Highly potent non-peptidic inhibitors of the HCV NS3/NS4A serine protease

Article abstract of DOI:10.1016/S0960-894X(02)00680-7

Screening of a diverse set of bisbenzimidazoles for inhibition of the hepatitis C virus (HCV) serine protease NS3/NS4A led to the identification of a potent Zn²⁺-dependent inhibitor (1). Optimization of this screening hit afforded a 10-fold more potent inhibitor (46) under Zn²⁺ conditions (K_i=27 nM). This compound (46) binds also to NS3/NS4A in a Zn²⁺ independent fashion (K_i=1 μM). The SAR of this class of compounds under Zn²⁺ conditions is highly divergent compared to the SAR in the absence of Zn²⁺, suggesting two distinct binding modes.

Full text of DOI:10.1016/S0960-894X(02)00680-7

Bioorganic & Medicinal Chemistry Letters 12 (2002) 3129–3133
Highly Potent Non-peptidic Inhibitors of the HCV NS3/NS4A
Serine Protease
David Sperandio,a,* Anthony R. Gangloff,a Joane Litvak,a Richard Goldsmith,a,y
Jason M. Hataye,a Vivian R. Wang,a Emma J. Shelton,a Kyle Elrod,a James W. Janc,a
James M. Clark,a Ken Rice,a,{ Steve Weinheimer,b Kap-Sun Yeung,b
Nicholas A. Meanwell,b Dennis Hernandez,b Andrew J. Staab,b Brian L. Venablesb
and Jeffrey R. Spencera
aCelera, 180 Kimball Way, South San Francisco, CA 94080, USA
bBristol-Myers Squibb Pharmaceutical Research Institute, 5 Research Parkway, PO Box 5100, Wallingford, CT 06492, USA
Received 29 May 2002; accepted 29 July 2002
Abstract—Screening of a diverse set of bisbenzimidazoles for inhibition of the hepatitis C virus (HCV) serine protease NS3/NS4A
led to the identification of a potent Zn2+-dependent inhibitor (1). Optimization of this screening hit afforded a 10-fold more potent
inhibitor (46) under Zn2+ conditions (Ki=27 nM). This compound (46) binds also to NS3/NS4A in a Zn2+ independent fashion
(Ki=1 mM). The SAR of this class of compounds under Zn2+ conditions is highly divergent compared to the SAR in the absence of
Zn2+, suggesting two distinct binding modes.
# 2002 Elsevier Science Ltd. All rights reserved.
The prevalence of chronic Hepatitis C virus (HCV)
infections in the US was recently determined to be
around 1.8% (2.7 million people), with about 15,000
new cases every year.1 Worldwide, HCVis estimated to
infect 170 million people.1 Over 85% of all cases
become chronic.2,3 Current therapies of HCVinfections
with interferons show only modest results and have ser-
ious side effects such as depression, fever, and fatigue.
P7-NS2-NS3-NS4A-NS4B-NS5A-NS5B), which is then
processed into the active viral proteins. Host cell proteases
are responsible for cleavages in the C-E1-E2-P7-NS2
region.7 The NS2-NS3 cleavage is performed by NS2,
most likely a Zn2+ containing metalloprotease.8 The
remaining processings of the NS3-NS4A-NS4B-NS5A-
NS5B fragment are all dependent on the NS3/NS4A
protease, which makes this chymotrypsin-like serine
protease a promising target.7 Highly potent substrate-
based peptidic inhibitors for the HCVNS3/NS4A pro-
tease were developed (Ac-Asp-d-Gla-Leu-Ile-Cha-Cys-
OH, Ki=1.5 nM).9 Non peptidic small molecule inhibi-
tors for NS3/NS4A were published by Rational Drug
Design Laboratories10 and Schering-Plough11 with
activities in the low micromolar range.
