A highly predictive 3D-QSAR model for binding to the voltage-gated sodium channel: Design of potent new ligands

Article abstract of DOI:10.1016/j.bmc.2013.11.049

A comprehensive comparative molecular field analysis (CoMFA) model for the binding of ligands to the neuronal voltage-gated sodium channel was generated based on 67 diverse compounds. Earlier published CoMFA models for this target provided μM ligands, but the improved model described here provided structurally novel compounds with low nM IC₅₀. For example, new compounds 94 and 95 had IC₅₀ values of 129 and 119 nM, respectively.

Full text of DOI:10.1016/j.bmc.2013.11.049

Bioorganic & Medicinal Chemistry 22 (2014) 95–104
Contents lists available at ScienceDirect
Bioorganic & Medicinal Chemistry
journal homepage: www.elsevier.com/locate/bmc
A highly predictive 3D-QSAR model for binding to the voltage-gated
sodium channel: Design of potent new ligands
Congxiang Zha a, George B. Brown b, Wayne J. Brouillette a,
⇑
a Department of Chemistry, The University of Alabama at Birmingham, Birmingham, AL 35294, United States
b Department of Psychiatry and Behavioral Neurobiology, The University of Alabama at Birmingham, Birmingham, AL 35294, United States
a r t i c l e i n f o
a b s t r a c t
Article history:
A comprehensive comparative molecular field analysis (CoMFA) model for the binding of ligands to the
neuronal voltage-gated sodium channel was generated based on 67 diverse compounds. Earlier published
Received 15 September 2013
Revised 17 November 2013
Accepted 25 November 2013
Available online 6 December 2013
CoMFA models for this target provided
lM ligands, but the improved model described here provided
structurally novel compounds with low nM IC50. For example, new compounds 94 and 95 had IC50 values
of 129 and 119 nM, respectively.
Ó 2013 Elsevier Ltd. All rights reserved.
Keywords:
Sodium channel
CoMFA model
Nanomolar ligands
1. Introduction
tained depolarization of ischemic brain tissues. Therefore, drugs
that selectively block voltage-sensitive Na+ channels (particularly
Voltage-gated sodium channels (VGSCs) are responsible for the
generation of conducted electrical signals in neurons and other
excitable cells.1–5 Upon polarization, the permeability of the cell
membrane to sodium ions increases dramatically over a period of
0.5 to hundreds of msec and then decreases to the baseline level
over a period of 2 ms to a few seconds. This biphasic behavior re-
sults from two experimentally separable gating processes that con-
trol ion channel function: activation, which controls the rate and
voltage dependence of the permeability increase following depo-
larization, and inactivation, which controls the rate and voltage
dependence of the subsequent return of the ion permeability to
the resting level during a maintained depolarization.
The properties of the sodium channel have been shown to be
important for several physiological and pathological functions.6–8
It not only is responsible for the initiation and conduction of neu-
ronal action potentials but can also influence neurotransmitter re-
lease from presynaptic vesicles. Furthermore, the maintenance of
ionic homeostasis of neurons is the basis for normal functions;
thus, any improper transportation of sodium ions can cause neu-
rons to malfunction. Meanwhile, in neuronal preparations very
small persistent Na+ currents arise from sustained or repeated
Na+ channel openings. These persistent currents last much longer
than the 1–5-ms duration of action potentials.5 It is likely that
the persistent current is involved with the repeated depolarizing
waves associated with seizures, spreading depression and the sus-
persistent Na+ currents) are a reasonable choice for treating epi-
lepsy, stroke, cerebral ischemia, and head trauma. Sodium chan-
nels have also been targeted for the design of numerous other
drugs, including local anesthetics, class I antiarrhythmics, anticon-
vulsants which have
a pharmacological profile like that of
diphenylhydantoin (DPH, or phenytoin), analgesics, and neuropro-
tective agents.9–17
While VGSCs are well recognized to exist in excitable tissues
such as nerve and muscle, in the past fifteen years they have been
found in non-excitable tissues, including a broadening number of
aggressive metastatic cancers,18 but they are not found in nonag-
gressive cancers or normal tissue. A growing body of evidence sug-
gests that VGSCs play an important pathological role during cancer
progression toward metastasis. Their role in regulating cellular
migration and invasion, and their potential utility as diagnostic
and therapeutic targets has drawn increasing attention. In fact,
known drugs that target sodium channels such as anticonvulsants,
local anesthetics, and antiarrhythmics have been observed to sig-
nificantly reduce the metastasis of tumor cells.19–24 The diverse
and growing number of important applications for this channel
as a drug target emphasizes the need for efficient drug design
methodologies.
In contrast to the widely and thoroughly studied physiological
and pharmacological properties of the sodium channel, the
3-dimensional (3-D) structure of the binding site is much less under-
stood due to difficulties in crystallizing the high molecular weight
transmembrane protein from mammalian cells. Historically, design
of drugs for sodium channel-related diseases has been carried out
⇑
Corresponding author. Tel.: +1 205 934 8288.
E-mail address: wbrou@uab.edu (W.J. Brouillette).
0968-0896/$ - see front matter Ó 2013 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.bmc.2013.11.049

96
C. Zha et al. / Bioorg. Med. Chem. 22 (2014) 95–104
largely by screening compound arrays. 3-D QSAR models generated
from activities of known compounds are attractive alternatives to
assist medicinal chemists in rational drug design when crystal
structures of protein targets are not available. The utility of such
models lies in their ability to accurately predict the biological
activities for newly designed ligands.
phobic groupings: two phenyl groups at the 5-position of the
hydantoin ring and a hydrophobic substituent at the N-3 position.
This also suggests that further spatial information beyond the
existing pharmacophore needs to be uncovered.
Recently, Brown and co-workers published two CoMFA models
for the neuronal voltage-gated sodium channel.27,28 In one model,
the authors combined the pharmacophore of the CoMFA model
generated by using a group of hydantoins as the training set, and
the scaffold of propofol, for new design. In the other model, the
authors combined the pharmacophore of local anesthetics, hydan-
toins and hydroxyphenylamides. These models offer new insights
about the binding of the blockers to the channel, but the most po-
tent newly designed compounds based on the models were micro-
molar blockers. As for most drug targets, it is desirable to develop
models that can be used to design ligands with IC50 values at the
low nanomolar or better level.
