CTLA4 gene polymorphisms are associated with chronic bronchitis

Study design

The present study is a two-phase study. First, we explored whether CTLA4 SNPs were associated with COPD and COPD-related phenotypes in a subset of the International COPD Genetics Network (ICGN) population, a large cohort of COPD probands (n = 606) and siblings (n = 1,285), whose methodological details have been described elsewhere 23–25. Secondly, we attempted to replicate our findings in an independent population that included COPD cases (n = 953) and controls (n = 956) recruited in Bergen (Norway), whose characteristics have also been reported 23, 25.

Populations and assessment

The clinical characteristics of the subjects in both cohorts included in the analysis of the chronic bronchitis phenotype are shown in table 1⇓. Subject recruitment and clinical characteristics have been described previously 23–25. Briefly, the first population represents the multicentre ICGN study; subjects with known COPD were recruited as probands, and siblings and available parents were ascertained through the probands. Inclusion criteria for probands were airflow limitation (post-bronchodilator forced expiratory volume in 1 s (FEV₁) <60% predicted and FEV₁/vital capacity (VC) <90% pred) at a relatively early age (45–65 yrs), a smoking history of ≥5 pack-yrs, and at least one eligible sibling (with a smoking history of ≥5 pack-yrs). COPD in siblings was defined by a post-bronchodilator FEV₁ <80% pred and FEV₁/VC <90% pred (generally consistent with Global Initiative for Chronic Obstructive Lung Disease (GOLD) stage II or greater). 606 pedigrees with 1,896 Caucasian individuals were genotyped in this analysis. Chronic bronchitis was defined as chronic cough and sputum production for ≥3 months·yr⁻¹ for the past 2 yrs. Among the 1,193 COPD cases of the ICGN subjects analysed, 471 (39.5%) had chronic bronchitis.

The cohort recruited in Bergen included COPD cases and controls; all subjects were either current or ex-smokers with ≥2.5 pack-yrs of smoking history. COPD was diagnosed if post-bronchodilator FEV₁ was <80% pred with FEV₁/forced vital capacity (FVC) <0.7. Controls were selected on the basis of FEV₁ >80% pred and FEV₁/FVC >0.7. A total of 953 COPD cases and 956 controls were included. Out of the 953 cases with COPD, there were 342 (35.9%) with and 511 (53.6%) without chronic bronchitis.

SNP selection

Linkage disequilibrium (LD) tagging of CTLA4 SNPs was performed using an r² threshold of 0.8 for the SNP selection with a pairwise method in the Haploview program 26. The tagging SNP selection was based on HapMap data for European-Americans (US residents with northern and western European ancestry, i.e. CEPH (Centre d'Etude du Polymorphisme Humain) samples from Utah (CEU)) with a minor allele frequency >5% from the public database 27. Accordingly, nine SNPs in CTLA4 region were selected. These SNP locations and characteristics are summarised in table 2⇓.

Genotyping

Genotyping in the two cohorts was performed with the Illumina array-based custom SNP genotyping platform (Illumina Inc., San Diego, CA, USA). Deviation from Hardy–Weinberg equilibrium for each SNP was assessed in controls by using the goodness-of-fit test with SAS software version 8.2 (SAS Institute, Inc., Cary, NC, USA); Hardy–Weinberg equilibrium for each SNP was also tested in the family data using PBAT version 3.5 (Golden Helix Inc, Bozeman, MT, USA) 28. All CTLA4 SNPs (p-values >0.05) were in Hardy–Weinberg equilibrium in both the family data and the control data. Mendelian inheritance was assessed in the family data using the PedCheck program 29, and individuals with pedigree inconsistencies for a particular SNP were removed from analysis.

Statistical analysis

In the ICGN study, we used Family-based Association Test (FBAT) version 1.7.3 30 to analyse single-SNP associations with COPD and chronic bronchitis (qualitative phenotypes). The analyses of quantitative traits (FEV₁ and FEV₁/VC) were performed with covariates including centre, age, sex, height and pack-yrs of cigarette smoking using PBAT version 3.5 (Golden Helix Inc, Bozeman, MT, USA) 28. We performed analyses using an additive genetic model in order to reduce multiple statistical testing. The additive model is equivalent to the classic transmission disequilibrium test. The risk allele was determined from the FBAT Z statistic. Values of p<0.05 from the association analyses were considered nominally significant. We estimated the number of independent SNPs for nine SNPs using the SNP spectral decomposition (SNPSpD) approach 31, and five SNPs were found to be independent. Thus, the adjusted significant p-value is <0.01 (i.e. 0.05/5) after correction for multiple testing. For the ICGN families, haplotypes were inferred and assessed using the HBAT command of the FBAT software with the use of Monte Carlo sampling 32. A minimum of 10 informative families were analysed using both FBAT and HBAT. Using a SNP sliding window approach (adjacent sets of two and three SNPs), the results of global and haplotype-specific association statistics were determined. Values of p<0.05 from the analyses were considered the nominal significant levels. Considering that 15 sliding window haplotypes were evaluated in our study, the adjusted significant p-value is 0.0033 (i.e. 0.05/15) based on a global p-value after Bonferroni correction for multiple testing. In consideration of high correlation among COPD, FEV₁, FEV₁/(F)VC and chronic bronchitis phenotypes, we chose not to correct for multiple testing of the phenotypes in order to avoid being over-conservative. For chronic bronchitis data, we used PBAT version 3.5 28 to estimate the power. At significance level of 0.05, and penetrances of 0.7, 0.4 and 0.1 for three genotypes, our study had 97.32%, 75.20%, 97.51%, 92.43%, 97.23%, 90.52%, 97.23%, 97.81% and 96.67% power to detect association for SNP 1–9, respectively.

