Anal. Chem. 2001, 73, 303-309
Quantitative Analysis of Etheno-2′-Deoxycytidine DNA Adducts Using On-Line Immunoaffinity Chromatography Coupled With LC/ES-MS/MS Detection Dean W. Roberts, Mona I. Churchwell, Frederick A. Beland, Jia-Long Fang, and Daniel R. Doerge*
National Center for Toxicological Research, Jefferson, Arkansas 72079
Etheno DNA adducts, including 3,N4-etheno-2′-deoxycytidine (etheno-dC), are promutagenic lesions present in normal animal and human tissues. These DNA adducts are believed to be important in the etiology of cancer. Existing methods for quantifying etheno-dC use 32Ppostlabeling. Although highly sensitive, postlabeling requires the use of an energetic radioisotope and considerable time and effort. The new methodology reported here permits automated quantification of trace levels of etheno-dC in crude DNA hydrolysates on the order of 5 adducts in 108 normal nucleotides from 100-µg samples of DNA. This was accomplished by using on-line immunoaffinity chromatography, a reverse-phase LC separation on graphitized carbon, tandem mass spectrometric detection, and an isotopically labeled internal standard. The automated procedures permitted analysis of 4 DNA hydrolysates/hr. The sensitivity using immunoaffinity cleanup was approximately 100-fold greater than that observed when using a silica-based trapping system. The validated method was applied to the analysis of etheno-dC in commercial calf thymus DNA, untreated mouse liver, and untreated rat liver DNA. The demonstrated level of performance suggests future applicability of this method in studies of cancer in humans and experimental animals. The etheno-derivatives of guanine (1,N2- and N2,3-etheno-2′deoxyguanosine), adenine (1,N6-etheno-2′-deoxyadenosine, ethenodA), and cytosine (3,N4-etheno-2′-deoxycytidine, etheno-dC), shown Figure 1, are promutagenic lesions in DNA.1 These adducts are produced either by endogenous metabolic processes, including lipid peroxidation,2 or by metabolic activation of exogenous chemicals, including vinyl chloride3 and urethane.4 Etheno-DNA adducts are persistent in target tissues of experimental animals,3,5 and it is likely that these adducts play an important role in human carcinogenesis caused by diet and lifestyle factors.6 * Corresponding author: Division of Biochemical Toxicology, 3900 NCTR Road, Jefferson, AR 72079. Phone: 870-543-7943. Fax: 870-543-7720. E-mail:
[email protected]. (1) Pandya, G. A.; Moriya, M. Biochemistry 1996, 35, 11487-11492. (2) Nair, J.; Vaca, C. E.; Velic, I.; Mutanen, M.; Valsta, L. M.; Bartsch, H. Cancer Epidemiol., Biomarkers Prev. 1997, 6, 597-601. (3) Swenberg, J. A.; Fedtke, N.; Ciroussel, F.; Barbin, A.; and Bartsch, H. Carcinogenesis 1992, 13, 727-729. (4) Fernando, R. C.; Nair, J.; Barbin, A.; Miller, J. A.; Bartsch, H. Carcinogenesis 1996, 17, 1711-1718. 10.1021/ac000866n CCC: $20.00 Published on Web 12/15/2000
© 2001 American Chemical Society
Figure 1. Structures of etheno-DNA adducts.
The 32P-postlabeling technique has been the principal analytical method used for measurement of etheno-dA and etheno-dC adducts in DNA from rodents and humans.7,8 This assay uses γ-32Plabeled adenosine triphosphate to incorporate a highly radioactive reporter group into nucleotides derived from enzymatic hydrolysis of DNA.9 Adducted nucleotides are then separated from normal nucleotides by TLC and visualized using radioautography. Although highly sensitive for any DNA adduct, this technique requires handling large amounts of an energetic radioisotope, requires extensive sample cleanup, is very time-consuming, has limited analyte resolving power (particularly in the TLC mode), cannot provide structural information about the adduct, and requires extensive method validation for reliable quantitative performance. In the referenced methods for analysis of etheno (5) Guichard, Y.; Ghissassi, F. E.; Nair, J.; Bartsch, H.; Barbin, A. Carcinogenesis 1996, 17, 1553-1559. (6) Nair, J.; Barbin, A.; Velic, I.; Bartsch, H. Mutat. Res. 1999, 424, 59-70. (7) Nair, J.; Barbin, A.; Guichard, Y.; Bartsch, H. Carcinogenesis 1995, 16, 613617. (8) Watson, W. P.; Aston, J. P.; Barlow, T.; Crane, A. E.; Potter, D.; Brown, T. IARC Scientific Publications; No. 150; International Agency for Research on Cancer: Lyon, France, 1999; pp 63-73. (9) Beach, A. C.; Gupta, R. C. Carcinogenesis 1992, 13, 1053-1074.
