LepRb+ cell–specific deletion of Slug mitigates obesity and nonalcoholic fatty liver disease in mice

Leptin exerts its biological actions by activating the long-form leptin receptor (LepRb). LepRb signaling impairment and leptin resistance are believed to cause obesity. The transcription factor Slug — also known as Snai2 — recruits epigenetic modifiers and regulates gene expression by an epigenetic mechanism; however, its epigenetic action has not been explored in leptin resistance. Here, we uncover a proobesity function of neuronal Slug. Hypothalamic Slug was upregulated in obese mice. LepRb+ cell–specific Slug-knockout (SlugΔLepRb) mice were resistant to diet-induced obesity, type 2 diabetes, and liver steatosis and experienced decreased food intake and increased fat thermogenesis. Leptin stimulated hypothalamic Stat3 phosphorylation and weight loss to a markedly higher level in SlugΔLepRb than in Slugfl/fl mice, even before their body weight divergence. Conversely, hypothalamic LepRb+ neuron–specific overexpression of Slug, mediated by AAV-hSyn-DIO-Slug transduction, induced leptin resistance, obesity, and metabolic disorders in mice on a chow diet. At the genomic level, Slug bound to and repressed the LepRb promoter, thereby inhibiting LepRb transcription. Consistently, Slug deficiency decreased methylation of LepRb promoter H3K27, a repressive epigenetic mark, and increased LepRb mRNA levels in the hypothalamus. Collectively, these results unravel what we believe to be a previously unrecognized hypothalamic neuronal Slug/epigenetic reprogramming/leptin resistance axis that promotes energy imbalance, obesity, and metabolic disease.


Introduction
Hypothalamic neural circuits play an essential role in the control of energy balance, body weight, and metabolic homeostasis. GWAS reveal that approximately 95% of human obesity-associated genes and pathways are related to the CNS (1), indicating that hypothalamus and brain dysfunctions are a primary risk factor for obesity and metabolic disease. Leptin is secreted from adipose tissues to relay information about peripheral energy storage and availability to the hypothalamus, and it promotes weight loss by decreasing food intake (2). It also stimulates a sympathetic nerve and fat thermogenesis axis to increase energy expenditure (3,4). Leptin is believed to exert its metabolic action by activating the long-form leptin receptor LepRb in the hypothalamus (5). In addition to suppressing feeding behavior, hypothalamic LepRb signaling, enhanced by Sh2b1, increases sympathetic nerve outflows to brown adipose tissue (BAT) to promote fat thermogenesis and energy expenditure (3). Impaired leptin action, referred to as leptin resistance, is an important risk for obesity, and leptin resistance impedes leptin therapy to combat obesity and related metabolic disease (2,4). Hence, molecular mechanisms underlying leptin resistance have gained increased attention. Both negative and positive regulators of LepRb signaling have been identified, and negative and positive regulator imbalances have been proposed to drive leptin resistance (4,6). Hypothalamic LepRb is downregulated in rodents with high fat diet-induced (HFD-induced) obesity, exacerbating leptin resistance (7)(8)(9). Of note, leptin resistance is persistent in diet-induced obesity, raising the possibility that epigenetic reprogramming may be a causal factor for leptin resistance. However, epigenetic modifications are poorly understood in LepRb neural circuits.
