Rap1 in the VMH regulates glucose homeostasis

The hypothalamus is a critical regulator of glucose metabolism and is capable of correcting diabetes conditions independently of an effect on energy balance. The small GTPase Rap1 in the forebrain is implicated in high-fat diet–induced (HFD-induced) obesity and glucose imbalance. Here, we report that increasing Rap1 activity selectively in the medial hypothalamus elevated blood glucose without increasing the body weight of HFD-fed mice. In contrast, decreasing hypothalamic Rap1 activity protected mice from diet-induced hyperglycemia but did not prevent weight gain. The remarkable glycemic effect of Rap1 was reproduced when Rap1 was specifically deleted in steroidogenic factor-1–positive (SF-1–positive) neurons in the ventromedial hypothalamic nucleus (VMH) known to regulate glucose metabolism. While having no effect on body weight regardless of sex, diet, and age, Rap1 deficiency in the VMH SF1 neurons markedly lowered blood glucose and insulin levels, improved glucose and insulin tolerance, and protected mice against HFD-induced neural leptin resistance and peripheral insulin resistance at the cellular and whole-body levels. Last, acute pharmacological inhibition of brain exchange protein directly activated by cAMP 2, a direct activator of Rap1, corrected glucose imbalance in obese mouse models. Our findings uncover the primary role of VMH Rap1 in glycemic control and implicate Rap1 signaling as a potential target for therapeutic intervention in diabetes.


INTRODUCTION 44
45 The brain has long been known as a key regulator of glucose metabolism (1-3). Recent studies have shown that the 46 brain is clearly capable of correcting diabetic conditions (1,2,(4)(5)(6). For example, direct infusion of leptin, insulin, 47 and fibroblast growth factors into the brain exhibits a remarkable antidiabetic effect in animal models of diabetes 48 (7)(8)(9)(10)(11)(12)(13)(14). Clinical and preclinical data demonstrate that pharmacological activation of hypothalamic KATP channels or 49 the serotonin 2C receptor improves glycemic control (15)(16)(17). In addition, deep brain stimulation was further shown 50 to enhance peripheral insulin sensitivity in humans with diabetes (18). Thus, a growing body of evidence strongly 51 suggests the brain as a promising yet unrealized therapeutic target for type 2 diabetes. To further materialize this 52 concept, it is of great interest to identify potentially druggable molecular targets mediating the brain's antidiabetic 53 effects. 54 55 One of the critical hypothalamic sites mediating glycemic control is the ventromedial nucleus of the hypothalamus 56 (VMH). The VMH has been recognized as a hypothalamic nucleus that possesses glucose-sensing neurons (19, 20) 57 and regulates glucose metabolism of peripheral tissues (21,22). Recent genetic and pharmacological studies have 58 demonstrated that multiple hormonal and neural signals regulate VMH neurons to alter glucose balance (23)(24)(25)(26)(27)(28)(29)(30)(31). 59 Further evidence supporting the role of VMH in glucose metabolism stems from optogenetic, electromagnetic and 60 chemogenetic studies demonstrating that manipulations of VMH neural activity influence blood glucose levels, 61 glucose tolerance and peripheral insulin sensitivity (27,(32)(33)(34)(35)(36)(37)(38). Consequently, VMH neurons are thought to be a 62 crucial mediator of the neural glucoregulatory mechanism. However, signaling mechanisms within VMH neurons 63 that mediate whole body glycemic control remain elusive. 64 65 Rap1 is a monomeric small GTPase belonging to the Ras family (39). Rap1, which is encoded by Rap1a and Rap1b,66 is ubiquitously expressed throughout the body. At the cellular level, Rap1 mediates various cellular functions, such 67 as proliferation, differentiation, adhesion, and motility (40). In the central nervous system (CNS), Rap1 is widely 68 expressed in broad areas of the CNS, including hypothalamic nuclei known to control energy and glucose 69 homeostasis (41,42). Rap1 in the CNS is activated in response to acute and chronic high-fat diet (HFD) feeding 70 (42,43). Furthermore, activation of Rap1 diminishes cellular actions of leptin, a crucial hormonal mediator that 71 maintains normal body weight in vitro and in vivo, thereby contributing to adiposity (43,44). In addition, genetic 72 deletion of Rap1 in the forebrain protects mice from HFD-induced metabolic disturbances, such as neural leptin 73 resistance, obesity, and glucose imbalance (42). Consistently, mice with global knockout of EPAC1, a GTP/GDP 74 exchange factor for Rap1 (an upstream activator of Rap1), are also protected from diet-induced obesity and insulin 75 resistance (45). As such, Rap1 signaling in the CNS has emerged as a crucial mediator for the effects of HFD 76 feeding, including the development of leptin resistance, obesity, and glucose imbalance. However, the exact CNS 77 sites and the specific neural populations where Rap1 mediates overnutrition-associated disorders remain to be 78 determined. Here, we defined the physiologic function of Rap1 expressed by the medial hypothalamus, a key CNS 79 site for the control of energy and glucose metabolism, by employing a combination of gain-of-function and loss-of-80 function genetics, pharmacology and glucose clamp studies. 81

