Lipid‐associated macrophages are more abundant in subcutaneous than visceral adipose tissue in patients with obesity

Abstract

Objective

Because white adipose tissue is infiltrated by several immune cells and their signature in individuals with obesity has not been fully characterized, we wanted to study the most abundant population, which is macrophages, a subtype of myeloid cell.

Methods

To address this objective, we performed transcriptomic analysis of subcutaneous adipose tissue (SAT)‐ and visceral adipose tissue (VAT)‐infiltrated CD11b⁺ myeloid cells from individuals with severe obesity.

Results

Our results showed that gene expression in human white adipose tissue–infiltrated CD11b⁺ myeloid cells was depot‐dependent. The expression of lipid‐associated macrophage biomarkers was higher in SAT‐ than VAT‐infiltrated CD11b⁺ cells (TREM2, CD9, GPNMB, CD68). In contrast, VAT‐infiltrated CD11b⁺ cells overexpressed genes associated with a perivascular M2‐like adipose tissue macrophage signature (LYVE1, TIMD4, MRC1). In addition, no classical gene expression polarization (M1 and M2) was shown when VAT and SAT CD11b⁺ cells were compared. Finally, high levels of CD248, a sensor of lipids associated with insulin resistance, were found to be overexpressed in SAT‐ compared with VAT‐infiltrated CD11b⁺ myeloid cells.

Conclusions

This study characterizes for the first time the macrophage biomarker signature in human VAT‐ and SAT‐infiltrated CD11b⁺ myeloid cells from individuals with severe obesity. Further studies are required to elucidate their potential role and specific function in the immunometabolism of individuals with obesity.

STUDY IMPORTANCE.

What is already known?

• Subcutaneous and visceral adipose tissue (SAT and VAT, respectively) have different biological characteristics.

• Obesity triggers adipose tissue remodeling and immune cells infiltration.

What does this study add?

• We demonstrate that white adipose tissue–infiltrated CD11b⁺ myeloid cells isolated from individuals with obesity have a depot‐dependent gene expression profile.

• We also observe that SAT‐infiltrated CD11b⁺ myeloid cells show a lipid‐associated macrophage (LAM) phenotype compared with those isolated from VAT.

How might these results change the direction of research or the focus of clinical practice?

• Our results regarding how LAMs (phagocytic cells with a lipid‐handling capacity) are more present in SAT to likely address correct adipose tissue remodeling and endocrine function could open a new therapeutic window to fight against obesity‐related metabolic complications.

INTRODUCTION

Obesity is a chronic, relapsing, progressive disease with a multifactorial etiology [1]. The worldwide incidence of obesity has tripled in the last four decades, making it one of the most prevalent diseases worldwide [2].

White adipose tissue (WAT) is an organ responsible for energy storage and comprises a heterogeneous cell population, such as preadipocytes, adipocytes, fibroblasts, and immune cells [3]. It also has a crucial endocrine role and can respond to metabolic changes [4, 5]. In humans, WAT can be classified according to its distribution into two main depots: visceral adipose tissue (VAT), which includes omental, mesenteric, retroperitoneal, gonadal, and pericardial WAT, and subcutaneous adipose tissue (SAT), which is located under the skin in the inguinal, axillar, and other areas. Both depots have been widely studied in obesity for their association with inflammation, insulin resistance, and cardiometabolic risk [6, 7].

It has been broadly described that an increase in WAT hypertrophy and hyperplasia causes local and systemic, low‐grade, chronic inflammation, or meta‐inflammation, that triggers the development of severe metabolic diseases [8, 9]. In obesity, an increase in triglyceride accumulation in adipocytes causes enlargement of WAT and, consequently, dysregulation of adipokine secretion. In addition, the secretion of WAT‐derived, proinflammatory cytokines promotes the migration and infiltration of immune cells into the WAT, which further contributes to inflammation [10]. This leads to chronic and self‐maintained, low‐grade inflammation associated with several obesity‐related pathologies [11].

Previous single‐cell studies have characterized the cellular composition of immune cells and furthered our understanding of their involvement in obesity [12, 13, 14, 15]. Moreover, previous studies in humans have mapped single‐cell genomic profiles onto spatial transcriptomic data to characterize the spatial patterning of WAT cellular composition [16, 17]. However, although a spatial understanding of obesity‐induced WAT‐infiltrated immune cells remodeling over the course of metabolic disruption has been developed in murine models [18], data in humans are lacking.

It has been reported that individuals with obesity show differences in adipose tissue‐infiltrated macrophages (ATMs) and their surface markers compared with normal‐weight individuals [19] and that the phenotype of ATMs is associated with changes in the metabolic signaling of the different adipose tissue depots [20, 21, 22, 23, 24, 25, 26, 27]. In an obesity context, ATMs can engulf lipids released by adipocytes, a process correlated with body mass index (BMI) mainly in VAT compared with SAT [22, 28, 29]. This process triggers proinflammatory signaling [30, 31] and other pathogenic mechanisms associated with the lipotoxic environment characteristic of obesity‐associated diseases [32, 33, 34]. Lipid‐associated macrophages (LAMs) are among the most abundant macrophage subtypes in chronic obesity and are typically located in the crown‐like structures (CLS) surrounding dead or dying adipocytes [17, 35]. LAMs are phagocytic cells with a lipid‐handling capacity that are absent during prenatal development or are at a very low abundance in healthy tissues but that appear in response to metabolic complications [36, 37, 38, 39].

Although the dynamics of WAT‐infiltrated immune cells during obesity are well documented, the molecular mechanisms regulating immune and metabolic dysfunction and their depot characterization within human WAT remain poorly understood. Here, we focus on WAT‐infiltrated CD11b⁺ cells, which mainly include macrophages but also other immune cells in a lower proportion, such as dendritic cells. We performed transcriptomic analysis of SAT‐ and VAT‐infiltrated CD11b⁺ myeloid cells from individuals with severe obesity and thoroughly characterized their gene expression signature.

METHODS

Study participants

The study was approved (code PI‐18‐078) by the ethical committee of the Hospital Germans Trias i Pujol (Badalona, Spain) and the ethical committee from Alcorcón Foundation University Hospital and Rey Juan Carlos University and was conducted in accordance with the guidelines of the Helsinki convention. Participants gave written informed consent before clinical data collection.

