DNA tandem lesions containing 8-oxo-7,8-dihydroguanine and formamido residues arise from intramolecular addition of thymine peroxyl radical to guanine

Article abstract of DOI:10.1021/tx0155909

Exposure of aerated aqueous solutions of dinucleoside monophosphates bearing both a pyrimidine base and a guanine residue to ionizing radiation leads to the formation of 8-oxo-7,8-dihydroguanine/formamido tandem lesions (8-oxodG/dF). Recent evidence for the formation of the latter damage within isolated DNA emphasized the possible biological relevance of this class of lesions. Therefore, an extensive mechanistic study of the formation of 8-oxodG/dF was carried out with thymine and guanine containing dinucleoside monophosphates (dGpdT and dTpdG). First, the peroxyl radical-induced degradation of guanine within dGpdT and dTpdG was studied in order to assess the possibility of intramolecular electron transfer between guanine and thymine peroxyl radicals. Then, the formation of a series of tandem lesions involving a formamido residue, thymine glycols, 8-oxodG, and oxazolone was monitored within aerated aqueous solutions of dTpdG and dGpdT exposed to γ-radiation. The absence of formation of tandem lesions other than 8-oxodG/dF in significant yield led to us propose a new mechanism involving addition of the thymine peroxyl radical to the guanine moiety. This received support from ¹⁸O labeling experiments.

Full text of DOI:10.1021/tx0155909

Chem. Res. Toxicol. 2002, 15, 445-454
445
DNA Ta n d em Lesion s Con ta in in g
8-Oxo-7,8-d ih yd r ogu a n in e a n d F or m a m id o Resid u es
Ar ise fr om In tr a m olecu la r Ad d ition of Th ym in e P er oxyl
Ra d ica l to Gu a n in e
Thierry Douki,* J ohann Rivie`re, and J ean Cadet*
DRFMC, Service de Chimie Inorganique et Biologique, Laboratoire des Le´sions des Acides
Nucle´iques, CEA/ Grenoble, 17, Avenue des Martyrs, F-38054 Grenoble Cedex 9, France
Received November 28, 2001
Exposure of aerated aqueous solutions of dinucleoside monophosphates bearing both a
pyrimidine base and a guanine residue to ionizing radiation leads to the formation of 8-oxo-
7,8-dihydroguanine/formamido tandem lesions (8-oxodG/dF). Recent evidence for the formation
of the latter damage within isolated DNA emphasized the possible biological relevance of this
class of lesions. Therefore, an extensive mechanistic study of the formation of 8-oxodG/dF was
carried out with thymine and guanine containing dinucleoside monophosphates (dGpdT and
dTpdG). First, the peroxyl radical-induced degradation of guanine within dGpdT and dTpdG
was studied in order to assess the possibility of intramolecular electron transfer between
guanine and thymine peroxyl radicals. Then, the formation of a series of tandem lesions
involving a formamido residue, thymine glycols, 8-oxodG, and oxazolone was monitored within
aerated aqueous solutions of dTpdG and dGpdT exposed to γ-radiation. The absence of formation
of tandem lesions other than 8-oxodG/dF in significant yield led to us propose a new mechanism
involving addition of the thymine peroxyl radical to the guanine moiety. This received support
from 18O labeling experiments.
Sch em e 1. Ch em ica l Str u ctu r e of th e Mod ified
Ba ses Stu d ied in Th is Wor k
In tr od u ction
Formation of oxidative base damage within DNA is
likely to be associated with deleterious cellular effects
including lethality and mutagenesis. Extensive studies
involving monomeric model systems such as bases and
nucleosides allowed the identification of the primary
produced radical intermediates (1, 2). In addition, isola-
tion of the final products followed by extensive spectro-
scopic analyses made possible their structural assign-
ment (3, 4). Altogether, oxidative degradation pathways
have been proposed for the four DNA bases and nucleo-
sides. The amount of available data on the chemistry of
bases within isolated and cellular DNA is growing with
the development of new analytical tools but remains to
be completed. In the recent years, model systems other
than single bases or nucleosides have been used. Isolation
of the products generated by oxidation of dinucleoside
monophosphates showed that in addition to the well-
known monomeric lesions, other damage involving two
adjacent bases, the so-called tandem lesions, was pro-
duced as the result of a single initial radical event (5). A
first class of tandem lesions involves addition of the
methyl-centered radical of a thymine residue to the
adjacent guanine base. Under anaerobic conditions, the
final product of the reaction was shown to exhibit a 8-(R-
thyminyl)-guanine structure (6-8). The formation of the
latter thymine-guanine adduct has also been observed
within isolated DNA (9). Addition of the methyl-centered
thymine radical to guanine may also occur in the pres-
ence of oxygen. Indeed, an adduct arising from the
addition of the latter thymine radical to the C4 position
of guanine followed by rearrangement of the purine ring
after addition of molecular oxygen has been isolated upon
type I photosensitization (10). The formation of other
thymine-guanine lesions involving the addition of the
C5 atom of a 6-hydroxy-5,6-dihydro-thymin-5-yl radical
to a guanine moiety has also been reported (11).
A second class of tandem lesions consists of two
adjacent monomeric base damage. A first example has
been reported by Box et al. who isolated degradation
products bearing both a 8-oxo-7,8-dihydroguanine (8-
oxoGua) and a formamido (dF) residue (Scheme 1) as
respective lesions of guanine and thymine upon exposure
of short oligonucleotides to X-rays in aerated aqueous
solution (5, 12-14). Evidence for the involvement of a
single initial event was obtained from kinetic consider-
ations. Recently, we reported that the two isomeric
* To whom correspondence should be addressed. (T.D.) Tel.: (33)4
38 78 31 91. E-mail: tdouki@cea.fr. (J .C.) Tel.: (33)4 38 78 49 87.
Fax: (33)4 38 78 50 90. E-mail: jcadet@cea.fr.
10.1021/tx0155909 CCC: $22.00 © 2002 American Chemical Society
Published on Web 02/27/2002

446 Chem. Res. Toxicol., Vol. 15, No. 3, 2002
Douki et al.
Sch em e 2. P r eviou sly P r op osed Mech a n ism for th e F or m a tion of 8-oxod G/d F Ta n d em Lesion sa
a
The key step is the intramolecular electron transfer from the guanine residue to the thymine peroxyl radical.
8-oxoGua/formamido tandem lesions (8-oxodGpdF and
currently accepted mechanism. Then, the formation of
new tandem lesions was monitored. The low level of the
latter degradation products led to the proposition of a
new mechanism for the formation of 8-oxodG/dF. This
involves the addition of the thymine peroxyl radical to
the C8 position of guanine, which was confirmed by 18O2
labeling experiments.
dFp8-oxodG), arising from °OH-mediated oxidation of
either guanine-pyrimidine or pyrimidine-guanine se-
quences, were also produced in isolated calf thymus DNA.
For this purpose, an assay involving an enzymatic
hydrolysis step followed by high-performance liquid
chromatography (HPLC)/tandem mass spectrometry
analysis was designed (15). The formation of dFp8-oxodG
has also been quantified within DNA using a 32P-post
labeling method (16). However, the reliability and the
high throughput of the HPLC-MS/MS assay together
with the possibility of detecting both dFp8-oxodG and
8-oxodGpdF allowed a more thorough investigation.
Using the latter method, the 8-oxodG/dF lesions were also
found to be produced in significant yield upon oxidation
of DNA under Fenton reaction conditions (15). Similar
results were recently obtained in short single-stranded
oligonucleotides (17). Dose-course study of the formation
of the lesions within DNA and comparison of their yield
of formation under various oxidative conditions including
γ-radiolysis, Fenton reaction, and type I photosensitiza-
tion (15) provided partial support for the mechanism
(Scheme 2) initially proposed by Box et al. (5). In the
initial step, an hydroxyl radical adds to the C5-C6
double bond of thymine. The resulting radical further
reacts with oxygen to produce a peroxyl radical. The
latter intermediate was then proposed to undergo an
intramolecular electron transfer from the guanine moiety,
leading to the formation of a thymine hydroperoxide on
one hand and the guanine radical cation on the other
hand. These transient species would then give rise to the
formamido residue and 8-oxoGua, respectively. This
mechanism predicts that a wide range of thymine and
guanine lesions might be involved in tandem lesions. The
latter type of damage may represent complex lesions for
the cell, resulting in a poor repair and a potentially high
mutagenicity. It thus appears necessary to get further
insights into the mechanism of formation of the tandem
lesions, especially of 8-oxodG/dF. Dinucleoside mono-
phosphates were chosen as the substrates for this inves-
tigation. Indeed, detection of tandem lesions within
isolated DNA is hampered by the necessity of quantita-
tive enzymatic release of the damage. In the first part of
the present work, thermal production of alkylperoxyl
radicals able to react with guanine and photochemical
generation of thymine peroxyl radicals within dinucleo-
side monophosphates were used as model reactions of the
Exp er im en ta l P r oced u r e
Ch em ica ls. 2,2′-Azobis(2-methylpropionamidine) hydrochlo-
ride (ABAP) and riboflavin were from Aldrich (Milwaukee, WI).
