Human subcutaneous and visceral adipose tissue exhibit distinct proteomic profiles irrespective of obesity status

To investigate depot-specific proteomic differences in humans, we performed quantitative proteomic analysis using data-independent acquisition (DIA) on paired abdominal subcutaneous (ASAT) and omental visceral adipose tissue (OVAT) samples from four non-obese (BMI < 30, Subject 1–4) and four obese (BMI > 35, Subject 5–8) individuals (Fig 1A) (S1 Table). Strikingly, pairwise comparison of protein abundance between the two depots within each subject revealed a two- to 10-fold higher percentage of enriched proteins in OVAT across all individuals (Fig 1B), indicating profound biological and functional divergence between these depots, independent of obesity status. Intersection analysis of ASAT- and OVAT-enriched proteins across the four subjects of the obese and non-obese groups identified 144 (ASAT non-obese), 138 (ASAT obese), 2,823 (OVAT non-obese), and 3,115 (OVAT obese) proteins, respectively (Fig 1C). The top 20 enriched proteins for each group are listed in Fig 1D. Pathway enrichment analyses demonstrated depot-specific molecular programs that were preserved between the obese and non-obese groups (Fig 1E). ASAT-enriched proteins were associated with reactive oxygen species metabolism, cellular detoxification, and gas transport, suggesting a state of high metabolic activity that generates oxidative stress in this tissue. In contrast, OVAT-enriched proteins from both lean and obese individuals showed enrichment in transcriptional and post-transcriptional regulatory processes such as RNA splicing, ribonucleoprotein (RNP) complex biogenesis, and protein-RNA complex organization, suggesting that it maintains a biosynthetically active environment, supporting dynamic RNA metabolism and efficient protein synthesis. Notably, these depot-specific pathway signatures were maintained regardless of obesity status (Fig 1E), suggesting that the fundamental identity and function of adipose tissue is primarily defined by its anatomical location, with obesity exacerbating rather than reprogramming these innate functions.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Human subcutaneous and visceral adipose depots maintain distinct proteomic signatures across metabolic states.

(A) Schematic overview of the experimental pipeline for DIA quantitative proteomic analysis of human ASAT and OVAT. Created by hand. (B) Scatter plots comparing protein abundances in ASAT vs. OVAT for each individual. n = 4 human subjects per group. Points above/below the dashed lines denote proteins enriched in ASAT or OVAT. Non-obese and obese groups were denoted with weak and strong color intensities, respectively. (C) Venn diagrams identify overlapping enriched proteins present in all 4 obese or non-obese subjects. (D) Lollipop plots rank the top 20 depot-enriched proteins from (C). (E) GO analysis (biological process) of ASAT- and OVAT-enriched proteins. Top 20 enriched pathways were shown for each group.

https://doi.org/10.1371/journal.pbio.3003700.g001

Mouse subcutaneous and visceral ASPCs exhibit depot-specific transcriptional plasticity in response to dietary transitions

The profound proteomic differences between human ASAT and OVAT prompted us to investigate the cellular origins of this depot-specific identity. We reasoned that ASPCs, as the reservoir for new adipocytes, are likely key determinants of these distinct tissue phenotypes. Therefore, we turned to a murine model to enable a high-resolution analysis of ASPC transcriptomic plasticity in response to metabolic challenge. To investigate the transcriptomic plasticity of subcutaneous and visceral ASPCs in response to transitions between CD and HFD, we set up the following five groups of feeding paradigms in C57BL/6J male mice: mice switched from CD to HFD at 6 weeks of age for either 12 (12wHFD) or 18 weeks (18wHFD); control groups maintained on CD for equivalent durations (12wCD, 18wCD); and a reversal group subjected to 12 weeks HFD followed by 6 weeks CD (12wHFD-6wCD) (Fig 2A). At the end of this schedule (i.e., 24 weeks of age), SVFs from iWAT and eWAT were isolated for scRNA-seq analysis. A total of 86,134 cells were identified after rigorous filtering of the 10 integrated datasets (S1A Fig). Unsupervised clustering identified expected cell populations, including ASPCs, immune lineages (macrophage, B cell, T cell, dendritic cell, neutrophil, natural killer cell, mast cell), and vascular components (pericyte, endothelial cell, and mesothelial cell) (S1B and S1C Fig). Further clustering of the ASPCs revealed Dpp4+&Pi16+ progenitors (ASPC1) and Pdgfrb+&Icam1+ preadipocytes (ASPC2) (Figs 2B and S1D). This dietary shift from CD to HFD induced a proportional shift from ASPC1 to ASPC2 at both 12 and 18 weeks. The effect was most pronounced in eWAT, with iWAT showing a less obvious trend, and was reversed upon returning to a CD (Figs 2C and S2). This suggests that the balance between progenitor and preadipocyte subpopulations is a dynamic equilibrium highly sensitive to and reversible by environmental nutritional signals. At the depot level, eWAT had a higher proportion of ASPC2 than iWAT across all dietary conditions (Fig 2C), suggesting a depot-specific difference in ASPC subpopulation composition.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. Mouse ASPCs of subcutaneous and visceral WAT show distinct transcriptomic plasticity in response to dietary switches.

(A) Schematic illustration of the feeding paradigms for the five groups of C57BL/6J male mice. CD, yellow pellet; HFD, blue pellet. Created by hand. (B) Uniform Manifold Approximation and Projection (UMAP) of integrated ASPC populations from the 10 scRNA-seq datasets. (C) Pie charts quantify the proportion of ASPC1 and ASPC2 in iWAT and eWAT. (D) Venn diagrams identify overlapping HFD-responsive genes in ASPCs from 12wHFD and 18wHFD groups. (E) Heatmaps show changes in HFD-responsive gene expression level in the 18wCD, 18wHFD, and 12wHFD-6wCD groups. Z-score was used for presentation. (F) Line plots show changes in the percentage of ASPCs expressing individual HFD-responsive genes. (G) GO analysis of the HFD-responsive genes in iWAT and eWAT. Top 10 enriched pathways were shown for each group.

https://doi.org/10.1371/journal.pbio.3003700.g002

To elucidate the depot-specific molecular underpinnings of adipose tissue plasticity in response to dietary transitions, we first performed differentially expressed gene (DEG) analysis on ASPCs of iWAT and eWAT between CD- and HFD-fed cohorts at the 12- (12wCD versus 12wHFD) and 18-week time points (18wCD versus 18wHFD). Then, we performed an intersection analysis of the upregulated and downregulated gene sets from the 12wHFD and 18wHFD groups to identify conserved transcriptional changes that persisted across both feeding durations. Using stringent thresholds (detection in ≥25% of cells, |log2FC| > 0.25, adjusted p < 0.05), we identified 87 HFD-responsive genes in iWAT ASPCs (56 upregulated and 31 downregulated) (Fig 2D). Consistent with the human proteomic analysis (Fig 1B and 1C), eWAT ASPCs exhibited a more extensive transcriptional response with 258 upregulated and 282 downregulated genes, indicating greater metabolic plasticity in the visceral depot. Importantly, both the expression magnitude (Fig 2E) and cellular prevalence (Fig 2F) of these HFD-responsive genes reverted toward baseline levels in the 12wHFD-6wCD switch group of both depots, demonstrating the dynamic adaptability of ASPCs to nutritional cues. Pathway enrichment analysis among HFD-responsive genes revealed distinct signaling pathways between iWAT and eWAT (Fig 2G), suggesting depot-specific transcriptional plasticity in response to dietary transitions.

