Leptin receptor-expressing neuron Sh2b1 supports sympathetic nervous system and protects against obesity and metabolic disease

Leptin stimulates the sympathetic nervous system (SNS), energy expenditure, and weight loss; however, the underlying molecular mechanism remains elusive. Here, we uncover Sh2b1 in leptin receptor (LepR) neurons as a critical component of a SNS/brown adipose tissue (BAT)/thermogenesis axis. LepR neuron-specific deletion of Sh2b1 abrogates leptin-stimulated sympathetic nerve activation and impairs BAT thermogenic programs, leading to reduced core body temperature and cold intolerance. The adipose SNS degenerates progressively in mutant mice after 8 weeks of age. Adult-onset ablation of Sh2b1 in the mediobasal hypothalamus also impairs the SNS/BAT/thermogenesis axis; conversely, hypothalamic overexpression of human SH2B1 has the opposite effects. Mice with either LepR neuron-specific or adult-onset, hypothalamus-specific ablation of Sh2b1 develop obesity, insulin resistance, and liver steatosis. In contrast, hypothalamic overexpression of SH2B1 protects against high fat diet-induced obesity and metabolic syndromes. Our results unravel an unrecognized LepR neuron Sh2b1/SNS/BAT/thermogenesis axis that combats obesity and metabolic disease.

A dipose hormone leptin critically regulates body weight and metabolism, and disruption of leptin/leptin receptor (LepR) signaling results in morbid obesity and severe metabolic disease 1 . LepR is widely expressed in the hypothalamus, including the preoptic area (POA), lateral hypothalamus, dorsomedial hypothalamus (DMH), ventromedial hypothalamus, and arcuate nucleus (ARC) 2,3 . Leptin exerts its anti-obesity action by activating LepR signaling in hypothalamic energy balance circuits 4 , but it remains elusive whether leptin regulates energy expenditure vs energy intake by similar or discrete pathways. Leptin signaling is mediated by tyrosine kinase JAK2 that interacts with long-form LepRb 1 . Of note, a number of negative regulators of JAK2, including SOCS3, PTP1B, RPTPe, and TCPTP, have been reported to promote obesity [5][6][7][8][9][10][11][12] , supporting the notion that JAK2 inhibitory molecules increase risk for leptin resistance, obesity, and metabolic disease. Interestingly, we identified a JAK2-binding protein Sh2b1 as a potent positive regulator of JAK2 (refs. 13,14 ); however, JAK2/Sh2b1 pathways in LepRexpressing neurons has not been explored in vivo.
We report that neuron-specific restoration of Sh2b1 expression reverses the obesity phenotypes of Sh2b1-null mice 39 , indicating that neurons mediate Sh2b1 actions on body weight and metabolism. Given that Sh2b1 augments LepRb/JAK2 signaling in cell cultures, we postulate that Sh2b1 might cell-autonomously increase the ability of LepR neurons to control energy balance and body weight, perhaps by directly enhancing leptin signaling. In this study, we generate and characterize LepR cell-specific Sh2b1 knockout (Sh2b1 ΔLepR ) mice. Sh2b1 ΔLepR mice, like global Sh2b1 knockout mice, develop obesity, insulin resistance, and liver steatosis. Remarkably, Sh2b1 deficiency in LepR neurons abrogates the ability of leptin to stimulate sympathetic nerves innervating brown adipose tissue (BAT), leading to BAT dysfunction and reduced core body temperature in Sh2b1 ΔLepR mice. Collectively, our results unveil an unrecognized leptin/Sh2b1/ sympathetic nerve/adipose thermogenesis axis that combats obesity, type 2 diabetes, and liver steatosis.

