Three-dimensional common-feature hypotheses for octopamine agonist 2-(arylimino)imidazolidines

Article abstract of DOI:10.1016/S0968-0896(01)00247-4

Three-dimensional pharmacophore hypotheses were built from a set of 10 octopamine (OA) agonist 2-(Arylimino)imidazolidines (AIIs), 2-(Arylimino)thiazolidines (AITs) and 2-(Arylimino)oxazolidines (AIOs). Among the 10 common-featured models generated by program Catalyst/HipHop, a hypothesis including a ring aromatic (RA), a positive ionizable (PI) and three hydrophobic aliphatic (HpA1) features was considered to be important in evaluating the OA-agonist activity. Active OA agonist 2,6-Et₂ AII mapped well onto all the RA, PI and HpAl features of the hypothesis. On the other hand, less active compounds were shown to be difficult to achieve the energetically favorable conformation which is found in the active molecules in order to fit the 3-D common-feature pharmacophore models. Taken together, 2,6-Et₂-Ph and foramidine structures are important as OA agonists. The present studies on OA agonists demonstrate that a RA, a PI and three HpAl sites located on the molecule seem to be essential for OA-agonist activity. Copyright

Full text of DOI:10.1016/S0968-0896(01)00247-4

Bioorganic & Medicinal Chemistry10 (2002) 117–123
Three-Dimensional Common-Feature Hypotheses for Octopamine
Agonist 2-(Arylimino)imidazolidines
Akinori Hirashima,a,* Masako Morimoto,b Eiichi Kuwano,a Eiji Taniguchic
and Morifusa Etod
aDepartment of Applied Genetics and Pest Management, Faculty of Agriculture, Graduate School, Kyushu University,
6-10-1 Hakozaki, Higashi-ku, Fukuoka 812-8581, Japan
bGraduate School of Bioresource and Bioenvironmental Sciences, Kyushu University, 6-10-1 Hakozaki, Higashi-ku,
Fukuoka 812-8581, Japan
cSchool of Agriculture, Kyushu Tokai University, Kumamoto 869-1404, Japan
dKyushu Women’s University, 1-1 Jiyugaoka, Yahatanishi-ku, Kita-Kyushu, Fukuoka 807-8586, Japan
Received 28 May2001; accepted 7 July2001
Abstract—Three-dimensional pharmacophore hypotheses were built from a set of 10 octopamine (OA) agonist 2-(Arylimino)imi-
dazolidines (AIIs), 2-(Arylimino)thiazolidines (AITs) and 2-(Arylimino)oxazolidines (AIOs). Among the 10 common-featured
models generated by program Catalyst/HipHop, a hypothesis including a ring aromatic (RA), a positive ionizable (PI) and three
hydrophobic aliphatic (HpAl) features was considered to be important in evaluating the OA-agonist activity. Active OA agonist
2,6-Et2 AII mapped well onto all the RA, PI and HpAl features of the hypothesis. On the other hand, less active compounds were
shown to be difficult to achieve the energeticallyfavorable conformation which is found in the active molecules in order to fit the
3-D common-feature pharmacophore models. Taken together, 2,6-Et2-Ph and foramidine structures are important as OA agonists.
The present studies on OA agonists demonstrate that a RA, a PI and three HpAl sites located on the molecule seem to be essential
for OA-agonist activity. # 2001 Elsevier Science Ltd. All rights reserved.
Introduction
dimensional information, one mayattempt to build a
hypothetical model of the active site that can provide
insight on the nature of the active site. Such a model is
known as a Hypo. Catalyst/Hypo is useful in building
3-D pharmacophore models from the activitydata and
conformational structure. It can be used as an alter-
native for QSAR methods because of easyvisualization
and high prediction.
Quantitative structure–activityrelationship (QSAR)
modeling is an area of research pioneered byHansch
and Fujita.1,2 The QSAR studyassumes that the differ-
ence of the molecules in the structural properties
experimentallymeasured accounts for the difference in
their observed biological or chemical properties.1ꢀ3 The
result of QSAR usuallyreflects as a predictive formula
and attempts to model the activityof a series of com-
pounds using measured or computed properties of the
compounds. More recently, QSAR has been extended
byincluding the three-dimensional information. In drug
discovery, it is common to have measured activity data
for a set of compounds acting upon a particular protein
but not to have knowledge of the three-dimensional
structure of the active site. In the absence of such three-
In a previous application, we described the use of Cat-
alyst/Hypo to derive a 4- and 5-feature hypothesis from
a set of 17 octopamine (OA) antagonists4 and 43 ago-
nists,5 in which binding assays were used, respectively.
Three-dimensional pharmacophore hypotheses were
built from a set of nine OA agonists responsible for the
inhibition of sex-pheromone production in Helicoverpa
armigera.6 These sets included a varietyof tpyes of
molecules, covering five orders of magnitude in activity.
For these type of training sets, the use of the hypothesis-
generation tool was appropriate. This tool Hypo builds
hypotheses (overlays of chemical features) for which the
*Corresponding author. Tel.: +81-92-642-2856; fax: +81-92-642-
2858; e-mail: ahirasim@agr.kyushu-u.ac.jp
0968-0896/02/$ - see front matter # 2001 Elsevier Science Ltd. All rights reserved.
PII: S0968-0896(01)00247-4

118
A. Hirashima et al. / Bioorg. Med. Chem. 10 (2002) 117–123
fit of individual molecules to a hypothesis can be corre-
lated with the molecule’s affinity. For OA antagonists
binding assays are good enough to evaluate their affi-
nity. Meanwhile, for OA agonists, they are not enough.
Thus, in order to evaluate OA-agonist activity, an ade-
nylate-cyclase assay was used. However, the high struc-
tural homologyamong the derivatives used in the
current studycombined with their smaller activityrange
makes this ‘quantitative’ hypothesis generation method
inappropriate. For this type of training set, the com-
mon-feature hypothesis generation, also called
HipHop,7 is more suitable. HipHop generates hypoth-
eses consisting onlyof identification and overlayof
common features (without the use of activitydata). The
aim of this work is to derive feature-based 3-D models
from a small set of 10 OA agonists using HipHop.
models were produced with the minimum permitted
interfeature spacing of 2.00 A generating alignments of
common features,7 which included the projected point
of RA, PI and HpAl.8
It was found that hypotheses contain good correlation
with RA, PI and HpAl. The characteristics of 10
hypotheses are listed in Table 3. All the hypotheses
contain five features with the ranking scores ranging
from 109.823 to 97.566. Hypotheses 1 and 2 consist of
the same common-feature functions of a RA, a PI and
three HpAls. The second group composes of hypotheses
3–5 and 7–8 which are characterized bya RA, a PI, two
HpAls and an Hp features. Another hypothese 6 is
characterized bya PI, three HpAls and an Hp features.
Hypotheses 9 and 10 consist of a PI, two HpAls and
two Hps. The rank score range over the 10 generated
hypotheses is 12.257. The small rank score range
observed here maybe due to two factors, namelymole-
cules in the training set are fairlyrigid and have a high
degree of structural homology. Due to the relatively
small range and owing, moreover, to the placement of
the identified hypotheses within this range, special care
was taken to test for chance correlation. The higher
the ranking score, the less likelyit is that the mole-
cules in the training set fit the hypothesis by a chance
correlation.
Results and Discussion
Assessment of 3-D hypothesis for OA-agonist activity
OA-agonist activities of test compounds at several con-
centrations were examined using the adenylate-cyclase
assaywhich was conducted on adult American cock-
roaches (P. americana L). The activityas OA agonist
was structure specific. AII 57 with 2,6-Et2-Ph sub-
stituent was the onlyfull agonist in this studyand all
other AIOs and AITs were partial agonists. A slight
modification of structure of 57 decreased the OA-ago-
nist activitydramatically: substituents at 2,6 positions
of the phenyl, the introduction of substituents at 4
position of the phenyl and the introduction of a het-
eroatom O or S to the imidazolidine. Hypotheses were
generated to explain the specificityof the OA agonists.