Bromine and potassium phosphate were purchased from Merck
(Darmstadt, Germany). Thymidylyl-(3′-5′)-2′-deoxyguanosine
(dTpdG) and 2′-deoxyguanosylyl-(3′-5′)-thymidine (dGpdT) were
prepared by liquid phase synthesis using a phosphotriester
approach. Water was deionized using a Milli-Q system. N-(2-
Deoxy-â-D-erythro-pentofuranosyl)formylamine-(3′-5′)-8-oxo-7,8-
dihydro-2′-deoxyguanosine (8-oxodGpdF) and 8-oxo-7,8-di-
hydro-2′-deoxyguanosylyl-(3′-5′)-N-(2-deoxy-â-D-erythro-pento-
furanosyl)formylamine (dFp8-oxodG) were synthesized as previ-
ously reported (15).
HP LC. Purification of the modified dinucleoside monophos-
phates was carried out on a Uptisphere ODB (250 mm × 4.6
mm i.d., 5 µm particle size) octadecylsilyl silica gel column
connected to a 7125 Rheodyne injector. The mobile phase was
delivered by a L6200 Merck-Hitachi intelligent pump in the
gradient mode. The initial eluent consisted of a 10 mM triethy-
lammonium acetate (pH 7) aqueous solution. After 5 min, the
proportion of acetonitrile was raised to 20% over a period of 40
min. The flow rate was set at 1 mL/min. The detection was
provided by a UV spectrophotometer set at 230 nm. HPLC
coupled to tandem mass spectrometry involved the use of an
Uptisphere ODB column (2 mm × 150 mm i.d., 3 µm particle
size) connected to a Shimadzu autosampler. A gradient of
acetonitrile in 2 mM ammonium formate was used at a flow
rate of 0.2 mL/min. The proportion of organic solvent rose from
0 to 20% within 25 min. The latter composition was maintained
for 5 min. The negative ion mass spectrometry detection was
provided by an API 3000 apparatus (Perkin-Elmer-Sciex, Thorn-
hill, Canada). For characterization purposes, the spectrometer
was used in either the single mass spectrometry (mass range
of 200-650) or the product ion scan mode. In the latter
1 Abbreviations: 8-oxodG, 8-oxo-7,8-dihydro-2′-deoxyguanosine; 8-ox-
oGua, 8-oxo-7,8-dihydroguanine; ABAP, 2,2′-azobis(2-methylpropiona-
midine) hydrochloride; BrOHdT, 5-bromo-6-hydroxy-5,6-dihydro-
thymidine (thymidine bromohydrin); dF, formamido residue; dG, 2′-
deoxyguanosine; dT, thymidine; dTGly, (5,6-dihydroxy-5,6-dihydro-
thymidine (thymidine glycol); dZ, 2,2-diamino-4-[(2-deoxy-â-D-erythro-
pentofuranosyl)amino]-5-(2H)-oxazolone (oxazolone); IT-ES-MS, ion
trap electrospray mass spectrometry.

Mechanism of Formation of DNA Tandem Lesions
Chem. Res. Toxicol., Vol. 15, No. 3, 2002 447
deoxyguanosine 3′-monophosphate]-), 310 (r.i.: 50%, [M - H
- C5NO4H5 - guanine]-), and 150 (r.i.: 100%, [guanine - H]-).
Ta ble 1. P a r a m eter s Used for th e HP LC-Ta n d em MS
Detection of th e Mod ified Din u cleosid e Mon op h osp h a tes
in th e MRM Mod e
3. Th ym idylyl-(3′-5′)-2,2-diam in o-4-[(2-deoxy-â-D-er yth r o-
p en tofu r a n osyl)a m in o]-5-(2H)-oxa zolon e (d Tp d Z) a n d 2,2-
Dia m in o-4-[(2-d eoxy-â-D-er yth r o-p en tofu r a n osyl)a m in o]-
5-(2H)-oxa zolon e-(3′-5′)-th ym id in e (d Zp d T). A saturated
solution of riboflavin in water (200 µL) was added to aqueous
solutions of either dGpdT or dTpdG (1 mg/mL, 2 mL). The
resulting solutions were then exposed under air bubbling for
15 min to the UVA radiation delivered by a lamp consisting of
two 15 W tubes. The irradiated mixtures were left overnight at
37 °C. The solutions were purified by HPLC, and the main peaks
were collected. For both dGpdT and dTpdG, the major product
(Rt, 18 and 19 min, respectively) was identified as dZpdT and
dTpdZ, respectively. Indeed, the UV spectrum of each of these
isolated compounds lacked the absorption bands corresponding
to the guanine moiety. In addition, both compounds were found
to release guanidine upon hot alkaline treatment. Final proof
of structure was obtained from IT-ES-MS analysis. The MS1
spectrum exhibited the expected pseudo-molecular ion at m/z
) 549 (r.i.: 30%), together with a more intense signal at m/z )
505 (r.i.: 100%), corresponding to the loss of CO2, a character-
istic behavior of oxazolone derivatives. For both dTpdZ and
dZpdT, fragmentation of the pseudo-molecular ion (m/z ) 549)
yielded fragments at m/z ) 505 (r.i.: 25%) and 477 (r.i.: 100%).
In contrast, fragmentation mass spectra of the ion at m/z ) 505
were different from one compound to the other. Indeed, dTpdZ
yielded fragments at m/z ) 379 (r.i.: 35%, [M - H - thymine]-)
and 263 (r.i.: 33%, [M - H - thymidine]-), whereas a single
daughter ion was observed at m/z ) 321 (r.i.: 45%, [thymidine
5′-monophosphate]-) in the fragmentation mass spectrum of the
m/z ) 505 ion arising from the loss of CO2 from dZpdT.
4. 5,6-Dih yd r oxy-5,6-d ih yd r ot h ym id ylyl-(3′-5′)-2,2-d i-
a m in o-4-[(2-d eoxy-â-D-er yth r o-p en tofu r a n osyl)a m in o]-5-
(2H)-oxa zolon e (d TGlyp d Z) a n d 2,2-Dia m in o-4-[(2-d eoxy-
â-D-er yth r o-p en tofu r a n osyl)a m in o]-5-(2H)-oxa zolon e-(3′-
5′)-5,6-d ih yd r oxy-5,6-d ih yd r oth ym id in e (d Zp d TGly). Solu-
tions of the isolated cis diastereoisomers of dTGlypdG and
dGpdTGly were photosensitized to UVA in the presence of
riboflavin as described for dTpdZ and dZpdT. Irradiated solu-
tions were purified by HPLC. In both cases, the major photo-
oxidation product was shown to be either dTGlypdZ or dZpdT-
Gly, on the basis of UV spectroscopic features, guanidine release,
and mass spectrometry analyses. For the latter experiments,
the pseudo-molecular ion was observed for all compounds at m/z
) 583. A fragment, corresponding to the loss of CO2, was also
detected at m/z ) 539. Fragmentation of the latter species
yielded for all compounds an ion at m/z ) 396 (r.i.: 100%, [M -
H - C5NO4H5]-). An additional signal was observed at m/z )
212 (r.i.: 15%, [5,6-dihydroxy-5,6-dihydrothymidine 5′-mono-
phosphate - C5NO4H5]-) and 281 (r.i.: 20%, [dZ 5′-monophos-
phate - CO2]-) for the cis diastereoisomers of dZpdTGly and
dTGlypdZ, respectively.
parent
ion
retention
time (min)
compd
dZpdT
daughter ions
505
505
586
586
424
424
505
505
539
539
620
620
321, 195
321, 263, 195
321, 166
460, 362, 195
240, 195
240, 195
240, 166
362, 166
396, 212
396, 281
14.4
dTpdZ
14.8
26.2
23.1
2.5
3.0
18.9
16.5
6.8, 18.7
6.5, 14.4
20.3, 21.6, 24.3
19.0, 21.7
8-oxodGpdT
dTp8-oxodG
dZpdF
dFpdZ
8-oxodGpdF
dFp8-oxodG
dZpdTGly
dTGlypdZ
8-oxodGpdTGly
dTGlyp8-oxodG
477, 212, 166
477, 362, 166
experiments, the pseudo-molecular ion of the compound of
interest was isolated and fragmented. The fragmentation mass
spectrum was then recorded (mass range of 100-650). In
quantitative studies, the mass spectrometer was used in the
multiple reaction monitoring (MRM) mode. The molecular ion
of the targeted compound was isolated on the first quadrupole,
and specific daughter ions were collected and quantified. The
specific monitored transitions and retention times of each of the
studied compounds are listed in Table 1.
