A Multi-step Virtual Screening Protocol for the Identification of Novel Non-acidic Microsomal Prostaglandin E2 Synthase-1 (mPGES-1) Inhibitors

Article abstract of DOI:10.1002/cmdc.201800701

Microsomal prostaglandin E₂ synthase-1 (mPGES-1) is a potential therapeutic target for the treatment of inflammatory diseases and certain types of cancer. To identify novel scaffolds for mPGES-1 inhibition, we applied a virtual screening (VS) p

Full text of DOI:10.1002/cmdc.201800701

Accepted Article
Title: A multi-step virtual screening protocol for identification of novel
non-acidic microsomal prostaglandin E2 synthase-1 (mPGES-1)
inhibitors
Authors: Suhaib Shekfeh, Burcu Çalışkan, Katrin Fischer, Tansu
Yalçın, Ulrike Garscha, Oliver Werz, and Erden Banoglu
This manuscript has been accepted after peer review and appears as an
Accepted Article online prior to editing, proofing, and formal publication
of the final Version of Record (VoR). This work is currently citable by
using the Digital Object Identifier (DOI) given below. The VoR will be
published online in Early View as soon as possible and may be different
to this Accepted Article as a result of editing. Readers should obtain
the VoR from the journal website shown below when it is published
to ensure accuracy of information. The authors are responsible for the
content of this Accepted Article.
To be cited as: ChemMedChem 10.1002/cmdc.201800701
Link to VoR: http://dx.doi.org/10.1002/cmdc.201800701
A Journal of
www.chemmedchem.org

10.1002/cmdc.201800701
ChemMedChem
FULL PAPER
A multi-step virtual screening protocol for identification of novel
non-acidic microsomal prostaglandin E2 synthase-1 (mPGES-1)
inhibitors
Suhaib Shekfeh,[a] Burcu Çalışkan,[a] Katrin Fischer,[b] Tansu Yalçın,[a] Ulrike Garscha,[b] Oliver Werz,[b]
and Erden Banoglu,*[a]
[a]
Dr. Suhaib Shekfeh; Assoc. Prof. Burcu Çalışkan; Tansu Yalçın; Prof. Erden Banoglu
Department of Pharmaceutical Chemistry
Faculty of Pharmacy, Gazi University
Ankara 06560, Turkey
E-mail: banoglu@gazi.edu.tr
[b]
Katrin Fischer; Dr. Ulrike Garscha; Prof. Oliver Werz
Department of Pharmaceutical/Medicinal Chemistry
Institute of Pharmacy, Friedrich-Schiller-University Jena
Philosophenweg 14, D-07743 Jena, Germany
Supporting information for this article is given via a link at the end of the document.
Abstract: Microsomal prostaglandin E2 synthase-1 (mPGES-1) is
considered as potential therapeutic target for treatment of
inflammatory diseases and certain types of cancer. In order to capture
novel scaffolds for mPGES-1 inhibition, we applied a virtual screening
protocol comprising molecular docking, fingerprints-based clustering
with diversity-based selection, protein-ligand interactions fingerprints,
and molecular dynamics (MD) simulations with Molecular Mechanic–
Poisson Boltzmann Surface Area (MM-PBSA) calculation. The hits
identified were carefully analyzed to ensure the selection of novel
scaffolds, which make stable interactions with key residues in the
mPGES-1 binding pocket and inhibit the catalytic activity of the
enzyme. As a result, we discovered two promising chemotypes 6 and
8 as non-acidic mPGES-1 inhibitors with IC50 of 1.2 and 1.3 μM,
PGI2 (prostacyclin) and thromboxane A2 (TXA2).[1a] The
isomerization of PGH2 to PGE2 is mediated by PGE2 synthases
(PGES), which include three members such as cytosolic PGE2
synthase (cPGES) and two microsomal PGE2 synthases
(mPGES-1 and mPGES-2) (Figure 1). While both cPGES and
mPGES-2 are constitutively expressed in a variety of tissues,
mPGES-1, which is functionally coupled with COX-2, is strongly
up-regulated under inflammatory conditions.[1a] Therefore, it is
believed that selective inhibition of mPGES-1 would selectively
interfere with the biosynthesis of pro-inflammatory PGE2, while
intervention with COX-1/2 by nonsteroidal anti-inflammatory
drugs (NSAIDs, such as indomethacin or ibuprofen) prevents the
production of all PGs. Hence, inhibition of mPGES-1 has been a
development of a pharmacological strategy for next generation
NSAIDs that allow selective interference with pro-inflammatory
PGE2, and is anticipated to overcome the common side effects
(i.e. gastrointestinal and cardiovascular) observed with current
NSAIDs due to suppression of homeostatic prostanoids.
respectively. Minimal structural optimization of
8 resulted in
compounds 15, 19 and 20 with promising improvement in the
inhibitory activity (IC50 = 0.3 - 0.6 μM). The unprecedented chemical
structures of 6 and 8, which are prone to further derivatization reveal
a new and attractive field for the development of mPGES-1 inhibitors
with potential anti-inflammatory and anticancer properties.
Introduction
The microsomal prostaglandin E2 synthase-1 (mPGES-1) has
been the object of extensive research for next generation anti-
inflammatory drugs during the last two decades.[1] mPGES-1 is an
inducible terminal enzyme in the biosynthesis of PGE2, which is
the main mediator of acute and chronic inflammation, fever and
pain (Figure 1).[1a] After its first discovery by Jakobsson and co-
workers,[2] mPGES-1 has rapidly become an attractive target for
pharmacological intervention with inflammation-related diseases,
such as arthritis, atherosclerosis, neuro-degenerative diseases
Figure 1. The Arachidonic acid pathway.
3]
and cancer.[1b, The biosynthesis of PGE2 is initiated by the
liberation of arachidonic acid (AA) from membrane phospholipids
by phospholipases A2. Cyclooxygenases (COX-1/2) then convert
free AA to PGH2, which is the junction point for the biosynthesis
of several structurally related PGs that are formed from PGH2 by
the action of their respective PG synthases. The PGs
biosynthesized by these pathways include PGE2, PGD2, PGF2α,
Figure 2 demonstrates several promising scaffolds that were
developed as mPGES-1 inhibitors during the last decade.[4] For
example, indole-carboxylic acid derivatives, which was evolved
from the structure of the 5-lipoxygenase-activating protein (FLAP)
1
This article is protected by copyright. All rights reserved.

10.1002/cmdc.201800701
ChemMedChem
FULL PAPER
inhibitor MK-886, effectively inhibit mPGES-1 activity in vitro.[5]
structures (4YL0, 4YL1, 4YL3, 4YK5) by investigating their
stability using MD simulations followed by MM-PBSA calculations.
