Modified guanines representing O6-alkylation by the cyclophosphamide metabolites acrolein and chloroacetaldehyde: Synthesis, stability, and ab initio studies

Article abstract of DOI:10.1021/tx0101503

Alkylation of DNA by acrolein and/or chloroacetaldehyde may result in the mutations that lead to the therapy-induced leukemia associated with cyclophosphamide (and ifosfamide) treatment. O⁶-(n-Propanalyl)guanine (O⁶-PAG) and O⁶-(ethanalyl)guanine (O⁶-EAG) were synthesized for use as authentic standards in investigations of DNA alkylation by acrolein and chloroacetaldehyde, respectively. Preparation of the O-methyl oximes of these aldehydes aided in confirming the structural assignments of O⁶-PAG and O⁶-EAG. HPLC was used to study the stability of O⁶-PAG under a variety of conditions. The decomposition of O⁶-PAG was attributed to an α,β-elimination reaction resulting in the formation of guanine and acrolein. In 0.1 M phosphate-DMSO (9:1), O⁶-PAG (1-10 mM) had a half-life of approximately h (pH 7.4, 37 °C). In 0.05 M Tris-DMSO (9:1), the apparent half-life of O⁶-PAG (1-10 mM) was approximately 16 h (pH 7.4, 37 °C). The increased lifetime under the latter conditions was attributed to a reversible reaction between Tris and the aldehydic functionality of O⁶-PAG to give a more stable oxazolidine. Under conditions similar to those that would be used for hydrolysis of DNA [0.1 M HCl-DMSO (98:2), pH 1.3, 70 °C, 30 min], there was an estimated 10-35% loss of O⁶-PAG. Under the same conditions, O⁶-EAG had apparent half-lives of 6.6 h (phosphate-DMSO) and 2.5 days (Tris-DMSO) and the estimated loss at pH 1.3 over 30 min (70 °C) was 15-20%. Ab initio quantum chemical calculations were used to understand the energy factors that underlie the occurrence of O- versus N-alkylations as well as possible, subsequent intramolecular cyclizations. Simulations of the free energies of reactions between acrolein and guanine indicated that N-alkylation was favored over O⁶-alkylation and that cyclizations to tautomers were most favorable if they involved the N-1 or NH₂ positions.

Full text of DOI:10.1021/tx0101503

380
Chem. Res. Toxicol. 2002, 15, 380-387
Mod ified Gu a n in es Rep r esen tin g O6-Alk yla tion by th e
Cyclop h osp h a m id e Meta bolites Acr olein a n d
Ch lor oa ceta ld eh yd e: Syn th esis, Sta bility, a n d a b In itio
Stu d ies
Narayanan Balu,† Michael P. Gamcsik,† Michael E. Colvin,‡ O. Michael Colvin,†
M. Eileen Dolan,§ and Susan M. Ludeman*,†
Duke Comprehensive Cancer Center, Department of Medicine, Duke University Medical Center,
Durham, North Carolina 27710, Biology and Biotechnology Research Program,
Lawrence Livermore National Laboratory, Livermore, California 94550, and
Cancer Research Center and Department of Medicine, University of Chicago, Chicago, Illinois 60637
Received September 11, 2001
Alkylation of DNA by acrolein and/or chloroacetaldehyde may result in the mutations that
lead to the therapy-induced leukemia associated with cyclophosphamide (and ifosfamide)
treatment. O6-(n-Propanalyl)guanine (O6-PAG) and O6-(ethanalyl)guanine (O6-EAG) were
synthesized for use as authentic standards in investigations of DNA alkylation by acrolein
and chloroacetaldehyde, respectively. Preparation of the O-methyl oximes of these aldehydes
aided in confirming the structural assignments of O6-PAG and O6-EAG. HPLC was used to
study the stability of O6-PAG under a variety of conditions. The decomposition of O6-PAG was
attributed to an R,â-elimination reaction resulting in the formation of guanine and acrolein.
In 0.1 M phosphate-DMSO (9:1), O6-PAG (1-10 mM) had a half-life of approximately 1 h (pH
7.4, 37 °C). In 0.05 M Tris-DMSO (9:1), the apparent half-life of O6-PAG (1-10 mM) was
approximately 16 h (pH 7.4, 37 °C). The increased lifetime under the latter conditions was
attributed to a reversible reaction between Tris and the aldehydic functionality of O6-PAG to
give a more stable oxazolidine. Under conditions similar to those that would be used for
hydrolysis of DNA [0.1 M HCl-DMSO (98:2), pH 1.3, 70 °C, 30 min], there was an estimated
10-35% loss of O6-PAG. Under the same conditions, O6-EAG had apparent half-lives of 6.6 h
(phosphate-DMSO) and 2.5 days (Tris-DMSO) and the estimated loss at pH 1.3 over 30 min
(70 °C) was 15-20%. Ab initio quantum chemical calculations were used to understand the
energy factors that underlie the occurrence of O- versus N-alkylations as well as possible,
subsequent intramolecular cyclizations. Simulations of the free energies of reactions between
acrolein and guanine indicated that N-alkylation was favored over O6-alkylation and that
cyclizations to tautomers were most favorable if they involved the N-1 or NH2 positions.
Sch em e 1
In tr od u ction
The anticancer agent cyclophosphamide (1) undergoes
a sequence of metabolic transformations leading to the
formation of 4-hydroxycyclophosphamide (2), aldophos-
phamide (3), and ultimately, phosphoramide mustard (4)
and acrolein (5) (Scheme 1) (1). Phosphoramide mustard
is responsible for cross-linking DNA (2) while acrolein is
most commonly associated with side effects such as
bladder cystitis (3). A competing, albeit minor (∼10%),
metabolic fate of 1 results in dechloroethylation and the
formation of 2-dechloroethylcyclophosphamide (6) and
chloroacetaldehyde (7) (Scheme 1). The occurrence of
neurotoxicity is attributed to this pathway (4, 5).
Recently, we provided the first evidence that O6-
alkylguanine-DNA alkyltransferase (AGT) plays an im-
portant role in protecting against the toxicity and mu-
tagenicity of 1 (6, 7). We also demonstrated that this
relationship between AGT and 1 is specifically correlated
with acrolein and not phosphoramide mustard (7). Con-
sidering that AGT repair of DNA involves removal of an
alkyl group from the O6 position in a guanine residue,
two pathways were presented by which acrolein might
generate such a repairable adduct. Acting as a nucleo-
phile, the guanylic oxygen could initiate a conjugate
* To whom correspondence should be addressed. Phone: (919) 681-
2808. Fax: (919) 684-5653. Email: s.ludeman@duke.edu.
† Duke Comprehensive Cancer Center and Department of Medicine,
Duke University Medical Center.
‡ Lawrence Livermore National Laboratory.
§ University of Chicago.
10.1021/tx0101503 CCC: $22.00 © 2002 American Chemical Society
Published on Web 02/08/2002

Modified Guanines Representing O6-Alkylation
Chem. Res. Toxicol., Vol. 15, No. 3, 2002 381
Sch em e 2a
mide treatment. Recent mutagenicity studies of oligo-
nucleotides containing the major adduct formed from
N-alkylation of guanine by acrolein suggest that this
lesion is not likely to be responsible for the mutagenic
effects associated with acrolein (17). While the formation
of chloroacetaldehyde derives from a minor metabolic
pathway for cyclophosphamide, it accounts for approxi-
mately half of the metabolic fate of the related anticancer
agent ifosfamide (4, 5). Thus, a determination of the total
contributions of these metabolites to the toxicity and
mutagenicity of the oxazaphosphorines remains an area
of active research.
