Discovery of D-amino acid oxidase inhibitors based on virtual screening against the lid-open enzyme conformation

Article abstract of DOI:10.1016/j.bmcl.2018.04.048

D-Amino acid oxidase (DAAO) inhibitors are typically small polar compounds with often suboptimal pharmacokinetic properties. Features of the native binding site limit the operational freedom of further medicinal chemistry efforts. We therefore initiated a structure based virtual screening campaign based on the X-ray structures of DAAO complexes where larger ligands shifted the loop (lid opening) covering the native binding site. The virtual screening of our in-house collection followed by the in vitro test of the best ranked compounds led to the identification of a new scaffold with micromolar IC₅₀. Subsequent SAR explorations enabled us to identify submicromolar inhibitors. Docking studies supported by in vitro activity measurements suggest that compounds bind to the active site with a salt-bridge characteristic to DAAO inhibitor binding. In addition, displacement of and interaction with the loop covering the active site contributes significantly to the activity of the most potent compounds.

Full text of DOI:10.1016/j.bmcl.2018.04.048

Bioorganic & Medicinal Chemistry Letters xxx (2018) xxx–xxx
Contents lists available at ScienceDirect
Bioorganic & Medicinal Chemistry Letters
journal homepage: www.elsevier.com/locate/bmcl
Discovery of D-amino acid oxidase inhibitors based on virtual screening
against the lid-open enzyme conformation
a
b
c
b
a
b
a,
⇑
ˇ
}
Bence Szilágyi , Ziga Skok , Anita Rácz , Rok Frlan , György G. Ferenczy , Janez Ilaš , György M. Keseru
a
Medicinal Chemistry Research Group, Research Centre for Natural Sciences, Hungarian Academy of Sciences, Magyar tudósok krt. 2. H-1117 Budapest, Hungary
Faculty of Pharmacy, University of Ljubljana, Ašker cˇ eva 7, 1000 Ljubljana, Slovenia
Plasma Chemistry Research Group, Research Centre for Natural Sciences, Hungarian Academy of Sciences, Magyar tudósok krt. 2, H-1117 Budapest, Hungary
b
c
a r t i c l e i n f o
a b s t r a c t
Article history:
D-Amino acid oxidase (DAAO) inhibitors are typically small polar compounds with often suboptimal
Received 30 March 2018
Revised 17 April 2018
Accepted 18 April 2018
Available online xxxx
pharmacokinetic properties. Features of the native binding site limit the operational freedom of further
medicinal chemistry efforts. We therefore initiated a structure based virtual screening campaign based
on the X-ray structures of DAAO complexes where larger ligands shifted the loop (lid opening) covering
the native binding site. The virtual screening of our in-house collection followed by the in vitro test of the
best ranked compounds led to the identification of a new scaffold with micromolar IC50. Subsequent SAR
explorations enabled us to identify submicromolar inhibitors. Docking studies supported by in vitro
activity measurements suggest that compounds bind to the active site with a salt-bridge characteristic
to DAAO inhibitor binding. In addition, displacement of and interaction with the loop covering the active
site contributes significantly to the activity of the most potent compounds.
Keywords:
D-Amino acid oxidase
Open-lid conformation
Virtual screening
Structure-activity relationship
Ó 2018 Elsevier Ltd. All rights reserved.
Introduction
tioned parallel with the isoalloxazine ring of the FAD cofactor. Fur-
ther interactions contributing to the binding of some of the ligands
D
-Amino acid oxidase (DAAO) metabolizes
at the glycine site on the N-methyl -aspartate (NMDA) receptor.
Several pieces of evidence suggest that increased DAAO activity
D
-serine, a co-agonist
are hydrogen bonds to the backbone of Gly313 and to the sidechain
of Tyr228 and stacking interactions with Tyr224. It is worth noting
that the binding pocket is highly polar at the Arg283 side while it is
predominantly hydrophobic at the opposite side (Fig. 2a).
1
D
2
D
-serine level3 are associated with schizophrenia.
,4
and decreased
Therefore, it has been suggested that the inhibition of DAAO may
result in an increase of brain -serine level and may have beneficial
effect on the positive, negative and cognitive symptoms of
schizophrenia.
Several DAAO inhibitors have been reported in the literature.
Representative examples are shown in Fig. 1. Most of the known
inhibitors are small and polar (Fig. 1a) similarly to the endogenous
ligands, D-amino acids. It is notable that DAAO works optimally
around pH 8 and the bound amino acids are deprotonated and
The small inhibitors bound to DAAO (Fig. 1a) are completely
buried in the protein binding site with no access to the bulk water.
This is thought to be the consequence of the switch of loop 216–
228 between an open conformation that allows the ligand to enter
and to leave the binding site, and a closed conformation adopted in
D
5
9
the complex.
X-ray structures with a lid-open conformation were reported10
with extended compounds 6 and 7 (Fig. 1b). The polar head of
these compounds (carboxylate in 6 and the hydroxy-pyridazinone
in 7) binds to Arg283, just as it was seen for small inhibitors. How-
ever, the annelated rings form additional van der Waals and polar
interactions and the connected aromatic ring opens the lid forming
non-polar interactions with the backbone and the hydrophobic
residues of the loop (Fig. 2b).
6
7
bear a negative charge rather than being zwitter ionic. In line with
this finding, DAAO inhibitors are deprotonable and a significant
amount of negatively charged microspecies is present around pH
8
=
8.
The important binding motifs of DAAO inhibitors include a neg-
atively charged moiety where two heteroatoms of the ligand form
a salt bridge with Arg283, and a planar, electron rich moiety posi-
As it has been demonstrated above the size of DAAO inhibitors
spans a large range. They include fragment sized polar compounds
most resemble to the endogenous ligands and larger lead-like com-
pounds whose binding is possible owing to the plasticity of loop
216–228. This loop serves as a lid that is closed when the bound
ligand is small and adopts an open conformation when the binding
⇑
Corresponding author.
E-mail address: keseru.gyorgy@ttk.mta.hu (G.M. Keser u} ).
https://doi.org/10.1016/j.bmcl.2018.04.048
960-894X/Ó 2018 Elsevier Ltd. All rights reserved.
0
Please cite this article in press as: Szilágyi B., et al. Bioorg. Med. Chem. Lett. (2018), https://doi.org/10.1016/j.bmcl.2018.04.048

