The design, preparation and SAR of novel small molecule sodium (Na +) channel blockers

Article abstract of DOI:10.1016/j.bmcl.2004.02.078

A parallel strategy incorporating predictive modeling of both sodium site 2 blocking activity and cytochrome P450 CYP2D6 enzyme activity as well as experimental data from ADME profiling (eADME) has been applied to the design of new small molecule sodium channel blockers. New structural motifs were identified, which combined sodium channel activity with decreased ADME liabilities. Compounds 10h (site 2, IC₅₀=531nM) and 7j (site 2, IC₅₀=149nM) were identified from two structural classes as sodium channel blockers with favorable in vitro eADME profiles.

Full text of DOI:10.1016/j.bmcl.2004.02.078

Bioorganic & Medicinal Chemistry Letters 14 (2004) 2025–2030
The design, preparation and SAR of novel small molecule
sodium (Na+) channel blockers
Mark A. Ashwell,* Jean-Marc Lapierre, Alan Kaplan, Jenny Li, Christopher Marr
and Jin Yuan
ArQule Inc., 19 Presidential Way, Woburn, MA 01801, USA
Received 28January 2004; revised 13 February 2004; accepted 13 February 2004
Abstract—A parallel strategy incorporating predictive modeling of both sodium site 2 blocking activity and cytochrome P450
CYP2D6 enzyme activity as well as experimental data from ADME profiling (eADME) has been applied to the design of new small
molecule sodium channel blockers. New structural motifs were identified, which combined sodium channel activity with decreased
ADME liabilities. Compounds 10h (site 2, IC50¼531 nM) and 7j (site 2, IC50¼149 nM) were identified from two structural classes
as sodium channel blockers with favorable in vitro eADME profiles.
Ó 2004 Elsevier Ltd. All rights reserved.
As a result of in vitro biological data generated while
profiling a targeted library of small molecules synthe-
sized for their ability to affect a range of voltage-gated
ion channels1 we identified a number of sodium channel
blockers. The in vitro screening showed them to be sub-
micromolar inhibitors of the binding of [3H]batracho-
toxin to the neurotoxin site 2 of the sodium (Naþ)
channel2 (e.g., 1, Fig. 1).
Currently there is considerable interest in sodium
channels as therapeutically relevant biological targets
with potential clinical utility3 so compounds from these
libraries were profiled further. Compounds 1a and 1b
were therefore selected as representative compounds and
functional electrophysiology studies in whole-cell patch-
clamp experiments conducted in cultured rat dorsal root
ganglion (DRG).4 Compound 1a was identified as a
potent nonselective blocker of both tetrodotoxin TTX-
resistant and TTX-sensitive current. 1b was found to be
considerably weaker (Fig. 1).
R1
However, further profiling of analogs with general
structure 1 across a range of in vitro ADME assays
revealed an unwanted inhibitory class effect against the
cytochrome P450 2D6 (CYP2D6). CYP2D6 is a minor
component (approx. 2%) of total cytochrome P450
content in human liver, yet is responsible for the
metabolism of a significant proportion (approx. 20%) of
drugs currently in clinical use.5 An additional compli-
cation with CYP2D6-mediated metabolism is due to
genetic polymorphism in humans.6
N
N
OH
1
OR2
In
CYP Peak Peak
2D6 TTXr TTXs
(% (% inhib (% inhib
Na
site 2
IC50
silico
CYP
2D6 Ki
(µM)
Compd R1 R2
inhib.@ @ 10
@ 10
(nM)
2µM)
µM)
µM)
1a
1b
CF3 Ph
893
0.6
87
61
57
We therefore considered CYP2D6 inhibition to be an
undesirable property for these potentially useful, sodium
channel blockers, and thus initiated a hit expansion
project to increase the structural diversity beyond that
of the original scaffold of 1. We sought to identify
new molecules, which maintained or improved in vitro
sodium blocking activity for site 2 and were less inhib-
itory to CYP2D6.
Cl allyl 401
0.1
87
20
NT
Figure 1. Predicted and measured properties of initial hits.
* Corresponding author. Tel.: +1-781-9940598; fax: +1-781-9940690;
e-mail: mashwell@arqule.com
0960-894X/$ - see front matter Ó 2004 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2004.02.078

