Characterization of a mutagenic DNA adduct formed from 1,2-dibromoethane by O6-alkylguanine-DNA alkyltransferase.

It has been proposed that the DNA repair protein O6-alkylguanine-DNA alkyltransferase increases the mutagenicity of 1,2-dibromoethane by reacting with it at its cysteine acceptor site to form a highly reactive half-mustard, which can then react with DNA (Liu, L., Pegg, A. E., Williams, K. M., and Guengerich, F. P. (2002) J. Biol. Chem. 277, 37920-37928). Incubation of Escherichia coli-expressed human alkyltransferase with 1,2-dibromoethane and single-stranded oligodeoxyribonucleotides led to the formation of covalent transferaseoligo complexes. The order of reaction determined was Gua>Thy>Cyt>Ade. Mass spectrometry analysis of the tryptic digest of the reaction product indicated that some of the adducts led to depurination with the release of the Gly136-Arg147 peptide cross-linked to a Gua at the N7 position, with the site of reaction being the active site Cys145 as established by chromatographic retention time and the fragmentation pattern determined by tandem mass spectrometry of a synthetic peptide adduct. The alkyltransferase-mediated mutations produced by 1,2-dibromoethane were predominantly Gua to Ade transitions but, in the spectrum of such rifampicin-resistant mutations in the RpoB gene, 20% were Gua to Thy transversions. The latter are likely to have arisen from the apurinic site generated from the Gua-N7 adduct. Support exists for an additional adduct/mutagenic pathway because evidence was obtained for DNA adducts other than at the Gua N7 atom and for mutations other than those attributable to depurination. Thus, chemical and biological evidence supports the existence of at least two alkyltransferase-dependent pathways for 1,2-dibromoethane-induced mutagenicity, one involving Gua N7-alkylation by alkyltransferase-S-CH2CH2Br and depurination, plus another as yet uncharacterized system(s).

O 6 -Alkylguanine-DNA alkyltransferase (AGT) 1 represents an important defense mechanism against the mutagenic, carcinogenic, and cytotoxic effects of simple alkylating agents (1)(2)(3). AGT repairs O 6 -alkylguanine adducts that are of major importance in initiating mutations and cytotoxicity. This repair process is mediated by transfer of the alkyl group from the O 6 atom of Gua to the active site Cys residue in the AGT protein. AGT is inactivated by the alkyl group transfer, and therefore additional repair requires new protein synthesis. Although an early study reported that the mutagenicity of 1,2dibromoethane (DBE) was not affected by enzymes that repair alkylation lesions (4), there is now convincing data that AGT paradoxically promotes the toxicity of DBE (5,6). Overexpression of human AGT (hAGT) or AGTs from other species enhances the mutagenicity and lethality of DBE in Escherichia coli (5)(6)(7)(8)(9) and induces growth retardation in mammalian cells (10).
Recently, we have shown that this toxicity is due to the interaction of DBE with the Cys acceptor site of AGT (Cys 145 in hAGT) (11). This Cys 145 residue is present in the hAGT protein as part of an extensive hydrogen-bonding network involving a water molecule and His 146 , Arg 147 , and Glu 172 (12). This environment causes Cys 145 to have a high reactivity with model electrophiles unrelated to alkylated DNA substrates and a very low pK a (13). It therefore reacts readily with DBE to generate a reactive intermediate, S-(2-bromoethyl)-Cys 145 -hAGT (11). This half-mustard intermediate is likely to cyclize forming an episulfonium ion that can react to form a covalent AGT-DNA adduct. Formation of DNA adducts is facilitated by the DNA binding properties of AGT.
DBE was used widely in gasoline as an anti-knock ingredient, as well as in pesticides and soil fumigants. Its use was drastically reduced due to reports of its toxicity. DBE is carcinogenic in rats, and toxic and mutagenic in microorganisms, plants, insects and humans (14 -16). Earlier studies with DBE have indicated that toxicity can be caused by metabolites formed via microsomal metabolism (17) and by the action of glutathione S-transferase (GST) (18 -20). The latter reaction generates S- (2-bromoethyl)glutathione, which undergoes a non-enzymatic dehalogenation and forms an episulfonium ion.
This ion then reacts rapidly with cellular nucleophiles such as DNA (21). The structures and mutagenic effects of major DBE-DNA adducts formed in this way have been studied extensively (19,22,23). Of these, S- [2-(N 7 -guanyl)ethyl]glutathione accounts for up to 95% of total DNA adducts (24) and its formation produces G:C to A:T transition mutations (25). GST was shown to enhance DBE genotoxicity in Salmonella typhimurium strains TA100 and TA1535, where mutations at specific guanines are needed to produce reversions (22,26,27). The extent to which the various pathways for activation of DBE (activation by microsomes, by GST or by AGT) contribute to its genotoxicity are not yet clear.
While previous studies provide a plausible model for the hAGT-mediated mechanism of DBE toxicity, there has been no direct evidence for the formation of hAGT-DNA adducts in vivo, structural characterization of hAGT-DNA adducts, or investigations of the types of mutations caused by these adducts. In the work described in this study, we describe evidence that AGT-DNA covalent cross-links are formed in cells, compare the reactivity of hAGT-DBE intermediates with different DNA bases, identify the mutations produced by DBE in cells lacking and expressing AGT, and characterize one of the hAGT-DNA adducts. Our results show that the predominant reaction is with Gua and that, in the presence of AGT, DBE treatment leads to a large increase in the frequency of G:C to A:T transitions and G:C to T:A transversions in E. coli. This evidence supports the involvement of both depurination and alternate mechanisms (e.g. error-prone DNA synthesis across sites) with hAGT adducts.
