Novel triazines as potent and selective phosphodiesterase 10A inhibitors

Article abstract of DOI:10.1016/j.bmcl.2012.07.076

The identification of highly potent and orally active triazines for the inhibition of PDE10A is reported. The new analogs exhibit low-nanomolar potency for PDE10A, demonstrate high selectivity against all other members of the PDE family, and show desired drug-like properties. Employing structure-based drug design approaches, we investigated the selectivity of PDE10A inhibitors against other known PDE isoforms, by methodically exploring the various sub-regions of the PDE10A ligand binding pocket. A systematic assessment of the ADME and pharmacokinetic properties of the newly synthesized compounds has led to the design of drug-like candidates with good brain permeability and desirable drug kinetics (t_1/2, bioavailability, clearance). Compound 66 was highly potent for PDE10A (IC₅₀ = 1.4 nM), demonstrated high selectivity (>200×) for the other PDEs, and was efficacious in animal models of psychoses; reversal of MK-801 induced hyperactivity (MED = 0.1 mg/kg) and conditioned avoidance responding (CAR; ID₅₀ = 0.2 mg/kg).

Full text of DOI:10.1016/j.bmcl.2012.07.076

Bioorganic & Medicinal Chemistry Letters 22 (2012) 5876–5884
Contents lists available at SciVerse ScienceDirect
Bioorganic & Medicinal Chemistry Letters
journal homepage: www.elsevier.com/locate/bmcl
Novel triazines as potent and selective phosphodiesterase 10A inhibitors q
Michael S. Malamas a, , , Hans Stange b, Rudolf Schindler b, Hans-Joachim Lankau b, Christian Grunwald b,
⇑
Barbara Langen b, Ute Egerland b, Thorsten Hage b, Yike Ni a, James Erdei a, Kristi Yi Fan a,§, Kevin Parris a,
,
Karen L. Marquis a, Steve Grauer a, Julie Brennan a, Rachel Navarra a, Radka Graf a, Boyd L. Harrison a, Albert
Robichaud a,à, Thomas Kronbach b, Menelas N. Pangalos a,–, Nicholas J. Brandon a,k, Norbert Hoefgen b,
a Pfizer Neuroscience Princeton, 865 Ridge Road, Monmouth Junction, NJ 08852, USA
b Biocrea, Meissner Strasse 191, 01445 Radebeul, Germany
a r t i c l e i n f o
a b s t r a c t
Article history:
Available online 31 July 2012
The identification of highly potent and orally active triazines for the inhibition of PDE10A is reported. The
new analogs exhibit low-nanomolar potency for PDE10A, demonstrate high selectivity against all other
members of the PDE family, and show desired drug-like properties. Employing structure-based drug
design approaches, we investigated the selectivity of PDE10A inhibitors against other known PDE iso-
forms, by methodically exploring the various sub-regions of the PDE10A ligand binding pocket. A system-
atic assessment of the ADME and pharmacokinetic properties of the newly synthesized compounds has
Keywords:
Schizophrenia
PDE10A inhibitors
Triazines
led to the design of drug-like candidates with good brain permeability and desirable drug kinetics (t1/2
,
MK-801
bioavailability, clearance). Compound 66 was highly potent for PDE10A (IC50 = 1.4 nM), demonstrated
high selectivity (>200ꢀ) for the other PDEs, and was efficacious in animal models of psychoses; reversal
of MK-801 induced hyperactivity (MED = 0.1 mg/kg) and conditioned avoidance responding (CAR;
ID50 = 0.2 mg/kg).
Ó 2012 Elsevier Ltd. All rights reserved.
Cyclic nucleotide phosphodiesterases (PDEs) are key regulators
of cellular signal transduction by hydrolyzing the 30, 50-monophos-
phate bond of the intracellular second messengers adenosine 30, 50-
monophosphate (cAMPa) and cyclic guanosine 30, 50-monophos-
phate (cGMP). There are 11 distinct phosphodiesterases, desig-
nated as PDE1 through PDE11.1 PDEs are classified by their
substrate specificity. Some selectively hydrolyze cAMP or cGMP,
while others hydrolyze both substrates.2,3 The PDEs are localized
to specific sub-cellular sites, which allows for fine spatial and tem-
poral control of the levels of the cyclic nucleotides.4,5 This feature is
thought to be an important contributing factor that allows the en-
zyme to influence selective intracellular signaling pathways in re-
sponse to different stimuli, in spite of the ubiquitous intracellular
distribution of the cyclic nucleotides. PDE10A is a dual substrate
PDE.6–9 PDE10A hydrolyzing both cAMP and cGMP, with a higher
affinity for cAMP (Km = 0.05 lM) than for cGMP (Km = 3 l
M).10
PDE10A is primarily a membrane bound enzyme containing a cat-
alytic domain in the C-terminal portion of the protein. Key residues
of the catalytic core form a well-defined hydrophobic clamp region
that positions the planar rings of the nucleotide for interaction
with an absolutely conserved glutamine residue (Gln716 in
PDE10A). Free rotation of this glutamine in PDE10A, and other dual
substrate enzymes, is believed to allow the binding of either cAMP
or cGMP in the substrate pocket.11 PDE10A mRNA is highly ex-
pressed only in brain and testes.7,8 In the brain, both PDE10A
mRNA and protein are specifically enriched in the medium spiny
neurons (MSNs) of the striatum.12 Intriguingly, PDE10A expression
looks to be fairly well conserved across species (mice, rats, dog,
macaque and humans), which could be important for the transla-
tion of any rodent models into human efficacy.13 PDE10A has been
suggested to play a key role in regulating MSN activity and in turn
striatal output as this region integrates dopaminergic and glutama-
Abbreviations: PDE10A, phosphodiesterase 10A; cAMP, cyclic adenosine mono-
phosphate; cGMP, cyclic guanine monophosphate; MSN, medium spiny neuron;
MK-801, dizocilpine maleate; MED, minimal effective dose; PDE, cyclic nucleotide
phosphodiesterase; LMA, locomotor activity; CAR, conditioned avoidance response.
q
Atomic coordinates of the PDE10 crystal structure with compounds 4 (4FCB) and
48 (4FCD) have been deposited in the Protein Data Bank, Research Collaboratory for
Structural Bioinformatics, Rutgers University, New Brunswick, New Jersey.