Recently, combination treatment with ribavarin,4
a
broad spectrum antiviral and pegylated interferon5 has
proven to be superior to monotherapy. Despite these
advances, a broadly effective antiviral therapy for HCV
is still elusive. An excellent review by Dymock
summarizes emerging therapies for HCV.6
Significant research efforts are currently directed
towards targeting viral enzymes. The HCVRNA gen-
ome is translated into a 9.5 kb polyprotein (C-E1-E2-
We recently published a new class of highly potent,
reversible and selective serine protease inhibitors.12 The
binding of these new inhibitors, which are derived from
a bis-benzimidazolemethane fragment, is dependent on
the presence of Zn2+. X-ray structure determination
revealed a Zn2+ ion at the catalytic site, which is coor-
dinated by His 57 and Ser 195 of the enzyme, and by
two imidazole-nitrogens of the bidentate ligand bis-
benzimidazole.12 We explored this binding motif for the
*Corresponding author. Tel.: +1-650-866-6544; fax +1-650-866-
6655; e-mail: david.sperandio@celera.com
yPresent address: Genentech, MS 18, 1 DNA Way, South San Fran-
cisco, CA 94080, USA
{Present address: Exelixis, 170 Harbor Way, South San Francisco, CA
94083, USA
0960-894X/02/$ - see front matter # 2002 Elsevier Science Ltd. All rights reserved.
PII: S0960-894X(02)00680-7

3130
D. Sperandio et al. / Bioorg. Med. Chem. Lett. 12 (2002) 3129–3133
inhibition of the HCVNS3/NS4A protease and
screened a library of a diverse set of bis-benzimidazoles.
Indeed, bis-benzimidazole 1 [CRA-6336 (APC-6336),
Fig. 1] showed an activity of 0.20 mM in the presence of
Zn2+ (12.5 mM).13 Sequestering of Zn2+ with excess
ethylenediamintetraacetate (EDTA, 25 mM) induced a
drop in binding affinity of about 800-fold. In contrast, the
presence of Zn2+ increased the affinity of peptidic inhib-
itors for HCVNS3/NS4A only by about 2-fold. Kinetic
studies showed that 1 is a competitive, active site directed
inhibitor of NS3/NS4A (data not shown). Screening hit 1
is over 100-fold selective for HCVNS3/NS4A compared
to the serine proteases chymotrypsin, tryspsin and elas-
tase. We have systematically explored the structure–
activity relationship (SAR) of screening hit 1 and were
able to identify bis-benzimidazoles with improved bind-
binding. The phosphonoalanine side chain SAR of 1 was
explored further in detail and resulted in the identification
of a novel series of active compounds under Zn2+ condi-
tions in the low micromolar range.14
The observation that the binding affinity of 1 shifts
three orders of magnitude in the presence or absence of
Zn2+, suggests a Zn2+ binding motif as reported.12,15
X-ray structure determination revealed this type of
binding motif in similar inhibitor–enzyme complexes of
trypsin and thrombin. Gradual deletion of the bisben-
zimidazole moiety in these inhibitor–enzyme complexes
resulted in a gradual loss of binding to thrombin or
trypsin. We have seen similar SAR trends in 1 (see com-
pounds 7–9, Fig. 3), which also support a binding mode
as reported.12 However, in the absence of structural
information, other zinc-mediated binding modes may
also be available to these inhibitors. The SAR shown in
Figures 1–3 suggests that the key binding element of 1
involves a Zn2+-mediated chelation of the benzimida-
zole scaffold and an electrostatic interaction of the acidic
phosphonalanine residues with NS3/NS4A. Deletion of
either binding element results in complete loss of activity.
ing affinity in the presence and absence of Zn2+
.
Studies to determine the key binding elements in
screening hit 1 are summarized in Figures 1–3. Masking
of one negative charge of the phosphonic acid moiety [–
PO3H2!–PO2(OMe)OH] results in a loss of binding
affinity in the presence of Zn2+ of almost one order of
magnitude. Surprisingly, the binding affinity in the pre-
sence of EDTA improved more than 4 times. Complete
deletion of the phosphonic acid as in the case of glycine
3 abolishes most of the activity under Zn2+ conditions
and all activity under EDTA conditions. Enantiopure 1
derived from either (R)- or (S)-phosphonoalanine did
not show a significant difference in binding in the
presence or absence of Zn2+ (data not shown).
