Gut bacteria enriched in obese mice produce leucic acid to counter obesity through HCAR2

Gut bacteria enriched in obese mice produce leucic acid to counter obesity through HCAR2 | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Article Gut bacteria enriched in obese mice produce leucic acid to counter obesity through HCAR2 Shu-Yong Lin, Zhi-Yu Chen, Kan Wang, Hao-Yu Qin, Yang Yang, Dong-Tai Liu, and 29 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-8045609/v1 This work is licensed under a CC BY 4.0 License Status: Under Review Version 1 posted You are reading this latest preprint version Abstract Gut microbiota modulates a wide range of physiological processes in the host. However, whether host genes shape microbial communities to affect host physiology remains poorly explored. Using Aida -/- mice that develop obesity due to increased intestinal fat absorption, we identify two Clostridium strains enriched in their fat-depleted gut environment. We demonstrate that the growth of these strains is suppressed by oleate both in vivo and in vitro , mirroring their selective expansion. Paradoxically, these strains protect against diet-induced obesity. Integrated metabolomics and 13 C-tracing reveal that these strains biosynthesise leucic acid, whose administration reduces adiposity and enhances lean mass and exercise capacity by suppressing lipid synthesis and promoting catabolism. By combining computational prediction with transcriptomic and human genetic evidence, we identify HCAR2 as a candidate receptor for leucic acid and confirm this by demonstrating direct binding and showing that HCAR2 is required for metabolic effects of leucic acid in vitro and in adipose tissue in vivo . We thus establish a homeostatic host-microbiome feedback loop wherein a host gene-defined metabolic niche selects for microbes that produce a signalling metabolite, which in turn activates a physiologically relevant receptor to suppress obesity. Biological sciences/Physiology/Metabolism/Fat metabolism Biological sciences/Physiology/Metabolism/Homeostasis Biological sciences/Microbiology/Bacteria/Bacterial host response Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 Figure 6 Main Coevolution between hosts and their resident microbes has fostered a mutual dependence 1-3 , in which the physiology in the host affects microbial communities, and vice versa 4-7 . Environmental perturbations, such as diet 8 , antibiotic exposure 9 and cold stress 10 , can elicit changes in this symbiosis and trigger a wide range of physiological changes in the host, including alterations in lipid metabolism, immune function and energy balance. In addition, emerging evidence shows that host genetics may have a critical impact on the microbiome composition 11-15 . For example, genome-wide association studies (GWAS) have linked the ABO blood group and FUT2 polymorphisms with luminal acetylgalactosamine (GalNAc) dynamics, which enrich Faecalibacterium prausnitzii 16-19 . Similarly, intestinal HIF-2α regulates lactate levels to influence Bacteroides vulgatus and Ruminococcus torques 20 . These studies have emphasised luminal metabolites as central mediators of the crosstalk between host genetic background and gut microbes. However, mechanistic links connecting host genetic background with resulting phenotypes and microbial composition/functions remain poorly defined. We previously established Aida as an anti-thrifty gene that limits intestinal lipid absorption and enhances energy expenditure to counter fat-induced obesity 21,22 . We hypothesised that fat depletion from the chyme resulting from Aida deletion would create an environment that favours specifically adapted bacterial taxa. Here, through systematic culturing of Aida -/- cecal content, we isolated two hitherto unidentified anaerobes, designated lipid-depletion-enriched (LDE)-1 (an Oscillospira strain) and LDE-2 (a tentative new genus within Lachnospiraceae ), both exhibiting oleate-sensitive growth suppression. Intriguingly, these strains function as feedback metabolic regulators, as administering either strain attenuated diet-induced obesity, suggesting a host-elicited microbial adaptation to counteract lipid dysregulation in Aida -/- mice. Metabolomic analysis revealed that both strains biosynthesise leucic acid (α-hydroxyisocaproic acid or 2-hydroxy-4-methylpentanoic acid) via conserved 2-hydroxyisocaproate dehydrogenases (HicDHs), which are responsible for the anti-obesity effects. Mechanistically, we identified the membrane receptor hydroxycarboxylic acid receptor 2 (HCAR2) as the receptor for leucic acid, which signals to suppress adipose lipid anabolism and promote catabolism. These findings define a feedback host-microbe-metabolite loop, in which host metabolic defects resulting from a genetic alteration reshape the gut microbiome, leading to the production of specific metabolites that rebalance host metabolism. Results AIDA depletion enriches specific gut bacterial species linked to adiposity To determine whether Aida depletion alters gut microbiota composition in the presumably altered chyme, we performed 16S rRNA gene sequencing on cecal contents from high-fat diet (HFD)-fed WT and Aida -/- mice (Supplementary Table 1). The overall bacterial composition at the phylum level did not change significantly in Aida -/- mice (Extended Data Fig. 1a). We also determined the richness and evenness of the bacterial species and found no significant differences (Extended Data Fig. 1b-f). We then assessed the beta diversity of operational taxonomic units (OTUs) using principal coordinate analysis (PCoA). Again, the Aida -/- samples exhibited a similar bacterial profile to that of the WT group (Fig. 1a). These data indicate that the deletion of Aida does not systematically disturb gut microbiota. However, we observed that 6 out of the 271 OTUs identified in the microbiota profiling exhibited significant abundance changes in Aida -/- mice (Fig. 1b). OTU68 ( Anaerotruncus colihominis ), OTU209 (unclassified member of the Lachnospiraceae family), and OTU228 (unclassified member of the genus of Clostridium )were enriched, while OTU21 ( Holdemania massiliensis ), OTU59 ( Adlercreutzia mucosicola ), and OTU141 (unclassified members of the genus of Ruminiclostridium ) were decreased (Fig. 1b and Extended Data Fig. 1g). Strikingly, analysis of metagenomic data from 355 normal-weight and 631 obese human fecal samples (Supplementary Table 2) revealed that OTU68 ( A. colihominis ) was significantly less abundant in obese compared to normal-weight individuals (Fig. 1c) and inversely correlated with body mass index (BMI) in obesity (Fig. 1c). Species phylogenetically related to other Aida -related OTUs, including Holdemania filiformis for OTU21 and Adlercreutzia equolifaciens for OTU59, showed no BMI association despite being mildly reduced in obesity (Extended Data Fig. 1h,i). In mice, OTU209 abundance in cecal contents strongly correlated with fat mass in Aida -/- background (Pearson r = 0.9369, P = .0059) (Fig. 1d; Extended Data Fig. 1j). These data suggest that OTU68 and OTU209 are associated with the development of obesity. To isolate bacterial strains linked to Aida -associated metabolic phenotypes, we cultured cecal content from HFD-fed Aida -/- mice under stringent anaerobic conditions (0% O 2 ), targeting the obligate anaerobes OTU68 and OTU209. Using nine distinct media, we obtained 254 isolates, resolving 30 taxonomically characterised species and more than 20 novel species based on 16S rRNA sequence identity thresholds (Extended Data Fig. 1k; Supplementary Table 3). Among these, strains LDE-1, which represents OTU68, and LDE-2, which represents OTU209, were prioritised for further functional studies. Scaffolded genome sequencing revealed that LDE-1 represented a new strain of A. colihominis (family Oscillospiraceae ; Supplementary note 1), while LDE-2 represented a putative new genus within the family Lachnospiraceae (Supplementary note 2). Whole-genome phylogeny corroborated these classifications, confirming the novelty of LDE-2 at the genus level. To define how Aida depletion reshapes the luminal lipid microenvironment to favour these strains, we quantified cecal fatty acid profiles via GC-MS. This revealed a significant reduction in oleate (C18:1) and palmitate (C18:0) in Aida -/- mice compared to WT controls (Fig. 1e). In contrast, laurate (C12:0), myristate (C14:0), stearate (C18:0), arachidate (C20:0), and arachidonate (C20:4) did not show significant change, indicating selective depletion of specific luminal LCFAs. Strikingly, in vitro growth assays demonstrated that oleate potently inhibited the growth of LDE-1 and LDE-2 at concentrations as low as 10 μM. As a comparison, Bacteroides uniformis, which did not show changes in abundance in the gut, remained resistant to oleate (Fig. 1f). In comparison, palmitate, stearate, and myristate, as well as short-chain (acetate, propionate, and butyrate) and medium-chain fatty acids (caprylate and decanoate), exhibited no suppressive effect (Fig. 1g-j). Critically, in vivo validation via gavage of oleate-rich olive oil recapitulated these findings, markedly reducing the abundance of OTU68 and OTU209 in Aida -/- mice (Fig. 1k). These data demonstrated that accelerated intestinal lipid absorption in Aida -/- mice depletes luminal LCFAs, particularly oleate, thereby alleviating the growth suppression of lipid-sensitive anaerobes and promoting their enrichment. LDE-1 and LDE-2 ameliorate diet-induced obesity and metabolic dysfunction in mice We next investigated whether the obesity-associated enrichment of LDE-1 and LDE-2 reflects a maladaptive bystander effect or a protective feedback response. To this end, we administered these strains to HFD-fed WT mice with prior subjection to antibiotic (ABX)-treatment (Fig. 2a). Strikingly, monocolonization with either strain robustly attenuated body weight gain, with LDE-1- and LDE-2-treated mice exhibiting a 10% reduction relative to controls at 8 weeks post-gavage (Fig. 2b). In contrast, the probiotic Escherichia coli Nissle 1917 (EcN) showed no significant anti-obesity effect (Extended Data Fig. 2a), underscoring the specificity of LDE-1 and LDE-2. At 5 weeks post-gavage, LDE-1 and LDE-2 administration markedly reduced gonadal white adipose tissue (gWAT) and inguinal WAT (iWAT) mass compared to controls (Fig. 2c,d). Histological analysis confirmed marked reductions in adipocyte hypertrophy in both gWAT and iWAT (Fig. 2e-h and Extended Data Fig. 2b-e). Notably, neither strain induced muscle wasting. LDE-2 even increased soleus muscle mass by 18% (Fig. 2i,j). Treadmill testing revealed a significant increase in running distance and prolongation of time to exhaustion in LDE-2-treated mice (Fig. 2k,l), while both strains improved grip strength (Fig. 2m). Importantly, there were no substantial changes in serum alanine transaminase (ALT) or blood urea nitrogen (BUN) levels (Fig. 2n,o), suggesting these strains did not cause liver or renal damage. In addition, food intake among the groups was similar (Fig. 2p), indicating that the observed anti-obesity effects are not caused by a consequence of bacterial gavage-induced anorexia. These results demonstrated that LDE-1 and LDE-2 uncouple obesity resistance from muscle wasting, a critical limitation of conventional weight-loss therapies. We further explored the long-term effects of LDE-2, which appeared to exert more beneficial effects than LDE-1, in mice maintained on HFD for 16 weeks following the last gavage. Notably, LDE-2 sustained 30% lower body weight gain (Extended Data Fig. 2f) and improved fasting glycemia (Fig. 2q). Hepatic steatosis was attenuated, with 22% reduced liver weight and steatosis scores reduced to 0 (Fig. 2r-t). A mild increase in rectus femoris muscle mass further underscored its musculoskeletal benefits (Extended Data Fig. 2g). Leucic acid is increased by LDE-1 and LDE-2 strains To identify factors mediating the metabolic effects of LDE-1 and LDE-2, we performed untargeted metabolomics on pre- and post-culture media using ultra-high-performance liquid chromatography-quadrupole time-of-flight mass spectrometry (UHPLC-QTOF-MS). Focusing on metabolites commonly upregulated by both strains, we identified six metabolites that were upregulated in both LDE-1 and LDE-2 cultures, including leucic acid, butyric acid, N-methyltryptamine, thymine, 4-hydroxy-3-methoxycinnamaldehyde and Tyr-Pro (Fig. 3a-c; Supplementary Table 4,5). To validate the in vivo enrichment of these metabolites, we analysed cecal contents of HFD-fed mice using gas chromatography-mass spectrometry (GC-MS). After testing trimethylsilyl (TMS), tert -butyldimethylsilyl (TBDMS), and sequential (TMS followed by trifluoroacetyl) derivatization approaches, we selected TMS based on peak resolution, sensitivity, and reproducibility for leucic acid, the most markedly changed metabolite, while maintaining compatibility with the detection of a broad spectrum of other metabolites (Extended Data Fig. 3a). Indeed, Aida -/- mice exhibited higher leucic acid abundance compared to WT controls (Extended Data Fig. 3b,c), confirming that AIDA depletion drives accumulation of this microbially-derived metabolite. Critically, structurally related hydroxy acids (2-hydroxybutyric acid, 3-hydroxybutyric acid, and 2-hydroxy-3-methylbutyric acid) and leucine, the precursor for leucic acid, were unchanged in Aida -/- mice (Extended Data Fig. 3d-g). To assess non-hydroxylated SCFAs, we used derivatisation-independent GC-MS, and found that butyric acid together with propionic acid, pentanoic acid, isobutyric acid, 2-methylbutanoic acid, and isovaleric acid, remained unchanged or even decreased (Extended Data Fig. 3h-m), underscoring the specificity of leucic acid induction among SCFA family. In addition, N-methyltryptamine, thymine, 4-hydroxy-3-methoxycinnamaldehyde, and Tyr-Pro were not detectable using either approach, consistent with their much lower confidence score compared to leucic acid and butyric acid in the untargeted LC-MS data (Supplementary Table 4,5). These data establish that AIDA-deficient mice enrich LDE-1 and LDE-2 that produce leucic acid. Leucic acid phenocopies the beneficial effects of LDE-1 and LDE-2 in mice To determine whether leucic acid mediates the anti-obesity effects of LDE-1 and LDE-2, we supplemented drinking water with 50 mM leucic acid, a dose in line with prior studies of branched-chain amino acid derivatives and short-chain fatty acids 23-27 . Leucic acid robustly attenuated HFD-induced weight gain in both WT and Aida -/- mice (Extended Data Fig. 4a). Strikingly, WT mice receiving leucic acid maintained body weights comparable to those of normal chow diet (NCD)-fed controls, achieving a 17% reduction in adiposity by week 12 (Fig. 3d). Consistent with the impact observed in LDE-1- or LDE-2-gavaged mice, gWAT and iWAT mass decreased by 46% and 50%, respectively (Fig. 3e,f; Extended Data Fig. 4b-e). Like administration of LDE-1 or LDE-2, leucic acid treatment did not cause muscle wasting (Extended Data Fig. 4f,g), accompanied by larger myofiber cross-sectional areas in the quadriceps (Extended Data Fig. 4h,i), longer distance to exhaustion on the treadmill, and greater grip strength (Fig. 3g,h). The enhanced grip strength was also observed in Aida -/- mice administered leucic acid, although their exercise endurance was not significantly changed (Extended Data Fig. 4j-l). Notably, leucic acid did not elevate levels of ALT or BUN in WT or Aida -/- mice, indicating no hepatorenal toxicity (Extended Data Fig. 5a-d). However, unlike LDE-2, leucic acid failed to improve fasting glycemia or reduce hepatic steatosis (Extended Data Fig. 5e,f), suggesting additional mediators in LDE-2 contribute to long-term metabolic benefits. To assess whether leucic acid has an impact on systemic energy metabolism to reduce adiposity and confer other benefits, WT mice were subjected to metabolic cage analysis. The leucic acid-treated mice exhibited a pronounced reduction in respiratory quotient (RQ) (Extended Data Fig. 5g), indicative of preferential lipid oxidation over catabolism of carbohydrate or amino acid. Leucic acid also induced an increase in basal rectal temperature (Fig. 3i), suggesting elevated heat production. These metabolic alterations occurred independently of changes in food intake, locomotor activity (Extended Data Fig. 5h,i), or faecal lipid excretion (Extended Data Fig. 5j,k), excluding malabsorption or anorexia as confounding factors. Bacterial HicDHs drive leucic acid production and are phylogenetically restricted in human gut clades To identify the enzymes responsible for leucic acid production, we performed Conserved Domain (CD) Searches 28 in strains LDE-1 and LDE-2. For reference, genes in these newly sequenced genomes were assigned arbitrary numeric identifiers during annotation. Both encode a branched-chain-amino-acid aminotransferase (BCAT) homolog for converting L-leucine to 2-ketoisocaproic acid (KIC) (Extended Data Fig. 6a,b). We identified six putative 2-hydroxyacid dehydrogenases (HicDHs) capable of reducing KIC to leucic acid (Supplementary Table 6,7). Phylogenetic analysis separated these into two distinct clusters. The proteins encoded by GM000059 and GM003503 in LDE-1, and GM002896 , GM003422, and GM003150 in LDE-2, formed a clade with various known D-HicDHs related proteins . The protein encoded by GM003311 in LDE-1 grouped with L-HicDH family proteins (Extended Data Fig. 6c). AlphaFold-predicted structures confirmed conservation, with low root mean square deviation (RMSD) values (0.928-1.907 Å) to reference enzymes (Extended Data Fig. 7a), and key nicotinamide adenine dinucleotide (NAD + ) co-substrate binding residues were strictly conserved (Extended Data Fig. 7b-g). Functional validation in E. coli revealed that only three HicDHs, GM003311 (LDE-1), GM003150 (LDE-2), and GM003422 (LDE-2), produced deuterated leucic acid from L-leucine-5,5,5-D3 (Extended Data Fig. 8a), despite being expressed at similar levels (Extended Data Fig. 8b). Enantiomer-specific triple quadrupole GC-MS analysis confirmed strict stereochemical partitioning. LDE-1 produced >98% L-leucic acid, while LDE-2 produced >98% D-leucic acid (Extended Data Fig. 8c,d), consistent with phylogenetic predictions. We assessed the human relevance of this pathway by probing the Unified Human Gastrointestinal Genome (UHGG) database 29 , and identified 27,237 putative homologs (≥50% similarity, ≥80% coverage) across 288 species. A high-confidence subset (344 homologs, ≥70% similarity) was enriched within 13 species from the Lachnospiraceae and Ruminococcaceae families, matching the lineages of LDE-2 and LDE-1, respectively (Extended Data Fig. 8e,f and Supplementary Table 8). This indicates leucic acid production is a specialised trait enriched in specific gut clades. Genomic analysis confirmed that this reductive pathway dominates in both strains, as we found no homologs for alternative routes such as leucine 2,3-aminomutase (LAM) or the branched-chain 2-keto acid dehydrogenase (BCKDH) complex (Extended Data Fig. 6a). This absence correlates with our observation of unchanged isobutyric acid and isovaleric acid levels (the end products of those pathways) in bacterial culture media (Supplementary Table 4,5) and the reduced levels of these acids in the ceca of Aida -/- mice (Fig. 3l,n). These findings establish HicDH-mediated reductive catabolism as the dominant leucine metabolic pathway in both strains, explaining their selective leucic acid production and metabolic effects. Leucic acid suppresses lipid storage through metabolic rewiring and enantioselective catabolic activation To elucidate the metabolic mechanism of leucic acid, we performed integrated proteomic and transcriptomic profiling of gWAT from leucic acid-treated WT mice. Multi-omics analysis identified 40 significantly downregulated candidates (|log 2 (fold change)| > 0.5, P < 0.05) enriched in lipid anabolic pathways, including triglyceride biosynthesis (Reactome: R-MMU-75109), fatty acid biosynthesis (GO:0006633) and lipid storage (GO:0019915) (Fig. 4a and Supplementary Table 9,10). Critically, leucic acid suppressed key enzymes catalysing triglyceride synthesis (DGAT2, MOGAT2) and fatty acid synthesis (ACSS2, FASN, SCD2 and ELOVL6) (Fig. 4b,c and Extended Data Fig. 9a). These findings were recapitulated in Aida -/- gWAT transcriptomes (Extended Data Fig. 9b-d and Supplementary Table 11) and validated by immunoblotting in both genotypes (Fig. 4d,e), confirming AIDA-independent suppression of lipid storage. In iWAT, leucic acid administration upregulated 14 genes in both WT and Aida -/- iWAT, enriched for pathways linked to muscle-like metabolic activation, such as muscle contraction (Reactome: R-MMU-397014, GO:0006936), myofibril (GO:0030016) and sarcomere (GO:0030017) (Extended Data Fig. 9e and Supplementary Table 12,13), indicative of adipose tissue beiging 30 . Strikingly, Gene Set Enrichment Analysis (GSEA) of Aida -/- iWAT demonstrated amplified thermogenic activation, with robust induction of thermogenesis (NES = 2.7437, adjusted P < .0001) and fatty acid degradation (NES = 2.3071, adjusted P < .0001) (Fig. 4f and Supplementary Table 14). These transcriptomic changes were biochemically validated by an increase in UCP1 protein abundance in both genotypes, correlating with the leucic acid-induced elevation of basal core body temperature (Fig. 3i), alongside a coordinated downregulation of lipid anabolism proteins DGAT2 and FASN via immunoblotting (Fig. 4g,h). We chose 50 μM, which was within the supplemented range in vivo (Extended Data Fig. 10a-g), to investigate direct cellular mechanisms. At this dose, leucic acid enantiomers significantly reduced lipid accumulation in oleate-challenged MEFs, with both enantiomers decreasing lipid stores by ∼60% in WT and ∼70% in Aida -/- cells (Fig. 5a,b). Notably, this potent lipid-lowering effect occurred despite the 1.9-fold higher lipid content in Aida -/- MEFs, which is consistent with AIDA being able to reduce lipid absorption 21 and suggests an AIDA-independent activity of leucic acid. We next investigated the catabolism of fatty acids using [U- 13 C]oleic acid tracing. In beige-like adipocytes, treatment with D-leucic acid significantly enhanced isotopic enrichment in α-ketoglutarate (36%) and citrate (47%), while L-leucic acid showed a weaker increase in α-ketoglutarate labelling (23%) and no citrate effect (Fig. 5c). This enantioselectivity was also found in Aida -/- MEFs. Both forms increased labelled α-ketoglutarate (D: 14%; L: 16%) and malate (D: 34%; L: 33%), but only D-leucic acid dramatically elevated proline (53%), a derivative of α-ketoglutarate (Extended Data Fig. 11a-d). Critically, unchanged total metabolite pools (Extended Data Fig. 11e-h) excluded biosynthesis alterations, instead implicating enhanced fatty acid-derived acetyl-CoA flux into the TCA cycle. Functional validation via oxygen consumption assays revealed that D-leucic acid markedly elevated respiration rate in WT and Aida -/- MEFs, while L-leucic acid induced subtler increases (Fig. 5d). Collectively, leucic acid exerts AIDA-independent suppression of lipid storage across adipose depots and genotypes, while exhibiting enantiomer-selective amplification of lipid catabolism in thermogenic adipocytes and Aida -/- cells, with a mechanistic divergence between gWAT and iWAT at physiologically relevant concentrations (Fig. 5e). HCAR2 mediates leucic acid-induced suppression of lipid accumulation To identify the molecular target through which leucic acid regulates lipid metabolism, we performed integrative target prediction for L- and D-enantiomers using complementary computational approaches, TargetNet employing deep learning on 3D features 31 and SwissTargetPrediction utilising similarity-based machine learning 32 (Supplementary Table 15). Intersection analysis revealed 13 consensus targets (Fig. 6a), mitigating the single-platform biases common in endogenous metabolite studies, with both enantiomers exhibiting full compliance with Lipinski's Rule of Five 33 (Extended Data Fig. 12a), indicating favourable drug-likeness. To prioritise these candidates based on human genetic evidence, we screened the GWAS Catalog 34 , which uncovered 44 obesity-associated intergenic or regulatory region SNPs near HCAR2 ( Hydroxycarboxylic Acid Receptor 2 ). In contrast, only one SNP was found near GRM2 ( Glutamate Metabotropic Receptor 2 ), and no significant associations were observed for other candidates (Fig. 6b and Supplementary Table 16). Transcriptomic analysis of adipose tissue from leucic acid-treated mice identified Hcar2 and Car4 ( Carbonic Anhydrase 4 ) as the sole genes that were significantly differentially expressed across both absolute abundance change and fold-change metrics (Fig. 6c). This convergence of genetic association and transcriptional responsiveness prioritised HCAR2 for mechanistic validation (Fig. 6d). Molecular docking suggested potential binding of HCAR2 for both D-leucic acid (ΔG ≈ -5.2 kcal/mol) and L-leucic acid (ΔG ≈ -5.3 kcal/mol) (Fig. 6e), with binding affinities comparable to established ligands niacin (ΔG ≈ -5.9 kcal/mol) and β-hydroxybutyrate (ΔG ≈ -4.8 kcal/mol) calculated under identical parameters (Extended Data Fig. 12b). We experimentally validated receptor engagement using 19 F-NMR ligand-detected binding assay 35,36 . Incubation of 4-fluoro-leucic acid with HCAR2-expressing cells induced a pronounced chemical shift (Δδ = 0.08 ppm), indicative of ligand-receptor engagement (Fig. 6f). By contrast, no shift occurred in HCAR1-expressing cells, confirming the specificity of HCAR2. The requirement of HCAR2 to mediate the role of leucic acid was further investigated through loss-of-function experiments. Leucic acid enantiomers reduced lipid accumulation in beige-like adipocytes and oleate-loaded MEFs expressing control shRNA, but this effect was abolished in Hcar2 -knockdown cells (Fig. 6g,h and Extended Data Fig. 12c-f). Similarly, Hcar2 knockdown eliminated leucic acid-stimulated oxygen consumption (Fig. 6i). For in vivo validation, we knocked down Hcar2 via AAV in one side of the iWAT, using the contralateral side injected with control shRNA as an internal control (Fig. 6j and Extended Data Fig. 12g). While Hcar2 knockdown alone did not alter baseline iWAT mass, leucic acid administration induced significant mass reduction and pronounced browning in control shRNA-treated iWAT. Crucially, the same treatment resulted in significantly attenuated fat mass reduction and diminished browning in contralateral Hcar2 -knockdown tissue relative to control tissues within the same animals (Fig. 6k,l). Collectively, these findings establish HCAR2 as the cognate receptor for leucic acid in mediating the suppression of lipid storage (Extended Data Fig. 12h). Discussion Our study reveals that the textural changes of chyme, caused by depletion of luminal fat as a result of AIDA deficiency, can critically influence microbial growth and metabolite production, which in turn plays a significant role in regulating host physiology. Specifically, the absence of AIDA in mice resulted in the selective enrichment of oleate-suppressed Clostridia strains LDE-1 and LDE-2. Paradoxically, these strains exert anti-obesity effects, as we found that gavage with LDE-1 or LDE-2 ameliorates obesity in mice fed a high-fat diet, forming a host-directed adaptive loop where microbial expansion counteracts host metabolic dysfunction. Mechanistically, we identified leucic acid as a key bioactive metabolite produced by HicDHs in these strains. Leucic acid counters obesity and enhances exercise performance. Moreover, we show that leucic acid suppresses lipid storage via divergent regulation of lipid anabolism and catabolism across adipose depots mediated by HCAR2. The selective enrichment of these beneficial strains is likely driven by a specific ecological niche, as varied abilities of the host to absorb different nutrients can alter the contents of residual nutrients in the chyme. Our analyses in vitro on the two strains LDE-1 and LDE-2 showed high sensitivity of the LDE strains to oleate, consistent with their enrichment in the fat-depleted gut of Aida -/- mice. Thus, AIDA deficiency, by promoting lipid absorption through ERAD-mediated degradation of re-esterification enzymes (GPAT3, MOGAT2 and DGAT2) 21 , creates a lipid-depleted luminal environment that favours the growth of oleate-sensitive bacteria. This model is further supported by the selective reduction of oleate and palmitate in the Aida -/- cecum, likely reflecting the preference of intestinal triglyceride synthesis enzymes for these acyl-CoAs 37,38 . Although dietary fatty acids generally exhibit antimicrobial properties, the particular sensitivity of LDE-1 and LDE-2 to oleate, a trait shared by bacteria like Lactobacillus iners 39 , coupled with their resistance to SCFAs or MCFAs that bypass intestinal re-esterification, defines a unique ecological niche. Leucic acid has been previously identified as a microbial metabolite that shows beneficial effects on metabolic dysfunction-associated steatohepatitis (MASH) 40 . We have found that leucic acid production by LDE-1 and LDE-2 represents a previously unrecognised mode of host-microbe crosstalk emerging from the changed luminal environment resulting from the genetic mutation of the host. The lipid-depleted conditions in Aida -/- gut create an ecological niche analogous to time-restricted feeding 40 . Our data show that leucic acid exerts striking metabolic benefits across both WT and Aida -/- genetic backgrounds, including reduction of adiposity and enhancement of exercise capacity. These effects stem from the combined roles of suppression of lipid synthesis in gWAT and promotion of lipid catabolism and thermogenesis in iWAT. The physiochemical properties of leucic acid salts facilitate HCAR2 engagement through efficient aqueous diffusion enabled by their hydrophilicity (TPSA 57-60 Å 2 , log P -0.8567 to 0.478). This mechanism gains human relevance from GWAS linking intergenic variants near the HCAR2 locus to obesity traits. Although physiological levels of leucic acid are modest in humans, these genetic associations likely represent conserved signalling through endogenous HCAR2 ligands such as β-hydroxybutyrate 41 and niacin 42 , positioning this pathway as an evolutionarily conserved regulator of energy homeostasis. Taken together, we have presented an exemplary mode of host-microbe relationship, in which the host physiological state can influence the community of gut microbes that exert feedback controls on the host. Methods Animals and ethics All experiments utilised male C57BL/6J mice. Homozygous Aida -/- mice and wild-type (WT) littermate controls were generated by crossing heterozygous Aida -/- parents as described previously 21 . WT mice administered LDE-1, LDE-2, or leucic acid were sourced from the Specific Pathogen-Free (SPF) facility at the Laboratory Animal Center of Xiamen University (Xiamen, China). For metabolic studies, 8-week-old male mice were fed a high-fat diet (HFD; D12492, Research Diets, New Brunswick, NJ, USA; 60% kcal from fat, primarily lard and soybean oil) ad libitum and housed under controlled conditions (temperature: 22-24 ℃; humidity: 55-60%; light/dark cycle: 12 h, lights on at 8:00 AM; enrichment: nesting materials replaced weekly). All animal procedures were approved by the Xiamen University institutional animal care and use committee (Approval No. XMULAC20220188). Cell culture Beige-like adipocytes were derived from 3T3-L1 preadipocytes (ATCC CL-173) via a beige protocol. Cells were maintained in an induction medium (DMEM/F12 1:1 medium with 10% FBS containing 5 μg/mL insulin (BS901, Biosharp, Hefei, China), 1 nM 3,3’,5-triiodo-l-thyronine (T3) (T-2877, Sigma-Aldrich, Taufkirchen, Germany), 125 μM indomethacin (S1723, Sigma-Aldrich), 0.5 mM isobutylmethylxanthine (IBMX) (I5879, Sigma-Aldrich) and 1 μM dexamethasone (DEX) (D4902, Sigma-Aldrich)) for 2 days. Cells were switched to maintenance medium (DMEM/F12 1:1 medium with 10% FBS supplemented with 5 μg/mL insulin and 1 nM T3) for 6 days, with medium refreshed every 48 h. Differentiated beige-like adipocytes were confirmed by multilocular lipid droplet formation. MEFs were maintained in DMEM with 10% FBS. Analysis of human faecal metagenomic data Publicly available metagenomic datasets (accession codes: CNP0003084, PRJEB12123) 43 were analysed, comprising 355 normal-weight (BMI 18.5-24.9) and 631 obese (BMI > 30) individuals. Taxon abundances were expressed as relative abundances, calculated by dividing the read count of each taxon by the total read count across all taxa. Zero values were substituted with the detection