Research ArticleImmunologyMetabolism Open Access | 10.1172/jci.insight.193837

Julia Chini,1,2,3 Nicole DeMarco,2 Dana V. Mitchell,4 Sam J. McCright,1,2,3 Kaitlyn M. Shen,5 Divyansi Pandey,1,2,3 Rachel L. Clement,1,2,3 Jessica Miller,2 Rajan Jain,5 Deanne M. Taylor,1,4 Mitchell A. Lazar,6,7 and David A. Hill1,2,3,7

1Department of Pediatrics, Perelman School of Medicine, University of Pennsylvania, Philadelphia, Pennsylvania, USA.

2Division of Allergy and Immunology, Children’s Hospital of Philadelphia, Philadelphia, Pennsylvania, USA.

3Institute for Immunology and Immune Health, Perelman School of Medicine, University of Pennsylvania, Philadelphia, Pennsylvania, USA.

4Department of Biomedical and Health Informatics, Children’s Hospital of Philadelphia, Philadelphia, Pennsylvania, USA.

5Departments of Medicine and Cell and Developmental Biology, Penn Cardiovascular Institute, Penn Epigenetics Institute,

6Department of Medicine, Division of Endocrinology, and

7Institute for Diabetes, Obesity and Metabolism, Perelman School of Medicine, University of Pennsylvania, Philadelphia, Pennsylvania, USA.

Address correspondence to: David A. Hill, Division of Allergy and Immunology, Children’s Hospital of Philadelphia, Abramson Research Building, 1208B, 3615 Civic Center Blvd., Philadelphia, Pennsylvania, 19104, USA. Phone: 215.590.2549; Email: hilld3@chop.edu.

Find articles by Chini, J. in: PubMed | Google Scholar

1Department of Pediatrics, Perelman School of Medicine, University of Pennsylvania, Philadelphia, Pennsylvania, USA.

2Division of Allergy and Immunology, Children’s Hospital of Philadelphia, Philadelphia, Pennsylvania, USA.

3Institute for Immunology and Immune Health, Perelman School of Medicine, University of Pennsylvania, Philadelphia, Pennsylvania, USA.

4Department of Biomedical and Health Informatics, Children’s Hospital of Philadelphia, Philadelphia, Pennsylvania, USA.

5Departments of Medicine and Cell and Developmental Biology, Penn Cardiovascular Institute, Penn Epigenetics Institute,

6Department of Medicine, Division of Endocrinology, and

7Institute for Diabetes, Obesity and Metabolism, Perelman School of Medicine, University of Pennsylvania, Philadelphia, Pennsylvania, USA.

Address correspondence to: David A. Hill, Division of Allergy and Immunology, Children’s Hospital of Philadelphia, Abramson Research Building, 1208B, 3615 Civic Center Blvd., Philadelphia, Pennsylvania, 19104, USA. Phone: 215.590.2549; Email: hilld3@chop.edu.

Find articles by DeMarco, N. in: PubMed | Google Scholar

1Department of Pediatrics, Perelman School of Medicine, University of Pennsylvania, Philadelphia, Pennsylvania, USA.

2Division of Allergy and Immunology, Children’s Hospital of Philadelphia, Philadelphia, Pennsylvania, USA.

3Institute for Immunology and Immune Health, Perelman School of Medicine, University of Pennsylvania, Philadelphia, Pennsylvania, USA.

4Department of Biomedical and Health Informatics, Children’s Hospital of Philadelphia, Philadelphia, Pennsylvania, USA.

5Departments of Medicine and Cell and Developmental Biology, Penn Cardiovascular Institute, Penn Epigenetics Institute,

6Department of Medicine, Division of Endocrinology, and

7Institute for Diabetes, Obesity and Metabolism, Perelman School of Medicine, University of Pennsylvania, Philadelphia, Pennsylvania, USA.

Address correspondence to: David A. Hill, Division of Allergy and Immunology, Children’s Hospital of Philadelphia, Abramson Research Building, 1208B, 3615 Civic Center Blvd., Philadelphia, Pennsylvania, 19104, USA. Phone: 215.590.2549; Email: hilld3@chop.edu.

Find articles by Mitchell, D. in: PubMed | Google Scholar

1Department of Pediatrics, Perelman School of Medicine, University of Pennsylvania, Philadelphia, Pennsylvania, USA.

2Division of Allergy and Immunology, Children’s Hospital of Philadelphia, Philadelphia, Pennsylvania, USA.

3Institute for Immunology and Immune Health, Perelman School of Medicine, University of Pennsylvania, Philadelphia, Pennsylvania, USA.

4Department of Biomedical and Health Informatics, Children’s Hospital of Philadelphia, Philadelphia, Pennsylvania, USA.

5Departments of Medicine and Cell and Developmental Biology, Penn Cardiovascular Institute, Penn Epigenetics Institute,

6Department of Medicine, Division of Endocrinology, and

7Institute for Diabetes, Obesity and Metabolism, Perelman School of Medicine, University of Pennsylvania, Philadelphia, Pennsylvania, USA.

