Discovery of novel α7 nicotinic acetylcholine receptor ligands via pharmacophoric and docking studies of benzylidene anabaseine analogs

Article abstract of DOI:10.1016/j.bmcl.2011.11.090

Based on pharmacophore elucidation and docking studies on interactions of benzylidene anabaseine analogs with AChBPs and α7 nAChR, novel spirodiazepine and spiroimidazoline quinuclidine series have been designed. Binding studies revealed that some of hydr

Full text of DOI:10.1016/j.bmcl.2011.11.090

Bioorganic & Medicinal Chemistry Letters 22 (2012) 1179–1186
Contents lists available at SciVerse ScienceDirect
Bioorganic & Medicinal Chemistry Letters
journal homepage: www.elsevier.com/locate/bmcl
Discovery of novel a7 nicotinic acetylcholine receptor ligands via
pharmacophoric and docking studies of benzylidene anabaseine analogs
⇑
David C. Kombo , Anatoly A. Mazurov, Joseph Chewning, Philip S. Hammond, Kartik Tallapragada,
Terry A. Hauser, Jason Speake, Daniel Yohannes, William S. Caldwell
Targacept Inc., 200 East First Street, Suite 300, Winston-Salem, NC 27101 4165, USA
a r t i c l e i n f o
a b s t r a c t
Article history:
Based on pharmacophore elucidation and docking studies on interactions of benzylidene anabaseine ana-
Received 3 September 2011
Revised 18 November 2011
Accepted 21 November 2011
Available online 30 November 2011
logs with AChBPs and
been designed. Binding studies revealed that some of hydrogen-bond donor containing compounds exhi-
bit improved affinity and selectivity for the 7 nAChR subtype in comparison with most potent metab-
olite of GTS-21, 3-(4-hydroxy-2-methoxybenzylidene)-anabaseine. Hydrophobicity and rigidity of the
ligand also contribute into its binding affinity. We also describe alternative pharmacophoric features
equidistant from the carbonyl oxygen atom of the conserved Trp-148 of the principal face, which may
a7 nAChR, novel spirodiazepine and spiroimidazoline quinuclidine series have
a
Keywords:
AChBPs
be exploited to further design diverse focused libraries targeting the
a7 nAChR.
Nicotinic acetylcholine receptor
Pharmacophore
a7
Ó 2011 Elsevier Ltd. All rights reserved.
Docking
Benzylidene anabaseine analogs
Spirodiazepine and spiroimidazoline
quinuclidine series
Neuronal nicotinic acetylcholine receptors (nAChR) are penta-
meric ligand-gated ion channels of broad distribution and structural
heterogeneity. Such diversity in both function and distribution
indicates involvement in a variety of neuronal processes and has
generated great interest in these acetylcholine receptors as targets
for therapeutic intervention in a number of pathological conditions
Subsequent to activation, the
a7 nAChR rearranges into desensi-
tized state resulting in loss of biological response.2
The a7 subunit is expressed at high levels in hippocampus and
cerebral cortex, brain regions involved in learning and memory.3
Accumulating physiological,4 pharmacological,5 and human genet-
ic6 experimental data suggest involvement of
a7 nAChR in the cog-
and diseases.1 The
a7 subtype (a homopentamer consisting of five
nitive deficits associated with a number of neuropsychiatric and
neurodegenerative diseases including schizophrenia and Alzhei-
mer’s disease. The growing body of evidence has highlighted the
a7 subunits) has been extensively studied in recent years. The
human
a7 subunit is approximately 50 kD in size and composed
of 502 amino acids, including a 22 amino acid signal peptide. The
protein may be divided into an extracellular N-terminal and a trans-
membrane domains. The N-terminal domain, composed of approx-
imately 200 amino acids forming ten b strands, encompasses the
ligand-binding domain. The transmembrane domain comprises
clinical need for the development of a7 nAChR agonists, which
may offer new approaches to the treatment of impaired cognitive
functions.
Acetylcholine-binding protein (AChBP) folds into a homopenta-
meric structure similar to the extracellular ligand-binding domain
four
channel. The
a
helices crossing the lipid bi-layer and constitutes the ion
of a7 nAChR. Several crystal structures of AChBPs and their vari-
a
7 nAChR is highly permeable to Ca2+ ions, and is
ants, in complex with nicotinic ligands have been solved.7–12 These
structural biology studies have shown that the molecular recogni-
tion process between ligands and nAChRs primarily occurs through
therefore characterized as a calcium channel. Agonist binding leads
to conformational changes via a quaternary symmetrical twist,
opening the hydrophobic gate and allowing passage of ions.
cation–p interactions, hydrogen-bonding interactions, p–p and
hydrophobic interactions as well.13–15 Ulens et al.16 have recently
shown that an AChBP crystal structure could be used as a surrogate
Abbreviations: ACHBP, acetylcholine-binding protein; DMXBA, 3-(2,4-dim-
ethoxybenzilidene)-anabaseine; HBD, hydrogen-bond donor; nAChR, nicotinic
acetylcholine receptor; Pdb, protein databank; rmsd, root-mean-squared-deviation;
ROC, receiver operating characteristic curve.
for docking, in attempt to identify novel
this, Taylor and co-workers have also used AChBP crystal struc-
tures to rationalize the binding mode of agonists to the
nAChR.12 Benzylidene anabaseine analogs exhibit functional selec-
tivity towards
nAChR17,18 with the 4-hydroxy-containing
a7 nAChR ligands. Prior to
a
7
⇑
Corresponding author. Tel.: +1 336 480 2127; fax: +1 336 480 2107.
