163 - Epigenetic landscape of 5-hydroxymethylcytosine and associations with gene expression in placenta.

5-hydroxymethylcystosine (5hmC) is produced through enzymatic oxidation of 5-methylcytosine (5mC). 5hmC is an intermediary in the DNA demethylation pathway, though recent evidence suggests that it acts as a functional epigenetic modification. The landscape of placental 5hmC and its role in gene expression is poorly understood. We hypothesize that 5hmC plays a key role in placental gene expression.




Objective:

We aimed to characterize the distribution of 5hmC across the placenta and identify genomic regions of 5hmC that play a direct role in gene expression.


Design/Methods:

Using 5hmC levels at CG dinucleotide (CpG) probes along with RNA-sequencing (RNA-seq) data, we assessed 5hmC distribution and its relationship with gene expression across 197 placenta samples from participants in the Rhode Island Child Health Study (RICHS). To assess the impact 5hmC has on cis gene expression (CpG within 1 Mb of gene), we implemented a genome-wide expression quantitative trait hydroxymethylation (eQTHM) analysis. We also performed differentially methylated region (DMR) analysis to identify contiguous stretches of CpGs that may regulate expression.

Results:
Although 5hmC levels across the placental epigenome were generally low, we identified ~47,000 loci with consistently elevated (systematic) 5hmC levels. Significant (p < 0.0001) depletion of systematic 5hmC was observed at CpG islands (CGI) and within the first exon, while enrichment was found at probes in “open sea” regions (CpG > 4 Mb from CGI) and within the gene body. eQTHM analysis produced 499 significant (empirical-p < 0.05) cis CpG-gene pairs. Most eQTHMs were located outside their associated gene, and in ”open seas”. eQTHMs most often positively correlated with expression (75.4%). We observed an enrichment of negatively-correlated eQTHMs in CGI and active transcription start sites (TSS), and a depletion in heterochromatic regions. Among positively-correlated pairs, we observed enrichment in enhancers and polycomb repressed regions, along with depletion in regions that flank poised TSS. Finally, we identified 107 DMRs (sidak p < 0.05) across 100 genes.


Conclusion(s):
Collectively, our results suggest that placental 5hmC is enriched in “open sea” and gene body regions and most often positively associated with placental gene expression. This study provides the most complete characterization of the distribution and potential impact of 5hmC in the human placenta and will be useful for future studies of the placental epigenome.