Synthesis, 3D-QSAR, and docking studies of 1-phenyl-1H-1,2,3-triazoles as selective antagonists for β3 over α1β2γ2 GABA receptors

Bioorganic & Medicinal Chemistry 15 (2007) 5090–5104

Synthesis, 3D-QSAR, and docking studies

of 1-phenyl-1H-1,2,3-triazoles as selective antagonists

for b3 over a1b2c2 GABA receptors

Mohammad Sayed Alam,^aJia Huang,^aFumiyo Ozoe,^a

Fumio Matsumura^band Yoshihisa Ozoe^a,*

^aDepartment of Life Science and Biotechnology, Faculty of Life and Environmental Science, Shimane University, Matsue,

Shimane 690-8504, Japan

^bDepartment of Environmental Toxicology and the Center for Environmental Health Sciences, University of California, Davis,

CA 95616, USA

Received 31 March 2007; revised 13 May 2007; accepted 15 May 2007

Available online 18 May 2007

Abstract—A series of 16 1-phenyl-1H-1,2,3-triazoles with substituents at both the 4- and 5-positions of the triazole ring were syn-

thesized, and a total of 49 compounds, including previously reported 4- or 5-monosubstituted analogues, were examined for their

ability to inhibit the speciﬁc binding of [³H]4⁰-ethynyl-4-n-propylbicycloorthobenzoate (EBOB), a non-competitive antagonist, to

human homo-oligomeric b3 and hetero-oligomeric a1b2c2 c-aminobutyric acid (GABA) receptors. Among all tested compounds,

the 4-n-propyl-5-chloromethyl analogue of 1-(2,6-dichloro-4-triﬂuoromethylphenyl)-1H-1,2,3-triazole showed the highest level of

aﬃnity for both b3 and a1b2c2 receptors, with K_ivalues of 659 pM and 266 nM, respectively. Most of the tested compounds showed

selectivity for b3 over a1b2c2 receptors. Among all 1-phenyl-1H-1,2,3-triazoles, the 4-n-propyl-5-ethyl analogue exhibited the high-

est (>1133-fold) selectivity, followed by the 4-n-propyl-5-methyl analogue of 1-(2,6-dibromo-4-triﬂuoromethylphenyl)-1H-1,2,3-tri-

azole with a >671-fold selectivity. The 2,6-dichloro plus 4-triﬂuoromethyl substitution pattern on the benzene ring was found to be

important for the high aﬃnity for both b3 and a1b2c2 receptors. Comparative molecular ﬁeld analysis (CoMFA) and comparative

molecular similarity indices analysis (CoMSIA) provided similar contour maps, revealing that an electronegative substituent at the

4-position of the benzene ring, a compact, hydrophobic substituent at the 4-position of the triazole ring, and a small, electronegative

substituent at the 5-position of the triazole ring play signiﬁcant roles for the high potency in b3 receptors. Molecular docking studies

suggested that the putative binding sites for 1-phenyl-1H-1,2,3-triazole antagonists are located in the channel-lining 2⁰-6⁰region of

the second transmembrane segment of b3 and a1b2c2 receptors. A diﬀerence in the hydrophobic environment at the 2⁰position

might underlie the selectivity of 1-phenyl-1H-1,2,3-triazoles for b3 over a1b2c2 receptors. The compounds that had high aﬃnity

for b3 receptors with homology to insect GABA receptors showed insecticidal activity against houseﬂies with LD₅₀values in the

pmol/ﬂy range. The information obtained in the present study should prove helpful for the discovery of selective insect control

chemicals.

1. Introduction

tein-coupled metabotropic receptors, to regulate the

function of the nervous system.¹The ionotropic GABA

receptors belong to a family of ligand-gated ion chan-

nels that include nicotinic acetylcholine receptors, gly-

cine receptors, and serotonin type 3 receptors.²The

mammalian ionotropic GABA receptors are heteropen-

tamers composed of ﬁve subunits forming a central

chloride ion channel. GABA binds to the agonist bind-

ing site of GABA receptors in the cell membrane and

inﬂuences the opening of the built-in channels to in-

crease the chloride ion inﬂux into the cell, which results

in suppression of nerve excitation.³To date, 19 subunits

(a_1–6, b_1–3, c_1–3, q_1–3, d, e, p, and h) of GABA receptors

c-Aminobutyric acid (GABA) is the major inhibitory

neurotransmitter ubiquitously distributed in the central

nervous system of both vertebrates and invertebrates.

GABA exerts its inhibitory eﬀect through two types of

membrane receptors, the ionotropic and the G pro-

Keywords: GABA receptor; Antagonist; 3D-QSAR; Homology model;

Phenyltriazoles.

*

Corresponding author. Fax: +81 852 32 6092; e-mail: ozoe-y@life.

shimane-u.ac.jp

doi:10.1016/j.bmc.2007.05.039

M. S. Alam et al. / Bioorg. Med. Chem. 15 (2007) 5090–5104

5091

have been identiﬁed in the nervous system, and have

been classiﬁed according to their sequence identity.^4,5

Although a number of diﬀerent subunit compositions

would have the potential to form the pentamers, only

a limited number of combinations have been identiﬁed.

The a1b2c2 subunit composition is the most abundant,

accounting for about 43% of GABA type A (GABA_A)

receptors in the rat brain.⁶Insect ionotropic GABA

receptors are structurally similar to mammalian

GABA_Areceptors, although the two have diﬀerent

pharmacologies. Insect GABA receptors are not only

localized in the central nervous system but also in the

peripheral nervous system⁷; therefore, anti-GABAergic

insecticides are thought to easily access these peripheral

GABA receptors and thereby exert their insecticidal

eﬀects. At present, three diﬀerent types of ionotropic

GABA receptor subunits have been cloned from several

diﬀerent insect species; that is, RDL (a subunit encoded

by the Rdl (resistant to dieldrin) gene), LCCH3 (ligand-

gated chloride channel 3), and GRD (the GABA_Aand

glycine receptor-like subunit of Drosophila). RDL,

which was ﬁrst isolated from a naturally occurring

dieldrin-resistant strain of Drosophila melanogaster,^8,9

is capable of forming a chloride channel.

Recently, we have also reported the synthesis and

structure–activity relationships of 4- or 5-substituted

1-phenyl-1H-1,2,3-triazoles (Fig. 1) as non-competitive

antagonists that showed signiﬁcant insecticidal action.¹⁹

In the present study, we further synthesized a new series

of 1-phenyl-1H-1,2,3-triazoles with substituents at both

the 4- and 5-positions of the triazole ring to study the

molecular mechanisms of the interaction of phenyl het-

erocyclic compounds with GABA receptors. The aﬃni-

ties of 1-phenyl-1H-1,2,3-triazoles, including the newly

and the previously synthesized series of analogues, for

recombinant human b3 and a1b2c2 GABA receptors

were determined by

a

competition assay using

[³H]EBOB as a probe. The human b3 receptor was used

here as an insect GABA receptor model, since human b3

and insect RDL GABA receptors were shown to have

homologous amino acid sequences in the putative non-

competitive antagonist binding site and similar pharma-

cological properties,^15,17,19and it should also be noted

that such a homo-oligomeric b3 ion channel has not

been identiﬁed in the mammalian brain.²⁰In the present

study, we also performed three-dimensional quantitative

structure–activity relationship (3D-QSAR) analysis to

predict favorable and unfavorable moieties of 1-phe-

nyl-1H-1,2,3-triazoles for high potency. Finally, we

modeled docking of 1-phenyl-1H-1,2,3-triazoles to

GABA receptors to predict the mechanisms of their recep-

tor selectivity. Here we report the results of these eﬀorts.

In order to pharmacologically characterize ionotropic

GABA receptors, a number of non-competitive antago-

nists such as picrotoxinin and 4⁰-ethynyl-4-n-propyl-

bicycloorthobenzoate (EBOB) (Fig. 1) have been

utilized as probes, and a number of structurally diverse

non-competitive antagonists have also been discovered

to date.¹⁰The binding site for the non-competitive antag-

onists is thought to be located within the channel pore,

where they bind to stabilize the channel in a closed form

and exert toxic eﬀects. Fipronil (Fig. 1) is an eﬀective,

commercially available insecticide that acts at insect

ionotropic GABA receptors as well as inhibitory gluta-

mate receptors (glutamate-gated chloride channels).¹¹

Fipronil reduced GABA-induced currents in Drosophila

RDL homo-oligomers stably expressed in a Drosophila

S2 cell line¹²and in native GABA receptors of cockroach

neurons.¹³Fipronil also potently competes for the EBOB

binding site in both native and recombinant GABA

receptors.^14–17The selectivity of ﬁpronil for insect versus

mammalian GABA receptors might be due to the diﬀer-

ence in their subunit compositions,¹⁶which indicates that

insect receptors diﬀer signiﬁcantly from their mamma-

lian counterparts with respect to the structure of the

non-competitive antagonist binding site. In a previous

study, we reported that diverse classes of phenyl hetero-

cyclic compounds also potently and selectively inhibit

[³H]EBOB binding to insect GABA receptors.¹⁸

2. Results and discussion

2.1. Synthesis of 4,5-disubstituted 1-(substituted phenyl)-

1H-1,2,3-triazoles

In the present study, a series of 16 4,5-disubstituted 1-

(substituted phenyl)-1H-1,2,3-triazoles (Table 1) were

synthesized by the 1,3-dipolar cycloaddition between a

phenyl azide and an appropriate unsymmetrical alkyne,

which gave a mixture of two regioisomers with substitu-

ents exchanged at the 4- and 5-positions (Scheme 1). The

regioisomers were separated by silica gel column chro-

matography. Among the disubstituted phenyltriazoles

containing alkyl groups at both the 4- and 5-positions

of the triazole ring, the regioisomers with a long alkyl

group at the 4-position were usually obtained in 1.3-

to 1.6-fold higher yields than their corresponding iso-

mers. The structures of the regioisomers of disubstituted

1-phenyl-1H-1,2,3-triazoles were determined on the ba-

sis of their ¹H NMR, mass spectrum, and elemental

analysis data as well as on the data from our previous

study.¹⁹Compounds 15 and 16 were synthesized from

Figure 1. Structures of EBOB, ﬁpronil, and 1-phenyl-1H-1,2,3-triazoles.

5092

M. S. Alam et al. / Bioorg. Med. Chem. 15 (2007) 5090–5104

Table 1. Potencies of 4,5-disubstituted 1-(2,6-disubstituted 4-triﬂuoromethylphenyl)-1H-1,2,3-triazoles in inhibiting [³H]EBOB binding to b3 and

a1b2c2 GABA receptors

Compound

R¹

R²

R³/R⁴

K_ivalue

RS^a

b3 (nM)

a1b2c2 (lM)

1

2

Et

Me

Et

Cl

Br

Cl

7.08 (4.18–11.60^b)

70.1 (44.6–110.1^b)

4.27 (2.50–7.00^b)

1.94 (1.17–3.08^b)

539 (378–779^b)

13.6 (7.3–23.6^b)

8.07 (5.19–12.25^b)

47.2 (31.3–70.4^b)

128 (78–208^b)

2.53 (1.49–4.17^b)

3450 (1980–7780^b)

0.659 (0.242–1.667^b)

66.7 (34.2–125.0^b)

1.06 (0.67–1.77^b)

>9.15

0.959 (0.619–1.564^b)

0.322 (0.187–0.558^b)

>9.15

150

>131

225

166

3

Et

Me

4

n-Pr

Me

5

n-Pr

Me

>17.0

>673

>1130

>194

>71.5

153

6

n-Pr

Et

>9.15

10

11

12

13

14

15

16

Et

n-Pr

CH₂OH

n-Pr

CH₂Cl

n-Pr

>9.15

n-Pr

CH₂OH

n-Pr

CH₂Cl

0.387 (0.208–0.778^b)

>9.15

>2.65

403

>137

0.266 (0.175–0.412^b)

>9.15

Compounds with a K_ivalue of >8.00 lM for b3 receptors and a K_ivalue of >9.00 lM for a1b2c2 receptors: 7 (R¹= Me, R²= n-Pr, R³/R⁴= Br), 8

(R¹= n-Pr, R²= Me, R³/R⁴= H), and 9 (R¹= Me, R²= n-Pr, R³/R⁴= H).

