Three-dimensional pharmacophore hypotheses of octopamine/tyramine agonists which inhibit [1-14C]acetate incorporation in Plodia interpunctella

Article abstract of DOI:10.1016/S0968-0896(02)00320-6

Three-dimensional pharmacophore hypotheses were built from a set of 36 octopamine (OA)/tyramine (TA) agonists responsible for the inhibition of sex-pheromone production in Plodia interpunctella. Among the ten chemical-featured models generated by a program Catalyst/Hypo, hypotheses including hydrogen-bond acceptor (HBA), hydrogen-bond acceptor aliphatic (HBAl), hydrophobic (Hp), hydrophobic aromatic (HpAr) and hydrophobic aliphatic (HpAl) features were considered to be important and predictive in evaluating OA/TA agonists. Active agonists mapped well onto all the features of the hypothesis such as HBA, HBAl, Hp, HpAr and HpAl features. On the other hand, inactive compounds were shown to be poorly capable of achieving an energetically favorable conformation shared by the active molecules in order to fit the 3-D chemical-feature pharmacophore models. Those hypotheses are considered to be used in designing new leads for hopefully more active compounds. Further research on the comparison of models from the agonists may help elucidate the mechanisms of OA/TA receptor-ligand interactions.

Full text of DOI:10.1016/S0968-0896(02)00320-6

Bioorganic & Medicinal Chemistry11 (2003) 95–103
Three-Dimensional Pharmacophore Hypotheses of Octopamine/
Tyramine Agonists which Inhibit [1-14C]Acetate Incorporation in
Plodia interpunctella
Akinori Hirashima,a,* Tomohiko Eiraku,b Yoko Shigetac and Eiichi Kuwanoa
aDivision of Bioresource and Bioenvironmental Sciences, Graduate School, Kyushu University, Fukuoka 812-8581, Japan
bGraduate School of Bioresource and Bioenvironmental Sciences, Kyushu University,
Fukuoka 812-8581, Japan
cDepartment of Bioresource and Bioenvironmental Sciences, School of Agriculture, Kyushu University,
Fukuoka 812-8581, Japan
Received 20 May2002; accepted 5 July2002
Abstract—Three-dimensional pharmacophore hypotheses were built from a set of 36 octopamine (OA)/tyramine (TA) agonists
responsible for the inhibition of sex-pheromone production in Plodia interpunctella. Among the ten chemical-featured models gen-
erated by a program Catalyst/Hypo, hypotheses including hydrogen-bond acceptor (HBA), hydrogen-bond acceptor aliphatic
(HBAl), hydrophobic (Hp), hydrophobic aromatic (HpAr) and hydrophobic aliphatic (HpAl) features were considered to be
important and predictive in evaluating OA/TA agonists. Active agonists mapped well onto all the features of the hypothesis such as
HBA, HBAl, Hp, HpAr and HpAl features. On the other hand, inactive compounds were shown to be poorlycapable of achieving
an energeticallyfavorable conformation shared bythe active molecules in order to fit the 3-D chemical-feature pharmacophore
models. Those hypotheses are considered to be used in designing new leads for hopefully more active compounds. Further research
on the comparison of models from the agonists mayhelp elucidate the mechanisms of OA/TA receptor–ligand interactions.
# 2002 Elsevier Science Ltd. All rights reserved.
Introduction
inhibitoryaction is a result of a receptor (separate from
the PBAN-receptor) which can be inhibited bypertussis
toxin.12 This provided evidence that this specific pher-
omonostatic-aminergic receptor is linked to a G-inhibi-
toryprotein. Female moths call conspecific males during
specific periods when theyemit their pheromone. The
major pheromone component of Indian meal moth Plodia
interpunctella was identified as (Z,E)-9,12-tetradecadienyl
acetate14ꢀ16 and the inhibitors of calling behaviour and
pheromone production have been reported.17
Production of the pheromone blend is under the regula-
tion of a neuropeptide termed pheromone biosynthesis
activating neuropeptide (PBAN).1ꢀ4 The direct action of
5—10
PBAN has been demonstrated bystudies in vitro
showing stimulation of pheromone production in the
presence of synthetic peptide by isolated pheromone
gland tissue. The exact tissue involved was delineated as
the intersegmental membrane which is situated between
the 8th and 9th abdominal segments.11,12 In Helicoverpa
armigera, we have shown that octopamine (OA) and
clonidine significantlyinhibit the pheromonotropic
action due to PBAN in intact moths and decapitated
moths, as well as pheromone gland incubations in
vitro.11ꢀ13 The inhibition was also reflected in a sig-
nificant inhibitoryeffect on intracellular cAMP levels
which were stimulated in the presence of PBAN. This
The biogenic monoamine OA, which has been found in
high concentrations in various insect tissues, is the
monohydroxylated analogue of the vertebrate hormone
noradrenaline. It has been found that OA is present in a
high concentration in various invertebrate tissues.18
This multifunctional and naturallyoccurring biogenic
amine has been well studied and established as: (1) a
neurotransmitter, controlling the fireflylight organ and
endocrine gland activityin other insects; (2) a neuro-
hormone, inducing mobilization of lipids and carbohy-
drates; (3) a neuromodulator, acting peripherallyon
*Corresponding author. Tel.: +81-92-642-2856; fax. +81-92-642-
2858; e-mail: ahirasim@agr.kyushu-u.ac.jp
0968-0896/03/$ - see front matter # 2002 Elsevier Science Ltd. All rights reserved.
