Immunochip analysis identifies association of the RAD50/IL13 region with human longevity

Summary Human longevity is characterized by a remarkable lack of confirmed genetic associations. Here, we report on the identification of a novel locus for longevity in the RAD50/IL13 region on chromosome 5q31.1 using a combined European sample of 3208 long‐lived individuals (LLI) and 8919 younger controls. First, we performed a large‐scale association study on 1458 German LLI (mean age 99.0 years) and 6368 controls (mean age 57.2 years) by targeting known immune‐associated loci covered by the Immunochip. The analysis of 142 136 autosomal single nucleotide polymorphisms (SNPs) revealed an Immunochip‐wide significant signal (PI mmunochip = 7.01 × 10–9) for the SNP rs2075650 in the TOMM40/APOE region, which has been previously described in the context of human longevity. To identify novel susceptibility loci, we selected 15 markers with PI mmunochip < 5 × 10–4 for replication in two samples from France (1257 LLI, mean age 102.4 years; 1811 controls, mean age 49.1 years) and Denmark (493 LLI, mean age 96.2 years; 740 controls, mean age 63.1 years). The association at SNP rs2706372 replicated in the French study collection and showed a similar trend in the Danish participants and was also significant in a meta‐analysis of the combined French and Danish data after adjusting for multiple testing. In a meta‐analysis of all three samples, rs2706372 reached a P‐value of PI mmunochip+Repl = 5.42 × 10−7 (OR = 1.20; 95% CI = 1.12–1.28). SNP rs2706372 is located in the extended RAD50/IL13 region. RAD50 seems a plausible longevity candidate due to its involvement in DNA repair and inflammation. Further studies are needed to identify the functional variant(s) that predispose(s) to a long and healthy life.


Summary
Human longevity is characterized by a remarkable lack of confirmed genetic associations. Here, we report on the identification of a novel locus for longevity in the RAD50/IL13 region on chromosome 5q31.1 using a combined European sample of 3208 long-lived individuals (LLI) and 8919 younger controls. First, we performed a large-scale association study on 1458 German LLI (mean age 99.0 years) and 6368 controls (mean age 57.2 years) by targeting known immune-associated loci covered by the Immunochip. The analysis of 142 136 autosomal single nucleotide polymorphisms (SNPs) revealed an Immunochip-wide significant signal (P Immunochip = 7.01 3 10 -9 ) for the SNP rs2075650 in the TOMM40/APOE region, which has been previously described in the context of human longevity. To identify novel susceptibility loci, we selected 15 markers with P Immunochip < 5 3 10 -4 for replication in two samples from France (1257 LLI, mean age 102.4 years; 1811 controls, mean age 49.1 years) and Denmark (493 LLI, mean age 96.2 years; 740 controls, mean age 63.1 years). The association at SNP rs2706372 replicated in the French study collection and showed a similar trend in the Danish participants and was also significant in a meta-analysis of the combined French and Danish data after adjusting for multiple testing. In a meta-analysis of all three samples, rs2706372 reached a P-value of P Immunochip+Repl = 5.42 3 10 À7 (OR = 1.20; 95% CI = 1.12-1.28). SNP rs2706372 is located in the extended RAD50/IL13 region. RAD50 seems a plausible longevity candidate due to its involvement in DNA repair and inflammation. Further studies are needed to identify the functional variant(s) that predispose(s) to a long and healthy life.
