Novel atypical antipsychotic agents: Rational design, an efficient palladium-catalyzed route, and pharmacological studies

Article abstract of DOI:10.1021/jm049629t

Using rational drug design to develop atypical antipsychotic drug candidates, we generated novel and metabolically stable pyrrolobenzazepines with an optimized pK_i 5-HT_2A/O₂ ratio. 5a, obtained by a new palladium-catalyzed

Full text of DOI:10.1021/jm049629t

J. Med. Chem. 2005, 48, 1705-1708
1705
Chart 1. Reference and Title Compounds and the
Three-Step Pd-Catalyzed Synthesis of 5a
Novel Atypical Antipsychotic Agents:
Rational Design, an Efficient
Palladium-Catalyzed Route, and
Pharmacological Studies
Giuseppe Campiani,*,§,# Stefania Butini,§,#
Caterina Fattorusso,|,# Francesco Trotta,§,#
Sandra Gemma,§,# Bruno Catalanotti,|,# Vito Nacci,§,#
Isabella Fiorini,§,# Alfredo Cagnotto,‡,#
Ilario Mereghetti,‡,# Tiziana Mennini,‡,#
Patrizia Minetti,† M. Assunta Di Cesare,†
M. Antonietta Stasi,† Stefano Di Serio,†
Orlando Ghirardi,† Ornella Tinti,† and
Paolo Carminati†
Dipartimento Farmaco Chimico Tecnologico, Via Aldo Moro,
and European Research Centre for Drug Discovery and
Development, Universita` di Siena, 53100 Siena, Italy,
Dipartimento di Chimica delle Sostanze Naturali,
Universita` di Napoli Federico II, Via D. Montesano 49,
80131 Napoli, Italy, Istituto di Ricerche Farmacologiche
Mario Negri, Via Eritrea 62, 20157 Milano, Italy, and
Sigma-Tau Industrie Farmaceutiche Riunite, spa, Via
Pontina Km 30,400, 00040 Pomezia, Italy
Received May 20, 2004
Abstract: Using rational drug design to develop atypical
antipsychotic drug candidates, we generated novel and meta-
bolically stable pyrrolobenzazepines with an optimized pKi
5-HT2A/D2 ratio. 5a, obtained by a new palladium-catalyzed
three-step synthesis, was selected for further pharmacological
and biochemical investigations and showed atypical anti-
psychotic properties in vivo. 5a was active on conditioned
avoidance response at 0.56 mg/kg, it had low cataleptic
potential and proved to be better than ST1899, clozapine, and
olanzapine, representing a new clinical candidate.