Hypothalamic Rap1 controls blood glucose but not body weight in overnutrition conditions 84
While Rap1 activity in the hypothalamus is increased in response to acute and chronic HFD feeding (42, 43), it 85 remains unclear whether increased activity of hypothalamic Rap1 plays a role in HFD-induced body weight gain 86 and glucose imbalance. To directly test this, we increased hypothalamic Rap1 activity by bilaterally injecting adeno-87 associated virus (AAV) expressing a constitutively active GTP-locked human Rap1a variant (46) (AAV-Rap1 V12 ) 88 into the medial hypothalamic area aimed at the VMH. As a control, AAV expressing GFP alone was injected. After 89 the injections of AAVs, the mice were challenged with a HFD to induce diet-induced obesity and hyperglycemia. 90 AAV-mediated ectopic expression of human Rap1 V12 was confirmed in the medial hypothalamic area including the 91 VMH by using qPCR (Supplemental Figure 1A), and GFP fluorescence was found in the VMH (Supplemental 92 Figure 1B), suggesting the VMH as a primary site of hypothalamic expression of Rap1 V12 . Hypothalamic expression 93 of Rap1 V12 resulted in a modest yet significant increase in total Rap1 activity (a 1.6 fold increase, Supplemental 94 Figure 1C). Prior to the onset of HFD feeding, there were no differences in body weight and blood glucose levels 95 between AAV-Rap1 V12 and their control mice (Supplemental Figure 1D and E). During HFD feeding, both AAV-96 Rap1 V12 and control mice were similar in body weight ( Figure 1A). However, blood glucose was markedly elevated 97 in AAV-Rap1 V12 mice compared to control animals ( Figure 1B). Rap1-induced hyperglycemia was not observed 98 under normal chow-fed conditions, as demonstrated by the results that forced activation of hypothalamic Rap1 has 99 no impact on body weight and blood glucose in lean animals (Supplemental Figure 1F and G). These data 100 collectively suggest that increased Rap1 activity in the hypothalamus sufficiently aggravates diet-induced 101 hyperglycemia without an effect on body weight. 102

103
The glycemic effect might be due to the pharmacological effects of overexpression of a constitutively active form 104 of Rap1 in the medial hypothalamus, and we next sought to investigate the physiological relevance of hypothalamic 105 Rap1. To do this, we decreased Rap1 within the hypothalamus using bilateral injection of AAV expressing Cre 106 recombinase into the medial hypothalamus aimed at the VMH of Rap1a and Rap1b double floxed mice (47) 107 (Rap1 ∆HYP ). For the control, we injected GFP-expressing AAVs into the medial hypothalamus (control). GFP 108 fluorescence was found in the VMH and a lesser extent in the DMH but not the ARC (Supplemental Figure 1H). 109 Consistently, Rap1a and Rap1b mRNAs were significantly reduced in the VMH of AAV-Cre injected mice 110 (Supplemental Figure 1I). While blood glucose levels of control mice were significantly elevated over the course 111 of HFD feeding, Rap1 ∆HYP mice did not show increased blood glucose in response to HFD feeding ( Figure 1D), 112 suggesting that genetic deletion of hypothalamic Rap1 prevents HFD-induced hyperglycemia. Interestingly, body 113 weight was not altered by hypothalamic deletion of Rap1 ( Figure 1C). Thus, AAV-mediated deletion of Rap1 genes 114 restricted to the hypothalamus protects mice against HFD-induced mild hyperglycemia independent of body weight. 115 These loss-of-function and gain-of-function data collectively highlight the primary role of hypothalamic Rap1 in 116 mediating hyperglycemia during HFD conditions. 117 118