Syn th esis of th e Mod ified Din u cleosid e Mon op h os-
p h a tes. 1. 5-Br om o-6-h yd r oxy-5,6-d ih yd r oth ym id ylyl-(3′-
5′)-2′-d eoxygu a n osin e (Br OH d Tp d G) a n d 2′-Deoxy-
gu a n osylyl-(3′-5′)-5-b r om o-6-h yd r oxy-5,6-d ih yd r ot h ym i-
d in e (d Gp Br OHd T). Bromine (0.8 equiv) suspended in 500 µL
of water was added to a solution of dGpdT (5 mL, 1 mg/mL)
cooled in an ice bath. After 30 min, air was bubbled to remove
traces of bromine. The product of the bromination reaction was
purified by HPLC. Peaks corresponding to the main reaction
products (retention time (Rt), 31 and 33 min) were observed on
the chromatogram in addition to unreacted dinucleoside mono-
phosphate (Rt, 27 min). They were identified by ion trap
electrospray mass spectrometry (IT-ES-MS) on an ion trap LCQ
apparatus (Thermoquest, San J ose, CA) as the two trans
diastereoisomers of dGpBrOHdT. The spectrum exhibited two
pseudo-molecular ions (m/z ) 666 and 668), corresponding to
the presence of the two main stable bromine isotopes (Mw ) 79
and 81). The fragmentation mass spectrum exhibited a base
peak at m/z ) 586, corresponding to the loss of HBr. A weaker
signal was observed at m/z ) 435 (relative intensity (r.i.): 15%,
[M - H - 8-bromoguanine]-). Similar results were obtained
with dTpdG. The respective retention times of the unmodified
dinucleoside monophosphate and the two diastereoisomers of
BrOHdTpdG were 25, 29, and 30 min, respectively.
2.
5,6-Dih yd r oxy-5,6-d ih yd r ot h ym id ylyl-(3′-5′)-2′-d e-
oxygu a n osin e (d TGlyp d G) a n d 2′-Deoxygu a n osylyl-(3′-5′)-
5,6-d ih yd r oxy-5,6-d ih yd r oth ym id in e (d Gp d TGly). Sodium
hydrogen carbonate was added to the solution of either dGp-
BrOHdT or BrOHdTpdG to reach pH 8. The resulting solutions
were left overnight at room temperature. The resulting mixture
was purified by HPLC. Two main fractions that were found to
contain each of the cis diastereoisomers of dGpdTGly were
isolated (Rt, 24 and 26 min). They were analyzed by IT-ES-MS.
The pseudo-molecular ion was observed at m/z ) 604. MS2
fragmentation spectra exhibited ions at m/z ) 586 (r.i.: 20%,
[M - H - H2O]-) and 461 (r.i.: 100%, [M - H- 143]-, loss of
OdCdN-CO-C(CH3)(OH)-CHdO, C5NO4H5). Similar results
were obtained for the synthesis of dTGlypdG from BrOHdTpdG.
Two fractions were isolated (Rt, 23 and 25 min). MS1 and MS2
IT-ES-MS spectra were similar to those of dGpdTGly. Differ-
ences between the thymine glycol derivatives arising from
dGpdT and dTpdG were only observed on MS3 spectra (frag-
mentation of the ion at m/z ) 461 from the MS2 spectrum).
Indeed, dTGlypdG yielded a single intense fragment at m/z )
346 (r.i.: 100%, [2′-deoxyguanosine 5′-monophosphate]-) while
dGpdTGly gave rise to ions at m/z ) 346 (r.i.: 55%, [2′-
P r ep a r a tion of Ca libr a ted Solu tion s of 8-Oxo-7,8-d ih y-
d r o-2′-d eoxygu a n osylyl-(3′-5′)-t h ym id in e (8-oxod Gp d T),
Th ym id ylyl-(3′-5′)-8-oxo-7,8-d ih yd r o-2′-d eoxygu a n osin e
(d Tp 8-oxod G), 8-Oxo-7,8-d ih yd r o-2′-d eoxygu a n osylyl-(3′-
5′)-5,6-d ih yd r oxy-5,6-d ih yd r oth ym id in e (8-oxod Gp d TGly),
a n d 5,6-Dih yd r oxy-5,6-d ih yd r oth ym id ylyl-(3′-5′)-8-oxo-7,8-
d ih yd r o-2′-d eoxygu a n osin e (d TGlyp 8-oxod G). Methylene
blue at a final optical density at 550 nm of 1 and FeSO4 (200
µL, 1 mM) were added to solutions of either dTpdG, dGpdT,
dTGlypdG, or dGpdTGly (2 mL). The resulting mixtures were
placed in a glass tube under a 500 W halogen lamp (Phillips).
Air was continuously bubbled during the 10 min irradiation
period. An aliquot fraction (5 µL) of each of the irradiated
solutions was diluted to 50 µL by addition of water. The modified
dinucleoside monophosphates were digested by sequential
incubation at pH 5 in the presence of nuclease P1 and phos-
phodiesterase II and pH 8 following addition of alkaline phos-
phatase and phosphodiesterase I. The content in 8-oxodG of the
resulting samples was determined by HPLC-EC using a previ-

448 Chem. Res. Toxicol., Vol. 15, No. 3, 2002
Douki et al.
ously described method (18). Fragmentation mass spectra of the
compounds were recorded by HPLC coupled to tandem mass
spectrometry. Three peaks, corresponding to diastereoisomers
exhibiting the same mass spectra, were observed on the chro-
matogram of dTGlyp8-oxodG and 8-oxodGpdTGly. In both cases,
the first eluting peak was broad, suggesting the coelution of two
diastereoisomers. Only one peak exhibiting the expected mass
was observed within oxidized solutions of dTpdG and dGpdT.
The pseudo-molecular and the main daughter ions were the
following:
dTp8-oxodG ([M - H]-) 586): 460 (r.i.: 70%, [M - H -
thymine]-); 362 (r.i.: 20%, [8-oxodG 5′-monophosphate]-); 195
(r.i.: 100%, [2-deoxyribose monophosphate]-); 166 (r.i.: 15%,
[8-oxoGua - H]-).
8-oxodGpdT ([M - H]-) 586): 419 (r.i.: 15%, [M - H -
8-oxoGua]-); 321 (r.i.: 15%, [thymidine 5′-monophosphate]-);
166 (r.i.: 100%, [8-oxoGua - H]-).
8-oxodGpdTGly ([M - H]-) 620): 477 (r.i.: 80%, [M - H -
C5NO4H5]-); 212 (r.i.: 20%, [5,6-dihydroxy-5,6-dihydrothymidine
5′-monophospate - C5NO4H5]-); 166 (100%, [8-oxoGua - H]-).
dTGlyp8-oxodG ([M - H]-) 620): 477 (100%, [M - H -
C5NO4H5]-); 362 (40%, [8-oxodG 5′-monophosphate]-); 166 (10%,
[8-oxoGua - H]-).
F igu r e 1. Fragmentation spectrum in the negative mode of
8-oxodGpdTGly obtained by triple quadrupole mass spectrom-
etry analysis. The parent ion was set at m/z ) 620.
