A comparison of mechanisms proposed for the conversion of mitomycins into mitosenes.

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.

Additional nucleotide derivatives of mitosenes. Synthesis and activity against parental and multidrug resistant L1210 leukemia.

Cytidine 5'-monophosphate and 5'-ara-CMP conjugates of 2,7-diaminomitosene, with the phosphate groups linked to C-1, were prepared by treating mitomycin C with the appropriate nucleotides. 5'-UMP conjugates were prepared from mitomycin A, 7 (M-83), and 8 (BMY-25282) by similar procedures. A conjugate could not be prepared from mitomycin C and 6-MPRP, but a sulfur-linked derivative was made with 6-MP ribonucleoside. The corresponding 1-hydroxy-2-aminomitosenes were prepared from the parent mitomycin analogues for structure-activity comparisons. All compounds were tested against L1210 murine leukemia in the MTT tetrazolium dye assay. In general, the conjugates were less potent than the parent mitomycins; however 5'-ara-CMP conjugate 14 derived from mitomycin C was more potent than the parent compound or any mitomycin tested except mitomycin A. It also was more potent than ara-C. This result establishes the value of this approach to prodrugs, at least in cell culture. Against a multi-drug-resistant L1210 cell line, all of the conjugates derived from mitomycin C were more potent than the parent compound. 6-Mercaptopurine ribonucleoside conjugate 15 was more active against the resistant cells than it was against the parental cell line.

Metal complexes of mitomycins.

The preparation of stable complexes between the N7-[2-(2-pyridyl)ethyl] and N7-(2-piperazinylethyl) derivatives of mitomycin C and metal ions such as Cu(II), Zn(II), and Pt(II) was accomplished. Mitomycin C did not form stable complexes, but it rearranged to a mitosene capable of complex formation. Some of these complexes had antitumor activity in mice. However, they were less active than mitomycin C. Weak associations between mitomycin C and metal ions were demonstrated by 13C and 15N NMR spectrometry.

Preparation and antitumor activity of 7-substituted 1,2-aziridinomitosenes.

7-Methoxy-1,2-aziridinomitosenes were prepared from mitomycin A and its N-methyl homologue by catalytic reduction followed by air oxidation. Treatment of these products with amines, including ammonia, ethylenimine, 2-methylethylenimine, propargylamine, and furfurylamine gave the corresponding 7-(substituted amino) derivatives. Screening of these compounds against P-388 leukemia in mice revealed some good activities. The more easily reduced compounds gave prolongation of life span comparable to that of mitomycin C, but their optimal doses were higher. Among these compounds, a methyl group on the aziridine nitrogen increased potency. The 7-amino derivatives, which were difficult to reduce to hydroquinones, were essentially inactive. The aziridinomitosenes were subjected to a Hansch-type analysis, but no statistically significant correlation was found.

Aminoglycoside antibiotics. 6. Chemical reactions of aminoglycosides with disodium carbenicillin.

Aminoglycoside antibiotics including kanamycin A, tobramycin, and the gentamicin C complex reacted with 1 mol of disodium carbenicillin to give products derived from acylation of their amino groups by the beta-lactam function of the carbenicillin. Amikacin was acylated by two carbenicillin units. Chromatographic analysis of fragments from the acid hydrolysis of these derivatives showed that the preferred site of acylation was in the 2-deoxystreptamine unit of the aminoglycosides. The two sites of acylation in amikacin were the 6'-amino group and the amino group in the aminohydroxybutyryl substituent. The derivatives had almost no antibacterial activity, and they were not toxic.

New 2-substituted indoloquinone mitomycin analogues.

Previously reported 2-(hydroxymethyl)indoloquinones, prepared as their acetates or carbamates, were less active than 2-methyl analogues in bacterial cultures and they had no activity in mice, despite functionality appropriate for DNA cross-linking. On the basis of the hypothesis that these compounds might have been too reactive chemically for selective alkylation of DNA, we prepared new analogues with substituents that could give variation in the reduction potential of the quinone ring, which might control their rate of bioactivation. The 5-methoxyindoloquinones were much more potent cytotoxics than mitomycin C against human tumor cell lines, but they were inactive against P388 leukemia in mice. Two 5-aziridinylindoloquinones were also more potent than mitomycin C against the cell lines and one of them was active in the P388 model upon in vivo assay. The corresponding 5-amino analogues were less potent than mitomycin C against both the cell lines and murine P388 leukemia. A 2-(1-hydroxyethyl)carbamate was prepared by a 20-step synthesis. It was about one-fourth as potent as mitomycin C against two cell lines.