PII: S0968-0896(02)00320-6

96
A. Hirashima et al. / Bioorg. Med. Chem. 11 (2003) 95–103
different muscles, fat body, and sensory organs such as
corpora cardiaca and the corpora allata, and (4) a cen-
trallyacting neuromodulator, influencing motor pat-
terns, habituation and even memoryin various
invertebrate species.19,20 The action of OA is mediated
through various receptor classes which are coupled to
G-proteins and is specificallylinked to an adenlyate
cyclase. Thus, the physiological actions of OA have
been shown to be associated with elevated levels of cyc-
lic AMP.21 Three different receptor classes OAR1,
OAR2A and OAR2B have been distinguished from
non-neuronal tissues.22 In the nervous system of locust,
a particular receptor class was characterized and estab-
lished as a new class OAR3 bypharmacological inves-
tigations of the [3H]OA binding site using various
agonists and antagonists.23ꢀ27 Recentlymuch attention
has been directed at the octopaminergic system as a
valid target in the development of safer and selective
pesticides.28ꢀ30 Structure–activitystudies of various
types of OA agonists and antagonists were reported
using the nervous tissue of the migratorylocust, Locusta
migratoria L.23ꢀ27 However, information on the struc-
tural requirements of these OA-agonists and antagonists
for OA-receptor ligands with high activityis still lim-
ited. Tyramine (TA) showed stronger inhibitory activity
(0.63 mM) of pheromone production than that of OA
(5.11 mM) in P. interpunctella.17 While it has been
clearlyshown that OA acts as a neuromediator in var-
ious insects, there have onlybeen suggestions that TA
might be a neuromediator in invertebrates.31 Conse-
quently, TA could not only be a biosynthetic precursor
to OA, but mayalso playan independent role as a
neuromediator, although it still remains to be clarified.
Thus, the pheromonostatic receptor, acting in a neuro-
modulatoryrole, represents a novel tpye of OA/TA
receptor. It is therefore of critical importance to provide
information on the pharmacological properties of this
OA/TA receptor types and subtypes.
Our interest in octopaminergic agonists was aroused by
the results of quantitative structure–activityrelation-
ships (QSARs) studyusing various phsyicochemical
parameters as descriptors32,33 or receptor surface mod-
els.34 Furthermore, molecular modeling and conforma-
tional analysis were performed in Catalyst/Hypo35 to
gain a better knowledge of the interactions between
octopaminergic antagonists36 and OAR3 in order to
identifythe conformations required for binding activity.
Similar procedure was repeated using OA agonists.37
The current work is aimed to generate 3-D chemical
function-based hypotheses from some set of OA/TA
agonists responsible for the inhibition of sex-pheromone
production in P. interpunctella, which would identify
specific and sensitive inhibitors of pheromone biosynth-
esis in the future drug design.
Results and Discussion
Assesment of 3-D-QSAR for inhibitory activity
A set of 36 molecules, that are responsible for the inhi-
bition of sex-pheromone production in P. interpunctella,
was selected as the target training set. Their chemical
structures and experimental activities are listed in Fig-
ure 1 and Table 1. 2-(Substituted benzylthio)-2-oxazo-
line (SBO) 4 had the highest potency, followed by SBO
derivative 10 and SBO 7, substituted with 3-CH3, 4-CH3
and 3-CF3, respectively, in inhibition of de novo pher-
omone production (Table 1). Activities of the agonists
Figure 1. Structures of OA/TA agonists in the training and test sets.

A. Hirashima et al. / Bioorg. Med. Chem. 11 (2003) 95–103
Table 1. OA/TA agonists and their biological activityused in this study
97
Compda
R1
R2
Mp (ꢁC)
Activity(mM) b
1
2
3
4
5
6
7
8
SBO
SBO
SBO
SBO
SBO
SBO
SBO
SBO
SBO
SBO
SBO
SBO
SBO
AIO
AIO
AIO
AIO
AIO
AIO
AIO
AIO
AIO
AIO
AIO
AIT
AIT
AIT
AIT
AIT
AIT
BAT
BAT
BAT
SBT
SBT
SBT
H
2-CH3
3-Cl
3-CH3
3-CH3
3-CH3
3-CF3
3-CH3O
4-F
4-CH3
4-CH3
4-CH3O
4-tBu
H
2-CH3
2-CH2CH3
2-CH(CH3)2
2,6-Cl2
H
H
H
Oil
Oil
Oil
Oil
Oil
Oil
Oil
Oil
Oil
Oil
Oil
Oil
Oil
132–134
Oil
61–63
89–91
175–176
Oil
1.97 (1.15–4.61)
0.25 (0.19–0.31)
50
H
5–CH3
4,40–(CH3)2
0.088 (0.057–0.153)
0.68 (0.55–0.82)
0.25 (0.19–0.34)
0.124 (0.082–0.183)
0.96 (0.75–1.24)
0.56 (0.44–0.70)
0.096 (0.070–0.129)
0.29 (0.22–0.38)
0.65 (0.52–0.81)
2.34 (1.25–5.24)
50
H
H
H
H
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
4,40–(CH3)2
H
H
50
4.67 (3.92–5.56)
6.46 (5.66–7.35)
50
2,6-(CH3)2
2-CH3,6-CH2CH3
2-CH3,6-CH(CH3)2
2,6-(CH2CH3)2
2-CH2CH3,6-CH(CH3)2
2,6-[CH(CH3)2]2
H
2,4-(CH3)2
2,4,6-(CH3)3
2,6-(CH3)2
2,6-(CH2CH3)2
2-CH2CH3,6-CH(CH3)2
2-CH3
50
102–104
Oil
3.07 (2.14–4.67)
5.83 (5.40–6.26)
1.65 (1.30–2.05)
0.28 (0.19–0.37)
1.02 (0.79–1.29)
50
2.12 (1.82–2.48)
3.03 (2.48–3.65)
1.38 (1.07–1.75)
0.27 (0.18–0.40)
0.82 (0.56–1.15)
50
172–174
103–105
169–170
174–176
106–108
99–101
169–171
72–74
138–140
119–120
65–66
102–103
Oil
Oil
Oil
168–169
Oil
Oil
Oil
Oil
167–168
3-CH3
2.20 (1.81–2.68)
50
50
2,3-(OCH3)2
2-CH3
3-CH3
50
50
4-CH3
AII
BOA
CAO
CDM
DIP
HQO
4.05 (2.66–6.76)
0.93 (0.74–1.21)
1.82 (1.09–3.38)
3.90 (3.01–4.90)
50
50
aSBOs 1–13 were prepared from oxazolidine-2-thione and substituted benzylamine in the presence of sodium hydride. The compounds reported here
have been prepared according to the ref 17. AIOs 14–24 were obtained by cyclodesulfurizing the corresponding thiourea with yellow mercuric oxide.
AITs 25–30 and BATs 31–33 were synthesized by cyclization of the corresponding thiourea with concd. Hydrochloric acid. AII 37 was prepared by
refluxing the corresponding aniline and 1-acetyl-2-imidazolidone in phosphoryl chloride followed by hydrolysis. BOA 38 was prepared from oxa-
zine-2-thione and m-methyl benzylamine in the presence of sodium hydride. CAO 39 were prepared from oxazolidine-2-thione and cinnamylamine in
the presence of sodium hydride. DIP 41 was obtained byrefluxing d-valerolactam and the corresponding aniline in phosphoryl chloride. HQO 42
was obtained from SBO and o-aminobenzoic acid in the presence of sodium hydride. CDM 40 was a gift from Nihon Nohyaku Co. Ltd. and used
after purification bycolumn chromatographyon silica gel.
bThe intersegments of P. interpunctella were incubated individuallyin 10 mL medium containing 0.5 mCi [1-14C]acetate in the presence of synthetic
Hez-PBAN (0.5 mM) and test compounds at room temperature for 3 h, maintaining the photoperiod. In order to measure the incorporation of
[1-14C]acetate into pheromone components, the glands were extracted in hexane, which was washed with water, and the amount of radioactivityof
the hexane extract was measured using a LSC after adding scintillation cocktail. The response obtained in control pheromone gland incubated with
Hez-PBAN (0.5 mM) alone is regarded as 100%. The inhibitoryactivityof OA was 5.11 mM.
are expressed as their ID50 values in mM and activities
range over three orders of magnitude (min 0.088 mM
and max 50 mM). This set included a varietyof types of
molecules and for this types of training set, the use of
the hypothesis generation tool was appropriate. This
tool builds hypotheses (overlays of chemical features)
for which the fit of individual molecules to a hypothesis
can be correlated with the molecule’s affinity.
generalized CHARMm-like force field38ꢀ41 implemented
in the program. It was found that hypotheses contain
good correlation with hydrogen-bond acceptor (HBA),
hydrogen-bond acceptor aliphatic (HBAl), hydrophobic
(Hp), hydrophobic aromatic (HpAr) and hydrophobic
aliphatic (HpAl). The characteristics, such as cost, root
mean square (RMS) and the regression constant r, of
the 10 lowest cost hypotheses are listed in Table 2. The
statistical relevance of the various hypotheses obtained
is assessed on the basis of their cost relative to the null
hypothesis and their r.32 The total fixed cost of the run
was 132.37 and the cost of the null hypothesis was
The 3-D-QSAR studywas performed with the Catalyst
(Version 4.0) package. The geometryof each compound
was built with a visualizer and optimized byusing the

98
A. Hirashima et al. / Bioorg. Med. Chem. 11 (2003) 95–103
Table 2. Predicted activity(ID 50) from 10 best hypotheses against actual inhibitory activity data for 36 OA/TA agonists
Comp.
Exp.a (mM)
Conf.a
Hypotheses
1
2
3
4
5
6
7
8
9
10
1
2
4
5rb
5sb
6
7
8
9
1.97
0.25
0.088
0.68
0.68
0.25
0.12
0.96
0.56
0.096
0.29
0.65
2.34
50
50
4.7
6.5
3.1
5.8
1.65
0.28
1.0
10
18
29
26
10
18
12
21
32
22
16
9
31
2
4
6
8
13
2
9
14
12
5
4
2
7
11
7
1.00
0.86
0.081
0.17
0.57
0.093
0.18
0.36
0.89
0.51
0.36
0.26
1.10
0.89
0.81
0.074
0.20
0.57
0.10
0.23
0.31
0.94
0.46
0.41
0.22
1.30
1.10
0.72
0.078
0.19
1.1
0.11
0.29
0.30
0.97
0.66
0.62
0.20
1.60
0.83
0.86
0.078
0.20
0.58
0.11
0.26
0.29
1.00
0.43
0.42
0.21
1.40
0.90
0.76
0.067
0.15
0.50
0.078
0.15
0.31
0.78
0.90
0.89
0.22
0.97
1.00
0.89
0.083
0.18
0.63
0.094
0.18
0.37
0.92
0.51
0.36
0.26
1.10
65
8.1
11
9.5
1.6
2.5
1.80
2.4
3.2
23
5.9
2.9
4.0
1.60
0.70
13
4.2
7.5
0.87
0.59
0.048
0.18
0.93
0.072
0.22
0.32
0.79
1.30
0.97
0.22
1.30
0.83
0.62
0.058
0.15
0.71
0.073
0.18
0.30
0.71
1.00
0.89
0.23
1.10
52
16
9.7
6.8
2.3
3.7
1.10
1.7
3.0
0.78
0.78
0.074
0.15
0.40
0.081
0.16
0.27
0.73
0.81
0.78
0.21
0.87
0.75
0.73
0.061
0.18
0.54
0.085
0.20
0.25
0.80
0.92
0.83
0.19
1.00
53
12
7.8
7.7
1.7
2.6
1.30
2.0
3.1
52
4.5
2.7
4.3
1.40
1.50
11
6.3
6.4
10
11
12
13
14
15
16
17
20
21
22
23
24
25
26
27
28
29
30
31
32
33
37
38
39
40
42
64
16
11
56
14
10
53
20
12
54
15
11
62
15
10
47
17
10
53
16
13
9.2
8.0
7.5
7.8
8.5
6.4
8.2
1.6
2.5
1.70
2.4
3.2
22
5.6
2.7
3.7
1.60
0.68
12
4.3
9.0
1.6
2.3
1.50
2.2
3.2
48
4.2
2.4
3.5
1.60
1.60
12
6.0
7.4
1.8
2.6
0.86
1.2
1.7
42
8.4
3.5
1.7
2.3
1.50
2.1
3.0
54
3.7
2.5
3.5
1.50
1.80
12
6.2
7.1
1.4
2.3
1.50
2.2
2.8
21
5.2
2.7
4.3
1.40
0.59
11
3.9
8.3
2.0
3.4
0.95
1.5
2.7
37
4.5
3.3
1.9
3.0
1.40
2.2
3.4
26
4.1
3.0
50
2.1
3.0
41
3.8
3.6
1.4
5.3
5.4
5.7
4.8
1.40
1.10
0.27
0.82
50
2.2
50
0.93
0.90
9.1
7.8
5.8
0.98
1.10
7.5
6.1
5.0
1.10
1.30
7.8
3.6
5.4
15
17
20
25
38
4
11
3.3
7.7
3.4
1.0
1.70
6.0
4.1
3.9
3.8
2.9
3.6
3.5
4.2
2.9
3.2
3.2
0.93
1.82
3.9
0.26
2.00
6.5
0.20
1.80
4.5
0.21
1.10
8.7
0.17
1.90
4.0
1.50
1.80
6.0
0.26
2.00
6.7
1.10
0.91
4.4
1.20
0.96
6.6
0.92
1.60
3.8
4
2
50
63
53
31
54
61
65
61
55
53
54
aExp., experimental data (ID50 in mM for inhibition of pheromone production); Conf., number of conformational models.
