Mass Spectrometry - American

786

Chem. Res. Toxicol. 1998, 11, 786-793

Quantification of 5-(Hydroxymethyl)uracil in DNA by Gas Chromatography/Mass Spectrometry: Problems and Solutions Christopher J. LaFrancois, Kin Yu, and Lawrence C. Sowers* Division of Pediatrics, City of Hope National Medical Center, 1500 East Duarte Road, Duarte, California 91010 Received December 23, 1997

Oxidation of the thymine methyl group results in the formation of 5-(hydroxymethyl)uracil (HmU). HmU is a recognized endogenous DNA damage product, and HmU levels in DNA are increased by oxidant stress. Previous studies have reported substantially conflicting values for HmU levels in DNA. In studies utilizing postlabeling methods, HmU levels have been reported to be as high as or higher than the levels of some of the more commonly described DNA oxidation damage products such as 8-oxoguanine. In some studies utilizing GC/MS methods, however, HmU has been undetectable. In acid solution, the hydroxymethyl group of HmU can undergo condensation reactions with carboxylic acids, alcohols, and amines. While HmU can be accurately measured by GC/MS, the first step in the preparation of samples for GC/MS analysis is acid hydrolysis of the DNA. Such hydrolysis would be expected to result in substantial derivatization of HmU. We have utilized chemically synthesized oligonucleotides containing a known amount of HmU as well as an isotopically enriched standard to investigate the chemical modification of HmU during the acid hydrolysis of DNA. We conclude that HmU levels reported by GC/MS following acid hydrolysis may be up to an order of magnitude lower than the actual levels. Further, we propose modifications to the standard hydrolysis protocols which maximize recovery of HmU prior to silylation and analysis by GC/MS.

Introduction Oxidative DNA damage is an established component of genomic instability (1-5). Substantial efforts from many laboratories have been devoted to identifying, quantifying, and studying the biochemical properties of a complex array of oxidative damage products (6-18). Such products may result from endogenous oxidation, radiation, or xenobiotic chemicals. Among these potential damage products is 5-(hydroxymethyl)uracil (HmU)1 which results from oxidation of the thymine methyl group. The biological consequences of HmU in eukaryotes are not fully understood; however, specific repair activities have been identified which remove HmU from DNA (8, 10). Increased HmU formation has been correlated with some human diseases, and HmU has been proposed as a marker of oxidative DNA damage (19-21). Several methods have been employed to measure the levels of oxidized DNA bases including GC/MS (gas chromatography/mass spectrometry), HPLC-EC (highperformance liquid chromatography-electrochemical detection), and postlabeling (22-38). In general, the HmU levels measured by GC/MS are substantially lower than the HmU levels measured by other methods, whereas levels of some other oxidized bases, such as 8-oxoguanine, * Corresponding author: phone, (626) 359-8111, x-3845; fax, (626) 301-8458; e-mail, [email protected]. 1 Abbreviations: GC/MS, gas chromatography/mass spectrometry; HPLC-EC, high-performance liquid chromatography-electrochemical detection; HmU, 5-(hydroxymethyl)uracil; HmdU, 5-(hydroxymethyl)2′-deoxyuridine.

are generally higher when measured by GC/MS. Such problems more likely result from sample preparation, and not from the GC/MS method itself. Because the biological significance of a selected modified base is often presumed to correlate with the measured levels of that lesion, accurate methods of quantification take on additional importance. The GC/MS method has been widely applied to measure the levels of modified DNA bases because of the definitive chemical identity revealed by mass spectra and the capacity to simultaneously measure a multitude of different modified bases in the same chromatographic run. The molar response of silylated HmU in the GC/ MS method is reported to be similar to other bases (30), suggesting that the lower amounts measured by GC/MS might result from loss during DNA hydrolysis and/or derivatization. Factors potentially responsible for loss of HmU during sample preparation, such as formic acid concentration, time, and hydrolysis temperature, have been addressed previously, primarily on an empirical basis (28, 30, 35, 36). The generation of artifacts during the preparation of DNA samples containing modified bases comprises an inherent problem in the measurement of all such adducts (22-38). It is generally presumed that the measured amount of a modified base reflects the amount present in the original sample. Such a presumption is often difficult to confirm experimentally. The strength of this presumption increases when one uses an isotopically enriched internal standard which would undergo the same chemical reactions as the analyte of interest.

S0893-228x(97)00233-6 CCC: $15.00 © 1998 American Chemical Society Published on Web 06/13/1998

Quantification of HmU in DNA by GC/MS

Chem. Res. Toxicol., Vol. 11, No. 7, 1998 787

Potential artifacts resulting from sample preparation may be further examined by using synthetic oligonucleotides containing known amounts of a particular modified base. Such synthetic methods are now available for HmU-containing oligonucleotides as well as several other oxidized bases (39-44). An additional advantage of synthetic oligonucleotides as analytical standards is that they may be analyzed by a variety of methods in parallel, thus allowing comparison and validation of such methods. By using both synthetic oligonucleotides containing a known amount of HmU and a stable-isotope-enriched standard (45), we have investigated the chemical reactivity of HmU during DNA hydrolysis. We have identified the pathways by which HmU is lost during sample preparation, prior to GC/MS analysis, and we propose experimental modifications which maximize recovery of HmU for analysis by GC/MS. Our results indicate that the amount of HmU reported previously in several studies using the GC/MS method might have been an order of magnitude lower than the actual amount present in the original sample.

