Answers are grounded in this paper and cite their sources.
Objectives: To characterize the hepatic lipid metabolic response to intermittent hypoxia (IH) exposure in obese mice and to nominate candidate genes for further testing. Methods: An in silico re-analysis of public transcriptomic data was performed. GSE1873 (liver, IH exposure in obese mice) served as the discovery dataset; after quantile normalization, differentially expressed genes (DEGs) were identified with the limma empirical Bayes-moderated t-test, implemented in Python 3.13 and cross-checked against R limma, and were intersected with a self-curated mouse lipid metabolism gene set. Hypergeometric enrichment and preranked gene set enrichment analysis (GSEA) were performed against the genome-wide background of all detected genes. Candidate genes were nominated from differential expression, co-expression topology, and single-gene discrimination, and were then examined in two external datasets. GSE49195 (ob/ob vs. lean wild-type liver) was used to describe the obesity baseline, and its concordance with the hypoxia effect profile was assessed. Results: Of 47 DEGs (29 upregulated, 18 downregulated), none reached FDR < 0.05 and they are therefore reported as exploratory candidates; 17 of them were lipid metabolism related. Cholesterol biosynthesis (9/28 detected genes, FE = 92.19, p = 2.22 × 10−16), lipogenesis and regulation of lipid synthesis (6/46, FE = 37.41), fatty acid biosynthesis (4/34, FE = 33.74), and lipid droplets and lipolysis (2/18, FE = 31.87) were over-represented. GSEA showed positive enrichment of cholesterol biosynthesis (NES = +2.618) and fatty acid biosynthesis (NES = +2.186), and negative enrichment of lipid droplets and lipolysis (NES = −1.975) and fatty acid β-oxidation (NES = −1.727); the PPAR signaling pathway showed a negative trend that did not survive correction (NES = −1.402, p = 0.034, FDR = 0.051). Eleven candidate genes were nominated: Idi1, Pmvk, Scd2, Acly, Msmo1, Cyp51, Sqle, Plin4, Me1, Elovl3, and Cidea. Me1 alone reached an apparent AUC of 1.00 and the 11-gene model 0.94 in repeated cross-validation, but the performance was not reproduced when feature selection was moved inside the cross-validation loop (mean AUC 0.41), indicating small-sample selection bias rather than validated prediction. Directional agreement in the external datasets was 8/8 and 9/10 among testable genes, no candidate reached an adjusted significance threshold in validation, and rank correlations of effect size were negligible to moderate (rho = −0.02 and 0.52). The obesity comparison identified 4866 DEGs, with lipogenesis (NES = +2.65) and lipid droplets and lipolysis (NES = +2.11) already upregulated, whereas cholesterol biosynthesis was unchanged (NES = −0.95); the two effect profiles showed negligible correlation across lipid metabolism genes (rho = −0.005). Conclusions: In obese mouse liver, IH exposure is accompanied by a transcriptional shift within the distal mevalonate/cholesterol biosynthetic segment and by the lower expression of lipid droplet and β-oxidation programmes. These findings are exploratory and hypothesis-generating; they do not establish increased cholesterol synthesis, lipid accumulation, or independence from SREBP-2/HMGCR regulation, and require protein- and metabolite-level confirmation.
No takes yet. Share an insight, caveat, or question.
Chen et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: