Ovarian Cycle-Specific Regulation of Adipose Tissue Lipid Storage by Testosterone in Female Nonhuman Primates

Abstract

Previous studies in rodents and humans suggest that hyperandrogenemia causes white adipose tissue (WAT) dysfunction in females, although the underlying mechanisms are poorly understood. In light of the differences in the length of the ovarian cycle between humans and rodents, we used a nonhuman primate model to elucidate the effects of chronic hyperandrogenemia on WAT function in vivo. Female rhesus macaques implanted with testosterone capsules developed insulin resistance and altered leptin secretion on a high-fat, Western-style diet. In control visceral WAT, lipolysis and hormone-sensitive lipase expression were upregulated during the luteal phase compared with the early follicular (menses) phase of the ovarian cycle. Hyperandrogenemia attenuated elevated lipolysis and hormone-sensitive lipase activity in visceral WAT during the luteal phase but not during menses. Under control conditions, insulin-stimulated Akt and Erk activation and fatty acid uptake in WAT were not significantly affected by the ovarian cycle. In contrast, testosterone treatment preferentially increased fatty acid uptake and insulin signaling at menses. The fatty acid synthase and glucose transporter-4 genes were upregulated by testosterone during the luteal phase. In summary, this study reveals ovarian stage-specific fluctuations in adipocyte lipolysis and suggests that male sex hormones increase and female sex hormones decrease lipid storage in female WAT.

Polycystic ovary syndrome (PCOS) is one of the most common endocrine disorders of reproductive age associated with infertility, menstrual cycle disturbances, hyperandrogenemia, and the metabolic syndrome (1). The latter includes increased visceral (V) adiposity and insulin resistance (2–4). Because testosterone (T) administration lowers insulin sensitivity in healthy women, it is believed that hyperandrogenemia is the leading cause of insulin resistance in PCOS women (5–7). The development of PCOS symptoms in lean women suggests that PCOS is not necessarily associated with obesity (8). Lean PCOS patients, however, are more insulin-resistant than lean controls (9). Furthermore, weight loss results in significant improvements of ovarian and metabolic symptoms in PCOS patients (8).

Although the mechanisms of androgen-dependent metabolic syndrome remain largely unknown, increasing evidence suggests a leading role for white adipose tissue (WAT) in the pathogenesis of PCOS. Chronic hyperandrogenemia can cause functional alterations in female WAT. For example, isolated subcutaneous (SC) adipocytes from women with PCOS displayed impaired insulin sensitivity, but normal insulin receptor binding and insulin-stimulated Akt phosphorylation (10–12). However, these studies examined adipocytes only from SC WAT and not from V WAT. Insulin inhibition of lipolysis, another measure of insulin sensitivity in adipocytes, is impaired in SC adipocytes from women with PCOS (13, 14). T reduced catecholamine-induced lipolysis and hormone-sensitive lipase (HSL) expression in differentiated human preadipocytes from SC WAT but not in V WAT (15). Furthermore, in V WAT of women with PCOS, protein levels of the CD36 free fatty acid (FFA) transporter were increased, and levels of HSL were decreased, whereas levels of the adipocyte triglyceride lipase (ATGL) were unchanged compared with controls (16).

Ovarian hormones regulate body weight, insulin sensitivity, and lipolysis in females (17–20). Because the effect of the ovarian cycle on WAT in the setting of hyperandrogenemia has not been described, we explored the effects of chronic T exposure on WAT function in female nonhuman primates (NHPs) at the different stages of the ovarian cycle under conditions of diet-induced obesity.

Materials and Methods

Animals

Starting prepubertally, female rhesus macaques received either control (cholesterol [C]; n = 6) or T implants (n = 6) as previously described (21). For ovarian cycle studies, C and T groups were subdivided into equal subgroups (n = 3). The 4 resultant experimental groups were luteal-phase control (LC), menses control (MC), luteal-phase testosterone-treated (LT), and menses testosterone-treated (MT). Animals were housed individually, and menses were detected daily, starting at 1 year of age, by swabbing the vaginal area with a cotton-tipped swab as described (21). There were no acyclic females used in this study, and there was no difference in the total length of the last complete menstrual cycle before tissue collection (Table 1) as well as the length of follicular and luteal phases of the cycle (22). Ovarian function of experimental animals was evaluated as described previously (22). All procedures described in this study were approved by the Oregon National Primate Research Center (ONPRC) Institutional Animal Care and Use Committee.

