Interstrand and intrastrand DNA-DNA cross-linking by 1,2,3,4-diepoxybutane: Role of stereochemistry

Article abstract of DOI:10.1021/ja051979x

1,2,3,4-Diepoxybutane (DEB) is a bifunctional electrophile capable of forming DNA-DNA and DNA-protein cross-links. DNA alkylation by DEB produces N7-(2′-hydroxy-3′,4′-epoxybut-1′-yl)-guanine monoadducts, which can then form 1,4-bis-(guan-7-yl)-2,3-butanediol (bis-N7G-BD) lesions. All three optical isomers of DEB are produced metabolically from 1,3-butadiene, but S,S-DEB is the most cytotoxic and genotoxic. In the present work, interstrand and intrastrand DNA-DNA cross-linking by individual DEB stereoisomers was investigated by PAGE, mass spectrometry, and stable isotope labeling. S,S-, R,R-, and meso-diepoxides were synthesized from L-dimethyl-2,3-O-isopropylidene-tartrate, D-dimethyl-2,3-O-isopropylidene- tartrate, and meso-erythritol, respectively. Total numbers of bis-N7G-BD lesions (intrastrand and interstrand) in calf thymus DNA treated separately with S,S-, R,R-, or meso-DEB (0.01-0.5 mM) were similar as determined by capillary HPLC-ESI⁺-MS/MS of DNA hydrolysates. However, denaturing PAGE has revealed that S,S-DEB produced the highest number of interchain cross-links in 5′-GGC-3′/3′-CCG-5′ sequences. Intrastrand adduct formation by DEB was investigated by a novel methodology based on stable isotope labeling HPLC-ESI⁺-MS/MS. Meso DEB treatment of DNA duplexes containing 5′-[1,7, NH₂-¹⁵N₃,2- ¹³C-G]GC-3′/3′-CCG-5′ and 5′-GGC-3′/ 3′-CC[¹⁵N₃,2-¹³C-G]-5′ trinucleotides gave rise to comparable numbers of 1,2-intrastrand and 1,3-interstrand bis-N7G-BD cross-links, while S,S DEB produced few intrastrand lesions. R,R-DEB treated DNA contained mostly 1,3-interstrand bis-N7G-BD, along with smaller amounts of 1,2-interstrand and 1,2-intrastrand adducts. The effects of DEB stereochemistry on its ability to form DNA-DNA cross-links may be rationalized by the spatial relationships between the epoxy alcohol side chains in stereoisomeric N7-(2′-hydroxy-3′,4′-epoxybut-1′-yl)- guanine adducts and their DNA environment. Different cross-linking specificities of DEB stereoisomers provide a likely structural basis for their distinct biological activities.

Full text of DOI:10.1021/ja051979x

Published on Web 09/22/2005
Interstrand and Intrastrand DNA-DNA Cross-Linking by
1,2,3,4-Diepoxybutane: Role of Stereochemistry
Soobong Park,† Christopher Anderson,† Rachel Loeber,†
Mahadevan Seetharaman,† Roger Jones,‡ and Natalia Tretyakova*,†
Contribution from the Cancer Center and the Department of Medicinal Chemistry, UniVersity of
Minnesota, Minneapolis, Minnesota 55455 and Department of Chemistry, Rutgers UniVersity,
Piscataway, New Jersey 08854
Received March 28, 2005; E-mail: trety001@umn.edu
Abstract: 1,2,3,4-Diepoxybutane (DEB) is a bifunctional electrophile capable of forming DNA-DNA and
DNA-protein cross-links. DNA alkylation by DEB produces N7-(2′-hydroxy-3′,4′-epoxybut-1′-yl)-guanine
monoadducts, which can then form 1,4-bis-(guan-7-yl)-2,3-butanediol (bis-N7G-BD) lesions. All three optical
isomers of DEB are produced metabolically from 1,3-butadiene, but S,S-DEB is the most cytotoxic and
genotoxic. In the present work, interstrand and intrastrand DNA-DNA cross-linking by individual DEB
stereoisomers was investigated by PAGE, mass spectrometry, and stable isotope labeling. S,S-, R,R-,
and meso-diepoxides were synthesized from L-dimethyl-2,3-O-isopropylidene-tartrate, D-dimethyl-2,3-O-
isopropylidene-tartrate, and meso-erythritol, respectively. Total numbers of bis-N7G-BD lesions (intrastrand
and interstrand) in calf thymus DNA treated separately with S,S-, R,R-, or meso-DEB (0.01-0.5 mM) were
similar as determined by capillary HPLC-ESI+-MS/MS of DNA hydrolysates. However, denaturing PAGE
has revealed that S,S-DEB produced the highest number of interchain cross-links in 5′-GGC-3′/3′-CCG-5′
sequences. Intrastrand adduct formation by DEB was investigated by a novel methodology based on stable
isotope labeling HPLC-ESI+-MS/MS. Meso DEB treatment of DNA duplexes containing 5′-[1,7, NH2-15N3,2-
13C-G]GC-3′/3′-CCG-5′ and 5′-GGC-3′/3′-CC[15N3,2-13C-G]-5′ trinucleotides gave rise to comparable numbers
of 1,2-intrastrand and 1,3-interstrand bis-N7G-BD cross-links, while S,S DEB produced few intrastrand
lesions. R,R-DEB treated DNA contained mostly 1,3-interstrand bis-N7G-BD, along with smaller amounts
of 1,2-interstrand and 1,2-intrastrand adducts. The effects of DEB stereochemistry on its ability to form
DNA-DNA cross-links may be rationalized by the spatial relationships between the epoxy alcohol side
chains in stereoisomeric N7-(2′-hydroxy-3′,4′-epoxybut-1′-yl)-guanine adducts and their DNA environment.
Different cross-linking specificities of DEB stereoisomers provide a likely structural basis for their distinct
biological activities.
sociated with clinical application of DNA alkylating drugs,6 they
remain at the front lines of cancer therapy, prompting further
investigations of their mechanisms with the goal of developing
novel DNA targeting agents.7-13
S,S-1,2,3,4-Diepoxybutane (S,S-DEB) is a simple bis-epoxide
cross-linking agent believed to be the active form of Ovastat
(L-threitol-1,4-bismethanesulfonate) used clinically against ad-
Introduction
DNA-DNA cross-linking is recognized as the primary
mechanism for the cytotoxic activity of many clinically useful
antitumor drugs, including nitrogen mustards, chloroethylni-
trosoureas, platinum agents, and mitomycin C1. Interchain DNA
adducts prevent DNA strand separation, blocking DNA replica-
tion, transcription, and repair and leading to cancer cell death
and the inhibition of tumor growth.1,2 On the other hand, some
nucleobase monoadducts and intrastrand DNA-DNA cross-
links can be bypassed by DNA polymerases, potentially resulting
in the insertion of incorrect nucleotides opposite the structurally
altered base.1,3-5 Despite their mutagenic side effects and, in
some cases, the development of secondary malignancies as-
(5) Kanuri, M.; Nechev, L. V.; Tamura, P. J.; Harris, C. M.; Harris, T. M.;
Lloyd, R. S. Chem. Res. Toxicol. 2002, 15, 1572-1580.
(6) Sunters, A.; Grimaldi, K. A.; Souhami, R. L.; Hartley, J. A. Nucleic Acids
Res. 1998, 26, 5617-5623.
(7) Sharma, S. K.; McLaughlin, L. W. J. Am. Chem. Soc. 2002, 124, 9658-
9659.
(8) Noronha, A. M.; Wilds, C. J.; Miller, P. S. Biochemistry 2002, 41, 8605-
8612.
(9) Wang, P.; Liu, R.; Wu, X.; Ma, H.; Cao, X.; Zhou, P.; Zhang, J.; Weng,
X.; Zhang, X. L.; Qi, J.; Zhou, X.; Weng, L. J. Am. Chem. Soc. 2003, 125,
1116-1117.
† University of Minnesota.
‡ Rutgers University.
(10) Richter, S. N.; Maggi, S.; Mels, S. C.; Palumbo, M.; Freccero, M. J. Am.
Chem. Soc. 2004, 126, 13973-13979.
(1) Rajski, S. R.; Williams, R. M. Chem. ReV. 1998, 98, 2723-2796.
(2) Skladanowski, A.; Konopa, J. Biochem. Pharmacol. 1994, 47, 2279-2287.
(3) Carmical, J. R.; Zhang, M.; Nechev, L.; Harris, C. M.; Harris, T. M.; Lloyd,
R. S. Chem. Res. Toxicol. 2000, 13, 18-25.
(11) Wilds, C. J.; Noronha, A. M.; Robidoux, S.; Miller, P. S. J. Am. Chem.
