Discovery of orally active pyrazoloquinolines as potent PDE10 inhibitors for the management of schizophrenia

Article abstract of DOI:10.1016/j.bmcl.2011.11.023

A series of pyrazoloquinoline analogs have been synthesized and shown to bind to PDE10 with high affinity. From the SAR study and our lead optimization efforts, compounds 16 and 27 were found to possess potent oral antipsychotic activity in the MK-801 induced hyperactive rat model.

Full text of DOI:10.1016/j.bmcl.2011.11.023

Bioorganic & Medicinal Chemistry Letters 22 (2012) 235–239
Contents lists available at SciVerse ScienceDirect
Bioorganic & Medicinal Chemistry Letters
journal homepage: www.elsevier.com/locate/bmcl
Discovery of orally active pyrazoloquinolines as potent PDE10 inhibitors for
the management of schizophrenia
Shu-Wei Yang a, , Jennifer Smotryski a, William T. McElroy a, Zheng Tan a, Ginny Ho a, Deen Tulshian a,
⇑
William J. Greenlee a, Mario Guzzi b, Xiaoping Zhang b, Deborra Mullins b, Li Xiao c, Alan Hruza c,
Tze-Ming Chan c, Diane Rindgen d, Carina Bleickardt b, Robert Hodgson b
a Department of Medicinal Chemistry, Merck Research Laboratories, 2015 Galloping Hill Road, Kenilworth, NJ 07033, United States
b Department of Neuroscience, Merck Research Laboratories, 2015 Galloping Hill Road, Kenilworth, NJ 07033, United States
c Department of Structural Chemistry, Merck Research Laboratories, 2015 Galloping Hill Road, Kenilworth, NJ 07033, United States
d Department of Drug Metabolism and Pharmacokinetics, Merck Research Laboratories, 2015 Galloping Hill Road, Kenilworth, NJ 07033, United States
a r t i c l e i n f o
a b s t r a c t
Article history:
A series of pyrazoloquinoline analogs have been synthesized and shown to bind to PDE10 with high affin-
ity. From the SAR study and our lead optimization efforts, compounds 16 and 27 were found to possess
potent oral antipsychotic activity in the MK-801 induced hyperactive rat model.
Ó 2011 Elsevier Ltd. All rights reserved.
Received 18 October 2011
Revised 4 November 2011
Accepted 7 November 2011
Available online 16 November 2011
Keywords:
PDE10
Inhibitors
Pyrazoloquinoline
Schizophrenia
Antipsychotic
Schizophrenia is a mental illness affecting ꢀ1% of the world’s
population.1 Although second generation antipsychotic drugs show
effectiveness in treating positive psychotic symptoms (e.g., para-
noid, visual delusions and hallucinations) of schizophrenia, they
are ineffective against negative symptoms (e.g., absence of speech
and lack of interest and drive) and cognitive disorder. However,
even with current drug therapy, ꢀ15% patients still have residual
positive symptoms, and only a small portion of patients under
treatment are able to lead independent lives.2 In addition, current
antipsychotic agents have shown adverse effects, including QT pro-
longation, weight gain, diabetes, and extrapyramidal syndrome
(involuntary muscle movements).3–5 There is an unmet medical
need for improved treatment of schizophrenia.
The phosphodiesterases (PDEs) are a family of degradative en-
zymes that hydrolyze the second messengers, cyclic adenosine
monophosphate (cAMP) and/or cyclic guanosine monophosphate
(cGMP), to terminate signal transduction. PDE10, a dual cAMP/
cGMP phosphodiesterase, is expressed at high levels in the striatal
medium spiny neurons, but is expressed at very low levels else-
where in the brain and other tissues.6 PDE10 inhibition mimics
D2 dopamine receptor antagonism in the indirect striatopallidal
output pathway through increasing cAMP and cGMP levels. This
antagonistic effect could increase the activity of the striatonigral
output and normalize the reduced striatal output that character-
izes schizophrenia.7
Papaverine, a potent PDE10A inhibitor, was utilized as a phar-
macological probe in mice, and exhibited efficacies in rat models
of schizophrenia.7,8 Several PDE10A inhibitors have been reported
to exhibit efficacies in a range of antipsychotic models including
the models established for negative symptoms and cognitive defi-
cits.7–9 Thus, selective PDE10 inhibitors provide a novel therapeutic
approach for the treatment of schizophrenia.9–11 Recent advances
in X-ray co-crystal structures of PDE10 have triggered structure-
based drug design for several pharmacophore models.10,11
From high-throughput screening of our chemical library, a ser-
ies of analogs with a pyrazoloquinoline scaffold were identified to
display PDE10A inhibitory activity. These compounds possessed
C-4 N-substituted moieties. To further change and optimize the
chemical properties of that series (4-aminopyridine series), chem-
istry was developed to prepare C-4 carbon-linked heterocyclic
analogs. In this letter, we will describe syntheses and structure–
activity relationships of C-4 carbon-linked pyrazoloquinoline
analogs. The SAR study resulted in identification of highly potent
and selective PDE10A ligands, demonstrating oral in vivo efficacy
in the MK-801 induced hyperactivity rat model for anti-psychosis.