Three different cohorts of patients were enrolled (Table 1). All patients were evaluated by endocrinologists and surgeons following the institutional protocol for bariatric surgery (BS), between October 2015 and September 2023, according to BS criteria (Spanish Position Statement between Obesity, Endocrinology, Diabetes and Surgery Societies). For cohort 1, 22 individuals (7 male individuals and 15 female individuals) with severe obesity (BMI > 35 kg/m²) were enrolled, and VAT and SAT from the same individual were collected during BS. Clinical data were collected at baseline and then at 6 and 24 months after surgery. Cohort 2 was composed of 27 patients (13 with severe obesity and 14 without obesity with BMI >35 or BMI <27, respectively). VAT and SAT biopsies were collected when they attended to BS for the cohort 1 or on occasion of consultation or minor surgery, mainly cholecystectomy for the latter. Cohort 3 included 22 patients with severe obesity (BMI > 35) who underwent BS at the Alcorcón Foundation University Hospital between June 2022 and June 2023. SAT biopsy was collected when they attended to BS and 3 months after surgery. CD11b⁺ myeloid cells were isolated from cohort 1 and total cells from cohorts 2 and 3. Exclusion criteria for all cohorts were having cancer, active infectious or inflammatory pathologies other than those related to obesity, and treatment with immunosuppressant drugs or suffering from other forms of immunosuppression.

TABLE 1.

Clinical parameters of cohorts 1, 2, and 3.

	Cohort 1 (n = 22)						Cohort 2 (n = 27)			Cohort 3 (n = 11)
	Time, mean ± SD			Sig., p value			Control (n = 14), mean ± SD	Obesity (n = 13), mean ± SD	Sig., p value	Time, mean ± SD		Sig., p value
	0 mo	6 mo	24 mo	0 vs. 6 mo	0 vs. 24 mo	6 vs. 24 mo				0 mo	3 mo
Body weight, kg	116.77 ± 25.77	87.99 ± 19.76	84.38 ± 18.81	***	***	NS	64.9 ± 9.40	112.2 ± 12.0	****	121.45 ± 17.69	97.15 ± 15.08	**
BMI	43.92 ± 6.75	33.56 ± 7.20	30.82 ± 9.96	****	***	**	24.7 ± 2.5	43.5 ± 3.9	****	42.28 ± 5.13	33.77 ± 4.03	***
Insulin, mIU/L	14.41 ± 9.23	7.28 ± 1.96	7.48 ± 3.24	**	*	NS	5.4 ± 1.3	16.5 ± 11.8	NS	20.60 ± 12.42	6.96 ± 2.64	**
Glucose, mg/dL	103.14 ± 21.17	92.48 ± 11.29	97.76 ± 13.46	**	NS	*	92.7 ± 15.4	106.5 ± 23.5	NS	104.55 ± 24.62	88.55 ± 10.68	*
HOMA‐IR, %	3.88 ± 2.70	1.67 ± 0.50	1.79 ± 0.76	**	*	NS	1.3 ± 0.4	4.8 ± 4.2	NS	5.26 ± 3.06	1,55 ± 0,70	***
HbA1c, %	39.52 ± 5.90	33.90 ± 4.84	35.95 ± 4.19	****	***	*	4.8 ± 0.8	5.3 ± 1.2	**	34.00 ± 4.17	32.82 ± 2.75	NS
TAG, mg/dL	128.19 ± 60.49	100.14 ± 34.37	97.95 ± 48.32	*	*	NS	72 ± 28	126 ± 34	**	129.64 ± 56.09	98.45 ± 19.43	0.097
Total‐c, mg/dL	203.65 ± 65.57	192.40 ± 38.48	179.25 ± 31.13	NS	NS	NS	190 ± 34	158 ± 18	*	150.36 ± 38.94	148.45 ± 17.74	NS
HDL‐c, mg/dL	47.14 ± 10.33	49.33 ± 7.66	56.68 ± 8.81	NS	***	**	68 ± 9	42 ± 8	****	37.73 ± 10.67	42.00 ± 11.58	NS
LDL‐c, mg/dL	131.90 ± 51.82	117.38 ± 39.77	100.70 ± 24.88	NS	0.0579	NS	108 ± 31	91 ± 14	NS	86.82 ± 27.57	86.82 ± 15.78	NS
CRP, mg/dL	9.95 ± 6.07	4.26 ± 3.45	2.46 ± 2.47	**	***	*

Note: Cohort 1: 22 patients living with obesity (7 male patients and 15 female patients). Cohort 2: 14 control patients and 13 patients living with obesity undergoing BS (all female patients). Cohort 3: 22 patients living with obesity undergoing BS (5 male patients and 6 female patients), at the moment of the surgery and after 3 months. Differences between controls and patients living with obesity were assessed using Student t test (normally distributed) or Mann–Whitney test (nonnormally distributed) for unpaired data. Normality was checked using the Shapiro–Wilk test.

Abbreviations: BS, bariatric surgery; CRP, C‐reactive protein; HbA1c, glycated hemoglobin; HDL‐c, high‐density lipoprotein cholesterol; HOMA‐IR, homeostasis model assessment of insulin resistance; LDL‐c, low‐density lipoprotein cholesterol; NS, not significant; Sig., statistical significance; TAG, triglycerides; Total‐c; total cholesterol.

^* p < 0.05.

^** p < 0.01.

^*** p < 0.001.

^**** p < 0.0001.

Human serological analysis

Serum samples were collected after 12‐h fasting and frozen at −20 °C at baseline and at 6 and 24 months after BS. Glucose and insulin levels, as well as lipid profiles (total cholesterol, high‐density lipoprotein cholesterol, low‐density lipoprotein cholesterol, and triglycerides), were measured in the certified core clinical laboratory. Homeostasis model assessment of insulin resistance (HOMA‐IR) was calculated as: $\text{HOMA}-\mathrm{IR}=\frac{\left[\text{Glucose}\frac{\mathrm{mg}}{\mathrm{dL}}\right]\times \left[\text{Insulin}\frac{\mathrm{m}.\mathrm{u}.\mathrm{int}}{\mathrm{dL}}\right]}{405}$.

