Role of the benzoyloxyl radical in DNA damage ... - ACS Publications

Mar 1, 1991 - Role of the benzoyloxyl radical in DNA damage mediated by benzoyl peroxide. James E. Swauger, Patrick M. Dolan, Jay L. Zweier, Periannan...
0 downloads 0 Views 1MB Size
Chem. Res. Toxicol. 1991,4, 223-228

223

Role of the Benzoyloxyl Radical in DNA Damage Mediated by Benzoyl Peroxide James E. Swauger,t Patrick M. Dolan,t Jay L. Zweier,t Periannan Kuppusamy,l and Thomas W. Kensler**t Division of Toxicological Sciences, Department of Environmental Health Sciences, and The Electron Paramagnetic Resonance Laboratories, Department of Medicine, Division of Cardiology, Johns Hopkins Medical Institutions, Baltimore, Maryland 21205 Received August 10, 1990 Benzoyl peroxide (BzPO) is both a tumor promoter and progressor in mouse skin; however, BzPO is neither an initiator nor a complete carcinogen in this tissue. Although not mutagenic, BzPO has been observed to produce strand breaks in DNA of exposed cells. These actions are presumed to be mediated by free-radical derivatives of BzPO. Previous studies suggested that the metabolism of BzPO in keratinocytes proceeds via the initial cleavage of the peroxide bond, yielding benzoyloxy radicals which, in turn, can either fragment to form phenyl radicals and carbon dioxide or abstract H atoms from biomolecules to yield benzoic acid. Benzoic acid is the major stable metabolite of BzPO produced by keratinocytes. In the present study we have investigated the role of BzPO and its metabolites in the generation of strand scissions in a cell-free system using (PX-174 plasmid DNA. In this system BzPO produced DNA damage that was dose-dependent over a concentration range of 0.1-1 mM and required the presence of copper but not other transition metals. By contrast, benzoic acid did not produce DNA damage in this system, either in the presence or in the absence of copper. The inclusion of spin trapping agents, (PBN), 3,5-dibromo-4-nitrosobenzenesulfonate, and nisuch as N-tert-butyl-a-phenylnitrone trosobenzene, in incubations was found to significantly reduce the extent of DNA damage generated via the copper-mediated activation of BzPO. Electron paramagnetic resonance spectroscopy studies suggested that the primary radical trapped by PBN following coppermediated decomposition of BzPO was the benzoyloxy radical. By contrast, formation of either phenyl radicals or carbon dioxide was not detected in this system. Compounds that serve as facile H donors, such as glutathione and ergothioneine, were also effective inhibitors of BzPOmediated DNA strand breakage. BzPO does not appear to readily undergo addition reactions with DNA in that no covalent binding of BzPO to DNA was produced in incubations of radiolabeled BzPO, calf thymus DNA, and Cu+. Collectively, these observations suggest BzPO may be activated to DNA-damaging intermediates via copper-catalyzed cleavage of the peroxide bond, resulting in the formation of the benzdyloxyl radical which may then produce labile sites in DNA through H-abstraction reactions.

Introduction Benzoyl peroxide (BzPO)' is a widely used free radical generating compound with an estimated annual production of approximately 7 million pounds. BzPO is used principally as a source of free radicals in the plastics and rubber industry; however, the compound is also found in use as a food additive and in certain nonprescription drugs. Concern over the potential exposure of the public to this peroxide arose as a result of data published which demonstrated that this compound was both a tumor promoter and progressor in mouse skin (1,2). The biological activity of BzPO in multistage carcinogenesis regimens in the skin of SENCAR mice has previously been ascribed to its ability to generate free-radical derivatives although the mechanisms by which BzPO enhances tumorigenesis remain to be elucidated. Nacht et al. (3) have demonstrated that benzoic acid is the major stable metabolite formed in skin explants following exposure to BzPO. Additional investigations from our laboratory utilizing spin trapping and

EPR techniques to charcaterize free-radical metabolites of BzPO in murine keratinocytes demonstrated the generation of PBN-alkoxy1 (unpublished observation) and DMPO-aryl radical adducts (4). Further, incubation of epidermal homogenates with [ 14C]carbonyl-labeledBzPO resulted in the conversion of 8-10% of the peroxide to l4COZ. However, no 14C02was formed from either [14C]ring-labeled BzPO or [ ''C]carbonyl-labeled benzoic acid ( 4 ) . As summarized in Scheme I, these results collectively suggest that in keratinocytes the peroxide bond undergoes cleavage to yield benzoyloxyl radicals which, in turn, can undergo either fragmentation to form both phenyl radicals and COz or H abstraction to yield benzoic acid. The detection of alkoxy1 and aryl radical species in isolated keratinocytes provides support for the hypothesis that free-radical metabolites are potentially responsible for the toxicologic activities of BzPO in vivo. DNA damage may play a role in both tumor promotion and enhancement of malignant conversion associated with

