IFNγ-CXCL9-CXCR3 axis links T cell recruitment to skeletal muscle remodelling in human physical frailty
Introduction: Physical frailty is increasingly prevalent in our ageing society, and results in higher morbidity and healthcare costs. Yet, targeted therapeutics are lacking because underlying mechanisms are poorly defined. Human cohort studies have linked inflammation with frailty, but do not define the mechanisms by which the ageing immune system affects skeletal muscle and physical function. Defining this immune-muscle interaction is therefore crucial for identifying therapeutics for human physical frailty. Methods: We first studied two independent human ageing cohorts (discovery n=22; validation n=33). Physical frailty was quantified with age-adjusted timed-up-and-go (TUG) residuals. Peripheral blood was analysed for plasma OLINK proteomics, high-dimensional spectral immunophenotyping and in a subset, single-cell RNA-sequencing of PBMCs with paired TCR profiling. We then examined an independent human skeletal-muscle cohort (n=340) by bulk RNA-sequencing to derive our proposed immune-muscle axis. Finally, we interrogated individual steps of this axis in primary human in vitro models using IFN{gamma}-conditioned myotube-fibroblast co-cultures, PBMC transwell migration assays with CXCL9 neutralisation (anti-CXCL9) and receptor CXCR3 antagonism (AMG487). Macrophage co-cultured with CD8 were sequentially co-cultured with myotube/fibroblast co-culture, supernatant was then analysed for fibrosis mediators. We used fixed-effect meta-analysis, multivariable models and sequential mediation for analyses. Results: Physical frailty, by single-cell analyses, was associated with a compartmentalised immune-remodelling state involving inflammatory, tissue-remodelling and trafficking programmes in myeloid and T-cells. This finding was validated with plasma proteomics of the same human cohort, where physical frailty was significantly positively associated with an interferon-gamma (IFN{gamma})/CXCR3 (IL-18R1, IFN{gamma}, CXCL9) module, within which plasma CXCL9 was the strongest correlate. CD8-TEMRA and {gamma}{delta} T-cells were most responsive to CXCL9 in physical frailty. This expanded CD8-TEMRA compartment was polyclonal, consistent with bystander IFN{gamma} production rather than a shared antigen-specific response. In human muscle, CXCL9 was significantly positively associated with T-cell module, and through this, fibrosis. Conversely, fibrosis was associated with macrophages. Sequential mediation was consistent with a CXCL9-T-cell-macrophage-fibrosis pathway. Finally, in in-vitro human primary cell experiments, IFN{gamma}-conditioned myotube-fibroblast cultures produced CXCL9 and attracted CD8-TEMRA subsets. Anti-CXCL9 and CXCR3 antagonism significantly reduced this migration, indicating that recruitment of CD8-TEMRA cells was dependent on the CXCL9-CXCR3 axis. Incubating with activated CD8 cells shifted macrophages towards mixed CD163+CD86+ and pro-inflammatory CD163-CD86+ phenotypes. These CD8-co-cultured macrophages in myotube/fibroblast co-culture were associated with increased fibrosis markers transforming growth factor beta-1 and plasminogen activator inhibitor-1. Conclusion: We identified a CD8-TEMRA T-cell migration to human muscle dependent on IFN{gamma}-CXCL9-CXCR3 axis that is associated with human TUG-defined physical frailty which presents potential therapeutic avenues.