To further improve the activity, we prepared a library of
analogues of 1 which have different electronic properties
and steric demand at the left-hand benzimidazole moi-
ety. The above-mentioned key binding elements were
retained (see Table 1). Overall, the SAR under Zn2+
conditions is unresponsive. Compared to parent com-
pound 1 (see also 35), electron withdrawing substituents
(10–18) and electron donating substituents (19–23) did
not produce more active leads. A methyl group seems
not to be tolerated in the R4 position (20). Similarly, in
the bridgehead monofluoro series (24–26), a relatively
flat SAR under Zn2+ conditions was observed. Bridge-
head difluorination (27–33) appears to reduce the activ-
ity under Zn2+ as well as under EDTA conditions.
Bridgehead difluoro-bisbenzimidazole 27 had an excep-
tionally high activity of 1.1 mM under EDTA condi-
tions. A survey of the N-alkyl substituent of the right-
hand benzimidazole also revealed a flat SAR (35–40).
The orientation of the phosphonoalanine side chain rela-
tive to the bis-benzimidazole scaffold appears to be crucial
for binding (see Fig. 2). Moving the phosphonoalanine
side chain from the 6-position in benzimidazole 5, to the
5-position (4), or to the 7 position (6) abolishes most of the
Studies by Pessi et al.9 of peptide-based inhibitors of
HCVNS3/NS4A revealed that charge–charge inter-
action at P6 and P5 in conjunction with a carboxylic
acid at the C-terminus is crucial for binding. We won-
dered if this binding interaction could be translated to
the bis-benzimidazole scaffold. Indeed, installation of an
additional phosphonoalanine side chain at the left hand
side resulted in a significant increase in binding both
under Zn2+ and EDTA conditions (41, see Fig. 4).
Figure 1. Screening hit 1 and analogues 2 and 3. In parentheses, Ki
2+
(mM) for inhibition of HCVNS3/NS4A in the presence of Zn
/
EDTA.13
Figure 2. Phosphonoalanine analogues of 5. In parentheses, Ki (mM)
Figure 3. Left-hand deletion SAR. In parentheses, Ki (mM) for inhibition
for inhibition of HCVNS3/NS4A in the presence of Zn 2+/EDTA.13
of HCVNS3/NS4A in the presence of Zn 2+/EDTA.13

D. Sperandio et al. / Bioorg. Med. Chem. Lett. 12 (2002) 3129–3133
3131
Table 1. SAR of the bis-benzimidazole scaffold. Ki (mM) for inhibition of HCVNS3/NS4A in the presence of Zn 2+ or EDTA13
Compd
R
R1
R2
R3
R4
R5
Ki (Zn2+
)
Ki (EDTA)
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
–CH2–
–CH2–
–CH2–
–CH2–
–CH2–
–CH2–
–CH2–
–CH2–
–CH2–
–CH2–
–CH2–
–CH2–
–CH2–
–CH2–
–CHF–
–CHF–
–CHF–
–CF2–
–CF2–
–CF2–
–CF2–
–CF2–
–CF2–
–CF2–
–CF2–
–CH2–
–CH2–
–CH2–
–CH2–
–CH2–
–CH2–
H
H
H
H
H
H
F
H
H
H
F
F
H
F
F
Cl
CF3
F
H
F
F
H
F
Me
H
OH
OMe
H
H
H
H
H
H
F
Cl
F
H
H
H
H
H
H
H
H
H
H
H
F
F
H
F
CH3
CH3
CH3
CH3
CH3
CH3
CH3
CH3
CH3
CH3
CH3
CH3
CH3
CH3
CH3
CH3
CH3
CH3
CH3
CH3
CH3
CH3
CH3
CH3
0.28
0.38
0.41
0.21
0.24
0.27
0.2
0.35
0.28
0.24
10
95
130
70
25
58
140
21
22
17
F
F
F
F
F
H
H
H
H
H
H
F
H
H
H
H
H
H
H
F
H
H
H
H
H
H
H
H
H
H
H
H
H
F
H
Me
H
H
OMe
H
F
98
>300
250
187
24
0.19
0.2
0.49
0.47
0.55
0.66
0.25
1.2
1.3
2.4
4
1.4
37
72
91
1.1
74
55
38
15
25
12
F
F
H
CONH2
H
H
F
H
H
H
H
H
F
F
F
F
2.9
H
H
H
H
H
H
H
Me
H
H
H
H
H
H
CH3
CH3
21
0.45
12
0.14
0.37
0.61
0.87
66
>300
>300
140
>300
>300
>300
i-Pr
cyclo-PrCH2
i-Bu
neo-Pentyl
i-Pentyl
H
H
Bridgehead difluorination (42) improves the activity to
73 nM under Zn2+ conditions and 15 mM under EDTA
conditions. While masking of the negative charge in the
carboxylic acid moieties as in 43 did not have a significant
effect, the introduction of benzyl esters as in 44 com-
pletely impairs binding to NS3/NS4A. This observation
is in line with the low activities of 2 and 3, which lack the
negative charges in the phosphonic acid moieties.