A set of 10 molecules, including 57 and its derivatives,
was selected as the target training set. Their experi-
mental biological activities are listed in Table 1. Among
the 10 molecules of the training set, 57 was chosen as a
reference compound, which was allowed to map all fea-
tures, and the other nine molecules were allowed to map
partiallyon the hypotheses (Table 2). Except for this
classification, the activities of the molecules were not
used in the analysis. This tool builds hypotheses (over-
lays of common features) for which the fit of individual
molecules to a hypothesis can be correlated with the
molecule’s activity. The 3-D hypothesis study was per-
formed with the Catalyst (version 4.0) package. The
geometryof each compound was built with a visualizer
and optimized byusing the generalized CHARMm-like
force field implemented in the program. A preparative
test was performed with hydrogen-bond acceptor
(HBA), hydrogen-bond acceptor lipid (HBAl), hydro-
gen-bond donor (HBD), hydrophobic (Hp), hydro-
phobic aromatic (HpAr), hydrophobic aliphatic (HpAl),
negative ionizable (NI), ring aromatic (RA) and positive
ionizable (PI).8 NI and PI were used rather than nega-
tive charge and positive charge in order to broaden the
search for deprotonated and protonated atoms or
groups at physiological pH. Using conformatinal pol-
ing,9 a representative familyof conformers was gener-
ated, within a 20 kcal/mol range of the computed
minimum, for each molecule. Potential hypothesis
OA agonists–receptor interaction
Comparison of the procedure and regression studies
shows that hypotheses 1, 3, 6 and 9 are the best models
among the four groups and are selected for further eva-
luation. Figures 1–3 depict AII 57, the most active
compound, AII 56 and AIO 49, which is an analogue of
57, mapped onto hypothesis 1, respectively. The mole-
cule 57 maps well onto the five features of hypothese 1
(Fig. 1), whereas two HpAls do not map on the iPr and
Me in 56 (Fig. 2). A PI of hypothesis 1 does not fit to
either 49 (Fig. 3) nor its AIT derivative 27 (data not
shown). The methyl group at 4 position of phenyl of 59
does not map to none of 5 features (Fig. 4), suggesting
that substituents at 4 position of phenyl are not favor-
able. It is, however, still to be clarified, preferablyusing
2,4,6-Et3 derivative which showed a high binding affi-
5
nityin a previous report but was not available in this
study. Taken together, 2,6-Et2-Ph and foramidine
structures are important as OA agonists. Generally,
more active molecules map well onto all the features of
the hypothesis (Fig. 1), and compounds that have low
activitymap poorlyto the hypothesis (Figs 2–4). Other
compounds in Table 1 with low activityalso do not fit
to these features.
A HpAl of hypothesis 1 is replaced by a Hp, leading to
hypothesis 3 and a RA of hypothesis 3 is replaced by a
HpAl, leading to hypothesis 6. A HpAl of hypothesis 5
is replaced bya Hp, leading to hypothesis 9. The small
range of rank score suggests that these hypotheses
were homogenous. Roughlyspeaking, hypotheses 1, 3, 6
and 9 have the good similarityin 3-D spatial shape
and therefore these hypotheses are considered to be
equivalent.

