Synthesis of 5-nitro-2-furancarbohydrazides and their cis- diamminedichloroplatinum complexes as bitopic and irreversible human thioredoxin reductase inhibitors

Article abstract of DOI:10.1021/jm050256l

The human selenoprotein thioredoxin reductase is involved in antioxidant defense and DNA synthesis. As increased thioredoxin reductase levels are associated with drug sensitivity to cisplatin and drug resistance in tumor cells, this enzyme represents a promising target for the development of cytostatic agents. To optimize the potential of the widely used cisplatin to inhibit the human thioredoxin reductase and therefore to overcome cisplatin resistance, we developed and synthesized four cis-diamminedichloroplatinum complexes of the lead 5-nitro-2-furancarbohydrazide 8 selected from high-throughput screening. Detailed kinetics revealed that the isolated fragments, 5-nitro-2-furancarbohydrazide and cisplatin itself, bind with micromolar affinities at two different subsites of the human enzyme. By tethering both fragments four nitrofuran-based cis-diamminedichloroplatinum complexes 13a-c and 20 were synthesized and identified as biligand irreversible inhibitors of the human enzyme with nanomolar affinities. Studies with mutant enzymes clearly demonstrate the penultimate selenocysteine residue as the prime target of the synthesized cJs-diamminedichloroplatinum complexes.

Full text of DOI:10.1021/jm050256l

7024
J. Med. Chem. 2005, 48, 7024-7039
Synthesis of 5-Nitro-2-furancarbohydrazides and Their
cis-Diamminedichloroplatinum Complexes as Bitopic and Irreversible Human
Thioredoxin Reductase Inhibitors
,X
Re´gis Millet,†, Sabine Urig,#, Judit Jacob,§ Eberhard Amtmann,| Jacques-Philippe Moulinoux,‡
Stephan Gromer,§ Katja Becker,# and Elisabeth Davioud-Charvet†,§,
*
UMR 8525 CNRS-Universite´ Lille II-Institut Pasteur de Lille, Institut de Biologie de Lille, 1 rue du Professor Calmette,
BP447 59021 Lille Cedex, France, Interdisciplinary Research Center, Giessen University, Heinrich Buff-Ring 26-32,
D-35392 Giessen, Germany, Department D060, German Cancer Research Centre, Im Neuenheimer Feld 280,
D-69120 Heidelberg, Germany, Groupe de Recherche en The´rapeutique Anticance´reuse, Universite´ Rennes I,
2 Av du Professeur Le´on Bernard, 35043 Rennes Cedex, France, and Biochemie-Zentrum der Universita¨t Heidelberg,
Im Neuenheimer Feld 504, D-69120 Heidelberg, Germany
Received March 21, 2005
The human selenoprotein thioredoxin reductase is involved in antioxidant defense and DNA
synthesis. As increased thioredoxin reductase levels are associated with drug sensitivity to
cisplatin and drug resistance in tumor cells, this enzyme represents a promising target for the
development of cytostatic agents. To optimize the potential of the widely used cisplatin to inhibit
the human thioredoxin reductase and therefore to overcome cisplatin resistance, we developed
and synthesized four cis-diamminedichloroplatinum complexes of the lead 5-nitro-2-furan-
carbohydrazide 8 selected from high-throughput screening. Detailed kinetics revealed that the
isolated fragments, 5-nitro-2-furancarbohydrazide and cisplatin itself, bind with micromolar
affinities at two different subsites of the human enzyme. By tethering both fragments four
nitrofuran-based cis-diamminedichloroplatinum complexes 13a-c and 20 were synthesized
and identified as biligand irreversible inhibitors of the human enzyme with nanomolar affinities.
Studies with mutant enzymes clearly demonstrate the penultimate selenocysteine residue as
the prime target of the synthesized cis-diamminedichloroplatinum complexes.
Introduction
Thioredoxin (Trx) and TrxR form the thioredoxin
system, which is involved in a magnitude of cellular
functions,4 including DNA synthesis and cell signaling,
rendering it a suitable target for chemotherapeutic
agents per se. Several independent knock-out5,6,7,8,9 or
mutant10 studies indicate the vital function of both
proteins from various species. Furthermore, recent
reports provide strong evidence for an important
role of the human thioredoxin system in cisPt-
resistance.11
The catalytic mechanism of large TrxRs is similar to
that of glutathione reductase (GR) by sharing similar
NADPH- and FAD-binding domains, a dimer interface
domain, and the thiol-disulfide redox active center Cys-
Val-Asn-Val-Gly-Cys located in the FAD domain of the
N-terminal part of each subunit (Scheme 1). The flow
of electrons during catalysis is from NADPH to FAD to
the active center disulfide and then to the C-terminal
redox center formed by a cysteine-selenocysteine couple
at the other TrxR subunit. A key feature of the proposed
catalytic mechanism of large TrxR is the motion of the
flexible C-terminal tail that is responsible for the
transport of electrons from the buried N-terminal redox
center to large substrates as Trx at the surface of the
protein (Scheme 1).12-16 Data from the three-dimen-
sional structure of mammalian TrxR,17 and from selec-
tive cleavage13 corroborate this mechanistic model.
Beyond the reduction of the natural substrate thio-
redoxin, the Cys497-Sec498 redox pair reacts with a
large variety of structurally diverse low Mr substrates.3
A major obstacle in anticancer chemotherapy is the
emergence of drug resistance, leading to a reemergence
of a tumor which initially responded well to treatment.
Cis-diamminedichloroplatinum(II), or cisplatin (cisPt),
is a commonly used drug in cancer therapy,1 particularly
prone to resistance, as an increase in dosage is only
feasible to a limited extend due to severe toxic side
effects. Many of these resistance mechanisms are di-
rectly or indirectly linked to the cellular redox system
protecting the cells by formation of drug conjugates for
export, by counteracting apoptotic signals and by other
mechanisms.2
One of the key players in the cellular redox system is
thioredoxin reductase (TrxR, EC 1.8.1.9).3 TrxR cata-
lyzes the NADPH-dependent reduction of oxidized thio-
redoxin (Trx(S)2) according to eq 1:
NADPH + Trx(S)2 h NADP+ + Trx(SH)2 (1)
* Correspondence to: Elisabeth Davioud-Charvet, delegate of
Centre National de la Recherche Scientifique, France, in the frame of
a French-German cooperation with the University of Heidelberg,
Germany. Phone: +49-6221-54-4188; Fax: +49-6221-54-5586; E-mail:
elisabeth.davioud@gmx.de.
† UMR 8525 CNRS-Universite´ Lille II-Institut Pasteur de Lille.
# Giessen University.
| German Cancer Research Centre.
‡ Universite´ Rennes I.
§ Biochemie-Zentrum der Universita¨t Heidelberg.
Both authors contributed equally to this work.
X Present address: I.C.P.A.L. 3, rue du Pr Laguesse, BP 83, F-59006
Lille, France.
10.1021/jm050256l CCC: $30.25 © 2005 American Chemical Society
Published on Web 10/07/2005

Human Thioredoxin Reductase Inhibitors
Journal of Medicinal Chemistry, 2005, Vol. 48, No. 22 7025
Scheme 1. Postulated Catalytic Mechanism of Human
Chart 1. Structures of the
TrxR and Other Large TrxRs for Trx Reduction3,13,53
5-Nitro-2-furancarbohydrazides 8, 14b and Related
Derivatives 12a-c, 19 and Their
cis-Diamminedichloroplatinum Complexes 13a-c and
20 as Inhibitors of Human Thioredoxin Reductase
TrxR’s unique C-terminal redox center allowing ir-
reversible inhibition. In theory, the advantage of such
a bitopic ligand is a more efficient TrxR inhibition than
either of the parent compounds alone. The binding
affinity of a linked compound is, in principle, the product
of the binding constants of the individual fragments plus
a term that accounts for the changes in binding affinity
that are due to the linker.22 An additional effect of DNA
destruction by cisPt is expected since DNA synthesis
and repair depend on the formation of desoxyribonucleo-
tides, which itself is dependent on Trx(SH)2. A syner-
gistic effect should result when taking into account
that resistance formation is to some degree dependent
on Trx and TrxR. It is noteworthy that many cisPt
analogues linked to various DNA ligands to improve
DNA-targeting ability, have been described and evalu-
ated,23-27 although the rationales have been different
from the approach proposed here.
In the present paper, the starting lead inhibitor was
selected from the primary high-throughput screening
of a general library of compounds (unpublished data)
in the hTrxR assay using an artificial disulfide sub-
strate, 5,5′-dithiobis(2-nitrobenzamide) (DTNBA).28 The
most active compound was identified to be a 5-nitro-2-
furancarbohydrazide derivative 8 with a naphthalene
moiety (Chart 1). Various nitroaromatic compounds,
including nitrofuran derivatives, are known to bind to
several disulfide reductases by acting both as inhibitors
and as substrates.29,30 Starting from the lead structure
we report on the synthesis of four platinum-chelating
ethanediamine ligands 12a-c and 19 linked to the lead
5-nitro-2-furancarbohydrazide derivative 8 and ana-
logues, and on the inhibitory potency of their corre-
sponding cis-diamminedichloroplatinum(II) complexes
13a-c and 20 (Chart 1). The four complexes were
synthesized by tethering the nitrofuran moiety to the
cisPt moiety with linkers at different positions to the
naphthalene group. The binding affinities of these four
In this sketch, only one reaction center, yet formed by both
subunits of the homodimeric mammalian TrxR (indicated by black
and gray lines), is shown. The flavin near the N-terminal redox
active site (Cys59 and Cys64) is provided by one subunit, and the
C-terminal redox active site of the same reaction center by the
other subunit (Cys′497 and Sec′498). The oxidized enzyme (Eox
)
is reduced to an EH2 species by NADPH.12 The N-terminal redox
active site exchanges the electrons with the C-terminal redox
active site of the opposite subunit.16,31 Additional reducing equiva-
lents provided by NADPH are taken up to yield an EH4-species.12
Selective digest experiments suggest that the reduced C-terminal
tail now moves to a more solvent exposed position.13,14 Oxidized
thioredoxin reacts with the reduced C-terminal tail’s selenolate
to yield a mixed selenenyl sulfide, which is cleaved by the adjacent
thiol,31 to yield reduced thioredoxin and the initial TrxR-EH2
species. Steady-state kinetics demonstrated an overall ping-pong
mechanism as indicated by this model.19 Note D. melanogaster
TrxR has a disulfide, not a selenenyl sulfide as in mammalian
TrxR.
Selenols have a much higher reactivity to bind heavy
metal ions than thiols.18 Therefore, as previously hy-
pothesized, the easily accessible C-terminally located
selenocysteine of hTrxR is most likely to be the major
target of a variety of more or less effective metal-
containing inhibitors.19,20,21 One of those inhibitors,
although rather weak, is cisPt.11
As the cellular mode of cisPt action is primarily the
formation of covalent adducts with DNA, it may con-
tribute to its effectiveness and to reduction of resistance
formation if its specificity toward hTrxR is increased.
The optimization of cisPt as hTrxR inhibitor could be
reached by following the biligand approach, i.e., linking
the cisPt-structure to another moiety that interacts with
the N-terminal redox-active site of the enzyme. During
catalysis this region can be assumed to be exposed as
the reduced C-terminal tail moves toward a more
solvent exposed position. The N-terminal interaction
might temporarily block the enzyme and should fur-
thermore trap the cisPt-moiety in close proximity to the

