Comparison of the Formation of 8-Hydroxy-2 '-deoxyguanosine and

Imperial College School of Medicine at St. Mary's, Norfolk Place, London W2 1PG, U.K.. Received August 29, 1997. The formation of 8-hydroxydeoxyguanos...
0 downloads 0 Views 178KB Size
Download PDF
420

Chem. Res. Toxicol. 1998, 11, 420-427

Comparison of the Formation of 8-Hydroxy-2′-deoxyguanosine and Single- and Double-Strand Breaks in DNA Mediated by Fenton Reactions Daniel R. Lloyd,* Paul L. Carmichael,§ and David H. Phillips Section of Molecular Carcinogenesis, Institute of Cancer Research, Haddow Laboratories, Cotswold Road, Sutton, Surrey SM2 5NG, U.K., and Department of Pharmacology and Toxicology, Imperial College School of Medicine at St. Mary’s, Norfolk Place, London W2 1PG, U.K. Received August 29, 1997

The formation of 8-hydroxydeoxyguanosine (8-OHdG) and both single- and double-strand breaks in DNA by Fenton-type reactions has been investigated. Salmon sperm DNA was exposed to hydrogen peroxide (50 mM) and one of nine different transition-metal ions (25 µM-1 mM). Modified DNA was isolated and subjected to analysis by liquid chromatography coupled to an electrochemical detection system (LC-ECD), to evaluate the formation of 8-OHdG. The highest yield of 8-OHdG was obtained following treatment of DNA with the chromium(III) Fenton reaction (a maximum of 19 400/106 nucleotides), followed by iron(II) (13 600), vanadium(III) (5800), and copper(II) (5200). The chromium(VI) Fenton reaction generated a moderate yield of 8-OHdG (3600/106 nucleotides), while the yield obtained in DNA treated with cobalt(II), nickel(II), cadmium(II), and zinc Fenton reactions was not significantly higher than in control incubations of DNA with hydrogen peroxide alone. Similar treatment of the double-stranded plasmid pBluescript K+ with hydrogen peroxide (1 mM) and each transitionmetal ion (1-100 µM) followed by quantitative agarose gel electrophoresis demonstrated that open-circle DNA, resulting from single-strand breaks, was generated in Fenton reactions involving all nine metal ions. In contrast, linear DNA was only formed in Fenton reactions involving chromium(III), copper(II), iron(II), and vanadium(III) ions. Formation of linear DNA, under conditions that generated relatively few single-strand breaks, suggests that these four transition-metal ions partake in Fenton reactions to generate true double-strand breaks. Furthermore, the generation of 8-OHdG exhibits a good correlation with the formation of doublestrand breaks, suggesting that they arise by a similar mechanism.

Introduction The ubiquitous nature of free radicals in vivo has led to the hypothesis that they may be involved in the etiology of a number of human pathologies (1-5). Of particular relevance are reactive oxygen species (ROS) since they are generated as a function of normal metabolism (6) and have been shown to cause extensive damage to essential cellular components (3, 7, 8). The highly electron-dense structure of DNA makes it an extremely good target for attack by many ROS, but especially by the hydroxyl radical which is a highly transient, reactive, and electrophilic species (9). The ability of the hydroxyl radical to damage DNA has therefore been the focus of extensive investigation in recent years. The number of DNA lesions generated in vivo via oxidative processes is estimated to be in excess of 10 000/ cell/day (10). A wide spectrum of these oxidative DNA lesions has been identified following exposure of DNA to ROS both in vitro and in vivo. These include single- and double-stranded DNA breaks and base modifications such as thymine glycol, 8-hydroxy-2′-deoxyguanosine (8* To whom correspondence and requests for reprints should be addressed. Fax: +44 181 770 7290. E-mail: [email protected]. § Imperial College School of Medicine at St. Mary’s.

OHdG),1 8-hydroxy-2′-deoxyadenosine (8-OHdA), and 5-hydroxy-2′-deoxycytidine (5-OHdC) which have been demonstrated to be promutagenic lesions (11-14). In addition, lesions generated in DNA following hydroxyl radical attack have been detected by 32P-postlabeling (15, 16). These lesions were identified as intrastrand crosslinks between adjacent purine bases; specifically, dimers between adjacent adenines and between adenine and guanine residues were identified (16). More recently, a number of other intrastrand cross-linked species have been identified, along with base-deoxyribose cross-links (17). There are numerous mechanisms by which ROS might be generated in vivo. One such mechanism is thought to be the Fenton reaction (3), which involves the reduction of hydrogen peroxide by a transition-metal ion to form the hydroxyl radical (18). Several transition-metal ions are able to partake in Fenton reactions to generate oxidative DNA lesions. For example, the formation of strand breaks and 8-OHdG has been demonstrated in DNA treated with hydrogen peroxide and iron or copper 1 Abbreviations: 8-OHdG, 8-hydroxy-2′-deoxyguanosine; LC-ECD, liquid chromatography-electrochemical detection; 8-OHdA, 8-hydroxy2′-deoxyadenosine; 5-OHdC, 5-hydroxy-2′-deoxycytidine; SK+, pBluescript K+ plasmid DNA.