Schizophrenia is a disabling disease that affects 1%
of the world population and is the second major psy-
chiatric disease.1 There are no specific characteristics
for the diagnosis of schizophrenia, and no single symp-
tom is consistently present in all patients.2 The symp-
toms most commonly associated with the disease are
positive symptoms and denote the presence of grossly
abnormal behavior. These include thought disorder,
delusions, and hallucinations. Less obvious than the
positive symptoms but equally serious are the negative
symptoms, namely, absence of normal behavior.2 These
include the flat or blunted affect (i.e., lack of emotional
expression), apathy, and social withdrawal. Cognitive
deficits include reduced working memory, attention, and
verbal fluency. Before the 1990s, antipsychotic drug
development focused exclusively on agents with sub-
stantial activity on dopamine receptors (typically halo-
peridol).3 Typical antipsychotic drugs carry a heavy side
effect burden, fail to manage negative symptoms (ha-
loperidol), and are ineffective in about one-third of
patients with schizophrenia. Over the past decade, the
development of new antipsychotic agents has shifted
from selective dopamine antagonist to compounds that
have a broader receptor affinity profile (the so-called
atypical antipsychotics), characterized by improved
clinical efficacy and fewer side effects.4 The first atypical
neuroleptic drug was clozapine, followed by olanzapine
(1 and 2, Chart 1). Although many atypical antipsy-
chotic drugs have recently been approved for the treat-
ment of schizophrenia,5 olanzapine is still invaluable for
psychosis, having high clinical efficacy. However, olan-
zapine may precipitate or unmask diabetes in suscep-
tible patients, and it has been associated with a 12%
increase in excessive appetite with respect to halo-
peridol.6
In the past 10 years, several novel strategies (gluta-
mate (mGluRs)2 and tachykinin receptors)7 have been
proposed for the development of atypical antipsychotics
with fewer side effects; nevertheless, the need for an
ideal antipsychotic agent continues to stimulate the
search for newer and safer drugs. We recently developed
a new class of atypical antipsychotics with the pyrrolo-
[2,1-b][1,3]benzothiazepine structure.8 The benzothiaz-
epine 3a (ST1899) (Chart 1) showed an atypical binding
profile, and in vivo pharmacological studies indicated
its pharmacological behavior to be superior to that of
* To whom correspondence should be addressed. Phone: 0039-0577-
234172. Fax: 0039-0577-234333. E-mail: campiani@unisi.it.
§ Dipartimento Farmaco Chimico Tecnologico, Universita` di Siena.
# European Research Centre for Drug Discovery and Development.
| Universita` di Napoli Federico II.
‡ Istituto di Ricerche Farmacologiche Mario Negri.
† Sigma-Tau Industrie Farmaceutiche Riunite.
10.1021/jm049629t CCC: $30.25 © 2005 American Chemical Society
Published on Web 07/28/2004

1706 Journal of Medicinal Chemistry, 2005, Vol. 48, No. 6
Letters
olanzapine. However, 3a may undergo several metabolic
pathways in vivo, including the oxidative transforma-
tion into its sulfoxide derivative, that could lead to a
partial inactivation, as shown for quetiapine9 and oc-
toclothepin.
compounds at the D2 receptor,8 we calculated the
conformational inversion energy barrier for the tricyclic
system of 5a-c, which was around 14 and 16 kcal/mol,
using MM (Discover, Insight2000.1, Accelrys, San
Diego) and semiempirical (MOPAC/PM3) optimization
methods, respectively. Neither the chlorine substituent
of 5b nor the methyl substituent of 5c affected the
energy inversion barrier values. On the other hand, the
pyrrolo[2,1-b]thieno[3,2-f][1,3]thiazepine system of 4
showed a lower conformational inversion energy barrier
(6.4 and 11.0 kcal/mol in MM and PM3 calculations,
respectively). Taken together, these results suggested
that the pyrrolo[1,2-b][2]benzazepine and pyrrolo[2,1-
b]thieno[3,2-f][1,3]thiazepine systems can undergo ring
inversion under physiological conditions, as happens for
many other tricyclic systems,9c resulting in A-fold and
B-fold types (negative and positive values of τ, respec-
tively; Table 3 in Supporting Information). We then did
a comprehensive conformational search, systematically
varying all rotatable bonds, to determine the corre-
sponding energy minima. The τN value required for
positioning the lone pair of the distal piperazine nitro-
gen to optimally fulfill the D2 pharmacophore was
established for both folds (BIO piperazine position;
Table 4 in Supporting Information). The resulting MM
conformers, as well as the BIO conformations, were used
as starting points for full semiempirical PM3 geometry
optimization. The first major result was that the BIO
conformations of 5a-c showed an increased energy
difference from the global minimum (∆EGM = 3.0 kcal/
mol; Table 4 in Supporting Information) with respect
to benzothiazepine-based derivatives such as 3a (∆EGM
= 1.8 kcal/mol in PM3 calculations8). This could be
explained by the following considerations. The pyrrolo-
[2,1-b][1,3]benzothiazepine system of 3a,b and the pyr-
rolo[1,2-b][2]benzazepine of 5a-c show the pyrrole
nitrogen at two different positions. In 3a,b the nitrogen
is closer to the bridged double bond (C9-C10) with
respect to the heterocyclic core system of 5a-c, and this
fact affects electronic conjugation between the vicinal
piperazine nitrogen (N1) and the bridged double bond.
