Pm. Nz@. Acad. Sci. USA Vol. 80, pp. 129-132, January 1983 Biochemistry Ascorbate-quinone interactions: Electrochemical, free radical, and cytotoxic properties (redox potential/semiquinone radical/cancer drug/electron transfer/ascites) RONALDPETHIG*,PETER R. C. GASCOYNE,JANE A. MCLAUGHLIN,ANDALBERT SZENT-GY~RGYI Labmatory of the National Foundation for Cancer Research at the Marine Biological Laboratory, Woods Hole. Massachusetts 02543 Contributed by Albert Scent-Gydrgyi, October 12,1982 ABSTRACT Standard midpoint potentials have been deter- mined for p-benxoquinone, methoxy-p-benzoquinone and 2,3-, 2,5-, and 2,6-dimethoxy-p-benxoquinones in aqueous solution. ESR studies have been made of the ascorbate and semiqninone radicals produced when these quinones interact with sodium as- corbate. Direct correlations are found between the electrocbem- ical potentials, generated semiquinone lifetimes, and cytotoxic action in Ehrlich ascites-bearing mice. This work follows from earher considerations of the oxidation- reduction and cytotoxic properties of some methoxyquinones (1) and ofcharge-transfer reactions and electronic delo~ization effects in biological systems (2-4). The compounds of interest in these particular studies have been the methoxy-substituted p-quinones, two of which (methoxy- and 2,6-dimethoxy-p-ben- zoquinone) occur naturally in wheat germ (5). Bachur et al (6) have suggested that the cytotoxic action of quinone anticancer drugs is mediated through their free-radical metabolites. An important first step in such activity is the pro- duction of relatively stable semiquinone free radicals. Radicals result from one-electron rather than two-electron redox reac- tions and the radical concentration and rate ofproduction should depend on the difference between the electrochemical poten- tials of the electron donor and acceptor molecules involved. In this work, electrochemical and ESR measurements have been made to determine the correlation between the cytotoxic activ- ity of several quinone/ascorbate mixtures and their ability to produce stable semiquinone radicals. MATERIALS AND METHODS Pure samples of ~-benzoquinone, methoxy-~-benzoquinone, and the three dimethoxy- (2,3-, 2,5-, 2,6-) p-benzoquinones, as well as the corresponding hydroquinones, were supplied by G. Fodor (West Virginia University, Morgantown). The redox potentials of the various quinone-hydroquinone couples were determined at 25oC by using electrodes of bright platinum wire (24 gauge). These were subjected to a nonoxi- dizing cleaning procedure (7) by placing them for 10 min in boiling 10% sodium bisulfite and then washing in distilled water. The reference half-cell consisted of a saturated solution of quinhydrone (Eastman Kodak) in nitrogen-saturated aqueous 0.01 M HC1/0.09 M KCI, and a 0.5 M KCl/5% agar salt bridge was used as the connection to the test half-cell The combined electrode and salt-bridge junction errors were measured to be less than 0.2 mV and, when calibrated against a platinized plat- inum/hydrogen electrode, the standard potential (pH 0) of the quinhydrone reference electrode was determined as 700 + 1 mV, in good agreement with the values 699.58 mV and 699.92 The publication costs of this article were defrayed in part by page charge payment. This article must therefore be hereby marked "a&e&se- tnent" in accordance with 18 U. S. C. 51734 solely to indicate this fact. mV obtained by earfier workers (89). The electrode potentials were measured by using a Keithley 616 digital electrometer (input impedance > 0.1 PfI) that had been calibrated to within 20.1 mV by using a Data Precision (model SlOQ) voltage cab- brator Fresh quinhydrone reference electrodes were prepared for each new experiment, which typically took less than 3 hr to perform. The standard midpoint potentials were obtained in two ways, either by placing various quinone/hydr~uinone mixtures in the test half-cell or by titrating quinone solutions against ascorbic acid (Sigma). For these measurements, both aqueous HCl/KCl and 10% methanol/go% HCl/KCl were used, and