Identification of novel microsomal prostaglandin E2synthase-1 (mPGES-1) lead inhibitors from Fragment Virtual Screening

Article abstract of DOI:10.1016/j.ejmech.2016.09.042

Identification of new microsomal prostaglandin E₂synthase-1 (mPGES-1) inhibitors is currently sought for the treatment of cancer and inflammation. Here we show the results of a Fragment Virtual Screening campaign using the X-ray crystal structu

Full text of DOI:10.1016/j.ejmech.2016.09.042

Accepted Manuscript
Identification of novel microsomal prostaglandin E synthase-1 (mPGES-1) lead
2
inhibitors from fragment Virtual Screening
Gianluigi Lauro, Michele Manfra, Silvana Pedatella, Katrin Fischer, Vincenza
Cantone, Stefania Terracciano, Alessia Bertamino, Carmine Ostacolo, Isabel Gomez-
Monterrey, Mauro De Nisco, Raffaele Riccio, Ettore Novellino, Oliver Werz, Pietro
Campiglia, Giuseppe Bifulco
PII:
S0223-5234(16)30773-5
10.1016/j.ejmech.2016.09.042
EJMECH 8908
DOI:
Reference:
To appear in: European Journal of Medicinal Chemistry
Received Date: 28 April 2016
Revised Date: 1 July 2016
Accepted Date: 13 September 2016
Please cite this article as: G. Lauro, M. Manfra, S. Pedatella, K. Fischer, V. Cantone, S. Terracciano,
A. Bertamino, C. Ostacolo, I. Gomez-Monterrey, M. De Nisco, R. Riccio, E. Novellino, O. Werz, P.
Campiglia, G. Bifulco, Identification of novel microsomal prostaglandin E synthase-1 (mPGES-1)
2
lead inhibitors from fragment Virtual Screening, European Journal of Medicinal Chemistry (2016), doi:
10.1016/j.ejmech.2016.09.042.
This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to
our customers we are providing this early version of the manuscript. The manuscript will undergo
copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please
note that during the production process errors may be discovered which could affect the content, and all
legal disclaimers that apply to the journal pertain.

ACCEPTED MANUSCRIPT

ACCEPTED MANUSCRIPT
Identification of novel microsomal prostaglandin E2
synthase‑1 (mPGES-1) lead inhibitors from
Fragment Virtual Screening
Gianluigi Lauro,a Michele Manfra,b Silvana Pedatella,e Katrin Fischer,d Vincenza Cantone,a,d
Stefania Terracciano,a Alessia Bertamino,a Carmine Ostacolo,c Isabel Gomez-Monterrey,c Mauro
De Nisco,b Raffaele Riccio,a Ettore Novellino,c Oliver Werz,d Pietro Campiglia,a,* and Giuseppe
Bifulcoa,*
aDepartment of Pharmacy, University of Salerno, via Giovanni Paolo II 132, 84084 Fisciano,
Italy
bDepartment of Science, University of Basilicata, Viale dell’Ateneo Lucano 10, 85100 Potenza,
Italy
cDepartment of Pharmacy, Università degli Studi di Napoli “Federico II”, Via Montesano 49,
80131 Napoli, Italy
dDepartment of Pharmaceutical/Medicinal Chemistry, Institute of Pharmacy, University of Jena,
Philosophenweg 14, D-07743 Jena, Germany
eDepartment of Chemical Sciences, University of Napoli Federico II, Via Cintia 4, I-80126
Napoli, Italy
1

ACCEPTED MANUSCRIPT
*To whom correspondence should be addressed: bifulco@unisa.it, Telephone: +39 (0)89
969741, Fax: +39 (0)89 969602, pcampiglia@unisa.it, Telephone: +39 (0)89 969242, Fax: +39
(0)89 969602
2

ACCEPTED MANUSCRIPT
ABSTRACT
Identification of new microsomal prostaglandin E2 synthase-1 (mPGES-1) inhibitors is
currently sought for the treatment of cancer and inflammation. Here we show the results of a
Fragment Virtual Screening campaign using the X-ray crystal structure of human mPGES-1
(PDB code: 4AL0). Among the fragments selected and biologically tested, 6 (9H-indeno[1,2-
b][1,2,5]oxadiazolo[3,4-e]pyrazin-9-one) showed the most promising mPGES-1 inhibitory
activity (~30% inhibition at 10 ꢀM). A minimal structure-based optimization of 6 led to
compounds 15, 20 and 21, with a promising enhancement of the inhibitory activity (IC50 = 4.6 ±
0.2 µM for 15; 2.4 ± 1.0 µM for 20; IC50 = 2.4 ± 0.8 µM for 21). The unprecedented chemical
core and the possibility of synthesizing novel derivatives reveal a new and attractive field of
action for the development of mPGES-1 inhibitors with potential anti-inflammatory and
anticancer properties.
KEYWORDS: Molecular docking, cancer, inflammation, mPGES-1 inhibitors, Virtual Screening
3

