Discovery of Novel Phosphodiesterase-2A Inhibitors by Structure-Based Virtual Screening, Structural Optimization, and Bioassay

Article abstract of DOI:10.1021/acs.jcim.6b00551

Phosphodiesterase-2A (PDE2A) is a potential therapeutic target for treatment of Alzheimer’s disease and pulmonary hypertension. However, most of the current PDE2A inhibitors have moderate selectivity over other PDEs. In the present study, we described the discovery of novel PDE2A inhibitors by structure-based virtual screening combining pharmacophore model screening, molecular docking, molecular dynamics simulations, and bioassay validation. Nine hits out of 30 molecules from the SPECS database (a hit rate of 30%) inhibited PDE2A with affinity less than 50 μM. Optimization of compound AQ-390/10779040 (IC₅₀ = 4.6 μM) from the virtual screening, which holds a novel scaffold of benzo[cd]indol-2(1H)-one among PDE inhibitors, leads to discovery of a new compound LHB-8 with a significant improvement of inhibition (IC₅₀ = 570 nM). The modeling studies demonstrated that LHB-8 formed an extra hydrogen bond with Asp808 and a hydrophobic interaction with Thr768, in addition to the common interactions with Gln859 and Phe862 of PDE2A. The novel scaffolds discovered in the present study can be used for rational design of PDE2A inhibitors with high affinity.

Full text of DOI:10.1021/acs.jcim.6b00551

Subscriber access provided by UNIV OF CALIFORNIA SAN DIEGO LIBRARIES
Article
Discovery of Novel Phosphodiesterase-2A Inhibitors by Structure-
based Virtual Screening, Structural Optimization, and Bioassay
Chen Zhang, Ling-Jun Feng, Yiyou Huang, Deyan Wu, Zhe Li, Qian Zhou, Yinuo Wu, and Hai-Bin Luo
J. Chem. Inf. Model., Just Accepted Manuscript • DOI: 10.1021/acs.jcim.6b00551 • Publication Date (Web): 05 Jan 2017
Downloaded from http://pubs.acs.org on January 5, 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 Chemical Information and Modeling 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 35
Journal of Chemical Information and Modeling
1
2
3
4
Discovery of Novel Phosphodiesterase-2A Inhibitors by Structure-based
5
6
7
Virtual Screening, Structural Optimization, and Bioassay
8
9
Chen Zhanga, LingꢀJun Fenga, Yiyou Huanga, Deyan Wua, Zhe Lia, Qian Zhoua, Yinuo Wua, *, and
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
HaiꢀBin Luoa, b,
*
a School of Pharmaceutical Sciences, Sun YatꢀSen University, Guangzhou, 510006, China
b Collaborative Innovation Center of High Performance Computing, National University of Defense
Technology, Changsha 410073, China
Abstract:
Phosphodiesteraseꢀ2A (PDE2A) is a potential therapeutic target for treatment of Alzheimer’s
disease and pulmonary hypertension. However, most of the current PDE2A inhibitors have the
moderate selectivity over other PDEs. In the present study, we described the discovery of novel
PDE2A inhibitors by structureꢀbased virtual screening combining pharmacophore model screening,
molecular docking, molecular dynamics simulations, and bioassay validation. Nine hits out of 30
molecules from the SPECS database (a hit rate of 30%) inhibited PDE2A with affinity less than 50
ꢀM. Optimization of compound AQ-390/10779040 (IC50 = 4.6 ꢀM) from the virtual screening,
which holds a novel scaffold of benzo[cd]indolꢀ2(1H)ꢀone among PDE inhibitors, leads to discovery
of a new compound LHB-8 with a significant improvement of inhibition (IC50 = 570 nM). The
modeling studies demonstrated that LHB-8 formed an extra hydrogen bond with Asp808 and a
hydrophobic interaction with Thr768, in addition to the common interactions with Gln859 and
Phe862 of PDE2A. The novel scaffolds discovered in the present study can be used for rational
design of PDE2A inhibitors with high affinity.
1
ACS Paragon Plus Environment

Journal of Chemical Information and Modeling
Page 2 of 35
1. Introduction
1
2
3
4
5
6
7
8
9
Phosphodiesterases (PDEs) are a super family of enzymes hydrolyzing the intracellular second
messengers cyclic guanosine and adenosine monophosphate (cGMP and cAMP).[1ꢀ2] It has been
shown that decreased cGMP and cAMP concentrations are the main courses of certain diseases.
PDEs, for their modulation of the level of cGMP and cAMP, are thus potential therapeutic targets for
treatment of many diseases.[3ꢀ5] Several PDEs inhibitors have been approved as clinical drugs. For
example, sildenafil (a PDE5 inhibitor) is widely used for treatment of male erectile dysfunction and
pulmonary hypertension.[6ꢀ9]
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
Human PDEs are encoded by 21 genes and divided into 11 families which have different
substrate specificity. Nevertheless, they share structural similarity in catalytic domains and possess a
highly conserved hydrophobic clamp and bimetal ion binding center at the active sites.[1,10ꢀ11] PDE2
is a dualꢀspecific enzyme that hydrolyzes both cGMP and cAMP,[12] and mostly distributes in the
central nervous system (CNS), with relatively low expression in peripheral tissues. Several studies
demonstrated that PDE2 modulates neuronal signalings involved in emotion, perception,
concentration, learning and memory, thus made it a potential therapeutic target for treatment of CNS
diseases such as Alzheimer’s disease, mild cognitive impairment, and depression.[13ꢀ14]
Although several efficient PDE2 inhibitors have been developed in recent years, there is still no
PDE2 inhibitor on the market now (Fig. S1. ESI).[15ꢀ21] The first generation of PDE2 inhibitors is
EHNA, which lacks the selectivity over other PDEs.[15ꢀ16] BAY-60-7550, a specific PDE2 inhibitor
with IC50 of 2 nM, showed good curative effect on the rodent behavioral models in learning and
memory, and improved the acquisition and consolidation phases of novel object memory in
ageꢀimpaired rats.[22] However, the pharmacokinetic properties of BAY-60-7550 are poor, which
limited its clinical application.[17] Thus, discovery of novel PDE2 inhibitors is highly in demand.
Structureꢀbased drug design has been considered as a rational strategy and widely applied in the
drug discovery in recent years. The Xꢀray crystal structure of PDE2 in complex with BAY-60-7550
was reported in 2013. It supported the conserved glutamine switch mechanism for substratesꢀPDEs
binding and revealed that a hydrophobic pocket may play an important role in the improvement of
2
ACS Paragon Plus Environment

Page 3 of 35
Journal of Chemical Information and Modeling
binding affinity and selectivity for PDE2.[20ꢀ22] This structural evidence may contribute to the
computerꢀaided inhibitor design for PDE2. Herein, our work on discovery of novel PDE2 inhibitors
started from a structureꢀbased virtual screening to select potential leads from the commercial
database SPECS. Nine hits out of 30 molecules (a hit rate of 30%) were identified to be PDE2
inhibitors by subsequent bioassay, 5 of which possess novel scaffolds. Further structural optimization
of AG-390/10779040 led to a new compound LHB-8, giving an IC50 of 570 nM. The novel scaffolds
for PDE2 inhibitors reported in this paper might be valuable for development of PDE2 selective
inhibitors.
1
2
3
4
5
6
7
8
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
2. Methods and materials
2.1. Molecular modeling
In the first step, a pharmacophore model was generated for an expeditious screening of the
smallꢀmolecule database SPECS, in order to significantly reduce the size of the database. In the
second step, molecular docking was applied for a preliminary prediction of binding modes and
energies between the protein and molecules. The molecules in the dataset were further reduced in
this step. In the last step, molecular dynamics (MD) simulations, together with the molecular
mechanism/PoisonꢀBoltzman surface area (MMꢀPBSA) method[23ꢀ25] were performed for a more
precise prediction of binding modes and corresponding binding energies. The 30 molecules with
proper binding patterns and top binding energies made up the final dataset, followed by subsequent
enzymatic bioassay.
2.1.1. Database preparation
The database SPECS (http://www.specs.net), consisting of almost 200,000 small molecules, was
selected for our virtual screening and filtered by Lipinski's Rule of Five[26] using the modeling
software Molecular Operating Environment (MOE 2008.10), to form the initial dataset Dataset0.
Then multiple conformations for compounds were generated with the "conformation import" method
in MOE.
2.1.2. Structure-based virtual screening
3
ACS Paragon Plus Environment

