Reactions of estradiol-2,3-quinone with deoxyribonucleosides: Possible insights in the reactivity of estrogen quinones with DNA

Article abstract of DOI:10.1021/tx015561y

Estrogen 2,3- and 3,4-quinones are reactive species toward nucleophiles and Michael acceptors. As such, they can bind to DNA and induce cellular damages. As an alkylation model, reactions of estradiol-2,3-quinone with deoxyribonucleosides were previously studied by mass spectrometry. In this work, estrogen-deoxyribonucleoside adducts were synthesized by reaction of 17β-estradiol-2,3-quinone with deoxyguanosine or deoxyadenosine and analyzed by NMR and LC-MSⁿ in order to determine the structure and the stereochemistry of the resulting covalent adducts. Although estradiol- and estrone-2,3-quinones were previously thought to give mainly stable adducts, identification of depurinating adducts with both nucleosides, i.e., 2-OHE₂-6(α,β)-N7Gua and 2-OHE₂-6(α,β)-N7Ade, was unambiguously obtained. This is of particular interest since depurinating adducts are generated from DNA, and therefore, their amount should be correlated to the parallel formation of apurinic sites, which might play an important role in the cancer initiation process. Besides, a byproduct, i.e., 2-hydroxy-11-oxoestradiol, corresponding to an unstable alkylation product of 2-hydroxyestradiol has been unambiguously identified and is indicative of a plausible addition process at the C9 position of catechol estrogens. The synthetic adducts will be useful as reference compounds to further elucidate the structure of adducts formed by reaction of estrogen metabolites with DNA or oligonucleotides.

Full text of DOI:10.1021/tx015561y

754
Chem. Res. Toxicol. 2002, 15, 754-764
Rea ction s of Estr a d iol-2,3-qu in on e w ith
Deoxyr ibon u cleosid es: P ossible In sigh ts in th e
Rea ctivity of Estr ogen Qu in on es w ith DNA
Odile Convert,*,† Christel Van Aerden,† Laurent Debrauwer,‡ Estelle Rathahao,‡
Huguette Molines,§ Franc¸oise Fournier,† J ean-Claude Tabet,† and Alain Paris*,‡
Laboratoire de Chimie Structurale Organique et Biologique, CNRS UMR7613, UPMC,
4 place J ussieu, 75252 Paris Cedex 05, France, Laboratoire des Xe´nobiotiques, INRA, BP 3,
31931 Toulouse, Cedex 09, France, and Laboratoire de Chimie des He´te´rocycles, CNRS UMR7611,
UPMC, 4 place J ussieu, 75252 Paris, Cedex 05, France
Received August 31, 2001
Estrogen 2,3- and 3,4-quinones are reactive species toward nucleophiles and Michael
acceptors. As such, they can bind to DNA and induce cellular damages. As an alkylation model,
reactions of estradiol-2,3-quinone with deoxyribonucleosides were previously studied by mass
spectrometry. In this work, estrogen-deoxyribonucleoside adducts were synthesized by reaction
of 17â-estradiol-2,3-quinone with deoxyguanosine or deoxyadenosine and analyzed by NMR
and LC-MSn in order to determine the structure and the stereochemistry of the resulting
covalent adducts. Although estradiol- and estrone-2,3-quinones were previously thought to
give mainly stable adducts, identification of depurinating adducts with both nucleosides, i.e.,
2-OHE2-6(R,â)-N7Gua and 2-OHE2-6(R,â)-N7Ade, was unambiguously obtained. This is of
particular interest since depurinating adducts are generated from DNA, and therefore, their
amount should be correlated to the parallel formation of apurinic sites, which might play an
important role in the cancer initiation process. Besides, a byproduct, i.e., 2-hydroxy-11-oxo-
estradiol, corresponding to an unstable alkylation product of 2-hydroxyestradiol has been
unambiguously identified and is indicative of a plausible addition process at the C9 position
of catechol estrogens. The synthetic adducts will be useful as reference compounds to further
elucidate the structure of adducts formed by reaction of estrogen metabolites with DNA or
oligonucleotides.
in vivo experiments carried out from mammary tissues
of rats treated with 4-hydroxy-estradiol (10, 11). These
authors have postulated that 4-hydroxy-catechol estro-
gens can give rise to depurinating adducts whereas
2-hydroxy-catechol estrogens lead only to stable adducts.
Therefore, this difference could explain the difference in
carcinogenic potency of these two types of catechol
estrogens. It was also shown that different types of
covalent adducts could be obtained according to the
reaction conditions and to the o-quinones reactivity. The
structures established for these adducts are consistent
with a DNA alkylation process involving a Michael
addition mechanism as predicted from the chemical
model studies. Thus, these authors have determined by
NMR that the reaction of estrone-2,3-quinone (2-OHE1-
Q)1 on deoxyguanosine dG essentially consists of a
regioselective attack of the exocyclic N2dG atom on the
prochiral C6 position of the estrogen skeleton (10) and
In tr od u ction
It is now established that estrogens are involved in
carcinogenesis either as hormones or as reactive species
after oxidation (1-4). Although the hormonal mecha-
nisms are known to mediate cell proliferation, growing
attention has been given in the past few years to tumor
initiation in relation to the modification of macrobiomol-
ecules, particularly DNA. Estrogens can undergo several
enzymatic or chemical oxidation processes, which may
lead to formation of highly electrophilic species, i.e.
o-quinones substrates. If these o-quinones are not deac-
tivated by conjugation to glutathione via glutathione-S-
transferases, they can bind covalently to DNA and so
induce the critical initiation step of chemical carcinogen-
esis. All the work on this topic has been recently reviewed
(5-7). A particular emphasis was given on reactivity of
4-hydroxy-catechol vs 2-hydroxy-catechol estrogens since
the first species have been reported to be carcinogenic
in male Syrian golden hamsters whereas the latter are
not (8, 9). These results were corroborated by those
obtained more recently by the group of Cavalieri using
1
Abbreviations: dA, deoxyadenosine; dG, deoxyguanosine; dC,
deoxycytidine; 2-OHE2, 2-hydroxyestradiol; 2-OHE1, 2-hydroxyestrone;
2-OHE2-Q, estradiol-2,3-quinone; 2-OHE1-Q, estrone-2,3-quinone;
2-OHE2-QM, 2-hydroxyestradiol quinone methide; 2-OHE1-QM, 2-hy-
droxyestrone quinone methide; DMSO, dimethyl sulfoxide; NOE,
nuclear Overhauser effect; TOCSY, total correlation spectroscopy;
HMQC, heteronuclear multiple quantum coherence; HMBC, hetero-
nuclear multiple bond correlation; CAD, collisionally activated decom-
position; LC-MS2, LC-MS/MS, liquid chromatography-tandem mass
spectrometry; LC-MS3, LC-MS/MS/MS, liquid chromatography-mass
spectrometry “to the third”.
* To whom correspondence should be addressed. (A.P.) Phone: 33
561 285 394. Fax: 33 561 285 244. E-mail: aparis@toulouse.inra.fr.
(O.C.) Phone: 33 144 273 115. Fax: 33 144 273 843. E-mail: oconvert@
ccr.jussieu.fr.
† Laboratoire de Chimie Structurale Organique et Biologique.
‡ Laboratoire des Xe´nobiotiques.
§ Laboratoire de Chimie des He´te´rocycles.
10.1021/tx015561y CCC: $22.00 © 2002 American Chemical Society
Published on Web 04/26/2002

Catechol Estrogen-Deoxynucleoside Adducts
Chem. Res. Toxicol., Vol. 15, No. 5, 2002 755
experiments were achieved by isolation of a fragment ion from
an MS2 experiment into the trap and excitation of this ion in
an MS “to the third” step. All MS and MSn analyses were
performed in the normal scan mode under automatic gain
control conditions. The mass spectrometer was interfaced to a
Thermo Separation P4000 gradient HPLC system equipped with
a Rheodyne 7725i injector and an UV1000 UV detector set at
280 nm. The samples were analyzed on a C18 Ultrabase
(250 × 2 mm) column from Life Sciences International (Eragny,
France) at a flow rate of 0.2 mL/min. The mobile phases are
the same as described above to perform semipreparative HPLC.
