|Home | About | Journals | Submit | Contact Us | Français|
The prevalence of obesity in older people is the leading cause of metabolic syndromes. Central neurons serving as homeostatic sensors for bodyweight control include hypothalamic neurons that express pro-opiomelanocortin (POMC) or neuropeptide-Y (NPY) and agouti-related protein (AgRP). Here we report an age-dependent increase of mammalian target of rapamycin (mTOR) signaling in POMC neurons that elevates the ATP-sensitive potassium (KATP) channel activity cell-autonomously to silence POMC neurons. Systemic or intracerebral administration of the mTOR inhibitor rapamycin causes weight loss in old mice. Intracerebral rapamycin infusion into old mice enhances the excitability and neurite projection of POMC neurons, thereby causing a reduction of food intake and bodyweight. Conversely, young mice lacking the mTOR negative regulator TSC1 in POMC neurons, but not those lacking TSC1 in NPY/AgRP neurons, were obese. Our study reveals that an increase in mTOR signaling in hypothalamic POMC neurons contributes to age-dependent obesity.
Obesity poses a growing risk for the middle-aged adult population. This phenomenon may have different causes including genetic predisposition, poor dietary habits and sedentary life-style. As the aging population increases, obesity has become a global health issue especially in developed counties (Marcellini et al., 2009). The increased prevalence of obesity among old people has emerged as the leading cause of metabolic syndromes such as type II diabetes, heart diseases and stroke (Mokdad et al., 2001). Hence, weight control for the elderly population is not just a cosmetic procedure; it will dramatically reduce the risk of obesity-related comorbidities, which are commonly associated with the aging process (Marcellini et al., 2009). As the obesity and diabetes epidemics continue to rise and the global population ages further, greater efforts are being devoted to understanding the mechanisms of age-dependent metabolic disorders (Freedman et al., 2002).
The hypothalamus is the control center for food intake and bodyweight (Berthoud and Morrison, 2008; Hill et al., 2008). Among hypothalamic neurons, the POMC neurons that express pro-opiomelanocortin (POMC) and secrete an anorexic neuropeptide melanocyte-stimulating hormone (α-MSH), a proteolytic product of POMC, and the NPY/AgRP neurons that express and secrete the orexic neuropeptides Neuropeptide-Y (NPY) and agouti-related protein (AgRP), are the key players in regulating food intake and energy homeostasis (Elias et al., 1998). Defects of this POMC-NPY/AgRP circuit cause serious abnormality in food intake and bodyweight control (Elmquist, 2001). Notwithstanding our knowledge about this hypothalamic circuit in the regulation of bodyweight in the normal physiological setting, how this circuit might be altered with aging is an open question.
The mammalian target of rapamycin (mTOR) is a serine/threonine kinase that integrates nutrients and hormonal signals to control cell growth and proliferation (Wullschleger, 2006). The mTOR activity is negatively regulated by the tuberous sclerosis complex (TSC) composed of TSC1 and TSC2; AKT activates mTOR by inhibiting TSC (Wullschleger et al., 2006). The mTOR inhibitor rapamycin, an FDA-approved drug for patients with organ transplant, has been considered for treatment of psychiatric disorders and metabolic disorders, and as a promising longevity-enhancing drug (Harrison et al., 2009). In addition to extending life span, reducing mTOR activity may improve symptoms in neurodegenerative diseases associated with aging, such as Alzheimer’s disease and Parkinson’s disease. This beneficial effect of reducing mTOR signaling might further improve the quality of life of the aging population (Garelick and Kennedy, 2011). Given that adult-onset obesity could result from hypothalamic neurodegeneration (Ryu et al., 2008; Xu et al., 2005) and leptin, an adipostatic hormone secreted by white adipocytes, fail to augment energy expenditure in older rodents, indicating leptin signaling may be attenuated with aging (Li et al., 1998), we wondered whether age-dependent obesity might be associated with leptin resistance due to hyperactive mTOR signaling in the hypothalamic neuronal circuit. In particular, whereas in young mice only ~10% of POMC neurons show active mTOR signaling and this fraction is increased to 18% upon leptin stimulation (Reeds et al., 2010), we wondered whether mTOR signaling is increased in POMC neurons of aging mice. If so, suppressing this excessive mTOR signaling via rapamycin administration may re-establish the hypothalamic circuit and ameliorate age-dependent obesity.
