IDH3A Deficiency Compromises Adaptive Thermogenesis and Exacerbates Obesity-Induced Metabolic Dysfunction via Impaired BCKDHA-Dependent BCAA Catabolism

Abstract

Adaptive thermogenesis in beige adipocytes is essential for maintaining energy homeostasis and preventing obesity. Emerging evidence suggests that human visceral adipose tissue harbors adipocytes with beige-like thermogenic features, enabling analysis of thermogenic gene regulation in humans. Isocitrate dehydrogenase 3A (IDH3A) is a rate-limiting enzyme of the tricarboxylic acid cycle, yet its function in adipocytes remains poorly defined. In this study, we examined IDH3A expression in human visceral adipose tissue and generated adipocyte-specific IDH3A knockout mice to investigate its role in beige adipocyte thermogenesis and metabolic regulation. IDH3A expression in human visceral fat was inversely associated with adiposity and adverse metabolic traits. Moreover, IDH3A expression was induced in human and mouse adipocytes following thermogenic stimulation. Adipocyte-specific IDH3A deletion in mice impaired beige fat thermogenic capacity, led to cold intolerance, and exacerbated diet-induced metabolic dysfunction. Mechanistically, IDH3A deficiency increased DNA methylation at the Bckdha promoter, resulting in the repression of this key branched-chain amino acid (BCAA) catabolic gene and impaired BCAA catabolism. Notably, restoring BCKDHA in IDH3A-deficient adipocytes rescued respiration and thermogenic function. Together, in addition to its canonical enzymatic role, our findings identify IDH3A as a critical regulator of BCAA catabolism that facilitates adaptive thermogenesis under metabolic stress conditions.

Article Highlights

• IDH3A expression in human visceral fat exhibits a negative correlation with metabolic dysfunction indicators.

• IDH3A is induced during thermogenic activation, yet its role in adipose tissue is not well characterized.

• Adipocyte-specific deletion of IDH3A impairs adaptive thermogenesis and metabolic homeostasis by disrupting BCKDHA-mediated BCAA catabolism.

• Overexpression of BCKDHA restores the thermogenic program in IDH3A-deficient beige adipocytes.

Graphical Abstract

Introduction

Obesity continues to rise globally and is tightly linked to type 2 diabetes and cardiometabolic disease (1). Recent expert opinion has emphasized that BMI measures alone are insufficient and that excess adiposity should be confirmed by additional anthropometric criteria or direct body fat measurement (2). Even individuals with normal or slightly elevated body weight may be at increased risk of obesity-related metabolic complications if they accumulate excess visceral adipose tissue (VAT) (3,4). Recent studies have suggested that human VAT harbors adipocytes with beige fat–like thermogenic characteristics (5). Activation of these inducible beige adipocytes increases energy expenditure through adaptive thermogenesis and ameliorates metabolic dysfunction (6,7). Unlike classical brown adipocytes, which are abundant in infants but decline postnatally, beige adipocytes develop after birth and emerge within white adipose tissue (WAT) when stimulated by cold exposure or β3-adrenergic agonists (8). However, the molecular mechanisms that sustain beige thermogenic capacity remain incompletely defined.

Branched-chain amino acid (BCAA) catabolism has been implicated in the acute cold-induced activation of thermogenic adipose tissue (9–11). Their oxidation is governed by the branched-chain α-keto acid dehydrogenase (BCKDH) complex, of which the E1 subunit encoded by BCKDHA is rate limiting (12). Impaired BCAA catabolism in adipose tissue is associated with reduced energy expenditure and worsened systemic metabolism (13,14). Thus, identifying the key regulators of BCAA catabolism in adipose tissue is of considerable physiological importance.

IDHs are enzymes of the tricarboxylic acid cycle (TCA) that catalyze the oxidative decarboxylation of isocitrate to α-ketoglutarate (αKG). Among IDH isoforms, IDH3 catalyzes an irreversible, rate-limiting step as a heterotetramer containing the essential catalytic α-subunit IDH3A (15,16). While roles for IDH1 and IDH2 in adipocyte biology have been described (17,18), the expression and function of IDH3A in adipose tissue remain largely undefined. Altered IDH1 activity has been implicated in epigenetic regulation of specific BCAA metabolic enzymes through DNA methylation (19,20). However, whether similar regulatory mechanisms exist in adipocytes has not been explored.

In this study, we examined IDH3A expression patterns in human and mouse adipose tissue under metabolic stress conditions. Using in vivo and in vitro loss-of-function models, we investigated the role of IDH3A in systemic metabolic regulation. Our findings indicate that IDH3A links BCAA catabolism to adaptive thermogenesis and metabolic homeostasis.

Research Design and Methods

Human Adipose Tissue Samples

Paired abdominal subcutaneous adipose tissue (SAT) and VAT were collected from 236 individuals undergoing bariatric surgery or cholecystectomy at Shanghai Jiao Tong University School of Medicine Affiliated Sixth People’s Hospital (July 2019 to August 2020). Participant criteria and clinical assessments were previously described (21), and detailed characteristics are provided in Supplementary Table 1. Visceral fat area (VFA) was quantified by MRI. The study was approved by the institutional ethics committee, and all participants provided written informed consent.