Other potent mPGES-1 inhibitors comprise various chemical
scaffolds
such
as
tri-substituted
phenanthrene-imidazoles,[6]
4-
biarylimidazoles,[7]
urea
derivatives,[8]
imidazoquinolines,[9] dihydropyrimidines,[10] and benzoxazole-
piperidines.[11] Although, these mPGES-1 inhibitors with distinct
chemical classes have been identified, none has been proven
clinically useful so far due to various problems (e.g. strong plasma
protein binding and high lipophilicity) as reviewed elsewhere.[4, 12]
Up to now, only a single clinical trial with an mPGES-1 inhibitor
(i.e. LY3023703 by Eli Lilly) has been reported,[13] while a second
compound from Glenmark Pharmaceuticals, namely GRC-27864,
has recently entered phase I clinical development for the potential
treatment of chronic inflammatory disorders such as osteoarthritis
and rheumatoid arthritis (Clinical Trials Identifier: NCT02179645).
To our knowledge, no structural information regarding both
compounds has yet been disclosed. Therefore, developing new
inhibitors of mPGES-1 with distinct scaffolds is a major challenge
in anti-inflammatory drug development. Based on these
considerations, we hereby report on the identification of new
inhibitors of human mPGES-1 with various scaffolds through a
multi-step virtual screening approach. We also show that these
newly identified chemotypes are prone to further structural
optimization for the development of potent mPGES-1 inhibitors as
potential anti-inflammatory and anticancer agents.
Figure 3. The workflow of the applied VS protocol.
Analysis of protein-ligand interactions in reported crystal
structures of mPGES-1
The recent elucidation of the first crystal structure of mPGES-1 in
an active form (PDB ID: 4AL1)[15] was rapidly followed by
elucidation of further high-resolution structures of human
mPGES-1-inhibitor complexes (PDB ID: 4BMP, 4YK5, 4YL1,
4YL3 and 4YL0)[16] to support the rational design of novel
mPGES-1 inhibitors.[17] With this aim, we retrospectively analyzed
the reliability of a variety of available mPGES-1 crystal structures
for use in our virtual screening study (Figure 4).
R1
R1
R3
R3
Cl
R4
COOH
N
N
N
R1
N
H
N
H
R2
R4
F
R2
Indole-carboxylic acid
Phenanthrenes-imidazoles
2,4-Biaryl-imidazoles
R2
O
N
N
R1
R2
R3
R1
R1
O
N
O
N
N
N
H
N
H
R2
Br
Tri-substituted Urea
Imidazo-quinolines
Benzoxazole-piperidines
Figure 2. The main scaffolds of potent mPGES-1 inhibitors developed during
the last decade.
Figure 4. Important protein-ligand interactions for co-crystallized inhibitors in
the published mPGES-1 crystal structures 4YL0, 4YL1, 4YL3, and 4YK5.
Results and Discussion
Virtual screening (VS) and molecular docking are well-established
computational methods that are appropriate to predict binding
interactions of small molecules with experimental protein
structures, and to retrieve putative small-molecule inhibitors from
large chemical databases.[14] However, some factors remain as
challenging obstacles for obtaining a true prediction of the ligand-
protein binding-affinity such as protein plasticity, water-mediated
binding interactions, presence of structural water molecules in the
active site and solvent entropy. Considering all these factors, we
herein describe a multi-step VS protocol (Figure 3), which
combines molecular docking, protein-ligand interactions profile,
molecular dynamics (MD) simulations, and Molecular Mechanic–
Poisson Boltzmann Surface Area (MM-PBSA) energy calculation
in order to improve the virtual screening accuracy. Our final
objective is to identify novel small-molecule mPGES-1 inhibitors
devoid of cross-activity against COXs. The selection step in our
protocol depended on preserving some key protein-ligand
interactions as noticed in the four mPGES-1-inhibitor crystal
Careful analysis of binding modes of the co-crystallized inhibitors
with mPGES-1 led to the recognition of a binding groove and
important residues participating in the protein-ligand interactions.
For example, X-ray crystal structures with PDB ID: 4YL0 and
4YL3 represent the mPGES-1 structures bound to
a
phenanthrene-imidazole derivative (MF-63) and a brominated
biaryl-imidazole derivative, respectively (see Figure S1A-B for
chemical structures). Both inhibitors bind between the two
mPGES-1 monomers A and B, and more precisely, between two
helical turns of α-4 from monomer A and α-1 from monomer B.[15,
18]
By examining the binding interactions, we mostly noticed
strong hydrogen bonds in the more polar region close to residues
S127 (from monomer A), H53 and R52 (from monomer B), while
aromatic rings extended inside both the deep hydrophobic groove
and binding groove above the co-substrate glutathione (GSH).
Therefore, strong hydrophobic interactions are made with
different residues such as A123 and S127 side chains (monomer
2
This article is protected by copyright. All rights reserved.

10.1002/cmdc.201800701
ChemMedChem
FULL PAPER
A) and R38, L39, F44, D49, H53, and R53 side chains (monomer
B) (Figure 4).
the receptor (Figure 4 and 5).[15-16] Other X-ray crystal structures
(PDB ID: 4YK5 and 4YL1) have captured the binding of two
indole-carboxylic acid derivatives (Figure S1C-D) to mPGES-1,
showing double salt bridges between the inhibitor's carboxylate
groups and the side chain of R52 from monomer B (Figure 4,
Figure S2). In both structures, bridging water assists in mediating
another interaction between the carboxylate and the neighboring
residue H53. The aromatic ring, attached to the indole's nitrogen,
extends in the binding groove above GSH, making hydrophobic
interactions with the side chains of R38, L39, and F44 (from
monomer B) and also with GSH. The fluorinated biaryl extension
goes down the hydrophobic binding pocket making strong
hydrophobic interaction with V128, Y130, and T131 (from
monomer A) and on the other side with Y28 and I32 (from
monomer B) (Figure S2, Figure 4).
In the case of MF63 (Figure S1A) in the crystal structure 4YL0,
one of the imidazole's nitrogens contributes also to a hydrogen
bond with H53, while the other may be involved either with
hydrogen bond to GSH or via a bridge of water molecules
connected with one of the ligand's cyanide groups (Figure 5A).
The planar aromatic tetracycle is involved with multiple
hydrophobic interactions with residues P124, S127 and V128
from monomer A. In case of the brominated biaryl-imidazole
derivative (Figure S1B) in 4YL3, the bi-aryl hydrophobic tail
interacts with P124, S127, V128, T131, L132 and L135 from
monomer A (Figure 5B). In both structures, water molecules
mediate protein-ligand interactions in order to fill distances and
satisfy hydrogen bond's donors/acceptors in both the ligand and
Figure 5. X-ray binding mode analysis of A) the phenanthrene-imidazole derivative (MF-63, Figure S1A) inside the binding pocket of mPGES-1 (PDB ID: 4YL0)
between two monomers; Monomer A is shown in blue, while monomer B is shown in pink; the residue R52 from 4YL0 (in pink) is superimposed on the same residue
from 4YL3 (the side chain in magenta) in order to show the difference of the side-chain conformation. B) The biaryl-imidazol derivative (Figure S1B) inside the
binding pocket of mPGES-1 (PDB ID: 4YL3).
