U.S. flag

An official website of the United States government

Dot gov

The .gov means it's official.
Federal government websites often end in .gov or .mil. Before sharing sensitive information, make sure you're on a federal government site.

Https

The site is secure.
The https:// ensures that you are connecting to the official website and that any information you provide is encrypted and transmitted securely.

Benchmark Dose Modeling Estimates of the Concentrations of Inorganic Arsenic That Induce Changes to the Neonatal Transcriptome, Proteome, and Epigenome in a Pregnancy Cohort

Julia E. Rager, Scott S. Auerbach, Grace A. Chappell, Elizabeth Martin, Chad M. Thompson, and Rebecca C. Fry
Chemical Research in Toxicology (2017). DOI: https://doi.org/10.1021/acs.chemrestox.7b00221 PMID: 28927277


Publication


Abstract

Prenatal inorganic arsenic (iAs) exposure influences the expression of critical genes and proteins associated with adverse outcomes in newborns, in part through epigenetic mediators. The doses at which these genomic and epigenomic changes occur have yet to be evaluated in the context of dose-response modeling. The goal of the present study was to estimate iAs doses that correspond to changes in transcriptomic, proteomic, epigenomic, and integrated multi-omic signatures in human cord blood through benchmark dose (BMD) modeling. Genome-wide DNA methylation, microRNA expression, mRNA expression, and protein expression levels in cord blood were modeled against total urinary arsenic (U-tAs) levels from pregnant women exposed to varying levels of iAs. Dose-response relationships were modeled in BMDExpress, and BMDs representing 10% response levels were estimated. Overall, DNA methylation changes were estimated to occur at lower exposure concentrations in comparison to other molecular endpoints. Multi-omic module eigengenes were derived through weighted gene co-expression network analysis, representing co-modulated signatures across transcriptomic, proteomic, and epigenomic profiles. One module eigengene was associated with decreased gestational age occurring alongside increased iAs exposure. Genes/proteins within this module eigengene showed enrichment for organismal development, including potassium voltage-gated channel subfamily Q member 1 (KCNQ1), an imprinted gene showing differential methylation and expression in response to iAs. Modeling of this prioritized multi-omic module eigengene resulted in a BMD(BMDL) of 58(45) μg/L U-tAs, which was estimated to correspond to drinking water arsenic concentrations of 51(40) μg/L. Results are in line with epidemiological evidence supporting effects of prenatal iAs occurring at levels <100 μg As/L urine. Together, findings present a variety of BMD measures to estimate doses at which prenatal iAs exposure influences neonatal outcome-relevant transcriptomic, proteomic, and epigenomic profiles.

Figures


Figure 1. BMD model distributions

BMD model distributions of (A) best curve fit models for each -omic endpoint category, (B) BMD estimates, and (C) BMDL estimates. BMD values represent estimated doses causing a 10% shift over the background rate of response.

Figure 2. Module eigengenes representing co-modulated molecules associated with U-tAs.

Module eigengenes representing co-modulated molecules associated with U-tAs across multi-omic signatures.
(A) Correlation matrix showing that module eigengenes (MEs) were highly correlated with several measures of arsenic exposure, including U-tAs (left half). Module eigengene correlations to birth outcomes are also shown (right half).
(B) BMD curve fit for the prioritized module eigengene (“MEgrey”) displaying the eigengene data points grouped by quintiles (red squares) with the best fitting curve plotted as a blue line. BMD estimates are represented by the black vertical lines closest to the right, and BMDL estimates are represented by the black vertical lines closest to the left.

Figure 3. Example molecules of interest showing concerted responses to prenatal iAs exposure.

Example molecules of interest showing concerted responses to prenatal iAs exposure, grouped within the prioritized module eigengene (“MEgrey”). Example molecules are shown that had methylation and/or expression levels that were (A) positively or (B) negatively correlated to MEgrey, a collective measure of methylation and/or expression levels of 984 co-regulated molecules. Methylation and expression levels are Z-score normalized.

Figure 4. Network showing organismal development signaling associated with molecules.

Network showing organismal development signaling associated with molecules within the prioritized multi-omic module eigengene (“MEgrey”). Molecules with iAs-associated changes are colored, and molecules with associated signaling are white.

Tables


Table 1. Characteristics of the Study Cohort

aDifferences in n are based on missing demographic data.
bLOD for DW-iAs = 0.456 μg As/L.

Table 2. Summary of Sample Counts, Quintile Groupings Used for Multi-omic BMD Modeling.

Summary of Sample Counts, Quintile Groupings Used for Multi-omic BMD Modeling, and Number of Molecules Meeting Statistical Criteriaa.
aStatistical criteria were adapted for each analysis to suite the differing platforms and value distributions, thus making the number of genes/proteins identified as significantly associated with U-tAs not directly comparable across molecular end points.
bCpG sites were previously identified by Rojas et al. (2015) and filtered here to include those annotated to known genes.
cmiRNAs were previously identified by Rager et al. (2014).
dmRNAs were previously identified by Rager et al. (2014) and filtered here to include mRNAs annotated to known genes. These 315 mRNAs were represented by 364 probeset IDs.
eProteins were previously identified by Bailey et al. (2014). These 111 proteins were represented by 114 array probes.

Toxicogenomics


Microarray Data

Gene Expression Omnibus (GEO) Series: GSE48355

Supplemental Materials


Supporting Information

Molecular endpoints showing dose-dependent changes associated with urinary total arsenic (U-tAs) and statistical associations to birth outcomes (Table S1);
molecules binned into the prioritized module eigengene of interest (MEgrey) through weighted gene co-expression network analysis (Table S2);
networks constructed with the co-regulated molecules in the prioritized module eigengene (MEgrey) (Table S3);
biological functions and disease signatures associated with co-regulated molecules in the prioritized module eigengene (MEgrey) (Table S4)