Characterization of nucleoside and DNA adducts formed by S-(1-Acetoxymethyl)glutathione and implications for dihalomethane - Glutathione conjugates

Article abstract of DOI:10.1021/tx010006h

S-(1-Acetoxymethyl)glutathione (GSCH₂OAc) was synthesized and used as a model for the reaction of glutathione (GSH)-dihaloalkane conjugates with nucleosides and DNA. Previously, S-[1-(N²-deoxyguanosinyl)methyl]GSH had been identified as the major adduct formed in the reaction of GSCH₂OAc with deoxyguanosine. GSCH₂OAc was incubated with the three remaining deoxyribonucleosides to identify other possible adducts. Adducts to all three nucleosides were found using electrospray ionization mass spectrometry (ESI MS). The adduct of GSCH₂OAc and deoxyadenosine was formed in yield of up to 0.05% and was identified as S-[1-(N⁷-deoxyadenosinyl)methyl]GSH. The pyrimidine deoxyribonucleoside adducts were formed more efficiently, resulting in yields of 1 and 2% for the GSCH₂OAc adducts derived from thymidine and deoxycytidine, respectively, but their lability prevented their structural identification by ¹H NMR. On the basis of the available UV spectra, we propose the structures S-[1-(N³-thymidinyl)methyl]GSH and S-[1-(N⁴-deoxycytidinyl)methyl]GSH. Because adduct degradation occurred most rapidly at alkaline and neutral pH values, an enzymatic DNA digestion procedure was developed for the rapid hydrolysis of DNA to deoxyribonucleosides at acidic pH. DNA digests were completed in less than 2 h with a two-step method, which consisted of a 15 min incubation of DNA with high concentrations of deoxyribonuclease II and phosphodiesterase II at pH 4.5, followed by incubation of resulting nucleotides with acid phosphatase. Analysis of the hydrolysis products by HPLC-ESI-MS indicated the presence of the thymidine adduct.

Full text of DOI:10.1021/tx010006h

600
Chem. Res. Toxicol. 2001, 14, 600-608
Ch a r a cter iza tion of Nu cleosid e a n d DNA Ad d u cts
F or m ed by S-(1-Acetoxym eth yl)glu ta th ion e a n d
Im p lica tion s for Dih a lom eth a n e-Glu ta th ion e Con ju ga tes
Glenn A. Marsch,‡,§ Ralf G. Mundkowski,| Brent J . Morris,‡,§ M. Lisa Manier,‡
Melanie K. Hartman,‡ and F. Peter Guengerich*,‡
Department of Biochemistry and Center in Molecular Toxicology, Vanderbilt University School of
Medicine, Nashville, Tennessee 37232, and Department of Chemistry and Physics,
Union University, J ackson, Tennessee 38305
Received J anuary 9, 2001
S-(1-Acetoxymethyl)glutathione (GSCH2OAc) was synthesized and used as a model for the
reaction of glutathione (GSH)-dihaloalkane conjugates with nucleosides and DNA. Previously,
S-[1-(N2-deoxyguanosinyl)methyl]GSH had been identified as the major adduct formed in the
reaction of GSCH2OAc with deoxyguanosine. GSCH2OAc was incubated with the three
remaining deoxyribonucleosides to identify other possible adducts. Adducts to all three
nucleosides were found using electrospray ionization mass spectrometry (ESI MS). The adduct
of GSCH2OAc and deoxyadenosine was formed in yield of up to 0.05% and was identified as
S-[1-(N7-deoxyadenosinyl)methyl]GSH. The pyrimidine deoxyribonucleoside adducts were
formed more efficiently, resulting in yields of 1 and 2% for the GSCH2OAc adducts derived
from thymidine and deoxycytidine, respectively, but their lability prevented their structural
identification by 1H NMR. On the basis of the available UV spectra, we propose the structures
S-[1-(N3-thymidinyl)methyl]GSH and S-[1-(N4-deoxycytidinyl)methyl]GSH. Because adduct
degradation occurred most rapidly at alkaline and neutral pH values, an enzymatic DNA
digestion procedure was developed for the rapid hydrolysis of DNA to deoxyribonucleosides at
acidic pH. DNA digests were completed in less than 2 h with a two-step method, which consisted
of a 15 min incubation of DNA with high concentrations of deoxyribonuclease II and
phosphodiesterase II at pH 4.5, followed by incubation of resulting nucleotides with acid
phosphatase. Analysis of the hydrolysis products by HPLC-ESI-MS indicated the presence of
the thymidine adduct.
Sch em e 1. Du a l P a th w a y for CH2Cl2 Meta bolism
In tr od u ction
CH2Cl2 is a solvent used in the United States at the
level of 3 × 108 kg/year (1, 2). Although CH2Cl2 is not
extremely toxic to humans, its biological effects are of
concern because of the risk of widespread exposure to the
compound in the chemical and other industries (3, 4). A
major concern is the possible carcinogenicity (2, 4);
CH2Cl2 is known to be tumorigenic in mice but not in
rats (5, 6). Due to the potential danger of CH2Cl2 to
workers in the chemical industry and others exposed to
the compound in large quantities, the Environmental
Protection Agency and the Occupational Safety and
Health Administration have regulated CH2Cl2 exposure
in the workplace (4).
The second pathway is through GSH conjugation by GSH
S-transferase (GST)1 enzymes and results in formation
of HCHO (7). It is likely that the genotoxicity of CH2Cl2
results from DNA binding by the electrophilic intermedi-
ate GSCH2Cl (7) rather than DNA binding by HCHO.
While this is an efficient detoxication method, DNA
adducts are probably formed as a side reaction of this
The detoxication of CH2Cl2 occurs through two compet-
ing pathways, both of which convert this hydrophobic
molecule to water-soluble products that can be easily
excreted (Scheme 1). The first of these pathways is an
oxidative P450 pathway that yields carbon monoxide (7).
* To whom correspondence should be addressed. Phone:
(615) 322-2261. Fax: (615) 322-3141. E-mail: guengerich@
toxicology.mc.vanderbilt.edu.
