Graphical abstract graphic file with name fx1.jpg [68]Open in a new tab Highlights * • CD9^+ Progs in vWAT relate to fibrosis and poor T2D remission post-weight loss surgery * • Fibrotic and activated CD9^+ Progs can be targeted with nintedanib and celecoxib * • Dual treatment halts vWAT fibrosis progression and improves metabolic homeostasis * • Mesothelial cells contribute to perilobular vWAT fibrosis via PDGFRα/TGF-β pathways __________________________________________________________________ Rebière et al. combine a series of human investigations with preclinical assays and underscore the pivotal role adipose progenitors play in shaping tissue functions, influencing metabolic well-being and the response to anti-obesity therapy. Moreover, they highlight mesothelial cells as important contributors to the fibrotic transformation of visceral adipose tissue. Introduction Obesity favors the onset of metabolic alterations and is associated with major comorbidities such as type 2 diabetes (T2D). White adipose tissue (WAT) and its metabolic functions are major contributors to whole-body energy homeostasis.[69]^1 As WAT is extensively remodeled and its functions are severely impaired during obesity, there is growing interest in targeting WAT alterations to combat obesity-induced dysmetabolism. WAT massively enlarges during obesity, and its site of expansion critically determines disease outcome. Notably, patients with higher visceral WAT (vWAT) mass are at a greater risk of developing metabolic dysfunction than those with predominantly subcutaneous fat accumulation.[70]^2^,[71]^3^,[72]^4 Extensive vWAT remodeling, including interstitial or perilobular fibrosis,[73]^5^,[74]^6 often accompanies its expansion. Fibrosis favors vWAT dysfunction—defective insulin signaling, altered secretome, and impaired tissue plasticity—which precipitates systemic metabolic alterations.[75]^1^,[76]^7 Although vWAT fibrosis has long been recognized, its underlying determinants have remained ill defined. We unveiled the key contribution of fibrogenic CD9^+ adipose tissue progenitors (Progs).[77]^8 Indeed, CD9^+ Progs are major extracellular matrix (ECM) producers in obesity, and their activation drives vWAT fibrosis and dysfunction.[78]^8^,[79]^9 Therefore, targeting CD9^+ Progs may limit vWAT fibrotic remodeling and metabolic deterioration. In patients with obesity, we showed that CD9^+ Progs in vWAT associate with fibrosis and poor T2D remission following weight loss surgery. Therefore, we aimed to pharmacologically inhibit CD9^+ Progs’ activation to treat obesity-induced vWAT fibrosis and related dysfunction. We identified key molecular signatures of their fibrogenic reprogramming, including inflammation-related genes, epithelial-to-mesenchymal transition (EMT)-related genes, and G2/M checkpoint regulators. To block these processes, we selected Food and Drug Administration (FDA)-approved drugs. Nintedanib (Nin), a tyrosine kinase inhibitor, was chosen for its anti-proliferative and anti-EMT activity in other contexts. Celecoxib (Cel), a cyclooxygenase-2 inhibitor, was selected for its anti-inflammatory properties.[80]^10^,[81]^11^,[82]^12 Co-administration of these drugs effectively suppressed CD9^+ Prog activation and limited vWAT fibrosis progression, thereby improving metabolic homeostasis in obese mice. Importantly, both Ly-6C^+ Progs and mesothelial cells (MCs) within the CD9^+ pool[83]^9 underwent a pro-fibrotic switch during obesity, which was significantly attenuated by the drug combination. These findings provide a promising therapeutic strategy to attenuate obesity-induced vWAT fibrosis and dysfunction. Results CD9^+ progenitor prevalence in the visceral WAT of patients with obesity associates with T2D and weight loss outcome We previously reported that CD9 expression defines two platelet-derived growth factor receptor (PDGFR)α^+ Prog populations in vWAT of mice (perigonadal fat) and humans (omental fat).[84]^8 During obesity, CD9^+ PDGFRα^+ progenitors (CD9^+ Progs) adopt a fibrogenic phenotype and accumulate in fibrotic vWAT, whereas CD9^− counterparts, committed to adipogenesis, tended to disappear.[85]^8 Using a cohort of 88 patients with severe obesity undergoing bariatric surgery (BS) ([86]Table S1), we assessed whether CD9^+ Prog prevalence in the omental WAT (oWAT) at the time of surgery (baseline) correlated with clinical phenotypes and metabolic response to BS ([87]Figures S1A and S1B). Consistent with prior finding,[88]^8 CD9^+ Prog prevalence strongly associated with baseline metabolic variables including glycemia, insulinemia, homeostatic model assessment of insulin resistance (HOMA-IR), hemoglobin A1c (HbA1c), adiponectinemia, and oWAT fibrosis ([89]Figures 1A and 1B). We also identified additional association with android fat mass (dual-energy X-ray absorptiometry, DXA) and age ([90]Figure 1A). CD9^+ Prog frequency remained significantly associated with android fat mass, insulin levels, and HOMA-IR after adjustment for T2D, age, or both ([91]Figure 1A). Figure 1. [92]Figure 1 [93]Open in a new tab CD9^+ progenitor frequency in visceral WAT is associated with loss of glucose control and bariatric surgery outcomes (A) Heatmap of multivariable linear regressions between bioclinical variables and CD9^+ progenitor percentage in omental white adipose tissue (oWAT), with or without adjustment (n = 88). Standardized adjusted β coefficient is presented; ∗p < 0.05, ∗∗p < 0.005, and ∗∗∗p < 0.0005. (B) Pearson correlation between hydroxyproline content and CD9^+ Progs in oWAT from individuals with obesity (n = 60). R and p values are indicated. (C–K) Analysis of T2D remission 1 year after bariatric surgery (BS) in patients with obesity and T2D (OBD), stratified by low or high CD9^+ Progs in oWAT at the time of BS (n = 23). (C) Comparison of full, partial, or no T2D remission rates (chi-squared test, n = 19). The ratio