Soc. 2004, 126, 9257-9265.
(12) Zhou, Q.; Duan, W.; Simmons, D.; Shayo, Y.; Raymond, M. A.; Dorr, R.
T.; Hurley, L. H. J. Am. Chem. Soc. 2001, 123, 4865-4866.
(13) Mitra, K.; Marquis, J. C.; Hillier, S. M.; Rye, P. T.; Zayas, B.; Lee, A. S.;
Essigmann, J. M.; Croy, R. G. J. Am. Chem Soc. 2002, 124, 1862-1863.
(4) Carmical, J. R.; Kowalczyk, A.; Zou, Y.; Van Houten, B.; Nechev, L. V.;
Harris, C. M.; Harris, T. M.; Lloyd, R. S. J. Biol. Chem. 2000, 275, 19482-
19489.
9
10.1021/ja051979x CCC: $30.25 © 2005 American Chemical Society
J. AM. CHEM. SOC. 2005, 127, 14355-14365
14355

A R T I C L E S
Park et al.
Scheme 1. Structures of DEB Stereoisomers: R,R-DEB (1a),
vanced ovarian cancer.14-16 Ironically, S,S-DEB, along with R,R-
and meso-DEB, is also the suspected carcinogenic metabolite
of 1,3-butadiene (BD), a major industrial chemical used in the
rubber and plastics industry and a recognized human carcino-
gen.17,18 In vivo metabolic activation of BD to DEB is catalyzed
by CYP2E1 and CYP2A6 monooxygenases. The first epoxi-
dation step yields (R)- and (S)-3,4-epoxy-1-butene (EB).19,20 EB
can then be hydrolyzed to 1-butene-3,4-diol or can undergo the
second oxidation to yield R,R-, S,S-, and meso-DEB.21 Although
DEB is a relatively minor metabolite of BD, available experi-
mental evidence suggests that it is responsible for many of the
adverse effects of BD. Biological studies indicate that DEB is
50-100-fold more genotoxic and mutagenic than its monoep-
oxide analogues, EB and 3,4-epoxy-1,2-butanediol,22,23 likely
a result of is ability to induce DNA-DNA and DNA-protein
cross-links. Efficient metabolism of BD to DEB in target tissues
of laboratory mice is thought to cause the extreme susceptibility
of this species to BD carcinogenesis.24
The formation of DNA-DNA cross-links by DEB is a two-
step process. Initial reactions of DEB with guanine nucleobases
in DNA produce N7-(2′-hydroxy-3′,4′-epoxybut-1′-yl)-guanine
(N7-HEBG) monoadducts, which can then be hydrolyzed to N7-
(2′,3′,4′-trihydroxybut-1′-yl)-guanine (N7-THBG) or can alky-
late neighboring nucleobases within the major groove of DNA
to form bifunctional DNA adducts.1,25 Although DNA-DNA
cross-linking is less common than monoadduct formation, it is
considered key for the ability of DEB to induce chromosomal
aberrations, mutagenicity, cytotoxicity, and tumorigenic effects.
Over 40 years ago, Lawley and Brookes first isolated guanine-
guanine conjugates of DEB, which were identified as 1,4-bis-
(guan-7-yl)-2,3-butanediol (bis-N7G-BD).26 This structural as-
signment was recently confirmed in a series of detailed NMR
experiments conducted in our laboratory.27 Millard and col-
laborators used denaturing PAGE to demonstrate that, in analogy
with nitrogen mustards, racemic DEB preferentially induces
interstrand DNA lesions at 5′-GNC sequences (N ) G or C).28,29
The formation of intrastrand cross-links by DEB has been
hypothesized,4 although no direct evidence for the presence of
such lesions in DEB-treated DNA is available in the literature.
Furthermore, the role of DEB stereochemistry in defining the
efficiency and selectivity of DNA-DNA cross-linking has not
been extensively explored.
S,S-DEB (1b), and meso-DEB (1c)
least in part, result from the presence of three distinct optical
isomers of the diepoxide (S,S-, R,R-, and meso-DEB, Scheme
1).
Biological studies reveal significant differences between the
ability of DEB stereoisomers to inactivate T7 coliphage,31 induce
chromosomal aberrations,32 and cause mutagenesis in maize.33
Among the three stereoisomers, S,S-DEB exhibits the most
potent genotoxicity and cytotoxicity, followed by R,R- and then
meso-DEB.31-33 Over 30 years ago, Feit and collaborators
employed density gradient centrifugation to show that meso-
DEB was less effective than S,S- and R,R-DEB in inducing
interstrand DNA lesions.31 This original investigation31 sug-
gested for the first time that the observed differences between
the biological effects of DEB optical isomers may be structural
in their origin.
In the present study, interstrand and intrastrand DNA-DNA
cross-linking by individual DEB stereoisomers was directly
investigated by a combination of mass spectrometry, gel
electrophoresis, and a novel methodology based on stable isotope
labeling of DNA. Our results provide evidence for stereospecific
interactions between DEB stereoisomers and their DNA target,
leading to differences in cross-linking specificity and, ultimately,
in their biological activity.
Experimental Section
Note: DEB is a suspected human carcinogen and must be handled
with adequate safety precautions.
Materials and General Methods. Dimethyl 2,3-O-isopropylidene-
L-tartrate, dimethyl 2,3-O-isopropylidene-D-tartrate, racemic DEB,
meso-erythritol, lithium aluminum hydride, methansulfonyl chloride,
pyridine, and methansulfonic acid were purchased from Sigma-Aldrich
(Milwaukee, WI, St. Louis, MO). CDCl3 was obtained from Cambridge
Isotope Laboratories (Andover, MA). Racemic bis-N7G-BD, N7-
THBG, and 15N5-N7-THBG were prepared as described previously.27
High-resolution mass spectra were obtained on a Bruker BioTOF II.
UV spectra were recorded with a Beckman DU 7400 spectrophotometer.
Proton NMR spectra were obtained on a 600 MHz Varian Inova NMR
spectrometer.
The ability of DEB to display both antitumor and carcinogenic
activity is of significant interest because of the widespread
human exposure to BD in automobile exhaust and in tobacco
smoke (20-75 µg per cigarette).30 This dual activity may, at
Gas chromatography-positive ion chemical ionization mass spec-
trometry (GC-MS) experiments were carried out on an HP5890 series
II gas chromatograph (Agilent Technologies, Palo Alto, CA). The GC
was fitted with a 30 m × 0.25 mm × 0.25 µm film thickness DB-
17MS column (Agilent Technologies) and a 2 m × 530 µm uncoated
deactivated fused silica precolumn. Helium was used as a carrier gas
with a constant flow of 0.7 mL/min. A 1 µL volume of each sample
was injected via splitless injection with a gooseneck splitless liner. The
splitless vent flow was 50 mL/min, the purge open time was 0.5 min,
and the injector temperature was 200 °C. GC oven temperature was
(14) Schmidmaier, R.; Oellerich, M.; Baumgart, J.; Emmerich, B.; Meinhardt,
G. Exp. Hematol. 2004, 32, 76-86.
(15) Hartley, J. A.; O’Hare, C. C.; Baumgart, J. Br. J. Cancer 1999, 79, 264-
266.
(16) Runowicz, C. D.; Fields, A. L.; Goldberg, G. L. Cancer 1995, 76, 2028-
2033.
(17) Melnick, R. L.; Kohn, M. C. Carcinogenesis 1995, 16, 157-163.
(18) Rice, J. M.; Boffetta, P. Chem. Biol. Interact. 2001, 135-136, 11-26.
(19) Malvoisin, E.; Roberfroid, M. Xenobiotica 1982, 12, 137-144.
(20) Himmelstein, M. W.; Turner, M. J.; Asgharian, B.; Bond, J. A. Toxicology
1996, 113, 306-309.
(21) Krause, R. J.; Elfarra, A. A. Arch. Biochem. Biophys. 1997, 337, 176-
184.
(22) Sasiadek, M.; Norppa, H.; Sorsa, M. Mutat. Res. 1991, 261, 117-121.
(23) Cochrane, J. E.; Skopek, T. R. Carcinogenesis 1994, 15, 713-717.
(24) Henderson, R. F.; Thornton-Manning, J. R.; Bechtold, W. E.; Dahl, A. R.
Toxicology 1996, 113, 17-22.
(25) Lawley, P. D.; Brookes, P. Nature 1965, 206, 480-483.
(26) Lawley, P. D.; Brookes, P. J. Mol. Biol. 1967, 25, 143-160.
(27) Park, S.; Tretyakova, N. Chem. Res. Toxicol. 2004, 17, 129-136.
(28) Millard, J. T.; White, M. M. Biochemistry 1993, 32, 2120-2124.
