Structure-activity relationship studies of a bisbenzimidazole-based, Zn2+-dependent inhibitor of HCV NS3 serine protease

Article abstract of DOI:10.1016/S0960-894X(01)00457-7

A survey of isosteric replacements of the phosphonoalanine side chain coupled with a process of conformational constraint of a bisbenzimidazole-based, Zn²⁺-dependent inhibitor of hepatitis C virus (HCV) NS3 serine protease resulted in the identification of novel series of active compounds with extended side chains. However, Zn²⁺-dependent HCV NS3 inhibition was relatively insensitive to the structural variations examined but dependent on the presence of negatively charged functionality. This result was interpreted in the context of an initial electrostatic interaction between protease and inhibitor that is subsequently consolidated by Zn²⁺, with binding facilitated by the featureless active site and proximal regions of the HCV NS3 protein.

Full text of DOI:10.1016/S0960-894X(01)00457-7

Bioorganic & Medicinal Chemistry Letters 11 (2001) 2355–2359
Structure–Activity Relationship Studies of a Bisbenzimidazole-
Based, Zn2+-Dependent Inhibitor of HCV NS3 Serine Protease
Kap-Sun Yeung,a,* Nicholas A. Meanwell,a Zhilei Qiu,a Dennis Hernandez,a
Sharon Zhang,a Fiona McPhee,a Steve Weinheimer,a James M. Clarkb
and James W. Jancb
aBristol-Myers Squibb Pharmaceutical Research Institute, 5 Research Parkway, POBox 5100, Wallingford, CT 06492, USA
bAxys Pharmaceuticals, 180 Kimball Way, South San Francisco, CA 94080, USA
Received 6 April 2001; accepted 21 June 2001
Abstract—A survey of isosteric replacements of the phosphonoalanine side chain coupled with a process of conformational con-
straint of a bisbenzimidazole-based, Zn2+-dependent inhibitor of hepatitis C virus (HCV) NS3 serine protease resulted in the
identification of novel series of active compounds with extended side chains. However, Zn2+-dependent HCV NS3 inhibition was
relatively insensitive to the structural variations examined but dependent on the presence of negatively charged functionality. This
result was interpreted in the context of an initial electrostatic interaction between protease and inhibitor that is subsequently con-
solidated by Zn2+, with binding facilitated by the featureless active site and proximal regions of the HCV NS3 protein. # 2001
Elsevier Science Ltd. All rights reserved.
Hepatitis C virus (HCV) is the major etiological agent
of non-A, non-B hepatitis that was identified in 1989.1 It
has been estimated that 1% of the global population is
infected with HCV, affecting approximately 4 million
people in the USA. A chronic HCV infection develops
in 75% of these individuals and has been linked to the
development of liver cirrhosis and hepatocellular carci-
noma. At present, the development of an effective vac-
cine remains an uncertain prospect and current therapy
for HCV disease using interferon in combination with
ribavirin is less than optimal due to toxicity and low
long term response rates.2
important drug discovery target that has been the sub-
ject of intensive study in the search for efficacious anti-
HCV drugs.2ꢀ4
As part of an effort directed towards the discovery of
small molecule inhibitors of HCV NS3 protease, we
sought to take advantage of a serine protease inhibitor
chemotype in which inhibition is dependent on the for-
mation of a ternary complex between enzyme, inhibitor,
and Zn2+.5 Panning of a library of bisbenzimidazole
derivatives as potential inhibitors of HCV NS3 identi-
fied the phosphonoalanine derivative 1 (APC-6336) as a
potent inhibitor of the protease that displayed an IC50
of 0.20 mM in the presence of Zn2+.6 In the absence of
Zn2+, APC-6336 (1) was over 800-fold less active with
an IC50 of 167 mM (Table 1).7
During the replication of HCV, proteolytic processing
of the nonstructural (NS) proteins at four specific sites
(NS3/NS4A, NS4A/NS4B, NS4B/NS5A, and NS5A/
NS5B) is carried out by a protease located within the
amino terminal of nonstructural protein 3 (NS3).3 The
HCV NS3 protease has been determined to be a serine
protease with a chymotrypsin/trypsin-like fold that
possesses a Ser-His-Asp catalytic triad. It has been
established that enzymatic activity is markedly enhanced
by a cofactor, NS4A.3 Since the HCV NS3 serine protease
plays an essential role in viral replication, it represents an
It was envisioned that the interaction of APC-6336 (1)
with HCV NS3 was such that the bisbenzimidazole core
occupied the active site of the enzyme and, together with
the catalytic serine hydroxyl and histidine imidazole
acting as the other two ligands, formed a tetrahedral
*Corresponding author. Tel.: +1-203-677-6897; fax: +1-203-677-
7702; e-mail: kapsun.yeung@bms.com
0960-894X/01/$ - see front matter # 2001 Elsevier Science Ltd. All rights reserved.
