POMC neuron METTL14/m6A/YTHDC1/YTHDF2 pathways safeguard energy balance, body weight, and metabolism Authors: Yuan Li, Min-Hyun Kim, Decheng Ren, Liangyou Rui Published: 2026 ABSTRACT METTL14 mediates N6-methyladenosine (m6A) RNA modification, while YTHDC1 and YTHDF2 specifically bind m6A-methylated RNA to regulate RNA fate. POMC neurons constitute the core of the central melanocortin circuit, and POMC deficiency causes obesity in both mice and humans. However, how m6A-based epitranscriptomics regulates melanocortin circuit function remains unclear. Here, we generated and characterized POMC neuron-specific knockout mice lacking Mettl14 (Mettl14ΔPOMC), Ythdc1 (Ythdc1ΔPOMC), or Ythdf2 (Ythdf2 ΔPOMC). Mettl14ΔPOMC and Ythdc1ΔPOMC mice develop hyperphagia, obesity, glucose intolerance, insulin resistance, and hepatic steatosis in both sexes under standard chow conditions, accompanied by POMC downregulation. Conversely, POMC neuron-specific overexpression of METTL14 or YTHDC1 protects against diet-induced obesity. In contrast, Ythdf2 ΔPOMC mice are resistant to obesity, revealing an m6A-dependent balance between YTHDC1 and YTHDF2. Mechanistically, the METTL14/YTHDC1 pathway is indispensable for embryonic POMC neurogenesis, while in adults YTHDC1 maintains melanocortin circuit integrity/function. METTL14 and YTHDC1 directly target POMC and ISL1 transcripts to regulate protein expression. POMC neuron-specific restoration of POMC reverses obesity and metabolic phenotypes in Mettl14ΔPOMC and Ythdc1ΔPOMC mice, defining an anti-obesity METTL14/m6A/YTHDC1/POMC axis. These findings identify METTL14 as the m6A writer for POMC/ISL1 and YTHDC1 and YTHDF2 as their readers, uncovering a critical role of m6A epitranscriptomic regulation in melanocortin circuit development and maintenance. Introduction Pro-opiomelanocortin (POMC) neurons in the arcuate nucleus (ARC) send projections to neurons in the paraventricular hypothalamus (PVH), constituting the core of the central melanocortin circuit that controls energy balance, body weight, and metabolic homeostasis. POMC is cleaved to produce α-melanocyte-stimulating hormone (α-MSH) that binds and activates melanocortin receptor type 4 (MC4R) and MC3R to suppress appetite1–4. POMC deficiency causes obesity and metabolic disease in both mice and humans1,2,5,6. Transcription factor ISL1 plays a critical role in stimulating POMC expression7–9; moreover, it also regulates embryonic POMC neurogenesis2,10. The melanocortin circuit, including POMC protein expression, represents an important therapeutic target for obesity treatment. N6-methyladenosine (m6A) modification of RNA has emerged as a critical mechanism governing cell proliferation, differentiation, and function11. METTL14 binds METTL3 to form an RNA m6A methyltransferase catalyzing m6A methylation (referred to as m6A writer)11,12. YTH domain-containing proteins YTHDC1, YTHDF2, and other family members, referred to as m6A readers, specifically bind m6A-methylated RNA to regulate RNA metabolism, including pre-mRNA splicing, RNA stability, and mRNA nuclear export and translation11,12. Global deletion of Mettl14, Mettl3, Ythdc1, or Ythdf2 causes embryonic lethality13–16, highlighting the essential role of the m6A system in supporting development and survival. Despite their importance, the role of m6A writers and readers in the central melanocortin circuit remains largely unexplored. To investigate their role, we have generated and characterized POMC neuron-specific knockout mice lacking Mettl14 (Mettl14ΔPOMC), Ythdc1 (Ythdc1ΔPOMC), or Ythdf2 (Ythdf2 ΔPOMC). In the current study, we demonstrate that both Mettl14ΔPOMC and Ythdc1ΔPOMC mice develop spontaneous obesity and metabolic disease in both sexes on standard chow conditions, accompanied by downregulation of POMC expression. In contrast, Ythdf2 ΔPOMC mice exhibit the opposite phenotype, revealing an m6A-controlled balance between YTHDC1 and YTHDF2. Mechanistically, deletion of Mettl14 or Ythdc1 disrupts embryonic POMC neurogenesis, thereby decreasing POMC neuron numbers. Notably, adult-onset deletion of POMC neuron Ythdc1 also causes spontaneous obesity; conversely, POMC neuron-specific overexpression of METTL14 or YTHDC1 in adult mice protects against high fat diet (HFD)-induced obesity and metabolic disorders. These results uncover a pivotal role of m6A-dependent YTHDC1 pathways in the maintenance of the melanocortin circuit in adult animals. Furthermore, we have identified ISL1 and POMC mRNAs as key targets of the METTL14 and YTHDC1. POMC neuron-specific restoration of POMC expression reverses obesity and metabolic phenotypes in both Mettl14ΔPOMC and Ythdc1ΔPOMC mice, thus revealing a previously unrecognized METTL14/m6A/YTHDC1/POMC axis and its role in restraining hyperphagia and excessive body weight gain. Taken together, these findings unveil a critical role of POMC neuron-intrinsic METTL14/YTHDC1/YTHDF2 pathways in governing both the development and the maintenance of the central melanocortin circuit. Results POMC neuron-specific overexpression of METTL14 protects against diet-induced obesity and metabolic disorders To examine dietary influence on the m6A system, male C57BL/6J mice were fed a HFD for 15 weeks. Both mRNA and protein levels of METTL14 and YTHDC1 were significantly reduced in the hypothalamus (Fig. 1a–c). Histological analysis unveiled significant reductions in ARC METTL14-expressing neurons and YTHDC1-expressing neurons in HFD-fed mice (Fig. 1d and Supplementary Fig. 1a). These results demonstrate that obesogenic factors suppress the hypothalamic m6A system.Fig. 1POMC neuron-specific overexpression of METTL14 protects against diet-induced obesity and metabolic disorders.a–d C57BL/6 J males were fed a HFD for 15 weeks. a Hypothalamic gene expression (normalized to 36B4 levels; n = 5 mice per group). b, c Hypothalamic extracts were immunoblotted with the indicated antibodies. Protein levels were normalized to GAPDH control (n = 4 mice per group). d METLL14 and YTHDC1 neurons were visualized by immunostaining and quantified in the ARC (normalized to area). Chow: n = 7, HFD: n = 5. e Cre-dependent AAV-CAG-DIO-METTL14 vector. f–l POMC-Cre males (7 weeks) received bilateral microinjections of AAV9-DIO-mCherry or AAV9-DIO-METTL14 vector into the ARC and then fed HFD. f Immunoblotting of ARC extracts 12 weeks after AAV transduction. g Body weight (n = 8 per group). h Fat and lean mass content 10 weeks after AAV injection. mCherry: n = 8, METTL14: n = 10. i H&E and Oil red O staining of tissue sections 12 weeks after AAV injection. Scale bar: 200 µm. j Tissue weights 12 weeks after AAV injection (n = 8 mice per group). k Liver TAG levels 12 weeks after AAV injection (normalized to liver weight). mCherry: n = 7, METTL14: n = 8. l GTT and ITT 10 weeks after AAV injection. GTT: mCherry: n = 8, METTL14: n = 6; ITT: mCherry: n = 8, METTL14: n = 8. Data are presented as mean ± SEM. g, l 2-way ANOVA/Bonferroni’s multiple comparisons test. a, c, d, h, j, k 2-tailed unpaired Student’s t test. POMC neuron-specific overexpression of METTL14 protects against diet-induced obesity and metabolic disorders. a–d C57BL/6 J males were fed a HFD for 15 weeks. a Hypothalamic gene expression (normalized to 36B4 levels; n = 5 mice per group). b, c Hypothalamic extracts were immunoblotted with the indicated antibodies. Protein levels were normalized to GAPDH control (n = 4 mice per group). d METLL14 and YTHDC1 neurons were visualized by immunostaining and quantified in the ARC (normalized to area). Chow: n = 7, HFD: n = 5. e Cre-dependent AAV-CAG-DIO-METTL14 vector. f–l POMC-Cre males (7 weeks) received bilateral microinjections of AAV9-DIO-mCherry or AAV9-DIO-METTL14 vector into the ARC and then fed HFD. f Immunoblotting of ARC extracts 12 weeks after AAV transduction. g Body weight (n = 8 per group). h Fat and lean mass content 10 weeks after AAV injection. mCherry: n = 8, METTL14: n = 10. i H&E and Oil red O