Journal of Chemical Information and Modeling
Page 4 of 35
Crystal structures of PDE2A with bound inhibitors were used to build a wellꢀperformed 3D
pharmacophore model for screening of Dataset0. Five crystal structures were downloaded from
Protein Data Bank (PDB ID: 4HTX, 4JIB, 4D08, 4D09, and 4C1I) [20ꢀ22, 27] and applied to establish
the pharmacophore model. On basis of superposition of these five crystal structures, the spatial
positions and interactions between the ligands and the protein were carefully analyzed and common
structures in ligands such as aromatic rings, hydrophobic side chains, hydrogen bond donor and
acceptor groups were defined as pharmacophore features. The volume of the amino acid residues was
also applied as an exclusion feature to filter the compounds in the dataset. A test set, consisting of 30
active compounds (IC50 < 10 ꢀM) and 174 inactive compounds (IC50 ≥ 10 ꢀM) which were
randomly picked from the CheMBL database (http://www.ebi.ac.uk/chembl), was built for the
Goodness of hit (GH) test.[28] The resulting pharmacophore model was utilized to filter compounds
in Dataset0 and to save as Dataset1.
1
2
3
4
5
6
7
8
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 false positive compounds in virtual screening may interfere with other detecting methods,
leading to fake hits or lead compounds.[29] Thus, PAINS (panꢀassay interfering compound
substructures) screening was applied to removing possible false positive compounds to build up
Dataset2.
Molecular docking was performed for the compounds in Dataset2. Those with higher scores
than the reference thresholds and optimal binding patterns made up Dataset3. Worthy to be
mentioned is that the invariant residue Gln859 in the binding pocket may have two conformations for
its side chain, as depicted in Fig. S2. Thus, two crystal structures (4HTX and 4C1I) were utilized as
reference structures for molecular docking. The two metal ions in the binding pocket were kept as it
is in the crystal structures of PDE families.[1ꢀ3] Water molecules coordinating the two metal ions were
also retained after which hydrogen atoms were complemented by using H++ webserver
(http://biophysics.cs.vt.edu/) and protonation states were corrected.
Next，MD simulations were carried out to more precisely predict their binding patterns. The
software AMBER 10.0[30] was used to simulate binding of molecules in Dataset3 with PDE2A. The
4
ACS Paragon Plus Environment

Page 5 of 35
Journal of Chemical Information and Modeling
docking poses of molecules in Dataset3 were first calculated for the partial atomic charges by using
the HartreeꢀFock method at the 6ꢀ31G* level with Gaussian 03[31]. Then Antechamber[32] was applied
to fit the restricted electrostatic potential (RESP) and assign the general amber force field (GAFF)
parameters. The amber03 force field was utilized for this protein. The force field parameters for Zn2+
and Mg2+ were assigned with the "nonꢀbond model" method.[33] This simple model reproduced the
structural and energetic properties of the solvated ions in MD simulations. The results of simulations
agreed with experimental results in general[34] in this and our previous study.[33, 35] His660 and
His696, which coordinate with two metal ions, were protonated at the δ positions. The oxygen atom
bridging the two metal ions was treated as a hydroxide ion. His656, which is closest to the hydroxide
ion and able to capture a proton, is treated as HIP (protonated histidine). An 8 Å TIP3P water box in
the form of a truncated octahedron with Na+ (for 4HTX) or Clꢀ (for 4C1I) ions was added for
neutralizing.
1
2
3
4
5
6
7
8
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
Such modified ligands and protein were then submitted to MD simulations. Eight ns MD
simulations were carried out in the NPT ensemble with a constant pressure of 1 atm and a constant
temperature of 300 K. The periodic boundary conditions were adopted, along with a 8 Å cutoff for
longꢀrange electrostatic interactions with the partial mesh Ewald (PME) method.[36ꢀ37 ] The SHAKE
algorithm[38ꢀ39] were utilized for restriction of all bonds involving hydrogen atoms, and thereby the
time step was set to 2 fs. Since the graphics processing units (GPUs) are available in performing
floatingꢀpoint calculations, an Intel Xeon E5620 CPU and an NVIDIA Tesla C2050 GPU were
applied to accelerate the process of MD simulations for each system. Subsequent routine included
the extraction of 100 snapshots of the last 1 ns trajectories and MMꢀPBSA binding free energy
calculations with default parameters assigned.
In light of the MMꢀPBSA method[23ꢀ25, 40ꢀ42], the binding free energy (ꢁGbind) can be calculated by
the following equation 1: in which Gcomp, Grec and Glig represent the free energies of complex,
receptor and ligand, respectively.
ꢁGbind = Gcomp ꢀ Grecꢀ Glig
(1)
Gcomp is calculated by the sum of the MM energy EMM, the solvation free energy Gsolv, comp, and
5
ACS Paragon Plus Environment

Journal of Chemical Information and Modeling
Page 6 of 35
the entropy contribution Scom, by equation 2. The same calculation can be applied to Grec and Glig.
1
2
3
4
5
6
7
8
G
comp = EMM + Gsolv, comp ꢀ TScom
Equation 3 can be obtained from above:
ꢁGbind = ꢁEMM + ꢁGsolvꢀ TꢁS
(2)
(3)
9
ꢁEMM represents the gas phase interaction energy and can be decomposed into EMM, comp, EMM, rec and
EMM, lig. Solvation free energy can be represented by the sum of the electrostatic solvation free energy
and nonpolar solvation free energy. The electrostatic solvation free energy, ꢁGPB, can be calculated
by the Poisson Boltzmann (PB) equation, while the nonpolar solvation free energy is proportional to
the solvation accessible surface area (SASA) and is calculated by the following equations:
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
ꢁGsolv = ꢁGPB + ꢁGnp
ꢁGnp = γ SASA + b
(4)
(5)
The calculation of the entropy contribution for the PDE2ꢀligand complexes was omitted, since it
is extremely timeꢀconsuming for large proteinꢀligand systems.
After MD simulations, the compounds in Dataset3 with negative binding energies at top rank
and optimal binding patterns as inspected visually formed Dataset4. Finally, all of the compounds in
Dataset4 were purchased from SPECS and subjected to bioassay for their inhibition on PDE2A.
2.2. Bioassay
2.2.1. Protein expression and purification
The recombinant pET15bꢀPDE2A plasmid for expression of the catalytic domain (residues
580ꢀ919) was subcloned and purified following protocols previously reported [34, 43ꢀ46], and was
transferred into E. colistrain BL21 (Codonplus, Stratagene). The E. colicells carrying the
recombinant plasmid were grown in an 2XYT medium (containing 100 ꢂg/mL ampicillin and 20
ꢂg/mL chloramphenicol) at 37 oC until OD600
=
0.6ꢀ0.8. Then, 0.1 mM
isopropylꢀβꢀDꢀthiogalactopyranoside was added to induce the PDE2A protein expression at 16 oC for
24 h. The nickel nitriloacetic acid (NiꢀNTA) column (Qiagen) was used to purify PDE2A proteins.
The concentration of the PDE2 fractions was estimated according to the absorbance at 280 nm
(calculated by the ProtParam software). A typical batch of purification yielded 10–20 mg PDE2A
6
ACS Paragon Plus Environment

Page 7 of 35
Journal of Chemical Information and Modeling
protein from a 1.0 L cell culture. The PDE2A proteins had purity greater than 90% as shown by
1
2
3
SDSꢀPAGE.
4
5
6
2.2.2. Enzymatic assays
3
7
8
The enzymatic activities of the catalytic domain of PDE2A were measured with HꢀcGMP as
9
the substrate in buffer of 50 mM Tris pH = 8.0, 10 mM MgCl2 and 0.5 mM dithiothreitol, 20,000 ꢀ
30,000 cpm 3HꢀcGMP per assay. Inhibitors solution (DMSO) in different concentration was added to
PDE2A enzyme. The enzymatic reaction was performed at room temperature of 25 °C for 15 min
and then terminated by addition of 0.2 M ZnSO4. The reaction product was precipitated out by 0.2 N
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
3
Ba(OH)2 and the unreacted HꢀcGMP remained in the supernatant. The radioactivity in the
supernatant was measured in 2.5 ml of Ultima Gold liquid scintillation cocktails (PerkinElmer) by a
PerkinElmer 2910 liquid scintillation counter. At least eight concentrations of inhibitors were used
for measurement of inhibitors IC50 value. Each measurement was repeated three times, and IC50
values were calculated by nonlinear regression. The mean values of the measurements were
considered as the final IC50 values with the SD values of the measurements. In this assay, EHNA
purchased from SIGMA was used as the reference compound.
2.3. General chemistry
1H NMR and 13C NMR spectra were recorded on a BrukerBioSpin GmbH spectrometer at 400.1
and 100.6 MHz, respectively. Coupling constants are given in Hz. Highꢀresolution mass spectra
(HRMS) were obtained on an ITꢀTOF mass spectrometer. Flash column chromatography was
performed using silica gel (200–300 mesh) purchased from Qingdao Haiyang Chemical Co. Ltd.
Thinꢀlayer chromatography was performed on precoated silica gel Fꢀ254 plates (0.25 mm, Qingdao
Haiyang Chemical Co. Ltd) and visualized with UV light. All the starting materials and reagents
were purchased from commercial suppliers and used directly without further purification.
2.3.1. General procedures[47] for synthesis of compounds LHB-1 to LHB-8
To the solution of Benzo[cd]indolꢀ2(1H)ꢀone (10 mmol) in CHCl3 (30 mL), chlorosulfonic Acid
(30.0 mmol) was added dropwise in 10 min. The mixture was then stirred at room temperature for 6
h and poured into ice water to quench. The residue was extracted with 10 mL dichloromethane twice,
7
ACS Paragon Plus Environment

