Bivalent inhibitors of glutathione S-transferase: The effect of spacer length on isozyme selectivity

Article abstract of DOI:10.1016/j.bmcl.2006.04.041

Glutathione S-transferases (GSTs) are cytosolic enzymes that catalyze the conjugation of glutathione with a variety of exogenous and endogenous electrophiles. High affinity, isozyme-specific inhibitors of GST are required for use as pharmacological tools

Full text of DOI:10.1016/j.bmcl.2006.04.041

Bioorganic & Medicinal Chemistry Letters 16 (2006) 3780–3783
Bivalent inhibitors of glutathione S-transferase: The effect
of spacer length on isozyme selectivity
Dean Y. Maeda,a,* Sumit S. Mahajan,b William M. Atkinsb and John A. Zebalaa
aSyntrix Biosytems, Auburn, WA 98001, USA
bDepartment of Medicinal Chemistry, University of Washington, Seattle, WA 98195, USA
Received 7 March 2006; revised 17 April 2006; accepted 17 April 2006
Available online 3 May 2006
Abstract—Glutathione S-transferases (GSTs) are cytosolic enzymes that catalyze the conjugation of glutathione with a variety of
exogenous and endogenous electrophiles. High affinity, isozyme-specific inhibitors of GST are required for use as pharmacological
tools as well as potential therapeutics. The design of selective inhibitors is hindered due to the broad substrate binding capabilities of
the GST enzymes. GSTs are dimeric enzymes, and therefore offer a unique discriminator for achieving inhibitor selectivity: the dis-
tance between binding sites on each monomer unit as a function of its quaternary organization. Bivalent analogs of the non-selective
GST inhibitor ethacrynic acid were prepared, and selectivity for the GST A1-1 isozyme over GST P1-1 (IC50 values of 13.7 vs
1022 nM, respectively) was achieved through the optimization of the spacer length between the ethacrynic acid ligand domains.
Ó 2006 Elsevier Ltd. All rights reserved.
Glutathione S-transferases represent a family of cytosol-
ic enzymes that catalyze the conjugation of glutathione
(GSH) to a variety of exogenous and endogenous elec-
trophiles,1,2 and are implicated in tumor cell resistance
toward chemotherapy agents.3 While recent studies have
focused on the GST P1-1 enzyme and its involvement in
cell proliferation,4 considerable interest has been paid to
the study of the related isozyme GST A1-1 and its role in
the regulation of lipid peroxides,5,6 which are important
regulators of apoptosis.7 A potent and selective inhibitor
of GST A1-1 would help further characterize its bio-
chemical role in acquired tumor cell resistance and
apoptosis, and potentially result in a useful adjuvant
to restore chemotherapeutic sensitivity in resistant
tumor cells.
neered through the optimization of the spacer length
between ligand domains in bivalent inhibitor.
a
‘Selective bivalency’ describes this concept of imparting
isozyme selectivity through spacer optimization to an
otherwise non-selective inhibitor. The concept of
selective bivalency is illustrated in Figure 1. In example
I, dimeric proteins A and B differ by the spatial
non-selective inhibitor
I.
Dimeric Protein A
Dimeric Protein B
bivalent inhibitor X
The difficulty in the rational design of selective inhibi-
tors for GST lies in their inherent ability to bind and
conjugate a wide variety of hydrophobic substrates.
GSTs are dimeric proteins and therefore provide a un-
ique discriminator for selectivity: the differences in dis-
tance and environment between binding sites on each
monomer unit. This would allow selectivity to be engi-
II.
Dimeric Protein A
Dimeric Protein B
bivalent inhibitor Y
III.
Keywords: Glutathione S-transferase inhibitor; Bivalent inhibitor;
Selective bivalency.
Dimeric Protein A
Dimeric Protein B
*
Corresponding author. Tel.: +1 253 833 8009; fax: +1 253 833
8127; e-mail: dmaeda@syntrixbio.com
Figure 1. The concept of selective bivalency.
0960-894X/$ - see front matter Ó 2006 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2006.04.041