Table 1.

Metabolic and Hormonal Characteristics of Experimental Animals^a

	Menses				Luteal Phase		Menses		Ovarian Cycle Length, d
	Fasting Insulin, mU/L	Glucose, mg/dL	Insulin Sensitivity, min−1/(mU/L)	T, ng/mL	E, pg/mL	P, ng/mL	E, pg/mL	P, ng/mL
Control	14.3 ± 1.8	60.2 ± 2.1	13.9 ± 3.7	0.1 ± 0.0	59.7 ± 8.1	7.0 ± 2.9	51.3 ± 3.1	0.2 ± 0.1	27 ± 0.7
T-treated	19.0 ± 2.7	58.2 ± 4.1	5.3 ± 1.9b	1.6 ± 0.2c	46.3 ± 6.4	3.7 ± 0.8b	51.3 ± 11.4	0.1 ± 0.1	27 ± 1.0

^aHormonal and metabolic measurements were conducted during the luteal phase or at menses. Values are presented as mean ± SEM (n = 6). For estrogen and progesterone measurement, the groups were further subdivided into ovarian cycle-specific groups (n = 3). Significant changes between control, C-treated and T-treated groups were determined by t test. Ovarian cycle length was determined as described previously (21).

^bP < .05.

^cP < .01.

Diets

Regular diet consisted of 2 daily meals of Purina LabDiet fiber-balanced monkey chow (15% calories from fat, 27% from protein, and 59% from carbohydrates; no. 5000; Purina Mills), supplemented with fruits and vegetables. A Western-style diet (WSD; 33% calories from fat, 17% from protein, 51% from carbohydrates; 5A1F; Purina Mills) was given ad libitum and each meal was supplemented with a high-sugar treat. Previous studies have shown that a WSD accelerates the development of obesity in rhesus macaques within the first 6 months (23, 24).

T implants

T implants have been previously described (21, 22). Normal T levels in prepubertal female rhesus macaques were determined using 1-year-old female rhesus macaques in the ONPRC colony. Their average T value (0.4 ng/mL) was multiplied by 3 to achieve the lower limit (1.2 ng/mL) and by 4 to achieve the upper limit (1.6 ng/mL) of target values in the T-treated animals. This was based on the clinical evidence that obese girls with hyperandrogenemia and women with PCOS have T levels about 3- to 4-fold higher than controls (25–27). T concentrations in serum were monitored weekly, and implants were replaced as described (21).

Hormone assays

T, leptin, and adiponectin levels were measured in the ONPRC Endocrine Technologies Support Core by ELISA (Immuno-Biological Laboratories, Inc). The sensitivity of the T assay was 0.083 ng/mL, and the intra- and interassay coefficients of variation for the assays were 8.9% and 13.5%, respectively. Serum estradiol and progesterone concentrations were determined weekly as described (21). Both hormones were assayed using an Immulite 2000 platform (Siemens Healthcare Diagnostics). As with many validated clinical platforms, the Immulite 2000 runs 3 quality control (QC) serum pools daily, and so no specific intra-assay QC data are available. The interassay coefficient of variation, reflecting variability in daily QC results over the period in which these assays were performed was 8.5% for estradiol and 9.4% for progesterone. Fasting serum insulin and glucose levels were determined as described (28).

Glucose tolerance tests

Glucose tolerance testing (GTT) was performed during the early follicular phase of the menstrual cycle when the animals were 6.5 years old and had been on the WSD for 12 months (the GTT method was previously described in Ref. 21). Each animal was sedated initially with telazol (tiletamine hydrochloride and zolazepam hydrochloride; Fort Dodge Animal Health) and subsequently with ketamine to maintain sedation. The protocol was based on that designed by Bergman et al (29). Dextrose (300 mg/kg) was infused iv through a catheter, and blood samples were taken from 15 minutes before to 3 hours after the glucose infusion. Tolbutamide (5 mg/kg) was infused iv 20 minutes after the dextrose to stimulate insulin secretion. All samples were immediately assayed for glucose using a YSI 2300 Stat Plus (YSI Inc) and subsequently for insulin by RIA (Linco human insulin RIA; Millipore Corp). The sensitivity of the insulin assay was 1 μU/mL, and the intra-assay coefficient of variation was 4.9%.

