Despite intense efforts, the cure rates of childhood and adult solid tumors are not satisfactory. Resistance to intensive chemotherapy is common, and targets for molecular therapies are largely undefined. We have found that the majority of childhood solid tumors, including rhabdoid tumors, neuroblastoma, medulloblastoma, and Ewing sarcoma, express an active DNA transposase, PGBD5, that can promote site-specific genomic rearrangements in human cells. Using functional genetic approaches, we discovered that mouse and human cells deficient in nonhomologous end joining (NHEJ) DNA repair cannot tolerate the expression of PGBD5. In a chemical screen of DNA damage signaling inhibitors, we identified AZD6738 as a specific sensitizer of PGBD5-dependent DNA damage and apoptosis. We found that expression of PGBD5, but not its nuclease activity–deficient mutant, was sufficient to induce sensitivity to AZD6738. Depletion of endogenous PGBD5 conferred resistance to AZD6738 in human tumor cells. PGBD5-expressing tumor cells accumulated unrepaired DNA damage in response to AZD6738 treatment and underwent apoptosis in both dividing and G₁-phase cells in the absence of immediate DNA replication stress. Accordingly, AZD6738 exhibited nanomolar potency against most neuroblastoma, medulloblastoma, Ewing sarcoma, and rhabdoid tumor cells tested while sparing nontransformed human and mouse embryonic fibroblasts in vitro. Finally, treatment with AZD6738 induced apoptosis and regression of human neuroblastoma and medulloblastoma tumors engrafted in immunodeficient mice in vivo. This effect was potentiated by combined treatment with cisplatin, including substantial antitumor activity against patient-derived primary neuroblastoma xenografts. These findings delineate a therapeutically actionable synthetic dependency induced in PGBD5-expressing solid tumors.

Epigenetic dependencies have become evident in many cancers. On the basis of antagonism between BAF/SWI-SNF and PRC2 in SMARCB1-deficient sarcomas, we recently completed the clinical trial of the EZH2 inhibitor tazemetostat. However, the principles of tumor response to epigenetic therapy in general, and tazemetostat in particular, remain unknown. Using functional genomics and diverse experimental models, we define molecular mechanisms of tazemetostat resistance in SMARCB1-deficient tumors. We found distinct acquired mutations that converge on the RB1/E2F axis and decouple EZH2-dependent differentiation and cell-cycle control. This allows tumor cells to escape tazemetostat-induced G1 arrest, suggests a general mechanism for effective therapy, and provides prospective biomarkers for therapy stratification, including PRICKLE1. On the basis of this, we develop a combination strategy to circumvent tazemetostat resistance using bypass targeting of AURKB. This offers a paradigm for rational epigenetic combination therapy suitable for translation to clinical trials for epithelioid sarcomas, rhabdoid tumors, and other epigenetically dysregulated cancers. Significance: Genomic studies of patient epithelioid sarcomas and rhabdoid tumors identify mutations converging on a common pathway for response to EZH2 inhibition. Resistance mutations decouple drug-induced differentiation from cell-cycle control. We identify an epigenetic combination strategy to overcome resistance and improve durability of response, supporting its investigation in clinical trials. See related commentary by Paolini and Souroullas, p. 903. This article is featured in Selected Articles from This Issue, p. 897

High-risk neuroblastoma, a pediatric tumor originating from the sympathetic nervous system, has a low mutation load but highly recurrent somatic DNA copy number variants. Previously, segmental gains and/or amplifications allowed identification of drivers for neuroblastoma development. Using this approach, combined with gene dosage impact on expression and survival, we identified ribonucleotide reductase subunit M2 (RRM2) as a candidate dependency factor further supported by growth inhibition upon in vitro knockdown and accelerated tumor formation in a neuroblastoma zebrafish model coexpressing human RRM2 with MYCN. Forced RRM2 induction alleviates excessive replicative stress induced by CHK1 inhibition, while high RRM2 expression in human neuroblastomas correlates with high CHK1 activity. MYCN-driven zebrafish tumors with RRM2 co-overexpression exhibit differentially expressed DNA repair genes in keeping with enhanced ATR-CHK1 signaling activity. In vitro, RRM2 inhibition enhances intrinsic replication stress checkpoint addiction. Last, combinatorial RRM2-CHK1 inhibition acts synergistic in high-risk neuroblastoma cell lines and patient-derived xenograft models, illustrating the therapeutic potential.

Predicting in vivo response to antineoplastics remains an elusive challenge. We performed a first-of-kind evaluation of two transcriptome-based precision cancer medicine methodologies to predict tumor sensitivity to a comprehensive repertoire of clinically relevant oncology drugs, whose mechanism of action we experimentally assessed in cognate cell lines. We enrolled patients with histologically distinct, poor-prognosis malignancies who had progressed on multiple therapies, and developed low-passage, patient-derived xenograft models that were used to validate 35 patient-specific drug predictions. Both OncoTarget, which identifies high-affinity inhibitors of individual master regulator (MR) proteins, and OncoTreat, which identifies drugs that invert the transcriptional activity of hyperconnected MR modules, produced highly significant 30-day disease control rates (68% and 91%, respectively). Moreover, of 18 OncoTreat-predicted drugs, 15 induced the predicted MR-module activity inversion in vivo. Predicted drugs significantly outperformed antineoplastic drugs selected as unpredicted controls, suggesting these methods may substantively complement existing precision cancer medicine approaches, as also illustrated by a case study. Significance: Complementary precision cancer medicine paradigms are needed to broaden the clinical benefit realized through genetic profiling and immunotherapy. In this first-in-class application, we introduce two transcriptome-based tumor-agnostic systems biology tools to predict drug response in vivo. OncoTarget and OncoTreat are scalable for the design of basket and umbrella clinical trials. This article is highlighted in the In This Issue feature, p. 1275

Large-scale combination drug screens are generally considered intractable due to the immense number of possible combinations. Existing approaches use ad hoc fixed experimental designs then train machine learning models to impute unobserved combinations. Here we propose BATCHIE, an orthogonal approach that conducts experiments dynamically in batches. BATCHIE uses information theory and probabilistic modeling to design each batch to be maximally informative based on the results of previous experiments. On retrospective experiments from previous large-scale screens, BATCHIE designs rapidly discover highly effective and synergistic combinations. In a prospective combination screen of a library of 206 drugs on a collection of pediatric cancer cell lines, the BATCHIE model accurately predicts unseen combinations and detects synergies after exploring only 4% of the 1.4M possible experiments. Further, the model identifies a panel of top combinations for Ewing sarcomas, which follow-up validation experiments confirm to be effective, including the rational and translatable top hit of PARP plus topoisomerase I inhibition. These results demonstrate that adaptive experiments can enable large-scale unbiased combination drug screens with a relatively small number of experiments. BATCHIE is open source and publicly available (https://github.com/tansey-lab/batchie).