|Home | About | Journals | Submit | Contact Us | Français|
Over the past century, prevalent models of energy and glucose homeostasis have been developed from a better understanding of the neural circuits underlying obesity and diabetes. From the early hypothalamic lesion reports to the more recent pharmacological and molecular/genetic studies, the hypothalamic melanocortin system has been shown to play a critical role in the regulation of metabolism. This review attempts to highlight contributions to our current understanding of how numerous neuromodulators (leptin, insulin, and serotonin) integrate with the central melanocortin system to coordinate alterations in energy and glucose balance.
Obesity has reached epidemic proportions in the United States, with the majority of the population now overweight and greater than a quarter of the population classified as obese (Flegal et al., 2010). Alarmingly this is a trend which is not limited to the United States; rather obesity and its co-morbidities such as type II diabetes mellitus are on the rise and pose a serious threat to public health around the world (Shaw et al., 2010; Wild et al., 2004; Zimmet et al., 2001). To understand the causes and to develop treatments for obesity and type II diabetes mellitus, it is first necessary to unravel how numerous neuropeptides, neurotransmitters, receptors, and the central intracellular signaling pathways regulate coordinated energy and glucose homeostasis. Several studies have determined that the hypothalamus is a key component in the regulation of metabolic homeostasis, integrating information regarding the body’s internal environment and orchestrating a series of coordinated endocrine, autonomic and behavioral responses that maintain metabolic homeostasis. This review outlines the current understanding of leptin, insulin, and serotonin action, key regulators of glucose and energy homeostasis, largely within the hypothalamic melanocortin system.
The first clinical description of hypothalamic-pituitary injury resulting in obesity was reported in 1840 (Mohr, 1840). Bernhard Mohr described his patient, a 57 year old woman who experienced rapid weight gain, became obese within one year, and suffered from multiple neurological deficits , reviewed here (Brobeck, 1946). After a few years with her deteriorating condition she died and although post-mortem analysis revealed tumor-like degeneration of the pituitary body likely producing increased intracranial pressure upon adjacent parts of the brain little else was known about the causes of her condition. In 1900, Joseph Babinski noted a condition characterized by feminine obesity and sexual infantilism resulting from a tumor of the pituitary (Babinski, 1900). A year later, Alfred Fröhlich described a rare childhood metabolic disorder characterized by obesity, growth retardation, and retarded development of the genital organs also resultant of a pituitary tumor (Fröhlich, 1901). Since the reports of the Babinski- Frohlich’s syndrome, debate as to whether this disorder was due to pituitary insufficiency or hypothalamic damage ensued (Erdheim, 1904). Forty years later, A.W Hetherington and Stephen Ranson performed a series of electrolytic lesions in the ventral medial hypothalamus of rats which replicated the obesity and hyperphagia previously observed in the Babinski- Frohlich’s syndrome (Hetherington and Ranson, 1940). Their work showed for the first time that the ventral medial hypothalamus, independent of any pituitary insufficiency, is required for proper energy homeostasis (Hetherington and Ranson, 1942). Collectively, these data highlight the necessity to understand the hypothalamic circuitry involved in the regulation of energy and glucose homeostasis.
Claude Bernard first suggested the idea of homeostasis nearly 150 years ago when he introduced the concept of milieu intérieur. Numerous examples of homeostatic mechanisms exist in mammals, such as those that regulate energy and glucose homeostasis, largely relying on a feedback loop of a signal/hormone which regulates activities such as food intake, metabolic rate, glucose uptake/disposal or production. Gordon Kennedy is credited with establishing one homeostatic paradigm known as the “lipostatic hypothesis” - which states a signal released into the circulation in proportion to body adipose stores acts to regulate feeding and overall energy balance (Kennedy, 1953). A seminal study which extended the “lipostatic hypothesis” and helped to delineate a possible circulating factor involved in the regulation of energy balance was performed by Doug Coleman while working at Jackson Laboratories (Coleman, 1973; Coleman and Hummel, 1969). He studied mice in which two independent spontaneous autosomal recessive mutations occurred and resulted in obesity and diabetes; reviewed here (Williams et al., 2009). Briefly, Coleman’s work identified a circulating factor which was absent in ob/ob mice that regulated energy and glucose homeostasis. Conversely, this circulating factor was made in excess in db/db mice yet these mice were resistant either because they lacked the receptor or the signaling cascade activated by this circulating factor was impaired. It wasn’t until another 40 years had passed before the adipose-derived peptide leptin was first described and identified as the circulating factor in Coleman’s studies which was required for proper energy and glucose balance (Zhang et al., 1994). Today, it is generally accepted that peripheral signals/hormones integrate with other factors such as the presence of food or social behavior to regulate food intake (Schwartz and Porte, 2005; Woods et al., 1998).
