with Dehydroascorbate Cause Single-Strand Breaks in Plasmid DNA

Chem. Res. Toxicol. 1997, 10, 271-278

271

Articles Intermediates Produced in the Reaction of Chromium(VI) with Dehydroascorbate Cause Single-Strand Breaks in Plasmid DNA Diane M. Stearns and Karen E. Wetterhahn* Department of Chemistry, 6128 Burke Laboratory, Dartmouth College, Hanover, New Hampshire 03755-3564 Received August 29, 1996X

Ascorbate (vitamin C) is a biological reductant of the human carcinogen chromium(VI). The product of this reaction is presumed to be dehydroascorbate. However, we have found that chromium(VI) can also react with dehydroascorbate. This reaction was monitored by UV/ visible and electron paramagnetic resonance (EPR) spectroscopies. In sodium acetate buffer at pH 3.8, the reaction of chromium(VI) and excess dehydroascorbate produced chromium(V) and chromium(IV) intermediates. At high reaction concentration, the chromium(V) intermediate formed an EPR silent dimer, which dissociated upon dilution to lower concentration. UV/ visible experiments at pH 3.8 demonstrated that manganese(II) catalyzed the disproportionation of chromium(IV) to chromium(V) and chromium(III). The ability of the reaction intermediates to induce strand breaks in pBR322 DNA was determined at pH 3.8 and pH 5.8. At pH 3.8, chromium(IV) appeared to be the major species responsible for induction of strand breaks because the time course for formation of strand breaks did not parallel that of chromium(V), and strand breaks were decreased in the presence of the chromium(IV) scavenger manganese(II). At pH 5.8, fewer strand breaks were observed; however, the time course for their formation followed that of chromium(V). There has been much effort devoted to identification of the intermediate responsible for the induction of strand breaks during reactions of chromium(VI) with biological reductants. The current results suggest that it is not a single type of species that universally produces the DNA strand breaks observed in different chromium(VI) systems and that the reactivity of intermediates will depend on the chosen experimental conditions. Understanding this variability in chromium(VI) reactions may help to resolve the conflicting results from in vitro studies that are aimed at deciphering mechanisms of chromium(VI)induced cancers.

Introduction Chromium(VI) is a human carcinogen that enters cells through the phosphate-sulfate transport channels and is reductively activated by small molecules such as ascorbate and glutathione and many enzyme systems (1). The intracellular metabolism of Cr(VI) produces reactive intermediates such as Cr(V), Cr(IV), free radicals, and the final product Cr(III). Several types of DNA lesions have been observed when Cr(VI) is reduced in the presence of DNA in vitro and in vivo, including Cr-DNA adducts, DNA-protein cross-links, single-strand breaks, DNA interstrand cross-links, and oxidized nucleotide bases (2). Chromium(VI) can induce forward and reversion point mutations, and chromosomal aberrations (reviewed in ref 3). However, it remains to be determined which of the genotoxic species are ultimately responsible for mutations and which DNA lesion(s) are primarily responsible for Cr-induced tumorigenesis. One active area of chromium research has been aimed at distinguishing among the intermediates Cr(V), Cr(IV), and free radicals in terms of their ability to produce a * Author to whom correspondence should be addressed. Telephone (603) 646-3413, Fax (603) 646-3946. X Abstract published in Advance ACS Abstracts, February 1, 1997.

S0893-228x(96)00152-X CCC: $14.00

DNA single-strand break. DNA strand breaks have been observed after Cr(VI) exposure in cultured cells (reviewed in ref 2) and in vivo, in rat liver (4) and in the liver and red blood cells of chick embryos (5, 6). A DNA strand break, if mis-repaired, may be involved in chromosomal (multiple-gene) mutations that can be observed as structural chromosomal aberrations (7). Chromium(VI) induces chromosomal aberrations in cultured cells and in vivo, in rodent lymphocytes and bone marrow cells (reviewed in ref 3). Chromosomal aberrations have been linked to some forms of cancer (8). The formation of a Cr-induced DNA strand break signals the presence of an oxidative pathway that may manifest itself genetically through chromosomal alterations, but also epigenetically by effects on cellular signal transduction and gene expression. There has been much interest generated in the reaction of Cr(VI) with ascorbate (vitamin C) since the observation that ascorbate is one of the major reductants of Cr(VI) in rodents and humans (9-11). Through spectroscopic experiments, we have shown that the reaction of Cr(VI) with ascorbate in vitro produced Cr(V) and carbon-based radicals when Cr(VI) was in excess of ascorbate, whereas Cr(IV), Cr(III), and ascorbate radical were produced in amounts increasing with ascorbate concentration (12, 13). © 1997 American Chemical Society

272 Chem. Res. Toxicol., Vol. 10, No. 3, 1997

Stearns and Wetterhahn

The current study was motivated by the observation that Cr(V) was detected only in the presence of excess Cr(VI) in the Cr(VI)/ascorbate reactions, which led us to predict that Cr(VI) may further oxidize what was presumed to be the final oxidation product, dehydroascorbate (DHA)1 (12). We have found that Cr(VI) does indeed react with DHA and its decomposition products (14). Thus Cr(VI) is capable of going beyond the presumed reaction (eq 1):

2Cr(VI) + 3AscH- f 2Cr(III) + 3DHA

(1)

To our knowledge, this is the first example of the oxidation of DHA by a transition metal. In the current study, EPR and UV/visible spectroscopic experiments have been used to monitor the reaction of Cr(VI) with DHA and to detect the presence of Cr(IV) and Cr(V) intermediates. The ability of these intermediates to cause single-strand breaks in pBR322 plasmid DNA was evaluated at pH 3.8 and pH 5.8. We found that DNA strand breaks were induced at both pH 3.8 and 5.8, but by comparing the data to results of spectroscopic experiments, it appears that the lesions were caused by different reactive intermediates.