For the purpose of identifying mutagenic lesions in
DNA caused by acrolein and chloroacetaldehyde, authen-
tic adducts are needed as standards. Herein, we report
the synthesis of the thermodynamic product resulting
from O6-alkylation of guanine by acrolein, O6-(n-propana-
lyl)guanine (8; “O6-PAG”, Scheme 2). Also included are
studies of the chemical stability of 8 under conditions
used for DNA incubation and digestion. Quantum chemi-
cal calculations are presented that compare the relative
favorabilities of O6-alkylated adducts such as 8 and 9
with N-alkylated guanines. We also report the synthesis
and study of O6-(ethanalyl)guanine (12; “O6-EAG”, Scheme
3) as representative of the adduct formed from O6-
alkylation of a guanine moiety in DNA by chloroacetal-
dehyde.
a
∆G in kcal/mol, COSMO method.
Sch em e 3a
Exp er im en ta l P r oced u r es
In general, reagents were purchased from Aldrich Chemical
Co., Milwaukee, WI, and were used without further purification.
Reactions that did not include water were carried out under a
slight positive pressure of nitrogen. Melting points were deter-
mined with a Fisher-J ohns melting point apparatus and are
uncorrected. A Labconco Freeze-Dry System/Freezone 4.5 was
a
∆G in kcal/mol, COSMO method.
1
used for lyophilizing samples. H (300 MHz) and 13C NMR (75
addition across the R,â-unsaturated system of acrolein
and/or it could directly add across the aldehydic moiety
of acrolein (6). The products given by these two mecha-
nisms are depicted as O6-alkylated guanines 8 and 9,
respectively, in Scheme 2. Considering the general
reactivity of R,â-unsaturated aldehydes, 8 is expected to
be the thermodynamic product and 9 the kinetic product
(8). Thus, even though an adduct such as 9 might form
faster than 8, it is anticipated that 9 would be unstable
due to the predicted lability of the hemiacetal moiety. In
fact, reactions between acrolein and different nitrogens
in guanine, guanosines, and guanyl residues in DNA
have led almost exclusively to the isolation of the
thermodynamic products derived from conjugate addition
(9-17). On the other hand, a Schiff base adduct between
crotonaldehyde (CH3CHdCHCHO) and the exocyclic NH2
of a guanyl residue in DNA has recently been reported,
and this would be the result of an initial, direct addition
of the amine across the aldehydic carbonyl (18).
The reactivity of chloroacetaldehyde (7) toward sub-
stitution reactions (with loss of chloride) suggests that
this is another cyclophosphamide metabolite capable of
alkylating nucleophilic sites in DNA, including the O6
position of a guanine residue. By way of comparison, O6-
allylguanine has been reported as a product derived from
the alkylation of DNA by allyl chloride (ClCH2CHdCH2),
a compound with chemical properties somewhat similar
to those of 7 (19).
MHz) were recorded on a Varian Mercury-300 spectrometer.
Samples were dissolved in DMSO-d6 containing 0.1% TMS as
an internal standard. Flash chromatography used 230-400
mesh silica gel (Merck). Preparative TLC used silica gel 60 F254
(Merck), 0.25, 0.5, and 1 mm plates (20 × 20 cm). The maximum
amount of crude product applied to each plate was approxi-
mately 25, 50, and 100 mg, respectively. Component visualiza-
tion was made using a UV lamp.
High-resolution fast atom bombardment (FAB) mass spectra
were obtained using a J EOL J MS SX-102 mass spectrometer.
Data were acquired in the scanning mode using a matrix
mixture of m-nitrobenzyl alcohol and PEG-600. Masses were
acquired 200 amu above and below the predicted mass of the
analyte.
HPLC analyses were conducted on a Waters (Milford, MA)
system equipped with a 996-Photodiode Array Detector, 474-
Scanning Fluorescence Detector, 515-HPLC Pumps and 717-
plus Autosampler interfaced with Waters Millennium32 Chro-
matography Manager Software. Some fluorescence spectra were
also recorded on a Shimadzu RF 1501 spectrometer. Some UV
spectra were also recorded on a Beckman DU-600 spectropho-
tometer.
O6-(3-Bu ten yl)gu a n in e (14). With modifications to pub-
lished syntheses of analogous compounds (20-22), freshly cut
metallic sodium (5.9 mmol, 0.14 g) was added slowly to neat
3-buten-1-ol (35 mmol, 3.0 mL) at 50 °C. When the sodium was
no longer visible (∼10 min), 2-amino-6-chloropurine (3.0 mmol,
0.50 g) was added and the mixture was heated at 110 °C for 10
h. The excess 3-buten-1-ol was removed by distillation and the
residue was flash chromatographed on silica gel (3 cm × 47 cm
column) using CHCl3-CH3OH (3:1) to obtain the product as a
white, crystalline solid [2.7 mmol, 0.56 g, 90% yield, mp 195-
Alkylation of DNA by acrolein and/or chloroacetalde-
hyde may result in the mutations that lead to the
therapy-induced leukemia associated with cyclophospha-
1
198 °C, Rf 0.68 in CHCl3-CH3OH (3:1)]. H NMR (DMSO-d6) δ

382 Chem. Res. Toxicol., Vol. 15, No. 3, 2002
Balu et al.
12.42 (br s, 1H, exchangeable with D2O, NH), 7.82 (s, 1H, C-8
proton), 6.23 (s, 2H, exchangeable with D2O, NH2), 5.86 (ddt, J
) 7, 10 and 17 Hz, 1H, CH)CH2), 5.22-5.07 (m, 2H, CHdCH2),
4.43 (t, 2H, J ) 7 Hz, OCH2), and 2.54 [apparent dt, J ) 7 and
7 Hz (plus 1 Hz fine splitting), 2H, OCH2CH2]. 13C NMR (DMSO-
d6) δ 159.5 (C-2 and C-6), 154.8 (C-4), 137.44 (C-8), 134.5 (CHd
CH2), 117.0 (CH)CH2), 113.4 (C-5), 64.50 (OCH2), and 32.87
(OCH2CH2). MS (FAB+) m/z calcd for C5H12N5O [M + H]+
206.1043, found 206.1042.
solid [5.2 mmol, 1.0 g, 88% yield, mp 205-207 °C, Rf 0.63 in
CHCl3-CH3OH (7:1)]. 1H NMR (DMSO-d6) δ 12.44 (br s, 1H,
exchangeable with D2O, NH), 7.84 (s, 1H, C-8 proton), 6.24 (s,
2H, exchangeable with D2O, NH2), 6.23-6.03 (m, 1H, CHdCH2),
5.49-5.23 (m, 2H, CHdCH2), and 5.00 (d, 2H, J ) 1.4 Hz,
OCH2). 13C NMR (DMSO-d6) δ 160.2 (C-2 and C-6), 155.7 (C-4),
138.4 (C-8), 134.1 (CHdCH2), 118.6 (CHdCH2), 114.1 (C-5), and
66.57 (OCH2). MS (FAB+) m/z calcd for C8H10N5O [M + H]+
192.0886, found 192.0886. Diode array UV spectra indicated λmax
) 281 nm with another peak at 240 nm.
O6-(n -P r op a n a lyl)gu a n in e (8; O6-P AG). Solid OsO4 (1.0
mmol, 0.25 g) was added to a solution of butenyl guanine adduct
14 (0.20 g, 1.0 mmol) in CH3OH-H2O (3:1, 5 mL). The mixture
was stirred for 10-15 min and then solid NaIO4 (1.5 mmol, 0.31
g) was added (23). After 1 h, the reaction mixture was vacuum
filtered through a fritted glass funnel, and the solids were
washed (CH3OH). The filtrate was concentrated on a rotary
evaporator, and the residue was placed under high vacuum.