Address correspondence to: David A. Hill, Division of Allergy and Immunology, Children’s Hospital of Philadelphia, Abramson Research Building, 1208B, 3615 Civic Center Blvd., Philadelphia, Pennsylvania, 19104, USA. Phone: 215.590.2549; Email: hilld3@chop.edu.

Find articles by McCright, S. in: PubMed | Google Scholar

1Department of Pediatrics, Perelman School of Medicine, University of Pennsylvania, Philadelphia, Pennsylvania, USA.

2Division of Allergy and Immunology, Children’s Hospital of Philadelphia, Philadelphia, Pennsylvania, USA.

3Institute for Immunology and Immune Health, Perelman School of Medicine, University of Pennsylvania, Philadelphia, Pennsylvania, USA.

4Department of Biomedical and Health Informatics, Children’s Hospital of Philadelphia, Philadelphia, Pennsylvania, USA.

5Departments of Medicine and Cell and Developmental Biology, Penn Cardiovascular Institute, Penn Epigenetics Institute,

6Department of Medicine, Division of Endocrinology, and

7Institute for Diabetes, Obesity and Metabolism, Perelman School of Medicine, University of Pennsylvania, Philadelphia, Pennsylvania, USA.

Address correspondence to: David A. Hill, Division of Allergy and Immunology, Children’s Hospital of Philadelphia, Abramson Research Building, 1208B, 3615 Civic Center Blvd., Philadelphia, Pennsylvania, 19104, USA. Phone: 215.590.2549; Email: hilld3@chop.edu.

Find articles by Shen, K. in: PubMed | Google Scholar

1Department of Pediatrics, Perelman School of Medicine, University of Pennsylvania, Philadelphia, Pennsylvania, USA.

2Division of Allergy and Immunology, Children’s Hospital of Philadelphia, Philadelphia, Pennsylvania, USA.

3Institute for Immunology and Immune Health, Perelman School of Medicine, University of Pennsylvania, Philadelphia, Pennsylvania, USA.

4Department of Biomedical and Health Informatics, Children’s Hospital of Philadelphia, Philadelphia, Pennsylvania, USA.

5Departments of Medicine and Cell and Developmental Biology, Penn Cardiovascular Institute, Penn Epigenetics Institute,

6Department of Medicine, Division of Endocrinology, and

7Institute for Diabetes, Obesity and Metabolism, Perelman School of Medicine, University of Pennsylvania, Philadelphia, Pennsylvania, USA.

Address correspondence to: David A. Hill, Division of Allergy and Immunology, Children’s Hospital of Philadelphia, Abramson Research Building, 1208B, 3615 Civic Center Blvd., Philadelphia, Pennsylvania, 19104, USA. Phone: 215.590.2549; Email: hilld3@chop.edu.

Find articles by Pandey, D. in: PubMed | Google Scholar

1Department of Pediatrics, Perelman School of Medicine, University of Pennsylvania, Philadelphia, Pennsylvania, USA.

2Division of Allergy and Immunology, Children’s Hospital of Philadelphia, Philadelphia, Pennsylvania, USA.

3Institute for Immunology and Immune Health, Perelman School of Medicine, University of Pennsylvania, Philadelphia, Pennsylvania, USA.

4Department of Biomedical and Health Informatics, Children’s Hospital of Philadelphia, Philadelphia, Pennsylvania, USA.

5Departments of Medicine and Cell and Developmental Biology, Penn Cardiovascular Institute, Penn Epigenetics Institute,

6Department of Medicine, Division of Endocrinology, and

7Institute for Diabetes, Obesity and Metabolism, Perelman School of Medicine, University of Pennsylvania, Philadelphia, Pennsylvania, USA.

Address correspondence to: David A. Hill, Division of Allergy and Immunology, Children’s Hospital of Philadelphia, Abramson Research Building, 1208B, 3615 Civic Center Blvd., Philadelphia, Pennsylvania, 19104, USA. Phone: 215.590.2549; Email: hilld3@chop.edu.

Find articles by Clement, R. in: PubMed | Google Scholar

1Department of Pediatrics, Perelman School of Medicine, University of Pennsylvania, Philadelphia, Pennsylvania, USA.

2Division of Allergy and Immunology, Children’s Hospital of Philadelphia, Philadelphia, Pennsylvania, USA.

3Institute for Immunology and Immune Health, Perelman School of Medicine, University of Pennsylvania, Philadelphia, Pennsylvania, USA.

4Department of Biomedical and Health Informatics, Children’s Hospital of Philadelphia, Philadelphia, Pennsylvania, USA.

5Departments of Medicine and Cell and Developmental Biology, Penn Cardiovascular Institute, Penn Epigenetics Institute,

6Department of Medicine, Division of Endocrinology, and

7Institute for Diabetes, Obesity and Metabolism, Perelman School of Medicine, University of Pennsylvania, Philadelphia, Pennsylvania, USA.