In a previous paper25 we developed a preliminary 3-D QSAR
(comparative molecular field analysis, or CoMFA) model for the
binding of ligands to the sodium channel that was trained using
a group of 5-phenylhydantoins. However, this model is highly lim-
ited since only a single class of compounds (hydantoins) was in-
cluded in the training set, which can probe only limited 3
dimensional steric and electrostatic space. To evaluate further
the importance of the hydantoin ring, and to explore spatial fea-
tures of related non-hydantoin compounds to increase structural
diversity, we previously reported a number of new types of sodium
channel ligands and a systematic SAR study for these com-
pounds.26 The results suggest that the high potency of ligands for
this channel may result from the contribution of distinct hydro-
To achieve this goal, we proposed to develop a more compre-
hensive 3-D QSAR model that can include structural information
associated with a much greater variety of structural types and drug
O
4
3
O
O
O
4
4
HN
4
4
HN
O
HN
3
NH
O
HN
3
3
3
NH
O
1
1
9
NH
1
2
NH
1
1
N
2
2
5
2
2
5
NH
5
5
6
10
6
5
6
O
6
O
8
7
O
7
8
9
7
7
11
1
8
2
3
4
4
4
3
O
O
4
4
R
6
3
O
O
O
1
3
3
N
HN
2
2
HN
NH
O
NH
1
1
1
R
2
2
NH
N
O
C4H9
7
O
O
R = C3H7, C5H11, C7H15
R = C3H7, C5H11, C6H13, C7H15, C9H19, Ph
13
12
14
16
15
8
10
6
7
9
5
4
4
4
3
4
O
O
3
HN
3
HN
1
3
OH
OH
1
1
6
NH2
1
O
O
2
N
2
N
2
2
OH
S
S
R
R
5
R
O
5
O
6
O
O
O
O
24
R = C7H15, C9H19
R = C3H7, C5H11, C7H15, C9H19
17
22
23
18
19
20
21
4
10
3
6
OH
9
6
7
R
O
O
1
3
H
5
1
3
9
OH
OH 8
9
N
2
N
N
7
2
8
4
5
6
2
1
8
7
R
4
H
H
O
R
R = C2H5, C5H11
R = C2H5, C5H11
27
R = C2H5, C5H11
25
29 30
28
26
4
3
8
O
7
2
7
OH
1
N
O
3
HN
4
9
1
R
NH
N
6
5
5
6
5
N
R
10
2
N
2
N
4
N
R
3
1
10
8
9
O
O
O
R = C2H5, C5H11
R = C5H11, C9H19
R = CH3, C5H11
33 34
35 36
31 32
15
6
12
O
5
4
12
8
11
7
N
4
11
6
8
1
3
2
10
5
14
16
9
7
1
3
N
13
O
10
9
2
O
38
37
Figure 1. Compounds used in the training set.

C. Zha et al. / Bioorg. Med. Chem. 22 (2014) 95–104
97
Cl
4
11
Cl
15
1
N
14
16
8
16
13
12
1
10
4
O
8
3 2
10
N
2
5
15
N
11
9
N
9
12
5
3
14
6
7
40
Cl Cl
N
6
7
13
Cinnarazine
39
Cl
Cl
12
11
N
4
Calmdazolium
8
N
Cl
CF3
3
10
OH
Cl
2
1
O
9
7
43
F
5
6
Fluperamide
4
Cl
O
11
15
1
N
14
16
8
13
12
O
10
3
2
3
1
12
11
10
N
O
O
4
9
5
N
4
8
N
Cl
7
42
3
6
N
H
OH
1
O
9
7
44
Flunarine
5
F
41
6
Loperamide
Felodipine
O
12
O
10
11
O
O
1
9
8
4
13
16
NO2
O
NO2
O
CN
3
5
14
O
O
N
O
15
6
3
1
3
2
1
7
O
O
2
4
O
O
2
4
Meo_Verapamil
N
H
45
N
H
46
47
Nifedipine
Nisodipine
O2N
5
4
12
O
11
10
H
N
9
10
9
8
O
O
O
4
11
O
O
1
4
13
16
3
2
3
1
8
CN
3
1
5
N
14
O
15
O
2
7
6
6
7
2
N
H
49
48
50
Nitrendipine
Prenylamine
Verapamil
H2N
1
3
1
O
2
2
O
O
2
S
NH2
1
S
3
NH2
O
O
3
N
53
Benzocaine
O
N
H
4
4
4
H
W-5
52
51
W-7
Fig. 1 (continued)
classes that all bind to the sodium channel. This includes local
anesthetics, anticonvulsants, and antiarrhythmics, which have
been proposed to bind to the same receptor site.29–34 In the present
study we used a training set of 67 sodium channel blockers, 36 of
which were synthesized in our laboratory and 31 sodium channel
blockers from the literature.35 Of the 31 compounds from the liter-
ature, 15 compounds are local anesthetics, one is the anticonvul-
sant DPH, and the other 15 compounds are Ca2+ and Na+ channel
mixed blockers (Fig. 1). The dependent variable is the IC50 value
high-quality model and its utility in designing new potent ligands
with low nanomolar IC50 values.
2. Methods
2.1. Biological data
In the sodium channel evaluation, the IC50, which represents the
concentration of compound required to displace 50% of specifically
bound [3H]BTX-B, was determined in an in vitro assay using rat
brain cerebral cortex synaptoneurosomes. For each IC50 value,
5–7 concentrations (triplicate determinations) were used. The
IC50 value was obtained from a Probit analysis of the data that plots
the log concentration versus the Probit number (converted from
percent binding of [3H]BTX-B for the samples). The concentration
where the Probit number equals 5 is the IC50 value. The activities
of compounds from the literature were obtained using an identical
obtained using the [3H]batrachtoxinin A 20- -benzoate ([3H]BTX-B)
a
sodium channel binding assay. All of the literature compounds
were taken from a single literature reference36 that used the same
[3H]BTX-B binding assay that we employed. It was anticipated that
data obtained using similar assays would reduce errors in the
model. CoMFA calculations were then carried out to correlate
the effects of changes in steric and electrostatic effects for the
67 compounds in the database. We describe here the resulting

98
C. Zha et al. / Bioorg. Med. Chem. 22 (2014) 95–104
4
7
5
O
N
O
O
N
5
O
2
7
1
4
2
H
N
4
N
6
6
7
N
3
3
N
6
3
11
O
1
H
8
2
5
1
O
O
54
9
10
Dibucaine
Bupivacaine
56
Ocaine
55
10
8
6
5
2
O
9
O
O
1
2
4
2
4
6
N
6
5
N
N
8
N
N
3
3
5
1
7
H
3
1
H
4
59
Lidocaine
7
58
Etidocaine
Diphenyamine
57
O
O
2
4
H
6
7
2
3
N
2
N
4
O
N
N
N
H
3
1
1
5
3
H
1
8
O
61
4
60
Mepivacaine
62
Piperocaine
Prilocaine
H2N
O
2
4
NH2
2
4
N
N
3
1
H
O
3
N
2
3
1
4
1
H
O
OH
65
63
Procaine
64
Propranol
Tocainide
9
O
10
11
5
1
7
O
4
O
6
13
5
8
O
N
3
N
2
O
3
N
N
N
O
7
2
1
12
H
8
H
4
6
H
66
Tetracaine
67
QX572
Fig. 1 (continued)
binding assay, which gave the same value for DPH (IC50 = 40
obtained in the present assay.
l
M)
the training set are shown in Figure 1, and lowest conformations
of the training set used in this study are listed in Table 1S
(Supplementary data). See Supplementary data—conformational
search and alignment for details.