Experimental Procedures General Methods. All buffers were treated with Chelex 100 resin, sodium form, and 100-200 mesh (Bio-Rad Laboratories, Richmond, CA) to remove contaminating iron. All buffers and solutions for plasmid DNA experiments contained 20 µM diethylenetriaminepentaacetic acid (DETAPAC) to slow reaction of any residual trace iron with dioxygen. The Cr(VI) source was K2Cr2O7, the Mn(II) source was MnCl2‚4H2O, and the Mg(II) source was MgCl2‚6H2O. DHA was used as received (Aldrich, Milwaukee, WI) after purity was verified by 13C NMR spectroscopy for solutions in D2O (15). For each experiment, DHA was dissolved in buffer, and reactions were initiated within 3 min. Caution: Chromium(VI) is a known human carcinogen, and both inhalation and skin contact of Cr(VI) and its reaction solutions should be avoided. Instrumentation. UV/visible spectra were obtained on a Perkin Elmer Lambda 2 spectrophotometer with temperature regulated at 25.0 ( 0.1 °C or 37.0 ( 0.1 °C by a circulating water bath. EPR spectra were recorded on a Bruker ESP-300 spectrometer with 100 kHz field modulation, 1.0 G modulation amplitude, 5.12 ms time constant, 9.769-9.772 GHz microwave frequency, 20 dB (2 mW) microwave power, a 3475-3575 G sweep width, one scan, and a 21 s scan time. The receiver gain was 400 and 1600 for the analysis of 20 and 3.0 mM Cr reactions, respectively. For the time course experiments of the 20 mM Cr reactions, the EPR cavity temperature was regulated at 25 ( 1 °C by a warm nitrogen stream apparatus developed in-house. EPR spectra were recorded for samples in glass capillaries as previously described (12). Reaction of Cr(VI) with Dehydroascorbate and Dilution of Stock Solutions for Spectroscopy or Plasmid DNA Reactions. Initial reactions of Cr(VI) and DHA were carried out at high concentrations to maximize the formation of the chromium intermediates. The NaOAc buffer concentration of 1.0 M was needed to maintain a constant pH over the course of the reaction. Stock reactions were run at room temperature (RT) with 20.0 mM Cr(VI) and 100 mM DHA in 1.0 M NaOAc (pH 3.8, 25 °C) in a total volume of 2.0 mL. Aliquots of the stock solution (0.20 mL) were taken at either 5, 15, 50, or 90 min and added to either (i) 0.80 mL of H2O (with 25 µM DETAPAC) to yield 4.0 mM total Cr and 20 mM total 1 Abbreviations used: DETAPAC, diethylenetriaminepentaacetic acid; DHA, dehydroascorbate; DKG, 2,3-diketogulonic acid; DMPO, 5,5dimethyl-1-pyrroline N-oxide; EHBA, 2-ethyl-2-hydroxybutyric acid; EPR, electron paramagnetic resonance; RT, room temperature.

DHA in 0.20 M NaOAc/20 µM DETAPAC (pH 3.85) or (ii) 0.80 mL of 0.198 M aqueous NaOH/25 µM DETAPAC to yield 4.0 mM total Cr and 20 mM total DHA in 0.20 M NaOAc/20 µM DETAPAC (pH 5.85). Aliquots of these 4.0 mM Cr solutions were immediately added to DNA for strand break analysis or to buffer for spectroscopic analysis. For spectroscopic studies, 0.60 mL of the 4.0 mM Cr solution was diluted into 0.20 mL of 0.20 M NaOAc buffer at either pH 3.8 or pH 6.0 to give 3.0 mM total Cr and 15 mM total DHA. The pH of the 3.0 mM Cr solutions was verified by pH meter and was found to be reproducible at pH 3.80 ( 0.05 and pH 5.85 ( 0.04, respectively (n ) 4). Dilutions from 20 to 4.0 mM Cr and subsequently to 3.0 mM Cr were carried out within 1 min. All reactions were initiated by addition of DHA and were run in air. The concentrations of Cr(V) observed by EPR spectroscopy were estimated by comparison of I∆H2 to a range of known concentrations of Na[CrO(EHBA)2] (16) in 0.10 M EHBA measured with identical parameters, where I ) equals peak to trough intensity and ∆H ) line width between maximum and minimum of the first derivative spectral peak. pBR322 DNA Strand Break Studies. For some studies, contaminating RNA and protein were removed from pBR322 DNA with the SpinBind Miniprep system (FMC BioProducts, Rockland, ME). However, strand breakage was found to be statistically equivalent between experiments using purified and unpurified plasmid DNA; therefore, the presented data represents combined experiments. pBR322 DNA (Gibco-BRL, Gaithersburg, MD) was precipitated with ethanol and 0.3 M NaOAc (pH 5.2), and resuspended in 0.20 M NaOAc/20 µM DETAPAC, at pH 3.8 or 6.0 (37 °C) at 0.10 mM DNA-P. Reactions were carried out with plasmid DNA by addition of 6 µL of the 4.0 mM Cr reactions to 2.0 µL of 0.10 mM DNA-P pBR322 DNA in 0.20 M NaOAc/20 µM DETAPAC (pH 3.8, 37 °C). Reaction solutions of 8 µL contained pBR322 DNA at 25 µM DNA-P (3 nM plasmid), 3.0 mM total Cr, 15 mM total DHA. DNA reactions were initiated by microcentrifugation of reactants and were run at 37 °C for 15 min, followed by termination in an ice water bath. A 2 µL portion of loading buffer (0.4% bromophenol blue, 0.4% xylene cyanol FF, 50% glycerol, 1 mM EDTA in water) was immediately added to each sample. Reaction samples (10 µL) were analyzed by electrophoresis on 1% agrose gels, in 0.5 × TBE buffer (45 mM Tris-HCl, 45 mM boric acid, 1.0 mM EDTA) at 104 V (32 mA) for 1.3 h. Gels were stained in aqueous solutions of ethidium bromide (0.5 µg/mL) for 2535 min, destained in H2O for 45-60 min, and photographed with Polaroid 55 film under short wavelength UV light. Negatives were scanned with a Helena Laboratories Quick Scan R&D densitometer connected to a Hewlett Packard HP3396A integrator to quantitate relative intensities of bands representing supercoiled (form I) and nicked circular (form II) plasmid DNA. Data were corrected for background relaxation in untreated DNA controls ( 5, the lactone ring opens to form 2,3-diketogulonic acid (DKG) (Scheme 1). The rate of decay increases with increasing pH and temperature. At pH 7, the half-life for ring opening of

