Transcriptome-wide Mendelian randomization during CD4+ T cell activation reveals immune-related drug targets for cardiometabolic diseases

Abstract

Immunity has shown potentials in informing drug development for cardiometabolic diseases, such as type 2 diabetes (T2D) and coronary artery disease (CAD). Here, we performed a transcriptome-wide Mendelian randomization (MR) study to estimate the putative causal effects of 11,021 gene expression profiles during CD4⁺ T cells activation on the development of T2D and CAD. Robust MR and colocalization evidence was observed for 162 genes altering T2D risk and 80 genes altering CAD risk, with 12% and 16% respectively demonstrating CD4⁺ T cell specificity. We observed temporal causal patterns during T cell activation in 69 gene-T2D pairs and 34 gene-CAD pairs. These genes were eight times more likely to show robust genetic evidence. We further identified 25 genes that were targets for drugs under clinical investigation, including LIPA and GCK. This study provides evidence to support immune-to-metabolic disease connections, and prioritises immune-mediated drug targets for cardiometabolic diseases.

Fig. 1. Study design of the dynamic and non-dynamic eQTL MR analysis.

The study included instrument selection and validation, outcome selection, dynamic and non-dynamic eQTL MR analysis, cell type-specific analysis and activation time-specific analysis.

Fig. 2. Presentation of MR results of causal effects of dynamic CD4⁺ T cell eQTLs on T2D and CAD and top target genes.

a Illustrating the number of all the MR estimates, MR estimate with Benjamini–Hochberg FDR corrected P value < 0.05 (two-sided) and with evidence of colocalization. The bar chart shows the number of MR signals with and without colocalization evidence according to the disease. b Validation in the results of the eight sensitivity analysis methods compared to the Wald ratio results. c A Manhattan plot illustrating the associations of genetically influenced gene expression derived from CD4⁺ T cells on T2D and CAD, respectively. The Y-axis shows the minus log 10 P values (two-sided) of the MR estimates. The labeled genes are the most significant genes with strong MR (FDR < 0.05) and colocalization evidence on each chromosome among the 162 genes affecting T2D and 80 genes affecting CAD. T2D type 2 diabetes, CAD coronary artery disease.

Fig. 3. Differential expression analysis of 162 genes causally affecting T2D risk.

a Number of genes among 162 genes differentially expressed and detected in different scRNA-seq data of diverse tissues from T2D patients. The number of samples from each group is as follows: Intestine (Control naive CD4⁺ T cells: n = 643, T2D naive CD4⁺ T cells: n = 1403, Control memory CD4⁺ T cells: n = 3300, T2D memory CD4⁺ T cells: n = 1509), Kidney (Control naive CD4⁺ T cells: n = 409, T2D naive CD4⁺ T cells: n = 1594, Control memory CD4⁺ T cells: n = 112, T2D memory CD4⁺ T cells: n = 866), Liver (Control naive CD4⁺ T cells: n = 741, T2D naive CD4⁺ T cells: n = 107, Control memory CD4⁺ T cells: n = 3264, T2D memory CD4⁺ T cells: n = 454), Lung (Control naive CD4⁺ T cells: n = 3411, T2D naive CD4⁺ T cells: n = 393, Control memory CD4⁺ T cells: n = 1210, T2D memory CD4⁺ T cells: n = 649), PBMC1 (Control naive CD4⁺ T cells: n = 2008, T2D naive CD4⁺ T cells: n = 3691, Control memory CD4⁺ T cells: n = 678, T2D memory CD4⁺ T cells: n = 773), PBMC2 (Control naive CD4⁺ T cells: n = 13072, T2D naive CD4⁺ T cells: n = 7301, Control memory CD4⁺ T cells: n = 4928, T2D memory CD4⁺ T cells: n = 4138), Wound-edge (Control naive CD4⁺ T cells: n = 689, T2D naive CD4⁺ T cells: n = 173, Control memory CD4⁺ T cells: n = 288, T2D memory CD4⁺ T cells: n = 880). b Genes validated in differential expression analysis. The color represents their average expression fold change levels. The size of a point represents the number of times it has been validated in different datasets. c UMAP embedding of scRNA-seq data of naive CD4⁺ T cells from T2D patients and healthy controls. d STK17B gene expression in naive CD4⁺ T cells.

Fig. 4. Comparison of dynamic and non-dynamic MR results of the 162 T2D genes and 80 CAD genes.

Red and yellow dots represent genes causally affecting T2D and/or CAD risk with colocalization evidence in the dynamic eQTL data or in any of the non-dynamic eQTL datasets. Gray dots indicate the gene is absent in a given database or the result was tested in a given database but did not show robust MR and colocalization evidence. a Genes causally affecting T2D risk that are only detected in dynamic eQTL data and for example, genes detected in dynamic or non-dynamic eQTL datasets. b Genes causally affecting CAD risk that were only detected in dynamic eQTL data and for example, genes detected in dynamic or non-dynamic eQTL datasets. c Distribution of the 162 T2D genes and 80 CAD genes in dynamic and non-dynamic datasets.

Fig. 5. Cell type-specific analysis results.

a Distribution of the number of genes with colocalization evidence and the number of genes for instruments in dynamic eQTL data across CD4⁺ T cell expression profiles. Numbers in the box represent the number of genes corresponding to each expression profile. b Correlation of gene expression in CD4⁺ T cells and that in other immune cell types (n = 91). c Identified genes were specific to CD4⁺ T cells and not highly expressed in any other immune cell types.

Fig. 6. Distribution of genes, pair-wise Z score and heterogeneity test of MR estimates across different activation time points.

a Proportion of the 162 T2D genes and 80 CAD genes by activation status and five time points. b Example genes that showed distinguish causal effects across activation time points. Effect estimates were odds ratio and 95% confidence interval of disease risk per unit change of the expression levels of the relevant gene. The error bars represent 95% confidence intervals. c Pair-wise Z score of gene–disease pairs across different activation time points. X-axis is Z-score from two-sided pair-wise Z test with degree of freedom of one. The Y-axis is minus log 10 P value of the pair-wise Z test. d Number and two-sided Fisher’s exact test of gene-disease pairs with or without robust MR evidence and time point-specific effects.