Analytical Chemistry, Vol. 73, No. 2, January 15, 2001 303
DNA adducts, considerable specificity has been added through the use of immunoaffinity purification7 or HPLC enrichment procedures8 of DNA hydrolysates prior to 32P-postlabeling, and sensitivity in the range of one adduct in 107-109 normal nucleotides has been reported. More recently, GC/MS and LC/ES-MS detection methods in conjunction with labeled internal standards have been used to measure etheno-dA adducts in human placenta at levels > 1 adduct in 106 normal nucleotides10 or etheno-guanine in human liver DNA at > 1 adduct in 107 normal nucleotides.11 These methods used off-line reverse-phase solid-phase extraction to purify etheno DNA adducts from milligram amounts of target DNA, and isotopically labeled analytes were used as internal standards to achieve accuracy and precision of measurement. No MS method for the quantitative analysis of etheno-dC in DNA has been reported. This laboratory reported previously a method that used automated on-line reverse-phase SPE coupled with LC/ES-MS for the quantitative analysis of 4-aminobiphenyl-C8-dG in mouse liver DNA.12 The sensitivity and selectivity of the method permitted adduct measurement at levels < 1 adduct in 107 normal nucleotides (< 10 pg on-column) from crude enzymatic hydrolysates of 100 µg of DNA. The high sensitivity of the method was due, in part, to efficient removal of the unmodified nucleosides through column-switching techniques. Drastically reduced MS responses were observed when these components, which are present at levels many million-fold greater than the adducts, were not removed. In a preliminary investigation, a similar on-line reverse-phase SPE method was applied to the measurement of etheno-DNA adducts in mouse liver using LC/ES-MS/MS.13 Although the method performed well for etheno-dA (LOQ on the order of 1 adduct in 108 normal nucleotides), significant interference from the DNA hydrolysate was encountered in the determination of etheno-dC (LOD, ca. 10 adducts in 106 normal nucleotides). The interference, which resulted primarily from the residual dA introduced into the mass spectrometer, was particularly problematic because of the identical molecular weights for dA and ethenodC (251 Da), the identical principal multiple reaction monitoring (MRM) transition (m/z 252 f 136) observed upon CID, and the incomplete LC resolution (see Figure 2A). This paper reports adaptation of our original on-line SPE methodology for DNA adduct analysis to include a reusable immunoaffinity column for the efficient cleanup of DNA hydrolysates, tandem MS detection, and the use of unique chromatographic properties of graphitized carbon media to maximize separation efficiency. These adaptations significantly eliminated interferences, greatly improved the sensitivity for quantitative determination of etheno-dC, and permitted analysis of untreated rat liver DNA from DNA samples of 100 µg or less. (10) Chen, H. J. C.; Chiang, L. C.; Tseng, M. C.; Zhang, L. L.; Ni, J.; Chung, F. L. Chem. Res. Toxicol. 1999, 12, 1119-1126. (11) Yen, T. Y.; Christova-Gueoguieva, N. I.; Scheller, N.; Holt, S.; Swenberg, J. A.; Charles, M. J. J. Mass Spectrom. 1996, 31, 1271-1276. (12) Doerge, D. R.; Churchwell, M. I.; Marques, M. M.; Beland, F. A. Carcinogenesis 1999, 20, 1055-1061. (13) Churchwell, M. I.; Doerge, D. R.; Roberts, D. W.; Fang, J. L.; Beland, F. A. Proc. 47th ASMS Conf. Mass Spectrom. Allied Top. 1999, 758-759.