ACCEPTED MANUSCRIPT
Introduction
Nonsteroidal anti-inflammatory drugs (NSAIDs) nowadays represent the most important
agents for the inflammation therapy. They act suppressing the biosynthesis of prostaglandins
(PGs), bioactive mediators involved in key physiological functions but also in several other
pathologic conditions, such as tumorigenesis[1]. The relationship between inflammation and
tumor progression is noteworthy[2-4], and the correlation between inflammatory events and the
development of pre-cancerous lesions at various anatomic sites has been established[5-7].
Proliferation, invasion and migration processes are promoted by inflammatory events and,
indeed, the identification of new anti-inflammatory drugs could be synergic in the context of
anti-cancer pharmacological strategies[8]. Nevertheless, NSAIDs cause several side effects due
to their action on cyclooxygenase (COX) targets, such as cardiovascular, gastrointestinal and
renal side[9], and then the development of safer alternatives especially for long-term therapies is
even more required[10, 11].
In this scenario, prostaglandin E2 synthases (PGES, namely mPGES-1, mPGES-2 and cPGES)
represent valuable targets for the development of new anti-inflammatory/anticancer agents with
reduced side effects. Specifically, PGES are terminal enzymes involved in the biosynthesis of the
inflammatory lipid mediator PGE2[12, 13]. In particular, while mPGES-2 and cPGES represent
the constitutive form, the inducible membrane-bound isoform mPGES-1 has become as a key
drug target in PGE2-related acute and chronic disorders[14], such as pain[15], fever[16],
rheumatoid arthritis[17], arthritis[18], inflammation[19], and cancer[20, 21]. Then, mPGES-1
inhibitors recently emerged as new valuable and safer drugs, avoiding the chemical conversion
of PGH2, enzymatically produced by the COXs, to PGE2[22]. Up to now, a number of mPGES-1
inhibitors featuring different chemical platforms have been developed[23]. The structure-based
4

ACCEPTED MANUSCRIPT
design of new inhibitors was supported in 2013 by the first X-ray crystal structure of the protein
with the identification of the active site[24], and subsequently by other crystal structures with
several known mPGES-1 inhibitors acting both as substrate (available crystal structures with
PDB codes: 4BPM, 4WAB, 4YK5, 4YL0, 4YL1, 4YL3)[25-27] and substrate/cofactor
competitors (PDB code: 4AL1)[24]. Also, new high-resolution X-ray structures of human
mPGES-1 in complex with new and potent inhibitors have been presented (PDB code: 5BQG,
5BQH, and 5BQI)[28]. This new information is crucial in a medicinal chemistry approach for the
structure-based design of novel inhibitors, and it could also serve to retrospectively analyze the
binding modes of known binders[22].
In the last few years, our research group has been involved in the computer-aided design,
chemical synthesis and biological evaluation of novel mPGES-1 inhibitors featuring different
chemical scaffolds[29-39]. With the aim of identifying novel mPGES-1 featuring unprecedented
molecular scaffolds, a Virtual Screening computational approach was undertaken using the
recent crystal structure of mPGES-1 (PDB code: 4AL0). Specifically, we here focused on the
fragment-based drug discovery (FBDD) approach[40], since it has recently emerged as a
significant tool in the field of medicinal chemistry for discovery of new hit compounds. In fact,
the identification of low-molecular-weight compounds (<300 Da) could allow the design of high
affinity ligands following a “linking” or “growing” approach[37, 41]. Starting from these
considerations, we report a Fragment Virtual Screening workflow leading to the identification of
new valuable chemical scaffolds, whose biological activities were detected by in vitro
experiments. Moreover, simple modifications on the most promising fragment prompted to the
chemical synthesis of a small set of derivatives,whose remarkable inhibitory activities pave the
way for the development of novel potent mPGES-1 inhibitors.
5

ACCEPTED MANUSCRIPT
Materials and Methods
Input files preparation for molecular docking. The chemical structures of the investigated
compounds were retrieved from the ZINC database [42, 43]. Specifically, we used the
“Fragment-Like – In Stock” subset, then accounting the commercially available compounds for
an immediate delivery. When the library was downloaded (2013), it was composed by ~4.4 × 105
compounds. For the subsequent docking calculations, all the structures were converted in the
.pdbqt format using OpenBabel software (version 2.3.2)[44], adding Gasteiger charges.
Protein 3D models were prepared using the Schrödinger Protein Preparation Wizard[45], starting
from the mPGES-1 X-ray structure in the active form (PDB code: 4AL0)[24]. A first set of
docking calculations was performed on the protein structure in absence of the cofactor GSH
(named mpges_1_no_gsh branch), while it was preserved for a second set of experiments
(mpges_1_with_gsh branch). Crystallized water molecules were removed, all hydrogens were
added, and bond orders were assigned. Protein .pdb files obtained were then converted in .pdbqt
format.
Molecular Docking. Docking calculations were performed using Autodock-Vina software[46].
In the configuration files linked to 3D structures of the proteins, we specified coordinates and
dimensions along x,y,z axes of the grids related to the sites of presumable pharmacological
interest. In particular, we chose the binding site between A and B chains for both the forms of the
protein, and a grid box size of 24×20×18 and centered at 10.304 (x), -11.033 (y), and -8.384 (z),
with spacing of 1.0 Å between the grid points, setting the exhaustiveness value to 8. For all the
6