In this study, we have found that mTOR signaling is elevated in the hypothalamic POMC neurons of old mice, causing silencing of these neurons due to up regulation of KATP channel activity accompanied with an aging-associated expression of the Kir6.2 pore-forming subunit of KATP channels. In support of the critical role of enhanced mTOR signaling in causing obesity, removal of the mTOR negative regulator TSC1 in POMC neurons of young mice elevated KATP channel activity, leading to silencing of POMC neurons and obliteration of leptin-induced release of the anorexic hormone α-MSH. Whereas TSC1 deletion in POMC neurons resulted in obesity of young mice, TSC1 deletion in NPY/AgRP neurons had no effect on neuronal excitability or bodyweight. Remarkably, infusion of the mTOR inhibitor rapamycin into the brain of aging mice caused a reduction of bodyweight. This intracerebral rapamycin infusion reduced food intake without altering the blood glucose level. It also suppressed KATP channel activity to increase repetitive firing of POMC neurons, and expanded the POMC neuronal projection into the paraventricular nucleus (PVN) involved in controlling food intake and bodyweight. Taken together with our finding that systemic rapamycin injection also reduces bodyweight of old mice, this study raises the prospect of potential therapeutic application of rapamycin to reduce midlife obesity.
Studies of POMC neurons from young rodents (typically < 3 months old) have shown that these neurons fire action potentials repeatedly so as to cause α-MSH secretion; elimination of action potential firing in POMC neurons abolishes α-MSH secretion (Bunel et al., 1990). It is an open question whether POMC neurons are still active in older rodents, which tend to display obesity and increased-adiposity. Our recording of GFP-labeled POMC neurons from transgenic mouse hypothalamic slices revealed that POMC neurons from young (1 month old) mice were electrically active (Figure 1A and D). In contrast, POMC neurons from aging (>6 months old) mice, which had gained more weight (Figure S1C), were silent (Figure 1B and D). As control for the health of brain slices from aging mice, recordings from neurons without GFP labeling from 12 months old POMC-GFP mice revealed that these neurons fired action potentials repeatedly, and some displayed rhythmic bursting characteristic of tuberoinfundibular neurons in the arcuate nucleus (Figure S1A and B) (Lyons et al., 2010). By surveying four different age groups of mice, we found a significant reduction of input resistance of POMC neurons from 6, 12 or 18 months old mice as compared to those from 1 month old mice (p<0.05 One-way ANOVA with Bonferroni post hoc test) (Figure 1F), leading to a reduction of action potential firing (Figure 1D). This silencing of POMC neuronal activity was accompanied with a more hyperpolarized resting potential (Figure 1C). We also found that the membrane capacitance, which provides a measure of the cell surface area, was significantly increased in POMC neurons from 18 months old mice (Figure 1E).
POMC neurons in the arcuate nucleus receive multiple synaptic inputs and display synaptic plasticity (Liu et al., 2012) in response to physiological changes (Dicken et al., 2012; Pinto et al., 2004). To ask whether synaptic plasticity may contribute to the POMC neuron silencing in aged mice, we applied the glutamate receptor antagonist DNQX (10 μM) and the GABA receptor antagonist picrotoxin (50 μM) to the bath solution. Blocking these transmitter receptors did not affect the resting potential or reduce action potential firing of POMC neurons from young (1 month old) mice (Figure 2A and C), nor did the antagonists for these transmitter receptors alter the resting potential of the silent POMC neurons from older (6 months old) mice (Figure 2B and C). Interestingly, glibenclamide, a specific KATP channel blocker, induced significant depolarization and caused these silent POMC neurons from aged mice to fire action potentials even in the presence of the transmitter receptor antagonists (Figure 2B and C). To test for the possibility that POMC neurons from old mice have increased expression of KATP channels composed of the pore-forming subunit Kir6.2 (KCNJ11) and the auxiliary sulfonylurea receptor subunit SUR1, we performed single-cell RT-PCR of POMC neurons from young mice (Figure 2D) and old mice (Figure 2E). We found an age-dependent up regulation of Kir6.2 but not SUR1 mRNA (Figure 2D-F). Thus, aging is accompanied with Kir6.2 expression and KATP channel-mediated silencing of POMC neurons.