304
Analytical Chemistry, Vol. 73, No. 2, January 15, 2001
EXPERIMENTAL SECTION Reagents. Labeled and unlabeled 3,N4-etheno-dC were prepared by the method of Green and Hathway14 and purified by HPLC. Normal nucleosides, calf thymus DNA, and all biochemical reagents were obtained from Sigma Chemical Co. (St. Louis, MO). DNA Modified in Vitro. Calf thymus DNA in 100 mM sodium cacodylate buffer (pH 7.5) was reacted overnight with several stoichiometric ratios of chloroacetaldehyde: DNA nucleosides (1.85 × 10-5, 1.85 × 10-4, and 1.85 × 10-3) according to the method of Young and Santella.15 The DNA was precipitated using ethanol and sodium chloride, washed with 70% ethanol, and dissolved in 5 mM Bis-Tris, 0.1 mM EDTA buffer (pH 7.1). DNA Modified in Vivo. Eight-week-old B6C3F1 untreated mice were killed by exposure to carbon dioxide. The livers were removed, nuclei were isolated,16 and DNA was prepared using a phenol-based isolation method17 and dissolved in 5 mM Bis-Tris, 0.1 mM EDTA (pH 7.1) buffer for analysis. Rat liver DNA from an untreated 13-week-old female Sprague-Dawley was similarly prepared. Enzymatic Hydrolysis of DNA. The in vitro- and in vivomodified DNA samples (ca. 100 µg in 100-200 µL total volume) were spiked with the internal standard (10-20 pg). Initially, the DNA was hydrolyzed enzymatically to nucleosides with DNase I, followed by alkaline phosphatase and snake venom phosphodiesterase (type VII, see ref 18). Because unsatisfactory results were obtained (see Results), subsequent hydrolyses were conducted with micrococcal nuclease, spleen phosphodiesterase (type II), and nuclease P1. Specifically, the samples were incubated with 2 U of micrococcal nuclease and 0.2 U of spleen phosphodiesterase overnight at 37 °C in 20 mM sodium succinate, 10 mM calcium chloride buffer (pH 6). These enzymes had previously been dialyzed against water, as described previously.19 Nuclease P1 (5 µg) was then added, and the hydrolysis was continued an additional 2 h. Complete hydrolysis of the DNA to nucleosides was verified using HPLC-UV. The dA present in some hydrolysates was converted to deoxyinosine by adding adenosine deaminase (0.01 units per sample, sufficient to hydrolyze > 99% of dA in 0.5 h, not shown). The etheno-dC adducts were then determined by direct injection into the LC system. Liquid Chromatography. The liquid handling system consisted of an autosampler (AS3500, Dionex, Sunnyvale, CA), two automated switching valves (TPMV, Rheodyne, Cotati, CA), two HPLC pumps (a Dionex GP40 quaternary gradient pump and a Hewlett-Packard 1050 pump, Palo Alto, CA), and a column heater (Hot Pocket, Keystone Scientific Inc., Bellfonte, PA). The chromatographic components used were a Hypercarb analytical column (2 × 100 mm, 5-µm particles, Keystone Scientific), a Hypercarb trap column (2 × 10 mm, 5-µm particles, Keystone Scientific), and a reusable immunoaffinity column containing (14) Green, T.; Hathway, D. E. Chem.-Biol. Interact. 1978, 22, 211-224. (15) Young, T. L.; Santella, R. M. Carcinogenesis 1988, 9, 589-592. (16) Basler, J.; Hastie, N. D.; Pietras, D.; Matsui, S.-I.; Sandberg, A. A.; Berezney, R. Biochemistry 1981, 20, 6921-6929. (17) Beland, F. A.; Fullerton, N. F.; Heflich, R. H. J. Chromatogr. 1984, 308, 121-131. (18) Heflich, R. H.; Morris, S. M.; Beranek, D. T.; McGarrity, L. J.; Chen, J. J.; Beland, F. A. Mutagenesis 1986, 1, 201-206. (19) Randerath, K.; Reddy, M. V.; Gupta, R. C. Proc. Natl. Acad. Sci. U.S.A. 1981, 6126-6129.