⇑
Corresponding author.
E-mail address: malamas.michael@gmail.com (M.S. Malamas).
These authors contributed equally to the work.
Present address: Sage Therapeutics, Cambridge, MA.
Present address: Vitae Pharmaceuticals, Fort Washington, PA.
Present address: AstraZeneca, Alderley Park, SK10 4TG, UK.
à
§
–
k
Present address: AstraZeneca Neuroscience iMED, Boston, MA.
Present address: Pfizer Neuroscience, Eastern Point Road, Groton CT.
0960-894X/$ - see front matter Ó 2012 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.bmcl.2012.07.076

M. S. Malamas et al. / Bioorg. Med. Chem. Lett. 22 (2012) 5876–5884
5877
tergic inputs from midbrain and cortical regions, respectively, to
action motoric and cognitive function. Dysfunction in cortical–stri-
atal neurotransmission has been implicated in the pathophysiol-
ogy of schizophrenia and Huntingdon’s disease, thus PDE10A
inhibition has been suggested as a therapeutic strategy for this dis-
ease.14–17
(in-house for 4) versus Tyr693 (reported for 3) is due to the differ-
ent sequence numbering of the two resolved X-ray crystal struc-
tures). Furthermore, the fused ring system of the ligand is
sandwiched between residues Phe719 and Phe686, making
p-
stacking interactions with Phe719 and hydrophobic contacts with
Phe686. Also, the ligand extends toward the invariant Gln716 res-
idue and makes a direct hydrogen bond contact between the pyr-
azine nitrogen and Gln716. The ligand is further stabilized within
the binding pocket with additional water-mediated hydrogen bond
interactions between the imidazole nitrogen and the residues
Tyr514 and Asp664 (not shown). Several quinoline analogs were
prepared and profiled (data not shown); however, despite the im-
proved selectivity of the compounds against PDE2, the added
molecular weight has negatively impeded on the ADME properties
rendering these compounds not suitable as CNS drug candidates.
Next, we turned our attention to other approaches to increase
selectivity against PDE2. Examination of the superimposed X-ray
crystals of PDE2 (PDE2A PDB code 1Z1l) catalytic domain and
PDE10A:1 (R9 = propyl, PDB code 3LXG) (Fig. 3) revealed that the
majority of the residues are identical in the active sites, except
two residues around position 6 (Val668Gln and Ser667Asp). Fur-
thermore, there is a large side chain movement for Ile682Ile close
to position 9. Previous SAR studies18,19 have suggested that various
substitutions (i.e., OMe, CHF2, CN, NH2, and NHSO2Me) at position
6 have only marginally affected the selectivity for PDE2. Also, it is
noteworthy mentioning that smaller groups at position 6 showed
tendency to be more selective against PDE2. At this point, we con-
sidered that a methodic SAR exploration of the various sub-regions
of the PDE10A binding pocket could offer the opportunity to opti-
mize the selectivity of our PDE10A inhibitors regarding PDE2
affinity.
In recent disclosures,18,19 we have described the design of pyr-
idyl- and phenyl-[2,3-e]pyrazines (1, 2; Fig. 1) as potent inhibitors
(IC50 ꢁ 50-500 pM) of PDE10A, with high selectivity (>500ꢀ) for
the other known PDE isoforms, with the exception of PDE2, which
are less selective (ꢁ50–200ꢀ). The reported research, which was
based on structure-based drug design approaches, has provided a
better understanding of the requirements needed to enhance the
ligand’s potency and selectivity. Furthermore, other reports20 on
PDE10A inhibitors (exemplified by compound 3; Fig. 1), which
are highly selective (ꢁ1000ꢀ) for all PDE isoforms, have revealed
that the enhanced PDE2 selectivity of this class of compounds
was attributed to the access of the ligand to a lipophilic pocket
near the entrance of the hydrophobic cleft that defines the com-
mon PDE substrate binding site. A comparison of the known PDE
crystal structures and profiling the inhibitor binding site suggested
that this class of PDE10A ligands occupy this lipophilic pocket in an
orientation that it is unique to PDE10A. The quinoline group of
PDE10A inhibitor 3, confirmed by X-ray crystallography studies,20
extends deep into this lipophilic pocket and makes a hydrogen
bond contact between the quinoline nitrogen and Tyr693 anchor-
ing the inhibitor within the pocket. Only PDE10A has a small gly-
cine residue at the entrance of this lipophilic pocket next to
conserved Gln716, while all other PDEs have larger residues at
the same position blocking the entrance of the ligand to this lipo-
philic ‘selectivity’ pocket. We have used this finding to further in-
crease the selectivity of our PDE10A inhibitors (1, 2) versus PDE2.