Recombinant C-terminal histidine-tagged wild-type hAGT was purified by Co 2ϩ -affinity chromatography using a BioCad Sprint Perfusion system (PerSeptive Biosystems) as described previously (28). E. coli TRG8 cells transformed with pIN-hAGT or an empty pIN vector (29) were grown at 37°C under the selection of 50 g/ml ampicillin and kanamycin (11) unless otherwise specified.
Most oligodeoxyribonucleotides (oligos) used in this study were synthesized by Invitrogen. The complementary 16-mer oligonucleotides 5Ј-d(TGCGTGAAGTGAGTGA)-3Јand 5Ј-d(TCACTCACTTCACGCA)-3Ј used for MS analysis were purchased from Midland Certified Reagent Co. (Midland, TX) (GF grade, identity, and purity confirmed by MALDI-TOF MS). These oligonucleotides were dissolved in H 2 O; equal concentrations of each were mixed, heated to 90°C, and then allowed to cool to room temperature (4 h time) to form a double-stranded oligomer for cross-linking studies.
Analysis of DBE-dependent AGT Binding to DNA in Vitro-Oligodeoxyribonucleotide 16-mers 5Ј-d(C) 16 -3Ј, 5Ј-d(TG) 8 -3Ј, or 5Ј-d(AG) 8 -3Ј were labeled with [␥-35 S]ATP␥S at the 5Ј-end as described previously (11). Radiolabeled oligonucleotides (6.5 pmol) were incubated with various unlabeled oligonucleotides (0 -1000 pmol) for 60 min at 37°C. The 15 l reaction mixtures also contained 20 mM DBE and wild-type hAGT (2 g) in AGT buffer (50 mM Tris (pH 7.6) and 0.1 mM EDTA). Reactions were terminated via the addition of 40 mM Tris (pH 6.8), 100 mM DTT, 2% SDS (w/v), 0.1% bromphenol blue (w/v) and 10% glycerol (v/v). The mixtures were then separated by polyacrylamide gel electrophoresis in the presence of sodium dodecyl sulfate (SDS-PAGE) using 12.5% gels and analyzed using a PhosphorImager SI system. The oligonucleotide bands were quantified using ImageQuant software (11). The percentage of oligonucleotide in the lower mobility hAGT-oligonucleotide complex forms was determined from these measurements and plotted against the concentrations of unlabeled oligonucleotides used in the reaction.
Oligonucleotide 16-mers 5Ј-d(TGCGTG*AAGGAGTGA)-3Ј including a Gua or an O 6 -methylguanine at the G* position were 5Ј-end labeled using [␥-35 S]ATP␥S. Mixtures of 6.5 pmol of [ 35 S]-labeled and 20 pmol of unlabeled oligonucleotides were added at 0 -70 min to a mixture of 2 g of hAGT protein and 20 mM DBE. After the addition of oligonucleotides, the incubation was allowed to proceed for another 60 min. Reaction mixtures were then separated by SDS-PAGE and analyzed using the PhosphorImager SI system.
Piperidine Cleavage Studies-The hAGT protein (8 g) was incubated with 100 pmol of 35 S-labeled 16-mer oligonucleotide 5Јd(C) 7 G(C) 8 -3Ј or 5Ј-d(T) 16 -3Ј in the presence or absence of 20 mM DBE or 1 mM DMS in AGT buffer. The 20-l incubation mixture was prepared in duplicate and the reactions were allowed to proceed for 1 h at 37°C. Piperidine (1 M) was then added to one set of the reaction mixtures, and these samples were heated at 90°C for 10 min. Aliquots of sample buffer (5 l) containing 0.25% bromphenol blue (w/v) and 40% (w/v) sucrose were then added to all samples. The mixtures were resolved by polyacrylamide gel electrophoresis in 40 mM Tris acetate, 1 mM EDTA, pH 8.0 (TAE-PAGE) using 20% gels. Gels were vacuum-dried and analyzed using a PhosphorImager SI system.
Identification of DBE-mediated Formation of Covalent DNA-AGT Complexes-E. coli TRG8 cells transformed with an empty pIN vector or wild-type pIN-hAGT were grown in 40 ml of LB at 37°C to an OD 600 of 0.5. The expression of the hAGT protein was induced using 0.2 mM IPTG. After a 1-h induction period, cells were exposed to concentrations of 0.035 or 0.2 mM DBE or vehicle (dimethylsufoxide (Me 2 SO)) for 90 min under constant shaking. Cells were collected by centrifugation and washed with 20 ml of 1ϫ M9 salt solution (90 mM Na 2 HPO 4 , 25 mM KH 2 PO 4 , 10 mM NaCl, and 20 mM NH 4 Cl). The cell pellets were stored at Ϫ80°C until analysis.
Genomic DNA from TRG8 cells was prepared using a Qiagen blood and cell culture DNA Maxi kit (Valencia, CA). The manufacturer's protocol for genomic DNA prepared from Gram-negative bacteria was adapted with minor modifications. Heat-inactivated RNase and lysozyme were added to buffer B1 to lyse the cells. Protease K was excluded to preserve the full-length hAGT protein, and NaCl (250 mM) was added to buffers B1 and B2 to reduce the noncovalent binding of proteins to DNA. E. coli genomic DNA was precipitated from the QF (Qiagen, Valencia, CA) elution solution using a mixture (v/v) of 70% isopropyl alcohol and 30% 0.4 M NaCl, followed by centrifugation at 14,000 ϫ g for 15 min. The DNA was dissolved in 100 -300 l of deionized water by rotation-mixing at room temperature overnight.