staining of tissue sections 12 weeks after AAV injection. Scale bar: 200 µm. j Tissue weights 12 weeks after AAV injection (n = 8 mice per group). k Liver TAG levels 12 weeks after AAV injection (normalized to liver weight). mCherry: n = 7, METTL14: n = 8. l GTT and ITT 10 weeks after AAV injection. GTT: mCherry: n = 8, METTL14: n = 6; ITT: mCherry: n = 8, METTL14: n = 8. Data are presented as mean ± SEM. g, l 2-way ANOVA/Bonferroni’s multiple comparisons test. a, c, d, h, j, k 2-tailed unpaired Student’s t test. To investigate the pathological role of m6A dysfunctions, we tested if METTL14 restoration might rescue diet-induced obesity. To this end, we generated a Cre-dependent AAV-CAG-DIO-METTL14 vector (Fig. 1e). Male POMC-Cre mice were bilaterally microinjected into the ARC with AAV9-CAG-DIO-METTL14 or AAV9-CAG-DIO-mCherry (control) vector and were subsequently fed an HFD for 12 weeks. Immunoblot and histological analyses verified an ARC- and POMC neuron-specific overexpression of METTL14 (Fig. 1f and Supplementary Fig. 1c, d). METTL14 overexpression profoundly reduced diet-induced obesity, as evidenced by a progressive and profound reduction of body weight in AAV9-CAG-DIO-METTL14-transduced mice compared with AAV9-CAG-DIO-mCherry mice (Fig. 1g). METTL14 overexpression significantly reduced fat mass (Fig. 1h). Consistently, adipocytes in both epididymal white adipose tissue (eWAT) and inguinal WAT (iWAT) were smaller in METTL14-overexpressing mice (Fig. 1i and Supplementary Fig. 1e), and the weights of the liver, eWAT, iWAT, and brown adipose tissue (BAT) were significantly reduced (Fig. 1j). POMC neuron-specific overexpression of METTL14 substantially reduced the number and size of hepatic lipid droplets, as revealed by H&E and Oil red O staining, as well as liver triacylglycerol (TAG) levels (Fig. 1i–k). METTL14 overexpression also markedly improved glucose intolerance and insulin resistance, as assessed by glucose tolerance test (GTT) and insulin tolerance test (ITT), respectively (Fig. 1l and Supplementary Fig. 1f). Together, these results reveal a previously unrecognized anti-obesity strategy through activation of METTL14 pathways in POMC neurons. POMC neuron-specific deletion of Mettl14 results in obesity and metabolic disorders To study the role of endogenous METTL14, we generated POMC neuron-specific Mettl14 knockout mice (Mettl14ΔPOMC) by crossing Mettl14f/f mice with POMC-Cre drivers. RNAscope analysis confirmed efficient and specific deletion of Mettl14 in POMC neurons (Supplementary Fig. 2a). Strikingly, both male and female Mettl14ΔPOMC mice developed severe obesity despite being maintained on a normal chow diet (Fig. 2a). Fat mass and the weights of iWAT, eWAT (in males), gonadal WAT (gWAT; in females), and BAT were significantly elevated in Mettl14ΔPOMC mice compared with Mettl14f/f littermates (Fig. 2b and Supplementary Fig. 2b). Histological analyses revealed adipocyte hypertrophy in iWAT and eWAT, as well as enlarged lipid droplets in BAT, indicative of BAT whitening and dysfunction (Fig. 2c and Supplementary Fig. 2c). Mettl14ΔPOMC mice exhibited hyperinsulinemia (Fig. 2d), a hallmark of insulin resistance. As expected, Mettl14ΔPOMC mice developed glucose intolerance and insulin resistance in both sexes, as assessed by GTT and ITT, respectively (Fig. 2e, f and Supplementary Fig. 2d, e). Hepatic insulin signaling was severely impaired in Mettl14ΔPOMC mice, as evidenced by reduced AKT phosphorylation (Fig. 2g, h). Hepatic lipid droplets, as assessed by H&E and Oil red O staining, and liver TAG levels were significantly elevated in Mettl14ΔPOMC mice (Fig. 2c, i). In addition, body length was increased in Mettl14ΔPOMC mice (Supplementary Fig. 2f), consistent with previous reports that disruption of central melanocortin signaling promotes increased linear length17. Pituitary morphology, size, and POMC cell numbers were comparable between Mettl14f/f and Mettl14ΔPOMC mice (Supplementary Fig. 3a, b), suggesting that the observed phenotypes are not attributable to pituitary dysfunction. Moreover, GTT and ITT were indistinguishable between young Mettl14ΔPOMC and Mettl14f/f mice prior to body weight divergence (Supplementary Fig. 3c), suggesting that the metabolic abnormalities arise largely secondary to obesity.Fig. 2POMC neuron-specific deletion of Mettl14 causes obesity and metabolic dysfunction.a Body weight under chow diet. Males: Mettl14f/f: n = 12, Mettl14ΔPOMC: n = 9; females: Mettl14f/f: n = 12, Mettl14ΔPOMC: n = 8. b Tissue weights at 20 weeks of age. Males: Mettl14f/f: n = 8, Mettl14ΔPOMC: n = 8; females: Mettl14f/f: n = 9, Mettl14ΔPOMC: n = 7. c H&E and Oil red O staining of tissue sections from males at 20 weeks. Scale bar: 200 µm. d Male overnight-fasted plasma insulin levels at 13 weeks (n = 6 per group). e, f GTT and ITT at 17 weeks. Male GTT: Mettl14f/f: n = 10, Mettl14ΔPOMC: n = 10; male ITT: Mettl14f/f: n = 7, Mettl14ΔPOMC: n = 10; female GTT: n = 7 mice per group, female ITT: Mettl14f/f: n = 12, Mettl14ΔPOMC: n = 7. g, h Western blot analysis of liver extracts from fasted male mice (20 weeks) injected with insulin. Insulin-stimulated phosphorylation of AKT (pSer473) was normalized to AKT. Mettl14f/f: n = 3, Mettl14ΔPOMC: n = 4. (i) Liver TAG levels at 20 weeks (normalized to liver weight). Males: Mettl14f/f: n = 8, Mettl14ΔPOMC: n = 7; females: n = 6 mice per group. Data are presented as mean ± SEM. a, e, f 2-way ANOVA/Bonferroni’s multiple comparisons test. b, d, h, i 2-tailed unpaired Student’s t test. POMC neuron-specific deletion of Mettl14 causes obesity and metabolic dysfunction. a Body weight under chow diet. Males: Mettl14f/f: n = 12, Mettl14ΔPOMC: n = 9; females: Mettl14f/f: n = 12, Mettl14ΔPOMC: n = 8. b Tissue weights at 20 weeks of age. Males: Mettl14f/f: n = 8, Mettl14ΔPOMC: n = 8; females: Mettl14f/f: n = 9, Mettl14ΔPOMC: n = 7. c H&E and Oil red O staining of tissue sections from males at 20 weeks. Scale bar: 200 µm. d Male overnight-fasted plasma insulin levels at 13 weeks (n = 6 per group). e, f GTT and ITT at 17 weeks. Male GTT: Mettl14f/f: n = 10, Mettl14ΔPOMC: n = 10; male ITT: Mettl14f/f: n = 7, Mettl14ΔPOMC: n = 10; female GTT: n = 7 mice per group, female ITT: Mettl14f/f: n = 12, Mettl14ΔPOMC: n = 7. g, h Western blot analysis of liver extracts from fasted male mice (20 weeks) injected with insulin. Insulin-stimulated phosphorylation of AKT (pSer473) was normalized to AKT. Mettl14f/f: n = 3, Mettl14ΔPOMC: n = 4. (i) Liver TAG levels at 20 weeks (normalized to liver weight). Males: Mettl14f/f: n = 8, Mettl14ΔPOMC: n = 7; females: n = 6 mice per group. Data are presented as mean ± SEM. a, e, f 2-way ANOVA/Bonferroni’s multiple comparisons test. b, d, h, i 2-tailed unpaired Student’s t test. To examine the role of METTL14 in adult mice, we generated an adult-onset, POMC neuron–specific Mettl14 deletion model. Mettl14f/f mice were crossed with tamoxifen-inducible POMC-CreERT drivers, which have been previously validated18, and the resulting Mettl14f/f;CreERT+/− mice were treated with tamoxifen. However, deletion efficiency was low, with a substantial proportion of POMC neurons retaining METTL14 expression, as assessed by RNAscope analysis (Supplementary Fig. 4a, b). Consistent with this limited deletion, body weight, fat content, glucose tolerance, and insulin sensitivity were largely unchanged in tamoxifen-treated Mettl14f/f;CreERT+/− mice (Supplementary Fig. 4c–f). It is likely that limited accessibility of CreERT to the Mettl14 floxed allele may underlie the inefficient recombination observed in this model. POMC neuron-specific overexpression of YTHDC1 protects against obesity and metabolic disease We sought to determine whether YTHDC1 mediates METTL14’s anti-obesity function. We generated Cre-dependent AAV9-CAG-DIO-YTHDC1 vector (Fig. 3a) and bilaterally injected the vector into the ARC of male POMC-Cre mice (AAV9-CAG-DIO-mCherry as control). The transduced mice were subsequently