Journal of Chemical Information and Modeling
Page 8 of 35
washed with brine, dried over anhydrous Na2SO4. The solvent was removed in vacuum to give the
crude product M1 (46%). The corresponding amine (1.2 mmol) was added to the M1 (1 mmol) in
pyridine (10 mL). After the mixture was stirred at room temperature for 24 h, the reaction was
diluted with ethyl acetate, washed with saturated sodium bicarbonate solution, and brine, dried over
MgSO4. The solvent was removed in vacuum. The crude product was purified by silica gel
chromatography with ethyl acetate/methanol (100:1) to give the final compound. Bioassay was then
performed for compounds LHB-1 to LHB-8 with the protocol described above.
1
2
3
4
5
6
7
8
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
2.3.2. N-(2-isopropylphenyl)-2-oxo-1,2-dihydrobenzo[cd]indole-6-sulfonamide (LHB-1)
Yellow solid, 68%; 1H NMR (400 MHz, DMSO) δ 11.12 (s, 1H), 9.80 (s, 1H), 8.55 (d, J = 8.3
Hz, 1H), 8.10 (d, J = 6.9 Hz, 1H), 7.93 – 7.80 (m, 1H), 7.15 (d, J = 4.4 Hz, 1H), 7.06 – 6.96 (m, 1H),
13
6.91 (d, J = 7.8 Hz, 1H), 2.96 (dt, J = 13.6, 6.8 Hz, 1H), 0.67 (d, J = 6.8 Hz, 6H); C NMR (101
MHz, DMSO) δ 168.76, 145.75, 142.68, 132.64, 132.59, 130.26, 129.49, 128.64, 127.88, 127.27,
126.68, 126.18, 126.06, 125.86, 124.78, 124.54, 104.62, 26.48, 23.21; LCꢀMS (ESI), m/z for
C20H18N2O3S ([M ꢀ H]ꢀ ): Calcd 365.43, found 365.4.
2.3.3. 2-oxo-N-(3,4,5-trimethoxyphenyl)-1,2-dihydrobenzo[cd]indole-6-sulfonamide(LHB-2)
Yellow solid, 53%; 1H NMR (400 MHz, DMSO) δ 11.15 (s, 1H), 10.34 (s, 1H), 8.67 (d, J = 8.3
Hz, 1H), 8.16 (d, J = 7.6 Hz, 1H), 8.10 (d, J = 6.9 Hz, 1H), 7.94 (t, J = 7.7 Hz, 1H), 7.06 (d, J = 7.6
13
Hz, 1H), 6.31 (s, 2H), 3.57 (s, 6H), 3.50 (s, 3H); C NMR (101 MHz, DMSO) δ 169.21, 153.36,
143.56, 134.31, 133.86, 131.14, 129.53, 127.69, 127.23, 126.35, 125.43, 124.85, 105.13, 97.97,
60.50, 56.16; LCꢀMS (ESI), m/z for C20H18N2O6S ([M ꢀ H]ꢀ ): Calcd 413.43, found 413.4.
2.3.4. N-(3,5-dichloro-4-hydroxyphenyl)-2-oxo-1,2-dihydrobenzo[cd]indole-6-sulfonamide
(LHB-3)
Yellow solid, 61%;1H NMR (400 MHz, DMSO) δ 11.20 (s, 1H), 10.53 (s, 1H), 10.04 (s, 1H),
8.62 (d, J = 8.4 Hz, 1H), 8.12 (d, J = 7.0 Hz, 1H), 8.07 (d, J = 7.6 Hz, 1H), 8.01ꢀ7.89 (m, 1H), 7.06
13
(d, J = 7.6 Hz, 1H), 6.96 (s, 2H); C NMR (101 MHz, DMSO) δ 169.16, 146.48, 143.82, 134.24,
131.31, 130.60, 129.31, 127.37, 127.01, 126.42, 125.55, 124.63, 123.01, 120.98, 105.16; LCꢀMS
(ESI), m/z for C17H10Cl2N2O4S ([M ꢀ H]ꢀ ): Calcd 408.24, found 408.2.
8
ACS Paragon Plus Environment

Page 9 of 35
Journal of Chemical Information and Modeling
2.3.5. N-(3-chloro-4-fluorophenyl)-2-oxo-1,2-dihydrobenzo[cd]indole-6-sulfonamide (LHB-4)
Yellow solid, 81%;1H NMR (400 MHz, DMSO) δ 11.18 (s, 1H), 10.74 (s, 1H), 8.64 (d, J = 8.3
Hz, 1H), 8.16 – 8.07 (m, 2H), 7.96 (dd, J = 8.4, 7.1 Hz, 1H), 7.24 (t, J = 9.1 Hz, 1H), 7.16 (dd, J =
6.5, 2.7 Hz, 1H), 7.05 (d, J = 7.6 Hz, 1H), 6.98 (ddd, J = 9.0, 4.1, 2.7 Hz, 1H); 13C NMR (101 MHz,
DMSO) δ 169.16, 143.84, 134.40, 131.35, 129.27, 127.36, 126.99, 126.47, 125.54, 124.58, 118.59,
118.41, 116.49, 109.21, 109.01, 105.13; LCꢀMS (ESI), m/z for C17H10ClFN2O3S ([M ꢀ H]ꢀ ): Calcd
375.79, found 375.7.
1
2
3
4
5
6
7
8
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
2.3.6. N-(3,4-difluorophenyl)-2-oxo-1,2-dihydrobenzo[cd]indole-6-sulfonamide (LHB-5)
Yellow solid, 72%; 1H NMR (400 MHz, DMSO) δ 11.19 (s, 1H), 10.77 (s, 1H), 8.65 (d, J = 8.4
Hz, 1H), 8.12 (t, J = 6.9 Hz, 2H), 8.02 – 7.90 (m, 1H), 7.24 (dd, J = 19.5, 9.2 Hz, 1H), 7.11 – 7.00
(m, 2H), 6.84 – 6.76 (m, 1H); 13C NMR (101 MHz, DMSO) δ 169.16, 143.84, 134.40, 131.35,
129.27, 127.36, 126.99, 126.47, 125.54, 124.58, 118.59, 118.41, 116.49, 109.21, 109.01, 105.13;
LCꢀMS (ESI), m/z for C17H10F2N2O3S ([M ꢀ H]ꢀ ): Calcd 359.33, found 359.3.
2.3.7. N-(3,5-bis(trifluoromethyl)phenyl)-2-oxo-1,2-dihydrobenzo[cd]indole-6-sulfonamide
(LHB-6)
1
Yellow solid, 62%; H NMR (400 MHz, DMSO) δ 11.40 (d, J = 57.7 Hz, 1H), 11.21 (s, 1H),
8.65 (t, J = 10.0 Hz, 1H), 8.25 (t, J = 10.0 Hz, 1H), 8.11 (d, J = 6.9 Hz, 1H), 7.98 (dd, J = 8.4, 7.1 Hz,
13
1H), 7.66 (s, 1H), 7.60 (s, 2H), 7.09 (d, J = 7.7 Hz, 1H); C NMR (101 MHz, DMSO) δ 169.05,
144.38, 140.22, 134.71, 131.83, 131.59, 131.50, 128.90, 127.43, 126.54, 126.30, 125.66, 124.60,
124.46, 121.89, 118.33, 116.84, 104.99; LCꢀMS (ESI), m/z for C19H10F6N2O3S ([M ꢀ H]ꢀ ): Calcd
459.35, found 459.3.
2.3.8. 4-amino-2-chlorophenyl 2-oxo-1,2-dihydrobenzo[cd]indole-6-sulfonate (LHB-7)
Yellow solid, 77%; 1H NMR (400 MHz, DMSO) δ 11.17 (s, 1H), 10.20 (s, 1H), 10.06 (s, 1H),
8.63 (d, J = 8.1 Hz, 1H), 8.11 (d, J = 6.4 Hz, 1H), 8.04 – 7.89 (m, 2H), 7.02 (d, J = 7.0 Hz, 1H), 6.93
(s, 1H), 6.74 (s, 2H); 13C NMR (101 MHz, DMSO) δ 169.18, 150.80, 143.48, 133.98, 131.09, 129.59,
127.57, 127.31, 126.38, 125.41, 124.78, 123.53, 122.10, 119.86, 117.25, 105.12; LCꢀMS (ESI), m/z
for C17H11ClN2O4S ([M ꢀ H]ꢀ ): Calcd 373.80, found 373.8.
9
ACS Paragon Plus Environment

