PUBLICATIONS
Below is a selection of highlight publications, for a full publication list visit ORCID
Our drug-sensitivity screen of 16 patient-derived ovarian cancer models (OCMs) unexpectedly found that inhibitors targeting the replication stress response (CHK1i, WEE1i, ATRi, PARGi) are not interchangeable. Therefore, to overcome redundancy in the replication stress response, we screened the OCMs with all two-, three- and four-drug combinations in a multiple-low-dose strategy. Low-dose CHK1i-ATRi was synergistic and had a potent anti-proliferative effect on 15 of the OCMs, with the potential to minimise treatment resistance and toxicity. This study therefore demonstrates the potential of the living biobank of OCMs as a drug discovery platform for HGSOC.
Winner of the journal’s 2021 outstanding paper prize! Here we modelled HR-deficient HGSOC using non-transformed human fallopian tube epithelial cells – the likely HGSOC cell-of-origin. We sequentially mutated TP53 and BRCA1 then overexpressed MYC. Loss of TP53 was sufficient to deregulate multiple cell cycle control networks and thereby initiate chromosome instability. Both transcriptional deregulation and karyotype diversity were dramatically exacerbated by loss of BRCA1 function, with whole-genome doubling events observed in independent p53/BRCA1-deficient lineages.
Expanding our research on PARG inhibition, 10 ovarian cancer cell lines and 32 patient-derived ovarian cancer models (OCMs) were screened for sensitivity to PARG inhibitor monotherapy. Four cell lines and seven OCMs were found to be sensitive, with sensitivity accompanied by markers of persistent DNA replication stress – in line with an inherent DNA replication vulnerability. A similar proportion of OCMs exhibited PARP and PARG inhibitor sensitivity. Importantly, four of the OCMs that were sensitive to PARG inhibition were resistant to the PARP inhibitor, consistent with sensitivity to these two agents being mostly non-overlapping.
This analysis finds that applying non-negative matrix factorisation (NMF) to the transcriptional profiles of 44 commonly used ovarian cancer cell lines exquisitely clusters them into five distinct classes, representative of the five main subtypes of epithelial ovarian cancer. This robust classification now informs the selection of the most appropriate pre-clinical models for all five histotypes. We also used machine learning to train a ‘transcriptional classifier’ using the subtype-specific metagenes defined by NMF. Testing of the classifier on our “living biobank” of ovarian cancer models demonstrated the future potential of such a classifier for subtyping new models, especially if clinical diagnosis is unavailable or uncertain.
Here, we first describe our “living biobank” of ovarian cancer models. Fifteen models are characterised by p53 profiling, exome sequencing and transcriptomics, and karyotyped using single-cell whole-genome sequencing. This proof-of-concept study demonstrates that our clinically annotated ex vivo cultures recapitulate the features of ovarian cancers. Drug profiling reveals cisplatin sensitivities consistent with patient responses, demonstrating that our workflow has the potential to generate personalized avatars with advantages over current pre-clinical models and the potential to guide clinical decision making.
Here, we show that deregulating MYC modulates multiple aspects of mitotic chromosome segregation. Cells overexpressing MYC have altered spindle morphology, take longer to align their chromosomes at metaphase and enter anaphase sooner. Proteomic analysis showed that MYC modulates multiple networks predicted to influence mitosis. This proteomic screen also identified MYC-modulated SUMO signalling components, namely UBC9 and RANBP2.
This is our first evaluation of a PARG inhibitor (PDD00017273; PARGi). PARGi sensitivity was accompanied by pan-nuclear gH2AX staining, indicating replication catastrophe. A “synthetic lethal” siRNA screen identified DNA replication genes, in particular TIMELESS, further supporting the role of replication catastrophe. While four ovarian cancer cell lines proliferated in both PARP and PARG inhibitors, Kuramochi proliferation was suppressed by PARG inhibition and OVCAR3 proliferation was suppressed by PARP inhibition. This differential sensitivity indicates that PARG inhibitors present a new opportunity to treat HR-proficient ovarian cancers not expected to respond to PARP inhibitors.