Aliquots of E. coli genomic DNA (4 g) were partially digested with DNase I (0.4 unit/g DNA) at 37°C for 30 min. The mixtures were resolved by electrophoresis using 15% SDS-PAGE and transferred to a polyvinylidene difluoride membrane. The presence of hAGT protein was determined using a specific monoclonal antibody (Clone MT 3.1; Lab Vision Corp., Fremont, CA) and visualized using the Phototope TM -HRP Detection Kit.
Analysis of Mutants Induced by DBE in the hisG Gene of S. typhimurium and the lacI Gene of E. coli-For studies of S. typhimurium, the strain YG7108 (derivative of TA1535, lacking both endogenous AGT genes (30)) was transformed with pIN and pIN-AGT vectors. For studies in E. coli, strain TRG8 (8) was transformed with pIBA2-AGT (which produces active AGT) or pIBA2-C145A, which produces an inactive mutant. These plasmids were constructed by PCR amplification from pIN-AGT and pIN-C145A plasmids (31) using the upper primer 5Ј-GC-ACTGACCGTCTAGATTAAAGAGGAGAAAGG-3Ј, which introduces a XbaI restriction site just upstream of the ribosome binding site and restores the open reading frame of AGT, containing 5 extra amino acid residues in pIN-AGT, to its wild-type condition, and the lower primer 5Ј-GTTGCGCAGAAGCTTCCACTCAGTTTCG-3Ј, which changes a BamHI site to a HindIII site downstream of the stop codon. The amplification reaction generated 680 bp DNA fragments that were cut with XbaI and HindIII. The restricted fragments were ligated to a pASK-IBA2 vector (IBA GmbH, Germany) previously cut with the same enzymes, to produce pIBA2-AGT and pIBA2-C145A.
The measurement of cell survival and mutation was according to the protocol described by Liu et al. (8) with minor modifications: for TRG8/ pIBA2-AGT transformants, anhydrotetracycline (100 ng/ml) was employed for induction of AGT expression. Aliquots of the cell cultures (0.5 ml) were exposed to various concentrations of DBE (0.035-1.2 mM) or to the solvent Me 2 SO for 30 -90 min at 37°C. For cell survival estimation, the cells were washed with 1ϫ M9 salts, diluted 10 2 to 10 5 fold, and plated on a non-restrictive medium (M9 minimal medium supplemented with 0.2% glucose (w/v), 50 g/ml each ampicillin and kanamy-cin, 40 g/ml histidine) (32) and incubated for 2-3 days at 37°C. The percentage survival was quantified by the number of colonies of bacteria exposed to DBE over that of untreated bacteria. For mutant isolation, after DBE treatment, undiluted cells were plated on M9 minimal medium plates (32) containing 0.7 mg/ml phenyl-␤-D-galactoside as the sole carbon source for selection of mutants that synthesize ␤-galactosidase constitutively (lacI Ϫ or lacO c ) or Vogel-Bonner minimal medium (33) with excess biotin and trace amounts of L-histidine for isolation of S. typhimurium hisG ϩ revertants (34). The plates were incubated at 37°C for 3 days, and the number of colonies on the plates was determined. The mutation frequency was calculated by the number of colonies grown on the selective media over 10 8 survivors grown on the glucose-containing plates.
The colonies that grew on selective plates were purified by streaking on Petri dishes containing the same selective agent. These bacterial colonies were used in a PCR amplification reaction using the cells directly instead of purified DNA as a template. After incubation for 10 min at 95°C to break the cells and release template DNA, Pfu ultra DNA polymerase was added, and amplification carried out with primers 5Ј-TGAATGTGAAACCAGTAACG-3Ј and 5Ј-GCTCACTGCCCGCTT-TCCA-3Ј for the analysis of E. coli lacI gene and 5Ј-GTTAGACAACAC-CCGCTTACGC-3Ј and 5Ј-GGAGGTGCGGATATGAGGTAGC-3Ј for the S. typhimurium hisG gene. The amplification products were purified using the QIAquick PCR purification kit (Qiagen) and both strands sequenced using the same primers employed for amplification in separate reactions at the Vanderbilt Core DNA Sequencing Facility, Vanderbilt University Medical Center.
Analysis of Rifampicin-resistant Mutants Induced by DBE-TRG8 cells were grown in 50 ml of LB media containing 0.2 mM IPTG to an OD 600 of 0.5. Cells were pelleted by centrifugation and resuspended in 2 ml of M9 salts. Aliquots of cells (0.5 ml) were then exposed to various concentrations of DBE (0.035-0.1 mM) or to the solvent Me 2 SO at 37°C for 90 min. The cells were washed with M9 salts and resuspended in 0.5 ml of the same solution. In order to derive rpoB gene mutants, cells were plated on LB media plates supplemented with 100 g/ml rifampicin. Undiluted cell suspensions (100 l) were plated either immediately after the treatment or following an additional overnight culture in LB media at 37°C. The cells were grown in a 37°C incubator for ϳ36 h until discrete colonies appeared. In addition, the overnight cultures were diluted 1:1.2 ϫ 10 7 -fold and plated onto LB plates lacking rifampicin to determine the number of viable cells. For cells plated immediately after treatment, a 1:10 4 -10 6 fold dilution was utilized. The mutation frequency of the rpoB gene in TRG8 cells was expressed as the number of rpoB mutants per 10 8 survivors.