Shortly after leptin was discovered, leptin’s cognate receptor was cloned (Tartaglia et al., 1995). The leptin receptor is a member of the type I cytokine receptor family and shares homology to glycoprotein 130, a transmembrane protein which forms a subunit of the interleukin-6 family of cytokines. Leptin has 6 identified receptors (LepRa-f), of which LepRb is defined as the longest isoform of the leptin receptor, containing the intracellular tyrosine residues necessary for activation of multiple intracellular signaling cascades (Chen et al., 1996; Lee et al., 1996; Myers et al., 2008; Takaya et al., 1996; Tartaglia et al., 1995). Upon ligand binding the leptin receptor (LepRb) dimerizes and triggers the activation of Janus activated kinases (JAKs) which phosphorylate the tyrosine residues on the cytoplasmic domain of the leptin receptor and thus initiating the multiple signaling cascades activated by leptin (Fig. 1). Peripheral leptin administration results in decreased food intake and body weight (Halaas et al., 1995). Interestingly intracerebroventricular leptin administration alone replicates the decrease in food intake and body weight observed after peripheral administration (Campfield et al., 1995; Halaas et al., 1997). Moreover, selective deletion of leptin receptors within the central nervous system (CNS) of mice recapitulates the db/db phenotype (Cohen et al., 2001). Together these data suggest that the CNS is a primary site of action for leptin to mediate effects of energy and glucose homeostasis. These data also led to the prevalent model of energy homeostasis in which leptin is not only an adipostatic secreted factor, but rather a metabolic cue that communicates directly to the brain the status of available energy stores (Schwartz and Porte, 2005). Importantly, leptin monotherapy was shown to correct a genetic deficiency of leptin in a small population of humans, supportive of a shared physiology from rodents to humans and suggesting that understanding leptin regulation of energy and glucose balance in rodents also advances an understanding of circuits involved in energy and glucose homeostasis in human physiology (Farooqi et al., 1999; Farooqi et al., 2002).
In 1869, Paul Langerhans was first to identify structures of unknown function throughout the pancreas resembling islands floating among acinar cells and later termed islets of langerhans. Initially it was thought that the pancreas was solely critical for proper digestion, however at the turn of the century, several researchers were making critical observations which clearly established a link between the islets and diabetes mellitus. Joseph von Mering and Oskar Minkowski made the first major discovery in 1889 when they reported severe and fatal diabetes resulting from the complete removal of the pancreas from dogs (von Mering and Minkowski, 1889). Eugene Opie subsequently reported hyalinosis of the islets of langerhans in patients suffering from diabetes mellitus suggesting degeneration of the islets contributed to the pathophysiology of the disease (Opie, 1900–1901a; b). Several attempts were made to identify what the islets were producing in an effort to treat diabetes; however it was not until the early 1920s when Frederick Banting and medical student Charles Best developed a method to isolate a viable extract from the pancreas of dogs while working in the lab space of John James Richard Macleod at the University of Toronto – reviewed here (Banting et al., 1922). The extract was found to contain isletin (later termed insulin) and injection of the extract into pancreatectomized dogs corrected diabetes (Banting and Best, 1922). In January 1922, after improvements in the quality and method of extraction, a Toronto teenager (L.T.) was the first successfully injected with Banting and Best’s extract and almost completely corrected the signs of diabetes. Interestingly, it is estimated that the islets of langerhans constitute less than 3% of the pancreas, yet their partial or entire destruction results in diabetes mellitus (Elayat et al., 1995; Orci et al., 1976; Overholser, 1925). Later research showed that the beta cells (β-cells) within pancreatic islets of langerhans play a critical role in mammalian physiology by synthesizing and secreting insulin.
The insulin receptor is a transmembrane tyrosine kinase receptor retaining endogenous enzyme activity to transfer phosphate groups from ATP to tyrosine residues on the intracellular target proteins (Feener et al., 1993; Lee and Pilch, 1994; Ullrich et al., 1985). The insulin receptor contains two alpha and two beta subunits linked by disulfide bonds (Ullrich et al., 1985). The beta subunits span the plasma membrane and constitute the intracellular domain of the receptor, while the alpha subunits are entirely extracellular and retain the ligand binding site (Fig. 1). Insulin binding to the extracellular alpha subunits causes the beta subunits to autophosphorylate, thus activating various intracellular proteins/kinases which regulate biological activity (Lee and Pilch, 1994; Ullrich et al., 1985).
Diabetes mellitus, resulting from impaired insulin signaling, leads to several complications including coronary artery disease, retinopathy, stroke, kidney failure, and ultimately death. However, it is clear that proper maintenance of blood glucose levels can ameliorate these conditions. Fortunately, multiple therapies in combination with insulin exist today to treat diabetes and the metabolic perturbations that accompany this disease. Although less is known about the role of insulin in the brain, growing evidence suggests that insulin signaling in the brain is equally important as signaling in the periphery for the maintenance of glucose and energy homeostasis (Hill et al., 2010; Konner et al., 2007; Woods et al., 1979). For instance, similar to leptin receptors, insulin receptors are widely expressed in the CNS and maybe critical for neuronal survival (Havrankova et al., 1978a; Havrankova et al., 1978b). Moreover, Woods, Porte, and colleagues showed that intracerebroventricular infusion of insulin reduced food intake and body weight (Woods et al., 1979). Mice lacking insulin receptors specifically in the CNS also displayed increased food intake, diet-induced obesity and mild insulin resistance (Bruning et al., 2000). Thus, increasing evidence suggests that in order to continue providing better therapeutics for the treatment of diabetes, we must fully understand the effects of insulin not only in the periphery but also in the brain. Given that leptin and insulin are both peripheral hormones which mediate effects on glucose and energy balance through circuits in the brain, the Elmquist laboratory and others has been focused on understanding the circuits which are required and/or sufficient for insulin and leptin action within the CNS.
