Proc. Natl. Acad. Sci. USAVol. 83, pp. 4514-4518, June 1986Medical Sciences

Role of hydrogen peroxide and hydroxyl radical formation in thekilling of Ehrlich tumor cells by anticancer quinones

(doxorubicin/mitomycin C/diaziridinylbenzoquinone/Ehrlich carcinoma cells)

JAMES H. DOROSHOWDepartment of Medical Oncology and Therapeutics Research, City of Hope National Medical Center, Duarte, CA 91010-0269

Communicated by Rachmiel Levine, January 17, 1986

ABSTRACT The cytotoxicity of the clinically importantantineoplastic quinones doxorubicin, mitomycin C, and diazi-ridinylbenzoquinone for the Ehrlich ascites carcinoma wassignificantly reduced or abolished by the antioxidant enzymescatalase and superoxide dismutase, the hydroxyl radical scav-engers dimethyl sulfoxide, diethylurea, and thiourea, and theiron chelators deferoxamine, 2,2-bipyridine, and diethyl-enetriaminepentaacetic acid. However, tumor cell killing by5-iminodaunorubicin, a doxorubicin analog with a modifiedquinone function that prohibits oxidation-reduction cycling,was not ameliorated by any of the free radical scavengerstested. Furthermore, treatment of intact tumor cells withdoxorubicin, mitomycin C, and diaziridinylbenzoquinone butnot 5-iminodaunorubicin generated the hydroxyl radical, or arelated chemical oxidant, in vitro in a process that requiredhydrogen peroxide, iron, and intact tumor cells. These resultssuggest that drug-induced hydrogen peroxide and hydroxylradical production may play a role in the antineoplastic actionof redox active anticancer quinones.

Quinone-containing antineoplastic agents, including doxo-rubicin (DOX), mitomycin C (MMC), and diaziridinylbenzo-quinone (AZQ), constitute one of the most important groupsof chemicals used in cancer chemotherapy because theypossess significant therapeutic action in the treatment ofmosthematologic malignancies as well as carcinomas of the lung,breast, ovary, brain, and gastrointestinal tract (1-3). Al-though the mechanism oftumor cell killing by these drugs hasbeen related to their effects on DNA structure or synthesis asa consequence of intercalation or alkylation (2-4), recentevidence suggests that cytotoxicity, at least for DOX, mayalso be due to lethal events occurring at extranuclear sites (5,6). Furthermore, despite studies demonstrating reactive ox-ygen production and DNA damage after cyclical reductionand oxidation of the quinone function in the presence ofpurified NADPH-cytochrome P-450 reductase (7), NADHdehydrogenase (8), and xanthine oxidase (9, 10), conclusiveevidence regarding a role of the quinone moiety in theanticancer action of these compounds for intact cells islacking.

Quite recently, however, production of potent oxidizingspecies, including the hydroxyl radical ( OH), has beendemonstrated during treatment of intact human MCF-7breast cancer cells with DOX (11). Furthermore, resistanceto the cytotoxic effect of DOX in this cell line apparentlyinvolves, at least in part, a significant enhancement ofintracellular antioxidant defense enzymes (12). These pre-liminary results, taken together with studies from this labo-ratory indicating that DOX, MMC, and AZQ stimulatesuperoxide anion and H202 production by microsomal,mitochondrial, and nuclear preparations from Ehrlich tumor

cells, strongly suggest that oxygen radical metabolism mightplay a role in the antineoplastic action of the anticancerquinones (13-15).For these reasons, the effect of a wide range of reactive

oxygen scavengers on the toxicity of the major antineoplasticquinones for Ehrlich carcinoma cells was examined. Agentsthat detoxify H202 or *OH, including proteins that do notpenetrate the cell surface, significantly reduced or abolishedthe cytotoxicity of these drugs. Thus, both extracellular andintracellular oxygen radical attack could be involved in thetumor cell toxicity produced by anticancer quinones. Inaddition, tumor cell killing was diminished by antioxidantsonly for drugs that generated reactive oxygen species fromintact cells, suggesting that cytotoxicity from H202 or *OH isa specific explanation for at least part of the antineoplasticeffect of these chemotherapeutic agents. The data for thisstudy have been published in abstract form (46, 47).