Previous studies have shown that increased mTOR activity causes hypertrophy in numerous cell types including neurons (Meikle et al., 2008). Given the suggestion that aging brains may have elevated mTOR activity (Garelick and Kennedy, 2011), we wondered whether the hypertrophy of POMC neurons in aged mice (Figure 1E) might be caused by increased mTOR signaling in POMC neurons. Consistent with previous findings of low mTOR activity in POMC neurons from young mice (Reed et al., 2010; Villanueva et al., 2009), our immunostaining of hypothalamic sections from the POMC-GFP transgenic mice with antibody against GFP to label POMC neurons and antibody against phospho-S6 (p-S6), an endogenous reporter of mTOR activity, revealed that only a small fraction of the GFP-labeled POMC neurons in one month old mice showed phospho-S6 immunoreactivity (Figure 3A-F), unlike NPY/AgRP neurons that displayed robust mTOR signaling in young mice (Figure S2). In contrast, the mTOR activity as indicated by phospho-S6 immunoreactivity was significantly elevated in POMC neurons of 12 months old mice (Figure 3G-M).
Given that POMC neurons and NPY/AgRP neurons are the two major appetite-controlling neurons in the hypothalamus, we tested whether obesity could be induced by activating mTOR signaling via conditional knockout of its upstream negative regulator TSC1 (Meikle et al., 2008) in either POMC neurons (Figure 4) or NPY/AgRP neurons (Figure 4 and S3). Consistent with a previous study (Mori et al., 2009), we found deleting Tsc1 in POMC neurons via Pomc-cre (Figure S4) but not in NPY/AgRP neurons via Agrp-cre caused obesity (Figure 4A and B). Moreover, we found that TSC1 is essential for maintaining the excitability of POMC neurons but not NPY/AgRP-neurons; conditional knockout of Tsc1 in POMC neurons silenced these neurons (Figure 4C), which could be induced to fire action potential via current injection (Figure S4), whereas conditional knockout of Tsc1 in NPY/AgRP neurons had no effect on their firing pattern (Figure 4C), resting membrane potential (Figure 4G) or neuronal size (Figure 4E). Recapitulating features of POMC neurons in aged mice (Figure 1), removal of the mTOR negative regulator TSC1 in POMC neurons resulted in hypertrophic soma (Figure 4D), hyperpolarized resting membrane potential (Figure 4F) and reduced excitability (Figure 4H). Since the PI3K signaling pathway has been proposed to silence POMC neurons through activation of KATP channels (Plum et al., 2006) and mTOR is downstream of PI3K in the signaling pathway, we wondered whether the elevated mTOR signaling caused silencing of POMC neurons by up regulating their KATP channel activity. To test this possibility, we dialyzed the neuron under patch-clamp whole-cell recording with an internal solution containing low (0.5 mM) MgATP and treated the hypothalamic slice with 300 μM diazoxide, a KATP channel opener, to estimate the total KATP channel activity in POMC neurons (Speier et al., 2005). We found that removing the mTOR negative regulator TSC1 indeed caused a significant increase of the total KATP channel conductance (Figure 4I). These results lend further support to the notion that elevation of mTOR signaling causes silencing of POMC neurons mainly by increasing the KATP channel activity.
Previous studies indicate that hypothalamic KATP channels regulate the blood glucose homeostasis: local application of glibenclamide to the arcuate nucleus reduces the ability of glucagon-like peptide 1 (GLP-1) to suppress hepatic gluconeogenesis (Sandoval, 2008), and hypothalamic KATP channel activation by infusing diazoxide, a specific KATP channel opener, to the 3rd ventricle suppresses glucose production thereby lowering blood glucose (Pocai et al., 2005). Having found that the increased mTOR signaling in POMC neurons from Pomc-cre;Tsc1-f/f mice caused KATP activation (Figure 4I), we asked whether the increased KATP currents in POMC neurons affect glucose homeostasis. To our surprise, at 3 months of age, those Pomc-cre;Tsc1-f/f mice that were about 50% heavier than their heterozygous littermates (Figure 4B) did not show signs of glucose intolerance; instead, they displayed a slight but significant improvement in glucose tolerance based on intraperitoneal glucose tolerance test (Figure 4J). Similar tests revealed no alterations of glucose tolerance in Agrp-cre;Tsc1-f/f mice (Figure 4K).