limit (1 × 10 -6 ), and log 10 transformation was applied. Covariates (sampling batch, sex, age) were assessed but showed no significant association with taxon abundances, thus, no adjustments were made. Gut microbiota analysis by 16S rRNA sequencing Cecal content samples were collected from 6 homozygous Aida -/- mice and 6 WT littermates after 4 months of HFD feeding. Genomic DNA was extracted using the MagBeads Fecal DNA Kit (116570400, MP Biomedicals, Santa Ana, CA, USA). The V3-V4 hypervariable regions of bacterial 16S rRNA genes were amplified and analysed using an Illumina HiSeq2500 platform and sequenced on an Illumina HiSeq2500 platform (Majorbio Bio-Pharm Technology Co. Ltd., Shanghai, China). Sequences with ≥ 97% similarity were assigned to the same operational taxonomic units (OTUs). Microbial alpha diversity indexes were obtained from the free online platform of the Majorbio Cloud Platform (www.majorbio.com). The 16S rRNA sequencing data have been deposited in the National Center for Biotechnology Information (NCBI) GenBank database under BioProject accession number PRJNA1311729. Quantitative PCR (qPCR) DNA from cecal content was extracted as above. Specific primers were designed for the amplification of fragments within 16S rRNA genes of OTU68 and OTU209, as well as universal primers targeting the 16S rRNA gene for normalization purposes 44 . The specificity of these primers was validated through in silico analysis and confirmed experimentally using known positive controls. OTU68: 187 bp product (F: 5′-GCGTCCGATTAGCCAGTTGG-3′; R: 5′-GCCTGCCTCTACTTCACTCAAG-3′). OTU209: 227 bp product (F: 5′-GCGTGGGTGAAGAAGCGCC-3′; R: 5′-TCTTGGAACTGCGTGGCTG-3′). Universal 16S: 177 bp product (F: 5′-ACTCCTACGGGAGGCAGCAGT-3′; R: 5′-ATTACCGCGGCTGCTGGC-3′). The qPCR was performed in triplicate using ChamQ SYBR qPCR Master Mix (Q312, Vazyme Biotech, Nanjing, China). Primer specificity was confirmed by Sanger sequencing of amplicons. Relative abundances were calculated using the 2 -ΔΔCt method normalised to universal 16S rRNA. Isolation of AIDA depletion-related bacterial strains Culture media preparation Nine agar plate formulations were used for anaerobic bacterial isolation: modified Gifu anaerobic medium (mGAM) and yeast casitone fatty acids broth (YCFA), with or without 1 g/L taurocholic acid sodium salt, respectively. mGAM was made by supplementing GAM with 10% bovine rumen fluid. Modified chopped meat medium (CMM), TSA medium (with 5% defibrinated sheep blood), rumen bacteria medium (TPY), modified reinforced clostridial medium (RCM) and modified PYG medium were also used. Media were autoclaved (121 ℃, 20 min), and 5% sheep blood was added post-sterilisation. Plates were pre-reduced in an anaerobic workstation (D500G, Gene Science, Wilmington, DE, USA; atmosphere: 85% N 2 , 5% CO 2 , 10% H 2 ) for at least 24 h before use. Strict anaerobic conditions of liquid media were generated by using an intelligent anaerobic culture preparation system (DLQ120-B, Beijing Ai Si Pu Technology Co., Ltd., Beijing, China). Cecal sample processing and anaerobic cultivation Cecal contents from 6-month HFD-fed Aida -/- mice were divided into two aliquots. Ethanol-treated fraction: Resuspended in 1 mL sterile PBS, incubated in 70% ethanol (4 h, 25 ℃), and filtered (0.22 µm) to remove host cells and debris. Sequential 10-fold dilutions (10 -3 to 10 -6 ) were plated (100 µL/plate). Untreated fraction: Directly diluted in PBS and plated. The plates were incubated at 37 ℃ in the anaerobic chamber (D500G, GeneScience). Plates were incubated at 37 ℃ under anaerobic conditions for up to 30 days. Colonies were monitored at 2, 4, 6, 8, 10, 15, 20, and 30 days, with six morphologically distinct colonies per plate isolated and subcultured for 2-7 days. Colony morphology and spatial distribution were documented to avoid redundancy. Bacterial identification via 16S rRNA gene sequencing Single colonies were lysed in 3 μL of NaOH/SDS buffer (0.8% NaOH, 1% SDS), diluted in 67 µL ddH 2 O, boiled (95 ℃, 30 min), and centrifuged (5000 g , 5 min). Supernatants were used as templates for PCR amplification of near-full-length 16S rRNA genes using universal primers 45 : 27F: 5'-AGAGTTTGATCCTGGCTCAG-3' and 1492R: 5'-GGTTACCTTGTTACGACTT-3'. PCR products (~1,500 bp) were Sanger-sequenced (Sangon Biotech Co., Ltd., Shanghai, China). Sequences were analysed using NCBI BLAST against a custom database of Aida -related OTUs, the Nucleotide Collection (nr/nt) and the 16S rRNA database (Bacteria and Archaea). Taxonomic assignments followed established thresholds with ≥ 98.7% identity to reference strains as the same species. LDE-1 and LDE-2 were deposited under the Budapest Treaty at the China General Microbiological Culture Center (CGMCC), Beijing, China. The assigned deposit numbers are CGMCC No.46618 for LDE-1 and CGMCC No.46619 for LDE-2. These strains are available for research purposes from CGMCC under the terms of the Material Transfer Agreement. Negative stain electron microscopy Bacterial isolates were resuspended in 20 μL of 0.1 M phosphate buffer. A portion of this suspension was carefully applied onto a carbon/formvar-coated grid and allowed to adhere for 15 s, blotted off and stained with 1% uranyl acetate for 15 s. Grids were air-dried for 3 h at 25 ℃ and imaged on a Hitachi HT-7800 transmission electron microscope (HT-7800, Hitachi, Tokyo, Japan). Images were captured at magnifications ranging from 4,000 × to 10,000 × at 80 kV. Bacterial genome sequencing and assembly DNA extraction and library preparation Genomic DNA was extracted from 1 g of pure LDE-1 or LDE-2 cultures and subjected to bacteria draftmap analysis by Novogene Co., Ltd. (Beijing, China). DNA was randomly sheared into short fragments, end-repaired, A-tailed, and ligated with an Illumina adapter. The DNA fragments with adapters were size-selected, PCR amplified and purified. The library was checked with Qubit and real-time PCR for quantification, and a bioanalyzer for size distribution detection. Sequencing and quality control Paired-end sequencing (2 × 150 bp) was performed on an Illumina NovaSeq 6000 platform (Novogene Co., Ltd., Beijing, China). The reads with > 40% low-quality bases (mass value ≤ 20), with > 10% unrecognised nucleotides (N), or adapter contamination (15 bp overlap, < 3 mismatches). Genome Assembly and Annotation The clean data were assembled with SOAPdenovo, SPAdes, and ABySS software. Contigs from all assemblies were integrated with CISA software and polished with GapCloser software to fill gaps. Final scaffolds (> 500 bp) were annotated using GeneMarkS. Functional annotation involved performing a whole-genome Blast search against GO and KEGG databases. Genome assemblies were deposited in NCBI GenBank under BioProject accession PRJNA1311729. Strain classification and phylogenetic analysis Taxonomic assignment of the strains was performed using a dual approach: 16s rRNA sequence analysis and the Genome Taxonomy Database Toolkit (GTDB-Tk). Full-length 16S rRNA sequences, extracted from whole-genome assemblies, were first queried against the GenBank/ENA/DDBJ and EzBioCloud databases (https://ezbiocloud.net/) to identify closest relatives. Sequence alignment was executed using ClustalX (v. 1.838), and phylogenetic trees were constructed using the neighbour-joining algorithm in MEGA11 46 with bootstrap support based on 1,000 replicates. For higher-resolution classification, draft genome assemblies were analysed with GTDB-Tk (v. 1.7.0) via the KBase platform (https://narrative.kbase.us/). This workflow leverages 120 conserved bacterial marker genes from the Genome Taxonomy Database (GTDB, v. R06-RS202), which integrates curated RefSeq and Genbank reference genomes. Strains were taxonomically assigned by placement into a domain-specific reference phylogeny, refined using relative evolutionary divergence (RED) thresholds and pairwise average nucleotide identity (ANI) values relative to GTDB reference genomes 47,48 . All GTDB-Tk analyses were finalised on 2024-05-08. Bacterial growth inhibition assays by fatty acids Fatty acid solution preparation Sodium salt solutions were prepared according to chain length. Acetate (S818277, Macklin Biochemical Technology, Shanghai, China), propionate (S817368, Macklin Biochemical Technology), and butyrate (S817488, Macklin Biochemical Technology) were dissolved in ddH 2 O to prepare 100 mM stocks, 0.22 µm-filtered, and assayed at 5 mM. Caprylate (S6096, Macklin Biochemical Technology), decanoate (629-25-4, MedChemExpress, Shanghai, China), and myristate (822-12-8, MedChemExpress) were dissolved in 2% DMSO in ddH 2 O to prepare 10 mM stocks, sonicated or vortexed, 0.22 µm nylon-filtered (84311ES03, Yeasen Biotechnology, Shanghai, China), and tested at 100 µM. Palmitate (S161420, Aladdin Scientific, Shanghai, China), stearate (822-16-2, Aladdin Scientific), and oleate (S817542, Macklin Biochemical Technology) were added into mGAM media, autoclaved to dissolve, tested at 10 µM or 100 µM. All solutions were validated precipitate-free. NaCl controls (matched to experimental concentrations) were prepared equivalently. Bacterial growth analysis LDE-1, LDE-2 and Bacteroides uniformis were inoculated in 96-well plates (1 mL per well) or jars (40 mL per jar) containing fatty acid working solutions or controls. LDE-1 was inoculated at a density of OD 600 = 0.6 and LDE-2 at a density of OD 600 = 0.15 to mGAM media. Cultures were normalised to an identical starting OD 600 . Media pH was verified to be pH 7.0 ± 0.2 pre-inoculation. Plates or jars were maintained in an anaerobic chamber (D500G, GeneScience) throughout incubation. Growth was monitored at 24 h intervals by measuring OD 600 . Background absorbance was corrected by subtracting cell-free supernatant OD 600 from total readings. Post-assay culture purity was confirmed via 16S rRNA sequencing. Bacterial preparation and oral administration Two-month-old male mice were randomised into experimental groups based on body weight and housed under SPF conditions. Mice were fed HFD for 3-5 days, followed by broad-spectrum antibiotic (ABX) treatment (1 g/L ampicillin, 0.5 g/L vancomycin, 1 g/L neomycin, 1 g/L metronidazole in drinking water) for 5-7 days to deplete gut microbiota. The LDE-1 culture was inoculated into fresh mGAM and cultured anaerobically (85% N 2 , 5% CO 2 , 10% H 2 ) at 37 ℃ for 48 h. After incubation, the culture was centrifuged (10,000 g , 8 min). The supernatant was discarded, and the bacterial pellet was washed with sterile PBS. The bacteria were then resuspended in fresh mGAM to achieve OD 600 = 1.3. The control group received the same volume of fresh mGAM administered by oral gavage, once every three days for ten administrations. The preparation of LDE-2 was similar to that of LDE-1, with the following modifications. The culture was inoculated at a 1:25 dilution, incubated for 60 h, centrifuged at 6,000 × g for 5 min, and resuspended to achieve OD 600 = 0.6. The control group received the same volume of sterile PBS administered by oral gavage, once daily for 14 days. The Escherichia coli Nissle 1917 (EcN) culture was inoculated at a 1:50 dilution into fresh LB liquid medium and incubated in a shaker at 37 ℃ for 18-22 h. The culture was then centrifuged at 6,000 × g for 5 min. The supernatant was discarded, and the bacterial pellet was washed with sterile PBS. The bacteria were resuspended in sterile PBS to achieve an OD 600 = 1.3. The control group received the same volume of sterile PBS administered by oral gavage, once daily for 10 days. After the last gavage, the mice that received LDE-1 were kept for an additional 10 days, while those that received LDE-2 were observed for 16 weeks before sacrifice to minimise any acute effects of gavage. The mice were fasted for 6 h (8:30 a.m. to 2:30 p.m.) before sacrifice. Administration of leucic acid in mice Two-month-old male mice were divided into two comparable groups. HFD feeding started at least 2 days before the leucic acid administration. DL-leucic acid (S161215, Aladdin Scientific, Shanghai, China) was solubilised in the drinking water to make a final concentration of 50 mM leucic acid (pH = 2.5). The drinking water was acidified using HCl (ADW) to the same pH as the control mice. The drinking water was replaced with distilled water containing 50 mM leucic acid or ADW every 5 days. The mice were fasted for 6 h (8:30 a.m. to 2:30 p.m.) before sacrifice. Biochemical analysis Blood and faecal parameters were measured in mice after a 6-h fast (8:30 a.m. to 2:30 p.m.). Blood glucose was quantified by a glucose meter (OneTouch UltraVue, Johnson & Johnson, New Brunswick, NJ, USA). Commercial kits were utilised to determine the levels of triglyceride (TAG, A110-1-1, Nanjing Jiancheng Bioengineering Institute, Nanjing, China), non-esterified fatty acids (NEFA, A042-1-1, Nanjing Jiancheng), alanine aminotransferase (ALT, C009-2-1, Nanjing Jiancheng) and blood urea nitrogen (BUN, C013-2-1, Nanjing Jiancheng) in accordance with the manufacturer’s protocols. Measurement of basal core body temperature Following one week of leucic acid or ADW administration to mice acclimated to HFD for 5 days, basal core body temperature was measured. Using a calibrated rectal probe, temperatures were recorded between 14:30 and 15:30 after a 4-h fast. Two independent measurements were taken per mouse at 48 h intervals, ensuring consistent probe insertion depth of 1.5 cm and a dwell time of 30 seconds. Mass-spectrometry-based metabolomics for bacterial secreted metabolites Sample preparation Untargeted metabolomics was performed on culture media from LDE-1 and LDE-2 strains (Metware Biotechnology Inc., Wuhan, China). Strains were grown in modified GAM medium to the exponential phase. Pre- and post-culture media were collected. A 150 μL extraction solution (acetonitrile: methanol = 1:4, v:v) containing 1 μg/mL of 2-chlorophenylalanine (internal standard) was added into a 50 μL sample. Then the sample was vortexed (3 min), centrifuged (12,000 g , 10 min, 4 ℃). Supernatant was collected and incubated at -20 ℃ (30 min), recentrifuged (12,000 g , 3 min, 4 ℃). 120 μL supernatant was aliquoted for LC-MS. LC-MS analysis LC-MS analysis was performed using UHPLC (Shimadzu Nexera LC-30A, Kyoto, Japan) and MS (SCIEX TripleTOF 6600+, Foster City, CA, USA) in positive/negative ionisation modes. Waters ACQUITY Premier HSS T3 Column (1.8 µm, 2.1 mm × 100 mm) using 0.1% formic acid in water as solvent A and 0.1% formic acid in acetonitrile as solvent B in the following gradient: 5 to 20% in 2 min, increased to 60% in the following 3 min, increased to 99% in 1 min and held for 1.5 min, then come back to 5% mobile phase B within 0.1 min, held for 2.4 min. The analytical conditions were as follows: column temperature, 40 ℃; flow rate, 0.4 mL/min; injection volume, 4 μL. Mass spectrometry data acquisition was operated using information-dependent acquisition (IDA) mode using Analyst TF software v. 1.7.1 (SCIEX, Concord, Canada). The source parameters (ion source Gas 1/Gas 2: 50 psi; curtain gas: 25 psi; temperature: 550 ℃; ion spray voltage floating: 5000 V or -4000 V; decluttering potential: ±60 V). TOF-MS scan parameters (mass range: 50-1000 Da; accumulation time: 200 ms; and dynamic background subtract: on). Product ion scan parameters (mass range: 25-1000 Da; accumulation time: 40 ms; collision energy: ±30; collision energy spread: 15; resolution: UNIT; charge state: 1 to 1; intensity: 100 cps; exclude isotopes within 4 Da; mass tolerance: 50 ppm; maximum number of candidate ions to monitor per cycle: 18). Data Processing Raw data were converted to mzXML format by ProteoWizard software. Peak extraction, alignment, and retention time correction were performed in XCMS program. Peak areas normalised using support vector regression (SVR). Low-abundance peaks (< 50% detection rate per group) were excluded. Metabolic identification information was obtained by searching in-house, public, and AI-predicted databases and MetDNA web server. The metabolomics data have been deposited to MetaboLights 49 repository with the study identifier MTBLS12947. Determination of leucic acid concentration by GC-MS The concentrations of leucic acid in mouse faeces, serum, and tissues were measured using GC-MS. To optimise this determination, different derivatisation approaches and different capillary columns were compared. Sample preparation involved homogenising approximately 50 mg of tissue or cecal content in a 20-fold volume of 80% (v:v) methanol aqueous solution containing 10 μg/mL 2-hydroxyoctanoic acid as an internal standard. The homogenate was centrifuged at 18,000 × g and 4 °C for 15 min. A 200 μL aliquot of the supernatant was dried using a vacuum centrifugal concentrator. The residue was reconstituted in 50 μL pyridine (270407, Merck, Darmstadt, Germany) via sonication. Derivatisation was then performed using one of three methods. The first method involved derivatisation with 50 μL N-methyl-N-(trimethylsilyl)trifluoroacetamide (MSTFA, M85387, Meryer, Shanghai, China) at 37 °C for 1 h to form trimethylsilyl (TMS) derivatives. The second method involved sequential derivatisation, first with 50 μL MSTFA at 37 °C for 1 h, followed by 50 μL N-Methyl-bis(trifluoroacetamide) (MBTFA, N814244, Macklin, Shanghai, China) at 37 °C for 1 h to form TMS and trifluoroacetyl (TFA) derivatives. The third method used 50 μL N- tert -butyldimethylsilyl-N-methyltrifluoroacetamide (MTBSTFA, M-108, Merck, Darmstadt, Germany) at 55 °C for 1 h to form tert -butyldimethylsilyl (TBDMS) derivatives. All derivatisation reactions occurred in a heating block. Following derivatisation, samples were centrifuged at 18,000 × g for 5 min, and the supernatant was transferred to GC vials for analysis. For the comparison of low-polarity (HP-5MS) and mid-polarity (DB-1701) capillary columns, TMS-derivatised samples prepared as described were analysed. Both GC-MS methods employed a 1 mL/min helium carrier gas flow rate, a 1 µL injection volume, a 10:1 split ratio, and a 250 °C injector temperature. However, the temperature programs were adjusted to account for the different column characteristics and ensure appropriate elution of the leucic acid di-TMS derivative. The HP-5MS method used an initial hold at 70 °C for 2 min, ramped at 7 °C per min to 140 °C, then ramped at 30 °C per min to 310 °C with a final hold for 9 min. The DB-1701 method used an initial hold at 80 °C for 2 min, ramped at 5 °C per min to 140 °C, then ramped at 10 °C per min to 200 °C, followed by a ramp at 30 °C per min to 290 °C with a final hold for 9 min. Detection was performed in combined scan/selected ion monitoring (SIM) mode. The SIM ions were set at m/z 103, 159, 233, and 261, with a solvent delay of 7 min. Indirect calorimetry Mice were acclimated to individual housing in metabolic chambers (CAB-16-1-EU, Sable Systems International, North Las Vegas, NV, USA) for 3 days to minimise stress from environmental changes. Following acclimation, oxygen consumption (VO 2 ), carbon dioxide production (VCO 2 ), and respiratory exchange ratio (RER) were measured over 48 h at 5-min intervals using high-resolution sensors. Food intake was estimated daily per cage and normalised to individual mouse intake by dividing total cage consumption by the number of mice. Body composition analysis Fat and lean mass were quantified in conscious mice using a 3-in-1 EchoMRI Composition Analyzer (100H, Echo Medical Systems, Houston, TX, USA). Triplicate measurements per mouse were performed and averaged to ensure precision. Muscle performance assessments Grip strength test Forelimb grip strength was measured using a digital force meter (47200, Ugo Basile, Gemonio, Italy). The maximum grip strength was measured for each mouse, with claws of forelimbs held against the device. Each mouse was tested three times to ensure reliable data collection. Treadmill exhaustion test A motorised treadmill (47300, Ugo Basile, Gemonio, Italy) was used to assess endurance. Mice were fasted for 1 h prior to an acclimatisation phase on the treadmill two days before the formal test (5° incline, 5 m/min speed). The formal test mirrored the acclimatisation protocol but with a higher speed limit (initial speed = 5 m/min, increased by 1 m/min until reaching 25 m/min and held constant). Exhaustion was defined as the mouse remaining on the electrified rear of the treadmill (> 5 s) despite mild stimulation. Total distance and time to exhaustion were recorded. Structure prediction and alignment for bacterial HicDHs Putative 2-hydroxyisocaproate dehydrogenase (HicDH) structures from LDE-1 and LDE-2 were modelled using AlphaFold2 50 . Predicted structures were aligned to experimentally resolved templates (PDB IDs: 1DXY [ Lactobacillus casei D-2-HicDH] 51 , 1HYH [ Weissella Confusa L-2-HicDH] 52,53 in PyMOL (v. 3.1.6.1, Schrödinger, New York, NY, USA). Root-mean-square deviation (RMSD) values were calculated by PyMOL using the entire protein structure. Phylogenetic analysis of putative HicDHs A neighbour-joining 55 tree was constructed in MEGA11 46 . Each of the protein sequences of the six putative HicDHs from LDE-1 and LDE-2 was used for BLAST of 3 proteins with top total scores, which belong to 3 different genera within NCBI refseq_protein . Three well-characterised HicDHs, including L-2-HicDH of Weissella confusa , D-2-HicDH of Lacticaseibacillus casei , and LDHD of Mus musculus, were additionally included. The percentage of replicate trees in which the associated taxa clustered together in the bootstrap test (500 replicates) is shown next to the branches 55 . The evolutionary distances were computed using Poisson correction 56,57 distances with complete deletion of gaps/missing data (279 positions retained). Cloning of putative 2-hydroxyacid dehydrogenases Putative 2-hydroxyacid dehydrogenases were identified through in silico screening of predicted proteins encoded by the draft genomes of strains LDE-1 and LDE-2. A batch conserved domain (CD)-Search against the NCBI CDD database (https://www.ncbi.nlm.nih.gov/Structure/bwrpsb/bwrpsb.cgi) 28 revealed six candidates annotated as 2-hydroxyisocaproate dehydrogenase (HicDH), HicDH-like proteins, or 2-hydroxyacid dehydrogenases. Coding sequences for these candidates were codon-optimised for heterologous expression in Escherichia coli using ExpOptimizer (NovoPro, https://www.novoprolabs.com/tools/codon-optimization), chemically synthesised (Tsingke Biotech, Beijing, China), and cloned into the expression vector pGEX-4T-1 for downstream functional characterisation. Determination of the enantiomer ic excess of leucic acid Enantiomeric excess analysis was performed with modifications to a previously described method 58 . Strains were cultured in mGAM medium and 4-Methyl-2-oxovaleric acid (1 mM, M104530, Aladdin Scientific, Shanghai, China), adjusted to pH 7.0 with NaOH, was added 12 h prior to sampling. Leucic acid was analysed as methyl esters by a trimethylsilyldiazomethane methylation approach. The methylation procedure was as follows. Leucic acid was extracted three times with diethyl ether after acidification with 0.2 M HCl. Pooled ether layers were dried under nitrogen gas. Residues were reconstituted in 90 μL methanol:ether (2:7, v:v), cooled to 4 ℃, and reacted with 10 μL trimethylsilyl-diazomethane (T140538, Aladdin Scientific) for 3 h at 25 ℃. The resulting leucic acid methyl esters, along with a commercial racemic mixture standard (D9509, TargetMol Chemicals, Shanghai, China), were quantified using a CP-Chirasil-Dex CB column (25 m × 0.25 mm × 0.25 μm) on an GC-MS system (Agilent 8890B GC system and 7000D MSD system, Agilent Technologies, Santa Clara, CA, USA) using helium at 1 mL/min (constant flow). Temperature program was as follows: 80 ℃ (2 min), ramp to 114 ℃ at 2 ℃/min, 194 ℃ at 10 ℃ (hold 3 min). Injection was performed with split mode (10:1), 1 μL volume. DL-leucic acid-methyl ester (T9509, TargetMol Chemicals, Shanghai, China) served as the external standard. Data were analysed using MassHunter Qualitative Analysis software v. 10.0 (Agilent Technologies). Stable isotope tracing of leucic acid production Bacterial culture and isotope labelling E. coli BL21 harbouring pGEX-4T-1-HicDHs were grown in LB medium (with 100 μg/mL ampicillin) at 37 ℃ to OD 600 = 0.6. Protein expression was induced with 0.5 mM isopropyl β-D-1-thiogalactopyranoside (IPTG), followed by incubation for 3 h at 37 ℃. After centrifugation, the cells were resuspended to OD 600 = 0.4 in fresh LB containing 1 mM L-leucine-5,5,5-D3 (HY-N0486S9, MedChemExpress) and cultured for 3 h. Metabolite extraction and derivatisation A 500 μL aliquot of the culture supernatant was dried using centrifugal freeze-drying and extracted with 200 μL methanol, sonicated (30 s, 20 kHz, 40 W) and centrifuged (18,000 g , 10 min). The supernatant (50 μL) was transferred to a glass vial, dried under nitrogen gas, and derivatised by sequential treatment with 50 μL pyridine containing 20 mg/mL methoxyamine hydrochloride (89803, Sigma-Aldrich; 37 ℃, 1.5 h; for methoximation of carbonyl groups) and 50 μL N-methyl-N-(trimethylsilyl) trifluoroacetamide (MSTFA; M-132, Merck, Darmstadt, Germany; 37 ℃, 1 h; for trimethylsilylation). GC-MS analysis Derivatised samples were analysed using a quadrupole GC-MS system (Agilent 7890B GC system and 5977B MSD System) equipped with an HP-5ms capillary column (30 m × 0.25 mm × 0.25 μm; Agilent Technologies). Helium carrier gas was maintained at a constant flow rate of 1 mL/min. The temperature program was as follows: 70 ℃ (2 min), ramp to 140 ℃ at 7 ℃/min, then ramp to 310 ℃ at 30 ℃/min (hold 9 min). Samples (1 μL) were injected in split mode (10:1 ratio). Electron impact ionisation (70 eV) was performed with mass spectra acquired in full-scan mode (m/z 50-650). Data were analysed using MassHunter (Agilent Technologies). Isotopomer distributions were quantified by integrating ion chromatograms (leucic acid: m/z 159; leucic acid-5,5,5-D3: m/z 162) corresponding to the trimethylsilylated leucic acid fragment, normalised to an external calibration curve generated using synthetic leucic acid-5,5,5-D3. The standard was synthesised via diazotisation of d3-leucine as previously described 59 . Tissue proteomic analysis by mass spectrometry Gonadal WAT (gWAT) of 8-week-old male mice were used for proteomic analysis. Mice were adapted to HFD-feeding for one week, following administration of 50 mM leucic acid in drinking water or acidified water with equal pH as a control. Mice were euthanised 2 weeks post-treatment, and tissues were immediately removed for preparation of protein extracts. Tissues were homogenised in 1% cholic acid (V900388, Sigma-Aldrich) at a 1:3 (w/v) tissue-to-buffer ratio using a homogeniser. Homogenates were sonicated (15 cycles, 30 Hz) and clarified by centrifugation at 14,000 × g for 15 min at 4 °C. The lipid layer was carefully discarded, and the infranatant was transferred to a new tube and centrifuged again under identical conditions. The final protein supernatant was collected, and its concentration was determined using BCA assay. For each sample, 10 μg of protein was subjected to in-solution trypsin digestion. Dried peptide samples were reconstituted in 10 μL of 0.1% formic acid. Peptide concentration was estimated by measuring OD 214 . Subsequently, 400 ng of peptides from each sample were separated online using a nanoElute nanoflow LC system coupled to a timsTOF HT mass spectrometer (Bruker corporation, Billerica, MA, USA) equipped with a CaptiveSpray ion source. Peptides were loaded onto a homemade C18 column (75 μm inner diameter × 35 cm, 1.9 μm particle size, 100 Å pore size) and eluted with a 60-min linear gradient from 3% to 35% acetonitrile in 0.1% formic acid at a constant flow rate of 300 nL/min. Mass spectra were acquired in data-independent acquisition parallel accumulation serial fragmentation (DIA-PASEF) mode. Raw data files were processed using Spectronaut 17.0 (Biognosys) in a directDIA+ mode against the UniProtKB Mus musculus proteome database (release 2023_01). Search parameters were set as follows: enzyme, trypsin/P; specificity, full; maximum missed cleavages, 2; fixed modification, carbamidomethylation (C); variable modifications, oxidation (M) and protein N-terminal acetylation. Cross-run normalisation was enabled based on the total peptide signal. All other parameters were set to the software default. Missing values were imputed with the lower detection limit, and log 2 -transformed intensities were used for fold-change calculation and volcano plot generation. The mass spectrometry proteomics data have been deposited to the ProteomeXchange Consortium (https://proteomecentral.proteomexchange.org) via the iProX partner repository 60,61 with the dataset identifier PXD068013. Transcriptomic profiling of adipose tissues gWAT and iWAT of 8-week-old male WT and Aida -/- mice were subjected to transcriptomic analysis. RNA-seq was performed by Novogene (Beijing, China) on the Novaseq platform (Illumina, San Diego, CA, USA) with paired-end 150 bp reads. Raw reads were trimmed of adapter sequences and filtered to remove low-quality bases (Phred score < 20) and contaminants using FastQC (v0.11.9). High-quality reads were aligned to the GRCm39 mouse reference genome (Mus musculus) using HISAT2 (v2.2.1), and transcript counts were generated with featureCounts (v2.0.6) using the following flags: featureCounts -p --countReadPairs -t exon -g gene_id. Differential gene expression analysis was performed in R (v4.3.2) using the limma-voom pipeline. Genes with |log 2 (fold change)| > 0.5 and P < 0.05 were classified as significantly differentially expressed. The RNA-sequencing data have been deposited in the NCBI Gene Expression Omnibus (GEO) under accession number GSE306938. Functional enrichment of differentially expressed genes Differentially expressed genes (DEGs) were defined as those with |log 2 (fold change)| > 0.5 and P < 0.05. DEGs underwent functional enrichment analysis using ShinyGO (v. 0.82) 62 with the following parameters: enrichment fold FDR < 0.05, pathway size 2-500 genes, organism database Mus musculus . Selected enriched GO terms and Reactome 63 pathways were visualised. Gene set enrichment analysis (GSEA) 64 was performed on the full expression dataset using the gseKEGG function in clusterProfiler R package (v4.8.3). To ensure statistical robustness, we conducted the analysis with the default number of permutations while maintaining default weight settings. After the analysis, we used gseaplot2 function to visualise the selected enrichment patterns. Immunoblot analysis of adipose tissues Gonadal and inguinal WAT from HFD-fed male mice treated with 50 mM leucic acid in drinking water for 6 weeks were minced and homogenized in ice-cold RIPA lysis buffer (20 mM Tris-HCl pH 7.5, 150 mM NaCl, 1 mM Na 2 EDTA, 1 mM EGTA, 1% NP-40, 1% sodium deoxycholate, 2.5 mM sodium pyrophosphate, 1 mM β-glycerophosphate) supplemented with 0.1% SDS and Phosphatase Inhibitor Cocktail I/II (HY-K0021/ HY-K0022, MedChemExpress), and Protease Inhibitor Cocktail (B14002, Selleck, Shanghai, China) using a tissue homogenizer (OSE-TH-02, TIANGEN, Beijing, China) at 7 m/s linear velocity for 5 cycles (40 s run/20 s pause per cycle). Homogenates were sonicated and centrifuged. Protein concentrations in the supernatant were determined by BCA assay (20201ES86, Yeasen, Shanghai, China), and proteins were denatured in SDS sample buffer at 70 °C for 15 min. 15 μg protein per lane were resolved by SDS-PAGE and analysed via immunoblotting. Primary antibodies were ACSS2 (1:3,000, 16087-1-AP, RRID: AB_2878216, Proteintech, Wuhan, China), DGAT2 (1:1,000, 17100-1-AP, RRID: AB_2918049, Proteintech), FABP4 (1:10,000, 15872-1-AP, RRID: AB_2102440, Proteintech), FASN (1:5,000, A22462PM, RRID: AB_2862543, ABclonal, Wuhan, China), MOGAT2 (1:500, 19514-1-AP, RRID: AB_10638917, Proteintech) and UCP1 (1:1,000, 72298, RRID: AB_2936479, Cell Signaling Technology, Danvers, MA, USA). ACSS2, FASN and FABP4 antibodies have been validated by knockdown/knockout, DGAT2 and MOGAT2 antibodies by overexpression positive controls, and UCP1 antibody by tissue-specific expression. Secondary antibodies were goat anti-rabbit IgG HRP (1:5,000, 31460, RRID: AB_228341, Thermo Scientific, Waltham, MA, USA) and goat anti-mouse IgG HRP (1:5,000, 31430, RRID: AB_228307, Thermo Scientific). Integrated intensities of protein bands were quantified by ImageJ. Metabolic flux analysis using [U- 13 C]oleic acid Metabolic flux analysis was performed in beige-like adipocytes and MEFs using [U- 13 C]oleic acid tracing. For beige-like adipocytes, 3T3-L1 cells were treated with 50 μM sodium salts of D-leucic acid, L-leucic acid, or NaCl vehicle control throughout differentiation and beigeing induction. MEFs were treated with the same compounds for 24 h. One hour prior to harvest, the medium was replaced with fresh medium containing 250 μM [U - 13 C]oleic acid (#490431, Sigma-Aldrich, Taufkirchen, Germany) alongside the respective leucic acid enantiomers or control. After incubation, cells were washed twice with 30 mL ice-cold saline, and culture dishes were flash-frozen in liquid nitrogen for 10 min. Cells were subsequently scraped into pre-chilled Eppendorf tubes using 1 mL ice-cold 80% methanol containing 10 μg/mL 2-hydroxyoctanoic acid (H862015, Macklin Biochemical Technology) as internal standard. Lysates were sonicated and centrifuged at 12,000 × g for 10 min at 4 °C. An 800 μL aliquot of the supernatant was lyophilised to complete dryness. The dried residue was derivatised by resuspension in 50 μL pyridine containing 20 mg/mL methoxyamine hydrochloride (89803, Sigma-Aldrich), followed by sonication and incubation at 37 °C for 1.5 h. Next, 50 μL N-Methyl-N-( tert -butyldimethylsilyl)trifluoroacetamide (MTBSTFA; M-108, Cerilliant, Round Rock, Texas, USA) was added, and the mixture was incubated at 55 °C for 1 h. After centrifugation, 60 μL of the derivatised supernatant was transferred for GC-MS analysis. GC-MS was conducted using an Agilent 7890B/5977B system. Samples (2 μL) were injected in split mode (1:10 ratio) with an inlet temperature of 250 °C. The GC oven temperature program initiated at 70 °C (2-min hold), ramped to 140 °C at 3 °C/min, then to 310 °C at 10 °C/min. Mass spectrometry operated in SIM mode, targeting [M-57] + fragments (loss of tert -butyldimethylsilyl group) and isotopologue-specific M+n ions (n = number of 13 C atoms) of tert -butyldimethylsilyl (TBDMS) derivatives. Isotopic natural abundance was corrected using IsoCor software (v. 2.2.3) 65 . Integrative target identification using orthogonal computational platforms Computational target prediction for L- and D-leucic acid enantiomers was performed using two orthogonal algorithmic platforms, Ensemble TargetNet 31 and SwissTargetPrediction 32 , to ensure robust identification of potential targets. Ensemble TargetNet is a deep learning model that generates binding probabilities based on three-dimensional molecular features, including ECFP6, MACCS, and Daylight fingerprints, where targets exceeding the probability threshold of 0.7 were retained as high-confidence candidates. Complementary to this, SwissTargetPrediction employs a hybrid approach combining two-dimensional and three-dimensional ligand similarity matching with machine learning algorithms, but due to its systematic underestimation of probability values, only the top 100 ranked targets were considered for further analysis. Lipinski's Rule of Five parameters, which evaluate drug-likeness through molecular weight, hydrogen bond donors, hydrogen bond acceptors, and log P 33 were computed directly from the TargetNet framework. The stereospecific structures of both enantiomers were defined using canonical SMILES strings with explicit stereochemistry annotations. L-leucate represented as CC(C)C[C@@H](C(=O)O)O and D-leucate as CC(C)C[C@H](C(=O)[O-])O, which were submitted as input for both prediction platforms to account for chiral specificity in target engagement. Molecular docking Molecular docking simulations were performed using AutoDock Vina 66 (v. 1.2.7) to elucidate the binding interactions between leucic acid enantiomers and the human HCAR2. The HCAR2 (PDB ID: 8H2G) cryo-EM structure was prepared by removing crystallographic water molecules, adding polar hydrogens, and assigning Kollman united-atom charges via AutoDockTools (v. 1.5.7) 67 . Leucic acid enantiomers as well as niacin and β-hydroxybutyrate were energy-minimised using the MM2 force field in Chem 3D (v. 22.0.0, Revvity, Waltham, MA, USA), with Gasteiger charges assigned using AutoDockTools (v. 1.5.7) 68 . A docking grid box (20 × 20 × 20 Å 3 ) was centred on the orthosteric binding site (coordinates: x = 115.895, y = 114.032, z = 160.702) to encompass key residues. Vina parameters included an exhaustiveness of 64, energy range of 4 kcal/mol, and 10 binding poses generated per run. The top-scoring pose (lowest predicted ΔG) for each target was selected. All visualisations and interaction analyses were conducted in PyMOL (v. 3.1.6.1, Schrödinger). Ligand-receptor binding assay via 19 F-NMR in live cells Ligand-receptor interactions were analysed using real-time 19 F NMR in live cells, adapted from established protocols 35,36 . Cells were harvested by EDTA-mediated detachment, washed twice with ice-cold PBS, and resuspended in DMEM supplemented with 2 mM EDTA and 10% OptiPrep (D1556, Sigma-Aldrich). 