Using molecular modeling docking calculations, we were able to
identify the attachment point (position 7) of the pyrazine scaffold
(4; Fig. 1) to append a similar quinoline group to that found in 3,
and generate PDE10A potent inhibitor 4 (IC50 = 2.9 nM) with
marked enhancement in PDE2 selectivity (>300ꢀ). To further sup-
port our molecular modeling calculations, compound 4 was then
crystallized with PDE10A, and as shown in Figures 2a–b (PDB ID
code 4FCB). In agreement with the molecular calculation studies
the quinoline group extends deep toward the PDE10A ‘selectivity’
pocket and makes a direct contact between the quinoline nitrogen
and Tyr683 (the difference in position numbering Tyr683
In this Letter, we investigate the structural requirements critical
for selectivity and disclose new triazine-based PDE10A inhibitors
with an improved selectivity profile against PDE2. These new tri-
azine inhibitors were developed in parallel with the pyridyl- and
phenyl-[2,3-e]pyrazines inhibitors (1, 2), linking together key
structure-activity relationship (SAR) findings and crystallographic
driven structural features.
The compounds needed to delineate the SAR for this study were
prepared according to synthetic Schemes 1–4. The described pyri-
dyl-triazines (Table 1) were prepared according to Schemes 1 and
2. In Scheme 1 and 2-chloro-pyridine 5 was treated with 2-bro-
mo-4-methyl imidazole in the presence of potassium hydroxide
to afford the coupled product 6. Reduction of the nitro group of 6
with stannous chloride produced aniline 7. Cyclization of 7 to pyr-
idyl-triazine 8 was accomplished with sodium nitrite and acetic
acid. Intermediate 8 served as a building block for the preparation
of synthetic libraries. Palladium-catalyzed cross coupling reaction
of 8 with any number of aryl and heteroaryl boronic acids (Suzuki
coupling21) in the presence of Pd(0) or Pd(II) catalyst under a variety
of well known conditions produced the desired products 9 (entries
30–42; Table 1). The 4-methoxy analogs (entries 43–45; Table 1)
were prepared according to Scheme 2. This synthetic protocol is clo-
sely related to that of Scheme 1; however, due to the decreased
reactivity of 10 toward nucleophilic displacement, 4-methyl imid-
azole was used to couple with 10. After separation of the 4:1 regio-
isomeric mixture, the major nitro intermediate 11 was reduced
with iron to the aniline 12. Cyclization of aniline 12 to pyrido-tri-
azine 13 was accomplished with sodium nitrite and acetic acid. Bro-
mination of 13 with N-bromosuccinimide produced bromo-
imidazole 14, which was further functionalized to the 4-methoxy
analogs 43-45 (Table 1) using the Suzuki protocol as before. The
synthetic approaches depicted in Scheme 3 and 4 were utilized
for the preparation of the highly functionalized phenyl-triazine
derivatives contained in Table 2. In Scheme 3, trifluoro-benzene
15 was treated with morpholine and potassium carbonate to
OMe
6
6
N
N
MeO
2
N
N9
MeO
N1
8
N
N
R9
R1
1: R9 = 4-(3-Me-pyridyl)
2: R1 = 4-(3-Me-pyridyl)
PDE10A IC50 = 11 nM
PDE10A IC50 = 0.2 nM
PDE2
IC50 = 38 nM
PDE2
IC50 = 83 nM
N
7
O
N
N
N
N
N
N
N
O
R1
4: R1 = propyl
3
PDE10A IC50 = 2.1 nM
PDE2 IC50 = 3620 nM
PDE10A IC50 = 2.9 nM
PDE2
IC50 = 883 nM
Figure 1. PDE10A inhibitors.

5878
M. S. Malamas et al. / Bioorg. Med. Chem. Lett. 22 (2012) 5876–5884
Figure 2a. Crystal structure (2.1 Å resolution; PDB ID code 4FCB) of PDE10A complexed with 4 highlighting the key hydrogen bond interactions between ligand and protein at
the invariant Gln716 with the pyrazine nitrogen and the water-mediated contacts with Tyr514. The fused ring system of the ligand is sandwiched between residues Phe719,
and Phe686, making p-stacking interactions with Phe719, pi-edge stacking with Phe686. The quinoline nitrogen makes a hydrogen bond interaction with Tyr663.
Figure 2b. Ligplot representation (2.1 Å resolution; PDB ID code 4FCB) of PDE10A complexed with 4 highlighting the hydrophobic and hydrogen bond interactions between
the ligand and the amino acid residues involved in the PDE10A binding pocket.
produce 16. Furthermore, aniline 17 was oxidized with sodium per-
borate to afford nitro benzene 18. Intermediates 15, 16 and18 were
coupled with 2-bromo-4-methyl imidazole, as described in
Scheme 1, to furnish products 19. Treatment of 19 with methanol
and potassium hydroxide afforded the methoxy analogs 20. Palla-
dium-catalyzed cross coupling reaction of 20 with any number of
aryl and heteroaryl boronic acids gave 21. Reduction of the nitro
intermediates 21 with ammonium formate and palladium on car-
bon afforded anilines 22, which were further cyclized to phenyl-tri-
azines 23 as before. The fluoro-methoxy analogs (entries 57–65, 70,
71 Table 2) were prepared according to Scheme 4. Phenol 24 was ni-
trated with fuming nitric acid to produce a mixture (1.5:1) of regioi-
somers 25 and 26, which were separated and used as building
blocks to generate the desired products 29 as described above in
Schemes 1 and 2. The difluoromethoxy analog (entry 69; Table 2)
was prepared from nitro-phenol 24 upon alkylation with sodium
chlorodifluoro acetate (forming difluoromethoxy group), and then
applying the same synthetic protocols as described above in
Schemes 1 and 2. Detailed experimental protocols of the com-
pounds described in this Letter have been reported.22

M. S. Malamas et al. / Bioorg. Med. Chem. Lett. 22 (2012) 5876–5884
5879
Figure 3. Superimposed X-ray crystals of PDE2A (grey; PDB ID code 1Z1 l) catalytic domain and PDE10A:1 (R = propyl) co-crystal (green; PDB ID code 3LXG) reveal that the
majority of the residues are identical in the active site, except two residues (yellow circle) Val668Gln and Ser667Asp. There is a large side chain movement for Ile682Ile
(magenta circle).
