Thiazolidinone-peptide hybrids as dengue virus protease inhibitors with antiviral activity in cell culture

Article abstract of DOI:10.1021/jm400828u

The protease of dengue virus is a promising target for antiviral drug discovery. We here report a new generation of peptide-hybrid inhibitors of dengue protease that incorporate N-substituted 5-arylidenethiazolidinone heterocycles (rhodanines and thiazolidinediones) as N-terminal capping groups of the peptide moiety. The compounds were extensively characterized with respect to inhibition of various proteases, inhibition mechanisms, membrane permeability, antiviral activity, and cytotoxicity in cell culture. A sulfur/oxygen exchange in position 2 of the capping heterocycle (thiazolidinedione-capped vs rhodanine-capped peptide hybrids) has a significant effect on these properties and activities. The most promising in vitro affinities were observed for thiazolidinedione-based peptide hybrids containing hydrophobic groups with K_i values between 1.5 and 1.8 μM and competitive inhibition mechanisms. Rhodanine-capped peptide hybrids with hydrophobic substituents have, in correlation with their membrane permeability, a more pronounced antiviral activity in cell culture than the thiazolidinediones.

Full text of DOI:10.1021/jm400828u

Subscriber access provided by NEW MEXICO STATE UNIV
Article
Thiazolidinone–Peptide Hybrids as Dengue Virus
Protease Inhibitors with Antiviral Activity in Cell Culture
Christoph Nitsche, Verena Schreier, Mira A.M. Behnam,
Anil Kumar, Ralf Bartenschlager, and Christian D.P. Klein
J. Med. Chem., Just Accepted Manuscript • DOI: 10.1021/jm400828u • Publication Date (Web): 01 Oct 2013
Downloaded from http://pubs.acs.org on October 5, 2013
Just Accepted
“Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted
online prior to technical editing, formatting for publication and author proofing. The American Chemical
Society provides “Just Accepted” as a free service to the research community to expedite the
dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts
appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been
fully peer reviewed, but should not be considered the official version of record. They are accessible to all
readers and citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered
to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published
in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just
Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor
changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers
and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors
or consequences arising from the use of information contained in these “Just Accepted” manuscripts.
Journal of Medicinal Chemistry is published by the American Chemical Society. 1155
Sixteenth Street N.W., Washington, DC 20036
Published by American Chemical Society. Copyright © American Chemical Society.
However, no copyright claim is made to original U.S. Government works, or works
produced by employees of any Commonwealth realm Crown government in the course
of their duties.

Page 1 of 69
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Thiazolidinone–Peptide Hybrids as Dengue Virus
Protease Inhibitors with Antiviral Activity in Cell
Culture
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
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
†
†
†
‡
Christoph Nitsche, Verena N. Schreier , Mira A. M. Behnam, Anil Kumar, Ralf
‡
†
Bartenschlager, Christian D. Klein*
†
Medicinal Chemistry, Institute of Pharmacy and Molecular Biotechnology IPMB, Heidelberg
University, Im Neuenheimer Feld 364, 69120 Heidelberg, Germany
‡
Department of Infectious Diseases, Molecular Virology, Heidelberg University, Im
Neuenheimer Feld 345, 69120 Heidelberg, Germany
ABSTRACT: The protease of dengue virus is a promising target for antiviral drug discovery. We
here report a new generation of peptide–hybrid inhibitors of dengue protease that incorporate Nꢀ
substituted 5ꢀarylidenethiazolidinone heterocycles (rhodanines and thiazolidinediones) as Nꢀ
terminal capping groups of the peptide moiety. The compounds were extensively characterized
with respect to inhibition of various proteases, inhibition mechanisms, membrane permeability,
antiviral activity, and cytotoxicity in cellꢀculture. A sulfur/oxygen exchange in position 2 of the
capping heterocycle (thiazolidinedioneꢀcapped vs. rhodanineꢀcapped peptide hybrids) has a
significant effect on these properties and activities. The most promising inꢀvitro affinities were
observed for thiazolidinedioneꢀbased peptide hybrids containing hydrophobic groups with Kiꢀ
1
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 2 of 69
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
values between 1.5 and 1.8 ꢀM and competitive inhibition mechanisms. Rhodanineꢀcapped
peptide hybrids with hydrophobic substituents have, in correlation with their membrane
permeability, a more pronounced antiviral activity in cellꢀculture than the thiazolidinediones.
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
INTRODUCTION
Recent estimates suggest that up to 390 million dengue virus (DEN) infections occur each year
of which around 100 million become clinically apparent in about 100 countries in Latin America,
1
ꢀ3
Southeast Asia, Central Africa and other regions all over the world. At the moment no
vaccines or antiviral agents are available, so that mosquito control is the sole approach to reduce
1
ꢀ4
infections in these countries. For the development of chemotherapeutical agents three viral
enzymes are considered as promising drug targets: the DEN protease, the methyltransferase and
4
the RNAꢀdependent RNAꢀpolymerase. As demonstrated for other, closely related viruses like
5
, 6
HCV, viral proteases represent druggable targets with therapeutic relevance.
Therefore,
inhibitors of the DEN protease are potential drug candidates for the treatment and prevention of
7
DENꢀrelated diseases.
The DEN genome consists of a singleꢀstranded RNA, encoding for three structural and seven
4, 7
nonꢀstructural proteins (NS1, NS2A, NS2B, NS3, NS4A, NS4B, NS5).
The multifunctional
NS3 protein provides a protease and a helicase function, localized in different domains. The NS3
protease domain requires the cytosolic core sequence of the NS2B protein as a cofactor to form
8
, 9
the catalytically competent complex NS2BꢀNS3. As in other flaviviral proteases, particularly
the closely related West Nile virus protease, the cofactor was found to be partially disordered and
4
, 10ꢀ
flexible in crystal structures in the absence of a substrate or a substrateꢀcompetitive inhibitor.
1
2
Instead of this ‘open’ conformation, the catalytically active ‘closed’ conformation with
2
ACS Paragon Plus Environment

