Dual inhibitors of the dengue and West Nile virus NS2B-NS3 proteases: Synthesis, biological evaluation and docking studies of novel peptide-hybrids

Article abstract of DOI:10.1016/j.bmc.2015.07.012

Dengue virus (DENV) and West Nile virus (WNV) are mosquito-borne arboviruses responsible for causing acute systemic diseases and severe health conditions in humans. The discovery of therapies capable to prevent infections or treat infected individuals rem

Full text of DOI:10.1016/j.bmc.2015.07.012

Accepted Manuscript
Dual inhibitors of the dengue and West Nile virus NS2B-NS3 proteases: Syn-
thesis, biological evaluation and docking studies of novel peptide-hybrids
Allan Bastos Lima, Mira A.M. Behnam, Yasmin El Sherif, Christoph Nitsche,
Sergio M. Vechi, Christian D. Klein
PII:
S0968-0896(15)00595-7
DOI:
http://dx.doi.org/10.1016/j.bmc.2015.07.012
BMC 12441
Reference:
Bioorganic & Medicinal Chemistry
To appear in:
Received Date:
Revised Date:
Accepted Date:
11 March 2015
8 July 2015
9 July 2015
Please cite this article as: Lima, A.B., Behnam, M.A.M., Sherif, Y.E., Nitsche, C., Vechi, S.M., Klein, C.D., Dual
inhibitors of the dengue and West Nile virus NS2B-NS3 proteases: Synthesis, biological evaluation and docking
studies of novel peptide-hybrids, Bioorganic & Medicinal Chemistry (2015), doi: http://dx.doi.org/10.1016/j.bmc.
2015.07.012
This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers
we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and
review of the resulting proof before it is published in its final form. Please note that during the production process
errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

Dual inhibitors of the dengue and West Nile virus
NS2B-NS3 proteases: Synthesis, biological evaluation
and docking studies of novel peptide-hybrids
Allan Bastos Limaa, Mira A. M. Behnama, Yasmin El Sherifa, Christoph Nitsche a, Sergio M.
Vechia,b, Christian D. Kleina
aMedicinal Chemistry, Institute of Pharmacy and Molecular Biotechnology IPMB, Heidelberg
University, Im Neuenheimer Feld 364, 69120 Heidelberg, Germany
bFederal University of Alagoas (UFAL), Campus Arapiraca, NCEx, 57309-005 Arapiraca,
Alagoas, Brazil
ABSTRACT:
Dengue virus (DENV) and West Nile virus (WNV) are mosquito-borne arboviruses
responsible for causing acute systemic diseases and severe health conditions in humans.
The discovery of therapies capable to prevent infections or treat infected individuals
remains an important challenge, since no vaccine or specific efficient treatment could be
developed so far. In this context, we present herein the synthesis, characterization,
biological evaluation and docking studies of novel peptide-hybrids based on 2,4-
thiazolidinedione scaffolds containing non-polar groups. The most promising compound
has an IC50 of 0.75 μM against WNV protease, which represents a seventyfold
improvement in activity compared to our previously reported compounds. Experimental
1

results and docking studies are in agreement with the hypothesis that a non-polar group
in the scaffold is important to obtain interactions between the inhibitors and a
hydrophobic pocket in the substrate recognition region of the DENV and WNV NS2B-
NS3 serine proteases.
Keywords: Dengue virus, West Nile virus, peptide-hybrids, protease inhibitors
2

1. Introduction
Dengue is a viral infection that is considered one of the seventeen neglected tropical
diseases, as classified by the World Health Organization (WHO). These diseases affect,
in particular, people living in underdeveloped and emergent tropical countries, where
children are generally the most vulnerable to infections.[1] DENV has however become
a threat also to developed countries.[2-4] It is transmitted by mosquitoes of the Aedes
spp., from which Aedes aegypti and Aedes albopictus are the most important vectors.[5,
6] Four serotypes of the virus (DENV-1, DENV-2, DENV-3 and DENV-4) have been
described, with a recent study indicating the discovery of a fifth subtype.[6-9] Since the
discovery and isolation in Japan (1943) of the serotype later know as serotype 1, the
distribution of the four serotypes has spread worldwide.[10] In 2010, for instance, the
virus victimized around 390 million people all over the world.[1, 2]
Similarly, WNV has progressively become a considerable public health concern. Even
though at a slower rate than DENV, WNV has infected thousands of people since its
discovery and the number of infections increased drastically during the last decade. The
virus was discovered in 1937 when it was first isolated from the blood of a Ugandan
native during an epidemiological investigation undertaken in the presumably endemic
zone.[11] Transmission occurs particularly by mosquitoes of the Culex spp., but the
virus can also be transmitted by some other species like Aedes and Anopheles.[12]
Among all of the mosquito-borne flaviviruses, WNV has the most widespread
geographical distribution and the largest vector and host range, which reinforces its
potential as a global health threat.[13] Besides several undesirable symptoms, WNV can
3

