Uranyl Acetate Causes DNA Single Strand Breaks In Vitro in the

Monica Yazzie, Shania L. Gamble, Edgar R. Civitello, and Diane M. Stearns*. Department of Chemistry, Northern Arizona University, Flagstaff, Arizona, ...
0 downloads 0 Views 157KB Size
Download PDF
524

Chem. Res. Toxicol. 2003, 16, 524-530

Uranyl Acetate Causes DNA Single Strand Breaks In Vitro in the Presence of Ascorbate (Vitamin C) Monica Yazzie, Shania L. Gamble, Edgar R. Civitello, and Diane M. Stearns* Department of Chemistry, Northern Arizona University, Flagstaff, Arizona, 86011 Received December 23, 2002

Uranium is a radioactive heavy metal with isotopes that decay on the geological time scale. People are exposed to uranium through uranium mining, processing, the resulting mine tailings, and the use of depleted uranium in the military. Acute exposures to uranium are chemically toxic to the kidney; however, little is known about chronic exposures, for example, if there is a direct chemical genotoxicity of uranium. The hypothesis that is being tested in the current work is that hexavalent uranium, as uranyl ion, may have a chemical genotoxicity similar to that of hexavalent chromium. In the current study, reactions of uranyl acetate (UA) and ascorbate (vitamin C) were observed to produce plasmid relaxation in pBluescript DNA. DNA strand breaks increased with increasing concentrations of a 1:1 reaction of UA and ascorbate but were not affected by increasing the ratio of ascorbate. Plasmid relaxation was inhibited by coincubation of reactions with catalase but not by coincubation with the radical scavengers mannitol, sodium azide, or 5,5-dimethyl-1-pyrroline-N-oxide. Reactions of UA and ascorbate monitored by 1H NMR spectroscopy showed formation of a uranyl ascorbate complex, with no evidence of a dehydroascorbate product. A previous study inferred that hydroxyl radical formation was responsible for oxidative DNA damage in the presence of reactions of uranyl ion, hydrogen peroxide, and ascorbate [Miller et al. (2002) J. Bioinorg. Chem. 91, 246-252]. Current results, in the absence of added hydrogen peroxide, were not completely consistent with the interpretation that strand breaks were produced by a Fenton type generation of reactive oxygen species. Data were also consistent with the interpretation that a uranyl ascorbate complex was catalyzing hydrolysis of the DNA-phosphate backbone, in a manner similar to that known for the lanthanides. These data suggest that uranium may be directly genotoxic and may, like chromium, react with DNA by more than one pathway.

Introduction Uranium is a heavy metal whose chemical toxicity is poorly understood. The element has 16 different isotopes, only three of which exist in significant amounts. Uranium238 is the most abundant and accounts for 99.27% of the naturally occurring metal. The natural abundances of U-235 and U-234 are 0.72 and 0.006%, respectively. All three of these isotopes are radioactive; however, their half-lives are on the geological time scale. The half-life for U-238 is 4.5 × 109 years, the t1/2 of U-235 is 7 × 108 years, and that of U-234 is 2.5 × 105 years. All three of these isotopes decay by R- and β-emission. The relevant oxidation states of uranium are 6+, 5+, 4+, and 3+, with hexavalent and tetravalent uranium being the most stable. Uranium exists in minerals and seawater in the hexavalent oxidation state. Its molecular geometry is almost always in the form of the linear uranyl dication, OdUdO2+. There are three major uses for uranium. The first use of the U-235 isotope was as the source of fissionable material in atomic weapons. More recently, uranium enriched in U-235 serves as fuel for nuclear reactors. Depleted uranium (DU), in which 50-70% of the U-235 has been extracted, is currently used in antitank weapons, tank armor, and ammunition rounds because of the * To whom correspondence should be addressed. Tel: (928)523-4460. Fax: (928)523-8111. E-mail: [email protected].

metal’s density. A minor use of uranium has been as a pigment to color ceramics and glass. All of these uses provide opportunities for occupational and environmental exposures. Uranium exposure has been associated with lung cancer, with most of the epidemiological data provided by mining exposures. In the U.S., the uranium mining industry started in 1947 in and around the Navajo Indian Reservation in the Colorado Plateau region of Arizona, New Mexico, Colorado, and Utah. Subsequently, uranium mining was linked to the development of lung cancer in the Navajo (1). However, the observed increases in lung cancer were attributed to radon, a radioactive gas that is formed upon radioactive decay of uranium-238, rather than the uranium itself (1). Although the ability of radon to cause lung cancer is well-established (2), the chemical carcinogenicity of uranium itself has not been ruled out. Occupational exposure to uranium through military action and in uranium processing facilities has been considered to be too low to cause significant cancer risks (3). However, a recent cumulative report found a slight risk of lung cancer in soldiers with the highest exposures to DU, but the risk was expected to be less than twice that of the general population (4). More recently, implantation of DU fragments in rats produced a significant increase in soft tissue sarcomas relative to tantalum fragments used as a foreign body control (5). At this time, no final conclusions can yet be drawn regarding cancer risks from DU exposure.