1
Abbreviations: CID, collision-induced dissociation; ESI, electro-
spray ionization; DNase, deoxyribonuclease; DTNB, 5, 5′-dithio-bis(2-
nitrobenzoic acid); ESI, electrospray ionization; GSCH2OAc, S-(1-
acetoxymethyl)GSH; GST, GSH S-transferase; PDE, phosphodiesterase;
SIM, selected ion monitoring. See Instructions to Authors for the
standard abbreviations of nucleosides and bases [Chem. Res. Toxicol.
(1998) 11, 5A-10A].
‡ Vanderbilt University.
§ Union University.
| Current address: Abteilung Klinische Pharmakologie, Institut fu¨r
Experimentelle und Klinische Pharmakologie und Toxikologie, Schill-
ingalee 70, Universita¨t Rostock, D-18055 Rostock, Germany.
10.1021/tx010006h CCC: $20.00 © 2001 American Chemical Society
Published on Web 04/26/2001

Glutathione-CH2-DNA Adducts
Sch em e 2. Ad d u ct Str u ctu r es of
Chem. Res. Toxicol., Vol. 14, No. 5, 2001 601
developed for an enzymatic digestion of duplex DNA to
nucleosides under mild acidic conditions (Scheme 3)
because of the lability of the dGuo and other DNA
adducts (9).
GSCH2OAc-Nu cleosid e P r od u cts
Exp er im en ta l P r oced u r es
Wa r n in g! GSCH2OAc is a potential mutagen and carcinogen
(9) and should be handled carefully according to appropriate
environmental safety and health protocols.
Ch em ica ls. GSH, nucleosides, and DNA base analogues were
purchased from Sigma Chemical Co. (St. Louis, MO) and used
without further purification. GSCH2OAc was prepared by
dissolving GSH (9.8 g, 32 mmol) in 40 mL of dry CH3OH and
then adding dropwise a solution of Na° (2.3 g, 0.10 mol, 3 equiv)
dissolved in 80 mL dry CH3OH (9). The resulting solution of
tri-anionic GSH was added dropwise, at room temperature, to
a
solution of bromomethyl acetate (Aldrich Chemical Co.,
Milwaukee, WI, 3.3 mL, 34 mmol) dissolved in 40 mL of
CH3OH, and the product was allowed to precipitate at 0 °C for
20 min (9). The product was collected by centrifugation
(3 × 103g, 10 min), the supernatant was decanted, and the
product was dried in vacuo. The recovered GSCH2OAc (15.4 g)
was stored desiccated at -80 °C and used without further
purification.
The dAdo adduct was synthesized by reacting dAdo (250 mg,
0.93 mmol) with GSCH2OAc (1.0 g, 2.6 mmol) in 50 mL of H2O.
The reaction was incubated at room temperature for 2.5 h, and
pathway (Scheme 1). The tumors that have been observed
in mice are proposed to occur by way of the GST pathway
and not the P450 pathway (8).
S-(1-Acetoxymethyl)GSH (GSCH2OAc) was developed
as a model compound for the potentially toxic S-(1-
halomethyl)GSH intermediates in the GST detoxication
pathway (9). The rationale for using GSCH2OAc is that
the acetate models the halogen leaving group of the
putative conjugate generated enzymatically. In contrast
to the halides, GSCH2OAc can be prepared, stored as a
dry powder, and used to generate adducts in aqueous
solutions. Use of this compound in the formation of DNA
adducts eliminates the requirement for GST enzymes.
Previous research identified a single adduct of the
GSCH2OAc to dGuo at the exocyclic N2 position, S-[1-
(N2-deoxyguanosinyl)methyl]GSH (Scheme 2) (9). Evi-
dence for adduct formation with dCyd had been observed
(9), but the product could not be characterized.
the reaction was stopped by adding 12.5 mL of 3.0
NH4CH3CO2.
M
The dThd adduct was synthesized by reacting dThd (300 mg,
1.2 mmol) with GSCH2OAc (850 mg, 2.2 mmol) in 50 mL of H2O.
The reaction was incubated at room temperature for 10 min and
was stopped by the addition of 12.5 mL of 3.0 M NH4CH3CO2.
The dCyd adduct was synthesized by reacting dCyd (64 mg,
0.28 mmol) with GSCH2OAc (100 mg, 0.26 mmol) in 12.5 mL of
H2O. The reaction was incubated at room temperature for 30
min, and another 100 mg of GSCH2OAc was added to the
solution, followed by a 30 min incubation. The reaction was
stopped by the addition of 2.5 mL of 3.0 M NH4CH3CO2.
Kin etic Stu d ies of GSCH2OAc a n d d Gu oCH2SG Sta bili-
ties. UV spectra were acquired at room temperature with a
OLIS/Cary 14 spectrophotometer (On-Line Instrument Systems,
Bogart, GA).
The goals of the present study were to synthesize,
purify, and characterize the structures of any other
nucleoside adducts using UV, MS, and NMR methodolo-
gies. Adduction of GSCH2OAc to double-stranded DNA
in vitro was also investigated. A rapid protocol was
Sch em e 3. Ra p id Digestion of DNA to Nu cleosid es

602 Chem. Res. Toxicol., Vol. 14, No. 5, 2001
Marsch et al.
For the estimation of the t1/2 of GSCH2OAc, a cuvette (1.0
mL, 1 cm path length) was filled with 0.98 mL of a solution of
0.10 M Tris-HCl (pH 8.0) and 20 mM 5,5′-dithio-bis(2-nitro-
benzoic acid) (DTNB) (at 23 °C), in a Cary 14/OLIS spectro-
photometer. A 10 µL aliquot of 10 mM GSH was added, and
the final A412 value (1.15) was reached within 0.1 min (allowing
for 5 s deadtime after manual mixing). The experiment was done
again but by quickly dissolving 14 mg of GSCH2OAc in 5.0 mL
of H2O and adding 30 µL of this solution to the cuvette to start
the assay). An exponential increase in A412 was recorded and
fit to a pseudo-first-order plot, assuming that the fit is for decay
of GSCH2OAc to GSH and without correcting for the reaction
of GSH with DTNB.
