Novel agmatine dipeptide inhibitors against the West Nile virus NS2B/NS3 protease: A P3 and N-cap optimization study

Article abstract of DOI:10.1016/j.ejmech.2012.12.043

This communication describes the synthesis and inhibitory activities of thirty-seven novel C-terminal agmatine dipeptides used as screening compounds to study the structure-activity relationship between active-site peptidomimetics and the West Nile virus (WNV) NS2B/NS3 serine protease. Our efforts lead to the discovery of a novel agmatine dipeptide inhibitor (compound 33, IC₅₀ 2.6 ± 0.3 μM) with improved inhibitory activity in comparison to the most potent inhibitor described in our recent report [IC₅₀ 4.7 ± 1.2 μM; Lim et al., Eur. J. Med. Chem. 46 (2011) 3130-3134]. In addition, our study cleared the contention surrounding the previous X-ray co-crystallization study and an enzyme inhibition report on the binding conformation adopted by active-site peptide aldehydes. Our data should provide valuable insights into the design of future peptidomimetic antivirals against WNV infections.

Full text of DOI:10.1016/j.ejmech.2012.12.043

European Journal of Medicinal Chemistry 62 (2013) 199e205
Contents lists available at SciVerse ScienceDirect
European Journal of Medicinal Chemistry
journal homepage: http://www.elsevier.com/locate/ejmech
Short communication
Novel agmatine dipeptide inhibitors against the West Nile virus
NS2B/NS3 protease: A P3 and N-cap optimization study
Huichang Annie Lim, Melgious Jin Yan Ang, Joma Joy, Anders Poulsen, Wenshi Wu,
*
Shi Chie Ching, Jeffrey Hill, Cheng San Brian Chia
Experimental Therapeutics Centre, Agency for Science, Technology and Research (A*STAR), 31 Biopolis Way, Nanos #03-01, Singapore 138669,
Singapore
a r t i c l e i n f o
a b s t r a c t
Article history:
This communication describes the synthesis and inhibitory activities of thirty-seven novel C-terminal
agmatine dipeptides used as screening compounds to study the structureeactivity relationship between
active-site peptidomimetics and the West Nile virus (WNV) NS2B/NS3 serine protease. Our efforts lead to
Received 11 October 2012
Received in revised form
11 December 2012
Accepted 25 December 2012
Available online 3 January 2013
the discovery of a novel agmatine dipeptide inhibitor (compound 33, IC50 2.6 ꢀ 0.3
inhibitory activity in comparison to the most potent inhibitor described in our recent report [IC50
M; Lim et al., Eur. J. Med. Chem. 46 (2011) 3130e3134]. In addition, our study cleared the
mM) with improved
4.7 ꢀ 1.2
m
contention surrounding the previous X-ray co-crystallization study and an enzyme inhibition report on
the binding conformation adopted by active-site peptide aldehydes. Our data should provide valuable
insights into the design of future peptidomimetic antivirals against WNV infections.
Ó 2013 Elsevier Masson SAS. All rights reserved.
Keywords:
Agmatine
Peptide
West Nile virus
NS2B/NS3
Serine protease inhibitor
1. Introduction
precursors into functional proteins responsible for viral replication
[9]. This trypsin-like protease is specific for protein substrates
The West Nile virus (WNV) is a member of the Flaviviridae
family of viruses transmitted by Culex mosquitoes [1]. Originating
from Africa, it has spread to humans in Asia, southern Europe, the
Middle East and North America [2,3]. Symptoms of infection
include fever, headaches, chills, diaphoresis, enlarged lymph nodes
and arthralgia [4]. Approximately 60% of these cases progress to
encephalitis with a fatality rate of 10% [1]. Between 1999 and 2011,
the United States Center for Disease Control reported more than
23,000 human infections in North America, resulting in 1251
deaths [5]. More recently, an outbreak occurred in North America,
with 1118 human infections and 41 fatalities [6]. There is currently
no vaccine or specific drug therapy to treat WNV infections, thus
creating an urgent need to develop a drug against this disease [7].
The WNV has a single-strand, 11-kb RNA genome which acts as
a messenger RNA for protein synthesis and a template for RNA
replication in mammalian host cells. The viral genome encodes
seven nonstructural proteins (NS1, NS2A, NS2B, NS3, NS4A, NS4B
and NS5) [8]. NS3 complexes with its co-factor NS2B to form
a serine protease in the host cell which hydrolyze polyprotein
containing two consecutive basic residues (Lys-Arg) at the P2 and
P1 positions respectively, cleaving the C-terminal end of arginine
(Fig. 1A) [10,11]. This unusual specificity is not found in many
mammalian proteases, making NS2B/NS3 a potential drug target
[12e14].