Activation of mTOR in POMC neurons causes an attenuation of sensitivity to leptin (Mori et al., 2009), which exerts its anorexic effect by stimulating α-MSH secretion from POMC neurons (Forbes et al., 2001). To test whether removal of TSC1 from POMC neurons attenuates the ability of leptin to induce α-MSH release, we measured α-MSH secretion from hypothalamic tissue explants from Pomc-cre;Tsc1-f/+ mice and Pomc-cre;Tsc1-f/f mice. Indeed, leptin failed to stimulate α-MSH secretion from Pomc-cre;Tsc1-f/f hypothalamic explants (Figure 5), while leptin applied together with 10 μM glibenclamide caused an increase of α-MSH secretion (Figure 5). These results indicate that the elevated KATP channel activity in POMC neurons lacking TSC1 reduced the ability of leptin to stimulate α-MSH secretion likely by silencing those POMC neurons.
As mTOR signaling in POMC neurons was significantly increased in aged mice, we next tested whether suppressing mTOR signaling by rapamycin can cause weight loss. Indeed, daily intraperitoneal injection of rapamycin at 5 mg/kg of bodyweight, the dose that has been shown previously to be effective for rapamycin to cross blood-brain barrier without causing bodyweight change in young adult mice (Meikle et al., 2008) (Figure 6A), reduced the bodyweight of 12 months old mice (Figure 6B). Because chronic systemic administration of rapamycin causes glucose intolerance and hypoinsulinemia (Yang et al., 2011), we infused rapamycin into the lateral ventricle in the brain through an osmotic pump to avoid potential complications of rapamycin actions in the periphery. Similar to systemic rapamycin injection, chronic intracerebral infusion of rapamycin significantly suppressed mTOR signaling in POMC neurons from 12 months old mice (Figure S5). Moreover, intracerebral rapamycin caused weight loss of 12 months old mice (Figure 6C). Those mice receiving intracerebral rapamycin infusion had normal glucose tolerance (Figure 6D), indicating that rapamycin had largely been confined within the central nervous system. Thus, the weight loss was due to reduced mTOR signaling in the central nervous system. Old mice receiving rapamycin infusion into the brain also exhibited a reduction in food intake (Fig. 6E). Whereas rapamycin suppressed mTOR signaling in NPY/AgRP neurons as well (Figure S2), it did not alter their biophysical properties nor did it halt the action potential firing (Figure S6), in contrast to the ability of rapamycin to enhance the excitability of POMC neurons (Figure 7). These findings provide further support for our hypothesis that the age-dependent elevation of mTOR signaling inactivates POMC neurons to cause excessive food consumption and bodyweight gain, raising the possibility that suppressing mTOR activity by rapamycin can restore POMC neurons’ anorexic effect in exerting weight control.
To directly assess the impact of intracerebral rapamycin infusion on neuronal activity, we recorded from POMC neurons of 12 months old mice that had rapamycin infused into the brain for three weeks. Similar to untreated old mice, POMC neurons from control old mice receiving vehicle-only from the pump were silent (Figure 7A). Applying 10 μM glibenclamide to block KATP channels restored neuronal excitability and action potential firing (Figure 7A). In contrast, POMC neurons from old mice receiving rapamycin infusion were more excitable and fired action potentials repeatedly (Figure 7B), had reduced KATP channel activity (Figure 7C) and more depolarized resting membrane potential (Figure 7D), thus recapitulating the features of POMC neurons from young mice (Figure 1). Interestingly, we found rapamycin not only restored the excitability but also reduced the soma size of POMC neurons (Figure 7E), consistent with previous reports of the mTOR effects on neuronal morphology (Jaworski et al., 2005).
