Single-cell RNA-sequencing of PBMCs highlights the central role of anti-SSA antibodies and its association with Interferon-stimulated genes in the expression of Sjögren's disease

Abstract

Introduction

We aimed at exploring transcriptomes of PBMCs from SICCA participants using single cell RNA sequencing (scRNA-Seq), to better understand the heterogeneity of Sjögren's disease (SjD).

Matériels et méthodes

Participants randomly selected from eligible SICCA-cohort subsets were divided into three groups: SSA+ primary SjD cases (SSA+SjD+), SSA- primary SjD cases (SSA-SjD+), and symptomatic controls (SjD-). We performed scRNA-seq in addition to surface protein (CITE-seq) using 10x Genomics’ Chromium Single Cell 5’ V1.1 chemistry on multiplexed samples and 10x Cell Ranger pipeline. Analytical pipeline (Seurat, R software) included Freemuxlet for demultiplexing, DoubletFinder for exclusion of intra-individual doublets, Harmony for batch correction, and EdgeR for gene differential expression (DE) analysis adjusting for age, sex, and genetic ancestry, and Monocle inference for trajectory analysis. We used Single-Cell Interpretable Tensor Decomposition (scITD) to examine gene expression patterns across multiple cell types and correlation to phenotypic characteristics.

Résultats

After filtering steps, 333 participants remained, including 101 SSA+SjD+, 47 SSA-SjD+, and 185 SjD-. Louvain clustering revealed 28 distinct cell clusters across all PBMCs. At a large scale, SSA+SjD+ status correlated with increase of transitional B cells, central memory CD4+ T cells, non-classical monocytes, and a reduced proportion of effector memory CD4+ T cells. Using principal component (PC) analysis on cell type abundance, SSA+SjD+ members differed from SSA-SjD+ and SjD- groups in the PC1/PC2 space. DE gene analysis showed an upregulation of type I interferon-stimulated genes (ISGs) in SSA+SjD+ participants. We observed very few DE genes for SSA-SjD+ versus SjD- participants.

Upon re-clustering of B cells, monocytes-dendritic (MD) cells, and T cells, we respectively identified 23, 27, and 38 different cell types. Within B cells, Dirichlet regression for differential abundance (DA) revealed that SSA positivity among SjD+ participants was linked to higher transitional B cells, while SSA-SjD+ status was associated with no specific cell cluster compared to SjD- group. Trajectory in B cells suggested a reinforced transition from transitional B cells to marginal zone B cells in SjD+ participants, particularly with anti-SSA autoantibodies.

In re-clustered MD cells, DA analysis identified an association between the presence of anti-SSA antibodies and both classical and non-classical monocytes expressing a high level of ISGs. MD trajectories highlighted increased differentiation into intermediate and non-classical monocytes in SSA+SjD+ compared to both SSA-SjD+ and control groups.

In scITD analysis, one factor strongly correlated with SjD status and the presence of anti-SSA antibodies was identified among the main factors of variability in gene expression. This factor was linked to ISGs, with many genes having correlated expression across B cells, T cells, and monocytes. This correlation pattern was most striking within the SSA+SjD+ participants in all 3 groups, and least present in SSA-SjD+ cells.

Conclusion

Our investigations confirmed the pivotal role of anti-SSA antibodies in SjD systemic expression and its strong correlation with ISGs. SSA-SjD+ participants did not show significant differences from symptomatic controls at the PBMC level. These findings add new pieces to the SjD puzzle, supporting the idea of two different pathogenesis between SjD with and without anti-SSA antibodies.