Pharmacophore Modeling and Three Dimensional Database Searching for Drug Design Using Catalyst: Recent Advances

Abstract

Perceiving a pharmacophore is the first essential step towards understanding the interaction between a receptor and a ligand. Once a pharmacophore is established, a beneficial use of it is 3D database searching to retrieve novel compounds that would match the pharmacophore. As the 3D searching technology has evolved over the years, it has been effectively used for lead optimization, combinatorial library focusing, as well as virtual high-throughput screening. This paper is an update to the original paper published in this journal earlier: Kurogi, Y, and Guner, O. F. "Pharmacophore Modeling and Three-Dimensional Database Searching for Drug Design Using Catalyst," in Current Medicinal Chemistry, 2001, 8(9), 1035-1055.

Full text

Current Medicinal Chemistry, 2004, 11, 763-771 1
0929-8673/04 $45.00+.00 © 2004 Bentham Science Publishers Ltd.
Pharmacophore Modeling and Three Dimensional Database Searching for
Drug Design Using Catalyst: Recent Advances
Osman Güner,
†
Omoshile Clement,
†
and Yasuhisa Kurogi
‡
†
Accelrys, 9685 Scranton Road, San Diego, CA 92121, USA
‡
Cambridge Isotope Laboratories, Inc., 50 Frontage Road, Andover, MA 01810, USA
Abstract: Perceiving a pharmacophore is the first essential step towards understanding the interaction between
a receptor and a ligand. Once a pharmacophore is established, a beneficial use of it is 3D database searching to
retrieve novel compounds that would match the pharmacophore. As the 3D searching technology has evolved
over the years, it has been effectively used for lead optimization, combinatorial library focusing, as well as
virtual high-throughput screening. This paper is an update to the original paper published in this journal
earlier: Kurogi, Y, and Güner, O. F. "Pharmacophore Modeling and Three-Dimensional Database Searching for
Drug Design Using Catalyst,” in Current Medicinal Chemistry, 2001, 8(9), 1035-1055.
RECENT ADVANCEMENT IN LIBRARY
ENUMERATION AND MULTICONFORMATION
DATABASE BUILDING USING CATDBLIBRARY
CatDBLibrary
This is a new technique for rapid enumeration of large
combinatorial libraries based on Markush structures, and
conversion to 3D multi-conformational searchable databases.
Details of the algorithm and applications will be published
elsewhere.[1]
RECENT STUDIES
1.1 Cyclooxygenase-2 (COX-2) Selective Inhibitors
Palomer and co-workers[2] employed Catalyst/HipHop’s
modeling approach[3] to investigate selectivity and activity
profiles of five COX-2 inhibitors of the phenylsulfonyl
tricyclic series (Chart 1). A 3D pharmacophore model for
COX-2 inhibition as deduced from common-feature
alignments of SC-558 2, Rofecoxib 3 [4],
DFU 4, and 5, has the following geometric features: a
sulfonyl S atom, an aromatic ring with a fixed position
orthogonal to the plane, and a second aromatic ring; both
aromatic rings lie at 290
0
± 10
0
dihedral angle from each
other. Including information derived from X-ray structures of
SC-558-bound COX-2 complex[5] with respect to the non-
conserved residue 523 important to COX-2 inhibition, the
ligand-based pharmacophore model was enhanced by placing
an exclusion volume inside one of the aromatic rings to
mimic location of the valine 523 side chain γ position (see
Fig. (1)).
The derived pharmacophore model was docked into the
active site of a known COX-2 structure containing the non-
selective NSAID Indomethacin (6; Chart 1), and found to be
consistent with known binding requirements in this
complex. Using this approach led to the identification of a
*Address correspondence to this author at the Accelrys, 9685 Scranton
Road, San Diego, CA 92121, USA; E-mail: osman@ accelrys.com
new, potent and selective COX-2 inhibitor 8 (IC
50
= 0.65
µM).
1.2 Mycobacterium avium Complex Dihydrofolate
Reductase (MAC DHFR)
The bacterial microorganism, Mycobacterium avium
complex (MAC), is a leading cause of disseminated
infections in 50-70% of patients with acquired
immunodeficiency syndrome (AIDS) [6-13]. These
organisms have significant resistance to most antibiotics and
anti-mycobacterial agents, hence an urgent need to identify
novel anti-mycobacterial agents targeted to these organisms.
Debnath and co-workers [14] applied Catalyst/Hypogen
3D QSAR modeling program[15] to develop pharmacophore
models based on 2,4-diamino-5-deazapteridine inhibitors of
Mycobacterium avium complex (MAC) and human
dihydrofolate reductase (hDHFR) (see Chart 2).
For this class of inhibitors, Catalyst/Hypogen yielded a
four-feature pharmacophore model - two hydrogen bond
acceptors (HBA), one hydrophobic feature (HYD) and a ring
aromatic feature (RA). To validate this model, the test
compounds were partitioned into three classes: very active,
moderately active, and inactive. Using this classification
scheme, the Hypogen models were able to correctly classify
100% of the training set compounds, and more importantly,
92% of the test compounds were also correctly classified.
This model was then employed against three known MAC
DHFR inhibitors with successful matching of ligand-target
binding features.
1.3 Non-Steroidal 5αα
αα
-Reductase Inhibitors
The enzyme 5α-Reductase has been linked to the
pathogenesis of benign prostate hyperplasia (BPH), prostatic
cancer and androgenic alopecia, acne and hirsutism [16], and
is also involved in the biosynthetic pathway for reduction of
testosterone (T) to dihydrotestosterone (DHT). Because this
enzyme is membrane-bound, there is no known crystal
structure, preventing a structure-based lead generation
approach. Ligand-based molecular modeling using

2 Current Medicinal Chemistry, 2004, Vol. 11, No. 6 Güner et al.
N
N
S
X
CF
3
O
H
2
N
O
O
S
O
H
3
C
O
R
R
O
CH
3
O
Cl
O
H
3
C
COOH
H
3
C
F
COOH
N
N
S
O
H
2
N
O
S
CF
3
1: X = CH
3
(Celecoxib)
2: X = 1 (SC-588)
3: R = H (Rofecoxib)
4: R = CH
3
(DFU)
5
6
7
Chart. (1). Reference compounds used in generating pharmacophore model for COX-2 inhibition [2].
Fig. (1). Catalyst-generated pharmacophore model for COX-2 inhibition aligned with protein-bound structure of SC-558 (green), and
a novel computer-designed COX-2 lead (red) [2].
pharmacophore models has been employed by Chern and co-
workers [17] to identify new inhibitors of the rat 5α-
Reductase enzyme.
N
Cl
S
O
O
O
O
8
In the study by Chern and co-workers [17], predictive 3D
pharmacophore models were generated from inhibitors of the
enzyme 5α-Reductase using the Catalyst/ Hypogen program.
The pharmacophore model obtained contained two hydrogen
bond acceptors (HBA) and three hydrophobic (HYD)
features. This model was aligned with a selected number of
structurally diverse active compounds, as well as reported
structures shown to have high potency against this enzyme.
It was shown that many of these test compounds fit the
pharmacophore features quite well.
The search for new inhibitors of this enzyme was
extended to isoflavones due largely to their promiscuous
affinities toward many biological targets. The isoflavones
exhibit activities such as phytotherapeutics, phytoalexins,
antimicrobials, oestrogenics, antioxidants, nerve growth
stimulant, and insecticides. Using the same pharmacophore
model against a small library of 151 isoflavonoids curated