4-fluoroleucic acid was synthesised via diazotisation of 4-fluoroleucine (BD271440, Bidepharm, Shanghai, China), purified by following a previously reported protocol 60 . The yield of 4-F-leucic acid from 4-F-leucine was found to be equivalent to the yield of leucic acid from leucine, as determined by GC-MS. The compound was purified via methyl tert -butyl ether extraction and dried under nitrogen, dissolved and neutralised with KOH to a 30 mg/mL stock (pH 7.0). For binding assays, 400 μL cell suspension was mixed with 50 μL fluorinated ligand, 5 μL potassium trifluoroacetate (100 mg/mL internal standard), and 50 μL D 2 O, incubated (10 min, 37 ºC), and analysed on a high-resolution NMR spectrometer (AVANCE-NEO-600, Bruker, Billerica, USA) equipped with a broadband 19 F probe. Acquisition parameters: 16 scans, 12 μs pulse width, no decoupling, ambient temperature. 19 F NMR spectra were processed in MestReNova software v. 15.0.0-34764 (Mestrelab Research, Santiago de Compostela, Spain) using automated phase correction, Whittaker smoothing, and internal reference calibration. Cellular respiration monitoring (Resipher) Respiration of intact cells under normal culture conditions (37 ℃, 5% CO 2 ) was performed using a non-invasive oxygen consumption rate monitoring system (Resipher 32X, Lucid Scientific Inc., Atlanta, GA, USA) 68 . Briefly, 5 × 10 3 cells were seeded in 96-well microplates. Cells were treated with 50 μM D- or L-leucic acid (sodium salt), or NaCl (control) for 36 h. The Resipher oxygen sensing lid was positioned upon the plate before monitoring. Oxygen concentration was measured in each well every minute at 37 ºC, and OCR was calculated based on the oxygen concentration during 15 min. Data were exported using the Resipher web application and OCR normalised to cell count. High-resolution respirometry (Oxygraph-O2k) High-resolution respirometry with intact cells was performed using the Oroboros Oxygraph-O2k (Oroboros Instruments, Innsbruck, Austria) 69 . 3T3-L1 preadipocytes were seeded into 6-well plate. Cells were grown to confluence and differentiated into mature adipocytes following the protocol as described above, with 50 μM D- or L-leucic acid (sodium salt), or NaCl as a control in the induction and maintenance media. Following full differentiation, these adipocytes were washed twice with PBS, trypsinised and centrifuged at 300 × g for 5 min, resuspended in DMEM (supplemented with non-essential amino acids, GlutaMax and sodium pyruvate). The Oroboros O2k chambers were pre-oxygenated, and the chamber temperature was set to 37 ℃. The oxygen concentration was calibrated before the experiment. An aliquot containing 2 × 10 6 cells was added to each chamber and allowed to stabilise for 10 min before measurement began. Baseline oxygen consumption rates were recorded. Sequentially, 2.5 μM oligomycin (s1478, Selleck), 0.5 μM FCCP (s8276, Selleck), 0.5 μM rotenone (s2348, Selleck) and 2.5 μM antimycin A (ab141904, Abcam, Cambridge, UK) were sequentially added into the chambers to characterise maximum mitochondrial oxygen consumption. Data were recorded and analysed using DatLab software (v. 7.4.0.4, Oroboros Instruments). Oxygen consumption rates (OCR) were normalised to the protein concentrations of the respective samples and expressed as pmol O 2 /s/mg protein. Unilateral AAV-mediated Hcar2 knockdown in iWAT AAV2/9 vectors were generated for in vivo knockdown of Hcar2 using a triple-plasmid system. The target sequences were Renilla as a control are 5’-GCTGTGTGTTCCGAGATGA-3’ for murine Hcar2 and 5’-GTAGCGCGGTGTATTATAC-3’ for Renilla (control). The transgene plasmid, bearing AAV2 inverted terminal repeats, contained mU6 promoter for shRNA expression. This was co-transfected into HEK293T cells alongside the pAAV2/9 rep-cap plasmid with sequences coding for REP (from type 2 AAV) and CAP (from type 9 AAV), and the pAdΔF6 adenoviral helper plasmid. Viral particles were harvested 60-72 h post-transfection. Cells underwent three cycles of freeze-thaw lysis (liquid nitrogen/37 °C), followed by precipitation of medium-derived virions with polyethylene glycol (PEG; final concentration of 8 % PEG-8000, 0.5 M NaCl; 4 °C overnight). Lysates and PEG pellets were combined, purified via discontinuous OptiPrep (D1556, Sigma-Aldrich) density gradient ultracentrifugation (17%, 25%, 40%, 60 %), and the 40% fraction was collected. Viruses were washed three times with PBS using 100 kDa ultrafiltration centrifugation (UFC910008, Millipore, Billerica, MA, USA). Genomic titers were quantified by qPCR. Twelve 6-week-old male mice were housed under standard conditions. Bilateral iWAT depots received differential injections. The right depot was administered 2 × 10 11 genomic copies of AAV2/9- mU6 - shRenilla (control) in 50 µL PBS across four sites, and the left depot received AAV2/9- mU6 - shHCAR2 (knockdown) identically. Following a 12-day recovery period, mice were acclimated to HFD for 4 days. Animals were then randomised into two weight-matched groups. One group received 50 mM leucic acid in drinking water, and the controls received pH-matched acidified water. Tissues were harvested 13 weeks post-injection for analysis. Statistics and reproducibility Statistical analyses were performed using GraphPad Prism (v. 9.0, GraphPad Software, La Jolla, CA, USA). Normality was assessed using D'Agostino-Pearson omnibus or Shapiro-Wilk tests to guide test selection. For two-group comparisons, two-tailed unpaired Student’s t -tests or Mann-Whitney tests were applied (depending on normality). Multi-group comparisons involving a single variable employed one-way ANOVA (parametric) or Kruskal-Wallis tests (non-parametric), with Tukey’s, Dunn’s or Dunnett's post-hoc corrections as detailed in figure legends. For experiments with two independent variables, two-way ANOVA with Dunnett's or Holm-Šídák’s post-hoc testing was utilised. Statistical significance between growth curves was determined by comparing fits using an extra sum-of-squares F-test for curve divergence 70 . To assess the statistical significance of differences among groups in Principal Coordinates Analysis (PCoA) plots, Analysis of Variance Using Distance Matrices (ADONIS), also known as Permutational Multivariate Analysis of Variance (PERMANOVA), was employed by using ClustVis tool 71 . Pearson's correlation coefficient ( r ) or Spearman's rank correlation coefficient ( ρ ) (depending on normality), along with their associated P -value, was utilised to evaluate the relationship between two variables. A significance threshold of P < 0.05 was applied throughout. All experiments included ≥ 3 biological replicates, with full statistical details provided in figures and source data files. Declarations Data availability The 16S rRNA sequencing data have been deposited in the National Center for Biotechnology Information (NCBI) GenBank database under BioProject accession number PRJNA1311729. Genome assemblies were deposited in NCBI GenBank under BioProject accession PRJNA1311729. The metabolomics data have been deposited to MetaboLights repository with the study identifier MTBLS12947. The RNA-sequencing data have been deposited in the NCBI Gene Expression Omnibus (GEO) under accession number GSE306938. The mass spectrometry proteomics data have been deposited to the ProteomeXchange Consortium (https://proteomecentral.proteomexchange.org) via the iProX partner repository with the dataset identifier PXD068013. Statistical source data and image source data are provided with this paper. Bacteria strains LDE-1 and LDE-2, and all other data that support the findings of this study, are available from the corresponding author upon reasonable request. The uncropped immunoblot and microscopy images (PDF) and statistical source data (Excel) have been deposited in a private Figshare repository and can be accessed via this link for review: https://figshare.com/s/3dea33aa5b47932a9a7b Acknowledgements We gratefully acknowledge Drs. Haojie Huang, Xiaoxing Li and Xiang Chen for their assistance with bacterial strain typing. We thank Dr. Shu-Hai Lin for guidance on metabolomics techniques, Dr. Baoding Zhang for molecular docking, and Dr. Guang Li for phylogenetic analysis. We are grateful to Drs. Changtao Jiang, Tong-Jin Zhao, Lemin Zheng, Shih-Chin Cheng and Weiqi Xu for their critical comments and suggestions. Technical support for animal studies was provided by the Laboratory Animal Center of Xiamen University. Finally, we would like to thank all members of the SCL laboratory for their contributions to this work. This study utilises data from the GWAS Catalog 34 . This work was supported by grants from the National Key Research and Development Program of China (2022YFA0806500, 2024YFC3407000), the National Natural Science Foundation of China (32571357, 31822027, 32271216, 82088102, 32422002, 32370052), the Fundamental Research Funds for the Central Universities (20720210110, 20720200069), Project “111" sponsored by the State Bureau of Foreign Experts and Ministry of Education of China (BP2018017), the Joint Funds for the Innovation of Science and Technology, Fujian province (2021Y9232), the XMU-Fujian Cancer Hospital cooperation grant for the Research Center of Metabolism and Tumor, the Joint Funds for the Innovation of Science and Technology, Fujian province (2024Y9603), and XMU Training Program of Innovation, and Entrepreneurship for Undergraduates (2020X892, 202410384095, 202510384048). Author information Authors and Affiliations State Key Laboratory of Cellular Stress Biology, School of Life Sciences, Faculty of Medicine and Life Sciences, Xiamen University, Xiamen, China Zhi-Yu Chen, Hao-Yu Qin, Yang Alan Yang, Dong-Tai Liu, Shu-Han Yang, Ximin Chi, Hao Guo, Feng Guo, Qi-Qian Han, Pu-Chun Zhu, Wen-Jing Wu, Ming Jiang, Ming-Xia Zhu, Cixiong Zhang, Lei Zhang, Yaying Wu, Ao Li, Luming Yao, Wei Hong, Tian-Yu Tang, Sheng-Cai Lin, Shu-Yong Lin Department of Gastroenterology, Clinical Research Center for Gut Microbiota and Digestive Diseases of Fujian Province, The National Key Clinical Specialty, Zhongshan Hospital of Xiamen University, School of Medicine, Xiamen University, Xiamen, China Kan Wang, Yang Song, Jianlin Ren, Hongzhi Xu, Shu-Yong Lin Department of Endocrine and Metabolic Diseases, Shanghai Institute of Endocrine and Metabolic Diseases, Ruijin Hospital, Shanghai Jiao Tong University School of Medicine, Shanghai, China Ruixin Liu, Chao Wu Shanghai National Clinical Research Center for Metabolic Diseases, Key Laboratory for Endocrine and Metabolic Diseases of the National Health Commission of the PR China, Shanghai Key Laboratory for Endocrine Tumor, State Key Laboratory of Medical Genomics, Ruijin Hospital, Shanghai Jiao Tong University School of Medicine, Shanghai, China Ruixin Liu, Chao Wu State Key Laboratory of Genetic Engineering, Shanghai Key Laboratory of Metabolic Remodeling and Health, Laboratory of Longevity and Metabolic Adaptations, Institute of Metabolism and Integrative Biology, Fudan University, Shanghai, China Terytty Yang Li, Lin Chen State Key Laboratory of Microbial Technology, Shandong University, Qingdao, China. Qinghua Zhao, Changyu Wang, Chang Liu, Shuang-Jiang Liu Department of Chemical Engineering, Tsinghua University, Beijing, China Ming Jiang Department of Endocrinology and Metabolism, Shanghai Clinical Centre for Diabetes, Shanghai Diabetes Institute, Shanghai Key Laboratory of Diabetes Mellitus, Shanghai Jiao Tong University Affiliated Sixth People’s Hospital, Shanghai, China Mingliang Zhang Contributions Conceptualisation: SCL and SYL. Methodology: SJL, CL, QZ, CW 6 , FG, ZYC, KW, HYQ, XC, HG, YS, MXZ, CZ, LY, AL, LZ, YW, MZ. Resources: SJL, CL, RL, JR, YS, HX, MZ. Investigation: ZYC, KW, DTL, HYQ, YAY, SHY, LC, MJ, PCZ. Formal analysis: ZYC, KW, DTL, HYQ, YAY, SHY, QQH, CW 3, 4 , TYL, SYL. Validation: SYL, ZYC, KW, HYQ, YAY. Visualisation: SYL, ZYC, XC, WJW. Writing - Original Draft: SYL, ZYC, HYQ, YAY. Writing - Review & Editing: SCL, SYL, SJL, HG, CL. Supervision: SYL, SCL, SJL, HX. Project administration: SYL, SCL, TYT, WH. Funding acquisition: SCL, SYL, JR, CL, HG, SHY. Corresponding author Correspondence to Shu-Yong Lin. Ethics declarations Competing interests The authors declare the following financial interests: The bacterial strains LDE-1 and LDE-2 are the subject of a patent deposition. The findings regarding the use of leucic acid against obesity are also intended for patent protection. The authors may potentially file related patent applications in the future, which could lead to the commercialisation of technology. References Groussin, M., Mazel, F., and Alm, E.J. (2020). Co-evolution and Co-speciation of Host-Gut Bacteria Systems. Cell Host Microbe 28 , 12-22. 10.1016/j.chom.2020.06.013. Mallott, E.K., and Amato, K.R. (2021). Host specificity of the gut microbiome. Nat Rev Microbiol 19 , 639-653. 10.1038/s41579-021-00562-3. Wilde, J., Slack, E., and Foster, K.R. (2024). Host control of the microbiome: Mechanisms, evolution, and disease. Science 385 , eadi3338. 10.1126/science.adi3338. Cani, P.D., and Van Hul, M. (2024). Gut microbiota in overweight and obesity: crosstalk with adipose tissue. Nat Rev Gastroenterol Hepatol 21 , 164-183. 10.1038/s41575-023-00867-z. Van Hul, M., and Cani, P.D. (2023). The gut microbiota in obesity and weight management: microbes as friends or foe? Nat Rev Endocrinol 19 , 258-271. 10.1038/s41574-022-00794-0. Cani, P.D., Van Hul, M., Lefort, C., Depommier, C., Rastelli, M., and Everard, A. (2019). Microbial regulation of organismal energy homeostasis. Nat Metab 1 , 34-46. 10.1038/s42255-018-0017-4. Backhed, F., Ding, H., Wang, T., Hooper, L.V., Koh, G.Y., Nagy, A., Semenkovich, C.F., and Gordon, J.I. (2004). The gut microbiota as an environmental factor that regulates fat storage. Proc Natl Acad Sci U S A 101 , 15718-15723. 10.1073/pnas.0407076101. Gentile, C.L., and Weir, T.L. (2018). The gut microbiota at the intersection of diet and human health. Science 362 , 776-780. 10.1126/science.aau5812. Suarez-Zamorano, N., Fabbiano, S., Chevalier, C., Stojanovic, O., Colin, D.J., Stevanovic, A., Veyrat-Durebex, C., Tarallo, V., Rigo, D., Germain, S., et al. (2015). Microbiota depletion promotes browning of white adipose tissue and reduces obesity. Nat Med 21 , 1497-1501. 10.1038/nm.3994. Chevalier, C., Stojanovic, O., Colin, D.J., Suarez-Zamorano, N., Tarallo, V., Veyrat-Durebex, C., Rigo, D., Fabbiano, S., Stevanovic, A., Hagemann, S., et al. (2015). Gut Microbiota Orchestrates Energy Homeostasis during Cold. Cell 163 , 1360-1374. 10.1016/j.cell.2015.11.004. Goodrich, J.K., Waters, J.L., Poole, A.C., Sutter, J.L., Koren, O., Blekhman, R., Beaumont, M., Van Treuren, W., Knight, R., Bell, J.T., et al. (2014). Human genetics shape the gut microbiome. Cell 159 , 789-799. 10.1016/j.cell.2014.09.053. Sanna, S., Kurilshikov, A., van der Graaf, A., Fu, J., and Zhernakova, A. (2022). Challenges and future directions for studying effects of host genetics on the gut microbiome. Nat Genet 54 , 100-106. 10.1038/s41588-021-00983-z. Nagarajan, A., Scoggin, K., Gupta, J., Threadgill, D.W., and Andrews-Polymenis, H.L. (2023). Using the collaborative cross to identify the role of host genetics in defining the murine gut microbiome. Microbiome 11 , 149. 10.1186/s40168-023-01552-8. Goodrich, J.K., Davenport, E.R., Beaumont, M., Jackson, M.A., Knight, R., Ober, C., Spector, T.D., Bell, J.T., Clark, A.G., and Ley, R.E. (2016). Genetic Determinants of the Gut Microbiome in UK Twins. Cell Host Microbe 19 , 731-743. 10.1016/j.chom.2016.04.017. Dzierozynski, L., Queen, J., and Sears, C.L. (2023). Subtle, persistent shaping of the gut microbiome by host genes: A critical determinant of host biology. Cell Host Microbe 31 , 1569-1573. 10.1016/j.chom.2023.09.007. Yang, H., Wu, J., Huang, X., Zhou, Y., Zhang, Y., Liu, M., Liu, Q., Ke, S., He, M., Fu, H., et al. (2022). ABO genotype alters the gut microbiota by regulating GalNAc levels in pigs. Nature 606 , 358-367. 10.1038/s41586-022-04769-z. Zhernakova, D.V., Wang, D., Liu, L., Andreu-Sanchez, S., Zhang, Y., Ruiz-Moreno, A.J., Peng, H., Plomp, N., Del Castillo-Izquierdo, A., Gacesa, R., et al. (2024). Host genetic regulation of human gut microbial structural variation. Nature 625 , 813-821. 10.1038/s41586-023-06893-w. Ruhlemann, M.C., Hermes, B.M., Bang, C., Doms, S., Moitinho-Silva, L., Thingholm, L.B., Frost, F., Degenhardt, F., Wittig, M., Kassens, J., et al. (2021). Genome-wide association study in 8,956 German individuals identifies influence of ABO histo-blood groups on gut microbiome. Nat Genet 53 , 147-155. 10.1038/s41588-020-00747-1. Qin, Y., Havulinna, A.S., Liu, Y., Jousilahti, P., Ritchie, S.C., Tokolyi, A., Sanders, J.G., Valsta, L., Brozynska, M., Zhu, Q., et al. (2022). Combined effects of host genetics and diet on human gut microbiota and incident disease in a single population cohort. Nat Genet 54 , 134-142. 10.1038/s41588-021-00991-z. Wu, Q., Liang, X., Wang, K., Lin, J., Wang, X., Wang, P., Zhang, Y., Nie, Q., Liu, H., Zhang, Z., et al. (2021). Intestinal hypoxia-inducible factor 2alpha regulates lactate levels to shape the gut microbiome and alter thermogenesis. Cell Metab 33 , 1988-2003 e1987. 10.1016/j.cmet.2021.07.007. Luo, H., Jiang, M., Lian, G., Liu, Q., Shi, M., Li, T.Y., Song, L., Ye, J., He, Y., Yao, L., et al. (2018). AIDA Selectively Mediates Downregulation of Fat Synthesis Enzymes by ERAD to Retard Intestinal Fat Absorption and Prevent Obesity. Cell Metab 27 , 843-853 e846. 10.1016/j.cmet.2018.02.021. Shi, M., Huang, X.Y., Ren, X.Y., Wei, X.Y., Ma, Y., Lin, Z.Z., Liu, D.T., Song, L., Zhao, T.J., Li, G., et al. (2021). AIDA directly connects sympathetic innervation to adaptive thermogenesis by UCP1. Nat Cell Biol 23 , 268-277. 10.1038/s41556-021-00642-9. Newgard, C.B., An, J., Bain, J.R., Muehlbauer, M.J., Stevens, R.D., Lien, L.F., Haqq, A.M., Shah, S.H., Arlotto, M., Slentz, C.A., et al. (2009). A branched-chain amino acid-related metabolic signature that differentiates obese and lean humans and contributes to insulin resistance. Cell Metab 9 , 311-326. 10.1016/j.cmet.2009.02.002. Canfora, E.E., Jocken, J.W., and Blaak, E.E. (2015). Short-chain fatty acids in control of body weight and insulin sensitivity. Nat Rev Endocrinol 11 , 577-591. 10.1038/nrendo.2015.128. Layman, D.K., and Walker, D.A. (2006). Potential importance of leucine in treatment of obesity and the metabolic syndrome. J Nutr 136 , 319S-323S. 10.1093/jn/136.1.319S. den Besten, G., van Eunen, K., Groen, A.K., Venema, K., Reijngoud, D.J., and Bakker, B.M. (2013). The role of short-chain fatty acids in the interplay between diet, gut microbiota, and host energy metabolism. J Lipid Res 54 , 2325-2340. 10.1194/jlr.R036012. Crozier, S.J., Kimball, S.R., Emmert, S.W., Anthony, J.C., and Jefferson, L.S. (2005). Oral leucine administration stimulates protein synthesis in rat skeletal muscle. J Nutr 135 , 376-382. 10.1093/jn/135.3.376. Wang, J., Chitsaz, F., Derbyshire, M.K., Gonzales, N.R., Gwadz, M., Lu, S., Marchler, G.H., Song, J.S., Thanki, N., Yamashita, R.A., et al. (2023). The conserved domain database in 2023. Nucleic Acids Res 51 , D384-D388. 10.1093/nar/gkac1096. Almeida, A., Nayfach, S., Boland, M., Strozzi, F., Beracochea, M., Shi, Z.J., Pollard, K.S., Sakharova, E., Parks, D.H., Hugenholtz, P., et al. (2021). A unified catalog of 204,938 reference genomes from the human gut microbiome. Nat Biotechnol 39 , 105-114. 10.1038/s41587-020-0603-3. Bostrom, P., Wu, J., Jedrychowski, M.P., Korde, A., Ye, L., Lo, J.C., Rasbach, K.A., Bostrom, E.A., Choi, J.H., Long, J.Z., et al. (2012). A PGC1-alpha-dependent myokine that drives brown-fat-like development of white fat and thermogenesis. Nature 481 , 463-468. 10.1038/nature10777. Yao, Z.J., Dong, J., Che, Y.J., Zhu, M.F., Wen, M., Wang, N.N., Wang, S., Lu, A.P., and Cao, D.S. (2016). TargetNet: a web service for predicting potential drug-target interaction profiling via multi-target SAR models. J Comput Aided Mol Des 30 , 413-424. 10.1007/s10822-016-9915-2. Daina, A., Michielin, O., and Zoete, V. (2019). SwissTargetPrediction: updated data and new features for efficient prediction of protein targets of small molecules. Nucleic Acids Res 47 , W357-W364. 10.1093/nar/gkz382. Lipinski, C.A., Lombardo, F., Dominy, B.W., and Feeney, P.J. (2001). Experimental and computational approaches to estimate solubility and permeability in drug discovery and development settings. Adv Drug Deliv Rev 46 , 3-26. 10.1016/s0169-409x(00)00129-0. Sollis, E., Mosaku, A., Abid, A., Buniello, A., Cerezo, M., Gil, L., Groza, T., Gunes, O., Hall, P., Hayhurst, J., et al. (2023). The NHGRI-EBI GWAS Catalog: knowledgebase and deposition resource. Nucleic Acids Res 51 , D977-D985. 10.1093/nar/gkac1010. Luchinat, E., Barbieri, L., Davis, B., Brough, P.A., Pennestri, M., and Banci, L. (2024). Ligand-Based Competition Binding by Real-Time (19)F NMR in Human Cells. J Med Chem 67 , 1115-1126. 10.1021/acs.jmedchem.3c01600. Dalvit, C., and Vulpetti, A. (2019). Ligand-Based Fluorine NMR Screening: Principles and Applications in Drug Discovery Projects. J Med Chem 62 , 2218-2244. 10.1021/acs.jmedchem.8b01210. Sukumaran, S., Barnes, R.I., Garg, A., and Agarwal, A.K. (2009). Functional characterization of the human 1-acylglycerol-3-phosphate-O-acyltransferase isoform 10/glycerol-3-phosphate acyltransferase isoform 3. J Mol Endocrinol 42 , 469-478. 10.1677/JME-09-0010. Lardizabal, K.D., Mai, J.T., Wagner, N.W., Wyrick, A., Voelker, T., and Hawkins, D.J. (2001). DGAT2 is a new diacylglycerol acyltransferase gene family: purification, cloning, and expression in insect cells of two polypeptides from Mortierella ramanniana with diacylglycerol acyltransferase activity. J Biol Chem 276 , 38862-38869. 10.1074/jbc.M106168200. Zhu, M., Frank, M.W., Radka, C.D., Jeanfavre, S., Xu, J., Tse, M.W., Pacheco, J.A., Kim, J.S., Pierce, K., Deik, A., et al. (2024). Vaginal Lactobacillus fatty acid response mechanisms reveal a metabolite-targeted strategy for bacterial vaginosis treatment. Cell. 10.1016/j.cell.2024.07.029. Zhang, Y., Wang, X., Lin, J., Liu, J., Wang, K., Nie, Q., Ye, C., Sun, L., Ma, Y., Qu, R., et al. (2024). A microbial metabolite inhibits the HIF-2alpha-ceramide pathway to mediate the beneficial effects of time-restricted feeding on MASH. Cell Metab 36 , 1823-1838 e1826. 10.1016/j.cmet.2024.07.004. Taggart, A.K., Kero, J., Gan, X., Cai, T.Q., Cheng, K., Ippolito, M., Ren, N., Kaplan, R., Wu, K., Wu, T.J., et al. (2005). (D)-beta-Hydroxybutyrate inhibits adipocyte lipolysis via the nicotinic acid receptor PUMA-G. J Biol Chem 280 , 26649-26652. 10.1074/jbc.C500213200. Lorenzen, A., Stannek, C., Lang, H., Andrianov, V., Kalvinsh, I., and Schwabe, U. (2001). Characterization of a G protein-coupled receptor for nicotinic acid. Mol Pharmacol 59 , 349-357. 10.1124/mol.59.2.349. Wu, C., Yang, F., Zhong, H., Hong, J., Lin, H., Zong, M., Ren, H., Zhao, S., Chen, Y., Shi, Z., et al. (2024). Obesity-enriched gut microbe degrades myo-inositol and promotes lipid absorption. Cell Host Microbe 32 , 1301-1314 e1309. 10.1016/j.chom.2024.06.012. Barman, M., Unold, D., Shifley, K., Amir, E., Hung, K., Bos, N., and Salzman, N. (2008). Enteric salmonellosis disrupts the microbial ecology of the murine gastrointestinal tract. Infect Immun 76 , 907-915. 10.1128/IAI.01432-07. Weisburg, W.G., Barns, S.M., Pelletier, D.A., and Lane, D.J. (1991). 16S ribosomal DNA amplification for phylogenetic study. J Bacteriol 173 , 697-703. 10.1128/jb.173.2.697-703.1991. Tamura, K., Stecher, G., and Kumar, S. (2021). MEGA11: Molecular Evolutionary Genetics Analysis Version 11. Mol Biol Evol 38 , 3022-3027. 10.1093/molbev/msab120. Parks, D.H., Chuvochina, M., Chaumeil, P.A., Rinke, C., Mussig, A.J., and Hugenholtz, P. (2020). A complete domain-to-species taxonomy for Bacteria and Archaea. Nat Biotechnol 38 , 1079-1086. 10.1038/s41587-020-0501-8. Parks, D.H., Chuvochina, M., Waite, D.W., Rinke, C., Skarshewski, A., Chaumeil, P.A., and Hugenholtz, P. (2018). A standardized bacterial taxonomy based on genome phylogeny substantially revises the tree of life. Nat Biotechnol 36 , 996-1004. 10.1038/nbt.4229. Yurekten, O., Payne, T., Tejera, N., Amaladoss, F.X., Martin, C., Williams, M., and O'Donovan, C. (2024). MetaboLights: open data repository for metabolomics. Nucleic Acids Res 52 , D640-D646. 10.1093/nar/gkad1045. Jumper, J., Evans, R., Pritzel, A., Green, T., Figurnov, M., Ronneberger, O., Tunyasuvunakool, K., Bates, R., Zidek, A., Potapenko, A., et al. (2021). Highly accurate protein structure prediction with AlphaFold. Nature 596 , 583-589. 10.1038/s41586-021-03819-2. Dengler, U., Niefind, K., Kiess, M., and Schomburg, D. (1997). Crystal structure of a ternary complex of D-2-hydroxyisocaproate dehydrogenase from Lactobacillus casei, NAD+ and 2-oxoisocaproate at 1.9 A resolution. J Mol Biol 267 , 640-660. 10.1006/jmbi.1996.0864. Niefind, K., Hecht, H.J., and Schomburg, D. (1995). Crystal structure of L-2-hydroxyisocaproate dehydrogenase from Lactobacillus confusus at 2.2 A resolution. An example of strong asymmetry between subunits. J Mol Biol 251 , 256-281. 10.1006/jmbi.1995.0433. Feil, I.K., Hendle, J., and Schomburg, D. (1997). Modified substrate specificity of L-hydroxyisocaproate dehydrogenase derived from structure-based protein engineering. Protein Eng 10 , 255-262. 10.1093/protein/10.3.255. Saitou, N., and Nei, M. (1987). The neighbor-joining method: a new method for reconstructing phylogenetic trees. Mol Biol Evol 4 , 406-425. 10.1093/oxfordjournals.molbev.a040454. Felsenstein, J. (1985). Confidence Limits on Phylogenies: An Approach Using the Bootstrap. Evolution 39 , 783-791. 10.1111/j.1558-5646.1985.tb00420.x. Zuckerkandl, E., and Pauling, L. (1965). Molecules as documents of evolutionary history. J Theor Biol 8 , 357-366. 10.1016/0022-5193(65)90083-4. Zuckerkandl, E., and Pauling, L.C. (1965). Evolutionary Divergence and Convergence in Proteins. Valentina Jurkaš, Christoph K. Winkler, Silvan Poschenrieder, Paulo Oliveira, Catarina C. Pacheco, Eunice A. Ferreira, Florian Weissensteiner, Piera De Santis, Selin Kara, Robert Kourist, et al. (2022). Expression and activity of heterologous hydroxyisocaproate dehydrogenases in Synechocystis sp. PCC 6803 ΔhoxYH. Engineering Microbiology 2 . https://doi.org/10.1016/j.engmic.2021.100008. Hu, D.X., O'Brien, M., and Ley, S.V. (2012). Continuous multiple liquid-liquid separation: diazotization of amino acids in flow. Org Lett 14 , 4246-4249. 10.1021/ol301930h. Chen, T., Ma, J., Liu, Y., Chen, Z., Xiao, N., Lu, Y., Fu, Y., Yang, C., Li, M., Wu, S., et al. (2022). iProX in 2021: connecting proteomics data sharing with big data. Nucleic Acids Res 50 , D1522-D1527. 10.1093/nar/gkab1081. Ma, J., Chen, T., Wu, S., Yang, C., Bai, M., Shu, K., Li, K., Zhang, G., Jin, Z., He, F., et al. (2019). iProX: an integrated proteome resource. Nucleic Acids Res 47 , D1211-D1217. 10.1093/nar/gky869. Ge, S.X., Jung, D., and Yao, R. (2020). ShinyGO: a graphical gene-set enrichment tool for animals and plants. Bioinformatics 36 , 2628-2629. 10.1093/bioinformatics/btz931. Milacic, M., Beavers, D., Conley, P., Gong, C., Gillespie, M., Griss, J., Haw, R., Jassal, B., Matthews, L., May, B., et al. (2024). The Reactome Pathway Knowledgebase 2024. Nucleic Acids Res 52 , D672-D678. 10.1093/nar/gkad1025. Subramanian, A., Tamayo, P., Mootha, V.K., Mukherjee, S., Ebert, B.L., Gillette, M.A., Paulovich, A., Pomeroy, S.L., Golub, T.R., Lander, E.S., and Mesirov, J.P. (2005). Gene set enrichment analysis: a knowledge-based approach for interpreting genome-wide expression profiles. Proc Natl Acad Sci U S A 102 , 15545-15550. 10.1073/pnas.0506580102. Millard, P., Delepine, B., Guionnet, M., Heuillet, M., Bellvert, F., and Letisse, F. (2019). IsoCor: isotope correction for high-resolution MS labeling experiments. Bioinformatics 35 , 4484-4487. 10.1093/bioinformatics/btz209. Eberhardt, J., Santos-Martins, D., Tillack, A.F., and Forli, S. (2021). AutoDock Vina 1.2.0: New Docking Methods, Expanded Force Field, and Python Bindings. J Chem Inf Model 61 , 3891-3898. 