Page 3 of 69
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
ordered cofactor domain is observed in crystal structures when a ligand is bound to the NS2Bꢀ
1
2ꢀ14
NS3 construct.
In this conformation, the cofactor contributes with a βꢀhairpin to the S and
2
4
S pockets, which are crucial for substrate recognition. Recently, crystal structures of two
3
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
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
1
4
inhibitors bound to the protease of DEN serotype 3 were published. Surprisingly, the broadꢀ
spectrum serine protease inhibitor aprotinin (BPTI) does not bind to the closed conformation of
4, 12, 14
NS2BꢀNS3,
whereas for a covalently bound tetrapeptidylꢀaldehyde inhibitor the closed
4
, 14
state was observed.
The closed state was also found predominantly in NMR structures of the
NS2BꢀNS3 construct in presence of a ligand; thereby a significant conformational change from
15ꢀ17
the open to the closed form was observed.
More recently, NMR experiments using a
NS2B/NS3 preparation without covalent linkage of the protease and the cofactor were presented.
These solution structures resemble the closed conformation of NS2BꢀNS3 in absence and
1
8
presence of an inhibitor, indicating that the previously observed open conformations are
artifacts which arise from the nonꢀnatural NS2BꢀNS3 construct with covalent linkage. However,
the binding of an inhibitor to alternative conformations of NS2BꢀNS3 may be
thermodynamically favored, which would explain the high affinity of aprotinin to the enzyme
4
, 14
and the observed crystal structure with this inhibitor.
This example implies that the closed
conformation is not necessarily the preferred structural target for the design and discovery of
inhibitors. Instead, the open conformation or an intermediate between the open and closed
conformations may also be relevant for inhibitor binding. Inhibitors that bind to these
conformations of the enzyme will not display a substrateꢀcompetitive binding behavior. An
additional difficulty in structureꢀbased drug design arises from the fact that under inꢀvivo
conditions the NS3 protein is part of a larger, membraneꢀassociated replication complex, which
1
9
might not properly resemble the NS2BꢀNS3 protein in isolated or crystallized form. In
3
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 4 of 69
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
particular, this caveat applies to the experimental characterization of enzyme inhibitors using the
isolated, recombinant NS2BꢀNS3 protein.
Considering these restrictions, empirical inꢀvitro and inꢀcellulo SAR studies of different
compound classes are required to obtain potent DEN protease inhibitors and improve our
understanding of their binding and inhibition modes. In the last years several compound series
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
2
0ꢀ27
for targeting the DEN protease have been published. Lowꢀmolecular weight inhibitors
without similarity to the natural substrate of the protease have, in most cases, uncertain or low
selectivity and potency, but can often be considered "drugꢀlike" with respect to permeability,
stability and ligand efficiency. Peptidic inhibitors that mimic the recognition sequence of the
2
8ꢀ35
protease and frequently incorporate electrophilic warheads
have increased potency, but often
carry intrinsic reactivity and pharmacokinetic liabilities due to highly basic side chains and,
consequently, pronounced polarity. However, some of these inhibitors have good inꢀvitro
affinities in the lower micromolar range and below in the case of boronic acid or
2
9
trifluoromethylketone as electrophilic warheads. Considering the successful discovery and
development of HCV protease inhibitors which are now in clinical use, substrateꢀbased peptides
appear as a promising starting point to obtain clinical development candidates after several
4
, 6, 36
optimization cycles.
The core challenge in the development of promising DEN protease
inhibitors is to combine the high affinity and selectivity of peptide inhibitors with the drugꢀ
likeness of smallꢀmolecular compounds without any highly reactive groups.
37
Peptide hybrids are an attractive possibility to approach this challenge. Recently we
2
0
described hybrids of arylcyanoacrylamides with short peptide sequences. These hybrids do not
contain a highly reactive warꢀhead and have a remarkable affinity against the DEN protease with
K ꢀvalues between 5 and 10 ꢀM for the best derivatives, but had no significant antiviral activity
i
4
ACS Paragon Plus Environment

Page 5 of 69
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
37
in cellꢀculture. Based on this partial, initial success we searched for alternative Nꢀterminal cap
moieties to increase the activity and to create membrane–permeable peptide hybrids with
increased antiviral activity in cellꢀculture. Herein we report Nꢀsubstituted 5ꢀ
arylidenethiazolidinone–peptide hybrids with an extensive evaluation of the thiazolidinedione
and rhodanine scaffolds. These 5ꢀmembered heterocycles have been previously explored as
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
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
2
2
smallꢀmolecular inhibitors of the DEN protease with moderate activity. Rhodanines,
thiazolidinediones and related heterocycles have been discussed extensively and controversially
2
2, 38ꢀ41
as key (or nuisanceꢀcausing) structures in medicinal chemistry.
These compounds have a
distinct binding profile to numerous targets which probably originates from a particularly high
potential to form intermolecular interactions, which is not related to aggregation or other
2
2
nonspecific, "promiscuous" inhibition mechanisms. Variations of the Nꢀsubstituent at position
3
of the heterocycle are a straightforward way to obtain series of compounds with different
hydrophobicity and to explore additional possibilities for intermolecular interaction. The
connection to the peptide can be achieved via the aromatic 5ꢀarylidene residue, as described
37
before.
RESULTS AND DISCUSSION
Chemistry. The chemical structure of the new series of peptide hybrids is highly influenced by
2
0, 22, 37
previously identified DEN protease inhibitors.
Tripeptides of the sequence ArgꢀLysꢀNle
with Cꢀterminally nonꢀsubstituted amide residues were combined with Nꢀterminal cap groups, as
3
7
this had been found in a screening of peptide hybrids as the most suitable approach. 5ꢀ
Arylidenerhodanines and ꢀthiazolidinediones were chosen as Nꢀterminal caps and combined via
the aromatic ring with the peptide chain using amide bonds, as this was previously reported as an
5
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 6 of 69
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
37
easy connection approach leading to affine compounds. To establish some additional
intermolecular interactions and to modulate the hydrophobicity, the 5ꢀmembered heterocycle was
further substituted at the nitrogen atom in position 3. As a proof of principle for our design
considerations, compounds with modified Nꢀterminal caps, alternative heterocycles, and
different peptide sequences were additionally synthesized.
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
The synthesis of peptide hybrids was executed by solidꢀphase peptide synthesis according to
the Fmoc protocol. Nꢀterminal cap molecules were connected to the presynthesized peptide
37
sequences using carboxylic acid functionalities as shown in Scheme 1. Connection between the
thiazolidinone scaffolds and the benzoic acid moiety was realized by a highꢀthroughput
2
2
Knoevenagel condensation with formylbenzoic acid under microwaveꢀassisted conditions.
Using acetic acid as solvent, all condensation products could be separated by filtration and were
used for the solidꢀphase peptide synthesis without any further purification steps. The geometry of
the double bond is assumed to be (Z), as was previously confirmed by Xꢀray crystal structures
2
2, 42, 43
for closely related analogs, which were prepared under identical conditions.
It appears that
the preferred geometry of the double bond is (Z) due to steric effects, with the sulfur atom in
44
position 2 of the heterocycle being cisꢀorientated to the aromatic ring. The absence of (E)
isomers was confirmed by LCꢀMS. The 5ꢀmembered Nꢀsubstituted heterocyclic rings were
obtained in different ways, depending on the nature of the substituent (R = alkyl, benzyl, aryl)
and the heterocycle itself (X = S, O). For rhodanines bearing alkyl or benzyl substituents, a
microwaveꢀassisted three component reaction in water using the amine, carbon disulfide and
4
5
chloroacetic acid was developed before. As this method was not adequately applicable for
4
5
arylamines, an alternative method was evaluated using bis(carboxymethyl)trithiocarbonate. Nꢀ
alkylꢀ and benzylthiazolidinediones were synthesized starting with deprotonation of 2,4ꢀ
6
ACS Paragon Plus Environment