E-mail address: david.kombo@targacept.com (D.C. Kombo).
a7
0960-894X/$ - see front matter Ó 2011 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2011.11.090

1180
David C. Kombo et al. / Bioorg. Med. Chem. Lett. 22 (2012) 1179–1186
derivatives being among the most potent of this congeneric series.
Crystallographic studies of the 3-(4-hydroxy-2-methoxybenzyli-
dene)-anabaseine (1a) a metabolite of 3-(2,4-dimethoxybenzylid-
ene)-anabaseine (GTS-21, 1b), in complexation with Aplysia
AChBP have shown that its OH group forms a hydrogen-bond to
a polar side chain triad made of Asp-164, Ser-166 and Ser-167,
which are all located in loop F.12 Using docking, and comparative
modeling, we found that a similar hydrogen-bond donor (HBD)
contributes to the interaction of these ligands with Bulinus AChBP
the cationic center by a distance approximately equal to the
average distance constraint required for binding to rat 7, Ac, Bt,
or Ls (about 5.6 Å), could theoretically satisfy the docking-derived
receptor-based pharmacophoric distance (approximately 12.9 Å)
required for binding to the a7 nAChR, as shown in Figure 2a. One
should note that in addition to the mere ligand HBD-cation
distance, the length of the N–H bond (1.01 Å), O–H bond (0.95 Å)
, and twice the optimum hydrogen-bond distance (ꢀ2.5 Å), should
be taken into account. The binding affinity of the designed library
a
and the
a
7 nAChR, whereas in the case of Lymnaea, the ligand OH
of spirodiazepine and spiroimidazoline quinuclidines to rat a7
group acts as a hydrogen bond acceptor instead, and interacts with
Gln-73 of the complementary face.
nAChR was predicted using pharmacophore models and molecular
docking as described herein.20 Scheme 1 shows the synthetic
scheme of the compounds made.
Pharmacophore elucidation studies were carried out to explore
quantitative 3D QSAR models to predict AChBP Kd and
a
7 nAChR Ki,
Compounds containing hydroxyl group (2a–g, 3a) or amino
group (3b) as a hydrogen bond donor have been synthesized by
coupling 3-amino-3-(aminomethyl)quinuclidine (4) with series
of aroylacetates,21 methyl 2-(cyanomethoxy)benzoate or (2-cya-
nophenoxy)acetonitrile. Spirodiazepine 2h without hydroxyl
moiety has been obtained by condensation of diamine 1 with
[3-(dimethylamino)-2-phenylprop-2-enylidene]-dimethylamm-
onium hexafluorophosphate. Heating in a microwave of diamine 1
with diethyl 2-phenylmalonate or methyl benzimidate provided
1,4-diazepindione 2i and phenylimidazoline 3c. Hydroxypheny-
limidazoline 3d has been carried out by cyclization of 3-amino-
3-(aminomethyl)quinuclidine (1) with 2-(2-hydroxyphenyl)-1,
3-benzoxazin-4-one.22
as described herein.19 Summary statistics for the best model de-
rived in each case are shown in Table 1 and indicate that based
on Pearson’s correlation coefficient and root-mean-square-devia-
tion, derived models show reasonable ability to predict binding.
The results obtained for the best 10 pharmacophore hypotheses
indicate that the most frequent chemical features observed were
positive charge, hydrogen-bond donor, hydrogen-bond acceptor,
ring aromatic and hydrophobic. Fischer randomization tests for
each pharmacophore model were carried out to demonstrate that
correlations observed between the predicted biological activity
and ligand molecular descriptors did not result from chance corre-
lation. In each case, Kd or Ki values were scrambled nine times; ran-
dom models were generated and compared to their non-random
counterpart. The significance thus derived for the best models
The experimentally observed binding data23 obtained for the
designed compounds are shown in Table 2. Results show that none
were 90%, for Bt, Ls, and
tained when screening a diverse library for rat
2.70, 1.75, 1.47, and 1.44 for the best model derived from Ac, rat
7, Bt, and Ls, respectively. The average distance constraint be-
tween the minimum value of 1.97 Å and maximum value of
9.30 Å obtained is about 5.6 Å.
a
7, and 70% for Ac. The enrichment ob-
of the designed compounds interact with the
more, none of the designed compounds exhibited binding neither
to the ganglion-type 3b4 nAChR nor to muscle-type nicotinic
receptors. HBD-containing compounds 3a, 3b and 3d, interact with
7 nAChR with Ki < 300 nM likewise, compounds 2a through 2g,
which might tautomerize into lactim,24 interact with the
7 sub-
a4b2 nAChR. Further-
a7 binding is
a
a
a
a
Figure 1 illustrates mapping of compounds 1a and 1b onto the
type. All active compounds exhibit a HBD-cation distance within
the longest range of 5.47–7.65 Å. Substitution within the aromatic
ring of compounds 2a–g slightly affects interaction with the recep-
tor, while potential conjugation of the aromatic ring and azome-
best pharmacophore models for each AChBP and the
a7 nAChR.