The hypothalamic arcuate nucleus includes proopiomelanocortin (POMC) and neuropeptide-Y/agouti-related peptide (NPY/AgRP) neurons. POMC is a precursor polypeptide which undergoes post-translational processing to produce several biologically active peptides called melanocortins (γ-melanocyte-stimulating hormone (MSH), adrenocorticotropic hormone (ACTH), β-lipotropin, α-MSH, corticotropin-like intermediate lobe peptide (CLIP), γ-lipotropin, β-endorphin, and β-MSH) (Chretien et al., 1979; Cochet et al., 1982; Nakanishi et al., 1979; Uhler and Herbert, 1983). Roger Cone and colleagues first cloned the melanocortin receptors in the 1990s (Adan and Gispen, 1997; Chhajlani et al., 1993; Chhajlani and Wikberg, 1992; Gantz et al., 1993a; Gantz et al., 1993b; Mountjoy et al., 1994; Mountjoy et al., 1992). There are 5 cognate melanocortin receptors (MC1–5 receptor) which are 7 transmembrane G-protein coupled receptors of the rhodopsin family. Several of the receptors have biologically active functions in the periphery such as regulating mammalian skin/hair color (MC1 receptor) (Jackson, 1993; Robbins et al., 1993), stimulation of glucocorticoid production (MC2 receptor) (Clark et al., 1993; Tsigos et al., 1993), and regulating the production of sebum in the exocrine glands (MC5 receptor) (Griffon et al., 1994). However, MC3 and MC4 receptors are the only melanocortin receptors both highly expressed within the brain and involved in the regulation of energy balance. For instance, mice deficient for MC3 receptors are characterized by modest increased weight gain and adiposity on a low-fat diet (Butler et al., 2000; Sutton et al., 2006). MC4 receptor null mice on a low-fat diet show striking hyperphagia, obesity, hyperinsulinemia, and increased lean body and fat mass (Butler et al., 2001; Sutton et al., 2006). Exposure of MC4 receptor null mice to a high-fat diet exacerbates the obesity and diabetes observed on a low-fat diet. Surprisingly, MC3 receptor null mice on a high-fat diet share similar characteristics to high-fat diet fed MC4 receptor null mice suggesting that MC3 receptor null mice may have an important role in regulation peripheral metabolism and energy balance (Sutton et al., 2006). Interestingly, re-expression of MC4 receptor in neurons of the paraventricular hypothalamus, a target of arcuate POMC and NPY/AgRP neurons, and a subpopulation of amygdala neurons, completely rescued the increased food intake and the majority (60%) of the obesity observed in MC4 receptor null mice (Balthasar et al., 2005). These data highlight the importance of melanocortin signaling within the paraventricular hypothalamus as well as support a model which includes a distributed network of melanocortin neurons that regulate energy homeostasis. It should be noted that humans with mutations in the MC4 receptor exhibit similar characteristics to MC4 receptor null mice suggesting a critical role of MC4 receptor in the regulation of energy balance in human physiology (Hinney et al., 1999; Vaisse et al., 1998; Yeo et al., 1998). Moreover, the work of O’Rahilly and colleagues showed that mutations in the MC4 receptor gene are now recognized as the most common monogenic form of human obesity described to date (Barsh et al., 2000; Farooqi et al., 2003). Thus these data define a key role of melanocortin signaling in the regulation of energy and glucose homeostasis.
