DHEA-Mediated Inhibition of the Pentose Phosphate Pathway Alters Oocyte Lipid Metabolism in Mice

Abstract

Women with polycystic ovary syndrome (PCOS) and hyperandrogenism have altered hormone levels and suffer from ovarian dysfunction leading to subfertility. We have attempted to generate a model of hyperandrogenism by feeding mice chow supplemented with dehydroepiandrosterone (DHEA), an androgen precursor that is often elevated in women with PCOS. Treated mice had polycystic ovaries, low ovulation rates, disrupted estrous cycles, and altered hormone levels. Because DHEA is an inhibitor of glucose-6-phosphate dehydrogenase, the rate-limiting enzyme in the pentose phosphate pathway, we tested the hypothesis that oocytes from DHEA-exposed mice would have metabolic disruptions. Citrate levels, glucose-6-phosphate dehydrogenase activity, and lipid content in denuded oocytes from these mice were significantly lower than controls, suggesting abnormal tricarboxylic acid and pentose phosphate pathway metabolism. The lipid and citrate effects were reversible by supplementation with nicotinic acid, a precursor for reduced nicotinamide adenine dinucleotide phosphate. These findings suggest that elevations in systemic DHEA can have a negative impact on oocyte metabolism and may contribute to poor pregnancy outcomes in women with hyperandrogenism and PCOS.

Up to 7% of reproductive-age women are diagnosed with polycystic ovary syndrome (PCOS) (1, 2), a heterogeneous disease that is characterized by hyperandrogenism, ovulatory dysfunction, and ovarian changes. Other features include insulin resistance and long-term risks of diabetes mellitus (3); up to 40% of women with PCOS exhibit glucose intolerance, and approximately 10% have type 2 diabetes mellitus (4). Furthermore, obesity is a common problem for reproductive-aged women. In a recent study, 55.8% of women 20–39 years of age were found to have a body mass index ≥ 25 (5). Obese women, particularly those with insulin resistance, frequently have decreased sex hormone-binding globulin leading to increased available androgens, as well as increased ovarian production of androgens (6). Given the ovulatory dysfunction experienced by PCOS and obese patients, assisted reproductive techniques are frequently successful in restoring ovulation and often lead to pregnancy. Stimulation of ovulation with clomiphene citrate has been reported to result in ovulation rates of 59%–75.1% and pregnancy rates of 15.4%-32% in PCOS women (7–9). Pregnancy rates following in vitro fertilization (IVF) for women ≤ 35 years of age with ovulatory dysfunction are 52% (10).

In addition to the ovulatory dysfunction that is characteristic of PCOS patients, there is some, albeit conflicting, data to suggest that the affected women have additional challenges to achieving a successful pregnancy. There are some reports that women with PCOS have an increased risk of miscarriage, as high as 35%-65% (11–13). Palomba et al (14) found the miscarriage rate to be 24.7% in PCOS women whereas only 8.7% of controls had miscarriages. More modest miscarriage rates of 14% in women with PCOS and 3% in controls have also been reported (15). Still other studies have found no difference in miscarriage rates between PCOS patients and controls (16–18). These conflicting findings may be due to the inherent heterogeneity of the PCOS patient population or use of different diagnostic criteria and definitions for clinical pregnancy and miscarriage. A meta-analysis of overweight and obese women that included women with and without PCOS found an increased risk of miscarriage with an odds ratio of 1.67 (19). Furthermore, Luke et al (20) found that clinical pregnancy and live birth rates decrease with increasing obesity with autologous oocytes; however, they found no difference with donor oocytes, supporting the oocyte environment as a cause for reduced success. Nonetheless, it is reasonable to suspect that the abnormal hormonal environment that leads to ovulatory dysfunction may also affect the quality of ovulated oocytes, and therefore, embryo competence, resulting in increased miscarriages. Thus, women with hyperandrogenism who do not meet PCOS criteria may also experience abnormalities in oocyte development.