In 2008, a cell permeable, serum stable, non-cytotoxic tripeptide
aldehyde inhibitor (Lys-Lys-Arg-H) was demonstrated to com-
petitively inhibit the WNV NS2B/NS3 protease at nanomolar po-
tency, bringing the field a step closer to the development of the first
WNV drug [15]. A year later, the X-ray crystal structure of the WNV
NS2B/NS3 protease complexed to the tripeptide aldehyde inhibitor,
2-naphthoyl-Lys-Lys-Arg-H, was solved [16]. The naphthoyl N-ter-
minal capping group (N-cap) was observed to be solvent-exposed
and did not interact with the protease (Fig. 1B). Intriguingly,
modifications to the N-cap drastically affected inhibitory activities,
with up to 30-fold differences in IC50 values [15]. The authors
attributed these conflicting observations to artifacts of crystal-
lization and using molecular dynamics and simulated annealing
calculations, proposed alternative ligand binding poses with the
N-cap binding to the protease S4 subsite [15].
Recently, we showed that agmatine dipeptide 1 (Fig. 2) could
competitively inhibit the WNV NS2B/NS3 protease with an IC50
* Corresponding author. Tel.: þ65 64070348; fax: þ65 64788768.
E-mail address: cschia@etc.a-star.edu.sg (C.S.B. Chia).
of 4.7 ꢀ 1.2
mM [17]. Peptides containing a P1 agmatine at their
0223-5234/$ e see front matter Ó 2013 Elsevier Masson SAS. All rights reserved.
http://dx.doi.org/10.1016/j.ejmech.2012.12.043

200
H.A. Lim et al. / European Journal of Medicinal Chemistry 62 (2013) 199e205
Fig. 1. (A) Schematic diagram of a peptide substrate showing the scissile bond (arrowed) and P1eP3 residues using conventional terminology [11]. Protease residues involved in
substrate binding are numbered based on the crystal structure 3E90.pdb. (B) Co-crystal structure (3E90.pdb) of naphthoyl-KKR-H (stick model) bound to the WNV NS2B/NS3
protease.
C-termini have been shown to inhibit various serine proteases,
including thrombin and furin [18,19]. These peptidomimetics
resemble peptide aldehydes without the reactive C-terminal alde-
hyde functionality. Although less potent, they do not cyclize in so-
lution and can be easily and rapidly synthesized using solid-phase
chemistry [20], making them ideal tool compounds for studying
enzymeeinhibitor interactions. To gain better insights into the
conflicting X-ray and enzyme inhibition data, we conducted
a structureeactivity relationship (SAR) study using thirty-seven
novel agmatine dipeptide NS2B/NS3 inhibitors with various P3
and N-cap analogs. We believe our findings will be invaluable to
medicinal chemists involved in the design and development of
novel WNV protease inhibitors.
the guanidino Nε involved in intra-molecular hydrogen bonding to
the carbonyl O of the 4-phenyl-phenacetyl (4-PhPhac) N-cap
(Fig. 3A). To probe the binding role of the P3 residue, we synthe-
sized compound 2 (Table 1) and discovered that a single residue
truncation abrogated its inhibitory activity (IC50 > 100
gesting that the P3 Lys played an important enzyme inhibitory role.
Changing the stereochemistry of P3 Lys to its enantiomer (com-
pound 3) also resulted in a loss of activity (IC50 > 100 M; Table 1),
mM), sug-
D
m
suggesting that the ligandeprotease binding was stereospecific.
Substituting the P3 Lys with non-polar (compounds 4e9), un-
charged polar (compounds 10e12) and anionic residues (com-
pounds 13 and 14) resulted in either a total loss or drastic reduction
in inhibitory activities, highlighting the specificity of the S3 binding
pocket (Table 1). Interestingly, compound 8 with a P3 norleucine
(Nle) retained some inhibitory activity (IC50 47.9 ꢀ 6.2
mM). This
2. Results and discussion
may be because Nle is an analog of Lys without the cationic Nε
amino group, suggesting that the amino moiety played an impor-
tant binding role within the S3 subsite (Fig. 3A). Intrigue by this, we
substituted the P3 Lys with another cationic residue, arginine
(compound 15). Assay results revealed that compound 15 suffered
a 4-fold decrease in inhibitory activity compared to 1 (IC50
2.1. Chemistry
The preparation of compound 1 can be found in a previous
report [17]. The synthetic procedure for the remaining compounds
was published recently [20] and is outlined in Scheme 1. Detailed
synthetic procedures are reported in the supplementary notes.