^aSelectivity for b3 versus a1b2c2 receptors (K_i^a1b2c2/K^b_i³Þ.

^b95% conﬁdence limit.

Scheme 1. Synthesis of 4,5-disubstituted 1-phenyl-1H-1,2,3-triazoles.

chlorination of their corresponding hydroxymethyl ana-

logues (13 and 14) with thionyl chloride in dimethyl-

formamide (DMF) (Scheme 1).

containing a long alkyl group at the 4-position of the tri-

azole ring showed higher aﬃnity for both b3 and a1b2c2

receptors than those containing a long alkyl group at the

5-position. This is consistent with our previous ﬁnding¹⁹

regarding the 4- or 5-monosubstituted phenyltriazoles

that the 4-substituted regioisomer has higher aﬃnity

for receptors than the 5-substituted regioisomer (Table

2). 1-(2,6-Dichloro-4-triﬂuoromethylphenyl)-1H-1,2,3-

triazoles containing an n-propyl group at the 4-position

and a methyl (4), a hydroxymethyl (13), or a chloro-

methyl (15) group at the 5-position displayed the highest

level of aﬃnity for both b3 and a1b2c2 receptors.

Replacement of the methyl group of 4 by a chloromethyl

group at the 5-position to aﬀord 15 resulted in only a

ꢀ3- and a ꢀ1.2-fold enhancement in aﬃnity for b3

and a1b2c2 receptors, respectively. This change in po-

tency resulted in a 2.4-fold increase in selectivity for

b3 over a1b2c2 receptors. When the methyl group was

replaced by a hydroxymethyl group to give 13, the aﬃn-

ity and selectivity of 13 remained almost the same as

those of 4. The aﬃnity levels of 4, 13, and 15 were in

the same range as those of 4-monosubstituted analogues

2.2. Aﬃnities of 4,5-disubstituted 1-phenyl-1H-1,2,3-

triazoles for b3 and a1b2c2 GABA receptors

The aﬃnities of 4,5-disubstituted 1-(substituted phenyl)-

1H-1,2,3-triazoles for human b3 and a1b2c2 receptors

were evaluated by binding assays using [³H]EBOB as a

radioligand. The obtained IC₅₀values were converted

to K_ivalues according to Cheng and Prusoﬀ.²¹The K_i

values of disubstituted phenyltriazoles (1–16) for b3

and a1b2c2 receptors are listed in Table 1. All of the

tested compounds, except 7–9, showed aﬃnity for b3

receptors, whereas most of the compounds showed no

or low aﬃnity for a1b2c2 receptors. Thus, most of the

tested compounds showed higher aﬃnity for b3 recep-

tors than for a1b2c2 receptors. This is also the case

for the 4- or 5-monosubstituted analogues (Table 2).¹⁹

Among the 4,5-disubstituted 1-phenyl-1H-1,2,3-triaz-

oles with diﬀerent alkyl substituents, the regioisomers

M. S. Alam et al. / Bioorg. Med. Chem. 15 (2007) 5090–5104

5093

Table 2. Potencies of 1-(2,6-dichloro-4-triﬂuoromethylphenyl)-4 (or 5)-(alkyl or phenyl)-1H-1,2,3-triazoles in inhibiting [³H]EBOB binding to b3 and

a1b2c2 GABA receptors

Compound

R¹

R²

K_ivalue

a1b2c2 (lM)

RS^b

b3 (nM)^a

17

18

19

21

23

24

25

26

27

28

29

30

31

n-Pr

H

2.14

757

0.470 (0.285–0.796^c)

>9.15

1.01 (0.52–2.26^c)

0.390 (0.219–0.714^c)

1.39 (0.92–2.23^c)

>9.15

1.98 (1.21–3.62^c)

>9.15

1.50 (0.84–3.07^c)

0.616 (0.379–1.043^c)

>9.15

220

n-Pr

H

>12.1

362

470

n-Bu

t-Bu

n-Pen

H

2.79

0.831

27.8

1390

392

H

50.0

>6.58

5.05

>10.6

127

n-Pen

H

n-Hex

H

n-Hex

H

867

c-Hex

(CH₂)₃Cl

Ph

11.8

1.63

128

H

378

H

Ph

>71.5

>19.3

57.1

H

Ph

475

20.3

>9.15

1.16 (0.74–1.93^c)

NH₂

Compounds with a K_ivalue of >8.00 lM for b3 receptors and a K_ivalue of >9.00 lM for a1b2c2 receptors: 20 (R¹= H, R²= n-Bu) and 22 (R¹= H,

R²= t-Bu).

^aK_ivalues for b3 receptors were calculated from IC₅₀values reported in our previous paper¹⁹according to Cheng and Prusoﬀ.²¹

^bSelectivity for b3 versus a1b2c2 receptors (K_i^a1b2c2/K^b_i³Þ.

^c95% conﬁdence limit.

Figure 2. Concentration-inhibition curves of 4 (A), 15 (B), and 21 (C) in [³H]EBOB binding assays for human b3 (j) and a1b2c2 (d) GABA

receptors. The data are means SD for at least three experiments, each performed in duplicate.

with the highest levels of aﬃnity, that is, the 4-n-propyl

(17), the 4-n-butyl (19), the 4-tert-butyl (21), and the 4-

chloropropyl (28) analogues (Table 2). Figures 2A and B

show the dose–response curves of the two most active

disubstituted phenyltriazoles (4 and 15) in b3 and

a1b2c2 GABA receptors.

a1b2c2 receptors and thus little change in selectivity.

Replacement of the 5-methyl group of 1 by an ethyl

group to give the 4,5-diethyl analogue (3) provided a

ꢀ2-fold increase in aﬃnity for b3 receptors, while the

aﬃnity for a1b2c2 receptors remained almost unchanged.

Removal of chlorine atoms at the 2- and 6-positions of

the phenyl group of 4 and 5 to give 8 and 9 resulted in

substantial loss of aﬃnity for both b3 and a1b2c2 recep-

tors. Replacement of the chlorine atoms of 4 with bro-

mine atoms gave 6 with a ꢀ7-fold decreased aﬃnity

for b3 receptors, although the selectivity increased <4-

fold as the aﬃnity for a1b2c2 receptors was substan-

tially lost. Taken together with the low potencies of

36–49 (Table 3), these ﬁndings indicate that the combi-

nation of 2,6-dichloro and 4-triﬂuoromethyl substitu-

tions on the phenyl group of phenyltriazoles is

important for high aﬃnity for both b3 and a1b2c2

receptors, although regarding the selectivity, the bromo

Compounds 10 and 12 were synthesized to examine the

eﬀect of increasing the length of the alkyl chain at the 5-

position of 4. The 4-n-propyl-5-ethyl analogue (10)

showed a 4-fold decreased aﬃnity for b3 receptors and

substantially lost aﬃnity for a1b2c2 receptors, which re-

sulted in a >6-fold increase in selectivity compared to 4,

while the 4,5-dipropyl analogue (12) showed a ꢀ66-fold

decreased aﬃnity for b3 receptors and lost aﬃnity for

a1b2c2 receptors, resulting in a <2-fold decreased selec-

tivity. Replacement of the 4-n-propyl group of 4 by an

ethyl group to produce the 4-ethyl-5-methyl analogue

(1) led to a ꢀ3-fold decreased aﬃnity for both b3 and

5094

M. S. Alam et al. / Bioorg. Med. Chem. 15 (2007) 5090–5104

Table 3. Potencies of 1-(2,4,6-trisubstituted phenyl)-4 (or 5)-n-butyl-1H-1,2,3-triazoles in inhibiting [³H]EBOB binding to b3 and a1b2c2 GABA

receptors

Compound

R¹

R²

R³/R⁴

R⁵

K_ivalue

RS^b

b3 (nM)^a

a1b2c2 (lM)

36

39

40

41

42

43

44

45

47

48

49

n-Bu

H

Cl

H

C„CH

Cl

34.7

7580

82.3

4080

31.2

783

>9.15

>264

>1.21

>111

>2.24

>293

>11.7

>2.87

>3.35

>1.73

>2.60

>4.46

n-Bu

H

n-Bu

H

Cl

H

n-Bu

H

Cl

Br

n-Bu

H

n-Bu

H

Br

n-Bu

H

3190

2730

5300

3520

2050

n-Bu

H

Br

H

CF₃

H

n-Bu

H

Cl

n-Bu

H

Compounds with a K_ivalue of >8 lM for b3 receptors and a K_ivalue of >9 lM for a1b2c2 receptors (MCB: C„C(CH₂)₂CO₂CH₃): 32 (R¹= n-Bu,

R²= H, R³/R⁴= Cl, R⁵= MCB), 33 (R₁= H, R²= n-Bu, R³/R⁴= Cl, R⁵= MCB), 34 (R¹= n-Bu, R²= H, R³/R⁴= H, R⁵= MCB), 35 (R¹= H,

R²= n-Bu, R³/R⁴= H, R⁵= MCB), 37 (R¹= H, R²= n-Bu, R³/R⁴= Cl, R⁵= C„CH), 38 (R¹= n-Bu, R²= H, R³/R⁴= H, R⁵= C„CH), and 46

(R¹= n-Bu, R²= H, R³/R⁴= H, R⁵= CF₃).

^aK_ivalues for b3 receptors were calculated from IC₅₀values reported in our previous paper¹⁹according to Cheng and Prusoﬀ.²¹

^bSelectivity for b3 versus a1b2c2 receptors (K_i^a1b2c2/K^b_i³Þ.

substitution at the 2- and 6-positions on the phenyl ring

is preferable to the chloro substitution (4 vs 6). The rank

orders of disubstituted phenyltriazoles to give high

aﬃnity for b3 and a1b2c2 receptors were the same as

follows: 4-n-Pr, 5-CH₂Cl P 4-n-Pr, 5-Me P 4-n-Pr,

5-CH₂OH P 4,5-diEt P 4-Et, 5-Me P 4-n-Pr, 5-Et for

b3 receptors and 4-n-Pr, 5-CH₂Cl P 4-n-Pr, 5-Me P

4-n-Pr, 5-CH₂OH P4,5-diEt P 4-Et, 5-Me ꢁ 4-n-Pr,

5-Et for a1b2c2 receptors. The rank order of selectivity

was as follows: 4-n-Pr, 5-Et P 4-n-Pr, 5-Me (2,6-Br₂-

4-CF₃-Ph) > 4-n-Pr, 5-CH₂Cl > 4,5-diEt > 4-n-Pr, 5-Me >

4-n-Pr, 5-CH₂OH > 4-Et, 5-Me. Thus, the 4-n-propyl

and 5-chloromethyl groups at the triazole ring of disub-

stituted phenyltriazoles are optimized substituents with

respect to high aﬃnity for b3 receptors (K_i= 659 pM),

but in terms of selectivity, the ethyl group is more eﬀec-

tive than a chloromethyl group as the 5-substituent.