We find that the ranked order of fitness to the pharmacophore
hypotheses and predicted Kd values for both compounds, for all
three AChBP models, is consistent with the trend in experimen-
tally-determined Kd values. Likewise, the ranked order of fitness
to the pharmacophore hypothesis and predicted Ki values for both
thine moiety provide
which might also tautomerize into lactim does not interact with
the rat 7 receptor. Although this compound satisfies the distance
p–p interactions. However, compound 2i,
a
compounds, for the rat
ranked order of experimentally-determined Ki values. Taken to-
gether, these results suggest that binding of anabaseine analogs
to these four proteins is mediated by cation–p, p–p, hydrophobic
and hydrogen-bonding interactions, which is in agreement with
the well-known nicotinic ligands pharmacophore,1 as well as dock-
ing studies (Figs. 2–5), and co-crystallographic data.
We have designed novel HBD-containing spirodiazepine and
spiroimidazoline series (Scheme 1), as ligands aimed at interacting
a
7 nAChR model, is consistent with the
requirement like its counterparts 2a–g, it exhibits reduced conju-
gation pattern and as a result, its tautomeric forms are the most
polar of the series, as expressed by their AlogP98 values (ꢁ2.33
and ꢁ3.74), which are the lowest in the series, as shown in Table
2. A similar trend was also observed when logD at pH 7.4 was used
instead of AlogP98 (data not shown). Reduced conjugation and
high polarity prevent the compound to interact with the hydro-
phobic and aromatic rings in the protein binding site, while such
interactions have been shown to be important both by ligand-
based pharmacophore modeling (Fig. 1 and Table 1) and receptor
structure-based docking studies described herein. On the whole,
with the
a7 nAChR. Our underlying reasoning was that quinucli-
dine analogs containing a HBD functional group separated from
Table 1
Summary statistics of the best ligand-based pharmacophore models
Pharmacophore
model
r
rmsd Confidence level
(%)
Chemical features
Distance constraint HBD-positive charge
(Å)
Enrichment
factor
Ac
0.77 1.02
70
HBA, HBD, positive charge &
hydrophobic
7.46–8.46
2.70
Bt
Ls
0.69 0.99
0.62 1.27
0.70 0.66
90
90
90
HBD & positive
HBA, HBD & positive charge
HBD, ring aromatic & positive charge
1.97–3.97
4.23–6.23
7.30–9.30
1.47
1.44
1.75
a7
The parameter r designates the Pearson correlation coefficient between the observed and predicted Kd (or Ki) values, rmsd designates the root-mean-squared-deviation
between observed and predicted Kd (or Ki) values. HBA and HBD designate hydrogen-bond acceptor and hydrogen-bond donor, respectively. The enrichment factor was
calculated using a test set of 493 diverse compounds, as described in methods.

David C. Kombo et al. / Bioorg. Med. Chem. Lett. 22 (2012) 1179–1186
1181
Figure 1. Heatmap of pharmacophore mapping & fitness for compounds 1a and 1b, derived from the best pharmacophore hypothesis obtained for rat
a7 Ac, Bt, and Ls.
Observed and predicted Kd values for AChBPs and Ki values for
a7 nAChR are also shown. Chemical features are colored as follows: green: hydrogen bond acceptor, red:
positive charge, magenta: hydrogen bond donor, and orange: aromatic ring.
for each ligand, the contribution of hydrophobic enclosures and
pairwise lipophilic interactions derived from docking made about
50% of the total Glide score. This result suggests a strong contribu-
tion of hydrophobic interactions in the overall free energy of
binding and is in agreement with the pharmacophore models
exhibits a longer HBD-cation distance (8.96 Å), in comparison to
the designed actives counterparts (5.47–7.65 Å). By contrast, com-
pounds 2h, 3c, 3e, which have a HBD-cation distance within the
shortest range of 3.52–4.72 Å, and thus lack a HBD that satisfies
the required distance constraint, do not interact with a7 nAChR.
obtained. Compound 3a (
binding affinity for
1300 nM) and 1a ( 7 nAChR Ki 420 nM), and demonstrated partial
agonism with an Emax of 53.0 4.4 % and an EC50 of 0.6 0.5 M, by
patch-clamp electrophysiology in rat 7 nAChR. Compound 1a
a
7 nAChR Ki 7.2 nM) exhibited better
Further examination of Table 2 indicates that on the whole,
designed compounds have a lower number of rotatable bonds
(1–3), as compared to the parent compounds 1a (4) and 1b (4). This
results in reduced ligand flexibility, which in turns would promote
binding affinity due to loss of configurational entropy. Thus, in
a7
than compounds 1b nAChR Ki
(a7
a
l
a

1182
David C. Kombo et al. / Bioorg. Med. Chem. Lett. 22 (2012) 1179–1186
Figure 4a. Designed compound 2b (in its lactim form) docked into rat a7 homology
model, is shown in ball and stick and colored green. The ligand OH group donates a
hydrogen-bond to the backbone carbonyl oxygen atom of Leu-119 (in the
complementary face), while its methoxy group accepts a hydrogen bond from
Gln-57 (in the complementary face). The cationic center donates a hydrogen-bond
to the backbone CO group of Trp-148, in the principal face. Neighboring amino-acid
side chains of the F-loop, such as Asp-164 and Ser-167, are shown in stick, while
residues making hydrogen-bond with the ligand are shown in ball and stick.
