Evolution, synthesis and SAR of tripeptide α-ketoacid inhibitors of the hepatitis C virus NS3/NS4A serine protease

Article abstract of DOI:10.1016/S0960-894X(01)00843-5

N-Terminal truncation of the hexapeptide ketoacid 1 gave rise to potent tripeptide inhibitors of the hepatitis C virus NS3 protease/NS4A cofactor complex. Optimization of these tripeptides led to ketoacid 30 with an IC₅₀ of 0.38 μM. The SAR of these tripeptides is discussed in the light of the recently published crystal structures of a ternary tripetide/NS3/NS4A complexes.

Full text of DOI:10.1016/S0960-894X(01)00843-5

Bioorganic & Medicinal Chemistry Letters 12 (2002) 705–708
Evolution, Synthesis and SAR of Tripeptide ꢀ-Ketoacid Inhibitors
of the Hepatitis C Virus NS3/NS4A Serine Protease
Stefania Colarusso, Benjamin Gerlach, Uwe Koch, Ester Muraglia, Immacolata Conte,
Ian Stansfield, Victor G. Matassa and Frank Narjes*
Department of Chemistry, IRBM, MRL Rome, Via Pontina Km 30.600, Pomezia, 00040 Rome, Italy
Received 3 October 2001; revised 4 December 2001; accepted 12 December 2001
Abstract—N-Terminal truncation of the hexapeptide ketoacid 1 gave rise to potent tripeptide inhibitors of the hepatitis C virus NS3
protease/NS4A cofactor complex. Optimization of these tripeptides led to ketoacid 30 with an IC50 of 0.38 mM. The SAR of these
tripeptides is discussed in the light of the recently published crystal structures of a ternary tripetide/NS3/NS4A complexes. # 2002
Elsevier Science Ltd. All rights reserved.
In the preceding paper, we described the design and
successful implementation of 2-amino-4,4-difluoro-
butyric acid (difluoroAbu) as a substitute for the cano-
nical cysteine found in substrates and product inhibitors
of the hepatitis C virus (HCV) NS3/NS4A serine pro-
tease.1 This protease is essential for the viral lifecycle
and therefore presents a prime target for the develop-
ment of antiviral drugs.2,3 Avoiding the reactive thiol
allowed the preparation of potent hexapeptide inhibi-
tors 1–3 of this enzyme, from which ketoacid 1 emerged
as an exceptionally potent inhibitor.4 Since polypep-
tides, especially those containing multiple carboxylates,
are unlikely to become orally bioavailable, cell-pene-
trant drugs, our next goal was to reduce the size and
anionic nature of the hexapeptides.
Modeling these inhibitors into the active site of the
enzyme guided the SAR, which led ultimately to the
discovery of potent tripeptide keto acids, containing
only one carboxylic acid. The findings will be discussed
in conjunction with the crystal structures obtained from
two tripeptides as ternary adducts with the protease and
its NS4A co-factor.5
Synthesis
The synthesis of ketoacids has been described in part,4
and examples are shown in Scheme 1. Condensation of
31, readily obtained from the parent amino acid,1 with
(cyanomethylene)phosphorane by modifying the pro-
cedure by Wasserman and co-workers,6 gave, after
removal of the Cbz group, amine 32 in high overall
yield. Coupling to CbzLeuOH and deprotection then
furnished 33, which was coupled to Boc-cyclopentyl
glycine. Ozonolysis of this intermediate in dichloro-
methane in the presence of methanol led to the crude
ketoester, which was hydrolyzed to ketoacid 30.
Coupling amine 32 to dipeptide CbzIle-LeuOH fol-
lowed by deprotection gave tripeptide phosphorane 34,
which was used for investigating the capping group
SAR. For example, 34 was alkoxycarbonylated with the
mixed succinimide carbonate obtained from methyl
4-(hydroxymethyl)-benzoate and N,N0-disuccinimidyl
carbonate.7 Ozonolysis and quenching the intermediate
cyano ketone in water gave keto acid 17. Hydrolysis of
the N-terminal ester then led to 18.
In this letter, we describe the structure–activity studies
leading from hexapeptide 1 to tripeptide inhibitors.
*Corresponding author. Tel.: +39-06-910-93-625; fax: +39-06-910-
93-654; e-mail: frank_narjes@merck.com
0960-894X/02/$ - see front matter # 2002 Elsevier Science Ltd. All rights reserved.
PII: S0960-894X(01)00843-5