The stability of S-[1-(N2-deoxyguanosinyl)methyl]GSH (9) as
a function of pH was studied by MS methods. Standard adduct
was dissolved in 50 mM NH4CH3CO2 and adjusted with either
NH4OH or CH3CO2H to pH 4.0, 5.0, 6.0, 7.0, 8.0, or 9.0. For
adduct incubated at a given pH value, aliquots were extracted
at specified times in a time course that extended to 4.2 h.
Samples were immediately lyophilized and resuspended in 1.0
mL of HPLC buffer A [2% CH3OH (v/v) in 10 mM NH4CH3CO2,
pH 4.5]. Samples were immediately placed in an Alliance 2690
Separations Module (Waters, Milford, MA), and the autosampler
injected 5 µL aliquots onto a C18 guard cartridge at 5 min
intervals. Adduct was eluted from the guard column with 95%
CH3OH (v/v) and was introduced on-line into a TSQ-7000 triple
stage quadrupole mass spectrometer (Finnigan MAT, San J ose,
CA) outfitted with an API-1 ESI source (vide infra for MS
parameters). Absorption at 260 nm was used to monitor total
Gua chromophore levels of eluted analyte from each injection.
Any sample-to-sample injection differences could be corrected
by normalizing the 260 nm UV peaks to unity. Intact adduct
levels in each injected sample were measured by selected ion
monitoring (SIM) at m/z 587 and 294 (doubly charged adduct).
Correction factors resulting from the normalized UV traces were
then applied to the ion level detected by the mass spectrometer,
with the zero time sample given an adduct level of unity (or
100%). While the absolute concentration of intact adduct cannot
be quantified by this procedure, the relative amount of adduct
remaining during the time course can.
and depended on the type of experiment conducted and the flow
rate of introduced sample. Typical parameters for a flow rate
of 20 µL min-1 from a syringe pump were N2 sheath gas, 54
psi; N2 auxiliary gas, 0 psi; spray voltage, 3.6 kV; capillary
temperature, 200 °C.
Data acquisition and spectral analysis were conducted using
Finnigan ICIS software, version 8.3.2, on a Digital Equipment
Co. Alpha workstation. When samples were introduced by direct
infusion, a Harvard Apparatus (Holliston, MA) model 22 syringe
pump was used at a flow of 10-20 µL min-1
.
For qualitative analysis, different scan modes were applied.
Simple full scans covered the m/z range from 100 to 2000 with
scan times up to 2 ms/amu (the range was extended to check
condensation products). Tandem MS experiments were con-
ducted with Ar at 2.5 mTorr as the collision gas and a collision
voltage of -22 V. Daughter scans were performed to screen the
fragmentation pattern of the parent ions of interest, and parent
scans were performed to identify the source molecules of
characteristic daughter ions. A Waters on-line UV monitor was
used (in parallel) to record analytes absorbing at 260 nm in order
to correlate ions with molecules bearing DNA base chromo-
phores.
HP LC-MS of Ad d u cts. Prior to purification, crude GSCH2-
OAc-nucleoside adduct samples were analyzed using HPLC-MS.
The autosampler and HPLC system consisted of an Alliance
2690 Separations Module from Waters. The buffers used for this
separation were identical to those used in preparative HPLC
purifications of nucleoside adducts. The flow rate was 200 µL
min-1. Analytical 250 mm C18 columns (5 µm particle diameter)
from several manufacturers were employed in the separation
of adducts in the overall course of this work. The gradient was
as follows: 0 to 6 min, B ) 0%; 6 to 8 min, B ) 0 to 5%; 8 to 36
min, B ) 5 to 25%; 36 to 54 min, B ) 25 to 100%; 46 to 54 min,
B ) 100%. From 0 to 6 min, the valve was switched to waste in
order to desalt samples on-line; at 6 min the valve switched back
to the mass spectrometer ESI source.
1
NMR of d Ad oCH2SG. H NMR spectra were acquired with
a Bruker DRX400 Avance instrument operating at 400.13 MHz
(Bruker, Billerica, MA). Nucleoside adducts, nucleosides, and
GSH were all dissolved in 2H2O. Proton scans were then
performed on dAdo, GSH, and the dAdoCH2SG adduct, also in
2H2O. The data were processed using a Silicon Graphics O2
computer with XWINN NMR software.
Sep a r a tion a n d P u r ifica tion of Nu cleosid e Ad d u cts.
Nucleoside adducts were separated using HPLC. Buffer A
contained CH3OH and 10 mM NH4CH3CO2 (pH 4.5) (2-98, v/v).
Buffer B contained CH3OH and 10 mM NH4CH3CO2 (pH 4.5)
(95-5, v/v). The flow rate was 3.0 mL min-1, and the octadecyl-
silane (C18) column used was a Beckman ODS Ultrasphere (5
µm, 10 × 250 mm) (Beckman, San Ramon, CA). Initial small-
scale separations were performed to determine the tRs of the
unreacted deoxynucleosides, possible N-acetylated nucleosides,
and the nucleoside adducts. Subsequent large-scale HPLC
separations were performed with 2.0 mL injection volumes of
nucleoside/GSCH2OAc reaction mixtures.
F or m a tion of GSCH2OAc-Mod ified Du p lex DNA. Double-
stranded calf thymus DNA (Sigma) was dissolved in H2O and
sonicated for 1 h in a Fisher FS9H sonicator bath (Fisher
Scientific, Pittsburgh, PA). The DNA was repurified by a phenol:
CHCl3:isoamyl alcohol (25:24:1, v/v/v) extraction followed by
three C2H5OH precipitations. Ethanolic DNA pellets were
resuspended in H2O and the purity and double-stranded char-
acter of the DNA were monitored by UV spectroscopy.
Following removal of CH3OH in vacuo, pooled adduct frac-
tions were concentrated using Bakerbond C18 cartridges (J . T.
Baker, Milford, MA); nucleoside adducts were eluted with 2.0
mL of a CH3OH:H2O mixture (60/40, v/v). CH3OH was quickly
removed using a Labconco vacuum centrifuge (Labconco, Kansas
City, MO). UV spectra of purified adducts were acquired to
verify identities of eluted materials.