2026
M. A. Ashwell et al. /Bioorg. Med. Chem. Lett. 14 (2004) 2025–2030
A virtual library design protocol was developed incor-
porating in silico pharmacophore and predictive ADME
screening. New designs were pre-screened in silico for
their ability to fit an in-house sodium channel blocker
pharmacophore model developed for site 2 binding and
for their inhibition of CYP2D6.7 Virtual library designs
were thus conceived, screened in silico to assist our
scaffold prioritization as well as to guide reagent selec-
tion, prior to preparation in a parallel chemistry para-
digm.
Further library synthesis is shown in Scheme 3. Com-
mercially available 2-[4-[(tert-butyl)oxycarbonyl]pipera-
zinyl]-2-phenylethylamines, 8, were coupled with a
variety of acid chlorides to give 9. Boc deprotection and
epoxide opening gave 10. In a similar manner, inter-
mediates 11 were obtained from the reaction of 8 with
sulfonyl chlorides. Deprotection and epoxide opening
provided 12.
Our screening approach determined the in vitro sodium
site 2 activity using a ligand-displacement assay mea-
suring the ability of the compound to inhibit the binding
of [3H]batrachotoxin to rat cerebral cortex membranes.
Single point screening at a fixed concentration was fol-
lowed by IC50 determination for active compounds. A
suite of ADME assays allowed us to assess additional
in vitro parameters such as solubility, CYP inhibition,
metabolic stability, etc.
Since the early SAR indicated that the R2 substituent of
1 played a minor role in affecting CYP2D6 inhibition we
focused on the removal of the benzhydryl chiral center
of 1a. Moving one of the piperazinyl nitrogens out of the
ring and subsequent amide functionalization provided
intermediates 3. Compounds based on this scaffold were
readily accessible,8 as shown in Scheme 1.
tert-Butyl-4-anilinotetrahydro-1(2H)-pyridine carboxyl-
ates (commercially available or prepared in one-step
from tert-butyl 4-oxo-1-piperidinecarboxylate)9 were
coupled with a variety of acid chlorides, providing
intermediates 3. Removal of the Boc group and reaction
of the resulting amine with commercial 2-(aryloxy-
methyl)oxiranes gave 4.
Initially we sought to determine the role of the diphe-
nylmethyl piperazine core of 1a in determining site 2
binding as well as its influence on CYP2D6 inhibition.
Significantly, the first array (4, Table 1) demonstrated
that replacement of one of the aromatic rings with an
amide group provided molecules with site 2 blocking
activity. Additionally, some compounds in this array
were somewhat more active (about 2- to 4-fold) than 1a
at blocking sodium site 2 (e.g., 4b and 4e).
Compounds 7 were prepared according to Scheme 2.
Commercially available 2-[4-[(tert-butyl)oxycarbonyl]
piperazinyl]-2-phenylacetic acids, 5, were reacted with
amines (HBTU coupling)10 to provide intermediates 6.
The Boc protecting group was removed and the free
amine reacted with 2-(aryloxymethyl)oxiranes as shown
to provide compounds of general structure 7.
From an analysis of the predicted in silico CYP2D6
inhibitions we did not expect a significant reduction
in inhibition from members of this array and as shown
in Table 1 the measured in vitro single point inhibi-
tion values at the screening concentration of 2 lM
O
R2
R1
O
N
N
R1
R1
R2
NH
N
N
b, c
a
OH
O
N
BOC
4
BOC
2
3
R3
Scheme 1. Reagents and conditions: (a) ArCOCl, HunigÕs base, CH2Cl2; (b) 4 M HCl in dioxane; (c) 2-(aryloxymethyl)oxirane, MeOH, 65 °C.
O
R1
R2
O
N
H
R1
R2
b, c
R1
CO2H
N
N
N
H
N
N
N
a
7
OH
O
N
BOC
5
BOC
6
R3
Scheme 2. Reagents and conditions: (a) amine, HBTU, HunigÕs base, DMF; (b) 4 M HCl in dioxane; (c) 2-(aryloxymethyl)oxirane, MeOH, 65 °C.