Because of ethical concerns, it has been difficult to study the effects of hyperandrogenism and PCOS on the quality of human oocytes and embryos. Therefore, most information comes from animal models. There have been many attempts to develop an animal model of PCOS, but due to the complexity and unknown etiology of the disease, there is no perfect model. A nonhuman primate model of PCOS has been developed by prenatal exposure to testosterone either early or late in gestation (21). Adult rhesus monkeys that underwent gonadotropin stimulation followed by IVF had similar rates of follicular development, serum FSH, estradiol, progesterone, and testosterone. Furthermore, the number of oocytes retrieved and fertilization rates were similar. However, follicular fluid estradiol and androstenedione were lower, and these correlated with decreased oocyte competence (21). Oocytes exposed to abnormal hormone concentrations were unable to progress beyond the 8-cell embryo stage (21). Other groups have used sc injection of dehydroepiandrosterone (DHEA) to create polycystic ovaries in rats (22, 23). Rodents exposed to DHEA also provide a model for hyperandrogenism and have some similar phenotypic features of PCOS.

In addition to its hormonal effects, DHEA is a known noncompetitive inhibitor of glucose-6-phosphate dehydrogenase (G6PDH), the rate-limiting step in the pentose phosphate pathway (26, 27). The pentose phosphate pathway leads to production of reduced nicotinamide adenine dinucleotide phosphate (NADPH), which is important for lipid metabolism, maintenance of the cellular redox state, and purine substrates for DNA synthesis (Figure 1). Our laboratory has demonstrated that inhibition of this pathway by DHEA prevents proliferation and decidualization of the uterine endometrium through a decrease in purine synthesis (28). Because decidualization is a required process for embryo implantation, this is one possible explanation for the increased miscarriage rate seen by some groups in PCOS patients.

Figure 1.

Glucose metabolism and generation of NADPH in the cumulus cell and oocyte. The pentose phosphate pathway metabolizes glucose to produce ribose-5-phosphate for DNA synthesis and NADPH for reduction, fatty acid synthesis, and cholesterol and steroid hormone production. When the rate-limiting enzyme of the pentose phosphate pathway, G6PDH, is inhibited by DHEA, a greater amount of glucose is shunted to glycolysis, leading to ATP and citrate production. Citrate is converted in the TCA cycle to malate to produce NADPH to support lipid synthesis in hormone-dependent tissues such as the oocyte. G6PDH, glucose-6-phosphate dehydrogenase.

We postulated that because the pentose phosphate pathway is essential for completion of oocyte maturation (29), DHEA might also result in poor oocyte competence and, therefore, poor embryo quality, which could lead to increased miscarriage rates. The oocyte obtains most of the energy it requires to complete the maturation process from the cumulus-oocyte complex (COC) via glycolysis and the tricarboxcylic acid (TCA) cycle (Figure 1). The cumulus cells metabolize glucose via glycolyis to produce pyruvate and lactate, the preferred substrates of the oocyte (30, 31). However, multiple studies have shown that denuded oocytes can also take up glucose (32–34) and have the ability to use it through the pentose phosphate pathway (29, 32). Previous work by our laboratory has shown that citrate (generated in the TCA cycle) and ATP levels in oocytes from diabetic mice correlate with oocyte quality; specifically, lower levels indicate decreased competence (33, 35). In addition to glycolysis, the TCA cycle, and the small amount of glucose that is metabolized via the pentose phosphate pathway (26, 27), the oocyte also obtains energy from β-oxidation of lipids. Utilization of fatty acids as an energy source is important during oocyte maturation and can lead to improved embryo development. Conversely, inhibition of this pathway results in decreased oocyte competence (36). Here we have evaluated the effect of DHEA on oocyte metabolism and demonstrate that elevated levels of DHEA lead to abnormal oocyte metabolism of glucose and lipids.

