Comparison of the levels of 8-hydroxyguanine in DNA as measured by gas chromatography mass spectrometry following hydrolysis of DNA by Escherichia coli Fpg protein or formic acid

Abstract

8-hydroxyguanine (8-OH-Gua) is one of many lesions generated in DNA by oxidative processes including free radicals. It is the most extensively investigated lesion, due to its miscoding properties and its potential role in mutagenesis, carcinogenesis and aging, and also to the existence of analytical methods using HPLC and gas chromatography mass spectrometry (GC/MS). Some studies raised the possibility of artifacts generated during sample preparation. We investigated several experimental conditions in order to eliminate possible artifacts during the measurement of 8-OH-Gua by GC/MS. Derivatization has been reported to produce artifacts by oxidation of guanine to 8-OH-Gua in acid-hydrolysates of DNA, although the extent of artifacts seems to depend on experimental conditions. For removal of 8-OH-Gua from DNA, we used either formic acid hydrolysis or specific enzymatic hydrolysis with Escherichia coli Fpg protein. Derivatization of enzyme-hydrolysates should not generate additional 8-OH-Gua because of the absence of guanine, which is not released by the enzyme, whereas guanine released by acid may be oxidized to yield 8-OH-Gua. The measurement of 8-OH-Gua in calf thymus DNA by GC/isotope-dilution MS (GC/IDMS) using these two different hydrolyses yielded similar levels of 8-OH-Gua. This indicated that no artifacts occurred during derivatization of acid-hydrolysates of DNA. Pyridine instead of acetonitrile and room temperature were used during derivatization. Pyridine reduced the level of 8-OH-Gua, when compared with acetonitrile, indicating its potential to prevent oxidation. Two different stable-isotope labeled analogs of 8-OH-Gua used as internal standards for GC/IDMS analysis yielded similar results. A comparison of the present results with the results of recent trials by the European Standards Committee for Oxidative DNA Damage (ESCODD) is also presented.

INTRODUCTION

DNA damage generated by oxygen-derived free radicals is implicated in mutagenesis, carcinogenesis and aging (1). Of these radicals, the hydroxyl radical (OH radical) is the most reactive and generates multiple modifications in DNA including base and sugar lesions, strand breaks and DNA-protein crosslinks (2–4). Accurate and reliable measurement of DNA damage is essential for understanding the mechanisms of formation of DNA lesions and their cellular repair, and thus to ascertain the consequences of oxidative stress in certain diseases. In recent years, one of these products, 7,8-dihydro-8-oxoguanine (8-hydroxyguanine) (8-OH-Gua, also called 8-oxoGua or 8-oxoG) has been extensively investigated, in part due to its miscoding properties and to the availability of a method using HPLC with electrochemical detection (ECD) for the measurement of its nucleoside form 8-hydroxy-2′-deoxyguanosine (8-OH-dGuo or 8-oxodG) following enzymatic hydrolysis of DNA (5–7). However, the measurement of a single product such as 8-OH-dGuo may be misleading, because free radicals generate many products in DNA at the same time (2–4), and because product yields depend on reaction conditions (3). Moreover, the endogenous levels of 8-OH-Gua measured in various laboratories significantly differ from one another, suggesting a possible laboratory-dependent variability in measurements (5,7).

8-OH-Gua is formed by the addition of an OH radical to the C-8 position of guanine in DNA and subsequent one-electron oxidation of the resulting OH-adduct radical of guanine (3,8). On the other hand, the OH-adduct radical of guanine can also undergo one-electron reduction to yield 2,6-diamino-4-hydroxy-5-formamidopyrimidine (FapyGua) (3,8). Thus, the relative yields of 8-OH-Gua and FapyGua as well as those of other products may depend on reaction conditions and the redox status of cells (3,9). Hence, the measurement of a single product such as 8-OH-Gua in DNA may give limited information on the initial radical attack, and distribution and biological properties of the other various products. Therefore, it may be misleading to measure 8-OH-Gua as a sole marker for oxidative stress. Recent data demonstrated the advantage of measuring multiple DNA modified bases (10,11).