The decreased degree of electronic conjugation found on
the 5a-c core system rendered the D2 bioactive con-
formation (BIO piperazine position) energetically dis-
favored. Accordingly, the BIO conformations of 4, in
which the pyrrole nitrogen is closer to the bridged
double bond, showed decreased conformational energies
(∼2.4 kcal/mol; Table 4 in Supporting Information). This
hypothesis is further supported by the analysis of charge
distribution on the PM3 optimized bioactive conforma-
tions of 3a, 4, and 5a-c and by τN values reported in
Table 4. Consequently, 5a-c were predicted to be on
the whole less potent at D2 receptors than benzothiaz-
epines. Our calculations showed that the BIO piperazine
position is equally energetically accessible for both folds
of the tricyclic systems of 4 and 5a-c (Figure 1; Table
4 in Supporting Information).
To provide drug therapies with prompter therapeutic
benefit for “resistant” schizophrenic patients, we decided
to investigate novel tricyclic systems with the aim of
developing an atypical drug candidate characterized by
improved efficacy (4 and 5a-c) and, presumably, meta-
bolic stability (5a-c). Here, we describe the design,
synthesis, and biological evaluation of novel benzazepine
antipsychotics, characterized by absence of the sulfur-
bridged atom of 3a,b sensitive to oxidative metabolism,
and their SARs for dopamine and serotonin receptor
affinity. Among the analogues synthesized and tested,
5a was selected for further biological investigation. We
discuss its development and a molecular modeling
study.
The set of compounds included in the molecular
modeling study was synthesized as described in Schemes
1 and 2 in the Supporting Information. The main goal
in the synthesis of the principal compound of the series,
5a, was the newly developed palladium-catalyzed three-
step pathway that provided the desired benzazepine
with high overall yield (Chart 1). The new tricyclic
compounds were subjected to binding experiments to
evaluate their affinity for rat serotonin and dopamine
receptor subtypes, and 5a was selected for thorough
investigation in vivo.
In a previous study, we demonstrated the possibility
of modulating the in vivo properties (typical/atypical)
of new potential tricyclic antipsychotics by specifically
adapting their potency toward D2 receptors.9a-c Our
strategy was based on the hypothesis that tricyclic
antipsychotics may interact with the D2 receptor active
site by adopting both folds of the tricyclic system and
that the chemical/physical properties of the tricyclic
skeleton can drive the binding mode determining which
aromatic ring is preferentially recognized as the “rel-
evant” one. To expand our SAR studies and to obtain
molecules with superior antipsychotic efficacy, we used
this hypothesis in designing the 4 and 5a-c (Chart 1),
based on a pyrrolo[2,1-b]thieno[3,2-f][1,3]thiazepine and
a pyrrolo[1,2-b][2]benzazepine skeleton, respectively.
Since early benzothiazepines, represented by 3a, may
undergo oxidative metabolism in vivo at the bridged-
sulfur atom and, as in the case of quetiapine and
octoclothepin, the oxidized metabolite may have weak
to negligible antipsychotic activity,9 the pyrrolo[1,2-b]-
[2]benzazepine system (5a-c) was investigated to ob-
tain potential antipsychotics characterized by higher
metabolic stability due to absence of the bridged-sulfur
atom. Calculated structural parameters reported in
Table 3 (Supporting Information) show that a butterfly-
like conformation of the tricyclic system, optimal for D2
receptor interaction,8 is maintained in both skeletons,
as confirmed by the X-ray structure of 5a (Figure 2,
Supporting Information). Starting from lead 3a, we
developed 4 and 5a-c rationally modifying several
structural features that can specifically affect D2 recep-
tor affinity, such as the distance between the centroid
of the relevant aromatic ring and the basic nitrogen.
To test the two possible binding modes for the new
In Figure 1 is reported the fit between the D2 receptor
3D pharmacophore and the newly developed com-
pounds, considering either the A- or the B-fold of the
tricyclic system, compared to the binding mode of the
reference 3b (Ki-D2 ) 0.43 nM8). When 5b binds D2
receptor adopting an A-fold, the chlorine atom favorably
interacts with the relevant polarized aromatic ring

Letters
Journal of Medicinal Chemistry, 2005, Vol. 48, No. 6 1707
Figure 1. Hypothetical D2 binding modes for 3b (ST1508), 4, and 5a-c. Structures are superimposed on the centroids of the
aromatic rings and a point 2.8 Å along the nitrogen lone pair vector. Nitrogens are in blue, carbons in green, chlorines in light-
green, sulfurs in yellow. Hydrogens were omitted except for the one protonating the distal nitrogen (white). ∆E ) EA-fold - EB-fold
.