cause fatal neurological diseases in humans, but remains asymptomatic in 80% of the
cases.[14]
There are currently no proven specific therapies for the prevention or treatment of
infections caused by DENV and WNV.[15, 16] The evolution of these diseases and
their consequences to the society trigger therefore the urgency for the development of
new therapies. For this purpose, non-structural proteins of the virus present great
potential. Our interest and focus here lies on the viral NS2B-NS3 serine protease.
DENV and WNV NS2B-NS3 proteases share several similarities in the binding pocket,
which may represent the possibility of developing dual inhibitors for both targets.
Compounds with activity against more than one viral pathogen have a number of
advantages: Synergies can be used in the clinical development stages (in particular,
safety studies have to be performed only once); the compounds could be used in cases
where the diagnosis (WNV or DENV infection) is not certain; and they could be used as
preventive measures when traveling into areas in which both DENV and WNV are
endemic. On the other hand, the catalytic sites of the flaviviral NS2B-NS3 proteases are
rather flat, making it a challenge to develop high-affinity inhibitors.[17] Hence, the
combined use of computational tools and in vitro assays enabled important advances in
understanding DENV and WNV proteases and may have a pivotal role in the design and
development of novel inhibitors for these enzymes.[18]
In this context, several inhibitors ranging from substrate-like to non-peptidic have
already been reported.[19-27] For the DENV NS2B-NS3 protease, for instance, Yin and
co-workers performed an extensive study on peptide hybrids.[20, 21] For the WNV
NS2B-NS3 protease, Stoermer and co-workers developed potent cationic inhibitors with
4

serum stability and cell permeability.[26] Potent peptidomimetic inhibitors were
developed by Hammamy and co-workers.[27] Using high-throughput screening,
Johnston and co-workers also identified a variety of non-peptidic inhibitors.[28]
Previous works from our group revealed 2,4-thiazolidinedione peptide-hybrids as
inhibitors of the DENV NS2B-NS3 protease serotype 2, with higher target affinities
than the previously tested analogs. [19] These compounds consist of a sequence of three
amino acids coupled to a non-peptidic N-terminal group which we denote as Cap. The
tripeptide sequence capped with benzoic acid at the N-terminus showed significant
inhibitory activity, and combination with substituted 2,4-thiazolidinediones led to a
further increase of activity. The most active peptide-hybrid thus found is comprised of a
sequence of amino acid residues (arginine, lysine, norleucine) capped with a 2,4-
thiazolidinedione moiety with a lipophilic substituent on the 3 position, presenting an
in-vitro IC50 of 2.5 μM. The most recent work from our group explored the C-terminal
position, evaluating the effect of exchanging the C-terminus residue of amino acid
(norleucine) to several other natural and artificial residues. Sequences containing
aromatic residues in that position generally presented higher activity and phenylglycine
was identified as the most active.[29] This sequence presented an IC50 of 3.3 μM when
capped with benzoic acid (Compound I) (Table 1). Combination with the most active
2,4-thiazolidinedione Cap previously found, which contains a cyclopentyl group as a
substituent in position 3 of the ring (Compound II) (Table 1), led to the expected
improvement in activity (IC50 = 0.6 μM). This structure was therefore considered for the
development of further inhibitors. The WNV NS2B-NS3 protease has also been
targeted in our previous works, but most of the compounds presented inhibition below
30% at screening conditions (enzyme 150 nM, inhibitor 50 μM, substrate 50 μM, Km =
212 μM).[19, 29] Compound I (Table 1) was one of the exceptions, being also the most
5

active with a percentage inhibition of 98,4 % against WNV. However, at that time, no
further investigation was carried out with these compounds against WNV protease.
Also from earlier investigations with our previously developed compounds, using a
homology model of the DENV 2 protease, docking studies indicated that the
phenylglycine of the peptide-hybrids presumably interacts with residues in the S1
pocket, while the arginine interacts with the S2 and the lysine with the S4 pocket. The
benzene ring of the N-terminal Cap appears to bind to a hydrophobic region near the S3
and S4 substrate binding pockets. This hydrophobic area extends to the cofactor
domain, where the 2,4-thiazolidinedione group would be interacting.[19, 29] Thus,
based on these studies, we reasoned that the Cap could be optimized in order to better
interact with the hydrophobic domain of the DENV and WNV NS2B-NS3 proteases.
Considering this, we here present the synthesis, characterization, biological evaluation
and docking studies of novel peptide-hybrids based on 2,4-thiazolidinedione scaffolds
containing non-polar groups. We also evaluated a small group of peptide-hybrids with
Caps containing alkynyl groups which had not been explored yet, but also showed
relevant inhibitory activity towards the proteases. All of the compounds presented here
showed inhibition at both DENV and WNV proteases and low inhibition at thrombin.
We see therefore an opportunity to develop at the same time more compounds that
could inhibit both targets, but also present low risk of interfering with the activity of
other serine proteases like thrombin.
2. Results and discussion
6

2.1. Chemistry
The synthesis of the targeted compounds 7a-s was accomplished according to
methodologies illustrated in Scheme 1. The potassium salt of 2,4-thiazolidinedione 2
can be easily prepared through the reaction of potassium hydroxide with commercially
available 2,4-thiazolidinedione 1 under reflux in ethanol. A diversity of the alkylated
2,4-thiazolidinediones scaffolds 3 could be then obtained by reacting alkyl halides with
the thiazolidinedione salt 2.[30] The Knoevenagel condensation of 3 with
formylbenzoic acid using acetic acid as solvent under microwave-assisted conditions
provided the 5-arylidenethiazolidine-2,4-dione derivatives 4.[19, 31] The insolubility of
the arylidene products in the reaction medium allowed the products to be filtrated out of
the solution with suitable purity for the last step of the synthesis. Considerations
towards the geometry of the 5-arylidene moiety were made previously, and the
preferred stereoisomer of the double bond was determined to be (Z), apparently due to
steric effects.[31, 32] The peptidic domain of the compounds was synthesized using the
Fmoc protocol.[25, 33] Therefore, each amino acid was coupled with the carboxylic
acid activated by HBTU coupled to the resin and deprotected with piperidine. The
synthesis was finished after the 5-arylidene compounds were coupled also with their
carboxylic acid activated by HBTU and the final products 7 were obtained after
cleavage with trifluoroacetic acid and purification. In total, 19 peptide-hybrids were
obtained. The compounds herein presented (except 2, 3a, 3b, 3f, 3h, 3l, 3o, 3q, 3r, 4a,
4f and 4l) have not been reported in the literature before.[19, 30]
7

