Discovery of potent selective bioavailable phosphodiesterase 2 (PDE2) inhibitors active in an osteoarthritis pain model. Part II: Optimization studies and demonstration of in vivo efficacy

Article abstract of DOI:10.1016/j.bmcl.2013.03.082

Selective phosphodiesterase 2 (PDE2) inhibitors are shown to have efficacy in a rat model of osteoarthritis (OA) pain. We identified potent, selective PDE2 inhibitors by optimizing residual PDE2 activity in a series of phosphodiesterase 4 (PDE4) inhibitors, while minimizing PDE4 inhibitory activity. These newly designed PDE2 inhibitors bind to the PDE2 enzyme in a cGMP-like binding mode orthogonal to the cAMP-like binding mode found in PDE4. Extensive structure activity relationship studies ultimately led to identification of pyrazolodiazepinone, 22, which was >1000-fold selective for PDE2 over recombinant, full length PDEs 1B, 3A, 3B, 4A, 4B, 4C, 7A, 7B, 8A, 8B, 9, 10 and 11. Compound 22 also retained excellent PDE2 selectivity (241-fold to 419-fold) over the remaining recombinant, full length PDEs, 1A, 4D, 5, and 6. Compound 22 exhibited good pharmacokinetic properties and excellent oral bioavailability (F = 78%, rat). In an in vivo rat model of OA pain, compound 22 had significant analgesic activity 1 and 3 h after a single, 10 mg/kg, subcutaneous dose.

Full text of DOI:10.1016/j.bmcl.2013.03.082

Bioorganic & Medicinal Chemistry Letters 23 (2013) 3443–3447
Contents lists available at SciVerse ScienceDirect
Bioorganic & Medicinal Chemistry Letters
journal homepage: www.elsevier.com/locate/bmcl
Discovery of potent selective bioavailable phosphodiesterase
2 (PDE2) inhibitors active in an osteoarthritis pain model.
Part II: Optimization studies and demonstration of in vivo efficacy
b
a
a
d
Mark S. Plummer a, , Joseph Cornicelli , Howard Roark , Donald J. Skalitzky , Charles J. Stankovic ,
Susan Bove b, Jayvardhan Pandit c, Annise Goodman a, James Hicks a, Aurash Shahripour a, David Beidler b,
Xiao Kang Lu b, Brian Sanchez b, Christopher Whitehead a, Ron Sarver e, Timothy Braden b,
Richard Gowan b, Xi Qiang Shen b, Katherine Welch b, Adam Ogden d, Nalini Sadagopan d, Heidi Baum d,
Howard Miller d, Craig Banotai e, Cindy Spessard e, Sandra Lightle e
⇑
a Department of Chemistry, Pfizer Global Research and Development, Ann Arbor, MI 48105, USA
b Inflammation Pharmacology, Pfizer Global Research and Development, Ann Arbor, MI 48105, USA
c Discovery Technology, Pfizer Global Research and Development, Groton, CT 06340, USA
d Pharmacodynamics and Drug Metabolism, Pfizer Global Research and Development, Ann Arbor, MI 48105, USA
e Discovery Technology, Pfizer Global Research and Development, Ann Arbor, MI 48105, USA
a r t i c l e i n f o
a b s t r a c t
Article history:
Available online 29 March 2013
Selective phosphodiesterase 2 (PDE2) inhibitors are shown to have efficacy in a rat model of osteoarthri-
tis (OA) pain. We identified potent, selective PDE2 inhibitors by optimizing residual PDE2 activity in a
series of phosphodiesterase 4 (PDE4) inhibitors, while minimizing PDE4 inhibitory activity. These newly
designed PDE2 inhibitors bind to the PDE2 enzyme in a cGMP-like binding mode orthogonal to the cAMP-
like binding mode found in PDE4. Extensive structure activity relationship studies ultimately led to iden-
tification of pyrazolodiazepinone, 22, which was >1000-fold selective for PDE2 over recombinant, full
length PDEs 1B, 3A, 3B, 4A, 4B, 4C, 7A, 7B, 8A, 8B, 9, 10 and 11. Compound 22 also retained excellent
PDE2 selectivity (241-fold to 419-fold) over the remaining recombinant, full length PDEs, 1A, 4D, 5,
and 6. Compound 22 exhibited good pharmacokinetic properties and excellent oral bioavailability
(F = 78%, rat). In an in vivo rat model of OA pain, compound 22 had significant analgesic activity 1 and
3 h after a single, 10 mg/kg, subcutaneous dose.
Keywords:
Phosphodiesterase 2 (PDE2) inhibitors
Structure based drug design
Osteoarthritis (OA) pain
Ó 2013 Elsevier Ltd. All rights reserved.
