Testing the BCL-2 Family to Direct Therapy and Predict Response.

Abstract SCI-14

Many, if not most, chemotherapeutic agents kill cancer cells via the mitochondrial apoptotic pathway. A simple model of this pathway breaks the killing into three steps. In Step 1, drug contacts target. In Step 2, prodeath signaling is generated. In Step 3, this pro-death signaling is interpreted by the BCL-2 family proteins at the mitochondrion, and when sufficient, the cancer cell is committed to the fate of death. Despite prominent exceptions, it is generally poorly understood why some cells are killed by chemotherapy and others are not. This is particularly perplexing in the case of most conventional chemotherapeutics which target ubiquitous elements such as DNA or microtubules. Most studies of determinants of chemotherapy induced cell death are focused on Steps 1 and 2. Relatively little is known about how mitochondrial preconditions affect chemosensitivity in cancer cells. We have developed a tool called BH3 profiling which exposes mitochondria to measured doses of BH3 domains, pro-death molecules of the BCL-2 family that ultimately carry pro-death signaling from Step 2 to the mitochondrion. By comparing mitochondrial dysfunction between different cells exposed to identical doses of BH3 domains, we can compare readiness of different cell types to undergo apoptosis. We have shown that this mitochondrial preset predicts response to conventional therapies of disparate mechanism, including etoposide, vincristine, and doxorubicin. Furthermore, BH3 profiling can predict response to agents such as ABT-737 which selectively kill cells dependent on anti-apoptotic proteins BCL-2, BCL-XL, and BCL-w. We have found BH3 profiling to be useful in many hematologic malignancies, including CLL, AML, ALL, lymphoma and myeloma. We have recently improved the BH3 profiling tool so that now it can be performed in a fluorescence activated cell sorting (FACS) format. The FACS format allows us to deconvolute complex, heterogeneous samples such are typically provided by clinical material. Our results suggest that heterogeneity of response to therapies, either within heterogeneous samples or between different tumors, relies to a very great extent on the mitochondrial preset measured by BH3 profiling.