2.2. Conformational search
All conformational searches were performed on a SGI Octane
using SYBYL software (Tripos Assoc., Inc.). In conformational
searches, two assumptions were made: (1) all molecules are in a
fully extended conformation, and (2) all chiral centers are R unless
the actual configuration was known. Based on the first assumption,
if a molecule had ring system(s) which belong to part of the ex-
tended chain between two ends of the molecule, extended confor-
mations which contain part of the ring system were used if they
belonged to the lowest energy conformations (within 2 kcal/mol
of the global energy minimum). For compounds containing the
diphenylmethyl fragment, only one low-energy conformation
was used for the diphenylmethyl group. For compounds with a ter-
minal phenyl ring, the conformation with the phenyl ring orienta-
tion corresponding to one of the two phenyl groups in the
diphenylmethyl compounds was included in the analysis. No con-
formations that had conformational energy >2 kcal/mol above the
global energy minimum were considered. Based on these assump-
tions and definitions, conformational searches using the ‘grid
search’ and ‘random search’ routines of SYBYL were performed
with the remaining rotatable bonds. The chemical structures of
2.3. CoMFA calculation
CoMFA, using default parameters except where noted, was cal-
culated in the QSAR option of SYBYL 6.4 on the SGI Octane com-
puter. The CoMFA grid spacing was 2.0 Å in the x, y and z
directions, and the grid region was automatically generated by
the CoMFA routine to encompass all molecules with an extension
of 4.0 Å in each direction. A sp3 carbon and a point charge of +1.0
were used as probes to generate the interaction energies at each
lattice point. The default value of 30 kcal/mol was used as the max-
imum electrostatic and steric energy cutoff.
CoMFA calculations were then performed on the training set.
The sodium channel binding activities of all compounds in the
training set are listed in Table 1. These calculations generated a
model with a cross-validated q2 = 0.765 and a non-cross-validated
R2 = 0.951.
We also used the non-cross-validated CoMFA model to predict
the sodium channel binding activities for the lowest energy confor-
mations of the test set, compounds 68–87, and the newly designed
compounds 88–97.

C. Zha et al. / Bioorg. Med. Chem. 22 (2014) 95–104
99
Table 1
for the neuronal voltage-gated sodium channel. Several of these
were selected for synthesis, and the methods employed are shown
in Scheme 1. Compounds 88 and 89 (Scheme 1) were synthesized
by reacting (3-aminopropyl)benzene with the appropriate alkyl
chloride in DMSO at room temperature. Compounds 90 and 91
were synthesized in the same manner starting with (2-amino-
ethyl)benzene. Compounds 92–97 (Scheme 1) were synthesized
by reacting N-(diphenylmethyl)piperazine with the appropriate
halide at 70–80 °C in DMSO.
The predicted and observed activities for the compounds in the training set
Compounds
Observed LogIC50
Predicted LogIC50
Residual
1
2
3
4
5
6
7
8
9
3.32
2.93
2.40
2.40
2.21
1.59
1.11
0.70
0.70
1.60
2.40
2.86
2.45
2.65
2.50
1.48
1.70
2.70
2.05
1.11
1.10
0.95
0.81
1.92
1.48
0.70
1.54
0.72
0.91
0.40
0.80
0.68
0.52
0.11
0.84
0.71
0.63
ꢀ0.08
0.85
ꢀ0.35
1.57
ꢀ0.22
0.55
ꢀ0.56
0.56
1.57
1.41
1.56
ꢀ0.52
0.52
1.34
1.08
2.86
0.73
1.69
0.78
0.15
0.54
1.68
2.28
1.11
1.73
2.04
1.19
2.30
0.53
0.59
3.18
2.56
2.44
2.31
2.38
1.88
1.31
1.08
0.77
1.92
2.56
2.61
2.17
2.49
1.91
1.18
1.68
2.66
2.11
1.13
1.12
1.29
0.97
1.77
1.69
1.03
1.33
1.00
0.87
0.50
0.89
0.55
0.37
0.05
0.78
0.57
0.71
ꢀ0.18
0.80
ꢀ0.31
1.66
ꢀ0.33
0.85
ꢀ0.69
0.47
1.40
0.96
1.53
ꢀ0.34
0.53
1.47
1.44
2.61
0.92
1.67
1.28
0.09
0.47
1.68
2.04
1.10
1.62
1.90
1.10
2.53
0.51
0.83
0.14
0.37
ꢀ0.04
0.09
ꢀ0.17
ꢀ0.29
ꢀ0.20
ꢀ0.38
ꢀ0.07
ꢀ0.32
ꢀ0.16
0.25
0.28
0.16
0.59
0.30
3. Results and discussion
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
3.1. Basic features of the model
The CoMFA calculation was carried out for the training set con-
taining the 67 compounds. A very good correlation with a cross-
validated q2 = 0.765 and a non-cross-validated R2 = 0.951 was
achieved. The predicted and observed log IC50 values for the com-
pounds in the training set are listed in Table 1. As shown in
Figure 2, the correlation between the predicted and the observed
values is excellent, with a correlation coefficient R2 = 0.946
(obtained from a regression analysis by Sigmaplot, and slightly
different from the non-cross-validated R2 = 0.951 obtained from
the CoMFA calculation).
The steric and electrostatic properties of the CoMFA model are
illustrated separately as stereoviews in Figures 3 and 4, respec-
tively. Compared to the known models,25,27,28 the new model is
much richer in spatial information regarding steric and electro-
static factors for the binding of ligands to the sodium channel, as
evidenced by the maps of either steric or electrostatic influences
on activity. Figure 3 describes the steric map, and yellow contours
represent the steric limit occupied by good ligands (additional ste-
ric occupancy beyond this contour is unfavorable), while green re-
gions represent areas where increased steric effects are favorable.
