The fat-derived hormone, leptin, is well known to regulate body weight. However, there is now substantial evidence that leptin also plays a primary role in the regulation of glucose homeostasis, independent of actions on food intake, energy expenditure or body weight. As such, leptin might have clinical utility in treating hyperglycemia, particularly in conditions of leptin deficiency, such as lipodystrophy and diabetes mellitus. The mechanisms through which leptin modulates glucose metabolism have not been fully elucidated. Leptin receptors are widely expressed in peripheral tissues, including the endocrine pancreas, liver, skeletal muscle and adipose, and both direct and indirect leptin action on these tissues contributes to the control of glucose homeostasis. Here we review the role of leptin in glucose homeostasis, along with our present understanding of the mechanisms involved. (J Diabetes Invest, doi: 10.1111/j.2040-1124.2012.00203.x, 2012)
The inheritable obese phenotype in the ob/ob and db/db mouse lines were discovered several decades ago1,2. Through cross-circulation experiments between ob/ob and db/db mice, Coleman postulated that ob/ob mice lacked a circulating satiety factor, whereas db/db mice lacked a functional responsive site to this factor3. The identity of this satiety factor remained unknown until 1994, when the ob gene was identified through positional cloning4. ob/ob mice possess a single nonsense mutation in the ob gene, resulting in the production of a truncated form of the protein product, leptin, and undetectable circulating leptin levels4,5. Injection of wild-type leptin can lower body weight in both ob/ob and wild-type mice5. Although rare, mutations in the human leptin gene have been identified in several families6–10, and similar to ob/ob mice, humans with homozygous null mutations in the leptin gene have undetectable circulating leptin levels, and are obese with a plethora of metabolic, reproductive and immune dysfunctions; these patients can be effectively treated with leptin replacement therapy7,11–14.
The discovery of the leptin receptor (Lepr), encoded by the db gene followed soon after leptin was identified15. The db gene encodes an alternatively spliced transcript, capable of producing six leptin receptor isoforms (Lepr-a to Lepr-f)16. db/db mice have an insertion mutation in the db gene that prevents the normal splicing of the Lepr-b isoform, resulting in a truncated intracellular signaling domain16,17. Lepr-b has the longest intracellular domain of the leptin receptor isoforms16,17, and signals through Janus kinase (JAK)/signal transducers and activators of transcription (STAT) pathways to influence transcription of target genes18–21. When back-crossed to the same genetic background, ob/ob and db/db mice have nearly identical phenotypes22,23, indicating that Lepr-b is primarily responsible for carrying out leptin action. Thus, Coleman’s prediction regarding the defective production of a satiety factor in ob/ob mice, and an unresponsive satiety centre in db/db mice was confirmed. It is now well known that leptin normally circulates in proportion to body fat24,25, and acts on hypothalamic neurons to inhibit food intake and increase energy expenditure, leading to a reduction in body weight5,26,27; thereby, the loss of leptin action results in hyperphagia, decreased energy expenditure and profound obesity.
The Role of Leptin In Glucose Homeostasis
Although leptin was originally recognized for its role as a satiety factor, it is now implicated in a wide variety of biological functions, including the regulation of glucose homeostasis. In addition to obesity, ob/ob and db/db mice have a phenotype similar to human type 2 diabetes, including hyperglycemia, hyperinsulinemia and insulin resistance, and thus have been widely used for decades as animal models of diabetes2,27–32. It can be postulated that the perturbed glucose metabolism that accompanies leptin or leptin receptor deficiency is secondary to obesity and hyperphagia. However, several lines of evidence show that leptin regulates glucose metabolism independent of its effects on body weight and food intake. First, hyperinsulinemia occurs in ob/ob mice29,31–33, along with transient hypoglycemia32,33, before the onset of insulin resistance, hyperglycemia and obesity. Likewise, early hyperinsulinemia is also common to rodents with disrupted leptin receptor function, including db/db mice, Zucker fatty (fa/fa) rats and corpulent rats30,34–37. To further examine the chronology of events in leptin deficiency, we examined the effect of acutely disrupting endogenous leptin action in wild-type mice using a leptin antagonist38. Adult mice treated with this antagonist developed fasting- and glucose-stimulated hyperinsulinemia, and insulin resistance within 3 days, without significant accompanying changes in body composition or body weight. Thus, when leptin action is disrupted or absent, perturbations in glucose homeostasis chronologically precede obesity. Similar to leptin deficient ob/ob mice, rodents and humans with a near or complete loss of adipose tissue (lipodystrophy) are also hypoleptinemic, and have hyperinsulinemia, insulin resistance and hyperglycemia39,40. Thus, regardless of adiposity, inappropriately low leptin levels result in perturbed glucose homeostasis.