Scheme 1. Synthesis of thiazolidinedione-based peptide hybrids containing non-polar
groups a) KOH, EtOH, reflux; b) Alkyl halide, DMF, reflux; c) 4-Formylbenzoic acid,
NH4OAc, AcOH, MW 180 °C, 5min; d) piperidine, DMF; e) Fmoc-Phg-OH, HBTU,
DIPEA, DMF; f) Fmoc-Lys(Boc)-OH, HBTU, DIPEA, DMF; g) Fmoc-Arg(Pbf)-OH,
HBTU, DIPEA, DMF; h) HBTU, DIPEA, DMF; i) TFA, TIPS, H2O.
The N-alkylation of the potassium salt of 2,4-thiazolidinedione with
(bromomethyl)cyclopropane, 2-bromonorbonane and 2-bromoadamantane to afford the
correspondent compounds was also attempted. The desired products of these reactions
could not be obtained. In the case of 2-bromoadamantane, steric hindrance may have
prevented the SN2 reaction to occur, whereas in the case of (bromomethyl)cyclopropane
and 2-bromonorbonane, the compounds may have undergone rearrangements or
decomposition leading to a complex mixture of products which could not be separated.
8

2.2. In vitro protease inhibition assays
The compounds herein synthesized were evaluated with a rster (fluorescence)
resonance energy transfer (FRET)-based assay against the DENV and WNV proteases.
[25] All peptide-hybrids presented inhibition higher than 98% in the initial screening
assay at DENV NS2B-NS3 protease serotype 2 (enzyme 100 nM, inhibitor 50 μM,
substrate 50 μM, Km= 105 μM). For the WNV NS2B-NS3 protease, all compounds
showed relative inhibition above 90%, and about half of them higher than 98% (enzyme
150 nM, inhibitor 50 μM, substrate 50 μM, Km = 212 μM). Therefore, the IC50 was
determined for all compounds at both proteases (Table 1). None of the compounds
presented relevant activity against thrombin (enzyme: 10 nM, inhibitor: 25 μM,
substrate: 50 μM, Km = 16 μM), except for 7f, presenting a relative inhibition of 39%.
Some of the 5-arylidene caps without peptidic moiety (compound series 4) were also
characterized at the proteases. These 5-arylidenes are structurally similar to previously
reported ones for which no significant inhibitory activity was observed.[19] As
expected, these substances were also poorly active at the DENV and WNV proteases,
with an inhibitory activity below 50% in the screening assay (inhibitor concentration:
50 μM) (cf. Supporting Information).
A number of investigations were previously carried out in order to determine a peptide
sequence that would fit into the binding site of the DENV and WNV proteases.[26, 27,
29, 34] For the protease from DENV, the most promising pharmacophore found in our
group so far is a peptide-hybrid incorporating the non-natural amino acid phenylglycine
(Phg), with the general structure Cap-Arg-Lys-Phg-NH2. The most active Cap
identified so far is a 2,4-thiazolidinedione moiety that contains a hydrophobic group
9

(cyclopentyl) as a substituent in the position 3 of the ring (compound II) (Table 1).[29]
Aiming to further explore this position, this structure was used as a model to the
selection of the non-polar groups for the scaffolds herein synthesized. On this basis,
hydrophobic substituents with variable size, flexibility and bulkiness were chosen to
build these novel scaffolds, and some interesting aspects could be observed. For the
protease from DENV, the most active Caps contain small substituents in position 3 of
the 2,4-thiazolidinedione ring, as can be noticed for 7a (IC50 = 0.47 μM) and 7i (IC50 =
0.62 μM), bearing isopropyl and cyclobutyl groups, respectively. Adding one methylene
group to the linear alkyl chain while maintaining the terminal structure led to a decrease
in activity of the corresponding compounds, 7b (IC50 = 1.01 μM) and 7j (IC50 = 0.78
μM), respectively. Insertion of a second methylene group into the chain led to a slightly
increased activity, as can be seen for compound 7c (IC50 = 0.90 μM), the extended
analog of 7a and 7b. A more elongated substituent was explored for the alkynyl
analogs. However, 7r (IC50 = 0.46 μM), with an ethyl linker, was identified as the most
active not only among this class, but among all peptide-hybrids presented here. The
higher homolog 7s (IC50 = 0.85 μM), did not show significant improvement towards 7r
or 7q (IC50 = 0.99 μM), which indicates that a longer and more flexible chain does not
necessarily improve activity.
Table 1. Inhibitory activity of thiazolidinedione-based peptide hybrids (6a-s) against
DENV and WNV proteases and thrombina
DENV
WNV
IC50 (μM)c
THR
R
No.
b
d
e
%
IC50 (μM)c
%
%
I[28]
95.0 ± 0.6
99.1 ± 0.6
3.32 ± 0.05
39.3 ± 1.4
94.4 ± 1.0
58.1 ± 1.8
n.i.
†
0.47 ± 0.09
3.59 ± 0.15
8.5 ± 1.8
7a
10