7026 Journal of Medicinal Chemistry, 2005, Vol. 48, No. 22
Millet et al.
complexes for TrxRs were measured in Trx and DTNB
disulfide reduction assays and were found to be effective
in nanomolar concentrations for irreversible binding to
TrxRs. To verify our proposed inhibitory mechanism of
the four cis-diamminedichloroplatinum(II) complexes
13a-c and 20, detailed kinetic studies were performed.
Apart from the authentic human TrxR, we also included
a C-terminally truncated enzyme, a mutant lacking the
final 16 amino acids (hTrxR∆-16) and a SecfCys
mutant form of hTrxR (Sec498Cys).
As it has previously been shown that the microenvi-
ronment of the C-terminal redox center is of importance
as well, we included C-terminal mutants of Drosophila
melanogaster thioredoxin reductase (DmTrxR),31 a cys-
teine homologue of the human enzyme, to analyze this
aspect with respect to the alkylating properties of the
most potent compound 20.
The data are in good agreement with previous observa-
tions made with nitrofuran and nitrophenyl deriva-
tives,30,34 which possess single-electron reduction po-
tentials between -255 and -287 mV.
Characterization of Inhibition by Lead Com-
pound 8 in the Steady-State. To compare the binding
of the inhibitor to the N- and/or the C-terminal parts of
TrxR, these studies were done using the wild-type
enzyme (wt-hTrxR) and the truncated enzyme (hTrxR∆-
16) lacking the second redox center (deletion of 16 amino
acids). The suitable TrxR assay based on Trx reduction
could not be used because Trx is only reduced at the
C-terminal redox center of wild-type enzyme. Thus,
DTNB was selected as artificial substrate to allow
comparison of the inhibition type of the lead compound
8 with wild-type TrxR and truncated enzyme. DTNB is
known to be reduced at both active sites of wild-type
enzymes, e.g. of Plasmodium falciparum TrxR, but with
14-fold reduced catalytic efficiency kcat/Km at the N-
terminal part.35 This mainly resulted from higher Km
value for DTNB and reduced kcat, as suggested from the
determined Km and kcat values for DTNB reduction at
the N-terminal binding site of the human truncated
TrxR. Using eight different substrate concentrations
(11-435 µM) in the absence of inhibitor, DTNB is
reduced by hTrxR∆-16 with Km and kcat values of 5305.9
( 67.5 µM and 0.79 s-1, versus values of 50.5 ( 1.6 µM
and 20.79 s-1, when reduced by wt-hTrxR. This accounts
for a 2825-fold reduced catalytic efficiency kcat/Km at the
N-terminal part of the human enzyme.
Lead Compound 8 Follows Uncompetitive Ki-
netics in Inhibition of Wild-Type Human TrxR. To
investigate the outcome of wt-hTrxR inhibition at vary-
ing DTNB concentrations (11-435 µM), kinetics were
first performed using saturating NADPH concentration
(200 µM) in the presence of inhibitor 8 (0-40 µM). The
experimental data for wild-type enzyme were analyzed
using eq 4 for uncompetitive inhibition (Figure 1A),
where [S] is the concentration of varied substrate, [I] is
the inhibitor concentration, Vmax is the maximal velocity,
Km is the Michaelis-Menten constant, and Ki is the
inhibition constant for I binding to the ES complex.
Results
Inhibitor Screening on hTrxR. In a high-through-
put screening 12 000 compounds were analyzed in
microtiter plates as potential hTrxR inhibitors at a
concentration of 25 µM in the presence of 500 µM
NADPH and 200 µM of the artificial disulfide substrate
DTNBA.28 Three known TrxR inhibitors, dichloroindo-
phenol,32 naphthazarin,21 and cisPt,11,33 were selected
as references. As judged from the IC50 values, only one
compound, N′-(2-naphthylmethyl)-5-nitro-2-furancarbo-
hydrazide 8 (IC50 value )16 µM), was found to be more
potent than the three reference inhibitors, dichloroindo-
phenol (IC50 ) 25 µM), naphthazarin (IC50 ) 20 µM),
and especially cisPt (no inhibition at 25 µM, IC50 . 50
µM) in the TrxR assay using 50 µM DTNBA and 200
µM NADPH. It is noteworthy to mention that any of
the intermediates 9a-c to 12a-c and 14a-c to 19,
synthesized after further chemical derivatization of the
lead structure 8, showed IC50 values below 40 µM under
the same conditions. The most potent TrxR inhibitors,
the lead inhibitor 8, and the four related cisPt analogues
13a-c and 20 were investigated in inhibition studies.
Nitrofuran Reductase Activity of Thioredoxin
Reductase. The ability of TrxR to reduce the nitrofuran
derivative was studied by following the oxidation of
NADPH in the presence of the lead inhibitor 8. The
nitrofuran reductase activity of TrxR was compared
with the low intrinsic NADPH oxidase activity of the
enzyme in the absence of nitrofuran (NF) (eq 2). TrxR
displayed a 21-fold higher NADPH oxidase activity in
the presence of 300 µM 8 (eq 3) (0.043 U mg-1 protein
in the absence versus 0.908 U mg-1 protein in the
presence of 300 µM nitrofuran 8).
Vmax[S]
v )
(4)
[I]
Km + [S] 1 +
(
)
Ki
The Ki value of 16.0 ( 0.8 µM for compound 8 was
determined as shown in Figure 1 (see inset in Figure
1A). The Cornish-Bowden plot (Figure 1B) as well as
the Dixon (Figure 1C) and Lineweaver-Burk (Figure
1D) plots for compound 8 was consistent with uncom-
petitive inhibition type, whereas expressions for com-
petitive and noncompetitive inhibition could not accu-
rately describe the observed relationship between Ki and
DTNB concentration. Under steady-state conditions, it
clearly appeared from the Cornish-Bowden plot (Figure
1B) that binding of the inhibitor was facilitated by
DTNB binding, revealing an induced cooperative effect.
A detailed analysis of this plot showed that, as DTNB
concentration increases, the constant Ki for the dissocia-
tion of I appeared by itself in a slope and reached a
maximum, approximately 12 µM, when [DTNB] is in
the Km range (see inset in Figure 1A).
2 O2 + NADPH + H+ f 2 O2-• + NADP+ + 2H+
(enzymatic) (2)
2 NF + NADPH + H+ f 2 NF-• + NADP+ + 2H+
(enzymatic) (3)
The Km (529 ( 80 µM) and kcat (2.5 ( 0.3 s-1 per
catalytic subunit) values were derived from measure-
ments at seven different substrate concentrations. Fol-
lowing NADPH consumption, 8 was reduced by TrxR
with a catalytic efficiency kcat/Km of 4.7 × 103 M-1 s-1
.

Human Thioredoxin Reductase Inhibitors
Journal of Medicinal Chemistry, 2005, Vol. 48, No. 22 7027
Figure 1. Characterization of the inhibition type of wild-type human placenta thioredoxin reductase by lead nitrofuran compound
8 in the steady-state. Inhibition of wild-type human placenta thioredoxin reductase by the lead derivative 8 in DTNB reduction
assay was characterized (Panel A), and the Ki value was determined (inset). Uncompetitive type of inhibition was confirmed by
Cornish-Bowden, Dixon, and Lineweaver-Burk plots (Panels B, C, D). DTNB concentrations were 10.9, 21.7, 43.4, 86.9, 173.8,
260.9, 347.6, and 434.5 µM. Inhibitor concentrations used were 0, 10, 20, and 40 µM.
Figure 2. Characterization of the inhibition type of truncated human placenta thioredoxin reductase by lead nitrofuran compound
8 in the steady-state. Inhibition of truncated human placenta thioredoxin reductase by the lead derivative 8 in the DTNB reduction
assay was characterized (Panel A), and the Ki value was determined (inset) by fitting the experimental data to the Michaelis-
Menten equation. Competitive type of inhibition was confirmed by Dixon and Lineweaver-Burk plots (Panels B and C). DTNB
concentrations were 0.17, 0.35, 0.43, 0.65, 0.87, 1.32, 1.74, 2.63, 3.48, and 5.21 mM. Inhibitor concentrations used were 0, 100,
220, and 250 µM.
Lead Compound 8 Follows Competitive Kinetics
in Inhibition of Truncated Human TrxR. Due to the
much higher Km DTNB value for the truncated enzyme
hTrxR∆-16, it was possible to study the inhibition of
hTrxR∆-16 by varying the DTNB concentration at
higher concentration level (0.6-6.0 mM). Kinetics were
performed using saturating NADPH concentration (200
µM) in the absence or in the presence of inhibitor 8 (0-
250 µM). Assuming simple competitive inhibition, the
data were fitted to eq 5 (Figure 2A):
Vmax[S]
v )
(5)
[I]
Km 1 +
+ [S]
(
)
Ki
The Ki value for 8 was determined as 136.5 ( 9.9 µM
with respect to DTNB (see inset in Figure 2A). The

