Tài liệu Báo cáo khoa học: Covalent binding to glutathione of the DNA-alkylating antitumor agent, S23906-1 doc

Covalent binding to glutathione of the DNA-alkylating antitumor
agent, S23906-1
Marie-He
´
le
`
ne David-Cordonnier
1
, William Laine
1
, Alexandra Joubert
1
, Christelle Tardy
1
,
Jean-Franc¸ois Goossens
2
, Mostafa Kouach
3
, Gilbert Briand
3
, Huong Doan Thi Mai
4
, Sylvie Michel
4
,
Francois Tillequin
4
, Michel Koch
4
, Ste

´
phane Leonce
5
, Alain Pierre
5
and Christian Bailly
1
1
INSERM U-524 et Laboratoire de Pharmacologie Antitumorale du Centre Oscar Lambret, IRCL, Lille, France;
2
Laboratoire de
Chimie Analytique, Faculte
´
des Sciences Pharmaceutiques et Biologiques, et
3
Laboratoire de Spectrome
´
trie de Masse,
Universite
´
de Lille, Lille, France;
4
Laboratoire de Pharmacognosie, Universite
´
Rene
´
Descartes (Paris 5), CNRS UMR8638,
Faculte
´
des Sciences Pharmaceutiques et Biologiques, Paris, France;

5
Division Recherche Cance
´
rologie,
Institut de Recherches SERVIER, Croissy sur Seine, France
The benzoacronycine derivative, S23906-1, was character-
ized recently as a novel potent antitumor agent through
alkylation of the N2 position of guanines in DNA. We show
here that its reactivity towards DNA can be modulated
by glutathione (GSH). The formation of covalent adducts
between GSH and S23906-1 was evidenced by EI-MS, and
the use of different GSH derivatives, amino acids and
dipeptides revealed that the cysteine thiol group is absolutely
required for complex formation because glutathione disul-
fide (GSSG) and other S-blocked derivatives failed to react
covalently with S23906-1. Gel shift assays and fluorescence
measurements indicated that the binding of S23906-1 to
DNA and to GSH are mutually exclusive. Binding of
S23906-1 to an excess of GSH prevents DNA alkylation.
Additional EI-MS measurements performed with the mixed
diester, S28053-1, showed that the acetate leaving group at
the C1 position is the main reactive site in the drug: a reaction
scheme common to GSH and guanines is presented. At the
cellular level, the presence of GSH slightly reduces the
cytotoxic potential of S23906-1 towards KB-3-1 epidermoid
carcinoma cells. The GSH-induced threefold reduction of
the cytotoxicity of S23906-1 is attributed to the reduced
formation of lethal drug–DNA covalent complexes in cells.
Treatment of the cells with buthionine sulfoximine, an
inhibitor of GSH biosynthesis, facilitates the formation of

drug–DNA adducts and promotes the cytotoxic activity.
This study identifies GSH as a reactant for the antitumor
drug, S23906-1, and illustrates a pathway by which GSH
may modulate the cellular sensitivity to this DNA alkylating
agent. The results presented here, using GSH as a biological
nucleophile, fully support our initial hypothesis that DNA
alkylation is the major mechanism of action of the promising
anticancer drug S23906-1.
Keywords: glutathione; DNA alkylation; acronycine; anti-
cancer drug; mechanism of action.
Introduction
The alkaloid acronycine (Fig. 1) was first isolated from the
bark of Acronychia baueri (also known as Sarcomeli-
cope simplicifolia), a Rutaceous tree widely distributed in
Australia [1,2]. This tetracyclic alkaloid was shown to be
moderately cytotoxic to a wide range of tumor cells in vitro
[3,4] and to display antitumor activities in vivo [5]. However,
clinical testing of acronycine itself showed a poor response
and the development of this compound was arrested in
the early 1980s. Nevertheless, the antitumor potential of
acronycine has stimulated the synthesis of numerous
analogues [6–8]. Recently, the benzoacronycine derivative,
S23906-1 (Fig. 1), was identified as a potent anticancer drug
with activity against a variety of human tumor xenograft
models in mice [9,10]. S23906-1 has been selected for
advanced preclinical evaluation.
From a mechanistic point of view, S23906-1 was recently
characterized as a DNA alkylating agent reacting irrevers-
ibly with guanine residues at the N2 position in double-
stranded DNA [11]. The covalent binding to DNA is

apparently responsible for the cytotoxic action [12] and the
capacity of the drug to trigger apoptosis in tumor cells
[13,14]. In the course of our ongoing studies aimed at
characterizing the interaction of S23906-1 with biologically
significant molecules, the reaction with glutathione (GSH)
was examined. The observation that a tricyclic analogue
of S23906-1 (i.e. 1,2-dihydroxy-1,2-dihydroacronycine
diacetate) reacts covalently with benzyl mercaptan to form
Correspondence to C. Bailly, INSERM U-524 et Laboratoire de
Pharmacologie Antitumorale du Centre Oscar Lambret, IRCL,
59045 Lille, France.
Fax: + 33 320 16 92 29, Tel.: + 33 320 16 92 18,
E-mail:
Abbreviations: BSO, buthionine sulfoximine; CD, circular dichroı
¨
sm;
Cys,
L
-cysteine; Cys-Gly, cysteine-glycine; EI-MS, electrospray ion-
ization mass spectroscopy; c-Glu-Cys, gamma-glutamic acid-cysteine;
Gln,
L
-glutamine; c-Glu-Gly, gamma-glutamic acid-glycine;
GS-DCE, S-dicarboxyethyl-glutathione; GS-Me, S-methyl-glutathi-
one; GS-NO, S-nitrosoglutathione; GS-SA, glutathione sulfonic acid;
GSH-O-Et, glutathione reduced ethyl ester; GSSG, oxidized gluta-
thione; IC
50
, 50% inhibitory concentration; MC, mitomycin C;
Met,

L
-methionine; N-Ac-Cys, N-acetyl-
L
-cysteine.
(Received 10 March 2003, revised 05 May 2003,
accepted 12 May 2003)
Eur. J. Biochem. 270, 2848–2859 (2003) Ó FEBS 2003 doi:10.1046/j.1432-1033.2003.03663.x
an S-linked adduct prompted us to initiate this work [6].
GSH, which is an abundant intracellular tripeptide (
L
-c-
glutamyl-
L
-cysteinyl-glycine), was used as a model bio-
nucleophile to investigate further the reactivity of S23906-1.
Direct experimental evidences for the formation of covalent
adducts between S23906-1 and GSH are reported here, and
a reaction scheme common to DNA and GSH is described.
Materials and methods
Chemicals and biochemicals
Synthesis of the benzoacronycine derivatives has been
reported previously [7]. The enantiomers S27589-1 and
S27590-1 were obtained from the racemate S23906-1 by
HPLC on a cellulose stationary phase (Chiralcel OC; Chiral
Technologies, Strasbourg, France). Synthesis of S28053-1
will be reported elsewhere, together with that of related
mixed esters and diacid hemiesters. Reduced glutathione
(GSH) and oxidized glutathione (GSSG), as well as
L
-cys-

teine (Cys), N-acetyl-
L
-cysteine (N-Ac-Cys),
L
-methionine
(Met),
L
-glutamine (Gln), gamma-glutamic acid-cysteine
(c-Glu-Cys), cysteine-glycine (Cys-Gly), gamma-glutamic
acid-glycine (c-Glu-Gly), S-methyl-glutathione (GS-Me),
S-nitrosoglutathione (GS-NO), S-dicarboxyethyl-glutathi-
one (GS-DCE), glutathione reduced ethyl ester (GSH-O-
Et), glutathione sulfonic acid (GS-SA) and buthionine
sulfoximine (BSO) were purchased from Sigma Aldrich.
DNA restriction fragments
The pBS plasmid was digested with PvuII and EcoRI and
the resulting 117-bp fragment was labeled at the EcoRI site
with [a-
32
P]dATP and avian myeloblastosis virus (AMV)
reverse transcriptase. Electrophoresis on a nondenaturing
6% (w/v) polyacrylamide gel served to remove excess
radioactive nucleotide, with the desired 3¢ end-labeled
product being cut out of the gel and eluted overnight in
500 m
M
ammonium acetate/10 m
M
magnesium acetate and
then ethanol precipitated.