In the Bergen cohort, a logistic regression model was used to investigate relationships of CTLA4 SNPs with COPD and chronic bronchitis. Analyses were performed using SAS software 8.2 with an additive genetic model. The odds ratios were estimated using the minor allele as the reference. Haplotype-based analyses were performed using the expectation-maximisation algorithm and score tests as implemented in Haplo.stats software 33. As we used the Bergen case–control cohort for replication of the confirmed results from the ICGN family-based association analysis, we did not perform correction of multiple testing for the association analysis of both single SNPs and haplotypes. In the Bergen cohort, we estimated an inflation factor λ from the genomic control method 34 using a panel of 221 unlinked SNPs genotyped previously and did not detect significant evidence of population stratification; the mean inflation factor of the genomic control SNPs was 1.027 23.

Visualisation of LD measures in the CTLA4 region was performed with the Haploview program 26. Haplotype blocks were defined with the confidence interval method described by Gabriel et al. 35.

Individual SNP association analyses

We evaluated nine SNPs in the CTLA4 region; the location and characteristics of those SNPs are summarised in table 2⇑. We did not detect any significant genetic associations of CTLA4 SNPs with the presence/absence of COPD in either the ICGN or the Bergen COPD cases and controls (table E1, online supplemetary material). In addition, there were no significant associations between CTLA4 SNPs and FEV₁ or FEV₁/(F)VC in either the ICGN families or the Bergen COPD cases (data not shown).

In contrast, we found significant associations with the other COPD-related phenotype investigated, chronic bronchitis. In the ICGN families, six out of the nine CTLA4 SNPs genotyped (SNPs rs926169, rs11571316, rs231775, rs231779, rs3087243 and rs231725) were significantly associated with the chronic bronchitis phenotype (table 3⇓; 0.0079≤p≤0.0432). Associations with SNPs rs231775 and rs3087243 (p = 0.0081 and 0.0079, respectively) were significant even after a correction for multiple testing.

We replicated the significant association with the chronic bronchitic phenotype in the COPD cases of the Bergen cohort for four of the six CTLA4 SNPs (rs11571316, rs231775, rs3087243 and rs231725; 0.0325≤p≤0.0408) identified in the ICGN study (table 3⇑). After evaluating the risk allele (table 3⇑), we found that three of the replicated significant SNPs (rs231775, rs3087243 and rs231725) for chronic bronchitis have the same directionality of association in the two populations.

We did not detect any significant association with chronic bronchitis in control subjects without COPD in the Bergen cohort (data not shown), suggesting that CTLA4 is associated with chronic bronchitis among COPD subjects.

LD block analysis

Figure 1⇓ shows pairwise LD r²-values for the nine SNPs in the CTLA4 region. In the ICGN study, we identified two haplotype blocks. Several of the significantly associated SNPs (rs926169, rs11571316, rs231775 and rs231777) were located in block 1, while three other significantly associated SNPs (rs231779, rs3087243 and rs231725) were located in block 2. In block 1, there were strong pairwise correlations between SNPs rs926169 and rs231775 (r² = 0.85), and between SNPs rs16840252 and rs231777 (r² = 0.86). In the Bergen cohort, the same two haplotype blocks were identified, with SNPs significantly associated with chronic bronchitis in each block.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 1—

Linkage disequilibrium (LD) map across the cytotoxic T-lymphocyte antigen (CTLA) 4 region. Pairwise LD plots of nine single nucleotide polymorphisms (SNPs) within the CTLA4 region in a) the International Chronic obstructive pulmonary disease Genetics Network family-based population and b) the Bergen case–control population. Values of r² (×100) are shown. ▪: r² = 1; □: r² = 0; squares in shades of grey: 0<r²<1 (the intensity of the grey is proportional to r²). LD block structure was estimated using the Haploview program.

Haplotype association analyses

Finally, we conducted adjacent two and three SNP haplotype analyses for chronic bronchitis in both cohorts (table 4⇓). In the ICGN data, 11 adjacent SNP combinations showed significant associations with chronic bronchitis (0.0029≤ global p-value ≤0.0457 and 0.0015≤ specific p-value ≤0.0286). Haplotypes 4–5 (rs16840252–rs231775) and 4–5–6 (rs16840252–rs231775–rs231777) for chronic bronchitis remained significant after Bonferroni correction for multiple testing. Similar results were observed in the COPD cases of the Bergen cohort, where we found nine haplotypes showing significant association with chronic bronchitis (0.0282≤ global p-value ≤0.0481 and 0.0186≤ specific p-value ≤0.0385).