7028 Journal of Medicinal Chemistry, 2005, Vol. 48, No. 22
Millet et al.
Chart 2. Structures of the Four Amino Acids:
Phenylglycine (21), 2-Naphthylalanine (22),
To obtain the platinum complexes 13a-c and 20, we
first prepared the protected NR-(ω-aminoethyl)amino
acid 5 as a building block. The synthesis of the poly-
amino acid side chain is outlined in Scheme 3 and is
based on previously published protocols with slight
modifications.36 We used the benzyl (N-benzyl or benzyl
ester) and the N-tert-butoxycarbonyl moieties as pro-
tecting groups. The first step was carried out to selec-
tively protect the ethylenediamine. Double acylation
was avoided by using a large excess of the nucleophilic
amine and purification yielded the monoprotected N-Boc
ethylenediamine 1 with 65% yield. The resulting amine
2 was used in a reductive amination reaction with
benzaldehyde and sodium borohydride. This temporary
protection of amine 2 by the N-benzyl protecting group
allowed in the next step the monosubstitution of amine
3 by 2-benzyl bromoacetate, hence preventing polyalkyl-
ation of amine 2. Catalytic hydrolysis using ammonium
formate and Pd (10% on charcoal) was applied to remove
the N-benzyl and the Z-protected ester. Zwitterion 4 was
finally protected by an N-tert-butoxycarbonyl moiety and
compound 5 which contains two equal N-protected
groups was used directly as a building unit for the
design of the platinum complexes.
Different Pt chelating ligands were investigated,
assuming that the locus where the linker chain is
attached can position the platinum atom in quite
different ways with respect to the potential nucleophilic
attack from the selenocysteine during the enzyme
catalytic cycle. To attach the platinum-chelating ethane-
diamine linker at different loci of the 5-nitro-2-furan-
carbohydrazide core, we selected four differently sub-
stituted (three naphthyl- and one phenyl-) amino acids
21-24 (Chart 2) to design the four Pt-ligands 12a-c
and 19, and their corresponding cis-diamminedichloro-
platinum(II) complexes 13a-c and 20. The phenyl-
amino acid 21 was chosen to evaluate the importance
of the naphthalene moiety for hTrxR recognition.
The 5-nitro-2-furancarbohydrazides 9a-c were pre-
pared by coupling the three racemic N-Boc amino acids
using the standard PyBOP methodology (Scheme 2).
After deprotection of the Boc group, the amines 10a-c
were next coupled with the polyamino acid 5 to yield
11a-c. Their Boc groups were removed by methanolic
HCl solution, and the platinum complexes 13a-c were
obtained by titrating the racemic polyamine hydro-
chloride 12a-c in the presence of equimolar amounts
of K2PtCl4 in DMF/H2O (1:3). Using this aqueous DMF
solution significantly increased the efficiency of the
platination reaction compared to the commonly used
aqueous methanol mixture. The last two steps of the
platinum synthesis were monitored by analytical HPLC.
For the synthesis of the cisPt complex 20 (Scheme 4),
the aromatic amine 17 was first prepared by coupling
the 5-nitro-2-furancarbohydrazide 7 and the N-Boc-
naphthoic acid 15 in the presence of PyBOP. PyBOP
was also used as coupling agent for the synthesis of
14a-c, the three analogues of 8. The deprotection of
the Boc group in 16 was carried out by using TFA in
CH2Cl2 and the coupling reaction of 17 with the N-
protected ethanediamine linker 5 was achieved in DMF
in the presence of PyBROP to produce 18. The last two
reactions to produce 20 were performed in a way similar
to that used for 13a-c. The conditions of complexation
1-Naphthylalanine (23), 6-Amino-2-naphthoic Acid (24)
competitive type of inhibition of hTrxR∆-16 by 8 was
derived from Dixon (Figure 2B) and Lineweaver-Burk
(Figure 2C) plots. At this stage, it was already obvious
that the kinetics were more complex (weakly sigmoidal
at low inhibitor concentration as DTNB concentration
increased) and a better curve fit was realized by using
the equation for inhibition of multisite-enzymes (data
not shown). The plot Km/Ki versus [I] could indeed be
interpreted using an exponential equation. This obser-
vation was further confirmed with complex 20 in the
following kinetics where marked sigmoidal curves were
obtained and could be fully analyzed.
Chemistry. From the selected lead structure 8, the
N′-(2-naphthylmethyl)-5-nitro-2-furancarbohydrazide, two
series of analogues 9a-c and 14a-c, as well as the
related ethanediamine ligands 12a-c and 19, and their
corresponding cis-diamminedichloroplatinum(II) com-
plexes 13a-c and 20 (Chart 1) were synthesized. The
design of the four cisPt complexes was oriented as
followed. Assuming that the C-terminal part of TrxR
moves to the external surface of the protein in the course
of the catalytic cycle, a rational approach is to design a
bitopic and irreversible TrxR inhibitor that would (i) be
recognized by the enzyme in a first sub-site, (ii) an-
chored by multiattachment in a second subsite revealed
after motion of the C-terminal part, and (iii) react with
the exposed selenocysteine. For this purpose, a coordi-
nating arm linking the 5-nitro-2-furancarbohydrazide
moiety to the cisPt unit is required. The selected
suitable arm consists of a spacer that ends by a
chelating ethylenediamine function. The linkage of the
arm to the 5-nitro-2-furancarbohydrazide was built as
followed. The methylene bridge of the 2-naphthylmethyl
substituent in the lead structure 8 was selected to
anchor the suitable arm. To give more flexibility to the
spacer the homo analogue was built by starting from
the racemic amino acid 2-naphthylalanine 22 (Chart 2)
instead of a 2-naphthylglycine. The regioisomer 1-naph-
thylalanine 23 was also selected as starting material
to vary the flexibility of the spacer. Then the phenyl-
glycine 21 was selected in order to evaluate the essential
requirement of the naphthyl moiety for optimal binding
to the target. Finally, the 6-amino-2-naphthoic acid 24
allowed the building of a more flexible and “linear”
spacer between the 5-nitro-2-furancarbohydrazide unit
and the chelating ethylenediamine function.
Our starting hTrxR inhibitor compound 8 was ob-
tained according to standard procedures for peptide
synthesis in solution (Scheme 2). Briefly, 5-nitro-2-furoic
acid was reacted with tert-butyl carbazate using PyBOP
as coupling agent. The Boc group was removed by
methanolic HCl solution resulting in 5-nitro-2-furan-
carbohydrazide hydrochloride 7, which was coupled with
2-naphthyl acetic acid to yield diacylhydrazine 8.

Human Thioredoxin Reductase Inhibitors
Journal of Medicinal Chemistry, 2005, Vol. 48, No. 22 7029
Scheme 2. Synthesis of 5-Nitro-2-furancarbohydrazide Derivativesa
a Reagents and conditions: (a) PyBOP, DIEA, tert-butyl carbazate, CH2Cl2, rt, 16 h; (b) MeOH, HCl, rt, 16 h; (c) PyBOP, DIEA, CH2Cl2,
N-Boc-amino acid, rt, 16 h; (d) PyBOP, DIEA, CH2Cl2, 2-naphthylacetic acid, rt, 16 h; (e) MeOH, HCl, rt, 16 h; (f) PyBOP, DIEA, CH2Cl2,
compound 5, rt, 24 h; (g) MeOH, HCl, rt, 16 h; (h) K2PtCl4, DMF/H2O 1:3, rt, 10 days.
Scheme 3. Synthesis of Polyaminocarboxylic Acid 5a
the high reactivity of DMSO toward platinum complexes
is responsible for slow chloro-ligand-DMSO exchange.27
From the proton NMR spectra, the signal correspond-
ing to the amine group of ethylenediamine appeared
strongly deshielded in the complexes (δ between 3.00
and 3.46 ppm for NH and δ between 2.50 and 3.40 ppm
for NH2). After complexation by K2PtCl4, we also
observed 50:50 diastereoisomeric mixtures, due to the
racemic parent polyamine and to δ and λ enantiomers
of the 1,2-diaminoethane ring, that can adopt an
envelope conformation, for the three Pt complexes 13a-
c, both from HPLC chromatograms and NMR spectra.
By contrast, the case did not apply for the Pt complex
20 starting from the nonracemic diamine 19. The 195Pt
NMR spectra were obtained for the four complexes and
showed 195Pt resonances at δ values of ca. -2346 ppm
(from PtCl62-). These values are in the same range as
those exhibited by previously reported ethylediamine-
linked Pt complexes in accordance with the mode of
coordination of the platinum.24,25 The geometry of
complexes is also supported by FTIR spectroscopy. The
IR spectral data showed bands for 13a-c at 477 and
for 20 at 474.4 cm-1 assigned to ν(Pt-N). The stoichi-
ometry of platinum in the complexes 13a-c and 20 was
measured by high resolution inductively coupled plasma-
a Reagents and conditions: (a) (Boc)2O, CH2Cl2, 0 °C, 3 h and
room temperature, 16 h; (b) (1) PhCHO, NEt3, MeOH, molecular
sieves 3 Å, rt, 0 °C, 4 h, (2) NaBH4, MeOH, rt, 16 h; (c) 2-benzyl
bromoacetate, DIEA, DMF, 0 °C, 1 h; (d) ammonium formate, Pd/
C, MeOH, reflux, 16 h; (e) (Boc)2O, NaOH, dioxane/H2O 4:1, rt,
16 h.
by K2PtCl4 were slightly modified by performing the
titration of diamine 19 at 60 °C to accelerate the
reaction. Once prepared, the cisPt complexes appeared
insoluble in water and MeOH, but dissolved rapidly in
DMF and DMSO. DMF was used as solvent for the
NMR and enzymic studies, as well as in all enzymatic
and biological studies, because, as previously reported,

7030 Journal of Medicinal Chemistry, 2005, Vol. 48, No. 22
Millet et al.
Scheme 4. Synthesis of 5-Nitro-2-furancarbohydrazide Derivatives 14a-c, 16-20a
a Reagents and conditions: (a) appropriate acid, PyBOP, DIEA, CH2Cl2, rt, 16 h; (b) CH2Cl2, TFA, rt, 16 h; (c) PyBROP, Compound 5,
DIEA, DMF, rt, 16 h; (d) MeOH, HCl, rt, 72 h; (e) K2PtCl4, DMF/H2O 1:3, 60 °C, 7 days.
Figure 3. Characterization of the inhibition of wild-type human placenta thioredoxin reductase by CisPt Complex 20 in the
steady-state. Inhibition of wild-type human placenta thioredoxin reductase by CisPt Complex 20 was characterized in the DTNB
reduction assay (Panels A and B) and the Trx reduction assay (Panel C). Competitive type of inhibition of wt-hTrxR in DTNB
reduction assay was evidenced from Dixon and Cornish-Bowden plots (Panels A and B). DTNB concentrations were 100, 200,
300, 400 and 500 µM. Inhibitor concentrations used were 0, 100, 200, 400, 600 and 800 nM. Panel A. The intersection point of all
lines in the upper left quadrant from Dixon plot, as typical for competitive inhibition, indicates a Ki value of 275 nM by extrapolation
to the abscissa. Panel B. The reversible competitive component of inhibition becomes evident by parallel linear regression curves
from the Cornish-Bowden plot. Panel C. The type of inhibition of wt-hTrxR by inhibitor 20 measured in Trx assay is complex,
as suggested from the Lineweaver-Burk plot showing the intersection point of the lines in the upper right quadrant (upper to
lower lines are for 500, 250, 125, 20 nM inhibitor 20). Trx concentrations were 8.0, 10.0, 14.9, 20.0, 30.3, 50.0, and 100.0 µM.
mass spectrometry (ICP-MS).37 Four samples corre-
sponding to Pt-complexes 13a-c and 20 were prepared
at fixed concentration and analyzed. Platinum concen-
tration was determined by ICP-MS for 13a-c and 20
solutions and was confirmed by comparison with cali-
brated platinum reference solutions.
Steady-State Kinetic Analysis of the Complexes.
The reversible inhibitory properties of the compounds
were analyzed in direct assays under steady-state
conditions, using either thioredoxin or the alternative
disulfide DTNB as substrate. All assays were started
by the addition of NADPH in order to minimize effects
caused by prereduction of the enzyme. The amount of
enzyme was kept low in order to render the nitrofuran
reduction negligible. As indicated by Dixon (1/v versus
[I], Figure 3A) and Cornish-Bowden ([S]/v versus [I],
Figure 3B) plots, all compounds exhibited competitive
type inhibition with respect to DTNB reduction. The
respective Ki values for competitive inhibition were
derived from the plots and calculated using eq 5.
The Ki values were 110 nM for compound 13a, 84 nM
for 13b, 120 nM for 13c, and 275 nM for compound 20.