Residual water was removed on a freeze-dryer and the crude
product was obtained as a solid (94 mg, ∼0.45 mmol, yield
∼45%). MS (FAB+) m/z calcd for C8H10N5O2 [M + H]+ 208.0835,
O6-(Eth a n a lyl)gu a n in e (12, O6-EAG). The title compound
was synthesized exactly as described for 8 using solid OsO4 (1.57
mmol, 0.40 g) and propenyl guanine adduct 16 (1.57 mmol, 0.30
g) in CH3OH-H2O (3:1, 10 mL). The mixture was stirred for 15
min, solid NaIO4 (2.36 mmol, 0.50 g), was added and stirring
was continued for 1 h (23). Following a workup as described for
8, the lyophilized material was taken up in CH3OH-CHCl3 (1:
1) and filtered; the clear filtrate was concentrated on a rotary
and then under high vacuum. The crude product was obtained
as an off-white solid (80 mg, ∼0.42 mmol, ∼26% yield, mp >250
°C). MS (FAB+) m/z calcd for C7H8N5O2 [M + H]+ 194.0679,
1
found 208.0835. H NMR (DMSO-d6/D2O, ca. 2:1) δ 9.8 (CHO),
1
8.1-7.8 (C-8 proton), 4.7-4.1 (m, OCH2), and CH2CHO obscured
by DMSO-d5 or HOD. Possible proton signals due to contribu-
tions from an aldehyde hydrate were obscured by the HOD
resonance.
found 194.0674. H NMR (DMSO-d6/D2O, ca. 2:1) δ 9.7 (CHO),
7.8 (C-8 proton), and 4.8 (OCH2). Possible proton signals due to
contributions from an aldehyde hydrate were obscured by the
HOD resonance. Diode array UV spectra indicated λmax ) 280
nm with another peak at 239 nm.
Crude product was dissolved in CH3OH-CHCl3 (1:1) and
filtered. HPLC analysis of this solution showed one major peak
with diode array UV spectra indicating λmax ) 293 nm with a
shoulder at 235 nm. The major peak was the only one that was
visible by fluorescence detection [in CH3CN (ca. 1 mg in 10 mL),
294 (excitation) and 353 (emission) nm].
E/Z-O6-(n -Eth a n a lyl)gu a n in e O-Meth yloxim e (E/Z-17;
E/Z-O6-EAG-oxim e). The title compound was made and puri-
fied as described for E/Z-15 using O-methylhydroxylamine
hydrochloride (0.21 mmol, 17 mg) in 0.2 M NaOH (1.5 mL) and
O6-EAG (12; ∼0.21 mmol, 40 mg) in DMSO (0.5 mL). Product
17 was obtained as a white solid in 19% yield [0.04 mmol, 10
mg, 35% E- and 65% Z-isomers, Rf 0.65 in CHCl3-CH3OH (7:3),
mp 200-206 °C]. Signal assignments and isomer composition
by 1H NMR were based on literature reports of spectral data
E/Z-O6-(n -P r op a n a lyl)gu a n in e O-Meth yloxim e (E/Z-15;
E/Z-O6-P AG-oxim e). O-Methylhydroxylamine hydrochloride
(0.45 mmol, 38 mg) was dissolved in 0.2 M NaOH (1 mL), and
the pH of the solution was then adjusted to 7.4. This was added
to a solution of O6-PAG (8; ∼0.45 mmol, 94 mg) in DMSO (0.5
mL), and the mixture was stirred for 3 h. The reaction mixture
was concentrated on a freeze-dryer, and the residue was purified
by preparative TLC on silica gel using CHCl3-CH3OH (7:3).
While the E- and Z-stereoisomers were observed at different Rf
values (0.7 and 0.6), the two bands on the preparative plates
were collected together. Product was desorbed from silica gel
using CHCl3-CH3OH (1:1), and 15 was obtained as a white solid
in 56% yield (0.25 mmol, 59 mg, 56% E and 44% Z-isomers, mp
range 165-195 °C with decomposition over 200 °C). Signal
assignments and isomer composition by 1H NMR were based
on literature reports of spectral data for similar oximes (24).
1H NMR (DMSO-d6) δ 12.40 (br s, 1H, exchanges with D2O, NH),
7.78 (s, 1H, C-8 proton), 7.45 (t, J ) 6 Hz, 0.56 H, ONdCH of
E-isomer), 6.85 (t, J ) 5 Hz, 0.44 H, ONdCH of Z-isomer), 6.20
(s, 2H, exchangeable with D2O, NH2), 4.50 (E) and 4.48 (Z) [two
t, J ) 7 (E) and 6 (Z) Hz, 2H total, OCH2], 3.75 (Z) and 3.69 (E)
[two s, 3H total, OCH3] and 2.69 (Z) and 2.59 (E) [two apparent
q, J ) 6 (Z) and 6 (E) Hz, 2H total, OCH2CH2]. 13C NMR
(DMSO-d6) δ 159.5 (C-2 and C-6), 156.31 (C-4), 148.4 and 148.1
(E/Z CdNO), 138.0 (C-8), 112.4 (C-5), 62.43, 62.29, 61.18, and
60.78 (E/Z OCH2 and OCH3), and 29.07 and 25.80 (E/Z CH2Cd
N). MS (FAB+) m/z calcd for C9H13N6O2 [M + H]+ 237.1101,
found 237.1107. UV (0.5 mg of product in 7.5 mL of CH3CN and
diluted with 7.5 mL of H2O): 282 and 240 nm; the extinction
coefficient at 282 nm was 7667 M-1cm-1 (mixture of isomers).
Fluorescence in CH3CN (ca. 1 mg in 10 mL): 286 (excitation)
and 348 (emission) nm.
1
for similar oximes (24). H NMR (DMSO-d6) δ 12.47 (br s, 1H,
exchangeable with D2O, NH), 7.86 (s, 1H, C-8 proton), 7.72
(apparent t, J ) 6 Hz, 0.35 H, ONdCH of E-isomer), 7.11
(apparent t, J ) 4 Hz, 0.65 H, ONdCH of Z-isomer), 6.31 (br s,
2H, exchangeable with D2O, NH2), 5.16 (Z) and 4.98 (E) [two d,
J ) 4 (Z) and 6 (E) Hz, 2H total, OCH2], and 3.86 (Z) and 3.80
(E) (two s, 3H total, OCH3). 13C NMR (DMSO-d6) δ 159.4 (C-2
and C-6), 155.4 (C-4), 148.5 and 146.2 (E/ Z CdNO), 138.3
(C-8), 113.2 (C-5), and 61.96, 61.55 and 60.01 (E/ Z OCH2 and
OCH3). MS (FAB+) m/z calcd for C8H11N6O2 [M + H]+ 223.0944,
found 223.0944. Diode array UV spectra indicated λmax ) 281
nm with another peak at 239 nm.
Kin etics. In general, samples were analyzed at 1-h intervals
using an HPLC unit with an auto-injector and temperature con-
trol. At each time point, chromatograms were generated using
UV and fluorescence detectors. Peak areas (using, in general,
fluorescence detection) were used to measure relative concentra-
tions of components as a function of time. Linear least-squares
fits of pseudo-first-order plots of the disappearance of individual
HPLC peaks provided the half-lives given in Table 1.
Phosphate-DMSO. An appropriate weight of the compound
of interest was dissolved in DMSO (0.1 mL) and this was then
diluted with 0.9 mL of 0.1 M phosphate buffer (potassium
phosphate monobasic) at pH 7.4. After mixing, the pH was
readjusted to 7.4.
Tris-DMSO. A buffer was prepared which was 50 mM in Tris,
0.1 mM in EDTA and 5 mM in dithiothreitol and the pH was
adjusted to 7.4. An appropriate weight of the compound of
interest was dissolved in DMSO (0.1 mL) and this was then
diluted with 0.9 mL of the Tris buffer. The pH was readjusted
to 7.4.
