Search

X-ray, NMR, and molecular mechanics studies on antitumor antibiotic porfiromycin (C16H20N4O5), a covalent binder of DNA, have been carried out to study the structure, conformation, and theoretical interactions with DNA. The crystal structure was solved by direct methods and refined to an R value of 0.052. The configurations at C(9), C(9a), C(1), and C(2) are S, R, S, and S, except for the orientation of the aziridine ring and (carbamoyloxy)methyl side chain. The five-membered ring attached to the aziridine ring adopts an envelope conformation. The solution conformation is similar to that observed in the solid state except for the (carbamoyloxy)methyl side chain. Monovalent and cross-linked models of the drug bound to DNA have been energetically refined by using molecular mechanics. The results indicate that, in the case of monocovalent binding, the drug clearly prefers a d(CpG) sequence rather than a d(GpC) sequence. In the case of the cross-linked model there is no clear-cut preference of d(CpG) over d(GpC), indicating that the binding preference of the drug may be kinetic rather than thermodynamic.

A series of 7-(2-substituted-ethyl)amino analogues of mitomycin C and porfiromycin was prepared and screened in standard antitumor systems. Certain of these analogues showed better activity than mitomycin C against P-388 leukemia, L-1210 leukemia, and/or B-16 melanocarcinoma in mice. Compounds also tested for their leukopenic effects in mice, the limiting toxicity of mitomycin C. Some of them were less leukopenic and some were more leukopenic than this clinical agent. No statistically significant correlations could be made between physicochemical properties and antitumor activities of the analogues.

Two mechanisms proposed for the acid-catalyzed conversions of mitomycins into mitosenes were investigated by deuterium incorporation methods. Four different mitomycins, an aziridinomitosene, and an N-acetylmitomycin all underwent the conversion in acetic acid-d with no incorporation of deuterium at C-1. This evidence suggests that the mechanism based on initial elimination of the elements of methanol to give an aziridinomitosene is more likely the correct one. The products of these reactions had considerable variation in the ratios of cis to trans isomers: 7-aminomitosanes gave a predominance of trans and 7-methoxymitosanes gave a predominance of cis. Treatment of mitomycin C with DCl in D2O gave predominantly cis product with about 45% deuterium exchange at C-1. The isomeric 2,7-diamino-1-methoxymitosenes previously obtained by treating mitomycin C in methanol containing acetic acid were found to have stereochemistry opposite to that originally assigned by us.

New mitomycin C and porfiromycin analogues were prepared by treating mitomycin A and N-methylmitomycin A with a variety of amines, including aziridines, allylamines, propargylamines, chloroalkylamines, hydroxyalkylamines, glycine derivatives, aralkylamines, and heterocyclic amines. All analogues were evaluated against P-388 murine leukemia and selected ones were examined for their leukopenic properties. Certain analogues were found to be superior to mitomycin C in potency, efficacy, and therapeutic ratio in the P-388 assay. The most active substituents at the mitosane 7 position included aziridine, 2-methylaziridine, propargylamine, furfurylamine, methyl glycinate, and 3-aminopyridine. Mitomycin A and the 7-aziridino, 7-(2-methylaziridino), and 3-aminopyridine analogues were less leukopenic than mitomycin C. Certain other analogues, including propargylamino and methyl glycinate, were highly leukopenic. The three compounds tested against B-16 melanoma in mice were significantly more effective than mitomycin C in this assay. Previously established structure--activity relationships were found inadequate to account for all of the new data.