From genome to genotype : regulation of the genome in glioblastoma multiforme and atrial fibrillation




Hall, Amelia Weber

Journal Title

Journal ISSN

Volume Title



The modern era of genomics has made sequencing a genome nearly routine. Genomics has amassed huge amounts of somatic and disease mutation data, as a result, the character sequence of the human genome has been extensively studied. This information is having an impact on the standard of care in the clinical sphere, with an increasing number of patients and clinicians turning to sequencing data as a determinant of treatment regimen. Knowledge of human protein coding genes and gene expression patterns is extensive, though not absolute. Venturing outside the relatively well-defined protein-coding regions of the genome, much is undetermined. Genome wide association studies (GWAS) have identified many genetic polymorphisms in non-coding regions on the genome that contribute to disease risk. Understanding the mechanisms by which a non-coding polymorphism can cause a phenotype demands an understanding of the physical organization and structure of chromatin in the eukaryotic nucleus. Gene expression data from primary gliolbastoma multiforme tumors (GBM) has uncovered the existence of four molecular subtypes, which affects prognosis and response to treatment. With the goal of gaining an understanding of transcriptional regulation in brain cancer, we profiled post-translational modifications of histone H3 in primary GBM tumors using ChIP-seq, and profiled gene expression in these tumors as well. We used a hidden Markov Model approach to abstract common co-occurrences of histone modifications into chromatin states. We were able to identify signatures consistent with known chromatin regulatory motifs, such as enhancers, and a bivalent state, marked by an active and repressive histone modification. These states regulated expression in a subtype-specific manner, with the proneural subtype showing a protective signature, and the mesenchymal and classical subtypes presenting a signature of invasive cellular migration and angiogenesis. The bivalent and enhancer states controlled a gene expression signature strongly suggestive of glioma stem cells (GSCs), the cells thought to be self-renewing in GBM. As part of profiling gene expression in primary GBMs, we performed RNA-sequencing in primary normal human astrocytes and six GBM-derived commercially available cell lines. We identified widespread differences in expression between tumors and cell lines, as well as a gene interaction network that is common to tumors and cell lines, dominated by chromatin remodelers and Rho guanine exchange factors. Finally, in a pilot study of 400 patients with atrial fibrillation (AF), we identified several SNPs associated with probability of success of cardiac ablation, a surgical therapy for AF. We propose that examining the local topology between a SNP of interest and any long-range contacts will help identify regulatory regions that allow a non-coding SNP to have an effect on gene expression, and thus phenotype


LCSH Subject Headings