A linear gradient was used with an increase from 0 to 25% B
from 0 to 5 min, 25 to 50% B from 5 to 30 min, and 50 to 100%
B from 30 to 40 min, then an isocratic step at 100% B was used
from 40 to 60 min.
NMR. NMR spectra were recorded on Bruker AM500 or
Avance DMX500 spectrometers, operating at a 1H resonance
frequency of 500 MHz. Samples were dissolved in 500 µL of
DMSO-d6 or CD3OD, at concentrations which may vary from
0.2 to 1 mM according to the amount collected by HPLC.
Chemical shifts are given in ppm/TMS by using the residual
solvent signals as internal references: 2.5 ppm for 1H and 39.5
ppm for 13C in DMSO, 3.31 ppm for 1H and 49.0 ppm for 13C in
CD3OD. Unless otherwise indicated, all 1H two-dimensional
experiments were acquired in the phase-sensitive mode with
the time-proportional phase incrementation of the initial pulse
(16). A relaxation delay of 1.2 to 2 s was used. 256-512 t1
increments and 2048 complex data points in t2 were recorded
for a spectral width of 5000 Hz in the two dimensions. Prior to
Fourier transform, the signal was multiplied by a shifted square
sine-bell window function. For TOCSY experiments (17), a
MLEV17 mixing scheme of different lengths (from 10 to 80 ms)
with a 10 kHz spin-locking field strength was used in order to
define direct and long-range correlations. In CD3OD solutions,
signal OH suppression was achieved by selective irradiation of
this signal during the relaxation delay. 2D heteronuclear
experiments were performed by using the standard Bruker
software. In 13C-1H HMQC and HMBC (18, 19) sequences,
delays were optimized for coupling constants around 140 and
10 Hz, respectively. One-dimensional NOE experiments were
recorded in difference mode by subtracting one spectrum with
irradiation on resonance and another one with irradiation off
resonance; in this experiment, a relaxation delay of 3 s was used.
gives rise to a mixture of diastereomer adducts, 2-OHE1-
6(R,â)-N2dG.
In previous works, we showed by using LC-MS (12, 13)
that the 2-catechol estrogen quinone, estradiol-2,3-
quinone (2-OHE2-Q), reacts with deoxyribonucleosides
dG, dA, and dC to give rise to different covalent adducts.
Liquid chromatography-tandem mass spectrometry (LC-
MS2) as well as liquid chromatography-mass spectrom-
etry “to the third” (LC-MS3) experiments carried out in
an ion trap mass spectrometer allowed us to analyze the
fragmentation patterns and discriminate the different
isomeric adducts. The present work concerns MS and
NMR characterizations of reaction products obtained by
reaction of 2-OHE2-Q with dG and dA to support the
previous propositions and to determine the stereochem-
istry of all the adducts formed. The goal is to provide
insights into the possible modes of covalent binding for
the catechol estrogens to DNA.
Ma ter ia ls a n d Meth od s
Ca u tion : Catechol estrogen quinones are hazardous chemicals
and should be handled carefully in accordance with statutory
procedures and health guidelines.
Ch em ica ls. All chemicals were purchased from Aldrich
(Saint Quentin Fallavier, France). LC-MS analyses used HPLC-
grade solvents from Scharlau (Barcelona, Spain) and water from
a Milli-Q system (Millipore, Saint Quentin en Yvelines, France).
Rea ction s of Estr a d iol-2,3-qu in on e w ith Deoxyr ibo-
n u cleosid es. Syntheses have been carried out according to
previously published procedures. 2-Hydroxyestradiol (2-OHE2)
was prepared by the method described by Gelbke et al. (14) and
2-OHE2-Q was obtained by oxidation of the corresponding
catechol with activated manganese dioxide as described by Stack
et al. (10), according to Abul-Hajj’s procedure (15). Addition
reactions between 2-OHE2-Q and deoxynucleosides, i.e., deoxy-
guanosine (dG), and deoxyadenosine (dA), were performed as
described previously (10). The atom numbering for the different
adducts corresponds to the one used in previous studies (10).
Sem ip r ep a r a tive HP LC. Adducts obtained from the reac-
tion between 2-OHE2-Q and dG or dA were quantitatively
purified by using semipreparative chromatographies. HPLC
were performed on a PU4100 gradient pump equipped with a
Rheodyne 7500 injector and connected to a PU4110 UV detector
set at 280 nm. The reaction products were purified on a C18
Ultrabase (250 × 7.5 mm, 5 µm) column from SFCC (Saint
Antoine, France) at a flow rate of 2 mL/min. The mobile phases
consisted in mixtures of methanol:water:acetic acid in the
following proportions: 10:90:0.2 for the solvent A and 90:10:
0.2 for the solvent B. Analyses were achieved at 35 °C using
the following gradient: a linear gradient was used with an
increase from 0 to 25% B on the first 5 min, 25 to 50% B from
5 to 35 min, then 50 to 70% B from 45 to 60 min, and 70 to
100% B from 60 to 61 min, and finally an isocratic step at 100%
B was maintained from 61 to 75 min. Concerning the yield of
adducts, UV recording at 280 nm was used to quantify the
amount purified in semipreparative conditions, making assump-
tions that molar extinction coefficients are rather comparable
between them but also to this of 2-OHE2.
Resu lts
Rea ction of 2-OHE2-Q w ith d G. LC-MS analysis has
revealed the presence of several adducts in the reaction
mixtures of 2-OHE2-Q with dG (Figure 1). Three stable
adducts G3, G4, and G5 (adducts with deoxyribose)
displaying MH+ ions at m/z 554 and three depurinating
adducts G1, G2, and G6 (adducts without deoxyribose) at
m/z 438 were detected. From the fragmentation patterns
observed in LC-MS2 and LC-MS3 experiments, it was
deduced that the C6 position of the catechol estrogen was
involved in the linkage with dG for adducts G1, G2, G3,
and G4 as described in Scheme 1. Furthermore, the
diagnostic charge-remote fragmentation process occur-
ring for adducts G5 and G6 suggested a base addition on
the aromatic ring of the catechol estrogen (C1 or C4
position).
However, to assess the chemical reactivity of 2-OHE2-Q
toward deoxyribonucleosides, it was necessary to eluci-
date by NMR the structures proposed in previous studies
and to determine their stereochemistry. Therefore, large-
scale preparations of authentic adducts were performed
and repeated purifications on semipreparative reversed
phase high-performance liquid chromatography were
carried out on reaction mixtures to collect fractions in a
sufficient amount for structural analyses.
Liqu id Ch r om a togr a ph y-Electr ospr a y Ma ss Spectr om -
etr y. Tandem mass spectrometry experiments were carried out
on a Finnigan LCQ (Thermo Quest, Les Ulis, France) quadru-
pole ion trap mass spectrometer equipped with electrospray
ionization source. Ionization was achieved by using a typical
spray needle voltage of 5.2 kV, toward the heated transfer
capillary held at 220 °C and 5 V as the counter electrode. Helium
was used as damping and relaxation gas as well as collision
target gas for sequential MSn experiments. MS2 experiments
were carried out by isolation and subsequent excitation of the
selected parent ion into the ion trap device. Sequential MS3

756 Chem. Res. Toxicol., Vol. 15, No. 5, 2002
Convert et al.
F igu r e 1. LC-MS2 analysis of the reaction mixture of 2-OHE2-Q with dG. Recording of ions at m/z 554 and m/z 438 reveals
chromatographic separation of complete and deglycosylated adducts, respectively.
Sch em e 1. P r od u cts Id en tified in th e Rea ction Mixtu r e of 2-OHE2-Q w ith DG by LC-MSn (*), by LC-MSn ,
a n d NMR (**), or by NMR (***)a
a
Amounts of the different adducts purified in semi-preparative conditions are given in brackets.