NH2
NO2
b
NO2
Cl
Me
a
MeO
N
N
Me
MeO
N
N
N
MeO
c
N
N
Br
7
Br
d
6
5
N
N
N
N
Me
MeO
N
N
Me
MeO
N
N
N
N
Ar
9
8
Br
Scheme 1. Reagents: (a) 2-Br-4-Me-imidazole, KOH, DMF; (b) SnCl2, EtOH; (c) NaNO2, AcOH; (d) Ar-B(OH)2, Pd(PPh3)4.
OMe
OMe
OMe
NO2
N
NO2
Cl
NH2
N
a
b
Me
N
N
Me
N
4:1 mixture
N
11
12
10
N
OMe
OMe
N
N
d
N
c
N
Me
N
N
Me
N
N
N
14
N
13
Br
Scheme 2. Reagents: (a) 4-Me-imidazole, KOH, DMF; (b) Fe/AcOH, EtOH; (c) NaNO2, AcOH; (d) NBS, AcCN.
Inhibitory potencies of our compounds were tested using
tivity against the PDE isoforms PDE2-PDE6, which have the poten-
tial for off-target liability. For matter of brevity, we are presenting
our SAR evolution only in respect to PDE10A and those PDEs. How-
in vitro inhibition of human recombinant PDE10A catalyzed cAMP
hydrolysis.19 We have routinely evaluated all compounds for selec-

5880
M. S. Malamas et al. / Bioorg. Med. Chem. Lett. 22 (2012) 5876–5884
F
NO2
F
R
1 = H, F
c
OMe
F
R1
NO2
N
15
NO2
N
d
c
a
R
1 = F
F
Me
R
Me
R
NO2
F
20
N
19
N
Br
Br
N
R
= H, F, CF3, morpholine
e
O
OMe
c
16
F
OMe
N
N
N
NO2
N
F
g
NO2
F
NH2
F
Me
R
b
Me
f
R
23
N
21
F3C
21
: NO2
22
: NH2
N
Ar
F3C
Ar
18
17
Scheme 3. Reagents: (a) morpholine, K2CO3, DMF; (b) NaBO3, H2O, AcOH; (c) 2-Br, 4-Me-imidazole, K2CO3, DMF; (d) KOH, MeOH; (e) Ar-B(OH)2, Pd(PPh3)4; (f) HCO2NH4, Pd-
C, MeOH; (g) NaNO2, AcOH.
F
R1
F
F
NO2
F
NO2
F
b
a
+
HO
F
R2
separate
HO
HO
F
regioisomers
27
NO2
R1
25
26
R1 = F or OMe
R2 = OMe or F
24
R1
N
N
N
NO2
d-f
c
Me
R2
N
R2
N
similar to schemes 1, 2 29
N
R
28
Br
Scheme 4. Reagents: (a) HNO3; (b) MeI, K2CO3; (c) 2-Br, 4-Me-imidazole, K2CO3, DMF; (d) Ar-B(OH)2, Pd(PPh3)4; (e) Fe, AcOH; (f) NaNO2, AcOH.
ever, if unexpected findings are emerging during the screening
against the other PDEs (1, 7–9, 11), we are presenting these data
appropriately. The results are compiled in Tables 1 and 2. The ini-
tial lead 1 (R9 = propyl, Fig. 1) was optimized19 by using structure-
based drug design techniques, primarily exploring the C9 and C6
positions of the pyridyl- (1) and phenyl- (2) pyrazines, respec-
tively. The reported SAR exploration has resulted in >100-fold in-
crease of the ligand’s binding affinity for PDE10A, with enhanced
selectivity >1000-fold for all other PDEs, with the exception of
PDE2, where only a small subset of compounds showed selectivity
ꢁ100-fold. The SAR studies have indicated that substitutions at po-
sition 6 (phenyl-pyrazines; 2) have contributed the most to the
selectivity for PDE2. Other substitution positions that have also af-
fected the selectivity versus PDE2 were positions 6 and 9 (pyridyl-
pyrazines; 1). As mentioned above, examination of the superim-
posed X-ray crystals of PDE2A (PDE2A PDB ID code 1Z1l) catalytic
domain and PDE10A: 1 (R9 = propyl; PDB ID code 3LXG; Fig. 3)
have revealed that the majority of the residues are identical in
the active site, except two residues around position 6 (Val668Gln
and Ser667Asp), with a large side chain movement for Ile682Ile
close to position 9. In this Letter, we focus on analogs that lack a
substitution at position 6 of the pyridyl-pyrazine (1) scaffold. It
is notable, that the C6 des-methyl pyrazine analogs are rather
weakly selective against PDE2 (data not shown). We have discov-
ered that the triazine nucleus (Table 1) offered the opportunity
to further investigate the PDE10A inhibitor potency and selectivity.
