Synergistic Activity of Lysine Deacetylase Inhibitors and Microtubule-Targeting Agents in AML

Background: Treatment options for elderly acute myeloid leukemia (AML) patients, despite various novel therapies, are still limited. Relapse is common after chemotherapy, and progression on lower intensity treatment with hypomethylating agents is inevitable in all patients even after initial response. Lysine deacetylase inhibitors (KDIs or HDAC inhibitors) have shown limited single-agent clinical activity in AML. Therefore, to develop potential novel treatment options for AML, we aimed to identify synergistic KDI-based drug combinations to build upon the single-agent activity and clinical-availability of KDIs. We performed RNA-interference (RNAi) drug modifier screens of the kinome and phosphatome with the KDI vorinostat in AML cells in order to identify targets to inhibit in combination with KDIs. Based on pathway analysis of RNAi screen results that highlighted an unexpected role for p38/JNK/SAPK signaling in vorinostat activity in AML, we explored the putative p38 substrate-selective MK2 inhibitor CMPD1 combined with KDIs, and demonstrated potent synergistic activity in AML regardless of molecular, cytogenetic or lineage background. As published evidence suggests that CMPD1 can also act as a microtubule-destabilizing/targeting agent (MTA), and that MK2 can associate with microtubules and cytoskeleton-associated proteins, we examined several MTAs in combination with KDIs in AML, including vinca alkaloid microtubule-destabilizing agents vinblastine/vincristine, and taxol as a microtubule-stabilizing agent.

Results: Vinblastine and vincristine dose-dependently synergized with vorinostat similarly to CMPD1 in all AML cell lines tested, with vinblastine being more potently synergistic than vincristine. Taxol only potentiated KDI activity in approximately half of AML cell lines. Importantly, preliminary data suggest preferential, dose-dependent synergy of KDIs and vinca alkaloid MTAs in the CD34+ progenitor AML cell fraction, as compared to the paired CD34-depleted fraction from the same primary AML sample in ex vivo culture. This preferential CD34+ activity was similar to that observed with CMPD1 combined with KDIs in ex vivo models. Consistent with synergistic activity of MTAs and KDIs, multiple RNAi screen hits also converged on microtubule dynamics. Subsequently, we identified the LKB1 signaling network, a critical integrator of cellular stress metabolism and hematopoietic stem cell quiescence/homeostasis, as a determinant of vorinostat activity in AML. Of 14 known kinases directly regulated by LKB1, 4 belong to microtubule-associated protein (MAP) or microtubule-affinity regulating kinases (MARK) families (e.g. screen hit MARK4). RNAi screen hits also converged on regulation of microtubule proteins, or directly on microtubule proteins themselves, (e.g. atypical MAPK4 and MAPK6 which phosphorylate microtubule-associated protein 2, or TAOK2 which has a kinase-independent role in stabilizing microtubules). Additionally, several RNAi hits such as PACSIN1 and -2, and MAST2 and -4 link the actin cytoskeleton to microtubule dynamics.

Discussion: Herein we show that several types of clinically-used MTAs (i.e. vinca alkaloids or taxanes) potently synergize with KDIs in AML, and that this synergy can occur preferentially in primary CD34+ cells, as compared to the CD34-negative AML compartment. Although at high in vitro doses microtubule-destabilizing agents such as vinca alkaloids, and microtubule-stabilizing agents such as taxanes, can both induce apoptosis via mitotic perturbation, many publications show that low therapeutic MTA doses can induce apoptosis in tumor cells in vivo independent of mitosis. Therefore, we hypothesize that clinically-available KDIs (i.e. vorinostat, panobinostat) and MTAs (i.e. vinca alkaloids or taxanes) represent novel combinations with the potential to target the AML progenitor pool even in a non-proliferate state, and thus have the potential to eradicate minimal residual disease.