⇑
Corresponding author. Tel.: +1 908 740 7291; fax: +1 908 740 7164.
E-mail address: shu-wei.yang@merck.com (S.-W. Yang).
0960-894X/$ - see front matter Ó 2011 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2011.11.023

236
R1,R2
S.-W. Yang et al. / Bioorg. Med. Chem. Lett. 22 (2012) 235–239
R1,R2
R1,R2
R1,R2
H
N
b
H
N
H
NH2
N
N
N
a
N
N
N
N
R1,R2
a
H
N
R1,R2
H
N
H
N
c
H
N
CHO
1
Cl
CN
N
9
NH2
S
R1,R2
8
b
EtO2C
7
NCS
R1,R2
H
3
4
N
N
N
c
R1,R2
2
R1,R2
H
H
N
H
N
N
N
e
R5,R6
d
N
N
N
10
HO2C
X
Cl
6
5
X = NR, O, or CH2
Scheme 1. Reagents and conditions: (a) NH4OH, CS2, EtOH, rt, 1 h; then
ClCH2CO2Na, rt, 1 h; then NH2NH2, rt; (b) NH2NH2, DCM, rt; (c) ethyl 2-chloro-
acetoacetate, EtOH, rt; (d) KOH in EtOH–H2O, reflux; (e) POCl3, 100 °C.
Scheme 2. Reagents and conditions: (a) KCN, 110 °C, DMSO; (b) DIBAL, toluene; (c)
amine, NaBH(OAc)3.
Table 1
The synthetic route to the pyrazoloquinoline core is summa-
rized in Scheme 1. The commercially available aniline (1) was
transformed to carbothio-hydrazine 3 in one pot. Alternatively 3
could be conveniently obtained from commercially available phe-
nyl isothiocyanate (2). Then 3 was coupled with ethyl 2-chloro-
acetoacetate and cyclized to form pyrazole 4.9,10 Ester 4 was fur-
ther hydrolyzed to acid 5, and cyclization could occur under neat
phosphorus(V) oxychloride condition at 100 °C to obtain 6.
The C-4 heterocyclic-methyl analogs (10) were synthesized as
described in Scheme 2. The C-4 chloro atom of 7 was replaced with
a cyano group using potassium cyanide to afford 8, followed by DI-
BAL reduction to give aldehyde 9. Reductive amination of 9 with
various heterocyclic amines afforded 10. This is the first report of
replacing the C-4 chloro atom with reactive one carbon unit for
further transformation.
SAR of the C-4 piperidinyl and pyrrolidinyl analogs
H
N
N
N
O
N
n
4
2
R
3
Compds
R
n
PDE10 Ki (nM)
Piperidines
11
12
13
14
15
H
4-OH
4-F
3-CF3
4-CF3
2
2
2
2
2
804
57
202
353
871
Target compounds were tested for their affinity at the cloned
human recombinant PDE10A1 and/or other PDE isozymes (BPS
Bioscience Inc.) by measuring their ability to compete with
[3H]cAMP (Amersham) as the radioligand.
