- Email: [email protected]
Purification and Recovery of Bulky Hydrophobic DNA Adducts
Analytical Biochemistry 272, 100 –106 (1999) Article ID abio.1999.4161, available online at http://www.idealibrary.com on
Purification and Recovery of Bulky Hydrophobic DNA Adducts Curt B. Norwood U. S. EPA AED, 27 Tarzwell Drive, Narragansett, Rhode Island 02882
Received March 19, 1999
For many years 32P postlabeling has detected DNA adducts at very low levels and yet has not been able to identify unknown adducts. Mass spectrometry offers substantially improved identification powers, albeit at some loss in detection limits. With this ultimate utilization of mass spectrometry in mind, the current research presents a new method to quantitatively purify bulky hydrophobic DNA adducts at levels that are pertinent to ongoing DNA adduct research in human health and environmental fields. This method was demonstrated with benzo[a]pyrene adducts. Purification was accomplished with the use of small columns (7.5-mm frits) with an 11 mg bed of polystyrene– divinlybenzene beads which retained the adducts while permitting the nonadducted nucleotides to be washed out with water. Subsequently, the adducts were eluted with 50% MeOH and the sample was reduced in volume in an evacuated centrifuge. Purification was demonstrated at adduct levels ranging from 4 adducts in 10 6 nonadducted nucleotides to 4 in 10 8. For these levels, analyses by capillary electrophoresis with sample stacking and UV detection determined that recoveries ranged from 91 to 54%, respectively. The adduct quantities isolated should be sufficient to allow the use of current MS capabilities that are linked on-line to separation methodologies such as capillary electrophoresis, capillary electrochromatography, and high-pressure liquid chromatography.
DNA adducts, the covalent modifications of DNA, continue to be of major importance, as evidenced by the many hundreds of research articles devoted to them every year. Initially, the focus on DNA adducts involved links to cancer, especially in humans. More recently, DNA adducts have been viewed as biomarkers of both exposure and effects, not only in humans (1), but also in other organisms (2– 8). Considerable research effort has addressed what have come to be 100
known as endogenous adducts, which in part may be due to lipid peroxidation (9). Because of their genotoxic properties, the importance of adducts to disease states other than carcinogenesis is now also recognized (10 – 12). Furthermore, DNA adducts are increasingly involved in risk evaluations (13–16) and the development of intervention strategies (17). Much of the current DNA adduct research relies on the 32 P-postlabeling analytical methodology (P32) 1 with autoradiography. P32 is especially appropriate in those situations where the genotoxic compounds are known. Furthermore, P32 is extremely sensitive and requires only very small amounts of DNA. However, P32 suffers from a poor ability to actually identify unknowns, not only detect them. Traditionally P32 has been used with autoradiography to reveal the migration pattern of various components on TLC plates. P32 has also been used with HPLC and on-line 32P (18) detection. In both cases any identification capabilities are limited to comparisons to known components that share similar elution characteristics. This and other shortcomings of P32 (19) are widely recognized and many pleas have been made for enhanced analytical techniques to overcome this deficiency (16, 20). A similar inability to identify unknown DNA adducts is shared by approaches that rely on fluorescence (21, 22), phosphorescence (23), immunochemical (24 –26), and electrochemical (27) detection. Both NMR and MS offer vastly improved identification abilities. However, nuclear magnetic resonance suffers from a requirement for considerably more sample amount than does MS. MS is especially useful if coupled with a separation scheme such as TLC (28), HPLC (29, 30), capillary electrophoresis (CE) (31–33), 1
Abbreviations used: 50% MeOH, methanol:deionized water(1:1, v/v); BaPdGMP, benzo[a]pyrene-29-deoxyguanosine 59-phosphate; CE, capillary electrophoresis; CEC, capillary electrochromatography; DI-H 2O, deionized water; P32, 32P-postlabeling; %rec, percent recovery. 0003-2697/99
PURIFICATION AND RECOVERY OF BULKY DNA ADDUCTS
capillary electrochromatography (CEC) (34), or gas chromatography (35), although this latter technique usually requires derivatization, which introduces other analysis problems (36). However, when MS monitors only selected ions or reactions, although its sensitivity is considerably enhanced, its identification powers are sharply reduced. Coupling of the separation scheme on-line with the detector provides greater ease of operation, better quantitation, less sample handling, and less chance of contamination. This article presents a methodology to purify bulky non-polar-adducted DNA mononucleotides (59-monophosphates) by removing a great excess of nonadducted DNA mononucleotides. The levels of adduction of the benzo[a]pyrene adducts ranged from 4 in 10 6 to 4 in 10 8 (4 adducts for 10 n nonadducted nucleotides). Although the results presented here relied on CE separations coupled, on line, with ultraviolet detection, many alternative separation schemes and detectors could have been utilized. The actual amounts of DNA should be sufficient to permit MS analyses and thus enable the identification of unknown adducts. MATERIALS AND METHODS
Two batches of DNA were prepared and hydrolyzed under sterile conditions. The DNA hydrolysis solution for salmon testes DNA (Sigma, St. Louis, MO) consisted of DNA at 3 mg/ml, 50 mM Tris, pH 8.6 (Sigma), 5 mM CaCl 2 and 4 mM MgCl 2. To initiate hydrolysis to generate 59-mononucleotides, two enzymes were added: SVPDEase I (Sigma P7027 0.002 U/ml of DNA solution) and DNase I (Sigma 3 kU/ml). This batch was incubated at 37°C for 4 days, The second batch of DNA contained calf thymus DNA (Sigma) that was again at 3 mg/ml in a hydrolysis solution that contained buffer and salt concentrations identical to those above. However, enzyme concentrations were lower— 0.0005 U and 0.8 kU/ml, for SVPDEase I and DNase I, respectively. This batch was incubated at 37°C for 8 days. The mononucleotide solutions were kept frozen except for brief thawing periods to permit removal of portions to be used in subsequent analyses. To produce benzo[a]pyrene–29-deoxyguanosine 59phosphate adducts (BaPdGMP), benzo[a]pyrene– diolepoxide was reacted with calf thymus DNA, enzymatically digested to 59-monophosphates, passed through a membrane that contained polystyrene– divinylbenzene beads to purify the BaPdGMP, and concentrated in a vacuum centrifuge, as previously described (37). A standard solution of this was obtained by diluting an aliquot of this solution by a factor of ;400 in methanol: deionized water (1:1, v/v) (50% MeOH). CE was performed with sodium carbonate/bicarbonate buffer (15 mM, pH 8.6) (Sigma) in uncoated fused silica capillary columns (780 –900 mm 3 0.375 mm o.d.