Materials and Methods Materials. Free base standards of the normal DNA bases, uracil, thymine, HmU, 6-azathymine, nuclease P1, bacterial alkaline phosphatase, and calf thymus DNA were purchased from Sigma Chemical Co., St. Louis, MO. Isotopically enriched water (18O-H2O, 98% enriched) was purchased from Cambridge Isotope Laboratories, Andover, MA. The preparation and characterization of unenriched HmdU as well as the isotopically enriched 5-(hydroxymethyl)-2′-deoxyuridine, containing 15N in the 1 and 3 ring positions and deuterium in the 5-methylene group and 2′′ positions, have been reported elsewhere (45). Both labeled and unlabeled HmdU were characterized by UV, NMR, and mass spectrometry, and the purity was verified by HPLC analysis. An oligonucleotide corresponding to the Dickerson dodecamer (46) containing one HmU residue in place of one thymine residue was prepared according to a published method (41). The sequence of the HmU-containing dodecamer, abbreviated as HDODEC, is 5′ CGCGAAT(HmU)CGCG. Instrumentation. GC/MS measurements were performed with a Hewlett-Packard 5890A GC interfaced to a 5970 mass selective detector. HPLC analysis of deoxynucleosides was performed with a Perkin-Elmer series IV liquid chromatograph interfaced with a Pharmacia-LKB 2140 spectral array detector. UV spectra were obtained with a Perkin-Elmer Lambda 3B double-beam ultraviolet-visible spectrophotometer. Methods. Oligonucleotides containing HmU were enzymatically digested using nuclease P1 followed by bacterial alkaline phosphatase as previously described (41). Oligonucleotides were incubated with nuclease P1 for 3 h followed by bacterial alkaline phosphatase overnight. Nucleoside digests were analyzed by HPLC using a Supelcosil LS-C18 reverse-phase column. The mobile phase consisted of a 0.1 M potassium phosphate buffer, pH 4.5, with a linear gradient of methanol from 5% to 50% over 15 min. GC/MS measurements were made using a Hewlett-Packard RP column. The injector temperature was set at 250 °C, and the detector interface was 280 °C. The oven temperature was started at 100 °C for 2 min and then ramped to 260 °C over 16 min. The final temperature was maintained for 2 min. The concentration of the self-complementary HmU-containing Dickerson dodecamer was measured by UV spectrophotometry using a calculated molar extinction coefficient (47) of 115 000 at 260 nm. The concentration of the isotope-labeled HmdU derivative was measured using a molar extinction coefficient of 10 300 at 265 nm, pH 2.0 (48). The extinction coefficients used for determining the concentration of solutions of HmU (49),

Figure 1. (A) Mass spectrum of silylated HmU, indicating the observed fragments and their identity. (B) Mass spectrum of the silylated formate ester of HmU and the corresponding fragments. uracil (49), thymine (49), and 6-azathymine (50) were 8100 at 261 nm (pH 7), 8200 at 259 nm (pH 7), 7900 at 265 nm (pH 7), and 5200 at 261 nm (0.1 M HCl), respectively. The concentration of calf thymus DNA in water was determined based upon the relationship 100 µg/mL ) 2.0 OD at 260 nm (51). The ratio of absorbance of the calf thymus DNA at 280 nm to 260 nm was 0.5. The amount of thymine per sample was calculated to be 9 × 10-8 mol/100 µg of DNA. Formic acid (88% and 96%) was obtained from Aldrich Chemical Co. A 60% aqueous solution of formic acid was prepared by diluting 4 mL of 88% formic acid with 2 mL of distilled water. Acid treatment of samples was conducted in Teflon-sealed 1-mL Hypo-vials (Pierce) containing 200 µL of the acid solution. Formic acid was removed under reduced pressure. Dried samples were derivatized with 100 µL of bis(trimethylsilyl)trifluoroacetamide containing 1% trimethylchlorosilane and 100 µL of dry acetonitrile. Acid treatment and derivatization were conducted at 150 °C for 40 and 30 min, respectively. Uracil was added as an internal standard prior to acid hydrolysis. To determine the amount of HmU in commercially available calf thymus DNA, the DNA was dissolved in water and the concentration was determined by measuring the UV spectrum. To each sample containing 100 µg of DNA was added 5 × 10-11 mol of the isotopically labeled standard. DNA samples were hydrolyzed for 1 h. Silylation of the hydrolyzed DNA samples was conducted in a volume of 70 µL. Approximately 2 µL was injected for GC/MS analysis in the SIM mode, monitoring the 255, 358 and 362 m/z ions.

Results The GC/MS Method Can Measure HMU Both Qualitatively and Quantitatively. The mass spectrum of silylated HmU (Figure 1A) displays a parent ion of 358 amu, which is also the largest fragment in the spectrum. Other fragments noted are those resulting from cleavage of the 5-substituent as indicated in Figure 1A. The relative molar response factor of the 358 amu ion of silylated HmU was measured in parallel with the

788 Chem. Res. Toxicol., Vol. 11, No. 7, 1998

Figure 2. Observed recovery of HmU, following formic acid treatment in different concentrations of formic acid, measured by GC/MS. The amount of HmU was normalized to the HmU signal observed in the control sample containing HmU and T which was not acid-hydrolyzed. Left: Data obtained for the free base, HmU. Center: Corresponding data obtained with the deoxynucleoside, HmdU. Right: Corresponding data obtained with the HmU-containing dodecamer, HDODEC. Error bars represent the standard deviation of three independent experiments.