of patients undergoing full, partial, and no T2D remission are presented. (D–K) Clinical parameters at baseline and 1 year post-BS are represented as mean ± SEM: (D) HbA1c (n = 18–23), (E) glycemia (n = 18–23), (F) insulinemia (n = 16–22), (G) HOMA-IR (n = 13–18), (H) BMI (n = 19–23), (I) fat mass (n = 18–22), (J) fat-free mass (n = 18–22), and (K) leptinemia (n = 18–22). Group comparisons used Student’s t test and generalized linear models adjusted for baseline values. p_adj: post hoc contrasts; plots show unadjusted data. ns, not significant. As CD9^+ Progs accumulation was associated with impaired glucose metabolism, we then tested whether CD9^+ Prog abundance correlated with T2D remission following BS. We thus only considered patients with T2D and stratified them into two groups based on the median of CD9^+ Prog frequency (e.g., “Low” versus “High” abundance of CD9^+ Progs) ([94]Figure S1B, Analysis #2). T2D remission is typically defined by normalized glycemia and HbA1c (<7 mmol/L and <6%, respectively) without glucose-lowering medication at 1 year post-BS.[95]^13 T2D remission was more frequent in the “Low” CD9^+ Progs group ([96]Figure 1C). Notably, these groups did not differ in baseline glycemic markers, medication, or T2D duration ([97]Figures 1D–1G, [98]S1C, and S1D). By contrast, at 1-year post-BS, patients with “Low” CD9^+ Progs frequency showed lower HbA1c ([99]Figure 1D), a trend toward reduced glycemia ([100]Figure 1E) and decreased insulinemia and HOMA-IR ([101]Figures 1F and 1G). Baseline parameters like surgery type, age ([102]Figures S1E and S1F), BMI, and body composition ([103]Figures 1H–1J) were comparable, but the “Low” group showed greater post-BS reductions in BMI, fat mass, and leptinemia ([104]Figures 1H–1K). Lipid and liver function markers remained similar at baseline and 1 year ([105]Figure S1G). Baseline-adjusted analyses ([106]Figures 1D–1K) confirmed that BMI and insulinemia reductions remained significant ([107]Figures 1F and H), suggesting that diabetes remission in the low-CD9^+ group may relate to greater BMI loss. Trends for other baseline-adjusted variables persisted, although statistical significance was not reached due to the limited sample size. Overall, CD9^+ Prog frequency in patients with severe obesity correlates with abdominal fat distribution, fibrosis, and glycemic control. In patients with T2D, high CD9^+ Prog frequency is associated with lower T2D remission and reduced BMI loss post-surgery. These results suggest that limiting CD9^+ Prog expansion and activation may improve vWAT remodeling and metabolic outcomes. To address this, we set out to develop a therapeutic strategy aimed at limiting the expansion and activation of CD9^+ Progs, which we subsequently tested in preclinical mouse models of obesity. A therapeutic drug combination inhibits the pro-fibrotic activation of Progs through in vitro experiments We first aimed to identify determinants of CD9^+ Prog fibrogenic fate during obesity by analyzing the transcriptional landscape. CD9^+ Progs were isolated from the epididymal WAT (EpiWAT) of lean and obese C3H mice, a strain prone to obesity-induced WAT fibrosis.[108]^8^,[109]^14 We found 1,529 genes upregulated and 954 downregulated in CD9^+ Progs from fibrotic obese WAT. Genes involved in inflammation, hypoxia, G2/M checkpoint, and EMT (a hallmark of myofibroblast differentiation[110]^15) were induced, whereas genes linked to stemness, adipogenesis, and oxidative phosphorylation were downregulated ([111]Figures 2A and 2B). These data indicate that CD9^+ Progs acquire a pro-fibrotic, inflammatory, and proliferative phenotype during obesity. To assess whether CD9^+ Prog activation preceded fibrosis, we analyzed vWAT after 7 days of high-fat diet (HFD), before histological fibrosis signs appeared ([112]Figures 2C and 2D). CD9^+ Progs already showed increased proliferation and density ([113]Figures 2E and 2F). Col1a1 and Ptgs2 mRNA expression were elevated, while other fibrotic markers remained unchanged ([114]Figure 2G). This indicates early activation of these pathways that could thus participate in the EpiWAT fibrotic transformation. Figure 2. [115]Figure 2 [116]Open in a new tab The fibrogenic phenotype of the progenitors is curbed with nintedanib and celecoxib (A) Gene set enrichment analysis of differentially expressed genes in CD9^+ progenitors isolated from lean and obese fibrotic EpiWAT of C3H mice. (B) Heatmap of selected significantly up- or downregulated genes in CD9^+ progenitors from lean and obese EpiWAT (n = 4; fold change (FC) > 1.5; adjusted p < 0.05). (C) EpiWAT mass in mice fed chow or HFD for 7 days (n = 5). (D) Representative bright-field and polarized images of picrosirius red-stained EpiWAT sections and adipocyte size distribution (scale bar, 100 μm; n = 3). (E) Flow cytometry plots of Ki-67 expression in CD9^+ progenitors and quantification of Ki-67^+ cells (n = 4–5). (F) CD9^+ progenitor count per gram of EpiWAT in chow- and HFD-fed mice (n = 4–5). (G) Relative mRNA expression of fibrosis markers in whole EpiWAT (normalized to chow; n = 5). (H) mRNA expression in CD9^+ progenitors from obese fibrotic EpiWAT treated with vehicle (Veh), nintedanib (Nin), and/or celecoxib (Cel) (n = 4). (I) mRNA expression in CD9^+ progenitors from human oWAT treated with Veh, Nin, and/or Cel (n = 4). (A and B) Linear models for microarray data (LIMMA) analysis with multiple-testing correction. (C–H) Data are represented as mean ± SEM. (C–G) Student’s t test. (H and I) One-way ANOVA with Newman-Keuls post hoc test. Since chronic inflammation is frequently associated with fibrosis[117]^16 and proliferation would likely increase the number of fibrogenic CD9^+ Progs, we reasoned that a combination of anti-inflammatory, anti-proliferative, and anti-fibrotic drugs could limit CD9^+ Prog activation. To enhance translational potential, we selected FDA-approved drugs. Ptgs2 (cyclooxygenase [COX]-2) was induced in CD9^+ Progs ([118]Figures 2B and G); thus, we chose the COX-2 inhibitor celecoxib (Cel). To target proliferation and fibrosis, we selected nintedanib (Nin), a tyrosine kinase inhibitor effective in idiopathic pulmonary fibrosis.