Gel-shift studies
A typical cross-linking reaction consisted of incubating
10 lL of radiolabeled DNA, 2 lL of buffer (10 m
M
Na
cacodylate, pH 7.0; Tris buffer was avoided owing to the
presence of reactive amine functions) and 10 lL of the drug
at the desired concentration in the dark at room tempera-
ture, during the period specified in the legend, prior to
adding 5 lL of a 50% glycerol solution containing tracking
dyes. To study the inhibition of DNA alkylation, S23906-1
(50 l
M
) was first incubated with an excess of GSH or
derivatives (400 l
M
)in20lL of ammonium acetate for 1 h
at 37 °C prior to the addition of the radiolabeled DNA for a
further 2-h incubation period at 37 °C. DNA samples were
resolved by electrophoresis under nondenaturing conditions
in 6% acrylamide gels for  5 h at 300 V at room
temperature in TBE buffer (89 m
M
Tris base/89 m
M
boric
acid/2.5-m
M
Na
2

EDTA, pH 8.3). Gels were transferred to
Whatman 3MM paper, dried under vacuum at 80 °C, and
then analyzed on a phosphorimager (Molecular Dynamics
445SI).
Circular dichro (CD)
The CD spectra were obtained using a J-810 Jasco
spectropolarimeter at 20 °C controlled by a PTC-424S/L
Peltier type cell changer (Jasco Inc., Easton MD, USA). A
quartz cell of 10-mm path length was used to obtain spectra
from 450 to 290 nm with a resolution of 0.1 nm. The drug,
S27590-1 (50 l
M
final concentration), was incubated in
1mL of 1m
M
ammonium acetate, pH 7.15, with or
without (control) 1 m
M
GSH, GSSG, Cys, N-Ac-Cys,
Cys-Gly, GS-Me or GS-NO (from a 100-m
M
stock solution
previously equilibrated at pH 8.0 using KOH) in 1 m
M
ammonium acetate. CD spectra were collected every 10 min
from 0 to 150 min.
Electrospray ionization mass spectroscopy (EI-MS)
The spontaneous hydrolysis of S28053-1 was monitored by
EI-MS using a drug solution of 250 l
M

in 200 lLof1m
M
ammonium acetate, pH 8.0, and analyzing the sample at the
various time-points specified in the figure legend. For the
alkylation reactions, the various GSH derivatives (100 l
M
)
were incubated for 16 h at 20 °C, either alone or with
100 l
M
S23906-1, S27589-1, S27590-1 or S28053-1 in
200 lLof1m
M
ammonium acetate, pH 8.0. Samples were
injected in a simple-quadruple MS API I (Perkin-Elmer
Sciex) equipped with an ion-spray (nebulizer-assisted elec-
trospray) source (Sciex, Toronto, Canada) using a needle
prewashed with methanol. The solutions were infused
continuously with a medical infusion pump (Model 11,
Harvard Apparatus, South Natick, USA) at a flow rate of
5 lLÆmin
)1
. Polypropylene glycol was used to calibrate the
quadrupole. Ion spray mass spectra were acquired at unit
resolution by scanning from m/z 200–800 with a step size of
0.1 Da and a dwell time of 2 ms. Twenty spectra were
summed and recorded at an orifice voltage of +50 V,
whereas the potential of the spray needle was held at
+5 kV.
Alkylation of plasmids and fluorescence measurements

Compound S23906-1 (100 l
M
) was incubated with or
without increasing amounts of GSH or GSSG pre-
equilibrated at pH 8.0 from 100 l
M
to 25 m
M
in 200 lL
Fig. 1. Structure of the racemic cis diacetate compound, S23906-1 and
the mixed ester analogue, S28053-1. Compounds S27590-1 and S27589-
1 correspond to the individual enantiomeric R,R and S,S forms of the
racemate S23906-1, obtained by chiral HPLC.
Ó FEBS 2003 Glutathione conjugation by S23906-1 (Eur. J. Biochem. 270) 2849
of incubation buffer (1 m
M
ammonium acetate) for 6 h at
room temperature prior to adding 10 lg of the plasmid and
a further 16 h of incubation at 37 °C. Free drug molecules
were separated from DNA cross-linked molecules by
phenol/chloroform/isoamyl alcohol (25 : 24 : 1) extraction
followed by addition of 5 lLof5-
M
NaCl and precipitation
of DNA using 1 mL of cold ethanol. After centrifugation at
13 800 g for 30 min, the DNA pellet was dried and
dissolved in 1 mL of incubation buffer. The fluorescence
of the compound covalently linked to DNA was measured
using an excitation wavelength of 354 nm and an emission
range from 420 to 650 nm. Spectra were recorded using a

SPEX spectrofluorimeter Fluorolog.
Cell culture
KB-3-1 epidermoid carcinoma cells [15] were grown in
DMEM (Dulbecco’s modified Eagle’s medium)-glutaMAX
medium (Gibco) supplemented with 10% fetal bovine
serum, penicillin (100 IUÆmL
)1
) and streptomycin (100Æ
lg/mL) in a humidified atmosphere at 37 °C under 5%
CO
2
. The cells were harvested by trypsinization and plated
24 h before treatment with the test drug.
Survival assay
The cytotoxicity of S23906-1 and the effects of GSH or
derivatives on the cytotoxicity of this compound were
assessed using a cell proliferation assay developed by
Promega (CellTiter 96
Ò
AQ
ueous
one solution cell prolifer-
ation assay). Briefly, 3 · 10
3
exponentially growing KB-3-1
cells were seeded in 96-well microculture plates for 24 h
prior to treatment with graded concentrations of S23906-1
and 1-m
M
GSH or derivatives, in six independent points.

After 72 h of incubation at 37 °C, 20 lL of the tetrazolium
dye was added to each well and the samples were incubated
for a further 2 h at 37 °C. Plates were analyzed on a
Labsystems Multiskan MS (type 352) reader at 492 nm.
Detection of S23906-1-DNA adducts in KB-3-1 cells
in the presence of GSH
Approximately 1.5 · 10
6
KB-3-1 cells were grown for 24 h
in 100-mm diameter dishes with 5 mL of culture medium,
prior to the addition of GSH or derivatives (1 m
M
each)
and S23906-1 (10 l
M
) for 24 h. The genomic DNA was
extracted from cells as described previously [11]. Briefly,
after the drug treatment, cells were collected by centrifuga-
tion (5 min, 188 g), washed twice with 10 mL of NaCl/P
i
buffer and resuspended in 2 mL of NaCl/P
i
containing
5m
M
MgCl
2
prior to the addition of 200 lLof10%SDS
and mild agitation for 5 min. Proteinase K (80 lLofa
10-mgÆmL

)1
stock solution) was added for a further 5 min
of mild agitation, and finally 200 lLof0.1
M
EDTA,
pH 7.5, was added and the mixture incubated for 4 h at
37 °C. After addition of 80 lLof5
M
NaCl, the DNA was
extracted using 3 mL of phenol/chloroform/isoamyl alcohol
(25 : 24 : 1) and centrifuged at 3000 g for 10 min, followed
by two extractions with 3 mL of chloroform/isoamyl
alcohol (24 : 1). Finally, the DNA was precipitated with
cold ethanol and collected by centrifugation at 19 000 g for
30 min. The pellet was then dissolved in 200 lLofH
2
Oand
treated for 2 h with 5 l
M
RNase A (10 mgÆmL
)1
)toavoid
RNA contamination. The DNA concentration was
estimated by spectrophotometry at 260 nm. The fluores-
cence of S23906-1 covalently linked to DNA was measured
using a SPEX Fluorolog spectrofluorimeter with an
excitation wavelength at 300 nm and an emission range
from 420 to 555 nm. A similar procedure was used to
detect S23906-1–DNA adducts in KB-3-1 cells previously
treated using BSO. KB-3-1 cells, prepared as described