Materials and Methods

Animals

The Animal Studies Committee at Washington University School of Medicine reviewed and approved all animal experiments. Animals were maintained according to the Guide for the Care and Use of Laboratory Animals provided by the Institute for Laboratory Animal Research. Female FVB/NJ mice were obtained from The Jackson Laboratory. Regular rodent chow was supplemented with 0.01%, 0.1%, or 0.6% (wt/wt) DHEA with or without 0.01% (wt/wt) nicotinic acid (Sigma). Three-week-old mice had access to food and water ad libitum. Mice were fed control or supplemented chow for 2 weeks. All experiments were performed at 5–6 weeks of age, with the exception of the ovarian histology and the estrous cycle evaluation. Separate sets of mice were used for the histology studies, estrous cycle monitoring, metabolic assays, body fat and lipid quantification, mitochondrial DNA (mtDNA) copy number, and IVF studies.

Histology

Ovaries were removed at 10 weeks of age, fixed in Bouin's solution, transferred to 70% ethanol, embedded in paraffin, and serial sectioned at a thickness of 10 μm. Every tenth section was kept and stained with hematoxylin and eosin. Follicles with visible nuclei were classified as primordial, primary, secondary, or antral as previously described (37, 38). Corpora lutea were also counted. Follicle counts from each classification were summed for each mouse.

Hormone concentration

Control and DHEA-fed mice underwent ovarian stimulation by ip injection of 10 IU pregnant mare serum gonadotropin (PMSG). DHEA, estradiol, and testosterone serum concentrations were measured 48 hours later. For progesterone concentrations, mice were given an ip injection of 10 IU human chorionic gonadotropin (hCG) 48 hours after PMSG injection, and blood was collected 24 hours later. Mice were anesthetized by ip administration of a 0.5 mg ketamine and 0.5 mg xylazine mixture, and blood was collected by cardiac puncture. Serum was separated by centrifugation and stored at −20°C. ELISAs were used to determine serum DHEA, estradiol, progesterone, and total testosterone concentrations (catalog no. 20-DHEHU-E01, 11-ESTHU-E01, 11-PROHU-E01, 11-TESHU-E01; Alpco). The detection limit of the DHEA assay is 0.108 ng/mL, intraassay as well as interassay coefficients of variation (CVs) are both <10%. The cross-reactivity of the DHEA assay is as follows: DHEA, 100%; DHEA-S, 0.0037%; testosterone, 0.002%; and < 0.08% for 15 other steroids tested. The estradiol assay has a detection limit of 10 pg/mL and intraassay as well as interassay CVs of ≤ 10.1%. The cross-reactivity of estradiol is as follows: estradiol, 100%; estriol, 1.6%; estrone, 1.3%; and progesterone and cortisol, 0.1%. The progesterone assay has a detection limit of 0.1 ng/mL and intraassay as well as interassay CVs of ≤ 12.6%. The cross-reactivity of progesterone is as follows: progesterone, 100%; 11α-OH-progesterone, 100%; deoxycorticosterone, 1.7%; and <0.5% for 13 other steroids. The total testosterone assay has a detection limit of 0.022 ng/mL, and intraassay and interassay CVs ≤ 10%. The cross-reactivity for the total testosterone assay is as follows: testosterone, 100%; 5α dihydrotestosterone, 5.2%; androstendione, 1.4%; androstandiol, 0.8%; progesterone, 0.5%; androsterone, 0.1%; and < 0.1% for DHEA and 9 other steroids.

Determination of estrous cycles and pregnancy rates

Female inbred mice reach sexual maturity at 6 to 8 weeks of age (39, 40). Therefore, vaginal smears were performed daily starting at 8 weeks of age, after puberty and establishment of regular estrous cycles. Vaginal cells were collected in 20 μL of PBS and evaluated for stage of estrus by light microscopy. Control and DHEA-fed mice were mated with males of known fertility until 3 litters were delivered or up to 12 weeks.