706
S. Colarusso et al. / Bioorg. Med. Chem. Lett. 12 (2002) 705–708
Scheme 1. (a) Ph3P¼CHCN, EDCI, HOBt, DCM, 82%; (b) NH4CO2, MeOH, Pd/C, 80%; (c) CbzLeuOH, EDCI, HOBt, DCM, 88%; NH4CO2,
MeOH, Pd/C, 78%; (d) Boc–c-pentylglycine, EDCI, HOBt, DCM; (e) O3, DCM, MeOH ꢀ78 ꢁC; 68%; (f) MeOH, 1 N NaOH; RP-HPLC; (g)
CbzIleLeuOH, EDCI, HOBt, DCM, 80%; NH4CO2, MeOH, Pd/C, 87%; (h) methyl 4-(hydroxymethyl)benzoate, DSC, NEt3, MeCN; evaporate,
add 34, DCM, 55%; (i) O3, DCM,ꢀ78 ꢁC; THF, H2O; RP-HPLC, 21%; (j) 1 N NaOH, MeOH; RP-HPLC, 56%.
Table 1. Truncation of inhibitor 1
Compd
IC50 (mM)
4
5.3
5
>50
6
7
8
>50
1.4
Scheme 2. (a) Ph3P¼CHCN, EDCI, DMAP DCM, 54%; (b) O3,
DCM, MeOH ꢀ78 ꢁC; NaBH4, MeOH, 62%; (c) HCl/EtOAc;
CbzLeuOH, EDCI, HOBt, i-Pr2NEt DCM, 76%; (d) NaOH, MeOH,
quant; (e) IBCF, NMM, THF; CH2N2, 79%; (f) AgBz, NEt3, MeOH,
78%; (g) HCl/EtOAc; CbzLeuOH, EDCI, HOBt, i-Pr2NEt DCM,
92%; (h) NaOH, MeOH, 27%; (i) PMB-tetrazole, BuLi, THF,
TMEDA, ꢀ98 ꢁC, 30%; (j) NaBH4, EtOH, 88%; (k) HCl/EtOAc;
CbzIleLeuOH, HATU, 2,6-lutidine, DCM, 92%; (l) CAN, MeCN,
H2O; DMP, t-BuOH, MeCN; RP-HPLC, 15%.
1.7
Results and Discussion
Data for inhibition of the NS3/NS4A complex were
obtained after 30-min incubation10 with the inhibitor.
All crude ketoacids were purified by HPLC with
separation of the diastereomers resulting from the use of
racemic difluoroAbu. The more active diastereomer
eluted first and was later shown by X-ray crystal-
lography to have the l-configuration at P1.5 The other
diastereomer was usually one order of magnitude less
active (results not shown).
Truncation of the P5 and P6 acid anchor, which is
usually crucial for activity in the hexapeptide product-
based inhibitors, was uniquely successful in the keto
acid series, giving rise to tetrapeptide ketoacid 4 (IC50
5.3 mM). The corresponding aldehyde 5 and product
acid 6 were essentially inactive (Table 1).
The building block for aldehyde 12 was synthesized as
described.1 Hydroxyester 10 was prepared as shown in
Scheme 2. Homologation at 35 using diazomethane8 led
to acid 11. Keto tetrazole 13 was prepared by reacting
amide 39 with the anion of the PMB-protected tetra-
zole.9 Reduction and deprotection led to hydroxy amine
40, which after coupling, removal of the PMB group
and oxidation of the alcohol, furnished 13.
Encouraged by this result, further truncation was
attempted. Initially the Cbz-protected compounds 7 and
8 were made, with the intention to partially mimic the
P4 residue with the Cbz-capping group. They proved to
be more active than tetrapeptide 4, probably due to a
better contact of the phenyl ring in the S4-site. More
importantly these inhibitors contain only
a single