Structure-activity relationships for mitomycin C and mitomycin A analogues.

A set of 30 mitomycin C and mitomycin A analogues, including five new compounds, was screened against three different solid human tumor cell lines using the MTT tetrazolium dye assay. A statistically significant correlation among antitumor activity, quinone reduction potential (E1/2), and the logarithm of the partition coefficient (log P) was obtained, with the most easily reduced and the most lipophilic compounds being the most potent. When these analogues were separated into mitomycin C and mitomycin A subsets, the former gave a correlation only with E1/2, whereas the latter (which differ little in their E1/2 values) gave a correlation only with log P. These correlations are in contrast to those made in the P388 leukemia assay in mice wherein the most active mitomycin C and mitomycin A analogues were the most hydrophilic ones. When the same compounds were tested against P388 leukemia cells in the MTT assay, the results were the same as those of the solid tumor assays. Thus, the substantial differences in relative potencies of mitomycins are related not to the kind of tumor cell, but to the type of assay performed, cell culture versus whole animal. No correlation was found between antitumor potency in the cell culture systems and calculated relative DNA binding strengths, probably because the limiting factors in antitumor potency of mitomycins appear to be tumor cell uptake (log P) and/or bioreductive activation (E1/2).

Molecular mechanics simulations on covalent complexes of mitomycin C and its analogues with left-handed DNA duplexes.

We present molecular mechanics simulations on covalent complexes between d(GCGCGCGCGC).d(GCGCGCGCGC) in the left-handed double helical forms (B and Z) and potent antitumor antibiotics mitomycin C and three of its analogues using the all atom force field in the framework of the program AMBER(UCSF). The energy-refined models of the complexes show interesting networks of hydrogen-bonding interactions between the drugs and DNA groups in the minor groove of the left-handed helices. The energy-refined models suggest that mitomycins could bind strongly to left-handed helices. This result might be relevant to the interpretation of earlier experiments which suggested that DNA bound by mitomycin C underwent a transition to a non-Z left-handed structure.

Development of new mitomycin C and porfiromycin analogues.

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.

Preparation and antitumor activity of new mitomycin A analogues.

A series of 26 mitomycin A analogues including 23 new ones was prepared by a variety of methods. The most useful methods were alkoxide exchange on mitomycin A and treatment of 7-hydroxymitosane with 3-substituted 1-phenyltriazenes. Many of the new analogues were superior to mitomycin C in the P388 leukemia assay and the more stringent subcutaneous B16 melanoma assay both in mice. Four of them gave long-term survivors in the latter assay. Quantitative correlations between log P and antitumor activity were not possible, but some guidelines for future analogue development are proposed.

Alkylation of DNA by C-10 of 2,7-diaminomitosene.

Mitomycin C and certain analogues alkylate DNA with their C-1 position and cross-link it by a second alkylation involving C-10. We now show that monoalkylation by C-10 (carbamate group) can occur for mitosene analogues that have no reactive C-1 functionality. Sodium dithionite reduction of 2,7-diaminomitosene or 2,7-diamino-1-hydroxymitosene in the presence of calf thymus DNA resulted in alkylation of the DNA to the extent of one molecule per 14 and 11 bases, respectively, although no covalent binding was observed on catalytic reduction. Reduction of each of these mitosenes by sodium dithionite in the presence of 2'-deoxyguanosine gave monoalkylation on the 2-amino group of this nucleotide. The 2,7-diaminomitosenes inhibited L-1210 leukemia cell colony formation in vitro at concentrations 3-4-fold greater (less potent) than mitomycin C. DNA single-strand breaks were also produced by each mitosene, but these lesions did not correlate with cytotoxicity and were less prominent than breaks produced by another monofunctional alkylating agent, methyl methanesulfonate. Mitosene-induced DNA strand breaks are probably due to excission-repair endonuclease activity and not from oxygen free radicals produced by redox cycling of the quinone moiety. There was no evidence of DNA-DNA cross-links by either 2,7-diaminomitosene.