2
B. Szilágyi et al. / Bioorganic & Medicinal Chemistry Letters xxx (2018) xxx–xxx
as the third target for docking. Protein and ligand preparations
1
4
were performed with Schrödinger’s tools with standard settings
1
4
and Glide was used for docking and scoring.
Hits were selected using Glide’s docking score. Docked poses of
top scored compounds were visually inspected and the best 16
compounds were subject to in vitro testing with
D
-kynurenine
assay. This procedure led to the identification of compound 8
Table 1) as a micromolar DAAO inhibitor. The docked pose of 8
1
5
(
and its interactions are shown in Fig. 3. Docking suggests that
the deprotonated carboxylic acid interacts with Arg283 forming a
salt-bridge as it is observed in the X-ray structure of several com-
plexes (see above). There is not enough space for the pyrrole-car-
boxamide moiety to be in the plane of the carboxylate group,
parallel with the isoalloxazine ring of FAD, and it leans towards
loop 216–228. An interesting feature of the docked compounds is
that their carboxamide moiety between the pyrrole rings adopts
cis conformation.
We performed an SAR study based on the newly identified
DAAO inhibitor, compound 8, with two objectives; first, to under-
stand the structural requirements of lid opening, and second, to
explore the interaction network around the inhibitor scaffold.
Compounds 20–28 were synthesized to explore how the inter-
action with the lid contributes to the activity. These compounds
either have reduced size presumably not interacting with the lid,
or they have varied substitution pattern on the terminal pyrrole
ring. As shown in Scheme 1, compounds 20–28 were obtained from
9. Reduction of 9 followed by acylation with the corresponding
acids or acid chlorides resulted 11–19. These compounds were
hydrolyzed in a mixture of dioxane and 1 M sodium hydroxide
solution resulting 20–27.
Fig. 1. Representative set of small (a) and extended (b) DAAO inhibitors.
site accommodates large ligands. The ability of large compounds to
bind and inhibit DAAO offers an opportunity to drug discovery
efforts to identify inhibitors with balanced properties. The polar
head group that provides strong binding primarily by interacting
with Arg283 and with some other polar groups (cf. Fig. 2) can be
grown to include various chemical groups allowing the fine-tuning
of the overall molecular properties. This recognition motivated our
work to search for lead-like DAAO inhibitors that bind to the lid-
open conformation of the enzyme. Our work started with struc-
ture-based virtual screening of our in-house compound library
against the lid-open conformation of DAAO.
1
6
In order to explore the impact of the polar interaction network,
specifically the role of the NH groups, methyl groups were added to
the two amide N-atoms and also to the middle pyrrole ring N-
atom. N-methylation affects the H-bonding ability of the com-
pounds, the space they fill and also their preferred conformation.
For the preparation (Scheme 2) of these N-methylated compounds
1
1–13
To find novel structural classes in-house compound library
of over 2000 unique compounds at the Faculty of Pharmacy,
University of Ljubljana was used for virtual screening. Compounds
were docked into three DAAO structures. Two were taken from
1
7
(36, 39, 40), 29 was converted to 31 through hydrogenation,
followed by acylation with 2,2,2-trichloro-1-(4,5-dibromo-1H-pyr-
rol-2-yl)ethan-1-one. To get 36, compound 31 was hydrolyzed to
carboxylic acid 32 than 34 was formed with sarcosine methyl ester
hydrochloride and HATU. Compound 36 was obtained from 34
10
10
complexes of open-lid proteins (PDB: 4QFC and 4QFD ). These
two structures have different sequences with missing residues
and they were used to create a complete structure that were used
Fig. 2. Key protein-ligand interactions identified for DAAO inhibitors. Small (a) and extended (b) inhibitors are exemplified by compounds 4 (PDB 3G3E) and 6 (PDB 4QFC),
respectively. Ligand carbon atoms are in black, protein amino-acids are shown in colored drops. Interactions are indicated by lines: magenta H-bond, green aromatic-aromatic
interaction.
Please cite this article in press as: Szilágyi B., et al. Bioorg. Med. Chem. Lett. (2018), https://doi.org/10.1016/j.bmcl.2018.04.048