10.1021/tx025685q CCC: $25.00 © 2003 American Chemical Society Published on Web 03/22/2003

DNA Strand Breaks by Uranyl Acetate and Ascorbate

There is a growing concern over the environmental exposures to uranium that have resulted from mine tailings left over from uranium extraction; yet, mechanisms behind uranium-induced health effects have not been deciphered. A weak link between birth defects, stillbirths, and adverse outcomes of pregnancy was suggested for Navajo women living near uranium mine tailings (6). In 1972, a study of cancer mortality was carried out for men living in counties of Colorado where uranium tailings were used as construction fill. No increases in lung cancers or leukemia were found; however, increases in prostate, pancreas, stomach, and colon cancer mortalities were deemed significant relative to rates in populations not exposed to uranium tailings (7). The authors noted that these cancers were not expected to be associated with ionizing radiation. More recently, significant increases in gastric cancer were reported for counties in New Mexico with significant deposits of uranium or uranium tailings (8). This author also noted that gastric cancer was not expected to be associated with radiation and suggested that the causative agents could have been the cocontaminants arsenic, cadmium, selenium, molybdenum, lead, or cyanide. Although health effects from environmental exposures are beginning to be documented, the potential chemical genotoxicity of uranium was not considered in any of these cases. We hypothesize that uranium may be chemically genotoxic because there are some parallels between the chemistry of uranium and the chemistry of chromium, which is a known human lung carcinogen. Chromium exists in the six oxidations states of 6+, 5+, 4+, 3+, 2+ and 0, which is similar to the uranium oxidation states of 6+, 5+, 4+, 3+, and 0. One difference, however, is that for chromium the 6+ and 3+ forms are the most stable, whereas for uranium the 6+ and 4+ states are stable. Chromium6+ requires metabolic activation for DNA damage. Ascorbate (Asc), GSH, and Cys are major reductants of chromate6+ (Cr6+) (9). Very little is known about the biological metabolism of uranium outside of three reports: uranium (UO22+) has been shown to catalytically oxidize Asc to dehydroascorbate (DHA) at low pH (0.992.00) in the presence of dioxygen (10), uranyl nitrate produced hydroxyl radicals in the presence of H2O2 at pH < 4 (11), and DU was recently found to cause oxidative damage to calf thymus DNA in the presence of H2O2 and Asc (12). We are testing the hypothesis that uranium may be activated by biomolecules to cause DNA damage, by pathways similar to those known for chromium(VI) genotoxicity. Consistent with this hypothesis, results reported here demonstrate that uranium and Asc, in the absence of H2O2, can produce single strand breaks in plasmid DNA in vitro.

Experimental Procedures Chemicals and Reagents. DU as uranyl acetate dihydrate (6159-44-0, UA) was obtained from Spectrum Chemical Mfg. Corp. (Gardena, CA) and used as received. A U234/U238 activity ratio of 0.12 was determined by R-spectrometry (Department of Geological Sciences, Florida State University). L-Ascorbic acid (Sigma-Aldrich Chemical Corp., St. Louis, MO) and tris(hydroxymethyl)aminomethane hydrochloride (Tris-HCl) (Fisher Scientific Company, Pittsburgh, PA) were treated with Chelex 100 resin (Bio-Rad Laboratories, Hercules, CA) to remove trace metals. Mannitol (Mallinckrodt, Phillipsburg, NJ) and sodium azide (Mallinckrodt) were used as received. Solutions of 5,5-