101
3 0.075 mm i.d.) (Polymicro Technologies, Phoenix, AZ). The column was maintained at ambient temperature. The CE system, which was assembled from individually purchased components, was operated at 15–17 kV provided by a Bertan (Hicksville, NY) Series 230 polarity-reversing high-voltage power supply. These conditions resulted in ;17 mA. CE analyses involved sample stacking, as previously described (37). Briefly, the column was approximately 21 filled (from the head) with sample (volume approximately 1.7 ml). High voltage was applied with polarity such that endoosmotic flow was toward the column head. When the monitored current reached ;16.7 mA, the polarity was reversed so that the endoosmotic flow carried components in the normal direction—toward the tail of the column. On-column UV absorbance was monitored with a Linear (Reno, NV) Model 200 UVIS detector located 50 mm from the tail of the column. Data were collected with a Crystal CE system (ATI, Franklin, MA). Purification of adducts was accomplished via liquid chromatography with columns whose design and use is described in the next five paragraphs. Empty column preparation. Empty 1-ml polypropylene columns (Quiagen, Chatsworth, CA) initially had only the bottom frit (proprietary material) in place. These 7.5-mm i.d. columns actually have a volume of ;6 ml if filled to the top of the flared out upper region. For the procedures described below, nylon gloves were worn and all utensils were very clean. The tapered column bottom was removed at a level ;1 mm below the bottom frit. The bottom 7.5-mm frit was pushed out the top with a rod and, along with the top frit, was placed in a container of methanol. The bottom of the empty column was swaged a bit smaller by firmly pressing on a microcentrifuge tube and leaving the assembly for at least a few hours. After removal of the swaging tube, the column was held with forceps and rinsed sequentially with hot tap water, deionized water (DI-H 2O), and methanol. Column bed preparation. A frit was removed from the methanol and very firmly pressed with a rod to the bottom of the column. Eighty ml(11 mg) of dry POROS 50 RI beads (polystyrene– divinylbenzene)(Perseptive Biosystems, Cambridge, MA) in a pipettor with a slightly enlarged tip orifice was dispensed into the column. A top frit was inserted into the column and quite firmly pressed into place. The resultant POROS bed thickness was ;1.5 mm. Methanol (3 ml) was added above the top frit and was allowed to drip via gravity. When the volume remaining above the top frit was ;1 ml, DI-H 2O was added to fill the column to the top and thus bring the volume to ;6 ml. The column was again allowed to drip under gravity until the remaining volume was ;3 ml. The column was rinsed out
102
CURT B. NORWOOD
with DI-H 2O to remove any POROS remaining above the top frit and refilled with 6 ml of DI-H 2O. The column was again allowed to drip until ;1 ml of DIH 2O remained. The column was again filled with DIH 2O and the drip rate was measured. Rates of from 3 to 7 s per drop were considered acceptable. Slower drip rates required removal of the frits and reassembly of the column. Faster drip rates could be slowed by either switching back to methanol and returning to DI-H 2O, or by pressing down more firmly on the top frit. Sample loading and rinse to remove nonadducted nucleotides. An acceptable column was emptied and then prefilled with between 0.3 and 1.0 ml of DI-H 2O (depending on the planned sample volume) and placed nearly horizontally into a holder. The appropriate volumes of DNA mononucleotides and standard BaPdGMP were added. The column was returned to an upright position to permit gravity dripping that continued until there was very little liquid remaining above the top frit. DI-H 2O (0.5 ml) was added to the column and again allowed to pass down through the column bed until the top frit was nearly dry. This 0.5-ml rinsing procedure was repeated two more times. To further wash out the nonadducted nucleotides, the top of the column was grasped with locking pliers and 0.8 ml of DI-H 2O was added to the column. Two drops was allowed to elute from the column, after which time the column was filled to the top with DI-H 2O and immediately inverted to pour out the liquid contents. The bottom of the now upside-down column was rinsed off with DI-H 2O. The column was righted and this filling and rinsing subprocedure was repeated four more times. The entire washing procedure just described was performed a total of six times. Adduct elution and collection. The column was suspended vertically just above a 0.6-ml preweighed, silanized microcentrifuge tube (PGC Scientifics, Gaithersburg, MD) that had been vortexed/rinsed with hot water, DI-H 2O, and methanol. Approximately 350 ml of 50% MeOH was added to the column to elute the adducted nucleotides. After a few drops had been collected, an additional 350 ml of 50% MeOH was added above the column. A total of ;570 ml was collected. Volume reduction. The microcentrifuge tube was placed in a clean vacuum centrifugation system (Savant, San Diego, CA). Slow evacuation conditions (considerable restriction in the line between the centrifuge and the vacuum pump) for 1 h resulted in a volume of ;150 ml. Subsequent fast evacuation for an additional hour resulted in a volume of ;25 ml. The exact volume was determined by weighing the tube and contents. An equal volume of methanol was added and the microcentrifuge tube was vortexed for 1 min prior to centrifugation and subsequent CE analysis. In some cases, to achieve a smaller final volume, this ;50-ml volume was
TABLE 1
Experimental Design Level of adduction
Adduct volume (ml)
DNA volume (ml)
Number of samples
4 in 10 6 4 in 10 7 4 in 10 8
25 25 4.17
120 1200 2000
8 4 3
further reduced in the vacuum concentrator to ;5 ml. An equal volume of MeOH was added prior to vortexing, centrifugation, and CE analysis. The two types of calculations used in this research, level of adduction and percent recovery (%rec), were based on data generated by CE analyses (with sample stacking) where the solvent was 50% MeOH. The first of these, level of adduction, required a determination of the proper volumes of BaPdGMP and mononucleotide solutions that should be mixed to achieve the desired level of adduction. This determination required a calculation of the moles of BaPdGMP in the standard solution and the moles of dGMP in an equal volume of the mononucleotides solution that had been diluted 3000-fold in 50% MeOH. To enable this calculation, separate CE analyses were performed on equal volume injections of each of these solutions. Figure 1 shows the resultant electropherograms with UV detection at 279 nm. To calculate the relative number of moles in the BaPdGMP and the undiluted mononucleotides solutions, the observed ratio of the peak areas of the BaPdGMP in the standard and the dGMP in the diluted mononucleotides solution was multiplied by the reciprocals of their respective extinction coefficients of 41,000 and 7770, the reciprocal of the dilution factor, the known value of 0.21 for the mole fraction of dGMP in calf thymus DNA, and the reciprocal of the observed retention time ratio for the two peaks (to correct for differences in ion mobility). This calculation showed that there were 1.8 adducts for every 10 5 nucleotides in equal volumes of the BaPdGMP and the undiluted DNA mononucleotides solution. If 25 ml of the BaPdGMP were mixed with 120 ml of the DNA mononucleotides, the resultant solution would contain ;4 adducts per 10 6 nonadducted nucleotides. This calculation does not require knowledge of the precise concentration of either solution. However, those concentrations are ;90 pg/ml and 3 mg/ml, respectively. For this study, samples were developed by mixing known quantities of the standard with known amounts of the DNA mononucleotides solution to achieve various levels of adduction. Table 1 details these volumes, the level of adduction, and the number of samples analyzed. Each sample was then treated with the above purification and concentration procedures. The second calculation type, %rec, again was based on sep-