In Table 1, we have aggregated selected analogs examining the ef-
fect of substitution at positions 2, 4 and 9 of the pyridyl-triazine
scaffold (Table 1). Pyridyl analog 30 was weakly active, while the
regioisomeric 31 was about 8–10ꢀ more active in both PDE10A
and PDE2. The similarly substituted methoxy analog 32 was simi-
lar in potency to that of 31 (32 vs 31). Thiazole 33 was about 2ꢀ
weaker in PDE2 to that of 31. The methyl-substituted phenyl ana-
logs 34 and 35 were similar in potency to that of 31, while the hal-
ogen substituted-phenyl analogs 36–38 were about fivefold more
potent in PDE10A. Interestingly, the dichloro analog 39, while it
did not show a change in PDE10A potency, it was sevenfold more
potent in PDE2 with an IC50 value of 2 nM (39 vs 34). The fluoro-
methoxy phenyl analog 40 was potent in both PDE10A and PDE2.
All phenyl analogs (34–40) were weakly active in the PDE3–PDE6
isoforms. An increase of the substitution size OEt versus OMe in
analog 41 resulted in about 10-fold loss of potency for both
PDE10A and PDE2 isoforms (41 vs 40). More pronounced was the
potency loss (ꢁ200-fold) for the bulkier isopropyl analog 42 (42
vs 40). Moving the methoxy group from position 2 to position 4
(analog 43) resulted in 10-fold potency increase for PDE10A with
15-fold concomitant potency decrease for PDE2 (43 vs 30). Fur-
thermore, the similarly C4 methoxy substituted pyridyl analogs
44 and 45 were found to be the most potent and selective ligands
of the pyridyl-triazine series of compounds, with IC50 values for
PDE10A ꢁ10 nM and >100–200 fold selectivity in the other
PDE2-PDE6 isoforms. Compound 45 was further evaluated in PDE
isoforms (PDE1, PDE7–PDE9, PDE11) and it was found to be weakly
active (>500–1000 fold selective; data not shown).

M. S. Malamas et al. / Bioorg. Med. Chem. Lett. 22 (2012) 5876–5884
5881
Table 1
Pyridyl-triazine analogs
R4
N
N
R2
N
N
N
R9
R2
R4
R9
IC50 (nM)
PDE4
a
Compd
PDE10A
PDE2
PDE3
PDE5
PDE6
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
OMe
OMe
OMe
OMe
OMe
OMe
OMe
OMe
OMe
OMe
OMe
OMe
OMe
H
H
H
H
H
H
H
H
H
H
H
H
H
4-Me-3-pyridyl
5-Me-3-pyridyl
5-OMe-3-pyridyl
2,4-Me, Me-thiazole
2-Me-Ph
4-F, 2-Me-Ph
2-Cl-Ph
2,4-Cl, Cl-Ph
388
142
14.4
14
—
—
—
—
>5000
7260
1240
>1000
>1000
750
1820
—
—
—
—
—
—
—
—
—
3410
34100
1680
>1000
>1000
1150
702
—
—
—
—
—
1570
1150
435
>1000
462
829
734
—
51.5
91.3
64.5
47.6
45.9
13.1
15
13.1
53.1
9.83
114
23
15.8
13.2
4.3
>10,000
>10,000
>10,000
>1000
>1000
4410
4820
—
—
—
3680
6960
15.1
11.8
2.06
2.99
28.6
101.6
1950
2055
974
2-Cl, 4-F-Ph
2,5-Cl, Cl-Ph
2-F, 5-OMe-Ph
2-F, 5-OEt-Ph
2-F, 5-O(CHMe2)-Ph
4-Me-3-pyridyl
3-Me-4-pyridyl
2-Me-3-pyridyl
H
2430
42.4
10.1
12.4
—
—
—
—
—
—
OMe
OMe
OMe
H
H
>10,000
>10,000
>10,000
>5000
>10,000
>5000
a
Means of IC50; n = 4.
Table 2
Phenyl-triazine analogss
R6
N
N
N
R8
N
R1
R1
R6
R8
IC50 (nM)
PDE4
a
Compd
PDE10A
PDE2
PDE3
PDE5
PDE6
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
2-Me-Ph
2-Cl-Ph
2-Cl-Ph
2-Me-Ph
OMe
OMe
OMe
OMe
OMe
OMe
OMe
OMe
OMe
OMe
OMe
OMe
OMe
OMe
OMe
OMe
OMe
OMe
OMe
OMe
OMe
OMe
OMe
OMe
F
H
H
OMe
OMe
OMe
OMe
OMe
OMe
morpholine
morpholine
morpholine
F
F
F
F
F
3.3
2.41
0.28
5.13
3.35
1.48
4.16
2.06
1.53
0.85
3.99
6.53
9.11
3.55
25.4
7.2
7.84
13.1
1.65
11.7
1.48
5.11
1.85
7.0
172
103
30.3
95.9
522
196
775
589
110
50.4
4.37
894
1998
1150
2375.5
1220
1270
1500
269
>10,000
>10,000
>5000
>10000
>1000
>10,000
>10,000
>10,000
>10,000
>10,000
>10,000
>10,000
>1000
176
70.8
1910
>10000
>1000
>10,000
>10,000
>10,000
2920
3670
3060
436
1110
>1000
1430
5070
4950
651
603
1980
1060
73.15
349
>1000
493
1822
1240
693
276
2-CF3-Ph
2-Me-3-pyridyl
4-Me-3-pyridyl
3-Me-4-pyridyl
2-Me-Ph
2-Cl-Ph
3-Me-4-pyridyl
2-OMe-Ph
2680
>10,000
>10,000
>1000
5380
3120
>10,000
>1000
>1000
>10,000
>10,000
>5000
>10,000
3600
4760
2290
>10,000
2130
4020
>10,000
>1000
>1000
>10,000
>10,000
2830
>5000
3270
3880
1790
5250
1260
909
3-OMe-Ph
3-F, 2-Me-Ph
4-Me-3-pyridyl
3-Me-4-pyridyl
2-Me-3-pyridyl
3,5-Me, Me-N-Me-pyrazole
3-Me-2-thienyl
2,4-Me, Me-thiazole
3-Me-4-pyridyl
4-Me-3-pyridyl
3-F, 2-Me-Ph
>1000
>1000
>10,000
>10,000
>10,000
>10,000
>10,000
>10,000
>10,000
>10,000
>5000
>10,000
>10,000
>10,000
>10,000
7720
F
F
F
F
3420
346.5
864
722
944
2460
CF3
CF3
CF3
OCF2
OMe
OMe
>10,000
>10,000
>5000
>10,000
>1000
3-Me-4-pyridyl
2-OMe-Ph
3-Me-2-thienyl
>10,000
>1000
2250
8.33
9.43
3390
35.9
>1000
2780
>1000
1230
F
3670
a
Means of IC50; n = 4.
