Immune cells lacking Y chromosome show dysregulation of autosomal gene expression

Abstract

Epidemiological investigations show that mosaic loss of chromosome Y (LOY) in leukocytes is associated with earlier mortality and morbidity from many diseases in men. LOY is the most common acquired mutation and is associated with aberrant clonal expansion of cells, yet it remains unclear whether this mosaicism exerts a direct physiological effect. We studied DNA and RNA from leukocytes in sorted- and single-cells in vivo and in vitro. DNA analyses of sorted cells showed that men diagnosed with Alzheimer's disease was primarily affected with LOY in NK cells whereas prostate cancer patients more frequently displayed LOY in CD4 + T cells and granulocytes. Moreover, bulk and single-cell RNA sequencing in leukocytes allowed scoring of LOY from mRNA data and confirmed considerable variation in the rate of LOY across individuals and cell types. LOY-associated transcriptional effect (LATE) was observed in ~ 500 autosomal genes showing dysregulation in leukocytes with LOY. The fraction of LATE genes within specific cell types was substantially larger than the fraction of LATE genes shared between different subsets of leukocytes, suggesting that LOY might have pleiotropic effects. LATE genes are involved in immune functions but also encode proteins with roles in other diverse biological processes. Our findings highlight a surprisingly broad role for chromosome Y, challenging the view of it as a "genetic wasteland", and support the hypothesis that altered immune function in leukocytes could be a mechanism linking LOY to increased risk for disease.

Keywords: Differential gene expression; LATE; LOY; LOY-associated transcriptional effects; Mosaic loss of chromosome Y.

Fig. 1

LOY associated transcriptional effect (LATE) in leukocytes in vivo. On the X-axes are the level of LOY mosaicism (estimated from DNA by SNP-array from three sorted cell populations) and on the Y-axes are the normalized level of gene expression estimated from the same samples using RNAseq. Panel a display the average expression of six genes located in the male-specific part of chromosome Y (MSY) as a function of LOY. Panel b shows corresponding analysis of 13 genes located in the pseudo-autosomal regions of chromosomes X and Y (PAR). Panels c, d illustrate the finding of autosomal LATE genes, i.e., genes located on other chromosomes showing reduced or increased abundance of transcripts in samples with LOY. Panel c displays the average expression of the ten most underexpressed autosomal LATE genes and the ten most over-expressed autosomal LATE genes is shown in panel d. Grey areas represent the standard error of linear regression models and beta (β) with confidence estimate (p) is shown. Mono. monocytes, Gran. granulocytes

Fig. 2

Differential expression (DE) of specific genes as an effect of LOY in NK cells and monocytes, studied by both RNAseq and scRNAseq. Panels a, b illustrate the level of DE in 206 and 60 autosomal LATE genes identified in each cell type after correction for multiple testing (FDR < 0.1) and with at least an average of 100 reads per gene in samples without LOY. Names are shown for LATE genes known to be linked with immune system functions and/or cancer and/or development of Alzheimer's disease. Panels c, d display DE observed in autosomal and PAR genes that was identified as LATE genes by both RNAseq and scRNAseq at a 0.05 α-level. MSY genes were excluded because of lack of expression in cells without chromosome Y. Independent identification of LATE was observed for 16 and 7 genes in the NK cells and monocytes, respectively. The autosomal gene LY6E and the PAR gene CD99 displayed LATE in both cell types and by both technologies

Fig. 3

Distribution of LOY in six populations of leukocytes in men diagnosed with Alzheimer’s disease or prostate cancer vs. controls. Leukocytes were sorted by FACS, followed by SNP-array genotyping for each cell fraction and calculation of the percentage of LOY. The numbers in parentheses under the X-axes in panels a, b denote the number of studied subjects for each cell type. Panels a, b show results from unadjusted analyses and panels c and d show results from logistic regression models adjusted for age and smoking. Ctrl. control, AD Alzheimer’s disease, PC prostate cancer, LR logistic regression, OR odds ratio, lymph. Lymphocytes, NK natural killer