Drugs that inhibit Bcl2-family survival proteins promise to change the landscape of cancer care. Cancers activate stress signals in the form of BH3-domain proteins like Bim as they escape cellular growth constraints and invade hostile environments. To remain viable, tumors use survival proteins like Bcl2, Bclx, Bclw and Mcl1 to sequester these BH3 proteins through specific protein-protein interactions (PPIs). This blocks their apoptotic signal but also renders such tumors continually dependent on this function. Drugs that competitively displace BH3 proteins from the survival protein sequestering them unleash a potent apoptotic signal. Indeed, venetoclax is a Bcl2 inhibitor that has demonstrated striking clinical efficacy, garnering FDA-approval for the treatment of chronic lymphocytic and adult myelogenous leukemias.These tumor types do not have Bcl2-activatingmutations but are empirically defined to be dependent on Bcl2 for their survival. In contrast, many solid tumors have heterogeneity in which survival protein they use to block BH3 signals, and the absence of biomarkers that predict sensitivity to this emerging drug class remains a barrier. Our objectives are to identify predictive biomarkers and develop diagnostic tools to leverage Bcl2-family inhibitors for clinical use. We created a national infrastructure to generate patient-derivedxenograftand cell line models of the lethal childhood tumor, neuroblastoma, and utilized innovative functional assays to define the Bcl2-family protein they depend on for survival. We discovered that neuroblastomas have endogenously activated Bim neutralized through a PPI with a single dominant survival protein. This sequesters Bim's apoptotic activity, but also encodes a continual dependency, and defines a mechanistic biomarker defining the survival protein required for viability. An unanticipated finding is that this survival dependency is highly cancer cell intrinsic and stable: consistent in patient-matched tumors from primary and metastatic sites, and at diagnosis and relapse. In tumors with Bim bound by Bcl2 (Bim:Bcl2 PPI), Bcl2 inhibitors like venetoclax are highly active in vitro and in vivo. In tumors with Bim bound by Mcl1 (Bim:Mcl1 PPI), Bcl2 inhibitors have no activity. Further, we find that all neuroblastomas with MAPK pathway mutations are in the Bim:Mcl1 class. Surprisingly, Mcl1 inhibitors also have no activity for this subset, despite displacing Bim from Mcl1. In these tumors, Bim is re-sequestered by Bclx, and all are exquisitely sensitive to combined Mcl1/Bclx inhibition. This demonstrates the robustness of our predictive Bim PPI biomarker that we will exploit to identify all survival dependency classes in neuroblastoma. We will define the extent to which the biomarker remains a stable intrinsic tumor feature that predicts selective vulnerability to Bcl2-family inhibitors. We also leverage therapeutic MAPK inhibitors to antagonize Mcl1 dependency. Importantly, we seek to develop in vitro diagnostic tools to identify predictive BimPPIs using proximity-ligation assays, and to credential genomic MAPK biomarkers to define Mcl1 dependence. Collectively, our work applies precision medicine approaches to assign Bcl2-family inhibitors for clinical use, and inform clinical trial designs, including predicting rational combination therapies.
The proposed research is relevant to public health since the resistance of tumors to available therapies remains the major cause of cancer mortality. A better understanding of howresistance develops, and the specific vulnerabilities each resistance mechanism poses, will provide new treatment opportunities. The inclusion of selective Bcl2 family inhibitors into clinical care, driven by this fundamental knowledge, will have broad reach in improving cancer outcomes.