ABAP Tr ea tm en t of d Tp d G a n d d Gp d T. Two sets of
conditions were used for the ABAP-mediated oxidation of dGpdT
and dTpdG. In a first series of experiments, ABAP was used at
concentrations of 0, 2.5, 5, 10, and 20 mM by addition of a freshly
prepared 1 M solution into tubes containing 2 mL of a 1 mM
aqueous solution of dinucleoside monophosphate. The samples
were left under constant air bubbling for 1 h at 37 °C. Then,
air bubbling was stopped and the samples were left at room
temperature overnight. ABAP was also used under much milder
conditions. The solutions of either dTpdG or dGpdT were placed
at 0 °C in an ice bath. Air was bubbled in the solution for 15
min. Then, a 1 mM solution of ABAP was added in order to
reach final concentrations of 0, 10, 20, 50, and 100 µM. The
samples were left under air bubbling in the ice bath for 1 h. Air
bubbling was stopped, and the samples were left at 4 °C for 2
h. They were then left at room temperature overnight. In both
cases, samples were analyzed by HPLC-MS/MS.
UV P h otolysis of d Gp Br OHd T. The two trans diastereoi-
somers of dGpBrOHdT were prepared from 5 mL of dGpdT (1
mM) and isolated by HPLC as described above. The mixture of
both diastereoisomers was solubilized in water and transferred
into a 3 mL quartz cell (1 cm optical path, 1 cm width). The
sample was then exposed to the UVC light emitted by a 2 × 15
W germicidal lamp (fluence rate: 3.5 kJ m-2 min-1) for 2 h
under constant stirring. The irradiated solution was then
concentrated under vacuum to 100 µL and analyzed by HPLC
coupled to mass spectrometry in the product ion scan mode. The
selected precursor ion was 505, corresponding to the pseudo-
molecular ion of 8-oxodGpdF. A peak was observed at the
expected retention time (18.9 min). The corresponding fragmen-
tation mass spectrum was the following: 505 (r.i.: 20%, [M -
H]-); 338 (r.i.: 10%, [M - H - 8-oxoGua]-); 240 (r.i.: 15%, [dF
5′-monophosphate]-); 166 (r.i.: 100%, [8-oxoGua - H]-).
γ-Ir r a d ia tion of d Gp d T a n d d Tp d G. Solutions of either
dGpdT or dTpdG (1 mM, 1 mL) that contained 10 mM potassium
phosphate (pH 7) were exposed to the γ-rays emitted by a 60Co
source (dose rate 20 Gy/min) in a glass tube under constant air
bubbling. The overall dose applied was either 0, 20, 40, 100,
200, or 400 Gy. Each sample was then analyzed in triplicate
for its content in lesions by HPLC-MS/MS. Irradiation of dGpdT
was also carried out under 18O-labeled molecular oxygen
atmosphere (96.7%, Eurisotop, St. Aubin, France). For this
purpose, the sample was degassed by bubbling argon followed
by a series of freezing/thawing steps under vacuum. Labeled
oxygen was then added into the tube. The latter was sealed and
exposed to γ-radiation for 30 min (overall dose, 600 Gy). The
content of the irradiated solution was then analyzed by HPLC-
MS in the single mass spectrometry, product ion scan, and MRM
modes.
Resu lts
Syn th etic a n d An a lytica l Asp ects. A series of
modified dinucleoside monophosphates were prepared for
this work using established synthetic pathways. They
were isolated by HPLC and characterized, mostly on the
basis of mass spectrometry analyses. Indeed, the latter
technique allowed the assignment of the chemical struc-
ture of the two base moieties of the modified dinucleoside
monophosphates through the study of the fragmentation
pattern. As a general trend, a first fragmentation in-
volved the loss of the 5′-end base. In addition, the
nucleoside 5′-monophosphate containing the 3′-end base
was obtained as a daughter ion. Additional fragmentation
pathways were observed for some lesions, as discussed
below. A key step in the present work was the synthesis
of the trans thymine bromohydrins (5-bromo-6-hydroxy-
5,6-dihydrothymine) of dTpdG and dGpdT. The corre-
sponding derivatives have been used as precursors in the
synthesis of thymidine and 2′-deoxycytidine hydroper-
oxides (19, 20). In the present work, reaction conditions
were optimized in order to prevent the addition of
bromine to the C8 position of the guanine residue of the
starting dinucleoside monophosphates. This was achieved
by lowering the reaction temperature and limiting the
conversion of thymine to ca. 60%. For each of the
dinucleoside monophosphates, the two trans diastereo-
isomers were isolated and unambiguously characterized
by mass spectrometry. The hydrolytic conversion into the
corresponding cis thymine glycol derivatives under slightly
alkaline conditions (21) was quantitative. The resulting
modified dinucleoside monophosphates (dTGlypdG and
dGpdTGly) exhibited similar fragmentation spectra, with
a major daughter ion at [M - H - 143]-, which could
not be accounted for by the loss of the 5′-end base. The
favored loss of 143 amu from dTGlypdG and dGpdTGly,
which was also observed for dTGlypdZ, dTGlyp8-oxodG,
dZpdTGly, and 8-oxodGpdTGly, can be rationalized in
terms of a rearrangement of the ring-opened form of the
thymine glycol moiety (Figure 1).
Another series of modified dinucleoside monophos-
phates involved degradation of guanine into its oxazolone
derivative. All corresponding compounds, dTpdZ, dZpdT,
dTGlypdZ, and dZpdTGly, were found to release guani-
dine upon alkaline treatment. In addition, these mol-

Mechanism of Formation of DNA Tandem Lesions
Chem. Res. Toxicol., Vol. 15, No. 3, 2002 449
F igu r e 3. Degradation of dGpdT and dTpdG upon incubation
with large concentrations of ABAP in aerated aqueous solution
at 37 °C. The amount of dinucleoside monophosphate was
inferred from HPLC-UV analysis. Dinucleoside monophosphates
were solubilized at 1 mM concentration in 10 mM, pH 7,
phosphate buffer. The reported values represent the average of
two independent determinations.
F igu r e 2. MS1 and MS2 mass spectra of dTGlypdZ obtained
by IT-MS analysis. The fragmentation spectrum was recorded
with the parent ion set at m/z ) 539 ([M - CO2 - H]-).
ecules lost CO2 upon ionization in the mass spectrometer
(Figure 2). These two features have been already ob-
served upon characterization of the 2′-deoxyribonucleo-
side of oxazolone (22, 23). It should be added that the
guanidine release was used to calibrate the solutions,
using a previously described HPLC/fluorescence assay
(24). Finally, solutions containing 8-oxodG derivatives
were prepared. The synthesis in large amount was not
attempted from these dinucleoside monophosphates be-
cause 8-oxodG is a minor oxidation product within small
model systems. Therefore, dTpdG, dGpdT, dTGlypdG,
and dGpdTGly were exposed to singlet oxygen, a specific
oxidant of guanine residues, in order to keep unmodified
the adjacent pyrimidine moiety. Ferrous ions were added
to the solution in order to increase the yield of 8-oxodG
derivatives by reduction of the transient guanine endo-
peroxide (25). The isolation of pure dinucleoside mono-
phosphates bearing both an oxazolone and a formamido
residue was not possible because of the lack of hydro-
phobicity of the compounds, which are likely to be almost
not retained on reverse phase HPLC columns. However,
γ-irradiated solutions of dGpdT and dTpdG contained a
degradation product exhibiting the expected molecular
weight and undergoing the loss of CO2 upon ionization.
The fragmentation mass spectra of the latter ion were
in agreement with the presence of a formamido moiety
since the main observed fragment corresponded to the
phosphorylated derivative of the latter nucleoside. There-
fore, these lesions were tentatively identified as dZpdF
and dFpdZ.
The modified dinucleoside monophosphates prepared
as reported above were then used to optimize the
response of the mass spectrometer operating in the
sensitive and specific MRM mode. The latter approach
is receiving major attention in the field of DNA damage
with a growing number of applications. Among them,
assays involving the detection of modified dinucleoside
monophosphates have been developed for tandem lesions
(15) and UV-induced pyrimidine dimeric photoproducts
(26, 27). In the MRM mode, the pseudo-molecular ion of
the targeted molecule ([M - H]-) is isolated in the first
quadrupole of the apparatus. It is then fragmented, and
specific fragments are quantified in the last quadrupole.
The ions used in the quantitative analyses of radiation-
induced degradation products of dGpdT and dTpdG are
listed in Table 1. It should be noted that the precursor
ion of the transition used for dZ-containing lesions
corresponded to [M - CO2 - H]-.