Figure 2. Effect of temperature on the graphitized carbon reverse-phase separation for the on-line IAC-LC/ES-MS/MS detection of ethenodC in chloroacetaldehyde-modified DNA. The transitions for labeled etheno-dC standard (m/z 255 f 139) are shown in the upper traces and unlabeled etheno-dC (m/z 252 f 136) are shown in the lower traces. The peak area and retention time (in minutes) are listed above each integrated peak. Chemically modified calf thymus DNA was hydrolyzed enzymatically to nucleosides and injected directly onto the liquid handling system using (A) left panel, graphitized carbon trap and analytical columns for the separation of the IAC eluate performed at ambient temperature (22 °C) and (B) right panel, the analogous separation with the analytical column held at 85 °C.
covalently bound monoclonal antibody with specificity for ethenodC. Isolation of Anti-Etheno-dC Monoclonal Antibody. A starter culture of the murine hybridoma clone designated 6f5, reported to produce IgG2, lambda isotype monoclonal antibodies with specificity for etheno-dC was kindly provided by Dr. Regina Santella, Columbia University.15 Hybridoma cells were propagated in 1:1 Dulbecco’s modified essential media: Ham’s F-12 media supplemented with 10% heat-inactivated fetal bovine serum, penicillin (50 U/ml), and streptomycin sulfate (50 mg/mL, GIBCO, Grand Island, NY). Cell culture supernatant containing monoclonal antibody was harvested and cellular debris was removed by centrifugation and filtration. Specificity of the monoclonal antibody for chloroacetaldehyde-modified DNA was verified by noncompetitive ELISA using calf thymus DNA modified with chloroacetaldehyde as solid-phase assay antigen.15 Flow-Through Preparation of Immunoaffinity Columns. The apparatus for on-line affinity isolation and covalent binding of monoclonal antibody to protein A consisted of a low pressure, 6-position, 7-port, mobile-phase selection valve (Rheodyne); an ISO-2000 pump (ChromTech, Apple Valley, MN); a 6-port, 2-position, high-pressure selection valve (Rheodyne) configured to put
the protein A column in-line or off-line; a UV detector (Waters 440, Milford, MA); a flow-through pH detector (Bradley-James Co., Santa Ana, CA); and a 300 psi backpressure fitting (ChromTech). Approximately 800 µL of Poros protein A affinity matrix (Perspective Biosystems, Framingham, MA) was slurry-packed in a 4.6 × 50 mm PEEK column (Perspective Biosystems), and the column matrix was cleaned by sequential washing with 50% methanol and 0.1 M glycine (pH 2.5). At a flow rate of 2 mL/min, the protein A column was equilibrated with 0.01 M sodium phosphate, 0.15 M sodium chloride, and 0.01 M EDTA, (isotonic buffered saline, IBS) to establish pH and A280 baselines, then equilibrated with 3 M NaCl, 10 mM boric acid, and 10 mM potassium phosphate (pH 8.9, loading buffer). Anti-etheno-dC monoclonal antibody was immobilized by pumping 50 mL of 6f5 culture supernatant that had been adjusted to contain 3M NaCl, 10 mM boric acid, and 10 mM potassium phosphate (pH 8.9) through the column at 2 mL/ min, followed by 10 mL of loading buffer and then approximately 20 mL of IBS until baseline A280 returned to zero. Antibody activity in the fractions that were collected during loading was negligible, which indicated that essentially all antibody had bound to the protein A column in a single pass at the loading level used. The column containing reversibly immobilized antibody was then Analytical Chemistry, Vol. 73, No. 2, January 15, 2001
305
Table 1. Schedule of Timed Events for Automated IAC-LC/ES-MS/MS Analysis of Etheno-dCa time (min)
mobile phase/flow rate
comments
0-3 0-6 3-6 6-10 10-15
PB at 0.2 mL/min 60%F/40% ACN at 0.2 mL/min PB at 2 mL/min 70%H2O/30% MeOH at 1 mL/min 60%F/40% ACN at 0.2 mL/min
10-15
PB at 1 mL/min
start method by injecting sample and washing IAC column to waste using pump 1 equilibrate trap and analytical columns using pump 2 wash IAC column to waste using pump 1 switch valve 2 and elute IAC onto trap column and wash trap using pump 1 switch valve 2 and backflush trap column onto analytical column using pump 2; start MS acquisition reequilibrate IAC column using pump 1
a The mobile phase content (PB ) phosphate buffer 0.01M containing 0.1% sodium azide, pH 7.2; F ) 0.1% aqueous formic acid, ACN ) acetonitrile, MeOH ) methanol) and flow rates are indicated along with the valve/pump switching events