0.1 M HCl-DMSO. An appropriate amount of O6-PAG (8) or
O6-EAG (12) was dissolved in DMSO (0.05 mL) and this was
diluted with 0.1 M HCl (2.95 mL). The solution was mixed, and
an aliquot (0.5 mL) was analyzed immediately by HPLC. The
remaining solution was heated to 70 °C, and aliquots were taken
O6-(Allyl)gu a n in e (16). With minor modifications to pub-
lished procedures (19, 20, 25), freshly cut metallic sodium (11.8
mmol, 0.14 g) was added slowly to neat allyl alcohol (118 mmol,
8.0 mL) at room temperature. When the sodium was no longer
visible (∼10 min), 2-amino-6-chloropurine (5.9 mmol, 1.0 g) was
added, and the mixture was heated at 100 °C for 10 h. The
excess allyl alcohol was removed by distillation, and the residue
was flash chromatographed on silica gel (3.5 cm × 37 cm) using
CHCl3-CH3OH (3:1) to obtain the product as a white, crystalline

Modified Guanines Representing O6-Alkylation
Chem. Res. Toxicol., Vol. 15, No. 3, 2002 383
Ta ble 1. Solu tion Sta bility of O6-P AG (8), O6-EAG (12) a n d O6-P AG-Oxim e (15)
compd
concentration (mM)
solution
pH; temp
stabilitya
8
8
8
15
8
8
8
8
8
8
1
5
10
5
1
8
10
5
0.8
1.6
5
5
0.8
1.6
0.1 M phosphate-DMSO (9:1)
0.1 M phosphate-DMSO (9:1)
0.1 M phosphate-DMSO (9:1)
0.1 M phosphate-DMSO (9:1)
50 mM Trise - DMSO (9:1)
50 mM Trise-DMSO (92:8)
50 mM Trise-DMSO (9:1)
DMSO
0.1 M HCl-DMSO (98:2)
0.1 M HCl-DMSO (98:2)
0.1 M phosphate-DMSO (9:1)
50 mM Trise-DMSO (9:1)
0.1 M HCl-DMSO (98:2)
0.1 M HCl-DMSO (98:2)
7.4; 37 °C
7.4; 37 °C
7.4; 37 °C
7.4; 37 °C
7.4; 37 °C
7.4; 37 °C
7.4; 37 °C
h; RT
1.3; 70 °C
1.3; 70 °C
7.4; 37 °C
7.4; 37 °C
1.3; 70 °C
1.3; 70 °C
τ1/2 ) 57 min (0.998)b
τ1/2 ) 53 min (0.999)b
τ1/2 ) 57 ( 9 min (0.994 ( 0.003)b,c
no loss over 18 hd
τ1/2 ) 15 ( 2 h (0.993 ( 0.001)c,d,f
τ1/2 ) 16 h (0.995)d,f,g
τ1/2 ) 17 h (0.993)d,f
τ1/2 ) 46 h (0.960)d
10-15% loss/30 mini
35% loss/30 mini
12
12
12
12
τ1/2 ) 6.6 h (0.999)d
τ1/2 ) 2.5 days (0.732)d,f
15-20% loss/30 ming,i
15% loss/30 ming,i
a
Half-lives were determined using HPLC and fluorescence detection to measure changes in peak area with time. Numbers in parentheses
b
are correlation coefficients. In general, experimental error limits are estimated at (15%. Kinetic data derived from time points taken
hourly over 4-5 h. c Average of two runs. Kinetic data derived from time points taken hourly over 14-24 h. e Buffer also contained
d
EDTA and dithiothreitol (see Kinetics section in the Experimental Procedures for details). f In Tris, aldehydes 8 and 12 likely participate
in a pseudo-equilibrium with a more stable oxazolidine. Thus, the given values are ‘apparent half-lives’ (see Adduct Stabilities in text for
g
h
i
details). UV detection was used to measure peak areas. Ambient pH. Stability determined through ‘2 point’ kinetic plots (change in
concentration between 30 and 60 min of reaction time). A high degree of uncertainty is associated with these determinations.
widely demonstrated to yield accurate chemical structures and
reaction energies for most molecules (29). The molecular entro-
pies and thermal corrections to the free energies were calculated
from the harmonic vibrational constants determined by analytic
energy second derivatives calculated at the gas-phase optimized
structures.
Sch em e 4
To estimate the effect of aqueous solvation on the relative
energies, single point energies were calculated at the B3LYP
optimized geometries using a dielectric continuum solvation
model [COSMO (30)] coupled to the DFT B3LYP wave function.
This method has been shown to yield accurate solvation energies
for a wide range of compounds. In general, the use of gas-phase
optimized molecular structures in the calculation of solution-
phase energies has been found to yield relative energies in very
good agreement with those calculated from solution-phase
optimized structures (30-33). Additionally, the solvation ener-
gies were calculated using an independent model, the Langevin
Dipole method (34), which approximates the surrounding aque-
ous medium as a series of point dipoles fixed to a grid. The
Langevin Dipole solvation energies were calculated using
ChemSol 2.1 (34), using Merz-Kollman (28) atomic charges
derived from the B3LYP/6-31G (34) wave function using the
polarizable continuum model (35). Overall, there was good
agreement between the solvation energies calculated by the
COSMO and ChemSol methods. All other calculations were
performed using Gaussian 98 (36).
at 0.5 and 1 h intervals and kept frozen (-20 °C) until HPLC
analysis.
HP LC Meth od a n d Da ta . Compounds were separated on a
Waters Spherisorb S5 ODS2 reverse phase column (4.6 mm ×
250 mm) using a gradient mobile phase of deionized water and
acetonitrile, starting with 100% deionized water and increasing
linearly to 30% acetonitrile over 30 or 40 min, then increasing
linearly to 80% acetonitrile over 10 min and equilibrating back
to 100% water over 10 min. The flow rate was 1 mL/min and
the column was at room temperature. Compounds were moni-
tored by (1) fluorescence detection at λexcitation ) 295 nm and
λemission ) 375 nm using a Waters 474-Scanning Fluorescence
Detector and (2) UV detection at 290 nm using a Waters 996-
Photodiode Array Detector. Using the gradient where acetoni-
trile was increased from 0 to 30% over 30 min, the following
retention times ((1 min) were obtained: 8 (O6-PAG), 19 min;
14, 39 min; 15 (O6-PAG-oxime), 27 min; and 16, 27 min. Using
the gradient where acetonitrile was increased from 0 to 30%
over 40 min, the following retention times were obtained: 8 (O6-
PAG), 20 ( 2 min; 15 (O6-PAG-oxime), 35 ( 3 min; 12 (O6-EAG),
17 ( 1 min; and 17 (O6-EAG-oxime), 28 ( 1 min.
Resu lts a n d Discu ssion
Syn th esis a n d Id en tifica tion of Ad d u cts O6-P AG
(8) a n d O6-EAG (12). Acrolein adduct O6-PAG (8) was
synthesized using variations on literature procedures for
other O6-derivitized guanines (19-22, 25). As shown in
Scheme 4, reaction of the sodium salt of 3-buten-1-ol with
2-amino-6-chloropurine gave intermediate 14. Oxidative
cleavage (23) of the alkene functionality in 14 provided
the desired aldehyde, O6-PAG (8). 1H NMR spectra of the
crude product showed a signal indicative of an aldehydic
proton; resonances for an alkene moiety were absent.
HPLC analysis using UV detection gave one major and
several small peaks, suggesting the presence of minor
impurities and/or equilibrium structures (e.g., 10 and/or
11, Scheme 2), which will be discussed in a later section.
The major product visible by UV was the only component
to give an intense signal upon detection with fluorescence
[a characteristic of O6- but not N-alkylated derivatives
of guanines (37)].
Qu a n t u m Ch em ica l Ca lcu la t ion s. The structures of all
compounds were optimized in the gas-phase using density
functional theory DFT with the Becke three parameter exchange
(26) and Lee-Yang-Parr correlation functional (B3LYP) (27)
with a 6-31+G(d,p) basis set (28), an approach that has been

384 Chem. Res. Toxicol., Vol. 15, No. 3, 2002
Balu et al.