19.1 ꢀ 2.0 vs. 4.7 ꢀ 1.2
mM; Table 1), suggesting that the enzyme’s S3
binding pocket was lysine specific. Modeling revealed that the
more bulky arginine side chain was too bulky to fit completely into
the S3 subsite due to steric interactions caused by Gln86 and
Met156 and the guanidino moiety could not make the same
hydrogen bond interactions like the Nε amino group of the P3 lysine
(Fig. 3B).
2.2. Biological activity
Inhibitory activities of the agmatine dipeptides are summarized
in Table 1. Compound 1 was modeled using the WNV NS2B/NS3
crystal structure 3E90.pdb complexed to the tripeptide aldehyde
inhibitor (2-naphthoyl-KKR-H) as a template. Structural analysis
revealed that the inhibitor adopted a ‘C’-shaped conformation with
Next, the P3 Lys was replaced with aromatic residues (com-
pounds 16e18). Compound 16 with a P3 His showed no inhibitory
activity while compounds 17 and 18 displayed weak inhibitory
activities (IC50 44.5 ꢀ 5.3 and 33.5 ꢀ 2.7
mM; Table 1). Docking
studies of compound 18 revealed that the Trp indole side chain
could participate in hydrophobic interaction and hydrogen bond to
the Ile155 side chain and carbonyl moiety respectively (Fig. 3C).
Based on the IC50s of compounds 2e18, we concluded that Lys was
the optimal residue for enzyme binding and our P3 mutation re-
sults were in agreement with the co-crystal structure published by
Robin and co-workers (3E90.pdb) [16].
N-cap
P3
P2
NH2
P1
O
NH
H
N
H
N
N
H
N
H
NH2
Our next series of inhibitors involved N-cap mutation. As
mentioned earlier, the carbonyl O of the N-cap was observed to be
involved in intra-molecular hydrogen bonding with the P1 guani-
O
O
ε
ꢀ
dino N moiety (w1.7 A). This oriented the aromatic portion of the
ꢀ
N-cap 3.6 A away from the agmatine guanidino group (distance
1
NH2
between the guanidino central C to the ring center) and could
therefore participate in intra-molecular cation-p stacking (Fig. 3A).
To investigate the role played by this intra-molecular hydrogen
bond, we removed the carbonyl functionality by converting
Fig. 2. Structure of compound 1 linked to an agmatine moiety (in blue) and demar-
cated into four regions (P1, P2, P3 and N-cap). (For interpretation of the references to
colour in this figure legend, the reader is referred to the web version of this article.)

H.A. Lim et al. / European Journal of Medicinal Chemistry 62 (2013) 199e205
201
H CO
3
O
H CO
3
H CO
3
O
H CO
3
O
O
BocHN
a, b
c
d, e
O
H
2
NBoc
NHBoc
NBoc
NHBoc
OCH
OCH
OCH
3
OCH
3
3
3
HN
HN
N
O
N
H
NH
2
H N
N
H
2
= polystyrene resin
f, e
H CO
3
O
BocHN
H N
2
2
2
O
NBoc
NHBoc
O
NH
NH
g, h
OCH
H
H
N
3
N
O
R
N
H N
2
N
H
N
H
N
H
N
H
P3
P3
2
O
O
P3 = appropriate amino acid
Scheme 1. Synthesis of agmatine dipeptides. Reagents and conditions: (a) resin swelling in DMF, 25 ꢁC, 1 h; (b) 1,4-diaminobutane, NaCNBH3, 1% AcOH in DMF (v/v), microwave,
60 ꢁC, 10 min, repeat; (c) N,N0-bis(tert-butoxy carbonyl)thiourea, DIPEA, DMF, microwave, 60 ꢁC, 10 min; (d) Fmoc-Lys(Boc)-OH, DIC, HOBt, 90% CH2Cl2 in DMF (v/v), microwave,
60 ꢁC, 10 min, repeat; (e) 20% piperidine in DMF (v/v), microwave, 60 ꢁC, 5 min; (f) Fmoc-P3-OH, DIC, HOBt, 90% CH2Cl2 in DMF (v/v), microwave, 60 ꢁC, 10 min; (g) Appropriate
carboxylic acid, DIC, HOBt, 90% CH2Cl2 in DMF (v/v), microwave, 60 ꢁC, 10 min; (h) 95% TFA in CH2Cl2 (v/v), 25 ꢁC, 30 min.