fed a HFD to induce obesity. Immunoblotting and histological analysis confirmed that YTHDC1 was modestly overexpressed selectively in POMC neurons (Fig. 3b, c and Supplementary Fig. 5a, b). Remarkably, POMC neuron-specific overexpression of YTHDC1 progressively and significantly decreased body weight (Fig. 3d), accompanied by a significant reduction in fat but not lean mass (Fig. 3e) and reduced adipocyte size in eWAT and iWAT (Fig. 3f and Supplementary Figs. 5c). Liver, eWAT, and iWAT weights, hepatic lipid droplet content, and liver TAG levels were all reduced by YTHDC1 overexpression (Fig. 3f–h). YTHDC1 overexpression modestly improved glucose intolerance but markedly mitigated insulin resistance (Fig. 3i and Supplementary Fig. 5d). It is likely that YTHDC1 overexpression influences glucose clearance by an additional insulin-independent mechanism. Together, these results unveil an important anti-obesity METTL14/YTHDC1 epitranscriptomic pathway in POMC neurons.Fig. 3POMC neuron-specific overexpression of YTHDC1 protects against HFD-induced obesity and metabolic dysfunction.a Cre-dependent AAV-CAG-DIO-YTHDC1 vector. b–i POMC-Cre males (7 weeks) were bilaterally microinjected into the ARC with AAV9-DIO-mCherry or AAV9-DIO-YTHDC1 vector and then fed HFD. b Immunoblotting of ARC extracts 12 weeks after AAV injection. c YTHDC1 and POMC levels were quantified and normalized to Actb levels (n = 4 per group). d Body weight. mCherry: n = 8, YTHDC1: n = 11. e Fat and lean mass in males 10 weeks post AAV injection (by pDexa). mCherry: n = 11, YTHDC1: n = 9. f H&E and Oil red O staining of tissue sections 12 weeks after AAV injection. Scale bar: 200 µm. g Tissue weights 12 weeks after AAV injection (n = 7 mice per group). h Liver TAG levels 12 weeks after AAV injection (normalized to liver weight; n = 7 per group). i GTT and ITT 10 weeks after AAV injection. mCherry: n = 8, YTHDC1: n = 9. Data are presented as mean ± SEM. d, i 2-way ANOVA/Bonferroni’s multiple comparisons test. c, e, g, h 2-tailed unpaired Student’s t test. POMC neuron-specific overexpression of YTHDC1 protects against HFD-induced obesity and metabolic dysfunction. a Cre-dependent AAV-CAG-DIO-YTHDC1 vector. b–i POMC-Cre males (7 weeks) were bilaterally microinjected into the ARC with AAV9-DIO-mCherry or AAV9-DIO-YTHDC1 vector and then fed HFD. b Immunoblotting of ARC extracts 12 weeks after AAV injection. c YTHDC1 and POMC levels were quantified and normalized to Actb levels (n = 4 per group). d Body weight. mCherry: n = 8, YTHDC1: n = 11. e Fat and lean mass in males 10 weeks post AAV injection (by pDexa). mCherry: n = 11, YTHDC1: n = 9. f H&E and Oil red O staining of tissue sections 12 weeks after AAV injection. Scale bar: 200 µm. g Tissue weights 12 weeks after AAV injection (n = 7 mice per group). h Liver TAG levels 12 weeks after AAV injection (normalized to liver weight; n = 7 per group). i GTT and ITT 10 weeks after AAV injection. mCherry: n = 8, YTHDC1: n = 9. Data are presented as mean ± SEM. d, i 2-way ANOVA/Bonferroni’s multiple comparisons test. c, e, g, h 2-tailed unpaired Student’s t test. Embryonic POMC neuron–specific deletion of Ythdc1 results in obesity and metabolic dysfunction We next aimed to investigate the functions of endogenous YTHDC1. To this end, we generated Ythdc1f/f mice using CRISPR/Cas9 (Fig. 4a) and subsequently produced POMC neuron-specific Ythdc1 knockout (Ythdc1ΔPOMC) mice by crossing Ythdc1f/f mice with POMC-Cre mice. Histological analysis confirmed a POMC neuron-restricted deletion, as evidenced by the loss of YTHDC1 expression in a large proportion of tdTomato lineage-labelled POMC neurons (Supplementary Fig. 6a, b). Strikingly, Ythdc1ΔPOMC mice of both sexes developed severe obesity despite being maintained on a normal chow diet (Fig. 4b). Consistently, fat mass and weights of iWAT, eWAT (in males), gWAT (in females), and BAT were significantly elevated in Ythdc1ΔPOMC mice compared with Ythdc1f/f littermates (Fig. 4c and Supplementary Fig. 6c). Histological examination revealed marked adipocyte hypertrophy in iWAT and eWAT, along with pronounced BAT whitening in Ythdc1ΔPOMC mice (Fig. 4d). Body length was also increased (Supplementary Fig. 6d). Similar to Mettl14ΔPOMC mice, Ythdc1ΔPOMC mice exhibited hyperinsulinemia, glucose intolerance, and insulin resistance in both sexes (Fig. 4e–g and Supplementary Fig. 6e). Insulin-stimulated phosphorylation of AKT in the liver was impaired in both male and female Ythdc1ΔPOMC mice (Fig. 4h–i and Supplementary Figs. 6f, g). Liver weight, hepatic lipid droplet accumulation, and liver TAG levels were all significantly elevated in Ythdc1ΔPOMC mice (Fig. 4d, j). Notably, glucose and insulin tolerance were comparable between Ythdc1ΔPOMC and Ythdc1f/f mice prior to divergence in body weight (Supplementary Fig. 6h), suggesting that obesity is the primary driver of the observed metabolic dysfunction.Fig. 4Embryonic deletion of Ythdc1 in POMC neurons causes obesity and metabolic dysfunction.a The targeted Ythdc1 allele. b Body weight. Males: Ythdc1f/f: n = 9, Ythdc1ΔPOMC: n = 15; females: Ythdc1f/f: n = 13, Ythdc1ΔPOMC: n = 8. c Tissue weights at 20 weeks. Males: Ythdc1f/f: n = 11, Ythdc1ΔPOMC: n = 13; females: Ythdc1f/f: n = 10, Ythdc1ΔPOMC: n = 11. d H&E and Oil red O staining of tissue sections from males at 20 weeks. Scale bar: 200 µm. e Male overnight-fasted plasma insulin levels at 10 weeks (n = 5 per group). f, g GTT and ITT at 17 weeks. Male GTT: Ythdc1f/f: n = 7, Ythdc1ΔPOMC: n = 10; Male ITT: Ythdc1f/f: n = 8, Ythdc1ΔPOMC: n = 10; female GTT and ITT: Ythdc1f/f: n = 10, female Ythdc1ΔPOMC: n = 11. h–i Western blot analysis of liver extracts from fasted male mice (20 weeks) following insulin injection. j Liver TAG levels at 20 weeks (normalized to liver weight). Males: n = 8 per group; females: n = 6 per group. Data are presented as mean ± SEM. b, f, g 2-way ANOVA/Bonferroni’s multiple comparisons test. c, e, i, j 2-tailed unpaired Student’s t test. Embryonic deletion of Ythdc1 in POMC neurons causes obesity and metabolic dysfunction. a The targeted Ythdc1 allele. b Body weight. Males: Ythdc1f/f: n = 9, Ythdc1ΔPOMC: n = 15; females: Ythdc1f/f: n = 13, Ythdc1ΔPOMC: n = 8. c Tissue weights at 20 weeks. Males: Ythdc1f/f: n = 11, Ythdc1ΔPOMC: n = 13; females: Ythdc1f/f: n = 10, Ythdc1ΔPOMC: n = 11. d H&E and Oil red O staining of tissue sections from males at 20 weeks. Scale bar: 200 µm. e Male overnight-fasted plasma insulin levels at 10 weeks (n = 5 per group). f, g GTT and ITT at 17 weeks. Male GTT: Ythdc1f/f: n = 7, Ythdc1ΔPOMC: n = 10; Male ITT: Ythdc1f/f: n = 8, Ythdc1ΔPOMC: n = 10; female GTT and ITT: Ythdc1f/f: n = 10, female Ythdc1ΔPOMC: n = 11. h–i Western blot analysis of liver extracts from fasted male mice (20 weeks) following insulin injection. j Liver TAG levels at 20 weeks (normalized to liver weight). Males: n = 8 per group; females: n = 6 per group. Data are presented as mean ± SEM. b, f, g 2-way ANOVA/Bonferroni’s multiple comparisons test. c, e, i, j 2-tailed unpaired Student’s t test. Adult-onset Ythdc1 deletion in POMC neurons causes obesity and metabolic disease Embryonic deletion might perturb developmental programs in Ythdc1 ΔPOMC mice. To circumvent developmental adaptations, Ythdc1f/f mice were crossed with POMC-CreERT drivers to produce Ythdc1f/f;CreERT+/− mice (i.e., wild-type). Adult Ythdc1f/f;CreERT+/− mice were treated with tamoxifen at 12 weeks of age to selectively disrupt Ythdc1 in POMC neurons. Histological analysis confirmed POMC neuron-specific ablation of Ythdc1, as evidenced by the loss of YTHDC1 expression in a large proportion of tdTomato lineage-labeled POMC neurons (Supplementary Fig. 7a, b). Tamoxifen-induced adult-onset deletion of Ythdc1 resulted in obesity even under normal chow diet conditions (Fig. 5a). Ablation of Ythdc1 substantially increased fat content, weights of eWAT, iWAT, BAT, and liver, and adipocyte hypertrophy in eWAT and iWAT, accompanied