DNA Strand Breaks by Cr(VI) and Dehydroascorbate

Chem. Res. Toxicol., Vol. 10, No. 3, 1997 273

Scheme 1

DHA is ∼45 min at 23-25 °C but shortens to ∼6-20 min at 37-40 °C (18, 19). However, at pH 3-5, DHA was determined to be stable in solution for at least 4 h (18). Spectroscopic Characterization of the Reaction of Chromium(VI) with DHA. Reactions of Cr(VI) with DHA were carried out at pH 3.8 for several reasons: to minimize the ring opening of DHA to DKG, to optimize the stability of the intermediates for spectroscopic detection, and to compare results using this system with results using the Cr(V)/Cr(IV) complexes of 2-ethyl-2hydroxybutyric acid (EHBA) (20, 21) that were obtained at acidic pH. Aliquots of reaction solutions were diluted for spectroscopic analysis and for reaction with DNA. Stock reactions of Cr(VI) and DHA at this high concentration gave the advantage that Cr(V) was produced in relatively large amounts, and solutions containing identical amounts of intermediates could be adjusted to a different pH simply by changing the pH of the dilution buffer. Thus, the pH changes would affect the reactivity and decay rates of intermediates, while being less complicated by different rates of intermediate formation. The reaction of 20 mM Cr(VI) with 100 mM DHA in 1.0 M NaOAc (pH 3.8, RT) produced a strong Cr(V) EPR signal (g ) 1.980, ∆Hpp ) 0.91 ( 0.06 G, A(53Cr) ) 17.9 ( 0.5 G ) 16.5 cm-1) (Figure 1A). At 25 °C the Cr(V) EPR signal grew over time to reach a maximum intensity at ∼15 min and had decayed by ∼65 min (Figure 1B). An identical reaction monitored by UV/visible spectroscopy showed the growth of an intermediate at 582 nm (Figure 2A), followed by decay to a Cr(III) product at 410 and 568 nm (Figure 2B). However, the intermediate detected by UV/visible spectroscopy did not track with that observed by EPR spectroscopy. The UV/visible intermediate reached maximum intensity at ∼50 min, and the reaction had gone to completion by 105 min. Chromium(V) is the only oxidation state of Cr directly detectable by EPR spectroscopy, but it was initially unclear if the UV/visible intermediate represented Cr(V) or Cr(IV). The identity of the UV/visible intermediate was determined by taking advantage of the fact that Mn(II) can be used to catalyze the disproportionation of Cr(IV) to Cr(V) and Cr(III) at acidic pH (22). If the UV/visible intermediate at 582 nm were Cr(IV), then a reaction run in the presence of Mn(II) should result in a loss of this intermediate. However, Mn(II) was found to have the opposite effect on the UV/visible spectrum (Figure 3). Reaction of 20 mM Cr(VI) with 100 mM DHA in the presence of 20 mM Mn(II) produced the intermediate at 582 nm at greater intensity than reactions without Mn(II), and the intermediate reached a maximum at ∼30

Figure 1. Chromium(V) observed by EPR spectroscopy for reaction of 20 mM Cr(VI) with 100 mM DHA in 1.0 M NaOAc at pH 3.8, 25 °C. (A) Representative Cr(V) signal (g ) 1.980) after reaction time of 17 min at 25 °C. (B) Intensity of Cr(V) EPR signal over time. Data shown represent triplicate reactions.

min, compared to 50 min for reactions without Mn(II). In the presence of Mn(II), the reaction had gone to completion by ∼60 min. Addition of 20 mM Mg(II) to the reaction of Cr(VI) and DHA had no effect on the UV/ visible spectrum (Figure 3), demonstrating that the observations were due to Mn(II) redox activity and not simply to changes in ionic strength. Thus, Mn(II) was reacting with Cr(IV) and was causing Cr(V) to be formed at a faster rate, and the UV/visible intermediate at 582 nm corresponded to Cr(V). Manganese(II) has been used to detect the formation of Cr(IV) during Cr(VI) reactions, through loss of the Mn(II) EPR signal (12). However, this EPR technique cannot be used to detect Cr(IV) at acidic pH, because Mn(II) is reformed catalytically (see Scheme 3, discussed below), thus no loss in the Mn(II) EPR signal would be expected to occur, and this was verified experimentally (data not shown). In reactions analyzed by EPR spectroscopy, Mn(II) did not inhibit the formation of monomeric Cr(V), which is consistent with the UV/visible results and our overall interpretation. The presence of Mn(II) caused line broadening of the Cr(V) EPR signal (data not shown), which precludes a direct quantitation of monomeric Cr(V). The presence of Mg(II) did not affect the EPR spectra (data not shown). Spectroscopic Characterization of Diluted Aliquots from Reactions of Chromium(VI) with DHA. It was concluded that Cr(V) was being detected by both EPR and UV/visible spectroscopies for the reaction of Cr(VI) with DHA. The difference in the EPR and UV/ visible time courses of Cr(V) formation can be explained by the assumption that only monomeric Cr(V) is detected by EPR spectroscopy. If the Cr(V) species were dimer-


Stearns and Wetterhahn

Figure 4. Intensity of Cr(V) EPR signal at g ) 1.980 in diluted solutions. Initial stock reactions of 20 mM Cr(VI) and 100 mM DHA in 1.0 M NaOAc, pH 3.8, were diluted to 3.0 mM total Cr and 15 mM total DHA in 0.20 M NaOAc/20 µM DETAPAC at pH 3.8 (A) or pH 5.8 (B) at various times. Spectra were acquired at RT, 3 min after dilution, with one scan. Values represent average ( SD with n ) 4-5 (5, 15, and 50 min) and n ) 2 (90 min). (*) Statistically significant from 5 min (p < 0.001) and from 15 min (p < 0.05) aliquots. (**) Statistically significant from 5 min and 15 min aliquots (p < 0.001).