CD11b⁺ myeloid cells isolation and RNA extraction from cohort 1

Immediately after surgical extraction, isolation of the stromal vascular fraction (SVF) from SAT and VAT biopsies was performed under sterile conditions. First, biopsies were washed with Hanks’ Balanced Salt Solution (Capricorn Scientific, HBSS‐2A), minced using a scalpel, and digested using 1 mg/mL of collagenase NB4 (SERVA, S1745401) in Hanks’ Balanced Salt Solution. The digestion was performed at 37°C for 1 h with stirring up every 10 min. Once the tissue was totally digested, digestion was stopped by putting the samples in ice. Then, the solution containing the SVF was filtered through a 100‐μm filter and centrifuged at 500g for 5 min at 4°C, and the supernatant was discarded. The obtained pellet was washed twice with phosphate‐buffered saline (PBS) plus 1% fetal bovine serum (FBS) and centrifuged at 500g for 5 min at 4°C. After washing, the resulting pellet containing the SVF was resuspended in cold magnetic‐activated cell sorting (MACS) buffer for magnetic labeling of the CD11b⁺ cells.

Magnetic labeling and separation were performed using a MidiMACS kit with LS column (Miltenyi Biotec, 130‐042‐301) according to the manufacturer's instructions. The eluted fraction enriched with CD11b⁺ cells was cooled in ice. An aliquot of 2 μL was taken, and then the fraction was centrifuged at 300g for 7 min at 4°C. Next, 4 mL of the supernatant was discarded, and the pellet was resuspended in the remaining volume, which was transferred to a 1.5‐mL conical tube and centrifuged at 11,000g for 5 min. The pellet was used to perform RNA extraction using a Single Cell RNA Purification Kit (Norgen Biotek Corp., 51800) according to the manufacturer's instructions.

The aliquots from SAT and VAT were labeled with 7‐AAD (Miltenyi Biotec, 130‐111‐568) and CD11b (Miltenyi Biotec, 130‐113‐797) and incubated with CountBright counting beads (Invitrogen, C36950) to quantify the viability and number of infiltrated CD11b⁺ cells according to the manufacturer's instructions. Data acquisition was performed using a FACSCanto II system (BD Biosciences) and FACSDiva software v9.0 (BD Biosciences) and analyzed in FlowJo version X.0.7 (FlowJo, LLC). Moreover, we checked the amount of CD11b⁺ cells in comparison to another immune marker to corroborate that our samples were enriched in this CD11b subpopulation (online Supporting Information).

Transcriptome analysis from cohort 1

The total quantity and integrity of the extracted RNA were evaluated using a Bioanalyzer system (Nano 6000 assay on Agilent's 2100 Bioanalyzer). RNA from each sample was used for microarray hybridization using Affymetrix array (Clariom D Assay, human, Thermo Fisher Scientific, Inc.) and processed using the Applied Biosystems GeneChip System 3000 (Thermo Fisher Scientific).

Sample processing was performed at the High Technology Unit facility at Vall d'Hebron Institut de Recerca in Barcelona, Spain, selecting samples with RNA concentrations ≥3 μg and RNA integrity with numbers >7. The cohort of individuals was clustered by glycemia levels and initially assigned to two groups: glycemia ≤100 mg/dL and glycemia >100 mg/dL.

Bioinformatics analysis from cohort 1

All statistical analyses of microarray data were performed using the R‐based software, R version 4.0.3 (R Project for Statistical Computing). Quality control was performed using the arrayQualityMetrics package 3.64 (Bioconductor). Background correction, probe set summarization, and normalization were performed using the oligo package with the most up‐to‐date annotation in Bioconductor version 3.12. A paired‐sample design comparing VAT and SAT from the same individual was applied. Subsequent differential expression analysis was performed using the limma package at the gene level, focusing on known genes (with assigned gene symbols). Transcripts were considered for further analyses if they matched the double criteria or false detection rate (FDR) of <0.05 and log (fold change) VAT versus SAT of >1.5. Exploratory inference of putatively affected biological functions was performed using GOrilla (Multi Knowledge Project) v2, harnessing Gene Ontology categories to perform pathway analyses. In‐depth functional enrichment analyses were performed using Gene Set Enrichment Analysis and the results visualized using the Enrichment Map tool in Cytoscape (Cytoscape Consortium) 2.10.3. Pre‐ranked–based analyses were performed using ranking by log2ratio or signed −log10 p values on relevant Molecular Signatures Database gene set collections (version 7.1, University of California, San Diego and Broad Institute).

Adipose tissue collection and RNA isolation and processing from cohorts 2 and 3

Whole VAT and SAT samples were obtained from cohort 2 at the time of surgery. SAT was obtained from cohort 3 at the time of surgery and 3 months after the intervention. Total RNA was then extracted from whole adipose samples using a standard column‐affinity based methodology (NucleoSpin RNA II, MACHEREY‐NAGEL). An amount of 500 ng of total RNA was retro‐transcribed into complementary DNA (cDNA) using random hexamer primers and MultiScribe Reverse Transcriptase (TaqMan Reverse Transcription Reagents, ThermoFisher Scientific), following the manufacturer's instructions. Platinum Quantitative PCR SuperMix‐UDG with ROX reagent (ThermoFisher Scientific) was used as master mix reagent, and expression levels of each gene of interest were assessed with the specific TaqMan probes (ThermoFisher Scientific). As an endogenous control, peptidylprolyl isomerase A was used. Quantitative reverse transcriptase‐polymerase chain reaction (RT‐PCR) was carried out in an ABI PRISM 7900HT Sequence Detection System (ThermoFisher Scientific) using the following conditions: 2 min at 50 °C, 10 min at 95 °C followed by 40 cycles of 15 s at 95 °C, and 1 min at 60 °C. Relative mRNA levels were determined using the 2^−ΔΔCt method and expressed as fold change in arbitrary units.