(29) Sawyer, G. A.; Frederick, E. D.; Millard, J. T. Chem. Res. Toxicol. 2004,
17, 1057-1063.
(30) Hecht, S. S. J. Natl. Cancer Inst. 1999, 91, 1194-1210.
(31) Verly, W. G.; Brakier, L.; Feit, P. W. Biochim. Biophys. Acta 1971, 228,
400-406.
(32) Matagne, R. Mutat. Res. 1969, 7, 241-247.
(33) Bianchi, A.; Contin, M. J. Heredity 1962, 53, 277-281.
9
14356 J. AM. CHEM. SOC. VOL. 127, NO. 41, 2005

DNA−DNA Cross-Linking by 1,2,3,4-Diepoxybutane
A R T I C L E S
held at 40 °C for 2 min and then ramped at 3 °C/min until 80 °C. The
CI source parameters were as follows: filament emission current, 178
µA; electron energy, 193 eV; ion source temperature, 250 °C. A mixture
of NH3 (4%) and CH4 (96%) was used as the reagent gas. For all
samples analyzed, m/z 104 [M + NH4]+ was monitored. The peak width
was 0.8 amu, and the scan time was 0.2 s.
S,S- and R,R-1,2,3,4-Diepoxybutane. Optically active S,S- and R,R-
DEB stereoisomers were prepared as described in the literature34-36
starting with dimethyl 2,3-O-isopropylidene-L-tartrate and dimethyl 2,3-
O-isopropylidene-D-tartrate, respectively.
S,S-1,2,3,4-Diepoxybutane: clear oil, bp 138-140 °C; 1H NMR
(CDCl3) δ (ppm) 2.70-2.72 (2H, m), 2.80-2.82 (2H, m), 2.86-2.88
(2H, m); 13C (CDCl3) δ (ppm) 51.4, 43.6.
R,R-1,2,3,4-Diepoxybutane: clear oil, bp 138-140 °C;1H NMR-
(CDCl3) δ (ppm) 2.70-2.72 (2H, m), 2.79-2.81 (2H, m), 2.86-2.88
(2H, m); 13C (CDCl3) δ (ppm) 51.5, 43.6.
30 membrane filters (Millipore - Bedford, MA), and the filtrates were
directly analyzed by HPLC-ESI+-MS/MS as described below.
Detection of Interstrand DEB Cross-Links by Denaturing PAGE.
Synthetic DNA 28-mers (5′-TAT ATA TTT ATA GGC TAT TAT
TAT ATT A) (+ strand), (5′-TAA TAT AAT AAT AGC CTA TAA
ATA TAT A) (- strand) were prepared by standard phosphoramidite
chemistry. Both strands were purified by 20% denaturing PAGE.38 The
(+) strand (200 pmol) was 5′-endlabeled with 32P in the presence of
[γ-32P]ATP/T4 polynucleotide kinase38 and then spiked with the
corresponding unlabeled DNA (20 nmol). The DNA was annealed to
the complementary (-) strand (22 nmol), and the resulting double
stranded DNA was dissolved in 0.3 mM NaOAc, pH 5.0, and treated
with 0, 10, 50, or 100 mM of R,R-, S,S-, racemic, or meso-DEB for 3
h at 37 °C. The DNA solution was dried under vacuum, and the residue
was dissolved in a 1:1 mixture of water and formamide (5 µL). The
solution was loaded onto 20% denaturing PAGE (0.4 mm, 41 cm ×
37 cm), run at 60 W and at ambient temperature. Radiolabeled DNA
bands were visualized with a Storm 840 phosphorimager. The cross-
linked products were detected as low-mobility bands on the gel, and
their identity was confirmed by capillary HPLC-ESI--MS/MS analysis
of the material eluted from the gel (M ) 17 254 (calculated), M )
17 254 (observed)). The bands were quantified by volume analysis,
and cross-linking efficiency was quantified from the intensity ratio of
the cross-linked band versus the band corresponding to single-stranded
DNA.
Capillary HPLC-ESI--MS/MS analysis of cross-linked DNA was
performed on an 1100 Agilent Technologies capillary HPLC-MSD
ion trap system operated in the negative ion mode. A Zorbax Extend-
C18 column (0.5 mm × 150 mm, 3.5 µm, Agilent Technologies) was
eluted at a flow rate of 12 µL/min, with a linear gradient of 15 mM
ammonium acetate (A) and acetonitrile (B). The mobile phase composi-
tion was kept at 1% B for 5 min, then increased to 25% B for 15 min,
and further to 25% B at 30 min. The mass range was m/z 500-1900,
and the target ion abundance was 30 000. ESI was achieved with a
spray voltage of +2.6 kV and a source temperature of 200 °C. The
nebulizing gas (N2) pressure was set to 15 psi, and the drying gas (N2)
flow rate was set to 5 L/min.
Stable Isotope Labeling to Quantify Intrastrand and Interstrand
Bis-N7G-BD Cross-Links in 5′-GGC Context. The DNA 28-mer (5′-
TAT ATA TTT ATA GGC TAT TAT TAT ATT A) (+ strand) and
the complementary strand (5′-TAA TAT AAT AAT A[1,7,NH2-15N3-
2-13C-G]C CTA TAA ATA TAT A) were synthesized by standard
phosphoramidite methodology. 1,7,NH2-15N3-2-13C-dG phosphoramidite
was prepared at Rutgers University as previously described.39 Both
oligodeoxynucleotides were purified by HPLC with an Agilent
Technologies HPLC system (model 1100) incorporating a UV diode
array detector and a semi-micro UV cell. A Supelcosil LC-18-DB
column (4.6 × 150 mm, 5 µm) was eluted at a flow rate of 1 mL/min
with a gradient of 150 mM ammonium acetate (A) and acetonitrile
(B). Solvent composition was changed from 5 to 15% B in 40 min.
HPLC fractions corresponding to full-length oligodeoxynucleotide
products were collected, combined, and dried under vacuum. DNA
purity and identity were determined by HPLC-ESI- MS (+ strand:
calcd M ) 8580.7, obsd M ) 8581.2; - strand: calcd M ) 8588.7,
obsd M ) 8589.5).
meso-1,2,3,4-Diepoxybutane. meso-1,2,3,4-Diepoxybutane was pre-
pared from meso-erythritol according to the published procedure.37 Clear
1
oil, bp 138-140 °C; H NMR (CDCl3) δ (ppm) 2.65-2.72 (2H, m),
2.75-2.82 (2H, m), 2.91-2.95 (2H, m). 13C (CDCl3) δ (ppm) 50.3,
44.2. The diastereomeric purity as determined by NMR and GC-MS
signal integration was ∼95%.
meso-1,4-Bis-(guan-7-yl)-2,3-butanediol. Guanosine (1.00 g, 3.53
mmol) was suspended in 20 mL of glacial acetic acid and stirred at 80
°C for 1 h. meso-DEB (300 µL, 3.88 mmol) was added to initiate the
reaction, and the mixture was stirred at 80 °C for 4 h. Once cooled to
room temperature, 100 mL of acetone/diethyl ether (1:4) were slowly
added over the course of 30 min under vigorous stirring. The resulting
light yellow solid containing the bifunctional lesion was filtered, rinsed
with diethyl ether, and dried under nitrogen. To remove the ribosyl
groups, the precipitate was resuspended in 25 mL of 1 N HCl solution
and stirred at 85 °C for 1 h. The mixture was cooled to room
temperature, and the solvent was evaporated, affording a white solid.
meso-Bis-N7G-BD was recrystallized from 1 N NaOH to give 37.8
mg of the product (5.5% yield). Chemical and diastereoisomeric purity
of the product (>95%) was established by HPLC, NMR spectroscopy,
and mass spectrometry. Standard solution concentrations were deter-
mined by UV spectrophotometry (ꢀ252 ) 15,700 at pH 1).27
Racemic and meso-15N10-1,4-Bis-(guan-7-yl)-2,3-butanediol. 15N5-
dGTP (3.3 mg, from Martek Biosciences, Columbia, MD) was dissolved
in 100 µL of glacial acetic acid. A 0.5 µL aliquot of either racemic or
meso-DEB was added, followed by incubation at 80 °C for 2 h. The
reaction mixture was cooled to room temperature, and ethyl acetate
was added to cause precipitate formation. The resulting solid was
removed, washed with ethyl acetate, and redissolved in 300 µL of 1 N
HCl. Hydrolysis was performed at 80 °C for 1 h to achieve depurination.
The solution was dried under reduced pressure, and the residue was
dissolved in 1 mL of water and purified by SPE using Waters SepPak
C18 cartridges. Standard solution concentrations were established by
HPLC-ESI+-MS/MS and HPLC-UV peak integration in comparison
with unlabeled bis-N7G-BD diastereomers.