Dual-energy x-ray absorptiometry scanning

Percent body fat was determined using dual-energy x-ray absorptiometry (DEXA) scanning as described (21). DEXA measurements were not controlled for the ovarian cycle stage. Monkeys were sedated with ketamine and positioned supine on the bed of a Hologic DEXA scanner (Discovery scanner; Hologic Inc). DEXA scanning was performed immediately before starting the WSD when the animals were 5.5 years old and again when the animals had been on the WSD for 16 months.

Western blotting

WAT procedures were described previously (30). Briefly, at necropsy, ∼2 g of V WAT (omental) and SC WAT (lower abdominal area) were collected in 50-mL tubes filled with 20 mL M199 medium (Invitrogen) (pH 7.4) at room temperature. Insulin and isoproterenol treatments of WAT were started within 30 minutes of necropsy. For Western blotting studies of insulin signaling, 100 ± 10-mg WAT explants were placed into a 12-well culture dish containing 2 mL incubation medium (M199 medium, 0.1% fatty acid-free BSA [Sigma-Aldrich], 20 mM HEPES [pH 7.4], supplemented with penicillin, streptomycin, and fungizone) and incubated at 37°C for 2 hours in an atmosphere of 5% CO₂ at 37°C in the presence of indicated concentrations of human insulin (Sigma-Aldrich). WAT was homogenized using a hand-held pestle motor homogenizer, boiled for 5 minutes, and centrifuged for 10 minutes to separate lipids (top layer) and cell membranes (pellet). Then 170 μL of liquid interface was transferred to fresh 1.5-mL tubes, supplemented with 0.1% bromophenol blue and 100 mM dithiothreitol, and boiled for 4 minutes, and 20–40 μg of proteins were loaded onto precast 10% SDS-PAGE gels (Bio-Rad). Protein concentration was determined before bromophenol blue and dithiothreitol were added to the tubes with a DC protein assay kit (Bio-Rad). Western blotting was performed using primary antibodies to Akt (Cell Signaling Technologies), Ser473-Akt (Invitrogen), Erk (Cell Signaling Technologies), and Thr202/Tyr204-Erk (Cell Signaling Technologies), HSL and ATGL (Santa Cruz Biotechnology) at a 1:500 dilution, and goat antibodies to actin (Santa Cruz Biotechnology) at a 1:5000 dilution. Secondary horseradish peroxidase-conjugated antibodies were used at a 1:5000 dilution. For quantification of in vivo HSL and ATGL levels and basal phosphorylation of Erk, WAT tissue obtained at necropsy was snap-frozen in liquid nitrogen and processed as described above.

Fluorescent microscopy

For imaging studies, small WAT explants were incubated free-floating in plastic 8-well chambers (Lab-Tek II chambered no. 1.5, German coverglass system; NalgeNunc International) filled with 0.4 mL incubation medium supplemented with indicated concentrations of insulin. BODIPY-500/510 C₁, C₁₂ (BODIPY-C12; Invitrogen) was prepared in advance by diluting a 2.5mM methanol stock solution in incubation medium to a final concentration of 10μM. To allow BODIPY-C12 to bind BSA, the 10μM BODIPY-C12 solution was incubated for 15 minutes, protected from light, in a 37°C water bath. During this incubation step, 2 μL ethidium homodimer (LIVE/DEAD Viability/Cytotoxicity Kit; Invitrogen) were mixed into each well containing WAT explants and incubation continued for 15 minutes. Eight-well chambers containing WAT explants free-floating in 0.4 mL incubation medium (with ethidium homodimer and hormones) were removed from the incubator, 100 μL of 10μM BODIPY-C12 solution was added to each well, mixed by repeated pipetting, and the chambers were incubated for an additional 10 minutes at 37°C. Labeling reactions were placed on ice, and WAT explants were brought to the bottom of the chamber with 8 × 8-mm squares of light stainless steel mesh (0.4 mm; TWP, Inc). Medium was removed by aspiration, and tissue was washed 3 times with ice-cold incubation medium. WAT explants were fixed at room temperature with 4% paraformaldehyde in PBS for 20 minutes, washed 4 times with PBS, and stored in PBS at 4°C, protected from light, for up to 48 hours before confocal analysis. Image recording and cell size determination were conducted using a Leica SP5 AOBS spectral confocal system as described (28).