Statistical analysis

The statistical analyses performed indicate a comparison between controls and treatments of each experimental model. One‐way ANOVA was performed when more than two groups were compared, followed by Tukey's post hoc analysis. Correlation analyses were conducted by age‐ and sex‐adjusted multiple linear regression. Multicollinearity was assessed by calculating the variance inflation factor (VIF) for each covariables in all regressions; no significant multicollinearity was detected in any case (VIF < 1.5). Standardized multiple regression model coefficients (β) and their p values for each variable of interest in their respective age‐ and sex‐adjusted models were retrieved. All analyses were performed using SPSS Statistics version 25.0 (IBM Corp.) and GraphPad Prism version 8.0.1 (GraphPad Software). Statistically significant differences were considered when the p value was <0.05.

RESULTS

WAT‐infiltrated CD11b⁺ cells from individuals with severe obesity show a depot‐dependent signature

The clinical data of the cohort of individuals with severe obesity analyzed in this study are described in Table 1. Viable CD11b⁺ myeloid cells were isolated from VAT and SAT from individuals with obesity, and in‐depth transcriptomic analyses were subsequently performed. Isolated cells were confirmed as CD11b⁺ by fluorescence‐activated cell sorting in both VAT and SAT, and equal amounts of cellular mRNA from both depots were used to perform the array (dataset available at the Gene Expression Omnibus [GEO] repository: GSE276225). It is important to highlight that CD11b⁺ myeloid cell data from SAT clustered separately from VAT and showed different transcriptomic data.

Differential gene expression analyses of CD11b⁺ cells in VAT and SAT from individuals with obesity revealed nominally significant differences in the gene expression profile (FDR p < 0.05, absolute fold change > 1.5; Figure 1A,B). Multiple testing corrections revealed that 2627 genes were significantly modulated between VAT and SAT (559 upregulated, 2068 downregulated, |fold change| > 1.2, FDR < 0.05; Figure 1A). ITLN1, UPK1B, TIMD4, CLDN1, ANXA8, IL2RA, PRG4, MSLN, ITLN2, and C4BPB were among the top 10 upregulated transcripts, whereas MCOLN3, CXCL14, TM4SF19‐DYNLT2B, ITGA3, TM4SF19, MMP7, MMP12, CHIT1, XGY2, and NTRK2 were among the top 10 downregulated transcripts (Figure 1A,B).

FIGURE 1.

Differential gene expression summary results between VAT‐ and SAT‐derived CD11b⁺ cells from individuals with severe obesity. (A) Volcano plot with significant genes (nominal p < 0.001, with |FC| > 1.2 in red; |FC| < 1.2 in blue), NS (nominal p > 0.001, with |FC| > 1.2 in green; |FC| < 1.2 in gray); top DEG symbols highlighted. (B) Heat map of the 50 top DEGs (among adjusted p < 0.05, |FC| > 1.2). DEG, differentially expressed gene; FC, fold change; NS, not significant; SAT, subcutaneous adipose tissue; VAT, visceral adipose tissue; WAT, white adipose tissue. [Color figure can be viewed at wileyonlinelibrary.com]

Higher expression of LAM markers in SAT‐ than VAT‐infiltrated CD11b⁺ cells

Bioinformatic analysis was performed to identify the most abundant subtype of macrophages in each depot according to the gene expression levels of different markers. Following the classification described in the current literature [18], we identified monocyte‐derived nonresident macrophages, resident macrophages (MAC1), proinflammatory macrophages (MAC2–3), pre‐LAMs as precursors of LAMs (MAC4), and mature LAMs (MAC5).

Our results indicated that SAT‐isolated CD11b⁺ cells showed higher expression levels of MAC4 and MAC5 biomarkers (TREM2, CD9, GPNMB, CD68) in MAC4/pre‐LAMs or MAC5/LAMs than those isolated from VAT (Figure 2). In contrast, VAT‐isolated CD11b⁺ cells overexpressed genes associated with the MAC1 or perivascular M2‐like ATM signature (high levels of LYVE1, TIMD4, MRC1 [CD206]; Figure 2). No classical polarization (M1 and M2) was shown when VAT‐ and SAT‐infiltrated CD11b⁺ cells were compared in terms of gene expression markers. However, a robust CXCL9‐SPP1 polarization related to immunosuppressor protumoral CD11b⁺ myeloid cells versus antitumoral CD11b⁺ myeloid cells was found (Table S1A). Moreover, elevated expression levels of CD248, a marker of lipids associated with insulin resistance, was found in SAT‐infiltrated CD11b⁺ cells compared with VAT‐infiltrated CD11b⁺ cells in individuals with severe obesity (Table S1B).

FIGURE 2.

Volcano plot with significant differentially expressed genes between VAT‐ and SAT‐derived CD11b⁺ cells from individuals with severe obesity (in red); highlighted gene symbols correspond to genes described as markers of MAC1 or stromal vascular matrix adipose tissue‐infiltrated macrophages, more highly expressed in VAT, and markers of LAMs, more highly expressed in SAT. LAM, lipid‐associated macrophage; MAC1, resident macrophages; SAT, subcutaneous adipose tissue; VAT, visceral adipose tissue. [Color figure can be viewed at wileyonlinelibrary.com]

Finally, when we analyzed the expression levels of the aforementioned LAM marker genes in cohort 2, we observed that they were induced in adipose tissue from patients with obesity compared with lean controls. Moreover, gene expression of these LAM markers was higher in SAT than in VAT (Figure 3).

FIGURE 3.

Relative transcript levels of selected LAM marker genes and mean relative levels (LAM marker score) in SAT and VAT from C or OB from cohort 2. Data are expressed as mean ± SEM and referenced to C SAT. *p < 0.05; **p < 0.01; ***p < 0.001, C vs. OB; #p < 0.05; ###p < 0.001, SAT vs. VAT. AU, quantitative PCR–determined arbitrary units; C, lean individuals; LAM, lipid‐associated macrophage; OB, individuals with obesity; SAT, subcutaneous adipose tissue; VAT, visceral adipose tissue.

Presence of LAM markers in SAT from patients with obesity was associated with clinical parameters before and after BS

When we analyzed the gene expression of the main LAM markers in our cohort 3 (Table 2), we saw that, at baseline, a negative correlation between TREM2 expression and body weight was observed and that this correlation was missed when we compared these two variables 3 months after the surgery. Interestingly, a positive correlation between CD68 expression and high‐density lipoprotein cholesterol levels was found before the surgery, whereas 3 months later the correlation turned to be negative. Moreover, an inverse correlation was identified 3 months after surgery between CD68 gene expression and body weight.

