Genome-wide analysis highlights contribution of immune system pathways to the genetic architecture of asthma

Abstract

Asthma is a chronic and genetically complex respiratory disease that affects over 300 million people worldwide. Here, we report a genome-wide analysis for asthma using data from the UK Biobank and the Trans-National Asthma Genetic Consortium. We identify 66 previously unknown asthma loci and demonstrate that the susceptibility alleles in these regions are, either individually or as a function of cumulative genetic burden, associated with risk to a greater extent in men than women. Bioinformatics analyses prioritize candidate causal genes at 52 loci, including CD52, and demonstrate that asthma-associated variants are enriched in regions of open chromatin in immune cells. Lastly, we show that a murine anti-CD52 antibody mimics the immune cell-depleting effects of a clinically used human anti-CD52 antibody and reduces allergen-induced airway hyperreactivity in mice. These results further elucidate the genetic architecture of asthma and provide important insight into the immunological and sex-specific relevance of asthma-associated risk variants.

Fig. 1. Overview of genetic and functional analyses.

A GWAS was first carried out using primary level data in the UK Biobank with ~9.5 million SNPs. In parallel, summary statistics were imputed based on publicly available GWAS data from the TAGC. The results were combined in a Z-score meta-analysis that included 88,486 asthma cases and 447,859 controls with 8,365,359 variants common to both datasets. 58 previously unknown genome-wide significant loci for asthma were identified in the meta-analysis. Combined with the eight previously unknown genome-wide significant loci from the GWAS analysis in the UK Biobank alone, a total of 66 previously unknown asthma susceptibility loci were identified. Follow up bioinformatics and eQTL analyses prioritized CD52 for in vivo functional validation studies.

Fig. 2. GWAS meta-analysis results for asthma.

Of the 66 previously unknown loci, 58 were significantly associated with asthma in the meta-analysis with the UK Biobank and TAGC (red dots). Eight loci were associated with asthma in the GWAS analysis with the UK Biobank alone but fell slightly below the genome-wide significance threshold in the meta-analysis with the TAGC (light blue dots). The Z-score meta-analysis included a total of 88,486 cases and 447,859 controls from the UK Biobank (64,538 asthma cases and 329,321 controls) and TAGC (23,948 asthma cases and 118,538 controls) and 8,365,359 SNPs common to both datasets. Genome-wide thresholds for significant (P = 5.0 × 10⁻⁸) and suggestive (P = 5.0 × 10⁻⁶) association are indicated by the horizontal red and dark blue lines, respectively. P-values are truncated at −log₁₀(P) = 30.

Fig. 3. Cumulative genetic burden and risk of asthma in the UK Biobank.

a Individuals in the highest decile for three weighted genetic risk scores (GRS) based on the known or previously unreported susceptibility loci have significantly elevated risk for asthma compared to the lowest decile. b–d Compared to females, cumulative genetic risk for asthma was more significantly pronounced in males across the entire spectrum of sex-specific weighted GRS constructed from all 212 (P-int = 6.1 × 10⁻¹⁸), 146 known (P-int = 3.0 × 10⁻¹⁸), and 66 previously unknown (P-int = 0.01) loci. Interactions with sex were also significant when comparing the highest versus lowest decile for the GRS constructed from all 212 (P-int = 7.3 × 10⁻¹⁰), 146 known (P-int = 1.9 × 10⁻¹⁰), and 66 previously unknown (P-int = 0.03) loci. Each decile in the GRS analyses included 39,386 subjects when both sexes were combined (a) or 18,318 males and 21,068 females when stratified by sex (b–d). Data are shown as ORs with 95% CIs, as calculated by logistic regression.

Fig. 4. Enrichment analyses with asthma-associated variants.

a GARFIELD analysis revealed highly significant 3 to 4-fold enrichment of asthma-associated variants colocalizing to DNase I hypersensitive sites (peaks) in several asthma-related tissues, such as blood (red box), epithelium (orange box), lung (blue box), and thymus (purple box). The radial plot shows fold enrichment in various available tissues using different GWAS significance thresholds for the inclusion of asthma-associated variants in the GARFIELD analysis, as calculated by logistic regression. The small dots on the inside of the outer most circle indicate whether the inclusion of asthma-associated variants at GWAS significance thresholds of 1.0 × 10⁻⁵ (one dot), 1.0 × 10⁻⁶ (two dots), 1.0 × 10⁻⁷ (three dots), 1.0 × 10⁻⁸ (four dots) in the GARFIELD analysis yielded significant enrichment in that tissue or cell type at P = 1.0 × 10⁻¹⁵. b Highly significant enrichment in DNase I hypersensitive sites was particularly evident in immune cells, such as GM12878 B cells, CD19⁺ B cells as well as CD3⁺, CD4⁺, and CD8⁺ T cells.

Fig. 5. Effect of an anti-CD52 antibody on lung function and inflammation in mice.

a The experimental protocol for testing the effect of a mouse anti-CD52 (αCD52) antibody on pulmonary function and inflammation was designed to mimic the clinical protocol used to treat multiple sclerosis patients with a human αCD52 antibody (alemtuzumab). Female BALB/cByJ mice were immunized on day 1 with 100 µg of house dust mite (HDM) in 2 mg of aluminum hydroxide (alum) by intraperitoneal (i.p.) injection. On day 7, mice were intraperitoneally administered with 500 µg of either the αCD52 antibody (Group 1, red bars; n = 8) or isotype control antibody (Group 2, black bars; n = 9), or phosphate-buffered saline (PBS) (Group 3, white bars; n = 4). On days 8, 9 and 10, mice in Groups 1 and 2 were intravenously (i.v.) administered 250 µg of the αCD52 antibody or isotype control antibody, respectively, and simultaneously challenged intranasally (i.n.) with 50 µg HDM. Mice in Group 3 were only challenged with PBS i.n. on days 8, 9, and 10. On day 11, lung function was measured by invasive plethysmography and leukocyte counts were determined in bronchial alveolar lavage (BAL) by flow cytometry. Compared to HDM-exposed mice receiving the control isotype antibody, HDM-exposed mice administered the αCD52 antibody had significantly reduced lung resistance (b) and improved dynamic compliance (c). HDM-exposed mice receiving the αCD52 antibody also had significantly decreased numbers of total cells, eosinophils, T cells, and neutrophils in BAL (d) as well as decreased inflammation and thickness of the airway epithelium (e) compared to HDM-exposed mice injected with the isotype control antibody. Scale bars equal 50 μm. Data are shown as mean ± SE and derived from two independent experiments yielding similar results. Two-tailed unpaired Student’s t-tests were used to determine statistically significant differences between groups. *P < 0.05, **P < 0.005; ***P < 0.0005; ****P < 0.0001.