Transcription factor Slug, also called Snail2 or Snai2, is a Snail family member -along with Snail1 and Snail3 -that controls gene expression by an epigenetic mechanism (10). Slug binds to target enhancers and promoters -E2 boxes: CACCTG or CAGGTG -via its C-terminal zinc finger domains, while its Nterminal SNAG domain recruits histone deacetylase 1 (HDAC1), HDAC2, lysine-specific demethylase 1 (LSD1), G9a, and/or enhancer of zeste homologue 2 (EZH2) to catalyze histone modifications on target chromatins (10)(11)(12). Slug and Snail1 have been well known to promote epithelial-mesenchymal transition (EMT) by epigenetically suppressing E-cadherin expression (10)(11)(12). Hepatic Slug promotes liver steatosis and nonalcoholic fatty liver disease (NAFLD) by epigenetically activating lipogenic genes (13). In this work, we report that Slug is expressed in a subset of hypothalamic neurons and is upregulated in obesity. LepRb cell-specific deletion of Slug protects against diet-induced leptin resistance, Leptin exerts its biological actions by activating the long-form leptin receptor (LepRb). LepRb signaling impairment and leptin resistance are believed to cause obesity. The transcription factor Slug -also known as Snai2 -recruits epigenetic modifiers and regulates gene expression by an epigenetic mechanism; however, its epigenetic action has not been explored in leptin resistance. Here, we uncover a proobesity function of neuronal Slug. Hypothalamic Slug was upregulated in obese mice. LepRb + cell-specific Slug-knockout (Slug ΔLepRb ) mice were resistant to diet-induced obesity, type 2 diabetes, and liver steatosis and experienced decreased food intake and increased fat thermogenesis. Leptin stimulated hypothalamic Stat3 phosphorylation and weight loss to a markedly higher level in Slug ΔLepRb than in Slug fl/ fl mice, even before their body weight divergence. Conversely, hypothalamic LepRb + neuron-specific overexpression of Slug, mediated by AAV-hSyn-DIO-Slug transduction, induced leptin resistance, obesity, and metabolic disorders in mice on a chow diet. At the genomic level, Slug bound to and repressed the LepRb promoter, thereby inhibiting LepRb transcription. Consistently, Slug deficiency decreased methylation of LepRb promoter H3K27, a repressive epigenetic mark, and increased LepRb mRNA levels in the hypothalamus. Collectively, these results unravel what we believe to be a previously unrecognized hypothalamic neuronal Slug/epigenetic reprogramming/leptin resistance axis that promotes energy imbalance, obesity, and metabolic disease. LepRb + cell-specific deletion of Slug mitigates obesity and nonalcoholic fatty liver disease in mice Slug sequences in Slug ΔLepRb mice. We detected hypothalamic Slug + LepRb + double-positive neurons in Slug fl/fl but not Slug ΔLepRb mice at 8 weeks of age on chow diet (Supplemental Figure 2A). Slug + LepRb + neuron number was underestimated because the Slug probes were unable to detect Slug + neurons expressing Slug at low levels. Additionally, Slug + LepRb + neurons were expected to be increased in obesity, given that HFD feeding increases hypothalamic Slug + neuron number. We placed Slug ΔLepRb , Slug fl/fl , and LepRb-Cre mice on a HFD. Body weight was indistinguishable between Slug fl/fl and LepRb-Cre mice (Figure 2A), so Slug fl/fl mice were used as a control in the following experiments. Both male and female Slug ΔLepRb mice were markedly resistant to HFDinduced obesity, and their body weights were significantly lower compared with sex-and age-matched Slug fl/fl mice ( Figure 2A). Whole-body fat content was substantially lower in Slug ΔLepRb than in Slug fl/fl mice on a HFD for 12 weeks ( Figure 2B). Lean mass was comparable in Slug ΔLepRb and Slug fl/fl mice (Supplemental Figure  2B). Individual white adipocyte size was smaller in Slug ΔLepRb than in Slug fl/fl mice ( Figure 2C). Inguinal white adipose tissue (iWAT), BAT, and liver weights were significantly lower in Slug ΔLepRb than in Slug fl/fl males on a HFD for 13 weeks ( Figure 2D). Likewise, iWAT, gonadal WAT (gWAT), BAT, and liver weights were also markedly lower in Slug ΔLepRb than in Slug fl/fl females ( Figure 2D). We also examined Slug ΔLepRb mice on a normal chow diet. Body weight was comparable between Slug ΔLepRb and Slug fl/fl mice at 8 weeks of age, but thereafter, Slug ΔLepRb mice progressively gained less body weight and, after 12 weeks of age, became significantly lighter than Slug ΔLepRb mice (Supplemental Figure 2C). Epididymal WAT (eWAT) weight was significantly lower in Slug ΔLepRb than in Slug fl/fl mice at 13 weeks of age (Supplemental Figure 2D). These results demonstrate, for the first time to our knowledge, that LepRb + cell-specific Slug deficiency protects against both diet-induced and age-associated obesity.