To whom correspondenceshould beaddresGd at theDeparc ment of Environmental Health Sciences, Johns Hopkins School of Hygiene and Public Health, 615 N. Wolfe St., Baltimore, MD 21205. Department of Environmental Health Sciences. Division of Cardiology.

*

Abbreviations: BzPO, benzoyl peroxide; PBN, N-tert-butyl-nphenylnitrone;DBNBS, 3,5-dibromo-4-nitrosobenzenesulfonate; DMPO, 5,5'-dimethyl-l-pyrrolineN-oxide;EPR, electron paramagnetic resonance; DMF, N,N-dimethylformamide;Quin-2/AM,2-1[2-(hiu(carhoxymethyl)amino]-5-methylphenoxy]methyl~-6-methoxy-8-[ bis(carhoxymethy1)aminolquinoline tetrakis(acetoxymethy1) ester.

0 1991 American Chemical Society

224

Chem. Res. Toxicol., Vol. 4, No. 2, 1991

Scheme I. Proposed Pathway for the Activation of Benzoyl Peroxide to Free-Radical and DNA-Damaging Species

BENZOYL PEROXIDE

Y- cu+l

hc u + z

BENZOYLOXYLRADICAL

/

BIOMOLECULES\

J

(R3 I

PIlFNYL RADICAL

DNA DAMAGE LIPIDPE~OXIDATION TIllOL OXIDATION

B m l C ACID

application of BzPO to mouse skin. Although not m u t a genic (9,BzPO has been demonstrated to cause s t r a n d breaks in a number of cell types including primary cultures of leukocytes and epithelial cells as well as immortalized epidermal cell lines (6-8). Hartley et al. (6)have suggested that the tumor-promoting activity of BzPO occurs through a selective and direct induction of strand breaks in normal keratinocytes resulting i n cytotoxicity which subsequently confers a g r o w t h a d v a n t a g e to the resistant, initiated keratinocyte population. Tumor progression, the conversion of b e n i g n papillomas i n t o m a l i g n a n t , r a p i d l y growing neoplasms, is believed to involve a second, discrete heritable event which m a y or m a y not be identical with events occurring d u r i n g initiation. The enhancement of m a l i g n a n t progression i n vivo has previously been demonstrated with BzPO (2)as well as with mutagens s u c h as urethane, N-methyl-N'-nitro-N-nitrosoguanidine, and 4nitroquinoline N-oxide (9). While the critical molecular target of BzPO has not been identified, it seems likely that the g e n o m e represents a potentially i m p o r t a n t t a r g e t i n mouse skin tumorigenesis. Therefore, we have utilized the induction of strand scissions in plasmid DNA as a model s y s t e m i n which to e x a m i n e the DNA-damaging activity of BzPO and its metabolites as well as to probe potential m e c h a n i s m s associated with the activation of BzPO.

Materials and Methods Materials. BzPO, benzoic acid, o-phenanthroline, sodium iodide, ergothioneine, glutathione, histidine, DBNBS, NB, CuCl, and ethidium bromide were purchased from Sigma Chemical Co. (St. Louis, MO). Quin-2/AM was purchased from Calbiochem (Sen Diego, California). [14C]Carbonyl-labeledBzPO (26 pCi/ pmol) and [l'C]carbonyl-labeled benzoic acid (56.7 pCi/pmol) were purchased from Research Products International Corp. (Mount Prospect, IL). [ ''C]Ring-labeled BzPO (29 pCi/pmol) was purchased from Amersham Corp. (Arlington Heights, IL). Perbenzoic acid was prepared as described by Braun (10). BzPO was initially cleavaged by sodium methoxide at low temperature. Perbenzoic acid was then liberated from sodium perbenzoate via acidification with 1 N sulfuric acid. Assay for DNA S t r a n d Breaks. DNA damage was measured by the conversion of closed circular supercoiled OX-174 DNA to open circular or linear forms. Briefly, 0.2 pg of OX-174 DNA (New England Biolabs, Beverly, MA) was incubated in 10 pM Tris, pH 7.4, for 20 min with and without the indicated concentrations of BzPO, transition metals, and inhibitors. DMF was used as the