WAT lipolysis

For lipolysis studies, 100 ± 10-mg WAT explants were placed into a 24-well culture dish containing 0.5 mL incubation medium (phenol red-free DMEM [Invitrogen], 0.5% BSA [Sigma-Aldrich], 20 mM HEPES [pH 7.4]), and incubated at 37°C free-floating for 2 hours in an atmosphere of 5% CO₂ at 37°C. Glycerol release was determined using the glycerol detection kit (Zen-Bio).

RNA isolation and quantitative RT-PCR analysis of gene expression

RNA was prepared from 100 mg whole WAT using an RNeasy lipid kit (QIAGEN, Inc). Equal amounts of input RNA were used for all PCR. RNA was reverse-transcribed using random-hexamer primers and treated with deoxyribonuclease. Quantitative RT-PCR (qRT-PCR) was performed using SYBR Green-I reagent, transcript-specific primers, and actin primers as an endogenous control. The Delta-Delta-Ct (ddCt) algorithm was used to calculate fold difference in transcript levels between samples, and amplification efficiencies for all primer pairs were verified to be equivalent over a range of template concentrations. qRT-PCR was performed using an ABI7900 thermocycler (Applied Biosystems, Inc).

Data analysis

Statistical analysis was performed using Prism version 4 (GraphPad Software, Inc) as indicated in the figure legends. Single-cell intracellular fluorescence associated with BODIPY-C12 was determined as previously described (28). The MINMOD Millennium computer program was used to determine glucose insulin sensitivity values (31). This program was designed to calculate these values based on the GTT protocol that was described by Bergman et al (29) and used in this study.

Results

To assess the effect of hyperandrogenemia on the metabolic function of WAT, female rhesus macaques received chronic, 6-year T or C treatments, starting prepubertally at 1 year of age (21). To induce diet-induced obesity, both experimental groups were switched, at 5.5 years of age, from a low-fat chow diet to a WSD rich in saturated fat and continued receiving T/C treatments for 18 months. Fat mass, the percentage of body fat gained on a WSD, and ovarian cycle length were not statistically different between C and T groups (Figure 1, A–C). However, T-treated animals showed reduced insulin sensitivity and reduced serum progesterone levels during the luteal phase (Table 1). The levels of circulating leptin were significantly higher in T-treated animals than in control animals receiving the WSD for 18 months (Figure 2B, shaded fields). Interestingly, the T and C groups had similar leptin levels before introduction of a WSD (Figure 2B, open fields). Total and high-molecular-weight serum adiponectin levels were slightly reduced in the T-treated groups compared with control animals before and during the WSD (Figure 2, C and D). Hyperandrogenemia thus exacerbated the effects of a WSD on insulin resistance and hyperleptinemia without any significant effect on adiposity.

Figure 1.

Chronic hyperandrogenemia does not accelerate diet-induced obesity in female rhesus monkeys. Panel A, Representative DEXA scans of control, C-treated and T-treated animals were performed immediately before starting the WSD, and again when the animals have been on the WSD for 16 months (see Materials and Methods for details). Images were analyzed as described in Materials and Methods. Panels B and C, Quantification of DEXA scans shows percent body fat gained on a WSD (panel B) and percent central fat after a WSD (panel C). The graphs show means ± SEM (n = 6). No significant differences between C- and T-treated groups were detected by t test. Panel D, The morphological appearance of V WAT explants stained with BODIPY-C12. Arrows indicate smaller, more active adipocytes that accumulate higher amounts of BODIPY-C12 than larger adipocytes (28). BV, blood vessels. Each quadrant is 1 mm across. Panel E, The size of V and SC adipocytes from LC, MC, LT, and MT animals. The graphs show means ± SEM (n = 3). No significant differences between C- and T-treated groups were detected using two-way ANOVA followed by Bonferroni post hoc t test.

Figure 2.