Molecular docking and enrichment studies
between the specific enzyme structure and its respective ligand.
The phenanthrene-imidazole ligand (MF63; Figure S1A) in 4YL0
with its voluminous aromatic tetracycle skeleton forces R52 to
shift 1.15 Å outwards as compared to the same residue in
structures 4YL3 and 4YL1 (Figure 5A). Consequently, docking
programs may fail to dock phenanthrene-imidazole derivatives
correctly to other crystal structures (4YK5, 4YL1, and 4YL3) due
to a steric clash between the phenanthrene-imidazole ring and the
R52, which would prevent the observed binding mode in the
crystal structure 4YL0. Similarly, the indole-carboxylic acid
derivative in structure 4YL1 shows a similar sensitivity for the right
orientation of R52 in order to establish the salt bridge interactions.
As a consequence, the correct binding mode of co-crystallized
inhibitor in 4YL1 (Cmpd. 3 in Figure S1D) in different crystal
structures was not always the top-scored solution but it appeared
We tested the reliability of the Glide docking program
(Schrödinger suite 2016) in order to analyze if the observed
binding modes of inhibitor ligands in mPGES-1 crystal complexes
(PDB ID: 4YK5, 4YL0, 4YL1, and 4YL3) were reproducible by
docking experiments using the respective protein structures.[16] All
crystal ligands were successfully docked inside their respective
crystal structures with low root-mean-square deviation (RMSD; 0
to 0.5 Å) to confirm the applicability of molecular docking
experiments (Table S1 in Supplementary Information). However,
the cross-docking of these four inhibitors in other crystal
structures did not always produce the correct binding modes due
to the distinct conformations of some flexible residues in different
crystal structures. For example, Arg52 (R52) adopts different
orientations in order to make the optimal hydrogen bonds network
3
This article is protected by copyright. All rights reserved.

10.1002/cmdc.201800701
ChemMedChem
FULL PAPER
as the second or third-ranked docking solution (Table S1 in the
Supplementary Information).
Enrichment can be generally assessed by the number of active
compounds detected at a given percent of the decoy set
(presumed non-binding molecules) by score-ranked poses.
Enrichment was calculated at 5%, 10%, and 20% of the ranked
docked decoys by applying Glide SP score for ranking (Table 2).
Another enrichment metrics, which was also estimated for the
enrichment studies, includes the area under the accumulation
curve (AUAC), and the area under the receiver operating
characteristic (ROC) curve. The area under the ROC curve
generally represents the inherent capability of the docking
protocol and the used crystal structure to discriminate between
the actives on one hand, and the inactives and the set of decoy
molecules on the other hand. The results for 4YL0 structure
indicate that 47.6% of the actives are retrieved when 20% of the
decoys are captured (Figure S3). That yields the highest
enrichment metrics for 4YL0 structure having area under the ROC
curve as 0.75, while the area under the ROC curve for structures
4YL3, 4YL1, and 4YK5 is 0.67, 0.64 and 0.63, respectively. This
difference in the performance can be simply explained by the
incorrect docking of the highly active phenanthrene-imidazole
derivatives in both structures 4YL3 and 4YK5 as pointed out
earlier.
Figure 6. The protein-ligand interactions for the VS-hits and compound 8
derivatives docked inside mPGES-1 crystal structure 4YL3.
N
O
O
N
N
N
N
HN
F
N
O
H
N
HN
N
H
N
O
1
2
F
N
N
O
NH
N
HN
N
N N
O
HN
S
O
O
H2N
N
Table 1. Enrichment metrics for glide docking of mPGES-1 inhibitors together
with 1000 decoy compounds from SCHROEDINGER’s decoys set.
O
3
4
6
H
O
N
S
O
Enrichment
5%
10%
20%
AUAC
ROC
O
[1000 Decoys +
Decoys
Decoys
Decoys
H
N
N
N
H
N
H
N
F
N
H
N
N
N
Inactives (IC50
≥
O
5 μM)]
O
Cl
5
4YL0
4YL3
4YK5
4YL1
16.1
14.35
10.45
11.63
31.2
23.7
47.6
33.7
0.72
0.63
0.60
0.61
0.75
0.67
0.63
0.64
O
S
O
O
N
18.34
19.84
28.3
N
O
N
S
NH
H2N
N
N
Cl
N
H
29.58
N
O
O
O
7
8
The virtual screening work-flow
H
N
O
The general applied VS protocol includes several steps that are
summarized in Figure 3. In the first step, a proper chemical library
was generated starting from the MolPort screening library with
adequate geometries, ionization states, conformations, and
tautomers using the LigPrep module in SCHROEDINGER 2016
suite. In the early step, the Lipinski’s rules of five (RO5) were
applied as filter in order to establish a library with compounds
displaying high drug-likeness.[19] Both crystal structures (4YL0
and 4YL3) were used in the VS study in order to take two different
conformations of the R52 side chain into account. In the first
structure-based VS tour, all drug-like compounds retrieved from
MolPort database (about 5 mio compounds) were subjected to
molecular docking to the mPGES-1 binding site using the crystal
structure 4YL0. Later, the top scored compounds (40,000
compounds) were docked again into another crystal structure,
namely 4YL3, in order to consider the two different conformations
of the R52. Then, a first diversity-based selection protocol was
applied to the docking poses that survived using both structural
MACCS- and pharmacophore-based fingerprints application in
SCHROEDINGER’s canvas to obtain 1000 docking complexes
O
N
S
HN
S
N
N
O
O
N
N
N
O
O
F
9
10
Figure 7. Chemical structures of the selected VS hits. Among them, the two
structures 6 and 8 were identified as promising mPGES-1 inhibitors in cell-free
activity assays with IC50 = 1.2 and 1.3 μM respectively.
with chemically diverse scaffolds. In the next step, the open-
source software PLIP (protein-ligand interaction profiler) was
used to process the final 1000 docking complexes for
characterization of the most diverse protein-ligand interaction
pattern.[20] As a result, a subset of 49 hits was selected based on
diversity observed in protein-ligand interaction profiles as well as
in chemical structures. Finally, ten candidates were ‘cherry-picked’
from this subset based on chemical intuition, literature knowledge
considering the geometrical and pharmacophoric features (i.e.,
4
This article is protected by copyright. All rights reserved.

10.1002/cmdc.201800701
ChemMedChem
FULL PAPER
non-acidic features) and similarity to the protein-interaction profile
of the mPGES-1-co-crystallized inhibitors (Figures 6 and 7). All
selected candidates were chosen to have good similarity
(Tanimoto index > 0.75) to the protein-interaction profile of the
mPGES-1-co-crystallized inhibitors, which provide one hydrogen
bond at least to residues of H53, R52 and S127, and at least two
hydrophobic interactions with the set of residues comprising F44,
P124, Y130, L39 or Y28 as presented in Figure 6. Accordingly, to
further investigate the stability of the binding modes of selected
compounds to mPGES-1 considering potential ligand-induced
conformational changes, we performed MD simulations (20 ns)
using the 3D structure of mPGES-1 (PDB ID: 4YL3 in complex
with these docked ligands), followed by MM-PBSA energy
decomposition per residue as will be discussed later.