Production and validation of SF1-specific Rap1 deficient mice 119
While Rap1 in the forebrain (42) and in the hypothalamus (Figure 1 in this study) plays a critical role in glycemic 120 regulation, the site of the effect remains to be established. We assume that VMH neurons mediate Rap1-dependent 121 glycemic control on the basis of the following observations: (1) the VMH is a well-established hypothalamic 122 nucleus including multiple distinct neural populations that influence whole-body glucose metabolism (20,33,48,123 49); (2) Rap1 is produced in the VMH ((42) and Figure 2A); (3) in Rap1 ∆CNS mice that have improved glucose 124 balance, Rap1 is depleted in the VMH (42); and (4) AAV-mediated manipulation of Rap1 occurs mostly in the 125 VMH but not in the arcuate nucleus (ARC) (Supplemental Figure 1B and 1H). Thus, we next sought to determine 126 whether Rap1 participates in VMH-mediated glycemic control. To this end, we generated mice lacking Rap1 127 specifically in VMH neurons, referred to herein as Rap1 ∆SF1 mice, by breeding Rap1a and Rap1b double floxed 128 mice (47) to the steroidogenic factor-1 (SF1) Cre line that expresses Cre recombinase only in SF1 neurons in the 129 brain (23). VMH SF1 neurons are a critical subset of VMH neurons for the control of leptin actions, energy, and 130 glucose homeostasis. We first confirmed Cre-mediated excision of the floxed Rap1 alleles in the hypothalamus 131 including the VMH (Supplemental Figure 2A). Immunohistochemical analyses for endogenous Rap1 exhibited 132 selective depletion of Rap1 protein in the VMH but not in the adjacent arcuate nucleus (ARC) (Figure 2A), which 133 was further confirmed by western blot analysis of the Rap1 protein ( Figure 2B and Supplemental Figure 3A). We 134 further confirmed that Rap1 was markedly reduced in SF1 positive cells in the VMH (Supplemental Figure 2B). In 135 addition, the vast majority of SF1 cells were concomitantly labeled with NeuN, a marker of postmitotic neurons 136 (50) (Supplementary Figure 2C), suggesting neuron-specific deletion of Rap1 in the VMH. Altogether, our results 137 demonstrate that Rap1 is produced in the VMH and that its deletion is restricted to SF1 neurons in Rap1 △ SF1 mice. 138 139 Improved glucose balance and peripheral insulin sensitivity in Rap1 ∆SF1 mice 140 Using Rap1 ∆SF1 mice, we directly examined whether Rap1 in the VMH has a role in systemic glucose balance. 141 Under normal chow condition, Rap1 ∆SF1 mice exhibited a significant decrease in blood glucose in both the fed and 142 fasted states ( Figure 2C) and lower serum insulin levels but not serum glucagon (Supplemental Figure 4). In 143 agreement with lower glycemia, Rap1 ∆SF1 animals had markedly improved glucose and insulin tolerance compared 144 to weight-and age-matched littermate controls ( Figure 2D and 2E). We further examined the effect of SF1 cell-145 specific Rap1 deletion on HFD-induced diabetic-like conditions. HFD-fed Rap1 ∆SF1 mice showed significantly 146 reduced blood glucose levels ( Figure 2F), improved glucose tolerance ( Figure 2G), and enhanced insulin sensitivity 147 ( Figure 2H) compared to control littermates under HFD conditions. Improved glucose balance was similarly 148 observed in HFD-fed female Rap1 △SF1 mice (Supplementary Figure 5B-D). Thus, these data suggest that SF1 cell-149 specific Rap1 deficiency protects mice from HFD-induced insulin resistance, hyperglycemia, and impaired glucose 150 tolerance. 151

152
To further characterize the mechanisms underlying the improved glucose and insulin homeostasis in Rap1 ΔSF1 mice, 153 hyperinsulinemic-euglycemic clamp studies were performed in HFD-fed Rap1 ∆SF1 and littermate control mice. 154 Compared to control animals, Rap1 ∆SF1 mice displayed a significantly higher glucose infusion rate (GIR) to maintain 155 euglycemia ( Figure 3A), indicating that Rap1 in the VMH neurons controls whole-body insulin actions. This 156 increase in GIR in Rap1 ∆SF1 mice seems to be due to an increase in insulin-induced peripheral glucose disposal 157 ( Figure 3B). Consistently, 2-deoxy-D-glucose (2DG) uptake in muscles and adipose tissue was significantly 158 increased in SF1 neuron-specific Rap1-deficient mice ( Figure 3C). As shown in Figure 3D, cellular insulin 159 sensitivity was also enhanced in the muscle of Rap1 ∆SF1 mice compared to littermate control mice, which is clearly 160 demonstrated by enhanced insulin-induced phosphorylation of AKT, a critical mediator of cellular insulin signaling 161 in Rap1 ∆SF1 mice. 162