Rifampicin-resistant cell clones from rifampicin-containing plates were suspended in 100 l of deionized water. Aliquots (2-5 l) of these suspensions were used as DNA templates in PCR. A section of the rpoB gene was amplified by PCR using 5Ј-TGGCCTGGTACGTTGAGA-G3Ј (forward primer) and 5Ј-AACCAGCGGCTTTACAGC-3Ј (reverse primer). The setting for the thermocycler were: 92°C for 4 min, 35 cycles of 92°C for 30 s, 52°C for 30 s, and 72°C for 30 s; 72°C for 5 min. Sequences of the PCR products were examined by DNA sequencing using the reverse primer carried out at the Macromolecular Core Facility of the Milton Hershey Medical Center, Pennsylvania State University.
MALDI-TOF Analysis of DBE-mediated AGT-DNA Adducts-A solution of calf thymus DNA (1 mg/ml) was sonicated using 6 pulses of 10 s each. DNA (45 g in 100 l of diisopropylethylamine buffer, pH 6.7) was incubated with 300 g of hAGT and 20 mM DBE (from a 500 mM stock in CH 3 CN) at 37°C for 1 h and then digested with trypsin (25 g) at 37°C overnight.
MALDI-TOF spectra were recorded on a Perseptives Voyager Elite instrument (Applied Biosystems, Framingham, MA) in the positive reflector mode. The accelerating voltage, grid voltage, and guide wire voltages were 20,000, 74.5%, and 0.05%, respectively. A 10 mg ml Ϫ1 solution of ␣-cyano-4-hydroxycinnamic acid in 60:40 CH 3 CN:2% aqueous CF 3 CO 2 H (v/v) was used as the matrix. Samples used for MALDI analysis were diluted in 0.10% CF 3 CO 2 H (v/v) and then desalted with ZipTip C18 pipette tips (Millipore, Milford, MA). Bradykinin and ␤-endorphin were used as external standards to calibrate the mass spectrometer.
Modification of Peptide Gly 136 -Arg 147 -The synthetic peptide was reacted with DBE and dGuo to generate modifications of the peptide with -CH 2 CH 2 -dGuo and -CH 2 CH 2 -(N 7 )Gua at Cys 145 of the peptide, for use as standards for analyzing the tryptic peptides derived from hAGT/ DBE/DNA reactions. The peptide (380 nmol, in 0.5 ml of 0.10 M NaHCO 3 buffer, pH 8.7) was mixed with dGuo and DBE (10 mM each) and allowed to react at 37°C for 3 h. Potassium phosphate buffer (100 l of a 1.0 M solution, pH 6.0) was added (final pH 7.0). One reaction was heated at 95-100°C for 60 min to cleavage the glycosidic bond of any guanyl-N 7 adduct. Another sample was analyzed directly for adducts containing dGuo.
HPLC/MS was performed using an Agilent 1100A system (Agilent) with a Phenomenex Jupiter octadecylsilane (C 18  ). The analysis indicated that the peptide-CH 2 CH 2 -Gua product eluted at t R 9.63 min and the peptide-CH 2 CH 2 -dGuo product eluted shortly thereafter at t R 9.9 min (separate runs), with the appropriate MH ϩ ions. The apparent yields (as judged by total ion current) were low (Ͻ10%); the most abundant products were peptide(s) cross-linked by Cys dimers (cystine) and by ethylene links. Collision-induced dissociation (CID) of the suspected ions yielded the appropriate b and y fragmentation patterns (see Fig. 5

and Supplementary Data).
Cross-linking of hAGT and Double-stranded Oligomer for MS Analysis of Depurinated Products-hAGT (10 nmol, in 0.40-ml total volume) was incubated with 23 nmol of double-stranded 16-mer oligonucleotide (from Midland, see above) and 20 mM DBE at 37°C for 60 min. The solution was dialyzed at 4°C over a 3-day period versus 100 volumes of 80 mM sodium acetate (pH 5.0) to remove residual DBE (3 changes). The hAGT precipitated and was recovered by centrifugation (10 3 ϫ g, 5 min). The pellet was suspended in 200 l of 0.10 M sodium phosphate (pH 7.6) containing 0.9% SDS (w/v) and heated at 90 -95°C for 60 min, to depurinate DNA adducts. A 100-l aliquot (1/2) was diluted to a volume of 2.0 ml in 0.10 M NaHCO 3 (pH 8.0), bringing the SDS concentration to 0.045% (w/v). Trypsin was added at a 1:10 ratio (trypsin: hAGT) by weight and incubated at 37°C for 14 h, followed by another addition of the same amount of trypsin and incubation for 7 h more at 37°C. SDS-PAGE showed extensive proteolysis, and the digested sample was used for HPLC/MS-ESI analysis under conditions similar to those used for the analysis of the standard peptide (Gly 136 -Arg 147 ) modified with DBE and dGuo (see above).