B. Szilágyi et al. / Bioorganic & Medicinal Chemistry Letters xxx (2018) xxx–xxx
3
2
Scheme 1. Synthetic route of compounds 8 and 20–28. Reagents and conditions: (i) Pd/C (10 w/w%), H , DMF, r.t., 24 h; (ii) TEA, acid chloride, DCM:DMF (6:1), 0 ? r.t.,
overnight or carboxylic acid, NMM, EDC, HOBT, DCM:DMF (1:1), 0 ? 45 °C, overnight (iii) dioxane: 1 M NaOH solution (1:1), r.t., 2 h or THF: 1 M LiOH solution (2:1), r.t.,
overnight.
Scheme 2. Synthetic route of compounds 36, 39, 40 and 46. Reagents and conditions: (i) Pd/C (10 w/w%), H
pyrrol-2-yl)ethan-1-one, Na CO , DMF, 100 °C, 24 h or 4,5-dibromo-1H-pyrrole-2-carboxylic acid, NMM, EDC, HOBT, DCM:DMF (1:1), 0 ? 45 °C, overnight; (iii) dioxane: 1 M
NaOH solution (1:1), r.t., 2 h or THF: 1 M LiOH solution (2:1), r.t., overnight (iv) MeI, Na CO , DMF, r.t., 24 h; (v) sarcosine methyl ester hydrochloride, HATU, DIPEA, DMF, r.t.,
overnight; (vi) glycine methyl ester hydrochloride, HATU, DIPEA, DMF, r.t., overnight; (vii) tetrabutylammonium iodide, NaH, dry DMF, 0 °C, 10 min, than MeI, r.t., overnight;
viii) glycine methyl ester hydrochloride, NMM, EDC, HOBT, DCM:DMF (1:1), 0 ? 45 °C, overnight.
2
, methanol, r.t., 2 h; (ii) 2,2,2-trichloro-1-(4,5-dibromo-1H-
2
3
2
3
(
via hydrolysis. In the cases of compounds 39–40, 31 was methy-
lated with methyl iodine than hydrolyzed to 35. Compound 35
was coupled with the corresponding amino acid esters, followed
by hydrolysis to get 39 and 40. The preparation of compound 46
started with the methylation of 29 and followed by an analogous
procedure described above.
Data in Table 1 show that active compounds all include a termi-
nal pyrrole ring with large substituents. Compounds with unsub-
stituted pyridine (23, 24) or without the terminal substituted
pyrrole (20, 25, 28) exhibit low inhibitory activity. Investigating
their docked poses (cf. Fig. 3), we observe the presence of a salt
bridge between the carboxylate group and Arg283, but the com-
pounds fail to form other interactions characteristic to high affinity
DAAO inhibitors. No stacking with the isoalloxazine ring of the FAD
Compounds designed, synthesized and tested as DAAO inhibi-
tors are shown in Table 1.
Please cite this article in press as: Szilágyi B., et al. Bioorg. Med. Chem. Lett. (2018), https://doi.org/10.1016/j.bmcl.2018.04.048