Semipreparative HPLC purification of the reaction
mixture of 2-OHE2-Q with dG allows isolation of five
fractions, F1 to F5 (Figure 2), which were submitted to
the fraction F3 by semipreparative HPLC (G7), has been
achieved by NMR.
2-OHE2-6r-N7Gu a (G1) a n d 2-OHE2-6â-N7Gu a (G2).
The F1 fraction, which accounts for 9% of the initial
amount of 2-OHE2-Q used in the reaction, was identified
by LC-MS as the G1 adduct (Figure 1). Its 1H NMR
spectrum in CD3OD makes evident the absence of signals
in the region from 4.6 to 3.5 ppm and consequently a loss
of the deoxyribose moiety. This observation is consistent
with the mass spectrometric results which provided an
MH+ ion observed at m/z 438 for this adduct correspond-
ing to a species in which the deoxyribose moiety is absent.
This allows us to confirm a structure of depurinating
adduct for G1, corresponding to the addition of the base
1
NMR analyses. Complete H and 13C NMR studies were
achieved on the F1, F4, and F5 fractions isolated as pure
1
samples. H NMR analysis was conducted on the F2 and
F3 fractions, which are mixtures of two and three main
components, respectively. However, we were not able to
isolate any fraction containing sufficient amount of
adducts G5 or G6 to perform NMR studies; consequently
they were characterized only by LC-MSn from their
characteristic fragmentation pattern. Adducts G3 and G4
were characterized only by LC-MSn, but the structure of
a structurally related compound, which was purified in

Catechol Estrogen-Deoxynucleoside Adducts
Chem. Res. Toxicol., Vol. 15, No. 5, 2002 757
the nature of this carbon provides evidence for determin-
ing the reaction site as the C6 position. Therefore, the
proton at 5.68 ppm (doublet of doublets with coupling
constants of 11.5 and 6.1 Hz) is assigned to a H6 proton.
Effectively, it presents characteristic TOCSY correlations
with aliphatic protons corresponding to H7, H8, H9, H14,
H15, and H16 protons of the steroid skeleton (Table 1).
Most of the 13C signals of this adduct (Table 1) are very
similar to the corresponding ones in the base or in the
catechol estrogen 2-OHE2, except for the carbons C6 and
C7 that are involved in the base addition.
Since the best way to assign unequivocally the struc-
tures derives from studies of nuclear Overhauser effects
between various neighboring protons (20), NOE experi-
ments were conducted to establish the site of linkage to
the guanine ring, as previously used in the structure
determination of similar thiol or glutathione conjugates
(21, 22). Characteristic NOEs are found between the
proton H6 at 5.68 ppm and the two protons at 6.10 and
7.58 ppm in one part, between the proton at 6.80 ppm
and one aliphatic proton at 2.30 ppm in another part.
These observations permitted assignment of the proton
at 2.30 ppm to the H11R proton and the three aromatic
protons as follows: H1 proton at 6.80 ppm, H4 proton at
6.10 ppm and H8-Gua at 7.58 ppm. Furthermore, the
occurrence of a NOE between the H6 and H8-Gua protons
indicates a spatial closeness for these protons and
consequently implicates a base linkage at the N7 atom
of dG that results in the loss of the deoxyribose due to a
destabilization of the glycosidic bond (23, 24). This
proposal is consistent with the absence of modification
in the H6 signal multiplicity when the spectrum is
recorded in DMSO (Figure 3B) instead of CD3OD. Be-
sides, the occurrence in the spectrum recorded in DMSO
of a broad signal at 6.50 ppm corresponding to two
protons excludes the possibility of a steroid-base coupling
at the NH2 exocyclic site of the dG base. Therefore, all
these results are consistent with the 2-OHE2-6-N7Gua
structure for the G1 depurinating adduct. Since the
protons located at C6 are prochiral, an attack of the dG
nucleophile at this position gives rise to the formation of
two diastereomers. The analysis of the pattern of coupling
constants for the H6 signals in both the diastereomer
adducts should allow us to assign their respective stereo-
chemistries as shown below.
F igu r e 2. Semipreparative purification of the reaction mixture
of 2-OHE2-Q with dG. Yield of adducts was obtained from peak
area integration with UV recording at 280 nm assuming that
molar extinction coefficients of the different adducts are similar.
The F2 fraction represented the same amount of
reaction products as F1 (9%). Its 1H NMR spectrum
recorded in CD3OD (Figure 4A) shows a clear splitting
of signals in the aromatic region (at least six signals from
8 to 6 ppm), in the H6 region (5.68 and 5.55 ppm) and in
the methyl region (0.82 and 0.71 ppm). From the relative
intensities of the two sets of peaks, it is possible to assign
the proportion of the two species present in this fraction
as 20/80. The minor signals correspond exactly to those
of the G1 adduct. Taking into account the analogies in
TOCSY correlations for both signals at 5.68 and 5.55
ppm, it is assumed that the two G1 and G2 diastereomer
adducts are present in the F2 fraction. This is consistent
with mass spectra, which display MH+ ions at m/z 438
and current HPLC retention times of 26.6 and 27.9 min
corresponding to the G1 and G2 adducts. Therefore, the
signal at 5.55 ppm appearing as a doublet (4.5 Hz) of
large signals is assigned to the proton H6 of the G2
adduct. As above, the multiplicity of this signal does not
change in DMSO (Figure 4B). Despite the chemical
instability of this fraction, a comparison with the F1
F igu r e 3. Regions 8-5 ppm of 1H NMR spectra of the 2-OHE2-
6R-N7Gua depurinating adduct (G1) in CD3OD (A) and in DMSO
(B).
moiety of deoxyguanosine to one molecule of 2-OHE2.
Taking into account the presence of one aromatic proton
(H8-Gua) in the guanine ring, the observation of three
singlets at 6.10, 6.80, and 7.58 ppm in the 1H NMR
spectrum (Figure 3A) rules out the possibility of a
reaction of nucleoside at one aromatic ring site (C1 or
C4) of the steroid skeleton. Furthermore, this spectrum
shows at 5.68 ppm one proton correlated in the HMQC
spectrum to a methine group at 51.9 ppm. The low-field
values of these 1H and 13C chemical shifts suggest a base-
steroid coupling at a benzylic position (C6 or C9), and

758 Chem. Res. Toxicol., Vol. 15, No. 5, 2002
Convert et al.
Ta ble 1. 13C a n d 1H Ch em ica l Sh ifts of th e G1 Ad d u ct (2-OHE2-6r-N7Gu a ), of th e G2 Ad d u ct (2-OHE2-6â-N7Gu a ) a n d of th e
G7 Ad d u ct (2-OHE2-6â-N2Gu a ) Id en tified in th e Rea ction Mixtu r e of 2-OHE2-Q w ith d Ga
δ
13C
(ppm)
G1
DMSO
δ 1H
(ppm)
G1
DMSO
G2
DMSO
G1
CD3OD
G2
CD3OD
G7
CD3OD
C1
C2
C3
C4
C5
C6
C7
C8
112.3
143.2
144.4
112.8
125.6
51.9
34.5
38.4
43.0
H1
6.68
5.96
6.80
6.80
6.10
6.88
6.43
6.88
6.55
H4
6.15
H
H
6R/H6â
7R/H7â
- /5.47
1.85/2.05
1.55
5.41/ -
1.72/1.92
1.49
- /5.68
1.82/2.24
1.68
5.55/ -
1.75/2.10
1.45
5.86/ -
1.80/2.16
1.34
H8â
H9R
C9
2.35
2.12
2.43
2.18
C10
C11
C12
C13
C14
C15
C16
C17
C18
C8-Gua
131.1
25.5
36.1
42.0
48.5
22.2
29.5
79.5
11.1
H
H
11R/H11â
12R/H12â
2.17/1.35
1.20/1.88
2.22/1.52
1.22/1.90
2.30/1.50
1.30/2.00
2.34/1.57
1.32/1.97
2.34/1.54
H14R
H
H
H17R
CH3
1.20
1.19
1.32
1.23
1.25
15R/H15â
16R-H16â
1.55/1.22
1.88/1.38
3.55
0.69
7.55
1.45/0.96
1.85/1.35
3.50
0.62
7.00
1.68/1.35
2.03/1.50
3.65
0.82
7.58
1.55/1.05
1.95/1.42
3.63
0.71
7.03
1.67/1.36
2.03/1.52
3.67
0.67
7.12
135.6
H8-Gua
a
13C and 1H chemical shifts are given in ppm/TMS by using the residual solvent signal as internal reference: 2.5 and 39.5 ppm,
respectively, in DMSO, 3.31 and 49.0 ppm, respectively, in CD3OD.
previously proposed from mass spectrometry results and
allows us to determine the stereochemistry of these
depurinating adducts. It is worth noting the existence of
large differences in chemical shifts of the aromatic, the
H6 protons and the 18-methyl groups observed between
the two diastereomers, due to ring current effects from
the base and, probably, conformational change in the B
cycle. The chemical shift values complement coupling
constants as an argument to give the stereochemistry of
new adducts.
