Adduction of catechol estrogens to nucleosides.

Article abstract of DOI:10.1016/S0039-128X(02)00070-3

We report the formation, detection, quantitation and structural characterization of products resulting from the adduction of deoxynucleosides (deoxyadenosine, deoxyguanosine, deoxycytidine and 5-methyldeoxycytidine) to the catechol estrogens (CE) of estrone, estradiol-17beta and estradiol-17 alpha. The crude products are obtained in a one-pot synthesis through oxidation of catechols to quinones and subsequent Michael-type reaction with the deoxynucleosides in acidic medium.In all experiments, adducts are detected by electrospray ionization mass spectrometry analysis after HPLC separation (LC/ESI/MS(n)). The two pyrimidines deoxycytidine and 5-methyldeoxycytidine yield only CE adducts to deoxynucleosides, which correspond to stable adducts on DNA. For purines, the results depend on the CE (2,3- or 3,4-catechols) used, the function and configuration on carbon 17 (ketone for estrone, alcohol for alpha and beta isomers of estradiol), and on the purine itself (deoxyadenosine or deoxyguanosine). Both stable adducts and deglycosylated adducts are formed, and therefore formation of stable adducts on DNA as well as the loss of purines from the DNA strands could be possible. MS(2) and MS(3) experiments prove to be relevant for further structural determinations, enabling in some cases the elucidation of the regiochemistry of adduction on the A and B rings of the steroid moiety.

Full text of DOI:10.1016/S0039-128X(02)00070-3

Steroids 67 (2002) 1091–1099
Adduction of catechol estrogens to nucleosides
Isabelle Jouanin, Laurent Debrauwer, Gwénola Fauglas, Alain Paris∗, Estelle Rathahao
Laboratoire des Xénobiotiques, INRA, 180 Chemin de Tournefeuille, BP 3, 31931 Toulouse Cedex 09, France
Received 26 June 2001; received in revised form 4 June 2002; accepted 11 June 2002
Abstract
We report the formation, detection, quantitation and structural characterization of products resulting from the adduction of deoxynucleo-
sides(deoxyadenosine, deoxyguanosine, deoxycytidineand5-methyldeoxycytidine)tothecatecholestrogens(CE)ofestrone, estradiol-17␤
and estradiol-17␣. The crude products are obtained in a one-pot synthesis through oxidation of catechols to quinones and subsequent
Michael-type reaction with the deoxynucleosides in acidic medium.
In all experiments, adducts are detected by electrospray ionization mass spectrometry analysis after HPLC separation (LC/ESI/MSn ).
The two pyrimidines deoxycytidine and 5-methyldeoxycytidine yield only CE adducts to deoxynucleosides, which correspond to stable
adducts on DNA. For purines, the results depend on the CE (2,3- or 3,4-catechols) used, the function and configuration on carbon 17
(ketone for estrone, alcohol for ␣ and ␤ isomers of estradiol), and on the purine itself (deoxyadenosine or deoxyguanosine). Both stable
adducts and deglycosylated adducts are formed, and therefore formation of stable adducts on DNA as well as the loss of purines from the
DNA strands could be possible. MS2 and MS3 experiments prove to be relevant for further structural determinations, enabling in some
cases the elucidation of the regiochemistry of adduction on the A and B rings of the steroid moiety.
© 2002 Elsevier Science Inc. All rights reserved.
Keywords: Catechol estrogens; Deoxynucleosides; DNA adducts; Steroid; Michael-type addition; LC/ESI/MSn
1. Introduction
relative binding affinity of E2␣ using uterine estrogen re-
ceptors or in vitro synthesized ␣ or ␤ receptor subtype are
in good agreement with the low estrogenic potency of E2␣
[9–11]. Further oxidation of the CE of E2␤ or E1 yield
semi-quinones and quinones, which are strong electrophiles
that can react with nucleophilic sites of DNA [12]. However,
on the contrary to biopotency evaluations, the influence of
the configuration of the 17-hydroxyl group on the chemical
reactivity of CE has never been determined until now, and to
our knowledge, there has been no study on the reactivity of
17␣-estradiol quinones towards DNA bases. Depending on
the reactive site involved, adducts can be retained on DNA
as stable adducts or decomposed as CE-bases adducts (deg-
lycosylated adducts), giving rise to the occurrence of abasic
sites on DNA [12]. As these processes could be implicated
in the initiation step of carcinogenesis, the knowledge of the
adducts, which are really produced in vivo, is of great in-
terest. Previously, only a few estrogen–nucleic acid adducts
were described [13–17], mainly with the 3,4-CE of E1 and
E2␤ in connection with their established carcinogenicity in
animal models [14] or with their supposed link to mammary
or endometrium cancer [18,19]. Very few studies have dealt
with chemical reactions and mechanisms that could explain
the formation of the different isomeric adducts [20,21].