163
In addition to Rap1 deletion in the VMH in the CNS, Rap1 was ablated in adrenal glands (Supplementary Figure  164 2A and 3B) as expected. Loss of Rap1 in the adrenal glands is unlikely to account for the glucose phenotype of 165 Rap1 △SF1 mice. First, basal and stress-induced corticosterone levels were similar between control and Rap1 △SF1 166 mice (Supplemental Figure 3D), indicating preserved adrenal function. Furthermore, the adrenal glands of Rap1 △ 167 SF1 mice were histologically indistinguishable from those of control mice (Supplemental Figure 3E). More 168 importantly, AAV-mediated deletion of Rap1 only in the VMH and its surrounding areas sufficiently induced the 169 glucose phenotype without affecting peripheral Rap1 ( Figure 1). These data indicate that the observed phenotype 170 in these mice is highly likely to be due to the absence of Rap1 in VMH SF1 neurons. 171 172 Energy balance and leptin responsiveness in Rap1 ∆SF1 mice 173 As forebrain-specific Rap1 knockout mice display a lean phenotype (42), it is possible that the remarkable glycemic 174 effects we observed in Rap1 ∆SF1 mice might be due to the potential confounding effects of Rap1 deficiency on body 175 weight or adiposity. We thus explored the effect of Rap1 in VMH SF1 neurons on energy balance by measuring the 176 body weight and adiposity of Rap1 ∆SF1 and their littermate control mice. We found no differences in body weight 177 and adiposity between the two groups irrespective of age, diet or sex ( Figure 4A-D, 5A-C and Supplemental Figure  178 5A). Despite an apparent lack of effect on body weight, Rap1 ∆SF1 mice exhibited increased food intake ( Figure 5D). 179 Concomitantly, parameters pertaining to energy expenditure were significantly increased in the mice with SF1 cell-180 specific Rap1 deletion compared with littermate controls by DEXA analysis ( Figure 5E-J). These data suggest that 181 Rap1 deletion in SF1 cells has no impact on body weight, which is likely to be accounted for by a simultaneous 182 increase in both food intake and energy expenditure by Rap1 deficiency in SF1 neurons. 183 184 Since Rap1 is known to inhibit the cellular action of leptin (42, 44, 45), a critical hormone that maintains normal 185 body weight and normoglycemia, we examined the effect of VMH Rap1 on leptin responsiveness. By assessing 186 cellular and anorectic responses to exogenously administered leptin (3 mg/kg, i.p., twice a day), we found that 187 Rap1 ∆SF1 mice showed greater weight loss in response to leptin compared to the control animals ( Figure 6A). In 188 addition, leptin was more effective at suppressing food intake in Rap1 ∆SF1 mice than in their controls ( Figure 6B), 189 suggesting the enhanced leptin responsiveness in Rap1 ∆SF1 mice. Under HFD conditions, the response to exogenous 190 leptin was diminished in the control mice ( Figure 6C and D), as previously shown (51). Rap1 ∆SF1 mice retained their 191 ability to respond to exogenous leptin in terms of leptin-induced suppression of food intake and reduction in body 192 weight ( Figure 6C and D). This finding concurs with the increased leptin-dependent phosphorylation of STAT3, a 193 critical mediator for leptin's metabolic effects, in the VMH but not in the arcuate nucleus of Rap1 ∆SF1 mice ( Figure  194 6E and F). These results suggest that despite having no effect on body weight/adiposity, Rap1 deficiency sensitizes 195 VMH SF1 neurons to exogenous leptin action in vivo. 196 197 Acute pharmacological inhibition of brain EPAC2 improves glucose balance independent of leptin 198 We sought to assess the potential translational relevance of CNS Rap1 inhibition. To test this, we used ESI-05, a 199 well-established selective inhibitor of EPAC2 (Rehmann, 2013;Tsalkova et al., 2012). EPAC2 is a GTP/GDP 200 exchange factor for Rap1 that serves as a direct activator for Rap1 and is highly enriched in the brain (52). We have 201 previously demonstrated that ESI-05 successfully inhibited endogenous Rap1 activity in the hypothalamus when 202 ESI-05 was centrally infused at a dose of 0.2 nmol (42, 53). The same dose of ESI-05 was directly infused into the 203 brains of HFD-fed C57BL/6 mice, and we assessed the effect of ESI-05 on glucose metabolism. Centrally 204 administered ESI-05 alleviated mild hyperglycemia in HFD-fed mice, as demonstrated by the lowered fed and 205 fasted blood glucose levels ( Figure 7A) and improved glucose tolerance ( Figure 7B) in the HFD-fed mice treated 206 with i.c.v. ESI-05 compared to levels in the vehicle-injected control HFD-fed mice. We confirmed the Rap1 207 dependency of ESI-05's glycemic effect by demonstrating that the glucose-lowering effect of ESI-05 was not 208 observed in Rap1 ∆CNS mice that lack both Rap1a and Rap1b in the forebrain including the hypothalamus ( Figure  209 7C). We further found that ESI-05-induced glucose lowering was completely abolished in Rap1 △SF1 mice ( Figure  210 7D), suggesting that Rap1 in VMH SF1 neurons mediates glucose reducing effect of ESI-05. Collectively, the acute 211 pharmacological inhibition of Rap1 signaling improves HFD-induced disordered glucose balance, which further 212 supports the genetic evidence described above. 213 214

