Reproductive Tissues Maintain Insulin Sensitivity in Diet-Induced Obesity

Mice were maintained on a mixed background: CD1/129SvJ/C57BL6. Pituitary-specific insulin receptor KO mice (PITIRKO) were produced by our group as described previously (2). αGSU-Cre-negative littermates were used as controls (wild type [WT]). Female PITIRKO mice and controls were fed a high-fat diet (HFD) or regular chow as described previously (2). Mice used in this study were female and were maintained with food and water ad libitum under a 14-h/10-h dark/light cycle. Procedures were approved by the Johns Hopkins Animal Care and Use Committee. Experiments were conducted with 5.5- to 6.5-month-old female mice (lean and DIO): lean WT (30.6 ± 1.0 g), lean PITIRKO (29.6 ± 1.8 g), WT DIO (44.3 ± 6.9 g), and PITIRKO-DIO (43.4 ± 6.4 g). Mice were tested for estrous cyclicity, and WT DIO mice were acyclic as reported previously (2).

Mice were fasted overnight, and blood was collected from mice via mandibular bleed at 10 a.m. All samples were collected by 10:30 a.m. Insulin, leptin, and glucagon were measured with the Milliplex Map Mouse Serum Adipokine Panel (Millipore, Billerica, MA), and insulin-like growth factor-I (IGF-I) was measured with a single-plex IGF-I panel (Millipore) on a Luminex 200IS platform (Luminex Corporation, Austin, TX). All samples were performed in one plate. The intra-assay coefficient of variation for each assay was between 5 and 9%.

Mice were fasted overnight and injected intraperitoneally with 2 g/kg body wt dextrose. Glucose was measured from tail blood at the times indicated using a One Touch Ultra glucometer. After insulin injection (Lilly, Indianapolis, IN) in lean and WT DIO mice, glucose was measured as described above.

Mice were fasted overnight, and insulin or 0.9% saline was injected intraperitoneally with different doses. Tissues were collected at 10, 15, 30, or 45 min after insulin injection and snap-frozen in liquid nitrogen. Protein was obtained and measured as described previously (2). To quantify the levels of phosphorylated AKT (pAKT), phosphorylated extracellular signal–related kinase (pERK), and AKT for different tissues of each individual animal after insulin or saline injection, Bio-Plex Phosphoprotein Detection Multiplex assays were used (Bio-Rad Laboratories, Hercules, CA. For quantification of phosphorylated Tyr (pTyr)-IRS1 (or total IRS1) levels in cell lysates, Milliplex Map Phospho IRS1 Mapmates (or Total IRS1 Mapmates) kits were used. Protein (10 μg) from each tissue was loaded into a 96-well microplate. Tissues from four to eight mice were measured independently and were not pooled. Assays were conducted using the Luminex 200 system. Values were assessed as mean fluorescent intensity. For quantification of the pTyr-IRS2 level, a PathScan Phospho-IRS-2 (panTyr) ELISA kit (Cell Signaling Technology) was applied according to the manufacturer’s instructions.

Identical amounts of protein were loaded and separated by SDS-PAGE gel and blotted onto nitrocellulose membranes. Signals were detected by enhanced chemiluminescence plus Western Blotting Detection System (GE Healthcare, Buckinghamshire, U.K.). Rabbit primary antibodies against pAKT (Ser473), AKT, pERK, ERK, phosphorylated Jun NH₂-terminal kinase (pJNK), Jun NH₂-terminal kinase (JNK), and actin were from Cell Signaling Technology. Rabbit anti-IRS1 was from Millipore. Goat anti-rabbit IgG (H+L)-HRP conjugate was from Bio-Rad Laboratories.

Pituitary and ovary RNA was extracted and reverse transcribed as previously described (4). For the expression of irs1, iQ SYBR Green quantitative PCR (Bio-Rad Laboratories) was performed, and irs1 primers were from the study by Murata et al. (5). gapdh was used as described previously (4). irs1 values were corrected with gapdh values. Reactions were performed using an iCycler iQ5 quantitative PCR machine (Bio-Rad Laboratories) as described previously (4).