Figure 2a. Compound 1a docked into rat
a7 homology model, is shown in green.
The ligand OH group donates a hydrogen-bond to Ser-36 (in the complementary
face). The cationic center donates a hydrogen bond to the backbone CO group of
Trp-148, in the principal face. Also shown are the hydrogen-bond donor to acceptor
distances (2.00 and 2.15 Å) and the distance between the backbone oxygen atom of
Trp-148 and side-chain oxygen atom of Ser-36 and Asp-164 (12.9 Å).
Figure 2b. Compound 1a docked into rat
stick and colored green. Trp-148, Tyr-187 and Tyr-194 (of the principal face) which
are involved in cation– interactions with the basic nitrogen of the ligand, are
a7 homology model, is shown in ball and
Figure 4b. Designed compound 2b (in its lactim form) docked into rat a7 homology
p
model, is shown in green. Trp-148, Tyr-187 (of the principal face), and Trp-55, Ala-
163, Leu-119, and Tyr-168 (of the complementary face) which are involved in
hydrophobic enclosures interactions with the ligand, are shown in space-filling
representation. Portions of the ligand which are involved in hydrophobic enclosures
are shown in ball and stick.
shown in space-filling representation.
Figure 5. Designed compound 3b docked into rat a7 homology model. The ligand
OH group donates a hydrogen-bond to Tyr-168 (in the complementary face), which
is located in the F-loop (shown in dark blue). Also shown are neighboring amino-
acid side chains such as Ser-167, Asp-164, and Ser-36. The cationic center donates
to the backbone CO group of Trp-148, in the principal face. Both hydrogen-bond
distances are shown in broken lines.
Figure 3. Pharmacophore mapping of designed compound 3a onto the best
pharmacophore hypothesis of rat
a7. The positive charge, HBD and aromatic ring
chemical features are shown in red, magenta and orange, respectively.

David C. Kombo et al. / Bioorg. Med. Chem. Lett. 22 (2012) 1179–1186
1183
illustrates the mapping of compound 3a onto the best pharmaco-
phore of rat 7.
The binding modes of compounds 2b and 3b into the rat
a
a7 pro-
O
O
O
tein binding site, as predicted via docking, are shown in Figures 4
and 5, respectively. Figure 4a shows that the OH group of the
lactim form of compound 2b donates a hydrogen-bond to Gln-57.
The oxygen atom of its methoxy group, accepts a hydrogen-bond
from the backbone NH group of Leu-119. This predicted binding
mode is similar to the one experimentally-observed in the co-crys-
tal structure of compound 1b (conformation B) with AChBP.12 Fur-
thermore, Figure 4b shows that the designed ligand 2b is involved
in hydrophobic enclosures with both aliphatic and aromatic resi-
dues of both the principal and complementary faces. Similar pat-
tern was also observed with other designed compounds. On the
O
O
N
N
N
N
N
N
N
N
N
3e
2i
3a
N
e
O
b
Ar
N
N
a
N
d
N
N
N
whole, designed compounds exhibited stronger cation–
p and
N
hydrophobic enclosures interactions than parent compounds 1a
and 1b. Figure 5 shows that the NH2 group of compound 3b do-
nates a hydrogen-bond to the OH group of Tyr-168, a residue lo-
cated in the F-loop region. This predicted binding mode is similar
to the one experimentally-observed in the co-crystal structure of
compounds 1b (conformation A) and 1a, in complex with AChBP.12
N
g
f
4
c
2a-g
2h
O
N
N
N
N
A closer look at the rat a7 binding site suggests the existence of
alternative homologous receptor-based pharmacophoric con-
straints which appear to be consistent with the type of interactions
involving parent compounds and designed ligands described here-
in. As shown in Figure 6, the side-chain oxygen atom of Gn-117,
Tyr-168, Asn-77, Ser-36, Ser-34 and Ser-167 (of the complemen-
tary face) are separated from the main-chain carbonyl oxygen
atom of the highly conserved Trp-148 (of the principal face) with
a distance of about 10.3, 11.1, 11.3, 12.9, 13.9, and 14.7 Å, respec-
tively. Likewise, a side-chain nitrogen atom of Arg-79 and Gln-57
(of the complementary face) is separated from the same oxygen
atom of Trp-148 with a distance of 10.3 and 11.6 Å, respectively.
Thus, side-chain heteroatoms of many amino acid residues
N
O
N
N
N
3c
3d
N
N
3b
Scheme 1. Synthesis of
a7 nAChR ligands. Reagents and conditions: (a)
ArCOCH2CO2Me, i-BuOH, 100 °C, overnight; (b) methyl 2-(cyanomethoxy)benzoate,
CS2 (one drop), 100–110 °C, overnight; (c) (2-cyanophenoxy)acetonitrile, CS2 (one
drop), 100–110 °C, overnight; (d) [3-(dimethylamino)-2-phenylprop-2-enylidene]-
dimethylammonium hexafluorophosphate, MeOH, reflux, overnight; (e) diethyl
2-phenylmalonate, 150 °C, 5 min; (f) methyl benzimidate hydrochloride, methanol,
150 °C (microwave), 5 min; (g) 4-hydroxycoumarin, t-butanol, 100 °C, overnight.
surrounding the binding site of the rat
a7 appear to be almost
equidistantly located from the carbonyl oxygen atom of Trp-148,
which is known to accept a quintessential hydrogen-bond donated
by the cationic center, which in turn is engaged in cation–p inter-
actions, as shown in Figures 2 and 4. Such arrangement can accom-
addition to the proper spatial arrangement of the cationic center
with respect to the hydrogen-bond donor group and hydrophobic
aromatic group, ligand rigidity contributes to the increase in bind-
ing affinity of designed ligands, in comparison to GTS-21. Figure 3
modate diverse ligands and multiple binding modes.