TABLE 2.

Age‐ and sex‐adjusted multiple linear regression analyses for expression levels of selected LAM marker genes and clinical variables of interest in cohort 3 at baseline or 3 months after BS.

	TREM2		CD9		GPNMB		CD68		CD36
	β (95% CI)	p value	β (95% CI)	p value	β (95% CI)	p value	β (95% CI)	p value	β (95% CI)	p value
Baseline (LAM markers and clinical variables)
Weight (PD)	−0.777 (−0.943 to −0.124)	0.047 *	−0.037 (−0.874 to 0.799)	0.933	0.072 (−0.793 to 0.936)	0.875	0.13 (−0.786 to 1.047)	0.788	0.462 (−0.371 to 0.905)	0.313
Height	−0.697 (−0.954 to 0.147)	0.150	0.145 (−0.8 to 0.909)	0.772	−0.15 (−0.928 to 0.829)	0.773	0.732 (−0.164 to 1.628)	0.153	0.001 (−0.922 to 0.903)	0.998
Weight (PS)	−0.737 (−0.932 to −0.156)	0.042 *	−0.100 (−0.863 to 0.663)	0.805	0.002 (−0.791 to 0.796)	0.995	0.128 (−0.711 to 0.967)	0.774	0.323 (−0.466 to 0.911)	0.449
BMI	−0.381 (−0.925 to 0.262)	0.283	−0.105 (−0.776 to 0.566)	0.769	0.126 (−0.567 to 0.819)	0.731	−0.275 (−0.991 to 0.44)	0.475	0.379 (−0.292 to 0.905)	0.305
Glucose	−0.387 (−0.955 to 0.281)	0.293	−0.026 (−0.724 to 0.673)	0.944	0.050 (−0.672 to 0.773)	0.895	−0.207 (−0.960 to 0.547)	0.608	0.085 (−0.664 to 0.834)	0.830
Insulin	0.209 (−0.716 to 1.135)	0.671	0.051 (−0.85 to 0.952)	0.915	−0.105 (−0.936 to 0.825)	0.831	0.253 (−0.723 to 1.229)	0.627	−0.002 (−0.973 to 0.969)	0.997
HOMA‐IR	−0.043 (−0.864 to 0.778)	0.921	0.021 (−0.768 to 0.810)	0.960	−0.047 (−0.863 to 0.77)	0.914	0.069 (−0.799 to 0.937)	0.880	0.014 (−0.836 to 0.863)	0.976
CRP	−0.329 (−0.932 to 0.374)	0.389	−0.004 (−0.719 to 0.711)	0.991	0.057 (−0.683 to 0.796)	0.885	−0.266 (−1.028 to 0.497)	0.516	0.043 (−0.726 to 0.812)	0.916
HbA1c	0.282 (−0.416 to 0.981)	0.454	−0.058 (−0.757 to 0.642)	0.876	0.05 (−0.675 to 0.774)	0.897	0.143 (−0.621 to 0.908)	0.724	0.162 (−0.583 to 0.906)	0.683
Hb	−0.277 (−0.961 to 0.507)	0.511	0.099 (−0.676 to 0.875)	0.809	−0.312 (−0.985 to 0.46)	0.454	−0.068 (−0.924 to 0.789)	0.881	0.019 (−0.819 to 0.857)	0.966
Creatinine	−0.310 (−0.917 to 0.550)	0.503	−0.238 (−0.975 to 0.599)	0.594	−0.141 (−0.902 to 0.738)	0.762	−0.412 (−0.9303 to 0.479)	0.395	0.296 (−0.598 to 0.919)	0.537
Albumin	−0.386 (−0.991 to 0.42)	0.379	0.257 (−0.542 to 0.956)	0.548	0.113 (−0.733 to 0.959)	0.800	−0.012 (−0.917 to 0.892)	0.980	0.133 (−0.745 to 0.912)	0.775
TAG	0.149 (−0.551 to 0.850)	0.689	0.511 (−0.056 to 0.978)	0.121	0.298 (−0.372 to 0.968)	0.412	0.294 (−0.425 to 1.012)	0.449	0.276 (−0.428 to 0.981)	0.467
HDL‐c	−0.536 (−0.981 to 0.110)	0.148	0.125 (−0.597 to 0.848)	0.744	−0.075 (−0.827 to 0.676)	0.850	0.801 (0.262 to 0.934)	0.023 *	0.058 (−0.725 to 0.841)	0.889
LDL‐c	0.266 (−0.429 to 0.962)	0.477	0.109 (−0.581 to 0.799)	0.767	0.163 (−0.546 to 0.872)	0.666	0.223 (−0.524 to 0.970)	0.577	0.164 (−0.574 to 0.902)	0.676
Total‐c	0.097 (−0.617 to 0.812)	0.797	0.256 (−0.407 to 0.919)	0.474	0.185 (−0.515 to 0.886)	0.620	0.436 (−0.251 to 1.124)	0.254	0.209 (−0.517 to 0.935)	0.590
GPT	−0.156 (−0.847 to 0.535)	0.672	−0.023 (−0.696 to 0.650)	0.949	0.044 (−0.652 to 0.741)	0.904	−0.283 (−0.995 to 0.428)	0.461	0.153 (−0.563 to 0.869)	0.687
GOT	−0.266 (−0.935 to 0.402)	0.461	−0.035 (−0.705 to 0.635)	0.922	0.049 (−0.644 to 0.742)	0.893	−0.345 (−1.038 to 0.347)	0.361	0.154 (−0.559 to 0.866)	0.685
GGT	0.136 (−0.603 to 0.874)	0.730	−0.454 (−0.987 to 0.178)	0.202	−0.397 (−0.978 to 0.284)	0.291	−0.102 (−0.887 to 0.683)	0.806	−0.166 (−0.927 to 0.595)	0.681
3 mo (LAM markers and clinical variables)
Weight (PD)	0.212 (−0.627 to 1.051)	0.636	−0.354 (−0.906 to 0.353)	0.360	−0.291 (−0.854 to 0.272)	0.344	−0.547 (−0.928 to 0.124)	0.154	0.585 (0.078 to 0.991)	0.058
Height	0.001 (−0.969 to 0.971)	0.998	−0.227 (−0.907 to 0.613)	0.613	0.022 (−0.663 to 0.707)	0.952	−0.842 (−0.948 to −0.206)	0.036 *	0.138 (−0.613 to 0.888)	0.730
Weight (3 mo)	0.170 (−0.548 to 0.888)	0.657	−0.269 (−0.881 to 0.344)	0.418	−0.166 (−0.666 to 0.334)	0.535	−0.673 (−0.912 to −0.225)	0.022 *	0.434 (−0.035 to 0.904)	0.113
BMI	0.148 (−0.624 to 0.919)	0.719	−0.319 (−0.966 to 0.327)	0.365	−0.294 (−0.799 to 0.211)	0.292	−0.552 (−0.914 to 0.035)	0.108	0.532 (0.068 to 0.995)	0.059
Glucose	−0.027 (−0.748 to 0.693)	0.943	−0.157 (−0.783 to 0.468)	0.637	0.031 (−0.478 to 0.539)	0.909	−0.328 (−0.944 to 0.289)	0.332	−0.138 (−0.691 to 0.415)	0.639
Insulin	0.019 (−0.752 to 0.790)	0.962	−0.273 (−0.923 to 0.378)	0.439	−0.288 (−0.789 to 0.213)	0.297	−0.353 (−1.011 to 0.306)	0.328	−0.125 (−0.720 to 0.469)	0.692
HOMA‐IR	0.021 (−0.712 to 0.754)	0.956	−0.240 (−0.863 to 0.382)	0.474	−0.221 (−0.712 to 0.27)	0.407	−0.358 (−0.977 to 0.261)	0.294	−0.137 (−0.700 to 0.426)	0.649
CRP	0.739 (0.195 to 0.928)	0.032 *	0.277 (−0.372 to 0.927)	0.430	0.204 (−0.319 to 0.728)	0.470	−0.123 (−0.826 to 0.58)	0.742	0.496 (0.019 to 0.973)	0.081
