Discovery of Novel Potent Reversible and Irreversible Myeloperoxidase Inhibitors Using Virtual Screening Procedure

Article abstract of DOI:10.1021/acs.jmedchem.7b00285

The heme enzyme myeloperoxidase (MPO) participates in innate immune defense mechanism through formation of microbicidal reactive oxidants. However, evidence has emerged that MPO-derived oxidants contribute to propagation of inflammatory diseases. Because of the deleterious effects of circulating MPO, there is a great interest in the development of new efficient and specific inhibitors. Here, we have performed a novel virtual screening procedure, depending on ligand-based pharmacophore modeling followed by structure-based virtual screening. Starting from a set of 727842 compounds, 28 molecules were selected by this virtual method and tested on MPO in vitro. Twelve out of 28 compounds were found to have an IC₅₀ less than 5 μM. The best inhibitors were 2-(7-methoxy-4-methylquinazolin-2-yl)guanidine (28) and (R)-2-(1-((2,3-dihydro-1H-imidazol-2-yl)methyl)pyrrolidin-3-yl)-5-fluoro-1H-benzo[d]imidazole (42) with IC₅₀ values of 44 and 50 nM, respectively. Studies on the mechanism of inhibition suggest that 28 is the first potent mechanism-based inhibitor and inhibits irreversibly MPO at nanomolar concentration.

Full text of DOI:10.1021/acs.jmedchem.7b00285

Subscriber access provided by NEW YORK UNIV
Article
Discovery of Novel Potent Reversible and Irreversible
Myeloperoxidase Inhibitors using Virtual Screening Procedure
Jalal Soubhye, Ibaa Chikh Alard, Iyas Aldib, Martine Prévost, Michel Gelbcke, Annelise De Carvalho, Paul
G. Furtmüller, Christian Obinger, Joerg Flemmig, Sara Tadrent, Franck Meyer, Alexandre Rousseau, Jean
Neve, Véronique Mathieu, Karim Zouaoui Boudjeltia, Francois M U Dufrasne, and Pierre G Van Antwerpen
J. Med. Chem., Just Accepted Manuscript • DOI: 10.1021/acs.jmedchem.7b00285 • Publication Date (Web): 03 Jul 2017
Downloaded from http://pubs.acs.org on July 3, 2017
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 82
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Discovery of Novel Potent Reversible and
Irreversible Myeloperoxidase Inhibitors using
Virtual Screening Procedure
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
Jalal Soubhye(1)*, Ibaa Chikh Alard(2), Iyas Aldib(1), Martine Prévost(3), Michel Gelbcke(1),
Annelise De Carvalho(4), Paul G. Furtmüller(5), Christian Obinger(5), Jörg Flemmig(6), Sara
Tadrent(1), Franck Meyer(7), Alexandre Rousseau(8), Jean Nève(1), Véronique Mathieu(4) Karim
Zouaoui Boudjeltia(8), François Dufrasne(1)§*, Pierre Van Antwerpen(1,9)§*
(1) Laboratoire de Chimie Pharmaceutique Organique, Faculté de Pharmacie, Université
Libre de Bruxelles, Campus de la Plaine, Boulevard du Triomphe, 1050 Bruxelles, Belgium.
(2) Laboratoire de Pharmacie Galénique et Biopharmacie, Faculté de Pharmacie, Université
Libre de Bruxelles, Campus de la Plaine, Boulevard du Triomphe, 1050 Bruxelles, Belgium.
(3) Laboratoire de Structure et Fonction des Membranes Biologiques, Université Libre de
Bruxelles, Campus de la Plaine, Boulevard du Triomphe, 1050 Bruxelles, Belgium.
(4) Laboratoire de Cancérologie et Toxicologie Expérimentale, Faculté de Pharmacie,
Université Libre de Bruxelles, Campus de la Plaine, Boulevard du Triomphe, 1050 Bruxelles,
Belgium.
(5) Department of Chemistry, BOKU−University of Natural Resources and Life Sciences,
Vienna, Austria.
(6) Institute for Medical Physics and Biophysics, Medical Faculty, University of Leipzig,
Haertelstrasse 16–18, 04107 Leipzig, Germany.
(7) Laboratory of Biopolymers and Supramolecular Nanomaterials, Faculty of Pharmacy,
Université Libre de Bruxelles (ULB), Campus de la Plaine, Boulevard du Triomphe, 1050
Bruxelles, Belgium.
(8) Laboratory of Experimentral Medicine, CHU Charleroi, A. Vesale Hospital, and
Université Libre de Bruxelles, 6110 MontignyꢀleꢀTilleul, Belgium.
(9) Analytical Platform of the Faculty of Pharmacy, Université Libre de Bruxelles, Campus
de la Plaine, Boulevard du Triomphe, 1050 Bruxelles, Belgium.
1
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 2 of 82
1
2
Abstract
3
4
5
6
7
8
The heme enzyme myeloperoxidase (MPO) participates in innate immune defense mechanism
through formation of microbicidal reactive oxidants. However, evidence has emerged that
MPOꢀderived oxidants contribute to propagation of inflammatory diseases. Because of the
deleterious effects of circulating MPO, there is a great interest in the development of new
efficient and specific inhibitors. Here, we have performed a novel virtual screening procedure,
depending on ligandꢀbased pharmacophore modeling followed by structureꢀbased virtual
screening. Starting from a set of 727,842 compounds, 28 molecules were selected by this
virtual method and tested on MPO in vitro. Twelve out of 28 compounds were found to have
an IC50 less than 5 µM. The best inhibitors were 2ꢀ(7ꢀmethoxyꢀ4ꢀmethylquinazolinꢀ2ꢀ
yl)guanidine (28) and (R)ꢀ2ꢀ(1ꢀ((2,3ꢀdihydroꢀ1Hꢀimidazolꢀ2ꢀyl)methyl)pyrrolidinꢀ3ꢀyl)ꢀ5ꢀ
fluoroꢀ1Hꢀbenzo[d]imidazole (42) with IC50 values of 44 and 50 nM, respectively. Studies on
the mechanism of inhibition suggest that 28 is the first potent mechanismꢀbased inhibitor and
inhibits irreversibly MPO at nanomolar concentration.
9
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. Introduction
The heme enzyme myeloperoxidase (MPO, EC 1.11.2.2) is a lysosomal protein that plays an
important role in the human innate immunity system. It is expressed in neutrophils and stored
in their azurophilic granules1ꢀ5. After phagocytosis of pathogens by the neutrophils, MPO is
able to catalyze the H2O2ꢀmediated oxidation of chloride to the powerful oxidizing agent
hypochlorous acid (HOCl), which leads to the oxidation (degradation) of biomolecules of
pathogens in the phagosome. Other halide (Brꢀ, Iꢀ but not Fꢀ) or pseudoꢀhalide (SCNꢀ) ions can
also act as substrates for MPO. Native ferric MPO [PorꢀFe(III)] is oxidized by H2O2 to
Compound I [•+PorꢀFe(IV)=O, i.e. oxoiron(IV) porphyryl radical]. Compound I directly
oxidizes SCNꢀ, Iꢀ, Brꢀ or Clꢀ to the corresponding (pseudo)hypohalous acids (HOX) thereby
2
ACS Paragon Plus Environment