Of the techniques used for measuring free radical damage to DNA, gas chromatography mass spectrometry (GC/MS) is the only one capable of measuring multiple modified bases from all four DNA bases in a single DNA sample (6,12). Recently, the artifactual formation of five modified bases from the corresponding intact DNA bases has been reported to occur during derivatization (i.e., trimethylsilylation) of formic acid-hydrolysates of DNA prior to GC/MS analysis (13,14). Of more than 20 modified bases that can be measured by GC/MS (12), only 8-OH-Gua, 5-hydroxycytosine (5-OH-Cyt), 8-hydroxyadenine (8-OH-Ade), 5-hydroxymethyluracil (5-OHMeUra) and 5-formyluracil have been reported to be artifactually formed (13,14). However, hydrolysis and derivatization conditions used differed significantly from those previously published. Furthermore, the wealth of relevant papers reporting the levels of these compounds in DNA from various sources were ignored, excluding a comparison between published results (13,14). Recent data confirmed that experimental conditions can affect the levels of modified bases in DNA as measured by GC/MS and artifactual formation of the aforementioned products could be avoided, if proper experimental conditions were used (15). Possible artifacts were recently discussed (15,16). Because 8-OH-Gua is measured in many laboratories as an indicator of oxidative DNA damage, a European Standards Committee for Oxidative DNA Damage (ESCODD) was established to resolve problems associated with the measurement of 8-OH-Gua. In two recent trials, samples of 8-OH-dGuo and calf thymus DNA from a stock source were distributed for analysis by a number of laboratories using a variety of analytical techniques. Results showed significant variations in the levels of 8-OH-dGuo between laboratories and techniques used for measurement (17,18).

Due to the biological importance of 8-OH-Gua, its accurate and reproducible measurement is of great importance. In the present study, we established and validated a method that meets these two objectives. The levels of 8-OH-Gua in calf thymus DNA were measured by gas chromatography/isotope-dilution mass spectrometry (GC/IDMS) following hydrolysis either by Escherichia coli Fpg protein (Fpg-Eco) or by formic acid. The aim was to compare the levels of 8-OH-Gua in DNA using two different hydrolysis methods. The rationale was that acidic hydrolysis releases modified DNA bases as well as guanine (and the three other intact bases), which may be oxidized during derivatization to yield additional 8-OH-Gua. In contrast, Fpg protein only releases modified purines and no intact bases, excluding a possible artifactual formation of 8-OH-Gua from guanine during derivatization.

MATERIALS AND METHODS

Materials

Materials for GC/MS were obtained as previously described (19,20). Calf thymus DNA was obtained from Sigma Chemical Company (St Louis, MO). Ultra high purity nitrogen was purchased from Matheson Gas Products (Parsippany, NJ). 2′-deoxyguanosine-1,3,7,9-¹⁵N₄-(2-amino-¹⁵N) (dGuo-¹⁵N₅) was purchased from Cambridge Isotope Laboratories (Andover, MA). 8-hydroxyguanine-1,3-¹⁵N₂-(2-amino-¹⁵N)-2-¹³C (8-OH-Gua-¹⁵N₃-¹³C) and 8-hydroxy-2′-deoxyguanosine-8-¹⁸O (8-OH-dGuo-¹⁸O) were synthesized by Dr Victor Nelson from Program Resources, Inc. (Frederick, MD) (20). dGuo-¹⁵N₅ and 8-OH-dGuo-¹⁸O were dissolved in water. 8-OH-Gua-¹⁵N₃-¹³C was dissolved in 0.01 N NaOH. The concentration of these compounds was calculated by weight and by measuring their absorption spectra using the following absorption coefficients: dGuo-¹⁵N₅, 13 000 M^–1 cm^–1 at 254 nm (21); 8-OH-dGuo-¹⁸O, 9908 M^–1 cm^–1 at 245 nm (22); 8-OH-Gua-¹⁵N₃-¹³C, 11 600 M^–1 cm^–1 at 283 nm (23). Fpg-Eco and E.coli Nth protein (Nth-Eco) were obtained as previously described (24,25).

Preparation of DNA samples

Calf thymus DNA was dissolved at 4°C in Na-phosphate buffer (pH 7.4) (0.3 mg/ml). The concentration of DNA was determined by UV spectroscopy (1 absorbance unit = 50 mg of DNA/ml). Aliquots of this solution containing 100 mg of DNA were dried under vacuum in a SpeedVac.