S0893-228x(97)00156-2 CCC: $15.00 © 1998 American Chemical Society Published on Web 04/17/1998

Oxidative DNA Damage Mediated by Fenton Reactions

ions (19). In addition, occupational exposure to transition-metals such as nickel and chromium has been shown to increase incidence of tumors in the respiratory tract (20); this has led to speculation that the carcinogenic effects of these metals might arise via a genotoxic mechanism involving the formation of ROS and subsequent generation of DNA lesions (21, 22). Indeed, there is growing evidence that transition-metal ions are able to interact with and alter genetic material via an oxidative mechanism involving the generation of ROS (3). Although it is well-established that ROS, and especially Fenton reactions, can cause a plethora of DNA lesions, it is unclear to what extent the lesions are formed by a common mechanism. Recently we reported the formation of putative intrastrand cross-links between adjacent bases in DNA following in vitro incubation with hydrogen peroxide and six different transition-metal ions in Fenton-type reactions and compared the formation of these lesions with the concomitant formation of DNA strand breaks (23). In that study we demonstrated that the generation of the intrastrand cross-links was via a mechanism independent of the formation of strand breaks. However the type of strand breaks formed in these Fenton reactions (i.e., single vs double) was not fully distinguished. In the present study we have expanded upon our earlier work and compared the formation of 8-OHdG with both single- and double-strand breaks in DNA following treatment in vitro in Fentontype reactions involving nine different transition-metal ions.

Experimental Section Materials. XL-1 Blue Escherichia coli and pBluescript K+ (SK+) plasmid DNA were obtained from Stratagene (Cambridge, U.K.). Salmon sperm DNA, alkaline phosphatase, and lysozyme (from chicken egg white) were obtained from Sigma Chemical Co. (Poole, Dorset, U.K.). Snake venom phosphodiesterase and deoxyribonuclease 1 (from bovine pancreas, grade II) were obtained from Boehringer Mannheim (Lewes, East Sussex, U.K.). Chemicals used in the Fenton-type treatment of DNA were obtained from previously mentioned sources (23). Copper(II) sulfate, sodium chromate(VI), cobalt(II) sulfate, iron(II) sulfate, nickel(II) sulfate, vanadium(III) chloride, cadmium(II) chloride, chromium(III) chloride, and zinc(II) sulfate, all analytical grade and over 97% pure, were freshly made up as stock solutions in deionized water under ambient conditions and used within approximately 15 min. Isolation of DNA. Commercially obtained salmon sperm DNA was purified further by phenol/chloroform extraction (24) and dissolved in deionized water. SK+ plasmid DNA was amplified in bacteria according to a method previously described (25). XL-1 Blue E. coli were inoculated with a solution of SK+, plated onto ampicillincontaining agar, and incubated at 37 °C for 18 h. Transformed colonies of cells were grown in L-broth (500 mL) for 18 h, and SK+ DNA was subsequently isolated using the alkali lysis method of Maniatis (25). Cell suspensions were harvested and resuspended in buffer (50 mM glucose, 25 mM Tris base, 10 mM EDTA, pH 8) containing lysozyme (5 mg/mL). Cells were then lysed by addition of an alkaline solution (200 mM NaOH, 1% sodium dodecyl sulfate, 20 mL), and the resulting solution was left to stand on ice for 10 min. An ice-cold solution of potassium acetate (5 M, pH 4.8, 15 mL) was then added, and the bacterial DNA was precipitated by leaving on ice for 10 min and subsequently pelleted by centrifugation. 2-Propanol (0.6 vol) was then added to the supernatant, and the SK+ DNA was precipitated by leaving at room temperature for 15 min. DNA

Chem. Res. Toxicol., Vol. 11, No. 5, 1998 421 was recovered by centrifugation, washed in 70% ethanol, and resuspended in Tris-EDTA buffer (10 mM Tris base, 1 mM EDTA, 8 mL). Supercoiled DNA was obtained by centrifugation in a cesium chloride density gradient. Cesium chloride (8 g) and ethidium bromide (8 mg) were added to the SK+ DNA solution, and the resulting solution was centrifuged (55 000 rpm, 18 h, 18 °C). The visible band of supercoiled plasmid was harvested, and ethidium bromide was removed by repeated extraction with water-saturated butanol. The resulting DNA solution was then extensively dialyzed against progressively weaker solutions of EDTA and finally against distilled water. Treatment of Salmon Sperm DNA. Salmon sperm DNA was treated with Fenton reagents as previously described (23); briefly, aliquots (500 µL, 0.5 mg/mL) were incubated with hydrogen peroxide (50 mM) and each transition-metal ion (25 µM-1 mM) for 1 h at 37 °C. Fenton reactions were terminated using immobilized catalase (100 µL) which was removed by centrifugation after 15 min at ambient temperature. DNA treated with Cr(VI), Co(II), Ni(II), Cd(II), and Zn(II) Fenton reactions was precipitated from the supernatant with sodium chloride and ethanol. In all other cases where DNA could not be precipitated due to extensive strand breakage (23), transitionmetal ions were removed from the supernatant by chelation with immobilized iminodiacetic acid (100 µL) for 15 min followed by centrifugation as before. LC-ECD Analysis of Salmon Sperm DNA. Analysis of 8-OHdG was carried out as described previously (26). DNA samples (0.5 mg in 900 µL of water, pH 7.0) were digested overnight with deoxyribonuclease I (150 µg) followed by a 6-h incubation with snake venom phosphodiesterase (12 µg, pH 9.0). The resulting nucleotides were converted to nucleosides with alkaline phosphatase (1.2 units). Hydrolysate (equivalent to 100 µg of DNA) was applied to a Nucleosil ODS column and eluted with ammonium acetate (50 mM) and acetic acid (50 mM) in 5% methanol at a flow rate of 1.0 mL/min. 8-OHdG was detected using an ESA Coulochem II electrochemical detector and eluted after 29-32 min. Unmodified nucleosides were detected by UV absorption using a Pharmacia UV-1 detector. The amount of 8-OHdG present in the DNA samples was calculated by measuring the area of the peaks obtained from both electrochemical and UV traces and comparing those obtained from DNA samples with those obtained from synthetic standards. Treatment and Analysis of pBluescript K+ DNA. Aliquots of plasmid DNA (10 µL, 10 ng/µL) were incubated with hydrogen peroxide (1 mM) and each transition-metal ion (1100 µM) for 15 min at ambient temperature. Samples were then immediately loaded, without further treatment or purification, onto a 1% agarose gel containing ethidium bromide as previously described (23). Following electrophoresis (3 h, 80 V) in Trisacetate buffer, gels were viewed under UV light and photographed using a UVP Imagestore 5000 video camera (Ultraviolet Products), and images were stored in TIFF format on floppy disk. Densitometry analysis of individual DNA bands was achieved using ImageQuant software.