Figure 8. The suggested binding mode of the identified mPGES-1 inhibitors 6 and 8, as docked inside the putative binding pocket of mPGES-1 (PDB ID: 4YL3).
Hydrogen bonds are shown in yellow, while the pi-pi stacking and hydrophobic interactions are represented by red dashes.
MD simulations: Analysis, MM-PBSA calculation, and energy
decomposition
The decomposition of the binding energy of the mPGES-1/VS-hits
complexes showed a significant contribution of aforementioned
key residues (S127, R126, H53, and R52) in establishing a
hydrogen bonds’ network between the ligand and the protein. The
most recognized hydrophobic residues in the decomposition
analysis appeared to be P124, A123, F44, L39, and Y130. The
list of key residues can be extended to T131, Q134, I32, and Y28.
The binding of 6, as an example, appeared to be mostly driven by
interactions with several hydrophobic residues such as P124, L39,
and I32, and A123 while the electrostatic and polar interactions
were contributed mainly by S127, R52, R126, and Q134. In the
case of 8, the hydrophobic interactions were contributed mainly
by Y130, P124, L39, and I32, while polar interactions were mainly
made with H53, R38, and S127. Moreover, the analysis of MD
snapshots with an explicit solvent model can also show water-
mediated interactions between the studied ligands (from the VS
selected compounds) and some key residues such as R53, H53,
or S127. For both 6 and 8, we observed an optimal orientation of
one aromatic ring close to the residues F44 and L39, while other
aromatic rings interact with other hydrophobic residues, e.g., Y28,
Y130, T131, and I32 (Figure 8).
MD simulations were performed on the selected VS hits in order
to prove the stability of the suggested binding mode and to
examine the dynamic variations of the ligand-protein interactions.
The time series of the RMSD of the docking complexes were
under 2.6 Å over 20 ns of MD simulation, and under 0.5 Å for the
simulated ligands in the putative binding pocket. To further
investigate the protein-ligand interactions over the MD trajectory,
the MM-PBSA method was used to estimate the binding energy
and to perform energy decomposition by residue (using
g_mmpbsa tool)[21], although an efficient algorithm for calculating
the entropy term is still lacking.[22] Additionally, the presence of the
bilayer membrane in the case of membrane proteins would make
the exact description of desolvation and calculation of entropy
term even more complex. Therefore, the application of MM-PBSA
binding energy calculation was not intended to reproduce the
exact binding free energy, but to estimate the energetic
contribution of the key protein residues to the protein-ligand
binding. The calculated free binding energies of the native co-
crystallized ligands in two mPGES-1 crystal structures (4YL0 and
4YL3) as well as three of the selected VS compounds (6, 7, and
8) are listed in Table S3, with the contributions from vdW,
electrostatic interaction, polar desolvation, and apolar desolvation
energy terms using MM-PBSA method.
Biological evaluation and SAR
The mPGES-1 inhibitory activity of the acquired compounds was
assessed using a well-established cell-free assay based on
5
This article is protected by copyright. All rights reserved.

10.1002/cmdc.201800701
ChemMedChem
FULL PAPER
incubation of microsomes from IL-1-activated human A549 cells,
which strongly express mPGES-1, with 20 µM PGH2 as
substrate.[23] The mPGES-1 inhibitor MK886 was used as a
reference compound (IC50 = 1.6 µM, not shown). Among the
tested compounds (1 and 10 μM, final concentration), 6 and 8
effectively inhibited the activity of mPGES-1 while the other
compounds were less efficient (Table 2). More detailed
concentration-response analysis revealed the IC50 values of 1.2 ±
0.2 µM for 6 and 1.3 ± 0.4 µM for 8 (Table 3). Because many
mPGES-1 inhibitors tend to inhibit also COX-1 (including
MK886),[3b, 24] we investigated the two hit compounds for COX-1
inhibition. However, 6 and 8 displayed no inhibitory effectiveness
against COX-1 activity in a cell-based assay using human
platelets up to 10 µM (Figure S4).
(17-20) compounds (Figure 9). We first focused on the
heterocyclic ring occupying the deep hydrophobic binding cavity
and obtained the closely related analogues having benzothiazole
(12) and benzoxazole (14) rings. Results from the biological
assessment of mPGES-1 inhibition by 12 and 14 indicated that 12
with a benzothiazole ring (IC50 = 2.3 µM) was superior to the
benzoxazole (14, IC50 = 7 µM) counterpart (Table 3). Next, we
investigated the position of the 3-acetylamino function and
obtained the 4-acetylamino congeners 15-17 with the hypothesis
of the gain of further favorable contacts with amino acids R52,
H53 and F44 at the upper part of the substrate binding cavity. As
seen from table 4, 15 with a benzothiazole ring as compared to
16 and 17 with benzoxazole and benzimidazole, respectively,
resulted in significant inhibition of mPGES-1 activity (IC50 = 0.6, 5
and >10 µM, respectively). Removal of the acetylamino function
in the benzothiazole series (18) caused a small decrease in the
inhibitory activity (IC50 = 1.4 µM). At this point, in particular, the
presence of the benzothiazole ring guaranteed the most potent
mPGES-1 inhibition, and additionally suggested that the presence
of more appropriate polar aromatic functions at the upper end of
the molecule might significantly contribute to the inhibitory activity
by establishing additional polar and - interactions with amino
acids at the cytosolic entrance (i.e., R52, H53, R126, F44).
Therefore, we searched the MolPort database for closely related
congeners leading to commercially available 19 and 20 with
benzofurazan and benzoxazolone rings, respectively, as a
replacement of the 4-acetylaminophenyl part in compound 15. As
expected, both compounds profoundly inhibited mPGES-1 activity
with IC50 values of 0.6 and 0.3 µM, thereby corroborating our
hypothesis (Table 3).
Table 2. Inhibitory effects of newly identified hit compounds 1 - 10 on PGE2
formation at concentrations of 1 µM and 10 µM using the IL-1-activated A549
cells expressing mPGES-1 and 20 µM PGH2 as substrate.
Remaining mPGES-1 activity[a] in % of control at
Cmpd.
1 M
10 M
1
2
86.7 ± 4.5
88.6 ± 1.2
99.1 ± 4.3
73.4 ± 2.3
90.2 ± 5.1
52.9 ± 1.0
83.4 ± 3.6
54.9 ± 2.0
80.9 ± 3.3
83.0 ± 2.0
63.1 ± 3.0
79.8 ± 2.5
83.8 ± 2.9
67.8 ± 1.9
74.9 ± 9.7
16.2 ± 1.0
60.9 ± 1.5
35.2 ± 1.8
65.5 ± 2.2
77.5 ± 1.2
3
4
5
6
7
8
9
10
[a] Data are given as means  SEM, n=3.
Table 3. In vitro inhibitory activities of newly identified mPGES-1 inhibitors under
cell-free conditions.
