Role of Melanocortin Signaling in Neuroendocrine and Metabolic Actions of Leptin in Male Rats With Uncontrolled Diabetes

Abstract

Although the antidiabetic effects of leptin require intact neuronal melanocortin signaling in rodents with uncontrolled diabetes (uDM), increased melanocortin signaling is not sufficient to mimic leptin's glucose-lowering effects. The current studies were undertaken to clarify the role of melanocortin signaling in leptin's ability to correct metabolic and neuroendocrine disturbances associated with uDM. To accomplish this, bilateral cannulae were implanted in the lateral ventricle of rats with streptozotocin-induced diabetes, and leptin was coinfused with varying doses of the melanocortin 3/4 receptor (MC3/4R) antagonist, SHU9119. An additional cohort of streptozotocin-induced diabetes rats received intracerebroventricular administration of either the MC3/4R agonist, melanotan-II, or its vehicle. Consistent with previous findings, leptin's glucose-lowering effects were blocked by intracerebroventricular SHU9119. In contrast, leptin-mediated suppression of hyperglucagonemia involves both melanocortin dependent and independent mechanisms, and the degree of glucagon inhibition was associated with reduced plasma ketone body levels. Increased central nervous system melanocortin signaling alone fails to mimic leptin's ability to correct any of the metabolic or neuroendocrine disturbances associated with uDM. Moreover, the inability of increased melanocortin signaling to lower diabetic hyperglycemia does not appear to be secondary to release of the endogenous MC3/4R inverse agonist, Agouti-related peptide (AgRP), because AgRP knockout mice did not show increased susceptibility to the antidiabetic effects of increased MC3/4R signaling. Overall, these data suggest that 1) AgRP is not a major driver of diabetic hyperglycemia, 2) mechanisms independent of melanocortin signaling contribute to leptin's antidiabetic effects, and 3) melanocortin receptor blockade dissociates leptin's glucose-lowering effect from its action on other features of uDM, including reversal of hyperglucagonemia and ketosis, suggesting that brain control of ketosis, but not blood glucose levels, is glucagon dependent.

Since its discovery more than 90 years ago, insulin has been the cornerstone of therapy for individuals with type 1 diabetes (1). Recent evidence suggests that like insulin, the adipocyte hormone leptin can also normalize blood glucose levels in rodent models of uncontrolled diabetes (uDM) when administered systemically at pharmacological doses (2, 3). The brain is implicated in this effect, because leptin is equally effective when administered centrally at low doses that have no effect when given systemically (4–8). In addition to hyperglycemia, uDM is associated with a wide range of metabolic, behavioral, autonomic, and neuroendocrine disturbances. These range from a pronounced increase of food intake (“diabetic hyperphagia”) to activation of the hypothalamic-pituitary-adrenal (HPA) axis, which along with increased glucagon secretion and insulin deficiency results in elevated hepatic glucose production (HGP) and ketogenesis (9–12). At the same time, both the reproductive and thyroid axes are inhibited (13). These neuroendocrine responses collectively resemble those induced by fasting (14), which, like uDM, is associated with combined insulin and leptin deficiency (15). Moreover, each of these neuroendocrine and metabolic consequences of uDM are corrected with low-dose intracerebroventricular (icv) leptin treatment (6, 16), suggesting a causal role for leptin deficiency in their genesis. Taken together, these considerations support the hypothesis that in uDM, the effect of leptin to normalize hyperglucagonemia contributes to the associated normalization of hyperglycemia, HGP, and ketoacidosis, because excess glucagon secretion is thought to promote each (3, 5, 6).

Leptin action in the central nervous system (CNS) also exerts potent effects on energy balance, some of which are proposed to involve the hypothalamic melanocortin pathway (17, 18). Leptin activates neurons in the hypothalamic arcuate nucleus (ARC) that express proopiomelanocortin (POMC) (19, 20), which release α-melanocyte-stimulating hormone, an endogenous agonist of neuronal melanocortin receptors (melanocortin 3 receptor [MC3R]/MC4R), and this leptin action inhibits food intake, reduces body weight, and improves glucose metabolism (17, 21). Conversely, leptin inhibits adjacent ARC neurons that express both Agouti-related peptide (AgRP), an antagonist of melanocortin signaling (22), and neuropeptide Y (NPY) (23), peptides that stimulate food intake while also causing glucose intolerance and insulin resistance (24–28). Owing in part to leptin deficiency, uDM is characterized by inhibition of POMC and activation of NPY/AgRP neurons, and leptin reverses both these responses (29). Moreover, the glucose-lowering effects of leptin in rats with uDM are blocked by icv pretreatment with a MC3/4R antagonist, suggesting that intact neuronal melanocortin signaling is required for this leptin effect (30). Yet icv infusion of a MC3/4R agonist does not ameliorate hyperglycemia in uDM, even at doses that potently inhibit food intake, indicating that increased melanocortin signaling alone cannot account for leptin's glucose-lowering effect (30). Consistent with this interpretation, leptin signaling in POMC neurons appears to be neither required nor sufficient to explain the actions of leptin in uDM (31).