2.3. Aﬃnities of 4- or 5-substituted 1-phenyl-1H-1,2,3-

triazoles for b3 and a1b2c2 GABA receptors

As for the series of monosubstituted phenyltriazoles,

their phenyl or alkyl groups at the 4- or 5-position of

the triazole ring were examined for their eﬀects on the

aﬃnities for a1b2c2 GABA receptors. Table 2 is the list

of the K_ivalues of 4- or 5-substituted 1-(2,6-dichloro-4-

triﬂuoromethylphenyl)-1H-1,2,3-triazoles, including the

previously reported data¹⁹for b3 GABA receptors for

comparison. The binding aﬃnity data revealed that the

Figure 3. CoMFA contour plots for a human b3 GABA receptor

model. The green contours indicate areas where sterically bulky

substituents increase receptor aﬃnity, whereas the yellow contours

indicate areas where sterically bulky groups are detrimental to the

aﬃnity of the 4-regioisomers is higher than that of the

5-regioisomers in every case. Most of the monosubsti-

potency. The blue contours indicate areas where electropositive groups

enhance aﬃnity for receptors, while the red contours represent areas

tuted analogues of 1-(2,6-dichloro-4-triﬂuoromethylphe-

nyl)-1H-1,2,3-triazoles exhibited higher aﬃnity for b3

receptors than for a1b2c2 receptors. Among the mono-

where electronegative moieties are favorable for receptor aﬃnity.

Compounds 15 and 16 are shown as reference molecules in (A) and

(B), respectively.

M. S. Alam et al. / Bioorg. Med. Chem. 15 (2007) 5090–5104

5095

substituted phenyltriazoles, the 4-tert-butyl analogue

(21) showed the highest aﬃnity for both b3

(K_i= 831 pM) and a1b2c2 (K_i= 390 nM) receptors

and a 470-fold selectivity for b3 versus a1b2c2 receptors.

Figure 2C shows the dose–response curves of this highly

active, selective phenyltriazole. Replacement of the

4-tert-butyl group of 21 with a C3–C6 unbranched

alkyl group or a phenyl group resulted in decreased

aﬃnity in the following order: t-Bu > n-Pr P n-Bu >

n-Pen > Ph > n-Hex for b3 receptors and t-Bu P

n-Pr P n-Bu ꢂ n-Pen > n-Hex ꢁ Ph for a1b2c2 recep-

tors. The selectivity for b3 versus a1b2c2 receptors

was as follows: t-Bu > n-Bu > n-Pr > Ph P n-Pen > n-

Hex. These ﬁndings indicate that long alkyl chains do

not conform to the binding site in either b3 or a1b2c2

receptors. Replacement of the 4-phenyl group of 29 by

a cyclohexyl group to give 27 led to a ꢀ11-fold and a

>6-fold increase in aﬃnity for b3 and a1b2c2 receptors,

respectively. Introduction of an amino group into the

5-position of the triazole ring of 29 to aﬀord 31 resulted

in ꢀ6-fold and >8-fold enhancements in aﬃnity for b3

and a1b2c2 receptors, respectively. When the distal-

end methyl group of the 4-n-butyl group of 19 was

replaced with a chlorine atom to produce the 4-n-chloro-

propyl analogue (28), the analogue did not exhibit a

marked increase in either aﬃnity or selectivity. Taken

together, these ﬁndings indicate that the tert-butyl group

at the 4-position of the triazole ring is an optimum sub-

stituent in terms of both high potency for b3 receptors

and selectivity for b3 versus a1b2c2 receptors, followed

by the chloropropyl and n-propyl groups.

failed to give appropriate contour maps, although the

cross-validated and non-cross-validated predictive

abilities were statistically signiﬁcant in both analyses

(data not shown). The reason why we failed to obtain

good contour plots in the case of a1b2c2 receptors

might be the low receptor aﬃnity of all active com-

pounds and the narrow range of the K_ivalues. To gen-

erate a CoMFA model for b3 receptors, we ﬁrst used 37

active phenyltriazoles in the analysis. The cross-vali-

dated (q²) and non-cross-validated (r²) correlation coef-

ﬁcients were 0.508 and 0.709, respectively. In this

analysis, as the diﬀerences between the predictive and

actual pK_i(ꢃlogK_i) values of four compounds (14, 25,

44, and 48) were more than one logarithmic unit, these

compounds were omitted from the ﬁnal analysis.

Although the reason why such compounds showed a

large deviation is not clear at this moment, it should

be noted that they all are analogues with low potency.

The summary of the ﬁnal CoMFA results is shown in

Table 4. The cross-validated and non-cross-validated

correlation coeﬃcients were 0.705 and 0.854, respec-

tively. The standard error of prediction (s_press) and the

standard error of estimate (s) were 0.677 and 0.476 for

leave-one-out cross-validated and non-cross-validated

analysis, respectively. The cross-validated correlation

coeﬃcient is a better criterion to judge the model quality

than the non-cross-validated correlation coeﬃcient, and

the above high q²value indicates the validity of the sug-

gested mode of interaction of phenyltriazoles with b3

receptors.

The CoMFA steric and electrostatic ﬁeld contour maps

for b3 receptors are shown in Figure 3. The green con-

tours indicate areas where sterically bulky substituents

increase receptor aﬃnity, whereas the yellow contours

indicate areas where sterically bulky groups are detri-

mental to the potency. The blue contours indicate areas

where electropositive groups enhance aﬃnity for recep-

tors, while the red contours represent areas where elec-

tronegative moieties are favorable for receptor aﬃnity.

The ﬁnding that the activities of 4, 13, 15, 17, and 21

are higher than those of their corresponding regioisom-

ers is due to the fact that the n-propyl or the tert-butyl

group at the 4-position of the triazole ring occupies

the green contours, whereas the n-propyl or the tert-bu-

tyl group at the 5-position of the same ring interacts

with the yellow contours. The phenyltriazoles contain-

ing a long substituent(s) at either the 4-, 5-, or both po-

sition(s), such as compounds 12, 23, 24, 26,29, and 30,

showed low aﬃnity due to the fact that the substituents

occupy the yellow contours. The red contours around

the 4-position of the benzene ring indicate that a more

negatively charged group at this site would enhance

the receptor aﬃnity. Compounds (e.g., 48) without an

electronegative substituent at the 4-position of the phe-

nyl ring or compounds (e.g., 36, 40, and 42) that have

a lower electronegative substituent (e.g., C„CH, Cl,

and Br) than a triﬂuoromethyl group showed lower

activity than 19, which has a triﬂuoromethyl group at

this position. Blue contours that appeared above and be-

low the benzene ring indicate that a more positive

charge is preferred on the benzene ring. Thus, the strong

electron-withdrawing substituents at the 2-, 4-, and

2.4. CoMFA and CoMSIA 3D-QSAR models

We performed 3D-QSAR analyses to predict the favor-

able and unfavorable moieties of 1-phenyl-1H-1,2,3-

phenyltriazoles for high potency in b3 receptors using

the CoMFA (comparative molecular ﬁeld analysis)²²

and CoMSIA (comparative molecular similarity indices

analysis)²³modules in the SYBYL program. 3D-QSAR

analyses such as CoMFA and CoMSIA are generally

applied to ﬁnd the common features of ligands that

are important for binding to receptors, and the informa-

tion obtained is usually used to design and develop

novel ligands. Both analyses are based on the assump-

tion that changes in the binding aﬃnities of ligands

are related to changes in diﬀerent ﬁelds surrounding

the molecules, such as the steric and electrostatic ﬁelds,

as well as the hydrophobic properties and hydrogen-

bond accepting and donating capabilities. The partial

least squares (PLS) analysis is a multivariate statistical

method that is used to generate QSAR and can be

applied for predicting the binding aﬃnity of ligands.

The CoMFA methodology was ﬁrst introduced by

Cramer et al.,²²and CoMSIA is an extension of the

CoMFA method reported by Klebe et al.²³The diﬀer-

ence between these two methods is in the implementa-

tion of the ﬁelds. In most cases the predictive ability

of CoMSIA is the same or better than, or complemen-

tary to that of CoMFA.^24–26

3D-QSAR studies were conducted successfully only on

b3 GABA receptors. Those on a1b2c2 GABA receptors

5096

M. S. Alam et al. / Bioorg. Med. Chem. 15 (2007) 5090–5104

Table 4. Results of CoMFA and CoMSIA of 1-phenyl-1H-1,2,3-triazole binding to b3 GABA receptors pK_i= a + [CoMFA or CoMSIA ﬁeld terms]

Analysis

a

C^a

n

s

r²

F_c,nꢃcꢃ1

Cross-validated

Relative contribution

s_press

q²

S^b

E^c

H^d

CoMFA

CoMSIA

6.928

6.601

2

5

33

35

0.476

0.332

0.854

0.936

87.755

84.787

0.677

0.757

0.705

0.668

0.694

0.262

0.306

0.465

—

0.273

Compounds 14, 25, 44, and 48 were excluded in CoMFA, and 14 and 48 were omitted in CoMSIA.

^aOptimum component.

^bSteric.

^cElectrostatic.

^dHydrophobic.

6- positions of the phenyl group might make the charge

distribution on the benzene ring relatively electroposi-

tive. This is consistent with the observation that phenyl-

triazoles (e.g., 38, 44, 46, or 48) without electronegative

atoms at the 2-, 4-, or 6-position on the benzene ring or

those (e.g., 6) having substituents with less electron-

withdrawing ability at these positions led to a decrease

or loss of aﬃnity for receptors compared to their corre-

sponding analogues. It is interesting to note that red

contours are present between the 4- and 5-substituents

of the triazole ring. The negative electrostatic regions

at this position probably arise due to the fact that sev-

eral phenyltriazoles, such as 13, 15, 28, and 31, contain

electronegative substituents at either the 4- or 5-position

of the triazole ring. The 3-chloropropyl group of 28 is

acceptable as the n-butyl group of 19, and the introduc-

tion of an amino group into the 5-position of 29 to pro-

duce 31 also led to increased receptor aﬃnity, as the

electronegative substituents probably occupy the elec-

tronegatively favorable regions. These ﬁndings allow us

to speculate that an appropriate electronegative substitu-

ent at these positions would increase receptor aﬃnity.

The CoMSIA steric and electrostatic stdev coeﬃcient

*

contour plots obtained from non-cross-validated PLS

analysis are shown in Figures 4A and B, respectively.

The green contours indicate regions where sterically

bulky substituents are favorable, while the yellow con-

tours show regions where sterically bulky substituents

are unfavorable for receptor aﬃnity. The blue contours

indicate areas where electropositive groups enhance

receptor aﬃnity in this region, while the interaction of

electronegative substituents with areas shown with red

contours increases aﬃnity. Information obtained from

CoMSIA contour maps (Fig. 4A) was similar to that ob-

tained from CoMFA contour maps with respect to steric

eﬀects, but one exception was also noted. The CoMSIA

contour indicates that there is another sterically disfa-

vored region located behind the sterically favored

regions indicated by the green contours around the

4-position of the triazole ring, whereas this region was

not observed in the CoMFA model. This ﬁnding indi-

cates that optimum sizes of substituents for aﬃnity for

receptors are present at both 4- and 5-positions. The

electrostatic contour map of CoMSIA (Fig. 4B) was

comparable to that of CoMFA (Fig. 3) with a few

exceptions. A small red contour between the 4- and 5-

positions of the triazole ring in CoMFA turned into

two large linked red contours in CoMSIA, and the red

contours around the triﬂuoromethyl group at the 4-po-

sition of the benzene ring became smaller. CoMSIA

predicts that the contribution of electronegative substit-

uents to the binding aﬃnity at the 4- and 5-positions of

the triazole ring is high, while the contribution of the

electronegative substituent at the 4-position of the ben-

zene ring is low. The electronegative groups of com-

pounds 13, 15, 16, and 28–31 at the 4- or 5-position of

the triazole ring might have contributed to the genera-

tion of red contours that are present near the triazole

ring.