LepRb + cell-specific ablation of Slug protects against HFD-induced type 2 diabetes and NAFLD. We placed Slug ΔLepRb and Slug fl/fl mice on a HFD and assessed their glucose metabolism. Overnight-fasted plasma insulin levels were significantly lower in Slug ΔLepRb than in Slug fl/fl male mice on a HFD for 10 weeks ( Figure 3A). Glucose tolerance tests (GTT) and insulin tolerance tests (ITT) were performed in male mice on a HFD for 13 weeks. Blood glucose levels were significantly lower in Slug ΔLepRb mice relative to Slug fl/fl mice after glucose and insulin injections ( Figure 3B). Similarly, Slug ΔLepRb female mice also displayed reduced plasma insulin levels after being fed a HFD for 10 weeks and improved glucose and insulin tolerances after being fed a HFD for 12 weeks, compared with Slug fl/fl females ( Figure 3, C and D). In line with these results, insulin stimulation increased liver Akt phosphorylation to a significantly higher level in Slug ΔLepRb mice relative to Slug fl/fl mice on a HFD for 13 weeks ( Figure 3E). Akt phosphorylation was low in Slug fl/fl mice due to HFD-induced insulin resistance. Liver weight ( Figure 2D) and liver triacylglycerol (TAG) levels ( Figure 3F) were significantly lower in Slug ΔLepRb than in Slug fl/fl mice on a HFD for 13 weeks, for both males and females. Hepatocyte lipid droplets were smaller and less abundant in Slug ΔLepRb mice relative to Slug fl/fl mice, as revealed by H&E and Nile red staining of liver sections ( Figure 3G and Supplemental Figure 2E). We also performed GTT and ITT on chow-fed Slug ΔLepRb and Slug fl/fl mice at 8 weeks of age, when obesity, type 2 diabetes, and NAFLD. Conversely, mediobasal hypothalamus (MBH) LepRb neuron-specific overexpression of Slug has the opposite effects. At the molecular level, Slug binds to the LepRb promoter and induces repressive histone methylations, thereby suppressing LepRb expression. These observations suggest hypothalamic LepRb-neuron Slug as a previously unrecognized epigenetic inducer of leptin resistance and obesity.

Results
Slug is expressed in a subset of hypothalamic neurons and upregulated in obesity. To explore Slug in the brain, we mapped Slug neuron distributions in the hypothalamus. Since anti-Slug antibodies were unable to detect endogenous Slug by immunostaining, we exploited Slug-LacZ reporter mice (Slug LacZ ) in which a β-galactosidase (β-gal) transgene was inserted into the Slug locus under the control of the Slug promoter (14). Slug-expressing cells (β-gal + ) can be readily detected in heterozygous Slug LacZ/+ mice by X-gal or anti-β-gal-antibody staining (14,15). We detected abundant β-gal + Slug-expressing cells in the dorsomedial hypothalamus (DMH), ventromedial hypothalamus (VMH), and arcuate nucleus (ARC) ( Figure 1A). By contrast, Slug-expressing cells were barely detectable in the cerebral cortex, hippocampus, and cerebellum (Supplemental Figure 1A; supplemental material available online with this article; https://doi.org/10.1172/JCI156722DS1). Hypothalamic Slug-expressing cells coexpressed the neuronal marker NeuN but not the astrocyte marker glial fibrillary acidic protein (GFAP) (Figure 1, B and C). In the VMH, approximately 99% of Slug-expressing cells were neurons, and Slug + neurons accounted for approximately 50% of the total neurons ( Figure 1D).
To test if Slug expression was influenced by nutritional states and body weight, we placed C57BL/6J male mice on a HFD for 15 weeks. Hypothalamic Slug mRNA levels were significantly higher in mice fed a HFD than in chow-fed mice ( Figure 1E), and were also significantly higher in ob/ob mice compared with agematched WT mice ( Figure 1F). In contrast, HFD feeding did not increase Slug mRNA abundance in the cerebral cortex and cerebellum (Supplemental Figure 1B). To extend these findings, we counted Slug + neurons in the hypothalamus. We placed Slug LacZ/+ mice on a HFD for 3 weeks and immunostained hypothalamic sections with anti-β-gal antibody. Slug + neuron number was significantly higher in the VMH and DMH of HFD-fed mice compared with chow-fed mice (Figure 1, G and H). Given that de novo neurogenesis is rare, the newly generated Slug + neurons in HFD-fed mice may have arisen from neurons that were originally Slug --a Slugto Slug + phenotype switch or transdifferentiation.
