Serum Response Factor (SRF) promotes actin cytoskeletal organization in adipocytes to support adaptive hypertrophic expansion and tissue remodeling during obesity in mice

Abstract

Background:

Adipocyte hypertrophy, the unique capacity of adipocytes to enlarge in response to energy surplus, is a crucial determinant of metabolic health during obesity. Nonetheless, the molecular mechanisms governing this adaptive growth remain incompletely characterized.

Methods:

Super-enhancer landscapes in adipocytes were mapped via H3K27ac chromatin immunoprecipitation sequencing analysis of adipocyte nuclei from mice fed either a standard chow diet or high-fat diet (HFD) to identify transcriptional regulators activated under obesogenic conditions. Functional validation was conducted through both in vitro and in vivo experiments, including adipocyte-specific gene deletion mouse models, followed by single-nucleus RNA sequencing.

Results:

Super-enhancer profiling identified Serum Response Factor (SRF) as a critical driver of actin cytoskeletal remodeling in adipocytes during obesity. SRF was shown to be both necessary and sufficient for regulation of actin cytoskeletal gene expression in 3T3-L1 adipocytes. Adipocyte-specific SRF ablation in mice led to reduced expression of actin cytoskeletal genes, disruption of actin filament organization, and impaired adipocyte enlargement under HFD feeding. Despite comparable body weight, SRF-deficient mice developed exacerbated insulin resistance and ectopic lipid accumulation in the liver and brown adipose tissue, indicative of compromised lipid storage within adipocytes. Single-nucleus RNA-seq further revealed that cell-intrinsic actin cytoskeletal defects in adipocytes propagated to tissue-level dysfunction, impairing vascularization and increasing inflammation.

Conclusion:

These findings establish SRF as a central regulator of actin cytoskeletal organization that promotes healthy adipocyte hypertrophy and adipose tissue remodeling. Enhancing SRF-dependent cytoskeletal remodeling in adipocytes may offer a therapeutic strategy to preserve metabolic health in obesity.

1. Introduction

Obesity is a widespread and growing global health problem, significantly contributing to metabolic diseases such as type 2 diabetes (T2D) and cardiovascular complications [1,2]. Adipose tissue plays a crucial role in obesity, serving not only as a primary energy reservoir but also as an essential endocrine organ that regulates systemic energy balance and nutrient homeostasis [3]. However, chronic obesity often leads to adipose tissue dysfunction, characterized by impaired lipid metabolism, cell death, inflammation and insulin resistance [4]. These pathological changes are key precursors to the development of T2D and metabolic syndrome [4], highlighting the importance of understanding the mechanisms driving adipose tissue dysfunction during obesity.

Although increased adiposity is commonly linked to a higher risk of metabolic diseases, the relationship between adiposity and metabolic health is multifaceted. One of the most critical determinants is the mode by which adipose tissue expands during obesity. Adipose tissue expands through two mechanisms: adipocyte hyperplasia, an increase in adipocyte number, and adipocyte hypertrophy, enlargement of existing adipocytes [5]. Hyperplasia is often regarded as protective, promoting healthy lipid storage and preserving cellular function, whereas hypertrophy is associated with adipocyte dysfunction, triggering inflammation, cellular stress, and insulin resistance [6,7]. Despite this well-established dichotomy, the molecular mechanisms that enable adipocytes to successfully expand for lipid storage without incurring cellular damage remain poorly understood. Furthermore, the pathways through which these mechanisms become compromised in obesity, leading to adipocyte dysfunction, have yet to be clearly elucidated.

Adipocyte expansion involves complex biological processes such as lipogenesis, membrane stretching, and increased biosynthesis of RNA and proteins [4,8,9]. The cytoskeleton provides mechanical support to accommodate these structural and functional changes. For example, significant cytoskeletal remodeling is required during adipogenesis to facilitate the differentiation of progenitor cells into mature adipocytes by directing the shaping of their unique spherical morphology [10,11]. In mature adipocytes, this remodeling results in a distinct cytoskeletal organization, characterized by a diffuse cortical actin network at the cell periphery surrounding the large lipid droplets [12]. We previously showed that the actin filament organization is markedly enhanced in hypertrophic adipocytes during obesity, suggesting a potential role in accommodating cellular expansion [13,14]. Nevertheless, the precise functions and physiological relevance of the activated actin network in hypertrophic adipocytes remain largely undefined. While TGF-β signaling has been identified as a key factor in actin remodeling within adipocytes [14], the primary regulators of actin dynamics during obesity are yet to be elucidated.

Super-enhancers are large clusters of enhancers densely bound by transcription factors and co-activators, collaboratively driving the robust transcription of genes critical for cell identity and function [15]. Therefore, the analysis of super-enhancers has been instrumental in uncovering gene regulatory mechanisms involving transcription factors and chromatin-associated proteins that govern cellular differentiation, responses, and disease-related processes. Super-enhancers critical for adipogenesis and associated with the cooperative actions of adipogenic transcription factors have been identified in cultured adipocytes derived from both mice and humans [16,17]. In addition, super-enhancers involved in regulating adipocyte state transitions such as the white-to-brown conversion, have also been characterized [18]. However, super-enhancers in adipocytes from in vivo tissues remain undefined, despite well-documented differences in chromatin states between in vitro and in vivo environments. Furthermore, given the dramatic morphological and functional changes adipocytes undergo during obesity-induced hypertrophy, defining alterations in super-enhancer profiles in this context could reveal fundamental, yet unexplored, gene regulatory mechanisms.

In this study, we identify super-enhancers that arise in adipocytes during obesity and are functionally linked to actin cytoskeleton organization, a hallmark feature elevated in obese adipocytes. We further demonstrate that Serum Response Factor (SRF) serves as a transcriptional regulator of actin cytoskeletal genes, activated by obesity-induced super-enhancers. While SRF is both necessary and sufficient for the expression of these genes in vitro, it is also critical for actin cytoskeletal remodeling in vivo to support healthy adipocyte expansion. Loss of SRF impairs this process, leading to defective adipocyte enlargement, increased cell death, ectopic lipid deposition, and systemic insulin resistance. Furthermore, SRF is essential for promoting adipose tissue vascularization during obesity, underscoring its role as a key regulator of adipocyte expansion and stromal cell crosstalk during adipose tissue remodeling.