Results Formation of 8-OHdG in Salmon Sperm DNA. Analysis of hydrolysates of salmon sperm DNA by LCECD gave rise to a chromatogram containing three major peaks detected by UV absorbance and one major peak detected by electrochemical oxidation. A typical chromatogram is represented in Figure 1. Coelution with synthetic standards demonstrates that peak A, with a retention time of approximately 9 min, is due to absorbance of deoxycytidine, while peak C (retention time 53 min) is due to absorbance of deoxyadenosine. Peak B (retention time 22 min) corresponds to deoxyguanosine and thymidine, which coelute in this chromatography

422 Chem. Res. Toxicol., Vol. 11, No. 5, 1998

Figure 1. LC-ECD chromatogram of a hydrolysate of salmon sperm DNA treated with a copper(II) Fenton reaction. The upper trace represents electrochemical detection, and the peak eluting at 32 min corresponds to the electrochemical oxidation of 8-OHdG. The lower trace represents UV absorbance, and the products eluting at 9, 22, and 53 min are, respectively, deoxycytidine, deoxyguanosine and thymidine (which coelute at 22 min), and deoxyadenosine.

system. Peak D (retention time 32 min) is due to electrochemical oxidation of 8-OHdG. Data from these chromatograms demonstrated that the generation of 8-OHdG in DNA treated in Fenton reactions was dependent upon the transition-metal ion used and on its concentration. The highest yield of 8-OHdG was generated in the chromium(III) Fenton reaction (Figure 2H). A steady increase in the yield of 8-OHdG was observed in this reaction as the concentration of chromium(III) ions was increased from 25 µM to 1 mM, where a maximum yield of 19 400/106 nucleotides was observed. Similarly, the generation of 8-OHdG in the iron(II) Fenton reaction increased and reached a maximum level of 13 600/106 nucleotides at 800 µM iron(II) sulfate (Figure 2D). However, the formation of 8-OHdG in the copper(II) Fenton reaction remained relatively constant as the concentration of copper(II) sulfate was increased; the maximum yield obtained was 5200/106 nucleotides (Figure 2A). The amount of 8-OHdG formed in DNA treated with the vanadium(III) Fenton reaction reached a maximum yield at 75 µM (5800/106 nucleotides) and declined rapidly at higher concentrations of vanadium(III) chloride (Figure 2F). Incubation of DNA with a chromium(VI) Fenton reaction generated a moderate yield of 8-OHdG, reaching a maximum of 3600/106 nucleotides (Figure 2B). Treatment of DNA in Fenton reactions involving cobalt(II) (Figure 2C), nickel(II) (Figure 2E), cadmium(II) (Figure 2G), and zinc(II) ions (Figure 2I) did not generate levels of 8-OHdG that were significantly higher than the yield found in DNA treated with hydrogen peroxide alone (900/106 nucleotides), although a small increase was observed at low concentrations of nickel(II) and zinc(II) (Figure 2E,I). Treatment of DNA with each transition-metal ion in the absence of hydrogen peroxide resulted in a level of between 10 and 250 8-OHdG/106 nucleotides. These values were, in some cases, higher than in untreated DNA (15/106 nucleotides) but were significantly less than in treatments with hydrogen peroxide alone. Control experiments were carried out to determine whether residual transition-metal ions in the DNA samples affected the detection of 8-OHdG by inhibiting DNA digestion. Each of the nine transition-metal ions was added to untreated DNA samples to a concentration of 1 mM, and the DNA was then purified by either ethanol precipitation or treatment with immobilized iminodiacetic acid, depending on which method had been used in the corresponding Fenton incubations. The DNA

Lloyd et al.

