Prediction of initial reduction potentials of ... - ACS Publications

Prediction of initial reduction potentials of compounds related to anthracyclines and implications for estimating cardiotoxicity. Y. Kawakami, and A. ...
0 downloads 0 Views 426KB Size
244

Chem. Res. Toxicol. 1990, 3, 244-247

Prediction of Initial Reduction Potentials of Compounds Related to Anthracyclines and Implications for Estimating Cardiotoxicity Y. Kawakamit and A. J. Hopfinger* Deparatment of Medicinal Chemistry and Pharmacognosy, University of Illinois a t Chicago, Box 6998, Chicago, Illinois 60680 Received December 14. 1989

Eighteen anthracyclines representing several different structural classes were found in the literature to have their respective redox potentials measured under near-identical conditions. The initial reductions were determined in terms of half-wave potentials (HWPs). The electronic structures of these 18 anthracyclines were optimized, and the resulting electronic descriptor measures were regressed against the HWPs. A very significant linear correlation (QSAR) was found between H W P and the lowest unoccupied molecular orbital (LUMO) energy by using both the CNDO/2 and MNDO one-electron molecular orbital methods. These findings suggest that it may be possible to estimate the H W P of an anthracycline in advance of its synthesis. Insofar as the redox potential, as represented by HWP, is an indicator of the relative degree of cardiotoxicity of an anthracycline, it may also be possible to rank relative cardiotoxicity for chemical classes of anthracycline analogues in terms of the identified QSAR.

Introduction The biochemical mechanism that controls the cardiotoxic behavior of anthracyclines, an important chemical class of anticancer agents, is not completely understood. However, there is increasing evidence that the redox properties of these molecules may be central to this toxicity (1). In particular, the redox behavior of the quinone moiety appears to lead to the production of semiquinone and reactive oxygen species (2). The reactive oxygen species include superoxide anion, hydrogen peroxide, and hydroxyl radicals produced by secondary reactions. NADH dehydrogenase is thought to play a major role in this metabolic activation of anthracyclines (3). The hydroxyl radicals have been observed to damage nucleic acids (4-6) and to cause lipid peroxidation (7,8). In turn, lipid peroxides have been shown to induce cardiotoxicity in rats (9), and rat hearts accumulate anthracyclines more than liver and skeletal muscles (10). There is also evidence that anthracyclines are reduced to the semiquinone form at Complex I of the mitochondrial electron transport chain (3). This finding might explain the injury to cardiac mitochondria by anthracyclines. Certain chromophore-modified anthracyclines show a correlation between the suppression of redox activity and concomitant cardiotoxic effects. Examples to date include 5-imino analogues (1, 4,11,12) and anthrapyrazole analogues (13-16). Relative to doxorubicin, the modification of the central quinone to a quasi-iminoquinone results in a considerably reduced superoxide dismutase sensitive oxygen consumption in a rat liver microsomal system and a much greater resistance to electrochemical reduction (16). Because of these data which suggest the redox properties of anthracyclines may be responsible for cardiotoxicity, different assays and physical experiments have been developed to characterize the biological and chemical nature of anthracycline reduction. These tests can be divided into three classes: measures of microsomal oxygen consump*Correspondence should be addressed to this author. 'Permanent address: Eisai Co., Ltd., Tsukuba Research Laboratories, 1-3, Tokodai 5 chome, Tsukuba, Ibaraki 300-26, Japan.

tion, assays for superoxide anion and hydroxyl radical generation, and electrochemical measurements of redox potentials. In particular, there appears to be a direct relationship between the degree of cardiotoxicity and the ease of reduction. This relationship is crucial to the significance of performing redox measurements and to the work we report here. It occurred to us that it might be possible to compute electronic properties of structurally diverse anthracyclines which correlate with measured redox potentials. To whatever extent this should prove possible, we might be able to predict the relative redox potentials of new anthracyclines in advance of their syntheses. This information could then be used in deciding whether or not to synthesize a new anthracycline analogue, at least with respect to possible cardiotoxicity. Of course, this is predicated upon ease of reduction being indicative of the degree of cardiotoxicity.