A. Hirashima et al. / Bioorg. Med. Chem. 10 (2002) 117–123
Table 1. OA agonists AITs, AIOs and AIIs used in this study
119
Compda
R
X
Mp (ꢁC)
Adenylate-cyclase activity
(relative to OA,%)b
1
2
3
4
5
6
7
8
Ph
2-Br-Ph
2-Et-Ph
3-NO2-Ph
4-Cl-Ph
4-CF3-Ph
4-Me-Ph
4-MeS-Ph
4-EtO-Ph
4-Et2N-Ph
4-(Bz)2N-Bz
2,3-Cl2-Ph
2-Cl,4-Br-Ph
2,4-F2-Ph
2-Br,4-Me-Ph
2-Me,4-Cl-Ph
2,4-Me2-Ph
2,4-(MeO)2-Ph
2,5-Cl2-Ph
2-MeO,5-Cl-Ph
2-MeO,5-Me-Ph
2,5-(MeO)2-Ph
2,6-Cl2-Ph
2-Cl,6-Me-Ph
2,6-Me2-Ph
2-Me,6-Et-Ph
2,6-Et2-Ph
2-Et,6-iPr-Ph
3,4-Cl2-Ph
3-Cl,4-Me-Ph
3-CF3,4-Cl-Ph
3,5-Cl2-Ph
2,3,4-Cl3-Ph
2,4,5-Cl3-Ph
2,4,6-Br3-Ph
2,4,6-Cl3-Ph
2,6-Me2,4-Br-Ph
2,4,6-Me3-Ph
2,3,4,5-Cl4-Ph
Ph
S
S
S
S
S
S
S
S
S
S
S
S
S
S
S
S
S
S
S
S
S
S
S
S
S
S
S
S
S
S
S
S
S
174–176
166–168
59–61
3.7ꢂ0.1
5.2ꢂ0.3
12.3ꢂ2.5
4.8ꢂ1.0
7.4ꢂ1.4
6.8ꢂ0.2
3.0ꢂ0.2
2.2ꢂ0.2
2.0ꢂ0.4
1.1ꢂ0.4
2.5ꢂ1.2
3.8ꢂ0.5
4.2ꢂ1.0
0.6ꢂ0.2
25.0ꢂ2.5
36.7ꢂ3.8
55.0ꢂ3.9
3.4ꢂ0
124–126
164–166
163–165
145–147
105–107
108–111
106–108
156–157
168–170
171–173
146–148
148–150
139–141
66–68
130–132
171–173
136–138
144–146
155–157
171–173
141–143
109–111
Oil
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
1.2ꢂ0.1
8.0ꢂ0.7
14.0ꢂ2.6
18.5ꢂ2.2
1.8ꢂ0.9
4.8ꢂ1.1
7.9ꢂ0.9
30.4ꢂ4.7
50.2ꢂ2.2
24.4ꢂ3.5
8.4ꢂ0.5
5.5ꢂ0.3
0.2ꢂ0.2
3.4ꢂ0.2
9.1ꢂ1.0
9.1ꢂ0.7
2.7ꢂ0.1
5.3ꢂ0.4
19.3ꢂ2.3
33.3ꢂ2.9
4.8ꢂ0.1
2.7ꢂ0.4
5.2ꢂ0.7
9.1ꢂ0.4
5.9ꢂ0.3
11.5ꢂ0.1
ꢀ0.3ꢂ0.6
7.9ꢂ0.6
10.3ꢂ1.2
15.6ꢂ0.4
50.5ꢂ7.1
42.5ꢂ0.8
13.9ꢂ1.4
6.1ꢂ0.4
33.0ꢂ14.8
47.7ꢂ7.6
47.6ꢂ2.7
18.8ꢂ8.7
104.0ꢂ7.2
10.9ꢂ3.3
59.1ꢂ10.2
72–74
115–117
137–139
107–109
116–117
197–199
194–196
188–190
176–178
142–144
42–44
99–101
185–187
132–134
Oil
61–63
89–91
175–176
147–148
Oil
102–104
Oil
103–105
172–174
169–170
144–145
103–104
159–160
149–150
200–201
168–169
234–235
174–175
S
S
S
S
S
S
O
O
O
O
O
O
O
O
O
O
O
O
N
N
N
N
N
N
N
N
2-Me-Ph
2-Et-Ph
2-iPr-Ph
2,6-Cl2-Ph
2,6-F2-Ph
2,6-Me2-Ph
2-Me,6-Et
2-Me,6-iPr
2,6-Et2-Ph
2-Et,6-iPr-Ph
2,6-iPr2-Ph
2-Me-Ph
2-Et-Ph
2,6-Me2-Ph
2-Me,6-Et-Ph
2-Me,6-iPr-Ph
2,6-Et2-Ph
2,6-iPr2-Ph
2,4,6-Me3-Ph
aAITs 1–39 were synthesized by cyclization of the corresponding N-arylthioureas with concentrated hydrogen chloride.10 AIOs 40–51 were obtained
bycyclodesulfurizing the corresponding N-arylthioureas with yellow mercuric oxide.11 AIIs 52–59 were prepared according to a reported method by
refluxing the corresponding substituted anilines and 1-acetyl-2-imidazolidone in phosphoryl chloride followed by hydrolysis.12
bThe adenylate-cyclase assay of test compounds was conducted at several concentrations on adult American cockroaches as shown in previous
report.10ꢀ13 The basal (control) and maximal adenylate-cyclase activities stimulated by OA (0.1 mM) were 26.2ꢂ5.6 and 612.2ꢂ127.5 pmol cAMP/
min/mg of protein, respectively. The maximal stimulatoryactivities (mostlyat 0.1 mM) of test compounds were calculated relative to OA (100%)
and control (0%).