DEB Treatment of Calf Thymus DNA and Cross-Link Analyses.
Calf thymus DNA (CT DNA, 1 mg/mL solution in 10 mM Tris-HCl
buffer, pH 7.2, in triplicate) was treated with R,R-, S,S-, or meso-DEB
(0.01-0.5 mM) at 37 °C for 24 h. The unreacted DEB was extracted
with diethyl ether (2 × 400 µL). Samples were spiked with known
amounts of racemic and meso-15N10-bis-N7G-BD (15-20 pmol) and
subjected to either neutral thermal or acid hydrolysis. Neutral thermal
hydrolysis was performed at 80 °C for 1 h, while acid hydrolysis was
achieved by heating in 0.1 N HCl (80 °C for 1 h). Partially depurinated
DNA backbone was removed by ultrafiltration through Microcon YM-
To obtain double-stranded DNA, (+) strand (20 nmol) and (-) strand
(22 nmol) were combined, dried under vacuum, and dissolved in 180
µL of 0.3 mM sodium acetate buffer, pH 5.0 containing 0.1 mM Tris,
and 0.5 mM NaCl. 10% excess of the (-) strand was used to ensure
that the (+) strand remained double-stranded throughout the duration
of the experiment (calculated Tm ) 43 °C). The solution was heated to
90 °C and allowed to slowly cool to room temperature to form double-
(34) Feit, P. W. J. Med. Chem. 1964, 7, 14-17.
(35) Mash, E. A.; Nelson, K. A.; Van Deusen, S.; Hemperly, S. B. Org. Synth.
1990, 68, 92-103.
(38) Sambrook, J.; Fritsch, E. F.; Maniatis, T. Molecular Cloning: A Laboratory
Manual; Cold Spring Harbor Press: 1989.
(36) Robbins, M. A.; Devine, P. N.; Oh, T. Org. Synth. 1999, 76, 101-109.
(37) Claffey, D. J. Synth. Commun. 2002, 32, 3041-3045.
(39) Shallop, A. J.; Gaffney, B. L.; Jones, R. A. J. Org. Chem. 2003, 68, 8657-
8661.
9
J. AM. CHEM. SOC. VOL. 127, NO. 41, 2005 14357

A R T I C L E S
Park et al.
stranded DNA. The DNA (1.5 nmol, 15 µL aliquots) was treated with
R,R-, S,S-, racemic, or meso-DEB (10 equiv) for 3 h at 25 °C and
precipitated with cold ethanol. The reaction mixtures were resuspended
in water, spiked with 1.5 pmol of racemic and meso-15N10-1,4-bis-
N7G-BD (internal standard for HPLC-ESI+-MS/MS), and subjected
to thermal or acid hydrolysis as described above. The hydrolysates were
directly analyzed by HPLC-ESI+-MS/MS. The relative contribu-
tion of intrastrand and interstrand cross-linking was determined from
the HPLC-MS/MS peak areas corresponding to bis-N7G-BD (m/z
389.2 [M + H] +f 152.1 [Gua + H] +) and [15N3, 13C1]-bis-N7G-BD
(393.2 [M + H]+f 156.0 [15N3,13C1-Gua + H]+, 152.1 [Gua + H] +),
respectively.
quadrupole mass spectrometer was used in all analyses. Chromato-
graphic separation was achieved with a Zorbax SB-C18 column (150
mm × 0.5 mm, 5 µm) eluted with 15 mM ammonium acetate, pH 5.5
(A) and acetonitrile (B). A linear gradient was from 0 to 10% B in 20
min at a flow rate of 10 µL/min, and the injection volume was 8 µL.
To minimize salt contamination, HPLC effluent from the first 3 min
of each run was diverted to waste. The mass spectrometer was operated
in the positive ion mode with nitrogen as a sheath gas (5 L/min).
Electrospray ionization was achieved at a spray voltage of 4.0 kV, and
the capillary temperature was 250 °C. Bis-N7G-BD cross-links were
detected in selected reaction monitoring mode as described below.
Bis-N7G-BD lesions formed in DEB-treated calf thymus DNA were
quantified by isotope dilution with racemic and meso-15N10-bis-N7G-
BD internal standards. Quantitative analyses were performed using
HPLC-ESI+ MS/MS peak areas corresponding to bis-N7G-BD (m/z
389.2 [M + H]+f 238.0 [M + H-Gua]+, 389.2 f 220.0 [M + H-Gua
Stable Isotope Labeling To Determine the Kinetics of Spontane-
ous Hydrolysis of Intrastrand and Interstrand Bis-N7G-BD in
Double-Stranded DNA. The DNA 28-mer (5′-TAT ATA TTT ATA
GGC TAT TAT TAT ATT A) (+ strand) and the complementary strand
(5′-TAA TAT AAT AAT A[1,7,NH2-15N3-2-13C-G]C CTA TAA ATA
TAT A) were prepared and annealed as described above. The 15N3-2-
13C-labeled duplex (15 nmol) was dissolved in 240 µL of 0.3 mM
NaOAc, pH 5.0 and treated with R,R-DEB (10 µL) for 3 h at 37 °C to
generate both interstrand and intrastrand bis-N7G-BD adducts. The
DNA was precipitated with cold ethanol, dried under nitrogen, and
redissolved in 125 µL of 10 mM Tris buffer containing 0.2 N NaCl.
Following the addition of 15N10-1,4-bis-N7G-BD internal standard (25
pmol), the solution was maintained at 37 °C. Aliquots (20 µL) were
removed following incubation for 0, 1, 3, 6, 9, 16, and 20 h and
immediately frozen at -20 °C. The hydrolysates were directly analyzed
by HPLC-ESI+-MS/MS as described below. The autosampler tray was
maintained at 4 °C to minimize further depurination. The amounts of
hydrolytically released intrastrand and interstrand bis-N7G-BD cross-
links were determined from HPLC-ESI+-MS/MS peak areas corre-
sponding to bis-N7G-BD (m/z 389.2f 238.0) and [15N3]-bis-N7G-BD
15
- H2O]+). Internal standards meso- and racemic N10-bis-N7G-BD
were analyzed analogously using the transitions m/z 399.2 [M + H]+f
243.0 [M + H-15N5-Gua]+, 399.2 f 225.0 [M + H-15N5-Gua - H2O]+.
Quantitation of N7-THBG was achieved using the reconstructed ion
chromatograms corresponding to the MS/MS transition: m/z ) 256.0
[M + H]+f 152.0 [M + H-THB]+ and the corresponding transition
m/z ) 261.0 [M + H]+f 157.0 [M + H-THB]+ of the 15N5-THBG
internal standard.40-42 Calibration curves were constructed by analyzing
standard solutions containing known amounts of each adduct and its
internal standard, followed by regression analysis of the HPLC-ESI+-
MS/MS peak area ratios.
Intrastrand and interstrand bis-N7G-BD cross-links originating from
15N3,13C1-dG containing 28-mer (5′-TAT ATA TTT ATA GGC TAT
TAT TAT ATT A) (+ strand), (5′-TAA TAT AAT AAT A[1,7,NH2-
15N3-2-13C-G]C CTA TAA ATA TAT A) (- strand) were quantified
using the transitions m/z ) 389.2 [M + H]+f 152.0 [Gua + H]+
(intrastrand) and 393.2 [M + H]+f 156.0 [15N313C1-Gua + H]+, 152.0
[Gua + H]+ (interstrand). The molar ratios between interstrand and
intrastrand bis-N7G-BD lesions were determined directly from the
corresponding HPLC-ESI+-MS/MS peak areas.
1,2-/1,3-Interstrand bis-N7G-BD cross-links originating from [1,7,-
NH2-15N3-2-13C]- dG containing 28-mer (5′-TAT ATA TTT ATA [1,7,-
NH2-15N3-2-13C-G] GC TAT TAT TAT ATT A) (+ strand), 5′-TAA
TAT AAT AAT AG CTA TAA ATA TAT A) (- strand) were
quantified following thermal or acid hydrolysis of interstrand DNA-
DNA cross-links isolated by denaturing PAGE. HPLC-ESI+-MS/MS
settings were as described in the preceding section. MS/MS transitions
m/z 389.2 [M + H]+f 152.0 [Gua + H]+ and m/z 393.2 [M + H]+f
155.0 [15N313C1-Gua + H]+, 152.0 [Gua + H]+ were used to quantify
1,3- and 1,2-interstrand bis-N7G-BD lesions, respectively.