by pronounced elevations in hepatic lipid droplet accumulation and TAG levels (Fig. 5b–e and Supplementary Fig. 7c). In addition, YTHDC1-deficient mice exhibited glucose intolerance and insulin resistance, as assessed by GTT and ITT, respectively (Fig. 5f). Notably, embryonic deletion of Ythdc1 produced more severe obesity and metabolic phenotypes than adult-onset deletion, raising the possibility that the METTL14/m6A/YTHDC1 pathway may play a pivotal role in both the development and the maintenance of the central melanocortin system.Fig. 5Adult-onset deletion of Ythdc1 in POMC neurons causes obesity and metabolic dysfunction.Ythdc1f/f and Ythdc1f/f;CreERT males were fed a standard chow diet and treated with tamoxifen at 12 weeks of age. a Male body weight. Ythdc1f/f: n = 8, Ythdc1f/f;CreERT: n = 11. b Fat and lean mass content 6 weeks after tamoxifen injection. Mettl14f/f: n = 11, Mettl14f/f;CreERT: n = 7. c H&E and Oil red O staining of tissue sections 8 weeks after tamoxifen injection. Scale bar: 200 µm. d Tissue weights 8 weeks after tamoxifen injection. Ythdc1f/f: n = 8, Ythdc1f/f;CreERT: n = 8. e Liver TAG levels 8 weeks after tamoxifen injection (normalized to liver weight). Ythdc1f/f: n = 7, Ythdc1f/f;CreERT: n = 9. f GTT and ITT 4 weeks after tamoxifen injection (n = 8 mice per group). AUC: area under the curve: a.u., arbitrary units. Data are presented as mean ± SEM. a, f 2-way ANOVA/Bonferroni’s multiple comparisons test. b, d, e, f 2-tailed unpaired Student’s t test. Adult-onset deletion of Ythdc1 in POMC neurons causes obesity and metabolic dysfunction. Ythdc1f/f and Ythdc1f/f;CreERT males were fed a standard chow diet and treated with tamoxifen at 12 weeks of age. a Male body weight. Ythdc1f/f: n = 8, Ythdc1f/f;CreERT: n = 11. b Fat and lean mass content 6 weeks after tamoxifen injection. Mettl14f/f: n = 11, Mettl14f/f;CreERT: n = 7. c H&E and Oil red O staining of tissue sections 8 weeks after tamoxifen injection. Scale bar: 200 µm. d Tissue weights 8 weeks after tamoxifen injection. Ythdc1f/f: n = 8, Ythdc1f/f;CreERT: n = 8. e Liver TAG levels 8 weeks after tamoxifen injection (normalized to liver weight). Ythdc1f/f: n = 7, Ythdc1f/f;CreERT: n = 9. f GTT and ITT 4 weeks after tamoxifen injection (n = 8 mice per group). AUC: area under the curve: a.u., arbitrary units. Data are presented as mean ± SEM. a, f 2-way ANOVA/Bonferroni’s multiple comparisons test. b, d, e, f 2-tailed unpaired Student’s t test. The POMC neuron METTL14/m6A/YTHDC1 pathway critically regulates food intake and energy expenditure To interrogate how the m6A system protects against obesity, we assessed food intake and energy expenditure in both gain-of-function (METTL14 and YTHDC1 overexpression) and loss-of-function (Mettl14 and Ythdc1 knockout) models described above. POMC neuron-specific overexpression of either METTL14 or YTHDC1 significantly reduced food intake while increasing core body temperature, particularly during the dark phase (Fig. 6a–d). Conversely, ablation of either Mettl14 or Ythdc1 in POMC neurons resulted in hyperphagia accompanied by a reduction in core body temperate (Fig. 6e–h). Consistently, adult-onset deletion of Ythdc1 in POMC neurons also induced hyperphagia and decreased core body temperature (Fig. 6i, j). We further evaluated energy balance using CLAMS. O2 consumption and CO2 production, normalized to lean mass, were significantly reduced in both Mettl14ΔPOMC and Ythdc1ΔPOMC mice, whereas food intake was increased (Fig. 6k–l and Supplementary Figs. 8a–d). ANCOVA analysis further confirmed a reduction in energy expenditure in both mutant models (Supplementary Fig. 8e, f). In line with these findings, expression of thermogenic genes in BAT and iWAT (e.g., Ucp1, Pgc1α, Dio2) was significantly reduced in Mettl14ΔPOMC and Ythdc1ΔPOMC mice (Supplementary Fig. 9), and UCP1 protein levels were reduced in Ythdc1ΔPOMC mice (Fig. 6m). Consequently, Ythdc1ΔPOMC mice exhibited exacerbated hypothermia following cold exposure compared with controls (Fig. 6n). These results suggest that the METTL14/m6A/YTHDC1 pathway in POMC neurons protects against obesity and associated metabolic dysfunction, at least in part, through safeguarding energy balance.Fig. 6POMC neuron-intrinsic METTL14/YTHDC1 pathways regulate energy balance.a–d Male POMC-Cre mice (7 weeks) received bilateral microinjections of AAV9-DIO-mCherry, AAV9-DIO-METTL14, or AAV9-DIO-YTHDC1 vector into the ARC and then fed HFD. a, b Food intake and rectal temperature 6 weeks after AAV transduction. mCherry: n = 7, METTL14: n = 6. c, d Food intake and rectal temperature 5 weeks after AAV transduction (n = 6 per group). e Food intake (males at 7 weeks maintained on a chow diet). Mettl14f/f: n = 9, Mettl14ΔPOMC: n = 8. f Male core body temperature at 7 weeks. Mettl14f/f: n = 11, Mettl14ΔPOMC: n = 7. g Male food intake at 7 weeks (on chow diet; n = 6 per group). h Male core body temperature at 7 weeks (n = 8 per group). i–j Male Ythdc1f/f and Ythdc1f/f;CreERT mice were treated with tamoxifen at 12 weeks of age (on standard chow diet). Food intake and rectal temperature 7 weeks after tamoxifen treatment. Ythdc1f/f: n = 6, Ythdc1f/f: n = 7. k O2 consumption and CO2 production in males at 7 weeks of age (normalized to lean mass). Mettl14f/f: n = 9, Mettl14ΔPOMC: n = 10. l Male O2 consumption and CO2 production at 7 weeks (normalized to lean mass). Ythdc1f/f: n = 12, Ythdc1ΔPOMC: n = 8. m Male BAT extracts were immunoblotted with the indicated antibodies at 20 weeks. n Rectal temperature of males (19 weeks old) upon cold exposure (4 °C). Ythdc1f/f: n = 6, Ythdc1ΔPOMC: n = 10. Data are presented as mean ± SEM. a, c, e, g, i, n 2-way ANOVA/Bonferroni’s multiple comparisons test. b, d, f, h, j, k, l 2-tailed unpaired Student’s t test. POMC neuron-intrinsic METTL14/YTHDC1 pathways regulate energy balance. a–d Male POMC-Cre mice (7 weeks) received bilateral microinjections of AAV9-DIO-mCherry, AAV9-DIO-METTL14, or AAV9-DIO-YTHDC1 vector into the ARC and then fed HFD. a, b Food intake and rectal temperature 6 weeks after AAV transduction. mCherry: n = 7, METTL14: n = 6. c, d Food intake and rectal temperature 5 weeks after AAV transduction (n = 6 per group). e Food intake (males at 7 weeks maintained on a chow diet). Mettl14f/f: n = 9, Mettl14ΔPOMC: n = 8. f Male core body temperature at 7 weeks. Mettl14f/f: n = 11, Mettl14ΔPOMC: n = 7. g Male food intake at 7 weeks (on chow diet; n = 6 per group). h Male core body temperature at 7 weeks (n = 8 per group). i–j Male Ythdc1f/f and Ythdc1f/f;CreERT mice were treated with tamoxifen at 12 weeks of age (on standard chow diet). Food intake and rectal temperature 7 weeks after tamoxifen treatment. Ythdc1f/f: n = 6, Ythdc1f/f: n = 7. k O2 consumption and CO2 production in males at 7 weeks of age (normalized to lean mass). Mettl14f/f: n = 9, Mettl14ΔPOMC: n = 10. l Male O2 consumption and CO2 production at 7 weeks (normalized to lean mass). Ythdc1f/f: n = 12, Ythdc1ΔPOMC: n = 8. m Male BAT extracts were immunoblotted with the indicated antibodies at 20 weeks. n Rectal temperature of males (19 weeks old) upon cold exposure (4 °C). Ythdc1f/f: n = 6, Ythdc1ΔPOMC: n = 10. Data are presented as mean ± SEM. a, c, e, g, i, n 2-way ANOVA/Bonferroni’s multiple comparisons test. b, d, f, h, j, k, l 2-tailed unpaired Student’s t test. The METTL14/m6A/YTHDC1 pathway enhances POMC and ISL1 biosynthesis in POMC neurons We next aimed to identify mRNA targets of METTL14 and YTHDC1 that mediate the metabolic function of the METTL14/m6A/YTHDC1 pathway. POMC neuron–specific overexpression of either METTL14 or YTHDC1 significantly increased POMC fiber density in the PVH, despite no change in POMC neuron numbers in the ARC (Fig. 7a–d), indicative of enhanced POMC/α-MSH production. Consistently, hypothalamic Pomc and Isl1 mRNA levels, but not those of AgRP and Npy, were significantly elevated in METLL14- and YTHDC1-overexpressing mice (Fig. 7e, f). The transcription