ACCEPTED MANUSCRIPT
investigated compounds, all open-chain bonds were treated as active torsional bonds. Docking
results were analyzed with Autodock Tools 1.5.7 and Maestro (version 10.2)[45]. Illustrations of
the 3D models were generated using VMD software[47] and Maestro (version 10.2)[45].
General Synthetic Information.
Materials. Inorganics, organic reagents, and solvents were commercial pure compounds and
used without further purification. TLC analyses were performed using silica gel plates (silica gel
60 F-254) visualized by UV light, fluorescent light, and iodine. Column chromatography was
carried out on silica gel (70-230 mesh). 1H and 13C NMR spectra were recorded on Varian Inova
500 MHz, BrukerAvance 400 MHz, and BrukerDrx 400 MHz spectrometers: chemical shifts in
ppm (δ) and J coupling constants in Hz. The following abbreviations are used to indicate the
multiplicity: s, singlet; d, doublet; m, multiplet; b, broad signal. Abbreviations, NBS: N-
bromosuccinimide; DMSO: dimethylsulfoxide.
Synthesis of Compound 18. Substituted phenol 17 (296 mg, 2 mmol) was dissolved in 3 ml of
10% aqueous sodium hydroxide solution in a 25 mL flask. Solutions of tetra-n-butylammonium
chloride (56 mg, 0.2 mmol) in 0.7 mL of dichloromethane and acyl chloride (232 µL, 2 mmol) in
2 mL of dichloromethane were prepared. After cooling all solutions at 0 °C, they were mixed at
once. The reaction mixture was kept under vigorous magnetic stirring at 0 °C for 15 min and
then poured over 50 mL of icy water. The organic layer was separated and the aqueous layer was
extracted twice with 40 mL of diethyl ether. The combined organic extracts were washed with
saturated NaCl solution. After drying on Na2SO4, the solvent was evaporated, and the residue
was purified by silica gel chromatography (CH2Cl2) to give the compound 18. (456 mg; 1.8
mmol; 90%). 1H NMR (400 MHz, CDCl3): δ 7.83 (d, J = 8.3 Hz, 1H), 7.49-7.40 (m, 3H), 7.35-
7

ACCEPTED MANUSCRIPT
7.26 (m, 4H), 3.19 (t, J = 5.5 Hz, 2H), 2.75 (t, J = 5.5 Hz, 2H). 13C NMR (125 MHz, CDCl3): δ
206.5, 164.6, 156.8, 156.0, 134.7, 133.9, 130.1, 128.8, 128.6, 125.0, 121.4, 119.7, 36.3, 25.6.
Calculated for C16H12O3: C, 76.18%; H, 4.79%; O, 19.03%; Found: C, 76.08%; H, 4.60%; N,
19.13%.
Synthesis of Compounds 13, 19. In a typical procedure, N-bromosuccinimide (671 mg, 3.6
mmol) was added to protected phenol (12, 18) (456 mg, 1.8 mmol) in 9 mL of
dimethylsulfoxide. The reaction mixture was stirred at 60 °C for 3 hours and then at 80 °C for 4
hours. The reaction mixture was cooled to room temperature and then poured into 200 mL of
water. The product was extracted with three portions of 100 mL of dichloromethane, and then
the combined organic extracts were dried (sodium sulfate), filtered and evaporated. The residue
was purified by silica gel chromatography (Etp:AcOEt 7:3) to yield compounds 13 and 19.
1
Compound 13: 328 mg (1.1mmol; 60%). H NMR (500 MHz, DMSO-d6): δ 8.22-8.13 (m, 3H),
8.02-7.99 (m, 2H), 7.80 (t, J = 7.7 Hz, 1H), 7.68-7.65 (m, 2H), 7.59 (s, 2H, OH);13C NMR (125
MHz, DMSO-d6): δ198.7, 198.6, 164.0, 152.7, 145.5, 141.3, 136.5, 132.2, 129.6, 127.6, 125.9,
125.3, 123.0, 88.9. Anal. Calcd. for C16H10O6: C, 64.43; H, 3.38. Found: C, 64.11; H,
1
3.57.Compound 19: 349 mg (1.2 mmol; 65%). H NMR (400 MHz, DMSO-d6): δ 8.19-8.09 (m,
3H), 8.02-7.95 (m, 2H, OH), 7.72 (t, J = 6.4 Hz, 1H), 7.66-7.56 (m, 4H). 13C NMR (100 MHz,
DMSO-d6): δ 196.9, 196.3, 164.4, 157.6, 140.7, 136.6, 134.9, 131.8, 130.9, 128.8, 126.4, 117.6,
88.4. Anal. Calcd. for C16H10O6: C, 64.43; H, 3.38. Found: C, 64.11; H, 3.57.
Synthesis of Compounds 15, 20, 21. In a typical procedure,a mixture of gem-diol (13, 19) (149
mg, 0.5 mmol) and furazan-3,4-diamine (14) (50 mg, 0.5 mmol) in 1.5 mL of ethanol and 1.5 mL
of glacial acetic acid was stirred at room temperature for 18 hours and then heated at reflux for 6
hours. The precipitated solid was filtered and washed with water to yield the compounds 15, 20
8