2. Methods

2.1. Animals

All animal experiments were conducted in accordance with protocols approved by the Institutional Animal Care and Use Committees (IACUC) at the Indiana University School of Medicine (IUSM) and Beth Israel Deaconess Medical Center (BIDMC). Mice were housed under a 12-hour light–dark cycle at 22 °C with ad libitum access to food and water. To generate inducible adipocyte-specific SRF knockout (SRF-AKO) mice, Srf-flox mice (Jackson Laboratory, 006658) were crossed with Adipoq-CreERT2 mice (Jackson Laboratory, 024671). For generation of SRF knockout mice specific to beige and brown adipocytes (SRF-BKO), Srf-flox mice were crossed with Ucp1-Cre mice (Jackson Laboratory, 024670). Tamoxifen (Sigma, T5648) was dissolved at 20 mg/mL in sunflower seed oil (Sigma, S5007) by incubation at 37 °C overnight with agitation. Littermates were housed together and subsequently allocated to experimental groups based on genotype: SRF-AKO (Srf^flox/flox; Adipoq-CreERT2⁺ or Ucp1-Cre⁺) or wild-type (WT) control (Srf^flox/flox; Cre-negative). Mice exhibiting fatal health problems or severe injury during experiments were designated for exclusion prior to analysis; however, no animals ultimately met these criteria. Investigators were generally aware of group allocation during animal handling and experimental procedures, while outcome assessments for selected experiments (e.g., transmission electron microscopy, TUNEL assay) were performed by personnel blinded to genotype. Mice received intraperitoneal (IP) injections of tamoxifen at a dose of 100 mg/kg body weight for 3 consecutive days. After a 1-week washout period, mice were subjected to cold exposure (4 °C) or high-fat diet (HFD) (Research Diets Inc., D12492i) feeding. For HFD-fed cohorts, an additional 3-day course of tamoxifen injections was administered 6 weeks after the start of the HFD. HFD experiments were performed by feeding 8-week-old mice a HFD for 12 to 16 weeks, and body weights were monitored weekly. At the end of experiments, mice were euthanized using CO₂, and tissues were collected and weighed, flash-frozen in liquid nitrogen, and stored at −80 °C until further analysis.

2.2. Cold challenge

A cold challenge was performed to activate thermogenesis in brown adipose tissue (BAT) and to induce beige adipocyte formation in inguinal white adipose tissue (iWAT), enabling evaluation of in vivo adipocyte thermogenic capacity and its potential contribution to phenotypes observed during HFD feeding experiments. Mice were individually housed with bedding in a rodent incubator (Power Scientific Inc., RIT33SD) maintained at 4 °C. Water and food were provided ad libitum throughout the experiment. Core body temperature was measured at indicated time points using a rectal probe (Physitemp BAT-12 with RET-3 probe). Upon completion of the experiment, tissues were harvested, flash-frozen in liquid nitrogen, and stored at −80 °C.

2.3. Indirect calorimetry

Metabolic parameters were assessed using indirect calorimetry in open-circuit Oxymax chambers, part of the Comprehensive Lab Animal Monitoring System (CLAMS; Columbus Instruments). Mice were singly housed and acclimated to the metabolic cages for 72 h, followed by 72 h of automated hourly recordings of oxygen consumption, carbon dioxide production, and other physiological parameters. Mice were maintained on a 12-h light–dark cycle at 23 °C with ad libitum access to food and water. Data were analyzed using ANCOVA via the CalR software [19].

2.4. Glucose homeostasis tests

For glucose tolerance tests, mice were fasted for 12 to 16 h and then injected IP with glucose (1 g/kg body weight). Blood samples were collected at 0, 10, 20, 30, 60, 90, and 120 min post-injection to measure glucose levels with a portable glucometer. For insulin tolerance tests, mice were fasted for 4 to 6 h and then injected IP with insulin (1 U/kg body weight). Blood glucose levels were measured at 0, 15, 30, 45 and 60 min following insulin injection. To assess fasting plasma insulin levels, blood was collected into EDTA-coated tubes after a 6-hour fast and analyzed using an ELISA kit (Crystal Chem) according to the manufacturer’s instructions.

2.5. Adipose tissue explant lipolysis assay

Upon harvest, mouse adipose tissues were minced into small pieces, equilibrated in serum-free DMEM/F12 medium supplemented with 2% fatty acid-free bovine serum albumin (BSA), and then transferred to fresh medium for lipolysis assays. Lipolysis was assessed under basal condition or after stimulation with 20 μM isoproterenol for 4 h at 37 °C. Glycerol released into the culture medium was quantified using a colorimetric assay kit (Sigma, F6428) and normalized to tissue weight.

2.6. Tissue histology and immunofluorescence staining

For histological analysis, adipose tissues were fixed in 10% neutral-buffered formalin overnight at 4 °C, washed in PBS, transferred to 70% ethanol, and submitted to the IUSM Histology Core for paraffin embedding and hematoxylin and eosin (H&E) staining.

For whole-mount immunofluorescence, adipose tissues were cut into small pieces, fixed in 10% neutral-buffered formalin overnight at 4 °C, and washed twice in PBS (30 min each). Samples were then permeabilized and blocked overnight at 4 °C in PBS containing 2.5% BSA and 1% Triton X-100. Primary antibodies targeting PECAM1 (BD Biosciences, 553370, 1:200) or F4/80 (Abcam, ab6640, 1:200) were diluted in blocking buffer and applied for 48 h at 4 °C. Unbound antibodies were removed with 4 washes, each lasting 1 h, in PBS. Tissues were incubated with Alexa Fluor 647-conjugated donkey anti-rat secondary antibody (Invitrogen, A-21247, 1:500) overnight at 4 °C. After secondary antibody removal with 4 successive 1-h PBS washes, tissues were stained with BODIPY (ThermoFisher, D3922, 2 μg/mL) overnight at room temperature on a rotator. Nuclei were counterstained with Hoechst 33342 (ThermoFisher, 1 μg/mL) for 30 min after 2 washes in PBS. Samples were mounted in anti-fade mounting medium (Electron Microscopy Sciences, 17989–50) and imaged with a Leica confocal microscope.

For phalloidin staining of isolated adipocytes, adipose tissues were minced and digested in PBS containing collagenase D (1.5 U/mL) and dispase II (2.4 U/mL) at 37 °C approximately for 40 min with intermittent shaking. Digested tissues were centrifuged at 300 ×g for 5 min at room temperature, and the floating mature adipocytes were collected and fixed in 2% paraformaldehyde (PFA) for 30 min at room temperature with gentle agitation. Fixed adipocytes were washed twice in wash buffer (PBS with 1% fatty acid–free BSA and 0.05% Tween-20), then permeabilized and blocked for 1 h at room temperature in PBS with 0.5% Triton X-100 and 2% fatty acid–free BSA. Samples were incubated overnight at 4 °C with an antibody against Perilipin 1 (PLIN1) (Abcam, ab61682; 1:200), followed by 1-h incubation at room temperature with Alexa Fluor 488-conjugated donkey anti-goat IgG (Invitrogen) and phalloidin (ThermoFisher, A12380). After washes, nuclei were stained with Hoechst 33342 (TheromoFisher) for 30 min. Samples were mounted on 0.5 mm chambered slides, sealed with coverslips, and imaged using a Leica confocal microscope.

2.7. Adipose tissue explant compression assay

Adipose tissues were harvested and fixed in 2% PFA at 4 °C overnight with shaking. Fixed tissues were washed twice with PBS for 30 min each at 4 °C and then placed between two glass slides. Compression was applied by securing both ends of the slides with binder clips for 90 min at 4 °C. Following compression, tissues were permeabilized and blocked overnight at 4 °C with gentle agitation in PBS containing 2.5% BSA and 1% Triton X-100. Samples were then stained with BODIPY (ThermoFisher, D3835, 2 μg/mL) in PBS at room temperature while rotating. After 4 repeated 1-hour PBS washes, tissues were counterstained with Hoechst 33342 (ThermoFisher, 1 μg/mL) for 30 min, followed by a final PBS wash. Stained tissues were mounted using anti-fade mounting medium (Electron Microscopy Sciences, 17989–50) and imaged using a Leica confocal microscope.