Methods (A) Electrochemical Methods. Two groups of workers (I, 16, 17) report the redox potentials of various anthracyclines in aqueous media with respect t o a standard aqueous saturated calomel electrode a t 25 "C. Thus, while the experimental procedures are somewhat different for the two groups, the same conditions have been used to make redox measurements. Table I contains the half-wave potentials (HWPs) that have been assigned to the initial reduction of the 18 anthracyclines considered in this study. The structures of these compounds are given in Figure 1. It should be noted that the structures in Table I are abbreviated by maintaining the chromophore, but simplifying the side chains. The half-wave potential method (16) is a particular polarographic experimental procedure that seems well suited for analyzing the reduction properties of anthracyclines. (B) Computational Methods. The molecular geometries of each of the 18 compounds in Table I were built by using standard bond lengths and angles with the CHEMLAB-II molecular modeling package (18). These structures were initially optimized by using the molecular mechanics MMFF option in CHEMLAB-11. This is an extended version of the MMZ force field (19). No electrostatic interactions were considered in these initial optimizations. The resulting optimized structures were then processed through the CNDO/2 semiempirical molecular orbital program (20) to obtain

0893-228~/90/2703-0244$02.50/0 0 1990 American Chemical Society

Initial Reduction Potentials of Anthracyclines

Chem. Res. Toxicol., Vol. 3, No. 3, 1990 245

Table I. Half-Wave Potentials (HWPs) and LUMO Energy Levels of Antitumor Anthracyclines -HWP obsd. V -HWP Peters Showalter predicted, V residuals LUMO, kcal/mol et al. et al. CNDO/2 MNDO compounds (1986)b (1986)c eq 1 eq 2 eq 2 eq 1 -9.7006 -50.2253 0.456 0.3437 0.3749 0.1123 0.0811 anthraquinonea (AQS) 0.64 4.9364 -39.5602 0.0444 0.0035 0.6364 0.5956 doxorubicin (DXR) 0.6297 0.6034 4.6013 -39.1837 -0.0997 0.53 -0.0734 4-de-0-methyl-11-deoxy-DXR 4.0456 -37.0601 0.54 0.6186 0.6474 -0.0786 -0.1074 11-deoxy-DXR 0.0444 4.9364 -39.5602 0.64 0.625 0.6364 0.5956 0.0035 daunorubicin (DAU) 0.6297 0.6034 4.6013 -39.1837 0.53 -0.0997 -0.0734 4-de-0-methyl-1I-deoxy-DAU 0.6186 0.6474 4.0456 -37.0601 0.54 -0.0786 -0.1074 11-deoxy-DAU -0.0214 5-imino-DAU 0.6964 0.6777 0.675 7.9321 -35.5963 -0.0026 N,N-dibenzyl-5-imino-DAU 0.745 0.6964 0.6777 0.0486 7.9321 -35.5963 0.0673 5-imino-DXR 0.72 0.6964 0.6777 0.0236 7.9321 -35.5963 0.0423 5-imino-DAU-01 0.70 0.6964 0.6777 0.0036 0.0223 7.9321 -35.5963 mitoxantrone (MTX) 0.775 0.6514 0.7464 0.136 5.6834 -32.2810 0.0287 anthrapyrazole (APZ) 1.0089 1.0147 -0.0079 1.001 -0.0137 23.5597 -19.3068 7-OH-APZ 1.040 1.0230 1.0359 0.0170 24.2617 -18.2854 0.0041 10-OH-APZ 1.0036 1.0235 0.983 -0.0206 -0.0405 23.2938 -18.8845 7,10-(OH),-APZ 1.0162 1.0364 1.045 0.0288 23.9223 -18.2577 0.0085 7,9,10-(OH)rAPZ 1.0722 1.0359 1.085 0.0128 0.0491 26.7212 -18.2849 1.047 7,8,10-(OH),-APZ 1.0182 1.0210 0.0288 24.0226 -19.0018 0.0260 a 2-Anthraquinonesulfonic

acid sodium salt.

O R ?

@@ R1

0

0

OH

NH

OH

@@

OH

CH30 R,

5-imino-DAU N,N-dbenzyl-5-imineDAU 5-imino-DXR 5-imino-DAU-ol

R2

doxorubicin (DXR)

OCH3 4-de-Omethyl-11-deoxy-DXR OH 1ldeoxy-DXR OCH3 daunorubicin (DAU) OCH3

OH

4-de-Omethyl-11-deoxy-DAU OH 1ldeoxy-DAU OCH3

H H

OH 0

See ref 1. cSee ref 16.