DENV
WNV
IC50 (μM)c
THR
R
No.
b
d
e
%
IC50 (μM)c
%
%
99.6 ± 0.5
1.01 ± 0.03
90.7 ± 0.4
95.4 ± 0.5
93.5 ± 2.1
6.24 ± 0.63
1.76 ± 0.10
2.03 ± 0.10
n.i.
7b
100.0 ± 0.4 0.90 ± 0.09
100.6 ± 2.8 0.58 ± 0.02
7.3 ± 2.1
9.7 ± 3.2
7c
7d
101.0 ± 1.2 0.51 ± 0.04
101.7 ± 4.6 0.92 ± 0.02
98.9 ± 0.5
92.8 ± 2.0
95.0 ± 1.1
1.91 ± 0.21
2.50 ± 0.10
1.12 ± 0.07
9.7 ± 4.1
38.9 ± 2.6
8.0 ± 3.7
7e
7f
99.3 ± 0.9
0.74 ± 0.05
7g
100.2 ± 1.8 0.65 ± 0.04
100.0 ± 0.4 0.62 ± 0.04
99.2 ± 0.6
95.3 ± 0.2
1.35 ± 0.13
1.69 ± 0.09
8.7 ± 2.2
4.7 ± 4.4
7i
99.7 ± 0.2
99.6 ± 0.3
0.78 ± 0.04
0.60 ± 0.01
95.9 ± 0.6
98.4 ± 0.9
1.39 ± 0.12
1.97 ± 0.23
8.0 ± 5.6
7j
II[28]
9.2 ± 2.45
100.2 ± 0.5 0.85 ± 0.08
100.6 ± 0.7 0.79 ± 0.09
98.8 ± 1.6
98.0 ± 0.5
99.6 ± 0.3
1.01 ± 0.06
1.71 ± 0.10
0.75 ± 0.04
10.1 ± 3.3
9.5 ± 2.8
6.9 ± 1.0
7k
7l
99.3 ± 0.9
99.9 ± 0.5
1.05 ± 0.05
1.92 ± 0.16
7m
100.9 ± 0.9
101.2 ± 1.8
1.13 ± 0.14
2.11 ± 0.15
7.3 ± 4.8
7n
7o
100.3 ± 0.8 0.83 ± 0.04
101.8 ± 2.2 0.74 ± 0.03
11.5 ± 2.9
106.7 ± 4.0
1.27 ± 0.06
6.6 ± 4.9
7p
99.2 ± 1.0
99.9 ± 0.2
0.99 ± 0.08
0.46 ± 0.02
93.1 ± 1.2
97.8 ± 1.7
2.36 ± 0.14
2.12 ± 0.05
10.9 ± 3.3
13.0 ± 2.5
7q
7r
103.1 ± 5.3 0.85 ± 0.03
105 ± 2.8
2.75 ± 0.10
10.4 ± 2.2
7s
11

† Identical peptidic sequence as other analogs, but capped with benzoyl moiety (Bz-
a
Lys-Arg-Phg). All measurements were carried out in triplicate. n.i. = no inhibition
bPercent inhibition of the DENV NS2B-NS3 protease serotype 2 (enzyme: 100 nM,
c
inhibitor: 50 µM, substrate: 50 µM, Km = 105 µM). IC50 values against DENV and
WNV NS2B-NS3 protease serotype 2 were determined at a substrate concentration of
d
50 μM. Percent inhibition of WNV NS2B-NS3 protease (enzyme: 150 nM, inhibitor:
e
50 μM, substrate: 50 μM, Km = 212 μM). Percent inhibition of thrombin (enzyme: 10
nM, inhibitor: 25 μM, substrate: 50 μM, Km = 16 μM).
For the WNV protease, the improvement in activity of the compounds was more
prominent in relation to our previous results. In this case, the IC50 previously published
for compound I, with a benzoyl Cap coupled to the arginine, was 58.1 μM.[29] When
the peptide sequence of compound I (Arg-Lys-Phg-NH2) was merged with the Caps
herein developed, the resulting compounds presented at least a tenfold higher activity
(Table 1). Structure-activity relationships similar to DENV protease activity was also
observed for the WNV protease, but the difference between the activities of compounds
bearing similar elongated substituent is sometimes negligible. In any case, Caps with an
ethyl spacer in the 3-substituent of the 2,4-thiazolidinedione, for instance 7c (IC50 =
1.76 μM), showed a somewhat higher activity than the analogs 7a (IC50 = 3.59 μM) and
7b (IC50 = 6.24 μM). The highest affinities, however, were observed for the
cyclopentylmethylene and cyclohexylmethylene analogs 7m (IC50 = 0.75 μM) and 7k
(IC50 = 1.01 μM). 7n, also bearing a cyclohexyl ring at the terminus of the substituent
chain, showed an IC50 of 1.13 μM, being the third most active analog within this series.
The most active compound 7m showed a seventyfold improvement in the activity at
12

WNV protease compared to the compounds reported previously.[29] It was also seven
times more active than 7b, the least active one. This indicates that size and flexibility of
the non-polar substituent on the Cap moiety influences the activity. Compounds bearing
alkynyl groups (7q-w) did not present significant activities, which supports the idea that
bulky groups capable of making extended hydrophobic interactions in that position of
the Cap are more important for the WNV than for the DENV protease.
2.3. Aprotinin fluorescence quenching assay and kinetic studies
To verify the binding mode at DENV serine protease, the aprotinin fluorescence
quenching assay was exemplarily performed with the most active compound against
DENV (7r). [35] The assay was performed as described before, using DENV protease
at 1 μM, aprotinin at 10 μM and different concentrations of the inhibitor (0, 0.5, 1, 2, 5,
10, 20, 30, 40, 48 μM).[19] This assay is based on the intrinsic fluorescence of Trp50 of
the DEN protease, which is located near the active site of the enzyme. Inhibitors
containing groups that are capable of absorbing radiation in the range of the tryptophan
emission quench this fluorescence by FRET if they bind in the active site and therefore
close to the Trp50 residue. Displacement of the inhibitor by aprotinin, a known DENV
protease inhibitor with relatively high affinity, will partially restore the fluorescence.
Compound 7r caused a concentration-dependent, nonlinear fluorescence quenching of
DENV protease (Figure 1). With the addition of aprotinin to the system, fluorescence
was significantly restored, indicating specific binding of 7r to the active site of DENV
protease.
13