Page 7 of 69
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
thiazolidinedione and isolation of the formed potassium salt followed by nucleophilic
4
6
substitution using alkyl or benzyl halides (Scheme 1). The 3ꢀ(pꢀtolyl)thiazolidineꢀ2,4ꢀdione
derivative was synthesized by oxidation of the corresponding 3ꢀ(pꢀtolyl)rhodanine with sodium
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
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
47
tungstate and hydrogen peroxide. 3ꢀPhenylꢀ and 3ꢀbenzylthiohydantoine were synthesized from
4
8
the corresponding isothiocyanates and methyl glycinate (Scheme 2). These compounds could
also be transformed into the 5ꢀarylidene derivatives using the same Knoevenagel reaction
conditions as for the other analogs. 2ꢀ[3ꢀ(4ꢀMethoxybenzyl)ꢀ4ꢀoxoꢀ2ꢀthioxothiazolidinꢀ5ꢀ
yl]acetic acid as precursor for compound 41 was synthesized according to a three component
49
reaction described in the literature (Scheme 3). For the synthesis of the unsaturated analog 2ꢀ
[
3ꢀ(4ꢀmethoxybenzyl)ꢀ4ꢀoxoꢀ2ꢀthioxothiazolidinꢀ5ꢀylidene]acetic acid an alternative three
5
0
component reaction was used (Scheme 3). This reaction was extended to the use of diꢀtertꢀ
butyl acetylenedicarboxylate, because the resulting tertꢀbutyl ester could easily be cleaved using
TFA in the following step.
All target compounds were tested in fluorimetric assays against the DEN and WNV proteases
and thrombin. Furthermore we included five of the most active peptide hybrids form the
previously reported series of compounds (I, II, III, IV, V) for correlation and comparison
purposes (Table 3). To confirm the results and to exclude falseꢀpositive results, selected
(
especially colored) compounds were additionally studied in an HPLCꢀbased DEN protease
5
1
assay.
Structure-Activity Relationships. 26 different substituents (including H) at the nitrogen atom
in position 3 (R according to Scheme 1) were evaluated for the rhodanine and thiazolidinedione
scaffolds, respectively, in the peptide hybrids shown in Table 1. The general tripeptide sequence
7
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 8 of 69
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
for all compounds in Table 1 is CapꢀArgꢀLysꢀNleꢀNH (Scheme 1). This sequence was identified
2
3
7
as the most active and selective one for the peptide hybrids evaluated before. The connection
3
7
via the arylidene system was arranged in paraꢀsubstitution, also as explored before.
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
All derivatives in Table 1 have only negligible activity against thrombin. Some rhodanine
derivatives with hydrophobic substituents (e.g. 9a, 23a, 24a, 25a, 26a) possess notable inhibitory
potency at the WNV protease with percentage of inhibition values above 97%. However, further
experiments at lower concentrations indicate that the expected IC ꢀvalues for all these
5
0
compounds are above 10 ꢀM, implicating DEN protease selectivity for these compounds. None
of the thiazolidinedione derivatives reached inhibition values of 90% or higher at the WNV
protease, indicating that the selectivity between the two viral targets is much higher for this
compound class than for the rhodanine derivatives. The relatively high %ꢀinhibition values of
some compounds in the WNV protease screen are representing a slightly lower target affinity
than %ꢀinhibition values in the same range for the DEN protease screen. This is due to the
difference in K , which is 212 ꢀM for the WNV protease FRET substrate and 105 ꢀM for the
m
DEN protease FRET substrate. The high selectivity of the compounds, especially between the
closely related DEN and WNV proteases, is clearly indicative of specific target–ligand
recognition, and makes a nonꢀspecific or promiscuous effect of the thiazolidinedione and
rhodanine heterocycles unlikely.
A high activity at the DEN protease (≥ 97% inhibition) was observed for the peptide hybrids
bearing hydrophobic substituents with a certain flexibility, e.g. butyl (7a, 7b), cyclopentyl (9a,
9b), cyclohexyl (10a, 10b) and all explored benzyl substituents (21aꢀ25a, 21bꢀ25b). Smaller
groups with lower hydrophobicity, like hydrogen (1a, 1b), methyl (2a, 2b) or hydroxyethyl (14a,
14b), and also substituents with a lack of flexibility, like all explored aryl groups (16a, 17a, 17b,
8
ACS Paragon Plus Environment

Page 9 of 69
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
1
8a, 19a, 20a) are less active. IC ꢀvalues were determined for all compounds with an inhibition
5
0
of at least 97% and for selected compounds from the former series (IꢀV) (Table 2). Generally,
the target affinities of the thiazolidinedione derivatives (X = O) were significantly higher than
that of the rhodanines (X = S) (Table 2). Within both compound classes the cyclopentylꢀ
substituted derivatives (9a, 9b) were most potent with an IC ꢀvalue of 6.1 ꢀM (9a, X = S) and
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
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
5
0
2
.5 ꢀM (9b, X = O), respectively.
To confirm that the sequence CapꢀArgꢀLysꢀNleꢀNH is the most suitable one for the peptide
2
hybrids, ten derivatives bearing the same highly affine Nꢀterminal cap moiety were additionally
evaluated (Table 3). For the inverse sequence Capꢀ ꢀArgꢀ ꢀLysꢀ ꢀNleꢀNH (27) a significantly
D
D
D
2
lower target affinity was observed. If the lysine residue is replaced by an ornithine (28) the same
observation was made. High activity could be observed for a substitution of the lysine residue by
an additional arginine (29). However, this derivative did not reach the activity of the derivative
containing a lysine residue (21b). A replacement of the arginine by lysine (31) and an
interchange of the arginine and lysine residues (32) did also not result in high affinities. The
importance of the norleucine residue is confirmed by compound 34, which is less active due to
the absence of the Cꢀterminal norleucine. Nearly all affinity is lost if the lysine residue is
additionally absent (35).
To confirm the paraꢀsubstitution at the aromatic system as the most promising alternative, two
metaꢀsubstituted derivatives were explored (37, 38), which are less active (Table 3). An
alternative connection of the peptide sequence via the nitrogen atom of the rhodanine scaffold in
position 3 bearing a hydrophobic arylidene moiety at position 5 did not result in highly active
compounds (39, 40). The same observation was made when the aromatic connection moiety
between the heterocycle in position 5 and the peptide chain was omitted (41, 42). Derivative 41
9
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 10 of 69
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
is active against thrombin with an IC ꢀvalue of 5.7 ꢀM (K = 3.8 ꢀM). Finally, thiohydantoin
5
0
i
22
moieties as alternative heterocycles for rhodanines and thiazolidinediones were evaluated.
However, the two explored derivatives (43, 44) were unconvincing (Table 3). In summary the
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
results shown in Table 3 underline the superiority of the CapꢀArgꢀLysꢀNleꢀNH sequence and a
2
paraꢀarylidene substituent in position 5 of the rhodanine and thiazolidinedione moieties,
respectively.
Six 5ꢀarylidenerhodanines and the corresponding six 5ꢀarylidenethiazolidinedione cap
molecules had only poor activity against the three assayed proteases, indicating that the
tripeptide moiety is a crucial part of the inhibitors (Table S1).
Studies on Inhibition Mechanisms. Detailed analyses with determination of the inhibition
mechanism were performed for all derivatives with a DEN protease inhibition higher than 97%.
Table 2 provides the K ꢀvalues for a competitive binding mode (enzymeꢀinhibitor complex) and
i
K ’ꢀvalues for an uncompetitive binding mode (enzymeꢀsubstrateꢀinhibitor complex) and the
i
observed inhibition mechanism. K ꢀ and K ’ꢀvalues were determined using Dixonꢀ and Cornishꢀ
i
i
5
2, 53
Bowdenꢀplots, respectively.
of the rhodanineꢀ (X = S) and thiazolidinedioneꢀbased (X = O) peptide hybrids was observed.
The binding of five rhodanine derivatives can be described with K ꢀ and additionally K ’ꢀvalues,
A remarkable difference concerning the inhibition mechanisms
i
i
resulting in a mixed inhibition model. For some of these compounds (21a, 22a, 23a) the K ꢀ and
i
K ’ꢀvalues are approximately in the same range, which argues for a nonꢀcompetitive (K = K ’)
i
i
i
mechanism. Only two rhodanine derivatives with aliphatic substituents bind as competitive
inhibitors, with inferior K ’ꢀvalues above 10 ꢀM. Rhodanineꢀbased peptide hybrids 12a, 24a and
i
25a display an almost uncompetitive inhibition with negligible K ꢀvalues far above 10 ꢀM. In
i
10
ACS Paragon Plus Environment