PURIFICATION AND RECOVERY OF BULKY DNA ADDUCTS
103
arate, equal volume, sample stacked, CE analyses of the BaPdGMP standard, and the sample. The BaPdGMP peak area at 344 nm for the standard was compared with that same peak area for a sample. Factoring in the initial volume of standard added and the final sample volume allowed a simple determination of the %rec. RESULTS AND DISCUSSION
This study presents results of an ongoing effort to purify unknown hydrophobic DNA adducts at levels of adduction in the low 10 28 range (or lower) in quantities that should be sufficient to permit their identification and quantitation by MS. The starting amounts of DNA and the final sample volumes have been selected so that at the lowest levels of adduction ;20% of the sample could be injected for separation of the components by the chosen method and detection by MS. This study, however, relies on a UV detector whose identification powers are vastly inferior to that of MS, but nonetheless sufficient for methods development with known compounds. Many of the individual steps detailed under Materials and Methods were instituted in response to specific problems that had been previously noted during the methods development. For example, gravity induced flow starts easily in columns assembled with frits that had been held in MeOH, but not with dry frits. Swaging of the column bottom was necessary to greatly reduce the flow around, rather than through, the bottom frit and thus reduce and lend reproducibility to the total flow rate. In addition, the unwanted retention of the nonadducted nucleotides was drastically reduced if the bottom of the column had been removed and the volume immediately below the bottom frit was flushed copiously during the rinsing procedures. Locking pliers were used to securely hold the column because if it had been dropped even a few centimeters, the column bed would have developed areas of high flow and the %rec would have suffered dramatically. The fixed wavelength electropherograms in Fig. 1 were recorded at 279 and 344 nm because of the local maxima in the UV absorption spectrum of BaPdGMP at those wavelengths. The mononucleotides do not absorb at 344 nm. Figure 1 clearly shows that the BaPdGMP elutes earlier than any of the four nucleotides. This time separation is important because of the great excess of mononucleotides in the actual samples. In fact, for the levels of adduction in this study, the samples were formulated with between 15,000 and 1,500,000 times as much of the nucleotides relative to the BaPdGMP as those demonstrated in Fig. 1. Figure 2 displays electropherograms at the same two wavelengths (279 and 344 nm) for samples that originally were at the three levels of adduction and subse-
FIG. 1. Electropherograms for sample-stacked injections into 80cm 3 75-mm i.d. columns with 15 mM sodium carbonate/bicarbonate buffer, pH 9.6, operated at 17 kV (17 mA). (A) Electropherogram at 279 nm of a 3000-fold dilution of calf thymus DNA hydrolyzed to its 29-deoxy-59-monophosphate nucleotides (A, C, T, and G) (approx 1.7 ng of DNA). (B and C) Electropherograms at 279 and 344 nm, respectively, of 150 pg of the BaPdGMP adduct (B).
quently purified by the methods of this study. The 279-nm electropherograms show the limited amounts of nonadducted mononucleotides remaining in the sample and by inference thus indicate the tremendous extent to which these samples have been purified. Even for the 10 28 sample, it is clear that far greater than
104
CURT B. NORWOOD
FIG. 2. Electropherograms of samples of DNA dosed at various levels of adduction and purified by the methods detailed in this article: (A–C) at 279 nm and (D–F) at 344 nm: (A and D) at 4 in 10 6, (B and E) at 4 in 10 7, and (C and F) at 4 in 10 8. Other electrophoresis conditions were the same as in Fig. 1.
99.99% of the nonadducted mononucleotides have been removed. To perform the %rec calculations, the 344-nm electropherogram peak was used because interferences with the unknown components are more problematic in the 279-nm electropherogram, especially at the lowest level of adduction (Fig. 2C). For each of the three levels of adduction in this study, Table 2 lists the range of final sample volumes and the mean %rec 6 standard deviation. In all cases
the final solvent was 50% MeOH and the volumes were determined by weight. The %rec calculations were in part based on the 344-nm CE peak area for the standard BaPdGMP. An injection volume of ;1.7 ml of this standard contained ;150 pg, which was well above the limit of detection (Fig. 1C). None of the %rec values exceeded 100. Each of the %rec determinations was based on two CE analyses of the sample and at least two CE analyses of the BaPdGMP.
PURIFICATION AND RECOVERY OF BULKY DNA ADDUCTS
105
ACKNOWLEDGMENTS
TABLE 2
Percent Recovery Level of adduction
Final sample volume (ml)
%rec 6 SD
4 in 10 6 4 in 10 7 4 in 10 8
63–79 67–71 5–12
91 6 5 81 6 9 54 6 3
Mention of a trade name does not constitute endorsement by the U.S. EPA. This method has not yet undergone Agency level review and thus should not be construed as an EPA-approved method. This article represents contribution NHEERL-NAR-No. 2055 of the U. S. EPA National Health and Environmental Effects Research Lab, Atlantic Ecology Division.
REFERENCES
The best recoveries were obtained for those samples that had the highest level of adduction. This may in part be due to the fact that volume reduction was the greatest for those samples with the lowest level of adduction. This extra amount of volume reduction was necessary to obtain a more quantifiable BaPdGMP peak on the electropherogram because the starting amount of BaPdGMP was 4 ml instead of the 25 ml used at the higher levels. This extra volume reduction was also partly responsible for the more apparent unresolved complex mixture of compounds that appears as a mound that elutes for a few minutes prior to the BaPdGMP and interferes with its quantitation (Figs. 2C and 2F). However, this mound is primarily due to as yet unknown constituents in the mononucleotides solution, an increased amount of which is used for those samples with lower levels of adduction. These unknown constituents are perhaps indicative of, or responsible for, the extra losses at lower levels of adduction. A possible explanation for this includes greater irreversible adsorption of the BaPdGMP to these constituents, which in turn are irreversibly bound to the PSDVB beads. Alternatively, a portion of these constituents could bind to the microcentrifuge walls and cause losses when the BaPdGMP in turn binds to them. Earlier research in fact showed that the adduct did bind to many materials used in filters and to proteins, which is the reason for the minimal enzyme levels used in the hydrolyses. In addition, losses could result during the CE stacking if some of the BaPdGMP was swept out from the head of the column because of incomplete separation from the high levels of other constituents. To achieve levels of adduction much lower than those used in this study, it will probably be necessary to utilize a different separation or detection system to eliminate the mound of interfering compounds. When analyzing real samples with unknown levels of adduction, the amount of volume reduction can be matched with the level of sensitivity of the detection system. However, it would be prudent to initially reduce the volume to ;35 ml (70 ml after an equal volume of MeOH is added). An analysis at that volume would use less than 3% of the sample, and if further volume reduction was necessary, it could then be performed. This purification method works equally well for 39- and 59monophosphates. It also works well for nucleosides, although the final percentage of methanol must be increased to elute the adducts from the column.