*Corrected for aprotinin intrinsic fluorescence.
Figure 1. Results of the aprotinin fluorescence quenching assay. The concentration-
dependent nonlinear quench of the DENV protease fluorescence intensity by 7r is
shown. The intrinsic fluorescence of the protease, caused by Trp50 in the vicinity of the
substrate binding region, is partially restored when aprotinin is added.
Studies on the inhibition mechanisms were performed for the most active compounds
against DENV and WNV serine proteases, 7r and 7m, respectively. The Ki values and
binding modes were determined using Cheng-Prusoff, Dixon and Cornish-Bowden
plots.[36-39] The experiments were carried out at four different substrate (50, 100, 150,
200 μM) and inhibitor (0, 0.25, 0.5, 1, 1.5, 2, 2.5, 3 μM for DENV, and 0, 0.5, 1, 2, 3, 4,
5, 6 μM for WNV) concentrations. Measurements at DENV protease were carried out in
triplicate. At WNV protease, triplicate measurements were repeated three times. For the
14

data analysis and generation of the plots, out of 96 values obtained for compound 7r at
DENV protease, two were excluded as outliers, and out of 288 obtained for compound
7m at WNV protease, 12 were excluded.
Using Cheng-Prusoff’s method, the Ki can be obtained as the intercept at the Y axis
from a plot of IC50 values as a function of the substrate concentration. For compound 7r
at DENV protease, the Ki thus determined is 0.40 ± 0,03 μM (Figure 2A). At WNV
protease, compound 7m has a Ki value of 1.30 ± 0.07 μM (Figure 3A). In both cases,
IC50 values increase in correspondence to substrate concentrations, indicating a
competitive inhibition mechanism.
Using the Dixon method, plots of the reciprocal velocity as a function of the substrate
concentration were generated and the average of the intercepts between the lines is
regarded as the Ki value. A Ki value of 0.10 ± 0.10 μM was obtained for compound 7r at
the DENV protease and a Ki value of 0.82 ± 0.3 μM for compound 7m at the WNV
protease (Figures 2B and 3B). These Ki values are in reasonable agreement with the
values determined via the Cheng-Prusoff method. On the Cornish-Bowden plots,
parallel lines indicate a competitive inhibition mechanism for both compounds (Figures
2C and 3C).
Figure 2. Kinetic studies for compound 7r at DENV protease. A: Cheng-Prusoff plot
with the Ki value of 0.40 ± 0,03. B: Dixon plot with the Ki value of 0.10 ± 0.10 μM. C:
15

Cornish-Bowden plot. Cheng-Prusoff and Cornish-Bowden plots indicate a competitive
inhibition mechanism.
Figure 3. Kinetic studies for compound 7m at WNV protease. A: Cheng-Prusoff plot
with the Ki value of 1.3 ± 0.3 μM. B: Dixon plot with the Ki value of 0.82 ± 0.3 μM. C:
Cornish-Bowden plot. Cheng-Prusoff and Cornish-Bowden plots indicate a competitive
inhibition mechanism.
2.4. Docking studies
Docking studies were performed with AutoDock Vina (cf. Supporting Information for
details) using DENV and WNV NS2B-NS3 serine proteases structures.[40] We used the
previously reported homology model for the DENV-2 protease and the X-ray structure
2FP7 for the WNV protease to dock all the compounds presented in Table 1.[17, 29] In
our analysis, considering the similarity of the compounds, we focused on the binding
mode of our most active ones against DENV and WNV proteases, 7r and 7m,
respectively. In general, the predicted binding modes for these compounds presented
some similarities to those proposed in our previous work, as summarized above.[29]
Considering the poses obtained for all ligands docked into the DENV protease, it was
16

noticed that the ligand residues phenylglycine and arginine, besides the Cap, do not
vary significantly in their orientations within the binding pockets. This, however, did
not apply to the lysine residue, which interacts either with the protein or with the ligand
itself. Another tendency observed, comparing the poses obtained for the DENV and the
WNV proteases, was that the binding modes for the ligands docked into the DENV
structure were fairly similar to each other, while for the WNV structure the variability
of the poses was more pronounced. Therefore, confidence for the DENV docking results
may be higher than for the WNV results.
The proposed binding modes for the most potent analogs 7r docked into DENV
protease and 7m into WNV protease are presented in Figures 4 and 5 with the
respective Kyte-Doolittle hydrophobicity scale mapped on the solvent-accessible
surface of the protein. For both DENV and WNV proteases the phenylglycine residue of
the ligand is placed in the S1 pocket, the arginine residue in the S2 and the Cap in the S3
and S4 pockets. On the other hand, lysine is located between the S1 and S2 pockets for
the DENV protease, while pointing to the ligand itself in the WNV protease, apparently
lacking a specific interaction with the enzyme. The conformation of the lysine sidechain
in the docking studies is relatively undefined and varies between individual ligands and
docking runs, whereas the positions of the phenylglycine and arginine moieties were
fairly conserved within their respective binding pockets. For the DENV protease, 21
protein residues are involved in interactions with the ligand within a distance of 3.9 Å,
while for the WNV protease there were 19 residues involved (cf. Table S2 in
Supporting Information).[41] From these residues, 16 were observed in both cases,
but their interactions were not always identical. A summary of the main intermolecular
interactions and some of their features is presented in Table S3 (cf. Supporting
17