Page 3 of 82
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
restoring the ferric state (Figure 1)6,7. Other electron donors such as tyrosine, serotonin or
tryptamine are oxidized via a oneꢀelectron pathway resulting in the formation of Compound II
[PorꢀFe(IV)ꢀOH, i.e. oxoiron(IV)]. Finally, reduction of Compound II (which is outside the
halogenation cycle) restores MPO in the native state (Figure 1)8. Although the reaction of
Compound I with SCNꢀ, Iꢀ and Brꢀ is faster than with Clꢀ, HOCl is considered as the essential
reaction product of MPO because of the high Clꢀ concentration in human blood plasma
compared to the other species9.
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
In some cases, MPO is released from neutrophils, producing HOCl in the circulation which
results in oxidative damages of the host tissues. These damages sometimes contribute to the
development of injuries, e.g. in the kidney, the central nervous system, lung or the
cardiovascular system3. The close relation between MPO activity and cardiovascular diseases
prompted a study on the roles of the MPO/H2O2/Clꢀ system in atherosclerosis. It has been
reported that MPO can oxidize lowꢀdensity lipoproteins (LDLs) thereby promoting an
inflammatory response in monocytes and endothelial cells3. This inflammation results in the
formation of foam cells in the artery walls, thus generating a necrotic center of
atherosclerosis. Some evidences highlighted the role of MPO in the oxidation of highꢀdensity
lipoproteins (HDLs), causing a decrease in their capacity to remove cholesterol from
atherosclerotic lesions10. It is well documented that the MPO/H2O2/Clꢀ system also contributes
to decrease the concentration of NO•, and activates the protease cascades and fibrin
deposition. As a result, the dysfunction of endothelial cells causes vulnerable plaques 11.
The role of the MPO/H2O2/Clꢀ system in the inflammatory syndromes makes MPO a
promising target for the development of new antiꢀinflammatory agents12. Several MPO
inhibitors were already identified. Most of these compounds were obtained by drug design
and comprise 3ꢀaminoalkyl fluoroindole derivatives13, aromatic hydroxamates14 and
indazoles15. Figure 2 illustrates the most potent MPO inhibitors discovered until now that can
3
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 4 of 82
1
2
3
inhibit MPO at the nanomolar range.
4
5
6
7
8
By now, the structures of human MPO unbound or in complex with fairly small anions
(halides, thiocyanate, cyanide) or covalentlyꢀbound to 2ꢀthioxanthine have been determined16,
17. The crystal structure of salicylhydroxamic acid (SHA) bound to human MPO has also been
described though no structure has been deposited in the PDB18. Recently, the Xꢀray structure
of human MPO complexed to 2ꢀ((3,5ꢀbis(trifluoromethyl)benzyl)amino)ꢀNꢀhydroxyꢀ4ꢀoxoꢀ
1,4ꢀdihydropyrimidineꢀ5ꢀcarboxamide (compound 1) has been solved (Figure 2)14. These
structural data were used in the docking experiments for three main purposes: 1) to predict the
interaction between the inhibitors and the active site 2) to implement structure activity
relationship (SAR) for designing potent MPO inhibitors19 and 3) to find ligands selected by
9
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
virtual screening from available databases20, 21
.
Up to now, two virtual screening experiments were done to determine new MPO inhibitor
scaffolds. The first one performed by Malvezzi et al.22 using GOLD docking program
screened the full Zinc database. The second was carried out by Aldib et al.20 using Glide® as
a software and ASINEX® as the ligand database. In our paper, four pharmacophore models
were designed based on the most potent MPO inhibitors using LigandScout®23ꢀ25. These
pharmacophore models were used to screen a filtered Zinc database. Compounds selected
from this virtual screening were docked into the heme cavity of MPO by means of the
LeadIT® docking program26 to screen the Zinc database for potential ligands27. The best
compound was selected as a hit, followed by the synthesis and evaluation of several
analogues.
2. Results
Pharmacophore design
4
ACS Paragon Plus Environment

Page 5 of 82
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
In this study, we applied a fourꢀstep computerꢀaided protocol to select compounds to be tested
in vitro (Figure 3). The whole process included successively a Lipinski’s rule filter, a
screening based on the derivation of a pharmacophore, a docking screening and a selection
based on various drugꢀrelevant properties (Figure 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
First, the 727,842 compounds of the Zinc database molecules (http://zinc.docking.org) were
screened following the Lipinski's rule filter to select drugꢀlike molecules using DruLiTo
software (ref : (http://www.niper.gov.in/pi_dev_tools/DruLiToWeb/DruLiTo_index.html).
This filter selected 494,915 compounds. In the second step, pharmacophore models, featuring
the 3D key chemical determinants required for binding to MPO, were generated using the
structural data of the seven most potent MPO inhibitory compounds known so far (Figure 2).
These molecules have been used alone or in combination. When each potent inhibitor was
used individually to generate pharmacophore models, a small number of compounds from the
494,915 drugꢀlike molecules previously filtered matched the pharmacophore features (less
than 25 compounds in total from the seven pharmacophores). In contrast, when using three or
more potent inhibitors to define a pharmacophore model, a large number of compounds
resulted from the screening (more than 20,000 compounds in all cases). Finally using all
possible pairs combining the seven most potent inhibitors, 21 pharmacophore models were
obtained which featured a pharmacophoreꢀfit score ranging from 50.9 to 68.4 %. About 11000
hit molecules were then retrieved from the pharmacophore screening against the 494,915
compounds of this filtered database.
Only the pharmacophore models that have a specificity ≥ 0.5 were kept, providing a number
of compounds less than 500 for each model. Four models met this criterion (named model 1
to 4) as they contain, 273, 185, 106, and 6 molecules respectively (570 compounds in total)
(Figure 3). They were obtained from the following pairs of active potent inhibitors: 613 and
428 for model 1, 114 and 613 for model 2, 4 and 2 for model 314 and 729 and 1 for model 4
5
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 6 of 82
1
2
3
4
5
6
7
8
9
(Figure 4). Model 1 comprises one hydrogen bond acceptor (HBA), one hydrogen bond
donor (HBD), one hydrophobic moiety (H), one positively charged group (PI), and 40
exclusion volumes (XVOLs). Model 2 contains one HBA, one aromatic group (AR), one
HBD, and 33 XVOLs. Model 3 encompasses two H sites, one HBA, and 41 XVOLs. Finally,
model 4 contains two HBA, two HBD, one AR, and 37 XVOLs (see Figure 4, Figures S3,
and Table S1).
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
Docking calculations
In the next stage, compounds issued from the pharmacophore screening procedure were
docked on the Xꢀray structure of human MPO complexed with thiocyanate (PDB 1DNU)
using the FlexX software.
To validate the docking procedure, compound 1 which has wellꢀestablished interactions with
MPO was docked (Figure 5). The bestꢀscore docked pose of 1 featured the pyrimidine ring
stacked on the pyrrole ring D of the heme, and the trifluoromethyl aromatic ring occupying
the hydrophobic pocket at the entrance of the active site. The hydroxamic acid group lied in
the center of the distal cavity, and both the carbonyl and hydroxyl oxygens formed hydrogen
bonds with Gln91 and His95 residues. One of the pyrimidine nitrogens provided the main
interaction with the heme propionate group, whereas the oxygen of the hydroxyl group was
well positioned for a hydrogen bond with Asp94. The trifluoromethyl aromatic ring of the
molecule extended toward the surface of the enzyme. These interactions formed in the docked
pose of 1 are quite similar to those in the crystallographic structure14 where its lowest rootꢀ
mean square deviation computed on all heavy atoms (RMSD) was 0.39 Å, allowing for the
validation of the docking procedure (Figure 5A). Compounds 6, 7 and 4 were also docked
according to our procedure. The best pose of these ligands featured the same set of
interactions as those predicted with Glide software13, 28 (Figure 5). The fluorophenyl moiety
6
ACS Paragon Plus Environment

