Abstract:
The Development of Hypoxia-selective PERK Inhibitor Prodrugs Lydia P. Liew PhD1, Way W. Wong MSc1, Dean C. Singleton PhD1, Stephen M.F. Jamieson PhD1, Jack U. Flanagan PhD1, Costas Koumenis PhD2, Michael P. Hay PhD1. 1Auckland Cancer Society Research Centre, University of Auckland, Private Bag 92019, Auckland, New Zealand. 2Perelman School of Medicine, University of Pennsylvania, Philadelphia, PA, USA. Purpose The unfolded protein response (UPR) is initiated in cells under endoplasmic reticulum (ER) stress. Within the tumor microenvironment, accumulation of unfolded or misfolded proteins results in a stress stimuli which is sensed by chaperone proteins in the ER to activate protein kinase R-like ER kinase (PERK) as well as inositol-requiring enzyme 1 (IRE-1) and activating transcription factor-6 (ATF-6). Under hypoxia-induced stress, the UPR is primarily mediated via the PERK pathway. Upon activation, PERK signals responses to alleviate cellular stress; in the event where these stresses cannot resolved, the cells enter apoptosis. Potent and selective PERK inhibitors (PERKi) have been developed, and leading examples (GSK2606414 and GSK2656157) demonstrate tumor growth inhibition in human tumor xenograft models. However, these PERKi also exhibit mechanism-based normal tissue toxicity. Although GSK’157 was advanced to Phase I studies, no further studies have been conducted, presumably because of an inadequate therapeutic index. In order to overcome this normal tissue toxicity, we propose to engender tumor selectivity by using a hypoxia-activated prodrug (HAP) approach. We envision inactive, non-toxic HAPs of PERK inhibitors will undergo selective activation in hypoxic tumour tissue to release the active drug. Methods Examination of the binding mode of GSK’414 in the active site of PERK (pdb4g31) highlighted several critical interactions involved in drug binding. We explored two design approaches (i) blocking interaction with the kinase β-strand hinge; and (ii) blocking interaction with the lipophilic specificity pocket. We prepared a series of analogues based on GSK’414 and the corresponding HAP analogues. We measured the stability of the prodrugs and their ability to fragment following reduction. We determined the ability of the analogues to inhibit phosphorylation of EIF2AK3 and explored their effect on HCT116 cells under oxic and anoxic conditions. Results Our attempts to design effective prodrugs that would disrupt the kinase β-strand hinge binding were unsuccessful. Model prodrugs did not demonstrate fragmentation after reduction and the presence of the prodrug unit precluded the synthesis of fully functional PERKi. In contrast, we were able to exploit structure-based design to successfully prepare new indolyl-pyrrolo[2,3-d]pyrimidine analogues with modifications in the lipophilic specificity domain designed to accommodate HAP triggers. The structure-activity relationships (SAR) for PERK inhibition were determined and nitroimidazole-based HAPs were prepared. The purity and stability of the HAPs were tested in biological media. The HAPs fragmented to release the effector following reductive activation in vitro. Conclusions While direct preparation of HAPs of the GSK clinical candidate was not possible, modification at the specificity pocket domain led to potent PERKi suitable for a HAP approach. The corresponding HAPs were significantly deactivated as PERKi and underwent reductive activation to release effector. Their biological evaluation and development continues. We are currently investigating an expanded series of PERKi, and are optimising trigger and linker options for our inhibitors.