2.8. Transmission electron microscopy

Fresh adipose tissues were fixed in 3% glutaraldehyde in 0.1 M sodium cacodylate (SC) buffer, rinsed, and post-fixed with 1% osmium tetroxide in SC buffer for 1 h. After rinsing, samples were dehydrated through a graded ethanol series (70–100%), then infiltrated with 100% acetone and a 1:1 acetone:resin mixture (Electron Microscopy Sciences, Embed 812) for 2 to 3 days. Sample vials were then uncapped and left open for 3 h to allow acetone to evaporate. Samples were subsequently embedded in fresh 100% resin and polymerized overnight at 60 °C. Ultrathin sections (80–90 nm) were cut, stained with uranyl acetate replacement stain (Electron Microscopy Sciences), and imaged using a Tecnai Spirit transmission electron microscope (ThermoFisher). Digital images were acquired using a charged-coupled device (CCD) camera (AMT).

2.9. Cell culture

3T3-L1 preadipocytes (ATCC) were maintained in DMEM (Invitrogen) supplemented with 10% bovine calf serum (HyClone) and 1% penicillin–streptomycin (Gibco). Upon reaching confluence, cells were cultured for an additional 2 days before initiating differentiation. Differentiation was induced by treating cells with DMEM containing 10% fetal bovine serum (FBS; Peak Serum), 5 μg/mL insulin, 500 μM isobutylmethylxanthine, and 1 μM dexamethasone for 48 h. Cells were then maintained in DMEM supplemented with 10% FBS and 5 μg/mL insulin for another 2 days, followed by maintenance in DMEM with 10% FBS alone for the remainder of the culture period.

For TGF-β treatment, recombinant mouse TGF-β1 (R&D Systems, 7666-MB-005) was reconstituted at 50 μg/mL in 4 mM HCl containing 0.1% BSA and added to culture medium at a final concentration of 1 ng/mL. For lentiviral transduction, viral particles were produced by cotransfecting HEK293T cells with lentiviral constructs and the packaging plasmids pMD2.G and psPAX2. Supernatants were harvested 48 h post-transfection and applied to 3T3-L1 adipocytes for 24 h. After infection, cells were maintained in fresh culture medium and used for downstream analyses 4 or 5 days later. For knockdown experiments, pLKO.1 vectors targeting a scrambled sequence (control) or mouse Srf (Sigma, TRCN0000054597, TRCN0000054594) were used. For overexpression, Srf cDNA was subcloned from pMSCV-SRF [20] into the pCDH1 vector. pCDH1-GFP was used as a control.

2.10. SRF chromatin immunoprecipitation sequencing (ChIP-seq)

SRF ChIP-seq was performed as previously described [21] with minor modifications. Differentiated 3T3-L1 adipocytes were washed twice with PBS and collected in nucleus preparation buffer (NPB; 10 mM HEPES, pH 7.5, 1.5 mM MgCl₂, 10 mM KCl, 250 mM sucrose, 0.1% NP-40, 0.2 mM DTT). Cells were homogenized using a dounce homogenizer with 15 strokes each using loose and tight pestles. Homogenates were filtered through 100 μm cell strainers and immediately cross-linked with 1% PFA for 10 min at room temperature with gentle shaking. Cross-linking was quenched by addition of 125 mM glycine for 10 min. Homogenates were centrifuged at 1000 ×g for 10 min, and nuclear pellets were washed once in NPB and again in PBS with 0.1% NP-40 before being resuspended in nuclear lysis buffer (NLB; 10 mM Tris-HCl, pH 8.0, 1 mM EDTA, 0.25% SDS). Chromatin was sheared using a Covaris E220 sonicator for 8 min and cleared by centrifugation at 13,000 rpm for 10 min at 4 °C. The supernatant containing fragmented chromatin was diluted in ChIP dilution buffer (16.7 mM Tris-HCl, pH 8.0, 1.2 mM EDTA, 167 mM NaCl, 1.1% Triton X-100, 0.01% SDS) and incubated overnight at 4 °C with 1 μg/mL SRF antibody (Santa Cruz Biotechnology, sc-335X). Immunoprecipitates were incubated with Protein A/G Dynabeads (Invitrogen) at 4 °C for 1 h. Beads were sequentially washed twice each with low-salt buffer (20 mM Tris-HCl, pH 8.0, 1 mM EDTA, 140 mM NaCl, 1% Triton X-100, 0.1% sodium deoxycholate, 0.1% SDS), high-salt buffer (same as above but with 500 mM NaCl), LiCl buffer (10 mM Tris-HCl, pH 8.0, 1 mM EDTA, 0.5% NP-40, 0.5% sodium deoxycholate, 250 mM LiCl), and TE buffer (10 mM Tris-HCl, pH 8.0, 1 mM EDTA). Chromatin was eluted and reverse cross-linked in ChIP elution buffer (10 mM Tris-HCl, pH 8.0, 50 mM EDTA, 0.1% SDS, 300 mM NaCl) supplemented with 0.8 mg/mL proteinase K and 10 μg/mL RNase A, incubated at 65 °C for 8 h. DNA was purified using AMPure XP beads (Beckman Coulter, A63881) according to the manufacturer’s instructions.

Sequencing libraries were prepared using the on-bead library preparation protocol [21]. Briefly, purified DNA was subject to end repair and phosphorylation using the End-It DNA End-Repair Kit (Epicenter, ER0720), followed by A-tailing with Klenow Fragment (NEB, M0212) and adaptor ligation with Quick Ligase (NEB, M2200). Libraries were then PCR-amplified for 18 cycles with PfuUltra II Hotstart PCR Master Mix (Agilent, 600850). Size selection of fragments (250–600 bp) was performed using E-Gel EX Agarose Gels (Invitrogen) followed by gel extraction with the MinElute Gel Extraction Kit (Qiagen). Library concentration and quality were assessed by Qubit fluorometry (Invitrogen) and Bioanalyzer (Agilent), respectively. Indexed libraries were pooled to a final concentration of 12 pM and sequenced on an Illumina NextSeq 500.

2.11. Protein isolation and Western blotting

Cells were lysed in radioimmunoprecipitation assay (RIPA) buffer containing 20 mM Tris-HCl (pH 8.0), 140 mM NaCl, 1% Triton X-100, 0.1% sodium deoxycholate, 0.1% SDS, and 1 mM EDTA, supplemented with EDTA-free protease inhibitor cocktail (Roche). Protein concentrations were determined using bicinchoninic acid (BCA) assay (Thermo Fisher Scientific, PI23228). Equal amounts of proteins were separated by 4–15% gradient SDS-PAGE and transferred to polyvinylidene difluoride (PVDF) membranes (Millipore). Membranes were incubated with primary antibodies followed by horseradish peroxidase (HRP)-conjugated secondary antibodies and developed using the Western Lightning ECL detection system (PerkinElmer, NEL103001EA). The following primary antibodies were used: SRF (Santa Cruz Biotechnology, sc-335X), ACTA2 (Abcam, ab5694), and GAPDH (Cell Signaling Technology, 2118S).