PII: S0960-894X(01)00457-7

2356
K.-S. Yeung et al. / Bioorg. Med. Chem. Lett. 11 (2001) 2355–2359
Table 1. Assay results for compounds 1–4 against HCV NS3 protease
domain [amino acids (aa) 1–183] in the presence of Zn2+7
This theme was further developed by examining the
series of amino acid sulfonamides depicted in Figure 2.
Analogues 9, 10, and 12 demonstrated similar inhibitory
activity to 5, presumably because the acidic acylsufona-
mide and attendant ortho carboxylate together per-
formed as an effective phosphonate mimic. However,
the mono methyl esters 7 and 8 retain significant inhi-
bitory properties, suggesting that non-charged polarity
is sufficient in this context, but diesters 6 and 11 are
notably less effective although still detectably active.
Comparing the results for 9 and 10 indicates that the
CONH2 moiety distal to the amino acid element does
not contribute significantly to inhibitory activity whilst
the data recorded for 9 and 128 reveal little dependence
on the separation of the charged terminus and the het-
erocyclic nucleus. The iminodiacetic acid9 derivative 13,
in which the dicarboxylate moiety is conformationally
more flexible and based on a less hydrophobic scaffold,
is only 2-fold less potent than 5.
Compound
n
IC50 (mM)
1 (APC-6336)
1
2
3
5
0.20a
0.87
0.82
0.51
2
3
4
aIC50=167 mM in the absence of Zn2+ and the presence of EDTA.
coordination complex with a Zn2+ atom.5 As a con-
sequence of the pseudo symmetrical topology, the
phosphonoanaline moiety could project into either the P
or P0 sites.
Further studies focused on exploring a combination of
conformational constraint within the context of an
extended side chain coupled with phosphonate replace-
ments as a means of enhancing potency. Tyrosyl acetic
acid and its carboxylated analogues have been used as
phosphonate isosteres10 and were selected as convenient
platforms with which to explore the presentation of
acidic functionality, as summarized in Figure 3. From
this series, the simple tyrosine-O-acetic acid analogue 17
emerged as the most potent compound, IC50=1 mM in
the presence of Zn2+. However, the introduction of
As one aspect of the development of SAR for this series
of HCV NS3 protease inhibitors, interactions of the
phosphonoalanine side chain of APC-6336 (1) with the
protease were examined in a broad fashion by a process
of isosteric replacement and side-chain rigidification
with simultaneous introduction of hydrophobicity, a
strategy that also served as an avenue to identify new
points of contact between inhibitor and protease. It was
anticipated that the results of these studies would lead
to the design of a new generation of non-metal-medi-
ated inhibitors that possessed more desirable structural
features, particularly reduced anionic character.
A survey of phosphonoalanine homologues was con-
ducted initially in order to determine the effect of vary-
ing carbon chain length on the potency of compound 1.
The results of this study are summarized in Table 1 and
reveal that sub-micromolar, Zn2+-dependent inhibitory
activity was largely maintained as the chain length was
increased. The fully extended AP-7 derivative 4 is only
about 2-fold less active than prototype 1 despite a
marked increase in conformational flexibility. Indeed,
the activity may be retained as a consequence of the
structural mobility of the side chain. However, replacing
the phosphoalanine moiety with either aspartate or
glutmate produced compounds that were over 200-fold
less potent than 1. Since a gem-dicarboxyl group may
more effectively mimic the doubly charged phosphonate
moiety, the g-carboxyglutamic acid analogue 5 was
synthesized and found to exhibit significant inhibitory
activity with an IC50 of 2.6 mM in the presence of Zn2+
(Fig. 1).