Gpc3 displays reciprocal expression dynamics in subcutaneous versus visceral WAT during metabolic adaptation in mice and humans

Among all HFD-responsive genes analyzed, Gpc3 emerged as the only gene that displayed reciprocal expression dynamics between iWAT and eWAT in response to HFD and CD transitions. In iWAT, both Gpc3 expression levels and the proportion of Gpc3-expressing ASPCs increased under HFD and decreased upon reversion to CD (Fig 3A and 3B). Conversely, eWAT demonstrated the complete inverse response pattern (Fig 3A and 3B). This depot-specific regulation was further confirmed at the mRNA level in whole-tissue analysis (Fig 3C) and the protein level in SVF (Fig 3D). Consistent with the murine data, DIA quantitative proteomic analysis of paired ASAT and OVAT from obese and lean donors revealed similar differences in GPC3 protein levels (Fig 3E). Notably, western blot analysis detected GPC3 protein in ASAT from two of the obese subjects (Subject 5 and 7) from the proteomics analysis as well as two additional obese subjects (Subject 9 and 10), with no expression observed in OVAT (Fig 3F). In addition, we found a modest positive association between GPC3 protein expression and BMI in the ASAT of the obese donor group (Fig 3G). A similar trend between Gpc3 mRNA level and body weight was seen in the iWAT of HFD-fed mice, whereas their eWAT exhibited a negative correlation (Fig 3H). Moreover, analysis of published single-nucleus RNA sequencing (snRNA-seq) datasets [12] from human subcutaneous and visceral adipose tissue further corroborated that GPC3 expression was predominantly localized to ASPCs (Fig 3I). Importantly, both fat depots exhibited reciprocal GPC3 expression patterns between lean (BMI 20–30) and obese (BMI 30–40) donors, mirroring the responses observed in mouse dietary interventions (Fig 3I). Collectively, these findings support an evolutionarily conserved, depot-specific role for Gpc3 in adaptative responses to metabolic challenges.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3. Gpc3 exhibits opposing expression patterns in subcutaneous vs. visceral WAT during metabolic adaptation in mice and humans.

(A) Bar graphs show mean Gpc3 expression in all mouse iWAT and eWAT ASPCs across dietary regimens. (B) UMAPs and bar charts show percentage of Gpc3-expressing ASPCs. (C) qPCR analysis of Gpc3 mRNA level in SVFs of iWAT and eWAT from the 18wCD, 18wHFD, and 12wHFD-6wCD groups. Mean ± SD, n = 3 animals per group, represented by a dot in the histogram. One-way ANOVA Tukey’s multiple comparison test was used to determine statistical significance. Results were normalized to the 18wCD group for better visualization of the fold changes. (D) FACS analysis of percentage of GPC3-expressing ASPCs in SVF of iWAT and eWAT from the 18wCD, 18wHFD, and 12wHFD-6wCD groups. ASPCs were identified by PDGFRA antibody. (E) Box and whiskers plots show GPC3 protein levels in ASAT and OVAT of obese and non-obese human subjects. (F) Western blots show GPC3 protein level in paired ASAT and OVAT from obese Subject 5–11. (G, H) Linear regression curves of Gpc3 protein (G) and mRNA (H) expression against BMI/body weight of human donors (G) and mice (H). Each dot represents individual human or mouse sample. Gpc3 mRNA level in (H) was assessed by qPCR. (I) Analysis of GPC3 expression in snRNA-seq datasets by Emont and colleagues [12]. Percentage below each UMAP of the ASPC population shows the proportion of ASPCs that expresses GPC3. SAT, subcutaneous adipose tissue; VAT, visceral adipose tissue. Bar charts show the mean GPC3 expression levels in all SAT and VAT ASPCs. Numerical data of (A), (C), (E), (G), (H), and (I) can be found in S1 Data, sheet “Fig 3”. Raw images of (F) can be found in S1 Raw Images.

https://doi.org/10.1371/journal.pbio.3003700.g003

ASPC-specific Gpc3 deletion selectively promotes HFD-induced iWAT expansion

To investigate the cell-autonomous role of Gpc3 in ASPCs, we obtained a mouse line harboring a conditional knockout allele of the Gpc3 gene (located on the X chromosome) with loxP sites flanking exon 3. Crossing these mice with PdgfraCre mice enabled the deletion of Gpc3 specifically in ASPCs. qPCR analysis confirmed significant reductions in Gpc3 mRNA levels in both iWAT and eWAT of PdgfraCre;Gpc3^flox mice (mutant) compared to the Gpc3^flox littermates (control) (Fig 4A). Consistently, ASPCs isolated from both depots showed comparable knockdown efficiency, with near-complete loss of Gpc3 expression (Fig 4B). Next, we subjected 6-week-old control and mutant mice to an 18-week HFD regimen. Mutant mice showed significantly greater weight gain and fat mass accumulation compared to both control and PdgfraCre mice (Fig 4C). At the tissue level, the wet weight of iWAT but not eWAT was significantly increased in the mutant mice (Fig 4D). Consistent with this observation, iWAT adipocytes in mutant mice were significantly larger in size, whereas eWAT adipocyte size remained comparable between groups (Fig 4E). Notably, control and mutant mice showed no differences in glucose tolerance, insulin sensitivity, respiration, energy expenditure, food intake, and ambulation (Figs 4F, 4G, and S3A), suggesting that the mutation primarily affects local adipose tissue expansion rather than systemic metabolic regulation. To identify molecular pathways driving depot-specific adipose remodeling, we performed quantitative DIA proteomics on iWAT and eWAT from HFD-fed control and mutant mice. We found that most of the pathways that are known to regulate adipose tissue expansion, including lipid uptake and storage, lipogenesis, adipogenesis, inflammation, extracellular matrix (ECM) remodeling, autophagy and lysosomal function, and hormonal regulation were upregulated in the iWAT of mutant mice (Fig 4H). In contrast, these pathways were either unchanged or downregulated in eWAT (Fig 4H), highlighting a fundamental depot-specific role for Gpc3 that promotes metabolically favorable remodeling in iWAT. Together, these results demonstrate that Gpc3 deletion in ASPCs drives depot-specific adipose expansion under HFD-feeding by selectively affecting iWAT homeostasis, possibly through coordinated upregulation of lipid metabolic, inflammatory, and remodeling pathways. snRNA-seq of iWAT from HFD-fed mice confirmed the efficacy of the genetic model, revealing a 3-fold reduction in the proportion of cells expressing Gpc3 (predominantly ASPCs) in mutants compared to controls (Fig 4I). While Gpc3 deletion did not drastically alter the proportional distribution of the two major ASPC subpopulations (Fig 4J), it profoundly altered their transcriptional programs as reflected by a shift in the top 5 enriched pathways within both populations (Fig 4K). Specifically, ASPC1 transitioned from a thermogenic profile in controls to an immunomodulatory state in mutants, whereas ASPC2 shifted its enrichment from ECM organization to pathways governing translational regulation (Fig 4K).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 4. Targeted deletion of Gpc3 in ASPCs selectively enhances iWAT expansion under HFD feeding condition.