MATERIALS AND METHODSMaterials. Doxorubicin hydrochloride was purchased from

Adria Laboratories (Wilmington, DE). 5-Iminodaunorubicinand AZQ were supplied by the Drug Synthesis and ChemistryBranch, Division of Cancer Treatment, National CancerInstitute, Bethesda, MD. MMC was obtained from BristolLaboratories (Syracuse, NY). Dimethyl sulfoxide (Me2SO),diethylenetriaminepentaacetic acid (DTPA), and superoxidedismutase (SOD) [2750 units/mg, as assayed by the methodof McCord and Fridovich (16)] were purchased from Sigma.Diethylurea, thiourea, and urea were obtained from Aldrich.2,2-Bipyridine was purchased from Baker. Dulbecco's mod-ified Eagle's medium (DME medium), Ham's nutrient mix-ture F12 (F12 medium), and heat-inactivated fetal calf serum(FCS) were from Grand Island Biological (Grand Island,NY). The synthetic, selenoorganic compound 2-phenyl-1,2-benzisoselenazol-3(2H)-one (PZ 51), which possesses gluta-thione peroxidase-like activity (17, 18), was a gift of E. Graf(A. Nattermann and Cie Gmb. H, Cologne, F.R.G.). Themean iron concentration of a 1:1 (vol/vol) mixture of DMEmedium/F12 medium containing 10% FCS was 4 AM whenfour different lots of each medium were combined andexamined by atomic absorption spectroscopy.

Cell Lines. Ehrlich-Lettre tumor cells were maintained byweekly passage of one million cells i.p. in 20-g femaleSwiss-Webster mice. For experiments examining the effectof anticancer quinones on cellular - OH production, tumorcells were harvested 5-6 days after implantation, washedtwice in 0.9% NaCl, exposed to hypotonic shock lysis toremove contaminating erythrocytes (19), and resuspended in

Abbreviations: Me2SO, dimethyl sulfoxide; DOX, doxorubicin;MMC, mitomycin C; AZQ, diaziridinylbenzoquinone; SOD, super-oxide dismutase; FCS, fetal calf serum; DTPA, diethylenetriamine-pentaacetic acid; PZ 51, 2-phenyl-1,2-benzisoselenazol-3(2H)-one;IC50, drug concentration producing a 50% inhibition of clonogeniccell growth.

4514

The publication costs of this article were defrayed in part by page chargepayment. This article must therefore be hereby marked "advertisement"in accordance with 18 U.S.C. §1734 solely to indicate this fact.

Dow

nloa

ded

by g

uest

on

Nov

embe

r 27

, 202

0

Proc. Natl. Acad. Sci. USA 83 (1986) 4515

a 1:1 (vol/vol) mixture of DME medium and F12 mediumcontaining 10% FCS (DME medium/F12 medium/FCS). Cellviability (routinely >95%) was confirmed by exclusion of0.1% trypan blue dye.For experiments examining the cytotoxic effect of

anticancer quinones, Ehrlich cells were transferred to tissueculture by plating 0.5 x 106 cells in 25 ml ofDME medium/F12 medium/FCS supplemented with 15 mM Hepes bufferand 1.2 g of sodium bicarbonate per liter at pH 7.4; the cellswere maintained in suspension culture at 37°C in a humidified5% CO2 in air atmosphere. Tumor cells were used for drugexperiments 3 days after plating 0.5 x 106 cells per 25 ml ofmedium in 75-cm2 plastic flasks.The catalase level of Ehrlich carcinoma cells in logarith-

mic-phase growth incubated for 1 hr with 3000 units of theenzyme per ml was also determined in triplicate. In theseexperiments, 1 x 108 tumor cells were homogenized with 20strokes of a Dounce homogenizer and centrifuged at 104,000x g and 4°C for 60 min; the supernatant was assayed forcatalase activity as described (20) and was compared, intriplicate, to paired samples incubated without exogenouscatalase.Drug Treatment and Clonogenic Cell Survival Determina-

tions. The antitumor effect of anticancer quinones wasexamined by exposure of Ehrlich cells in the tissue cultureflask to a constant (36 ul) volume of drug for 1 hr; treatmentof the cells with either the anticancer agent or a free radicalscavenging agent or both always occurred in a total volumeof 35 ml ofDME medium/F12 medium/FCS. Briefly, 3 daysafter plating, paired groups of Ehrlich tumor cells in suspen-sion culture (1-2 x 106 cells per ml) exponentially growing inDME medium/F12 medium/FCS were treated with eitherDME medium/F12 medium/FCS (10 ml) or an equal volumeof medium with the scavenger under investigation for 1 hr;cells were then exposed to either 36 ,ul of medium or to anequal volume of the anticancer quinone for 1 hr. At thecompletion of drug exposure, cells were washed twice infresh medium and resuspended in 5 ml ofDME medium/F12medium/FCS. The Ehrlich cells were counted in a hemo-cytometer and cell viability (in all cases >95%) was measuredby trypan blue dye exclusion.