Tissues were paraffin sectioned by Johns Hopkins University Pathology Core Facility. Sections were stained with rat anti-mouse F4/80 (1:100) and Mac-2 (1:200), and markers for macrophages (6) were purchased from Cedarlane (Burlington, Ontario, Canada). Secondary antibody Alexa Fluor 594 goat anti-rat IgG (H+L) (1:200) was from Invitrogen.

All data were analyzed with unpaired two-tailed Student t test using Prism Software (GraphPad Software, Inc.) and are expressed as means ± SEM. P < 0.05 was defined as statistically significant.

We measured fasting serum insulin, leptin, glucagon, and IGF-I levels in mice fed an HFD and in chow-fed control littermates. Insulin and leptin levels were elevated in DIO mice relative to controls as reported previously (Fig. 1A and B) (2,7). DIO mice had lower serum glucagon levels than did controls (Fig. 1C); IGF-I levels were unchanged in DIO mice relative to those in lean controls. (Fig. 1D). DIO mice exhibited glucose intolerance after d-dextrose injection in as little as 15 min (Fig. 1E).

We measured the time course of the response of circulating glucose levels to insulin in order to identify a time at which we could measure early insulin-signaling events independent of insulin-induced hypoglycemia. We injected overnight-fasted lean mice with insulin (0.5, 1, 1.5, and 3 units/kg body wt) and measured glucose at 10, 30, and 45 min after injection. Glucose levels declined, as expected, at the 30- and 45-min time points for all doses of insulin. No significant decrease in glucose levels was observed at 10 min for any of the doses of insulin (Fig. 2). Additionally, the highest levels of pAKT in the pituitary and ovary were observed at this time (Supplementary Fig. 1). Therefore, for all subsequent studies (except those specifically addressed) a 10-min time point was used to avoid the effects of hypoglycemia on signaling pathway activity.

For determination of whether reproductive tissues (hypothalamus, pituitary, and ovary) are comparable in insulin sensitivity with classical energy storage tissues such as liver, muscle, and fat, insulin (0.5, 1, or 1.5 units/kg body wt) was injected intraperitoneally into fasted lean WT female mice. At 10 min, tissues were collected, and pAKT, pERK, and total AKT levels were quantified by Luminex assays. Results are expressed as fold change compared with data from saline-injected littermates. Liver and muscle showed significantly increased pAKT levels at 0.5 units/kg body wt insulin (Fig. 3A and C), demonstrating that intraperitoneal insulin administration achieves circulating insulin levels at 10 min adequate for inducing signaling in these tissues. However, the pituitary and ovary first demonstrated AKT phosphorylation at the higher insulin dose of 1.5 units/kg body wt (Fig. 3E and G). No change in pAKT levels was observed in the hypothalamus at any of the doses (Fig. 3I). There was no significant increase in pERK in pituitary, ovary, or hypothalamus in response to insulin (Fig. 3F, H, and J). Muscle and liver exhibited increased pERK levels at 1 and 1.5 units/kg body wt insulin but not 0.5 units/kg body wt insulin (Fig. 3B and D). Because no changes in ERK signaling were observed in the reproductive tissues at these doses, subsequent analyses focused on changes in the phosphatidylinositol (PI) 3-kinase signaling pathway. Meanwhile, total AKT levels were also measured, with no observed difference between insulin- and vehicle-injected lean mice (Fig. 3K, L, and M).

In addition to Luminex assays of insulin signaling, we conducted Western blot analysis to explore insulin-signaling pathways in reproductive tissues and the energy storage tissues (Fig. 4A). We found similar tissue-specific phosphorylation patterns in response to insulin using Western blot analysis. We analyzed insulin signaling in the DIO mice by Western blot, and both the pituitary and ovary responded to insulin with increased phosphorylation of AKT. Furthermore, higher basal levels of pAKT were observed in the pituitary and ovary of DIO female mice compared with those in lean WT mice (Fig. 4B). However, insulin resistance was clearly observed in liver in DIO mice with dramatic reduction in the levels of insulin-stimulated pAKT compared with those seen in lean mice (Fig. 4B).