(A) Gpc3 mRNA levels in iWAT and eWAT of control and mutant mice. Mean ± SD, n = 3 for both groups, represented by a dot in the histogram. Unpaired t test was used to determine statistical significance. Results were normalized to the control group for better visualization of the fold changes. (B) Gpc3 mRNA levels in ASPCs isolated from iWAT and eWAT of control and mutant mice. Mean ± SD, n = 8 independent experiments for both groups, represented by a dot in the histogram. Unpaired t test was used to determine statistical significance. Results were normalized to the control group for better visualization of the fold changes. (C) Body weight and fat mass (assessed by a small animal body composition analyzer) curves of control and mutant mice during HFD feeding. Mean ± SEM, n = 8, 13, and 19 for the PdgfraCre, Gpc3^flox and PdgfraCre;Gpc3^flox groups, respectively. Two-way ANOVA was used to determine statistical significance. p-values indicated in the graphs are for the independent variables. p-values for the interaction are 0.0663 (B) and 0.1211 (C), respectively. (D) Wet weights of iWAT and eWAT of control and mutant mice by the end of HFD feeding. Mean ± SD, n = 6–10 animals, represented by a dot in the histogram. Unpaired t test was used to determine statistical significance. (E) HE staining images and size-based distribution of adipocytes in iWAT and eWAT of control and mutant mice during CD feeding. Mean ± SD, n = 7–9 animals, represented by a dot in the histogram. Two-way ANOVA Sidak’s multiple comparison test was used to determine statistical significance. All reported p-values compared the values of Gpc3^flox and PdgfraCre;Gpc3^flox groups within each adipocyte size category. (F) Glucose and insulin tolerance tests in HFD-fed control and mutant mice. Mean ± SD, n = 6 and 4 for control and mutant groups, respectively. (G) Respiratory exchange ratio (RER) and energy expenditure in HFD-fed control and mutant mice. Mean ± SEM, n = 10 and 7 for control and mutant groups, respectively. Energy expenditure was analyzed by the ANCOVA test using body weight as a covariate (p = 0.76). (H) ssGSEA scores of pathways related to adipose expansion in iWAT and eWAT. (I) UMAPs show Gpc3 expression in all cell populations of iWAT from Gpc3^flox and PdgfraCre;Gpc3^flox groups. (J) UMAPs and composition of ASPC1 and 2 in Gpc3^flox and PdgfraCre;Gpc3^flox iWAT. (K) GO analysis of the top five pathways that are enriched in ASPC1 and 2 of Gpc3^flox and PdgfraCre;Gpc3^flox iWAT. Numerical data of (A), (B), (C), (D), (E), (F), and (G) can be found in S1 Data, sheet “Fig 4”.

https://doi.org/10.1371/journal.pbio.3003700.g004

Of note, Gpc3 mRNA levels showed a decreasing trend in the brain, but not in the pancreas, of mutant mice (S3B Fig). To determine if brain Gpc3 contributes to the phenotype, we generated Camk2aCre;Gpc3^flox mice to delete Gpc3 specifically in forebrain neurons. On the same HFD regimen, these mice showed no differences in body weight or fat mass compared to Gpc3^flox controls (S3C Fig), indicating that brain Gpc3 does not play a major role in the observed obesity. Consistent with its location on the X chromosome, qPCR analysis revealed that Gpc3 mRNA abundance was significantly higher in both iWAT and eWAT from wild-type female mice compared to their male counterparts at 4 months of age (S3D Fig), suggesting that Gpc3 is a candidate escape gene from X-inactivation. It is worth noting that the expression of Mir717, an obesity associated microRNA [21] encoded within the third intron of the Gpc3 gene, was unaffected in our model (S3E Fig).

Gpc3 constrains iWAT expansion in a cell-autonomous manner during HFD feeding

To investigate the cellular mechanisms underlying Gpc3-mediate regulation of iWAT and eWAT expansion, we crossed PdgfraCreER^T (control) and PdgfraCreER^T;Gpc3^flox (mutant) mice with an mTmG reporter strain, which labels Cre-expressing cells and their progeny with GFP while non-recombined cells express tdTomato [22], enabling lineage trace the ASPCs for quantitative analysis. Six-week-old mice received five consecutive daily tamoxifen injections, which efficiently reduced Gpc3 mRNA in both iWAT and eWAT (S4A Fig), and then were placed on a 7-week HFD regimen (Fig 5A). In line with the non-inducible deletion model, mutant mice exhibited significantly greater weight gain than controls (Fig 5B). Strikingly, iWAT from mutant mice showed a significantly higher proportion of GFP+ traced adipocytes derived from ASPCs, along with reduced adipocyte size (Fig 5C), which could be due to the enhanced differentiation potential in Gpc3-deficient ASPCs. In addition, Gpc3-deficient (GFP+) adipocytes in the mutant iWAT were significantly smaller than the control (Tomato+) adipocytes within the same tissue microenvironment (S4B Fig). In contrast, eWAT displayed no differences in the proportion or size of traced adipocytes between genotypes (Fig 5C), consistent with a depot-specific role for Gpc3 in iWAT under HFD. Notably, non-traced (Tomato+) adipocytes were unaffected in both depots (Fig 5C), demonstrating that Gpc3 acts in a cell-autonomous manner to constrain tissue expansion. When using the same injection protocol while maintaining the mice on CD (Fig 5D), an increasing trend in body weight gain was also observed in mutant mice, though statistical significance was not reached (Fig 5E). The only difference from the HFD condition was that mutant mice showed significantly higher proportions of GFP+ traced adipocytes in both iWAT and eWAT (Fig 5F), suggesting a diet-dependent role of Gpc3 in regulating eWAT expansion. Of note, the actual proportions of GFP+ traced adipocytes were higher in iWAT (average 2.5% versus 1.5%) and eWAT (average 6% versus 2.5%) of HFD-fed control mice compared with CD condition, despite their visual abundance in the CD-fed mice due to smaller cell size, which is in line with a previous study showing that HFD-animals have higher rates of adipogenesis [23].

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 5. Gpc3 cell-autonomously restricts iWAT expansion during HFD feeding.