Results
Sh2b1 ΔLepR mice spontaneously develop obesity. To determine the role of Sh2b1 in LepR neurons, we generated Sh2b1 ΔLepR mice (Sh2b1 f/f ;LepR-Cre +/+ ) by crossing Sh2b1 f/f mice with LepR-Cre drivers. LepR-Cre mice were characterized previously 40,41 . Mice were in a C57BL/6J background and fed a standard chow diet. Sh2b1 ΔLepR male and female mice progressively became heavier than sex/age-matched Sh2b1 f/f and LepR-Cre mice (Fig. 1a). Fat content was dramatically higher in Sh2b1 ΔLepR males and females relative to sex/age-matched Sh2b1 f/f and LepR-Cre mice (Fig. 1b). Both gonadal and inguinal white adipose tissue (WAT) depots were significantly larger in Sh2b1 ΔLepR relative to Sh2b1 f/f and LepR-Cre mice ( Supplementary Fig. 1a). Individual white adipocyte size was substantially larger in Sh2b1 ΔLepR than in LepR-Cre mice (Fig. 1c). Lean mass was not significantly different between Sh2b1 ΔLepR and LepR-Cre mice ( Supplementary Fig. 1b). To gain insight into the underlying mechanism, we analyzed energy balance. Food intake was relatively normal (Fig. 1d). O 2 consumption and CO 2 production (per mouse) were also not significantly different between Sh2b1 ΔLepR and Sh2b1 f/f mice (Fig. 1e, Supplementary Fig. 1c). Of note, O 2 consumption and CO 2 production, after normalization to body weight, were significantly lower in Sh2b1 ΔLepR males and females relative to sex/age-matched Sh2b1 f/f mice ( Supplementary Fig. 1d, e). Core body temperature was significantly lower in Sh2b1 ΔLepR males and females relative to sex/age-matched LepR-Cre or Sh2b1 f/f mice (Fig. 1f, g). We further confirmed that Sh2b1 ΔLepR mice had lower body temperature using E-Mitters, and Sh2b1 ΔLepR locomotor activity was relatively normal (Fig. 1h). These data indicate that Sh2b1 in LepR neurons is indispensable for the maintenance of both body weight and core body temperature.
Sh2b1 ΔLepR mice develop insulin resistance and liver steatosis. Obesity promotes type 2 diabetes and nonalcoholic fatty liver disease (NAFLD), prompting us to assess insulin sensitivity and hepatic lipid content. Sh2b1 ΔLepR males and females developed hyperglycemia and hyperinsulinemia compared to sex/age-matched LepR-Cre mice at 19-20 weeks of age (Fig. 2a). In glucose (GTT) or insulin (ITT) tolerance tests, blood glucose levels were markedly higher in Sh2b1 ΔLepR males and females relative to sex/ age-matched LepR-Cre or Sh2b1 f/f mice (Fig. 2b). Consistently, insulin-stimulated phosphorylation of Akt (pThr308 and pSer473) in liver and skeletal muscle was substantially lower in Sh2b1 ΔLepR than in LepR-Cre mice (Fig. 2c). Sh2b1 ΔLepR mice also developed severe liver steatosis, as demonstrated by markedly increased lipid droplet number and size and triacylglycerol (TAG) levels in the liver (Fig. 2d, e). These results suggest that Sh2b1 in LepR neurons combats against insulin resistance, type 2 diabetes, and NAFLD.
Adult-onset ablation of hypothalamic Sh2b1 results in obesity. We recently reported that neuronal Sh2b1 promotes brain development 16 . To distinguish between brain development-dependent and -independent actions of Sh2b1 on body weight and metabolism, we generated adult-onset, hypothalamus-specific Sh2b1 knockout mice by bilaterally microinjecting AAV1-hSyn-Cre vectors into the mediobasal hypothalami (MBH) of Sh2b1 f/f males at 10 weeks of age. AAV1-hSyn-green fluorescent protein (GFP) vectors were used as control. Bilateral MBH injections were histologically verified ( Supplementary Fig. 2a). MBH-specific ablation of Sh2b1 substantially increased body weight and fat content (Fig. 3a, b). As an additional control, AAV1-hSyn-Cre vectors were bilaterally injected into the MBH of wild-type C57BL/6 males. There was no difference in body weight and fat content between the AAV1-hSyn-Cre and AAV1-hSyn-GFP groups ( Supplementary Fig. 2b-c). O 2 consumption and CO 2 production (per mouse) were not significantly different between the AAV1-hSyn-Cre and AAV1-hSyn-GFP groups (Fig. 3c). Nonetheless, O 2 consumption and CO 2 production, after normalization to body weight, were significantly lower in AAV1-hSyn-Cre relative to AAV1-hSyn-GFP groups in the dark phase ( Supplementary Fig. 2d). Core body temperature was significantly lower in the AAV1-hSyn-Cre mice in the dark cycle (Fig. 3d). Notably, MBH-specific ablation of Sh2b1 significantly increased food intake (Fig. 3e). Like Sh2b1 ΔLepR mice, adult-onset and MBH-specific Sh2b1 knockout mice developed hyperinsulinemia, glucose intolerance, and insulin resistance (Fig. 3f, g).
