Figures and data
In vitro activation and phenotyping of CD4+ T cells under varying costimulation conditions.
A) Overview of the experiment. B) Activation rates of sorted naive and memory CD4+ T cells, measured by expression of CD69 determined by flow cytometry, for different costimulation conditions. mIgG2a = IgG2a isotype control.
Global changes in gene expression under different costimulation conditions.
A) Principal component analysis of RNA-seq samples, coloured by stimulation condition and shaded by memory/naive subset. B) Count of differentially expressed genes between costimulation conditions and control conditions, broken down by UP- vs DOWN-regulation, naive/memory subset and control condition (shown in gray boxes above the plot, costimulation conditions are compared to aCD3-stimulated-only controls, and aCD3 is compared to mIgG2a isotype controls). C) Count of differentially expressed genes between alternative costimulation conditions (aCD6, aCD27 and aICOS) and the aCD28-costimulated condition, broken down by UP- vs DOWN-regulation and naive/memory subset.
Number of costimulation-biased genes in memory and naive CD4+ T cells under different alternative costimulation conditions compared to CD28 costimulation,
estimated using two different methods (LM-based and shape-based).
Pathway and transcription factor enrichment analysis.
A) Hallmark and B) KEGG terms significantly enriched (FDR < 0.01) in costimulation-biased genes in at least one alternative costimulation condition NES=Normalized Effect Size, padj=Benjamini-Hochberg adjusted q-value C) Enrichment of transcription factor motifs in the CD6 and CD28 costimulation conditions compared to CD3-alone. Yellow dots indicate significant costimulation motifs, defined as motifs which are significantly enriched (FDR < 0.05) both in peaks that are upregulated in the CD6 condition compared to all peaks and compared to peaks that are upregulated in either the CD6 or CD28 conditions.
Computational follow-up of key pathways.
A) Estimated proliferation based on conserved proliferation gene signature for each costimulation condition and cell type. B) Differential activity statistics for inferred metabolic rates after costimulation with CD6 or CD28. Green line is the estimated fit, OxPhos reactions highlighted in blue. C,D) Induction of costimulation molecule transcripts in naive (C) and memory (D) CD4+ T cells under different costimulation conditions. Text in the boxes are in the form “log fold change (p-value).”
The impact of alternative costimulation on risk variants for inflammatory bowel disease.
A) Enrichment of heritability in open chromatin regions upregulated in different costimulation conditions (coefficient with 95% CIs). B) Enrichment of heritability in regions specifically upregulated in CD28 or in one or more alternative costimulation conditions (CD28-specific and alternative-specific peaks), in both CD28 and at least one alternative costimulation condition (shared peaks) or not upregulated in activated T cells in any condition (neither). C) The 8q24 IBD GWAS locus, causality posteriors for candidate causal variants [1], and estimated log-fold changes (with 95% CIs) in the CD28 and CD6 conditions as compared to aCD3-only control. Dashed line shows the location of the highest posterior variant. D) Enrichment of IBD associations colocalizing with gene expression in an eQTL study of CD28-costimulated CD4+ T cells [4] among genes upregulated during activation (in any costimulation condition) or specifically upregulated by alternative costimulation and not by CD28.
Costimulation-specific inflammatory bowel disease risk genes
High-confidence IBD risk genes that were significantly up- or down-regulated only in alternative costimulation (and not in CD28-costimulated cells), or detected with opposite direction of effect between CD28 and at least one other costimulation condition. ↑/↓ indicate that this gene was significantly upregulated/downregulated in this costimulation compared to the aCD3-alone condition.
Experimental validation of key findings.
A) Induction of costimulation receptors on the cell by CD28 and CD6 costimulation, measured by the ratio of mean fluorescence intensity (MFI) in costimulated cells to the MFI in the CD3-alone condition. B) Relative change in RNA levels (measured by RNA-seq) and of protein concentration in supernatant (measured by a multiplex immunoassay) between CD6 and CD28 costimulated cells. C) Quantification of intracellular lysosome abundance and lysosome secretion under different costimulation conditions, measured using the MFI of intracellular and extracellular LAMP-1. D) Quantification of autophagic flux in different costimulation conditions.
Costimulation-biased genes by the shape method.
Gene names are displayed for the top 5 and bottom 5 transcripts ranked by the logFC between the non-canonical costim and CD28. Dashed line is the slope 1 diagonal. The non-canonical significant-bias transcripts are colored in red.
Costimulation-biased genes by the linear model (LM) method.
The top 5 and bottom 5 transcripts by Z score are labelled. Dashed line is the slope 1 diagonal, while green line is the linear model fit. Significant transcripts are colored in red, and sized by their absolute Z score.
Global changes in chromatin accessibility under different costimulation conditions.
A) Principal component analysis of ATAC-seq samples, coloured by stimulation condition and shaded by memory/naive subset. B) Count of differentially accessible regions (DAR) between costimulation conditions and control conditions, broken down by UP- vs DOWN-regulation, naive/memory subset and control condition (shown in gray boxes above the plot, costimulation conditions are compared to aCD3-stimulated-only controls, and aCD3 is compared to mIgG2a isotype controls). C) Count of differentially accessible regions between alternative costimulation conditions (aCD6, aCD27 and aICOS) and the aCD28-costimulated condition, broken down by UP- vs DOWN-regulation and naive/memory subset. D) Number of costimulated-biased genes in memory and naive CD4+ T cells under different alternative costimulation conditions compared to CD28 costimulation, estimated using two different methods (LM-based and shape-based).
KEGG plot of lysosome genes in naive CD6 v. CD28 costimulation.
Positive Z (red) are CD6-upregulated, negative (green) are CD28-upregulated. Figure produced using the R package pathview.
A highlighting of differences in cytokine levels between costimulations.
Filled dots represent genes specific to the non-canonical costim based on the shape-method (rule based). The list of cytokines (used for highlighting purposes) was obtained from literature (Santoso et al. 2020).
Gating strategy and example expression histograms for measuring LAMP1 expression.
Gating strategy and example expression histograms for measuring autophagic flux expression.