above, were treated for 24 h with increasing concentrations
of BSO prior to treatment with S23906-1. Genomic DNA
was extracted and the fluorescence quantified as described
above.
Determination of the intracellular GSH content
KB-3-1 cells (1 · 10
6
cells/dish) were incubated in the
presence or absence of BSO for 24 h, as described above.
Cells were then collected, washed with 10 mL of NaCl/P
i
and resuspended in 200 lLofNaCl/P
i
prior to lysis by
freezing and thawing twice. The ThioGlo-1
TM
(Calbiochem)
reagent (10 l
M
) was then added to the lysate (from a fresh
10-m
M
stock solution in dimethylsulfoxide) and the fluor-
escence was immediately measured with an excitation
wavelength of 360 nm and an emission wavelength of
400–650 nm.
Results
Molecular studies
Two complementary techniques, EI-MS and CD, were
deployed to study the reaction of the benzoacronycine

derivative, S23906-1, with GSH. MS is particularly well
suited for detecting the covalent adducts formed between
S23906-1 and the tripeptide. This is shown in the mass
spectrum given in Fig. 2A, with well-resolved peaks at
MH
+
¼ 308, 406, 448 and 490 corresponding to GSH and
to the diol, mono-acetate and di-acetate forms of the
parent drug, respectively. The latter three species reflect the
hydrolysis of the compound in solution (see the results
below with the analogue S28053-1). In addition, the
incubation of S23906-1 with GSH for 16 h in 1-m
M
ammonium acetate yielded two species with a mass of
MH
+
¼ 695 and 737. They corresponded to the expected
mass for the covalent GSH-drug adducts, with either one
remaining acetate (MH
+
¼ 737) or an OH group
(MH
+
¼ 695), at the C2 position. Direct evidence is given
below for a reaction with GSH implicating the C1 position.
These two peaks at MH
+
¼ 737 and 695 provide strong
evidence that the drug forms covalent complexes with the
tripeptide.

Similar MS experiments were performed after reacting
S23906-1 with various GSH analogues differently substi-
tuted or with amino acids and dipeptides each representing
a portion of the
L
-c-glutamyl-
L
-cysteinyl-glycine parent
compound. Typical mass spectra obtained with Cys,
N-acetyl-Cys, c-Glu-Cys, Cys-Gly and GSH-O-Et are
presented in Figs 2B–2F, respectively. In all cases with the
compounds bearing a free SH group, we were able to detect
the formation of covalent adducts with S23906-1. The
reaction is particularly strong with Cys because, in this case,
2850 M H. David-Cordonnier et al. (Eur. J. Biochem. 270) Ó FEBS 2003
the Cys–drug adducts are detected readily (either the
mono-acetate form at MH
+
¼ 551 or the OH form at
MH
+
¼ 509), with very little drug remaining unbound
(species at MH
+
¼ 406 (diol) or 448 (monoacetate). The
reaction is also very pronounced with the dipeptide Cys-Gly
(Fig. 2E), whereas the level of adducts is much smaller (but
still observed) with the c-Glu-Cys dipeptide (Fig. 2D). This
suggests that the drug preferentially recognizes the Gly side
of the tripeptide.

In contrast, absolutely no covalent adducts were detected
with other GSH-related compounds lacking the free thiol
group. The compounds tested are listed in Table 1. The list
includes the oxidized form GSSG and the S-protected GSH
derivatives GS-Me, GS-NO, GS-DCE and GS-SA, but also
smaller compounds such as Met, Gln and c-Glu-Gly. None
of these compounds was able to react covalently with
S23906-1, indicating that the SH group is strictly required
for the formation of covalent adduct with the benzoacro-
nycine derivative.
The second method used to provide evidence for the
formation of complexes between S23906-1 and GSH is CD.
Because this spectroscopic approach requires optically
active molecules, we did not use S23906-1 itself but the
pure enantiomeric forms S27590-1 and S27589-1, which are
the two cis enantiomers with the acetate groups located
either above or below the plane of the aromatic chromo-
phore (Fig. 1). These two compounds are equitoxic toward
different tumor cell lines (data not shown) and also react
equally well with DNA. This is shown in Fig. 3 from the gel-
shift experiments performed with a 117-bp radiolabeled
DNA substrate incubated with graded concentrations of
S27589-1 or S27590-1 for 2 h (Fig. 3A) or with a single dose
of each compound for up to 2 h (Fig. 3B). The band of
DNA showing a retarded electrophoretic mobility reflects
the formation of covalent adducts, as recently described
[11]. It is clear from these kinetic experiments that the two
cis enantiomers present equal capacities to react covalently
with DNA.
The reaction of compounds S27589-1 and S27590-1 with

GSH was then monitored by CD. As shown in Fig. 4A, the
CD spectrum of S27590-1 is altered upon reaction with
GSH, but no change occurs with GSSG. Monitoring the
changes of the CD signal at 300 nm allows us to distinguish
the SH and S-protected compounds. With the former, such
as GSH, Cys and Cys-Gly (open symbols in Fig. 4B), the
CD amplitude at 300 nm largely decreases. With the latter,
including GS-Me and GS-NO, for example (filled symbols
in Fig. 4B), variations of the CD sign are extremely limited.
This method thus provides an easy way to distinguish
reactive (free SH) vs. nonreactive (S-protected) species.
Although this technique cannot distinguish between binding
and bonding, the data nicely complement the MS results to
support the formation of stable complexes between GSH
and the benzoacronycine derivatives.
We have previously demonstrated that S23906-1 reacts
covalently with DNA [11] and we now show that it also
forms stable adducts with GSH. Two potential targets have
thus been identified. Competition experiments were per-
formed to determine whether the bonding to GSH can
prevent DNA alkylation. For this purpose, the drug was
first incubated for 1 h with an excess of GSH or related
compounds containing a free SH group (such as GSH-O-Et,
the dipeptide Cys-Gly or the amino acid Cys) or an
S-protected function (such as GSSG or GS-NO) prior to
addition of the 117-bp radiolabeled DNA fragment. After a
2-hreactionat37°C, the DNA samples were analyzed by
electrophoresis on polyacrylamide gels. The results are
displayed in Fig. 5. The alkylation of DNA is visualized by
an important gel shift and the degree of retardation depends

on the capacity of the drug to react with the competing
GSH-related product. GSH binding to native DNA and to
alkylated DNA was insignificant. Preincubation of S23906-
1 with Cys totally prevented the formation of DNA–drug
adducts. An incomplete inhibition of DNA–S23906-1
complex formation was also observed with GSH and the
SH-containing compounds GSH-O-Et, Cys-Gly and, to a
lesser extent, c-Glu-Cys and N-Ac-Cys. It is important to
note that despite the large excess of GSH compared with the
drug (400-l
M
GSH vs. 50-l
M
S23906-1), the reactivity of
the drug towards DNA was not completely abolished.
This can be important at the cellular level.
Fig. 2. EI-MS analysis of the alkylation of glutathione (GSH) and its
derivatives by S23906-1. S23906-1 (100 l
M
) was incubated with 100 l
M
GSH (M ¼ 307, panel A),
L
-cysteine (Cys) (M ¼ 121, panel B),
N-acetyl-
L
-cysteine (N-Ac-Cys) (M ¼ 163, panel C), gamma-glutamic
acid-cysteine (c-Glu-Cys) (M ¼ 250, panel D), cysteine-glycine (Cys-
Gly) (M ¼ 178, panel E) or glutathione reduced ethyl ester (GSH-
O-Et) (M ¼ 335, panel F) for 16 h at room temperature in 1 m