Oocyte collection

After stimulation with PMSG, ovaries were removed and placed in M2 media (Sigma). COCs were obtained by rupture of antral follicles with a syringe and needle. To collect denuded germinal vesicle (GV) oocytes, the cumulus cells were removed by gentle pipetting. To collect metaphase II oocytes, mice were primed with PMSG and hCG. Twenty-four hours later, the oviducts were removed and the ampulla of the oviducts was punctured to release the clutch of ovulated oocytes. At the time of collection, oocytes from animals receiving the same treatment were pooled for further analysis. In a subset of the animals, blood was collected prior to oocyte collection, as described above.

Metabolite and enzyme assays

Denuded oocytes were frozen on a glass slide with liquid nitrogen and dried overnight at −35°C under vacuum within 1 hour following removal of cumulus cells. Individual oocytes were extracted in nanoliter volume under oil. Assays were designed to measure ATP, citrate, G6PDH, or hydroxyacyl-coenzyme A dehydrogenase, type II (HADH2) by linking with nicotinamide adenine dinucleotide/reduced nicotinamide adenine dinucleotide or NADP/NADPH, which were enzymatically amplified in a cycling reaction to allow products to be measured by fluorometry, as previously described (41). Sample values are determined based on the average of internal standards that have <3% variation.

Body fat quantification

Body composition (liquids, fat, and lean tissue) was determined by magnetic resonance imaging (EchoMRI 3-in-1, Echo Medical Systems) acquisition of 3-dimensional body images. Magnetic resonance images were performed following the 2-week diet on the morning of oocyte collection for 4,4-difluoro-4-bora-3a,4a-diaza-s-indacene (BODIPY) staining.

BODIPY immunofluorescence

Denuded GV oocytes were washed in PBS, fixed on a glass slide with 3% paraformaldehyde for 1.5 hours, washed with PBS, stained with 1 μg/mL BODIPY 493/503 (Invitrogen) lipophilic dye for neutral lipid for 1 hour, washed with PBS, and mounted with Vectashield (Vector Laboratories). Images were taken by confocal microscopy at ×20 magnification, and fluorescence intensity was measured in an image of an entire oocyte with Image J software (NIH).

mtDNA copy number quantification

The procedure to isolate and quantify mtDNA was performed as previously described (35, 42).

In vitro fertilization (IVF)

Oocytes were recovered from superovulated female mice 13 hours after 10 U of hCG injection. Sperm were collected from cauda epididymides of 10-week-old male mice and capacitated in vitro at 37°C for 1 hour. Capacitated sperm at a concentration of approximately 10⁵ sperm/mL were coincubated with metaphase II oocytes for 6 hours, and unbound sperm were subsequently washed away. After 24 hours incubation the embryos were observed under light microscopy. The development of 2-cell stage was considered successful fertilization. Two-cell embryos were counted the next morning and unfertilized oocytes were removed. The 2-cell embryos were cultured for 3 additional days to the blastocyst stage.

Statistics

Data are represented as mean ± SEM. Hormone concentrations, estrous cycle length, pregnancy rates, and body fat were analyzed with ANOVA followed by Dunnett's or Bonferroni's post hoc analysis (GraphPad Prism 5 for comparisons between control and experimental values). χ² test was used to analyze the IVF data.

For the metabolic and enzyme assays and BODIPY measurements, 3 to 4 mice were used in each experiment and multiple oocytes from the same mouse were measured. Mouse identification numbers for each oocyte were not tracked; therefore, in order to control for the correlation among the oocytes from the same mouse, mouse identification numbers were generated and simulation analyses were performed. Mixed effects models were used for the comparisons between the control and experiment groups. This modeling approach will account for correlation within groups by including the random effects in the model. The overall error distribution of the mixed model was evaluated. Stata was used for this statistical analysis.

P values of < 0.05 were considered significant. At least 3 independent experiments were performed for all data, with the exception of the IVF data.