OH H H

@@ OH 0 NHCH& mitoxantrone (MTX)

0

anthraquinone (AQS)

o

HWP = O.OBO(LUM0) CND0/2:

NCH& Z H

anthrapyrazole (APZ) 7-OH-APZ 7-OH 10-OH-APZ 10-OH 7,l O-(OH)z-APZ 7,lO-(OH)z 7,9,10-(OH)3-APZ 7,9,10-(OH)3 7,8,10-(OH)3-APZ 7,8,10-(OH)3 Figure 1. Structures of the 18 anthracyclines considered in the analyses. partial atomic charges. The initially optimized structures with atomic partial charges were then reoptimized in MMFF taking into account electrostatic interactions in the optimization. The reoptimized structures were then rerun through CNDO/2 and also optimized, yet a third time, by using MNDO (21). Sets of electronic descriptors including highest occupied molecular orbial energy (HOMO), lowest unoccupied molecular orbital energy (LUMO), (HOMO - LUMO) energy, and partial atomic charges were generated from both the CNDO/2 and MNDO calculations. All combinations of the electronic descriptors were correlated against the observed HWPs (1, 16) by using multidimensional linear regression analysis.

Results and Discussion The LUMO energy levels from both the CNDO/2 and

+ 0.538

(1)

N = 18 R = 0.957 SD = 0.066 F = 175.5

and HWP = O.OBl(LUM0)

q$fs03H

NHCH2+

MNDO calculations strongly correlate with the observed HWP. The correlation equations are

+ 1.414

(2)

MNDO: N = 18 R = 0.963 SD = 0.057 F = 206.3 The CND0/2 and MNDO LUMO energies are expressed in kilocalories per mole while the HWPs are given in volts. N is the number of compounds, R is the correlation coefficient, SD is the standard deviation of fit, and F is the statistical significance of fit. The observed and predicted HWPs and differences for HWP are given as part of Table I along with the computed values of the LUMO energies. Figure 2 contains plots of (a) eq 1 and (b) eq 2 as well as the observed HWP values in HWP versus LUMO space. It can be argued that the 18 compounds studied separate into two major clusters-namely, the anthrapyrazoles and the rest of the anthracyclines. Closer analysis indicates that the non-anthrapyrazoles separate into 5-imino analogues and the remainder of the compounds. This clustering behavior is unfortunate in terms of applicability of regression analysis. It would be better if the HWP and LUMO values were uniformly distributed over their respective ranges. Nevertheless, the regression equations are quite significant. Moreover, the clustering behavior indicates that HWP is, as one expects, largely characteristic of the central ring structure and only depends in a minor way on ring substituents. The corresponding clustering of LUMO values indicates that the same i s true for LUMO. The findings of this study are consistent with the combined experimental and theoretical studies of Tempczyk et al. (22),where superoxide anion radical formation of anthraquinone antibiotics may occur by single-electron transfer. Quinones form an important group of substrates for several enzymes, including NADPH-cytochrome P-450 reductase, NADH dehydrogenase, xanthine oxidase, and NADH-cytochrome b5 reductase and undergo one-electron

246 Chem. Res. Toxicol., Vol. 3, No. 3, 1990

Kawakami and Hopfinger 91441-19-9; 7,9,10-(OH)3-APZ, 91441-42-8; 7,8,10-(OH),-APZ, 91441-43-9.

References

10

(1) Peters, J. H., Gordon, G. R., Kashiwase, D., Lown, J. W., Yen,

3

0.4

AOS

4 10

0

20

10

30

LUMO by CNDO ( k c a h o l ) 12

b

g T

,

/

L1

-HWP

5-IMINOs

m/

DXRa,DAUo

04 .60

.50

-40

.30

-20

-10

LUMO b y MNDO (kcalimol)

Figure 2. LUMO versus -HWP for (a) the CNDO/2 study and (b) the MNDO analysis.

reduction (23-27). Half-wave potentials, considered an approximate value of the free energy of reduction, were found to be negatively correlated with both the results of the microsomal oxygen consumption test and the production of superoxide anion in the chemical test system in this study ( I ) . It is possible that the LUMO value indicates the ease of initial one-electron reduction from the correlation between HWP and LUMO, on the basis of one-electron MO methods. Moreover, the finding that a single calculated molecular property, LUMO, can be used to explain the variance in HWP for a structurally diverse set of anthracyclines may have useful drug design applications. It should be possible to predict the HWP and thereby infer the cardiotoxicity of a new anthracycline relative to the 18 investigated in this work, in advance of its synthesis. It was the primary objective of this study.