Information). In both cases an intramolecular - stacking interaction between the
aromatic rings of the phenylglycine and the Cap may be considered. The distances of
the centroids of these rings are 5.7 Å and 6.1 Å for the DENV and WNV protease,
respectively. This interaction could stabilize this ligand conformation favoring the other
interactions described in Table S2. Tyrosines from the proteases (Tyr215 at DENV and
Tyr161 at WNV) may also be involved in - stacking interactions. The role of -
stacking is coherent with the structure activity relationships observed previously.[29]
The docking studies support the experimental results, which indicate the importance of
a non-polar substituent in the N-terminal Cap. Accordingly, hydrophobic interactions
with residues from the NS3–NS2B interface should have a significant effect in the
binding affinity of the ligand. For the WNV protease, docking indicates that bulky non-
polar substituents are expected to have higher affinity, whereas the DENV protease
prefers smaller, less flexible but still hydrophobic groups.[42]
18

19

Figure 4. Close up view of the docked ligand-protein complex predicted with AD Vina
for the DENV protease. Ligand 7r is shown in green. (a) Protein surface is colored by
amino acid hydrophobicity using the Kyte-Doolittle scale. Change from blue to white to
dark orange indicates increase in hydrophobicity.[43] Negatively charged surface areas
are shown in red. The positions of the binding pockets S1, S2, S3 and S4 and the catalytic
triad are indicated. The protein is presented in the serine protease standard orientation,
i.e., looking into the active site cleft. (b) Detailed view of the main interactions
presented in the proposed binding mode. Labels of the NS2B and NS3 residues are
shown in blue and red, respectively, labels for the catalytic triad are black, distances of
the intermolecular hydrogen bonds are in magenta, and the centroids of the aromatic
rings, perhaps involved in - stacking interactions, are colored in orange with
respective distances labeled in black. The graphical representation of the binding mode
was generated using Chimera.
20

Figure 5. Close up view of the docked ligand-protein complex predicted with AD Vina
for the WNV protease. Ligand 7m is shown in dark green. See legend of Figure 4 for
further details.
21

3. Conclusion
Peptide-hybrids are dual inhibitors of the DENV and WNV proteases. The present work
elucidates the SAR of Caps bearing hydrophobic groups in position 3 of the 2,4-
thiazolidinedione ring. The most active compounds against DENV protease, for
instance 7a, 7d and 7r, presented lower activity against WNV protease. On the other
hand, the most active compounds against WNV protease, for instance 7k and 7m,
showed reasonable affinity towards DENV protease. None of the compounds, except for
7f, presented relevant activity against thrombin. We therefore see an opportunity to
develop dual inhibitors of the WNV and DENV proteases. The SAR exploration of the
Caps resulted in good improvement of the activity against WNV protease. Results from
docking simulations provide insight into the interactions between the inhibitors and the
proteases, and explain the importance of having a non-polar group in the Cap to interact
with the hydrophobic pocket. Future work will be directed towards the discovery of new
Caps that include hydrophobic moieties which bind even more effectively to the
corresponding binding pocket, and combine these with further optimized peptidic
groups.
4. Experimental
4.1. Chemistry
All reagents and solvents were obtained from commercial suppliers and were used
without further purification. All chemicals for the precursor synthesis were obtained
22

from Sigma-Aldrich (Germany) and were of analytical grade. The protected amino acids
were purchased from Orpegen (Germany) and Carbolution Chemicals (Germany).
HBTU and Rink amide AM resin (capacity 0.63 mmol/g) were purchased from Iris
Biotech (Germany). NMR spectra of intermediates were recorded on Varian NMR
instrument at 300 MHz, 300 K in acetone-d6 (δ ppm: 2.09), CDCl3 (δ ppm: 7.26) or
DMSO-d6 (δ ppm: 2.50). NMR spectra of peptide-hybrids were recorded on a Varian
NMR instrument at 500 MHz, 300 K in deuterium oxide-d2 (δ ppm: 4.79). Chemical
shifts (δ) are given in parts per million (ppm) and the solvents were used as reference
peaks. Residuals of nondeuterated solvents were used as internal standard. Coupling
constants (J) are given in hertz (Hz). Mass spectra were measured on a Bruker
micrOTOF-Q II (HR-ESI) instrument. Flash chromatography was performed on a
Biotage Isolera One purification system using silica gel (0.060–0.200 mm) cartridges
(KP-Sil) and UV monitoring at 254 and 280 nm. The reaction progress was determined
by thin layer chromatography on Merck silica gel plates 60 F254 (UV detection).
Microwave synthesis was done using a Monowave 300 synthesis reactor with IR
temperature sensor from Anton Paar. Purity of the compounds used in biological assays
was determined by LC-MS using an Agilent 1200 HPLC system with a MWD detector
combined with a Bruker micrOTOF-Q II instrument on an RP-18 column (ReproSil-
Pur-ODS-3, Dr. Maisch GmbH, Germany, 3 μm, 50 mm × 2 mm).
4.2. General procedure A: Synthesis of thiazolidine-2,4-dione potassium salt (2)
and 3-alkylthiazolidine-2,4-diones (3) with exemplary compound data.
The compounds were prepared according to the procedure previously described.[19, 30]
To a refluxing solution of thiazolidine-2,4-dione (43 mmol) in ethanol (25 mL) was
23