It is generally thought that molecular alignment changes

exert less eﬀect on the results in CoMSIA, which also af-

fords smoother and more easily interpretable contour

plots using Gaussian type distance dependence with

the molecular similarity indices in the analysis.²³In

addition to the steric and electrostatic ﬁelds of CoMFA,

CoMSIA also deﬁnes clear hydrophobic and hydrogen-

bond donor/acceptor descriptor ﬁelds. In the present

CoMSIA PLS analysis, we considered steric, electro-

static, and hydrophobic ﬁeld descriptors to obtain a b3

receptor model, since most of the substituents of our

compounds contain no hydrogen-bond donor or accep-

tor atom. Thirty-seven active phenyltriazoles were used

to generate a CoMSIA b3 receptor model, and two com-

pounds (14 and 48) showed a large deviation from this

model and were excluded in the ﬁnal analysis. The

excluded compounds are the compounds that also

showed a large deviation in the CoMFA. The reason

for the large deviation is obscure at this time. The results

of the ﬁnal CoMSIA are summarized in Table 4. The

leave-one-out cross-validated PLS analysis coeﬃcient

(q²) was 0.668, and the standard error of prediction

(s_press) was 0.757. The optimum number of components

obtained from the cross-validation linear regression

coeﬃcient was used to produce the non-cross-validated

model. The non-cross-validated PLS analysis yielded a

high correlation coeﬃcient (r²) of 0.936 with a low stan-

dard error estimate (s) of 0.332.

The CoMSIA hydrophobic ﬁeld map is shown in Figure

4C. The yellow contours in this ﬁgure represent hydro-

phobic favorable regions that enhance the aﬃnity for

receptors, while the white contours show hydrophobic

unfavorable regions. Several favorable and unfavorable

regions appeared in the hydrophobic property contour

map. The presence of yellow contours behind the white

contour around the 4-position of the triazole ring indi-

cates that an optimum hydrophobicity of substituents

for high potency is present at this position. At the same

time, the presence of a yellow contour in front of

the white contour at the 5-position of the triazole ring

indicates that small rather than large hydrophobic

M. S. Alam et al. / Bioorg. Med. Chem. 15 (2007) 5090–5104

5097

negative substituents at the 2-, 4-, and 6-positions of the

phenyl ring. This ﬁnding could explain that there was

low or no activity of phenyltriazoles having low electro-

negative substituents or of phenyltriazoles without

substituents at the 2-, 4-, and 6-positions of the benzene

ring.

Both CoMFA and CoMSIA provided consistent PLS

statistical results and contour maps for b3 receptors,

with only a few exceptions. In CoMSIA, a higher non-

cross-validated correlation coeﬃcient (r²) and a lower

standard error of estimate (s) were obtained than in

CoMFA, although the cross-validated correlation coef-

ﬁcient (q²) and the standard error of prediction (s_press

)

were almost similar. Regarding the relative ﬁeld contri-

butions, the CoMFA results showed that steric ﬁelds

made greater contributions to the aﬃnity of compounds

than electrostatic ﬁelds, whereas the CoMSIA results

showed that electrostatic ﬁelds made a greater contribu-

tion than steric and hydrophobic ﬁelds. It is also noted

that the steric and hydrophobic ﬁeld contributions were

almost the same in CoMSIA. CoMSIA apparently per-

formed better to explain the structure–activity relation-

ships of phenyltriazoles in b3 receptors than CoMFA,

because of additional information on the hydrophobic

ﬁeld in the 3D-QSAR model.

2.5. Binding sites and selectivity of 1-phenyl-1H-1,2,3-

triazoles for b3 over a1b2c2 GABA receptors

We performed homology modeling studies of human b3

and a1b2c2 GABA receptors in an attempt to predict

the binding site of 1-phenyl-1H-1,2,3-triazoles as well

as to ﬁnd out the cause of their selectivity for b3 over

a1b2c2 receptors. The homology models were built

using an atomic model of the closed nicotinic acetylcho-

˚

line receptor channel, resolved to 4 A by electron

microscopy of the crystalline postsynaptic membrane

of Torpedo marmorata²⁷as a template. Both GABA

and nicotinic acetylcholine receptors belong to a Cys-

loop receptor family of ligand-gated ion channels that

share structural features; for example, they consist of

ﬁve subunits, each of which has four a-helical trans-

membrane segments (TM1-TM4). The TM2s from the

ﬁve subunits are assembled to form a channel lumen.

The amino acid sequences of TM2 of GABA receptor

subunits are shown in Figure 5A. The subunit corre-

spondence of the a1b2c2 receptor to the nicotinic acetyl-

choline receptor is depicted in Figure 5B. Figure 6A

shows amino acids from the 2⁰to 6⁰position of the

TM2 channel lumen of the homology model of the hu-

man b3 homopentamer with 15, one of the most active

analogues, automatically docked using the chemistry

and bioinformatics package MOE. The T6⁰methyl

groups of the homopentamer show favorable hydropho-

bic interactions with the n-propyl group of the 4-posi-

tion of the triazole ring, and one of the A2⁰methyl

groups of the homopentamer shows favorable hydro-

phobic interactions with the phenyl group. The T6⁰

hydroxyl group probably forms hydrogen bonds with the

nitrogen atoms of the 2- and 3-positions of the triazole

ring. Introduction of an alkyl chain longer than the n-

propyl group or a substituent bulkier than the tert-butyl

Figure 4. CoMSIA model for the human b3 GABA receptor. (A)

Steric contour plots. The green contours indicate regions where

sterically bulky substituents are favorable, while the yellow contours

show regions where sterically bulky substituents are unfavorable for

receptor aﬃnity. (B) Electrostatic contour plots. The blue contours

indicate regions where the electropositive groups enhance receptor

aﬃnity, while the interaction of electronegative substituents with the

areas indicated by red contours increases aﬃnity. (C) Hydrophobic

contour plots. The yellow contours in this ﬁgure indicate hydrophobic

favorable regions that enhance the aﬃnity for receptors, while the

white contours show hydrophobic unfavorable regions. Compound 15

is shown as a reference molecule.

substituents are favorable for high aﬃnity at this posi-

tion. A hydrophobic substituent of excessive length is

not favorable at the 4-position of the triazole ring, be-

cause the sterically unfavorable region (Fig. 4A) is

located behind the hydrophobic favorable region

(Fig. 4C). The two other yellow contours above and be-

low the benzene ring indicate that the hydrophobic nat-

ure of the benzene ring would increase the receptor

aﬃnity, which could be achieved by introducing electro-

5098

M. S. Alam et al. / Bioorg. Med. Chem. 15 (2007) 5090–5104

our homology model will be needed so as to incorporate

all analogues into the model.

It has been shown that a b subunit is essential for the

binding of non-competitive antagonists to heteromeric

GABA_Areceptors, and that the b3 subunit alone can

form a channel that is suﬃcient for antagonist bind-

ing.^16,28It is believed that the binding site for non-com-

petitive antagonists is located within the channel pore.

Recent site-directed mutagenesis studies of various GA-

BA_Areceptors have reported on the molecular localiza-

tion of the antagonist binding site, and the amino acid

residues responsible for the binding have been shown

to exist in the cytoplasmic half of TM2 of the chan-

nels.^29–35On the basis of site-directed mutagenesis and

molecular modeling results of the b3 homopentamer,

Chen et al. reported with reference to information from

other studies that widely diverse non-competitive antag-

onists bind to the same site of the receptor, and pointed

out that A2⁰and T6⁰play important roles for antagonist

binding, with L9⁰playing a supplemental role.³⁵In an-

other model of the b3 receptor provided by Ci et al.,

L3⁰, A2⁰, and T6⁰were postulated to play more impor-

tant roles than L9⁰for interaction with ﬁpronil-related

non-competitive antagonists.³⁶The A2⁰S mutation of

b3 receptors caused a reduced aﬃnity for EBOB bind-

ing, whereas the T6⁰V mutation abolished the binding,

which indicates that the 6⁰amino acid has more intense

eﬀects than the 2⁰amino acid on EBOB binding to these

receptors.³⁷Even if the ﬁndings regarding the putative

binding site in our model are mostly consistent with

these other research results, it should be noted that dif-

ferent binding orientations from those of ﬁpronil and its

related antagonists were observed for 1-phenyl-1H-

1,2,3-triazole antagonists.^35,36Fipronil and its closely

related analogues showed levels of potency almost iden-

tical to those of the GABA receptors obtained from Rdl

(A2⁰S) mutant and wild-type houseﬂies, whereas the 4-

tert-butyl analogue of phenyltriazole (21) showed lower

aﬃnity for the GABA receptors of Rdl mutant house-

ﬂies than for those of wild-type houseﬂies.³⁸The alkyl

group at the 4-position of the triazole ring of phenyl-

triazoles and the triﬂuoromethylsulﬁnyl group of ﬁpro-

nil are diﬀerent in terms of electron density,¹⁹which

Figure 5. Amino acid sequences of the putative binding site and

subunit correspondence. (A) Amino acid residues of TM2 of GABA

receptor subunits. Pore-facing amino acids are underlined. Accession

Nos. human a1 subunit = NM_174540; human b2 subunit =

NM_000813; human c2 subunit = NM_000816; human b3 sub-

unit = NM_000814; houseﬂy RDL subunit = AB177547. The positions

of amino acid residues are index-numbered from the N-terminal side of

TM2 according to Miller.⁴⁶(B) Schematic presentation of subunit

correspondence between the nicotinic acetylcholine receptor and the

a1b2c2 GABA receptor. Agonist binding sites are indicated by arrows.

group at the 4-position of the triazole ring would result

in a collision with the channel wall, which would make

the compounds less potent than the parent compound,

as predicted in 3D-QSAR studies. Most of the com-

pounds performed similar interactions with the putative

binding site within the channel pore, whereas some of

the compounds (6, 13, and 17) were docked into a diﬀer-

ent binding site or in a diﬀerent orientation. For in-

stance, the 2,6-dibromo-4-triﬂuoromethylphenyl group

of 6 was involved in hydrophobic interactions with the

T6⁰and L9⁰methyl groups, and the 5-methyl-4-n-pro-

pyl-1H-1,2,3-triazole moiety was outside the putative

binding site. The 4-n-propyl group of 17 was involved

in hydrophobic interactions with the T6⁰and L9⁰methyl

groups, and the 2,6-dichloro-4-triﬂuoromethylphenyl

group was outside the putative binding site. The reason

why these compounds bound to the diﬀerent sites is

uncertain at the present moment. The reﬁnement of

Figure 6. Docking of 1-phenyl-1H-1,2,3-triazoles into GABA receptors. Compound 15 was automatically docked into (A) homopentameric b3

receptors and (B) heteropentameric a1b2c2 receptors.

M. S. Alam et al. / Bioorg. Med. Chem. 15 (2007) 5090–5104

5099

might be responsible for the binding of ﬁpronil to b3

receptors in a diﬀerent orientation. A deﬁnitive conclu-

sion regarding the binding orientation of antagonists

must await a high-resolution characterization of the

structure of ligand-gated ion channels.

The channel lumen from the 2⁰to 6⁰position of the

homology model of the human a1b2c2 heteropentamer

is shown in Figure 6B, and the most active compound

(15) was docked into the putative binding site. Favor-

able hydrophobic contacts were observed for the T6⁰

methyl groups with the alkyl substituent at the 4-posi-

tion of the triazole ring and for the V2⁰and A2⁰methyl

groups with the 2,6-dichloro-4-trifuoromethylphenyl

group. All of the compounds except 25 displayed the

same binding orientation within the channel pore. The

long n-hexyl group of 25 was involved in hydrophobic

interactions with the V2⁰and T6⁰methyl groups and

the triazole ring with the L9⁰methyl groups, and the

2,6-dichloro-4-trifuoromethylphenyl group resided out-

side the putative binding site. Perret et al. reported that

a major interaction of chemically reactive phenylpyraz-

ole antagonists occurred at the V257 position of rat a1

subunit,³¹which is equivalent to the V2⁰position of hu-

man a1 subunit (Fig. 5A). The same amino acid had

been suggested to be involved in binding the antagonist

picrotoxinin by the substituted-cysteine accessibility

method.²⁹Recombinant rat a1b2c2 receptors contain-

ing the T6⁰F mutation in the a1, b2, or c2 subunit were

shown to be insensitive to picrotoxinin.³⁰The results of

our docking studies are in agreement with those of these

site-directed mutagenesis studies.