GSCH2OAc (67 mg, 180 µmol) was added to DNA (2.2 mg,
7.0 µmol DNA bases) in 2.0 mL of 83 mM sodium acetate buffer
(pH 5.0). The reaction progressed for 1 h at 37 °C. The DNA
was precipitated by C2H5OH precipitation, washed twice with
70% C2H5OH (v/v), and dried in vacuo. A mixture of porcine
spleen deoxyribonuclease (DNase) II (EC 3.1.22.1, Sigma) and
bovine spleen phosphodiesterase (PDE) II (EC 3.1.16.1, Sigma)
was used to degrade duplex DNA to 3′-phosphate nucleotides
in 15 min at 37 °C. The concentrations used were the follow-
ing: DNA, 1 mM (bases); sodium acetate buffer (pH 5.0), 83
mM; MgSO4, 8.3 mM; NaCl, 20 mM; DNase II, 1 unit (µg
Concentrated nucleoside adduct samples were diluted to 4.0
mL with H2O and injected into the HPLC in two separate 2.0
mL repurification runs. The nucleoside fraction was collected
from each of the runs, and CH3OH was removed in vacuo.
Adduct fractions were concentrated using Bakerbond cartridges
(vide supra) and the eluted material was concentrated to dryness
in vacuo.
DNA)-1; PDE II, 0.74 units (mg DNA)-1
.
Three nonspecific acid phosphatases proved useful in hydro-
lyzing GSCH2OAc-modified DNA to nucleosides (all from
Sigma): Type I wheat germ, 15-20 units (mg DNA)-1; Type II
white potato, 3.00 units (mg DNA)-1; and Type IV white potato,
11.5 units (mg DNA)-1. Enzymes and any nondigested DNA
were removed with Amicon Centricon-30 membranes (Amicon,
Lexington, MA, Mr cutoff 30 kDa) and the filtrate was stored
at -80 °C until further analysis. The lability of the GSCH2OAc-
MS of Ad d u cts. MS was performed using a Finnigan TSQ-
7000 triple stage quadrupole mass spectrometer with a standard
API-1 electrospray ionization source outfitted with a 150 µm
inner diameter deactivated fused silica capillary. The mass
spectrometer was operated in the positive ion mode. MS
detection parameters were optimized using dGuo for a standard

Glutathione-CH2-DNA Adducts
Chem. Res. Toxicol., Vol. 14, No. 5, 2001 603
F igu r e 1. Stability of reagent and a nucleoside adduct. (A) Decomposition of GSCH2OAc. (B) Stability of S-[1-(N2-deoxyguanosinyl)-
methyl]GSH at varying pH values. See Experimental Procedures for details.
derived adducts necessitated a rapid, efficient digest with all
aspects of the experiment performed in 1 day.
Resu lts
Sta bilities of GSCH2OAc a n d d Gu oCH2SG Ad -
d u ct. In pH 8.0 Tris-HCl buffer, the half-life of GSCH2-
OAc was 11 s (Figure 1A). The reaction kinetics were best
fit to an exponential decay curve, C(t) ) Coe-kt, suggesting
a pseudo-first-order reaction in which GSCH2OAc is
hydrolyzed to GSH (partially GS- at pH 8.0), which in
turn reacts with DTNB.
HPLC-purified dGuoCH2SG was observed to be alka-
line labile (Figure 1B). Using an exponential decay as a
best fit, the half-lives of dGuo adduct at pH 5.0, 6.0, 8.0,
and 9.0 were estimated to be 38, 14, and 1.8 h and 19
min, respectively, in qualitative agreement with previous
work in this laboratory.2,3 The instability of dGuo adduct
under alkaline conditions led to the development of a
rapid DNA digest protocol at low pH, affording the
opportunity to study adducts in double-stranded DNA
(vide infra).
HP LC P u r ifica tion of Ad d u cts. Initial reaction
conditions were designed using studies with the dGuo
adduct, because this had been most extensively charac-
terized (9). The adduct S-[1-(N2-deoxyguanosinyl)methyl]-
GSH was the only one detected in the reaction of dGuo
with GSCH2OAc, using HPLC monitored by UV or MS.
Following the reaction of each deoxyribonucleoside
with GSCH2OAc, components of the reaction mixtures
were analyzed and subsequently purified by HPLC.4 The
initial analytical HPLC was followed by a series of
preparative injections (Figure 2). The peak eluting at 11.4
min in part A was shown by UV and MS spectra to be a
F igu r e 2. HPLC chromatograms of reactions of GSCH2OAc
with deoxynucleosides, followed by separation with a semi-
preparative C18 (octadecylsilane) column. Absorbance was
monitored at 260 nm (HPLC buffers at pH 4.5). (A) Injection of
50 µL of GSCH2OAc plus dCyd reaction mixture. The inset
shows injection of 1.0 mL of the same reaction mixture. (B)
Injection of 50 µL of GSCH2OAc plus dThd reaction mixture.
The inset shows an injection of 2.0 mL of the reaction mixture.
(C) Injection of 50 µL of GSCH2OAc plus dAdo reaction mixture.
The inset shows an injection of 2.0 mL of the reaction mixture.
2
Mu¨ller, M., Doerschuk, B., and Guengerich, F. P., unpublished
results. The results were similar to those described here but were
compromised by the apparent presence of an acetyl group in the GSH
moiety that was inadvertently formed due to too extensive treatment
with GSCH2OAc in the synthetic reaction.3
dCyd adduct (vide infra). In the case of the dAdo, the
peaks at tR 25 and 26 min were both collected; prelimi-
nary MS identified the tR 25 min peak as an acetylated
dAdo derivative3 and the tR 26 min peak as the putative
adduct of interest. The peak at 22 min in the dThd
chromatogram was tentatively identified as the dThd
adduct (vide infra). For the larger-scale reactions, the
efficiencies of dCyd, dThd, and dAdo adduct formation
3
With concentrations of GSCH2OAc higher than used here, the
products were sometimes complicated by the presence of acetylated
products, presumably resulting from attack of a deoxyribose alcohol
or the GSH amine by the acetyl group of GSCH2OAc.