Swauger et al. vehicle for BzPO and was added to a final concentration of 25%. All incubations were carried out in 1.5-mL microtubes a t 37 "C. Following incubation, the samples were immediately loaded on a 1?% agarose gel containing 40 mM Tris, 20 mM sodium acetate, and 2 mM EDTA and electrophoresed in a horizontal slab gel apparatus. Gels were stained with a solution of 1pg/mL ethidium bromide for approximately 5 min, washed in distilled water for 10 min, and irradiated with UV light a t 254 nm. The gels were then photographed and the negatives scanned with a Beckman DU-7 spectrometer a t 560 nm. The percentage of DNA in each form was calculated by integrating the area under the peaks. Covalent Binding t o Calf T h y m u s D N A a n d Bovine Ser u m Albumin. [14C]Ring-labeled BzPO, [14C]carbonyl-labeled BzPO, and [14C]carbonyl-labeledbenzoic acid (1mM) were incubated in the presence of freshly prepared CuCl(O.1 mM) with either 1mg/mL calf thymus DNA (Sigma, type I) or bovine serum albumin (Sigma, fraction V) in a final volume of 2 mL for 1 h. DNA incubations were stopped via the addition of 7 mL of phenol containing 0.1 TO8-hydroxyquinoline. These samples were extracted a minimum of five times with 10 mL of phenol to ensure the complete removal of all noncovalently bound radioactivity. Samples were finally extracted with 10 mL of H,O-saturated ether. DNA was pelleted with ethanol overnight a t -20 "C and solubilized in 0.5 mL of 1 N NaOH, and aliquots were removed for the determination of bound radioactivity and DNA content (11). Protein incubations were stopped by the addition of 10 mL of absolute methanol. These samples were extracted five times with 5 mL of methanol/ether (3:l) to remove all noncovalently bound radioactivity. Protein pellets were solubilized in 0.5 mL of 1 N NaOH, and aliquots were taken for the determination of protein content (12) and radioactivity. H P L C Analysis of B z P O Metabolites. [ 14C]Ring-labeled BzPO (1mM; 0.45 pCi) was incubated in the presence or absence of freshly prepared CuCl(O.l mM) for 20 min under conditions identical with those used in the plasmid DNA nicking assay. Separation and identification of metabolites of BzPO were carried out by reverse-phase high-performance liquid chromatography using a Waters 820 workstation and a Du Pont Zorbax ODS semipreparative column (9.4 mm X 25 cm) preceded by a Waters Guardpack C-18 precolumn. Metabolites were eluted a t a flow rate of 1.5 mL/min using an 85:15 methanol/water mobile phase. One-minute fractions were collected and counted in a Packard 300C scintillation spectrometer. EPR Studies. EPR spectra were recorded a t room temperature by using an IBM-Bruker ER 300 spectrometer operating a t X band with a TM,,, cavity. The spectrometer settings are indicated in the figure legend. Microwave frequency and magnetic field were measured, respectively, with an EIP 575 source locking microwave counter and a Bruker ER 035M NMR gaussmeter. The spin traps, DMPO and PBN (Aldrich Chemical Co., Milwaukee, WI), were used a t final concentrations of 50 and 30 mM, respectively. DMPO was vacuum distilled in the dark two times and stored desiccated a t -80 "C, while PBN was used without further purification. Care was taken to ensure that the DMPOcontaining solutions were subject to minimal light-induced degradation.

Results BzPO is activated to a DNA-damaging species i n the presence of Cu+ as evidenced by the near-complete conversion of the closed circular, supercoiled +X-174 plasmid DNA to the open circular form over the 20-min incubation period (Figure 1). Longer incubation periods lead to the complete degradation of the plasmid DNA ( n o t shown). B y contrast, BzPO alone i n D M F / T r i s buffer w i t h o u t added copper had no effect on the supercoiled DNA. T h i s activation of BzPO is clearly m e t a l - d e p e n d e n t as ophenanthroline, a copper chelator, blocks the generation of single-strand D N A breaks. However, other metals such as m o l y b d e n u m , zinc, and calcium, as well as b o t h ferric and ferrous iron, failed to activate BzPO to DNA-damaging species i n t h i s system. In addition, an iron-nitriloacetic acid chelate, shown by EPR to actively metabolize hydrogen peroxide to hydroxyl radical prior to use i n t h i s

