Peter Stirling

Mutations in ARID1A, the DNA-binding subunit of the BAF chromatin remodeling complex, occur in about 7% of all cancers. Loss of ARID1A in cancer results in the dysregulation of epigenetic and transcriptional programs, and disturbs normal genome integrity maintenance resulting in genome instability. Genome instability, which consists of a higher frequency of mutation rates, is a hallmark feature of cancer that contributes to initiation and progression of the disease. The accumulation of RNA:DNA hybrid structures in the genome, called R-loops, has become increasingly appreciated as a form of stress that contributes to genome instability.R-loops regulate a variety of processes within the cell, including mitochondrial DNA replication, splicing and gene expression. To provide a better understanding of the breadth of factors that regulate R-loop in human cells, I used machine learning and CRISPR screening approaches to identify R-loop binding and modulating proteins (RLBPs) in human cells. This analysis identified features unique to RLBPs, produced a comprehensive functional network of R-loop associated factors, and identified novel RLBPs in FXR1, LIG1 and ARID1A. Validation work revealed that ARID1A prevents R-loop mediated DNA replication stress via the regulation of TOP2A localization.To gain insight on new therapeutic avenues against ARID1A-deficient cancers, I conducted genome-wide dropout CRISPR screens to identify synthetic lethal partners of ARID1A. I focused on the validation of one of the hits, KEAP1, and confirmed that ARID1A- deficient cells are sensitive to inhibition of KEAP1. I found that inhibition of KEAP1 in ARID1A-KO cells results in an exacerbation of genome instability phenotypes. While the mechanism remains unclear, my data suggests that NRF2-independent functions of KEAP1 are likely responsible for the synthetic lethality between ARID1A and KEAP1.Overall, my results highlight the diversity of proteins that regulate R-loop homeostasis in human cells and uncover a previously unrecognized role for ARID1A in R-loop mitigation and the maintenance of genome integrity. I also document for the first time that cells lacking ARID1A are sensitive to KEAP1 inhibition, resulting in a new body of knowledge on the consequences of ARID1A deficiency and informing on a potentially novel approach for the treatment of such tumours.

View record

Chromatin is involved in all aspects of genome function and its dynamic structure is a key factor in many fundamental cellular processes such as transcription, DNA replication, and DNA repair. A variety of mechanisms regulate chromatin structure including the incorporation of histone variants. H2A.Z, encoded by the non-essential HTZ1 gene in budding yeast, is an evolutionarily conserved histone variant that replaces H2A in 5-10% of nucleosomes in a reaction catalyzed by the SWR1 complex, a conserved ATP-dependent chromatin remodeler. Specifically, H2A.Z is incorporated near centromeres, at the border of heterochromatic domains, and at the majority of transcription start sites. In this dissertation I utilized a combination of genome-wide approaches along with traditional molecular biology techniques to uncover how H2A.Z’s unique identity as a histone variant contributes to gene expression regulation in budding yeast. First, I identified three amino acid regions in the C-terminal half of H2A.Z that can confer specific H2A.Z-identity to H2A. Remarkably, the combination of only 9 amino acid changes, the H2A.Z M6 region, K79 and L81 (two amino acids in the α2-helix), were sufficient to fully recapitulate wild-type growth in genotoxic stress. Furthermore, combining H2A.Z K79 and L81, the M6 region, and the C-terminal tail was sufficient for expression of H2A.Z-dependent heterochromatin-proximal genes and GAL1 derepression. I then determined the impact of H2A.Z incorporation on gene expression during DNA replication stress by generating genome-wide mRNA transcript profiles for wild-type, htz1Δ, swr1Δ, and htz1Δswr1Δ mutants before and after exposure to hydroxyurea (HU). My findings revealed that removing H2A.Z incorporation did not result in mass dysregulation of gene expression, however H2A.Z was required for proper wild-type expression of several HU-induced genes. Collectively my dissertation uncovered the amino acids responsible for H2A.Z-specific identity and explored the importance of this unique identity in gene induction during cellular stress.