4
B. Szilágyi et al. / Bioorganic & Medicinal Chemistry Letters xxx (2018) xxx–xxx
Table 1
DAAO inhibitory activity of designed compounds.
Compound
Structure
% inhibition at 20
100 (0)%
l
Ma
IC50 [l
M]a
8
5.2 (0.6)
2
2
0
1
39 (0)%
–
100 (0)%
8.8 (2.3)
2
2
100 (0)%
6.7 (1.8)
2
2
3
4
45 (2)%
18(2)%
–
–
2
2
5
6
8 (1)%
–
100 (0)%
0.39 (0.02)
2
2
7
8
33 (8)%
35 (4)%
–
–
3
3
4
4
6
9
0
6
100 (0)%
0 (8)%
0.49 (0.06)
–
–
–
0 (3)%
32 (1)%
a
All measurements were performed in duplicates; standard deviation is shown in parenthesis.
and limited favorable interactions by the central part of the ligands
are detected. Importantly, these compounds do not reach and open
the lid formed by loop 216–228. A comparison of their structure
and activity to those of their analogues 8 and 26 indicates that
the lack of interaction with loop 216–228 is primarily responsible
for their low activity.
Best activity in the series is obtained for compounds 26 and 36,
both containing Br-substituents on the terminal pyrrole ring. It is
Please cite this article in press as: Szilágyi B., et al. Bioorg. Med. Chem. Lett. (2018), https://doi.org/10.1016/j.bmcl.2018.04.048

B. Szilágyi et al. / Bioorganic & Medicinal Chemistry Letters xxx (2018) xxx–xxx
5
Fig. 3. Binding pose (a) and interactions (b) of 8 as proposed by docking. Panel (a): Ligand carbon atom are in grey, FAD carbon atoms are in light blue and protein carbon
atoms are in green. Interactions are indicated by dashed lines: yellow H-bond, cyan halogen-bond. Tyr55 is not shown for the sake of clarity. Panel (b): Ligand carbon atoms
are in black, protein amino-acids are shown in colored drops. Interactions are indicated by lines: magenta H-bond, green aromatic-aromatic, orange halogen-bond.
notable that compound 22 having a single Br-substituent at the 4th
position and unsubstituted at the 5th has an order of magnitude
lower activity than the 4,5 disubstituted compound 26. The impor-
tance of the substituent at the 5th position is underlined by the
methyl group is close to loop 48–63. The effect of this steric clash
observed for 39 and 40 was not possible to assess a priori since the
conformation of this loop either varies (see e.g. PDB codes 3ZNQ
and 4QFC) or is partially missing (3CUK and 4QFD) in available
crystal structures that indicate structural flexibility. The complete
loss of activity observed with the methylation of the amide
between the pyrrole rings (compare compounds 26 and 36 with
39 and 40) suggests that this loop is unable to adopt a conforma-
tion that allows the binding pocket to accommodate the methy-
lated amide group.
In summary, DAAO inhibitors were identified starting with the
structure-based virtual screening of our in-house library of ꢀ2000
compounds using protein structures with lid-open conformation.
Testing of designed analogues allowed us to explore the SAR of
the series and to identify submicromolar inhibitors. They have a
polar head group characteristic to DAAO inhibitors; in addition,
they have a linear shape and are larger than most reported inhibi-
tors. According to SAR and docking studies their tail interacts with
the lid and this interaction is essential for the activity of the
compounds.
activity measured for compounds 8 (IC50 = 5.2
lM) and 21 (IC50 =
8
.7 lM) containing 5-Cl and 5-methyl substituents, respectively.
They both have hydrophobic groups at the 5th position just like
compounds 26 and 36 with Br at the 5th position. The 10-fold
increase in activity of the two latter compounds can be attributed
to the presence of Br that is larger than either Cl or Me on one
hand, and is able to form a halogen bond with Ser223 backbone
carbonyl, on the other hand.
The methylation of the central pyrrole ring leads to a significant
loss of activity (compare compounds 26 and 46). This methyl group
prevents the H-bond formation with Gly313 backbone and, accord-
ing to the docked poses, it causes steric clash with the protein
backbone. Methylation of the amide nitrogen close to the carboxyl
end of the molecules does not affect activity according to the com-
parison of compounds 26 and 36. Docking studies suggest that this
amide can adopt both cis and trans conformation without signifi-
cantly affecting the position of the terminal substituted pyrrole
ring in either 26 or 36. The trans amide conformation is proposed
to be preferred for amide (26) and the cis conformation is proposed
to be preferred for methylamide (36).
Acknowledgements
This research was supported by the bilateral Hungarian-Slove-
nian science and technology cooperation under Grant Number
TÉT 16-1-2016-0069 and BI-HU/17-18-009.
The amide group between the pyrrole rings is predicted by
docking to adopt cis conformation. This results in a bend molecule
that is accommodated in the binding pocket extended by the
movement of loop 216–228 (lid opening). The presence of a sec-
ondary cis amide is notable, since amides have high preference
A. Supplementary data
18
for the trans conformation both in proteins
and in small
1
9
Supplementary data associated with this article can be found, in
the online version, at https://doi.org/10.1016/j.bmcl.2018.04.048.
molecules. Nevertheless, there are examples where ligands bind
to proteins with cis amide conformation (PDB: 1HTG, 1AJ6, 1YBH)
showing that intermolecular interactions are able to compensate
for the energy increase due to amide trans-cis transformation. Con-
verting secondary amides to tertiary amides increases the propen-
sity for cis conformation. Therefore, the methylation of the amide
that is assumed to adopt cis-conformation when bound may be
able to increase activity by facilitating the adoption of cis confor-
mation. Another effect of methylation, however, is the increase of
bulkiness. Docking of compounds 39 and 40 shows that this
References
1
2
.
.
Wolosker H. Mol Neurobiol. 2007;36:152.
Burnet PWJ, Eastwood SL, Bristow GC, et al. Mol Psychiatry. 2008;13:658.
3. Hashimoto K, Fukushima T, Shimizu E, et al. Arch. Gen. Psychiatry.
003;60:572–576.
2
4
5
.
.
Bendikov I, Nadri C, Amar S, et al. Schizophrenia Res.. 2007;90:41.
Smith SM, Uslaner JM, Hutson PH. Open Med Chem J. 2010;4:3.
6. Harris CM, Pollegioni L, Ghisla S. Eur J Biochem. 2001;268:5504.
Please cite this article in press as: Szilágyi B., et al. Bioorg. Med. Chem. Lett. (2018), https://doi.org/10.1016/j.bmcl.2018.04.048