Journal of Chemical Information and Modeling
Page 10 of 35
2.3.9. 4-amino-3-chlorophenyl 2-oxo-1,2-dihydrobenzo[cd]indole-6-sulfonate (LHB-8)
1
2
3
1
Yellow solid, 81%, H NMR (400 MHz, DMSO) δ 11.16 (s, 1H), 9.76 (s, 2H), 8.72 (d, J = 8.4 Hz,
4
5
6
7
8
1H), 8.14 (d, J = 6.9 Hz, 1H), 8.03 – 7.86 (m, 2H), 7.21 (d, J = 8.5 Hz, 1H), 7.03 (d, J = 7.6 Hz, 1H),
13
6.81 (dd, J = 8.5, 2.4 Hz, 1H), 6.70 – 6.59 (m, 1H). C NMR (101 MHz, DMSO) δ 169.26, 152.46,
9
143.18, 133.11, 130.57, 130.52, 130.37, 129.01, 127.58, 127.07, 126.44, 125.15, 123.56, 119.20,
115.51, 104.95, 31.14.HRMS (ESIꢀTOF) m/z: [M + Na]+Calcd for C17H11N2O4SCl 397.0020; Found
397.0030.
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
3. Results and discussion
3.1. Identification of PDE2 inhibitors with novel scaffolds by structure-based virtual screening
and bioassay
A method combining pharmacophore model screening, molecular docking, molecular dynamics
simulations was utilized to screen hits. The schematic diagram of the virtual screening strategy in
this study is presented in Fig. 1.
Insert Fig. 1
The database SPECS of about 200,000 small molecules was preꢀfiltered by Lipinski’s Rule of
Five with the "conformation import" method in MOE 2008.10, leading to formation of Dataset0 that
contains 112,755 molecules passed the filtrations. Dataset0 was rapidly screened by the
pharmacophore model and then filtered by PAINS. Our initial pharmacophore model with 4 features
was generated by the consensus strategy from the superposition of five PDE2A crystal structures
(Fig. S6. ESI). However, too many molecules in Dataset0 passed the pharmacophore screening and
apparently this pharmacophore model is overꢀsimplified. We manually modified the model by
adding more features and performed the GH test by using known compounds. The pharmacophore
model was tuned by 16 hits out of the 30 active compounds and 6 hits out of the 174 inactive
compounds in the GH test set. The final GH test score of the pharmacophore model was 0.66, which
is significantly better than 0.5, a threshold value regarded as a good model,[48ꢀ50] thus implying that
our pharmacophore model is suitable for screening purpose. The final pharmacophore model
10
ACS Paragon Plus Environment

Page 11 of 35
Journal of Chemical Information and Modeling
contained 7 features and the exclusion feature, as shown in Fig. 2. After the pharmacophore
1
2
3
4
5
screening, 2,757 molecules fitted all the features in the model and were reserved as Dataset1.
Insert Table 1 and Insert Fig. 2
6
7
8
To avoid interference of false positive compounds with our subsequent screening, a PAINS
9
screening on compounds in Dataset1 was performed using the online program Free AMDEꢀTox
Filtering Tool (FAFꢀDrugs3, http://fafdrugs3.mti.univꢀparisꢀdiderot.fr),[29] leading to 2,603 of the
2,757 compounds, which were deposited to Dataset2.
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 order to obtain the optimal parameters for reliable molecular docking, the inhibitors in the
crystal structures (PDB ID: 4HTX and 4C1I) were redocked back to the structures. It turned out that
the Surflexꢀdock[51] method embedded in TriposSybyl 2.0 was suitable for PDE2A. The average
RMSD values between the original Xꢀray pose and the top 10 docking poses were less than 1.5 Å for
the ligands BAY-60-7550 and EHNA, implying the reliability of the docking. Thus, the identical
parameters were used for the docking screening of molecules in Dataset2. Fifty conformations were
randomly generated for each molecule. The means of top ten docking scores for 4HTX (11.13) and
4C1I (6.33) were used as the thresholds in the docking screening. Since the threshold of 11.13 for
4HTX was too high for most compounds in Dataset2 to pass the docking screening, it was reduced
to 5.40 that was obtained from the test docking of the similar crystal structure 4JIB.
In the hydrophobic clamp of the catalytic site of PDE2, the substrate cAMP or cGMP was
sandwiched between Phe862 and Phe830 and formed hydrogen bonds with Gln859. Similar
interactions were also observed for competitive inhibitors of PDE2. Thus, Gln859 and Phe862 are
recognized as two conservative amino acid residues during the receptor–ligand recognitions. In
addition, the side chain of Gln859 may exist in two conformations, which induces distinct binding
patterns in the PDE2A crystal structures. Accordingly, both the crystal structures of 4HTX and 4C1I
were utilized during our docking screening. Only molecules fitting the pharmacophore model were
retained as potential PDE2 inhibitors. We chose the top 211 compounds (134 molecules based on the
crystal structure of 4HTX and 77 molecules based on 4C1I) in view of higher docking scores than
the thresholds and appropriate patterns to compose Dateset3 for further study. The binding poses of
11
ACS Paragon Plus Environment

Journal of Chemical Information and Modeling
Page 12 of 35
some molecules are shown in Fig. S3.
1
2
3
4
5
6
7
8
9
To evaluate the stability of each system of PDE2 in complex with small molecules, molecular
dynamics simulations were conducted in Amber 10.0. The results of 8 ns MD simulations and the
MMꢀPBSA binding free energy calculations with the stable trajectories implied that 30 systems met
the requirements sufficiently to cover both stable binding patterns and higher energies. The RMSD
values of the backbone atoms for 4HTX and 4C1I over MD simulations as references are shown in
Fig. S4. As a result, these 30 compounds were retained to constitute Dataset4. The RMSD curves
during MD simulations and predicted binding energies of the selected 30 compounds with PDE2A
are shown in Fig. S5 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
The 30 compounds were purchased from SPECS for subsequent bioassay. EHNA was used as
the reference compound and has an IC50 of 2.6 ꢀM which was comparable to the literature value of
0.8 ꢀM[15]. Nine of 30 tested compounds were confirmed to be PDE2A inhibitors, and inhibited ≥ 50%
PDE2 activity at their 50 ꢀM. Three of them even showed ≥ 50% inhibition at their 10 ꢀM (Fig.3).
Insert Fig. 3 and Fig. 4
3.2. New scaffolds for PDE2 inhibitors and their binding patterns during MD simulations
Among the 9 PDE2 inhibitors we identified, compound AGꢀ690/12244899 gave the best IC50
for PDE2. However, its scaffold is similar to cAMP and cGMP[52] and may have low selectivity over
other PDE families. Thus, AQ-390/10779040 and AG-690/10776061, which share the same novel
scaffold of benzo[cd]indolꢀ2(1H)ꢀone, were selected for further structural modification. The
inhibitory profiles of these two compounds with an unambiguous doseꢀdependent effect are shown in
Fig. 4 (AQ-390/10779040, IC50 = 4.6 ꢀM; AG-690/10776061, IC50 = 9.8 ꢀM).
The binding modes of the new PDE2 inhibitors after 8 ns MD were analyzed for further
structural modification. The predicted binding patterns of 4 new inhibitors are illustrated in Fig. 5.
AQ-390/10779040 and AG-690/10776061 interact with PDE2A via two hydrogen bonds and π–π
stacking interactions (picture 1 and picture 2, Fig. 5) The molecules are located in the hydrophobic
clamp: residue Phe862, Gln859, Phe830 are on one end, while Leu770 and Ile866 form another side
12
ACS Paragon Plus Environment

