Early macrophage response to obesity encompasses Interferon Regulatory Factor 5 regulated mitochondrial architecture remodelling

Abstract

Adipose tissue macrophages (ATM) adapt to changes in their energetic microenvironment. Caloric excess, in a range from transient to diet-induced obesity, could result in the transition of ATMs from highly oxidative and protective to highly inflammatory and metabolically deleterious. Here, we demonstrate that Interferon Regulatory Factor 5 (IRF5) is a key regulator of macrophage oxidative capacity in response to caloric excess. ATMs from mice with genetic-deficiency of Irf5 are characterised by increased oxidative respiration and mitochondrial membrane potential. Transient inhibition of IRF5 activity leads to a similar respiratory phenotype as genomic deletion, and is reversible by reconstitution of IRF5 expression. We find that the highly oxidative nature of Irf5-deficient macrophages results from transcriptional de-repression of the mitochondrial matrix component Growth Hormone Inducible Transmembrane Protein (GHITM) gene. The Irf5-deficiency-associated high oxygen consumption could be alleviated by experimental suppression of Ghitm expression. ATMs and monocytes from patients with obesity or with type-2 diabetes retain the reciprocal regulatory relationship between Irf5 and Ghitm. Thus, our study provides insights into the mechanism of how the inflammatory transcription factor IRF5 controls physiological adaptation to diet-induced obesity via regulating mitochondrial architecture in macrophages.

Fig. 1. IRF5 expression is associated with ATM metabolic adaptation to short-term caloric excess.

a, b Wild-type (WT) mice and mice with myeloid-deficiency of IRF5 (IRF5-KO) were placed on (a) 12-week long-term (LT-) high-fat diet (HFD) or (b) 4-week short-term (ST-)HFD. F4/80⁺ adipose tissue macrophages (ATM) were sorted from epididymal fat pads (EpiWAT) for RNA-seq. Differentially expressed genes between genotypes were used for gene ontology analysis (n = 4 per genotype, Wald test p-value <0.05, Log₂FC > 1.0). Heatmaps represent fold-enrichment (Binomial test of gene list compared to reference genome, p values in Fig. S1A, B). c Irf5 expression in EpiWAT from C57BL/6 J mice on normal chow diet (NCD), ST-HFD or LT-HFD (n = 10 per group; Kruskal–Wallis multiple comparisons, left-to-right **p = 0,0063; **p = 0,0015). d JC1-Green and JC1-Red fluorescence to assess mitochondrial mass (MT Mass) and membrane potential (mΔΨ) in ATMs of C57BL/6 J mice on NCD, ST-HFD or LT-HFD (n = 10 per group on NCD/ST-HFD; n = 7 on LT-HFD; one-way ANOVA, **p = 0.0079; ****p = 0,000094). e Oxygen consumption rate (OCR) from ATMs of C57BL/6 J mice on NCD, ST-HFD or LT-HFD. Cells treated with Oligomycin (Oli), carbonyl cyanide 4-(trifluoromethoxy) phenylhydrazone (FCCP) and Rotenone/Antimycin A (Rot/AA) (n = 4 on NCD, n = 9 on ST-HFD, n = 8 on LT-HFD). f Correlation between IRF5 and JC1-Green MFI (Linear regression, Pearson correlation R² = 0.36; p = 0.02) and JC1-Red/JC1-Green MFI (R² = 0.22; p = 0.04) in ATMs of C57BL/6 J mice on NCD or ST-HFD (n = 9 per group). g Correlation between Irf5 and F4/80 expression in EpiWAT of C57BL/6 J mice on NCD or ST-HFD (n = 10 per group). h IRF5 MFI in F4/80^Lo and F4/80^Hi ATMs of C57BL/6 J mice on NCD or ST-HFD (n = 10 per group, two-tailed unpaired t-test, **p = 0.004). i Flow cytometry contour plots of JC1-Green and JC1-Red in F4/80^Hi ATMs of C57BL/6 J on NCD or ST-HFD. MFI of JC1-Green and JC1-Red in F4/80^Hi ATMs (n = 10 mice per group, two-tailed unpaired t-tests, *p = 0.01; ***p = 0.0009). j MFI of JC1-Red and JC1-Red/JC1-Green in F4/80^Hi EpiWAT ATMs of WT and IRF5-KO mice on ST-HFD (n = 10 for WT and 12 for IRF5-KO, two-tailed unpaired t-tests, **p = 0.0032; *p = 0.019). k tSNE plot of F4/80 and JC1-Red MFI on the stromal vascular fraction of EpiWAT, tSNE plot of JC1-Red MFI on F4/80^Hi populations from mice on ST-HFD. Data presented as mean ± SEM. Source Data file provided.

