The initial steps of the activation of tirapazamine (TPZ, 1, 3-amino-1,2,4-benzotriazine 1,4-N,N-dioxide) under hypoxic conditions consist of the one-electron reduction of 1 to radical anion 2 and the protonation of 2 at O(N4) or O(N1) to form neutral radicals 3 and 4, respectively. There are some questions, however, as to whether radicals 3 and/or 4 will then undergo N–OH homolyses 3 → 5 + ·OH and 4 → 6 + ·OH or, alternatively, whether 3 and/or 4 may react by dehydration and form aminyl radicals via 3 → 11 + H2O and 4 → 12 + H2O or phenyl radicals via 3 → 17 + H2O. These outcomes might depend on the chemistry after the homolysis of 3 and/or 4, that is, dehydration may be the result of a two-step sequence that involves N–OH homolysis and formation of ·OH aggregates of 5 and 6 followed by H-abstraction within the ·OH aggregates to form hydrates of aminyls 11 and 12 or of phenyl 17. We studied these processes with configuration interaction theory, perturbation theory, and density functional theory. All stationary structures of OH aggregates of 5 and 6, of H2O aggregates of 11, 12, and 17, and of the transition state structures for H-abstraction were located and characterized by vibrational analysis and with methods of electron and spin-density analysis. The doublet radical 17 is a normal spin-polarized radical, whereas the doublet radicals 11 and 12 feature quartet instabilities. The computed reaction energies and activation barriers allow for dehydration in principle, but the productivity of all of these channels should be low for kinetic and dynamic reasons. With a view to plausible scenarios for the generation of latent aryl radical species without dehydration, we scanned the potential energy surfaces of 2–4 as a function of the (O)N1–Y (Y = C5a, N2) and (O)N4–Z (Z = C4a, C3) bond lengths. The elongation of any one of these bonds by 0.5 Å requires less than 25 kcal/mol, and this finding strongly suggests the possibility of bimolecular reactions of the spin-trap molecules with 2–4 concomitant with triazene ring-opening.

Heterocyclic N-oxides are an interesting class of antitumor agents that selectively kill the hypoxic cells found in solid tumors. The hypoxia-selective activity of the lead compound in this class, tirapazamine, stems from its ability to undergo intracellular one-electron reduction to an oxygen-sensitive drug radical intermediate. In the presence of molecular oxygen, the radical intermediate is back-oxidized to the parent molecule. Under hypoxic conditions, the extended lifetime of the drug radical intermediate enables its conversion to a highly cytotoxic DNA-damaging intermediate via a “deoxygenative” mechanism involving the loss of oxygen from one of its N-oxide groups. The natural product myxin is a phenazine di-N-oxide that displays potent antibiotic activity against a variety of organisms under aerobic conditions. In light of the current view of heterocyclic N-oxides as agents that selectively operate under hypoxic conditions, it is striking that myxin was identified from Sorangium extracts based upon its antibiotic properties under aerobic conditions. Therefore, we set out to examine the molecular mechanisms underlying the biological activity of myxin. We find that myxin causes bioreductively activated, radical-mediated DNA strand cleavage under both aerobic and anaerobic conditions. Our evidence indicates that strand cleavage occurs via a deoxygenative metabolism. We show that myxin displays potent cytotoxicity against the human colorectal cancer cell line HCT-116 under both aerobic and anaerobic conditions that is comparable to the cell-killing properties of tirapazamine under anaerobic conditions. This work sheds light on the processes by which the naturally occurring aromatic N-oxide myxin gains its potent antibiotic properties under aerobic conditions. Furthermore, these studies highlight the general potential for aromatic N-oxides to undergo highly cytotoxic deoxygenative metabolism following enzymatic one-electron reduction under aerobic conditions.

The initial steps of the activation of tirapazamine (TPZ, 1, 3-amino-1,2,4-benzotriazine 1,4-N,N-dioxide) under hypoxic conditions consist of the one-electron reduction of 1 to radical anion 2 and the protonation of 2 at O(N4) or O(N1) to form neutral radicals 3 and 4, respectively. There are some questions, however, as to whether radicals 3 and/or 4 will then undergo N–OH homolyses 3 → 5 + ·OH and 4 → 6 + ·OH or, alternatively, whether 3 and/or 4 may react by dehydration and form aminyl radicals via 3 → 11 + H2O and 4 → 12 + H2O or phenyl radicals via 3 → 17 + H2O. These outcomes might depend on the chemistry after the homolysis of 3 and/or 4, that is, dehydration may be the result of a two-step sequence that involves N–OH homolysis and formation of ·OH aggregates of 5 and 6 followed by H-abstraction within the ·OH aggregates to form hydrates of aminyls 11 and 12 or of phenyl 17. We studied these processes with configuration interaction theory, perturbation theory, and density functional theory. All stationary structures of OH aggregates of 5 and 6, of H2O aggregates of 11, 12, and 17, and of the transition state structures for H-abstraction were located and characterized by vibrational analysis and with methods of electron and spin-density analysis. The doublet radical 17 is a normal spin-polarized radical, whereas the doublet radicals 11 and 12 feature quartet instabilities. The computed reaction energies and activation barriers allow for dehydration in principle, but the productivity of all of these channels should be low for kinetic and dynamic reasons. With a view to plausible scenarios for the generation of latent aryl radical species without dehydration, we scanned the potential energy surfaces of 2–4 as a function of the (O)N1–Y (Y = C5a, N2) and (O)N4–Z (Z = C4a, C3) bond lengths. The elongation of any one of these bonds by 0.5 Å requires less than 25 kcal/mol, and this finding strongly suggests the possibility of bimolecular reactions of the spin-trap molecules with 2–4 concomitant with triazene ring-opening.