[119]^11 We next developed an in vitro assay using EpiWAT Progs from obese mice, treated with vehicle (Veh), Nin, Cel, or their combination (N+C). Both Nin and Cel reduced expression of several fibrosis-related genes, with the combination showing the strongest effect ([120]Figure 2H). Using human oWAT Progs from patients with obesity, we confirmed that Nin significantly reduced most fibrosis markers, while Cel decreased COL1A1 and COL3A1 expression ([121]Figure 2I). Again, the drug combination yielded the most pronounced anti-fibrotic response, suggesting that their combined effect is more potent in reducing cell fibrosis. Combined Nin and Cel treatment curbs fibrosis development in obese mice We assessed whether the drug combination could limit obesity-induced vWAT fibrosis in vivo using C3H mice fed a high-fat diet (HFD). First, we asked whether our drug combination could prevent fibrosis development by initiating pharmacological treatment early. Treatment began 1 week after HFD onset ([122]Figure 3A), a time point already associated with glucose intolerance ([123]Figure S2A) and moderate insulin resistance ([124]Figures S2B and S2C). After 4 weeks of treatment with Veh, Nin, Cel, or both (Nin+Cel), body weight, fat and lean mass were similar across groups ([125]Figures 3B–3D). In line, the masses of various tissues, including EpiWAT, inguinal white adipose tissue (IngWAT), liver, and kidneys, remained unchanged among the groups ([126]Figures 3E–3H). However, while each drug alone had a moderate impact, the combination of Nin+Cel significantly improved glucose tolerance ([127]Figure 3I). Although basal fasted glycemia was similar among groups ([128]Figure 3J), the reduction in insulin levels ([129]Figure 3J) suggest enhanced insulin sensitivity. Figure 3. [130]Figure 3 [131]Open in a new tab The combination of nintedanib and celecoxib treatment improved glucose and lipid metabolism in vivo (A) Experimental design: 1 week after initiating high-fat diet (HFD), C3H male mice were treated for 4 weeks with vehicle (Veh), nintedanib (Nin), celecoxib (Cel), or Nin+Cel. (B–D) Body weight, fat mass, and lean mass measured pre- and post-treatment (n = 8–10); dashed lines indicate mean values from age-matched chow-fed controls. (E–H) Mass of EpiWAT, IngWAT, liver, and kidney post-treatment (n = 8–10). Dashed lines indicate mean values from age-matched chow-fed controls. (I) Glucose tolerance test (GTT) and area under the curve (AUC) (n = 8–10). (J) Fasting glycemia (n = 8–10) and insulinemia (n = 8–9). (K) Liver histology (H&E; scale, 50 μm) and steatosis quantification (n = 4–5); dashed lines represent chow-fed controls. (L–N) VO[2] normalized to lean mass, food intake, and respiratory exchange ratio (RER) in Veh vs. Nin+Cel groups (n = 5). (O) Plasma glycerol levels pre- and post-CL 316,243 injection (n = 5–6). (B–O) Data are represented as mean ± SEM. (B–J) One-way ANOVA with Newman-Keuls post hoc test; (K, M, and O) Student’s t test; (L and N) two-way repeated-measures ANOVA. A second independent experiment confirmed these effects: at baseline, before receiving either the Veh or the Nin+Cel combination, no significant differences were detected in glucose tolerance, insulin levels, or insulin sensitivity among HFD-fed mice ([132]Figures S2A–S2C). After treatment, Nin+Cel improved glucose tolerance ([133]Figure S2D), lowered glucose and insulin levels ([134]Figure S2E), and restored insulin sensitivity to levels seen in lean controls ([135]Figure S2F), confirming the significant efficacy of the drug combination. In addition, Nin+Cel treatment significantly reduced liver steatosis ([136]Figure 3K). Consistent with reduced lipid accumulation, metabolic analysis revealed improved lipid utilization, as reflected by lower respiratory exchange ratios ([137]Figure 3N), indicating increased fatty acid oxidation. These changes occurred despite unchanged oxygen consumption and food intake ([138]Figures 3L–3M). In addition, increased glycerol release in response to β3-adrenergic stimulation ([139]Figure 3O) provided further evidence of improved lipid metabolism. Together, these findings suggest that Nin+Cel treatment not only improves glucose clearance but also improves lipid utilization. We then focused our analysis on EpiWAT fibrosis. Histological analysis showed that Nin+Cel abolished collagen deposition in EpiWAT, restoring tissue architecture similar to that of lean controls ([140]Figures 4A, 4B, and 4D). This was accompanied by reduced Col1a1 and Lox mRNA expression ([141]Figure 4E), independent of changes in EpiWAT mass ([142]Figure 3E) or adipocyte size ([143]Figures 4B and 4C). Inflammation was also alleviated: while all treatments reduced macrophage proliferation ([144]Figure 4F), only Nin and Nin+Cel significantly decreased macrophage accumulation. Notably, the combination therapy uniquely suppressed the expression of inflammatory markers Tnf, Spp1, and Lgals3 ([145]Figures 4G and 4H). Supporting these findings, RNA sequencing (RNA-seq) of CD64^+CD45^+ macrophages revealed an enrichment of resident macrophage gene signatures and a repression of lipid-associated macrophage profiles in response to Nin+Cel ([146]Figures S3A–S3C). Together, these results suggest that the reduction in macrophage numbers was also driven by diminished recruitment following combination treatment. Figure 4. [147]Figure 4 [148]Open in a new tab Combination treatment with nintedanib and celecoxib exerts anti-fibrotic effects in adipose tissue (A) Hydroxyproline content in EpiWAT of HFD-fed C3H mice treated with Veh, Nin, Cel, or Nin+Cel (n = 