Chem. Res. Toxicol., Vol. 16, No. 4, 2003 525

Figure 1. Comparison of the reaction of Asc (1.0 mM) with either UA (1.0 mM) or Cr (1.0 mM) in the presence of pBluescript SK+ plasmid (0.30 mM DNA-P) in 0.10 M Tris-HCl (pH 6.9, 37 °C). Reactions were incubated for 30 min prior to gel electrophoresis. (A) Representative gel: lane 1, untreated DNA; 2, Asc; 3, UA; 4, UA + Asc; 5, Cr; 6, Cr + Asc. (B) Quantitation of % plasmid relaxation relative to supercoiled DNA per lane. Data represent mean ( SE for n ) 4-5 experiments. dimethyl-1-pyrroline-N-oxide (DMPO, Sigma) were purified by filtration over activated charcoal. Catalase from bovine liver, without thymol, 2200 U/mg (Sigma), was used as received or denatured by boiling stock solutions of 10 mg/mL in Tris-HCl (pH 7.0, 37 °C) for 5 min. The pBluescript SK+ plasmid was obtained by transformation of Escherichia coli strain TOP10F′ followed by purification with a QIAfilter plasmid Mega Kit (Qiagen, Inc., Valencia, CA). Gel Electrophoresis. Reactions of UA (0.1-1 mM) and Asc (0.1-5 mM) were carried out in 0.10 M Tris-HCl (pH 7.0, 37 °C) in the presence of pBluescript SK+ plasmid (0.30 mM DNAP, 300 µg) for 30 min in the dark at 37 °C. In some experiments, UA, Asc, and DNA were coincubated with active or denatured catalase (30 µg/mL or 66 U/mL) or with the radical scavengers mannitol, sodium azide, or DMPO at 0.50 mM concentrations. For the 1.0 mM UA reactions, the final reaction pH was measured to be 6.9 (37 °C); for the 0.10 and 0.50 mM UA reactions, the final reaction pH remained at 7.0 (37 °C). Relaxation of supercoiled plasmid DNA was observed by gel electrophoresis on 1% agarose gels with 0.5× TBE running buffer (45 mM Tris, 45 mM boric acid, 1 mM EDTA) at 120 V. Gels were stained with ethidium bromide, destained with water, visualized on an ULTRA‚LU h M Electronic Dual Transiluminator, and digitally photographed with a Panasonic CCD Digital Camera wv-BP332 equipped with a Rainbow TV Zoom Lens. The % plasmid relaxation relative to supercoiled DNA was quantified from digital images using UN-SCAN-IT gel software version 5.1 (Silk Scientific Corp., Orem, UT). Small differences in % plasmid nicking were sometimes observed between subsets of experiments and were likely due to differences in removing trace metals from stock solutions of Asc or minor differences in solution pH values, which would affect the extent of UA oligomerization. Data displayed in Figure 2B represent the summation of reactions of UA and Asc with DNA for all experiments (n ) 9-27). 1H NMR Spectroscopy. Reactions of 50 mM UA and either 50 mM Asc or DHA were carried out in D2O at room temperature. 1H NMR spectra were acquired on a 400 MHz Varian

526

Chem. Res. Toxicol., Vol. 16, No. 4, 2003

Figure 2. Effect of concentration on the reaction of UA (0.101.0 mM) with 1 equiv of Asc in the presence of pBluescript SK+ plasmid DNA (0.30 mM DNA-P) in 0.10 M Tris-HCl (pH 6.97.0, 37 °C). Reactions were incubated for 30 min prior to gel electrophoresis. (A) Representative gel: lane 1, untreated DNA; 2, 0.10 mM Asc; 3, 0.10 mM UA; 4, 0.10 mM reaction; 5, 0.50 mM Asc; 6, 0.50 mM UA; 7, 0.50 mM reaction; 8, 1.0 mM Asc; 9, 1.0 mM UA; 10, 1.0 mM reaction. (B) Quantitation of % plasmid relaxation relative to supercoiled DNA per lane. Data represent mean ( SE for n ) 9-27 experiments. (***) Values for plasmid relaxation from UA + Asc reactions were significantly different from both Asc alone and UA alone at p < 0.001. Differences between Asc alone and UA alone were not statistically significant at any concentration. Mercury VX Spectrometer at 25 °C with 56 scans and standard proton parameters. Hazards. UA is a radioactive compound and a possible carcinogen. Ethidium bromide is a mutagen. Inhalation of and skin contact with these chemicals should be avoided, and they should be disposed of properly. Aqueous ethidium bromide waste was purified using a Schleicher and Schuel ethidium bromide waste reduction system (VWR International, West Chester, PA).

Results The hypothesis being tested was that if uranium has a reactivity with biological reducing agents that is similar to Cr(VI), then uranium may cause direct DNA damage, similar to Cr(VI). This hypothesis was tested by evaluating the ability of UA to cause single strand breaks in plasmid DNA in the presence of Asc. Chromium(VI) has been shown to relax plasmid DNA in the presence of Asc with more strand breaks occurring for reactions in TrisHCl buffer than in HEPES buffer (13). Incubation of 1.0 mM UA with 1.0 mM Asc in 0.10 M Tris-HCl (pH 6.9, 37 °C) was compared to reaction of Cr(VI) with Asc at equivalent concentrations. The reaction of UA with Asc in the presence of DNA caused 80% plasmid relaxation or 6-8-fold higher levels of strand breaks than reactions of DNA with either Asc or UA alone (Figure 1A, lane 4 vs lanes 2 and 3, Figure 1B), and the UA-Asc reaction caused 5-fold more plasmid relaxation than the equivalent Cr(VI) reaction (Figure 1A, lane 4 vs lane 6, Figure 1B). The background plasmid relaxation observed for Asc reactions with DNA (Figure 1A, lane 2) is presumed to