Pyrrolidines
16
17
18
H
1
1
1
24
3
32
3-OH (R)
3-CH2OH
The unsubstituted-piperidinyl analog (11) showed PDE10 bind-
ing affinity with a Ki of 804 nM. Introduction of a hydroxy group at
the 4 position of the piperidine ring led to a ꢀ14-fold increase of
PDE10 binding affinity (12, Ki 57 nM, Table 1). The analogs with
4-fluoro or 3- or 4-trifluoromethyl piperidine (13–15) did not show
desired potency (>100 nM). Three pyrrolidinylmethyl analogs were
prepared to explore the relationship of reduced C-4 ring size and
PDE10 affinity. The unsubstituted-pyrrolidinyl analog 16 displayed
improved PDE10 affinity (Ki 24 nM) compared to that of piperidinyl
11. Hydroxyl substituted analog 17 further improved the PDE10
binding affinity to 3 nM. The SAR trend is consistent with the
SAR observed for piperidinyl analog 12. However 3-hydroxymethyl
substitution did not improve binding affinity in this series.
To explore whether the oxygen atom inside the C-4 ring could
be tolerated, the morpholinemethyl analog (19–23, Table 2) was
prepared. The unsubstituted-morpholine analog 19 exhibited
highly potent PDE10 binding affinity with a Ki of 1 nM. Because
19 displayed high PDE10A binding affinity, it was soaked with
PDE10A crystals for an X-ray co-crystallography study (deposition
number: PDB code 3UI7). Compound 19 bound with PDE10A in the
catalytic pocket (Figs. 1 and 2). The conserved glutamine Gln-726
formed strong bidentate hydrogen bonds with N-H and N-9 of
the pyrazoloquinoline ring. The tricyclic ring of 19 was clamped
by two phenylanalines, Phe-696 and Phe-729. The morpholine
moiety extended into the deep pocket formed by Y524, H525,
F696, and I692, and the oxygen atom formed an additional hydro-
gen bond with histidine (His-525). These binding interactions
probably contributed to its high binding affinity. The methanethio
group of methionine Met-713 was close to C-7. Therefore C-7
Table 2
SAR of the C-4 morpholinyl and piperazinyl analogs
H
N
N
N
O
6
N
X
R
3
2
\
Compds
X
R
PDE10 Ki (nM)
19
20
21
22
23
24
25
26
O
O
O
O
H
2-Me
2-CH2OH
2,2-di-Me
cis 2,6-di-Me
H
1
3
0.6
171
26
49
9
O
NH
NH
NH
2 @O
2-Me (R)
30
substitution might clash with Met-713, and reduce PDE10A bind-
ing affinity. The C-8 methyl group of 19 was proximate to Gly-
725, and this might contribute to its good selectivity over other
PDE isozymes. Other PDE isozymes possessed larger amino acids
(e.g., Leu, Ile, etc.) instead of glycine at position 725, and the large

S.-W. Yang et al. / Bioorg. Med. Chem. Lett. 22 (2012) 235–239
237
Table 3
SAR of the A-ring modification of 27
8
H
N
N
7
R1,R2
A
N
6
4
N
O
Compds
R1, R2
PDE10 Ki (nM)
28
29
27
30
31
32
33
34
35
36
6-OMe
6-CF3
6-OMe, 8-Me
6-OH, 8-Me
6-O(CH2)2OMe, 8-Me
6-OiPr, 8-Me
6-OCF3, 8-Me
6,8-di-OMe
153
104
5
2
1
8
19
14
41
37
Figure 1. X-ray structure of PDE10 and 19 (yellow).
6-Cl, 8-Me
6-Br, 8-OCF3
were prepared using different substituted anilines (1) as precur-
sors. The C-5 substitution showed potential reduction of the stabil-
ity of the compound (data not shown), and was not desired. The C-7
substitution usually resulted in reduction of the binding affinity
due to clashing with Met713 residue of PDE10. Therefore only C-6
and C-8 substitutions were investigated. Both analogs 28 and 29
without 8-substitution showed poor binding affinity, Ki >100 nM.
Compared to 27, removal of the 8-methyl group (28) led to a
ꢀ20-fold reduction of the PDE10 binding affinity. This indicated
that hydrophobic interaction between 8-methyl group and PDE10
assisted binding. Therefore 8-aliphatic substitution was preferred.
Replacing 6-OMe group with other alkyl ethers such as O-iPr or
OCF3 resulted in slightly reduced affinity (Ki values of 8 and
19 nM for 32 and 33, respectively). Replacing the 8-methyl group
of 27 with a methoxyl group lead to slightly reduced affinity (34,
Ki 14 nM). PDE10 binding affinity diminished when the 6-position
was substituted with halogen (Cl or Br) such as analogs 35 and 36.