b5r has R configuration and 5s S configuration.
171.34. The cost range between best hypothesis 1 and
null hypothesis was 19.97. The cost range over the 10
generated hypotheses was 2.04. Hypotheses 1 and 2
consist of the same chemical-feature functions as an
HBA, an HBAl, an Hp, an HpAl and an HpAr. The Hp
of hypotheses 1 and 2 is replaced by an HpAl, leading to
hypotheses 5 and 7.
ated hypothesis and that of the null hypothesis, the more
likelyit is that the hypothesis reflects a chance corre-
lation.41 The correlation between observed and estimated
inhibitoryactivities is satisfactoryin the three hypotheses.
The predicted activities are in the right order and parallel
the values actuallyobserved (Table 2).
Table 3. Characteristics of 10 lowest cost hypotheses from 36 OA/
TA agonists (cost of ideal hypothesis: 132.37, cost of null hypothesis:
171.34)
Validation of the hypothesis
The hypotheses are used to estimate the activities of the
training set. Those activities are derived from those
conformers displaying the smallest RMS deviations
when projected onto the hypothesis. Hypotheses 1–10
shared five common features located at almost exactly
the same 3-D coordinates. The qualityof the correlation
among the data in the training set is given bythe RMS
score which was normalized bythe log (uncertainty) and
r. All calculated activities from 10 best hypotheses and
the number of generated conformations for each mole-
cule are listed in Table 2. Even though hypothesis 1 has
a lower cost value than others and the RMS index for
hypothesis 1 is verysmall as 0.873, theyhave nearlyno
difference in r and RMS (Table 3). Roughlyspeaking,
the greater the difference between the cost of the gener-
Hypotheses
Featurea
Cost RMS r
1
2
3
4
5
1
2
3
4
5
6
7
8
HBA HBAl Hp HpAl HpAr 151.37 0.873 0.855
HBA HBA Hp HpAl HpAr 152.05 0.879 0.852
HBA HBAl Hp HpAl HpAr 152.14 0.909 0.841
HBAl HBAl Hp HpAl HpAr 152.35 0.883 0.851
HBA HBAl HpAl HpAl HpAr 152.49 0.915 0.839
HBA HpAl Hp
Hp HpAr 152.52 0.909 0.841
HBA HBAl HpAl HpAl HpAr 152.74 0.934 0.831
HBA HBA HpAl HpAl HpAr 152.85 0.920 0.837
HBAl HBAl HpAl HpAl HpAr 153.25 0.912 0.840
HBAl HBAl Hp HpAl HpAl 153.41 0.924 0.835
9
10
aHBA, hydrogen-bond acceptor; HBAl, hydrogen-bond acceptor ali-
phatic; Hp, hydrophobic; HpAl, hydrophobic aliphatic; HpAr,
hydrophobic aromatic.

A. Hirashima et al. / Bioorg. Med. Chem. 11 (2003) 95–103
99
The small cost range observed here maybe due to two
factors, namelymolecules in the training set are fairly
rigid and have a high degree of structural homology.
Due to the relativelysmall range between the costs for
an ideal versus null hypothesis and due moreover to the
placement of the identified hypotheses within this range,
special care was taken to test for chance correlation.
The hypothesis 1 turns out to be the best measure in the
test set for the whole training set as attested bythe rea-
sonablygood correlation between observed and esti-
mated activitities. Hence, the hypothesis 1 was regressed
using each molecule in its most chemicallyreasonable
conformation. All test compounds 3, 8, 19, 34–36 and
41 were overestimated byhypothesis 1, which had the
highest r (0.855) and the smallest RMS. The type of
compounds used in preparing Catalyst hypothesis is still
limited and it will need improvement to design new
molecules.
Figure 2. Mapping of 4 onto hypothesis 1, which contains an HBA
(green right), an HBAl (green left), an Hp (light blue), an HpAr (blue)
Receptor–Drug Interaction
group of 2 dose not overlap with anyfeatures of
hypothesis 1 (Fig. 4). Thus, the activity of 2 is much
lower than 4. In 2-(arylimino)thiazolidine (AIT) with
shorter bridge between phenyl and oxazoline rings than
SBO (Fig. 5), the phenyl group of 29 overlaps with the
HpAr feature of hypothesis 1, whereas the bridge nitro-
gen atom serves as an HBA, between a phenyl and
thiazolidine rings. However, the HBAl of hypothesis 1
does not fit with 29, leading to lower inhibitoryactivity
than 4. The Hp and HpAl features overlap neatlywith
the two ethyls, respectively. AIO 29 has a phenyl and
oxazolidine ring directlyconnected bynitrogen and
thus, is less flexible than SBO 4, because 29 has less (11)
conformational flexibilitythan that (29) of 4.