Modeling and Three Dimensional Database Searching for Drug Design Using Catalyst Current Medicinal Chemistry, 2004, Vol. 11, No. 6 3
N
NN
N
H
R
2
R
3
R
4
R
5
R
6
R
1
NH
2
H
2
N
N
NN
N
H
R
2
R
1
NH
2
H
2
N
N
NN
X
R
2
R
3
R
4
R
5
R
6
R
1
NH
2
H
2
N
Chart. (2). Analogs of 2,4-diamino-5-deazapteridine compounds employed in pharmacophore modeling investigation of MAC DHFR
[14].
O
O
O
OO
OMe
OMe
O
O
O
O
OB n
OBn
OMe
O
O
O
OBn
OB n
OM e
O
O
O
OMe
OMe
OM e
O
O
OH
OH
OH
O
O
OOH
O
OH
HO
O
OOH
O
OH
O
HO
OH
O
O
OH
HO
O
OH
OH
48.7 26.1 µM
31.0 22.7 µM
27.4 11.9 µM
31.0 15.8 µM
15.7 8.9 µM
23.1 22.3 µM
6.9 4.4 µM
26.2 15.2 µM
Chart. (3). Structures of new isoflavone derivatives identified via pharmacophore model search, and experimental IC
50
values
measured against rat 5α-Reductase [17].
from the NCI database, these workers [17] identified 37
compounds. Biological testing of some of these compounds
(see Chart 3) yielded low micromolar affinities for rat 5α-
Reductase, suggesting their potentials as a new class of 5α-
Reductase inhibitors [17].
1.4 Mesangial Cell Proliferation Inhibitors
Proliferation of mesangial cells (MC) are implicated in
glomerular diseases including IgA nephropathy,
membranoproliferative glomerulonephritis, lupus nephritis

4 Current Medicinal Chemistry, 2004, Vol. 11, No. 6 Güner et al.
N
S
N
N
H
O
P
O
OO
O
N
N
H
O
P
O
OO
Cl
N
S
N
H
O
P
O
OO
O
O
N
H
O
P
O
OO
9
10
11
12
Chart. (4). Training set of four mesangial cell proliferation inhibitors used as input to the Catalyst/HipHop program in generating the
chemical feature-based pharmacophore model [19].
Fig. (2). Conformation of 10 representing the best flexible fit to the Catalyst-generated pharmacophore model for MC proliferation
inhibition; fit value = 7.0 [19].
and diabetic nephropathy [18]. Inhibitors of MC
proliferation offer therapeutic treatments for these diseases,
but to date no known selective inhibitor has been reported
[19]. Kurogi and co-workers [19] employed a molecular
modeling approach to identify novel MC proliferation
inhibitors using the Catalyst/HipHop program. A training
set of four MCP inhibitors (see Chart 4) was used to build a
common-feature pharmacophore model. The pharmacophore
model derived from these compounds contained seven
chemical features: three hydrogen bond acceptors (HBA) and
four hydrophobic groups (HYD). This model was aligned
with compound 10 using best fit (Chart 4) and found to give
the highest fit value of 7.0 (see Fig. (2)). The model was
used as a 3D search query against a Catalyst/Maybridge 3D
database, which led to the identification of four new MCP
inhibitors. More importantly, these new compounds were
found to only weakly inhibit normal cells, in contrast to the
high toxicity exhibited by the training set compounds. Of
the four new candidates identified in the pharmacophore
search of the Maybridge database, compound 13 was found
to be the most promising selective MCP inhibitor [19].
1.5 Candida Albicans MyristoylCoA: Protein N-
Myristoyltransferase Inhibitors
Myristoyl CoA: protein N-myristoyltransferase (NMT) is
a cytosolic monomeric enzyme, which catalyzes the transfer
of cellular fatty acid myristate from myristoylCoA to the N-
terminal glycine amine of several eukaryotic proteins [20].
Genetic studies have established a link between NMT and
growth and survival of organisms that cause systemic fungal
infections in immuno-suppressed humans [21-23].
Mutagenesis, depeptidisation and structure-specificity
experiments were conducted to identify the essential
functionalities for molecular recognition of NMT [24]. This
led to the identification of a number of highly potent

Modeling and Three Dimensional Database Searching for Drug Design Using Catalyst Current Medicinal Chemistry, 2004, Vol. 11, No. 6 5
H
2
N
H
N
N
H
H
N
N
H
H
N
N
H
H
N
NH
2
CH
3
O
O
O
OH
CH
3
O
OH
O
NH
2
O
O
OH
O
14
peptide-based inhibitors of this enzyme such as 14.
However, as most peptide-based inhibitors, compounds like
14 have poor in vitro antifungal activity, hence the need to
identify non-peptidic NMT inhibitors.
O
ON
H
N
H
N
O
OO
13
In the search for non-peptidic NMT inhibitors, Kulkarni
and Karki [25] employed pharmacophore-based 3D QSAR
modeling available in the Catalyst/Hypogen program, to
evaluate 44 peptidic NMT inhibitors with varying activities.
Several hypotheses were constructed and tested against the
set of peptidic NMT inhibitors. The pharmacophore model
chosen contained four chemical features – two hydrogen
bond acceptors (HBA), one hydrophobic group (HYD) and
one positive ionizable group (PI). This model was compared
with the crystal structure of NMT [26] and found to closely
mimic the binding requirements for NMT-inhibitor
interactions (see Table 1).
Table 1. Comparison of Pharmacophore Model Features
Derived from Peptidic NMT Inhibitors by
Catalyst/Hypogen and NMT Active Site Residues
[25]
Pharmacophore Features Active site residues
1. Hydrophobic Met-454, Thr-205, Asn-169
2. HBA1 Gly-418, His-221, Asp-417
3. HBA2 Asp-417, Gly-416
4. Positive Ionisable (PI) Asp-417, Asp-108, Asp-106
1.6 5-HT
3
Partial Agonists
The biological functions of 5-HT receptors are poorly
understood, however evidence suggests that these receptors
may be implicated in depression [27], control of circadian
rhythms [28], and muscle relaxations [29]. Currently, there
are about fourteen serotonin receptor subtypes, and interest
in the 5-HT
3
is derived primarily because this receptor has
been linked to many important physiological impairments
such as pain, memory loss, depression, anxiety, drug
addiction and psychosis. It is expected that agonists and
antagonists of this receptor will be useful in treating these
impairments [30].
Rault and co-workers first described a pharmacophore
model for partial agonists of the serotonin 5-HT
3
receptor
[31]. The serotonin 5-HT
3
partial agonist model was derived
exclusively using the Catalyst pharmacophore modeling
program. This approach considered chemical functions
common to compounds in the training set and their
correlations with activity. It used pattern recognition across
the diverse set of conformations of the input ligands, in
identifying a common set of chemical features. However,
this method may have some drawbacks when applied to a
system with very subtle chemical differences. Additional
computational work was required to ascertain if this model
was truly representative of the binding constraints for partial
agonism against this receptor subtype.
In the follow-up study [32], a combination of 3D QSAR
techniques, Catalyst/ Hypogen and CoMFA, as well as
conformational quality assessment from molecular
dynamics, semi-empirical methods (MNDO94, PM3), and
X-ray crystallographic information, were all applied to
evaluate structure-activity correlations for 75 serotonin 5-
HT
3
ligands. The study concluded that (a) the previously
determined pharmacophore model for 5-HT
3
partial agonists
was correct, (b) each chemical feature identified by the
Catalyst-generated pharmacophore model are also found in
the CoMFA model, and (c) the combination of these
approaches provided further validation that the
pharmacophore model derived for 5-HT
3
partial agonists is
indeed very descriptive of binding requirements at this
receptor active site.
1.7 5-HT
7
antagonists
There are about fourteen serotonin receptor subtypes, and
5-HT
7
is one of the recently identified subtypes [33]. There
already exists three known 5-HT
7
receptor antagonists – SB-
258719 [34], SB-269970, and DR4004 [35], which were
discovered largely via high throughput screening (HTS)
experiments.