Cell survival after drug treatment for the tumor cells wasthen determined by soft agar cloning as described (21).Dilutions of 1000 and 10,000 cells were prepared in theresuspension medium and plated in triplicate in 35 x 10 and60 x 15 mm Petri dishes, respectively. For these studies, thefeeder layer consisted of DME medium/F12 medium/FCSand 0.5% agar; the overlayer contained the cells, DME

medium/F12 medium/FCS, and 0.3% agarose. Colonies of>50 cells were counted 6 days after plating. The percentageof clonogenic cell survival has been expressed as the numberof colonies produced by drug-treated cells divided by thenumber of colonies produced by control cells (corrected forcloning efficiency) x 100. The plating or cloning efficiency ofthe Ehrlich cells in this system [which is defined as the ratioof the number of colonies counted divided by the number ofcells initially plated (22)] ranged from 40% to 75%. Detailedcontrol experiments revealed that in the concentrations usedfor these studies none of the free radical scavenging agentsproduced any significant effect on the plating efficiency ofthetumor cells.Measurement of * OH Production. The formation of * OH, or

an oxidizing species with the chemical reactivity of * OH, byEhrlich tumor cells treated with anticancer quinones wasassessed by measurement ofCH4 production from Me2SO byusing a modification of the method of Repine et al. (23, 24) aspublished recently (8). For these experiments, the 2-mlmixtures contained 2 x 107 tumor cells and 100 mM Me2SOin DME medium/F12 medium/FCS, and the concentrationsof reactive oxygen scavengers and anticancer quinonesshown in Table 3; reactions were initiated by addition of theanticancer quinone into the gas-tight, siliconized reactionvessels by injection through the red rubber stopper. Incuba-tion at 37°C was for 2 hr in a shaking water bath and reactionswere terminated by placing the experimental tubes on meltedice. A 500-,lI sample of the headspace gas from each tube(total volume, 1 ml) was assayed for methane concentration.

Statistical Analyses. Data were analyzed with the two-tailedStudent's t test for independent means [not significant, P >0.05 (25)]. Data have been expressed as the mean ± 1 SEM.

RESULTS

Effect of Radical Scavengers and Antioxidant Enzymes onAnticancer Quinone Cytotoxicity. The initial experimentsinvestigated whether several potent OH scavengers alteredthe cytotoxicity of various antineoplastic quinones (Table 1).Diethylurea, thiourea, and Me2SO significantly inhibitedtumor cell killing by MMC, AZQ, and DOX; althoughthiourea was the most potent inhibitor of cytotoxicity,Me2SO produced complete protection against cell killing byDOX at a concentration of 300 mM, and diethylurea pro-duced almost total protection for cells treated with DOX orMMC. It is important to point out that Me2SO, itself, mayproduce many biophysical changes in tumor cells in additionto scavenging *OH and that these effects might make cells

Table 1. Effect of reactive oxygen scavengers on Ehrlich carcinoma cell kill by anticancer quinonesReactive oxygen scavenger, % of control colonies

HI HI Diethyl-Anticancer Conc., Catalase, catalase,* SOD, SOD,* Me2SO Thiourea, urea, Urea, PZ 51,quinone AhM None 3000 U/ml 3000 U/ml 50 tzg/ml 50 ,tg/ml 300 mM 450 mM 200 mM 200 mM 200 mM 10 ,iMMMC 2.5 50 ± 3 88 ± St 58 ± 5 89 ± 4* 56 ± 3 60 ± lt 69 ± 3§ 98 ± 6§ 84 ± 10§ 43 ± 4 86 ± 4§AZQ 7.5 46 ± 3 72 ±1l 49 ± 1 51 ± 1 48 ± 2 63 ± 1§ 65 ± 1§ 94 ± 4§ 64 ± 2§ 52 ± 3 83 ± 3§DOX 1.5 51 ± 2 75 ± 1§ 52 ± 2 72 ± 2§ 58 ± 4 104 ± 1§ 71 ± 1§ 86 ± 3§ 87 ± 3§ 48 ± 3 94 ± 6§All anticancer quinones except AZQ were initially dissolved fresh in glass-distilled, deionized water and protected from light before use. AZQ