Page 7 of 82
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
of 4 occupied the hydrophobic pocket at the entrance of the active site. Noticeably, the
fluorophenyl group of 1 defined as an AR in the pharmacophore models 2 and 4 was not the
aromatic moiety, the pyrimidine, making the stacking with the pyrrole subunit of heme in the
Xꢀray structure or the docking (Figure 5). The indole and benzodioxole groups of 6 and 4,
respectively, were defined as H groups in the pharmacophore model 1 for 6 and in
pharmacophore models 1 and 3 for 4, where they formed a πꢀπ interaction with the pyrrole
ring in the docking experiments. In contrast, the indole groups of 6 and 7 were presented as
AR groups in pharmacophore models 2 and 4, respectively, indeed make aromatic interactions
in their docked poses.
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
From the docked compounds issued from the pharmacophore screening procedure, only
ligands featuring a docking score < ꢀ8.0 kcal/mol were kept. These ligands represent the 10 %
best scoring molecules from models 1, 2 and 3 (24, 20, 12 and 3 compounds, respectively)
and the 50% best scoring molecules from model 4 (4 molecules). The inspection of the
docked poses of these selected molecules showed that most of them make interactions with
Glu102, Thr100 and/or Gln91.
Toxicological Evaluation and DrugꢀLike Profile
Finally, the OSIRIS® property explorer tool (www.organicꢀchemistry.org/prog/peo) was used
to predict drugꢀlikeness, toxicity risk, mutagenicity, irritant, and tumorigenic effects of the
selected molecules. Only the compounds passing this filter were considered for future studies
(Figure 3). From this selection, 16, 8, 5 and 1 compounds from models 1 to 4 successfully
passed through the last filtering step, respectively. Two of these compounds were repeated in
two models to give a total of 28 molecules.
Validation of the screening procedure
7
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 8 of 82
1
2
3
4
5
6
7
8
9
To assess the specificity and the sensitivity of this protocol, a set of 20 molecules of various
chemical structures containing 10 active and 10 inactive compounds on MPO was chosen
from the literature (the results of the method validation are illustrated in Table 1). All the
active compounds were selected by Lipinski’s filter and OSIRIS explorer (except 14 which
were derived from dihydroꢀnaphtalene). As a result, the pharmacophore screening procedure
selected all active compounds except thioxanthine derivative, with a total selectivity of 90%
and nine out of ten active compounds have passed the docking filter. In addition, regarding
the 10 inactive molecules, none passed all screening filters indicating that these filters can
retain possible inactive compounds.
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
Finally, the selected molecules were filtered for Pan Assay Interference Compounds (PAINS)
by the online filter (http://zinc15.docking.org/patterns/home/). Only 34 and 51 did not pass
this filter (Table S4).
Inhibition of MPO Chlorination Activity
After the selection of compounds through virtual screening, their activities were assessed by
the taurine chloramine assay12 and their IC50 values were determined (Table 2). The in vitro
tests revealed that 12 compounds have an IC50 smaller than 5 µM. Among them, 9, 2 and 1
compounds belonged to the pharmacophore models 1, 2 and 4, respectively (no molecule
associated to model 3). The best activities were found for model 1.
Compounds 28, 42 and 55 have attracted our attention owing to their low IC50 ranging from
44 to 130 nM. Remarkably, they displayed docking features similar to those of 5, 6 and 4, in
particular an ionic interaction of their ammonium or guanidinium group with Glu102 (see
Figure 6). Among the tested candidates, 5 drugs were found to have IC50 values less than 1
µM including 36, 41, 43 and 48 (Table 2). Other drugs were of particular interest such as the
fluoroquinolones (29 and 30) due to good predicted interactions with the active site of the
8
ACS Paragon Plus Environment

Page 9 of 82
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
enzyme, and the absence of inhibition. Pharmacophore modeling and docking experiments
predicted that 38, 40, 44, 47, 49 and 51 had a high potency, but their inhibitory properties
were low. These results suggest that other parameters such as the redox properties (of some of
these compounds) play an important role in the inhibition of MPO, emphasizing the limitation
of the virtual screening protocol29.
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
In the light of these results, we decided to use Glide, another docking program20, to dock the
12 active compounds in order to combine information about the binding modes predicted by
both software. Overall the binding modes are rather comparable with, in most cases, the same
aromatic moiety of each compound found to be stacked onto the pyrrole ring (Table S2).
There are however a few exceptions such as compounds 34 and 43 with IC50 values of 1.15
and 0.35 µM, respectively, which feature a stacking of an aromatic moiety onto the heme in
the docked pose produced by Glide though no stacking is observed with FlexX. In contrast the
phtalazine structure of compound 36 (IC50 = 0.9 µM) is found to bind in a parallel mode on
the heme pyrrole with FlexX only whereas Glide favors the interaction of the hydrazine group
with Glu102.
Virtual evaluation of the active 12 compounds on mutated MPO (Glu102→Gln102)
In order to buttress the conjecture that this inhibitor indeed binds the MPO active site, the 12
most active compounds were docked in MPO with, Glu102, one residue shown by docking to
interact with several inhibitors, substituted in silico into Gln (Figures S5). For most
compounds carrying a positive charge, their binding modes predicted by Glide differ in the
MPO variant compared to the wildꢀtype and sustain the role of Glu102 in contributing to the
binding of those compounds in the active site. In contrast, the binding mode of most
compounds carrying no positive charge is not affected by Glu102 substitution.
Steadyꢀstate Activity Studies on the Mechanism of Inhibition
9
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 10 of 82
1
2
3
4
5
6
7
8
9
Myeloperoxidase inhibitors are classified in three main groups, namely (i) irreversible
inhibitors (e.g. thioxanthine), (ii) compounds that promote accumulation of Compound II
without binding to the active site (e.g. dapsone) and (iii) compounds that bind reversibly to
the active site of MPO (e.g. SHA). Compounds from the second category may lose their
inhibitory effects in a physiological environment because of the abundance of better
peroxidase substrates such as ascorbate and tyrosine that restore ferric MPO by reduction of
Compound II14. In order to decipher the mechanism of action of these inhibitors, ligands with
IC50 < 2.5 µM (Compounds 28, 34, 35, 36, 41, 42, 43, 48, and 55) were selected and their
interaction with MPO was studied in the absence and presence of hydrogen peroxide. In
practice, after 10 minutes of incubation, the enzyme was diluted 100ꢀfold, and the residual
chlorination activity was measured. Each inhibitor alone caused about 10ꢀ30 % inactivation of
MPO, whereas hydrogen peroxide alone minimally affected activity. However, 28 or 36 in the
presence of hydrogen peroxide caused an almost complete loss in enzyme activity. Chloride
did not prevent enzyme inactivation by these compounds (Table 3). By contrast, combination
of the other compounds with hydrogen peroxide did not cause more than 30 % inhibition.
These data clearly suggest that MPO uses hydrogen peroxide to oxidize 28 and 36, thereby
irreversibly inactivating the enzyme17.
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
Furthermore, we added 5 µM of 28 (a model of irreversible inhibitor) and 43 (a model of
reversible inhibitor) to MPO in the presence of 30 µM of guaiacol. After incubation with or
without hydrogen peroxide for 20 minutes, the reaction mixture was diluted 200 times and the
residual peroxidase activity was assessed by following the increase of absorbance at 470 nm
for 10 min14. Figure 7 demonstrates that the activity of MPO in the presence of 28 cannot be
recovered by dilution as long as H2O2 was present. In the absence of H2O2, there was no
inhibition of MPO, clearly underlining that 28 is an irreversible mechanismꢀbased inhibitor.
By contrast, the effect of metoclopramide on the residual activity of MPO both in the
10
ACS Paragon Plus Environment