Due to the fact that chromatographic (silica gel and
reverse phase) purifications of O6-PAG were unsatisfac-
tory, the structural integrity of this compound was
established by generating a derivative of 8 with O-
methylhydroxylamine. The resultant oxime was much
more stable than its aldehydic precursor and, therefore,
was readily chromatographed and characterized (NMR,
MS) as compound 15, the E- and Z-isomers of the
O-methyl oxime of O6-PAG (Scheme 4). When the oxime
reaction was followed by HPLC, the peak assigned to O6-
PAG rapidly disappeared upon addition of O-methylhy-
droxylamine. This was accompanied by the appearance
of a peak attributable to oxime 15 (as identified with
authentic material).
As shown in Scheme 4, O6-EAG (12) was made by a
pathway parallel to that for O6-PAG with the previously
reported O6-allyl 16 as the intermediate (20, 25). In the
1H spectrum of 16, the presence of a singlet at δ 7.84 for
the C-8 proton was particularly useful in determining the
isomeric integrity of this compound. It has been shown
that 16 and related O6-allylguanines are subject to a
rearrangement wherein the allylic group shifts from the
O6 to the C-8 position; however, this rearrangement can
be suppressed with appropriate control of temperature
and reaction times (37).
Oxidation (23) of 16 provided O6-EAG (12) and, as with
O6-PAG, this compound was difficult to isolate in pure
form. Crude O6-EAG showed an aldehyde resonance by
proton NMR and a major component by HPLC that
displayed strong fluorescence. As with O6-PAG, the
possibility of O6-EAG interconverting with one or more
tautomers was considered (e.g., 13, Scheme 3), as dis-
cussed in a later section. Conversion of 12 to an oxime
with O-methylhydroxylamine provided the more stable
and readily characterized E/ Z-17, the diastereomeric
O-methyl oximes of O6-EAG (Scheme 4). As with O6-PAG,
the oxime reaction was followed by HPLC. Addition of
O-methylhydroxylamine resulted in the disappearance
of the peak assigned to O6-EAG concomitant with the
appearance of a signal for oxime 17.
Ad d u ct Sta bilities. The stability of O6-PAG (8) was
studied under a variety of conditions using HPLC to
monitor changes in component concentration (Table 1).
The loss of O6-PAG was accompanied by the appearance
of guanine as identified through the use of authentic
material. The formation of guanine (and acrolein) could
be accounted for by an R,â-elimination reaction initiated
by base-catalyzed removal of a proton R to the carbonyl
in 8. On the basis of this likely mechanism, it could be
predicted that the decomposition of O6-PAG would be
dependent on a variety of conditions including buffer-
type and pH. Such condition-dependent rates are known
for aldophosphamide (3) which undergoes an analogous
R,â-elimination to give acrolein and phosphoramide
mustard (4, Scheme 1) (1, 38-42).
As shown in Table 1, the apparent lifetime of O6-PAG
in Tris is much longer than that in phosphate. This is
not surprising in view of the fact that Tris is known to
effectively enhance the stability of aldophosphamide (3)
(38, 39). In a reversible reaction, Tris adds across the
aldehydic functionality in 3 and, presumably, in O6-PAG
to give an oxazolidine product (38, 39). The loss of the
carbonyl moiety decreases the acidity of the hydrogens
on the adjacent methylene group and this disfavors any
R,â-elimination reaction of the oxazolidine. In the inter-
conversion of the aldehydic species and the oxazolidine,
higher concentrations of Tris favor the oxazolidine and,
conversely, lower concentrations favor the aldehyde.
Thus, injection of the O6-PAG/Tris reaction mixture onto
an HPLC column (using Tris-free eluent) results in a
significant dilution of Tris, and the pseudo-equilibrium
between oxazolidine and aldehyde shifts to greatly favor
O6-PAG. The free aldehyde (or its hydrate) is the species
that is then detected. While Tris is likely the major
contributor to the rate retardation seen in these experi-
ments, the presence of dithiothreitol in this buffer may
also play some role. Again, with aldophosphamide, thiols
are known to add in a reversible manner across the
aldehydic moiety producing more stable hemithioacetals
(1, 38, 43). The apparent stability of O6-PAG in this
particular Tris buffer system (containing EDTA and
dithiothreitol) was of particular interest because we have
used it for past studies of DNA alkylation (7, 44) and we
expect to use it for future investigations of O6-guanyl
alkylation by acrolein. The good stability of O6-PAG (via
its oxazolidine) under these conditions suggests that if
such an adduct is formed in DNA, it should convert to
the oxazolidine and, therefore, survive long enough to be
detected.
Under conditions similar to those used in DNA diges-
tion (0.1 M HCl, 70 °C, 30 min), O6-PAG was reasonably
stable with an estimated decomposition rate of 10-35%/
30 min (Table 1). A high degree of uncertainty was
associated with these determinations as each kinetic plot
employed just two time points. Nevertheless, it was clear
from the chromatograms that a substantial portion of the
starting material survived the hydrolysis conditions.
Assuming that the major loss of O6-PAG was a result of
base-catalyzed elimination, it was consistent that acidic
conditions would favor stability, even at the temperature
extreme of 70 °C.
Aldophosphamide (3) is readily converted to stable
oximes through reactions with hydroxylamines (24, 45).
In anticipation that this chemistry might have some
applications to the ultimate goal of detecting O6-alkylated
residues in DNA, the stability of the O-methylhydroxyl-
oxime of O6-PAG (15) was studied in a limited manner.
HPLC analyses of solutions of 15 at pH 7.4 in phosphate/
DMSO showed no signs of decomposition over 18 h at 37
°C (Table 1).
Under similar conditions, O6-EAG (12) was more stable
than O6-PAG (Table 1) but, as with O6-PAG, the loss of
O6-EAG was accompanied by the appearance of guanine.
The likely mechanism of decomposition for O6-EAG is an
irreversible substitution reaction whereby a nucleophile
attacks the carbon adjacent to the carbonyl resulting in
the formation of an R-substituted acetic acid and the
liberation of guanine. The rate of displacement can be
modulated by various factors including nucleophile
strength (e.g., water versus phosphate versus Tris) and
changes in the electrophilicity and accessibility of the
If the fragmentation of O6-PAG to guanine and acrolein
were reversible, the rate of reaction would also be
dependent on the concentrations of the reactants. Within
experimental error limits ((15%), no such dependency
was observed in these kinetic studies; however, no
precautions were taken to prevent or minimize the loss
of the volatile acrolein from the reaction mixture. Thus,
the half-lives given in Table 1 reflect irreversible frag-
mentation rates but that may be a function of experi-
mental design.

Modified Guanines Representing O6-Alkylation
Chem. Res. Toxicol., Vol. 15, No. 3, 2002 385
Sch em e 5a
Ta ble 2. Rela tive En er gies of th e Gu a n in e-Acr olein a n d
Gu a n in e-Ch lor oa ceta ld eh yd e Ad d u ctsa
aqueous phase
gas phaseb
COSMO
∆Gd
(kcal/mol)
ChemSol
∆Ge
(kcal/mol)
∆H
(kcal/mol)
∆Gc
(kcal/mol)
compd
8
9
18
19
20
8.91
18.25
0.00
4.04
0.91
7.20
17.89
0.00
3.36
-0.45
10.71
21.85
2.95
1.07
0.00
11.35
24.04
2.65
-3.38
0.00
10
11
21
22
23
24
34.57
16.82
2.58
31.02
0.00
34.45
16.57
1.72
30.66
0.00
28.21
15.74
0.00
25.32
2.03
23.43
14.66
0.00
20.24
1.28
12.79
12.65
13.34
11.24
12
13
0.00
9.21
0.00
11.27
0.00
6.14
0.00
9.07
a
Energies are computed relative to the most stable isomer
b
among the monofunctional and cyclic subsets of adducts. Gas-
phase energies calculated at the B3LYP/6-31+G (28) level of
theory. c All energies were calculated at the gas-phase B3LYP/
6-31+G (28) optimized geometries. Thermal enthalpy corrections
and entropy terms were calculated from the B3LYP/6-31+G (28)
a
∆G in kcal/mol, COSMO method.
d
harmonic vibrational frequencies. Calculated using the conduc-
tor-like screening solvation model (COSMO) using the B3LYP/
6-31+G (28) wave function. e Calculated using the Langevin Dipole
solvent model (34) (ChemSol) using Merz-Kollman (28) atomic
charges derived from the B3LYP/6-31G (34) wave function using
the polarizable continuum model (35).
alpha carbon (as in conversion of the carbonyl to an
oxazolidine with Tris).