Cross-linking of hAGT and Double-stranded Oligomer for Nuclease Digestion and MS Analysis-An incubation was set up as in the case of the depurination experiment, except that the amounts of each component were doubled (0.80 ml volume). After 60 min, 200 l of 1 M NaHCO 3 (pH 8.7) and 100 l of 0.4 M dithiothreitol were added, and the reaction was allowed to stand at room temperature for 2 h, to conjugate residual DBE. The sample was dialyzed overnight (at 4°C) versus 200 ml of 50 mM Tris-HCl buffer (pH 8.7) containing 15 mM MgCl 2 . The sample was then incubated for 4 h (37°C) with 80 g of phosphodiesterase I; then 80 g of alkaline phosphatase was added, and the incubation proceeded for another 4 h at 37°C. The sample was dialyzed overnight versus 100 volumes of 50 mM potassium phosphate buffer (pH 8.0) and hAGT was recovered by applying the sample to a 0.4 ml Ni 2ϩ -nitrilotriacetate column pre-equilibrated with 20 mM potassium phosphate (pH 8.0), washing with 15 ml of the equilibration buffer, and eluting with the same buffer containing 200 mM imidazole (at room temperature). The eluted fractions containing hAGT (as measured by A 280 measurements, recording UV spectra on a Cary 14/OLIS spectrophotometer (OLIS, Bogart, GA)) were pooled (total volume ϳ3 ml) and dialyzed versus 250 ml of 10 mM NaHCO 3 buffer (pH 8.5) overnight at 4°C. The sample was concentrated to dryness by lyophilization and reconstituted in 1/10 the original volume. Trypsin was added (2.5 g, ϳ1:10 ratio to hAGT by weight) and the reaction proceeded overnight at 37°C; another 2.5 g of trypsin was added and the reaction continued for 8 h at 37°C. The sample was frozen at Ϫ20°C until HPLC/MS analysis (10 days later).

Relative Reactivity of Bases with hAGT and DBE-
The formation of a covalent complex between hAGT, DBE, and 5Ј-[ 35 S]-labeled oligonucleotides (11) was used to investigate the relative reactivity of the four deoxyribonucleosides in forming hAGT-DNA adducts. The ability of unlabeled 16-mers consist-ing of repeats of a single nucleoside or two nucleosides to prevent the formation of this radioactive complex was used to determine their reactivity (Fig. 1). As shown in Fig. 1A, hAGT was linked to [ 35 S]d(C) 16 forming complexes with either one (complex 1) or two (complex 2) hAGT proteins (Fig. 1A). Approximately 28% of the d(C) 16 was present in complexed forms (in the absence of any unlabeled oligonucleotide), and the amount of radiolabeled DNA-hAGT complexes declined as the levels of unlabeled d(C) 16 , d(T) 16 , or d(A) 16 increased from 0 to 350 pmol (Fig. 1A). d(T) 16 competed most effectively and the d(A) 16 was least effective (Fig. 1B), indicating an apparent order of reactivity of TϾCϾA for hAGT-DNA conjugation. Studies with d(G) 16 could not be carried out due to problems in the synthesis of such homopolymers; therefore, reactivity of Gua was compared with Thy using [ 35 S]d(TG) 8 in the presence of unlabeled d(T) 16 and d(TG) 8 (Fig. 1, C and D). The latter was more effective indicating that Gua is more reactive than Thy. Taken together, these data indicate that the overall order of reactivity for the four bases is GuaϾThyϾCytϾAde. Similar results were obtained using other combinations of labeled and unlabeled oligonucleotides; e.g. we confirmed that Thy is more reactive than Ade by comparing the reactivity of d(TG) 8 and d(AG) 8 (Fig. 1, E and F).
AGT repairs O 6 -methylguanine and the possibility was considered that the DBE-mediated conjugation of the protein might be greatly influenced by the presence of an O 6 -methyl-Gua. This experiment is complicated by the possibility that hAGT would repair the O 6 -methylGua lesion rather than react with DBE. Therefore, hAGT was incubated with DBE for 0 -60 min prior to addition of oligonucleotides, 5Ј- activated hAGT-Cys 145 SCH 2 CH 2 Br, which is converted to hAGT-Cys 145 -SCH 2 CH 2 OH in the absence of DNA or other reaction substrates (11).
On the basis of many previous studies of reactions of alkylating agents with Gua in DNA and particularly the episulfonium ion formed from S-(2-bromoethyl)glutathione (21,24), reaction with the N 7 position of Gua is a potentially major site of attack. Lesions where Gua is modified at N 7 can be identified by using the ability of piperidine to cleave at apurinic sites derived from thermal release of the adduct (22,35). Therefore 5Ј-[ 35 16 was reacted with DBE and hAGT and also with DMS, which causes alkylation prevalently on the N 7 -position of guanine (35) to act as a positive control. The products were then treated with piperidine and separated by TAE-PAGE (Fig. 3). No alteration in mobility was seen when d(T) 16 was used but a small part of the 5Ј-d(C) 7 G(C) 8 -3Ј was cleaved to form the expected 7-mer when reacted with both DBE and hAGT. More extensive cleavage was observed with the 5Ј-d(C) 7 G(C) 8 -3Ј after it had been reacted with DMS (Fig.  3). These results are consistent with at least part of the covalent adducts formed by DBE-activated hAGT with Gua being at the N 7 position.