these were deoxygenated by perfusion with pure nitrogen. The concentrations of the quinones, hydroquinones, and ascorbic acid in the test half-cell were determined spectro- scopically by using a Beckman model 35 spectrophotometer in parallel experiments. The spectroscopic data obtained are given in Table I. Depending on solubility, quinone concentrations of 0.1-0.6 mM were used. The free radicals produced in the interaction of the various quinones with ascorbic acid were investigated by using a Varian El09 (ESR) spectrometer. For each measurement, a solution of 10 mM sodium ascorbate (Sigma) at pH 7.8 was flow mixed, using a vortex-flow mixing chamber (Varian E-249), with an aqueous 0.5 mM quinone solution. As with the electrochemical studies, solutions were deoxygenated by perfusion with pure nitrogen before mixing. The dynamics of the production and decay of the ascorbate and semiquinone radicals at 25oC was investigated for times between 10-r to IC? set by using the stopped-flow technique. During these dynamic measurements, the spectrometer was locked to a characteristic absorption line of the ESR spectrum by using the Varian {model E-272B) field/ frequency lock. Spin concentrations and g vahres were cali- brated against aqueous MnCl, and Wurster's blue perchlorate (g = 2.003) solutions, respectively. In the animal studies, female CD, mice (Charles River Breeding Laboratories) were inoculated with approximately IO' Ehrlich ascites tumor cells and randomized into groups of 10 mice each. This cell line is maintained in our laboratory by ~~splan~tion into CD, mice at 7-day intervals. Then, 24 hr after inoculation of the mice, the first of a 7-day course of twice- daily 0.25-ml intraperitoneal injections of a quinone/ascorbate mixture, which had been mixed only minutes earlier, was ad- ministered. On the 8th day, the mice were sacrificed and ex- amined for the presence of ascites. The quinone doses were based on the following criteria: methoxy-, relatively nontoxic and soluble, the dosages used were well below lethal; 2,3- and 2,6dimethoxy, dosages used were below those causing weight loss; 2,5-dimethoxy-, a saturated solution was used; benzoqui- none was given in half the lethal dosage. * Permanent address: School of Electronic Engineering Science, Uni- versity College ofNorth Wales, Bangor, Gwynedd, United Kingdom. 129 130 Biochemistry: Pethig e6 al: 750 700 9 E 650 .2 $ 600 2 550 500 450 I , , I 1 I 1 - I.0 -0.5 0 0.5 1.0 1.5 2.0 Log EJ- PH4 E`lG. 1. Electrochemical potential at 25oC in 10% methanol/901 H&KC1 as a function of the quinone (Q)~y~oquinone (QHal con- centration ratio, o , Quinone/hy~oquinone mixtures; o, quinone/as- corbic acid titrations. Curves: 1, p-benxoquinone; 2, methoxyquinone; 3, 2,3&methoxyquinone; 4, 2,5-dimethoxyquinone; 5, 2,6-dimeth- oxyquinone. RESULTS After conversion of the potentials measured with reference to the quinhydrone electrode to normal eIectrode potentials (hy- drogen electrode, pH 0), the electrochemical data for the 10% methanol solutions were plotted in the form of the Nernst equa- tion for the five quinones studied (Fig. 1). The slopes of the straight-line plots were found to be 29.5 c 0.5 mV, consistent with the formation of two-electron redox couples. The standard midpoint potentials derived from these plots are given in Table I, together with the results obtained using aqueous HCI/KCI. For the monomethoxy-, 2,3-, and 2,6-dimethoxy compounds, the quinone/hydroquinone mixtures gave results in good agreement with those of the ascorbic acid titrations. However, because ascorbic acid only partiahy reduced the 2,6-dimethoxy compound, the use of the spectroscopic data (Table 1) was re- quired to determine the precise quinone/hydroquinone ratios. These results show that at pH 2 the dominant products formed from two-electron reduction of the quinones by ascorbic acid are the corresponding hydroquinones and dehydroascorbic acid, For 2,5-dimethoxy quinone (the least soluble compound studied), the titrations against ascorbic acid did not give con- sistent and reproducible electrochemical results. If, to maxi- mize the balance, the highest concentration (aO.5 mM) of the a Proc. Natl Acad. Sci. USA 80 (1983) FSG, `2, ESRsi~alsof qui~one/ascorba~ mixtures. Cuwes: a and b, ~rba~/2,~dime~oxyquinone 5 and 25 sac after mixing; c and d, aacorbate/2,3-dimethoxyquinone 5 and 25 set aiter mixing. A, spec- tral featureof the ascorbate radical used in kinetic studies; Q, spectral feature of the aemiquinone radical used in kinetic studies; $ , g = 2.0052. 