Uncovering the Pathology of Rheumatoid Arthritis With Single Cell Immunoprofiling

Abstract

Rheumatoid arthritis (RA) is a chronic multi-systemic autoimmune disorder affecting nearly 25 million people worldwide, yet its underlying causes remain unclear. Genetic studies of patients with RA have highlighted the role of dysfunction in the adaptive immune system, particularly among CD4+ T cell populations. While defining the precise CD4+ T cell subsets that are dysregulated in RA patients is critical to deciphering pathogenesis, much of the work in the field has relied on animal models or derives from bulk analyses of immune cells, which can lead to overlooking rare or transitional cell types due to the heterogenous nature of immune populations. The recent introduction of high dimensional single-cell analyses have improved the ability to resolve complex mixtures of cells; however, identifying disease-associated cell types or cell states in patient samples remains challenging due to technical and inter-individual variation. In particular, case-control analysis of disease using single cell data requires a quantitative approach to determining which cells provide the most information (and which cells are uninformative) while accounting for confounding effects from batch or technical variation; properly grouping those cells into biologically relevant populations, and then determining whether the abundance of these populations is statistically different between cases and controls. Mixed effects modeling of Associations of Single Cells (MASC) is a novel reverse single cell association strategy to determine if a cellular subpopulation is associated with case-control status while controlling for technical confounders and biological covariates. This method revealed important changes in the abundance of disease-associated immune and stromal populations – specifically an expansion of cytotoxic CD4+ T cells and HLA+ sublining fibroblasts in RA. Compared to peripheral blood, synovial fluid and synovial tissue samples from RA patients were significantly enriched for both of these populations, indicating that these cell types are present in high abundance at the specific locus of RA pathogenesis. The methods developed for the analysis of single cell data are broadly applicable, support performing association testing with high-dimensional single cell data, and can help identify other cellular populations that are critical to rheumatic disease pathogenesis.