Figure 1. Compound 5 and, in parentheses, the IC50 (mM) as an inhi-
bitor of the HCV NS3 protease domain (aa 1–183) in the presence of
Figure 2. Compounds 6–13 and, in parentheses, their IC50 (mM) as
inhibitors of HCV NS3 protease domain (aa 1–183) in the presence of
Zn2+ 7,8
.
Zn2+ 7
.

K.-S. Yeung et al. / Bioorg. Med. Chem. Lett. 11 (2001) 2355–2359
2357
isomer, 23. Isophthalate analogues 25 and 26, in which
the additional carboxyl group may function as the a-
carboxylate group of 1, afforded the best potency in this
series.
The inhibitors of HCV NS3 protease described in this
survey exhibit a relatively flat SAR profile. Although
several novel replacements for the phosphonoalanine
moiety of 1 were identified that retained reasonable
potency, none surpassed the activity of the prototype or
its close homologues. Whilst the activity of compound
17, with an extended tyrosine-O-acetic acid side chain, is
particularly interesting, it is not clear that this repre-
sents the result of a specific binding interaction with the
HCV NS3 protein. The results described here are remi-
niscent of the results reported for a series of inhibitors
of HCMV protease designed to take advantage of the
concept of Zn2+-stabilization of the inhibitor–protease
complex.11 There appears to be a significant contribu-
tion of electrostatic interactions in the vicinity of the
active site to the recognition of substrates by HCV NS3
protease.12ꢀ14 The substrate recognition groove and
binding pockets of HCV NS3 lack the deep invagina-
tions more typically observed with mammalian serine
proteases. As a consequence, a plausible explanation for
the results reported herein relies on an initial attractive
electrostatic interaction between protease and inhibitor
that is subsequently consolidated and stabilized by the
zinc ion. The flat and featureless environment of the
active site and adjacent substrate recognition sites will
likely require a more subtle and well defined stereo-
chemical presentation of a combination of com-
plementary functionality and hydrophobic elements if a
more coherent SAR is to be observed and more potent
inhibitors are to be identified.
Figure 3. Compounds 14–21 and, in parentheses, their IC50 (mM)
against full-length NS3 in the presence of Zn2+ 7
(a) Same activity
against NS3 protease domain (aa 1–183); (b) Activity against NS3
protease domain (aa 1–183).
.
additional polar functionality to the phenyl ring led to
decreased potency (compare 17 with 16, 19, and 20).
The inhibitory activity shows little sensitivity to the
relative topology of the acetic acid moiety since the
para-phenoxyacetic acids 16 are only 2- to 3-fold more
potent than 19 or 20. However, the conformational
constraint incorporated into tyrosine-O-acetic acid 17
suggests that it is not functioning as a direct mimic of 1,
assuming a similar binding mode of the bis-benzimida-
zole core.5 It is plausible that 17 identifies a specific
binding interaction more closely associated with the
acidic moiety of 4 distal to the heterocyclic nucleus.
The bisbenzimidazole core acids 27 and 28 were pre-
pared in a similar fashion to the method reported in ref
6. The side chains were coupled to the core acid either as
the preformed TMS ester using MSTFA, or as the
In order to determine the importance of the a-carboxyl
group of the tyrosine-O-acetic acid inhibitors, the tyr-
amine analogue 18 was prepared. This compound was
an order of magnitude less potent than prototype 17, a
circumstance not altered by increasing the lipophilicity
of the carboxyl terminus, as in the acylsulfonamide 21.
These results suggest a substantial contribution by the
a-carboxyl group to the HCV NS3 inhibitory activity of
this chemotype.
As a complementary extension to this structure–activity
survey, constraining the flexibility of the side chain at
the a-position was also examined. An aminobenzoic
acid template provided a convenient means to explore
this concept and the compounds prepared are depicted
in Figure 4. From this exercise, it was found that a
meta-aminobenzoic acid, 24, was superior to the para
Figure 4. Compounds 22–26 and, in parenthesis, their IC50 (mM)
against NS3 protease domain (aa 1–183) in the presence of Zn2+ 7
.

2358
K.-S. Yeung et al. / Bioorg. Med. Chem. Lett. 11 (2001) 2355–2359
methyl or benzyl ester followed by deprotection under
basic hydrolytic or hydrogenolysis conditions (Scheme 1).
