Replication study
We replicated the top findings in whole blood in two independent cohorts. After replication and meta-analysis, one CpG associated with ClinR and another one CpG associated with ComR were replicated (meta P<1.14×10-7). The ClinR-associated CpG (cg13378519) was located on Chr 1 and close to gene PEX11B(Peroxisomal Biogenesis Factor 11 Beta). This CpG was lower methylated in ClinR subjects compared to PersA (Figure 3a). A similar trend was also found in nasal brushes although the association was not statistically significant (Figure 3b). The methylation levels at this CpG also differed between ClinR and healthy subjects in Lifelines cohort (Figure 3c), indicating that DNA methylation status of asthma remission subjects is not equal to that of healthy subjects. The ComR-associated CpG (cg24788483) was located on chr10 and close to gene HABP2(Hyaluronan Binding Protein 2) and TCF7L2 (Transcription Factor 7 Like 2). This CpG was lower methylated in ComR subjects compared to PersA in whole blood, but showed no obvious difference in nasal brushes (Figure S4). In whole blood cis-eQTM analysis, this CpG was significantly negatively correlated to the expression level ofTCF7L2 gene (P=9.12×10-7).
To evaluate potential SNP effects within the probe for the two replicated CpG sites, the β value distributions were visually assessed in the discovery cohorts (Figure S5), and no bimodal distribution was detected indicating no underlying SNP effect within the probe. The regional co-methylation plots of genes annotated to replicated CpG sites are shown in the Figure S6, which showed that cg13378519 is located in the promoter region of PEX11B gene, and cg24788483 is located in heterochromatin region of TCF7L2 gene. We did not identify any SNP that was associated with the two CpG sites using an online database18.
We verified if the results of the two replicated CpG sites were affected by inhaled corticosteroids (ICS), known to affect DNA methylation at multiple sites in the genome (Figure S7). The boxplots showed that the methylation levels in the remission groups (both ClinR and ComR) were significantly lower than both persistent asthma with ICS usage group and persistent asthma without ICS usage group, which indicated that the results were not affected by the use of ICS.
We also compared our results from the discovery cohort with the published DNA methylation study of asthma remission in bronchial biopsies (Vermeulen et al 10). One CpG (cg13525448), located close to LBX1 (Ladybird Homeobox 1) gene and TLX1 (T Cell Leukemia Homeobox 1) gene, in their top list of remission vs persistent asthma reached nominal significance (P<0.05) and showed the same direction of effect in our results of ComR in whole blood (Table S13). Two genes of their DMRs associated with ClinR, at the gene PTCHD3 ( patched domain-containing protein 3) and LOC100507389 , were also present in our results of DMRs associated with ClinR in nasal brushes (Table S4).