QSAR modeling is an area of research pioneered by
Hansch and Leo,42 and Hansch and Fujita.43 QSAR
attempts to model the activityof a series of compounds
using measured or computed properties of the com-
pounds. More recently, QSAR has been extended by
including 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-
dimensional information, one can attempt to build a
hypothetical model of the receptor that can provide
insight about receptor characteristics. Such a model is
known as a Hypo, which provides three-dimensional
information about a putative receptor site. Catalyst/
Hypo was useful in building 3-D pharmacophore mod-
els from the binding activitydata and conformational
structure. It can be used as an alternative for QSAR
methods.
The present studies on OA/TA agonists demonstrate
that two HBA sites and three hydrophobic sites located
on the molecule seem to be essential in inhibition of
pheromone production of P. interpunctella. Generally,
Figure 2 depicts the conformation of the most active
SBO 4 in the training set mapped onto the lowest cost
hypothesis 1. The molecule maps well onto all five
hypotheses features in a similar way in hypotheses 1–10
(data not shown) and therefore these hypotheses are
considered to be equivalent. Roughlyspeaking,
hypotheses 1–10 have good similarity in 3-D spatial
shape. The phenyl group of 4 overlaps with the HpAr
feature of hypothesis 1, whereas the oxygen and nitro-
gen atoms of oxazolidine ring serve as an HBA and an
HBAl, respectively. The Hp feature overlaps neatly with
the methyl and the HBAl with methylenethio group,
respectively. Thus, the most active OA/TA agonist 4 in
the training set maps closelywith the statisticallymost
significant hypothesis 1, which is characterized by five
features (Fig. 2) and the predicted activityof 4 by
hypothesis 1 is reasonable (Table 2). Meanwhile, the
oxygen and nitrogen atoms of oxazolidine ring in SBO
10 serve as an HBAl and an HBA, respectively(Fig. 3).
The Hp does not fit slightlywith the methyl group of 10,
which shows lower activitythan 4. Similarly, the methyl
Figure 3. Mapping of 10 onto hypothesis 1, which consists of an HBA
(green left), an HBAl (green right), an Hp (light blue), an HpAr (blue)
and an HpAl (blue).

100
A. Hirashima et al. / Bioorg. Med. Chem. 11 (2003) 95–103
to investigate further the mechanisms of OA/TA recep-
tor–ligand interactions.
Conclusions
The inhibitors of pheromone production have been
reported, which showed also OA-agonist activities.17
Those OA agonist activities were nonadditive with
respect to the activityof a maximallyeffective con-
centration of OA.29 The activities were inhibited by
several OA antagonist, the rank-order abilityof which
to block the OA agonists was identical to the rank-order
abilityof the same antagonists to block OA. These
results suggest that the activities of those compounds
are due to their OA-agonist action. Although the inhi-
bitoryactivities of OA agonists were in mM range for P.
interpunctella, theywere in nM range for H. armigera.48
The result suggests spiecies specificityof the activities,
which still remains to be clarified for development of
pest-control agents.
Figure 4. Mapping of 2 onto hypothesis 1, which consists of an HBA
(green right), HBAl (green left), an Hp (light blue), an HpAr (blue)
and an HpAl (blue).
more active molecules map onto all the features of the
hypothesis. Conversely, compounds that are estimated
to have low activitymap poorlyto the hypothesis.
Three-dimensional pharmacophore hypotheses were
built from a training set of 36 OA/TA agonists.
Hypotheses were obtained from this study and applied
to map with the active or inactive compounds to explain
the mechanism of OA/TA receptor–ligand interactions.
Important features were found such as HBA, HBAl,
Hp, HpAr and HpAl of the surface-assessable models.
Two HBAs and three hydrophobic features are the
minimum components of an effective OA/TA agonistic
binding hypothesis. Meanwhile, three-dimensional
pharmacophore hypotheses were built from a set of 43
agonists against OAR3 in locust nervous tissue.49
Among the 10 chemical-featured models generated by
program Catalyst/Hypo, a hypothesis including an
HBA, three Hps and an HpAl features was considered
to be important and predictive in evaluating OAR3
agonists. It was found that more active agonists map
well onto all the features of the hypotheses, meanwhile
for some inactive compounds, their lack of binding affi-
nityis primarilydue to their inabilityto achieve an
energeticallyfavorable conformation shared bythe
active compounds in order to fit the 3-D common fea-
ture pharmacophore keys.
44ꢀ47
The predictive character of this five-feature hypothesis
was further assessed using candidate molecules, whose
structures are shown in Fig. 1, outside of the training set
(Table 4). The best statisticallysignificant hypothesis 1
was applied to access the activities of those OA agonists
and the predicted values of the molecules are listed in
Table 4. It will need improvement to design new mole-
cules. It is necessaryto include various types of com-
pounds with varietyof activityto obtain better
hypothesis and predictions before designing new mole-
cules in the future. The hypothesis 1 is considered to be
used in designing new leads for hopefullymore active
compounds. Further research on the comparison of the
3-D hypotheses from more potent OA/TA agonists as
well as those generated from the corresponding data
from various insect might be interesting and stimulating
Based upon this study, several three-dimensional phar-
macophore models for the OA/TA agonists–receptor
interactions have been proposed. Those hypotheses are
considered to be useful in designing new leads for
hopefullymore active compounds, although the num-
bers of compounds tested are still limited. Hence, a fur-
ther comparison studyof 3-D hpyothesis models of
agonists responsible for the inhibition of sex-pheromone
production in P. interpunctella is in progress and expec-
ted to clarifythe mode of action of these compounds
acting on the OA/TA receptor. Such work will surely
help elucidate the mechanisms of OA/TA receptor–
ligand interactions. The above hypotheses studies show
that agonists with certain substituents can be potential
ligands to OA/TA receptors. Theymayhelp to point the
waytowards developing extremelypotent and relatively
specific OA/TA agonists, leading to potential pest-con-
Figure 5. Mapping of 29 onto hypothesis 1, which consists of an HBA
(green right), an HBA (green left), an Hp (light blue), an HpAr (blue)
and an HpAl (blue).