Yazzie et al.

be due to the presence of trace iron that was not removed by Chelex pretreatment. Results were consistent with a chemical reactivity that was responsible for DNA strand breaks rather than R-particle decay. Radioactive decay is independent of chemical reactivity; therefore, if R-particle decay was solely responsible for plasmid relaxation, then UA itself should have produced the same amount of plasmid nicking as that observed for reactions of UA with Asc (Figure 1A, lane 3 vs lane 4). A minor contribution of uranium radioactive decay could not be ruled out since a slight amount of plasmid relaxation was observed for UA alone (Figure 1A, lane 3); however, this could also be due to trace iron given that the amount of radioactive decay that would be expected over a 30 min period for isotopes with half-lives of 105-109 years would be negligible. Alternatively, uranyl ion could weakly catalyze the hydrolysis of DNA in a manner similar to that shown for the lanthanides (14) (vide infra). The formation of DNA single strand breaks was proportional to the reaction concentration of UA with a stoichiometric amount of Asc. Plasmid DNA was incubated with UA, Asc, or reactions of UA and Asc at increasing concentrations of 0.10, 0.50, or 1.0 mM UA and Asc (Figure 2A,B). At each concentration level, the amount of strand breaks observed was significantly greater for the UA-Asc reaction than for either reagent separately, and the amount of strand breaks increased on the order of 0.10 mM (33.1 ( 4.9%, n ) 15) < 0.50 mM (56.4 ( 4.9%, n ) 13) < 1.0 mM (83.5 ( 3.9%, n ) 11) (Figure 2A, lanes 4, 7, and 10; Figure 2B). Strand breaks induced by Asc alone ranged from 17 to 22%, and strand breaks induced by UA alone ranged from 18 to 24% across the three tested concentrations (Figure 2B). Results are consistent with the interpretation that the presence of both UA and Asc was necessary for the generation of DNA strand breaks. It was hypothesized that a reactive intermediate could be responsible for generation of DNA strand breaks. Attempts were made to test this hypothesis by determining the effect of preincubation of uranium and Asc solutions before reaction with DNA. However, precipitation of reaction products over extended reaction times produced a decrease in plasmid relaxation (data not shown) that masked any decrease in plasmid relaxation that could have occurred from loss of reactive intermediates. The effect of the ratio of Asc to uranium on DNA strand breaks was also explored. Reactions of 0.10 mM UA with Asc concentrations of 0.050-0.50 mM were carried out in the presence of DNA plasmid in 0.10 M Tris-HCl (pH 7.0, 37 °C) in triplicate (Figure 3). Increasing concentrations of Asc produced a low background level of 7-9% plasmid relaxation (Figure 3A, lanes 2-5, Figure 3B), and UA alone produced 27% relaxation (Figure 3A, lane 6, Figure 3B). A higher level of 42-45% plasmid relaxation was observed for the reactions lanes; however, increasing the ratios of Asc to UA did not significantly change the amount of strand breaks produced (Figure 3, lanes 7-10, Figure 3B). These results differed from those obtained with Cr(VI). In the Cr(VI) Asc reactions, the presence of excess Asc quenched the formation of strand breaks, presumably by quenching the high valent Cr species or free radicals (13). Therefore, current results were not consistent with the interpretation that free radical intermediates were causing single strand breaks

DNA Strand Breaks by Uranyl Acetate and Ascorbate

Figure 3. Effect of stoichiometry on the reaction of UA (0.10 mM) with varying equivalents of Asc in the presence of pBluescript SK+ plasmid DNA (0.30 mM DNA-P) in 0.10 M Tris-HCl (pH 7.0, 37 °C). Reactions were incubated for 30 min prior to gel electrophoresis. (A) Representative gel: lane 1, untreated DNA; 2, 0.050 mM Asc; 3, 0.10 mM Asc; 4, 0.20 mM Asc; 5, 0.50 mM Asc; 6, 0.10 mM UA; 7, 0.10 mM UA + 0.050 mM Asc; 8, 0.10 mM UA + 0.10 mM Asc; 9, 0.10 mM UA + 0.20 mM Asc; 10, 0.10 mM UA + 0.50 mM Asc. (B) Quantitation of % plasmid relaxation relative to supercoiled DNA per lane. Data represent mean ( SE for n ) 3 experiments.

