Senescent immune cells accumulation promotes brown adipose tissue dysfunction during aging

BAT promotes non-shivering thermogenesis and supports metabolic health, but with age it exhibits increased adiposity, inflammatory immune infiltration, and reduced thermogenesis, contributing to obesity and age-related metabolic disorders. Immunosenescence is implicated in aging of solid organs, yet the contribution of senescent immune cells to BAT aging is unclear. The sympathetic nervous system (SNS) directly innervates BAT, releasing noradrenaline to drive lipolysis and adaptive thermogenesis. Emerging evidence highlights interactions between the nervous and immune systems in adipose tissue, but how sympathetic neurons and immune cells cooperate in BAT during aging—when sympathetic innervation is impaired—remains unresolved. S100A8, an alarmin from the S100 family released by neutrophils and monocytes, signals via TLR4 and AGER to promote cell activation and recruitment. Single-cell RNA-seq studies identified S100A8 as commonly increased in aged tissues including BAT, suggesting a potential role in BAT aging. This study addresses whether senescent S100A8+ immune cells accumulate in aged BAT, how they affect sympathetic innervation and thermogenesis, and whether targeting this axis can rejuvenate BAT function.

Prior work shows BAT thermogenic capacity declines with age, accompanied by inflammatory remodeling. Immunosenescence contributes to aging of solid organs including adipose tissue. SNS innervation is essential for BAT activation; neuro-immune crosstalk in adipose has been reported, including neuro-mesenchymal units influencing ILC2s and macrophage-derived signals modulating innervation and thermogenesis. Aging associates with impaired sympathetic innervation of BAT. S100A8/A9 are alarmins that bind TLR4 and AGER to modulate inflammation; scRNA-seq data in aged rats identified increased S100A8 across tissues, including BAT. Together, these studies motivate investigating S100A8+ senescent immune cells as drivers of BAT aging through effects on sympathetic innervation.

• Bioinformatics: Reanalyzed published scRNA-seq dataset of BAT from young (5 months) and aged (27 months) rats to identify immune populations and gene programs associated with aging. Performed clustering, subclustering, GO/KEGG enrichment, and marker analysis to identify S100a8-expressing immune subsets.
• Mouse models: Used male C57BL/6J mice at 2, 8, and 15 months; S100a8-Cre-EGFP reporter mice; Tlr4 knockout (Tlr4-/-); Tlr4fl/fl crossed with Adipoq-Cre for adipocyte-specific Tlr4 deletion (Tlr4ΔAdipo); NOD-SCID mice for human cell xenotransplantation. All mice were maintained under SPF conditions.
• Isolation and adoptive transfer of immune cells: From 15-month-old mouse bone marrow, sorted S100A8+ and S100A8− immune cells by flow cytometry; validated senescence markers (p16, p21, γH2AX) and SASP (Tnf, Ifng). Labeled with PKH26 for tracking. Transferred intravenously into 2-month-old mice; assessed tissue infiltration and metabolic effects.
• Bone marrow chimera: Isolated Lin−Sca-1+ c-Kit+ HSPCs from S100a8-Cre-EGFP mice; transplanted into lethally irradiated WT recipients; monitored GFP+ immune cell reconstitution and BAT infiltration over aging. Also generated Tlr4−/− bone marrow into WT chimeras to parse immune vs adipocyte TLR4 contributions.
• AAV interventions: Intra-BAT injections of AAV-Fabp4-ShRbm3 (RBM3 knockdown in adipocytes) in 2-month-old mice; AAV-Fabp4-RBM3 (RBM3 overexpression) and AAV-ShS100a8 (S100a8 knockdown) in 15-month-old mice. Scramble/AAV-NC controls used.
• Protein/compound treatments: Systemic recombinant S100A8 protein (2.5 μg per mouse, twice weekly for 4 weeks). TLR4 inhibitor TAK-242 in vitro. Paquinimod (1 mg/kg/day by oral gavage, three times per week) in 15-month-old mice for 4 weeks to 6 months under NCD; 12 weeks under HFD.
• Phenotyping: Indirect calorimetry (CLAMS) for VO2, energy expenditure, RER, activity at room temperature and cold (4 °C). Cold exposure tests at 6 °C to measure core body temperature. GTT and ITT protocols for metabolic assessment. Body weight, food intake, tissue weights, histology (H&E) for adipose and liver.
• Tissue and cellular assays: Flow cytometry of BAT SVF to quantify S100A8+ CD3+ T cells and CD11b+ myeloid cells; immunofluorescence for S100A8, TH, TUBB3, CD3, CD11b, CD31; immunoblotting for UCP1, PGC-1α, p16, p21, TH, phospho-TH, TUBB3, RBM3, NRP1, EPHA7; ELISA for serum and tissue S100A8, TNF, IL-6. SA-β-gal staining for tissue senescence.
• In vitro adipocyte studies: Differentiated C3H10T1/2 brown adipocytes treated with recombinant S100A8, S100A9, or LPS; RNA-seq to identify DEGs; assessed TLR4-p38/ERK signaling, RBM3 regulation, and effects on thermogenic vs neuron-related gene programs.
• RIP-seq and RNA-seq: RIP-seq using anti-RBM3 to identify binding targets; RNA-seq after siRbm3 knockdown to define RBM3-dependent DEGs; cross-analysis to identify RBM3-bound DEGs enriched in axon guidance and related pathways.
• mRNA stability and binding: Predicted RBM3-binding motifs in 3'UTRs; RIP-qPCR for RBM3 binding to Nrp1 and Epha7; actinomycin D chase assays to measure mRNA decay rates upon RBM3 knockdown.
• Neuron–adipocyte coculture: PC12 sympathetic-like neurons cocultured with RBM3-overexpressing brown adipocytes with or without siNrp1/siEpha7; neurite outgrowth assessed by TUBB3 staining.
• Human studies: Collected peripheral blood from healthy male young (16–32) and older (56–69) donors; quantified circulating S100A8+ immune cells and T cells; analyzed GTEx correlation between S100A8 and p21 (KRAS) expression. Isolated human S100A8+ or S100A8− immune cells and transferred into NOD-SCID mice; assessed BAT infiltration (human CD45), TH, thermogenic proteins, senescence markers, infrared thermography, and cold tolerance.
• Statistics: Student’s t test, one-way/two-way ANOVA with post hoc tests, ANCOVA for energy expenditure controlling for body weight, Pearson’s correlation; significance at P<0.05.