HbA1c	0.159 (−0.599 to 0.918)	0.693	−0.001 (−0.679 to 0.677)	0.997	−0.064 (−0.604 to 0.476)	0.823	0.166 (−0.529 to 0.860)	0.655	0.134 (−0.457 to 0.725)	0.670
Hb	−0.588 (−1.943 to 0.767)	0.423	−0.052 (−0.931 to 0.920)	0.937	−0.048 (−0.905 to 0.957)	0.929	−0.389 (−0.966 to 0.886)	0.568	−0.267 (−0.936 to 0.826)	0.647
Creatinine	0.568 (−0.167 to 1.303)	0.174	−0.054 (−0.801 to 0.693)	0.891	−0.014 (−0.612 to 0.584)	0.964	0.115 (−0.658 to 0.889)	0.778	0.058 (−0.602 to 0.717)	0.869
Albumin	−0.701 (−0.922 to −0.180)	0.034 *	−0.493 (−0.931 to 0.045)	0.115	−0.427 (−0.839 to −0.014)	0.082	−0.36 (−0.981 to 0.262)	0.294	−0.247 (−0.792 to 0.298)	0.404
TAG	−0.234 (−0.950 to 0.483)	0.543	−0.360 (−0.954 to 0.233)	0.273	−0.015 (−0.535 to 0.505)	0.956	0.022 (−0.655 to 0.699)	0.951	0.155 (−0.409 to 0.718)	0.607
HDL‐c	−0.469 (−1.229 to 0.292)	0.267	−0.911 (−0.921 to −0.612)	* 0.001	−0.497 (−0.959 to −0.035)	0.043 *	−0.698 (−0.926 to −0.130)	0.047 *	0.046 (−0.606 to 0.698)	0.893
LDL‐c	0.318 (−0.344 to 0.980)	0.377	0.323 (−0.249 to 0.895)	0.305	0.144 (−0.340 to 0.629)	0.577	0.002 (−0.643 to 0.647)	0.995	0.253 (−0.262 to 0.768)	0.368
Total‐c	0.013 (−0.691 to 0.716)	0.973	−0.201 (−0.804 to 0.402)	0.534	−0.100 (−0.591 to 0.391)	0.702	−0.311 (−0.915 to 0.293)	0.346	0.270 (−0.242 to 0.781)	0.336
GPT	−0.076 (−0.793 to 0.640)	0.84	0.076 (−0.556 to 0.708)	0.821	0.081 (−0.423 to 0.585)	0.762	0.387 (−0.208 to 0.982)	0.243	−0.368 (−0.858 to 0.122)	0.185
GOT	−0.247 (−0.933 to 0.439)	0.503	0.003 (−0.624 to 0.629)	0.994	0.101 (−0.394 to 0.597)	0.700	0.245 (−0.381 to 0.871)	0.468	−0.295 (−0.804 to 0.214)	0.293
GGT	0.597 (−0.316 to 1.509)	0.241	0.221 (−0.659 to 0.910)	0.637	0.286 (−0.398 to 0.970)	0.439	−0.194 (−0.911 to 0.726)	0.691	0.842 (0.355 to 0.933)	0.012 *
Baseline levels (LAM markers) vs. 3 mo (clinical variables)
Height	−0.697 (−0.954 to 0.147)	0.150	0.145 (−0.800 to 0.909)	0.772	−0.150 (−0.912 to 0.829)	0.773	0.732 (−0.164 to 0.962)	0.153	0.001 (−0.902 to 0.902)	0.998
Weight	−0.726 (−0.924 to −0.212)	0.028 *	−0.016 (−0.731 to 0.699)	0.967	−0.021 (−0.761 to 0.719)	0.958	0.271 (−0.491 to 0.903)	0.509	0.259 (−0.486 to 0.900)	0.517
BMI	−0.748 (−0.932 to −0.178)	0.037 *	0.008 (−0.756 to 0.772)	0.984	0.139 (−0.645 to 0.923)	0.738	−0.021 (−0.862 to 0.821)	0.963	0.492 (−0.244 to 0.922)	0.232
Glucose	−0.358 (−0.904 to 0.328)	0.340	0.009 (−0.698 to 0.716)	0.981	−0.379 (−0.905 to 0.297)	0.308	0.480 (−0.212 to 0.917)	0.216	−0.244 (−0.982 to 0.495)	0.539
Insulin	−0.436 (−0.915 to 0.281)	0.272	−0.152 (−0.899 to 0.596)	0.702	−0.395 (−0.912 to 0.331)	0.322	0.933 (0.819 to 0.971)	<0.001 *	−0.351 (−0.912 to 0.420)	0.402
HOMA‐IR	−0.423 (−0.910 to 0.256)	0.261	−0.144 (−0.855 to 0.566)	0.703	−0.428 (−0.910 to 0.245)	0.253	0.929 (0.654 to 0.976)	0.001 *	−0.369 (−0.909 to 0.354)	0.350
CRP	0.01 (−0.777 to 0.797)	0.981	−0.552 (−0.9189 to 0.084)	0.133	−0.349 (−0.908 to 0.391)	0.386	−0.182 (−0.9005 to 0.64)	0.677	−0.232 (−0.908 to 0.564)	0.585
HbA1c	−0.079 (−0.860 to 0.702)	0.848	0.149 (−0.595 to 0.894)	0.706	0.191 (−0.576 to 0.957)	0.641	0.312 (−0.484 to 0.9108)	0.468	0.401 (−0.352 to 0.915)	0.331
Hb	−0.224 (−0.966 to 0.921)	0.769	0.771 (−0.502 to 2.044)	0.274	0.614 (−0.756 to 0.998)	0.409	0.918 (−0.079 to 0.944)	0.109	0.456 (−0.907 to 0.991)	0.560
Creatinine	−0.144 (−0.900 to 0.713)	0.752	−0.317 (−0.9113 to 0.480)	0.461	−0.531 (−0.929 to 0.233)	0.215	0.735 (0.001 to 0.947)	0.090	−0.257 (−0.913 to 0.616)	0.582
Albumin	−0.568 (−0.919 to 0.054)	0.117	0.555 (−0.039 to 0.9148)	0.110	0.345 (−0.357 to 0.904)	0.367	0.559 (−0.120 to 0.9237)	0.150	0.407 (−0.308 to 0.912)	0.301
TAG	−0.295 (−0.901 to 0.424)	0.448	0.477 (−0.153 to 0.9107)	0.181	0.336 (−0.369 to 0.904)	0.381	−0.013 (−0.808 to 0.782)	0.975	0.642 (0.028 to 0.925)	0.080
HDL‐c	−0.927 (−0.943 to −0.420)	0.009 *	0.221 (−0.583 to 0.9024)	0.607	−0.058 (−0.906 to 0.790)	0.897	0.487 (−0.34 to 0.9315)	0.286	0.126 (−0.751 to 0.903)	0.786
LDL‐c	0.045 (−0.67 to 0.761)	0.905	0.013 (−0.675 to 0.701)	0.972	0.174 (−0.527 to 0.874)	0.642	0.155 (−0.595 to 0.904)	0.698	0.267 (−0.447 to 0.981)	0.487
Total‐c	−0.440 (−0.908 to 0.199)	0.219	0.219 (−0.451 to 0.889)	0.543	0.202 (−0.496 to 0.900)	0.588	0.365 (−0.345 to 0.9074)	0.348	0.429 (−0.242 to 0.910)	0.250
GPT	0.443 (−0.212 to 0.909)	0.227	−0.288 (−0.959 to 0.383)	0.428	−0.454 (−0.910 to 0.194)	0.212	0.244 (−0.510 to 0.999)	0.546	−0.695 (−0.925 to −0.139)	0.044 *
GOT	0.305 (−0.383 to 0.992)	0.414	−0.136 (−0.824 to 0.552)	0.710	−0.334 (−0.901 to 0.343)	0.366	0.172 (−0.584 to 0.927)	0.670	−0.531 (−0.916 to 0.106)	0.147
GGT	−0.086 (−0.912 to 0.946)	0.875	−0.414 (−0.9359 to 0.531)	0.419	−0.122 (−0.914 to 0.902)	0.821	−0.176 (−0.9263 to 0.911)	0.760	0.077 (−0.991 to 0.914)	0.891