Next, we turned our attention to the evaluation of phenyl-fused
triazines. In Table 2, we have aggregated selected analogs examin-
ing substitutions at positions 1, 6 and 8. Mostly, we have retained
the methoxy group at position 6 of the phenyl-triazine scaffold,
since it had played a key role to the potency and selectivity of
the pyridyl-triazine PDE10A inhibitors, as described above. Tolyl

5882
M. S. Malamas et al. / Bioorg. Med. Chem. Lett. 22 (2012) 5876–5884
Figure 4a. Crystal structure (2.0 Å resolution; PDB ID code 4FCD) of PDE10A complexed with 48 highlighting the key hydrogen bond interactions between ligand and protein
at the invariant Gln716 with both the triazine nitrogens and the 6-OMe group, and the water-mediated contacts with Asp664 and Thr623. The fused ring system of the ligand
is sandwiched between residues Phe719, and Phe686, making p-stacking interactions with Phe719, p-edge stacking with Phe686. The C9 position 2-Cl-Ph rotates off the
fused-tricyclic ring system about 90 degrees.
Figure 4b. Ligplot representation (2.0 Å resolution; PDB ID code 4FCD) of PDE10A complexed with 48 highlighting the hydrophobic and hydrogen bond interactions between
the ligand and the amino acid residues involved in the PDE10A binding pocket.
analog 46 was a potent PDE10A inhibitor (IC50 = 3.3 nM) and about
60-fold selective against PDE2. Similar potency was also observed
for the chloro-phenyl analog 47. Nevertheless, both of these ana-
logs inhibited PDE4 with IC50 values of 176 and 71 nM, respec-
tively. Introduction of a methoxy group at position 8 (entry 48)
resulted in about 10-fold potency increase for PDE10A (48 vs 47),
and a marked decrease (200ꢀ) of the PDE4 potency. Unexpectedly,
compound 48 showed good potency for the PDE6 isoform
(IC50 = 73 nM). In contrast, the analogous trifluoromethyl-phenyl
analog 50 was about 10-fold weaker in PDE10A (50 vs 48), but
inactive in PDE6, and the other tested PDEs. Dimethoxy analogs
(entries 51–53) with various pyridyl groups at position 1 have
showed similar PDE10A potency to that of 50, and better selectivity
against PDE2. Amongst them, compound 53 was the most selective
(ꢁ300ꢀ) against PDE2. Substitution at position 8 with a morpho-
line moiety (entries 54–56) resulted in similar PDE10A potency
to that of the C8 methoxy analogs but with significant decrease
in selectivity against PDE2 (56 vs 53). The C8 substituted morpho-
line 56 was equipotent to both PDE10A and PDE2 isoforms, while
the analogous C8 substituted methoxy analog 53 was about
>280-fold selective against PDE2. Furthermore, introduction of a
fluorine group at position 8 (entries 57-65) exhibited similar
PDE10A potency to that of the C8 methoxy analogs, while either re-
tained or even enhanced selectivity against all tested PDE isoforms.
Introduction of the more stable trifluoromethyl group (metaboli-
cally more stable than methoxy) at position 8 resulted in about

M. S. Malamas et al. / Bioorg. Med. Chem. Lett. 22 (2012) 5876–5884
5883
Table 3
(Figs. 4a–b; PDB ID code 4FCD) revealed that the C9 2-chloro-
phenyl projects toward the opened space of the PDE10A binding
pocket, in agreement with our molecular modeling calculations,
and it is positioned practically in a vertical orientation relatively
to the fused-ring system. Furthermore, the fused ring system of
the ligand is sandwiched between residues Phe719 and Phe686,
Pharmacokinetic profile of selected compounds
Compd
51
52
53
66
Species
Dose; mg/kg
%F
Rats
iv 2/po 10
100
2.0
4
0.4
6590/
21,306
Rats
iv 2/po 10
100
0.8
14
0.5
3001/
8786
Rats
iv 2/po 10
100
0.9
13
0.4
2781/
18,681
Rats
iv 2/po 1
53
4.5
5
1.3
622/
1605
t1/2 (h)
making
p-stacking interactions with Phe719 and hydrophobic
CL (mL/min/kg)
Vss (L/kg)
AUCa brain/AUC
contacts with Phe686. Also, the ligand extends toward the invari-
ant Gln716 residue and makes two direct hydrogen bond interac-
tions between the C6 methoxy group and the position 5 nitrogen
of the triazine nucleus, as well as a water mediated contact be-
tween the position 4 nitrogen of the triazine and Gln716. The li-
gand is further stabilized within the binding pocket with
additional water-mediated hydrogen bonding interactions be-
tween the imidazole nitrogen and the residues Tyr514 and
Asp664.