Thus Figure 3 reveals one major steric region, near the n-alkyl
group of 32, where additional steric bulk is favorable. Regarding
the electrostatic properties in Figure 4, the model reveals a major
favorable positively charged region in the central part of the
map, and several favorable negatively charged regions. The major
positively charged region is nicely occupied by the piperazine
nitrogens of 32.
0.02
0.04
ꢀ0.06
ꢀ0.02
ꢀ0.02
ꢀ0.34
ꢀ0.16
0.15
ꢀ0.21
ꢀ0.33
0.21
ꢀ0.28
0.04
ꢀ0.10
ꢀ0.09
0.13
0.15
0.06
0.06
0.12
ꢀ0.08
0.10
0.05
ꢀ0.04
ꢀ0.09
0.11
ꢀ0.30
0.13
0.09
0.17
0.45
0.03
ꢀ0.18
ꢀ0.01
ꢀ0.13
ꢀ0.36
0.25
3.2. The predictive ability of the comprehensive model
To test the predictive ability of this new CoMFA model, a test set
containing 20 compounds (68–87), all sodium channel blockers
with known activity that were not used in the training set, was
constructed. The structures of these compounds are shown in
Figure 5. These compounds were selected from the same literature
reference35 that was used to provide some of our training set
compounds, which can limit errors caused by different testing
methodologies.
ꢀ0.19
0.02
ꢀ0.50
0.06
0.07
0.00
0.24
0.01
0.11
0.14
0.08
Both our current and previously published CoMFA models25
were used to predict the sodium channel binding activities for
the compounds in the test set. The results are listed in Table 2.
From the results it is clear that the current model provides signif-
icantly better predictive ability than the old model for most com-
pounds in the test set. For example, test set compound 77 had an
experimentally measured log IC50 = ꢀ0.22, while the old model
gave a predicted log IC50 = 0.75 (residual = ꢀ0.97) and the new
model gave a predicted log IC50 = 0.09 (residual = ꢀ0.31). As true
for 77, most compounds in the test set had much greater accuracy
of prediction by the new model as compared to the old model. Of
the 20 compounds in the test set, activities for only 68, 72, and
81 were better predicted by the old model versus the new one.
ꢀ0.23
0.02
ꢀ0.24
q
2 = 0.765 for crossvalidation; 2 = 0.951 for non-crossvalidation. The optimum
r
number of components is 5.
2.4. Chemistry
The newly generated CoMFA model was used to design a num-
ber of new compounds predicted to have potent binding affinities

100
C. Zha et al. / Bioorg. Med. Chem. 22 (2014) 95–104
F
F
DMSO
R1
Cl
N
+
R1 NH2
H
F
F
88
89
PhCH2CH2CH2
PhCH2CH2
R1
=
=
R1
H
N
+
NH2
BrCH2(CH2)nPh
n = 1,2
n
90 n=1
n=2
91
R
Cl
N
NH
N
N R
DMSO, 70-80 °C
R =
O
O
O
94
93
92
F
97
96
95
F
Scheme 1. Synthesis of newly designed ligands.
Figure 3. Stereoview for compound 32 posed in the steric map of the CoMFA
model. Green: increased steric bulk here results in increased activity; Yellow: limit
of steric space. Increased bulk here decreases activity. For 32, the n-C5H11 alkyl
group at the ‘top’ occupies a sterically favorable region, while the remainder of 32
fills much of the available favorable steric space.
Figure 2. The correlation between the predicted and the observed activities for the
training set.
tematic changes were made on structural templates that
contained favorable features of the CoMFA model. The model was
then used to predict the effect on activity for subtle changes in
presentation of charge and steric effects for each hypothetical new
3.3. Design of new, potent sodium channel ligands
In order to use the current CoMFA model to design new com-
pounds, we employed a ‘computational SAR’ approach, where sys-

C. Zha et al. / Bioorg. Med. Chem. 22 (2014) 95–104
101
as 32 is favorable. As shown in Table 3, in one SAR study we utilized a
simple template containing these features and systematically varied
the length of the linker between the two hydrophobic end groups
(compounds SAR-1–SAR-6). This revealed a chain length optimum
at n = 5. Similarly, we explored a variety of other structural effects,
including the location, nature, and number of positive charges with-
in the structure, and the size, nature, and location of various hydro-
phobic groups, among others.
Based on the information revealed by these types of studies, we
designed several new sodium channel ligands that were subse-
quently synthesized, compounds 88–97. Their structures and the
synthetic procedures employed are shown in Scheme 1, and the
observed and predicted IC50 values are given in Table 4. Several
of these contained the N-diphenylmethylpiperazine fragment, but
we also focused on simpler structures. These newly designed com-
pounds (88–97) were predicted by the model to span a range of
IC50 values from 251 to 1000 nM. The experimentally determined
IC50 values in most cases were well predicted by the model, and
the prediction in general did not differ from the observed value
by more than 2-fold (with the exception of 90).
Figure 4. Stereoview for compound 32 posed in the electrostatic map of the CoMFA
model. Blue: region where positive (+) charge increases activity; Red: region where
negative (ꢀ) charge increases activity. The piperazine nitrogens of compound 32
occupy favorable positively charged areas near the central region of the map.
compound. For example, many of the more potent compounds con-
tained the diphenylmethylamino fragment, and the CoMFA steric
model (Fig. 3) suggested that steric bulk at each end of a ligand such
Similar to cinnarizine (40), a nanomolar compound used in the
training set for the CoMFA calculation, compounds 92–97 in Table 4
O
H
H2N
O
N
S
N
O
O
N
H
N
H
O
O
O
O
70 Butethamine
69 Carticaine
Benzoxinate
68
O
O
O
O
N
N
N
O
O
O
O
O
O
71
72
Dilazep
Dicyclomine
CF3
O
N
O
HN
N
OH
75
S
N
O
F
N
N
N
Pheniramine
N
73
Droperidol
74
Fluphenazine
O
F
HN
N
O
N
N
N
N
O
F
76
Haloperidol
78
Dimethisoquine
Cl
77
Spiroperidol
H2N
O
N
N
H
O
N
O
O
80
Hexylamine
O
79
Phenoxybenzamine
81 Butacaine
H2N
O
O
O
O
N
O
O
H
N
N
N
H
83
Meprylcaine
Phenacaine
82
Propoxycaine
84
NH2
N
N
O
O
O
O
O
N
N
O
OH
Pramoxin
86
N
N
85
Prazosin
Diphenidol
O
87
Figure 5. Compounds used in the test set.

102
C. Zha et al. / Bioorg. Med. Chem. 22 (2014) 95–104
Table 2
designed that do not contain this moiety. Of these, compounds
88, 90, and 91 are low nanomolar (nM) ligands. The simplest was
compound 90, which is a dialkylamine containing only one phenyl
group at each end. To the best of our knowledge, compound 90 is
among the smallest known low nM blockers of this channel.