Table 2
Percent inhibition, standard error, of control radioligand binding to nAChRs of the designed spirodiazepine and spiroimidazoline quinuclidines tested at 5
l
M in radioactive
displacement assays
Compound Structure
a7
a
4b2
a3b4
a1bc
d
AlogP98
HBD-cation NROT
distance (Å)
5.47
O
95 3 [3]
(140 7 [3])
HN
2a
15 1 [3]
NT
NT
NT
NT
NT
NT
0.54
2
3
3
N
N
O
HN
87 3 [3] (110
[1])
2b
7
1 [3]
0.52
0.52
5.53
5.59
N
N
O
N
O
HN
93 7 [3] (87
[1])
2c
12 1 [3]
O
N
(continued on next page)

1184
David C. Kombo et al. / Bioorg. Med. Chem. Lett. 22 (2012) 1179–1186
Table 2 (continued)
Compound Structure
a7
a4b2
a3b4
a1bc
d
AlogP98
HBD-cation
distance (Å)
NROT
O
HN
O
98 3 [6] (42
[1])
2d
42 14 [6]
63 0.4 [3]
38 5 [3]
0.52
5.55
5.61
3
N
N
O
HN
95 3 [6] (77
[1])
2e
62 15 [6]
89 0.45 [3]
28 6 [3]
1.25
0.26
2
2
S
Br
N
N
O
HN
HN
98 2 [3] (58
[1])
2f
20 1 [3]
NT
NT
NT
NT
5.60
S
N
N
N
O
84 4 [3] (150
[1])
2g
14 3 [3]
ꢁ0.07
ꢁ0.17
5.60
3.52
2
1
N
N
O
H
N
2h
2i
27 5 [3]
6
2
2 [3]
1 [3]
NT
NT
NT
NT
N
N
O
H
N
ꢁ2.33; ꢁ3.74;
ꢁ2.33
5.10; 5.46; 6.97;
6.15
2; 2;
3
19 1 [3]
O
O
N
H
93 6 [9] (7
[4])
3
N
3a
3b
32 11 [9]
55 3 [6]
15 10 [6]
0.30
5.75
7.65
2
1
N
N
H
OH
O
H
N
96 1 [3] (120
[1])
9
1 [3]
47 3 [3]
38 2 [3]
ꢁ0.20
N
H2N
N
3c
N
1
9 [3]
11 2 [3]
NT
NT
NT
NT
ꢁ 0.05
ꢁ0.29
4.64
1
2
NH
NH
N
N
77 3 [3] (290
[1])
3d
4
4
1 [3]
2 [3]
5.62 ;4.66
N
OH
N
3e
1b
33 3 [3]
13 1 [3]
26 5 [3]
0.54
3.03
4.72
NA
1
4
N
N
O
N
H
O
(3900 770
[3])
(1300 550 [4]) (51 10 [3])
(2500 270 [3])
N
O

David C. Kombo et al. / Bioorg. Med. Chem. Lett. 22 (2012) 1179–1186
1185
Table 2 (continued)
Compound Structure
a7
a
4b2
a3b4
a1bc
d
AlogP98
HBD-cation
distance (Å)
NROT
N
OH
106 3 [3] (420 97 1 [3]
[1]) (19 6 [4])
75 1 [3]
(970 110 [4])
1a
72 3 [3]
2.81
8.96
4
N
O
Values in parentheses represent Ki standard error in nM. The number of replicates is listed beside each result in brackets. The
a
7 radioligand [3H]-methyllycaconitine was
4b2 heterologously
used for
a
7 binding studies on rat hippocampal membranes and the nicotinic radioligand [3H]-epibatidine was used for binding studies at human
a
expressed in SH-EP1 cells, ganglion-type nicotinic receptors on SH-SY5Y cellular membranes and muscle-type nicotinic receptors on TE-671 cellular membranes. See
Experimental Section for details. NT = not tested. For comparison purpose, corresponding data for compounds 1a and 1b are also shown. The distance between the HBD and
the cationic center for the lowest-energy conformation is also shown. For compound 2i, the three plausible tautomeric forms are considered. The calculated partition
coefficient between the organic and the aqueuous phase expressed in terms of AlogP98, and the total number of rotatable bonds (NROT) for each compound, as derived from
Pipeline Pilot (Accelrys Inc., San Diego, CA, 2006), are also listed.
Figure 6. Conserved Trp-148 of the principal face and surrounding amino acid side-chains of the complementary face which are likely to interact with designed ligands. Also
shown are the distances between the main chain carbonyl oxygen atom of Trp-148 and a side chain heteroatom of another amino acid residue.