Chronic hyperandrogenemia exaggerates the effect of a WSD on leptin secretion in female rhesus monkeys. Panel A, Experimental design: beginning at 1 year of age, animals were chronically treated with T (closed circles) or C (open circles). At 5.5 years of age, animals were switched to a WSD and continued receiving T and C treatments for 18 months (year 7). Panels B–D, The levels of serum leptin (panel B), total adiponectin (panel C), and high-molecular-weight adiponectin (panel D) were determined before introduction of a WSD (years 1 to 5.5) and during 18 months on a WSD (years 5.5 to 7, shaded areas). The graphs show means ± SEM (n = 6). *, P < .05, two-way repeated-measures ANOVA followed by Bonferroni post hoc t test.

Morphological and functional studies of WAT were conducted using V WAT and SC WAT explants obtained after 18 months on the WSD. Animals were subdivided into groups according to the stage of the ovarian cycle (luteal phase or at menses), and adipocyte size was determined using a fluorescently labeled FFA tracer (28). Adipocyte size was not statistically different between LC, MC, LT, and MT groups, although the latter showed a trend to larger adipocytes (Figure 1, D and E). Hyperandrogenemia thus had no significant effect on the development of diet-induced hypertrophy of V and SC WAT.

Basal and isoproterenol-stimulated lipolysis, as a function of glycerol release, was assessed in vitro, using freshly isolated WAT explants. The magnitude of basal lipolysis in both WAT depots varied depending on the stage of the menstrual cycle and previous T treatment. In the control group, glycerol levels were elevated during the luteal phase and reduced at menses in V WAT and SC WAT (Figure 3, A and D, open bars). In V WAT, but not in SC WAT, in vivo T treatment greatly attenuated the stimulatory effect of the luteal phase on lipolysis (Figure 3, A and D, filled bars). The expression of HSL in V WAT during the menstrual cycle in T-treated and control animals positively correlated with the magnitude of lipolysis (Figure 3, A and C). The expression of ATGL in V WAT and the magnitude of the isoproterenol-stimulated lipolysis in both depots showed no correlation with the stage of the menstrual cycle (Figure 3, F, B, and E). Due to the small amount of SC WAT in the experimental animals, HSL and ATGL levels in this depot were not determined.

Figure 3.

Ovarian cycle-specific regulation of lipolysis in WAT of female rhesus monkeys. V WAT (panels A–C and F) and SC WAT (panels D and E) explants were incubated in vitro in the absence (panels A and D) or presence (panels B and E) of 10 μM isoproterenol, and glycerol release (panels A, B, D, and E) and the expression of HSL (panel C) and ATGL (panel F) in V WAT were determined as described in Materials and Methods. The graphs show means ± SEM (n = 3). *, P < .05, two-way ANOVA followed by Bonferroni post hoc t test. Abbreviation: a.u., arbitrary units.

Insulin-dependent FFA uptake in adipocytes relies on activation of Akt and Erk signaling pathways (32–34). To investigate the potential ovarian-specific regulation of insulin signaling, V WAT explants obtained after 18 months on the WSD were incubated with various doses of insulin ex vivo, and the levels of activated phospho-Akt and phospho-Erk were measured by Western blotting. In parallel experiments, FFA uptake into WAT was quantified using a fluorescently labeled FFA tracer (Figure 1D) (28). Under control conditions, insulin-dependent Akt and Erk phosphorylation and FFA uptake in V WAT were not significantly different between the 2 stages of the ovarian cycle (Figure 4, A–C, MC and LC). In contrast, in vivo T treatment preferentially increased Akt phosphorylation in response to insulin (Figure 4A, MT). The increase in Akt phosphorylation correlated with enhanced FFA uptake in V WAT, in a dose-dependent manner, at menses but not in the luteal phase (Figure 4C, MT and LT). T treatment also preferentially increased Erk phosphorylation in response to a physiological, 100pM concentration of insulin (Figure 4B, MT). T treatment also enhanced FFA uptake in SC WAT at menses but not during the luteal phase (data not shown). Thus, in the absence of ovarian hormones, previous in vivo T exposure enhanced insulin signaling and increased FFA uptake in WAT.

Figure 4.

