Differential DNA methylation in umbilical cord blood of infants exposed to low levels of arsenic in utero

Abstract

Background: There is increasing epidemiologic evidence that arsenic exposure in utero, even at low levels found throughout much of the world, is associated with adverse reproductive outcomes and may contribute to long-term health effects. Animal models, in vitro studies, and human cancer data suggest that arsenic may induce epigenetic alterations, specifically by altering patterns of DNA methylation.

Objectives: In this study we aimed to identify differences in DNA methylation in cord blood samples of infants with in utero, low-level arsenic exposure.

Methods: DNA methylation of cord-blood derived DNA from 134 infants involved in a prospective birth cohort in New Hampshire was profiled using the Illumina Infinium Methylation450K array. In utero arsenic exposure was estimated using maternal urine samples collected at 24-28 weeks gestation. We used a novel cell mixture deconvolution methodology for examining the association between inferred white blood cell mixtures in infant cord blood and in utero arsenic exposure; we also examined the association between methylation at individual CpG loci and arsenic exposure levels.

Results: We found an association between urinary inorganic arsenic concentration and the estimated proportion of CD8+ T lymphocytes (1.18; 95% CI: 0.12, 2.23). Among the top 100 CpG loci with the lowest p-values based on their association with urinary arsenic levels, there was a statistically significant enrichment of these loci in CpG islands (p = 0.009). Of those in CpG islands (n = 44), most (75%) exhibited higher methylation levels in the highest exposed group compared with the lowest exposed group. Also, several CpG loci exhibited a linear dose-dependent relationship between methylation and arsenic exposure.

Conclusions: Our findings suggest that in utero exposure to low levels of arsenic may affect the epigenome. Long-term follow-up is planned to determine whether the observed changes are associated with health outcomes.

Figure 1

Locus-by-locus examination of differentially methylated CpG sites based on U-As levels. (A) Manhattan plot for total U-As, where points represent the -log₁₀ (p-value) testing the null hypothesis of no difference in methylation across quartiles of arsenic exposure, adjusted for maternal age at delivery, infant sex, and urinary creatinine levels. Red and blue lines indicate –log₁₀(1 × 10^–4) and –log₁₀(0.05), respectively. (B) Location of the top 100 CpGs associated with U-As on the basis of p-value (top 100 probes) compared with all CpGs on the methylation array (all probes). N. shore and N. shelf refer to CpG island shore and shelf regions, respectively, that are upstream of a CpG island region. S. shore and S. shelf refer to CpG island shore and shelf regions, respectively, that are downstream of a CpG island region.

Figure 2

Crude plots of DNA methylation for cg08884395 (A) and cg27514608 (B) by quartiles of total U-As for the 134 study subjects. Red lines denote the within-quartile median methylation status.