LepRb + cell-specific ablation of Slug protects against diet-induced obesity. We set out to investigate hypothalamic Slug functions by generating and characterizing conditional Slug knockout mice. Considering the essential role of LepRb + neurons in the control of body weight, we generated LepRb + cell-specific Slug knockout (Slug ΔLepRb ) mice by crossing Slug fl/fl mice with Lep-Rb-Cre mice. Slug fl/fl mice and LepRb-Cre mice -with Cre knockin at the LepRb locus -were described previously (13,16). In Lep-Rb-Cre mice, Cre is expressed in cells expressing LepRb but not in cells expressing short forms of leptin receptors (16). We validated Slug ΔLepRb mice using RNAscope assays. Slug probes were designed to hybridize mRNA fragments encoded by the deleted higher in Slug ΔLepRb than in Slug fl/fl mice in the dark phase ( Figure  4B). We assessed energy expenditure using metabolic cages. O 2 consumption and CO 2 production, normalized to lean mass, were significantly higher in Slug ΔLepRb mice relative to Slug fl/fl mice in both males and females, particularly in the dark phase ( Figure 4, C-E). However, ANCOVA calculations did not reveal a significant difference between the 2 groups, possibly due to a low animal number limiting ANCOVA statistical power under this condition.
BAT and beige fat protect against obesity by increasing energy expenditure (17), prompting us to examine adipose thermogenic programs in Slug ΔLepRb mice. We placed Slug fl/fl and Slug ΔLepRb mice their body weights were comparable. GTT and ITT were indistinguishable between Slug ΔLepRb and Slug fl/fl mice (Supplemental Figure  2F). Thus, LepRb + cell-specific ablation of Slug protected against diet-induced obesity, type 2 diabetes, and NAFLD.
LepRb + cell-specific ablation of Slug decreases food intake and increases adipose thermogenesis. We postulated that Slug ΔLepRb mice might resist obesity through decreasing their food intake, increasing their energy expenditure, or both. Supporting this notion, we found that food intake was substantially lower in Slug ΔLepRb males and females relative to sex-and age-matched Slug fl/fl mice ( Figure 4A). Body temperature, an energy expenditure index, was significantly

(B and C)
Representative hypothalamic images (n = 3 mice per group). Hypothalamic sections were prepared from Slug LacZ/+ mice and coimmunostained with antibodies to β-gal, NeuN, and GFAP as indicated. (D) VMH cell subpopulations were counted (n = 3 mice). (E) Hypothalamic Slug mRNA levels (normalized to 36B4 expression) in C57BL/6J male mice on a HFD for 15 weeks (n = 8 mice per group). a.u., arbitrary units. (F) Hypothalamic Slug mRNA levels in WT and ob/ob male mice at 14 weeks of age (n = 8 mice per group). (G and H) Hypothalamic sections were prepared from Slug LacZ/+ male mice (HFD for 3 weeks) and stained with anti-β-gal antibody. β-gal neurons in the VMH, DMH, and ARC were counted (n = 3 mice per group). Data are presented as mean ± SEM. *P < 0.05, 2-tailed unpaired Student's t test.
against the sympathetic nerve marker tyrosine hydroxylase (TH), were significantly higher in Slug ΔLepRb than in Slug fl/fl mice ( Figure 5, A and B). Slug ΔLepRb mice displayed increased recruitment of beige adipocytes; iWAT expression of beige adipocyte markers -Ucp1, Pgc1α, and PPARγ -was significantly higher in Slug ΔLepRb than in Slug fl/fl mice after 5 days of cold exposure ( Figure 5D). To assess adaptive thermogenesis in vivo, we placed Slug ΔLepRb mice at ambient cold temperature, 4 0 C, and monitored body core temperature on a HFD for 12-15 weeks. HFD feeding induced whitening of BAT in Slug fl/fl mice, as illustrated by enlarged lipid droplets (Figure 5A). Remarkably, Slug ΔLepRb mice were completely resistant to HFD-induced BAT whitening ( Figure 5A). BAT Ucp1 protein and mRNA levels were significantly higher in Slug ΔLepRb than in Slug fl/fl mice ( Figure 5, A-C). Sympathetic nerve inputs are known to increase BAT thermogenesis. Sympathetic nerve innervations, as assessed by immunostaining of BAT sections with antibody  through the rectum. Body core temperature was considerably higher in Slug ΔLepRb mice relative to Slug fl/fl mice following cold exposure ( Figure 5E). These results unveil what we believe to be a previously unrecognized hypothalamic Slug/sympathetic nerve/thermogenic fat axis.
