Transcriptomic analysis of immune cells in a multi-ethnic cohort of systemic lupus erythematosus patients identifies ethnicity- and disease-specific expression signatures

Abstract

Systemic lupus erythematosus (SLE) is an autoimmune disease in which outcomes vary among different racial groups. We leverage cell-sorted RNA-seq data (CD14+ monocytes, B cells, CD4+ T cells, and NK cells) from 120 SLE patients (63 Asian and 57 White individuals) and apply a four-tier approach including unsupervised clustering, differential expression analyses, gene co-expression analyses, and machine learning to identify SLE subgroups within this multiethnic cohort. K-means clustering on each cell-type resulted in three clusters for CD4 and CD14, and two for B and NK cells. To understand the identified clusters, correlation analysis revealed significant positive associations between the clusters and clinical parameters including disease activity as well as ethnicity. We then explored differentially expressed genes between Asian and White groups for each cell-type. The shared differentially expressed genes across cells were involved in SLE or other autoimmune-related pathways. Co-expression analysis identified similarly regulated genes across samples and grouped these genes into modules. Finally, random forest classification of disease activity in the White and Asian cohorts showed the best classification in CD4+ T cells in White individuals. The results from these analyses will help stratify patients based on their gene expression signatures to enable SLE precision medicine.

Multi-context genetic modeling of transcriptional regulation resolves novel disease loci

“…10 . We also examined, for the first time in a large-scale TWAS context, a single-cell RNAseq data set from the California Lupus Epidemiology Study (CLUES) 19 , 20 . The CLUES data set contained 9592 genes measured in 8 cell types in peripheral blood from 90 individuals.…”

“…We ran FUSION-TWAS using the default recommended settings, with reference data from the 1000 genomes project 81 . TWAS weights were trained on the GTEx v7 dataset 2 as well as the CLUES 20 single-cell RNAseq dataset of PBMCs. For a given gene-context-trio, we ran up to 5 TWAS—1) context-by-context, 2) UTMOST, 3) CONTENT(Shared), 4) CONTENT(Specific), and 5) CONTENT(Full).…”

“…We evaluate the performance of CONTENT over simulated data sets, GTEx 2,11,13 , and a single-cell RNA-Seq data set [19][20][21] . We show in simulations that CONTENT captures a greater proportion of the heritable component of expression than previous methods (at minimum over 22% more), and that CONTENT successfully distinguishes the specific and shared components of genetic variability on expression.…”

Research perspectives—Pipelines to human tendon transcriptomics

“…• Body mass index (BMI): if limited medical history is available, restricting donor inclusion criteria to "normal" BMI can decrease the Male, female Other demographic factors should be considered and included in a balanced dataset, particularly race and ethnicity, to account for ethnicity-specific expression signatures. 13 If this and other demographic information is available but not an exclusion factor, we recommend reporting this information in publications and repositories.…”

Augmented glucose dependency of autoreactive B cells provides a treatment target for lupus

“…The B-cell, plasma cell, T-cell, activated T-cell, and myeloid-cell signatures were derived from Mouse CellScan, a tool for the identification of cellular origin of mouse gene-expression datasets. For human data, gene expression analysis was performed on publicly available transcriptomic data (GEO accession: GSE164457) 60 from participants recruited from the California Lupus Epidemiology Study. Briefly, PBMCs were isolated from 120 lupus patients, and were sorted into populations of CD19 + B cells and CD4 + T cells for bulk RNA-seq.…”