Although pituitary and ovary were less sensitive to insulin than muscle and liver, we observed that they maintain insulin sensitivity in obese mice in which insulin resistance was demonstrated in the energy storage tissues. To examine the differences between the tissues in pAKT signaling with higher throughput, accuracy, and sensitivity, we used four to eight animals per tissue per treatment and applied each individual protein into 96-well microplates to perform Luminex signaling assays. Since AKT phosphorylation in the reproductive tissues only increased at higher insulin doses in the lean mice, 3 units/kg body wt insulin were also included in our study. Pituitary-specific insulin receptor conditional KO mice were examined as a control to verify that the pAKT signaling is mediated through the insulin receptor. After 12 weeks of HFD, WT DIO mice showed a blunted response to insulin in liver and muscle compared with that in the lean WT mice (Fig. 5). Basal levels of pAKT in the pituitary and ovary were also significantly elevated in DIO mice relative to lean mice, and insulin treatment increased levels of pAKT to equal magnitudes in both lean and DIO mice (Fig. 5A and B). The increased basal level of pAKT and the response to insulin in pituitary were confirmed to be mediated by the insulin receptor by using PITIRKO-DIO mice that had no significant increase of basal or insulin-stimulated pAKT levels in the pituitary relative to lean mice. Since the insulin receptor is intact in the liver, muscle, and ovary, these tissues respond to insulin nearly identically to WT DIO mice (Fig. 5B, D, and E). The hypothalamus did not respond to insulin at even the 3 units/kg body wt dose (Fig. 5C).

IRS1 and IRS2 are scaffolding proteins that transduce signals from the insulin receptor to the PI 3-kinase and mitogen-activated protein kinase (MAPK) signaling pathways. They are also proposed as loci of protein modifications that produce insulin resistance (8). pTyr is a marker of insulin sensitivity (9), and pTyr-IRS1 levels were measured. We observed that pituitary pTyr-IRS1 levels do not change in response to insulin treatment in either lean or DIO mice (Fig. 6A). As for pAKT, the basal levels of pTyr-IRS1 in the ovary are elevated in DIO mice relative to those in lean mice (Fig. 6B). Additionally, insulin stimulated an increase in pTyr-IRS1 in the ovary in both lean and DIO mice in a dose-dependent manner. Liver and muscle exhibited insulin resistance at the level of IRS1 in DIO mice compared with lean mice (Fig. 6C and D). The insulin receptors remained intact in the ovary, liver, and muscle in the PITIRKO mice; therefore, insulin-induced IRS1 signaling in these tissues closely resembles that seen in WT mice.

Since IRS1 does not mediate insulin signaling in the pituitary, we also measured pTyr-IRS2 levels. Both the pituitary and the ovary showed higher basal levels of pTyr-IRS2 in DIO mice compared with lean controls (Fig. 6E and F). Furthermore, pTyr-IRS2 levels were increased in both the ovary and pituitary in response to insulin in both lean and WT DIO mice. In contrast, while pTyr-IRS2 levels were increased in liver and muscle in response to insulin in lean controls, there was no increase observed in DIO mice (Fig. 6G and H). Both the basal increase in pTyr-IRS2 levels in DIO mice and the insulin-stimulated increase in pTyr-IRS2 levels were absent in the pituitaries of PITIRKO mice, confirming that both are mediated by the insulin receptor. As expected, there was no difference in the responses of WT DIO and PITIRKO-DIO mice in liver, muscle, or ovary. Since we did not observe any change in pTyr-IRS1 in pituitary but did in ovary, we examined the mRNA and protein level of IRS1 in pituitary and ovary and found that both were significantly lower in the pituitary (Fig. 6I–K).

Others have shown that HFD increased macrophage infiltration in energy storage tissues (10,11), which can induce pJNK activation of liver and muscle in DIO mice (7,12,13). pJNK activation is a major contributor to insulin resistance (7). Since basal pAKT levels of pituitary and ovary were elevated in DIO versus lean mice, we chose to investigate whether macrophage infiltration and JNK phosphorylation were different between reproductive tissues and the energy storage tissues. Using immunofluorescence and Western blot analysis, we observed that the pituitary and ovary of DIO mice had levels of macrophage infiltration and pJNK similar to those of their lean littermates (Fig. 7A and B). In contrast, adipose tissue exhibited dramatically increased staining of the macrophage-specific protein F4/80 (Fig. 7A) in the DIO mice. Another macrophage-specific protein, Mac2, showed a distribution similar to that of F4/80 (data not shown). Liver and adipose tissue expressed strong staining for pJNK in the DIO mice relative to the lean controls (Fig. 7B). No change in total JNK was observed in examined tissues with HFD conditions (Fig. 7B).

