Synthesis of mRNA in eukaryotes involves the coordinated action of many enzymatic processes, including initiation, capping, elongation, splicing, cleavage, and poly-adenylation. These processes are carried out by mega-dalton complexes which assemble and disassemble within the dynamic milieu of the nucleus. Due to the stochastic nature of biochemical reactions, individual cells exhibit a range of possible responses, resulting in heterogeneous or 'noisy' gene expression. By imaging the synthesis and processing of single RNAs in living cells, one can shed light on the underlying molecular behavior and also gain insight into the effects of noise on disease progression. In this talk, I will discuss how stochastic transcription and RNA synthesis can lead to cancer-related phenotypes. In recent years, mutations in the general splicing machinery have appeared in numerous malignancies. In particular, the heterozygous missense mutation in the 3' splice site recognition factor U2AF1 (S34F) correlates with a poorer prognosis and displays the genetic signature of an oncogene. We demonstrate that this mutation is a gain of function mutation which acts in a dominant manner to alter splicing and transcription kinetics. At the physiological level, the mutation results in a subset of cells which are resistant to DNA damage from genotoxic treatments. However, this resistance only occurs in a subset of cells. This subset is determined by the stochastic nature of expression and splicing activity of the mutant allele, which alters a small cohort of isoforms related to the DNA damage response.

Nature exploits many protein folds, but we have only recently begun to understand the consequences of amino acid changes in a utilitarian manner. The conservation and co-variation of residues across homologous protein sequences throws light on the natural strategies employed in protein stability, and we can use statistical measures as a means to understand them. We have adapted relative entropy and mutual information metrics to reliably predict stabilizing mutations, and have used combinatorial approaches to further our understanding of the meanings of amino acid consensus and correlations. Previously, a consensus variant of TIM (cTIM) suffered overall destabilization and impaired activity. However, a curated database of sequences yielded a variant with a native-like fold and activity (ccTIM). Since the variants differed minimally at the sequence level, we suspected the scrambling of statistical correlations in cTIM and developed combinatorial libraries to interrogate the exact nature of the differences between cTIM and ccTIM. Preliminary data points to global stabilization of the protein fold as a basis for the rescue of cTIM. We have developed a consensus-screening algorithm that has been applied to candidate TIM sequences from different branches of phylogeny in an attempt to exploit amino acid conservation & co-variation. Our results show a ~90% probability in identifying stabilizing mutations, with successful application to proteins with different folds. While we have begun to make rational mutations in ccTIM in an attempt to understand stability from the angle of destabilization of the fold, we hope to use positional binomial libraries and DNA-shuffling to further our knowledge of protein stability in this context.