Figure 2. Representative spectra for reaction of 20 mM Cr(VI) with 100 mM DHA in 1.0 M NaOAc (pH 3.8) at 25.0 °C observed by UV/visible spectroscopy over time. (A) Formation of intermediate at 582 nm from 0 to 54 min, at 3-min intervals. (B) Decay of intermediate at 582 nm to form Cr(III) product(s) at 410 and 568 nm from 57 to 108 min, at 3-min intervals. Figure 5. Initial absorbance at 579 nm in UV/visible spectra after the stock reaction of 20 mM Cr(VI) and 100 mM DHA was diluted to pH 3.8 (A) or pH 5.8 (B) at different time intervals. Aliquots of the initial reaction were diluted to 3.0 mM total Cr and 15 mM total DHA in 0.20 M NaOAc/20 µM DETAPAC. Absorbance readings were acquired at 37 °C, 1.1 min after dilution. Data are shown as average ( range for duplicate experiments.

Figure 3. Change in absorbance at 582 nm over time for UV/ visible spectra of reactions of 20 mM Cr(VI) with 100 mM DHA in 1.0 M NaOAc (pH 3.8) in the presence and absence of 20 mM MnCl2‚4H2O or 20 mM MgCl2‚6H2O. Data are shown for duplicate scans of (O) reaction of Cr(VI) and DHA; (b) reaction in the presence of Mn(II); and (×) reaction in the presence of Mg(II). Reactions were carried out at 25.0 °C.

izing, presumably through oxo-atom bridges, the two d1 atoms could antiferromagnetically couple to give an EPR silent species (23, 24). UV/visible spectroscopy, on the other hand, could detect total Cr(V). This interpretation was further supported by spectroscopic studies of the diluted reaction solutions. Reactions of 20 mM Cr(VI) and 100 mM DHA were diluted to 3.0 mM Cr after 5, 15,

50, or 90 min. EPR spectra showed Cr(V) at all time points for dilutions to either pH 3.8 or pH 5.8 (Figure 4), but the strongest EPR signal was detected for the 50 min dilutions at either pH. The Cr(V) EPR signals for solutions diluted to pH 3.8 and pH 5.8 were identical to that of the stock reaction solution, except that the Cr(V) signals at pH 3.8 and pH 5.8 were slightly narrower, ∆Hpp ) 0.83 ( 0.05 G as compared to 0.91 G for the 20 mM reaction. The UV/visible spectra of diluted solutions were consistent with the EPR spectra. For the diluted solutions the λmax shifted slightly to 579 nm. The absorbance at 579 nm was highest immediately after dilution and decayed over time (data not shown). At 1.1 min after dilution, a higher absorbance was observed for the 50 min dilutions than for the 5 and 15 min dilutions (Figure 5) at both pH 3.8 and pH 5.8. Thus, in diluted solutions the time courses for both EPR and absorbance measurements of Cr(V) agree that the highest levels of Cr(V) occur at 50 min. Comparison of the intensities of the Cr(V) EPR signals (Figure 4) with the UV/visible absorbances (Figure 5)


Figure 6. Representative agrose gel showing single-strand breaks (form II) induced in pBR322 DNA (25 µM DNA-P) after incubations with reaction solutions diluted to 3.0 mM total Cr and 15 mM total DHA in 0.20 M NaOAc/20 µM DETAPAC, pH 3.8. Aliquots of the initial reaction [20 mM Cr(VI) and 100 mM DHA in 1.0 M NaOAc, pH 3.8, 25 °C] were diluted at 5, 15, 50, and 90 min. Subsequent incubations with DNA were carried out for 15 min at 37 °C. Lane numbers: (1) DNA alone; (2) 3 mM Cr(VI); (3) 15 mM DHA; (4) 5 min aliquot; (5) 15 min aliquot; (6) 50 min aliquot; (7) 90 min aliquot; (8) DNA MW marker.

shows that there is a larger difference between pH 3.8 and pH 5.8 for the EPR spectra than for the UV/visible spectra. However, direct comparison between UV/visible spectra at the different pH values is misleading because the extinction coefficient for Cr(V) is not known and may be pH dependent. In summary, the spectroscopic data suggested that in the 20 mM Cr reactions an EPR silent dimer of Cr(V) was formed as the concentration of Cr(V) increased in the first 50 min of the reaction. When reactions were diluted to 3.0 mM Cr, the dimeric Cr(V) dissociated to form monomeric Cr(V). DNA Strand Breaks Produced by Reaction of Cr(VI) with DHA at pH 3.8 and pH 5.8. Plasmid relaxation experiments were carried out to determine if the intermediates formed during the reaction of Cr(VI) with DHA could cause single-strand breaks. The intermediates Cr(V) (20), Cr(IV) (21), and free radicals (13) have all been proposed to cause plasmid relaxation under various conditions. In the current study, we have taken advantage of the time course for the formation of Cr(V) to test its ability to cause single-strand breaks. It was hypothesized that, if Cr(V) were causing strand breaks, then an aliquot of the stock reaction diluted when maximum Cr(V) was formed should cause the most plasmid relaxation. From the EPR and UV/visible spectroscopic experiments of the 3 mM Cr dilutions (Figures 4 and 5), the highest amount of total Cr(V) was produced at 50 min. Therefore, aliquots of stock reactions [20 mM Cr(VI) and 100 mM DHA at pH 3.8, RT] were taken at 5, 15, 50, or 90 min and diluted to 4 mM Cr and 20 mM DHA at pH 3.8. Diluted solutions were immediately reacted with pBR322 at final concentrations of 3.0 mM Cr, 15 mM DHA, and 0.025 mM DNA-P in 0.20 M NaOAc/20 µM DETAPAC buffer, pH 3.8, 37 °C. The DNA incubations were carried out for 15 min at 37 °C and analyzed by gel electrophoresis. For incubations at pH 3.8, the maximum amount of plasmid relaxation was found to occur with the 5 min aliquot, with decreasing strand breaks occurring for each subsequent time point (Figure 6). Although staggered reactions analyzed on the same gels showed more strand breaks induced with the 5 min aliquot than with the 15 min aliquot, the quantitation of multiple replicates by scanning densitometry showed that within experimental error the strand breaks produced at pH 3.8 with the 5 and 15 min aliquots were statistically equivalent (Figure 7A). The 5 min