120
A. Hirashima et al. / Bioorg. Med. Chem. 10 (2002) 117–123
Table 2. Characteristics for the common feature hypothesis run
Compd
Confsa
Features/confsa
Principalb
MaxOmitFeatc
27
49
52
53
54
55
56
57
58
59
30
30
4
20
11
20
18
36
15
9
10.93
11.00
9.00
1
1
1
1
1
1
1
2
1
1
1
1
1
1
1
1
1
0
1
1
9.30
10.64
11.65
11.83
12.00
11.27
12.44
aAbbreviations: Confs, number of conformers; Features/confs, total
number of features divided bythe number of conformers (summed
over the entire familyof conformers).
bPrincipal=1 means that this molecule must map onto the hypotheses
generated bythe search procedure. Partial mapping is allowed. Prin-
cipal=2 means that this is a reference compound. The chemical fea-
ture space of the conformers of such a compound is used to define the
initial set of potential hypotheses.
cMaxOmitFeat=1 means a feature of a compound maynot be map-
ped on a hypothesis model. MaxOmitFeat=0 means all features of a
compound are mapped on a hypothesis model.
Figure 2. Mapping of 56 onto hypothesis 1, which contains a RA
(orange), a PI (red) and three HpAls (blue).
Figure 3. Mapping of 49 onto hypothesis 1, which contains a RA
(orange), a PI (red) and three HpAls (blue).
Figure 1. Mapping of 57 onto hypothesis 1, which contains a RA
(orange), a PI (red) and three HpAls (blue).
Table 3. Results of the common feature hypothesis run
Hypotheses
Featurea
Rank Score
Direct Hitb
Partial Hitb
1
2
3
4
5
6
7
8
9
10
RA
RA
RA
RA
RA
PI
RA
RA
PI
PI
PI
PI
PI
HpAl
HpAl
HpAl
HpAl
HpAl
HpAl
HpAl
HpAl
HpAl
HpAl
HpAl
HpAl
HpAl
HpAl
HpAl
HpAl
HpAl
HpAl
Hp
HpAl
HpAl
Hp
Hp
Hp
Hp
Hp
Hp
Hp
109.823
106.505
103.423
103.352
103.352
103.269
100.105
100.043
98.988
0111111101
0111111101
0111111101
0111111101
0111111101
0111111101
0111111101
0111111101
0111111101
0111111101
1000000010
1000000010
1000000010
1000000010
1000000010
1000000010
1000000010
1000000010
1000000010
1000000010
PI
HpAl
PI
PI
HpAl
HpAl
PI
Hp
Hp
97.566
aAbbreviations: RA, ring aromatic; PI, positive ionizable; HpAl, hydrophobic aliphatic; Hp, hydrophobic.
bDirect Hit, all the features of the hypothesis are mapped. Direct Hit=1 means yes and Direct Hit=0 is no; Partial Hit, partial mapping of the
hypothesis. Partial Hit=1 means yes and Partial Hit=0 means no. Each number refers to a molecule in Table 2 (same order).