Previous study has shown that removal of TSC1 from POMC neurons reduces their projection to the target areas such as the PVN that mediates the control of food intake and bodyweight (Mori et al., 2009). To quantify the POMC neuronal projection, we measured the GFP-labeled POMC neurite density within the PVN region from 12 months old POMC-GFP transgenic mice that had received intracerebral rapamycin infusion for 3 weeks. Indeed, we found that mice with rapamycin infusion had more extensive POMC neurites in the PVN region than control mice with vehicle-only infusion (Figure 7 F-H).
Our study shows that mTOR signaling in hypothalamic POMC neurons is elevated in aged mice and we propose that this heightened mTOR activity in POMC neurons contributes to midlife obesity via two mechanisms: silencing POMC neurons by activating KATP channels and reducing POMC neurite projection to its downstream target such as the PVN (Figure 8).
As POMC neurons exert their anorexic effect via α-MSH secretion, and secretion of the anorexic α-MSH is highly dependent on the POMC neuronal excitability, modulating POMC neuron electrical activity can have a dramatic effect on bodyweight. Previous study has shown that nutrient overload such as high fat diet silences the POMC neurons via activation of peroxisome proliferator-activated receptor γ (PPAR-γ) thereby reducing reactive oxygen species. Pharmacological block of PPAR-γ re-activates those silent POMC neurons and reduces food intake (Diano et al., 2011). In our study of aging-dependent obesity, we found that elevating mTOR signaling causes KATP channel activation to silence POMC neurons (Figure 1 to to4).4). It would be interesting to test whether the PPAR-γ mediated neuronal silencing involves KATP channel activation.
KATP channels can be activated under injury such as hypoxia (Ballanyi, 2004). Since brain slice preparations from old mice are more fragile, we need to consider the possibility that neurons with higher KATP channel density may have been more resistant to the insults during the slice preparation. This possibility seems unlikely for the following reasons: First, slices from old mice treated with rapamycin had less KATP channel activity in POMC neurons, which were also more excitable than POMC neurons from old mice receiving vehicle-only infusion (Figure 7A and B). Second, recording from neurons without GFP expression, which are hence not POMC neurons, in the arcuate nucleus of old POMC-GFP mice revealed that those neurons showed excitability similar to that seen in young mice (Figure S1). These controls provide validation for our conclusion that aging is associated with reduced excitability of POMC neurons.
POMC neurons in the hypothalamic arcuate nucleus are heterogeneous; they may exhibit different responses to hormones and neurotransmitters. Thus, altering mTOR activity will likely change the responsiveness of POMC neurons to a number of different inputs (Williams et al., 2010). mTOR can be activated by multiple ways such as branch-chained amino acids (Cota et al., 2006), high protein diet (Ropelle et al., 2008), leptin and insulin (Blouet et al., 2008). Previous studies have found that short-term activation of mTOR by acute leucine infusion (Cota et al., 2006) or through inhibiting the AMP-activated protein kinase (AMPK), an endogenous mTOR suppressor, reduces food intake and bodyweight (Ropelle et al., 2008). However, other studies have shown that genetic manipulation to increase mTOR activity by deleting the Tsc1 gene in POMC neurons results in hyperphagic obesity in mice (Mori et al., 2009). In our study, we found that sustained elevation of mTOR activity due to conditional knockout of Tsc1 in POMC neurons resulted in obesity of young mice (Figure 4). It takes at least two weeks of rapamycin treatment to cause bodyweight loss. Moreover, chronic mTOR inhibition by intracerebral rapamycin infusion suppressed appetite and reduced bodyweight of old mice (Figure 6). It should be noted that POMC neurons of young healthy mice rarely show any elevated mTOR activity (Reed et al., 2010; Villanueva et al., 2009). Whereas leptin increases mTOR activity in the basomedial hypothalamus including the arcuate nucleus (Blouet et al., 2008; Cota et al., 2006), leptin only modestly activates mTOR in POMC neurons (Reed et al., 2010). Our finding of elevated mTOR signaling in POMC neurons of old mice accounts for the effect of rapamycin on midlife obesity, and is further substantiated by recapitulating the effects of elevated mTOR signaling on silencing POMC neuronal activity and causing bodyweight gain in young mice with conditional knockout of Tsc1 in POMC neurons.