Further supporting the body weight independence of leptin action on glucose homeostasis, leptin replacement therapy can improve glucose metabolism in mice and humans with either congenital leptin deficiency or lipodystrophy7,11–14,27,39,40. Leptin administration reduces circulating insulin and glucose levels in ob/ob mice, indicative of increased insulin sensitivity, to a greater extent than pair-feeding41–43, showing that the actions of leptin on glucose homeostasis cannot simply be explained by reduced food intake. Furthermore, we have shown that leptin can lower circulating insulin and glucose levels in ob/ob mice within 1 or 2 days, before changes in body weight occur44,45. We also showed that temporary leptin therapy has longer lasting effects on blood glucose than food intake in ob/ob mice46. Finally, low doses of leptin that do not alter body weight or food intake can normalize circulating insulin and glucose levels in ob/ob mice27,47. These studies firmly establish that leptin replacement has a more potent effect on glucose metabolism than body weight in leptin deficient animals.
Perhaps the most compelling evidence of the profound effect of leptin on glucose homeostasis is that leptin administration can normalize blood glucose levels in non-obese rodent models of insulin deficient, type 1 diabetes. Leptin infusion or gene therapy, can reverse hyperglycemia without a detectable rise in circulating insulin levels in streptozotocin (STZ)-treated rats48–54 and mice55–57, non-obese diabetic (NOD) mice49,58, insulin deficient Akita mice59,60 and BioBreeding rats with virally-induced β-cell destruction61. Leptin therapy also normalizes water intake and urine output, and reverses glycosuria, hyperketonemia and hyperphagia in insulin deficient rodents50,52,54,58, indicating improved overall health of these animals. Despite the anorexigenic effect of leptin, pair-feeding studies have shown that the glucose lowering in response to leptin therapy cannot be explained by decreased food intake49,52,58. Furthermore, the reversal of glycosuria with leptin therapy rules out euglycemia induced by increased glucose output in urine, and supports a direct antidiabetic effect of leptin on glucose metabolism.
The normalization of blood glucose levels by leptin therapy in insulin deficient rodents correlates with increased insulin sensitivity48–50,55,58,62. We found that STZ-diabetic mice treated with leptin have heightened insulin sensitivity, even compared with non-diabetic controls, and thus postulated that the profound insulin sensitizing effect of leptin in this model might compensate for residual insulin levels in STZ-treated rodents55. Indeed, low-dose insulin administration that was ineffective alone was found to dramatically reduce glucose levels in STZ-diabetic mice when combined with a dose of leptin that was ineffective alone63. In addition, leptin therapy decreases levels of counter regulatory hormones49,50,55,58, which could contribute to the glucose-lowering action of leptin. In insulin deficient rodents leptin administration has been found to decrease corticosterone and growth hormone (GH) levels, hormones which impair insulin sensitivity50,55. Leptin also robustly decreases glucagon levels in this model49,58. Hyperglucagonemia is a common characteristic of diabetes, and is a requisite for hyperglycemia in several models of insulin deficiency64. Interestingly, suppression of endogenous glucagon65 or antagonism of glucagon receptor signaling decreases hyperglycemia in STZ-diabetic rats66–68. Furthermore, mice with genetic knockout of glucagon receptors are resistant to developing STZ-induced diabetes69. Thus, the potent suppressive effect of leptin on glucagon levels might contribute to restoration of euglycemia in leptin-treated insulin deficient rodents.
Circulating leptin levels rapidly fall after the decrease in insulin levels induced by STZ administration in rats, and are acutely restored by insulin injection70. Low plasma leptin levels are also observed in people with newly-diagnosed type 1 diabetes, before the initiation of insulin therapy, and are subsequently elevated by insulin therapy71. Interestingly, by continuously administering a dose of leptin that prevents the decrease in leptin levels in STZ-treated rats, hyperphagia72, insulin resistance73 and hyperglucagonemia73 are prevented. It is therefore intriguing to consider that key metabolic disturbances associated with insulin deficiency are actually secondary to the underlying hypoleptinemia. Importantly, despite reversal of hyperphagia, insulin resistance and hyperglucagonemia, restoration of physiological leptin levels does not reverse hyperglycemia in STZ-treated rats73. Thus, the glucose lowering effect of leptin in the context of insulin deficient diabetes seems to require supraphysiological leptin levels. Taken together, the profound effect of leptin deficiency and leptin administration on glucose metabolism in ob/ob mice and rodent models of type 1 diabetes and lipodystrophy provide ample evidence that leptin plays a primary role in glucose homeostasis.