Based on these observations, the current studies were conducted to clarify the role of melanocortin signaling in leptin-mediated reversal of metabolic and neuroendocrine dysfunction associated with uDM. In addition, because leptin inhibits the high levels of AgRP expression and release characteristic of uDM, whereas melanocortin agonists, such as α-melanocyte-stimulating hormone, do not, we hypothesized that increased AgRP signaling might explain the failure of melanocortin agonists to ameliorate diabetic hyperglycemia. Thus, we predicted that mice lacking AgRP would be 1) protected against hyperglycemia in uDM, and 2) susceptible to the antidiabetic effect of increased melanocortin signaling. We report that although the MC3/4R antagonist SHU9119 blocked the effect of leptin to normalize blood glucose levels, it only partially attenuated the leptin-mediated normalization of both plasma glucagon and ketone body levels. Because glucagon is a key driver of ketogenesis (32, 33), these findings suggest that a melanocortin-independent action of leptin inhibits glucagon secretion and thereby reduces ketone body production. Consistent with this, increased central melanocortin signaling failed to lower blood glucose, plasma glucagon, or ketone body levels in uDM, despite robustly attenuating food intake. Lastly, because AgRP deficiency had no impact on diabetic hyperglycemia and did not enhance the effect of increased melanocortin signaling to lower blood glucose, increased release of this peptide does not appear to explain differences in the antidiabetic effect of increased leptin vs melanocortin signaling.

Materials and Methods

Animals

All procedures were performed in accordance with National Institutes of Health Guidelines for the Care and Use of Animals and were approved by the Animal Care Committee at the University of Washington. All animals were housed individually in a temperature-controlled room with a 12-hour light, 12-hour dark cycle under specific pathogen-free conditions and had ad libitum access to standard laboratory chow (PMI Nutrition International) and water unless otherwise stated.

Adult male Wistar rats were obtained from Harlan Laboratories. The AgRP knockout (KO) mouse line has previously been generated (34) and was kindly provided to us by Dr Streamson Chua, Jr, (Albert Einstein College of Medicine). AgRP KO and wild-type (WT) mice were generated from homozygous matings and genotyped as previously described (34), and the deletion of AgRP neurons was validated by measuring hypothalamic expression of AgRP levels using real-time PCR.

Surgery

Adult male Wistar rats underwent bilateral cannulation of the lateral ventricle (Alzet, DURECT Corp) under isoflurane anesthesia at stereotaxic coordinates: 1.5 mm lateral, 0.8 mm posterior to bregma, and 3.5 mm below the skull surface (35). Four days later after streptozotocin (STZ) administration to induced uDM, animals were implanted with an osmotic minipump (Alzet, DURECT Corp) sc that was connected to one of the lateral ventricle cannula to enable icv infusion of either vehicle (physiological saline) or the MC3/4R antagonist SHU9119 (Bachem) at 1 of 3 doses (0.5, 5, and 25 nmol/d). A second osmotic minipump was connected to the remaining cannula in the same surgical session to enable simultaneous icv infusion of leptin (1 μg/d) or its vehicle (PBS [pH 7.8]). The connecting line from the osmotic minipump to the pump was filled with 0.9% saline and cut at such a length that animals did not receive the prescribed drug until day 5.

To test the sufficiency of melanocortin action to correct the neuroendocrine and metabolic disturbances associated with STZ-induced diabetes (STZ-DM), a separate cohort of diabetic rats was implanted with a sc osmotic minipump connected to a single lateral ventricle cannula to infuse either the MC3/4R agonist melanotan-II (MTII) (10 μg/d; Bachem) or its vehicle (sterile water).

For mouse studies, animals underwent cannulation of the lateral ventricle (Alzet, DURECT Corp) under isoflurane anesthesia at stereotaxic coordinates: 1.3 mm lateral, 0.7 mm posterior to bregma, and 2.1 mm below the skull surface (36). After induction of STZ-DM, AgRP-deficient and WT control mice were implanted with an osmotic minipump sc that was connected to the lateral ventricle to enable direct infusion of either vehicle (sterile water), MTII (10 μg/d; Bachem), or leptin (1 μg/d, Dr Parlow; National Hormone Peptide Program) into the brain for up to 11 days.

