MicroRNA-21 modulates brown adipose tissue adipogenesis and thermogenesis in a mouse model of polycystic ovary syndrome

Polycystic ovary syndrome (PCOS) is the most common endocrine disorder in reproductive-age women, with a prevalence ranging from 4–21% depending on the diagnostic criteria used [1]. PCOS is characterized by oligo- or anovulation, polycystic ovaries and clinical and/or biochemical signs of hyperandrogenism. Women with PCOS have higher incidence of multiple cardiovascular risk factors such as obesity, elevated blood pressure, renal injury, dyslipidemia, and insulin resistance [2–7]. Obesity is the most common of these risk factors, with the majority of women with PCOS being obese or overweight [8, 9] and having increased central adiposity [10]. Women with PCOS have lower postprandial thermogenesis, which is positively associated with insulin resistance [11]. Moreover, supraclavicular skin temperature is inversely correlated with serum testosterone level in PCOS women [12]; however, the detailed molecular mechanisms involved in androgen-mediated altered thermogenic responses in PCOS are not fully understood.

Brown adipose tissue (BAT) dissipates energy in the form of heat by uncoupling the mitochondrial respiratory chain and ATP synthesis, in contrast to white adipose tissue (WAT), which stores excess energy [13]. WAT and BAT dysfunction contributes to the development of obesity-related metabolic disturbances due to impairment in the regulatory circuits of fuel storage and oxidation [14]. Inducing BAT activity can mitigate obesity-related metabolic dysregulation by increasing energy utilization via thermogenesis [15, 16]. Thus, BAT plays an important role in weight management and metabolic homeostasis.

Thermogenesis in mitochondria-rich brown adipocytes involves a series of steps that begin with the release of free fatty acids (FFAs) by lipolysis and continue with their transport to the mitochondria to undergo β-oxidization by carrier proteins such as fatty acid-binding proteins (FABPs) and carnitine palmitoyltransferase 1 (CPT1) [17, 18]. Uncoupling protein 1 (UCP1) is a proton channel found in the inner mitochondrial membrane that serves to uncouple the mitochondrial electron transport chain from ATP production, directing it to thermogenesis and resulting in the release of significant amounts of chemical energy in the form of heat [19]. Therefore, UCP1 is the main thermogenic protein expressed in the BAT [20]. The BAT’s structure and thermogenic functions are regulated by a number of transcription factors. The members of CCAAT/enhancer-binding protein family (C/EBPs) and peroxisome proliferator-activated receptor gamma (PPARγ) are master transcriptional regulators of both WAT and BAT adipogenesis, which is the differentiation of adipocyte precursors (preadipocytes) to mature adipocytes [21, 22]. The thermogenic function of BAT is primarily regulated by transcription machinery that controls the expression of thermogenic genes in brown adipocytes such as peroxisome proliferative activated receptor gamma coactivator 1 alpha (PGC-1α; gene Ppargc1a) [23], PR domain containing 16 (PRDM16) [24], and cell-death inducing DNA fragmentation factor-like effector A (CIDE-A) [25]. Type 2 deiodinase (DIO2) is an important thermogenic marker that catalyzes the formation of thyroid hormone triiodothyronine (T3) from thyroxine [26], which is one of the major mechanisms controlling BAT activity by inducing mitochondrial biogenesis [27].

Dysregulation of many microRNAs (miRNAs) in metabolic tissues, including the adipose tissue, contributes to the development of obesity and its associated complications [28, 29]. MiRNAs are small, non-coding RNAs composed of 21–25 nucleotides that suppress gene expression at the posttranscriptional level via mRNA degradation or translational repression [30]. Several miRNAs modulate BAT thermogenic activity by regulating brown fat lineage determination and differentiation in a positive or negative manner [31, 32]. One miRNA of particular interest in both PCOS and obesity is miRNA-21 (miR-21). Higher levels of miR-21 are found in the WAT of obese rodents and humans [33, 34]. Furthermore, miR-21 mimic induces thermogenic gene expression in both WAT and BAT and alleviates high-fat diet-induced obesity in mice [35]. Moreover, circulating miR-21 is increased in women with PCOS [36–39]. However, the role of miR-21 in androgen-induced adipose tissue dysfunction and altered thermogenic responses in PCOS is still unknown.