Figure 7. Single-strand breaks induced in pBR322 DNA upon incubation with diluted reaction solutions of 3.0 mM total Cr and 15 mM total DHA in 0.20 M NaOAc/20 µM DETAPAC, pH 3.8 or 5.8, 37 °C. Reactions were carried out as described in Figure 6 caption. Values represent average ( SD (n ) 2-12) after subtraction of background strand breaks in DNA controls. (**) indicates strand breaks significantly higher than those for 50 or 90 min aliquots (p < 0.001). (*) indicates strand breaks significantly higher than those for 5, 15, or 90 min aliquots (p < 0.05).

aliquots produced ∼2- and 4-fold more strand breaks than the 50 and 90 min aliquots, respectively. Significantly less single-strand breaks were produced by the reaction aliquots diluted to pH 5.8 (Figure 7B); however, at pH 5.8 it was the aliquot taken at 50 min that produced the most strand breaks, 2-3-fold higher than those with the 5, 15, or 90 min aliquots (discussed further below). The results of the plasmid relaxation experiments at pH 3.8 suggested that Cr(V) may not be the major species responsible for cleavage of DNA, because maximum strand breaks did not occur under conditions producing maximum Cr(V). This observation pointed to the involvement of Cr(IV) in the formation of DNA strand breaks. The ability of Cr(IV) to cause plasmid relaxation was tested for reactions of Cr(VI) and DHA carried out in the presence of Mn(II). The UV/visible experiments (Figure 3) showed that, at the early reaction times, the presence of Mn(II) resulted in increased formation of Cr(V), presumably by facilitating the disproportionation of Cr(IV) to Cr(V) and Cr(III). It was hypothesized that if Cr(IV) were causing strand breaks at pH 3.8, then the presence of Mn(II) should decrease plasmid relaxation for these time points, whereas if Cr(V) were responsible for strand breaks, Mn(II) may increase plasmid relaxation. Stock reactions were carried out as above, with 20 mM Cr(VI), 100 mM DHA, and either 20 mM Mn(II) or 20 mM Mg(II) at pH 3.8 at room temperature. Aliquots were taken at 5 or 15 min, diluted, and reacted with pBR322 DNA at pH 3.8 as described above. For the 5 min aliquots, strand breaks in the presence of Mn(II) were reduced by 79% relative to those in reactions without Mn(II), and for the 15 min aliquots, strand breaks were reduced by 69% with Mn(II) (Figure 8). The presence of Mg(II) did not affect the reaction of Cr(VI) and DHA (Figure 3), but it did affect the production of strand breaks in the DNA incubations (Figure 8). Magnesium(II) was not as efficient as Mn(II) in inhibiting plasmid relaxation. For the 5 and 15 min aliquots, Mg(II) decreased strand breaks by 49% and 51%, respectively (Figure 8). It must be inferred from these Mg(II) control reactions that the ability of Mn(II) to decrease strand breaks in these experiments may be a combination of reductive scavenging of Cr(IV) and cationic association of Mn(II) with the anionic phosphate backbone of DNA,


Stearns and Wetterhahn Scheme 2

Scheme 3

Figure 8. Effect of Mn(II) and Mg(II) on single-strand breaks induced in pBR322 DNA (25 µM DNA-P) after incubation with diluted reaction solutions of 3.0 mM total Cr, 15 mM total DHA, and 3.0 mM Mn(II) or 3.0 mM Mg(II) in 0.20 M NaOAc/20 µM DETAPAC, pH 3.8, 37 °C. Reactions were carried out as described in Figure 3 caption and aliquots diluted at 5 or 15 min. DNA incubations were carried out for 15 min at 37 °C and analyzed by gel electrophoresis. Values represent average ( SD (n ) 3-12) after subtraction of background strand breaks in DNA controls. (a) Statistically significant from Cr(VI) + DHA reactions (p < 0.001) and reactions + Mg(II) (p < 0.01). (b) Statistically significant from Cr(VI) + DHA reactions (p < 0.001). (c) Statistically significant from Cr(VI) + DHA reactions (p < 0.001) and reactions + Mg(II) (p < 0.05). (d) Statistically significant from Cr(VI) + DHA reactions (p < 0.001).

which could block interactions between DNA and reactive chromium species. After taking this cation effect into consideration, the plasmid relaxation experiments and spectroscopic characterization of this system (Figures 3-5) are still fully consistent with the interpretation that Cr(IV) is the major species producing strand breaks for the early dilution times at pH 3.8. The use of Mn(II) in these experiments also leads to a caveat: in the absence of control reactions and without spectroscopic experiments under the exact conditions used to induce DNA damage, the ability of Mn(II) to decrease Cr(VI)-induced strand breaks should be interpreted with caution (25). At pH 3.8, the formation of DNA strand breaks did not track with the formation of detectable Cr(V). In contrast, at pH 5.8 (Figure 7), the strand breaks were highest under conditions that produced maximum Cr(V), as detected by both EPR (Figure 4) and UV/visible spectroscopy (Figure 5). The effect of Mn(II) on strand breaks at pH 5.8 was inconclusive because control reactions of Mn(II) + DHA produced significant strand breakage (data not shown). This was likely due to the reaction of Mn(II) with DKG (12), which would be present at pH > 5. Like sorbitol and related ligands (26), DKG may coordinate Mn(II) and enhance the metal’s reactivity with oxygen, which would produce high-valent Mn/peroxo species and increase the formation of DNA strand breaks. The spectroscopic and plasmid relaxation results are consistent with the interpretation that Cr(V) is more reactive than Cr(IV) at pH 5.8; however, further experimentation is required to confirm this.