One of the melanocortins, α-MSH - an MC3/4 receptor agonist, and its analogs potently inhibits food intake in the brain via activation of central melanocortin neurons (Fan et al., 1997; Ludwig et al., 1998; Rossi et al., 1998; Tsujii and Bray, 1989). Interestingly, an increase in serum leptin results in both an increase in POMC gene expression and the resulting peptide α-MSH (Mizuno et al., 1998; Schwartz et al., 1997; Thornton et al., 1997). Moreover, acute administration of leptin results in increased markers of cellular activation in POMC neurons (Elias et al., 1999). Furthermore, leptin administration in an acute slice preparation results in a leptin-induced depolarization of arcuate POMC neurons concomitant with activation of a putative mixed-cation conductance (Cowley et al., 2001; Hill et al., 2008; Qiu et al., 2010; Williams et al., 2010). A recent study extended these observations of the acute activation of POMC neurons by leptin suggesting the involvement of a TrpC channel in the leptin-induced depolarization (Qiu et al., 2010). However the subunits comprising the TrpC channel are currently unknown and future studies will undoubtedly delineate the properties of this leptin-activated mixed-cation conductance in arcuate POMC cells. Adjacent to arcuate melanocortin neurons are the NPY/AgRP cells. NPY has been associated with a number of physiological responses in the brain however a primary effect of NPY appears to be a reduction in food intake (Stanley and Leibowitz, 1985). In addition, AgRP colocalizes with NPY in arcuate neurons and functions as an inverse agonist to the melanocortin receptor (Chai et al., 2003; Hahn et al., 1998; Nijenhuis et al., 2001; Tota et al., 1999). Opposite to the actions of leptin in POMC neurons, leptin administration has been shown to inhibit gene transcription of NPY/AgRP in the hypothalamus (Schwartz et al., 1996a; Schwartz et al., 1996b; Stephens et al., 1995). Leptin application also results in a decrease in GABAergic inhibitory post synaptic currents (IPSCs) onto POMC neurons resulting in a disinhibition of POMC cellular activity (Cowley et al., 2001). Due to previous observations which showed an inhibition of mediobasal hypothalamic neurons by leptin and a colocalization of GABA and NPY in synaptic boutons which synapse onto POMC neurons, it has largely been speculated that leptin hyperpolarizes and suppresses the cellular activity of NPY/AgRP neurons via a PI3K dependent activation of an ATP-sensitive potassium conductance (Cowley et al., 2001; Spanswick et al., 1997). Moreover, leptin application to rat NPY neurons, identified via single-cell PCR, resulted in a hyperpolarization which mirrored the previous reports on unidentified mediobasal hypothalamic neurons providing evidence for a leptin-induced inhibition of NPY/AgRP neurons (van den Top et al., 2004). Together, these data form the basic model of leptin regulation in NPY/AgRP and POMC neurons of the hypothalamic arcuate nucleus (Cone et al., 2001; Cowley et al., 2001). Also, these data are an indicator that the next era of obesity research has begun with the use of mouse genetics to test models of energy and glucose homeostasis.
Identification and cloning of leptin, leptin receptors, and melanocortin receptors were key steps to the start of the “molecular era” of obesity research which incorporates the power of mouse genetics to test varying circuit models utilizing Cre-loxP (cyclization recombination-locus of X over P1) technology to perform genetic manipulation in a cell specific manner. Leptin receptors are widely expressed in the brain, however an enriched expression is observed within the mediobasal hypothalamus including the arcuate POMC neurons (Elias et al., 1999; Elmquist et al., 1998b). In a first attempt to tease apart leptin action on neural circuits in the brain a simple model was proposed. If the melanocortin POMC neurons are the primary site for mediating the effects of leptin to maintain energy homeostasis then deleting leptin receptors in POMC neurons alone should result in altered energy balance. Selective deletion of leptin receptors in POMC neurons resulted in a modest obesity in male and female mice (Balthasar et al., 2004). Interestingly, no effect was observed on food intake; rather the increased body weight was dependent on a decreased energy expenditure and locomotor activity (Fig. 2). Deletion of leptin receptors in POMC neurons also produced no or mild effects on glucose homeostasis (Balthasar et al., 2004; Shi et al., 2008). This was surprising given the fact that leptin deficient mice expressing a transgene of the POMC gene under the control of a neuron-specific promoter exhibited normalized glucose levels and insulin resistance (Mizuno et al., 2003). In subsequent studies consistent with these leptin data, selective deletion of suppressor of cytokine signaling 3 (SOCS3) in POMC cells resulted in enhanced leptin sensitivity and also improved glucose homeostasis on a chow diet while weight gain was reduced and insulin sensitivity and glucose homeostasis was improved on a high energy diet (Kievit et al., 2006). Importantly, another recent study examined the role of leptin receptors in adjacent AgRP neurons (van de Wall et al., 2008). Van de Wall and colleagues found that selective deletion from AgRP neurons also resulted in hyperphagia, decreased locomotor activity, and modest obesity (Fig. 2), supporting the basic model of leptin action postulated from electrophysiological data described in the prior section. These data also suggest another cell population within the arcuate other than NPY/AgRP neurons contributes to the leptin-induced regulation of glucose balance and additional studies were needed in order to determine the role of melanocortin neurons in glucose homeostasis.