factor ISL1 is known to play a critical role in embryonic POMC neurogenesis as well as in the regulation of Pomc transcription7,8. Accordingly, YTHDC1-overexpressing mice exhibited increased POMC protein levels in the ARC (Fig. 3b, c). In contrast, adult-onset deletion of Ythdc1 in POMC neurons significantly reduced POMC fiber density in the PVH, hypothalamic Pomc and Isl1 mRNA abundance, and ISL1 and POMC protein levels in the ARC, without affecting POMC neuron numbers (Fig. 7g–i and Supplementary Fig. 10a). In line with these findings, hypothalamic expression of Isl1 and Pomc was also significantly decreased in Mettl14ΔPOMC and Ythdc1ΔPOMC mice (Supplementary Fig. 10b).Fig. 7METTL14/YTHDC1 pathways enhance ISL1 and POMC biosynthesis.a–f POMC-Cre males were bilaterally injected into the ARC with AAV9-CAG-DIO-METTL14, AAV9-CAG-DIO-YTHDC1, or AAV9-CAG-DIO-mCherry vector and the fed HFD for 12 weeks. a–d ARC POMC neuron numbers and PVH POMC fiber density were analyzed histologically using anti-POMC antibody and quantified. b: mCherry: n = 4, METTL14: n = 4. d: mCherry: n = 4, YTHDC1: n = 5. e, f Hypothalamic gene expression 12 weeks after AAV injection (normalized to 36B4 levels). e mCherry: n = 6, METTL14: n = 6. f: mCherry: n = 6, YTHDC1: n = 8. g–i The hypothalamus was harvested 8 weeks after tamoxifen injection. g Hypothalamic sections were stained with anti-POMC antibody. ARC POMC neuron numbers and PVH POMC fiber density were quantified (n = 5 mice per group). h Gene expression (normalized to 36B4 levels; n = 8 mice per group). i ARC extracts were immunoblotted with the indicated antibodies. j–l m6A modification of Isl1 and Pomc mRNAs in the hypothalamus (normalized to inputs). j n = 5 per group. k n = 3 per group. l C57BL/6J males were fed a HFD for 15 weeks. Isl1: n = 3 per group, Pomc: n = 4 per group). m, n ISL1 or POMC plasmids were cotransfected with METTL14 and YTHDC1 plasmids into N2a cell culture. m YTHDC1-bound Isl1 and Pomc mRNAs were measured by YTHDC1-linked RIP and normalized to inputs (n = 3 per group). n Cell extracts were immunoblotted with the indicated antibodies. o POMC or ISL1 plasmids were cotransfected with METTL14 and/or YTHDC1 plasmids into N2a cells. POMC and ISL1 translations were assessed using OPP assays (normalized to inputs; n = 3 per group). Data are presented as mean ± SEM. b, d–h, j–m, o 2-tailed unpaired Student’s t test. METTL14/YTHDC1 pathways enhance ISL1 and POMC biosynthesis. a–f POMC-Cre males were bilaterally injected into the ARC with AAV9-CAG-DIO-METTL14, AAV9-CAG-DIO-YTHDC1, or AAV9-CAG-DIO-mCherry vector and the fed HFD for 12 weeks. a–d ARC POMC neuron numbers and PVH POMC fiber density were analyzed histologically using anti-POMC antibody and quantified. b: mCherry: n = 4, METTL14: n = 4. d: mCherry: n = 4, YTHDC1: n = 5. e, f Hypothalamic gene expression 12 weeks after AAV injection (normalized to 36B4 levels). e mCherry: n = 6, METTL14: n = 6. f: mCherry: n = 6, YTHDC1: n = 8. g–i The hypothalamus was harvested 8 weeks after tamoxifen injection. g Hypothalamic sections were stained with anti-POMC antibody. ARC POMC neuron numbers and PVH POMC fiber density were quantified (n = 5 mice per group). h Gene expression (normalized to 36B4 levels; n = 8 mice per group). i ARC extracts were immunoblotted with the indicated antibodies. j–l m6A modification of Isl1 and Pomc mRNAs in the hypothalamus (normalized to inputs). j n = 5 per group. k n = 3 per group. l C57BL/6J males were fed a HFD for 15 weeks. Isl1: n = 3 per group, Pomc: n = 4 per group). m, n ISL1 or POMC plasmids were cotransfected with METTL14 and YTHDC1 plasmids into N2a cell culture. m YTHDC1-bound Isl1 and Pomc mRNAs were measured by YTHDC1-linked RIP and normalized to inputs (n = 3 per group). n Cell extracts were immunoblotted with the indicated antibodies. o POMC or ISL1 plasmids were cotransfected with METTL14 and/or YTHDC1 plasmids into N2a cells. POMC and ISL1 translations were assessed using OPP assays (normalized to inputs; n = 3 per group). Data are presented as mean ± SEM. b, d–h, j–m, o 2-tailed unpaired Student’s t test. To determine whether RNA m6A modification links METTL14 and YTHDC1 to POMC and ISL1 synthesis, we measured m6A enrichment on Isl1 and Pomc transcripts using methylated RNA immunoprecipitation assays (MeRIP). As expected, overexpression of the m6A writer METTL14 increased the abundance of m6A-modified Isl1 and Pomc mRNAs, whereas ablation of endogenous Mettl14 produced the opposite effect (Fig. 7j, k). Consistently, in N2a neuronal cells, METTL14 overexpression similarly enhanced m6A-methylation of Isl1 and Pomc transcripts (Supplementary Fig. 10c). Importantly, HFD feeding―which suppresses hypothalamic METTL14 expression―significantly reduced m6A methylation of Isl1 and Pomc mRNAs in the ARC (Fig. 7l). These findings uncover a previously unrecognized diet-METTL14-m6A-Isl1/Pomc regulatory axis. We next examined whether YTHDC1 directly binds Isl1 and Pomc transcripts in an m6A-dependent manner using YTHDC1-linked RNA immunoprecipitation (RIP). In N2a neuronal cells overexpressing YTHDC1, YTHDC1 robustly associated with Isl1 and Pomc mRNAs, and co-overexpression of the m6A writer METTL14 further enhanced YTHDC1 binding to both transcripts (Fig. 7m). These results indicate that METTL14 and YTHDC1 function, respectively, as the m6A writer and reader for Isl1 and Pomc mRNAs in neurons. To define the functional consequences of YTHDC1 binding, we next measured POMC and ISL1 synthesis in N2a cells. Co-overexpression of METTL14 and YTHDC1 synergistically increased POMC protein levels (Fig. 7n), whereas METTL14 overexpression alone was sufficient, likely via endogenous YTHDC1, to elevate ISL1 protein abundance (Supplementary Fig. 10d). In line with these findings, overexpression of METTL14 and YTHDC1 increased Pomc mRNA stability, as determined by transcription inhibition using actinomycin D; in contrast, METTL14 exerted a limited effect on Isl1 mRNA stability(Supplementary Fig. 10e). We next measured POMC and ISL1 translations using O-propargyl-puromycin (OPP) pulldown assays19,20. Overexpression of METTL14 and YTHDC1 markedly enhanced POMC translation in N2a cells, as evidenced by increased levels of OPP-labeled POMC protein (Fig. 7o and Supplementary Fig. 10f). Similarly, METTL14 overexpression significantly promoted ISL1 translation (Fig. 7o and Supplementary Fig. 10g). Collectively, these results uncover a METTL14-m6A-YTHDC1-ISL1/POMC axis. Restoration of POMC expression rescues obesity and metabolic phenotypes in Mettl14ΔPOMC and Ythdc1ΔPOMC mice To verify the pathological contribution of POMC deficiency in Mettl14ΔPOMC and Ythdc1ΔPOMC mice, Cre-dependent AAV-DIO-POMC, AAV-DIO-α-MSH, or AAV-DIO-mCherry (control) vectors were injected into the ARC. To circumvent potential developmental confounds, an adult-onset Ythdc1 deletion model was employed to assess the effect of POMC neuron–specific restoration of POMC and α-MSH secretion. AAV-DIO-POMC and AAV-DIO-α-MSH vectors have been validated previously21, and we further confirmed POMC neuron-specific restoration of POMC or α-MSH production in Mettl14ΔPOMC and Ythdc1ΔPOMC mice by histological analyses (Fig. 8a). As expected, Mettl14ΔPOMC and Ythdc1ΔPOMC mice developed severe obesity and metabolic dysfunction. Remarkably, restoration of POMC largely reversed obesity, glucose intolerance, insulin resistance, hepatic steatosis, and hyperphagia in both models (Fig. 8b–g and Supplementary Fig. 11a–h). Restoration of α-MSH similarly rescued these pathological phenotypes in Ythdc1ΔPOMC mice (Fig. 8h–j and Supplementary Fig. 11i–l). These findings define a critical role of the POMC neuron METTL14/m6A/YTHDC1/POMC axis in maintaining energy homeostasis and safeguarding against obesity and metabolic disease.Fig. 8POMC neuron-specific reconstitution of POMC protein expression rescues the obesity phenotypes in Mettl14ΔPOMC and Ythdc1ΔPOMC mice.AAV-DIO-POMC, AAV-DIO-α-MSH, or