View record

Proteostasis is critical for cell survival. Exposure to stressors such as DNA damage, heat, mutations, or ageing can accumulate proteins in non-native conformations. The accumulation of these misfolded proteins, through either gain- or loss-of-function, disrupts vital cellular processes. This is notably evident in neurodegenerative diseases such as Alzheimer’s, and Parkinson's which are linked to protein aggregation pathology. Therefore, cells strive to maintain a healthy proteome which is achieved through a complex network of protein quality control (PQC) circuits. Traditionally, PQC was thought to rely on two major pathways: molecular chaperones for refolding and proteolytic systems for degradation. However, this perspective has been challenged with the discovery of protein sequestration. Under severe stress conditions that overwhelm PQC systems, misfolded proteins are spatially sequestered to specialized and distinct membrane-less inclusions within the cell. This spatial compartmentalization represents a crucial aspect of PQC, facilitating the storage of disassembled or non-native proteins until they can be either refolded or degraded in a regulated manner. The model organism Saccharomyces cerevisiae or budding yeast has three well-established sequestration sites – the Insoluble PrOtein Deposit (IPOD), the JUxtaNuclear Quality control (JUNQ) site and the IntraNuclear Quality control (INQ) site. Although substantial efforts have contributed to identifying key proteins involved in the formation and dissolution of these sequestration sites, our understanding of the regulation and dynamics of INQ remains incomplete.In my thesis, I provide a comprehensive analysis of all proteins that have been found at INQ. Furthermore, I explore different pathways and stressors to expand our knowledge of the factors governing INQ formation. To do so, I first characterize Rpd3, a histone deacetylase, as a novel INQ marker and use its localization as a proxy for INQ levels. Through microscopy and genetic approaches, I demonstrate that INQ formation of Rpd3 is a general response to DNA damage. Furthermore, I explore the role of the post-translational modification (PTM) SUMOylation at INQ and document the DNA damage induced SUMOylation of chaperone Btn2 and Hsp42. Additionally, I engineer a non-SUMOylatable Btn2 and dissect its role in INQ clearance, placing it at the nexus of protein degradation and refolding at INQ.

View record

DNA replication is a vulnerable time for genome stability maintenance. Endogenous add oncogenic stressors can challenge replication by fostering conflicts with transcription and stabilizing DNA:RNA hybrids called R-loops. The capacity for unscheduled R-loops to collapse stalled replication forks and induce genome instability has become apparent. In this work, I provide an overview of the causes and cellular responses to replication stress, detailing the importance of replication checkpoint signaling, replication fork remodelling, and the clinical relevance to cancer development. Recently progress has been made towards identifying R-loop and conflict regulators, however, our understanding of this regulation is incomplete. Building off past findings that utilized a genome-wide interaction screen in budding yeast lacking key R-loop degrading RNase H enzymes, I explored a network of potential R-loop tolerance genes. PCNA is a complex replisome component with over 200 binding partners and was identified as a hit alongside RAD18 and ATAD5 (Elg1 homolog), key factors required for PCNA ubiquitin regulation. Once ubiquitinated, PCNA catalyzes several downstream repair pathways. I validated the roles of these genes in R-loop biology, identifying an accumulation of DNA:RNA hybrids in several human cell lines after gene silencing and in conjugation-incompetent PCNA K164R mutants. In both RAD18- and ATAD5-deficient cells, I observed higher levels of replication stress, DNA damage, and transcription-replication conflicts, dependent on R-loops, transcription, and replication. To gain mechanistic insight into how RAD18 tolerates transcriptional stress, I utilized stressors aphidicolin and pladienolide B to stall replication and disrupt the R-loop landscape respectively. I found that RAD18 localization was dependent on these transcriptional stressors and R-loops. I explored the mechanistic connection between RAD18 and members of the Fanconi anemia (FA) pathway, showing DNA:RNA hybrid epistasis. RAD18 was important for recruitment of FANCD2, the key effector molecule with established roles in resolving R-loop-induced stress, to R-loop prone loci and common fragile sites. These findings expand upon a published connection between RAD18 and the FA pathway by establishing a mechanism involving RAD18 in opposing transcription-associated genome instability through FANCD2 recruitment. The result of my work establishes PCNA ubiquitination as a novel regulatory axis for the tolerance of transcriptional stress.