Ovarian cycle-specific regulation of insulin signaling and FFA uptake in WAT of female rhesus monkeys. V WAT explants were incubated in vitro in the presence of indicated concentrations of insulin, and phospho-Akt (pAkt) (panel A), phospho-Erk (pErk) (panel B), and FFA uptake (panel C) in adipocytes were determined as described in Materials and Methods. The graphs show means ± SEM (n = 3). Statistical differences between groups were detected by two-way ANOVA followed by Bonferroni post hoc t test. *, P < .05; ***, P < .001.

To assess whether adipogenic genes are regulated during the ovarian cycle, we analyzed expression of WAT genes involved in adipogenesis and lipid and glucose metabolism. In V WAT, the fatty acid synthase (FASN) and glucose transporter-4 (GLUT4) genes were significantly upregulated by T exposure during the luteal phase but not at menses (Figure 5A). In contrast, leptin gene expression was elevated by T exposure at menses but not during the luteal phase (Figure 5A). In SC WAT, none of the genes examined were affected by T exposure, although the leptin gene expression was modestly increased during menses, but the difference did not achieve statistical significance (Figure 5B). Thus, chronic T exposure, combined with a WSD, leads to the upregulation of liponeogenic and leptin gene expression in V WAT of NHPs.

Figure 5.

Chronic hyperandrogenemia stimulates lipogenic gene expression in female V WAT. mRNA from V WAT (panel A) and SC WAT (panel B) explants from MC (open bars), MT (light gray bars), LC (dark gray bars), and LT (black bars) animals was subjected to qRT-PCR gene expression analysis as described in Materials and Methods. Values are expressed as fold change relative to MC levels (1.0). The graphs show means ± SEM (n = 3). Statistical differences between groups were detected by two-way ANOVA. *, P < .05.

Discussion

The present study identifies ovarian cycle-dependent fluctuations in the lipolytic activity of WAT. It appears that factors released during the luteal phase enhance lipolysis via induction of HSL in WAT. Taken together, our data suggest that the ovarian hormones estrogen and/or progesterone are involved in the observed cycle-specific regulation of lipolysis. Our data are consistent with previous reports demonstrating that estrogens protect rodents and humans from high-fat diet-induced obesity and insulin resistance by stimulating adipocyte lipolysis (17–19). Studies in humans suggest that estrogens can either inhibit (35) or stimulate (18) lipolysis. Progesterone was reported to be a potent lipolytic hormone as well (36, 37). Because progesterone levels were significantly reduced during the luteal phase after T treatment (Table 1), we propose that lower progesterone levels, possibly in combination with hyperinsulinemia, contribute to reduced lipolysis in V WAT of PCOS patients.

We hypothesize that the natural fluctuations of ovarian hormones are essential for an optimal lipolytic response of WAT in females. That rodents and primates significantly differ in the length of the ovarian cycle (4 vs 27 days on average) raises the question whether rodents represent an appropriate model of the human reproductive cycle. The present NHP study is more relevant to describing the natural lipolytic state of human WAT in vivo. Because animals were exposed to the same physical environment, the present approach eliminates the variability due to possible confounding differences in diet and lifestyle seen in human patients.

Given that hyperandrogenemia in women is associated with insulin resistance (5, 6) and, in some cases, with obesity (2, 38, 39), the present study uncovers a potential pathophysiological mechanism of metabolic changes seen in women with PCOS. Our demonstration that in vivo T exposure suppressed the potentiating action of the luteal phase on V WAT lipolysis and HSL expression further expands previous studies that showed reduced levels of HSL and suppressed lipolysis in V WAT and SC WAT of PCOS patients (15, 16). However, HSL gene expression changes in response to T treatment did not correlate with changes in protein levels, suggesting that the ovarian cycle may control HSL protein or mRNA stability in adipocytes.

Although a defect in adipocyte lipolysis may explain obesity associated with PCOS, almost half of PCOS patients are not obese (40). The present study also demonstrated that chronic T treatment triggers insulin resistance without overt obesity. Although adipocyte size was not significantly affected by T treatment, the increased levels of circulating leptin and Erk phosphorylation observed in T-treated animals signal the ongoing development of WAT dysfunction. Because Erk activation has been associated with WAT hypertrophy and obesity (41, 42), the present report, for the first time, links chronic hyperandrogenemia to Erk activation in female WAT.