Effective BAT thermogenic machinery is associated with favorable metabolic phenotypes and efficient energy expenditure [20]. Therefore, strategies aimed at increasing BAT activity and thermogenic function may be an effective therapeutic approach for attenuating the metabolic derangements observed in PCOS and other metabolic disorders with increased adiposity, such as fatty liver disease and type 2 diabetes. We aim to test the hypothesis that miR-21 controls brown adipocyte differentiation and adaptive thermogenesis in BAT by targeting key regulators of both adipogenesis and thermogenesis in a mouse model of PCOS. Our research is highly clinically relevant because it adds to our understanding of the multifunctional roles that miR-21 plays in metabolic pathways regulation by identifying its effect on the regulation of BAT thermogenic function in PCOS.

Obesity constitutes a major cardiovascular risk factor in women with PCOS [51]. Expansion of both WAT and BAT due to obesity is reported to be associated with impaired thermogenic responses in humans and animal models [14, 52, 53]. Despite similar resting energy expenditure in PCOS subjects and controls, reduced postprandial thermogenesis has been reported in both lean and obese subjects [11], indicating a possible role of androgens in altered thermogenic activity in PCOS that is independent of obesity. However, the detailed molecular mechanisms involved in this effect are not fully understood. Several miRNAs have been shown to be involved in the regulation of BAT structure and thermogenic function [31, 32]. Increased BAT thermogenic function is associated with favorable metabolic outcomes; consequently, BAT activation or transplant has been proposed as an anti-obesity therapy, and exciting studies in rodents and humans have shown beneficial effects in obesity and glucose homeostasis [54–57]. Moreover, BAT transplant or cold-mediated activation improved multiple reproductive and metabolic traits in rodent models of PCOS [58–62]. Furthermore, combined WAT, BAT, and brain AR genetic ablation protects against androgen excess-mediated obesity and hepatic steatosis in a mouse model of PCOS [63]. Therefore, targeting specific miRNAs to increase BAT function could be an effective therapeutic approach. MiR-21, which is increased in the circulation of PCOS women [36, 37], is upregulated by obesity in brown and white primary adipocytes, as well as subcutaneous and epididymal WAT [33, 34]. Furthermore miR-21 mimic induces the expression of genes involved in browning and thermogenesis in 3T3-L1 adipocytes and the BAT of obese mice [35]. Based on these premises, we performed an exhaustive study of the role of miR-21 whole body energy homeostasis and BAT molecular and histological changes triggered by excess androgens in a well-stablished mouse model of PCOS [64].

The most significant findings in our study of the role of miR-21 in a mouse model of PCOS are: (1) miR-21 genetic ablation mitigated DHT-mediated increase in body weight and lean mass, but it did not modify DHT-mediated increase in fat or BAT mass, (2) miR-21 genetic ablation attenuated DHT-mediated increase in BAT AR protein expression, (3) DHT treatment of miR21KO mice decreased O2 consumption, CO2 production, and RER, without affecting EE, (4) miR-21 genetic ablation abolished DHT-mediated decrease in caloric food intake, (5) miR-21 genetic ablation abolished DHT-mediated increase in sleep time, (6) miR-21 genetic ablation decreased the expression of some adipogenesis markers in BAT of DHT-treated mice, (7) miR-21 genetic ablation decreased the expression of some extracellular matrix remodeling markers in BAT of DHT-treated mice, (8) miR-21 genetic ablation reversed DHT-mediated increases in the expression of some genes related to thermogenesis in BAT.

There are two main types of adipose tissues in mammals: WAT and BAT. BAT is a developmentally formed adipose depot and is prominent in the interscapular region of rodent and human infants. Classical interscapular brown adipocytes are derived from myogenic Myf5-positive precursors and has high mitochondrial density and expression of thermogenic genes [20]. Brown adipocytes are also characterized by the presence of small multilocular lipid droplets serving as fuel reservoirs for mitochondrial respiration. The WAT, on the other hand, has low mitochondrial density, low expression of thermogenic proteins, and is characterized by adipocytes with a single large lipid droplet. Thereby, the main function of WAT is to store excess energy in the form of triglycerides, while the main function of BAT, on the other hand, is to mediate non-shivering thermogenesis and to dissipate excess fuel energy as heat. Therefore, BAT leads the physiological adaptation to cold and whole-body energy homeostasis [65].

