Variation in ERAP2 has opposing effects on severe respiratory infection and autoimmune disease

IntroductionERAP2 is an aminopeptidase involved in immunological antigen presentation. Genotype data in human samples from before and after the Black Death, an epidemic due to Yersinia pestis, have marked changes in population allele frequency of the common single nucleotide polymorphism (SNP) rs2549794. This SNP in strong linkage disequilibrium with a key splicing SNP in ERAP2 (rs2248374) and this suggests that variation at ERAP2 may be relevant for protection from infection. rs2549794 is also associated with Crohns disease and findings imply balancing selection between infection and autoimmune disease at this locus. There have been no large-scale prospective case-control studies of variation at ERAP2 and infection. MethodsThis study aimed to explore the association between variation at ERAP2 and a) infection, b) autoimmune disease, and c) parental longevity as a proxy for lifespan. Genome Wide Association Studies (GWAS) of these outcomes were identified in contemporary cohorts (UK Biobank, FinnGen, and GenOMICC). Effect estimates were extracted for rs2549794 and rs2248374. Additionally, cis expression and protein quantitative trait loci (QTLs) for ERAP2 were used in Mendelian randomisation analyses. ResultsAcross all cohorts, the T allele (minor allele frequency of 0.4-0.5) of rs2549794 showed evidence of association with respiratory infection (odds ratio; OR for pneumonia 1.03; 95% CI 1.01-1.05; p = 0.014). Effect estimates were larger in bacterial rather than viral infection and larger for more severe phenotypes (OR for critical care admission with pneumonia 1.08; 95% CI 1.02-1.14, p = 0.008, OR for death from pneumonia 1.07; 95% CI 1.01-1.12; p = 0.014). In contrast, opposing effects were identified for Crohns disease (OR 0.86; 95% CI 0.82-0.90, p = 8.6 x 10-9) and type 1 diabetes (OR 0.95; 95% CI 0.90-0.99, p = 0.02). No strong evidence for association was identified for sepsis. Carriage of the T allele was associated with increased age of parental death (beta in Z-scored years across both parents age at death 0.01, 95% CI 0.004-0.017, p = 0.002). Similar results were identified for rs2248374. In Mendelian randomisation analyses, increasing transcription or protein levels of ERAP2 were strongly associated with protection from respiratory infection, with opposing effects identified on Crohns disease and type 1 diabetes. Increased expression of ERAP2 was associated with reduced parental longevity. ConclusionsVariation at ERAP2 is associated with severe respiratory infection in modern societies, with an opposing association with Crohns disease and type 1 diabetes. These data support the hypothesis that changes in allele frequencies in ERAP2 observed at the time of the Black Death reflect protection from infection, and suggest ongoing balancing selection at this locus driven by autoimmune and infectious disease