Human Thioredoxin Reductase Inhibitors
Journal of Medicinal Chemistry, 2005, Vol. 48, No. 22 7031
Table 1. Time-Dependent Inhibition of Wild Type hTrxR and
hTrxR Sec498Cys Mutant by cis-Diamminedichloroplatinum
Complexes 13a-c and 20 (IC50 values)a
nM to 210 nM in Trx reduction assay). These IC50 values
found in the submicromolar range in wt-hTrxR assays
revealed complex 20 as the most effective for inactiva-
tion. The flexibility of the spacer introduced in complex
20 seems to play an important role in the rate of the
reaction between the cisPt unit and the selenocysteine
of this enzyme. The inhibition of the mutant (Sec498Cys)
hTrxR was by at least 2 orders of magnitude less
effective. This result strongly supports the hypothesis
that platinum complexes interact specifically with the
selenocysteine at the C-terminal Cys-Sec redox pair.
Interestingly, the inhibition of the mutant was more
effective in the presence of its natural substrate Trx
than in the presence of DTNB, an observation that will
be discussed below. The IC50 values of the four inves-
tigated compounds were found to be 5 to 16-fold higher
in the hGR than in the hTrxR assays, indicating a
relative TrxR specificity for our compounds.
Reversibility studies of the inhibited hTrxR were
tested for all four compounds. After a 10 min-incubation
period of hTrxR with 2 µM inhibitor in the presence of
100-200 µM NADPH more than 90% of all enzyme
molecules were inhibited. Addition of 10 mM dithio-
threitol (DTT) for 2 h and exhaustive dialysis did not
restore TrxR activity (less than 5%). This suggests that
the inhibition is irreversible and that the formation of
a stable covalent bond between the enzyme and the
inhibitor is involved in the inactivation process.
IC50 (µM)
8
13a
0.16
0.10
13b
0.14
0.03
24.0
230
195
0.95
13c
0.21
0.08
21.0
235
170
1.0
20
wt-hTrxR Trx assay
wt-hTrxR DTNB assay
Sec498Cys Trx assay
Sec498Cys DTNB assay
hTrxR∆-16 DTNB assay 215
hGR 4.1
nd
18
0.12
0.11
38.0
130
145
1.0
nd ∼250b
nd
265
520
2.6
a DTNB assay: 5 nM wild-type hTrxR and 100 nM mutant
hTrxR, respectively, were preincubated for 10 min at 25 °C in
phosphate buffer in the presence of 200 µM NADPH with or
without inhibitor; the reaction was started by adding 3 mM DTNB.
Trx assay: 30 nM wild-type hTrxR and 1.3 µM mutant hTrxR,
respectively, were preincubated for 10 min at 25 °C in assay buffer
in the presence of 100 µM NADPH with or without inhibitor; the
reaction was started by adding 20 µM hTrxC72S. DMF was used
in the control assays. All values are mean values of at least three
independent determinations which differed by less than 15%.
b Compound 13a showed a strong absorbance at 340 nm which
limited exact IC50 determination.
These values revealed that the 2-naphthylmethyl unit
was the optimal substituent recognized by hTrxR in
agreement with the substitution pattern of the lead
structure 8. The results obtained with thioredoxin
as substrate were more complex (Figure 3C). Dixon,
Cornish-Bowden, and Lineweaver-Burk curves of
inhibited wild-type and mutant (Sec498Cys) hTrxR
showed an intersection point in the upper right quad-
rant, but a clear classification of inhibition type was not
possible.
In the presence of compound 13c and in the absence
of any disulfide substrate, the Sec498Cys hTrxR mutant
showed a strong NADPH oxidation reaction (20 mU/
mg), but not in the control assay with DMF instead of
inhibitor, which is most likely due to an induced oxidase
activity of the alkylated enzyme. This effect had been
previously observed with 1-chloro-2,4-dinitrobenzene-
alkylated human thioredoxin reductase.38
Time-Dependency and Irreversibility of Inhibi-
tion of hTrxR. As shown above, it was confirmed that
NADPH-reduction is a prerequisite for irreversible TrxR
inhibition by cisplatinum complexes. On incubation with
the complexes, human TrxR is inhibited in a time-
dependent manner and the inhibition follows pseudo-
first-order kinetics. The process for enzyme inactivation
may be represented by the following eq 6:
Relative IC50 Values from Time-Dependent In-
activation. A general feature of disulfide reductases
is their particular susceptibility toward many nucleo-
philic compounds in the NADPH-reduced state. This
holds true for TrxR. Less than 10% inhibition was
observed when the wild-type enzyme was preincubated
with the compounds in the absence of NADPH. For
comparable results we incubated the enzyme for 10 min
with NADPH and inhibitor and started the assay
by adding the respective disulfide substrate. Apart
from wt-hTrxR we also analyzed the hTrxR mutant
(Sec498Cys) and the truncated hTrxR∆-16 enzyme to
verify the preferential effects on selenocysteine at the
C-terminal redox center. Inhibition of the closely related
enzyme human glutathione reductase (hGR, EC 1.8.1.7.)
was also determined to analyze enzyme specificity
toward inhibitors. The IC50 values for the four com-
pounds 13a, 13b, 13c, and 20 were calculated from
dose-response curves and are summarized in Table 1.
The IC50 values of the lead 5-nitro-2-furancarbohy-
drazide derivative 8 were added for comparison. Under
the same conditions, no time-dependent inhibition of the
enzyme was observed in the presence of the reversible
hTrxR inhibitor 8. Wild-type hTrxR was inhibited by
all four complexes in the nanomolar range (IC50 values
from 30 nM to 110 nM in DTNB reduction assay; 120
KI,SN represents the dissociation constant of the
inhibitor at the N-terminal redox active site, and KI,SC
the corresponding constant at the C-terminal redox
active site, respectively. ki is the first-order rate constant
for irreversible inactivation. It is however unlikely that
both parts of the inhibitor are firmly bound to the
enzyme simultaneously. The main function of the nitro-
furan moiety is to guide the cisPt-ligand to its site of
action and to trap it there until the reaction with the
selenolate has taken place. In good accordance with the
experimental data, this setting allows to apply the
derivation by Kitz and Wilson,39 for irreversible inac-
tivation. Under the assumption that the concentration
of inhibitor (I) is much higher than the stoichiometric
amount of enzyme (E0), it is possible to calculate the
remaining amount of active enzyme (E) at time t via eq
7, according to the derivation by Kuo and Jordan:40
E
ln
(E )
ki
1 + KI,SN[I]-1 + KI,SNKI,SC[I]-2
0
) - kobs(t) )
(7)
t

7032 Journal of Medicinal Chemistry, 2005, Vol. 48, No. 22
Millet et al.
Figure 4. Time-dependent inactivation of wild-type human thioredoxin reductase by CisPt Complexes 20 and 13a-c. The time-
dependency of inactivation of wild-type hTrxR is shown in Panels A-C for CisPt complex 20, in Panel D for CisPt complexes 13a,
in Panel E for 13b, and in Panel 4F for 13c. Panels A-C: Different aliquots of pretreated wt-hTrxR enzyme were tested for
residual activity at different time periods: 0.5, 5, 10, 15, 30, and 60 min. Data for inhibitor 20 are presented in the primary plot
from Panel A. The dissociation constants for inhibitor 20 from the N-terminal redox active site and from the C-terminal redox
active site (KI,SN, KI,SC), respectively, and the first-order rate constants for irreversible inactivation ki were determined. The kobs
data versus [I] were fitted to the eq 7 in the text, resulting in the plot given in Panel B. The t1/2 values versus 1/[inhibitor]-2 were
fitted to the eq 8 in the text, resulting in the plot given in Panel C. Inhibitor 20 concentrations used were 0, 8.2, 12.3, 16.4, and
24.6 µM. Panels D: Inhibitor 13a concentrations used were 0, 4.6, 9.1, 15.2, and 23.0 µM. Panels E: Inhibitor 13b concentrations
used were 0, 7.3, 13.7, and 18.3 µM. Panels F: Inhibitor 13c concentrations used were 0, 9.1, 15.2, and 23.0 µM.
The observed t1/2 for each concentration of inhibitor
is given by eq 8:
the mutant (Sec498Cys) hTrxR, it was assumed that
the very reactive C-terminal selenolate of wt-hTrxR
attack the platin atom of the cis-diamminedichloro-
platinum(II) moiety of 13a-c and 20. In the case of
Drosophila melanogaster enzyme (DmTrxR) the serine
residues flanking the C-terminal Cys residues are
responsible for activating the Cys to match the catalytic
efficiency of a Sec-Cys pair as found in hTrxR.31 This is
due to the low pKa values of the selenol in hTrxR41 and
the decreased pKa of Cys in wild-type DmTrxR. Conse-
quently, a variation in the pKa value of the cysteines in
wild-type and mutant DmTrxRs should influence the
concentration of the nucleophilic thiolates and thereby
the effectiveness of the inhibitor complexes. To validate
this assumption, the reactivity of the most potent
inhibitor 20, with respect to alkylation, was compared
in assays with wild-type recombinant D. melanogaster
enzyme and C-terminal mutants.31 In these mutants,
either the critical Cys was replaced by a Sec (Cys490Sec),
or the surrounding serine residues in the wild-type
enzyme (SCCS) were replaced by glycines (GCCG), a
residue found in the human C-terminal active center
(GCUG). As shown previously,31 the flanking polar
serine residues at the C-terminal Cys-Cys-redox active
site enhance the cysteine reactivity significantly com-
pared to the nonpolar glycine mutants. The prereduced
t1/2 ) t1∞/2(1 + KI,SN‚[I]-1 + KI,SNKI,SC‚[I]-2
where t∞1/2 ) (ln 2)/ki.
)
(8)
The primary plot (% residual activity versus pre-
incubation time) defining time-dependent inhibition of
wild-type hTrxR is shown in Figure 4A. At early stages,
the inactivation occurred by a pseudo-first-order process
following eq 7 (Figure 4B). In the later stages, the
inactivation kinetics became distinctly sigmoidal (as
shown e.g. for inhibitor 13b in Figure 4E). In accordance
with eq 8, a secondary plot of t1/2 (obtained from the
early stages of inactivation) against [I]-2 resulted in a
linear relationship at low concentration of inhibitor 20
(Figure 4C). For inhibitor 20, a first-order rate constant
ki of 0.96 min-1 as well as KI,SC and KI,SN dissociation
constant values of 30 nM and of 42 mM were determined
from plots in Figure 4A,B. The inactivation process of
hTrxR by the other cisPt complexes 13a-c obeyed the
same eq 7 (Figures 4D-F).
Effects on Drosophila melanogaster Thioredox-
in Reductase. The results shown above indicate that
our cisPt complexes act as active site-directed irrevers-
ible inhibitors as expected. From data gathered with

Human Thioredoxin Reductase Inhibitors
Journal of Medicinal Chemistry, 2005, Vol. 48, No. 22 7033
Figure 5. Time-dependent inactivation of wild-type D. melanogaster thioredoxin reductase and of SCUS mutant by CisPt complex
20. The time-dependency of inactivation of wild-type DmTrxR and of SCUS mutant is shown in Panels A and B for CisPt complex
20, respectively. Different aliquots of pretreated wt-hTrxR enzyme were tested for residual activity at different time periods: 0,
5, 10, 15, 30, and 60 min. Panel A: Inhibitor concentrations used were 0, 10, 25, and 50 µM. Panel B: Inhibitor concentrations
used were 0, 5, 10, and 25 µM.
enzymes were inactivated in a time- and concentration-
dependent manner. The wild-type (SCCS) enzyme was
inactivated the fastest, the GCCG-mutant the slowest,
as predicted. Detailed kinetics are given below.
Inhibition by the four complexes 13a-c and 20
followed pseudo-first-order reaction kinetics. A semi-
logarithmic plot of the fraction of noninhibited enzyme
activity ln(vi/v0) versus preincubation time yielded linear
curves with increasing slopes, equivalent to the appar-
ent rate constant of irreversible inhibition (kobs). The
kobs values for 20 and wild-type enzyme ranged between
6.9 × 10-3 min-1 and 24.7 × 10-3 min-1 within the log-
linear range of the inhibition curve, and a maximal ki
value of 1.7 × 10-3 s-1 (Figure 5A). A double reciprocal
replot of kobs versus [I] was fitted to the linear relation-
ship (eq 9):
reactivity state of the terminal cysteines in the wt
enzyme, as suggested earlier.31 The bimolecular rate
constant ki/KI for the Dm GCCG mutant of 8.5 M-1 s-1
could only be calculated from time-dependent inactiva-
tion experiment performed with inhibitor in DMSO
solution instead of DMF. The influence of DMSO was
directly observed on the ki value which was evaluated
as 5-fold higher than in the presence of DMF under
same conditions, i.e., 0.05 min-1 versus 0.013 min-1. It
is well-known that exchange of one chloride ligand by
a DMSO molecule induces a positive charge at the metal
atom, thus increasing the reactivity toward thiol attack
and consequently the rate of the inactivation process.
Such DMSO effects were previously reported to enhance
both the cytotoxicity and inhibition of DNA synthesis.27
Cytotoxicity. A panel of human and mouse tumor
lines was tested for cytotoxic effects of compounds 13a,
13b, 13c, and 20 in dose response experiments (Table
2). For comparison, cisPt was tested on the same cell
lines. Tumor cell lines were selected upon their sensitiv-
ity to cisPt (Calu-6, highly sensitive; SK-MEL-25,
moderately sensitive; and MCF-7, almost resistant). A
35-fold higher concentration of cisPt was needed to kill
MCF-7 cells (IC50 ) 63.7 µM) as compared to Calu-6
(IC50 ) 1.8 µM). In contrast, the concentration of 13a,
13b, 13c and 20 required to kill MCF-7 cells had only
to be doubled (IC50 ranging from 46.3 to 55.9 µM) in
comparison to Calu-6 (IC50 ranging from 25.0 to 29.1
µM). Furthermore, the absolute concentration of each
of the four platinum complexes required to kill MCF-7
was lower than for cisPt itself, indicating a higher
potency to inhibit the hTrxR/Trx-dependent growth of
the most resistant cell lines.
ki[I]
kobs
)
(9)
KI + [I]
The secondary plots expressing kobs as a function of
inhibitor concentration, as illustrated for complex 20
and wild-type enzyme (inset in Figure 5A), showed
hyperbolic curves allowing determination of ki as 0.102
min-1 and second-order rate constants ki/KI as 10.8 M-1
s-1. The most potent inhibitor of the human enzyme as
judged from IC50 values, complex 20, was selected as a
model in our inactivation studies with mutant enzymes.
The DmTrxR CysfSec mutant enzyme (SCUS) is alkyl-
ated by cisPt complex 20 with a higher efficiency than
the wild-type enzyme (SCCS), in agreement with a
higher bimolecular rate constant ki/KI (17.9 versus 10.8
M-1 s-1) (Figure 5B). This result confirmed that the
C-terminal active site of DmTrxR is the locus of alkyl-
ation by cis-Pt derivatives. The DmTrxR mutant in
which the surrounding serine residues were replaced
by glycine exhibited ki value which is 7.8-fold lower than
that of the wild-type serine-containing enzyme, i.e.,
0.102 min-1 (SCCS) and 0.013 min-1 (GCCG), respec-
tively. The resulting t1/2 values were determined as 6.8
and 53.3 min, respectively. This finding supports the
influence of the serine residues 488 and 491 on the
Table 2. Antiproliferative Effects of the Platinum Complexes
13a-c and 20 on Various Cancer Cell Lines
IC50 (µM)a
cell line
MCF-7
SK-MEL-25 15.9 (3.2) 15.0 (3.0) 14.0 (2.8) 15.1 (3.0)
Calu-6 25.0 (13.9) 29.1 (16.1) 26.5 (14.7) 27.6 (15.3) 1.8 (1)
13a
55.9 (0.9) 52.0 (0.8) 46.3 (0.7) 53.1 (0.8) 63.7 (1)
5.0 (1)
13b
13c
20
cisPt
a The ratio IC50 13a-c-20/IC50 cisPt for each derivative is given in
parentheses.