The glucose-lowering of ESI-05 is independent of leptin action 215
Rap1 deficiency in the forebrain (42) or in the VMH (this study) simultaneously enhances leptin responsiveness 216 and improves glucose balance. As leptin is known to improve glucose balance, we clarified the potential role of 217 leptin in the glucose lowering effect of ESI-05. Centrally administered ESI-05 significantly lowered blood glucose 218 levels ( Figure 7E) and improved glucose tolerance ( Figure 7F) in leptin-deficient obese mice (ob/ob mice), which 219 is comparable to the effect in leptin-intact dietary obese mice ( Figure 7A and B), suggesting that brain Rap1-220 mediated glycemic regulation occurs without leptin. Collectively, our data clearly demonstrate that acute inhibition 221 of Rap1 signaling in the brain improves glucose balance by a leptin-independent mechanism. 222 223 DISCUSSION 224

225
The CNS has emerged as an attractive target for diabetes intervention (1-6), although such an approach has not yet 226 fully materialized. Thus, it is of paramount interest to identify druggable targets within the CNS glucoregulatory 227 mechanism that would potentially create a novel therapeutic intervention to mitigate diabetic conditions. Our 228 findings reveal a molecular pathway in the hypothalamus that mediates whole-body glucose balance, as we 229 demonstrate that hypothalamic Rap1 is sufficient to produce alterations in whole-body glucose balance. Specifically, 230 activation of Rap1 in the hypothalamus exaggerated hyperglycemia in diet-induced obesity. In contrast, 231 hypothalamic loss of Rap1 decreased hyperglycemia in dietary obesity. The glycemic phenotype was observed 232 without altered body weight, suggesting the primary role of Rap1 in glucoregulatory function. In addition, we 233 provide a proof-of-concept for the potential of targeting Rap1 signaling within the CNS to improve glucose 234 imbalance and induce antidiabetic effects. Our data collectively suggest that hypothalamic Rap1 is a molecular 235 pathway for the control of glucose metabolism and mediates HFD-induced glucose imbalance, thereby making it a 236 potential target for therapeutics. 237