Ab In itio Stu d ies of O- ver su s N-Alk yla tion s a n d
Ta u tom er F or m a tion . To date, the major products
isolated from reactions between acrolein and guanine (or
deoxyguanosine) arise from conjugate addition at the N-1
(18) and, especially, the exocyclic NH2 (19) positions on
the guanyl ring (Scheme 5). Each of these adducts is
isolated as its more stable tautomer derived from in-
tramolecular cyclization: 18 f 21 and 19 f 23 (Scheme
5) (9, 10, 15). The formation of similar tautomeric
structures was considered for O6-PAG (8) and O6-EAG
(12), but no experimental evidence was obtained to
support the occurrence of any such reactions (Scheme 2,
8 f 10 and/or 11; Scheme 3, 12 f 13).
To enhance our experimental studies of alkylated
guanines, ab initio quantum chemical calculations were
used to understand the energy factors that underlie the
occurrence of O- versus N-alkylations as well as possible,
subsequent intramolecular cyclizations (Tables 2 and 3;
Schemes 2, 3, and 5). For each reaction of interest, the
reaction free energy was calculated using quantum
chemical density functional methods and including aque-
ous solvation effects using implicit solvent models
(COSMO and ChemSol, see Experimental Procedures for
details).
For guanine-acrolein and guanine-chloroacetaldehyde
adducts, relative enthalpies and free energies in the gas
phase as well as relative free energies in the aqueous
phase (COSMO and ChemSol methods) are given in
Table 2. Reaction free energies for each alkylation and
subsequent cyclization reaction are given in Table 3 (both
gas and aqueous phase, Cosmo and ChemSol methods).
Because the COSMO method is currently believed to give
the most accurate solvation energies, these are the
numbers reported in Schemes 2, 3, and 5. In all cases, a
lower (or more negative) energy indicates a more favored
structure (Table 2) or reaction (Table 3 and the schemes).
The reaction free energies indicate which reactions are
thermodynamically favored, and therefore, they predict
the equilibrium distribution of products. If, however,
there are significant energy barriers to these reactions,
Ta ble 3. En er gies of th e Rea ction s In volvin g th e
F or m a tion of Gu a n in e-Acr olein (5) a n d
Gu a n in e-Ch lor oa ceta ld eh yd e(7) Ad d u cts
aqueous phase
gas phasea
COSMO ChemSol
∆H
∆Gb
∆Gc
∆Gd
reaction
(kcal/mol) (kcal/mol) (kcal/mol) (kcal/mol)
guanine + 5 f 8
guanine + 5 f 9
guanine + 5 f 18
guanine + 5 f 19
guanine + 5 f 20
8 f 10
-11.07
-1.73
-19.98
-15.93
-19.06
22.40
4.66
0.91
11.60
-6.29
-2.92
-6.74
25.74
7.87
8.42
19.56
0.66
-1.23
-2.29
12.57
0.10
11.81
24.50
3.11
-2.92
0.46
10.34
1.57
8 f 11
18 f 21
18 f 22
19 f 23
20 f 24
guanine + 7 f 12
12 f 13
-0.67
27.76
-7.30
8.63
4.43
9.21
0.21
-7.88
17.44
-3.96
8.42
14.11
6.14
-4.39
15.85
2.93
9.50
14.91
9.07
29.15
-4.87
11.60
7.21
11.27
a
Gas-phase energies calculated at the B3LYP/6-31+G (28) level
b
of theory. All energies were calculated at the gas-phase B3LYP/
6-31+G (28) optimized geometries. Thermal enthalpy corrections
and entropy terms were calculated from the B3LYP/6-31+G (28)
harmonic vibrational frequencies. c Calculated using the conduc-
tor-like screening solvation model (COSMO) using the B3LYP/6-
d
31+G (28) wave function. Calculated using the Langevin Dipole
solvent model (34) (ChemSol) using Merz-Kollman (28) atomic
charges derived from the B3LYP/6-31G (34) wave function using
the polarizable continuum model (35).
a different product distribution may be kinetically fa-
vored.
As shown in Scheme 2 for O6-alkylation reactions
between acrolein and guanine, the COSMO quantum
chemical calculations predict that the conjugate addition
reaction to form product 8 is favored (∆G ) 8.4 kcal/mol)
relative to direct addition to form product 9 (∆G ) 19.6
kcal/mol). As described below, however, any of the three
possible N-alkylation reactions is significantly favored

386 Chem. Res. Toxicol., Vol. 15, No. 3, 2002
Balu et al.
over the O-alkylations (Scheme 5, adducts 18, 19, and
20).
Con clu sion
O6-PAG (8) and O6-EAG (12) were synthesized for use
as authentic standards in our continuing investigations
of DNA alkylation by acrolein and chloroacetaldehyde.
Studies of the chemical stabilities of these adducts
suggest that guanyl residues in DNA that have been
alkylated by acrolein or chloroacetaldehyde will have
lifetimes sufficient for detection. No experimental support
was obtained for the formation of tautomeric structures
from O6-PAG or O6-EAG.
Ab initio quantum chemical simulations of the free
energies of reactions between acrolein and guanine
indicate that N-alkylation is favored over O6-alkylation
and that subsequent cyclization reactions are most
favorable if they involve the N-1 or NH2 positions. When
drawing conclusions from these data, however, it must
be remembered that the calculated free energies reflect
which reactions are thermodynamically favored. If there
are significant energy barriers to these reactions, the
actual product distribution could be different as a result
of kinetic rather than thermodynamic control. Further-
more, the electronic changes that occur upon alkylation
of guanine are not the same as those that derive from
alkylation of N-9 substituted guanines (as in DNA).
Predictions of relative product formation in DNA itself
would be useful and, to this end, ab initio calculations of
the type presented here are in progress for models of
alkylated DNA.
After an initial R,â-conjugate addition, O6-PAG (8) can
undergo intramolecular cyclization via nucleophilic at-
tack on the carbonyl to form one or more cyclic tautomers
(10 and 11, Scheme 2). Interestingly, tautomer 11 with
a seven-membered ring is favored over the six-membered
ring 10; however, both reactions 8 f 10 and 8 f 11 are
endothermic.
For N-alkylation reactions between acrolein and gua-
nine, Scheme 5 shows the possible alkylation products
and their attendant cyclic tautomers. The three possible
initial alkylation reactions are found to be very similar
in energy, with reaction at N-7 (to give 20) slightly
favored over that at N-1 (to give 18) or the exocyclic NH2
(to give 19). While intermediate 18 could cyclize to 21 or
22, only 21 has been reported (9, 10). This is in agreement
with the COSMO and ChemSol predictions which show
that the difference in free energies between the forma-
tions of 21 and 22 is 20-25 kcal/mol in favor of 21 (Table
3). Adduct 19 is found to have a favorable reaction energy
to form 23; in contrast, the cyclization of adduct 20 to
give seven-membered ring 24 is predicted to have an
unfavorable reaction free energy (Table 3).