8–10). (B) Representative bright-field and polarized light images of picrosirius red-stained EpiWAT (scale, 100 μm). (C–E) (C) Mean adipocyte size (n = 6); (D) fibrosis per adipocyte (arbitrary units) (n = 5–6); (E) mRNA expression in EpiWAT (n = 7–8). (F) Percentage of Ki-67^+ macrophages and macrophage density (CD45^+CD64^+CD31^−) in EpiWAT (n = 6–8). (G–H and J) mRNA expression in EpiWAT (n = 7–8). (I) Insulin-stimulated p(S473)AKT and pan-AKT levels in EpiWAT; quantification relative to basal (n = 5). (K–M) Flow cytometry of EpiWAT progenitors: (K) left, representative fluorescence-activated cell sorting (FACS) plot showing percentage of CD9^+PDGFRα^+Ki-67^+; right, the percentages of CD9^+PDGFRα^+Ki-67^+ are represented as mean ± SEM. (L) CD9^+ progenitor density and (M) CD9^− preadipocyte density (n = 6–8). (N) Hydroxyproline content in IngWAT (n = 8–10). (O–Q) mRNA expression in IngWAT (n = 5–8). (R–T) IngWAT progenitor analysis (n = 6–8): (R) percentage of PDGFRα^+Ki-67^+, (S) PDGFRα^+ cell density, and (T) percentage of CD36^+ progenitors. (A–T) Data are represented as mean ± SEM. (A, D–H, and J–T) One-way ANOVA with Newman-Keuls post hoc test; (C and I) Student’s t test. In line with the improved inflammation in EpiWAT, Nin+Cel treatment also enhanced EpiWAT function, evidenced by increased insulin-stimulated protein kinase B (AKT) phosphorylation ([149]Figure 4I) and elevated Adipoq and Leptin expression ([150]Figure 4J). We next examined whether our drug combination impacts on EpiWAT progenitor phenotype. We previously described that a progenitor imbalance occurs with an accumulation of pro-fibrotic CD9^+ Progs and a progressive loss of CD9^− preadipocytes over time in the EpiWAT of HFD-fed C3H mice.[151]^8 Here, Nin (alone or combined) reduced CD9^+ Prog proliferation and numbers ([152]Figures 4K and 4L), while Cel (alone or combined) preserved CD9^− preadipocytes ([153]Figure 4M), supporting a favorable shift in progenitor balance. Together, the co-administration of Nin and Cel showed therapeutic potential by effectively curtailing macrophage-associated inflammation, obesity-induced progenitor proliferation, fibrosis development, and associated dysfunctions. Although fibrosis is absent in IngWAT at this stage,[154]^8 we evaluated drug effects there. No differences were found in adipocyte size ([155]Figures S2G and S2H), collagen deposition ([156]Figures 4N and [157]S2G, S2I), or fibrosis markers ([158]Figure 4O). However, Adipoq and Leptin mRNA levels increased with Nin or Nin+Cel ([159]Figure 4P), while Tnf (with Nin or Nin+Cel) and Lgals3 (with Cel alone) expression declined in specific groups ([160]Figure 4Q). However, insulin sensitivity remained unchanged ([161]Figure S2J). At the progenitor level, though their proliferation was reduced ([162]Figure 4R), their density ([163]Figure 4S) remained unaffected. To further investigate, we examined preadipocytes in IngWAT using the CD36^+ marker[164]^17 ([165]Figure S2K) and found no significant differences in preadipocyte abundance ([166]Figure 4T). Overall, these results highlight depot-specific responses, with EpiWAT being more responsive to Nin+Cel treatment. Combined Nin and Cel treatment suppresses fibrosis progression in obese mice We next assessed whether Nin+Cel treatment could limit the progression of established vWAT fibrosis. To this end, the drug combination was administered after 5 weeks of HFD ([167]Figure 5A), when fibrosis was already present ([168]Figure 5B). A 3-week treatment was deemed sufficient, as hydroxyproline levels approximatively doubles between 5 and 8 weeks of HFD (not shown), and previous data ([169]Figure 4A) suggested a ∼50% reduction in fibrosis with treatment. At the end of 3 weeks of treatment, body weight ([170]Figure 5C) was similar between groups. Liver analysis revealed unchanged liver mass ([171]Figure 5D) but reduced lipid deposition in Nin+Cel-treated mice ([172]Figures 5E and 5F). In EpiWAT, Nin+Cel significantly reduced collagen content without affecting EpiWAT fat mass and adipocyte size ([173]Figures 5G–5K). Although fibrosis was not fully reversed, progression was effectively halted, even as fat accretion continues to progress. Furthermore, this reduction in fibrosis was associated with improved glucose tolerance ([174]Figures 5L, 5M, and [175]S4A–S4F) and insulin sensitivity ([176]Figure 5N), restoring glucose levels to those seen in lean mice ([177]Figure 5L). Figure 5. [178]Figure 5 [179]Open in a new tab The combination of nintedanib and celecoxib blocks the progression of established EpiWAT fibrosis (A) Experimental design: C3H male mice fed HFD for 5 weeks and then treated with Veh or Nin+Cel for 3 weeks. (B) Hydroxyproline content in EpiWAT from chow- or HFD-fed mice (n = 4–5). (C and D) Body weight and liver mass after treatment (n = 10); dashed lines: age-matched chow-fed mice. (E and F) Liver H&E staining (scale, 50 μm) and steatosis quantification (n = 5). Dashed lines: age-matched chow-fed mice. (G and H) EpiWAT mass and hydroxyproline content (n = 10); dashed line in (H): baseline HFD-fed mice. Dashed lines: age-matched chow-fed mice. (I) Representative picrosirius red-stained EpiWAT images (scale, 100 μm). (J and K) Adipocyte size and fibrosis per adipocyte in EpiWAT (n = 7–8). (L and M) GTT, AUC, fasted glycemia, and insulinemia (n = 4–10). (N) Insulin tolerance test (ITT) in Veh-vs. Nin+Cel-treated mice (n = 5–10). (O–Q) EpiWAT progenitor analysis: (O) left, representative FACS plot showing the percentage of CD9^+PDGFRα^+Ki-67^+; right, the percentages of CD9^+PDGFRα^+Ki-67^+ are represented as mean ± SEM. (P) CD9^+ progenitor density and (Q) PDGFRα^+CD9^− preadipocyte density (n = 8). (R and S) Macrophage counts and Ki-67^+ percentage among macrophages in EpiWAT (n = 8). (T) IngWAT mass (n = 10). (U and V) IngWAT histology (picrosirius red; scale, 100 μm), adipocyte size, and fibrosis per adipocyte (n = 4–6). (W–Y) IngWAT progenitor analysis: (W) PDGFRα^+ cell density, (X) percentage of PDGFRα^+Ki-67^+, and (Y) percentage of CD36^+ progenitors (n = 8). (Z) Spearman correlation between CD36^+ progenitors and IngWAT mass (n = 42). R and p values are indicated. (B–Y) Data are represented as mean ± SEM. (B–J and O–Y) Student’s t test; (K–N) one-way ANOVA with Newman-Keuls post hoc test; (Z) Spearman correlation. As observed earlier in the context of fibrosis development ([180]Figures 4K and 4L), Nin+Cel also reduced CD9^+ Progs proliferation ([181]Figure 5O) and numbers ([182]Figure 5P), though CD9^− preadipocyte density remained unchanged ([183]Figure 5Q). In this context of established fibrosis, CD9^− preadipocytes were already markedly reduced at the start of the treatment (after 5 weeks of HFD).