predominant ions of several other pyrimidines including thymine, uracil, and 6-azathymine. The molar response factors of thymine (255 amu), uracil (241 amu), and HmU (358 amu), relative to 6-azathymine (256 amu), were determined to be 7.3, 5.1, and 4.7, respectively. The magnitudes of these molar response factors are very sensitive to instrument tuning conditions and must be determined prior to each series of experiments. These data confirm that the GC/MS method can measure HmU levels both quantitatively and qualitatively. Substantial Loss of HmU Is Observed Following Formic Acid Hydrolysis. When DNA bases are analyzed by GC/MS, the glycosidic bond is cleaved by acid hydrolysis to liberate the free base. Aqueous formic acid is usually employed because it is volatile and it is believed to result in minimal destruction of the DNA bases. We treated a standard solution containing HmU and thymine with 60%, 88%, and 96% formic acid at 150 °C for 40 min (Figure 2, left). A substantial loss of the HmU signal was noted with a maximum decline in 96% formic acid. With the deoxynucleoside, HmdU, similar destruction of HmU was observed (Figure 2, center). HmdU may be directly converted to the silylated free base; however, the yield of this reaction under standard silylation conditions is only approximately 36% (Figure 2, center). A Synthetic Oligonucleotide Can Be Prepared Containing a Known Amount of HmU. To examine the recovery of HmU from DNA, we constructed a synthetic 12-base self-complementary oligonucleotide containing one HmU and one thymine residue. The sequence of the oligonucleotide corresponds to the Dickerson dodecamer (46). To verify the composition of the oligonucleotide, it was enzymatically digested with nuclease P1 and bacterial alkaline phosphatase. The digestion products were analyzed by HPLC using a diode array detector. Under standard digestion conditions, the dodecamer containing only normal bases was completely digested to the 2′-deoxynucleosides. For the normal dodecamer, four HPLC peaks were observed at 270 nm corresponding to dC, dG, dT, and dA (Figure 3A). The identification of the chromatographic peaks was verified by examination of the UV spectra of each peak as well as cochromatography with authentic standards. In the HmU-containing dodecamer, one thymine is replaced by one HmU residue so that thymine and HmU

LaFrancois et al.

Figure 3. Analysis of the deoxynucleoside content of synthetic oligonucleotides by HPLC (270 nm) following enzymatic hydrolysis: (A) synthetic dodecamer containing only normal bases; (B) synthetic dodecamer in which one thymine residue was replaced by one HmU residue.

are equimolar. When using the conditions which resulted in complete digestion of the normal dodecamer, the HmUcontaining dodecamer is not completely digested (Figure 3B). As expected, replacement of one thymine residue by HmU results in a thymine peak which is roughly onehalf the size of the thymine peak in the normal dodecamer. Three additional peaks are seen in the chromatogram. The largest of the new peaks corresponds to HmdU as expected. Additional peaks correspond to the 5′-monophosphate of HmdU and the Tp(HmU) dinucleotide monophosphate. The relative amounts of the HmdU, HmdUMP, and Tp(HmU) peaks represent roughly 70%, 15%, and 15% of the total amount of HmU in the oligonucleotide. The amount of the normal bases, as well as the total HmU present in the three HmU-containing chromatographic peaks, is consistent with the expected composition of the HmU-containing dodecamer. The Recovery of HmU from DNA Following Acid Hydrolysis Is Very Low. The synthetic oligonucleotide containing one thymine and one HmU residue was then used to investigate the amount of HmU recovered following acid hydrolysis using three different formic acid concentrations. The amount of HmU recovered was determined by examining the relative areas of the silylated thymine and HmU peaks. As shown in Figure 2 (right), the amount of HmU recovered was not more than 30% using 60% formic acid and decreased with increasing formic acid concentration. The loss of HmU following acid hydrolysis is similar for the free base, deoxynucleoside, and HmU residue in a synthetic oligonucleotide at a given formic acid concentration. Formic Acid Reacts with HmU To Generate the Formate Ester. We then conducted a series of experiments to determine the fate of HmU under formic acid hydrolysis conditions. A sample of pure HmU was heated in 88% formic acid for 30 min at 150 °C. Following evaporation of the solvent and silylation, the reaction products were analyzed by GC/MS with the mass spectrometer operating in the scan mode. A substantial new GC peak was observed which eluted prior to the silylated HmU peak. The mass spectrum of this peak displayed a parent ion of 314 amu. This mass, as well as the additional fragment peaks shown in Figure 1B, is con-


sistent with the silylated formate ester of HmU. No such peak was observed following formic acid treatment of thymine, uracil, or 6-azathymine. To confirm the identity of this peak, HmU was heated in 13C-enriched formic acid. A similar GC peak was obtained; however, the mass spectrum displayed a parent ion of 315 amu and an M - 15 fragment at 300 amu. Unenriched fragments at 285, 269, and 255 amu were also observed. These data confirm that the product derived from HmU formic acid treatment contains the formate carbon, and the fragmentation pattern observed in the mass spectrum is consistent with the assignment of the formate ester structure as shown in Figure 1B. The formate ester of HmU is unstable and undergoes facile regeneration of HmU. We were unable to isolate the formate ester using either silica gel chromatography or HPLC methods. Therefore, we could not use the formate ester of HmU as a quantitative surrogate marker of HmU in DNA. Alternatively, we observed that treatment of samples containing the formate ester of HmU with water resulted in hydrolysis of the formate ester with subsequent regeneration of HmU. HmU Can Be Regenerated Following Formic Acid Treatment. The above studies demonstrate that substantial quantities of the formate ester of HmU are generated during formic acid treatment. The peak corresponding to the formate ester was observed following formic acid treatment of the free base and deoxynucleoside, as well as the HmU-containing oligonucleotide. Furthermore, the size of the formate ester peak, monitored at 285 amu, increased with increasing formic acid concentration. Having also demonstrated that the formate ester was easily hydrolyzed, generating HmU, we investigated conditions which might maximize the regeneration of HmU following acid hydrolysis. We examined dilute HCl, aqueous ammonia, and water at various times and temperatures from ambient to 150 °C. We observed that the maximum amount of HmU could be recovered by hydrolyzing the formic acid-treated sample with water at 60 °C in a sealed vial. Although the formate ester is completely lost by heating the sample in water at 60 °C for 10 min, we generally heated the samples for 30 min to ensure complete hydrolysis. We then repeated the experiment presented in Figure 2, in which HmU, HmdU, and the HmU-containing oligonucleotide were heated with different formic acid concentrations. This time, however, we heated the samples in water at 60 °C for 30 min following formic acid treatment in order to hydrolyze the HmU-formate ester back to the parent HmU. Recovery of up to 90% was observed in the experiment with the free base, and maximum recovery of HmU from the oligonucleotide (71%, Figure 4, right) was observed following hydrolysis in 88% formic acid. The Formation of HmU Derivatives Proceeds through Nucleophilic Attack at the Methylene Carbon. Based upon previously published studies (41, 52, 53), it was anticipated that the reaction products of HmU from acid treatment would proceed through nucleophilic attack at the methylene carbon as shown in Figure 5. While the reaction pathway shown in Figure 5 corresponds to an SN1 mechanism, as originally proposed by Scheit (53), it could as well be an SN2 process. A partial positive charge on the methylene carbon would be stabilize by the benzyl-like system, accounting for the