Animal Experiments

Mice on a C57BL/6J background were maintained under standard conditions with ad libitum access to food and water. Adipocyte-specific Idh3a knockout (Idh3a AKO) mice were generated by crossing floxed Idh3a (Idh3a^F/F; GemPharmatech Co., Ltd.) with adiponectin-Cre^+/− mice (provided by Dr. Suzhen Chen) to delete exons 2–8 (primer sequences in Supplementary Table 3). Idh3a^F/F/adiponectin-Cre^−/− littermates were used as controls. Metabolic stress models included cold exposure, with mice housed at 4°C for 3 or 7 days; β3-adrenergic stimulation, with CL316,243 (1 mg/kg C5976; Sigma-Aldrich) administered intraperitoneally; and 60% high-fat diet (HFD) feeding (D12492; Research Diets) in 8-week-old male mice for 22 weeks, with body weight measured weekly. For indirect calorimetry, mice were individually housed at 22°C, and metabolic parameters were monitored for 24–48 h using a Columbus Instruments system. Body composition of conscious mice was measured by MRI (EchoMRI). All experimental procedures were approved by the institutional animal care committee.

Serum Chemistry Analysis

Serum was isolated by centrifugation (4,000 rpm for 15 min). Insulin was measured via ELISA (ImmunoDiagnostics, Ltd.). Metabolite levels were quantified using a commercial kit (BCAA, ab83374; Abcam) or ultra-performance liquid chromatography–tandem mass spectrometry (Metabo-Profile Biotechnology [Shanghai]). Serum nonesterified fatty acids (NEFAs) were measured with a Wako Diagnostics kit (633-52001). Liver tissue was homogenized in lysis buffer, and triglycerides (E1013; Applygen) and malondialdehyde (MDA) (A003-1-2; Nanjing Jiancheng) were quantified using commercial kits.

Primary Cell Isolation, Culture, and Differentiation

Stromal vascular fractions (SVFs) were isolated from inguinal WAT (iWAT) of 8-week-old mice and from human SAT and differentiated as described previously (22,23). In brief, mouse SVFs were induced for 48 h and maintained for 96 h, then switched to DMEM with 10% FBS on day 6 for downstream analyses. Human abdominal SAT SVFs were induced for 96 h and maintained for an additional 96 h. Browning was induced in mature human adipocytes by forskolin (10 μmol/L for 6 h).

Lentivirus and Adenovirus Transfection

Lentivirus was packaged as described (24) using pLKO.1-puro vectors carrying Idh3a-targeting shRNA (GGTGGTGTTCAGACAGTAATT). Lentivirus was transduced into beige adipocytes on day 3 of differentiation. For adenoviral knockdown, SVF-derived adipocytes (day 4) from Idh3a^F/F mice were infected with adenovirus-Cre (Adv-Cre) or control virus (OBiO Technology [Shanghai]).

Mitochondrial Stress Test and DNA Copy Number

Oxygen consumption rate (OCR) was measured using a Seahorse XFe96 Flux Analyzer (103015-100; Agilent). On day 5, adipocytes were seeded into 96-well plates and incubated in XF base medium (25 mmol/L glucose, 2 mmol/L pyruvate, 2 mmol/L glutamine) for 1 h. Following basal respiration assessment, oligomycin (2 μmol/L), carbonyl cyanide p-trifluoromethoxyphenylhydrazone (FCCP) (1 μmol/L), and rotenone/antimycin A (1 μmol/L) were sequentially injected. OCR was normalized to total protein content per well. Genomic DNA was extracted from mature adipocytes using QIAamp Fast DNA Tissue Kit (51404; QIAGEN). Quantitative RT-PCR was then used to determine the ratio of mitochondrial gene mt-Nd1 to nuclear gene Rbm15.

RNA Extraction, Gene Expression Analysis, and RNA Sequencing

Total RNA was extracted from adipocytes and tissues using TRIzol reagent (15596018; Invitrogen) according to the manufacturer’s instructions, and concentration was measured with NanoDrop (Thermo Fisher Scientific). cDNA was synthesized using PrimeScript RT Reagent Kit (RR047B; Takara). Quantitative RT-PCR was performed using SYBR qPCR Master Mix (Q511-02; Vazyme) on the Bio-Rad C1000. Primers are listed in Supplementary Table 4. RNA integrity was assessed using an Agilent 2100 Bioanalyzer (Agilent Technologies). Libraries were prepared with the TruSeq RNA Sample Prep Kit (RS-122-2001; Illumina) and sequenced on the NovaSeq 6000 platform at SinoTech Genomics. Reads were aligned to GRCm38.102, and differential gene expression was analyzed using edgeR with absolute fold change >1.5 and P < 0.05.

Western Blot Analysis

Adipocytes and tissues were lysed in radioimmunoprecipitation assay buffer with protease and phosphatase inhibitors. Equal amounts of protein were separated by 10% SDS-PAGE and transferred to membranes. Blots were incubated with antibodies against IDH3A (15909-1-AP; Proteintech), UCP1 (ab10983; Abcam), tubulin (T6199; Sigma-Aldrich), HSP90 (4874S; Cell Signaling Technology [CST]), and BCKDHA (90198T; CST). Secondary horseradish peroxidase–conjugated antibodies (7074 and 7076; CST) and ECL chemiluminescent regent (WBKLS0500; Millipore) were used for detection on a Bio-Rad imaging system.

Cold Tolerance Test

For acute cold exposure, 8-week-old male mice were housed at 30°C for 14 days, then exposed to 4°C. Rectal temperature was measured at 0–5 h postexposure. For prolonged cold exposure, 14-week-old male mice maintained on a standard chow diet were gradually exposed to 16°C for 1 day, 10°C for 1 day, and 4°C for 3 days. Rectal temperature was recorded at the end of the 3-day exposure to 4°C.