Page 13 of 35
Journal of Chemical Information and Modeling
(only Leu770 is shown in the figure). For AG-670/40741585 (picture 3, Fig. 5), two hydrogen bonds
are formed with Gln859 and Tyr827, respectively. The π–π stacking against Phe862 is also observed.
The interactions between AG-690/12244899 and PDE2A include two hydrogen bonds with Gln859
and a stack against Phe862 (picture 4, Fig. 5). Besides, the molecule stretches into the hydrophobic
1
2
3
4
5
6
7
8
9
pocket which was mentioned above.
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
Insert Fig. 5
In summary, the binding patterns of these four molecules are very similar. One or two hydrogen
bonds with Gln859 and π–π stacking with Phe862 are formed by all of them, which are commonly
considered as conservative interactions for PDE2 inhibitors. All but AG-670/40741585 reach the
hydrophobic pocket in one end and occupy it. This binding mode may indicate a new way for
discovery of novel PDE2 inhibitors.
3.3. Design and synthesis of potent PDE2 inhibitors with the benzo[cd]indol-2(1H)-one scaffold
Based on our experimental results and the binding patterns analysis, we selected the scaffold of
AQ-390/10779040 and AG-690/10776061 for further structural modification (Scheme 1). According
to the binding modes, we kept the benzo[cd]indolꢀ2(1H)ꢀone core to retain the hydrogen bond
interaction with Gln859 and π–π stacking with Phe862, and initially focused on modification of the
phenyl group. We aimed to improve inhibition by adding interactions between the substituent groups
and PDE2.
Insert Scheme 1
Different substituent groups were attached to the phenyl group to explore the effect on the
inhibition (Table 2). Compared with AQ-390/10779040 and AG-690/10776061, LHB-1 and LHB-2
exhibited decrease in potency whereas six other compounds of LHB-4 to LHB-8 demonstrated
potency enhancement to a certain degree. We observed that the halogen groups on the phenyl group
improved inhibition while the methoxy or methyl groups didn’t impact significantly. The best
compound was LHB-8 giving the IC50 value of 570 nM, which exhibited 8ꢀfold increase in the
potency to the hit compound AQ-390/10779040. The inhibitory profiles of these compounds clearly
showed doseꢀdependent pattern (Fig. S7). PAINS screening was applied to these derivatives and only
13
ACS Paragon Plus Environment

Journal of Chemical Information and Modeling
Page 14 of 35
LHB-3 failed to pass the test.
1
2
3
Insert Table 2
4
5
Molecular docking method was used to explore interactions of LHB-8 with PDE2 for the
6
7
8
9
structureꢀactivity relationship (SAR) analysis. The best docked pose of LHB-8 from the
Surflexꢀdock method was selected and superposed on that of AQ-390/10779040 (Fig.6). In addition
to the common interactions with Gln859 and Phe862, LHB-8 forms an extra hydrogen bond with
Asp808. To further explore contribution of each component of LHB-8 to the inhibition, the binding
free energy was calculated with the MMꢀPBSA method. Stable MD simulations trajectories were
utilized for data extracting and binding free energy calculating. The predicted binding free energies
were ꢀ29.00 and ꢀ32.23 kcal/mol for the complexes of PDE2-AQ-390/10779040 and PDE2-LHB-8,
respectively, in agreement with their IC50 values (4.6 ꢀM and 0.57 ꢀM). The energies for
electrostatic interaction (ꢁGele) and polar contribution to solvation (ꢁGele, sol) of the complex of
PDE2ꢀ LHB-8 are 6.8 and 5.2 kcal/mol lower than those for AQ-390/10779040. We speculated that
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 polar contribution may play an important role for the enhanced inhibition.
Insert Fig. 6
Energy decomposition for the binding free energies was also performed to evaluate the
contribution of each residue in the binding pocket of PDE2 with the MMꢀPBSA method (Fig. 7).
Usually, a residue is considered to be important for recognition of ligands if the interaction energy
with ligand is lower than ꢀ1 kcal/mol. Our results suggest that the following residues may be
important for inhibition of AQ-390/10779040 and LHB-8 on PDE2: Ala771 (ꢀ2.51 and ꢀ2.49
kcal/mol), Asp808 (ꢀ1.33 and ꢀ1.94 kcal/mol), Ile826 (ꢀ1.45 and ꢀ1.63 kcal/mol), Phe830 (ꢀ1.65 and
ꢀ1.18 kcal/mol), Gln859 (ꢀ4.22 and ꢀ4.19 kcal/mol), Phe862 (ꢀ3.42 and ꢀ3.40 kcal/mol) and Ile866
(ꢀ0.93 and ꢀ0.95 kcal/mol). Among these residues, Gln859 and Phe862 mostly contributed to the total
binding free energies, in consistence with the binding patterns in which all ligands formed hydrogen
bonds with Gln859 and π–π stacking with Phe862. Thr768 and Asp808 in the hydrophobic pocket
made the major difference in energy contribution. Thr768 and Asp808 contribute to the PDE2ꢀ
LHB-8 complex with the binding free energies of 0.7 and 0.6 kcal/mol lower than those of
14
ACS Paragon Plus Environment

Page 15 of 35
Journal of Chemical Information and Modeling
PDE2ꢀAQ-390/10779040, which is consistent with the previous result that the polar contribution
accounts for the improved inhibition against PDE2. From the MD binding patterns, one end of
LHB-8 is located in the hydrophobic clamp and another end reached the hydrophobic pocket to form
a hydrogen bond with Asp808 (Fig. 6). Compared with AQ-390/10779040, LHB-8 adopted a
slightly different conformation for stronger interaction with Thr768. The energy decomposition of
residues helps to interpret the binding patterns from the MD simulations and is supplementary to the
binding affinity results.
1
2
3
4
5
6
7
8
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
Insert Fig. 7
4. Conclusions
A structureꢀbased virtual screening method combining pharmacophore model screening,
molecular docking, MD simulations, and bioassay was performed for discovery of novel PDE2A
inhibitors with new scaffolds in this study. Thirty compounds from the SPECS database were picked
out by the screening and 9 of them exhibited the inhibition of
< 50 ꢂM on PDE2. Five kinds of new
scaffolds were discovered out of the 9 hits. Structural optimization of AQ-390/10779040 (IC50 = 4.6
ꢂM) led to discovery of a new compound LHB-8 with an improved affinity of 570 nM. The strategy
we used in this study makes a good concession between computational cost and improvement of the
hit ratio of drug discovery and may have a wide application in further rational drug design.
■ ASSOCIATED CONTENT
Supporting Information
The Supporting Information is available free of charge on the ACS Publications website.
■ AUTHOR INFORMATION
Corresponding Authors
* Eꢀmail: wyinuo3@mail.sysu.edu.cn (Y.W)
* Eꢀmail: luohb77@mail.sysu.edu.cn; Fax: +86ꢀ20ꢀ3994 3000 (H.B.L.).
Notes
15
ACS Paragon Plus Environment

Journal of Chemical Information and Modeling
Page 16 of 35
The authors declare no competing financial interest.
1
2
3
4
5
6
■Acknowledgements
7
8
9
This work was supported by Natural Science Foundation of China (81522041, 21572279, 81373258,
and 21402243), Guangdong Science Foundation (2014A020210009), Science Foundation of
Guangzhou City (2014J4100165), Research Fund for the Doctoral Program of Higher Education of
China (20130171110096), Fundamental Research Funds for the Central Universities (Sun YatꢀSen
University) (No. 13ykpy10), and Guangdong Province Higher Vocational Colleges & Schools Pearl
River Scholar Funded Scheme (2016).We also thank Special Program for Applied Research on Super
Computation of the NSFCꢀGuangdong Joint Fund (the second phase) for providing the
supercomputing service.
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
References
[1] Manallack, D. T.; Hughes, R. A.; Thompson, P. E. The next generation of phosphodiesterase
inhibitors: Structural clues to ligand and substrate selectivity of phosphodiesterases. J. Med. Chem.
2005, 48, 3449ꢀ3462.
[2] Jeon, Y. H.; Heo, Y. S.; Kim, C. M.; Hyun, Y. L.; Lee, T. G.; Ro, S.; Cho, J. M.
Phosphodiesterase: overview of protein structures, potential therapeutic applications and recent
progress in drug development. Cell. Mol. Life Sci. 2005, 62, 1198ꢀ1220.
[3] Bender, A. T.; Beavo, J. A. Cyclic nucleotide phosphodiesterases: Molecular regulation to
clinical use. Pharmacol. Rev. 2006, 58, 488ꢀ520.
[4] Beavo, J. A.; Brunton, L. L. Cyclic nucleotide research ꢀ still expanding after half a century. Nat.
Rev. Mol. Cell Bio. 2002, 3, 710ꢀ718.
[5] Lugnier, C. Cyclic nucleotide phosphodiesterase (PDE) superfamily: A new target for the
development of specific therapeutic agents. Pharmacol. Therapeut. 2006, 109, 366ꢀ398.
[6] Setter, S. M.; Iltz, J. L.; Fincham, J. E.; Campbell, R. K.; Baker, D. E. Phosphodiesterase 5
inhibitors for erectile dysfunction. Ann. Pharmacother. 2005, 39, 1286ꢀ1295.
16
ACS Paragon Plus Environment