the N-terminus into a free amine (compound 19; Table 1). Exper-
imental results were striking, showing an 8-fold decrease in
inhibitory activity compared to compound 1 (IC50 38.2 ꢀ 5.7 vs
compound 1 (IC50 8.5 ꢀ 0.9 vs 4.7 ꢀ 1.2
mM; Table 1), suggesting that
the distal phenyl ring of 4-PhPhac was also involved in enzyme
binding. Next, we investigated the binding role of the N-cap Phac
group of compound 29. As observed earlier, the aromatic ring was
situated w3.6 A from the P1 guanidino moiety (distance between
the ring center to the central guanidino C), allowing favorable intra-
4.7 ꢀ 1.2
mM). A plausible explanation could be that the cationic
ꢀ
guanidino moiety was repulsed by the positively-charged N-ter-
minus, resulting in an increase in conformational energy and
a decrease in binding affinity. This supports the ‘C’-shaped binding
pose adopted by 2-naphthoyl-KKR-H in the X-ray co-crystal
structure [16]. Replacing the biphenyl moiety with non-aromatic
alkyl groups should also abrogate inhibitory potencies by remov-
molecular cation-
3-phenylpropanoyl and 4-phenylbutanoyl functionalities (com-
pounds 30 and 31; Table 1) should disrupt the catione interactions
p stacking to occur. Substituting Phac with
p
and hence reduce their inhibitory activities. As expected, compound
ing the cation-
p
stacking. To test this hypothesis, the 4-PhPhac N-
30 exhibited a 5-fold decrease in activity compared to compound 29
cap was replaced with six alkyl moieties (compounds 20e25).
Starting with an acetyl capping group (compound 20), an approx-
imate 4-fold reduction in inhibitory activity was observed com-
(IC50 42.2 ꢀ 6.0 vs. 8.5 ꢀ 0.9
m
M) while compound 31 displayed only
M; Table 1). Intrigued,
a slight reduction in potency (IC50 11.2 ꢀ 1.0
m
we modeled compounds 30 and 31 into the protease followed by
pared to compound 1 (IC50 18.2 ꢀ 2.5 vs 4.7 ꢀ 1.2
m
M), supporting
energy minimization. Results showed that the phenyl ring of com-
pound 30 was situated w4.1 A away from the P1 guanidino moiety
ꢀ
our hypothesis. Computer modeling revealed that the methyl group
of the acetyl N-cap could participate in hydrophobic interaction
with the protease Ile155 side chain (Fig. 3D). Lengthening the acetyl
with propanoyl, butanoyl and 3-methylbutanoyl groups did not
significantly affect inhibitory activities (compounds 21e23;
Table 1). However, substituting them for branched acyl moieties
caused a slight reduction in activities (compounds 24 and 25;
Table 1). Modeling of compound 25 with a trimethylacetyl N-cap
suggested that this could be due to steric hindrance between the
bulky trimethylacetyl group and the protease Ile155 side chain
(Fig. 3E).
(measured from the ring center to the central C of the guanidino
ꢀ
group), compared to 3.6 A for compound 29. This could have
reduced intra-molecular cationep interactions with the P1 guani-
dino moiety (Fig. 3G). For compound 31, the propylene chain shifted
the phenyl ring away from the P1 guanidino moiety and oriented it
ꢀ
3.8 A toward the NH of Thr132 (Fig. 3H). As aromatic rings can act as
hydrogen bond acceptors, this could have resulted in favorable
dipolee
p
interactions and may explain why compound 31 was
M; Table 1).
4-fold more potent than 30 (IC50 11.2 ꢀ1.0 vs. 42.2 ꢀ 6.0
m
In the last series of inhibitors, ring-substitution studies were
conducted on the N-cap Phac group. Shifting the 4-phenyl ring of
compound 1 to the 3-position (compound 32; Table 1) resulted in
Next, we substituted the N-caps with aromatic benzoyl and
naphthoyl moieties (compounds 26e28; Table 1). Surprisingly,
their inhibitory activities were approximately 2-fold less potent
than the acetyl analog (compound 20). Modeling of compound 27
showed that the aromatic N-cap was largely solvent-exposed
(Fig. 3F) and the observed reduction in activity could be due to its
structural rigidity which prevented the naphthoyl group from ori-
enting itself into a more favorable position for hydrophobic in-
teractions with Ile155.