was reconstituted in a 20% (vol/vol) dimethylacetamide stock solution; when used for these experiments, the final concentration ofdimethylacetamide was never above 0.02% (vol/vol), which had no effect on tumor cloning efficiency. All reactive oxygen scavengers wereprepared in tissue culture medium except PZ 51, which was initially dissolved in concentrated form in 100% (vol/vol) Me2SO; in theseexperiments, the final concentration ofMe2SO, when used with PZ 51 was 5 mM, which was below the protective threshold for this agent. Resultsrepresent the mean ± SEM of three to six experiments for each drug with every scavenger; significance was determined by using the two-tailedt test for independent means. P values are indicated for the comparison between clonogenic cell survival after treatment with the anticancerquinone in the presence and the absence of a reactive oxygen scavenger. Conc., concentration; U, units.*Catalase and SOD were heat-inactivated (HI) by autoclaving for 60 min.tP < 0.05.tP < 0.02.§p < 0.01.

Medical Sciences: Doroshow

Dow

nloa

ded

by g

uest

on

Nov

embe

r 27

, 202

0

4516 Medical Sciences: Doroshow

more resistant to DOX toxicity (26). However, recent inves-tigations have demonstrated no amelioration of DNA strandbreakage by Me2SO in cells exposed to DOX (27). Further-more, urea, a structural analog of thiourea and diethylureathat is an ineffective -OH trapping agent (23), produced nosignificant effect on anticancer quinone cytotoxicity. Theseresults, taken together, support the hypothesis that *OHformation contributed to Ehrlich tumor cell killing in theabove experiments.

Catalase and SOD, scavengers of H202 and superoxideanion, respectively, significantly inhibited tumor cell killing;since the heat-denatured proteins were not protective, it isunlikely that these results were simply due to a nonspecificprotein effect or to alterations in drug uptake by the antiox-idant enzymes. Furthermore, over a wide range of drugconcentrations, catalase significantly inhibited tumor cellkilling for MMC, AZQ, and DOX (Fig. 1). It is of someinterest that the drug concentrations found to inhibitclonogenic cell survival by 50% (IC50; Table 1) approximatequite closely the peak plasma concentrations of these agentsafter i.v. administration in man (1-3, 28, 29). At the IC50 orlower, catalase produced substantial protection against cellkilling; at higher drug concentrations, the extracellularantioxidant enzyme inhibited cytotoxicity to a lesser, albeitsignificant, extent (Fig. 1). Inhibition of drug-related cyto-toxicity by catalase occurred even though no evidence wasfound for the uptake of catalase into Ehrlich tumor cells inthese experiments (data not shown). These studies suggestthat the anticancer quinones may kill tumor cells by multiplemechanisms; however, H202-mediated events could predom-inate at lower drug concentrations.As shown in Table 1, PZ 51, a lipophilic organoselenium

compound that possesses glutathione peroxidase-like activ-ity for intact cells and that can effectively detoxify H202 inthe presence of glutathione (17, 18), significantly reduced thecytotoxicity of the anticancer quinones for Ehrlich cells. Asa cytoprotective agent, PZ 51 was superior to catalase withrespect to cell killing by DOX and AZQ.

In addition to scavengers of OH and H202, three ironchelating agents were examined to determine whether theycould affect tumor cell killing by anticancer quinones (Table2). Each of these agents complexes iron in a form that isunavailable for participation in reactions generating *OH (30,31); however, DTPA is incapable of penetrating the cellsurface (32), bipyridine enters cells freely (31), and deferox-amine enters only certain mammalian cells but is capable ofassociating closely with the plasma membrane (33-35).Deferoxamine and bipyridine reduced the toxicity of MMC,

100

80

to

LuJ

0

60

0

0

20

0 I)I.O)IJe)D Dto4.. V

DOXORUBICIN (pM) DI

Table 2. Effect of iron chelating agents on Ehrlich carcinomacell kill by anticancer quinones

Iron chelator, % of control colony survival

Anticancer Conc., Deferoxamine, DTPA, Bipyridine,quinone IiM None 50,MM 150 AM 150 gMMMC 2.5 50 ± 3 83 ± 8* 70 ± lt 60 ± ltAZQ 7.5 46 ± 3 60 ± 3t 46 ± 9 60 ± ltDOX 1.5 51 ± 2 72 ± 2t 62 ± 2* 62 ± lt

These experiments were performed as described in the legend toTable 1; results represent the mean ± SEM of three to six experi-ments for each drug with every iron chelating agent. P values areindicated for the comparison between clonogenic cell survival aftertreatment with the anticancer quinone in the presence and theabsence of an iron chelating agent. Conc., concentration.*P < 0.05.tP < 0.01.