Results

DHEA dietary supplementation as a model for anovulation

Many other investigators have used sc injection of DHEA to produce a rodent model of PCOS (22, 43–45). We sought to establish a simple, diet-based model in which mice are fed DHEA-supplemented chow. Analysis of ovarian morphology revealed that mice fed both the control diet and the DHEA-supplemented diet had follicles at various stages of development. The ovaries from mice on the 0.6% DHEA diet had large cystic structures (Figure 2), similar to what has been described previously in rats and mice treated with sc DHEA (22, 43, 44, 46, 47). The number of secondary follicles was slightly, but not significantly, higher in the DHEA-treated animals than the controls. There were also slightly, but again not significantly, fewer corpora lutea in the DHEA-supplemented animals than the controls (data not shown).

Figure 2.

Representative hematoxylin-eosin-stained ovarian sections.

Previous work has demonstrated that sc injection of DHEA prevented ovulation and resulted in persistent estrus or diestrus in rodents. DHEA can be converted to other androgens and estrogens that disrupt the normal feedback of the hypothalamus-pituitary-ovary axis, and serum DHEA levels, as well as other androgens and estrogens, are often increased in women with PCOS (24, 25). We first assessed whether or not inhibition of ovulation occurred in our DHEA-fed mice. Mice fed regular chow ovulated 20.2 ± 1.5 oocytes whereas mice fed 0.01%, 0.1%, and 0.6% DHEA-supplemented chow ovulated 10.8 ± 2.8, 15 ± 2, and 7.1 ± 1.4 oocytes following superovulation, respectively. We also measured serum hormone levels after stimulating ovulation and found that DHEA in the diet led to dose-dependent increases in serum DHEA, estradiol, and testosterone concentrations (Figure 3, A–C). The levels of estradiol in the unstimulated mice were below the level of detection. DHEA-supplemented mice had lower levels of serum progesterone after ovulation than the controls (Figure 3D).

Figure 3.

Hormone concentrations in control and DHEA-fed mice. DHEA (panel A), estradiol (panel B), and testosterone (panel C) were measured 48 hours after stimulation with PMSG. D, Progesterone was measured 24 hours after injection with hCG. Values are mean ± SEM. *, P < .05; **, P < .01 compared with control; n = 4–7 mice.

Before examining any effects of the DHEA-supplemented diet on the estrous cycle, we first sought to confirm that the cycles were normal before DHEA supplementation. We evaluated cycles for 8 days in pubertal females in each treatment group while on regular rodent chow and found that all groups had regular estrous cycles of 5.625 ± 0.31 days. After 2 weeks on the DHEA diets, the mice consuming 0.01% DHEA had cycles that were significantly longer than mice on control diet. Mice fed 0.1% DHEA or 0.6% DHEA remained in metestrus or diestrus for the entire evaluation period (Figure 4A). Mice fed 0.01% DHEA-supplemented chow were able to produce pups while on the diet, but the litters had fewer pups than the controls. The mice on 0.1% or 0.6% DHEA did not deliver any pups (Figure 4B and data not shown). Taken together, these data indicate that chow supplementation with DHEA is an effective means of inducing an anovulation phenotype in mice.

Figure 4.

Estrous cycles and pregnancy rates of control and DHEA-fed mice. A, Average length of estrous cycle determined by vaginal smears taken daily for 8 days before and 15 days after diet. All mice in the 0.1% DHEA and 0.6% DHEA groups after diet remained in the same stage of the estrous cycle. B, Number of pups per litter while on DHEA diet. Values are mean ± SEM. *, P < .05; **, P < .001 compared with control; n = 4 mice.