S. Colarusso et al. / Bioorg. Med. Chem. Lett. 12 (2002) 705–708
707
carboxylate and have reduced molecular weight. As with
ketoacid 1, the potency of these tripeptides is due to the
slow dissociation rate of the covalent adduct, whereas the
affinity of product inhibitors 2, and also aldehyde 3 is
determined by high electrostatically driven association
rates.11
Also, ketotetrazole 13, envisioned as a stable and phar-
macologically more acceptable replacement of the
ketoacid, was inactive.
These observations can be explained on the basis of the
properties of the active site. In the X-ray crystal struc-
ture of NS3/4A bound with inhibitor 8, the expected
covalent bond between the catalytic serine (S139) and
the ketone is formed. However, the configuration at the
ketal carbon is different from that usually observed,
where the oxyanion is bound in the oxyanion hole.5
Here this cavity, formed by the backbone of Ala137,
Ser138 and Ser139, is occupied by the carboxylate. The
conserved Lys136 proximal to the active site favors this
interaction (Fig. 1),5 which may explain the specific need
for a ketoacid at this position. Due to the limited size of
the oxyanion hole and its rigidity, the tetrazole moiety
can not be accommodated there in this orientation.
Compound 8 was then used as a starting point for
optimization of the tripeptide series.
Attempts to modify or replace the ketoacid moiety were
unsuccessful (Table 2). Ketoester 9, hydroxy-acid 10,
homologated acid 11 and aldehyde 12 were inactive,
pointing to a crucial role played by the carbonyl group
and the acid in the ketoacid moiety.
Table 2. C- and N-terminal SAR of tripeptide ketoacids
The SAR in the capping group was investigated only
briefly (Table 2). Conversion of the benzyl carbamate to
the benzyl urea gave a less active compound. The Cbz
group could be replaced by the Boc group, without loss
of potency, as in 15. That the lipophilic part of the cap-
ping group makes a significant contribution to binding
was shown with acetyl amide 16, which is 16 times less
active than 15. In the X-ray crystal structure the capping
group interacts with Val158 (Fig. 1), thus explaining the
preference for hydrophobic residues. Modeling suggested
that an acid in the para-position of the Cbz ring could
exploit the same electrostatic interactions of the P6-Glu
with Arg123. Consistent with this hypothesis, acid 18
showed an improvement in potency (IC50 0.44 mM),
while methyl ester 17 retained the same potency as 8.
No.
R1
R2
IC50 (mM)
1.7
8
Cbz
9
Cbz
Cbz
Cbz
Cbz
Cbz
>50
>50a
>50
>50b
>50c
3.3
10
11
12
13
14
15
16
17
18
Lipophilic contact with the enzyme is also essential in
the P2-position, as replacement of Cha or Leu in 7 and 8
shows (Table 3). The glycine compound 19 is devoid of
activity, probably also for conformational reasons.
Going from Ala (20, IC50 16 mM) to the more hydro-
phobic Abu (21, IC50 3.0 mM) and then to 7 and 8, it
Boc
Ac
1.0
16
1.7
0.44
aMixture of four diastereomers.
bCompound contains Cha in P2.
c2:1 mixture of diastereomers.
Figure 1. Inhibitor 8 bound in the active site of the NS3/NS4A complex.