Page 17 of 35
Journal of Chemical Information and Modeling
[7] Liu, Y. G.; Shakur, Y.; Yoshitake, M.; Kambayashi, J. Cilostazol (Pletal (R)): A dual inhibitor
of cyclic nucleotide phosphodiesterase type 3 and adenosine uptake. Cardiovascular Drug Rev. 2001,
19, 369ꢀ386.
1
2
3
4
5
6
7
8
9
[8] Giembycz, M. A. Phosphodiesterase 4 inhibitors and the treatment of asthma ꢀ Where are we
now and where do we go from here? Drugs 2000, 59, 193ꢀ212.
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
[9] Zhu, J.; Mix, E.; Winblad, B. The antidepressant and Antiinflammatory effects of rolipram in the
central nervous system. Cns Drug Rev. 2001, 7, 387ꢀ398.
[10]Card, G. L.; England, B. P.; Suzuki, Y.; Fong, D.; Powell, B.; Lee, B.; Luu, C.; Tabrizizad, M.;
Gillette, S.; Ibrahim, P. N.; Artis, D. R.; Bollag, G.; Milburn, M. V.; Kim, S. H.; Schlessinger, J.;
Zhang, K. Y. J. Structural basis for the activity of drugs that inhibit phosphodiesterases. Structure
2004, 12, 2233ꢀ2247.
[11]Beavo, J. A. Cyclic nucleotide phosphodiesterases: Functional implications of multiple isoforms.
Physiol. Rev. 1995, 75, 725ꢀ748.
[12]Conti, M.; Beavo, J. Biochemistry and physiology of cyclic nucleotide Phosphocliesterases:
Essential components in cyclic nucleotide signaling. Annu. Rev.Biochem. 2007, 76, 481ꢀ511.
[13]Boess, F. G.; Hendrix, M.; van der Staay, F. J.; Erb, C.; Schreiber, R.; van Staveren, W.; de
Vente, J.; Prickaerts, J.; Blokland, A.; Koenig, G. Inhibition of phosphodiesterase 2 increases
neuronal cGMP, synaptic plasticity and memory performance. Neuropharmacology 2004, 47,
1081ꢀ1092.
[14]Martinez, S. E.; Wu, A. Y.; Glavas, N. A.; Tang, X. B.; Turley, S.; Hol, W. G. J.; Beavo, J. A.
The two GAF domains in phosphodiesterase 2A have distinct roles in dimerization and in cGMP
binding. P. Natl. Acad. Sci. USA 2002, 99, 13260ꢀ13265.
[15]Podzuweit, T.; Nennstiel, P.; Mueller, A. Isozyme selective inhibition of cGMPꢀstimulated
cyclic nucleotide phosphodiesterases by erythroꢀ9ꢀ(2ꢀHydroxyꢀ3ꢀnonyl) adenine. Cell. Signal. 1995,
7, 733ꢀ738.
[16]RivetꢀBastide, M.; Vandecasteele, G.; Hatem, S.; Verde, I.; Benardeau, A.; Mercadier, J.ꢀJ.;
Fischmeister, R. CGMPꢀstimulated cyclic nucleotide phosphodiesterase regulates the basal calcium
17
ACS Paragon Plus Environment

Journal of Chemical Information and Modeling
Page 18 of 35
current in human atrial myocytes. J.Clin. Invest. 1997, 99, 2710ꢀ2718.
1
2
3
4
5
6
7
8
9
[17]Rutten, K.; Prickaerts, J.; Hendrix, M.; van der Staay, F. J.; Sik, A.; Blokland, A.
Timeꢀdependent involvement of cAMP and cGMP in consolidation of object memory: Studies using
selective phosphodiesterase type 2, 4 and 5 inhibitors. Eur. J. Pharmacol. 2007, 558, 107ꢀ112.
[18]Chambers, R. J.; Abrams, K.; Garceau, N. Y.; Kamath, A. V.; Manley, C. M.; Lilley, S. C.; Otte,
D. A.; Scott, D. O.; Sheils, A. L.; Tess, D. A.; Vellekoop, A. S.; Zhang, Y.; Lam, K. T. A new
chemical tool for exploring the physiological function of the PDE2 isozyme. Bioorg. Med. Chem.
Lett. 2006, 16, 307ꢀ310.
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
[19]Seybold, J.; Thomas, D.; Witzenrath, M.; Boral, S.; Hocke, A. C.; Burger, A.; Hatzelmann, A.;
Tenor, H.; Schudt, C.; Krull, M.; Schutte, H.; Hippenstiel, S.; Suttorp, N. Tumor necrosis
factorꢀalphaꢀdependent expression of phosphodiesterase 2: role in endothelial hyperpermeability.
Blood 2005, 105, 3569ꢀ3576.
[20]Buijnsters, P.; De Angelis, M.; Langlois, X.; Rombouts, F. J. R.; Sanderson, W.; Tresadern, G.;
Ritchie, A.; Trabanco, A. A.; VanHoof, G.; Van Roosbroeck, Y.; Andres, J.ꢀI. StructureꢀBased
Design of a Potent, Selective, and Brain Penetrating PDE2 Inhibitor with Demonstrated Target
Engagement. Acs. Med. Chem. Lett. 2014, 5, 1049ꢀ1053.
[21]Rombouts, F. J. R.; Tresadern, G.; Buijnsters, P.; Langlois, X.; Tovar, F.; Steinbrecher, T. B.;
Vanhoof, G.; Somers, M.; Andres, J.ꢀI.; Trabanco, A. A. Pyrido[4,3ꢀe][1,2,4]triazolo[4,3ꢀa]pyrazines
as Selective, Brain Penetrant Phosphodiesterase 2 (PDE2) Inhibitors. Acs. Med. Chem. Lett. 2015, 6,
282ꢀ286.
[22]Zhu, J.; Yang, Q.; Dai, D.; Huang, Q. Xꢀray Crystal Structure of Phosphodiesterase 2 in
Complex with a Highly Selective, Nanomolar Inhibitor Reveals a BindingꢀInduced Pocket Important
for Selectivity. J. Am. Chem. Soc. 2013, 135, 11708ꢀ11711.
[23]Massova, I.; Kollman, P. A. Combined molecular mechanical and continuum solvent approach
(MMꢀPBSA/GBSA) to predict ligand binding. Perspect. Drug Discov. 2000, 18, 113ꢀ135.
[24]Hou, T.; Wang, J.; Li, Y.; Wang, W. Assessing the Performance of the MM/PBSA and
MM/GBSA Methods. 1. The Accuracy of Binding Free Energy Calculations Based on Molecular
18
ACS Paragon Plus Environment

Page 19 of 35
Journal of Chemical Information and Modeling
Dynamics Simulations. J. Chem. Inf. Model. 2011, 51, 69ꢀ82.
1
2
3
4
5
6
7
8
[25]Liu, M.; Yuan, M.; Luo, M.; Bu, X.; Luo, H.ꢀB.; Hu, X. Binding of curcumin with glyoxalase I:
Molecular docking, molecular dynamics simulations, and kinetics analysis. Biophys. Chem. 2010,
147, 28ꢀ34.
9
[26]Lipinski, C. A.; Lombardo, F.; Dominy, B. W.; Feeney, P. J., Experimental and computational
approaches to estimate solubility and permeability in drug discovery and development settings. Adv.
Drug Del. Rev. 1997, 23, 3ꢀ25.
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
[27]Plummer, M. S.; Cornicelli, J.; Roark, H.; Skalitzky, D. J.; Stankovic, C. J.; Bove, S.; Pandit, J.;
Goodman, A.; Hicks, J.; Shahripour, A.; Beidler, D.; Lu, X. K.; Sanchez, B.; Whitehead, C.; Sarver,
R.; Braden, T.; Gowan, R.; Shen, X. Q.; Welch, K.; Ogden, A.; Sadagopan, N.; Baum, H.; Miller, H.;
Banotai, C.; Spessard, C.; Lightle, S. Discovery of potent, selective, bioavailable phosphodiesterase
2 (PDE2) inhibitors active in an osteoarthritis pain model, Part I: Transformation of selective
pyrazolodiazepinone phosphodiesterase 4 (PDE4) inhibitors into selective PDE2 inhibitors. Bioorg.
Med. Chem. Lett. 2013, 23, 3438ꢀ3442.
[28]Osman, F. G. D.; Henry, R. Metric for analyzing hit lists and pharmacophores. In.
Phamacophore Perception, Development, and Use in Drug Design; Guner., O. F., Ed.; International
University Line: La Jolla, CA, 2000; pp 193ꢀ210.
[29]Baell, J. B.; Holloway, G. A. New Substructure Filters for Removal of Pan Assay Interference
Compounds (PAINS) from Screening Libraries and for Their Exclusion in Bioassays. J. Med.
Chem.2010, 53, 2719ꢀ2740.
[30]Case, D. A.; Darden, T. A.; Cheatham, T. E. III; Simmerling, C. L.; Wang, J.; Duke, R. E.; Luo,
R.; Walker, R. C.; Zhang, W.; Merz, K. M.; Roberts, B. P.; Hayik, S.; Roitberg, A.; Seabra, G.;
Swails, J.; Götz, A. W.; Kolossváry, I.; Wong, K. F.; Paesani, F.; Vanicek, J.; Wolf, R. M.; Liu, J.;
Wu, X.; Brozell, S. R.; Steinbrecher, T.; Gohlke, H.; Cai, Q.; Ye, X.; Wang, J.; Hsieh, M. J.; Cui, G.;
Roe, D. R.; Mathews, D. H.; Seetin, M. G.; SalomonꢀFerrer, R.; Sagui, C.; Babin, V.; Luchko, T.;
Gusarov, S.; Kovalenko, A.; Kollman, P. A. AMBER 10.0, University of California, San Francisco.
19
ACS Paragon Plus Environment

