Sex differences in adenosine deaminase activity associate with disparities in SARS-CoV-2 innate immunity

Abstract

Females demonstrate elevated type-I interferon production and a stronger antiviral immune response; however, the mechanisms underlying sex-based differences in antiviral immunity are incompletely understood. We previously reported that low adenosine deaminase (ADA) activity perturbs the methylation-based transcriptional silencing of endogenous retroviral elements (hERV), which stimulates IFN-Stimulated Genes (ISG) and primes antiviral immunity. Here we demonstrate lower ADA activity in females compared to their male counterparts, which correlated with higher hERV and ISG expression in female lungs. Sex differences in ADA2 were linked to the number and expression profiles of blood and lung-derived monocyte populations. Single-cell RNA sequencing of respiratory cells from patients with COVID-19 showed a significant female bias in hERV-ISG signatures, and implicated IL-18 as a driver of sex-specific ADA2 expression. Observations in healthy and COVID-19 cohorts indicate that higher ADA activity is associated with suppressed antiviral innate immunity in the male respiratory tract, which may drive adverse COVID-19 outcomes.

Keywords: Immunology; Virology.

Graphical abstract

Figure 1

Healthy female individuals exhibit lower peripheral ADA and higher tissue hERV (A) Demographic table of the 500 Functional Genomes (500FG) cohort. (B–F) Total, EHNA-sensitive (i.e., ADA1) or EHNA-resistant (i.e., ADA2) adenosine deaminase enzyme activity measured in peripheral blood from healthy individuals and stratified by (B) sex, (C) contraceptive use, (D) age, (E) BMI, and (F) smoking history. Data are represented as individual points with bar = mean. (G) Schematic representation of the mechanism by which ADA suppresses hERV gene methylation and type-I IFN. (H) Cohort description for bulk measurements of hERV transcripts from 134 human tissue samples from Oncobox Atlas of Normal Tissue Expression (ANTE, Suntsova et al. (2019) Sci Data). (I) Total hERV expression in solid organs stratified by sex. Data are represented as individual points with bar = mean. (J) Volcano plot shows the differential expression of individual hERV transcripts in the lung. (K) Differential expression of hERV transcripts in the esophagus and stomach. Gray/purple delineate significance cutoff per legend. (L) Heatmaps of differentially expressed hERV transcript in lung (n = 5 males, 3 females), esophagus (n = 6 males, 6 females), and stomach (n = 8 males, 6 females). Stats = with univariate linear regression, n = 474 (Figures 1B–1F), multivariate linear regression, covariates = individual, tissue, sex (Figure 1I), differential gene expression analysis with Mann-Whitney post-hoc analysis (Figures 1J and 1K). CPM = counts per million. See also Figure S1, Table S1.

Figure 2

Sex differences in ADA2 activity and expression are linked to monocyte populations (A) Demographics table for the EMBARC COVID-19 cohort. (B) Forest plot of the odds ratio for ICU admission or mortality in patients with EMBARC COVID-19. Error = 95% CI, vertical rectangle = mean log odds ratio (C–D) Adenosine deaminase activity in patients stratified by (C) COVID-19 status or (D) COVID-19 status and sex. Data are represented as individual points with bar = mean. (E) Demographic table of Ren et al. COVID-19 cohort. (F) Two-dimensional tSNE plot of single-cell RNAseq analysis of COVID-19 BALF and sputum (n = 24), stratified by cell type (left) and relative ADA2 mRNA expression (right). (G) ADA2 mRNA expression in monocytes and macrophages from patients with COVID-19, stratified by patient sex. (H) ADA2 mRNA expression in patients with COVID-19, stratified by monocyte clusters and patient sex. (I and J) ADA2 activity measured in bulk populations of monocytes (CD14⁺) (I) or monocyte subsets (J) by flow cytometry, (500FG, n = 474). Data are represented as individual points with bar = mean. Stats = univariate logistic regression (Figure 2B); multivariate logistic regression, covariates = age, sex, BMI (Figures 2C and 2I left), covariates = age, BMI (Figures 2D and 2I right; Figure 2J); covariates = patient (Figures 2G and 2H). See also Figure S2, Table S2.

Figure 3

Expression of hERV and ISG signatures in the lungs of patients with COVID-19 (A) Two-dimensional tSNE representation of single-cell RNAseq analysis of BALF and sputum from patients with COVID-19, stratified by cell type (left) and relative hERV transcript levels (right). (B) Volcano plot of gene ontology and biological pathway (GOBP) terms differentially expressed between BALF/sputum-derived single epithelial cells from male and female patients with COVID-19. (C) Scatterplots of hERV transcript levels relative to ISG transcript levels in epithelial cells without detectable COVID-19 transcripts. Linear regression with a 95% confidence interval is shown. (D) ISG expression stratified by patient sex. Stats = Mann-Whitney nonparametric (Figure 3B), univariate linear regression (Figure 3C), multivariate linear regression, covariates = patient (Figure 3D). See also Figures S3 and S4.

Figure 4

IL-18 expressed by lung epithelial promote ADA2 expression by monocytes (A and B) Variation of (A) IL-18, and (B) IL-1b cytokine concentration by sex (top) or ADA2 activity (bottom), in blood of healthy individuals in the 500FG cohort (n = 408). (C) Two-dimensional tSNE representation of single-cell RNAseq analysis of COVID-19 patient BALF/sputum, stratified by cell type (left) and relative IL-18 expression (right). (D) IL-18 mRNA expression in BALF/sputum-derived lung epithelial cells, stratified by patient sex. Data are represented as individual points with bar = mean. Stats = multivariate linear regression, covariate = sex (Figures 4A and 4B); multivariate linear regression, covariates = patient (Figure 4D). See also Figure S5.