In summary, a high-quality comprehensive 3-D QSAR (CoMFA)
model was generated by using a training set of 67 sodium channel
ligands (36 synthesized in this laboratory and 31 selected from the
literature). This is the first published 3D-QSAR model for this target
that allows design of low nM ligands. The newly generated CoMFA
model was used to guide design for new sodium channel ligands
predicted to have nanomolar IC50 values, and several structurally
simple examples were synthesized. The IC50 values were well pre-
dicted by the model, suggesting the potential of this and related
models for guiding the design of new sodium channel ligands that
are potential drugs with applications in epilepsy, neuroprotection,
analgesia, arrhythmias, and/or metastatic cancer.
The predicted activity and actual activity for compounds in the test set by the old
model25 and current model
Compound Observed
LogIC50
Current model
Old model
Predicted Residual Predicted
LogIC50
Residual
LogIC50
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
0.46
1.71
1.74
0.45
ꢀ0.22
ꢀ0.13
0.57
1.52
0.08
ꢀ0.22
0.53
0.85
1.65
0.43
1.38
0.88
0.23
1.33
1.00
0.53
0.60
1.63
0.94
0.48
0.26
0.50
0.01
1.78
0.25
0.09
1.02
0.86
1.84
0.54
1.16
0.93
0.64
1.24
1.19
0.82
ꢀ0.16
0.08
0.80
0.52
2.13
0.53
1.19
0.07
0.90
0.02
2.18
0.89
0.75
1.82
1.19
0.83
0.44
0.72
0.93
1.29
1.44
2.41
1.07
ꢀ0.06
ꢀ0.42
1.21
ꢀ0.74
ꢀ0.29
ꢀ1.03
0.55
ꢀ0.66
ꢀ0.81
ꢀ0.97
ꢀ1.29
ꢀ0.34
0.82
ꢀ0.03
ꢀ0.48
ꢀ0.63
0.56
ꢀ0.26
ꢀ0.17
ꢀ0.31
ꢀ0.49
ꢀ0.01
ꢀ0.19
ꢀ0.11
0.22
ꢀ0.01
0.66
ꢀ0.05
ꢀ0.41
0.09
ꢀ0.05
ꢀ1.06
ꢀ0.11
ꢀ1.41
ꢀ0.54
4. Experimental section
Melting points were recorded on an Electrothermal melting
point apparatus and are uncorrected. IR spectra were recorded on
Bruker Vector 22 and Arid-Zone™ Bomem MB-series spectrome-
ters. Elemental analyses were performed by Atlantic Microlabs of
Norcross, GA. 1H and 13C NMR spectra were recorded on a Bruker
(ARX series) NMR spectrometer operating at 300.1 MHz (for 1H),
and 75.4 MHz (for 13C). The MS spectra were recorded on a Per-
kin-Elmer SCIEX API triple-quadrupole mass spectrometer using
electrospray ionization. Analytical chromatography was performed
ꢀ0.19
ꢀ0.19
Correlation: current model, Y = 0.69X + 0.32, r2 = 0.718; old model, Y = 0.41X + 0.76,
r
2 = 0.16.
Table 3
The predicted effect (CoMFA) of number of atoms between the diphenylmethylamino
group and a second hydrophobic group
using Whatman PE Sil G/UV silica gel plates (250
lm). Flash chro-
matography was performed on J.T. Baker silica gel (40
l
m). All
commercially obtained reagents were used as received unless
otherwise noted.
N
n
H
4.1. General procedure for the preparation of compounds 88–91
To a solution of alkylamine (2 eqs) in DMSO (10 mL) was added
the appropriate alkyl halide (1 equiv). The mixture was stirred at
room temperature for 12 h until the halide disappeared from
TLC. Ethyl acetate (100 mL) was added and the mixture was
washed with 5% Na2CO3 (3 ꢁ 30 mL). The organic layer was dried
(Na2SO4) and evaporated to give a crude product, which was puri-
fied on a silica gel column (40% ethyl acetate/55% hexanes/5%
Et3N). In this manner were prepared the following compounds.
Compound
n
IC50 (nM)_
SAR-1
SAR-2
SAR-3
SAR-4
SAR-5
SAR-6
1
2
3
4
5
6
3160
760
1330
562
345
560
4.2. 5,5-Di-(4-fluorophenyl)-N-(3-phenylpropyl)pentanamine
(88)
Table 4
The observed and the predicted ic50 values for newly designed compounds synthe-
sized in Schemes 1
From 3-phenylpropylamine (0.482 g, 3.56 mmol), and 4,4-di-(4-
fluorophenyl)butyl chloride (0.500 g, 1.72 mmol) was prepared 88
(0.488 g, 75%) as a thick liquid (Rf = 0.65). 1H NMR (CDCl3) d 7.29–
6.91 (m, 13H, 3Ph), 3.85 (t, J = 7.5 Hz, 1H, CH(PhF)2), 2.66–2.57 (m,
6H, CH2NHCH2 & CH2Ph), 2.03–1.96 (m, 2H, CH2CH2CH(PhF)2),
1.83–1.73 (m, 2H, CH2CH2Ph), 1.46–1.36 (m, 3H, NH & CH2CH2-
CH(Ph(F))2); 13C NMR (CDCl3) d 163.4, 160.0, 142.1, 140.8, 140.7,
129.5, 129.4, 128.8, 128.7, 126.3, 115.8, 115.5, 50.0, 49.5, 49.2,
33.89, 33.86, 31.1, 28.0; MS 380 (M+1)+; IR 3058 (N-H) cmꢀ1; Anal.
Calcd for C22H29N: C, 79.13; H, 7.17; N, 3.69. Found: C, 79.39; H,
7.19; N, 3.75.
Compound
Predicted IC (nM)
Observed IC50 (nM)
50
88
89
90
91
92
93
94
95
96
97
360
801
1000
443
280
283
251
260
340
470
229 [104–382]a
1230 [506–2200]
340 [200–500]
252 [96–420]
156 [82–240]
283 [192–420]
128 [56–260]
119 [45–198]
192 [106–340]
257 [120–446]
a
Numbers in brackets represent 1 stand deviation.