Figure 9. Chemical structures of analogues of compound 8.
Cmpd.
IC50 (µM)[a]
1.2 ± 0.2
1.3 ± 0.4
2.3 ± 1.3
7.0 ± 2.1
0.6 ± 0.1
Cmpd.
16
IC50 (µM)
5.0 ± 0.4
>10
Docking studies and molecular dynamic (MD) simulations
To provide further insights into the interaction of 15, 19, and 20
with mPGES-1, we performed docking studies in combination with
MD simulations (20 ns) using the mPGES-1 crystal structure
(PDB code: 4YL3). According to the docking and MD simulations
performed on the inhibitors/protein complexes inside lipid bilayer
(POPC model), we clearly observed stable - interactions
between the benzothiazole aromatic rings of all three compounds
and the residue Y130 from mPGES-1 structure (Figure 10).
Moreover, the enhanced protein-ligand interactions are mainly
noticed inside the catalytic pocket above GSH where 15, 19 and
20 interact with residues H53, D49, R126 or S127 by hydrogen
bonds, while the aromatic rings of 19 and 20 (benzofurazan in 19
and benzoxazolone in 20) establish additional - stacking or
charge- interactions with the residues F44 or R126, respectively.
MD studies of both 19 and 20 offer an overview of stable hydrogen
6
8
17
12
14
15
18
1.4 ± 0.7
0.6 ± 0.2
0.3 ± 0.1
19
20
[a] The IC50 values are given as mean ± SEM of n = 3-4 determinations.
Based on the VS results obtained, we selected 8 for further
structural optimization to deduce preliminary structure-activity
relationships in a small series of synthesized (12, 14-16, see
supporting information for synthesis) or commercially available
6
This article is protected by copyright. All rights reserved.

10.1002/cmdc.201800701
ChemMedChem
FULL PAPER
bond with H53 (for 19) or S127 (for 20), stabilized by one or two
additional hydrogen bonds with R126 (for 19) or D49 (for 20).
Moreover, both compounds were able to maintain stable -
stacking with at least two of the aromatic residues of the binding
pocket such as Y28 and Y130 (for 19), or Y130 and F44 (for 20).
Figure 10. Binding mode analysis of 15 (A), 19 (B) and 20 (C) during interaction with mPGES-1 (PDB code: 4YL3) considering membrane residues. Main interactions
are represented schematically considering MD simulations in the time window 0-20 ns. Hydrogen bonds are represented by yellow dashes, while the pi-pi stacking
interactions are represented by red dashes.
1 activity. Depending on the study of various crystal structures of
mPGES-1, it appears that the binding to the region around
Conclusions
glutathione is quite important in order to block the access of the
substrate PGH2 in which the hit compounds 6 and 8 efficiently
occupy. In addition, starting from 8, we have performed a
preliminary SAR study with a small series of closely related
structures, and obtained 15, 19 and 20 with an improved activity
(IC50 = 0.3 - 0.6 µM). Computational data demonstrated that the
more potent benzothiazole derivatives 15, 19 and 20 were able to
gain several interactions at the upper part of the substrate binding
site without affecting the original binding mode of 8. As a result,
our protocol was proven successful for identifying two promising
scaffolds that warrant further chemical development as non-acidic
mPGES-1 inhibitors.
There is currently a strong interest in the development of mPGES-
1 inhibitors because of their promising potential as safer and
effective therapeutics in inflammatory diseases. In recent years,
we see an increase in the number of VS studies for the discovery
of novel mPGES-1 inhibitors, which utilize ligand- and structure-
based approaches with distinct methodologies. For example,
Lauro et al. recently reported the structure-based virtual fragment
growing optimization for the development of new mPGES-1
inhibitors leading to sulfonamide-based mPGES-1 inhibitors.[17a]
In another recent study, Rörsch et al. demonstrated that a
multistep VS methodology on a rather small library of compounds
(360,169 compounds from Asinex) can be successfully used for
identification of non-acidic mPGES-1 inhibitors.[25] Additional
studies from Schuster lab showed that urea-based virtual
screening hits, identified from 450,000 structures of NCI and
Specs databases, would be a good starting point for further
development of mPGES-1 inhibitors.[26] Apart from others, He et
al. identified novel mPGES-1 inhibitor structures in nM range by
applying a three-step VS scheme combining both docking and
molecular mapping on the Specs library of 197,211
compounds.[27] Based on the aforementioned potential of VS
approaches, we hereby have implemented a rapid VS study on a
large number of compounds from vendor libraries for identification
of novel mPGES-1 inhibitor chemotypes using the recent
mPGES-1 structures co-crystallized with some distinct and potent
mPGES-1 inhibitors. The protocol starts with docking of a huge
screening library comprising 5 million drug-like compounds,
followed by fingerprints-based clustering and diversity-based
selection. Subsequent steps were performed to ensure the
protein-ligand interaction fingerprints as observed in four crystal
structures. Out of ten candidates that were tested, the
sulfonamide-containing compounds 6 and 8 were identified as
potent suppressors of PGE2 biosynthesis without inhibiting COX-
Experimental Section
Computational studies
Docking: Crystal structures preparation and enrichment studies
Four scaffolds of active mPGES-1 inhibitors were presented in four
recently published crystal structures (PDB ID: 4YK5, 4YL0, 4YL1 and
4YL3) in which 4YK5 and 4YL0 are in complex with two biaryl-indoles;
4YL0 is in complex with phenanthrene imidazole (MF63), while the inhibitor
structure in 4YL3 is a biaryl-imidazole (Compd. 3 in Fig. S1B)[16] The PDB
coordinates of the X-ray complexes 4YK5, 4YL0, 4YL1 and 4YL3 were
obtained from the protein data bank website as a biological assembly of
three identical chains. The biological assembly files were imported into
Maestro software and merged in order to obtain the homotrimer of
mPGES-1 structures. Charges and bond orders were assigned, hydrogens
were added to the heavy atoms, and all waters were deleted. Docking
studies were carried out in one of the ligand binding sites of the crystal
structures 4YK5, 4YL0, and 4YL3. Glide energy grids were calculated
inside a box centered on the co-crystallized ligand’s centroid with
dimensions of 15 Å.[28]
7
This article is protected by copyright. All rights reserved.

10.1002/cmdc.201800701
ChemMedChem
FULL PAPER
The Schrödinger decoy set, which consists of 1000 drug-like compounds
with an average molecular weight of 400 Daltons, was used to validate the
docking protocol and the suitability of the used crystal structures.[29] The
enrichment studies were carried out by docking Chembl compounds with
reported bioactivity against mPGES-1 (from Chembl 21 database[30]) after
seeding them with Schrodinger decoys set. We considered all reported
structures with Ki or IC50 less than 1 μM as part of the mPGES-1 active
inhibitors (303 structures), while the compounds with IC50 > 5 μM as a set
of inactives (412 structures). The interactions between the protein with the
ligand in the crystal structures were characterized using the open-source
program ‘protein–ligand interaction profiler’ (PLIP).[20]
ligand, and water-ions using V-rescale algorithm (a modified Brendesen
thermostat) in order to reach a temperature of 310 K. Leap-frog integrator
was used with linear constraint solver (LINCS) algorithm applied to all
bonds. The cutoff for short-range electrostatics and van der Waals
interactions was 1.2-nm. The second phase of NPT equilibrium was
performed for 500 ps (250,000 steps) using similar conditions like the
previous NVT equilibrium (Position restrains and short-range interactions’
cutoff), except using Nose-Hoover thermostat for temperature coupling at
310 K. The temperature coupling was also applied to the exact four groups
of the system. After the two equilibration phases, a production simulation
is performed with the same parameters as NPT run but without position
restrains.
