Genotype-phenotype correlation of T-cell subtypes reveals senescent and cytotoxic genes in Alzheimer's disease

Abstract

Recent studies identifying expression quantitative trait loci (eQTLs) in immune cells have uncovered important links between disease risk alleles and gene expression trends in monocytes, T cells and other cell types. However, these studies are generally done with young, healthy subjects, limiting the utility of their findings for age-related conditions such as Alzheimer's disease (AD). We have performed RNA sequencing on four T-cell subsets in genome-wide genotyped and well-characterized AD subjects and age- and sex-matched controls from the Religious Orders Study/Memory and Aging Project. We correlated gene expression data with AD neuropathological traits and with single-nucleotide polymorphisms to detect eQTLs. We identified several significant genes involved in T-cell senescence and cytotoxicity, consistent with T-cell RNA sequencing studies in aged/AD cohorts. We identified unexpected eQTLs previously associated with neuropsychiatric disease traits. Finally, we discovered that pathways related to axon guidance and synaptic function were enriched among trans-eQTLs in coding regions of the genome. Our data strengthen the potential link between T-cell senescence and age-related neurodegenerative disease. In addition, our eQTL data suggest that T-cell phenotypes may influence neuropsychiatric disorders and can be influenced by genes involved in neurodevelopmental processes.

Figure 1

Patterns of T-cell gene expression, co-expression and correlation with AD pathological traits. (A) Overview of study design, including the number of AD and control subjects, cell sorting strategy and data collection. Created with Biorender.com. (B) Heatmap displaying numbers of differentially expressed genes (DEGs) when compared between given cell types and AD status. n = 96, FDR-adjusted P-value < 0.05. (C) Selected genes (with immune or disease significance) associated with AD pathological traits (shape legend shown at top right) at a nominal P < 0.05 for each cell type, arranged according to the cellular localization of protein product.

Figure 2

Relative abundance of WGCNA co-expression modules by cell type and pathway analysis. (A) Relative expression of modules 1, 2 and 3 are shown for four T-cell subtypes (red = CD4 + CD45RO−, green = CD4 + CD45RO+, cyan = CD8 + CD45RO−, magenta = CD8 + CD45RO+). (B) Results of Enrichr pathway analysis on modules 1, 2 and 3. Dot size reflects the number of module genes in a given pathway, location on the X axis represents the proportion of the GO pathway represented by module genes and dot color represents P-value. Terms on the left are from the GO Biological Processes 2021 database.

Figure 3

Distribution of trans-eQTL across the genome and between T-cell subtypes. (A) Venn diagrams showing the distribution of unique genes (top) and SNPs (bottom) among the four T-cell subsets. Genes found in trans-eQTL are mainly shared among all T cells, while SNPs are more subtype-specific. (B) Manhattan plots showing the distribution of SNPs from the trans-eQTL in each cell type, with SNPs at –log₁₀(P) > 14 annotated.

Figure 4

Disease association of trans-eQTL. (A) Stacked bar graph showing numbers of trans-eQTL whose SNP is found to associate with certain categories of disease traits as found in the GWAS catalog. (B) Circos plots displaying the distribution of disease-associated trans-eQTL by chromosome. Color legend for GWAS categories is found on the bar graph.

Figure 5

Pathway analysis of trans-eSNPs within the coding region of a gene. For SNPs that fell within the coding region of a gene, pathway analysis was run with Enrichr against the GO Biological Processes 2021 database for each cell type, (A) CD4 + CD45RO−, (B) CD4 + CD45RO+, (C) CD8 + CD45RO− and (D) CD8 + CD45RO+. These SNPs were highly associated with pathways related to synapse organization and function, axon guidance and neurodevelopment.