Figure 4. Observed recovery of HmU following formic acid treatment in different concentrations of formic acid and hydrolysis of the formate ester of HmU in water prior to derivatization, measured by GC/MS. The amount of HmU was normalized to the HmU signal observed in the control sample containing HmU and T which was not acid-hydrolyzed. Left: Data obtained for the free base, HmU. Center: Corresponding data obtained with the deoxynucleoside, HmdU. Right: Corresponding data obtained with the HmU-containing dodecamer, HDODEC. Error bars represent the standard deviation of three independent experiments.

reaction selectivity at the 5 versus 5′ positions. The higher formic acid concentration would result in greater formation of the formate ester, with less HmU remaining, consistent with the data shown in Figure 2. As the formate ester can be hydrolyzed regenerating HmU, a greater amount of recovery of HmU would be anticipated following hydrolysis in 88% versus 60% formic acid. The reduced recovery of HmU following hydrolysis in 96% formic acid suggests that additional, nonrecoverable derivatives may also form. The results of our studies reported thus far suggested that all HmU derivatives observed, including the residual HmU measured following acid hydrolysis, would proceed through nucleophilic attack at the methylene carbon. To investigate this hypothesis, we prepared a solution of 60% formic acid by adding 98% 18O-enriched water to 88% formic acid. This enriched aqueous formic acid was used to hydrolyze a sample of the HmU-containing oligonucleotide. Following lyophilization, the sample was silylated and analyzed by GC/MS. As shown in the results of the control experiment, Figure 6A, the relative intensity of the mass spectral lines in the isotope cluster of silylated HmU is consistent with the relative abundance of these lines calculated based upon the elemental composition of the molecule and the natural abundance of heavier isotopes of the atoms which comprise the molecule (54). The reaction scheme shown in Figure 5 suggests that, if the hydrolysis was conducted in 18O-enriched water, 18O would be incorporated into the HmU analyzed by GC/MS. Consistent with this expectation, the abundance of the 358 amu ion decreased substantially, with a concomitant increase of the 360 amu ion (Figure 6B). Following hydrolysis, the enrichment of the water in the hydrolysis medium with 18O was measured by GC/ MS to be 36 ( 3%. On the basis of the increase in the intensity of the 360 amu ion of the HmU peak and the decrease of the 358 amu ion, we calculate that the extent of the HmU enrichment with 18O is 37 ( 3%. As the degree of enrichment of 18O in the hydrolysis medium is the same as the degree of enrichment of the HmU following hydrolysis, we conclude that all of the HmU molecules present in the original oligonucleotide proceeded through nucleophilic attack at the methylene


LaFrancois et al.

Figure 5. Reaction pathways for HmU during acid hydrolysis. R is H, deoxyribose, or a deoxyribose monophosphate residue in DNA. R′ may be derived from any alcohol, sugar, or other carboxylic acid present in the hydrolysis solution.

DNA were prepared to which 5 × 10-11 mol of isotopically enriched HmdU was added. Following hydrolysis in 88% formic acid for 1 h at 150 °C, the formic acid was removed under reduced pressure and replaced with 200 µL of water. The samples were then heated at 60 °C for 30 min. The water was removed under reduced pressure, and the samples were then derivatized and analyzed by GC/MS. The mass spectrometer was operated in the SIM mode monitoring the 255, 358, and 362 amu ions. Based upon the ratio of the observed peaks at 358 and 362 amu, the amount of HmU present in the original sample was determined to be (6.3 ( 1.0) × 10-11 mol. A 100-µg sample of calf thymus DNA contains 9 × 10-8 mol of thymine, and therefore, the amount of HmU present in the calf thymus DNA was determined to be 70 ( 11 mol of HmU/105 mol of thymine.

Discussion

Figure 6. Mass spectral isotope cluster of the predominant fragment for silylated HmU. (A) Theoretical (left, narrow) and observed (right, wide) lines for silylated HmU fragment corresponding to the presence of heavier isotopes at natural abundance. (B) As above, except the observed intensities of the fragment lines correspond to silylated HmU generated following hydrolysis in 18O-enriched aqueous formic acid.

carbon and that the distribution of products derived from HmU depends on the relative proportions of molecules which can compete for reaction, including water, formic acid, or other molecules such as alcohols, sugars etc. (denoted by R′ in Figure 5) present in the hydrolysis medium. Regeneration of HmU and an Isotope Standard Allow Accurate HmU Quantification. Having established the chemical mechanisms by which HmU is derivatized during formic acid hydrolysis and having devised a strategy for the regeneration of HmU prior to GC/MS analysis, we conducted a series of experiments to measure the amount of HmU in calf thymus DNA. A series of six samples containing 100 µg of calf thymus

The data presented here confirms earlier reports that GC/MS can provide definitive identification of HmU in complex biological samples. Furthermore, the GC/MS method can be used to quantitatively determine HmU levels in a given sample. The problem associated with measurement of HmU in DNA, however, derives from the hydrolysis of the glycosidic bond prior to GC/MS analysis. Several previous studies have discussed the apparent degradation of HmU during acid hydrolysis; however, these reports have been largely empirical in nature, focusing on experimental parameters such as the formic acid concentration, temperature, and time of hydrolysis (28, 30, 35, 36). Some of these studies have reached conflicting conclusions as discussed below. We find that the major derivative of HmU following formic acid treatment is the formate ester (Figures 1 and 5). The formate ester derivative is also a substantial product formed by acid hydrolysis of HmdU as well as the HmU-containing oligonucleotide. Unfortunately, the formate ester is not sufficiently stable nor reproducibly formed to be used as a surrogate marker for measuring HmU levels in complex samples. Using 88% formic acid for hydrolysis, only 10% of the HmU present in the original sample survives hydrolysis and is therefore available for measurement by GC/MS.