The use of PyBroP/iPr2NEt in DMF appeared to be
more generally useful for the coupling than other
reagents (e.g., PyBoP/iPr2NEt, HATU/iPr2NEt, EDC/
HOBT and DPPA/Et3N). All coupling reactions were
performed under a nitrogen atmosphere to minimize air
oxidation of the bridgehead methylene carbon. The final
products were purified by either reverse-phase pre-
parative HPLC (H2O/MeOH/0.1% TFA) or flash
chromatography on silica gel using EtOAc/iPrOH/aq
NH3 as the eluent.
benzoate (prepared from the Diel–Alder reaction
between 2-(trimethylsilyloxy)-1,3-cyclohexadiene and
ethyl tetrolate16) as the starting material, respectively
(Scheme 2).
The tyrosine-O-acetic acid side chain of compound 17
was prepared by O-alkylation of N-tBoc-tyrosine-a-
methyl ester with methyl bromoacetate. The tyramine
side chains of 18 and 21 was prepared by O-alkylation
of 4-hydroxyphenylacetonitrile followed by CoCl2-
mediated NaBH4 reduction17 of the nitrile (Scheme 3).
The side chains of compounds 22–26 were prepared
from the coupling nitrobenzoic acid or mono-methyl 5-
nitroisophthalate to the amino acid methyl ester via the
acid chloride (SOCl2), followed by reduction of the
nitro group.
The side chains of compounds 6–13 were prepared from
the coupling of the a-benzyl ester of N-tBoc-aspartic (or
glutamic) acid with methyl 2-(aminosulfonyl)benzoate
using EDC/DMAPin CH 2Cl2, or with iminodiacetic
acid dimethyl ester using PyBroP/iPr2NEt in DMF.
The side chains of compounds 14 and 16 were prepared
from methyl 5-methyl salicylate in five steps with the
addition of diethyl acetamidomalonate to the inter-
mediate benzylbromide as the key reaction.15 The side
chains of 19 and 20 were prepared analogously using 4-
methylsalicyclic acid and ethyl 2-methyl-4-hydroxy-
Acknowledgements
We thank Dr. Kenneth D. Rice (Axys), and Dr. William
Harte and Dr. Michael A. Walker (Bristol-Myers
Squibb) for interesting discussions. We also thank Dis-
covery Analytical Sciences (BMS, Wallingford) for
1
collecting some of the H NMR spectra.
References and Notes
1. Houghton, M. In Fields Virology, 3rd ed.; Field, B. N.,
Knipe, D. M., Howley, P. M., Eds.; Raven: New York, 1996;
Vol. 1, p 1035.
2. (a) Reding, M. T. Expert Opin. Ther. Pat. 2000, 10, 1201.
(b) Walker, M. A. Drug Disc. Today 1999, 4, 518.
3. (a) De Francesco, R.; Steinkuhler, C. Curr. Top. Microbiol.
Immunol. 2000, 242, 149. (b) Perni, R. B. Drug News Perspect.
2000, 13, 69. (c) Bartenschlager, R. J. Viral Hepatitis 1999, 6,
16.
Scheme 1. General synthesis of compounds 2–26.
4. Shoemaker, K. R. Curr. Opin. AntiInfect. Invest. Drugs
2000, 1, 559 and references cited therein.
5. (a) Katz, B. A.; Clark, J. M.; Finer-Moore, J. S.; Jenkins,
T. E.; Johnson, C. R.; Ross, M. J.; Luong, C.; Moore, W. R.;
Stroud, R. M. Nature 1998, 391, 608. (b) Clark, J. M.; Jenkins,
T. E.; Katz, B. A.; Stroud, R. M. US Patent 5,693,515, 1997;
Chem. Abstr. 1997, 128, 45242.
6. Hataye, J. M.; Rice, K.; Shelton, E. J.; Spencer, J. R.;
Wang, V. R. Patent WO 0020400, 2000; Chem. Abstr. 2000,
132, 279216.
Scheme 2. (a) (i) MeI (for 14) or methyl bromoacetate (for 16),
K2CO3, DMF, rt; (ii) NBS, Bz2O2, CCl4, reflux; (iii) diethyl acet-
amidomalonate, NaOEt, EtOH, rt; (b) (i) 1 N HCl/H2O, reflux; (ii)
48% HBr, reflux (for 14) or concd H2SO4, MeOH, reflux (for 16).