A. Hirashima et al. / Bioorg. Med. Chem. 10 (2002) 117–123
121
Graphical examination of the 10 hypotheses shows that
there are four major families of models depending
mainlyon the location and the orientation of the pro-
jected point of the RA, PI, HpAl and Hp. It was found
that more active OA agonists map well onto all the fea-
tures of the hypotheses. For some inactive compounds,
their lack of affinityis primarilydue to their inabilityto
achieve an energeticallyfavorable conformation shared
bythe active compounds. Taken together, a RA, a PI
and three HpAls located on the molecule seem to be
essential for activityof OA agonists.
Experimental
Chemicals
OA [2-amino-1-(4-hydroxyphenyl)ethanol], theophylline
(1,3-dimethylxanthine) and ethylene glycol bis(b-ami-
noethyl ether)-N,N,N0,N0-tetraacetic acid (EGTA) were
purchased from Nacalai Tesque (Kyoto, Japan); GTP
was from Sigma Chemical Co. (St. Louis, USA); ATP
disodium salt was from Kohjin Co. (Tokyo, Japan);
lithium aluminum hydride (LAH) was from Chemetall
GmbH (Frankfurt, Germany).
Figure 4. Mapping of 59 onto hypothesis 1, which contains a RA
(orange), a PI (red) and three HpAls (blue).
Radiochemical
Conclusions
The cAMP radioimmunoassay(RIA) kit (cord RPA
509) was purchased from Amersham International
(Buckinghamshire, England).
In rational drug design process, it is common that the
biological activitydata of a set of compounds acting
upon a particular protein is known, while information
of the three-dimensional structure of the protein active
site is absent. A three-dimensional pharmacophore
hypothesis that is consistent with known data should be
useful and predictive in evaluating new compounds and
directing further synthesis. A pharmacophore model
postulates that there is an essential three-dimensional
arrangement of functional groups that a molecule must
possess to be recognized bythe active site. It collects
common features distributed in 3-D space that is inten-
ded to represent groups in a molecule that participates
in important interactions between drugs and their active
sites. Hence, a pharmacophore model provides crucial
information about how well the common features of a
subject molecule overlap with the hypothesis model. It
also informs the abilityof molecules to adjust their
conformations in order to fit an active site with energe-
ticallyreasonable conformations. Such characterized
3-D models conveyimportant information in an intuitive
manner.
Methods
Synthesis of test compounds. All compounds were pre-
pared using published methods. 2-(Arylimino)thiazoli-
dines (AITs) 1–39 were synthesized by cyclization of the
corresponding thiourea with concd hydrogen chloride.10
2-(Arylimino)oxazolidines (AIOs) 40–51 were obtained
bycyclodesulfurizing the corresponding thiourea with
yellow mercuric oxide.11 2-(Arylimino)imidazolidines
(AIIs) 52–59 were prepared according to a reported
method byrefluxing the corresponding substituted ani-
lines and 1-acetyl-2-imidazolidone in phosphoryl chlo-
ride followed by hydrolysis.12 The structures of the
1
compounds were confirmed by H and 13C NMR mea-
sured with a JEOL JNM-EX400 spectrometer at
400 MHz, tetramethyl silane (TMS) being used as an
1
internal standard for H NMR, and elemental analysis.
Insects. Males and females of P. americana were used
indiscriminately, as their nervous systems exhibited no
gross structural or neurochemical differences. The
insects were reaꢁred under crowded conditions in this
laboratoryat 28 C with a photoperiod of 12 h light:12 h
dark and at a relative humidityof 65–70% for more
than 7 years; they were provided with an artificial
mouse diet (Oriental Yeast Co., Chiba, Japan) and
water ad libitum.
The present work shows how a set of activities of var-
ious OA agonists maybe treated statisticallyto uncover
the molecular characteristics which are essential for high
activity. These characteristics are expressed as common
features disposed in three-dimensional space and are
collectively termed a hypothesis. Hypotheses were
obtained and applied to map the active or inactive
compounds. Important features such as a RA, a PI and
three HpAls of the surface-assessable models were
found for OA agonists. Theyare the minimum compo-
nents of a hypothesis for effective OA agonists.