D. Y. Maeda et al. / Bioorg. Med. Chem. Lett. 16 (2006) 3780–3783
3781
orientation of the binding sites on the individual mono-
Br
Br
mer units. The non-selective monovalent inhibitor will
bind to each monomer unit binding site with equal affin-
ity, and thus exhibit no selectivity. In example II, two
non-selective inhibitors are bound together by a spacer
of optimized length, resulting in bivalent inhibitor X
with enhanced affinity to dimeric protein A. The spacer
length chosen is too short to span the gap between the
binding sites of dimeric protein B, therefore bivalent li-
gand X does not exhibit the same increase in affinity
for dimeric protein B, resulting in an overall increase
in selectivity for dimeric protein A. In example III, the
spacer length in bivalent inhibitor Y is increased such
that it is able to bind both dimeric proteins A and B.
In this case, selectivity for either dimeric protein is not
observed.
i, i
ii, iv, v
(79%)
(53%)
MeO
O
MeO
O
3
2
NH2
NH2
FmocHN
NHFmoc
vi
(91%)
O
O
HO
HO
5
4
While the concept of selective bivalency was utilized to
prepare potent and selective inhibitors of human glyox-
alase I,8 its use as a method to develop GST inhibitors
was first investigated by Lyon et al.9 in which bivalent
inhibitors based on S-nitrophenyl glutathione conju-
gates exhibited a 100-fold increase in affinity and a 10-
fold increase in isozyme selectivity for GST P1-1. In or-
der to develop selective bivalent inhibitors for GST A1-
1, the crystal structures of GST A1-110 and GST P1-111
co-crystallized with the non-selective GST inhibitor eth-
acrynic acid 112 (EA, Fig. 2) were analyzed. We hypoth-
esized that GST A1-1 selectivity could be achieved by
exploiting the spatial orientation differences between
the EA binding sites of GST A1-1 and P1-1. Bivalent
inhibitors comprised of EA ligand domains coupled to
various amino acid spacers linked to a bivalent core
were prepared and evaluated for inhibition against
GST A1-1 and GST P1-1.
Scheme 1. Synthesis of bivalent core 5. Reagents and conditions: (i) N-
bromosuccinimide, benzoyl peroxide; (ii) diethyl phosphite; (iii)
sodium azide; (iv) triphenylphosphine; (v) 2 N aqueous HCl, reflux;
(vi) Fmoc-OSu.
followed by reduction with triphenylphosphine14 and
subsequent acid-catalyzed ester hydrolysis resulted in
diamino acid 4. The free amines were then protected
using N-(9-fluorenylmethoxycarbonyl)oxysuccinimide
(Fmoc-OSu) to yield the protected bivalent core 5. The
bivalent core was then loaded onto Tentagel SRAM
resin using O-benzotriazole-N,N,N,N0-tetramethyluro-
nium hexafluorophosphate (HBTU) as the coupling
agent with N-hydroxybenzotriazole (HOBt) as an
additive. Initial synthetic efforts to prepare the bivalent
inhibitors used Fmoc-protected spacer amino acids,
but HPLC and mass spectrometric analyses of the crude
products indicated that there were substantial amounts
of by-products containing one or two additional spacer
amino acid residues. A phthaloyl-based protection
scheme resulted in higher product purities, and this
protection strategy was then adopted for the bivalent
inhibitor series. Phthaloyl-protected amino acid spacers
of different lengths were then coupled to the bivalent
core using HBTU and HOBt. The phthaloyl group
was then removed using 1 M hydrazine in EtOH, and
the carboxyl group of EA was coupled to the resulting
free amine using HBTU and HOBt. In bivalent inhibitor
7, EA was directly coupled to the bivalent core molecule
to investigate the effects of having no spacer amino acid
present. In contrast, bivalent inhibitor 13 was
synthesized with a spacer consisting of three repeating
4-aminobutyric acid units to measure the effects of a
lengthy spacer between the EA ligand domains. As a
control, the monovalent analog 6 was also synthesized
using commercially available Fmoc-3-aminomethyl
benzoic acid as the monovalent core. The inhibitors
were then cleaved from the solid support by treatment
with 95% TFA/H2O and purified by semi-preparative
HPLC. The purified compounds were then evaluated
for inhibition of GST A1-1 and GST P1-1 mediated
GSH conjugation to 1-chloro-2,4-dinitrobenzene
(CDNB)15. GST A1-1 and GST P1-1 were expressed
in Escherichia coli and purified as described previously.9
The results are shown in Table 1.
The method used to prepare the bivalent inhibitors
based on EA was adapted to a solid-phase strategy to
allow for synthetic ease as well as the potential for future
combinatorial library development. This was accom-
plished by utilizing 3,5-diaminomethylbenzoic acid as
the bivalent core, which was synthesized in its protected
form as shown in Scheme 1. Methyl 3,5-diaminomethyl
benzoate 2 was treated with N-bromosuccinimide and
catalytic amounts of benzoyl peroxide, followed by
diethyl phosphite and diisopropylethylamine to yield
dibromo ester 3.13 Treatment with sodium azide,
O
Cl
O
H
N
EA
EA
O
Cl
O
N
H
N
H
HO
N
H
EA
H N
2
O
H N
2
O
7
1: Ethacrynic acid (EA)
6
O
O
H
N
H
N
EA
EA GABA
GABA
3
3
(CH )
2 n
(CH )
2 n
N
H
N
H
N
H
N
H
EA
EA
H N
2
O
H N
2
O
13
8: n = 1
9: n = 2
10: n = 3
11: n = 5
12: n = 7
O
H
N
GABA =
Figure 2. Ethacrynic acid and bivalent inhibitors based on ethacrynic
acid.