7034 Journal of Medicinal Chemistry, 2005, Vol. 48, No. 22
Millet et al.
DNA Interaction Studies. The four tested sub-
stances reacted strongly with DNA at a concentration
of 1 mM (Figure 6). No DNA was visible in the gel any
longer. This is probably due to extended cross linking
which prevents the DNA from entering the gel. At a
drug concentration of 0.1 mM, a partial cross linking
was obtained. The amount of DNA and the mobility of
the individual bands were found to be reduced. At 0.01
mM of the tested compound, DNA samples were indis-
tinguishable from control DNA. A similar pattern was
obtained with cisPt. Partial interaction with DNA was
observed at 0.1 mM and no reaction at 0.01 and 0.001
mM.
chain in order to increase the binding to the target
according to a biligand approach. Irreversible inactiva-
tion was desired to optimize the cytotoxicity in cancer
cells producing high levels of GSH and Trx.42-44
Screening and Chemistry. The screening of a
12 000-members library of chemicals resulted in only
one inhibitor of human TrxR, the lead 5-nitro-2-furan-
carbohydrazide 8, with inhibitory potencies higher than
those of known inhibitors selected as references. Previ-
ous studies have demonstrated that nitrofurans are
potent inhibitors of the disulfide reductase family,
including glutathione reductase29 and trypanothione
reductase34 which do not possess a second redox center
in the C-terminal part of the protein. Nitroaromatic
compounds including nitrofurans, nitrobenzenes and
nitroimidazoles are known to exert their cytotoxic effects
through their nitroreductase activities. The targets
belong to the flavin-dependent oxidoreductase family
that catalyzes one- or two-electron reducing reactions.
Nitrofurans act as redox-cyclers; they were also de-
scribed as subversive substrates or turncoat inhibitors
of these flavoenzymes. In the presence of yeast GR,
numerous nitrofurans showed an uncompetitive inhibi-
tion type with respect to both NADPH and GSSG.29
Among the TrxR family, only the small TrxR from
Arabidopsis thaliana was reported to reduce nitrofurans
so far.30 The site where reduction of nitrofurans takes
place in the flavoprotein structure is not yet identified.
New compounds were designed from the lead nitrofuran
derivative 8 as bitopic inhibitors, a reversible nitrofuran
attached to a cisPt moiety to allow the irreversible
binding of the C-terminal Sec in hTrxR, both to optimize
TrxR inhibition and to overcome cisPt resistance. It is
well documented that the attachment of small unteth-
ered molecules only binding in the micromolar to
millimolar range to proximal subsites of target proteins
leads to bitopic molecules with submicromolar affinity.45
Figure 6. Effects of Complexes 13a-c, 20 and cisPt on
electrophoretic mobility of DNA. A mix of DNA fragments was
incubated with platinum compounds at a concentration of 1,
0.1, 0.01, or 0.001 mM, as indicated. DNA samples were
separated by electrophoresis on an agarose gel, followed by
staining with ethidium bromide. The gel was photographed
under UV light. Binding of platinum compounds to DNA is
visualized by shift of electrophoretic mobility in agarose.
Discussion
While nitroaromatic derivatives are not the preferred
candidates for drug development, the very low mu-
tagenic properties reported for 5-nitro-2-furohydrazide
derivatives among three structure-related series of
5-nitrofuran derivatives prompted us to investigate the
chemistry of the lead 5-nitro-2-furancarbohydrazide 8
to develop hTrxR inhibitors as potential anticancer
drugs.46
Interpretation of Steady-State Kinetics with
Nitrofuran Carbohydrazide 8 and the Four CisPt
Complexes. The uncompetitive kinetics on wild-type
hTrxR when DTNB is varied showed that prior DTNB
binding to the reduced enzyme promotes association of
the lead nitrofuran 8 to another site. The dissociation
constant Ki for formation of the reversible complex was
16 µM, underlining the significant affinity of the lead
inhibitor for a yet unidentified binding site of hTrxR.
The competitive kinetics observed for DTNB reduction
catalyzed by hTrxR∆-16 implied that DTNB binding to
the reduced truncated enzyme decreases affinity for the
lead compound 8 at the N-terminal binding site. Thus,
these studies are consistent in suggesting that the
binding of the lead compound 8 is promoted by prior
association of the substrate to the C-terminal part of
reduced wt-hTrxR. Furthermore, DTNB competes with
the lead compound 8, both are small electron acceptors,
for the N-terminal binding site of the truncated enzyme.
The common cytostatic agent cisPt is clinically used
for the treatment of many malignancies. However, many
initially sensitive tumors develop resistances against
this drug after the first cycles of treatment, requiring
an increase in cisPt dosage to maintain antitumoral
activity. Increase in dosage is however limited by the
severe side effects of the drug, particularly myelo-,
neuro- and nephrotoxicity. Therefore, many efforts in
designing novel Pt-based drugs have been made to
overcome resistance, to increase the solubility in water
and to cure a broader range of cancers. Enzyme inhibi-
tion may contribute to undesirable and toxic side effects
as consequences of chemotherapy or may reinforce the
antitumoral effects by interaction with additive or
synergistic targets for cancer therapy. Platinum com-
plexes are often very reactive toward the cysteine
residue of GSH, which detoxifies these compounds by a
rapid binding mechanism.42 Indeed, cisPt itself il-
lustrates the positive combined effects of DNA binding
and enzyme inhibition, in particular in hTrxR inhibi-
tion.11 Formation of the GS-Pt adduct, which is a major
route for elimination of cisPt, is also a toxic event for
the cancer cells after irreversible inactivation of hTrxR,
more potent than with the parent cisPt itself, by
disturbing the redox equilibrium.33 The aim in our
studies was the design of cisPt derivatives by attaching
a motif known to recognize hTrxR to the ethanediamine

Human Thioredoxin Reductase Inhibitors
Journal of Medicinal Chemistry, 2005, Vol. 48, No. 22 7035
The last result was also confirmed by the time-depend-
ent inactivation experiments. After attachment of the
cisPt moiety to the lead compound 8, the resulting four
cisPt complexes 13a-c and 20 were found to inactivate
the authentic wt-hTrxR with Ki in the nanomolar range
(84 nM to 275 nM). This is the successful result of the
biligand strategy to design potent inhibitors of large
TrxRs.
Time-Dependent Inactivation of Large TrxRs by
the Four CisPt Complexes. IC50 values were mea-
sured from 10 min preincubation assays with mutant
TrxRs in addition to the wild-type TrxR. In these assays,
the submicromolar IC50 values determined in wt-hTrxR
assays and the IC50 values determined in the micro-
molar range confirmed the selenocysteine as the main
target of the cisPt complexes. The lowest IC50 values
(21-38 µM) obtained in Trx reduction assays versus the
high IC50 values (130-265 µM) obtained in DTNB
reduction assays with the Sec498Cys mutant are in
agreement with competition at the N-terminal active
site between DTNB, but not Trx, and the four cisPt
complexes.
In the present study, we observed a potent and
irreversible inhibition of authentic human TrxR by the
four cisPt complexes derived from 5-nitro-2-furancar-
bohydrazide 8. Inactivation increased with time of
preincubation in enzyme-NADPH-inhibitor solu-
tions and was not reversible after dialysis. Evaluation
yielded a rate constant ki of 0.96 min-1, describing the
conversion of the preformed reversible hTrxR‚inhibitor
20 complex to an irreversibly inhibited enzyme (dis-
sociation constant value KI,SC of 30 nM for the cis-
diamminedichloroplatinum (II) moiety of complex 20 at
the C-terminal redox center). With regard to the C-
terminal binding site and the mechanism of inhibition,
the selenocysteine is predicted to be the target of
alkylation by the four cisPt complexes since its replace-
ment by a cysteine or its absence in the truncated
enzyme led to a drastic decrease in inactivation. Finally,
the determined inhibition parameters evidenced the
improved potency of the inactivation process by the
bitopic inhibitor 20 and its analogues compared to cisPt
itself.
Cooperative Effect Induced by Nitrofuran Car-
bohydrazide 8 and the Four cisPt Complexes. A
clear induced cooperative effect in the course of hTrxR
inactivation by inhibitors 13a-c and 20 could be
observed for the first time. This observation is not really
surprising since a 3D-structure model of the mam-
malian TrxR-Trx complex was suggested to explain the
electron transfer from NADPH to the disulfide of the
substrate by involving the motion of the C-terminal tail
without large conformational changes. So far the only
three-dimensional structure of a large TrxR is that of
rat TrxR SeCys498Cys mutant‚NADP+ complex deter-
mined to 3.0 Å resolution by X-ray crystallography.17
The overall structure is similar to that of glutathione
reductase, including conserved amino acid residues
binding the cofactors FAD and NADPH. The structure
of the reduced enzyme shows a surface-exposed confor-
mation of the C terminal tail. By contrast, the oxidized
enzyme is proposed to have its last terminal residues
buried at the active site since they are not accessible to
the protease13,14 or to chemical modification.15 The
absence of drastic steric clashes in the conformational
changes of the very C-terminal tail in the step EH2 f
EH4 might explain why cooperative effects were not
observed in kinetics with both NADPH and Trx.
Specificity of hTrxR inhibition. To estimate the
specificity of the inhibition, the cisplatinum complexes
were also tested on the structurally and mechanistically
closely related, but selenium-free hGR. While all resi-
dues directly interacting with the substrate glutathione
disulfide (GSSG) in GR are conserved in hTrxR, the
failure of GSSG to act as substrate for TrxR strongly
suggests that these residues are not exposed to allow
its binding. In our studies, human GR was found far
less susceptible (by a factor of 5 to 16) to inhibition than
hTrxR. These data again indicate that the cisplatinum
complexes bind covalently and specifically to the ex-
posed selenocysteine residue. The inhibition of hGR
might be related to nitrofuran recognition at the very
conserved N-terminal active site.
Structure-Activity Relationships and Cytotox-
icities. The structure-activity relationships of this first
generation of cisPt complexes showed that little chemi-
cal modifications were made to observe a broad spec-
trum of inhibitory potencies and cytotoxicities. An
apparently conflicting finding was that our four cisPt
complexes induced a potent irreversible inhibition of the
isolated hTrxR whereas their cytotoxic effects on three
different tumor cell lines were only moderate, even
though they are more effective than cisPt itself on the
most resistant cell line MCF-7. Several explanations
may account for these observations. First, the primary
target of cisPt is DNA. After covalent linkage of cisPt
by the nitrofuran via a spacer, the capability of the
newly synthesized cisPt derivatives to bind DNA was
shown not to be altered. Interaction with DNA was
observed at a similar effective concentration as with
cisPt. In contrast, the cytotoxic effect of cisPt was
significantly more pronounced in sensitive cell lines.
Therefore, the differences in antitumoral activity cannot
be attributed to reactivity with DNA. In intact cells
DNA is packed with a huge amount of proteins. There-
fore, the accessibility for platinum compounds may be
hindered. The sterically less demanding cisPt may reach
DNA in the nucleus much easier. In addition, transport
of platinum compounds through the plasma membrane
is a sensitive process which might be affected by the
tested compounds. Alternatively, either the four cisPt
complexes possess amide bonds that could be cleaved
in tumoral cells, or the structure of the four complexes
is not appropriate for high cell permeability, or adducts
with glutathione are rapidly extruded out off the cells.
It might furthermore be possible that the alkylated
hTrxR is reactivated by an unknown intracellular
mechanism. Nevertheless, the higher cytotoxicity of the
four cisPt complexes versus cisPt against the breast
cancer MCF-7 strain supports that TrxR indeed con-
tributes to cell growth, drug sensitivity, and DNA repair
in tumor cells, particularly in resistant cell lines.
Conclusion
We have demonstrated that the four cisPt complexes
based on the lead nitrofuran carbohydrazide 8 are
potent and specific inhibitors of large TrxRs. The four
cisPt complexes revealed to be valuable tools to study