The Elmquist, Lowell and Chua laboratories further examined the role of leptin signaling in the arcuate nucleus to regulate glucose balance by generating a mouse homozygous for a FLPe-reactivatable, leptin receptor null allele [Lepr(neo/neo)] (Coppari et al., 2005). The Lepr(neo/neo) mice were similar to the db/db mouse in that they are hyperphagic, obese, hyperglycemic, hyperinsulinemic, infertile, and hypoactive. Unilateral injection of an adeno-associated virus expressing FLPe-recombinase into the arcuate nucleus of Lepr(neo/neo) mice resulted in a modest decrease in food intake and bodyweight. Unilateral reactivation of Leprs in the arcuate nucleus restored deficits in glucose homeostasis such that hyperinsulinemia was improved and blood glucose levels were normalized demonstrating the sufficiency of arcuate leptin signaling in maintaining glucose homeostasis. Similar results were found when leptin receptors were expressed via adenoviral injection into the arcuate nucleus of Koletsky rats (fak/fak) which are deficient for the leptin receptor and recent evidence by the same group suggests a leptin-induced hypothalamic regulation of hepatic insulin sensitivity is dependent upon the hepatic vagal branch (German et al., 2009; Morton et al., 2005; Morton et al., 2003). Christain Bjorbaek’s group recently investigated whether overexpression of LepRb in POMC cells alone was sufficient to regulate energy and glucose homeostasis (Huo et al., 2009). They found that leptin receptor expression in POMC neurons of a db/db mouse resulted in a decreased food intake and energy expenditure concomitant with a modest reduction in body weight. Importantly, overexpression of leptin receptors in POMC neurons alone normalized the high blood glucose levels observed in the db/db mouse. Together these data suggest that direct leptin action in POMC neurons is required for the maintenance of energy homeostasis through energy expenditure and locomotor regulation and is not required but maybe sufficient for the maintenance of proper glucose homeostasis (Fig. 2). Intriguingly, a recent report suggests that leptin alone can normalize blood glucose levels of type I diabetics and keep the blood glucose levels in the normal range for extended periods of time possibly via regulation of the peripheral hormone glucagon (Yu et al., 2008). It remains to be determined if direct leptin action in POMC neurons may underlie the dramatic effect of leptin to ameliorate insulin insensitivity in diabetes mellitus.
The relatively modest effects observed on body weight in mice lacking leptin receptors in POMC neurons suggested that the effects of leptin require a distributed network of neurons. In support of multiple sites of leptin action, selective deletion of leptin receptors in adjacent ventral medial hypothalamic steroidogenic factor-1 (SF-1) neurons resulted in an obesity which mirrored that observed in the POMC-LepR null mouse (Bingham et al., 2008; Dhillon et al., 2006). Moreover, deletion of leptin receptors in POMC and SF1 neurons resulted in an additive effect on body weight. The leptin-induced acute activation of SF1 neurons was similar to the acute effects of leptin observed in POMC neurons such that most SF1 positive neurons were depolarized resulting in an increased firing rate in response to leptin – an effect not observed in SF1-LepR null mice (Fig. 2). However, future studies are needed to assess whether these data support a similar leptin-induced activation of a putative mixed-cation conductance in SF1 cells. Interestingly, SF1-LepR null mice fed a high-fat diet gained considerably more weight than their wildtype counterpart. The gain in body weight of SF1-LepR null mice on a HFD was also greater than predicted by leptin deficiency or HFD alone. Together, these data suggest that direct leptin action within the ventral medial hypothalamus plays an important role in energy homeostasis through modifying energy expenditure and notably protects against HFD-induced obesity possibly by regulation of diet-induced thermogenesis (Fig. 2). It should be noted that Heatherington and Ranson’s electrolytic lesion studies in the 1940s were largely inclusive of the ventral medial hypothalamus. Therefore these data are essentially a modern (21st century) correlate of the Heatherington and Ranson lesion studies and support a requirement of the ventral medial hypothalamus for proper energy balance. These data also highlight growing evidence for a distributed network of leptin responsive neurons within the hypothalamus contributing to the central actions of leptin to regulate energy and glucose homeostasis.
Balthasar and colleagues (Balthasar et al., 2005), reported that expression of MC4 receptors in the PVN and amygdala is responsible for the reduction in food intake mediated by melanocortin signaling and not for the effects on metabolism. Moreover, local and long-term antagonism of melanocortin signaling in the paraventricular hypothalamus, dorsal medial hypothalamus, and lateral hypothalamus, by use of an adeno-associated virus expressing Agouti under the control of the cytomegalovirus promoter, differentially induced hyperphagia and accelerated body weight gain without apparent changes in central neuropeptide systems that are known to stimulate food intake (Kas et al., 2004). Thus, melanocortin action may be seen as being distributed, across numerous CNS nuclei. While a significant amount of research has focused on the hypothalamic actions of both leptin and the melanocortins, projections from hypothalamic melanocortin neurons to the hindbrain have also been implicated in the modulation of energy homeostasis. Similar to the hypothalamus, the melanocortins act in concert with leptin to affect both food intake and