(A, D) Schematic illustration of the experimental setup for HFD (A) and CD (D) feeding conditions. Created by hand. (B, E) Body weight gain of control and mutant mice during HFD (B) and CD (E) feeding. Mean ± SEM, n = 6, 4, 5, and 8 for the HFD and CD PdgfraCreER^T;mTmG and PdgfraCreER^T;Gpc3^flox;mTmG groups, respectively. Two-way ANOVA was used to determine statistical significance. p-values for the interaction are 0.6333 (B) and 0.0773 (E), respectively. (C, F) Representative confocal images and quantification of adipocyte labeling in HFD (C) and CD (F)-fed iWAT and eWAT. Mean ± SD, n = 3-6 animals per group, represented by a dot in the histogram. Unpaired t test and Two-way ANOVA Sidak’s multiple comparison test were used to determine statistical significance. All reported p-values for the Two-way ANOVA Sidak’s multiple comparison test compared the values of PdgfraCreER^T;mTmG and PdgfraCreER^T;Gpc3^flox;mTmG groups within each adipocyte size category. Small and large adipocytes were determined based on the median size in each comparison. HFD iWAT, small: <7,000 µm², large: ≥ 7,000 µm²; HFD eWAT, small: <10,000 µm², large: ≥10,000 µm²; CD iWAT, small: <1,500 µm², large: ≥1,500 µm²; CD eWAT, small: <3,000 µm², large: ≥ 3,000 µm². Numerical data of (B), (C), (E), and (F) can be found in S1 Data, sheet “Fig 5”.

https://doi.org/10.1371/journal.pbio.3003700.g005

Gpc3-deficient ASPCs in HFD-fed iWAT show enhanced differentiation potential and reduced proliferative capacity

To investigate the mechanisms underlying enhanced hyperplasia in Gpc3-deficient ASPCs, we isolated ASPCs from iWAT and eWAT of HFD-fed control and mutant mice for in vitro primary culture. By days in vitro (DIV) 7, mutant iWAT-derived adipocytes exhibited significantly greater Oil Red O-stained lipid accumulation (Fig 6A), along with increased expression of adipogenic (Pparg, Cebpa) and lipogenic (Dgat2) markers during differentiation (Fig 6B). These changes were accompanied by reduced BrdU incorporation (Fig 6C) and Ccnd1 expression (Fig 6D), indicating impaired proliferative capacity. In contrast, eWAT-derived adipocytes from HFD-fed mice showed no differences in lipid accumulation (Fig 6E) or proliferative capacity (Fig 6G and 6H), despite elevated expression of adipogenic and lipogenic markers (Fig 6F). Under CD condition, control and mutant iWAT-derived adipocytes showed comparable differentiation capacity (Fig 6I) and similar Pparg, but not Cebpa or Dgat2 expression (Fig 6J). Notably, mutant eWAT-derived adipocytes showed reduced differentiation capacity (Fig 6K) with decreased expression of adipogenic and lipogenic markers (Fig 6L). Together, these data demonstrate that Gpc3 deletion in ASPCs selectively enhances adipogenic potential while suppressing proliferation in iWAT under HFD condition.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 6. Gpc3 deficiency promotes differentiation while suppressing proliferation in iWAT ASPCs under HFD condition.

(A, E, I, K) Oil Red O staining and quantification of iWAT (A, I) and eWAT (E, K) ASPC-derived adipocytes from HFD- (A, E) or CD-fed (I, K) control and mutant mice on DIV7. Mean ± SD, n = 3 independent experiments, represented by a dot in the histogram. Unpaired t test was used to determine statistical significance. Results were normalized to the control group for better visualization of the fold changes. (B, F, J, L) Pparg, Cebpa, and Dgat2 mRNA levels of control and mutant mice during HFD- (B, F) or CD-fed (J, L) iWAT (B, J) and eWAT (F, L) ASPC differentiation on DIV0, 4, and 7. Mean ± SD, n = 3 independent experiments. Two-way ANOVA Sidak’s multiple comparison test was used to determine statistical significance. All reported p-values compared the values of Gpc3^flox and PdgfraCre;Gpc3^flox groups at each time point. Results were normalized to the DIV0 group for better visualization of the fold changes. (C, G) BrdU incorporation in iWAT (C) and eWAT (G) ASPC from HFD-fed control and mutant mice on DIV2. Cells were exposed to BrdU during DIV0-2. Mean ± SD, n = 3 independent experiments, represented by a dot in the histogram. Unpaired t test was used to determine statistical significance. Results were normalized to the control group for better visualization of the fold changes. (D, H) Ccnd1 mRNA levels in iWAT (D) and eWAT (H) ASPC from HFD-fed control and mutant mice on DIV2. Mean ± SD, n = 3 independent experiments, represented by a dot in the histogram. Unpaired t test was used to determine statistical significance. Results were normalized to the control group for better visualization of the fold changes. Numerical data of (A), (B), (C), (D), (E), (F), (G), (H), (I), (J), (K), and (L) can be found in S1 Data, sheet “Fig 6”.

https://doi.org/10.1371/journal.pbio.3003700.g006

Gpc3 governs iWAT differentiation through canonical Wnt signaling

To elucidate the molecular mechanisms underlying the iWAT-specific effects of Gpc3, we performed comparative pathway enrichment analysis of our mouse proteomic data, focusing on pathways known to be regulated by GPC3. We found that the Wnt signaling pathway was specifically downregulated in mutant iWAT (Fig 7A), consistent with Gpc3’s role as a Wnt activator in hepatocellular carcinoma [24,25]. In contrast, other pathways, including Hedgehog, Hippo, and fibroblast growth factor signaling remained unaffected or exhibited alterations only in mutant eWAT (Fig 7A). Supporting this observation, Wnt-related proteins, including total β-catenin (CTNNB1) were predominantly upregulated in mutant iWAT, while no consistent pattern was observed in eWAT (Fig 7B) (S2 Table). In adipocytes derived from iWAT ASPCs, both active and total β-catenin protein levels were significantly reduced in HFD-fed PdgfraCre;Gpc3^flox (mutant) mice compared to Gpc3^flox (control) mice (Fig 7C). In our human proteomic data, the WNT signaling pathway was upregulated in ASAT compared to OVAT under obese, but not lean condition (Fig 7D). In addition, ASAT from obese patients showed higher WNT signaling activity than that from lean patients, whereas no difference was observed in OVAT (Fig 7D).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 7. iWAT-specific Wnt pathway dysregulation in Gpc3 mutant mice.