DNA Damage by Benzoyl Peroxide 1

2

3

4

Chem. Res. Toxicol., Vol. 4, No. 2, 1991 225 5

1

OC

Figure 1. Effect of copper-mediated activation of benzoyl peroxide on a x - 1 7 4 plasmid DNA. DNA damage was determined by measuring the conversion of closed circular (CC) supercoiled DNA to an open circular (OC) form. Lane 1: CbX-174 DNA in DMF/Tris. Lane 2: 9X-174 DNA BzPO in DMF/Tris. Lane 3: CbX-174 DNA + Cu+ in DMF/Tris. Lane 4: CbX-174 DNA + BzPO + Cu+ in DMF/Tris. Lane 5: CbX-174 DNA + BzPO Cu+ + o-phenanthroline in DMF/Tris. The concentrations used were as follows: BzPO, 1 mM; CuCl, 0.1 mM, o-phenanthroline, 33 mM; and Tris, 10 pM.

+

+

-

(9

"DNA damage was measured as described under Materials and Methods. Values are the mean f SE of triplicate determinations.

z

f

2

U

lo0T

Table I. Effect of Free-Radical Scavengers on CoppedBenzoyl Peroxide Mediated DNA Damagea 9% closed % circular DNA protection incubation mixture 81.1 f 2.3 Tris buffer 16.5 f 1.3 cu+/ 3zPO 84 cu+/ 3zPO + o-phenanthroline (37 mM) 70.9 f 1.9 56 cu+/ 3zPO + o-phenanthroline (1 mM) 52.8 f 2.2 51 49.6 f 2.0 cu+/ 3zPO + desferal (25 mM) 100 84.9 f 1.5 cu+/ 3zPO + Quin-B/AM (1 mM) 100 85.1 f 3.2 cu+/ 3zPO + glutathione (10 mM) 77 66.0 f 8.6 cu+/ 3zPO + glutathione (1 mM) 39.4 f 5.9 36 cu+/ 3zPO + glutathione (0.1 mM) 78 66.9 f 3.4 cu+/ 3zPO + ergothionine (1 mM) 100 81.6 f 5.4 cu+/ 3zPO + NaI (1 mM) 91 75.5 f 4.2 cu+/ 3zPO + histidine (1 mM) 51.0 f 11.2 53 cu+/ 3zPO + PBN (30 mM) 48 47.2 f 1.6 cu+/ 3zPO + DBNBS (30 mM) 0 12.9 f 2.7 cu+/ 3zPO + DMPO (50 mM)

BENZOIC ACID

1mM BzPO

4

0

% m

60

.0

9

20 m

Figure 2. Dose-response curves for the copper-mediated activation of benzoyl peroxide, perbenzoic acid, and benzoic acid. DNA damage was determined by measuring the conversion of closed circular supercoiled DNA to an open circular form following a 20-min incubation. The concentration of CuCl utilized was 0.1 mM.

system, failed to activate BzPO to DNA-damaging species (not shown). The doseresponse relationship for the copper-mediated activation of BzPO to DNA-damaging intermediates was investigated and is depicted in Figure 2. DNA strand breaks resulting from the copper-mediated activation of BzPO occurred over concentrations ranging between 0.1 and 1 mM. Similar concentratinos of BzPO have been demonstrated to cause strand breaks in several different types of cells in culture (6-8). The activities of benzoic acid, the primary stable metabolite of BzPO produced in keratinocytes, as well as perbenzoic acid, a related peracid, were also examined in this system. Neither benzoic acid nor perbenzoic acid was found to produce DNA damage in the absence of copper. However, unlike benzoic acid, perbenzoic acid readily generated detectable levels of DNA damage in the presence of copper. As derived from Figure 2, the ECM for producing a single strand break is approximately 30 pM for perbenzoic acid and is 25-fold higher or approximately 700 pM for BzPO. Correspondingly, perbenzoic acid, which is also an epidermal tumor promoter (13,14), is a much more potent oxidizer than BzPO in skin. As presented in Table I, DNA strand breakage mediated by BzPO and copper can be inhibited by a variety of agents. Reduced glutathione and ergothioneine, both naturally occurring sulfhydryl compounds, as well as other radical-scavengingagents such as histidine and iodide were found to block the DNA-damaging effects of BzPO at 1 mM concentrations. Notably, glutathione protected plasmid DNA even at subphysiologic, micromolar con-

loo

1mM BzPO

80

--

60

--

40

--

20

--

+ 0.1 mM Cu+

c

Figure 3. HPLC analysis of stable metabolites of benzoyl peroxide produced in the presence of copper. Upper panel: BzPO in DMF/Tris. Lower panel: BzPO + Cu+ in DMF/Tris. The concentrations utilized were as follows: BzPO, 1 mM; CuCI, 0.1 mM.