Adenylate-cyclase assay. The adenylate-cyclase assay
was conducted on adult American cockroaches (P.
americana L) as shown in previous report.10ꢀ13 Thoracic

122
A. Hirashima et al. / Bioorg. Med. Chem. 10 (2002) 117–123
nerve cords of P. americana were homogenized (15 mg/
mL) in a 6 mM Tris-maleate buffer (pH 7.4) byusing a
chilled microtube homogenizer (S-203, Ikeda Sci.,
Tokyo, Japan) as shown in previous report. The homo-
genate was diluted (1 mg/mL) in 6 mM Tris-maleate,
and then centrifuged at 120,000g and 4 ꢁC for 20 min.
The supernatant was discarded, the pellet being resus-
pended byhomogenizing (1 mg/mL) in the buffer, and
again centrifuged at 120,000g and 4 ꢁC for 20 min. The
resulting pellet (P2) resuspended in the buffer was
equivalent to the starting amount (15 mg/mL). The
adenylate-cyclase activity was measured according to
Nathanson’s procedure under optimal conditions10ꢀ13 in
a test tube containing 200 mL of 120 mM Tris-maleate
(pH 7.4, including 15 mM theophylline, 12 mM MgCl2
and 0.75 mM EGTA), 60 mL of the P2 fraction and 30
mL of each synthesized compound solution in poly-
ethylene glycol. An appropriate solvent control was run
in parallel. The enzyme reaction (5 min at 30 ꢁC) was
initiated byadding 10 mL of a mixture of ꢁ3 mM GTP
and 60 mM ATP, stopped byheating at 90 C for 2 min
and then centrifuged at 1000g for 15 min to remove the
insoluble material. The cAMP level in the supernatant
was measured byRIA. 10,11,13 Protein concentration was
determined bythe Lowrymethod, 14 using bovine serum
albumin (Sigma, St. Louis, USA) as the standard.
Enzyme activity in each assay was corrected using OA
as a reference. The maximal stimulatoryactivity(mostly
at 0.1 mM) was calculated relative to OA (100%) and
control (0%).
within a given energy range. Catalyst provides two types
of conformational analysis: fast and best quality. The
best option was used, specifying 250 as the maximum
number of conformers. The molecules associated with
their conformational models was submitted to Catalyst
hypothesis generation. Hypotheses approximating the
pharmacophore were described as a set of features dis-
tributed within a 3-D space. This process onlycon-
sidered surface accessible functions such as HBA,
HBAl, HBD, Hp, HpAr, HpAl, RA, NI and PI.
HipHop provides feature-based alignment of a collec-
tion of compounds without considering activity. It
matches the chemical features of a molecule, against
drug candidate molecules. HipHop takes a collection of
conformational models of molecules and a selection of
chemical features, and produces a series of molecular
alignments in a varietyof standard file formats. HipHop
begins byidentifying configurations of features common
to a set of molecules. A configuration consists of a set of
relative locations in 3-D space and associated feature
types. A molecule matches the configurations if it pos-
sesses conformations and structural features that can be
superimposed within a certain tolerance from the corre-
sponding ideal locations. HipHop also maps partial
features of molecules in the alignment set. This provi-
sion gives the option to use partial mapping during the
alignment. Partial mapping allows to identifylarger,
more diverse, more significant hypotheses and align-
ment models without the risk of missing compounds
that do not map to all of the pharmacophore features.
Misses, the number of molecules which do not have to
map to all features in generated hypotheses, Feature-
Misses, the the number of maximal molecules which do
not have to map to each feature in generated hypoth-
eses, and CompleteMisses, the number of molecules
which do not have to map to anyfeature in a given
hypothesis, were set as 3, 2 and 2, respectively.
Hypothesis generation
All experiments were conducted on a Silicon Graphics
O2, running under the IRIX 6.5 operating system.