Page 11 of 69
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
summary, the inhibition mechanism of the rhodanineꢀbased peptide hybrids is largely
uncompetitive. The opposite result was found for the thiazolidinedione derivatives: All analyzed
compounds have K ꢀvalues lower than 4 ꢀM and only three compounds had K ’ꢀvalues lower
i
i
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
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
than 10 ꢀM. These three derivatives (9b, 24b, 25b) are "mixed model" inhibitors with a clear
preference for a competitive inhibition mechanism. The other thiazolidinedioneꢀbased peptide
hybrids can be considered as purely competitive inhibitors. Figure 1 shows the differences
between the two explored compound classes for the phenylethylꢀsubstituted derivatives 26a and
26b. The reasons for these findings are difficult to explain. One may speculate about additional
interactions of the thiocarbonyl function, resulting in formation of a ternary enzyme–substrate–
inhibitor complex. The thiazolidinediones may have the possibility for more polar interactions
using the additional carbonyl function, for example by hydrogen bonds to the enzyme, resulting
in higher affinity.
For the previously reported series of compounds, K ꢀvalues between 4.0 and 14.5 ꢀM with a
i
competitive inhibition mechanism were observed (Table 2), clearly indicating a significant inꢀ
vitro affinity improvement for the thiazolidinedioneꢀbased peptide hybrids.
Additional kinetic analyses did not indicate timeꢀdependent inhibition modes or instability of
the peptide hybrids over some hours (data not shown). To check our results, the K ꢀvalue for the
i
most active compound 9b was confirmed by an HPLCꢀbased assay (data not shown).
Aprotinin Competition Assay. To exclude a nonspecific binding mode, a fluorescence
quenching assay developed by Bodenreider et al., which can discriminate between specific and
54
nonspecific binders, was used. This assay is based on the intrinsic fluorescence of Trp50 of the
DEN protease, which is located near the active site of the enzyme. An inhibitor that specifically
11
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 12 of 69
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
binds to the protease quenches this fluorescence by FRET if a structural element of the inhibitor
is able to absorb radiation in the range of the tryptophan emission. The fluorescence will be
partially restored if aprotinin, which is a highly affine protease inhibitor without any tryptophan
residues, replaces the other inhibitor from the active site. This was already shown for the
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
3
7
previously reported peptide hybrids. The thiazolidinedioneꢀbased peptide hybrids are perfectly
applicable for this method with an absorption maximum around 330 nm, so that the experiments
were carried out with derivatives 21b and 26b (Figure S1). Both compounds show a
concentration dependent nonlinear decrease of the registered fluorescence. If aprotinin was
additionally added, a significant fraction of the fluorescence was restored, clearly indicating
specific binding of the inhibitors at the active site.
Molecular Modeling. All experiments were carried out using the protease of DEN serotype 2.
No active structure of the serotype 2 protease coꢀcrystallized with an inhibitor is available at
present. An active structure that displays a ‘closed’ conformation of the DEN protease serotype 3
with a coꢀcrystallized tetrapeptidyl aldehyde inhibitor was recently reported by Noble et al. (pdb
1
4
code: 3U1I). We used the 3U1I structure to obtain an idea of how the investigated peptide
hybrids may interact with the target protein, because 3U1I is the only available structure of the
catalytically competent complex NS2BꢀNS3 of DEN protease. Some conformational and
biochemical differences were observed for the proteases of the DEN serotypes, especially
regarding the role of the cofactor domain. However, choosing the serotype 3 structure for
preliminary modeling studies appears to be reasonable, because the active sites are highly
8
, 11, 55, 56
conserved and the substrate specificity is practically identical for all serotypes.
As
representative compound for the thiazolidinedione series shown in Table 1, the most active
12
ACS Paragon Plus Environment

Page 13 of 69
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
derivative 9b (R = cyclopentyl, X = O) was selected. The results are shown in Figure 2. Similar
to the inhibitorꢀenzyme complex in the 3U1I structure, the S and S pockets are occupied by two
1
2
basic side chains of 9b. This correlates with other findings, e.g. for substrate specificity of the
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
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
55, 57, 58
DEN protease.
The lysine residue is placed in the S and the arginine residue in the S
1 2
pocket with the possibility of hydrogen bonds or salt bridges to Asp129 and Asp75, respectively.
The norleucine residue is flexibly arranged near the S ’ pocket, interacting with a hydrophobic
1
surface formed by two valine side chains of the NS3 domain. The Nꢀterminal cap moiety binds
near the S pocket, which is formed by residues of the NS3 protease and the NS2B cofactor. The
3
arylidene moiety acts as a relatively rigid linker for the heterocycle and the hydrophobic
cyclopentyl substituent, to reach a hydrophobic moiety formed by two valine side chains of the
NS3 domain and a methionine and isoleucine residue of the cofactor NS2B. This could explain
the remarkable preference for hydrophobic substituents connected to position 3 of the
heterocycle. One may speculate about a stabilization of the ‘closed’ conformation by these
interactions of the Nꢀterminal cap moiety with the hydrophobic residues of the protease and
cofactor domains. In summary, the docking results provide some interesting information about
possible interactions of the peptide hybrids in the competitive binding mode.
Permeability. An often discussed restriction of peptideꢀbased inhibitors is their lack of
5
9
membrane permeability. Nevertheless, membrane permeability is a basic requirement for DEN
protease inhibitors, and should be taken into account for future drug candidates. Therefore, we
estimated the passive membrane permeability behavior of the investigated compounds by an
easily applicable, inꢀvitro permeability model. The parallel artificial membrane permeability
assay (PAMPA) is an established and reliable method for predicting the passive cell permeability
13
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 14 of 69
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
60
of compounds. For our studies we used advanced preꢀcoated trilayer (lipid/oil/lipid) artificial
membranes. These have been described to yield permeability data in good correlation with
6
1
results from cellꢀbased Cacoꢀ2 permeability assays. Table 4 provides the PAMPA results for 24
analyzed new peptide hybrids and five compounds from the former series. Five drugs
(famotidine, amiloride, furosemide, phenytoin and caffeine) with known permeation behavior
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
61
were used as reference compounds. For some of the new derivatives, high mass retentions in
the range of 30% were found, indicating high surface adsorptions and accumulation within the
artificial membrane (Table 4). The best permeability could be observed for rhodanine derivative
–6
9
a (R = cyclopentyl, X = S) with a P ꢀvalue (permeability in 10 cm/s) of 2.31. By way of
e
comparison, for caffeine and phenytoin as compounds with high membrane permeability and
gastrointestinal absorption P ꢀvalues of 12.9 and 10.3 were found, respectively. Other
e
compounds with very high membrane permeability, like ibuprofen, dilitiazem or imipramine
6
1
have P ꢀvalues in the same range. This demonstrates that a value of 2.31 can be considered as a
e
good starting point for peptideꢀbased molecules, indicating the ability of some of the compounds
to pass biological membranes and enter cells. The substituent at the heterocycle in position 3 (R)
appears to have a particular influence on the permeation behavior, as compounds bearing
hydrophobic aliphatic substituents in this position have a higher permeability. The rhodanineꢀ
based peptide hybrids have a higher permeability than the thiazolidinedioneꢀbased compounds.
However, two of the analyzed rhodanine derivatives bearing the nonpolar chlorobenzyl (25a)
and phenylethyl (26a) substituents were insoluble in PBS buffer at the required concentration of
2
00 ꢀM. A fineꢀtuning of hydrophobicity with respect to the two endꢀpoints permeability and
solubility is therefore necessary.
14
ACS Paragon Plus Environment