In our previous letter we reported that pyrazolodiazepinone
inhibitors selective for PDE2 were identified from structure activity
studies. We were interested in pursuing peripherally restricted
PDE2 inhibitors as a potential treatment for OA pain. The impetus
for our interest originated in an indications discovery effort to
identify novel analgesic mechanisms, where we observed antinoci-
ceptive activity when employing selective PDE2 inhibitors. The role
of PDE2 in the production and modulation of pain is inferred from a
number of observations including that PDE2 is highly expressed in
the brain, spinal cord, dorsal root ganglia (DRG) as well as several
non-pain related tissues.1 Additionally the non-hydrolyzable cGMP
analog, 8-bromo-cGMP, has been shown to reduce nociceptive
behavior in rats at low doses in formalin-induced inflammatory
pain.2 Selective inhibition of PDE2 elevates cGMP in cells or subcel-
lular compartments where it is located.3 Collectively this suggests
raising cGMP levels through PDE2 inhibition may provide a poten-
tial treatment for OA pain. We were particularly interested in
peripheral PDE2 inhibition in the DRG as a mechanism to avoid po-
tential but unknown side effects of a centrally acting agent.
The design premise was that diazepinones could be made PDE2
selective by virtue of a cGMP-like binding mode to an active site
glutamine, Gln859 in PDE2. This cGMP-like binding mode is
orthogonal to that of the diazepinone cAMP-like binding mode in
PDE4 due to a glutamine switch.4 The presence of the cGMP-like
binding mode was confirmed by a crystal structure of a PDE2 selec-
tive inhibitor 1 bound to the active site of the PDE2 catalytic do-
main, PBP ID 4JIB (Fig. 1).
We identified diazepinone PDE2 inhibitors with submicromolar
potency and modest selectivity (33-fold) versus PDE4 as described
in the previous paper. However, these inhibitors suffered from
poor microsomal stability. At this point we shifted our strategy,
abandoning biaryl substituted imidazoles, as found in 1, for a more
metabolically stable 3,4-dimethoxyphenyl group and worked to
⇑
Corresponding author.
E-mail address: marksplummer@gmail.com (M.S. Plummer).
0960-894X/$ - see front matter Ó 2013 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.bmcl.2013.03.082

3444
M. S. Plummer et al. / Bioorg. Med. Chem. Lett. 23 (2013) 3443–3447
HO
O
NH
N
N
N
N
1
N
PDE2 IC50 = 45nM
PDE4B IC50 = 378
Selectivity PDE4b/PDE2 = 8.4
HLM t1/2 (min) = 6.9
Figure 1. The PDE2 selective inhibitor, 1, crystallized with the PDE2 catalytic domain, PBP ID 4JIB.
exploit the cGMP-like orthogonal binding mode to produce nano-
molar potent PDE2 inhibitors with greater than 100 fold selectivity
versus other PDEs.
ity over PDE4. The 2,4-dimethoxy analog 7 provided even greater
PDE2 inhibition (PDE2 IC50 = 14 nM) and 200-fold selectivity. The
requirement of a polar hydrogen bond acceptor is illustrated by
the carboxyl functionality in 8 a potent PDE2 inhibitor with 675-
fold selectivity. Indeed, a variety of hydrogen bond acceptor het-
erocyclic biaryl groups as in 9–12 provided potent PDE2 inhibition
with greater than 100-fold selectivity over PDE4 (Table 1). These
heterocyclic biaryl groups also provided a structural basis for the
modification of physicochemical properties (LogP, LogD, and aque-
ous solubility) which yielded significantly improved human liver
microsomal stability as compared to the more hydrophobic
analogs.
The X-ray structure of 1 reveals several points to differentiate
PDE2 inhibition from PDE4 inhibition. Firstly the inhibitor 1 binds
in an orthogonal manner to PDE2 wherein the hydroxyethyl side
chain points to the interior of the binding pocket, with a hydrogen
bond interaction between the hydroxyl group of the inhibitor and
Gln812. In PDE4, the residue corresponding to Gln812 is a proline
which cannot form such a hydrogen bond. This additional interac-
tion and the size of the surrounding pocket are points of differen-
tiation. The crystal structure also reveals several hydrophobic
residues adjacent to the phenyl group of the phenylimidazole moi-
ety at the entrance to the binding pocket, suggesting an additional
area to further optimize PDE2 inhibition and selectivity.