Table 1. Binding Affinities for 5-HT2A, 5-HT2C, D1, D2, and D3 Receptors of 4 and 5a-c, and Reference Compounds
Ki ( SD,a nM
D1
pKi ratio
5-HT2A/D2
compd
R
R1
5-HT2A
D2
D3
5-HT2C
log Y b
4
2.15 ( 0.32
7.25 ( 1.2
1.35 ( 0.08
4.99 ( 1.1
4.0 ( 1.0
10.0 ( 1.0
0.65 ( 0.1
164 ( 22.0
4.91 ( 0.5
125 ( 12
8.40 ( 1.0
578 ( 46
25.0 ( 3.9
224 ( 23
-
-
1.07
1.30
1.24
1.38
1.17
1.21
1.18
0.82
6.70
3.20
3.91
2.33
4.69
3.89
4.98
9.14
5a
5b
5c
H
Cl
H
H
H
47.4 ( 2.6
230 ( 18
71.1( 12
24.2 ( 2.7
88.6 ( 23
39.0 ( 5.9
319 ( 45.0
8.30 ( 0.5
18.0 ( 1.5
3.8 ( 0.35
Me
923 ( 306
69.0 ( 17.0
250 ( 57.0
17.2 ( 4.5
4.80 ( 1.0
15.0 ( 1.2
1, olanzapine
2, clozapine
3a, ST1899c
haloperidol
85.0 ( 3.5
353 ( 35.0
19.7 ( 1.3
318 ( 59.0
-
-
-
-
a Values are the mean ( SD of three determinations and indicate the concentration giving half-maximal inhibition of [3H]ketanserin
(5-HT2A), [3H]SCH 23390 (D1), [3H]spiperone (D2), and [3H]mesulergine (5-HT2C) binding to rat tissue homogenate and [3H]-7-OH-DPAT
(D3) binding to Sf9 cell membranes. b The log Y score was calculated as in ref 8. c From ref 8.
pocket of the receptor but the piperazine ring is unfa-
vorably positioned; on the other hand, adopting the
B-fold, 5b is able to optimally orientate the piperazine
ring but completely lacks a favorable interaction with
the relevant aromatic ring pocket. Indeed, the dipole
determined by the fused pyrrole ring of 5b produced a
vector perpendicular to that given by the dipole of the
chloro-substituted benzo-fused system (Decipher, In-
sight2000.1, Accelrys). Adopting both folds, 5b (Ki-D2
) 71.1 nM, Table 1) could not optimally fulfill the D2
pharmacophore, as occurs for 3b (Ki-D2 ) 0.43 nM;
Figure 1). When the chlorine atom of 5b is replaced by
a hydrogen, the resulting 5a lost another D2 receptor
interaction point and 5a was predicted to have a lower
polarization of thiophene (Decipher, Insight2000.1, Ac-
celrys) is still favorable for D2 receptor interaction,
mimicking the chloro-substituted ring of 3b. As a
consequence, adopting a B-fold, 4 accommodates the
piperazine ring in the right orientation, still keeping
favorable π-π interactions with the relevant aromatic
ring pocket of the receptor (Figure 1). In line with our
hypothesis, 4 showed an increased D2 receptor affinity
(4, Ki-D2 ) 8.4 nM). 5a was also tested on 5-HT2A
recombinant human receptors ([3H]ketanserin) and
showed a Ki of 5.4 nM.