centrations. The inclusion of various spin trapping agents was also observed to significantly protect the DNA from BzPO/Cu+-mediated damage. PBN, at a concentration of 30 mM, reduced the percentage of closed circular DNA converted to the open circular form by approximately 50%. Other spin trapping agents such as DBNBS were also found to provide significant protection when utilized at millimolar concentrations. The metal chelators 0phenanthroline and Quin-2/AM afforded 56 and 100% protection, respectively, at 1mM concentrations. Another metal chelator, desferal, was found to provide approximately 50% protection when utilized at a concentration of 25 mM, although because of the high concentration required this effect may be due to a radical-scavenging mechanism rather than copper chelation. The Cu+-dependentmetabolism of BzPO to both stable and radical metabolites was characterized under the incubation conditions utilized for the DNA nicking assay. HPLC of metabolites of BzPO produced in the presence

Swauger et al.

226 Chem. Res. Toxicol., Vol. 4, No. 2, 1991

Table 11. Covalent Binding of Benzoyl Peroxide or Benzoic Acid to DNA and Protein" specific activity nmol of nmol of bound/mg of bound/mg of compound protein DNA [14C]ring-labeledbenzoyl peroxide 13.4 f 5.5 not detectable [14C]carbonyl-labeledbenzoyl 20.3 f 2.9 not detectable peroxide [14C]carbonyl-labeledbenzoic acid not detectable not detectable Radiolabeled compounds (1 mM) were incubated in the presence of CuCl (0.1 mM) with either 1 mg/mL calf thymus DNA or 1 mg/mL serum albumin for 1 h, and covalent binding was determined. Not detectable is 50.5 nmol/mg. Values are the mean f SE of quadruplicate incubations.

G.

D.

L 3430

3455

I

I

1

3400

3505

3530

MAGNETIC F I E L D (GAUSS)

Figure 4. Electron paramagnetic resonance analysis of freeradical metabolites of benzoyl peroxide produced in the presence of copper. (A) 30 mM PBN, 1 mM BzPO, and 0.1 mM Cu+ in DMF/Tris. (B)Computer simulation of spectrum shown in (A). (C) 30 mM PBN and 1mM BzPO in DMFITris. (D)30 mM PBN and 0.1 mM Cu+ in DMF/Tris. (E) 30 mM PBN, 0.1 mM Cu2+, and 1 mM phenylhydrazine in DMF/Tris. The instrument settings were as follows: modulation frequency, 100 kHz; modulation amplitude, 0.5 G; scan time, 1.0 min; microwave power, 20 mW; microwave frequency, 9.772 GHz.

of CuCl demonstrated that benzoic acid was the only stable metabolite produced under these conditions (Figure 3). EPR studies indicated that an alkoxy1 radical adduct was formed from BzPO in the presence of copper and the spin trap PBN (Figure 4A). Omission of either copper (Figure 4C) or BzPO (Figure 4D) from the incubations abolished the EPR spectra. The spectrum in Figure 4A was stimulated and is shown in Figure 4B. The derived splitting constants for this PBN adduct are aN = 13.47, aH = 1.84. Inasmuch as there are no literature values for the PBNbenozyloxyl adduct in DMF/Tris, comparable incubations were conducted in benzene (not shown). In this case, the observed splitting constants were aN = 13.31, aH = 1.34 for the PBN adduct: these values closely match those previously reported for the PBN-benzoyloxyl radical adduct in benzene (15-17). Further, comparable reactions with BzPO and Cu+ constituted in DMF/phosphate buffer produced identical spectra, suggesting that secondary formation and trapping of Tris radicals were not interfering with the PBN adduct spectral assignment. Shown in Figure 4E, the phenyl radical-PBN adduct ( a N= 15.34, aH = 3.4) (produced from Cu2+and phenylhydrazine) can be readily distinguished from the benzoyloxyl radical-PBN adduct. No evidence for aryl radical adducts (i.e., phenyl) derived from BzPO was obtained with either the spin trap PBN or DMPO in this system. Taken together, these data suggest that the principle PBN adduct formed under the incubation conditions of the DNA nicking assay is the benzoyloxyl adduct. Additional confirmation that minimal decomposition of benzoyloxyl radicals to phenyl radicals and CO2 occurs in the plasmid DNA nicking assay was obtained by constituting incubations with [carbonyl14C]BzP0and assaying for the evolution of 14C02.Unlike the situation with cytosol prepared from keratinocytes (41, no radiolabeled C02was released during the Cu+-mediated