Hypothalamic LepRb-neuron Slug promotes leptin resistance. We postulated that hypothalamic Slug might promote obesity by inducing leptin resistance. Considering that hyperleptinemia is often associated with leptin resistance, we measured blood leptin levels. Leptin levels were substantially lower in Slug ΔLepRb males and females relative to sex-and age-matched Slug fl/fl mice on HFD (Supplemental Figure 4A). To exclude body-weight influence on leptin secretion, we measured leptin levels in Slug ΔLepRb and Slug fl/fl mice on a chow diet at 8 weeks of age, when their body weights were comparable. Plasma leptin levels were still significantly lower in Slug ΔLepRb mice ( Figure 7A). To assess leptin sensitivity in vivo, we treated Slug ΔLepRb and Slug fl/fl mice on a chow diet at 7 weeks of age -when their body weights were similar -with leptin for 3 days and monitored body weight changes. Leptin treatments decreased body weight to a significantly higher degree in Slug ΔLepRb mice compared with Slug fl/fl mice ( Figure 7B). To complement these findings, we tested whether MBH LepRb neuron-specific overexpression of Slug inhibits leptin actions. AAV-hSyn-DIO-Slug or AAV-hSyn-DIO-mCherry vectors were bilaterally microinjected into the MBH of Slug ΔLepRb mice on a chow diet. The AAV-transduced mice were treated with leptin for 4 days, and their body weights were monitored. Leptin decreased body weight to a significantly lesser degree in hSyn-DIO-Slug/Slug ΔLepRb mice than in hSyn-DIO-mCherry/Slug ΔLepRb mice ( Figure 7C). To corroborate these results, we assessed hypothalamic leptin signaling in these mice. Slug ΔLepRb and Slug fl/fl male mice -8 weeks old and on a chow diet -were fasted overnight and i.p. injected with leptin 45 minutes before hypothalamic extracts were taken and were prepared and immunoblotted with anti-phospho-Stat3 antibody. Body weight (Supplemental Figure 4B) and baseline Stat3 phosphorylation ( Figure 7D ARC and VMH were significantly higher in Slug ΔLepRb mice relative to Slug fl/fl mice (Supplemental Figure 4C). To test if MBH LepRb + neuron-specific overexpression of Slug suppressed leptin signaling, we bilaterally microinjected AAV-CAG-DIO-Slug or AAV-CAG-DIO-mCherry vectors into the MBH of Slug ΔLepRb (i.e., Slug fl/fl ;LepRb-Cre +/-) mice on a chow diet at 9 weeks of age. Two weeks later, the mice were fasted overnight and stimulated with leptin, and hypothalamic extracts were immunoblotted with anti-phospho-Stat3 antibody. Body weight (Supplemental Figure  4D) and baseline Stat3 phosphorylation ( Figure 7E) were comparable between CAG-DIO-Slug/Slug ΔLepRb and CAG-DIO-mCherry/Slug ΔLepRb mice, but leptin-stimulated phosphorylation of Stat3 was significantly lower in CAG-DIO-Slug/Slug ΔLepRb mice ( Figure  7, E and F). Taken together, these results demonstrate that LepRb + neuron-intrinsic Slug cell-autonomously suppresses leptin signaling to induce leptin resistance, which leads to obesity.

Slug directly suppresses LepRb expression by an epigenetic mechanism.
We next set out to identify molecular targets of hypothalamic Slug. We isolated the hypothalamus from Slug ΔLepRb and Slug fl/fl mice for Affymetrix GeneChip analysis (GSE217748). We found 180 genes upregulated by more than 1.25-fold (P < 0.05) and 70 genes downregulated by more than 25% (P < 0.05) in Slug ΔLepRb mice (Supplemental Figure 5, A and B). These putative targets were annotated to multiple pathways, including leptin signaling pathways (Supplemental Figure 5C). Interestingly, LepRb expression was substantially upregulated in Slug ΔLepRb mice (Supplemental Figure  5, A and B). By quantitative PCR (qPCR), we confirmed that hypothalamic LepRb mRNA levels were significantly higher in Slug ΔLepRb than in Slug fl/fl mice that had been on a HFD for 15 weeks (Supplemental Figure 6A). To exclude body-weight influence on LepRb expression, we measured LepRb abundance in Slug ΔLepRb and Slug fl/fl mice on chow diet at 8 weeks of age, when their body weights were female Slug fl/fl , n = 4; female Slug ΔLepRb , n = 4. (C-E) O 2 consumption, CO 2 production (normalized to lean mass), and respiratory exchange ratio (RER) at 9 weeks of age (n = 4 mice per group). Data are presented as mean ± SEM. *P < 0.05, 2-tailed unpaired Student's t test.