4
With each of the nucleosides, efforts were made to improve the
reaction yields with the addition of N,N-dimethylformamide to improve
the stability of the reagent GSCH2OAc. However, yields of each
nucleoside adduct were lowered when any of this cosolvent (or
Me2SO) was used, possibly reflecting the need for SN1 reactions.

604 Chem. Res. Toxicol., Vol. 14, No. 5, 2001
Marsch et al.
F igu r e 4. ESI-MS of HPLC-purified dAdoCH2SG Adduct. (A)
Direct infusion at 10 µL min-1 with a full ion scan in positive
chemical ionization mode. The trace of the adduct ion is shown
at m/z 571. Ion fragmentation of dAdoCH2SG, after loss of
deoxyribose, is shown at m/z 455. (B) ESI-MS/MS of purified
dAdoCH2SG, showing fragmentation of the m/z 571 parent ion.
Introduction of the sample was also by direct infusion at 10 µL
min-1
.
F igu r e 3. UV-vis spectra of dAdoCH2SG and methylated dAdo
analogues in H2O at pH 4.5. λmax values are in parentheses.
were approximately 2, 2, and 0.05% of the starting
nucleoside, respectively, but some smaller-scale prepara-
tions showed pyrimidine adduct yields of 11%.
Ch a r a cter iza tion of d Ad o Ad d u ct. The on-line UV
spectrum of the dAdo adduct indicated a λmax at 271 nm
(Figure 3). This is significantly red-shifted from the dAdo
standard, λmax 256 nm. Spectra of methylated base
analogues were obtained for comparison. The UV spectra
of N1-CH3 Ade, N3-CH3 Ade, N6-CH3 dAdo, and N7-CH3
Ade had λmax values at 253, 269, 258, and 267 nm,
respectively (Figure 3). The literature indicates λmax
values (at neutral to acidic pH) of 259 (10, 11), 270 (12),
and 271 (11, 13) for N1-, N3-, and N7-alkyl (d)Ado,
respectively, indicating a lack of effect of the ribosyl
moiety on the Ade spectra. The dAdoCH2SG adduct was
characterized by a UV spectrum similar to those of N3-
CH3 dAdo and N7-CH3 dAdo.
F igu r e 5. 1H NMR downfield aromatic regions of dAdo and
dAdoCH2SG (2H2O). (A) dAdo. (B) dAdoCH2SG.
1
MS of the purified dAdo adduct (Figure 4, panel A)
showed a parent ion (MH+) at m/z 571. The fragment at
m/z 455 represents a loss of deoxyribose, and the frag-
ment at m/z 308 indicates the presence of GSH in the
sample. CID of the m/z 571 peak (Figure 4, panel B) and
analysis by MS/MS showed the loss of deoxyribose (-116
Da) and the presence of GSH. In addition, the fragment
proton disappears from a H NMR spectrum (14). Such
facile exchange has also been documented for N7-alkyl
dAdo adducts (15).
On the basis of all of this information, the dAdo adduct
was determined to form through the N7 position and
assigned the structure S-[1-(N7-deoxyadenosinyl)methyl]-
GSH (Scheme 2).
+
at m/z 320 represents a GSCH2 carbocation fragment
Ch a r a cter iza tion of d Th d Ad d u ct. The online UV
spectrum of the dThd adduct indicated a λmax at 266 nm
(Figure 6), corresponding to a red shift of 2 nm with
respect to standard dThd (λmax 264 nm). Spectra of
methylated base analogues were obtained for comparison
(Figure 6). The UV spectra for O2-CH3 dThd, N3-CH3
dThd, and O4-CH3 dThd exhibited lowest-energy λmax
values of 250, 260, and 274 nm, respectively. The adduct
spectrum is most similar to the N3-CH3 dThd and O4-
CH3 dThd spectra.
and the m/z 252 fragment confirmed the presence of dAdo
in the parent compound.
The 1H NMR spectrum of the dAdo adduct (2H2O)
appeared to indicate a mixture of dAdo and GSH spectra,
except for the downfield, aromatic region (Figure 5). The
only significant difference between the two spectra was
the absence of the C8 proton resonance in the adduct
spectrum, indicating that the adduct was formed at either
the N7 or C8 position on the purine ring. Tomasz first
showed that alkylation at the N7 position of Guo induces
C8 proton exchange with the solvent, and thus the H8
The crude dThd plus GSCH2OAc reaction mixture was
analyzed by HPLC-ESI-MS, and an eluate at 36 min was

Glutathione-CH2-DNA Adducts
Chem. Res. Toxicol., Vol. 14, No. 5, 2001 605
F igu r e 8. HPLC-ESI-MS of the dCydCH2SG adduct. The pH
of the HPLC buffers was maintained at 4.5. Left: Mass
chromatograms of crude dCyd plus GSCH2OAc reaction mixture,
monitoring elution of dCyd adduct (A), dCyd (B), and the total
ion current (C). (D) HPLC/MS of 26 min eluate. (E) HPLC-ESI-
MS/MS of the dCyd ion.
F igu r e 6. UV spectra of methylated dThd analogues. λmax
values are in parentheses. All spectra were recorded at pH 4.5
in 18% aqueous CH3OH (v/v).
NMR spectra of the labile dThd adduct were unsuccess-
ful.
Ch a r a cter iza tion of d Cyd Ad d u ct. The λmax of the
dCyd adduct was 271 nm, in agreement with previous
work in this laboratory (9). This λmax is similar to that
reported for N4-ethyl Cyd (272 nm) at neutral or alkaline
pH but not N3-ethyl Cyd (279 nm) (16). HPLC-ESI-MS
of crude nucleoside adduct mixtures showed the dCyd
adduct eluting 2 min before dCyd (Figure 8, panels A-C).
