Smoking-associated AHRR demethylation in cord blood DNA: impact of CD235a+ nucleated red blood cells

BACKGROUND: Numerous studies have demonstrated that DNA methylation levels in the aryl hydrocarbon receptor repressor (AHRR) gene measured in cord blood are significantly associated with prenatal tobacco smoke exposure and can be used as a fetal exposure biomarker. The mechanism driving this demethylation has not been determined and it is unclear if all cord blood cell types are impacted. Nucleated red blood cells (nRBCs/CD235a+ cells) are developmentally immature RBCs that display genome-wide hypomethylation and are observed at increased frequency in the cord blood of smoking mothers. We tested if AHRR methylation levels in CD235a+ nRBCs or nRBC counts influenced AHRR methylation in whole cord blood. METHODS: Cord blood was collected from smoking (n = 34) and nonsmoking (n = 19) mothers and DNA was prepared from whole cord blood, isolated CD235a+ nRBCs, and CD14+ monocytes. AHRR methylation in cord blood DNA was measured using Illumina 850K arrays (cg05575921, chr5:373378). Pyrosequencing was used to compare methylation levels among cord blood, CD235a+, and CD14+ cells. We measured nRBC percentages using conventional complete blood counts and estimated percent nRBCs by a deconvolution model. RESULTS: Methylation levels in AHRR were significantly lower in nRBCs relative to whole cord blood and CD14+ monocytes. While AHRR methylation levels in the cell types were significantly correlated across all subjects, methylation values at the chr5:373378 CpG averaged 14.6% lower in nRBCs (range 0.4 to 24.8%; p = 3.8E−13) relative to CD14+, with nonsmokers showing a significantly greater hypomethylation (− 4.1%, p = 1.8E−02). Methylation level at the AHRR chr5:373378 CpG was strongly associated with self-reported smoking in both CD14+ monocytes (t test p = 5.7E−09) and nRBCs (p = 4.8E−08), as well as cotinine levels (regression p = 1.1E−07 and p = 3.6E−04, respectively). For subjects with whole blood 850K data, robust linear regression models adjusting for estimated cell type composition, either including nRBCs counts or estimates, modestly increased the association between smoking and cg05575921 methylation. CONCLUSIONS: Prenatal smoke exposure was highly significantly associated with AHRR methylation in cord blood, CD14+ monocytes, and CD235a+ nRBCs. AHRR methylation levels in nRBCs and nRBC counts had minimal effect on cord blood methylation measurements. However, regression models using estimated nRBCs or actual nRBC counts outperformed those lacking these covariates. ELECTRONIC SUPPLEMENTARY MATERIAL: The online version of this article (10.1186/s13148-019-0686-1) contains supplementary material, which is available to authorized users.