Quantification and statistical analysis

All measurements were taken from distinct biological samples. Unless otherwise noted, all statistical analyses were performed using GraphPad Prism (version 9) and tests described in the figure legends. In the case of multiple groups, a one- or two-way ANOVA with post-hoc tests were used to determine statistical significance. When only two groups were compared, nonparametric Mann–Whitney tests were used to determine statistical significance. Gene expression and metabolomic heatmaps were plotted as Z-scores using R(4.0.3) and ComplexHeatmap (2.6.2). The species-by-species t-test was applied for metabolomics data to identify the top differentially regulated metabolites that passed the nominal threshold p-values. For multiple comparisons, the Sidak procedure was used for false discovery rate (FDR) correction. Statistical analysis of energy balance was performed by ANCOVA with lean body mass as a co-variate (Mina et al., 2018). No statistical method was used to predetermine sample size. Unblinded analysis of histology was performed by the investigators. All data are expressed as mean ± SEM, unless otherwise specified.