The full-scan mass spectrum showed several prominent
ion peaks (Figure 8D). The mass chromatogram of the
HPLC run suggested that the analyte corresponding to
m/z 439 did not precisely coelute with the adduct. Upon
collision with Ar gas (Figure 8E), the adduct parent ion
yielded fragments with m/z 431, 308 (weak), and 285,
corresponding to adduct minus deoxyribose, GSH, and
adduct minus deoxyribose and the glutamyl residue of
GSH, respectively. The dCyd adduct decomposed quickly
at room temperature and proved refractory to NMR
structure determination.
DNA Digests. The lability of the dGuoCH2SG adduct
at alkaline and neutral pH necessitated the development
of a relatively rapid DNA digest to nucleosides under acid
conditions (Figure 1B). Although we did not analyze the
kinetics of hydrolysis of the other adducts, the similarity
of the assigned structures and initial work on separation
of products suggested similar behavior. Porcine spleen
DNase II and bovine spleen PDE II efficiently digested
DNA to nucleotides in 15 min. Four different nonspecific
acid phosphatases, Type I from wheat germ and Types
II, IV, and VII from white potato, were then used to
digest the nucleotides to nucleosides in as little as 1.7 h.
In this examination, Types I, II, and IV acid phos-
phatases worked best and yielded all four nucleosides in
approximately equimolar concentrations (data not shown).
F igu r e 7. HPLC-ESI-MS of dThdCH2SG adduct. The pH of
HPLC buffers was maintained at pH 4.5. (A) Full scan of dThd
adduct (m/z 562) acquired on-line after elution from analytical
reversed-phase C18 column. (B) Fragmentation of the m/z 562
dThd adduct ion by HPLC-ESI-MS/MS.
identified as the dThdCH2SG adduct (m/z 562). Full-scan
ESI-MS of the analyte showed strong ion peaks at m/z
273 and m/z 290 (Figure 7). These fragments were
verified to be adduct-related by CID and analysis by MS/
MS (Figure 7). The m/z 273 and 290 fragments corre-
sponded to GSH minus the sulfhydryl and dThdCH2SH,
respectively. In addition, fragmentation of adduct re-
leased dThd (m/z 243). Several attempts to obtain 1H

606 Chem. Res. Toxicol., Vol. 14, No. 5, 2001
Marsch et al.
F igu r e 10. HPLC-ESI-MS and HPLC-ESI-MS/MS of GSCH2-
OAc-modified calf thymus DNA digested to nucleosides with
DNase II, PDE II, and acid phosphatase Type IV. (A-C) Mass
chromatograms monitored at m/z 243 and 562, corresponding
to dThd and dThdCH2SG. (D, E) Full scans of the eluted adduct
(tR 33.5 min). The adduct ion m/z 562 was selected for the CID
tandem MS experiment (E).
F igu r e 9. Acid DNA digests of calf thymus DNA. The upper
two panels (A, B) show the HPLC chromatograms of nucleotide
and nucleoside standards. The lower panel (C) shows the
chromatogram of DNA digested to nucleosides by DNase II, PDE
II, and wheat germ acid phosphatase Type I (Beckman Ultra-
sphere ODS analytical column).
DNA digests in citrate, acetate, and succinate buffers
were performed at pH values of 5.0, 5.5, or 6.0 to optimize
reaction conditions. Acetate buffer at pH 5.0 was found
to work well for rapid reaction with few reaction byprod-
ucts and a high ratio of nucleosides to nucleotides. A 2-h
digest in sodium acetate buffer (pH 5.0) is shown in
Figure 9, with an 8:1 ratio of nucleosides to nucleotides
(right-hand portion of the chromatogram shows a level
baseline where GSCH2OAc adducts should elute).
product, presumably GSCH2X (X ) halide). Bonse et al.
(17) prepared model compunds in the series RSCH2X
(where R is a simple alkyl or aryl group) and measured
the stabilities in 1,4-dioxane solutions containing low
concentrations of H2O, where the half-lives were on the
order of hours. Dekant (18) prepared S-(1-chloromethyl)-
Cys and estimated that the t1/2 in aqueous buffer was <4
s at 0 °C. Stourman and Armstrong have recently
estimated the t1/2 of CH3SCH2Cl in 90% H2O to be ∼11
ms.5 We continued to utilize a model alkylating agent
we developed, GSCH2OAc (9). This compound can be
readily prepared and stored. Its t1/2 was estimated to be
11 s at pH 8.0 (Figure 1), which allows this reagent to
be practical in model reactions with nucleosides because
of the apparent stability of the acetate relative to the
chloride.
HPLC-ESI-MS of DNA modified with GSCH2OAc and
digested to nucleosides indicated the presence of dThd
lesions. The mass chromatogram (Figure 10) showed a
m/z 562 peak, corresponding to a dThd adduct. The dThd
adduct coeluted with the unreacted dThd at 33.5 min
with a 0.25 mL min-1 flow rate. CID of the m/z 562 parent
ion efficiently fragmented the dThd adduct into dThd
+
nucleoside (m/z 243) and GSCH2 (m/z 320). Thus, the
We consider GSCH2OAc to be a reasonable model and
employed this reagent in our studies. A liablity is the
deomonstrated tendency to acetylate when very high
concentrations are used.3
compound in the peak eluted at 33.5 min contained both
GSH and nucleoside, verifying its identity as a dThd DNA
lesion.
d Ad o Ad d u ct. Using the extinction coefficient for Ado
(ꢀ260 ) 13 400 M-1 cm-1) as a basis, we estimate that the
yield for dAdoCH2SG formation was only 0.05%, yet the
100 µg obtained were sufficient for structure determina-
tion. By comparing the UV spectrum of the dAdo adduct
Discu ssion
One approach to better understanding the mutagenic-
ity of GSH conjugates of CH2Cl2 and other dihalo-
methanes is the characterization of deoxyribonucleoside
adducts and ultimately DNA adducts. A major technical
obstacle is the preparation of the relevant electrophilic
5
Stourman, N. V., and Armstrong, R. N., unpublished results
(estimated by appearance of thiolate absorbance at 240 nm).