S]d(C) 7 G(C) 8 -3Ј or [ 35 S]-d(T)
MS of hAGT Modifications-A number of technical difficulties are associated with analysis of nucleic acid-protein crosslinks. MS of peptide-nucleic acid complexes has been very limited because optimal conditions developed for either peptides or nucleic acids (either MALDI-TOF or ESI) are negated by the presence of the other component. Chemical methods of digestion of the complexes are too harsh to preserve most linkages. Enzymatic methods have the problem of achieving digestion with steric hindrance, i.e. the protein (or peptide) may block nucleases and the nucleic acid may block proteases. Another problem specific to the work with hAGT is the decreased solubility following modification of Cys 145 .
Several approaches were used in efforts to characterize hAGT interactions with DNA. Obtaining mass spectra of a tryptic peptide (Gly 136 -Arg 147 ) with Cys 145 -CH 2 CH 2 OH after reaction with DBE alone was relatively straightforward, and we had previously reported both MALDI-TOF and ESI spectra (11). We were able to recover and identify the peptide using a more sensitive instrument in this study and use CID to fragment the peptide to the expected b and y ion series (see Supplementary Data). We also found evidence for alkylation of other Cys residues of hAGT under extensive conditions, including a residue in peptide 148 -165 (presumably due to alkylation of Cys 150 , modified ϳ 10% of the extent of G136-R147).
hAGT was reacted with calf thymus DNA in the presence of 20 mM DBE for 1 h and then subjected to trypsin digestion. In the MALDI-TOF spectrum (Fig. 4), a peak with an m/z of 1493 corresponded to the G136-R147 fragment cross-linked to a gua-nine base. A peak at m/z 1359 corresponded to the 2-hydroxyethyl adduct at Cys 145 described above.
Preliminary ESI analysis was also consistent with the presence of the ϩ2 and ϩ3 charged ions of this adduct. This data would correspond to depurination, presumably of an N 7 -guanyl adduct. Although a number of other procedures were tried (see Supplementary Data), we found that we could best characterize this adduct by optimizing the protocols to those presented in the Experimental Procedures section. Briefly, a 60-min incubation of hAGT with DBE and a double-stranded oligonucleotide was followed by a 60 min heat step (95°C, neutral pH), solubilization of the denatured protein in SDS with heating, dilution to lower the SDS concentration, trypsin digestion, and HPLC/MS-ESI with comparison to a sample prepared from the reaction of the synthetic peptide (Gly 136 -Arg 147 ) with DBE and dGuo at alkaline pH. The same product was identified from an experiment in which hAGT, DBE, and the double-stranded 16-mer were incubated followed by reaction with DTT at alkaline pH to conjugate residual DBE, dialysis at alkaline pH, digestion with phosphodiesterase I and then alkaline phosphatase (at alkaline pH), isolation of the hAGT using Ni 2ϩ -nitrilotriacetate chromatography (elution with imidazole), dialysis, lyophilization, and digestion with trypsin, The resulting data are presented in Fig. 5 and confirm the depurination product.
The sample prepared from the incubation of synthetic peptide (Gly 136 -Arg 147 ), DBE, and dGuo was also used as a reference material to search for hAGT-derived peptide cross-linked to dGuo and other nucleosides in the above sample that had been digested enzymatically. Only in one experiment could we tentatively identify dGuo nucleoside adducts (and were unable to further verify this by fragmentation analysis), and we did not detect any of the other three nucleosides in adducts. The absence of these adducts may be due to limits of sensitivity, which may vary depending upon the nucleoside. Further, the standard peptide Gly 136 -Arg 147 -CH 2 CH 2 -dGuo adduct (see Supplementary Data) apparently degraded to the corresponding Gua adduct during handling and storage.
Formation of hAGT-DNA Adducts in Vivo-E. coli TRG8 cells, which lack endogenous AGT protein, were transformed with an empty vector control plasmid or a plasmid leading to expression of hAGT. The cells were treated with DBE or vehicle (Me 2 SO) for 90 min under constant shaking. Genomic DNA was isolated using a procedure in which any non-covalent hAGT-DNA complexes were disrupted by the presence of the detergent and 250 mM NaCl. The DNA was digested with DNase I, separated by SDS-PAGE, and then subjected to immunoblot analysis using a monoclonal antibody specific for hAGT. A signal was detected only in the cells expressing hAGT and treated with DBE and was much stronger in the cells treated with the higher concentration of DBE (Fig. 6). The smear is probably due to incomplete digestion of all of the DNA since a low concentration of DNase was used. This also may account for the presence of complexes in Fig. 6 that clearly contain two hAGT molecules, which were mostly noticeable in cells treated with 0.35 mM DBE for 90 min. However, this may also indicate that two hAGT molecules are linked to each other. This is consistent with findings that binding of AGT to DNA is highly cooperative placing two molecules in close proximity to each other (36) and such dimers were seen in reaction of hAGT with DBE and oligos in vitro (11) (Fig. 1). The covalent AGT-DNA adduct was not seen when DNA was prepared from cells exposed to Me 2 SO vehicle or 1,2-dichloroethane, a chemical that does not react with hAGT in vitro, nor form hAGT-DNA adducts (data not shown).
Determinations of Mutation Spectra Caused by DBE and hAGT-The ability of AGT to mediate mutations by DBE at G:C pairs was demonstrated using S. typhimurium YG7108, a strain derived from TA1535 with additional alterations that inactivate endogenous AGT genes (30). The strain contains mutation hisG46 imparting a requirement for histidine that results from an inactivating mutation forming a CCC codon. This can be reverted by any base change at the first base and by a G:C to A:T transition and G:C to T:A transversion at the second base of the codon. The expression of hAGT caused a large increase in the formation of revertants in response to DBE (Fig. 7A). The hisG gene of some mutants obtained was PCR amplified and sequenced. In all cases, the mutations mediated by hAGT were G:C to A:T transitions in the first or the second position (29% were TCC and 71% were CTC).