metabolic rate, through actions in the nucleus of the solitary tract, area postrema and vagus nerve. Recent work involving the injection of the MC3/4 receptor agonist MTII directly into the NTS at relatively low concentrations resulted in an increased metabolic and cardiac rate with higher doses of MTII producing a reduction in food intake and acute body weight loss (Skibicka and Grill, 2009b). Use of the MC3/4 receptor antagonist SHU, injected into the NTS, has been demonstrated to antagonize the effect of leptin injected into the arcuate nucleus, blocking the leptin dependent reduction in meal size and frequency (Zheng et al.). Interestingly, leptin may also be modulating food intake and metabolic rate through a direct action on neurons of the hindbrain. In particular, hindbrain delivery of leptin results in tachycardia and hyperthermia coupled with a reduction in food intake that is blocked by hindbrain application of the MC3/4 receptor antagonist SHU (Grill et al., 2002; Skibicka and Grill, 2009a). The mechanism of leptin action in the hindbrain, specifically in the NTS, has been suggested to involve a reduction in food intake that results from changes in meal size that results from the potentiation of the vagal gastric distention signal. (Emond et al., 2001; Grill et al., 2002; Huo et al., 2007; Schwartz and Moran, 2002). Recent work by Roger Cone and colleagues suggests CCK activates POMC neurons of the NTS and results in a reduction in food intake, an effect which also requires MC4 receptor activation downstream of POMC neurons in the hindbrain (Fan et al., 2004). Furthermore, leptin potentiates the NTS response to a gastric preload and to the satiation signal CCK, enhancing the food intake reducing actions of these stimuli. In agreement with this observation, neurons of the dorsal vagal complex have been shown to be directly responsive to leptin, resulting in changes in acute neuronal activity (Williams and Smith, 2006; Williams et al., 2007). Finally, knock down of leptin receptor expression recapitulated the results obtained from hinbrain leptin injection, producing hyperphagia for both chow and high fat diet coupled with a decreased sensitivity to CCK resulting in increased weight gain (Hayes et al.). Collectively, these data describing leptin and melanocortin action in the hindbrain, highlight the distributed nature of satiation signaling within the CNS.
Multiple signaling cascades are activated by leptin with each arm contributing to different facets of energy and glucose homeostasis (Fig. 1). There are several reports supportive of a role for one such signaling cascade, phosphoinositide 3-kinase (PI3K), in the effects of leptin and insulin action within the arcuate. For example, Barsh and colleagues utilized a fluorescent reporter for PI3K activity targeted to AgRP or POMC neurons (Wanting Xu et al., 2005). They found that leptin and insulin both triggered membrane accumulation of the PI3K fluorescent reporter which could be blocked by pretreatment with PI3K inhibitors. Additionally, a recent report showed that insulin increased phosphatidylinositol (3,4,5)-triphosphate (PIP3) levels in the arcuate nucleus of mice including POMC neurons – an effect absent in mice that the PIP3 phosphatase, PTEN, was disrupted selectively in POMC cells (Plum et al., 2006). Moreover, a subpopulation of neurons in the mediobasal hypothalamus is hyperpolarized by both leptin and insulin, an effect dependent upon a PI3K induced-activation of an ATP-sensitive potassium channel (Spanswick et al., 1997; Spanswick et al., 2000). Leptin and insulin both increase the phosphorylation of Akt and other PI3K signaling intermediates in the arcuate nucleus (Mirshamsi et al., 2004). Importantly, peripheral or central administration of leptin or insulin also activates IRS-2 associated PI3K in the hypothalamus of rats and PI3K inhibitors block the ability of icv leptin or insulin to reduce food intake (Niswender et al., 2003; Niswender et al., 2001). Together, these data suggest that the weight reducing effects of leptin and insulin depends on the PI3K activity within the hypothalamus and leptin and insulin act in parallel to stimulate PI3K activity in POMC neurons. Thus we investigated the role of class IA PI3K in leptin and insulin action on POMC neurons in a cell-type specific manner (Hill et al., 2008). We used a model in which a mouse was globally deficient for the P85β regulatory subunit and the P85α regulatory subunit was selectively deleted from POMC neurons in a Cre-dependent manner. These animals differed from those with a selective STAT3 deletion in POMC cells such that animals which lacked PI3K in POMC neurons had normal body weights food intake and glucose levels (Xu et al., 2007). However these animals failed to suppress food intake in response to acute leptin administration. Moreover, leptin failed to modify the cellular activity of POMC neurons lacking PI3K regulatory subunits. These data are supportive of the concept that leptin signaling is comprised of multiple cascades constituting the metabolic effects of leptin action. This study was also the first to delineate the requirement of PI3K signaling in POMC neurons for the acute effects of insulin on arcuate POMC neurons. Interestingly insulin and leptin both modify arcuate cellular activity via PI3K signaling, however leptin depolarizes while insulin hyperpolarizes POMC cells via a putative mixed cation channel and a Katp conductance, respectively. At first glance, it may seem illogical that leptin activates while insulin inhibits arcuate POMC neurons via the same signaling cascade, PI3K. However it was unclear whether the acute effects of leptin and insulin are observed in similar or distinct populations of POMC neurons.