Journal of Chemical Information and Modeling
Page 20 of 35
2008.
1
2
3
4
5
6
7
8
9
[31]Frisch, M. J.; Trucks, G. W.; Schlegel, H. B.; Scuseria, G. E.; Robb, M. A.; Cheeseman, J. R.;
Montgomery, J. A.; Vreven, T.; Kudin, K. N.; Burant, J. C.; Millam, J. M.; Iyengar, S. S.; Tomasi, J.;
Barone, V.; Mennucci, B.; Cossi, M.; Scalmani, G.; Rega, N.; Petersson, G.A.; Nakatsuji, H.; Hada,
M.; Ehara, M.; Toyota, K.; Fukuda, R.; Hasegawa, J.; Ishida, M.; Nakajima, T.; Honda, Y.; Kitao, O.;
Nakai, O.; Klene, M.; Li, X.; Knox, J. E.; Hratchian, H. P.; Cross, J. B.; Bakken, V.; Adamo, C.;
Jaramillo, J.; Gomperts, R.; Stratmann, R. E.; Yazyev, O.; Austin, A. J.; Cammi, R.; Pomelli, C.;
Ochterski, J. W.; Ayala, P. Y.; Morokuma, K.; Voth, G. A.; Salvador, P.; Dannenberg, J. J.;
Zakrzewski, V. G.; Dapprich, S.; Daniels, A. D.; Strain, M. C.; Farkas, O.; Malick, D. K.; Rabuck, A.
D.; Raghavachari, K.; Foresman, J. B.; Ortiz, J. V.; Cui, Q.; Baboul, A. G.; Clifford, S.; Cioslowski,
J.; Stefanov, B. B.; Liu, G.; Liashenko, A.; Piskorz, P.; Komaromi, I.; Martin, R. L.; Fox, D. J.;
Keith, T.; AlꢀLaham, M. A.; Peng, C. Y.; Nanayakkara, A.; Challacombe, M.; Gill, P. M. W.;
Johnson, B.; Chen, W.; Wong, M. W.; Gonzalez, C.; Pople, J. A. Gaussian 03, Revision E.01;
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
Gaussian, Inc.: Pittsburgh PA, 2004
.
[32]Wang, J.; Wang, W.; Kollman, P. A.; Case, D. A. Automatic atom type and bond type
perception in molecular mechanical calculations. J. Mol. Graph. Model. 2006, 25, 247ꢀ260.
[33]Li, Z.; Cai, Y. H.; Cheng, Y. K.; Lu, X.; Shao, Y. X.; Li, X. S.; Liu, M.; Liu, P. Q.; Luo, H.ꢀB.
Identification of Novel Phosphodiesteraseꢀ4D Inhibitors Prescreened by Molecular
DynamicsꢀAugmented Modeling and Validated by Bioassay. J. Chem. Inf. Model. 2013, 53,
972ꢀ981.
[34]Stote, R. H.; Karplus, M. Zinc binding in proteins and solution: A simple but accurate
nonbonded representation. Proteins 1995, 23, 12ꢀ31.
[35]Li, Z.; Lu, X.; Feng, L. J.; Gu, Y.; Li, X. S.; Wu, Y. U.; Luo, H. B. Molecular dynamicsꢀbased
discovery of novel phosphodiesteraseꢀ9A inhibitors with nonꢀpyrazolopyrimidinone scaffolds. Mol.
Biosyst. 2015, 11, 115ꢀ125.
[36]Essmann, U.; Perera, L.; Berkowitz, M. L.; Darden, T.; Hsing, L.; Pedersen, L. G.A smooth
20
ACS Paragon Plus Environment

Page 21 of 35
Journal of Chemical Information and Modeling
particle mesh Ewald method. J. Chem. Phys. 1995, 103, 8577ꢀ8593.
1
2
3
4
5
[37]Darden, T.; York, D.; Pedersen, L. Particle mesh Ewald: an N.log( N) method for Ewald sums in
large systems. J. Chem. Phys. 1993, 98, 10089ꢀ10092.
6
7
8
9
[38]Miyamoto, S.; Kollman, P. A. SETTLE: an analytical version of the SHAKE and RATTLE
algorithm for rigid water models. J. Comput. Chem. 1992, 13, 952ꢀ962.
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
[39]Ryckaert, J. P.; Ciccotti, G.; Berendsen, H. J. C. Numerical integration of the Cartesian
equations of motion of a system with constraints: molecular dynamics of nꢀalkanes. J. Comput. Phys.
1977, 23, 327ꢀ341.
[40]Wang, J. M.; Hou, T. J.; Xu, X. J. Recent Advances in Free Energy Calculations with a
Combination of Molecular Mechanics and Continuum Models. Curr. ComputꢀAid. Drug 2006, 2,
287ꢀ306.
[41]Xu, L.; Sun, H. Y.; Li, Y. Y.; Wang, J. M.; Hou, T. J. Assessing the Performance of MM/PBSA
and MM/GBSA Methods. 3. The Impact of Force Fields and Ligand Charge Models. J. Phys. Chem.
B 2013, 117, 8408ꢀ8421.
[42]Hou, T. J.; Wang, J. M.; Li, Y. Y.; Wang, W. Assessing the Performance of the Molecular
Mechanics/Poisson Boltzmann Surface Area and Molecular Mechanics/Generalized Born Surface
Area Methods. II. The Accuracy of Ranking Poses Generated From Docking. J. Comput. Chem.
2011, 32, 866ꢀ877.
[43]Shao, Y. X.; Huang, M.; Cui, W.; Feng, L. J.; Wu, Y.; Cai, Y.; Li, Z.; Zhu, X.; Liu, P.; Wan, Y.;
Ke, H.; Luo, H.ꢀB. Discovery of a phosphodiesteraseꢀ9A inhibitor as a potential hypoglycemic agent.
J. Med. Chem. 2014, 57, 10304ꢀ130313.
[44]Su, T.; Zhang, T.; Xie, S.; Yan, J.; Wu, Y.; Li, X.; Huang, L.; Luo, H.ꢀB. Discovery of novel
PDE9 inhibitors capable of inhibiting Aβ aggregation as potential candidates for the treatment of
Alzheimer's disease. Sci. Rep. 2016, 6, 21826.
[45]Shang, N.N.; Shao, Y.X.; Cai, Y.H.; Guan, M.; Huang, M.; Cui, W.; He, L.; Yu, Y.J.; Huang, L.;
Li, Z.; Bu, X. Z.; Ke, H.; Luo, H.ꢀB. Discovery of 3ꢀ(4ꢀhydroxybenzyl)ꢀ1ꢀ(thiophenꢀ2ꢀyl)
chromeno[2,3ꢀc]pyrrolꢀ9(2H)ꢀone as a phosphodiesteraseꢀ5 inhibitor and its complex crystal
21
ACS Paragon Plus Environment

