Mining Novel Candidate Imprinted Genes Using Genome-Wide Methylation Screening and Literature Review

Large-scale transcriptome and methylome data analyses obtained by high-throughput technologies have been enabling the identification of novel imprinted genes. We investigated genome-wide DNA methylation patterns in multiple human tissues, using a high-resolution microarray to uncover hemimethylated CpGs located in promoters overlapping CpG islands, aiming to identify novel candidate imprinted genes. Using our approach, we recovered ~30% of the known human imprinted genes, and a further 168 candidates were identified, 61 of which with at least three hemimethylated CpGs shared by more than two tissue types. Thirty-four of these candidate genes are members of the protocadherin cluster on 5q31.3; in mice, protocadherin genes have non-imprinted random monoallelic expression, which might also be the case in humans. Among the remaining 27 genes, ZNF331 was recently validated as an imprinted gene, and six of them have been reported as candidates, supporting our prediction. Five candidates (CCDC166, ARC, PLEC, TONSL, and VPS28) map to 8q24.3, and might constitute a novel imprinted cluster. Additionally, we performed a comprehensive compilation of known human and mice imprinted genes from literature and databases, and a comparison among high-throughput imprinting studies in humans. The screening for hemimethylated CpGs shared by multiple human tissues, together with the extensive review, appears to be a useful approach to reveal candidate imprinted genes.