M. A. Ashwell et al. /Bioorg. Med. Chem. Lett. 14 (2004) 2025–2030
2027
O
R1
O
R2
R1
N
H
R2
N
N
H
N
c, d
a
N
N
OH
O
BOC
9
10
R1
NH2
N
R3
O O
N
BOC
O O
S
H
R1
R2
R1
R2
S
N
H
N
N
c, d
N
N
b
8
N
BOC
OH
O
12
11
R3
Scheme 3. Reagents and conditions: (a) acid chloride, HunigÕs base, CH2Cl2; (b) sulfonyl chloride, pyridine, 60 °C; (c) 4 M HCl in dioxane;
(d) 2-(aryloxymethyl)oxirane, MeOH, 65 °C.
Table 1. SAR of functionalized N-phenyl-N-(piperidin-4-yl)benzamides
R1
O
N
R2
N
OH
4
O
R3
Compound
R1
R2
R3
Na site 2 IC50
(nM)
In silico CYP
2D6 Ki (lM)
CYP 2D6 inhi- HLM stability
bition (% inhi- (% remaining
bition @ 2 lM) @ t ¼ 30 min)
CYP 3A4
(% inhibition
@ 2 lM)
4a
4b
4c
4d
4e
4f
H
H
H
F
H
OMe
OMe
OMe
OMe
OMe
OMe
261
183
2281
194
191
214
2
34
73
24
33
71
87
OMe
Cl
0.7
71
1
28 90
H
55
6834 8
77
41
79
F
OMe
Cl
3
8
3897
F
2
follow a similar trend to the predicted Ki values.
Important SAR trends were indicated, such as a ten-
dency for lower CYP2D6 inhibition when R2¼H (4a
and 4d).
assay (HLM) we sought to identify additional alterna-
tive scaffolds without these liabilities.
One such design, 7, in which an amide group has been
introduced to replace one of the phenyl rings of the
diphenylmethyl core of 1, was expected to have reduced
CYP2D6 inhibition based on our predictive model. In
addition a good pharmacophore fit encouraged us to
synthesize several building blocks 6 and to use these to
furnish a library of compounds 7. The sodium site 2
binding assay showed that many examples of 7 had
good blocking profiles (Table 2). The most potent
Having made the important discovery that the diphen-
ylmethyl piperazine scaffold was not a pre-requisite for
site 2 activity we pursued virtual designs based on
the N-alkylated N-phenylpiperidin-4-amine core of 2.
However, since many of the compounds in Table 1 were
additionally also potent CYP3A4 inhibitors and had
low stability in the isolated human liver microsome

2028
M. A. Ashwell et al. /Bioorg. Med. Chem. Lett. 14 (2004) 2025–2030
Table 2. SAR of N-benzyl-2-phenyl-2-(piperazin-1-yl)acetamides
R1
O
N
H
N
N
R2
7
OH
O
R3
Compound
R1
R2
R3
Na site 2 IC50
(nM)
In silico CYP
2D6 Ki (lM)
CYP 2D6 (%
inhibition @
2 lM)
HLM stability CYP 3A4
(% remaining
(% inhibition
@ t ¼ 30 min) @ 2 lM)
7a
7b
7c
7d
7e
7f
Me
Me
Me
CF3
CF3
CF3
CF3
OMe
F
H
2,6-Di-Me
4-OMe
325
579
351
296
429
355
359
571
254
149
343
214
2
4
3
1
2
2
4
5
5
2
3
7
26
10
14
24
13
860
0
41
10
44
65
74
53
H
OMe
H
2,6-Di-Me
2,6-Di-Me
4-OMe
60
62
H
2868
62
OMe
OMe
OMe
H
2,6-Di-Me
4-OMe
7g
7h
7i
43
64
88
74
2,6-Di-Me
4-OMe
9
0
849
57
7j
F
H
2,6-Di-Me
2,6-Di-Me
4-OMe
20
11
10
38
70
81
7k
7l
F
OMe
OMe
52
10
F
compounds in this series were those that had electron-
withdrawing groups at R1 especially fluorine (7i–l).
ment with our expectations. Indeed many examples in
this array were potent site 2 sodium channel blockers
(below 500 nM) with little inhibition of CYP2D6 (below
25%).
The predicted in silico CYP2D6 inhibition values for 7
showed a trend towards reduced inhibition across the
series compared to 1. Experimentally CYP2D6 single
point percentage inhibitions were below 25% in agree-
However, in general, members of this series were also
CYP3A4 inhibitors with low to medium HLM stability
Table 3. SAR of N-(2-phenyl-2-(piperazin-1-yl)ethyl)benzamides
O
R2
N
H
N
N
OH
O
10
R3
Compound
R2
R3
Na site 2 IC50
(nM)
In silico CYP
2D6 Ki (lM)
CYP 2D6 (%
inhibition @
2 lM)
HLM stability
(% remaining
@ t ¼ 30 min)
CYP 3A4
(% inhibition
@ 2 lM)
10a
10b
10c
10d
10e
10f
10g
10h
10i
H
OMe
OMe
Cl
745
871
5384
253
2081
721
363
531
880
4
22
12
74
12
78 69
9
879
4-OMe
4-OMe
4-CF3
4-CF3
2-OMe
2-OMe
4-F
8
23
95
94
47
3
96
OMe
Cl
20
40
89
7
OMe
Cl
93
39
62
6
9
61
22
90
H
OMe
3
33
15
4-F
5