6
B. Szilágyi et al. / Bioorganic & Medicinal Chemistry Letters xxx (2018) xxx–xxx
7
.
.
.
Kurtz KA, Rishavy MA, Cleland WW, Fitzpatrick PF.
000;122:12896.
Orgován Z, Ferenczy GG, Steinbrecher T, Szilágyi B, Bajusz D, Keser u} GM. J
Comput Aided Mol Des. 2018;32:331.
J
Am Chem Soc.
13. Brus B, Košak U, Turk S, et al. J Med Chem. 2014;57:8167.
14. Schrödinger Suite. 2017À1; Schrödinger. New York, NY: LLC; 2017.
15. Song Z, Ogaya T, Ishii K, Ichiba H, Iizuka H, Fukushima T. J Health Sci.
2010;56:341.
2
8
9
Todone F, Vanoni MA, Mozzarelli A, et al. Biochemistry. 1997;36:5853.
16. Available from AKos Consulting
& Solutions. (http://www.akosgmbh.de/
10. Terry-Lorenzo RT, Chun LE, Brown SP, et al. Biosci Rep. 2014;34:487.
1
1. Cotman AE, Trampu zˇ M, Brvar M, et al. Arch Pharm. 2017;350:8.
1
2. Ilaš J, Pe cˇ ar S, Hockemeyer J, Euler H, Kirfel A, Müller CE. J Med Chem.
AKosSamples/), ID: AKOS006034828.
17. Lautru S, Song L, Demange L, et al. Angew Chem Int Ed. 2012;51:7454.
18. Weiss MS, Jabs A, Hilgenfeld R. Nat Struct Biol. 1998;5:676.
2
005;48:2108.
19. Brameld KA, Kuhn B, Reuter DC, Stahl M. J Chem Inf Model. 2008;48:1.
Please cite this article in press as: Szilágyi B., et al. Bioorg. Med. Chem. Lett. (2018), https://doi.org/10.1016/j.bmcl.2018.04.048