However, our pharmacophore modeling and molecular docking
studies described herein have neglected protein flexibility, and
assumed that the protein is rigid. One may envision that due to
molecular motions of the F-loop, side-chain oxygen atoms of
Ser-166 and Asp-164, which are located at about 20.5 and 19.4 Å
from the main chain carbonyl oxygen atom of Trp-148 in the
homology model used herein, may reach a position closer to the
conserved Trp residue, and thus meet the pharmacophoric
requirement.
Acknowledgments
We sincerely thank Dr. Gary Byrd for high resolution LCMS and
Christopher Helper for binding affinity studies. We also thank
Merouane Bencherif, Craig Miller, and Heather Savelle for helpful
discussions, suggestions and comments.
References and notes
In conclusion, we have used pharmacophore elucidation and
molecular docking to study the interactions of benzylidene
1. Lippiello, P. M.; Bencherif, M.; Hauser, T. A.; Jordan, K. G.; Letchworth, S. R.;
Mazurov, A. A. Expert Opin. Drug Discov. 2007, 2, 1185.
2. Law, R. J.; Henchman, R. H.; McCammon, A. Proc. Natl. Acad. Sci. U.S.A. 2005, 102,
6813.
3. Marks, M. J.; Collins, A. C. Mol. Pharmacol. 1982, 3, 554.
4. Freedman, R.; Hall, M.; Adler, L. E.; Leonard, S. Biol. Psychiatry 1995, 38, 22.
5. Bertrand, D.; Gopalakrishnan, M. Biochem. Pharmacol. 2007, 74, 1155.
6. Freedman, R.; Coon, H.; Myles-Worsley, M.; Orr-Urtreger, A.; Olincy, A.; Davis,
A.; Polymeropoulos, M.; Holil, J.; Hopkins, J.; Hoff, M.; Rosenthal, J.; Waldo, M.
C.; Reimherr, F.; Wender, P.; Yaw, J.; Young, D. A.; Breese, C. R.; Adams, C.;
Patterson, D.; Adler, L. E.; Kruglyak, L.; Leonard, S.; Byerley, W. Proc. Natl. Acad.
Sci. U.S.A. 1997, 94, 587.
7. Brejc, K.; van Dijk, W. J.; Klaassen, R. V.; Schhurmans, M.; van de Oost, J.; Smit,
A. B.; Sixma, T. K. Nature 2001, 411, 269.
8. Unwin, N. J. Mol. Biol. 2005, 346, 967.
9. Bourne, Y.; Talley, T. T.; Hansen, S. B.; Taylor, P.; Marchot, P. EMBO J. 2005, 24, 1512.
10. Celie, P. H.; Rossum-Fikkert, S. E.; van Dijk, W. J.; Brejc, K.; Smit, A. B.; Sixma, T.
K. Neuron 2004, 41, 907.
anabaseine congeners with AChBPs and rat
found that binding affinity is primarily mediated via electrostatic
interactions (hydrogen-bond, -cation and ) and hydrophobic
interactions as well. In particular, a hydrogen bond donor feature
appears to contribute to ligand binding to AChBPs and 7 nAChR.
a7 nAChR. We have
p
p–p
a
We have used this pharmacophore to design novel spirodiazepine
and spiroimidazoline quinuclidine series. As predicted, docking-
derived binding modes of these ligands are strikingly reminiscent
of those observed in co-crystal structures of benzylidene anabase-
ine analogs with AChBP. Experimentally-observed data have
shown that the designed compounds selectively bind to rat
a7
nAChR. In particular, compound 3a binds 7 nAChR with a much
a
better affinity than both GTS-21 and its active metabolite, and
demonstrates partial agonism. Our studies have also revealed the
existence of alternative pharmacophoric features equidistant from
the carbonyl oxygen atom of the conserved Trp-148 of the princi-
pal face, which may be exploited to design diverse focused libraries
11. Celie, P. H.; Klaassen, R. V.; Rossum-Fikkert, S. E.; van Elk, R.; van Nierop, P.;
Smit, A. B.; Sixma, T. K. J. Biol. Chem. 2005, 280, 26457.
12. Hibbs, R. E.; Sulzenbacher, G.; Shi, J.; Talley, T. T.; Conrod, S.; Kem, W. R.; Taylor,
P.; Marchot, P.; Bourne, Y. EMBO J. 2009, 28, 3040.
13. Mecozzi, S.; West, A. P.; Dougherty, D. A. Proc. Natl. Acad. Sci. U.S.A. 1996, 93,
10566.
14. Schmitt, J. D.; Sharples, C. G.; Caldwell, W. S. J. Med. Chem. 1999, 42, 3066.
targeting the a7 nAChR.

1186
David C. Kombo et al. / Bioorg. Med. Chem. Lett. 22 (2012) 1179–1186
15. Dougherty, D. A. J. Org. Chem. 2008, 16, 3667.
was heated at 100 °C overnight, cooled to ambient temperature and
concentrated. The residue was purified by preparative HPLC to yield 2a
trifluoroacetate (105 mg, 53%). 1H NMR (CD3OD, 300 MHz) d 8.07d (d, 2H), 7.72
(m, 1H), 7.60 (m, 2H), 4.23 (dd, 2H, J = 65, 12 Hz), 3.94 (dd, 2H, J = 28, 10 Hz),
3.59–3.38 (m, 6H), 2.49 (s, 1H), 2.35–2.10 (m, 4H). High resolution LSMS, m/e
284.1767, C17H22N3O, calcd 284.1763.