Insulin-stimulated phosphorylation of hepatic Akt was lower in AAV1-hSyn-Cre-transduced than in AAV1-hSyn-GFP-transduced Sh2b1 f/f mice (Fig. 3h). MBH-specific Sh2b1 knockout mice also developed severe liver steatosis, as demonstrated by elevated levels of hepatocyte lipid droplets (Oil Red O staining of liver sections) and high levels of liver TAG (Fig. 3i). These data indicate that Sh2b1 in the MBH regulates body weight, metabolism, food intake, and/or body temperature independently of its action on brain development.
To determine whether MBH-specific overexpression of human SH2B1 protects against obesity, AAV9-CAG-SH2B1 or AAV9-CAG-GFP (control) vectors were bilaterally injected into the MBH of C57BL/6J males. Mice were fed an HFD to induce obesity. Recombinant SH2B1 was detected in AAV9-CAG-SH2B1-transduced but not AAV9-CAG-GFP-transduced mice ( Supplementary Fig. 3a, b). Body weight and fat content were significantly lower in the AAV9-CAG-SH2B1 group relative to the AAV9-CAG-GFP group (Fig. 4a, b). O 2 consumption and CO 2 production (per mouse) were not significantly different between the GFP and SH2B1 groups (Fig. 4c). Upon normalization to body weight, O 2 consumption and CO 2 production were significantly higher in the AAV9-CAG-SH2B1 group relative to the AAV9-CAG-GFP group ( Supplementary Fig. 3c). Core body temperature was significantly higher in the AAV9-CAG-SH2B1 group (Fig. 4d). MBH-specific overexpression of SH2B1 substantially ameliorated HFD-induced insulin resistance and glucose intolerance, as assessed by ITT, GTT, and insulin-stimulated phosphorylation of Akt (Fig. 4e, f). Overexpression of SH2B1 in the hypothalamus also blocked HFDinduced liver steatosis, as demonstrated by a marked reduction in lipid droplet number and size and liver TAG levels in the AAV9-CAG-SH2B1 relative to AAV9-CAG-GFP groups (Fig. 4g). These data further confirm that hypothalamic Sh2b1, perhaps in LepR neurons, protects against obesity, type 2 diabetes, and NAFLD in adult mice.
LepR neuron Sh2b1 is required for brown fat thermogenesis.
Given that BAT and beige fat promote adaptive thermogenesis, energy expenditure, and weight loss, we examined the impact of hypothalamic Sh2b1 on BAT activity. Ablation of Sh2b1 in either LepR neurons (Sh2b1 ΔLepR mice) or the MBH (AAV1-hSyn-Cretransduced Sh2b1 f/f mice) caused whitening of BAT (e.g. enlarged lipid droplets) and dramatic downregulation of uncoupling protein 1 (Ucp1) (Fig. 5a). Ucp1 protein and mRNA were barely detectable in Sh2b1 ΔLepR mice at 22 weeks of age (Fig. 5b, c). Ucp1 expression in inguinal WAT markedly decreased in Sh2b1 ΔLepR mice ( Supplementary Fig. 1g). Of note, absolute expression levels of Ucp1 was markedly higher in BAT than in WAT. Conversely, MBH-specific overexpression of SH2B1 reversed HFD-induced a b g Body weight (g) Weeks: Core temp (°C) whitening of BAT and downregulation of Ucp1 (Fig. 5a). Ucp1 protein and mRNA levels were substantially higher in AAV9-CAG-SH2B1-transduced relative to AAV9-CAG-GFP-transduced mice (Fig. 5b, c). Ucp1 mediates adaptive thermogenesis and energy expenditure 42 . Accordingly, Sh2b1 ΔLepR mice displayed markedly lower core body temperature compared to LepR-Cre mice upon cold exposure (on a chow diet), and mice with MBHspecific overexpression of SH2B1 (on a HFD) had the opposite effects (Fig. 5d). Thus, we uncovered an unrecognized hypothalamic Sh2b1/BAT axis that critically regulates adaptive thermogenesis and core body temperature.