2-OHE2-6(r,â)-N2d G (G3/G4). Several attempts have
been made for the purification of G3 and G4 by semi-
preparative HPLC. Unfortunately, these adducts could
never been obtained in sufficient purified amounts for
NMR analysis, very likely owing to degradation occurring
during the purification process, leading in particular to
the related G7 adduct (see further in the text). Thus, the
only structural information available was generated by
mass spectrometry from freshly prepared adduct mix-
tures. As indicated in Figure 1, both G3 and G4 display
MH+ ions at m/z 554, indicating that they are stable
adducts. As previously reported (12), the MS/MS analysis
of both the MH+ ions of G3 and G4 resulted in identical
product ion spectra, displaying m/z 438, m/z 287, and m/z
268 as the most abundant product ions. The m/z 438 is
formed by loss of deoxyribose with proton transfer from
the sugar to the guanine moiety of the adduct, whereas
the m/z 287 and m/z 268 ions are the result of the
cleavage of the steroid-base bond. The CAD mass spectra
obtained from MS3 on the m/z 438 ion isolated from the
decomposition of the previous m/z 554 MH+ ion also
displayed two complementary ions (m/z 287 and 152)
arising from the cleavage of the steroid-base linkage.
These fragmentation patterns are known to characterize
adducts with a base attachment on the C6 atom of the
steroid (10-13). In absence of NMR information, and as
the CAD spectra obtained from MS2 and MS3 experi-
ments were identical for G3 and G4, it may be concluded
that they correspond to diastereoisomeric adducts in
which dG is linked to the C6 atom of the steroid via its
N2 exocyclic atom. In this case, the nucleophilic attack
does not induce the destabilization of the glycosidic bond
of the nucleoside. Thus, 2-OHE2-6R-N2dG and 2-OHE2-
F igu r e 4. Regions 8-5 ppm of 1H NMR spectra of the fraction
F2 in CD3OD (A) and in DMSO (B): main components are 20%
of 2-OHE2-6R-N7Gua (G1) and 80% of 2-OHE2-6â-N7Gua (G2)
depurinating adducts.
fraction allowed us to assign all the protons of this G2
adduct in DMSO and in CD3OD (Table 1).
In these compounds, a half-chair conformation is
usually assumed for the B ring (25) of the steroid
skeleton, and the coupling constants of an H6 proton with
the two H7 protons are largely dependent on its position,
R or â (26). Given the coupling constants values observed
for the H6 proton, 11.5 and 6.1 Hz for G1, 4.5 and less
than 1 Hz for G2, it is possible to determine the stereo-
chemistry of these adducts corresponding to a base-
steroid linkage by the N7dG atom on the face R of the
steroid for the G1 adduct, 2-OHE2-6R-N7Gua and on the
face â of the steroid for the G2 adduct, 2-OHE2-6â-N7Gua
(Scheme 1).
At this stage, the NMR analysis of the F1 and F2
fractions provides a clear confirmation of the structures

Catechol Estrogen-Deoxynucleoside Adducts
Chem. Res. Toxicol., Vol. 15, No. 5, 2002 759
Ta ble 2. 13C a n d 1H Ch em ica l Sh ifts in CD3OD of th e P r od u cts F 4 (2-Hyd r oxy-6-oxo-estr a d iol) a n d
a
F 5 (2-Hyd r oxy-11-oxo-estr a d iol) in Com p a r ison to th e Ca tech ol Estr ogen 2-OHE2
δ
13C
(ppm)
δ 1H
2-OHE2
F4
F5
(ppm)
2-OHE2
6.70
F4
6.84
F5
6.30
C1
C2
C3
C4
C5
C6
C7
C8
113.4
143.8
143.8
116.5
128.9
29.9
28.6
40.4
45.4
128.9
27.7
38.0
44.3
51.3
24.0
30.7
82.5
11.7
112.7
153.2
145.1
114.3
126.0
199.8
44.6
41.9
44.2
143.3
26.7
37.5
-
114.9
146.9
145.0
116.8
126.4
24.6
21.0
43.5
43.5
129.2
213.7
51.3
48.0
43.4
23.9
30.8
80.4
12.0
H1
H4
6.40
7.38
6.60
H
H
6R/H6â
7R/H7â
H8â
H9R
2.70/2.70
1.18/1.78
1.28
2.60/2.80
1.75/2.18
2.05
2.19/2.55
1.87
2.44
C9
2.02
2.18
C10
C11
C12
C13
C14
C15
C16
C17
C18
H
H
11R/H11â
12R/H12â
2.15/1.35
1.18/1.88
2.32/1.57
1.35/2.01
2.18/2.42
51.1
23.7
30.6
82.2
11.4
H14R
15R/H15â
16R/H16â
H17R
CH3
1.10
1.35
1.90
H
H
1.60/1.28
1.95/1.45
3.65
1.68/1.35
2.06/1.57
3.68
1.75/1.43
2.10/1.55
3.66
0.77
0.78
0.75
a
13C and 1H chemical shifts are given in ppm/TMS by using the residual solvent signal at 3.31 and 49.0 ppm, respectively, as internal
reference.
6â-N2dG are proposed as structures of the adducts G3
and G4.
of this function on the steroid skeleton has been deduced
from a complete analysis by 2D homo- or heteronuclear
spectra (Table 2). In particular, the long-range correla-
tions observed in the HMBC spectra between the CO
signal at 199.8 ppm and the H4, H7 protons for F4,
between the CO signal at 213.7 ppm and the H11 protons
for F5 have allowed us to characterize these compounds
as the 2-hydroxy-6-oxo-estradiol for the F4 fraction and
the 2-hydroxy-11-oxo-estradiol for the F5 one. On the
basis of the observed coupling constants, the stereochem-
istry is assumed to be preserved in the C8, C9, and C14
positions of the steroid skeleton for both compounds.
The positive electrospray mass spectra of F4 and F5
display respectively a peak at m/z 303 for [M + H]+ and
a [M - H]+ peak at m/z 301 (data not shown). As expected
for F4, the direct protonation of the carbonyl group takes
place at C6 (Scheme 2). Conversely, in the case of F5,
[M - H]+ ions observed at m/z 301 constitute the main
species representing the sample molecules. This can be
explained considering the loss of hydride from both the
benzylic and enolic sites during the H+ approach as
reported in Scheme 2, the resulting [M - H]+ species
being stabilized by conjugation.
2-OHE2-6â-N2Gu a (G7). The yield of adduct in the F3
fraction represented nearly 5% of the amount of 2-OHE2-Q
used to perform the reaction. Despite the complexity of
the 1H NMR spectrum seen for this fraction, a comparison
of the two regions 8.0-5.0 and 1.0-0.6 ppm with the
corresponding ones for the F1 and F2 fractions makes
evident the presence of at least three components, namely
the G1, G2 adducts and a third compound. Taking into
account the values of the 1H chemical shifts and the mass
spectra displaying MH+ ions at m/z 438, a depurinating
adduct may be proposed for this new product, called G7.