View record

RNA splicing mutants have been broadly implicated in genome stability, but mechanistic links are often unclear. Two predominant models have emerged: one involving changes in gene expression that perturb other genome maintenance factors and another in which genotoxic DNA:RNA hybrids, called R-loops, impair DNA replication. Recent efforts in whole genome sequencing have identified splicing factor mutations in several cancers, suggesting that splicing disruption may be a common mechanism involved in oncogenesis. To understand how splicing factor mutations contribute to genetic instability (GIN) in budding yeast, I selected strains with mutations in core snRNP complexes involved in establishing the splicing reaction and characterized GIN phenotypes to find that mitotic defects, and in some cases R-loop accumulation, are causes of GIN. I observed evidence of R-loop induced DNA damage in some cases, while all splicing mutants tested caused GIN through aberrant splicing of the TUB1 transcript, the protein product of which, α-tubulin, is critical in forming the mitotic spindle. GIN is exacerbated by loss of the spindle-assembly checkpoint protein Mad1, and moreover, removal of the intron from the α-tubulin gene TUB1 restores genome integrity. To gain functional insights to how HSH155 could influence GIN in the context of cancer progression, I studied five cancer-associated SF3B1 point mutations in the yeast ortholog HSH155. While the splicing activity in Hsh155 and SF3B1 were conserved, I did not observe measurable phenotypes in the yeast mutant strains. Thus, I used isogenic NALM6 human leukemia cell lines to investigate how a specific SF3B1 hotspot mutation, H662Q contributes to GIN. My data indicate that GIN occurs in two ways: 1) by induction of R-loop-mediated replication stress either directly or indirectly through suboptimal expression of an R-loop modulating factor, and 2) aberrant splicing of the multifunctional protein DYNLL1, which may potentially perturb double strand break repair pathway choice. The results of my study show how differing penetrance and selective effects on the transcriptome in yeast and human splicing factors contribute to GIN through R-loop accumulation and altered gene expression, adding to a growing body of evidence that splicing factors play a key role in genome maintenance across species.

View record

Chromosome instability (CIN) is characterized by an increased rate of the unequal distribution of DNA between daughter cells. Such changes in chromosome structure or number can occur due to both mitotic defects leading to aneuploidy and DNA damage-induced chromosome rearrangements. Previous large-scale screens for CIN genes in the model organism Saccharomyces cerevisiae identified DIS3, which encodes a catalytic component of the core RNA exosome complex, as a novel CIN gene. Mutations in human DIS3 have been identified in roughly 11% of multiple myeloma (MM) cases. I sought to recapitulate MM-associated point mutations at conserved sites in yeast cells, in order to understand the mechanism of emergent CIN in MM. I have found that MM-associated DIS3 exonuclease mutations do increase the frequency of CIN. A temperature sensitive DIS3 mutant accumulates DNA:RNA hybrids, however analysis of DNA damage foci by microscopy revealed no increase in double-strand breaks in any of the tested strains. Yeast DIS3 exonuclease mutants experience growth retardation, temperature sensitivity, and an altered cell cycle. Microarray analysis of one MM mutant has additionally demonstrated downregulation of cell cycle components, consistent with the potential for mitotic defects, in addition of upregulation of a host of metabolic pathways. Further, genetic interaction profiling by synthetic genetic array indicates MM-associated DIS3 mutations synthetically interact with rRNA processing proteins, as well as a host of mitotic regulators and metabolic pathways, particularly those involved in spindle and kinetochore function. Further, I verify that DIS3 mutants have a functional spindle assembly checkpoint, and are in fact resistant to microtubule poisons. Finally, I discover that the fitness defects induced by these mutations can be abrogated through culturing on media containing only a non-fermentable carbon source, suggesting that growth on poor carbon sources may also rescue CIN.Together, these results demonstrate extensive phenotypic consequences of MM-associated point mutations in DIS3, and support a model for CIN in DIS3 mutants involving defects in mitotic progression.

View record