ACCEPTED MANUSCRIPT
1
and 21. Compound 15: 76 mg (0.22 mmol; 45%), H NMR (400 MHz, DMSO-d6): δ 8.31-8.21
(m, 3H), 8.11-8.01 (m, 1H), 7.91 (d, J = 8.1 Hz, 1H), 7.83 (dd, J = 10.7, 4.2 Hz, 1H), 7.74-7.66
(m, 2H). 13C NMR (100 MHz, DMSO-d6): δ 186.1, 165.8, 165.5, 164.3, 151.6, 150.0, 149.7,
139.1, 136.4, 134.0, 131.7, 131.6, 131.5, 130.5, 124.0, 117.5. Anal. Calcd. for C18H8N4O4: C,
62.80; H, 2.34; N, 16.27. Found: C, 62.72; H, 2.10; N, 16.08. Compounds 20 and 21: A mixture
of compounds 20 and 21 in ratio 1:0.88 respectively (0.25 mmol; 50%) was purified by silica gel
chromatography (Etp:AcOEt 7:3). Compound 20: 1H NMR (500 MHz, DMSO-d6): δ 8.45 (d, J =
8.3 Hz, 1H), 8.25-8.19 (m, 3H), 8.03 (dd, J = 8.3, 1.9 Hz, 1H), 7.85-7.78 (m, 1H), 7.71-7.64 (m,
13
2H). C NMR (125 MHz, DMSO-d6): δ 185.1, 164.7, 163.3, 161.7, 158.2, 153.6, 142.1, 138.3,
137.4, 135.1, 132.2, 130.6, 129.6, 128.8, 127.2, 118.9. Anal. Calcd. for C18H8N4O4: C, 62.80; H,
2.34; N, 16.27. Found: C, 62.63; H, 2.10; N, 16.10. Compound 21: 1H NMR (500 MHz, DMSO-
d6): δ 8.35 (d, J = 1.8 Hz, 1H), 8.26-8.20 (m, 3H), 7.94 (dd, J = 8.3, 1.8 Hz, 1H), 7.84-7.78 (m,
1H), 7.70-7.63 (m, 2H). 13C NMR (100 MHz, DMSO-d6): δ 184.7, 165.6, 163.9, 162.2, 157.3,
155.7, 142.3, 138.3, 137.9, 135.9, 130.9, 130.5, 130.1, 129.7, 127.8, 118.4. Anal. Calcd. for
C18H8N4O4: C, 62.80; H, 2.34; N, 16.27. Found: C, 62.59; H, 2.19; N, 16.09.
Synthesis of Compounds 16, 22, 23. In a typical procedure, protected compounds (15, 20, 21)
(177 mg, 0.51 mmol) were dissolved in MeOH/THF (1:1) and stirred with sodium thiophenolate
(101 mg, 0.77 mmol). The reaction was heated to reflux for 2.5 h and precipitated solid was
filtered and purified by silica gel chromatography (CH2Cl2) to yield compounds 16, 22, and 23.
Compound 16: 115 mg (0,48 mmol; 95%). 1H NMR (500 MHz, DMSO-d6): δ 7.85-7.72 (m, 1H),
7.63-7.54 (m, 1H), 7.26-7.18 (m, 1H). 13C NMR (100 MHz, DMSO-d6): δ 184.1, 165.5, 152.3,
149.3, 148.2, 141.1, 138.2, 136.1, 128.0, 124.5, 117.6. Anal. Calcd. for C11H4N4O3: C, 55.01; H,
1.68; N, 23.33. Found: C, 55.12; H, 1.88; N, 23.03. Compound 22: 115 mg (0.48 mmol; 95%).
9

ACCEPTED MANUSCRIPT
1H NMR (500 MHz, DMSO-d6): δ 8.31 (s, 1H), 7.94 (d, J = 8.5 Hz, 1H), 7.20 (d, J = 7.8 Hz,
13
1H). C NMR (100 MHz, DMSO-d6): δ 183.0, 167.9, 163.6, 162.5, 153.7, 142.8, 133.2, 128.1,
124.5, 115.8. Anal. Calcd. for C11H4N4O3: C, 55.01; H, 1.68; N, 23.33. Found: C, 55.19; H, 1.89;
N, 23.01. Compound 23: 115 mg (0.48 mmol; 95%). 1H NMR (400 MHz, DMSO-d6): δ 8.59 (bs,
1H, OH), 8.15 (d, J = 6.2 Hz, 1H), 7.37 (d, J = 6.2 Hz, 1H), 7.26 (bs, 1H). 13C NMR (125 MHz,
DMSO-d6): δ 187.4, 167.5, 165.1, 164.3, 156.0, 155.9, 145.4, 135.9, 128.7, 128.4, 112.5. Anal.
Calcd. for C11H4N4O3: C, 55.01; H, 1.68; N, 23.33. Found: C, 55.08; H, 1.97; N, 23.05.
Bioactivity assays
Cell-free mPGES-1 activity assay
Microsomes of IL-1β-stimulated A549 cells were used a as source for mPGES-1. Expression of
mPGES-1, preparation of microsomes and determination of mPGES-1 activity was performed as
described previously[48]. In brief, A549 cells were treated with Il-1β (1 ng/ml) for 48 h, cells
were harvested, sonicated and the homogenate was subjected to differential centrifugation at a)
10,000×g for 10 min and b) 174,000×g for 1 h at 4 °C. The microsomal fraction (pellet) was
resuspended in 1 ml homogenization buffer (0.1 M potassium phosphate buffer, pH 7.4, 1 mM
phenylmethanesulfonyl fluoride, 60 µg/mL soybean trypsin inhibitor, 1 µg/mL leupeptin, 2.5
mM glutathione, and 250 mM sucrose), the total protein concentration was determined, and
microsomes were diluted in potassium phosphate buffer (0.1 M, pH 7.4) containing 2.5 mM
glutathione. Test compounds (or DMSO as vehicle) were added, and after 15 min at 4 °C
reaction (100 µl total volume) was initiated by addition of 20 µM PGH2. After 1 min at 4 °C, 100
µl of stop solution (40 mM FeCl2, 80 mM citric acid, and 10 µM 11β-PGE2) were added. PGE2
was separated by solid-phase extraction and analyzed by RP-HPLC as described previously[48].
10