Hydrolysis with formic acid

For hydrolysis with formic acid, aliquots of 8-OH-Gua-¹⁵N₃-¹³C (2 pmol) and dGuo-¹⁵N₅ (2.5 µmol) were added to DNA samples (100 µg) as internal standards and the samples were dried under vacuum in a SpeedVac. Three independent measurements were carried out for each data point. Instead of 8-OH-Gua-¹⁵N₃-¹³C, aliquots of 8-OH-dGuo-¹⁸O (2 pmol) were added to some samples. Upon acidic hydrolysis, the latter compound yields 8-OH-Gua-¹⁸O, which is also used as an internal standard for 8-OH-Gua. The use of dGuo-¹⁵N₅ permits the determination of the DNA amount by GC/IDMS. Upon hydrolysis, this compound yields guanine-¹⁵N₅, which is used as an internal standard for quantification of guanine in DNA, and consequently for the determination of the DNA amount (15). The results of DNA determination by this method and by UV absorbance measurements correlated well with each other. DNA samples were hydrolyzed with 0.5 ml of 60% formic acid in evacuated and sealed tubes at 140°C for 30 min. The hydrolysates were frozen in vials placed in liquid nitrogen and then lyophilized for 18 h. For derivatization of DNA hydrolysates, 60 µl of a mixture of nitrogen-bubbled bis(trimethylsilyl)trifluoroacetic acid (BSTFA) (containing 1% trimethylchlorosilane) and pyridine (1:1, v/v) were added to the vials. The samples were purged individually with ultra high purity nitrogen, vortexed and then tightly sealed under nitrogen with Teflon-coated septa. The derivatization was carried out at 23°C for 2 h by vigorously shaking the vials. A small portion of the samples (20 µl) was removed and placed in vials used for injection of samples onto the GC-column. Vials were purged with nitrogen and tightly sealed with septa.

Hydrolysis with E.coli Fpg protein

Dried DNA samples were dissolved in 100 µl of Na-phosphate buffer (final concentration 50 mM, pH 7.4) containing 100 mM KCl, 1 mM EDTA and 0.1 mM dithiothreitol. Following addition of Fpg-Eco, three replicates of each sample were incubated at 37°C. The amount of the enzyme and time of incubation varied depending on the experiments. Details are given in the figure legends. As controls, DNA samples were incubated with heat-inactivated enzyme (140°C for 30 min) or without the enzyme. In another instance, Nth-Eco was used instead of Fpg-Eco as a further control. Following incubation, DNA was precipitated with 260 µl of cold ethanol (at –20°C), and the samples were kept at –20°C for 2 h. An aliquot of 8-OH-Gua-¹⁵N₃-¹³C (2 pmol) was added as an internal standard to each sample. Samples were centrifuged at 15 000 g for 30 min at 4°C. DNA pellets and supernatant fractions were separated. Ethanol was removed from supernatant fractions under vacuum in a SpeedVac. The supernatant fractions were lyophilized for 18 h, and then derivatized with 0.6 ml of a mixture of nitrogen-bubbled BSTFA and pyridine (1:1, v/v) at room temperature for 2 h. After cooling, the samples were centrifuged at 5000 g for 30 min to precipitate the salt. The clear supernatant fractions were removed and placed in vials used for injection of samples onto the GC-column. Vials were purged with nitrogen and tightly sealed with septa.

Analysis by gas chromatography mass spectrometry

The derivatized samples were analyzed by GC/IDMS with selected-ion monitoring (SIM) using a gas chromatograph (Hewlett-Packard Model 5890 Series II) mass spectrometer (Hewlett-Packard Model 5989A MS Engine) system equipped with an automatic sampler. The column was a fused silica capillary column (12.5 m × 0.2 mm i.d.) coated with cross-linked 5% phenylmethylsilicone gum phase (film thickness, 0.33 µm) (Hewlett-Packard, Palo Alto, CA). Ultra high purity helium was used as the carrier gas. The injection port and the GC/MS interface were kept at 250 and 280°C, respectively. The ion source temperature was 280°C. The oven temperature of the gas chromatograph was programmed from 130 to 280°C at a rate of 8°C/min after 2 min at 130°C. The column head pressure was 65 kPa. SIM was performed in the electron ionization mode at 70 eV using the characteristic ions of the trimethylsilyl derivatives of 8-OH-Gua, guanine and their stable isotope-labeled analogs (12,19). An aliquot of each sample (4 µl) was injected into the injection port of the gas chromatograph using the split mode of injection with a split ratio of 10 to 1.