To investigate the relationship of the steric bulkiness of the C-4
ring of the molecule and PDE10 binding affinity, several bridged
analogs including bridged pyrrolidine (37 and 40), piperidine
(38), and morpholine (39 and 41) were prepared (Table 4). Overall
these compounds showed significantly reduced PDE10 binding
affinities, especially for 37 and 39, ꢀ46–50-fold reduction, com-
pared to those of the non-bridged analogs (16 and 19, respec-
tively). The morpholine analogs (39 and 41) again showed better
potency, compared to the pyrrolidine or piperidine analogs (37,
38, 40). Therefore the strategy of increasing steric bulkiness did
not lead to compounds with potent binding affinity.
Figure 2. Binding mode and molecular interactions for 19 and PDE10 based on X-
ray crystallography.
amino acids might clash with the C-8 methyl group of 19 and re-
duce binding affinity. The co-crystal structure explained some of
the SAR and further assisted our future SAR development.
Additional SAR was conducted for the substituted morpholine
analogs (20–23). Slight variation of 2-substituents (methyl or
hydroxylmethyl for 20 and 21, respectively) on the morpholine ring
was tolerated. However binding affinity significantly dropped for
the di-methyl analogs (22 and 23). The C-4 piperazinemethyl ana-
logs (24–26) were also prepared for SAR investigation. Among
them, the piperazin-2-one analog (25) showed the best binding
affinity (9 nM). In general, polar substituents or replacements
(OH, NH, O, or @O) on the C-4 cycloalkyl ring improved PDE10 bind-
ing affinity. This could be explained based on the X-ray co-crystal
structure. In the X-ray structure, the C-4 cycloalkyl ring resided in
the large hydrophilic pocket containing water molecules. Therefore
hydrophilic substitution on the C-4 ring provided better binding
interaction compared to that of non-polar substitution.
Some compounds with potent PDE10 binding affinities were
chosen for selectivity evaluation against other PDE isozymes (Table
5). In general these compounds were highly selective (>1000-fold)
over PDE1A3, PDE2A, PDE3A, PDE4B2, PDE5A1, PDE7A1, PDE8A1,
and PDE9A1. They were also selective over PDE6 and PDE11A3
with ꢀ320- to 6000-fold selectivity.
Some compounds with high PDE10 binding affinity were evalu-
ated for schizophrenia rat models based on inhibition of MK-801
induced hyperactivity. MK-801 is a NMDA-receptor antagonist,
which produces schizophrenia-like symptoms in rats.13 Selected
compounds were screened at 30 mg/kg oral dosing, and the ones
that showed reduction of hyperactivity in rats were further evalu-
ated at 10 and 3 mg/kg (Table 6). Compounds 12, 16–21, and 27 re-
versed hyperactivity in the MK-801 treated rats, while 24–26 did
not show efficacy in the same model. All of the active compounds
except 18 reduced the hyperactivity higher than 100%, compared
The homomorpholinemethyl analog (27),12 a homolog of 19,
retained potent binding affinity (Ki 5 nM). To investigate the A-ring
SAR, A-ring modified homomorpholine analogs (28–36, Table 3)

238
S.-W. Yang et al. / Bioorg. Med. Chem. Lett. 22 (2012) 235–239
Table 4
reduced in vivo efficacy in the MK-801 treated rat model. Hence,
addition of the hydroxyl group to the C-4 heterocyclic was not
desired.
SAR of the C-4 bridged bi-cyclic analogs
R2
H
N
Compound 16 did not show good AUC in the rat pharmacoki-
N
netic (PK) study (0.3
for PK and in vitro safety studies. Compound 27 displayed reason-
able rat AUC (3.8 M h) and brain/plasma ratio (1.6) in the PK
lM h). Compound 27 was further profiled
N
R1
l
Q
study. Compound 27 showed potent PDE10 inhibitory activity with
Ki = 5 nM and >520-fold selectivity over other PDE isozymes. It
showed minimal inhibition (5% at 10 lM) in hERG assay (Ionwork),
Compds
R1, R2
Q
PDE10 Ki (nM)
and did not show significant induction effect in the hPXR reporter
gene assay (0.17-fold relative to rifampicin). Compound 27 dis-
played no inhibition of CYP P450 enzymes (2D6, 2C9, and 3A4)
up to 20 lM. Compound 27 exhibited reasonable pharmacokinetic
profile with about 10-fold therapeutic window against hypoloco-
37
6-OMe, 8-Me
N
N
N
N
N
1200
38
39
40
6-OMe, 8-Me
6-OMe, 8-Me
6-Cl, 8-Me
265
46
OH
motion side effect (MED 30 mg/kg).