[184]^8 Although still effective at limiting CD9^+ Prog accumulation and the progression of established vWAT fibrosis, Nin+Cel treatment did not allow to rescue the loss of CD9^− preadipocytes. Regarding inflammation, although macrophage numbers decreased ([185]Figure 5R), their proliferation is found unchanged between the two conditions at the end of the treatment period ([186]Figure 5S). In IngWAT, no differences in fibrosis or adipocyte size were detected ([187]Figures 5U and 5V), but Prog count ([188]Figure 5W) and proliferation were reduced ([189]Figure 5X), alongside an increase in CD36^+ cells ([190]Figure 5Y) and IngWAT mass ([191]Figures 5T and 5Z), suggesting a pro-adipogenic effect in this depot. In our cohort of patients with obesity and T2D, elevated CD9^+ Progs frequency, measured prior to any weight loss intervention, was associated with impaired body weight loss and reduced T2D remission. We then evaluated whether our pharmacological approach could optimize weight loss response in mice. Specifically, we investigated whether preconditioning with the anti-fibrotic Nin+Cel before weight loss induction could influence animal outcomes ([192]Figure S4G). To this end, C3H mice were fed an HFD for 4 weeks, reaching an obese state characterized by increased adiposity and marked glucose intolerance but without detectable EpiWAT fibrosis. The mice were then treated with the drug combination (Nin+Cel) until improved glycemia was detected after 2 weeks of treatment ([193]Figure S4K). After this 6 weeks period, HFD feeding and the drug treatment were stopped, and weight loss was induced by shifting animals to a standard chow diet as previously described.[194]^18 Mice pretreated with Nin+Cel showed greater body weight and fat mass loss ([195]Figures S4H and S4I), with unchanged lean mass ([196]Figure S4J), and improved glycemia, insulinemia, and insulin sensitivity ([197]Figures S4K–S4M). Glucose tolerance was also enhanced ([198]Figure S4N). Notably, the initial glucose-lowering effect of Nin+Cel adds to the glucose-lowering effect achieved through weight loss. Thus, these findings suggest that targeting CD9^+ Progs and fibrosis may optimize weight loss response and glucose control. Fibro-inflammatory Progs and mesothelial cells (MCs) undergo obesity-induced fibrotic transformation Our data suggest that limiting CD9^+ Prog numbers via combined Nin and Cel treatments curbs vWAT fibrosis. As previously reported, CD9^+ Progs include CD9^+ Ly-6C^+ fibro-inflammatory progenitors (FIPs) and CD9^+ Ly-6C^− MCs,[199]^9 which may respond differently to the anti-inflammatory/anti-fibrotic drugs. In lean C3H animals, both populations were present in EpiWAT, with FIPs comprising ∼61% and MCs ∼16% of Progs ([200]Figure 6A). RNA-seq confirmed enrichment of MC markers (e.g., Msln, Krt19, Upk3b, and Lrrn4) in CD9^+ Ly-6C^− cells. ([201]Figure 6A). Gene profiling in obese mice revealed distinct enrichment in gene ontology (GO) biological processes: FIPs upregulated prostanoid and prostaglandin synthesis pathways (including Ptgs2 that encodes COX-2), and MCs upregulated collagen-related genes ([202]Figure 6B). While FIPs induce a fibrosis gene signature in obesity ([203]Figure 6D), the expression of ECM components was even higher in MCs ([204]Figure 6C), suggesting that MCs may also play an important role in vWAT fibrosis. This observation prompted us to further investigate MCs in the EpiWAT of lean and obese C3H mice. Figure 6. [205]Figure 6 [206]Open in a new tab Mesothelial cells and fibro-inflammatory progenitors undergo a fibrotic transformation in EpiWAT of obese C3H mice (A) Left: gating strategy for FIPs (GP38^+CD9^+Ly-6C^+CD45^−CD31^−) vs. mesothelial cells (MCs, GP38^+CD9^+Ly-6C^−CD45^−CD31^−). Right: heatmap of MC marker expression (Msln, Krt19, Upk3b, and Lrrn4) in sorted FIPs and MCs (n = 5). (B) GO term enrichment analysis of genes differentially expressed between FIPs and MCs from EpiWAT of HFD-fed mice. (C) Heatmap of ECM-related gene expression upregulated in FIPs vs. MCs (n = 5, FC > 1.5, adjusted p < 0.05). (D) mRNA expression of fibrosis marker genes in FIPs and MCs; Msln in MCs (n = 5). (E) Flow cytometric quantification of MSLN^+ cells per gram of EpiWAT in chow- and HFD-fed mice (n = 6). (F) Representative picrosirius red staining of EpiWAT perilobular area (scale, 50 μm) and quantification of collagen deposition (n = 4–5). (G and H) Two-photon microscopy of EpiWAT from Krt19-CreERT; tdTomato mice fed HFD. (G) Top view: Tomato^+ MCs (red) and second harmonic generation (SHG)-collagen (green) at tissue surface. (H) Side view: mesothelial layer closely associated with collagen fibers (scale, 20 μm). (I) Tomato^+ cell density in EpiWAT from control (Krt19-CreERT; tdTomato) and Krt19-αK (Krt19-CreERT; tdTomato; PDGFRαK) mice (n = 5–7). (J) mRNA expression of fibrosis genes in sorted Tomato^+ MCs from controls and Krt19-αK (n = 3–4). (K and L) Human WAT histology from patients with obesity (n = 3). (K) Representative oWAT images showing COL1 (green), MSLN (red), DAPI (blue), and bright field (gray); scale, 20 μm. (L) MSLN expression on mesWAT and oWAT but not on scWAT surfaces; autofluorescence (green), DAPI (blue), and bright field (gray); scale, 130 μm. (D–F, I, and J) Data are represented as mean ± SEM. (A–C) LIMMA analysis with adjusted p values. (D) One-way ANOVA with Newman-Keuls test. (D–J) Student’s t test. Similar to FIPs, MCs exhibit increased expression of genes associated with myofibroblastic transformation in the EpiWAT of obese C3H mice ([207]Figure 6D), suggesting that both MCs and FIPs contribute to the fibrotic remodeling of EpiWAT. Concurrently, the expression of the identity marker Msln is downregulated in MCs ([208]Figure 6D), indicating that obesity-induced WAT fibrosis may involve mesothelial-to-mesenchymal transition (MMT), as previously described.