RESULTS

The objective of this study was to determine and compare the levels of 8-OH-Gua in calf thymus DNA following either enzymatic hydrolysis with Fpg-Eco or chemical hydrolysis with formic acid. Trimethylsilylation that follows the hydrolysis to obtain the volatile trimethylsilyl derivative of 8-OH-Gua for GC/MS analysis may artifactually generate additional 8-OH-Gua from guanine released by acidic hydrolysis (13,14). In contrast, 8-OH-Gua should not be generated during trimethylsilylation of the supernatant fractions of Fpg-Eco-hydrolysates of DNA because of the absence of guanine, which is not released by the enzyme. The comparison of the levels of 8-OH-Gua measured using these two different hydrolysis methods should reveal whether the trimethylsilylation of acid-hydrolysates of DNA generates 8-OH-Gua from guanine under a given set of conditions. In this work, we modified the derivatization conditions in search of measures to prevent guanine oxidation. Besides BSTFA, which is the actual derivatization agent, pyridine was used instead of acetonitrile in the derivatization mixture. The rationale was that pyridine is an aromatic compound and thus may prevent oxidation of guanine in acid-hydrolysates during derivatization. Ethanethiol has also been used as an oxidation-preventing compound (26). However, this compound is quite inconvenient to use because of its strong odor. Pyridine may also serve as an appropriate solvent for intact and modified bases. This is especially true for 8-OH-Gua and guanine, which may not be completely soluble during derivatization at room temperature (23). Previously, trifluoroacetic acid was used to dissolve guanine and 8-OH-Gua during derivatization (23). However, this compound interferes with the analysis by GC/MS of pyrimidine-derived DNA lesions (15).

In this study, a variety of experimental conditions were used to ensure the consistency and reproducibility of the results. Trimethylsilylation was done at room temperature for 2 h. Derivatization at room temperature was reported to prevent artifactual formation of 8-OH-Gua from guanine in acid-hydrolysates of DNA (15,23,27). On the other hand, it was shown that the levels of other modified bases such as 5-OH-Cyt, 5-OH-Ura, 8-OH-Ade and 5-OHMe-Ura are not affected by changes in derivatization temperature from room temperature to 120°C, when proper experimental conditions are used (15).

Figure 1 illustrates the levels of 8-OH-Gua measured in two different lots of calf thymus DNA measured by GC/IDMS following hydrolysis by either Fpg-Eco or formic acid. The levels of 8-OH-Gua measured in DNA incubated without the enzyme or with heat-inactivated Fpg-Eco were similar, indicating no excision by the inactive protein (Fig. 1, columns 1a–d and 2a–c). Active Nth-Eco, which does not excise 8-OH-Gua, was also used to provide another control experiment. The levels observed following treatment of DNA with this enzyme were similar to those obtained without the enzyme or with inactivated Fpg-Eco (Fig. 1, columns 3a–c). This confirms that Nth-Eco does not excise 8-OH-Gua. Incubation with active Fpg-Eco significantly increased the level of 8-OH-Gua found in the supernatant fractions (columns 4a–e). Different amounts of enzyme and varying incubation times were used. Details are given in the figure legend. The lowest level of the enzyme/DNA ratio (4 µg/100 µg) was chosen on the basis of the recently described optimal digestion conditions for Fpg-Eco (28). At greater enzyme/DNA ratios, no dependence of the excised levels on the enzyme amount and incubation time was observed, likely due to saturating amounts of the enzyme used. Insignificant differences between the levels of 8-OH-Gua in two different lots of calf thymus DNA were discernable. The levels of 8-OH-Gua observed using acidic hydrolysis were not significantly different from those obtained using Fpg-Eco hydrolysis (compare columns 4a–e with 5a–e in Fig. 1). In one instance, 8-OH-dGuo-¹⁸O, which yields 8-OH-Gua-¹⁸O upon acidic hydrolysis, was used as an internal standard instead of 8-OH-Gua-¹⁵N-¹³C to test whether the use of a different internal standard would result in significantly different levels of 8-OH-Gua, when acidic hydrolysis was used. This experiment was performed, since differences may result from preparation of the solutions of these compounds. Thus, 8-OH-dGuo-¹⁸O is readily soluble in water, whereas 0.01 N NaOH is required to dissolve 8-OH-Gua-¹⁵N-¹³C, resulting in different absorption coefficients (see Materials and Methods) (22,23). No significant difference between the levels of 8-OH-Gua was observed when these two compounds were used as internal standards (compare columns 5a–c and 5e with 5d in Fig. 1). It should be pointed out that the level of the internal standard 8-OH-Gua-¹⁵N-¹³C or 8-OH-dGuo-¹⁸O added to DNA samples was similar to the level of 8-OH-Gua in DNA. The former was no more than twice the level of the latter or vice versa to ensure the linearity of the response of the mass spectrometer to the quantities of these compounds. A linear relationship was found between the ratio of the amounts of these compounds and the ratio of the ion currents of their trimethylsilyl derivatives, when these ratios varied from 0.25 to 2.5 (23,29).