AAV-DIO-mCherry vector was injected into the ARC of male Mettl14ΔPOMC and Ythdc1ΔPOMC (tamoxifen-induced) mice, and body weight and metabolism were modified following AAV transduction. a Hypothalamic sections were stained with anti-POMC or anti-α-MSH antibody 4 weeks after AAV transduction. Scale bar: 200 µm. b Body weight. mCherry: n = 6, POMC: n = 6. c GTT and ITT 8 weeks after AAV transduction. mCherry: n = 6, POMC: n = 6. d Food intake 9 weeks after AAV transduction. mCherry: n = 6, POMC: n = 6. e Body weight. mCherry: n = 6, POMC: n = 7. f GTT and ITT 7 weeks after AAV transduction. mCherry: n = 6, POMC: n = 7. g Food intake 8 weeks after AAV transduction. mCherry: n = 6, POMC: n = 7. h Body weight. mCherry: n = 6, α-MSH: n = 5. i GTT and ITT 7 weeks after AAV transduction. mCherry: n = 6, α-MSH: n = 5. j Food intake 8 weeks after AAV transduction. mCherry: n = 6, α-MSH: n = 5. Data are presented as mean ± SEM. b–j 2-way ANOVA/Bonferroni’s multiple comparisons test. POMC neuron-specific reconstitution of POMC protein expression rescues the obesity phenotypes in Mettl14ΔPOMC and Ythdc1ΔPOMC mice. AAV-DIO-POMC, AAV-DIO-α-MSH, or AAV-DIO-mCherry vector was injected into the ARC of male Mettl14ΔPOMC and Ythdc1ΔPOMC (tamoxifen-induced) mice, and body weight and metabolism were modified following AAV transduction. a Hypothalamic sections were stained with anti-POMC or anti-α-MSH antibody 4 weeks after AAV transduction. Scale bar: 200 µm. b Body weight. mCherry: n = 6, POMC: n = 6. c GTT and ITT 8 weeks after AAV transduction. mCherry: n = 6, POMC: n = 6. d Food intake 9 weeks after AAV transduction. mCherry: n = 6, POMC: n = 6. e Body weight. mCherry: n = 6, POMC: n = 7. f GTT and ITT 7 weeks after AAV transduction. mCherry: n = 6, POMC: n = 7. g Food intake 8 weeks after AAV transduction. mCherry: n = 6, POMC: n = 7. h Body weight. mCherry: n = 6, α-MSH: n = 5. i GTT and ITT 7 weeks after AAV transduction. mCherry: n = 6, α-MSH: n = 5. j Food intake 8 weeks after AAV transduction. mCherry: n = 6, α-MSH: n = 5. Data are presented as mean ± SEM. b–j 2-way ANOVA/Bonferroni’s multiple comparisons test. The METTL14/m6A/YTHDC1 pathway promotes embryonic POMC neurogenesis We noticed that embryonic deletion of Ythdc1 resulted in more severe obesity than adult-onset deletion (Fig. 4b vs Fig. 5a), raising the possibility that the METTL14/m6A/YTHDC1 pathway might safeguard the central melanocortin system beyond enhancing POMC synthesis. Supporting this notion, the number of POMC neurons in the ARC was markedly reduced in both Mettl14ΔPOMC and Ythdc1ΔPOMC mice compared with wild-type controls (Fig. 9a, b), whereas AgRP neuron numbers remained unchanged (Supplementary Fig. 12a). Consistently, POMC fiber density in the PVH was substantially diminished in both mutant models (Fig. 9a, b). Importantly, a significant reduction in POMC neuron number was already evident in Mettl14ΔPOMC pups at postnatal day 10 (P10) (Supplementary Fig. 12b), further supporting a critical role of the METTL14/YTHDC1 pathway in the development of the central melanocortin system.Fig. 9METTL14/YTHDC1 pathways enhance embryonic POMC neurogenesis.a, b Hypothalamic sections from males (8 weeks) were stained with anti-POMC antibody. ARC POMC neurons and PVH POMC fiber density were quantified. Scale bar: 200 µm. Mettl14f/f and Mettl14∆POMC:ARC: n = 6; Mettl14f/f: PVH: n = 6, Mettl14∆POMC: n = 4. Ythdc1f/f and Ythdc1∆POMC: n = 6 mice per group. c, d POMC neurons in the ARC were histologically visualized at P10 using the tdTomato reporter and anti-POMC antibody, and were subsequently counted. Control: n = 4, Mettl14∆POMC: n = 4, control: n = 6, Ythdc1∆POMC: n = 4. Scale bar: 200 µm. e, f Hypothalamic sections from P1 pups were stained with BrdU. BrdU+tdTomato+ cells were counted and normalized to total tdTomato+ cells (n = 4 pups per group). Scale bar: 200 µm. g Hypothalamus m6A-modified mRNAs at E18 (normalized to inputs; n = 3 embryos per group). h Hypothalamic gene expression at E18 (normalized to 36B4 levels; Mettl14f/f and Mettl14∆POMC: n = 6 embryos per group; Ythdc1f/f: n = 6, Ythdc1∆POMC: n = 7). i Hypothalamic extracts were immunoblotted with the indicated antibodies at P1day. Data are presented as mean ± SEM. b, d, f, g, h 2-tailed unpaired Student’s t test. METTL14/YTHDC1 pathways enhance embryonic POMC neurogenesis. a, b Hypothalamic sections from males (8 weeks) were stained with anti-POMC antibody. ARC POMC neurons and PVH POMC fiber density were quantified. Scale bar: 200 µm. Mettl14f/f and Mettl14∆POMC:ARC: n = 6; Mettl14f/f: PVH: n = 6, Mettl14∆POMC: n = 4. Ythdc1f/f and Ythdc1∆POMC: n = 6 mice per group. c, d POMC neurons in the ARC were histologically visualized at P10 using the tdTomato reporter and anti-POMC antibody, and were subsequently counted. Control: n = 4, Mettl14∆POMC: n = 4, control: n = 6, Ythdc1∆POMC: n = 4. Scale bar: 200 µm. e, f Hypothalamic sections from P1 pups were stained with BrdU. BrdU+tdTomato+ cells were counted and normalized to total tdTomato+ cells (n = 4 pups per group). Scale bar: 200 µm. g Hypothalamus m6A-modified mRNAs at E18 (normalized to inputs; n = 3 embryos per group). h Hypothalamic gene expression at E18 (normalized to 36B4 levels; Mettl14f/f and Mettl14∆POMC: n = 6 embryos per group; Ythdc1f/f: n = 6, Ythdc1∆POMC: n = 7). i Hypothalamic extracts were immunoblotted with the indicated antibodies at P1day. Data are presented as mean ± SEM. b, d, f, g, h 2-tailed unpaired Student’s t test. To directly examine embryonic POMC neurogenesis, we employed tdTomato-based lineage-tracing approaches. Mettl14ΔPOMC and Ythdc1ΔPOMC mice were crossed with Rosa26LSL-tdTomato+/- reporter mice to genetically label POMC-expressing cells. Histological analysis revealed a significant reduction in tdTomato+ POMC neurons in Mettl14ΔPOMC pups at P1 and P10, as well as in Ythdc1ΔPOMC pups at P10 (Fig. 9c, d). Because no differences in apoptosis were detected between knockout and control groups, we next assessed POMC progenitor proliferation using BrdU incorporation assays. POMC progenitors were labelled at embryonic day 10.5 (E10.5), and BrdU+tdTomato+ POMC neurons were quantified at P1. The number of BrdU+tdTomato+ cells were significantly reduced in both Mettl14ΔPOMC and Ythdc1ΔPOMC animals (Fig. 9e-f). Collectively, these findings indicate that the progenitor-intrinsic METTL14/YTHDC1 pathway critically supports embryonic POMC neurogenesis and melanocortin circuit development. We next explored the underlying molecular mechanisms by assessing m6A modification and expression of key hypothalamic transcriptional regulators that govern POMC neurogenesis, including ISL1, SIX3, Nkx2.1, MASH1, TBX3, and PRDM121,22,23. m6A methylation of Isl1, Six3, Prdm12, and Pomc mRNAs was significantly inhibited in Mettl14ΔPOMC embryos (Fig. 9g), indicating that METTL14 serves as the m6A writer for these transcripts. Moreover, Isl1, and Six3 mRNA expression were significantly suppressed in both Mettl14ΔPOMC and Ythdc1ΔPOMC embryos at E18 (Fig. 9h). Consistently, ISL1 protein levels were significantly reduced in Mettl14ΔPOMC mice at both P1 and 8 weeks of age (Fig. 9i and Supplementary Fig. 12c). Together, these findings suggest that the METTL14/m6A/YTHDC1 pathway safeguards POMC neurogenesis and melanocortin circuit development, at least in part, by sustaining ISL1 and POMC expression. YTHDF2 in POMC neurons promotes obesity Notably, the obesity phenotype was more severe in Ythdc1ΔPOMC mice than in Mettl14ΔPOMC mice (Fig. 10a), raising the possibly that additional m6A readers might promote obesity and thereby counteract the action of YTHDC1 in POMC neurons. Among these, YTHDF2 is known to mediate m6A-dependent mRNA decay. To test this hypothesis, we generated POMC neuron-specific Ythdf2 knockout mice (Ythdf2ΔPOMC) by crossing Ythdf2f/f mice with POMC-Cre drivers. Under normal chow diet conditions, Ythdf2ΔPOMC mice exhibited significantly reduced body weight compared with Ythdf2f/f littermates (Fig. 10b). In line with these results, Ythdf2ΔPOMC mice were resistant to HFD-induced obesity, as evidenced by marked reductions in body weight, total fat mass, eWAT and iWAT weights, and white adipocyte size (Fig. 10c-g). Hepatic TAG levels and lipid droplet accumulation were also markedly reduced in Ythdf2ΔPOMC mice (Fig. 10f-g). Moreover, glucose intolerance and insulin resistance were markedly improved, as assessed by GTT and ITT (Fig. 10h). Ythdf2ΔPOMC mice consumed less food and displayed significantly elevated core body temperature during the dark phase, indicative of increased energy expenditure (Fig. 10i-j). Together, these results suggest that POMC neuron YTHDF2 promotes obesity and metabolic dysfunction, at least in part, by disrupting energy balance.Fig. 10POMC neuron-specific deletion of Ythdf2 protects against obesity.