Understanding dominant-negative states of the cohesin complex

Genome instability is an enabling characteristic of cancer cells that also presents a therapeutic opportunity. The cohesin multi-protein complex plays essential roles in sister chromatid cohesion, chromatin organization, and DNA damage repair and is mutated in several cancer types. In this thesis, I evaluate the cohesin complex as a potential anti-cancer target. Cohesin is a challenging therapeutic target because it is essential for viability. However, if a portion of cohesin is trapped on chromatin, it could result in a dominant negative effect that is lethal in cancer cells but tolerated in healthy cells. This rationale is based on the observation that PARP inhibitors and topoisomerase poisons trap their targets on DNA, producing a genotoxic gain-of-function effect that can kill cancer cells with DNA replication or repair defects. The potential of disrupting cohesin dynamics as anti-cancer approach was observed while attempting to “humanize” the cohesin complex in the yeast, Saccharomyces cerevisiae. Heterologous expression of human cohesin subunits in wildtype yeast cells results in slow growth, cohesion dysregulation, sensitivity to DNA-damaging agents and synthetic lethality with genome instability genes. These effects were caused by yeast–human hybrid cohesin complexes loading on chromatin. I hypothesized that individual cohesin missense mutations could elicit a similar phenotype and selectively kill cells exhibiting genomic instability. I generated and screened a random point mutagenesis library of yeast Smc1 for proteoforms that cause dominant synthetic lethality in cohesin-compromised cells. Mapping the specific amino acid substitutions onto the three-dimensional structure of Smc1 revealed mutations clustering in the ATPase domain. Characterization of these mutations showed that they cause mutant Smc1–Smc3 dimers to load onto DNA, leading to replication stress and dominant synthetic lethality in genetic backgrounds analogous to certain tumour types. This work provides mechanistic insight into dominant-negative proteoforms of cohesin and establishes a molecular scale map and platform to inform in silico docking and small-molecule discovery efforts aimed at phenocopying cohesin trapping. The data and analysis in this thesis demonstrate the power of dominant phenotypic screens to model potential therapeutics and provides a roadmap for future work developing cohesin as an anti-cancer therapeutic target.