The hydrophobic area adjacent to the phenyl ring was especially
attractive as the metabolically more stable 3,4-dimethoxyphenyl
group led directly to design of 3-aryl-4-methoxyphenyl moieties,
enabling the exploration of the adjacent hydrophobic region. This
phenyl substitution improved selective inhibition of PDE2 over
PDE4 as seen in 2 and 3 when compared with 1 (Table 1). These
compounds were conveniently prepared by late-stage Suzuki cou-
pling of the corresponding 3-bromo-4-methoxyphenyldiazepinone
(V). The general synthetic route to the key intermediate (V) was
described in our previous paper. Synthesis of the required amino
ketone was accomplished by reaction of 3-bromo-4-methoxychlo-
roacetophenone with sodium biformylamide followed by acid
hydrolysis to give 3-bromo-4-methoxyphenylaminoketone hydro-
chloride III (Scheme 1).5
To explore interactions with Gln812 more fully, we targeted
hydroxypropyl substituted pyrazoles to establish the degree of
selectivity for PDE2 versus PDE4, because this functionality was
previously found to impart superior PDE2/PDE4 selectivity and
had greater metabolic stability than the corresponding hydroxy-
ethyl analogs. We postulated that improved selectivity would di-
rectly translate to the more potent hydroxyethyl pyrazole series.
Recall, the aryl group of the 3-aryl-4-methoxyphenyl substituent
was designed to interact with the rather hydrophobic exit from
the PDE2 catalytic active site, suggesting that hydrophobic biaryl
moieties would result in increased PDE2 inhibition and selectivity.
Surprisingly, the most hydrophobic residues, as in 2–5, had the
poorest PDE2 activity in the series (Table 1). Addition of slightly
more polar functionality gave the 2,5-dimethoxy analog, 6, result-
ing in a threefold increase in PDE2 inhibition and 176-fold selectiv-
The recently published PDE2A (215-900) crystal structure6 pro-
vides insight into the significant SAR observed with the distal aryl
ring that extends out of the catalytic site. The structure shows that
PDE2A exists as a dimer with the H-loops of the catalytic subunits
packing against each other at the dimer interface, occluding the
catalytic sites. The model predicts that cGMP binding to the GAF-
B domain, facilitates movement of the H-loops, so that they no
longer pack at the dimer interface allowing access to the catalytic
site. We speculate that the distal aryl ring in our inhibitors may
interact either with residues in the H-loop or some aspect of the
catalytic domain in the adjacent strand of the dimer.
Four of the most PDE2-selective analogs, 7 and 10–12, were
synthesized with the more potent but potentially less selective
hydroxyethyl substituted pyrazole moiety to provide 13–16 (Ta-
ble 2). As expected, three of the hydroxyethyl compounds had sig-
nificantly improved PDE2 potency compared with the
hydroxypropyl series, (7 vs 15; 10 vs 13; 11 vs 14). The remaining
matched pair (12, 16) had equivalent potency. In addition, selectiv-
ity for PDE2 inhibition was relatively poor for 13 and 16 as PDE4
inhibition increased. Interestingly, the dimethylaminopyridine
analog 14 displayed excellent PDE2 potency and selectivity versus
PDE4, PDE2 IC50 = 1.3 nM; selectivity, 410-fold. Consequently,
additional methoxy- and amino- pyridine and pyrimidine analogs,
17–22, were synthesized. All of these compounds had low nano-
molar PDE2 inhibition and greater than 140-fold selectivity over
PDE4. Further evaluation of these compounds against a broad
PDE panel showed that 22 exhibited excellent selectivity against
human recombinant, full length, partially purified PDEs. Com-
pound 22 demonstrated excellent PDE2 selectivity, greater than
1000-fold over PDEs 1B, 3A, 3B, 4A, 4B, 4C, 7A, 7B, 8A, 8B, 9, 10,
11, and between 241-fold to 419-fold selectivity versus PDEs 1A,
4D, 5 and 6.

M. S. Plummer et al. / Bioorg. Med. Chem. Lett. 23 (2013) 3443–3447
3445
Table 1
3-Aryl-4-methoxy phenyl analogs in the hydroxypropyl series
OH
O
NH
N
N
N
R
O
Compound
R=
Catalytic domain PDE2 IC50 (nM)
505
Catalytic domain PDE4B IC50 (nM)
100,000
HLM t1/2 (min)
Selectivity PDE4/PDE2
S
F
F
2
51
>200
3
4
483
586
100,000
8160
120
48
>200
14
F
O
O
5
6
7
156
46
2070
8110
82
61
13
176
O
O
14
100,000
120
>200
O
O
8
9
40
70
27,000
6190
120
120
675
88
OH
N
O
N
N
10
11
12
36
18
7
6740
5660
1920
120
42
187
323
263
N
N
O
O
N
N
120
Each IC50 is the average of at least two independent determinations. Each independent curve consisted of 10 half log dilutions with the 50% inhibition level at approximately
the center of the dilution scheme. Each point on the curve was independently determined in duplicate.