In previous work reported by Dizdaroglu and coworkers, which utilized 88% formic acid, background HmU levels were reported as undetectable (25). More recently, when using 60% formic acid, that group reports increased HmU levels (30). Our results are in accord with these previous studies in that HmU recovery is greater when the sample is treated with 60% rather than 88% formic acid (Figure 2) if water regeneration is not performed. On the basis of our results, we estimate that previous studies, in which acid hydrolysis with 88% formic acid was utilized, may have underestimated the actual HmU levels by approximately 1 order of magnitude. The reactivity of HmU under acidic conditions as reported here is not unexpected. In 1959, Cline et al. demonstrated that, under acidic conditions, HmU could condense with carboxylic acids, alcohols, and amines to form esters, ethers, and amines, respectively (52). In 1966, Scheit argued the existence of a resonancestabilized methylene cation formed selectively by the hydroxymethyl group of HmU derivatives under acidic conditions (53). Indeed, the selective reactivity of the 5-hydroxymethyl group in acetic acid allowed the regiospecific acetylation of unprotected HmdU. Selective acetylation of the HmdU hydroxymethyl group provided a pathway for the efficient generation of an HmUphosphoramidite needed for synthesis of HmU-containing oligonucleotides using standard phosphoramidite chemistry (41). The inherent reactivity of HmU derivatives under acidic conditions is an obvious problem which must be faced in order to measure HmU levels by GC/MS. The nature of the HmU derivatives formed would be dependent upon the composition of the original sample. Formic acid treatment of the HmU free base would give predominantly the formate ester; however, the generation of free sugars and other molecules upon acid hydrolysis of DNA would be expected to result in a complex array of products. The use of the synthetic HmU-containing oligonucleotide reported here provided a sufficient quantity of a consistent sample containing a known amount of HmU for an examination of HmU recovery under a multitude of conditions. Inspection of the mass spectrum of HmU derived from hydrolysis of the HmU-containing oligonucleotide in 60% formic acid containing 18O-enriched water revealed the incorporation of 18O into HmU (Figure 6). The incorporation of 18O from water into HmU under acidic conditions is consistent with nucleophilic attack at the methylene carbon as proposed by Scheit (53). Furthermore, as the degree of incorporation of 18O into the HmU is the same as the level of 18O present in the water of the hydrolysis medium, we conclude that all of the HmU molecules measured undergo nucleophilic attack (Figure 5). The methylene carbon is subject to attack by nucleophiles which are present in the hydrolysis medium, including water, formate, and other molecules such as alcohols and sugars (R′, Figure 5). The conditions which maximize recovery of HmU for GC/MS analysis from the synthetic oligonucleotides utilized here call for hydrolysis in 88% formic acid followed by water treatment at 60 °C for 30 min. This result may at first seem contradictory to studies which show greater HmU levels when DNA is hydrolyzed in 60% rather than 88% formic acid. The explanation, however, proceeds from the reaction scheme outlined in


Figure 5. In 88% formic acid, a greater amount of the formate ester would form, at the expense of other derivatives such as ethers. Ethers and non-formate esters would not hydrolyze as easily during the aqueous regeneration of HmU. Indeed, the formic acid could be considered a transient protecting group with hydrolysis prior to silylation allowing recovery of approximately 70% of the HmU from the synthetic oligonucleotide studied here. Previously, Djuric et al. discussed the effect of formic acid treatment on the GC/MS determination of HmU (28). They found that the HmU peak area could drop by more than 20-fold following heating in formic acid. Although the use of an isotopically enriched internal standard of HmU could be used to adjust for acid-induced loss of HmU, the resulting peak areas were reduced to the limit of detection. As an alternative, it was proposed that enzymatic hydrolysis could be substituted for acid hydrolysis. The liberated deoxynucleosides were isolated with a reverse-phase cartridge and converted directly to the silylated free base using standard silylation conditions. We have reported that an oligonucleotide containing HmU in the sequence GpHmUpG was completely digested using nuclease P1 and bacterial alkaline phosphatase (41). In contrast, we find here that the Tp(HmU) dinucleotide is somewhat resistant to cleavage by nuclease P1. Under conditions which result in the complete digestion of a dodecamer of the same sequence containing only normal bases, only 70% of the HmU residues are recovered as HmdU (Figure 3B). It has been established by Falcone and Box (55) that nuclease P1 digestion rates vary considerably among the possible dinucleotide sequences and that the TpT step is the slowest. Oxidized bases further reduce the rate of enzymatic cleavage of some dinucleotides. The reduced recovery of HmdU in the sequence reported here likely reflects the sequence dependence of the nuclease reaction. The recovery of HmdU from oxidized DNA following enzymatic digestion is less efficient than for normal DNA bases in some sequences, and the recovery appears to be sequencedependent. In our study, the direct conversion of HmdU to silylated HmU, without intervening acid hydrolysis, is approximately 36% (Figure 2), and the overall recovery of HmU would be somewhat less if the enzymatic digestion was incomplete. Consistent with these findings, we report that the background level of HmU in calf thymus DNA is approximately 70 HmU residues/105 thymine residues, whereas Djuric et al. reported a value of 33 HmU residues/105 thymine residues (28). Alternatively, formic acid hydrolysis followed by aqueous treatment, as described here, allows recovery of approximately 70% of the HmU present in the original oligonucleotide. The results reported here suggest that formic acid hydrolysis in 88% formic acid, followed by hydrolysis of the formate ester, would be faster and more reliable than the enzymatic method. Recently, Douki et al. examined conditions for the measurement of HmU by GC/MS following irradiation of DNA in solution (36). They concluded that 88% formic acid was superior to 60% formic acid for the recovery of HmU. This is in contrast to the results of Dizdaroglu and co-workers who abandoned hydrolysis in 88% formic acid in favor of 60% formic acid (30). The explanation provided by Douki et al. was that the degree of dissocia-


