Anorexigenic and Orexigenic Hormone Modulation of Mammalian Target of Rapamycin Complex 1 Activity and the Regulation of Hypothalamic Agouti-Related Protein mRNA Expression

Dysregulation of appetite or loss of caloric sensing can have a major impact on health and quality of life. When appetite suppression is blunted, obesity and associated comorbidities ensue [ 1 ]. Alternatively in conditions such as cancer, sepsis and ageing, appetite is suppressed, which exacerbates the associated cachexia/sarcopenia [ 2 ]. Considering the worldwide projections for obesity and the ageing population, it is vital to understand the molecular mechanisms regulating appetite and energy metabolism as this information may yield suitable pharmacological targets for the treatment of obesity and wasting. Energy homeostasis is maintained, in part, through regulation of body fat stores via feedback signals arising from fat depots that are sensed by neurons in the hypothalamus. For example, leptin and insulin, made and released from adipocytes and pancreatic beta cells, respectively, circulate in proportion to body fat levels and access the basomedial hypothalamus via specialized uptake systems [ 3 , 4 ]. In contrast, ghrelin, an orexigenic hormone, is produced predominantly in the stomach and readily transported into the hypothalamus, although ghrelin is also ex-pressed and released by a small population of hypothalamic neurons potentially allowing a local, fine control of neuronal function [ 5 ]. These hormones control food intake and energy balance by regulating orexigenic and anorexigenic peptide levels and release by neurons in the arcuate nucleus (ARC) of the hypothalamus [ 3 , 6 ]. Insulin and leptin suppress expression of orexigenic agouti-related peptide (AgRP) and neuropeptide Y (NPY) and promote anorexigenic pro-opiomelanocortin (POMC)-derived peptide production. Conversely ghrelin, which increases appetite and fat deposition, increases NPY and AgRP and reduces POMC neuropeptide expression and release, respectively [ 5 , 7 ]. The precise mechanisms involved in the regulation of hypothalamic peptide expression and release remain to be fully determined. Recent studies have indicated a key role for the class 1A phosphoinositide-3-kinases (PI3Ks) in the signalling cascades linking leptin and insulin with modulation of ARC neuron neuropeptide expression and electrical activity [ 3 , 4 ]. Thus, PI3K is an important integration node for leptin and insulin action on ARC neurons. However, the role of downstream mediators and the contributions of additional signalling pathways that act to modify PI3K-driven outputs in ARC neurons are unclear. An important signalling component downstream from PI3K in the hypothalamus is the mammalian target of rapamycin complex 1 (mTORC1), which activates the S6 kinases (S6K1 and S6K2). Refeeding-, amino acid- or leptin treatment increases ARC mTORC1-S6K activity in association with appetite suppression, actions attenuated by the specific mTORC1 inhibitor, rapamycin [ 8 , 9 ]. Furthermore, mTORC1 activity is predominantly located to the ARC and highly colocalized to leptin receptor-expressing NPY/AgRP and, to a lesser extent, POMC neurons [ 8 , 9 ]. These data are consistent with ARC mTORC1-S6K activity involvement in anorexigenic signalling. Indeed, recent studies have demonstrated that decreased hypothalamic mTORC1-S6K signalling is causally linked to diet-induced obesity and leptin resistance [ 10 , 11 , 12 ]. However, mice globally deficient in S6K are lean, insulin sensitive and resistant to obesity induced by high-fat diets [ 13 ]. This may be explained by the observation that hyperactive mTORC1 activity in liver and muscle is associated with insulin resistance and metabolic dysfunction [ 14 , 15 ]. Taken together, these data indicate that dysregulated mTORC1 activity either centrally or peripherally has significant consequences for metabolic status, with the final outcome also dependent on the site of modification. Ghrelin stimulates food intake by acting as an agonist on the growth hormone secretagogue receptor 1a (GHSR1a) within the hypothalamus [ 16 , 17 ]. Central administration of ghrelin increases ARC AgRP and NPY mRNA levels and induces c-fos and Egr-1 in NPY/AgRP ARC neurons, indicative of increased neuronal activity [ 7 , 18 , 19 ]. The appetite-promoting effects of peripheral and central ghrelin treatment have largely been attributed to stimulation of AMP-activated protein kinase (AMPK), an enzyme activated during fuel deficiency to promote catabolic and inhibit anabolic pathways [ 20 , 21 ]. Indeed, hypothalamic expression of a constitutively active form of AMPK or direct pharmacological activation of hypothalamic AMPK using the AMP analogue, AICAR, increases food intake and expression of AgRP and NPY mRNA [ 20 , 22 ]. Importantly, AMPK activation inhibits mTORC1 activity in a variety of cell types [ 23 ]. However, it is unclear whether ghrelin-mediated induction of hypothalamic AgRP/NPY gene expression involves AMPK and/or mTORC1 signalling. It is also uncertain whether mTORC1 is a downstream signalling node in hypothalamic neurons, capable of integrat