The imbalance in food storage and energy expenditure associated with adipose tissue dysfunction is a major mechanism of the metabolic disturbances induced by obesity [14]. Weight gain and increased central/total body fat mass are key clinical features that characterize the majority of women with PCOS and animal models of hyperandrogenemia [10, 66–69]. According to the Androgen Excess-PCOS Society (AE-PCOS), PCOS should be first considered a disorder of androgen excess or hyperandrogenism [70]. Hyperandrogenism is present in more than 80% of women with PCOS, and those women have a worse cardiovascular and metabolic profile than normoandrogenic PCOS subjects [1, 71–73]. In the current study, female mice treated with the non-aromatizable androgen DHT from both miR21KO and WT strains had significantly higher body weight, fat mass, and BAT mass compared to their vehicle-treated littermates. The increase in body weight was attenuated in miR21KO mice due to the decrease in lean mass. BAT miR-21 expression was not modulated in response to DHT in mice in our study. Notably, androgens suppress miR-21 expression in breast cancer cells [74], whereas they increase its expression in LNCaP prostate cancer cells [75]. The findings of those and our studies suggest that androgens have a cell- and/or sex-specific effect on miR-21 expression.

In PCOS women, circulatory miRNAs, including miR-21, are differentially expressed compared to control subjects [36–39]. Moreover, androgens upregulate endometrial AR expression [76]. Our data show that the effects of DHT on body composition were associated with an increase in BAT AR protein, which was reduced in miR21KO mice, indicating that miR-21 has a positive regulatory effect on BAT AR expression. Interestingly, miR-21 can either negatively or positively regulate AR expression depending on the disease or tissue. While miR-21-3p mimics reduce AR expression in cardiac fibroblasts in a diabetic cardiac fibrosis model in male rats [77], miR-21 mimics upregulate AR protein expression in prostate cancer cells [78].

The obesogenic effect of DHT in female mice that we observed was associated with a similar reduction in EE and activity, which was represented by the significant reduction in the locomotor activity and total distance traveled in both genotypes. Androgen administration in male mice was found to decrease locomotor activity despite favorable effects on thermogenic genes expression in skeletal muscle [79]. Notably, androgen-mediated increase in sleep time in DHT-treated mice was attenuated by miR-21 genetic ablation. Women with PCOS experience longer daytime sleepiness, among other sleep disorders, compared to control individuals, likely due to increased melatonin secretion or poor mental health profile [80]. Several miRNAs have recently been proposed to control sleep–wake regulation in humans and animal experimental models [81, 82]. The lack of an effect of miR-21 on EE and locomotor activity suggests that miR-21-mediated sleep time modulation is likely due to central sleep behavioral effects rather than its effect on energy balance.

Interestingly, caloric food intake was reduced by DHT only in WT mice, while it was not affected by DHT in miR21KO mice. Energy balance is maintained when energy from food intake is equal to energy expenditure; moreover, caloric restriction is a potential strategy to counteract excessive weight gain in obesity [83]. Increased adiposity leads to the release of endocrine signals that exert negative feedback in the brain to decrease food intake in conditions of energy surplus [84, 85], hence the decreased caloric food intake in DHT-treated WT mice could be a compensatory feedback mechanism to overcome the excess energy storage following chronic DHT treatment. It is also possible that the decrease in food intake may be a secondary consequence of decreased physical activity; however, in the DHT-treated miR21KO there was no change in caloric food intake despite a similar reduction in physical activity. Interestingly, a possible direct effect of androgens on the regulation of food intake was reported [86]. The loss of this effect in the DHT-treated miR21KO mice indicates an impairment of this potential mechanism and a further energy imbalance in those mice.