the binding to NS3/NS4A only under Zn2+ conditions.
However, in conjunction with a third phosphonoalanine
moiety at the right-hand position as in compound (46),
the binding to NS3/NS4A improved to 27 nM under
Zn2+ conditions and 1 mM under EDTA conditions.
While trends in the SAR under EDTA conditions lead-
ing to 46 and trends in the SAR of peptidic inhibitors
are similar, we do not have evidence for similar binding
modes of 46 and peptidic inhibitors.
Further elaboration of the right-hand phosphonoala-
nine side chain with another phosphonoalanine as in
(45) (see Fig. 5) to mimic the P5/P6 interaction improved
The synthesis of 46 is summarized below: Esterification
of N-Boc serine followed by conversion to the dehy-
droalanine and reaction with P(OH)(OBn)2 in the
presence of DBU gave orthogonally protected
phosphonoalanine (Scheme 1). Ester cleavage afforded
Figure 5. Bis- and tris-phosphonoalanine analogues of 27 (mixtures of
diastereoisomers). In parentheses, Ki (mM) for inhibition of HCVNS3/
NS4A in the presence of Zn2+ or EDTA.13
Figure 4. Bis-phosphonate analogues of 1. In parentheses, Ki (mM) for
inhibition of HCVNS3/NS4A in the presence of Zn 2+/ EDTA.13

3132
D. Sperandio et al. / Bioorg. Med. Chem. Lett. 12 (2002) 3129–3133
In summary, the binding affinity of screening hit 1
under Zn2+ conditions was improved by the introduc-
tion of a phosphonoalanine side chain on the left-hand
side in conjunction with elaboration of the right-hand
phosphonoalanine to a bisphosphonate 46. It was found
that the specific display of negative charges presented by
the phosphonic acid moieties is responsible for most of
the binding in the side chain. Attempts to replace the
phosphonic acid moieties by either less-charged groups
or by groups which can be masked by a prodrug
approach were reported14 and resulted in only moder-
ately active compounds. Sequestering of Zn2+ with
EDTA results in a drop of activity in 46 from 27 nM to
1 mM. Divergent trends in the SAR were found under
EDTA conditions compared to Zn2+ conditions. This
implies two different binding modes. Both binding
modes depend heavily on the polyanionic character of
this class of inhibitors, which is reminiscent of the SAR
of substrate based peptidic inhibitors of NS3/
NS4A.9,16À18
Scheme 1. Synthesis of 47 and 48: (a) TMSCH2N2, MeOH/benzene,
95%; (b) PPh3, DEAD, toluene, rt, 85%; (c) P(OH)(OBn)2, DBU,
neat, 65%; (d) LiOH, THF, H2O, 80%; (e) TFA, CH2Cl2, 68%;
(f) PyBop, DMF, NMM, 93%; (g) TFA, CH2Cl2, 95%.