Glucose metabolism in DHEA-exposed oocytes

To assess possible metabolic consequences of DHEA supplementation, we first examined ATP and citrate levels in oocytes from mice fed 0.01% DHEA and 0.1% DHEA. We focused the metabolic studies on the 2 lower DHEA concentrations given the large hormonal changes seen with 0.6% DHEA. ATP levels in denuded GV oocytes did not differ significantly between mice on control diet and those on 0.01% DHEA-supplemented chow (Figure 5A). However, citrate levels were significantly lower in both DHEA diet groups (Figure 5B). Because DHEA is an inhibitor of G6PDH, we compared activity of this enzyme between oocytes from mice fed control and those on DHEA-supplemented diets. The 0.1% DHEA diet led to significant inhibition of G6PDH activity (assessed by measurement of NADPH production) to 87% of normal (Figure 5C). These data demonstrate that both the TCA cycle and the pentose phosphate pathway were affected in the oocytes of mice ingesting DHEA.

Figure 5.

Metabolite level and enzyme activity in single denuded oocytes from control and DHEA-fed mice. A, ATP level; *, P < .05 between control and 0.1% DHEA. B, Citrate level; *, P < .05 between control and 0.01% DHEA; **, P < .001 between control and 0.1% DHEA. C, G6PDH enzyme activity; *, P < .05 between control and 0.1% DHEA; n = 3–4 mice. DO, denuded oocytes.

Lipid metabolism in DHEA-exposed oocytes

Inhibition of the pentose phosphate pathway leads to decreased ribose-5-phosphate and NADPH. Given that NADPH is used during multiple steps of the lipid synthesis process, we measured the effect of DHEA on lipid synthesis. Mice fed the 0.1% DHEA diet had significantly lower levels of whole-body fat than those fed control chow (Figure 6). Additionally, BODIPY staining of lipid droplets revealed that denuded GV oocytes from the DHEA-fed mice had significantly less lipid than oocytes from control-fed mice (Figure 7A). We sought to reverse the DHEA-mediated inhibition of NADPH production by addition of a precursor, nicotinic acid. Mice fed a diet supplemented with 0.1% DHEA and 0.01% nicotinic acid had higher body weight (22.25 ± 0.57 g) than those fed 0.1% DHEA alone (15.92 ± 0.98 g) and were similar in weight to the controls (20.37 ± 0.58 g). Nicotinic acid supplementation also led to normalization of lipid levels (Figure 7A) and citrate levels (Figure 8) in DHEA-exposed oocytes.

Figure 6.

Percent body fat in control and DHEA-fed mice. *, P < .01 between control and 0.1% DHEA; n = 9–10 mice.

Figure 7.

Lipid quantification in control, DHEA-, and DHEA/NA-fed mice. A, Single denuded GV oocytes BODIPY quantification by Image J. *, P < .05; n = 3–4 mice. B, Representative confocal microscopy images of single denuded GV oocytes. NA, nicotinic acid.

Figure 8.

Supplementation with nicotinic acid rescues DHEA-mediated decreases in citrate levels in single denuded GV oocytes. *, P < .05 between control and 0.1% DHEA; n = 3–4 mice. NA, nicotinic acid,

Because β-oxidation and the TCA cycle are important processes in oocyte maturation and they both occur in the mitochondria, we measured the mtDNA copy number in the GV oocytes as a marker for mitochondrial function. There was no significant difference in the mtDNA copy number between the oocytes from mice fed control or DHEA-supplemented chow (data not shown). We also measured the enzyme activity of HADH2, a key enzyme in the β-oxidation pathway. There was not a significant difference in the activity of HADH2 between control and DHEA-exposed oocytes (data not shown). Therefore, the DHEA-exposed oocytes should have had normally functioning mitochondria capable of both TCA cycle metabolism and β-oxidation, suggesting the defect is in the pentose phosphate pathway.