Confluent GT1-7 cells, in 6-well dishes, were serum starved for 3 h prior to hormone: leptin 50 n M (R&D Systems, Abingdon, UK), insulin 50 n M (Novo Nordisk, Crawley, UK), or ghrelin 100 n M (Tocris Bioscience, Bristol, UK) and/or drug: wortmannin 100 n M (Sigma-Aldrich, Gillingham, UK), LY294002 10 μ M , Akti 10 μ M , rapamycin 100 n M , compound C 40 μ M , STO-609 10 μ M (all Merck Chemicals Ltd., Nottingham, UK) and A-769662 50 μ M (Ascent Scientific, Bristol, UK) treatment for various times, as described in the Results section. All drugs were dissolved in saline or 0.1% DMSO, unless otherwise stated. Total protein lysates from the cells were subjected to SDS-PAGE, electrotransferred to a nitrocellulose membrane. Non-specific protein binding sites were blocked by a 60-min incubation in 5% (w/v) skimmed milk in Tris-buffered saline supplemented with 0.5% (v/v) Tween 20. The membrane was then incubated overnight at 4°C using 5% milk solution supplemented with the appropriate primary antibody. Following three washes with 5% milk solution, the membranes were incubated with 5% milk solution supplemented with horseradish peroxidase-conjugated anti-IgG antibody (Fisher Scientific, UK) for 60 min at room temperature. After three further washes with 5% milk solution and two washes with Tris-buffered saline, the immunoreactive proteins were identified by enhanced chemiluminescence (GE Healthcare, UK). Primary antibodies used were: p-S6K (p85 S6K (Thr412) and p70 S6K (Thr389); 1:1,000), p-AMPKα (Thr172; 1:1,000), p-ACC (Ser79; 1:1,000), ACC (1:1,000) all from New England Biolabs, Hitchin, UK; S6K (1:1,000; Cambridge Bioscience, Cambridge, UK), Actin (1:5,000; Sigma-Aldrich) and AMPKα1 (1:5,000) and AMPKα2 (1:10,000), both of which were kind gifts from Grahame Hardie (University of Dundee). Protein bands on gels were quantified by densitometry using Aida Image Data Analyzer software (version 3), where total density was determined with respect to constant area, background was subtracted and average relative band density was calculated. Phosphoprotein levels are presented as normalized values in relation to control and as a ratio of non-phosphorylated total levels at equivalent time points.

GT1-7 cells were serum starved for 3 h prior to calcium imaging using a conventional MetaMorph imaging system (Universal Imaging Corporation, Marlow, UK) with a Zeiss Axiovert 200 inverted epifluorescence microscope equipped with a ×40 oil immersion objective. Cells were loaded with fura-2 AM (Sigma-Aldrich) for 45 min at 37°C in dark conditions in incubation medium containing 135 m M NaCl, 5 m M KCl, 1 m M MgCl 2 , 2.5 m M glucose, 10 m M leucine, 1 m M CaCl 2 , 10 m M HEPES, pH 7.4). Cells were maintained at 35°C and selected for measurement by placement of Triton-X100 regions of interest around their somata. The fluorescence emission at 510 nm was collected for alternating excitations at 340 and 380 nm and expressed as a ratio of emission fluorescence (340/380 nm). The ratiometric images were collected at 5- to 30-second intervals. This was increased to 2-second intervals during the first 5 min of drug application. Data from 7 separate experiments were derived from the somata of 5–15 cells within a field and are presented as percentage of change in fluorescent ratio, which is proportional to the intracellular calcium concentration.

Male or female adult C57BL6/J mice were maintained on a 12-hour light-dark cycle with free access to water and chow. All animal care protocols and procedures were performed in accordance to the Animal Scientific Procedures Act (1986) and with approval of the University of Dundee Animal Ethics Committee. Animals were fasted for 1 h prior to experiments. The preparation of hypothalamic slices was as previously described [ 24 ]. In brief, horizontal 400-μm coronal brain slices were prepared using a Vibratome (Intracel, Royston, UK) and hypothalamic ARC wedges were cut and incubated in aCSF ± A-769662 (50 μ M ), ghrelin (100 n M ), insulin (50 n M ) or leptin (50 n M ) ± rapamycin (100 n M ) or compound C (40 μ M ), or combinations thereof, for 3 h. Following tissue homogenization and RNA isolation, reverse transcription-PCR was performed using Superscript II (Life Technologies, Paisley, UK) and random primers. AgRP gene expression was then assayed by real-time PCR using a premixed mouse AgRP primer-probe set purchased from Applied Biosystems (Foster City, Calif., USA).