1. Poirier, M. C. (1997) Environ. Health Perspect. 105, 907–912. 2. Pfau, W. (1997) Biomarkers 2, 145–151. 3. Schilderman, P. A. E. L., Hoogewerff, J. A., vanSchooten, F. J., Maas, L. M., Moonen, E. J. C., vanOs, B. J. H., vanWijnen, J. H., and Kleinjans, J. C. S. (1997) Environ. Health Perspect. 105, 322–330. 4. Mathieu, A., Payne, J. F., Fancey, L. L., Santella, R. M., and Young, T. L. (1997) J. Toxicol. Environ. Health 51, 1– 4. 5. Harvey, J. S., and Parry, J. M. (1997) Mutagenesis 12, 153– 158. 6. Karakoc, F. T., Hewer, A., Phillips, D. H., Gaines, A. F., and Yuregir, G. (1997) Biomarkers 2, 303–309. 7. Canova, S., Degan, P., Peters, L. D., Livingstone, D. R., Voltan, R., and Venier, P. (1998) Mutat. Res. Fundam. Mol. Mech. Mutat. 399, 17–30. 8. Harvey, J. S., and Parry, J. M. (1998) Mutat. Res. Fundam. Mol. Mech. Mutat. 399, 31– 42. 9. Burcham, P. C. (1998) Mutagenesis 13, 287–305. 10. Deflora, S., Izzotti, A., Randerath, K., Randerath, E., Bartsch, H., Nair, J., Balansky, R., Vanschooten, F., Degan, P., Fronza, G., Walsh, D., and Lewtas, J. (1996) Mutat. Res. Rev. Genet. Toxicol. 366, 197–238. 11. Poirier, M. C., and Weston, A. (1996) Environ. Health Perspect. 104(Suppl. 5), 883– 893. 12. Elespuru, R. K. (1996) Mutat. Res. Rev. Genet. Toxicol. 365(1-3), 191–204. 13. Dale, C. M., and Garner, R. C. (1996) Food Chem. Toxicol. 34, 905–919. 14. Motykiewicz, G., Perera, F. P., Santella, R. M., Hemminki, K., Seemayer, N. H., and Chorazy, M. (1996) Toxicol. Lett. 88, 169 – 173. 15. La, D. K., and Swenberg, J. A. (1996) Mutat. Res. Rev. Genet. Toxicol. 365(1–3), 129 –146. 16. Nestmann, E. R., Bryant, D. W., and Carr, C. J. (1996) Regul. Toxicol. Pharmacol. 24(1, Part 1), 9 –18. 17. Christiani, D. C. (1996) Environ. Health Perspect. 104(Suppl. 5), 921–925. 18. Koskinen, M., Rajaniemi, H., and Hemminki, K. (1997) J. Chromatogr. B 691, 155–160. 19. Baan, R. A., Steenwinkel, M. J. S. T., vanAsten, S., Roggeband, R., and vanDelft, J. H. M. (1997) Mutat. Res. Fundam. Mol. Mech. Mutat. 378, 41–50. 20. Phillips, D. H. (1997) Mutat. Res. Fundam. Mol. Mech. Mutat. 378, 1–12. 21. vanBommel, M. R., de Jong, A. P. J. M., Tjaden, U. R., Irth, H., and van der Greef, J. (1996) J. Chromatogr. A 755, 205– 210. 22. Zamzow, D., Lin, C. H., Small, G. J., and Jankowiak, R. (1997) J. Chromatogr. A 781, 73– 80. 23. Li, M., and Hurtubise, R. J. (1998) Anal. Lett. 31, 491–510.
106
CURT B. NORWOOD
24. Culp, S. J., Roberts, D. W., Talaska, G., Lang, N. P., Fu, P. F., Lay, J. O., Teitel, C. H., Snawder, J. E., VonTungeln, L. S., and Kadlubar, F. F. (1997) Mutat. Res. Fundam. Mol. Mech. Mutat. 378, 97–112. 25. Randerath, K., Sriram, P., Moorthy, B., Aston, J. P., Baan, R. A., vandenBerg, P. T. M., Booth, E. D., and Watson, W. P. (1998) Chem.-Biol. Interact. 110, 85–102. 26. Shinozaki, R., Inoue, S., and Choi, K. S. (1998) Cytometry 31, 300 –306. 27. Bol, S. A. M., deGroot, A. J. L., Tijdens, R. B., Meerman, J. H. N., Mullenders, L. H. F., and vanZeeland, A. A. (1997) Anal. Biochem. 251, 24 –31. 28. OConnor, P., Fremont, S., Schneider, J., Dowty, H., Jaeger, M., Talaska, G., and Warshawsky, D. (1997) J. Chromatogr. B 700, 49 –57. 29. Tretyakova, N. Y., Chiang, S. Y., Walker, V. E., and Swenberg, J. A. (1998) J. Mass Spectrom. 33, 363–376.
30. Tretyakova, N., Sangaiah, R., Yen, T. Y., Gold, A., and Swenberg, J. A. (1997) Chem. Res. Toxicol. 10, 1171–1179. 31. Barry, J. P., Norwood, C., and Vouros, P. (1996) Anal. Chem. 68, 1432–1438. 32. Schrader, W., and Linscheid, M. (1997) Arch. Toxicol. 71, 588 – 595. 33. Deforce, D. L. D., Lemiere, F., Esmans, E. L., DeLeenheer, A., and VandenEeckhout, E. G. (1998) Anal. Biochem. 258, 331–338. 34. Ding, J. M., Szeliga, J., Dipple, A., and Vouros, P. (1997) J. Chromatogr. A 781, 327–334. 35. Ranasinghe, A., Scheller, N., Wu, K. Y., Upton, P. B., and Swenberg, J. A. (1998) Chem. Res. Toxicol. 11, 520 –526. 36. Norwood, C., and Vouros, P. (1993) in Mass Spectrometry— Clinical and Biomedical Applications (Desiderio, D., Ed.), Vol. 2, pp. 89 –133. IEEE Comp. Soc., Los Alamitos, CA. 37. Norwood, C., Jackim, G., and Cheer, S. (1993) Anal. Biochem. 213, 194 –199.