Figure 7. Correlation between the total binding energy (DG_binding

)

determined by docking studies and the experimentally determined

aﬃnity (pK_i) of 1-phenyl-1H-1,2,3-triazoles for human b3 GABA

receptors. Compound numbers are indicated.

would make the molecules less stable in the channel pore

of a1b2c2 receptors than in that of b3 receptors. This

might be one of the reasons that 1-phenyl-1H-1,2,3-triaz-

oles select b3 over a1b2c2 GABA receptors.

Finally, to validate the suggested mode of interaction

between 1-phenyl-1H-1,2,3-triazoles and the homology

model of human b3 receptors, we examined the relation-

ship between the total binding energies (DG_binding) calcu-

lated by the MOE program and the experimentally

determined aﬃnity (pK_i) values. We ﬁrst randomly

chose 17 analogues for this validation so as to include

compounds with a broad range of potency (K_i=

0.659–392 nM). As a result, a good correlation (r²=

0.80) was obtained between DG_bindingand the pK_ivalues

of 14 selected 1-phenyl-1H-1,2,3-triazoles with a wide

range of potency (Fig. 7), although three compounds

(6, 13, and 17) that failed to bind to the 2⁰and 6⁰posi-

tions were omitted from the plot for the reason that

most of the site-directed mutagenesis studies using

GABA and glycine receptors identiﬁed the 2⁰and 6⁰

amino acids as amino acids responsible for antagonist

binding. Further, we tested three more compounds

(10, 29, and 31) to see whether the addition of these

compounds deteriorates the correlation, and this addi-

tion did not signiﬁcantly aﬀect the correlation. This ﬁnd-

ing indicates the validity of our model. As for a1b2c2

receptors, a plot of DG_bindingvalues against pK_ivalues

yielded a correlation (r²= 0.77) for 8 compounds (15,

17, 21, 23, 25, 27, 28, and 31), although only a limited

number and a narrow range of pK_ivalues for 1-phe-

nyl-1H-1,2,3-triazoles were available (plot not shown).

How can we explain the selectivity of 1-phenyl-1H-1,2,3-

triazoles for b3 over a1b2c2 receptors? The b3 GABA

receptor is a homopentamer in which the channel-lining

pores at the 2⁰and 6⁰positions are formed by only ala-

nine and threonine residues, respectively, thereby mak-

ing the channel lumen symmetrical, so any parallel

alignment of phenyltriazoles to the channel pore axis

at this putative binding site might be favorable for

hydrophobic interactions and for the formation of

hydrogen bonds between the hydroxyl group of T6⁰

and the nitrogen atoms of the triazole ring of 15. In con-

trast, the channel-lining pore at the 2⁰position of the

heteropentameric a1b2c2 GABA receptor is formed by

two valines, two alanines, and one serine residue, which

makes the channel pore unsymmetrical. Therefore, the

speciﬁc alignment of phenyltriazoles in the channel pore

might be necessary for favorable hydrophobic interac-

tion with the V2⁰and A2⁰methyl groups. The docking

energy of 15 in b3 receptors was found to be lower

(ꢃ5.66 kcal/mol) than that in a1b2c2 receptors

(ꢃ4.26 kcal/mol), and similar relations were also

observed for other analogues, indicating that 1-phenyl-

1H-1,2,3-triazoles are more stable in b3 receptor

channels than in a1b2c2 GABA receptor channels. The

higher docking energy in a1b2c2 receptors might be

due to the steric repulsion of the bulky isopropyl side

chain of V2⁰with the phenyl groups of 1-phenyl-1H-

1,2,3-triazoles, which also keeps the triazole ring far from

the hydroxyl groups of T6⁰and thereby prevents the for-

mation of hydrogen bonds in the case of 15. This factor

2.6. Conclusions

In the current study, we synthesized a series of 4,5-

disubstituted 1-phenyl-1H-1,2,3-triazoles and examined

their antagonist potencies (aﬃnities) in human b3 and

a1b2c2 GABA receptors to study the structure–activity

relationships. Compound 15 exhibited the highest level

of aﬃnity for both b3 and a1b2c2 receptors, which

was the same level as that of the most potent monosub-

stituted analogue (21). Most of the analogues were more

potent in b3 receptors than in a1b2c2 receptors. The 4-

n-propyl-5-ethyl analogue (10) exhibited the highest

selectivity. The n-propyl group at the 4-position

and the chloromethyl group at the 5-position of

5100

M. S. Alam et al. / Bioorg. Med. Chem. 15 (2007) 5090–5104

1-(2,6-dichloro-4-triﬂuoromethylphenyl)-1H-1,2,3-tria-

zole are optimum substituents with respect to conferring

high aﬃnity for b3 receptors (K_i= 659 pM), but in terms

of selectivity for b3 versus a1b2c2 receptors, the ethyl

group (RS >1133-fold) is preferable to the chloromethyl

group as the 5-substituent. 3D-QSAR studies revealed

that a small hydrophobic 4-substituent and a hydropho-

bic or hydrophilic 5-substituent smaller than the 4-sub-

stituent are favorable for high potency at b3 receptors,

together with electronegative substituents at the 2-, 4-,

and 6-positions on the benzene ring. Docking studies

using a b3 GABA receptor homology model showed

that the amino acid residues at the 2⁰- and 6⁰-positions

play an important role in binding 1-phenyl-1H-1,2,3-

triazoles. The ﬂexibility of the alignment of molecules

in the channel lumen and a speciﬁc hydrophobic envi-

ronment at the 2⁰-position of b3 homopentamers that

are brought about by the composition of the same ﬁve

subunits could be one of the reasons that 1-phenyl-1H-

1,2,3-triazoles select b3 over a1b2c2 receptors.

same as that described in our previous paper.¹⁹Brieﬂy,

2,6-dichloro-4-triﬂuoromethylphenyl azide was prepared

from 2,6-dichloro-4-triﬂuoromethylaniline (670 mg,

2.91 mmol) by the treatment with concd HCl (50 mL), so-

dium nitrite (200 mg, 2.91 mmol), and sodium azide

(190 mg, 2.91 mmol) in water (50 mL), and the mixture

was stirred overnight at room temperature. After comple-

tion of the reaction, the mixture was extracted with

CH₂Cl₂, and the solvent was evaporated. The residue

was chromatographed on silica gel (n-hexane), aﬀording

2,6-dichloro-4-triﬂuoromethylphenyl azide as a red oil.

2,6-Dichloro-4-triﬂuoromethylphenyl azide (600 mg,

2.38 mmol), an unsymmetrical alkyne (2.38 mmol), and

toluene (3 mL) were placed in a sealed tube and heated

at 110 ꢁC for 16 h. After the reaction, the mixture was par-

titioned between water and ether. The ether layer was

dried with sodium sulfate overnight. The ether was evap-

orated, and the residue was subjected to silica gel column

chromatography (n-hexane/EtOAc = 30:1), thus aﬀord-

ing the 4,5-dialkyl isomers of the desired compound.

Compounds 1–14 were synthesized by this method.

The K_ivalues of compounds such as 4, 13, 15, 17, 21,

and 28 in the range of 0.66 to 2.53 nM in b3 receptors

are comparable to or smaller than the reported IC₅₀

values of the GABA receptor antagonist/commercial

insecticide ﬁpronil^14,16,39,40and those of its 4-tert-butyl

and 4-isopropyl analogues in b3 or native houseﬂy recep-

tors.¹⁷The interaction of 1-phenyl-1H-1,2,3-triazoles

with houseﬂy GABA receptors causes signiﬁcant insecti-

cidal activity,¹⁹and the most two potent analogues (15

and 21) demonstrated considerable insecticidal activity

(LD₅₀s = 5.17 and 1.86 pmol/ﬂy,¹⁹respectively) against

houseﬂies. The information obtained from the 3D con-

tour maps of CoMFA and CoMSIA of b3 receptors to-

gether with the homology models of b3 and a1b2c2

receptors should prove helpful for understanding the

mechanisms of the interaction of non-competitive antag-

onists with GABA receptors, and might aﬀord useful

information for designing new insect control chemicals.

3.1.3. 1-(2,6-Dichloro-4-triﬂuoromethylphenyl)-4-ethyl-5-

methyl-1H-1,2,3-triazole (1). Yield 23%, mp 71.2–

72.1 ꢁC. H NMR (CDCl₃) d 1.37 (3H, t, J = 7.6 Hz,

CH₂CH₃), 2.77 (2H, q, J = 7.6 Hz, CH₂CH₃), 2.11

(3H, s, CH₃), 7.78 (2H, s, Ph). EIMS m/z (%) 41 (100),

240 (34), 280 (82), 283 (55), 323 (M, 8), 325 (M+2, 5).

Anal. Calcd for C₁₂H₁₀Cl₂F₃N₃: C, 44.47; H, 3.11; N,

12.96%. Found: C, 44.73; H, 3.20; N, 12.66%.

1

3.1.4. 1-(2,6-Dichloro-4-triﬂuoromethylphenyl)-5-ethyl-4-

methyl-1H-1,2,3-triazole (2). Yield 17%, mp 64.6–

1

65.7 ꢁC. H NMR (CDCl₃) d 1.08 (3H, t, J = 7.6 Hz,

CH₂CH₃), 2.51 (2H, q, J = 7.6 Hz, CH₂CH₃), 2.42

(3H, s, CH₃), 7.78 (2H, s, Ph). EIMS m/z (%) 41 (100),

280 (80), 282 (53), 323 (M, 3.2), 325 (M+2, 2.2). Anal.

Calcd for C₁₂H₁₀Cl₂F₃N₃: C, 44.47; H, 3.11; N,

12.96%. Found: C, 44.83; H, 3.29; N, 12.50%.

3.1.5. 1-(2,6-Dichloro-4-triﬂuoromethylphenyl)-4,5-diethyl-

1

3. Experimental

3.1. Synthesis

1H-1,2,3-triazole (3). Yield 37%, mp 68.9–69.6 ꢁC. H

NMR (CDCl₃) d 1.07 (3H, t, J = 7.7 Hz, CH₂CH₃-5),

1.38 (3H, t, J = 7.6 Hz, CH₂CH₃-4), 2.51(2H, q,

J = 7.7 Hz, CH₂CH₃-5), 2.78 (2H, q, J = 7.6 Hz,

CH₂CH₃-4), 7.78 (2H, s, Ph). EIMS m/z (%) 294 (100),

309 (11), 337 (M, 3), 339 (M+2, 2). Anal. Calcd for

C₁₃H₁₂Cl₂F₃N₃: C, 46.17; H, 3.58; N, 12.43%. Found:

C, 46.17; H, 3.62; N, 12.11%.

3.1.1. General. ¹H NMR spectra were obtained on a

JEOL JNM-A 400 spectrometer, and chemical shifts (d

values) are given in ppm using tetramethylsilane

(TMS) as an internal standard. Mass spectra were taken

on a Hitachi M80-B spectrometer using the electron im-

pact method (70 eV), and the fragment data are reported

as m/z. Melting points were determined using a Yana-

gimoto MP 500D apparatus and are uncorrected.

[Propyl-2,3-³H]EBOB (47.5 or 45.0 Ci/mmol) was

purchased from Perkin-Elmer Life Sciences, Inc. (Bos-

ton, MA). Phenyltriazoles 17–49 were available from

our earlier studies.^18,19General chemicals were pur-

chased from Wako Pure Chemical Industries, Ltd (Osa-

ka, Japan) unless otherwise noted.