Its NMR characteristics are three aromatic protons, a
doublet (4.0 Hz) of large signals at 5.86 ppm presenting
TOCSY correlations observed for a H6 proton and a
methyl signal at 0.67 ppm (Table 1). On the basis of the
H6 proton multiplicity, it is suggested that this additional
adduct G7 might correspond to a base attachment on the
â face of the steroid by the 2-NH2 exocyclic group.
Surprisingly, although it represents about 5% of the
amount of 2-OHE2-Q used in the reaction, this depuri-
nating adduct, 2-OHE2-6â-N2Gua, was not detected by
LC-MSn at a retention time that should correspond to
the F3 peak one. Therefore, it would be secondary formed
by a process involving the deoxyribose hydrolysis of the
corresponding stable adducts G3 or G4 unambiguously
characterized by LC-MSn only. This hydrolytic step might
be due probably to the presence of acetic acid in the
HPLC solvents used in the semipreparative HPLC pro-
cedures.
2-OHE2-1(4)-N2d G (G5). This adduct displaying MH+
ions at m/z 554 (Figure 1) results from the addition of
dG on the A ring of 2-OHE2 without loss of deoxyribose.
Accordingly, the linkage of dG must occur via its exocyclic
N2 atom, the only nucleophilic site of dG that does not
induce destabilization of the sugar-base bond during the
addition process on the steroid. The CAD spectrum of the
m/z 554 MH+ ion displays a unique product ion at m/z
438 (data not shown), formed by the loss of the sugar
moiety of the molecule as previously described (12).
Sequential MS3 experiments carried out on the m/z 438
fragment ion from G5 leads to the CAD spectrum pre-
sented in Figure 5. This spectrum displays major char-
acteristic fragment ions at m/z 324, 312, 298, 286, and
272. The formation of these fragment ions can be at-
tributed to charge-remote fragmentation processes, oc-
curring from species with a highly stable charged site
and giving rise to thermal decompositions in which the
positive charge is not involved (27-29). These processes
are frequently observed and have been extensively
studied on various molecules including steroids and bile
2-Hyd r oxy-6-oxo-estr a d iol (F4) a n d 2-Hyd r oxy-11-
oxo-estr a d iol (F 5). The fractions F4 and F5, accounting
respectively for 12 and 1% of the amount of 2-OHE2-Q
used in the synthesis of adducts, correspond to pure
compounds presenting only two aromatic protons in the
1H NMR spectra. Furthermore, the comparison of their
13C NMR spectra with those of the starting material
2-OHE2 (m/z 289) makes evident the presence of ad-
ditional signals at 199.8 and 213.7 ppm for F4 and F5,
respectively, as well as the disappearance of a CH2 group
in the region 20-30 ppm. These observations suggest
that in both cases a methylene group in 2-OHE2 has been
chemically modified into a carbonyl group. The position

760 Chem. Res. Toxicol., Vol. 15, No. 5, 2002
Convert et al.
Sch em e 2. P r op osed Mech a n ism s for th e F or m a tion of MH+ Ion s for F 4 a n d [M - H]+ Ion s for F 5 Usin g
P ositive Electr osp r a y Ion iza tion
acids (30, 31). The formation of the m/z 324 ion is
initiated by the steroid C ring opening due to a hydrogen
migration from the C9 to the C12 atom of the steroid and
followed by the cleavage of the C11-C12 bond with
hydrogen rearrangement resulting in a methyl-naphtha-
lene like structure as indicated in Scheme 3A. The m/z
312 ion originates in the charge-remote cleavages of the
C9-C11 and C8-C14 bonds (Scheme 3B) according to a
concerted mechanism as previously described (10). The
m/z 298 ion can be produced by a hydrogen migration
from the C9 to the C12 position of the steroid, leading to
the C ring opening, followed by cleavage of the C6-C7
and C8-C9 bonds as represented in Scheme 3C. This
constitutes an alternative to the mechanism previously
proposed for the formation of this ion from 4-OHE1-1-
N7Gua (10). The m/z 286 ion from 2-OHE2-1(4)-N7Gua
very likely possesses the same structure as described for
4-OHE1-1-N7Gua (10): this confirms that this ion is not
related to a radical cation form of the steroid. A mech-
anism for the formation of this ion involves the cleavage
of the C7-C8 and C9-C10 bonds of the steroid B ring
and is accompanied by hydrogen migration as presented
in Scheme 3D. Finally, the formation of the m/z 272 ion
can be obtained from a hydrogen migration from the C8
to the C9 position of 2-OHE2 resulting in the cleavage of
the C9-C10 and C6-C7 bonds (Scheme 3E). Note that
in this case the hydrogen migration process should affect
the charged site location.
Despite the lack of NMR information, this adduct could
be clearly characterized as 2-OHE2-1(4)-N2dG on the
basis of its CAD mass spectrum and the well documented
decomposition schemes described for A ring substituted
catechol estrogens (10-13).
2-OHE2-1(4)-N7Gu a (G6). The G6 adduct produces an
MH+ ion at m/z 438, indicative of the loss of the deoxy-
ribose moiety. Accordingly, the nucleophilic site of dG
involved in this adduct is the N7 nitrogen atom. Submit-
ted to collisional excitation in the ion trap, the m/z 438
parent ion of G6 decomposes into the same fragment ions
as observed for the m/z 438 MH+ ion from adduct G5 (data
not shown). The fragmentation pattern observed on the
CAD spectrum of G6 is not influenced by the steroid-base
linkage on the A ring, since adducts G5 and G6 both give
rise to highly stable charged sites which lead to the same
charge remote fragmentation processes. On the basis of
these data and previously reported results (10-12), the
linkage site of dG could be located on the A ring of the
steroid and G6 was identified as 2-OHE2-1(4)-N7Gua.
Rea ction of 2-OHE2-Q w ith d A. A similar study was
carried out with deoxyadenosine (dA). LC-MS experi-
ments have allowed us to characterize two stable adducts
A1 (tR ) 36.7 min) and A3 (tR ) 39.3 min) by their MH+
ions at m/z 538 and two depurinating adducts A2 (tR
)
38.2 min) and A4 (tR ) 38.7 min) at m/z 422 (not shown).
As above, successive purifications of the reaction mixture
have been carried out using semipreparative HPLC. Two
fractions F1′ and F2′, which accounted respectively for 44
and 6% of the amount of 2-OHE2-Q used in the reaction,
were isolated as pure compounds but the fraction F3′
(<1%) contains several products (data not shown).
The occurrence of the m/z 152 fragment ion (Figure 5)
confirms the attachment of Gua on the steroid A ring,
and its formation can be explained considering the
formation of an ion-dipole complex intermediate. The
cleavage of the steroid base bond can occur by hydrogen
transfer from the C11 to the C1 carbon atom of the
steroid, leading to an ion-neutral complex species (in
brackets in Scheme 4), which gives rise to the liberation
of the m/z 152 ion after proton transfer from the charged
steroid to the neutral base moiety of the intermediate.
1
2-OHE2-6r-N6d A (A1 or A3). A H NMR spectrum of
the F1′ fraction was recorded in DMSO (not shown) but
it was not possible to realize 2D experiments since the
solution decomposed overnight. Of particular impor-
F igu r e 5. CAD mass spectrum obtained from daughter ion at m/z 438 isolated from MH+ ion of G5 (m/z 554). Interpretation of the
main fragments is given in Schemes 3 and 4.