2,3-CE are of interest since they are produced by most
Catechol estrogens (CE) can be produced in human
cells through aromatic hydroxylation of estrone (E1) and
estradiol-17␤ (E2␤) [1,2]. Estradiol-17␣ (E2␣) is the main
metabolite of E2␤ found in the bovine species [3,4] but is
only a minor metabolite in humans [5]. This derivative is
qualitatively submitted to the same main metabolic path-
ways as E2␤ and E1, which result in the biosynthesis of
2-methoxy derivatives [6]. Compared to E2␤, E2␣ is par-
tially devoid of estrogenic activity [7,8]. Studies on the
Abbreviations: CE, catechol estrogens; LC/ESI/MS, liquid chromato-
graphy–electrospray ionization–mass spectrometry; E1, estrone (estra-
1,3,5(10)-trien-3-ol-17-one); E2␣, estradiol-17␣ (estra-1,3,5(10)-trien-
3,17␣-diol); E2␤, estradiol-17␤ (estra-1,3,5(10)-trien-3,17␤-diol); 2-OH-
E2␤, 2-hydroxyestradiol-17␤ (estra-1,3,5(10)-trien-2,3,17␤-triol); 4-OH-
E2␤, 4-hydroxyestradiol-17␤ (estra-1,3,5(10)-trien-3,4,17␤-triol); 2-OH-
E2␣, 2-hydroxyestradiol-17␣ (estra-1,3,5(10)-trien-2,3,17␣-triol); 4-OH-
E2␣, 4-hydroxyestradiol-17␣ (estra-1,3,5(10)-trien-3,4,17␣-triol); 2-OH-
E1, 2-hydroxyestrone (estra-1,3,5(10)-trien-2,3-diol-17-one); 4-OH-E1,
4-hydroxyestrone (estra-1,3,5(10)-trien-3,4-diol-17-one); RT, retention
time; dA, deoxyadenosine; dG, deoxyguanosine; dC, deoxycytidine;
5-Me-dC, 5-methyl-deoxycytidine; TIC, total ionic current
∗
Corresponding author. Tel.: +33-561-28-53-94;
fax: +33-561-28-52-44.
E-mail address: aparis@toulouse.inra.fr (A. Paris).
0039-128X/02/$ – see front matter © 2002 Elsevier Science Inc. All rights reserved.
PII: S0039-128X(02)00070-3

1092
I. Jouanin et al. / Steroids 67 (2002) 1091–1099
cell types in larger amounts than 3,4-CE [21,22] and have
also been shown to generate CE–DNA adducts [23]. More-
over, using in vitro methodologies, such 2,3-CE adducts on
oligonucleotides can induce A → T transversions, A → G
or G → A transitions as well as deletions depending on the
DNA-polymerases (␣, ␤, ␦) used for the elongation steps
[24,25].
ative HPLC, using a Diol column (Lichrospher OH,
250 mm × 21.2 mm, 5 ␮m, Interchim, Montluçon, France)
and pre-column (Lichrospher OH, 33 mm × 7.8 mm, 5 ␮m,
Interchim). Isocratic elution was achieved at 10 ml/min us-
ing dichloromethane/isopropanol (97:3, v/v). UV detection
was recorded at 280 nm.
2.3. Syntheses
In order to fully characterize chemical structures, we
have synthesized various adducts by chemical synthesis,
leading to a larger amount of products than by biological
means. To obtain an easier structural analysis of the iso-
meric adducts synthesized, we have used deoxynucleosides
instead of deoxynucleotides. We have re-examined the re-
activity of 2,3- and 3,4-CE of E1, E2␣ and E2␤ towards
four different deoxynucleosides on a systematic basis in
order to determine which kind of adduct could be formed,
and to assess their quantitative distribution. Because these
adducts and the reactants are sensitive to oxidation, the
crude reaction mixtures were directly analyzed by liquid
chromatography–electrospray ionization mass spectrometry
Adducts were prepared as previously described [15,21].
2,3-CE (resp. 3,4-CE) were oxidized to quinones in acetoni-
trile at −40 ◦C (resp. 0 ◦C) with 8 eq. of activated manganese
(IV) oxide. After 10 min, the suspension was filtered under
vacuum through a Whatman GF/B filter on a solution of
monohydrated nucleoside (6 eq.) in acetic acid/water (1:1,
v/v). Stirring was maintained for 6 h at room temperature
(+26 ◦C) under argon. After addition of 1 mg of l-ascorbic
acid, the solution was evaporated to dryness, and the or-
ange oil obtained was kept under argon at −20 ◦C until
analysis.
n
(LC/ESI/MS ) as described before [23]. General schemes
2.4. LC/ESI/MSn analyses
are proposed to explain the regiochemistry of the adduction
process by comparison with previously published results
[12,20,21,23].
An Ultrabase C18 reversed phase column (250 mm ×
2 mm, 5 ␮m, SFCC, Eragny, France) was used at a flow rate
of 0.2 ml/min with the following gradient: from 0 to 5 min,
linear gradient from 0 to 15% B and then, linear gradient
to get 50% B at 30 min and then 100% B at 40 min. The
latter was maintained until 60 min. Eluent compositions
were as follows: (A) CH3OH/H2O/CH3COOH (10:90:0.2,
v/v/v) and (B) CH3OH/H2O/CH3COOH (90:10:0.2, v/v/v).
Quantification of adducts was achieved by integration of the
corresponding chromatographic peaks on the appropriate
reconstituted ion chromatogram. The results were expressed
as percentages of the total adducts detected considering
that all the adducts display the same mass spectrometric
response under positive electrospray ionization conditions.