added a hot solution of potassium hydroxide (45 mmol) in ethanol (25 mL). After
additional refluxing for 30 min, the mixture was cooled to room temperature and the
precipitate was filtered and washed with cold ethanol. The obtained potassium 2,4-
dioxothiazolidin-3-ide (2) (5 mmol) was refluxed with alkyl halides (5.5 mmol) in DMF
(15 mL) for 3–4 h. After cooling to room temperature and addition of water (50 mL),
the crude product was extracted 3 times with ethyl acetate, washed with brine, and
purified by flash chromatography (cyclohexane/ethyl acetate).
4.2.1. 3-(3,3-dimethylbutyl)thiazolidine-2,4-dione (3e)
Compound 3e was prepared according to standard procedure A by using compound 2
1
and 1-chloro-3,3-dimethylbutane. Obtained an orange solid in 68% yield. H NMR
(300 MHz, CDCl3, δ ppm): 0.96 (s, 9H), 1.47 (m, 2H), 3.63 (m, 2H), 3.92 (s, 2H);
HRMS (ESI): m/z [M+Na]+ calcd for C9H15NNaO2S: 224,0721, found: 224,0717.
4.3. General procedure B: Synthesis of 5-Arylidenethiazolidine-2,4-diones (4) with
exemplary compound data.
A mixture of substituted thiazolidine-2,4-dione (1 mmol), ammonium acetate (2 mmol),
and formylbenzoic acid (1.1 mmol) in acetic acid (2 mL) was heated in a microwave
reactor at 180 °C for 5 min. The mixture was cooled for 30 min at 4 °C, and the
resulting precipitate was filtered, washed with water and cold ether and dried by air
flow.
4.3.1. 4-[(3-(3,3-dimethylbutyl)-2,4-dioxothiazolidin-5-ylidene)methyl]benzoic acid
(4e)
24

Compound 4e was prepared according to standard procedure B by using compound 3e.
1
Obtained pale yellow crystals in 86% yield. H NMR (300 MHz, DMSO-d6, δ ppm):
0.94 (s, 9H), 1.48 (m, 2H), 3.66 (m, 2H), 7.74 (d, J = 8.4, 2H), 7.97 (s, 1H), 8.06 (d, J =
8.4, 2H); HRMS (ESI): m/z [M–H]– calcd for C17H19NO4S: 333,1035, found: 332,096.
ASSOSSIATED CONTENT
Supporting Information
Synthesis of peptide hybrids and peptidic assay substrates, HRMS data for all
synthesized peptide-hybrids, expression and purification of the viral proteases, DENV
and WNV protease inhibition assay, inhibitory activity of selected Caps against DENV
and WNV proteases, analytical data for all compounds, and details of computational
studies.
ABBREVIATIONS USED
DENV, dengue virus; DIPEA, diisopropylethylamine; HBTU, O-(benzotriazol-1-yl)-
N,N,N′,N′-tetramethyluronium hexafluorophosphate; MW, microwave; RFU, relative
fluorescence units; THR, thrombin; WNV, West Nile virus
AUTHOR INFORMATION
25

Corresponding Author.
*Tel.: (+)49-6221-544875; Fax: (+)49-6221-546430; E-mail address: c.klein@uni-
heidelberg.de
ACKNOWLEDGMENT
We thank Heiko Rudy for measuring ESI high resolution spectra, Lena Weigel for
performing LC-MS analytics and Sabine Volz for technical support. A.B.L and S.M.V.
appreciate financial support by a research fellowship from the National Council for
Scientific and Technological Development (CNPq), Brazil. M.B. appreciates financial
support by the German Academic Exchange Service (DAAD). The project was
sponsored by the Deutsche Forschungsgemeinschaft, KL-1356/3-1.
REFERENCES
[1] World Health Organization (WHO). Global strategy for dengue prevention and
control 2012-2020, World Health Organization, Geneva, Switzerland, 2012, pp. vi, 43.
[2] S. Bhatt, P.W. Gething, O.J. Brady, J.P. Messina, A.W. Farlow, C.L. Moyes, J.M.
Drake, J.S. Brownstein, A.G. Hoen, O. Sankoh, M.F. Myers, D.B. George, T. Jaenisch,
G.R. Wint, C.P. Simmons, T.W. Scott, J.J. Farrar, S.I. Hay, The global distribution and
burden of dengue, Nature, 496 (2013) 504-507.
26

[3] O. Brathwaite Dick, J.L. San Martin, R.H. Montoya, J. del Diego, B. Zambrano,
G.H. Dayan, The history of dengue outbreaks in the Americas, The American journal of
tropical medicine and hygiene, 87 (2012) 584-593.
[4] M.G. Guzman, S.B. Halstead, H. Artsob, P. Buchy, J. Farrar, D.J. Gubler, E.
Hunsperger, A. Kroeger, H.S. Margolis, E. Martinez, M.B. Nathan, J.L. Pelegrino, C.
Simmons, S. Yoksan, R.W. Peeling, Dengue: a continuing global threat, Nature
reviews. Microbiology, 8 (2010) S7-16.
[5] I. Kautner, M.J. Robinson, U. Kuhnle, Dengue virus infection: epidemiology,
pathogenesis, clinical presentation, diagnosis, and prevention, J Pediatr., 131 (1997 )
516-524.
[6] E A Henchal, J.R. Putnak, The dengue viruses, Clin. Microbiol. Rev., 3 (1999) 379-
396.
[7] K.L. Barr, B.D. Anderson, G.L. Heil, J.A. Friary, G.C. Gray, D.A. Focks, Dengue
serotypes 1–4 exhibit unique host specificity in vitro, Virus Adaptation and Treatment,
4 (2012) 65–73.
[8] R.F. Mercado-Curiel, H.A. Esquinca-Aviles, R. Tovar, A. Diaz-Badillo, M.
Camacho-Nuez, L. Munoz Mde, The four serotypes of dengue recognize the same
putative receptors in Aedes aegypti midgut and Ae. albopictus cells, BMC
microbiology, 6 (2006) 85.
[9] N. Vasilakis, M. Ooi, M. Rabaa, S. Mayer, S. Widen, D. Perera, A. Hickey, A.
Travassos Da Rosa, K. Bossart, R. Tesh, E. Holmes, J. Cardos, The daemon in the
forest-emergence of a new dengue serotype in southeast Asia., III. International
Conference on Dengue and Dengue Haemorrhagic FeverBangkok, 2013.
[10] E. Konishi, G. Kuno, In memoriam: Susumu Hotta (1918-2011), Emerg Infect Dis,
19 (2013) 843-844.
27

