Integrating single-cell multi-omics to understand immune cell dynamics in systemic lupus erythematosus

Abstract

Systemic lupus erythematosus (SLE) is a complex autoimmune disease characterised by heterogeneous clinical features and dysregulated immune responses. This thesis aims to understand the cellular and genetic mechanisms driving SLE pathology by leveraging two single-cell multi-omic datasets. The first part of the thesis focuses on a longitudinal cohort of SLE patients treated with rituximab, a CD20+ B cell-depleting therapy, to study immune cell dynamics in relation to treatment response. The second part investigates the genetic regulation of immune cell gene expression in a large cross-sectional cohort of SLE patients with paired whole genome sequencing (WGS) and single-cell RNA-sequencing (scRNA-seq) data. In the first part, I developed a robust single-cell multi-omics analysis pipeline to ensure quality control and integration of gene expression, surface protein, and immune receptor data. Longitudinal analysis from nine lupus patients revealed that clinical response to rituximab was marked by selective depletion and incomplete recovery of age-associated and switched memory B cells. Repopulated naïve B cells in rituximab responders showed decreased NF-kappaB pathway activity and lower BAFFR expression, suggesting a reset of inflammatory programming. Furthermore, non-B cell changes included increased activation and cytotoxicity-related gene expression in CD4 central memory and double-negative T cells. These findings provide insight into treatment-induced changes in immune composition and function that may underlie variable patient outcomes. In the second part, I generated and analysed paired single-cell multi-omic data and WGS data from 281 SLE patients. sc-eQTL mapping was performed across 15 immune cell types, identifying 3,803 unique eGenes, with substantial variation in genetic regulation across cell types. Shared and cell type-specific effects were observed, with functionally related lineages exhibiting greater eQTL overlap. Additionally, conditionally independent eQTL signals were identified, highlighting regulatory variants not captured in primary analyses. Colocalisation between these eQTLs and SLE GWAS loci revealed 144 sc-eQTLs with evidence of shared genetic architecture, including both broadly shared eQTLs (e.g., RNASET2, ORMDL3) and highly cell type-specific ones (e.g., ARL17B in classical monocytes). These results emphasise the utility of high-resolution eQTL mapping in immune cells to dissect the genetic underpinnings of SLE. Collectively, these findings underscore the power of single-cell multi-omics to resolve the complexity of SLE and offer a blueprint for dissecting immune cell regulation in disease. This work establishes a scalable framework for integrating single-cell and genomic data in complex diseases and contributes valuable resources and methodology to the field of immunogenomics.