Even though all hydroxyethyl compounds had reduced human
microsomal stability compared with their hydroxypropyl counter-
parts, our immediate goal was to establish the unprecedented
pharmacology of PDE2 inhibition for OA pain. Thus, we evalu-
ated 22 for bioavailability in rats prior to testing in an efficacy
model.
Compounds were evaluated for pharmacokinetic properties in
male Sprague–Dawley rats. Two animals were dosed either intra-
venously (1 mg/kg) or orally (10 mg/kg), and pharmacokinetic
parameters were determined. The pyrazolodiazepinone PDE2
inhibitors 10 and 22 generally provided good oral bioavailability
with clearance and half-lives reflecting in vitro microsomal meta-
bolic stability (Table 3). The bioavailability of 10 and 22 would sug-
gest that oral dosing could be readily used in animal efficacy
models. However, not all of the compounds studied (data not
shown) had such good PK data, therefore, we chose to use subcu-
taneous (SC) dosing to provide internal consistency within our
in vivo studies. Compound 22 has a cLogP = 2.0 and a polar surface
area of 124 with a molecular weight of 464 g/mol which are out-
side the limits of preferred CNS property space, and thus not ex-

3446
M. S. Plummer et al. / Bioorg. Med. Chem. Lett. 23 (2013) 3443–3447
O
CHO
Cl
Br
O
N
O
CHO
a
O
b
NH3 Cl
Br
Br
OCH3
(II)
OCH3
(I)
(III)
HO
HO
THPO
+
O
O
O
N
N
NH
N
NH
N
N
OH
NO2
N
c,d
e
N
N
Br
HetAr
O
O
(VI)
(IV)
(V)
Scheme 1. Reagents and conditions: (a) sodium biformylamide, CH3CN, room temperature, 77%; (b) 6 N aqueous HCl/glacial acetic acid/reflux/60%; (c) EDAC/HOBt/DMF/
room temperature, 82%; (d) SnCl2ÁH2O/EtOH/reflux, 66%; (e) (PPh3)4Pd, 2M Na2CO3, ArB(OH)2 or HetArB(OH)2, DMF, 120–150 °C, 30–70%, microwave.
Table 2
3-Aryl-4-methoxy phenyl analogs in the hydroxyethyl series
HO
O
NH
N
N
N
R
O
Compound
R=
Catalytic domain PDE2 IC50 (nM)
Catalytic domain PDE4B IC50 (nM)
110
HLM t1/2 (min)
Selectivity PDE4/PDE2
16
N
N
13
6.8
59
N
N
O
14
15
1.3
3.4
534
741
74
43
410
217
O
O
N
16
9.3
138
36
15
O
N
N
N
N
17
18
1.2
1.4
410
265
33
39
341
179
N
N
O
N
N
O
19
1.8
418
57
229
O
N
N
20
21
22
2.5
2.5
3.3
695
353
562
36
18
34
283
144
168
N
N
N
O
N
N
N

M. S. Plummer et al. / Bioorg. Med. Chem. Lett. 23 (2013) 3443–3447
3447
Table 3
Pharmacokinetic parameters for pyrazolodiazepinone PDE-2 inhibitors
Compound
IV T1/2 (h)
Cl (mL/min/kg)
Vdss (mL/kg)
PO AUC (ng h/mL)
PO Tmax (h)
%F
HO
O
N
N
NH
N
0.53
70
1100
1660
2.25
75
N
N
10
22
0.73
1.5
43
27
2000
3020
1160
4530
0.5
3.0
31
78
was greater than that of a single oral dose of the COX-2 inhibitor,
rofecoxib. The data suggests that PDE2 inhibition is a novel ap-
proach to the treatment of osteoarthritis pain.
Plasma levels of 22 at one hour post dose were 1190 ng/mL
(2.5 lM). Based on the rat plasma free fraction of 3.2%, this plasma
60
50
40
30
20
10
0
Vehicle
22 10mg/kg, SC
Rofecoxib (10 mg/kg, PO)
level corresponds to a free concentration of 38 ng/mL (80 nM). This
is well above the in vitro PDE2 IC50 of 3.3 nM and below the IC50 of
other PDEs suggesting the efficacy is due to PDE2 inhibition. The
total brain exposure was 214 ng/mL (475 nM), corresponding to a
free brain concentration of 1.9 ng/mL (4.3 nM) based on a rat free
brain fraction of 0.9%. Thus, the observed analgesic effect appears
to be primarily driven by PDE2 inhibition in the periphery.