A. Hirashima et al. / Bioorg. Med. Chem. 11 (2003) 95–103
101
Table 4. Predicted activityof OA/TA agonists from hypothesis 1
dering stage were pupated between pieces of paper car-
ton and the resulting pupae were sexed and males and
females were emerged separately. Emerged virgin
females were staged according to age.
Compound
Conf.
ID50 (mM)
Experimental
Error
Predicted
3
28
4
2
53
42
19
15
50
50
50
50
50
50
50
22
44
38
27
9.8
25
29
ꢀ2.2
ꢀ1.1
ꢀ1.3
ꢀ1.9
ꢀ5.1
ꢀ2.0
ꢀ1.7
18
19
34
35
36
41
In vitro pheromone-production bioassay
Compounds were tested for inhibitoryspecificityusing a
modified radiochemical bioassayto monitor de novo
pheromone production.17 Abdominal tips, containing
the eighth and ninth abdominal segments with the
attached intersegmental membrane, were removed from
1-day-old virgin females under sterile conditions during
the first-third h scotophase, using a dim red light for
illumination. After preincubation in Pipes-buffered
incubation medium48 for 30 min, the intersegments were
dried on tissue paper and then transferred individually
to 10 mL medium containing 0.5 mCi [1-14C]acetate in
the presence or absence of 0.5 mM synthetic Hez-PBAN
and test compounds. All incubations for pheromone
production were performed at room temperature,
maintaining the photoperiod. After the required incu-
bation period (3 h) in order to measure the incorpora-
tion of [1-14C]acetate into pheromone components, the
glands were extracted in hexane, which was washed with
water, and the amount of radioactivityof the hexane
extract was measured using a liquid scintillation counter
(LSC, Beckman LS 6500 multipurpose liquid scintilla-
tion analyzer) after adding scintillation cocktail (Clear-
sol I). ID50s were calculated bya sigmoidal curve-fitting
program designed for log dose-probit activityanalyses
using a Macintosh personal computer system.
Conf., number of conformational models. In case predicted activityis
overestimated, error is obtained bycalculating experimental activity
divided bypredicted value and indicated byminus.
trol agents. In order to optimize the activities of these
compounds as OA/TA agonists, more detailed experi-
ments are in progress.
Experimental
Synthesis of OA/TA agonists
The compounds reported here have been prepared
according to the ref 17. SBOs 1–13 were prepared from
oxazolidine-2-thione and substituted benzylamine in the
presence of sodium hydride. 2-(Arylimino)oxazolidines
(AIOs) 14–24 were obtained bycyclodesulfurizing the
corresponding thiourea with yellow mercuric oxide.
AITs 25–30 and 2-(substituted benzylamino)-2-thiazo-
lines (BATs) 31–33 were synthesized by cyclization of
the corresponding thiourea with concd hydrochloric
acid. 2-(2,6-Diethylphenylimino)imidazolidine (AII) 37
was prepared byrefluxing the corresponding aniline and
1-acetyl-2-imidazolidone in phosphoryl chloride fol-
lowed by hydrolysis. 2-(3-methyl benzylthio)-2-oxazine
(BOA) 38 was prepared from oxazine-2-thione and m-
methyl benzylamine in the presence of sodium hydride.
2-(Cinnamylthio)-2-oxazoline (CAO) 39 was prepared
from oxazolidine-2-thione and cinnamylamine in the
presence of sodium hydride. 2-(2,6-Diethylphenylimi-
no)piperidine (DIP) 41 was obtained byrefluxing d-
valerolactam and the corresponding aniline in phos-
phoryl chloride. 4H-Quinazolin-4-one derivative (HQO)
42 was obtained from SBO and o-aminobenzoic acid in
the presence of sodium hydride. The structures of the
Computational details
Hypothesis generation. All calculation was conducted on
a Silicon Graphics O2, runnning under the IRIX 6.5
operating system. Hypotheses generation and its func-
tionality is available as part of Catalyst/Hypo modeling
environment from Accelyrys (San Diego, USA). Mole-
cules were edited using the Catalyst 2D/3-D visualizer.
The Catalyst model treats molecular structures as tem-
plates consisting of strategicallypositioned chemical
functions that will bind effectivelywith complementary
functions on receptors. The biologicallymost important
functions are deduced from a set of compounds that
cover a broad range of activity. Catalyst automatically
generated conformational models for each compound
using the Poling Algorithm.38ꢀ40 Diverse conforma-
tional models for each compound were generated such
that the conformers covered accessible conformational
space defined within 20 kcal of the estimated global
minimum. The models emphasized a conformational
diversityunder the constraint energythreshold above
the estimated global minimum based on use of the
CHARMm force field.38ꢀ41 Molecular flexibilityis taken
into account byconsidering each compound as a col-
lection of conformers representing a different area of
conformational space accessible to the molecule within
a given energy range. Catalyst provides two types of
conformational analysis: fast and best quality. Fast
option was used, specifying 250 as the maximum num-
ber of conformers.
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 byelemental analy-
tical data. Chlordimeform (CDM, 96% pure) 40 was a
gift from Nihon Nohyaku Co. Ltd. (Osaka, Japan) and
used after purification bycolumn chromatographyon
silica gel.