Page 15 of 69
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Antiviral Activity in Cell-Culture and Cytotoxicity. A set of 26 compounds from Table 1
and Table 3 with high, moderate and low affinity towards the DEN protease and different
molecular structures was additionally screened in an infectionꢀbased cell culture model that
utilizes a DEN serotype 2ꢀderived reporter virus genome and the Huhꢀ7 human hepatoma cell
line. The latter stably expresses a firefly luciferase reporter gene for accurate quantification of
cytotoxicity. Because no cytotoxic effects were observed in initial experiments at concentrations
below 50 ꢀM, 18 compounds (including five from the previously reported series) were selected
for additional experiments at higher concentrations (Figure 3). The results of the virus reporter
gene assay are summarized in Table 5. Antiviral activity was observed in a promising
concentration range: The best antiviral activity in cellꢀculture was found for the rhodanineꢀbased
peptide hybrid 10a, bearing a cyclohexyl moiety at the heterocycle, with an EC ꢀvalue of 16.7
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
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
5
0
ꢀM. This derivative is also nonꢀtoxic at the highest dengue reporter gene assay concentration of
50 ꢀM (Figure 3). High activity was also observed for the corresponding phenylethylꢀsubstituted
derivative 26a with EC = 25.6 ꢀM (nonꢀtoxic at 50 ꢀM). In correlation with the previously
50
discussed results (permeability, inꢀvitro activity), compounds with hydrophobic substituents have
higher activity. Consequently, the protonꢀ (1a, 1b), methylꢀ (2a) and ethanolylꢀsubstituted (14a,
14b) derivatives possess no relevant antiviral activity (EC >50 ꢀM). A tendency for higher
5
0
antiviral activities (and cytotoxicity at higher concentrations) could be observed for the
rhodanine derivatives compared to the thiazolidinedione derivatives, although the rhodanines are
less active against the isolated protease. An explanation for this result might be the higher
hydrophobicity and better permeability (in correlation with the PAMPA results) and the observed
additional mode of intermolecular interaction by an uncompetitive inhibition mechanism
(enzyme–substrate–inhibitor complex).
15
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 16 of 69
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
Surprisingly, the ornithine congener 28, which is less active and less permeable than the
corresponding lysine derivative 21b, has significantly higher antiviralꢀactivity (26.8 ꢀM vs. >50
ꢀM). A similar observation is made for compounds 41 and 42. These two derivatives show
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
nearly no inhibition of the DEN protease in vitro, but possess antiviral activity. Because of these
unexpected findings, we considered an interference of the inhibitors with the luciferase activity,
which could lead to falseꢀpositive results. This possibility was studied as follows: Both
luciferases were stably expressed in cell lines that were treated for a short period with the
compounds. Afterwards the activities of the renilla and firefly luciferases were analyzed
seperately. The results are provided in Figure S2. An interference with both luciferases was
observed for compound 42, rendering the results of the cytotoxicity and dengue virus reporter
gene assays nonꢀsignificant (Table 5, Figure 3). Other compounds display only minor
interferences (Figure S2).
To further confirm our results, six compounds (50 ꢀM) were additionally analyzed in a dengue
virus reduction assay that is not based on luciferase activity (plaque assay). The results are given
in Figure 4 and indicate no significant reduction of the viral titer for the two compounds II and
III from the previous series and for the thiazolidinedioneꢀbased peptide hybrids 25b and 28. In
contrast, the two rhodanineꢀbased peptide hybrids 10a and 25a caused a significant decrease of
viral titers. Compound 25a is somewhat cytotoxic at the relevant concentration of 50 ꢀM (Figure
3
), which might contribute to the more pronounced reduction of viral replication compared to
10a. The latter compound has a cytotoxic effect at concentrations above 100 ꢀM and causes a
/3 reduction of viral replication at a concentration of 50 ꢀM. These results indicate an
advantageous antiviral/cytotoxic activity ratio for rhodanineꢀbased peptide hybrids bearing
2
16
ACS Paragon Plus Environment

Page 17 of 69
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
hydrophobic aliphatic moieties (e.g. 10a) in comparison to benzyl substituted analogs such as
25a.
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
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
CONCLUSIONS
A promising new generation of peptide hybrids as DEN protease inhibitors was explored and
extensively analyzed towards their possible biochemical and biological potential of interaction. It
was confirmed that two basic side chains (arginine and lysine) are sufficient to realize high target
affinity and selectivity with K ꢀvalues below 2 ꢀM. It is also evident that reactive, electrophilic
i
moieties which form covalent bonds to the catalytic serine are not essential to obtain high
affinities. From the structureꢀactivity relationships of our previous and current works the
following consequences for DEN protease peptide hybrid inhibitors can be drawn: The
connection between the thiazolidinone heterocycle and the peptide should be accomplished by
relatively rigid arylidene moieties using paraꢀsubstitution. The CapꢀArgꢀLysꢀNleꢀNH motif
2
appears to be optimal for high target affinity. The character of the terminal substituent at the
nitrogen atom of the thiazolidinone heterocycle in position 3 is very important for the target
affinity, the permeability and the antiviral activity in cellꢀculture. More hydrophobic substituents,
with a tendency towards aliphatic groups, have increased membrane permeability and higher inꢀ
vitro and inꢀcellulo activities. The oxygen–vs.–sulfur exchange in position 2 of the 4ꢀ
thiazolidinone scaffold has a significant influence on various properties of the compounds: The
rhodanine (X = S) derivatives are somewhat less potent at the isolated enzyme (IC never below
5
0
6
ꢀM), show a mixed inhibition mode with significant uncompetitive contribution, are slightly
better permeable and possess higher activity in cell culture. The thiazolidinedione (X = O)
17
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 18 of 69
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
derivatives are predominantly substrateꢀcompetitive inhibitors with higher inꢀvitro affinities (Ki
up to 1.5 ꢀM) but inferior permeation and antiviral activity. A striking observation is the fact that
minor changes in the structure of the inhibitor, particularly the sulfurꢀoxygen exchange, cause
considerably changes in the biochemical interaction profile, by changing the binding mode from
competitive to uncompetitive. Future work will aim to increase the target affinity further,
accompanied by modifications of the peptide backbone and basic side chains to improve the
pharmacokinetic properties of the compounds. It may also be considered to specifically optimize
the compounds towards an uncompetitive or nonꢀcompetitive binding mode, given the higher
antiviral activities of the respective inhibitors described here.
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
EXPERIMENTAL SECTION
Expression and Purification of the Viral Proteases. The DEN (serotype 2) and WNV
5
8, 62
NS2BꢀNS3 protease constructs were described before.
Transformation of the pET28a
plasmid (Novagen, Germany), expression in E. coli BL 21 λ (DE3) cells and purification by
2
1, 63
nickel affinity chromatography was done according to the protocol described by Steuer et al.
For both viral enzymes the cofactor is covalently connected to the protease domain by a glycineꢀ
serine linker (GGGGSGGGG).
Dengue and West Nile Virus Protease Assays. The DEN and WNV protease assays were
2
0, 37, 51, 63
performed as described previously.
In short, continuous enzymatic assays were
performed in black 96 well Vꢀbottom plates (Greiner BioꢀOne, Germany) and monitored using a
BMG Labtech Fluostar OPTIMA microtiter fluorescence plate reader at an excitation
wavelength of 320 nm, and a monitored emission wavelength of 405 nm. The inhibitors (final
18
ACS Paragon Plus Environment