• scRNA-seq reanalysis of rat BAT identified increased T cells and neutrophils with age, enriched for inflammatory and adhesion pathways, with S100a8 markedly upregulated in both lineages. S100A8+ T cells increased ~150-fold in aged vs young rats; S100A8+ neutrophils and macrophages also increased.
• In mice, S100A8 protein increased in aged BAT SVF, with elevated frequencies of S100A8+ CD3+ T cells and S100A8+ CD11b+ myeloid cells by flow cytometry and immunostaining.
• Bone marrow origin and BAT tropism: Transferred PKH26-labeled BM-derived S100A8+ immune cells preferentially accumulated in BAT over iWAT, eWAT, and liver. In S100a8-Cre-EGFP HSPC bone marrow chimeras, GFP+ CD3+ T cells and CD11b+ myeloid cells accumulated in BAT over time (2→6→12 months).
• Functional impact: Adoptive transfer of S100A8+ BM immune cells to young mice reduced BAT UCP1 and PGC-1α protein/mRNA, increased p16/p21, decreased oxygen consumption and energy expenditure, lowered core body temperature during cold challenge, and increased lipid droplet size. BAT-specific knockdown of S100a8 in aged mice increased thermogenic gene expression and improved cold tolerance.
• Mechanism via adipocyte TLR4: Recombinant S100A8 decreased BAT thermogenic markers in WT but not Tlr4−/− mice. In chimeras, adipocyte-specific Tlr4 deletion (Tlr4ΔAdipo) abrogated S100A8 suppression of Ucp1/Ppargc1a, whereas Tlr4 deletion in hematopoietic cells did not, indicating S100A8 acts primarily on adipocyte TLR4.
• Neuroimmune adipose interface: S100A8+ immune cells localized adjacent to TH+ and TUBB3+ sympathetic fibers in aged BAT. TH intensity negatively correlated with S100A8 intensity (R = −0.785, P = 0.0001). Transfer of S100A8+ immune cells reduced TH, phospho-TH (Ser40), and TUBB3, indicating impaired sympathetic innervation.
• S100A8 suppresses RBM3 in adipocytes via TLR4–p38: RNA-seq of S100A8-treated adipocytes showed DEGs enriched in neuron projection and synaptic plasticity pathways. S100A8 reduced RBM3 in vivo and in vitro; TAK-242 (TLR4 inhibitor) and p38 inhibition restored RBM3, whereas ERK inhibition did not.
• RBM3 governs sympathetic innervation: BAT-specific Rbm3 knockdown (AAV-ShRbm3) decreased TH/p-TH/TUBB3, lowered UCP1, and impaired cold tolerance. RBM3 overexpression enhanced innervation and thermogenesis and rescued S100A8+ immune cell-induced deficits, restoring VO2 and cold tolerance and lowering p16/p21.
• RBM3 targets axon guidance genes: RIP-seq identified 4349 RBM3-binding peaks; cross with RBM3-knockdown RNA-seq yielded 909 RBM3-bound DEGs enriched in axon guidance, focal adhesion, endocytosis. RBM3 stabilizes mRNAs of Nrp1 and Epha7 by binding their 3'UTRs; RBM3 knockdown decreased their stability and protein levels. In coculture, siNrp1 or siEpha7 abrogated RBM3-driven neurite outgrowth.
• Human relevance: Older male donors had higher circulating S100A8+ immune cells and S100A8+ T cells; S100A8 expression correlated with p21 (KRAS) in human blood (R=0.47, p<0.0001). Transplantation of human S100A8+ immune cells into NOD-SCID mice led to BAT infiltration (human CD45+), reduced TH, decreased UCP1, increased P16/P21 and SA-β-gal, and reduced thermogenesis and core temperature during cold exposure.
• Therapeutic targeting: Paquinimod, an S100A8/A9–TLR4 interaction inhibitor, reduced S100A8+ T cell and myeloid infiltration in aged BAT, restored TH and UCP1/PGC-1α, lowered P16/P21, reduced body weight without altering food intake, improved fasting glucose under NCD, and under HFD attenuated weight gain, improved glucose tolerance and insulin sensitivity, reduced eWAT and liver mass, adipocyte hypertrophy, and hepatic steatosis.