Further examination of the docked model of compound 1
revealed that its 4-PhPhac N-cap was situated between two ‘cliffs’
formed by Thr132 and Ile155 of the enzyme (Fig. 3A). The phenyl
ring distal to the N-cap carbonyl could take part in hydrophobic
interaction with the enzyme’s Pro131 five-membered ring. To
investigate further, we replaced the 4-PhPhac with a phenacetyl
(Phac) group (compound 29; Table 1). Protease inhibition results
revealed an approximate 2-fold decrease in activity compared to
a
2-fold decrease in inhibitory activity (IC50 4.7
ꢀ
1.2 vs.
9.9 ꢀ 2.3 M; Table 1). This may be due to loss of the hydrophobic
m
interaction between the phenyl ring and the enzyme’s Pro131 ring
(Fig. 3I). Placing a Cl moiety at the 2-, 3- and 4-positions of the N-
cap Phac (compounds 33e35; Table 1) yielded different inhibitory
results. The most potent inhibitor, compound 33, with a 2-Cl sub-
stituent was 2-fold more potent than compound 1 (IC50 2.6 ꢀ 0.3 vs.
4.7 ꢀ 1.2
mM respectively; Table 1). Our molecular model suggested
that the 2-Cl group could either participate in dipoleedipole
interaction with the side chain hydroxyl of Thr132 or have
increased hydrophobic interactions with Ile155 (Fig. 3J). The 3-Cl
analog (compound 34; IC50 9.6 ꢀ 2.0
mM), like compound 32, was
likely to be sterically hindered if pointing toward Thr132 and is
hence likely to point in the opposite direction (Fig. 3K). This was
likely to orient the 3-Cl moiety too far to interact with Thr132.

202
H.A. Lim et al. / European Journal of Medicinal Chemistry 62 (2013) 199e205
Table 1 (continued)
Table 1
Inhibitory data of agmatine dipeptide inhibitors with the sequence: (N-cap)-P3-Lys-
agmatine.
Compound
N-cap
P3
IC50 (mM)
O
P3
P2 (Lys)
P1 (agmatine)
28
Lys
32.5 ꢀ 4.7
O
H
N
H
N
N-cap
P3
NH2
N
H
N
H
O
O
NH
29
30
Lys
Lys
8.5 ꢀ 0.9
3
NH2
O
42.2 ꢀ 6.0
Compound
N-cap
P3
IC50 (mM)
1
2
4-PhPhac
e
Lys
CH3 (acetyl)
4.7 ꢀ 1.2
>100
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
4-PhPhac
4-PhPhac
4-PhPhac
4-PhPhac
4-PhPhac
4-PhPhac
4-PhPhac
4-PhPhac
4-PhPhac
4-PhPhac
4-PhPhac
4-PhPhac
4-PhPhac
4-PhPhac
4-PhPhac
4-PhPhac
H
D
-Lys
>100
>100
Gly
Ala
Val
Leu
Nle
Pro
Asn
Gln
Ser
Asp
Glu
Arg
His
Phe
Trp
Lys
O
53.7 ꢀ 6.8
31
32
Lys
Lys
11.2 ꢀ 1.0
9.9 ꢀ 2.3
>100
>100
47.9 ꢀ 6.2
>100
O
>100
>100
>100
>100
>100
19.1 ꢀ 2.0
Cl
>100
O
44.5 ꢀ 5.3
33.5 ꢀ 2.7
38.2 ꢀ 5.7
33
34
Lys
Lys
2.6 ꢀ 0.3
9.6 ꢀ 2.0
O
20
Lys
18.2 ꢀ 2.5
O
Cl
Cl
O
21
22
Lys
Lys
19.1 ꢀ 3.3
18.5 ꢀ 4.1
O
35
Lys
4.8 ꢀ 1.0
O
O
CH3
O
36
37
38
Lys
Lys
Lys
6.8 ꢀ 1.2
9.2 ꢀ 1.6
6.9 ꢀ 1.3
23
24
Lys
Lys
19.2 ꢀ 4.2
23.8 ꢀ 4.8
O
O
O
H3C
H3C
O
25
26
Lys
Lys
27.1 ꢀ 5.0
38.6 ꢀ 5.3
O
Unsurprisingly, compounds 32 and 34 exhibited almost identical
inhibitory potencies (IC50 9.9 ꢀ 2.3 vs. 9.6 ꢀ 2.0 M respectively).