Address correspondence to: David A. Hill, Division of Allergy and Immunology, Children’s Hospital of Philadelphia, Abramson Research Building, 1208B, 3615 Civic Center Blvd., Philadelphia, Pennsylvania, 19104, USA. Phone: 215.590.2549; Email: hilld3@chop.edu.

Find articles by Miller, J. in: PubMed | Google Scholar

1Department of Pediatrics, Perelman School of Medicine, University of Pennsylvania, Philadelphia, Pennsylvania, USA.

2Division of Allergy and Immunology, Children’s Hospital of Philadelphia, Philadelphia, Pennsylvania, USA.

3Institute for Immunology and Immune Health, Perelman School of Medicine, University of Pennsylvania, Philadelphia, Pennsylvania, USA.

4Department of Biomedical and Health Informatics, Children’s Hospital of Philadelphia, Philadelphia, Pennsylvania, USA.

5Departments of Medicine and Cell and Developmental Biology, Penn Cardiovascular Institute, Penn Epigenetics Institute,

6Department of Medicine, Division of Endocrinology, and

7Institute for Diabetes, Obesity and Metabolism, Perelman School of Medicine, University of Pennsylvania, Philadelphia, Pennsylvania, USA.

Address correspondence to: David A. Hill, Division of Allergy and Immunology, Children’s Hospital of Philadelphia, Abramson Research Building, 1208B, 3615 Civic Center Blvd., Philadelphia, Pennsylvania, 19104, USA. Phone: 215.590.2549; Email: hilld3@chop.edu.

Find articles by Jain, R. in: PubMed | Google Scholar

1Department of Pediatrics, Perelman School of Medicine, University of Pennsylvania, Philadelphia, Pennsylvania, USA.

2Division of Allergy and Immunology, Children’s Hospital of Philadelphia, Philadelphia, Pennsylvania, USA.

3Institute for Immunology and Immune Health, Perelman School of Medicine, University of Pennsylvania, Philadelphia, Pennsylvania, USA.

4Department of Biomedical and Health Informatics, Children’s Hospital of Philadelphia, Philadelphia, Pennsylvania, USA.

5Departments of Medicine and Cell and Developmental Biology, Penn Cardiovascular Institute, Penn Epigenetics Institute,

6Department of Medicine, Division of Endocrinology, and

7Institute for Diabetes, Obesity and Metabolism, Perelman School of Medicine, University of Pennsylvania, Philadelphia, Pennsylvania, USA.

Address correspondence to: David A. Hill, Division of Allergy and Immunology, Children’s Hospital of Philadelphia, Abramson Research Building, 1208B, 3615 Civic Center Blvd., Philadelphia, Pennsylvania, 19104, USA. Phone: 215.590.2549; Email: hilld3@chop.edu.

Find articles by Taylor, D. in: PubMed | Google Scholar |

1Department of Pediatrics, Perelman School of Medicine, University of Pennsylvania, Philadelphia, Pennsylvania, USA.

2Division of Allergy and Immunology, Children’s Hospital of Philadelphia, Philadelphia, Pennsylvania, USA.

3Institute for Immunology and Immune Health, Perelman School of Medicine, University of Pennsylvania, Philadelphia, Pennsylvania, USA.

4Department of Biomedical and Health Informatics, Children’s Hospital of Philadelphia, Philadelphia, Pennsylvania, USA.

5Departments of Medicine and Cell and Developmental Biology, Penn Cardiovascular Institute, Penn Epigenetics Institute,

6Department of Medicine, Division of Endocrinology, and

7Institute for Diabetes, Obesity and Metabolism, Perelman School of Medicine, University of Pennsylvania, Philadelphia, Pennsylvania, USA.

Address correspondence to: David A. Hill, Division of Allergy and Immunology, Children’s Hospital of Philadelphia, Abramson Research Building, 1208B, 3615 Civic Center Blvd., Philadelphia, Pennsylvania, 19104, USA. Phone: 215.590.2549; Email: hilld3@chop.edu.

Find articles by Lazar, M. in: PubMed | Google Scholar

1Department of Pediatrics, Perelman School of Medicine, University of Pennsylvania, Philadelphia, Pennsylvania, USA.

2Division of Allergy and Immunology, Children’s Hospital of Philadelphia, Philadelphia, Pennsylvania, USA.

3Institute for Immunology and Immune Health, Perelman School of Medicine, University of Pennsylvania, Philadelphia, Pennsylvania, USA.

4Department of Biomedical and Health Informatics, Children’s Hospital of Philadelphia, Philadelphia, Pennsylvania, USA.

5Departments of Medicine and Cell and Developmental Biology, Penn Cardiovascular Institute, Penn Epigenetics Institute,

6Department of Medicine, Division of Endocrinology, and

7Institute for Diabetes, Obesity and Metabolism, Perelman School of Medicine, University of Pennsylvania, Philadelphia, Pennsylvania, USA.

Address correspondence to: David A. Hill, Division of Allergy and Immunology, Children’s Hospital of Philadelphia, Abramson Research Building, 1208B, 3615 Civic Center Blvd., Philadelphia, Pennsylvania, 19104, USA. Phone: 215.590.2549; Email: hilld3@chop.edu.