The majority of the analogs shown in Tables 1 and 2, have dem-
onstrated good microsomal stability (t1/2 > 30 min) in both human
and rodent microsomes, have desired TPSA values (ꢁ50–65 Å2),
and good brain permeability (40%). Evaluation of selected PDE10
inhibitors in the MDR1-MDCK permeability assay revealed that
the compounds showed high level of permeability [apical to baso-
lateral (A–B) Papp and basolateral to apical (B–A) Papp] and lacked
efflux properties (data not shown).
plasma
a
AUC h*ng/mL
Table 4
Reversal of MK-801 induced hyperactivity and stereotyped sniffing
Compd
MEDa
% Decrease in
MEDa
%Decrease
IC50
hyperactivity
(mg/kg po)
hyperactivityb sniffing
in sniffingb (nM)c
(mg/kg
po)
44
48
51
52
53
66
2.5
0.5
0.25
1
0.5
0.1
55.4
70.3
90.2
79.9
71.7
40.5
16.53
2.5
0.5
0.25
1
0.5
0.1
20
34.1
55.7
71.3
69.8
57.4
18.4
16.28
10.1
0.28
1.48
4.16
2.06
1.48
Clozapine 20
n.i.d
The evaluation of the pharmacokinetic properties of selected
compounds in rats after oral administration is summarized in Ta-
ble 3. All compounds showed low to moderate clearance, low vol-
ume of distribution, good oral bioavailability, moderate
elimination terminal half-life (t1/2) and good brain exposure.
Next, we evaluated this class of PDE10A inhibitors in models
thought to be predictive of antipsychotic activity in humans.
Reversal of MK-801 induced hyperactivity in Rats: A repre-
sentative subset of high affinity PDE10A inhibitors was selected
for investigation of their antipsychotic-like potential in vivo. To en-
able the characterization of the antipsychotic activity we utilized a
MK-801 induced hyperactivity reversal model where we have pro-
filed the PDE10A mechanism previously.23 MK-801 (0.1 mg/kg ip)
significantly increased horizontal activity (t test: p <0.001) and in-
duced stereotyped sniffing (t test: p <0.001) in female rats. These
behaviors were significantly reversed by the positive control which
in this case was the atypical antipsychotic clozapine. All evaluated
PDE10A inhibitors also reversed hyperactivity and stereotyped
sniffing in a dose dependent and significant manner. The minimal
effective dose (MED) of the novel PDE10A inhibitors can be seen in
Table 4. Compound 66 was the most active analog with a MED va-
lue of 0.1 mg/kg.
a
Minimal effective dose that significantly reduced MK-801-induced hyperac-
tivity or stereotyped sniffing, respectively, (p <0.05).
b
p <0.001, n = 6.
Inhibition of PDE10 activity from Tables 1 & 2.
ni: no inhibition at PDE10.
c
d
threefold potency increase for PDE10A (66 vs 61, 67 vs 60) with a
modest twofold selectivity increase against PDE2. The C8 difluoro-
methoxy analog 69 was about 4-fold less potent (69 vs 66). Lastly,
we have introduced a fluorine group at position 6 (entries 70, 71).
The 2-methoxy phenyl analog 70 showed the highest selectivity
against PDE2 (>400-fold), while the thienyl analog 71 was only
threefold selective against PDE2 (71 vs 64). From the SAR findings
described above, one can conclude that substitutions directed at
the various sub-regions of the PDE10A ligand binding pocket influ-
ence collectively at various degree the selectivity of the compound
against other known PDE isoforms, but most prominently against
PDE2.
In support of our molecular modeling studies and design of
PDE10A ligands, we have crystallized compound 48 with human
PDE10A protein. Examination of the 48:PDE10A crystal structure
Figure 5. Compound 66 was evaluated in the conditioned avoidance response in male Sprague–Dawley rats using a 60-min pretreatment interval and an oral route of
administration. Data are expressed as average (means S.E.M.; A) avoidance, escape, and response failures observed over 50 trials (n = 8 animals/group). ⁄Statistical difference
in avoidance and escape response from vehicle-treated controls (p = 0.05). Nonlinear regression analysis (B) calculated an ID50 equivalent to 0.204+/ꢂ0.033 mg/kg; p = 0.05.

5884
M. S. Malamas et al. / Bioorg. Med. Chem. Lett. 22 (2012) 5876–5884
4. Houslay, M. D.; Schafer, P.; Zhang, K. Y. Drug Discov. Today 2005, 10, 1503.
Disruption of Conditioned Avoidance Responding (CAR) in
5. Bender, A. T.; Beavo, J. A. Pharmacol. Rev. 2006, 58, 488.
6. Loughney, K.; Snyder, P. B.; Uher, L.; Rosman, G. J.; Ferguson, K.; Florio, V. A.
Gene 1999, 234, 109.
7. Fujishige, K.; Kotera, J.; Omori, K. Eur. J. Biochem. 1999, 266, 1118.
8. Fujishige, K.; Kotera, J.; Michibata, H.; Yuasa, K.; Takebayashi, S.; Okumura, K.;
Omori, K. J. Biol. Chem. 1999, 274, 18438.
9. Soderling, S.; Bayuga, S. J.; Beavo, J. A. Proc. Natl. Acad. Sci. U.S.A. 1999, 96, 7071.
10. Soderling, S.; Beavo, J. A. Curr. Opin. Cell Biol. 1999, 12, 174.
11. Zhang, K.; Card, G.; Suzuki, Y.; Artis, R.; Fong, D.; Gillette, S.; Hsieh, D.; Neiman,
J.; West, B.; Zhang, C.; Milburn, M.; Kim, S.-H.; Schlessinger, J.; Bollag, G. Mol.
Cell 2004, 15, 279.