10.1021/acs.jcim.1c00203. Goodsell, D.S., and Olson, A.J. (1990). Automated docking of substrates to proteins by simulated annealing. Proteins 8 , 195-202. 10.1002/prot.340080302. Triolo, M., and Khacho, M. (2024). Protocol to monitor live-cell, real-time, mitochondrial respiration in mouse muscle cells using the Resipher platform. STAR Protoc 5 , 103330. 10.1016/j.xpro.2024.103330. Doerrier, C., Garcia-Souza, L.F., Krumschnabel, G., Wohlfarter, Y., Meszaros, A.T., and Gnaiger, E. (2018). High-Resolution FluoRespirometry and OXPHOS Protocols for Human Cells, Permeabilized Fibers from Small Biopsies of Muscle, and Isolated Mitochondria. Methods Mol Biol 1782 , 31-70. 10.1007/978-1-4939-7831-1_3. Motulsky, H., Christopoulos, A., and ProQuest. (2004). Fitting models to biological data using linear and nonlinear regression : a practical guide to curve fitting, 1st Edition (Oxford University Press). Metsalu, T., and Vilo, J. (2015). ClustVis: a web tool for visualizing clustering of multivariate data using Principal Component Analysis and heatmap. Nucleic Acids Res 43 , W566-570. 10.1093/nar/gkv468. Additional Declarations Yes there is potential Competing Interest. The authors declare the following financial interests: The bacterial strains LDE-1 and LDE-2 are the subject of a patent deposition. The findings regarding the use of leucic acid against obesity are also intended for patent protection. The authors may potentially file related patent applications in the future, which could lead to the commercialisation of technology. Supplementary Files SupplementaryTable118.xlsx Supplementary Table 1-18 Additionalinformation.docx Cite Share Download PDF Status: Under Review Version 1 posted You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-8045609","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Article","associatedPublications":[],"authors":[{"id":546343240,"identity":"c33c3923-c65d-4e81-b835-55896447a300","order_by":0,"name":"Shu-Yong Lin","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAAs0lEQVRIiWNgGAWjYBACPmYQWQHhSBClhQ2s5QxJWkAEYxtJWth5D3/mnXc42uAA88HbPAx2eUQ4jC9NmndbWu6GA2zJ1jwMycVEaOExY+bdZgPUwmMmzcNwILGBCC3Gn3nnSAC18H8jWouBNG8D2BY2orWYSc45lpY78zCbseUcg2TCWvj5zxh/eFNzOLfvePPDG28q7AhrAQEmHhAJjlMDYtQDAeMPIhWOglEwCkbBCAUA7KoxXJDqKyEAAAAASUVORK5CYII=","orcid":"https://orcid.org/0000-0002-4419-1713","institution":"Xiamen University","correspondingAuthor":true,"prefix":"","firstName":"Shu-Yong","middleName":"","lastName":"Lin","suffix":""},{"id":546343241,"identity":"35935ac2-6ebe-4f93-b573-f1e4c9b55c62","order_by":1,"name":"Zhi-Yu Chen","email":"","orcid":"","institution":"Xiamen University","correspondingAuthor":false,"prefix":"","firstName":"Zhi-Yu","middleName":"","lastName":"Chen","suffix":""},{"id":546343242,"identity":"bdc0f603-6e7d-4003-b18b-4968d01d3f81","order_by":2,"name":"Kan Wang","email":"","orcid":"","institution":"Xiamen University","correspondingAuthor":false,"prefix":"","firstName":"Kan","middleName":"","lastName":"Wang","suffix":""},{"id":546343243,"identity":"5ec849bf-ec1b-45f6-9c98-f3a47d5fb3bf","order_by":3,"name":"Hao-Yu Qin","email":"","orcid":"","institution":"Xiamen University","correspondingAuthor":false,"prefix":"","firstName":"Hao-Yu","middleName":"","lastName":"Qin","suffix":""},{"id":546343244,"identity":"2f7ae1d4-0a92-4f8d-a932-b162bc13f52f","order_by":4,"name":"Yang Yang","email":"","orcid":"","institution":"Xiamen University","correspondingAuthor":false,"prefix":"","firstName":"Yang","middleName":"","lastName":"Yang","suffix":""},{"id":546343245,"identity":"aac0a8ca-1487-41e6-8768-41dc0b6c7cba","order_by":5,"name":"Dong-Tai Liu","email":"","orcid":"https://orcid.org/0000-0002-7293-9717","institution":"Xiamen University","correspondingAuthor":false,"prefix":"","firstName":"Dong-Tai","middleName":"","lastName":"Liu","suffix":""},{"id":546343246,"identity":"e162a5e3-e406-4901-82b0-c83029f22b17","order_by":6,"name":"Shu-Han Yang","email":"","orcid":"","institution":"Xiamen University","correspondingAuthor":false,"prefix":"","firstName":"Shu-Han","middleName":"","lastName":"Yang","suffix":""},{"id":546343247,"identity":"def739a3-58cd-4e88-ac8a-6c111bc3c139","order_by":7,"name":"Ximin Chi","email":"","orcid":"","institution":"Xiamen University","correspondingAuthor":false,"prefix":"","firstName":"Ximin","middleName":"","lastName":"Chi","suffix":""},{"id":546343248,"identity":"4ec508d1-2fb6-433b-b9f2-0c7dbb3e8488","order_by":8,"name":"Ruixin Liu","email":"","orcid":"","institution":"Shanghai Institute of Endocrine and Metabolic Diseases, Ruijin Hospital, Shanghai Jiao Tong University School of Medicine, Shanghai, China","correspondingAuthor":false,"prefix":"","firstName":"Ruixin","middleName":"","lastName":"Liu","suffix":""},{"id":546343249,"identity":"d6d9ceb1-4e67-43c6-a0d7-73663ae6e88e","order_by":9,"name":"Chao Wu","email":"","orcid":"","institution":"Department of Endocrine and Metabolic Diseases, Shanghai Institute of Endocrine and Metabolic Diseases, Ruijin Hospital, Shanghai Jiao Tong University School of Medicine","correspondingAuthor":false,"prefix":"","firstName":"Chao","middleName":"","lastName":"Wu","suffix":""},{"id":546343250,"identity":"04681a87-2fb4-4c75-bea9-84aa6575b785","order_by":10,"name":"Terytty Yang Li","email":"","orcid":"https://orcid.org/0000-0003-3684-7935","institution":"Fudan University","correspondingAuthor":false,"prefix":"","firstName":"Terytty","middleName":"Yang","lastName":"Li","suffix":""},{"id":546343251,"identity":"ab1999a7-662f-4622-ae0b-23a46bd5d4ad","order_by":11,"name":"Lin Chen","email":"","orcid":"","institution":"Fudan University","correspondingAuthor":false,"prefix":"","firstName":"Lin","middleName":"","lastName":"Chen","suffix":""},{"id":546343252,"identity":"d0446a07-5f24-46a5-9092-8a52c2687bc9","order_by":12,"name":"Hao Guo","email":"","orcid":"","institution":"Xiamen University","correspondingAuthor":false,"prefix":"","firstName":"Hao","middleName":"","lastName":"Guo","suffix":""},{"id":546343253,"identity":"110b7352-b38d-4ec2-a975-54f15c0a0050","order_by":13,"name":"Pu-Chun Zhu","email":"","orcid":"","institution":"Xiamen University","correspondingAuthor":false,"prefix":"","firstName":"Pu-Chun","middleName":"","lastName":"Zhu","suffix":""},{"id":546343254,"identity":"5ec6a55a-d42b-45fa-bd08-4d9b023b5a30","order_by":14,"name":"Wen-Jing Wu","email":"","orcid":"","institution":"Xiamen University","correspondingAuthor":false,"prefix":"","firstName":"Wen-Jing","middleName":"","lastName":"Wu","suffix":""},{"id":546343255,"identity":"e45bd6d1-aca4-4f88-b52a-91780a86dda6","order_by":15,"name":"Yang Song","email":"","orcid":"","institution":"The Chinese University of Hong Kong","correspondingAuthor":false,"prefix":"","firstName":"Yang","middleName":"","lastName":"Song","suffix":""},{"id":546343256,"identity":"dbb43483-bb91-4fe1-b244-7ddf2dabb668","order_by":16,"name":"Qinghua Zhao","email":"","orcid":"","institution":"Shandong University","correspondingAuthor":false,"prefix":"","firstName":"Qinghua","middleName":"","lastName":"Zhao","suffix":""},{"id":546343257,"identity":"24bfa656-e133-4647-9fb3-e4ce198364eb","order_by":17,"name":"Changyu Wang","email":"","orcid":"","institution":"Shandong University","correspondingAuthor":false,"prefix":"","firstName":"Changyu","middleName":"","lastName":"Wang","suffix":""},{"id":546343258,"identity":"e75dfe93-d2d7-414d-be89-580cce6cb74f","order_by":18,"name":"Ming Jiang","email":"","orcid":"","institution":"Xiamen University","correspondingAuthor":false,"prefix":"","firstName":"Ming","middleName":"","lastName":"Jiang","suffix":""},{"id":546343259,"identity":"8cfe4964-3307-4f53-9a68-ca097de8afa7","order_by":19,"name":"Mingxia Zhu","email":"","orcid":"https://orcid.org/0000-0002-1858-5655","institution":"School of Life Sciences, Xiamen University","correspondingAuthor":false,"prefix":"","firstName":"Mingxia","middleName":"","lastName":"Zhu","suffix":""},{"id":546343260,"identity":"7e042c92-3504-4000-a665-f4d1d443e07a","order_by":20,"name":"Cixiong Zhang","email":"","orcid":"","institution":"State Key Laboratory of Cellular Stress Biology, Innovation Center for Cell Signaling Network, School of Life Sciences, Xiamen University, Xiamen, Fujian 361102","correspondingAuthor":false,"prefix":"","firstName":"Cixiong","middleName":"","lastName":"Zhang","suffix":""},{"id":546343261,"identity":"64f76e2e-92f0-4265-80d2-21eb2b769e98","order_by":21,"name":"Lei Zhang","email":"","orcid":"","institution":"State Key Laboratory of Cellular Stress Biology, School of Life Sciences, Faculty of Medicine and Life Sciences, Xiamen University","correspondingAuthor":false,"prefix":"","firstName":"Lei","middleName":"","lastName":"Zhang","suffix":""},{"id":546343262,"identity":"81f2ddee-4054-4eed-989e-d1087d32133c","order_by":22,"name":"Yaying Wu","email":"","orcid":"","institution":"Xiamen University","correspondingAuthor":false,"prefix":"","firstName":"Yaying","middleName":"","lastName":"Wu","suffix":""},{"id":546343263,"identity":"cd82b455-29f5-4ab3-a1fd-97814e165b15","order_by":23,"name":"Ao Li","email":"","orcid":"","institution":"Xiamen University","correspondingAuthor":false,"prefix":"","firstName":"Ao","middleName":"","lastName":"Li","suffix":""},{"id":546343264,"identity":"61851ec3-047e-4c65-a0ab-2e2685cc874d","order_by":24,"name":"Luming Yao","email":"","orcid":"","institution":"State Key Laboratory of Cellular Stress Biology, Innovation Center for Cell Signaling Network, School of Life Sciences, Xiamen University, Xiamen, Fujian 361102","correspondingAuthor":false,"prefix":"","firstName":"Luming","middleName":"","lastName":"Yao","suffix":""},{"id":546343265,"identity":"8b721800-dbe8-4905-adda-08e75a0677b0","order_by":25,"name":"Mingliang Zhang","email":"","orcid":"","institution":"Shanghai Diabetes Institute, Shanghai Key Laboratory of Diabetes Mellitus, Shanghai Clinical Center for Diabetes, Shanghai Jiao Tong University Affiliated Sixth People's Hospital.","correspondingAuthor":false,"prefix":"","firstName":"Mingliang","middleName":"","lastName":"Zhang","suffix":""},{"id":546343266,"identity":"09b9b866-0d55-47e3-a384-c4ec28a418d8","order_by":26,"name":"Feng Guo","email":"","orcid":"","institution":"Xiamen University","correspondingAuthor":false,"prefix":"","firstName":"Feng","middleName":"","lastName":"Guo","suffix":""},{"id":546343267,"identity":"bde9686f-4833-415b-8a21-d35127af4d0a","order_by":27,"name":"Qi-Qian Han","email":"","orcid":"","institution":"Xiamen University","correspondingAuthor":false,"prefix":"","firstName":"Qi-Qian","middleName":"","lastName":"Han","suffix":""},{"id":546343268,"identity":"e26be8f2-ba52-4991-b2db-0eedb0650303","order_by":28,"name":"Wei Hong","email":"","orcid":"","institution":"Xiamen University","correspondingAuthor":false,"prefix":"","firstName":"Wei","middleName":"","lastName":"Hong","suffix":""},{"id":546343269,"identity":"b594deb1-c391-460c-b67c-7d255b7ee84c","order_by":29,"name":"Tian-Yu Tang","email":"","orcid":"https://orcid.org/0000-0002-3339-7564","institution":"Xiamen University","correspondingAuthor":false,"prefix":"","firstName":"Tian-Yu","middleName":"","lastName":"Tang","suffix":""},{"id":546343270,"identity":"666c3858-ba59-45ec-878f-cc0ee4416c4a","order_by":30,"name":"Jianlin Ren","email":"","orcid":"","institution":"Xiamen University","correspondingAuthor":false,"prefix":"","firstName":"Jianlin","middleName":"","lastName":"Ren","suffix":""},{"id":546343271,"identity":"38f95b3e-e4fb-4477-aec0-2742f66e1635","order_by":31,"name":"Chang Liu","email":"","orcid":"https://orcid.org/0000-0001-9398-3701","institution":"Shandong University","correspondingAuthor":false,"prefix":"","firstName":"Chang","middleName":"","lastName":"Liu","suffix":""},{"id":546343272,"identity":"db78e564-597a-47ba-933b-2fc39388421c","order_by":32,"name":"Hongzhi Xu","email":"","orcid":"","institution":"zhongshan hospital, xiamen university","correspondingAuthor":false,"prefix":"","firstName":"Hongzhi","middleName":"","lastName":"Xu","suffix":""},{"id":546343273,"identity":"6da1eee5-a5fe-40a1-b029-a9b3b1681238","order_by":33,"name":"Shuang-Jiang Liu","email":"","orcid":"https://orcid.org/0000-0002-7585-310X","institution":"Shandong University","correspondingAuthor":false,"prefix":"","firstName":"Shuang-Jiang","middleName":"","lastName":"Liu","suffix":""},{"id":546343274,"identity":"a65d493a-4468-482d-a4ac-04ba74592af7","order_by":34,"name":"Sheng-Cai Lin","email":"","orcid":"https://orcid.org/0000-0003-1993-8376","institution":"Xiamen University","correspondingAuthor":false,"prefix":"","firstName":"Sheng-Cai","middleName":"","lastName":"Lin","suffix":""}],"badges":[],"createdAt":"2025-11-06 08:37:04","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-8045609/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-8045609/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":96259999,"identity":"66a30b67-bd43-49c3-80dc-c6e8d586f9ec","added_by":"auto","created_at":"2025-11-19 07:56:16","extension":"jpg","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":270342,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eAIDA depletion enriches gut bacterial taxa linked to obesity.\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003ea,\u003c/strong\u003e Beta-diversity of cecal microbiota in WT and \u003cem\u003eAida\u003c/em\u003e\u003csup\u003e\u003cem\u003e-/-\u003c/em\u003e\u003c/sup\u003e mice visualised by principal coordinates analysis (PCoA) of Bray-Curtis distances at the operational taxonomic unit (OTU) level. Statistical significance was tested by Analysis of Variance Using Distance Matrices (ADONIS).\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eb,\u003c/strong\u003e Volcano plot of differentially abundant OTUs (\u003cem\u003eAida\u003c/em\u003e\u003csup\u003e\u003cem\u003e-/-\u003c/em\u003e\u003c/sup\u003e vs. WT). Significance of differences was assessed by multiple two-tailed unpaired Mann-Whitney test.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003ec,\u003c/strong\u003e Clinical relevance of \u003cem\u003eAnaerotruncus colihominis\u003c/em\u003e (OTU68). Faecal metagenomic data from 355 normal-weight (BMI 18.5-24.9) and 631 obese (BMI \u0026gt; 30) individuals were analysed. Violin plots (left) depict the relative abundances of \u003cem\u003eA. colihominis\u003c/em\u003e. Medians (solid lines) and interquartile ranges (dashed lines) are shown. Significance was assessed using two-tailed unpaired Mann-Whitney test. Correlation analyses (right) show Spearman’s rank correlation between BMI and taxon abundances in obese individuals. Correlation coefficients (\u003cem\u003eρ\u003c/em\u003e) and \u003cem\u003eP\u003c/em\u003e values are annotated.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003ed,\u003c/strong\u003e Pearson correlation analysis between fat mass (%) and relative abundance of OTU209. Shaded regions indicate 95% confidence intervals (CI). Correlation coefficients (\u003cem\u003er\u003c/em\u003e) and \u003cem\u003eP\u003c/em\u003e values are annotated.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003ee,\u003c/strong\u003e Abundance of fatty acid species in the luminal content of the mice cecum. Cecal contents were subjected to GC-MS analysis. The abundances of major fatty acid species in the luminal content were expressed as relative ratios to D-fructose. Data represent individual mouse (\u003cem\u003en\u003c/em\u003e = 5) with median ± 95% CI. Statistical differences were assessed by two-way ANOVA followed by Holm-Šídák’s post-hoc testing.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003ef,\u003c/strong\u003e Growth kinetics of LDE-1 (left), LDE-2 (middle), and \u003cem\u003eBacteroides uniformis\u003c/em\u003e (right) in basal medium with or without 10 μM sodium oleate. Vertical error bars indicate standard error of the mean (SEM) of three biological replicates. Solid lines denote best-fit logistic growth curves derived from nonlinear regression. Curve differences relative to vehicle controls are assessed by an extra sum-of-squares \u003cem\u003eF\u003c/em\u003e-test.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eg-j,\u003c/strong\u003e Growth kinetics of LDE-1 and LDE-2 in basal media supplemented with \u003cstrong\u003eg,\u003c/strong\u003e 100 μM sodium palmitate (C16:0), stearate (C18:0), or oleate (C18:1); \u003cstrong\u003eh,\u003c/strong\u003e 100 μM sodium myristate (C14:0); \u003cstrong\u003ei\u003c/strong\u003e, 5 mM sodium acetate (C2:0), propionate (C3:0), or butyrate (C4:0); \u003cstrong\u003ej\u003c/strong\u003e. 100 μM sodium caprylate (C8:0) or decanoate (C10:0). Vertical error bars indicate SEM of three biological replicates. Solid lines denote best-fit logistic growth curves derived from nonlinear regression. Curve differences relative to vehicle controls are assessed by extra sum-of-squares \u003cem\u003eF\u003c/em\u003e-test.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003ek,\u003c/strong\u003e Olive oil gavage reduced OTU68 and OTU209 in \u003cem\u003eAida\u003c/em\u003e\u003csup\u003e\u003cem\u003e-/-\u003c/em\u003e\u003c/sup\u003e mice. Mice received daily oral gavage of 200 μL saline or olive oil for 14 days. OTU abundances in cecal faeces were quantified by quantitative PCR (qPCR) and normalised to total bacterial 16S rRNA gene levels. Data represent individual mice (\u003cem\u003en\u003c/em\u003e = 7 per group), with median ± 95% CI. Statistical significance was determined by two-tailed unpaired Mann-Whitney test.\u003c/p\u003e\n\u003cp\u003e****\u003cem\u003eP\u003c/em\u003e \u0026lt; 0.0001, **\u003cem\u003eP\u003c/em\u003e \u0026lt; 0.01, *\u003cem\u003eP\u003c/em\u003e \u0026lt; 0.05, N.S. not significant.\u003c/p\u003e","description":"","filename":"1.jpg","url":"https://assets-eu.researchsquare.com/files/rs-8045609/v1/35d9aeaa71feb806f14be10f.jpg"},{"id":96259700,"identity":"ae8ce4df-d5fa-4e46-a63f-51f40e179661","added_by":"auto","created_at":"2025-11-19 07:56:00","extension":"jpg","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":238953,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eLDE-1 and LDE-2 ameliorate diet-induced obesity and metabolic dysfunction in mice.\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003ea,\u003c/strong\u003eExperimental design for oral gavage of LDE-1 and LDE-2, or phosphate-buffered saline (PBS) in high-fat diet (HFD)-fed mice.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eb, \u003c/strong\u003eBody weight trajectories of mice treated with LDE-1, LDE-2, or PBS. Slopes were analysed by simple linear regression analysis. Data are mean ± SEM, \u003cem\u003en\u003c/em\u003e = 6 mice per group.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003ec,d, \u003c/strong\u003eMass of inguinal white adipose tissue (iWAT; \u003cstrong\u003ec\u003c/strong\u003e) and gonadal white adipose tissue (gWAT; \u003cstrong\u003ed\u003c/strong\u003e). Data represent individual mice (\u003cem\u003en\u003c/em\u003e = 6 per group), with mean ± 95% CI. Differences were assessed by one-way ANOVA followed by Dunnett's test.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003ee-h,\u003c/strong\u003e Adipocyte morphology in gWAT. Representative hematoxylin and eosin (H\u0026amp;E)-stained sections from LDE-1- (\u003cstrong\u003ee\u003c/strong\u003e) or LDE-2-treated in (\u003cstrong\u003eg\u003c/strong\u003e) mice. Adipocyte cross-sectional areas were quantified for LDE-1 (\u003cstrong\u003ef\u003c/strong\u003e; \u003cem\u003en\u003c/em\u003e = 7 or 8 mice) and LDE-2 (\u003cstrong\u003eh\u003c/strong\u003e; \u003cem\u003en\u003c/em\u003e = 5 mice). Violin plots show median (solid line) and quartiles (dashed lines). Statistical significance was determined by two-tailed unpaired Mann-Whitney test.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003ei,j,\u003c/strong\u003eSoleus (\u003cstrong\u003ei\u003c/strong\u003e) and quadriceps (\u003cstrong\u003ej\u003c/strong\u003e) muscle mass. Data are individual values (\u003cem\u003en\u003c/em\u003e = 6 mice per group) with mean ± 95% CI. Differences were analysed by one-way ANOVA.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003ek\u003c/strong\u003e-\u003cstrong\u003em,\u003c/strong\u003e Exercise capacity assessment. Tolerance to exhaustion (distance, \u003cstrong\u003ek\u003c/strong\u003e; time, \u003cstrong\u003ei\u003c/strong\u003e) and forelimb grip strength (\u003cstrong\u003em\u003c/strong\u003e). Data represent individual mice (\u003cem\u003en\u003c/em\u003e = 6 per group) with mean ± 95% CI. One-way ANOVA followed by Dunnett’s test.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003en,o,\u003c/strong\u003eSerum alanine aminotransferase (ALT; \u003cstrong\u003en\u003c/strong\u003e) and blood urea nitrogen (BUN; \u003cstrong\u003eo\u003c/strong\u003e) levels. Data are individual values (\u003cem\u003en\u003c/em\u003e = 6 mice per group) with mean ± 95% CI. One-way ANOVA.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003ep,\u003c/strong\u003eDaily food intake per cage. Food intake was measured per cage (2 cages per group, each housing 3 mice) daily for 3 consecutive days. Individual mouse values were calculated by dividing cage-level intake by the number of mice per cage. Data are estimated individual mouse values with mean ± 95% CI. One-way ANOVA.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eq-s\u003c/strong\u003e, Long-term metabolic effects of LDE-2 treatment. Fasting blood glucose (\u003cstrong\u003eq\u003c/strong\u003e) and liver-to-body weight ratios (\u003cstrong\u003er\u003c/strong\u003e) at 12 weeks post-treatment. Data are individual values (\u003cem\u003en\u003c/em\u003e = 5 mice per group) with mean ± 95% CI. Differences were assessed by two-tailed unpaired Student’s \u003cem\u003et\u003c/em\u003e-test. (\u003cstrong\u003es\u003c/strong\u003e) Representative H\u0026amp;E-stained liver sections.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003et\u003c/strong\u003e, Hepatic steatosis quantification. Steatosis scores reflect the percentage of hepatocytes containing lipid droplets. Data represent individual mice (\u003cem\u003en\u003c/em\u003e= 5 per group), with median ± 95% CI. Statistical significance was determined by two-tailed unpaired Mann-Whitney test.\u003c/p\u003e\n\u003cp\u003e****\u003cem\u003eP\u003c/em\u003e \u0026lt; 0.0001, **\u003cem\u003eP\u003c/em\u003e \u0026lt; 0.01, *\u003cem\u003eP\u003c/em\u003e\u0026lt; 0.05.\u003c/p\u003e","description":"","filename":"2.jpg","url":"https://assets-eu.researchsquare.com/files/rs-8045609/v1/c0a808a34b02617eb4c09175.jpg"},{"id":96259578,"identity":"2d4a57a1-efcc-42c2-8cd2-068fe6a262b8","added_by":"auto","created_at":"2025-11-19 07:55:55","extension":"jpg","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":182167,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eLeucic acid is the active metabolite that phenocopies bacterial benefits.\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003ea,b,\u003c/strong\u003e Heatmaps of metabolites detected in pre- and post-culture media of LDE-1 (\u003cstrong\u003ea\u003c/strong\u003e) and LDE-2 (\u003cstrong\u003eb\u003c/strong\u003e) by untargeted metabolomics. Metabolites with variable importance in the projection (VIP) score \u0026gt; 1 are shown.Values are z-score normalised (row-wise mean-centred and scaled by standard deviation). Metabolites significantly enriched in post-culture medium of both LDE-1 and LDE-2 are indicated.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003ec,\u003c/strong\u003eVolcano plots of shared and strain-specific metabolite production by comparing pre- and post-culture media of LDE-1 and LDE-2 by untargeted metabolomics. Metabolites significantly increased (FDR \u0026lt; 1) in post-culture media of LDE-1 and LDE-2 (\u003cem\u003en\u003c/em\u003e = 3 biological replicates per group). The differences in the metabolite abundance between the pre-culture and post-culture media were assessed using two-tailed unpaired Student’s \u003cem\u003et\u003c/em\u003e-test with false discovery rate (FDR) correction (LDE-1: \u0026lt; 0.2%; LDE-2: \u0026lt; 2%).\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003ed,\u003c/strong\u003e Body weight changes in mice fed a normal chow diet (NCD), HFD, or HFD supplemented with leucic acid (initiated 5 weeks post-HFD). Data are mean ± SEM, \u003cem\u003en\u003c/em\u003e = 3 (NCD + ADW), 5 (HFD + ADW), and 5 (HFD + leucic acid) mice. Slopes compared by simple linear regression analysis.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003ee,f,\u003c/strong\u003eGonadal white adipose tissue (gWAT; \u003cstrong\u003ee\u003c/strong\u003e) and inguinal white adipose tissue (iWAT; \u003cstrong\u003ef\u003c/strong\u003e) mass. Data represent individual mice (\u003cem\u003en\u003c/em\u003e = 3-5 per group) with mean ± 95% CI. Differences assessed by one-way ANOVA followed by Tukey’s multiple comparisons test.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eg,h,\u003c/strong\u003eFunctional metabolic assessments. \u003cstrong\u003eg,\u003c/strong\u003eForelimb grip strength. \u003cstrong\u003eh,\u003c/strong\u003e Treadmill exhaustion distance. Data are individual values (\u003cem\u003en\u003c/em\u003e= 6-10 mice per group) with mean ± 95% CI. Unpaired two-tailed Student’s \u003cem\u003et\u003c/em\u003e-test.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003ei,\u003c/strong\u003eLeucic acid increases basal body temperature in mice. Basal rectal temperature was measured in HFD-fed WT mice after 1 week of treatment with leucic acid or ADW (\u003cem\u003en\u003c/em\u003e = 10 mice per group). Data are means of two measurements per mouse ± 95% CI. Unpaired two-tailed Student’s \u003cem\u003et\u003c/em\u003e-test.\u003c/p\u003e\n\u003cp\u003e****\u003cem\u003eP\u003c/em\u003e \u0026lt; 0.0001, ***\u003cem\u003eP\u003c/em\u003e \u0026lt; 0.001, **\u003cem\u003eP\u003c/em\u003e\u0026lt; 0.01, *\u003cem\u003eP\u003c/em\u003e \u0026lt; 0.05.\u003c/p\u003e","description":"","filename":"3.jpg","url":"https://assets-eu.researchsquare.com/files/rs-8045609/v1/0e89f47d8dab6b44ddcd62ed.jpg"},{"id":96259603,"identity":"904ef759-f70b-4a46-bd8c-c1418ec747b6","added_by":"auto","created_at":"2025-11-19 07:55:57","extension":"jpg","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":246557,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eLeucic acid suppresses lipid storage through depot-divergent effects.\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003ea,\u003c/strong\u003e Multi-omics analysis of leucic acid-responsive pathways in gWAT. Upper: Venn diagram of differentially expressed genes (DEGs; |log\u003csub\u003e2\u003c/sub\u003e(fold change)| \u0026gt; 0.5, \u003cem\u003eP\u003c/em\u003e \u0026lt; 0.05) from proteomic and transcriptomic profiles of gWAT from HFD-fed WT mice treated with leucic acid (Leuc) or ADW (control) for 2 weeks. Lower: enrichment analysis (ShinyGO) of downregulated candidates in lipid anabolic pathways.\u0026nbsp;\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eb,\u003c/strong\u003e Volcano plots of mRNA (left; \u003cem\u003en\u003c/em\u003e = 4 per group) and protein (right; \u003cem\u003en\u003c/em\u003e = 3 per group) abundance changes in gWAT (Leuc vs. ADW). Significantly altered molecules (\u003cem\u003elimma-voom\u003c/em\u003e;\u003cem\u003e \u003c/em\u003e|log\u003csub\u003e2\u003c/sub\u003e(fold change)| \u0026gt; 0.5, \u003cem\u003eP\u003c/em\u003e \u0026lt; 0.05) in red (up) or blue (down).\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003ec,\u003c/strong\u003e Heatmap of fatty acid and triglyceride synthesis enzymes in gWAT of leuc- or ADW-treated WT mice.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003ed,e,\u003c/strong\u003e Leucic acid suppresses lipid synthesis machinery in gWAT. \u003cstrong\u003ed,\u003c/strong\u003e Immunoblots of key lipogenic proteins (FASN, ACSS2, DGAT2, MOGAT2) in gWAT lysates from HFD-fed WT and \u003cem\u003eAida\u003c/em\u003e\u003csup\u003e\u003cem\u003e-/-\u003c/em\u003e\u003c/sup\u003e mice treated with 50 mM leucic acid or ADW control for 6 weeks. FABP4 served as loading control (15 μg protein per lane). \u003cstrong\u003ee,\u003c/strong\u003e Quantification of immunoblot signals expressed relative to the mean of respective ADW control groups. Data represent individual mice (\u003cem\u003en\u003c/em\u003e = 5 WT, \u003cem\u003en\u003c/em\u003e = 6 \u003cem\u003eAida\u003c/em\u003e\u003csup\u003e\u003cem\u003e-/-\u003c/em\u003e\u003c/sup\u003e) with median ± 95% CI. Statistical significance for protein levels was determined by one-way ANOVA followed by Holm-Šídák's test when ANOVA indicated significant differences (\u003cem\u003eP\u003c/em\u003e \u0026lt; 0.05).\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003ef,\u003c/strong\u003e Gene Set Enrichment Analysis (GSEA) of thermogenesis pathway in leuc-treated \u003cem\u003eAida\u003c/em\u003e\u003csup\u003e\u003cem\u003e-/-\u003c/em\u003e\u003c/sup\u003e iWAT. Normalised Enrichment Score (NES) and statistical significance shown.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eg,h,\u003c/strong\u003e Leucic acid modulates lipid synthesis and thermogenic programs in iWAT. \u003cstrong\u003eg,\u003c/strong\u003e Immunoblots of lipid synthesis regulators (FASN, DGAT2) and thermogenic marker UCP1 in iWAT lysates from mice described in \u003cstrong\u003ed\u003c/strong\u003e. FABP4 served as loading control (15 μg protein per lane). \u003cstrong\u003eh,\u003c/strong\u003e Quantification of immunoblot signals using methodology identical to \u003cstrong\u003ee\u003c/strong\u003e, with statistical analysis of DGAT2 levels performed via Kruskal-Wallis test followed by Dunn's test due to violation of normality assumptions. \u0026nbsp;\u003c/p\u003e\n\u003cp\u003e****\u003cem\u003eP\u003c/em\u003e \u0026lt; 0.0001, ***\u003cem\u003eP\u003c/em\u003e \u0026lt; 0.001, **\u003cem\u003eP\u003c/em\u003e \u0026lt; 0.01.\u003c/p\u003e","description":"","filename":"4.jpg","url":"https://assets-eu.researchsquare.com/files/rs-8045609/v1/90c2cea15cc1e039d1edd30b.jpg"},{"id":96259742,"identity":"c8e7a6f1-8b8f-48d6-9f62-069ce257de09","added_by":"auto","created_at":"2025-11-19 07:56:02","extension":"jpg","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":123698,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eLeucic acid directly enforces lipid catabolism in an enantioselective manner.\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003ea,\u003c/strong\u003e Lipid droplet accumulation in WT and \u003cem\u003eAida\u003c/em\u003e\u003csup\u003e\u003cem\u003e-/-\u003c/em\u003e\u003c/sup\u003e mouse embryonic fibroblasts (MEFs) pre-treated with 50 μM D- or L-leucic acid (sodium salt) or NaCl (control) for 36 h, then 250 μM oleic acid for 12 h. Stained with BODIPY 493/503. Scale bar: 20 μm.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eb,\u003c/strong\u003e Quantification of lipid droplet fluorescence intensity (\u003cem\u003en\u003c/em\u003e = 326-419 cells per group from 2 fields). Data normalised to WT control. RFU: relative fluorescence intensity. Violin plots show median (solid line) and quartiles (dashed lines). Kruskal-Wallis followed by Dunn’s test.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003ec,\u003c/strong\u003e Isotopic enrichment (M+2 species) in 3T3-L1 derived beige-like adipocytes. Cells pre-treated with 50 μM D- or L-leucic acid or NaCl for 24 h, then 250 μM [U-\u003csup\u003e13\u003c/sup\u003eC]oleic acid for 1 h. Data are presented as individual values with mean ± 95% CI, two-way ANOVA followed by Dunnett’s test.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003ed,\u003c/strong\u003e Oxygen consumption rate (OCR) in MEFs treated with 50 μM D- or L-leucic acid. MEFs were monitored using the Resipher continuous oxygen consumption reader every 15 min for 1 h, and shown as individual values (4 replicate wells for each group, each with 4 time points recorded) with mean ± 95% CI, two-way ANOVA followed by Dunnett’s test.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003ee,\u003c/strong\u003e Schematic diagram of the effect of leucic acid on lipid metabolism across adipose depots.\u003c/p\u003e\n\u003cp\u003e****\u003cem\u003eP\u003c/em\u003e \u0026lt; 0.0001, ***\u003cem\u003eP\u003c/em\u003e \u0026lt; 0.001, *\u003cem\u003eP\u003c/em\u003e \u0026lt; 0.05.\u003c/p\u003e","description":"","filename":"5.jpg","url":"https://assets-eu.researchsquare.com/files/rs-8045609/v1/dda8744e679e4f750fcb9209.jpg"},{"id":96259612,"identity":"61261059-2e94-418e-a34b-ed18bbdb63fc","added_by":"auto","created_at":"2025-11-19 07:55:59","extension":"jpg","order_by":6,"title":"Figure 6","display":"","copyAsset":false,"role":"figure","size":373191,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eHCAR2 mediates leucic acid-induced suppression of lipid accumulation.