To explore the mechanisms involved in increased BAT mass and its associated reduction in EE, we assessed BAT de novo lipogenesis by measuring the protein expression of two FA synthesis markers. Increased lipid deposition in the BAT induces its involution, which is characterized by a reduction in both mitochondrial mass and BAT thermogenic activity [87]; hence, inhibiting de novo lipogenesis could be a potential therapeutic target to prevent the loss of BAT thermogenic capacity. Our data indicate that ACC, the FA synthesis marker, was downregulated by DHT only in WT mice. This effect could be a compensatory mechanism to attenuate excessive adipose tissue expansion by decreasing de novo lipogenesis. Under normal diet conditions, female rats express higher BAT levels of FAS compared to males [88], explaining the reduction in FAS expression in response to androgens in the DHT-treated WT mice in our PCOS model. Moreover, sex-specific differences in BAT FAS expression tend to have opposite expression patterns in response to diet-induced obesity [88]. Excitingly, our miR21KO mice showed an opposite effect of DHT on FA synthesis markers expression, where they have a significant upregulation in BAT FAS protein levels, indicating increased lipogenesis. Notably, FAS inhibitors have been reported to inhibit food intake [89]. This could explain the genotype-specific effect of DHT on caloric food intake, which was only decreased in DHT-treated WT mice and was accompanied by comparable decreases in BAT FAS protein levels.

The role of miRNAs as critical regulators of lipid synthesis and FA oxidation has been reported [90]. Our results indicate that miR-21 inhibits de novo lipogenesis in BAT from DHT-treated mice, as FAS is upregulated by DHT in the absence of miR-21. There have been no studies on the effect of miR-21 on adipose tissue de novo lipogenesis; however, this finding is supported by miR-21's ability to inhibit lipopolysaccharide-induced lipid accumulation and lipid-laden foam cell formation in macrophages [91]. Notably, silencing miR-21 expression reverses the latter effect via targeting TLR-4 signaling [91], which is a known inducer of de novo lipogenesis in the liver [92, 93]. Moreover, lipid deposition and foam cell formation in macrophages can be mediated by de novo lipid synthesis, in addition to the well-known classical mechanism in which foam cells are produced by cholesterol uptake from oxidized LDL [94]. On the other hand, miR-21 was also reported to induce hepatic steatosis and promote the progression of NAFLD [90], implying that miR-21's effect on lipogenesis is tissue- and disease-specific.

The RER, also known as the respiratory quotient, is a useful tool for estimating the fuel source used for energy production based on the number of oxygen molecules required for the oxidation of glucose versus fatty acids. Decreasing RER values trending towards a minimum values of 0.7 indicate that fatty acids are the primary oxidative substrate, whereas increasing RER values towards a maximum value of 1.0 indicate that carbohydrates are the primary energy substrate [95]. In the light phase, the RER was significantly increased by DHT only in WT mice, indicating a tendency towards increased carbohydrate utilization. Moreover, despite no change in RER in the dark phase in WT-DHT mice, miR21KO-DHT mice showed a significant reduction in RER, indicating relying more on lipids as the energy source. The increase in BAT de novo lipogenesis (higher fatty acid synthesis markers expression), decrease in carbohydrate utilization, and increased reliance on lipids as an energy source in our miR21KO-DHT mice, as indicated by the reduced RER, could be explained by an energy surplus caused by an increase in whole-body fat content following 12 weeks of DHT treatment in our mice. This observation is supported by the finding that miR-21 modulation has no effect on cellular lipid content in H9C2 cardiomyocytes under normal nutrient conditions [96]. In the presence of lipid excess, however, miR-21-5p overexpression reduced the lipid-induced increase in cellular lipid content, whereas its suppression increased cellular lipid accumulation and decreased glucose consumption [96].