These results were further supported by studies of the mutation spectra of mutants produced by 1 mM DBE in the lacI gene. Expression of active AGT, but not the inactive C145A mutant, increased the frequency of such mutations by more than 100-fold (Fig. 6B). All 44 mutants analyzed were single base pair substitutions, 43 were transitions and predominantly (41/43) G:C to A:T base pairs. There were also two A:T to G:C transitions, and the other base substitution corresponded to a G:C to T:A transversion. These mutations were distinct from those spontaneous mutants detected in cells not containing hAGT and were not randomly distributed along the lacI open reading frame. Instead, six nucleotide residues clustered two thirds of the substitutions where two or more mutations were found. Mutation of the mid base of codon 222 was the most frequent (nine independent mutants), followed by codons 118 (six mutants), 68 and 58 (four mutants each), 27 (three mutants) and 187 (two mutants).
In order to obtain a more complete picture of the mutation spectrum, the occurrence of rifampicin resistant mutants, which result from alterations in the rpoB gene, was examined. The rpoB gene, which encodes the ␤-subunit of RNA polymerase II, has been widely used as a marker to examine the spectra of mutants arising endogenously or through induction by exogenous reagents (38,39). The frequency of rifampicin-resistant (Rif r ) mutants produced by DBE was greatly increased by the expression of hAGT (Fig. 7C). Sequence analysis of the mutants (Table I) indicated that in DBE-treated cells expressing hAGT nearly all of the mutations were G:C to A:T transitions (76%) or G:C to T:A transversions (20%). The spectrum of mutations was quite different from that observed in cells lacking hAGT (but treated with DBE) where 39% of mutations correspond to A:T to C:G transversions, 33% of G:C to A:T transitions, 17% A:T to T:A tranversions, 5.5% A:T to G:C transitions and 5.5% G:C to T:A transversions (Table I). The sites of mutation in cells expressing hAGT were also different from those in cells lacking hAGT ( Table I). Exposure of cells to solvent Me 2 SO did not cause any significant increase in mutations in the rpoB gene irrespective of hAGT status. Also, the spectra of background mutants in TRG8 cells did not reveal any specific induction of G:C to A:T or G:C to T:A mutations (results not shown) and was similar to that seen in a recent large study (39). DISCUSSION Previous work by other groups established the paradoxical enhancement of the mutagenicity of DBE by expression of AGT (5-7). These results were confirmed in one of our laboratories (8), and subsequently we reported the cross-linking of AGT to DNA, with the reaction initiated by the conjugation of AGT with DBE (11). The reaction of AGT with DBE is due to the high reactivity of the Cys 145 thiolate anion with alkylating agents (13) and the formation of a half-mustard and probably episulfonium ion (21) that can cross-link to DNA (11). However, further information about how this process could lead to mutations was not available. We now provide evidence for an N 7 -Gua/depurination mechanism to explain part of the mutations.
As shown in Fig. 1, the intermediate formed by the reaction of DBE with hAGT in vitro can react with all four nucleotides. Reaction was strongest with Gua and the second most reactive base in vitro was Thy. These bases are the sites of AGT reaction (2,40), but the difference in reaction may simply result from the availability of appropriate nucleophilic sites at locations that can react with the episulfonium ion intermediate. Although reaction with Gua was the most favored, the in vitro studies showing reaction with all four bases contrast somewhat with the mutations seen in vivo which were virtually exclusively at G:C pairs. The Cys 145 acceptor site of hAGT is located in an internal pocket and nucleoside "flipping" induced by the AGT protein is needed to allow the potential substrate to be placed in proximity to this residue (12,41). Even allowing for the additional size of the activated DBE adduct and the possible distortion in the protein that may arise from its formation, it is unlikely that covalent attachment of the activated AGT could occur readily without this change on the DNA structure. Many aspects of the mechanism by which AGT finds lesions in DNA are unclear, for example whether it samples the DNA by flipping every base or is directed toward some sites by minor distortions in the DNA. The preference for Gua seen in our in vitro experiments may therefore be magnified in cellular DNA, which is predominantly double stranded. The presence of an O 6 -methylguanine residue had no effect on the covalent attachment of DBE-activated AGT to DNA in vitro (Fig. 2). This result is not very surprising because hAGT does not bind much more strongly to DNA containing O 6 -methylguanine than to DNA without methyl adducts (36,42,43). However, the result makes the important point that the formation of hAGT-DBE-DNA crosslinks is not dependent on other processes such as endogenous or exogenous methylation generating sites for AGT action.
Definitive proof that AGT can be cross-linked to DNA in the presence of DBE is provided by the MS results. First, the data shown in Figs. 4 and 5 support the assignment as hAGTderived peptide 136 -147 bound to -CH 2 CH 2 -Gua through modification at Cys 145 . Such release of the Gua adduct under conditions of neutral heating is very consistent with the evidence shown for the hot piperidine treatment results in Fig. 3. The MS results are only consistent with the linkage of the Gua at the N 7 atom, leading to depurination. N 3 -Ade adducts also depurinate, but we did not find MS evidence for any Adecontaining adducts when we searched the expected m/z ions.