6 Current Medicinal Chemistry, 2004, Vol. 11, No. 6 Güner et al.
Table 2. Matrix Distances (Å) Characterizing 5-HT Pharmacophore Model Derived [41] by Catalyst/Hypogen
Aromatic Ring
a
Feature Hydrophobic gp Basic center PI-1 PP-2 PI-1 PP-2
Hydrophobe
Basic center 10.6
Aromatic ring PI-1 4.1 6.6
Aromatic ring PP-1 4.2 8.3 3.0
Aromatic ring PI-2 8.5 4.1 5.1 5.3
Aromatic ring PP-2 8.6 4.8 4.9 5.1 3.4
a
PI = initial point (heavy atom location) for the ligand; PP = projection point on the receptor
The first pharmacophore model for 5-HT
7
antagonists
was recently described by López-Rodríguez and co-workers
[36], using the Catalyst/Hypogen program. Thirty
compounds culled from literature sources and in-house
synthesized compounds were subjected to pharmacophore-
based 3D QSAR modeling. The derived pharmacophore
model contained four chemical features in a specific 3-
dimensional orientation. As deduced by the Catalyst
program, the minimally-required features important to 5-HT
7
antagonism are: a hydrophobic group (HYD), a positively
charged center (P+), a hydrogen bond acceptor (HBA) and a
ring aromatic group (RA).
N
N
N
O
15
As a predictive model, this model gives a correlation
coefficient r = 0.921 for the fourteen training compounds.
To validate this model, a set of naphtholactams and
naphthosultams (Chart 6) with measured affinities against 5-
HT
7
, were used as test compounds. A number of compounds
mapping to all features in the pharmacophore model were
found to be fairly potent (pKi > 6.5) while compounds
missing one or more features in the model showed moderate
inactivity (pKi < 6.5), further suggesting that the derived
pharmacophore model provides a good approach to
understanding 5-HT
7
antagonism. Of the active compounds
in the test set, compound 15 was selected as the most
promising antagonist for this receptor subtype.
1.8 Inhibition of 5-HT Serotonin Reuptake
Several studies have appeared in the literature describing
pharmacophoric patterns for inhibition of 5-HT serotonin
reuptake [37-40]. These models provide a topological
description of the required chemical features for this process.
A recent study by Rault and co-workers [41] described a
pharmacophore-based 3Q QSAR modeling of a diverse set of
serotonin reuptake inhibitors, in the search for new ligands
possessing unique or multiple controlled affinities across
two serotonin subtypes – 5-HT and 5-HT
3
[41]. Using data
available from the literature, an activity-based
pharmacophore model was derived by the Catalyst/Hypogen
program. The model contained four geometric features – a
hydrophobic group, a basic amino group, and two aromatic
rings. This model is in agreement with that proposed by
Rupp and co-workers [39]. The distance constraints between
these features are provided in Table 2 below.
A test set of 28 compounds were used to evaluate the
predictive ability of this model. Of special interest was how
this model can account for the stereochemical difference in
the R and S-configurations of citralopram 16, a commercially
available selective 5-HT serotonin reuptake inhibitor.
Alignment of the 5-HT pharmacophore model with
citralopram 16, found that S-citralopram had a higher fit
score (and hence potentially more active) than R-citralopram,
in accord with findings that this drug is mainly active in its
S-form [42]. The ability of the hypothetical model to
differentiate between stereoisomers and its effect on binding
affinity makes this model quite useful in designing novel 5-
HT selective inhibitors.
The study also compared pharmacophore models derived
for inhibitors of 5-HT serotonin reuptake [41], and partial
agonists of 5-HT
3
[31b]. Both models have a basic amine
center and an aromatic ring in common.
O
F
N
CH
3
H
3
C
NC
16
To tune the 5-HT model towards more selective ligands,
it is imperative that there be two orthogonal ring aromatics.
To modulate this requirement, a genetic function-based
QSAR model was derived using various 1D, 2D as well as
3D descriptors [44]. Spatial descriptors of the Jurs shadow
indices were found repeatedly in the multiple models

Modeling and Three Dimensional Database Searching for Drug Design Using Catalyst Current Medicinal Chemistry, 2004, Vol. 11, No. 6 7
S
N
N
O
N
H
S
N
N
O
N
H
S
N
N
O
NH
2
17
18
19
Chart. (5). Structures of 5-HT selective ligands (17 and 18) newly identified in the study [35].
N
N
N
O
O
N
N
N
(CH
2
)n
Cl
O
X
N
N
N
O
O
NN
N
O
(CH
2
)n
X
Fig. (3). Structural motifs of pyridazinones used in pharmacophore modeling of α
1
AR antagonists [49].
generated by this approach, underscoring their importance to
modulating inhibition of this receptor. Of the multiple
models derived from this method, those with three variables
have r
2
= 0.84, and s = 0.52 (n = 19), while the four
variable models are statistically improved with r
2
= 0.90,
and s = 0.43 (n = 19).
To take advantage of the differences between the
pharmacophore model derived for 5-HT and 5-HT
3
in the
design of more selective 5-HT serotonin reuptake inhibitor,
Rault and co-workers [41] investigated compounds 17 and
18 with oriented aromatic rings. These compounds are
analogs of the non-selective but highly potent serotonin
reuptake inhibitor 19 (Chart 5). Fit scores computed for the
alignment of these compounds against the 5-HT and 5-HT
3
models was found to be highest for the E-configuration of
17. Based on this finding, 17 was synthesized and found to
have good affinity and selectivity for the 5-HT receptor. This
study also revealed that the crucial basic amine center should
be a secondary amine rather than a primary amine. This
finding led to the synthesis and testing of the racemate of
18. The measured experimental affinity was in close
agreement with predictions by the model [41].
1.9 αα
αα
1
- and αα
αα
2
- Adrenergic Receptor Antagonists
Adrenergic receptors (ARs) are membrane-bound proteins
and members of the G-protein coupled receptor superfamily.
They mediate many effects of the sympathetic nervous
system, and are activated by catecholamines, adrenaline and
noradrenaline [45]. There are three distinct α
1
-AR subtypes:
α
1A
, α
1B
and α
1D
[46], and four subtypes for α
2
-ARs:
α
2A
,
α
2B
, α
2C
, and α
2D
[47]. Cloning studies [48] have
shown that α
1
and α
2
-ARs have many features in common
hence the similarity in their mechanism of action. In the
search for new ligands with affinity and selectivity for α
1
-
ARs relative to α
2A
ARs, Botta and co-workers [49]
conducted a ligand-based pharmacophore modeling technique
in order to identify new ligands satisfying both of these
requirements.
In the study, a training set consisting of analogs of
Pyridazin-3(2H)-one (see Fig. (3)), supplemented by
literature data to accommodate structural diversity and higher
range of activity, were used. The Catalyst/Hypogen-derived
model contained five chemical features – a positive ionizable
group, three hydrophobic groups, and a hydrogen bond
acceptor, with r = 0.92 and rmsd = 0.89 (n = 24). This
model successfully predicted the binding affinities for a large
range of compounds from both the training set and a test set.
The test compounds are shown in Chart 6. Comparative
results between experimental and predicted activities for
these compounds are collected in Table 3.
1.10 Tyrosine Kinase Inhibitors
Computational approaches to drug design often involve a
combination of techniques and tools. This provides a
complementary approach to identifying and optimizing
novel lead candidates for any target of interest. The