was then subjected to the digestion conditions above. The efficiency of DNA digestion was measured by comparing peak areas of unmodified nucleosides from samples containing 1 mM transition-metal ions with samples that had not been further treated with transition-metal ions. The peak areas remained very constant (81-94% efficient digestion compared to untreated DNA) suggesting that detection of 8-OHdG was not significantly affected by residual transition-metal ions in the DNA samples or the extraction procedure used to purify the DNA. An exception was chromium(III) which reduced the efficiency of DNA digestion to 67%. However, this transition-metal ion gave the highest yield of 8-OHdG and the most pronounced dose response; hence, the less efficient digestion does not appear to have affected the results significantly. Generation of Single- and Double-Strand Breaks. In its native, undamaged state, the double-stranded SK+ plasmid exists in a tightly compact “supercoiled” conformation, and as such it has a relatively high electrophoretic mobility. Upon formation of one or more singlestrand breaks, the supercoiled tertiary structure is disrupted resulting in an open-circle conformation and a reduced electrophoretic mobility in agarose. Linear DNA, formed either by double-strand breaks or closely opposed single-strand breaks, has a mobility intermediate between that of the supercoiled and open-circular conformations of plasmid DNA. Hence analysis by agarose gel electrophoresis yields information regarding the type of strand breakage occurring in each Fenton reaction by separation of the different conformations of plasmid DNA (Figure 3). Incubations of SK+ DNA with Fenton reactions demonstrated that the open-circle conformation was generated in reactions involving all nine transition-metal ions tested. In five out of nine cases [Fenton reactions involving chromium(VI), cobalt(II), nickel(II), cadmium(II) and zinc(II)] the yield of open-circle DNA reached near maximum at 50 µM and remained relatively constant at higher metal ion concentrations (Figure 4). The chromium(VI) Fenton reaction, for example, formed increasing amounts of open-circular DNA with increasing metal ion concentration, and this was accompanied by a concomitant and proportional decrease in remaining supercoiled DNA; however, linear DNA was not formed in this reaction. The cobalt(II) Fenton reaction gave a similar distribution of reaction products, with the yield of opencircular DNA reaching a near-maximum level of 57% with 20 µM cobalt sulfate and remaining very constant as the concentration was increased; once again, linear DNA was not formed (Figure 4C). Similarly, in the nickel(II) (Figure 4E), cadmium(II) (Figure 4G), and the zinc(II) (Figure 4I) Fenton reactions, the yield of opencircular DNA increased with increasing concentration of each transition-metal ion up to 50 µM, reaching maxima at 79%, 70%, and 69%, respectively, while a proportional decrease was observed in remaining supercoiled DNA and the yield of linear DNA remained at near-zero levels. In contrast, the other four Fenton reactions [involving copper(II), iron(II), vanadium(III) and chromium(III), ions] generated significant yields of linear DNA and different reaction profiles. For example, the generation of open-circle DNA in the copper(II) Fenton reaction (Figure 4A) reached a maximum level of 27% at a concentration of 20 µM copper(II) sulfate, at which point the yield of linear DNA accounted for 38% of the total.

Oxidative DNA Damage Mediated by Fenton Reactions

Chem. Res. Toxicol., Vol. 11, No. 5, 1998 423

Figure 2. Effect of transition-metal ion concentration on the formation of 8-OHdG in Fenton reactions. Salmon sperm DNA was exposed to 50 mM hydrogen peroxide and between 25 and 1000 µM transition-metal ion as follows: A, copper(II) sulfate; B, sodium chromate(VI); C, cobalt(II) sulfate; D, iron(II) sulfate; E, nickel(II) sulfate; F, vanadium(III) chloride; G, cadmium(II) chloride; H, chromic(III) chloride; I, zinc(II) chloride. Data points represent the mean ( SEM for duplicate determinations of duplicate samples. Where error bars are not apparent, they are within the range obscured by the data point.

rapid decline in remaining supercoiled DNA; under these reaction conditions, the generation of linear DNA reached a level of 21%, and the yield increased at higher concentrations of iron(II) sulfate at the expense of the recovery of open-circular DNA (Figure 4D). The vanadium(III) (Figure 4F) and chromium(III) (Figure 4H) Fenton reactions also generated high levels of linear DNA, while the yield of open-circular DNA reached maxima at 67% and 23%, respectively, of the total DNA and declined as the metal ion concentration increased. Figure 3. Agarose gel electrophoresis of SK+ plasmid DNA following treatment with an iron(II) Fenton reaction. Lane 1 shows untreated SK+ DNA, while lane 2 shows DNA treated with 1 mM hydrogen peroxide only. Lane 3 represents DNA treated with the restriction enzyme BamHI which converts SK+ DNA selectively to the linear conformation. Remaining lanes (4-8) represent SK+ treated with 1 mM hydrogen peroxide and 1, 2, 5, 10, and 20 µM iron(II) sulfate, respectively.

The yield of linear DNA increased further at higher concentrations of this transition-metal ion. Similarly, at a relatively low concentration of iron(II) sulfate (10 µM), the generation of open-circle DNA reached a maximum level at 63% of total DNA and was accompanied by a

Incubation of SK+ DNA with each transition-metal ion (100 µM, the highest concentration used in these experiments), in the absence of hydrogen peroxide, demonstrated that in all but two cases, the generation of strand breaks did not occur in these incubations. The exceptions were the iron(II)- and vanadium(III)-treated DNA, where strand breaks were demonstrated by the generation of 50% and 67% open-circle DNA and 45% and 1% linear DNA, respectively. The levels of nonsupercoiled DNA observed in these control samples are much less than those observed in the presence of hydrogen peroxide since in both cases the proportion of linear and open-circular

424 Chem. Res. Toxicol., Vol. 11, No. 5, 1998

Lloyd et al.

Figure 4. Effect of transition-metal ion concentration on the conversion of supercoiled SK+ plasmid to open-circular and linear conformations in Fenton reactions. SK+ DNA was exposed to 1 mM hydrogen peroxide and between 1 and 100 µM transition-metal ion as follows: A, copper(II) sulfate; B, sodium chromate(VI); C, cobalt(II) sulfate; D, iron(II) sulfate; E, nickel(II) sulfate; F, vanadium(III) chloride; G, cadmium(II) chloride; H, chromic(III) chloride; I, zinc(II) chloride. Data points represent the mean ( SEM for the analyses of triplicate samples. Where error bars are not apparent, they are within the range obscured by the data point.