Catechol Estrogen-Deoxynucleoside Adducts
Chem. Res. Toxicol., Vol. 15, No. 5, 2002 761
Sch em e 3. P r op osed Decom p osition Mech a n ism s of th e m /z 438 P a r en t Ion Cor r esp on d in g to a n Ad d u ct on
A Rin g Lea d in g to Ion s a t m /z 324 (A), 312 (B), 298 (C), 286 (D), a n d 272 (E)
Sch em e 4. P r op osed Mech a n ism for th e F or m a tion of Ion s a t m /z 152 fr om th e P a r en t Ion a t m /z 438
Cor r esp on d in g to a n Ad d u ct on A Rin g
tance is the observation of characteristic sugar signals:
a triplet at 6.36 ppm for the H1′ proton and four
multiplets between 4.6 and 3.5 ppm for the H4′, H3′ and
H5′, H5′′ protons. The spectrum is very similar to the
one described by Stack et al. (10) as the adduct 2-OHE1-
6(R,â)-N6dA in the reaction of dA with 2-OHE1-Q. The
presence of a NH doublet at 7.93 ppm suggests a reaction
of the base at the C6 position by the N6dA atom for this
stable adduct displaying MH+ ions at m/z 538. The NMR
analysis was restricted to the assignment of protons
involved in the base addition process, namely the H6, the
aromatic, the NH or NH2 protons as well as the methyl
groups (Table 3). Taking into account the line width of
the multiplet at 5.62 ppm assigned to the H6 proton, it
is possible to propose for this adduct a base linkage on
the R face of the steroid, 2-OHE2-6R-N6dA. However, due
to the rapid degradation, it was not possible to identify
by LC-MS this compound as the A1 or A3 adduct.

762 Chem. Res. Toxicol., Vol. 15, No. 5, 2002
Convert et al.
Ta ble 3. 1H Ch em ica l Sh ifts of Ad d u cts Id en tified in th e Rea ction Mixtu r e of 2-OHE2-Q w ith d Aa
adducts
solvent
H6
H1/H4
H2/H8-dA
CH3
NH/NH2
2-OHE1-6(R,â)-N6dA (10)
DMSO
DMSO
DMSO
CD3OD
CD3OD
5.58
5.62
5.65
5.73
5.87
6.68/6.56
6.63/6.52
6.80/6.30
6.92/6.44
6.82/6.00
8.31/8.27
8.32/8.22
8.17/7.46
8.26/7.45
8.21/8.08
0.81
0.70
0.61
0.70
0.83
7.94/ -
7.93/ -
- /7.21
2-OHE2-6R-N6dA
2-OHE2-6â-N7Ade
2-OHE2-6â-N7Ade
2-OHE2-6R-N7Ade
a
As a reference, the first line of this table corresponds to the stable adduct 2-OHE1-6(R,â)-N6dA obtained by Stack et al. (10). Chemical
shifts are given in ppm/TMS by using the residual solvent signal as internal reference at 2.5 ppm for DMSO and 3.31 ppm for CD3OD.
2-OHE2-6â-N7Ad e a n d 2-OHE2-6r-N7Ad e (A2, A4).
In the 1H NMR spectrum of the F2′ fraction in DMSO
(not shown), the characteristic sugar signals are absent.
So, in agreement with the mass spectrometry results
(MH+ ion observed at m/z 422), the F2′ compound should
correspond to a depurinating adduct. Beside a broad
signal corresponding to two protons at 7.21 ppm, the
spectrum shows a methyl signal at 0.61 ppm, four
aromatic protons, and one doublet at 5.64 ppm with
TOCSY correlations characteristic of a H6 proton (Table
3). No modification is observed for this H6 proton when
the spectrum is recorded in CD3OD. These observations
are consistent with a reaction of the base at the C6
position by the N7dA atom, leading to a loss of the
deoxyribose moiety. Given the above results, the multi-
plicity of the H6 proton (doublet of two large signals)
allows us to propose the following structure, 2-OHE2-6â-
N7Ade for this adduct, which should correspond to the
adduct A2 or A4 detected by LC-MS2 at m/z 422.
m/z 554) (12, 13), but they could not be purified for a
subsequent NMR analysis. However, the structure of the
depurinating adduct G7, which very likely results from
the hydrolysis of one of these stable adducts in the
purification step, has been obtained by NMR. We know
that N-glycosidic bonds are susceptible to pH variations.
The acid-catalyzed hydrolysis of the base-sugar linkage
has already been described from guanine (33) or adenine
(34) nucleosides. Loss of a base from DNA by hydrolysis
of the N-glycosidic bond is known to be one of the most
frequent types of spontaneous alterations of the cell
genome (35). If we consider that G7 is resulting from G3
or G4, then yields we have obtained for stable adducts
(nearly 5%) are not very different from those given by
Stack et al. (10).
The two diastereomer depurinating adducts G1 (2-
OHE2-6R-N7Gua) and G2 (2-OHE2-6â-N7Gua) have been
analyzed by LC-MS and NMR and the stereochemistry
of these two adducts has been determined by NMR.
According to the UV profile and the reconstructed ion
chromatograms of the m/z 438 diagnostic ion (Figure 1),
the abundance ratio [G1]/[G2] of these adducts in the
crude reaction mixture could be estimated to be 20/80.
This result might be explained by a higher stability of
the G2 adduct in the acidic conditions of the reaction.
Furthermore, the presence of the 2-hydroxy-11-oxo-
estradiol in most fractions of the reaction mixtures might
indicate the parallel formation of unstable adducts at the
C9 position. Adducts G5 and G6 detected in LC-MS
experiments as respectively stable or depurinating ad-
duct (12, 13) are formed via a base addition on the C1 or
C4 aromatic position of the steroid skeleton (Figure 1).
Unfortunately, they could not be isolated for a NMR
characterization and therefore seem to be formed as
minor adducts under the reaction conditions we have
used.
Two new keto steroids were isolated as byproducts at
the same time as deoxynucleosides adducts. The forma-
tion of the F4 compound, 2-hydroxy-6-oxo-estradiol, may
be explained by a simple oxidation of the starting steroid
2-OHE2 since it is also found when the reaction was
carried out without any deoxynucleoside addition in the
reaction medium. On the other hand, the F5 compound,
2-hydroxy-11-oxo-estradiol, might derive from a precursor
adduct corresponding to reaction of the base at the C9
position as shown in Scheme 1, since this compound was
not detected when the reaction was conducted without
dG or dA. Such a C9 adduct was proposed in the reaction
between 2- or 4-OHE1 and glutathione as a result of the
isomerization of these catechol estrogen quinones to
highly electrophilic p-quinone methides 2- or 4-OHE1-
QM2 (36). These adducts at the C9 position were de-
scribed as unstable and decomposed into 9(11)-dehydro-
hydroxyestrones (5). In the present case, the degradation
product 9(11)-dehydro-2-OHE2 might be further oxidized
to give the F5 compound. As support for this proposal, it
1
The H NMR spectrum of the F3′ fraction in CD3OD is
rather complex. However, on the basis of the signal
integrals, this mixture should contain in addition to free
sugar and free base, a very small amount of an adduct
characterized by a doublet of doublets at 5.87 ppm, four
singlets in the aromatic region and a methyl signal at
0.83 ppm. In this fraction, from the values of these
chemical shifts and from the multiplicity of the H6 proton
at 5.87 ppm, and by comparison with the dG case, it is
possible to assume the presence of the diastereomer
depurinating adduct 2-OHE2-6R-N7Ade.
Discu ssion
The o-quinones formed from oxidation of the catechols
are considered as strong electrophiles and can react in
vitro or in vivo with nucleophilic residues (10, 11). Using
the chemical conditions optimized by Stack et al. (10),
covalent adducts between 2-OHE2 and the deoxyribo-
nucleosides dG, dA, and dC were obtained in minute
quantities and could be characterized only by electro-
spray ionization mass spectrometry (12, 13). The reaction
between catechol estrogen quinones and deoxyribonu-
cleosides gives rise to more or less stable covalent
adducts, resulting from an attack of the base on specific
electrophilic sites of the estrogen skeleton. These sites
are the aromatic positions C1 or C4 (ring A) and the
benzylic positions C6 or C9 (ring B) since it is known that
the 2-OHE2-o-quinone can form two quinone methides,
2-OHE2-QM1 and 2-OHE2-QM2 (32) (Scheme 1).
The above results demonstrate that the nucleophilic
attack of the deoxyribonucleoside dG principally takes
place at the C6 atom of the estradiol-2,3-quinone skel-
eton, via both exocyclic N2dG and endocyclic N7dG atoms.