Previous studies

Thus far, the most clearly documented genetic risk factor for COPD is severe α₁-antitrypsin deficiency. Many previous studies have identified a variety of genetic variants in other genes that were associated with COPD, but these results have not often been replicated in other cohorts. To our knowledge, no previous study has investigated CTLA4 in relation to COPD-related phenotypes. Our results show that CTLA4 is associated with chronic bronchitis among COPD cases. Interestingly, the CTLA4 gene is located near the region of chromosome 2q with linkage to COPD-related phenotypes 9–12. It is noteworthy that we did not find associations of CTLA4 SNPs with the presence of COPD or the severity of airflow limitation assessed by FEV₁ and FEV₁/FVC.

Potential mechanisms

T-lymphocytes play a key role in the pathogenesis of COPD and chronic bronchitis. CTLA4 is a crucial gene in the control of T-cell activation 20. We speculate that the CTLA4 polymorphisms may compromise the regulatory function of CTLA4, and that a weakened inhibitory function of CTLA4 could facilitate T-cell proliferation and differentiation, thus leading to airway inflammation and the symptoms of chronic cough and phlegm production that define chronic bronchitis 20. This would be in keeping with the recently proposed hypothesis of an abnormal acquired immune response in the pathogenesis of COPD 36 and would be supported by the following documented findings. CTLA4 controls a potentially damaging expansion of T-cells by arresting activated T-cells 20. However, this CTLA-mediated anergic check-point can be released by simultaneous stimulation of TCR/CD28 and the polarising cytokines, interleukin (IL)-12 and IL-4, which lead to CD4+ T-cell differentiation towards T-helper cell (Th) type 1 and 2, respectively 20. Th1 CD4+ cells and cytotoxic T-cell-1 CD8+ cells are increased in COPD and chronic bronchitis 13, 15, 16. These cells secrete interferon (IFN)-γ and tumour necrosis factor-α, which activate macrophages. Activated macrophages secrete IL-12, which in turn activates the transcription factor STAT4 37. There is an increased number of activated STAT4 and IFN-γ⁺ cells in bronchial biopsies and bronchoalveolar lavage fluid in patients with COPD, and their increasing expression correlates with decreasing lung function 37. If STAT4 signals in the lung overcome the inhibitory role of CTLA4 38 or the latter is genetically compromised, tissue inflammation and damage may ensue and COPD including chronic bronchitis may develop. It is of interest that one of the SNPs associated with chronic bronchitis (rs231775) is a nonsynonymous coding polymorphism, which results in a change of the amino acid at position 17 from Thr to Ala. This polymorphism is in the signal peptide and is absent from the mature protein. It could affect the efficiency of post-translational modifications other than cleavage, potentially resulting in inefficient processing in the endoplasmic reticulum, leading to reduced surface expression 39. There are several reports showing the functional effect of this polymorphism on T-cell function. Mäurer et al. 40 examined the T-cell response from healthy donors homozygous for either A or G at nucleotide position 49 (amino acid position 17). They found that the G allele affected the CTLA4 downregulation of T-cell activation. In addition, blockade of CTLA4 ligation using soluble anti-CTLA4 monoclonal antibody did not augment proliferation as much in T-cells from G/G-expressing individuals as from A/A-expressing ones. Similar results have also been shown by Sun et al. 41. All these studies show consistent results and it has to be noted that the association that we identified with chronic bronchitis is with the G allele. Thus, this polymorphism might result in reduced expression of CTLA4, which could lead to the development of chronic bronchitis. However, we note that this variant may also be a marker for an un-genotyped variant in tight LD with this variant in the CTLA4 gene.

Strengths and limitations of our study

Our study has several strengths and limitations that deserve comment. Among the former is the fact that we can exclude population stratification because family-based association analyses, as we performed in the ICGN study, are robust against population stratification, and the genomic control methods used in our case–control cohort did not detect significant population stratification. A second strength of our experimental design is that the two study populations were phenotyped for chronic bronchitis using the same questionnaire, thus reducing possible phenotype heterogeneity. Thirdly, we replicated the genetic associations found in the ICGN cohort in an independent population, an approach that is considered the gold standard for demonstrating the relevance of a candidate gene for complex diseases 42, 43. Finally, two SNPs (rs231775 and rs3087243) were still significantly associated with chronic bronchitis after conservative correction for multiple comparisons, thus reinforcing the validity of our observations.

Conversely, our study has a limitation that requires discussion. We did not find significant associations of CTLA4 SNPs with the presence of COPD or the degree of airflow limitation assessed by FEV₁ and FEV₁/FVC. It is possible that the multiple mechanisms that could contribute to airflow obstruction in COPD impaired our ability to find significant associations with COPD; alternatively, it is possible that CTLA4 is a genetic determinant of only chronic bronchitis, and not of airflow obstruction.

In conclusion, we performed a genetic association study and identified variants in the CTLA4 gene that were associated with chronic bronchitis in COPD cases, although not with COPD directly. Although the precise role of CTLA4 in the pathogenesis of COPD including chronic bronchitis is still to be elucidated, our findings open a new avenue for COPD research.