Hypotheses generation was applied against previously
described data sets and their functionalityis available as
part of Molecular Simulations Incorporated’s Cata-
lyst/HipHop (version 4.0) modeling environment
(Burlington, USA). Molecules were edited using the
Catalyst 2-D/3-D visualizer. Catalyst automatically
generated conformational models for each compound
using the Poling Algorithm.9,15,16 The number of con-
formations needed to produce a good representation of
a compound’s conformational space depends on the
molecule. Conformation-generating algorithms were
adjusted to produce a diverse set of conformations,
avoiding repetitious groups of conformations all repre-
senting local minima. The conformations generated
were used to align common molecular features and
generate pharmacophoric hypotheses. HipHop used
conformations generated to align chemicallyimportant
functional groups common to the molecules in the study
set. A pharmacophoric hypothesis then was generated
from these aligned structures.
Acknowledgements
This work was supported in part bya Grant-in-Aid for
Scientific Research from the Ministryof Education,
Science and Culture of Japan.
References and Notes
1. Hansch, C.; Leo, A. In Exploring QSAR: Fundamentals and
Applications in Chemistry and Biochemistry; American Chemi-
cal Society: Washington DC, 1995.
2. Hansch, C.; Fujita, T. J. Am. Chem. Soc. 1964, 86, 1616.
3. Golender, V. E.; Vorpagel, E. R. In 3-D-QSAR in Drug
Design: Theory, Methods, and Applications; Kubinyi, H., Ed.;
ESCOM Science Publishers: The Netherlands, 1993; p 137.
4. Pan, C.; Hirashima, A.; Kuwano, E.; Eto, M. J. Molec.
Model. 1997, 3, 455.
5. Hirashima, A.; Pan, C.; Kuwano, E.; Taniguchi, E.; Eto,
M. Bioorg. Med. Chem. 1999, 7, 1437.
6. Hirashima, A.; Rafaeli, A.; Gileadi, C.; Kuwano, E. J. Mol.
Graphics Mod. 1999, 17, 43.
The models emphasized a conformational diversity
under the constraint of 20 kcal/mol energythreshold
above the estimated global minimum based on use of
the CHARMm force field.9,15ꢀ17 Molecular flexibility
was taken into account byconsidering each compound
as a collection of conformers representing a different
area of conformational space accessible to the molecule
7. Barnum, D.; Greene, J.; Smellie, A.; Sprague, P. J. Chem.
Inf. Comput. Sci. 1996, 36, 563.

A. Hirashima et al. / Bioorg. Med. Chem. 10 (2002) 117–123
123
8. Greene, J.; Kahn, S.; Savoj, H.; Sprague, P.; Teig, S. J.
Chem. Inf. Comput. Sci. 1994, 34, 1297.
9. Smellie, A.; Teig, S. L.; Towbin, P. J. Comp. Chem. 1994,
16, 171.
13. Hirashima, A.; Yoshii, Y.; Eto, M. Pestic. Biochem. Phy-
siol. 1992, 44, 101.
14. Lowry, O. H.; Rosebrough, N. J.; Farr, A. L.; Randall,
R. J. J. Biol. Chem. 1951, 193, 265.
10. Hirashima, A.; Tarui, H.; Eto, M. Biosci. Biochem. Bio-
tech. 1994, 58, 1206.
15. Smellie, A.; Kahn, S. D.; Teig, S. L. J. Chem. Inf. Comp.
Sci. 1995, 35, 285.
11. Hirashima, A.; Pan, C.; Katafuchi, Y.; Taniguchi, E.; Eto,
M. J. Pestic. Sci. 1996, 21, 419.
16. Smellie, A.; Kahn, S. D.; Teig, S. L. J. Chem. Inf. Comp.
Sci. 1995, 35, 295.
12. Nathanson, J. A.; Kaugars, G. J. Med. Chem. 1989, 32,
1795.
17. Brooks, B. R.; Brucolleri, R. E.; Olafson, B. D.; States, D. J.;
Swaminathan, S.; Karplus, M. J. Comput. Chem. 1983, 4, 187.