Find articles by Hill, D. in: PubMed | Google Scholar |

Published February 10, 2026 - More info

Published in Volume 11, Issue 6 on March 23, 2026
JCI Insight. 2026;11(6):e193837. https://doi.org/10.1172/jci.insight.193837.
© 2026 Chini et al. This work is licensed under the Creative Commons Attribution 4.0 International License. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/. Published February 10, 2026 - Version history
Received: March 25, 2025; Accepted: January 28, 2026 View PDF Abstract

Dysfunctional white adipose tissue contributes to the development of obesity-related morbidities, including insulin resistance, dyslipidemia, and other metabolic disorders. Adipose tissue macrophages (ATMs) accumulate in obesity and play both beneficial and harmful roles in the maintenance of adipose tissue homeostasis and function. Despite their importance, the molecules and mechanisms that regulate these diverse functions are not well understood. Lipid-associated macrophages (LAMs), the dominant subset of obesity-associated ATMs, accumulate in crown-like structures and are characterized by a metabolically activated and proinflammatory phenotype. We previously identified CD9 as a surface marker of LAMs. However, the contribution of CD9 to the activation and function of LAMs during obesity is unknown. Using a myeloid-specific CD9-KO model, we show that CD9 supports ATM-adipocyte adhesion and crown-like structure formation. Furthermore, CD9 promotes the expression of profibrotic and extracellular matrix remodeling genes. Loss of myeloid CD9 reduces adipose tissue fibrosis, increases visceral adipose tissue accumulation, and improves global metabolic outcomes during diet-induced obesity. These results identify CD9 as a causal regulator of pathogenic LAM functions, highlighting CD9 as a potential therapeutic target for treating obesity-associated metabolic disease.

Graphical Abstract Introduction

White adipose tissue undergoes dramatic remodeling in response to excess caloric intake, resulting in tissue expansion, inflammation, and fibrosis. This pathological remodeling promotes insulin resistance and ectopic lipid deposition in peripheral organs, predisposing individuals to the development of metabolic sequelae such as metabolic dysfunction–associated fatty liver disease, atherosclerosis, and type 2 diabetes (1). Although newer therapeutics, including glucagon-like peptide-1 (GLP-1) receptor agonists, can induce meaningful weight loss, emerging evidence indicates that the inflammatory and fibrotic features of adipose tissue remodeling often persist (2–6). Defining the mechanisms that drive adipose tissue remodeling is therefore essential for developing more effective strategies to prevent and treat metabolic disease.

In response to excess caloric intake, adipose tissue initially undergoes adaptive expansion through adipocyte hypertrophy and hyperplasia, thereby increasing its storage capacity for lipids. However, with prolonged overnutrition, adipocyte hypertrophy can outpace angiogenesis, leading to tissue hypoxia and adipocyte death (7). Adipocyte dysfunction and apoptosis promote the infiltration and proliferation of immune cells, particularly adipose tissue macrophages (ATMs). ATMs accumulate in crown-like structures surrounding dead and dying adipocytes, where they support tissue homeostasis through the efferocytosis of apoptotic adipocytes and the uptake of free fatty acids released during adipocyte lipolysis (8–12). At the same time, ATMs can contribute to adipose tissue dysfunction by producing pro-inflammatory and profibrotic mediators (10, 13–19). Despite their central role in adipose tissue remodeling, the molecular pathways that direct beneficial versus pathogenic ATM functions remain poorly understood.

During obesity, a subset of ATMs acquires a metabolically activated phenotype characterized by increased lipid accumulation and processing, fatty acid–dependent cellular metabolism, and proinflammatory cytokine production (9, 10, 13, 14, 20). This population, termed lipid-associated macrophages (LAMs), arises largely from recruited monocytes and becomes the dominant ATM subset in adipose tissue during obesity (19). LAMs localize to crown-like structures surrounding dead and dying adipocytes, where they mediate efferocytosis and lipid handling in part through the lipid receptor TREM2 (11, 21, 22).

We previously identified CD9 as a surface marker of LAMs in both mice and humans (11, 23). CD9 is a tetraspanin molecule that recruits and stabilizes proteins such as integrins, cytokine receptors, and metalloproteinases to tetraspanin-enriched microdomains (24–28). Through these interactions, CD9 contributes to cellular adhesion, migration, signal transduction, and extracellular vesicle formation (24, 26, 29–32). Since our initial report, CD9 has been widely used to identify LAMs across multiple other tissues, including the liver and aorta (33–38). Scar-associated macrophages (SAMs), which accumulate during liver and lung fibrosis, also share features with LAMs, including the expression of CD9 (36, 37, 39, 40). Despite the strong association between macrophage CD9 expression and metabolic dysfunction, the influence of CD9 on LAM functions during obesity remains unknown.