Mechanisms of Leptin Action on Glucose Homeostasis
Leptin receptors, including the Lepr-b isoform, are widely expressed throughout the central nervous system and peripheral tissues18. Numerous studies have been carried out to try to identify the specific target tissues that mediate leptin action on glucose metabolism. From this body of work, both the hypothalamus and several extra-hypothalamic sites within the brain have emerged as major targets of leptin. Others have recently provided thorough reviews focused on how these specific leptin-activated neural pathways regulate glucose homeostasis74–76. In addition, direct leptin signaling in peripheral tissues can also modulate glucose metabolism. The direct and indirect actions of leptin on peripheral target tissues that are likely to contribute to glucose homeostasis are reviewed here, and summarized in Figures 1 and 2, respectively.
Leptin Action on Pancreatic β-Cells
Leptin, either through gene delivery or recombinant peptide administration, rapidly lowers circulating insulin levels in ob/ob mice44,45,77–79, but is ineffective in db/db mice79,80. In fact, leptin administration lowers insulin secretion in vivo within minutes, and simultaneously causes an acute rise in blood glucose levels, indicating that decreased insulin levels are not secondary to increased insulin sensitivity79. Leptin has an inhibitory effect on insulin synthesis as well; a single leptin injection in ob/ob mice decreases preproinsulin messenger ribonucleic acid (mRNA) in islets within 24 h, by alterations in transcription factor binding to the insulin promoter44. A potential mechanism is the induction of JAK/STAT-mediated expression of suppressor of cytokine signaling 3 (SOCS3) in β-cells by leptin, which subsequently inhibits preproinsulin gene transcription81. Leptin administration has also been reported to acutely decrease glucose-stimulated insulin levels in normal rodents82, albeit the effect is less robust than in ob/ob mice45.
Leptin binding, Lepr-b transcript expression, and functional Lepr-b signaling have been shown in pancreatic islets or β-cells from mice, rats and humans80,81,83–85. In vitro studies support a direct suppressive action of leptin on basal- and glucose-stimulated insulin gene expression and secretion in β-cells (Figure 1)44,79–81,84–88. Leptin robustly inhibits insulin secretion from isolated islets and the perfused pancreas of ob/ob mice80,84. Most studies using islets or perfused pancreata from non-leptin deficient animals, and β-cell lines show an inhibitory effect of leptin on insulin secretion79,80,86,89–93, although in a few reports leptin administration had no effect87,94–96, or even stimulated97,98 insulin secretion. The reasons for these apparent discrepancies are not clear.
Our group previously showed that in vitro leptin can reduce glucose transporter 2 (GLUT2) phosphorylation, glucose transport and intracellular adenosine triphosphate (ATP) levels99. In addition, leptin activates ATP-sensitive potassium (KATP) channels and hyperpolarizes β-cells, thereby decreasing intracellular calcium concentrations84,85,100. Leptin has also been shown to suppress cyclic adenosine monophosphate (cAMP)-induced insulin secretion, through activation of phosphodiesterase-3B (PDE3B)86,87,101. Leptin also inhibits protein kinase C (PKC)-induced insulin secretion91,102. Supporting this, leptin can inhibit acetylcholine103, and glucagon-like peptide 1 (GLP-1)-stimulated insulin secretion86. Thus, leptin might directly suppress insulin secretion by inhibiting glucose sensing and KATP channel closure, and by inhibiting cAMP and PKC-mediated insulin secretion in β-cells.
To uncover the physiological role of direct leptin action on β-cells in vivo, we used the Cre-lox system to generate mice with disrupted Lepr-b signaling in β-cells104. Mice with a disrupted signaling domain of the Lepr gene in β-cells and the hypothalamus showed hyperinsulinemia and fasting hypoglycemia, supporting a physiological role of leptin to inhibit insulin secretion from β-cells104. Furthermore, these mice showed glucose intolerance, impaired glucose-stimulated insulin secretion, insulin resistance and mild obesity. Despite partial hypothalamic recombination of the Lepr gene, these mice maintained normal anorexigenic responses to leptin administration, indicating that the phenotype was not a result of increased food intake. Interestingly, we found that insulin resistance was secondary to hyperinsulinemia in mice with disrupted Lepr signaling in β-cells; administration of metformin to these mice improved insulin sensitivity, but was unable to ameliorate hyperinsulinemia, whereas administration of diazoxide to inhibit insulin secretion ameliorated both hyperinsulinemia and insulin resistance105. In a study carried out by Morioka et al.106, mice with a pancreatic and duodenal homeobox-1 (Pdx1)-cre-mediated loss of Lepr expression in the pancreas showed hyperinsulinemia, but in contrast to our studies, had elevated glucose-stimulated insulin secretion, and improved glucose tolerance. Thus, whether the loss of inhibitory leptin signals in β-cells alone results in impaired or improved glucose homeostasis is unclear. Because the Pdx1-cre used by Morioka et al. was recently found to be expressed in the brain107, and the Pdx1 promoter is active in the gut108, Lepr expression might have been disrupted in extrapancreatic sites of their mice, potentially contributing to differences between our two mouse models. Another possible explanation is the presence of mild obesity in our mice, suggesting that an additional insult is required to observe deleterious effects of disrupted β-cell leptin signaling. Interestingly, when the mice of Morioka et al.106 were fed a high-fat diet, they showed impaired glucose tolerance and glucose stimulated insulin secretion, in agreement with our mouse model.