Insect culture
The colonyof P. interpunctella was raised on a diet of
80% ground rice, 10% glycerin, 5% brewer’s yeast and
ꢁ
5% honeyat 28 C and 70% RH in a 14:10 (light/dark)
photoperiod as reported previously.50 Larvae of wan-

102
A. Hirashima et al. / Bioorg. Med. Chem. 11 (2003) 95–103
The molecules associated with their conformational
models was submitted to Catalyst hypothesis genera-
tion. The present work shows how a set of various OA/
TA agonists, responsible for the inhibition of sex-pher-
omone production in P. interpunctella, maybe treated
statisticallyto uncover the molecular characteristics
which are essential for high activity. These character-
istics are expressed as chemical features disposed in
three-dimensional space and are collectivelytermed a
hypothesis. Hypotheses approximating the pharmaco-
phore are described as a set of features distributed
within a 3-D space. This process onlyconsidered surface
accessible functions such as HBA, HBAl, hydrogen-
bond donor (HBD), Hp, HpAr, HpAl, negative charge,
positive charge, ring aromatic (RA), negative ionizable
(NI) and positive ionizable (PI).51 A preparative test
was performed with these features. NI and PI were used
rather than negative charge and positive charge in order
to broaden the search for deprotonated and protonated
atoms or groups at physiological pH. Furthermore, in
order to emphasize the importance of an aromatic
group corresponding to the phenol moietyof test com-
pounds, RA which consists of directionalitywas chosen
to be included in the subsequent run. The hypothesis
generator was restricted to select onlyfive features due
to the molecule’s flexibilityand functional complexity.
For molecules larger than dipeptides, Catalyst often will
find five-feature hypotheses automatically, but for
smaller molecules, three- or four-feature hypotheses
might be in the majority. Since hypotheses with more
features are more likelyto be stereospecific and gen-
erallymore restrictive models, the total features min
value was set to 5 in order to force Catalyst to search
for five-feature hypotheses.32
chemist assess the validityof a hypothesis. One is the cost
of an ideal hypothesis, which is a lower bound on the cost
of the simplest possible hypothesis that still fits the data
perfectly. The other is the cost of the null hypothesis,
which presumes that there is no statisticallysignificant
structure in the data, and that the experimental activities
are normallydistributed about their mean. Generally, the
greater the difference between the two costs, the higher the
probabilityfor finding useful models. In terms of hypoth-
esis significance, a generated hypothesis with a cost that is
substantiallybelow that of the null hypothesis is likelyto
be statisticallysignificant and bears visual inspection. 52
Acknowledgements
We thank Dr. Ada Rafaeli (Department of Stored Pro-
ducts, Pheromone Research Lab, Volcani Centre,
Israel) for valuable suggestions in rearing P. inter-
punctella. 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. Raina, A. K. Ann. Rev. Entomol. 1993, 38, 329.
2. Ma, P. W. K.; Roelofs, W. Insect Biochem. Molec. Biol.
1995, 25, 467.
3. Rafaeli, A.; Gileadi, C. Invert. Neurosci. 1997, 3, 223.
4. Jurenka, R. A. Arch. Insect Biochem. Physiol. 1996, 33, 245.
5. Soroker, V.; Rafaeli, A. Insect Biochem. 1989, 19, 1.
6. Rafaeli, A.; Soroker, V.; Kamensky, B.; Raina, A. K. J.
Insect Physiol. 1990, 36, 641.
7. Arima, R.; Takahara, K.; Kadoshima, T.; Numazaki, F.;
Ando, T.; Uchiyama, M.; Nagasawa, H.; Kitamura, A.;
Suzuki, A. Appl. Entomol. Zool. 1991, 26, 137.
8. Jurenka, R. A.; Jacquin, E.; Roelofs, W. L. Proc. Natl.
Acad. Sci. U.S.A. 1991, 88, 8621.
9. Fonagy, A.; Matsumoto, S.; Schoofs, L.; Loof, A.; De
Mitsui, T. Biotech. Biochem. 1992, 56, 1692.
10. Matsumoto, S.; Ozawa, R.; Nagamine, T.; Kim, G.-H.;
Uchiumi, K.; Shono, T.; Mitsui, T. Biosci. Biotech. Biochem.
1995, 59, 560.
11. Rafaeli, A.; Gileadi, C. Insect Biochem. Mol. Biol. 1995,
25, 827.
12. Rafaeli, A.; Gileadi, C. Insect Biochem. Mol. Biol. 1996,
26, 797.
13. Rafaeli, A.; Gileadi, C.; Fan, Y.; Meixun, C. J. Insect
Physiol. 1997, 43, 261.
14. Kuwahara, Y.; Kitamura, C.; Takahashi, S.; Hara, H.;
Ishii, S.; Fukami, H. Science 1971, 171, 801.
15. Brady, U. E.; Tumlinson, J. H.; Brownlee, R. G.; Silver-
stein, R. M. Science 1971, 171, 802.
16. Zhu, J.; Ryne, C.; Unelinus, C. R.; Valeur, P. G.; Lof-
stedt, C. Entomol. Exp. Appl. 1999, 92, 137.
17. Hirashima, A.; Eiraku, T.; Watanabe, Y.; Kuwano, E.;
Taniguchi, E.; Eto, M. Pest Manag. Sci. 2001, 57, 713.
18. Axelrod, J.; Saavedra, J. M. Nature 1977, 265, 501.
19. Evans, P. D. In Reviews in Comparative Molecular Neuro-
biology; Heller, S. R., Ed.; Birkhauser: Basel, 1993; p 287.
20. Evans, P. D. In Reviews in Comprehensive Insect Physiol-
ogy Biochemistry Pharmacology, Kerkut, G. A., Gilbert, G.,
Eds.; Pergamon: Oxford, 1985; Vol. 11, p 499.