16. Ulens, C.; Akdemir, A.; Jongejan, A.; van Elk, R.; Bertrand, S.; Perrakis, A.; Leurs, R.;
Smit, A. B.; Sixma, T. K.; Bertrand, D.; de Esch, I. J. P. J. Med. Chem. 2009, 52, 2372.
17. Papke, R. L.; Meyer, E. M.; Lavieri, S.; Bollompally, S. R.; Papke, T. A.; Horenstein,
N. A.; Itoh, Y.; Papke, J. K. Neuropharmacology 2004, 46, 1023.
18. Stokes, C.; Papke, J. K.; Horenstein, N. A.; Kem, W. R.; McCormack, T. J.; Papke, R.
L. Mol. Pharmacol. 2004, 66, 14.
6-Phenylspiro[1,3-dihydro-1,4-diazepine-2,30-quinuclidine] (2h): 3-Amino-3-
(aminomethyl)quinuclidine (1) (75 mg, 0.5 mmol) and [3-(dimethylamino)-2-
phenylprop-2-enylidene]-dimethylammonium hexafluorophosphate (100 mg,
0.4 mmol) were dissolved in methanol (5 ml). The reaction mixture was
refluxed overnight and concentrated. The residue was purified by preparative
HPLC to yield 2h trifluoroacetate (75 mg (20%). 1H NMR (CD3OD, 300 MHz) d
0.15 (s, 1H), 8.01 (s, 1H), 7.52–7.38 (m, 5H), 4.50 (d, 1H, J = 12 Hz), 3.73 (dd, 2H,
J = 27, 12 Hz), 3.60–3.38 (m, 5H), 2.22–1.98 (m, 5H).
193D-QSAR Pharmacophore models aimed at predicting ligand binding to AChBPs
and rat a7 were built using the datasets described in ; Talley, T. T.; Yalda, S; Ho,
K.-Y.; Tor, Y.; Soti, F. S.; Kem, W. R.; Taylor, P. Biochemistry 2006, 45, 8894, and in
; Slavov, S. H.; Radzvilovits, M.; LeFrancois, S.; Stoyanova-Slavova, I.; Sot, F.;
Kem, W. R.; Katritzky, A. R. Eur. J. Med. Chem. 2010, 45, 2433, respectively. The
training sets used to build Ac, Bt, Ls, and rat a7 binding models were comprised
of 21, 20, 20, and 20 compounds, respectively. In the later case, adopting their
terminology, the ID of each benzylidene anabaseine analog selected to be
included in the training set was: 1.1, 1.11, 1.13, 1.14, 1.15, 1.2, 1.22, 1.23, 1.26,
1.27, 1.3, 1.36, 1.37, 1.4, 1.41, 1.42, 1.43, 1.44, 1.45, 1.5, 1.6, 1.7, 1.8, and 1.9.
Models were developed using Discovery Studio (Accelrys Inc., San Diego, CA,
2006). Variable weights to the chemical features were automatically assigned.
Ligand conformational models were generated using the BEST option, with an
energy cutoff of 20 kcal/mol and the maximum number of conformations set to
250. Fischer randomization test was used to demonstrate that derived models
were not generated by chance, that is, that a true correlation exists between the
biological activity and the structural features. Binding affinity prediction for
designed ligands was carried out by computing pharmacophore hypothesis
fitness values and Kd estimates using ‘ligand pharmacophore mapping’ module
6-Phenylspiro[1,4-diazepane-2,30-quinuclidine]-5,7-dione (2i): A mixture of 3-
amino-3-(aminomethyl)quinuclidine (1) (75 mg, 0.5 mmol) and diethyl 2-
phenylmalonate (118 mg, 0.5 mmol) were heated at 150 °C in microwave for
5 min. The reaction mixture was cooled to rt and purified by preparative HPLC
to yield 2i trifluoroacetate (98 mg, 24%). 1H NMR (CD3OD, 300 MHz) d 7.46–
7.22 (m, 5H), 4.30 (s, 1H), 4.02 (dd, 2H, J = 50, 13 Hz), 3.53 (dd, 2H, J = 53,
12 Hz), 3.53–3.15 (m, 4H), 2.23–1.98 (m, 5H). High resolution LCMS m/e
300.1723, C17H22N3O2, calcd 300.1712.
2-(3-Hydroxybenzofuran-2-yl)spiro[1,5-dihydroimidazole-4,30-quinuclidine] (3a):
A mixture of 3-amino-3-(aminomethyl)quinuclidine (1) (75 mg, 0.5 mmol),
methyl 2-(cyanomethoxy)benzoate (96 mg, 0.5 mmol) and one drop of carbon
disulfide was heated in a sealed vial at 100–110 °C overnight. The reaction
mixture was cooled to rt and purified by preparative HPLC to yield 3a
trifluoroacetate (98 mg, 48%). 1H NMR (CD3OD, 300 MHz) d 7.98 (d, 1H), 7.65
(m, 2H), 7.43 (m, 2H), 4.26 (dd, 2H, 76, 12), 3.77 (dd, 2H, J = 35, 14), 3.60–3.38
(m, 4H), 2.50–2.33 (m, 2H), 2.15 (m, 3H).