The study identifies a bone marrow-derived, pro-inflammatory, senescent S100A8+ immune cell population (notably T cells and neutrophils) that accumulates in aged BAT and forms a neuroimmune adipose interface with sympathetic fibers and adipocytes. Local S100A8 secretion acts on adipocyte TLR4 to downregulate RBM3 via p38 signaling, leading to dysregulation of axon guidance genes (e.g., Nrp1, Epha7), reduced sympathetic innervation (TH/TUBB3), and diminished thermogenesis. Functional experiments demonstrate sufficiency (adoptive transfer) and necessity (BAT S100a8 knockdown; RBM3 overexpression) in modulating BAT aging phenotypes. The findings shift the paradigm from inflammatory cytokine-SASP-centric models to a mechanism where senescent immune-derived S100A8 primarily acts on adipocytes to perturb neuron–adipocyte crosstalk. Human data show increased circulating S100A8+ immune cells with age and xenotransplant experiments recapitulate BAT dysfunction, supporting translational relevance. Pharmacologic blockade of S100A8/TLR4 with paquinimod reverses BAT innervation defects and improves systemic metabolism, highlighting a therapeutic avenue for age-related metabolic decline.

This work demonstrates that senescent, bone marrow-derived S100A8+ immune cells accumulate in aged BAT, where their S100A8 secretion acts through adipocyte TLR4 to suppress RBM3, destabilize axon guidance mRNAs (Nrp1, Epha7), impair sympathetic innervation, and reduce thermogenesis. Enhancing adipocyte RBM3 or inhibiting S100A8–TLR4 interactions (paquinimod) restores innervation and BAT function and improves metabolic outcomes. These findings establish the S100A8–TLR4–RBM3 axis as a key driver of BAT aging and a promising target to treat age-associated metabolic disorders. Future directions include delineating upstream recruitment cues (e.g., CXCL1–CXCR2 axis) for S100A8+ immune cells to BAT, assessing sex-specific effects, testing long-term safety and efficacy of S100A8/TLR4 inhibition in diverse models, and translating these strategies to human interventions.

• Sex limitation: All experiments were conducted in male mice; female responses may differ.
• Model constraints: Heavy reliance on murine models and immunodeficient mice for human cell transfer; human BAT physiology may not fully align.
• Partial mechanistic scope: While adipocyte TLR4 is implicated, potential contributions from other receptors (e.g., AGER) or cell types were not exhaustively tested.
• Local vs systemic S100A8: Serum S100A8 did not change markedly; local concentration and kinetics within BAT niches were inferred but not directly quantified at single-cell resolution.
• In vitro findings: S100A8 did not directly suppress thermogenic gene expression in isolated adipocytes, suggesting context dependence mediated by tissue architecture or innervation not fully recapitulated in vitro.
• Limited human data: Cross-sectional blood analysis and short-term xenotransplantation provide associative and sufficiency evidence but not longitudinal causality in humans.
• Sample sizes and duration: Some experiments have modest n and limited duration; long-term effects and safety of paquinimod in aged animals require further study.