7. The activities of the inhibitors against NS3 protease
domain (aa 1–183) were measured using the assay conditions
reported in ref 6. Activity assay against the full length NS3/4A
(aa 1027–1711) was performed using conditions similar to
above and with 30 nM of the enzyme and 12.5 mM of ZnCl2.
The enzyme was expressed and purified as described: Hama-
take, R.; Wang, H.-G. H.; Butcher, J. A.; Bifano, M.; Clark,
G.; Hernandez, D.; Zhang, S.; Racela, J.; Standring, D.;
Colonno, R. Intervirology 1996, 39, 249.
8. In the absence of Zn2+ and the presence of 10 mM of
EDTA, the compounds reported in this letter in general
exhibited an IC50 of >100 to >300 mM. Detectable Zn-inde-
pendent activities of a few examples (like compounds 9 and
12) against HCV NS3 protease domain (aa 1-183) were
observed, which were 10- to >20-fold less than the corre-
sponding activities in the presence of Zn2+. These observations
Scheme 3. (a) (i) Methyl bromoacetate, K2CO3, DMF, rt; (ii) NaBH4,
CoCl2, MeOH/THF, rt (for 18); (b) (i) Boc2O, Et3N, CH2Cl2, rt; (ii)
NaOH (1 N, aq), MeOH, rt; (iii) PhSO2NH2, EDC, DMAP, CH2Cl2,
rt; (iv) 4 N HCl, dioxane, rt (for 21).

K.-S. Yeung et al. / Bioorg. Med. Chem. Lett. 11 (2001) 2355–2359
2359
were studied separately in more detail but this did not lead to
any tractable results.
Bailey, M.; Fazal, G.; Goulet, S.; Halmos, T.; Laplante, S.;
Maurice, R.; Poirier, M.; Poupart, M.-A.; Thibeault, D.;
Wernic, D.; Lamarre, D. Bioorg. Med. Chem. Lett. 1998, 8,
1713.
14. A combination of anionic and lipophilic interactions with
protein surface has been utilized in the design of potent and
selective inhibitors of human leukocyte elastase: Regan, J.;
McGarry, D.; Bruno, J.; Green, D.; Newman, J.; Hsu, C.-Y.;
Kline, J.; Barton, J.; Travis, J.; Choi, Y. M.; Volz, F.; Pauls,
H.; Harrison, R.; Zilberstein, A.; Ben-Sasson, S. A.; Chang,
M. J. Med. Chem. 1997, 40, 3408.
15. Leonard, F.; Wajngurt, A.; Tachannen, W.; Klein, M. J.
Med. Chem. 1967, 10, 478.
16. Freskos, J. N.; Morrow, G. W.; Swenton, J. S. J. Org.
Chem. 1985, 50, 805.
17. Satoh, T.; Suzuki, S.; Suzuki, Y.; Miyaji, Y.; Imai, Z.
Tetrahedron Lett. 1969, 52, 4555.
9. We had attempted to use the iminodiacetic acid as a build-
ing block to prepare dendritic acidic side chains; cf. Boger,
D. L.; Goldberg, J.; Jiang, W.; Chai, W.; Ducray, P.; Lee,
J. K.; Ozer, R. S.; Andersson, C.-M. Bioorg. Med. Chem.
1998, 6, 1347.
10. Burke, T. R., Jr.; Yao, Z.-J.; Zhao, H.; Milne, G. W. A.;
Wu, L.; Zhang, Z.-Y.; Voigt, J. H. Tetrahedron 1998, 54, 9981.
11. Dhanak, D.; Burton, G.; Christmann, L. T.; Darcy, M. G.;
Elrod, K. C.; Kaura, A.; Keenan, R. M.; Link, J. O.; Peishoff,
C. E.; Shah, D. H. Bioorg. Med. Chem. Lett. 2000, 10, 2279.
12. Koch, U.; Biasiol, G.; Brunetti, M.; Fattori, D.; Pallaora,
M.; Steinkuhler, C. Biochemistry 2001, 40, 631.
13. (a) Ingallinella, P.; Altamura, S.; Bianchi, E.; Taliani, M.;
Ingenito, R.; Cortese, R.; De Francesco, R.; Steinkuhler, C.;
Pessi, A. Biochemistry 1998, 37, 8906. (b) Llinas-Brunet, M.;