P er oxyl Ra d ica l-In d u ced Oxid a tion of d Gp d T
a n d d Tp d G. To further investigate the reaction of a 6(5)-
hydroxy-5(6)-peroxy-5,6-dihydrothymine moiety with an
adjacent guanine base, as previously proposed in the
mechanism of formation of 8-oxodG/dF tandem lesions,
peroxyl radicals produced by thermal decomposition of
ABAP were used as model systems together with dTpdG
and dGpdT as substrates. The former chemical generates
2-methyl-3-yl-priopionamidine radicals, which further
react with oxygen to yield the related peroxyl radicals.
Each molecule of ABAP gives rise to two identical
radicals, with release of nitrogen. Therefore, in contrast
to other systems, only one class of radical species is
involved. In addition, ABAP is a water soluble radical
precursor, a feature required in order to react with
dinucleoside monophosphates. Peroxyl radicals were
found to be very efficient damaging species of the guanine
moieties of dGpdT and dTpdG. Indeed, when used at high
concentrations (up to 20 mM), an almost complete
degradation of both dinucleoside monophosphates was
observed (Figure 3). The major product obtained under
these conditions was identified as a modified dinucleoside
monophosphate bearing an oxazolone and a thymine
moiety. The rate of formation of the latter lesion and of
the corresponding 8-oxodG analogue was determined
within solutions of dGpdT and dTpdG treated by ABAP
under milder conditions (ABAP concentration lower than
100 µM). The level of 8-oxodG was found to slightly
decrease while that of oxazolone increased in a quadratic
relationship with respect to the ABAP concentration
(Figure 4). These features are very similar to recent
observations made upon type I photosensitization of 2′-
deoxyguanosine (Ravanat and Cadet, manuscript in
preparation). Altogether, the present results show that
peroxyl radicals may undergo electron abstraction from
the guanine moiety of dinucleoside monophosphates.
F or m a tion of d F p 8-oxod G a n d 8-oxod Gp d F u p on
P h ot olysis of t h e Br om oh yd r in s of d Tp d G a n d
d Gp d T. Previous results have led to the conclusion that
formation of the 8-oxodG/dF tandem lesions upon expo-
sure to γ-radiation arose from the initial formation of the

450 Chem. Res. Toxicol., Vol. 15, No. 3, 2002
Douki et al.
F igu r e 5. Formation of oxazolone-containing single lesions
upon exposure of dGpdT and dTpdG to increasing doses of
γ-radiation. Dinucleoside monophosphates were solubilized at
1 mM concentration in 10 mM, pH 7, phosphate buffer. The
reported values, expressed as number of lesions per 103 mol-
ecules of dinucleoside monophosphate, represent the average
( standard deviation of three independent determinations.
F igu r e 4. Formation of guanine oxidation products of dTpdG
upon incubation at 0 °C with a low amount of ABAP. Dinucleo-
side monophosphates were solubilized at 1 mM concentration
in 10 mM, pH 7, phosphate buffer. The reported values,
expressed as number of lesions per 103 molecules of dinucleoside
monophosphate, represent the average ( standard deviation of
three independent determinations.
Ta ble 2. Ra d iolytic Yield of F or m a tion of Oxid a tion
P r od u cts of d Gp d T a n d d Tp d G u p on Exp osu r e to
γ-Ra d ia tion in Aer a ted Aqu eou s Solu tion a
The low yield of 8-oxodG upon γ-irradiation of dG was
proposed to be accounted for by a reaction between the
oxidizing and the reducing radicals of guanine (28).
Ongoing studies indicate that secondary oxidation of
8-oxodGuo by the guanine oxidizing radical is also
involved (Ravanat and Cadet, manuscript in prepara-
tion). The present observation of the formation of 8-ox-
odG-containing dinucleoside monophosphates in signifi-
cant yield suggests that these reactions are much less
efficient with dGpdT and dTpdG. The high yield of
8-oxodG/dF tandem lesions also indicates that the reac-
tion between the 8-oxodG moieties and the guanine
oxidizing radical is much less efficient that with isolated
nucleosides. These observations may be accounted for by
the electrostatic repulsion of the negative charge of
phosphodiester linkage of dinucleoside monophosphates.
Altogether, both oxazolone and 8-oxodG were found to
be primary °OH radical-induced degradation products of
guanine moieties in dinucleoside monophosphates, as
confirmed by the linearity of their dose-course formation
(Figure 5 for dZpdT and dTpdZ).
F or m a tion of 8-oxod G- a n d Oxa zolon e-Con ta in -
in g Ta n d em Lesion s u p on Exp osu r e of d Tp d G a n d
d Gp d T to γ-Ra d ia tion . The formation of 8-oxodGpdF
and dFp8-oxodG was first monitored in aerated aqueous
solutions of dinucleoside monophosphates exposed to
γ-radiation. They were both found to be produced in a
linear relationship with respect to the dose (Figure 6),
in agreement with previous data (5). A drastic sequence
effect was observed since the radiolytic yield of 8-oxo-
dGpdF (G ) 8.2 ( 0.9 nmol J -1) was higher than that of
dFp8-oxodG (G ) 0.6 ( 0.1 nmol J -1). The predominant
formation of 8-oxodG/dF tandem lesion at GT with
respect to TG sequence has already been observed within
γ-irradiated isolated DNA (15). Interestingly, a similar
sequence effect has been reported for oligonucleotides and
DNA concerning the hydrogen abstraction from a 2-deoxy-
ribose moiety by a vicinal uracil-5-yl radical produced by
photochemical decomposition of 5-bromouracil (29, 30).
Similar conformational effects are thus likely to be
involved in both reactions, in agreement with the produc-
tion of a thymine radical located at the C5/C6 positions
as the initial event in the formation of 8-oxodGpF. The
tandem lesions exhibiting both formamido and oxazolone
dinucleoside monophosphate
dGpdT
dTpdG
8-oxodG/dT
dZ/dT
8-oxodG/dF
8-oxodG/dTGly
3.25 ( 0.32
5.90 ( 0.12
8.17 ( 0.87
0.10 ( 0.01
1.38 ( 0.07
8.73 ( 0.24
0.63 ( 0.14
0.09 ( 0.01
a
Results are expressed as nmol/J .
5-hydroxy-6-yl and/or the 6-hydroxy-5-yl radicals of
thymine (and cytosine) (15). To confirm this hypothesis,
the bromohydrins of dGpdT were used as precursors of
the latter transient species. Indeed, UVC irradiation of
dGpBrOHdT leads to the homolytic cleavage of the C-Br
bond, yielding the oxidizing pyrimidine radical that is
also obtained by addition of °OH to the C6 position of
the thymine moiety. Analysis of the resulting sample by
HPLC-MS/MS unambiguously showed the presence of
8-oxodGpdF among the photolysis products (yield, 5%).
Indeed, the fragmentation mass spectrum of the com-
pound eluting at the expected retention time was identi-
cal to that of synthetically prepared 8-oxodGpdF. The
formation of other products was also observed. However,
they were found to be only modified on the thymine
moiety, as shown by the loss of unmodified guanine upon
fragmentation during HPLC-MS/MS analysis.
F or m a tion of Degr a d ed Gu a n in e Sin gle Lesion s
u p on Exp osu r e of d Tp d G a n d d Gp d T to γ-Ra d ia -
tion . The formation of a wide set of modified dinucleoside
monophosphates was monitored within aerated aqueous
solutions of dGpdT and dTpdG exposed to γ-radiation.
First, the formation of oxidation products exhibiting
either a 8-oxodG or an oxazolone moiety on one hand and
thymine on the other hand was quantified. In both cases,
oxazolone was found to be the main hydroxyl radical-
induced product when compared to 8-oxodG (Table 2).
The ratio between the yields of the two former lesions
was 1.8 ( 0.4 and 6.3 ( 0.8 for dGpdT and dTpdG,
respectively. A similar trend has been observed upon
exposure of aerated aqueous solution of 2′-deoxygua-
nosine to γ-radiation (28). However, a ratio of 62 was then
found between the yield of oxazolone (measured as
guanidine releasing compound) and the yield of 8-oxodG.