Evidence suggests that leptin also inhibits insulin secretion through central mechanisms (Figure 2). Mice with neuronal disruption of leptin signaling show hyperinsulinemia, whether body weight is unaltered109 or increased110. Intracerebroventricular (ICV) administration of leptin either as a peptide or as a gene therapy lowers insulin levels111–115, although in one study the effect was modest116, whereas in another, leptin did not lower meal-stimulated insulin levels compared with pair-fed controls43. The effect of central leptin action on insulin levels is consistent with increased sympathetic tone to β-cells. Indeed, a recent study showed that acute ICV leptin administration suppressed glucose-stimulated insulin secretion in a manner dependent on activation of the sympathetic nervous system (SNS)113. One SNS-mediated mechanism of leptin action on the β-cell has been proposed to occur through sympathetic inhibition of bioactive osteocalcin secretion from osteoblasts, which in turn reduces β-cell insulin secretion109. Interestingly, one study found that an inhibitory effect of leptin on glucose stimulated insulin secretion was observed in vagotomized, but not intact rats, and was abolished by sympathectomy117. Thus, although studies indicate that the SNS partly mediates the inhibitory action of leptin on β-cells, this might be counterbalanced by the parasympathetic nervous system (PNS).
Collectively, these findings indicate that both direct and centrally-mediated leptin action inhibit β-cell insulin secretion. Insulin secretion from pancreatic β-cells promotes lipid storage and leptin synthesis in adipocytes, creating a bidirectional regulatory loop between β-cells and adipocytes, previously termed the adipoinsular axis83,118. The in vivo studies carried out by our group and by others show that disruption of this feedback loop results in hyperinsulinemia, which can lead to perturbations in glucose homeostasis. Thus, the adipoinsular axis might physiologically act to protect normal glucose homeostasis from environmental triggers that promote hyperinsulinemia.
Leptin Action on Pancreatic α-Cells
Several lines of evidence implicate a role of leptin in inhibiting glucagon secretion from pancreatic α-cells. Circulating glucagon levels are elevated in leptin-deficient ob/ob mice119,120, and this is corrected by leptin administration120. Furthermore, leptin administration or gene therapy reverses the hyperglucagonemia present in animal models of type 1 diabetes49,55,58,73. Recent evidence indicates that leptin can suppress glucagon secretion by directly acting on pancreatic α-cells. The expression of Lepr transcript has been shown in an α-cell line, and Lepr-b immunoreactivity was shown in mouse and human α-cells121. Application of leptin to isolated mouse α-cells induces phosphorylation and nuclear translocation of STAT3122. Stimulation of α-tumor cell 1 (TC1) cells and mouse α-cells under low-glucose conditions in the presence of leptin hyperpolarizes membrane cell potential, leading to decreased electrical activity121. Furthermore, leptin suppresses calcium oscillations in mouse and human α-cells from intact islets, and decreases glucagon secretion from mouse islets121. Administration of leptin in vivo and in vitro reduces islet preproglucagon mRNA and intracellular glucagon content of mouse islets122. Leptin did not suppress glucagon secretion in the presence of a phosphoinositide-3-kinase (PI3K) inhibitor, or in db/db islets, indicating a Lepr-b mediated PI3K dependent mechanism. As insulin can inhibit glucagon secretion, it can be postulated that the suppressive effect of leptin on glucagon secretion could be indirectly mediated through changes in insulin secretion; however, this is unlikely due to the inhibitory effect that leptin has on β-cell insulin secretion.
In addition to a direct suppressive action of leptin on α-cell glucagon secretion, ICV leptin administration also reverses hyperglucagonemia in STZ-treated rodents50,57, and suppresses glucagon content and preproglucagon transcript levels in pancreata of STZ-treated mice57. In contrast, leptin administration has also been shown to enhance hypoglycemia-induced glucagon secretion in rats through activation of the sympathetic nervous system117. This effect was not observed when leptin was perfused in the rat pancreas, supporting an indirect mode of action117. Therefore, central leptin action might have differential effects on glucagon secretion when under different metabolic stressors, such as hypoglycemia.