Purification and Recovery of Bulky Hydrophobic DNA Adducts Curt B. Norwood U. S. EPA AED, 27 Tarzwell Drive, Narragansett, Rhode Island 02882
Received March 19, 1999
For many years 32P postlabeling has detected DNA adducts at very low levels and yet has not been able to identify unknown adducts. Mass spectrometry offers substantially improved identification powers, albeit at some loss in detection limits. With this ultimate utilization of mass spectrometry in mind, the current research presents a new method to quantitatively purify bulky hydrophobic DNA adducts at levels that are pertinent to ongoing DNA adduct research in human health and environmental fields. This method was demonstrated with benzo[a]pyrene adducts. Purification was accomplished with the use of small columns (7.5-mm frits) with an 11 mg bed of polystyrene– divinlybenzene beads which retained the adducts while permitting the nonadducted nucleotides to be washed out with water. Subsequently, the adducts were eluted with 50% MeOH and the sample was reduced in volume in an evacuated centrifuge. Purification was demonstrated at adduct levels ranging from 4 adducts in 10 6 nonadducted nucleotides to 4 in 10 8. For these levels, analyses by capillary electrophoresis with sample stacking and UV detection determined that recoveries ranged from 91 to 54%, respectively. The adduct quantities isolated should be sufficient to allow the use of current MS capabilities that are linked on-line to separation methodologies such as capillary electrophoresis, capillary electrochromatography, and high-pressure liquid chromatography.
DNA adducts, the covalent modifications of DNA, continue to be of major importance, as evidenced by the many hundreds of research articles devoted to them every year. Initially, the focus on DNA adducts involved links to cancer, especially in humans. More recently, DNA adducts have been viewed as biomarkers of both exposure and effects, not only in humans (1), but also in other organisms (2– 8). Considerable research effort has addressed what have come to be 100
known as endogenous adducts, which in part may be due to lipid peroxidation (9). Because of their genotoxic properties, the importance of adducts to disease states other than carcinogenesis is now also recognized (10 – 12). Furthermore, DNA adducts are increasingly involved in risk evaluations (13–16) and the development of intervention strategies (17). Much of the current DNA adduct research relies on the 32 P-postlabeling analytical methodology (P32) 1 with autoradiography. P32 is especially appropriate in those situations where the genotoxic compounds are known. Furthermore, P32 is extremely sensitive and requires only very small amounts of DNA. However, P32 suffers from a poor ability to actually identify unknowns, not only detect them. Traditionally P32 has been used with autoradiography to reveal the migration pattern of various components on TLC plates. P32 has also been used with HPLC and on-line 32P (18) detection. In both cases any identification capabilities are limited to comparisons to known components that share similar elution characteristics. This and other shortcomings of P32 (19) are widely recognized and many pleas have been made for enhanced analytical techniques to overcome this deficiency (16, 20). A similar inability to identify unknown DNA adducts is shared by approaches that rely on fluorescence (21, 22), phosphorescence (23), immunochemical (24 –26), and electrochemical (27) detection. Both NMR and MS offer vastly improved identification abilities. However, nuclear magnetic resonance suffers from a requirement for considerably more sample amount than does MS. MS is especially useful if coupled with a separation scheme such as TLC (28), HPLC (29, 30), capillary electrophoresis (CE) (31–33), 1
Abbreviations used: 50% MeOH, methanol:deionized water(1:1, v/v); BaPdGMP, benzo[a]pyrene-29-deoxyguanosine 59-phosphate; CE, capillary electrophoresis; CEC, capillary electrochromatography; DI-H 2O, deionized water; P32, 32P-postlabeling; %rec, percent recovery. 0003-2697/99
PURIFICATION AND RECOVERY OF BULKY DNA ADDUCTS
capillary electrochromatography (CEC) (34), or gas chromatography (35), although this latter technique usually requires derivatization, which introduces other analysis problems (36). However, when MS monitors only selected ions or reactions, although its sensitivity is considerably enhanced, its identification powers are sharply reduced. Coupling of the separation scheme on-line with the detector provides greater ease of operation, better quantitation, less sample handling, and less chance of contamination. This article presents a methodology to purify bulky non-polar-adducted DNA mononucleotides (59-monophosphates) by removing a great excess of nonadducted DNA mononucleotides. The levels of adduction of the benzo[a]pyrene adducts ranged from 4 in 10 6 to 4 in 10 8 (4 adducts for 10 n nonadducted nucleotides). Although the results presented here relied on CE separations coupled, on line, with ultraviolet detection, many alternative separation schemes and detectors could have been utilized. The actual amounts of DNA should be sufficient to permit MS analyses and thus enable the identification of unknown adducts. MATERIALS AND METHODS
Two batches of DNA were prepared and hydrolyzed under sterile conditions. The DNA hydrolysis solution for salmon testes DNA (Sigma, St. Louis, MO) consisted of DNA at 3 mg/ml, 50 mM Tris, pH 8.6 (Sigma), 5 mM CaCl 2 and 4 mM MgCl 2. To initiate hydrolysis to generate 59-mononucleotides, two enzymes were added: SVPDEase I (Sigma P7027 0.002 U/ml of DNA solution) and DNase I (Sigma 3 kU/ml). This batch was incubated at 37°C for 4 days, The second batch of DNA contained calf thymus DNA (Sigma) that was again at 3 mg/ml in a hydrolysis solution that contained buffer and salt concentrations identical to those above. However, enzyme concentrations were lower— 0.0005 U and 0.8 kU/ml, for SVPDEase I and DNase I, respectively. This batch was incubated at 37°C for 8 days. The mononucleotide solutions were kept frozen except for brief thawing periods to permit removal of portions to be used in subsequent analyses. To produce benzo[a]pyrene–29-deoxyguanosine 59phosphate adducts (BaPdGMP), benzo[a]pyrene– diolepoxide was reacted with calf thymus DNA, enzymatically digested to 59-monophosphates, passed through a membrane that contained polystyrene– divinylbenzene beads to purify the BaPdGMP, and concentrated in a vacuum centrifuge, as previously described (37). A standard solution of this was obtained by diluting an aliquot of this solution by a factor of ;400 in methanol: deionized water (1:1, v/v) (50% MeOH). CE was performed with sodium carbonate/bicarbonate buffer (15 mM, pH 8.6) (Sigma) in uncoated fused silica capillary columns (780 –900 mm 3 0.375 mm o.d.