Nucleotide derivatives of 2,7-diaminomitosene.

Treatment of mitomycin C with pyrimidine nucleotides in acidic media produced derivatives of 2,7-diaminomitosene in which C-1 was covalently bound to the phosphate group of the nucleotides. On reduction, these derivatives liberated the nucleotides and a mitomycin intermediate that alkylated DNA. Their reduction in the presence of 2'-deoxyguanosine produced some bifunctional alkylation as did mitomycin C. They were readily taken up by L1210 leukemia cells, in which they showed potent cytotoxicity. These properties suggest that they are acting as prodrugs capable of conversion into two active species. The uridylate derivative showed activity comparable to that of mitomycin C against P-388 leukemia in mice.

Mitomycin C and porfiromycin analogues with substituted ethylamines at position 7.

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.

Mitomycin C analogues with increased metal complexing ability.

Twenty-three new mitomycin C analogues designed to have increased metal complexing ability were synthesized and tested against P388 leukemia in mice. Their ability to complex Cu(II) was revealed by the shifts in their UV absorption spectra caused by this metal. One analogue was clearly more active than mitomycin C in the antitumor assay and two others had good activity. Correlation between antitumor activity and Cu(II) complexing ability was ambiguous. The most active compounds were either not complexers or they were complexers limited to the 2-(2-pyridyl)alkyl type substituent on N7. A variety of amino acid substituents on N7 showed only weak antitumor activity.

Novel antitumor 2-cyanoaziridine-1-carboxamides.

A set of 20 2-cyanoaziridine-1-carboxamides was synthesized from 2-cyanoaziridine and appropriate isocyanates. These compounds were active against a variety of solid and hematological tumor cells in culture, including strains resistant to doxorubicin and mitoxantrone. Their potencies in these assays correlated with the lipophilicity of substituents. The N-phenyl derivative was more potent and equally effective to imexon, a cyclized 2-cyanoaziridine-1-carboxamide of clinical interest, against cloned fresh human tumors.

1,4-disubstituted anthracene antitumor agents.

Three different types of 1,4-disubstituted anthracenes were synthesized, and their cytotoxicity in a panel of tumor cells was compared with that of the corresponding 9,10-disubstituted anthracenes. The panel contained human myeloma, melanoma, colon, and lung cancer cells and sensitive and multidrug-resistant murine L1210 leukemia cells. These compounds had [[(dimethylamino)ethyl]amino]methyl, N-[(dimethylamino)ethyl]carbamoyl, and carboxaldehyde (4,5-dihydro-1H-imidazol-2-yl)hydrazone side chains. The 1,4-diamide was more potent across the tumor panel than the corresponding 9,10-isomer, but the 1,4-diamine and the 1,4-hydrazone were less potent than their 9,10-isomers. Although the 1,4-hydrazone was active against P388 leukemia in mice, it was inactive against L1210 leukemia. Within each pair of compounds, the one with greater average potency against tumor cells gave a greater increase in the transition melt temperature of DNA.

6- and 7-substituted 2-[2'-(dimethylamino)ethyl]-1,2-dihydro-3H-dibenz[de,h] isoquinoline-1,3-diones: synthesis, nucleophilic displacements, antitumor activity, and quantitative structure-activity relationships.

