Computational Prediction And Experimental Validation Of Novel Mouse Imprinted Genes
Epigenetic modifications, including DNA methylation and covalent modifications to histone tails, are major contributors to the regulation of gene expression. These changes are reversible, yet can be stably inherited, and may last for multiple generations without change to the underlying DNA sequence. Genomic imprinting results in expression from one of the two parental alleles and is one example of epigenetic control of gene expression. So far, 60 to 100 imprinted genes have been identified in the human and mouse genomes, respectively. Identification of additional imprinted genes has become increasingly important with the realization that imprinting defects are associated with complex disorders ranging from obesity to diabetes and behavioral disorders. Despite the importance imprinted genes play in human health, few studies have undertaken genome-wide searches for new imprinted genes. These have used empirical approaches, with some success. However, computational prediction of novel imprinted genes has recently come to the forefront. I have developed generalized linear models using data on a variety of sequence and epigenetic features within a training set of known imprinted genes. The resulting models were used to predict novel imprinted genes in the mouse genome. After imposing a stringency threshold, I compiled an initial candidate list of 155 genes. A subset of these genes was tested for evidence of imprinting using allele-specific restriction digests in either brain or placenta. Of the 10 genes tested in placenta, 2 showed evidence of maternal allele-specific expression. I also designed a custom microarray to test a total of 563 genes predicted as imprinted at lower stringency levels. Of these 563 genes, I experimentally tested 32 in placenta and 8 in brain, resulting in the identification of an additional 5 novel imprinted genes in placenta. This study is the first to demonstrate the utility of epigenetic marks in the prediction of imprinted genes. Furthermore, specific combinations of epigenetic marks were commonly found within particular regions relative to the transcriptional start sites of imprinted genes, implicating their placement and localization in the imprinting mechanism.
dissertation or thesis