Glucose and Insulin Tolerance Tests

For the glucose tolerance test, mice fed an HFD for 16 weeks were fasted overnight for 15 h and injected intraperitoneally with glucose (1 g/kg body weight). For the insulin tolerance test, mice fed an HFD for 17 weeks were fasted for 6 h and received an intraperitoneal injection of insulin (1.8 units/kg body weight). Tail blood glucose was measured at 0–120 min using a glucometer (Roche).

Histology and Immunohistochemistry Analysis

Adipose and liver tissues were fixed in 4% paraformaldehyde, paraffin embedded, and stained with hematoxylin-eosin (G1076; Servicebio) or UCP1 antibody (ab10983; Abcam). Images were inspected and photographed using a Pannoramic Digital Slide Scanner (3DHISTECH).

Adeno-Associated Virus Production and Injection

Adeno-associated serotype 9 viruses expressing shRNA targeting Idh3a (AAV-shIdh3a) were obtained from OBiO Technology. Viral particles (2 × 10¹⁰ for chow diet, 4 × 10¹⁰ for HFD) were bilaterally injected into iWAT (23). Mice were analyzed 4 weeks postinjection.

Methyl-Capture Sequencing

Genomic DNA was extracted using the DNeasy Kit (QIAGEN). Libraries were prepared with the SureSelect XT Methyl-Seq Kit and hybridized to 84-Mb capture panels (∼3.7 million CpG sites). Bisulfite conversion was performed using an EZ DNA Methylation Kit (Zymo Research). Sequencing was conducted on a NovaSeq 6000 (SinoTech Genomics). Differential methylation was analyzed using a methylKit (false discovery rate <0.05, absolute methyl differentiation| ≥10%).

External Database Analysis

External validation was performed using the Adipose Tissue Knowledge Portal (25), which integrates clinical, transcriptomic, and proteomic data from >6,000 individuals across multiple cohorts and adipose depots. The portal’s summary section provides a weighted analysis of results across all cohorts for each trait. In the clinical module, Spearman correlations between candidate genes and clinical parameters are reported and grouped by anthropometric, circulating, and tissue-specific categories, with statistical significance and the number of contributing cohorts provided.

Statistical Analysis

Statistical analyses were conducted using GraphPad Prism 8 and SPSS version 27. Data are presented as mean ± SEM. Two-group comparisons used unpaired, two-tailed Student t tests. One-way ANOVA with Tukey post hoc test was used for comparisons among multiple groups, and two-way ANOVA with Bonferroni correction was applied for two-factor experiments. Correlations were assessed by Pearson or Spearman tests. Significance was defined as P < 0.05.

Data and Resource Availability

The data sets generated and/or analyzed in the current study are available from the corresponding authors upon reasonable request.

Results

IDH3A Expression in Human Visceral Fat Is Negatively Correlated With Metabolic Dysfunction Traits

In our bariatric surgery cohort (21), participants with high VFA exhibited significantly worse metabolic profiles than those with low VFA, despite similar BMI and age (Supplementary Table 2). VAT transcriptomic analysis showed that participants with high VFA had lower expression of thermogenesis- and TCA-related gene sets (Fig. 1A), suggesting suppression of these metabolic programs in association with metabolic disturbance.

Figure 1.

IDH3A expression in human visceral fat is negatively correlated with metabolic dysfunction traits. A: Schematic workflow comparing VAT transcriptomes between BMI- and age-matched individuals in the lower vs. upper quartiles of VFA, followed by gene set enrichment analysis (lower quartile: VFA <115.75 cm², n = 12; upper quartile: VFA >207 cm², n = 12). B: Venn diagram presenting the 38 shared upregulated genes derived from the upregulated genes in the iWAT of mice that were successively treated with CL316,243, exposed to cold, and subjected to swimming and in human adipocytes treated with forskolin. C: Gene coexpression analysis of IDH3A in human VAT. Genes that were positively coexpressed with IDH3A (Spearman r ≥0.4) were subjected to Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analysis. D–I: Spearman correlation of IDH3A mRNA expression in human VAT with BMI, fat percentage, fat mass, 2-h postprandial plasma insulin, 2-h C-peptide (CP), and fatty liver index (FLI) (n = 166). J: Spearman correlation of IDH3A mRNA expression in human subcutaneous and omental adipose depots with clinical parameters and serum biochemical indicators from Zhong et al. (25). Summary results represent a weighted analysis across all available cohorts for each trait. Gene expression levels are presented as fragments per kilobase of exon per million mapped reads (FPKM) derived from RNA-seq. Linear regression analysis and Spearman correlation analysis were conducted, and r and P values are depicted in the figure. **P < 0.01, ***P < 0.001, ****P < 0.0001. circ, circulating; CRP, C-reactive protein; FDR, false discovery rate; HOMA-IR, HOMA of insulin resistance; RT, room temperature; NES, normalized enrichment score; TG, triglyceride; WHR, waist-to-hip ratio.

We analyzed public RNA sequencing (RNA-seq) data sets of thermogenically stimulated adipose tissues (cold-treated, β3-agonist–treated, or exercise-treated mouse iWAT [Gene Expression Omnibus accession no. GSE86338] and forskolin-reated human adipocytes [Gene Expression Omnibus accession no. GSE129153]) to identify regulators of inducible thermogenesis. Across these conditions, overlapping upregulated genes were identified (Fig. 1B), including known thermogenic regulators Chchd10 and Apoo (26,27). Notably, IDH3A, a catalytic TCA enzyme, was similarly induced, prompting us to examine its relevance in our VAT cohort.

