Mutations in the SMARCA4/BRG1 gene resulting in complete loss of its protein (BRG1) occur frequently in non-small cell lung cancer (NSCLC) cells. Currently, no single therapeutic agent has been identified as synthetically lethal with SMARCA4/BRG1 loss. We identify AURKA activity as essential in NSCLC cells lacking SMARCA4/BRG1. In these cells, RNAi-mediated depletion or chemical inhibition of AURKA induces apoptosis and cell death in vitro and in xenograft mouse models. Disc large homologue-associated protein 5 (HURP/DLGAP5), required for AURKA-dependent, centrosome-independent mitotic spindle assembly is essential for the survival and proliferation of SMARCA4/BRG1 mutant but not of SMARCA4/BRG1 wild-type cells. AURKA inhibitors may provide a therapeutic strategy for biomarker-driven clinical studies to treat the NSCLCs harbouring SMARCA4/BRG1-inactivating mutations.

Polyploid giant cancer cells (PGCC) are common in tumors and have been associated with resistance to cancer therapy, tumor relapse, malignancy, immunosuppression, metastasis, cancer stem cell production, and modulation of the tumor microenvironment. However, the molecular mechanisms that cause these cells to form are not yet known. In this study, we discover that Aurora kinases are synergistic determinants of a switch from the proliferative cell cycle to polyploid growth and multinucleation in lung cancer cell lines. When Aurora kinases were inhibited together, lung cancer cells uniformly grew into multinucleated PGCCs. These cells adopted an endoreplication in which the genome replicates, mitosis is omitted, and cells grow in size. Consequently, such cells continued to safely grow in the presence of antimitotic agents. These PGCC re-entered the proliferative cell cycle and grew in cell number when treatment was terminated. Thus, PGCC formation might represent a fundamental cellular response to Aurora kinase inhibitors and contributes to therapy resistance or tumor relapse. Significance: These findings provide a novel insight about how cancer cells respond to Aurora kinase inhibitors and identify a new mechanism responsible for resistance to these agents and other antimitotic drugs.

SW044248, identified through a screen for chemicals that are selectively toxic for NSCLC cell lines, was found to rapidly inhibit macromolecular synthesis in sensitive, but not in insensitive cells. SW044248 killed approximately 15% of a panel of 74 NSCLC cell lines and was non-toxic to immortalized human bronchial cell lines. The acute transcriptional response to SW044248 in sensitive HCC4017 cells correlated significantly with inhibitors of topoisomerases and SW044248 inhibited topoisomerase 1 (Top1) but not topoisomerase 2. SW044248 inhibited Top1 differently than camptothecin and camptothecin did not show the same selective toxicity as SW044248. Elimination of Top1 by siRNA partially protected cells from SW044248, although removing Top1 was itself eventually toxic. Cells resistant to SW044248 responded to the compound by upregulating CDKN1A and siRNA to CDKN1A sensitized those cells to SW044248. Thus, at least part of the differential sensitivity of NSCLC cells to SW044248 is the ability to upregulate CDKN1A.

Polyploid giant cancer (PGCC) state is a common response of cancer cells to various stressors including chemotherapy, irradiation, hypoxia and viral infection. Upon stress, PGCCs adopt an endoreplication in which the genome replicates, mitosis is omitted, and cells grow in size, leading to resistance. How accessible endoreplication is to a cell, therefore, directly translates to its resistance to therapy. In this study, we hypothesized that endoreplication and PGCC state are more plausible to cancer cells that already have a higher ploidy content prior to therapy. To test this hypothesis, we developed a comprehensive three-tier framework consisting of i) computational, ii) in vitro and iii) in silico components. We designed a set of ordinary differential equations (ODEs) to model how cells enter and exit the PGCC state in response to a given stressor. We engineered how this in silico approach interacts with in vitro experiments into a broadly applicable software solution called CLONEID. The software uses computer vision to monitor phenotypic changes in cell and nuclear size from standard bright-field microscopy and classify cells into PGCC and non-PGCC states. We used CLONEID to test various therapeutic agents for their ability to select for a stable near-tetraploid population in a set of near-diploid cell lines. Through spontaneous cell fusions, we also obtained tetraploid breast cancer cells matched with their parental lines. Altogether, this framework, now, enables us i) to monitor the PGCC state experimentally ii) in matched cell lines with differential DNA content iii) to model the successful entrance and exit rates to and from the endoreplication state, respectively. As the first application of this framework, we tested the ability of our matched triple-negative breast cancer (TNBC) lines (SUM-159 and MDA-MB-231) to access the PGCC state upon treatment with 18 commonly-used chemotherapy agents. We observed that gemcitabine caused continued cell growth without cell division in both tetraploid SUM-159 and MDA-MB-231 cells whereas near-diploid parental cells were hypersensitive to the treatment. Consequently, tetraploid cancer cells continued to safely grow in the presence of gemcitabine. Furthermore, these PGCCs re-entered the proliferative cell cycle and grew in cell number when treatment is terminated. Gemcitabine (GEM)-based chemotherapy is a standard treatment for patients with TNBC although its efficacy is limited mainly due to drug resistance. Moreover, such poor response rate is coupled with severe side effects that frequently leads to the failure of vital organs, serious secondary diseases, and co-morbidities. Our findings suggest that ploidy is a predictive biomarker for gemcitabine sensitivity. Thus, we expect our findings and three-component framework strategy to help stratify the TNBC patient population by their response to gemcitabine. In addition, our mathematical modelling approach has the promising potential to inform personalized dose optimization and to effectively decrease administered gemcitabine dose for a subset of patients, which would alleviate therapy-associated side effects and lethalities. Citation Format: Vural Tagal, Jackson P. Cole, Daria Miroshnychenko, Andriy Marusyk, Noemi Andor. Ploidy as a predictive biomarker for gemcitabine sensitivity in triple-negative breast cancers [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2023; Part 2 (Clinical Trials and Late-Breaking Research); 2023 Apr 14-19; Orlando, FL. Philadelphia (PA): AACR; Cancer Res 2023;83(8_Suppl):Abstract nr LB175.