**Background**
White adipose tissue (WAT) is a highly plastic organ that expands or shrinks in response to caloric supply and demand. Disturbed WAT remodeling leads to changes in cell composition, increasing the risk of insulin resistance, type 2 diabetes, and other cardiometabolic complications. While single-cell technologies have identified novel specialized cells in adipose tissue, most studies are limited by small cohort sizes (fewer than 15 individuals) and qualitative differences between platforms. This study aimed to create a comprehensive cellular meta-map of human WAT by integrating newly generated and publicly available single-cell (scSeq) and single-nucleus (snSeq) RNA sequencing data, and to link cell types to metabolic health using spatial transcriptomics and bulk transcriptomic data from large clinical cohorts.
**Methods**
The authors retrieved scSeq, snSeq, and spatial transcriptomic (STx) data from ten published reports and combined them with unpublished data from four additional cohorts (Massier et al. #1–4), totaling 17 datasets. These included 401,320 quality-filtered cells/nuclei from 103 samples of 83 donors (age range 22–77 years, BMI range 17–55 kg/m²). Individual datasets were processed using Seurat v4.1, and a Jaccard index was calculated to compare marker genes between studies. Data integration was performed using scANVI and four integrative frameworks (rPCA, BBKNN, Harmony, scVI), with scVI selected for final analysis based on benchmarking metrics (ARI, kBET, LISI). Cell types were annotated using marker genes and validated by flow cytometry. Spatial deconvolution of STx data was performed using cell2location and five other tools. Cell-cell communication was inferred using CellChat. To link cell types to clinical parameters, the authors deconvolved bulk transcriptomic data from eight independent studies (864 individuals) using BisqueRNA, and correlated cell type abundances with anthropometric measures, HOMA-IR, circulating lipids, leptin, fat cell volume, and lipolysis. Effects of weight loss were assessed in two bariatric surgery cohorts (n=52).
**Key Results**
The meta-map identified four major cell classes: adipocytes, fibroblast and adipogenic progenitors (FAPs), vascular cells, and immune cells. FAPs constituted the largest class (~40% of total cells), followed by adipocytes and immune cells (each ~20%), and vascular cells (~15%). Within immune cells, 11 lymphoid and 16 myeloid subtypes were identified, including novel M2-like macrophage subtypes (myC08, myC12) and redox-regulatory metabolic macrophages (Mox, myC15). Vascular cells included 12 subtypes, with vC05 expressing preadipocyte markers and vC08/vC09 showing anti-angiogenic and pro-angiogenic profiles, respectively. Subcutaneous FAPs separated into 17 clusters, with pseudo-time analysis revealing two differentiation trajectories; route 1 was recapitulated during in vitro adipogenesis, while route 2 was not. Omental FAPs showed similar signatures. Adipocyte snSeq data displayed inconsistent heterogeneity between studies, and STx data correlated better with bulk RNAseq of isolated adipocytes than snSeq. Cell-cell communication analysis suggested FAPs relay multiple signals to M2-like macrophages. Spatial deconvolution revealed that vascular and myeloid cells were concentrated in specific tissue areas, with FAPs showing distinct localization patterns (e.g., sfC12 near endothelial cells, sfC08 near LAMs). Deconvolution of bulk transcriptomic data from 864 individuals identified two clusters of cell types: cluster A (enriched for FAPs and vascular cells) associated positively with a metabolically beneficial profile, while cluster B (enriched for immune cells including LAMs, Mmes, and M2-like macrophages) associated negatively. Weight loss after bariatric surgery normalized most cell types in clusters A and B, except capillary endothelial cells (vC0).
**Clinical Implications**
This comprehensive cellular map of human WAT provides a framework for understanding how specific cell types contribute to metabolic health and disease. The identification of cell types associated with insulin resistance and obesity (e.g., LAMs, Mmes, CD55/PI16-expressing adipose precursors) and those linked to beneficial metabolic profiles (e.g., intermediate FAPs, capillary endothelial cells) may inform the development of targeted therapies for obesity and type 2 diabetes. The finding that weight loss normalizes many of these cell types underscores the dynamic nature of WAT and the potential for lifestyle or surgical interventions to reverse maladaptive cellular changes. The publicly available annotation models and deconvolution tools will facilitate future studies on WAT in diverse populations and therapeutic contexts.