15
(m/z 393.2 f 238.0; 242.0), respectively. N10-Bis-N7G-BD internal
standard was detected analogously by monitoring the transition m/z
399.2 f 243.0. To determine the total number of bis-N7G-BD adducts
in this DNA, several aliquots were analyzed following quantitative
release of the cross-links by thermal hydrolysis (70 °C for 1 h). First-
order constants (k) were calculated from the slope of the plot of ln[c/
c0] ) -kt, where t is the incubation time, c is the concentration of
bis-N7G-BD remaining in DNA at the time t, and c0 is the starting
concentration of bis-N7G-BD in DNA.
Stable Isotope Labeling Experiments to Distinguish between 1,2
(GC) and 1,3 (GNC) Interstrand Cross-Linking by DEB. DNA 28-
mer (5′-TAT ATA TTT ATA [1,7,NH2-15N3-2-13C-G] GC TAT TAT
TAT ATT A) (+ strand) and its unlabeled complement were prepared
as described above. Double-stranded DNA (7 nmol) was treated with
S,S-, R,R-, racemic, or meso-DEB (50 mM) in 0.3 mM sodium acetate
buffer (pH 5, total volume ) 72 µL). The reaction mixture was
incubated at 37 °C for 3 h, dried under vacuum, and separated by 20%
denaturing PAGE (2,000 V, for 2.5 h). The position of the slowly
migrating band containing [1,7,NH2-15N3-2-13C]-labeled interstrand
DNA cross-links was estimated based on the analogous band from the
unlabeled DNA reaction (20 nmol duplex, treated with a 100× molar
excess of S,S-DEB), which was visualized by UV shadowing. The DNA
bands containing interstrand DEB cross-links were excised, and the
DNA was extracted by standard “crash and soak” methods (10 mM
Tris-Cl, 1 mM EDTA, pH 7.6; overnight).38 The DNA was desalted
by SPE on Waters C18 SepPak cartridges eluted with methanol/water,
dried, and dissolved in 15 µL of water. DEB-induced N7-guanine cross-
links were released by thermal or acid hydrolysis as described above.
The molar ratio of 1,3-interstrand and 1,2-interstrand G-G DEB cross-
links was established from the HPLC-ESI+-MS/MS peak area ratio
corresponding to [1,7,NH2-15N3-2-13C]-labeled and unlabeled bis-N7G-
BD, respectively.
Results
DEB stereoisomers induce similar numbers of bis-N7G-
BD adducts. Bifunctional alkylation of DNA by DEB gives
rise to 1,4-bis-(guan-7-yl)-2,3-butanediol (bis-N7G-BD) adducts
(2-4 in Scheme 2).27 Because N-7 alkylation of guanine in
DNA generates a quaternized nitrogen, destabilizing the gly-
cosidic bond, these lesions are hydrolytically labile and can be
selectively released from the DNA backbone as free bases upon
heating at neutral pH. We have previously analyzed the
formation of racemic bis-N7G-BD lesions in D,L-DEB-treated
DNA by HPLC-ESI+-MS/MS analysis of thermal DNA
hydrolysates.27 Bis-N7G-BD and the corresponding N7-(2′,3′,4′-
(40) Tretyakova, N. Y.; Chiang, S. Y.; Walker, V. E.; Swenberg, J. A. J Mass
Spectrom. 1998, 33, 363-376.
(41) Koc, H.; Tretyakova, N. Y.; Walker, V. E.; Henderson, R. F.; Swenberg,
J. A. Chem. Res. Toxicol. 1999, 12, 566-574.
Capillary HPLC-ESI+-MS/MS. An Agilent 1100 capillary HPLC
system (Wilmington, DE) interfaced to a Finnigan TSQ Quantum triple
(42) Oe, T.; Kambouris, S. J.; Walker, V. E.; Meng, Q.; Recio, L.; Wherli, S.;
Chaudhary, A. K.; Blair, I. A. Chem. Res. Toxicol. 1999, 12, 247-257.
9
14358 J. AM. CHEM. SOC. VOL. 127, NO. 41, 2005

DNA−DNA Cross-Linking by 1,2,3,4-Diepoxybutane
A R T I C L E S
Scheme 2. Structures of Stereoisomeric 1,4-bis-(guan-7-yl)-2,3-butanediol (bis-N7G-BD) Adducts
trihydroxybut-1′-yl)guanine (N7-THBG) monoadducts were
formed in a concentration-dependent fashion, with similar
amounts of N7-N7-G cross-links and monoadducts produced
at low exposure concentrations (0.05-0.1 mM DEB) and N7-
THBG predominating at higher DEB exposures.27
DEB to induce interstrand DNA-DNA cross-links. 32P end-
labeled DNA duplexes containing a complementary 5′-
GGC-3′/3′-CCG-5′ trinucleotide flanked by AT-rich regions
(5′-TAT ATA TTT ATA GGC TAT TAT TAT ATT A)
(+ strand), (5′-TAA TAT AAT AAT AGC CTA TAA ATA
TAT A) (- strand) were incubated with individual DEB
stereoisomers. 5′-GGC-3′ sequences have been previously
identified as the target sites for DEB-induced interstrand DNA
cross-linking.28,29 Denaturing PAGE of DEB-treated duplexes
has revealed the formation of interchain linkages as indicated
by the appearance of a new DNA band with approximately half
the mobility of the corresponding single strands (Figure 2).
ESI--MS analysis of this material following extraction from
the gel confirmed that it represented cross-linked duplexes
containing a single butanediol linkage (calculated M ) 17 254;
observed M ) 17 254, Figure 3). For each DEB stereoisomer,
the cross-linked band increased in intensity as the DEB
concentration was raised from 10 to 50 mM and further to 100
mM (Figure 2). Furthermore, interstrand cross-linking efficiency
In the present work, the formation of diastereomeric bis-N7G-
BD lesions 2-4 in double-stranded DNA treated with individual
DEB stereoisomers was investigated in order to compare their
ability to induce bifunctional DNA adducts. G-G DEB
conjugates were quantified as free bases following their release
from the DNA backbone,27 thus providing the total amounts of
guanine-guanine butanediol cross-links generated by S,S-, R,R-,
and meso-DEB, irrespective of their type (interstrand or intras-
trand). Low DEB concentrations (0.01-0.5 mM) were employed
to maximize cross-linking selectivity and to mimic physiological
exposures. Accurate and specific quantification of stereoisomeric
bis-N7G-BD adducts was achieved by isotope dilution capillary
HPLC-ESI+-MS/MS using synthetic racemic and meso-15N10-
bis-N7G-BD internal standards.27 While the two optically active
bis-N7G-BD lesions (2 and 3 in Scheme 2) are enantiomeric
and thus have identical HPLC retention times, the meso cross-
link (4 in Scheme 2) elutes slightly later from the HPLC column
(see Supporting Information).
Bis-N7G-BD amounts increased in a dose-dependent manner
in double-stranded DNA (calf thymus DNA) treated with 0.01-
0.5 mM S,S-, R,R-, or meso-DEB (Figure 1). Adduct quantities
ranged between 0.001 and 0.3 per 103 normal guanines,
depending on exposure concentration (Figure 1). However, the
number of cross-linked lesions was only moderately affected
by diepoxide stereochemistry (Figure 1). Similarly, the amounts
of N7-THBG monoadducts in these DNA samples were
comparable, regardless of the DEB isomer used (results not
shown). Taken together, these results rule out stereospecific
differences between the total reactivity of DEB isomers toward
double-stranded DNA.
Figure 1. Formation of 1,4-bis-(guan-7-yl)-2,3-butanediol lesions in calf
thymus DNA treated with individual stereoisomers of DEB (0.01-0.5 mM).
Bis-N7G-BD adduct amounts (intrastrand + interstrand) were determined
by capillary HPLC-ESI+-MS/MS analysis of DNA hydrolysates.
S,S-DEB is the most efficient interstrand DNA-DNA cross-
linker. We next examined the ability of S,S-, R,R-, and meso-
9
J. AM. CHEM. SOC. VOL. 127, NO. 41, 2005 14359

A R T I C L E S
Park et al.
(Figure 4). In contrast, intrastrand G-G DEB lesions originate
exclusively from the unlabeled (+) strand and thus will not
contain the 15N3,13C1 label (Figure 4). HPLC-ESI+-MS/MS
analysis of DNA hydrolysates can readily distinguish between
[15N3,13C1]-labeled and unlabeled bis-N7G-BD conjugates (m/z
393.2 and 389.2 (M + H)+, respectively) and thus can determine
the relative extent of intrastrand and interstrand cross-linking
by DEB. A 10% molar excess of the (-) strand was employed
to ensure that the (+) strand remained hybridized to its
complement throughout the duration of the experiment.
The [15N3,13C1]-labeled duplex was treated with S,S-, R,R-,
racemic, or meso-DEB (10 molar equiv), followed by HPLC-
ESI+-MS/MS analyses of bis-N7G-BD and [15N3,13C1]-bis-N7G-
BD in DNA hydrolysates. Extracted ion chromatograms cor-
responding to unlabeled G-G DEB cross-links (Figure 5A)
contain a prominent peak with HPLC retention time and MS/
MS fragmentation pattern identical to those of synthetic bis-
N7G-BD (m/z 389.2 f 238.0 (M + H-Gua)+, 220 (M + H-Gua
- H2O)+, and 152 (Gua + H)+, see Figure 5A, insert).27 The
corresponding [15N3,13C1]-labeled lesions are observed at m/z
393.2 (Figure 5B), exhibiting two series of ions consistent with
the presence of both guanine and 15N3,13C1-guanine in the
molecule: m/z 393.2 [M + H]+f m/z 242.0 (M + H-Gua)+,
238.0 (M + H-[15N3, 13C1]-Gua)+, 224.0 (M + H-Gua - H2O)+,
220.0 (M + H-[15N3,13C1]Gua - H2O)+, 156.0 ([15N3, 13C1]-
Gua + H)+, and 152.0 (Gua + H)+ (Figure 5B, inset).
The formation of unlabeled bis-N7G-BD (M ) 388.2, m/z
389.2 [M + H]+) following meso-DEB treatment of the 5′-GGC-
3′/3′-CC[15N3,13C1]G-5′ duplex (Figure 5A) provides the first
direct evidence for the formation of intrastrand DNA lesions
by DEB. Similar results were obtained when bis-N7G-BD
adducts were released by acid hydrolysis (0.1 N HCl, 1 h at 70
°C), ruling out the possibility of artifactual bis-N7G-BD
formation by reactions of N7-(2′-hydroxy-3′,4′-epoxy-but-1′-
yl)guanine free bases released during thermal hydrolysis of
DEB-treated DNA.46 In contrast, our attempts to quantify
interstrand and intrastrand bis-N7G-BD lesions following their
spontaneous hydrolysis at physiological temperature (37 °C)
failed to provide reproducible results because of the significant
differences in their hydrolytic stability (see below).
Importantly, the relative contribution of intrastrand G-G
cross-linking to the total number of bis-N7G-BD adducts was
strongly dependent on diepoxide stereochemistry, accounting
for 4% of total bis-N7G-BD lesions for S,S-DEB, 19% for R,R-
DEB, and 51% for meso-DEB (Table 1). The presence of
significant numbers of intrastrand G-G lesions in meso-DEB-
treated DNA (Table 1) is consistent with the inefficient
interstrand adduct formation by this stereoisomer (Figure 2).
In contrast, S,S-DEB specifically forms interchain G-G cross-
links but induces few intrastrand bis-N7G lesions (Figure 2,
Table 1). R,R-DEB appears to have intermediate cross-linking
specificity (Table 1). We propose that the three DEB isomers
give rise to similar numbers of N7-(2′-hydroxy-3′,4′-epoxybut-
1′-yl)guanine monoadducts, which have different fates in DNA,
depending on their stereochemistry (Scheme 3). The epoxy ring
of S,S monoadducts is subject to an SN2-type nucleophilic attack
Figure 2. 20% Denaturing PAGE of synthetic DNA duplexes (5′-TAT
ATA TTT ATA GGC TAT TAT TAT ATT A, (+) strand) following their
incubation with R,R- (lanes 2-4), S,S- (lanes 5-6), racemic (lanes 8-10),
and meso-DEB (lanes 11-13). DEB concentrations were as follows: 0 mM
(lane 1, control DNA); 10 mM (lanes 2, 5, 8, and 11); 50 mM (lanes 3, 6,
9, and 12); and 100 mM (lanes 4, 7, 10, 13). Interstrand cross-links (x-
links) are observed as slowly moving bands on the gel.
differed significantly for S,S, R,R, and meso isomers of DEB
(Figure 2). The number of interstrand cross-links was the highest
for S,S-DEB, followed by racemic, R,R-, and meso-diepoxide
(Figure 2). Quantitative densitometry analysis indicated that,
at the highest exposure level (100 mM), S,S-DEB induced 8-fold
and 10-fold more interstrand lesions than did R,R- and meso-
diepoxides, respectively. These results demonstrate that S,S-
DEB is a more effective interstrand DNA-DNA cross-linker
than R,R- and meso-DEB, consistent with its potent cytotoxic-
ity.31,32
meso-DEB produces 1,2-intrastrand N7G-N7G cross-
links. Our observation of similar total numbers of bis-N7G-
BD cross-links in DNA treated with different DEB stereoisomers
(Figure 1), but a greater ability of S,S-DEB to induce interstrand
DNA cross-links (Figure 2), can potentially be explained by
the formation of intrastrand bis-N7G-BD lesions by meso- and
R,R-DEB, but not S,S-DEB. Unfortunately, such intrastrand
DNA cross-linking cannot be analyzed by standard methods.
Denaturing PAGE43 cannot directly detect intrastrand lesions
because their gel mobility is similar to that of the corresponding
single strands. Mass spectrometry analysis of nuclease di-
gests44,45 is not readily applicable to N7-guanine lesions because
of their hydrolytic lability, which leads to inadvertent depuri-
nation during enzymatic digestion.
We developed a novel strategy based on stable isotope
labeling-mass spectrometry methodology (Figure 4) to sepa-
rately quantify intrastrand and interstrand bis-N7G-BD cross-
links in DEB-treated DNA. A DNA duplex containing a single
5′-GGC-3′ trinucleotide flanked by two AT-rich sequences was
prepared. Only native bases were incorporated into the (+)
strand, while the (-) strand contained a single [15N3,13C1]-
guanine opposite GGC (Figure 4):
Since the only guanine present in the (-) strand has the
15N3,13C1 isotope tag, any interstrand G-G conjugates origi-
nating from this duplex must possess the 15N3,13C1 label, with
a corresponding (+4) mass shift detectable by mass spectrometry
(44) Zeng, Y.; Wang, Y. J. Am. Chem. Soc. 2004, 126, 6552-6553.
(45) Wang, Y.; Taylor, J. S.; Gross, M. L. Chem. Res. Toxicol. 1999, 12, 1077-
1082.
(46) Tretyakova, N. Y.; Sangaiah, R.; Yen, T. Y.; Swenberg, J. A. Chem. Res.
Toxicol. 1997, 10, 779-785.
(43) Millard, J. T.; Weidner, M. F.; Kirchner, J. J.; Ribeiro, S.; Hopkins, P. B.
Nucleic Acids Res. 1991, 19, 1885-1891.
9
14360 J. AM. CHEM. SOC. VOL. 127, NO. 41, 2005

DNA−DNA Cross-Linking by 1,2,3,4-Diepoxybutane
A R T I C L E S
Figure 3. ESI- mass spectrum of interstrand cross-linked DNA 28-mer (5′-TAT ATA TTT ATA GGC TAT TAT TAT ATT A) (+ strand) extracted from
the slowly moving band on denaturing PAGE (Figure 2). The observed molecular weight (M ) 17 254) is consistent with the presence of a single bis-
N7G-BD cross-link (calculated, M ) 17 254).
Figure 4. Stable isotope labeling HPLC-ESI+-MS/MS approach to distinguish between intrastrand and interstrand bis-N7G-BD cross-link formation by
DEB in 5′-GGC-3′ sequences.
by the N7-guanine in the complementary DNA strand, leading
to the preferential formation of interstrand bis-N7G cross-links
(Scheme 3). meso-DEB derived intermediates similarly partici-
pate in nucleophilic substitution by the neighboring N7-guanine
in the same DNA strand, giving rise to equal numbers of
intrastrand and interstrand bis-N7G-BD lesions (Scheme 3).
and requires considerable conformational flexibility of DNA
to accommodate a butanediol linker.
Because gel electrophoresis based studies reported in the
literature28,29 employed very high DEB concentrations (300-
1000 mM) and did not examine actual DNA adduct structures,
we decided to re-examine sequence preferences for interstrand
cross-linking by DEB using a novel, stable isotope labeling
based approach. A DNA 28-mer was prepared in which the 5′
guanine in the (+) strand was labeled with 15N3,13C1:
DEB-induced interstrand lesions involve distal guanines
within 5′-GGC context. In their original publication, Brookes
and Lawley proposed that DEB cross-links the opposite DNA
strands within 5′GC dinucleotides.25,47 More recent studies
utilizing gel electrophoresis contradict this original hypothesis,
suggesting that, in analogy with nitrogen mustards, interstrand
cross-linking by DEB involves the distal guanines within 5′GNC
context.28 This sequence preference is perhaps surprising given
the fact that a four-carbon chain of DEB (4-5 Å) must span
the distance between the N7-guanine atoms in 5′GNC (8.9 Å)
1,3-/1,2-Interstrand cross-linking in this duplex can be
distinguished by the molecular weights of the resulting G-G
conjugates: 1,3-bis-N7G adducts, but not 1,2-bis-N7G adducts,
9
J. AM. CHEM. SOC. VOL. 127, NO. 41, 2005 14361

A R T I C L E S
Park et al.