BAT adipogenesis is critical for BAT structure and thermogenic function. Decreased capacity of compensatory de novo adipogenesis is associated with increased central obesity [97]. Failure of adipocyte differentiation leads to impaired ability of the adipose tissue to expand to accommodate energy surplus leading to insulin resistance and lipodystrophy in which lipids are deposited into multiple organs across the body [98]. Our data indicate that the effect of androgens on the expression of key transcriptional regulators of adipogenesis and adipocytes differentiation, such as adiponectin, PPARγ, C/EBPα, C/EBPβ [99, 99, 99, 99, 99, 99], and MMP9 are modulated by miR-21. Adiponectin, an adipose tissue-derived adipokine, plays a protective role in various physiological conditions including thermogenesis [100]. Adiponectin enhances cold-induced subcutaneous WAT (scWAT) browning [101] and its genetic ablation showed impaired adaptive thermogenic program following cold exposure or high-fat diet -induced obesity in both scWAT and BAT [101, 102], and PPARγ expression downregulation [102]. PPARγ is another adipogenesis marker that plays a pivotal role in adipocyte differentiation, and its deficiency is associated with impaired BAT thermogenesis, lipodystrophy, and a worsened metabolic phenotype [103–105]. Despite preserved adiponectin mRNA and protein expression following DHT in WT mice, its mRNA was downregulated by DHT in miR21KO mice, which also showed significantly lower adiponectin and PPARγ protein levels compared to DHT-treated WT mice. Our data is in accordance with others indicating the potential role of miR-21 in preadipocyte differentiation. MiR-21 overexpression during adipocyte differentiation increases both adiponectin mRNA and protein expression in 3T3-L1 adipocytes [106], and the use of miR-21 mimics or its overexpression induces PPARγ expression in the BAT of obese mice [35] and human adipose tissue-derived mesenchymal stem cells (hASCs), respectively [107]. Among the C/EBPs family members, C/EBPβ enhances the differentiation of 3T3-L1 preadipocytes toward the BAT rather than the WAT phenotype [108]. While one study showed that C/EBPα is required for differentiation of white, but not brown, adipose tissue [109], other studies indicate its critical function in BAT adipocyte differentiation by regulating BAT thermogenic and adipogenic gene expression [110]. Notably, C/EBPβ regulates adipose differentiation by activating the expression of PPARγ and other adipogenic genes, including C/EBPα [111]. Interestingly, C/EBPβ mRNA expression and C/EBPα activation were only reduced in DHT-treated miR21KO mice. Moreover, the increased C/EBPβ activation (evidenced by increased phosphorylation) by DHT in the WT mice was lost by miR-21 genetic ablation, implying that C/EBPα and C/EBPβ are possible indirect miR-21 targets.

Adipose tissue structure and function are maintained with the help of extracellular matrix (ECM), and its remodeling plays a critical role in preadipocyte differentiation and adipocyte growth [112]. The imbalance between matrix metalloproteinases (MMPs) and their endogenous inhibitors, tissue inhibitors of metalloproteinases (TIMPs), is involved in obesity development via modulating adipocyte differentiation and disrupting adipose tissue ECM turnover [113]. Silencing of miR-21, but not miR-224, reduced MMP9 protein levels after cerebral ischemia [114]. Despite other studies indicating that MMP9 inhibition can reduce fat mass and adipose tissue hypertrophy, pharmacologic inhibition of MMP2 and MMP9 reduces preadipocyte differentiation in vitro and reduce PPARγ induction [115, 116]. In addition to its reported regulatory effect on adipogenic differentiation [107], miR-21 can modulate angiogenesis through the modulation of VEGF-A [117], which has a positive effect on BAT thermogenic function [50]. Despite the lack of significant changes in VEGF-A expression by DHT or miR-21 genetic ablation, MMP9 expression was downregulated only in DHT-treated miR21KO mice. Our data suggest that miR-21 function is required for the proper development and differentiation of brown adipocytes and, in turn, its thermogenic function. Our study shows for the first time the effect of miR-21 genetic ablation on BAT adipogenesis in PCOS.

The fatty acid-binding proteins (FABPs) are a family of small cytoplasmic proteins, that are highly expressed in both adipocytes and macrophages and hence play an important role in inflammation related to lipid metabolism [118]. The adipocyte-FABP (also known as FABP4 or aP2) functions as a lipid chaperone that regulates FFAs trafficking and signaling and reduction in its activity or expression is associated with a metabolically favorable phenotype in human and animal models [119, 120]. Our mice did not show any significant changes in FABP4 expression by DHT in either genotype; however, FABP4 basal levels were higher in vehicle-treated miR21KO mice compared to their WT counterparts.