Electrospray ionization was achieved using the following
conditions: needle voltage (5 kV), heated capillary tempera-
ture (230 ◦C), heated capillary voltage (26 V), and lens tube
2. Experimental
2.1. Materials
2.1.1. HPLC
Preparative HPLC was performed on a Thermo Separation
Products P1000XR pump equipped with a UV 3000 detector
(Thermo Quest, Les Ulis, France).
2.1.2. LC/ESI/MSn
Structural analyses were carried out using HPLC on a
Thermo Separation Products P4000 pump fitted with a Rheo-
dyne 7725i injector. UV detection was achieved at 280 nm
with a Thermo Separation Products UV 1000 detector. Mass
spectrometry was performed on a Finnigan LCQ (Thermo
Quest) quadrupole ion trap mass spectrometer fitted with an
ESI source.
n
offset (0 V). MS experiments were carried out using typi-
cal isolation window widths from 1.0 to 1.5 amu and exci-
tation voltages from 0.75 to 1.25 V. All data were recorded
using the AGC (automated gain control) mode. Helium was
n
used as the collisional gas for MS experiments.
2.2. Chemicals
E1 (estra-1,3,5(10)-trien-3-ol-17-one), E2␣ (estra-1,3,5
(10)-trien-3,17␣-diol), E2␤ (estra-1,3,5(10)-trien-3,17␤-
diol), monohydrated deoxynucleosides [deoxyadenosine
(dA), deoxyguanosine (dG), deoxycytidine (dC), 5-methyl-
deoxycytidine (5-Me-dC)], Fremy’s salt and activated man-
ganese (IV) dioxide were purchased from Sigma–Aldrich
Company (L’Isle d’Abeau, France). The 2,3- and 3,4-CE
were prepared from the corresponding estrogens by oxida-
tion with Fremy’s salt, according to previously published
methods [26]. We set up CE purification by prepar-
3. Results
3.1. General considerations for elucidating the
adduction of deoxynucleosides
The reaction of E2␤-2,3-quinone with deoxynucleosides
was extensively studied in previous works [23,27]. In this
work, we have studied the possible Michael-type adduction
of four deoxynucleosides (dA, dG, dC, 5-Me-dC) with the
2,3-quinones of E1 and E2␣, and with the 3,4-quinones of

I. Jouanin et al. / Steroids 67 (2002) 1091–1099
1093
Fig. 1. Adduction of nucleosides with CE: reaction of dG with E2␣-3,4-quinone.
E1, E2␤ and E2␣, in acidic medium (Fig. 1). These quinones
were obtained in situ by oxidation of the corresponding cat-
echols (CE), using one of the two methods employed in the
literature [20,21]. LC/ESI/MS and MS experiments were
carried out for the detection and analysis of adducts in crude
mixtures.
According to the nucleophilic site involved in the
Michael-type reaction, two different types of adducts could
be formed: CE-nucleosides and CE-bases (Fig. 2), the latter
resulting from the glycosidic bond breaking during the reac-
tion. From each crude reaction mixture, TIC chromatograms
were obtained by LC/ESI/MS (Fig. 3a), and various adducts
were separated. Their characterization was based on the
reconstituted ion chromatograms corresponding to the
m/z ratio of their quasi-molecular MH+ ions (Fig. 3b
and c). The repeatability of the experiments was assessed
with E2␣-2,3-Q on the basis of duplicate experiments for
each nucleoside used. The qualitative composition of the
reaction mixtures was found to be the same, although slight
variations in the relative quantities of adducts were ob-
served with variation coefficients being less than 12%. For
every quinone/deoxynucleoside pair studied, the relative
n
Fig. 2. Fragmentation patterns of adducts synthesized from 4-OH-E2␤ and dG.

1094
I. Jouanin et al. / Steroids 67 (2002) 1091–1099
Fig. 3. LC/ESI/MS and LC/ESI/MSn analyses of adducts synthesized from 4-OH-E2␤ and dG. Recordings are obtained for TIC (a), reconstituted ion
chromatogram of m/z 554 quasi-molecular ions (CE-nucleoside) (b), reconstituted ion chromatogram of m/z 438 quasi-molecular ions (CE-base) (c), MS3
spectrum of m/z 554 ions (d), and MS2 spectrum of m/z 438 ions (e).
quantification of each adduct was carried out according
to the HPLC procedure described in the experimental
section.
rings of the steroid skeleton, as indicated in Fig. 4. As a mat-
ter of fact, most of the adducts can be further identified by
comparing their fragmentation patterns determined by MS2
and MS3 experiments to those of similar adducts unambigu-
ously identified in previous works [21,23]. Schematically,
for adducts in which the nucleoside is linked to the B ring
of the steroid, the CAD spectra displays ions yielded by the
breaking of the steroid–base bond, which corresponds to the
protonated nucleoside and the steroid cation, as described
in Figs. 2 and 3d. In the case of a steroid–base linkage on A
ring, the bonding between the two partners is stronger and
charge-remote cross-ring breakings are observed, as shown
in Figs. 2 and 3e [21,23].