DNA reaches much higher levels, even when using lower concentrations of iron(II) or vanadium(III) ions. Indeed, at concentrations of 50 µM and above of both these transition-metal ions and in the presence of hydrogen peroxide, not only is the plasmid converted to linear DNA, visualized by a discrete band on the agarose gel, but the extent of strand breakage is such that the plasmid is broken down into smaller fragments and appears on the gel as a diffuse smear of DNA (Figure 3, lanes 7 and 8).

Discussion In the present study we have demonstrated the generation of 8-OHdG in salmon sperm DNA and single- and double-strand breaks in SK+ DNA exposed to Fenton reactions involving nine different transition-metal ions. The yield of all three types of oxidative damage was found to be dependent upon the transition-metal ion used in the Fenton reaction and the concentration of that metal ion. To our knowledge, the relationship between the formation of these types of oxidative damage mediated by the involvement of a number of different transition-metal ions has not been reported. However, we have previously investigated the ability of the same transition-metal ions to partake in Fenton reactions and form strand breaks

and putative intrastrand cross-links in salmon sperm DNA (23). In that study, we demonstrated extensive degradation of salmon sperm DNA in Fenton reactions involving copper(II), iron(II), vanadium(III), and chromium(III) ions (23). However, the method of analysis used was not sufficiently sensitive to distinguish between the formation of single- and double-strand breaks. In the present study we have extended upon these earlier observations by using the plasmid relaxation assay, a very sensitive method for detecting the formation of DNA strand breaks in vitro. In this assay, the formation of open-circle DNA can only arise via single-strand breakage. Linear DNA could result from the formation of a true double-strand break (i.e., scission of both strands of DNA in one single event). However, it might also be due to the formation of many single-strand breaks such that two are formed independently but closely on opposing strands. In the copper(II), iron(II), vanadium(III), and chromium(III) Fenton reactions (those that converted supercoiled plasmid DNA to the linear conformation) the yield of linear DNA reached significant levels (8-38%), while at the same concentration of transition-metal ions the yield of open-circle DNA reached a maximum (1667%). In contrast, the other Fenton reactions generated higher yields of open-circle DNA (up to 79%) without forming linear DNA. Hence, if the formation of linear

Oxidative DNA Damage Mediated by Fenton Reactions

Chem. Res. Toxicol., Vol. 11, No. 5, 1998 425

Table 1. Comparison of the Formation of Putative Intrastrand Cross-Links, 8-OHdG, and Single- and Double-Strand Breaks in Fenton-Treated DNA

a

transitionmetal ion

generation of putative intrastrand cross-linksa

generation of 8-OHdG

salmon sperm DNA degradation (total strand breaks)a

proportion of single- vs double-strand breaks

copper(II) chromium(VI) cobalt(II) iron(II) nickel(II) vanadium(III) cadmium(II) chromium(III) zinc(II)

+++ + ++ + + + -

++ + +++ ++ +++ -

+++ + +++ +++ + +++ -

double > single single only single only double . single single only double . single single only double . single single only

Data from Lloyd et al. (23).

DNA was due solely to the formation of closely opposed single-strand breaks, one would expect linear DNA to be formed in those Fenton reactions where the yield of opencircle DNA exceeded the maximum yield obtained in reactions involving the metal ions mentioned above. Hence the formation of linear DNA in these reactions must be due, at least in part, to the formation of true double-strand breaks. A further extension to our earlier work was to investigate the formation of 8-OHdG in Fenton-type reactions. 8-OHdG is a very commonly measured oxidative base lesion, and a number of methods have been developed to analyze its content in DNA following exposure to ROS both in vivo and in vitro, such as the LC-ECD method employed here and analysis by 32P-postlabeling (27). The relative ease of its detection has thereby helped to establish 8-OHdG as a “benchmark” biomarker for in vivo oxidative damage. The LC-ECD system used in the present study was sufficiently sensitive to determine the 8-OHdG content in the Fenton-treated salmon sperm DNA samples. However, it was a less sensitive assay than the plasmid relaxation assay used to measure strand break formation, and as such the Fenton conditions used to generate 8-OHdG in salmon sperm DNA were more vigorous. Hence the concentrations of Fenton reagents used in the plasmid relaxation assay were significantly lower, which allowed a more detailed analysis of plasmid DNA treated with the lowest concentrations of metal ions. Indeed, it is evident from the graphs in Figure 3 that at concentrations of 20-50 µM and above, the formation of strand breaks reached a maximum and remained at that level as the metal ion concentration increases. In contrast, at the same concentration of metal ions in the LC-ECD analysis of salmon sperm DNA, the yield of 8-OHdG continued to increase. It was preferable to use 1 mM rather than 50 mM hydrogen peroxide in the treatment of SK+ plasmid DNA since the latter caused a detectable level of strand breakage in the absence of transition-metal ions. However, to investigate whether a higher concentration of hydrogen peroxide could have reduced the formation of oxidative DNA damage, by a mechanism involving scavenging of hydroxyl radicals (28, 29), an iron(II) Fenton reaction involving 50 mM rather than 1 mM hydrogen peroxide was carried out, and the resulting strand breaks in SK+ DNA were analyzed by electrophoresis as before. No significant difference was observed upon increasing the hydrogen peroxide concentration in this manner (data not shown). It may be that this similarity is because the concentration of metal ions used in each Fenton reaction was varied accordingly such that the hydrogen peroxide,