Page 11 of 82
Journal of Medicinal Chemistry
1
2
3
4
5
presence and absence of H2O2 was similar to that of DMSO, indicating that 43 acts as
reversible MPO inhibitor (Figure 7A).
6
7
8
9
In order to see whether 43 can bind at the active site with a relatively low dissociation rate,
another reversibility test was done. Tyrosine was used as a MPO substrate that converts MPO
Compound I into Compound II and Compound II into native MPO, acting as a substrate for
Compound I and II. DMSO was employed as a negative control while 43 in absence of
tyrosine was used as a positive one. After adding 10 µM H2O2, the MPO activity was
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
monitored by measuring the absorbance of 3,3′,5,5′ꢀtetramethylbenzidine (TMB) during 10
minutes. As shown in Figure 7B, 43 can inhibit MPO even in the presence of tyrosine
indicating that the inhibitor binds the active site, preventing tyrosine to enter the active pocket
and oxidize Compound II to ferric enzyme.
TransientꢀState Kinetic Studies
Next, we probed the interaction of 5 compounds with the relevant redox intermediates of
MPO, i.e. Compound I and Compound II (Figure 1)9, by using multimixing stoppedꢀflow
spectroscopy. In detail, we aimed to study the mechanism of interaction involving irreversible
inhibitors 28 and hydralazine, reversible inhibitor 43 and inactive compounds 40 and 29.
Compound I was preꢀformed and probed for interaction with these molecules. Concerning 28
and hydralazine, an unusual behavior was observed (Figure 8 and S2). As with most
inhibitors, Compound I was reduced via a oneꢀelectron reaction to Compound II, reflected by
an increasing absorbance at 456 and 632 nm (Figure 8A and C). These reactions were fast
with an apparent bimolecular rate constant k3 of 1.98 X 106 Mꢀ1 sꢀ1 and 1.7 X 106 Mꢀ1 sꢀ1 for 28
and 36, respectively (Table 4). Finally, the Compound II spectrum was converted
(independently of the concentration of 28) (Figure 8D) to a spectrum with a Soret maximum
at 432 nm (accompanied by hypochromicity) and peaks at 570 and 626 nm. This spectrum
11
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 12 of 82
1
2
3
cannot be assigned to native MPO or Compound III (Figure 8B), but, interestingly, resembles
4
5
6
that observed for MPO with covalently bound 2ꢀthioxanthine17.
7
8
9
With 40 and 43, there was a direct and fast transition of Compound I to Compound II (Soret
maximum at 456 nm) with clear isosbestic points (Figure 9A). The reactions were
monophasic (inset to Figure 9A) and from the slope of the linear plot of kobs values versus
inhibitor concentration, the apparent bimolecular rate constant (k3) of Compound I reduction
was calculated (Figure 9C). Both molecules are good electron donors for Compound I (Table
3). As typical reversible inhibitors, 43 reduces Compound II back to ferric MPO (Figure 9B
and D) at a slow rate (k4 = 180 Mꢀ1 sꢀ1). The ratio of k3/k4 is high and as a consequence
Compound II accumulates thereby reducing the chlorination activity of MPO.
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 behavior of the two inactive molecules 40 and 29 was completely different. The
successive transformations Compound I → Compound II →ferric MPO in the presence of 40
was very fast. The calculated apparent bimolecular rate constants of both reactions were k3 =
5.2 × 106 Mꢀ1 sꢀ1 and k4 = 1.7 ×104 Mꢀ1 sꢀ1. Therefore, 40 does not mediate Compound II
accumulation and thus (reversible) inhibition of the chlorination activity (Table 4). By
contrast, 29 was shown to be a bad electron donor for Compound I (k3 = 3.1 × 104 Mꢀ1 sꢀ1) and
Compound II reduction to the ferric state could not be measured. However, there was a
steadyꢀstate shift to Compound II similar to that seen with 28 and 36 (Figure S8). However,
in contrast to the latter, 29 reacts very slowly with Compound I (Table 4) and thus cannot
compete with chloride (140 mM) for Compound I reduction.
Determination of partition ratio for mechanismꢀbased inhibitors
The transientꢀstate kinetic studies on 28 and 36 together with the steadyꢀstate activity tests
indicated that these ligands are mechanismꢀbased irreversible inhibitors of MPO. The
mechanism of irreversible interaction might include modification of the prosthetic group (e.g.
12
ACS Paragon Plus Environment

Page 13 of 82
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
covalent bond formed between inhibitor and heme) or of the protein, whereas a reversible
inhibitor (e.g. 43) is released from the (unmodified) enzyme. Here, we probed the impact of
different ratios of concentration of [inhibitor]/[MPO] on the remaining MPO activity. Figure
10 demonstrates that the remaining MPO activity decreased as a linear function of
[inhibitor]/[MPO] ratio and completely disappeared at [inhibitor]/[MPO] ratios of 67 and 110
for 28 and 36, respectively. This suggests that on average one individual enzyme is able to
cycle 67 times before being irreversibly inhibited by 28. A completely different picture was
obtained with the reversible inhibitor 43 (Figure 10) as no decrease of remaining MPO
activity at increasing [43]/[MPO] ratios is observed.
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
Inhibition of LDL oxidation
Finally, we investigated the effect of these compounds on the MPOꢀmediated oxidation of
LDL by using an enzymeꢀlinked immunosorbent assay (ELISA) based on a mouse
monoclonal antibody (Mab AG9) that specifically recognizes MPOꢀoxidized ApoBꢀ100 on
LDL. The obtained IC50 values are summarized in Table 3. Most of these molecules inhibit
LDL oxidation at low µM concentrations. The hierarchy of IC50 values correlated with that of
the taurine chlorination inhibition assay, but the absolute IC50 values obtained from the LDL
oxidation assay were slightly increased. 28 exhibits the best LDL oxidation inhibitory activity
with a calculated IC50 value of 90 nM. Interestingly, in this assay 43 did not show inhibitory
activity.
Oxidation products of 36 and 28
The oneꢀelectron oxidation of 36 by horseradish peroxidase (HRP) is known to produce
hydralazyl radicals which then decompose to form various products or react with molecular
oxygen to generate reactive oxygenꢀcentered radicals38. Here, we incubated MPO with 36 in
phosphateꢀbuffered saline (PBS), started the reaction by adding H2O2 and analyzed the
13
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 14 of 82
1
2
3
4
5
6
7
8
reaction mixture by LCꢀHRMS after 60 min of incubation. Phtalazine (MH+ = 131.0606, error
= ꢀ1.8 ppm) was the only detected reaction product (Figure S10A). Upon H2O2ꢀmediated
oxidation of 36 by HRP, this intermediate was also found as main reaction product.
9
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
In case of 28, two reaction products were detected by LCꢀHRMS. Regarding the first
derivative, the amino group of the guanidine was eliminated giving N'ꢀ(7ꢀmethoxyꢀ4ꢀ
methylquinazolinꢀ2ꢀyl)formimidamide (28ꢀOXꢀ1) (MH+ = 217.0926, error = 0.42 ppm)
(Figure S10B). The second detected compound was formed via the elimination of diamineꢀ
methylene group from the side guanidine group to give 2ꢀaminoꢀ7ꢀmethoxyꢀ4ꢀ
methylquinazoline (28ꢀOXꢀ2) (MH+ = 190.0973, error = 1.6 ppm) (Figure S10C). It is
worthy to note that phtalazine and 28ꢀOXꢀ1 could not be detected when the experiment was
performed without adding H2O2, while the metabolite 28ꢀOXꢀ2 was also found in the absence
of H2O2, although in very low amounts.
Optimization of 28 Derivatives
Following the inhibition of taurine chlorination, compound 28 provides the best activity. This
result prompted us to improve its activity. A series of Nꢀ(4ꢀmethylꢀquinazolinꢀ2ꢀyl)ꢀ
guanidines (56ꢀ62) was designed based on the features of the docking poses of 28. The
introduction of halogen atoms (F, Cl or Br) at positions 6 and 7 as well as displacement of
methoxy group for hydrogen was probed by docking experiments (structures provided in table
5). It has resulted that candidate 56 (without HBA group) featured a shifted stacking pose,
while compounds 58 and 59 (Br or Cl on position 6) had no stacking pose. Ligands 57 and 60
with fluorine on position 6 and 7 show the same interactions as 28 but with a lower affinity.
Subsequently, Nꢀ(4ꢀMethylꢀquinazolinꢀ2ꢀyl)ꢀguanidine analogues were synthesized through
Skraup reaction by incorporating the relevant aniline at a high dilution level, which led to the
14
ACS Paragon Plus Environment