Due to the very low reaction yield, structural infor-
mation was generated by mass spectrometry. Moreover,
direct LC/ESI/MS analyses of the crude mixtures were
performed in order to avoid the degradation of adducts
that occurred while trying to isolate them for further struc-
tural identification. Taking these limitations into account,
our conclusions about the regiochemistry of the adduction
were deduced from the well-known chemistry of the 1,4-
and 1,6-Michael-type reactions, and from the interpretation
n
of the experimental MS data. Considering the estrogen
n
quinones studied in this work, the Michael reaction theoret-
ically allows five possible sites of adduction on the A and B
Yet, in some cases, the fragmentations observed in MS
were unusual and we could not precisely determine which

I. Jouanin et al. / Steroids 67 (2002) 1091–1099
1095
Fig. 4. Structures of 2,3-quinones, 3,4-quinones and quinone methides. Arrows indicate electrophilic sites for 1,4- and 1,6-Michael-type adductions.
site of adduction was more likely to correspond to a defined
structure, especially for the pyrimidic nucleosides.
(Table 1). In the reaction between E2␣-3,4-Q or E2␤-3,4-Q
and dC, two and three adducts were detected, one of them
being the major product accounting for 84 and 78%, respec-
tively. Most often, the regiochemistry of adduction was diffi-
3.2. Reaction of dC and 5-Me-dC with CE
n
cult to determine, due to non-elucidated MS data (Table 1).
Depending on the quinone, one to five CE-nucleosides
were formed whereas no CE-base was detected (Table 1).
Indeed, E2␤-2,3-Q yielded only one adduct with dC and
5-Me-dC whereas four or five adducts were formed with
E1-3,4-Q, E1-2,3-Q and E2␣-2,3-Q, with no major adduct
3.3. Reaction of dG with CE
CE-nucleosides and CE-bases were detected with all
quinones (Table 2). The CE-base accounted for more than
Table 1
LC/ESI/MSn data and quantification of the CE-nucleosides obtained from dC and 5-Me-dC and expected sites of adduction on the quinone moiety, as
deduced from the chemistry of Michael-type reaction and LC/ESI/MSn data
Pyrimidines
Estrogens
3,4-CE-quinones
E1-3,4-Q
2,3-CE-quinones
E1-2,3-Q
E2␤-3,4-Q
m/z 514
E2␣-3,4-Q
m/z 514
E2␤-2,3-Qa
m/z 514
E2␣-2,3-Q
m/z 514
dC
m/z 512
m/z 512
RTb
%
Sitec
RT
%
Site
RT
%
Site
RT
%
Site
RT
%
Site
RT
%
Site
24.0
26.0
33.0
41.0
34
13
29
24
1
1
1
1
24.5
28.5
33.5
78
11
11
1
1
1
29.0
31.5
16
84
1
1
26.0
27.0
29.5
31.0
31
16
42
11
1
1
1
1
25.5
100
3
22.0
25.8
27.0
35.0
29
21
31
19
1
1
1
1
5-Me-dC
m/z 526
m/z 528
m/z 528
m/z 526
m/z 528
m/z 528
24.5
35.0
39.0
40.0
41.5
22
13
19
35
11
2
2
2
2
2
39.0
39.7
53
47
2
2
29.0
39.5
40.0
54
25
21
2
2
2
28.0
30.5
32.0
40.0
40.5
21
35
14
14
16
2
2
2
2
2
26.4
100
2
27.5
29.0
36.0
41.0
41.5
27
17
30
15
11
2
2
2
2
2
a Reported in [23,27].
b Retention times (RT) are given in minutes.
c 1: C1, C2, C6␣, C6␤, C9; 2: C1, C4, C6␣, C6␤, C9; 3: C6␣, C6␤, C9 (B ring).

1096
I. Jouanin et al. / Steroids 67 (2002) 1091–1099
Table 2
LC/ESI/MSn data and quantification of adducts obtained from dG and dA and expected sites of adduction on the quinone moiety, as deduced from the
chemistry of Michael-type reaction and LC/ESI/MSn data
Purines
Estrogens
3,4-CE-quinones
E1-3,4-Q
2,3-CE-quinones
E1-2,3-Q
E2␤-3,4-Q
m/z 554
E2␣-3,4-Q
m/z 554
E2␤-2,3-Qa
m/z 554
%
E2␣-2,3-Q
m/z 554
dG CE-nucleosides m/z 552
m/z 552
RTb
%
Sitec
RT
%
Site RT
%
Site RT
%
Site RT
Site RT
%
Site
26.0 25
5
25.0 15
1
1
31.2 69
5
32.5 95
3
1
29.5 14
30.7 48
4
4
6
29.0 77
34.0 20
3
3
31.8
2
35.6
4
CE-bases
m/z 436
m/z 438
m/z 438
m/z 436
m/z 438
m/z 438
35.5 75
5
36.1 83
5
39.5 31
5
29.5
5
26.6
6
4
4
6
26.0
3
3
27.8 22
36.3
6
dA CE-nucleosides m/z 536
26.0 92
m/z 538
m/z 538
m/z 536
m/z 538
m/z 538
3
3
27.0 11
29.5 80
39.0
5
5
3
29.0 27
34.8 16
36.5 42
42.5 11
1
1
1
1
40.0 18
41.0 78
3
3
36.7 81
39.3 16
4
4
36.0 75
40.5 22.5
3
3
40.5
8
9
CE-bases
m/z 528
m/z 528
m/z 528
m/z 528
m/z 528
m/z 528
–
–
–
–
–
–
32.0
4
3
34.5
4
3
38.2
38.7
2
1
4
4
30.0
37.5
2
0.5
3
3
a Reported in [23,27].
b Retention times (RT) are given in minutes.
c 1: C1, C2, C6␣, C6␤, C9; 3: C6␣, C6␤, C9 (B ring); 4: C6␣, C6␤ (B ring); 5: C1, C2 (A ring); 6: C1, C4.