It is important to consider various scenarios to reconcile the discrepancy that short-term mTOR activation causes weight loss while chronic activation of mTOR causes obesity. Depending on the dosage and duration, rapamycin may affect glucose homeostasis in different ways (Blagosklonny, 2011b). Rapamycin has also been proposed to have a hormetic effect on aging and aging-related diseases (Blagosklonny, 2011a). Short-term rapamycin transiently affects intracellular signaling activated by nutrient overload; it does not affect bodyweight but abrogates the anorexia induced by nutrients such as leucine (Cota et al., 2006). On the other hand, long-term rapamycin treatment has a profound impact on neuronal morphology and functions. Ion channels typically have slow turnover rates (Crane and Aguilar-Bryan, 2004), and any loss of neurite projection may require a few days to be restored (Keck et al., 2011). Similarly, Mori and colleagues have shown that activating mTOR in POMC neurons by deleting the Tsc1 gene also results in enlarged POMC neuron soma and reduced neurite projections to the PVN, and chronic intraperitoneal injection of high doses of rapamycin restores the projection (Mori et al., 2009). In our study, we have identified that aging is one of the physiological factors to activate mTOR in POMC neurons and the genetic activation of mTOR in conditional knockout mice recapitulates the physiological consequences of aging by activating mTOR in POMC neurons.
Glucose intolerance and diabetes are commonly associated with obesity (Mokdad et al., 2001). Paradoxically, we did not observe such a deleterious metabolic dysfunction in Pomc-cre;Tsc1-f/f mice, despite the fact that they were obese and hyperphagic. Instead, these mice had an improved glucose tolerance (Figure 4J). Interestingly, a recent study has also revealed that direct leptin action in POMC neurons regulates glucose homeostasis independent of body weight (Berglund et al., 2012). Other than regulating bodyweight and appetite, the arcuate nucleus is well documented as a central regulator for glucose homeostasis. To keep a constant glucose supply to the brain, those hypothalamic neurons that are located outside the blood-brain-barrier can monitor the glucose level in the periphery and send feedback commands to visceral organs such as the liver and pancreas through the descending autonomic system to counter-balance any fluctuation of the glucose supply to the brain (Parton et al., 2007). Moreover, activating KATP channels in hypothalamic neurons has been shown to improve glucose metabolism as infusing diazoxide to the 3rd ventricle suppresses hepatic gluconeogenesis (Pocai et al., 2005) and central GLP-1 reduces peripheral glucose levels by activating KATP channels in POMC neurons (Sandoval et al., 2008). As obesity impairs glucose tolerance, we suspect that the improved glucose tolerance in the Pomc-cre;Tsc1-f/f mice might be more significant in pair-feeding experiments that ensure matched-bodyweight.
Previous study has shown that conditional knockout of Phosphatase and Tensin Homolog (PTEN), a lipid phosphatase that antagonizes the function of PI3K, in POMC neurons causes hyperphagic obesity, raising the possibility that removal of PTEN results in an increase of PIP3, which could interact with KATP channels on the plasma membrane to increase their activity (Plum et al., 2006). In our study, we found that elevated mTOR signaling in POMC neurons increased KATP current, and this heightened KATP channel activity silenced POMC neurons and reduced leptin-stimulated α-MSH secretion; Pomc-cre;Tsc1-f/f mice also exhibited a hyperphagic obese phenotype (Figure 4 and and5).5). Since increasing PIP3 level on the plasma membrane activates PI3K, a canonical activator of mTOR (Wullschleger et al., 2006), it is expected that deleting PTEN in POMC neurons may also activate mTOR, and this elevated mTOR activity could activate KATP channels, as we have shown in our study. Whereas deleting PTEN increases plasma PIP3 thereby prolonging the opening time of KATP channels (Plum et al., 2006), the elevated mTOR signaling in POMC neurons lacking TSC1 likely causes an increase in KATP channel density, because the maximum KATP current level is doubled in the presence of diazoxide (Figure 4I). Since PIP3 and diazoxide share the same common mechanism for KATP channel activation due to increased open time (Koster et al., 1999), the maximum KATP current in POMC neurons without TSC1 should remain unchanged if activating mTOR were to increase KATP current by generating PIP3. It is of interest to note that deleting PTEN in POMC neurons also results in hypertrophic soma as in POMC neurons with an elevated mTOR activity (Mori et al., 2009). It thus seems likely that activation of the PI3K pathway will have effects similar to those caused by elevating mTOR activity, likely an increase of KATP channel density, in addition to an increase of channel open time due to an increase of phosphoinositides such as PIP3.