Figure 5. HPLC-ESI+-MS/MS analysis of 1,4-bis-(guan-7-yl)-2,3-butanediol (bis-N7G-BD) adducts in DNA hydrolysates derived from meso-DEB-treated
duplex (5′-TAT ATA TTT ATA GGC TAT TAT TAT ATT A) (+ strand), (5′-TAA TAT AAT AAT A[1,7,NH2-15N3-2-13C-G]C CTA TAA ATA TAT A)
(- strand). Extracted ion chromatograms corresponding to intrastrand (unlabeled) and interstrand (15N3,13C1-labeled) bis-N7G-BD cross-links were obtained
by monitoring the transitions m/z 389.2 [M + H]+ f 152.0 [Gua + H]+ and m/z 393.2 [M + H]+f 156.0 [15N313C1-Gua + H]+, 152.0 [Gua + H]+,
respectively. Inset: MS/MS spectra corresponding to unlabeled (A) and 15N3,13C1-labeled bis-N7G-BD (B).
Table 1. Quantitative Analysis of 1,3-Interstrand, 1,2-Interstrand, and 1,2-Intrastrand 1,4-Bis-(guan-7-yl)-2,3-butanediol Cross-Links
Generated in 5′-GGC-3′/3′-CCG-5′ Sequence Context
a Average ( SD of three measurements. It should be noted that although the total yields of bis-N7G-BD adducts in the 5′GGC sequence differ depending
on DEB stereochemistry (Table 1), they are similar in calf thymus DNA (Figure 1). This can be explained by the presence of other potential cross-linking
sites (in addition to 5′-GGC) in genomic DNA.
should bear the 15N3,13C1 isotope tag. Following incubation with
individual DEB stereoisomers, interstrand cross-linked duplexes
were isolated by denaturing PAGE. Bis-N7G-BD lesions were
quantitatively released by acid hydrolysis, and the molar ratio
between 1,3- ([15N3, 13C1]-labeled) and 1,2- (unlabeled) inter-
strand adducts was established from their HPLC-ESI+-MS/
MS peak areas (m/z 393.2 and 389.2, respectively). We found
that, for all three DEB stereoisomers, the majority of interstrand
bis-N7G lesions derived from the alkylation of distal guanines
on opposite DNA strands of the 5′-GGC sequence (Table 1).
Only R,R-DEB was capable of inducing 1,2-interstrand lesions
(13% of the total number of bis-N7G-BD adducts, Table 1).
1,3-Interstrand cross-linking specificity exhibited by DEB in
5′-[15N3,13C1-G] GC trinucleotides originates from the cross-
linking step, rather than from the sequence-dependent formation
of monoalkylated intermediates. Indeed, HPLC-MS/MS analy-
sis of N7-THBG lesions in the same samples provides no
evidence for the preferential formation of 15N3,13C1-labeled N7-
THBG (results not shown). We conclude that N7-(2′-hydroxy-
3′,4′-epoxybut-1′-yl)-guanine (N7-HEBG) monoadducts are
produced in a sequence-independent manner at all guanine bases
throughout a DNA duplex and are primarily hydrolyzed to N7-
THBG. Only monoadducts produced in a suitable sequence
context can give rise to biologically relevant bifunctional lesions
(5′-GNC for interstrand cross-linking and 5′-GG for intrastrand
cross-linking). In summary, our results confirm common
sequence preferences for interstrand DNA cross-linking by DEB
and by the antitumor nitrogen mustards28,48 and reveal further
9
14362 J. AM. CHEM. SOC. VOL. 127, NO. 41, 2005

DNA−DNA Cross-Linking by 1,2,3,4-Diepoxybutane
A R T I C L E S
Scheme 3. Formation of Intrastrand and Interstrand DNA Cross-Linking by DEB Stereoisomers
stereospecific differences between DNA-DNA cross-linking
by the optical isomers of DEB.
Intrastrand and interstrand bis-N7G-BD lesions have
different stabilities in DNA. N-7 alkylation of guanine in DNA
generates a positive charge at the modified nitrogens, leading
to spontaneous depurination of the adducted nucleobase and the
concurrent formation of apurinic sites.49 We previously reported
that bis-N7G-BD lesions are more stable in DNA than the
corresponding N7-guanine monoadducts.27 In the present work,
we employed a stable isotope labeling approach to directly
compare the hydrolytic stability of intrastrand and interstrand
bis-N7G-BD adducts in double-stranded DNA. The DNA 28-
mer containing a single 5′-[1,7, NH2-15N3,2-13C-G]GC-3′/3′-
CCG-5′ and 5′-GGC-3′/3′-CC[1,7,NH2-15N3,2-13C-G]-5′ trinu-
cleotide was treated with R,R-DEB to generate both interstrand
and intrastrand cross-links (Table 1). DEB-treated DNA was
precipitated with cold ethanol, redissolved in buffer, and
incubated at physiological conditions. Aliquots were removed
at specific times, and the kinetics of the spontaneous release of
interstrand and intrastrand bis-N7G-BD nucleobase conjugates
from this duplex was determined by HPLC-ESI+-MS/MS
analysis of [15N3,13C1]-bis-N7G-BD and unlabeled bis-N7G-
BD, respectively. Another aliquot was completely hydrolyzed
to establish the starting numbers of intrastrand and interstrand
bis-N7G-BD cross-links in this DNA. First-order kinetic analysis
(Figure 6) indicated that the half-life of interstrand N7G-N7G
DEB cross-links in double-stranded DNA is 147 h, while the
t1/2 of intrastrand bis-N7G-BD adducts at the same conditions
is 35 h. The experiment was repeated 4 separate times, and the
results consistently demonstrated faster hydrolytic release of
intrastrand bis-N7G-BD adducts from the DNA backbone. This
noticeable difference in stability may be explained by a greater
charge density in intrastrand bis-N7G-BD adducts which contain
Figure 6. First-order kinetics for spontaneous depurination of intrastrand
and interstrand R,R-bis-N7G-BD lesions from double-stranded DNA. Bis-
N7G-BD adducts were generated by treating 15N3,13C-labeled DNA duplex
(5′-TAT ATA TTT ATA GGC TAT TAT TAT ATT A) (+ strand) with
R,R-DEB. DNA was precipitated with cold ethanol and incubated at
physiological conditions (37 °C, 200 mM NaCl). The amounts of hydro-
lytically released intrastrand and interstrand bis-N7G-BD cross-links were
determined from HPLC-ESI+-MS/MS peak areas corresponding to bis-
N7G-BD (m/z 389.2f 238.0) and [15N3]-bis-N7G-BD (m/z 393.2 f 242.0),
respectively.
two neighboring, positively charged N7-guanine adducts.50 For
example, Wogan and coauthors observed a decreased stability
of the N7-guanine adducts of aflatoxin in DNA that had a higher
number of adducts per DNA base pair.51 In contrast, N7-
alkylguanines within 1,3-interstrand bis-N7G-BD lesions are
located in different DNA strands and are flanked by another
base pair, decreasing the “adduct density” as compared with
the corresponding 1,2-instrastrand lesions. Alternatively, the
presence of a neighboring abasic site may facilitate the second
hydrolytic step because of an increased solvent accessibility of
the partially depurinated DNA strand.
Discussion
Our results presented above indicate that chirality can be an
important factor in determining DNA-DNA cross-linking
specificity of bifunctional electrophiles. We found that bifunc-
(47) Brookes, P.; Lawley, P. D. J. Chem. Soc. 1961, 3923-3927.
(48) Rink, S. M.; Solomon, M. S.; Taylor, M. J.; Rajur, S. B.; McLaughlin, L.
W.; Hopkins, P. B. J. Am. Chem. Soc. 1993, 115, 2551-2557.
(49) Singer, B.; Grunberger, D. Molecular Biology of Mutagens and Carcino-
gens; Plenum Press: New York and London, 1983.
(50) Gates, K. S.; Nooner, T.; Dutta, S. Chem. Res. Toxicol. 2004, 17, 839-
856.
(51) Groopman, J. D.; Croy, R. G.; Wogan, G. N. Proc. Natl. Acad. Sci. U.S.A
1981, 78, 5445-5449.