Insulin and Leptin Increase S6K Phosphorylation Levels in GT1-7 Cells To explore the mechanisms by which leptin, insulin and ghrelin alter mTORC1 signalling, we utilized changes in S6K phosphorylation (p-S6K) levels as a surrogate marker for mTORC1 activity in GT1-7 cells. Previous studies have shown that leptin and insulin increase activity through the PI3K signalling pathway in these cells [ 24 ] and that regulation of PI3K-dependent signalling is important for modulation of mTORC1-dependent S6K activity [ 26 ]. Immunoblots for pS6K generally exhibit two bands; the upper band represents p-85 S6K (Thr412) and the lower band p-70 S6K (Thr389). Control experiments demonstrated that leptin and insulin do not alter total levels of S6K over this time period (data not shown), whereas both hormones increase p-S6K levels (fig. 1a, b ). In order to confirm a role for PI3K in leptin- and insulin-induced S6K phosphorylation, GT1-7 cells were pre-treated with vehicle or the PI3K inhibitors, wortmannin (100 n M ) or LY294002 (10 μ M ) for 20 min prior to leptin or insulin stimulation (in the continued presence of inhibitor or vehicle). Basal S6K phosphorylation was not significantly affected by wortmannin, whereas LY294002 almost completely suppressed basal p-S6K (fig. 1a–f ). Both inhibitors were effective blockers of the insulin- (fig. 1a, c, e ) and leptin- (fig. 1b, d, f ) mediated increase in p-S6K. The ability of LY294002 to suppress p-S6K more than wortmannin may be due to this concentration of LY294002 directly inhibiting mTORC1 activity more than 100 n M wortmannin [ 27 , 28 ]. Increased mTORC1 activity elicited by growth factors is reported to be dependent on PI3K signalling through the PKB/Akt pathway [ 29 ]. Therefore, to determine whether insulin and leptin utilize this pathway, GT1-7 cells were pre-incubated (as above) with the selective PKB inhibitor Akti [ 30 ] or the highly selective mTORC1 inhibitor, rapamycin [ 28 ]. The PKB inhibitor, Akti (10 μ M ), prevented the insulin- (fig. 2a, c ) and leptin-mediated (fig. 2b, d ) increases in p-S6K, with the inhibitor having no significant effect on basal p-S6K. In contrast, 100 n M rapamycin abolished basal, insulin- and leptin-induced S6K phosphorylation (fig. 2e–h ). Thus, leptin and insulin increase S6K phosphorylation in GT1-7 cells, utilizing a pathway that depends upon the activity of the PI3K-PKB-mTORC1 pathway.

We have demonstrated that leptin and insulin increase whereas ghrelin or direct activation of AMPK using A-769662 decrease mTORC1 activity, respectively, in GT1-7 cells. Consequently, these results suggest that, at least, part of the final hypothalamic output of anorexigenic and orexigenic hormone-mediated signalling may depend on integration of their actions on upstream (i.e. PI3K-PKB vs. AMPK) signalling pathways that converge on mTORC1-S6K. We therefore determined whether increased AMPK activity could attenuate the anorexigenic hormone-driven increase in mTORC1 activity. Both the insulin- and leptin-mediated increases in S6K phosphorylation were significantly reduced in the presence (with 30 min pre-incubation) of A-769662 (fig. 6a–c ). This shows that direct stimulation of AMPK mimics the effect of the mTORC1 inhibitor, rapamycin, on anorexigenic hormone-induced S6K phosphorylation. However, pre-incubation of GT1-7 cells with 100 n M ghrelin had no significant effect on the insulin-mediated increase in S6K phosphorylation (fig. 6d, e ), perhaps reflecting the smaller effect of ghrelin on AMPK activity observed in comparison to A-769662 (fig. 4e ). Hence, in GT1-7 cells, the inhibitory signal elicited by ghrelin, presumably through increased AMPK activity, is too weak to overcome the stimulatory insulin signal (via PI3K-PKB) to mTORC1, suggesting that the insulin signal dominates. In contrast, the stimulatory signal driven by leptin to mTORC1 is weaker than that of insulin, and pre-incubation with ghrelin (100 n M ) inhibits the leptin signal (fig. 6f ).