Glutathione-CH2-DNA Adducts
Chem. Res. Toxicol., Vol. 14, No. 5, 2001 607
to that of the methylated base analogues, we were able
to determine that the N1 and N6 positions were unlikely
sites of adduction, whereas the absorbance spectra of the
N3- and N7-alkyl analogues were similar to the spectrum
of the dAdo adduct derived from GSCH2OAc.
a rough comparison of ion currents suggests that ∼1
dThd in 103 was modified.6 A particularly pertinent
question is the in vivo applicability of the GSCH2OAc-
DNA adducts, which were formed at pH 5.0 and have
short lifetimes at neutral to alkaline pH. However, it is
important to note that the stability in the nucleosides
may be much less than in double-stranded DNA, as is
the case with some other DNA adducts (20). We have not
yet attempted to estimate stabilities within DNA.
Con clu sion s. In conclusion, the adducts generated
with GSCH2OAc are considerably different from those we
have observed with half-mustards of the form GSCH2-
CH2X (21-25). With the ethylene series, the predominant
reactions are SN2-like, with an episulfonium ion inter-
mediate (26) and the N7 atom of dGuo the most reactive
site in DNA (21-25). We hypothesize that the reactions
with GSCH2OAc may have more SN1 character. Future
efforts will involve the extension of strategies for the
rapid hydrolysis of DNA, the characterization of GSH-
containing DNA adducts, and more sensitive analysis by
HPLC-MS or other methods.
Definitive evidence for the structure of the dAdo adduct
was provided by the absence of the C8 proton resonance
2
in the NMR spectrum recorded in H2O, consistent with
the known exchange of an N7alkyl dAdo adduct (15). The
dAdo adduct, determined to form at the N7 position, was
therefore characterized as S-[1-N7-deoxyadenosinyl)-
methyl]GSH (Scheme 2). Although the N7 atom of dAdo
is not as nucleophilic as the N7 atom of dGuo, it is a
major site of alkylation on Ado by several electrophiles
(15, 19).
d Th d Ad d u ct. The dThd adduct was produced in a
yield of 2% (based on the absorbance and ꢀ260 ) 9500 M-1
cm-1) but the nucleoside adduct proved to be rather
labile. MS showed that the adduct was formed, and a
characteristic pattern of fragmentation occurred between
the GSH and the sulfhydryl group of its Cys residue. This
fragmentation produced the ion at m/z 290, representing
dThd with the addition of the methylene bridge and the
sulfhydryl. The other resulting ion at m/z 273 represents
loss of the sulfhydryl group (-32) from GSH.
Due to the instability of the dThdCH2SG adduct,
attempts at 1H NMR were unsuccessful. When the UV
spectrum of the adduct was compared to that of standard
dThd, the adduct showed a 2 nm red shift, indicating a
small change in the aromatic character of the chromo-
phore. Among the methylated base-adduct analogues, the
most likely site of adduction is at the N3 position of the
pyrimidine ring, which would result in an unchanged
resonance structure. The UV spectrum of O2-CH3 Thd is
very dissimilar to the dThd adduct, and the lowest energy
Ack n ow led gm en t. This work was supported in part
by U.S. Public Health Service Grants (USPHS) R35
CA44353 and P30 ES00267. M.K.H. was supported in
part by USPHS T32 ES07028. We thank M. Mu¨ller, R.
Their, and B. Doerschuk for their earlier contributions
to this effort.
Refer en ces
(1) Kari, F. W., Foley, J . F., Seilkop, S. K., Maronpot, R. R., and
Anderson, M. W. (1993) Effect of varying exposure regimens on
methylene choride-induced lung and liver tumors in female
B6C3F1 mice. Carcinogenesis 14, 819-826.
(2) Hanson, D. (1991) OSHA to cut methylene chloride exposure
limits. Chem. Eng. News 69, 8.
(3) Agency for Toxic Substances and Disease Registry (1993) Toxi-
cological Profile for Methylene Chloride, U.S. Department Health
& Human Services, Atlanta.
(4) Hanson, D. (1997) Methylene chloride exposure sharply limited.
Chem. Eng. News 75, 7.
(5) Green, T. (1997) Methylene chloride induced mouse liver and lung
tumours: an overview of the role of mechanistic studies in human
safety assessment. Hum. Exp. Toxicol. 16, 3-13.
(6) Huff, J ., Bucher, J ., and Barrett, J . C. (1996) Methylene chloride.
Science 272, 1083-1084.
(7) Kubic, V. L., and Anders, M. W. (1978) Metabolism of dihalo-
methanes to carbon monoxidesIII. Studies on the mechanism of
the reaction. Biochem. Pharmacol. 27, 2349-2355.
(8) Andersen, M. E., Clewell, H. J ., III, Gargas, M. L., Smith, F. A.,
and Reitz, R. H. (1987) Physiologically based pharmacokinetics
and the risk assessment process for methylene chloride. Toxicol.
Appl. Pharmacol. 87, 185-205.
(9) Thier, R., Pemble, S. E., Taylor, J . B., Humphreys, W. G.,
Persmark, M., Ketterer, B., and Guengerich, F. P. (1993) Expres-
sion of mammalian glutathione S-transferase 5-5 in Salmonella
typhimurium TA1535 leads to base-pair mutations upon exposure
to dihalomethanes. Proc. Natl. Acad. Sci. U.S.A. 90, 8576-8580.
(10) Fujii, T., Itaya, T., Tanaka, F., Saito, T., and Hisata, H. (1990)
Purines. XLVI. Preparation of 1-ethyladenine from adenosine.
Chem. Pharm. Bull. 38, 3463-3465.
(11) Singer, B., Sun, L., and Fraenkel-Conrat, H. (1974) Reaction of
adenosine with ethylating agents. Biochemistry 13, 1913-1920.
(12) Fujii, T., Saito, T., and Nakasaka, T. (1989) Purines. XXXIV.
3-Methyladenosine and 3-methyl-2′-deoxyadenosine: their syn-
thesis, glycosidic hydrolysis, and ring fission. Chem. Pharm. Bull.
37, 2601-2609.