75% for reactions involving E1-3,4-Q and E2␤-3,4-Q. In
all the other experiments, CE-nucleosides corresponded to
more than 66% of the total amount of adducts. Moreover,
we noticed one major adduct for E1-2,3-Q and E2␣-2,3-Q
(95 and 77%, respectively). The adduction reaction of dG
seemed to be very selective with all the quinones used,
except E2␤-2,3-Q, since we detected no more than one or
two CE-nucleosides and one CE-base (Table 2). On the
contrary, the reaction of E2␤-2,3-Q was less selective: three
CE-nucleosides and three CE-bases were produced. Indeed,
the 2,3-quinones tended to give alkylation preferentially
through their B ring, whereas the regiochemistry was on
the A ring of 3,4-quinones. For quinones such as E1-3,4-Q,
E2␣-3,4-Q, E2␤-2,3-Q and E2␣-2,3-Q, the general regio-
chemistry proposed was the same for the CE-nucleosides
and their corresponding CE-bases (Table 2).
for E2␣-2,3-Q, there was one major CE-nucleoside, account-
ing for 92, 80, 78, 81, and 75%, respectively, of the total
amount of adducts detected (Table 2). Concerning the de-
duced regiochemistry, adductions occurred on the B ring ex-
cept for two of the CE-nucleosides formed with E2␤-3,4-Q
(Table 2).
4. Discussion
4.1. Analysis of the chemical reactivity
In the past, the reaction of deoxynucleosides on steroid
quinones was not investigated on CE derivatives of E2␣,
probably because these compounds were considered to be
very minor metabolites in humans. Therefore, we acquired
new results concerning the 17␣ isomers of 2,3- and 3,4-CE
and provided additional observations concerning CE of E1
and E2␤. In our experiments, we used the deoxynucleo-
sides dA, dG, dC, and also 5-Me-dC. 5-Me-dC is obtained
on DNA through enzymatic methylation of dC. It plays a
role in the transcription regulation during the cellular dif-
ferentiation and prevents the expression of specific genes
[28,29]. Although much less occurring than dA, dG and dC,
its chemical modification can be hazardous, and then, deter-
mine profiles of mutational hot spots [30].
3.4. Reaction of dA with CE
We detected at least two and at the most four CE-nucleo-
sides that represented more than 96% of the total amount of
adducts (Table 2). E2␣-3,4-Q qualitatively yielded the larger
number of CE-nucleosides with four adducts and was the
only 3,4-quinone to yield a CE-base. Indeed, no CE-base
was detected for E1-3,4-Q and E2␤-3,4-Q. For E1-3,4-Q,
E2␤-3,4-Q, and their corresponding 2,3-quinones and also

I. Jouanin et al. / Steroids 67 (2002) 1091–1099
1097
Analysis of the chemical reactivity was based on the struc-
tural characteristics of both reagents: the quinone and the
deoxynucleoside. On one hand, quinones are double elec-
trophilic enones for the 1,4- and 1,6-Michael-type reactions.
They have two theoretical activated sites each (Fig. 4), since
the adduction on positions 5 and 10 is unlikely because of the
junction between A and B rings. There also exists two tau-
tomeric quinoid forms called quinone-methides [12] for each
quinone (QM1 and QM2). These are simple enones, with
and 5-Me-dC (Table 1). This indicates that the exocyclic
nitrogen is preferentially involved in the chemical reaction
(more nucleophilic and less hindered alkylation site). Thus,
on a DNA strand, dC could be chemically modified by co-
valent bonding of a steroid. Moreover, when exposed to the
following quinones: E2␤-2,3-Q, E2␤-3,4-Q and E2␣-3,4-Q,
a specific adduction could occur since we have obtained only
one major adduct (Table 1). On the contrary, up to five iso-
mers could be obtained on exposition of dC or 5-Me-dC to
the other quinones.