in the uranium reactions because excess Asc did not inhibit plasmid relaxation. Hydrogen peroxide was reported to be a product of the reaction of uranyl ion and Asc in the presence of oxygen at low pH (10). The possibility that H2O2 could be a reactive intermediate contributing to DNA strand break formation in the current system was explored by evaluating the effect of active and denatured catalase on plasmid relaxation. Reactions of equivalent amounts of UA and Asc at either 0.10 or 0.50 mM concentrations were carried out with DNA in the presence and absence of bovine catalase (Figure 4). Catalase induced 43 and 41% decreases in plasmid relaxation for the 0.10 and 0.50 mM reactions, respectively. Decreases of 13 and 7% were observed for corresponding reactions in the presence of denatured catalase; however, those values were not statistically significant from reactions without denatured catalase. Control reactions of iron(III) chloride with Asc and H2O2 at 0.050 mM concentrations showed a 29% decrease in plasmid relaxation in the presence of catalase that was lost upon denaturation of the catalase (Figure 4). These observations were consistent with the interpretation that at least some of the plasmid relaxation was due to generation of H2O2. It was also hypothesized that if reactive oxygen species were responsible for the induction of DNA strand breaks then the coincubation of UA and Asc with free radical scavengers should decrease plasmid relaxation. Reactions of 0.50 mM UA and 0.50 mM Asc were coincubated with plasmid DNA in the presence of 0.50 mM mannitol, sodium azide, or DMPO. The scavengers by themselves did not induce plasmid relaxation (Figure 5A, lanes 2-4, Figure 5B). The presence of the mannitol, azide, and DMPO produced slight decreases in the 45% strand breaks caused by reactions of uranium and Asc (Figure

Chem. Res. Toxicol., Vol. 16, No. 4, 2003 527

Figure 4. Effect of catalase (66 U/mL) on the reactions of UA (0.10 or 0.50 mM) with Asc (0.10 or 0.50 mM) in the presence of pBluescript SK+ plasmid (0.30 mM DNA-P) in 0.10 M TrisHCl (pH 7.0, 37 °C). Control reactions consisted of iron(III) chloride (0.050 mM) with Asc (0.050 mM) and H2O2 (0.050 mM). Reactions were incubated for 30 min prior to gel electrophoresis. (*) Values for plasmid relaxation for reactions in the presence of catalase were significantly different from values for reactions in the absence of catalase at p < 0.05. Differences between reactions in the presence and absence of denatured catalase were not statistically significant. Data represent mean ( SE for n ) 3-7 experiments.

5A, lanes 6-8 vs 5); however, these decreases were not statistically significant (Figure 5B). Mannitol reduced the plasmid relaxation caused by UA alone (Figure 5A, lane 10 vs 9, Figure 5B), presumably by coordinating to uranyl ion, whereas azide and DMPO had no significant effects (Figure 5A, lanes 11 and 12 vs 9, Figure 5B). The presence of mannitol, azide, and DMPO slightly reduced the strand breaks induced by Asc alone (Figure 5A, lanes 14-16 vs 13, Figure 5B); however, this decrease was also statistically insignificant. In conclusion, the radical scavengers mannitol, azide, and DMPO had no effect on plasmid relaxation induced by reactions of UA with Asc, further supporting the interpretation that reactive intermediates were not responsible for DNA strand breaks. The products of the reaction of UA with Asc were probed by 1H NMR spectroscopy. If Fenton type chemistry was occurring, then DHA should be a detectable product of the reaction between UA and Asc. It was previously shown that uranyl cation could catalytically oxidize Asc to DHA in the presence of dioxygen at pH e 2 (10). The extent to which this reaction could be observed by 1H NMR spectroscopy at neutral pH was measured in the current study. Reactions for 1H NMR spectroscopy were carried out at higher concentrations than reactions in the presence of DNA in order to increase detection levels. Reactions of 50 mM UA with 50 mM Asc in D2O showed a slight upfield shift in the acetate peak from 2.3 ppm in unreacted UA to 2.25 ppm for a 3 min reaction (Figure 6, C vs A). The spectrum was unchanged after 60 min, and after 1 week, the acetate resonance broadened slightly and shifted upfield to 2.05 ppm (data not shown). The Asc resonances at 4.40, 3.90, and 3.65 ppm, corresponding to the C5 and C6 protons, broadened and shifted to 4.8, 4.2, and 3.7 ppm, respectively, in the UAAsc reaction (Figure 6, B vs C). These results are consistent with formation of a uranium-Asc complex. Asc coordination most likely occurs through the ring hydroxyl groups on C2 and C3 because the extent of chemical shift between free Asc and reacted Asc increased C6 < C5 < C4 (Figure 6B,C). DHA was expected as a product in this reaction; however, the DHA resonances for C5 and C6 at 4.5 and 4.05-4.2 ppm, respectively, were not observed