Here we show that chromosome mis-segregation induces metabolic collapse and apoptosis, mediated by the p38 stress response kinase. Inhibiting p38 elevates Hif-1a, boosts glycolysis, and limits metabolic collapse, in turn allowing expansion of aneuploid clones. Because hypoxia and aneuploidy are both barriers to tumour progression, the ability of Hif-1α to promote cell survival following chromosome mis-segregation raised the possibility that aneuploidy tolerance coevolves with adaptation to hypoxia.
This work provides a mechanistic rationale explaining why pharmacological inhibition of Bcl-xL synergizes with paclitaxel. The selective Bcl-xL inhibitor WEHI-539 sensitised cells to mitotic blockers, such as paclitaxel and nocodazole, and second generation anti-mitotics targeting Eg5, Cenp-E, and Plk1, but not to agents that accelerate mitotic progression, e.g. Aurora A, Aurora B and Mps1 inhibitors. This differential is due to compensation by pro-survival Mcl-1; during a mitotic delay induced by mitotic blockers, Mcl-1 is degraded such that survival becomes critically dependent on Bcl-xL. By contrast, when cells are driven through mitosis, Mcl-1 remains intact, sustaining viability despite inhibition of Bcl-xL.
We used our competing networks model to design a genome-wide siRNA screen to dissect the death in mitosis pathway in response to antimitotic drugs. We screened for siRNAs that enhanced viability in response to paclitaxel and subsequently identified MYC as a major determinant of mitotic cell fate. We found that MYC sensitizes cancer cells to mitotic blockers and mitotic drivers by regulating the expression of an apoptotic network. MYC primes paclitaxel-induced apoptosis by suppressing pro-survival Bcl-xL and upregulating the pro-death factors Bim, Bid and Noxa.
In this study, we characterised a novel Mps1 inhibitor, AZ3146, and used it to probe the role of the catalytic activity of Msp1 during mitosis. From this work we were able to propose a model whereby Mps1 transphosphorylation is required to recruit Mad2 and CENP-E to unattached kinetochores. This subsequently contributed to defining the important role of Msp1 in activation of the spindle assembly checkpoint, and provided the field with a useful tool to dissect the checkpoint.
Here we describe our pivotal competing networks model, whereby mitotic cell fate is dictated by two networks, one involving a “death in mitosis” pathway by caspase activation, the second protecting cyclin B1 from degradation. We used a high-throughput automated time-lapse light microscopy approach to systematically analyse over 10,000 single cells from 15 cell lines in response to three different classes of antimitotic drug. By developing cell fate profiles to present the complex data in a visually intuitive manner, we were able to test a number of hypotheses. This resulting model has provided the field with an intellectual framework to further dissect how an abnormal mitosis induces apoptosis.
Here we characterise the first selective Aurora B inhibitor, ZM447439. We showed that Aurora B activity was not required for chromosome condensation or the ability of kinetochores to bind microtubules. Rather, we discovered that Aurora B promotes correct kinetochore-microtubule interactions, and by recruiting Bub1 and BubR1 to kinetochores, is an upstream component of the checkpoint. This publication not only provided novel insight into fundamental principles of mitotic regulation but also provided the community with a useful research tool.
By searching for homologues of yeast genes, this research discovered three mammalian checkpoint components, Bub1, BubR1 and Bub3, and showed that Bub1 and BubR1 localise to kinetochores via binding Bub3. Using a dominant-negative, it was found that compromising Bub1 function not only bypassed the checkpoint but also accelerated progress through an unperturbed mitosis. Moreover, it fundamentally changed post-mitotic behaviour following spindle damage: whereas checkpoint-proficient cells died, checkpoint-deficient cells survived. These observations opened the checkpoint field and also implicated the checkpoint as a key mediator of mitotic cell fate.