3.1.6. 1-(2,6-Dichloro-4-triﬂuoromethylphenyl)-5-methyl-

4-n-propyl-1H-1,2,3-triazole (4). Yield 40%, oily liquid.

¹H NMR (CDCl₃) d 0.98 (3H, t, J = 7.3 Hz, (CH₂)₂CH₃),

1.79 (2H, sex, J = 7.3 Hz, CH₂CH₂CH₃), 2.10 (3H, s,

CH₃), 2.72 (2H, t, J = 7.3 Hz, CH₂CH₂CH₃), 7.78 (2H,

s, Ph). EIMS m/z (%) 280 (100), 282 (65), 337 (M, 6),

339 (M+2, 4). Anal. Calcd for C₁₃H₁₂Cl₂F₃N₃: C,

46.17; H, 3.58; N, 12.43%. Found: C, 46.32; H, 3.56;

N, 12.15%.

3.1.2. General procedure for the synthesis of 4,5-disubsti-

tuted 1-phenyl-1H-1,2,3-triazoles (Scheme 1). The proce-

dure for synthesis of disubstituted phenyltriazoles was the

3.1.7. 1-(2,6-Dichloro-4-triﬂuoromethylphenyl)-4-methyl-

5-n-propyl-1H-1,2,3-triazole (5). Yield 30%, mp 113.5–

114.4 ꢁC. H NMR (CDCl₃) d 0.88 (3H, t, J = 7.5 Hz,

1

M. S. Alam et al. / Bioorg. Med. Chem. 15 (2007) 5090–5104

5101

(CH₂)₂CH₃), 1.46 (2H, sex, J = 7.5 Hz, CH₂CH₂CH₃),

2.41 (3H, s, CH₃), 2.46 (2H, t, J = 7.5 Hz, CH₂CH₂CH₃),

7.78 (2H, s, Ph). EIMS m/z (%) 27 (100), 281 (47), 283

(32), 337 (M, 10), 339 (M+2, 6.6). Anal. Calcd for

C₁₃H₁₂Cl₂F₃N₃: C, 46.17; H, 3.58; N, 12.43%. Found:

C, 46.24; H, 3.56; N, 12.51%.

Calcd for C₁₄H₁₄Cl₂F₃N₃: C, 47.75; H, 4.01; N, 11.93%.

Found: C, 47.88; H, 4.06; N, 11.75%.

3.1.14. 1-(2,6-Dichloro-4-triﬂuoromethylphenyl)-4,5-di-n-

propyl-1H-1,2,3-triazole (12). Yield 33%, mp 69.8–

1

70.9 ꢁC. H NMR (CDCl₃) d 0.87 (3H, t, J = 7.5 Hz,

(CH₂)₂CH₃-5), 1.00 (3H, t, J = 7.5 Hz, (CH₂)₂CH₃-4),

1.42 (2H, sex, J = 7.5 Hz, CH₂CH₂CH₃-5), 1.82 (2H,

sex, J = 7.5 Hz, CH₂CH₂CH₃-4), 2.47 (2H, t,

J = 7.5 Hz, CH₂CH₂CH₃-5), 2.72 (2H, t, J = 7.5 Hz,

CH₂CH₂CH₃-4), 7.78 (2H, s, Ph). EIMS m/z (%) 308

(100), 310 (66), 365 (M, 3), 367 (M+2, 2). Anal. Calcd

for C₁₅H₁₆Cl₂F₃N₃: C, 49.20; H, 4.40; N, 11.47%.

Found: C, 48.97; H, 4.42; N, 11.17%.

3.1.8. 1-(2,6-Dibromo-4-triﬂuoromethylphenyl)-5-methyl-

4-n-propyl-1H-1,2,3-triazole (6). Yield 14%, oily liquid.

¹H NMR (CDCl₃) d 0.97 (3H, t, J = 7.4 Hz, (CH₂)₂CH₃),

1.80 (2H, sex, J = 7.4 Hz, CH₂CH₂CH₃), 2.10 (3H, s,

CH₃), 2.72 (2H, t, J = 7.4 Hz, CH₂CH₂CH₃), 7.98

(2H, s, Ph). EIMS m/z (%) 43 (100), 368 (25), 370

(50), 372 (25), 425 (M, 2.4), 427 (M+2, 4.8), 429 (M+4,

2.4).

3.1.15. 1-(2,6-Dichloro-4-triﬂuoromethylphenyl)-5-hydroxy-

methyl-4-n-propyl-1H-1,2,3-triazole (13). Yield 26%,

3.1.9. 1-(2,6-Dibromo-4-triﬂuoromethylphenyl)-4-methyl-

5-n-propyl-1H-1,2,3-triazole (7). Yield 9%, 104.4–

105.9 ꢁC. H NMR (CDCl₃) d 0.83 (3H, t, J = 7.5 Hz,

(CH₂)₂CH₃), 1.42 (2H, sex, J = 7.5 Hz, CH₂CH₂CH₃),

2.34 (3H, s, CH₃), 2.40 (2H, t, J = 7.5 Hz, CH₂CH₂CH₃),

7.91 (2H, s, Ph). EIMS m/z (%) 32 (100), 368 (48), 370

(96), 372 (48), 425 (M, 1.8), 427 (M+2, 3.6), 429 (M+4,

1.8).

87.3–87.7 ꢁC. ¹H NMR (CDCl₃)

d 0.97 (3H, t,

1

J = 7.4 Hz, (CH₂)₂CH₃), 1.79 (2H, sex, J = 7.4 Hz,

CH₂CH₂CH₃), 2.16 (1H, s, CH₂OH), 2.76 (2H, t,

J = 7.4 Hz, CH₂CH₂CH₃), 4.60 (2H, s, CH₂OH), 7.77

(2H, s, Ph). EIMS m/z (%) EIMS 296 (100), 298 (68),

353 (M,1.1), 355 (M+2, 0.7). Anal. Calcd for

C₁₃H₁₂Cl₂F₃N₃O: C, 44.09; H, 3.42; N, 11.86%. Found:

C, 44.53; H, 3.53; N, 11.44%.

3.1.10. 5-Methyl-4-n-propyl-1-(4-triﬂuoromethylphenyl)-

1H-1,2,3-triazole (8). Yield 45%, mp 50.4–51.7 ꢁC. H

1

3.1.16. 1-(2,6-Dichloro-4-triﬂuoromethylphenyl)-4-hydroxy-

methyl-5-n-propyl-1H-1,2,3-triazole (14). Yield 28%,

120.5–121.0 ꢁC. ¹H NMR (CDCl₃) d 0.89 (3H, t,

J = 7.5 Hz, (CH₂)₂CH₃), 1.50 (2H, sex, J = 7.5 Hz,

CH₂CH₂CH₃), 2.17 (1H, s, CH₂OH), 2.56 (2 H, t,

J = 7.5 Hz, CH₂CH₂CH₃), 4.86 (2H, s, CH₂OH), 7.81

(2H, s, Ph). EIMS m/z (%) 43 (100), 296 (12), 298 (8),

353 (M, 0.5), 355 (M+2, 0.3). Anal. Calcd for

C₁₃H₁₂Cl₂F₃N₃O: C, 44.09; H, 3.42; N, 11.86%. Found:

C, 44.17; H, 3.37; N, 11.82%.

NMR (CDCl₃) d 1.01 (3H, t, J = 7.5 Hz, (CH₂)₂CH₃),

1.77 (2H, sex, J = 7.5 Hz, CH₂CH₂CH₃), 2.32 (3H, s,

CH₃), 2.69 (2H, t, J = 7.5 Hz, CH₂CH₂CH₃), 7.64 (2H,

d, J = 8.4 Hz, Ph), 7.82 (2H, d, J = 8.4 Hz, Ph). EIMS

m/z (%) 212 (100), 226 (44), 241 (18), 269 (M, 5). Anal.

Calcd for C₁₃H₁₄F₃N₃: C, 57.99; H, 5.24; N, 15.61%.

Found: C, 57.91; H, 5.24; N, 15.28%.

3.1.11. 4-Methyl-5-n-propyl-1-(4-triﬂuoromethylphenyl)-

1

1H-1,2,3-triazole (9). Yield 36%, oily liquid. H NMR

3.1.17. 5-Chloromethyl-1-(2,6-dichloro-4-triﬂuoromethyl-

phenyl)-4-n-propyl-1H-1,2,3-triazole (15) (Scheme 1). A

mixture of 13 (128 mg, 0.36 mmol), thionyl chloride

(1.6 g, 1 mL), and DMF (0.5 mL) was stirred at 85 ꢁC

for 30 min. The reaction mixture was concentrated and

subjected to silica gel column chromatography (n-hex-

ane/EtOAc = 20:1) to give 15 (90 mg, 0.24 mmol) as a

yellowish solid. Yield 70%, 53.9–54.5 ꢁC. ¹H NMR

(CDCl₃) d 1.01 (3H, t, J = 7.4 Hz, (CH₂)₂CH₃), 1.85

(2H, sex, J = 7.4 Hz, CH₂CH₂CH₃), 2.81 (2H, t,

J = 7.4 Hz, CH₂CH₂CH₃), 4.45 (2H, s, CH₂Cl), 7.81

(2H, s, Ph). EIMS m/z (%) 240 (54), 242 (34), 294

(100), 306 (90), 308 (56), 371 (M, 2.1), 373 (M+2, 1.9),

(M+4, 0.7). Anal. Calcd for C₁₃H₁₁Cl₃F₃N₃: C, 41.91;

H, 2.98; N, 11.28%. Found: C, 41.66; H, 3.16; N,

10.86%.

(CDCl₃) d 0.79 (3H, t, J = 7.5 Hz, (CH₂)₂CH₃), 1.40

(2H, sex, J = 7.5 Hz, CH₂CH₂CH₃), 2.31 (3H, s, CH₃),

2.59 (2H, t, J = 7.5 Hz, CH₂CH₂CH₃), 7.52 (2H, d,

J = 8.4 Hz, Ph), 7.74 (2H, d, J = 8.4 Hz, Ph). EIMS m/

z (%) 212 (100), 226 (15), 241 (12), 269 (M, 6). Anal.

Calcd for C₁₃H₁₄F₃N₃: C, 57.99; H, 5.24; N, 15.61%.

Found: C, 57.92; H, 5.26; N, 15.36%.

3.1.12. 1-(2,6-Dichloro-4-triﬂuoromethylphenyl)-5-ethyl-

4-n-propyl-1H-1,2,3-triazole (10). Yield 49%, oily liquid.

¹H NMR (CDCl₃) d 1.00 (3H, t, J = 7.5 Hz, CH₂CH₃),

1.06 (3H, t, J = 7.5 Hz, (CH₂)₂CH₃), 1.82 (2H, sex,

J = 7.5 Hz, CH₂CH₂CH₃), 2.53 (2H, q, J = 7.5 Hz,

CH₂CH₃), 2.73 (2H, t, J = 7.5 Hz, CH₂CH₂CH₃), 7.78

(2H, s, Ph). EIMS m/z (%) 240 (36), 242 (25), 294 (100),

296 (65), 351 (M, 3), 353 (M+2, 2). Anal. Calcd for

C₁₄H₁₄Cl₂F₃N₃: C, 47.75; H, 4.01; N, 11.93%. Found:

C, 48.11; H, 4.08; N, 11.45%.

3.1.18. 4-Chloromethyl-1-(2,6-dichloro-4-triﬂuoromethyl-

phenyl)-5-n-propyl-1H-1,2,3-triazole (16). Compound 16

was synthesized by a method similar to that used for

3.1.13. 1-(2,6-Dichloro-4-triﬂuoromethylphenyl)-4-ethyl-

5-n-propyl-1H-1,2,3-triazole (11). Yield 33%, oily liquid.