101
3 0.075 mm i.d.) (Polymicro Technologies, Phoenix, AZ). The column was maintained at ambient temperature. The CE system, which was assembled from individually purchased components, was operated at 15–17 kV provided by a Bertan (Hicksville, NY) Series 230 polarity-reversing high-voltage power supply. These conditions resulted in ;17 mA. CE analyses involved sample stacking, as previously described (37). Briefly, the column was approximately 21 filled (from the head) with sample (volume approximately 1.7 ml). High voltage was applied with polarity such that endoosmotic flow was toward the column head. When the monitored current reached ;16.7 mA, the polarity was reversed so that the endoosmotic flow carried components in the normal direction—toward the tail of the column. On-column UV absorbance was monitored with a Linear (Reno, NV) Model 200 UVIS detector located 50 mm from the tail of the column. Data were collected with a Crystal CE system (ATI, Franklin, MA). Purification of adducts was accomplished via liquid chromatography with columns whose design and use is described in the next five paragraphs. Empty column preparation. Empty 1-ml polypropylene columns (Quiagen, Chatsworth, CA) initially had only the bottom frit (proprietary material) in place. These 7.5-mm i.d. columns actually have a volume of ;6 ml if filled to the top of the flared out upper region. For the procedures described below, nylon gloves were worn and all utensils were very clean. The tapered column bottom was removed at a level ;1 mm below the bottom frit. The bottom 7.5-mm frit was pushed out the top with a rod and, along with the top frit, was placed in a container of methanol. The bottom of the empty column was swaged a bit smaller by firmly pressing on a microcentrifuge tube and leaving the assembly for at least a few hours. After removal of the swaging tube, the column was held with forceps and rinsed sequentially with hot tap water, deionized water (DI-H 2O), and methanol. Column bed preparation. A frit was removed from the methanol and very firmly pressed with a rod to the bottom of the column. Eighty ml(11 mg) of dry POROS 50 RI beads (polystyrene– divinylbenzene)(Perseptive Biosystems, Cambridge, MA) in a pipettor with a slightly enlarged tip orifice was dispensed into the column. A top frit was inserted into the column and quite firmly pressed into place. The resultant POROS bed thickness was ;1.5 mm. Methanol (3 ml) was added above the top frit and was allowed to drip via gravity. When the volume remaining above the top frit was ;1 ml, DI-H 2O was added to fill the column to the top and thus bring the volume to ;6 ml. The column was again allowed to drip under gravity until the remaining volume was ;3 ml. The column was rinsed out
102
CURT B. NORWOOD
with DI-H 2O to remove any POROS remaining above the top frit and refilled with 6 ml of DI-H 2O. The column was again allowed to drip until ;1 ml of DIH 2O remained. The column was again filled with DIH 2O and the drip rate was measured. Rates of from 3 to 7 s per drop were considered acceptable. Slower drip rates required removal of the frits and reassembly of the column. Faster drip rates could be slowed by either switching back to methanol and returning to DI-H 2O, or by pressing down more firmly on the top frit. Sample loading and rinse to remove nonadducted nucleotides. An acceptable column was emptied and then prefilled with between 0.3 and 1.0 ml of DI-H 2O (depending on the planned sample volume) and placed nearly horizontally into a holder. The appropriate volumes of DNA mononucleotides and standard BaPdGMP were added. The column was returned to an upright position to permit gravity dripping that continued until there was very little liquid remaining above the top frit. DI-H 2O (0.5 ml) was added to the column and again allowed to pass down through the column bed until the top frit was nearly dry. This 0.5-ml rinsing procedure was repeated two more times. To further wash out the nonadducted nucleotides, the top of the column was grasped with locking pliers and 0.8 ml of DI-H 2O was added to the column. Two drops was allowed to elute from the column, after which time the column was filled to the top with DI-H 2O and immediately inverted to pour out the liquid contents. The bottom of the now upside-down column was rinsed off with DI-H 2O. The column was righted and this filling and rinsing subprocedure was repeated four more times. The entire washing procedure just described was performed a total of six times. Adduct elution and collection. The column was suspended vertically just above a 0.6-ml preweighed, silanized microcentrifuge tube (PGC Scientifics, Gaithersburg, MD) that had been vortexed/rinsed with hot water, DI-H 2O, and methanol. Approximately 350 ml of 50% MeOH was added to the column to elute the adducted nucleotides. After a few drops had been collected, an additional 350 ml of 50% MeOH was added above the column. A total of ;570 ml was collected. Volume reduction. The microcentrifuge tube was placed in a clean vacuum centrifugation system (Savant, San Diego, CA). Slow evacuation conditions (considerable restriction in the line between the centrifuge and the vacuum pump) for 1 h resulted in a volume of ;150 ml. Subsequent fast evacuation for an additional hour resulted in a volume of ;25 ml. The exact volume was determined by weighing the tube and contents. An equal volume of methanol was added and the microcentrifuge tube was vortexed for 1 min prior to centrifugation and subsequent CE analysis. In some cases, to achieve a smaller final volume, this ;50-ml volume was
TABLE 1
Experimental Design Level of adduction
Adduct volume (ml)
DNA volume (ml)
Number of samples
4 in 10 6 4 in 10 7 4 in 10 8
25 25 4.17
120 1200 2000
8 4 3
further reduced in the vacuum concentrator to ;5 ml. An equal volume of MeOH was added prior to vortexing, centrifugation, and CE analysis. The two types of calculations used in this research, level of adduction and percent recovery (%rec), were based on data generated by CE analyses (with sample stacking) where the solvent was 50% MeOH. The first of these, level of adduction, required a determination of the proper volumes of BaPdGMP and mononucleotide solutions that should be mixed to achieve the desired level of adduction. This determination required a calculation of the moles of BaPdGMP in the standard solution and the moles of dGMP in an equal volume of the mononucleotides solution that had been diluted 3000-fold in 50% MeOH. To enable this calculation, separate CE analyses were performed on equal volume injections of each of these solutions. Figure 1 shows the resultant electropherograms with UV detection at 279 nm. To calculate the relative number of moles in the BaPdGMP and the undiluted mononucleotides solutions, the observed ratio of the peak areas of the BaPdGMP in the standard and the dGMP in the diluted mononucleotides solution was multiplied by the reciprocals of their respective extinction coefficients of 41,000 and 7770, the reciprocal of the dilution factor, the known value of 0.21 for the mole fraction of dGMP in calf thymus DNA, and the reciprocal of the observed retention time ratio for the two peaks (to correct for differences in ion mobility). This calculation showed that there were 1.8 adducts for every 10 5 nucleotides in equal volumes of the BaPdGMP and the undiluted DNA mononucleotides solution. If 25 ml of the BaPdGMP were mixed with 120 ml of the DNA mononucleotides, the resultant solution would contain ;4 adducts per 10 6 nonadducted nucleotides. This calculation does not require knowledge of the precise concentration of either solution. However, those concentrations are ;90 pg/ml and 3 mg/ml, respectively. For this study, samples were developed by mixing known quantities of the standard with known amounts of the DNA mononucleotides solution to achieve various levels of adduction. Table 1 details these volumes, the level of adduction, and the number of samples analyzed. Each sample was then treated with the above purification and concentration procedures. The second calculation type, %rec, again was based on sep-