To investigate the functional role of CD9 in LAMs during obesity, we developed and characterized a myeloid-specific CD9-KO mouse model. Using this model, we identified a role for CD9 in mediating macrophage adhesion to adipocytes, which is critical for crown-like structure formation during obesity. Moreover, we found that CD9 promotes the expression of fibrosis-associated genes in macrophages. Consequently, loss of CD9 in myeloid cells led to attenuated adipose tissue remodeling and fibrosis, resulting in improved systemic metabolic outcomes during diet-induced obesity. Together, these findings establish CD9 as a pathogenic regulator of LAM function, suggesting that CD9 may be a potential therapeutic target for obesity-associated metabolic disease.

Results

CD9+ ATMs display a transcriptional profile enriched for extracellular matrix remodeling and adhesion pathways. We previously compared the transcriptional profiles of CD9+ ATMs and Ly6c+ monocytes and found that CD9+ ATMs were enriched for proinflammatory and lysosomal-dependent lipid metabolism gene programs (23). During obesity, CD9+ ATMs are thought to arise from monocytes via a CD9– ATM transitional state (11, 41). To further investigate the role of CD9 in ATMs, we compared the transcriptomes of sort-purified CD9+ and CD9– ATMs isolated from the visceral epididymal white adipose tissue (eWAT) of WT mice fed a high-fat diet (HFD) for 12 weeks (23). We found that 97 genes were upregulated and 84 genes were downregulated in CD9+ ATMs compared with CD9– ATMs (Figure 1A). CD9– ATMs demonstrated increased expression of genes encoding pattern recognition receptors (Cd209b, Cd209f, and Cd209g) and genes associated with antiinflammatory and perivascular macrophages (Lyve1, Retnla, Cd163, and Ednrb) (42). In contrast, CD9+ ATMs were enriched for genes previously identified in LAMs and SAMs, including Spp1, Gpnmb, Mmp12, Fabp5, Lpl, Cd36, and Lgals3 (9, 11, 13, 14, 36, 43) (Figure 1, A and B).

Figure 1

CD9+ adipose tissue macrophages have increased expression of extracellular matrix remodeling and adhesion genes. Bulk RNA-Seq analysis of CD9+ and CD9– adipose tissue macrophages (ATMs) sorted from epididymal white adipose tissue (eWAT) of WT male mice fed an HFD for 12 weeks (n = 3–5). (A) Volcano plot of gene expression in CD9+ versus CD9– ATMs showing the log2 fold-change (FC; x axis) and adjusted P value (–log10 FDR; y axis) of genes. Significantly differentially expressed genes (log2 FC > 0.58 and FDR < 0.05) are shown in purple. (B) Differential expression of genes previously associated with CD9+ macrophages (log2 FC of CD9+ ATMs/CD9– ATMs). Genes are listed in order from highest to lowest log2 FC. (C) Gene set enrichment analysis showing all Reactome pathways significantly upregulated in CD9+ ATMs compared with CD9– ATMs. (D) Heatmap of differentially expressed genes (FDR > 0.1) from Extracellular Matrix Organization Pathway (Reactome: R-MMU-1474244). P2RY1, p2Y purinoreceptor 1.

Gene set enrichment analysis revealed that CD9+ ATMs were enriched for pathways involved in extracellular matrix (ECM) remodeling, including programs related to both ECM formation and degradation (Figure 1, C and D). Furthermore, CD9+ ATMs were enriched for genes associated with cell adhesion, including Ncam1 (CD56) and integrin-related interactions (Figure 1C). These data indicate that CD9+ ATMs have a transcriptional profile characterized by enhanced expression of ECM-remodeling and adhesion-associated genes.

CD9 stabilizes ATM-adipocyte interactions in vivo. During obesity, CD9+ LAMs accumulate in adipose tissue in crown-like structures surrounding dead or dying adipocytes (44–46). We therefore investigated whether CD9 regulates ATM accumulation and localization in adipose tissue during obesity. To assess the functional role of CD9 in macrophages, we first generated a Cd9 conditional KO mouse line (Cd9fl/fl) by inserting loxP sites in the endogenous Cd9 locus (Supplemental Figure 1A; supplemental material available online with this article; https://doi.org/10.1172/jci.insight.193837DS1). Cd9fl/fl mice were then crossed to mice expressing Cre recombinase under the endogenous Lyz2 promoter (LysMCre) to generate a myeloid-specific CD9-KO (CD9-MKO) mouse model (47). Successful insertion of loxP sites was confirmed by DNA PCR genotyping (Supplemental Figure 1B). A reduction in gene expression of Cd9 and cell-surface expression of CD9 was confirmed by qPCR and flow cytometry of unpolarized BM-derived macrophages (BMDMs) (Supplemental Figure 1, C and D). Because CD9-MKO BMDMs exhibited only partial CD9 deletion, we also crossed Cd9fl/fl mice with Vav-iCre mice to generate a hematopoietic-wide CD9 KO (48). BMDMs from Cd9fl/fl Vav-iCre+/– mice had complete ablation of CD9 expression, thereby confirming the efficiency and fidelity of the Cd9fl/fl conditional allele (Supplemental Figure 1, E and F).