2.12. RNA isolation and quantitative real-time PCR

RNA was isolated from cells or tissues using TRIzol reagent (Invitrogen) according to standard procedures. cDNA was synthesized from 500 ng of total RNA using the High-Capacity cDNA Reverse Transcription Kit (Applied Biosystems) following the manufacturer’s instructions. Quantitative real-time PCR was carried out using SYBR Green Master Mix (Applied Biosystems) on a QuantStudio 5 Real-Time PCR System. Expression levels were normalized to the housekeeping gene Rplp0, and fold changes were calculated using the ΔΔCt method. The primers used for qRT-PCR are provided in Table S1.

2.13. Bulk RNA sequencing

Extracted RNA was purified by on-column DNA digestion using TURBO DNase kit (Invitrogen). For library construction, 100 ng of purified RNA was processed by NEBNext rRNA Depletion Kit (New England BioLabs, Inc.) to remove ribosomal RNA. Next, RNA was converted to cDNA using Maxima Reverse Transcriptase (Thermo Fisher Scientific) and NEBNext mRNA Second Strand Synthesis kit. Following purification based on size selection using AMPure XP beads (Beckman Coulter), sequencing libraries were generated through tagmentation using Nextera XT DNA Library Preparation kit and subsequent PCR amplification. The quantity and quality of the resulting libraries were assessed using Qubit fluorometry (Invitrogen) and Bioanalyzer (Agilent), respectively. Finally, the libraries were sequenced on an Illumina NovaSeq 6000.

2.14. Single-nucleus RNA sequencing

Single-nucleus RNA-sequencing was performed as previously described [22]. Briefly, nuclei were isolated from frozen tissues by pulverization using mortars and pestles in liquid nitrogen and then homogenized in chilled dounce homogenizers containing NPB buffer added with 10 mM vanadyl ribonucleoside complex (VRC, New England Biolabs), 1× Halt protease inhibitor (Thermo Fisher), 1 mM DTT, and 0.4 U/μL NxGen RNase inhibitor (Biosearch Technologies).

Tissue homogenates were filtered through 100 μm cell strainers, and centrifuged at 200 xg for 10 min at 4 °C. The nuclear pellets were resuspended in PBS containing 0.1% NP-40 and 5 mM VRC (PBS-N-VRC). Hoechst 33342 (ThermoFisher, 2 μg/mL) was added to the homogenates to label nuclei. Following centrifugation at 200 xg for 5 min at 4 °C, the nuclear pellets were resuspended in PBS-N-VRC and filtered through 40 μm cell strainers prior to sorting. The nuclear suspension was then subjected to sorting on a BD FACSAria Fusion system to eliminate debris and nuclear multiplets based on forward and side scatter profiles and Hoechst fluorescence. Sorted nuclei collected in PBS-N-VRC were centrifuged at 200 xg for 5 min at 4 °C in a swinging-bucket rotor. The resulting pellet was resuspended in PBS containing 0.1% NP-40 to achieve a concentration of 1000 nuclei/μL. Nuclei were counted using a hemocytometer, and 10,000 nuclei per sample were loaded onto the 10× Genomics Chromium Controller. Libraries were prepared using the Chromium Single Cell 3’ Reagent Kit v3.3 (10× Genomics) and sequenced on an Illumina NovaSeq 6000.

2.15. Bioinformatic analysis

2.15.1. ChIP-seq data analysis

Raw sequencing reads underwent quality assessment, adapter removal, and filtering using fastp [23]. The cleaned reads were aligned to the mm10 mouse genome with Bowtie2 [24], and PCR duplicates as well as low-quality reads were removed using SAMtools [25]. Peak calling was performed with MACS2 [26]. For comparative analysis, peaks from all samples were merged using BEDOPS [27], and read counts per peak were quantified using featureCounts [28]. Only peaks with log counts per million (CPM) greater than 1 across all samples were retained for downstream differential analysis using edgeR [29]. Motif enrichment was analyzed de novo with HOMER [30] and MEME [31]. Genomic annotation of peaks and pathway enrichment analyses were conducted via GREAT [32]. ChIP-seq coverage tracks were visualized on the WashU Epigenome Browser [33] using BigWig files generated with deepTools [34], normalized by sequencing depth.

2.15.2. Super-enhancer analysis

Super-enhancers were identified through reanalysis of our previously published H3K27ac ChIP-seq data from adipocyte nuclei [14]. Peak calling was performed independently for both chow- and HFD-fed mice using the Rank Ordering of Super-Enhancers (ROSE) algorithm [15], originally developed by the Young lab. Peaks overlapping blacklisted genomic regions [35] were excluded using BEDTools [36] intersect, and the remaining peaks were subsequently converted to GFF format for downstream analysis. ROSE was executed with a stitching distance of 12.5 kb and a transcription start site (TSS) exclusion zone of 2.5 kb. Super-enhancer ranking plots were generated using GraphPad Prism, and functional pathway analysis was performed using GREAT [32].

2.15.3. Bulk RNA-seq analysis

RNA-seq reads were quality-filtered and trimmed using fastp [23], then aligned to the mm10 reference genome with STAR [37]. Gene-level quantification was performed with featureCounts [28]. Genes exhibiting low expression (CPM ≤ 1 in more than half of the samples) were excluded prior to differential expression analysis with edgeR [29]. Differentially expressed genes were defined by and log₂fold change >0.5 and FDR < 0.05. Pathway enrichment analysis was conducted using Metascape [38], and heatmaps were generated with Morpheus (https://software.broadinstitute.org/morpheus/).

2.15.4. snRNA-seq analysis

The snRNA-seq data analysis pipeline was performed as previously described [22]. Briefly, for each sample, raw sequencing data were processed using Cell Ranger (10× Genomics, v7.1.0) which performs demultiplexing, alignment to the reference genome, filtering of reads, barcode and unique molecular identifier (UMI) counting, and generates gene-barcode matrices. To further remove ambient RNA contamination, CellBender v0.3.0 [39] was applied. The resulting counts were imported into R and processed with Seurat 5.3.0 [40] for downstream analyses.

Additional cell-level filtering was performed to retain nuclei with at least 500 UMIs and 250 detected genes, and with a mitochondrial gene expression ratio below 15%. Genes expressed in fewer than 10 nuclei and all mitochondrial genes were excluded. Doublets were identified and removed using scDblFinder [41]. For integration across samples, data were merged, normalized using SCTransform, and integrated using Harmony [42]. The integrated dataset was visualized using uniform manifold approximation and projection (UMAP) for dimensionality reduction and clustered at a resolution of 0.4. Clusters were annotated as distinct cell types based on canonical marker gene expression. To achieve higher resolution, specific cell type clusters were reclustered as needed. To assess intercellular communication within adipose tissue, CellChat [43] was performed to analyze ligand-receptor interactions in each sample, comparing between SRF-AKO mice and WT controls within each fat depot. For visualization, the scCustomize R package [44] was used to generate plots.

