We combined the data from two COPD case-control fine-mapping studies on chromosome 2q as an initial screening step and validated the findings in two family-based studies of COPD. The multistudy fine-mapping strategy identified XRCC5
as a potential COPD-susceptibility gene. Although SNP rs3821104 was not significant in all of the cohorts individually, the direction of effect was consistent in the two case-control and two family-based studies, and the combined P
value remained significant even when adjusted for multiple testing. Associations with XRCC5
were more pronounced in stratified analyses of younger subjects with COPD, which corresponds to the initial linkage findings in the Boston Early-Onset COPD Study, which recruited probands with severe airflow obstruction at an early age (5
). Genotype imputation pointed to a 25-kb region of XRCC5
most strongly associated, although the specific functional variant or variants have yet to be identified, despite resequencing of the exonic regions of XRCC5
The product of XRCC5
is an ATP-dependent DNA helicase involved in DNA double-strand break repair and immunoglobulin V(D)J rearrangement (26
is also known as Ku80 or Ku86. There are several potential mechanisms for the role of XRCC5
in the development of COPD, specifically early-onset COPD. Pulmonary emphysema has been described as an aging-related phenomenon (27
). For example, mice deficient in the antiaging gene klotho
develop emphysema among other manifestations of early senescence (29
mice also develop early aging (30
), although the lungs of these mice have not been specifically examined for emphysema (P. Hasty, personal communication).
Ku80/86 has been identified as an autoantigen in systemic lupus erythematosus and other autoimmune diseases (31
). Several groups have proposed an autoimmune component to COPD (32
). These studies have identified antiendothelial cell and antielastin antibodies, but not antinuclear antibodies that are relevant in systemic lupus erythematosus and other rheumatic diseases. However, an older study did find a higher incidence of low-titer antinuclear antibody positivity in patients with chronic bronchitis compared with control subjects (36
In mouse lung epithelial cell lines, Maeda and colleagues demonstrated that Ku80 coimmunoprecipitated with thyroid transcription factor-1 and poly(ADP-ribose) polymerase-2 (37
). This complex activated the surfactant protein-B (Sftpb) promoter in mouse lung epithelial cells and in Hela cells, but not the other surfactant protein genes in the murine cell line. Our group and others have found association between variants in SFTPB and COPD in human populations (11
Despite the previously published association with SERPINE2
and subsequent replication study, which used the same study populations as the current analysis (7
), we did not identify SERPINE2
based on the screening methods used in the current analysis. Several factors may explain this discrepancy, including differences in sample sizes, especially regarding the NETT-NAS analysis. Both of the prior papers used a family-based study (Boston Early-Onset COPD Study and ICGN, respectively) as the initial screening sample, with follow-up in the case-control samples (NETT-NAS and Norway), whereas the current study used the opposite approach of screening in the two case-control samples and serial replication in the two family-based studies. Moreover, the strongest associations for SERPINE2
SNPs in the Norway study were with quantitative spirometric traits, and not COPD status, which was the main phenotype used in the present analysis. The differences in results demonstrate that genetic association studies of carefully selected candidate genes— including multiple SNPs in the gene and additional quantitative as well as qualitative phenotypes—may provide different insights than regional fine mapping or even GWAS. In addition, a case-control study from the United Kingdom failed to demonstrate association with SERPINE2
). This lack of replication is common throughout complex trait genetics.
The linkage regions from the Boston Early-Onset COPD Study and the ICGN Study overlapped, but did not correspond exactly at the margins. SNP selection strategies were based on different tagging approaches in the NETT-NAS and the Norway studies. Genotyping was conducted independently, although both studies used the Illumina GoldenGate genotyping platform. Therefore, the fine mapping SNP sets are not the same in the two case-control studies used in the initial screening step. This is a limitation to our analysis. However, we were able to use the overlapping set of 790 SNPs to identify significant associations. Genotype imputation has been proposed as a method to combine data from different SNP sets, for example from different genomewide SNP chips (41
). However, imputed genotypes are clearly less accurate than direct SNP genotyping (42
), as evidenced by the large number of SNPs that failed imputation in our two case-control cohorts. Therefore, we elected to use SNP imputation only in a limited region as a confirmatory step, instead of imputing genotypes across the entire linkage region.
The differences in phenotype definitions across the study samples, specifically regarding subject ages and COPD severity, is a potential limitation to our study, which may have contributed to the heterogeneity of the association across the included populations. The associations with XRCC5 were more prominent in younger individuals, which is not surprising given the initial linkage in families recruited through a proband with severe, early-onset COPD. However, we did find significant evidence of association in subjects of older ages in the NETT-NAS Study. The family members in both the Boston Early-Onset COPD Study and the ICGN Study represent a broader range of ages and lung function (). Although possibly more important in severe early-onset COPD, the association with XRCC5 may be generalizable to subjects with a wider range of COPD severity. Although XRCC5 may be a COPD-susceptibility gene, we did not find evidence that XRCC5 modified lung function in subjects with established COPD in NETT, Norway, or the Lung Health Study. As with all complex traits, additional replication studies will be required to confirm the role of XRCC5 variants in COPD susceptibility. Studies in other racial groups may also be important, given the variation in minor allele frequency of SNP rs3821104 across racial groups in the HapMap project. In addition, functional studies will be required to determine the potential impact of XRCC5 variants on COPD pathophysiology.