Despite more than 20 years of research into the genetic basis of human longevity, only alleles in the APOE and FOXO3 genes have repeatedly been shown to be associated with survival to very advanced ages (Sch€ achter et al., 1994;Willcox et al., 2008;Flachsbart et al., 2009;Soerensen et al., 2010;Deelen et al., 2013). APOE and FOXO3 were initially detected in candidate-driven case-control investigations, but APOE has since then been confirmed in a number of genome-wide association studies (GWAS). In addition, a single nucleotide polymorphism (SNP) on chromosome 5q33.3 was recently identified in a GWAS meta-analysis on long-lived individuals (LLI) aged ≥ 90 years (Deelen et al., 2014). Besides the detection of APOE and the 5q33.3 locus, longevity GWAS have been relatively unsuccessful and have failed to reveal novel associations with genome-wide significance or sufficient reproducibility (Deelen et al., 2011(Deelen et al., , 2014Nebel et al., 2011). Here, we performed a large-scale candidate gene study by targeting established immune-associated loci present on the Immunochip (Trynka et al., 2011). The Immunochip was designed to perform fine-mapping of GWAS loci of major immune-mediated diseases using data from the 1000 Genomes Project and other sequencing initiatives. The application of the array in this study is based on the hypothesis that a wellfunctioning immune system and efficient anti-inflammatory networks are potent longevity-assurance mechanisms (Franceschi et al., 2007). We employed the Immunochip to screen 1458 German LLI and 6368 younger controls in a discovery phase (panel A in Table S1) for novel longevity loci followed by replication in 1750 LLI and 2551 younger controls from France and Denmark (panel B in Table S1).
After applying conservative and established quality filters to the German longevity sample, 142 136 autosomal SNPs were available for association analysis (see Appendix S1). We used a predefined threshold of P = 6.15 9 10 À7 to define statistical significance for the Immunochip-wide analysis, based on the Bonferroni correction for the number of linkage disequilibrium (LD)-independent markers on the Immunochip (see Appendix S1). The comparison of the case-control frequencies yielded an Immunochip-wide significant association signal for the SNP rs2075650 in the TOMM40/APOE region (P Immunochip = 7.01 9 10 À9 ; OR = 0.69; 95% CI = 0.60-0.78; Table 1, Figs S1 and S2a). This SNP is in moderate LD (r 2 = 0.52) with SNP rs429358 that defines the wellestablished APOE e4 allele (Deelen et al., 2011), and hence, rs2075650 was not considered for replication. With regard to the FOXO3 and the 5q33.3 loci, the Immunochip is uninformative due to poor coverage (Fig. S2b,c). Eighty-four other SNPs showed P Immunochip < 5 9 10 À4 in the discovery Immunochip analysis (Table S2). Of these, we selected 15 markers for replication, each one representing the best SNP for a specific associated region defined by the clumping procedure (see Appendix S1, Table S3). The replication analysis was performed in two longevity samples from France and Denmark (France: 1257 LLI and 1811 controls; Denmark: 493 LLI and 740 controls; panel B in Table S1). The signal at SNP rs2706372, located in a region encompassing RAD50 (radiation sensitive, Saccharomyces cerevisiae homolog) and IL13 (Interleukin 13), replicated in the French sample (P Repl-France = 2.69 9 10 À3 ; OR = 1.21; 95% CI = 1.07-1.38; P Repl-France; adj = 0.04 (corrected for 15 tests)) ( Table S4). In the smaller Danish sample, the allelic effect of rs2706372 showed a similar trend (P Repl-Denmark = 0.08; OR = 1.19; 95% CI = 0.98-1.45; P Repl-Denmark; adj = 1 (corrected for 15 tests)). In the combined French-Danish replication sample, meta-analysis association analysis yielded a P-value of 4.95 9 10 À4 (OR = 1.21; 95% CI = 1.09-1.34; P Repl-France-Denmark; adj = 0.0074 (corrected for 15 tests)). In a metaanalysis of the German discovery and French-Danish replication samples, rs2706372 reached a P-value of P Immunochip+Repl = 5.42 9 10 À7 ( Table 1). Estimates of odds ratios for rs2706372 were consistent across all three studies (OR Germany = 1.19; OR Denmark = 1.19; OR France = 1.21; statistical metric of heterogeneity I 2 = 0.0), supporting the validity of the association finding.