We next assessed how leptin and insulin may acutely interact within POMC cells of the mediobasal hypothalamus by combining histological techniques with patch-clamp electrophysiology (Williams et al., 2010). We found that leptin and insulin receptors appeared to be distributed in an exclusive rostro-caudal and medio-lateral patterns. Additionally, the acute effects of leptin and insulin mirrored these anatomical patterns. Moreover, all POMC neurons which were acutely inhibited by insulin failed to show the characteristic leptin-induced depolarization. In all these data suggest a segregation of the acute effects of leptin and insulin within arcuate POMC cells (Fig. 2). However, a recent report suggested that both acute leptin and insulin responses can be observed in a small subset of arcuate POMC cells suggestive that this segregation may not be absolute (Al-Qassab et al., 2009). Al-Qassab and colleagues also found that the effects of leptin and insulin in POMC and NPY/AgRP neurons maybe dependent upon the various catalytic isoforms of Class Ia PI3K. For instance, selective deletion of P110β from POMC cells resulted in central leptin resistance, increased adiposity, and diet-induced obesity, while POMC P110α null mice experienced minor energy expenditure deficits except when exposed to HFD. Moreover, the weight reducing effects of central leptin administration was absent in POMC P110β null mice. Accordingly, the leptin-induced depolarization of POMC cells requires P110β, while P110α and P110β are both required to mediate the insulin-induced hyperpolarization in POMC cells. Similarly, P110β appeared more important in AgRP neurons for the regulation of energy balance such that selective deletion of P110β resulted in hypophagia, leanness, and resistance to diet-induced obesity. AgRP P110β null mice also experienced an increased sensitivity to the weight reducing effects of centrally administered leptin. Interestingly, both P110α and P110β are required for the acute effects of insulin on AgRP cells. However, contrary to the prevailing basic model of the acute effects of leptin and insulin in the arcuate nucleus the authors failed to observe a leptin-induced hyperpolarization of NPY/AgRP cells, and instead reported an insulin-induced depolarization of AgRP cells (Cone et al., 2001; Cowley et al., 2001; van den Top et al., 2004). Moreover, deletion of P110α or P110β or pharmacological inhibition of PI3K in NPY/AgRP cells revealed an insulin-induced hyperpolarization which was dependent upon a Katp channel. Clearly further investigation is needed in order to delineate the disparity of acute effects on cellular activity between the current study and previous work (Al-Qassab et al., 2009; Cowley et al., 2001; van den Top et al., 2004). However together these data are supportive of a segregation of leptin and insulin responses within POMC neurons and is intriguing given the variety of downstream targets of POMC cells and the differing metabolic phenotypes observed from selective deletion of either leptin or insulin receptor in POMC cells (Baker and Herkenham, 1995; Elias et al., 1999; Elias et al., 1998; Elmquist et al., 1998a; Swanson and Kuypers, 1980; Zheng et al., 2005).
Deletion of leptin receptors in POMC neurons results in moderate obesity without any effect on glucose homeostasis (Balthasar et al., 2004). Deletion of insulin receptors alone from POMC neurons fails to influence energy or glucose homeostasis (Konner et al., 2007). We recently reported that leptin and insulin alter the cellular activity in disparate subgroups of arcuate POMC cells, however these data did not address the potential crosstalk of leptin and insulin signaling within POMC neurons as it relates to signal transduction pathways which regulate genomic transcription and may ultimately regulate energy and glucose homeostasis (Williams et al., 2010). Thus we designed experiments to assess the requirement of insulin and leptin receptors to regulate energy and glucose balance within POMC neurons (Hill et al., 2010). Interestingly deletion of both insulin and leptin receptors in POMC neurons resulted in modest effects on body weight with more profound effects on glucose balance inducing hepatic insulin resistance and severe diabetes not seen in either single deletion alone. The double deletion also resulted in reduced fertility and elevated androgen levels in female mice in pattern reminiscent of polycystic ovarian syndrome (PCOS). Thus the double knockout unmasks a liver phenotype as well as a role of leptin/insulin signaling in POMC neurons to regulate fertility. Moreover, these data suggest that POMC neurons can regulate glucose and insulin indepedent of body weight. In support of divergent actions on body weight and glucose homeostasis Roger Unger and colleagues recently showed that a single injection of leptin normalizes the deranged homeostasis of rodents with type 1 diabetes, an effect that may be dependent on leptin signaling in POMC neurons (Yu et al., 2008).
