Abstract:
‡a PR-104A was initially developed as a hypoxia-activated prodrug. It was subsequently characterized as the first substrate which undergoes aldo-keto reductase 1C3 (AKR1C3) mediated nitro reduction, to generate its cytotoxic metabolites. This oxygen-insensitive, off-target activation of PR-104A has led to its development in the use of AKR1C3 overexpressing neoplasms such as acute myeloid leukaemia (AML). The potential overexpression of AKR1C3 in leukaemia-initiating stem cells (LSCs), coupled with overexpression in the amplified pool of malignant AML cells presents a likely dual approach for targeting two distinct leukaemic cell populations with the use of an optimal AKR1C3 prodrug, which is likely to lead to better management of the disease. This project sought to design and characterize AKR1C3 prodrugs which are potentially better substrates for AKR1C3 than PR-104A. Three analogues of PR-104A were designed by molecular modeling methods, by replacing the carboxamide side chain of PR-104A with novel chemotypes that were shown by crystal structure evidence to better engage in the lipophilic cavity of the AKR1C3 active site. The library of prodrugs analyzed was expanded to include other nitrobenzamide mustard compounds in order to better understand the structure-activity relationship between substituents on the common nitrobenzamide mustard core and influences on AKR1C3-specific kinetic parameters and cytotoxicity. The general trend was for substrates with lipophilic side chains to confer improved affinity and catalysis with regards to AKR1C3. Substituents on the mustard arms were shown to influence affinity with the more lipophilic dibromo (Br/Br) mustard substrates showing better affinities for AKR1C3 than the bromomesylate (Br/OMs) mustard substrates. The three substrates with the newly designed cyclic substituted carboxamide side chains showed improved kinetic parameters relative to PR-104A, suggesting better engagement of the lipophilic cavity of AKR1C3 can lead to improved AKR1C3 kinetics. Cytotoxicity against AKR1C3 overexpressing cells was largely determined by the type of mustard present, with the Br/OMs mustard prodrugs showing greater cytotoxicity than the Br/Br mustard derivatives. The lack of a C3 benzene substitution seemed to provide prodrugs with a greater cytotoxic potential against AKR1C3-expressing cells compared to the ortho-nitro substituted derivatives. Two of the most potent prodrugs against AKR1C3- expressing cells in a 3D model of tumour cell growth were such mononitro mustard prodrugs (SN 34454 and SN 34118) which contained a methylpiperazine carboxamide side chain and differed only in relation to the mustard substituents. The prodrugs with the newly designed side chains failed to show improvements in cytotoxicity against AKR1C3 overexpressing cells relative to PR-104A, in both 2D and 3D models of cell growth, despite the improved kinetic parameters. Based on kinetic and cytotoxicity data, SN 34454 and SN 34118 were identified as two AKR1C3 prodrugs with improved affinities and potencies relative to PR-104A. Analysis of efficacy of SN 34454 and SN 34118 in vivo in a mixed AKR1C3 xenograft tumour model failed to establish a superior efficacy of either prodrug relative to PR-104A. Further analysis needs to be carried out in the design of more optimal in vivo experiments to accurately determine the level of cell kill that occurs as a result of diffusion of cytotoxic metabolites from AKR1C3 overexpressing (activator) cells to adjacent (target) cells lacking AKR1C3 expression (i.e. to better determine the bystander activity of each prodrug). Determination of the level of bystander cell kill is crucial in an AML setting as an ideal AKR1C3 prodrug for AML is postulated to possess a minimal bystander potential, being selectively cytotoxic to AKR1C3 overexpressing AML cells and LSCs, with minimal collateral damage to adjacent target cells.