Fig. 2. Increased mitochondrial respiration in IRF5-deficient macrophages alters adipose tissue phenotype and function upon short-term high-fat diet.

a Oxygen consumption rate (OCR) from extracellular flux analysis, following Oligomycin (Oli), carbonyl cyanide 4-(trifluoromethoxy) phenylhydrazone (FCCP) and Rotenone/Antimycin A (Rot/AA) treatments, from epididymal white adipose tissue (EpiWAT) magnetically sorted F4/80⁺ cells from WT and IRF5-KO mice on normal chow diet (NCD; left), short-term (ST-) high-fat diet (HFD; middle) and long-term (LT-)HFD (right). Energetic plot with Extracellular acidification rate (ECAR) and OCR from maximal respiration (n = 4 mice per genotype for NCD, n = 6 WT and 8 IRF5-KO mice for ST-HFD and n = 5 WT and 4 IRF5-KO mice for LT-HFD). b OCR from extracellular flux analysis, following Oligo, FCCP and Rot/AA treatments, from EpiWAT stromal vascular fraction (SVF) of WT and IRF5-KO mice on ST-HFD (n = 5 mice per genotype). c OCR from extracellular flux analysis under basal conditions and following addition of glucose, performed on the EpiWAT SVF of WT and IRF5-KO mice on ST-HFD (n = 5 mice per genotype, ***p = 0.0002 and **p = 0.0018 two-way ANOVA). d Maximal OCR from Fig. S4B of fatty acid oxidation (FAO) test on EpiWAT SVF from WT (n = 7) and IRF5-KO (n = 9) mice on ST-HFD (*p = 0.035, one-way ANOVA). Palm palmitate, ETO etomoxir. e Representative images of hematoxylin and eosin (HE) staining for adipocyte size and MAC2/DAPI immunostaining to visualise crown-like structures (white arrows), on EpiWAT sections of WT and IRF5-KO mice on ST-HFD (scale bar = 100 um). Right: adipocyte size quantification on the HE staining (data pooled from biologically independent replicates, n = 20 WT and 21 IRF5-KO mice, two-tailed unpaired t-test, ***p = 0.0001). f Crown-like structure quantification on MAC2/DAPI immunostained EpiWAT sections of WT and IRF5-KO mice on ST-HFD (n = 14 WT and 17 IRF5-KO mice, two-tailed unpaired t-test Welch’s correction, *p = 0.0472). g Glucose uptake assay performed on EpiWAT explants from WT and IRF5-KO mice on ST-HFD (n = 9 WT and 10 IRF5-KO mice, two-tailed unpaired t-test Welch’s correction, p = 0.038). Data presented as mean ± SEM. Source Data file provided.

Fig. 3. IRF5-deficient hyperoxidative phenotype is cell intrinsic, inducible and reversible in mature bone-marrow-derived macrophages.