O
O
PDE10: Ki: 5 nM
H
other PDEs: >520 fold
N
N
229
rat PK: AUC: 3763 nM.h (0-6 h)
Brain: 42 ng/g (6 h)
N
O
Brain/Plasma ratio: 1.56 (6 h)
Rat Hyperactivity: MED: 3 mg/kg
Rat Hypolocomotion: MED: 30 mg/kg
Cyp Inh. (2D6, 3A4, 2C9): clean, >20 µM
hERG (iw): 5.3% inhibition (10 uM)
hPXR: 0.17 fold to rifampicin @ 1 µM
Protein binding (rat): 97%
41
6-Cl, 8-Me
60
N
O
27
Hepatocyte clearance:
Table 5
Binding affinity of some selective compounds versus other PDE isozymes
8.3 uL/m/M cells (Human)
<1 uL/m/M cels (Rat)
Compds PDE5A1 Ki PDE6 Ki PDE11A3 Ki
Other PDEsaKi
6 uL/m/M cells (Monkey)
(
l
M)
(
l
M)
(
l
M)
(lM)
In conclusion, a novel series of potent PDE10 inhibitors based
on the pyrazoloquinoline scaffold were discovered. Several of these
compounds showed Ki levels in the single digit nM range, and one
compound (21) showed Ki at sub nM level. These compounds
showed good selectivity over other PDE isozymes. Among the po-
tent PDE10 inhibitors identified, compounds 16 and 27 demon-
strated potent oral anti-hyperactivity efficacy in the MK-801 rat
model, with MED values of 1 and 3 mg/kg, respectively. Compound
27 was selected for further in vivo safety studies in animals.
17
19
20
21
27
30
31
32
6.4
4.7
2
1.9
>10
>10
>10
ND
>10
>10
>10
>10
10.6
8.9
14.6
6.6
1.2
1.6
4.9
2.3
8.3
0.3
2.1
2.4
5.9
5
9.7
20.8
24.4
>10
>10
>10
a
Other PDE isozymes: PDE1A3, PDE2A, PDE3A, PDE4B2, PDE7A1, PDE8A1,
PDE9A1.
Acknowledgments
Table 6
The authors wish to acknowledge Drs. Mark Liang, Jianshe
Kong, Jesse Wong, Ms. Teresa Andreani, and Mr. Meng Tao for prep-
aration of the intermediates, the NMR group for structure confir-
mation of some analogs, Dr. Xiaoming Cui and DMPK group for
acquiring hPXR and pharmacokinetic data, and Drs. Steve Sorota
and Tony Priestley for providing hERG data.
In vivo screening data of selected analogs in anti-hyperactivity rat models
Compds
3 mpk
10 mpk
30 mpk
12
16
17
18
19
20
21
24
25
26
27
35
Inactive
96%a
Inactive
107%
Inactive
Inactive
131%
88%
Inactive
NT
NT
136%
122%
102%
38%
109%
130%
104%
Inactive
Inactive
Inactive
103%
87%
Inactive
Inactive
Inactive
Inactive
Inactive
NTb
References and notes
1. Saha, S.; Chant, D.; Welham, J.; McGrath, J. PLoS Med. 2005, 2, e141.
2. Cancro, R. Schizophrenia. In Treatments of Psychiatric Disorders: A Task Force
Report of the American Psychiatric Association; American Psychiatric Association:
Washington, DC, 1989; pp 1485–1606.
NT
NT
NT
120%
Inactive
66%c
Inactive
3. Taylor, D. M. Acta Psychiatr. Scand. 2003, 107, 85.
4. Casey, D. E. Am. J. Med. 2005, 118(Suppl. 2), 15S.
5. Newcomer, J. W. CNS Drugs 2005, 19(Suppl. 1), 1.
The percentages indicate the reduced percentage of the induced hyperactivity.
a
MED: 1 mg/kg.