washed with 10 mL of 0.2 M triethanolamine (pH 8.5, TEA) followed by 10 mL of freshly prepared dimethylpimelimidate (DMP, Pierce Co., Rockford, IL), 10 mg/mL in TEA to provide a covalent link between amino groups on the Fc portion of the immunoglobulin and protein A. Excess DMP was quenched by washing with 5 mL of 0.2 M monoethanolamine (pH 8) followed by 10 mL of PBS containing 0.02% (w/v) sodium azide as an antibacterial for storage. To ensure consistency, the 800-µL batch of protein A with covalently bound anti-etheno-dC monoclonal antibody was expelled from the 4.6 × 50 column, mixed, and used to slurry-pack five replicate 2.1 × 30 mm PEEK columns (Perspective Biosystems) at 2000 psi using IBS containing 0.02% (w/v) sodium azide as the mobile phase. The immunoaffinity (IAC) columns were stored at 4 °C in a buffer containing 0.01 M sodium phosphate and 0.02% sodium azide (pH 7.20). Timed Events for On-Line IAC-LC/ES-MS/MS. Gradient pump 1 and valve 1 were used for sample injection, cleanup, and regeneration of the IAC column, and isocratic pump 2 and valve 2 were used to load then backflush the trap column to the analytical column during analysis and to keep a constant flow of mobile phase going into the mass spectrometer during sample loading and washing. The sample was loaded onto the IAC column and was washed for 6 min with phosphate buffer (0.01 M containing 0.1% sodium azide, pH 7.2). Then valve 2 was switched and the material that was retained on the IAC was eluted onto the trap column (Hypercarb, 2 × 10 mm) with 30% aqueous methanol for 4 min. When valve 2 was switched at 6 min, the concentrated sample zone on the trap column was backflushed onto the analytical column (Hypercarb, 2 × 100 mm, 5 micron) at 0.2 mL/min with a mobile phase consisting of 40% acetonitrile in 0.1% aqueous formic acid. The separated sample components were then eluted into the mass spectrometer for data acquisition. When the 5-min acquisition was completed, the valves were switched to the original positions. At this point, the IAC column was equilibrated with starting mobile phases and the process could be repeated. Equilibration of the trap column occurred concurrently with the IAC loading/washing steps. These events are presented in Table 1, and a complete depiction of the plumbing configuration for IAC-LC/MS was previously published20. This method permitted automated analysis of four crude DNA hydrolysates/hr through software-directed commands controlling the mass spectrometer, quaternary LC pump, and switching valves. Mass Spectrometry. A Quattro LC triple quadrupole mass spectrometer (Micromass, Manchester, England) equipped with an ES interface was used with a source block temperature of 150 °C and desolvation temperature of 450 °C. Nitrogen gas was used 306 Analytical Chemistry, Vol. 73, No. 2, January 15, 2001
as the desolvation gas (750 L/hr) and the nebulizing gas (90 L/hr). Argon was used as the collision gas, at a collision cell pressure of 1.5 × 10-3 mBar. Positive ions were acquired in MRM mode (dwell time, 0.3 s; span, 0.02 Da; and interchannel delay, 0.03 s) for the M + H+-to-BH2+ transitions for etheno-dC (m/z 252/136) and the internal standard 15N3-etheno-dC (m/z 255/139). The cone voltage was 20 V for the etheno-dC transitions, and the collision energy was 15 eV. Preparation and Characterization of 15N-etheno-dC Internal Standard. Uniformly labeled 15N3-dC was obtained from Cambridge Isotope Laboratories (Andover, MA). Chloroacetaldehyde was reacted with 15N3-dC to form 15N3-etheno-dC, which was purified using preparative HPLC. The chemical concentration was determined spectrophotometrically, using LC with UV detection (260 nm) in comparison to an authentic sample of the unlabeled analogue. The isotopic distribution was determined using full-scan LC/ES-MS (m/z 100-600). The labeled etheno-dC was 96.4% 15N3 and 3.6% 15N2, and no unlabeled isotopomer was detected ( 0.999), and the slope was essentially unity (not shown). The performance of the method was validated using two pools of chemically modified calf thymus DNA hydrolysates with and without addition of unlabeled etheno-dC standard. The precision of the method was evaluated by analyzing 13 µg of DNA and 100 pg of the labeled internal standard injected on-column on two different days. On the first day, the determined amount was 4.2 ( 0.1 pg etheno-dC/µg DNA (3% RSD, n)4) and on another day, it was 4.3 ( 0.4 pg etheno-dC/µg DNA (9% RSD, n)4). This corresponds to 5.3 etheno-dC adducts per 106 normal nucleotides. The intra- and interassay precision and accuracy of the method were evaluated by adding a comparable amount of unlabeled 308 Analytical Chemistry, Vol. 73, No. 2, January 15, 2001