AZQ, and DOX, whereas DTPA did not reduce Ehrlichtumor cell killing by AZQ (Table 2). These experimentssuggest that the site of interaction between iron and theanticancer quinone (intracellular, extracellular, or cell mem-brane) may vary from drug to drug. However, since the levelof iron in the tissue culture media (4 kkM), which could alsobe present in vivo under certain conditions (36), has beendemonstrated to be sufficient to catalyze *OH production inthe presence of H202 (36), these experiments support thepremise that iron-related reactive oxygen metabolism may beinvolved in tumor killing by anticancer quinones. Finally,none of the -OH scavengers, antioxidant enzymes, or ironchelators tested in Tables 1 and 2 produced any significanteffect on the cytotoxicity of 5-iminodaunorubicin (IC50 = 4MtM), an anthracycline analog that does not generate freeradicals (data not shown). Further, the cytotoxicity of AZQwas not inhibited by SOD (Table 1). Thus, the antioxidantproteins and chemicals used in this study did not produce ageneral, nonspecific inhibition of drug-related tumor cellkilling in these experiments.

Effect of Anticancer Quinones on *OH Production by EhrlichCarcinoma Cells. To explain the effect of OH scavengers andantioxidant proteins on anticancer quinone cytotoxicity, *OHformation after treatment of intact cells with these drugs wasmeasured. In previous studies from this laboratory, MMC,AZQ, and DOX but not 5-iminodaunorubicin have beenshown to stimulate cyanide-resistant oxygen consumption byintact Ehrlich tumor cells and superoxide anion formation byEhrlich tumor microsomes, mitochondria, and nuclei (13-15). By using a well-described assay for *OH [or an oxidant

i-SCONTROL\ ---oCATALASE

I'''

FIG. 1. Effect of catalase onthe cytotoxicity of MMC (Left),DOX (Center), and AZQ (Right)for Ehrlich carcinoma cells. Re-sults represent the mean ± SEMof three to six experiments foreach drug dose with and withoutcatalase pretreatment. For each

,____,___,___,___,______ drug, catalase significantly re-2.5 5.0 7.5 100 12.5 15.0 duced cytotoxicity (at least P <

IAZIRIDINYLBENZOQUINONE (suM) 0.05) for every dose tested.

Proc. Natl. Acad. Sci. USA 83 (1986)

MITOMYCIN C (,uM)

Dow

nloa

ded

by g

uest

on

Nov

embe

r 27

, 202

0

Proc. Natl. Acad. Sci. USA 83 (1986) 4517

species with identical chemical characteristics (23, 24)], itwas found that the agents capable of inhibiting tumor cellkilling by anticancer quinones significantly decreased -OHproduction by intact, drug-treated cells (Table 3). Overall,thiourea was the most powerful inhibitor studied; this may beexplained by the ability of this compound to react with anddetoxify H202 and -OH (37). In these experiments, no -OHwas detected in the absence of an anticancer quinone; -OHproduction was abolished similarly by heat-denaturation ofthe carcinoma cells (Table 3). Hydroxyl radical productionwas studied using drug concentrations substantially higherthan those employed for cytotoxicity experiments. This wasnecessary in large part because the detecting system usedhere for OH is relatively inefficient. As shown (8), onlyabout 1% of the superoxide anion formation from oxidation-reduction cycling of anticancer quinones results in theproduction of CH4 from Me2SO. The limited yield of CH4 isprobably due to various factors, which include the multiplereaction products ofMe2SO with the *OH and the possibilitythat CH4 produced intracellularly may only imperfectlytraverse the plasma membrane (38).Tumor cells treated with 5-iminodaunorubicin (250 AuM; n

= 4) produced no detectable methane from Me2SO (data notshown). These experiments, as well as the cytotoxicitystudies with 5-iminodaunorubicin, are in agreement withrecent investigations that have demonstrated that the imino-substitution on the daunorubicin quinone effectively elimi-nates superoxide anion or H202 production by this compoundafter reaction with NADPH-cytochrome P-450 reductase (7)or NADH dehydrogenase (8). It is noteworthy that the IC50for 5-iminodaunorubicin (4 ,M) was nearly 3-fold higher thanthat for DOX; the IC50 for daunorubicin, itself, againstEhrlich tumor cells in this laboratory is 0.5-1.0 ,M. Since5-iminodaunorubicin does not appear to produce oxygen freeradicals in Ehrlich cells, the 4- to 8-fold difference in IC50between daunorubicin and the analog might represent a freeradical component involved in tumor cell killing by the parentanthracycline.