Similarly, compound 35 with a 4-Cl moiety also exhibited identical
M respec-
m
potency to compound 1 (IC50 4.8 ꢀ 1.0 vs. 4.7 ꢀ 1.2
m
O
tively), suggesting that the 4-phenyl ring could be replaced with
4-Cl moiety (Fig. 3L). The 4-Cl was too far to interact with the hy-
droxyl group of Thr132 or the side chain of Ile155 unlike compound
33 and this may account for its weaker IC50 (4.8 ꢀ 1.0 vs.
27
Lys
33.3 ꢀ 3.9
2.6 ꢀ 0.3
mM respectively). Substituting Cl with a methyl group at

H.A. Lim et al. / European Journal of Medicinal Chemistry 62 (2013) 199e205
203
Fig. 3. Stick models of selected compounds docked onto the WNV NS2B/NS3 catalytic site using 3E90.pdb as a template; (A) Cpd. 1; (B) Cpd. 15 showing steric hindrance caused by
Gln86 and Met156; (C) Cpd. 18 showing interactions between P2 Trp and Ile155; (D) Cpd. 20 showing the proximity of the acetyl N-cap to Ile155; (E) Cpd. 25 showing steric clash
between its N-cap with Ile155; (F) Cpd. 27 with its solvent-exposed N-cap; (G) Cpd. 30 showing possible intra-molecular cationep interactions between the N-cap and the P1
guanidino moiety; (H) Cpd. 31 N-cap interacting with Thr132 NH (CPK model); (I) Cpd. 32 with its N-cap oriented between Thr132 and Ile155; (J) Cpd. 33 showing its 2-Cl N-cap
interacting with Thr132; (K) Cpd. 34 with its 3-Cl N-cap between Thr132 and Ile155; (L) Cpd. 35 with its 4-Cl N-cap too far away to interact with Thr132.
the 2-, 3- and 4-positions (compounds 36e38; Table 1) yielded
a similar pattern of results albeit at reduced potencies. For example,
replacing the 2-Cl (compound 33) with 2-methyl (compound 36)
resulted in an approximate 3-fold reduction in inhibitory activity
(IC50 2.6 ꢀ 0.3 vs. 6.8 ꢀ 1.2
mM respectively). A possible explanation
may be due to the loss in dipoleedipole interaction with the side
chain hydroxyl of Thr132 when 2-Cl is replaced with a methyl
group (Fig. 3J). Interestingly, replacing 3-Cl (compound 34) with

204
H.A. Lim et al. / European Journal of Medicinal Chemistry 62 (2013) 199e205
3-methyl (compound 37) did not result in significant changes in
M respectively). This was
probably because ring substituents at the meta position could not
interact with the protease (see Fig. 3I and K). The final analog tes-
ted, compound 38, exhibited a slight reduction in activity compared
(2H, s, AreCH2eCOe), 4.26e4.28 (2H, m, lysine
a
-Hs), 7.28e7.40
23.8, 23.9,
activity (IC50 9.6 ꢀ 2.0 vs. 9.2 ꢀ 1.6
m
(4H, m, aromatics); 13C NMR (100 MHz, CD3OD)
d
27.0, 27.4, 28.0, 28.1, 32.0, 32.4, 39.9, 40.5, 40.6, 41.2, 48.5, 48.7,
48.9, 128.5, 130.1, 130.5, 133.2, 134.6, 135.5, 158.3, 173.7, 174.2,
174.7. ESI-TOF-MS: m/z calc C25H44ClN8O3 (M þ Hþ) 539.3225,
found 539.3221.
to compound 1 (IC50 6.9 ꢀ 1.3 vs. 4.7 ꢀ 1.2
mM respectively). A
plausible explanation could be that replacing the 4-phenyl group
with a methyl could have reduced the hydrophobic interaction
between the 4-Ph and Pro131.