Next, we placed CD9-MKO and control (Cd9fl/fl) mice on an HFD for 12 weeks and examined the visceral eWAT depot. Since female mice are known to be resistant to adipose tissue inflammation during diet-induced obesity (49–51), initial studies were performed in male mice. Immunofluorescence staining of eWAT sections revealed decreased expression of the pan-macrophage markers F4/80 and CD68 in CD9-MKO HFD mice compared with control HFD mice. In addition, the number of F4/80+ and CD68+ cells were also reduced in CD9-MKO HFD mice compared with control HFD mice (Figure 2A and Supplemental Figure 2A). Further, CD9-MKO HFD eWAT also displayed reduced expression of the LAM markers TREM2 and OPN compared with control HFD eWAT (Supplemental Figure 2, B and C). Overall, these data indicate that myeloid-specific deletion of CD9 decreases LAM accumulation in eWAT during diet-induced obesity.

Figure 2

CD9 promotes adipose tissue macrophage accumulation, crown-like structure formation, and adipocyte interactions. Analysis of epididymal white adipose tissue (eWAT) from control (Cd9fl/fl) and CD9-MKO (Cd9fl/fl LysMCre+/–) male mice fed an HFD for 12 weeks. (A) Representative images and quantification of immunofluorescence staining of F4/80 and DAPI in eWAT tissue sections. Data shown as relative MFI of F4/80 (n = 6–8) and number of F4/80+ DAPI+ cells per mm2 (n = 8). Scale bar: 200 μm. (B) Quantification by flow cytometry of adipose tissue macrophages (ATMs) from control and CD9-MKO HFD mice. Data are shown as frequency of total CD45+ immune cells (n = 15–16) and cells per gram of tissue (n = 13–16). (C) Representative high-resolution images of eWAT tissue sections from control and CD9-MKO HFD mice stained with DAPI (blue), PLIN1 (green), F4/80 (red), and CD9 (yellow) (n = 8). Scale bar: 50 μm. (D) F4/80+ crown-like structures (CLSs) per mm2 quantified from images from samples shown in A (n = 6–8). (E) 3D image reconstruction of PLIN1 (green), F4/80 (red), and CD9 (yellow) from control HFD mice (n = 8). Scale bar: 50 μm (left) or 20 μm (right). (F) Representative images of adipocyte fraction isolated from control HFD mice eWAT and stained for Hoechst (blue), BODIPY (green), F4/80 (red), and CD9 (yellow) (n = 3). Scale bar: 50 μm. (G and H) Relative expression of Adgre1 (F4/80), Cd68, Cd9, Trem2, and Spp1 (OPN) in stromal vascular fraction (n = 6–9; G) and adipocyte fraction (n = 6–9; H) isolated from eWAT. Data are from 2 (A, B, D, and E) or 3 (C, and F–H) independent pooled experiments. Data presented as mean ± SEM (A–C, G, and H). qPCR shown as ΔCt relative to a normalization factor relative to control HFD samples. Unpaired 2-tailed Student’s t test (A–C, G, and H). ns, not significant; *P < 0.05, **P < 0.01, ***P < 0.001.

Paradoxically, flow cytometric analyses of the stromal vascular fraction revealed no difference in the frequency or number of total ATMs between CD9-MKO and control HFD mice, despite a significant reduction in CD9 expression in ATMs and other myeloid cells in CD9-MKO HFD mice (Figure 2B, Supplemental Figure 3, and Supplemental Figure 4, A–E). Likewise, the total number of CD45+ immune cells and the distribution of other immune cell subsets in the stromal vascular fraction were unchanged (Supplemental Figure 4, F and G).

To explain the discrepancy between our immunofluorescence and flow cytometry findings, we hypothesized that a subset of ATMs might be lost during stromal vascular fraction preparation. Prior work has shown that some ATMs can remain tightly adherent to adipocytes and therefore partition into the adipocyte fraction, which is typically discarded during flow cytometry processing (52). To assess whether CD9+ ATMs associate with adipocytes, we performed high-resolution imaging of eWAT from control HFD and CD9-MKO HFD mice. Consistent with our prior studies (11, 23), CD9+ ATMs localized to crown-like structures surrounding adipocytes, including those lacking Perilipin-1 (PLIN1) expression, a feature of dying adipocytes (Figure 2C and Supplemental Figure 5) (53). In addition, CD9-MKO HFD mice exhibited fewer F4/80+ crown-like structures compared with control HFD mice (Figure 2, C and D). Three-dimensional reconstruction further confirmed that CD9+ ATMs reside in close proximity to adipocytes and are thus appropriately positioned for direct adhesion (Figure 2E).