We noticed that both mouse and human LepR promoters contain putative Slug response elements in the form of E2 boxes (Supplemental Figure 6B). To test whether Slug directly represses the LepR promoter, we constructed mouse LepR-promoter luciferasereporter plasmids and cotransfected the reporter plasmids with Slug plasmids into the hypothalamic cell line GT1-7. Slug dose-dependently suppressed LepR-promoter luciferase activity ( Figure 7I). To test if Slug directly binds to the LepR promoter, we isolated the similar (Supplemental Figure 4B). Hypothalamic LepRb mRNA levels were still significantly higher in Slug ΔLepRb than in Slug fl/fl mice ( Figure 7G). To determine whether LepRb + neuron-specific overexpression of Slug inhibits LepRb expression, we bilaterally microinjected AAV-CAG-DIO-Slug or AAV-CAG-DIO-mCherry vectors into the MBH of Slug ΔLepRb (i.e., Slug fl/fl ;LepRb-Cre +/-) mice at 9 weeks of age on a chow diet and isolated the hypothalamus 2 weeks later. Body weight was comparable between CAG-DIO-Slug/Slug ΔLepRb and CAG-DIO-mCherry/Slug ΔLepRb mice (Supplemental Figure  4D), but hypothalamic-LepRb mRNA levels were significantly and showed that MBH LepRb + neuron-specific overexpression of Slug was sufficient to induce obesity, glucose intolerance, insulin resistance, and NAFLD in mice on a chow diet. We confirmed these findings using a distinct Cre-dependent AAV-CAG-DIO-Slug vector. We observed that HFD feeding not only increased hypothalamic Slug expression but also promoted a conversion of hypothalamic Slugneurons to Slug + neurons. We therefore consider aberrant upregulation of hypothalamic Slug as a previously unrecognized causal factor for obesity and metabolic disease.
Leptin resistance has been well documented to drive obesity progression. We found that LepRb + neuron-intrinsic Slug directly induced leptin resistance. Plasma leptin levels were markedly lower in Slug ΔLepRb than in Slug fl/fl mice both before and after body weight divergence. Leptin stimulation increased hypothalamic-Stat3 phosphorylation and body-weight loss to a significantly higher level in Slug ΔLepRb than in Slug fl/fl mice on chow diet at 8-9 weeks of age, when their body weights were similar. This indicates that leptin resistance was a causal factor for, rather than a consequence of, obesity in these models. Conversely, MBH LepRb + neuron-specific overexpression of Slug blunted the ability of leptin to stimulate hypothalamic-Stat3 phosphorylation and to decrease body weight. We propose that Slug-induced leptin resistance in the hypothalamus is a causal factor for obesity and its associated disorders (Supplemental Figure 7). However, these data do not exclude the possibility that Slug may induce obesity by additional mechanisms.