PURIFICATION AND RECOVERY OF BULKY DNA ADDUCTS
103
arate, equal volume, sample stacked, CE analyses of the BaPdGMP standard, and the sample. The BaPdGMP peak area at 344 nm for the standard was compared with that same peak area for a sample. Factoring in the initial volume of standard added and the final sample volume allowed a simple determination of the %rec. RESULTS AND DISCUSSION
This study presents results of an ongoing effort to purify unknown hydrophobic DNA adducts at levels of adduction in the low 10 28 range (or lower) in quantities that should be sufficient to permit their identification and quantitation by MS. The starting amounts of DNA and the final sample volumes have been selected so that at the lowest levels of adduction ;20% of the sample could be injected for separation of the components by the chosen method and detection by MS. This study, however, relies on a UV detector whose identification powers are vastly inferior to that of MS, but nonetheless sufficient for methods development with known compounds. Many of the individual steps detailed under Materials and Methods were instituted in response to specific problems that had been previously noted during the methods development. For example, gravity induced flow starts easily in columns assembled with frits that had been held in MeOH, but not with dry frits. Swaging of the column bottom was necessary to greatly reduce the flow around, rather than through, the bottom frit and thus reduce and lend reproducibility to the total flow rate. In addition, the unwanted retention of the nonadducted nucleotides was drastically reduced if the bottom of the column had been removed and the volume immediately below the bottom frit was flushed copiously during the rinsing procedures. Locking pliers were used to securely hold the column because if it had been dropped even a few centimeters, the column bed would have developed areas of high flow and the %rec would have suffered dramatically. The fixed wavelength electropherograms in Fig. 1 were recorded at 279 and 344 nm because of the local maxima in the UV absorption spectrum of BaPdGMP at those wavelengths. The mononucleotides do not absorb at 344 nm. Figure 1 clearly shows that the BaPdGMP elutes earlier than any of the four nucleotides. This time separation is important because of the great excess of mononucleotides in the actual samples. In fact, for the levels of adduction in this study, the samples were formulated with between 15,000 and 1,500,000 times as much of the nucleotides relative to the BaPdGMP as those demonstrated in Fig. 1. Figure 2 displays electropherograms at the same two wavelengths (279 and 344 nm) for samples that originally were at the three levels of adduction and subse-
FIG. 1. Electropherograms for sample-stacked injections into 80cm 3 75-mm i.d. columns with 15 mM sodium carbonate/bicarbonate buffer, pH 9.6, operated at 17 kV (17 mA). (A) Electropherogram at 279 nm of a 3000-fold dilution of calf thymus DNA hydrolyzed to its 29-deoxy-59-monophosphate nucleotides (A, C, T, and G) (approx 1.7 ng of DNA). (B and C) Electropherograms at 279 and 344 nm, respectively, of 150 pg of the BaPdGMP adduct (B).
quently purified by the methods of this study. The 279-nm electropherograms show the limited amounts of nonadducted mononucleotides remaining in the sample and by inference thus indicate the tremendous extent to which these samples have been purified. Even for the 10 28 sample, it is clear that far greater than
104
CURT B. NORWOOD
FIG. 2. Electropherograms of samples of DNA dosed at various levels of adduction and purified by the methods detailed in this article: (A–C) at 279 nm and (D–F) at 344 nm: (A and D) at 4 in 10 6, (B and E) at 4 in 10 7, and (C and F) at 4 in 10 8. Other electrophoresis conditions were the same as in Fig. 1.
99.99% of the nonadducted mononucleotides have been removed. To perform the %rec calculations, the 344-nm electropherogram peak was used because interferences with the unknown components are more problematic in the 279-nm electropherogram, especially at the lowest level of adduction (Fig. 2C). For each of the three levels of adduction in this study, Table 2 lists the range of final sample volumes and the mean %rec 6 standard deviation. In all cases
the final solvent was 50% MeOH and the volumes were determined by weight. The %rec calculations were in part based on the 344-nm CE peak area for the standard BaPdGMP. An injection volume of ;1.7 ml of this standard contained ;150 pg, which was well above the limit of detection (Fig. 1C). None of the %rec values exceeded 100. Each of the %rec determinations was based on two CE analyses of the sample and at least two CE analyses of the BaPdGMP.
PURIFICATION AND RECOVERY OF BULKY DNA ADDUCTS
105
ACKNOWLEDGMENTS
TABLE 2
Percent Recovery Level of adduction
Final sample volume (ml)
%rec 6 SD
4 in 10 6 4 in 10 7 4 in 10 8
63–79 67–71 5–12
91 6 5 81 6 9 54 6 3
Mention of a trade name does not constitute endorsement by the U.S. EPA. This method has not yet undergone Agency level review and thus should not be construed as an EPA-approved method. This article represents contribution NHEERL-NAR-No. 2055 of the U. S. EPA National Health and Environmental Effects Research Lab, Atlantic Ecology Division.
REFERENCES
The best recoveries were obtained for those samples that had the highest level of adduction. This may in part be due to the fact that volume reduction was the greatest for those samples with the lowest level of adduction. This extra amount of volume reduction was necessary to obtain a more quantifiable BaPdGMP peak on the electropherogram because the starting amount of BaPdGMP was 4 ml instead of the 25 ml used at the higher levels. This extra volume reduction was also partly responsible for the more apparent unresolved complex mixture of compounds that appears as a mound that elutes for a few minutes prior to the BaPdGMP and interferes with its quantitation (Figs. 2C and 2F). However, this mound is primarily due to as yet unknown constituents in the mononucleotides solution, an increased amount of which is used for those samples with lower levels of adduction. These unknown constituents are perhaps indicative of, or responsible for, the extra losses at lower levels of adduction. A possible explanation for this includes greater irreversible adsorption of the BaPdGMP to these constituents, which in turn are irreversibly bound to the PSDVB beads. Alternatively, a portion of these constituents could bind to the microcentrifuge walls and cause losses when the BaPdGMP in turn binds to them. Earlier research in fact showed that the adduct did bind to many materials used in filters and to proteins, which is the reason for the minimal enzyme levels used in the hydrolyses. In addition, losses could result during the CE stacking if some of the BaPdGMP was swept out from the head of the column because of incomplete separation from the high levels of other constituents. To achieve levels of adduction much lower than those used in this study, it will probably be necessary to utilize a different separation or detection system to eliminate the mound of interfering compounds. When analyzing real samples with unknown levels of adduction, the amount of volume reduction can be matched with the level of sensitivity of the detection system. However, it would be prudent to initially reduce the volume to ;35 ml (70 ml after an equal volume of MeOH is added). An analysis at that volume would use less than 3% of the sample, and if further volume reduction was necessary, it could then be performed. This purification method works equally well for 39- and 59monophosphates. It also works well for nucleosides, although the final percentage of methanol must be increased to elute the adducts from the column.