8 Current Medicinal Chemistry, 2004, Vol. 11, No. 6 Güner et al.
N
H
N
N
H
O
O
N
N
OMe
OMe
R
N
N
(CH
2
)n
N
N
Ph
O
O
OMe
RN
N
(CH
2
)n
N
N
Ph
O
NH
2
COCH
3
N
N
(CH
2
)
4
N
O
OMe
N
N
OMe
N
H
O
N
N
(CH
2
)
4
OMe
N
O
O
N
N
OMe
N
H
O
N
N
(CH
2
)
4
N
O
N
N
N
N
OMe
O
N
N
NH
2
O
CH
3
Ph
N
N
(CH
2
)
4
OMe
N
O
O
MeO
MeO
N
N
NH
2
N
N
O
O
20
25: n=2, R=Cl 26: n=2, R=H
27: n=3, R=Cl 27: n=1, R=H
21: n=1, R=H 22: n=2, R=H
23: n=3, R=H 24: n=1, R=H
29
31
33
35
30
32
34
36
Chart. (6). Structures of α
1
-adrenoceptor antagonists culled from the literature and used as training set in pharmacophore generation
[49].
combination of pharmacophore modeling (Catalyst),
molecular fields analysis (CoMFA) and molecular similarity
indices analysis (CoMSIA), were recently applied by Xu and
co-workers to quantitate ligand-binding requirements for the
receptor tyrosine kinases [50].
The study involved two classes of compounds –
benzylidene malononitriles, and indolin-2-ones (see Chart 7)
that have shown strong inhibitory potency against the
human epidermal growth factor receptor-2 (HER2), a
member of the tyrosine kinase receptor family [51, 52]. It is
well known that for a reliable and effective application of
CoMFA or CoMSIA, the precursor alignments must be
precise [53]. Molecular alignments were performed using
simple atom-by-atom overlays (rigid alignment), and also by
a feature-based alignment generated using the Catalyst
software.
In the pharmacophore elucidation step, a total of 31
compounds – 19 malononitriles and 12 indolin-2-ones –
were submitted to the Catalyst/Hypogen program. The
model obtained contained four chemical features – one
hydrogen bond acceptor, one hydrogen bond donor, an
aliphatic hydrophobic group, and an aromatic hydrophobic

Modeling and Three Dimensional Database Searching for Drug Design Using Catalyst Current Medicinal Chemistry, 2004, Vol. 11, No. 6 9
Table 3. Actual and Calculated αα
αα
1
-Adrenergic Receptor Binding Affinities for Arylpiperazines Collected from Literature
Sources, as Calculated by Catalyst Hypothesis Model [49]
Ki / nM Ki / nM
Compound Actual calculated Compound actual calculated
20 0.21 0.11 29 0.34 0.36
21 2395 1200 30 374 140
22 24.6 15 31 258 380
23 46.7 11 32 516 620
24 42.7 11 33 0.8 0.46
25 15.6 9.2 34 11.9 19.0
26 147 68 35 17.5 6.7
27 7.9 0.86 36 0.74 9.8
28 2308 1000
CN
R
1
O
R
2
HO
R
3
O
S
N
N
N
N
O
N
O
R
1
N
O
NNCHO
NO
N
R
2
R
2
R
1
E-c onformer
Z-conform er
Chart. (7). Structures of benzylidene malononitriles and indolin-2-ones used in the study [50].
feature. The training set compounds were aligned onto this
model, and the aligned molecules were submitted to
CoMFA and CoMSIA experiments to generate the predictive
3D molecular field- and molecular similarity based models.
The results obtained from the CoMFA and CoMSIA
experiments [50] based on Catalyst-generated alignments are
noteworthy: (a) with only steric and electrostatic fields, the
CoMFA and CoMSIA statistics yielded q
2
≤ 0.656, (b)
including hydrogen bond acceptor and donor, lowers the
statistical significance q
2
to ≤ 0.575, (c) however, including
hydrophobic properties in the field analysis improved the
model with q
2
≥ 0.70. The overall best models were those
derived from CoMSIA which included fields, electrostatics,

10 Current Medicinal Chemistry, 2004, Vol. 11, No. 6 Güner et al.
O
N
N
O
O
H
3
C
R
2
R
3
R
1
CH
3
O
N
N
O
OR
1
H
3
C
R
2
R
3
R
4
ON
N
O
O
H
3
C
R
1
CH
3
ON
N
O
R
1
O
H
3
C
R
2
ON
N
H
3
C
O
OMe
Chart. (8). Sendai Virus inhibitors based on uracil, pyrimidinone and uridine derivatives [57].
hydrogen bond donors and acceptor and hydrophobic
properties. This supports the feature-based model derived by
the Catalyst program, which included many of these same
features considered important for this class of protein
tyrosine kinase inhibitors.
1.11 Inhibitors of Parainfluenza 1 (Sendai) Virus (PIV)
Sendai virus (SV) is a murine subtype of parainfluenza 1
viruses (PIVs), and a member of the family of
Paramyxoviridae. PIVs are important human pathogens
implicated in upper and lower respiratory tract infections
[54, 55]. As part of a search for antiviral agents, a general
anti-HIV screening of a series of 6-(oxiranylmethyl)uracil
analogs identified several of these compounds with low to
non-existing toxicities, and strong inhibitory activity against
the Sendai virus screen [56]. In order to ascertain the binding
requirements for SV inhibition, derivatized uracils,
pyrimidinones, and uridines with potentially high affinity
and selectivity towards SV, were synthesized and tested
against this virus. Examples of these compounds are listed
in Chart 8.
Following the syntheses and biological testing of these
compounds, a pharmacophore-based evaluation of the SAR
in this class of new antiviral agents was performed to probe
the binding site on SV. To perform this experiment, a set of
22 SV inhibitors spanning an activity spectrum ~3.7 log
units, was used as input to the Catalyst/Hypogen program.
The selected model from the ensemble results provided by
Catalyst contained four geometric features – three hydrogen
bond acceptors, and a hydrophobic group. This model
estimated activities of both training and test compounds
close to their experimental values. In addition, the model is
stereo chemically intuitive in its ability to differentiate
between geometric isomers, as demonstrated by compound
37 (see Fig. (4)). This further demonstrates the utility of the
model in identifying geometric features important to
inhibition of SV replication.
O
N
N
O
O
H
3
C
CH
3
Ph
O
N
N
O
O
H
3
C
CH
3
Ph
 