IDH3A mRNA was modestly higher in VAT from the low-VFA group (Supplementary Fig. 1A). Genome-wide coexpression analysis of the VAT transcriptome showed that genes positively correlated with IDH3A were enriched in thermogenic and metabolic regulatory pathways (Fig. 1C). Conversely, IDH3A expression in VAT was inversely correlated with multiple adverse metabolic indicators, including BMI, body fat percentage, fat mass, 2-h postprandial insulin, C-peptide levels, and fatty liver index (Fig. 1D–I). External database analyses (25) further supported these inverse relationships across different cohorts and adipose depots (Fig. 1J). These data link higher IDH3A expression to thermogenic gene programs and a more favorable metabolic profile.

IDH3A Expression Is Induced in Human and Mouse Adipocytes During Thermogenic Activation

We profiled IDH3A expression across tissues and found its enrichment in adipose depots, with the highest levels observed in brown adipose tissue (BAT), followed by iWAT (Supplementary Fig. 1B and C). Single nucleus RNA-seq (snRNA-seq) (28) of human WAT showed that IDH3A is predominantly expressed in adipocytes (Fig. 2A), with the highest levels in visceral adipocyte subclusters hAd4 and hAd6 (Fig. 2B) associated with lipogenic and thermogenic programs. To determine whether thermogenic signaling influences IDH3A expression, we treated human primary adipocytes with forskolin and observed a significant upregulation of IDH3A (Fig. 2C and D).

Figure 2.

Thermogenic stimulation induces IDH3A expression in white adipose depots. A: Dot plot of the expression of IDH3A across human WAT clusters from the Single Cell Portal. The size of the dot corresponds with the percentage of cells expressing IDH3A in each cluster, and the color represents the average expression level of IDH3A. B: Violin plot of the distribution of IDH3A in subpopulations of human visceral adipocytes. C: Relative IDH3A and UCP1 mRNA levels in human white adipocytes treated with forskolin (10 μmol/L) for 6 h (n = 3–4). D: Relative IDH3A and UCP1 protein levels in human white adipocytes treated with forskolin (10 μmol/L) for 6 h (n = 4). E: Dot plot of the expression of IDH3A across mouse WAT clusters. The size of the dot corresponds with the percentage of cells expressing IDH3A in each cluster, and the color represents the average expression level of IDH3A. F: Analysis of public small nuclear RNA-seq data (E-MTAB-8562) derived from Adipoq-tdTomato–positive adipocyte nuclei isolated from male mice housed at thermoneutrality, room temperature (RT), and after cold exposure at 8°C for 4 days (29). The colors in represent the expression of Idh3a (red), Ucp1 (blue), or both (purple). G: Relative Idh3a mRNA levels in iWAT of mice exposed to 4°C for 7 days (n = 6 mice). H: Relative IDH3A and UCP1 protein levels in iWAT of mice exposed to 4°C for 7 days (n = 5 mice). Bar graph data are mean ± SEM. **P < 0.01, ***P < 0.001 by unpaired, two-tailed Student t test. ASPC, adipose stem and progenitor cells; LEC, lymphatic endothelial cell; nk, natural killer; SMC, smooth muscle cell; UMAP, uniform manifold approximation and projection.

The murine snRNA-seq data set similarly showed Idh3a enriched in the adipocyte fraction of WAT (Fig. 2E). In an independent data set (E-MTAB-8562) (29), Idh3a expression closely paralleled Ucp1 induction during cold exposure (Fig. 2F), supporting its association with the thermogenic program. Moreover, chronic cold exposure in vivo markedly upregulated IDH3A in iWAT, with minimal change in BAT (Fig. 2G and H and Supplementary Fig. 1D and E), suggesting depot-specific inducibility. Collectively, these cross-species data show that IDH3A is adipocyte enriched and inducible by thermogenic stimulation, supporting a functional role in adipocyte thermogenic regulation.

Adipocyte-Specific Depletion of IDH3A Impairs Adaptive Thermogenesis and Energy Expenditure

To elucidate the functional role of IDH3A in adipose tissue, we generated Idh3a AKO mice (Supplementary Fig. 2A) and confirmed efficient IDH3A deletion in adipose tissues (Supplementary Fig. 2B–H). Chow diet–fed animals were first acclimated at thermoneutrality for 2 weeks and then housed individually at 22°C for indirect calorimetry. No differences were observed in food intake or locomotor activity between AKO and control mice (Fig. 3A and B). However, AKO mice exhibited significantly reduced heat production, Vo₂, and Vco₂, with unchanged respiratory exchange ratio (Fig. 3C–H and Supplementary Fig. 3A–D), indicating impaired systemic energy expenditure.

Figure 3.