[209]^19^,[210]^20 MCs, identified as mesothelin (MSLN)-positive cells in the EpiWAT ([211]Figure S5A), increased their density in the fibrotic EpiWAT ([212]Figure 6E). This suggests that MCs expand beyond fat mass growth and accumulate in fibrotic EpiWAT. Consistently, we observed increased thickening of perilobular collagen deposition in the EpiWAT of obese mice ([213]Figure 6F). To further study MCs, we used a tamoxifen-inducible Cre mouse model under the control of the Krt19 gene promoter (Krt19-creERT) mice[214]^21^,[215]^22 crossed with the tdTomato reporter strain that expresses tdTomato upon Cre-mediated recombination ([216]Figure S5B). After tamoxifen treatment, ∼20% of MCs expressed Tomato ([217]Figure S5C). In obese Krt19-CreERT; tdTomato mice, two-photon laser scanning microscopy showed MCs lining the fat pad surface, closely associated with collagen fibers (second-harmonic generation) ([218]Figures 6G and 6H). This supports that MCs could contribute to ECM production and organization in the vWAT. As the PDGFRα pathway is a key driver of WAT fibrosis,[219]^8^,[220]^23^,[221]^24 we wondered whether MCs could respond to the activation of this pathway. We confirmed PDGFRα expression on MCs ([222]Figure S5D) and used conditional knockin mice in which Cre recombination drives the expression of a constitutively active PDGFRαK mutant isoform (D842V mutation in the kinase domain) under the control of the endogenous PDGFRα promoter (αK).[223]^25 We then generated Krt19-creERT; αK; tdTomato animals (called Krt19-αK) in which both the PDGFRαK mutant and tdTomato are expressed in a subset of MCs ([224]Figure S5E). These mice had increased Tomato^+ MCs in EpiWAT ([225]Figure 6I), indicating that the D842V mutation likely enhances MC proliferation and survival in agreement with previous study.[226]^23 Furthermore, qPCR of sorted MCs expressing PDGFRαK showed upregulation of Col1a1 and Col3a1 ([227]Figure 6J), supporting a pro-fibrotic role of activated PDGFRα in MCs. These findings suggest that PDGFRα activation in MCs could promote a fibrotic phenotype contributing to obesity-induced WAT fibrosis development.[228]^8^,[229]^25 Then, we wanted to address the relevance of our mouse findings to MCs isolated from human oWAT. In primary human MCs, transforming growth factor β (TGF-β) triggered morphological changes shifting from their typical cobblestone shape to a fibroblastic appearance and upregulated fibrosis genes while downregulating mesothelial markers (MSLN, KRT19, and LRRN4) ([230]Figures S6A–S6C). Furthermore, public datasets from MCs isolated from oWAT of lean individuals and patients with obesity showed that genes associated with inflammation, hypoxia, EMT, and glycolysis were significantly upregulated with obesity ([231]Figure S6D). Concurrently, the expression of MSLN, KRT19, and LRRN4 was decreased ([232]Figure S6E). Together, this supports that MMT[233]^26 may occur in oWAT during human obesity. We then investigated whether Nin+Cel could modulate fibrosis driven by TGF-β and PDGFRα activation. Both TGF-β and platelet-derived growth factor-AA (PDGF-AA) treatments reduced the expression of mesothelial markers MSLN and LRRN4, with TGF-β also downregulating KRT19. Nin+Cel counteracted these effects in response to PDGF-AA but not TGF-β ([234]Figure S6F). Regarding COL1A1 and FN1 expression, only COL1A1 was reduced following Nin+Cel treatment under both TGF-β and PDGF-AA stimulation ([235]Figure S6G). Analysis of fibro-inflammatory cytokines revealed that, while PDGF-AA had no impact on Interleukin-6, inhibin subunit beta A (INHBA), or connective tissue growth factor (CTGF) expression, TGF-β-induced upregulation of these markers was attenuated by Nin+Cel ([236]Figure S6H). Tenascin C (TNC) and PAI-1 were upregulated by TGF-β, and this induction persisted despite Nin+Cel treatment. In contrast, PDGF-AA induced a milder increase in TNC and PAI-1, with Nin+Cel inhibiting only TNC expression. Although TGF-β plays a more prominent role in driving fibrosis compared to PDGF-AA, Nin+Cel can counteract the fibrotic effects of both. Specifically, Nin+Cel more strongly inhibits PDGFRα-driven pathways, while also providing partial protection against TGF-β-induced fibrosis. We further examined the histological organization of the mesothelium in human oWAT from patients with obesity. oWAT exhibits a distinct lobular architecture, with fibrosis deposition notably occurring at the surface of the small lobules during obesity.