LepR neuron Sh2b1 mediates leptin stimulation of the SNS.
Considering the pivotal role of the sympathetic nervous system (SNS) in BAT activation, we assessed the impact of Sh2b1 deficiency on the ability of leptin to stimulate sympathetic nerve transmissions in BAT. Both male and female Sh2b1 ΔLepR mice developed hyperleptinemia (Fig. 6a). Leptin stimulated phosphorylation of hypothalamic Stat3 to a lower degree in Sh2b1 ΔLepR relative to Sh2b1 f/f mice (Fig. 6b, c). Conversely, MBH-specific overexpression of SH2B1 augmented leptin-stimulated phosphorylation of hypothalamic Stat3 (Fig. 6b). To directly assess sympathetic nerve activity (SNA), we electrophysiologically recorded BAT SNA in Sh2b1 ΔLepR mice at both 6 weeks (prior to the onset of obesity) and 12 weeks of age. Baseline SNA was significantly lower in Sh2b1 ΔLepR relative to Sh2b1 f/f mice (Fig. 6d, e). In agreement with the previous reports 43 , central injection of leptin progressively and markedly increased SNA in wild type (i.e. Sh2b1 f/f ) mice (Fig. 6d, f). It is likely that multiple synaptic modifications and/or polysynaptic transmissions contribute to a delayed onset of the leptin action on SNA. Strikingly, deletion of Sh2b1 in LepR neurons completely abrogated the ability of leptin to stimulate SNA in Sh2b1 ΔLepR mice (Fig. 6f). These results indicate the Sh2b1 branch of LepR signaling pathways is required for leptin to stimulate the SNS. We next set out to examine neuronal activity (c-Fos expression as a surrogate marker) in the central sympathetic network, focusing on the POA, DMH, ARC, paraventricular hypothalamus (PVH), and rostral raphe pallidus (rRPa). These regions are known to control sympathetic outflows to BAT 44 . Cold exposure rapidly and robustly increased the number of c-Fos neurons in Sh2b1 f/f mice (Fig. 6g, h). Sh2b1 deficiency substantially suppressed cold-stimulated neuronal activation in the POA, DMH, and rRPa of Sh2b1 ΔLepR mice (Fig. 6g,   h). In the ARC, neural activity was significantly lower in Sh2b1 ΔLepR relative to Sh2b1 f/f mice at both 22 and 4°C (Fig. 6h).