1. Poirier, M. C. (1997) Environ. Health Perspect. 105, 907–912. 2. Pfau, W. (1997) Biomarkers 2, 145–151. 3. Schilderman, P. A. E. L., Hoogewerff, J. A., vanSchooten, F. J., Maas, L. M., Moonen, E. J. C., vanOs, B. J. H., vanWijnen, J. H., and Kleinjans, J. C. S. (1997) Environ. Health Perspect. 105, 322–330. 4. Mathieu, A., Payne, J. F., Fancey, L. L., Santella, R. M., and Young, T. L. (1997) J. Toxicol. Environ. Health 51, 1– 4. 5. Harvey, J. S., and Parry, J. M. (1997) Mutagenesis 12, 153– 158. 6. Karakoc, F. T., Hewer, A., Phillips, D. H., Gaines, A. F., and Yuregir, G. (1997) Biomarkers 2, 303–309. 7. Canova, S., Degan, P., Peters, L. D., Livingstone, D. R., Voltan, R., and Venier, P. (1998) Mutat. Res. Fundam. Mol. Mech. Mutat. 399, 17–30. 8. Harvey, J. S., and Parry, J. M. (1998) Mutat. Res. Fundam. Mol. Mech. Mutat. 399, 31– 42. 9. Burcham, P. C. (1998) Mutagenesis 13, 287–305. 10. Deflora, S., Izzotti, A., Randerath, K., Randerath, E., Bartsch, H., Nair, J., Balansky, R., Vanschooten, F., Degan, P., Fronza, G., Walsh, D., and Lewtas, J. (1996) Mutat. Res. Rev. Genet. Toxicol. 366, 197–238. 11. Poirier, M. C., and Weston, A. (1996) Environ. Health Perspect. 104(Suppl. 5), 883– 893. 12. Elespuru, R. K. (1996) Mutat. Res. Rev. Genet. Toxicol. 365(1-3), 191–204. 13. Dale, C. M., and Garner, R. C. (1996) Food Chem. Toxicol. 34, 905–919. 14. Motykiewicz, G., Perera, F. P., Santella, R. M., Hemminki, K., Seemayer, N. H., and Chorazy, M. (1996) Toxicol. Lett. 88, 169 – 173. 15. La, D. K., and Swenberg, J. A. (1996) Mutat. Res. Rev. Genet. Toxicol. 365(1–3), 129 –146. 16. Nestmann, E. R., Bryant, D. W., and Carr, C. J. (1996) Regul. Toxicol. Pharmacol. 24(1, Part 1), 9 –18. 17. Christiani, D. C. (1996) Environ. Health Perspect. 104(Suppl. 5), 921–925. 18. Koskinen, M., Rajaniemi, H., and Hemminki, K. (1997) J. Chromatogr. B 691, 155–160. 19. Baan, R. A., Steenwinkel, M. J. S. T., vanAsten, S., Roggeband, R., and vanDelft, J. H. M. (1997) Mutat. Res. Fundam. Mol. Mech. Mutat. 378, 41–50. 20. Phillips, D. H. (1997) Mutat. Res. Fundam. Mol. Mech. Mutat. 378, 1–12. 21. vanBommel, M. R., de Jong, A. P. J. M., Tjaden, U. R., Irth, H., and van der Greef, J. (1996) J. Chromatogr. A 755, 205– 210. 22. Zamzow, D., Lin, C. H., Small, G. J., and Jankowiak, R. (1997) J. Chromatogr. A 781, 73– 80. 23. Li, M., and Hurtubise, R. J. (1998) Anal. Lett. 31, 491–510.
106
CURT B. NORWOOD
24. Culp, S. J., Roberts, D. W., Talaska, G., Lang, N. P., Fu, P. F., Lay, J. O., Teitel, C. H., Snawder, J. E., VonTungeln, L. S., and Kadlubar, F. F. (1997) Mutat. Res. Fundam. Mol. Mech. Mutat. 378, 97–112. 25. Randerath, K., Sriram, P., Moorthy, B., Aston, J. P., Baan, R. A., vandenBerg, P. T. M., Booth, E. D., and Watson, W. P. (1998) Chem.-Biol. Interact. 110, 85–102. 26. Shinozaki, R., Inoue, S., and Choi, K. S. (1998) Cytometry 31, 300 –306. 27. Bol, S. A. M., deGroot, A. J. L., Tijdens, R. B., Meerman, J. H. N., Mullenders, L. H. F., and vanZeeland, A. A. (1997) Anal. Biochem. 251, 24 –31. 28. OConnor, P., Fremont, S., Schneider, J., Dowty, H., Jaeger, M., Talaska, G., and Warshawsky, D. (1997) J. Chromatogr. B 700, 49 –57. 29. Tretyakova, N. Y., Chiang, S. Y., Walker, V. E., and Swenberg, J. A. (1998) J. Mass Spectrom. 33, 363–376.
30. Tretyakova, N., Sangaiah, R., Yen, T. Y., Gold, A., and Swenberg, J. A. (1997) Chem. Res. Toxicol. 10, 1171–1179. 31. Barry, J. P., Norwood, C., and Vouros, P. (1996) Anal. Chem. 68, 1432–1438. 32. Schrader, W., and Linscheid, M. (1997) Arch. Toxicol. 71, 588 – 595. 33. Deforce, D. L. D., Lemiere, F., Esmans, E. L., DeLeenheer, A., and VandenEeckhout, E. G. (1998) Anal. Biochem. 258, 331–338. 34. Ding, J. M., Szeliga, J., Dipple, A., and Vouros, P. (1997) J. Chromatogr. A 781, 327–334. 35. Ranasinghe, A., Scheller, N., Wu, K. Y., Upton, P. B., and Swenberg, J. A. (1998) Chem. Res. Toxicol. 11, 520 –526. 36. Norwood, C., and Vouros, P. (1993) in Mass Spectrometry— Clinical and Biomedical Applications (Desiderio, D., Ed.), Vol. 2, pp. 89 –133. IEEE Comp. Soc., Los Alamitos, CA. 37. Norwood, C., Jackim, G., and Cheer, S. (1993) Anal. Biochem. 213, 194 –199.