That states of extreme positive energy balance and peripheral insulin resistance result in reproductive dysfunction is becoming increasingly clear. To investigate the response of reproductive tissues to hyperinsulinemia, we used a mouse model of infertility associated with DIO.

Initial analysis demonstrated that reproductive tissues required higher doses of insulin to activate downstream signaling pathways than do the classic energy storage tissues, suggesting that reproductive tissues were significantly less sensitive to insulin than classical energy storage tissues (Fig. 3). These initial studies were performed in lean female mice and suggest that fundamental tissue-specific differences in insulin sensitivity exist. Under normal physiologic conditions, insulin signaling in the gonadotropin-releasing hormone neuron and pituitary does not play an important role in reproductive function, since deletion of the insulin receptor in these neurons have no impact on fertility (2,14). A role for pathologic insulin signaling in states of insulin excess could occur, however, since insulin signaling was activated in the reproductive tissues at higher doses (Fig. 3). Consistent with this assertion, mice on 60% HFD for 12 weeks develop hyperinsulinemia associated with infertility and elevated luteinizing hormone and testosterone levels (2).

Curiously, the hypothalamus exhibited no change in AKT or IRS1 phosphorylation after insulin at any time point examined (Fig. 3I and Supplementary Figs. 1 and 4). The hypothalamus is a putative target of insulin in feeding behavior and discrete nuclei, or whole hypothalami, as previously shown in male rats and mice, are activated by insulin (15–17). However, these studies generally used intracerebroventricular infusions of insulin or used much higher peripheral doses of insulin than were used in our studies performed in female mice. Alternatively, the effects of insulin in the hypothalamus may have been masked by an abundance of nonresponse cells in the fragment. In DIO mice, there was a blunted response to insulin signaling at the level of AKT in both liver and muscle demonstrating classical insulin resistance (Fig. 5D and E). In contrast, the pituitary and ovary exhibited elevated basal levels of pAKT and were not resistant to insulin activation of AKT in the DIO mice (Fig. 5A and B). In the pituitary, the significance of AKT activation is not known. While insulin administration causes a robust stimulation of AKT in a gonadotrope cell line (18), whether AKT signaling plays a role in luteinizing hormone secretion remains to be determined. In the ovary, AKT activation has been found to play a role in insulin induction of 5α-reductase gene expression (19) and 17α-hydroxylase activity (20), leading to an increase in ovarian androgen biosynthesis. Interestingly, polymorphisms in the Akt2 gene have been found to be associated with PCOS in women (21). Insulin and IGF-I–induced progesterone production via MAPK pathways in the human ovary have also been described (22), with insulin stimulating the p38 family of MAPKs to induce this effect. We did not see an effect of insulin on ERK activation in the ovary but did not investigate for activation of the p38 family of MAPKs in the mice; thus, lack of activation of the entire MAPK pathway cannot be excluded by our studies.

To explore why the pituitary and the ovary maintain insulin sensitivity in the obese state, we measured activation of the IRS proteins (IRS1 and IRS2). Pituitary and ovary exhibited an elevation in basal pTyr-IRS2 levels in DIO mice compared with lean mice (Fig. 6E and F) that was similar to the tissue-specific pattern of AKT phosphorylation. In contrast with the liver and muscle, insulin was able to induce a further increase in pTyr-IRS2 levels in the pituitary and ovary from DIO mice (Fig. 6). Phosphorylation events were mediated by the insulin receptor—not the IGF-I receptor—since increases in pituitary pAKT or pTyr-IRS2 did not occur in PITIRKO mice (Figs. 5A and 6E). However, signaling characteristics in the ovary, liver, and muscle were nearly identical to those seen in WT mice. Our study is in agreement with phosphorylation patterns seen using a paradigm of chronic treatment with either insulin or human chorionic gonadotropin/insulin in rats to mimic chronic hyperinsulinemia (23,24). An increase in the IRS–PI 3-kinase/pAKT signaling pathway in whole ovary extracts was observed, while the liver and muscle became insulin resistant as indicated by attenuated insulin-induced pAKT levels. Others have reported a female rat model of hyperinsulinemia that exhibited a distinctly exhibited a distinctly PCOS appearance that was similar to our model of DIO-related infertility with high luteinizing hormone levels (23,24).

