Tracking clonal dynamics of CD8 T cells and immune dysregulation in progression of systemic lupus erythematosus with nephritis

Abstract

The fluctuating nature of disease activity in systemic lupus erythematosus (SLE), alternating between flares and remissions, poses substantial challenges for its effective management. The use of current biomarkers for monitoring SLE is limited in clinical settings owing to insufficient comprehension of the complex immune involvement underlying the disease course. Here, therefore, we profiled peripheral blood mononuclear cells at both stable and exacerbation states (total of n = 19) from six patients with SLE and 32 healthy donors using integrated single-cell RNA and T cell receptor (TCR) sequencing. To validate our findings, we analyzed two independent external datasets: bulk RNA sequencing and TCR data from 79 controls and 62 patients with SLE and single-cell RNA sequencing data from 99 healthy controls and 162 patients with SLE. Our analysis revealed cell type-specific activation of interferon-related genes in SLE grouped into four clusters, with elevated activity in disease-associated immune cells. Among these, atypical B cells associated with autoantibody production exhibited distinct differentiation patterns compared with conventional memory B cells, driven by heightened interferon signaling in SLE. Notably, clonal expansion of effector CD8 T cells emerged as a key driver of disease exacerbation, as indicated by reduced TCR diversity. Specific CD8 T cell clonotypes expanded during flare states, transitioning to effector phenotypes that exhibited heightened cytotoxicity and amplified interferon signaling, strongly correlating with tissue damage and flare severity. Our findings establish a critical link between interferon-driven mechanisms and cytotoxic T cell dysfunction in SLE flares, offering potential targets for therapeutic intervention and predictive biomarkers.

Fig. 1. Characterization of the immune landscape in SLE.

a A schematic of the single-cell transcriptome with TCR sequence profiling of immune cells in PBMCs from healthy controls (CTL) and patients with SLE. b UMAP visualization of immune cell types. c Scaled average expression of marker genes across ten immune cell types, with the dot size indicating the proportion of expressing cells and the color gradient indicating expression levels. d A bar plot showing the cell type proportions by disease state (left) and the total number of immune cells (right). The lines on the bar indicate the interquartile range (IQR), spanning from the 25th percentile (Q1) to the 75th percentile (Q3). Statistically significant increases are marked with orange stars for SLE and green for CTL. e A heat map of 100 IFN-related genes across different cell types across three different studies. log₂ fold changes are shown, with genes significantly overexpressed in SLE marked in red (right). f,g A comparison of the IFN C2 score in CD8 T cells (f) and the IFN C3 score in cM (g) across controls and patients with SLE. The box plots show the distribution of IFN scores from three different studies (left). The box represents the IQR, spanning from Q1 to Q3, with the line inside the box indicating the median. The whiskers extend to the smallest and largest values within 1.5 times the IQR. UMAP visualization of IFN scores across immune cells (right). The color gradient indicates low to high scores. **P < 0.01, ***P < 0.001 and ****P < 0.0001.

Fig. 2. Novel subtype of atypical B cell with elevated IFN pathway.

a A UMAP visualization of B cell subtypes: ABC, memory B cells (B mem), naive B cells (B naive), plasma cells (PC) and plasmablasts (PB). b Scaled average expression of marker genes across B cell subtypes, with the dot size indicating the proportion of expressing cells and the color gradient indicating the expression levels. c The proportion of B cell subtypes by disease state. The lines on the bar indicate the IQR, spanning from Q1 to Q3. Statistically significant increases are marked with orange stars for SLE and green for CTL. d Violin plots showing the distribution of IFN C1 cluster scores in B cell subtypes for controls (CTL) and patients with SLE in two studies: Perez et al. (left) and our study (right). The lines within the violin indicate the IQR, spanning from Q1 to Q3. e The ratio of ABC to B mem within total B cells, categorized by disease. The box represents the IQR, spanning from Q1 to Q3, with the line inside the box indicating the median. The whiskers extend to the smallest and largest values within 1.5 times the IQR. f A UMAP plot illustrating the inferred differentiation trajectory of non-plasma B cells, with color-coding from early (blue) to late (red) stages (left). UMAP plots for subtypes and TBX21 expression levels (right). The color gradient indicates the expression level. g The regulon specificity score (RSS) of TFs in ABC. TFs are sorted by RSS rank and ABC-specific TFs, appearing exclusively in the top 30 for ABCs, are highlighted in red. h TF activity changes across pseudotime for B cell subtypes in B naive to ABC (left) and B naive to B mem (right). The color gradient represents the progression through pseudotime, with the lines indicating the trend of TF activity. i Violin plots of STAT1 expression and TF activity in non-plasma B cell subtypes. Lines within the violin indicate the IQR, spanning from Q1 to Q3. ***P < 0.001 and ****P < 0.0001.