BMDMs from WT and IRF5-KO mice were treated with lipopolysaccharides (LPS) or palmitate (Palm) for 24 h. a Oxygen consumption rate (OCR) following addition of glucose (Glu) from extracellular flux analysis in Fig. S5A. (LPS ***p = 0.0001 and Palm **p = 0.008, two-tailed unpaired t-tests) (n = 5 per genotype). b OCR from extracellular flux analysis with following oligomycin (Oli), carbonyl cyanide 4-(trifluoromethoxy) phenylhydrazone (FCCP) and Rotenone/Antimycin A (Rot/AA) treatments (left) and area under the curve (AUC) (right) (n = 3 per genotype; *p = 0.038, and *p = 0.013 two-tailed unpaired t-test). c BMDMs from C57BL/6 J mice were treated with an IRF5-decoy peptide (DP) or a vehicle (Veh) and with palmitate (Palm). Energetic plot (left) of maximal respiration from extracellular flux analysis (right) and AUC (middle) (n = 3 per condition, two-tailed unpaired t-test, **p = 0.0022). d BMDMs from IRF5-KO mice were transfected with an IRF5 adenovirus (adIRF5) and treated with palmitate. Energetic plot (left) of maximal respiration from mitochondrial stress test (right) and AUC (middle) (n = 3 for KO + adIRF5 and KO + adIRF5 + Palm; n = 2 for KO, two-tailed unpaired t-test between KO + adIRF5 and KO + adIRF5 + Palm, *p = 0.01). Data presented as mean ± SEM. Source Data file provided.

Fig. 4. IRF5-KO alters TCA cycle metabolite concentrations and mitochondrial structural components in macrophages in response to lipotoxicity.

Bone-marrow-derived macrophages (BMDMs) from WT and IRF5-KO mice were treated with either bacterial lipopolysaccharides (LPS) or palmitate (Palm) for 2 or 24 h. Targeted metabolomics analyses were carried out to quantify intracellular tricarboxylic acid (TCA) cycle metabolites (n = 5–6 per condition) and electron microscopy was carried out to evaluate mitochondrial structural characteristics (n = 3 per condition). a Principal component analysis (PCA) on TCA cycle metabolites in all conditions. Palm 2 h condition separated (right). b PCA on TCA cycle metabolites in WT and IRF5-KO BMDMs stimulated with Palm for 2 h. c Variable weighting from PCA, percent variance contribution to principal component (PC)1, upon 2 h of Palm treatment. Lactate (Lac) intracellular concentration in WT and IRF5-KO BMDMs treated with Palm for 2 h (n = 6 per condition, **p = 0.003, two-tailed unpaired t-test). Aco aconitate, Gln glutamine, Gly glycerate, Asp aspartate, PEP phosphoenol pyruvate, Suc succinate, 3PG 3-phospho glycerate, AKG a-ketoglutarate, AHG a-hydroxyglutarate, Mal malate, Pyr pyruvate, Fum fumarate, Ita itaconate, Glt glutamate. d Schematic representation of metabolites with increased (red) or decreased (blue) abundance in IRF5-KO relative to WT BMDMs following treatment with Palm. Cit citrate, Iso isocitrate. e Extracellular acidification rate (ECAR) from extracellular flux analysis under Basal (Bas) and glucose-stimulated (Glu) conditions, of WT and IRF5-KO BMDMs treated with Palm for 2 h (n = 5 per condition, two-tailed unpaired t-tests, **p = 0.006, *p = 0.01). f Oxygen consumption rate (OCR) from extracellular flux analysis, with oligomycin (Oli), carbonyl cyanide 4-(trifluoromethoxy) phenylhydrazone (FCCP) and Rotenone/Antimycin A (Rot/AA) administration, performed on WT and IRF5-KO BMDMs, treated with Palm for 2 h (n = 5 per condition). g Electron micrograph and magnified inlet of BMDMs from WT and IRF5-KO mice after 2 h Palm treatment. Mitochondria are marked by ‘m’. h Mitochondrial cristae (white arrows) in electron micrograph and length and number of cristae of BMDMs from IRF5-KO and WT mice following Palm treatment for 2 h (*p = 0.01, *p = 0.04 unpaired t-test). Cell were derived from n = 3 independent animals per genotype and individual mitochondria were analysed in each sample (n = 26, 23 and 20 for WT and n = 11, 16 and 16 for IRF5-KO). Data presented as mean ± SEM. Source Data file provided.