4.3. 4,4-Di-(4-fluorophenyl)-N-(2-phenylethyl)butanamine (89)
From 2-phenylethylamine (0.180 g, 1.49 mmol) and 2-phenyl-
ethyl bromide (0.174 g, 0.620 mmol) was prepared 89 (0.192 g,
85%) as a liquid (Rf = 0.35): 1H NMR (CDCl3) d 7.25–6.82 (m, 13H,
all contain the diphenylmethylpiperazine fragment. While these
compounds are all potent sodium channel ligands, of greater
interest are the structurally much simpler compounds, 88–91, we

C. Zha et al. / Bioorg. Med. Chem. 22 (2014) 95–104
103
3Ph), 5.91 (s, 2H, NH2+), 3.75 (t, 1H, CHPh2), 2.89–2.76 (m, 4H,
2NCH2), 2.65 (t, 2H, PhCH2), 1.96–1.87 (m, 2H, CH2CHPh2), 1.47–
1.38 (m, 2H, CH2CH2CHPh2); 13C NMR (CDCl3) d 163.4, 160.2,
140.7, 140.6, 129.5, 129.4, 129.1, 129.0 127.0, 115.9, 115.6, 50.4,
49.9, 48.9, 35.1, 33.7, 27.1; MS 366 (M+1)+; IR 3040 (N-H), 2936
(Ar-H) cmꢀ1; Anal. Calcd for C24H25NF2: C, 75.17; H, 7.10; N, 3.65.
Found: C, 74.90; H, 6.96; N, 3.73.
(1.40 g, 64%) as a white solid (solvent for column: 90% hexanes +
8% ethyl acetate + 2% triethylamine, Rf = 0.40): mp 101–103 °C;
1H NMR (CDCl3) d 7.77–7.18 (m, 15H, 3Ph), 6.87 (s, 1H, NHCO),
4.23 (s, 1H, CHPh2), 3.55–2.50 (m, 2H, CH2NHCO), 2.62–2.44 (m,
10H, N(CH2)2(CH2)2NCH2); 13C NMR (CDCl3) d 167.4, 142.7, 134.7,
131.3, 128.5, 127.9, 127.0, 126.9, 76.3, 56.3, 53.1, 52.0, 36.3; IR
3334 (N-H), 1639 (C@O) cmꢀ1; MS 400 (M+1)+; Anal. Calcd for
C
26H28N3O: C, 78.16; H, 7.32; N, 10.52. Found: C, 78.00; H, 7.27;
4.4. N-(2-Phenylethyl)-3-phenylpropanamine (90)36
N, 10.29.
From 3-phenylpropylamine (0.180 g, 1.33 mmol) and 2-phenyl-
ethyl bromide (0.103 g, 0.556 mmol) was prepared 90 (0.130 g,
78%) as a liquid (Rf = 0.45): 1H NMR (CDCl3) d 9.12 (s, 2H, NH2+),
7.23–7.05 (m, 10H, 2Ph), 2.73 (t, 4H, 2NCH 2), 2.56 (t, 4H, 2PhCH2),
1.97–1.87 (m, 2H, PhCH2CH2CH2); 13C NMR (CDCl3) d; MS 240
4.9. 1-(4-(4-t-Butylphenyl)-4-oxobutyl)-4-(diphenylmethyl)-
piperazine (94)
From
4-chloro-1-(4-t-butylphenyl)-1-butanone
(0.500 g,
2.09 mmol) and 1-(diphenylmethyl)piperazine (1.20 g, 4.76 mmol)
was prepared 94 (0.488, 51%) (solvent for column: 90% hexanes +
8% ethyl acetate + 2% triethylamine, Rf = 0.60) as a thick liquid.
1H NMR (CDCl3) d 7.90–7.88 (2H, Ph), 7.47–6.91 (m, 12H, 3Ph),
4.18 (s, 1H, CHPh2), 2.96 (t, J = 7.2 Hz, 2H, CH2CO), 2.46–2.27 (m,
10H, N(CH2)2(CH2)2NCH2), 2.20–1.96 (m, 2H, CH2CH2CO), 1.34 (s,
9H, t-Bu); 13C NMR (CDCl3) d 200.1, 157.0, 143.2, 134.9, 128.9,
(M+1)+; IR 3061 (N-H), 2946 (Ar-H) cmꢀ1
.
4.5. N,N-Di-(3-phenylpropyl)amine (91)36
From 3-phenylpropylamine (0.180 g, 1.33 mmol) and 3-phenyl-
propyl bromide (0.110 g, 0.552 mmol) was prepared 91 (0.077 g,
55%) as a liquid (Rf = 0.43): 1H NMR (CDCl3) d 9.09 (s, 2H, NH2+),
7.39–7.22 (m, 10H, 2Ph), 2.86 (t, 4H, 2NCH 2), 2.69 (t, 4H, 2PhCH2),
2.12–2.02 (m, 4H, 2PhCH2CH2); 13C NMR (CDCl3) d 141.0, 128.9,
128.7, 126.6, 47.8, 33.4, 28.6; MS 254 (M+1)+; IR 3066 (N-H),
128.5, 128.3, 127.3, 125.9; MS 456 (M+1)+; IR 1683 (C@O) cmꢀ1
Anal. Calcd for C31H38N2O2: C, 81.90; H, 8.42; N, 6.16. Found: C,
82.00; H, 8.53; N, 6.21.
;
2931 (Ar-H) cmꢀ1
.
4.10. 1-[4,4-Bis(4-flurophenyl)]butyl-4-(diphenylmethyl)piper-
azine (95)
4.6. General procedure for the preparation of 1-alkyl-4-(diphen-
ylmethyl)piperazine
From 1,10-(4-chlorobutylidene)-bis-(4-fluorobenzene) (0.500 g,
1.78 mmol) and 1-(diphenylmethyl)piperazine (1.00 g, 3.97 mmol)
was prepared 95 (0.755 g, 85%) (solvent for column: 68% hexanes +
30% ethyl acetate + 2% triethylamine, Rf = 0.50) as a white solid: mp
118–120 °C; 1H NMR (CDCl3) d 7.40–6.90 (m, 18H, 4Ph), 4.19 (s, 1H,
CHPh2), 3.85 (t, J = 7.8 Hz, 1H, CH(PhF)2), 2.39–2.30 (m, 10H,
N(CH2)2(CH2)2NCH2), 2.01–1.93 (q, J = 7.8 Hz, 2H, CH2CH(PhF)2),
1.44–1.35 (m, 2H, CH2CH2CH(Ph(F))2); 13C NMR (CDCl3) d 162.9,
159.7, 142.8, 140.6, 129.1, 129.0, 128.4, 127.9, 126.9, 115.4,
115.1, 76.3, 58.5, 53.5, 51.9, 49.7, 33.9, 25.3; MS 497 (M+1)+; IR
2930 (Ar-H) cmꢀ1; Anal. Calcd for C33H34F2N2: C, 79.81; H, 6.90;
N, 5.64. Found: C, 79.79; H, 6.93: N, 5.67.