whether at 1 or 50 mM, was in comparable excess in both cases. An interesting observation from the LC-ECD analysis was that in some cases the yield of 8-OHdG declined dramatically at higher concentrations of transition-metal ions. This is particularly evident in the Fenton reactions involving higher concentrations of iron(II) and vanadium(III) ions; chromatograms of these DNA samples exhibited greatly reduced peak areas corresponding to both 8-OHdG and unmodified nucleosides (data not shown), suggesting that the DNA degradation was so extensive that individual nucleotides were being degraded and therefore losing their UV activity. Control experiments described in the Results section, where we found that residual transition-metal ions remaining in the DNA samples following Fenton treatment did not affect the DNA digestion, cannot account for the apparent lack of unmodified nucleosides. Strand breakage alone cannot account for this observation since the process does not involve breaking down deoxyribose. This implies that there are other active mechanisms occurring under these vigorous Fenton conditions that are able to degrade DNA to such a large extent. Indeed, a similar observation was reported in our previous study, where we found that the yield of putative intrastrand cross-links decreased at higher concentrations of copper(II), iron(II), and vanadium(III) ions; this reduced yield was attributed to the excessive formation of strand breaks, effectively preventing the formation of cross-links between adjacent bases. A summary of results obtained in this study is shown in Table 1. The formation of 8-OHdG in Fenton-treated salmon sperm DNA in the present study correlates well with the generation of total strand breaks under the same conditions reported in our previous study. Those transition-metal ions that took part in Fenton reactions to generate the most extensive degradation of salmon sperm DNA, i.e., copper(II), iron(II), vanadium(III), and chromium(III), were also the ones that generated the highest yield of 8-OHdG. Results obtained in the present study, involving a more sensitive method of detecting strand breaks, demonstrate that the formation of 8-OHdG correlates more specifically with the generation of doublerather than single-strand breaks. This is in contrast to another report which demonstrates good agreement between the formation of 8-OHdG and single-strand breaks when copper(II)/iron(II) ions were incubated with different concentrations of hydrogen peroxide (30). However, a recent study, comparing the formation of 8-OHdG and single-strand breaks in DNA treated with either γ-irradiation, Fe(II)-EDTA/H2O2, Cu(II)/H2O2, or peroxynitrite, demonstrated different ratios of 8-OHdG:strand

426 Chem. Res. Toxicol., Vol. 11, No. 5, 1998

break formation depending on the oxidizing agent used (31). This suggested differences in the mechanism forming either type of oxidative DNA damage, possibly due to variation in the spectrum of DNA-damaging agents generated by each oxidizing agent and/or differences in their sequence specificity (31). This is in better agreement with our current data, where the lack of correlation between the formation of 8-OHdG and single-strand breaks suggests that they are formed by independent mechanisms. However, a number of different oxidizing systems were used in that study, and these systems might be expected to differ in the types of ROS generated. In this and our earlier study (23) we used the same ROSgenerating reaction, differing only in the transition-metal ion used to reduce hydrogen peroxide. Therefore we suggest that the differences observed in the formation of 8-OHdG relative to strand breaks are due to the binding specificity of each transition-metal ion. Indeed, it is well-known that transition-metal ions are sequencespecific in their binding to nucleic acids and differ in their ability to complex with different nucleotide moieties (32, 33). This hypothesis supports the “site-specific” mechanism proposed by Chevion (34), which may account for the differences observed in oxidative damage generated by different transition-metal ions. This mechanism involves the binding of a particular metal ion to DNA and subsequent cyclic reduction and reoxidation of that metal ion. This would allow repeated reactions with hydrogen peroxide and hence a site-specific generation of several hydroxyl radicals to cause scission of both DNA strands. The correlation observed between the formation of doublestrand breaks and 8-OHdG suggests that they might both be formed via this site-specific mechanism. In contrast, those transition-metal ions that generated only singlestrand breaks might act in a “random-hit” mechanism, similar to the mode of action of ionizing radiation (34). Metal ions that are poor DNA-binders, such as nickel(II) and chromium(VI), would be less likely to generate high concentrations of hydroxyl radicals directly adjacent to the DNA and more likely to result in the random distribution of hydroxyl radicals in solution via a free metal ion-mediated Fenton reaction. While some authors have suggested that the concentrations of Fenton reagents present in tissues are so low as to render the reaction too slow to be biologically significant (35, 36), there are numerous mechanisms by which high levels of transition-metal ions can enter the body (for example, following occupational exposure to airborne metal particles) and subsequently invade cells (21, 37, 38) leading to an oxidative burst and the production of high levels of hydrogen peroxide (21). Therefore we suggest that the experimental conditions used in the present study may indeed reflect the concentrations present locally in some cells, allowing Fenton reaction-mediated oxidative DNA damage to occur. Results from our earlier study, which showed that the formation of intrastrand cross-links occurred by a mechanism independent to the degradation of DNA caused by strand breaks, complement our present data (23). The strand breakage demonstrated in salmon sperm DNA in that study correlated well with the formation of doublestrand breaks and 8-OHdG in the present study, while it appears that the putative intrastrand cross-links are generated via a mechanism independent of the other types of oxidative damage reported in either study (Table

Lloyd et al.