Acknowledgment. Y.K. gratefully acknowledges the financial support of Eisai Co., Ltd. for his industrial sabbatical leave to UIC. This work was carried out with financial support from the Laboratory of Computer-Aided Molecular Modeling and Design at UIC. All calculations were done by using the CHEMLAB-11 molecular modeling package. Registry No. AQS, 84-48-0; DXR, 23214-92-8; 4-de-0methyl-11-deoxy-DXR, 81382-05-0; 11-deoxy-DXR, 71800-89-0; DAU, 20830-81-3; 4-de-O-methyl-ll-deoxy-DAU, 92237-37-1; 11-deoxy-DAU, 84325-15-5; 5-imino-DAU, 72983-78-9; N,N-dibenzyl-5-imino-DAU, 126754-59-4; 5-imino-DXR, 84275-95-6; B-imino-DAU-ol,105929-26-8 MTX, 65271-80-9; APZ, 91440-30-1; 7-OH-APZ, 88303-60-0; 10-OH-APZ, 88303-63-3; 7,10-(OH),-APZ,

S.-F., and Plambeck, J. A. (1986) Redox Activities of Antitumor Anthracyclines Determined by Microsomal Oxygen Consumption and Assays for Superoxide Anion and Hydroxyl Radical Generation. Biochem. Pharmacol. 35, 1309-1323. (2) Lown, J. W. (1983) The 'Mechanism of Action of Quinone Antibiotics. Mol. Cell Biochem. 55, 17-26. (3) Davies, K. J. A., and Doroshow, J. H. (1986) Redox Cycling of Anthracyclines by Cardiac Mitochondria. I. Anthracycline Radical Formation by NADH Dehydrogenase. J . Biol. Chem. 261, 3060-3067. (4) Lown, J. W., Chen, H.-H., Plambeck, J. A., and Acton, E. M. (1979) Diminished Superoxide Anion Generation by Reduced 5Iminodaunorubicin Relative to Daunorubicin and the Relationship to Cardiotoxicity of the Anthracycline Antitumor Agents. Biochem. Pharmacol. 28, 2563-2568. (5) Berlin, V., and Haseltine, W. A. (1981) Reduction of Adriamycin to a Semiquinone-Free Radical by NADPH Cytochrome P-450 Reductase Produces DNA Cleavage in a Reaction Mediated by Molecular Oxygen. J. Biol. Chem. 256,4747-4752. (6) Favaudon, V. (1982) On the Mechanism of Reductive Activation in the Mode of Action of Some Anticancer Drugs. Biochimie 64, 457-464. (7) Mimnaugh, E. G., Kennedy, K. A., Trush, M. A., and Sinha, B. K. (1985) Membrane Lipid Peroxidation in Isolated Rat Nuclei. Cancer Res. 45, 3296-3301. (8) Thayer, W. S. (1984) Serum Lipid Peroxides in Rats Treated Chronically with Adriamycin. Biochem. Pharmacol. 33, 2259-2263. (9) Mizukami, M., Aono, J., Saki, K., Hata, S., and Nakano, M. (1984) Experimental Myocardial Infarction: Effects of a Lipid Peroxide, 13-Hydroperoxy Linoleic Acid on Coronary Circulation in Rats. Artneim-Forsch. 34 (I), 569-577. (10) Peters, J. H., Gordon, G. R., Kashiwase, D., and Acton, E. M. (1981) Tissue Distribution of Doxorubicin and Doxorubicinol in Rats Receiving Multiple Doses of Doxorubicin. Cancer Chemother. Pharmacol. 7, 65-71. (11) Acton, E. M., and Tong, G. L. (1981) Synthesis and Preliminary Antitumor Evaluation of 5-Iminodoxorubicin. J. Med. Chem. 24, 669-673. (12) Tong, G. L., Henry, D. W., and Acton, E. M. (1979) 5-Iminodaunorubicin. Reduced Cardiotoxic Properties in an Antitumor Antracycline. J . Med. Chem. 22, 36-39. (13) Showalter, H. D. H., Johnson, J. L., Werbel, L. M., Leopold, W. R., Jackson, R. C., and Elslager, E. F. (1984) 5-[(Aminoalkyl)amino]-Substituted Anthra[l,9-cd]pyrazol-6(2H)-onesas Novel Anticancer Agents. Synthesis and Biological Evaluation. J . Med. Chem. 27, 253-255. (14) Showalter, H. D. H., Johnson, J. L., Hoftiezer, J. M., Werbel, L. M., Schillis, J. L., and Plowman, J. (1984) 5-[(Aminoalkyl)amino]-Substituted Antra[l,9-cd]pyrazol-6(2H)-onesas Novel Anticancer Agents. Am. Assoc. Clin. Res. Abstr., 352-352. (15) Fry, D. W., Boritzki, T. J., Besserer, J. A., and Jackson, R. C. (1985) I n vitro DNA Strand Scission and Inhibition of Nucleic Acid Synthesis in L1210 Leukemia Cells by a New Class of DNA Complexes, the Antra[1,9cd]pyrazol-6(2H)-ones(Antrapyrozoles). Biochem. Pharmacol. 34, 3499-3508. (16) Showalter, H. D. H., Fry, D. W., Leopold, W. R., Lown, J. W., Plambeck, J. A., and Reszka, K. (1986) Design, Biochemical Pharmacology, Electrochemistry and Tumour Biology of AntiTumour Anthrapyrazoles. Anti-Cancer Drug Des. 1, 73-85. (17) Rao, G. M., Lown, J. W., and Plambeck, J. A. (1977) Electrochemical Studies of Antitumour Antibiotics. 11. Polarographic and cyclic Voltammetric Studies of Mitomycin C. J. Electrochem. SOC.124, 195-204. (18) Pearlstein, R. A. (1988) CHEMLAB-II User's Guide-V 10.0, Chemlab Inc., 1780 Wilson Dr., Lake Forest, IL 60045. (19) Allinger, N. L. (1977) Conformational Analysis. 130. MHz. A Hydrocarbon Force Field Utilizing VI and V2 Torsional Terms. J . Am. Chem. Soc. 99, 8127-8132. (20) Pople, J. A., and Beveridge, D. C. (1970) Approximate Molecular Orbital Theory, McGraw-Hill, New York. (21) Dewar, M. J. S., and Thie!, W. (1977) Ground States of Molecules. 38. The MNDO Method. Approximations and Parameters. J . Am. Chem. SOC.99, 4899-4908.