Fig. 5. IRF5 binds to and regulates expression of genes that control mitochondrial structure and metabolism in bone-marrow-derived macrophages and adipose tissue macrophages in response to metabolic stress.

a Clustering analysis on RNA sequencing from bone-marrow-derived macrophages (BMDM) from IRF5-KO and WT mice treated for 2 or 24 h with bacterial lipopolysaccharides (LPS) or palmitate (Palm) and epididymal white adipose tissue (EpiWAT) F4/80⁺ macrophages (ATMs) from IRF5-KO and WT mice following short-term (ST-) or long-term (LT-) high-fat diet (HFD). Clustering analyses was applied to genes differentially expressed between genotypes in at least one condition. b Gene ontology (GO) term enrichment, related to mitochondria and lipotoxicity. Genes from differentially regulated clusters in panel a and Fig S7A (Binomial test of gene list compared to species reference genome). c Publicly available chromatin immunoprecipitation (ChIP) seq of IRF5 in BMDMs treated with LPS for 120 min was procured. Peaks of interest were determined as either at or upstream of transcription start sites (TSS). Annotated genes were subjected to gene ontology (GO) enrichment analyses. d Differential analysis between genotypes per treatment: 2- or 24-h treatment with LPS or Palm and in EpiWAT ATMs following ST- or LT-HFD (Wald test p-value <0.05, equivalent to −log₁₀ p-value > 1.3, and Log₂FC > 1.0). Venn diagram of differentially expressed genes between genotypes, per treatment condition. Percentage refers to proportion of genes in overlap. Gene track from ChIP-seq in c. of GHITM gene which overlaps all conditions and also bound by IRF5. e Correlative analyses of IRF5 expression and the expression of previously identified overlapping genes in d; and notably GHITM expression in ATMs from IRF5-competent mice fed a ST- or LT-HFD (Pearson’s correlation, Pearson r = −0.83, two-tailed p = 0.009). Data presented as mean ± SEM. Source Data file provided.

Fig. 6. IRF5 and GHITM are highly expressed and reciprocally regulated in epididymal white adipose tissue macrophages and monocytes.

a Single-cell RNA sequencing of the epididymal white adipose tissue (EpiWAT) stromal vascular fraction (SVF) of C57BL/6 J mice following 6 or 12 weeks of high-fat feeding (Jaitin et al., 2019). Macrophages and monocytes were identified and expression of IRF5 and of GHITM were projected onto tSNE plots per cell type and duration of high-fat feeding. b Heatmap of mean expression values of IRF5, GHITM, ABCG1, SYCE2, FNIP2, ATF5 and LRRC27 over time and by cell type [monocytes (Mono) or macrophages (Mac)]. c Correlative analyses between IRF5 expression and expression of GHITM, ABCG1, SYCE2, FNIP2, ATF5 and LRRC27 at the single-cell level (Pearson’s correlation, Pearson r; two-tailed ***p < 0.0001 and **p = 0.004). Heatmap of IRF5 and GHITM expression, each line represents a single cell. d Heatmap of single-cell expression of IRF5, GHITM, ABCG1, SYCE2, FNIP2, ATF5 and LRRC27 over time and by cell type, each line represents a single cell. e Gene expression of GHITM in bone-marrow-derived macrophages (BMDMs) from mice with myeloid-restricted Cas9-GFP expression, treated with lipofection agent (Ctrl) or with a guide RNA (gRNA) targeting GHITM (gGHITM) and with or without Palm treatment for 2 h (n = 5 for Ctrl, n = 6 for other conditions, one-way ANOVA. ***p = 0.0003, left *p = 0.0423, right *p = 0.0167). Western blotting against GHITM in the same experimental design, quantification and blot in Fig. S8D, S8E (n = 2 per condition). f Oxygen consumption rate (OCR) from extracellular flux analysis in BMDMs with or without Palm treatment following transfection with gGHITM or with lipofection agent alone (Ctrl). Oligomycin (Oli), carbonyl cyanide 4-(trifluoromethoxy) phenylhydrazone (FCCP) and Rotenone/Antimycin A (Rot/AA) were administered (n = 5 for Ctrl, Ctrl+Palm and gGHITM; n = 8 for gGHITM + Palm). g Maximal respiration from extracellular flux analysis (n = 5 for Ctrl, Ctrl + Palm and gGHITM; n = 8 for gGHITM+Palm; one-way ANOVA, ****p = 0.000065 and **p = 0.0051). h Maximal respiration from extracellular flux analysis on Palm-treated BMDMs following transfection with a gRNA targeting IRF5 (gIRF5), double transfection with gGHITM and gIRF5 or with lipofection agent alone (Ctrl) (n = 3 per condition; one-way ANOVA, *p = 0.0428). Data presented as mean ± SEM. Source Data file provided.