Discussion Chromium(VI) reacts with dehydroascorbate to produce Cr(V) and Cr(IV) as reactive intermediates. The organic product(s) of this reaction has not been determined. The most likely site for oxidation is the hydroxyl group of carbon 5 in the bicyclic form of DHA or the hydroxyl groups of carbons 5 and 6 in the monocyclic form

(Scheme 1). The EPR parameters for Cr(V) are consistent with an oxo chromium in a five-coordinate square pyramidal geometry with oxygen donor ligands (27). The g value of 1.980 and A(53Cr) hyperfine of 17.9 G (16.5 cm-1) are identical to that observed in the reaction of Cr(VI) with ascorbate (12). The λmax of 579-582 nm in the UV/visible spectra of reaction solutions is also consistent with Cr(V). Although few Cr(V) complexes have been characterized, a λmax range of 506-519 nm was observed for chromium(V) complexes of EHBA and other tertiary hydroxy acids (16). Thus, both EPR and UV/ visible spectra are consistent with the assignment that they represent Cr(V). Mechanisms consistent with the spectroscopic observations are presented in Schemes 2 and 3. Scheme 2 outlines a simplified mechanism for the reaction of Cr(VI) with DHA in the absence of Mn(II). The first step is a two-electron reduction of Cr(VI) to form Cr(IV) and “oxidized DHA” (eq 2). The intermediate Cr(IV) may disproportionate to form Cr(V) and Cr(III) (eq 3). When Cr(V) is produced in sufficient amounts, an equilibrium may be formed between monomeric and dimeric species (eq 4). The number of bridging oxo groups is most likely two based on a representative structure (23). Upon dilution of stock solutions, the dimer may dissociate (eq 5). The Cr(V) intermediate is reduced to Cr(III) by DHA (eq 6); however, reduction of dimeric Cr(V) by DHA may also occur. In the presence of Mn(II) (Scheme 3), the Cr(IV) intermediate is reduced to Cr(III) with subsequent formation of Mn(III) (eq 8). The Mn(III) product then reacts with another molecule of Cr(IV) to produce Cr(V), regenerating Mn(II) (22) (eq 9). The Cr(V) intermediate may dimerize and/or be reduced to Cr(III) by DHA (eq 6). The overall reaction (eq 7) is the same in the presence and in the absence of Mn(II). This mechanism at pH 3.8 (Scheme 3) differs from that proposed for the reaction of Cr(VI) and ascorbate at pH 7 (12) because the Mn(II) reaction with Cr(IV) is catalytic in acidic solution (22), but at pH 7, Mn(II) was not observed to be regenerated (12), thus eq 9 does not occur at pH 7. Although the reaction between ascorbic acid and Cr(VI) has been thoroughly studied, the further reaction of Cr(VI) with the presumed end product DHA has not been reported prior to our original prediction (12). Kinetic studies of the reaction of Cr(VI) with ascorbate used excess ascorbate (28-30). Our spectroscopic experiments suggest that Cr(VI) reacts more slowly with DHA than with ascorbate; therefore, with the conditions used


in the previous kinetic studies this reaction would have been disfavored. The intermediates produced in the reaction of Cr(VI) with DHA caused strand breaks in plasmid DNA. One of the most interesting observations in this study was that there was a pH effect for the production of DNA damage that was related to the time course of the reaction. The combination of plasmid relaxation experiments with spectroscopic analysis of the reaction intermediates suggests that Cr(IV) may be the dominant species producing the extensive strand breakage observed at pH 3.8, whereas Cr(V) may produce the modest strand breakage observed at pH 5.8. In a prior study of DNA strand breaks induced by Cr(V) and Cr(IV) complexes of EHBA at pH 3.7 (20), it was concluded that Cr(IV) may be more damaging to DNA than Cr(V). However, those experiments were not carried out at pH > 5, and it is also difficult to rule out the possibility that Cr(IV) formed from minor disproportionation of [Cr(V)O(EHBA)2]- may be contributing to strand breaks rather than Cr(V) per se. The results of the current study suggest that at higher pH the Cr(IV) species in the DHA reaction may be destabilized so that other decomposition pathways, for example, ligand oxidation, may compete with pathways leading to DNA strand breaks. Presumably Cr(V) is destabilized as well, since EPR and UV/visible studies showed less Cr(V) at pH 5.8 than at pH 3.8, and the best conditions that produced strand breaks at pH 5.8, dilutions after 50 min reaction, resulted in 5-fold lower strand breaks than the 5 min dilutions at pH 3.8. The data presented here have provided a working hypothesis that can now be tested at higher pH, namely, that Cr(V) will be more likely than Cr(IV) to cause strand breaks under physiological conditions. The Cr/EHBA system that has been found to cause DNA strand breaks at low pH and to be mutagenic in the Ames test (19) suggests that activity detected at low pH may be relevant to damage induced under physiological conditions. Many investigations have been aimed at determining the reactive species responsible for Cr(VI)-induced DNA strand breaks with ascorbate (13, 31) and glutathione (31) as Cr reductants. Unfortunately, comparisons are often made between experiments carried out with different reactant concentrations, buffers, and incubation times, which complicate the interpretation. Furthermore, conclusions are often drawn in the absence of spectroscopic data. We propose that it is not a single type of reactive species that universally produces the strand breaks observed in different Cr(VI) systems. The DNA damaging agent will depend on the chosen conditions. This proposal may help to resolve many of the conflicting data from in vitro studies. In this work, results are consistent with the interpretation that Cr(IV) may be more active toward strand breakage than Cr(V) at low pH; however, the opposite trend appears to occur at higher pH. Under different conditions with Cr(VI) and ascorbate, we reported that carbon-based radicals may be more active than Cr(V) (13). Carbon-based free radicals were not observed in the current reactions (data not shown). Strand breaks were not observed in the Cr(VI)/glutathione system under conditions that produced Cr(V) and resulted in Cr-DNA adducts (32, 33). However, with a different Cr(VI)/glutathione concentration range and longer incubation times, Kortenkamp and co-workers observed DNA strand breaks and apurinic/ apyrimidinic (AP) sites that were oxygen-dependent (31). Experiments by these authors also showed that the