To further confirm the role of LepR neuron Sh2b1 in regulating the SNS, we measured the levels of tyrosine hydroxylase (TH), a sympathetic nerve marker, in BAT. We previously validated that anti-TH antibody specifically recognizes TH in immunostaining 45 . Immunoreactivity to TH was dramatically lower in Sh2b1 ΔLepR relative to LepR-Cre mice (Fig. 5a). Likewise, TH levels in BAT were also markedly reduced by MBH-specific ablation of Sh2b1 (AAV-Cre vs AAV-GFP groups) (Fig. 5a). Conversely, MBHspecific overexpression of SH2B1 increased BAT TH levels (AAV-SH2B1 vs AAV-GFP) (Fig. 5a). Collectively, these results unveil an LepR neuron Sh2b1 preserves adipose SNS integrity in aging. Sh2b1 ΔLepR mice develop obesity in an age-dependent manner, prompting us to examine age-associated SNS degeneration. BAT TH levels were normal in young Sh2b1 ΔLepR mice prior to 8 weeks of age (Fig. 7a, b)   detectable in Sh2b1 ΔLepR mice at 22 weeks of age (Fig. 7a, b). To confirm sympathetic degeneration, we assessed the levels of class III β-tubulin, a neuronal marker, using antibody TUJ1. TUJ1 immunoreactivity in BAT was abundant in Sh2b1 f/f but not Sh2b1 ΔLepR mice at 22 weeks of age (Fig. 7c). Of note, Ucp1 downregulation followed the course of SNS deterioration in Sh2b1 ΔLepR mice (Fig. 7d). Next, we asked whether Sh2b1 deficiency in LepR neurons worsens SNS degeneration in white adipose tissue (WAT). Because sympathetic innervation of WAT is sparse and difficult to be detected 46,47 , we assessed phosphorylation of hormonesensitive lipase (HSL), a surrogate marker for SNS activation. HSL phosphorylation (pSer563 and pSer660) was normal or slightly higher in Sh2b1 ΔLepR mice at 3 weeks of age; thereafter, HSL phosphorylation decreased progressively and became barely detectable at 22 weeks of age (Fig. 7e). Collectively, these results suggest that Sh2b1 in LepR neurons is involved in preserving the SNS in both BAT and WAT during aging.
Deletion of POMC neuron Sh2b1 is unable to induce obesity. We next aimed to further map Sh2b1 target neurons. Brain sections were prepared from wild type and global Sh2b1 knockout (negative control) mice and stained with anti-Sh2b1 antibody. Sh2b1 was detected in hypothalamic cells in wild-type mice ( Supplementary Fig. 4a). To confirm these results, we generated Sh2b1-Cre knockin mice by inserting an IRES-eGFP-2A-Cre cassette into the Sh2b1 locus 3′ to the STOP codon ( Supplementary  Fig. 4b). GFP levels in Sh2b1-Cre mice were below detection thresholds. To facilitate detection of Sh2b1 neurons, Sh2b1-Cre drivers were crossed with Rosa-mTmG reporter mice to genetically label Sh2b1 neurons with mGFP in Sh2b1-Cre;Rosa-mTmG mice. We found that mGFP (a marker for expression of endogenous Sh2b1) was expressed in most of hypothalamic cells in Sh2b1-Cre;Rosa-mTmG but not Rosa-mTmG mice (Supplementary Fig. 4c). In line with these findings, Sh2b1 protein is detected in the entire brain by immunoblotting 16 . To confirm that proopiomelanocortin (POMC) and AgRP neurons express Sh2b1, hypothalamic sections were prepared from Sh2b1-Cre;mTmG mice and immunostained with antibodies to POMC and AgRP. Both POMC and AgRP neurons expressed mGFP (Supplementary Fig. 4d).

Discussion
We herein identify LepR neurons as key Sh2b1 targets that mediate Sh2b1 protection against obesity, type 2 diabetes, and NAFLD. We demonstrated that LepR neuron-specific deletion of Sh2b1, or adult-onset deletion of Sh2b1 in the hypothalamus (containing LepR neurons), resulted in severe obesity, insulin resistance, and liver steatosis. Conversely, MBH-specific overexpression of SH2B1 ameliorated HFD-induced obesity and metabolic syndromes. Leptin stimulation of the hypothalamic JAK2/Stat3 pathway was impaired in Sh2b1 ΔLepR mice, supporting the notion that Sh2b1 is an endogenous sensitizer for leptin action, perhaps by enhancing JAK2 activation. Remarkably, ablation of Sh2b1 in LepR neurons abrogated the ability of leptin to stimulate sympathetic nerves projecting to BAT. Likewise, adult-onset, MBH-specific ablation of Sh2b1 also impaired sympathetic transmissions in BAT. BAT became whitening and impaired in adaptive thermogenesis in both Sh2b1 ΔLepR mice and mice with MBH-specific ablation of Sh2b1, presumably owing to adipose SNS-deficits. Consequently, core body temperature was low and cold tolerance was impaired in both Sh2b1 ΔLepR mice and MBH-specific Sh2b1 knockout mice. These findings define LepR neuron Sh2b1 as a critical central regulator of thermogenesis and body temperature. Thus, we unveil an unrecognized leptin/LepR neuron Sh2b1/SNS/BAT/thermogenesis/body temperature axis. However, we cannot exclude the possibility that hypothermic Sh2b1 may increase thermogenesis and body temperature by an additional leptin-independent mechanism. For instance, Sh2b1 may enhance the ability of interleukin-6, a wellknown pyrogenic cytokine, to increase thermogenesis and body temperature through enhancing the JAK2/Stat3 pathway. Furthermore, hypothalamic Sh2b1 may increase body temperature by a SNS-independent mechanism, perhaps by enhancing the ability of hypothalamic-pituitary-thyroid axis to increase thermogenesis and body temperature. Given that the SNS/BAT pathway increases energy expenditure and weight loss, the LepR neuron Sh2b1/SNS/BAT/thermogenesis pathway is expected to mediate leptin stimulation of energy expenditure. Notably, food intake was relatively normal in Sh2b1 ΔLepR mice, raising the possibility that leptin may regulate food intake and energy expenditure by Sh2b1-independent and Sh2b1-dependent pathways, respectively.