The raised levels of S6K phosphorylation by insulin and leptin were inhibited by the presence of rapamycin, strongly indicating that the kinase responsible for S6K phosphorylation in GT1-7 cells is mTORC1. The increase in mTORC1 activity driven by insulin and leptin in GT1-7 cells was shown to require signalling through the PI3K-PKB pathway. LY294002 and wortmannin, commonly used inhibitors of PI3K, inhibited leptin- and insulin-induced S6K phosphorylation in GT1-7 cells. One major caveat regarding these compounds concerns their ability to also inhibit mTOR, the proposed downstream effector of the S6K pathway. However, whereas LY294002 has been reported to inhibit mTOR at concentrations similar to those required for inhibition of PI3K, wortmannin has only been shown to inhibit mTOR at micromolar concentrations [ 28 ]. Additionally, the PKB inhibitor, Akti, is a selective inhibitor of PKB, with very little direct effect on other kinase activity, including mTOR and S6K [ 28 , 30 ]. Taken together these data indicate that leptin- and insulin-induced S6K phosphorylation is, at least partly, due to the activation of the PI3K-PKB pathway. Additionally, previous work has demonstrated that centrally or peripherally administered leptin and insulin increase hypothalamic S6K phosphorylation, in conjunction with their actions to reduce food intake and body weight [ 8 , 9 ]. However, the mechanism by which these hormones increase hypothalamic mTORC1 activity has not been fully explored, with both induction of PI3K signalling and actions secondary to increased neuronal activity plausible. The data presented here support the notion that PI3K and PKB activity are required for insulin and leptin modulation of mTORC1 in hypothalamic neurons and indicate that both hormones utilize the canonical growth factor signalling pathway to increase mTORC1 activity (fig. 8 ). The orexigenic action of ghrelin is likely associated with GHSR1a-dependent signalling in the hypothalamus [ 16 , 17 , 34 ] and recent studies strongly suggest that increased hypothalamic AMPK activity underpins ghrelin-induced food intake [ 33 , 36 ]. Furthermore, activity of the upstream AMPK kinase, CaMKK2, is a central component of ghrelin-mediated regulation of food intake and neuropeptide expression [ 33 ]. Thus, CaMKK2 KO mice exhibited decreased hypothalamic AMPK activity, which was associated with reduced AgRP and NPY mRNA levels and loss of response to ghrelin. Moreover, inhibition of hypothalamic CaMKK2 in control mice by intracerebroventricular application of the CaMKK2 inhibitor, STO-609, inhibited food intake and induced weight loss. Therefore, it has been proposed that ghrelin mediates orexigenic actions via hypothalamic GHSR1a activation, increasing intracellular neuronal calcium, which drives CaMKK2 to raise AMPK activity [ 37 ]. Consequently, we used GT1-7 cells to explore how this pathway is involved in hypothalamic ghrelin signalling. Previous work has indicated that ghrelin directly depolarizes and increases the firing rate of hypothalamic arcuate neurons, including NPY-containing neurons [ 5 , 38 ]. Ghrelin also raises [Ca 2+ ] i levels in neurons and glia [ 39 , 40 , 41 ]. However, the mechanisms by which ghrelin increase [Ca 2+ ] i appear complex, with reports of direct depolarization of neurons initiating channel-mediated extracellular calcium entry [ 38 , 39 , 40 ] and release of calcium from thapsigargan-sensitive intracellular stores [ 41 ]. Here we demonstrated that although ghrelin does not depolarize or excite GT1-7 hypothalamic neurons, it does induce a rise in [Ca 2+ ] i , which may in part be due to increased calcium entry during action potential firing. As we were focused on determining the main constituents of the signalling pathway utilized by ghrelin to alter mTORC1 activity, we have not determined the exact sources of the raised [Ca 2+ ] i in this study. Ghrelin also increased AMPK activity, as measured by increased p-AMPK and p-ACC and by direct kinase activity assay in GT1-7 cells, with the ghrelin-mediated increase in p-AMPK prevented by the presence of the CaMKK2 inhibitor, STO-609. The increase in AMPK activity also coincided, temporally, with a decrease in mTORC1 activity, with the latter outcome blocked by the presence of the AMPK inhibitor, compound C. At present, STO-609 and compound C are the best available inhibitors of CaMKK2 and AMPK, respectively. However, a major caveat regarding both of these drugs is their relative lack of selectivity, with inhibition of a wide range of kinases reported at concentrations that inhibit CaMKK2 and AMPK [ 28 ]. Nevertheless, these data are consistent with the model described above whereby ghrelin raises intracellular calcium levels, increasing activity of CaMKK2, thus raising the levels of p-AMPK and increasing AMPK activity. Higher AMPK activity in these cells was therefore predicted to inhibit mTORC1 activity and antagonize the effects of the anorexigenic hormones on mTORC1 sign