Figure 1.

Levels of 8-OH-Gua in calf thymus DNA as measured by GC/IDMS under various conditions. 1, no enzyme; a, 3 h incubation, lot 1 of calf thymus DNA; b, 1 h incubation, lot 1; c, 1 h, lot 2; d, 1 h, lot 2. 2, heat inactivated Fpg-Eco; a, 4 µg, 1 h, lot 1 of calf thymus DNA; b, 4 µg, 1 h, lot 2; c, 4 µg, 1 h, lot 2. 3, active Nth-Eco; a, 4 µg, 1 h, lot 1 of calf thymus DNA; b, 4 µg, 1 h, lot 2; c, 4 µg, 1 h, lot 2. 4, active Fpg-Eco; a, 5 µg, 1 h plus 5 µg, 2 h (total 10 µg), lot 1 of calf thymus DNA; b, 5 µg, 3 h, lot 1; c, 4 µg, 1 h, lot 1; d, 4 µg, 1 h, lot 2; e, 4 µg, 1 h, lot 2. 5, formic acid hydrolysis; a–c, lot 1 of calf thymus DNA; d, lot 1, 8-OH-Gua-¹⁸O as internal standard; e, lot 2. Each bar represents the mean value ± standard deviation from three independent measurements. Columns 4a–e and 5a–e are statistically different from columns 1a–3c (p <0.05).

Figure 2 illustrates the mean values and standard deviations of the levels of 8-OH-Gua, which were calculated using all the results in each group of experiments shown in Figure 1. This graph indicates that there was no statistical difference between the samples incubated without enzyme and those incubated with inactivated enzyme or Nth-Eco (Fig. 2, columns 1–3). Furthermore, it is clear that there was no statistical difference between the samples hydrolyzed with active Fpg-Eco and those hydrolyzed with formic acid (columns 4 and 5) and that the levels of 8-OH-Gua released by active Fpg-Eco and formic acid (columns 4 and 5) were significantly greater than those found in the control samples (columns 1–3).

Figure 2.

Average of the values in Figure 1. 1, no enzyme (n = 12); 2, heat-inactivated Fpg-Eco (n = 9); 3, active Nth-Eco (n = 9); 4, active Fpg-Eco (n = 15); 5, formic acid hydrolysis (n = 15). Columns 4 and 5 are statistically different from columns 1–3 (P <0.05).

Moreover, the levels of 8-OH-Gua found in this work were compared with those previously reported using commercial calf thymus DNA. This comparison is illustrated in Figure 3. Values that are off scale in this figure are given on the graph. Some of the previous values were given in ‘number of lesions/10⁵ guanines’. These values were converted to ‘number of lesions/10⁶ DNA bases’ by assuming that guanine constitutes 21.5% of mammalian DNA (30). The formula used was: 1 lesion/10⁵ guanines = 1/0.465 (or 2.15) lesions/10⁶ DNA bases. Besides the levels of 8-OH-Gua measured by GC/MS, some levels that were measured by HPLC under a variety of conditions are relevant here and are also shown. The majority of these measurements were obtained using commercially available calf thymus DNA. For comparison, Figure 3 also shows several levels of 8-OH-Gua in other mammalian tissues that were measured along with those in calf thymus DNA in the same study using HPLC-ECD following formic acid hydrolysis or enzymatic hydrolysis of DNA (31). The level of 8-OH-Gua that was measured using high performance liquid chromatography tandem mass spectrometry (HPLC/MS/MS) and rat liver DNA (32) is also shown. There is a wealth of data on the measurement of 8-OH-Gua in DNA by HPLC-ECD. It is beyond the scope of this paper to compare all 8-OH-Gua levels measured by this technique with those measured in this work. However, the data from the first and second ESCODD trials, which involved several European laboratories, are shown in Figure 4 (17,18). These measurements were performed by GC/MS, HPLC, HPLC/MS/MS, and immunological and ³²P-labeling techniques using either commercial calf thymus DNA or pig liver DNA.