2,Gl.imethoxy compound was used in the titration, the elec- trochemical data did not produce a linear Nemst pIot. Spec- troscopic measurements for the range 0.05-0.5 mM gave in- consistent results, suggesting that factors such as dimerization or molecular association may have complicated the measure- ments. For this reason, the standard midpoint potential for the 2,5dimethoxy compound was obtained solely from measure- ment of quinone/hydroquinone mixtures. It was also found that the reducing power of ascorbic acid toward the 2,5-dimethoxy was less than that for the 2,6-isomer. With increasing pH, the electrochemical measurements were more difficult to carry out. Problems associated with rapid oxidation of the hydroquinones increased with increasing pH and, for this reason, the electro- chemical potentials given for pH 7.4 in Table 1 are based on adjustment of the pH 2 results. We have also obtained spectroscopic data using chloroform as the solvent and our results for monomethoxy quinone and 2,bdimethoxyquinone are in good agreement with those of Cosgrove et al (5). We consider that the peak at 332 nm (a = 1,670) reported for 2,6-dimethoxyquinone by Braude (11) and the extinction coefficient of 3,390 reported by Ku'Loek (12) for the peak around 370 nm for 2,5-dimethoxyquinone are in error. Examples of the ESR signals obtained 5 and 25 set after mixing of the quinones (pH 6) with sodium ascorbate (pH 7.8) are shown in Fig. 2 for 2,6- and 2,5-dimethoxyquinones. The pH of the mixtures was 7.6. The first signal to appear was that for the ascorbate free radical, and it was followed by that for the semiquinone radical; the identification and time courses of de- cay of these two radicals are shown in Figs. 2 and 3 and given in Table 2. We consider that the most significant result relevant Table 1. Standard midpoint potentials (E,) and spectroscopic data at 25oC E, @H 0) 90% MeOH/lO% &n A nlax f&l* HCI,`KCl HCl/KCI ($3 7.4) Quinone Hydroquinone p'Benzoqu.inone 699 700 261 296 (3501,248 (22,500) 290 (2,730) Methoxyquinone 601 603 163 370(1,400), 257 (16,300) 290 (3,390) 2,3"D~me#oxyquinone 621 622 183 4~(1,240~,257 (14,200) 284 (3,000) 2,5-Dimethoxyquinone 472 475 34 378 (5401, 286 (21,500) 296 (3,400) 2,6-Dimethoxyquinone 516 521 78 390 (5831, 286 (15,300) 290 (2,740) Ascorbic acid+ 47 *In 10% MeOH/90% HCI/KCI. + From ref. 10. Biochemistry: Pethig et al Proc. Natl. Acad. Sci. USA 80 (1983) 131 A ,.-.-.-.-.w*- o ??? ? 50 (00 150 200 Time (set) FIG. 3. Time course of decay of ascorbate (A) and semiquinone (B) free radicals produced by interaction of aacorbate with 2,3- (0) and 2,6- (0) dimethoxyquinones at 25oC and pH 7.6. to the animal studies is that, in the reaction with sodium as- corbate, 2,6- and 2,5-dimethoxyquinones produced relatively stable semiquinone radicals and the longest lived ascorbate rad- icals. Ifsemiquinone radicals were produced in the interactions with p-benzoquinone or with the methoxyquinone, then their lifetimes were less than 0.1 sec. The results obtained to date in the determination of cytotoxic efficacy of the various quinone/ascorbate mixtures toward as- cites cells are summarized in Table 3. Benzoquinone also was investigated, with and without ascorbate, and found to be in- active. Neither the quinones nor ascorbate when administered separately exhibited any appreciable cytotoxic activity, and only the 2,5- and 2,6-dimethoxyquinone/ascorbate mixtures were effective in eliminating Ehrlich ascites in a high percentage of the mice tested. Even from these incomplete studies with mice, there is a striking correlation between the