4.2. Biological activities
4.2.1. WNV NS2B/NS3 enzyme inhibition assay
3. Conclusion
WNV NS2B/NS3 inhibitory assays were based on published work
[10] and performed in a buffer at pH 8.0 containing TriseHCl
(10 mM), CHAPS (1 mM) and glycerol (20% v/v). The enzyme
(20 nM) and varying concentrations of inhibitor were next added
and pre-incubated at 25 ꢁC for 1 h. The reaction was initiated by the
addition of the fluorogenic peptide substrate Pyr-RTKR-AMC
In this communication, we demonstrated that agmatine di-
peptides can be utilized as tool compounds for SAR studies on
ligandeenzyme interactions. We were able to confirm that the
optimal P3 residue for enzyme inhibition was Lys, supporting an
earlier screening effort involving a small tripeptide aldehyde li-
brary [21]. Our data also support both the proposed X-ray co-
crystal structure published by Robin and co-workers (3E90.pdb)
[16] and the peptide aldehyde enzyme inhibition data [15]. In the
former, we believe the inhibitor N-cap did not interact directly
with the protease (Fig. 1B) and was not due to an artifact of crys-
tallization. The seemingly conflicting conclusions based on the two
papers [15,16] were possibly due to inadequate SAR studies on the
inhibitor N-cap. Based on our experimental results, we conclude
that certain N-caps indeed have minimal interaction with the
protease (as in the cases shown in Figs. 1B and 3F), while some
N-caps could orient themselves favorably to interact with the
protease as exemplified by compound 33 (Fig. 3J). Finally, by syn-
thesizing and screening thirty-seven analogs of compound 1, we
were able to identify a novel agmatine dipeptide with a two-fold
improvement in inhibitory activity (compound 33). Taken
together, our data should provide medicinal chemists with deeper
insights into the design of new drugs for the treatment of WNV
infections.
(Bachem, Switzerland) to make a final concentration of 20
mM.
The reaction components were shaken for 5 s and the reaction
progress monitored at 37 ꢁC by measuring the increase in fluo-
rescence (lex355 nm and lem460 nm) every 45 s for 1 h on
a SpectraMax Gemini XS plate reader (USA). Experiments were
conducted in duplicates. IC50 values were derived by fitting the
initial velocity against the log [inhibitor] with a sigmoidal dose
response curve using GraphPad Prism 5 software (USA). Nona-
Arg-NH2 peptide (GenScript, USA) was used as positive control.
D-
4.3. Molecular modeling and visualization software
The WNV protease X-ray structure PDB entry 3E90 [16] was
downloaded from the Protein Data Bank [www.pdb.org] and pre-
pared with the protein preparation wizard in Maestro 9.3 (Schrö-
dinger, USA) using standard settings. This included the addition of
hydrogen atoms, bond assignments, removal of water molecules
ꢀ
>7 A from the ligand, protonation state assignment, optimization of
the hydrogen bond network and restrained minimization using the
OPLS2005 force field [22]. The co-crystallized, covalently-bound
inhibitor was used as a template for modeling the conformation
and orientation in the binding site of the agmatine inhibitors. The
inhibitoreprotein complex was finally energy-minimized using
4. Experimental protocols
4.1. Chemistry
ꢀ
Macromodel 9.9 (Schrödinger, USA). All residues >7 A from the
All reagents and solvents were obtained from commercial
sources and were used without further purification. Acetonitrile
was purchased from Merck KGaA (Germany). Fmoc-protected
amino acids and HBTU were bought from GL Biochem (China).
CD3OD was purchased from Cambridge Isotope Laboratories (USA).
All other reagents were purchased from SigmaeAldrich (USA).
Crude target inhibitors were purified using a reverse-phase C18
column (Waters X-bridge) on a high performance liquid chroma-
tography (HPLC) system with an ultraviolet detector set at 215 nm
(Shimadzu Prominence). The mobile phase consisted of solvent A
(water) and solvent B (acetonitrile). The gradient started with 1%
solvent B for 5 min which was increased to 11% in 40 min. All target
compounds were characterized by electrospray ionization-time of
flight-mass spectrometry (ESI-TOF-MS; Agilent 6224 TOF) using
Mass Hunter software. NMR spectra were recorded on a Bruker
Ultrashield 400þ spectrometer in CD3OD. Chemical shifts were
ligand were constrained before the complex was subjected to 500
steps of Polak-Ribière Conjugate Gradient [23] energy mini-
mization using the OPLS2005 force field and GB/SA continuum
solvation method [24]. Model visualization was done using
Chem3D Ultra v.10 software (ChembridgeSoft, USA).
Acknowledgements
We thank A*STAR Biomedical Research Council for financial
support, Dr. Manfred Raida for mass spectrometric support, Dr.
Thomas Keller and Dr. Michael Entzeroth for critical reading of the
manuscript.
Appendix A. Supplementary data
Supplementary data related to this article can be found at
http://dx.doi.org/10.1016/j.ejmech.2012.12.043.
expressed as
d (ppm) relative to TMS.
4.1.1. Synthesis of compounds
References
The preparation of compound 1 can be found in a previous
report [17]. The general synthetic outline of compounds 2e38 is
shown in Scheme 1 and is based on a previous report [20]. Syn-
thetic details can be found in the supplementary notes. Spectral
[1] T. Solomon, N. Engl. J. Med. 351 (2004) 370e378.