ACCEPTED MANUSCRIPT
Results and Discussion
Structural information. mPGES-1 is a glutathione-dependent membrane protein located on
the endoplasmic reticulum and structurally organized as homotrimer, with three equivalent active
site cavities within the membrane-spanning region at each monomer interface[24]. Each
asymmetric monomer is characterized by a four-helix bundle motif, while each active site is
toward the cytoplasmic part of the protein, between the N-terminal parts of helix II and IV of one
monomer and the C-terminal part of helix I and the cytoplasmic domain of the adjacent
monomer (Figure 1).
The analysis of the mPGES-1 crystal structure (PDB code: 4AL0)[24] disclosed several
regions in the binding site that could be conveniently targeted by a functionalized ligand (Figure
1). Firstly, a binding groove is between the GSH binding site and a molecular surface nearby the
cytoplasmic part of the protein, mainly composed by aromatic (B:Phe44, B:His53) and polar
(B:Arg52) residues (colored in cyan, Figure 1).
A potential binder could establish π-π contacts with these aromatic groups, as occurred for the
co-crystallized LVJ[25], 4DV, 4DZ, 4U8, 4U9[26], and as we proposed for dihydropyrimidin-
2(1H)-one based mPGES-1 inhibitors[36].
The cofactor (GSH) is in a profound cavity mainly characterized by polar residues, and it
adopts a U-shape due to the strong interactions between its two terminal carboxylic functions and
the positively charged residues in the deeper part of the binding site (B:Arg38, A:Arg73)
(colored in yellow, Figure 1). Importantly, the phenol group in the side chain of A:Tyr130 is
involved in a π-stacking with the gamma peptide linkage between the cysteine and the glutamate
11

ACCEPTED MANUSCRIPT
of GSH. This key residue could be targeted by a binder through a π-π interaction and/or polar/H-
bond interactions with the phenol hydroxyl moiety in the side chain.
Figure 1. mPGES-1 structure: a) molecular surface representation of the mPGES-1 trimer (chain A blue, chain B
red, chain C orange); b) molecular surface focused on the mPGES-1 binding site colored in orange, yellow, and cyan
(see text for details), dummy atoms define the external part of endoplasmic reticulum membrane (above red atoms)
and cytoplasmic (below blue atoms) sides; c,d) secondary structure focused to the mPGES-1 binding site (chain A
12

ACCEPTED MANUSCRIPT
blue, chain B red, chain C orange); glutathione (GSH) cofactor and key-residues in the mPGES-1 binding site are
represented in sticks (C: green, O red, N, blue, H light gray). (2-column fitting image)
Moreover, Ser127 on chain A represents another fundamental residue, since it was supposed to
be involved in the catalytic process behind the isomerization of PGH2 to PGE2[24]. Finally,
moving from the external part of endoplasmic reticulum membrane to the cytoplasmic part of the
protein, a binding groove is identifiable at the intersection between helix 1 of chain B and helix 4
of chain A, with polar (A:Gln134), aliphatic (B:Val24) and aromatic (B:Tyr28) residues, and
could be bound by long molecular functions (colored in orange, Figure 1).
Fragment Virtual Screening and biological evaluation.The identification of novel mPGES-
1 inhibitors was conducted following a multi-step structure-based Fragment Virtual Screening
approach, using the information arising from the analysis of the protein crystal structure
crystallized by Sjögren et al. (PDB code: 4AL0)[24]. In a previous study, we have compared
different mPGES-1 protein crystal structures highlighting the limited induced fit of the protein
after ligand binding, and the applicability of semi-flexibile molecular docking experiments
(namely ligands considered as flexible, while protein as rigid)[37]. The workflow is represented
in Figure 2.
The fragment library was docked onto the mPGES-1 protein structure after removal of the
cofactor (glutathione, GSH; this branch was named mpges_1_no_gsh). After several selection
steps (Steps 2-4, Figure 2), the most promising compounds were also docked on the protein in
the presence of the cofactor (mpges_1_with_gsh branch) (Step 5, Figure 2).
Previous studies[49] highlighted that mPGES-1 inhibitors could bind the protein either
occupying only the substrate binding site (prostaglandin H2, PGH2) or partially displacing GSH,
then filling part of the cofactor binding site as far as extending to the substrate active site. These
13

ACCEPTED MANUSCRIPT
two different binding modes seem to dramatically affect the potency, and in particular the
accommodation in both the sites (substrate and cofactor sites) could lead to high affinity
inhibitors[49]. Even if it is unlikely that a low affinity compound (such as a fragment) could
interfere with the binding of the cofactor, we considered this possibility in silico in view of the
subsequent optimization steps aimed to the generation of small molecules starting from
fragments.
Figure 2. Fragment Virtual Screening workflow for the identification of novel mPGES-1 inhibitors (1-column
fitting image)
The identification of the most promising binding poses represented the starting point for the
design of optimized compounds, able to bind this protein form with high affinity respecting the
14