(a, b) Body weights (replotting Figs. 4A and 5A). Males: Control: n = 21 (Mettl14f/f: n = 12, Ythdc1f/f: n = 9), Mettl14ΔPOMC: n = 9, Ythdc1ΔPOMC: n = 15. Females: Control: n = 25 (Mettl14f/f: n = 12, Ythdc1f/f: n = 13), Mettl14ΔPOMC: n = 8, Ythdc1ΔPOMC: n = 8. (b, c) Male body weight. (b) Ythdf2f/f: n = 11, Ythdf2ΔPOMC: n = 10. (c) Ythdf2f/f: n = 10, Ythdf2ΔPOMC: n = 9. (d) Fat and lean mass content 9 weeks after HFD, normalized to body weight. Ythdf2f/f: n = 7, Ythdf2ΔPOMC: n = 6. (e) Tissue weights 10 weeks after HFD. Ythdf2f/f: n = 8, Ythdf2ΔPOMC: n = 6. (f) H&E and Oil red O staining of tissue sections 10 weeks after HFD. Scale bar: 200 µm. (g) Liver TAG levels 10 weeks after HFD (normalized to liver weight; n = 6 per group). (h) GTT and ITT 8 weeks after HFD. GTT: Ythdf2f/f: n = 8, Ythdf2ΔPOMC: n = 6; ITT: n=6 per group. (i, j) Male food intake and rectal temperature 10 weeks after HFD (n = 6 per group). Data are presented as mean ± SEM. a–c, h, i 2-way ANOVA/Bonferroni’s multiple comparisons test. d, e, g, j 2-tailed unpaired Student’s t test. POMC neuron-specific deletion of Ythdf2 protects against obesity. (a, b) Body weights (replotting Figs. 4A and 5A). Males: Control: n = 21 (Mettl14f/f: n = 12, Ythdc1f/f: n = 9), Mettl14ΔPOMC: n = 9, Ythdc1ΔPOMC: n = 15. Females: Control: n = 25 (Mettl14f/f: n = 12, Ythdc1f/f: n = 13), Mettl14ΔPOMC: n = 8, Ythdc1ΔPOMC: n = 8. (b, c) Male body weight. (b) Ythdf2f/f: n = 11, Ythdf2ΔPOMC: n = 10. (c) Ythdf2f/f: n = 10, Ythdf2ΔPOMC: n = 9. (d) Fat and lean mass content 9 weeks after HFD, normalized to body weight. Ythdf2f/f: n = 7, Ythdf2ΔPOMC: n = 6. (e) Tissue weights 10 weeks after HFD. Ythdf2f/f: n = 8, Ythdf2ΔPOMC: n = 6. (f) H&E and Oil red O staining of tissue sections 10 weeks after HFD. Scale bar: 200 µm. (g) Liver TAG levels 10 weeks after HFD (normalized to liver weight; n = 6 per group). (h) GTT and ITT 8 weeks after HFD. GTT: Ythdf2f/f: n = 8, Ythdf2ΔPOMC: n = 6; ITT: n=6 per group. (i, j) Male food intake and rectal temperature 10 weeks after HFD (n = 6 per group). Data are presented as mean ± SEM. a–c, h, i 2-way ANOVA/Bonferroni’s multiple comparisons test. d, e, g, j 2-tailed unpaired Student’s t test. To test whether YTHDF2 functionally counteracts YTHDC1, both proteins were expressed in N2a neuronal cells. While YTHDC1 alone increased ISL1 and POMC levels, and coexpression of YTHDF2 attenuated these effects (Supplementary Figs. 13a-b). Similar to YTHDC1, YTHDF2 bound to POMC transcripts, as demonstrated by RIP assays (Supplementary Fig. 13c). Importantly, YTHDF2 overexpression reduced the amount of YTHDC1-bound POMC mRNA (Supplementary Fig. 13d). Collectively, these findings reveal opposing actions of YTHDF2 and YTHDC1 in POMC neurons and define an m6A reader-specific regulatory axis that governs energy balance and body weight homeostasis. Discussion While the central melanocortin system is well known to regulate energy balance and body weight in both mice and humans, a key knowledge gap remains regarding the role of RNA modifications and RNA-binding proteins in regulating embryonic melanocortin circuit development and in maintaining its integrity and function in adulthood. In the current study, we defined distinct functions of the METTL14/YTHDC1 epitranscriptomic pathway in embryonic melanocortin circuit development and its maintenance in adulthood, and uncovered opposing functions for YTHDC1 and YTHDF2 in POMC neurons. We demonstrated that embryonic deletion of either Mettl14 or Ythdc1 substantially reduced POMC neuron numbers due to impaired progenitor cell proliferation, resulting in severe obesity and metabolic dysfunction in adult mice. METTL14 and YTHDC1 directly targeted ISL1 mRNA and enhanced ISL1 protein expression, suggesting that a METTL14/m6A/YTHDC1/ISL1 epitranscriptomic axis contributes, at least in part, to the metabolic phenotypes observed in Mettl14ΔPOMC and Ythdc1ΔPOMC mice. Because POMC-expressing progenitor cells also give rise to AgRP and other hypothalamic neuron subpopulations10,24,25, and future studies will be required to determine whether alterations in these lineages contribute to the observed phenotypes. Moreover, as Mettl14 and Ythdc1 are also deleted in nucleus of the solitary tract (NST) POMC neurons, we cannot exclude a contribution of the NST METTL14/YTHDC1 pathway to certain aspects of the phenotype. We also observed, unexpectedly, that Ythdc1ΔPOMC mice were prone, whereas Ythdf2ΔPOMC mice were resistant, to obesity. In cell culture, YTHDC1 competed with YTHDF2 for binding to POMC mRNA,with YTHDC1 enhancing POMC synthesis and YTHDF2 antagonizing this effect. These findings support an m6A-dependent balance between YTHDC1 and YTHDF2 that shapes melanocortin circuit function (Supplementary Fig. 14). Importantly, adult-onset deletion of Ythdc1 in POMC neurons caused hyperphagia, obesity, and metabolic disorders, whereas overexpression of YTHDC1 or METTL14 in adult mice―overexpression was restricted to ARC but not NST POMC neurons―protected against HFD-induced obesity and metabolic dysfunction. These findings firmly establish a pivotal role of the METTL14/YTHDC1 epitranscriptomic pathway in adulthood by maintaining melanocortin circuit function. METTL14 and YTDHC1 directly targeted POMC mRNA to enhance POMC protein expression, and hypothalamic POMC expression was markedly reduced in Mettl14ΔPOMC and Ythdc1ΔPOMC mice. Restoration of POMC expression reversed obesity and metabolic phenotypes, uncovering a previously unrecognized anti-obesity METTL14/m6A/YTHDC1/POMC epitranscriptomic pathway. Nonetheless, additional non-POMC RNA targets of METTL14 and YTHDC1 may also contribute to these phenotypes and remain to be identified. During preparation of this manuscript, Shen et al reported mice with pan-hypothalamic deletion of Mettl14, Ythdc1, or Ythdf226. Although both studies identified a similar role of METTL14 and YTHDC1 in embryonic POMC neurogenesis, differences emerge between pan-hypothalamic and POMC neuron-specific loss of Ythdc1 and Ythdf2. In Shen et al., pan-hypothalamic deletion of Ythdc1 produced only mild metabolic alterations, and pan-hypothalamic Ythdf2 knockout mice appeared normal. In contrast, POMC neuron-specific ablation of Ythdc1 in the present study caused severe obesity and metabolic dysfunction, whereas POMC neuron-specific deletion of Ythdf2 conferred resistance to obesity and metabolic disorders. We propose that YTHDC1 and YTHDF2 exert cell type–specific and opposing functions in POMC versus non-POMC hypothalamic neurons. Consequently, deletion of Ythdc1 or Ythdf2 in non-POMC neurons in pan-hypothalamic models may mask or counteract the phenotypes arising from their loss in POMC neurons. Further studies will be required to directly test this hypothesis. We further show that obesogenic signals from HFD feeding reduced expression METTL14 and YTHDC1 in the hypothalamus, as well as m6A modification of Isl1 and Pomc mRNAs. These findings indicate that the METTL14 -m6A-YTHDC1-ISL1/POMC epitranscriptomic pathway is disrupted by obesogenic environments, thereby