Mechanism of Formation of DNA Tandem Lesions
Chem. Res. Toxicol., Vol. 15, No. 3, 2002 451
F igu r e 8. Mass spectrum of 8-oxodGpdF and 8-oxodGpdT
obtained upon γ-irradiation of dGpdT in the presence of 18O-
labeled molecular oxygen. The irradiated solution was injected
onto the HPLC-MS system with the spectrometer operating in
the MS1 mode. The chromatogram was simultaneously recorded
between two mass ranges, namely, 500-515 and 580-600 amu.
Identification of the products was inferred from both their mass
spectra and retention times.
F igu r e 6. Formation of dTGly/8-oxodG tandem lesions upon
exposure of dTpdG and dGpdT to increasing doses of γ-radiation.
Dinucleoside monophosphates were solubilized at 1 mM con-
centration in 10 mM, pH 7, phosphate buffer. The reported
values, expressed as number of lesions per 103 molecules of
dinucleoside monophosphate, represent the average ( standard
deviation of three independent determinations.
La belin g of 8-oxod Gp d F a n d 8-oxod Gp d T u p on
γ-Ir r a d ia tion of d Gp d T in 18O2-Sa tu r a ted Aqu eou s
Solu tion . To gain further insight into the mechanism
of formation of 8-oxodGpdF, γ-irradiation of dGpdT was
then carried out in aqueous solution saturated with 18O2.
The irradiated samples were analyzed by HPLC-MS/MS
for their content in 8-oxodG-containing dinucleoside
monophosphates. First, single mass spectrometry detec-
tion allowed to determine the level of 18O incorporation
into 8-oxodGpdF and 8-oxodGpdT. The pseudo-molecular
ion ([M - H]-) of the former molecule was found at m/z
) 505, 507, and 509 in a 20/43/37 ratio (Figure 8). This
can be accounted for by the involvement of two oxygen
atoms in the mechanism of formation of 8-oxodGpdF. In
contrast, no significant labeling of 8-oxodGpdT was
observed. The lack of oxygen incorporation into single
8-oxodG lesions is in agreement with the fact that the
oxygen atom at the C8 position arises from the initial
addition of an hydroxyl radical, the latter produced from
water. To identify the labeled position in 8-oxodGpdF,
the mass spectrometer was then used in the product ion
scan mode, which allowed the recording of the fragmen-
tation mass spectra of pseudo-molecular ions at m/z )
505 ([M - H]-), 507 ([M - H + 2]-), and 509 ([M - H +
4]-). Interestingly, fragmentation of 8-oxodGpdF yields
ions containing either the formamido (m/z ) 240, [dF 5′-
monophosphate]-) or the 8-oxoGua moiety (m/z ) 166,
[8-oxoGua - H]-). Therefore, it was shown that 8-
oxodGpdF molecules containing only one 18O atom were
all labeled on their 8-oxodG residue. Lesions containing
two 18O atoms were labeled on both the 8-oxodG and the
dF moieties. These results were confirmed by more
quantitative MRM analyses (Figure 9). In control experi-
ments, similar HPLC-MS/MS measurements were per-
formed on dGpdT samples exposed to γ-radiation in 16O2-
saturated solution. It was found that the abundance of
the M + 2 and M + 4 pseudo-molecular ions for both
8-oxodGpdT and 8-oxodGpdF was negligible.
F igu r e 7. Formation of dTGly/dZ tandem lesions upon expo-
sure of dTpdG and dGpdT to increasing doses of γ-radiation.
Dinucleoside monophosphates were at 1 mM concentration and
solubilized in 10 mM, pH 7, phosphate buffer. The reported
values, expressed as number of lesions per 103 molecules of
dinucleoside monophosphate, represent the average ( standard
deviation of three independent determinations.
moieties were found to be produced upon exposure of
aqueous solutions of dinucleoside monophosphates to
γ-radiation. However, the dose-course formation of
dZpdF and dFpdZ within irradiated solutions of dGpdT
and dTpdG, respectively, was found to be quadratic.
The formation of tandem lesions containing both a
8-oxodG and a thymine glycol residue was then quanti-
fied (Figure 6). It should be pointed out that this class of
damage consists of a mixture of diastereoisomers, well-
resolved by HPLC, due to the presence of two asymmetric
carbon atoms on the thymine glycol moiety. Both dTG-
lyp8-oxodG and 8-oxodGpdTGly were detected in γ-ir-
radiated samples of dTpdG and dGpdT, respectively.
However, their yield of formation was low, when com-
pared to dFp8-oxodG and 8-oxodGpdF. Tandem lesions
bearing both an oxazolone and a thymine glycol moiety
were also quantified. They were easily detected, even
though produced in significantly lower yield than the
corresponding thymine-containing modified dinucleoside
monophosphates. Their dose-course formation was found
to be quadratic (Figure 7), as already observed for dFpdZ
and dZpdF, strongly suggesting that these lesions re-
sulted from two separate radical events.
Discu ssion
Identification of the radiation-induced products of DNA
components has been carried out for decades on simple
monomeric compounds such as nucleosides and bases.
Recently, more relevant model systems such as dinucleo-
side monophosphates have been used. These allowed the
discovery, by Box and co-workers (5, 12-14), of oxidative
pathways leading to the degradation of two adjacent

452 Chem. Res. Toxicol., Vol. 15, No. 3, 2002
Douki et al.
degradation of dGpdT and dTpdG through a one electron
oxidation process, as inferred from the formation of
oxazolone as the major product. Altogether, evidence has
been obtained for the possible involvement of a thymine
peroxyl radical-induced oxidation of guanine in the
formation of 8-oxodG/dF tandem lesions.
As a consequence, the formation of a large series of
tandem lesions arising from the decomposition of thymine
hydroperoxides on one hand and the fate of the guanine
radical cation on the other hand might be predicted.
Thymine glycols (dTGly) were chosen as target lesions
of thymine since they have been reported to be major
final stable products arising from the decomposition of
thymine hydroperoxides (19). Tandem lesions carrying
dTGly on one hand and either 8-oxodG or oxazolone on
the other hand were thus studied. dTGly/8-oxodG tandem
lesions were actually found to be produced in very low
yield upon exposure of solutions of dTpdG and dGpdT to
γ-radiations. They were shown to be generated as pri-
mary products, on the basis of their linear dose-course
formation. In contrast, the related oxazolone-containing
tandem lesions studied (dTGly/dZ) were found to be “two
hits” degradation products as inferred from the quadratic
relationship of their formation with increasing doses.
Similar observations were made for the tandem lesions
bearing both an oxazolone and a formamido residue.
It can be concluded from the results presented above
that 8-oxodGpdF and dFp8-oxodG are the major tandem
lesions produced upon exposure of dGpdT and dTpdG,
respectively, to γ-rays in aqueous aerated solution.
Therefore, the diversity of radiation-induced tandem
lesions is not as large as predicted from the mechanism
involving intramolecular electron transfer between a
thymine peroxyl radical and a guanine residue. To
investigate a possible alternative mechanism, experi-
ments were carried out in order to establish the origin
of the oxygen atom at position 8 of the 8-oxodG moiety
in 8-oxodG/dF. Indeed, the overwhelming formation of
8-oxodG/dF may be accounted for by the addition of the
5(6)-hydroxyl-6(5)-peroxyl-5,6-dihydrothymine radical to
the guanine moiety, followed by the rearrangement of the
resulting peroxide (Scheme 3). This alternative mecha-
nism would lead to the direct formation of 8-oxodG,
without formation of oxazolone. Similarly, thymine per-
oxyl radical would be transiently converted into an alkoxy
radical, without formation of hydroperoxides. This might
therefore account for the high yield of formamido rem-
nant. Indeed, 5(6)-hydroxy-6(5)-oxyl-5,6-dihydroythymine
radicals have been shown to be precursors of formamido
residue via a â-scission (19, 37). In contrast, 8-oxoGua/
dTGly are produced in low yield by the reduction of the
alkoxy radical.
