcium homeostasis may contribute to the potentially unique phenotype of insulin resistance in patients with PCOS. Future studies focusing on the interaction between these pathways in skeletal muscle may unravel the molecular mechanism of insulin resistance in PCOS and help develop strategies to prevent the increased risk of early onset of type 2 diabetes in these women.

This study has received grants from the Novo Nordisk Foundation.

We acknowledge discussions of quantitative real-time PCR data with Lene Christiansen (University of Southern Denmark).