M. A. Ashwell et al. /Bioorg. Med. Chem. Lett. 14 (2004) 2025–2030
Table 4. SAR of N-(2-phenyl-2-(piperazin-1-yl)ethyl)phenylsulfonamides
R1
2029
O O
S
R2
N
H
N
N
OH
O
12
R3
Compound
R1
R2
R3
Na site 2 IC50 In silico CYP 2D6
(nM)
CYP 2D6 HLM stability
(% inhibition (% remaining @
3A4 inhibi-
tion (%)
Ki (lM)
@ 2 lM)
t ¼ 30 min)
53
12a
12b
12c
CF3
H
3-Me
2,5-Di-OMe
2-Cl
Br
Br
F
95a
522
76a
10
30
10
27
25
56
76
100
77
4892
CF3
a % Inhibition at 1 lM.
although SAR trends showed that the introduction of
methoxy groups at R2 or R3 was detrimental to the
overall profile (e.g., compare 7l to 7k and 7c to 7a)
suggesting strategies to follow beyond lead expansion.
From these series 7j and 10h represent new structural
motifs, which together with their associated SAR may
provide sodium channel site 2 blockers of therapeutic
value.
The most potent compound in this set, 7j was a potent
site 2 binder (IC50¼149 nM) with no CYP2D6 inhibi-
tion, good HLM stability, and low CYP3A4 inhibition.
Acknowledgements
In another of our templates we reversed the amide of 7,
providing structure 10. The predicted CYP2D6 values
indicated that this would result in a series with a
decrease in CYP2D6 inhibition. This proved to be the
case when these compounds were synthesized and
screened in vitro. Individual compounds were identified,
which combined good blocking activity at site 2 with
low CYP2D6 inhibition (e.g., 10e and 10d in Table 3).
Again, as with scaffold 7, additional eADME screening
revealed important SAR trends such as the detrimental
effect of a 4-OMe group as R3, which resulted in reduced
HLM stability and increased CYP3A4 inhibition
(compare 10i and 10h).
We thank Dr. Toan Nguyen for his molecule modeling
assistance and Dr. Steve Gallion for his project leader-
ship. In addition we also wish to acknowledge the
many valuable contributions made to the voltage-gated
ion channel project at ArQule, without which work
described in this paper would not have been possible.
References and notes
1. (a) Gallion, S.; Hardy, L.; Sheldon, A. Curr. Drug Discov.
2002, 1, 25–27; (b) Sections of this work have been
presented; http://www.chimolecularmed.com/postconpdf/
mmmtrack3/mark_ashwell.pdf
2. Brown, G. B. J. Neurosci. 1986, 6, 2064–2070.
3. Anger, D.; Madge, D. J.; Mulla, M.; Riddal, D. J. Med.
Chem. 2001, 44, 115–137.
4. Compounds were screened against the TTXr current
recorded from dissociated rat DRG cells. Each compound
was screened at a concentration of 10 lM and applied to
each cell for 4 min. In addition compounds were screened
against the TTXs sodium current. For more details see
Elliott, A. A.; Elliott, J. R. J. Physiol. 1993, 463, 39–56.
5. Tredger, J. M.; Stoll, S. Hosp. Pharmacist 2002, 9, 167–
173.
6. Bertilsson, L. Clin. Pharmacokinet. 1995, 29, 192–209.
7. The predictive competitive inhibition model for CYP2D6
was constructed from data using probe substrates and an
in house compound set. The output for this model is given
as a binding constant or Ki (expressed in lM). The
technical details of these models will be published
elsewhere.
It was expected from the in silico CYP2D6 model that
extending the design to the sulfonamides 12 would also
provide lower levels of CYP2D6 inhibition. This proved
to be the case. Site 2 blockers where prepared which had
reduced inhibition of CYP2D6 (12a and 12b). However,
12a and 12b were found to be significant CYP3A4
inhibitors and so these were not pursued further (Table
4).
In conclusion we have utilized a strategy incorporating
parallel in silico modeling of sodium site 2 blocking
activity and CYP2D6 inhibition to identify new scaf-
folds. Using this approach together with in vitro bio-
logical screening we have identified a number of series of
potent sodium channel blockers with a range of ADME
profiles. Two series (7 and 10) were of particular interest
for their overall favorable activity and ADME profiles.

2030
M. A. Ashwell et al. /Bioorg. Med. Chem. Lett. 14 (2004) 2025–2030
8. All final compounds were prepared in a parallel chemistry
9. (a) Apodaca, R.; Xiao, W. Org. Lett. 2001, 3, 1745–1748;
(b) Mattson, R.; Pham, K. M.; Leuck, D. J.; Cowen, K. A.
J. Org. Chem. 1990, 55, 2552–2554.
10. Knorr, R.; Trzeciak, A.; Bannwarth, W.; Gillessen, D.
Tetrahedron Lett. 1989, 30, 1927–1930.
fashion, purified by RP-HPLC and characterized by
HPLC–MS, yields ranged from 10–80% based on recovered
mass. In those cases were stereoisomers were generated
no attempt was made to separate individual components.