Fig. 7. IRF5 binds to GHITM and regulates mitochondrial activity in human monocytes and adipose tissue macrophages.

a CD14⁺ Monocytes from patients with type-2 diabetes (T2D; n = 5) were sorted based on expression of IRF5 for RNA-seq. Differential analyses were paired by patient and carried out on IRF5⁺ versus IRF5^- monocytes (n = 5, Wald test p-value < 0.05). b Gene ontology (GO) term enrichment from upregulated and downregulated genes in IRF5⁺ versus IRF5^- monocytes. c Expression of Ghitm in IRF5^- and IRF5⁺ monocytes (n = 5, *p = 0.039, two-tailed paired t-test). d Irf5 and Ghitm counts in white adipose tissue (WAT) macrophages and monocytes, from public dataset of scRNA-seq of the stromal vascular fraction (SVF) of patients that are lean or with obesity (two-tailed unpaired t-test, ****p < 0.0001). Heatmap of single-cell expression of Irf5 and Ghitm from monocytes (Mon) and macrophages (Mac), each line represents a single cell. e Proportion of Ghitm + (blue) and Ghitm− (red) cells in 10-cell bins by increasing Irf5 expression. f Correlation of Ghitm and Irf5 mean expression per bin (Pearson’s correlation Pearson R² = 0.28, two-tailed p < 0.0001). g Ghitm expression in CD14⁺ human visceral adipose tissue macrophages (vATMs). Samples were stratified based on expression of Irf5 into IRF5^Lo versus IRF5^Hi expressors (IRF5^Lo n = 7 and IRF5^Hi n = 6, two-tailed unpaired t-test, p = 0.13). h Correlation of IRF5 MFI, JC1-Green (mitochondrial mass, Mt Mass), JC1-Red (mΔΨ), and mΔΨ/mass in vATM from patients with obesity and in monocytes from patients with T2D (Mono) (n = 11 for monocytes, n = 9 for ATMs, Pearson’s correlation, Pearson r shown, p-values in Fig S9D). i Immunofluorescence staining of IRF5 (red), oxidative phosphorylation (OXPHOS) enzyme complexes (green) and CD14 (cyan) in human monocytes from patients with T2D. Nuclei stained visualised with DAPI (blue). Samples were separated based on IRF5 localisation, either nuclear (Nuc) or cytoplasmic (Cyt). Quantification of OXPHOS staining in Nuc and Cyt samples (n = 10 per condition, two-tailed unpaired t-test, *p = 0.0335). j University of California Santa Cruz (UCSC) genome browser^– (http://genome.ucsc.edu) tracks at the GHITM locus. JASPAR2020 tracks visualise transcription factor binding sites for IRF5. BLUEPRINT^, to visualise RNA expression and H3K27 acetylation marks in LPS-treated and -untreated human monocyte-derived macrophages (session link). Data presented as mean ± SEM. Source Data file provided.

Fig. 8. IRF5 transcriptional interaction with the mitochondrial structural protein GHITM limits adipose tissue macrophage oxidative capacity.

This mechanism alters mitochondrial cristae structures in adipose tissue macrophages to influence tissue adaptation in diet-induced obesity. Created with BioRender.com.