Acknowledgements
We thank Dr. Michael Venuti and Dr. Jochen Knolle
for interesting discussions. We also thank Dr. Liling
Fang for analytical support and Mr. Mitchell Lee for
technical assistance.
Scheme 2. Synthesis of 49: (a) 40% MeNH2 in H2O, 100 ꢀC, 82%;
(b) (t-BuO)2CHNMe2, toluene, 80 ꢀC, 78%; (c) H2, Pd/C, MeOH/
HOAc, 87%; (d) MeO(CN)CH2CO2Me, EtOH, 80 ꢀC, 77%; (e) neat,
150 ꢀC, 63%; (f) (PhSO2)2NF, MeOH, 70 ꢀC, 80%.
References and Notes
1. Cohen, C. Science 1999, 285, 26.
2. Cuthbert, J. Clin. Microbiol. Rev. 1994, 7, 505.
3. El-Serag, H.; Mason, A. Arch. Int. Med. 2000, 160, 3227.
4. Mutchnick, M. G.; Lindsay, K. L.; Schiff, E. R.; Cum-
mings, G. D.; Appelman, H. D.; Peleman, R. R.; Silva, M.;
Roach, K. C.; Simmons, F.; Milstein, S.; Gordon, S. C.;
Ehrinpreis, M. N. N. Engl. J. Med. 1998, 339, 1493.
5. Heathcote, E. J.; Shiffman, M. L.; Cooksley, W. G.; Dush-
eiko, G. M.; Lee, S. S.; Balart, L.; Reindollar, R.; Reddy,
R. K.; Wright, T. L.; Lin, A.; Hoffman, J.; De Pamphilis, J. N.
Engl. J. Med. 2000, 343, 1673.
6. Dymock, B. Emerg. Drugs 2001, 6, 13.
7. Bartenschlager, R.; Lohmann, V. J. Gen. Virol. 2000, 81,
1631.
8. Bieroni, L.; Saltonini, E.; Fipaldini, C. J. Virol. 1997, 71,
4665.
Scheme 3. Synthesis of 46: (a) LiOH, MeOH/H2O, 90 ꢀC, 60%; (b) 47,
PyBop, DIPEA, DMF, rt, 62%; (c) TFA, rt, 89%; (d) 48, PyBop,
DIPEA, DMF, rt, 79%; (e) LiOH, MeOH/H2O, 61%; (f) Pd/C,
MeOH/HCl, 10%.
the acid 47, which was coupled into the N-deprotected
phosphonoalanine to give 48 after TFA cleavage.
9. Ingallinella, P.; Altamura, S.; Bianchi, E.; Talinani, M.;
Ingenito, R.; Cortese, I.; De Francesco, R.; Steinkuhler, C.;
Pessi, A. Biochemistry 1998, 37, 8906.
Scaffold 49 was prepared via a thermal cyclocondensa-
tion at 150 ꢀC of a phenylendiamine and the right-hand
benzimidazole acetic acid methyl ester (Scheme 2).
Bridgehead fluorination was accomplished by heating
with N-fluoro-bisphenylsulphonimide in MeOH.
10. Kakiuchi, N.; Komoda, Y.; Komoda, K. FEBS Lett.
1998, 421, 217.
11. Chu, M.; Mierzwa, R.; Truumees, I. Tetrahedron Lett.
1996, 37, 7229.
12. Katz, B.; Clark, J.; Finer-Moore, J.; Jenkins, T.; Johnson,
C.; Ross, M.; Luong, C.; Moore, W.; Stroud, R. Nature 1998,
391, 608.
13. Ki0 determinations. HCVprotease inhibition studies were
performed in 50 mM HEPES buffer (pH 7.5), 50% glycerol,
0.005% polyoxyethylenesorbitan monolaurate (Tween-20),
and 10% DMSO. A synthetic 17 amino acid NS4A surrogate
(Ac-KKKGSVVIVGRIILSGR-CONH2) was included in the
assay at 40 mM. HCVprotease (12 nM active sites, recombi-
nant HCVprotease consisted of the first 183 amino acids of
the NS3 coding region) was incubated with inhibitor, present
Methyl ester cleavage of 49 followed by coupling with
47 afforded the left-hand side phosphonoalanine
(Scheme 3). TFA cleavage of the right hand ester and
coupling with 48 gave fully protected 46. Deprotection
to 46 was accomplished by cleavage of the methyl esters,
followed by global benzyl ester cleavage using Pd/C. All
compounds for assay were purified by reverse-phase
HPLC (H2O/MeCN/0.1% TFA).