528

Chem. Res. Toxicol., Vol. 16, No. 4, 2003

Figure 5. Effect of mannitol, sodium azide, and DMPO (0.50 mM each) on the reaction of UA (0.50 mM) with Asc (0.50 mM) in the presence of pBluescript SK+ plasmid (0.30 mM DNA-P) in 0.10 M Tris-HCl (pH 7.0, 37 °C). Reactions were incubated for 30 min prior to gel electrophoresis. (A) Representative gel: lane 1, untreated DNA; 2, mannitol; 3, sodium azide; 4, DMPO; 5, reaction of UA + Asc; 6, reaction + mannitol; 7, reaction + azide; 8, reaction + DMPO; 9, UA; 10, UA + mannitol; 11, UA + azide; 12, UA + DMPO; 13, Asc; 14, Asc + mannitol; 15, Asc + azide; 16, Asc + DMPO. (B) Quantitation of % plasmid relaxation relative to supercoiled DNA per lane. Data represent mean ( SE for n ) 4 experiments.

in the reaction spectrum (Figure 6, D vs C). The control reaction of 50 mM UA with 50 mM DHA showed a shift in the acetate peak from 2.3 ppm in unreacted UA to 2.13 ppm (Figure 6, E vs A); however, the DHA C5 and C6 resonances did not shift or broaden (Figure 6, D vs E), suggesting that there was no interaction between DHA and UA. Thus, the major product in the reaction of UA with Asc under these conditions appears to be a UAAsc complex; however, the stoichiometry cannot be determined from these data. These data are consistent with the interpretation that Fenton type chemistry is not a major reaction pathway in the development of DNA single strand breaks at neutral pH. The effect of excess Asc on the 1H NMR spectrum of the UA-Asc complex was also measured (Figure 7). UA (50 mM) was reacted with 0.5, 1, 2, and 5 equiv of Asc in D2O. Increasing equivalents of Asc served to sharpen the acetate peak and shift it upfield from 2.3 ppm in UA alone to 2.0 ppm in the 5:1 Asc:UA spectrum (Figure 7), which is consistent with the liberation of acetate upon coordination of Asc to uranyl cation. The three Asc resonances, 3.6-4.8 ppm, sharpened and shifted upfield

Yazzie et al.

Figure 6. 1H NMR spectroscopic analysis of the reaction of UA (50 mM) with either Asc (50 mM) or DHA (50 mM) in D2O. (A) UA; (B) Asc; (C) UA + Asc; (D) DHA; (E) UA + DHA. The peak at 3.2 ppm labeled with an asterisk was a solvent impurity used as a Y-scale internal standard.

with increasing Asc (Figure 7); however, a single set of resonances for these protons was observed in each spectrum, which is consistent with the rapid ligand exchange rate expected for uranyl cation.

Discussion Metals can cause DNA damage by two general mechanisms, indirectly by free radical generation (Fenton type chemistry) or through direct interactions. In the case of uranium, the free radical mechanism may be summarized by Scheme 1, in which Asc and dioxygen serve to promote catalytic cycling of uranium between U(VI) and U(IV) with the liberation of H2O2. Reaction of H2O2 with U(IV) may generate the DNA-damaging hydroxyl radical. This mechanism has been observed at low pH (10) but was below 1H NMR spectroscopy detection limits at neutral pH in the current study. A direct interaction for uranyl cation and DNA is outlined in Scheme 2. In this case, a uranyl-Asc complex may interact with the negatively charged DNA phosphate backbone. Withdrawal of electron density stabilizes the

DNA Strand Breaks by Uranyl Acetate and Ascorbate

Chem. Res. Toxicol., Vol. 16, No. 4, 2003 529 Scheme 2. Direct Mechanism for Uranium-Induced DNA Strand Breaks

Figure 7. 1H NMR spectroscopic analysis of the reaction of UA (50 mM) with varying Asc (0-250 mM) in D2O.