9
J. AM. CHEM. SOC. VOL. 127, NO. 41, 2005 14363

A R T I C L E S
Park et al.
tional DNA alkylation by S,S-DEB within the 5′-GGC target
sequence produced almost exclusively 1,3-interstrand bis-N7G-
BD lesions, while meso-DEB gave rise to similar numbers of
interstrand and instrastrand bis-N7G adducts (Table 1). R,R-
DEB exhibited intermediate specificity but was the only isomer
to induce a significant number of 1,2-interstrand cross-links
(Table 1). Although 1,2-intrastrand cross-linking by DEB has
been previously hypothesized,4 our experiments provide the first
direct evidence for the formation of such lesions in DEB-treated
DNA. The relative contribution of intrastrand DNA alkylation
by DEB in genomic DNA is likely to be greater than that in
5′GGC trinucleotides examined here because of the statistically
higher frequency of 5′-GG dinucleotides in general DNA
sequences.
Interchain DNA-DNA cross-linking by therapeutic bifunc-
tional alkylating agents is considered the primary mechanism
for their antitumor activity, while intrastrand lesions can exhibit
both cytotoxic and promutagenic properties.1,4 Interstrand
DNA-DNA adducts are among the most toxic nucleobase
lesions because they inhibit DNA replication and transcription,
and their enzymatic removal in cells is problematic because of
the loss of template information in both DNA strands. This
requires a complex repair process usually involving both
nucleotide excision repair and recombination repair,52 which can
lead to DNA rearrangements and deletions.53 Therefore, the
preferential formation of interstrand DNA-DNA lesions by S,S-
DEB is consistent with its potent cytotoxicity and its ability to
induce chromosomal aberrations.15 On the other hand, intrastrand
bis-alkylation of DNA by meso-DEB is likely to play a role in
mutagenesis.4 Importantly, our recent HPLC-ESI-MS/MS
results for bis-N7G-BD in liver DNA of laboratory animals
exposed to the metabolic precursor of DEB, 1,3-butadiene,
provide evidence for the formation of both diastereomers of
bis-N7G-BD in vivo (Loeber and Tretyakova, manuscript in
preparation).
The observed differences between the ability of DEB isomers
to induce DNA-DNA lesions are likely caused by different
orientations of functional groups in stereoisomeric N7-(2′-
hydroxy-3′,4′-epoxybut-1′-yl)-guanine (N7-HEBG) intermedi-
ates (Scheme 3). In S,S- and R,R-N7-HEBG, the epoxy ring
and the 2′-hydroxy group reside on one side of the plane formed
by the carbon chain, while, in the meso isomer, the epoxy
oxygen and the 2′-OH are located on different sides of the plane
(Scheme 3), potentially influencing the site of the second
alkylation. A recent NMR study of N6-(2′,3′,4′-trihydroxybut-
1′-yl)-deoxyadenosyl adducts has revealed distinct stereospecific
interactions between the hydroxyl groups of the trihydroxybutyl
side chain and the neighboring nucleobases.54 In the S,S adduct,
hydrogen bond formation was observed between the â-OH of
the side chain and the O4 of thymine in the opposite strand,
while, for the R,R adduct, hydrogen bonding was formed
between the γ-OH and the thymine O4 in the opposite DNA
strand.54 Although no NMR data is currently available for
stereoisomeric N7-HEBG intermediates, examination of mo-
lecular models suggests that hydrogen bonding between the 2′-
hydroxy group in S,S- and R,R-N7-HEBG and the N-3 of the
3′-neighboring guanine may position the 3′,4′-oxirane ring in a
favorable orientation for the SN2-type nucleophilic attack by
the N-7-guanine in the opposite strand (Figures S-5 and S-6).
In contrast, hydrogen bonding between the 2′-hydroxyl and 3′-
neighboring guanine in meso-DEB-induced S,R and R,S epoxy
alcohol intermediates stabilizes the conformation in which the
oxirane oxygen faces the N7-guanine in the opposite strand (e.g.,
Figure S-7). As a result, SN2 attack by the 3′-neighboring
guanine is preferred, shifting the cross-linking ratio in favor of
intrastrand lesions (Scheme 3).
Our stable isotope labeling experiments confirm the prefer-
ential formation of interstrand G-G butanediol lesions of DEB
between the guanine bases two residues apart in the DNA duplex
(1,3-cross-linking), with only the R,R isomer inducing measur-
able numbers of 1,2-interstrand lesions (Table 1). These results
are in agreement with an earlier study by Millard et al.28 who
used hot piperidine cleavage to reveal the alkylation sites within
DNA duplexes treated with racemic DEB. Further theoretical
and structural studies are needed to help explain this unusual
sequence specificity. The distance between the N7-guanine
atoms in 5′-GNC sequences of B-DNA (8.9 Å) is several
angstroms too long as compared with the length of the four-
atom tether of DEB. It is possible that the initially formed
cationic N7-(2′-hydroxy-3′,4′-epoxybut-1′-yl)-guanine interme-
diates (Figures S5-8) induce a conformational change in DNA,
bringing the second epoxide closer to the target G for interstrand
cross-link formation.55,56 Alternatively, the formation of N7-
HEBG may alter the chemical reactivity of neighboring guanine
bases via electronic or structural mechanisms, making the
nucleophilic attack by the distal G more favorable. Irrespective
of the mechanism, the resulting 1,3-bis-N7G-BD lesions must
distort the DNA duplex in order to be accommodated. Indeed,
our exonuclease resistance studies show that the 3′-exonuclease
activity of E. coli Polymerase I is blocked one nucleotide ahead
of the interstrand bis-N7G-BD lesion, suggesting that the DNA
duplex undergoes a structural change in the vicinity of the cross-
link (Figure S-9, S-10). This local distortion of the DNA duplex
may serve as a recognition signal for proteins involved in cross-
link repair. Studies are now underway to analyze the structural
perturbations induced by N7-(2′-hydroxy-3′,4′-epoxybut-1′-yl)-
guanine lesions and to examine the biological processing of
intrastrand and interstrand bis-N7G-BD lesions in the presence
of DNA polymerases and DNA repair proteins.
Acknowledgment. We thank Ms. Colleen Murray (University
of Minnesota) for her help with generating molecular models
of DNA duplexes containing stereoisomeric N7-(2′-hydroxy-
3′,4′-epoxybut-1′-yl)-guanine adducts and bis-N7G-BD cross-
links, Prof. Yusuf Abul-Hajj (University of Minnesota Depart-
ment of Medicinal Chemistry) for critical review of the
manuscript, and Gregory Janis for editorial help. Funding for
this research was from the Leukemia Research Foundation and
a grant from the National Cancer Institute (CA-100670). Rachel
Loeber is supported by an NIH Chemistry-Biology Interface
Training grant (T32-GM08700).
Supporting Information Available: Synthesis of R,R-, S,S-,
and meso-DEB (S-1); proton NMR spectra of synthetic DEB
(52) Cole, R. S. Proc. Natl. Acad. Sci. U.S.A. 1973, 70, 1064-1068.
(53) Zheng, H.; Wang, X.; Warren, A. J.; Legerski, R. J.; Nairn, R. S.; Hamilton,
J. W.; Li, L. Mol. Cell Biol. 2003, 23, 754-761.
(54) Merritt, W. K.; Scholdberg, T. A.; Nechev, L. V.; Harris, T. M.; Harris, C.
M.; Lloyd, R. S.; Stone, M. P. Chem. Res. Toxicol. 2004, 17, 1007-1019.
(55) Fan, Y.-H.; Gold, B. J. Am. Chem. Soc. 1999, 121, 11942-11946.
(56) Remias, M. G.; Lee, C. S.; Haworth, I. S. J. Biomol. Struct. Dyn. 1995,
12, 911-936.
9
14364 J. AM. CHEM. SOC. VOL. 127, NO. 41, 2005

DNA−DNA Cross-Linking by 1,2,3,4-Diepoxybutane
A R T I C L E S
stereoisomers (S-2), GC-MS traces or R,R-, S,S-, and meso-
DEB (S-3); HPLC-ESI+ MS/MS methods for bis-N7G-BD
diastereomers (S-4); molecular models of stereoisomeric N7-
(2′-hydroxy-3′,4′-epoxybut-1′-yl)-guanine adducts in 5′-GGC
trinucleotides (S-5-8), and exonuclease blockage sites in a
duplex containing racemic 1,4-bis-(guan-7-yl)-2,3-butanediol
cross-link (S-9, S-10). This material is available free of charge
via the Internet at http://pubs.acs.org.
JA051979X
9
J. AM. CHEM. SOC. VOL. 127, NO. 41, 2005 14365