Note: Standardized multiple regression model coefficients (β) and their 95% CI and p values are provided.

Abbreviations: BS, bariatric surgery; CRP, C‐reactive protein; GGT, gamma‐glutamyl aminotransferase; GOT, glutamic‐oxaloacetic transaminase/aspartate aminotransferase; GPT, glutamic‐pyruvic transaminase/alanine aminotransferase; Hb, total hemoglobin; Hb1Ac, glycated hemoglobin; HDL‐c, high‐density lipoprotein cholesterol; HOMA‐IR, homeostasis model assessment of insulin resistance; LAM, lipid‐associated macrophage; LDL‐c, low‐density lipoprotein cholesterol; PD, pre‐dietary intervention; PS, pre‐surgery; TAG, triglycerides; Total‐c, total cholesterol.

^*Bold: Statistically significant adjusted correlations (p < 0.05).

Finally, when we compared the gene expression levels of LAM markers at baseline and the clinical parameters 3 months after surgery to explore the potential prognosis of our markers in the success of the surgery, an inverse correlation was found between TREM2 baseline levels and body weight and BMI, and a positive correlation was detected between CD68 levels at baseline and insulin and HOMA‐IR 3 months after BS.

DISCUSSION

This study is the first to report transcriptomic analysis of SAT‐ and VAT‐infiltrated CD11b⁺ myeloid cells from individuals with severe obesity to characterize their potentially different phenotypes according to depot. We found that WAT‐infiltrated CD11b⁺ cells from individuals with severe obesity had a depot‐dependent gene expression profile, including different expression levels of biomarkers related to several subtypes of macrophages.