\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003ea,\u003c/strong\u003e Integrative target prediction for leucic acid enantiomers. Venn diagram shows 13 consensus targets shared by Ensemble TargetNet (ETN; probability \u0026gt; 0.7) and SwissTargetPrediction (STP; top 100 ranked targets). Magenta labels: high-confidence candidates.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eb,\u003c/strong\u003e Genetic associations of \u003cem\u003eHCAR2\u003c/em\u003e (12q24.31) and \u003cem\u003eGRM2\u003c/em\u003e (3p21.2) loci with obesity traits. Forest plot showing effect sizes (β) and 95% CI of intergenic variants identified through GWAS Catalogue screening. Traits: Waist-to-hip ratio (WHR), Body fat percentage (Body fat %), ratio of triglycerides to phosphoglycerides (TG ratio), and body mass index (BMI).\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003ec,\u003c/strong\u003e Transcriptional response to leucic acid in adipose tissue. Dual volcano plots assess differential expression by fold-change magnitude (|log\u003csub\u003e2\u003c/sub\u003e(fold change)| \u0026gt; 0.5, \u003cem\u003eP \u003c/em\u003e\u0026lt; 0.05) and absolute abundance change (|ΔCPM| \u0026gt; 10, \u003cem\u003eP\u003c/em\u003e \u0026lt; 0.05). \u003cem\u003eHcar2\u003c/em\u003e and \u003cem\u003eCar4\u003c/em\u003e (magenta) are the only significantly altered genes by both metrics (\u003cem\u003elimma-voom\u003c/em\u003e; \u003cem\u003en\u003c/em\u003e = 3-4 per group).\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003ed,\u003c/strong\u003e Target identification workflow.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003ee,\u003c/strong\u003e Predicted binding modes of D-leucate (yellow) and L-leucate (green) within the orthosteric pocket of HCAR2. Molecular docking simulations were performed using the cryo-EM structure of niacin-bound human HCAR2 in complex with the Gi heterotrimer (PDB ID: 8H2G) as the structural template. Key interacting residues are labelled, with polar residues highlighted in red. Binding free energies were calculated using AutoDock Vina.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003ef,\u003c/strong\u003e \u003csup\u003e19\u003c/sup\u003eF-NMR ligand-detected binding assay in intact cells. Spectra of 4-fluoroleucic acid with HCAR2-expressing cells show a significant chemical shift. Reference: trifluoroacetic acid (TFA).\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eg,\u003c/strong\u003e \u003cem\u003eHcar2\u003c/em\u003e knockdown abolishes lipid reduction in beige-like adipocytes. Representative BODIPY (lipids, yellow) and MitoTracker (mitochondria, red) images. Scale bar: 20 μm.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eh,\u003c/strong\u003e Quantification of lipid contents in \u003cem\u003eHcar2\u003c/em\u003e-knockdown cells. Data are presented as individual cells (\u003cem\u003en\u003c/em\u003e = 12-13 per group) with median ± 95% CI. Two-way ANOVA followed by Holm-Šídák's test.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003ei, \u003c/strong\u003e\u003cem\u003eHcar2\u003c/em\u003e-dependent oxygen consumption. Maximum mitochondrial oxygen consumption in beige-like adipocytes. Data are presented as individual experiments (\u003cem\u003en\u003c/em\u003e = 3) with median ± 95% CI, two-way ANOVA followed by Dunnett’s test.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003ej,\u003c/strong\u003e Experimental design for unilateral AAV-mediated \u003cem\u003eHcar2\u003c/em\u003e knockdown in iWAT. WT mice received in situ injections of AAV-\u003cem\u003eshHcar2\u003c/em\u003e or AAV-\u003cem\u003eshRenilla\u003c/em\u003e as a control at four sites per iWAT pad (contralateral control). Two weeks post-injection, mice were switched from NCD to HFD. After 1 week of HFD adaptation, leucic acid or vehicle ADW treatment was administered for 12 weeks prior to tissue collection.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003ek,l,\u003c/strong\u003e \u003cem\u003eHcar2\u003c/em\u003e-dependent attenuation of leucic acid-induced iWAT mass reduction and browning. \u003cstrong\u003ek,\u003c/strong\u003e Representative images for iWAT depots. \u003cstrong\u003el,\u003c/strong\u003e Mass ratios of\u0026nbsp;\u003cem\u003eHcar2\u003c/em\u003e-knockdown iWAT to contralateral \u003cem\u003eshRenilla\u003c/em\u003e-treated iWAT in the same animals. Data shown as individual points with mean ± 95% CI; \u003cem\u003en\u003c/em\u003e = 6 mice; statistical significance determined by two-tailed Student's \u003cem\u003et\u003c/em\u003e-test with Welch's correction.\u003c/p\u003e\n\u003cp\u003e****\u003cem\u003eP\u003c/em\u003e \u0026lt; 0.0001, **\u003cem\u003eP\u003c/em\u003e \u0026lt; 0.01, *\u003cem\u003eP\u003c/em\u003e \u0026lt; 0.05.\u003c/p\u003e","description":"","filename":"6.jpg","url":"https://assets-eu.researchsquare.com/files/rs-8045609/v1/4751bac35bec27d61f44b49c.jpg"},{"id":96261262,"identity":"89580d9f-534b-4bba-a970-1c5b9f64a296","added_by":"auto","created_at":"2025-11-19 07:58:01","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":4165581,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-8045609/v1/e83f04a7-f7f6-4acd-a5c3-2d1dc9593801.pdf"},{"id":96259905,"identity":"90bc68fe-ee64-4948-aa45-358271fd9e63","added_by":"auto","created_at":"2025-11-19 07:56:15","extension":"xlsx","order_by":1,"title":"","display":"","copyAsset":false,"role":"supplement","size":16263425,"visible":true,"origin":"","legend":"Supplementary Table 1-18","description":"","filename":"SupplementaryTable118.xlsx","url":"https://assets-eu.researchsquare.com/files/rs-8045609/v1/5c8856749bf5f2f947badd36.xlsx"},{"id":96259691,"identity":"1a45665d-96bd-40e1-889a-e2d96a376054","added_by":"auto","created_at":"2025-11-19 07:56:00","extension":"docx","order_by":2,"title":"","display":"","copyAsset":false,"role":"supplement","size":4035201,"visible":true,"origin":"","legend":"","description":"","filename":"Additionalinformation.docx","url":"https://assets-eu.researchsquare.com/files/rs-8045609/v1/82132875b25b99301a9ab38f.docx"}],"financialInterests":"\u003cb\u003eYes\u003c/b\u003e there is potential Competing Interest.\nThe authors declare the following financial interests: The bacterial strains LDE-1 and LDE-2 are the subject of a patent deposition. The findings regarding the use of leucic acid against obesity are also intended for patent protection. The authors may potentially file related patent applications in the future, which could lead to the commercialisation of technology.","formattedTitle":"Gut bacteria enriched in obese mice produce leucic acid to counter obesity through HCAR2","fulltext":[{"header":"Main","content":"\u003cp\u003eCoevolution between hosts and their resident microbes has fostered a mutual dependence\u003csup\u003e1-3\u003c/sup\u003e, in which the physiology in the host affects microbial communities, and vice versa\u003csup\u003e4-7\u003c/sup\u003e. Environmental perturbations, such as diet\u003csup\u003e8\u003c/sup\u003e, antibiotic exposure\u003csup\u003e9\u003c/sup\u003e and cold stress\u003csup\u003e10\u003c/sup\u003e, can elicit changes in this symbiosis and trigger a wide range of physiological changes in the host, including alterations in lipid metabolism, immune function and energy balance. In addition, emerging evidence shows that host genetics may have a critical impact on the microbiome composition\u003csup\u003e11-15\u003c/sup\u003e. For example, genome-wide association studies (GWAS) have linked the \u003cem\u003eABO\u003c/em\u003e blood group and \u003cem\u003eFUT2\u003c/em\u003e polymorphisms with luminal acetylgalactosamine (GalNAc) dynamics, which enrich \u003cem\u003eFaecalibacterium prausnitzii\u003c/em\u003e\u003csup\u003e16-19\u003c/sup\u003e. Similarly, intestinal HIF-2α regulates lactate levels to influence \u003cem\u003eBacteroides vulgatus\u003c/em\u003e and \u003cem\u003eRuminococcus torques\u003c/em\u003e\u003csup\u003e20\u003c/sup\u003e. These studies have emphasised luminal metabolites as central mediators of the crosstalk between host genetic background and gut microbes. However, mechanistic links connecting host genetic background with resulting phenotypes and microbial composition/functions remain poorly defined.\u003c/p\u003e\n\u003cp\u003eWe previously established \u003cem\u003eAida\u003c/em\u003e as an anti-thrifty gene that limits intestinal lipid absorption and enhances energy expenditure to counter fat-induced obesity\u003csup\u003e21,22\u003c/sup\u003e. We hypothesised that fat depletion from the chyme resulting from \u003cem\u003eAida\u003c/em\u003e deletion would create an environment that favours specifically adapted bacterial taxa. Here, through systematic culturing of \u003cem\u003eAida\u003c/em\u003e\u003csup\u003e-/-\u003c/sup\u003e cecal content, we isolated two hitherto unidentified anaerobes, designated lipid-depletion-enriched (LDE)-1 (an \u003cem\u003eOscillospira\u003c/em\u003e strain) and LDE-2 (a tentative new genus within \u003cem\u003eLachnospiraceae\u003c/em\u003e), both exhibiting oleate-sensitive growth suppression. Intriguingly, these strains function as feedback metabolic regulators, as administering either strain attenuated diet-induced obesity, suggesting a host-elicited microbial adaptation to counteract lipid dysregulation in\u003cem\u003e\u0026nbsp;Aida\u003csup\u003e-/-\u003c/sup\u003e\u003c/em\u003emice. Metabolomic analysis revealed that both strains biosynthesise leucic acid (α-hydroxyisocaproic acid or 2-hydroxy-4-methylpentanoic acid) via conserved 2-hydroxyisocaproate dehydrogenases (HicDHs), which are responsible for the anti-obesity effects. Mechanistically, we identified the membrane receptor hydroxycarboxylic acid receptor 2 (HCAR2) as the receptor for leucic acid, which signals to suppress adipose lipid anabolism and promote catabolism. These findings define a feedback host-microbe-metabolite loop, in which host metabolic defects resulting from a genetic alteration reshape the gut microbiome, leading to the production of specific metabolites that rebalance host metabolism.\u0026nbsp;\u003c/p\u003e"},{"header":"Results","content":"\u003cp\u003e\u003cstrong\u003eAIDA depletion enriches specific gut bacterial species linked to adiposity\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eTo determine whether \u003cem\u003eAida\u003c/em\u003e depletion alters gut microbiota composition in the presumably altered chyme, we performed 16S rRNA gene sequencing on cecal contents from high-fat diet (HFD)-fed WT and \u003cem\u003eAida\u003csup\u003e-/-\u003c/sup\u003e\u003c/em\u003e mice (Supplementary Table 1). The overall bacterial composition at the phylum level did not change significantly in \u003cem\u003eAida\u003csup\u003e-/-\u003c/sup\u003e\u003c/em\u003e mice (Extended Data\u0026nbsp;Fig. 1a). We also determined the richness and evenness of the bacterial species and found no significant differences (Extended Data Fig. 1b-f). We then assessed the beta diversity of operational taxonomic units (OTUs) using principal coordinate analysis (PCoA). Again, the \u003cem\u003eAida\u003c/em\u003e\u003cem\u003e\u003csup\u003e-/-\u003c/sup\u003e\u003c/em\u003e samples exhibited a similar bacterial profile to that of the WT group (Fig. 1a). These data indicate that the deletion of \u003cem\u003eAida\u003c/em\u003e does not systematically disturb\u0026nbsp;gut microbiota.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eHowever, we observed that 6 out of the 271 OTUs identified in the microbiota profiling exhibited significant abundance changes in \u003cem\u003eAida\u003c/em\u003e\u003csup\u003e-/-\u003c/sup\u003e mice (Fig. 1b). OTU68 (\u003cem\u003eAnaerotruncus colihominis\u003c/em\u003e), OTU209 (unclassified member of the \u003cem\u003eLachnospiraceae\u003c/em\u003e family), and OTU228 (unclassified member of the genus of \u003cem\u003eClostridium\u003c/em\u003e)were enriched, while OTU21 (\u003cem\u003eHoldemania\u003c/em\u003e\u003cem\u003e\u0026nbsp;massiliensis\u003c/em\u003e), OTU59 (\u003cem\u003eAdlercreutzia\u003c/em\u003e\u003cem\u003e\u0026nbsp;mucosicola\u003c/em\u003e), and OTU141 (unclassified members of the genus of \u003cem\u003eRuminiclostridium\u003c/em\u003e) were decreased (Fig. 1b and Extended Data\u0026nbsp;Fig. 1g). Strikingly, analysis of metagenomic data from 355 normal-weight and 631 obese human fecal samples (Supplementary Table 2) revealed that OTU68 (\u003cem\u003eA. colihominis\u003c/em\u003e) was significantly less abundant in obese compared to normal-weight individuals (Fig. 1c) and inversely correlated with body mass index (BMI) in obesity (Fig. 1c). Species phylogenetically related to other \u003cem\u003eAida\u003c/em\u003e-related OTUs, including \u003cem\u003eHoldemania filiformis\u003c/em\u003e for OTU21 and \u003cem\u003eAdlercreutzia equolifaciens\u003c/em\u003e for OTU59, showed no BMI association despite being mildly reduced in obesity (Extended Data\u0026nbsp;Fig. 1h,i). In mice, OTU209 abundance in cecal contents strongly correlated with fat mass in \u003cem\u003eAida\u003c/em\u003e\u003csup\u003e-/-\u003c/sup\u003e background (Pearson \u003cem\u003er\u003c/em\u003e = 0.9369, \u003cem\u003eP\u003c/em\u003e = .0059) (Fig. 1d; Extended Data\u0026nbsp;Fig. 1j). These data suggest that OTU68 and OTU209 are associated with the development of obesity.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eTo isolate bacterial strains linked to \u003cem\u003eAida\u003c/em\u003e-associated metabolic phenotypes, we cultured cecal content from HFD-fed \u003cem\u003eAida\u003csup\u003e-/-\u003c/sup\u003e\u003c/em\u003emice under stringent anaerobic conditions (0% O\u003csub\u003e2\u003c/sub\u003e), targeting the obligate anaerobes OTU68 and OTU209. Using nine distinct media, we obtained 254 isolates, resolving 30 taxonomically characterised species and more than 20 novel species based on 16S rRNA sequence identity thresholds (Extended Data Fig. 1k; Supplementary Table 3). Among these, strains LDE-1, which represents OTU68, and LDE-2, which represents OTU209, were prioritised for further functional studies. Scaffolded genome sequencing revealed that LDE-1 represented a new strain of \u003cem\u003eA. colihominis\u003c/em\u003e (family \u003cem\u003eOscillospiraceae\u003c/em\u003e; Supplementary note 1), while LDE-2\u0026nbsp;represented a putative new genus within the family \u003cem\u003eLachnospiraceae\u003c/em\u003e (Supplementary note 2). Whole-genome phylogeny corroborated these classifications, confirming the novelty of LDE-2 at the genus level.\u003c/p\u003e\n\u003cp\u003eTo define how \u003cem\u003eAida\u003c/em\u003e depletion reshapes the luminal lipid microenvironment to favour these strains, we quantified cecal fatty acid profiles via GC-MS. This revealed a significant reduction in oleate (C18:1) and palmitate (C18:0) in \u003cem\u003eAida\u003csup\u003e-/-\u003c/sup\u003e\u003c/em\u003e mice compared to WT controls (Fig. 1e). In contrast, laurate (C12:0), myristate (C14:0), stearate (C18:0), arachidate (C20:0), and arachidonate (C20:4) did not show significant change, indicating selective depletion of specific luminal LCFAs. Strikingly, \u003cem\u003ein vitro\u003c/em\u003e growth assays demonstrated that oleate potently inhibited the growth of LDE-1 and LDE-2 at concentrations as low as 10 μM. As a comparison, \u003cem\u003eBacteroides uniformis,\u003c/em\u003e which did not show changes in abundance\u0026nbsp;in the gut, remained resistant to oleate (Fig. 1f). In comparison, palmitate, stearate, and myristate, as well as short-chain (acetate, propionate, and butyrate) and medium-chain fatty acids (caprylate and decanoate), exhibited no suppressive effect (Fig. 1g-j). Critically, \u003cem\u003ein vivo\u003c/em\u003e validation via gavage of oleate-rich olive oil recapitulated these findings, markedly reducing the abundance of OTU68 and OTU209 in\u003cem\u003e\u0026nbsp;Aida\u003csup\u003e-/-\u003c/sup\u003e\u003c/em\u003e mice (Fig. 1k). These data demonstrated that accelerated intestinal lipid absorption in \u003cem\u003eAida\u003c/em\u003e\u003csup\u003e-/-\u003c/sup\u003e mice depletes luminal LCFAs, particularly oleate, thereby alleviating the growth suppression of lipid-sensitive anaerobes and promoting their enrichment.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eLDE-1 and LDE-2 ameliorate diet-induced obesity and metabolic dysfunction in mice\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eWe next investigated whether the obesity-associated enrichment of LDE-1 and LDE-2 reflects a maladaptive bystander effect or a protective feedback response.\u0026nbsp;To this end, we administered these strains to HFD-fed WT mice with prior subjection to antibiotic (ABX)-treatment (Fig. 2a). Strikingly, monocolonization with either\u0026nbsp;strain robustly attenuated body weight gain, with LDE-1- and LDE-2-treated mice exhibiting a 10% reduction relative to controls at 8 weeks post-gavage (Fig. 2b). In contrast, the probiotic \u003cem\u003eEscherichia coli\u003c/em\u003e Nissle 1917 (EcN) showed no significant anti-obesity effect (Extended Data\u0026nbsp;Fig. 2a), underscoring the specificity of LDE-1 and LDE-2. At 5 weeks post-gavage, LDE-1 and LDE-2 administration markedly reduced gonadal white adipose tissue (gWAT) and inguinal WAT (iWAT) mass compared to controls (Fig. 2c,d). Histological analysis confirmed marked reductions in adipocyte hypertrophy in both gWAT and iWAT (Fig. 2e-h and Extended Data\u0026nbsp;Fig. 2b-e). Notably, neither strain induced muscle wasting. LDE-2 even increased soleus muscle mass by 18% (Fig. 2i,j). Treadmill testing revealed a significant increase in running distance and prolongation of time to exhaustion in LDE-2-treated mice (Fig. 2k,l), while both strains improved grip strength (Fig. 2m). Importantly, there were no substantial changes in serum alanine transaminase (ALT) or blood urea nitrogen (BUN) levels (Fig. 2n,o), suggesting these strains did not cause liver or renal damage. In addition, food intake among the groups was similar (Fig. 2p), indicating that the observed anti-obesity effects are not caused by a consequence of bacterial gavage-induced anorexia. These results demonstrated that LDE-1 and LDE-2 uncouple obesity resistance from muscle wasting, a critical limitation of conventional weight-loss therapies.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eWe further explored the long-term effects of LDE-2, which appeared to exert more beneficial effects than LDE-1, in mice maintained on HFD for 16 weeks following the last gavage. Notably, LDE-2 sustained 30% lower body weight gain (Extended Data\u0026nbsp;Fig. 2f) and improved fasting glycemia (Fig. 2q). Hepatic steatosis was attenuated, with 22% reduced liver weight and steatosis scores reduced to 0 (Fig. 2r-t). A mild increase in rectus femoris muscle mass further underscored its musculoskeletal benefits (Extended Data\u0026nbsp;Fig. 2g).\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eLeucic acid\u0026nbsp;\u003c/strong\u003e\u003cstrong\u003eis increased by LDE-1 and\u0026nbsp;\u003c/strong\u003e\u003cstrong\u003eLDE-2 strains\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eTo identify factors mediating the metabolic effects of LDE-1 and LDE-2, we performed untargeted metabolomics on pre- and post-culture media using ultra-high-performance liquid chromatography-quadrupole time-of-flight mass spectrometry (UHPLC-QTOF-MS). Focusing on metabolites commonly upregulated by both strains, we identified six metabolites that were upregulated in both LDE-1 and LDE-2 cultures, including leucic acid, butyric acid, N-methyltryptamine, thymine, 4-hydroxy-3-methoxycinnamaldehyde and Tyr-Pro (Fig. 3a-c; Supplementary\u0026nbsp;Table 4,5). \u0026nbsp;\u003c/p\u003e\n\u003cp\u003eTo validate the \u003cem\u003ein vivo\u003c/em\u003e enrichment of these metabolites, we analysed cecal contents of HFD-fed mice using gas chromatography-mass spectrometry (GC-MS). After testing trimethylsilyl (TMS), \u003cem\u003etert\u003c/em\u003e-butyldimethylsilyl (TBDMS), and sequential (TMS followed by trifluoroacetyl) derivatization approaches, we selected TMS based on peak resolution, sensitivity, and reproducibility for leucic acid, the most markedly changed metabolite, while maintaining compatibility with the detection of a broad spectrum of other metabolites (Extended Data\u0026nbsp;Fig. 3a). Indeed, \u003cem\u003eAida\u003csup\u003e-/-\u003c/sup\u003e\u003c/em\u003e mice exhibited higher leucic acid abundance compared to WT controls (Extended Data\u0026nbsp;Fig. 3b,c), confirming that AIDA depletion drives accumulation of this microbially-derived metabolite. Critically, structurally related hydroxy acids (2-hydroxybutyric acid, 3-hydroxybutyric acid, and 2-hydroxy-3-methylbutyric acid) and leucine, the precursor for leucic acid, were unchanged in \u003cem\u003eAida\u003csup\u003e-/-\u003c/sup\u003e\u003c/em\u003e mice (Extended Data\u0026nbsp;Fig. 3d-g). To assess non-hydroxylated SCFAs, we used derivatisation-independent GC-MS, and found that butyric acid together with propionic acid, pentanoic acid, isobutyric acid, 2-methylbutanoic acid, and isovaleric acid, remained unchanged or even decreased (Extended Data\u0026nbsp;Fig. 3h-m), underscoring the specificity of leucic acid induction among SCFA family. In addition, N-methyltryptamine, thymine, 4-hydroxy-3-methoxycinnamaldehyde, and Tyr-Pro were not detectable using either approach, consistent with their much lower confidence score compared to leucic acid and butyric acid in the untargeted LC-MS data (Supplementary Table 4,5). These data establish that AIDA-deficient mice enrich LDE-1 and LDE-2 that produce leucic acid.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eLeucic acid phenocopies the beneficial effects of LDE-1 and LDE-2 in mice\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eTo determine whether leucic acid mediates the anti-obesity effects of LDE-1 and LDE-2, we supplemented drinking water with 50 mM leucic acid, a dose in line with prior studies of branched-chain amino acid derivatives and short-chain fatty acids\u003csup\u003e23-27\u003c/sup\u003e. Leucic acid robustly attenuated HFD-induced weight gain in both WT and \u003cem\u003eAida\u003csup\u003e-/-\u003c/sup\u003e\u003c/em\u003e mice (Extended Data\u0026nbsp;Fig. 4a). Strikingly, WT mice receiving leucic acid maintained body weights comparable to those of normal chow diet (NCD)-fed controls, achieving a 17% reduction in adiposity by week 12 (Fig. 3d). Consistent with the impact observed in LDE-1- or LDE-2-gavaged mice, gWAT and iWAT mass decreased by 46% and 50%, respectively (Fig. 3e,f; Extended Data\u0026nbsp;Fig. 4b-e). Like administration of LDE-1 or LDE-2, leucic acid treatment did not cause muscle wasting (Extended Data\u0026nbsp;Fig. 4f,g), accompanied by larger myofiber cross-sectional areas in the quadriceps (Extended Data\u0026nbsp;Fig. 4h,i), longer distance to exhaustion on the treadmill, and greater grip strength (Fig. 3g,h). The enhanced grip strength was also observed in \u003cem\u003eAida\u003csup\u003e-/-\u003c/sup\u003e\u003c/em\u003e mice administered leucic acid, although their exercise endurance was not significantly changed (Extended Data\u0026nbsp;Fig. 4j-l). Notably, leucic acid did not elevate levels of ALT or BUN in WT or \u003cem\u003eAida\u003csup\u003e-/-\u003c/sup\u003e\u003c/em\u003e mice, indicating no hepatorenal toxicity (Extended Data\u0026nbsp;Fig. 5a-d). However, unlike LDE-2, leucic acid failed to improve fasting glycemia or reduce hepatic steatosis (Extended Data\u0026nbsp;Fig. 5e,f), suggesting additional mediators in LDE-2 contribute to long-term metabolic benefits.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eTo assess whether leucic acid has an impact on systemic energy metabolism to reduce adiposity and confer other benefits, WT mice were subjected to metabolic cage analysis. The leucic acid-treated mice exhibited a pronounced reduction in respiratory quotient (RQ) (Extended Data\u0026nbsp;Fig. 5g), indicative of preferential lipid oxidation over catabolism of carbohydrate or amino acid. Leucic acid also induced an increase in basal rectal temperature (Fig. 3i), suggesting elevated heat production. These metabolic alterations occurred independently of changes in food intake, locomotor activity (Extended Data\u0026nbsp;Fig. 5h,i), or faecal lipid excretion (Extended Data\u0026nbsp;Fig. 5j,k), excluding malabsorption or anorexia as confounding factors. \u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eBacterial HicDHs drive leucic acid production and are phylogenetically restricted in human gut clades\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eTo identify the enzymes responsible for leucic acid production, we performed Conserved Domain (CD) Searches\u003csup\u003e28\u003c/sup\u003e in strains LDE-1 and LDE-2. For reference, genes in these newly sequenced genomes were assigned arbitrary numeric identifiers during annotation. Both encode a branched-chain-amino-acid aminotransferase (BCAT) homolog for converting L-leucine to 2-ketoisocaproic acid (KIC) (Extended Data\u0026nbsp;Fig. 6a,b). We identified six putative 2-hydroxyacid dehydrogenases (HicDHs) capable of reducing KIC to leucic acid (Supplementary Table 6,7). Phylogenetic analysis separated these into two distinct clusters. The proteins encoded by \u003cem\u003eGM000059\u003c/em\u003e and \u003cem\u003eGM003503\u003c/em\u003e in LDE-1, and \u003cem\u003eGM002896\u003c/em\u003e, \u003cem\u003eGM003422,\u003c/em\u003e and \u003cem\u003eGM003150\u003c/em\u003e in LDE-2, formed a clade with various known D-HicDHs related proteins\u003cem\u003e.\u003c/em\u003e The protein encoded by \u003cem\u003eGM003311\u003c/em\u003e in LDE-1 grouped with L-HicDH family proteins (Extended Data\u0026nbsp;Fig. 6c). AlphaFold-predicted structures confirmed conservation, with low root mean square deviation (RMSD) values (0.928-1.907 Å) to reference enzymes (Extended Data\u0026nbsp;Fig. 7a), and key nicotinamide adenine dinucleotide (NAD\u003csup\u003e+\u003c/sup\u003e) co-substrate binding residues were strictly conserved (Extended Data\u0026nbsp;Fig. 7b-g). Functional validation in \u003cem\u003eE. coli\u003c/em\u003e revealed that only three HicDHs, \u003cem\u003eGM003311\u003c/em\u003e (LDE-1), \u003cem\u003eGM003150\u0026nbsp;\u003c/em\u003e(LDE-2), and \u003cem\u003eGM003422\u0026nbsp;\u003c/em\u003e(LDE-2), produced deuterated leucic acid from L-leucine-5,5,5-D3 (Extended Data\u0026nbsp;Fig. 8a), despite being expressed at similar levels (Extended Data\u0026nbsp;Fig. 8b). Enantiomer-specific triple quadrupole GC-MS analysis confirmed strict stereochemical partitioning. LDE-1 produced \u0026gt;98% L-leucic acid, while LDE-2 produced \u0026gt;98% D-leucic acid (Extended Data\u0026nbsp;Fig. 8c,d), consistent with phylogenetic predictions.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eWe assessed the human relevance of this pathway by probing the Unified Human Gastrointestinal Genome (UHGG) database\u003csup\u003e29\u003c/sup\u003e, and identified 27,237 putative homologs (≥50% similarity,\u0026nbsp;≥80% coverage) across 288 species. A high-confidence subset (344 homologs,\u0026nbsp;≥70% similarity) was enriched within 13 species from the \u003cem\u003eLachnospiraceae\u003c/em\u003e and \u003cem\u003eRuminococcaceae\u003c/em\u003e families, matching the lineages of LDE-2 and LDE-1, respectively (Extended Data Fig. 8e,f and Supplementary Table 8). This indicates leucic acid production is a specialised trait enriched in specific gut clades.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eGenomic analysis confirmed that this reductive pathway dominates in both strains, as we found no homologs for alternative routes such as leucine 2,3-aminomutase (LAM) or the branched-chain 2-keto acid dehydrogenase (BCKDH) complex (Extended Data\u0026nbsp;Fig. 6a). This absence correlates with our observation of unchanged isobutyric acid and isovaleric acid levels (the end products of those pathways) in bacterial culture media (Supplementary Table 4,5) and the reduced levels of these acids in the ceca of \u003cem\u003eAida\u003csup\u003e-/-\u003c/sup\u003e\u003c/em\u003emice (Fig. 3l,n). These findings establish HicDH-mediated reductive catabolism as the dominant leucine metabolic pathway in both strains, explaining their selective leucic acid production and metabolic effects.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eLeucic acid suppresses lipid storage through metabolic rewiring and enantioselective catabolic activation\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eTo elucidate the metabolic mechanism of leucic acid, we performed integrated proteomic and transcriptomic profiling of gWAT from leucic acid-treated WT mice. Multi-omics analysis identified 40 significantly downregulated candidates (|log\u003csub\u003e2\u003c/sub\u003e(fold change)| \u0026gt; 0.5, \u003cem\u003eP\u003c/em\u003e \u0026lt; 0.05) enriched in lipid anabolic pathways, including \u003cem\u003etriglyceride biosynthesis\u0026nbsp;\u003c/em\u003e(Reactome: R-MMU-75109), \u003cem\u003efatty acid biosynthesis\u003c/em\u003e (GO:0006633) and \u003cem\u003elipid storage\u003c/em\u003e (GO:0019915) (Fig. 4a and Supplementary Table 9,10). Critically, leucic acid suppressed key enzymes catalysing triglyceride synthesis (DGAT2, MOGAT2) and fatty acid synthesis (ACSS2, FASN, SCD2 and ELOVL6) (Fig. 4b,c and Extended Data\u0026nbsp;Fig. 9a). These findings were recapitulated in \u003cem\u003eAida\u003csup\u003e-/-\u003c/sup\u003e\u003c/em\u003e gWAT transcriptomes (Extended Data\u0026nbsp;Fig. 9b-d and Supplementary Table 11) and validated by immunoblotting in both genotypes (Fig. 4d,e), confirming AIDA-independent suppression of lipid storage.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eIn iWAT, leucic acid administration upregulated 14 genes in both WT and \u003cem\u003eAida\u003csup\u003e-/-\u003c/sup\u003e\u003c/em\u003e iWAT, enriched for pathways linked to muscle-like metabolic activation, such as \u003cem\u003emuscle contraction\u003c/em\u003e (Reactome: R-MMU-397014, GO:0006936), \u003cem\u003emyofibril\u003c/em\u003e (GO:0030016) and \u003cem\u003esarcomere\u003c/em\u003e (GO:0030017) (Extended Data\u0026nbsp;Fig. 9e and Supplementary Table 12,13), indicative of adipose tissue beiging\u003csup\u003e30\u003c/sup\u003e. Strikingly, Gene Set Enrichment Analysis (GSEA) of \u003cem\u003eAida\u003csup\u003e-/-\u003c/sup\u003e\u003c/em\u003e iWAT demonstrated amplified thermogenic activation, with robust induction of \u003cem\u003ethermogenesis\u003c/em\u003e (NES = 2.7437, adjusted \u003cem\u003eP\u003c/em\u003e \u0026lt; .0001) and \u003cem\u003efatty acid degradation\u003c/em\u003e (NES = 2.3071, adjusted \u003cem\u003eP\u003c/em\u003e \u0026lt; .0001) (Fig. 4f and Supplementary Table 14). These transcriptomic changes were biochemically validated by an increase in UCP1 protein abundance in both genotypes, correlating with the leucic acid-induced elevation of basal core body temperature (Fig. 3i), alongside a coordinated downregulation of lipid anabolism proteins DGAT2 and FASN via immunoblotting (Fig. 4g,h).\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eWe chose 50 μM, which was within the supplemented range\u003cem\u003e\u0026nbsp;in vivo\u003c/em\u003e (Extended Data Fig. 10a-g), to investigate direct cellular mechanisms. At this dose, leucic acid enantiomers significantly reduced lipid accumulation in oleate-challenged MEFs, with both enantiomers decreasing lipid stores by ∼60% in WT and ∼70% in \u003cem\u003eAida\u003csup\u003e-/-\u003c/sup\u003e\u003c/em\u003e cells (Fig. 5a,b). Notably, this potent lipid-lowering effect occurred despite the 1.9-fold higher lipid content in \u003cem\u003eAida\u003csup\u003e-/-\u003c/sup\u003e\u003c/em\u003e MEFs, which is consistent with AIDA being able to reduce lipid absorption\u003csup\u003e21\u003c/sup\u003e and suggests an AIDA-independent activity of leucic acid.\u003c/p\u003e\n\u003cp\u003eWe next investigated the catabolism of fatty acids using [U-\u003csup\u003e13\u003c/sup\u003eC]oleic acid tracing. In beige-like adipocytes, treatment with D-leucic acid significantly enhanced isotopic enrichment in α-ketoglutarate (36%) and citrate (47%), while L-leucic acid showed a weaker increase in α-ketoglutarate labelling (23%) and no citrate effect (Fig. 5c). This enantioselectivity was also found in \u003cem\u003eAida\u003csup\u003e-/-\u003c/sup\u003e\u003c/em\u003e MEFs. Both forms increased labelled α-ketoglutarate (D: 14%; L: 16%) and malate (D: 34%; L: 33%), but only D-leucic acid dramatically elevated proline (53%), a derivative of α-ketoglutarate (Extended Data\u0026nbsp;Fig. 11a-d). Critically, unchanged total metabolite pools (Extended Data\u0026nbsp;Fig. 11e-h) excluded biosynthesis alterations, instead implicating enhanced fatty acid-derived acetyl-CoA flux into the TCA cycle.