238
The remarkable glycemic effect of hypothalamic Rap1 could be related to its action within the VMH neurons. The 239 VMH is classically known as a critical brain site for the control of glucose metabolism (21,33,48,49,54). 240 Modulation of neural activity of the VMH neurons, in particular, SF1-positive neurons, has been shown to influence 241 glucose metabolism of peripheral tissues. Indeed, optogenetic activation of VMH SF1 neurons increased blood 242 glucose by inducing counterregulatory responses (33, 38), whereas chemogenetic activation of VMH SF1 neurons 243 enhanced peripheral insulin sensitivity and increased glucose disposal (34). In line with these previous studies, our 244 data show that genetic ablation of Rap1 from SF1 neurons significantly decreased blood glucose, remarkably 245 reduced serum insulin levels, and robustly improved glucose and insulin tolerance. As Rap1 expression remains 246 intact in the pancreas of Rap1 ΔSF1 mice (Supplementary Figure 2A and 3C), a marked reduction in insulin level is 247 likely due to the central effect of Rap1 deficiency. Furthermore, Rap1 ΔSF1 mice had an increased glucose infusion 248 rate under the insulin clamp condition, suggesting enhanced whole-body insulin sensitivity. Consistently, cellular 249 insulin signaling in skeletal muscle was significantly enhanced by Rap1 deficiency in VMH neurons. In addition, 250 enhanced glucose uptake was further observed in the skeletal muscle and fat of Rap1 ΔSF1 mice. Our findings, taken 251 together with the data demonstrating that Rap1 ΔSF1 mice had no significant impact on serum levels of hormones that 252 promote glucose mobilization, Rap1 deficiency in the VMH is likely to improve glucose balance by promoting 253 glucose disposal. 254 255 A single dose of insulin caused a greater fall in blood glucose in Rap1 ΔSF1 mice ( Figure 2E and 2H and 256 Supplementary Figure 5D). As insulin-induced hypoglycemia is usually limited by the counterregulatory 257 mechanism that is initiated in part by the brain including the VMH neurons, our data may indicate impaired 258 counterregulation to the insulin stimulus. Given that the VMH is the critical site for the counter-regulatory 259 mechanism (38, 55) and that Rap1 has emerged as a key regulator of glucose metabolism, it is possible that Rap1 260 in the VMH has a role in the counter-regulatory process. In addition, we found that ICV glucose-induced c-fos 261 induction was significantly attenuated in the VMH, but not in the ARC, of mice receiving ICV ESI-05 262 (Supplementary Figure 6). These results may indicate potential roles of VMH Rap1 signaling in glucose sensing 263 mechanisms and are worth future investigations. Our findings collectively highlight the importance of Rap1 in 264 mediating VMH-dependent glycemic regulation. 265 266 While the VMH is known to play a major role in glycemic control, this nucleus also mediates other metabolic 267 functions, including the regulation of energy balance (i.e., food intake, energy expenditure, adiposity and body 268 weight) (23,28,30,31,34,35,38). We examined the metabolic phenotypes of Rap1 deficiency in the VMH to 269 determine the potential specificity of Rap1 signaling for improving glucose balance and found that VMH Rap1 was 270 not involved in determining body weight irrespective of diet, sex and age, although it influenced both arms of energy 271 balance by increasing food intake and energy expenditure (O2 consumption, CO2 production, heat production, and 272 physical activities), thereby offsetting the effect on body weight. Rather, VMH Rap1 deletion decreased blood 273 glucose and enhanced cellular insulin signaling in peripheral tissues. Furthermore, Rap1 ΔSF1 mice showed increased 274 peripheral glucose utilization in HFD-induced obesity, suggesting the specific role of VMH Rap1 in glucoregulatory 275 functions. Interestingly, our previous study demonstrated that mice with Rap1 deficiency in the broader area of the 276 brain (using a Cre line expressing CaMKII-driven Cre recombinase with specific activity in the forebrain including 277 multiple hypothalamic nuclei) exhibited markedly improved leptin responsiveness, reduced body weight and 278 adiposity, and decreased food intake (42), all of which suggest the role of forebrain Rap1 in participating in energy 279 balance. Compared to the prior study (42), our findings define the physiologic functions of Rap1 in VMH SF1 280 neurons relative to the broader forebrain areas, as we propose that VMH Rap1 is a molecular pathway capable of 281 selectively modulating glucose metabolism without having an effect on body weight. 282

283
Supporting the specific role of Rap1 in the VMH for glycemic regulation, signaling molecules that modulate and 284 mediate Epac-Rap1 signaling also selectively exhibit their glucoregulatory role within VMH SF1 neurons. The 285 heterotrimeric G protein Gs (Gsα) leads to activation of Epac-Rap1 signaling by promoting cAMP production (56). 286 Interestingly, mice with Gsα deficiency in VMH SF1 neurons display phenotypes akin to those of VMH Rap1 287 knockout mice: enhanced leptin responsiveness and improved glucose balance without altering body weight (29). 288 Similarly, VMH SF1 neuron-specific deletion of SOCS3, a major downstream mediator of Epac-Rap1 signaling, 289 also exhibits the same phenotypes, such as increased leptin responsiveness and markedly improved glucose 290 homeostasis in the absence of an effect on body weight (24). These striking similarities in the phenotypes and the 291 fact that Rap1 is biochemically linked to these signaling molecules collectively suggest the critical role of the Rap1 292 pathway as an intracellular signaling modality in the VMH neurons capable of determining whole body glucose 293 balance. 294 295 Rap1 ∆SF1 mice showed markedly enhanced cellular and anorectic responses to leptin, suggesting that Rap1 296 deficiency sensitizes leptin action in VMH SF1 neurons in vivo. Because leptin increases peripheral glucose uptake 297 and corrects hyperglycemia when directly infused into the VMH (57, 58), it is possible that the effects observed in 298 Rap1 ∆SF1 mice are due to enhanced sensitivity of leptin. Although possible, this seems unlikely given that a similar 299 glycemic effect was produced in the absence of leptin. Furthermore, despite improved leptin responsiveness, Rap1 300 deficiency in the VMH did not have an effect on body weight regardless of sex, diet, or age. This is somewhat 301 surprising, considering that leptin is a major determinant of body weight. Nevertheless, while the precise role of 302 increased leptin action of Rap1 ∆SF1 mice is unknown, future studies are warranted to investigate a specific role of 303 leptin in mediating the metabolic phenotypes of Rap1 ∆SF1 mice. 304 305 Lastly, one important implication arising from this study is that targeting the CNS Rap1 pathway might have the 306 potential to yield antidiabetic effects, as our pharmacological study demonstrates that ESI-05 significantly 307 decreased fed and fasted blood glucose levels and improved glucose tolerance in HFD-induced obese mice. 308 Identification of existing and novel chemical compounds that can modulate hypothalamic Rap1 activity may offer 309 a novel therapeutic opportunity to improve type 2 diabetes. 310