A number of transcription factors are known to play a pivotal role in the regulation of adaptive thermogenesis on the transcriptional level. In addition to its ability to trigger myogenic (Myf5⁺) and white adipose tissue precursor’s differentiation into brown and beige adipocytes, respectively, PRDM16 induces the expression of UCP1 and PGC-1α [121, 122]. PGC-1α is another transcriptional coactivator that regulates mitochondrial biogenesis and oxidative metabolism [123]. UCP1 overexpression in white adipocytes can drive adipocyte beiging [123]. CIDE-A is another lipid-droplet-associated protein that transcriptionally regulates UCP1 function [25]. Our data indicate no changes in the expression of either PRDM16 or PGC-1α by DHT in both genotypes. However, despite higher basal CIDE-A levels in miR21KO mice, both the mRNA and protein levels of this transcription factor were significantly downregulated by DHT only in DHT-treated miR21KO mice, which also showed lower CIDE-A protein levels compared to DHT-treated WT mice. Despite a lack of studies on the effect of miR-21 role on the expression of thermogenic targets in PCOS, the use of miR-21 mimics in obesity induced by high-fat diet was associated with upregulation in multiple genes involved in thermogenesis, including PRDM16, PGC-1α, UCP1, and CIDE-A [35].

Thyroid hormones constitute a major mechanism for controlling BAT activity. DIO2 enzyme mediates the formation of the T3 from thyroxine in the BAT [26]. T3 transcriptionally regulates UCP1 and other mitochondrial biogenesis genes including PGC-1α, nuclear respiratory factor 1 (NRF1), and cytochrome c (Cyt c) by binding to its nuclear receptor in brown adipocytes [27, 124–126]. DIO2 activation is associated with increased energy expenditure in BAT [127], and its deletion is associated with impaired BAT thermogenesis following cold exposure despite normal plasma T3 levels [126]. Our data show that UCP1 levels did not change in response to DHT; however, downregulation of DIO2 mRNA levels was observed in both genotypes. Our findings are consistent with those showing that despite a normal plasma T3 concentration and normal basal UCP1 levels, cold-exposed DIO2 knockout (DIO2-KO) mice had impaired BAT thermogenesis. In the same study, brown adipocytes from DIO2-KO mice showed an attenuation in norepinephrine-induced increase in UCP1 mRNA due to impaired cAMP generation [126]. In our study, the effect of DHT on DIO2 expression was similar in both miR21KO and WT mice. Notably, miR-21 positively regulates DIO3 while it has no effect on DIO2 in basal cell carcinoma cells [128].

Steps involved in mitochondrial thermogenesis include the release of FA from lipid droplets by lipolysis followed by its transport to the inner mitochondrial membrane to be involved in β-oxidation, a major step involved in BAT thermogenesis [129]. Long-chain fatty acids (LCFA), constitute a major fraction of fatty acids delivered to target tissues [130]. LCFAs activate UCP1’s heat generation function in the BAT [131]. Elovl3, also known as cold-induced gene (Cig30), is a member of a mammalian gene family that is involved in the biosynthesis of very long chain fatty acids (VLCFAs) in the C20-C24 range and is mainly expressed in liver, skin, and BAT [132]. In addition to playing a critical part in the development and recruitment of brown adipocytes, Elovl3 is also substantially elevated to function in thermogenesis after exposure to cold [49]. The levels of Elovl3 are correlated with the oxidative capacity of FA in the brown adipocytes [133]. Our data indicates similar downregulation of Elovl3 gene expression in the BAT of both miR21KO and WT genotypes, indicating another mechanism of impaired thermogenic responses in PCOS independent of miR-21 levels. FA release by lipolysis is under control of perilipin, a protein family that coats the lipid droplets of adipocytes and act as barrier against the lipolytic action of lipases [134, 135]. Perilipin-1, the predominant isoform of perilipin expressed in white and brown adipocytes [136], was not modified by DHT treatment or miR-21 genetic ablation.

The carnitine palmitoyltransferase 1 (CPT1) is the rate-limiting enzyme in mitochondrial FA oxidation as it facilitates mitochondrial FA transport for β-oxidation. CPT1B is the most abundant isoform in BAT; however, increased CPT1A isoform expression in brown adipocytes is also reported to increase FA oxidation, lipolysis, and UCP1 activity [137]. Our data show that both CPT1A mRNA and CPT1B protein expression were upregulated by DHT in WT mice, and this effect was lost by miR-21 genetic ablation. These effects could be attributed to the attenuation in DHT-mediated upregulation in AR in DHT-treated miR21KO mice. Notably in male mice, AR is involved in energy homeostasis as its blockade results in lower energy efficiency and fat metabolism following chronic exercise by suppressing CPT1 expression in the skeletal muscle [138]. Therefore, the increase in CPT1 in our DHT-treated WT mice could be a homeostatic mechanism to maintain energy balance in response to obesity by burning more FAs.