ascorbate system was not identical to that of glutathione, in that low Cr(VI) and ascorbate concentrations produced oxygen-dependent strand breaks and AP sites, but that the oxygen dependence was lost at higher reactant concentrations (31). In spite of this, it was proposed that only high-valent chromium oxo/peroxo species would cause strand breaks (31). What these disparate results may also mean for Cr(VI)-induced genotoxicity is that strand breaks detected in different tissues or cultured cells may not have been caused by the same reactive intermediates. Nor should it be assumed that there must be only one chemical product from DNA cleavage; for example, different reactive species may favor reactions at different sites in the DNA, and both pH and buffer may influence the pathway of sugar ring degradation that produces the strand break. Investigations aimed at characterization of the degradation products should clarify these issues (31, 34, 35). The reaction of Cr(VI) with dehydroascorbate is introduced here as another system that can be used in conjunction with the ascorbate and glutathione systems to evaluate the role of intermediates in Cr-induced DNA damage. These experiments serve as one more example of how the formation and stability of reactive intermediates strongly depend on reaction conditions and demonstrate that, without spectroscopic analysis under the identical conditions found to produce DNA lesions, one could easily be misled in the assignment of the species responsible for DNA damage. The species producing damage, the extent of damage, and the type of DNA lesion produced will be dependent on the chosen experimental conditions. Understanding this variability in Cr(VI) reactions is crucial for interpretation of in vitro studies that are aimed at deciphering the molecular mechanisms of Cr(VI)-induced cancers.

Acknowledgment. This investigation was supported by PHS Grants CA34869, ES07167 (K.E.W.), and CA59292 (D.M.S.). The EPR spectrometer was purchased with funding from NSF Grant CHE-8701406.

References (1) Connett, P. H., and Wetterhahn, K. E. (1983) Metabolism of the carcinogen chromate by cellular constitutents. Struct. Bonding 54, 93-124. (2) Cohen, M. D., Kargacin, B., Klein, C. B., and Costa, M. (1993) Mechanisms of chromium carcinogenicity and toxicity. Crit. Rev. Toxicol. 23, 255-281. (3) De Flora, S., Bagnasco, M., Serra, D., and Zanacchi, P. (1990) Genotoxicity of chromium compounds. A review. Mutat. Res. 238, 99-172. (4) Tsapakos, M. J., Hampton, T. H., and Wetterhahn Jennette, K. (1981) The carcinogen chromate induces DNA cross-links in rat liver and kidney. J. Biol. Chem. 256, 3623-3626. (5) Cupo, D. Y., and Wetterhahn, K. E. (1985) Modification of chromium(VI)-induced DNA damage by glutathione and cytochromes P-450 in chicken embryo hepatocytes. Proc. Natl. Acad. Sci. U.S.A. 82, 6755-6759. (6) Hamilton, J. W., and Wetterhahn, K. E. (1986) Chromium(VI) induced DNA damage in chick embryo liver and blood cells in vivo. Carcinogenesis 7, 2085-2088. (7) DeMarini, D. M., Brockman, H. E., de Serres, F. J., Evans, H. H., Stankowski, L. F., Jr., and Hsie, A. W. (1989) Specific-locus mutations induced in eukaryotes (especially mammalian cells) by radiation and chemicals: a perspective. Mutat. Res. 220, 11-29. (8) Solomon, E., Borrow, J., and Goddard, A. D. (1991) Chromosome aberrations and cancer. Science 254, 1153-1160. (9) Suzuki, Y. (1988) Reduction of hexavalent chromium by ascorbic acid in rat lung lavage fluid. Arch. Toxicol. 62, 116-122. (10) Standeven, A. M., and Wetterhahn, K. E. (1992) Ascorbate is the principal reductant of chromium(VI) in rat lung ultrafiltrates and cytosols, and mediates Cr-DNA binding in vitro. Carcinogenesis 13, 1319-1324.