Interestingly, pTyr-IRS1 levels in the pituitary were not changed by insulin treatment in lean or DIO mice (Fig. 6A and Supplementary Fig. 2), although this may be a function of the paucity of IRS1 in the pituitary as assessed by Western blot or Luminex assay (Fig. 6I, J, and K). Analysis of the IRS2 KO mice has also identified a similar predominant role of IRS2 in the pituitary without any detection of pTyr-IRS1 using pooled pituitaries (3). At the 10-unit dose of insulin, activation of IRS1 signaling was observed in the pituitary of both WT and PITIRKO mice, suggesting IGF-I receptor activation (Supplementary Fig. 3).

While there were significant differences in the pituitary in the phosphorylation of IRS1 and IRS2 to insulin, in the ovary, the two IRS proteins exhibited similar responses to insulin in both lean and DIO mice (Fig. 6A, B, E, and F). Insulin stimulated an increase in ovarian pTyr-IRS1 and pTyr-IRS2 in lean mice. Compared with those in lean mice, basal levels of ovarian pTyr-IRS1 and pTyr-IRS2 were elevated in DIO mice, with a further increase in levels of ovarian pTyr-IRS1 and pTyr-IRS2 after insulin administration. This preservation of insulin signaling in the setting of peripheral insulin resistance suggests that ovarian IRS is not susceptible to the mechanisms proposed to underlie tissue resistance to insulin, including serine phosphorylation and degradation of IRS (8). These data suggest that differences at this proximal level of the insulin-signaling pathway between the ovary and pituitary and the energy storage tissues underlie the retained sensitivity to insulin in obese mice.

The DIO mice are chronically hyperinsulinemic, which differs from studies that used acute or chronic insulin stimulation as a model (25,26). In the latter study, lean rats were injected with insulin for 22 days. However, these rats retained cyclicity, which may be a result of unchanged gonadotropin-releasing hormone sensitivity of the pituitary observed in this model. It is not clear whether this is a species-specific difference or due to the differences between the models of hyperinsulinemia (DIO versus insulin injection). However, recent work in a DIO rat model (27) showed that 120 days of HFD was associated with elevated circulating luteinizing hormone levels, elevated basal pAKT levels, and a retained responsiveness to acute insulin treatment—all of which were observed in our mouse DIO model (Figs. 4B and 5) (2). In contrast to our model, insulin resistance in the rat ovary developed after 180 days on HFD and luteinizing hormone levels normalized. The authors of the study also measured interleukin-1B, tumor necrosis factor-α, and pJNK expression in the ovary and found no increase in DIO versus lean females at 120 days but a significant increase after 180 days in DIO rats (27). These obesity-induced stress responses have been proposed as mechanisms for the development of insulin resistance in other tissues (8), and suppression of JNK signaling in liver and muscle has been shown to ameliorate insulin resistance in obese mice (28,29). The time course of cytokine expression and insulin resistance in the ovary observed by Akamine et al. (27) supports our contention that a lack of macrophage infiltration combined with a lack of increase in cytokine signaling in the pituitary and ovary may contribute to retained insulin sensitivity in the DIO mouse (Fig. 7). As our model exhibits retained insulin signaling with estrous acyclicity, this difference suggests that excessive insulin signaling may lead to anovulation and associated lack of estrous cycling. The development of ovarian insulin resistance observed by Akamine et al. may be a function of species differences (rat versus mouse). Alternatively, there are clearly time-dependent changes that occur between 120 and 180 days in the rat that could emerge with extended analysis of the DIO mice. We favor the former hypothesis, since we have observed retained insulin signaling in the ovaries and pituitaries of mice on an HFD for over 6 months (data not shown), suggesting that the protection from insulin resistance in these tissues may be more durable in the mouse than in the rat.