etheno-dC, 3.9 pg/µg of DNA, to the same pooled DNA hydrolysate. On the first day, the determined value for the fortified sample was 8.6 ( 0.2 pg etheno-dC/µg DNA (111% of the spike amount, RSD 2%, n)4), and on the second day, it was 8.6 ( 0.4 pg ethenodC/µg DNA (111% of the spike amount, RSD 5%, n)4). Precision was also evaluated using a second DNA hydrolysate that contained a lower adduct level. In this case, the determined value was 1.4 ( 0.05 pg etheno-dC/µg DNA (3% RSD, n)4) or 1.8 etheno-dC adducts per 106 normal nucleotides for an injection of 35 µg of DNA. A final evaluation of method performance near the LOQ was performed using liver DNA from an untreated female rat (see Figure 4). The rat liver DNA contained 0.085 pg etheno-dC/µg DNA ,which is equivalent to 11 adducts in 108 normal nucleotides. An approximately equal concentration of etheno-dC was added to this sample (0.12 pg/µg of etheno-dC from addition of 5 pg to 41 µg of DNA). On day 1, the determined value was 0.249 ( 0.008 pg/µg of DNA (121% of the spike amount, RSD 3%, n)3), and on day 2, the value was 0.239 ( 0.015 pg/µg of DNA (116% of the spike amount, RSD 6%, n)3). These accuracy and precision determinations were comparable to those obtained from DNA samples containing higher adduct levels.
Figure 4. Measurement of etheno-dC in untreated CD rat liver DNA hydrolysate processed using adenosine deaminase. The transitions for labeled etheno-dC standard (m/z 255 f 139) are shown in the upper traces, and unlabeled etheno-dC (m/z 252 f 136) are shown in the lower traces. The peak area and retention time (min) are listed above each integrated peak. A: the left panel shows the responses for labeled and unlabeled etheno-dC standards in the presence of 1000-fold excess dA following treatment with adenosine deaminase. B: the center panel shows the responses for a blank injection containing only labeled etheno-dC and dA. C: the right panel shows the responses by a DNA hydrolysate (100 µg) that were obtained from the liver of an untreated female rat that contained approximately 9 etheno-dA in 108 normal nucleotides.
The method was also applied to the analysis of a commercial salmon testes DNA sample. Duplicate analyses showed 8.7 and 8.6 etheno-dC in 107 normal nucleotides. Analyses were conducted for other commercial DNA samples (calf thymus, human placenta, and Clostridium perfringes), all of which contained etheno-dC below the detection limit. These observations suggest that unidentified factors in the processing and handling of animals and the isolated DNA may affect the measured etheno DNA adduct content of putatively normal and commercially available DNA. CONCLUSIONS This report details new methodology that improves analytical performance in the determination of etheno-dC in DNA hydrolysates. The beneficial aspects include (1) the use of a durable immunoaffinity column to capture and clean up the target analyte from extraneous components, (2) a chromatographic separation on graphitized carbon that takes advantage of the unique retentive characteristics and temperature stability of this phase, (3) the high sensitivity of triple quadrupole tandem mass spectrometry in the multiple-reaction monitoring mode, (4) an isotopically labeled internal standard for robust method performance, and (5) automated method control of all pumping and valve operations for
unattended high throughput analysis relative to the existing 32Ppostlabeling technique. On the basis of the method performance described above for the optimized method, it is reasonable to predict precise and accurate quantification of etheno-dC at levels in DNA with adduct content at or below 5 adducts in 108 normal nucleotides with a signal:noise ratio of g 10. The high sample throughput possible and the sensitivity and specificity of mass spectrometric detection demonstrated here should be particularly advantageous for large-scale mechanistic studies of carcinogenesis in animal models and epidemiological studies of human cancer. ACKNOWLEDGMENT We gratefully acknowledge the gift of mouse hybridoma cells from Dr. Regina Santella, Columbia University, and the assistance of Dr. William Tolleson, NCTR, with cell culturing. This research was supported in part by Interagency Agreement no. 224-93-0001 between NCTR/FDA and the National Institute for Environmental Health Sciences/National Toxicology Program. Received for review July 27, 2000. Accepted November 1, 2000. AC000866N Analytical Chemistry, Vol. 73, No. 2, January 15, 2001
309