(A) ssGSEA scores of pathways regulated by GPC3 in iWAT and eWAT of HFD-fed control and mutant mice. (B) Heatmaps show changes in Wnt-related protein levels. Z-score was used for presentation. (C) Western blots show active non-phospho and total β-catenin protein levels in adipocytes derived from iWAT ASPCs of HFD-fed control and mutant mice. Mean ± SD, n = 3 independent experiments, represented by a dot in the histogram. Unpaired t test was used to determine statistical significance. Results were normalized to the control group for better visualization of the fold changes. (D) ssGSEA scores of WNT signaling pathway from human proteomic data. Numerical data of (C) can be found in S1 Data, sheet “Fig 7”. Raw images of (C) can be found in S1 Raw Images.

https://doi.org/10.1371/journal.pbio.3003700.g007

To determine whether the effects of Gpc3 on ASPC differentiation are Wnt-dependent, we knocked down Axin1, the gene that encodes the scaffold protein essential for β-catenin phosphorylation and degradation, in iWAT ASPCs from HFD-fed control and mutant mice. Successful Axin1 knockdown (Fig 8A) increased active non-phospho β-catenin levels (Fig 8B) and reversed the enhanced adipogenic potential of mutant ASPCs (Fig 8C). Two-way ANOVA revealed a significant genotype-by-treatment interaction for adipogenesis (Fig 8C), suggesting Gpc3 influences the functional state of the destruction complex. Activation of Wnt signaling with either recombinant Wnt3a (rWnt3a) protein (Fig 8D and 8E) or CHIR99021 (a GSK3 inhibitor that stabilizes β-catenin) (Fig 8F and 8G) increased active non-phospho β-catenin levels and suppressed adipogenesis in both control and mutant ASPCs. Conversely, inhibition of Wnt signaling with XAV939 (a tankyrase inhibitor that promotes β-catenin degradation) in control and mutant ASPCs had the opposite effects (Fig 8H and 8I). The genotype-by-treatment interaction terms for adipogenesis of these three treatments were not significant (Fig 8E, 8G, and 8I), suggesting these agents act in parallel with Gpc3. Collectively, these findings establish Gpc3 as a major, though not exclusive, regulator of the canonical Wnt pathway in ASPCs under HFD feeding condition (Fig 8J). Notably, the observed reduction in mutant iWAT ASPC proliferation (Fig 6C) may reflect a trade-off with their enhanced differentiation capacity.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 8. Gpc3 regulates iWAT differentiation via canonical Wnt Signaling.

(A) Axin1 mRNA levels of HFD-fed iWAT ASPCs on DIV7 from control and mutant mice. Mean ± SD, n = 3 independent experiments, represented by a dot in the histogram. Unpaired t test was used to determine statistical significance. (B) Western blots show active non-phospho β-catenin protein levels of scrambled and Axin1 siRNA-treated HFD-fed iWAT ASPCs from control and mutant mice on DIV7. siRNA was applied during DIV0-2. Mean ± SD, n = 3 independent experiments, represented by a dot in the histogram. Unpaired t test was used to determine statistical significance. Results were normalized to the scrambled siRNA group for better visualization of the fold changes. (C) Oil Red O staining and quantification of scrambled and Axin1 siRNA-treated HFD-fed iWAT ASPCs from control and mutant mice on DIV7. Mean ± SD, n = 3 independent experiments, represented by a dot in the histogram. Two-way ANOVA Sidak’s multiple comparison test was used to determine statistical significance. Results were normalized to the vehicle-treated control group for better visualization of the fold changes. (D) Western blots show active non-phospho β-catenin protein levels in vehicle (PBS with 1% BSA)- and rWnt3a-treated iWAT ASPCs from HFD-fed mutant mice on DIV7. rWnt3a treatment was applied during DIV0-3. Mean ± SD, n = 3 independent experiments, represented by a dot in the histogram. Unpaired t test was used to determine statistical significance. Results were normalized to the vehicle group for better visualization of the fold changes. (E) Oil Red O staining and quantification of vehicle- and rWnt3a-treated iWAT ASPCs from HFD-fed control and mutant mice on DIV7. Mean ± SD, n = 3 independent experiments, represented by a dot in the histogram. Two-way ANOVA Sidak’s multiple comparison test was used to determine statistical significance. Results were normalized to the vehicle-treated control group for better visualization of the fold changes. (F) Western blots show active non-phospho β-catenin protein levels in vehicle (0.004% DMSO)- and CHIR99021-treated iWAT ASPCs from HFD-fed control and mutant mice on DIV7. CHIR99021 treatment was applied during DIV0-7. Mean ± SD, n = 3 independent experiments, represented by a dot in the histogram. Unpaired t test was used to determine statistical significance. Results were normalized to the vehicle group for better visualization of the fold changes. (G) Oil Red O staining and quantification of vehicle- and CHIR99021-treated iWAT ASPCs from HFD-fed control and mutant mice on DIV7. Mean ± SD, n = 3 independent experiments, represented by a dot in the histogram. Two-way ANOVA Sidak’s multiple comparison test was used to determine statistical significance. Results were normalized to the vehicle-treated control group for better visualization of the fold changes. (H) Western blots show β-catenin protein levels in vehicle (DMSO)- and XAV939 (0.4 µM)-treated iWAT ASPCs from HFD-fed control and mutant mice on DIV7. XAV939 treatment was applied during DIV0-3. Mean ± SD, n = 3 independent experiments, represented by a dot in the histogram. Unpaired t test was used to determine statistical significance. Results were normalized to the vehicle group for better visualization of the fold changes. (I) Oil Red O staining and quantification of vehicle- and XAV939-treated iWAT ASPCs from HFD-fed control and mutant mice on DIV7. Mean ± SD, n = 3 independent experiments, represented by a dot in the histogram. Unpaired t test was used to determine statistical significance. Results were normalized to the vehicle-treated control group for better visualization of the fold changes. (J) Schematic of the depot-specific regulation of adipogenesis by Gpc3 via the canonical Wnt pathway. Created by hand. Numerical data of (A), (B), (C), (D), (E), (F), (G), (H), and (I) can be found in S1 Data, sheet “Fig 8”. Raw images of (B), (D), (F), and (H) can be found in S1 Raw Images.

https://doi.org/10.1371/journal.pbio.3003700.g008

Human adipose tissue samples

Paired ASAT and OVAT collection was approved by the Peking University Third Hospital Medical Science Research Ethics Committee (M2024238) and performed in compliance with all relevant ethical regulations, with written informed consent obtained from all participants. The study was conducted in accordance with the principles expressed in the Declaration of Helsinki. General information of the participants was listed in S1 Table. Excess adipose tissue was collected during standard surgical procedures, with no additional interventions required.