Scheme 1. Indirect Mechanism for Uranium-Induced DNA Strand Breaks

phosophodiester moiety toward nucleophilic attack by water or hydroxyl anion, resulting in DNA hydrolysis. This direct mechanism is best promoted through lanthanide metals (14); however, uranyl cation has been reported to promote hydrolysis of phosphodiesters in the presence of N-hexadecyl-N,N′,N′-trimethylethylenediamine (HTMED) at pH 5 (15). The results of the current study are not consistent with any one exclusive mechanism leading to uraniuminduced strand breaks. Catalase was able to reduce strand breaks by ∼40% (Figure 4), which is consistent

with generation of H2O2; however, neither radical scavengers (Figure 5) nor excess Asc (Figure 3) had any significant effects on strand breaks induced by uranium Asc reactions, and DHA, the expected reaction product, was not detectable by 1H NMR spectroscopy (Figure 6). Precipitation of reaction products in Tris buffer precluded the possibility of studying time-course reactions to assess the effect of reactive intermediates on plasmid relaxation. In light of the observed stability of the UAAsc complex, the possibility of a direct DNA interaction cannot be ruled out. The interaction of uranyl cation and Asc has not been completely characterized. A previous spectroscopic study reported observation of a series of monomeric and hydrolytic species at pH, with the major species proposed to be UO2RH+ and UO2(RH)2, where R ) Asc anion, at pH < 4.5 (16). The authors also suggested that Tris buffer competed with Asc for coordination to uranyl cation, a possibility that has not been ruled out in the current study. Further studies to characterize uranium-Asc complexes and the DNA products of these reactions are ongoing. There are a growing number of studies documenting the possible genotoxicity of uranium, although molecular mechanisms have not been proposed. The injection of uranyl fluoride enriched to 18.9% in U-235 produced chromosomal aberrations in the form of breaks, gaps, and polyploids in BALB/c mice (17). Cytogenic analysis of peripheral blood lymphocytes was reported for groups of smokers who were exposed or not exposed to uranyl compounds in a nuclear fuel manufacturing facility. Lymphocytes from the exposed smokers had the most chromosomal aberrations and also showed exchanges and ring chromosomes that were not observed in either nonexposed smokers or nonexposed nonsmokers (18). Neither of these two studies distinguished between the chemical and the radiological effects of uranium on chromosome damage. Three studies have tried to discriminate between the chemical and the radiological effects of uranium. Uranyl nitrate induced micronuclei, chromosomal aberrations, and sister chromatid exchange in Chinese hamster ovary cells (19). The authors attributed the cytogenic damage to uranium directly, rather than R-particle emission, because radioactivity was not detectable by Geiger or scintillation counting at the doses used. Another study addressed the genotoxicity of uranium directly by measuring the transformation of human osteosarcoma TE85 cells by a DU complex. Exposure to uranyl chloride produced a 9.6-fold increase in transformed cells and an

530

Chem. Res. Toxicol., Vol. 16, No. 4, 2003

increase in sister chromatid exchange relative to untreated cells (20). Using microdosimetric assessment and computer simulation, the authors calculated that only 0.0014% of the cell nuclei were hit by R-particles, thus ruling out the effects of radiation on the observed cellular damage. A recent study elegantly explored the effects of DU on formation of dicentric chromosomal aberrations and neoplastic transformation in human osteoblast cells (21). Using equal concentrations of uranyl nitrate sources with three different levels of radioactivity, Miller and coworkers demonstrated that neoplastic transformations were proportional to the radioactivity. However, the authors noted that chemical genotoxicity of uranyl cation was not ruled out. The results of the current work, showing strand breaks induced by uranyl cation in the presence of Asc, suggest that the chromosomal aberrations observed in prior reports (19, 21) could be at least partly nonradiological in origin. Although dicentric aberrations are consistent with R-irradiation, DNA strand break-inducing, or “radiomimetic” molecules such as bleomycin and pesticides can also induce dicentric aberrations (22, 23). In summary, reaction of UA with Asc was found to induce single strand breaks in plasmid DNA in vitro; in other words, uranium was found to be chemically genotoxic. DNA damage was observed in the absence of added hydrogen peroxide. The enhancement of uranium-induced strand breaks in the presence of Asc was not consistent with a radiological mechanism, although a radiological contribution is not ruled out. At this time, there are two possible molecular mechanisms that could result in a chemically induced strand break: a reductive Fenton type free radical mechanism or metal-catalyzed DNA hydrolysis through a uranyl-Asc complex. Further studies are in progress to characterize the DNA damage and to assess the relevance of these observations to in vivo systems. This study contributes to the growing number of data suggesting that future research should consider both chemical and radiological endpoints when assessing occupational and environmental risks from uranium exposures.

Acknowledgment. This work was supported by NCI Grant No. CA96302 (D.M.S.), the Arizona Board of Regents Biotechnology and Human Welfare Program (D.M.S.), and the NIH Minority Student Development Program, Grant No. GM56931 (M.Y., S.L.G.). We thank Professor James Cowart, Florida State University, for measurement of the UA isotope ratios.

References (1) Samet, J. M., Kutvirt, D. M., Waxweiler, R. J., and Key, C. R. (1984) Uranium mining and lung cancer in Navajo men. N. Engl. J. Med. 310, 1481-1484. (2) Jostes, R. F. (1996) Genetic, cytogenic, and carcinogenic effects of radon: a review. Mutat. Res. 340, 125-139.