Oocyte competency from DHEA-treated mice

Two separate IVF experiments were performed with a total of 13 mice fed regular chow and 24 mice fed chow supplemented with 0.01% DHEA. The fertilization rate (number of 2-cell embryos per number of oocytes) was 87.4% and 94.1% in the control and 0.01% DHEA group, respectively. The relative risk for fertilization in the DHEA-treated group was 1.078 with CI 1.011–1.149. The progression to blastocyst rate (number of blasts/number of 2-cells) was 49.3% and 20.3% in the control and 0.01% DHEA group, respectively. The DHEA-exposed group had a relative risk of 0.4121 (0.3061–0.5548) for progression to blastocyst. Although the oocytes from DHEA-exposed mice fertilized normally, the ability to develop into blastocysts is significantly decreased, suggesting decreased competency.

Discussion

Here we have demonstrated that supplementing regular rodent chow with DHEA generates a mouse model that mimics several aspects of androgen-induced anovulation in humans. Exposing mice to DHEA in their diets for 2 weeks (correlating with the 10–16 day period of folliculogenesis in mice [Refs. 48 and 49]) negatively affected ovulation, likely through dysregulation of the hypothalamic-pituitary-ovary axis. Using diet supplementation, we were able to generate mice with serum levels of DHEA that are elevated to a similar degree as is observed in some women with PCOS. Whereas supplementation with 0.01% DHEA resulted in a 1.9-fold increase in DHEA serum levels in the mice, women with PCOS were reported to have 2.6-fold higher serum concentrations of DHEA than controls (50). Similar to the androgen-induced dysregulation of ovulation observed in women, the DHEA-exposed mice had prolonged estrous cycles with oligoovulation or anovulation. Analogous to the clinical finding that stimulation of ovulation can be successful in this patient population, we observed that the decreased ovulation, suggested by abnormal estrous cycles, in DHEA-fed mice was overcome with the use of gonadotropin stimulation.

Previous groups have shown similar ovulatory dysfunction in rodents exposed to sc DHEA injections of 6 mg/100g body weight for 20 days. Treatment with DHEA prevented ovulation and resulted in persistent estrous or diestrous states that were responsive to exogenous gonadotropins and returned to normal cycles after withdrawal of DHEA. Serum LH was much lower in rats receiving DHEA injections, whereas either no change or an increase in serum FSH was seen (43, 46). Familiari et al (51) found that DHEA-treated mice had higher androgen levels, but much lower 17β-estradiol levels. These studies were done in unstimulated mice with atretic follicles, in contrast to our finding of increased estradiol after ovarian hyperstimulation and potential rescue of follicles from atresia. The administration of DHEA has been shown to lead to atresia of follicles, as well as abnormal mitochondria, decreased actin, and degeneration in cultured granulosa cells (52). In DHEA-treated rodents with formation of large ovarian cysts, it is thought that the DHEA plays a critical role in atresia of these follicles and the oocyte degenerates first followed by the granulosa cells, whereas in physiologic atresia of follicles, the granulosa cells undergo apoptosis first (53–55). Although different parameters were used to identify ovulatory dysfunction (LH, FSH, stimulated and unstimulated hormone concentrations, and follicular atresia), they all suggest a similar phenotype. Our study is the first to report hormonal and ovarian changes with oral administration similar to those seen with sc injection.

In addition to its hormonal effects, our data suggest that DHEA can alter oocyte metabolism through inhibition of the pentose phosphate pathway and alteration of lipid metabolism. The importance of the pentose phosphate pathway in oogenesis has been illustrated in multiple publications. For example, glucose metabolized via the pentose phosphate pathway is important for nuclear maturation in mouse oocytes (56), as well as the progression of meiosis and cleavage of blastocysts (57, 58). In the COC, glucose metabolism to produce pyruvate via glycolysis and the pentose phosphate pathway has been shown to be necessary to prevent oocyte aging after ovulation (59). Our findings may help explain the observations of Shinohara et al (45) that DHEA treatment of rats resulted in a dose-dependent increase in the percentage of degenerated oocytes.