Page 15 of 82
Journal of Medicinal Chemistry
1
2
3
corresponding 1,2ꢀdihydroꢀ2ꢀmethylquinoline. Treatment with cyanoguanidine gave the
4
5
6
substituted Nꢀ(4ꢀmethylꢀquinazolinꢀ2ꢀyl)ꢀguanidines 56ꢀ62 (Scheme 1)39.
7
8
9
These Nꢀ(4ꢀmethylꢀquinazolinꢀ2ꢀyl)ꢀguanidine derivatives were assayed by the taurine
chloramine test to determine the respective IC50 values (Table 5). The most active molecules
were 60 and 28 that possessed F and MeO substituents, respectively, at position 7. In addition,
halogen atoms at the same position conferred a higher activity than compounds with a
halogen atom at position 6. Finally, the absence of a functional group on the aromatic ring of
56 resulted in the lowest activity.
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
Compounds in Table 5 showed a variation of activity as a function of halogen atom following
the order: F > Cl > Br. Docking experiments shed light on the in vitro results (Figure 11). All
compounds from 60 to 62 functionalized with a halogen atom at position 7 featured stacking
pose while those substituted with Br or Cl at position 6 featured a shifted stacking pose (58
and 59). However, all Nꢀ(4ꢀmethylꢀquinazolinꢀ2ꢀyl)ꢀguanidine derivatives interacted with
Glu102 and propionate of the active site through the guanidinium group. Regarding 60, 61
and 62, Thr100 appeared to be part of the anchoring site of MPO.
The effects of compounds 28 and 42 on human neutrophils
In comparison to the HOClꢀproducing activity observed in the absence of the inhibitor (the
value was set to 100%) no clear inhibitory effect could be found when the concentration of
compound 28 was up to 20 µM. In fact, while the addition of 1 or 5 µM of 28 led to a highly
significant reduction of the relative chlorinating activity to about 67 – 71 %, the addition of
10 – 20 µM of the inhibitor, the relative HOClꢀproducing activity of MPO was even slightly
higher than in the negative control. In contrast, starting at 50 µM of 28, a constant drop in the
relative chlorinating MPO activity was observed. At the highest tested inhibitor concentration
15
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 16 of 82
1
2
3
(1 mM), about 20 % of the original relative HOCl production was observed. Based on the
4
5
6
obtained data, an IC50 value of about 93.1 µM was calculated (Figure 12).
7
8
9
For compound 42, up to 50 µM of the inhibitor, no clear effect on the relative chlorinating
MPO activity was observed. Yet at higher inhibitor concentration, a clear concentrationꢀ
dependent decrease in the relative HOCl production rate was detected. At 2 mM of 42, the
relative MPO activity dropped to about 8.6 % of the value observed in the absence of the
inhibitor. From the data, an IC50 value of about 313.6 µM was calculated (Figure 12).
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
ADME/T properties
Absorption, distribution, metabolism, excretion and toxicity (ADME/T) experimental
properties of the most active compounds were obtained from admetSAR®
(http://lmmd.ecust.edu.cn:8000/) database40. It is noteworthy that 4 selected compounds
during the docking were drugs and were expected to be safe. However, in the absence of data,
the properties were predicted using the same software. The predicted and experimental
ADME/T properties of the active compounds are listed in Table 6. All the compounds can be
absorbed from the intestinal tract as suggested by the LogP values ranging between 0.54 and
1.06 (Table S6). In addition, the toxicity values of our compounds are rather low possibly due
to the OSIRIS filter which excluded the molecules with high risk of toxicity.
The in vitro growth inhibition of the active compounds in normal human dermal fibroblast
(NHDF) was assessed by MTT assay to evaluate the potential cytotoxicity of the new
inhibitors in a cellular model. The results are shown in Table 6 and demonstrate that
compound 28 inhibits the growth of these cells at a concentration (IC50) of 17 µM, which is
about 400 times higher than the concentration of the MPO inhibitory effect (0.044 µM). The
other compounds do not seem to be cytotoxic, at least up until 50 µM (Figure S6A).
16
ACS Paragon Plus Environment

Page 17 of 82
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Furthermore, the effect of compounds 28 and 42 on the vitality of neutrophil was assessed. It
is found that up to 1 mM, 28 did not significantly increase the apoptosis rate in the
neutrophils. Still at concentrations of 500 µM or higher, the FSC/SSC plots indicate a small
tendency for a higher necrosis rate in the cells (not shown). Again no significant effect of
compound 42 on the apoptosis rate was observed in the tested concentration range (Figure
S6B).
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
Effect of MPO inhibitors on lactoperoxidase
In order to assess the specificity, the activity of the potent compounds was also tested on
lactoperoxidase (LPO) which is a homologous human peroxidase41 (Table 7). Compounds 34,
36 and 41 showed similar activity on both LPO and MPO, whereas compounds 28, 60 and 42,
which represent the best irreversible and reversible inhibitors of MPO, showed a 100ꢀtimes
reduced activity on LPO.
Because of the small molecular weight of the active compounds, these might be active also on
other targets (enzymes or receptors). The information obtained from PubChem database
showed that 28 and its analogs have no activity on mammalian enzymes or receptors,
however, on Plasmodium falciparum 3D7. Information about 39, 42, 45 and 55 activity on
different targets is available in PubChem database (Table S5).
3.
Discussion
The present study aimed at evaluating a new virtual screening approach consisting of ligandꢀ
based pharmacophore modeling and structureꢀbased virtual screening in order to design new
potent MPO inhibitors.
Four pharmacophore models were generated to screen a subset of the Zinc database
containing compounds that match the Lipinski’s rules. The pharmacophore model obtained
17
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 18 of 82
1
2
3
4
5
6
7
8
9
using 6 and 4 provided the largest pool of active molecules: 9 hits out of 15 tested compounds
displayed IC50 less than 5 µM, five of them being active at nanomolar concentrations.
According to the docking experiment, this model highlighted the specific interactions of each
candidate with the active site. The ionizable basic groups in both 6 and 4 form the same
interaction in the active site, i.e. a salt bridge with Glu102. In addition, their aromatic groups
form πꢀπ interactions with the heme. The presence of these two types of interaction may
explain the success of pharmacophore model 1 as the other 3 models do not feature aromatic
and amine groups in both compounds. Furthermore, the HBD or HBA functions of the potent
inhibitors exhibit a different interaction with the enzyme (i.e. the HBD of 6 and 7 forms
hydrogen bond with Glu102, while that of 1 faces the heme iron). This observation could be
rationalized by considering that the HBD group in 1 contained in a hydroxamic acid group
that is known to have an excellent affinity for heavy metals such as iron, whereas the amino
group of 6 and the amide group of 7 do not possess this property.
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
All the ligands issued from our protocol share an aromatic ring, a HBA on the aromatic group
and a HBD or a positively charged group at the alkyl chain. Following the docked poses, the
aromatic moiety of most ligands is involved in a πꢀstacking with the pyrrole D of the heme,
and their positively charged group or HBD interacts with Glu102. Noticeably several
compounds selected from the virtual docking based on their scoring values showed a low
activity and some of them even had none (i.e. 29, 30, 38, 40, 44, 46, 47, 49 and 51). Other
compounds such as serotonin7 and 5ꢀcarboxytryptamine29 were also previously reported to be
inactive despite their good affinity9. These observations were explained by prevailing
weaknesses of the scoring function used in the docking programs to estimate the experimental
affinities19. This is indeed observed here as the ꢀG score values and the activity of the
inhibitors correlate poorly. Another possible cause for this poor correlation found for some
compounds could be rationalized by their potential to react with MPO which cannot be
18
ACS Paragon Plus Environment