311
In summary, our results offer genetic and pharmacological evidence suggesting that hypothalamic Rap1 signaling, 312 especially in the VMH, is a pivotal pathway for the control of glucose homeostasis. The findings also unveiled an 313 antidiabetic effect of Rap1 inhibition in diet-induced hyperglycemic mice. Thus, we propose that EPAC2-Rap1 314 signaling in the hypothalamus could serve as a potential molecular target for therapeutic interventions to mitigate 315 diabetic conditions. 316

Physiological measurements 337
Body weight was measured weekly. Blood samples were collected via the saphenous vein from 4-hour-fasted mice 338 without anesthesia. Serum was isolated after centrifugation (5000 x g for 10 min) at 4°C and stored at -80°C. Blood 339 glucose levels were determined in freshly withdrawn blood from the tail vain by using a One Touch Ultra Blood 340

Glucose Meter. Plasma insulin was analyzed with a Milliplex MAP Mouse Metabolic Hormone Magnetic Bead 341
Panel Kit. For glucose tolerance tests, food was removed at 5 p.m. to initiate a 18-hour fast. The following morning 342 at 11 a.m., mice were injected D-Glucose (1.5 g/kg), and blood glucose was measured at the indicated time periods 343 from tail vein. For insulin tolerance tests, food was removed at 9 a.m. for 4-hour-fasting. At 1 p.m., insulin (1 U/kg) 344 was injected intraperitoneally, and blood glucose was measured at the indicated periods from tail vain. Similarly, 345 we performed a glucose tolerance test for ESI-05 -treated mice. Intracerebroventricular (i.c.v.) surgery was carried 346 out on C57BL/6 mice fed a high-fat diet for 21 weeks, Rap1 ΔCNS mice or 12-week-old ob/ob mice. One week after 347 the i.c.v. surgery, the mice were injected with vehicle or ESI-05 (0.2 nmol/mouse) at 5 p.m. and food was removed 348 at a same time. The following morning at 9 a.m., the mice were injected again with vehicle or ESI-05. After 2 hours 349 of last bolus injection, a glucose tolerance test was performed. 350 351

Body composition and energy expenditure measurements 352
Whole body composition was measured using NMR imaging (EchoMRI). Body weight-and body composition-353 matched 9-month-old control and Rap1 ΔSF1 mice fed normal chow were used. Mice were first acclimatized to the 354 metabolic cages and housed individually for 3 days before measurements were taken. Metabolic parameters, 355 including O2 consumption, CO2 production, respiratory exchange ratio, heat production, ambulatory activity and 356 food intake, were determined by using the Columbus Instruments Comprehensive Lab Animal Monitoring System 357 (CLAMS). 358 359

Leptin sensitivity test 360
Mice were singly housed and acclimatized for 1 week prior to the study. Body weight-and adiposity-matched 361 normal chow-fed 24-week-old control and Rap1 ΔSF1 mice were injected intraperitoneally with vehicle (Dulbecco's 362 PBS, dPBS, Sigma-Aldrich, D8537) twice a day (5 p.m. and 9 a.m.) for 4 consecutive days. Three days after the 363 last vehicle treatment, mice were injected intraperitoneally with leptin (3 mg/kg, Harbor-UCLA Research and 364 Education Institute) twice a day for 4 consecutive days. Food intake and body weight were measured daily. Similarly, 365 we performed a leptin sensitivity test for HFD-fed mice. Control and Rap1 ΔSF1 mice were placed on a high-fat diet 366 for 20 weeks, and then the mice were injected with vehicle or leptin twice a day. Body weight and food intake were 367 measured daily. 368 369

Cannula implantation, AAV injection and treatments 370
Mice were anesthetized with isoflurane and positioned in a stereotaxic frame. For cannula implantation, the skull 371 was exposed, and a 26-gauge single stainless steel guide cannula (C315GS-5-SPC, Plastics One, Roanoke, VA, 372 USA) was implanted into the third cerebral ventricles (−0.45 mm from bregma, ±0.9 mm lateral, −2.5 mm from the 373 skull). The cannula was secured to the skull with screws and dental cement. After i.c.v. cannulation, the mice were 374 housed singly and given at least 1 week to recover. On experimental days, the mice were infused with 1 μL of each the hypothalamus, AAV-DJ-Rap1 V12 -GFP or AAV-DJ-GFP was injected into the hypothalamus of C57BL/6J mice 381 similarly as described above. The mice were given at least 1 week to recover. 382 383

Hyperinsulinemic-euglycemic clamps: 384
Hyperinsulinemic euglycemic clamp studies were performed at the Mouse Metabolomic and Phenotyping Core 385 (MMPC) at BCM as described previously (60, 61). Mice were anesthetized, and a midline neck incision was made 386 to expose the jugular vein. A microcannula was inserted into the jugular vein, threaded into the right atrium, and 387 anchored at the venotomy site. Mice were allowed to recover for 4 days with ad libitum access to water and food. 388 Following an overnight fast, the conscious mice received a primary infusion (10 uCi) and then a constant rate 389 intravenous infusion (0.1 uCi/min) of chromatography-purified [3-3H]-glucose using a syringe infusion pump. For 390 determination of basal glucose production, blood samples were collected after 50 and 60 min of labeled glucose 391 infusion. After 60 min, mice were infused continuously for 2 hours with human insulin (xx mU/kg/min). 392 Simultaneously, 25% glucose was infused using another infusion pump at a rate adjusted to maintain the blood 393 glucose level at 100-140 mg/dl (euglycemia). Blood glucose concentration was measured every 10 min by a 394 glucometer. Glucose production rate, peripheral glucose disposal rate, and glucose infusion rate were then 395 calculated. To estimate insulin-stimulated glucose uptake in individual tissues 2-[14C]-deoxyglucose (2DG) was 396 administered as a bolus (10 uCi) at 45 min before the end of the clamps. At the end mice were euthanized and 397 tissues were snap freeze using liquid nitrogen for tissue specific glucose uptake. Glucose uptake in different tissues as follows: after brief anesthetization with isoflurane, mice were decapitated, and the whole brain was removed. 406 Frontal sections of the hypothalamus were prepared using a brain matrix (1 mm-thick), and the VMH and ARC 407 were micro-dissected under a fluorescent stereomicroscope, frozen immediately in dry ice, and stored at 80°C. 408 Equal amounts of the samples were separated by SDS-PAGE and transferred to a nitrocellulose membrane by 409 electroblotting. The following primary antibodies were used for western blot assays: phospho-Akt antibody (1:1,000, 410 Cell Signaling Technology, 4060), Akt antibody (1:1,000, Cell Signaling Technology, 2920), Rap1 antibody 411 (1:1,000, Santa Cruz Biotechnology, sc-398755) and antibody against β-actin (1:1,000, Cell Signaling Technology, 412 4970S). After incubation with primary antibodies for 24-72 hours at 4°C, the membranes were incubated with the 413 following secondary antibodies conjugated to a fluorescent entity: IRDye 680RD goat anti-rabbit IgG (LI-COR 414 Biosciences, 926-68071) and/or IRDye 800CW goat anti-mouse IgG (LI-COR Biosciences, 926-32210) with gentle 415 agitation for 1 hour at room temperature. To measure the fluorescence intensity, the Odyssey IR imaging system 416 (LI-COR Biosciences) was used. 417 418

Statistics 448
The data are presented as the mean ± SEM. Statistical analyses were performed using GraphPad Prism for a two-449 tailed unpaired Student's t-test, or one-or two-way ANOVA followed by post hoc Tukey's or Bonferroni's tests. P 450 < 0.05 was considered to be statistically significant. 451 452

Study approval 453
All procedures to maintain and use the mice followed protocols reviewed and approved by the Animal Care and 454