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eFunctional validation via oxygen consumption assays revealed that D-leucic acid markedly elevated respiration rate in WT and \u003cem\u003eAida\u003csup\u003e-/-\u003c/sup\u003e\u003c/em\u003e MEFs, while L-leucic acid induced subtler increases (Fig. 5d). Collectively, leucic acid exerts AIDA-independent suppression of lipid storage across adipose depots and genotypes, while exhibiting enantiomer-selective amplification of lipid catabolism in thermogenic adipocytes and \u003cem\u003eAida\u003csup\u003e-/-\u003c/sup\u003e\u003c/em\u003e cells, with a mechanistic divergence between gWAT and iWAT at physiologically relevant concentrations (Fig. 5e).\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eHCAR2 mediates leucic acid-induced suppression of lipid accumulation\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eTo identify the molecular target through which leucic acid regulates lipid metabolism, we performed integrative target prediction for L- and D-enantiomers using complementary computational approaches, TargetNet employing deep learning on 3D features\u003csup\u003e31\u003c/sup\u003e and SwissTargetPrediction utilising similarity-based machine learning\u003csup\u003e32\u003c/sup\u003e (Supplementary Table 15). Intersection analysis revealed 13 consensus targets (Fig. 6a), mitigating the single-platform biases common in endogenous metabolite studies, with both enantiomers exhibiting full compliance with Lipinski's Rule of Five\u003csup\u003e33\u003c/sup\u003e (Extended Data\u0026nbsp;Fig. 12a), indicating favourable drug-likeness.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eTo prioritise these candidates based on human genetic evidence, we screened the GWAS Catalog\u003csup\u003e34\u003c/sup\u003e, which uncovered 44 obesity-associated intergenic or regulatory region SNPs near \u003cem\u003eHCAR2\u003c/em\u003e (\u003cem\u003eHydroxycarboxylic Acid Receptor 2\u003c/em\u003e). In contrast, only one SNP was found near \u003cem\u003eGRM2\u003c/em\u003e (\u003cem\u003eGlutamate Metabotropic Receptor 2\u003c/em\u003e), and no significant associations were observed for other candidates (Fig. 6b and Supplementary Table 16). Transcriptomic analysis of adipose tissue from leucic acid-treated mice identified \u003cem\u003eHcar2\u003c/em\u003e and \u003cem\u003eCar4\u003c/em\u003e (\u003cem\u003eCarbonic Anhydrase 4\u003c/em\u003e) as the sole genes that were significantly differentially expressed across both absolute abundance change and fold-change metrics (Fig. 6c). This convergence of genetic association and transcriptional responsiveness prioritised HCAR2 for mechanistic validation (Fig. 6d).\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eMolecular docking suggested potential binding of HCAR2 for both D-leucic acid (ΔG ≈ -5.2 kcal/mol) and L-leucic acid (ΔG ≈ -5.3 kcal/mol) (Fig. 6e), with binding affinities comparable to established ligands niacin (ΔG ≈ -5.9 kcal/mol) and β-hydroxybutyrate (ΔG ≈ -4.8 kcal/mol) calculated under identical parameters (Extended Data\u0026nbsp;Fig. 12b). We experimentally validated receptor engagement using \u003csup\u003e19\u003c/sup\u003eF-NMR ligand-detected binding assay\u003csup\u003e35,36\u003c/sup\u003e. Incubation of 4-fluoro-leucic acid with HCAR2-expressing cells induced a pronounced chemical shift (Δδ = 0.08 ppm), indicative of ligand-receptor engagement (Fig. 6f). By contrast, no shift occurred in HCAR1-expressing cells, confirming the specificity of HCAR2.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eThe requirement of HCAR2 to mediate the role of leucic acid was further investigated through loss-of-function experiments. Leucic acid enantiomers reduced lipid accumulation in beige-like adipocytes and oleate-loaded MEFs expressing control shRNA, but this effect was abolished in \u003cem\u003eHcar2\u003c/em\u003e-knockdown cells (Fig. 6g,h and Extended Data Fig. 12c-f). Similarly, \u003cem\u003eHcar2\u003c/em\u003e knockdown eliminated leucic acid-stimulated oxygen consumption (Fig. 6i). For \u003cem\u003ein vivo\u003c/em\u003e validation, we knocked down \u003cem\u003eHcar2\u003c/em\u003e via AAV in one side of the\u0026nbsp;iWAT, using the contralateral side injected with control shRNA as an internal control (Fig. 6j and Extended Data\u0026nbsp;Fig. 12g). While \u003cem\u003eHcar2\u003c/em\u003e knockdown alone did not alter baseline iWAT mass, leucic acid administration induced significant mass reduction and pronounced browning in control shRNA-treated iWAT. Crucially, the same treatment resulted in significantly attenuated fat mass reduction and diminished browning in contralateral \u003cem\u003eHcar2\u003c/em\u003e-knockdown tissue relative to control tissues within the same animals (Fig. 6k,l). Collectively, these findings establish HCAR2 as the cognate receptor for leucic acid in mediating the suppression of lipid storage (Extended Data Fig. 12h).\u0026nbsp;\u003c/p\u003e"},{"header":"Discussion","content":"\u003cp\u003eOur study reveals that the textural changes of chyme, caused by depletion of luminal fat as a result of AIDA deficiency, can critically influence microbial growth and metabolite production, which in turn plays a significant role in regulating host physiology. Specifically, the absence of AIDA in mice resulted in the selective enrichment of oleate-suppressed \u003cem\u003eClostridia\u003c/em\u003e strains LDE-1 and LDE-2. Paradoxically, these strains exert anti-obesity effects, as we found that gavage with LDE-1 or LDE-2 ameliorates obesity in mice fed a high-fat diet, forming a host-directed adaptive loop where microbial expansion counteracts\u0026nbsp;host metabolic dysfunction. Mechanistically, we identified leucic acid as a key bioactive metabolite produced by HicDHs in these strains. Leucic acid counters obesity and enhances exercise performance. Moreover, we show that leucic acid suppresses lipid storage via divergent regulation of lipid anabolism and catabolism across adipose depots mediated by HCAR2.\u003c/p\u003e\n\u003cp\u003eThe selective enrichment of these beneficial strains is likely driven by a specific ecological niche, as varied abilities of the host to absorb different nutrients can alter the contents of residual nutrients in the chyme. Our analyses \u003cem\u003ein vitro\u003c/em\u003e on the two strains LDE-1 and LDE-2 showed high sensitivity of the LDE strains to oleate, consistent with their enrichment in the fat-depleted gut of \u003cem\u003eAida\u003csup\u003e-/-\u003c/sup\u003e\u003c/em\u003e mice.\u0026nbsp;Thus, AIDA deficiency, by promoting lipid absorption through ERAD-mediated degradation of re-esterification enzymes (GPAT3, MOGAT2 and DGAT2)\u003csup\u003e21\u003c/sup\u003e, creates a lipid-depleted luminal environment that favours the growth of oleate-sensitive bacteria. This model is further supported by the selective reduction of oleate and palmitate in the \u003cem\u003eAida\u003c/em\u003e\u003csup\u003e-/-\u003c/sup\u003e cecum, likely reflecting the preference of intestinal triglyceride synthesis enzymes for these acyl-CoAs\u003csup\u003e37,38\u003c/sup\u003e. Although dietary fatty acids generally exhibit antimicrobial properties, the particular sensitivity of LDE-1 and LDE-2 to oleate, a trait shared by bacteria like \u003cem\u003eLactobacillus iners\u003c/em\u003e\u003csup\u003e39\u003c/sup\u003e, coupled with their resistance to SCFAs or MCFAs that bypass intestinal re-esterification, defines a unique ecological niche.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eLeucic acid has been previously identified as a microbial metabolite that shows beneficial effects on metabolic dysfunction-associated steatohepatitis (MASH)\u003csup\u003e40\u003c/sup\u003e. We have found that leucic acid production by LDE-1 and LDE-2 represents a previously unrecognised mode of host-microbe crosstalk emerging from the changed luminal environment resulting from the genetic mutation of the host. The lipid-depleted conditions in \u003cem\u003eAida\u003c/em\u003e\u003csup\u003e-/-\u003c/sup\u003e gut create an ecological niche analogous to time-restricted feeding\u003csup\u003e40\u003c/sup\u003e. Our data show that leucic acid exerts striking metabolic benefits across both WT and \u003cem\u003eAida\u003csup\u003e-/-\u003c/sup\u003e\u003c/em\u003e genetic backgrounds, including reduction of adiposity and enhancement of exercise capacity. These effects stem from the combined roles of suppression of lipid synthesis in gWAT and promotion of lipid catabolism and thermogenesis in iWAT.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eThe physiochemical properties of leucic acid salts facilitate HCAR2 engagement through efficient aqueous diffusion enabled by their hydrophilicity (TPSA 57-60 Å\u003csup\u003e2\u003c/sup\u003e, log\u003cem\u003eP\u003c/em\u003e -0.8567 to 0.478). This mechanism gains human relevance from GWAS linking intergenic variants near the \u003cem\u003eHCAR2\u003c/em\u003e locus to obesity traits. Although physiological levels of leucic acid are modest in humans, these genetic associations likely represent conserved signalling through endogenous HCAR2 ligands such as β-hydroxybutyrate\u003csup\u003e41\u003c/sup\u003e and niacin\u003csup\u003e42\u003c/sup\u003e, positioning this pathway as an evolutionarily conserved regulator of energy homeostasis. Taken together, we have presented an exemplary mode of host-microbe relationship, in which the host physiological state can influence the community of gut microbes that exert feedback controls on the host. \u0026nbsp;\u0026nbsp;\u003c/p\u003e"},{"header":"Methods","content":"\u003cp\u003e\u003cstrong\u003eAnimals and ethics\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eAll experiments utilised male C57BL/6J mice. Homozygous \u003cem\u003eAida\u003csup\u003e-/-\u003c/sup\u003e\u003c/em\u003e mice and wild-type (WT) littermate controls were generated by crossing heterozygous \u003cem\u003eAida\u003csup\u003e-/-\u003c/sup\u003e\u003c/em\u003e parents as described previously\u003csup\u003e21\u003c/sup\u003e. WT mice administered LDE-1, LDE-2, or leucic acid were sourced from the Specific Pathogen-Free (SPF) facility at the Laboratory Animal Center of Xiamen University (Xiamen, China). For metabolic studies, 8-week-old male mice were fed a high-fat diet (HFD; D12492, Research Diets, New Brunswick, NJ, USA; 60% kcal from fat, primarily lard and soybean oil) ad libitum and housed under controlled conditions (temperature: 22-24 ℃; humidity: 55-60%; light/dark cycle: 12 h, lights on at 8:00 AM; enrichment: nesting materials replaced weekly). All animal procedures were approved by the Xiamen University institutional animal care and use committee (Approval No. XMULAC20220188).\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eCell culture\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eBeige-like adipocytes were derived from 3T3-L1 preadipocytes (ATCC CL-173) via a beige protocol. Cells were maintained in an induction medium (DMEM/F12 1:1 medium with 10% FBS containing 5 μg/mL insulin (BS901, Biosharp, Hefei, China), 1 nM 3,3’,5-triiodo-l-thyronine (T3) (T-2877, Sigma-Aldrich, Taufkirchen, Germany), 125 μM indomethacin (S1723, Sigma-Aldrich), 0.5 mM isobutylmethylxanthine (IBMX) (I5879, Sigma-Aldrich) and 1 μM dexamethasone (DEX) (D4902, Sigma-Aldrich)) for 2 days. Cells were switched to maintenance medium (DMEM/F12 1:1 medium with 10% FBS supplemented with 5 μg/mL insulin and 1 nM T3) for 6 days, with medium refreshed every 48 h. Differentiated beige-like adipocytes were confirmed by multilocular lipid droplet formation. MEFs were maintained in DMEM with 10% FBS.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAnalysis of human faecal metagenomic data\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003ePublicly available metagenomic datasets (accession codes: CNP0003084, PRJEB12123)\u003csup\u003e43\u003c/sup\u003e were analysed, comprising 355 normal-weight (BMI 18.5-24.9) and 631 obese (BMI \u0026gt; 30) individuals. Taxon abundances were expressed as relative abundances, calculated by dividing the read count of each taxon by the total read count across all taxa. Zero values were substituted with the detection limit (1 × 10\u003csup\u003e-6\u003c/sup\u003e), and log\u003csub\u003e10\u003c/sub\u003e transformation was applied. Covariates (sampling batch, sex, age) were assessed but showed no significant association with taxon abundances, thus, no adjustments were made.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eGut microbiota analysis by 16S rRNA sequencing\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eCecal content samples were collected from 6 homozygous \u003cem\u003eAida\u003csup\u003e-/-\u003c/sup\u003e\u003c/em\u003e mice and 6 WT littermates after 4 months of HFD feeding. Genomic DNA was extracted using the MagBeads Fecal DNA Kit (116570400, MP Biomedicals, Santa Ana, CA, USA). The V3-V4 hypervariable regions of bacterial 16S rRNA genes were amplified and analysed using an Illumina HiSeq2500 platform and sequenced on an Illumina HiSeq2500 platform (Majorbio Bio-Pharm Technology Co. Ltd., Shanghai, China). Sequences with ≥ 97% similarity were assigned to the same operational taxonomic units (OTUs). Microbial alpha diversity indexes were obtained from the free online platform of the Majorbio Cloud Platform (www.majorbio.com). The 16S rRNA sequencing data have been deposited in the National Center for Biotechnology Information (NCBI) GenBank database under BioProject accession number PRJNA1311729.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eQuantitative PCR (qPCR)\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eDNA from cecal content was extracted as above. Specific primers were designed for the amplification of fragments within 16S rRNA genes of OTU68 and OTU209, as well as universal primers targeting the 16S rRNA gene for normalization purposes\u003csup\u003e44\u003c/sup\u003e. The specificity of these primers was validated through in silico analysis and confirmed experimentally using known positive controls. OTU68: 187 bp product (F: 5′-GCGTCCGATTAGCCAGTTGG-3′; R: 5′-GCCTGCCTCTACTTCACTCAAG-3′). OTU209: 227 bp product (F: 5′-GCGTGGGTGAAGAAGCGCC-3′; R: 5′-TCTTGGAACTGCGTGGCTG-3′). Universal 16S: 177 bp product (F: 5′-ACTCCTACGGGAGGCAGCAGT-3′; R: 5′-ATTACCGCGGCTGCTGGC-3′). The qPCR was performed in triplicate using ChamQ SYBR qPCR Master Mix (Q312, Vazyme Biotech, Nanjing, China). Primer specificity was confirmed by Sanger sequencing of amplicons. Relative abundances were calculated using the 2\u003csup\u003e-ΔΔCt\u003c/sup\u003e method normalised to universal 16S rRNA.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eIsolation of AIDA depletion-related bacterial strains\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e\u003cem\u003eCulture media preparation\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003eNine agar plate formulations were used for anaerobic bacterial isolation: modified Gifu anaerobic medium (mGAM) and yeast casitone fatty acids broth (YCFA), with or without 1 g/L taurocholic acid sodium salt, respectively. mGAM was made by supplementing GAM with 10% bovine rumen fluid. Modified chopped meat medium (CMM), TSA medium (with 5% defibrinated sheep blood), rumen bacteria medium (TPY), modified reinforced clostridial medium (RCM) and modified PYG medium were also used. Media were autoclaved (121 ℃, 20 min), and 5% sheep blood was added post-sterilisation. Plates were pre-reduced in an anaerobic workstation (D500G, Gene Science, Wilmington, DE, USA; atmosphere: 85% N\u003csub\u003e2\u003c/sub\u003e, 5% CO\u003csub\u003e2\u003c/sub\u003e, 10% H\u003csub\u003e2\u003c/sub\u003e) for at least 24 h before use. Strict anaerobic conditions of liquid media were generated by using an intelligent anaerobic culture preparation system (DLQ120-B, Beijing Ai Si Pu Technology Co., Ltd., Beijing, China).\u003c/p\u003e\n\u003cp\u003e\u003cem\u003eCecal sample processing and anaerobic cultivation\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003eCecal contents from 6-month HFD-fed\u003cem\u003e\u0026nbsp;Aida\u003csup\u003e-/-\u003c/sup\u003e\u003c/em\u003e mice were divided into two aliquots. \u0026nbsp;Ethanol-treated fraction: Resuspended in 1 mL sterile PBS, incubated in 70% ethanol (4 h, 25 ℃), and filtered (0.22 µm) to remove host cells and debris. Sequential 10-fold dilutions (10\u003csup\u003e-3\u003c/sup\u003e to 10\u003csup\u003e-6\u003c/sup\u003e) were plated (100 µL/plate). Untreated fraction: Directly diluted in PBS and plated. The plates were incubated at 37 ℃ in the anaerobic chamber (D500G, GeneScience). Plates were incubated at 37 ℃ under anaerobic conditions for up to 30 days. Colonies were monitored at 2, 4, 6, 8, 10, 15, 20, and 30 days, with six morphologically distinct colonies per plate isolated and subcultured for 2-7 days. Colony morphology and spatial distribution were documented to avoid redundancy.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cem\u003eBacterial identification via 16S rRNA gene sequencing\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003eSingle colonies were lysed in 3 μL of NaOH/SDS buffer (0.8% NaOH, 1% SDS), diluted in 67 µL ddH\u003csub\u003e2\u003c/sub\u003eO, boiled (95 ℃, 30 min), and centrifuged (5000 \u003cem\u003eg\u003c/em\u003e, 5 min). Supernatants were used as templates for PCR amplification of near-full-length 16S rRNA genes using universal primers\u003csup\u003e45\u003c/sup\u003e: 27F: 5'-AGAGTTTGATCCTGGCTCAG-3' and 1492R: 5'-GGTTACCTTGTTACGACTT-3'. PCR products (~1,500 bp) were Sanger-sequenced (Sangon Biotech Co., Ltd., Shanghai, China). Sequences were analysed using NCBI BLAST against a custom database of \u003cem\u003eAida\u003c/em\u003e-related OTUs, the Nucleotide Collection (nr/nt) and the 16S rRNA database (Bacteria and Archaea). Taxonomic assignments followed established thresholds with ≥ 98.7% identity to reference strains as the same species.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eLDE-1 and LDE-2 were deposited under the Budapest Treaty at the China General Microbiological Culture Center (CGMCC), Beijing, China. The assigned deposit numbers are CGMCC No.46618 for LDE-1 and CGMCC No.46619 for LDE-2. These strains are available for research purposes from CGMCC under the terms of the Material Transfer Agreement.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eNegative stain electron microscopy\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eBacterial isolates were resuspended in 20 μL of 0.1 M phosphate buffer. A portion of this suspension was carefully applied onto a carbon/formvar-coated grid and allowed to adhere for 15 s, blotted off and stained with 1% uranyl acetate for 15 s. Grids were air-dried for 3 h at 25 ℃ and imaged on a Hitachi HT-7800 transmission electron microscope (HT-7800, Hitachi, Tokyo, Japan). Images were captured at magnifications ranging from 4,000 × to 10,000 × at 80 kV.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eBacterial genome sequencing and assembly\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e\u003cem\u003eDNA extraction and library preparation\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003eGenomic DNA was extracted from 1 g of pure LDE-1 or LDE-2 cultures and subjected to bacteria draftmap analysis by Novogene Co., Ltd. (Beijing, China). DNA was randomly sheared into short fragments, end-repaired, A-tailed, and ligated with an Illumina adapter. The DNA fragments with adapters were size-selected, PCR amplified and purified. The library was checked with Qubit and real-time PCR for quantification, and a bioanalyzer for size distribution detection.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cem\u003eSequencing and quality control\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003ePaired-end sequencing (2 × 150 bp) was performed on an Illumina NovaSeq 6000 platform (Novogene Co., Ltd., Beijing, China). The reads with \u0026gt; 40% low-quality bases (mass value ≤ 20), with \u0026gt; 10% unrecognised nucleotides (N), or adapter contamination (15 bp overlap, \u0026lt; 3 mismatches).\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cem\u003eGenome Assembly and Annotation\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003eThe clean data were assembled with SOAPdenovo, SPAdes, and ABySS software. Contigs from all assemblies were integrated with CISA software and polished with GapCloser software to fill gaps. Final scaffolds (\u0026gt; 500 bp) were annotated using GeneMarkS. Functional annotation involved performing a whole-genome Blast search against GO and KEGG databases. Genome assemblies were deposited in NCBI GenBank under BioProject accession PRJNA1311729.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eStrain classification and phylogenetic analysis\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eTaxonomic assignment of the strains was performed using a dual approach: 16s rRNA sequence analysis and the Genome Taxonomy Database Toolkit (GTDB-Tk). Full-length 16S rRNA sequences, extracted from whole-genome assemblies, were first queried against the GenBank/ENA/DDBJ and EzBioCloud databases (https://ezbiocloud.net/) to identify closest relatives. Sequence alignment was executed using ClustalX (v. 1.838), and phylogenetic trees were constructed using the neighbour-joining algorithm in MEGA11\u003csup\u003e46\u003c/sup\u003e with bootstrap support based on 1,000 replicates. For higher-resolution classification, draft genome assemblies were analysed with GTDB-Tk (v. 1.7.0) via the KBase platform (https://narrative.kbase.us/). This workflow leverages 120 conserved bacterial marker genes from the Genome Taxonomy Database (GTDB, v. R06-RS202), which integrates curated RefSeq and Genbank reference genomes. Strains were taxonomically assigned by placement into a domain-specific reference phylogeny, refined using relative evolutionary divergence (RED) thresholds and pairwise average nucleotide identity (ANI) values relative to GTDB reference genomes\u003csup\u003e47,48\u003c/sup\u003e. All GTDB-Tk analyses were finalised on 2024-05-08.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eBacterial growth inhibition assays by fatty acids\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e\u003cem\u003eFatty acid solution preparation\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003eSodium salt solutions were prepared according to chain length. Acetate (S818277, Macklin Biochemical Technology, Shanghai, China), propionate (S817368, Macklin Biochemical Technology), and butyrate (S817488, Macklin Biochemical Technology) were dissolved in ddH\u003csub\u003e2\u003c/sub\u003eO to prepare 100 mM stocks, 0.22 µm-filtered, and assayed at 5 mM. Caprylate (S6096, Macklin Biochemical Technology), decanoate (629-25-4, MedChemExpress, Shanghai, China), and myristate (822-12-8, MedChemExpress) were dissolved in 2% DMSO in ddH\u003csub\u003e2\u003c/sub\u003eO to prepare 10 mM stocks, sonicated or vortexed, 0.22 µm nylon-filtered (84311ES03, Yeasen Biotechnology, Shanghai, China), and tested at 100 µM. Palmitate (S161420, Aladdin Scientific, Shanghai, China), stearate (822-16-2, Aladdin Scientific), and oleate (S817542, Macklin Biochemical Technology) were added into mGAM media, autoclaved to dissolve, tested at 10 µM or 100 µM. All solutions were validated precipitate-free. NaCl controls (matched to experimental concentrations) were prepared equivalently.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cem\u003eBacterial growth analysis\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003eLDE-1, LDE-2 and \u003cem\u003eBacteroides uniformis\u003c/em\u003e were inoculated in 96-well plates (1 mL per well) or jars (40 mL per jar) containing fatty acid working solutions or controls. LDE-1 was inoculated at a density of OD\u003csub\u003e600\u003c/sub\u003e = 0.6 and LDE-2 at a density of OD\u003csub\u003e600\u003c/sub\u003e = 0.15 to mGAM media. Cultures were normalised to an identical starting OD\u003csub\u003e600\u003c/sub\u003e. Media pH was verified to be pH 7.0 ± 0.2 pre-inoculation. Plates or jars were maintained in an anaerobic chamber (D500G, GeneScience) throughout incubation. Growth was monitored at 24 h intervals by measuring OD\u003csub\u003e600\u003c/sub\u003e. Background absorbance was corrected by subtracting cell-free supernatant OD\u003csub\u003e600\u003c/sub\u003e from total readings. Post-assay culture purity was confirmed via 16S rRNA sequencing.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eBacterial preparation and oral administration\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eTwo-month-old male mice were randomised into experimental groups based on body weight and housed under SPF conditions. Mice were fed HFD for 3-5 days, followed by broad-spectrum antibiotic (ABX) treatment (1 g/L ampicillin, 0.5 g/L vancomycin, 1 g/L neomycin, 1 g/L metronidazole in drinking water) for 5-7 days to deplete gut microbiota.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eThe LDE-1 culture was inoculated into fresh mGAM and cultured anaerobically (85% N\u003csub\u003e2\u003c/sub\u003e, 5% CO\u003csub\u003e2\u003c/sub\u003e, 10% H\u003csub\u003e2\u003c/sub\u003e) at 37 ℃ for 48 h. After incubation, the culture was centrifuged (10,000 \u003cem\u003eg\u003c/em\u003e, 8 min). The supernatant was discarded, and the bacterial pellet was washed with sterile PBS. The bacteria were then resuspended in fresh mGAM to achieve OD\u003csub\u003e600\u003c/sub\u003e = 1.3. The control group received the same volume of fresh mGAM administered by oral gavage, once every three days for ten administrations. The preparation of LDE-2 was similar to that of LDE-1, with the following modifications. The culture was inoculated at a 1:25 dilution, incubated for 60 h, centrifuged at 6,000 \u003cem\u003e×\u003c/em\u003e \u003cem\u003eg\u003c/em\u003e for 5 min, and resuspended to achieve OD\u003csub\u003e600\u003c/sub\u003e = 0.6. The control group received the same volume of sterile PBS administered by oral gavage, once daily for 14 days. The \u003cem\u003eEscherichia coli\u003c/em\u003e Nissle 1917 (EcN) culture was inoculated at a 1:50 dilution into fresh LB liquid medium and incubated in a shaker at 37 ℃ for 18-22 h. The culture was then centrifuged at 6,000 \u003cem\u003e×\u0026nbsp;g\u003c/em\u003e for 5 min. The supernatant was discarded, and the bacterial pellet was washed with sterile PBS. The bacteria were resuspended in sterile PBS to achieve an OD\u003csub\u003e600\u003c/sub\u003e = 1.3. The control group received the same volume of sterile PBS administered by oral gavage, once daily for 10 days. After the last gavage, the mice that received LDE-1 were kept for an additional 10 days, while those that received LDE-2 were observed for 16 weeks before sacrifice to minimise any acute effects of gavage. The mice were fasted for 6 h (8:30 a.m. to 2:30 p.m.) before sacrifice.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAdministration of leucic acid in mice\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eTwo-month-old male mice were divided into two comparable groups. HFD feeding started at least 2 days before the leucic acid administration. DL-leucic acid (S161215, Aladdin Scientific, Shanghai, China) was solubilised in the drinking water to make a final concentration of 50 mM leucic acid (pH = 2.5). The drinking water was acidified using HCl (ADW) to the same pH as the control mice. The drinking water was replaced with distilled water containing 50 mM leucic acid or ADW every 5 days. The mice were fasted for 6 h (8:30 a.m. to 2:30 p.m.) before sacrifice.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eBiochemical analysis\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eBlood and faecal parameters were measured in mice after a 6-h fast (8:30 a.m. to 2:30 p.m.). Blood glucose was quantified by a glucose meter (OneTouch UltraVue, Johnson \u0026amp; Johnson, New Brunswick, NJ, USA). Commercial kits were utilised to determine the levels of triglyceride (TAG, A110-1-1, Nanjing Jiancheng Bioengineering Institute, Nanjing, China), non-esterified fatty acids (NEFA, A042-1-1, Nanjing Jiancheng), alanine aminotransferase (ALT, C009-2-1, Nanjing Jiancheng) and blood urea nitrogen (BUN, C013-2-1, Nanjing Jiancheng) in accordance with the manufacturer’s protocols. \u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eMeasurement of basal core body temperature\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eFollowing one week of leucic acid or ADW administration to mice acclimated to HFD for 5 days, basal core body temperature was measured. Using a calibrated rectal probe, temperatures were recorded between 14:30 and 15:30 after a 4-h fast. Two independent measurements were taken per mouse at 48 h intervals, ensuring consistent probe insertion depth of 1.5 cm and a dwell time of 30 seconds.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eMass-spectrometry-based metabolomics for bacterial secreted metabolites\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e\u003cem\u003eSample preparation\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003eUntargeted metabolomics was performed on culture media from LDE-1 and LDE-2 strains (Metware Biotechnology Inc., Wuhan, China). Strains were grown in modified GAM medium to the exponential phase. Pre- and post-culture media were collected. A 150 μL extraction solution (acetonitrile: methanol = 1:4, v:v) containing 1 μg/mL of 2-chlorophenylalanine (internal standard) was added into a 50 μL sample. Then the sample was vortexed (3 min), centrifuged (12,000 \u003cem\u003eg\u003c/em\u003e, 10 min, 4 ℃). Supernatant was collected and incubated at -20 ℃ (30 min), recentrifuged (12,000 \u003cem\u003eg\u003c/em\u003e, 3 min, 4 ℃). 120 μL supernatant was aliquoted for LC-MS.\u003c/p\u003e\n\u003cp\u003e\u003cem\u003eLC-MS analysis\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003eLC-MS analysis was performed using UHPLC (Shimadzu Nexera LC-30A, Kyoto, Japan) and MS (SCIEX TripleTOF 6600+, Foster City, CA, USA) in positive/negative ionisation modes. Waters ACQUITY Premier HSS T3 Column (1.8 µm, 2.1 mm × 100 mm) using 0.1% formic acid in water as solvent A and 0.1% formic acid in acetonitrile as solvent B in the following gradient: 5 to 20% in 2 min, increased to 60% in the following 3 min, increased to 99% in 1 min and held for 1.5 min, then come back to 5% mobile phase B within 0.1 min, held for 2.4 min. The analytical conditions were as follows: column temperature, 40 ℃; flow rate, 0.4 mL/min; injection volume, 4 μL. Mass spectrometry data acquisition was operated using information-dependent acquisition (IDA) mode using Analyst TF software v. 1.7.1 (SCIEX, Concord, Canada). The source parameters (ion source Gas 1/Gas 2: 50 psi; curtain gas: 25 psi; temperature: 550 ℃; ion spray voltage floating: 5000 V or -4000 V; decluttering potential: ±60 V). TOF-MS scan parameters (mass range: 50-1000 Da; accumulation time: 200 ms; and dynamic background subtract: on). Product ion scan parameters (mass range: 25-1000 Da; accumulation time: 40 ms; collision energy: ±30; collision energy spread: 15; resolution: UNIT; charge state: 1 to 1; intensity: 100 cps; exclude isotopes within 4 Da; mass tolerance: 50 ppm; maximum number of candidate ions to monitor per cycle: 18).\u003c/p\u003e\n\u003cp\u003e\u003cem\u003eData Processing\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003eRaw data were converted to mzXML format by ProteoWizard software. Peak extraction, alignment, and retention time correction were performed in XCMS program. Peak areas normalised using support vector regression (SVR). Low-abundance peaks (\u0026lt; 50% detection rate per group) were excluded. Metabolic identification information was obtained by searching in-house, public, and AI-predicted databases and MetDNA web server. The metabolomics data have been deposited to MetaboLights\u003csup\u003e49\u003c/sup\u003e repository with the study identifier MTBLS12947.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eDetermination of leucic acid concentration by GC-MS\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe concentrations of leucic acid in mouse faeces, serum, and tissues were measured using GC-MS. To optimise this determination, different derivatisation approaches and different capillary columns were compared.