2.16. Statistical analysis

Statistical analyses were performed using GraphPad Prism 9 and R. Data are presented as mean ± SEM. Two-tailed unpaired Student’s t-test was used for pair-wise comparisons. Assumptions of normality and homogeneity of variance were tested using the Shapiro-Wilk and F-tests, respectively. A p value <0.05 was considered statistically significant unless otherwise specified. Differential expression analyses were conducted with edgeR [29] using the exact test as described by the package’s exactTest function.

3. Results

3.1. Adipocyte super-enhancers are associated with actin cytoskeletal genes in obesity

We previously developed a transgenic reporter mouse called Nuclear Tagging and Translating Ribosome Affinity Purification (NuTRAP), which allows cell type-specific transcriptomic and epigenomic profiling [21]. Using this model, our prior work revealed that adipocytes acquire myofibroblast-like gene expression features and exhibit reduced expression of metabolic genes during obesity [14]. To investigate whether these changes reflect shifts in adipocyte cellular identity in obesity, we analyzed super-enhancers using our previously generated H3K27ac ChIP-seq data from adipocyte nuclei isolated from eWAT of chow- or HFD-fed mice [14]. We identified 864 and 842 super-enhancers in adipocytes from chow- and HFD-fed mice, respectively, with the majority (689) shared between the two conditions (Fig. 1A–C). Gene ontology analysis showed that super-enhancers from both conditions are associated with biological pathways related to insulin response and lipid metabolism (Fig. 1D, top panel), in line with the core metabolic functions of adipocytes. These super-enhancers encompass well-established adipocyte genes such as Cebpa, Adipoq, Lep, Scd1, Plin1, and Abca1 (Fig. 1A, B). In addition, the shared super-enhancers were enriched near genes involved in cell-substrate adherens junction and focal adhesion (Fig. 1D, bottom panel), highlighting the importance of these structural elements in maintaining adipocyte identity and function. Overall, these results indicate that super-enhancer activity in adipocytes remains largely unchanged across dietary conditions. However, we observed a subset of super-enhancers that exhibited HFD-specific changes. For example, super-enhancer signals at Cebpa were reduced, while those at Lep were increased in HFD-fed mice (Fig. 1E). Notably, super-enhancers from HFD-fed mice showed unique enrichment for pathways related to the actin cytoskeleton, actomyosin, and stress fibers (Fig. 1D, bottom right), with associated genes including Sorbs1, Actg1, Mylk, Src, and Actn4 (Fig. 1B, F). Together, these findings suggest that while the core cellular identity of adipocytes is preserved, obesity specifically activates genes involved in actin cytoskeleton remodeling.

Fig. 1.

Identification of adipocyte super-enhancers associated with obesity. (A–C) Identification of super-enhancers (SE) in adipocyte nuclei from (A) chow-fed and (B) HFD-fed male mice using the ROSE algorithm, which stitches together nearby H3K27ac ChIP-seq peaks, ranks them by total signal intensity, and designates those above the inflection point of the ranked signal curve as super-enhancers (shown in red). (C) Venn diagram showing the overlap of super-enhancers identified in adipocyte nuclei from chow-fed and HFD-fed mice. (D) Gene Ontology (GO) analysis using GREAT of enriched pathways—Biological Process (top) and Cellular Component (bottom)—associated with genes linked to identified super-enhancers. Actin-related pathways enriched in HFD-associated super-enhancers are highlighted in bold. (E–F) Genomic tracks of H3K27ac ChIP-seq signals at representative adipocyte super-enhancer regions: (E) Cebpa and Lep, and (F) Sorbs1.

3.2. SRF is necessary and sufficient for actin cytoskeletal gene expression in adipocytes in vitro

To identify transcription factors that potentially bind super-enhancers regulating actin cytoskeletal genes in adipocytes, we performed motif enrichment analysis on promoter and enhancer elements marked by H3K27ac within super-enhancers identified in HFD-fed mice. This analysis revealed significantly enrichment of conserved SRF-binding motifs within these elements, consistent with the established role of SRF in regulating cytoskeletal gene expression and muscle development (Fig. 2A). Using SRF ChIP-seq in differentiated 3T3-L1 adipocytes treated with TGFβ1, a known inducer of actin cytoskeletal gene activation [14], we identified de novo DNA motifs enriched within SRF-bound regions. These motifs closely matched the canonical SRF motif, supporting SRF’s precise recruitment to key regulatory elements of actin cytoskeletal genes (Fig. 2B). Pathway analysis of genes nearest to SRF-binding peaks showed significant enrichment in pathways related to actin filament bundles, actomyosin and stress fiber (Fig. 2C). The Acta2 gene exemplifies direct SRF binding upon TGFβ1 stimulation, with the binding site located within a super-enhancer region (Fig. 2D). To assess the functional role of SRF in regulating actin cytoskeletal gene expression in adipocytes, we manipulated Srf expression in differentiated 3T3-L1 adipocytes. Lentivirus-mediated knockdown of Srf using shRNA significantly reduced the expression of key cytoskeletal genes, including Acta2 and Col1a1 (Fig. 2E). In contrast, SRF overexpression led to a significant upregulation of Acta2, Tagln, and Col1a1 (Fig. 2F). Consistently, ACTA2 protein levels were also elevated following SRF overexpression (Fig. 2G). Taken together, these results demonstrate that SRF drives actin cytoskeletal gene expression in adipocytes, highlighting its essential regulatory role.

Fig. 2.

Critical role of SRF in regulating actin cytoskeletal gene expression in adipocytes in vitro. (A) Motif identified by MEME that is enriched in H3K27ac peaks within the HFD-associated super-enhancer regions. (B) Alignment of the SRF binding motif identified de novo from SRF ChIP-Seq in TGFβ1-treated 3T3L1 adipocytes, compared to the canonical SRF motif from the HOMER database. (C) Pathway analysis of genes associated with SRF ChIP-seq peaks. (D) Genomic tracks showing SRF and H3K27ac ChIP-seq signals at the Acta2 locus, highlighting SRF binding induced by TGFβ1 treatment (indicated by a black arrow) within an adipocyte super-enhancer region. (E–G) In vitro loss- and gain-of function experiments in 3T3-L1 adipocytes. Gene expression analysis of cytoskeletal genes following (E) Srf knockdown (shSrf, n = 3 per condition, total N = 6) and (F) overexpression (Srf OE, n = 3 per condition, total N = 6). (G) Western blot analysis of SRF and ACTA2 protein levels upon Srf overexpression (Srf OE, n = 2 per condition, total N = 4). A two-tailed Student’s t-test was used for statistical analysis. *P < 0.05, **P < 0.01, ***P < 0.001.