7036 Journal of Medicinal Chemistry, 2005, Vol. 48, No. 22
Millet et al.
identical as reported earlier, i.e. 10-27 U/mg for wild-type
hTrxR in the DTNB-assay,19,49 and of 170 U/mg for hGR in
the GSSG-assay,47 respectively. The newly described hTrxR∆-
16 mutant’s activity was below the detection limit in the
thioredoxin reduction assay and reached 1.4% (0.41 U/mg) of
the authentic enzyme activity in the DTNB-reduction assay,
a value similar to data observed for the rat TrxRSec498Cys41
and the respective human enzyme of this work (0.35 U/mg).
All reagents used were of the highest commercially available
purity and obtained from Alfa, Boehringer, Serva, and Sigma,
Germany.
Inhibitor Screening with Human Thioredoxin Reduc-
tase. For preliminary inhibitor studies and high-throughput
inhibitor screening in microtiter plates, DTNBA disulfide28 was
used as artificial substrate of hTrxR according to the reported
procedures for hTrxR inhibition studies.50 The formation of
5-thio-2-nitrobenzamide liberated from DTNBA reduction was
followed by measuring the absorbance at 416 nm and plotted
as a function of time in both the absence and presence of 50
µM inhibitor. The DTNBA disulfide concentration of the 5 mM
stock solution in DMSO was adjusted spectrophotometrically
by using ꢀ327 nm of 15,600 ( 80 mM-1 cm-1. All reactions with
hTrxR were performed in 100 mM sodium phosphate, 2 mM
EDTA, pH 7.0. The standard assay mixture for HTS contained
500 µM NADPH, 200 µM DTNBA, and 25 µM inhibitor and
the reaction was started with native enzyme (8.0 × 10-4 units
in 100 µL total volume). The commercially available TrxR
inhibitor 2,6-dichloroindophenol was used as reference inhibi-
tor. Confirmation of the positive hits and evaluation of further
intermediates of synthetic procedures was performed spectro-
photometrically by measuring IC50 values in duplicate in the
presence of 50 µM DTNBA, increasing inhibitor concentrations
(5-50 µM) and 1% DMSO as final concentration. The reaction
was started by adding native enzyme (7.2 × 10-3 units in 500
µL total volume).
TrxR-Catalyzed Nitrofuran Reductase Activity. The
ability of TrxR to reduce the nitrofuran lead compound was
assayed at 25 °C by monitoring the oxidation of NADPH at
340 nm (ꢀ340 ) 6.22 mM-1 cm-1). The assays, in a total volume
of 1 mL, contained TrxR-buffer (100 mM potassium phosphate,
2 mM EDTA, pH 7.4) and 0.386 units TrxR. The nitrofuran
was dissolved in DMSO, and the NADPH-oxidase activity was
measured at 10 different concentrations (10-400 µM) in the
presence of 1% DMSO. For the determination of Km and Vmax
values, the steady-state rates were graphically fitted by using
nonlinear regression analysis software (Kaleidagraph) to the
Michaelis-Menten equation, and the turnover number kcat and
the catalytic efficiency kcat/Km were calculated. The initial rate
of intrinsic NADPH oxidase activity of P. falciparum TrxR was
not subtracted from the rates measured in the presence of the
nitrofuran, since it proved negligible in comparison to the
TrxR-catalyzed nitrofuran reductase activity.
the mechanism of these multisite enzymes. These
results encouraged us to explore new linkages and to
vary spacer lengths between the cisPt moiety and the
TrxR inhibitors for designing a second generation of
potential anticancer drugs. TrxR may be of importance
for the development of new platinum complexes which
circumvent resistance to cisplatinum. The additional low
but significant inhibition of human GR might be a
positive property to exploit for the design of potential
anticancer multitarget drugs that aim at overcoming
drug resistance. Work is now in progress to optimize
hTrxR binding while maintaining DNA alkylation prop-
erties by preparing new bidrugs based on other lead
structures that act act as irreversible hTrxR inhibitors
with more appropriate pharmacokinetic properties.
Experimental Section
Abbreviations. AA, amino acid, DMSO, dimethyl sulfoxide;
DTNB, 5,5′-dithiobis(2-nitrobenzoic acid); DTNBA, 5,5′-dithio-
bis{N-[3-(dimethylamino)propyl]-2-nitrobenzamide}; GSH, re-
duced glutathione; GSSG, glutathione disulfide; GR, gluta-
thione reductase (EC 1.8.1.7.); hGR, human glutathione re-
ductase; PHPLC, purity determined by HPLC; PyBrop, bromo-
tris-pyrrolidino-phosphonium hexafluorophosphate; TLC, thick-
layer chromatography; Trx, thioredoxin; Trx(S)2, oxidized
thioredoxin; Trx(SH)2, reduced thioredoxin; TrxR, thioredoxin
reductase (EC 1.8.1.9.); hTrxR, human thioredoxin reductase;
TOFMS, time-of-flight mass spectrometry; ICPMS, induced
coupled plasma mass spectrometry.
Materials. Human cytosolic thioredoxin reductase (hTrxR)
was purified from placenta as described by Gromer et al.19
Recombinant mutant thioredoxin reductase (hTrxRSec498Cys)
and hTrxR∆-16 were used for comparative and mechanistic
studies and prepared by standard techniques (Urig et al.,
unpublished data). In brief, the hTrxR-1 gene was amplified
by PCR from a human cDNA library. Site-directed mutagen-
esis (QuickChange, Stratagene) was used for the replace-
ment of selenocysteine by cysteine (hTrxRSec498Cys), and
C-terminal truncation (hTrxR∆-16) was done using standard
PCR techniques. The products were cloned into the pQE30
expression vector (Qiagen) and propagated in E. coli M15 cells.
Protein was expressed and subsequently purified over a Ni-
NTA column (Novagene) via its N-terminal hexahistidyl-tag
(Urig et al., unpublished data). Recombinant D. melanogaster
thioredoxin reductase and two mutants, referred to their
C-terminal tetrapeptide sequence (-SCCS ) wild-type, -SCUS,
and GCCG) were prepared as described earlier.31 Recombinant
human glutathione reductase (hGR) was produced and isolated
according to Nordhoff et al.47 Recombinant E. coli thioredoxin
(EcTrx) was kindly provided by Prof. Charles Williams, Ann
Arbor, MI. Recombinant human mutant thioredoxin (hTrxC72S)
and D. melanogaster thioredoxin-2 (DmTrx) were produced and
purified as previously described21,48 and allowed to study the
inhibitory effects with the respective physiological substrate.
All enzymes used were pure as judged by silver stained SDS-
PAGE. Flavoenzyme concentrations were determined spectro-
Inhibition of Wild-Type and of Truncated Human
TrxR by Lead Compound 8 in the Steady-State. The
DTNB reducing activity of wild-type and of truncated human
TrxR was assayed by monitoring the formation of thio-
nitrobenzoate, TNB-, at 412 nm (ꢀ412 nm ) 13.6 mM-1 cm-1
)
due to oxidation of NADPH. The assay mixtures, in a total
volume of 0.5 mL, contained TrxR-buffer, and 1.8 milliunits
wild-type TrxR and 3.4 milliunits of truncated human hTrxR∆-
16, respectively. For determination of Km and Vmax, the data
were fitted to the Michaelis-Menten equation using the
computerized least-squares regression program Kaleidagraph.
Evaluation of inhibition type and inhibition constants for
inhibitor 8 was done in duplicate experiments using 200 µM
NADPH. Wild-type TrxR activity was measured at different
concentrations of inhibitor 8 (0-40 µM) in the presence of
varying concentrations of DTNB (11-435 µM). Truncated
human hTrxR∆-16 activity was measured at different concen-
trations of inhibitor 8 (0-250 µM) in the presence of varying
concentrations of DTNB (0.6-6.0 mM). Inhibition was mea-
sured as a function of DTNB concentration, and the data were
fitted to the appropriate equation by using nonlinear regres-
sion analysis software (Kaleidagraph).
photometrically at 463 nm using an ꢀ value of 11.3 mM-1 cm-1
.
In the case of DmTrxR-SCUS, atomic absorbance spectroscopy
was applied as FAD-containing, but prematurely terminated
protein prevents a reliable spectrophotometric determina-
tion.3,12,47 Thioredoxin concentrations were measured enzy-
matically by end point determination at 340 nm. Two different
assays were applied to determine thioredoxin reductase activ-
ity, the DTNB-reduction assay, where 1 unit of TrxR activity
is defined as the NADPH-dependent production of 2 µmol
2-nitro-5-thiobenzoate per min (ꢀ412 nm 13.6 mM-1 cm-1), and
the thioredoxin reduction assay where 1 unit of TrxR activity
is defined as the consumption of 1 µmol NADPH per min
(ꢀ340 nm ) 6.22 mM-1 cm-1). Similarly, one unit of GR activity
is defined as the consumption of 1 µmol NADPH per min
(ꢀ340 nm ) 6.22 mM-1 cm-1). Specific activities were essentially

