Figure 1.
Relationships between DNA methylation and gene expression.
A) Cis regulation was defined by the correlation of the methylation level at the promoter region of a gene with expression level of the gene. B) Trans regulation was defined by the correlation of a methylation level at the promoter region of a gene with expression level of other genes. C) Potential relationships between cis and trans regulations. There are two potential causal mechanisms of cis and trans connections: Model I, where the methylation level regulates trans gene expression via the cis gene expression, and Model II, where Trans gene expression regulates the cis gene via controlling its methylation level. It is also possible that cis and trans connections are independently regulated by a factor X.
Figure 2.
The causality test of trans methyl-mRNA pairs.
A) and B) are causality test results for the causal model whereby methylation regulates trans gene expression (methylation → cis gene expression
→trans gene expression
) in control and COPD data sets, respectively. The Y-axis is the –log10 of the p-values for the Spearman correlation between
and
and the X-axis is –log10 of the p-values for the Spearman correlation between
and
. A causal relationship (methylation
→ cis gene expression
→trans gene expression
) was defined if the p-value of
was <0.0001 and the p-value of
was>0.01 (see Methods for details). A total of 30,177 and 362,095 causal pairs were inferred in control and COPD samples, respectively. C) and D) are the causality test results for the causal model whereby trans gene expression regulates methylation variation (trans gene expression
→methylation
→ cis gene expression
) in control and COPD data sets, respectively. The Y-axis is the –log10 of the p-values for the Spearman correlation
and the X-axis is –log10 of the p-values for the Spearman correlation
. A causal relationship (trans gene expression
→methylation
→ cis gene expression
) was defined if the p-value of
was <0.0001 and the p-value of
was>0.01 (see Methods for details). A total of 1,241 and 19,173 causal pairs were inferred in control and COPD, respectively.
Figure 3.
The numbers of downstream genes regulated by DNA methylation level variation follow a scale-free distribution (a linear relationship in log-log plots).
A) The numbers in control; B) The numbers in COPD.
Figure 4.
Comparing characteristics of key regulators with 5 COPD severity related traits in LGRC.
A) Comparing lung DNA methylation profiles of key regulators with 5 COPD severity related traits by Spearman correlation. At the Fisher's exact test p-value <0.05, the DNA methylation level variations of 3 key regulators, ACSF3, SELO, and EPAS1, were correlated with all 5 COPD severity related traits. B) Comparing downstream genes of key regulators with gene signature sets for 5 COPD severity related traits by the hypergeometric test. At the Fisher’s exact test p-value<0.05, only the key regulator EPAS1's downstream genes significantly overlapped with gene signature sets for all 5 COPD severity related traits.
Figure 5.
Gene expression levels of Epas1 and Vegfa were lower in chronic smoking mice than non-smoking age-matched mice at the time when COPD develops in different mouse models.
A) Gene expression levels of Epas1 and Vegfa in C57BL/6J mice that develop COPD after 6 months chronic exposure to cigarette smoke. B) Gene expression levels of Epas1 and Vegfa in A/J mice that develop COPD after 2 months chronic exposure to cigarette smoke. The t-test was used to compare Epas1 or Vegfa expression levels in mice with or without chronic smoke exposure.
Table 1.
EPAS1 siRNA signatures in human and mouse endothelial cells overlap with multiple COPD disease severity related signature sets.