tion of the formic acid was higher in the less concentrated solution (60% formic acid) and that the dissociated form of the acid was resulting in undescribed degradation of the HmU. The greater amount of measured HmU following hydrolysis in 88% formic acid reported by Douki et al. is apparently contrary to our results as reported in Figure 2, in which we observed that more HmU was recovered from the synthetic HmU-containing oligonucleotide using 60% formic acid (with no water recovery), as well as the results of Dizdaroglu and co-workers. The resolution of this seeming contradiction probably lies in the fact that Douki et al. introduced a “cleanup” step following formic acid hydrolysis and prior to silylation. This cleanup step involved aqueous extraction in calcium carbonate followed by HPLC purification. We suggest that the hydrolysis of DNA in 88% formic acid generates more of the formate ester. The aqueous cleanup step introduced by Douki et al. inadvertently hydrolyzed the formate ester of HmU back to HmU prior to silylation. If this explanation is indeed correct, then we are in accord with Douki et al., that hydrolysis in 88% formic acid, followed by aqueous treatment, maximizes the amount of HmU recovered. One of the original aims of this study was to determine why HmU levels as measured by GC/MS were substantially lower than those measured by other methods such as postlabeling. It is apparent from the results of this study that acid hydrolysis of HmU-containing DNA generates substantial amounts of the formate ester as well as other potential products. If one uses an internal standard which does not undergo similar reactivity, such as 6-azathymine, the observed levels would understandably represent an underestimate of the actual amount of HmU present in the original DNA sample. To measure HmU by GC/MS following formic acid hydrolysis, two additional factors must be considered: (1) hydrolysis of the formate ester of HmU must be induced prior to silylation in order to maximize the resulting HmU signal, and (2) an isotopically enriched internal standard must be used to control for the variable reactivity of HmU under hydrolysis conditions. While isotopically enriched HmU would be better than 6-azathymine, labeled HmdU would be the best as it would allow for control of the cleavage of the glycosidic bond as well. Having incorporated the above considerations into a measurement of HmU levels in synthetic oligonucleotides containing HmU, as well as commercially available calf thymus DNA, we conclude that some of the previous studies using the GC/MS method may have underestimated HmU levels by an order of magnitude. Therefore, the higher HmU levels reported by Frenkel et al., using postlabeling methods, probably more accurately reflect background levels of HmU in DNA isolated from cells (27). Notably, Frenkel et al. reported that the background levels of HmU in some systems were equal to or greater than the levels of 8-oxoguanine (27). Akman et al. measured the levels of several oxidized base in human chromatin using hydrolysis in 60% formic acid (29). The ratio of 8-oxoguanine to HmU was approximately 4. Here, we demonstrated that the yield of HmU from a synthetic oligonucleotide using 60% formic acid with no water recovery was approximately 35%. It is therefore likely that background levels of HmU and 8-oxoguanine are quite similar in human chromatin.

LaFrancois et al.

In conclusion, we suggest that synthetic oligonucleotides and isotopically enriched deoxynucleosides be used to validate GC/MS methods for the measurement of modified bases, especially in cases where sample preparation might conceivably generate artifacts. Synthetic oligonucleotide standards would also prove to be valuable in comparing various analytical methods. In particular, we demonstrate that previous sample preparation methods for the measurement of HmU levels by GC/MS have resulted in significant underestimates. It is likely that HmU is as abundant at background levels as the more widely discussed 8-oxoguanine.

Acknowledgment. This work was supported in part by the National Institutes of Health Grants GM 50351, GM41336, and CA33572.