F igu r e 9. Repartition of the 16O and 18O atoms in 8-oxodGpdF
produced upon irradiation of dGpdT in the presence of 18O. The
reported values represent the proportion (expressed as %) of
each of the 16O- and 18O-labeled (M + 2) 8-oxoGua and dF
moieties in the 8-oxodGpdF molecules exhibiting pseudo-mo-
lecular ions at m/z ) 505 (M), 507 (M + 2), and 509 (M + 4).
bases after one initial radical event. Indeed, tandem
lesions exhibiting both a 8-oxodG and a formamido
residue arising from guanine on one hand and thymine
or cytosine on the other hand have been identified within
short oligonucleotides and DNA exposed to low LET
radiation in aerated aqueous solutions. Until now, the
accepted mechanism (5, 15) for the formation of 8-oxodG/
dF tandem lesions involved the initial addition of a single
°OH to the C5-C6 double bond of a thymine residue,
followed by oxygen fixation on the resulting radical. This
was mainly inferred from experiments involving oxida-
tion of DNA under various conditions (15). In the present
work, a more direct evidence was provided by the study
of the photolysis of thymine bromohydrins. Indeed,
homolytic cleavage of the C-Br bond of the latter
compounds leads to the formation of the radical arising
from the addition of °OH at the C6 position of thymine.
Under these conditions, 8-oxodGpdF was actually char-
acterized as a UV-induced degradation product of dGp-
BrOHdT. The second key step in the proposed mecha-
nism of formation of 8-oxodG/dF tandem lesion is the
intramolecular electron abstraction from guanine by a
thymine peroxyl radical. The possibility for such a
reaction to occur has been already strongly suggested by
the study of methylperoxyl radicals arising from the
radiolysis of aqueous solutions containing dimethyl sul-
foxide (31). This has received additional support from
experiments using ABAP as a source of alkylperoxyl
radical. The latter oxidizing compound has previously
been used in mutagenesis experiments involving trans-
fected plasmids (32), and guanine was found to be a major
target. This has also been visualized by sequencing
techniques (33, 34). In addition, ABAP-treated DNA was
found to contain a much higher level of 8-oxodG than of
5-hydroxy-2′-deoxycytidine (35). The overwhelming deg-
radation of guanine has also been observed upon treat-
ment of DNA with tert-butoxyl radicals (36). These
results demonstrate that even though peroxyl radicals
are weaker oxidants than reactive species such as hy-
droxyl radicals, they are able to undergo one electron
oxidation of guanine. This selectivity might be accounted
for by the low oxidation potential of guanine with respect
to other DNA bases. This was confirmed in the present
study since ABAP was found to induce a very efficient
Evidence was provided for this alternative mechanism
of formation of 8-oxodG/dF tandem lesions by an experi-
ment involving irradiation of dGpdT in the presence of
18O2. A very efficient labeling of 8-oxodGpdF was observed
(80%), in agreement with a major role played by molec-
ular oxygen in the formation of the tandem lesion. HPLC-
MS/MS analysis allowed to show that either one or two
18O atoms were incorporated. However, the 8-oxoGua
residue was labeled in both cases. This definitively shows
that the oxygen atom of the keto group of 8-oxodG in
8-oxodG/dF is provided by molecular oxygen, likely
through the addition of the thymine peroxyl radical.
Indeed, conversion of the guanine radical cation into
8-oxodG requires hydration and would thus yield unla-

Mechanism of Formation of DNA Tandem Lesions
Chem. Res. Toxicol., Vol. 15, No. 3, 2002 453
Sch em e 3. P r esen tly P r op osed Mech a n ism for th e F or m a tion of 8-oxod G/d F Ta n d em Lesion sa
a
The described pathway only details the reaction occurring upon addition of ˚OH to the C6 position of thymine. The inset shows the
final steps of the reaction upon initial addition of ˚OH to the C5 position of thymine. The key step is the addition of the peroxyl radical
to the C8 position of guanine. The asterisks represent the position of the oxygen atoms arising from the molecular oxygen.
beled 8-oxodG moiety. Involvement of hydroxyl radicals
can be ruled out for similar reasons. The latter point was
emphasized by the observation that the 8-oxodGpdT
single lesion, produced by addition of °OH to dGpdT, was
not labeled in the presence of 18O2. The second 18O atom
present in the M + 4 derivative of 8-oxoGpdF was found
to be located on the formamido residue. However, less
than half of the 18O-labeled 8-oxodGpdF molecules carried
this second stable isotope. This may be accounted for by
the site of the initial addition of the hydroxyl radical to
the C5-C6 thymine double bond. Indeed, formation of
the formamido residue involves the fragmentation of the
pyrimidine ring, with loss of the C5, C4, N3, and C2
atoms, together with their respective substituents. There-
fore, formation of 8-oxodGpdF from a C5-peroxyl radical
would lead to the loss of one of the two 18O atoms arising
from the addition of oxygen. In contrast, involvement of
C6-peroxyl derivative would yield 8-oxodGpdF bearing
two 18O atoms. Interestingly, the abundance of the M +
2 pseudo-molecular ion was slightly higher than that of
the M + 4. This suggests that the amount of tandem
lesions arising from the initial addition of °OH to the C6
position of thymine is higher than that produced by the
occurrence of the latter reaction at C5. This contrasts
with the respective reactivity of the C5 and C6 positions
of thymine to which °OH adds in a 60/35 ratio (3). It
might therefore be proposed that the 6-hydroxy-5-per-
oxyl-5,6-dihydrothymine radical leads to a higher yield
of 8-oxodGpdF tandem lesions than its 5-hydroxy-6-
peroxyl analogue.
tion of guanine. It might therefore be proposed that the
addition of a thymine peroxyl radical to a vicinal guanine
residue is favored by entropic factors. The identification
of the latter mechanism for the formation of 8-oxodG/dF
allows us to restrict the unidentified candidates for the
tandem lesions likely to be found within irradiated DNA.
Indeed, only degradation products arising from the
thymine alkoxy radicals, such as the diastereoisomers of
5-hydroxy-5-methylhydantoin nucleosides, might be pro-
duced in significant yield. This would account for the
results obtained within DNA, showing a likely formation
of 8-oxodG-containing exonucleases-resistant tandem
lesions other than 8-oxodG/dF (15). Indeed, 5-hydroxy-
5-methylhydantoin was found to be resistant to the action
of snake venom and calf spleen phosphodiesterases (38).
Refer en ces
(1) O’Neill, P., and Chapman, P. W. (1985) Potential repair of free
radical adducts of dGMP and dG by a series of reductants. A pulse
radiolysis study. Int. J . Radiat. Biol. 47, 71-80.
(2) Steenken, S. (1989) Purine bases, nucleosides, and nucleotides:
aqueous solution redox chemistry and transformation reactions
of their radical cations and e- and OH adducts. Chem. Rev. 89,
503-520.
(3) Cadet, J ., Berger, M., Douki, T., and Ravanat, J .-L. (1997)
Oxidative damage to DNA: formation, measurement and biologi-
cal significance. Rev. Physiol. Biochem. Pharmacol. 131, 1-87.
(4) Breen, A. P., and Murphy, J . A. (1995) Reactions of oxyl radicals
with DNA. Free Radical Biol. Med. 18, 1033-1077.
(5) Box, H. C., Dubzinski, E. E., Freund, H. G., Evans, M. S., Patrzyc,
H. B., Wallace, J . C., and Maccubin, A. E. (1993) Vicinal lesions
in X-irradiated DNA? Int. J . Radiat. Biol. 64, 261-263.
(6) Box, H. C., Budzinski, E. E., Dawidzik, J . B., Gobey, J . S., and
Freund, H. G. (1997) Free radical-induced tandem base damage
in DNA oligomers. Free Radical Biol. Med. 23, 1021-1030.
(7) Romieu, A., Bellon, S., Gasparutto, D., and Cadet, J . (2000)
Synthesis and UV photolysis of oligodeoxynucleotides that contain
5-(phenylthiomethyl)-2′-deoxyuridine: a specific photolabile pre-
cursor of 5-(2′-deoxyuridilyl)methyl radical. Org. Lett. 2, 1085-
1088.
Altogether, these data strongly support the occurrence
of an intramolecular addition of a thymine peroxyl radical
to vicinal guanine in the formation of 8-oxodGpdF
tandem lesions. Interestingly, alkylperoxyl radicals pro-
duced in the bulk of the solution upon ABAP thermal
decomposition were found to undergo one electron oxida-

454 Chem. Res. Toxicol., Vol. 15, No. 3, 2002
Douki et al.
(8) Bellon, S., Gasparutto, D., Romieu, A., and Cadet, J . (2001)
5-(Phenylthiomethyl)-2′-deoxyuridine as an efficient photoreactive
precursor to generate single and multiple lesions within DNA
fragments. Nucleosides, Nucleotides Nucleic Acids 20, 967-971.
