Dynamic BH3 profiling-poking cancer cells with a stick

ABSTRACT

The vast majority of efforts in precision medicine for cancer try to link static genetic information to tumor biology and from there predict clinical response. Dynamic BH3 profiling offers an alternative functional approach by measuring death signaling induced by specific drugs in tumors from patients ex vivo to predict clinical response.

Since the introduction of Gleevec® for the treatment of chronic myeloid leukemia (CML), novel targeted therapies have represented a breakthrough in clinical oncology. For instance, several newly US Food and Drug Administration (FDA)-approved chemotherapeutic agents, including numerous small-molecule kinase inhibitors, are being used in oncology to treat different forms of leukemia, lung cancer, gastrointestinal stromal tumor, melanoma, breast cancer, and many other malignancies. However, with this growing armamentarium of cancer therapies we face a new problem: how to assign the right drug to the right patient. Enormous efforts have been devoted to discover new predictive biomarkers and improve precision medicine, mostly through genomic approaches, with some major successes.^1-3 However, in many cases there is still a lack of good predictors. This is likely due to cancer's complex signaling networks, mutations, and adaptation to treatment.^4,5

Cancer can be considered a complex system. One feature of a complex system is that its evolution and response to perturbations can be difficult to predict from initial conditions. All of the –omic technologies, whether genomic, proteomic, metabolomic, or transcriptomic, essentially represent different static, initial conditions of components of the complex cancer cell. It is therefore not surprising that there would be limits to the power of these tools to predict clinical response to therapy.

When confronted by complexity, an alternative strategy to learn about a system is to make strategic perturbations and observe the dynamic evolution of the system. For example, when wildlife expert Steve Irwin wanted to find out how a wild animal (complex system) might respond to aggression, he did not measure initial conditions—he poked it with a stick (strategic perturbation)! In addressing the problem of predicting whether a tumor will be killed in vivo by a particular treatment, we have turned to exposing the cancer cell ex vivo to the precise agent(s) in question and making rapid measurements of the induced death signaling. We call this strategy dynamic BH3 profiling⁶ (Fig. 1).

Figure 1.

Precision medicine with dynamic BH3 profiling. Patient tumor biopsies are converted to a single-cell suspension and briefly treated ex vivo with a number of drugs or combinations. After 6-24 hours, the cells are gently permeabilized to allow contact of BH3 peptides with mitochondria, and mitochondrial permeabilization is measured, e.g., by JC-1 dye fluorescence. Those agents that increase mitochondrial sensitivity to peptides are those that induce death signaling. Such agents are given priority for patient treatment. Adapted with permission from REF 6, Elsevier.

BH3 profiling exposes mitochondria to proapoptotic BH3 peptides and measures the amount of mitochondrial outer membrane permeabilization (MOMP). When a large magnitude of MOMP is induced by small amounts of peptide, we consider the mitochondrion (and the cell containing it) to be relatively “primed” for apoptotic death.⁷ We have found that baseline apoptotic priming predicts clinical response to conventional chemotherapeutics.^8,9 In dynamic BH3 profiling (DBP), we first expose the intact cells to one or more agents, and then ask whether the treatment causes an increase in apoptotic priming—that is, an increased sensitivity of the mitochondria to the BH3 peptide.

In practice, we obtain a single-cell suspension from a patient sample (or cell line), expose the cells ex vivo to treatments of interest, and after a short incubation (typically 6-24 hours) the cells are permeabilized with digitonin and exposed to the BIM BH3 peptide.¹⁰ This synthetic 20-mer oligopeptide is derived from the BH3 domain found in BIM, a BH3-only proapoptotic member of the BCL-2 family. Death signaling, measured by increased priming, should be initiated only in those cells that are dependent on the particular pathway that is targeted by the inhibitor. The first question we needed to address was “Does early (e.g., measured within 16 hours) death signaling predict frank cell death, even when cell death does not occur until a few days later?” We initially tested this in over a hundred cell line experiments, and found that it did.⁶ In subsequent murine and human clinical experiments, we found that DBP could also predict the in vivo response to therapy.

Using cell lines and clinical experiments, we find that early drug-induced death signaling measured by dynamic BH3 profiling predicts chemotherapy response across many cancer types and many agents, including combinations of chemotherapies, showing its versatility and applicability. One can imagine different clinical contexts in which such a tool might be useful. In the case of a single drug or combination and many patients, DBP could be used to stratify patients according to their likelihood to respond. This might be valuable in situations where only a minority of patients with a particular cancer respond to a certain therapy. DBP might offer a way to identify sensitive patients prospectively so that a majority of treated patients would respond. In a distinct context, a physician might need to choose among many therapies for a single patient. Given its flexibility, DBP might be used to identify the best therapy for that patient based on identifying the therapy that promotes the greatest death signaling in tumor cells ex vivo. This is the ultimate goal of precision medicine—to match the right drug with the right patient.

DBP can be performed in less than 24 hours. It is therefore consistent with the pace of decision making that confronts oncologists, even for the most aggressive cancers. Since it is a probe of function, however, it requires fresh or viably frozen tissues. Although this might present a challenge to initial testing, it should be remembered that all biopsies start out viable and unfixed. If DBP, or indeed another functional assay, can demonstrate its superior utility in clinical decision making, those fresh biopsies will likely be forthcoming. The prospective clinical testing required to sufficiently validate this type of approach is just beginning.

It is ironic that despite an appreciation of the complexity and dynamism of the cancer cell, the first step of the pathology department on receipt of a specimen is usually to kill all of the cells by fixation. We can think of no more important objective of the study of a patient's tumor sample than to choose the best therapy possible. We look forward to the day when dynamic, functional assays become an integral part of more effective selection of therapies for the individual cancer patient.

Disclosure of potential conflicts of interest

No potential conflicts of interest were disclosed.