References (1) Imlay, J. A., and Linn, S. (1988) DNA damage and oxygen radical toxicity. Science 240, 1302-1309. (2) Loeb, L. A. (1989) Endogenous carcinogenesis: molecular oncology into the twenty-first century - Presidential address. Cancer Res. 49, 5489-5496. (3) Mullaart, E., Lohman, P. H. M., Berends, F., and Vijg, J. (1990) DNA damage metabolism and aging. Mutat. Res. 237, 189-210. (4) Ames, B. N., Shigenaga, M. K., and Hagen, T. M. (1993) Oxidants, antioxidants, and the degenerative diseases of aging. Proc. Natl. Acad. Sci. U.S.A. 90, 7915-7922. (5) Ames, B. N., Gold, L. S., and Willett, W. C. (1995) The causes and prevention of cancer. Proc. Natl. Acad. Sci. U.S.A. 92, 52585265. (6) Dizdaroglu, M. (1984) The use of capillary gas chromatographymass spectrometry for identification of radiation-induced DNA base damage and DNA base-amino acid cross-links. J. Chromatogr. 295, 103-121. (7) Teebor, G. W., Frenkel, K., and Goldstein, M. S. (1984) Ionizing radiation and tritium transmutation both cause formation of 5-hydroxymethyl-2′-deoxyuridine in cellular DNA. Proc. Natl. Acad. Sci. U.S.A. 81, 318-321. (8) Hollstein, M. C., Brooks, P., Linn, S., and Ames, B. N. (1984) Hydroxymethyluracil DNA glycosylase in mammalian cells. Proc. Natl. Acad. Sci. U.S.A. 81, 4003-4007. (9) Frenkel, K., Cummings, A., Solomon, J., Cadet, J., Steinberg, J. J., and Teebor, G. W. (1985) Quantitative determination of the 5-(hydroxymethyl)uracil moiety in the DNA of gamma-irradiated cells. Biochemistry 24, 4527-4533. (10) Cannon-Carlson, S. V., Gokhale, H., and Teebor, G. W. (1989) Purification and characterization of 5-hydroxymethyluracil-DNA glycosylase from calf thymus. J. Biol. Chem. 264, 13306-13312. (11) Frenkel, K., and Klein, C. B. (1993) Methods used for analysis of “environmentally” damaged nucleic acids. J. Chromatogr. 618, 289-314. (12) Mellac, S., Fazakerley, G. V., and Sowers, L. C. (1993) Structures of base pairs with 5-(hydroxymethyl)-2′-deoxyuridine in DNA determined by NMR spectroscopy. Biochemistry 32, 7779-7786. (13) Grollman, A. P., and Moriya, M. (1993) Mutagenesis by 8-oxoguanine: an enemy within. TIG 9, 246-249. (14) Wallace, S. S. (1994) DNA damages processed by base excision repair: biological consequences. Int. J Radiat. Biol. 66, 579-589. (15) Feig, D. I., Sowers, L. C., and Loeb, L. A. (1994) Reverse chemical mutagenesis: Identification of the mutagenic lesions resulting from reactive oxygen species-mediated damage to DNA. Proc. Natl. Acad. Sci. U.S.A. 91, 6609-6613. (16) Vilpo, J. A., and Vilpo, L. M. (1995) Restriction, methylation and ligation of 5-hydroxymethyluracil-containing DNA. Mutat. Res. 316, 123-131. (17) Fazakerley, G. V., and Boulard, Y. (1995) DNA mismatches and modified bases. Methods Enzymol. 261, 145-163. (18) Henle, E., and Linn, S. (1997) Formation, prevention and repair of DNA damage by iron/hydrogen peroxide. J. Biol. Chem. 272, 19095-19098. (19) Djuric, Z., Heilbrun, L. K., Reading, B. A., Boomer, A., Valeriote, F. A., and Martino, S. (1991) Effects of a low-fat diet on levels of oxidative damage to DNA in human peripheral nucleated blood cells. J. Natl. Cancer Inst. 83, 766-769. (20) Frenkel, K., Karkoszka, J., Kim, E., and Taioli, E. (1993) Recognition of oxidized DNA bases by sera of patients with inflammatory diseases. Free Radical Biol. Med. 14, 483-494.

Quantification of HmU in DNA by GC/MS (21) Djuric, Z., Heilbrun, L. K., Simon, M. S., Smith, D., Luongo, D. A., LoRusso, P. M., and Martino, S. (1996) Levels of 5-hydroxymethyl-2′-deoxyuridine in DNA from blood as a marker of breast cancer. Cancer 77, 691-696. (22) Fuciarelli, A., Wegher, B. J., Gajewski, E., Dizdaroglu, M., and Blakely, W. F. (1989) Quantitative measurement of radiationinduced base products in DNA using gas chromatography-mass spectrometry. Radiat. Res. 119, 219-231. (23) Kresbach, G. M., Itani, M., Saha, M., Rogers, E. J., Vouros, P., and Giese, R. W. (1989) Ester and related derivatives of ring N-pentafluorobenzylated 5-hydroxymethyluracil. Hydrolytic stability, mass spectral properties, and trace detection by gas chromatogaphy-electron capture detection, gas chromatographyelectron-capture negative ion mass spectrometry, and movingbelt liquid chromatography-electron-capture negative ion mass spectrometry. J. Chromatogr. 476, 423-438. (24) Annan, R. S., Kresbach, G. M., Giese, R. W., and Vouros, P. (1989) Trace detection of modified DNA bases via moving-belt liquid chromatography-mass spectrometry using electrophoric derivitization and negative chemical ionization. J. Chromatogr. 465, 285-296. (25) Gajewski, E., Rao, G., Nackerdien, Z., and Dizdaroglu, M. (1990) Modification of DNA bases in mammalian chromatin by radiationgenerated free radicals. Biochemistry 29, 7876-7882. (26) Cattley, R. C., Dietze, S. R., Richardson, F. C., and Popp, J. A. (1990) Detection of 5-hydroxymethyl-2′-deoxyuridine in DNA from calf thymus, Hela cells, and rat liver nuclei. Nucleosides Nucleotides 9, 179-187. (27) Frenkel, K., Zhong, Z., Wei, H., Karkoszka, J., Patel, U., Rashid, K., Georgescu, M., and Solomon, J. J. (1991) Quantitative highperformance liquid chromatography analysis of DNA oxidized in vitro and in vivo. Anal. Biochem. 196, 126-136. (28) Djuric, Z., Luongo, D. A., and Harper, D. A. (1991) Quantitation of 5-(hydroxymethyl)uracil in DNA by gas chromatography with mass spectral detection. Chem. Res. Toxicol. 4, 687-691. (29) Akman, S. A., Doroshow, J. H., Burke, T. G., and Dizdaroglu, M. (1992) DNA base modifications induced in isolated human chromatin by NADH dehydrogenase-catalyzed reduction of doxorubicin. Biochemistry 31, 3500-3506. (30) Nackerdien, Z., Olinski, R., and Dizdaroglu, M. (1992) DNA base damage in chromatin of gamma-irradiated cultured human cells. Free Radical Res. Commun. 16, 259-273. (31) Claycamp, H. G. (1992) Phenol sensitization of DNA to subsequent oxidative damage in 8-hydroxyguanine assays. Carcinogenesis 13, 1289-1292. (32) Cadet, J., Incardona, M.-F., Odin, F., Molko, D., Mouret, J.-F., Poiverelli, M., Faure, H., Ducros, V., Tripier, M., and Favier, A. (1993) Measurement of oxidative base damage to DNA by using HPLC-32P-postlabeling and GC/MS-selective ion monitoring assays. IARC 124, 271-276. (33) Hamberg, M., and Zhang, L.-Y. (1995) Quantitative determination of 8-hydroxyguanine and guanine by isotope dilution mass spectrometry. Anal. Biochem. 229, 336-344. (34) Ravanat, J.-L., Turesky, R. J., Gremaud, E., Trudel, L. J., and Stadler, R. H. (1995) Determination of 8-oxoguanine in DNA by gas chromatography-mass spectrometry and HPLC-electrochemical detection: overestimation of the background level of the oxidized base by the gas chromatography-mass spectrometry assay. Chem. Res. Toxicol. 8, 1039-1045. (35) Swarts, S. G., Smith, G. S., Miao, L., and Wheeler, K. T. (1996) Effects of formic acid hydrolysis on the quantitative analysis of radiation-induced DNA base damage products assayed by gas chromatography. Radiat. Environ. Biophys. 35, 41-53. (36) Douki, T., Delatour, T., Paganon, F., and Cadet, J., (1996) Measurement of oxidative damage at pyrimidine bases in gammairradiated DNA. Chem. Res. Toxicol. 9, 1145-1151.