Page 19 of 82
Journal of Medicinal Chemistry
1
2
3
4
5
modeled by the docking. Despite these weaknesses, our procedure produced 21 active out of
24 tested compounds (88%), and 12 had an IC50 less than 5 µM (50%).
6
7
8
9
A kinetic study based on active (28, 36 and 43) and inactive (40 and 29) ligands was carried
out in order to understand the mechanism of inhibition of the active inhibitors and the absence
of activity of the others. In general, a compound is endowed with a high activity when it is
able to react fast with Compound I and slowly with Compound II leading to the accumulation
of Compound II which does not participate in the halogenation cycle (k3>>k4)7, 27. This
classical phenomenon was found for 43 (Figure 9A and B)(Table 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
Initially, also oxidation of 28 and 36 follows the classical peroxidase cycle. Both molecules
are oxidized by Compounds I and II, however, the resulting reaction products (i.e. radicals)
apparently shift the enzyme out of the peroxidase cycle into the Compound III state.
Normally, the latter can only be formed from ferric or ferrous MPO with superoxide (k6) or
dioxygen (k7) respectively (Figure 1). It seems that the substrate radicals react with ferric
MPO and reduce it to ferrous enzyme (k5) or alternatively, reduce dioxygen to superoxide that
reacts with ferric MPO to Compound III. Only partial recovery of ferric MPO after H2O2
consumption demonstrated that these molecules are mechanismꢀbased inhibitors since they
irreversibly inhibited MPO in a H2O2ꢀdependent reaction. Irreversible inhibition was also seen
in peroxidase activity assays in the presence of competing electron donors like guaiacol.
Complete inhibition of MPO by 28 and 36 is achieved by a [inhibitor]/[MPO] ratio of 67 and
110, respectively. In contrast, all other tested inhibitors showed a completely different
behavior, i.e. they followed the peroxidase cycle and their inhibitory activity did not depend
on the presence of hydrogen peroxide.
In the case of 40, the values of k3 and k4 did not show a sufficient k3/k4 ratio to promote
accumulation of Compound II and as a result the ligand has no activity. 29 slowly reacts with
19
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 20 of 82
1
2
3
4
5
6
7
8
9
Compound I (k3 at the order of 104 Mꢀ1sꢀ1), and consequently, cannot compete with the natural
electron donor, chloride anion. Thus, our kinetic results showed that the inactive compounds
selected by our screening method either react in a very fast way with both Compound I or II
or are poor substrates of MPO. This indicates that a potent MPO inhibitor must be a good
electron donor for Compound I capable of shifting the activity of the enzyme to the
peroxidation cycle. In some cases, the subsequent products of this reaction further react with
the redox intermediates of MPO to irreversibly block the enzyme as 28 and 36 do.
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
In general, predictions of biological activities based on docking experiments are below 50%
for most of the tested enzymes42. For MPO, this was shown to drop to 10%20. Our study
suggests that the ligandꢀbased pharmacophore modeling may contribute to improving the
results (50% with our procedure versus 10% with structureꢀbased virtual screening),
especially when model 1 is used due to the similarity of the chemical and physiochemical
properties of both tested inhibitors (6 and 4). It has been demonstrated that the compound
must have one oxidable group in order to promote Compound II formation43. Ligandꢀbased
pharmacophore modeling may contribute in selecting compounds with such groups. Malvezzi
et al used also ligandꢀbased pharmacophore modeling as a virtual screening procedure to find
new MPO inhibitors. However, they used only SHA as a pharmacophore model and
discovered nine compounds derived from phenylhydrazide (4 compounds with IC50 < 5
ꢁM)20. Despite the fact that the hit rate of their virtual screening was high compared to
structureꢀbased virtual screening20, all were derived from the same scaffold. Notably, to reach
the heme cavity of MPO, compounds have to go through a rather narrow channel lined with
hydrophobic patches. In that respect, our procedure which included a Lipinski’s rule filter
may have improved the results by selecting the compounds that have suitable molecular mass
and lipophilicity.
20
ACS Paragon Plus Environment

Page 21 of 82
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
It has been documented that MPOꢀdependent LDL oxidation plays an important role in
atherogenesis. In addition, there are numerous pieces of evidence that the complex of MPO
with ApoBꢀ100 of LDL is mainly formed through electrostatic interactions and modifies the
chemical and steric requirements for efficient penetration of substrates and/or inhibitors into
the binding site41. In this context, the inhibiting potency of the new inhibitors on LDL
oxidation carried out by the MPO/H2O2/Clꢀ system has been assessed. The results
demonstrated that the compounds with IC50 > 5 µM cannot inhibit the oxidation of LDL
carried out by MPO. This decrease in inhibitory activity in the LDL assay compared to the
taurine test may be caused by the hindrance of the catalytic site during binding to LDL,
thereby providing some protective effect42. The heme group of MPO that is responsible for
the enzymatic activity is indeed located in a distal hydrophobic cavity with a narrow ovalꢀ
shaped opening. Hence, the formation of a complex macromolecular structure between MPO
and LDL may modify the chemical and steric requirements for good penetration of substrates
and/or inhibitors into the heme cavity45. Parameters such as charge and lipophilicity of
compounds have already been evoked to explain similar results19.
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 molecular meachanism of modification of MPO by 36 and 28 remains elusive.
Upon oxidation by Compounds I and II both molecules are oxidized to radicals. In the case of
36, the primary amine is oxidized38 followed by transfer of the radical site to the aromatic ring
and loss of the diamino group. Then, either the resulting intermediate loses another electron,
forming a stable phtalazine, or might react with the heme group leading to irreversible
modification and inhibition of MPO (Figure 13A). A similar oxidation pathway can be
proposed for 28. The corresponding oneꢀelectron oxidation product could lose one amino
group or a diamineꢀmethylene group forming two radical intermediates which in turn react
with the heme or lose another electron to give stable molecules (Figure 13B).
21
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 22 of 82
1
2
3
4
5
6
7
8
9
In order to further optimize the activity of 28 (the best hit and the mechanismꢀbased inhibitor)
several derivatives were synthesized and tested. The results showed that a halogen atom on
the aromatic ring is very important for the inhibitory activity. According to docking, this
group features a hydrogen bond with His95 and/or Gln91 (Figure 11). However, this
interaction is achieved only when the HBA is on the position 7 of the quinazoline. In addition,
the activity increase follows the hierarchy F > Cl > Br. This can be explained by an increase
of the volume from F to Br and/or increasing the electronegativity from Br to F.
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
Of course inhibition of MPO might dampen the antimicrobial activity of MPO during
phagocytosis of neutrophils. In fact, the effects of compounds 28 and 42 on human
neutrophils indicate that these compounds also efficiently inhibit intracellular MPO. Yet the
almost 104ꢀ and 105ꢀtimes higher IC50 values for compounds 28 and 42 strongly indicate a
much weaker effect on intracellular MPO than on the extracellular enzyme, maybe due to
restricted membrane permeability. Thus these two inhibitors have only limited effect on the
antimicrobial activity of neutrophils.
The predicted and experimental ADME/T profiles and the selectivity in other peroxidases of
the new inhibitors make these molecules promising hits. Four of these inhibitors are
commercially available drugs with known ADME/T profiles. Compound 43 which is
commonly used to treat and prevent nausea and vomiting has an IC50 of 0.35 µM on MPO. It
has been demonstrated that the plasma level of this safe drug (pregnancy category: B) after
one treating oral dose is 0.2ꢀ0.4 µM which is of the same order as its MPO46. In addition, it is
found that MPO inhibitors may be useful in cardiovascular diseases (CVD). Ruggeri et al
reported that compound 12 is useful for treatment of CVD34. However, it is found in this
study that 36, used for CVD in the cases of high blood pressure and heart failure, has an
activity 2ꢀtimes higher than 12. The concentration of 36 in plasma after one oral dose was
found to be 3.8ꢀ6.7 µM which corresponds to its IC50 value on MPO activity47.
22
ACS Paragon Plus Environment