 (Z)-37
 (E)-37
(ED
50
(expt) 0.38 µM)
(ED
50
(calc) 0.57 µM)
(ED
50
(calc) 160 µM)
(ED
50
(expt) 193 µM)
Fig. (4 ). Experimental and predicted activities using
pharmacophore model derived for Sendai Virus inhibitors [57].
1.12 Inhibitors of Rhinovirus Replication
Absorption, along with distribution, metabolism and
excretion, plays a very significant role in the clinical efficacy
of any drug candidate. Predicting drug absorption is one of
the major challenges of computational chemistry in the 21
st
century, and elucidation of this problem will greatly impact
the cost of drug development as lead candidates that will fail
can be quickly identified and eliminated [58]. The number of
recent publications addressing this area of interest attests to
its great importance [58, 59].
N
N
R
2
NH
2
R
3
R
1
38
Although limited in scope, Caco-2 cells are a useful in
vitro assay system for predicting human absorption [60].
Ekins and co-workers [61] have applied four QSAR
modeling approaches - Catalyst/Hypogen, CoMFA,

Modeling and Three Dimensional Database Searching for Drug Design Using Catalyst Current Medicinal Chemistry, 2004, Vol. 11, No. 6 11
O
O
HO
O
O
O
O
HO
O
NRB 03849
IC
50
562 nM
NRB 13785
IC
50
616.5 nM
NR B 03731
IC
50
178.2 nM
NRB 03689
IC
50
56.0 nM
NR B 03742
IC
50
660.25 nM
O
O
O
O
N
N
H
N
O
H
H
Chart. (9). Structures of new steroidal compounds identified via pharmacophore model search, and experimental IC
50
values
measured against Human CYP17 [64].
VolSurf, and GFA optimized MS-WHIM, to study the
passive intestinal absorption for a set of structurally
homologous 2-amino benzimidazoles (38), which are potent
inhibitors of rhinovirus and enterovirus replication [61].
HO
N
N
39
Each of the 3Q QSAR modeling technique was used to
generate “permeability” models. The techniques were
evaluated on the basis of their ability to yield the most
predictive model. Of the 28 analogs of the 2-
aminobenzimidazoles (38), 19 were used to build the
models, and 9 were employed as a test set to validate the
models. The Catalyst “permeability pharmacophore”
contained three geometric features – a hydrogen bond donor,
a hydrogen bond acceptor and an aromatic ring. Such a
simplistic model of three required binding features (training
set: r
2
= 0.83; test set: r
2
= 0.94) was found to be the most
predictive among the four techniques applied in the study, in
terms of the correlation between observed and calculated
affinities. The CoMFA model, which also consisted of three
regions (training set: r
2
= 0.96; test set: r
2
= 0.83), and
VolSurf (training set: r
2
= 0.76; test set: r
2
= 0.83) were
equally successful in building predictive models, although
the correlation with experimental affinities was slightly
lower than those estimated by the Catalyst model. The least
successful technique for this system was the GFA-optimized
MS-WHIM method (test set: r
2
= 0.46).
1.13 Inhibitors of Human Cytochrome P450 17αα
αα
-
Hydroxylase-17,20-Lyase
While only about 50% of breast tumors are hormone-
sensitive, approximately 90% of prostate tumors are
androgen-dependent. The enzyme cytochrome P450
monooxygenase 17α-hydroxylase-17,20-lyase (CYP17 or
17-lyase), catalyzes the two key sequential reactions in the
biosynthesis of androgens such as testosterone [62]. This

12 Current Medicinal Chemistry, 2004, Vol. 11, No. 6 Güner et al.
Fig. (5). Alignment of pharmacophore model for steroidal CYP17 inhibitors with NRB 03689 identified from the Maybridge database
(IC
50
(expt) = 56.0 nM) [64].
N
R
R
O
N
O
N
R
R
R
N
R
R
R
HO
O
X
N
R
R
ON
R
O
R
N
R
R
Fig. (6). Examples of derivatives of propargylamines and non-propargylamines (Series I) used in the study [67].
enzyme is a target for the treatment of prostate cancer (PC),
an androgen-dependent disease. Ketoconazole, an antifungal
agent and a modest CYP17 inhibitor, has been used
clinically for the treatment of PC [63]. However,
Ketoconazole has been withdrawn from use because of liver
toxicity and other side-effects, due mainly to its inhibition
of other CYP enzymes. This highlights the need for
discovery and development of more potent and specific
CYP17 inhibitors.
Njar and co-workers [64] recently reported a
pharmacophore-based investigation of steroidal (see, e.g. 39)
and non-steroidal inhibitors of CYP17. Pharmacophore
models were generated to explain the putative binding
requirements for the two classes of human CYP17
inhibitors. Common chemical features in the steroid and
non-steroid human CYP17 enzyme inhibitors, as deduced by
the Catalyst/HipHop program, included 1-2 hydrogen bond
acceptors, and 3 hydrophobic groups. For azole-steroidal
ligands, the 3β–OH group of ring-A and N-3 of the azole
ring attached to ring-D at C-17 act as hydrogen bond
acceptors. A model that permits hydrogen bond interaction
between the azole functionality on ring-D and the enzyme is
consistent with experimental deductions for Type II CYP17
inhibitors where a sixth ligating atom interacts with Fe
II
of
heme. In general, pharmacophore models derived for steroid
and non-steroidal compounds bear striking similarities with
all azole sites mapping the HBA functionality, and 3
hydrophobic features describing the hydrophobic interactions
between the ligands and the enzyme. Using the
pharmacophore model derived for azole-steroidal inhibitors
as a 3D search query against the Catalyst/Maybrdige
database, several new steroidal compounds were identified
(see Chart 9; and Fig. (5)). Biological testing of some of
these compounds found five with low to high nanomolar
inhibitory potency against the human CYP17 enzyme.
1.14 Inhibitors of AChE and MAO
In the search for the treatment of Alzheimer’s Disease
(AD), a neuro-degenerative disease, it was hypothesized that
compounds with dual inhibitory roles against Acetycholine
esterase (AChE) and Monoamine oxidase (MAO) will be
most effective [65-67]. In one such study, Sterling and co-
workers [67] synthesized and conducted biological testing of
a series of carbamate and propargylamine derivatives (see,
e.g. Fig. (6) for their potential dual activity roles against
both Acetylcholine esterase (AChE) and Monoamine oxidase
(MAO) enzymes.
These workers [67] evaluated the putative binding
requirements for MAO inhibitory activity using a selection
of five anti-MAO compounds as input to the
Catalyst/HipHop program. In the model generation
experiment, all training set compounds were submitted in S-
configuration optical forms, higher weightings (priority)
were assigned to three of the five training set compounds,
and a new carbamate feature was defined and added into the
Catalyst feature dictionary.