Induction of diabetes

Rats received either 2 consecutive daily sc injections of STZ (40 mg/kg·body weight) to induce uDM or vehicle (NaCit [pH 4.5]) as previously described (6, 37). For mice, 2 sc injections of STZ spaced 3 days apart (150 mg/kg body weight) were given to induce uDM, whereas controls received vehicle (NaCit [pH 4.5]) and remained nondiabetic (38). Animals were defined as diabetic after 2 consecutive days when the fed blood glucose level was more than 200 mg/dL.

Experimental protocols

Body weight, food intake, and blood glucose levels were measured daily in the fed state at 10 am.

Body composition analysis

Measures of body lean and fat mass were determined in live, conscious animals using quantitative magnetic resonance spectroscopy (EchoMRI-700TM; Echo MRI) (39) using the University of Washington Nutrition Obesity Research Center Energy Balance and Glucose Metabolism Core.

Tissue processing, blood collection, and assay

Tissue samples were rapidly dissected, immediately frozen on dry ice, and stored at −80°C. Blood samples for plasma hormonal measures were collected from trunk blood in appropriately treated tubes (6, 37) and centrifuged; the plasma was removed, aliquoted, and stored at −80°C for subsequent assay. Plasma insulin and leptin levels were determined by ELISA (Crystal Chem), ketone bodies using a colorimetric kit (Wako Chemicals), glucagon using a RIA kit (Linco Research), and T₄ levels using an enzyme immunosorbent assay (MP Biomedicals).

Reverse transcription-polymerase chain reaction

RNA was extracted from tissue samples according to manufacturers' instructions (MRC), quantitated by spectrophotometry at 260 nm (Nanodrop 1000; Thermo Scientific), and reverse transcribed with avian myeoloblastosis virus reverse transcriptase (1 μg) (Promega). Real-time PCR was subsequently performed on an ABI Prism 7900 HT (Applied Biosystems) as previously described. Expression levels of each gene were normalized to a housekeeping gene (18S RNA) and expressed as a percentage of vehicle-vehicle (veh-veh) controls. Nontemplate controls were incorporated into each PCR run.

Statistical analyses

Results are expressed as mean ± SEM. Statistical analyses were performed using Statistica (version 7.1; StatSoft, Inc). A one-way ANOVA with a least significant difference post hoc test was used to compare mean values between multiple groups, and a 2-sample unpaired 2-tailed Student's t test was used for 2-group comparisons. In all instances, probability values of less than 0.05 were considered significant.

Results

Antidiabetic effects of icv leptin in uDM require melanocortin signaling

As expected, within 24 hours of STZ administration, animals exhibited markedly reduced plasma insulin levels (Figure 1A) and pronounced hyperglycemia that persisted throughout the study in animals receiving vehicle (Figure 1B). Consistent with previous reports (4–8), continuous icv leptin infusion fully normalized blood glucose levels in STZ-DM rats, and this effect was blocked by icv coinfusion of the MC3/4R antagonist, SHU9119 (25 nmol/d) (Figure 1B) (30). Interestingly, infusion of SHU9119 had no effect on food intake or blood glucose in the absence of leptin treatment (data not shown), despite previous work from our lab and elsewhere (24, 40) showing that in nondiabetic rats, even low doses of SHU9119 potently stimulate food intake. These observations suggest that in uDM, neuronal melanocortin signaling is already reduced sufficiently (due to leptin deficiency) to preclude any further decrease induced by a melanocortin receptor antagonist. Even at 5- and 50-fold lower doses (5.0 and 0.5 nmol/d icv), SHU9119 blunted the glucose-lowering effects of icv leptin (Figure 1B). This being said, the disruption of leptin's antidiabetic effect was modest in STZ-DM animals receiving the lowest dose of SHU9119 and was likely explained by the more complete attenuation of leptin's ability to normalize diabetic hyperphagia (Figure 1, B and C). In addition, the effect of leptin to reduce body fat stores in STZ-DM rats was blocked by SHU9119, and yet leptin's ability to preserve lean body mass was retained, except at the highest SHU9119 dose (Figure 1, E and F). Taken together, these data support the conclusion that the glucose-lowering action of leptin in the brain requires intact melanocortin signaling, but they do not address 1) whether increased melanocortin signaling can explain this leptin effect, or 2) the extent to which effects of leptin on metabolic and neuroendocrine parameters (other than blood glucose) involve melanocortin signaling.

