A novel prodrug activation system, endogenous in human tumor cells, is described. A latent enzyme-prodrug system is switched on by a simple synthetic, small molecule co-substrate. This ternary system is inactive if any one of the components is absent. CB 1954 [5-(aziridin-1-yl)-2,4-dinitrobenzamide] is an antitumor prodrug that is activated in certain rat tumors via its 4-hydroxylamine derivative to a potent bifunctional alkylating agent. However, human tumor cells are resistant to CB 1954 because they are unable to catalyze this bioactivation efficiently. A human enzyme has been discovered that can activate CB 1954, and it has been shown to be commonly present in human tumor cells. The enzyme is NQO2 [NAD(P)H quinone oxidoreductase 2], but its activity is normally latent, and a nonbiogenic co-substrate such as NRH [nicotinamide riboside (reduced)] is required for enzymatic activity. There is a very large (100-3000-fold) increase in CB 1954 cytotoxicity toward either NQO2-transfected rodent or nontransfected human tumor cell lines in the presence of NRH. Other reduced pyridinium compounds can also act as co-substrates for NQO2. Thus, the simplest quaternary salt of nicotinamide, 1-methyl-3-carboxamidopyridinium iodide, was a co-substrate for NQO2 when reduced to the corresponding 1,4-dihydropyridine derivative. Increased chain length and/or alkyl load at the 1-position of the dihydropyridine ring improved specific activity, and compounds more active than NRH were found. However, little activity was seen with either the 1-benzyl or 1-(2-phenylethyl) derivatives. A negatively charged substituent at the 3-position of the reduced pyridine ring also negated the ability of these compounds to act as cosubstrates for NQO2. In particular, 1-carbamoylmethyl-3-carbamoyl-1,4dihydropyridine was shown to be a co-substrate for NQO2 with greater stability than NRH, with the ability to enter cells and potentiate the cytotoxicity of CB 1954. Furthermore, this agent is synthetically accessible and suitable for further pharmaceutical development. NQO2 activity appears to be related to expression of NQO1 (DT-diaphorase), an enzyme that is known to have a favorable distribution toward certain human cancers. NQO2 is a novel target for prodrug therapy and has a unique activation mechanism that relies on a synthetic co-substrate to activate an apparently latent enzyme. Our findings may reopen the use of CB 1954 for the direct therapy of human malignant disease.