Mouse adipose tissue samples

All animal procedures were conducted in accordance with protocols approved by the Institutional Animal Care and Use Committees (IACUC) of Peking University (Protocol #Psych-XieM-2) and the Chinese Institute for Brain Research (Protocol #CIBR-IACUC-035), ensuring full compliance with ethical guidelines for animal research. The study was conducted in accordance with the Guidelines for the ethical review of laboratory animal welfare People’s Republic of China National Standard GB/T 35892‐2018. Gpc3^flox mice (T052331) were obtained from GemPharmatech (Nanjing, China). C57BL/6J (000664), PdgfraCre (013148), PdgfraCreER^T (018280), and mTmG (007676) mice were obtained from the Jackson Laboratory. Six-week-old mice were fed with CD (1010097, Jiangsu Xietong Pharmaceutical Bio-engineering Co., Jiangsu, China) or 60% HFD (D12492, Research Diet). All mice were maintained in a temperature-controlled environment with regulated humidity, standard 12:12 hour light/dark cycle, and free access to food and purified water. For genetic lineage tracing, tamoxifen (Sigma-Aldrich, T5648) was prepared in corn oil (Solarbio, C7030) and administered via intraperitoneal injection at 2 mg/day for 5 consecutive days. Fat mass was measured using a small animal body composition analyzer (QMR12-060H-I, Niumag Analytical Instrument Corporation, Suzhou, China). Male mice ranging from 6 to 24 weeks of age were used in the present study.

Human and mouse adipose sample preparation for DIA quantitative proteomics and data analysis

Human and mouse adipose tissues were snap-frozen in liquid nitrogen post-excision and stored at −80 °C. Tissues were pulverized in liquid nitrogen and lysed in SDT buffer (4% SDS, 100 mM Tris-HCl, pH 7.6, 100 mM NaCl, 100 mM DTT). Lysates were sonicated on ice for 5 min, heated at 95 °C for 10 min, and centrifuged at 12,000g for 15 min at 4 °C. Supernatants were alkylated with 100 mM iodoacetamide (Sigma-Aldrich) in the dark at room temperature (RT) for 1 hour. Proteins were precipitated with four volumes of pre-chilled acetone at −20 °C for 2 h, pelleted at 12,000g for 15 min at 4 °C, washed with ice-cold acetone, and dissolved in 8 M urea with 100 mM triethylammonium bicarbonate (pH 8.5; Sigma-Aldrich). Protein concentrations were measured using the Bradford assay (Bio-Rad) with bovine serum albumin as standard. Proteins (20 µg) were digested with trypsin (Promega) at a 1:50 ratio in 100 mM triethylammonium bicarbonate at 37 °C for 4 h, then overnight. Peptides were acidified to pH < 3 with formic acid, desalted on C18 Sep-Pak columns (Waters), eluted in 70% acetonitrile with 0.1% formic acid, and lyophilized. Peptides were reconstituted in 0.1% formic acid and analyzed by DIA on a Thermo Fisher Orbitrap Astral mass spectrometer with a Vanquish Neo UHPLC system. Separation used a PepMap Neo C18 column (150 µm × 15 cm, 2 µm; Thermo Fisher Scientific) with a C18 pre-column (5 mm × 300, 5 µm) at 50 °C, using a 22-min gradient (4%–99% mobile phase B: 80% acetonitrile, 0.1% formic acid). The mass spectrometer operated in DIA mode (spray voltage 1.9 kV, ion transfer tube 290 °C), scanning precursor ions (m/z 380–980) at 240,000 resolution and fragment ions (m/z 150–2000) at 80,000 resolution, with 300 DIA windows, 25% normalized collision energy, and 3 ms maximum injection time.

Raw DIA data were processed using DIA-NN (v1.8) against a UniProt database, with 10 ppm precursor and 0.02 Da fragment mass tolerances, cysteine alkylation, and up to one missed cleavage. Peptide Spectrum Matches were filtered at 99% confidence, with FDR <0.01). Protein quantification data from adipose tissue samples were imported from a.csv file and processed to remove duplicate gene entries, retaining unique gene identifiers. All intensity values were transformed using log2(x + 1) to stabilize variance and normalize the data distribution. For paired human sample analysis, protein abundance differences between ASAT and OVAT were evaluated using data from four lean and four obese individuals. Protein enrichment in ASAT or OVAT was determined based on the direction and magnitude of expression differences between paired samples. For comparisons between lean and obese individuals in humans, and between CD and HFD groups in mice, statistical significance was assessed using unpaired two-sample t-tests. Analyses were restricted to proteins with sufficient non-missing values and non-zero variance. Differentially abundant proteins were annotated based on fold change and p-value thresholds, and categorized as enriched in the respective groups. Differentially enriched proteins were subjected to Gene Ontology (GO) enrichment analysis. Gene symbols were mapped to Entrez Gene IDs using a standard annotation database. Enrichment analysis for biological process terms was performed using inclusive p-value and q-value thresholds to ensure comprehensive coverage of significant terms. The top enriched GO terms were visualized in a dot plot to highlight key biological processes. Single-sample Gene Set Enrichment Analysis (ssGSEA) was performed to quantify pathway activity scores across individual samples. Enrichment scores were visualized as boxplots stratified by group.

The human proteomic study was designed as a targeted, hypothesis-generating investigation to identify the most robust and large-magnitude proteomic differences, rather than a fully powered, population-level discovery effort. We acknowledge that our sample size (n = 4 per group) presents a challenge for statistical power; however, this design is strategically focused on uncovering only the most pronounced biological signals, which are more likely to be biologically relevant and less likely to be false positives, even within a smaller cohort.

Mouse SVF cell isolation

iWAT and eWAT were dissected, washed twice with HBSS, then minced and digested in 1 mg/ml collagenase type I (Sigma-Aldrich, C0130) dissolved in digestion buffer (0.1 M HEPES-Na, 120 mM NaCl, 50 mM KCl, 5 mM D-glucose, 1 mM CaCl₂, 1.5% BSA) for 45 min at 37°C with continuous shaking (150 rpm). Digestion was quenched by adding an equal volume of growth medium (Advanced DMEM/F-12 [Gibco, 12634028] supplemented with 10% FBS, 1% GlutaMAX [Thermo, 35050061], 1% MEM NEAA [Thermo, 11140050], and 1% penicillin-streptomycin [Gibco, 15140122]). The suspension was sequentially filtered through 100-μm and 40-μm cell strainers (Fisher Scientific), with centrifugation (450g for 10 min) between steps. Erythrocyte lysis was performed by resuspending the pellet in red blood cell lysis buffer (5 min, 37°C), followed by neutralization with growth medium and final filtration with the 40-μm strainer.