M
ammonium acetate, pH 7.15, prior to performing MS measurements
in the positive ion mode. Among the different species identified for
each spectrum, three species present the same molecular weight and
correspond to the uncomplexed molecules: (·) the diol form
(MH
+
¼ 406) and (.) the mono-acetate form (MH
+
¼ 448) derived
from spontaneous hydrolysis of (*) the parent di-acetate form
(MH
+
¼ 490). Covalent binding of the drug to GSH (MH
+
¼ 308)
or its derivatives gives adducts where one (ß)ortwo(fl) acetate groups
of S23906-1 are lost. In panel B, the peak at MH
+
¼ 241 corresponds
to [2xCys]H
+
.InpanelG,peaksatMH
+
¼ 336 and 669 correspond
to GSH-O-Et and [2xGSH-O-Et]H
+
, respectively.
Ó FEBS 2003 Glutathione conjugation by S23906-1 (Eur. J. Biochem. 270) 2851
Table 1. Summary data. The Ô+Õ refers to SH-containing glutathione (GSH) derivatives which interact with S23906-1 to modify the spectrum of

circular dichroı
¨
sm, form a covalent adduct identified by MS, compete with DNA in gel-shift assays, inhibit the alkylation of genomic DNA by
S23906-1 and reduce the cytotoxicity of S23906-1 in the survival assay using KB-3-1 cell line. The Ô0Õ refers to the S-protected GSH derivatives which
neither interact with S23906-1 nor modulate its cellular activity. +/–, Weak effect; ND, not determined. Cys,
L
-cysteine; Cys-Gly, cysteine-glycine;
c-Glu-Cys, gamma-glutamic acid-cysteine; Gln,
L
-glutamine; c-Glu-Gly, gamma-glutamic acid-glycine; GS-DCE, S-dicarboxyethyl-glutathione;
GS-Me, S-methyl-glutathione; GS-NO, S-nitrosoglutathione; GS-SA, glutathione sulfonic acid; GSH-O-Et, glutathione reduced ethyl ester;
GSSG, oxidized glutathione; Met,
L
-methionine; N-Ac-Cys, N-acetyl-
L
-cysteine.
Compound
Circular
dichroı
¨
sm MS
Gel-shift
competition
Alkylation of
genomic DNA
Survival
assay
GSH + + + + +
Cys + + + + +
N-Ac-Cys + + + + +

c-Glu-Cys ND + +/– ND +
Cys-Gly + + + ND +
GSH-O-Et ND + + + ND
GSSG 0 0 0 0 0
Met ND 0 0 0 0
Gln ND 0 0 0 0
c-Glu-Gly ND 0 0 ND 0
GS-Me 0 0 0 0 0
GS-NO 0 0 0 0 0
GS-DCE ND 0 0 0 0
GS-SA ND 0 0 ND ND
Fig. 3. Concentration- and time-dependence for the alkylation of DNA
by S27589 and S27590 using electromobility shift assays. (A) The drug
at the indicated micromolar concentration was incubated with a 117-
bp radiolabeled DNA fragment for 2 h at room temperature prior to
PAGE on a nondenaturing 6% gel. (B) The drug at a fixed concen-
tration of 50 l
M
was reacted with DNA for up to 120 min. In both (A)
and (B), experiments were performed in 1 m
M
sodium cacodylate,
pH 7.0, and the free and bound DNA forms were separated by PAGE
on nondenaturing 6% gels. Control tracks labeled ÔDNAÕ contained
no drug; (b)and(f) refers to bound and free DNA, respectively.
Fig. 4. Circular dichroı
¨
sm (CD) analysis of the interaction of glutathi-
one (GSH) or its derivatives with the diacetate compound S27590. (A)
CD spectra of S27590 (50 l

M
) incubated alone (thinner line), with 1 m
M
GSH (broken line) or with 1 m
M
glutathione disulfide (GSSG) (thick
line) for 90 min at 20 °C. In (B), the intensity of the CD band centered
at 300 nm, characteristic of the GSH–drug complexes, is plotted vs.
time. S27590 (50 l
M
)wasincubatedwith1-m
M
GSH (n), GSSG (m),
L
-cysteine (Cys) (s), N-acetyl-
L
-cysteine (N-Ac-Cys) (e), cysteine-
glycine (Cys-Gly) (h), S-methyl-glutathione (GS-Me) (d)orS-meth-
ylglutathione (GS-NO) (j)in1-m
M
ammonium acetate, pH 7.15.
2852 M H. David-Cordonnier et al. (Eur. J. Biochem. 270) Ó FEBS 2003
In sharp contrast to that observed with GSH, no
inhibition of DNA alkylation was detected with the
S-protected compounds, in particular GSSG, GS-NO, GS-
DCE or GS-SA. This is again a strong indication that the
SH group of the tripeptide is required for covalent complex
formation with the benzoacronycine derivative. The fact
that the inhibitory effect is significantly more pronounced
with dipeptide Cys-Gly than with c-Glu-Cys suggests that

the drug may establish contact with the Gly side of the
peptide to react with the adjacent SH group on the central
Cys residue, in agreement with the MS data presented
above. These experiments indicate that the same reactive site
on the drug molecule is implicated in the covalent binding to
DNA and GSH. The two targets compete for the same
electrophilic species.
The competition between DNA and GSH was investi-
gated further by fluorescence spectroscopy. The drug was
incubated for 6 h with GSH or GSSG prior to addition of
the DNA (in this case a supercoiled plasmid was used) and
the reaction was performed for 15 h at 37 °C. Unbound
drug molecules were then extracted with phenol/chloroform
and the remaining DNA was precipitated with ethanol. The
fluorescence spectrum of the DNA-bound S23906-1 species
was then recorded. The results shown in Fig. 6A clearly
reveal that the preincubation of the drug with GSSG had
no effect on the capacity of the drug to alkylate DNA,
whereas the incubation with GSH considerably reduced the
extent of drug–DNA covalent complex formation. This is
also shown in the concentration-dependence plot in
Fig. 6B. A concentration of 1-m
M
GSH (i.e. close to the
intracellular level) reduces, by about 40%, the capacity of
S23906-1 to bond to DNA. At 10-m
M
GSH, the extent of
DNA alkylation by S23906-1 is reduced by 70%. Alto-
gether, the gel-shift assays (Fig. 5) and fluorescence data

(Fig. 6) unambiguously reveal that the reaction of S23906-1
with GSH and DNA is mutually exclusive. Bonding to the
peptide target modulates binding to the nucleic acid and
vice versa. However, it should be noted that even a massive
GSH concentration, such as 10 m
M
(exceeding the physio-
logical concentration), does not completely abolish DNA
alkylation.
On the basis of structure–DNA alkylation and structure–
activity relationships, we have recently shown that the
reactive site on the S23906-1 molecule is the C1 position
bearing a leaving acetate group [12]. The same reactive site
must be involved in the reaction with the thiol function of
GSH. To investigate this further, we performed a detailed
MS analysis with an analogue of S23906-1 bearing two
different ester groups at the C1 and C2 positions. The mixed
diester S28053-1 (Fig. 1) bears an acetate group at C1 and a
hemisuccinate group at C2, enabling a facile distinction
between the two ester groups by means of MS. This
compound, used here specifically to investigate the mech-
anism of action of the parent compound, is as potent as
S23906-1 at alkylating DNA (data not shown).
Figure 7 shows the hydrolysis data for a solution of
S28053-1 incubated for up to 20 h in the ammonium acetate
Fig. 6. Effects of glutathione (GSH) and glutathione disulfide (GSSG)
on DNA alkylation by S23906-1. (A) The fluorescence-emission spectra
were recorded after incubation of S23906-1 with GSH or GSSG for
6 h at 37 °C followed by the addition of plasmid DNA (10 lg) for a
further 15-h incubation period at 37 °C. The DNA was then extracted,

precipitated and resuspended in buffer for measurement of fluores-
cence at an excitation wavelength of 354 nm. S23906-1 (100 l
M
) alone
(plain line), or with 25 m
M
GSH (long dash) or GSSG (short dash). (B)
Variation (%) of the fluorescence intensity at 520 nm for DNA-bound
S23906-1 in the presence of increasing concentrations of GSH (d)or
GSSG (h). The fluorescence of the drug–DNA complexes in the
controls (without GSH or GSSG) was considered to be 100%.
Fig. 5. Competitive binding of S23906-1 to DNA and SH-containing
molecule. The drug (50 l
M
) was incubated with or without glutathione
(GSH) or its derivatives (400 l
M
each) for 1 h at 37 °C prior to the
addition of the 117-bp radiolabeled DNA fragment for a further 2 h of
incubation at 37 °C. The drug free (f) and bound (b)DNAformswere
separated by PAGE on a 6% nondenaturing gel.
Ó FEBS 2003 Glutathione conjugation by S23906-1 (Eur. J. Biochem. 270) 2853
buffer required for the MS analysis. The mass spectra
recorded at 0 and 5 min correspond to a freshly made
solution of S28053-1. The intact drug is, of course,
predominantly found (MH
+
¼ 562), but a tiny amount of
the diol (MH
+

