Differential DNA methylation marks and gene comethylation of COPD in African-Americans with COPD exacerbations

Abstract

Background: Chronic obstructive pulmonary disease (COPD) is the third-leading cause of death worldwide. Identifying COPD-associated DNA methylation marks in African-Americans may contribute to our understanding of racial disparities in COPD susceptibility. We determined differentially methylated genes and co-methylation network modules associated with COPD in African-Americans recruited during exacerbations of COPD and smoking controls from the Pennsylvania Study of Chronic Obstructive Pulmonary Exacerbations (PA-SCOPE) cohort.

Methods: We assessed DNA methylation from whole blood samples in 362 African-American smokers in the PA-SCOPE cohort using the Illumina Infinium HumanMethylation27 BeadChip Array. Final analysis included 19302 CpG probes annotated to the nearest gene transcript after quality control. We tested methylation associations with COPD case-control status using mixed linear models. Weighted gene comethylation networks were constructed using weighted gene coexpression network analysis (WGCNA) and network modules were analyzed for association with COPD.

Results: There were five differentially methylated CpG probes significantly associated with COPD among African-Americans at an FDR less than 5 %, and seven additional probes that approached significance at an FDR less than 10 %. The top ranked gene association was MAML1, which has been shown to affect NOTCH-dependent angiogenesis in murine lung. Network modeling yielded the "yellow" and "blue" comethylation modules which were significantly associated with COPD (p-value 4 × 10-10 and 4 × 10-9, respectively). The yellow module was enriched for gene sets related to inflammatory pathways known to be relevant to COPD. The blue module contained the top ranked genes in the concurrent differential methylation analysis (FXYD1/LGI4, gene significance p-value 1.2 × 10-26; MAML1, p-value 2.0 × 10-26; CD72, p-value 2.1 × 10-25; and LPO, p-value 7.2 × 10-25), and was significantly associated with lung development processes in Gene Ontology gene-set enrichment analysis.

Conclusion: We identified 12 differentially methylated CpG sites associated with COPD that mapped to biologically plausible genes. Network module comethylation patterns have identified candidate genes that may be contributing to racial differences in COPD susceptibility and severity. COPD-associated comethylation modules contained genes previously associated with lung disease and inflammation and recapitulated known COPD-associated genes. The genes implicated by differential methylation and WGCNA analysis may provide mechanistic targets contributing to COPD susceptibility, exacerbations, and outcomes among African-Americans.

Trial registration: Trial Registration: NCT00774176 , Registry: ClinicalTrials.gov, URL: www.clinicaltrials.gov , Date of Enrollment of First Participant: June 2004, Date Registered: 04 January 2008 (retrospectively registered).

Keywords: Chronic obstructive pulmonary disease; DNA methylation; Microarray; Smoking; Weighted gene coexpression network analysis.

Figures

Fig. 1

Subject- and Probe-Level Quality Control Chart. Quality control of the PA-SCOPE methylation dataset included Probe-level controls and Subject-level quality controls (see Methods for details). Final analysis included 93 COPD cases and 269 smoking controls. SNP-Under-Probe refers to probes containing a CpG within 5 base pairs upstream or downstream of a known genomic SNP. Repeat-Under-Probe refers to probes that mapped to genomic repeat regions

Fig. 2

Differentially Methylated CpG Probes Associated with COPD. Differential methylation analysis revealed 12 CpG sites in 12 genes significantly associated with COPD with an FDR-corrected p-value less than 0.10. Difference in mean percent methylation represents the difference in mean methylation between COPD cases and smoking controls. The y-axis represents the negative log of the association p-value from linear mixed models adjusted for age, gender, pack years of smoking, assay batch, and cell type. The name of the nearest gene is included with each of the top five CpG results

Fig. 3

Manhattan Plot of Differential Methylation Analysis Results. Differential methylation analysis results presented by chromosomal location (x-axis). The y-axis represents the negative log of the association p-value from linear mixed models adjusted for age, gender, pack years of smoking, assay batch, and cell type. The name of the nearest gene is included with each of the top five CpG results

Fig. 4

WCGNA Module Trait Relationship Heatmap. Heatmap showing comethylation module correlation with phenotypic trait and associated p-value for these correlations within PA-SCOPE. Positive or negative correlation magnitude with COPD is presented with p-value for the correlation with COPD in parenthesis. The yellow and blue modules were both significantly associated with COPD affection status, labeled “COPD”