Nuclei isolation for snRNA-seq

Following dissection, iWAT was minced with scissors in 500 µl of ice-cold Nuclei Isolation Buffer, which consists of 250 mM sucrose, 10 mM HEPES, 1.5 mM MgCl₂, 10 mM KCl, 0.001% NP-40, 0.2 mM DTT, and 1 U/µl RNase inhibitor in DEPC-treated water. The minced tissue was then homogenized using a 2 ml Dounce homogenizer and filtered through a 70-µm cell strainer. The homogenate was centrifuged at 500g for 5 min at 4 °C to pellet the nuclei. The resulting nuclei pellet was resuspended in a Nuclei Resuspension Buffer containing 2% BSA, 1.5 mM MgCl₂, and 1 U/µl RNase inhibitor in PBS. All buffers were sterile-filtered before use.

scRNA-seq and snRNA-seq library preparation and sequencing

Isolated nuclei were stained with 10 µg/ml 7-AAD (Invitrogen, A1310) and sorted using a BD FACSMelody4-Way Cell Sorter (BD Biosciences) equipped with a 100-µm nozzle. Isolated cells were stained with 7-AAD (5 μg/ml) for viability assessment prior to FACS sorting. Cell and nuclei concentration was determined using a Countstar Rigel S5 Automated Cell Counter. For single-cell/nucleus capture, approximately 20,000 viable cells/nuclei were loaded onto a 10× Genomics Chromium Controller using the Single Cell 3′ Reagent Kit v3.1 (10× Genomics) with 3′ v3 chemistry. GEM-RT was conducted at 53 °C for 45 min followed by 85 °C for 5 min, with a 4°C hold. cDNA was recovered through emulsion lysis and purified using DynaBeads with SPRIselect reagent (Thermo Scientific), followed by PCR amplification. Library preparation involved enzymatic fragmentation, size selection, and sequential addition of P5/P7 adapters and i5/i7 indexes via end repair, A-tailing, and ligation. Final libraries were quantified and pooled for 150 bp paired-end sequencing on an Illumina NovaSeq 6000, targeting ~50,000 raw reads per cell/nucleus. cDNA quality was verified at each stage using an Agilent 2100 Bioanalyzer.

scRNA-seq and snRNA-seq data analysis

Raw FASTQ files were aligned to the mm10 reference genome (Ensembl annotation) using Cell Ranger 7.0.0 with intronic reads included. Initial gene count matrices were processed through CellBender 0.2.2 to remove ambient RNA contamination and empty droplets [46]. Subsequent quality control was performed in Seurat 4.3.0 [47], applying the following filtration criteria: 200−7,500 detected genes, 500−75,000 UMIs, and <10% mitochondrial RNA content. DoubletFinder 2.0.3 was used to exclude cell doublets [48]. To address sequencing depth heterogeneity, we performed comprehensive data preprocessing. Initial normalization was conducted using the LogNormalize method, scaling each cell’s gene counts by its total UMI count (factor = 10,000) followed by natural log transformation. Highly variable genes were identified through variance stabilizing transformation with mean-variance modeling. Expression values were subsequently scaled and centered using the ScaleData function (mean = 0, standard deviation = 1). We also applied sctransform to regularize the data using a negative binomial model, reducing technical artifacts while preserving biological variation [49,50]. To integrate multiple datasets and address batch effects, we applied the Harmony algorithm to analyze and adjust principal components, aligning data across different batches. This approach effectively reduced technical variations caused by experimental conditions and processing times while maintaining biological heterogeneity. The method enabled robust integration of cells from diverse datasets while preserving biologically relevant differences [6,51]. DEGs were identified using Seurat’s “FindMarkers” function with the following thresholds: |log2FC| > 0.25, adjusted p-value <0.05, and expression in >25% of cells. Cell type annotation was performed by integrating biological knowledge, established cell-type-specific markers, and top marker genes from the PanglaoDB database. The same “FindMarkers” function was applied to detect DEGs across different experimental conditions. Gene set enrichment analysis was subsequently performed using the ‘gsva’ function from the GSVA package to compute enrichment scores for each sample.

RNA isolation and qPCR analysis

RNA was extracted with the TRIzol reagent (Invitrogen, 15596018). For cDNA synthesis, 1 μg of total RNA was reverse transcribed using the H Minus First Strand cDNA Synthesis Kit (Thermo Scientific, K1652). Quantitative PCR was performed using PowerUp SYBR Green Master Mix (Applied Biosystems, A25742) on a CFX96 Real-Time PCR Detection System (Bio-Rad). Gene expression levels were normalized to Ywhaz reference gene using the ΔΔCt method. For Mir717 quantification, total RNA was reverse-transcribed with a kit using a miRNA-specific stem-loop primer and a U6 snRNA primer (Servicebio). The resulting cDNA was subjected to qPCR with SYBR Green Master Mix and Mir717-specific primers. All samples were normalized to U6 snRNA.

The following mouse-specific primer sequences were used:

1. Gpc3: 5′-GGAGCAAGACGTGACCTGAA-3′, 5′-GCATACGGCCACAGTCCTTA-3′
2. Pparg: 5′-GTACTGTCGGTTTCAGAAGTGCC-3′, 5′-ATCTCCGCCAACAGCTTCTCCT-3′
3. Cebpa: 5′-CAAGAACAGCAACGAGTACCG-3′, 5′-GTCACTGGTCAACTCCAGCAC-3′
4. Dgat2: 5′- CTGTGCTCTACTTCACCTGGCT-3′, 5′- CTGGATGGGAAAGTAGTCTCGG-3′
5. Ccnd1: 5′-AACTACCTGGACCGCTTCCT-3′, 5′- CCACTTGAGCTTGTTCACCA-3′
6. Ywhaz: 5′-CAGTAGATGGAGAAAGATTTGC-3′, 5′-GGGACAATTAGGGAAGTAAGT-3′
7. Mir717 forward: 5′-ACACCTCAGCTGGGCTCAGACAGAGATACC-3′
8. Mir717 stem-loop RT primer: 5′-CTCAACTGGTGTCGTGGAGTCGGCAATTCAGTTGAGAGAGAGG-3′
9. Universal reverse primer: 5′-TGGTGTCGTGGAGTCG-3′
10. U6 forward primer:5′-CTCGCTTCGGCAGCACA-3′
11. U6 reverse primer: 5′-AACGCTTCACGAATTTGCGT-3′

FACS

SVF cells isolated from iWAT and eWAT of C57BL/6J mice from the three dietary groups were resuspended in 2% FBS/PBS containing anti-mouse CD16/CD32 Fc Block (clone 2.4G2, BD) and incubated on ice for 20 min. Cells were then incubated with CD140a (PDGFRA) Antibody-APC (Thermo Fisher, 17-1401-81) and GPC3 (Thermo, MA5-17083) antibody while rotating at 4 °C for 30 min. Following incubation, cells were washed three times with 2% FBS/PBS and stained with 7-AAD to exclude dead cells. Samples were then subjected to cell analysis and sorting using a Beckman Coulter MoFlo XDP flow cytometer. The samples were loaded and run at a set flow rate, while the instrument’s threshold and scatter parameters were adjusted to identify individual cells. Forward scatter (FSC) and side scatter were used to assess cell size and internal complexity. Doublets were identified and excluded by comparing the height and width parameters of FSC.