Rapamycin has been found to affect the expression of Kv1.1 and Kv4.2 in dendrites of hippocampal neurons (Lee et al., 2011; Raab-Graham et al., 2006). Here we provide another example of how mTOR regulates neuronal activity by controlling ion channel density. Under physiological conditions, the ion channel density in neurons is tightly regulated (Ma and Jan, 2002). For example, when Parton et al. (2007) expressed in transgenic mice a dominant negative mutant form of Kir6.2 under the POMC promoter, POMC neurons in these transgenic mice nonetheless exhibit normal levels of KATP channel density (Parton et al., 2007). Ion channel density may be controlled at several different levels including transcription, translation, trafficking, and quality control of the endoplasmic reticulum (ER) (Ma et al., 2001). We found that POMC neurons from old mice express the transcripts for KATP channel subunits Kir6.2 and SUR1, for the most common KATP channel composition in neurons (van den Top et al., 2007). As functional KATP requires the co-assembly of Kir6.2 and SUR1 (Schwappach et al., 2000; Zerangue et al., 1999), the up regulation of Kir6.2 in POMC neurons in older mice is likely to increase the expression of KATP channels, leading to hyperpolarization and neuronal silencing (Figure 2). Moreover, constitutive mTOR activation that results in excessive protein translation could lead to ER stress (Reiling and Sabatini, 2006), and ER stress may silence brain endothelial cells by increasing the activity of Kir2.1 channels (Kito et al., 2011). Interestingly, multi-unit recording in the hypothalamic suprachiasmatic nucleus of aging rats has revealed a reduction in the amplitude of the electrical rhythm (Nakamura et al., 2011). The aging process has also been shown to modulate ion channels such as the expression of Kv1.1 and Kv1.2 in Purkinje neurons (Zhang et al., 2010). It would be of interest to test in future studies whether the age-dependent elevation of mTOR signaling causes ER stress in POMC neurons, and if ER stress or other aspects of mTOR signaling would unleash KATP channel trafficking or in some other ways increase KATP channel density, and ultimately reduce POMC neuron excitability.
We have shown that inhibiting mTOR by infusing rapamycin can promote POMC neuronal projections to their target region, the PVN (Figure 7). POMC neurons exert their anorexigenic effects on neurons expressing melanocortin 4 receptor (MC4R), a mandatory receptor for mediating the α-MSH effect in vivo (Vaisse et al., 1998). The expansion of POMC neuronal projection to the PVN with MC4R expression is likely one of the mechanisms for rapamycin to reduce midlife obesity. Multiple studies have revealed the impact of mTOR signaling on neuronal morphology. For example, rapamycin injection restores axon projection in Pomc-cre;Tsc1-f/f mice (Mori et al., 2009). Other studies have shown that the AKT-TSC-mTOR pathway plays a pivotal role in axon/dendrite polarity, axon/dendrite growth and projection (Choi et al., 2008). Activating mTOR by the AKT-TSC pathway up regulates SAD kinase, a kinase that determines the fate of neurite development by phosphorylating tau protein (Kishi et al., 2005; Wildonger et al., 2008). In the visual system of fruit flies, increased TSC-TOR signaling cell autonomously affects photoreceptor axon guidance (Knox et al., 2007). Recent study also has shown that deleting the autophagy gene 7 (Atg7) in POMC neurons reduces neurite projection to the PVN (Coupé and Bouret, 2012). Interestingly, Atg7 is inhibited by mTOR (Wyttenbach et al., 2008). Hence, the elevated mTOR signaling in POMC neurons of aged mice may suppress Atg7 and reduce neurite projection. Another study has found that deleting the LKB1 kinase, another suppressor of mTOR, in POMC neurons also reduces POMC neuronal projections to the PVN (Claret et al., 2011). Interestingly, aging is associated with numerous alterations of neuronal morphology, such as the hypertrophic soma of projection neurons with impoverished neuropiles in the limbic system (Machado-Salas and Scheibel, 1979) and the disruption of synaptic circuits of olfactory sensory neurons (Richard et al., 2010). Another fertile ground for future studies would be the age-dependence of neuronal morphology regulation.