To the solution of diphenylmethyl piperazine in DMSO (8 mL)
was added alkyl halide. The mixture was stirred at 70–80 °C for
10 h. The reaction mixture was transferred to a 250-mL separatory
funnel, ethyl acetate (100 mL) was added, and the mixture was
washed with 5% Na2CO3 (2 ꢁ 30 mL), water (2 ꢁ 50 mL) and brine
(50 mL). The organic layer was dried (Na2SO4) and evaporated to
dryness to give a crude product. The residue was purified by flash
silica gel chromatography to afford pure product. In this manner
were prepared the following compounds.
4.7. 1-[4-(3,4-Dimethylphenyl)-4-oxobutyl]-4-
(diphenylmethyl)piperazine (92)
4.11. 1-(Diphenylmethyl)-4-(3-phenylpropyl)piperazine (96)
From 4-chloro-1-(3,4-dimethylphenyl)-1-butanone (1.00 g,
4.75 mmol) and 1-diphenylmethylpiperazine (2.20 g, 8.73 mmol)
was prepared 92 (1.30 g, 70%) as a white solid (solvent for column:
90% hexanes + 8% ethyl acetate + 2% triethylamine, Rf = 0.65). The
amine was dissolved in ethyl acetate (10 mL), and into the ethyl
acetate solution was bubbled HCl gas. The solvent was removed,
and the residue was recrystallized from ethanol/benzene to give
a white solid (0.49 g): mp 185–187 °C; 1H NMR (DMSO-d6) d
10.35 (b, 1H, NH+), 7.75–7.20 (m, 13H, 3Ph), 4.48 (s, 1H, CHPh2),
3.49–3.45 (d, J = 12 Hz, 2H, H+N(CH)2(CH2)2N), 3.16–2.85 (m, 6H,
N(CH)2(CH2)2NCH2), 2.85–2.81 (d, J = 12 Hz, 2H, N(CH)2(CH2)2N),
2.44–2.36 (m, 2H, NHCH2CH2CH2), 2.29 (s, 6H, 2Me), 2.20–2.08
(m, 2H, NHCH2CH2CH2). 13C NMR (CDCl3) d 198.7, 143.7, 141.9,
137.5, 134.5, 130.4, 129.5, 129.3, 128.0, 126.1, 75.5, 56.8, 62.6,
48.7, 35.6, 31.3, 20.5, 20.2, 18.4; IR 1682 (C@O) cmꢀ1; MS 427
(M+1)+; Anal. Calcd for C29H34N2O⁄HCl⁄H2O: C, 72.40; H, 7.75; N,
5.82; Cl, 7.37. Found: C, 72.73; H, 7.73; N, 5.82; Cl, 7.42.
From 3-phenylpropyl bromide (0.120 g, 0.61 mmol) and
1-(diphenylmethyl)piperazine (0.360 g, 1.43 mmol) was prepared
96 (0.185 g, 81%) (solvent for column: 78% hexanes + 20% ethyl
acetate + 2% triethylamine, Rf = 0.45) as a white solid: mp 83–85 °C;
1H NMR (CDCl3) d 7.46–7.20 (m, 15H, 3Ph), 4.27 (s, 1H, CHPh2),
2.74–2.46 (m, 12H, N(CH2)2(CH2)2NCH2 & CH2Ph2), 1.90–1.80
(m, 2H, CH2CH2Ph); 13C NMR (CDCl3) d 143.4, 142.7, 129.0, 128.9,
128.8, 128.5, 127.4, 76.8, 58.6, 54.0, 52.5, 34.3, 29.2; IR 3031
(Ar-H) cm-1; MS 371 (M+1)+; Anal. Calcd for C26H30N2: C, 84.28;
H, 8.16; N, 7.56. Found: C, 84.19; H, 8.11; N, 7.69.
4.12. 1-(Diphenylmethyl)-4-(2-phenylethyl)piperazine (97)
From 3-phenylethyl bromide (0.249 g, 1.36 mmol) and
1-(diphenylmethyl)piperazine (0.343 g, 1.36 mmol) was prepared
97 (0.170 g, 35%) (solvent for column: 88% hexanes + 10% ethyl
acetate + 2% triethylamine, Rf = 0.25) as a white solid: mp
1
127–129 °C; H NMR (CDCl3) d 7.48–7.21 (m, 15H, 3Ph), 4.28 (s, 1H,
4.8. 1-[2-(N-Benzoylamino)ethyl]-4-(diphenylmethyl)-1-
piperazine (93)
CHPh2), 2.85-2.53 (m, 12H, N(CH2)2(CH2)2NCH2 & CH2Ph2); 13C
NMR (CDCl3) d 143.2, 140.8, 129.1, 128.9, 128.8, 128.3, 127.3,
126.5, 76.6, 60.9, 53.9, 52.2, 34.00; IR 3030 (Ar-H) cmꢀ1; MS 357
(M+1)+; Anal. Calcd for C25H28N2: C, 84.23; H, 7.92; N, 7.86. Found:
C, 84.03; H, 8.04; N, 7.78.
From N-2-chloroethylbenzamide (1.00 g, 5.45 mmol) and
1-diphenylmethylpiperazine (2.39 g, 10.9 mmol) was prepared 93

104
C. Zha et al. / Bioorg. Med. Chem. 22 (2014) 95–104
4.13. Sodium channel binding assay
Supplementary data
The details for the sodium channel binding assay were reported
in a previous paper.25 Briefly, synaptoneurosomes were prepared
from rat cerebral cortex as follows: Cerebral cortex slices (gray
matter) were obtained by removing the white matter and other
subcortical structure. The slices (about 1 g) were then homoge-
nized in 2 mL of incubation buffer containing 130 mM choline
chloride, 50 mM HEPES, 5.5 mM glucose, 0.8 mM MgSO4, and
5.4 mM KCl (adjusted to pH = 7.4 with Tris base). The tissue was
homogenized with 10 full strokes of a glass–glass homogenizer.
The tissue preparation was then transferred to a centrifuge tube.
The homogenizer was rinsed with 3 mL of the incubation buffer,
which was combined with the preparation. The preparation was
then centrifuged at 1000g for 15 min at 4 °C. The pellet was re-sus-
pended in a total volume of 20 mL of the incubation buffer and
transferred to a 50-mL homogenizer. Three full strokes were used
to homogenize the tissue. The resulting suspension was then fil-
tered, first with three layers of nylon mesh by gravity, and then
with Whatman 4 filter paper by house vacuum. The filtrate was
then centrifuged at 1000g for 30 min at 4 °C. The obtained pellet
was re-suspended with 5 mL isotonic buffer, and more isotonic
buffer was used as needed to adjust the suspension to have an
absorbance of 1 at 280 nm. The isotonic suspension was stored in
a freezer at ꢀ70 °C. Prior to the binding assay, the tissue was
thawed and the suspension was centrifuged, and the pellet was
re-suspended in the HEPES incubation buffer (same volume as
the isotonic buffer used to store the tissue).