Adipocyte-specific IDH3A deletion causes cold intolerance and reduced energy expenditure in mice. A and B: Eight-week-old male mice fed a chow diet were exposed to 30°C for 14 days and then were individually placed in the metabolic cages at 22°C for 24 h. Food intake (A) and locomotor activity (B) were measured (n = 5). C–H: Whole-body heat production (C and D), Vo₂ (E and F), and Vco₂ (G and H) of mice in A and B were analyzed (n = 5 mice). I: Eight-week-old male mice fed a chow diet were exposed to 30°C for 14 days and then were placed at 4°C for acute cold challenge, with rectal temperature recorded at indicated times (Idh3a^F/F mice, n = 8; Idh3a AKO mice, n = 9). J: Fourteen-week-old male mice fed a chow diet were first exposed to 16°C for 1 day, followed by exposure to 10°C for 1 day, then 4°C for 3 days. Rectal temperature was measured after 3 days of exposure to 4°C (n = 6 mice). K and L: Hematoxylin-eosin staining of iWAT from mice exposed to 4°C for 3 days (K), and immunohistochemistry staining of UCP1 in iWAT (L) (scale bar = 100 μm). UCP1-positive staining area was quantified using ImageJ in a blinded manner (three mice per group; three randomly selected nonoverlapping fields per mouse; total nine fields per group). M: Relative mRNA levels of thermogenesis and adipogenesis-related genes in iWAT (Idh3a^F/F mice, n = 7; Idh3a AKO mice, n = 9). N: The protein levels of IDH3A and UCP1 within iWAT from mice exposed to 4°C for 3 days. Data in C, E, G, and I were analyzed using two-way ANOVA with Bonferroni multiple comparisons test; other data were analyzed by unpaired, two-tailed Student t test. Bar graph data are mean ± SEM. *P < 0.05, **P < 0.01, ***P < 0.001.

During an acute cold challenge, AKO mice tended to exhibit lower core body temperatures than controls (Fig. 3I). Under prolonged cold exposure, this difference became more pronounced (Fig. 3J). Histological analysis of iWAT revealed that AKO fat depots were predominantly unilocular, whereas controls displayed multilocular beige adipocytes (Fig. 3K). Accordingly, UCP1 expression was markedly reduced (Fig. 3L–N), along with other thermogenic markers, while the adipogenic marker remained unchanged (Fig. 3M). Only a modest reduction in thermogenic gene expression was observed in BAT (Supplementary Fig. 4A and B), suggesting that IDH3A is more critical for beige fat thermogenesis than for BAT.

Local IDH3A knockdown in iWAT via AAV-shIdh3a similarly blunted thermogenesis. After 3 days of cold exposure, AAV-shIdh3a–treated iWAT appeared visually whitened and showed significantly reduced thermogenic gene expression (Supplementary Fig. 5A–E), recapitulating the AKO phenotype. Thus, IDH3A is required for maintaining beige fat thermogenic capacity and energy expenditure during cold adaptation.

IDH3A Deficiency Exacerbates Diet-Induced Metabolic Dysfunction

AKO mice fed a 60% HFD for 22 weeks (Supplementary Fig. 6A) had similar body weight gain and food intake to controls (Fig. 4A and B), while fat mass showed a trend toward being higher (Fig. 4C). Paradoxically, the weights of dissected iWAT and epididymal WAT were reduced. The downregulation of lipogenic genes in WAT suggests limited lipid storage in WAT and ectopic lipid redistribution, consistent with the increased liver weight in AKO mice (Fig. 4D and Supplementary Fig. 6B). Furthermore, AKO mice displayed whitening of BAT and hepatic lipid accumulation, and their WAT contained a higher proportion of small adipocytes (Fig. 4E and Supplementary Fig. 6C). These small adipocytes did not show elevated thermogenic markers (Supplementary Fig. 6D), indicating that they did not acquire a beige phenotype. In AKO WAT, inflammatory marker expression remained normal, but fibrosis-related transcripts were increased, further indicating impaired adipose expansion (Supplementary Fig. 6E). Consistently, circulating NEFAs, hepatic triglycerides, and MDA were significantly elevated, indicative of increased ectopic lipid burden and hepatic lipid peroxidation (Supplementary Fig. 7A and B and Fig. 4F).

Figure 4.

Adipocyte-specific IDH3A deletion exacerbates HFD-induced metabolic dysfunction. A: Body weight curve of Idh3a^F/F and Idh3a AKO mice fed an HFD (Idh3a^F/F mice, n = 7; Idh3a AKO mice, n = 9). B: Food intake of mice in A (Idh3a^F/F mice, n = 7; Idh3a AKO mice, n = 8). C: The average fat and lean masses of Idh3a^F/F and Idh3a AKO mice after 22 weeks of HFD (Idh3a^F/F mice, n = 8; Idh3a AKO mice, n = 8). D: The weights of BAT, iWAT, epididymal WAT [eWAT], and liver in Idh3a^F/F and Idh3a AKO mice after 22 weeks of HFD (Idh3a^F/F mice, n = 6; Idh3a AKO mice, n = 8). E: Hematoxylin-eosin staining of BAT and liver in mice after 22 weeks of HFD (scale bar = 100 μm). F: Hepatic triglyceride (TG) levels in mice (Idh3a^F/F mice, n = 7; Idh3a AKO mice, n = 8). G and H: Glucose tolerance test (GTT) conducted in mice after 15 weeks of HFD and its area under the curve (AUC) (n = 6 mice). I and J: Insulin tolerance test (ITT) conducted in mice after 18 weeks of HFD and its AUC (Idh3a^F/F mice, n = 7; Idh3a AKO mice, n = 8). K: Serum insulin levels in mice fasted for 16 h (Idh3a^F/F mice, n = 7; Idh3a AKO mice, n = 7). L–Q: Male mice after 14 weeks of HFD were individually placed in the metabolic cages at 22°C for 24 h and then intraperitoneally injected with CL316,243 (1 mg/kg body weight), followed by monitoring for another 24 h. Whole-body Vo₂ (L), Vco₂ (N), and heat production (P) of mice were analyzed. Regression-based analyses of absolute Vo₂, Vco₂, and heat production in relation to body weight (M, O, and Q) following intraperitoneal injection with CL316,243 were conducted (Idh3a^F/F mice, n = 5; Idh3a AKO mice, n = 6). Data in A, G, I, L, N, and P were analyzed using two-way ANOVA with Bonferroni multiple comparisons test; data in M, O, and Q were analyzed using two-sided ANCOVA; other data were analyzed by unpaired, two-tailed Student t test. Bar graph data are mean ± SEM. *P < 0.05, **P < 0.01.