Rats: Disruption of the ‘avoidance’ response in CAR is another pre-
clinical model predictive of antipsychotic activity.24,25 In the rat
CAR model (Fig. 5), 66 decreased avoidance responding with a sig-
nificant treatment effect at 0.17 mg/kg. Data are expressed as aver-
age (means S.E.M.) avoidance, escape, and response failures
observed over 50 trials (n = 8 animals/group). Nonlinear regression
analysis calculated an ID50 equivalent to 0.204+/ꢂ0.033 mg/kg;
p = 0.05.
In this Letter, we have described the discovery of new pyridyl-
and phenyl-triazine PDE10A inhibitors, further extending the SAR
studies of previously disclosed19 pyridyl- and phenyl- pyrazine
PDE10A inhibitors. Applications of structure-based drug design
techniques and molecular modeling have led to the identification
of low-nanomolar PDE10A inhibitors with excellent selectivity
against the other members of the PDE family. Particular attention
was directed to the selectivity property of the PDE10A inhibitors
against the PDE2 isoform, by methodically exploring the various
sub-regions of the PDE10A ligand binding pocket to maximize li-
gand selectivity. A systematic assessment of the ADME and phar-
macokinetic properties of the newly synthesized compounds has
led to the design of drug-like candidates with good brain perme-
ability and desirable drug kinetics (t1/2, bioavailability, clearance).
Compound 66 was highly potent for PDE10A (IC50 = 1.4 nM), dem-
onstrated high selectivity (>250ꢀ) for the other PDEs (data not
shown for PDE1, PDE7–PDE9, PDE11 isoforms in Table 2), and
was efficacious in animal models of psychoses; reversal of MK-
801 induced hyperactivity (MED = 0.1 mg/kg) and conditioned
avoidance responding (CAR; ID50 = 0.204+/ꢂ0.033 mg/kg).
12. Seeger, T. F.; Bartlett, B.; Coskran, T. M.; Culp, J. S.; James, L. C.; Krull, D. L.;
Lanfear, J.; Ryan, A. M.; Schmidt, C. J.; Strick, C. A.; Varghese, A. H.; Williams, R.
D.; Wylie, P. G.; Menniti, F. S. Brain Res. 2003, 985, 113.
13. Coskran, T.; Morton, D.; Menniti, F.; Adamowicz, W.; Kleiman, R.; Ryan, A.;
Strick, C.; Schmidt, C.; Stephenson, D. J. Histochem. Cytochem. 2006, 54, 1205.
14. Kaiya, H. Prostaglandins Leukot. Essent. Fatty Acids 1992, 46, 33.
15. Muly, E. C.; Smith, Y.; Allen, P.; Greengard, P. Psychopharmacol. Bull 2002, 36,
92.
16. Garver, D. L.; Carl Johnson, C.; Kanter, D. R. Life Sci. 1987, 1982, 31.
17. Giampa, C.; Laurenti, D.; Anzilotti, S.; Bernardi, G.; Menniti, F.; Fusco, F. PLoS
One 2010, 5, e13417.
18. Hoefgen, N.; Stange, H.; Rudolf Schindler, R.; Lankau, H.-J.; Grunwald, C.;
Langen, B.; Egerland, U.; Tremmel, P.; Pangalos, M. N.; Marquis, K. L.; Hage, T.;
Harrison, B. L.; Malamas, M. S.; Brandon, N. J.; Kronbach, T. J. Med. Chem. 2010,
53, 4399.
19. Malamas, M. S.; Ni, Y.; Erdei, J.; Stange, H.; Schindler, R.; Hans-Joachim Lankau,
H.-J.; Grunwald, C.; Fan, K.; Parris, K.; Langen, B.; Egerland, U.; Hage, T.;
Marquis, K.; Harrison, B.; Robichaud, A.; Kronbach, T.; Pangalos, M.; Brandon,
N.; Hofgen, N. J. Med. Chem. 2011, 54, 7621.
20. Verhoest, P.; Chapin, D.; Corman, M.; Fonseca, K.; Harms, J.; Hou, X.; Marr, E.;
Menniti, F.; Nelson, F.; O’Connor, R.; Pandit, J.; Proulx-LaFrance, C.; Schmidt, A.;
Schmidt, C.; Suiciak, J.; Liras, S. J. Med. Chem. 2009, 52, 5188.
21. Miyaura, N.; Suzuki, A. Chem. Rev. 1995, 95, 2457.
22. Stange, H.; Langen, B.; Egerland, U.; Hoefgen, N.; Priebs, M.; Malamas, M. S.;
Edrei, J.; Ni, Y. U.S. Patents 0120762, 0120763, 2010.
23. Grauer, S.; Pulito, V.; Navarra, R.; Kelly, M.; Kelley, C.; Graf, R.; Langen, B.;
Logue, S.; Brennan, J.; Jiang, L.; Charych, E.; Egerland, U.; Liu, F.; Marquis, K.;
Malamas, M. S.; Hage, T.; Comery, T.; Brandon, N. J. Pharm. Exp. Ther. 2009, 331,
574.
References and notes
24. Li, M.; Fletcher, P. J.; Kapur, S. Neuropsychopharmacology 2007, 32, 263.
25. Smith, A.; Li, M.; Becker, S.; Kapur, S. Neuropsychopharmacology 2004, 29, 1040.
1. Manallack, D. T.; Hughes, R. A.; Thompson, P. E. J. Med. Chem. 2005, 48, 3449.
2. Odingo, J. O. Exp. Opin. Ther. Patents 2005, 15, 773.
3. Rotella, D. Nat. Rev. Drug Discov. 2002, 1, 674.