¼ 406) and the mono-ester species
(MH
+
¼ 520 for the C2 hemisuccinate ester and
MH
+
¼ 448 for the C1 acetate ester) can be detected. With
time, the proportions of the diol species gradually increase
and, after 20 h, the diester compound S28053-1 is almost
undetectable. For the two monoester species, the one bearing
the acetate group is always present in a smaller amount
compared to that harboring the hemisuccinate ester func-
tion. It should be noted that different peaks with a mass of
+14 (peaks at 406 + 14, 448 + 14 or 520 + 14, marked
with different symbols in the Fig. 7) are detected and
correspond to the reaction of the drug with methanol used to
wash the needle of the quadrupole just prior to the injection.
The asymmetric compound, S28053-1, was then incuba-
ted with GSH overnight at room temperature in 1-m
M
ammonium acetate buffer, pH 7.15, and the resulting
mixture was analyzed by MS. A typical mass spectrum is
presented in Fig. 8A and the main peaks have been
attributed to the different species obtained. In addition to
the unreacted compounds (GSH at MH
+
¼ 308.2 and
S28053-1 at MH
+
¼ 562) and the above mentioned

hydrolysis products at MH
+
¼ 406 (diol) and
MH
+
¼ 520 (mono hemisuccinate ester), different species
corresponding to the GSH reaction products were detected.
The main peak, at MH
+
¼ 695.3, corresponds to the
expected mass for the adduct between GSH and the drug in
its alcohol form, i.e. without the ester group on the C2
position. By analogy with findings for DNA, we know that
this adduct arises from the reaction of the monoester
compound with GSH. A trans-esterification process con-
verts the C2 ester into a C1 ester, which then immediately
reacts with GSH to form the expected C2-OH adduct
Fig. 7. EI-MS analysis of the hydrolysis of S28053. A250-l
M
drug
solution prepared in 1 m
M
ammonium acetate, pH 7.15, was analyzed
at the indicated incubation time (from 0 to 20 h in panels A–H).
S28053 (MH
+
¼ 562) hydrolyzes in the mono-hemisuccinate form (Ö)
(MH
+
¼ 520), in the mono-acetate form (s)(MH

+
¼ 448), and in
the diol form (h)(MH
+
¼ 406). Three other peaks correspond to the
reactivity of methanol (M ¼ 32) (used to wash the needle of the
quadrupole) towards the mono-hemisuccinate form (MH
+
¼ 534, r),
the mono-acetate form (MH
+
¼ 448, .)andthediolform
(MH
+
¼ 406, e), giving the M +14 species.
Fig. 8. Alkylation reaction of S28053 with glutathione (GSH). (A)
EI-MS analysis of the alkylation reaction of S28053 with GSH. A
solution containing 100 l
M
of the drug was incubated with 100 l
M
GSH for 16 h at room temperature in 1 m
M
ammonium acetate,
pH 7.15, prior to EI-MS analysis in positive ion mode. The structures
of the main species identified are indicated. (B) Reaction scheme for the
hydrolysis of S28053 and its reaction with GSH. The different species
indicated, with the corresponding mass, are seen in the EI-MS spectra
of Fig. 7 and/or Fig. 8A.
2854 M H. David-Cordonnier et al. (Eur. J. Biochem. 270) Ó FEBS 2003

(Fig. 8B). The trans-esterification pathway has been clearly
demonstrated by NMR with the related mono-acetate
compound [12]. This C2-OH adduct cannot derive from the
diol compound, which is totally inert towards DNA [11], or
with GSH (data not shown). The peak corresponding to a
mass of MH
+
¼ 809.5 represents the GSH-bound
C2-hemisuccinate ester adduct, as depicted in the reaction
scheme in Fig. 8B. Some of the smaller peaks in the mass
spectrum shown in Fig. 8A remain to be attributed. Some
may correspond to side reaction products after the trans-
esterification process. For example, the peak at
MH
+
¼ 737.4 corresponds to the expected mass for the
GSH-bound C2-acetate adduct.
The reaction scheme presented in Fig. 8B summarizes the
MS data and illustrates the hydrolysis and GSH reaction
pathways for the compound S28053-1. The reactive site on
the benzoacronycine derivative is the C1 position bearing
the leaving acetate group. This group is absolutely essential.
Its replacement with a nonleaving group (e.g. a methoxy
substituent) totally prevents the reaction with the electro-
philic species, be it GSH or DNA [12].
Cellular studies
The reaction of S23906-1 with GSH may modulate the
cellular response to the benzoacronycine derivative by
quenching the alkylation of DNA, thereby decreasing the
formation of potentially lethal DNA–drug covalent

adducts. To test this hypothesis, we investigated the effect
of GSH on drug-induced cytotoxicity using the KB-3-1
epidermoid carcinoma cell line. The cells were incubated
with graded concentrations of S23906-1 for 72 h in the
presence or absence of GSH, and the cytotoxic effect was
measured using a conventional colorimetric assay. In these
experiments, a solution containing the test drug and GSH
was prepared and then added rapidly to the cellular
medium. In the presence of the GSH, the cytotoxic effect
was slightly reduced, by a factor of about three. The 50%
inhibitory concentration (IC
50
) value of 2.2 l
M
measured
with S23906-1 alone was increased to 7.1 l
M
in the presence
of 1-m
M
GSH. For comparison, in parallel experiments
performed with the conventional N7-DNA alkylator,
mechlorethamine hydrochloride, we observed a reduction
of cytotoxicity by a factor of about six. In this case, the IC
50
value of 4.2 l
M
determined with mechlorethamine alone
increased to 24.8 l
M

in the presence of mechlorethamine
and 1 m
M
GSH. The tripeptide exerts a mildly negative
effect on the cytotoxicity of the benzoacronycine derivative,
S23906-1.
If a significant proportion of the drug reacts with GSH,
there should be less active drug available to alkylate the
genomic DNA in cells. The amount of drug–DNA covalent
complexes in cells can be estimated by fluorescence meas-
urements, taking advantage of the specific fluorescence of
the benzoacronycine chromophore. To evaluate the effect of
GSH on the formation of DNA adducts in whole cells, KB-
3-1 cells were treated with 10 l
M
S23906-1 for 1 h in the
presence or absence of GSH or its derivatives. The genomic
DNA was subsequently isolated by phenol extraction and
precipitated prior to measurement of the drug–DNA
adducts by fluorescence. As shown in Fig. 9, the fluores-
cence spectra of a DNA solution obtained from the KB-3-1
cells treated with S23906-1 reveal the presence of drug
molecules attached to the genomic DNA. In the presence of
GSH, the level of drug–DNA adducts decreased signifi-
cantly, whereas it remained unchanged in the presence of
GSSG. Again, an effect was observed with the SH-
containing analogs, but not with the S-blocked GSH
derivatives. For example, a marked decrease of the fluor-
escence intensity at 500 nm was detected when cells were
treated with a combination of S23906-1 and Cys. On the