activation of BzPO in the DMF/Tris buffer system. Many genotoxic agents are known to form DNA adducts. Thus, the ability of BzPO to covalently modify target biomolecules was also investigated. Either [ 14C]ring-or [ 14C]carbonyl-labeled BzPO was incubated in the presence of copper with either calf thymus DNA or albumin for 1 h and covalent binding to these macromolecules determined. As shown in Table 11, substantial covalent modification of protein, but not DNA, was detected. By contrast, [ '*C]ring-labeled benzoic acid did not covalently modify either macromolecule. Further, essentially equivalent binding to protein was observed with either [14C]ringor [ 14C]carbonyl-labeledBzPO, again indicating the involvement of the benzoyloxyl radical in the reactivity of BzPO. These findings suggest that DNA adduct formation with BzPO is unlikely to occur in cells.

Discussion The involvement of transition metals in the activation of organic peroxides and hydroperoxides to reactive intermediates is well-known. For example, copper ions can activate hydrogen peroxide to form hydroxyl radicals (18, 19). The activation of BzPO by metal ions and specifically by copper has been previously described by Terakado et al. (20), who demonstrated that BzPO-mediated lipid peroxidation in microsomes prepared from rabbit dental pulp was enhanced by the presence of cations such as Cu2+ and Fe2+. However, in contrast to their observations, we find that iron, irrespective of its redox or chelation status, does not drive the activation of BzPO to DNA-damaging species. The reasons for this discrepancy and, more importantly, the physiological role of any transition metal in the cellular activation of BzPO, while plausible, are unknown. Spin trapping agents, such as PBN and DBNBS, have been utilized historically in EPR spectroscopy to overcome the technical difficulties associated with monitoring the generation of transient radical species. In our studies, spin traps were incorporated into incubations in an attempt to inhibit the interaction of the presumed radical metabolites of BzPO with plasmid DNA. As can be seen in Table I, PBN as well as other spin trapping agents such as DBNBS were found to ameliorate the damage generated via copper-mediated activation of BzPO. These data provide strong suggestive evidence that in fact radical species are responsible for the observed DNA damage. Other agents known to have a high affinity for oxygen-centered radicals, such as iodide and histidine, were also found to provide considerable protection in this system. Taken together, these observations indicate that BzPO is activated to a DNA-damaging species via the metal-catalyzed cleavage of the peroxide bond, resulting in the formation of an oxygen-centered radical, presumably the benzoyloxyl radical.

D N A Damage by Benzoyl Peroxide

Further investigations, utilizing spin trapping and EPR techniques to characterize free-radical metabolites of BzPO under conditions similar to those utilized in the DNA damage assays, have suggested that a single species, the benzoyloxyl radical, is trapped by PBN. This data is consistent with the studies demonstrating that copper does not drive the liberation of '%02from [lT]carbonyl-labeled BzPO. Collectively, these findings suggest that only the benzoyloxyl radical is formed from BzPO in the presence of copper and that degradation of this species to the phenyl radical does not readily occur in this simple chemical system. As shown in Table I, glutathione was the most effective inhibitor of DNA damage identified. Glutathione provided almost complete protection at concentrations clearly within normal physiological levels, presumably by participating as the H-atom donor to benzoyloxyl radicals. Interaction between glutathione and BzPO in vivo has been previously demonstrated by Perchellet et al. (21),who reported significant increases in oxidized glutathione in epidermal cells treated with BzPO. Hydrogen abstraction from glutathione can yield oxidized glutathione (22). Topical application of reduced glutathione has been demonstrated to inhibit both spontaneous and chemically induced tumor progression (23). However, it remains to be determined whether glutathione or other endogenous thiols such as ergothioneine provide protection against BzPO-mediated DNA damage in vivo as well. The inherent reactivities of the radical derivatives of BzPO, benzoyloxyl and phenyl radicals, would apparently preclude the possibility of their formation in the cytosol and subsequent migration into the nucleus to elicit DNA damage. Clearly, the avid capacity of this molecule to bind to protein (Table 11) and the obvious preponderance of such targets in the cytoplasm would necessitate formation of such a species in close proximity to the DNA in order to result in the direct formation of strand breaks. Several authors have previously suggested the existence of a sitespecific Fenton mechanism with regard to the coppermediated activation of hydrogen peroxide, according to which the binding of transition metals to the biological target is a prerequisite for the production of damage (19, 24,25). As copper is normally found bound to DNA at a concentration of approximately 0.3 pg/mg of DNA (26), the existence of such a direct activation pathway for BzPO in keratinocytes is likely. In studies to be reported elsewhere, we have observed that addition of metal chelators to primary cultures of keratinocytes blocks DNA strand scission mediated by BzPO. However, the potential formation of strand breaks in keratinocytes by BzPO through indirect mechanisms cannot, as yet, be excluded. A rapid, although transient, increase in cytosolic calcium levels in normal human bronchial epithelial cells has been observed following exposure to concentrations of BzPO comparable to those utilized in our studies (27). This change has been suggested to reflect a redistribution of intracellular calcium (28). Such increases in cytosolic calcium levels have been associated with activation of calcium-dependent endonucleases and as a result may provide a second potential mechanism through which BzPO may induce DNA damage in cells (29, 30). However a t this time the relative contributions of either the direct or indirect mechanisms to DNA damage induced by BzPO in keratinocytes is unknown. Studies to address these issues are currently ongoing in this laboratory. Acknowledgment. This work has been supported by grants from the National Institutes of Health (CA 44530 and ES 07141)and the Procter and Gamble Co. The EPR