In related experiments, -OH production by whole tumorcells treated with AZQ did not appear to involve the super-oxide anion because SOD produced no significant inhibitionof methane formation from Me2SO (Table 3). Recent studiesby Szmigiero and colleagues (39) support the concept that

Table 3. Effect of anticancer quinones on *OH production byEhrlich carcinoma cells

Methane production,Reactive pmol/120 minoxygen Conc. or MMC, AZQ, DOX,

scavenger activity 250 AM 50,.M 250,uM

None 97 ± 9 120 ± 7 98 ± 6Catalase 3000 U/ml 54 ± 3* 36 ± 18t 0 ± otHI catalaset 3000 U/ml 77 ± 13 130 ± 45 87 ± 2SOD 50,ug/ml 58 ± 2* 120 ± 2 29 ± 29*HI SOD 50,ug/ml 76 ± 2 90 ± 7 89 ± 2Thiourea 200 mM 21 ± 21* 39 ± 19t 29 ± 29*Urea 200 mM 76 ± 2 100 ± 9 88 ± 2Deferoxamine 50,M 21 ± 21* 62 ± 3t 21 ± 21t

Results are the mean ± SEM of three to five experiments. Nomethane was detected in the absence of the anticancer quinones (n= 6) or in the presence of MMC, AZQ, or DOX after denaturationof the tumor cells by heat (n = 3). P values are indicated for thecomparison of methane production in the presence and the absenceof the reactive oxygen scavenger. Conc., concentration; U, units.*P < 0.05.tp < 0.01.tCatalase and SOD were heat-inactivated (HI) by autoclaving for 60min.

AZQ undergoes an initial 2-electron rather than 1-electronreduction in vivo with the subsequent production of H202 inintact cells. This could explain why tumor cell killing and-OH production were not affected by SOD.

DISCUSSIONIt has been shown in these experiments that the cytotoxicityofDOX, MMC, and AZQ, three clinically important quinone-containing anticancer agents, can be reduced or abolished bychemicals and antioxidant proteins that detoxify or limit theformation of the -OH and H202. Previous experiments withDOX (9), MMC (10), or AZQ (39) have focused on thepossibility that oxidation-reduction cycling of these antican-cer quinones could produce DNA strand scission. However,the similarity between cytotoxicity studies and experimentsdemonstrating - OH production strongly suggests that drug-stimulated superoxide anion, H202, and *OH production playan important role in tumor cell killing as well as DNA damageby these structurally dissimilar anticancer quinones.The finding that the antioxidant proteins catalase and SOD

diminished anticancer quinone cytotoxicity is of particularinterest because enzymes of this size are unlikely to traversethe tumor cell plasma membrane (40). Prior studies haveshown that the uptake of exogenous SOD by Ehrlich ascitescells is minimal (41); furthermore, in these experiments noevidence was found for the uptake of catalase into Ehrlichtumor cells in suspension culture. Since H202 is freelypermeable across cell membranes, the protective effect ofexogenous catalase could result from a reduction in theintracellular peroxide concentration by the extracellularenzyme. This conclusion is supported by the results ofexperiments using the organoselenium compound PZ 51.Since intracellular glutathione is required for this agent tobreak down H202 (17, 18) and since PZ 51 was superior tocatalase in protecting cells against AZQ and DOX, it may besuggested that intracellular H202 production is stronglyassociated with the cytotoxic effect of anticancer quinoneswhen these drugs are tested at their IC50. However, as Powisand colleagues have shown (42), superoxide anion itself doesnot easily cross the tumor cell membrane, and, yet, SODpartially protected Ehrlich cells from the toxic effects ofDOX and MMC. Thus, anticancer quinone cytotoxicity in thepresent studies, as well as the recently described effects ofimpermeable, polymer-bound DOX (5, 6), could also involveH202 or superoxide anion produced at or near the tumor cellsurface as well as intracellularly.The mechanism of *OH production by anticancer quinones