The G3 and G4 stable adducts, 2-OHE2-6(R,â)-N2dG
(Figure 1), corresponding to attack at the C6 position by
the N2dG atom, have been detected by LC-MS (MH+ at

Catechol Estrogen-Deoxynucleoside Adducts
Chem. Res. Toxicol., Vol. 15, No. 5, 2002 763
is worth noting that a “transient” adduct was detected
by H NMR with no signal in the sugar region but one
on female CD-1 mice neonates treated with 2- and
4-hydroxyestradiol that both 2- and 4-hydroxylated es-
tradiol are carcinogenic metabolites (45). Furthermore,
the present results might provide reference data to
elucidate the structure of the different stable estrogen
adducts found with oligodeoxynucleotides. They may,
therefore, help us to understand at the molecular level
how they could potentially disrupt cellular functions
irreversibly (46). Work is now in progress to characterize
structure of adducts resulting from alkylation of catechol
estrogens to DNA sequence models.
1
sharp and two broad signals in the aromatic region.
These observations might indicate a steric compression
in this region as expected between the H8-Gua and H1
protons for an adduct at the C9 position, as observed by
Stack et al. (10) for the adduct at the C1 position. The
possibility of addition at the C9 position is in good
agreement with recent mass spectrometry results ob-
tained in similar reactions with catechol estrogens deu-
terated at the C6 and C7 positions.2 Attempts at isolation
and further spectral characterization of this “transient”
adduct were not successful.
Ack n ow led gm en t. This work was partly supported
for the C. Van Aerden’s thesis by the Institut Scientifique
Roussel, which is greatly acknowledged. We thank Dr.
E. Cavalieri for helpful discussions.
The NMR data indicate that the nucleophilic attack
of the deoxyribonucleoside dA takes place on the C6 atom
of the estradiol-2,3-quinone, via both N6dA and N7dA
atoms and gives rise to stable and depurinating adducts.
Amount of stable adducts, 2-OHE2-6(R,â)-N6dA, cor-
respond to about half of the yield published previously
(10).
Refer en ces
(1) Overall evaluations of carcinogenicity: an updating of IARC
monographs volumes 1 to 42 (1987) IARC Monographs on the
Evaluation of Carcinogenic Risk to Humans, Suppl 7, pp 272-
310, Lyon, France.
(2) Hormonal contraception and post-menopausal hormonal therapy
(1999) IARC Monographs on the Evaluation of Carcinogenic Risks
to Humans, Vol. 72, Lyon, France.
(3) Yager, J . D. (2000) Endogenous estrogens as carcinogens through
metabolic activation. J . Natl. Cancer Inst. Monogr. 27, 67-73.
(4) Dickson, R. B., and Stancel, G. M. (2000) Estrogen receptor-
mediated processes in normal and cancer cells. J . Natl. Cancer
Inst. Monogr. 27, 135-145.
(5) Bolton, J . L., Pisha, E., Zhang, F., and Qiu, S. (1998) Role of
quinoids in estrogen carcinogenesis. Chem. Res. Toxicol. 11,
1113-1127.
(6) Bolton, J . L., Trush, M. A., Penning, T. M., Dryhurst, G., and
Monks, T. J . (2000) Role of quinones in toxicology. Chem. Res.
Toxicol. 13, 136-160.
(7) Cavalieri, E., Frenkel, K., Liehr, J . G., Rogan, E., and Roy, D.
(2000) Estrogens as endogenous genotoxic agents-DNA adducts
and mutations. J . Natl. Cancer Inst. Monogr. 27, 75-93.
(8) Liehr, J . G., Fang, W. F., Sirbasku, D. A., and Ari-Ulubelen, A.
(1986) Carcinogenicity of catechol estrogens in Syrian hamsters.
J . Steroid Biochem. 24, 353-356.
(9) Li, J . J ., and Li, S. A. (1987) Estrogen carcinogenesis in Syrian
hamster tissues: role of metabolism. Fed. Proc. 46, 1858-1863.
(10) Stack, D. E., Byun, J ., Gross, M. L., Rogan, E. G., and Cavalieri,
E. L. (1996) Molecular characteristics of catechol estrogens
quinones in reactions with deoxyribonucleosides. Chem. Res.
Toxicol. 9, 851-859.
In conclusion, all the results show clearly that the
reaction of the quinones of 2-hydroxy-catechol estrogens
with the deoxyribonucleosides dG and dA may lead to
depurinating covalent adducts as well as stable adducts.
At this stage, it is worth noting that the estrogen-2,3-
quinones were previously (10) expected to form mainly
stable adducts, whereas the estrogen-3,4-quinones were
shown to give rise to formation of depurinating adducts.
Furthermore, analysis of the NMR coupling constants
allows us to establish the stereochemistry of the adduct
linkages. The apurinic adducts, i.e., 2-OHE2-6(R,â)-
N7Gua, 2-OHE2-6â-N2Gua, and 2-OHE2-1-N7Gua or
2-OHE2-4-N7Gua, and a degradation product, 2-hydroxy-
11-oxo-estradiol, generated from transient adducts have
been also identified. These compounds could be consid-
ered as a signature of the formation of unstable adducts
on DNA, particularly at the C9 position of catechols, and
also of a C6/C9 competition for the attack of the nucleo-
phile. Besides, such compounds or their metabolic deriva-
tives should be considered as complementary biomarkers
of adducts recently described for the 4-hydroxyestrogens
in cancer initiation risk assessment (37, 38).
(11) Cavalieri, E. L., Stack, D. E., Devanesan, P. D., Todorovic, R.,
Dwivedy, I., Higginbotham, S., J ohansson, S. I., Patil, K. D.,
Gross, M. L., Gooden, J . K., Ramanathan, R., Cerny, R. L., and
Rogan, E. G. (1997) Molecular origin of cancer: catechol estrogen-
3,4-quinones as endogeneous tumor initiators. Proc. Natl. Acad.
Sci. U.S.A. 94, 10937-10942.
(12) Van Aerden, C., Debrauwer, L., Tabet, J . C., and Paris, A. (1998)
Analysis of nucleoside-estrogen adducts by LC-ESI-MS-MS.
Analyst 123, 2677-2680.
(13) Van Aerden, C., Debrauwer, L., Paris, A., Molines, H., Convert,
O., and Tabet, J . C. (1998) Isomeric differentiation of nucleoside-
2-OH-estradiol adducts by ESI-MS/MS in an ITMS. Adv. Mass
Spectrom. 14, 177.
(14) Gelbke, H. P., Haupt, O., and Knuppen, R. (1976) A simple
chemical method for the synthesis of catechol estrogens. Steroids
21, 205-218.
(15) Abul-Hajj, Y. J . (1984) Synthesis of 3,4-estrogen-o-quinone. J .
Steroid Biochem. 21, 621-622.
(16) Marion, D., and Wuthrich, K. (1983) Application of phase sensitive
two-dimensional correlated spectroscopy for measurements of 1H-
1H spin-spin coupling constants in proteins. Biochem. Biophys.
Res. Commun. 113, 967-974.
(17) Bax, A., and Davis, D. G. (1985) MLEV-17 Two-dimensional
homonuclear magnetization transfer spectroscopy. J . Magn. Res.
65, 355-360.
(18) Bermel, W., Griesinger, C., Kessler, H., and Wagner, G. (1987)
Assignment of carbonyl groups and sequence analysis in peptides
by inverse correlation using long-range couplings Magn. Res.
Chem. 25, 325-326.
Formation of depurinating adducts from 2-OHE2 may
constitute an important argument concerning the sus-
pected incidence of the resulting apurinic sites in DNA
in cancer initiation process as hypothesized for 4-hy-
droxylated estrogens (7, 11). These sites might be re-
sponsible for mistakes in DNA replication and play a
crucial role in cancer initiation process as shown for
mouse skin c-H-ras mutations induced by dibenzo[a,l]-
pyrene adduction to DNA (39, 40) or for the human p53
gene mutation in codon 249 found in hepatocarcinomas
(41, 42) that could be attributed to the formation of the
aflatoxin B1-N7Gua adduct released from DNA (43).