Modeling and Three Dimensional Database Searching for Drug Design Using Catalyst Current Medicinal Chemistry, 2004, Vol. 11, No. 6 13
Fig. (7). (a) Putative five-featured pharmacophore for MAO inhibition derived for this class of inhibitors (green = hydrophobe, yellow
= carbamate moiety, magenta = hydrogen bond donor), and (b) alignment of two inactive compounds with the Catalyst/HipHop-
derived MAOI pharmacophore model [67].
The pharmacophore model derived for the putative
binding requirements against this enzyme consisted of the
following features: a carbamate group (yellow), two
hydrophobes (cyan) and two hydrogen bond donors
(magenta) (see Fig. (7a)). The carbamate feature in the
pharmacophore model imparts an orientational preference for
the propargylamine moiety in the interaction at the enzyme’s
active site. For the less potent (inactive) members of the
training set, the important propargylamine moiety was
shown to be misoriented when aligned with the
pharmacophore model (Fig. (7b)), hence accounting for their
weaker inhibitory affinities for this enzyme.
1.15 VLA-4 (
αα
αα
4
ββ
ββ
1) Antagonists
The antigen
α
4
β
1 (VLA-4) is a member of the integrin
family involved in the migration of white blood cells to
inflammation points in the body. The α4-integrin-dependent
adhesion pathways are critical target sites for therapeutic
treatments of asthma [68], multiple sclerosis [69], and
rheumatoid arthritis[70]. An early study [71] had identified a
minimal peptidic sequence of Leu-Asp-Val that encodes cell
adhesion activity against vascular cell adhesion molecule-1
(VCAM-1). A capped form of this peptide was found [72] to
be potent against a sheep model of allergic broncho-
constriction.
A search for non-peptidic inhibitors was conducted by
Singh and co-workers
[73], who employed a computational
approach to the design of novel and potent inhibitors of this
enzyme. Their approach involved the building of a
pharmacophore model using the Catalyst/Hypogen program,
and using the model as a search query against a chemical
compound database. The hits retrieved from the database
search were then tested for VLA-4 inhibitory activity.
Using Val-Leu-Asp as a template, the building block
consisted of 4-[N’-2-methylphenyl)ureido]phenylacetyl
(PUPA), and a carboxyl group. These features were found to
be critical for potent inhibition of this enzyme. Structurally
modified analogs of the peptidic template were evaluated
using Catalyst/Hypogen. The pharmacophore model derived
was used as a search query to identify new non-peptidic
compounds. Of the 416 hits retrieved from in silico database
screening, 170 were non-peptides. 12 of these were selected
on the basis of synthetic feasibility and commercial
availability, and tested biologically. Several compounds
were found to exhibit low micromolar to high nanomolar

14 Current Medicinal Chemistry, 2004, Vol. 11, No. 6 Güner et al.
potencies. The most potent compound identified is 40 (IC
50
~ 1.3 nM), and is shown aligned to the Hypogen model in
Fig. (8).
Fig. (8). Newly identified potent VLA-4 inhibitor (40; IC
50
~ 1.3
nM) aligned to the Catalyst/Hypogen model derived for VLA-4
inhibition [73].
NEW PHARMACOPHORE PATENTS USING
CATALYST
We have previously provided a review of two patent
applications filed by Biogen and Peptide Therapeutics
between 1998 and 1999, using the Catalyst/Hypogen
algorithm in building pharmacophore models for VLA-4
inhibitors [74] and Hepatitis C NS3 protease inhibitors
[75],
respectively. Hereafter, we list any new patent filings made
after 2000.
SELECTIVE SEROTONIN RE-UPTAKE
INHIBITORS
In 2002, Ekins and co-workers[76] filed a patent
application on the pharmacophore model for the elucidation
of inhibitory potency against CYP2D6 and design of
selective serotonin re-uptake inhibitors (SSRI). SSRIs are
antidepressant agents, with potential application in the
treatment of aggression disorders, anxiety disorders, panic
disorders, obsessive compulsive disorder, post traumatic
stress disorders, acute stress disorders, cognitive disorders,
etc. The patent describes a novel screening method for
identifying SSRIs with weak inhibitory potency against
many other cytochrome P450 enzymes, especially CYP2D6.
In addition, the work describes a method adopted for
generating a pharmacophore model for SSRIs. The
pharmacophore model was generated using the
Catalyst/Hypogen program, and consists of three geometric
features: a hydrogen bond donor, a hydrogen bond acceptor
and a hydrophobic group. The model yielded a novel SSRI,
which does not possess any inhibitory potency against
CYP2D6.
ACKNOWLEDGMENTS
The authors wish to acknowledge Dr. Jiabo Li and Dr.
Marvin Waldman for authorship of the CatDBLibrary
algorithm. We also thank Dr. Jon Sutter, Dr. Adrea T.
Mehl, Dr. Jeff Nauss, and Dr. Eric Jamois for their review of
this manuscript and helpful comments.
REFERENCES
[1] Li, J.; Waldman, M.; Sutter, J.; Freeman, C.F.; Clement, O.
CatDBLibrary: Ultra-rapid Conformation Generation for
Combinatorial Libraries. Manuscript in preparation.
[2] Palomer, A.; Cabré, F.; Pascual, J.; Campos, J.; Trujillp, M.A.;
Entrena, A.; Gallo, M.A.; García, L.; Mauleón, D.; Espinosa, A.
J. Med. Chem., 2002, 45, 1402-1411.
[3] (a) Barnum, D.; Green, J.; Smellie, A.; Sprague, P. J. Chem. Inf.
Comput. Sci. 1996, 36, 563-571; (b) Greene, J.; Kahn, S.; Savoj,
H.; Sprague, P.; Teig, S. J. Chem. Inf. Comput. Sci. 1994, 34,
1297-1308; (c) Sprague, P.W. In Perspectives in Drug Discovery
and Design, Vol. 3. Muller, K. ed., ESCOM Science Publishers B.
V., 1995, 1-20; (d) Clement, O.; Mehl, A.T. HipHop: In
Pharmacophore perception, development, and use in drug design,
Güner, O.F. ed., International University Line, La Jolla, CA, 2000,
71-84.
[4] Ehrich, E.; Dallob, A.; van Hecken, A.; Depre, M.; DeLepeleire,
I.; Brideau, C.; Hilliard, D.; Tanaka, W.; Mukhopadhyay, S.;
Seibold, J.; DeSchepper, P.; Gertz, D. Arthritus Rheum. 1996,
39(9), s81 (Abstract 328).
[5] Kurumbail, R.G.; Stevens, A.M.; Gierse, J.K.; McDonald, J.J.;
Stegeman, R.A.; Pak, J.Y.; Gildehaus, D.; Miyashiro, J.M.;
Penning, T.D.; Siebert, K.; Isakson, P. C.; Stallings, W.C. Nature,
1996, 384, 644-648.
[6] Schurmann, D.; Nightingale, S.D.; Bergmann, F.; Ruf, B. Infection,
1997, 25, 274-280.
[7] Benson, C. AIDS Res. Hum. Retroviruses, 1994, 10, 913-916.
[8] Churchyard, G.J.; Grant, A.D. S. Afr. Med. J., 2000, 90, 472-476.
[9] Mary-Krause, M.; Rabaud, C.; Jouan, M.; Obadia, M.; de la
Blanchardiere, A.; Raffi, F.; May, T. Pathol. Biol. 2000, 48, 495-
504.
[10] Ghassemi, M.; Andersen, B.R.; Reddy, V.M.; Gangadharam, P.R.;
Spear, G.T.; Novak, R.M. J. Infect. Dis. 1995, 171, 68-73.
[11] Reddy, V.M. Front. Biosci. 1998, 3, d525-d531.
[12] Barrow, W.W. Front. Biosci. 1997, 2, d387-d400.
[13] Horsburgh, C.R. Jr. N. Engl. Med. 1991, 324, 1332-1338.
[14] Debnath, A.K. J. Med. Chem. 2002, 45, 41-53.
[15] (a) Hong, L.; Sutter, J.; Hoffmann, R. In Pharmacophore
perception, development, and use in drug design, Güner, O. F. ed.,
International University Line, La Jolla, CA, 2000, 173-189; (b)
Kurogi, Y.; Güner, O.F. Curr. Med. Chem. 2001, 8, 1035-1055.
[16] Metcalf, B.W.; Levy, M.A.; Holt, D.A. Trends Pharmacol. Sci.
1989, 10, 491-495.
[17] Chen, G.S.; Chang, C.-S.; Kan, W.M.; Chang, C.-L.; Wang, K.C.
J. Med. Chem. 2001, 44, 3759-3763.
[18] Haas, C.S.; Schöcklmann, O.H.; Lang, S.; Kralewski, M.; Sterzel,
R.B. J. Nephrol. 1999, 12, 405-415.
[19] Kurogi, Y.; Miyata, K.; Okamura, T.; Hashimoto, K.; Tsutsumi,
K.; Nasu, M.; Moriyasu, M. J. Med. Chem. 2001, 44, 2304-2307.
[20] Johnson, D.R.; Bhatnagar, R.S.; Knoll, L.J.; Gordon, J.I. Ann. Rev.
Biochem. 1994, 63, 869.
[21] Duromio, R.J.; Towler, D.A.; Heuckeroth, R.O.; Gordon, J.I.
Science 1989, 243, 796.
[22] Weinberg, R.A.; McWherter, C.A.; Freeman, S.K.; Wood, D.C.;
Gordon, J.I. Mol. Microbiol. 1995, 16, 241.
[23] Lodge, J.K.; Jackson-Machelski, E.; Toffaletti, D.L.; Perfect, J.R.;
Gordon, J.I. Proc. Natl. Acad. USA 1994, 91, 12008.
[24] McWhether, C.A.; Rocque, W.J.; Zupec, M.E.; Freeman, S.K.;
Brown, D.L.; Devadas, B.; Getman, D.P.; Sikorski, J.A.; Gordon,
J.I. J. Biol. Chem. 1997, 272, 11874.
[25] Karki, R.G.; Kulkarni, V.M. Eur. J. Med. 2001, 36, 147-163.
[26] Weston, S.A.; Camble, R.; Colls, J.; Rosenbrock, G.; Taylor, I.;
Egerton, M.; Tucker, A.D.; Tunnicliffe, A.; Mistry, A.; Mancia,
F.; Fortelle, E.; Irwin, J.; Bricogne, G.; Pauptit, R.A. Nature Struct.
Biol. 1998, 213, 5.