Chem. Res. Toxicol., Vol. 4, No. 2, 1991 227

laboratories are supported by NIH Grants HL 17655 and HL 38324. T.W.K. is the recipient of Research Career Development Award CA 01230.

References (1) Slaga, T. J., Klein-Szanto, A. P., Triplett, L., and Yotti, L. P. (1981)Skin tumor-promoting activity of benzoyl peroxide, a widely used free radical generating compound. Science 213, 1023-1025. (2) OConnell, J. F., Klein-Szanto, A. P., DiGiovanni, D. M., Fries, J. W., and Slaga, T. J. (1986)Enhanced malignant progression of mouse skin tumors by the free-radical generator benzoyl peroxide. Cancer Res. 46, 2863-2865. (3) Nacht, S.,Yeung, D., Beasley, J. N., Anjo, M. D., and Maibach, H. J. (1981)Benzoyl peroxide: Percutaneous penetration and metabolic disposition. J. Am. Acad. Dermatol. 4,31-37. (4) Kensler, T. W., Egner, P. A., Swauger, J. E., Taffe, B. G., and Zweier, J. L. (1988)Formation of free radicals from benzoyl peroxide in murine keratinocytes. Proc. Am. Assoc. Cancer Res. 29, 150. (5) Ishidate, M., Sofuni, T., Yoshikawa, K., Hayashi, M., Nohmi, T., Sawada, M., and Matsuoka, A. (1984) Primary mutagenicity screening of food additives currently used in Japan. Food Chem. Toxicol. 22,623-636. (6) Hartley, J. A., Gibson, N. W., Zwelling, L. A., and Yuspa, S. H. (1985)Association of DNA strand breaks with accelerated terminal differentiation in mouse epidermal cells exposed to tumor promoters. Cancer Res. 45,4864-4870. (7) Saladino, A. J., Willey, J. C., Lechner, J. F., Grafstrom, R. C., Laveck, M., and Harris, C. C. (1985)Effects of formaldehyde, acetaldehyde, benzoyl peroxide and hydrogen peroxide on cultured normal bronchial epithelial cells. Cancer Res. 45, 2522-2526. (8) Birnboim, H. C. (1986)DNA strand breaks in human leukocytes induced by superoxide anion, hydrogen peroxide and tumor promoters are repaired slowly compared to breaks induced by ionizing radiation. Carcinogenesis 7, 1511-1517. (9) Hennings, H., Shores, R., Wenk, M., Spangler, E. F., Tarone, R., and Yuspa, S. H. (1983)Malignant conversion of mouse skin tumors is increased by tumor initiators and unaffected by tumor promoters. Nature 304,67-69. (10) Braun, G. (1941)Organic Syntheses (Blatt, A. H., Ed.) Collect. Vol. 1, p 431,Wiley, New York. (11) Giles, K. W., and Meyers, A. (1965)An improved diphenylamine method for the estimation of deoxyribonucleic acid. Nature 206,93. (12)Bradford, M.M. (1976)A rapid and sensitive method for the quantitation of microgram quantities of protein utilizing the principle of protein-dye binding. Anal. Biochem. 72,248-254. (13)Van Duuren, B. L., Nelson, N., Oris, L., Palmes, E. D., and Schmitt, F. L. (1963)Carcinogenicity of epoxides, lactones, and peroxy compounds. J. Natl. Cancer Znst. 31,41-55. (14) Bock, F. G., Meyers, H. K., and Fox, H. W. (1975)Cocarcinogenic activity of peroxy compounds. J. Natl. Cancer Znst. 55, 1359-1361. (15)Janzen, E. G., Evans, C. A., and Nishi, Y. (1972)Kinetic study of radical production in the thermal decomposition of benzoyl peroxide by electron spin resonance spin trapping. Rate (and substituent effect on the rate) of spin trapping by competitive scavenging of benzoyloxyl radicals with galvinoxyl. J. Am. Chem. SOC.94,8236-8238. (16) Janzen, E. G., and Blackburn, B. J. (1968)Detection and identification of short-lived free radicals by an electron spin resonance trapping technique. J. Am. Chem. SOC.90,5909-5910. (17) Bevington, J. C., Fridd, P. F., and Tabner, B. J. (1982)An electron spin resonance study of some trapped primary radicals. J. Chem. Soc., Perkin Trans. 2, 1389-1391. (18) Rowley, D. A., and Halliwell, B. (1983)Superoxide-dependent and ascorbate-dependent formation of hydroxyl radicals in the presence of copper salts: A physiologically significant reaction? Arch. Riochem. Biophys. 225,279-284. (19) Samuni, A,, Aronovitch, J., Godinger, D., Chevion, M., and Czapski, G. (1983)On the cytotoxicity of vitamin C and metal ions. Eur. J. Biochem. 137, 119-124. (20) Terakado, M., Yamazaki, M., Tsujimoto, T., Kawashima, T., Nagashima, K., Ogawa, J., Fujita, Y., Sugiya, H., Sakai, T., and Furuyama, S. (1984)Lipid peroxidation as a possible cause of benzoyl peroxide toxicity in rabbit dental pulp. J. Dental Res. 63,901-905.