Page 23 of 82
Journal of Medicinal Chemistry
1
2
3
4
4.
Conclusion
5
6
7
8
9
In summary, 12 active MPO inhibitors were identified. Most of them act simply as electron
donors toward this important oxidoreductase, blocking efficiently and reversibly its
halogenation activity. Compound 28 was found to be a mechanismꢀbased inhibitor of high
inhibitory potency. It is oxidized by the relevant enzymatic redox intermediates Compounds I
and Compound II to the corresponding oneꢀelectron oxidation products that attack the enzyme
and irreversibly modify its structure. Actually, 28 and 4 are the first mechanismꢀbased
inhibitors which inhibit MPO at nanomolar concentration. By pharmacomodulation, further
active compounds were obtained. This approach of combining ligandꢀbased pharmacophore
modeling and structureꢀbased virtual screening seems to be promising to obtain new scaffolds
of MPO inhibitors.
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. Experimental Section:
Synthesis
1Hꢀ and 13CꢀNMR spectra were taken on a Bruker Avance 300 MHz spectrometer
(Wissemburg, France) at 293 K. Chemical shifts (δ) are given in parts per million (ppm)
relative to DMSOꢀd6 or CDCl3, and the coupling constants are expressed in hertz. Infrared
spectroscopic analysis was performed with a Shimadzu (Kyoto, Japan) IRAffinityꢀ1
spectrophotometer equipped with an ATR device, and the peak data are given in cmꢀ1. All
reactions were followed by thinꢀlayer chromatography (TLC) carried out on Fluka (Bornem,
Belgium) PET foils silica gel 60, and compounds were visualized by UV. Column
chromatographies were performed with EchoChrom MP silica 63ꢀ200 from MP Biomedicals
(Santa Ana, CA). Organic solutions were dried over Na2SO4 and concentrated with a Buchi
rotatory evaporator (Flawil, Switzerland). Starting materials aniline, 4ꢀchloroaniline, 3ꢀ
chloroaniline, 4ꢀbromoaniline, 3ꢀbromoaniline and 4ꢀ(trifluoromethyl)aniline were available
23
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 24 of 82
1
2
3
4
5
6
7
8
9
from TCI (Japan). 1ꢀCyanoguanidine was purchased from SigmaꢀAldrich (Bornem, Belgium).
Purity was determined by liquid chromatography (LC) with diodeꢀarray detection (DAD)
(Agilent) on a 150 mm × 4.6 mm Symmetry C18 column at a mobile phase flow rate of 1
mL/min. The mobile phase was a mixture of methanol (350 mL) and a KH2PO4 solution (0.07
M in water, 650 mL) adjusted to pH 3.0 with a 34 wt % H3PO4 solution. The chromatograms
were extracted at maximum absorption wavelengths by using the Max Plot extraction mode.
The purity was ≥ 95% for all compounds.
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
General procedure of synthesis of Nꢀ(4ꢀmethylquinazolineꢀ2ꢀyl)ꢀguanidine derivatives39
To a solution of aniline derivative (10 mmol) in freshly distilled acetone (17 mL) was added
iodine (1.0 g, 4 mmol). The resultant solution was heated to 130 °C for 2 days. The solution
was cooled to room temperature and concentrated in vacuo. The residue was dissolved in
dichloromethane, washed with a 5% sodium bicarbonate and with brine. The solvent removed
in vacuo to give 2.0 g of brown oil. Purification on silica gel via flash column
chromatography (ethyl ether/ petroleum ether 3:97) gave 1,2ꢀdihydroꢀ2,2,4ꢀtrimethylquinoline
derivatives. The resulting compound (1 mmol) was dissolved in dry acetonitrile (0.4 mL) and
added to a solution of hydrochloric acid in diethylether (2 M, 1 mL, 2 mmol). The resultant
solution was stirred at room temperature until precipitation. The brown crystals were filtered,
washed with dry hexane and dried in vacuo. The hydrochloride salt (0.9 mmol) was
suspended in a mixture of water (1.5 mL) and EtOH (1 mL), and dicyandiamide (75 mg, 0.9
mmol) was added. The resultant solution was refluxed for 1 day, and then decanted hot,
removing the brown oil that formed during the reaction. The solution was adjusted to pH 10ꢀ
11 with 2M potassium hydroxide and concentrated in vacuo until a precipitate began to form.
The solution was then filtered and the precipitate washed with cold water. The cured
compound was suspended in chloroform by the aid of sonicator. Then the suspension was
filtered and the precipitate washed with chloroform.
24
ACS Paragon Plus Environment

Page 25 of 82
Journal of Medicinal Chemistry
1
2
Nꢀ(4ꢀMethylꢀquinazolinꢀ2ꢀyl)ꢀguanidine (56)
3
4
5
6
The titled compound was synthesized according to the general procedure to give a yellowish
7
8
9
1
solid (40 mg, 20% yield). H NMR (DMSOꢀd6) δ 8.06 (d, 1H, J = 9 Hz, Hꢀ8), 7.93 (t, J= 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
Hz, 1H, Hꢀ7), 7.90 (t, 1H, J= 3 Hz, Hꢀ6), 7.67 (d, 1H, J = 9 Hz, Hꢀ5), 2.94 (s, 3H, CH3). 13C
NMR (DMSOꢀd6) δ 159.5 (Cꢀguanidine*), 157.3 (Cꢀ2*), 155.3 (Cꢀ4), 148.8 (Cꢀ8a), 135.2 (Cꢀ
7), 128.7 (Cꢀ8), 127.4 (Cꢀ6), 126.8 (Cꢀ5), 125.3 (Cꢀ4a), 21.9 (CH3). IR (FT) 3325, 3143,
2197, 2150, 1622, 1503, 1333, 1248, 1026, 927 cmꢀ1.
Nꢀ(6ꢀFluoroꢀ4ꢀmethylꢀquinazolinꢀ2ꢀyl)ꢀguanidine (57)
The titled compound was synthesized according to the general procedure to give a yellowish
1
solid (37 mg, 17% yield). H NMR (DMSOꢀd6) δ 8.17 (ddd, 1H, J = 9, 9, 3 Hz, Hꢀ7), 8.03
13
(dd, 1H, J = 9, 4 Hz, Hꢀ8), 7.94 (dd, 1H, J = 9, 3 Hz, Hꢀ5), 2.83 (s, 3H, CH3). C NMR
(DMSOꢀd6) δ 167.4 (d, J= 233 Hz, Cꢀ6), 158.9 (Cꢀguanidine*), 157.9 (Cꢀ2*), 153.1 (Cꢀ4),
150.3 (Cꢀ8a), 136.2 (d, J= 18.9 Hz, Cꢀ7), 132.2 (d, J = 20 Hz, Cꢀ5), 130.6 (d, J = 7 Hz, Cꢀ8),
126.5 (Cꢀ4a), 22.7 (CH3). IR (FT) 3325, 3143, 2197, 2150, 1622, 1503, 1333, 1248, 1026,
927 cmꢀ1.
Nꢀ(6ꢀChloroꢀ4ꢀmethylꢀquinazolinꢀ2ꢀyl)ꢀguanidine (58)
The titled compound was synthesized according to the general procedure to give a yellowish
1
solid (84 mg, 36% yield). H NMR (DMSOꢀd6) δ 8.22 (d, 1H, J = 9 Hz, Hꢀ8), 7.86 (m, 2H,
Hꢀ5, 7), 2.95 (s, 3H, CH3). 13C NMR (DMSOꢀd6) δ 158.6 (Cꢀguanidine*), 157.1 (Cꢀ2*), 155.0
(Cꢀ4), 148.6 (Cꢀ8a), 136.4 (Cꢀ7), 132.4 (Cꢀ6), 130.6 (Cꢀ8*), 129.1 (Cꢀ5*), 123.9 (Cꢀ4a), 21.4
(CH3). IR (FT) 3325, 3143, 2197, 2150, 1622, 1503, 1333, 1248, 1026, 927 cmꢀ1.
Nꢀ(6ꢀBromoꢀ4ꢀmethylꢀquinazolinꢀ2ꢀyl)ꢀguanidine (59)
25
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 26 of 82
1
2
3
4
5
6
7
8
9
The titled compound was synthesized according to the general procedure to give a yellowish
solid (59 mg, 21% yield). 1H NMR (DMSOꢀd6) δ 8.40 (d, J = 3 Hz, 1H, Hꢀ5), 7.99 (dd, J = 9,
3 Hz, 1H, Hꢀ7), 7.85 (d, J= 9 Hz, 1H, Hꢀ8), 2.93 (s, 3H); IR (FT) 3325, 3143, 2197, 2150,
1622, 1503, 1333, 1248, 1026, 927 cmꢀ1.
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
Nꢀ(7ꢀFluoroꢀ4ꢀmethylꢀquinazolinꢀ2ꢀyl)ꢀguanidine (60)
The titled compound was synthesized according to the general procedure to give a yellowish
solid (37 mg, 17% yield). 1H NMR (CDCl3) δ 7.31 (dd, 1H, J = 9, 6 Hz, Hꢀ5), 7.23 (dd, 1H, J
= 9, 3 Hz, Hꢀ8), 6.97 (ddd, 1H, J = 9, 9, 3 Hz, Hꢀ6), 5.19 (s, 4H, 2NH2), 2.05 (s, 3H, CH3).
13C NMR (CDCl3) δ 160.6 (d, J= 143 Hz, Cꢀ7), 157.7 (Cꢀguanidine*), 156.6 (Cꢀ2*), 133.3
(Cꢀ4), 127.7 (d, J= 7 Hz, Cꢀ8a), 114.2 (d, J = 6 Hz, Cꢀ4a), 112.3 (d, J = 10 Hz, Cꢀ5), 110.9 (d,
J = 26 Hz, Cꢀ6), 103.9 (d, J = 23 Hz, Cꢀ8), 24.3 (CH3). IR (FT) 3325, 3143, 2197, 2150,
1622, 1503, 1333, 1248, 1026, 927 cmꢀ1.
Nꢀ(7ꢀChloroꢀ4ꢀmethylꢀquinazolinꢀ2ꢀyl)ꢀguanidine (61)
The titled compound was synthesized according to the general procedure to give a yellowish
solid (56 mg, 24% yield). 1H NMR (DMSOꢀd6) δ 8.27(d, J = 9Hz, 1H, Hꢀ5), 8.19 (s, 1H, Hꢀ
8), 7.78 (d, J = 9 Hz, 1H, Hꢀ6), 2.99 (s, 3H); 13C NMR (DMSOꢀd6) δ 159.9(Cꢀ2*), 157.9 (Cꢀ
guanidine*), 151.7 (Cꢀ4), 149.7 (Cꢀ8a), 136.5 (Cꢀ7), 132.6 (Cꢀ8), 131.3 (Cꢀ6), 127.9 (Cꢀ5),
125.4 (Cꢀ4a), 21.9 (CH3). IR (FT) 3325, 3143, 2197, 2150, 1622, 1503, 1333, 1248, 1026,
927 cmꢀ1.
Nꢀ(7ꢀBromoꢀ4ꢀmethylꢀquinazolinꢀ2ꢀyl)ꢀguanidine (62)
The titled compound was synthesized according to the general procedure to give a yellowish
solid (42 mg, 15% yield). 1H NMR (DMSOꢀd6) δ 8.09 (d, J = 9 Hz, 1H, Hꢀ6), 7.89 (s,1H, Hꢀ
8), 7.68 (d, J = 9 Hz, 1H, Hꢀ5), 2.93 (s, 3H); 13C NMR (DMSOꢀd6) δ 159.1(Cꢀ2*), 157.5 (Cꢀ
26
ACS Paragon Plus Environment