Our targeted immune gene approach on a combined European sample of 3208 LLI and 8919 controls resulted in the identification of a novel association for longevity in the RAD50/IL13 region on chromosome 5q31.1. The lead SNP rs2706372 is located in the intronic region of the RAD50 gene and is in strong LD with other associated SNPs close to IL13 and IL5. The actual association signal extends even further to include additional genes (Fig. 1). At this point, this observation renders it difficult to assess which gene is actually affected by the association, although RAD50 is a plausible candidate. The protein encoded by RAD50 is highly similar to Saccharomyces cerevisiae Rad50 which is involved in repairing DNA double-strand breaks. Similarly, the human RAD50 is integrated in a functional DNA-binding complex (Kinoshita et al., 2015) that is important for recombination, repair, and genomic stability (Trujillo et al., 1998). Hence, it is conceivable that variation in RAD50 could positively influence longevity by increasing DNA stability.  (Soerensen et al., 2010). Chr: chromosome of marker; association boundaries: association boundaries for each index SNP (see Appendix S1). Genomic positions were retrieved from NCBI's dbSNP build v141 (genome build hg19); dbSNP id: rs ID of the index SNP; A1: minor allele; A2: major allele; AF: allele frequency of A1 estimated from Immunochip (German population); key genes: candidate gene(s) in the region; P/OR: P-value and corresponding allelic odds ratio and 95% confidence interval with respect to A1. We used a significance level of 6.15 9 10 À7 for the statistical association analysis of the Immunochip in the discovery experiment and in the combined experiment of discovery and replication based on Bonferroni correction (see Appendix S1). For each panel, numbers of LLI/controls are displayed in parentheses; I 2 : statistical metric of heterogeneity. I 2 ranges from 0 to 100% and is considered low for values 0-25%.
Alternatively, it could exert its effect via the direct modulation of cytokine expression; recent evidence suggests at least two possible avenues. First, in dendritic cells RAD15 was found to activate-upon sensing viral DNA-the transcription factor NF-jB, thus leading to the production of pro-inflammatory IL-1b (Roth et al., 2014). Second, the RAD50 gene harbors at its 3' end an evolutionarily highly conserved locus control region (LCR; Lee et al., 2003;Li et al., 2010) that regulates the expression of the neighboring cytokine genes IL-4, IL-13, and IL-5 ( Fig. 1) in Th2 cells (Kelly & Locksley, 2000). Variants in the LCR were found to be associated with asthma (Li et al., 2010). Taken together, these findings indicate that the RAD50 locus may very well contribute to longevity via its role in inflammation and immunity. Nevertheless, it is still possible that the RAD50 signal is a result of its LD with other markers within the observed association boundaries. Multiple SNPs in the extended RAD50/IL13 region were previously identified as susceptibility factors for various chronic inflammatory diseases such as Crohn's disease, psoriasis, asthma, and atopic dermatitis (Rioux et al., 2000;Li et al., 2008Li et al., , 2010Paternoster et al., 2011). Further studies are therefore needed to identify the functional variant(s) and the underlying molecular mechanisms that predispose(s) to a long and healthy life.   (Table 1 and Appendix S1). Shown are the -log 10 P-values from the Immunochip analysis (P Immunochip ) of the German longevity discovery panel (panel A in Table S1) with regard to the physical location of markers. Purple diamond: lead SNP; filled circles: analyzed SNPs where the fill color corresponds to the strength of linkage disequilibrium (r 2 ) with the lead SNP (for color coding see legend in the upper right corner of the plot); blue line: recombination intensity (cM/Mb). Positions and gene annotations are according to NCBI's build 37 (hg19). Plot was generated using LocusZoom (Pruim et al., 2010).

Supporting Information
Additional Supporting Information may be found online in the supporting information tab for this article.
Table S1 LLI/control panels used in the analysis.        Table 1).
RAD50/IL13 as a new locus for longevity, F. Flachsbart et al.