contrary, the association S23906-1 + c-Glu-Gly gave a
level of drug–DNA adducts similar to that obtained with
the drug alone. The reduced cytotoxic activity of S23906-1
in the presence of GSH can thus be explained by a decreased
capacity to alkylate DNA. Indirectly, these experiments
support the idea that the formation of drug–DNA covalent
complexes is responsible for the cytotoxic action [12].
Finally, we evaluated the effect of BSO on the formation
of drug–DNA covalent complexes in cells. BSO decreases
GSH synthesis by specifically inhibiting c-glutamylcysteine
synthetase [16], the rate-limiting step in GSH biosynthesis.
BSO is commonly used to deplete cells in GSH, thereby
potentiating the cytotoxic action of GSH-reactive anti-
cancer drugs, such as melphalan and camptothecin [17]. The
depleting effect of BSO in KB-3-1 cells was visualized by
fluorescence using the fluorescent thiol reagent, ThioGlo-
1
TM
. This maleimide derivative produces a highly fluorescent
product upon its reaction with thiol groups, therefore
providing a simple and sensitive assay for estimating the
GSH content in cell homogenates [18]. The decrease of the
fluorescence peak centered at 510 nm, observed upon
adding increasing concentrations of BSO to the KB-3-1
cells, reflects the GSH-depleting effect (Fig. 10A). The cells
were first incubated with BSO for 24 h prior to adding
10-l
M
S23906-1 and, after a further 1 h of incubation, the
DNA was extracted and the level of drug–DNA adducts

formed in situ was estimated by fluorescence. The fluores-
cence spectra presented in Fig. 10B reveal that the number
of drug–DNA complexes increases with increasing concen-
trations of BSO. Depleting the cells with BSO reduces the
intracellular concentration of GSH and, consequently,
Fig. 9. Inhibition of the formation of S23906-1–DNA covalent com-
plexes in cells in the presence of glutathione (GSH) or its derivatives.
Fluorescence-emission spectra of DNA extracted from KB-3-1 epi-
dermoid cells treated for 1 h with 10 l
M
S23906-1 or in the presence of
1-m
M
GSH, glutathione disulfide (GSSG),
L
-cysteine (Cys), N-acetyl-
L
-cysteine (N-Ac-Cys), glutathione reduced ethyl ester (GSH-O-Et),
S-methyl-glutathione (GS-Me) or gamma-glutamic acid-glycine
(c-Glu-Gly). The excitation wavelength was set at 300 nm and
the emission range from 420 to 650 nm.
Ó FEBS 2003 Glutathione conjugation by S23906-1 (Eur. J. Biochem. 270) 2855
permits a more pronounced alkylation of DNA. The
capacity of the drug to react with the genomic DNA is
inversely proportional to the concentration of GSH. The
higher the GSH level, the lower the amount of drug–DNA
adduct and vice versa. The increased cytotoxic effect
observed in the presence of 1-m
M
BSO is weak (about

twofold) but consistent (Fig. 10C). Altogether, these experi-
ments demonstrate that GSH reduces the cytotoxic action
of S23906-1 by decreasing the formation of lethal DNA–
drug covalent adducts.
Discussion
The compound S23906-1, a diester derivative of 1,2-
dihydrobenzo[b]acronycine, has been recently identified as
a highly potent and promising antitumor agent [10]. The
pharmacological profile of this drug is unusual in the sense
that it is markedly active in orthotopic models of human
solid tumors, even by the oral route, but only moderately
active against murine transplantable tumors [9]. It is now
well established that DNA is a potential target for this
compound. We have recently demonstrated that the drug
alkylates the N2 position of guanine residues exposed in the
minor groove of double helical DNA [11]. Moreover, a very
recent structure–activity study strongly suggests that the
formation of DNA–S23906-1 covalent complexes is respon-
sible for the cytotoxic action [12]. Nevertheless, the
antitumor activity of a given drug rarely (and probably
never) relies on the interaction with a single molecular
target. The possible implication of other targets in the
cytotoxic action of S23906-1 must be kept in mind.
The relative intracellular abundance of GSH (0.5–
10 m
M
) and the nucleophilicity of its thiolate ion prompted
us to investigate its reactivity with S23906-1. The possibility
that bonding of the drug to GSH decreases the extent of
DNA adducts and/or other cellular nucleophiles, allowed us

confirm that alkylation is the principal mechanism of action
of S23906-1. The variety of experimental data reported in
the present study, summarized in Table 1, fully demon-
strates that the hypothesis was correct: S23906-1 does bind
covalently to GSH. The complex formation requires the SH
group of the Cys residue of the tripeptide. The drug reacts
neither with GSSG nor with other S-blocked derivatives
such as GS-NO, GS-DCE or GS-SA. A free SH group is
required but it is not sufficient because not all SH-
containing compounds react covalently with S23906-1.
For example, additional MS analyses (not presented here)
indicated that the drug does not bind covalently to
dithiothreitol or the SH protein, thioredoxin. The cysteine
SH group of GSH is necessary, but not sufficient, for
covalent complex formation. The use of the model dipep-
tides suggests that the drug preferentially recognizes the
Cys-GlymoietyofGSH.
The experiments performed with the human KB-3-1
carcinoma cell line indicate that GSH modulates the
cellular response to S23906-1 by inhibiting DNA alkyla-
tion, thereby decreasing the formation of potentially lethal
DNA adducts. At first sight, this observation may have
important biological implications because in vivo the drug
will encounter large quantities of GSH and other thiol-
containing substrates before it can reach the nucleus of
tumor cells. However, despite this reactivity towards thiols,
S23906-1 presents very high antitumor activities and even
shows curative effects in vivo in certain tumor models
[9,10].
Although covalent binding to GSH is frequently

observed with DNA alkylating agents, the type of adducts
formed and their biological effects often vary significantly
Fig. 10. Effect of buthionine sulfoximine (BSO) on the cytotoxicity of
S23906-1 and the formation of S23906-1–DNA covalent complexes in
cells. (A) Effect of BSO on intracellular glutathione (GSH) contents.
KB-3-1 cells were treated without (plain line) or with 0.1-, 0.25-, 0.5-, 1-
or 10-m
M
BSO (dashed lanes) for 24 h at 37 °C prior to lysis, and the
intracellular glutathione (GSH) contents were quantified using
the ThioGlo-1
TM
reagent. The excitation wavelength was 360 nm. The
fluorescence intensity was expressed as a percentage of the control
value (plain lane). (B) Fluorescence emission spectra of DNA extracted
from KB-3-1 cells treated or untreated with increasing concentrations
of BSO (0.1, 0.25, 0.5, 1 or 10 m
M
) (dashed lines) for 24 h prior to the
addition of 10 l
M
S23906-1 for 1 h (plain line). The excitation wave-
length was set at 300 nm. The fluorescence intensity was expressed as a
percentage of the control value. (C) Growth-inhibition curves for (j)
S23906-1 alone and (n) S23906-1 in the presence of 1-m
M
BSO. KB-3-
1 cells were treated for 72 h with S23906-1 prior to measuring the
viability using a conventional tetrazolium-based assay.
2856 M H. David-Cordonnier et al. (Eur. J. Biochem. 270) Ó FEBS 2003

from one drug to another. The drugs can be classified
into two categories, depending on the positive or negative
contribution of the GSH–drug adducts to the DNA
reactivity. The first group of antitumor agents activated
by reaction with GSH includes, for example, the natural
product leinamycin which reacts with thiols to generate an
electrophilic episulfonium ion then reacting with the N7
position of guanines in DNA [19]. This group also includes
the promising anticancer drug, irofulven (6-hydroxymethyl-
acylfulvene, also known as MGI 114 or HMAF), which is
thought to alkylate amines in DNA. This semisynthetic
derivative of illudin S is currently in phase II chemothera-
peutic clinical trials for a variety of solid tumors [20,21].
Reaction of irofulven with GSH activates a diene inter-
mediate for nucleophilic attack of DNA [22,23]. Another
prominent member of this class of thiol-activated antitumor
agents is the distamycin-a-bromoacrylic derivative PNU-
166196, designated brostallicin, which is currently under-
going phase II clinical trials [24]. The cytotoxic action of this
DNA minor-groove-binding agent is significantly enhanced
in the presence of GSH and this effect is believed to
originate from the formation of GSH–brostallicin covalent
adducts [25]. S23906-1 clearly does not belongs to this group
as, in the present case, the GSH–benzoacronycine adducts
reduce the cytotoxic action of the drug. It should be
included in the second group of compounds for which the
covalent binding to GSH inactivates the antitumor agent.
Similar behaviors have been reported with a variety of
DNA alkylators, especially the nitrogen mustards [26,27],
but these drugs have no real, inherent, affinity for DNA and