Fig. 3. Distinct T cell activation and TCR diversity in SLE.

a A UMAP visualization for cell composition by αβT cell subtypes: CD4⁺ naive (CD4 naive), central memory (CD4 Tcm), effector memory (CD4 Tem), regulatory (CD4 Treg), cytotoxic (CD4 CTL), cycling (CD4 cycling), CD8⁺ naive (CD8 naive), GZMK⁺ memory (CD8 GZMK), effector memory (CD8 Tem) and cycling (CD8 cycling) T cells. b The scaled average expression of marker genes across ten αβT cell subtypes, with dot size indicating the expressing cell proportion and the color gradient showing expression levels. c The proportion of αβT cell subtypes by disease state. The lines on the bar indicate the IQR, spanning from Q1 to Q3. Statistically significant increases are marked with orange stars for SLE and green for CTL. d Box plots comparing TCR signaling and T cell cytotoxicity in CD8 Tem cells across controls (CTL) and patients with SLE in three studies. The box represents the IQR, spanning from Q1 to Q3, with the line inside the box indicating the median. The whiskers extend to the smallest and largest values within 1.5 times the IQR. e A UMAP visualization delineates TCR clonotype distribution across αβT cells. Cells are colored based on clonotype expansion status. f Box plots showing TCR diversity (Shannon index) in CD8 T cell subtypes across controls and patients with SLE in two studies: Ota et al. (left) and our study (right). The box represents the IQR, spanning from Q1 to Q3, with the line inside the box indicating the median. The whiskers extend to the smallest and largest values within 1.5 times the IQR. g Heat maps of MHC-I and Galectin signaling strength from myeloid cells to CD8 T cell subtypes. The color gradient represents the relative strength of signaling. h Ligand–receptor communication probability of MHC-I (cM to CD8 T) (left) and Galectin signaling (pDC to CD8 T) (right). The color gradient represents communication probability and the empty dots indicate nonsignificant interactions. **P < 0.01, ***P < 0.001 and ****P < 0.0001.

Fig. 4. Expansion of clonally restricted cytotoxic CD8 T cell in SLE flare state.

a A schematic representation of CD8 T cell profiling in SLE flare using scRNA-seq and scTCR-seq. The timeline illustrates longitudinal tracking of BF and FL states in patients. b A UMAP visualization for cellular composition of CD8 T cells during flares, categorized by cell types (left) and disease status (right). c The correlation between the proportion of CD8 Tem cells among CD8 T cells measured by scRNA-seq and flow cytometry across BF and FL samples (left). A box plot showing the fold change of CD8 Tem cell proportion (FL relative to BF) in scRNA-seq data (right). Each dot represents an individual patient. The box represents the IQR, spanning from Q1 to Q3, with the line inside the box indicating the median. The whiskers extend to the smallest and largest values within 1.5 times the IQR. d Plots showing the IFN C2 cluster score in CD8 Tem cells during different states of disease activity. A violin plot for our dataset comparing BF and FL states (left). Lines within the violin indicate the IQR, spanning from Q1 to Q3.A box plot from Ota et al. across different SLEDAI states (right). The box represents the IQR, spanning from Q1 to Q3, with the line inside the box indicating the median. The whiskers extend to the smallest and largest values within 1.5 times the IQR. HDA, high disease activity; LDA, low disease activity; MDA, moderate disease activity. e TCR diversity and clonotype analysis in CD8 T cells by flare status. The plots show TCR diversity (left), the number of unique clonotypes (middle) and the proportion of the top ten clonotypes in CD8 T cells (right), comparing BF and FL states. The lines connecting paired samples from the same patient indicate changes between BF and FL time points. The box represents the IQR, spanning from Q1 to Q3, with the line inside the box indicating the median. The whiskers extend to the smallest and largest values within 1.5 times the IQR. f A UMAP visualization showing the density of expanded TCR clonotypes in CD8 T cells during flares. g An alluvial plot illustrating the distribution of expanded clonotypes across CD8 T cell subtypes in BF and FL states. h Bar plots showing the proportion of TCR clonotype in CD8 T cell subtypes of each pair of BF and FL time points across patients, with clonotypes categorized by color. i Violin plots comparing T cell cytotoxicity (left) and IFN C2 cluster scores (right) between expanded and non-expanded clonotypes in BF and FL states. The lines within the violin indicate the IQR, spanning from Q1 to Q3. j Heat maps representing the similarity index between samples from different time points in CD8 naive (left) and CD8 Tem cells (right), with patient numbers indicating sample origin. The color gradient represents the similarity index. ***P < 0.001 and ****P < 0.0001.