Taken together, the evidence indicates that leptin tonicly inhibits glucagon synthesis and secretion from α-cells, through direct leptin signaling in α-cells (Figure 1), and through leptin-responsive hypothalamic relays (Figure 2). However, whether suppression of glucagon levels alone can account for the antidiabetic action of leptin is unclear. One study giving a low dose of leptin that reversed hyperglucagonemia in STZ-treated rats did not substantially improve glycemia73, suggesting that glucagon suppression is not sufficient for the glucose-lowering effect of leptin therapy in insulin deficient rodents, but this warrants further investigation.
Leptin Action on Hepatocytes
In ob/ob mice, leptin administration profoundly alters hepatic gene expression123. Short-term leptin administration in normal rodents does not alter basal hepatic glucose production124, but alters hepatic glucose fluxes under hyperinsulinemic conditions124,125. Although the majority of studies show that leptin administration in rodents enhances insulin mediated suppression of hepatic glucose production38,48,124,126,127, the effects of leptin on hepatic glucose flux pathways are conflicting. Leptin has been found to promote58,124,125,128 and decrease129 hepatic glycogen storage, as well as promote124–126 and suppress50 hepatic gluconeogenesis. One study reported that leptin administration in ob/ob mice increased hepatic glucose production and increased glucose-6-phosphatase (G6Pase) activity while simultaneously inhibiting phosphoenolpyruvate carboxykinase (PEPCK) activity127. Thus, the effects of leptin on hepatic glucose metabolism are complex and likely dependent on the current metabolic state.
Lepr-b expression, has been shown in hepatic cell lines130,131, isolated hepatocytes from rats and pigs132, and in mouse liver18,55,133. Furthermore, we and others have demonstrated specific leptin binding or functional leptin signaling in hepatocytes45, hepatocyte cell lines45,130,131,134, and rat and mouse liver19,135,136. Application of leptin to hepatocytes often has simultaneously opposing effects on insulin signaling pathways131,134,135. As an example, in one study, perfusion of leptin in isolated rat livers enhanced insulin stimulated insulin receptor (IR) phosphorylation, and promoted insulin receptor substrate-2 (IRS-2) binding to PI3K, while simultaneously inhibiting that of insulin receptor substrate-1 (IRS-1)135. In contrast, another study showed that application of leptin to rat hepatoma cells enhanced insulin stimulated IRS-1 association with PI3K, while inhibiting that of IRS-2131. Furthermore, we found administration of leptin to a hepatic cell line, and to ob/ob mice in vivo enhanced hepatic insulin stimulated IR phosphorylation and paradoxically increased protein tyrosine phosphatase-1B (PTP1B) expression, a negative regulator of both insulin and leptin action45,137.
Alone, leptin has been reported to promote glycogen storage in hepatocytes by inhibiting glycogen phosphorylase and glycogen synthase kinase 3 (GSK3) in perfused rat liver and hepatic cell lines131,135,138, and to inhibit glucose production in response to gluconeogenic precursors in the perfused rat liver and isolated hepatocytes135,139. Leptin can also inhibit glucagon action in primary rat hepatocytes140 and the perfused rat liver139, an effect possibly mediated through activation of PI3K and PDE3B140. Collectively, these studies show that leptin directly modulates insulin signaling and glucose flux, but the effects of leptin are highly variable. Perhaps contributing to these seemingly contradictory findings is that the effect of leptin on hepatocytes appears to be dependent on duration of leptin pretreatment131, species132 and nutritional status141,142. Indeed, one study found that perfusion of livers during the postprandial state inhibited epinephrine-stimulated glucose production, whereas perfusion of livers during the postabsorptive phase stimulated glucose release142.
To examine the physiological role of hepatic leptin signaling, Cohen et al.110 used a Cre-lox approach to knock out total Lepr expression in livers of mice; they found no discernible differences in body weight or glucose metabolism in non-fasted conditions. We subsequently used a similar Cre-lox approach to generate mice with a liver-specific disruption of the signaling domain of Lepr-b, and examined whether these mice showed perturbations in glucose metabolism under varied metabolic conditions133. Although these mice had normal glucose homeostasis under basal, fasted conditions, during an oral glucose tolerance test these mice were protected from glucose intolerance induced by aging or high-fat feeding. Furthermore, during a hyperinsulinemic-euglycemic clamp, these mice showed enhanced hepatic insulin sensitivity when compared to wild-type littermate controls133. Thus, while in vitro effects of leptin on hepatocytes are highly variable, our in vivo evidence suggests that under hyperinsulinemic conditions, leptin has a direct antagonizing effect on hepatic insulin sensitivity (Figure 1). Interestingly, in insulin deficient STZ-treated mice, the loss of hepatic leptin signaling has no effect on the glucose lowering ability of leptin therapy55.