thus over-expression of GSH leads to competition between
concentrations of nucleophiles. In the case of S23906-1, the
affinity for DNA probably pulls the competition between
the DNA nucleophile and GSH towards the former. This
may explain the continuing biological activity and DNA-
alkylation ability of the agent, even in the presence of high
levels of GSH. Covalent binding to GSH has also been
observed with the drug mitomycin C (MC) [28,29]. The
formation of GSH–MC conjugates competes with DNA
alkylation, as is the case with S23906-1. Ternary GSH–MC–
guanine N2 DNA adducts have been isolated and charac-
terized [30]. The antitumor drug, cisplatinum, also reacts
covalently with GSH and cysteine [31,32], as well as the
pyrrolobenzodiazepine dimers, which form interstrand
covalent DNA crosslinks and similarly alkylate guanines
at their N2 positions in the duplex minor groove. Their
cytotoxicity is also modulated by GSH and other nonpro-
tein thiols, where reversible adducts are formed [33,34].
The formation of covalent adducts between S23906-1 and
GSH can be viewed as a detoxification system. Considering
that the intracellular concentration of S23906-1 in the
tumors is probably very low, whereas the intracellular
concentration of GSH is relatively high (generally >1 m
M
),
one can assume that the formation of GS–benzoacronycine
adducts will be favored and that these adducts can be
eliminated by different transporters, such as the multidrug-
resistance-associated protein (MRP) which functions as a
GSH S-conjugate carrier in leukemia and in lung carcinoma

cells [35–37]. However, the fate of the GS-S23906-1 adducts
are as yet unknown and at this stage we cannot eliminate the
possibility that these glutathionyl conjugates remain capable
of alkylating macromolecules and thus serve as a transport
form and a reservoir for the drug. For example, the
monoGSH-conjugate of cyclophophasmide can reform the
tumor active metabolite 4-hydroxy-cyclophophasmide [38].
Although the GSH-S23906-1 adducts are apparently very
stable and probably eliminated as such, we cannot reject
the possibility that in a specific cell/tissue environment, the
conjugates hydrolyze to generate a potentially alkylating
drug species. More work on the fate and distribution of the
GSH-conjugates of the benzoacronycine is required. More
information is also needed on the reactivity of the drug
towards GSH S-transferases which often participate in the
development of drug resistance. The role of GSH-dependent
DNA repair should also be investigated to better compre-
hend the mechanism of action of S23906-1. There is also the
possibility that the GSH–S23906-1 adducts detected here
using an in vitro system, either do not form or form only
very weakly in vivo. For example, the formation of cisplatin–
glutathione adducts was found in vitro, but not in vivo after
concomitant administration of cisplatin and glutathione to
rats and cancer patients [39].
The primary objective of the present study was to
investigate further the reactivity of S23906-1 towards the
bionucleophiles DNA and GSH. The covalent binding of
the drug to GSH reduced the formation of DNA adducts
and slightly decreased the cytotoxic potential of the
molecule. There is now little room for doubt that DNA is

an important target of S23906-1 and the reaction mechan-
ism clearly implicates the C1 functionality. The present
study, demonstrating the formation of covalent adducts
between GSH and the antitumor drug S23906-1, sets several
directions for further works to enhance our understanding
of the mechanism of action of this promising anticancer
agent.
Acknowledgements
The authors thanks Dr John A. Hickman (Head of the Cancer
Research Division, Institut de Recherches Servier) for stimulating
discussions and useful comments on the manuscript and Dr B. Serkis
(Division of Physicochemistry, Institut de Recherches Servier) for the
separation of the two cis enantiomers of S23906-1. This work was
carried out under the support of a Servier research grant to C. B. and a
fellowship to M H. D C. from the Association pour la Recherche sur
le Cancer.
References
1.Hughes,G.K.,Lahey,F.N.,Price,J.R.&Webb,L.J.(1948)
Alkaloids of the Australian rutaceae. Nature 162, 223–224.
2. Svoboda, G., Poore, G.A., Simpson, P.J. & Boder, G.B. (1966)
Alkaloids of Acronychia Baueri Schott I. Isolation of the alkaloids
and a study of the antitumor and other biological properties of
acronycine. J. Pharm. Sci. 55, 759–768.
3. Gout,P.W.,Dunn,B.P.&Beer,C.T.(1971)Effectsofacronycine
on nucleic acid synthesis and population growth in mammalian
tumor cell cultures. J. Cell Physiol. 78, 127–138.
4. Dunn, B.P., Gout, P.W. & Beer, C.T. (1973) Effects of the anti-
neoplastic alkaloid acronycine on nucleoside uptake and incor-
poration into nucleic acids by cultured L5178Y cells. Cancer Res.
33, 2310–2319.

5. Dorr, R.T., Liddil, J.D., Von Hoff, D.D., Soble, M. & Osborne,
C.K. (1989) Antitumor activity and murine pharmacokinetics of
parenteral acronycine. Cancer Res. 49, 340–344.
Ó FEBS 2003 Glutathione conjugation by S23906-1 (Eur. J. Biochem. 270) 2857
6. Elomri, A., Mitaku, S., Michel, S., Skaltsounis, A.L., Tillequin, F.,
Koch, M., Pierre
´
, A., Guilbaud, N., Le
´
once, S., Kraus-Berthier,
L., Rolland, Y. & Atassi, G. (1996) Synthesis and cytotoxic and
antitumor activity of esters in the 1,2-dihydroxy-1,2-dihydro-
acronycine series. J. Med. Chem. 39, 4762–4766.
7. Costes, N., Le Deit, H., Michel, S., Tillequin, F., Koch, M.,
Pfeiffer, B., Renard, P., Le
´
once, S., Guilbaud, N., Kraus-Berthier,
L., Atassi, G. & Pierre
´
, A. (2000) Synthesis and cytotoxic and
antitumor activity of benzo[b]pyrano[3,2-h]acridin-7-one ana-
logues of acronycine. J. Med. Chem. 43, 2395–2402.
8. Michel, S., Seguin, E. & Tillequin, F. (2002) Structure–activity
relationships in the acronycine series. Curr.Med.Chem.9, 1689–
1700.
9. Guilbaud, N., Kraus-Berthier, L., Meyer-Losic, F., Malivet, V.,
Chacun, C., Jan, M., Tillequin, F., Koch, M., Pfeiffer, B., Atassi,
G.,Hickman,J.&Pierre
´
, A. (2001) Marked antitumor activity of

a new potent acronycine derivative in orthotopic models of human
solid tumors. Clin. Cancer Res. 7, 2573–2580.
10. Guilbaud, N., Le
´
once, S., Tillequin, F., Koch, M., Hickman, J.A.
&Pierre
´
, A. (2002) Acronycine derivatives as promising antitumor
agents. Anticancer Drugs 13, 445–449.
11. David-Cordonnier, M.H., Laine, W., Lansiaux, A., Kouach, M.,
Briand, G., Pierre
´
,A.,Hickman,J.A.&Bailly,C.(2002)Alky-
lation of guanine in DNA by S23906-1, a novel potent antitumor
compound derived from the plant alkaloid acronycine. Biochem-
istry 41, 9911–9920.
12. Thi Mai, H.D., Gaslonde, T., Michel, S., Tillequin, F., Koch,
M., Bongui, J B., Elomri, A., Seguin, E., Pfeiffer, B., Renard,
P., David-Cordonnier, M H., Tardy, C., Laine, W., Bailly, C.,
Kraus-Berthier, L., Le
´
once, S., Hickman, J.A. & Pierre
´
,A.
(2003) Structure activity relationships and mechanism of action of
antitumor benzo[3,2-H]acridin-7-one analogs of acronycine.
J. Med. Chem in press.
13. Le
´
once, S., Perez, V., Lambel, S., Peyroulan, D., Tillequin, F.,