SAT‐infiltrated CD11b⁺ cells showed higher expression levels of biomarkers related to LAMs compared with those isolated from VAT. Due to their high expression of the scavenging receptor CD36, LAMs have been suggested to sequester adipocyte‐derived fatty acids [13]. These LAMs have also been shown to be important for efferocytosis, because TREM2 directly binds to the phospholipids of apoptotic cells [40, 41]. Genetic ablation of TREM2 enhanced adipocyte hypertrophy and altered tissue remodeling by dysregulating gene expression of lipid‐associated biomarkers of LAMs [13, 42]. In addition, a recent study showed that TREM2 is necessary for the BS‐induced reversal of disease [43]. The authors reported that TREM2 is required for dampening inflammation and effective efferocytosis of lipid‐laden hepatocytes by macrophages. In addition, these researchers postulated that BS restores AKT‐PI3K signaling, which is required for effective suppression of inflammation and oxidative phosphorylation [43]. In our study, SAT‐infiltrated CD11b⁺ cells showed a strong enrichment in the LAM signature, suggesting a relevant role in functions such as lipid sensing, lipid uptake, and lipogenesis. Moreover, genes associated with MAC5 (LGALS3, MMP12) and CLS (APOE, LRP1, LPL, and APP) were more highly expressed in SAT‐ than VAT‐infiltrated CD11b⁺ cells, which is compatible with the presence of a greater number of or a stronger functionality of LAMs in SAT based on observed bulk RNA‐sequencing transcriptional steady states. This could explain why SAT maintains a healthier phenotype than VAT during adipose tissue remodeling in obesity.

Previous studies showing the cellular landscape of human adipose tissue indicated a higher abundance of LAMs in individuals with obesity compared with lean individuals [12, 17, 44]. This suggests that adipose tissue remodeling in obesity may induce the metabolic phenotype of this macrophage subclass, although no different depots from the same individual have been previously studied at this level. In addition, a recent study [18] quantified the spatiotemporal dynamics of WAT macrophage infiltration and differentiation in WAT remodeling and suggested a critical role of LAMs in the formation of CLS in early obesity.

In the present study, VAT‐infiltrated CD11b⁺ cells showed an enrichment in the MAC1 or perivascular M2‐like ATM signature (high levels of LYVE1, TIMD4, MRC1 [CD206]). In contrast, there was no clear difference in MAC2–3 high proinflammatory gene signatures between VAT and SAT CD11b⁺ myeloid cells (high levels of IL6, TNF, IL1B, and CCL2) or monocyte‐derived macrophages (characterized by VCAN, HLA‐DRA, and LYZ). Another type of polarity described in tumor‐associated macrophages is CXCL9‐SPP1 [45]. Interestingly, we found differential expression of these two genes in VAT, with high CXCL9 and low SPP1, which would indicate that VAT‐infiltrated CD11b⁺ cells may have an enhanced proinflammatory profile, partially overlapping with M1‐like properties. It also presents overall enrichment with chronic inflammation signatures, such as those seen in inflammatory bowel disease, ulcerative colitis, allograft rejection, or Leishmania parasitic infection [46]. In addition, a signal for energy taxis (microbial attractants) was enriched in VAT, indicating chronic inflammation in the tissue.

Considering that having a higher amount of VAT is a cardiometabolic risk factor [6, 7], the increase in the LAM markers found in patients with obesity, and particularly in SAT compared with VAT, might suggest a role of this type of macrophages in a better adipose tissue remodeling for the adaptation to lipotoxicity and metabolic disturbances.

In addition, considering that the lack of TREM2 is related with adipocyte hypertrophy and altered adipose tissue remodeling [42], the inverse correlation observed in patients with obesity between SAT TREM2 gene expression and the body weight before the surgery might suggest a protective role of these LAMs in an obesogenic context. Indeed, the higher expression of this LAM marker was associated with a lower BMI 3 months after the surgery.

In conclusion, the present study demonstrated that WAT‐infiltrated CD11b⁺ myeloid cells isolated from individuals with severe obesity had a depot‐dependent gene expression profile. Moreover, SAT‐infiltrated CD11b⁺ cells showed a LAM phenotype compared with those isolated from VAT. Further studies are required to shed light on the potential role and function of these cell populations in the immunometabolism of individuals with obesity.

FUNDING INFORMATION

This work was supported by grant PID2020‐114953RB‐C21 funded by Ministerio de Ciencia, Innovación y Universidades, Agencia Estatal de Investigación, and the European Regional Development Fund (ERDF), by grant PID2023‐148783OB‐I00 funded by Ministerio de Ciencia, Innovación y Universidades, Agencia Estatal de Investigación, and ERDF, by grant CB06/03/0001 funded by Centro de Investigación Biomédica en Red Fisiopatología de la Obesidad y Nutrición and Instituto de Salud Carlos III, by grant 2021SGR00367 funded by the Government of Catalonia to Laura Herrero, and by grants PI17/00145 and PI20/00807 funded by Instituto de Salud Carlos III and ERDF to David Sánchez‐Infantes. Rubén Cereijo is a Serra Húnter Fellow (Government of Catalonia).

Supporting information

Table S1. (A) Comparison of adipose tissue macrophage biomarkers gene expression between visceral adipose tissue (VAT) and subcutaneous adipose tissue (SAT)‐derived CD11b⁺ cells from individuals with severe obesity. (B) Comparison of CD248 gene expression values between VAT and SAT‐derived CD11b⁺ cells from individuals with severe obesity.

Figure S1. (A) Boxplot showing nonsignificant differences (p > 0.05) in ITGAM gene expression (encoding CD11b protein) between visceral adipose tissue (VAT) and subcutaneous adipose tissue (SAT) CD11b positive cells. Normalized log2 intensity values obtained by microarray assay performed on total RNA extracted from fat depot‐derived cells were compared. (B) Relative transcript levels of CD11b, CD3D (T cell marker gene) and CD1c (dendritic cell marker gene) in samples isolated from SAT of participants from Cohort 1. Data is expressed as mean ± SEM and referenced to CD11b; mean fold change difference relative to CD11b levels is shown above each bar. AU: qPCR‐determined arbitrary units. ***p < 0.001 versus CD11b relative transcript levels.

Reyes‐Farias M, Fernández‐García P, Corrales P, et al. Lipid‐associated macrophages are more abundant in subcutaneous than visceral adipose tissue in patients with obesity. Obesity (Silver Spring). 2025;33(8):1543‐1554. doi: 10.1002/oby.24323

DATA AVAILABILITY STATEMENT

The data that support the findings of this study are available from the corresponding author upon reasonable request.