Figure 3.

Levels of 8-OH-Gua in calf thymus DNA and in DNA of various other sources obtained by derivatization of hydrolysates under various conditions (numbers with arrows indicate values, which are out of scale in this graph). 1, at room temperature, Fpg-Eco hydrolysis, this study; 2, at room temperature, formic acid hydrolysis, this work; 3 and 4, at room temperature and 120°C, respectively (15); 5 and 6, at room temperature and 90°C, respectively (23); 7 and 8, at 130°C and 130°C following prepurification of DNA hydrolysates, respectively (34); 9 and 10, at 130°C and 130°C following prepurification of DNA hydrolysates, respectively (13,34); 11 and 12, at room temperature and 90°C, respectively (27); 13 and 14, at room temperature and room temperature plus ethanethiol, respectively (26); 15 and 16, at 90°C and 90°C plus ethanethiol, respectively (26); 17 and 18, at 80°C and 80°C plus N-phenyl-1-naphthylamine (33); 19, HPLC-ECD, enzymatic hydrolysis (34); 20, HPLC-ECD, formic acid hydrolysis (31); 21, HPLC-ECD, enzymatic hydrolysis (35); 22, HPLC-ECD, formic acid hydrolysis, rat organ DNA (31); 23, HPLC/MS/MS, enzymatic hydrolysis, rat liver DNA (32).

Figure 4.

Levels of 8-OH-Gua in commercial calf thymus DNA measured in this work and by ESCODD trials (numbers with arrows indicate values, which are out of scale in this graph). 1, at room temperature, Fpg-Eco hydrolysis, this study; 2, at room temperature, formic acid hydrolysis, this work; 3, first ESCODD trial (mean value), GC/MS/HPLC, calf thymus DNA (17); 4, first ESCODD trial (mean value), HPLC-ECD, enzymatic hydrolysis, calf thymus DNA (17); 5, first ESCODD trial, GC/MS/HPLC, pig liver DNA (17); 6, first ESCODD trial, HPLC-ECD, enzymatic hydrolysis, pig liver DNA (17); 7–11, second ESCODD trial, GC/MS, calf thymus DNA (18); 12–25, second ESCODD trial, HPLC, calf thymus DNA (18); 26–28, second ESCODD trial, HPLC/MS/MS, immunological, and ³²P-labeling, respectively, calf thymus DNA (18).

DISCUSSION

The results show that there is no statistical difference between the levels of 8-OH-Gua measured in DNA under the conditions of this study using chemical hydrolysis with formic acid and enzymatic hydrolysis with Fpg-Eco. This clearly indicates that (i) under the strict experimental conditions used in this study, 8-OH-Gua was not formed artifactually from free guanine in formic acid-hydrolysates of DNA during derivatization at room temperature and (ii) Fpg-Eco can remove most if not all 8-OH-Gua residues present in DNA at background levels. Small amounts of 8-OH-Gua were released during the incubation of DNA without the enzyme, with inactivated Fpg-Eco or active Nth-Eco, which does not release 8-OH-Gua. This may indicate the instability of the glycosidic bond between the sugar moiety and 8-OH-Gua in DNA under the mild incubation conditions used. The extent of incubation from 30 min to 3 h did not affect these amounts. This may indicate that 8-OH-Gua was not formed during incubation by oxidation of guanine in DNA. The similarity of the levels observed following hydrolysis by active Fpg-Eco during the time period from 30 min to 3 h supports this notion.