The relative enthalpies for various conformers of 21
and 23 have been reported using semiempirical methods
that find 21 is favored over 23 by about 1 kcal/mol when
using gas-phase PM3 and AM1 calculations (46). The
reverse is found using the MNDO formalism where the
enthalpy of formation for 23 is 1.3 kcal/mol lower than
that of 21 (46). The MNDO result is in close agreement
with our more accurate ab initio quantum chemical, gas-
phase, free energy calculations that show 23 to be lower
in energy by 1.7 kcal/mol compared to 21 (Table 2). In
the aqueous-phase, we predict the reverse of this order-
ing, with 21 being favored by 1.3 (ChemSol) to 2.0
(COSMO) kcal/mol (Table 2).
We hypothesize that O6-alkylation of guanine by ac-
rolein will play a role in the mutagenicity associated with
cyclophosphamide (and ifosfamide) therapy. The syn-
thetic adducts reported here will be used to further
investigations of this hypothesis.
Ack n ow led gm en t. We would like to thank Dr.
Henry S. Friedman (Duke University Medical Center),
Dr. Robert F. Struck (Southern Research Institute), and
Dr. Anthony E. Pegg (The Milton S. Hershey Medical
Center) for their contributions in recognizing the poten-
tial for alkylation of O6-guanyl groups by acrolein. This
work was supported in part by Public Health Service
Grants CA68438 (S.M.L.), CA16783, CA692460, and
CA52857 (O.M.C.), CA71627 (M.E.D.), and CA57725
(M.E.D.) from the National Cancer Institute (Department
of Health and Human Resources) and by DAMD 17-99-
1-9175 (M.P.G.). The work at Lawrence Livermore Na-
tional Laboratory was performed under contract W-7405-
ENG-48 from the U.S. Department of Energy. The NMR
spectroscopy and mass spectrometry results were ac-
quired through the Shared Instrumentation Facility of
the Duke University Medical Center.
We also predicted the free energies for the reaction of
chloroacetaldehyde (7) with guanine at the O6 position
as shown in Scheme 3. As with acrolein O6-alkylation
reactions, we find the O6-alklyation by 7 to be energeti-
cally unfavored, with ∆G > 14 kcal/mol. Similarly, a
subsequent intramolecular cyclization to 13 is predicted
to be endothermic.
In summary, the predicted free energies for the reac-
tions of acrolein with guanine indicate that alkylations
at the N-1, NH2, and N-7 positions of guanine are favored
over alkylation at O6. Further, for the subsequent cy-
clization reactions, we predict that only the reactions to
form cyclic products including N-1 and NH2 (i.e., 21 and
23) are energetically favored. This latter prediction
correlates with in vitro experimental results: the major
products isolated from reactions between acrolein and
guanine are 21 and 23 (9, 10). A comparison of the
aqueous-phase free energies of 21 and 23 predict that
21 is slightly favored over 23. The same order of fa-
vorability in the guanine adducts would not necessarily
be found for guanosine analogues; however, it can be
noted that only the guanosine analogue of 23 has been
identified in the DNA of rodents and humans (47). The
authors of this latter study stated that although their
results suggested that the guanosine analogue of 23 was
formed preferentially in vivo, it was possible that gua-
nosine structures corresponding to 21 were not detected
because they were more unstable or more efficiently
repaired.
Refer en ces
(1) Ludeman, S. M. (1999) The chemistry of the metabolites of
cyclophosphamide. Curr. Pharm. Des. 5, 627-643.
(2) Colvin, M., and Chabner B. A. (1991) Alkylating agents. In Cancer
Chemotherapy: Principles and Practice (Chabner, B. A., and
Collins, J . M., Eds.) pp 276-313, J . L. Lippincott, Philadelphia.
(3) Tolly, D. A. (1977) The effect of N-acetyl cysteine on cyclophos-
phamide cystitis. Brit. J . Urol. 49, 659-661.
(4) Williams, M. L., and Wainer, I. W. (1999) Cyclophosphamide
versus ifosfamide: “To use ifosfamide or not to use, that is the
three dimensional question”. Curr. Pharm. Des. 5, 665-672.
(5) Kerbusch, T., de Kraker, J ., Keizer, H. J ., van Putten J . W. G.,
Groen, H. J . G., J ansen, R. L. H., Schellens, J . H. M., and Beijnen,
J . H. (2001) Clinical pharmacokinetics and pharmacodynamics
of ifosfamide and its metabolites. Drug Dispos. 40, 41-62.

Modified Guanines Representing O6-Alkylation
Chem. Res. Toxicol., Vol. 15, No. 3, 2002 387
(6) Friedman, H. S., Pegg, A. E., J ohnson, S. P., Loktionova, N. A.,
Dolan, M. E., Modrich, P., Moschel, R. C., Struck, R., Brent, T.
P., Ludeman, S., Bullock, N., Kilborn, C., Keir, S., Dong, Q., and
Bigner, D. D. (1999) Modulation of cyclophosphamide activity by
O6-alkylguanine-DNA alkyltransferase. Cancer Chemother. Phar-
macol. 43, 80-85.
(7) Cai, Y., Wu, M. H., Ludeman, S. M., Grdina, D. J ., and Dolan,
M. E. (1999) Role of O6-alkylguanine-DNA alkyltransferase in
protecting against cyclophosphamide-induced toxicity and mu-
tagenicity. Cancer Res. 59, 3059-3063.
(28) Besler, B. H., Merz, K. M., and Kollman, P. (1990) Atomic charges
derived from semiempirical methods. J . Comput. Chem. 11, 431-
439.
(29) J ensen, F. (1999) Introduction to Computational Chemistry, J ohn
Wiley and Sons, New York.
(30) Barone, V., and Cossi, M. (1998) Quantum calculation of molecular
energies and energy gradients in solution by a conductor solvent
model. J . Phys. Chem. A 102, 1995-2001.
(31) Tran, N. L., and Colvin, M. E. (2000) The prediction of biochemical
acid dissociation constants using first principles quantum chemi-
cal simulations. J . Mol. Structure, Theochem. 532, 127-137.
(32) Schuurmann, G. (2000) Prediction of Henry’s law constant of
benzene derivatives using quantum chemical continuum-solvation
models. J . Comput. Chem. 21, 17-34.
(33) Alagona, G., Ghio, C., Nagy, P. I., and Durant, G. J . (1999)
Theoretical studies on the continuum solvation of some N,N′-
dimethyl- and N-methyl-N′-acetyl-guanidine and guanidinium
conformers. J . Phys. Chem. A 103, 1857-1867.
(34) Florian, J ., and Warshel, A. (1997) Langevin dipoles model for
ab initio calculations of chemical processes in solution: Param-
etrization and application to hydration free energies of neutral
and ionic solutes and conformational analysis in aqueous solution.
J . Phys. Chem. B 101, 5583-5595.
(35) Cossi, M., Barone, V., Cammi, R., and Tomasi, J . (1996) Ab initio
study of solvated molecules: A new implementation of the
polarizable continuum model. Chem. Phys. Lett. 255, 327-335.
(36) Frisch, M. J ., Trucks, G. W., Schlegel, H. B., Scuseria, G. E., Robb,
M. A., and Cheeseman, J . R., et al. (1998) Gaussian 98, Revision
A.4, Gaussian Inc., Pittsburgh.
(37) Frihart, C., and Leonard, N. (1973) Allylic rearrangement from
O6 to C-8 in the Guanine Series. J . Am. Chem. Soc. 95, 7174-
7175.
(8) Clayden, J ., Greeves, N., Warren, S., and Wothers, P. (2001)
Organic Chemistry, pp 227-242, Oxford University Press, New
York.
(9) Chung, F. L., Young, R., and Hecht, S. S. (1984) Formation of
cyclic 1,N2-propanodeoxyguanosine adducts in DNA upon reaction
with acrolein or crotonaldehyde. Cancer Res. 44, 990-995.
(10) Eder, E., Hoffman, C., Bastian, H., Deininger, C., and Scheck-
enbach, S. (1990) Molecular mechanisms of DNA damage initiated
by alpha, beta-unsaturated carbonyl compounds as criteria for
genotoxicity and mutagenicity. Environ. Health Perspect. 88, 99-
106.