Abnormalities in the pentose phosphate pathway can reduce oocyte maturation, and perhaps affect the TCA cycle, by a lack of purine substrates and decreased production of NADPH. Nutt et al (60) have shown that NADPH production from the pentose phosphate pathway is important to prevent oocyte apoptosis through caspase-2. Although most NADPH is produced by the pentose phosphate pathway (61), a smaller amount is also produced from malate in the TCA cycle (Figure 1). To compensate for inhibition of the pentose phosphate pathway, glucose may be shunted through glycolysis to produce citrate, which is then converted to malate and NADPH. This would explain our findings of minimal differences in ATP levels, but significantly decreased citrate levels (Figure 5). We have previously shown that this degree of change in citrate levels leads to relevant biological changes (35, 41, 62). Only a small portion of glucose is utilized through the pentose phosphate pathway in cumulus cells (30), whereas the pentose shunt appears to be more active in the oocyte of multiple species (29, 58, 63), further highlighting the importance of inhibition of this pathway. Our data support that the pentose phosphate pathway in the oocyte is critical for normal metabolism. However, the cumulus cells surrounding the oocyte may also have a role, although likely of lesser importance, given the minor amount of glucose these cells metabolize via the pentose phosphate pathway. In addition, oocytes exposed to 0.01% DHEA showed decreased progression to the blastocyst stage, suggesting that DHEA negatively affects oocyte competency.

We propose that despite increased NADPH production through citrate metabolism, the DHEA-exposed oocyte is unable to fully compensate for the decreased pentose phosphate pathway contribution. Therefore, pathways that are dependent on NADPH will be negatively affected. One such pathway is lipid synthesis; multiple enzyme steps of lipid synthesis require NADPH. Consistent with this idea, we found that DHEA treatment resulted in decreases in both whole-body fat and oocyte lipid content. This decrease was not due to excess fatty acid metabolism (because activity of the β-oxidation pathway enzyme HADH2 was not affected), indicating that lipid synthesis was impaired. Further supporting an idea that the reduction in lipid synthesis was due to a lack of NADPH is our finding that supplementation with nicotinic acid, a precursor of NADPH, was able to reverse the citrate and lipid effects in the oocytes.

The relationship between androgens and adipose tissue is complex. Several studies in rodents, humans, and adipocyte cell lines suggest that testosterone and dihydrotestostone have antiadipogenic effects such as inhibition of preadipocyte differentiation, decreased fatty acid synthesis, and increased lipolysis (6). Furthermore, androgen receptor-null male mice developed late-onset obesity, suggesting that androgens lead to a decrease in fat (6). This is also similar to our model in which the DHEA-exposed mice had decreased body fat. However, the adipocytes from women with obesity behave differently. Insulin resistance and hyperinsulinemia, frequently associated with obesity, alter ovarian steroidogenesis and sex hormone-binding globulin production. These changes result in excess androgens that further alter insulin and body fat (6). Our model provides a mechanism to explore the effects of androgens on the oocyte, independent of obesity.

In conclusion, our work suggests that DHEA negatively affects oocyte quality. The exact mechanisms by which maternal DHEA exposure alters oocyte competence are not completely understood, but abnormal lipid metabolism appears to play an important role. Future studies will focus on understanding the impact of DHEA on lipid metabolism and lipid carrier proteins, as well as effects on fertilization potential and embryo development of oocytes exposed to DHEA and nicotinic acid. Our findings may be clinically significant given that DHEA is a commonly used dietary supplement and is often recommended to older women undergoing IVF to achieve pregnancy. Reports thus far indicate a potential improvement in pregnancy rates in these women taking DHEA. However, it is unclear how DHEA may lead to improved oocyte and embryo quality in an older population. It is possible that DHEA may have a bimodal effect that leads to poorer quality oocytes in young healthy women, but improves outcomes in women with diminished ovarian reserve. A better understanding of the role DHEA plays in oocyte and blastocyst metabolism may lead to therapeutic targets in women with PCOS or hyperandrogenism, as well as greater insight into the effect of DHEA supplementation in women with diminished ovarian reserve.