ACCEPTED MANUSCRIPT
original fragment positions and filling the unoccupied regions thanks to the additional chemical
groups. On the other hand, in order to facilitate the identification of active fragments prone to the
optimization, we also evaluated the binding mode of the identified hits in the presence of GSH
(mpges1_with_gsh branch), then finally selecting the compounds able to advantageously bind the
protein in both the forms.
In details, the Virtual Screening workflow started with the molecular docking of the starting
fragment library (~4.4 ×105 compounds) on the mPGES-1 protein structure built removing the
cofactor (mpges_1_no_gsh). The first selection filter concerned the most energetically favored
compounds, saving all the fragments whose predicted binding energies were comprised in a
range of 1.5 kcal/mol from the best identified value (Step 2, Figure 2). For the selected fragments
(2332 compounds), the related Ligand Efficiency[50] values were computed. This parameter
links the estimated binding energy with the molecular size and could advantageously guide the
optimization process of small chemical probes, leading to potent small molecules respecting the
Lipinsky rules[51]. Compounds with Ligand Efficiency values better than -0.500 kcal/mol
passed this filter, then restricting the number to 254 promising molecules (Step 3, Figure 2). In
view of the subsequent step of fragment-to-lead optimization, we carefully analyzed the binding
modes of the identified fragment docking poses, with the aim of selecting only those respecting
the key interactions with the receptor counterpart. The selected fragments were supposed to
advantageously interact in the cofactor (GSH) binding site, respecting part of the contacts
detected for the cofactor in the protein structure. In particular, different set of ligand-receptor
contacts were considered (Step 4, Figure 2):
•
•
π-π interaction with A:Tyr130;
polar contacts with A:Arg126 and/or B:Arg38 and/orA:Arg73.
15

ACCEPTED MANUSCRIPT
Depending on the starting positions in the cofactor binding site, the fragments could be
optimized using a “growing” approach on the cytosplasmic part (e.g. C-domain) or toward the
external binding groove (A:Gln134, B:Val24, B:Tyr28) belonging to the mPGES-1 binding site.
After that, the identified fragments (75 compounds) were then submitted to another docking
round, considering the protein form in the presence of the cofactor (mpges_1_with_gsh branch)
and evaluating their binding mode (Step 5, Figure 2). In this case, the following groups of
interactions were set for the selection:
•
a) edge-to-face π-π interaction with B:Phe44 and/or B:His53, and polar contacts with
A:Ser127
•
alternatively: b) edge-to-face π-π interaction with A:Tyr130, and polar contacts with
GSH.
In this way, in the a) case, the interacting fragments could be optimized moving from the
cytoplasmic to the most external part of the protein, reaching the external binding groove; in the
b) case, the fragment could grow toward the cytoplasmic part, with the aim of establishing
additional edge-to-face π- π interactions with B:Phe44 and/or B:His53.
The final stage of the workflow concerned the selection of the most promising fragments for the
biological evaluation and the eventual subsequent structure-based chemical optimization (Step 6,
Figure 2). The compounds identified for both the branches were then further visually inspected
to check which positions could be modified without perturbing the original binding modes. 11
compounds (1-11, Chart 1) were finally selected,purchased and biologically tested. To assess the
ability of compounds 1-11 (Chart 1) to interfere with the activity of mPGES-1, a cell-free assay
using the microsomal fractions of interleukin-1β (IL-1β)-stimulated A549 cells (as source for
mPGES-1) was applied. In a first screening round, all the compounds were solubilized in DMSO
16

ACCEPTED MANUSCRIPT
and tested at a final concentration of 10 ꢀM. Biological data are summarized in Figure 3. Among
the tested fragments, 6 (9H-indeno[1,2-b][1,2,5]oxadiazolo[3,4-e]pyrazin-9-one) showed the
most promising inhibitory activity (~30% of inhibition), while interestingly, 3 and 11 showed
only ~20% of inhibition (Figure 3).
Chart 1. Chemical structures of selected fragments 1-11.
17

ACCEPTED MANUSCRIPT
Figure 3. mPGES-1 remaining activity in the presence of compounds 1-11 at 10 ꢀM final concentration. Data are
given as mean ± S.E.M, n = 3. (1-column fitting image)
Fragment-to-lead optimization of compound 6 - Chemistry and biological evaluation. The
most active identified fragment 6 was then modified following two distinct optimization
schemes, starting from the analysis of the different binding modes from the two mPGES-1
structures considered for the docking calculations (mpges_1_no_gsh and mpges_1_with_gsh
branches) (Figure 4).
18

ACCEPTED MANUSCRIPT
Figure 4. Different binding modes of 6 (colored by atom types: C orange, O red, N blue, polar H light gray) in the
mPGES-1 binding site (represented in ribbons, blue for chain A, red for chain B) in the: a) absence, and b) presence
of the cofactor (GSH). Residues in the active site are represented in sticks (colored by atom types: C green, N blue,
O red, S yellow, H light gray, GSH depicted in green). H-bonds are represented in green dotted lines, while π-π
interactions are depicted with cyan dotted lines. (2-column fitting image)
Firstly, following the hypothesis that the optimized derivatives could act as substrate/cofactor
competitive inhibitors, we noticed that 6 occupied the cofactor binding site establishing the face-
to-face with A:Tyr130 and two hydrogen bonds with A:Arg73 and A:Arg126 (Figure 4A),
suggesting the introduction of chemical substituents able to fill the external part of the binding
site.
Also, we carefully investigated the binding mode of 6 hypothesizing the substrate
displacement and identifying two main binding modes in the GSH binding site (Figure 4B,
Figure S1, Supporting Information). In the first one (Figure 4B), the oxadiazole moiety on ring D
and the aromatic ring A of 6 established edge-to-face π-π interactions with B:Phe44 and
B:His53, suggesting in this case the insertion of chemical groups able to making contacts with
the residues belonging to the external binding groove (e.g. A:Tyr130). The alternative docking
pose of 6 showed an edge-to-face π-π interaction with A:Tyr130, a π-cation with A:Arg126 and
contacts with the cofactor GSH, while the remaining portion of the fragment pointed toward the
cytoplasmic part of the protein (Figure S1, Supporting Information).
The analysis of the docking poses of the starting fragment 6 then showed that the insertion of
substituents on the available positions of ring A could represent the starting point for the
synthesis of optimized derivatives. With this aim, we developed a new and rapid synthetic route
allowing the regioselective introduction of a hydroxyl as an attachment point on the aromatic
19