has not been addressed directly in these studies. Priorinvestigations, however, suggest that cyclical reduction andoxidation of the quinone moiety of these drugs is involved inproviding sufficient H202 for *OH production to proceed(7-10). In fact, the precise chemical pathway of *OH forma-tion may differ between the drugs investigated in theseexperiments. However, *OH production in each case ap-peared to involve the interaction of an iron catalyst withH202. The source and biochemical form of the iron involvedin these experiments is still a matter of conjecture. Recentstudies have shown that Ehrlich cells contain =100 ,uM ironthat is essentially all in protein-bound form (32). However,since DTPA and deferoxamine reduced the cytotoxicity ofMMC and DOX, the 4 ,uM extracellular iron in the tissueculture media may itself have been the proximate source ofthe iron catalyst in these studies. Results with SOD,deferoxamine, DTPA, and bipyridine suggest that either theiron-catalyzed Haber-Weiss reaction (30) or a direct inter-action between the MMC or DOX semiquinone and H202 toform *OH in the presence of iron (9) could explain the findingswith these drugs; the mechanism of AZQ-induced . OH

Medical Sciences: Doroshow

Dow

nloa

ded

by g

uest

on

Nov

embe

r 27

, 202

0

4518 Medical Sciences: Doroshow

formation, although involving H202, may be more difficult todefine.Whatever the precise mechanism involved, the oxidizing

power of the *OH radical (43) could exert a powerful oxidantstress at multiple cellular sites in tumors treated withquinone-containing antineoplastic drugs. The potentially le-thal events that could be produced by H202 or * OH in tumorcells are numerous but include deleterious effects on cellularcalcium hemeostasis (44) and energy production (45) as wellas diminished reproductive potential (31).The real possibility that oxygen free radicals are involved

in the therapeutic action of the anticancer quinones MMC,AZQ, and DOX suggests, finally, that strategies aimed ataltering the defenses of the tumor cell against oxidation maybe a novel method of enhancing the effectiveness as well asthe spectrum of activity for this widely used class ofantineoplastic compounds.

This study was supported by Grant CA 31788 from the NationalCancer Institute and by the Leukemia Society of America.

1. Young, R. C., Ozols, R. F. & Myers, C. E. (1981) N. Engl. J.Med. 305, 139-153.

2. Crooke, S. T. & Bradner, W. T. (1976) Cancer Treat. Rev. 3,121-139.

3. Bender, J. F., Grillo-Lopez, A. J. & Posada, J. G., Jr. (1983)Invest. New Drugs 1, 71-84.

4. DiMarco, A. (1982) in Cancer Medicine, eds. Holland, J. F. &Frei, E. (Lea & Febiger, Philadelphia), pp. 872-906.

5. Tritton, T. R. & Yee, G. (1982) Science 217, 248-250.6. Tokds, Z. A., Rogers, K. E. & Rembaum, A. (1982) Proc.

Natl. Acad. Sci. USA 79, 2026-2030.7. Bachur, N. R., Gordon, S. L., Gee, M. V. & Kon, H. (1979)

Proc. NatI. Acad. Sci. USA 76, 954-957.8. Doroshow, J. H. (1983) Cancer Res. 43, 4543-4551.9. Bates, D. A. & Winterbourn, C. C. (1982) FEBS Lett. 145,

137-142.10. Gutteridge, J. M. C., Quinlan, G. J. & Wilkins, S. (1984)

FEBS Lett. 167, 37-41.11. Sinha, B. K., Muindi, J. R. F., Batist, G. & Myers, C. E.

(1985) Proc. Am. Assoc. Cancer Res. 26, 225 (abstr.).12. Batist, G., Cowan, K. H., Curt, G., Katki, A. & Myers, C. E.

(1985) Clin. Res. 33, 576 (abstr.).13. Doroshow, J. H. (1981) Proc. Am. Assoc. Cancer Res. 22, 203

(abstr.).14. Doroshow, J. H. (1982) Clin. Res. 30, 69 (abstr.).15. Doroshow, J. H. (1984) Clin. Res. 32, 58 (abstr.).16. McCord, J. M. & Fridovich, I. (1969) J. Biol. Chem. 244,

6049-6055.17. Muller, A., Cadenas, E., Graf, P. & Sies, H. (1984) Biochem.

Pharmacol. 33, 3235-3239.18. Wendel, A., Fausel, M., Safayhi, H., Tiegs, G. & Otter, R.