Initial Reduction Potentials of Anthracyclines (22) Tempczyk, A., Tarasiuk, J., Ossowicki, T., and Borowski, E. (1988) An Alternative Concept for the Molecular Nature of the Peroxidating Ability of Antracycline Anti-Tumor Antibiotics and Antraacenodiones. Anti-Cancer Drug Des. 2, 371-385. (23) Bachur, N. R., Gordon, S. L., Gee, M. V., and Kon, H. (1979) NADPH Cytochrome P-450 Reductase Activation of Quinone Anticancer Agents to Free Radicals. Proc. Natl. Acad. Sci U.S.A. 76,954-957. (24) Lown, J. W., and Chen, H.-H. (1981) Electron Paramagnetic Resonance Characterization and Conformation of Daunorubicin Semiquinone Intermediate Implicated in Anthracycline Metabolism, Cardiotoxicity, and Anticancer Action. Can. J. Chem. 59,

Chem. Res. Toxicol., Vol. 3, NO.3, 1990 247 3212-3217. (25) Karasch, E. D., and Novak, R. F. (1981) Anthracenedione Activation by NADPH-Cytochrome P-450 Reductase; Comparison with Antracyclines. Biochem. Pharmacol. 30, 2881-2884. (26) Karasch, E. D., and Novak, R. F. (1983) Bis(alky1amino)anthracenedione Antineoplastic Agent Metabolic Activation by NADPH-Cytochrome P-450 Reductase and NADH Dehydrogenase: Diminished Activity Relative to Antracyclines. Arch. Biochem. Biophys. 224,682-694. (27) Doroshow, J. H. (1983) Effect of Anthracycline Antibiotics on Oxygen Radical Formation in Rat Heart. Cancer Res. 43, 460-471.