\u003c/p\u003e\n\u003cp\u003eSample preparation involved homogenising approximately 50 mg of tissue or cecal content in a 20-fold volume of 80% (v:v) methanol aqueous solution containing 10 μg/mL 2-hydroxyoctanoic acid as an internal standard. The homogenate was centrifuged at 18,000 × \u003cem\u003eg\u003c/em\u003e and 4 °C for 15 min. A 200 μL aliquot of the supernatant was dried using a vacuum centrifugal concentrator. The residue was reconstituted in 50 μL pyridine (270407, Merck, Darmstadt, Germany) via sonication. Derivatisation was then performed using one of three methods. The first method involved derivatisation with 50 μL N-methyl-N-(trimethylsilyl)trifluoroacetamide (MSTFA, M85387, Meryer, Shanghai, China) at 37 °C for 1 h to form trimethylsilyl (TMS) derivatives. The second method involved sequential derivatisation, first with 50 μL MSTFA at 37 °C for 1 h, followed by 50 μL N-Methyl-bis(trifluoroacetamide) (MBTFA, N814244, Macklin, Shanghai, China) at 37 °C for 1 h to form TMS and trifluoroacetyl (TFA) derivatives. The third method used 50 μL N-\u003cem\u003etert\u003c/em\u003e-butyldimethylsilyl-N-methyltrifluoroacetamide (MTBSTFA, M-108, Merck, Darmstadt, Germany) at 55 °C for 1 h to form \u003cem\u003etert\u003c/em\u003e-butyldimethylsilyl (TBDMS) derivatives. All derivatisation reactions occurred in a heating block. Following derivatisation, samples were centrifuged at 18,000 × \u003cem\u003eg\u003c/em\u003e for 5 min, and the supernatant was transferred to GC vials for analysis.\u003c/p\u003e\n\u003cp\u003eFor the comparison of low-polarity (HP-5MS) and mid-polarity (DB-1701) capillary columns, TMS-derivatised samples prepared as described were analysed. Both GC-MS methods employed a 1 mL/min helium carrier gas flow rate, a 1 µL injection volume, a 10:1 split ratio, and a 250 °C injector temperature. However, the temperature programs were adjusted to account for the different column characteristics and ensure appropriate elution of the leucic acid di-TMS derivative. The HP-5MS method used an initial hold at 70\u0026nbsp;°C for 2 min, ramped at 7 °C per min to 140 °C, then ramped at 30 °C per min to 310 °C with a final hold for 9 min. The DB-1701 method used an initial hold at 80 °C for 2 min, ramped at 5 °C per min to 140 °C, then ramped at 10 °C per min to 200 °C, followed by a ramp at 30 °C per min to 290 °C with a final hold for 9 min. Detection was performed in combined scan/selected ion monitoring (SIM) mode. The SIM ions were set at m/z 103, 159, 233, and 261, with a solvent delay of 7 min.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eIndirect calorimetry\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eMice were acclimated to individual housing in metabolic chambers (CAB-16-1-EU, Sable Systems International, North Las Vegas, NV, USA) for 3 days to minimise stress from environmental changes. Following acclimation, oxygen consumption (VO\u003csub\u003e2\u003c/sub\u003e), carbon dioxide production (VCO\u003csub\u003e2\u003c/sub\u003e), and respiratory exchange ratio (RER) were measured over 48 h at 5-min intervals using high-resolution sensors. Food intake was estimated daily per cage and normalised to individual mouse intake by dividing total cage consumption by the number of mice.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eBody composition analysis\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eFat and lean mass were quantified in conscious mice using a 3-in-1 EchoMRI Composition Analyzer (100H, Echo Medical Systems, Houston, TX, USA). Triplicate measurements per mouse were performed and averaged to ensure precision.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eMuscle performance assessments\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e\u003cem\u003eGrip strength test\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003eForelimb grip strength was measured using a digital force meter (47200, Ugo Basile, Gemonio, Italy). The maximum grip strength was measured for each mouse, with claws of forelimbs held against the device. Each mouse was tested three times to ensure reliable data collection.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cem\u003eTreadmill exhaustion test\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003eA motorised treadmill (47300, Ugo Basile, Gemonio, Italy) was used to assess endurance. Mice were fasted for 1 h prior to an acclimatisation phase on the treadmill two days before the formal test (5° incline, 5 m/min speed). The formal test mirrored the acclimatisation protocol but with a higher speed limit (initial speed = 5 m/min, increased by 1 m/min until reaching 25 m/min and held constant). Exhaustion was defined as the mouse remaining on the electrified rear of the treadmill (\u0026gt; 5 s) despite mild stimulation. Total distance and time to exhaustion were recorded.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eStructure prediction and alignment for bacterial HicDHs\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003ePutative 2-hydroxyisocaproate dehydrogenase (HicDH) structures from LDE-1 and LDE-2 were modelled using AlphaFold2\u003csup\u003e50\u003c/sup\u003e. Predicted structures were aligned to experimentally resolved templates (PDB IDs: 1DXY [\u003cem\u003eLactobacillus casei\u003c/em\u003e D-2-HicDH]\u003csup\u003e51\u003c/sup\u003e, 1HYH [\u003cem\u003eWeissella Confusa\u0026nbsp;\u003c/em\u003eL-2-HicDH]\u003csup\u003e52,53\u003c/sup\u003e in PyMOL (v. 3.1.6.1, Schrödinger, New York, NY, USA). Root-mean-square deviation (RMSD) values were calculated by PyMOL using the entire protein structure. \u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003ePhylogenetic analysis of putative HicDHs\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eA neighbour-joining\u003csup\u003e55\u003c/sup\u003e tree was constructed in MEGA11\u003csup\u003e46\u003c/sup\u003e. Each of the protein sequences of the six putative HicDHs from LDE-1 and LDE-2 was used for BLAST of 3 proteins with top total scores, which belong to 3 different genera within NCBI \u003cem\u003erefseq_protein\u003c/em\u003e. Three well-characterised HicDHs, including L-2-HicDH of \u003cem\u003eWeissella confusa\u003c/em\u003e, D-2-HicDH of \u003cem\u003eLacticaseibacillus casei\u003c/em\u003e, and LDHD of \u003cem\u003eMus musculus,\u003c/em\u003e were additionally included. The percentage of replicate trees in which the associated taxa clustered together in the bootstrap test (500 replicates) is shown next to the branches\u003csup\u003e55\u003c/sup\u003e. The evolutionary distances were computed using Poisson correction\u003csup\u003e56,57\u003c/sup\u003e distances with complete deletion of gaps/missing data (279 positions retained).\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eCloning of putative 2-hydroxyacid dehydrogenases\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003ePutative 2-hydroxyacid dehydrogenases were identified through in silico screening of predicted proteins encoded by the draft genomes of strains LDE-1 and LDE-2. A batch conserved domain (CD)-Search against the NCBI CDD database (https://www.ncbi.nlm.nih.gov/Structure/bwrpsb/bwrpsb.cgi)\u003csup\u003e28\u003c/sup\u003e revealed six candidates annotated as 2-hydroxyisocaproate dehydrogenase (HicDH), HicDH-like proteins, or 2-hydroxyacid dehydrogenases. Coding sequences for these candidates were codon-optimised for heterologous expression in\u003cem\u003e\u0026nbsp;Escherichia coli\u003c/em\u003e using ExpOptimizer (NovoPro, https://www.novoprolabs.com/tools/codon-optimization), chemically synthesised (Tsingke Biotech, Beijing, China), and cloned into the expression vector pGEX-4T-1 for downstream functional characterisation.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eDetermination of the\u003c/strong\u003e\u003cstrong\u003e\u0026nbsp;enantiomer\u003c/strong\u003e\u003cstrong\u003eic excess of leucic acid\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eEnantiomeric excess analysis was performed with modifications to a previously described method\u003csup\u003e58\u003c/sup\u003e. Strains were cultured in mGAM medium and 4-Methyl-2-oxovaleric acid (1 mM, M104530, Aladdin Scientific, Shanghai, China), adjusted to pH 7.0 with NaOH, was added 12 h prior to sampling. Leucic acid was analysed as methyl esters by a trimethylsilyldiazomethane methylation approach. The methylation procedure was as follows. Leucic acid was extracted three times with diethyl ether after acidification with 0.2 M HCl. Pooled ether layers were dried under nitrogen gas. Residues were reconstituted in 90 μL methanol:ether (2:7, v:v), cooled to 4 ℃, and reacted with 10 μL trimethylsilyl-diazomethane (T140538, Aladdin Scientific) for 3 h at 25 ℃. The resulting leucic acid methyl esters, along with a commercial racemic mixture standard (D9509, TargetMol Chemicals, Shanghai, China), were quantified using a CP-Chirasil-Dex CB column (25 m × 0.25 mm × 0.25 μm) on an GC-MS system (Agilent 8890B GC system and 7000D MSD system, Agilent Technologies, Santa Clara, CA, USA) using helium at 1 mL/min\u0026nbsp;(constant flow). Temperature program was as follows: 80 ℃ (2 min), ramp to 114 ℃ at 2 ℃/min, 194 ℃ at 10 ℃ (hold 3 min). Injection was performed with split mode (10:1), 1 μL volume. DL-leucic acid-methyl ester (T9509, TargetMol Chemicals, Shanghai, China) served as the external standard. Data were analysed using MassHunter Qualitative Analysis software v. 10.0 (Agilent Technologies). \u0026nbsp; \u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eStable isotope tracing of leucic acid production\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e\u003cem\u003eBacterial culture and isotope labelling\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003e\u003cem\u003eE. coli\u0026nbsp;\u003c/em\u003eBL21 harbouring pGEX-4T-1-HicDHs were grown in LB medium (with 100 μg/mL ampicillin) at 37 ℃ to OD\u003csub\u003e600\u003c/sub\u003e = 0.6. Protein expression was induced with 0.5 mM isopropyl β-D-1-thiogalactopyranoside (IPTG), followed by incubation for 3 h at 37 ℃. After centrifugation, the cells were resuspended to OD\u003csub\u003e600\u003c/sub\u003e = 0.4 in fresh LB containing 1 mM L-leucine-5,5,5-D3 (HY-N0486S9, MedChemExpress) and cultured for 3 h.\u003c/p\u003e\n\u003cp\u003e\u003cem\u003eMetabolite extraction and derivatisation\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003eA 500 μL aliquot of the culture supernatant was dried using centrifugal freeze-drying and extracted with 200 μL methanol, sonicated (30 s, 20 kHz, 40 W) and centrifuged (18,000 \u003cem\u003eg\u003c/em\u003e, 10 min). The supernatant (50 μL) was transferred to a glass vial, dried under nitrogen gas, and derivatised by sequential treatment with 50 μL pyridine containing 20 mg/mL methoxyamine hydrochloride (89803, Sigma-Aldrich; 37 ℃, 1.5 h; for methoximation of carbonyl groups) and 50 μL N-methyl-N-(trimethylsilyl) trifluoroacetamide (MSTFA; M-132, Merck, Darmstadt, Germany; 37 ℃, 1 h; for trimethylsilylation).\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cem\u003eGC-MS analysis\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003eDerivatised samples were analysed using a quadrupole GC-MS system (Agilent\u0026nbsp;7890B GC system and 5977B MSD System) equipped with an HP-5ms capillary column (30 m × 0.25 mm × 0.25 μm; Agilent Technologies). Helium carrier gas was maintained at a constant flow rate of 1 mL/min. The temperature program was as follows: 70 ℃ (2 min), ramp to 140 ℃ at 7 ℃/min, then ramp to 310 ℃ at 30 ℃/min (hold 9 min). Samples (1 μL) were injected in split mode (10:1 ratio). Electron impact ionisation (70 eV) was performed with mass spectra acquired in full-scan mode (m/z 50-650). Data were analysed using MassHunter (Agilent Technologies). Isotopomer distributions were quantified by integrating ion chromatograms (leucic acid: m/z 159; leucic acid-5,5,5-D3: m/z 162) corresponding to the trimethylsilylated leucic acid fragment, normalised to an external calibration curve generated using synthetic leucic acid-5,5,5-D3. The standard was synthesised via diazotisation of d3-leucine as previously described\u003csup\u003e59\u003c/sup\u003e.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eTissue proteomic analysis by mass spectrometry\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eGonadal WAT (gWAT) of 8-week-old male mice were used for proteomic analysis. Mice were adapted to HFD-feeding for one week, following administration of 50 mM leucic acid in drinking water or acidified water with equal pH as a control. Mice were euthanised 2 weeks post-treatment, and tissues were immediately removed for preparation of protein extracts. Tissues were homogenised in 1% cholic acid (V900388, Sigma-Aldrich) at a 1:3 (w/v) tissue-to-buffer ratio using a homogeniser. Homogenates were sonicated (15 cycles, 30 Hz) and clarified by centrifugation at 14,000 × \u003cem\u003eg\u0026nbsp;\u003c/em\u003efor 15 min at 4 °C. The lipid layer was carefully discarded, and the infranatant was transferred to a new tube and centrifuged again under identical conditions. The final protein supernatant was collected, and its concentration was determined using BCA assay.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eFor each sample, 10 μg of protein was subjected to in-solution trypsin digestion. Dried peptide samples were reconstituted in 10 μL of 0.1% formic acid. Peptide concentration was estimated by measuring OD\u003csub\u003e214\u003c/sub\u003e. Subsequently, 400 ng of peptides from each sample were separated online using a nanoElute nanoflow LC system coupled to a timsTOF HT mass spectrometer (Bruker corporation, Billerica, MA, USA) equipped with a CaptiveSpray ion source. Peptides were loaded onto a homemade C18 column (75 μm inner diameter × 35 cm, 1.9 μm particle size, 100 Å pore size) and eluted with a 60-min linear gradient from 3% to 35% acetonitrile in 0.1% formic acid at a constant flow rate of 300 nL/min.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eMass spectra were acquired in data-independent acquisition parallel accumulation serial fragmentation (DIA-PASEF) mode. Raw data files were processed using Spectronaut 17.0 (Biognosys) in a directDIA+ mode against the UniProtKB \u003cem\u003eMus musculus\u003c/em\u003e proteome database (release 2023_01). Search parameters were set as follows: enzyme, trypsin/P; specificity, full; maximum missed cleavages, 2; fixed modification, carbamidomethylation (C); variable modifications, oxidation (M) and protein N-terminal acetylation. Cross-run normalisation was enabled based on the total peptide signal. All other parameters were set to the software default. Missing values were imputed with the lower detection limit, and log\u003csub\u003e2\u003c/sub\u003e-transformed intensities were used for fold-change calculation and volcano plot generation. The mass spectrometry proteomics data have been deposited to the ProteomeXchange Consortium (https://proteomecentral.proteomexchange.org) via the iProX partner repository\u003csup\u003e60,61\u003c/sup\u003e with the dataset identifier PXD068013.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eTranscriptomic profiling of adipose tissues\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003egWAT and iWAT of 8-week-old male WT and\u0026nbsp;\u003cem\u003eAida\u003csup\u003e-/-\u003c/sup\u003e\u003c/em\u003e mice were subjected to transcriptomic analysis. RNA-seq was performed by Novogene (Beijing, China) on the Novaseq platform (Illumina, San Diego, CA, USA) with paired-end 150 bp reads. Raw reads were trimmed of adapter sequences and filtered to remove low-quality bases (Phred score \u0026lt; 20) and contaminants using FastQC (v0.11.9). High-quality reads were aligned to the GRCm39 mouse reference genome (Mus musculus) using HISAT2 (v2.2.1), and transcript counts were generated with featureCounts (v2.0.6) using the following flags: featureCounts -p --countReadPairs -t exon -g gene_id. Differential gene expression analysis was performed in R (v4.3.2) using the \u003cem\u003elimma-voom\u003c/em\u003e pipeline. Genes with |log\u003csub\u003e2\u003c/sub\u003e(fold change)| \u0026gt; 0.5 and \u003cem\u003eP\u003c/em\u003e \u0026lt; 0.05 were classified as significantly differentially expressed. The RNA-sequencing data have been deposited in the NCBI Gene Expression Omnibus (GEO) under accession number GSE306938.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eFunctional enrichment of differentially expressed genes\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eDifferentially expressed genes (DEGs) were defined as those with |log\u003csub\u003e2\u003c/sub\u003e(fold change)| \u0026gt; 0.5 and \u003cem\u003eP\u003c/em\u003e \u0026lt; 0.05. DEGs underwent functional enrichment analysis using ShinyGO (v. 0.82)\u003csup\u003e62\u003c/sup\u003e with the following parameters: enrichment fold FDR \u0026lt; 0.05, pathway size 2-500 genes, organism database \u003cem\u003eMus musculus\u003c/em\u003e. Selected enriched GO terms and Reactome\u003csup\u003e63\u003c/sup\u003e pathways were visualised. Gene set enrichment analysis (GSEA)\u003csup\u003e64\u003c/sup\u003e was performed on the full expression dataset using the gseKEGG function in clusterProfiler R package (v4.8.3). To ensure statistical robustness, we conducted the analysis with the default number of permutations while maintaining default weight settings. After the analysis, we used gseaplot2 function to visualise the selected enrichment patterns.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eImmunoblot analysis of adipose tissues\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eGonadal and inguinal WAT from HFD-fed male mice treated with 50 mM leucic acid in drinking water for 6 weeks were minced and homogenized in ice-cold RIPA lysis buffer (20 mM Tris-HCl pH 7.5, 150 mM NaCl, 1 mM Na\u003csub\u003e2\u003c/sub\u003eEDTA, 1 mM EGTA, 1% NP-40, 1% sodium deoxycholate, 2.5 mM sodium pyrophosphate, 1 mM β-glycerophosphate) supplemented with 0.1% SDS and Phosphatase Inhibitor Cocktail I/II (HY-K0021/ HY-K0022, MedChemExpress), and Protease Inhibitor Cocktail (B14002, Selleck, Shanghai, China) using a tissue homogenizer (OSE-TH-02, TIANGEN, Beijing, China) at 7 m/s linear velocity for 5 cycles (40 s run/20 s pause per cycle). Homogenates were sonicated and centrifuged. Protein concentrations in the supernatant were determined by BCA assay (20201ES86, Yeasen, Shanghai, China), and proteins were denatured in SDS sample buffer at 70 °C for 15 min. 15 μg protein per lane were resolved by SDS-PAGE and analysed via immunoblotting. Primary antibodies were ACSS2 (1:3,000, 16087-1-AP, RRID: AB_2878216, Proteintech, Wuhan, China), DGAT2 (1:1,000, 17100-1-AP, RRID: AB_2918049, Proteintech), FABP4 (1:10,000, 15872-1-AP, RRID: AB_2102440, Proteintech), FASN (1:5,000, A22462PM, RRID: AB_2862543, ABclonal, Wuhan, China), MOGAT2 (1:500, 19514-1-AP, RRID: AB_10638917, Proteintech) and UCP1 (1:1,000, 72298, RRID: AB_2936479, Cell Signaling Technology, Danvers, MA, USA). ACSS2, FASN and FABP4 antibodies have been validated by knockdown/knockout, DGAT2 and MOGAT2 antibodies by overexpression positive controls, and UCP1 antibody by tissue-specific expression. Secondary antibodies were goat anti-rabbit IgG HRP (1:5,000, 31460, RRID: AB_228341, Thermo Scientific, Waltham, MA, USA) and goat anti-mouse IgG HRP (1:5,000, 31430, RRID: AB_228307, Thermo Scientific). Integrated intensities of protein bands were quantified by ImageJ.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eMetabolic flux analysis using [U-\u003csup\u003e13\u003c/sup\u003eC]oleic acid\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eMetabolic flux analysis was performed in beige-like adipocytes and MEFs using [U-\u003csup\u003e13\u003c/sup\u003eC]oleic acid tracing. For beige-like adipocytes, 3T3-L1 cells were treated with 50 μM sodium salts of D-leucic acid, L-leucic acid, or NaCl vehicle control throughout differentiation and beigeing induction. MEFs were treated with the same compounds for 24 h. One hour prior to harvest, the medium was replaced with fresh medium containing 250 μM [U\u003cstrong\u003e-\u003csup\u003e13\u003c/sup\u003e\u003c/strong\u003eC]oleic acid (#490431, Sigma-Aldrich, Taufkirchen, Germany) alongside the respective leucic acid enantiomers or control.\u003c/p\u003e\n\u003cp\u003eAfter incubation, cells were washed twice with 30 mL ice-cold saline, and culture dishes were flash-frozen in liquid nitrogen for 10 min. Cells were subsequently scraped into pre-chilled Eppendorf tubes using 1 mL ice-cold 80% methanol containing 10 μg/mL 2-hydroxyoctanoic acid (H862015, Macklin Biochemical Technology) as internal standard. Lysates were sonicated and centrifuged at 12,000 × \u003cem\u003eg\u003c/em\u003e for 10 min at 4 °C. An 800 μL aliquot of the supernatant was lyophilised to complete dryness.\u003c/p\u003e\n\u003cp\u003eThe dried residue was derivatised by resuspension in 50 μL pyridine containing 20 mg/mL methoxyamine hydrochloride (89803, Sigma-Aldrich), followed by sonication and incubation at 37 °C for 1.5 h. Next, 50 μL N-Methyl-N-(\u003cem\u003etert\u003c/em\u003e-butyldimethylsilyl)trifluoroacetamide (MTBSTFA; M-108, Cerilliant, Round Rock, Texas, USA) was added, and the mixture was incubated at 55 °C for 1 h. After centrifugation, 60 μL of the derivatised supernatant was transferred for GC-MS analysis.\u003c/p\u003e\n\u003cp\u003eGC-MS was conducted using an Agilent 7890B/5977B system. Samples (2 μL) were injected in split mode (1:10 ratio) with an inlet temperature of 250 °C. The GC oven temperature program initiated at 70 °C (2-min hold), ramped to 140 °C at 3 °C/min, then to 310 °C at 10 °C/min. Mass spectrometry operated in SIM mode, targeting [M-57]\u003csup\u003e+\u003c/sup\u003e fragments (loss of \u003cem\u003etert\u003c/em\u003e-butyldimethylsilyl group) and isotopologue-specific M+n ions (n = number of \u003csup\u003e13\u003c/sup\u003eC atoms) of \u003cem\u003etert\u003c/em\u003e-butyldimethylsilyl (TBDMS) derivatives. Isotopic natural abundance was corrected using IsoCor software (v. 2.2.3)\u003csup\u003e65\u003c/sup\u003e.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eIntegrative target identification using orthogonal computational platforms\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eComputational target prediction for L- and D-leucic acid enantiomers was performed using two orthogonal algorithmic platforms, Ensemble TargetNet\u003csup\u003e31\u003c/sup\u003e and SwissTargetPrediction\u003csup\u003e32\u003c/sup\u003e, to ensure robust identification of potential targets. Ensemble TargetNet is a deep learning model that generates binding probabilities based on three-dimensional molecular features, including ECFP6, MACCS, and Daylight fingerprints, where targets exceeding the probability threshold of 0.7 were retained as high-confidence candidates. Complementary to this, SwissTargetPrediction employs a hybrid approach combining two-dimensional and three-dimensional ligand similarity matching with machine learning algorithms, but due to its systematic underestimation of probability values, only the top 100 ranked targets were considered for further analysis. Lipinski's Rule of Five parameters, which evaluate drug-likeness through molecular weight, hydrogen bond donors, hydrogen bond acceptors, and log\u003cem\u003eP\u003c/em\u003e\u003csup\u003e33\u003c/sup\u003e were computed directly from the TargetNet framework. The stereospecific structures of both enantiomers were defined using canonical SMILES strings with explicit stereochemistry annotations. L-leucate represented as CC(C)C[C@@H](C(=O)O)O and D-leucate as CC(C)C[C@H](C(=O)[O-])O, which were submitted as input for both prediction platforms to account for chiral specificity in target engagement.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eMolecular docking\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eMolecular docking simulations were performed using AutoDock Vina\u003csup\u003e66\u003c/sup\u003e (v. 1.2.7) to elucidate the binding interactions between leucic acid enantiomers and the human HCAR2. The HCAR2 (PDB ID: 8H2G) cryo-EM structure was prepared by removing crystallographic water molecules, adding polar hydrogens, and assigning Kollman united-atom charges via AutoDockTools (v. 1.5.7)\u003csup\u003e67\u003c/sup\u003e. Leucic acid enantiomers as well as niacin and β-hydroxybutyrate were energy-minimised using the MM2 force field in Chem 3D (v. 22.0.0, Revvity, Waltham, MA, USA), with Gasteiger charges assigned using AutoDockTools (v. 1.5.7)\u003csup\u003e68\u003c/sup\u003e. A docking grid box (20 × 20 × 20 Å\u003csup\u003e3\u003c/sup\u003e) was centred on the orthosteric binding site (coordinates: x = 115.895, y = 114.032, z = 160.702) to encompass key residues. Vina parameters included an exhaustiveness of 64, energy range of 4 kcal/mol, and 10 binding poses generated per run. The top-scoring pose (lowest predicted ΔG) for each target was selected. All visualisations and interaction analyses were conducted in PyMOL (v. 3.1.6.1, Schrödinger).\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eLigand-receptor binding assay via \u003csup\u003e19\u003c/sup\u003eF-NMR in live cells\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eLigand-receptor interactions were analysed using real-time \u003csup\u003e19\u003c/sup\u003eF NMR in live cells, adapted from established protocols\u003csup\u003e35,36\u003c/sup\u003e. Cells were harvested by EDTA-mediated detachment, washed twice with ice-cold PBS, and resuspended in DMEM supplemented with 2 mM EDTA and 10% OptiPrep (D1556, Sigma-Aldrich). 4-fluoroleucic acid was synthesised via diazotisation of 4-fluoroleucine (BD271440, Bidepharm, Shanghai, China), purified by following a previously reported protocol\u003csup\u003e60\u003c/sup\u003e. The yield of 4-F-leucic acid from 4-F-leucine was found to be equivalent to the yield of leucic acid from leucine, as determined by GC-MS. The compound was purified via methyl \u003cem\u003etert\u003c/em\u003e-butyl ether extraction and dried under nitrogen, dissolved and neutralised with KOH to a 30 mg/mL stock (pH 7.0). For binding assays, 400 μL cell suspension was mixed with 50 μL fluorinated ligand, 5 μL potassium trifluoroacetate (100 mg/mL internal standard), and 50 μL D\u003csub\u003e2\u003c/sub\u003eO, incubated (10 min, 37 ºC), and analysed on a high-resolution NMR spectrometer (AVANCE-NEO-600, Bruker, Billerica, USA) equipped with a broadband \u003csup\u003e19\u003c/sup\u003eF probe. Acquisition parameters: 16 scans, 12 μs pulse width, no decoupling, ambient temperature. \u003csup\u003e19\u003c/sup\u003eF NMR spectra were processed in MestReNova software v. 15.0.0-34764 (Mestrelab Research, Santiago de Compostela, Spain) using automated phase correction, Whittaker smoothing, and internal reference calibration.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eCellular respiration monitoring (Resipher)\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eRespiration of intact cells under normal culture conditions (37 ℃, 5% CO\u003csub\u003e2\u003c/sub\u003e) was performed using a non-invasive oxygen consumption rate monitoring system (Resipher 32X, Lucid Scientific Inc., Atlanta, GA, USA)\u003csup\u003e68\u003c/sup\u003e. Briefly, 5 × 10\u003csup\u003e3\u003c/sup\u003e cells were seeded in 96-well microplates. Cells were treated with 50 μM D- or L-leucic acid (sodium salt), or NaCl (control) for 36 h. The Resipher oxygen sensing lid was positioned upon the plate before monitoring. Oxygen concentration was measured in each well every minute at 37 ºC, and OCR was calculated based on the oxygen concentration during 15 min. Data were exported using the Resipher web application and OCR normalised to cell count.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eHigh-resolution respirometry (Oxygraph-O2k)\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eHigh-resolution respirometry with intact cells was performed using the Oroboros Oxygraph-O2k (Oroboros Instruments, Innsbruck, Austria)\u003csup\u003e69\u003c/sup\u003e. 3T3-L1 preadipocytes were seeded into 6-well plate. Cells were grown to confluence and differentiated into mature adipocytes following the protocol as described above, with 50 μM D- or L-leucic acid (sodium salt), or NaCl as a control in the induction and maintenance media. Following full differentiation, these adipocytes were washed twice with PBS, trypsinised and centrifuged at 300 × \u003cem\u003eg\u003c/em\u003e for 5 min, resuspended in DMEM (supplemented with non-essential amino acids, GlutaMax and sodium pyruvate). The Oroboros O2k chambers were pre-oxygenated, and the chamber temperature was set to 37 ℃. The oxygen concentration was calibrated before the experiment. An aliquot containing 2 × 10\u003csup\u003e6\u003c/sup\u003e cells was added to each chamber and allowed to stabilise for 10 min before measurement began. Baseline oxygen consumption rates were recorded. Sequentially, 2.5 μM oligomycin (s1478, Selleck), 0.5 μM FCCP (s8276, Selleck), 0.5 μM rotenone (s2348, Selleck) and 2.5 μM antimycin A (ab141904, Abcam, Cambridge, UK) were sequentially added into the chambers to characterise maximum mitochondrial oxygen consumption. Data were recorded and analysed using DatLab software (v. 7.4.0.4, Oroboros Instruments). Oxygen consumption rates (OCR) were normalised to the protein concentrations of the respective samples and expressed as pmol O\u003csub\u003e2\u003c/sub\u003e/s/mg protein.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eUnilateral\u003c/strong\u003e \u003cstrong\u003eAAV-mediated \u003cem\u003eHcar2\u003c/em\u003e knockdown in iWAT\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eAAV2/9 vectors were generated for \u003cem\u003ein vivo\u003c/em\u003e knockdown of \u003cem\u003eHcar2\u003c/em\u003e using a triple-plasmid system. The target sequences were \u003cem\u003eRenilla\u003c/em\u003e as a control are 5’-GCTGTGTGTTCCGAGATGA-3’ for murine \u003cem\u003eHcar2\u003c/em\u003e and 5’-GTAGCGCGGTGTATTATAC-3’ for \u003cem\u003eRenilla\u003c/em\u003e (control). The transgene plasmid, bearing AAV2 inverted terminal repeats, contained \u003cem\u003emU6\u003c/em\u003e promoter for shRNA expression. This was co-transfected into HEK293T cells alongside the pAAV2/9 rep-cap plasmid with sequences coding for REP (from type 2 AAV) and CAP (from type 9 AAV), and the pAdΔF6 adenoviral helper plasmid. Viral particles were harvested 60-72 h post-transfection. Cells underwent three cycles of freeze-thaw lysis (liquid nitrogen/37 °C), followed by precipitation of medium-derived virions with polyethylene glycol (PEG; final concentration of 8 % PEG-8000, 0.5 M NaCl; 4 °C overnight). Lysates and PEG pellets were combined, purified via discontinuous OptiPrep (D1556, Sigma-Aldrich) density gradient ultracentrifugation (17%, 25%, 40%, 60 %), and the 40% fraction was collected. Viruses were washed three times with PBS using 100 kDa ultrafiltration centrifugation (UFC910008, Millipore, Billerica, MA, USA). Genomic titers were quantified by qPCR.\u003c/p\u003e\n\u003cp\u003eTwelve 6-week-old male mice were housed under standard conditions. Bilateral iWAT depots received differential injections. The right depot was administered 2 × 10\u003csup\u003e11\u003c/sup\u003e genomic copies of AAV2/9-\u003cem\u003emU6\u003c/em\u003e-\u003cem\u003eshRenilla\u0026nbsp;\u003c/em\u003e(control) in 50 µL PBS across four sites, and the left depot received AAV2/9-\u003cem\u003emU6\u003c/em\u003e-\u003cem\u003eshHCAR2\u003c/em\u003e (knockdown) identically. Following a 12-day recovery period, mice were acclimated to HFD for 4 days. Animals were then randomised into two weight-matched groups. One group received 50 mM leucic acid in drinking water, and the controls received pH-matched acidified water. Tissues were harvested 13 weeks post-injection for analysis.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eStatistics and reproducibility\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eStatistical analyses were performed using GraphPad Prism (v. 9.0, GraphPad Software, La Jolla, CA, USA). Normality was assessed using D'Agostino-Pearson omnibus or Shapiro-Wilk tests to guide test selection. For two-group comparisons, two-tailed unpaired Student’s \u003cem\u003et\u003c/em\u003e-tests or Mann-Whitney tests were applied (depending on normality). Multi-group comparisons involving a single variable employed one-way ANOVA (parametric) or Kruskal-Wallis tests (non-parametric), with Tukey’s, Dunn’s or Dunnett's post-hoc corrections as detailed in figure legends. For experiments with two independent variables, two-way ANOVA with Dunnett's or Holm-Šídák’s post-hoc testing was utilised. Statistical significance between growth curves was determined by comparing fits using an extra sum-of-squares F-test for curve divergence\u003csup\u003e70\u003c/sup\u003e. To assess the statistical significance of differences among groups in Principal Coordinates Analysis (PCoA) plots, Analysis of Variance Using Distance Matrices (ADONIS), also known as Permutational Multivariate Analysis of Variance (PERMANOVA), was employed by using ClustVis tool\u003csup\u003e71\u003c/sup\u003e. Pearson's correlation coefficient (\u003cem\u003er\u003c/em\u003e) or Spearman's rank correlation coefficient (\u003cem\u003eρ\u003c/em\u003e) (depending on normality), along with their associated \u003cem\u003eP\u003c/em\u003e-value, was utilised to evaluate the relationship between two variables. A significance threshold of \u003cem\u003eP\u003c/em\u003e \u0026lt; 0.05 was applied throughout. All experiments included ≥ 3 biological replicates, with full statistical details provided in figures and source data files.