Supplementary data associated with this article can be found, in
the online version, at http://dx.doi.org/10.1016/j.bmc.2013.11.049.
References and notes
1. Catterall, W. A.; Wood, J. N.; Baker, M. Curr. Opin. Pharmacol. 2001, 1, 17.
2. Catterall, W. A. J. Bioenerg. Biomembr. 1996, 28, 219.
3. Hille, B. Ionic Channels of Excitable Membranes; Sinauer Associates: Sunderland,
MA, 1991.
4. Catteral, W. A. Trends Neurosci. 1986, 9, 7.
5. Yu, F. H.; Catterall, W. A. Genome Biol. 2003, 4, 207.
6. Taylor, C. P.; Narasimhan, L. S. Adv. Pharmacol. 1997, 39, 47.
7. Yao, C.; Williams, A. J.; Lu, X. C.; Price, R. A.; Cunningham, B. S.; Berti, R.;
Tortella, F. C.; Dave, J. R. Neurotoxicity Res. 2003, 5, 245.
8. Brown, A. M.; Schwindt, P. C.; Crill, W. E. J. Neurophysiol. 1994, 71, 2562.
9. Ragsdale, D. S.; Avoli, M. Brain Res. Rev. 1998, 26, 16.
10. Campell, T. J.; Williams, K. M. Br. J. Clin. Pharmacol. 1998, 46, 307.
11. Sheldon, R. S.; Duff, H. J.; Thakore, E.; Hill, R. J. J. Pharmacol. Exp. Ther. 1994, 268,
187.
12. Urenjak, J.; Obrenovitch, T. P. Pharmacol. Rev. 1996, 48, 21.
13. Catterall, W. A. Mol. Pharmacol. 1981, 20, 356.
14. di Vadi, P. P.; Hamann, W. Anesthesia 1998, 53, 808.
15. Deffois, A.; Fage, D.; Carter, C. Neurosci. Lett. 1996, 220, 117.
16. Waxman, S. G.; Cummins, T. R.; Dib-Hall, S.; Fjell, J.; Black, J. A. Muscle Nerve
1999, 22, 1177.
17. Courtney, K. R. J. Pharmacol. Exp. Ther. 1980, 213, 114.
18. Hille, B. J. Gen. Physiol. 1977, 69, 497.
19. Fraser, S. P.; Diss, J. K.; Chioni, A. M.; Mycielska, M.; Pan, H.; Yamaci, R. F., et al
Clin. Cancer Res. 2005, 11, 5381 [PubMed: 16061851].
20. Onganer, P. U.; Djamgoz, M. B. J. Membr. Biol. 2005, 204, 67.
21. Yang, M.; Kozminski, D. J.; Modak, R.; Calhoun, J. D.; Isom, L. L., et al Breast
Cancer Res. Treat. 2012, 134, 603.
22. Brackenbury, W. J. Channels (Austin) 2012, 6, 352.
23. Fraser, S. P.; Salvador, V.; Manning, E. A.; Mizal, J.; Altum, S.; Raza, M., et al J.
Cell. Physiol. 2003, 195, 479.
24. Abdul, M.; Hoosein, N. Anticancer Res. 2001, 21, 2045.
25. Brown, M. L.; Zha, C. C.; Van Dyke, C. C.; Brown, G. B.; Brouillette, W. J. J. Med.
Chem. 1999, 42, 1537.
26. Zha, C.; Brown, G. B.; Brouillette, W. J. J. Med. Chem. 2004, 47, 6519.
27. Brown, M. L.; Eidam, H. A.; Paige, M.; Jones, P. J.; Patel, M. K. Bioorg. Med. Chem.
2009, 17, 7056.
In the sodium channel binding assay, this suspension of syna-
ptoneurosomes (ꢂ1 mg of protein) from rat cerebral cortex was
incubated for 60 min at 25 °C with the test compound (seven dif-
ferent concentrations spanning the IC50) in a total volume of
320 l lg/mL of scorpion ve-
L containing 10 nM [3H]BTX-B and 50
nom. Incubations were terminated by dilution with ice-cold buffer
and filtration through a Whatman GF/C filter paper under vacuum
condition, and the filters were washed four times with ice-cold
buffer. Filters were counted in a Beckman scintillation counter.
Specific binding was determined by subtracting the nonspecific
28. Wang, Y.; Jones, P. J.; Batts, T. W.; Landry, V.; Patel, M. K.; Brown, M. L. Bioorg.
Med. Chem. 2009, 17, 7027.
29. Catterall, W. A. Trends Pharmacol. Sci. 1987, 8, 57.
30. Ragsdale, D. S.; McPhee, J. C.; Scheuer, T.; Catterall, W. A. Science 1994, 265,
1724.
binding, which was measured in the presence of 300 lM veratri-
31. Li, H.-L.; Galue, A.; Meadows, L.; Ragsdale, D. S. Mol. Pharmacol. 1999, 55, 134.
32. Ragsdale, D. S.; McPhee, J. C.; Scheuer, T.; Catterall, W. A. Proc. Natl. Acad. Sci.
U.S.A. 1998, 93, 9270.
33. Qu, Y.; Rogers, J.; Tanade, T.; Scheuer, T.; Catterall, W. A. Proc. Natl. Acad. Sci.
U.S.A. 1995, 92, 11839.
34. Sheldon, R. S.; Hill, R. J.; Taouis, M.; Wilson, L. M. Mol. Pharmacol. 1991, 472,
373.
35. McNeal, E. T.; Lawandowski, G. A.; Daly, J. W.; Creveling, C. R. J. Med. Chem.
1985, 28, 381.
dine, from the total binding of [3H]BTX-B. All experiments were
performed in triplicate and included a control tube containing
40 lM DPH. The allowed IC50 value of DPH was 40 4 (10%) lM,
and an assay would be repeated if the IC50 value was beyond this
range, which was uncommon. The IC50 values of test compounds
were determined from a Probit analysis of the dose–response curve
and excluded doses producing less than 10% or greater than 90%
inhibition.
36. Glennon, R. A.; Alordeppey, S. Y.; Ismaiel, A. M.; El-Ashmawy, M. B.; Fisher, J. B.;
Howie, K. B. J. Med. Chem. 1994, 37, 1214.