The Use of a BH3 Mimetic Toolkit Reveals a Switch of Myeloma Patients Dependency from BCL2 at Diagnosis to MCL1 at Relapse

Targeting anti-apoptotic proteins of the BCL2 family by BH3 mimetic small molecules is a new attractive therapeutic approach in hematological malignancies. The BH3 mimetics trigger apoptosis by mimicking the BH3-only proteins and promoting the release of pro-apoptotic BCL2 family members from anti-apoptotic proteins. Thus, the specific BH3 mimetics targeting BCL2, BCLXL or MCL1 exploit the dependency on these different anti-apoptotic proteins to kill tumor cells. Multiple myeloma is considered as a disease mainly dependent on MCL1 for survival based on the MCL1 dependency observed in human myeloma cell lines (HMCL), which are derived from the extramedullary stage of the disease and not representative of the diagnosis stage. Because no study up to date has reported the dependence of primary cells, we studied the dependency on BCL2 anti-apoptotic proteins both in a large collection of HMCL and in a cohort of 24 myeloma patients. We used a BH3 mimetic toolkit that includes venetoclax (ABT-199), A1155463 and A1210477, which target BCL2, BCLXL and MCL1 respectively.

Using a panel of 34 HMCLs, we demonstrated that 23% of HMCL were highly sensitive to venetoclax (LD50 <0.3 mM) all of them harboring a t(11;14) translocation and therefore dependent on BCL2, whereas only 3% of HMCL was efficiently killed by A-1155463 (LD50 <0.3 mM) showing that a minor subset of HMCL is highly BCLXL-dependent. We found that the majority of HMCLs (56%) were efficiently killed by A1210477; this result is in total agreement with the BH3 profiling performed with the MS1 peptide, specific for targeting MCL1 (r=0.74, p=0.0001), confirming the specificity of A1210477 as a specific MCL1 BH3 mimetic. Among the BCL2 dependent HMCLs, one third were also MCL1 dependent, indicating co-dependency to both molecules.

We also determined the dependency of primary myeloma cells by treating " ex-vivo" 24 consecutive bone marrow samples from myeloma patients (12 at diagnosis and 12 at relapse) with the respective BH3 mimetic for 20 hours. Patient characteristics showed that the age, sex and % of plasma cells from myeloma samples either at diagnosis or relapse were not significantly different. Analysis of patients at diagnosis revealed three stricking findings. First, a large group of patients (50%) were resistant to the three BH3 mimetics indicating that they were not dependent on any of the main anti-apoptotic proteins. Second, 41% of patients were sensitive to venetoclax, among them 20% were also sensitive to A1210477 indicating a co-dependency to BCL2/MCL1. Third only 8 % of patients showed a strict MCL1 dependency. Interestingly, at relapse 65% of myeloma patients showed a MCL1 dependency. Among them 25% were also killed by venetoclax while none of the myeloma samples presented a strict dependency to BCL2. While no dependency on BCLXL has been found at diagnosis, 24% of patients at relapse were efficiently killed by A1155463. Finally, only 16% of myeloma samples at relapse were resistant to the three BH3 mimetics compared to the 50% found at diagnosis. We hypothesize that the increase of the priming in favor of MCL1 or BCLXL at relapse could be the results of either the influence of drug treatments or a clonal selection.

Our study demonstrated that the dependency of myeloma cells on a particular anti-apoptotic BCL2 molecule differs with regards to the stage of the disease; At diagnosis, patients were mainly not dependent on BCL2, BCLXL or MCL1 or BCL2 dependent, whereas at relapse they were mainly MCL1 dependent or co-dependent. Finally, our results in myeloma suggested that the vulnerability of the tumor cells to the MCL1 BH3 mimetic rises during the progression of the disease while the resistance of myeloma cells to conventional drugs increased. Overall our results open the way to the use of BH3 mimetic targeted therapies not only at myeloma relapse but also at diagnosis.