We demonstrated that Slug induced leptin resistance by epigenetically repressing LepRb transcription. We confirmed the previous reports that hypothalamic LepRb expression is decreased in diet-induced obesity (7)(8)(9)18). It is not unexpected that reduced LepRb expression results in leptin resistance and obesity. In line with this notion, db/+ mice with haploinsufficiency of LepRb, in certain genetic backgrounds, are prone to dietinduced obesity and metabolic disorders (19,20). Neuronal restoration of LepRb in db/db mice dose-dependently reverses obesity, metabolic disorders, and fertility dysfunctions (21). Increased expression of hypothalamic LepRb is linked to resistance to dietinduced obesity and infertility in female mice (22). Notably, both mouse and human LepRb promoters contain putative Slug binding sites, and we verified that Slug directly bound to the LepRb promoter in the hypothalamus using ChIP. In cell culture, Slug directly repressed LepRb promoter activity. In mice, LepRb + neuronspecific ablation of Slug increased -whereas MBH LepRb + neuron-specific overexpression of Slug decreased -LepRb expression in the hypothalamus. These findings indicate that LepRb + neuron-intrinsic Slug directly suppressed LepRb expression in vitro and in vivo. At the chromatin level, Slug deficiency decreased H3K27me2 and H3K27me3 levels, which are repressive epigenetic marks, while increasing H3K27ac levels, which is an active epigenetic mark, in the hypothalamus. Consistent with this finding, hypothalamic LepRb promoter H3K27me2 and H3K27me3 levels were increased in HFD-fed mice, correlated with upregulation of hypothalamic Slug. In line with these observations, Slug has been reported to bind to and recruit multiple histone methyltransferases and/or demethylases (10,13). Based on these findings, we propose that obesogenic factors upregulate Slug in the hypothalamus. Slug recruits epigenetic modifiers to induce repressive epigenetic modifications on the LepRb promoter/enhancer, resulting in suppression hypothalamus from WT (i.e., Slug +/+ ) and whole-body Slug knockout (i.e., Slug -/-, negative control) mice and performed ChIP-qPCR assays. Slug +/+ and Slug -/mice were fed a HFD for 14-16 weeks to increase Slug expression in Slug +/+ mice. Slug -/mice, like Slug ΔLepRb mice, were resistant to HFD-induced obesity (Slug +/+ : 35.9 ± 1.13 g, n = 12; Slug -/-: 24.26 ± 0.39 g, n = 10, P < 0.05). We detected an abundant occupancy of hypothalamic Slug on the LepR promoter in Slug +/+ but not Slug -/mice ( Figure 7J).
Slug has been known to regulate histone modifications in target promoters and enhancers (13), prompting us to assess histone 3 lysine-27 dimethylation (H3K27me2), H3K27 trimethylation (H3K27me3), and H3K27 acetylation (H3K27ac) in the LepR promoter. We placed Slug +/+ and Slug -/male mice on a HFD for 14-16 weeks, increasing hypothalamic Slug expression, and isolated the hypothalamus for ChIP-qPCR. Remarkably, LepR promoter H3K27me2 and H3K27me3 levels, which are repressive epigenetic marks, were significantly lower, while H3K27ac levels, which is an active epigenetic mark, were significantly higher in Slug -/mice relative to Slug +/+ mice ( Figure 7K). Taken together, these results suggest that Slug directly bound to the LepRb promoter and epigenetically repressed LepRb expression, leading to leptin resistance.
Given that hypothalamic Slug is upregulated in obesity, we postulated that aberrant Slug might increase LepR promoter H3K27me2/3 levels in diet-induced obesity. We placed C57BL/6J male mice on a HFD for 10 weeks and measured H3K27me2/3 levels in the hypothalamus. LepRb promoter H3K27 methylations were significantly higher in HFD-fed than in chow-fed mice (Supplemental Figure 6C). HFD feeding has been reported to decrease hypothalamic LepRb expression in both mice and rats (7)(8)(9)18). We confirmed that MBH LepR expression was lower in HFD-fed mice than in chow-fed mice (Supplemental Figure 6D). Collectively, these results suggest that Slug epigenetically suppressed LepRb expression, contributing to leptin resistance and obesity (Supplemental Figure 7).

Discussion
In this study, we uncovered -as far as we know -a previously unrecognized obesity-prone action of Slug in hypothalamic neurons, particularly LepRb + neurons, and provided multiple lines of genetic and physiological evidence to establish a pivotal role of hypothalamic Slug in the control of body weight and metabolism. We found that both male and female Slug ΔLepRb mice with a LepRb + cell-specific deletion of Slug were profoundly resistant to HFDinduced, as well as age-associated, obesity, insulin resistance, glucose intolerance, and NAFLD. Slug ΔLepRb mice ate less than Slug fl/fl mice, which explains the obesity-resistant phenotype. Slug ΔLepRb mice resisted BAT whitening and maintained BAT thermogenesis on a HFD, and they recruited more beige adipocytes than Slug fl/fl mice upon cold exposure. Body core temperature was higher and cold tolerance was improved in Slug ΔLepRb mice relative to Slug fl/fl mice. Consistently, whole body energy expenditure, as assessed by O 2 consumption and CO 2 production and normalized to lean mass, was higher in Slug ΔLepRb mice. These observations unveil what we believe to be a previously unrecognized LepRb-neuron Slug/sympathetic nerve/adipose-thermogenesis axis. To directly demonstrate the proobesity action of hypothalamic LepRb-neuron Slug, we generated Cre-dependent, neuron-specific AAV-hSyn-DIO-Slug vectors, Leptin stimulation of weight loss. Mice were i.p. injected with leptin (0.25 mg/kg body weight) twice -once at 6 pm and again at 12 amdaily for 3-4 days, and body weight was recorded.