(1984) Biochem. Pharmacol. 33, 3241-3245.19. Goldstein, I., Brai, M., Osler, A. G. & Weissmann, G. (1973)

J. Immunol. 111, 33-44.20. Doroshow, J. H., Locker, G. Y. & Myers, C. E. (1980) J.

Clin. Invest. 65, 128-135.

21. Chu, M. Y. & Fisher, G. A. (1968) Biochem. Pharmacol. 17,753-767.

22. Roper, P. R. & Drewinko, B. (1976) Cancer Res. 36,2182-2188.

23. Repine, J. E., Eaton, J. W., Anders, M. W., Hoidal, J. R. &Fox, R. B. (1979) J. Clin. Invest. 64, 1642-1651.

24. Repine, J. E., Pfenninger, 0. W., Talmage, D. W., Berger,E. M. & Pettijohn, D. E. (1981) Proc. Natl. Acad. Sci. USA78, 1001-1003.

25. Armitage, P. (1971) in Statistical Methods in Medical Research(Blackwell, Oxford, United Kingdom), pp. 104-126.

26. Kharasch, N. & Thyagarajan, B. S. (1983) Ann. N.Y. Acad.Sci. 411, 391-402.

27. Pommier, Y., Zwelling, L., Mattern, M., Erickson, L., Ker-rigan, D., Schwartz,' R. & Kohn, K. (1983) Cancer Res. 43,5718-5724.

28. Van Hazel, G., Scott, M., Rubin, J., Moertel, C., Eagan, R.,O'Connell, M. & Kovach, J. S. (1983) Cancer Treat. Rep. 67,805-810.

29. Schilsky, R., Kelly, J. A., Ihde, D. C., Howser, D. M.,Cordes, R. & Young, R. C. (1982) Cancer Res. 42, 1582-1586.

30. Rosen, H. & Klebanoff, S. J. (1981) Arch. Biochem. Biophys.208, 512-519.

31. Mello-Filho, A. C., Hoffman, E. & Meneghini, R. (1984)Biochem. J. 218, 273-275.

32. Arslan, P., DiVirgilio, F., Beltrame, M., Tsien, R. & Pozzan,T. (1985) J. Biol. Chem. 260, 2719-2727.

33. Hershko, C. (1978) Blood 51, 415-423.34. Robbins, E., Fant, J. & Norton, W. (1972) Proc. Natl. Acad.

Sci. USA 69, 3708-3712.35. Byers, B., Arceneaux, J., Gaines, C. & Sciortino, C. V. (1975)

in Development of Iron Chelators for Clinical Use, eds.Anderson, W. F. and Hiller, M. C. (U.S. Dept. of Health,Education, and Welfare, Washington, DC), pp. 213-228.

36. Flitter, W., Rowley, D. A. & Halliwell, B. (1983) FEBS Lett.158, 310-312.

37. Cederbaum, A. I., Dicker, E., Rubin, E. & Cohen, G. (1979)Biochemistry 18, 1187-1191.

38. Klein, S., Cohen, G. & Cederbaum, A. (1981) Biochemistry 20,6006-6012.

39. Szmigiero, L., Erickson, L. C., Ewig, R. A. & Kohn, K. W.(1984) Cancer Res. 44, 4447-4452.

40. Petkau, A., Kelly, K., Chelack, W. & Barefoot, C. (1976)Biochem. Biophys. Res. Commun. 70, 452-458.

41. Michelson, A. M. & Puget, K. (1980) Acta Physiol. Scand.Suppl. 492, 67-80.

42. Powis, G., Svingen, B. A. & Appel, P. (1981) Mol. Pharmacol.20, 387-394.

43. Ward, P. A., Till, G. O., Kunkel, R. & Beauchamp, C. (1983)J. Clin. Invest. 72, 789-801.

44. Harris, E., Booth, R. & Cooper, M. (1982) FEBS Lett. 146,267-269.

45. Tyler, D. (1975) Biochim. Biophys. Acta 396, 335-346.46. Doroshow, J. H. & de Vries, L. (1984) Proc. Am. Assoc.

Cancer Res. 25, 300 (abstr.).47. Doroshow, J. H. (1985) Proc. Am. Assoc. Cancer Res. 26, 225

(abstr.).

Proc. Natl. Acad. Sci. USA 83 (1986)

Dow

nloa

ded

by g

uest

on

Nov

embe

r 27

, 202

0