Modeling and Three Dimensional Database Searching for Drug Design Using Catalyst Current Medicinal Chemistry, 2004, Vol. 11, No. 6 15
[27] (a) Yau, J.L.W.; Noble, J.; Widdowson, J.; Seckl, J.R. Mol. Brain
Res. 1997, 45, 182; (b) Shimizu, M.; Nishida, A.; Zensho, H.;
Yamawaki, S. J. Pharmacol. Exp. Ther. 1996, 279, 1551.
[28] Ying, S.-W.; Rusak, B. Brain Res. 1997, 755, 246.
[29] (a) Terron, J.A.; Falcon-Neri, A. Br. J. Pharmacol. 1999, 127,
609; (b) de Vries, P.; Visser, P.A.; Heiligers, J.P.C.; Villalon, C.M.;
Saxena, P.R. Naunyn-Schmiedeberg’s Arch. Pharmacol. 1999,
359, 331.
[30] (a) Martin, G.R.; Humphrey, P.P.A. Neuropharmacol. 1994, 33,
261-273; (b) Hoyer, D.; Clarke, D.E.; Fozard, J.R.; Harting, P.R.;
Martin, G.R.; Mylecharane, E.J.; Saxena, P.R.; Humphrey, P. A.
Pharmacol. Rev. 1994, 46, 157-203; (c) Zifa, E.; Fillion, G. 5-
Hydroxytryptamine Receptors. Pharmacol. Rev. 1992, 44, 401-
458; (d) Saxena, P. R. Pharmac. Ther. 1995, 66, 339-368.
[31] (a) Bureau, R.; Lancelot, J.C.; Prunier, H.; Rault, S. Quant. Struct.-
Act. Relat. 1996, 15, 373-381; (b) Daveu, C.; Bureau, R.; Baglin,
I.; Prunier, H.; Lancelot, J.C.; Rault, S. J. Chem. Inf. Comput. Sci.
1999, 39, 362-369.
[32] Bureau, R.; Daveu, C.; Baglin, I.; Sopkova-De Oliveira Santos, J.;
Lancelot, J.-C.; Rault, S. J. Chem. Inf. Comput Sci. 2001, 41, 815-
823.
[33] Eglen, R.M.; Jasper, J.R.; Chang, D.J.; Martin, G.R. Trends
Pharmacol. Sci. 1997, 18, 104.
[34] Forbes, I.T.; Dabbs, S.; Duckworth, D.M.; Jennings, A.J.; King,
F.D.; Lovell, P.J.; Brown, A.M.; Collin, L.; Hagan, J.J.;
Middlemiss, D.N.; Riley, G.J.; Thomas, D.R.; Upton, N. J. Med.
Chem. 1998, 41, 655.
[35] Kikuchi, C.; Nagaso, H.; Hiranuma, T.; Koyama, M. J. Med.
Chem. 1999, 42, 533.
[36] López-Rodríguez, M.L.; Porras, E.; Benhamu, B.; Ramos, J.A.;
Morcillo, M.J.; Lavandera, J.L. Biorg. Med. Chem. Lett. 2000, 10,
1097-1100.
[37] Lindberg, U.H.; Thorberg, S.O.; Bengtsson, S.; Renyi, A.L.; Ross,
S.B.; Ogren, S.O. J. Med. Chem. 1978, 21, 448-456.
[38] Humber, L.G.; Lee, D.; Rakhit, S.; Treasurywala, A.M. J. Mol.
Graphics 1985, 3, 84-89.
[39] Rupp, A.; Kovar, K.A.; Beuerle, G.; Ruf, C.; Folkers, G. Pharm.
Helv. Acta. 1994, 68, 235-244.
[40] Hoepping, A.; Johnson, K. M.; George, C.; Flippen-Anderson, J.;
Kozikowski, A. P. J. Med. Chem. 2000, 43, 2064-2071.
[41] Bureau, R.; Daveu, C.; Lancelot, J.-C.; Rault, S. J. Chem. Inf.
Comput. Sci. 2002, 42, 429-436.
[42] Stinson, S.C. Counting on chiral drugs. Special Report 1998, 76, 1-
136. http://pubs.acs.org/hotartcl/cenear/980921/drugs.html.
[43] Daveu, C.; Bureau, R.; Baglin, I.; Prunier, H.; Lancelot J.C.; Rault
S. J. Chem. Inf. Comput. Sci. 1999, 39, 362-369.
[44] Cerius2 v 4.5, Accelrys, 9685 Scranton Road, San Diego, CA
92121.
[45] Price, D.T.; Schwinn, D.A.; Lomasney, J.W.; Allen, L.F.; Caron,
M.G.; Lefkowitz, R.J. J. Urol. 1993, 250, 546-551.
[46] Hieble, J.P.; Ruffolo, Jr. R.R. Prog. Drug Res. 1996, 47, 81-90.
[47] Strader, C.D.; Sigal, I.S.; Dixon, R.A. FASEB J. 1989, 29, 123-
127.
[48] Strader, C.D.; Sigal, I.S.; Dixon, R.A. Trends Pharmacol. Sci.
1989, 26-30.
[49] Barbaro, R.; Betti, L.; Botta, M.; Corelli, F.; Giannaccini, G.;
Maccari, L.; Manetti, F.; Strappaghetti, G.; Corsano, S. J. Med.
Chem. 2001, 44, 2118-2132.
[50] Zhu, L.L.; Hou, T.J.; Chen, L.R.; Xu, X.J. J. Chem. Inf. Comput.
Sci. 2001, 41, 1032-1040.
[51] Osherov, N.; Gazit, A.; Gilon, C.; Levitzki, A. J. Biol. Chem. 1993,
268, 11134-11142.
[52] Sun, L.; Tran, N.; Tang, F.; App, H.; Hirth, P.; McMahon, G.;
Tang, C. J. Med. Chem. 1998, 41, 2588-2603.
[53] Martin, Y.C.; Bures, M.G.; Danaher, E.A.; DeLazzer, J.; Lico, I.;
Pavlik, P.A. J. Comput. Aided Mol. Des. 1993, 7, 587-619.
[54] Morris, J.A.; Blount, R.E., Jr.; Savage, R.E. Proc. Exp. Biol. Med.
1956, 192, 544-549.
[55] (a) Chanock, R.M. J. Exp. Med. 1956, 104, 555-563; (b) Chanock,
R.M.; Parrott, R.H.; Cook, K. N. Engl. J. Med. 1958, 258, 207-213.