Since CD9+ ATMs were closely associated with adipocytes, we hypothesized that CD9+ ATMs may be found within the adipocyte fraction. Indeed, we observed that CD9+ F4/80+ ATMs could be found attached to adipocytes within the floating adipocyte fraction (Figure 2F). To test whether ATM-intrinsic CD9 influences this retention, we quantified macrophage markers in both the stromal vascular fraction and adipocyte fraction from eWAT of control HFD and CD9-MKO HFD mice. Consistent with flow cytometry data, there were minimal differences in the expression of pan-macrophage markers (Adgre1 [F4/80] and Cd68) or LAM markers (Cd9, Trem2, and Spp1 [OPN]) in the stromal vascular fraction (Figure 2G). However, all 5 markers were reduced in the eWAT adipocyte fraction from CD9-MKO HFD mice compared with control HFD mice (Figure 2H). These findings demonstrate that macrophage-intrinsic CD9 promotes ATM-adipocyte interactions and is essential for the efficient formation of crown-like structures during obesity.

CD9 regulates integrin expression and macrophage-adipocyte adhesion. To investigate the functional role of CD9 in macrophage-adipocyte adhesion, we utilized an in vitro model in which BMDMs were metabolically activated with palmitate. Palmitate-treated BMDMs recapitulate key transcriptional and functional features of obesity-associated ATMs (9, 20). However, it is not known whether CD9 expression is induced by this metabolic stimulus. Metabolic activation with palmitate strongly induced the expression of Cd9 transcript and surface CD9 protein expression in BMDMs (Figure 3, A and B, and Supplemental Figure 6, A–C). CD9 deletion did not affect BMDM survival, differentiation, or proliferation of palmitate-treated BMDMs (Supplemental Figure 6, D–F). Further, palmitate also induced Cd9 expression in sort-purified blood monocytes from WT mice fed a normal chow diet (NCD; Figure 3C). Together, these results suggest that CD9 upregulation is a general feature of monocyte and macrophage metabolic activation.

Figure 3

CD9 promotes macrophage integrin expression and adhesion. (A and B) BMDMs from control (Cd9fl/fl) and CD9-MKO (Cd9fl/fl LysMCre+/–) mice were treated with BSA or palmitate for 24 hours. (A) Expression of Cd9 was assessed by qPCR (n = 6). Shown as expression relative to BSA controls (Ctrl). (B) Surface expression of CD9 shown as relative MFI compared with BSA controls. BMDMs were gated as live CD45+ CD11b+ F4/80+ CD64+ cells (n = 4). (C) Blood monocytes isolated from WT mice fed a normal chow diet were treated with BSA or palmitate for 24 hours. Expression of Cd9 was assessed by qPCR (n = 7–8). Monocytes were gated using the gating strategy in Supplemental Figure 3. (D) Expression of integrins Itgax and Itgav by qPCR in BMDMs from control and CD9-MKO mice treated with BSA or palmitate (n = 6). (E) Flow cytometry of CD11c (Itgax) and CD51 (Itgav) shown as MFI relative to control BMDMs treated with palmitate (n = 5). (F) Adhesion assay of control or CD9-MKO BMDMs on differentiated 3T3-L1 adipocytes (n = 6). (G) Relative expression of Itgax and Itgav in epididymal white adipose tissue (eWAT) adipocyte fraction isolated from control and CD9-MKO mice fed an HFD (n = 6–9). (H and I) Representative images and quantification of immunofluorescence staining of CD11c (H) and CD51 (I) in eWAT. Data shown as relative MFI (n = 8–9). Scale bar: 200 μm. Data shown as combined data from 2–3 independent experiments (C, G–I) or representative data from 3 independent experiments (A, B, and D–F). qPCR is shown as ΔCt relative to Hprt normalized to BSA controls. Data presented as mean ± SEM (A–I); 2-way ANOVA with Fisher’s LSD test (A, B, and D) or unpaired 2-tailed Student’s t test (C, and E–I). ns, not significant; **P < 0.01, ***P < 0.001, ****P < 0.0001.

Given the established role of CD9 in regulating cell-cell adhesion, we hypothesized that CD9 may promote macrophage-adipocyte adhesion by recruiting and stabilizing integrins at the cell surface (26, 30, 31). Consistent with this idea, metabolic activation of BMDMs with palmitate increased expression of the integrins Itgax (CD11c) and Itgav (CD51) in a CD9-dependent manner (Figure 3D). Further, palmitate-treated BMDMs from CD9-MKO mice had reduced surface levels of CD11c and CD51 compared with palmitate-treated control BMDMs (Figure 3E). As it has previously been described that ATMs form tight adhesions to adipocytes via integrin interactions, we next assessed whether CD9 affects this interaction (52, 54). We found that CD9-MKO BMDMs had reduced adhesion to 3T3-L1 adipocytes compared with control BMDMs, identifying a key role for CD9 in macrophage-adipocyte interactions (Figure 3F).