ACCEPTED MANUSCRIPT
ring A of 6 (Scheme 1), specifically leading to its fragment derivatives 16, 22, 23, prone to be
modified for the optimization step.
In detail, substituted analogues of the tetracycle 6 were obtained by following the protocol
described by the Bratton et al.[52] with appropriate modification to reduce the reaction times
(Scheme 1). Namely, the synthesis of derivative 16 started from the electrophilic bromination of
12 with N-bromosuccinimide, followed by oxidation with dimethyl sulfoxide, to give the 7-
substituted ninhydrin 13 in 60% yield.
Condensation of the latter with furazan-3,4-diamine (14) in a mixture of ethanol and acetic
acid afforded the substituted tetracycle 15. Although concomitant formation of the 4-substituted
regioisomer of 15 was possible, this product was not isolated from the reaction mixture.
Indeed, the reaction of 15 with sodium thiophenolate in MeOH/THF (1:1) provided the phenol
16 in 95% yield. This deprotection was preferred to the one earlier reported, involving the use of
cesium carbonate in tetrahydrofuran due to really shorter reaction time (from 5 days to only 2.5
h) with same yields[53]. It is noteworthy that the phenolic derivative 16 is useful for the
introduction of additional functionality. In details, Mitsunobu reaction conditions could be
employed for the introduction of a range of alkyl and aryl groups into the tetracycle obtained.
The procedure above described was exploited to obtain the isomeric C-6 and C-7 hydroxy
substituted analogs. Therefore, after the protection of the hydroxyl group of the commercially
available indanone 17 with benzoyl chloride[54], the synthetic strategy provided a mixture of
two tetracycles 20 and 21 in ratio 1:0.88 respectively through the intermediate gem-diol 19.
The two isomers were readily separated and purified by silica gel chromatography (Etp:AcOEt
7:3), then the ester cleavage produced the corresponding compounds 22 and 23 with a 95%
yields in both cases.
20

ACCEPTED MANUSCRIPT
We were intrigued by compounds 15, 20, 21 produced during the synthesis as the protected
analogues of 16, 22, 23, respectively, for which docking calculations showed promising binding
modes in the mPGES-1 binding site. In particular, the ester function on ring A guaranteed the
correct distance between the original core and the new aromatic function, and the gain of further
favorable contacts while respecting the binding modes previously described for compound 6. In
details, in the absence of the cofactor we detected the growing of the starting fragment toward
the external part of the binding site, with the establishment of an additional edge-to-face π-π
interaction with with A:Phe44 for 16 (Figure 5A), and with A:Tyr130 (Figure 5B-C) for 20 and
21. In the presence of the cofactor, compounds 15, 20 and 21 confirmed the fundamental edge-
to-face π-π interaction with B:Phe44/B:His53, respecting the binding mode of fragment 6 as
reported in figure 4B. In addition, the attachment of the benzyloxy group on the ring A of 6
determined additional edge-to-face π-π interactions with A:Tyr130 (Figure 5D-F). Furthermore,
we also found further binding modes of 15, 20, and 21 respecting the alternative docking pose of
6 (as reported in Figure S1, Supporting Information), in which the original core was oriented
toward the A:Tyr130 (Figure S2, Supporting Information), while the benzyloxy substituents
established edge-to-face π-π with B:Phe44/B:His53. Notably, in previous attempts[36] and after
a retrospective analysis of the binding mode of the inhibitor LVJ[25, 36], we highlighted the
importance of this key-interaction for the structure-based drug design and optimization of new
mPGES-1 inhibitors. As expected, for compounds 16, 22, and 23 the insertion of the hydroxyl
did not dramatically perturb the proposed binding modes of the starting fragment 6 (Figure S3,
Supporting Information).
Chart 2. Chemical structure of compounds 6, 15, 16, 20-23.
21

ACCEPTED MANUSCRIPT
Scheme 1. Chemical synthesis of compounds 15, 16, 20-23.
22

ACCEPTED MANUSCRIPT
Results from the assessment of mPGES-1 inhibition by compounds 15, 16, 20-23 corroborated
the computational outcomes, and specifically compounds 15, 20 and 21 confirmed their ability to
interfere with mPGES-1 activity (% of inhibition: ~ 70% for 15; ~ 70% for 20; ~ 85% for 21, at
10 µM). The related IC50 values substantiate these results, obtaining promising IC50 values in the
low micromolar range (IC50 = 4.6 ± 0.2 µM for 15; IC50 = 2.4 ± 1.0 µM for 20; IC50 = 2.4 ± 0.8
µM for 21) (Figure 6). As expected, compounds 16, 22, and 23 showed an inhibitory activity
comparable with that of starting fragment 6 (% of inhibition = 30% for 22, 15% for 23, tested at
10 µM), while a lower value was found for 16 (5% of inhibition at 10 µM).
23