Yazzie et al. (3) Priest, N. D. (2001) Toxicity of depleted uranium. Lancet 357, 244-246. (4) Royal Society (2002) The health effects of depleted uranium munitions: a summary. J. Radial. Prot. 22, 131-139. (5) Hahn, F. F., Guilmette, R. A., and Hoover, M. D. (2002) Implanted depleted uranium fragments cause soft tissue sarcomas in the muscles of rats. Environ. Health Perspect. 110, 51-59. (6) Shields, L. M., Wiese, W. H., Skipper, B. J., Charley, B., and Benally, L. (1992) Navajo birth outcomes in the Shiprock uranium mining area. Health Phys. 63, 542-551. (7) Mason, T. J., Fraumeni, J. F., and McKay, F. W. (1972) Uranium mill tailings and cancer mortality in Colorado. J. Natl. Cancer Inst. 49, 661-664. (8) Wilkinson, G. S. (1985) Gastric cancer in New Mexico counties with significant deposits of uranium. Arch. Environ. Health 40, 307-312. (9) Standeven, A. M., and Wetterhahn, K. E. (1991) Is there a role for reactive oxygen species in the mechanism of chromium(VI) carcinogenesis? Chem. Res. Toxicol. 4, 616-625. (10) Taqui Khan, M. M., and Martell, A. E. (1969) Kinetics of metal ion and metal chelate catalyzed oxidation of ascorbic acid. IV. Uranyl ion catalyzed oxidation. J. Am. Chem. Soc. 91, 4668-4672. (11) Hamilton, M. M., Ejnik J. W., and Carmichael, A. J. (1997) Uranium reactions with hydrogen peroxide studies by EPR-spin trapping with DMPO. J. Chem. Soc., Perkin Trans. 2, 2491-2494. (12) Miller, A. C., Stewart, M., Brooks, K., Shi, L., and Page, N. (2002) Depleted uranium-catalyzed oxidative DNA damage: absence of significant alpha particle decay. J. Inorg. Biochem. 91, 246-252. (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) Franklin, S. J. (2001) Lanthanide-mediated DNA hydrolysis. Curr. Opin. Chem. Biol. 5, 201-208. (15) Moss, R. A., Bracken, K., and Zhang, J. (1997) Actinide (uranyl) hydrolysis of phosphodiesters. Chem. Commun. 563-564. (16) Singh, H. B., Havel J., and Sommer, L. (1969) Studies on the reaction of uranium(VI) with ascorbic acid. Collect. Czech. Chem. Commun. 34, 3277-3296. (17) Hu, Q., and Shoupeng, Z. (1990) Induction of chromosomal aberrations in male mouse germ cells by uranyl fluoride containing enriched uranium. Mutat. Res. 244, 209-214. (18) Prabhavathi, P. A., Fatima, S. K., Rao, S., and Reddy, P. P. (2000) Analysis of chromosomal aberration frequencies in the peripheral blood lymphocytes of smokers exposed to uranyl compounds. Mutat. Res. 466, 37-41. (19) Lin, R. H., Wu, L. J., Lee, C. H., and Kin-Shiau, S. Y. (1993) Cytogenic toxicity of uranyl nitrate in Chinese hamster ovary cells. Mutat. Res. 319, 197-203. (20) Miller, A. C., Blakely, W. F., Livengood, D., Whittaker, T., Xu, J., Ejnik, J. W., Hamilton, M. M., Parlette, E., St. John, T., Gerstenberg, H. M., and Hsu, H. (1998) Transformation of human osteoblast cells to the tumorigenic phenotype by depleted uraniumuranyl chloride. Environ. Health Perspect. 106, 465-471. (21) Miller, A. C., Xu, J., Stewart, M., Brooks, K., Hodge, S., Shi, L., Page, N., and McClain, D. (2002) Observation of radiation-specific damage in human cells exposed to depleted uranium: dicentric frequency and neoplastic transformation as endpoints. Radiat. Prot. Dosim. 99, 275-278. (22) Bolzan, A. D., Bianchi, N. O., Larramendy, M. L., and Bianchi, M. S. (1992) Chromosomal sensitivity of human lymphocytes to bleomycin. Influence of antioxidant enzyme activities in whole blood and different blood fractions. Cancer Genet. Cytogenet. 64, 133-138. (23) Garaj-Vrhovac, V., and Zeljezic, D. (2002) Assessment of genome damage in a population of Croatian workers employed in pesticide production by chromosomal aberration analysis, micronucleus assay and comet assay. J. Appl. Toxicol. 22, 249-255.

TX025685Q