Human Thioredoxin Reductase Inhibitors
Journal of Medicinal Chemistry, 2005, Vol. 48, No. 22 7037
Time-Dependent Inactivation of Human Thioredoxin
Reductase by the Four cisPt Complexes. Relative IC50
Values from Time-Dependent Inactivation of Wild-Type
and Mutant Human TrxRs. Apart from the inhibitor screen-
ing studies all further inhibition studies were performed in a
total volume of 500 µL at 25 °C in 100 mM potassium
phosphate, 2 mM EDTA, pH 7.4. Either the DTNB assay (3
mM DTNB and 200 µM NADPH; ꢀ412 nm ) 13.6 mM-1 cm-1 for
the formation of one TNB anion) or the Trx assay (20 µM
hTrxC72S and 100 µM NADPH; ꢀ340 nm ) 6.22 mM-1 cm-1) was
used. IC50 values given in Table 1 were measured when hTrxR
(5 nM) or hTrxRSec498Cys (100 nM) was incubated for 10 min
with NADPH and varying concentrations of inhibitor (10 mM
stock solution in DMF) at 25 °C. The assays were started with
DTNB and hTrxC72S, respectively, and DMF was used in
control assays.
Time-Dependent Inactivation of Wild-Type Human
TrxR. For determining rate constants of hTrxR inactivation
TrxR activity was monitored over the time by following a
preincubation protocol. To a final volume of 500 µL TrxR buffer
were added at 25 °C 100 µM NADPH, 0-36.6 µM inhibitor,
and 1.83 µM TrxR. All reaction mixtures contained 2% DMF.
At different time points 5 µL aliquots of each reaction mixture
were removed, and the residual activity was measured in the
standard Trx reduction assay at 25 °C.51
Inhibition of Human TrxR by the Four cisPt Com-
plexes in the Steady-State. For determining type and
constants of wt-hTrxR inhibition by varying concentrations
of inhibitor 13a-c and 20 (0-800 nM), the reactions were
started following NADPH addition without preincubation. The
standard DTNB reduction assay mixtures contained 200 µM
NADPH, varying concentrations of DTNB (100-500 µM), 10
nM wt-hTrxR. The DTNB reduction was monitored at 412 nm,
following TNB formation. The standard Trx reduction assay
mixtures contained 100 µM NADPH, varying concentrations
of hTrxC72S (8-100 µM), 20 nM wt-hTrxR. Initial Trx
reduction rates were recorded from NADPH oxidation at 340
nm and 25 °C. To study the inhibition of hTrxRSec498Cys in
the presence of 20 to 250 µM inhibitor, the standard DTNB
reduction mixtures contained 100 nM mutant enzyme and
varying concentrations of DTNB (100-4000 µM), while Trx
reduction was monitored in assays using 1.3 µM mutant
enzyme and 10 to 100 µM Trx.
Inhibition of Glutathione Reductase. The inhibition of
human glutathione reductase was tested at 25 °C in 47 mM
potassium phosphate, 1 mM EDTA, 200 mM KCl, pH 6.9. After
preincubation of 2-10 mU/mL enzyme with 100 µM NADPH
and varying inhibitor concentrations, the assay was started
with 100 µM GSSG and the consumption of NADPH was
monitored as the decrease in absorbance at 340 nm.47 IC50
values were determined in comparison with DMF-containing
controls.
CO2 atmosphere. Cytotoxicity was determined as described
previously.52 One mg/mL stock solutions of the platinum
complexes in DMF were prepared. Cisplatin (Alfa, Karlsruhe)
was dissolved in water at a concentration of 0.5 mg/mL. A
maximum concentration of 50 µg/mL was applied to the cells
for all compounds. Cells were grown in 96-well plates at a
density of 2 × 106 cells/plate. One day later, test compounds
were added in quadruplicate. Serial 1:2 dilutions were pre-
pared directly on the plates using multichannel pipets. The
resulting inhibitor concentrations were 50, 25, 12.5, 6.25, 3.1,
1.6, and 0.8 µg/mL. After incubation for 96 h, cells were fixed
with 3% formaldehyde and stained with 1% crystal violet. The
amount of bound crystal violet, which is linearly dependent
on the number of adherent, surviving cells was determined in
an Antos 2001 ELISA reader at 595 nm. In the case of MCF-7
cells, which grow inhomogeneously, the dye was eluted with
200 µL ethanol/1% acetic acid before absorption determination.
IC50-values were taken from dose-response curves after curve
fitting.
DNA Interaction Studies. Stock solutions of tested sub-
stances in DMF (10 mM) were prepared and used within 1 h.
Gene ruler 1kb DNA Ladder (MBI) was diluted with water to
a final concentration of 50 µg/mL. DNA samples (50 µL each)
were mixed with the tested substances to a final concentration
of 1, 0.1 and 0.01 µM. cisPt was added to a concentration of
0.1, 0.01, and 0.001 mM. One sample was treated with 10%
DMF as a control. Samples were incubated at room temper-
ature for 16 h. Aliquots of 25 µL were electrophoresed into a
1.4% agarose gel (100 mM Tris/phosphate buffer) at 5 V/cm
for 3 h. The gel was stained with ethidium bromide and
photographed under UV light.
Acknowledgment. This work was supported by
Institut Pasteur de Lille (Fellowship attributed to R.M.),
by the Centre National de la Recherche Scientifique-
Forschungsgemeinschaft Funds 02N40/0800 (E.D.-C,
S.G., K.B.) and by the Deutsche Forschungsgemein-
schaft (Be 1540/6-2 to K.B., GR 2028/1-1 to S.G.). The
authors are grateful to Athel Cornish-Bowden and
Maria Luz Cardenas for fruitful discussions. They also
thank Anick Lemaire (CNRS France, Lille) and Elisa-
beth Fischer (Giessen University) for their excellent
contribution in the primary screening of the 12 000
compound-based library and technical assistance, re-
spectively. Mathieu Sauty and Julien Furrer are ac-
knowledged for recording ICP-MS spectra (Service Eaux
Environnement, Institut Pasteur de Lille, Dr. Pierre
Thomas) and 195Pt NMR spectra (Organisch-Chemisches
Institut, Heidelberg University), respectively.
Time-dependent Inactivation of D. melanogaster
Thioredoxin Reductase. DmTrxR (2 µL of 123.9 µM enzyme
solution) was incubated at 25 °C in a final volume of 50 µL of
100 mM potassium phosphate buffer pH 7.4, with 160 µM
NADPH and the inhibitor at different concentrations (2% final
DMF concentration was kept constant) for 60 min. Final
concentrations of inhibitors 13a-c and 20 in the preincubation
mixture were 0-10-25-50 µM. The inactivation process was
stopped by diluting a 5 µL aliquot of each incubation mixture
into a UV cuvette containing buffer to measure the residual
Trx reducing activity. The reaction mixture consisted of 100
mM potassium phosphate buffer pH 7.4, 2 mM EDTA, 100 µM
NADPH, 1 mM GSSG, and 43.6 µM DmTrx (0.5% DMF final).
The loss of D. melanogaster TrxR activity was determined by
monitoring Trx reduction as a function of time of the pre-
incubation mixture.51 NADPH oxidation was monitored at
340 nm.
Supporting Information Available: Experimental pro-
cedures, spectrosopic data, and elemental analyses for the
preparation and the characterization of compounds 1 to 20 are
available free of charge via the Internet at http://pubs.acs.org.
References
(1) Wong, E.; Giandomenico, C. M. Current status of platinum-based
antitumor drugs. Chem. Rev. 1999, 99, 2451-2466.
(2) Reedijk, J. Why does Cisplatin reach Guanine-n7 with competing
s-donor ligands available in the cell? Chem. Rev. 1999, 99, 2499-
2510.
(3) Gromer, S.; Urig, S.; Becker, K. The thioredoxin systemsfrom
science to clinic. Med. Res. Rev. 2004, 24, 40-89.
(4) Holmgren, A. Thioredoxin and glutaredoxin systems. J. Biol.
Chem. 1989, 264, 13963-13966.
(5) Krnajski, Z.; Gilberger, T. W.; Walter, R. D.; Cowman, A. F.;
Muller, S. Thioredoxin reductase is essential for the survival of
Plasmodium falciparum erythrocytic stages. J. Biol. Chem. 2002,
277, 25970-25975.
(6) Missirlis, F.; Ulschmid, J. K.; Hirosawa-Takamori, M.; Gronke,
S.; Schafer, U.; Becker, K.; Phillips, J. P.; Jackle, H. Mitochon-
drial and cytoplasmic thioredoxin reductase variants encoded
by a single Drosophila gene are both essential for viability. J.
Biol. Chem. 2002, 277, 11521-11526.
Cytotoxicity Assays. Calu-6 (adenocarcinoma, lung), SK-
MEL 25 (melanoma), and MCF-7 (mamma carcinoma) cells
were obtained from the American Type Culture Collection and
were cultivated according to instructions of the supplier at 37
°C in Dulbecco’s modified Eagle’s medium (DMEM) in a 5%