[11] K.C. Smithburn, T.P. Hughes, A.W. Burke, J.H. Paul, A Neurotropic Virus
Isolated from the Blood of a Native of Uganda, American Journal of Tropical Medicine,
20 (1940) 471-472.
[12] D.J. Gubler, The continuing spread of West Nile virus in the western hemisphere,
Clinical infectious diseases : an official publication of the Infectious Diseases Society of
America, 45 (2007) 1039-1046.
[13] H. Weissenbock, Z. Hubalek, T. Bakonyi, N. Nowotny, Zoonotic mosquito-borne
flaviviruses: worldwide presence of agents with proven pathogenicity and potential
candidates of future emerging diseases, Veterinary microbiology, 140 (2010) 271-280.
[14] L.E. Davis, R. DeBiasi, D.E. Goade, K.Y. Haaland, J.A. Harrington, J.B. Harnar,
S.A. Pergam, M.K. King, B.K. DeMasters, K.L. Tyler, West Nile virus neuroinvasive
disease, Annals of neurology, 60 (2006) 286-300.
[15] K.L. Tyler, Current developments in understanding of West Nile virus central
nervous system disease, Current opinion in neurology, 27 (2014) 342-348.
[16] D. Weiskopf, M.A. Angelo, J. Sidney, B. Peters, S. Shresta, A. Sette,
Immunodominance Changes as a Function of the Infecting Dengue Virus Serotype and
Primary versus Secondary Infection, Journal of virology, 88 (2014) 11383–11394.
[17] P. Erbel, N. Schiering, A. D'Arcy, M. Renatus, M. Kroemer, S.P. Lim, Z. Yin, T.H.
Keller, S.G. Vasudevan, U. Hommel, Structural basis for the activation of flaviviral
NS3 proteases from dengue and West Nile virus, Nature structural & molecular biology,
13 (2006) 372-373.
[18] C. Nitsche, S. Holloway, T. Schirmeister, C.D. Klein, Biochemistry and Medicinal
Chemistry of the Dengue Virus Protease, Chemical reviews, (2014).
28

[19] C. Nitsche, V.N. Schreier, M.A. Behnam, A. Kumar, R. Bartenschlager, C.D.
Klein, Thiazolidinone-peptide hybrids as dengue virus protease inhibitors with antiviral
activity in cell culture, Journal of medicinal chemistry, 56 (2013) 8389-8403.
[20] Z. Yin, S.J. Patel, W.L. Wang, W.L. Chan, K.R. Ranga Rao, G. Wang, X. Ngew,
V. Patel, D. Beer, J.E. Knox, N.L. Ma, C. Ehrhardt, S.P. Lim, S.G. Vasudevan, T.H.
Keller, Peptide inhibitors of dengue virus NS3 protease. Part 2: SAR study of
tetrapeptide aldehyde inhibitors, Bioorganic & medicinal chemistry letters, 16 (2006)
40-43.
[21] Z. Yin, S.J. Patel, W.L. Wang, G. Wang, W.L. Chan, K.R. Rao, J. Alam, D.A.
Jeyaraj, X. Ngew, V. Patel, D. Beer, S.P. Lim, S.G. Vasudevan, T.H. Keller, Peptide
inhibitors of Dengue virus NS3 protease. Part 1: Warhead, Bioorganic & medicinal
chemistry letters, 16 (2006) 36-39.
[22] S.M. Tomlinson, S.J. Watowich, Anthracene-based inhibitors of dengue virus
NS2B-NS3 protease, Antiviral research, 89 (2011) 127-135.
[23] M. Ezgimen, H. Lai, N.H. Mueller, K. Lee, G. Cuny, D.A. Ostrov, R.
Padmanabhan, Characterization of the 8-hydroxyquinoline scaffold for inhibitors of
West Nile virus serine protease, Antiviral research, 94 (2012) 18-24.
[24] J.E. Knox , N.L. Ma , Z. Yin , S.J. Patel , W. Wang , W. Chan , K. R. Ranga Rao ,
G. Wang , X. Ngew , V. Patel , D. Beer , S.P. Lim , S.G. Vasudevan , T.H. Keller,
Peptide Inhibitors of West Nile NS3 Protease: SAR Study of Tetrapeptide Aldehyde
Inhibitors, Journal of medicinal chemistry, 49 (2006) 6585–6590.
[25] C. Nitsche, C.D. Klein, Fluorimetric and HPLC-Based Dengue Virus Protease
Assays Using a FRET Substrate, in: SpringerLink (Ed.) Antiviral Methods and
Protocols, Humana Press, New York, 2013 pp. 221-236
29