Western blot

Protein extraction was performed by adding 0.3 mL of 100% ethanol to the interphase and lower organic phase obtained from TRIzol lysates, followed by centrifugation at 2,000g for 5 min at 4 °C. The upper phase was collected and proteins were precipitated with isopropanol (12,000g, 10 min, 4 °C). The resulting protein pellet was washed sequentially with: (1) 0.3 M guanidine hydrochloride (Solarbio, G8070) in 95% ethanol (two washes), (2) 100% ethanol (one wash), and then air-dried. Proteins were denatured in 200 μL of 1% SDS (Solarbio, S8010) at 50°C. Protein concentration was determined using a BCA assay kit (Solarbio, PC0020). For immunoblotting, 15 μg of protein was resolved by SDS-PAGE and transferred to PVDF membranes (Bio-Rad, 1620177). Membranes were developed with chemiluminescent HRP substrate (Millipore, WBKLS05000) and visualized using an Amersham ImageQuant 800 system. Band intensity quantification was performed using ImageJ software (v1.53k). The following antibodies were used: GPC3 (Santa Cruz, sc-390587), non-phospho β-Catenin (CST, #8814), total β-Catenin (CST, #9562), GAPDH (CST, #2118).

SVF cell primary culture for adipogenic differentiation

Isolated cells were resuspended in basal medium and plated in culture dishes. Prior to plating, viable cell numbers were quantified using a hemocytometer (Neubauer chamber) with trypan blue exclusion. Cells were then seeded in 12-well plates at 7.5 × 10⁴ cells/mL in a final volume of 2 mL per well. Upon reaching ~90% confluency, differentiation was initiated by replacing the medium with differentiation cocktail containing: advanced DMEM/F12, 5% FBS, 1% GlutaMAX, 1% NEAA, 1% P/S, 1% ITS (Gibco, 41400045), 33 μM d-biotin (Sigma-Aldrich, B4639), 17 μM pantothenate (Sigma-Aldrich, P5155), 0.5 mM IBMX (Sigma-Aldrich, I5879), 1 μM dexamethasone (Sigma-Aldrich, D4902), 1 μM rosiglitazone (Sigma-Aldrich, R2408), and 2 nM T3 (Sigma-Aldrich, T6397). After 4 days, cells were switched to maintenance medium (advanced DMEM/F12 with 2% FBS, 33 μM d-biotin, 17 μM pantothenate, 1 μM dexamethasone, 10 μg/mL insulin, 2 mM GlutaMAX, 0.1 mM NEAA, and 100 U/mL P/S) for 3 days. Cells were collected on DIV 0, 4, and 7 for analysis. GSK3 inhibitor CHIR99021 (MedChemExpress, HY-10182) and Tankyrase inhibitor XAV939 (MedChemExpress, HY-15147) were applied from DIV 0–7.

Oil Red O staining and quantification

Cells were washed three times with PBS to remove culture medium, followed by fixation with 4% PFA for 10 min at RT. After fixation, cells were stained with Oil Red O solution (Sigma-Aldrich, O1391) for 1 h at RT with gentle agitation. Unbound stain was removed by thorough washing with distilled water until the effluent became clear. Stained lipid droplets were imaged using an Axio Observer inverted microscope (Zeiss). Bound Oil Red O was then extracted with 100% isopropanol (15 min, RT) and absorbance measured at 500 nm using a microplate reader.

Hematoxylin and eosin (HE) staining of adipose samples and adipocyte quantification

Fresh iWAT and eWAT samples were fixed in 4% paraformaldehyde (PFA) overnight at 4 °C, followed by PBS washing. After sequential ethanol dehydration, tissues were paraffin-embedded and sectioned at 5 μm thickness. Sections were stained with HE solution (Servicebio, G1003) and imaged using an Axio Observer inverted microscope (Zeiss). Adipocyte size quantification was performed using AdipoCount software (http://www.csbio.sjtu.edu.cn/bioinf/AdipoCount/) through the following computational steps: (1) image thresholding to generate binary images and segment adipocyte membranes based on pixel intensity; (2) edge detection with histogram equalization to enhance contrast; and (3) post-processing combining thresholding and edge detection outputs for connected region analysis to determine individual adipocyte areas.

Immunofluorescence staining of adipose samples

Paraffin-embedded tissue sections were deparaffinized, rehydrated, and subjected to heat-induced antigen retrieval in a microwave oven. After blocking endogenous peroxidase and non-specific binding sites, sections were incubated overnight with primary antibody at 4 °C. Detection was achieved using an HRP-conjugated secondary antibody followed by tyramide signal amplification. Nuclei were counterstained with DAPI, autofluorescence was quenched, and slides were mounted with anti-fade mounting medium. Following antibodies were used: DPP4 (Abcam, #187048), PDGFRB (CST, #3169).

Whole-body metabolic status assessment

Indirect calorimetry was conducted using a comprehensive laboratory animal monitoring system (Columbus Instruments, Columbus, OH). After a 3-day acclimation period in the metabolic cages, data were collected from individually-housed mice over a 24-hour automated recording session. Mice had ad libitum access to food and water throughout the acclimation and data collection phases. Measurements were collected continuously and are presented as hourly averages across the light and dark phases. The experiments were conducted in adult mice fed a HFD starting at 6 weeks of age for 18 weeks (final age ~ 4 weeks).

Glucose tolerance test

After a 6-hour fast with free access to water, mice received an intraperitoneal injection of 50% (w/v) glucose solution at a dose of 2 g/kg body weight. Blood glucose levels were measured from tail vein samples at 0, 15, 30, 60, 90, and 120 min post-injection using a handheld glucometer (Accu-Chek, Roche Diagnostics, IN, USA).

Insulin tolerance test

Following a 6-hour fast, mice received an intraperitoneal injection of human insulin (Eli Lilly, Cat# 00002821501) at 1 U/kg body weight. Blood glucose was measured from tail vein samples at 0 (baseline), 15, 30, 60, 90, and 120 min post-injection using a handheld glucometer (Accu-Chek, Roche Diagnostics, IN, USA).

Mouse adipose tissue whole-mount imaging

Mice were anesthetized and perfused with PBS prior to euthanasia. iWAT and eWAT were dissected and fixed overnight in 1% PFA at 4 °C. Following fixation, tissues were washed in PBST (PBS containing 0.01% Tween-20), embedded in 5% low-melting agarose, and sectioned into 100–400 μm slices using a Leica VT1200 Vibratome. Samples were imaged with an Airyscan 2 LSM 900 confocal microscope (ZEISS) and images were processed using the Imaris software (Bitplane version 9.0.1).

BrdU assay

iWAT and eWAT SVF cells were stimulated with induction medium for 12 and 24 hours, respectively, before labeling with 10 μM BrdU (Roche, 11647229001) for 30 min. Cells were then fixed for DNA denaturation using the FixDenat solution, followed by incubation with anti-BrdU-POD antibody for 90 min at RT. After three washes with PBS-based buffer, tetramethylbenzidine substrate was added, and the reaction was incubated at RT until sufficient color development. Absorbance at 450 nm was measured using a microplate reader, with signal intensity reflecting DNA synthesis as a proxy for proliferation.

Statistics

Statistical analyses were performed using GraphPad Prism version 8.0.0. Statistical methods were specified in Figure legends. The level of statistical significance was assigned as *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001.