In summary, our study has demonstrated an important role of mTOR signaling in POMC neurons in developing age-dependent obesity. Up regulation of mTOR signaling in the hypothalamic POMC neurons causes an increase of KATP channel activity to silence POMC neurons and reduce their anorexigenic output, by suppressing leptin-stimulated α-MSH secretion and by reducing POMC neuronal projections to the target regions (Figure 8). Moreover, rapamycin may provide beneficial effects on aging-related metabolic disorders, by reducing midlife obesity.
This study was approved by the IACUC of the UCSF. At least 3 animals were used for every single experiment. Details about the mouse strain origins and drug preparation are described in the Supplemental Information.
A brain infusion kit (Alzet) was implanted into the right lateral ventricle attached to an osmotic minipump (Model1004 osmotic pump, Alzet). Minipumps and tubing was filled with rapamycin (10 mg/ml) or vehicle only (60% PEG400, 30% cremaphor and 10% DMSO). Detailed procedures are described in the Supplemental Information.
Experimental procedure for intraperitoneal glucose tolerance test is described in the Supplemental Information.
250 μm thick brain slices containing arcuate nucleus were prepared as described previously (Sternson et al., 2005). KATP currents in POMC neurons were measured as the glibenclamide-sensitive slope conductance of a voltage ramp as described previously (Plum et al., 2006; Speier et al., 2005). Briefly, POMC neurons were dialyzed with the same intracellular solution with low Mg2ATP and simultaneously the slices were perfused with diazoxide (Sigma) for ten minutes, and then glibenclamide was added to block KATP currents. Detailed experimental procedures are described in the Supplemental Information.
Single cell RT-PCR: cDNA synthesis single-cell PCR were prepared as described earlier (Liss et al., 1999). An RT-PCR kit (Superscript III, Invitrogen) was used to generate first strain cDNA using random hexamers and the cDNA library for each sample was used for multiplex PCR and nested PCR. Primer sequences were adapted as described earlier (Liss et al., 1999; Price et al., 2008).
The hypothalamic explants containing the arcuate nucleus were incubated for 45 min in 250 μl aCSF then transferred to solutions containing 50 nM leptin (Sigma) or 50 nM leptin plus 10 μM glibenclamide. At the end of each period, the aCSF was frozen until it was assayed for α-MSH by a fluorescent immunoassay (Phoenix Pharmaceuticals). Detailed experimental procedures are described in the Supplemental Information.
Brain sections (20 μm) were stained with primary antibodies against GFP (chicken 1:200, Aves) and phospho-S6 (rabbit, 1:200, Cell Signaling) followed by Alexa dye-tagged secondary antibodies (donkey 1:500, Invitrogen). See Supplemental Information for detailed experimental procedures.
Statistical analyses were performed with Prism software (Graphpad Software) using the Fisher’s Exact test, one-way ANOVA or 2-way repeated measures ANOVA with Bonferroni post-hoc multi-comparison test and Student’s t-test for pair-wise comparisons. p<0.05 was considered statistically significant. GFP-positive neurite densities within the PVN region were first converted to binary file then further quantified by Image J (NIH).
We thank members of the Jans, Xu and Vaisse laboratories at UCSF for discussions, Dr. Chris Bohlen and Xiuming Wong at UCSF for confirming the rapamycin activity, Dr. James Warne at UCSF for measuring α–MSH in tissue explants, Dr. Grant Li at UCSF for providing the Pomc-cre, Tsc1-flox and ZeG mouse lines and Dr. Jeffrey Friedman at Rockefeller University for providing the POMC-GFP mouse line. This work was supported by American Diabetes Association Mentor-Based Fellowship 7-06-MN-29 (to S.B.Y.), NIDDK summer student training grant (to G.B.) and the NIH grant MH065334 (to L.Y.J.). Y.N.J. and L.Y.J. are investigators of the Howard Hughes Medical Institute.
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.