¹H NMR (CDCl₃) d 0.80 (3H, t, J = 7.6 Hz, (CH₂)₂CH₃),

1.29–1.41 (5H, m, CH₂CH₃and CH₂CH₂CH₃), 2.40 (2H,

t, J = 7.6 Hz, CH₂CH₂CH₃), 2.71 (2H, q, J = 7.6 Hz,

CH₂CH₃), 7.71 (2H, s, Ph). EIMS m/z (%) 240 (39), 242

(27), 294 (100), 296 (66), 351 (M, 3), 353 (M+2, 2). Anal.

1

15. Yield 86%, 84.5–85.1 ꢁC. H NMR (CDCl₃) d 0.92

(3H, t, J = 7.7 Hz, (CH₂)₂CH₃), 1.53 (2H, sex,

J = 7.7 Hz, CH₂CH₂CH₃), 2.58 (2H, t, J = 7.7 Hz,

CH₂CH₂CH₃), 4.80 (2H, s, CH₂Cl), 7.81 (2H, s, Ph).

EIMS m/z (%) 240 (65), 242 (43), 306 (100), 308 (67),

371 (M, 7.2), 373 (M+2, 7), 375 (M+4, 2.4). Anal. Calcd

5102

M. S. Alam et al. / Bioorg. Med. Chem. 15 (2007) 5090–5104

for C₁₃H₁₁Cl₃F₃N₃: C, 41.91; H, 2.98; N, 11.28%. Found:

C, 41.84; H, 3.04; N, 11.08%.

containing 20% FBS was added to the dish. After a

24-h incubation, selection was started by replacing

the old medium with complete DMEM containing

G418 (1 mg/mL). After a ꢀ2 week selection, resistant

colonies were trypsinized in cloning cylinders and

transferred to 24-well plastic plates. Individual mono-

clonal cell lines were expanded and cultured in the same

medium.

3.2. Generation of Drosophila Schneider 2 (S2) cells

stably expressing human b3 GABA receptors

S2 cells expressing human b3 GABA receptors were gen-

erated as previously described.¹⁹Brieﬂy, the cDNA

encoding the b3 subunit of the human GABA receptor

(GenBank Accession No. NM_000814) was inserted

into the Drosophila expression vector pMT (Invitrogen,

Carlsbad, CA) to produce pMT-hsb3, in which the

expression of b3 cDNA is driven by the metallothionein

promoter and induced by copper sulfate. S2 cells were

co-transfected with pMT-hsb3 and the selection vector

pCoHygro by a calcium phosphate transfection method,

using a DES kit (Invitrogen) according to the manufac-

turer’s instructions. The transfected cells were main-

tained for ꢀ3 weeks in selection medium containing

the antibiotic hygromycin B. Hygromycin B-resistant

polyclonal cells were used for the binding assays as de-

scribed in Section 3.5.

The obtained cell lines were analyzed for the presence of

subunit mRNAs by reverse transcription (RT) of total

RNA followed by polymerase chain reaction (PCR). To-

tal RNA was isolated from transfected HEK-293 cell

lines using an Isogen kit (Nippon Gene, Toyama, Ja-

pan). RT was performed using an AMV reverse trans-

criptase kit (Life Sciences, St. Petersburg, FL) to

synthesize cDNAs of a1, b2, and c2 subunits. The RT

products were then ampliﬁed by PCR using sets of

primers annealing to the 5⁰and 3⁰ends of the genes.

PCR products were checked by electrophoresis in 2%

agarose gel. Cell lines that were positive for the presence

of a1, b2, and c2 transcripts were examined for their

ability to speciﬁcally bind [³H]EBOB. Incorporation of

c2 subunit into the ion channel was conﬁrmed by

[³H]diazepam binding assays as described by Hawkin-

son et al.⁴¹

3.3. Construction of plasmid vectors containing human a1,

b2, and c2 GABA receptor subunits

Plasmid

vectors,

pCDM8-bovine(Bt)GABA_Aa1

(NM_174540), which produces the amino acid sequence

of the human a1 subunit, pCDM8-human(Hs)GABA_Ab2

(NM_000813), and pCDM8-human(Hs)GABA_Ac2

(NM_000816), were obtained from the Neuroscience

Research Center of Merck Sharp & Dohme Ltd (Harlow,

Essex, UK). We changed the expression vector from

pCDM8 to pcDNA3 (Invitrogen) for technical conve-

nience. The 1.9 kb HindIII–XbaI BtGABA_Aa1, the

1.5 kb XbaI–XbaI HsGABA_Ab2, and the 1.5 kb

HindIII–HindIII HsGABA_Ac2 fragments were inserted

into the corresponding sites of pcDNA3 to produce

pcDNA3a1, pcDNA3b2, and pcDNA3c2, respectively.

3.5. Membrane preparation and [³H]EBOB binding

assays

The membranes of S2 cells expressing b3 receptors were

prepared as previously described.¹⁹Brieﬂy, the S2 cells

were grown in Schneider’s Drosophila medium (Gibco)

containing 10% FBS and hygromycin B (300 lg/mL)

in a shaking incubator (50 rpm) at 28 ꢁC. S2 cells were

grown at a viability of more than 96% and a density

of 6–20 · 10⁶cells/mL of culture medium for 3–4 days.

Copper sulfate was added to the medium at a concentra-

tion of 500 lM/mL 24 h before the preparation of mem-

branes. The S2 cells were homogenized in 10 mM Tris–

HCl buﬀer containing 250 mM sucrose (pH 7.5) with a

glass-Teﬂon homogenizer. The homogenate was centri-

fuged at 25,000g for 20 min. The resulting pellets were

suspended in 10 mM sodium phosphate buﬀer contain-

ing 300 mM NaCl (pH 7.5) and used immediately for

the binding assays.

3.4. Stable expression of heteropentameric a1b2c2 GABA

receptors in HEK-293 cells

Human embryonic kidney-293 (HEK-293) cells were

maintained in Dulbecco’s modiﬁed Eagle’s medium

(DMEM; Gibco, Gaithersburg, MD) supplemented

with 10% fetal bovine serum (FBS; Gibco) (complete

DMEM) in a humidiﬁed incubator (5% CO₂and

37 ꢁC). For stable expression of proteins, the cells were

transfected with vectors by the lipofection method and

selected with the antibiotic G418 (Geneticin; Sigma–Al-

drich, St. Louis, MO) according to the manufacturer’s

instructions. Brieﬂy, aliquots of pcDNA3a1 (0.5 lg),

pcDNA3b2 (0.5 lg), and pcDNA3c2 (5 lg) were added

to 100 lL of OPTI-MEM I reduced serum medium

(Gibco) and then mixed with 100 lL of OPTI-MEM I

medium containing 4 lL of Lipofectamine (Invitrogen).

The mixture was incubated at room temperature for

30 min, and then 0.8 mL of OPTI-MEM I medium

was added. This solution was then overlaid onto

2 · 10⁵HEK 293 cells, which had been deprived of ser-

um by washing with OPTI-MEM I medium, and incu-

bated for 5 h. After the incubation, 1 mL of DMEM

The membranes of HEK-293 cells expressing a1b2c2

receptors were prepared according to Yagle et al.⁴²

The HEK-293 cells were cultured in complete DMEM

and G418 (1 mg/mL) in 60-mm dishes to 100% conﬂu-

ent growth, and used for membrane preparation. The

cells were harvested by scraping with 1 mL of a con-

ditioned medium and centrifuged at 2000g for

10 min. The pelleted cells were suspended in 50 mM

sodium phosphate buﬀer containing 200 mM NaCl

(pH 7.4) and centrifuged at 2000g for 10 min. The

washed cells were resuspended in the buﬀer and dis-

rupted with a Polytron homogenizer (Kinematica, Lu-

cerne, Switzerland). The homogenate was centrifuged

at 25,000g for 30 min, and the resulting pellet was sus-

pended in the buﬀer and used immediately for the

binding assays.

M. S. Alam et al. / Bioorg. Med. Chem. 15 (2007) 5090–5104

5103

[³H]EBOB binding assays were carried out as previously

described.¹⁹Brieﬂy, S2 cell membranes (50 lg protein)

expressing b3 receptors were incubated with various con-

centrations of phenyltriazoles and 0.5 nM [³H]EBOB in

1 mL of 10 mM sodium phosphate buﬀer containing

300 mM NaCl (pH 7.5) at 22 ꢁC for 70 min. HEK-293

cell membranes (150 lg protein) expressing a1b2c2

receptors were incubated with varying concentrations

of phenyltriazoles and 1 nM [³H]EBOB in 1 mL of

50 mM sodium phosphate buﬀer containing 200 mM

NaCl (pH 7.4) at 25 ꢁC for 90 min. The protein content

was determined by the method of Bradford.⁴³After the

incubation, the mixture was ﬁltered through Whatman

GF/B ﬁlters and rapidly rinsed twice with 5 mL of the

incubation buﬀer (10 ꢁC) using a cell harvester (M-24;

Brandel, Gaithersburg, MD). The radioactivity of mem-

brane-bound [³H]EBOB was measured with a liquid scin-

tillation counter. Non-speciﬁc binding was determined in

the presence of 5 lM unlabeled EBOB. Experiments

were repeated at least three times. The IC₅₀values were

estimated from the mean values using the standard pro-

bit method. The K_ivalues were calculated using the

Cheng–Prusoﬀ equation K_i= IC₅₀/(1 + [radioligand]/

K_d).²¹The K_d(apparent dissociation constant) values

of EBOB for b3 (5.0 nM) and a1b2c2 (10.7 nM) recep-

tors were separately determined by Scatchard analysis.

predictive value of the models was ﬁrst evaluated by

leave-one-out cross-validation. The cross-validated

correlation coeﬃcient (q²) was calculated using the

P

equation q²= 1 ꢃ [ (Y_predꢃ Y_obs)²/ (Y_obsꢃ Y_mean)²],

where Y_pred, Y_obs, and Y_meanare the predicted, observed,

and mean aﬃnity (pK_i) values of molecules, respec-

P

2

tively.

(Y_predꢃ Y_obs

)

is the predictive residual

sum of squares (PRESS). The optimum number of

components corresponds to the highest q²and the

lowest PRESS values in the cross-validated PLS anal-

ysis. The ﬁnal 3D-QSAR models were obtained from

non-cross-validated calculations using the optimum

number of components. The CoMFA and CoMSIA

results were graphically represented by ﬁeld contribu-

tion maps using the ﬁeld type ‘stdev coeﬀ’.

*

3.7. Homology modeling and docking

The transmembrane segments of human b3 and a1b2c2

GABA receptors were modeled using the cryo-electron

microscopy structure of the nicotinic acetylcholine

˚

receptor of T. marmorata obtained at 4 A resolution

(PDB entry 1OED).²⁷The sequences of the a1, b2, b3,

and c2 subunits were obtained from the protein Data-

bank database (Accession Nos. a1, NM_174540; b2,

NM_000813; c2, NM_000816; b3, NM_000814). The

subunit correspondence between the nicotinic acetylcho-

line and GABA receptors was assigned based on func-

tional homology (Fig. 5B).⁴⁴The multiple sequence

alignment was performed with the MOE 2006.08 pro-

gram (Chemical Computing Group, Montreal, Canada)

using the PAM250 substitution matrix, and the long

loop between TM3 and TM4 was deleted because of

the lack of structure data in the template. About 20

intermediary models were generated by MOE, and the

raw model with the best packing quality was further re-

ﬁned until the root mean square gradient became 0.01,

according to the molecular mechanics minimization pro-

tocol using the Merck molecular ﬁeld force MMFF94x⁴⁵

in MOE. In the mean time, the validation tests of stereo-

chemical quality were conducted using MOE. Only 3

residues from the constructed b3 model and 4 from

the a1b2c2 model fell outside of the acceptable regions

of the Ramachandran plot (data not shown). All bond

angles, bond lengths, and Ca chiralities were also found

to be adequate. The structures of 1-phenyl-1H-1,2,3-

triazoles were built as described above. All the com-

pounds were automatically docked into the potential

binding site of b3 and a1b2c2 receptors separately using

the dock program in MOE. Each antagonist was sub-

jected to conformational analysis, placement, and phar-

macophore ﬁltering, and was scored in terms of

hydrophobic, ionic, and hydrogen-bond contacts, which

were generally favorable interactions. A ﬁnal docked

representation of the potential binding mode of the li-

gands was chosen based on selection of the compound

that possessed the best docking score within the most

populated cluster with the lowest docking energy.