Page 19 of 69
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
concentration 50 ꢀM, from 10 mM stock solutions in DMSO) were preincubated for 15 min with
the DEN protease (100 nM) or WNV protease (150 nM), respectively. Afterwards, the reaction
was initiated by the addition of the substrate to a final concentration of 50 ꢁM. FRET substrates
AbzꢀNleꢀLysꢀArgꢀArgꢀSerꢀ3ꢀ(NO )Tyr and AbzꢀGlyꢀLeuꢀLysꢀArgꢀGlyꢀGlyꢀ3ꢀ(NO )Tyr were
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
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
2
2
used for DEN and WNV protease, respectively. The activity of the enzyme was determined as
the slope per second (RFU/s) and monitored for 15 min. Determinations of percentage inhibition
were calculated in relation to a positive control (without inhibitor) and performed in triplicate.
For the HPLCꢀbased DEN protease assay 10 ꢀl of 4% TFA were added to every well after 15
min of enzymatic reaction, and the samples were analyzed by a Jasco HPLC system with an FPꢀ
2020 plus fluorescence detector (excitation: 320 nm, emission: 405 nm) with an RPꢀ18 column
Phenomenex Luna C18(2) (5 ꢀm, 150 x 3 mm) using the following conditions: flow rate: 1.0
ml/min, eluent A: water (0.1% TFA), eluent B: acetonitrile (0.1% TFA), gradient: 10% B (1
5
1
min), 55% B (9.5 min), 95% B (9.6 min), 10% B (12.6 min), 10% B (15 min).
Thrombin Assay. The thrombin assay was performed as a continuous fluorimetric assay in
black 96 well Vꢀbottom plates (Greiner BioꢀOne, Germany), using a BMG Labtech Fluostar
OPTIMA microtiter fluorescence plate reader and operating at an excitation wavelength of
355 nm and an emission wavelength of 460 nm. The inhibitors (final concentration 25 ꢀM, from
10 mM stock solutions in DMSO) were preincubated with thrombin (10 nM) for 15 min. The
cleavage reaction was initiated by addition of the BocꢀValꢀProꢀArgꢀAMC substrate (Bachem,
Germany) at a final concentration of 50 ꢁM. The assay buffer was used according to the
6
4
literature as 50 mM TrisꢀHCl pH 7.5, 150 mM NaCl and 0.05% Tween 20. The activity of
thrombin was determined as the slope per second (RFU/s) and monitored for 10 min.
19
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 20 of 69
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
Determinations of percentage inhibition were calculated in relation to a positive control (without
inhibitor) and performed in triplicate.
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Determination of K -, K ’- and IC -values. The assay conditions were generally used as
i
i
50
described before. The assays were performed using different inhibitor (0, 0.5, 1, 2, 3, 4, 5, 6 ꢀM)
and substrate concentrations (50, 100, 150, 200 ꢀM) in triplicate. Compounds from the former
series of peptide hybrids were assayed at concentrations of 5, 10, 15, 20, 30, 40 and 50 ꢀM.
Observed values were adjusted using correction factors due to the inner filter effect of the FRET
5
1, 65
substrate.
IC ꢀvalues were calculated for a substrate concentration of 50 ꢀM using Prism 5.0
5
0
(
GraphPad Software, Inc.). For K ꢀvalue determination data were plotted 1/v (reciprocal
i
enzymatic activity) against inhibitor concentration (Dixon plot) for the four different measured
substrate concentrations. For K ’ꢀvalue determination the data were plotted S/v (substrate
i
concentration divided by enzymatic activity) against inhibitor concentration (CornishꢀBowden
5
2, 53
plot) for the four measured substrate concentrations.
Aprotinin Competition Assay. This tryptophan quenching assay was performed as described
5
4
by Bodenreider et al. before. In short, the DEN protease (4 ꢀM) was incubated with inhibitors
63
at different concentrations (0, 5, 10, 20, 30, 50 ꢀM) using the assay buffer described before.
Additionally, the protease was incubated with the inhibitor (50 ꢀM) and aprotinin (10 ꢀM)
together. As negative control, the autofluorescence of the enzyme and aprotinin without inhibitor
at the same concentrations was additionally determined. The fluorescence of the assay samples
was monitored on a Tecan Safire II instrument using an excitation wavelength of 280 nm and an
emission wavelength of 330 nm. All determinations were performed in triplicate.
20
ACS Paragon Plus Environment

Page 21 of 69
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Docking. The calculations were performed on an Intel(R) Core(TM)2 Quad CPU Q9450 @
6
6
2
.66 GHz running open SuSE 11.0, using GOLD 5.1 and its graphical interface Hermes 1.4.
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
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
The crystal structure of the NS2BꢀNS3 protease of DEN serotype 3 in complex with a
1
4
tetrapeptidyl aldehyde was extracted from the dimer structure 3U1I. All waters and sulfate ions
6
7
were removed and the structure was further prepared using Dock Prep (Chimera). All
hydrogens were added and the tetrapeptidyl aldehyde ligand was extracted from the structure.
The binding site was defined in a radius of 10 Å around the extracted ligand and automatic
cavity detection was used. Ten different solutions were calculated for every ligand and the
6
7
docking results with the highest GOLD scores were aligned and visualized using Chimera.
Parallel Artificial Membrane Permeability Assay (PAMPA). Concentration determinations
were carried out on a Tecan Safire II instrument using 96 well Uꢀbottom polystyrene plates
(Greiner BioꢀOne, Germany) and alternatively a Jasco HPLC system with a single wavelength
UV detector and an RPꢀ18 column (ReproSilꢀPurꢀODS, Dr. Maisch GmbH, Germany, 3 ꢁm, 50
x 2 mm), using acetonitrile (0.1% TFA) and water (0.1% TFA) with a standard linear gradient.
For all experiments phosphate buffered saline (PBS), which was purchased from SigmaꢀAldrich
(Germany), was used, resulting in 0.01 M phosphate buffer, 0.0027 M potassium chloride and
0.137 M sodium chloride at pH 7.4. Calibration curves were generated for all compounds
including the references (famotidine, amiloride, furosemide, phenytoin, caffeine) at 10, 25, 50,
100, 150, 200 ꢀM using the HPLCꢀUV system (detection at 254 nm) and the UV plate reader (at
330 nm and 390 nm), respectively. 254 nm (HPLC) was used for famotidine, amiloride,
furosemide, phenytoin, caffeine, 41 and 42, I, II, IV and V, 330 nm for compounds 7a, 9a, 10a,
21
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 22 of 69
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
7
b, 9b, 10b, 15b, 17b, 24b, 25b, 26b, 27, 28, 38 and III, and 390 nm for 12a, 13a, 18a, 20a, 23a
2
and 43, respectively. All correlation coefficients (R ) were found to be at least 0.99 for the
calibration curves. The preꢀcoated PAMPA plate system was purchased from BD Bioscience
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
(Germany) and stored at ꢀ20 °C till use. Before use, the plates were warmed to room temperature
for 1 h. 300 ꢀl of the 200 ꢀM compound solutions in PBS (final concentration of DMSO: 2%)
were dispensed in triplicate into the donor plate. 200 ꢀl of PBS buffer were added to all wells of
the acceptor plate which was then carefully placed on the donor plate. The plate system was
covered and incubated at room temperature for 5 h. Afterwards the plates were separated and the
solutions of donor and acceptor plates were transferred into 96 well polystyrene Uꢀbottom plates
(Greiner BioꢀOne, Germany) and analyzed by UV absorption using the plate reader and the
HPLC system. The concentrations of the compounds in donor and acceptor plates were
determined using the previously generated calibration curves. Calculations of the permeability
6
1
(Pe) and the mass retention (R) were done according to the literature.
Dengue Replication and Cytotoxicity Assay in Huh-7 Cell Culture with Renilla and
Firefly Luciferase Reporter Genes. The inhibitors were reconstituted as 10 mM stock solutions
in DMSO, aliquoted and stored at ꢀ20 °C until use. For testing at higher concentrations (100, 200
and 400ꢁM), the concentration of the stock solutions were 40 mM in DMSO.The stock solutions
were freshly diluted before each experiment in DMEM (Dulbecco’s Modified Minimal Essential
Medium (GIBCO, Invitrogen) containing 2 mM
nonessential amino acids (GIBCO, Invitrogen), 100 ꢀg/ml penicillin (GIBCO, Invitrogen), 100
g/ml streptomycin (GIBCO, Invitrogen) and 10% fetal calf serum (heat inactivated at 56 °C for
0 minutes (GIBꢀCO, Invitrogen)) in twoꢀfold dilution series to prepare solutions of adequate
Lꢀglutamine (GIBCO, Invitrogen), 1x
ꢀ
3
22
ACS Paragon Plus Environment