New 2-[2'-(dimethylamino)ethyl]-3H-dibenz[de,h]isoquinoline-1,3-diones with substituents at the 6- and 7-positions were prepared. Nucleophilic aromatic displacement was a key reaction in the syntheses. Ten of the new compounds were more potent than the unsubstituted compound, azonafide, in a panel of tumor cells including human melanoma and ovarian cancer and murine sensitive and MDR L1210 leukemia. They also were less cardiotoxic in cell culture. Four of these compounds were not cross-resistant with the MDR leukemia, and one of them, 6-ethoxyazonafide, was nearly as potent against solid tumor cells as leukemia cells. These compounds also had good potency against human breast, colon, and lung cancer cells, including doxorubicin and mitoxantrone resistant cell lines. Advantages of the new analogues over azonafide were less in vivo, but 6-ethoxyazonafide was more effective against L1210 leukemia and subcutaneous B16 melanoma in mice. Although correlations of antitumor potency in cells and physicochemical properties of substituents were not found, there were statistically significant correlations of DNA melt transition temperature (delta Tm) with potency in solid tumor cells and sensitive and MDR resistant L1210 leukemia cells for 6-substituted azonafides and with solid tumors for 7-substituted azonafides.

Mitomycin C analogues with secondary amines at position 7.

A series of mitomycin C analogues with secondary amines at position 7 was prepared from mitomycin A. Eleven of the 20 new compounds in this series were more active than mitomycin C against P-388 murine leukemia, and 2 of these 11 were significantly less leukopenic. The two substituents conferring these superior properties were 4-formylpiperazine and 2-cyanoaziridine. No quantitative correlations could be made among antitumor activities and physicochemical properties of the analogues, although the relative ease of quinone reduction might be related to the good potencies (minimum effective doses) of many of them.

Mitomycin C analogues with aryl substituents on the 7-amino group.

A series of 30 different N7-phenyl-substituted mitomycin C analogues, including 25 new compounds, was prepared from mitomycin A. Seven of these compounds were clearly superior to mitomycin C in activity against P-388 murine leukemia. The para- and the meta-substituted derivatives were subjected to Hansch analysis, which revealed that the lipid-water distribution coefficient pi was the only significant factor in determining antitumor potency (MED). The substituent electronegativity factor sigma was statistically insignificant in determining potency, despite the good correlation of sigma p with the polarographic quinone-reduction potential. These results suggest that diffusion into the tumor cell or access to the receptor is more important than bioreductive activation in determining antitumor potency for this particular group of mitosanes . Fifteen new mitomycin C analogues with heterocycles on the 7-amino group also were prepared. Two of them, containing pyrazolyl and aminopyridyl substituents, were more active than mitomycin C against P-388 murine leukemia. No broad correlations could be made among the antitumor potencies and physicochemical properties for this type of analogue.

Cardiotoxicity of mitomycin A, mitomycin C, and seven N7 analogs in vitro.

The alkylating antitumor agents mitomycin A (MMA), mitomycin C (MMC), and seven N7 analogs were compared in terms of their cardiotoxic and antitumor activity in vitro. Neonatal rat-heart myocytes were sensitive to five of the compounds studied, including MMA, 7-dimethylamidinomitosane (BMY-25282), 7-(N-methyl-piperazinyl)-mitosane (RR-194), N7-(4-iodophenyl)-MMC (RR-208), and N7-(4-hydroxyphenyl)-MMC (M-83) in order of descending molar potency. MMA and RR-208 possessed the greatest cytotoxic potency against 8226 human myeloma tumor cells in vitro. Two of the nine mitomycins studied, BMY-25282 and M-83, showed greater cytotoxic potency for heart cells. For these two agents, the ratio of the 50% inhibitory concentration in heart cells to that in 8226 myeloma cells was 50 and 32, respectively. For the other analogs, the tumor-cell cytotoxic potency was much higher (ranging from 200 to 7,000). For the nine mitomycin compounds, a correlation was found between heart-cell toxicity and low reduction potentials (E1/2 values) ranging from -0.16 to -0.37 V. Thus, as the reduction potential decreased (easier reducibility), the cardiotoxic potency in vitro increased (r = 0.81). In contrast, mitomycins with reduction potentials of higher than -0.37 V were much less potent cardiotoxins. Thus, mitomycin C (E1/2 = -0.45 V) was noncardiotoxic even when tested at concentrations 100-fold above those pharmacologically achievable in humans. Mitomycin C also failed to enhance doxorubicin (Adriamycin) cardiotoxicity in vitro. Importantly, no correlation was found between the reduction potential and the antitumor activity of the nine analogs (n = 0.51), in this small series.