In normal rats, ICV leptin administration mimics the acute effects of peripheral leptin administration on hepatic glucose flux during hyperinsulinemia124,125, showing that leptin action on the liver can also be mediated through central mechanisms. ICV leptin administration enhances hepatic insulin sensitivity in ob/ob mice, and normal and diet-induced obese rats125,143,144. Similarly, reconstitution of Lepr-b in the hypothalamus of fa/fa rats enhances insulin-mediated suppression of hepatic glucose production145. Central leptin administration or gene delivery also influences hepatic expression of genes controlling glucose flux50,51,57,59,125. ICV leptin administration in normal rats inhibits glycogenolysis, but stimulates gluconeogenesis, the net effect resulting in enhanced insulin suppression of hepatic glucose production125. In agreement with this, the effect of ICV leptin on hepatic gluconeogenesis was shown to occur through a melanocortin dependent pathway, whereas the effect of leptin on glycogenolysis was unaffected by melanocortin blockade146. Furthermore, the stimulatory effect of leptin on gluconeogenesis in normal rodents is consistent with increased SNS tone, and antiglycolytic action of leptin on liver metabolism is consistent with increased PNS tone. In support of this, the effect of ICV leptin on hepatic glucose fluxes in normal rats was mimicked by pharmacological stimulation of the SNS125, whereas reconstitution of Lepr-b in the hypothalamus of fa/fa rats increased hepatic insulin sensitivity, in a manner dependent on hepatic vagal innervation145. Similarly, the ability of leptin to improve glucose tolerance was mildly attenuated by hepatic vagotomy in mice with a muscle-specific overexpression of a dominant negative insulin-like growth factor-1 (IGF-1) receptor (MKR mice), which is a model of type 2 diabetes147.
Collectively, these studies show that the physiological effects of leptin on the liver are complex. It appears that under hyperinsulinemic conditions, direct leptin action on the liver antagonizes hepatic insulin signaling (Figure 1). However, indirect hepatic actions of leptin through autonomic efferents from the brain enhance insulin-mediated suppression of glucose production, but can have differential effects on glycogenolysis and gluconeogenesis (Figure 2). Thus, the net direct and indirect effects of leptin on the liver in vivo might depend on factors, such as the current metabolic state and the level of SNS or PNS stimulation.
Leptin Action on Skeletal Muscle
In skeletal muscle, a major contributor to insulin-stimulated glucose disposal, the effect of leptin is unclear. Although some studies have found no change55,124,127,148, or even decreased glucose uptake after acute leptin administration in vivo149, most studies show that leptin stimulates glucose uptake and insulin sensitivity in skeletal muscle129,149–153. Several studies have reported Lepr-b expression in skeletal muscle154,155, yet the direct action of leptin on skeletal muscle is also controversial (Figure 1). Some studies indicate that application of leptin to muscle cell lines156,157 and isolated soleus muscle149,151,158 stimulates glucose uptake and glucose utilization either alone or in the presence of insulin; however, others have reported inhibition of insulin-stimulated glucose metabolism157,159–161, or no effect on muscle glucose uptake162,163. In soleus muscle isolated from ob/ob mice, leptin inhibited insulin stimulated glycogen synthesis160,161, but had no effect on glycogen synthesis in soleus isolated from wild-type mice162. Thus, the direct action on skeletal muscle appears dependent on metabolic status. Further adding to the complexity of leptin action, in a skeletal muscle cell line, the response to leptin was dependent on the duration of leptin exposure157. There is a substantial degree of insulin and leptin cross-talk in skeletal muscle. Leptin exposure has been shown to induce IRS-2 phosphorylation and PI3K activation156,164, and incubation with leptin alone has been shown to stimulate glucose uptake through a PI3K dependent mechanism157 in muscle cell lines. Acute leptin administration in vivo was shown to directly stimulate skeletal muscle adenosine monophosphate-activated protein kinase (AMPK), which alters lipid partitioning by promoting fatty acid oxidation148. Changes in lipid partitioning are also observed after leptin application to muscle cell lines162,165. Although acute leptin stimulation of AMPK did not alter glucose uptake148, chronic hyperleptinemia decreases lipid accumulation in muscle in vivo, and simultaneously enhances muscle insulin sensitivity, an effect mimicked by pharmacological stimulation of AMPK166–168. Thus, chronic exposure of skeletal muscle to leptin might have long-term effects on insulin sensitivity.