=
=
two possible C O/C C conjugations. Because of the junc-
tion between A and B rings, only two activated sites are free
for 1,4- and 1,6-Michael-type reactions on 2-OHE-QM1,
2-OHE-QM2 and 4-OHE-QM2. 4-OHE-QM1 has two avail-
able sites for the 1,4-adduction. Concerning the C6 position,
the stereoselectivity of adduction can be ␣ or ␤. Therefore,
quinones have five possible sites of adduction which are C1,
C4, C6␣, C6␤, C9 for 2,3-quinones, and C1, C2, C6␣, C6␤, C9
for 3,4-quinones. C2 position could be disfavored because
it is a soft electrophile while the attacking nucleophile has
hard character. We can suppose that the C9 position is dis-
favored for steric hindrance since it corresponds to the junc-
tion between B and C rings. Yet, we recently demonstrated
the formation of a CE-nucleoside adduct involving a linkage
at the C9 position between dG and E2␤-2,3-Q deuterated on
C6␣, C6␤, and C7␣ [31]. Moreover, the possible adduction at
the C9 position has been reported by Bolton for glutathione
conjugates [32]. On the other hand, nucleosides possess sev-
eral nucleophilic sites: exocyclic N4, intracyclic N3 and O
for dC. Both purines have nucleophilic intracyclic N1, N3,
N7, C8 and an exocyclic amine (N6 for dA, N2 for dG). In
addition, dG has a supplementary nucleophilic oxygen atom.
We considered that it was the exocyclic N that was respon-
sible for the formation of CE-nucleoside adducts, while the
involvement of the intracyclic N7 of purines could explain
the formation of CE-bases by breaking the glycosidic bond
[21].
4.3. Adduction of dG with CE-quinones
Stack et al. [21] have described a CE-nucleoside (C6–N2)
with E1-2,3-Q (40% yield) and a CE-base (C1–N7) with
E1-3,4-Q (10% yield). Three CE-nucleosides and three
CE-bases on both rings A and B of E2␤-2,3-Q were also
previously described [23]. In this work, both types of
adducts were present, indicating that both types of DNA
modifications could happen. Indeed, the results obtained for
3,4-quinones (mainly CE-bases for E1-3,4-Q and E2␤-3,4-Q,
CE-nucleoside for E2␣-3,4-Q, A ring regiochemistry) indi-
cated that abasic sites could be generated preferentially on
dG sites of DNA when exposed to E1-3,4-Q or E2␤-3,4-Q.
Besides, we showed they were not the only possible al-
terations, since CE-nucleosides were also detected, thus
providing complementary information to what was previ-
ously described [21]. For 2,3-quinones, CE-nucleosides are
considered as the main type of resulting adducts. Moreover,
there is a preponderant adduct that involves the B ring of
E1-2,3-Q, as described previously [21], and E2␣-2,3-Q.
Yet, no major adduct has been detected with E2␤-2,3-Q
(Table 2). We deduce that these quinones can produce a
specific covalent modification of the dG sites on DNA and
that the ␤ isomer of estradiol shows specific results.
The fact that nucleosides add preferentially on the B ring
of 2,3-quinones while adducts can be formed from adduction
on A ring of 3,4-quinones as well, can be explained by the
involvement of different quinoids (Fig. 4). For 3,4-quinones,
the quinone methide 4-OHE-QM1, which could lead to ad-
duction at C6 position, is said to be disfavored [12], therefore
enabling the reaction to occur on the A ring. However, this
regioselectivity has been mainly observed with dG, showing
the influence of the nucleoside rather than that of the quinoid.
Quinones of E2␤ gave CE-nucleosides on both A and B rings
while E2␣ and E1 seemed to be more regioselective on one
ring, i.e. B ring for 2,3-quinones and A ring for 3,4-quinones
(Table 2). This result is in favor of a long distance effect of
the C17 configuration (ketone or ␣/␤ hydroxyl).
n
The systematic use of LC/ESI/MS for performing anal-
yses of the crude reaction mixtures has enabled us to answer
the following questions:
(i) How many adducts are formed for a given pair
quinone/nucleoside?
(ii) Which type of adduct is detected (CE-nucleoside or
CE-base)?
(iii) What is the regiochemistry of the adduction?
4.2. Adduction of dC and 5-Me-dC with CE-quinones
Adduction of 5-Me-dC with CE-quinones was never men-
tioned in the literature until now. Concerning dC, no adduc-
tion was reported in acidic medium [21]. Under reductive
conditions, C1–N4 and C2–N4 adducts of dC with E1-3,4-Q
were described [20]. We previously reported the characteri-
zation of adduct of dC with E2␤-2,3-Q involving the B ring
of the steroid moiety [23]. In this work, we have observed
the exclusive formation of CE-nucleoside adducts with dC
4.4. Adduction of dA with the CE-quinones
A single CE-nucleoside (C6–N6) with E1-2,3-Q (80%
yield) and no adduct with E1-3,4-Q were reported [21]. Be-
sides, formation of two CE-nucleosides and two CE-bases
(diastereoisomers on positions C6␣ and C6␤), was previously

1098
I. Jouanin et al. / Steroids 67 (2002) 1091–1099
evidenced with E2␤-2,3-Q [27]. In this work, we have
obtained more than 96% of CE-nucleosides in all our ex-
periments with dA, but we have detected no CE-bases
at all with E1-3,4-Q and E2␤-3,4-Q. Consequently, we
can notice that the reaction of dA with E1-3,4-Q yields
adducts that are exclusively CE-nucleosides. This brings
complementary results on the reactivity of 3,4-quinones
on dA, compared to what was expected from the literature
[21]. E2␣-3,4-Q shows a different reactivity from the other
3,4-quinones (one CE-base with a B ring regiochemistry,
four major CE-nucleosides of undetermined regiochem-
istry), thus underlining the specific behavior of the 17␣
isomer of estradiol. The formation of one and two CE-bases
is also detected on the B ring of 2,3-quinones, indicating
that abasic sites could also result from the reaction of dA
sites on DNA with estrogen–2,3-quinones, although to a
much lesser extent. From all these observations, the stable
covalent modification of dA on DNA could be indeed the
major result of the exposition to all the quinones.
bation. Such cellular events could appear in parallel to DNA
damage as recently shown with 4-hydroxyequilenin [33].