3.3. SRF is not required for adipose tissue thermogenesis

Thermogenesis in BAT and cold-inducible beige adipocytes within WAT is a key adaptive mechanism that can limit adipose tissue expansion during obesogenic conditions by increasing energy expenditure [45]. Although SRF has been shown to inhibit brown and beige adipogenesis in in vitro cell culture models [46], yet its function in regulating adipose tissue thermogenesis in vivo remains to be elucidated. To determine whether SRF contributes to brown and beige thermogenesis and/or white adipocyte expansion, we generated SRF-AKO mice by crossing Adipoq-CreERT2 with Srf-floxed mice. This tamoxifen-inducible mouse model minimizes potential confounding effects on de novo adipogenesis, allowing us to directly evaluate SRF function in mature adipocytes. Following tamoxifen administrations, SRF-AKO mice were exposed to cold (4 °C) for 7 days to robustly activate thermogenesis in BAT and beige adipocytes within iWAT. We found no significant differences in the expression of key thermogenic genes, including Ucp1, Pgc1a, and Cox7a1, between SRF-AKO mice and WT control mice in either BAT or iWAT (Fig. S1A, B). Because SRF-AKO induces SRF deletion in all types of adipocytes, we next generated brown and beige adipocyte-specific SRF knockout (SRF-BKO) mice by crossing Ucp1-Cre with Srf-floxed mice. To evaluate acute thermogenic responses, SRF-BKO mice were exposed to cold (4 °C) for 6 h, resulting in no significant differences in body weight, core body temperature or thermogenic gene expression compared to controls (Fig. S1C–F). During prolonged cold exposure for 7 days, SRF-BKO mice showed comparable body weights and a trend toward lower body temperatures, although the difference did not reach statistically significance (Fig. S1G, H). Consistently, no significant differences were found in thermogenic gene expression in BAT or iWAT (Fig. S1I, J) nor in adipocyte size or morphology in iWAT or eWAT (Fig. S1K, L). These findings suggest that SRF in mature adipocytes may be dispensable for regulating adipocyte thermogenic activity in BAT and iWAT.

3.4. SRF is critical for actin filament formation and cellular expansion during obesity in vivo

Having excluded the role of SRF in adipose tissue thermogenesis, we next investigated its function in white adipocytes during obesity. Following tamoxifen injections, SRF-AKO mice were fed a HFD for 3 months, with an additional tamoxifen dose administered at 6 weeks to induce knockout in newly formed adipocytes. SRF-AKO mice did not exhibit significant differences in body weight compared to WT controls (Fig. 3A). RNA-seq analysis of eWAT and iWAT revealed a pronounced downregulation of actin cytoskeletal and collagen genes in SRF-AKO mice (Figs. 3B, S2A). Pathway enrichment analyses indicated significant suppression of biological processes related to supramolecular fiber organization, focal adhesion or actin filament-based processes in the WAT of SRF-AKO mice (Fig. S2B). Building on our previous findings that adipocytes develop complex actin filament structures during obesity [14], we examined the impact of SRF-AKO on actin filament formation by performing phalloidin staining on isolated primary adipocytes. Z-stacked confocal microscopy imaging showed fragmented and poorly developed actin filaments in SRF-AKO adipocytes, in contrast to the dense actin filament networks observed in WT mice (Fig. 3C). To confirm that the reduction in phalloidin staining was due to cytoskeletal rather than general adipocyte developmental defects, we co-stained adipocytes with phalloidin and PLIN1, a lipid droplet marker. While PLIN1 staining was comparable between WT and SRF-AKO adipocytes, phalloidin staining was markedly reduced in SRF-AKO adipocytes in both eWAT and iWAT (Fig. 3D, E). These findings demonstrate that SRF is essential for maintaining proper actin cytoskeletal architecture in adipocytes during obesity. Furthermore, the marked reduction in adipocyte size in both eWAT and iWAT from SRF-AKO mice (Fig. 3F, G) suggest that SRF supports adipocyte hypertrophy by preserving actin cytoskeletal integrity.

Fig. 3.

Role of SRF in actin filament structure and cellular expansion in adipocytes in vivo during obesity. (A) Body weight trajectories of wild-type (WT, n = 11) and tamoxifen-inducible, adipocyte-specific Srf KO (SRF-AKO, n = 9) male mice during HFD feeding. Total N = 20. (B) Heatmap showing the relative expression of cytoskeletal and collagen genes in eWAT from WT (n = 6) and SRF-AKO male mice (n = 5). Total N = 11. (C) Phalloidin staining of actin filaments in isolated adipocytes from WT and SRF-AKO male mice (scale bar: 20 μm). (D–E) Co-staining of isolated adipocytes with phalloidin (red) and PLIN1 (green) from both eWAT and iWAT of female mice. Scale bar: 100 μm, with quantification of phalloidin signal intensity. (F–G) Representative H&E-stained adipose tissue sections from (F) eWAT and (G) iWAT of WT and SRF-AKO male mice, with quantification of average adipocyte size. Scale bar: 100 μm. A two-tailed Student’s t-test was used for statistical analysis. **P < 0.01, ***P < 0.001, ****P < 0.0001.

3.5. SRF loss in adipocytes leads to defective glucose homeostasis during HFD-induced obesity

We next investigated the impact of SRF loss in adipocytes on systemic energy homeostasis. Although SRF-AKO mice showed no impairments in cold-induced thermogenesis, the possibility remained that SRF could influence diet-induced thermogenesis. To test this, we subjected SRF-AKO mice to a HFD for 3 months and assessed metabolic parameters via indirect calorimetry. SRF-AKO mice exhibited no significant alterations in systemic metabolic parameters, including oxygen consumption, carbon dioxide production, energy expenditure, respiratory exchange ratio, food intake, or locomotor activity (Fig. S3A–G). Consistently, there were no significant differences in thermogenic gene expression in BAT (Fig. S3H). These results indicate that adipocyte SRF does not influence whole-body energy balance and is dispensable for both cold- and diet-induced adipose thermogenesis.

To further explore the metabolic consequences of adipocyte SRF loss, especially considering the reduced adipocyte size and compromised cytoskeletal integrity, we examined its role in systemic glucose homeostasis. Glucose tolerance tests revealed a trend toward impaired glucose clearance in SRF-AKO mice following HFD feeding (Fig. 4A), and insulin tolerance tests demonstrated clear insulin resistance in these mice (Fig. 4B). Linear regression analysis of fasting insulin levels against body weight showed steeper slopes in SRF-AKO mice (Fig. 4C), indicating a trend toward increased insulin resistance relative to body weight. In addition, SRF-AKO mice exhibited significantly reduced eWAT and iWAT weights and enlarged liver size (Fig. 4D), accompanied by increased lipid accumulation in BAT and ectopic lipid deposition in the liver (Fig. 4E–F). Together, these data suggest that SRF-AKO mice develop partial lipodystrophy, leading to ectopic lipid redistribution and contributing to systemic insulin resistance and impaired glucose homeostasis.

Fig. 4.

Impaired systemic glucose homeostasis and partial lipodystrophy phenotype in SRF-AKO mice during obesity. (A–C) Defective glucose homeostasis in SRF-AKO (n = 9) compared to WT (n = 11) male mice during obesity (total N = 20), assessed by (A) glucose tolerance test (GTT, 1 g/kg body weight glucose), (B) insulin tolerance test (ITT, 1 U/kg body weight insulin), and (C) fasting insulin levels (ng/mL) relative to body weight. (D) Tissue weight (% of body weight) of eWAT, iWAT, and BAT, and liver in WT (n = 11) and SRF-AKO (n = 9) male mice. Total N = 20. (E–F) Representative H&E-stained sections of (E) BAT and (F) liver from WT and SRF-AKO male mice. Scale bar: 100 μm. A two-tailed Student’s t-test was used for statistical analysis. *P < 0.05, **P < 0.01.