[237]^27 Indeed, picrosirius red staining in oWAT of patients with obesity showed intense perilobular collagen deposition ([238]Figure S6I). This structured pattern of fibrosis led us to explore whether specific cell types could contribute to fibrosis. MSLN^+ cells were aligned along lobular surfaces ([239]Figure S6J), co-localizing with collagen I ([240]Figure 6K), suggesting a structural role in ECM deposition. This mesothelial pattern was observed in mesenteric but not subcutaneous fat ([241]Figure 6L), suggesting MC involvement in depot-specific fibrosis. Together, our data identify MCs as key contributors to visceral adipose fibrosis in both mice and humans, potentially via PDGFRα and TGF-β signaling. MCs and fibro-inflammatory Progs are both responsive to the drug combination Having identified that both FIPs and MCs produce collagens and fibrotic material, we then assessed the impact of Nin+Cel co-treatment on FIP and MC biology in our model of established WAT fibrosis ([242]Figure 5). Using flow cytometry analysis, we observed that FIP and MC density and proliferation were markedly lowered in response to Nin+Cel treatment ([243]Figures 7A and 7B). We also found that Msln expression was elevated in MCs isolated from the EpiWAT of mice receiving the combined treatment ([244]Figure 7C), suggesting that Nin+Cel limited MMT. We then performed transcriptional profiling of FIPs and MCs isolated from the EpiWAT of Veh- or Nin+Cel treated animals. In response to the drug combination, 167 genes and 178 genes were significantly downregulated in FIPs and in MCs, respectively. Notably, 62 downregulated genes were common between FIPs and MCs ([245]Figure 7D). Importantly, gene set enrichment analysis ([246]Figure 7E) revealed decreased expression of the fibrosis-associated gene signature that we had identified as upregulated in fibrotic CD9^+ Progs ([247]Figure 2A). In FIPs and MCs, Nin+Cel targets pathways associated with EMT, inflammation, or hypoxia ([248]Figure 7F), and, accordingly, many ECM-related genes were found downregulated ([249]Figure 7G). Consistently, we measured reduced perilobular fibrosis in EpiWAT from mice treated with Nin+Cel compared to the Veh-treated group ([250]Figure 7H). Together, this analysis unveiled the efficacy of the combined administration of Nin and Cel in limiting the fibrogenic activation of both MCs and FIPs, explaining the reduced fibrosis deposition and improved EpiWAT function in obese C3H. Figure 7. [251]Figure 7 [252]Open in a new tab FIPs and MCs lowered their fibrotic potential in mice treated with nintedanib and celecoxib (A) Density of CD9^+Ly-6c^+ FIPs and CD9^+Ly-6c^− MCs in EpiWAT of HFD-fed C3H mice treated with vehicle or Nin+Cel (treatment start after 5 weeks of HFD, n = 8). (B) Percentage of Ki-67^+ FIPs and MCs (n = 8). (C) Msln mRNA expression in MCs sorted from EpiWAT (n = 6). (D) Venn diagram of genes downregulated in FIPs and MCs following Nin+Cel treatment. (E) Gene set enrichment analysis (GSEA) plots showing enrichment of CD9^+ Prog fibrosis-associated signatures among genes downregulated in FIPs/MCs by the combined treatment (Phantasus). (F) Pathway enrichment analysis of downregulated genes in FIPs and MCs (Phantasus). (G) Heatmap of ECM-associated genes downregulated in sorted FIPs and MCs (fold change > 1.5, adjusted p < 0.05; n = 5). (H) Representative images (bright-field and polarized light) and quantification of perilobular collagen in EpiWAT from HFD-fed mice treated with Veh or Nin+Cel (n = 5–7; scale bar, 50 μm). (A–C and H) Data are represented as mean ± SEM. (A and B) One-way ANOVA with Newman-Keuls test; (C and H) Student’s t test; (D–G) LIMMA analysis with adjusted p values. Discussion Obesity, a major global health issue, is prevalent on all continents and requires innovative strategies to address its metabolic complications. WAT malfunction, especially visceral depots (vWAT), plays a pivotal role in obesity-related metabolic dysfunctions. Consequently, extensive research efforts have focused on the WAT to uncover novel therapeutic opportunities[253]^28. Fibrosis significantly contributes to vWAT dysfunction, thereby playing a role in the development of metabolic complications associated with obesity.[254]^7 Additionally, WAT fibrosis impairs the effectiveness of interventions such as BS.[255]^27^,[256]^29 Therefore, understanding the mechanisms underlying vWAT fibrosis holds promise to unveil additional therapeutic strategies. We previously highlighted the pivotal role of CD9^+ adipose Progs in vWAT fibrosis.[257]^8 Expanding on this, we now confirm in a larger human cohort that CD9^+ Progs correlate with vWAT fibrosis and impaired glycemic control in humans, as well as with age. After adjusting for potential confounders such as age and gender, we confirmed a strong correlation between insulin resistance markers, visceral fat accumulation (android obesity), and CD9^+ Prog accumulation in oWAT. These findings suggest that CD9^+ Progs may be involved in the interplay between visceral obesity and the development of T2D. Notably, in patients with T2D undergoing BS, higher CD9^+ Prog levels were linked to reduced BMI loss and limited diabetes remission. While these two effects are intertwined and difficult to separate, our observations suggest that CD9^+ Prog accumulation may reflect the extent of adverse tissue remodeling at the onset of weight loss. This remodeling, potentially via modulation of adipocyte function, could influence an individual’s capacity to lose fat mass during a weight loss intervention, which, in turn, may contribute to T2D remission. Thus, our findings not only reinforce the strong association between CD9^+ Progs, fibrosis, and glucose control but also highlight the potential of targeting CD9^+ Progs as a potential therapeutic strategy. As a proof of concept, to therapeutically target CD9^+ Progs, we investigated the combined use of FDA-approved Nin and Cel. This dual treatment reduced CD9^+ Prog expansion and fibrotic activation, preserving vWAT structure and improving metabolic outcomes in obese mice. Moreover, pre-treatment with Nin+Cel enhanced fat loss and glucose tolerance upon weight loss induction. These effects could be partly attributed to improved β3-adrenergic signaling, which is known to promote lipolysis and fat mass loss.[258]^30 Importantly, the initial glucose-lowering effect of Nin+Cel appears to complement the glucose-lowering effect achieved through subsequent weight loss. Such approaches could potentially maximize the beneficial effects of weight loss and thus help reduce inter-individual variability in the outcomes of anti-obesity strategies.