Figure 1.

Antidiabetic effects of CNS leptin in STZ-DM rats require melanocortin signaling. Plasma insulin (A), fed blood glucose levels (B), mean daily food intake (C), body weight change (D), percent body fat (E), and lean body mass (F) in STZ-induced diabetic animals receiving either vehicle (STZ-veh-veh) or leptin (STZ-veh-lep), or vehicle and the MC3/4R antagonist, SHU9119 (STZ-SHU-lep), at either 0.5, 5, or 25 nmol/d (n = 7/group). Data represent mean ± SEM. *, P < .05 vs STZ-veh-veh; #, P < .05 vs STZ-veh-lep.

Role of melanocortin signaling in the effect of CNS leptin to normalize hyperglucagonemia in STZ-DM

Besides diabetic hyperglycemia and hyperphagia, leptin also corrects most metabolic and neuroendocrine abnormalities associated with uDM (6), including normalization of hyperglucagonemia, an effect that may contribute to leptin's glucose-lowering effects (3, 5, 6). A key question, therefore, is whether the effect of leptin to normalize plasma glucagon levels also requires melanocortin signaling. Consistent with previous reports (5, 6), continuous icv leptin infusion restored elevated plasma glucagon levels to normal in STZ-DM rats (Figure 2A). Unexpectedly, leptin-mediated normalization of glucagon levels was not blunted by icv infusion of SHU9119 at a low dose and was only partially attenuated at the higher dose (Figure 2A), despite complete reversal of leptin's glucose-lowering effect (Figure 1B). These findings suggest that 1) reversal of hyperglucagonemia is insufficient to ameliorate hyperglycemia in uDM, and 2) leptin normalizes glucagon levels in uDM, at least in part, via a mechanism that is independent of melanocortin signaling.

Figure 2.

Effect of CNS leptin in STZ-DM rats to correct hyperglucagonemia and ketosis occurs, in part, independent of melanocortin signaling. Plasma glucagon (A), ketone body levels (B), correlation between plasma glucagon and ketone body levels (C), plasma corticosterone (D), and thyroid levels (E) in STZ-induced diabetic animals receiving either vehicle (STZ-veh-veh) or leptin (STZ-veh-lep), or vehicle and the MC3/4R antagonist, SHU9119 (STZ-SHU-lep), at either 0.5, 5, or 25 nmol/d (n = 7/group) at the completion of the study. Data represent mean ± SEM. *, P < .05 vs STZ-veh-veh; #, P < .05 vs STZ-veh-lep.

Because hyperglucagonemia is implicated in the effect of uDM to increase hepatic production of ketone bodies as well as glucose, our evidence that CNS melanocortin signaling is required for leptin reversal of hyperglycemia but not hyperglucagonemia raises interesting questions regarding the control of ketosis by the brain. Specifically, we hypothesized that although blockade of neuronal MC3/4Rs completely prevents the effect of leptin to normalize blood glucose levels in rats with STZ-DM (30), it will not fully block leptin's effect to lower plasma ketone bodies. As expected, elevated plasma levels of ketone bodies in STZ-DM rats were normalized by icv leptin infusion, and similar to what was observed for circulating glucagon levels, this leptin effect was only partially attenuated by icv coinfusion of SHU9119 (Figure 2B). These findings raise the possibility that a melanocortin-independent mechanism contributes to leptin-mediated inhibition of glucagon secretion and that plasma ketone body levels change in parallel with the degree of glucagon elevation. In support of this model, plasma levels of glucagon were strongly predictive of plasma ketone body levels across all groups (r = 0.55; P < .001) (Figure 2C).

We previously reported that in rats with STZ-DM, icv leptin also normalizes plasma levels of corticosterone, a marker of HPA axis activity (6). In the current studies, we again found that icv leptin effectively lowered elevated circulating corticosterone levels in uDM. However, this effect was not blunted at the lowest dose of the MC3/4R antagonist, despite blocking leptin's antidiabetic effects, suggesting that reversal of hypercorticosteronemia is insufficient to ameliorate hyperglycemia in uDM (Figure 2D). Unlike leptin-mediated suppression of plasma glucagon and ketone body levels, however, leptin-mediated normalization of corticosterone levels was completely blocked at the highest dose of SHU9119, suggesting that melanocortin signaling appears to be required for the ability of leptin to reverse HPA axis activation induced by uDM. In contrast, although icv leptin effectively raised low circulating levels of T₄ into the normal range in rats with uDM, this effect was only partially attenuated with the MC3/4R antagonist (Figure 2E). These findings suggest that although the melanocortin pathway participates in leptin-mediated regulation of both the HPA and hypothalamic-pituitary-thyroid (HPT) axes, a melanocortin-independent pathway contributes importantly to the effect of leptin on the latter but not the former.