Virtual screening: Library preparation, docking and filtration
MM-PBSA calculation
MolPort’s screening library is a well-maintained database of commercially
available screening compounds integrated with ZINC library, and
synchronized with the most warehouse databases of prominent chemical
companies. The MolPort screening library, which contains a collection of
6.5 million synthetic compounds from 21 chemical vendors (as
downloaded in March 2016), has been imported into Canvas software[31]
in order to build the 3D coordinates and calculate the physio-chemical
properties. The imported structures have been filtered according to
Lipinski’s rules of five (log P in the range −0.4 to +5.6, molecular weight in
the range 150 to 500, polar surface area (PSA) ≤ 140, hydrogen bonds
donors ≤ 5, hydrogen bonds acceptors ≤ 10).[19] The drug-like selected 5
million chemical compounds have been prepared using LigPrep[32] in order
to use for generation of all the possible isomers and tautomers, lowest
energy ring conformations, and also all possible ionization states
(protonation/deprotonation) at Ph = 7.0 ± 1 using the Epik software.[33] The
generated isomers/tautomers were docked using GLIDE[34] in one of the
binding pockets of the crystal 4YL0 inside the generated grid and Glide SP
score as the fitness function. The top 40,000 docking solutions with highest
Glide scores were later docked again inside the crystal structure of 4YL3
in order to perform ensemble docking. From both docking results, we have
selected the top-scored 10,000 docking solutions in order to select a list of
virtual screening hits. The selected 10,000 docking solution were imported
in Canvas software[35] in order to generate two kinds of fingerprints such
as structure-based and pharmacophore-based fingerprints. In the final
For the calculation of MM-PBSA energy, 50 snapshots were extracted from
the last nanosecond of 20 nanoseconds MD trajectory, which means using
the interval of 20 ps from the production trajectory between 19 to 20 ns.
The program g_mmpbsa[21b] was used for estimating the binding free
energies, and later to perform the energy decomposition in order to obtain
the contribution per residue to the binding energy. The energy components
(∆EMM, ∆GP-SOLV and ∆GNP-SOLV) for individual atoms were calculated in the
bound and the unbound form, and subsequently their contribution to the
binding energy of residue x, abbreviated as ∆R_BE(x) is calculated. ∆EMM
is the average molecular mechanics potential energy in vacuum, while
∆GSOLV is the free energy of solvation and is divided into polar term ∆GP-
SOLV non-polar term ∆GNP-SOLV. The calculation was performed depending
on the solvent accessible surface area (SASA) model, which assumes
linear correlation between SASA and the non-polar solvation energy. As
the entropy contribution is not included in this protocol, the calculated
energy is not expected to be comparable to the absolute binding energy,
but to relative binding energy. In MM-PBSA, the binding energy is
evaluated according to the equation below:
∆Gbind = ∆EMM + ∆GP-SOLV + ∆GNP-SOLV -T.∆S = ∆EVDW + ∆EELE + ∆GP-SOLV
∆GNP-SOLV -T.∆S
+
The Entropy term (-T.∆S) wasn’t calculated during this study. ∆GP-SOLV is
the polar solvation contribution calculated by solving the non-linear
Poisson-Boltzmann (PB) equation using the open-source program
APBS.[42] The values for the solute and solvent dielectric constants were
chosen to be 7 and 80, respectively.[43] The non-polar solvation free energy,
∆GNP-SOLV was estimated by the solvent accessible surface area (SASA)
step,
a clustering and diversity-based selection was applied using
Tanimoto similarity metric in order to choose 1,000 unique scaffolds from
the list of the top ten thousand docking solutions, and later the protein-
ligand interactions were applied on the chosen docking complexes also
using the open-source ‘protein–ligand interaction profiler’ (PLIP) as a
stand-alone program.[20]
using a water probe radius of 1.4 Å, according to the equation ∆GNP-SOLV
=
γ.SASA + b where the constants γ and b were set to 0.022 kJ/mol/Å2 and
3.84 kJ/mol, respectively.[21b]
Molecular dynamics (MD) simulation
In order to obtain the coordinates of the mPGES-1 crystal structures (PDB
ID: 4YL0 and 4YL3) immersed in bi-layer of POPC (monounsaturated 1-
palmitoyl-2-oleoyl-sn-glycero-3-phosphocholine) membrane, the MD
system preparation in SCHRODINGER suite 2016 was used,[36] and the
generated system was later compared to the membrane location in OPM
database (4YL3.pdb is available in OPM).[37] The MD simulation was
performed using the open-source software Gromacs 5.1[38] as the protein
atoms were parameterized by the amber99sb force field, while the LIPID14
amber’s special force field for lipids was used for the lipid bi-layer
membrane.[39] The docked ligand/inhibitor was parameterized by GAFF
force field using amber14’s antechamber tool,[40] and converted to
Gromacs coordination and topology format using ACPYPE software.[41]
For the preparation for the molecular dynamics system, the protein was
centered inside a cubic box with dimensions 9.50, 9.50 and 8.50 nm.
TIP3P model of water molecules has been used when the system was
solvated on both sides of the membrane. The system was neutralized
using a mixture of the chloride Cl- and sodium Na+ ions to provide 100 mM
salt concentration. After energy minimization, a short NVT equilibrium was
applied for 100 ps (50,000 steps with a 2 fs time-step) with position
restrains applied on the protein’s backbone. Temperature coupling is
performed on four groups of the simulated system: Protein, lipids (POPC),
Biological assays
Determination of mPGES-1 activity
Microsomal preparations of A549 cells expressing mPGES-1 were
prepared as previously described.[23] In brief, A549 cells were cultivated in
DMEM medium containing FCS (2%) and IL-1β (2 ng/mL) for 72 h (37 °C,
5% CO2). Cells were harvested and resuspended in homogenization buffer
consisting
of
potassium
phosphate
(0.1
M,
pH
7.4),
phenylmethanesulfonyl fluoride (1 mM), soybean trypsin inhibitor (60
µg/mL), leupeptin (1 µg/mL), glutathione (2.5 mM), and sucrose (250 mM).
After shock-freezing of the cells in liquid nitrogen, sonication (3×20 s), and
differential centrifugation at 10,000×g (10 min, 4 °C) and 174,000×g (60
min, 4 °C), the pellets were resuspended in homogenization buffer.