Journal of Chemical Information and Modeling
Page 22 of 35
structure. Biochem. Pharmacol. 2014, 89, 86ꢀ98.
1
2
3
4
5
6
7
8
9
[46]Meng, F.; Hou, J.; Shao, Y. X.; Wu, P. Y.; Huang, M.; Zhu, X.; Cai, Y. H.; Li, Z.; Xu, J.; Liu, P.
Q.; Luo, H.ꢀB.; Wan, Y.; Ke, H. Structureꢀbased discovery of highly selective phosphodiesteraseꢀ9A
inhibitors and implications for inhibitor design. J. Med. Chem. 2012, 55, 8549ꢀ8558.
[47]Xue, X.; Zhang, Y.; Liu, Z.; Song, M.; Xing, Y.; Xiang, Q.; Wang, Z.; Tu, Z.; Zhou, Y.; Ding,
K.; Xu, Y. Discovery of Benzo[cd]indolꢀ2(1H)ꢀones as Potent and Specific BET Bromodomain
Inhibitors: StructureꢀBased Virtual Screening, Optimization, and Biological Evaluation. J. Med.
Chem. 2016, 59, 1565ꢀ1579.
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
[48]Vadivelan, S.; Sinha, B. N.; Irudayam, S. J.; Jagarlapudi, S. A. R. P. Virtual screening studies to
design potent CDK2ꢀcyclin A inhibitors. J. Chem. Inf. Model. 2007, 47, 1526ꢀ1535.
[49]Boppana, K.; Dubey, P. K.; Jagarlapudi, S. A. R. P.; Vadivelan, S.; Rambabu, G. Knowledge
based identification of MAOꢀB selective inhibitors using pharmacophore and structure based virtual
screening models. Eur. J. Med. Chem. 2009, 44, 3584ꢀ3590.
[50]Shih, K.ꢀC.; Lin, C.ꢀY.; Zhou, J.; Chi, H.ꢀC.; Chen, T.ꢀS.; Wang, C.ꢀC.; Tseng, H.ꢀW.; Tang,
C.ꢀY. Development of Novel 3DꢀQSAR Combination Approach for Screening and Optimizing
BꢀRaf Inhibitors in silico. J. Chem. Inf. Model. 2011, 51, 398ꢀ407.
[51]Jain, A. N.Surflex: Fully automatic flexible molecular docking using a molecular
similarityꢀbased search engine. J. Med. Chem. 2003, 46, 499ꢀ511.
[52]Lim, F. P. L.; Dolzhenko, A. V. 1,3,5ꢀTriazineꢀbased analogues of purine: From isosteres to
privileged scaffolds in medicinal chemistry. Eur. J. Med. Chem. 2014, 85, 371ꢀ390.
22
ACS Paragon Plus Environment

Page 23 of 35
Journal of Chemical Information and Modeling
Table 1. Statistical parameters for the pharmacophore screening of the test set in the GH test.
1
2
3
4
5
6
7
8
9
Parameters
Total molecules in dataset (D)
Total number of actives in dataset (A)
Total hits (Ht)
Value
204
30
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
22
Active hits (Ha)
16
% Yield of actives [(Ha/Ht)*100]
% Ratio of actives [(Ha/A)*100]
Enrichment factor (EF) [(Ha*D)/(Ht*A)] a
False negatives [AꢀHa]
72.7
53.3
4.95
14
False positives [HtꢀHa]
6
Goodness of hit test score(GH test score)b
0.66
a EF = (Ha*D)/(Ht*A).
b GH test score = [Ha*(3A+Ht)/(4Ht*A)]*[1ꢀ(HtꢀHa)/(DꢀA)].
23
ACS Paragon Plus Environment

Journal of Chemical Information and Modeling
Page 24 of 35
Table 2. SAR of modified compounds based on the hit compounds AQ-390/10779040 and
1
2
3
AG-690/10776061.
4
5
6
7
8
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
a
PDE2A IC50
Compound
(µM)
AQ-390/10779040
AG-690/10776061
4.62± 0.78
9.78 ± 0.34
> 10
LHB-1
LHB-2
LHB-3
LHB-4
HN
OMe
OMe
> 10
OMe
3.61 ± 0.39
1.83 ± 0.29
LHB-5
1.58 ± 0.29
LHB-6
LHB-7
LHB-8
0.98 ± 0.29
2.91 ± 0.34
0.57 ± 0.03
a EHNA was used as the reference compound with an IC50 of 2.60 ꢀM.
24
ACS Paragon Plus Environment

Page 25 of 35
Journal of Chemical Information and Modeling
Table 3. Predicted free energies (kcal/mol) for binding of AQ-390/10779040 or LHB-8 to human
1
2
3
PDE2 by the MMꢀPBSA method.
4
5
6
Energy Terms
AQ-390/10779040
LHB-8
7
8
(kcal/mol)
9
a
ꢁGele
ꢀ22.61 ± 2.45
ꢀ37.74 ± 2.55
ꢀ4.06 ± 0.12
35.41 ± 2.04
ꢀ29.00 ± 2.53
ꢀ29.38 ± 3.37
ꢀ39.45 ± 2.36
ꢀ4.00 ± 0.11
40.59 ± 2.24
ꢀ32.23 ± 1.98
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
b
ꢁGvdw
c
ꢁGnonpol, sol
d
ꢁGele. solc
e
ꢁGbind, prede
aꢁGele, electrostatic interactions calculated using the MM force field.
bꢁGvdw, van der Waals’ contributions from MM.
cꢁGnonpol, sol, the nonpolar contribution to solvation.
dꢁGele.sol, the polar contribution to solvation.
eꢁGbind, pred = ꢁGele + ꢁGvdw + ꢁGnonpol, sol + ꢁGele, sol, the predicted binding free energies with the
entropic contribution omitted.
25
ACS Paragon Plus Environment

Journal of Chemical Information and Modeling
Page 26 of 35
Figure Captions
1
2
3
4
5
6
7
8
Fig. 1.Workflow of the virtual screening process, combining the pharmacophore model screening,
molecular docking, MD simulations, and bioassay. Nine hits were identified as PDE2 inhibitors in
further bioassay.
9
Fig. 2. The pharmacophore model used in the screening of PDE2A inhibitors (1) and the
pharmacophore model mapped with the ligands in the training set (2). Don|Acc: hydrogen bond
donor or acceptor; Acc2|Don2: projection of the acceptor or the donor; Don|Acc|Hyd: a unified
feature of the Don|Acc or the hydrophobic centroid one; PiR|Aro: Pi ring center or aromatic center;
Hyd|PiR: hydrophobic centroid or Pi ring center; PiN: normal ring center. The exclusion feature is
not represented in the figure.
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
Fig. 3. Chemical structures and affinities of the nine PDE2A inhibitors along with their SPECS IDs.
Fig. 4. The inhibitory profiles of hit compounds AQ-390/10779040 and AG-690/10776061 on
PDE2A.
Fig. 5. Binding modes of four most representative inhibitors within the active site pocket of human
PDE2A (PDB ID: 4HTX, color in cyan, key residue in magenta) after 8 ns MD simulations. The
molecules (carbon in slate, oxygen in red, sulfur in yellow, nitrogen in blue and chlorine in green)
are AQꢀ390/10779040 (1), AGꢀ690/10776061 (2), AGꢀ670/40741585 (3), and AGꢀ690/12244899 (4),
respectively. Hydrogen bonds are represented by dash lines.
Fig. 6. The binding mode of LHB-8 (carbon in slate, oxygen in red, sulfur in yellow, nitrogen in blue
and chlorine in green) within the active site pocket of human PDE2A (PDB ID: 4HTX, color in cyan,
key residue in magenta)after superposition on the binding pose of AQ-390/10779040 (carbon in
green, oxygen in red, sulfur in yellow and nitrogen in blue). Hydrogen bonds are represented by dash
lines.
Fig. 7. Decomposition of the key residue contributions to the binding free energies for the complexes
of PDE2 with AQ-390/10779040 and LHB-8.
Scheme 1. Reagents and conditions: a) Chlorosulfonic acid, rt, 6h; b) Pyridine, rt overnight.
26
ACS Paragon Plus Environment

Page 27 of 35
Journal of Chemical Information and Modeling
1
2
3
4
5
6
7
8
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
Graphic content
743x320mm (96 x 96 DPI)
ACS Paragon Plus Environment

Journal of Chemical Information and Modeling
Page 28 of 35
1
2
3
4
5
6
7
8
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
Fig. 1. Workflow of the virtual screening process, combining the pharmacophore model screening, molecular
docking, MD simulations, and bioassay. Nine hits were identified as PDE2 inhibitors in further bioassay.
339x235mm (96 x 96 DPI)
ACS Paragon Plus Environment

Page 29 of 35
Journal of Chemical Information and Modeling
1
2
3
4
5
6
7
8
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
Fig. 2. The pharmacophore model used in the screening of PDE2A inhibitors (A) and the pharmacophore
model mapped with the ligands in the training set (B). Don|Acc: hydrogen bond donor or acceptor;
Acc2|Don2: projection of the acceptor or the donor; Don|Acc|Hyd: a unified feature of the Don|Acc or the
hydrophobic centroid one; PiR|Aro: Pi ring center or aromatic center; Hyd|PiR: hydrophobic centroid or Pi
ring center; PiN: normal ring center. The excluded volume features are not represented in the figure.
313x143mm (96 x 96 DPI)
ACS Paragon Plus Environment