Next, to determine whether CD9 deletion also alters integrin expression in vivo, we assessed Itgax and Itgav levels in eWAT of control and CD9-MKO HFD mice. Consistent with our in vitro data, deletion of CD9 reduced Itgax and Itgav expression in the eWAT adipocyte fraction (Figure 3G). On the other hand, Itgax but not Itgav was decreased in the eWAT stromal vascular fraction of CD9-MKO HFD mice compared with control HFD mice (Supplemental Figure 6G). Immunofluorescence staining similarly showed decreased CD11c and CD51 protein expression in CD9-MKO HFD eWAT compared with controls (Figure 3, H and I). Overall, these results suggest that CD9 regulates integrin expression in metabolically activated macrophages, both in vitro and in vivo, providing a potential mechanism by which CD9 influences macrophage-adipocyte interactions.

Myeloid-intrinsic CD9 mediates adipose tissue remodeling during obesity. Given that ATMs are known to have both beneficial and pathological effects on adipose tissue adaptation in obesity, we next assessed how myeloid-specific CD9 deletion influences adipose tissue remodeling during obesity. To investigate this, we examined eWAT from CD9-MKO and control mice fed either an NCD or an HFD for 12 weeks. CD9-MKO mice exhibited slightly increased eWAT depots following HFD, but not NCD, compared with controls (Figure 4A). Histological analysis further demonstrated larger adipocyte size in CD9-MKO HFD eWAT compared with controls (Figure 4, B and C, and Supplemental Figure 7, A and B).

Figure 4

Myeloid-intrinsic CD9 alters adipose tissue remodeling during obesity. Epididymal white adipose tissue (eWAT) was collected from male control (Cd9fl/fl) and CD9-MKO (Cd9fl/fl LysMCre+/–) mice fed either an NCD or HFD for 12 weeks. (A) eWAT mass (n = 10–15). (B) Representative H&E staining of eWAT in control or CD9-MKO HFD mice (n = 11–12). Scale bar: 400 μm. (C) Frequency distribution and mean adipocyte diameter calculated from H&E images in B (n = 11–12). (D and E) Bulk RNA-Seq was performed on eWAT depots from control and CD9-MKO mice (n = 4). (D) Volcano plot showing the log2 fold-change (FC; x axis) and adjusted P value (–log10 FDR; y axis) of genes between CD9-MKO HFD and control HFD groups. Significantly differentially expressed genes (log2 FC > 0.58 and FDR < 0.05) are shown in purple. (E) Pathway enrichment analysis showing the most significant (by normalized enrichment score) nonredundant pathways downregulated in eWAT. Pooled data from 3 independent experiments (A–C) or data from 1 experiment (D and E). Data presented as mean ± SEM (A and C); 1-way ANOVA with Fisher’s LSD test (A) or unpaired 2-tailed Student’s t test between means (C). ns, not significant; *P < 0.05, ****P < 0.0001. Avg, average; MMP, matrix metalloproteinases; NOXs, NADPH oxidases.

To further assess how loss of CD9 influences adipose tissue remodeling, we performed bulk RNA-Seq of eWAT from CD9-MKO HFD mice and control HFD mice. We identified 1,873 genes upregulated and 2,204 genes downregulated in the eWAT of CD9-MKO HFD mice compared with control HFD mice (Figure 4D). Gene set enrichment analysis revealed that the top downregulated pathways in CD9-MKO HFD eWAT were associated with inflammatory cell activation and function, ECM remodeling, and cell adhesion (Figure 4E). Consistent with our previous data, expression of Itgax and Itgav was reduced in whole eWAT from CD9-MKO HFD mice compared with control HFD mice (Supplemental Figure 7, C and D). In contrast, the top upregulated pathways included those related to cellular metabolism and mitochondrial function (Supplemental Figure 7E). Collectively, these findings demonstrate that myeloid-intrinsic CD9 broadly regulates inflammatory, adhesive, and profibrotic remodeling programs in adipose tissue during diet-induced obesity.

Myeloid-intrinsic CD9 promotes adipose tissue fibrosis during diet-induced obesity. Given that CD9+ ATMs were enriched for ECM remodeling genes and that these pathways were downregulated in CD9-MKO HFD eWAT compared with controls, we next investigated whether loss of myeloid-intrinsic CD9 affected eWAT fibrosis during obesity. Histological analysis demonstrated reduced collagen deposition in CD9-MKO HFD eWAT compared with controls (Figure 5A). Next, we investigated whether this reduction in fibrosis was associated with changes in the transcriptional profile of eWAT adipocyte fraction. Fibrosis-associated genes (Lgals3, Pdgfb, Mmp12, Gpnmb, and Adam8) were decreased in the adipocyte fraction of CD9-MKO HFD mice compared with control HFD mice (Figure 5B). Notably, with the exception of Lgals3, these genes have either very low or no expression in adipocytes, suggesting that the reduction in expression of these genes primarily reflects the loss of CD9+ ATMs adhering to adipocytes within the adipocyte fraction (55).

Figure 5

Myeloid-intrinsic CD9 promotes adipose tissue fibrosis and macrophage profibrotic gene expression during obesity. Epididymal white adipose tissue (eWAT) was collected from male control (Cd9fl/fl) and CD9-MKO (Cd9fl/fl LysMCre+/–) mice fed an HFD for 12 weeks. (A) Representative images and quantification of Pic