Both aging and obesity are associated with impaired adipose sympathetic nerve transmissions [48][49][50] , but the underlying mechanism is poorly understood. We found that BAT sympathetic nerve fibers deteriorated age-dependently and became undetectable in Sh2b1 ΔLepR mice after 22 weeks of age. Following the course of sympathetic nerve degeneration, BAT progressively lost Ucp1 expression and thermogenic capability. Age-associated sympathetic nerve degeneration also occurred in WAT in Sh2b1 ΔLepR mice. Of note, BAT sympathetic innervation and transmission were normal in young Sh2b1 ΔLepR mice prior to 8 weeks of age. These results suggest that LepR neuron Sh2b1 is ΔLepR and Sh2b1 f/f males (6 weeks) were treated with leptin. Hypothalamic extracts were immunoblotted with antibodies against phospho-Stat3 (pTyr705) or Stat3. Stat3 phosphorylation was normalized to total Stat3 levels. Adult C57BL/6J males were bilaterally injected into the MBH with AAV9-CAG-SH2B1β or AAV9-CAG-GFP vectors, fed an HFD for 10 weeks, fasted overnight, and centrally injected with leptin for 15 min. Phosphorylation of hypothalamic Stat3 was assessed. c n = 3 per group. a.u. arbitrary units. d-f BAT sympathetic nerve activity (SNA). d Representative traces. e Baseline SNA at 6 (n = 3 mice per group) and 12 (n = 4 mice per group) weeks of age. f Time courses of leptin response. Leptin was centrally administrated into Sh2b1 ΔLepR (n = 3) and Sh2b1 f/f (n = 3) males at 6 weeks of age. g, h Sh2b1 ΔLepR and Sh2b1 f/f males (7 weeks) were exposed to 22 or 4°C for 4 h. Hypothalamic sections were stained with anti-Fos antibody. g Representative c-Fos staining of hypothalamic sections. Scale bar: 200 μm. h Number of c-Fos neurons in different hypothalamic areas. 4°C: n = 4 mice per group, 22°C: n = 3 mice per group. Data are presented as mean ± SEM. *p < 0.05, two-tailed unpaired Student's t-test (a, c, e, f) or two-way ANOVA (h). Source data are provided as a Source Data file.
NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-020-15328-3 ARTICLE NATURE COMMUNICATIONS | (2020) 11:1517 | https://doi.org/10.1038/s41467-020-15328-3 | www.nature.com/naturecommunications not required for adipose SNS development; rather, it plays an important role in preserving the adipose SNS against degeneration. Given that Sh2b1 is expressed broadly in the hypothalamus, it is not surprising that the neuronal activity of the central sympathetic network, particularly in the POA, DMH, and rRPa, is inhibited in Sh2b1 ΔLepR mice. These results suggest that hypothalamic Sh2b1 preserves adipose SNS integrity and sympathetic transmissions by a top-down mechanism, perhaps by enhancing leptin and/or other hormone and neuropeptide signaling in the central sympathetic network. We acknowledge that our data do not exclude the possibility that adipose SNS deterioration may be secondary to obesity in Sh2b1 ΔLepR mice. Additional studies are warranted to further characterize hypothalamic Sh2b1 circuits that protect against adipose SNS degeneration. The leptin resistance phenotype of Sh2b1 ΔLepR mice supports a concept that Sh2b1 is an endogenous enhancer of leptin sensitivity in vivo. Strikingly, ablation of Sh2b1 in LepR neurons abolished leptin stimulation of the adipose SNS, indicating that Sh2b1 mediates leptin actions on the SNS and energy expenditure. We acknowledge that we did not detect significant difference in energy expenditure (normalized to mice) between Sh2b1 ΔLepR and Sh2b1 f/f mice. We also did not observe the RER circadian rhythm. We speculate that assay-related stress and/or other factors may influence energy expenditure and the circadian rhythm, thereby masking difference between these two groups. Notably, the hypothalamic PI 3-kinase pathway was reported to mediate leptin stimulation of the SNS 51,52 . Sh2b1 mediates leptin stimulation of the PI 3-kinase pathway by recruiting IRS proteins to JAK2 13 , raising the possibility that the LepRb/Sh2b1/PI 3-kinase pathway may mediate leptin stimulation of the SNS/BAT/thermogenesis axis. However, we do not exclude the possibility that Sh2b1 in LepR neurons may promote SNS activity and maintenance by additional mechanisms. For instance, Sh2b1 enhances BDNF and insulin signaling in cell cultures [16][17][18]53 . BDNF and insulin, like leptin, also activate the SNS/BAT energy expenditure axis [16][17][18]54,55 . Sh2b1 may increase the ability of insulin and/or BDNF to stimulate energy expenditure. Of note, deletion of neither Sh2b2 nor Sh2b3 results in obesity 56,57 , suggesting that the Sh2b1 actions on energy balance and body weight cannot be replaced by other Sh2b family members.
We found that Sh2b1 was widely expressed in the entire brain, including hypothalamic POMC neurons and AgRP neurons. Sh2b1 ΔPOMC mice with POMC neuron-specific ablation of Sh2b1, unlike Sh2b1 ΔLepR mice, did not develop obesity and insulin resistance. These results are not unexpected, considering that ablation of LepR in POMC neurons, AgRP neurons, or both only slightly affects body weight and metabolism 58,59 . Ablation of LepR in Ghrh, Htr2c, or Prlh neurons also fails to cause obesity 60,61 . We postulate that Sh2b1 deficiency in POMC neurons may activate a compensatory mechanism in other Sh2b1 subpopulations, which masks the action of POMC neuron Sh2b1 on body weight and metabolism.
In conclusion, we unravel an unrecognized LepR neuron Sh2b1/SNS/adipose energy expenditure axis that combats against obesity, type 2 diabetes, and NAFLD. LepR neuron Sh2b1 mediates leptin stimulation of the SNS and supports preservation of adipose SNS against degeneration. The Sh2b1/SNS/fat axis may serve as a potential therapeutic target for the treatment of obesity and metabolic disease.

Methods
Animals. Sh2b1 f/f and LepR-Cre (knockin of Cre in the LepR 3′UTR) mice (C57BL/ 6 background) were characterized previously 40,62 . Sh2b1 f/f mice were crossed with LepR-Cre mice to generate Sh2b1 ΔLepR mice (Sh2b1 f/f ;Cre +/+ ). Because Cre levels in LepR-Cre +/− mice are insufficient to delete target genes 41 Stereotaxic microinjection. Mice were anesthetized with isoflurane and mounted on an Ultra Precise Small Animal Stereotaxic Alignment System (Model 963, David KOPF Instruments, Tujunga, CA). Body temperature was maintained within a normal range during the entire procedures using a thermal pad. The skull was exposed to identify the bregma and lambda, and a small opening in the skull was made using a bone drill at coordinates (mm): −1.5 (a-p), ±0.4 (m-l), and −5.8 (d-v). AAV vectors (0.5 µl) were injected into the hypothalamus in 10 min using UltraMicroPumps with SYS-Micro4 Controller (UMP3-2, World Precision Instruments Inc, Sarasota, FL). AAV1.hSyn.HI.eGFP-Cre and AAV1.hSyn.eGFP (control) were purchased from the Penn Vector Core, University of Pennsylvania School of Medicine. Cre expression is under the control of the neuron-specific human synapsin promoter. We prepared AAV9-CAG-SH2B1 and AAV9-CAG-GFP (control) vectors in which SH2B1β expression is under the control of the constitutively active synthetic CAG promoter.