[2] L.H. Gould, E. Fikrig, J. Clin. Invest. 113 (2004) 1102e1107.
[3] L.D. Kramer, J. Li, P.-Y. Shi, Lancet Neurol. 6 (2007) 171e181.
[4] R.T. Johnson, Curr. Clin. Top. Infect. Dis. 22 (2002) 52e60.
[5] www.cdc.gov/ncidod/dvbid/westnile/index.htm.
[6] J. Kaiser, Science 337 (2012) 1030.
data of compound 33: 1H NMR (400 MHz, CD3OD)
d
1.29e1.85
(16H, m, lysine side chains), 2.80e3.23 (8H, m, agmatine), 3.78

H.A. Lim et al. / European Journal of Medicinal Chemistry 62 (2013) 199e205
205
[7] M.S. Diamond, Antiviral Res. 83 (2009) 214e227.
[17] H.A. Lim, J. Joy, J. Hill, C.S.B. Chia, Eur. J. Med. Chem. 46 (2011) 3130e
3134.
[18] M.R. Wiley, N.Y. Chirgadze, D.K. Clawson, T.J. Craft, D.S. Gifford-Moore,
N.D. Jones, J.L. Olkowski, A.L. Schacht, L.C. Weir, G.F. Smith, Bioorg. Med. Chem.
Lett. 5 (1995) 2835e2840.
[19] G.L. Becker, F. Sielaff, M.E. Than, I. Lindberg, S. Routhier, R. Day, Y. Lu,
W. Garten, T. Steinmetzer, J. Med. Chem. 53 (2010) 1067e1075.
[20] M.J.Y. Ang, H.A. Lim, C.S.B. Chia, Tetrahedron Lett. 53 (2012) 5511e5514.
[21] J.E. Knox, N.L. Ma, Z. Yin, S.J. Patel, W.-L. Wang, W.-L. Chan, K.R.R. Rao,
G. Wang, X. Ngew, V. Patel, D. Beer, S.P. Lim, S.G. Vasudevan, T.H. Keller,
J. Med. Chem. 49 (2006) 6585e6590.
[8] S. Mukhopadhyay, R.J. Kuhn, M.G. Rossmann, Nat. Rev. Microbiol. 3 (2005) 13e22.
[9] I. Radichev, S.A. Shiryaev, A.E. Aleshin, B.I. Ratnikov, J.W. Smith,
R.C. Liddington, A.Y. Strongin, J. Gen. Virol. 89 (2008) 636e641.
[10] T.A. Nall, K.J. Chappell, M.J. Stoermer, N.X. Fang, J.D. Tyndall, P.R. Young,
D.P. Fairlie, J. Biol. Chem. 279 (2004) 48535e48542.
[11] I. Schechter, A. Berger, Biochem. Biophys. Res. Commun. 27 (1967) 157e162.
[12] K.J. Chappell, M.J. Stoermer, D.P. Fairlie, P.R. Young, J. Biol. Chem. 281 (2006)
38448e38458.
[13] K.J. Chappell, M.J. Stoermer, D.P. Fairlie, P.R. Young, Curr. Med. Chem. 281
(2008) 2771e2784.
[14] C.G. Noble, Y.-L. Chen, H. Dong, F. Gu, S.P. Lim, W. Schul, Q.-Y. Wang, P.-Y. Shi,
Antiviral Res. 85 (2010) 451e462.
[22] G.A. Kaminski, R.A. Friesner, J. Tirado-Rives, W.L. Jorgensen, J. Phys. Chem. B
105 (2001) 6474e6487.
[15] M.J. Stoermer, K.J. Chappell, S. Liebscher, C.M. Jensen, C.H. Gan, P.K. Gupta,
W.-J. Xu, P.R. Young, D.P. Fairlie, J. Med. Chem. 51 (2008) 5714e5721.
[16] G. Robin, K. Chappell, M.J. Stoermer, S.-H. Hu, P.R. Young, D.P. Fairlie,
J.L. Martin, J. Mol. Biol. 385 (2009) 1568e1577.
[23] E. Polak, G. Ribière, Revenue Francaise informat. Recherche opérationelle,
Série Rouge 16 (1969) 35e43.
[24] W. Hasel, T.F. Hendrickson, W.C. Still, Tetrahedron Comput. Methodol. 1
(1988) 103e116.