promoting obesity and its associated diseases (Supplementary Fig. 14). Thus, this m6A-centered epitranscriptomic pathway represents a potential therapeutic target for obesity and related diseases. Methods Animals Mettl14f/f and Ythdf2f/f mice were previously described16,19,27. Mettl14f/f and Ythdf2f/f mice were crossed with POMC-Cre drivers (JAX #005965) to obtain Mettl14ΔPOMC (Mettl14f/f;POMC-Cre+/-) and Ythdf2ΔPOMC (Ythdf2f/f;POMC-Cre+/-) mice (C57BL/6 background), respectively. Rosa26LSL-tdTomato mice (JAX 007914) were crossed with POMC-Cre or Mettl14ΔPOMC mice to obtain tdTomatoPOMC mice (Rosa26LSL-tdTomato+/-;POMC-Cre+/-) or tdTomatoPOMCMettl14ΔPOMC mice (Rosa26LSL-tdTomat +/-;Mettl14f/f;POMC-Cre+/-), respectively. POMC-CreERT2 mice were crossed with Mettl14f/f mice to generate Mettl14f/f;CreERT mice. Mice were housed on a 12 h light-dark cycle at ambient temperature of 25 0C and humidity of 30% and were fed ad libitum a normal chow diet (9% fat; TestDiet, St. Louis, MO) or a HFD (60% fat in calories; Research Diets, New Brunswick, NJ). Ythdc1 knockout mice Ythdc1f/+ ES cells were obtained from the European Mouse Mutant Cell Repository (EuMMCR, EPD0113_2_D09) to generate Ythdc1f/f mice (C57BL/6 J background). Ythdc1f/f mice were crossed with POMC-Cre or Mettl14f/f;CreERT drivers to obtain Ythdc1ΔPOMC mice (Ythdc1f/f;POMC-Cre+/-) and Ythdc1f/f;CreERT (Mettl14f/f;POMC-CreERT+/-) mice, respectively. Adult-onset deletion of Mettl14 or Ythdc1 Mettl14f/f;CreERT and Ythdc1f/f;CreERT males were treated with tamoxifen (100 mg/kg body weight, 5 times, 5 days apart) starting at 8 or 12 weeks of age. Mettl14f/f and Ythdc1f/f mice were treated with tamoxifen as control. AAV9-CAG-DIO-METTL14, AAV9-CAG-DIO-METTL14, AAV-DIO-POMC, and AAV-DIO-α-MSH vectors and stereotaxic microinjection We prepared Cre-dependent AAV-CAG-DIO-METTL14 (human), AAV-CAG-DIO-YTHDC1 (human), or AAV-CAG-DIO-mCherry vector. AAV-DIO-POMC and AAV-DIO-α-MSH vectors (dj serotype), which are available from Addgene (POMC: Cat# 239042; α-MSH: Cat#239044), were provided by Dr. Qingchun Tong (University of Texas Health Science Center at Houston). POMC-Cre males (7 weeks old) were isoflurane-anesthetized and mounted on an Ultra Precise Small Animal Stereotaxic Alignment System (David KOPF Instruments, Tujunga, CA). A small opening was made in the skull. AAV vector (0.3 µl) was bilaterally microinjected into the ARC (mm: −1.5 a-p, ±0.3 m-l, −5.9 d-v), using an UltraMicroPump with SYS-Micro4 Controller (UMP3-2, World Precision Instruments Inc, Sarasota, FL). One week later, mice were fed an HFD for 12 weeks. RNAScope RNAscope assays were performed by the Microscopy, Imaging, and Cellular Physiology Core, Michigan Diabetes Research Center, as described previously28. Mettl14 and Pomc transcripts were visualized by Pomc (ACD #314081) and Mettl14 probes (ACD #1166331-C2), respectively, using RNAscope® Multiplex Fluorescent in situ assay Reagent V2 Kit (Cat# 323110). CLAMS Body composition was measured using EchoMRI (Houston, TX). Energy expenditure was measured using the Comprehensive Laboratory Monitoring System (CLAMS, Columbus Instruments) in the University of Michigan Metabolic, Physiological, and Behavioral Core. Data was analyzed using ANCOVA by Notational Mouse Metabolic Phenotyping Center (MMPC): http://www.mmpc.org/shared/regression.aspx. Fat and lean mass Fat content and lean mass were measured using a dual-energy X-ray absorptiometry pDexa (Norland Stratec). Heads, necks, and tails were not included in data analysis. GTT and ITT Mice were fasted overnight (GTT) or 6 h (ITT) and intraperitoneally injected with glucose (2 g/kg body weight, GTT) or human insulin (1 unit/kg body weight, ITT). Blood samples were collected from tail veins to measure blood glucose levels at 0, 15, 30, 60, and 120 min after injection. Plasma insulin levels were measured using mouse insulin ELISA kits (CRYSTAL CHEM, Downers Grove, IL). Liver TAG Liver samples were homogenized in 1% acetic acid and extracted using 80% chloroform/methanol (2:1). The organic fraction was dried in a chemical hood, resuspended in a KOH (3 M)/ethanol solution, incubated at 70 °C for 1 h, and resuspended in MgCl2 (0.75 M). Aqueous fractions were used to measure TAG levels using a TAG assay kit (Pointe Scientific Inc., Canton, MI). Immunoblotting Tissues or cell cultures were homogenized in ice-cold RIPA buffer (50 mM Tris HCl, pH 7.5, 0.5% Nonidet P-40, 150 mM NaCl, 2 mM EGTA, 1 mM Na3VO4, 100 mM NaF, 10 mM Na4P2O7, 1 mM phenylmethylsulfonyl fluoride, 10 μg/ml aprotinin, 10 μg/ml leupeptin). Tissue or cell extracts were resolved by SDS-PAGE and immunoblotted with the indicated antibodies (Supplementary Table 1). Full uncropped scans were provided in the Source data file. Immunostaining Brain sections were prepared using Leica cryostat (Leica Biosystems Nussloch GmbH, Nussloch, Germany), and immunostained with the indicated antibodies (Supplementary Table 1). Images were visualized using BX51 Microscope (Olympus, Tokyo, Japan) and a DP72 digital camera (Olympus, Tokyo, Japan). POMC fiber density in the PVH was quantified using ImageJ and normalized to view area. H&E and Oil red O staining were described previously28. BrdU incorporation Dams were intraperitoneally injected with BrdU (80 mg/kg body weight, in PBS) on E10.5, E11.5, and E12.5. Brains were dissected at P1, fixed in 4% paraformaldehyde for 2 h, cryoprotected in 30% sucrose for 48 h at 4 °C, and stained with anti-BrdU antibody. Quantitative real time RT-PCR (qPCR) Total RNA was extracted using TRIzol reagents (Invitrogen Life Technologies, Carlsbad, CA). The first-strand cDNAs were synthesized using random primers and M-MLV reverse transcriptase (Promega, Madison, WI). qPCR was performed using Radiant™ SYBR Green 2X Lo-ROX qPCR Kits (Alkali Scientific, Pompano Beach, FL), a StepOnePlus RT PCR System (Life Technologies Corporation, NY, USA), and respective primers (Supplementary Table 2). MeRIP Total RNA was extracted using TRIzol reagents. RNA (10 μg) was incubated at 4 °C with anti-m6A antibody (1 μl) for 2 h and then with protein-A agarose beads (10 μl) for an additional 2 h (with rotations). The beads were washed with PBS, and RNA was extracted using TRIzol reagent to measure m6A-marked Isl1 and Pomc mRNAs by qPCR (normalized to inputs). O-propargyl-puromycin (OPP)-mediated pulldown assay N2a neurons (ATCC, CCL-131) were grown in DMEM supplemented with 10% FBS. Mouse Isl1 cDNA was prepared from hypothalamic mRNA using RT-PCR and inserted into pcDNA3.1(+) vectors. Forward: 5′-ATGGATCCACAGATATGGGAGACATGGGCGATC-3′; reverse: 5′-ATATCTAGACATTAGAACAGACTTCATGCGCTT-3′. Isl1, POMC, METTL14, and YTHDC1 plasmids were cotransfected into N2a cells using Lipofectamine 3000. Two days later, POMC and ISL1 protein translations were measured by OPP pulldown assays as we described previously19. RNA stability assay Isl1 or POMC plasmids were cotransfected into N2a cells with METTTL14 and YTHDC1 plasmids. After 20 h, cells were treated with actinomycin D (5 µg/ml) for various durations. Total RNA was extracted to measure Isl1 or POMC mRNAs levels by qPCR (normalized to 36B4 levels). Statistical analysis Data were presented as means ± SEM. Differences were analyzed by a 2-tailed Student’s t-test (2 groups) and ANOVA/Bonferroni post-test (>2 groups) using GraphPad Prism 8. A P value less than 0.05 was considered statistically significant. Survival rates were calculated using the Kaplan-Meier method. Ethics Animal research complied with all relevant ethical regulations and conducted following the protocols approved by the University of Michigan Institutional Animal Care and Use Committee (IACUC). Reporting summary Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article. Supplementary information Supplementary Information Reporting Summary Transparent Peer Review file Supplementary Information Reporting Summary Transparent Peer Review file Source data Source Data Source Data