Effect of increased melanocortin signaling on diabetic hyperglycemia in STZ-DM rats

Despite clear evidence that intact melanocortin signaling is required for leptin's glucose-lowering effects, previous studies suggest that icv administration of a melanocortin agonist cannot mimic this effect of leptin (30). Thus, melanocortin signaling appears to be necessary, but not sufficient, to explain the CNS effect of leptin on diabetic hyperglycemia. Consistent with this conclusion, we replicated earlier evidence (30) that icv administration of the melanocortin receptor agonist, MTII, fails to attenuate diabetic hyperglycemia in rats with STZ-DM relative to vehicle-treated diabetic controls, despite transient attenuation of diabetic hyperphagia (Figure 3, B and C). As a consequence, STZ-DM animals receiving MTII lost more weight than vehicle-treated controls with STZ-DM, comparable with that of STZ-veh pair-fed controls (Figure 3D).

Figure 3.

Administration of a melanocortin agonist is not sufficient to ameliorate diabetic hyperglycemia in STZ-DM rats. Plasma insulin (A), fed blood glucose levels (B), mean daily food intake (C), and body weight change (D) in nondiabetic or STZ-induced diabetic rats receiving either icv vehicle and fed ad libitum (STZ-veh-AL) or pair-fed (STZ-veh-PF), or the melanocortin agonist, MTII (STZ-MTII) (n = 6–8/group). Data represent mean ± SEM. *, P < .05 vs veh-veh; #, P < .05 vs STZ-MTII.

Effect of increased melanocortin signaling on metabolic and neuroendocrine parameters in rats with uDM

We next examined whether increased melanocortin signaling is sufficient to mimic the effect of leptin to correct other neuroendocrine and metabolic abnormalities characteristic of uDM. As expected, plasma levels of glucagon, ketone bodies, and corticosterone were elevated in rats with STZ-DM that received vehicle relative to nondiabetic controls, whereas plasma thyroid levels were reduced (Figure 4, A–D). However, icv administration of MTII failed to alter any of these responses to uDM (Figure 4, A–D). Increased melanocortin signaling, therefore, fails to mimic any of the beneficial actions of leptin on metabolic and neuroendocrine perturbations of uDM. Moreover, consistent with our earlier observations, we found that plasma glucagon levels were strongly predictive of plasma ketone body levels across veh-veh-, STZ-veh-, and STZ-MTII-treated animals. (r = 0.425; P < .05). Interestingly, however, this relationship was disrupted when STZ-veh-pair-fed (PF) animals were included in the analysis (r = 0.289; P = not significant). Specifically, STZ-veh-PF animals exhibited higher plasma ketone body levels for the same level of plasma glucagon. One possible explanation for this further elevation of ketone body levels is the effect of food restriction to raise circulating norepinephrine levels in rodents (41), and because norepinephrine infusion during somatostatin-induced insulin deficiency in humans results in an augmented and sustained increase in ketone body production (42), future studies are warranted to investigate this hypothesis. Together, these findings raise the intriguing possibility that fasting and uDM drive ketogenesis via mechanisms that are at least partially distinct and separate from one another.

Figure 4.

Administration of a melanocortin agonist to STZ-DM rats is not sufficient to ameliorate other metabolic abnormalities associated with uDM. Plasma glucagon (A), ketone body levels (B), plasma corticosterone (C), and thyroid levels (D) in nondiabetic or STZ-induced diabetic rats receiving either icv vehicle and fed ad libitum (STZ-veh-AL) or pair-fed (STZ-veh-PF), or the melanocortin agonist, MTII (STZ-MTII), at the completion of the study (n = 6–8/group). Data represent mean ± SEM. *, P < .05 vs veh-veh; #, P < .05 vs STZ-MTII.