Page 23 of 69
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
concentration, which were used later as 4x stock solutions. Huhꢀ7 cells stably expressing a firefly
luciferase reporter gene were seeded at 5000 cells/well into white clearꢀbottom 96 well plates
6
8
(
Greiner BioOne) in a volume of 100 ꢀl DMEM. To reduce the effects due to medium
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
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
evaporation, the outer row of wells were filled with medium but were not used for the
experiment. After overnight incubation of the cells in a 37 °C incubator, 50 ꢀl of 4x compound
stocks in DMEM were added to the cells at various final concentrations (50, 25, 12.5, 3.13, 1.56
ꢀM for the virus replication assay and additionally 100, 200, 400 ꢀM for the determination of
cytotoxicity). Four hours later, additional 50 ꢀl DMEM or DMEM containing dengue 2 renilla
6
8
luciferase reporter virus at a multiplicity of infection of 10 were added to the cells. The
cytotoxicity was measured using compound treated cells which were uninfected but incubated
with the compound for similar time as infected cells. The cells were incubated at 37 °C for
further 48 h and harvested by removing the media, washing 1x with PBS and adding 50 ꢀl of
luciferase lysis buffer (1% Triton Xꢀ100, 25 mM glycylglycine, 15 mM magnesium sulfate, 4
mM EGTA and 1 mM DTT, pH 7.8) and kept frozen at ꢀ20 °C until measurement.
The luciferase reporter activity was measured using a Mithras LB940 plate luminometer
(Berthold Technologies). The plates were thawed and kept at 4° C until measurement. The
cytotoxicity was measured by firefly reporter assay by adding 75 ꢁl assay buffer (25 mM
glycylglycine, 15 mM magnesium sulfate, 4 mM EGTA, 1 mM DTT, 2 mM ATP and 15 mM
potassium phosphate, pH 7.8), containing 70 ꢁM luciferin to each well and measuring the
luminescence for one second/well in the plate luminometer. Virus replication was estimated by
renilla luciferase reporter activity by the addition of 50 ꢀl assay buffer (25 mM glycylglycine, 15
mM magnesium sulfate, 4 mM EGTA and 15 mM potassium phosphate, pH 7.8), containing 5
23
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 24 of 69
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
ꢀ
g/ml coelenterazine (PJK) and the luminescence was measured for one second/well in the plate
luminometer.
For each inhibitor concentration, the viral replication was measured in at least four
independent wells and the replication was expressed as percentage in relation to the control
(DMSOꢀtreated) cells. The cytotoxicity for each concentration was measured as the mean of
three independent wells and expressed as a percentage of the mean of the DMSO control. The
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
EC ꢀvalues were estimated using Prism 5.0 (GraphPad Software, Inc.), using nonꢀlinear
50
regression analysis. Compounds with EC ꢀvalues higher than the highest concentration tested
50
(50 ꢀM) were reported as >50 ꢀM.
Virus Reduction Assay (Plaque Assay). Huhꢀ7 cells were seeded into 12ꢀwell plates (1.5 x
5
10 cells/well) in 1 ml complete DMEM. After overnight incubation of the cells at 37 °C,
medium was replaced with DMEM containing appropriate concentration of the assayed
compound. Four hours later the cells were infected for one hour with dengue reporter virus (MOI
of 5) in the presence of the compound in a volume of 200 ꢁl. Later the medium was replaced
with DMEM containing appropriate concentration of the test compound and 48 h later the
supernatant was harvested, filtered through a 0.4 ꢁm filter and stored at ꢀ80 °C. The virus titer
was determined by plaque assay on Vero cells. The values are expressed as the mean of three
independent experiments with standard deviation.
Chemistry. All chemicals for the precursor synthesis were obtained from SigmaꢀAldrich
(Germany) and Alfa Aesar, Johnson Matthey (Germany) and were of analytical grade. No further
purification steps were performed unless indicated. All solvents were used as obtained from the
24
ACS Paragon Plus Environment

Page 25 of 69
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
commercial sources. The protected amino acids were purchased from Orpegen (Germany),
Novabiochem, Merck (Germany) and SigmaꢀAldrich (Germany). HBTU and Rink amide resin
(capacity 0.63 mmol/g) were purchased from Iris Biotech (Germany). NMR spectra were
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
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
recorded on Varian NMR instruments at 300 or 500 MHz, 300 K in CDCl , acetoneꢀd or
3
6
DMSOꢀd . Chemical shifts (δ) are given in parts per million (ppm). Residuals of nonꢀdeuterated
6
solvents were used as internal standard. Coupling constants (J) are given in Hertz (Hz). Mass
spectra were measured on Finnigan MAT 8200 (EI) and Bruker micrOTOFꢀQ II (HRꢀESI)
instruments. Combustion elemental analysis was performed by double determination using a
Foss Heraeus Vario EL analyzer. Flash chromatography was performed on a Biotage Isolera One
purification system using silica gel (0.060ꢀ0.200 mm) cartridges (KPꢀSil) 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 LCꢀMS using an Agilent 1200 HPLC system with a MWD
detector combined with a Bruker micrOTOFꢀQ II instrument on an RPꢀ18 column (ReproSilꢀ
PurꢀODSꢀ3, Dr. Maisch GmbH, Germany, 3 ꢀm, 50 x 2 mm).
Procedure A: Synthesis of Peptide Hybrids and Peptidic Assay Substrates. Peptidic DEN
and WNV protease assay substrates and Nꢀterminally capped peptide hybrids were synthesized
37
as described before. In short, the Nꢀterminal cap (3 equivalents) as unprotected carboxylic acid
was coupled to the desired peptide sequence using the Fmoc protocol. After cleavage from the
resin and precipitation in cold diethyl ether, all crude peptides were purified by preparative
HPLC on an ÄKTA Purifier, GE Healthcare (Germany), with an RPꢀ18 pre and main column
25
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 26 of 69
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
(
Rephospher, Dr. Maisch GmbH, Germany, C18ꢀDE, 5 ꢁm, 30 x 16 mm and 120 x 16 mm).
After freezeꢀdrying, the peptides were characterized by HRꢀESI and purity was determined by
LCꢀMS. All evaluated peptide hybrids were obtained with a purity of at least 95% unless
indicated otherwise.
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Procedure B: Synthesis of 3-Alkyl- and 3-Benzylrhodanines with Exemplary Compound
Data. A suspension of amine (10 mmol), sodium hydroxide (22 mmol) and carbon disulfide (11
mmol) in water or ethanol (10 ml) was reacted in a microwave reactor for 5 minutes at 100 °C.
After automated cooling to 40 °C, chloroacetic acid (11 mmol) was added and the mixture was
reacted again at 100 °C for 5 min. After cooling (40 °C), concentrated hydrochloric acid (3 ml)
was added and the reaction was finished at 120 °C for 30 minutes. The crude product was
extracted with ethyl acetate (if ethanol was used as solvent, water was added before extraction)
and purified by flash chromatography (cyclohexane/ethyl acetate). When glycine or
phenylalanine were used as amine component, the first reaction step was carried out at room
temperature for 16 h, followed by the second step for additional 3 h and a final refluxing in
aqueous hydrochloric acid for another 16 h. After extraction these two compounds were used as
crude products without further purification.
1
3
-Cyclopentyl-2-thioxo-1,3-thiazolidin-4-one. Colorless solid (59% yield). HꢀNMR (300
MHz, CDCl ): δ = 1.53–1.67 (m, 2H), 1.80–2.00 (m, 4H), 2.03–2.17 (m, 2H), 3.81 (s, 2H), 5.31
3
13
(quint, J = 8.7 Hz, 1H) ppm; CꢀNMR (75 MHz, CDCl ): δ = 25.5, 27.5, 34.2, 57.8, 173.8,
3
+
2
02.5 ppm; MS (EI, 70 eV): m/z (%): 201.0 (76) [M ]; Anal. Calcd for C H NOS : C, 47.73; H,
8
11
2
5
.51; N, 6.96. Found: C, 47.73; H, 5.58; N, 6.97.
26
ACS Paragon Plus Environment