3782
D. Y. Maeda et al. / Bioorg. Med. Chem. Lett. 16 (2006) 3780–3783
Table 1. Evaluation of compounds 1, 6–13 on the inhibition of GST A1-1 and GST P1-1
b
Selectivity A1-1/P1-1a
˚
Distance (A)
Compound
IC50 SE (nM)
GST A1-1
GST P1-1
1c
6
5000
13,850 1475
32.7 4.6
4000
0.8
1.1
30
N.A.
N.A.
19.28
26.97
27.91
32.04
37.06
42.03
56.23
15,046 1149
992.4 180
704 38
7
8
24
4
29
75
9
13.7 1.8
14.3 1.4
39.3 5.0
52.1 5.3
98.9 7.8
1022 136
624 109
10
11
12
13
44
46.1
31.0 0.3
142
2
1.2
0.6
1.4
7
IC50 values are the average SEM of three independent experiments.
a Selectivity ratio = (GST P1-1 IC50) ꢀ (GST A1-1 IC50).
b Distances between the aromatic groups of the EA ligand domains were estimated using ChemDraw 3D (CambridgeSoft, Cambridge, MA) with the
bivalent inhibitors drawn and rendered in the fully extended state.
c IC50 values for EA obtained from Ploemen et al. (Ref. 12).
Upon evaluation, the monovalent EA-based inhibitor 6
exhibited non-selective inhibition of GST A1-1 and GST
P1-1 with IC50 values of 13,580 and 12,500 nM, respec-
tively. The higher IC50 values seen for the monovalent
control would suggest that any improvement in inhibi-
tion observed for the bivalent analogs can be attributed
to factors other than favorable interaction between the
linker core with the enzyme. While bivalent inhibitors
7–10 all exhibited increased selectivity for GST A1-1
over GST P1-1, bivalent inhibitor 9 exhibited the highest
degree of selectivity (75-fold), as well as the lowest IC50
for GST A1-1 (13.7 nM). Incubation of GST A1-1 in the
presence of inhibitor 9 for a period of 2 h did not yield
any conjugation products by mass spectrometric analy-
sis (data not shown), suggesting that bivalent ligand 9
is not inhibiting GST A1-1 via covalent modification.
prepared and evaluated for GST A1-1 and P1-1 inhibi-
tion. Bivalent inhibitor 9 exhibited markedly increased
selectivity for GST A1-1 with respect to EA as well as
the monovalent analog. To our knowledge, bivalent
inhibitor 9 represents a GST A1-1 inhibitor with the
highest affinity and selectivity reported in the literature
to date, as well as an example in which the concept of
selective bivalency was used to impart isozyme selectiv-
ity to a non-selective inhibitor between two dimeric pro-
teins belonging to the same enzyme family.
Acknowledgments
The authors gratefully acknowledge Robert P. Lyon for
helpful technical discussions and editing of the
manuscript. This work was supported by Grants
GM-62284GM, T32 GM07750 and R43CA92800 from
the National Institutes of Health.
The estimated distance between the aromatic rings of
the pendant EA ligand domains in 9 corresponds well
to the measured separation distances of the bound EA
molecules in the crystal structure of GST A1-1 (27.91
˚
vs 26.0 A, respectively). As depicted in Figure 1, we
References and notes
hypothesize that bivalent inhibitor 9 is able to bind in
a bivalent manner to GST A1-1, resulting in a large in-
crease in binding affinity in regard to both the parent EA
and the monovalent control. Due to its length, it is pos-
tulated that bivalent inhibitor 9 is unable to bridge the
distance between the GST P1-1 binding sites, therefore
precluding bivalent (but not monovalent) interaction
with GST P1-1. The increase in binding affinity to
GST A1-1 resulted in a large increase in overall selectiv-
ity for GST A1-1 over GST P1-1. This selectivity is lost
as the spacer is lengthened, as depicted in the IC50 values
for bivalent inhibitors 11, 12, and 13. In these cases, we
postulate that the distance between the EA ligand do-
mains was long enough to allow bivalent interaction
with GST P1-1, resulting in a dramatic increase in
GST P1-1 binding affinity and subsequent loss of GST
A1-1 selectivity.
1. Mannervik, B. Adv. Enzymol. Relat. Areas Mol. Biol.
1985, 57, 357.
2. Mannervik, B.; Danielson, U. H. CRC Crit. Rev. Biochem.
1988, 23, 283.
3. Townsend, D. M.; Tew, K. D. Oncogene 2003, 22, 7369.
4. Ruscoe, J. E.; Rosario, L. A.; Wang, T.; Gate, L.;
Arifoglu, P.; Wolf, C. R.; Henderson, C. J.; Ronai, Z.;
Tew, K. D. J. Pharmacol. Exp. Ther. 2001, 298, 339.
5. Hurst, R.; Bao, Y.; Jemth, P.; Mannervik, B.; Williamson,
G. Biochem. J. 1998, 332, 97.
6. Yang, Y.; Cheng, J.-Z.; Singhal, S. S.; Saini, M.; Pandya,
U.; Awasthi, S.; Awasthi, Y. C. J. Biol. Chem. 2001, 276,
19220.
7. Yang, Y.; Sharma, R.; Sharma, A.; Awasthi, S.; Awasthi,
Y. C. Acta Biochim. Pol. 2003, 50, 319.
8. Zheng, Z.-B.; Creighton, D. J. Org. Lett. 2003, 5, 4855.
9. Lyon, R. P.; Hill, J. J.; Atkins, W. M. Biochemistry 2003,
42, 10418.
10. Cameron, A. D.; Sinning, I.; L’Hermite, G.; Olin, B.;
Board, P. G.; Mannervik, B.; Jones, T. A. Structure 1995,
3, 717.
11. Oakley, A. J.; Rossjohn, J.; Bello, M. L.; Caccuri, A. M.;
Federici, G.; Parker, M. W. Biochemistry 1997, 36, 576.
In summary, the concept of selective bivalency was
applied to the non-selective inhibitor EA to impart selec-
tivity for GST A1-1. Bivalent inhibitors with varying
spacer lengths between the EA ligand domains were