Role of AgRP in diabetic hyperglycemia and the response to MTII and leptin

One possible explanation for these findings is that increased hypothalamic synthesis and release of the MC3/4R inverse agonist AgRP both drives hyperglycemia in uDM and abrogates the beneficial effect of MTII. Consistent with published work (18, 43), we found that hypothalamic expression of Npy and AgRP were markedly elevated (P < .05), whereas hypothalamic expression of Pomc was reduced (P < .05), in WT mice with STZ-DM relative to nondiabetic controls. Further, although each of these changes in hypothalamic neuropeptide gene expression was reversed by treatment with leptin (P < .05 for each), icv administration of MTII did not have this effect (data not shown). The observation that leptin inhibits NPY/AgRP neurons in uDM but MTII does not offers a potential explanation for failure of icv MTII to replicate the antidiabetic effects of leptin. Based on this reasoning, we hypothesized that mice lacking AgRP would be 1) protected against hyperglycemia in uDM, and 2) more susceptible to the antidiabetic effects of increased melanocortin or leptin signaling. To test this hypothesis, we induced uDM with STZ (150 mg/kg × 2) in AgRP KO mice and their WT littermate controls 1 week after cannulation of the lateral ventricle. Diabetic mice of each genotype then received chronic icv infusion of either 1) vehicle, 2) MTII (10 μg/d), or 3) leptin (1 μg/d). We found that the effect of STZ to raise levels of blood glucose did not differ between genotypes (Figure 5A), suggesting that AgRP signaling is not required for this response. To investigate the response to increased melanocortin signaling in these mice, MTII was administered icv to both genotypes at a dose that effectively reduced body weight in a separate cohort of nondiabetic controls (data not shown). We found that icv MTII failed to attenuate diabetic hyperglycemia in either WT or AgRP KO mice (Figure 5B), suggesting that even in the absence of AgRP, increased melanocortin signaling per se is unable to mimic the beneficial effect of leptin on diabetic hyperglycemia. We next determined whether AgRP signaling is required for leptin's antidiabetic actions. Consistent with previous reports (5, 31, 38), we found that leptin administration directly into the brain normalized blood glucose levels in STZ-DM WT mice (Figure 5C). In AgRP KO mice, this antidiabetic was not only fully preserved, but the animals were somewhat more sensitive to the glucose-lowering effects of leptin (Figure 5C).

Figure 5.

Administration of a melanocortin agonist and leptin on glycemia in STZ-DM AgRP KO mice. Blood glucose levels in STZ-induced diabetic AgRP KO mice and their littermate controls that receive icv administration of either vehicle (A), the melanocortin agonist, MTII (B), or leptin (C) (n = 4–7/group). Data represent mean ± SEM. *, P < .05 vs AgRP KO.

Discussion

Previous studies have demonstrated that leptin's antidiabetic effects in uDM require melanocortin signaling, but whether melanocortin signaling contributes to leptin-mediated correction of other metabolic and neuroendocrine disturbances characteristic of uDM is unknown. Here, we report that although leptin's glucose-lowering effects were blocked by coinfusion of the MC3/4R antagonist, SHU9119, the effects of leptin to normalize plasma glucagon and ketone body levels were only partially reversed. These data suggest that a melanocortin-independent action of leptin plays a key role to inhibit glucagon secretion and that this effect, in turn, explains the variation in ketone body levels. Consistent with these findings, increased CNS melanocortin signaling failed to lower blood glucose, glucagon, or ketone body levels. Furthermore, our findings suggest that increased CNS melanocortin signaling per se is insufficient to mimic leptin's ability to ameliorate diabetic hyperglycemia, even when AgRP is absent. Taken together, these data suggest that AgRP is not a major driver of diabetic hyperglycemia, that leptin's glucose-lowering actions involve mechanisms additional to increased melanocortin signaling, and that leptin-mediated inhibition of hyperglucagonemia plays a major role to normalize diabetic ketosis but not hyperglycemia.