[56] Saladino, R.; Danti, M.C.; Mincione, E.; Crestini, C.; Palamara,
A.T.; Savini, P.; Marini, S.; Botta, M. Bioorg. Med. Chem. Lett.
1998, 8, 1833-1838.
[57] Saladino, R.; Crestini, C.; Palamara, A.T.; Danti, M.C.; Manetti, F.;
Corelli, F.; Garaci, E.; Botta, M. J. Med. Chem. 2001, 44, 4554-
4562.
[58] Grass, G.M. Adv. Drug Del. Rev. 1997, 23, 199-219.
[59] (a) van de Waterbeemd, H.; Camenenisch, G.; Folkers, G.;
Raevsky, O.A. Quant. Struct. Act. Relat. 1996, 15, 480-490; (b)
Segarra, V.; Lopez, M.; Ryder, H.; Palacios, J.M. ibid., 1999, 18,
474-481; (c) Bravi, G.; Wikel, J.H. ibid., 2000, 19, 39-49; (d)
Clark, D.E. J. Pharm. Sci. 1999, 88, 807-814; (e) Kelder, J.;
Grootenhuis, P.D.J.; Bayada, D.M.; Delbressine, L.P.C.; Ploemen,
J.-P. Pharm. Res. 1999, 16, 1514-1519; (f) Krarup, L.H.;
Christensen, I.T.; Hovgaard, L.; Frokjaer, S. ibid. 1998, 15, 972-
978; (g) Lipinski, C.A.; Lombardo, F.; Dominy, B.W.; Feeney, P.J.
Adv. Drug Del. Rev. 1997, 23, 3-25; (h) Norinder, U.; Osterberg,
T.; Artursson, P. Eur. J. Pharm. Sci. 1999, 8, 49-56; (i) Palm, K.;
Luthman, K.; Ungell, A.-L.; Strandlund, G.; Artursson, P. J.
Pharm. Sci. 1996, 85, 32-39; (j) Palm, K.; Stenberg, P.; Luthman,
K.; Artursson, P. Pharm. Res. 1997, 14, 568-571; (k) Palm, K.;
Luthman, K.; Ungell, A.-L.; Strandlund, G.; Beigi, F.; Lundahl, P.;
Artursson, P. J. Med. Chem. 1998, 41, 5382-5392; (l) Sugawara,
M.; Takekuma, Y.; Yamada, H.; Kobayashi, M.; Iseki, K.;
Miyazaki, K. J. Pharm. Sci. 1998, 87, 960-966; (m) Wessel, M.D.;
Jurs, P.C.; Tolan, J.W.; Muskal, S.M. J. Chem. Inf. Comput. Sci.
1998, 38, 726-735; (n) Niwa, T. J. Chem. Inf. Comput. Sci. 2003,
43, 113-119.
[60] Kramer, S.D. Pharm. Sci. Technol. Today 1999, 2, 373-380.
[61] Ekins, S.; Durst, G.L.; Stratford, R. E.; Thorner, D.A.; Lewis, R.;
Loncharich, R.J.; Wikel, J. H. J. Chem. Inf. Comput. Sci. 2001, 41,
1578-1586.
[62] Hall, P.F. J. Steroid Biochem. Mol. Biol. 1991, 40, 527-532.
[63] Trachtenberg, J.; Halpern, N.; Pont, A. J. Urol. 1983, 130, 152-
153.
[64] Clement, O.; Freeman, C.M.; Hartmann, R.W.; Handratta, V.D.;
Vasaitis, T.S.; Brodie, A.M.H.; Njar, V.C.O. J. Med. Chem. 2003,
in press.
[65] Fink, D.M.; Palermo, M.G.; Bores, G.M.; Huger, F.P.; Kurys, B.E.;
Merriman, M.C.; Olsen, G.E.; Petko, W.; O’Malley, G.J. Bioorg.
Med. Chem. Lett. 1996, 6, 625-630.
[66] Bruhlmann, C.; Ooms, F.; Carrupt, P.-A.; Testa, B.; Catto, M.;
Leonetti, F.; Altomare, C,; Carotti, A. J. Med. Chem. 2001, 44,
3195-3198.
[67] Sterling, J.; Herzig, Y.; Goren, T.; Finkelstein, N.; Lerner, D.;
Goldenberg, W.; Miskolczi, I.; Molnar, S.; Rantal, F.; Tamas, T.;
Toth, G.; Zagyva, A.; Zekany, A.; Lavian, G.; Gross, A.;
Friedman, R.; Razin, M.; Huang, W.; Krais, B.; Chorev, M.;
Youdin, M.B.; Weinstock, M. J. Med. Chem. 2002, 45, 5260-5279.
[68] Abraham, W.M.; Sielczak, M.W.; Ahmed, A.; Cortes, A.;
Lauredo, I.T.; Kim, J.; Pepinsky, B.; Benjamin, C.D.; Leone, D.R.;
Lobb, R.R.; Weller, P.F. J. Clin. Invest. 1994, 93, 776-787.
[69] Chabot, S.; Williams, G.; Yong, V.M. J. Clin. Invest. 1997, 100,
604-612.
[70] Leger, O.J.; Yednock, T.A.; Tanner, L.; Horner, H.C.; Hines,
D.K.; Keen, S.; Saldanha, J. Hum. Antibodies 1997, 8, 3-6.
[71] Komoriya, A.; Green, L.J.; Mervic, M.; Yamada, S.S.; Yamada,
K.M.; Humphries, M.J. J. Biol. Chem. 1991, 266, 15075-15079.
[72] Lin, K.C.; Ateeq, H.S.; Hsiung, H.S.; Chong, L.T.; Zimmerman,
C.N.; Castro, A.; Lee, W.-C.; Hammond, C.E.; Kalkunte, S.; Chen,
L.-L.; Pepinsky, R.B.; Leone, D.R.; Sprague, A.G.; Abraham,
W.M.; Gill, A.; Lobb, R.R.; Adams, S.P. J. Med. Chem. 1999, 42,
920-934.
[73] Singh, J.; van Vlijmen, H.; Liao, Y.; Lee, W.-C.; Cornebise, M.;
Harris, M.; Shu, I.-H.; Gill, A.; Cuervo, J.H.; Abraham, W.M.;
Adams, S.P. J. Med. Chem. 2002, 45, 2988-2993.
[74] Singh, J.; Zheng, Z.; Sprague, P.; Van Vlihmen, H.; Castro, A.C.;
Adams, S.P. WO9804913.
[75] Hart, T.; Quibell, M. WO 9846630.
[76] Ekins, S. US2002/0013372 A1.