Sex-specific differences in peripheral blood leukocyte transcriptional response to LPS are enriched for HLA region and X chromosome genes

Abstract

Sex-specific differences in prevalence are well documented for many common, complex diseases, especially for immune-mediated diseases, yet the precise mechanisms through which factors associated with biological sex exert their effects throughout life are not well understood. We interrogated sex-specific transcriptional responses of peripheral blood leukocytes (PBLs) to innate immune stimulation by lipopolysaccharide (LPS) in 46 male and 66 female members of the Hutterite community, who practice a communal lifestyle. We identified 1217 autosomal and 54 X-linked genes with sex-specific responses to LPS, as well as 71 autosomal and one X-linked sex-specific expression quantitative trait loci (eQTLs). Despite a similar proportion of the 15 HLA genes responding to LPS compared to all expressed autosomal genes, there was a significant over-representation of genes with sex by treatment interactions among HLA genes. We also observed an enrichment of sex-specific differentially expressed genes in response to LPS for X-linked genes compared to autosomal genes, suggesting that HLA and X-linked genes may disproportionately contribute to sex disparities in risk for immune-mediated diseases.

Figure 1

Study overview. Where shown, HLA and X-linked genes were analyzed together with the remaining autosomal genes but are discussed separately in the manuscript and are therefore shown here separate from the autosomes to follow the manuscript flow. (A) Flowchart of differential expression analysis. Analyses of autosomal genes in the combined sample (main effect and sex interaction) is shown in lavender, analyses of X-linked genes in males and females separately (autosomal genes included for comparison) are shown in blue and orange, respectively. (B) Flowchart of eQTL mapping in each of the four groups.

Figure 2

Violin plots of exemplar genes with significant (FDR < 0.05) LPS treatment by sex interaction effects on gene expression indicating different responses to LPS treatment in males and females. Interaction p-values are shown. For each gene, expression levels for each treatment and sex are shown on the x-axis (male untreated, male LPS-treated, female untreated, female LPS-treated). Each boxplot inside the violin shows the 1st-3rd interquartile range; the horizontal line shows the median expression value. (A) CCL19, Chemokine ligand 19; (B) CCL20, chemokine ligand 20/macrophage inflammatory protein-3; (C) IL1A, Interleukin 1 alpha; (D) CXCR2, Interleukin 8 receptor, beta; (E) IRF7, Interferon regulatory factor 7; (F) IRF1, Interferon regulatory factor 1. Interaction P-values are shown. Note the different y-axis scales.

Figure 3

Scatterplot of fold change responses to LPS of genes located on (A) the autosomes (N = 12,475) and (B) the X chromosome (N = 393). Only genes with significant responses are shown. At FDR < 0.05, genes colored in blue were observed only in males, genes colored in orange were observed only females, and genes colored in purple were observed in both sexes.

Figure 4

Scatterplot of fold change for genes with significant (FDR < 0.05) sex by LPS treatment interaction effects (n = 1055). Genes colored in blue have a larger absolute magnitude of response to LPS in males compared to females; genes colored orange have a larger absolute magnitude of response in females compared to males. Labeled genes have a fold change of > 6 or < − 4.

Figure 5

Venn Diagrams with (A) Number of significant eGenes by condition (lfsr < 0.05) on the autosomes, and (B) number of significant eGenes by condition (lfsr < 0.05) on the X chromosome.

Figure 6

Meta-analysis forest plots showing the effect size (x-axis) of exemplar gene-SNP pairs for each treatment condition (y-axis; M LPS: expression data in males in the LPS-treated samples; M Veh: expression data in males in the untreated controls; F LPS: expression data in females in the LPS-treated samples; F Veh: expression data in females in the untreated controls). Black boxes show the mean effect size; and the size of the box is proportional to the inverse of the standard error (larger boxes indicate a smaller standard error). The gray lines show the 95% confidence intervals. All eQTLs shown are significant in at least one condition at an lfsr < 0.05. (A) Example of an eQTL shared across all conditions. HLA-DQA2, Human leukocyte antigen DQ alpha 2. (B) Example of a sex-specific eQTL (males), NOG, Noggin. (C) Example of a treatment-specific eQTL (vehicle only), KIAA1324, estrogen-induced gene 121 protein. (D) Example of a treatment-specific eQTL (LPS only). IL36RN, interleukin 36 receptor antagonist. (E) Example of a treatment specific and sex specific eQTL (Female LPS only), NOXA1, NADPH oxidase activator 1. (F) Example of a treatment specific and sex specific eQTL (Male LPS only), IL1R1, interleukin 1 receptor type 1.

Figure 7

Boxplots illustrating four HLA genes with significant sex by treatment interactions. (A) HLA-A and (B) HLA-B are examples of genes where the magnitude of response differs between males and females. (C) HLA-DQA1 and (D) HLA-DQB1 show opposite directions of response to LPS treatment in males and females.