Deviation of gene expression from normal levels has been associated with diseases. Both under- and overexpression of genes could lead to deleterious biological consequences. Dosage balance has been proposed to be a key issue of determining gene expression phenotype. Gene deletion or overexpression of any component in a protein complex produces abnormal phenotypes. As a result, interacting partners should be co-expressed to avoid dosage imbalance effects. The strength of transcriptional co-regulation of interacting partners is supposed to reflect gene dosage sensitivity. Although many cases of dosage imbalance effects have been reported, the molecular attributes determining dosage sensitivity remain unknown. This thesis uses a protein structure analysis protocol to explore the molecular basis of gene dosage sensitivity, and studies the post-transcriptional regulation of dosage sensitive genes.
Solvent-exposed backbone hydrogen bond (SEBH or called as dehydron) provides a structure marker for protein interaction. Protein structure vulnerability, defined as the ratio of SEBHs to the overall number of backbone hydrogen bonds, quantifies the extent to which protein relies on its binding partners to maintain structure integrity. Genes encoding vulnerable proteins need to be highly co-expressed with their interacting partners. Protein structure vulnerability may hence serves as a structure marker for dosage sensitivity. This hypothesis is examined through the integration of gene expression, protein structure and interaction data sets. Both gene co-expression and protein structure vulnerability are calculated for each interacting subunits from human and yeast complexes. It turns out that structure vulnerability quantifies dosage sensitivity for both temporal phases (yeast) and tissue-specific (human) patterns of mRNA expression, determining the extent of co-expression similarity of binding partners.
Highly dosage sensitive genes encode proteins which are vulnerable to water attack. They are subject to tight post-transcriptional regulation. In human, this extra regulation is achieved through extensive microRNA targeting of genes coding for extremely vulnerable proteins. In yeast, on the other hand, our results imply that such a regulation is likely achieved through sequestration of the extremely vulnerable proteins into aggregated states. The 85 genes encoding extremely vulnerable proteins contain the five confirmed yeast prions. It has been proposed that yeast prion protein aggregation could produce multiple phenotypes important for cell survival in some particular circumstances. These results suggest that extremely vulnerable proteins resorting to aggregation to buffer the deleterious consequences of dosage imbalance. However, a rigorous proof will require a structure-based integration of information drawn from the interactome, transcriptome and post-transcriptional regulome.