cytotoxicity of the quinone/ascorbate mixtures and their ability to produce long- lived semiquinone radicals. Since at pH 7.4 the hydroquinones are likely to be rapidly oxidized to the corresponding quinones, hydroquinone/ascorbate mixtures were also given to the mice. The possible advantage of this was considered to be that pro- duction of the semiquinone radical would begin after, rather than before, injection of the mixture into the mice. CONCLUSIONS The electrochemical potentials given in Table 1 indicate that the two-electron reduction of 2,5- and 2,6-dimethoxyquinones Table 2. Peak concentrations and half-lives of ascorbate and semiauinone radicals pBenzcquinone Methoxyquinone 2,3-Dimethoxyquinone 2,5-Dimethoxyquinone 2.6-Dimethoxvauinone tion, mM Aecor- Semi- bate quinone >0.5 NO 0.25 NO 0.81 0.069 0.054 0.002 0.081 0.008 Half-life, eec Ascor- Semi- bate quinone 0.6 NO 4.5 NO 1.7 3.0 19.0 140.0 21.0 195.0 NO, not observable after 0.1 sec. Table 3. Summary of current resulta: Cytotoxic activity toward Ehrlich as&es-bearing mice of quinones with and without aacorbate Hydro- Ascor- Mice W Quinone, quinone, bate, tasted, ascites mM mM mM no. free Methoxy 2,3-Dimethoxy 2,5-Dimethoxy 2,6-Dimethoxy Aecorbate alone Control+ 3.6 750 10 10 3.6 10 20 18.0 750 20 0 18.0 10 20 0.6 750 10 0 0.6 750 10 0 0.6 10 0 =0.5* 750 10 50 ==0.5* 750 10 60 =0.5* 10 0 3.0 750 10 100 3.0 10 0 1.5 750 60 83 1.5 10 0 1.5 375 20 95 1.5 200 10 50 1.5 100 10 10 750 10 0 100 9 * Saturated. + Saline injected. by ascorbate will occur less rapidly than of the other quinones studied. This should favor the production of relatively stable radicals formed by one-electron reduction, in agreement with the ESR results. Because semiquinone radicals are more stable in basic than in acidic solution (13), the presence of transient one-electron reduction processes would not be expected to greatly modify the 29.5-mV gradient of the equilibrium Nemst plots of Fig. 1. There appears to be a direct correlation between the cyto- toxic action of the quinone/ascorbate preparations and the gen- eration of long-lived semiquinone free radicals (Tables 2 and 3). Sodium ascorbate was used as the nontoxic electron donor in these studies. The results of this work predict that, if the mo- lecular nature and hence the electrochemical potential of the electron donor is changed, then a corresponding modification of the quinone acceptor will be necessary to maintain the degree of one-electron reduction. This principle may have general ap- plication in the design of quinone-based anticancer drugs. The results presented here also lend support to the hypothesis (6) that semiquinone radicals have important cytotoxic properties. However, the real objective of our work is an increased un- derstanding of living processes at the submolecular (electronic) dimension. We wish to acknowledge valuable discussions with, and the careful preparative work of, Professor Gabor Fodor, without which this work would not have been possible, and we thank Mr. Richard Meany for his work in the animal studies. 1. Szent-Gybrgyi, A. (1982) Znt. J. Quantum Chem: Quant. Biol. Symp. 9,27-30. 2. Szent-GyBrgyi, A. (1976) Electronic Biology and Cancer (Dek- ker, New York). 3. Pethig, R. & Szent-GyBrgyi, A. (1977) Proc. N&Z. Acad. Sci. USA 74, 226-228. 4. Gascoyne, P. R. C., Pethig, R. & Szent-Gycrgyi, A. (1981) Proc. Nat/. Acad. Sci. USA 78, 261-265. 5. Cosgrove, D. J., Daniels, D. G. H., Whitehead, J. K. & Goulden. J. D. S. (1952)J. Chem. Sot., 4821-4823. 132 Biochemistry: Pethig et al. PI-oc. Natl. Acad. Sci. USA 80 (1983) 6. Bachur, N. R., Gordon, S. L. & Gee, M. V. (1976) Cancer Res. 10. Ball, E. G. (1937)]. Siol Chem. 118,219-239. x4,1745-1750. 11. Braude, E. A. (1945)J. Chem. Sot., 490-497. 7. Coons, C. C. (1931) Ind. Eng. Chem. Anal Ed. 3,402-407. 12. Bu'Lock, J. D. (1955)J. Chem. Sot., 575-577. 8. Harned, H. S. & Wright, D. D. (1933) I. Am. Chem. Sot. 55, 13. Yamazaki, 1. & Piette, L. H. (1965)J. Am. Chem. Sot. 87,986- 4849-4857. 990. 9. Hovorka, F. & Dearing, W. C. (1935)J.Am. Chem. Sot. 57,446- 453.