(9) Box, H. C., Patrzyc, H. B., Dawidzik, J . B., Wallace, J . C., Freund,
H. G., Iijima, H., and Budzinski, E. E. (2000) Double base lesions
in DNA X-irradiated in the presence or absence of oxygen. Radiat.
Res. 153, 442-446.
(10) Delatour, T., Douki, T., Gasparutto, D., Brochier, M.-C., and
Cadet, J . (1998) A novel vicinal lesion obtained from the oxidative
photosensitization of TpdG: characterization and mechanistic
aspects. Chem. Res. Toxicol. 11, 1005-1013.
(11) Box, H. C., Budzinski, E. E., Dawidzik, J . B., Wallace, J . C., and
Iijima, H. (1998) Tandem lesions and other products in X-
irradiated DNA oligomers. Radiat. Res. 149, 433-439.
(12) Budzinski, E. E., Dawidzik, J . D., Wallace, J . C., Freund, H. G.,
and Box, H. C. (1995) The radiation chemistry of d(CpGpTpA) in
the presence of oxygen. Radiat. Res. 142, 107-109.
(13) Schroder, E., Budzinski, E. E., Wallace, J . C., Zimbrick, J . D.,
and Box, H. C. (1995) Radiation chemistry of d(ApCpGpT). Int.
J . Radiat. Biol. 68, 509-523.
(14) Box, H. C., Badwidzik, J . B., Wallace, J . C., Hauer, C. R., Stack,
R. F., Rajecki, M. J ., and Budzinski, E. E. (1999) An overall
perspective of X-ray damage to a DNA oligomer mediated by
reactive oxygen species. Radiat. Res. 152, 575-582.
(15) Bourdat, A.-G., Douki, T., Frelon, S., Gasparutto, D., and Cadet,
J . (2000) Tandem base lesions are generated by hydroxyl radicals
within isolated DNA in aerated aqueous solution. J . Am. Chem.
Soc. 122, 4549-4556.
(16) Maccubin, A. E., Iijima, H., Ersing, N., Dawidzik, J . B., Patrzyc,
H. B., Wallace, J . C., Budzinski, E. E., Freund, H. G., and Box,
H. C. (2000) Double-base lesions are produced in DNA by free
radicals. Arch. Biochem. Biophys. 375, 119-123.
(17) Patrzyc, H. B., Dawidzik, J . B., Budzinski, E. E., Iijima, H., and
Box, H. C. (2001) Double lesions are produced in DNA oligomers
by ionizing radiation and by metal-catalyzed H2O2 reactions.
Radiat. Res. 155, 634-636.
Biochem. Biophys. 374, 118-127.
(24) Ravanat, J .-L., Berger, M., Benard, F., Langlois, R., Ouellet, R.,
van Lier, J . E., and Cadet, J . (1992) Phthalocyanine and naph-
thalocyanine photosensitized oxidation of 2′-deoxyguanosine:
distinct type I and type II products. Photochem. Photobiol. 55,
809-814.
(25) Ravanat, J .-L., and Cadet, J . (1995) Reaction of singlet oxygen
with 2′-deoxyguanosine and DNA. Isolation and characterization
of the main oxidation products. Chem. Res. Toxicol. 8, 379-388.
(26) Douki, T., Court, M., Sauvaigo, S., Odin, F., and Cadet, J . (2000)
Formation of the main UV-induced thymine dimeric lesions
within isolated and cellular DNA as measured by HPLC-MS/MS.
J . Biol. Chem. 275, 11678-11685.
(27) Douki, T., and Cadet, J . (2001) Individual determination of the
yield of the main-UV induced dimeric pyrimidine photoproducts
in DNA suggests
a high mutagenicity of CC photolesions.
Biochemistry 40, 2495-2501.
(28) Douki, T., Spinelli, S., Ravanat, J .-L., and Cadet, J . (1999)
Hydroxyl radical-induced degradation of 2′-deoxyguanosine under
reducing conditions. J . Chem. Soc., Perkin Trans. 2, 1875-1880.
(29) Hutchinson, F. (1973) The lesions produced by ultraviolet light
in DNA containing 5-bromouracil. Quantum Rev. Biophys. 6, 201-
246.
(30) Sugiyama, H., Tsutumi, Y., and Saito, I. (1990) Highly sequence
selective photoreaction of 5-bromouracil-containing deoxyhexa-
nucleotides. J . Am. Chem. Soc. 112, 6720-6721.
(31) Milligan, J . R., Ng, J . Y.-Y., Wu, C. C. L., Aguilera, J . A., Ward,
J . F., Kow, Y. W., Wallace, S. S., and Cunningham, R. P. (1996)
Methylperoxyl radicals as intermediates in the damage to DNA
irradiated in aqueous dimethyl sulfoxide with gamma rays.
Radiat. Res. 146, 436-443.
(32) Valentine, M. R., Rodriguez, H., and Termini, J . (1998) Mutagen-
esis by peroxy radical is dominated by transversions at deoxy-
guanosine: Evidence for the lack of involvement of 8-oxo-dG(1)
and/or abasic site formation. Biochemistry 37, 7030-7038.
(33) Harkin, L. A., Butler, L. M., and Burcham, P. C. (1997) Role of
GfT transversions in the mutagenicity of alkylperoxyl radicals:
induction of alkali-labile sites in bacteriophage M13mp19. Chem.
Res. Toxicol. 10, 575-581.
(18) Frelon, S., Douki, T., Ravanat, J .-L., Tornabene, C., and Cadet,
J . (2000) High performance liquid chromatography - tandem
mass spectrometry measurement of radiation-induced base dam-
age to isolated and cellular DNA. Chem. Res. Toxicol. 13, 1002-
1010.
(19) Wagner, R. J ., van Lier, J . E., Berger, M., and Cadet, J . (1994)
Thymidine hydroperoxides: structural assignment, conforma-
tional features and thermal decomposition in water. J . Am. Chem.
Soc. 116, 2235-2242.
(34) Rodriguez, H., Valentine, M. R., Holmquist, G. P., Akman, S. A.,
and Termini, J . (1999) Mapping of peroxyl radical induced damage
on genomic DNA. Biochemistry 38, 16578-16588.
(35) Douki, T., Bretonnie`re, Y., and Cadet, J . (2000) Protection against
radiation-induced degradation of DNA bases by polyamines.
Radiat. Res. 153, 29-35.
(20) Tremblay, S., Douki, T., Cadet, J ., and Wagner, R. (1999)
2′-Deoxycytidine glycols, a missing link in the free radical-
mediated oxidation of DNA. J . Biol. Chem. 274, 20833-20838.
(21) Cadet, J ., Ulrich, J ., and Te´oule, R. (1975) Isomerization and new
specific synthesis of thymine glycol. Tetrahedron 31, 2057-2061.
(22) Cadet, J ., Berger, M., Buchko, G. W., J oshi, P. C., Raoul, S., and
Ravanat, J .-L. (1994) 2,2-Diamino-4-[(3,5-di-O-acetyl-2-deoxy-â-
D-erythro-pentofuranosyl)amino]-5-(2H)-oxazolone: a novel and
predominant radical oxidation product of 3′,5′-di-O-acetyl-2′-
deoxyguanosine. J . Am. Chem. Soc. 116, 7403-7404.
(36) Mahler, H.-C., Schulz, I., Adam, V., Grimm, G. N., Saha-Mo¨ller,
C. R., and Epe, B. (2001) tert-Butoxyl radicals generate mainly
7,8-dihydro-8-oxoguanine in DNA. Mutat. Res. 461, 289-299.
(37) Decarroz, C., Wagner, J . R., van Lier, J . E., Murali, K. C., Riesz,
P., and Cadet, J . (1986) Sensitized photooxidation of thymidine
by 2-methyl-1,4-naphthoquinone. Characterization of stable prod-
ucts. Int. J . Radiat. Biol. 50, 491-505.
(38) Gasparutto, D., Ait-Abbas, M., J aquinod, M., Boiteux, S., and
Cadet, J . (2000) Repair and coding properties of 5-hydroxy-5-
methylhydantoin nucleosides inserted into DNA oligomers. Chem.
Res. Toxicol. 13, 575-584.
(23) Ravanat, J .-L., Remaud, G., and Cadet, J . (2000) Measurement
of the main photooxidation products of 2′-deoxyguanosine using
chromatographic methods coupled to mass spectrometry. Arch.
TX0155909