Cold-tolerance test. Empty cages with no bedding materials were precooled at 4°C in a rodent environmental chamber (RIS33SD, Innovative Solutions). Mice were fasted overnight, transferred to the precooled cage, and housed individually. They had free access to water, but there was no food for them during the cold-exposure experiment. Core body temperature was measured hourly via the rectum.
Plasma insulin and leptin measurement, GTT, and ITT. Blood samples were collected from tail veins. Plasma insulin and leptin were measured using insulin and leptin ELISA kits (Crystal Chem), respectively. For GTT, mice were fasted overnight and i.p. injected with glucose (2 g/kg body weight), and blood glucose was measured 0, 15, 30, 60, and 120 minutes after injection. For ITT, mice were fasted for 6 hours and i.p. injected with insulin (0.75 U/kg).
Fat content and energy expenditure. Fat content and lean body mass (normalized to body weight) were measured using a dualenergy X-ray absorptiometry pDexa (Norland Stratec). Energy expenditure was measured by indirect calorimetry using the Windows Oxymax Equal Flow System (Columbus Instruments). Volume of O 2 consumption (VO 2 ) and volume of CO 2 production (VCO 2 ) were normalized to body lean mass. Additionally, ANCOVA analysis was performed following instructions by the National Institute of Diabetes and Digestive and Kidney Diseases (NIDDK) Mouse Metabolic Phenotyping Centers (www.MMPC.org/shared/regression. aspx). Data were analyzed using linear regression analysis to assess the impact of covariates on energy expenditure. Average VO 2 and VCO 2 values were used for ANCOVA, and either lean mass or body weight were set as covariates.
Liver TAG levels. Liver samples were homogenized in 1% acetic acid and extracted by chloroform-methanol (2:1). The organic phase was dried by evaporation and dissolved in isopropanol. TAG levels were measured using a TAG assay kit (Pointe Scientific Inc.) and normalized to liver weight.
There are limitations in this study. Slug-associated epigenetic modifiers mediating LepRb-promoter histone modifications remain elusive. A cause-effect relationship between LepRb-promoter histone modifications and leptin resistance needs to be further confirmed. Contribution of Slug-based epigenetic reprogramming of leptin pathways to obesity needs to be quantified. Signaling pathways coupling obesogenic factors to Slug upregulation remain to be identified. HFD feeding was reported to increase, rather than decrease, LepRb expression in some hypothalamic areas (30), and the underlying mechanisms for the opposing actions of HFD on LepRb expression remain unknown. Nonetheless, this work has defined what we believe to be a new Slug-elicited epigeneticreprogramming paradigm in the hypothalamus and laid a foundation for future studies to address these questions.
Arbor, Michigan, USA) were grown in DMEM containing 5 mM glucose and 10% FBS at 5% CO 2 and 37°C. GT1-7 cells were transiently cotransfected with pGL3-LepR luciferase reporter plasmids and appropriate other expression vectors, using polyethylenimine (Sigma-Aldrich). Luciferase activity was measured 48 hours after transfection using a kit (Promega) and normalized to β-gal internal control.
Data availability. The authors declare that the data supporting the findings of this study are available within the article and supplemental files.
Statistics. Data were presented as mean ± SEM. Differences between 2 groups were analyzed by 2-tailed Student's t test. Comparisons between more than 2 groups or variables were analyzed by 1-way or 2-way ANOVA and/or Tukey's post hoc test using GraphPad Prism 8. A P value of less than 0.05 was considered significant.
Study approval. Animal research complied with all relevant ethnic regulations. Animal experiments were conducted following the protocols approved by the University of Michigan IACUC.

Author contributions
MHK, YL, QZ, and LJ conducted experiments; MHK and LR designed experiments and wrote the manuscript; and MHK, MGM, WW, and LR edited the manuscript.