NT: Not tested.
MED: 3 mg/kg.
b
6. (a) Loughney, K.; Snyder, P. B.; Uher, L.; Rosman, G. J.; Ferguson, K.; Florio, V. A.
Gene 1999, 234, 109; (b) Soderling, S. H.; Bayuga, S. J.; Beavo, J. A. Proc. Natl.
Acad. Sci. U.S.A. 1999, 96, 7071; (c) Fujishige, K.; Kotera, J.; Michibata, H.; Yuasa,
K.; Takebayashi, S.; Okumura, K.; Omori, K. J. Biol. Chem. 1999, 274, 18438.
7. (a) Siuciak, J. A.; Chapin, D. S.; Harms, J. F.; Lebel, L. A.; James, L. C.; McCarthy, S.
A.; Chambers, L. K.; Shrikehande, A.; Wong, S. K.; Menniti, F. S.; Schmidt, C. J.
Neuropharmacology 2006, 51, 386; (b) Siuciak, J. A.; McCarthy, S. A.; Chapin, D.
S.; Fujiwara, R. A.; James, L. C.; Williams, R. D.; Stock, J. L.; McNeish, J. D.; Strick,
C. A.; Menniti, F. S.; Schmidt, C. J. Neuropharmacology 2006, 51, 374.
8. Siuciak, J. A.; Strick, C. A. Drug Discovery Today 2006, 3, 527.
9. For review, see: (a) Menniti, F. S.; Chappie, T. A.; Humphrey, J. M.; Schmidt, C. J.
Curr. Opin. Investig. Drugs 2007, 8, 54; (b) Menniti, F. S.; Faraci, W. S.; Schmidt, C.
J. Nat. Rev. Drug Disc. 2006, 5, 660; (c) Kehler, J.; Kilburn, J. P. Expert Opin. Ther.
Patents 2009, 19, 1715; (d) Chappie, T.; Humphrey, J.; Menniti, F.; Schmidt, C.
Curr. Opin. Drug Discov. Devel. 2009, 12, 458.
c
to that of vehicle and were further evaluated. In the further testing,
only compounds 16, 19, 20, and 29 showed anti-hyperactive effi-
cacy (reducing 107%, 131%, 88%, and 120%, respectively). Finally
minimum effective doses of the most potent analogs (16 and 27)
were determined as 1 and 3 mg/kg, respectively. Although the hy-
droxyl substitution (17, 19) enhanced PDE10 binding affinity com-
pared to that of the unsubstituted analogs (16, 21), it however

S.-W. Yang et al. / Bioorg. Med. Chem. Lett. 22 (2012) 235–239
239
10. Chappie, T. A.; Humphrey, J. M.; Allen, M. P.; Estep, K. G.; Fox, C. B.; Lebel, L. A.;
Liras, S.; Marr, E. S.; Menniti, F. S.; Pandit, J.; Schmidt, C. J.; Tu, M.; Williams, R.
D.; Yang, F. V. J. Med. Chem. 2007, 50, 182.
11. Verhoest, P. R.; Chapin, D. S.; Corman, M.; Fonseca, K.; Harms, J. F.; Hou, X.;
Marr, E. S.; Menniti, F. S.; Nelson, F.; O’Connor, R.; Pandit, J.; Proulx-LaFrance,
C.; Schmidt, A. W.; Schmidt, C. J.; Suiciak, J. A.; Liras, S. J. Med. Chem. 2009, 52,
5188.
12. Compound 27: 1H NMR (CDCl3): d 1.86 (tt, J = 6.1, 5.8 Hz, 2H), 2.74 (m, 2H), 2.82
(t, J = 5.8 Hz, 2H), 2.86 (s, 3H), 2.87 (s, 3H), 3.61 (m, 2H), 3.79 (t, J = 6.1 Hz, 2H),
3.93 (s, 3H), 4.39 (s, 2H), 7.32 (d, J = 2.8 Hz, 1H), 7.66 (d, J = 2.8 Hz, 1H), 11.41
(br s, 1H, NH). ESI-MS: 341 [M+H]+, C19H24N4O2.
13. Rung, J. P.; Carlsson, A.; Markinhuhta, K. R.; Carlsson, M. L. Prog.
Neuropsychopharmacol. Biol. Psychiatry 2005, 29, 827.