AKO mice exhibited a slight impairment in glucose tolerance but a markedly worsened insulin tolerance, with elevated fasting insulin indicative of insulin resistance (Fig. 4G–K). Upon β3-adrenergic stimulation with CL316,243, AKO mice exhibited a blunted increase in Vo₂, Vco₂, and heat generation compared with controls, while the respiratory exchange ratio remained unchanged (Fig. 4L–Q and Supplementary Fig. 8). These results suggest that IDH3A contributes to metabolic homeostasis and adrenergic thermogenic responsiveness during HFD.

AAV-mediated IDH3A knockdown in mice fed an HFD (Supplementary Fig. 9A) recapitulated key aspects of the AKO phenotype, including increased liver weight, reduced WAT mass, BAT whitening, more severe hepatic steatosis, and worsened insulin resistance with a trend toward glucose intolerance (Supplementary Fig. 9B–F). These findings reinforce the key role for IDH3A in preserving adipose tissue function and metabolic homeostasis during chronic nutrient excess.

IDH3A Is Essential for Beige Adipocyte Thermogenesis In Vitro

To determine whether IDH3A regulates adaptive thermogenesis in a cell-autonomous manner, we conducted in vitro loss-of-function experiments. SVFs were isolated from iWAT of Idh3a^F/F mice, differentiated into mature beige adipocytes, and transduced with Adv-Cre to achieve IDH3A knockdown (Fig. 5A). IDH3A knockdown did not impair adipogenesis, as the expression of adipogenic markers (Cebpa, Pparg, and Fabp4) remained unchanged (Fig. 5B and C). Thermogenic genes (Ucp1, Ppargc1α, Cidea, and Cox6a) were significantly downregulated in IDH3A-deficient adipocytes (Fig. 5D), and UCP1 protein levels were markedly reduced (Fig. 5E), indicating an attenuation of the thermogenic program.

Figure 5.

IDH3A deficiency attenuates the thermogenic program in beige adipocytes. A: Schematic illustration of differentiation of beige adipocytes in vitro. B: Bright field images of primary beige adipocytes with IDH3A knockdown and control. C: Relative mRNA levels of adipogenesis-related genes in primary beige adipocytes with IDH3A knockdown and control (n = 4). D: Relative mRNA levels of thermogenesis-related genes in primary beige adipocytes with IDH3A knockdown and control (n = 4). E: The protein levels of UCP1 in primary beige adipocytes with IDH3A knockdown and control. F–H: OCR of primary beige adipocytes with IDH3A knockdown and control (n = 8). Data were analyzed by unpaired, two-tailed Student t test. Bar graph data are mean ± SEM. *P < 0.05, **P < 0.01, ***P < 0.001. D, day; con, control.

Consistent with the transcriptional changes, IDH3A-deficient adipocytes exhibited impaired mitochondrial respiration, evidenced by a lower FCCP-stimulated maximal OCR (Fig. 5F–H). Notably, mitochondrial content and biogenesis markers were unchanged (Supplementary Fig. 10A and B), indicating reduced oxidative capacity without loss of mitochondrial abundance.

IDH3A Deficiency Disrupts Systemic BCAA Homeostasis

Given the thermogenic and metabolic defects observed with IDH3A ablation, we next explored potential underlying molecular mechanisms. IDHs have been implicated in epigenetic regulation of gene expression through effects on αKG availability and on one-carbon metabolism that influence methyl-donor supply (30,31). We therefore asked whether IDH3A deficiency alters metabolites involved in epigenetic regulation under thermogenic stimulation. Targeted metabolomics of isoproterenol-stimulated beige adipocytes revealed a significant reduction in intracellular αKG in IDH3A-deficient cells compared with controls (Supplementary Fig. 11).

We next performed RNA-seq and methyl-capture sequencing (MC-seq) in control versus IDH3A-deficient beige adipocytes. Transcriptomic analysis identified 801 downregulated genes (Fig. 6A). MC-seq revealed hypermethylation affecting 3,298 genes (Fig. 6B). Notably, 86 genes were both repressed and hypermethylated in IDH3A-deficient adipocytes (Fig. 6C), and these overlapping targets were enriched in the valine, leucine, and isoleucine (BCAA) degradation pathway (Fig. 6D).

Figure 6.

Adipocyte-specific IDH3A ablation disrupts systemic BCAA homeostasis. A–C: Beige adipocytes with IDH3A knockdown (KD) were subjected to RNA-seq (A) and MC-seq (B) on day 6 of differentiation. The Venn diagram (C) depicts the 86 shared genes derived from RNA-seq and MC-seq (n = 3). D: Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analysis was conducted for the 86 shared genes in C. E: Serum BCAA levels in mice exposed to 4°C for 3 days analyzed by commercial kits (Idh3a^F/F mice, n = 7; Idh3a AKO mice, n = 8). F–H: Serum valine, leucine, and isoleucine levels in mice exposed to 4°C for 3 days analyzed by ultra-performance liquid chromatography–tandem mass spectrometry (n = 6 mice). I: Serum BCAA levels in mice following 22 weeks of HFD were analyzed by commercial kits (Idh3a^F/F mice, n = 7; Idh3a AKO mice, n = 8). Data were analyzed by unpaired, two-tailed Student t test. Bar graph data are mean ± SEM. *P < 0.05, **P < 0.01, ***P < 0.001. Con, control; meth.diff, methylation differentiation.