ACCEPTED MANUSCRIPT
From a structural point of view, the here reported novel mPGES-1 inhibitors were discovered
performing molecular docking experiments on both the protein forms in the absence and in the
presence of the cofactor (glutathione, GSH). In parallel, new experimental information arose
from the analysis of the recently released crystal structures of mPGES-1 co-complexed with high
affinity inhibitors (PDB codes: 4BPM, 4WAB, 4YK5, 4YL0, 4YL1, 4YL3, 5BQG, 5BQH,
5BQI), confirming the presence of GSH at its site. Furthermore, the only one co-crystallized
inhibitor acting as partial substrate/cofactor competitor (PDB code: 4AL1) is a close analogue of
GSH (bis-phenyl-GSH), thus maintaining the same network of interactions of the original
cofactor at its site. Taken together, these experimental data point out that the displacement of the
cofactor is possible only if the binding site is occupied by a similar ligand respecting the binding
mode reported for GSH. These structural outcomes indicate the substrate displacement
hypothesis for the set of compounds here reported, and these information will be accounted for
the future in silico design of optimized derivatives.
24

ACCEPTED MANUSCRIPT
Figure 5. a,d) 3D docking models of 15 (colored by atom types: C cyan, O red, N blue, polar H light gray); b,e)
20 (colored by atom types: C pink, O red, N blue, polar H light gray), and c,f) 21 (colored by atom types: C violet, O
red, N blue, polar H light gray) in the mPGES-1 binding site in the absence (a,b,c) and in the presence (d,e,f) of the
cofactor GSH. Residues in the active site are represented in sticks (colored by atom types: C green, N blue, O red, S
yellow, H light gray). H-bonds are represented in green dotted lines, while π-π interactions are depicted with cyan
dotted lines. (2-column fitting image)
25

ACCEPTED MANUSCRIPT
Figure 6. a) mPGES-1 remaining activity in the presence of compounds 15, 16, 20-23 at 10 ꢀM final
concentration. b, c, d) Concentration-response curves of compounds 15, 20, 21 for inhibition of mPGES-1 activity,
respectively. Data are given as mean ± S.E.M., n = 3. (2-column fitting image)
Conclusion
Here we reported about the computer-aided identification, biological evaluation, and
optimization of the novel mPGES-1 small fragment inhibitor 6 following a Fragment Virtual
Screening approach.
In detail, from a multi-step Fragment Virtual Screening 11 compounds were selected,
purchased and biologically evaluated. Starting from the most active fragment 6 (~30% of
inhibition at concentration of 10 ꢀM), we have developed a synthetic route aimed at the
26

ACCEPTED MANUSCRIPT
preparation of its derivatives 16, 22, and 23, with the introduction of a hydroxy substituent in
different positions as possible attachment points for the future optimization steps.
By means of docking calculations, we also evaluated the binding modes of the synthesized
protected analogues 15, 20, and 21 featuring a benzyloxy chemical function attached to the
original chemical core. Computational data showed the ability of these compounds of gaining
several interactions while not affecting the original binding modes of 6. Encouraged by the
computational results, we biologically evaluated 15, 20, and 21 in vitro, revealing promising
inhibitory activities against mPGES-1 in the low micromolar range (IC50 = 4.6 ± 0.2 µM for 15;
IC50 = 2.4 ± 1.0 for 20; IC50 = 2.4 ± 0.8 µM for 21), disclosing 6 as a new interesting chemical
core for the development of mPGES-1 inhibitors.
Thanks to the rapid and efficient synthetic route developed, these results pave the way for the
design of new inhibitors, firstly considering the possibility of attaching a large set of substituents,
exploiting the hydroxy groups introduced on the different available positions on ring A of 6.
Furthermore, different substitution patterns on the terminal aromatic function belonging to the
benzyloxy substituent will be also evaluated. Moreover, the data here shown will provide
valuable information for the design of further optimized derivatives as novel mPGES-1 inhibitors
for the treatment of cancer and inflammation.
27

ACCEPTED MANUSCRIPT
ASSOCIATED CONTENT
Supporting Information. NMR data, and docking models. This material is available free of
charge via the Internet at http://pubs.acs.org.
AUTHOR INFORMATION
Corresponding Author
*To whom correspondence should be addressed: bifulco@unisa.it, Telephone: +39 (0)89
969741, Fax: +39 (0)89 969602; pcampiglia@unisa.it, Telephone: +39 (0)89 969242, Fax: +39
(0)89 962804
Author Contributions
The manuscript was written through contributions of all authors. All authors have given approval
to the final version of the manuscript.
Funding Sources
This work was supported by the University of Salerno (Italy), and by the Associazione Italiana
per la Ricerca sul Cancro (AIRC) (grants IG 2012—IG_12777 and IG 2015—IG_17440 to
Bifulco Giuseppe).
Notes
ACKNOWLEDGMENT
G.L. acknowledges fellowship support from Associazione Italiana per la Ricerca sul Cancro
(AIRC) (grant IG 2012—IG_12777).
28