21. Nathanson, J. A. Mol. Pharm. 1985, 28, 254.
22. Evans, P. D. J. Physiol. 1981, 318, 99.
Validation of the hypothesis. During a hypothesis gen-
eration run, Catalyst considers and discards many
thousands of models. It attempts to minimize a cost
function consisting of two terms. One penalizes the
deviation between the estimated activities of the training
set molecules and their experimental values. The other
penalizes the complexityof the hypothesis. The overall
assumption used is based on Occam’s razor, that
between otherwise equivalent alternatives, the simplest
model is best. Simplicityis defined using the minimum
description length principle from information theory.
The overall cost of a hypothesis is calculated by sum-
ming the cost function consisting of three terms (weight
cost, error cost, and configuration cost). Weight cost is
a value that increases in a Gaussian form as the feature
weight in a model deviates from an idealized value of
2.0. Error cost is a major value that increase as RMS
difference between estimated and measured activities. It
is obtained bycalculating predicted activitydivided by
experimental value, when predicted activityis under-
estimated. In case predicted activityis overestimated, it is
obtained bycalculating experimental activitydivided by
predicted value and indicated byminus. Configuration
cost is a fixed cost which depends on the complexityof
the hypothesis, equal to entropy of the hypothesis space.
Besides providing a numerical score for each generated
hypothesis, Catalyst provides two numbers to help the
23. Roeder, T.; Nathanson, J. A. Neurochem. Res. 1993, 18, 921.

A. Hirashima et al. / Bioorg. Med. Chem. 11 (2003) 95–103
103
24. Roeder, T.; Gewecke, M. Biochem. Pharm. 1990, 39, 1793.
25. Roeder, T. Life Science 1992, 50, 21.
40. Smellie, A.; Kahn, S. D.; Teig, S. L. J. Chem. Inf. Comp.
Sci. 1995, 35, 295.
26. Roeder, T. Br. J. Pharm. 1995, 114, 210.
27. Roeder, T. Eur. J. Pharm. 1990, 191, 221.
28. Jennings, K. R.; Kuhn, D. G.; Kukel, C. F.; Trotto, S. H.;
Whiteney, W. K. Pest. Biochem. Physiol. 1988, 30, 190.
29. Hirashima, A.; Yoshii, Y.; Eto, M. Pestic. Biochem. Phy-
siol. 1992, 44, 101.
41. Brooks, B. R.; Brucolleri, R. E.; Olafson, B. D.; States,
D. J.; Swaminathan, S.; Karplus, M. J. Comput. Chem. 1983,
4, 187.
42. Hansch, C.; Leo, A. In Exploring QSAR: Fundamentals
and Applications in Chemistry and Biochemisry; American
Chemical Society: Washington, DC, 1995.
30. Ismail, S. M. M.; Baines, R. A.; Downer, R. G. H.;
Dekeyser, M. A. Pest. Sci. 1996, 46, 163.
31. Robertson, H. A.; Juorio, A. V. Int. Rev. Neurobiol. 1976,
19, 173.
32. Hirashima, A.; Shinkai, K.; Pan, C.; Kuwano, E.; Tani-
guchi, E.; Eto, M. Pest. Sci. 1999, 55, 119.
33. Pan, C.; Hirashima, A.; Tomita, J.; Kuwano, E.; Tani-
guchi, E.; Eto, M. Internet J. Sci.-Biol. Chem. 1997, 1 (http://
www.netsci-journal.com/97v1/97013/index.htm).
34. Hirashima, A.; Pan, C.; Tomita, J.; Kuwano, E.; Tani-
guchi, E.; Eto, M. Bioorg. Med. Chem. 1998, 6, 903.
35. Catalyst Tutorial, Version 4.0; Molecular Simulations
Inc.: San Diego, CA, 1998.
43. Hansch, C.; Fujita, T. J. Am. Chem. Soc. 1964, 86, 1616.
44. Hirashima, A.; Rafaeli, A.; Gileadi, C.; Kuwano, E. J.
Mol. Graphics Mod. 1999, 17, 43.
45. Hirashima, A.; Morimoto, M.; Kuwano, E.; Taniguchi,
E.; Eto, M. Bioorg. Med. Chem. 2002, 10, 117.
46. Hirashima, A.; Eiraku, T.; Shigeta, Y.; Kuwano, E.;
Taniguchi, E.; Eto, M. Internet Electron. J. Mol. Des 2002, 1,
37 (hyttp://www.biochempress.com).
47. Hirashima, A.; Morimoto, M.; Ohta, H.; Kuwano, E.;
Taniguchi, E.; Eto, M. Int. J. Mol. Sci. 2002, 3, 56 (http://
www.biochempress.com).
48. Rafaeli, A.; Gileadi, C.; Hirashima, A. Pestic. Biochem.
Physiol. 1999, 65, 194.
36. Pan, C.; Hirashima, A.; Kuwano, E.; Eto, M. J. Mol.
Model. 1997, 3, 455.
49. Hirashima, A.; Pan, C.; Kuwano, E.; Taniguchi, E.; Eto,
M. Bioorg. Med. Chem. 1999, 7, 1437.
37. Hirashima, A.; Pan, C.; Kuwano, E.; Taniguchi, E.; Eto,
M. Bioorg. Med. Chem. 1999, 7, 1437.
38. Smellie, A.; Teig, S. L.; Towbin, P. J. Comp. Chem. 1994,
16, 171.
39. Smellie, A.; Kahn, S. D.; Teig, S. L. J. Chem. Inf. Comp.
Sci. 1995, 35, 285.
50. Rafaeli, A.; Gileadi, C. J. Stored Prod. Res. 1995, 31, 243.
51. Greene, J.; Kahn, S.; Savoj, H.; Sprague, P.; Teig, S. J.
Chem. Inf. Comput. Sci. 1994, 34, 1297.
52. Hoffmann, R.; Bourguignon, J.-J. In Complete Listing of
Case Studies by Application Area; Molecular Simulations Inc.
(http://www.msi.com/solutions/cases/notes/CCKB_Peb_full.html).