Similar to prior leptin and insulin studies, genetic mouse models have been used to describe a role for the serotonergic (5HT) system in the regulation of feeding behavior and body weight. Serotonin has been shown to have profound effects on energy homeostasis (Nonogaki et al., 1998; Tecott et al., 1995). This has led to the development of therapeutics, such as sibutramine and d-fenfluramine (d-Fen), which target the serotoninergic system in order to achieve reductions in body weight by reducing food intake and increasing energy expenditure (Bray et al., 1999; Rowland and Carlton, 1986). Serotoninergic terminals converge on the arcuate nucleus which is supportive of an endogenous role of serotonin in regulating POMC and NPY neurons (Kiss et al., 1984; Zhou et al., 2005). Lora Heisler and colleagues found that similar to leptin action in the arcuate nucleus, serotonin activates POMC neurons while inhibiting NPY neurons (Heisler et al., 2002; Heisler et al., 2006). Moreover, the serotonin induced activation of POMC neurons is thought to be mediated by the 5HT2CR, while the inhibition of NPY neurons may be dependent on the 5HT1BR (Heisler et al., 2002; Heisler et al., 2006). The similarity between leptin and serotonin regulation of arcuate neuronal activity suggests that serotonin induced-regulation of melanocortin neurons may be at least partly responsible for the serotonin induced effects on food intake and energy expenditure. In support of this idea, mice with a global 5HT2CR deficiency are both hyperphagic and obese (Nonogaki et al., 1998; Tecott et al., 1995). Interestingly, d-Fen’s anorexigenic effects are dependent upon 5HT2CR activation (Vickers et al., 1999). Heisler and colleagues also demonstrated that the reduction in food intake requires MC4 receptors downstream of arcuate POMC neurons (Heisler et al., 2002; Heisler et al., 2006; Lam et al., 2008). Clearly, serotonin can modify the activity of arcuate melanocortin neurons, however the involvement of 5HT2CRs in the regulation of POMC neurons and the sufficiency of serotonin signaling in POMC neurons to regulate energy homeostasis was unclear. In an attempt to determine the sufficiency of serotonin signaling in POMC neurons we generated a mouse model that is globally deficient for 5HT2CR expression by inserting a loxP-flanked transcription blocker (loxTB) into the 5HT2CR gene (2C null) (Xu et al., 2008). We then mated these 2C null mice to POMC-Cre mice which resulted in 5HT2CR reactivation selectively in POMC neurons (2C/POMC). Selective expression of 5HT2CRs in POMC neurons normalized the hyperphagia, hyperactivity, obesity, and attenuated pharmacological responses to 5-HT therapeutics observed in 5HT2CR nulls. These data suggest that expression of 5HT2CRs selectively in POMC neurons is sufficient to normalize defects in food intake and weight gain observed in 2C null mice and suggests an important role of serotonergic signaling in POMC neurons to maintain energy homeostasis. In addition to obesity and hyperphagia, mice lacking 5HT2CR develop hyperinsulinemia and impaired glucose tolerance (Nonogaki et al., 1998). Moreover, 5HT2CR agonists improve glucose sensitivity and insulin resistance in a model of DIO (Zhou et al., 2007). Zhou and colleagues also showed that similar to the effects of 5HT2CR agonists on food intake and weight gain, 5HT2CR agonists require MC4 receptors for improvement in glucose homeostasis. Interestingly, mice deficient for both 5HT2CR and leptin (OB2C) exhibited an additive effect on impaired glucose tolerance suggesting a synergistic action of leptin and serotonin to regulate glucose homeostasis (Wade et al., 2008). Future experiments will have to delineate if the 5HT22CR induced improvements in diabetes are dependent upon central melanocortin action. Also, as previously discussed, recent evidence in our laboratory suggests that the acute effects of leptin on POMC neurons is distinct from the acute effects of insulin. It is currently unclear whether the effects of serotonin on arcuate POMC neurons will segregate to one of these populations or whether serotonin may be acting on an entirely separate subpopulation of POMC neurons, ultimately adding another level of complexity to our understanding of leptin, insulin, and serotonin action within arcuate POMC neurons. Thus, although much progress has been made in understanding the role of serotonin in regulating energy and glucose homeostasis, many questions remain and await further investigation.
In summary, since the seminal observations linking obesity with hypothalamic dysfunction, much progress has been made in understanding the neural circuitry controling metabolic homeostasis. Numerous hormones, including insulin and leptin along with neuronal signals such as serotonin and the melanocortins have been shown to act on specific neuronal populations to modulate food intake, glucose homeostasis and energy expenditure. Within the medial basal hypothalamus, POMC and NPY/AGRP neurons have been identified as important intergrators of metabolically relevant neuronal and hormonal inputs which act antagonistically to influence energy balance. Through the use of a molecular genetic approach, neuron specific manipulation of leptin, insulin, serotonin and melanocortin receptors has revealed the necessity and sufficiency of metabolic signaling in the maintenance of homeostasis in specific nuclei not only within the hypothalamus and the POMC – AGRP/NPY circuit but in other structures including the hindbrain. Both receptor deletion and reactivation studies have subsequently described a distributed network of neurons that responds to these salient metabolic cues. Lastly, further studies investigating the mechanism of action of these cues have revealed specific cascades, such as signaling through PI3K, responsible for select actions of leptin and insulin in both the arcuate and ventral medial hypothalamic nuclei of the hypothalamus. Thus, while considerable work remains to be done, our understanding of the functional neuroanatomy and physiology of the networks controlling metabolic homeostasis has advanced considerably, presenting us with novel targets for the development of therapeutic interventions targeted for the treatment of diabetes and obesity.
This work was supported by the National Institutes of Health (K01 DK087780 to KWW, K99 DA024719 to MMS, and DK53301, MH61583, DK081185 and DK71320 to JKE) and by an ADA Smith Family Foundation Pinnacle Program Award to JKE.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.