Page 27 of 69
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
1
3
-(4-Chlorobenzyl)-2-thioxo-1,3-thiazolidin-4-one. Pale yellow solid (80% yield). HꢀNMR
13
(
300 MHz, CDCl ): δ = 3.98 (s, 2H), 5.13 (s, 2H), 7.27 (m, 2H), 7.38 (m, 2H) ppm; CꢀNMR
3
(
75 MHz, CDCl ): δ = 35.6, 46.9, 128.7, 130.6, 133.1, 134.2, 173.7, 200.8 ppm; MS (EI, 70 eV):
3
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
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
+
m/z (%): 256.8 (85) [M ]; Anal. Calcd for C H ClNOS : C, 46.60; H, 3.13; N, 5.43. Found: C,
1
0
8
2
46.48; H, 3.43; N, 5.23.
-(2-Phenylethyl)-2-thioxo-1,3-thiazolidin-4-one. Pale yellow solid (65% yield). HꢀNMR
1
3
13
(
300 MHz, CDCl ): δ = 2.93 (m, 2H), 3.92 (s, 2H), 4.19 (m, 2H), 7.20–7.33 (m, 5H) ppm; Cꢀ
3
NMR (75 MHz, CDCl ): δ = 32.6, 35.2, 45.7, 126.8, 128.5, 128.9, 137.3, 173.4, 200.9 ppm; MS
3
+
(
EI, 70 eV): m/z (%): 236.9 (76) [M ]; Anal. Calcd for C H NOS : C, 55.67; H, 4.67; N, 5.90.
1
1
11
2
Found: C, 55.67; H, 4.82; N, 5.88.
Procedure C: Synthesis of 3-Arylrhodanines with Exemplary Compound Data. A
suspension of amine (5 mmol) and bis(carboxymethyl)trithiocarbonate (5.5 mmol) in water was
reacted in a microwave reactor at 160 °C for 15 min. After automated cooling to 40 °C the crude
product was extracted with ethyl acetate and purified by flash chromatography
(cyclohexane/ethyl acetate). For methyl 2ꢀaminohexanoate and ethanolamine as amine
components this procedure was also used, instead of procedure B.
1
3
-Phenyl-2-thioxo-1,3-thiazolidin-4-one. Pale yellow solid (32% yield). HꢀNMR (300 MHz,
13
CDCl ): δ = 4.18 (s, 2H), 7.19 (m, 2H), 7.52 (m, 3H) ppm; CꢀNMR (75 MHz, CDCl ): δ =
3
3
+
3
6.3, 128.3, 129.6, 129.8, 134.9, 173.4, 201.1 ppm; MS (EI, 70 eV): m/z (%): 208.9 (100) [M ];
Anal. Calcd for C H NOS : C, 51.65; H, 3.37; N, 6.69. Found: C, 50.51; H, 3.42; N, 6.33.
9
7
2
1
3
-(4-Methoxyphenyl)-2-thioxo-1,3-thiazolidin-4-one. Pale yellow solid (39% yield). Hꢀ
13
NMR (300 MHz, CDCl ): δ = 3.84 (s, 3H), 4.16 (s, 2H), 7.02 (m, 2H), 7.10 (m, 2H) ppm; Cꢀ
3
27
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 28 of 69
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
NMR (75 MHz, CDCl ): δ = 36.2, 55.5, 114.9, 127.2, 129.4, 160.3, 173.5, 201.5 ppm; MS (EI,
3
+
7
0 eV): m/z (%): 238.9 (100) [M ]; Anal. Calcd for C H NO S : C, 50.19; H, 3.79; N, 5.85.
10
9
2 2
Found: C, 50.24; H, 3.73; N, 5.82.
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
3
-(2-Hydroxyethyl)-2-thioxo-1,3-thiazolidin-4-one. Pale yellow oil (17% yield). HꢀNMR
13
(300 MHz, CDCl ): δ = 3.90 (t, J = 5.5 Hz, 2H), 4.01 (s, 2H), 4.25 (t, J = 5.5 Hz, 2H) ppm; Cꢀ
3
NMR (75 MHz, CDCl ): δ = 35.4, 46.6, 59.9, 174.6, 201.9 ppm; MS (EI, 70 eV): m/z (%): 177.0
3
+
(
60) [M ]; Anal. Calcd for C H NO S : C, 33.88; H, 3.98; N, 7.90. Found: C, 33.29; H, 4.12; N,
5
7
2 2
7.20.
Procedure D: Synthesis of 3-Alkyl- and 3-Benzylthiazolidine-2,4-diones and Precursor
with Exemplary Compound Data. To a refluxing solution of thiazolidineꢀ2,4ꢀdione (43 mmol)
in ethanol (25 ml) was added a hot solution of potassium hydroxide (45 mmol) in ethanol (25
ml). After additional refluxing for 30 min the mixture was cooled to 0 °C and the precipitate was
filtered and washed with cold ethanol. The obtained potassium 2,4ꢀdioxothiazolidinꢀ3ꢀide (5
mmol) was refluxed with alkyl or benzyl halides (5.5 mmol) in DMF (15 ml) for 3ꢀ4 h. After
cooling to room temperature and addition of water (40 ml) the crude product was extracted with
ethyl acetate, washed with brine and purified by flash chromatography (cyclohexane/ethyl
acetate).
1
3-Methylthiazolidine-2,4-dione. Pale yellow solid (89% yield). H NMR (300 MHz, CDCl ):
3
+
δ = 3.12 (s, 3H), 3.95 (s, 2H) ppm; HRMS (ESI): m/z [M+Na] calcd for C H NNaO S:
4
5
2
153.9933, found: 153.9943.
28
ACS Paragon Plus Environment

Page 29 of 69
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
1
3
-Butylthiazolidine-2,4-dione. Pale yellow oil (71% yield). HꢀNMR (300 MHz, CDCl ): δ =
3
0
.93 (t, J = 7.3 Hz, 3H), 1.33 (m, 2H), 1.58 (m, 2H), 3.62 (m, 2H), 3.93 (s, 2H) ppm; HRMS
+
(
ESI): m/z [M+Na] calcd for C H NNaO S: 196.0403, found: 196.0428.
7
11
2
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
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
1
3
-(4-Methylbenzyl)thiazolidine-2,4-dione. Pale yellow oil (88% yield). HꢀNMR (300 MHz,
CDCl ): δ = 2.32 (s, 3H), 3.92 (s, 2H), 4.73 (s, 2H), 7.13 (m, 2H), 7.30 (m, 2H) ppm; HRMS
3
+
(
ESI): m/z [M+H] calcd for C H NO S: 222.0583, found: 222.0583.
1
1
12
2
Procedure E: Synthesis of 3-(4-Methylphenyl)thiazolidine-2,4-dione. The corresponding 3ꢀ
(
4ꢀmethylphenyl)rhodanine (2.5 mmol) was dissolved in ethyl acetate, followed by the addition
of sodium tungstate (0.25 mmol) and hydrogen peroxide (25 ml, 30% in water). After stirring at
room temperature for 15 h, the organic layer was washed two times with brine and a solution of
sodium sulfite. After removal of the solvent the crude product was purified by flash
1
chromatography (cyclohexane/ethyl acetate) to get a colorless solid (14% yield). HꢀNMR (300
MHz, CDCl ): δ = 2.40 (s, 3H), 4.12 (s, 2H), 7.13 (m, 2H), 7.31 (m, 2H) ppm; HRMS (ESI): m/z
3
+
[
M+Na] calcd for C H NNaO S: 230.0246, found: 230.0289.
1
0
9
2
Procedure F: Synthesis of 3-Benzyl-2-thiohydantoin and 3-Phenyl-2-thiohydantoin. A
mixture of the corresponding benzyl or phenyl isothiocyanate (5 mmol), methyl glycinate
hydrochloride (5 mmol) and triethylamine (5 mmol) in diethyl ether (20 ml) was stirred at room
temperature for 15 h. The resulting precipitate was filtered, washed with diethyl ether and used
as crude product without further purification.
29
ACS Paragon Plus Environment