[259]^31 While our aim was to target CD9^+ Prog cells, recent single-cell RNA sequencing (scRNA-seq) of vWAT Progs unveiled the considerable degree of diversity within this population. Hepler et al.[260]^9 indeed showed that CD9^+ Progs encompass two major subsets: CD9^+ Ly-6C^+ cells, characterized by fibrotic, pro-inflammatory, and anti-adipogenic traits, and CD9^+ Ly-6C^− cells, which are MCs. MCs form a single-layered structure lining the vWAT (but not subcutaneous fat depots) and express specific markers, including MSLN and keratin 19 (KRT19). The mesothelium assumes a multifaceted role in various physiological functions.[261]^32^,[262]^33 Primarily, it acts as a protective physical barrier from mechanical damage and infection, while also regulating intra-organ fluid transport.[263]^34 Under specific circumstances, like peritoneal dialysis or liver injury, MCs can undergo detrimental MMT, resulting in the production of fibrosis-related molecules.[264]^20^,[265]^35^,[266]^36. In obesity, MCs have been shown to secrete chemokines that may contribute to inflammation in human omental adipose tissue.[267]^37 Recent studies have also highlighted potential adipogenic regulatory functions of the mesothelium in human adipose tissue[268]^38 and identified subsets that correlate with the metabolic status of patients with obesity.[269]^39 Despite these findings, their role in obesity-induced fibrosis remains poorly understood. Here, we show that the expansion and fibrotic transformation of vWAT during obesity development are accompanied by a marked increase in MC density along with elevated fibrosis marker expression and increased perilobular fibrosis. This phenomenon may be associated with the upregulation of TGF-β or PDGFRα signaling,[270]^7 both of which are known to be stimulated in obesity[271]^8^,[272]^24 and to promote fibrosis.[273]^23^,[274]^25 These findings strongly suggest that MCs hold the potential to contribute to perilobular fibrosis in vWAT. While the specific impact of perilobular fibrosis on adipose tissue function is unclear, its potential to exacerbate tissue dysfunction cannot be ruled out. Notably, in human vWAT, we observed that the mesothelial layer not only is present on the surface of the organ but also surrounds the structural units known as lobules. This observation further suggests that MC-related perilobular fibrosis may contribute to unhealthy adipose tissue expansion under obesogenic conditions as shown for peri-adipocyte fibrosis.[275]^7 Importantly, the drug combination effectively inhibits the obesity-induced fibrotic transformation of both CD9^+ Ly-6c^+ Progs and Ly-6c^− MCs in the C3H mouse model, limiting their expansion and transcriptional activation. Nin+Cel may induce changes in MCs to support their native phenotype and have beneficial effects on maintaining vWAT homeostasis. Given the reciprocal relationship between fibrogenesis and adipogenesis,[276]^40^,[277]^41 we investigated whether Nin+Cel treatment could also modify the expansion patterns of EpiWAT and IngWAT. Specifically, we examined whether our intervention could mitigate adipogenic potential. In EpiWAT, we and others have previously identified CD9^− Progs as preadipocytes, notably using adipogenesis assays.[278]^8^,[279]^9 Analysis of the CD9^− Prog pool in visceral adipose tissue (AT) revealed that the anti-fibrotic treatment prevented the loss of these preadipocytes when administered prophylactically. However, when Nin+Cel was administered after fibrosis was established and CD9^− Prog loss had already occurred, the treatment failed to regenerate this pre-adipocyte pool. Therefore, the anti-fibrotic treatment does not directly affect adipogenic capacity in EpiWAT. In contrast, in IngWAT, Nin+Cel promoted CD36^+ Prog expansion and increased IngWAT mass without altering adipocyte size, indicating a depot-specific pro-adipogenic effect that could potentially contribute to healthier adipose tissue expansion and metabolic status.[280]^40 In summary, our findings underscore the therapeutic potential of targeting fibrogenic Progs to preserve adipose tissue function. This substantiates the pivotal role Progs play in shaping tissue functions and influencing metabolic well-being. Moreover, we shed light on an additional role for MCs in vWAT fibrosis. Future studies using lineage tracing and targeted manipulation are needed to better define the peculiar role of the mesothelium in visceral fat homeostasis. Limitations of the study This study has several limitations. First, the stratification of patients based on CD9^+ Prog frequency in omental adipose tissue was performed using samples collected at the time of surgery. While longitudinal monitoring during weight loss would have been informative, post-operative biopsies of visceral fat are ethically constrained. Second, all in vivo experiments were performed in male mice, precluding the assessment of gender-specific responses. Finally, while fibrotic MCs are likely contributors to the impaired remodeling observed in EpiWAT of obese mice, their specific role could not be directly assessed due to the low recombination efficiency of the Krt19-CreERT2 mouse model. Resource availability Lead contact Requests for further information, resources, and reagents should be directed to and will be fulfilled by the lead contact, Geneviève Marcelin (genevieve.marcelin@inserm.fr). Materials availability This study did not generate new unique reagents. Data and code availability * • Data: The microarray analysis and RNA-seq data have been deposited with the dataset identifier EMBL-EBI: E-MTAB-13669 and EMBL-EBI: E-MTAB-13670. * • Code: This paper does not report original code. * • Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request. Acknowledgments