The terminal and rate-limiting step of mitochondrial respiration is the reduction of molecular oxygen to water via cytochrome c oxidase (COX) using cytochrome c as substrate [139]. COX, a complex of the mitochondrial respiratory chain, is a dimer consisting of 13 individual subunits per monomer [140]. We assessed the expression of multiple COX subunits of significant expression and functional activity in the BAT, including Cox4i1, Cox4i2, Cox7a1 [141], and Cox8b [142]. Our data indicate that although DHT did not induce any significant changes in the assessed COX subunits in WT mice, COX4 protein expression was significantly downregulated by DHT in miR21KO mice, indicating less mitochondrial activity and hence BAT thermogenic function regulation by miR-21 genetic ablation. Several miRNAs are reported to modulate electron transport chain components, including COX4 in a negative or positive manner [46]. While miR-338 downregulates COX4 leading to reduced mitochondrial activity in sympathetic neurons [47], knocking down miR-182 and miR-203 leads to lower COX7 and COX8 expression in primary brown adipocytes [48]. Our data is the first to indicate a possible regulatory effect of miR-21 on COX4 expression in the BAT in a PCOS model.

Adaptive de novo lipogenesis inhibition and increased expression of adipogenesis and thermogenic markers is a mechanism for achieving energy balance in conditions of obesity and energy surplus. In our PCOS model, androgen-induced obesity was associated with the downregulation of lipogenesis markers, increased adipogenesis (higher C/EBPβ activity), and upregulation of CPT1A and CPT1B key thermogenesis and mitochondrial biogenesis markers. MiR-21 genetic ablation in DHT-treated mice, on the other hand, increased de novo lipogenesis while inhibiting BAT differentiation, as evidenced by a significant decrease in mRNA and/or protein expression of key BAT adipogenic genes such as adiponectin, PPARγ, and C/EBPβ, which was accompanied by downregulation of CIDE-A and COX4 thermogenic markers. In addition, the adaptive increase in CPT1A and CPT1B was lost upon miR-21 genetic ablation. Several miRNAs have been reported to regulate brown fat differentiation and thermogenic function, either positively or negatively. However, this is the first study to investigate the role of miR-21 in altered thermogenic responses caused by excess androgens in a PCOS model, with miR-21 genetic ablation abolishing androgen-mediated beneficial effects on BAT thermogenic markers expression. The latter effect could be explained in part by attenuating androgen-mediated BAT AR upregulation in DHT-treated miR21KO mice. In males, AR activation has been shown to have beneficial thermogenic functions [138]. On the other hand, androgens have been reported to have deleterious effects in WAT in females. Androgens, via AR activation, inhibit adipogenesis in non-obese women's human adipose stem cells (hASCs) [143], which is postulated to promote ectopic lipid deposition and lipotoxicity, and its administration to women inhibits subcutaneous abdominal adipose lipolysis by inhibiting hormone-sensitive lipase [144]. It is possible that the negative obesogenic effects of long-term hyperandrogenemia on total body fat mass are mitigated in part by the positive AR thermogenic action in BAT. This conclusion is supported by data demonstrating disease and cell-specific differential regulation of AR and miR-21. While miR-21 downregulates AR leading to a deleterious effect on cardiac fibrosis [77], in prostate cancer, miR-21 increases AR protein expression and is involved in tumor pathogenesis [78]. Future studies using depot-specific adipose tissue modulation of miR-21 levels are needed to identify specific androgen effects on WAT and BAT, which could be a novel therapeutic approach for the treatment of PCOS-associated metabolic derangements.

Our findings strongly suggest that miR-21 function is required for normal BAT adipogenesis in addition to its roles in thermogenic machinery and whole-body homeostasis. Moreover, our findings suggest that miR-21 plays a critical role in the molecular adaptive responses of BAT to hyperandrogenemia in PCOS. Furthermore, increased circulatory miR-21 levels in PCOS could be a compensatory mechanism that helps mitigate systemic androgen-mediated metabolic dysfunction, providing a new theoretical foundation for the development of miRNA-based therapeutics for PCOS-related adipose tissue dysfunction.