Michel,S.,Koch,M.,Pfeiffer,B.,Atassi,G.,Hickman,J.A.&
Pierre
´
, A. (2001) Induction of cyclin E and inhibition of DNA
synthesis by the novel acronycine derivative S23906-1 precede the
irreversible arrest of tumor cells in S phase leading to apoptosis.
Mol. Pharmacol. 60, 1383–1391.
14. Kluza, J., Lansiaux, A., Wattez, N., Hildebrand, M.P., Le
´
once, S.,
Pierre
´
, A.,Hickman, J.A. & Bailly, C. (2002) Induction of apop-
tosis in HL-60 leukemia and B16 melanoma cells by the
acronycine derivative S23906-1. Biochem. Pharmacol. 63, 1443–
1452.
15. Le
´
once, S., Pierre
´
,A.,Anstett,M.,Perez,V.,Genton,A.,Bizzari,
J.P. & Atassi, G. (1992) Effects of a new triazinoaminopiperidine
derivative on adriamycin accumulation and retention in cells dis-
playing P-glycoprotein-mediated multidrug resistance. Biochem.
Pharmacol. 44, 1707–1715.
16. Bailey, H.H. (1998) L-S,R buthionine sulfoximine: historical
development and clinical issues. Chem. Biol. Interact. 111–112,
239–254.
17. Gamcsik, M.P., Kasibhatla, M.S., Adams, D.J., Flowers, J.L.,
Colvin, O.M., Manikumar, G., Wani, M., Wall, M.E., Kohlha-

gen, G. & Pommier, Y. (2001) Dual role of glutathione in mod-
ulating camptothecin activity: depletion potentiates activity, but
conjugation enhances the stability of the topoisomerase I–DNA
cleavage complex. Mol. Cancer Ther. 1, 11–20.
18. Kagan, V.E., Yalowich, J.C., Borisenko, G.G., Tyurina, Y.Y.,
Tyurin,V.A.,Thampatty,P.&Fabisiak,J.P.(1999)Mechanism-
based chemopreventive strategies against etoposide-induced acute
myeloid leukemia: free radical/antioxidant approach. Mol. Phar-
macol. 56, 494–506.
19. Gates, K.S. (2000) Mechanisms of DNA damage by leinamycin.
Chem. Res. Toxicol. 13, 953–956.
20. Berg, W.J., Schwartz, L.,Yu, R., Mazumdar, M. & Motzer, R.J.
(2001) Phase II trial of irofulven (6-hydroxymethylacylfulvene) for
patients with advanced renal cell carcinoma. Invest. New Drugs 19,
317–320.
21. Dowell, J.E., Johnson, D.H., Rogers, J.S., Shyr, Y., McCullough,
N., Krozely, P. & DeVore, R.F. (2001) A phase II trial of
6-hydroxymethylacylfulvene (MGI-114, irofulven) in patients with
advanced non-small cell cancer previously treated with chemo-
therapy. Invest. New Drugs 19, 85–88.
22. McMorris, T.C. (1999) Discovery and development of sesqui-
terpenoid derived hydroxymethylacylfulvene: a new anticancer
drug. Bioorg. Med. Chem. 7, 881–886.
23. Wolkenberg, S.E. & Boger, D.L. (2002) Mechanisms of in situ
activation for DNA-targeting antitumor agents. Chem. Rev. 102,
2477–2495.
24. Cozzi, P. (2001) A new class of cytotoxic DNA minor groove
binders: alpha-halogenoacrylic derivatives of pyrrolecarbamoyl
oligomers. Farmaco 56, 57–65.
25. Geroni, C., Marchini, S., Cozzi, P., Galliera, E., Ragg, E.,

Colombo, T., Battaglia, R., Howard, M., D’Incalci, M. & Broggini,
M. (2002) Brostallicin, a novel anticancer agent whose activity is
enhanced upon binding to glutathione. Cancer Res. 62, 2332–2336.
26. Dirven, H.A., van Ommen, B. & van Bladeren, P.J. (1996) Glu-
tathione conjugation of alkylating cytostatic drugs with a nitrogen
mustard group and the role of glutathione S-transferases. Chem.
Res. Toxicol. 9, 351–360.
27. Rothbarth, J., Vahrmeijer, A.L. & Mulder, G.J. (2002) Modula-
tion of cytostatic efficacy of melphalan by glutathione: mechan-
isms and efficacy. Chem. Biol. Interact. 140, 93–107.
28. Paz, M.M., Das, A., Palom, Y., He, Q.Y. & Tomasz, M. (2001)
Selective activation of mitomycin A by thiols to form DNA cross-
links and monoadducts: biochemical basis for the modulation of
mitomycin cytotoxicity by the quinone redox potential. J. Med.
Chem. 44, 2834–2842.
29. Sharma, M. & Tomasz, M. (1994) Conjugation of glutathione and
other thiols with bioreductively activated mitomycin C. Effect
of thiols on the reductive activation rate. Chem. Res. Toxicol. 7,
390–400.
30. Sharma, M., He, Q.Y. & Tomasz, M. (1994) Effects of glutathione
on alkylation and cross-linking of DNA by mitomycin C. Iso-
lation of a ternary glutathione–mitomycin–DNA adduct. Chem.
Res. Toxicol. 7, 401–407.
31. Mistry, P., Lee, C. & McBrien, D.C. (1989) Intracellular meta-
bolites of cisplatin in the rat kidney. Cancer Chemother. Phar-
macol. 24, 73–79.
32. Volckova, E., Dudones, L.P. & Bose, R.N. (2002) HPLC
determination of binding of cisplatin to DNA in the presence of
biological thiols: implications of dominant platinum-thiol binding
to its anticancer action. Pharm. Res. 19, 124–131.

33. Thurston, D.E., Bose, D.S., Thompson, A.S., Howard, P.W.,
Leoni, A., Croker, S.J., Jenkins, T.C., Neidle, S., Hartley, J.A. &
Hurley, L.H. (1996) Synthesis of sequence-selective C8-linked
pyrrolo[2,1-c][1,4]benzodiazepine DNA interstrand cross-linking
agents. J. Org. Chem. 61, 8141–8147.
34.Gregson,S.J.,Howard,P.W.,Hartley,J.A.,Brooks,N.A.,
Adams, L.J., Jenkins, T.C., Kelland, L.R. & Thurston, D.E.
(2001) Design, synthesis, and evaluation of a novel pyrrolo-
benzodiazepine DNA-interactive agent with highly efficient cross-
linking ability and potent cytotoxicity. J. Med. Chem. 44, 737–748.
35. Jedlitschky, G., Leier, I., Buchholz, U., Center, M. & Keppler, D.
(1994) ATP-dependent transport of glutathione S-conjugates
by the multidrug resistance-associated protein. Cancer Res. 54,
4833–4836.
36. Zaman, G.J., Lankelma, J., van Tellingen, O., Beijnen, J., Dekker,
H., Paulusma, C., Elferink, R.P., Baas, F. & Borst, P. (1995) Role
2858 M H. David-Cordonnier et al. (Eur. J. Biochem. 270) Ó FEBS 2003
of glutathione in the export of compounds from cells by the
multidrug-resistance-associated protein. Proc. Natl Acad. Sci.
USA 92, 7690–7694.
37. Borst, P., Evers, R., Kool, M. & Wijnholds, J. (2000) A family of
drug transporters: the multidrug resistance-associated proteins.
J. Natl Cancer Inst. 92, 1295–1302.
38. Dirven, H.A., Venekamp, J.C., van Ommen, B. & van
Bladeren, P.J. (1994) The interaction of glutathione with
4-hydroxycyclophosphamide and phosphoramide mustard, stu-
died by 31P nuclear magnetic resonance spectroscopy. Chem. Biol.
Interact. 93, 185–196.
39. Bernareggi, A., Torti, L., Facino, R.M., Carini, M., Depta, G.,
Casetta, B., Farrell, N., Spadacini, S., Ceserani, R. & Tognella, S.

(1995) Characterization of cisplatin-glutathione adducts by liquid
chromatography-mass spectrometry. Evidence for their formation
in vitro but not in vivo after concomitant administration of cis-
platin and glutathione to rats and cancer patients. J. Chromatogr.
B Biomed. Appl. 669, 247–263.
Ó FEBS 2003 Glutathione conjugation by S23906-1 (Eur. J. Biochem. 270) 2859