D. Sperandio et al. / Bioorg. Med. Chem. Lett. 12 (2002) 3129–3133
3133
at varying concentrations, for 30 min at room temperature
(21–24 ꢀC) in 96-well microtiter plates. Typically, the evalua-
tion of a given inhibitor’s potency was determined under two
distinct sets of experimental conditions. In one set of assay
conditions, ZnCl2 was supplied at 12.5 mM. In parallel experi-
ments, the inhibitor was evaluated without Zn(II) supple-
mentation and with 25 mM EDTA included in the standard
assay. After pre-incubation, reactions were initiated with the
addition of the quenched fluorescence substrate, Ac-DED(E-
dans)EE-Abu[COO]-ASK(Dabcyl)-CONH2 (final concentra-
tion was 1.5 mM) (Taliani, M.; Bianchi, E.; Narjes, F.;
Fossatelli, M.; Urbani, A.; Steinkuhler, C.; De Francesco, R.;
Pessi, A. Anal. Biochem, 1996, 240, 60). The hydrolysis of this
substrate was monitored fluorometrically (filter pair: excita-
tion 355 nm, emission 485 nm) using an FMAX Kinetic
Microplate Reader (Molecular Devices, Sunnyvale, CA,
USA). The velocity of the HCVprotease catalyzed reaction
was obtained from the linear portion of the progress curves.
Apparent inhibition constants, Ki0, were calculated from the
velocity data generated at the various inhibitor concentrations
using the software package, Batch Ki (Biokin Ltd., Pullman,
WA, USA) (Kuzmic, P.; Sideris, S.; Cregar, L. M.; Elrod, K.
C.; Rice, K. D.; Janc, J. W. Anal. Biochem. 2000, 281, 62).
Batch Ki provides a parametric method for the determination
of inhibitor potency using a transformation of the tight bind-
ing inhibition model described by Morrison (Morrison, J. F.
Biochem. Biophys. Acta 1969, 185, 269). A decapeptide was
used as a positive control: H2N-E-D-V-V-L-C-Tic-Nle-S-Y-
OH: 298 nM in the presence of Zn2+, 724 nm in the presence
of EDTA, literature value: 340 nM, Landro, J. A.; Raybuck,
S. A.; Luong, J. P. J.; O’Malley, E. T.; Harbeson, S. L.; Mor-
genstern, K. A.; Rao, G.; Livingston, D. J. Biochemistry 1997,
36, 9340.
14. Yeung, K.-S.; Meanwell, N.; Qiu, Z.; Hernandez, D.;
Zhang, S.; McPhee, F.; Weinheimer, S.; Clark, J.; Janc, J.
Bioorg. Med. Chem. Lett. 2001, 11, 2355.
15. Katz, B.; Luong, C. J. Mol. Biol. 1999, 292, 669.
16. Barbato, G.; Cicero, D.; Nardi, M.; Steinkuhler, C.; Cor-
tese, R.; Francesco, R. D. J. Mol. Biol. 1999, 289, 371.
17. Cicero, D.; Barbato, G.; Koch, U.; Ingallinella, P.; Bian-
chi, E.; Nardi, M.; Steinkuhler, C.; Cortese, R.; Matassa, V.;
Francesco, R. D.; Pessi, A.; Bazzo, R. J. Mol. Biol. 1999, 289,
385.
18. LaPlante, S.; Cameront, D.; Aubry, N.; Lefebre, S.;
Kukolj, G.; Maurice, R.; Thibeault, D.; Lamarre, D.; Llinas-
Brunet, M. J. Biol. Chem. 1999, 274, 18618.