7038 Journal of Medicinal Chemistry, 2005, Vol. 48, No. 22
Millet et al.
(7) Missirlis, F.; Phillips, J. P.; Jackle, H. Cooperative action of
antioxidant defense systems in Drosophila. Curr. Biol. 2001, 11,
1272-1277.
(8) Jakupoglu, C.; Przemeck, G. K.; Schneider, M.; Moreno, S. G.;
Mayr, N.; Hatzopoulos, A. K.; de Angelis, M. H.; Wurst, W.;
Bornkamm, G. W.; Brielmeier, M.; Conrad, M. Cytoplasmic
thioredoxin reductase is essential for embryogenesis but dis-
pensable for cardiac development. Mol. Cell Biol. 2005, 25,
1980-1988.
(9) Conrad, M.; Jakupoglu, C.; Moreno, S. G.; Lippl, S.; Banjac, A.;
Schneider, M.; Beck, H.; Hatzopoulos, A. K.; Just, U.; Sinowatz,
F.; Schmahl, W.; Chien, K. R.; Wurst, W.; Bornkamm, G. W.;
Brielmeier, M. Essential role for mitochondrial thioredoxin
reductase in hematopoiesis, heart development, and heart
function. Mol. Cell Biol. 2004, 24, 9414-9423.
(10) Anestål, K.; Arne´r, E. S. Rapid induction of cell death by
selenium-compromised thioredoxin reductase 1 but not by the
fully active enzyme containing selenocysteine. J. Biol. Chem.
2003, 278, 15966-15972.
(11) Sasada, T.; Nakamura, H.; Ueda, S.; Sato, N.; Kitaoka, Y.; Gon,
Y.; Takabayashi, A.; Spyrou, G.; Holmgren, A.; Yodoi, J. Possible
involvement of thioredoxin reductase as well as thioredoxin in
cellular sensitivity to cis-diamminedichloroplatinum (II). Free
Radic. Biol. Med. 1999, 27, 504-514.
(12) Arscott, L. D.; Gromer, S.; Schirmer, R. H.; Becker, K.; Williams,
C. H., Jr. The mechanism of thioredoxin reductase from human
placenta is similar to the mechanisms of lipoamide dehydro-
genase and glutathione reductase and is distinct from the
mechanism of thioredoxin reductase from Escherichia coli. Proc.
Natl. Acad. Sci. U.S.A. 1997, 94, 3621-3626.
(13) Gromer, S.; Wissing, J.; Behne, D.; Ashman, K.; Schirmer, R.
H.; Flohe, L.; Becker, K. A hypothesis on the catalytic mecha-
nism of the selenoenzyme thioredoxin reductase. Biochem. J.
1998, 332 (Pt 2), 591-592.
(14) Zhong, L.; Arne´r, E. S.; Ljung, J.; A° slund, F.; Holmgren, A. Rat
and calf thioredoxin reductase are homologous to glutathione
reductase with a carboxyl-terminal elongation containing a
conserved catalytically active penultimate selenocysteine resi-
due. J. Biol. Chem. 1998, 273, 8581-8591.
(15) Lee, S. R.; Bar-Noy, S.; Kwon, J.; Levine, R. L.; Stadtman, T.
C.; Rhee, S. G. Mammalian thioredoxin reductase: oxidation of
the C-terminal cysteine/selenocysteine active site forms a thio-
selenide, and replacement of selenium with sulfur markedly
reduces catalytic activity. Proc. Natl. Acad. Sci. U.S.A. 2000,
97, 2521-2526.
(16) Bauer, H.; Massey, V.; Arscott, L. D.; Schirmer, R. H.; Ballou,
D. P.; Williams, C. H., Jr. The mechanism of high Mr thioredoxin
reductase from Drosophila melanogaster. J. Biol. Chem. 2003,
278, 33020-33028.
(17) Sandalova, T.; Zhong, L.; Lindqvist, Y.; Holmgren, A.; Schneider,
G. Three-dimensional structure of a mammalian thioredoxin
reductase: implications for mechanism and evolution of a
selenocysteine-dependent enzyme. Proc. Natl. Acad. Sci. U.S.A.
2001, 98, 9533-9538.
(18) Kaim, W.; Schwederski, B. Bioanorganische Chemie. Zur Funk-
tion chemischer Elemente in Lebensprozessen, 2nd ed.; Teubner
Verlag: Stuttgart, 1995; p 460.
(19) Gromer, S.; Arscott, L. D.; Williams, C. H., Jr.; Schirmer, R. H.;
Becker, K. Human placenta thioredoxin reductase. Isolation of
the selenoenzyme, steady-state kinetics, and inhibition by
therapeutic gold compounds. J. Biol. Chem. 1998, 273, 20096-
20101.
(20) Becker, K.; Herold-Mende, C.; Park, J. J.; Lowe, G.; Schirmer,
R. H. Human thioredoxin reductase is efficiently inhibited by
(2,2′:6′,2′′-terpyridine)platinum(II) complexes. Possible implica-
tions for a novel antitumor strategy. J. Med. Chem. 2001, 44,
2784-2792.
(21) Irmler, A.; Bechthold, A.; Davioud-Charvet, E.; Hofmann, V.;
Re´au, R.; Gromer, S.; Schirmer, R. H.; Becker, K. Disulfide
Reductases - Current Developments. Flavins and Flavoproteins
2002; Agency for Scientific Publications: Berlin, 2002; pp 803-
815.
(22) Jencks, W. P. On the attribution and additivity of binding
energies. Proc. Natl. Acad. Sci. U.S.A. 1981, 78, 4046-4050.
(23) Pasini, A. A doxorubicin-Pt(II) complex with antitumor activity.
Synthesis, characterization and reactions with guanosine and
DNA. Gazz. Chim. Ital. 1987, 117, 763-768.
(26) Perez, J. M.; Lopez-Solera, I.; Montero, E. I.; Brana, M. F.;
Alonso, C.; Robinson, S. P.; Navarro-Ranninger, C. Combined
effect of platination and intercalation upon DNA binding of novel
cytotoxic Pt-bis(naphthalimide) complexes. J. Med. Chem. 1999,
42, 5482-5486.
(27) von Nussbaum, F.; Miller, B.; Wild, S.; Hilger, C. S.; Schumann,
S.; Zorbas, H.; Beck, W.; Steglich, W. Synthesis of 1-(2-amino-
phenyl)isoquinolines and the biological activity of their cis-
dichloro platinum(II) complexes. J. Med. Chem. 1999, 42, 3478-
3485.
(28) Davioud-Charvet, E.; Becker, K.; Landry, V.; Gromer, S.; Loge,
C.; Sergheraert, C. Synthesis of 5,5′-dithiobis(2-nitrobenzamides)
as alternative substrates for trypanothione reductase and
thioredoxin reductase: a microtiter colorimetric assay for inhibi-
tor screening. Anal. Biochem. 1999, 268, 1-8.
(29) Cenas, N. K.; Bironaite, D. A.; Kulys, J. J.; Sikhova, N. M.
Interaction of nitrofurans with glutathione reductase. Biochim.
Biophys. Acta 1991, 1073, 195-199.
(30) Miskiniene, V.; Sarlauskas, J.; Jacquot, J. P.; Cenas, N. Nitro-
reductase reactions of Arabidopsis thaliana thioredoxin reduc-
tase. Biochim. Biophys. Acta 1998, 1366, 275-283.
(31) Gromer, S.; Johansson, L.; Bauer, H.; Arscott, L. D.; Rauch, S.;
Ballou, D. P.; Williams, C. H., Jr.; Schirmer, R. H.; Arne´r, E. S.
Active sites of thioredoxin reductases: why selenoproteins? Proc.
Natl. Acad. Sci. U.S.A. 2003, 100, 12618-12623.
(32) Mau, B. L.; Powis, G. Mechanism-based inhibition of thioredoxin
reductase by antitumor quinoid compounds. Biochem. Pharma-
col. 1992, 43, 1613-1620.
(33) Arne´r, E. S.; Nakamura, H.; Sasada, T.; Yodoi, J.; Holmgren,
A.; Spyrou, G. Analysis of the inhibition of mammalian thio-
redoxin, thioredoxin reductase, and glutaredoxin by cis-
diamminedichloroplatinum (II) and its major metabolite, the
glutathione-platinum complex. Free Radic. Biol. Med. 2001, 31,
1170-1178.
(34) Blumenstiel, K.; Schoneck, R.; Yardley, V.; Croft, S. L.; Krauth-
Siegel, R. L. Nitrofuran drugs as common subversive substrates
of Trypanosoma cruzi lipoamide dehydrogenase and trypano-
thione reductase. Biochem. Pharmacol. 1999, 58, 1791-1799.
(35) Gilberger, T. W.; Bergmann, B.; Walter, R. D.; Muller, S. The
role of the C-terminus for catalysis of the large thioredoxin
reductase from Plasmodium falciparum. FEBS Lett. 1998, 425,
407-410.
(36) Muller, D.; Zeltser, I.; Bitan, G.; Gilon, C. Building units for
N-backbone cyclic peptides. 3. Synthesis of protected N(alpha)-
(omega-aminoalkyl)amino acids and N(alpha)-(omega-carboxy-
alkyl)amino acids. J. Org. Chem. 1997, 62, 411-416.
(37) Thomas, P.; Finnie, J. K.; Williams, J. G. Feasibility of identi-
fication and monitoring of arsenic species in soil and sediment
samples by coupled HPLC- inductively coupled plasma mass
spectrometry. J. Anal. At. Spectrom. 1997, 12, 1367-1372.
(38) Arne´r, E. S.; Bjornstedt, M.; Holmgren, A. 1-Chloro-2,4-dinitro-
benzene is an irreversible inhibitor of human thioredoxin
reductase. Loss of thioredoxin disulfide reductase activity is
accompanied by a large increase in NADPH oxidase activity. J.
Biol. Chem. 1995, 270, 3479-3482.
(39) Kitz, R.; Wilson, I. B. Esters of methanesulfonic acid as irrevers-
ible inhibitors of acetylcholinesterase. J. Biol. Chem. 1962, 237,
3245-3249.
(40) Kuo, D. J.; Jordan, F. Active site directed irreversible inactiva-
tion of brewers' yeast pyruvate decarboxylase by the conjugated
substrate analogue (E)-4-(4-chlorophenyl)-2-oxo-3-butenoic acid:
development of a suicide substrate. Biochemistry 1983, 22,
3735-3740.
(41) Zhong, L.; Holmgren, A. Essential role of selenium in the
catalytic activities of mammalian thioredoxin reductase revealed
by characterization of recombinant enzymes with selenocysteine
mutations. J. Biol. Chem. 2000, 275, 18121-18128.
(42) Jansen, B. A.; Brouwer, J.; Reedijk, J. Glutathione induces
cellular resistance against cationic dinuclear platinum anti-
cancer drugs. J. Inorg. Biochem. 2002, 89, 197-202.
(43) Kearns, P. R.; Hall, A. G. Glutathione and the response of
malignant cells to chemotherapy. Drug Discovery Today 1998,
3, 113-121.
(44) Baker, A.; Payne, C. M.; Briehl, M. M.; Powis, G. Thioredoxin,
a gene found overexpressed in human cancer, inhibits apoptosis
in vitro and in vivo. Cancer Res. 1997, 57, 5162-5167.
(45) Hajduk, P. J.; Meadows, R. P.; Fesik, S. W. Discovering high-
affinity ligands for proteins. Science 1997, 278, 497-499.
(46) Kato, Y.; Inouye, T.; Jyosui, S.; Kakimoto, K.; Iyehara-Ogawa,
H.; Tsuruta, S. Structure-dependent variation in the mutagenic,
prophage-inducing and antibacterial activities of 5-nitro-2-
furamide derivatives. Mutat. Res. 1984, 140, 169-174.
(47) Nordhoff, A.; Bucheler, U. S.; Werner, D.; Schirmer, R. H.
Folding of the four domains and dimerization are impaired by
the Gly446fGlu exchange in human glutathione reductase.
Implications for the design of antiparasitic drugs. Biochemistry
1993, 32, 4060-4066.
(24) Palmer, B. D.; Lee, H. H.; Johnson, P.; Baguley, B. C.; Wickham,
G.; Wakelin, L. P.; McFadyen, W. D.; Denny, W. A. DNA-directed
alkylating agents. 2. Synthesis and biological activity of platinum
complexes linked to 9-anilinoacridine. J. Med. Chem. 1990, 33,
3008-3014.
(25) Gibson, D.; Gean, K. F.; Ben-Shoshan, R.; Ramu, A.; Ringel, I.;
Katzhendler, J. Preparation, characterization, and anticancer
activity of a series of cis-PtCl2 complexes linked to anthra-
quinone intercalators. J. Med. Chem. 1991, 34, 414-420.

Human Thioredoxin Reductase Inhibitors
Journal of Medicinal Chemistry, 2005, Vol. 48, No. 22 7039
(48) Bauer, H.; Kanzok, S. M.; Schirmer, R. H. Thioredoxin-2 but
not thioredoxin-1 is a substrate of thioredoxin peroxidase-1 from
Drosophila melanogaster: isolation and characterization of a
second thioredoxin in D. Melanogaster and evidence for distinct
biological functions of Trx-1 and Trx-2. J. Biol. Chem. 2002, 277,
17457-17463.
(51) Gromer, S.; Merkle, H.; Schirmer, R. H.; Becker, K. Human
placenta thioredoxin reductase: preparation and inhibitor stud-
ies. Methods Enzymol. 2002, 347, 382-394.
(52) Amtmann, E.; Zoller, M.; Wesch, H.; Schilling, G. Antitumoral
activity of a sulphur-containing platinum complex with an acidic
pH optimum. Cancer Chemother. Pharmacol. 2001, 47, 461-
466.
(53) Zhong, L.; Arne´r, E. S.; Holmgren, A. Structure and mechanism
of mammalian thioredoxin reductase: the active site is a redox-
active selenolthiol/selenenylsulfide formed from the conserved
cysteine-selenocysteine sequence. Proc. Natl. Acad. Sci. U.S.A.
2000, 97, 5854-5859.
(49) Holmgren, A.; Bjornstedt, M. Thioredoxin and thioredoxin
reductase. Methods Enzymol. 1995, 252, 199-208.
(50) Salmon-Chemin, L.; Buisine, E.; Yardley, V.; Kohler, S.; Debreu,
M. A.; Landry, V.; Sergheraert, C.; Croft, S. L.; Krauth-Siegel,
R. L.; Davioud-Charvet, E. 2- and 3-substituted 1,4-naphtho-
quinone derivatives as subversive substrates of trypanothione
reductase and lipoamide dehydrogenase from Trypanosoma
cruzi: synthesis and correlation between redox cycling activities
and in vitro cytotoxicity. J. Med. Chem. 2001, 44, 548-565.
JM050256L