(11) Eder, E., and Hoffman, C. (1993) Identification and characteriza-
tion of deoxyguanosine adducts of mutagenic beta-alkyl-substi-
tuted acrolein congeners. Chem. Res. Toxicol. 6, 486-494.
(12) Eder, E., Scheckenbach, S., Deininger, C., and Hoffman, C. (1993)
The possible role of alpha, beta-unsaturated carbonyl compounds
in mutagenesis and carcinogenesis. Toxicol. Lett. 67, 87-103.
(13) Hoffman, C., Bastian, H., Wiedenmann, M., Deininger, C., and
Eder, E. (1989) Detection of acrolein congener-DNA adducts
isolated from cellular systems. Arch. Toxicol. Suppl. 13, 219-
223.
(14) Khullar, S., Varaprasad, C. V., and J ohnson, F. (1999) Postsyn-
thetic generation of a major acrolein adduct of 2′-deoxyguanosine
in oligomeric DNA. J . Med. Chem. 42, 947-950.
(15) Nechev, L. V., Harris, C. M., and Harris, T. M. (2000) Synthesis
of nucleosides and oligonucleotides containing adducts of acrolein
and vinyl chloride. Chem. Res. Toxicol. 13, 421-429.
(16) VanderVeen, L. A., Hashim, M. F., Nechev, L. V., Harris, T. M.,
Harris, C. M., and Marnett, L. J . (2001) Evaluation of the
mutagenic potential of the principal DNA adduct of acrolein. J .
Biol. Chem. 276, 9066-9070.
(17) Yang, I. Y., Hossain, M., Miller, H., Khullar, S., J ohnson, F.,
Grollman, A., and Moriya, M. (2001) Responses to the major
acrolein-derived deoxyguanosine adduct in Escherichia coli. J .
Biol. Chem. 276, 9071-9076.
(38) Zon, G., Ludeman, S. M., Brandt, J . A., Boyd, V. L., Ozkan, G.,
Egan, W., and Shao, K.-L. (1984) NMR spectroscopic studies of
intermediary metabolites of cyclophosphamide. A comprehensive
kinetic analysis of the interconversion of cis- and trans-4-
hydroxycyclophosphamide with aldophosphamide and the con-
comitant partitioning of aldophosphamide between irreversible
fragmentation and reversible conjugation pathways. J . Med.
Chem. 27, 466-485.
(39) Borch, R. F., and Getman, K. M. (1984) Base-catalyzed hydrolysis
of 4-hydroperoxycyclophosphamide: Evidence for iminocyclophos-
phamide as an intermediate. J . Med. Chem. 27, 485-490.
(18) Wang, M., McIntee, E. J ., Cheng, G., Shi, Y., Villalta, P. W., and
Hecht, S. S. (2001) A Schiff base is a major DNA adduct of
crotonaldehyde. Proc. Am. Assoc. Cancer Res. 42, 2544.
(19) Eder, E., Lutz, D., and J orns, M. (1987) Allylic compounds bind
directly to DNA: Investigation of the binding mechanisms in
vitro. Chem.-Biol. Interact. 61, 97-108.
(20) Griffin, R. J ., Arris, C. E., Bleasdale, C., Boyle, F. T., Calvert, A.
H., Curtin, N. J ., Dalby, C., Kanugula, S., Lembicz, N. K., Newell,
D. R., Pegg, A. E., and Golding, B. T. (2000) Resistance-modifying
agents. 8. Inhibition of O6-alkylguanine-DNA alkyltransferase by
O6-alkenyl-, O6-cycloalkenyl-, and O6-(2- oxoalkyl)guanines and
potentiation of temozolomide cytotoxicity in vitro by O6-(1-
cyclopentenylmethyl)guanine. J . Med. Chem. 43, 4071-4083.
(21) Chae, M. Y., McDougall, M. G., Dolan, M. E., Swenn, K., Pegg,
A. E., and Moschel, R. C. (1994) Substituted O6-benzylguanine
derivatives and their inactivation of human O6-alkylguanine-DNA
alkyltransferase. J . Med. Chem. 37, 342-347.
(22) Moschel, R. C., McDougall, M. G., Dolan, M. E., Stine, L., and
Pegg, A. E. (1992) Structural features of substituted purine
derivatives compatible with depletion of human O6-alkylguanine-
DNA alkyltransferase. J . Med. Chem. 35, 4486-4491.
(23) Thaisrivongs, S., Pals, D. T., Turner, S. R., and Kroll, L. T. (1988)
Conformationally constrained renin inhibitory peptides: gamma-
lactam-bridged dipeptide isostere as conformational restriction.
J . Med. Chem. 31, 1369-1376.
(24) Zon, G., Ludeman, S. M., Sweet, E. M., Egan, W., and Phillips,
L. R. (1982) O-Methylhydroxylamine as a new trapping reagent
for quantitative studies of 4-hydroxycyclophosphamide and al-
dophosphamide. J . Pharm. Sci. 71, 443-446.
(40) Kwon, C. H., Maddison, K., LoCastro, L., and Borch, R. F. (1987)
Accelerated decomposition of 4-hydroxycyclophosphamide by hu-
man serum albumin. Cancer Res. 47, 1505-1508.
(41) Low, J . E., Borch, R. F., and Sladek, N. E. (1983) Further studies
on the conversion of 4-hydroxyoxazaphosphorines to reactive
mustards and acrolein in inorganic buffers. Cancer Res. 43, 5815-
5820.
(42) Borch, R. F., and Millard, J . A. (1987) The mechanism of
activation of 4-hydroxycyclophosphamide. J . Med. Chem. 30, 427-
431.
(43) Dirven, H. A., Venekamp, J . C., van Ommen, B., and van
Bladeren, P. J . (1994) The interaction of glutathione with 4-hy-
droxycyclophosphamide and phosphoramide mustard, studied by
31P nuclear magnetic resonance spectroscopy. Chem.-Biol. Inter-
act. 93, 185-196.
(44) Cai, Y., Wu, M. H., Xu-Welliver, M., Pegg, A. E., Ludeman, S.
M., and Dolan, M. E. (2000) Effect of O6-benzylguanine on
alkylating agent-induced toxicity and mutagenicity in Chi-
nese hamster ovary cells expressing wild-type and mutant
O6-alkylguanine-DNA alkyltransferases. Cancer Res. 60, 5464-
5469.
(45) Ludeman, S. M., Shulman-Roskes, E. M., Wong, K. K., Han, S.
Y., Anderson, L. W., Strong, J . M., and Colvin, O. M. (1995) Oxime
derivatives of the intermediary oncostatic metabolites of cyclo-
phosphamide and ifosfamide: Synthesis and deuterium labeling
for applications to metabolite quantification. J . Pharm. Sci. 84,
393-398.
(46) Boerth, D. W., Eder, E., Hussain, S., and Hoffman, C. (1998)
Structures of acrolein-guanine adducts: A semi-empirical self-
consistent field and nuclear magnetic resonance spectral study.
Chem. Res. Toxicol. 11, 284-294.
(47) Nath, R. G., and Chung, F.-L. (1994) Detection of exocyclic
1,N6-propanodeoxyguanosine adducts as common DNA lesions
in rodents and humans. Proc. Natl. Acad. Sci. U.S.A. 91,
7491-7495.
(25) Eder, E. (2001) University of Wurzburg, Germany, personal
communication. The NMR data listed in ref 19 likely contains a
typographical error for the assignment of the C-8 proton in O6-
allylguanine.
(26) Becke, A. D. (1993) Density-functional thermochemistry. III. The
role of exact exchange. J . Chem. Phys. 98, 5648-5652.
(27) Lee, C., Yang, W., and Parr, R. G. (1988) Development of the Colle-
Salvetti correlation-energy formula into a functional of the
electron density. Phys. Rev. B 37, 785-789.
TX0101503