Thermogenic adipose tissue contributes to systemic BCAA clearance during cold challenge (10). In line with this, circulating valine, leucine, and isoleucine levels were elevated in AKO mice (Fig. 6E–H). The AAV-mediated knockdown model showed a similar increase (Supplementary Fig. 12A). Likewise, serum BCAA levels were also elevated in AKO mice after 22 weeks of HFD (Fig. 6I).

IDH3A Modulates the Thermogenic Program via Bckdha-Mediated BCAA Catabolism

Adipose tissue can influence systemic BCAA homeostasis in a depot- and state-dependent manner (32). In IDH3A-deficient beige adipocytes, BCAA catabolic enzyme genes were markedly downregulated, whereas BCAA transporter genes remained largely unchanged (Fig. 7A). Similarly, iWAT from both chronically cold-exposed AKO and AAV-shIdh3a mice showed reduced expression of BCAA catabolic genes, notably Bckdha and Bckdhb (Fig. 7B and Supplementary Fig. 12B). Consistent with this, metabolomic analysis revealed trends toward accumulation of BCKAs, with increased intracellular KIV in IDH3A knockdown beige adipocytes and elevated circulating KMV in cold-exposed Idh3a AKO mice (Supplementary Fig. 13A and B).

Figure 7.

IDH3A modulates the thermogenic gene program via BCKDHA-mediated BCAA catabolism. A: Heat map of the genes related to BCAA degradation in beige adipocytes with IDH3A knockdown (n = 3). B: Relative mRNA levels of the genes related to BCAA catabolism in iWAT from mice exposed to 4°C for 3 days (Idh3a^F/F mice, n = 8; Idh3a AKO mice, n = 9). C and D: The percentage of DNA methylation at individual methylated cytosine sites of Bckdha (C) and Bckdhb (D) in the iWAT of Idh3a^F/F and AKO mice was analyzed through pyrosequencing (n = 4 mice). E and F: OCR of primary beige adipocytes with IDH3A knockdown and BCKDHA overexpression (n = 6). G: The protein levels of IDH3A, UCP1, and endogenous BCKDHA in beige adipocytes with IDH3A knockdown and BCKDHA overexpression (n = 3). Data in E–G were analyzed using one-way ANOVA with Tukey multiple comparisons test; other data were analyzed by unpaired, two-tailed Student t test. Bar graph data are mean ± SEM. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. Con, control; OE, overexpression; NC, normal control.

BCKDHA and BCKDHB encode the E1 subunits of the BCKDH complex, which catalyzes the rate-limiting step in BCAA catabolism (33). Promoter hypermethylation typically represses transcription (34); therefore, we examined DNA methylation status of these genes. In iWAT from cold-exposed Idh3a AKO mice, the Bckdha promoter was significantly hypermethylated, whereas Bckdhb methylation was unchanged (Fig. 7C and D). These findings suggest that IDH3A may regulate BCKDHA expression by modulating its promoter methylation status, thereby influencing BCAA catabolism.

We then tested whether restoring BCKDHA in IDH3A-deficient adipocytes could rescue thermogenic function. BCKDHA overexpression partially increased mitochondrial respiration and UCP1 protein levels (Fig. 7E–G). Furthermore, in human VAT, IDH3A expression positively correlated with the expression of BCAA catabolic genes (Fig. 8A–E). Collectively, these data identify IDH3A as a crucial regulator of BCAA catabolism, operating primarily through BCKDHA, and highlight its importance in sustaining beige adipocyte thermogenesis and systemic energy metabolism.

Figure 8.

IDH3A expression correlates with BCAA catabolic gene expression in human VAT. A–E: Pearson correlation analysis of IDH3A expression in human VAT with the genes related to BCAA catabolism (BCAT2, BCKDHA, BCKDHB, DLD, and DBT) (n = 236). F: Illustration of a working model in which adipocyte IDH3A deficiency is associated with increased Bckdha promoter methylation and reduced BCKDHA expression, consistent with suppression of the adipose BCAA catabolic program and elevated circulating BCAA levels. Under cold exposure or β3-adrenergic stimulation, IDH3A deficiency is accompanied by downregulation of the thermogenic gene program and mitochondrial respiration (with UCP1 shown as a representative readout) and reduced adrenergic thermogenic response and energy expenditure. During HFD feeding, adipocyte IDH3A deficiency is associated with elevated serum BCAA levels and adverse metabolic outcomes, including increased insulin resistance and hepatic lipid accumulation. Gene expression levels are presented as fragments per kilobase of exon per million mapped reads (FPKM) derived from RNA-seq. Linear regression analysis and Pearson correlation analysis was conducted, and r and P values are depicted in the figure. Created with BioRender.com.

Discussion

In this study, IDH3A expression in human VAT is associated with more favorable metabolic traits. IDH3A deficiency compromises thermogenic capacity and exacerbates diet-induced metabolic dysfunction in association with reduced BCKDHA expression and impaired BCKDHA-dependent BCAA catabolism. Restoring BCKDHA partially rescued thermogenic defects in IDH3A-deficient adipocytes. Consistent with these findings, we identify IDH3A as a previously unrecognized regulator of beige fat thermogenesis and metabolic fitness under stress.