The lack of elevated JNK signaling in DIO mice relative to controls (Fig. 7B) differs from the findings of a recent study that proposed a role for pituitary JNK signaling in the regulation of glucose homeostasis (30). These investigators noted an increase in JNK-induced c-Jun phosphorylation in DIO male mice relative to chow-fed controls. These differences are likely due to strain and sex differences between our studies.

We focused on PI 3-kinase AKT, which is thought to more closely mediate metabolic signaling, whereas the MAPK pathway mediates growth and development (31), and because our studies did not show significant changes in ERK phosphorylation between conditions. This does not preclude important roles for ERK signaling in reproductive tissues, and in fact, ERK mediates essential signaling events in both the pituitary (32) and ovary (33) as a result of G-protein receptor activation. In addition, a role for non-ERK MAPK signaling has been implicated in ovarian progesterone synthesis (22). Impaired glucose uptake and diminished accumulation of lactate were observed in granulosa cells of women with anovulatory PCOS compared with control women, suggesting that granulosa cells in women with PCOS exhibit resistance to the metabolic effects of insulin (34). While our studies did not distinguish between signaling in theca and granulosa cells, our results would be consistent with retained signaling in the theca cells. Therefore, the retained insulin responsiveness in DIO females (Figs. 5 and 6) may be manifest at the level of the theca cell even in the face of resistance in the granulosa cells, resulting in increased steroidogenesis (35–37).

Persistent ovarian sensitivity to insulin in the context of peripheral insulin resistance has been implicated in the pathophysiology of PCOS. Human studies indicate that ovarian steroidogenesis involves AKT activation (18,19) and that steroid metabolism is preserved in women with hyperinsulinemia and PCOS (2,38,39). Our results, which demonstrate maintained insulin-induced Akt phosphorylation in the ovary in the setting of hyperinsulinemia, suggest that persistent AKT phosphorylation may be contributing to the pathologic steroidogenesis in women with PCOS. Furthermore, women with type 1 diabetes, who are often treated with supraphysiologic doses of insulin on a chronic basis, have a two- to threefold increased risk of development of PCOS. It is noteworthy that this subset of women with PCOS has hyperandrogenemia, implying abnormal ovarian steroid metabolism, without evidence of elevated antimullerian hormone (produced by small ovarian follicles), luteinizing hormone-to-follicle-stimulating hormone ratio, or sulfated dihydroepiandrosterone as seen in PCOS (40).

Hence, it is likely that a hormonal milieu that differs between PCOS and type 1 diabetes is responsible for disruption of ovarian and adrenal function and that persistent AKT signaling cannot alone account for the ovarian dysfunction seen in PCOS. As a further indication of the link between hyperinsulinemia and ovarian steroidogenesis, hyperinsulinemic states induced by hyperinsulinemic-euglycemic clamps lead to increased ovarian androgen production in lean women with PCOS, demonstrating an acute induction of ovarian steroidogenesis by insulin (41).

In summary, we have shown that the insulin responsiveness of the pituitary and ovary is uniquely preserved in mice with peripheral insulin resistance (Fig. 8). Our study is the first to suggest that the preservation of insulin signaling in the pituitary and ovary of the DIO mouse occurs at a proximal point in insulin signaling, i.e., at the level of the IRS proteins. Further delineation of these pathways will provide insight into the role of hyperinsulinemia in the development of PCOS.

The authors acknowledge funding support from the Endocrine Fellow Foundation (to S.W.), the National Institutes of Health (NIH) (R01 HD44608 to A.W.), and the Eunice Kennedy Shriver National Institute of Child Health and Human Development/NIH through cooperative agreement (U54 HD41859) as part of the Specialized Cooperative Centers Program in Reproduction and Infertility Research. They also acknowledge NIH grant P60DK079637 (Diabetes Research Training Center [DRTC]).

No potential conflicts of interest relevant to this article were reported.

S.W. developed the project, researched data, and wrote the manuscript. S.D. contributed to the writing of the manuscript. F.W. helped plan the project. A.W. helped develop the project, performed some experiments, and wrote the manuscript.

The authors are grateful for the critical review of the manuscript by Dr. Sally Radovick (Johns Hopkins University).