Chem. Res. Toxicol., Vol. 11, No. 7, 1998 793 (37) Herbert, K. E., Evans, M. D., Finnegan, M. T. V., Farooq, S., Mistry, N., Podmore, I. D., Farmer, P., and Lunec, J. (1996) A novel HPLC procedure for the analysis of 8-oxoguanine in DNA. Free Radical Biol. Med. 20, 467-473. (38) Collins, A., Cadet, J., Epe, B., and Gedik, C. (1997) Problems in the measurement of 8-oxoguanine in human DNA. Report of a workshop, DNA oxidation, held in Aberdeen, UK, 19-21 January, 1997. Carcinogenesis 18, 1833-1836. (39) Bodepudi, V., Iden, C. R., and Johnson, F. (1991) An improved method for the preparation of the phosphoramidites of modified 2′-deoxynucleotides: incorporation of 8-oxo-2′-deoxy-7H-guanosine into synthetic oligomers. Nucleosides Nucleotides 10, 755761. (40) Beaucage, S., and Iyer, R. (1993) The synthesis of modified oligonucleotides by the phosphoramidite approach and their applications. Tetrahedron 49, 6123-6194. (41) Sowers, L. C., and Beardsley, G. P (1993) Synthesis of oligonucleotides containing 5-(hydroxymethyl)-2′-deoxyuridine at defined sites. J. Org. Chem. 58, 1664-1665. (42) Matray, T. J., Haxton, K. J., and Greenberg, M. M. (1995) The effects of the ring fragmentation product of thymidine C5-hydrate on phosphodiesterases and Klenow (exo-) fragment. Nucleic Acids Res. 23, 4642-4648. (43) Tardy-Planechaud, S., Fujimoto, J., Lin, S. S., and Sowers, L. C. (1997) Solid phase synthesis and restriction endonuclease cleavage of oligodeoxynucleotides containing 5-(hydroxymethyl)-cytosine. Nucleic Acids Res. 25, 553-558. (44) Fujimoto, J., Tran, L., and Sowers, L. C. (1997) Synthesis and cleavage of oligodeoxynucleotides containing a 5-hydroxyuracil residue at a defined site. Chem. Res. Toxicol. 10, 1254-1258. (45) LaFrancois, C. J., Fujimoto, J., and Sowers, L. C. (1998) Synthesis and characterization of isotopically enriched pyrimidine deoxynucleoside oxidation damage products. Chem. Res. Toxicol. 11, 75-83. (46) Dickerson, R. E., and Drew, H. R. (1981) Structure of a B-DNA dodecamer II. Influence of base sequence on helix structure. J. Mol. Biol. 149, 761-786. (47) Cantor, C. R., Warshaw, M. M., and Shapiro, H. (1970) Oligonucleotide interactions. III. Circular dichroism studies of the conformation of deoxyoligonucleotides. Biopolymers 9, 1059-1077. (48) Shiau, G. T., Schinazi, R. F., Chen, M. S., and Prusoff, W. H. (1980) Synthesis and biological activities of 5-(hydroxymethyl, azidomethyl, or aminomethyl)-2′-deoxyuridine and related 5′substituted analogues. J. Med. Chem. 23, 127-133. (49) Handbook of Biochemistry and Molecular Biology, 3rd. ed. (1975) (Fasman, G. D., Ed.) Nucleic Acids Vol. 1, CRC Press Inc., Cleveland, OH. (50) Chang, P. K. (1958) Synthesis of some 5-alkyl-6-azauracils. J. Org. Chem. 23, 1951-1953. (51) Sambrook, J., Fritsch, E. F., and Maniatis, T. (1989) Molecular Cloning, a Laboratory Manual, 2nd ed., p C.1, Cold Spring Harbor Laboratory Press, Plainview, NY. (52) Cline, R. E., Fink, R. M., and Fink, K. (1959) Synthesis of 5-substituted pyrimidines via formaldehyde addition. J. Am. Chem. Soc. 81, 2521-2527. (53) Scheit, K. H. (1966) Die synthese der 5′-diphosphate von 5-methyluridin, 5-hydroxymethyl-uridin und 3,4-dimethyl-uridin. (Synthesis of the 5′-diphosphate of 5-methyluridine, 5-(hydroxymethyl)uridine and 3,5-dimethyluridine.) Chem. Ber. 99, 3884-3890. (54) McCloskey, J. A. (1990) Calculation of isotopic abundance distributions. Methods Enzymol. 193, 882-886. (55) Falcone, J. M., and Box, H. C. (1997) Selective hydrolysis of damaged DNA by nuclease P1. Biochim. Biophys. Acta 1337, 267-275.

TX970233C

Mass Spectrometry - American

Recommend Documents