Accumulating evidence suggests that human VAT retains a degree of thermogenic and metabolic plasticity (35–37). Consistent with this, our bulk RNA-seq analysis of human VAT revealed that increased VFA and poorer metabolic status were associated with coordinated suppression of thermogenesis and TCA gene programs. Moreover, we found a consistent correlation between IDH3A levels and thermogenic gene expression, as well as key clinical metabolic indicators. In the mouse, adipocyte-specific loss of IDH3A impaired cold-induced thermogenesis and energy expenditure, further supporting the critical role of IDH3A in maintaining mitochondrial function and adaptive thermogenesis. Future research should elucidate the precise functions and regulatory mechanisms of IDH3A in human adipocytes.

Mitochondria are traditionally seen as bioenergetic organelles, but growing evidence has shown that mitochondrial metabolism also shapes adaptive programs like thermogenesis (38). For example, cardiolipin synthase deficiency compromises the adipocyte thermogenic program by modulating an endoplasmic reticulum stress response factor that inhibits nuclear transcription (39). Likewise, ATP-citrate lyase facilitates WAT browning by preventing TCA overload and metabolic stress (40). In light of these findings, our study identifies IDH3A, a canonical TCA enzyme, as a novel regulator of thermogenic programming. Both genetic knockout and cell-autonomous experiments demonstrated that IDH3A is essential for maintaining thermogenic competence in beige adipocytes, supporting the notion that mitochondrial enzymatic function extends beyond intermediary metabolism to directly modulate thermogenic gene expression. Notably, the requirement for IDH3A in thermogenic regulation differed between adipose depots. Beige adipocytes in iWAT depended on IDH3A for inducible, adaptive thermogenesis, whereas loss of IDH3A in BAT caused only modest impairment of thermogenic activity. This contrast likely reflects fundamental differences in regulatory architecture, as BAT is constitutively thermogenic with high basal UCP1 expression and mitochondrial content, while beige adipocytes require dynamic metabolic reprogramming in response to external stimuli (8,41). Similar depot-specific differences have been observed with other thermogenic regulators, such as PRDM16 and SOX4 (6,42).

Integrated transcriptomic and DNA methylation analyses in beige adipocytes revealed that IDH3A knockdown repressed genes involved in BCAA catabolism, accompanied by Bckdha promoter hypermethylation. Consistent with these associations, IDH3A-deficient mice exhibited elevated serum BCAA levels after chronic cold exposure, indicating impaired systemic BCAA homeostasis. These findings identify IDH3A as a novel regulator of BCAA metabolism, connecting nutrient use to thermogenic capacity. The role is supported by emerging evidence that BCAA catabolism in thermogenic adipose tissue is highly condition dependent. During cold exposure, BAT from chow diet–fed mice enhances BCAA uptake, and genetic disruption of mitochondrial BCAA import/catabolism impairs thermogenesis and systemic BCAA clearance (9). However, it should be noted that recent in vivo flux analyses indicated that despite this high uptake, BCAA-derived carbon contributes modestly to BAT TCA fueling during thermogenesis compared with fatty acids and glucose (43). This implies that BCAA metabolism in thermogenic adipocytes acts more as an inducible metabolic module, coupling thermogenic activation to broader mitochondrial and metabolite homeostasis.

Under chronic nutrient excess, BCAA metabolism may take on additional roles beyond thermogenesis. Recent studies have emphasized that BAT BCAA metabolism involves substantial nitrogen flux and the synthesis of BCAA-derived metabolites that impact systemic metabolic health. HFD/obesity attenuate this pathway, compromising the beneficial effects of BCAA metabolism in adipose tissue (44). Elevated circulating BCAAs correlate with obesity, insulin resistance, and cardiovascular disease (45–47), and their reduction correlates with improved metabolic outcomes (48). This dysregulation in BCAA metabolism is closely related to our HFD phenotype that included worsened insulin sensitivity and a lipid overflow signature (elevated NEFAs and increased hepatic triglycerides and MDA). Meanwhile, BCAAs can directly contribute carbon to lipogenic precursor pools in adipocytes, supporting de novo lipogenesis and lipid storage (49). Thus, repression of BCAA metabolism under IDH3A deficiency likely impairs WAT lipid synthesis/expandability during prolonged HFD, contributing to elevated lipid ectopic accumulation, and the discrepancy between a trend toward increased whole-body fat mass and reduced weights of discrete WAT depots. Together, these findings illustrate the role of IDH3A in regulating BCAA metabolism to maintain metabolic homeostasis, linking thermogenesis and lipid metabolism under conditions of nutrient overload.

A limitation of our human cohort analysis is the sex imbalance in the extreme quartile VAT subgroup, which was predominantly female and possibly reflects the bariatric surgery recruitment setting. Our transcriptomic and DNA methylation analyses highlighted suppression of the BCAA catabolic program following IDH3A loss, but alternative downstream targets were not fully explored. Additionally, isotope tracing or flux analyses were not performed to directly measure the contribution of BCAA-derived carbon-to-mitochondrial oxidation. Finally, future studies integrating stable isotope flux approaches with chromatin profiling will be required to elucidate how IDH3A couples metabolism to epigenetic remodeling.

In summary, IDH3A emerges as a crucial regulator of beige adipocyte thermogenesis and systemic metabolic homeostasis under metabolic stress. Mechanistically, IDH3A preserves BCKDHA expression by preventing hypermethylation at the BCKDHA promoter, linking mitochondrial metabolism to adaptive thermogenesis. Collectively, these findings highlight IDH3A as a promising therapeutic target for obesity and related metabolic disorders.

This article contains supplementary material online at https://doi.org/10.2337/figshare.31371067.

Funding Statement

This study was financially supported by National Natural Science Foundation of China grants 82370851, 82270906, and 82300980.