Page 27 of 82
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
guanidine*), 153.5 (Cꢀ4), 150.7 (Cꢀ8a), 134.2 (Cꢀ8), 130.8 (Cꢀ6), 128.5 (Cꢀ7), 127.2 (Cꢀ5),
126.0 (Cꢀ4a), 22.4 (CH3). IR (FT) 3325, 3143, 2197, 2150, 1622, 1503, 1333, 1248, 1026,
927 cmꢀ1.
9
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
Data Sets
For the virtual screening campaigns, the ZINC database v.8 was downloaded from the ZINC
Web site (http://zinc.docking.org/). This database is composed of 727842 small synthetic
chemicals and consists of compounds. These compounds were transformed into a
LigandScout ® database using the idbgenꢀtool of LigandScout®. The selected compounds
(Zinc codes of these compounds are available in Table S1) were purchased from Aurora Fine
Chemicals Ltd. The remaining selected compounds were available from Fagron (Belgium).
Pharmacophore Modeling
DruLiTo
software
(ref:
http://www.niper.gov.in/pi_dev_tools/DruLiToWeb/DruLiTo_index.html) was used to filter
the Zinc database molecules against the Lipinski's rules that is the selected molecules have no
more than 5 hydrogen bond donors and 10 hydrogen bond acceptors, a molecular mass less
than 500 g molꢀ1 and a Moriguchi's logP (MLogP) < 5.
The
pharmacophore
models
were
constructed
using
LigandScout®3.12
(www.inteligand.com). In a shared feature pharmacophore model generation, LigandScout
generates pharmacophore models from the chemical functionalities of the training compounds
and aligns the molecules according to their pharmacophores. In our study, the search for a
pharmacophore using the structural data of all seven highly potent MPO inhibitory
compounds (Figure 2) was not successful. The models obtained using each of the seven
inhibitors selected only compounds analogous to this inhibitor (so no new scaffold was
found). The search made combining the seven inhibitors per different groups of three
27
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 28 of 82
1
2
3
4
5
6
7
8
molecules selected very large ensembles of hits because of the low numbers of the shared
functional groups. Only the models obtained using the seven inhibitors in pairs produced
pharmacophores with the highest number of shared chemical functionalities.
9
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 quality of the pharmacophore models was quantitatively evaluated by calculating the
selectivity (eq 1) and specificity (eq 2) for each model using known active and inactive
compounds.
ꢀꢁꢂꢃꢄ ꢅꢆꢇꢈꢉꢊ
Eq1
Eq2
selectivity = ꢅꢋꢋ ꢇꢌꢊ ꢇꢊꢍꢇꢊꢄ ꢆꢁꢎꢏꢁꢂꢃꢄꢍ
ꢀꢁꢂꢃꢄ ꢈꢃꢅꢆꢇꢈꢉꢊ
specificity = ꢅꢋꢋ ꢇꢌꢊ ꢇꢊꢍꢇꢊꢄ ꢆꢁꢎꢏꢁꢂꢃꢄꢍ
Docking experiments
Docking was performed with FlexX program included in LeadIT (version 2.1.8) or with the
Glide program (version 7.4).
The Xꢀray structure of human myeloperoxidase complexed to thiocyanate (PDB code 1DNU)
was used as the target structure for the docking studies with FlexX16. The protein dimer was
loaded into LeadIT ® 2.1.8 and chains B and D were selected to form the receptor
components. The binding site was defined by choosing thiocyanate as a reference. Amino
acids within 11 Å of the ligand were selected as binding site (the residues 81 to 120 and 330
to 337). The Xꢀray water and other ligand molecules were removed from the active site. The
ligand input files were prepared according to the following procedure. The initial 3D
structures of the ligands were generated by use of the LigandScout ® procedure48. The
prediction of different protonation states of all ligands was achieved with the protonation state
option in LeadIT ® software. For the stereochemistry, the options R/S and pseudo R/S were
chosen in stereo mode to cover all the possible stereo models of the ligands. Docking was
performed with FlexX program included in LeadIT (version 2.1.8), no additive filters were
28
ACS Paragon Plus Environment

Page 29 of 82
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
applied. Hydrogen bond donors and acceptors, phenyl centers, and hydrophobic features are
defined similarly in FlexX as pharmacophore features. The FlexX scoring function was used.
The ligand binding was driven by hybrid approach (enthalpy and entropy). The remaining
parameters were set to their default values. The experiment was validated using the crystal 1
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
ligand49ꢀ50
.
The Xꢀray structure of human MPO complexed to compound 1 (PDB code 4C1M) was used
as the receptor targets in docking studies51. The Xꢀray water and other ligand molecules were
removed from the active site. The initial 3D structures of the ligands were generated with the
Ligprep module (Schrodinger, LLC, New York, NY, 2017) and the ligand partial charges
were ascribed by use of the OPLS force field. The Epik program was used to predict different
protonation states of all ligands52. The Glide XP docking protocol and scoring function were
used. The remaining parameters were set to their default values.
In Silico Toxicological Evaluation and DrugꢀLike Profile
The in silico toxicity and drugꢀlike profile of the selected molecules were calculated using the
Program OSIRIS Property Explorer. Data were generated onꢀline in the Osiris Program,
accessed by the link (http://www.organicꢀchemistry.org/prog/peo/) and represented by
toxicity risks (mutagenic, irritant, tumorigenic and reproductive effects), drug likeness and
drugꢀscore. Drug likeness was calculated based on equation summing up score values of the
fragments present in the molecule under investigation. The fragments were identified from a
list of 5,300 distinct substructure fragments with associated drug likeness scores. Drugꢀscore
was calculated combining the drug likeness, cLogP, logS, molecular weight and toxicity risks
data. The compound was selected when all the following criteria were achieved: no
tumorigenicity risk, no reproductive effect risk, no mutagenicity risk, no irritating effect risk,
drugꢀlikeness ≤ꢀ0.05, and drugꢀscore ≥0.5.
29
ACS Paragon Plus Environment