\u0026nbsp;\u003c/p\u003e"},{"header":"Declarations","content":"\u003cp\u003e\u003cstrong\u003eData availability\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe 16S rRNA sequencing data have been deposited in the National Center for Biotechnology Information (NCBI) GenBank database under BioProject accession number PRJNA1311729. Genome assemblies were deposited in NCBI GenBank under BioProject accession PRJNA1311729. The metabolomics data have been deposited to MetaboLights repository with the study identifier MTBLS12947. The RNA-sequencing data have been deposited in the NCBI Gene Expression Omnibus (GEO) under accession number GSE306938. The mass spectrometry proteomics data have been deposited to the ProteomeXchange Consortium (https://proteomecentral.proteomexchange.org) via the iProX partner repository with the dataset identifier PXD068013. Statistical source data and image source data are provided with this paper. Bacteria strains LDE-1 and LDE-2, and all other data that support the findings of this study, are available from the corresponding author upon reasonable request. The uncropped immunoblot and microscopy images (PDF) and statistical source data (Excel) have been deposited in a private Figshare repository and can be accessed via this link for review: https://figshare.com/s/3dea33aa5b47932a9a7b\u003c/p\u003e\u003cp\u003e\u003cstrong\u003eAcknowledgements\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eWe gratefully acknowledge\u0026nbsp;Drs. Haojie Huang, Xiaoxing Li and Xiang Chen for their assistance with bacterial strain typing. We thank Dr. Shu-Hai Lin for guidance on metabolomics techniques, Dr. Baoding Zhang for molecular docking, and Dr. Guang Li for phylogenetic analysis. We are grateful to Drs. Changtao Jiang, Tong-Jin Zhao, Lemin Zheng, Shih-Chin Cheng and Weiqi Xu for their critical comments and suggestions. Technical support for animal studies was provided by the Laboratory Animal Center of Xiamen University. Finally, we would like to thank all members of the SCL laboratory for their contributions to this work. This study utilises data from the GWAS Catalog\u003csup\u003e34\u003c/sup\u003e. This work was supported by grants from the National Key Research and Development Program of China (2022YFA0806500, 2024YFC3407000), the National Natural Science Foundation of China (32571357, 31822027, 32271216, 82088102, 32422002, 32370052), the Fundamental Research Funds for the Central Universities (20720210110, 20720200069), Project “111\" sponsored by the State Bureau of Foreign Experts and Ministry of Education of China (BP2018017), the Joint Funds for the Innovation of Science and Technology, Fujian province (2021Y9232), the XMU-Fujian Cancer Hospital cooperation grant for the Research Center of Metabolism and Tumor, the Joint Funds for the Innovation of Science and Technology, Fujian province (2024Y9603), and XMU Training Program of Innovation, and Entrepreneurship for Undergraduates (2020X892, 202410384095, 202510384048).\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAuthor information\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eAuthors and Affiliations\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eState Key Laboratory of Cellular Stress Biology, School of Life Sciences, Faculty of Medicine and Life Sciences, Xiamen University, Xiamen, China\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eZhi-Yu Chen, Hao-Yu Qin, Yang Alan Yang, Dong-Tai Liu, Shu-Han Yang, Ximin Chi, Hao Guo, Feng Guo, Qi-Qian Han, Pu-Chun Zhu, Wen-Jing Wu, Ming Jiang, Ming-Xia Zhu, Cixiong Zhang, Lei Zhang, Yaying Wu, Ao Li, Luming Yao, Wei Hong, Tian-Yu Tang, Sheng-Cai Lin, Shu-Yong Lin\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eDepartment of Gastroenterology, Clinical Research Center for Gut Microbiota and Digestive Diseases of Fujian Province, The National Key Clinical Specialty, Zhongshan Hospital of Xiamen University, School of Medicine, Xiamen University, Xiamen, China\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eKan Wang, Yang Song, Jianlin Ren, Hongzhi Xu, Shu-Yong Lin\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eDepartment of Endocrine and Metabolic Diseases, Shanghai Institute of Endocrine and Metabolic Diseases, Ruijin Hospital, Shanghai Jiao Tong University School of Medicine, Shanghai, China\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eRuixin Liu, Chao Wu\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eShanghai National Clinical Research Center for Metabolic Diseases, Key Laboratory for Endocrine and Metabolic Diseases of the National Health Commission of the PR China, Shanghai Key Laboratory for Endocrine Tumor, State Key Laboratory of Medical Genomics, Ruijin Hospital, Shanghai Jiao Tong University School of Medicine, Shanghai, China\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eRuixin Liu, Chao Wu\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eState Key Laboratory of Genetic Engineering, Shanghai Key Laboratory of Metabolic Remodeling and Health, Laboratory of Longevity and Metabolic Adaptations, Institute of Metabolism and Integrative Biology, Fudan University, Shanghai, China\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eTerytty Yang Li, Lin Chen\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eState Key Laboratory of Microbial Technology, Shandong University, Qingdao, China.\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eQinghua Zhao, Changyu Wang, Chang Liu, Shuang-Jiang Liu\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eDepartment of Chemical Engineering, Tsinghua University, Beijing, China\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eMing Jiang\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eDepartment of Endocrinology and Metabolism, Shanghai Clinical Centre for Diabetes, Shanghai Diabetes Institute, Shanghai Key Laboratory of Diabetes Mellitus, Shanghai Jiao Tong University Affiliated Sixth People’s Hospital, Shanghai, China\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eMingliang Zhang\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eContributions\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eConceptualisation: SCL and SYL. Methodology: SJL, CL, QZ, CW\u003csup\u003e6\u003c/sup\u003e, FG, ZYC, KW, HYQ, XC, HG, YS, MXZ, CZ, LY, AL, LZ, YW, MZ. Resources: SJL, CL, RL, JR, YS, HX, MZ. Investigation: ZYC, KW, DTL, HYQ, YAY, SHY, LC, MJ, PCZ. Formal analysis: ZYC, KW, DTL, HYQ, YAY, SHY, QQH, CW\u003csup\u003e3, 4\u003c/sup\u003e, TYL, SYL. Validation: SYL, ZYC, KW, HYQ, YAY. Visualisation: SYL, ZYC, XC, WJW. Writing - Original Draft: SYL, ZYC, HYQ, YAY. Writing - Review \u0026amp; Editing: SCL, SYL, SJL, HG, CL. Supervision: SYL, SCL, SJL, HX. Project administration: SYL, SCL, TYT, WH. Funding acquisition: SCL, SYL, JR, CL, HG, SHY.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eCorresponding author\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eCorrespondence to Shu-Yong Lin.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eEthics declarations\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eCompeting interests\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe authors declare the following financial interests: The bacterial strains LDE-1 and LDE-2 are the subject of a patent deposition. The findings regarding the use of leucic acid against obesity are also intended for patent protection. The authors may potentially file related patent applications in the future, which could lead to the commercialisation of technology.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\n\u003cli\u003eGroussin, M., Mazel, F., and Alm, E.J. (2020). Co-evolution and Co-speciation of Host-Gut Bacteria Systems. Cell Host Microbe \u003cem\u003e28\u003c/em\u003e, 12-22. 10.1016/j.chom.2020.06.013.\u003c/li\u003e\n\u003cli\u003eMallott, E.K., and Amato, K.R. (2021). Host specificity of the gut microbiome. Nat Rev Microbiol \u003cem\u003e19\u003c/em\u003e, 639-653. 10.1038/s41579-021-00562-3.\u003c/li\u003e\n\u003cli\u003eWilde, J., Slack, E., and Foster, K.R. (2024). Host control of the microbiome: Mechanisms, evolution, and disease. Science \u003cem\u003e385\u003c/em\u003e, eadi3338. 10.1126/science.adi3338.\u003c/li\u003e\n\u003cli\u003eCani, P.D., and Van Hul, M. (2024). Gut microbiota in overweight and obesity: crosstalk with adipose tissue. Nat Rev Gastroenterol Hepatol \u003cem\u003e21\u003c/em\u003e, 164-183. 10.1038/s41575-023-00867-z.\u003c/li\u003e\n\u003cli\u003eVan Hul, M., and Cani, P.D. (2023). The gut microbiota in obesity and weight management: microbes as friends or foe? Nat Rev Endocrinol \u003cem\u003e19\u003c/em\u003e, 258-271. 10.1038/s41574-022-00794-0.\u003c/li\u003e\n\u003cli\u003eCani, P.D., Van Hul, M., Lefort, C., Depommier, C., Rastelli, M., and Everard, A. (2019). Microbial regulation of organismal energy homeostasis. Nat Metab \u003cem\u003e1\u003c/em\u003e, 34-46. 10.1038/s42255-018-0017-4.\u003c/li\u003e\n\u003cli\u003eBackhed, F., Ding, H., Wang, T., Hooper, L.V., Koh, G.Y., Nagy, A., Semenkovich, C.F., and Gordon, J.I. (2004). The gut microbiota as an environmental factor that regulates fat storage. Proc Natl Acad Sci U S A \u003cem\u003e101\u003c/em\u003e, 15718-15723. 10.1073/pnas.0407076101.\u003c/li\u003e\n\u003cli\u003eGentile, C.L., and Weir, T.L. (2018). The gut microbiota at the intersection of diet and human health. Science \u003cem\u003e362\u003c/em\u003e, 776-780. 10.1126/science.aau5812.\u003c/li\u003e\n\u003cli\u003eSuarez-Zamorano, N., Fabbiano, S., Chevalier, C., Stojanovic, O., Colin, D.J., Stevanovic, A., Veyrat-Durebex, C., Tarallo, V., Rigo, D., Germain, S., et al. (2015). Microbiota depletion promotes browning of white adipose tissue and reduces obesity. Nat Med \u003cem\u003e21\u003c/em\u003e, 1497-1501. 10.1038/nm.3994.\u003c/li\u003e\n\u003cli\u003eChevalier, C., Stojanovic, O., Colin, D.J., Suarez-Zamorano, N., Tarallo, V., Veyrat-Durebex, C., Rigo, D., Fabbiano, S., Stevanovic, A., Hagemann, S., et al. (2015). Gut Microbiota Orchestrates Energy Homeostasis during Cold. Cell \u003cem\u003e163\u003c/em\u003e, 1360-1374. 10.1016/j.cell.2015.11.004.\u003c/li\u003e\n\u003cli\u003eGoodrich, J.K., Waters, J.L., Poole, A.C., Sutter, J.L., Koren, O., Blekhman, R., Beaumont, M., Van Treuren, W., Knight, R., Bell, J.T., et al. (2014). Human genetics shape the gut microbiome. Cell \u003cem\u003e159\u003c/em\u003e, 789-799. 10.1016/j.cell.2014.09.053.\u003c/li\u003e\n\u003cli\u003eSanna, S., Kurilshikov, A., van der Graaf, A., Fu, J., and Zhernakova, A. (2022). Challenges and future directions for studying effects of host genetics on the gut microbiome. Nat Genet \u003cem\u003e54\u003c/em\u003e, 100-106. 10.1038/s41588-021-00983-z.\u003c/li\u003e\n\u003cli\u003eNagarajan, A., Scoggin, K., Gupta, J., Threadgill, D.W., and Andrews-Polymenis, H.L. (2023). Using the collaborative cross to identify the role of host genetics in defining the murine gut microbiome. Microbiome \u003cem\u003e11\u003c/em\u003e, 149. 10.1186/s40168-023-01552-8.\u003c/li\u003e\n\u003cli\u003eGoodrich, J.K., Davenport, E.R., Beaumont, M., Jackson, M.A., Knight, R., Ober, C., Spector, T.D., Bell, J.T., Clark, A.G., and Ley, R.E. (2016). Genetic Determinants of the Gut Microbiome in UK Twins. Cell Host Microbe \u003cem\u003e19\u003c/em\u003e, 731-743. 10.1016/j.chom.2016.04.017.\u003c/li\u003e\n\u003cli\u003eDzierozynski, L., Queen, J., and Sears, C.L. (2023). Subtle, persistent shaping of the gut microbiome by host genes: A critical determinant of host biology. Cell Host Microbe \u003cem\u003e31\u003c/em\u003e, 1569-1573. 10.1016/j.chom.2023.09.007.\u003c/li\u003e\n\u003cli\u003eYang, H., Wu, J., Huang, X., Zhou, Y., Zhang, Y., Liu, M., Liu, Q., Ke, S., He, M., Fu, H., et al. (2022). ABO genotype alters the gut microbiota by regulating GalNAc levels in pigs. Nature \u003cem\u003e606\u003c/em\u003e, 358-367. 10.1038/s41586-022-04769-z.\u003c/li\u003e\n\u003cli\u003eZhernakova, D.V., Wang, D., Liu, L., Andreu-Sanchez, S., Zhang, Y., Ruiz-Moreno, A.J., Peng, H., Plomp, N., Del Castillo-Izquierdo, A., Gacesa, R., et al. (2024). Host genetic regulation of human gut microbial structural variation. Nature \u003cem\u003e625\u003c/em\u003e, 813-821. 10.1038/s41586-023-06893-w.\u003c/li\u003e\n\u003cli\u003eRuhlemann, M.C., Hermes, B.M., Bang, C., Doms, S., Moitinho-Silva, L., Thingholm, L.B., Frost, F., Degenhardt, F., Wittig, M., Kassens, J., et al. (2021). Genome-wide association study in 8,956 German individuals identifies influence of ABO histo-blood groups on gut microbiome. Nat Genet \u003cem\u003e53\u003c/em\u003e, 147-155. 10.1038/s41588-020-00747-1.\u003c/li\u003e\n\u003cli\u003eQin, Y., Havulinna, A.S., Liu, Y., Jousilahti, P., Ritchie, S.C., Tokolyi, A., Sanders, J.G., Valsta, L., Brozynska, M., Zhu, Q., et al. (2022). Combined effects of host genetics and diet on human gut microbiota and incident disease in a single population cohort. Nat Genet \u003cem\u003e54\u003c/em\u003e, 134-142. 10.1038/s41588-021-00991-z.\u003c/li\u003e\n\u003cli\u003eWu, Q., Liang, X., Wang, K., Lin, J., Wang, X., Wang, P., Zhang, Y., Nie, Q., Liu, H., Zhang, Z., et al. (2021). Intestinal hypoxia-inducible factor 2alpha regulates lactate levels to shape the gut microbiome and alter thermogenesis. Cell Metab \u003cem\u003e33\u003c/em\u003e, 1988-2003 e1987. 10.1016/j.cmet.2021.07.007.\u003c/li\u003e\n\u003cli\u003eLuo, H., Jiang, M., Lian, G., Liu, Q., Shi, M., Li, T.Y., Song, L., Ye, J., He, Y., Yao, L., et al. (2018). AIDA Selectively Mediates Downregulation of Fat Synthesis Enzymes by ERAD to Retard Intestinal Fat Absorption and Prevent Obesity. Cell Metab \u003cem\u003e27\u003c/em\u003e, 843-853 e846. 10.1016/j.cmet.2018.02.021.\u003c/li\u003e\n\u003cli\u003eShi, M., Huang, X.Y., Ren, X.Y., Wei, X.Y., Ma, Y., Lin, Z.Z., Liu, D.T., Song, L., Zhao, T.J., Li, G., et al. (2021). AIDA directly connects sympathetic innervation to adaptive thermogenesis by UCP1. Nat Cell Biol \u003cem\u003e23\u003c/em\u003e, 268-277. 10.1038/s41556-021-00642-9.\u003c/li\u003e\n\u003cli\u003eNewgard, C.B., An, J., Bain, J.R., Muehlbauer, M.J., Stevens, R.D., Lien, L.F., Haqq, A.M., Shah, S.H., Arlotto, M., Slentz, C.A., et al. (2009). A branched-chain amino acid-related metabolic signature that differentiates obese and lean humans and contributes to insulin resistance. Cell Metab \u003cem\u003e9\u003c/em\u003e, 311-326. 10.1016/j.cmet.2009.02.002.\u003c/li\u003e\n\u003cli\u003eCanfora, E.E., Jocken, J.W., and Blaak, E.E. (2015). Short-chain fatty acids in control of body weight and insulin sensitivity. Nat Rev Endocrinol \u003cem\u003e11\u003c/em\u003e, 577-591. 10.1038/nrendo.2015.128.\u003c/li\u003e\n\u003cli\u003eLayman, D.K., and Walker, D.A. (2006). Potential importance of leucine in treatment of obesity and the metabolic syndrome. J Nutr \u003cem\u003e136\u003c/em\u003e, 319S-323S. 10.1093/jn/136.1.319S.\u003c/li\u003e\n\u003cli\u003eden Besten, G., van Eunen, K., Groen, A.K., Venema, K., Reijngoud, D.J., and Bakker, B.M. (2013). The role of short-chain fatty acids in the interplay between diet, gut microbiota, and host energy metabolism. J Lipid Res \u003cem\u003e54\u003c/em\u003e, 2325-2340. 10.1194/jlr.R036012.\u003c/li\u003e\n\u003cli\u003eCrozier, S.J., Kimball, S.R., Emmert, S.W., Anthony, J.C., and Jefferson, L.S. (2005). Oral leucine administration stimulates protein synthesis in rat skeletal muscle. J Nutr \u003cem\u003e135\u003c/em\u003e, 376-382. 10.1093/jn/135.3.376.\u003c/li\u003e\n\u003cli\u003eWang, J., Chitsaz, F., Derbyshire, M.K., Gonzales, N.R., Gwadz, M., Lu, S., Marchler, G.H., Song, J.S., Thanki, N., Yamashita, R.A., et al. (2023). The conserved domain database in 2023. Nucleic Acids Res \u003cem\u003e51\u003c/em\u003e, D384-D388. 10.1093/nar/gkac1096.\u003c/li\u003e\n\u003cli\u003eAlmeida, A., Nayfach, S., Boland, M., Strozzi, F., Beracochea, M., Shi, Z.J., Pollard, K.S., Sakharova, E., Parks, D.H., Hugenholtz, P., et al. (2021). A unified catalog of 204,938 reference genomes from the human gut microbiome. Nat Biotechnol \u003cem\u003e39\u003c/em\u003e, 105-114. 10.1038/s41587-020-0603-3.\u003c/li\u003e\n\u003cli\u003eBostrom, P., Wu, J., Jedrychowski, M.P., Korde, A., Ye, L., Lo, J.C., Rasbach, K.A., Bostrom, E.A., Choi, J.H., Long, J.Z., et al. (2012). A PGC1-alpha-dependent myokine that drives brown-fat-like development of white fat and thermogenesis. Nature \u003cem\u003e481\u003c/em\u003e, 463-468. 10.1038/nature10777.\u003c/li\u003e\n\u003cli\u003eYao, Z.J., Dong, J., Che, Y.J., Zhu, M.F., Wen, M., Wang, N.N., Wang, S., Lu, A.P., and Cao, D.S. (2016). TargetNet: a web service for predicting potential drug-target interaction profiling via multi-target SAR models. J Comput Aided Mol Des \u003cem\u003e30\u003c/em\u003e, 413-424. 10.1007/s10822-016-9915-2.\u003c/li\u003e\n\u003cli\u003eDaina, A., Michielin, O., and Zoete, V. (2019). SwissTargetPrediction: updated data and new features for efficient prediction of protein targets of small molecules. Nucleic Acids Res \u003cem\u003e47\u003c/em\u003e, W357-W364. 10.1093/nar/gkz382.\u003c/li\u003e\n\u003cli\u003eLipinski, C.A., Lombardo, F., Dominy, B.W., and Feeney, P.J. (2001). Experimental and computational approaches to estimate solubility and permeability in drug discovery and development settings. Adv Drug Deliv Rev \u003cem\u003e46\u003c/em\u003e, 3-26. 10.1016/s0169-409x(00)00129-0.\u003c/li\u003e\n\u003cli\u003eSollis, E., Mosaku, A., Abid, A., Buniello, A., Cerezo, M., Gil, L., Groza, T., Gunes, O., Hall, P., Hayhurst, J., et al. (2023). The NHGRI-EBI GWAS Catalog: knowledgebase and deposition resource. Nucleic Acids Res \u003cem\u003e51\u003c/em\u003e, D977-D985. 10.1093/nar/gkac1010.\u003c/li\u003e\n\u003cli\u003eLuchinat, E., Barbieri, L., Davis, B., Brough, P.A., Pennestri, M., and Banci, L. (2024). Ligand-Based Competition Binding by Real-Time (19)F NMR in Human Cells. J Med Chem \u003cem\u003e67\u003c/em\u003e, 1115-1126. 10.1021/acs.jmedchem.3c01600.\u003c/li\u003e\n\u003cli\u003eDalvit, C., and Vulpetti, A. (2019). Ligand-Based Fluorine NMR Screening: Principles and Applications in Drug Discovery Projects. J Med Chem \u003cem\u003e62\u003c/em\u003e, 2218-2244. 10.1021/acs.jmedchem.8b01210.\u003c/li\u003e\n\u003cli\u003eSukumaran, S., Barnes, R.I., Garg, A., and Agarwal, A.K. (2009). Functional characterization of the human 1-acylglycerol-3-phosphate-O-acyltransferase isoform 10/glycerol-3-phosphate acyltransferase isoform 3. J Mol Endocrinol \u003cem\u003e42\u003c/em\u003e, 469-478. 10.1677/JME-09-0010.\u003c/li\u003e\n\u003cli\u003eLardizabal, K.D., Mai, J.T., Wagner, N.W., Wyrick, A., Voelker, T., and Hawkins, D.J. (2001). DGAT2 is a new diacylglycerol acyltransferase gene family: purification, cloning, and expression in insect cells of two polypeptides from Mortierella ramanniana with diacylglycerol acyltransferase activity. J Biol Chem \u003cem\u003e276\u003c/em\u003e, 38862-38869. 10.1074/jbc.M106168200.\u003c/li\u003e\n\u003cli\u003eZhu, M., Frank, M.W., Radka, C.D., Jeanfavre, S., Xu, J., Tse, M.W., Pacheco, J.A., Kim, J.S., Pierce, K., Deik, A., et al. (2024). Vaginal Lactobacillus fatty acid response mechanisms reveal a metabolite-targeted strategy for bacterial vaginosis treatment. Cell. 10.1016/j.cell.2024.07.029.\u003c/li\u003e\n\u003cli\u003eZhang, Y., Wang, X., Lin, J., Liu, J., Wang, K., Nie, Q., Ye, C., Sun, L., Ma, Y., Qu, R., et al. (2024). A microbial metabolite inhibits the HIF-2alpha-ceramide pathway to mediate the beneficial effects of time-restricted feeding on MASH. Cell Metab \u003cem\u003e36\u003c/em\u003e, 1823-1838 e1826. 10.1016/j.cmet.2024.07.004.\u003c/li\u003e\n\u003cli\u003eTaggart, A.K., Kero, J., Gan, X., Cai, T.Q., Cheng, K., Ippolito, M., Ren, N., Kaplan, R., Wu, K., Wu, T.J., et al. (2005). (D)-beta-Hydroxybutyrate inhibits adipocyte lipolysis via the nicotinic acid receptor PUMA-G. J Biol Chem \u003cem\u003e280\u003c/em\u003e, 26649-26652. 10.1074/jbc.C500213200.\u003c/li\u003e\n\u003cli\u003eLorenzen, A., Stannek, C., Lang, H., Andrianov, V., Kalvinsh, I., and Schwabe, U. (2001). Characterization of a G protein-coupled receptor for nicotinic acid. Mol Pharmacol \u003cem\u003e59\u003c/em\u003e, 349-357. 10.1124/mol.59.2.349.\u003c/li\u003e\n\u003cli\u003eWu, C., Yang, F., Zhong, H., Hong, J., Lin, H., Zong, M., Ren, H., Zhao, S., Chen, Y., Shi, Z., et al. (2024). Obesity-enriched gut microbe degrades myo-inositol and promotes lipid absorption. Cell Host Microbe \u003cem\u003e32\u003c/em\u003e, 1301-1314 e1309. 10.1016/j.chom.2024.06.012.\u003c/li\u003e\n\u003cli\u003eBarman, M., Unold, D., Shifley, K., Amir, E., Hung, K., Bos, N., and Salzman, N. (2008). Enteric salmonellosis disrupts the microbial ecology of the murine gastrointestinal tract. Infect Immun \u003cem\u003e76\u003c/em\u003e, 907-915. 10.1128/IAI.01432-07.\u003c/li\u003e\n\u003cli\u003eWeisburg, W.G., Barns, S.M., Pelletier, D.A., and Lane, D.J. (1991). 16S ribosomal DNA amplification for phylogenetic study. J Bacteriol \u003cem\u003e173\u003c/em\u003e, 697-703. 10.1128/jb.173.2.697-703.1991.\u003c/li\u003e\n\u003cli\u003eTamura, K., Stecher, G., and Kumar, S. (2021). MEGA11: Molecular Evolutionary Genetics Analysis Version 11. Mol Biol Evol \u003cem\u003e38\u003c/em\u003e, 3022-3027. 10.1093/molbev/msab120.\u003c/li\u003e\n\u003cli\u003eParks, D.H., Chuvochina, M., Chaumeil, P.A., Rinke, C., Mussig, A.J., and Hugenholtz, P. (2020). A complete domain-to-species taxonomy for Bacteria and Archaea. Nat Biotechnol \u003cem\u003e38\u003c/em\u003e, 1079-1086. 10.1038/s41587-020-0501-8.\u003c/li\u003e\n\u003cli\u003eParks, D.H., Chuvochina, M., Waite, D.W., Rinke, C., Skarshewski, A., Chaumeil, P.A., and Hugenholtz, P. (2018). A standardized bacterial taxonomy based on genome phylogeny substantially revises the tree of life. Nat Biotechnol \u003cem\u003e36\u003c/em\u003e, 996-1004. 10.1038/nbt.4229.\u003c/li\u003e\n\u003cli\u003eYurekten, O., Payne, T., Tejera, N., Amaladoss, F.X., Martin, C., Williams, M., and O\u0026apos;Donovan, C. (2024). MetaboLights: open data repository for metabolomics. Nucleic Acids Res \u003cem\u003e52\u003c/em\u003e, D640-D646. 10.1093/nar/gkad1045.\u003c/li\u003e\n\u003cli\u003eJumper, J., Evans, R., Pritzel, A., Green, T., Figurnov, M., Ronneberger, O., Tunyasuvunakool, K., Bates, R., Zidek, A., Potapenko, A., et al. (2021). Highly accurate protein structure prediction with AlphaFold. Nature \u003cem\u003e596\u003c/em\u003e, 583-589. 10.1038/s41586-021-03819-2.\u003c/li\u003e\n\u003cli\u003eDengler, U., Niefind, K., Kiess, M., and Schomburg, D. (1997). Crystal structure of a ternary complex of D-2-hydroxyisocaproate dehydrogenase from Lactobacillus casei, NAD+ and 2-oxoisocaproate at 1.9 A resolution. J Mol Biol \u003cem\u003e267\u003c/em\u003e, 640-660. 10.1006/jmbi.1996.0864.\u003c/li\u003e\n\u003cli\u003eNiefind, K., Hecht, H.J., and Schomburg, D. (1995). Crystal structure of L-2-hydroxyisocaproate dehydrogenase from Lactobacillus confusus at 2.2 A resolution. An example of strong asymmetry between subunits. J Mol Biol \u003cem\u003e251\u003c/em\u003e, 256-281. 10.1006/jmbi.1995.0433.\u003c/li\u003e\n\u003cli\u003eFeil, I.K., Hendle, J., and Schomburg, D. (1997). Modified substrate specificity of L-hydroxyisocaproate dehydrogenase derived from structure-based protein engineering. Protein Eng \u003cem\u003e10\u003c/em\u003e, 255-262. 10.1093/protein/10.3.255.\u003c/li\u003e\n\u003cli\u003eSaitou, N., and Nei, M. (1987). The neighbor-joining method: a new method for reconstructing phylogenetic trees. Mol Biol Evol \u003cem\u003e4\u003c/em\u003e, 406-425. 10.1093/oxfordjournals.molbev.a040454.\u003c/li\u003e\n\u003cli\u003eFelsenstein, J. (1985). Confidence Limits on Phylogenies: An Approach Using the Bootstrap. Evolution \u003cem\u003e39\u003c/em\u003e, 783-791. 10.1111/j.1558-5646.1985.tb00420.x.\u003c/li\u003e\n\u003cli\u003eZuckerkandl, E., and Pauling, L. (1965). Molecules as documents of evolutionary history. J Theor Biol \u003cem\u003e8\u003c/em\u003e, 357-366. 10.1016/0022-5193(65)90083-4.\u003c/li\u003e\n\u003cli\u003eZuckerkandl, E., and Pauling, L.C. (1965). Evolutionary Divergence and Convergence in Proteins.\u003c/li\u003e\n\u003cli\u003eValentina Jurka\u0026scaron;, Christoph K. Winkler, Silvan Poschenrieder, Paulo Oliveira, Catarina C. Pacheco, Eunice A. Ferreira, Florian Weissensteiner, Piera De Santis, Selin Kara, Robert Kourist, et al. (2022). Expression and activity of heterologous hydroxyisocaproate dehydrogenases in Synechocystis sp. PCC 6803 \u0026Delta;hoxYH. Engineering Microbiology \u003cem\u003e2\u003c/em\u003e. https://doi.org/10.1016/j.engmic.2021.100008.\u003c/li\u003e\n\u003cli\u003eHu, D.X., O\u0026apos;Brien, M., and Ley, S.V. (2012). Continuous multiple liquid-liquid separation: diazotization of amino acids in flow. Org Lett \u003cem\u003e14\u003c/em\u003e, 4246-4249. 10.1021/ol301930h.\u003c/li\u003e\n\u003cli\u003eChen, T., Ma, J., Liu, Y., Chen, Z., Xiao, N., Lu, Y., Fu, Y., Yang, C., Li, M., Wu, S., et al. (2022). iProX in 2021: connecting proteomics data sharing with big data. Nucleic Acids Res \u003cem\u003e50\u003c/em\u003e, D1522-D1527. 10.1093/nar/gkab1081.\u003c/li\u003e\n\u003cli\u003eMa, J., Chen, T., Wu, S., Yang, C., Bai, M., Shu, K., Li, K., Zhang, G., Jin, Z., He, F., et al. (2019). iProX: an integrated proteome resource. Nucleic Acids Res \u003cem\u003e47\u003c/em\u003e, D1211-D1217. 10.1093/nar/gky869.\u003c/li\u003e\n\u003cli\u003eGe, S.X., Jung, D., and Yao, R. (2020). ShinyGO: a graphical gene-set enrichment tool for animals and plants. Bioinformatics \u003cem\u003e36\u003c/em\u003e, 2628-2629. 10.1093/bioinformatics/btz931.\u003c/li\u003e\n\u003cli\u003eMilacic, M., Beavers, D., Conley, P., Gong, C., Gillespie, M., Griss, J., Haw, R., Jassal, B., Matthews, L., May, B., et al. (2024). The Reactome Pathway Knowledgebase 2024. Nucleic Acids Res \u003cem\u003e52\u003c/em\u003e, D672-D678. 10.1093/nar/gkad1025.\u003c/li\u003e\n\u003cli\u003eSubramanian, A., Tamayo, P., Mootha, V.K., Mukherjee, S., Ebert, B.L., Gillette, M.A., Paulovich, A., Pomeroy, S.L., Golub, T.R., Lander, E.S., and Mesirov, J.P. (2005). Gene set enrichment analysis: a knowledge-based approach for interpreting genome-wide expression profiles. Proc Natl Acad Sci U S A \u003cem\u003e102\u003c/em\u003e, 15545-15550. 10.1073/pnas.0506580102.\u003c/li\u003e\n\u003cli\u003eMillard, P., Delepine, B., Guionnet, M., Heuillet, M., Bellvert, F., and Letisse, F. (2019). IsoCor: isotope correction for high-resolution MS labeling experiments. Bioinformatics \u003cem\u003e35\u003c/em\u003e, 4484-4487. 10.1093/bioinformatics/btz209.\u003c/li\u003e\n\u003cli\u003eEberhardt, J., Santos-Martins, D., Tillack, A.F., and Forli, S. (2021). AutoDock Vina 1.2.0: New Docking Methods, Expanded Force Field, and Python Bindings. J Chem Inf Model \u003cem\u003e61\u003c/em\u003e, 3891-3898. 10.1021/acs.jcim.1c00203.\u003c/li\u003e\n\u003cli\u003eGoodsell, D.S., and Olson, A.J. (1990). Automated docking of substrates to proteins by simulated annealing. Proteins \u003cem\u003e8\u003c/em\u003e, 195-202. 10.1002/prot.340080302.\u003c/li\u003e\n\u003cli\u003eTriolo, M., and Khacho, M. (2024). Protocol to monitor live-cell, real-time, mitochondrial respiration in mouse muscle cells using the Resipher platform. STAR Protoc \u003cem\u003e5\u003c/em\u003e, 103330. 10.1016/j.xpro.2024.103330.\u003c/li\u003e\n\u003cli\u003eDoerrier, C., Garcia-Souza, L.F., Krumschnabel, G., Wohlfarter, Y., Meszaros, A.T., and Gnaiger, E. (2018). High-Resolution FluoRespirometry and OXPHOS Protocols for Human Cells, Permeabilized Fibers from Small Biopsies of Muscle, and Isolated Mitochondria. Methods Mol Biol \u003cem\u003e1782\u003c/em\u003e, 31-70. 10.1007/978-1-4939-7831-1_3.\u003c/li\u003e\n\u003cli\u003eMotulsky, H., Christopoulos, A., and ProQuest. (2004). Fitting models to biological data using linear and nonlinear regression : a practical guide to curve fitting, 1st Edition (Oxford University Press).\u003c/li\u003e\n\u003cli\u003eMetsalu, T., and Vilo, J. (2015). ClustVis: a web tool for visualizing clustering of multivariate data using Principal Component Analysis and heatmap. Nucleic Acids Res \u003cem\u003e43\u003c/em\u003e, W566-570. 10.1093/nar/gkv468.\u003c/li\u003e\n\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":true,"hideJournal":false,"highlight":"","institution":"","isAcceptedByJournal":false,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"nature-portfolio","isNatureJournal":true,"hasQc":false,"allowDirectSubmit":false,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"","title":"Nature Portfolio","twitterHandle":"","acdcEnabled":false,"dfaEnabled":false,"editorialSystem":"ejp","reportingPortfolio":"","inReviewEnabled":true,"inReviewRevisionsEnabled":false},"keywords":"","lastPublishedDoi":"10.21203/rs.3.rs-8045609/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-8045609/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"Gut microbiota modulates a wide range of physiological processes in the host. However, whether host genes shape microbial communities to affect host physiology remains poorly explored. Using \u003ci\u003eAida\u003csup\u003e-/-\u003c/sup\u003e\u003c/i\u003e mice that develop obesity due to increased intestinal fat absorption, we identify two \u003ci\u003eClostridium\u003c/i\u003e strains enriched in their fat-depleted gut environment. We demonstrate that the growth of these strains is suppressed by oleate both \u003ci\u003ein vivo\u003c/i\u003e and \u003ci\u003ein vitro\u003c/i\u003e, mirroring their selective expansion. Paradoxically, these strains protect against diet-induced obesity. Integrated metabolomics and \u003csup\u003e13\u003c/sup\u003eC-tracing reveal that these strains biosynthesise leucic acid, whose administration reduces adiposity and enhances lean mass and exercise capacity by suppressing lipid synthesis and promoting catabolism. By combining computational prediction with transcriptomic and human genetic evidence, we identify HCAR2 as a candidate receptor for leucic acid and confirm this by demonstrating direct binding and showing that HCAR2 is required for metabolic effects of leucic acid \u003ci\u003ein vitro\u003c/i\u003e and in adipose tissue \u003ci\u003ein vivo\u003c/i\u003e. We thus establish a homeostatic host-microbiome feedback loop wherein a host gene-defined metabolic niche selects for microbes that produce a signalling metabolite, which in turn activates a physiologically relevant receptor to suppress obesity.","manuscriptTitle":"Gut bacteria enriched in obese mice produce leucic acid to counter obesity through HCAR2","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-11-19 07:42:19","doi":"10.21203/rs.3.rs-8045609/v1","editorialEvents":[],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"nature-metabolism","isNatureJournal":true,"hasQc":false,"allowDirectSubmit":false,"externalIdentity":"natmetab","sideBox":"Learn more about [Nature Metabolism](http://www.nature.com/natmetab/)","snPcode":"","submissionUrl":"","title":"Nature Metabolism","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"ejp","reportingPortfolio":"Nature Research","inReviewEnabled":true,"inReviewRevisionsEnabled":false}}],"origin":"","ownerIdentity":"fdb287e5-bfa2-46d7-981f-a64f4d2fe142","owner":[],"postedDate":"November 19th, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"under-review","subjectAreas":[{"id":58118785,"name":"Biological sciences/Physiology/Metabolism/Fat metabolism"},{"id":58118786,"name":"Biological sciences/Physiology/Metabolism/Homeostasis"},{"id":58118787,"name":"Biological sciences/Microbiology/Bacteria/Bacterial host response"}],"tags":[],"updatedAt":"2025-12-10T02:10:25+00:00","versionOfRecord":[],"versionCreatedAt":"2025-11-19 07:42:19","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-8045609","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-8045609","identity":"rs-8045609","version":["v1"]},"buildId":"8U1c8b4HqxoKbykW_rLl7","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}