278 Chem. Res. Toxicol., Vol. 10, No. 3, 1997 (11) Capellmann, M., and Bolt, H. M. (1992) Chromium(VI) reducing capacity of ascorbic acid and of human plasma in vitro. Arch. Toxicol. 66, 45-50. (12) Stearns, D. M., and Wetterhahn, K. E. (1994) Reaction of chromium(VI) with ascorbate produces chromium(V), chromium(IV), and carbon-based radicals. Chem. Res. Toxicol. 7, 219-230. (13) Stearns, D. M., Kennedy, L. J., Courtney, K. D., Giangrande, P. H., Phieffer, L. S., and Wetterhahn, K. E. (1995) Reduction of chromium(VI) by ascorbate leads to chromium-DNA binding and DNA strand breaks in vitro. Biochemistry 34, 910-919. (14) Stearns, D. M., and Wetterhahn, K. E. (1993) Reduction of carcinogenic chromium(VI) by dehydroascorbate and its decomposition products. J. Inorg. Biochem. 51, 430. (15) Tolbert, B. M., and Ward, J. B. (1982) Dehydroascorbic acid. In Ascorbic Acid: Chemistry, Metabolism and Uses (Seib, P. A., and Tolbert, B. M., Eds.) pp 101-123, American Chemical Society, Washington, DC. (16) Krumpolc, M., and Rocˇek, J. (1979) Synthesis of stable chromium(V) complexes of tertiary hydroxy acids. J. Am. Chem. Soc. 101, 3206-3209. (17) Sisley, M. J., and Jordon, R. B. (1992) Kinetic study of the reaction of 3-methylascorbic acid with aqueous iron(III). Inorg. Chem. 31, 2137-2143. (18) Bode, A. M., Cunningham, L., and Rose, R. C. (1990) Spontaneous decay of oxidized ascorbic acid (dehydro-L-ascorbic acid) evaluated by high-pressure liquid chromatography. Clin. Chem. 36, 18071809. (19) Huelin, F. E. (1949) Investigations on the stability and determination of dehydroascorbic acid. Aust. J. Sci. Res. Ser. B. 2, 346354. (20) Farrell, R. P., Judd, R. J., Lay, P. A., Dixon, N. E., Baker, R. S. U., Bonin, A. M. (1989) Chromium(V)-induced cleavage of DNA: are chromium(V) complexes the active carcinogens in chromium(VI)-induced cancers? Chem. Res. Toxicol. 2, 227-229. (21) Barr-David, G., Hambley, T. W., Irwin, J. A., Judd, R. J., Lay, P. A., Martin, B. M., Bramley, R., Dixon, N. E., Hendry, P., Ji, J.Y., Baker, R. S. U., and Bonin, A. M. (1992) Suppression by vanadium(IV) of chromium(V)-mediated DNA cleavage and chromium(VI/V)-induced mutagenesis. Synthesis and crystal structure of the vanadium(IV) complex (NH4)[V(O){HOC(Et2)COO}{OC(Et2)COO}]. Inorg. Chem. 31, 4906-4908. (22) Ghosh, M. C., Gelerinter, E., and Gould, E. S. (1992) Electron transfer. 111. Disproportionation of carboxylato-bound chromium(IV). Catalysis by manganese(II). Inorg. Chem. 31, 702-705. (23) Nishino, H., and Kochi, J. K. (1990) Unusually stable chromium(V) perfluoropinacolate complexes. Inorg. Chim. Acta 174, 93102.

Stearns and Wetterhahn (24) Farrell, R. P., and Lay, P. A. (1992) New insights into the structures and reactions of chromium(V) complexes: implications for chromium(VI) and chromium(V) oxidations of organic substrates and the mechanisms of chromium-induced cancers. Commun. Inorg. Chem. 13, 133-175. (25) Shi, X., Mao, Y., Knapton, A. D., Ding, M., Rojanasakul, Y., Gannett, P. M., Dalal, N., and Liu, K. (1994) Reaction of Cr(VI) with ascorbate and hydrogen peroxide generates hydroxyl radicals and causes DNA damage: role of a Cr(IV)-mediated Fenton-like reaction. Carcinogenesis, 15, 2475-2478. (26) Richens, D. T., Smith, C. G., and Sawyer, D. T. (1979) Sorbitol and related polyol complexes of manganese (II), -(III), and -(IV): redox and oxygenation equilibria. Inorg. Chem. 18, 706-712. (27) Barr-David, G., Charara, M., Codd, R., Farrell, R. P., Irwin, J. A., Lay, P. A., Bramley, R., Brumby, S., Ji, J.-Y., and Hanson, G. R. (1995) EPR characterisation of the CrV intermediates in the CrVI/V oxidations of organic substrates and of relevance to Crinduced cancers. J. Chem. Soc., Faraday Trans. 91, 1207-1216. (28) Baˆldea, J., and Munteanu, L. (1980) The kinetics of the oxidation of ascorbic acid by chromate Studia Univ. Babes-Bolyai. Chem. 25, 24-31. (29) Banas´, B. (1981) A stopped-flow study of the oxidation reaction of ascorbic acid with chromic acid. Inorg. Chim. Acta 53, L13L15. (30) Connett, P. H., and Wetterhahn, K. E. (1985) In vitro reaction of the carcinogen chromate with cellular thiols and carboxylic acids. J. Am. Chem. Soc. 107, 4282-4288. (31) Kortenkamp, A., Casadevall, M., and Da Cruz Fresco, P. (1996) The reductive conversion of the carcinogen chromium(VI) and its role in the formation of DNA lesions. Annals Clin. Lab. Sci. 26, 160-175. (32) Aiyar, J., Borges, K. M., Floyd, R. A., and Wetterhahn, K. E. (1989) Role of chromium(V), glutathione thiyl radical and hydroxyl radical intermediates in chromium(VI)-induced DNA damage. Toxicol. Environ. Chem. 22, 135-148. (33) Borges, K. M., Boswell, J. S., Liebross, R. H., and Wetterhahn, K. E. (1991) Activation of chromium(VI) by thiols results in chromium(V) formation, chromium binding to DNA and altered DNA conformation. Carcinogenesis 12, 551-561. (34) Bose, R. N., and Fonkeng, B. S. (1996) Mechanisms of DNA damage by chromium carcinogens. J. Chem. Soc., Chem. Commun. 2211-221. (35) Sugden, K. D., and Wetterhahn, K. E. (1996) Identification of the oxidized products formed upon reaction of chromium(V) with thymidine nucleotides. J. Am. Chem. Soc. 118, 10811-10818.

TX9601521

with Dehydroascorbate Cause Single-Strand Breaks in Plasmid DNA

Recommend Documents