Benzazepine 5a was identified as a potential atypical
antipsychotic agent, and its pharmacological profile was
confirmed in vivo. Tests were performed using 5a as a
free base, using the same experimental conditions as
for clozapine, olanzapine, and 3a.8
D2 receptor affinity (5a, Ki-D2 ) 578 nM; 5b, Ki-D2
)
71.1 nM). Taking into account previous 3D-SAR studies
on a series of antipsychotics,9c we introduced a methyl
substituent at C1 of the pyrrole ring. As reported in
Figure 1, the resulting 5c has further unfavorable
Blockade of spontaneous locomotor activity, antago-
nism of climbing elicited by the direct dopamine agonist
apomorphine in mice, and reduction of conditioned
avoidance response (CAR) in rats are robust and repro-
ducible models, sensitive to D2 receptor antagonists, for
predicting therapeutic efficacy against positive symp-
features, assuming either an A- or B-fold (5c, Ki-D2
)
923 nM). On the other hand, when the benzo-fused ring
of 3a is replaced by an isosteric thiophene ring (4), the

1708 Journal of Medicinal Chemistry, 2005, Vol. 48, No. 6
Letters
Table 2. In Vivo Pharmacological Profile of 5a and Inhibition of Different Behavioral Responses after Oral Administration of the
Test and Reference Compounds
5-MeO-DMT-induced
head twitches
apomorphine
climbing
spontaneous
locomotor activity
CAR
µmol/kga
CAT
µmol/kga
compd
mg/kg
µmol/kga
mg/kg
µmol/kga
mg/kg
µmol/kga
mg/kg
mg/kg
1, olanzapineb
2, clozapineb
3a, ST1899b
5a
1.45
5.35
1.99
0.12
4.65
16.4
6.69
0.43
2.69
5.72
1.64
0.14
8.61
17.5
5.51
0.50
2.81
5.50
0.51
1.60
8.99
16.8
1.71
5.73
1.46
4.89
1.09
0.56
4.67
15.0
3.66
2.00
21.2
67.9
>100
>100
>100
>306
>336.2
>357.4 (>179c)
a Results are expressed as ED50 b From ref 8. c CAT/CAR for 5a.
.
toms. These behavioral assays have been employed in
vivo to demonstrate 5a dopamine antagonist activity
and indirectly its ability to interact with the mesolimbic
dopaminergic system, as predictive of 5a antipsychotic
potential. Rat spontaneous locomotor activity was sig-
nificantly reduced by 5a, with ID50 ) 1.60 mg/kg (Table
2). The ability to antagonize apomorphine-induced
climbing behavior in mice was also evaluated. Oral 5a,
prior to 1.3 mg/kg apomorphine challenge, caused dose-
related suppression of apomorphine-induced climbing
behavior (Table 2) with ED50 )0.50 µmol/kg, much lower
than 3a (ED50 ) 5.5 µmol/kg), olanzapine (ED50 ) 8.61
µmol/kg), and clozapine (ED50 ) 17.5 µmol/kg). In light
of these results, 5a was selected for further studies.
The antipsychotic potential of 5a was measured by
evaluation of the CAR in rats. In the shuttle-box test,
oral 5a suppressed CAR with no significant effects on
escape responses (data not shown). As shown in Table
2, 5a was more potent than clozapine, olanzapine, and
3a (ED50: 5a, 2.0 µmol/kg; clozapine, 15 µmol/kg;
olanzapine, 4.67 µmol/kg; 3a, 3.69 µmol/kg). Further-
more, 5a antagonizes 5-HT2A receptors in vivo. After sc
injection (6 min) of 5-methoxy-N,N-dimethyltryptamine
(5-MeO-DMT, 10 mg/kg, sc), we measured the number
of 5-MeO-DMT-induced head twitches for 15 min. 5a
was administered orally 60 min before 5-MeO-DMT, and
the data were compared with clozapine, olanzapine, and
3a. 5a antagonized 5-MeO-DMT-induced head twitches
In conclusion, these results demonstrate the rational
design of the novel and highly potent atypical antipsy-
chotic agent 5a (ST2329). Its receptor affinity profile
suggests a complex interaction with cortical receptors
involved in regulation of the activity of prefrontal
cortical cells innervated by ventral tegmental area
neurons, like 5-HT2A, and dopaminergic receptor sub-
types. The 5-HT2A/D2 ratio (expressed by log Y ) 3.20)
is particularly favorable, and preliminary in vivo studies
confirmed pharmacological effects superior to those of
olanzapine and clozapine on 5-OMe-DMT-induced head
twitches, apomorphine-induced climbing, and spontane-
ous locomotor activity in animals, CAT, and CAR. 5a
displayed atypical antipsychotic activity at 0.56 mg/kg,
an oral dose much lower than that of olanzapine and
3a. The atypical antipsychotic 5a, characterized by
improved efficacy and better metabolic stability than
3a and olanzapine and obtained by an elegant three-
step synthesis, was selected for thorough pharmacologi-
cal investigation as a potential clinical candidate.