228 Chem. Res. Toxicol., Vol. 4, No. 2, 1991 (21) Perchellet, J. P., Perchellet, E. M., Orten, D. K., and Schneider, B. A. (1985) Decreased ratio of reduced/oxidized glutathione in

mouse epidermal cells treated with tumor promoters. Carcinogenesis 7, 503-506. (22) Medeiros, M. G., Wefers, H., and Sies, H. (1987) Generation of excited species catalyzed by horseradish peroxidase or hemin in the presence of reduced glutathione and HzOz. Free Radical Biol. Med. 3, 107-110. (23) Rotstein, J. B., and Slaga, T. J. (1987) Effect of exogenous glutathione on tumor progression in the murine skin multistage carcinogenesis model. Carcinogenesis 9, 1547-1551. (24) Marx, G., and Chevion, M. (1985) Site-specific modification of albumin by free radicals. Biochem. J. 236, 397-400. (25) Hintz, P., and Kalyanaraman, B. (1986) Metal ion-induced activation of molecular oxygen in pigmented polymers. Biochim. Biophys. Acta 883, 41-45. (26) Hardy, K. J., and Bryan, S. E. (1975) Localization and uptake

Swauger et al. of copper into chromatin. Toxicol. Appl. Pharmacol. 33,62-69. (27) Smith, M. W., Miyashita, M., Jones, R. T., Willey, J. C., Harris, C. C., and Trump, B. F. (1988) The effect of Hz02 and formaldehyde on [Ca+z]iof normal and transformed human bronchial epithelial cells. FASEB J . 2, A407. (28) Trump, B. F., Berezesky, I. K., Smith, M. W., Phelps, P. C., and Elliget, K. A. (1989) The relationship between cellular ion deregulation and acute and chronic toxicity. Toricol. Appl. Pharmacol. 97, 6-22. (29) McConkey, D. J., Hartzell, P., Nicotera, P., Wyllie, A. H., and Orrenius, S. (1988) Stimulation of endogenous endonuclease activity in hepatocytes exposed to oxidative stress. Toxicol. Lett. 42, 123-130. (30) McConkey, D. J., Hartzell, P., Duddy, S. K., Hakansson, H., kills and Orrenius, S. (1988) 2,3,7,8-Tetrachlorodibenzo-p-dioxin

immature thymocytes by Ca+z-mediated endonuclease activation. Science 242, 256-259.