Both ICV and intravenous short-term leptin administration stimulate glucose uptake in skeletal muscle in normal mice129, indicating that leptin action on skeletal muscle can also be indirectly mediated through central pathways (Figure 2). Leptin injection in the hypothalamus acutely stimulates glucose uptake, utilization and fatty acid oxidation in skeletal muscle148,150. Interestingly, this occurred when leptin was injected directly into the ventromedial hypothalamus150,169, and was blocked by a melanocortin receptor antagonist169. Furthermore, injection of leptin into the lateral hypothalamus alters the expression and activity of metabolic enzymes in skeletal muscle, including Akt and AMPK170. The effect of leptin on skeletal muscle is diminished by denervation129,148 or sympathetic blockade148,170. When leptin is administered chronically through an ICV route in insulin deficient rodents, glucose utilization is increased, as well as the expression of Glut4 in skeletal muscle50,57,171. However, central leptin administration or gene therapy has also been reported to have no effect59 or to decrease Glut4 expression51. In one study, ICV leptin increased Glut4 expression in white gastrocnemius, but not soleus muscle57, whereas the acute stimulation of glucose uptake by hypothalamic leptin injection was more prominent in the soleus than the extensor digitorum longus150, indicating that leptin might differentially regulate glucose metabolism in oxidative and glycolytic muscle fibres. Collectively, these studies suggest that leptin enhances insulin sensitivity and glucose uptake in skeletal muscle through central relays, whereas direct leptin action on skeletal muscle might enhance or oppose insulin action depending on metabolic state.
Leptin Action on Adipocytes
Unlike the hepatic and muscular actions of leptin, leptin administration is well known to inhibit insulin action in white adipose tissue (WAT). Paradoxically, both prolonged and acute administration of leptin in vivo have been shown to stimulate glucose uptake in brown adipose tissue (BAT), but not WAT127,129,152,172. The effect of leptin on WAT and BAT are thereby likely to partially mediate the effect of leptin on glucose homeostasis.
Expression of Lepr-b has been shown in BAT and WAT from mice173,174, and primary BAT and cultured brown adipocytes from rats175. Furthermore, leptin-induced STAT phosphorylation and translocation has been shown in brown and white adipocytes, the effect of which is absent in fa/fa brown adipocytes and attenuated in white adipocytes after Lepr-b knockdown by antisense RNA175,176. Isolated white adipocytes incubated with leptin become desensitized over several hours to insulin-induced glucose uptake and glycogen synthesis177. This desensitizing effect of leptin is dose- and time-dependent, and reversible after removal of leptin. A similar inhibition of insulin-stimulated glucose uptake was reported in a brown adipocyte cell line178. There is significant cross-talk between leptin and insulin signaling in adipocytes; leptin inhibits insulin stimulated phosphorylation of IR and GSK3, and binding of insulin to its receptor in isolated white adipocytes179,180, and reduces insulin-stimulated IR kinase activity, and IRS-1 phosphorylation in a brown adipocyte cell line178. Of note, leptin application also reduces leptin gene expression in adipocytes, possibly through suppression of the stimulatory effect of insulin on leptin synthesis181. One in vivo study found that mice with a knockdown of adipocyte leptin receptors had increased body weight and adiposity, indicative of enhanced adipogenic action of insulin in adipose tissue. In addition these mice had impaired glucose tolerance and insulin sensitivity176 suggesting that the direct inhibitory effect of leptin on insulin action in WAT is important for normal glucose homeostasis.
The inhibitory effect of leptin on insulin signaling is also observed in WAT explanted from rats treated with ICV leptin for 7 days, indicating that central leptin action induces prolonged changes to white adipocytes to inhibit insulin signaling179. In contrast to WAT, central leptin administration or gene therapy can acutely stimulate glucose uptake in BAT50,127,129,150, and increase expression of Glut4 and uncoupling proteins (UCP)-1 and -351,59. The effect of leptin on BAT glucose metabolism is abolished by denervation150, and mediated by the SNS182,183. Interestingly, in ob/ob mice, ICV leptin acutely stimulated whole body glucose turnover, and this correlated with enhanced glucose uptake in BAT, but not skeletal muscle or WAT, indicating that leptin might acutely stimulate whole-body glucose metabolism through central mediated effects on BAT127. Collectively, these studies show that leptin inhibits insulin signaling in adipocytes primarily through direct action (Figure 1), whereas central leptin-responsive relays mediate leptin action on glucose uptake in BAT (Figure 2).