The results suggest that pyridine in the derivation mixture helps prevent artifactual formation of 8-OH-Gua. Comparison to a previous study that used acetonitrile instead of pyridine under similar conditions (15) attests to this fact (Fig. 3, columns 1–3). The previous level was ∼2-fold greater that those observed in this work (20 versus 10 lesions/10⁶ DNA bases). Pyridine along with BSTFA was used previously for derivatization of formic acid-hydrolysates of calf thymus DNA to determine the level of 8-OH-Gua by GC/IDMS (33). N-phenyl-1-naphthylamine was used as an antioxidant to prevent guanine oxidation. However, the derivatization temperature was 80°C and the level of 8-OH-Gua observed was significantly greater than in this paper, although the use of N-phenyl-1-naphthylamine reduced the level of 8-OH-Gua (Fig. 3, columns 17 and 18). In another instance, ethanethiol was used as an oxidation-preventing compound during derivatization (26). It reduced the level of 8-OH-Gua at both room temperature and 90°C derivatization (Fig. 3, columns 13–16). However, the levels obtained at room temperature were 2–3-fold greater than that in this work (columns 13 and 14). It should be pointed out that ethanethiol has a strong odor and caution is recommended during its use (26). The level of 8-OH-Gua observed in this work was similar to those obtained by derivatization at room temperature or at 130°C following prepurification of formic acid-hydrolysates of DNA (23,34) (Fig. 3, columns 5 and 8). In another instance, however, a 2-fold greater level (24 lesions/10⁶ DNA bases) was reported following prepurification (column 10) (13). The levels obtained without prepurification were very high, suggesting that oxygen may not have been removed from the derivatization mixtures (Fig. 3, columns 7 and 9) (13,34,35).

A recent study reported the levels of 8-OH-Gua in rat kidney DNA, which were measured by HPLC-ECD as 8-OH-Gua or 8-OH-dGuo following either formic acid hydrolysis or enzymatic hydrolysis with nuclease P1 and alkaline phosphatase (31). No difference between the two methods of hydrolysis was observed, indicating that formic acid hydrolysis does not cause artifacts. In the same study, formic acid hydrolysis was also used to determine the level of 8-OH-Gua by HPLC-ECD in calf thymus DNA. Interestingly, the level found (13 lesions/10⁶ DNA bases) is similar to that in this paper and to that measured previously using HPLC-ECD after enzymatic hydrolysis (11 lesions/10⁶ DNA bases) (Fig. 3, columns 20 and 19, respectively) (34). However, in another instance, a greater level (17 lesions/10⁶ DNA bases) was reported using HPLC-ECD after enzymatic hydrolysis (Fig. 3, column 21) (35). The same method led to lower levels of 8-OH-Gua in rat liver (2–5 lesions/10⁶ DNA bases), some of which are shown in Figure 3 (column 22). The use of HPLC/MS/MS instead of HPLC-ECD provided a similar level of 8-OH-Gua in rat liver (4.5 lesions/10⁶ DNA bases) (Fig. 3, column 23) (32). These results show that rat liver DNA measured under similar experimental conditions had lower levels of 8-OH-Gua than commercial calf thymus DNA, although these differences may have been due to different DNA isolation techniques. Recently, it was shown that DNA may be oxidized during isolation and precautions must be taken to prevent oxidation of guanine, which is most prone to oxidation among DNA bases (7). Differences in DNA isolation techniques were thought to cause the wide range of 8-OH-Gua levels in DNA of various sources reported using HPLC-ECD (7).