3.6. Molecular modeling, CoMFA, and CoMSIA

3D-QSAR studies

The structures of all phenyltriazoles were constructed

based on the X-ray crystal data of 21¹⁹using the CRY-

SIN module in the molecular modeling software SYB-

YL ver. 7.1 (Tripos, St. Louis, MO) and were fully

optimized by the semi-empirical molecular orbital meth-

od AM1. All the molecules were superposed at the 4-po-

sition of the phenyl group and the 1- and 4-positions of

the triazole ring using the Fit Atoms menu of SYBYL.

A conventional CoMFA was performed to study the

3D-QSAR of phenyltriazoles in b3 receptors using steric

and electrostatic ﬁelds in SYBYL. All CoMFA calcula-

tions were performed with the SYBYL standard setup,

that is, steric and electrostatic ﬁelds described by Len-

nard-Jones and Coulomb potentials, respectively, dielec-

tric constant 1/r, and cutoﬀ 30 kcal mol^ꢃ1using an sp³

carbon atom with a charge of +1 as a probe atom.

CoMSIA was performed using the QSAR module in

SYBYL. Steric, electrostatic, and hydrophobic contri-

butions were evaluated using a common probe atom

as used in CoMFA. The steric contribution was calcu-

lated using the third power of the atomic radii. The

electrostatic ﬁeld was determined by the application of

semi-empirical AM1 MOPAC charges, and the hydro-

phobic ﬁeld was derived from atom-based parameters.

The attenuation factor (a) was deﬁned as 0.3 and used

as a default value.

The CoMFA and CoMSIA ﬁeld descriptors were used as

independent variables, and pK_ivalues were used as depen-

dent variables in the PLS analysis to derive 3D-QSAR

models. The K_ivalues of 1-phenyl-1H-1,2,3-triazoles in

Tables 1–3 were converted to pK_i(ꢃlogK_i) values. The

3.8. Insecticidal assays

The insecticidal activity of a representative disubsti-

tuted 1-phenyl-1H-1,2,3-triazole (15) was determined

5104

M. S. Alam et al. / Bioorg. Med. Chem. 15 (2007) 5090–5104

following the previously described method.¹⁹Brieﬂy,

diﬀerent concentrations of 15 in acetone (1 lL) were ap-

plied to the thoraces of adult female houseﬂies (WHO/

SRS strain, 3–5 days after emergence) that were pre-

treated with piperonyl butoxide (10 lg in 1 lL of ace-

tone) 1 h before the application of 15. Thirty

houseﬂies were used for each dosage. The houseﬂies

were maintained with sugar and water and were kept

at 25 ꢁC. The mortality rate was determined after 24 h,

and the experiments were repeated three times. The

LD₅₀value was calculated from the mean mortality

using the standard probit method.

17. Sammelson, R. E.; Caboni, P.; Durkin, K. A.; Casida, J.

E. Bioorg. Med. Chem. 2004, 12, 3345–3355.

18. Ozoe, Y.; Yagi, K.; Nakamura, M.; Akamatsu, M.;

Miyake, T.; Matsumura, F. Pestic. Biochem. Physiol.

2000, 66, 92–104.

19. Alam, M. S.; Kajiki, R.; Hanatani, H.; Kong, X.; Ozoe,

F.; Matsui, Y.; Matsumura, F.; Ozoe, Y. J. Agric. Food

Chem. 2006, 54, 1361–1372.

20. Sieghart, W.; Sperk, G. Curr. Top. Med. Chem. 2002, 2,

795–816.

21. Cheng, Y.; Prusoﬀ, W. H. Biochem. Pharmacol. 1973, 22,

3099–3108.

22. Cramer, R. D., III; Patterson, D. E.; Bunce, J. D. J. Am.

Chem. Soc. 1988, 110, 5959–5967.

23. Klebe, G.; Abraham, U.; Mietzner, T. J. Med. Chem.

1994, 37, 4130–4146.

Acknowledgments

24. Bo¨hm, M.; Sturzebecher, J.; Klebe, G. J. Med. Chem.

¨

1999, 42, 458–477.

25. Labrie, P.; Maddaford, S. P.; Fortin, S.; Rakhit, S.; Kotra,

L. P.; Caudreault, R. C. J. Med. Chem. 2006, 49, 7646–

7660.

26. Kulkarni, R. G.; Srivani, P.; Achaiah, G.; Sastry, G. N.

J. Comput. Aided Mol. Des. 2007, 21, 155–166.

27. Miyazawa, A.; Fujiyoshi, Y.; Unwin, N. Nature 2003, 423,

949–955.

We thank Merck Sharp & Dohme Ltd for supplying the

GABA_Areceptor plasmids. Part of this work was con-

ducted at the Department of Molecular and Functional

Genomics, Center for Integrated Research in Science,

Shimane University.

References and notes

28. Salany, A.; Zezula, J.; Tretter, V.; Seighart, W. Mol.

Pharmacol. 1995, 48, 385–391.

1. Waldvogel, H. J.; Billinton, A.; White, J. H.; Emson, P.

C.; Faull, R. L. M. J. Comp. Neurol. 2004, 470, 339–356.

2. Akabas, M. H. Int. Rev. Neurobiol. 2004, 62, 1–43.

3. Rabow, L. E.; Russek, S. J.; Farb, D. H. Synapse 1995, 21,

189–274.

4. Barnard, E. A.; Skolnick, P.; Olsen, R. W.; Mo¨hler, H.;

Sieghart, W.; Biggio, G.; Braestrup, C.; Bateson, A. N.;

Langer, S. Z. Pharmacol. Rev. 1998, 50, 291–313.

5. Bonnert, T. P.; McKernan, R. M.; Farrar, S.; Le Bour-

29. Xu, M.; Covey, D. F.; Akabas, M. H. Biophys. J. 1995, 69,

1858–1867.

30. Gurley, D.; Amin, J.; Ross, P. C.; Weiss, D. S.; White, G.

Receptors Channels 1995, 3, 13–20.

31. Perret, P.; Sarda, X.; Wolﬀ, M.; Wu, T.-T.; Bushey, D.;

Goeldner, M. J. Biol. Chem. 1999, 274, 25350–25354.

32. Bell-Horner, C. L.; Dibas, M.; Huang, R.-Q.; Drewe, J.

A.; Dillon, G. H. Mol. Brain Res. 2000, 76, 47–55.

33. Zorov, B. S.; Bregestovski, P. D. Biophys. J. 2000, 78,

1786–1803.

`

delles, B.; Heavens, R. P.; Smith, D. W.; Hewson, L.;

Rigby, M. R.; Sirinathsinghji, D. J. S.; Brown, N.;

Waﬀord, K. A.; Whiting, P. J. Proc. Natl. Acad. Sci.

U.S.A. 1999, 96, 9891–9896.

34. Buhr, A.; Wagner, C.; Fucks, K.; Seighart, W.; Sigel, E.

J. Biol. Chem. 2001, 276, 7775–7781.

35. Chen, L.; Durkin, K. A.; Casida, J. E. Proc. Natl. Acad.

Sci. U.S.A. 2006, 103, 5185–5190.

36. Ci, S.; Ren, T.; Su, Z. J. Mol. Model. 2007, 13,

457–464.

37. Hisano, K.; Ozoe, F.; Huang, J.; Kong, X.; Ozoe, Y.

Invert. Neurosci. 2007, 7, 39–46.

38. Ozoe, Y.; Ishikawa, K.; Tomiyama, S.; Ozoe, F.; Kozaki,

T.; Scott, J. G. In Synthesis and Chemistry of Agrochem-

icals Series VIII; Lyga, J. W.; Theodoritis, G., Eds.;

American Chemical Society: Washington, DC, 2007; ACS

Symposium Series 948, pp 39–50.

6. McKernan, R. M.; Whiting, P. J. Trends Neurosci. 1996,

19, 139–143.

7. Buckingham, S. D.; Sattelle, D. B.. In Comprehensive

Molecular Insect Science; Gilbert, L. I., Iatrou, K., Gill, S.

S., Eds.; Elsevier Pergamon: Amsterdam, 2004; Vol. 5, pp

107–142.

8. ﬀrench-Constant, R. H.; Mortlock, D. P.; Shaﬀer, C. D.;

MacIntyre, R. J.; Roush, R. T. Proc. Natl. Acad. Sci.

U.S.A. 1991, 88, 7209–7213.

9. ﬀrench-Constant, R. H.; Steichen, J. C.; Rocheleau, T. A.;

Aronstein, K.; Roush, R. T. Proc. Natl. Acad. Sci. U.S.A.

1993, 90, 1957–1961.

39. Hainzl, D.; Casida, J. E. Proc. Natl. Acad. Sci. U.S.A.

1996, 93, 12764–12767.

10. Ozoe, Y.; Akamatsu, M. Pest Manag. Sci. 2001, 57,

923–931.

40. Caboni, P.; Sammelson, R. E.; Casida, J. E. J. Agric. Food

Chem. 2003, 51, 7055–7061.

11. Zhao, X.; Yeh, J. Z.; Salgado, V. L.; Narahashi,

T. J. Pharmacol. Exp. Ther. 2004, 310, 192–201.

12. Grolleau, F.; Sattelle, D. B. Br. J. Pharmacol. 2000, 130,

1833–1842.

41. Hawkinson, J. E.; Drewe, J. A.; Kimbrough, C. L.; Chen,

J.-S.; Hogenkamp, D. J.; Lan, N. C.; Gee, K. W.; Shen,

K.-Z.; Whittemore, E. R.; Woodward, R. M. Mol.

Pharmacol. 1996, 49, 897–906.

13. Zhao, X.; Salagado, V. L.; Yeh, J. Z.; Narahashi,

T. J. Pharmacol. Exp. Ther. 2003, 306, 914–924.

14. Cole, L. M.; Nicholson, R. A.; Casida, J. E. Pestic.

Biochem. Physiol. 1993, 46, 47–54.

15. Ratra, G. S.; Casida, J. E. Toxicol. Lett. 2001, 122, 215–

222.

42. Yagle, M. A.; Martin, M. W.; de Fiebre, C. M.; de Fiebre,

N. C.; Drewe, J. A.; Dillon, G. H. Neurotoxicology 2003,

24, 817–824.

43. Bradford, M. M. Anal. Biochem. 1976, 72, 248–254.

44. Campagna-Slater, V.; Weaver, D. F. J. Mol. Graphics

Modell. 2007, 25, 721–730.

16. Ratra, G. S.; Kamita, S. G.; Casida, J. E. Toxicol. Appl.

Pharmacol. 2001, 172, 233–240.

45. Halgren, T. A. J. Comput. Chem. 1996, 17, 490–519.

46. Miller, C. Neuron 1989, 2, 1195–1205.

Article Doi

Synthesis, 3D-QSAR, and docking studies of 1-phenyl-1H-1,2,3-triazoles as selective antagonists for β3 over α1β2γ2 GABA receptors

DOI: 10.1016/j.bmc.2007.05.039

Source and publish data:

Authors:

Article abstract of DOI:10.1016/j.bmc.2007.05.039

Full text of DOI:10.1016/j.bmc.2007.05.039

Products guided by the article

R&D Labs maybe for 133152-04-2

Relevant to this article

Hot Product