3.6. SRF-deficient adipocytes display impaired structural integrity and increased fragility

Given the marked reduction in adipocyte size and fat mass in SRF-AKO mice, we hypothesized that abnormally increased lipid release might underlie the lipodystrophic phenotype. Lipolysis assays using adipose tissue explants from HFD-fed mice revealed enhanced lipolysis in eWAT under both basal and isoproterenol-stimulated conditions and increased stimulated lipolysis in iWAT (Fig. 5A). Because cytoskeletal integrity is essential for maintaining cellular structure and intracellular organelle organization, its disruption in adipocytes from SRF-AKO mice prompted ultrastructural analysis of adipose tissue using transmission electron microscopy. This revealed ruptured lipid droplet membranes and lipid leakage into the interstitial space (Fig. 5B), indicative of compromised structural integrity. Consistent with this, SRF-AKO mice showed significantly increased adipocyte cell death compared to WT controls (Fig. 5C), suggesting that structural breakdown contributes to cell death. To directly assess susceptibility to mechanical stress, we subjected adipose tissue explants to compression. While eWAT from WT mice remained largely intact, eWAT from SRF-AKO mice exhibited pronounced structural disruption. Similar, though less severe, damage was also observed in SRF-AKO iWAT (Fig. 5D), further supporting increased mechanical fragility in the absence of SRF. Taken together, these findings demonstrate that SRF is critical for maintaining structural integrity and ensuring cell survival under mechanical stress during obesity.

Fig. 5.

Compromised structural integrity and increased cellular fragility in SRF-deficient adipocytes. (A) Basal and isoproterenol (ISO)-stimulated lipolysis, measured by glycerol release from 4-h eWAT and iWAT explants from WT and SRF-AKO male mice (n = 3 per condition, total N = 12). (B) Representative transmission electron microscopy (TEM) images of eWAT from WT and SRF-AKO female mice, showing ruptured adipocyte membranes (indicated by black arrows). Scale bar: 5 μm. (C) Quantification of apoptotic cells in eWAT and iWAT from WT (eWAT, n = 3; iWAT, n = 5) and SRF-AKO (eWAT, n = 3; iWAT, n = 3) male mice by terminal deoxynucleotidyl transferase dUTP nick-end labeling (TUNEL) staining. (D) BODIPY staining of eWAT and iWAT from WT and SRF-AKO male mice following 1.5-h of compression. Scale bar: 100 μm. A two-tailed Student’s t-test was used for statistical analysis. *P < 0.05, **P < 0.01.

3.7. Impaired adipocyte expandability disrupts vascular integrity and cell-cell communication in adipose tissue

To determine the impact of impaired adipocyte expandability on adipose tissue homeostasis, we performed snRNA-seq on eWAT and iWAT from HFD-fed SRF-AKO and WT mice. From a total of 34,457 nuclei integrated across all conditions, we identified all major mouse adipose tissue cell types based on their respective marker expression, as previously described (Fig. 6A, B) [22]. Alterations in relative cell type proportions between SRF-AKO and WT mice were more pronounced in eWAT than in iWAT, reflecting the observed tissue structure disruptions in eWAT (Figs. 6C, S4A). Consequently, we focused our subsequent analysis on eWAT.

Fig. 6.

Impaired vascular integrity and altered cell-cell communication in adipose tissue driven by loss of SRF in adipocytes. (A) UMAP visualization of 34,457 single nuclei isolated from eWAT and iWAT of WT and SRF-AKO male mice (total N = 4), with annotated cell types. (B) Violin plots of cell type-specific marker gene expression across all identified cell types. (C) Relative proportions of each cell type in eWAT and iWAT from WT and SRF-AKO male mice. (D–E) Whole-mount staining of eWAT from WT and SRF-AKO mice with Hoechst (blue), BODIPY (green), and either F4/80 (red, D) or PECAM1 (red, E). Scale bar: 100 μm. (F) Circle plots from CellChat analysis showing altered cell-cell communication in eWAT of SRF-AKO male mouse compared to WT. Increases in interaction number (left) and strength (right) are shown in red; decreases are shown in blue.

Macrophage populations were markedly expanded in SRF-AKO mice compared to WT controls (Figs. 6C, S4A). Subclustering revealed 3 macrophage subtypes, including perivascular-like (PVM), non-perivascular-like (NPVM), and lipid-associated macrophages (LAM), as previously described [22,47], with comparable proportions in each depot in SRF-AKO and WT mice (Fig. S4B–D). These findings imply a general increase in total macrophages rather than selective subtype expansion. Whole-mount F4/80 staining confirmed enhanced macrophage infiltration and abundant crown-like structures containing large lipid-laden macrophages in SRF-AKO eWAT (Fig. 6D), whereas iWAT showed only modest, non-significant differences (Fig. S5A). Collectively, these data indicate that impaired adipocyte structural integrity in SRF-AKO mice promotes macrophage activation and inflammation in eWAT.

Another cell population showing notable changes in SRF-AKO mice was the vascular compartment, comprising endothelial cells (ECs), pericytes (PCs), and smooth muscle cells (SMCs), all of which were reduced (Figs. 6C, S4A). As we previously described [22], we identified 4 EC subtypes through subclustering, including venous (VenEC), arterial (ArtEC), capillary (CapEC) and lipid-associated (LipEC) ECs (Fig. S6E, F). CapECs were markedly reduced, while LipECs were increased in SRF-AKO mice (Fig. S4G), reflecting a shift in endothelial subtype composition along with an overall decline in total EC numbers. Whole-mount PECAM1 staining confirmed disrupted vascular architecture with fragmented, abnormally curved capillaries in SRF-AKO eWAT (Fig. 6E), while iWAT vasculature remained largely unaffected (Fig. S5B). These observations implicate that mechanical dysfunction of SRF-deficient adipocytes contributes to the impairment of overall vascular structure within adipose tissue.

To further elucidate alterations in intercellular communication that may underlie the macrophage accumulation and vasculature abnormalities observed in SRF-AKO mice, we employed CellChat analysis of ligand–receptor interactions among cell types in eWAT. Although overall intercellular communication remained robust in both WT and SRF-AKO eWAT (Fig. S6A, B), SRF-AKO adipocytes exhibited reduced interaction strength, particularly with vascular cells such as ECs and PCs/SMCs (Figs. 6F, S6C), in line with the structural defects observed in the vasculature. Conversely, interactions between adipocytes and mesothelial cells were elevated in SRF-AKO mice, along with increased interactions between immune cells and mesothelial cells (Figs. 6F, S6C). This pattern potentially reflects a shift toward a pro-inflammatory tissue state (Fig. 6E). Taken together, these findings suggest that impaired adipocyte expandability compromises vascular integrity and rewires intercellular signaling networks, thereby promoting inflammatory remodeling of adipose tissue.

4. Discussion

Hypertrophic expansion is a key process that allows adipocytes to store excess energy, yet it is often associated with metabolic dysfunction in obesity [6]. The molecular mechanisms that promote healthy and adaptive hypertrophy, however, remain poorly understood. In this study, we employed super-enhancer profiling to identify SRF as a key transcription factor regulating actin cytoskeletal remodeling, a central hallmark of adipocytes during obesity. Using adipocyte-specific SRF-deficient mice, we demonstrate that the actin cytoskeletal integrity is required for healthy adipocyte expansion under obesogenic conditions.