Microsomes were diluted in potassium phosphate buffer (0.1 M, pH 7.4)
with glutathione (2 mM) and pre-incubated with the test compounds or
vehicle (0.1% DMSO) on ice for 15 min. After stimulation for 1 min at 4 °C
with 20 µM PGH2 as substrate, the reaction was terminated by addition of
stop solution containing FeCl3 (40 mM), citric acid (80 mM), and 11β-PGE2
(10 µM, as internal standard) and analyzed for PGE2 by RP-HPLC as
reported before.[23]
8
This article is protected by copyright. All rights reserved.

10.1002/cmdc.201800701
ChemMedChem
FULL PAPER
Chem 2016, 59, 194-205; c) Z. Zhou, Y. Yuan, S. Zhou, K. Ding, F.
Zheng, C. G. Zhan, Bioorg Med Chem Lett 2017, 27, 3739-3743.
D. Li, N. Howe, A. Dukkipati, S. T. Shah, B. D. Bax, C. Edge, A.
Bridges, P. Hardwicke, O. M. Singh, G. Giblin, A. Pautsch, R. Pfau,
G. Schnapp, M. Wang, V. Olieric, M. Caffrey, Cryst Growth Des
2014, 14, 2034-2047.
C. A. Lipinski, J Pharmacol Toxicol Methods 2000, 44, 235-249.
S. Salentin, S. Schreiber, V. J. Haupt, M. F. Adasme, M. Schroeder,
Nucleic Acids Res 2015, 43, W443-447.
a) P. A. Kollman, I. Massova, C. Reyes, B. Kuhn, S. Huo, L. Chong,
M. Lee, T. Lee, Y. Duan, W. Wang, O. Donini, P. Cieplak, J.
Srinivasan, D. A. Case, T. E. Cheatham, 3rd, Acc Chem Res 2000,
33, 889-897; b) R. Kumari, R. Kumar, C. Open Source Drug
Discovery, A. Lynn, J Chem Inf Model 2014, 54, 1951-1962.
a) M. Karplus, J. N. Kushick, Macromolecules 1981, 14, 325-332; b)
S. Kassem, M. Ahmed, S. El-Sheikh, K. H. Barakat, J Mol Graph
Model 2015, 62, 105-117; c) J. Zhang, H. Zhang, T. Wu, Q. Wang,
D. van der Spoel, J Chem Theory Comput 2017, 13, 1034-1043.
A. Koeberle, U. Siemoneit, U. Buhring, H. Northoff, S. Laufer, W.
Albrecht, O. Werz, J Pharmacol Exp Ther 2008, 326, 975-982.
A. Koeberle, U. Siemoneit, H. Northoff, B. Hofmann, G. Schneider,
O. Werz, Eur J Pharmacol 2009, 608, 84-90.
F. Rorsch, I. Wobst, H. Zettl, M. Schubert-Zsilavecz, S. Grosch, G.
Geisslinger, G. Schneider, E. Proschak, J Med Chem 2010, 53, 911-
915.
B. Waltenberger, K. Wiechmann, J. Bauer, P. Markt, S. M. Noha, G.
Wolber, J. M. Rollinger, O. Werz, D. Schuster, H. Stuppner, J Med
Chem 2011, 54, 3163-3174.
S. He, C. Li, Y. Liu, L. Lai, J Med Chem 2013, 56, 3296-3309.
R. A. Friesner, J. L. Banks, R. B. Murphy, T. A. Halgren, J. J. Klicic,
D. T. Mainz, M. P. Repasky, E. H. Knoll, M. Shelley, J. K. Perry, D.
E. Shaw, P. Francis, P. S. Shenkin, J Med Chem 2004, 47, 1739-
1749.
T. A. Halgren, R. B. Murphy, R. A. Friesner, H. S. Beard, L. L. Frye,
W. T. Pollard, J. L. Banks, J Med Chem 2004, 47, 1750-1759.
A. Gaulton, L. J. Bellis, A. P. Bento, J. Chambers, M. Davies, A.
Hersey, Y. Light, S. McGlinchey, D. Michalovich, B. Al-Lazikani, J.
P. Overington, Nucleic Acids Res 2012, 40, D1100-1107.
Schrödinger Release 2016-2: Canvas, LLC, New York, NY, 2017.
Schrödinger Release 2016-2: LigPrep, LLC, New York, NY, 2017.
J. C. Shelley, A. Cholleti, L. L. Frye, J. R. Greenwood, M. R. Timlin,
M. Uchimaya, J Comput Aided Mol Des 2007, 21, 681-691.
Schrödinger Release 2016-2: Glide, LLC, New York, NY, 2017.
Schrödinger Release 2016-3: Desmond Molecular Dynamics
System, New York, NY, 2016. Maestro-Desmond Interoperability
Tools, Schrödinger, New York, NY, 2016.
Schrödinger Small-Molecule Drug Discovery Suite 2016, LLC, New
York, NY, 2017.
M. A. Lomize, A. L. Lomize, I. D. Pogozheva, H. I. Mosberg,
Bioinformatics 2006, 22, 623-625.
M. J. Abraham, T. Murtola, R. Schulz, S. Pall, J. C. Smith, B. Hess,
E. Lindahl, SoftwareX 2015, 1-2, 19-25.
Acknowledgements
[18]
We would like to acknowledge the Turkish Academy of Sciences (TÜBA)
and the Deutsche Forschungsgemeinschaft SFB1278 “Polytarget” for
financial support. Suhaib Shekfeh acknowledges the Scientific and
Technological Research Council of Turkey (TUBITAK)-BIDEB 2221
Fellowships for Visiting Scientists for the fellowship. One of the GPUs used
in this study was a donation from Nvidia Corporation.
[19]
[20]
[21]
Keywords: Microsomal prostaglandin E2 synthase-1 •
prostaglandin • virtual screening • docking • inflammation
[22]
[23]
[24]
[25]
References:
[1]
[2]
[3]
a) C. D. Funk, Science 2001, 294, 1871-1875; b) B. Samuelsson, R.
Morgenstern, P. J. Jakobsson, Pharmacol Rev 2007, 59, 207-224.
P. J. Jakobsson, S. Thoren, R. Morgenstern, B. Samuelsson, Proc
Natl Acad Sci U S A 1999, 96, 7220-7225.
a) I. Gomez, N. Foudi, D. Longrois, X. Norel, Prostaglandins Leukot
Essent Fatty Acids 2013, 89, 55-63; b) A. Koeberle, O. Werz,
Biochem Pharmacol 2015, 98, 1-15; c) K. Larsson, P. J. Jakobsson,
Prostaglandins Other Lipid Mediat 2015, 120, 161-165.
[26]
[27]
[28]
[4]
[5]
H. H. Chang, E. J. Meuillet, Future Med Chem 2011, 3, 1909-1934.
D. Riendeau, R. Aspiotis, D. Ethier, Y. Gareau, E. L. Grimm, J. Guay,
S. Guiral, H. Juteau, J. A. Mancini, N. Methot, J. Rubin, R. W.
Friesen, Bioorg Med Chem Lett 2005, 15, 3352-3355.