Hyperglycemia and ketosis are hallmarks of uDM, and hyperglucagonemia is implicated in both (9, 11, 33). For example, suppression of glucagon with either somatostatin or leptin in either humans with type 1 diabetes or in rodents with chemically induced diabetes lowers elevated levels of blood glucose and reverses diabetic ketosis (2, 3, 32, 44), whereas glucagon receptor-deficient mice are resistant to both hyperglycemia or ketosis after STZ (45, 46). Interestingly, leptin action in the brain can also normalize elevated levels of both glucose and ketone bodies in rodents with STZ-DM, but details of how this occurs are lacking. Importantly, these effects of central leptin are associated with reversal of hyperglucagonemia (5, 6), but neither the extent to which glucagon lowering mediates leptin's antidiabetic effects nor the neurocircuits that underlie leptin regulation of glucagon secretion are known. Although previous evidence suggests that leptin action on POMC neurons lowers plasma glucagon levels in nondiabetic animals (47), a recent report shows that the effect of leptin to suppress hyperglucagonemia in uDM does not require leptin signaling in POMC neurons (31). An alternative possibility is that the glucagon-suppressing effects of leptin involve the increase of melanocortin signaling that results from inhibition of AgRP neurons. Our current findings indicate that the effect of leptin to normalize hyperglucagonemia is only partially attenuated by coinfusion of a MC3/4R antagonist, whereas this intervention fully blocked leptin's ability to normalize blood glucose levels. Thus, a melanocortin-independent pathway of leptin action does appear to be involved, because the effect of leptin to normalize plasma ketone body levels was only partially attenuated by MC3/4R blockade. These results suggest that 1) a melanocortin-independent mechanism contributes to leptin inhibition of glucagon secretion, 2) this effect drives leptin-mediated inhibition of hepatic ketone body production, and 3) mechanisms additional to reversal of hyperglucagonemia contribute to leptin's glucose-lowering effects.

Among the findings in support of these conclusions is the positive correlation between plasma glucagon and ketone body levels, indicating that the extent to which leptin's effects on glucagon were blocked by central melanocortin receptor blockade were paralleled by similar changes in ketone body levels. Furthermore, icv infusion of a melanocortin receptor agonist failed to attenuate either hyperglucagonemia or ketosis in uDM, whereas icv leptin fully normalized both. These findings are consistent with previous evidence that normalization of plasma glucagon levels in uDM with either a physiological leptin replacement protocol (37) or by icv administration of the antidiabetic peptide brain-derived neurotrophic factor (BDNF) (48) normalize both plasma glucagon and ketone body levels without normalizing hyperglycemia in rats with uDM. Collectively, these data suggest that reversal of hyperglucagonemia in uDM can be mediated centrally and is sufficient to reverse diabetic ketoacidosis but not hyperglycemia. Moreover, the underlying CNS mechanism involves both melanocortin dependent and independent mechanisms.

Interestingly, although melanocortin signaling appears to be required for leptin's glucose-lowering effects, both melanocortin dependent and independent mechanisms appear to mediate leptin's actions on food intake and most other metabolic or neuroendocrine derangements in uDM. Consistent with this interpretation, leptin reduces, but does not normalize, food intake in nonobese Mc4r-deficient mice (49). Moreover, leptin is capable of reversing the powerful effects of uDM on both the HPA and HPT axes. As in fasting (14), plasma levels of T₄ are reduced in uDM, and this effect is reversed by central leptin administration (16), presumably via stimulation of TRH neurons in the hypothalamic paraventricular nucleus (50). Our finding that the effect of leptin to reverse low circulating thyroid levels in rats with uDM was strongly attenuated by coinfusion of the MC3/4R antagonist adds to existing evidence that melanocortin signaling stimulates TRH neurons in response to leptin and thereby activates the HPT axis (50, 51). However, our finding that leptin-mediated normalization of thyroid levels was not completely blocked by the MC3/4R antagonist supports previous evidence that leptin can also act directly to stimulate TRH neurons in the hypothalamic paraventricular nucleus (eg, in a melanocortin-independent manner) (52). A similar arrangement has been advanced to explain leptin-mediated inhibition of the HPA axis, with some effects being melanocortin dependent, whereas others are not (53). However, because MC3/4R blockade can block leptin's glucose-lowering effects without attenuating leptin-mediated normalization of corticosterone levels, it suggests that suppression of hypercorticosteronemia alone in uDM cannot explain leptin's antidiabetic effects.

At first glance, it may seem confusing that although neuronal melanocortin signaling is required for leptin's glucose-lowering effects in uDM, increased melanocortin signaling is unable to mimic this leptin effect (30). After all, restoring leptin action selectively to POMC neurons (by reactivating leptin receptors on POMC neurons) lowers blood glucose and plasma glucagon levels, and improves hepatic insulin sensitivity, in mice that otherwise lack leptin receptors (47). This finding differs from our work, however, in that we studied animals with combined insulin and leptin deficiency (which characterizes uDM), and previous work shows that leptin signaling in POMC neurons is neither required nor sufficient to mediate most leptin's glucose-lowering effects in this setting (31). One potential explanation for these disparate observations is that POMC cell activation by leptin effectively increases insulin sensitivity in liver (47) or other tissues, but such an effect is of limited benefit in the setting of severe insulin deficiency. In addition, previous work from the Rossetti group found that in normal, nondiabetic rats, icv administration of pharmacological doses of leptin fails to affect the overall rate of HGP (54). However, rather than having no effect on hepatic glucose metabolism, it did stimulate gluconeogenesis (via activation of the central melanocortin pathway) and simultaneously inhibit glycogenolysis (via a melanocortin independent pathway) (54). Moreover, icv leptin infusion reversed the insulin resistance induced by short-term exposure of rats to a high-fat diet by reducing both glycogenolysis and gluconeogenesis (55). In each case, these leptin effects were attributed to CNS-mediated changes of hepatic insulin sensitivity. Our previously published work demonstrates that leptin action in the brain normalizes blood glucose levels in a rodent model of uDM (6). Importantly, this leptin effect involves a novel, insulin-independent mechanism characterized by reduced rates of HGP, an effect that is therefore distinct from mechanisms involving changes of hepatic insulin sensitivity. Whether leptin-mediated suppression of HGP in uDM involves decreases of glycogenolysis, gluconeogenesis, or both, and whether such effects are melanocortin dependent, remains to be determined.