Acknowledgment. This study was supported by
grants from MIUR (Rome, Grant PRIN 2003) and
Sigma-Tau (Pomezia). The authors thank Dr. Gianluca
Giorgi, University of Siena, for the X-ray analysis of 5a.
The authors also thank Mrs. Helen Ampt for manuscript
editing.
Supporting Information Available: Experimental de-
tails for 4 and 5a-c, Tables 3 and 4, Figure 2, and Schemes
1 and 2. This material is available free of charge via the
Internet at http://pubs.acs.org.
at doses much lower than that of olanzapine (ED50
)
0.43 and 4.65 µmol/kg, respectively). Higher doses of
clozapine and 3a were necessary to achieve the same
effect (ED50: 3a, 6.7 µmol/kg; 2, 16.4 µmol/kg). 5a also
has a limited propensity to elicit catalepsy (CAT). The
degree of catalepsy is often used to predict the incidence
of extrapyramidal motor disorders. 5a was administered
orally to rats at doses of 100 mg/kg, and the catalepsy
response was evaluated after 60, 120, 180, 240, and 300
min. Like clozapine (>100 mg/kg, >306 µmol/kg, po) but
unlike olanzapine, which under the same experimental
conditions induced catalepsy in 50% of animals 240 min
after oral administration of 21 mg/kg (ED50 ) 67.9 µmol/
kg),8 5a had low cataleptogenic potential (ED50: >100
mg/kg, >357.4 µmol/kg). Thus, compared to olanzapine,
5a has a higher threshold for inducing catalepsy that
may by analogy translate into lower clinical EPS li-
ability. This limited propensity to elicit catalepsy results
in a large spread between doses, possibly inducing
catalepsy and blocking the avoidance response (CAT/
CAR > 179). These results suggest a preferential ability
of 5a to modulate mesolimbic instead of nigrostriatal
dopaminergic neurotransmission, highlighting its atypi-
cality and low propensity to induce unwanted extra-
pyramidal motor disturbances at therapeutically useful
doses.
References
(1) Work Group on Schizophrenia. Practice guidelines for the
treatment of patients with schizophrenia. Am. J. Psychiatry
1997, 154, 1-63.
(2) Wickelgren, I. A route to treating schizophrenia? Science 1998,
281, 1264-1265.
(3) Carlsson, A. The current status of the dopamine hypothesis of
schizophrenia. Neuropsychopharmacology 1988, 1, 179-186.
(4) Campbell, M.; Young, P. I.; Bateman, D. N. The use of atypical
antipsychotics in the menagement of schizophrenia. Br. J.
Pharmacol. 1999, 47, 13-22.
(5) Kelleher, J. P.; Centorrino, F.; Albert, M. J.; Baldessarrini, R.
J. Advances in atypical antipsychotics for the treatment of
schizophrenia. CNS Drugs 2002, 16, 249-261.
(6) Koller, E. A.; Doraiswamy, P. M. Olanzapine-associated diabetes
mellitus. Pharmacotherapy 2002, 22, 841-852.
(7) Web site: http://www.abpi.org.uk/publications/publication_de-
tails/targetSchizophrenia/section5_asp.
(8) Campiani, G.; Butini, S.; Fattorusso, C.; Catalanotti, B.; Gemma,
S.; Nacci, V.; Morelli, E.; Cagnotto, A.; Merenghetti, I.; Mennini,
T.; Carli, M.; Minetti, P.; Di Cesare, M. A.; Mastroianni, D.;
Scafetta, N.; Galletti, B.; Stasi, M. A.; Castorina, M.; Pacifici,
L.; Vertechy, M.; Di Serio, D.; Ghirardi, O.; Tinti, O.; Carminati,
P. Pyrrolo[1,3]benzothiazepine-Based Serotonin and Dopamine
Receptor Antagonists. Molecular Modeling, Further Structure-
Activity Relationship Studies, and Identification of Novel Atypi-
cal Antipsychotic Agents. J. Med. Chem. 2004, 47, 143-157, and
references therein.
(9) Web site: http://www.cc.nih.gov/phar/updates/99sept-oct.htm.
JM049629T