Leptin Action on Counter Regulatory and Insulin Mimetic Hormones
In addition to direct and centrally-mediated effects of leptin on the endocrine pancreas to influence insulin and glucagon levels, leptin might also indirectly regulate glucose metabolism by altering levels of other hormones that regulate glucose metabolism. Leptin is well known to inhibit synthesis and secretion of corticosterone73,184–187, which could thereby increase insulin sensitivity. Furthermore, a role of leptin in GH secretion has been identified. Although studies indicate that leptin enhances GH secretion188–191, we found that in STZ-treated mice, leptin therapy robustly suppressed circulating GH levels, which might enhance peripheral insulin sensitivity55. Leptin might also improve glycemia indirectly by increasing levels of the glucose-lowering hormone, GLP-1192. In addition, leptin administration in STZ-treated rats was found to elevate circulating IGF1, which in turn could have an insulin mimetic effect on peripheral tissues49. Finally, leptin therapy has also been found to increase circulating IGF binding protein-2 (IGFBP2) levels in ob/ob mice47, which we have since confirmed193. Interestingly, adenoviral-mediated overexpression of IGFBP2 to ob/ob mice mimics the effect of leptin administration, by reducing glucose and insulin levels, indicating that some of the metabolic action of leptin might be mediated through IGFBP247. As a whole, it appears that most effects of leptin on altering circulating hormone levels lead to an overall increase in insulin sensitivity.
The Therapeutic Potential of Leptin
As a result of the profound ability of leptin to modulate glucose homeostasis, leptin holds therapeutic potential for the treatment of metabolic disorders. Clinical trials showed that leptin had modest weight-reducing effects in some obese individuals194, and leptin therapy was recently reported to have minimal metabolic benefit in obese, type 2 diabetic patients195. It appears as though obesity is typically associated with elevated leptin levels and leptin resistance, thereby limiting the effectiveness of exogenous leptin. However, leptin therapy might be useful in combination with other means of weight loss, such as diet and exercise, to help reduce regaining weight196–201. In contrast, leptin therapy in humans with rare congenital leptin deficiency markedly reduces body weight and adiposity, and ameliorates the metabolic dysfunction of these patients7,11–14. Leptin also proved to ameliorate hyperinsulinemia, insulin resistance and hyperglycemia in another subset of leptin-deficient individuals, namely patients with general or partial lipodystrophy39,202. Thus, as previously speculated, leptin administration seems to be most effective in conditions with abnormally low circulating leptin levels203.
Another metabolic disorder that is accompanied by inappropriately low leptin levels is type 1 diabetes71, and a rapidly increasing number of studies support the fact that leptin has therapeutic properties in rodent models of insulin deficiency48–55,57–62. Although insulin therapy restores circulating leptin levels in type 1 diabetic patients71, studies in rodents show that higher doses of leptin might have additional therapeutic effects on glycemia. Leptin is now being tested clinically as an adjunct to insulin therapy for type 1 diabetes. Interestingly, occasional leptin injections in NOD mice receiving continuous insulin infusion were found to improve glycemia, compared with insulin alone, even when leptin injections were combined with a lower insulin dose58. This indicates that the addition of leptin to insulin therapy regimens for patients with type 1 diabetes might allow for tighter glycemic control, with less-frequent insulin dosing. However, with the potential benefit of combined leptin and insulin cotherapy, is a potential danger of hypoglycemia. Our work has shown that as a result of enhanced insulin sensitivity, a dose of insulin that was only modestly effective in lowering blood glucose in STZ-treated mice was nearly lethal when combined with leptin therapy55. Furthermore, given emerging evidence of the inhibitory effect of leptin on α-cell glucagon secretion121 and circulating glucagon levels49,50,55,57,58,73, leptin might interfere with counter-regulatory responses to hypoglycemia, a mechanism that is already impaired in patients with long standing type 1 diabetes. Thus, caution should be used when testing leptin with insulin in patients with type 1 diabetes. Also of potential concern are the immune-related actions of leptin204. NOD mice with a loss of function mutation in the Lepr gene have reduced incidence of diabetes205,206, whereas leptin administration has been reported to accelerate diabetes onset in NOD mice207. Nevertheless, the potent glucose-lowering effect of leptin in insulin deficient rodents justifies the careful assessment of leptin therapy in humans with diabetes.
Approximately two decades since its discovery, the so-called anti-obesity hormone, leptin, has now been established as a key regulator of glucose homeostasis, both in rodents and humans. Both leptin deficiency and leptin resistance have a profound impact on metabolism in rodents and humans. Leptin replacement can dramatically improve metabolism in cases of leptin deficiency, and now even shows promise as a therapy for insulin deficient type 1 diabetes. Leptin has multiple, complex actions on insulin-sensitive tissues and the hormones of the endocrine pancreas, all of which likely contribute to glucose homeostasis. Further investigation of the mechanisms of leptin action on metabolism is warranted, both to enhance our understanding of metabolic regulation, and to fully exploit the therapeutic potential of leptin.
H. C. D. is the recipient of a Canada Graduate Scholarship from the Natural Sciences and Engineering Research Council of Canada. F. K. H. is funded by the Canadian Diabetes Association. T. J. K. is the recipient of a senior scholarship from the Michael Smith Foundation of Health Research. Our work in this area is supported by the Canadian Institutes of Health Research. There are no financial conflicts to declare.