D. Y. Maeda et al. / Bioorg. Med. Chem. Lett. 16 (2006) 3780–3783
3783
12. Ploemen, J. H.; Ommen, B. V.; Bladeren, P. J. V. Biochem.
Pharmacol. 1990, 40, 1631.
13. Liu, P.; Chen, Y.; Deng, J.; Tu, Y. Synthesis 2001, 14,
2078.
14. Martin, V. V.; Lex, L.; Keana, J. F. W. Org. Prep. Proced.
Int. 1995, 27, 117.
15. Briefly, the inhibitors were dissolved in DMSO to yield a
2 mM stock solution and then serially diluted with water
to yield the final test concentrations (final DMSO
concentration <1%). The GST enzymes (20 nM) were
incubated in the presence or absence (vehicle control) of
test inhibitors with GSH and CDNB. The concentrations
of GSH and CDNB used were at their respective KM
values (A1-1: 350 and 720 lM; P1-1: 150 and 820 lM).
Rates of product formation were obtained by measuring
absorption at 340 nM for 1 min on Beckmann DU 7400
spectrophotometer. Inhibitor concentrations spanned four
orders of magnitude and experiments were carried out in
triplicate. The resulting data were then fitted to sigmoidal
dose–response curves using GraphPad Prism (GraphPad
Software Inc., San Diego, CA) to determine the 50%
inhibitory concentration.