708
S. Colarusso et al. / Bioorg. Med. Chem. Lett. 12 (2002) 705–708
Table 3. P2 and P3 SAR of tripeptide ketoacids
To optimize the hydrophobic interactions in the S3 site,
a variety of natural and non-natural hydrophobic
amino acids were screened. Cyclopentylglycine emerged
as the optimal substituent in this position (30, IC50 0.38
mM), being nearly equipotent to the glutamic acid con-
taining compound 24.
No.
X
P3
P2
IC50 (mM)
In summary, we have described the evolution of tripep-
tide ketoacid inhibitors from a polyanionic hexapeptide,
showing that the P5 and P6 acids are not needed for
activity. Also the P4 residue could be eliminated,
although the capping group fulfills part of the role of
this group. The structural data collected helped in
understanding the observed SAR. We believe that these
results may provide a basis for the difficult task of find-
ing more potent and less peptidic inhibitors of NS3/
NS4A protease.
19
20
21
22
23
24
25
26
27
28
29
30
Cbz
Cbz
Cbz
Cbz
Cbz
Boc
Boc
Boc
Boc
Boc
Boc
Boc
Ile
Ile
Ile
Ile
Gly
Ala
Abu
Phe
>100
15.6
3.28
9.4
Ile
Val
16.0
0.30
0.33
0.46
2.10
1.63
8.90
0.38
Glu
Glu
Val
Gln
Asp
Ala
DifluoroAbu
Leu
Leu
Leu
Leu
Leu
Leu
c-Pentylgly
becomes clear that activity is related to the hydro-
phobicity of the side chain. Alkyl is preferred over
phenyl (cf., 7 to 22, IC50 9.4 mM), and b-branched
hydrophobic amino acids such as Val (23, IC50 16 mM)
are excluded by the architecture of this site. These
observations are consistent with earlier SAR in the
product inhibitor hexapeptide series.12
Acknowledgements
The authors thank Mirko Brunetti, Mauro Cerretani,
Sergio Serafini, Sergio Altamura and Christian Stein-
kuhler for providing the assay data, the analytical
chemistry department for assistance, and Janet Clench
and Michael Rowley for valuable discussions.
Recalling that the NS5A/5B substrate for NS3 has
cysteine in P1 and P2, substitution with our cysteine
mimetic was also tried in this position. Inhibitor 24
proved to be equipotent to 25. The difluoromethyl
group probably acts here as a purely hydrophobic
group, although electrostatic components may have a
role in binding.13,14
References and Notes
1. Narjes, F.; Koehler, K. F.; Koch, U.; Gerlach, B.;
Colarusso, S.; Steinkuhler, C.; Brunetti, M.; Altamura, S.;
De Francesco, R.; Matassa, V. G. Bioorg. Med. Chem. Lett.
2002, 12, 701.
2. Kolykhalov, A. A.; Mihalik, K.; Feinstone, S. M.; Rice,
C. M. J. Virol. 2000, 74, 2046.
In the crystal structure (Fig. 1), the lipophilic P2 group
covers the catalytic ion pair His57-Asp81, thus stabiliz-
ing it in two ways: (1) by reducing its conformational
flexibility, and (2) by shielding it from solvent. We pre-
sume that this may influence active site binding through
subtle changes in the pKa of the amino acids involved.15
3. Perni, R. B. Drug News Perspect. 2000, 13, 69.
4. Narjes, F.; Brunetti, M.; Colarusso, S.; Gerlach, B.; Koch,
U.; Biasiol, G.; Fattori, D.; De Francesco, R.; Matassa, V. G.;
Steinkuhler, C. Biochemistry 2000, 39, 1849.
5. Di Marco, S.; Rizzi, M.; Volpari, C.; Walsh, M. A.; Narjes,
F.; Colarusso, S.; De Francesco, R.; Matassa, V. G.; Sollazzo,
M. J. Biol. Chem. 2000, 275, 7152.
6. Wasserman, H. H.; Ho, W.-B. J. Org. Chem. 1994, 59, 4364.
7. Ghosh, A. K.; Duong, T. T.; McKee, S. P.; Thompson, W.
J. Tetrahedron Lett. 1992, 33, 2781.
8. Podlech, J.; Seebach, D. Angew. Chem., Int. Ed. Engl. 1995,
34, 471.
9. Satoh, Y.; Marcopolus, N. Tetrahedron Lett. 1995, 36, 1759.
10. Urbani, A.; Bianchi, E.; Narjes, F.; Tramontano, A.; De
Francesco, R.; Steinkuhler, C.; Pessi, A. J. Biol. Chem. 1997,
272, 9204.
11. Koch, U.; Biasiol, G.; Brunetti, M.; Fattori, D.; Pallaoro,
M.; Steinkuhler, C. Biochemistry 2001, 40, 631.
12. Ingallinella, P.; Altamura, S.; Bianchi, E.; Taliani, M.;
Ingenito, R.; Cortese, R.; Francesco, R. D.; Steinkuehler, C.;
Pessi, A. Biochemistry 1998, 37, 8906.
Both hydrophobic and hydrophilic groups are accepted
in P3. Replacing isoleucine in 8 with valine or glutamic
acid, both found in substrates of NS3, increased
potency (Table 3). The interactions in the S3 site were
quite specific. Converting glutamic acid into glutamine
as in 27 or aspartic acid as in 28 resulted in a 6- or 10-
fold decrease in potency, respectively. Likewise, repla-
cement of Val with Ala gave a much weaker compound
(29, IC50 8.9 mM).
Modeling studies, confirmed by the X-ray structure,
showed that the lipophilic groups interact with Cys159
and Val132, and the carboxylate groups can have elec-
trostatic interactions with Lys136 (Fig. 1). The longer
side-chain of Glu versus Asp allows the carboxylate to
be in close vicinity to Lys136 and the extra methylene
augments the hydrophobic contact.
13. Dunitz, J. D.; Taylor, R. Chem. Eur. J. 1997, 3, 89.
14. Erickson, J. A.; McLoughlin, J. J. Org. Chem. 1995, 60, 1626.
15. Barbato, G.; Cicero, D. O.; Cordier, F.; Narjes, F.; Ger-
lach, B.; Sambucini, S.; Grzesiek, S.; Matassa, V. G.; De
Francesco, R.; Bazzo, R. EMBO J. 2000, 19, 1195.