Another possible explanation for our findings is that leptin can regulate the melanocortin signaling pathway in 2 ways: by activating POMC neurons to increase the hypothalamic release of melanocortin agonists of MC4R while also inhibiting the release of AgRP, an inverse agonist of MC4R. Accordingly, the effect of SHU9119 to block the ability of leptin to normalize blood glucose levels in uDM can potentially be explained by the combined effects of leptin stimulation of POMC neurons (to increase melanocortin signaling) and leptin inhibition of NPY/AgRP neurons (which also increases melanocortin signaling). However, icv MTII failed to attenuate diabetic hyperglycemia in AgRP-deficient mice and the degree of hyperglycemia induced by STZ was comparable between WT control and in AgRP-deficient mice in the absence of MTII treatment, demonstrating that AgRP is not a key driver of diabetic hyperglycemia.

Yet another possibility is that intact leptin signaling is required for the melanocortin pathway to function properly. Consistent with our findings, previous evidence suggests that leptin effects on cardiovascular function (30) and renal sympathetic nerve activity (56) also requires melanocortin signaling. Yet in rodent models of reduced or deficient leptin signaling, 1) MC3/4R activation fails to mimic the effect of leptin (30, 57), and 2) the response to MC3/4R activation is either attenuated or absent relative to nondiabetic, WT controls (30, 57). Activation of pathways that are normally inhibited by leptin (eg, melanin-concentrating hormone and orexin, in addition to NPY and AgRP), and suppression of other pathways that are normally stimulated by leptin (eg, cocaine-amphetamine-related transcript and BDNF) may therefore block the effect of MC3/4 agonists (48).

The identity of the melanocortin-independent pathway(s) that contribute to leptin's antidiabetic effects are as yet unknown but may include inhibitory γ-aminobutyric acid (GABAergic; vesicular GABA transporter-positive) neurons. Such neurons have recently emerged as key targets of leptin action, because deletion of leptin receptors from GABAergic neurons causes marked obesity and hyperglycemia (31) relative to deletion of leptin receptors from individual neuropeptide-containing populations (eg, POMC [58] or NPY/AgRP [59]) or brain areas (eg, VMN) (60, 61). Moreover, reexpression of leptin receptors specifically in GABAergic neurons is sufficient to restore most of leptin's antidiabetic effects in uDM (31). Within the hypothalamus, GABAergic neurons are found in key areas for metabolic regulation including the ARC, dorsomedial hypothalamus, and lateral hypothalamic area, and future studies are warranted to determine their contribution to leptin's glucose-lowering effects and/or other metabolic benefits in uDM. The VMN is of particular interest with respect to the melanocortin-independent effects of leptin, because 1) leptin administration directly into the VMN is sufficient to normalize diabetic hyperglycemia and hyperglucagonemia in STZ-DM (38), 2) BDNF signaling limited to the VMN attenuates diabetic hyperglycemia via a mechanism associated with inhibition of glucagon secretion and reduced rates of HGP (48), and 3) leptin activates BDNF neurons in this brain area (62), yet only a subset of these BDNF-expressing neurons appear to express Mc4r (63).

In summary, our data suggest that in diabetic animals, leptin-mediated suppression of hyperglucagonemia occurs via both melanocortin dependent and independent mechanisms and that the degree of glucagon inhibition is a potent determinant of the degree of elevation in the plasma level of ketone bodies but not glucose. Furthermore, our findings suggest that increased CNS melanocortin signaling per se is insufficient to mimic leptin's ability to ameliorate diabetic hyperglycemia, even when AgRP is absent.