For comparison, Figure 4 illustrates the levels of 8-OH-Gua measured in this study and those reported by two ESCODD trials. The first ESCODD trial using several techniques to measure the level of 8-OH-Gua in commercial calf thymus DNA and in pig liver DNA provided mean values between 13.5 and 20 lesions/10⁶ DNA bases and between 11 and 42 lesions/10⁶ DNA bases, respectively (Fig. 4, columns 3–6) (17). The GC/MS values, some of which also included measurements following guanase digestion to remove guanine (36), varied from 8 to 34 lesions/10⁶ DNA bases, whereas the HPLC-ECD values were in the range of 4 to 46 lesions/10⁶ DNA bases. It should be noted that the laboratories involved in these measurements used the same batch of commercial calf thymus DNA. All the levels of 8-OH-Gua in calf thymus DNA obtained by the second ESCODD trial are shown in Figure 4 (columns 7–28). These include the values measured by GC/MS, HPLC and other techniques. The 8-OH-Gua levels measured by GC/MS varied from 21.5 to 86.7 lesions/10⁶ DNA bases (or 10–40.3 lesions/10⁵ guanines) (columns 7–11), whereas those measured by HPLC were in the range of 10.5 to 39.4 lesions/10⁶ DNA bases (or 4.9 to 18.3 lesions/10⁵ guanines) (columns 12–25). Three values obtained by other techniques were from 12.3 to 37.2 lesions/10⁶ DNA bases (or 5.7 to 17.3 lesions/10⁵ guanines) (columns 26–28). The values measured in the second ESCODD trial by GC/MS were generally greater than those measured by HPLC. In some instances, however, both techniques provided similar values (compare columns 7, 8 and 11 with columns 12, 15, 24 and 25 in Fig. 4). The lower values reported by ESCODD are similar to those found in this work. Especially, the values found in the present study are remarkably similar to some of those reported by both ESCODD trials, as the comparison of the columns 1, 2, 6, 16, 18, 20 and 26 in Figure 4 clearly shows. The tendency of certain GC/MS measurements to yield higher than expected values in the second ESCODD trial is thought to result from a calibration problem (18). It was suggested that the internal standard, stable isotope-labeled 8-OH-Gua, might pose a problem, because 8-OH-dGuo in DNA must first be hydrolyzed to free 8-OH-Gua. In this study, both labeled 8-OH-Gua as a free base (8-OH-Gua-¹⁵N₃-¹³C) and its nucleoside form (8-OH-dGuo-¹⁸O) were used as internal standards. In both instances, similar levels of 8-OH-Gua were observed. This indicates that the nucleoside form of the internal standard is fully hydrolyzed to the free base by acidic hydrolysis and should pose no problem. This is also in agreement with a recent study reporting the full recovery of 8-OH-Gua under the conditions of acidic hydrolysis, which were similar to those used in this work (31). Another possible source of bias suggested by ESCODD was that guanine may be oxidized during the lyophilization of formic acid-hydrolysates (18). However, the results of the present study indicate that this is not true, because both Fpg-Eco hydrolysis and acidic hydrolysis provided similar results. This suggests that, under the conditions of this study, lyophilization does not cause oxidation of guanine in acid-hydrolysates of DNA. The ESCODD trials clearly show that different results may be obtained in different laboratories using different techniques and experimental conditions (17,18).

CONCLUSIONS

The following conclusions can be drawn from the results of this study and their comparison with previously published data: (i) There is no significant difference between the levels of 8-OH-Gua in DNA measured by GC/IDMS using Fpg-Eco hydrolysis and formic acid hydrolysis of DNA. This observation and the fact that Fpg-Eco does not release guanine from DNA suggest that 8-OH-Gua was not formed as an artifact during derivatization of formic acid-hydrolysates of DNA at room temperature under the conditions used in this study. (ii) Fpg-Eco may be used instead of formic acid for removal of 8-OH-Gua prior to GC/MS analysis to determine the background levels in DNA. However, the large amounts of pure Fpg-Eco required for this purpose may make this procedure cost prohibitive to a laboratory. (iii) The results suggest that pyridine inhibits the oxidation of guanine to yield 8-OH-Gua during derivatization of formic acid-hydrolysates of DNA and serves as a proper solvent for 8-OH-Gua and guanine. (iv) The values of 8-OH-Gua observed in this work are remarkably similar to some of the values previously observed either by GC/MS with formic acid hydrolysis or by HPLC-ECD with enzymatic hydrolysis and to some of those obtained during ESCODD trials. This suggests that both GC/MS and HPLC techniques may provide similar levels of 8-OH-Gua, if unbiased experimental conditions are used. (v) The results in this study and those by others suggest that there is no need for a tedious and time-consuming prepurification of DNA hydrolysates (13,14,34) to prevent potential artifactual formation of 8-OH-Gua or any other modified bases (15). Artifacts of derivatization may depend on experimental conditions, especially on the complete exclusion of oxygen. (vi) A comparison of the results of the present study with previous results and those obtained by ESCODD trials indicates that different results may be obtained in different laboratories. Moreover, different techniques may yield different results. There is an urgent need for a consensus in the measurement techniques and conditions between laboratories. The ESCODD trials are an important step in this direction.