1). This interesting observation sheds light on an earlier study from our laboratory concerning DNA damage in the livers of patients suffering from Wilson’s disease and primary hemochromatosis (26); consequences of these diseases include inefficient excretion of copper and iron, respectively, and a resultant accumulation of these metals in the liver. DNA lesions with similar chromatographic mobility to the putative intrastrand cross-links reported earlier were present in hepatic DNA from these patients, while the levels of 8-OHdG, determined using the same method of analysis as was used here, were no higher than in age-matched controls (26). The apparent difference in the mechanisms of formation of these types of DNA damage is in good agreement with the combined data from this and our earlier study. Although HPLC analysis revealed that the bulky lesions identified by postlabeling were not the same as the adenine-adenine or adenine-guanine dimers identified previously (16), the possibility remains that they are structurally related to some of the other DNA lesions either demonstrated in our study (23) or identified by Randerath et al. (17). In summary, we have demonstrated the formation of 8-OHdG and single- and double-strand breaks in DNA following treatment with a number of transition-metal ions in Fenton-type reactions and found that, while the formation of 8-OHdG correlates with the concomitant formation of double-strand breaks, the generation of single-strand breaks correlated with neither of these types of DNA damage. These differences in product distribution may be due to a site-specific versus randomhit mechanism depending upon how each transitionmetal ion associates with DNA. Together with our previous studies (16, 23, 26), and those of others (15, 17, 39-41) which reported the formation, both in vitro and in vivo, of bulky oxidative lesions (putative intrastrand cross-links), the present study contributes to increasing the understanding of the complex fundamental mechanisms by which these and other oxidative lesions are formed. Further studies investigating the formation and characterization of this important endogenous source of DNA damage are continuing.

Acknowledgment. D. R. Lloyd gratefully acknowledges receipt of a studentship from the Institute of Cancer Research.

References (1) Feig, D. I., Reid, T. M., and Loeb, L. A. (1994) Reactive oxygen species in tumorigenesis. Cancer Res. 54, 1890s-1894s. (2) Gutteridge, J. M. (1992) Aging and free radicals. Med. Lab. Sci. 49, 313-318. (3) Halliwell, B., and Gutteridge, J. M. C. (1985) The importance of free radicals and catalytic metal ions in human disease. Mol. Aspects Med. 8, 89-193. (4) Miesel, R., Zuber, M., Sanocka, D., Graetz, R., and Kroeger, H. (1994) Effects of allopurinol on in vivo suppression of arthritis in mice and ex vivo modulation of phagocytic production of oxygen radicals in whole human blood. Inflammation 18, 597-612. (5) Varma, S. D., Devamanoharan, P. S., and Morris, S. M. (1995) Prevention of cataracts by nutritional and metabolic antioxidants. Crit. Rev. Food Sci. Nutr. 35, 111-129. (6) Gutteridge, J. M. (1994) Biological origin of free radicals, and mechanisms of antioxidant protection. Chem.-Biol. Interact. 91, 133-140. (7) Cheeseman, K. H., Beavis, A., and Esterbauer, H. (1988) Hydroxyl radical-induced iron-catalysed degradation of 2-deoxyribose. Biochem. J. 252, 649-653. (8) Carmichael, P. L., and Hipkiss, A. R. (1991) Differences in susceptibility between crystallins and nonlenticular proteins to

Oxidative DNA Damage Mediated by Fenton Reactions

(9) (10) (11)

(12)

(13)

(14)

(15)

(16)

(17)

(18) (19) (20) (21) (22) (23)

(24) (25)

copper and hydrogen peroxide-mediated peptide bond cleavage. Free Radical Res. Commun. 15, 101-110. Pryor, W. A. (1988) Why is the hydroxyl radical the only radical that commonly adds to DNA? Free Radical Biol. Med. 4, 219223. Ames, B. N., and Saul, R. L. (1987) Oxidative DNA damage, cancer and aging. Ann. Intern. Med. 107, 526-545. Basu, A. K., Loechler, E. L., Leadon, S. A., and Essigmann, J. M. (1989) Genetic effects of thymine glycol: site-specific mutagenesis and molecular modeling studies. Proc. Natl. Acad. Sci. U.S.A. 86, 7677-7681. Cheng, K. C., Cahill, D. S., Kasai, H., Nishimura, S., and Loeb, L. A. (1992) 8-hydroxyguanine, an abundant form of oxidative DNA damage, causes G-T and A-C substitutions. J. Biol. Chem. 267, 166-172. Wood, M. L., Esteve, A., Morningstar, M. L., Kuziemko, G. M., and Essigmann, J. M. (1992) Genetic effects of oxidative DNA damage: comparative mutagenesis of 7,8-dihydro-8-oxoguanine and 7,8-dihydro-8-oxoadenine in Escherichia coli. Nucleic Acids Res. 20, 6023-6032. Feig, D. I., Sowers, L. C., and Loeb, L. A. (1994) Reverse chemical mutagenesis: identification of the mutagenic lesion resulting from reactive oxygen species-mediated damage to DNA. Proc. Natl. Acad. Sci. U.S.A. 91, 6609-6613. Randerath, K., Yang, P.-F., Danna, T. F., Reddy, R., Watson, W. P., and Randerath, E. (1991) Bulky adducts detected by 32Ppostlabeling in DNA modified by oxidative damage in vitro. Comparison with rat lung I-compounds. Mutat. Res. 250, 135144. Carmichael, P. L., Ni She´, M., and Phillips, D. H. (1992) Detection and characterization by 32P-postlabeling of DNA adducts induced by a Fenton-type oxygen radical-generating system. Carcinogenesis 13, 1127-1135. Randerath, K., Randerath, E., Smith, C. V., and Chang, J. (1996) Structural origins of bulky oxidative DNA adducts (type II I-compounds) as deduced by oxidation of oligonucleotides of known sequence. Chem. Res. Toxicol. 9, 247-254. Fenton, H. J. H. (1894) Oxidation of tartaric acid in the presence of iron. J. Chem. Soc. 899-910. Tachon, P. (1989) Ferric and cupric ions requirement for DNA single-strand breakage by hydrogen peroxide. Free Radical Res. Commun. 7, 1-10. IARC. (1990) IARC Monographs on the Evaluation of Carcinogenic Risks to Humans: Vol. 49, Chromium, Nickel and Welding, IARC, Lyon, France. Costa, M. (1991) Molecular mechanisms of nickel carcinogenesis. Annu. Rev. Pharmacol. Toxicol. 31, 321-337. 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. Lloyd, D. R., Phillips, D. H., and Carmichael, P. L. (1997) Generation of putative intrastrand cross-links and strand breaks in DNA by transition metal ion-mediated oxygen radical attack. Chem. Res. Toxicol. 10, 393-400. Gupta, R. C. (1984) Nonrandom binding of the carcinogen N-hydroxy-2-acetylaminofluorene to repetitive sequences of rat liver DNA in vivo. Proc. Natl. Acad. Sci. U.S.A. 81, 6943-6947. Maniatis, T., Fritsch, E. F., and Sambrook, J. (1989) Molecular Cloning: A Laboratory Manual, 2nd ed., Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY.