[6]
a) B. Cote, L. Boulet, C. Brideau, D. Claveau, D. Ethier, R. Frenette,
M. Gagnon, A. Giroux, J. Guay, S. Guiral, J. Mancini, E. Martins, F.
Masse, N. Methot, D. Riendeau, J. Rubin, D. Xu, H. Yu, Y.
Ducharme, R. W. Friesen, Bioorg Med Chem Lett 2007, 17, 6816-
6820; b) A. Giroux, L. Boulet, C. Brideau, A. Chau, D. Claveau, B.
Cote, D. Ethier, R. Frenette, M. Gagnon, J. Guay, S. Guiral, J.
Mancini, E. Martins, F. Masse, N. Methot, D. Riendeau, J. Rubin, D.
Xu, H. Yu, Y. Ducharme, R. W. Friesen, Bioorg Med Chem Lett 2009,
19, 5837-5841.
[29]
[30]
[31]
[32]
[33]
[7]
[8]
T. Y. Wu, H. Juteau, Y. Ducharme, R. W. Friesen, S. Guiral, L.
Dufresne, H. Poirier, M. Salem, D. Riendeau, J. Mancini, C. Brideau,
Bioorg Med Chem Lett 2010, 20, 6978-6982.
[34]
[35]
J. F. Chiasson, L. Boulet, C. Brideau, A. Chau, D. Claveau, B. Cote,
D. Ethier, A. Giroux, J. Guay, S. Guiral, J. Mancini, F. Masse, N.
Methot, D. Riendeau, P. Roy, J. Rubin, D. Xu, H. Yu, Y. Ducharme,
R. W. Friesen, Bioorg Med Chem Lett 2011, 21, 1488-1492.
a) T. Shiro, K. Kakiguchi, H. Takahashi, H. Nagata, M. Tobe, Bioorg
Med Chem 2013, 21, 2868-2878; b) T. Shiro, H. Takahashi, K.
Kakiguchi, Y. Inoue, K. Masuda, H. Nagata, M. Tobe, Bioorg Med
Chem Lett 2012, 22, 285-288.
[36]
[37]
[38]
[39]
[40]
[9]
C. J. Dickson, B. D. Madej, A. A. Skjevik, R. M. Betz, K. Teigen, I.
R. Gould, R. C. Walker, J Chem Theory Comput 2014, 10, 865-879.
J. Wang, W. Wang, P. A. Kollman, D. A. Case, J Mol Graph Model
2006, 25, 247-260.
A. W. Sousa da Silva, W. F. Vranken, BMC Res Notes 2012, 5, 367.
N. A. Baker, D. Sept, S. Joseph, M. J. Holst, J. A. McCammon, Proc
Natl Acad Sci U S A 2001, 98, 10037-10041.
a) M. Jafari, F. Mehrnejad, R. Aghdami, N. Chaparzadeh, Z.
Razaghi Moghadam Kashani, F. Doustdar, J Chem Inf Model 2017,
57, 929-941; b) J. Lee, S. W. Jung, A. E. Cho, Langmuir 2016, 32,
1782-1790.
[10]
[11]
G. Lauro, M. Strocchia, S. Terracciano, I. Bruno, K. Fischer, C.
Pergola, O. Werz, R. Riccio, G. Bifulco, Eur J Med Chem 2014, 80,
407-415.
a) G. B. Arhancet, D. P. Walker, S. Metz, Y. M. Fobian, S. E.
Heasley, J. S. Carter, J. R. Springer, D. E. Jones, M. J. Hayes, A. F.
Shaffer, G. M. Jerome, M. T. Baratta, B. Zweifel, W. M. Moore, J. L.
Masferrer, M. L. Vazquez, Bioorg Med Chem Lett 2013, 23, 1114-
1119; b) D. P. Walker, G. B. Arhancet, H. F. Lu, S. E. Heasley, S.
Metz, N. M. Kablaoui, F. M. Franco, C. E. Hanau, J. A. Scholten, J.
R. Springer, Y. M. Fobian, J. S. Carter, L. Xing, S. Yang, A. F.
Shaffer, G. M. Jerome, M. T. Baratta, W. M. Moore, M. L. Vazquez,
Bioorg Med Chem Lett 2013, 23, 1120-1126.
[41]
[42]
[43]
[12]
[13]
A. Koeberle, S. A. Laufer, O. Werz, J Med Chem 2016, 59, 5970-
5986.
Y. Jin, C. L. Smith, L. Hu, K. M. Campanale, R. Stoltz, L. G. Huffman,
Jr., T. A. McNearney, X. Y. Yang, B. L. Ackermann, R. Dean, A.
Regev, W. Landschulz, Clin Pharmacol Ther 2016, 99, 274-284.
C. McInnes, Curr Opin Chem Biol 2007, 11, 494-502.
[14]
[15]
T. Sjogren, J. Nord, M. Ek, P. Johansson, G. Liu, S. Geschwindner,
Proc Natl Acad Sci U S A 2013, 110, 3806-3811.
[16]
J. G. Luz, S. Antonysamy, S. L. Kuklish, B. Condon, M. R. Lee, D.
Allison, X. P. Yu, S. Chandrasekhar, R. Backer, A. Zhang, M.
Russell, S. S. Chang, A. Harvey, A. V. Sloan, M. J. Fisher, J Med
Chem 2015, 58, 4727-4737.
[17]
a) G. Lauro, P. Tortorella, A. Bertamino, C. Ostacolo, A. Koeberle,
K. Fischer, I. Bruno, S. Terracciano, I. M. Gomez-Monterrey, M.
Tauro, F. Loiodice, E. Novellino, R. Riccio, O. Werz, P. Campiglia,
G. Bifulco, ChemMedChem 2016, 11, 612-619; b) M. A. Schiffler, S.
Antonysamy, S. N. Bhattachar, K. M. Campanale, S.
Chandrasekhar, B. Condon, P. V. Desai, M. J. Fisher, C. Groshong,
A. Harvey, M. J. Hickey, N. E. Hughes, S. A. Jones, E. J. Kim, S. L.
Kuklish, J. G. Luz, B. H. Norman, R. E. Rathmell, J. R. Rizzo, T. W.
Seng, S. J. Thibodeaux, T. A. Woods, J. S. York, X. P. Yu, J Med
9
This article is protected by copyright. All rights reserved.

10.1002/cmdc.201800701
ChemMedChem
FULL PAPER
Entry for the Table of Contents
Novel scaffolds offer new hope against inflammation: Many pathological conditions involve inflammation, including rheumatoid
arthritis, asthma and cancer. mPGES-1, a safer alternative to COX inhibition, represents an attractive therapeutic target for
inflammation-related diseases. By using VS approaches and small library synthesis, we hereby identified novel scaffolds that warrant
further chemical development as non-acidic mPGES-1 inhibitors.
10
This article is protected by copyright. All rights reserved.