Acknowledgments
We thank the European Commission for financial support
(contract B6-7920/98/000823).
References
[1] Yager JD, Liehr JG. Molecular mechanisms of estrogen carcino-
genesis. Annu Rev Pharmacol Toxicol 1996;36:203–32.
[2] Zhu BT, Conney AH. Functional role of estrogen metabolism in
target cells: review and perspectives. Carcinogenesis 1998;19:1–27.
[3] Velle W. Studies on estrogens in cattle. Acta Endocrinol 1963;28:
186–91.
[4] Mellin TN, Erb RE. Estrogens in the bovine. J Dairy Sci 1965;48:
687–700.
[5] Adlercreutz H, Luukkainen T. Gas-chromatographic studies on
estrogens in the bile of pregnant women. Biochem Biophys Acta
1965;97:134–41.
[6] Williams KIH, Layne DS. Metabolism of 17␣-estradiol-6,7-3H by
nonpregnant women. J Clin Endocr 1967;27:159–64.
[7] Velle W. Endogenous anabolic agents in farm animals. Environ Qual
Saf Suppl 1976;5:159–70.
[8] Clark HH, Williams M, Upchurch S, Eriksson H, Helton E,
Markarevich BM. Effects of estradiol-17␣ on nuclear occupancy of
the estrogen receptor, stimulation of nuclear type II sites and uterine
growth. J Steroid Biochem 1982;16:323–8.
[9] Blair RM, Fang H, Branham WS, Hass BS, Dial SL, Moland
CL, et al. The estrogen receptor relative binding affinities of 188
natural and xenochemicals: structural diversity of ligands. Toxicol
Sci 2000;54:138–53.
[10] Kuiper GG, Carlsson B, Grandien K, Enmark E, Häggblad J, Nilsson
S, et al. Comparison of the ligand binding specificity and transcript
tissue distribution of estrogen receptors ␣ and ␤. Endocrinology
1997;138:863–70.
From the results obtained with 22 out of 24 experiments,
we can conclude that, among the total number of adducts
detected, CE-nucleosides correspond to the most frequently
obtained adducts. Thus, we can suspect that the major al-
teration of DNA results from the covalent adduction of the
nucleosides dC, 5-Me-dC, dA and dG with CE-quinones,
and that different regioisomers can be produced. Yet, in 8
out of these 22 experiments, we have also detected a signif-
icant amount of CE-bases. This suggests that the alteration
of DNA by generation of apurinic sites can also be possible.
From this work, it can be deduced that pyrimidic dC and
5-Me-dC cannot eliminate a CE-base from a DNA strand.
The two experiments for which the CE-base adducts are pre-
ponderant have implicated dG and the 3,4-quinones of E1
n
and E2␤. Besides, for 34 adducts detected by LC/ESI/MS ,
we are able to predict the steroid ring (A or B) involved in the
linkage between the quinone and the nucleoside used. These
general conclusions evidence the influence of various factors
on the adduction reaction such as the nature of the nucleo-
side/quinone pair used or the C17 configuration of steroids,
but remain difficult to explain. However, these results could
be particularly useful for comparison with adducts purified
from biological samples obtained from biopsies, cell incuba-
tion or with adducts coming from in vitro adduction experi-
ments performed on calf thymus DNA. Indeed, the reaction
conditions used in our chemical model studies have been de-
scribed to accurately predict the DNA adducts formed in vivo
[12,14,21]. Although some long distance specific effects on
the alkylation process that are linked to the hydroxyl stereo-
chemistry at the C17 position of CE have been revealed, we
can hypothesize that 17␣ isomers of CE could represent a
valuable chemical model to perform in vitro DNA alkylation
studies. In fact, due to the weak estrogenic potency of E2␣
[10,11], this method could be an easy way to uncouple such
cellular metabolic disruptions and DNA repairing steps due
to CE adduction processes from other cellular events medi-
ated via estrogen receptors as it is the case with E2␤ incu-
[11] Barkhem T, Carlsson B, Nilsson Y, Enmark E, Gustafsson J-A,
Nilsson S. Differential response of estrogen receptor ␣ and estrogen
receptor ␤ to partial estrogen agonist/antagonists. Mol Pharmacol
1998;54:105–12.
[12] Bolton JL, Pisha E, Zhang F, Qiu S. Role of quinoids in estrogen
carcinogenesis. Chem Res Toxicol 1998;11:1113–27.
[13] Nutters LM, Ngo EO, Abul-Hajj YJ. Characterization of DNA
damage induced by 3,4-estrone-o-quinone in human cells. J Biol
Chem 1991;266:16380–6.