Chem. Res. Toxicol., Vol. 11, No. 5, 1998 427 (26) Carmichael, P. L., Hewer, A., Osborne, M. R., Strain, A. J., and Phillips, D. H. (1995) Detection of bulky DNA lesions in the liver of patients with Wilson’s disease and primary haemochromatosis. Mutat. Res. 326, 235-243. (27) Devanaboyina, U., and Gupta, R. C. (1996) Sensitive detection of 8-hydroxy-2′-deoxyguanosine in DNA by 32P-postlabeling assay and the basal levels in rat tissues. Carcinogenesis 17, 917924. (28) Imlay, J. A., Chin, S. M., and Linn, S. (1988) Toxic DNA damage by hydrogen peroxide through the Fenton reaction in vivo and in vitro. Science 240, 640-642. (29) Luo, Y., Han, Z., Chin, S. M., and Linn, S. (1994) Three chemically distinct types of oxidants formed by iron-mediated Fenton reaction in the presence of DNA. Proc. Natl. Acad. Sci. U.S.A. 91, 1243812442. (30) Toyokuni, S., and Sagripanti, J.-L. (1996) Association between 8-hydroxy-2′-deoxyguanosine formation and DNA strand breaks mediated by copper and iron. Free Radical Biol. Med. 20, 859864. (31) Kennedy, L. J., Moore, K., Jr., Cauldfield, J. L., Tannenbaum, S. R., and Dedon, P. C. (1997) Quantitation of 8-oxoguanine and strand breaks produced by four oxidizing agents. Chem. Res. Toxicol. 10, 386-392. (32) Eichhorn, G. L., and Shin, Y. A. (1968) Interaction of metal ions with polynucleotides and related compounds. XII. The relative effect of various metal ions on DNA helicity. J. Am. Chem. Soc. 90, 7323-7328. (33) Pezzano, H., and Podo, F. (1980) Structure of binary complexes of mono- and polynucleotides with metal ions of the first transition group. Chem. Rev. 80, 366-401. (34) Chevion, M. (1988) A site-specific mechanism for free radical induced biological damage: the essential role of redox-active transition metals. Free Radical Biol. Med. 5, 27-37. (35) Bilinski, T., Krawiec, Z., Liczmanski, A., and Litwinska, J. (1985) Is hydroxyl radical generated by the Fenton reaction in vivo? Biochem. Biophys. Res. Commun. 130, 533-539. (36) Fee, J. A. (1982) Is superoxide important in oxygen poisoning? Trends Biochem. Sci. 7, 84-86. (37) Le´onard, A., and Lauwerys, R. R. (1980) Carcinogenicity and mutagenicity of chromium. Mutat. Res. 76, 227-239. (38) Nechay, B. R., Nanninga, L. B., and Nechay, P. S. E. (1986) Vanadyl(IV) and vanadate(V) binding to selected endogenous phosphate, carboxyl, and amino ligands; calculations of cellular vanadium species distribution. Arch. Biochem. Biophys. 251, 128138. (39) Chang, J., Watson, W. P., Randerath, E., and Randerath, K. (1993) Bulky DNA adduct formation induced by Ni(II) in vitro and in vivo as assayed by 32P-postlabeling. Mutat. Res. 291, 147-159. (40) Randerath, E., Watson, W. P., Zhou, G.-D., Chang, J., and Randerath, K. (1995) Intensification and depletion of specific bulky renal DNA adducts (I-compounds) following exposure of male F344 rats to the renal carcinogen ferric nitrilotriacetate (FeNTA). Mutat. Res. 341, 265-279. (41) Randerath, E., Zhou, G.-D., and Randerath, K. (1996) Organspecific oxidative DNA damage associated with normal birth in rats. Carcinogenesis 17, 2563-2570; corrigendum (1997) Carcinogenesis 18, 857-866.

TX970156L