Highly potent, highly selective, orally bioavailable, PDE2 inhib-
itors were discovered through optimization of weak PDE2 activity
present in a historic series of PDE4 inhibitors. The PDE2 inhibitors
bind orthogonally to PDE2 in a cGMP-like mode when compared to
the cAMP-like binding mode observed for this class of inhibitor
when bound to PDE4. This binding mode has been confirmed by
an inhibitor bound crystal structure in the catalytic domain of
PDE2. Replacement of metabolically-labile imidazole biaryls with
3-heterobiaryl-4-methoxyphenyl moieties in the pyrazolodiazepi-
nones has resulted in potent, highly selective, metabolically stable,
peripherally restricted, bioavailable PDE2 inhibitors. One of these
inhibitors, compound 22, demonstrated significant in vivo activity
1 and 3 h after a single dose in a rat model of osteoarthritis pain.
These results establish selective PDE2 inhibition as a potential
new mechanism by which to treat osteoarthritis pain.
44%
*
79%
*
79%
*
100%
1 h
3 h
-10
Time Post Dose
Figure 2. On day 27 post MMT surgery, rats were assessed for change in hind paw
weight distribution and distributed into 3 dose groups. The following day (day 28
post surgery) the rats were administered compound or vehicle (sc) and reassessed 1
and 3 h post dose. Data are expressed as mean SEM. Statistically significant
differences were determined by one-way ANOVA followed by Dunnett’s test. (⁄P
<0.05 vs vehicle at same time point). N = 6 rats per group.
pected to show good brain penetration.7 Additionally, the half-life
of 22 in human liver microsomes was only 34 min while in rat liver
microsomes it was >60 min. Collectively, these data along with the
pharmacokinetic data indicate that 22 should serve as a good tool
compound to evaluate the effects of peripheral PDE2 inhibition in
rat in vivo models of OA pain. The expected target site of activity is
the dorsal root ganglia, which are found outside of the CNS blood
brain barrier.
In vivo activity of 22 was evaluated in the rat medial meniscal
transection (MMT) model of osteoarthritis pain (Fig. 2).8 The
MMT model surgically transects the medial meniscus in the right
femorotibial joint of the rat. The mechanical disruption of the joint
simulates OA-like pain causing the animal to shift weight bearing
to the left limb. This shift is measured by placing the animal in a
device that assesses weight distribution to each hind limb. Nonste-
roidal anti-inflammatory drugs (NSAIDs), cyclooxygenase 2 (COX2)
inhibitors, and acetaminophen have been shown to alleviate the
imbalance of weight distribution in this model. The time course re-
sults of a single, subcutaneous administration of 22 reveal signifi-
cant inhibition of change in hind paw weight distribution versus
vehicle at both the 1 and 3 h post-dose time points and showed
sustained activity over the course of the study (Fig. 1). The effect
References and notes
1. Gupta, R.; Kumar, G.; Kumar, R. S. Methods Find Exp. Clin. Pharmacol. 2005, 27,
101.
2. Tegeder, I.; Schmidtko, A.; Niederberger, E.; Ruth, P.; Geisslinger, G. Neurosci.
Lett. 2002, 332, 146.
3. de Vente, J.; Markerink-van Ittersum, M.; Vles, J. S. H. J. Chem. Neuroanat. 2006,
31, 263.
4. Zhang, K. Y. J.; Card, G. L.; Suzuki, Y.; Artis, D. R.; Fong, D.; Gillette, S.; Hsieh, D.;
Neiman, J.; West, B. L.; Zhang, C.; Milburn, M. V.; Kim, S.-H.; Schlessinger, J.;
Bollag, G. A. Mol. Cell 2004, 15, 279.
5. (a) Suzuki, M.; Iwasaki, T.; Muneji, M.; Okumura, K.; Matsumoto, K. J. Org. Chem.
1973, 38, 3571; (b) Maeda, S.; Suzuki, M.; Iwasaki, T.; Matsumoto, K.; Iwasawa,
Y. Chem. Pharm. Bull. 1984, 32, 2536.
6. Pandit, J.; Forman, M. D.; Fennell, K. F.; Dillman, K. S.; Menniti, F. S. PNAS 2009,
106, 18225.
7. Wager, T. T.; Hou, H.; Verhost, P. R.; Villalobost, A. ACS Chem. Neurosci. 2010, 1,
435.
8. Hook, K. E.; Juneau, P. L.; Connor, J. R.; Kilgore, K. S. Osteoarthritis Cartilage 2006,
14, 1041.