[14] Cavalieri EL, Stack DE, Devanesan PD, Todorovic R, Dwidedy
I, Higginbotham S, et al. Molecular origin of cancer: catechol
estrogen–3,4-quinones as endogenous tumor initiators. Proc Natl
Acad Sci USA 1997;94:10937–42.
[15] Dwidedy I, Devanesan P, Cremonesi P, Rogan E, Cavalieri E.
Synthesis and characterization of estrogen 2,3- and 3,4-quinones.
Comparison of DNA adducts formed by the quinones versus horse-
radish peroxidase-activated catechol estrogens. Chem Res Toxicol
1992;5:828–32.
[16] Akanni A, Tabakovic K, Abul-Hajj YJ. Estrogen–nucleic acid
adducts: reaction of 3,4-estrone o-quinone with nucleic acid bases.
Chem Res Toxicol 1997;10:477–81.
[17] Abul-Hajj YJ, Tabakovic K, Tabakovic I. An estrogen–nucleic acid
adduct. Electroreductive intermolecular coupling of 3,4-estrone-o-
quinone and adenine. J Am Chem Soc 1995;117:6144–5.
[18] Liehr JG, Ricci MJ. 4-Hydroxylation of estrogens as marker of
human mammary tumors. Proc Natl Acd Sci USA 1996;93:3294–6.

I. Jouanin et al. / Steroids 67 (2002) 1091–1099
1099
[19] Liehr JG, Ricci MJ, Jefcoate CR, Hannigan EV, Hokanson JA, Zhu
BT. 4-Hydroxylation of estradiol by human uterine myometrium
and myoma microsomes: implications for the mechanism of uterine
tumorigenesis. Proc Natl Acad Sci USA 1996;93:9220–4.
[20] Akanni A, Abul-Hajj YJ. Estrogen–nucleic acid adducts: reaction
of 3,4-estrone-o-quinone radical anion with deoxyribonucleosides.
Chem Res Toxicol 1997;10:760–6.
[21] Stack DE, Byun J, Gross ML, Rogan EG, Cavalieri EL. Molecular
characteristics of catechol estrogen quinones in reactions with
dexyribonucleosides. Chem Res Toxicol 1996;9:851–9.
[22] Tabakovic K, Gleason WB, Ojala WH, Abul-Hajj YJ. Oxidative
transformation of 2-hydroxyestrone. Stability and reactivity of 2,3-
estrone quinone and its relationship to estrogen carcinogenicity. Chem
Res Toxicol 1996;9:860–5.
[23] Van Aerden C, Debrauwer L, Tabet JC, Paris A. Analysis of nucleo-
side-estrogen adducts by LC/ESI/MS. Analyst 1998;123:2677–80.
[24] Terashima I, Suzuki N, Itoh S, Yoshizawa I, Shibutani S. Mutagenic
specificity of model estrogen–DNA adducts in mammalian cells.
Biochemistry 1998;37:8803–7.
[27] Van Aerden C, Debrauwer L, Paris A, Molines H, Convert O,
Tabet JC. In: Karjalainen EJ, Hesso AE, Jabnen JE, Karjalainen UP,
editors. Isomeric differentiation, nucleoside–2-OH estradiol adducts
by ESI-MS/MS in an ITMS. Adv Mass Spectrom 1998;14. New
York: Elsevier.
[28] Privat E, Sowers LC. Photochemical deamination and demethylation
of 5-methylcytosine. Chem Res Toxicol 1996;9:745–50.
[29] Klein CB, Costa M. DNA methylation, heterochromatin and
epigenetic carcinogens. Mutat Res 1997;386:163–80.
[30] Denissenko MF, Chen JX, Tang MS, Pfeifer GP. Cytosine methylation
determines hot spots of DNA damage in the human p53 gene. Proc
Natl Acad Sci USA 1997;94:3893–8.
[31] Rathahao E, Clodic G, Debrauwer L, Jouanin I, Van Aerden
C, Molines H, et al. In: Gelpi E, editor. Investigation of the
regio- and stereoselectivity of the reaction of 2-OH estradiol with
deoxyribonucleosides using deuterium labelled steroids and liquid
chromatography–tandem ion mass spectrometry. Adv Mass Spectrom
2000;15. New York: Wiley.
[32] Bolton JL, Shen L. p-Quinone methides are the major decomposition
products of catechol estrogen o-quinones. Carcinogenesis
1996;17:925–9.
[33] Chen Y, Liu X, Pisha E, Constantinou AI, Hau Y, Shen L, et al.
A metabolite of equine estrogens, 4-hydroxyequilenin, induces DNA
damage and apoptosis in breast cancer cell lines. Chem Res Toxicol
2000;13:342–50.
[25] Terashima I, Suzuki N, Dasaradhi L, Tan C-K, Downey K, Shibutani
S. Translesional synthesis on DNA templates containing an estrogen
quinone-derived adduct: N2-(2-hydroestron-6-yl)-2ꢀ-deoxyguanosine
and N6-(2-hydroestron-6-yl)-2ꢀ-deoxyadenosine. Biochemistry 1998;
37:13807–15.
[26] Gelbke HP, Haupt O, Knuppen R. A simple chemical method for
the synthesis of catechol estrogens. Steroids 1973;21:205–18.