However, not all of the binding at Gua can be explained by the guanyl N 7 -modification. In other experiments (not shown), hAGT and calf thymus DNA were reacted with [ 14 C]DBE. The stoichiometry is somewhat problematic because of the difficulty in correcting for hydroxyethyl-hAGT (not bound to DNA), but in several experiments a mean of ϳ1/3 of the radioactivity bound to DNA (precipitated by centrifugation after 16 h at 10 5 ϫ g (19) remained bound after neutral thermal depurination. No N 7 -Gua adducts should have remained following the 60 min heat step (95°C) at neutral pH (24,44). Additional evidence for the existence of other DNA adducts was obtained in the piperidine cleavage experiments where only a fraction of the material was cleaved (Fig. 3). All of the studies on the mutation spectra are consistent with these mutations being derived from a Gua alteration caused by DBE in the presence of AGT. The presence of AGT caused a massive increase in His ϩ revertants in E. coli GWR109 (derived from F26) (11) similar to the effect of exposure to Nmethyl-NЈ-nitro-N-nitrosoguanidine, which is well-known to produce predominantly G to A transitions (45). All of the re-versions of hisG46 in S. typhimurium YG7108 were G to A transitions. Similarly, G:C sites represented the overwhelming majority of mutants in the lacI gene (41/43) and in the rpoB gene (24/25).
The rpoB gene is now firmly established as a suitable target for the facile detection of all classes of mutations (39). Both G:C to A:T transitions and G:C to T:A transversions leading to rifampicin resistance were observed in AGT-mediated response to DBE (Table I). The G to T transversions, which represent 20% of the mutations we observed in the rpoB gene are consistent with depurination of the Gua adduct and subsequent processing of the abasic site (46,47). However, in both the rpoB and the lacI genes, the most common mutations were G to A transitions. At present, the lesion leading to this type of mutation is unknown but as discussed above, there is clearly one or more additional adducts that we were not able to identify in the MS experiments. Although currently we have no evidence for the formation of O 6 -Gua adducts, it is worth pointing out that DBE treatment not only converts AGT into a DNA damaging agent but it also depletes the AGT activity and would therefore retard the repair of any O 6 -Gua adducts that are substrates for AGT.
The results described here provide strong support for the hypothesis that AGT increases the mutagenicity of DBE by reacting with it to form a reactive complex that leads to the attachment of AGT to DNA. Such a complex was detected in vivo (Fig. 6) after treatment with DBE and unequivocal evidence for a guanine adduct formed by the reaction of DBE and hAGT with DNA in vitro was obtained by MS analysis (Figs. 4  and 5). Also, 20% of the DBE-induced rifampicin resistant mutations that were dependent on hAGT expression were G:C to T:A transversion mutations that are likely to have arisen from the depurination of N 7 adducts to generate abasic sites. Such sites are cytotoxic and highly mutagenic giving such FIG. 7. Effect of hAGT in mutation frequency in response to DBE. A, S. typhimurium YG7108 transformed with pIN or pIN-AGT vectors as indicated was treated with DBE for 30 min and the frequency of His ϩ revertants determined. B, E. coli TRG8 expressing the wild type (filled symbols) or mutant C145A AGT (open symbols) contained in the pIBA2 construct in the presence (squares), or the absence (circles) of the transcriptional inducer anhydrotetracycline (AhTc) were used and mutations in the LacI gene were scored. C, E. coli TRG8 cells transformed with pIN or pIN-hAGT as indicated were exposed to 0 -0.1 mM DBE for 90 min. Rifampicin resistant mutants were isolated on LB plates containing 100 g/ml rifampicin. The mutation frequency of the rpoB gene was calculated based upon the number of survivors. transversions if not properly repaired. At present, we cannot definitively identify the adduct or adducts that give rise to the majority of the mutants which were G:C to A:T transitions and this is the subject of ongoing investigation. These could arise from an AGT protein-DNA cross-link at either the O 6 or N 2 positions. The extent to which by-pass DNA polymerases (37,48) are able to extend past such protein-DNA cross-links is unclear and there may be an initial "repair" event that removes part of this bulky adduct and leaves a lesion that is more readily by-passed. In addition to the possible effect from abasic sites described above, the large increase in cell killing by DBE mediated via AGT (5-9) may be due to the difficulties of DNA replication on a template that contains protein-DNA cross-links.
A scheme for the role of AGT in the genotoxicity of DBE is shown in Fig. 8. These mechanisms may provide a very efficient way of causing DNA damage, because the reactive species generated at the active site of AGT can be directed to DNA by the DNA binding properties of the protein. It will be important to determine the extent to which AGT-mediated effects contribute to the toxicity of DBE as compared with the other known activation mechanism (GST) by using cells/tissues containing physiological levels of these activation mechanisms (Fig. 8). Our observations of the AGT-dependent mutation spectra combined with those previously published of the GST-and microsomal mediated mutations will provide tools to carry out this analysis. Future studies are also needed to test the role that other DNA repair pathways play in protecting from DBE (4,6).
In summary, we have characterized the existence of a Gua-N 7 adduct involving hAGT and its Cys 145 and provide evidence that depurination can lead to mutations in bacterial cells. The work also provides evidence that this is not the exclusive pathway for AGT-enhanced mutations induced by DBE (Fig. 8). Further studies on the non-apurinic site mechanism are ongoing, as well as work with other bis-electrophiles that might involve similar mechanisms. These phenomena are likely to occur in mammalian cells as well as bacteria, although the relative contribution of the pathways shown in Fig. 8 remains to be elucidated.