Interestingly, when considering estrogen-3,4-quinones,
Chakravarti et al. (44) were able to measure significant
amounts of 4-hydroxyestradiol-1-N3-Ade and 4-hydroxy-
estradiol-1-N7-Gua adducts only 1 h after treatment of
mouse skin with estradiol-3,4-quinone, and they noticed
that the H-ras oncogene displayed abundant A to G
mutations 6 h after treatment. Our results concerning
the ability of 2-hydroxyestrogens to generate abasic sites
to a significant level support the recent evidence obtained
2
L. Debrauwer et al., manuscript submitted.

764 Chem. Res. Toxicol., Vol. 15, No. 5, 2002
Convert et al.
(19) Bax, A., and Marion, D. (1988) Improved resolution and sensitivity
in 1H-detected Heteronuclear Multiple-Bond Correlation Spec-
troscopy. J . Magn. Res. 78, 186-191.
(20) Neuhaus, D., and Williamson, M. (1989) The Overhauser effect
in structural and conformational analysis, pp 353-420, VCH
publishers.
(35) Kunkel, T. A., Schaaper, R. M., and Loeb, L. A. (1983) Depuri-
nation - induced infidelity of deoxyribonucleic acid synthesis with
purified deoxyribonucleic acid replication proteins in vitro. Bio-
chemistry 22, 2378-2384.
(36) Iverson, S. L., Shen, L., Anlar, N., and Bolton, J . L. (1996)
Bioactivation of estrone and its catechol metabolites to quinoid-
glutathione conjuguates in rat liver microsomes. Chem. Res.
Toxicol. 9, 492-499.
(37) Todorovic, R., Devanesan, P., Higginbotham, S., Zhao, J ., Gross,
M. L., Rogan, E., and Cavalieri, E. (2001) Analysis of potential
biomarkers of estrogen-initiated cancer in the urine of Syrian
golden hamsters treated with 4-hydroxyestradiol. Carcinogenesis
22, 905-911.
(38) Cavalieri, E. L., Kumar, S., Todorovic, R., Higginbotham, S.,
Badawi, A. F., and Rogan, E. G. (2001) Imbalance of estrogen
homeostasis in kidney and liver of hamsters treated with estra-
diol: implications for estrogen-induced initiation of renal tumors.
Chem. Res. Toxicol. 14, 1041-1050.
(39) Chakravarti, D., Pelling, J . C., Cavalieri, E. L., and Rogan, E. G.
(1995) Relating aromatic hydrocarbon-induced DNA adducts and
c-H-ras mutations in mouse skin papillomas: the role of apurinic
sites. Proc. Natl. Acad. Sci. 92, 10422-10426.
(40) Chakravarti, D., Mailander, P. C., Cavalieri, E. L., and Rogan,
E. G. (2000) Evidence that error-prone DNA repair converts
dibenzo[a,l]pyrene-induced depurinating lesions into mutations:
formation, clonal proliferation and regression of initiated cells
carrying H-ras oncogene mutations in early preneoplasia. Mutat.
Res. 456, 17-32.
(21) Tabakovic, K., Gleason, W. B., Ojala, W. H., and Abul-Hajj, Y. J .
(1996) Oxidative transformation of 2-hydroxyestrone. Stability
and reactivity of 2,3-estrone quinone and its relationship to
estrogen carcinogenecity. Chem. Res. Toxicol. 9, 860-865.
(22) Cao, K., Stack, D. E., Ramanathan, R., Gross, M. L., Rogan, E.
G., and Cavalieri, E. L. (1998) Synthesis and structure elucidation
of estrogen quinones conjuguated with cysteine, N-acetylcysteine
and glutathione. Chem. Res. Toxicol. 11, 909-916.
(23) York, J . L. (1981) Effect of the structure of the glycone on the
acid-catalysed of adenine nucleosides. J . Org. Chem. 46, 2171-
2173.
(24) Pfau, W., Broskstedt, U., Schulze, C., Neurah, G., and Marquardt,
H. (1996) Characterization of the major DNA adduct formed by
the food mutagen 2-amino-3-methyl-9H-pyrido[2,3-b] indole
(MeAaC) in primary rat hepatocytes. Carcinogenesis 12, 2727-
2732.
(25) Duax, W. L., Rohrer, D. C., Blessing, R. H., Strong, P. D., and
Segaloff, A. (1979) Acta Cristallogr., Sect. B 35, 1096-1098.
(26) Bischofberger, K., Bull, J . R., and Chalmers, A. A. (1987) Analyses
of the 500 MHz 1H spectra of the 3-methyl ethers of estrone,
estradiol and their 14a-methyl analogues. Magn. Res. Chem. 25,
780-783.
(27) Tomer, K. B., Crow, F. W., and Gross, M. L. (1983) Location of
double bond position in unsaturated fatty acids by negative ion
MS/MS. J . Am. Chem. Soc. 105, 5487-5488.
(28) Gross, M. L. (1992) Charge-remote fragmentations: method,
mechanism and applications. Int. J . Mass Spectrom. Ion Process
118/119, 137-165.
(29) Adams, J . (1990) Charge-remote fragmentations: analytical
applications and fundamental studies. Mass Spectrom. Rev. 9,
141-186.
(30) Tomer, K. B., and Gross, M. L. (1988) Fast atom bombardment
and tandem mass spectrometry for structure determination:
remote site fragmentation of steroid conjugates and bile salts.
Biomed. Environ. Mass Spectrom. 15, 89-98.
(31) Griffiths, W. J ., Zhang, J ., and Sjo¨vall, J . (1994) Charge-remote
fragmentation of sulphated and glucuronidated bile acids and
their 2-aminoethanesulphonic acid derivatives. Rapid Commun.
Mass Spectrom. 8, 227-236.
(41) Hsu, I. C., Metcalfe, R. A., Sun, T., Welsh J . A., Wang, N. J ., and
Harris, C. C. (1991) Mutational hotspot in the p53 gene in human
hepatocellular carcinomas. Nature 350, 427-428.
(42) Bressac, B., Kew, M., Wands, J ., and Ozturk, M. (1991) Selective
G
to T mutations of p53 in hepatocellular carcinoma from
southern Africa. Nature 350, 429-431.
(43) Groopman, J . D., Croy, R. G., and Wogan, G. N. (1981) In vitro
reactions of aflatoxin B1-adducted DNA. Proc. Natl. Acad. Sci.
78, 5445-5449.
(44) Chakravarti, D., Mailander, P. C., Li, K. M., Higginbotham, S.,
Zhang, H. L., Gross, M. L., Meza, J . L., Cavalieri, E. L., and
Rogan, E. G. (2001) Evidence that a burst of DNA depurination
in SENCAR mouse skin induces error-prone repair and forms
mutations in the H-ras gene. Oncogene 20, 7945-7953.
(45) Newbold, R. R., and Liehr, J . G. (2000) Induction of uterine
adenocarcinoma in CD-1 mice by catechol estrogens. Cancer Res.
60, 235-237.
(46) Terashima, I., Suzuki, N., Dasaradhi, L., Tan, C.-K., Downey, K.
M., and Shibutani, S. (1998) Translesional synthesis on DNA
templates containing an estrogen quinone-derived adduct: N2-
(2-hydroxyestron-6-yl-2′-deoxyguanosine and N6-(2-hydroxyestron-
6-yl)-2′-deoxyadenosine. Biochemistry 37, 13807-13815.
(32) Bolton, J . L., and Shen, L. (1996) p-Quinones are the major
decomposition products of catechol estrogen o-quinones. Carcino-
genesis 17, 925-929.
(33) Zoltewicz, J . A., and Clark, D. F. (1972) Kinetics and mechanism
of the hydrolysis of guanosine and 7-methylguanosine nucleosides
in perchloric acid. J . Org. Chem. 37, 1193-1197.
(34) York, J . L. (1981) Effect of the structure of the glycon on the acid-
catalyzed hydrolysis of adenine nucleosides. J . Org. Chem. 46,
2171-2173.
TX015561Y