Previous studies have established SRF as a key regulator of actin cytoskeleton organization across various tissues, particularly in muscle and cardiac tissue, where it orchestrates gene expression programs governing cell shape, motility, and differentiation [48]. However, in adipocytes, SRF has only been studied in vitro, where it has been shown to play an anti-adipogenic role during differentiation [20] and act as a negative regulator of brown adipocyte differentiation and thermogenesis [46,49]. Mice with a global knockout of MKL1, a co-activator of SRF, exhibit enhanced adipose thermogenesis, particularly through browning of white adipose tissue, suggesting a potential role for SRF in thermogenic regulation [50]. However, the specific tissue and cellular environments mediating these effects have yet to be clearly identified. Our study, utilizing both pan-adipocyte-specific SRF knockout (Adiponectin-Cre) and thermogenic adipocyte-specific SRF knockout (Ucp1-Cre) models, provides strong evidence that SRF does not play a major role in adipose thermogenesis under acute or chronic cold exposure and during diet-induced thermogenesis. While we cannot entirely exclude the possibility that SRF contribute to alternative, non-sympathetic modes of adipose activation, these results suggest that the thermogenic role of SRF described in previous studies may primarily pertain to in vitro systems, non-adipose tissues, or adipocyte progenitors, rather than mature adipocytes. In contrast, we identify a distinct and critical role for SRF in mature white adipocytes during hypertrophic expansion in obesity, specifically in regulation of actin filament organization. Using both in vitro and adipocyte-specific in vivo models, we show that SRF drives the transcriptional activation of actin cytoskeletal genes, orchestrating cytoskeletal remodeling and enabling proper adipose tissue adaptation. Collectively, these findings establish SRF as a key mediator of adipose tissue remodeling in obesity.

The actin cytoskeleton is a highly dynamic structure that enables cells to sense and adapt to various mechanical stresses, such as stretching and compression [51]. In endothelial cells, for instance, cyclic stretch induces re-alignment of actin filaments perpendicular to the stretch direction, thereby strengthening barrier function and maintaining structural integrity [52]. Fibroblasts similarly remodel their actin networks under mechanical strain, forming stress fibers that enhance cellular stiffness and resilience [53]. These mechanoadaptive responses serve to reinforce cell architecture, dissipate mechanical stress and protect tissues from deformation or damage. In adipocytes undergoing substantial hypertrophic expansion during obesity, increased actin filament networks have been observed both in our previous work and by others [13,14,54]. However, the precise functional significance of this actin filament remodeling in adipocytes has remained unclear. The present study demonstrates that actin cytoskeletal remodeling in hypertrophic adipocytes functions as a protective adaptation to mechanical stress, improving cellular resilience and maintaining structural stability. Disruption of this process by SRF knockout impaired lipid droplet integrity, as visualized by electron microscopy, and led to increased lipid leakage, evidenced by elevated lipolysis in adipose tissue explants. Furthermore, SRF-deficient adipose tissue exhibited significant structural disruption when subjected to mechanical compression ex vivo. These defects limited adipocyte expandability, resulting in lipid spillover into BAT and liver. Because thermogenic function was preserved in SRF-AKO mice, this ectopic lipid accumulation is likely a secondary consequence of the restricted expansion capacity of iWAT and eWAT rather than impaired thermogenesis. This lipid redistribution contributes to defective glucose homeostasis. Together, these findings establish a critical role for the actin cytoskeleton in preserving adipocyte architecture and mechanical stability, thereby supporting secure lipid storage during obesity-associated tissue remodeling.

Not only compromising adipocyte structural integrity, defective actin cytoskeleton also broadly perturbed adipose tissue homeostasis. Loss of adipocyte SRF promoted local inflammation, as characterized by an expansion of macrophage populations. Adipose tissue vascularization was also notably disrupted in SRF-AKO mice, marked by reduced EC populations, increased LipECs, and abnormal vascular architecture. These findings suggest that proper actin cytoskeletal organization in adipocytes is essential not only for mechanical resilience but also for coordinating stromal cell function and vascular remodeling during tissue expansion, as evidenced by our CellChat analysis revealing altered intercellular communication within adipose tissue. Whether SRF mediates these processes via secreted factors such as growth factors and cytokines, or through direct adipocyte-endothelial interactions involving membrane-associated proteins, remains to be determined.

Although increased adipocyte size is generally linked to metabolic dysfunction in obesity [55], adipocytes display substantial functional heterogeneity. Notably, individuals with similarly hypertrophic adipocytes can exhibit divergent metabolic outcomes, with some developing insulin resistance while others retain insulin sensitivity [56]. This suggests that the capacity for adipocyte expansion, rather than absolute cell size, may play a more critical role in determining adipose tissue function [57]. Supporting this notion, collagen VI-deficient mice exhibit adipocyte hypertrophy without metabolic impairments, due to a permissive extracellular matrix (ECM) that facilitates unrestricted cellular expansion, thereby representing metabolically healthy obesity (MHO) [58]. In contrast, our study demonstrates that smaller adipocytes but with impaired expandability in SRF-AKO mice result in defective adipose tissue homeostasis, ectopic lipid accumulation and systemic insulin resistance, which are hallmarks of metabolically unhealthy obesity (MUO) [59]. These findings identify intracellular actin cytoskeletal remodeling, in coordination with ECM remodeling, as a critical factor governing adipocyte expandability. Whether mechanical expandability differs between MHO and MUO in humans remains an important open question for future studies. In conclusion, this study identifies SRF as a key regulator of cytoskeletal remodeling required for adaptive, healthy adipose tissue expansion and suggests that enhancing SRF-dependent actin cytoskeleton dynamics in adipocytes may offer a novel therapeutic strategy to improve metabolic health in obesity.

4.1. Limitations of the study

While our study provides strong evidence that SRF orchestrates actin cytoskeletal remodeling to enable adaptive adipocyte hypertrophy during obesity, several considerations warrant discussion. First, our super-enhancer profiling was performed in eWAT, and while we expect similar regulatory mechanisms in iWAT and other fat depots, additional profiling would further substantiate this generalization. Second, although SRF-AKO mice exhibited features of partial lipodystrophy [60], our analyses focused primarily on adipose tissue and liver; examining skeletal muscle, circulating metabolic parameters such as plasma lipids or leptin, and pancreatic β-cell function could provide a more comprehensive view of systemic metabolic consequences. Third, most experiments were performed in male mice, though similar phenotypes in females suggest a largely sex-independent role for SRF that merits further validation. Lastly, all studies were conducted in mice, and it remains to be determined whether these mechanisms are fully conserved in human adipocytes.

Appendix A. Supplementary data

Supplementary data to this article can be found online at https://doi.org/10.1016/j.metabol.2026.156548.

Data availability

The raw and processed data generated in this study are available at the Gene Expression Omnibus (GEO) repository under accession number GSE300990. The adipocyte-specific H3K27ac ChIP-seq datasets from epididymal white adipose tissue (eWAT) of chow- and HFD-fed mice, previously deposited under accession number GSE153120, were reanalyzed in this study.