The amount of triglyceride stored in WAT is influenced by the opposing action of lipolysis and insulin-dependent FFA uptake. To test whether the ovarian cycle influences insulin action, we measured activation of Akt and Erk, the principal kinases in the major signaling pathways downstream of the insulin receptor that are known to regulate FFA uptake in adipocytes (33). In control animals, both kinases showed similar levels of phosphorylation during the luteal phase and at menses. In vivo T treatment, however, had a striking effect on Akt activation in that it enhanced the magnitude of insulin-mediated Akt phosphorylation and simultaneously accelerated FFA uptake in WAT (Figure 4). This concurrent enhancement of FFA uptake and insulin signaling via Akt was specific for menses, suggesting that luteal-phase hormones may antagonize the effect of T on insulin signaling in WAT. This finding, to the best of our knowledge, is the first demonstration of the ovarian cycle stage-specific regulation of insulin signaling by female and male sex hormones in WAT. Erk activation observed only at the lower dose of insulin (Figure 4B, MT) suggests that higher doses of insulin suppress Erk phosphorylation, possibly through the activation of an unknown phosphatase or via the negative feedback influence of the Akt pathway on Erk phosphorylation. Because Erk activation was suppressed at higher doses of insulin, it is unlikely that the Erk pathway significantly contributes to FFA uptake in female V WAT of NHPs.

We previously observed that 48-hour ex vivo T treatment of WAT explants from ovariectomized, but not from intact, females resulted in elevated FFA uptake in adipocytes (Figure 5B in Ref. 30). Although this previous study recapitulated the effects of the ovarian cycle on WAT lipid uptake in vitro, the present work provides additional evidence of the opposing regulation of lipid uptake and insulin signaling by female and male sex hormones in vivo. Gene expression analysis revealed that T treatment enhanced, in an ovarian stage-dependent manner, the expression of a subset of adipogenic genes in V WAT. In contrast, FFA uptake, insulin signaling, and leptin gene expression were enhanced at menses. We speculate that T plays a key role in priming V WAT for a greater lipogenic response during menses.

Although our findings reveal new mechanisms by which hyperandrogenemia contributes to WAT metabolism, the exact etiology of PCOS remains unclear. There is evidence for the direct contribution of hyperandrogenemia to insulin resistance (5–7). On the other hand, the opposite relationship has also been demonstrated, as in the case of ovarian androgen production stimulated by insulin resistance (43). Our experimental model was designed to test the direct effects of hyperandrogenemia on WAT function, but the model does not address the effects of insulin resistance per se on ovarian androgen production. In addition, there are arguments for and against the direct role of adiposity in the pathogenesis of PCOS. For example, elevated androgen levels are often associated with adiposity (44), whereas on the other hand, there is a subphenotype of PCOS that includes women that are lean but insulin-resistant compared with control populations, albeit less insulin-resistant compared with obese PCOS counterparts (9). Our collective knowledge about the role of WAT in PCOS, reviewed elsewhere (8), has shown that WAT plays just part of the role in the pathophysiology of this complex disease syndrome.

In summary, the present report provides mechanistic understanding of sex-hormone–mediated WAT disorders associated with systemic alterations of metabolism such as insulin resistance and obesity (45). Three potential defects in WAT function, including reduced lipolysis, enhanced FFA uptake, and upregulation of FASN and GLUT4 genes involved in de novo lipogenesis, were identified by the present study. Because these factors are positive regulators of lipid storage in adipocytes, they may independently, or together, contribute to adipocyte hypertrophy, insulin resistance, and greater adiposity seen in PCOS patients. We suggest that hyperandrogenemia alters the normal ovary-dependent equilibrium between FFA storage in and FFA mobilization from WAT, leading to development of obesity and insulin resistance in PCOS women consuming a WSD (Figure 6).

Figure 6.

Modulation of lipid metabolism in WAT by the ovarian cycle and T. The model explains the effect of the normal ovarian cycle (A) and the ovarian cycle of hyperandrogenemic females (B) on FFA uptake (arrows in) and lipolysis (arrows out) in WAT. In normoandrogenic females, adipocyte lipolysis fluctuates throughout the cycle, reaching its maximum during the luteal phase and its minimum at menses. T treatment blocks lipolysis during the luteal phase and enhances FFA uptake at menses. The predicted net effect of these alterations is a greater propensity for adipocyte hypertrophy and obesity.