(13) Fujii, T., and Saito, T. (1990) Purines. XLII. Synthesis and
glycosidic hydrolysis of 7-alkyladenosines leading to an alternative
synthesis of 7-alkyladenines. Chem. Pharm. Bull. 38, 1886-1891.
(14) Tomasz, M. (1970) Extreme lability of the C-8 proton: a conse-
quence of 7-methylation of guanine residues in model compounds
and in DNA and its analytical application. Biochim. Biophys. Acta
199, 18-28.
λ
max of the dThd adduct is blue-shifted 8 nm with respect
to O4-CH3 Thd. It is highly unlikely that adduction at
any position of an intact pyrimidine would induce a large
increase in energy of π-π* transitions, and we therefore
propose that the dThd adduct is S-[1-(N3-thymidinyl)-
methyl]GSH (Scheme 2).
d Cyd Ad d u ct. The dCyd adduct was also highly
unstable at room temperature and under the pH condi-
tions tested, from pH 5.0 to 8.0, but its UV spectrum
could be acquired on-line during HPLC purification. The
adduct spectrum was very similar to the UV spectrum
of unmodified dCyd nucleoside, with the adduct λmax blue-
shifted by just 1 nm with respect to the nucleoside λmax
(271 vs 272 nm for dCyd). The similarity of the spectra
suggest little perturbation of the Cyt π electrons, a
condition that would result from alkylation at the exo-
cyclic N4 position of the cytosine moiety (16). Thus, we
propose that the dCyd adduct is S-[1-(N4-deoxycytidinyl)-
methyl]GSH (Scheme 2).
Detection of Ad d u cts in Dou ble-Str a n d ed DNA.
The dThd adduct was detected in calf thymus DNA
treated with GSCH2OAc, as shown by the MS evidence
(Figure 10). The concentration is difficult to estimate, but
6
Surprisingly, dGuoCH2SG adducts in double-stranded DNA were
never detected, although the nucleoside formed adduct complexes in
yields higher than 5% in earlier (9) and subsequent work. Steric
hindrance may have blocked binding by GSCH2OAc, but another
possibility is that the PDE II or acid phosphatase may not have been
able to catalyze the hydrolysis of DNA with bulky dGuoCH2SG lesions.
In our hands, the dAdoCH2SG adduct was moderately stable (Figure
1), although the very low nucleoside yield might have precluded its
detection in a less-sensitive duplex DNA reaction.

608 Chem. Res. Toxicol., Vol. 14, No. 5, 2001
Marsch et al.
(15) Iyer, R. S., Voehler, M. W., and Harris, T. M. (1994) Adenine
adduct of aflatoxin B1 epoxide. J . Am. Chem. Soc. 116, 8863-
8869.
(16) Sun, L., and Singer, B. (1974) Reaction of cytidine with ethylating
agents. Biochemistry 13, 1905-1913.
(17) Bo¨hme, V. H., Fischer, H., and Frank, R. (1949) Darstellung und
Eigenschaften der R-halogenierten Thioa¨ther. Ann. Chem. 563,
54-72.
(18) Hashmi, M., Dechert, S., Dekant, W., and Anders, M. W. (1994)
Bioactivation of [13C]dichloromethane in mouse, rat, and human
liver cytosol: 13C nuclear magnetic resonance spectroscopic
studies. Chem. Res. Toxicol. 7, 291-296.
(19) Mandel, H. G., Straw, J . A., Farmer, P. B., and Martin, J . (1989)
Chromatographic detection of 7-methyladenine in urine of rats
administered N-methylnitrosourea: a potential marker for moni-
toring exposure to methylating carcinogens. Carcinogenesis 10,
757-762.
(20) Kusmierek, J . T., Folkman, W., and Singer, B. (1989) Synthesis
of N2,3-ethenodeoxyguanosine, N2,3-deoxyguanosine 5′-phosphate,
and N2,3-ethenodeoxyguanosine 5′-triphosphate. Stability of the
glycosyl bond in the monomer and in poly(dG,ꢀdG-dC). Chem. Res.
Toxicol. 2, 230-233.
(21) Ozawa, N., and Guengerich, F. P. (1983) Evidence for formation
of an S-[2-(N7-guanyl)ethyl]glutathione adduct in glutathione-
mediated binding of 1,2-dibromoethane to DNA. Proc. Natl. Acad.
Sci. U.S.A. 80, 5266-5270.
(22) Koga, N., Inskeep, P. B., Harris, T. M., and Guengerich, F. P.
(1986) S-[2-(N7-Guanyl)ethyl]glutathione, the major DNA adduct
formed from 1,2-dibromoethane. Biochemistry 25, 2192-2198.
(23) Inskeep, P. B., Koga, N., Cmarik, J . L., and Guengerich, F. P.
(1986) Covalent binding of 1,2-dihaloalkanes to DNA and stability
of the major DNA adduct, S-[2-(N7-guanyl)ethyl]glutathione.
Cancer Res. 46, 2839-2844.
(24) Cmarik, J . L., Humphreys, W. G., Bruner, K. L., Lloyd, R. S.,
Tibbetts, C., and Guengerich, F. P. (1992) Mutation spectrum and
sequence alkylation selectivity resulting from modification of
bacteriophage M13mp18 with S-(2-chloroethyl)glutathione. Evi-
dence for a role of S-[2-(N7-guanyl)ethyl]glutathione as a muta-
genic lesion formed from ethylene dibromide. J . Biol. Chem. 267,
6672-6679.
(25) Kim, M.-S., and Guengerich, F. P. (1997) Synthesis of oligonucle-
otides containing the ethylene dibromide-derived DNA adducts
S-[2-(N7-guanyl)ethyl]glutathione, S-[2-(N2-guanyl)ethyl]gluta-
thione, and S-[2-(O6-guanyl)ethyl]glutathione at a single site.
Chem. Res. Toxicol. 10, 1133-1143.
(26) Peterson, L. A., Harris, T. M., and Guengerich, F. P. (1988)
Evidence for an episulfonium ion intermediate in the formation
of S-[2-(N7-guanyl)ethyl]glutathione in DNA. J . Am. Chem. Soc.
110, 3284-3291.
TX010006H