Related Experiment Video
Updated: Oct 28, 2026
Adipocyte-Specific ATAC-Seq with Adipose Tissues Using Fluorescence-Activated Nucleus Sorting
Published on: March 17, 2023
Human Subcutaneous Adipose Transcriptomes Link Mitochondrial Transcription to Metabolic Heterogeneity in Obesity
Shuang Bai1, Meng Qi2,3, Shuang Wu1
1School of Kinesiology and Health, Capital University of Physical Education and Sports, Beijing 100191, China.
Background:
Metabolically healthy obesity and metabolically unhealthy obesity may occupy different positions along a continuum of adipose-tissue dysfunction, but stable molecular distinctions remain uncertain.
Methods:
We integrated three public human subcutaneous-adipose-tissue transcriptomic datasets while preserving participant and platform independence. GSE244118 was used for discovery, GSE152991 was restricted to 14 participants not overlapping with the discovery cohort, and GSE55200 provided cross-platform supportive evidence. We performed differential-expression analysis, correlation-adjusted Reactome testing, direction-aware cross-dataset integration, discovery-defined module scoring, and adipose-tissue deconvolution.
Results:
Fifty genes met conventional thresholds for metabolically unhealthy versus metabolically healthy obesity in the discovery dataset, but none passed the prespecified fold-change-aware test or strict cross-dataset replication. Among 1570 shared Reactome pathways, 409 reached false-discovery rate < 0.05 in signed-Z integration and 387 had concordant directions across all datasets. Metabolically unhealthy obesity consistently showed lower mitochondrial translation, respiratory-chain, and energy-metabolism programs, accompanied by secondary increases in extracellular-matrix remodeling. A discovery-defined mitochondrial module was lower in metabolically unhealthy obesity in synthesis restricted to non-discovery datasets under both rank and z-mean scoring. The 14-person holdout alone did not establish a significant module difference, and the extracellular-matrix module synthesis was nonsignificant. Ordered cross-sectional patterns and composition sensitivity described tissue-level associations.
Conclusions:
Metabolic obesity heterogeneity was therefore more reproducible at pathway and module levels than at the single-gene level. Unresolved clinical confounding and bulk tissue composition limit functional and causal interpretation.