Body core temperature and locomotor activity. Mice were anesthetized with isoflurane, and a G2 E-Mitter (870-0010-01, Starr life Sciences Corp, Oakmont, PA) was surgically implanted in the abdominal cavity. Locomotor activity and core body temperature were monitored using ER-4000 Energizer/Receiver (Bend, OR). Data were analyzed using Vitalview software (Starr life Sciences Corp, Oakmont, PA).
Plasma insulin and leptin, GTT, and ITT. Blood samples were collected from tail veins. Plasma insulin and leptin were measured using insulin and leptin ELISA kits (CRYSTAL CHEM, Downers Grove, IL), respectively. For GTT, mice were fasted for 16 h and intraperitoneally injected with glucose (2 g/kg body weight), and blood glucose was measured in 0, 15, 30, 60, and 120 min after injection. For ITT, mice were fasted for 6 h and intraperitoneally injected with insulin (0.7 U/kg for Sh2b1 ΔLepR mice, 0.6 U/kg for AAV-Cre or AAV-SH2B1β transduced mice), and blood glucose was measured in 0, 15, 30, and 60 min after injection.
Fat content and energy expenditure. Fat content and lean body mass were measured using a dual-energy X-ray absorptiometry pDexa (Norland Stratec). Heads, neck, and tails were not included in the pDexa analysis. Energy expenditure was measured by indirect calorimetry (Windows Oxymax Equal Flow System, Columbus Instruments, Columbus, OH).
Body composition and liver TAG levels. Mice were euthanized and organs were harvested and weighted. Liver samples were homogenized in 1% acetic acid and extracted using chloroform:methanol (2:1). The organic phase was dried via evaporation and dissolved in isopropanol. TAG levels were measured using a TAG assay kit (Pointe Scientific Inc., Canton, MI) and normalized to liver weight.
Immunostaining. Brain and BAT sections were prepared using a Leica cryostat (Leica Biosystems Nussloch GmbH, Nussloch, Germany), and immunostained with the indicated antibodies (Supplemental Table 1). Images were visualized using a BX51 Microscope (Olympus, Tokyo, Japan) and a DP72 digital camera (Olympus, Tokyo, Japan).
Electrophysiological recordings. Mice (5 or 11 weeks) were implanted with intracerebroventricular (icv) cannula and subjected to recordings after a 1-week recover period. Mice were anesthetized with ketamine (29 mg/kg body weight, intramuscular). BAT sympathetic nerve fibers were recorded using AxoScope 10.2 (San Jose, CA). Leptin (0.5 μl/mouse, 1 mg/ml) was injected via icv cannula. Nerve activity was amplified (NL104), filtered (NL 125/126, Neurolog, Digitimer Ltd), passed 100-1000 Hz, digitized (CED 1401, Cambridge Electronic Design, Cambridge, UK), and analyzed offline using Spike2 software (Cambridge Electronic Design, Cambridge, UK). To quantify SNA, the number of action potentials crossing a pre-set threshold was determined per second. The threshold was calculated at twice the baseline nerve activity for all experiments.
Statistical analysis. Data were presented as means ± SEM. Differences between two groups were analyzed with two-tailed Student's t-test, and differences between more than two groups were analyzed using one-way and two-way analysis of variance (ANOVA) and Bonferroni posttest using GraphPad Prism 7. A P value less than 0.05 was considered significant.
Reporting summary. Further information on research design is available in the Nature Research Reporting Summary linked to this article.