as implemented in Discovery Studio. Further validation of the
pharmacophore model was carried out by predicting a diverse library of 493
compounds, comprised of 127 actives ( 7 Ki 6 500 nM) and 366 decoys (a7
a7
a
Ki > 500 nM). The performance of the model to rank these compounds with
respect to their binding affinity, was evaluated by means of the area under the
receiver operating characteristic (ROC) curve, and the enrichment curve. The
ROC accuracy obtained was 0.62, and the enrichment factor of 1.75. If Yield
designates the ratio of the number of actives recovered to the total hit list size,
2-(3-Aminobenzofuran-2-yl)spiro[1,5-dihydroimidazole-4,30-quinuclidine] (3b): A
mixture of 3-amino-3-(aminomethyl)quinuclidine (1) (75 mg, 0.5 mmol), (2-
cyanophenoxy)acetonitrile (79 mg, 0.5 mmol) and one drop of carbon disulfide
was heated in a sealed vial at 100–110 °C overnight. The reaction mixture was
cooled to rt and purified by preparative HPLC to yield 3b trifluoroacetate
(145 mg, 68%). 1H NMR (CD3OD, 300 MHz) d 7.96 (d, 1H), 7.64 (t, 1H), 7.50 (d,
1H), 7.36 (t, 1H), 4.20 (dd, 2H, J = 60, 12 Hz, 2H), 3.79 (dd, 2H, J = 22, 12 Hz, 2H),
3.63–3.38 (m, 4H), 2.42 (m, 2H), 2.17 (m, 3H). High resolution LCMS m/e
290.1707, C17H21N4O, calcd 290.1715.
the enrichment is equal to Yield ⁄(D/A), where
D is the total number of
compounds in the dataset (493) and A the total number of active compounds in
the dataset (127)
20. Docking studies were carried out using Glide 5.5 (Schrodinger, Inc., 101 SW
Main Street, Suite 1300, Portland, OR 97204), as described in; Friesner, R. A.;
Banks, J. L.; Murphy, R. B.; Halgren, T. A.; Klicic, J. J.; Mainz, D. T.; Repasky, M. P.;
Knoll, E. H.; Shelley, M.; Perry, J. K.; Shaw, D. E.; Francis, P.; Shenkin, P. S. J. Med.
Chem. 2004, 47, 1739; Halgren, T. A.; Murphy, R. B.; Friesner, R. A.; Beard, H. S.;
Frye, L. L.; Pollard, W. T.; Banks, J. L. J. Med. Chem. 2004, 47, 1750. To validate
the procedure, a self-docking of 1b and its metabolite 1a into their cognate
Aplysia AChBP with the structure, obtained from Hibbs, R. E.; Sulzenbacher, G.;
Shi, J.; Talley, T. T.; Conrod, S.; Kem, W. R.; Taylor, P.; Marchot, P.; Bourne, Y.
EMBO J. 2009, 28, 3040, was carried . Glide-derived docked poses were within
an average rmsd of 1.3 Å with respect to the native co-crystalized
2-Phenylspiro[1,5-dihydroimidazole-4,30-quinuclidine] (3c):
A mixture of 3-
amino-3-(aminomethyl)quinuclidine (1) (475 mg, 3.0 mmol) and methyl
benzimidate hydrochloride (602 mg, 3.5 mmol) in methanol (3 ml) was
heated in microwave at 150 °C for 5 min. The reaction mixture was
concentrated in vacuo. The residue was purified by preparative HPLC to yield
3c (0.3 g, 28%). 1H NMR (CD3OD, 300 MHz) d 8.00 (d, 2H), 7.85 (m, 1H), 7.69
(m,2H), 4.33 (dd, 2H, J = 75, 11), 3.80 (dd, 2H), 3.59–3.37 (m, 4H), 2.59–2.40 (m,
2H), 2.09 (m, 3H). High resolution LCMS m/e 242.1649,
C15H20N3, calcd
242.1657.
conformation. Homology models of the extracellular domain of rat
a
7 nAChR
2-Spiro[1,5-dihydroimidazole-4,30-quinuclidine]-2-ylphenol (3d): 3-Amino-3-
(aminomethyl)quinuclidine (1) (75 mg, 0.5 mmol) and 4-hydroxycoumarin
(81 mg, 0.5 mmol) were dissolved in i-butanol (3 ml). The reaction mixture
was heated at 100 °C overnight, cooled to ambient temperature and
concentrated. The residue was purified by preparative HPLC to yield 3d
trifluoroacetate (15 mg, 8%). 1H NMR (CD3OD, 300 MHz) d 7.88 (d, 1H), 7.66 (m,
1H), 7.12 (m, 2H), 4.29 (dd, 2H, J = 79, 10), 4.78 (dd, 2H), 3.56–3.37 (m, 4H),
2.50–2.33 (m, 2H), 2.16 (m, 3H).
were obtained using MODELLER (Accelrys Inc., San Diego, CA, 2006), as
described in Eswar, N.; Marti-Renom, M. A.; Webb, B.; Madhusudhan, M. S.;
Eramian, D.; Shen, M.; Pieper, U.; Sali, A. Comparative Protein Structure
Modeling With MODELLER, In Current Protocols in Bioinformatics, John Wiley &
Sons; 2006, Supplement 15, 5.6.1-5.6.30, 200; Marti-Renom, M. A.; Stuart, A.;
Fiser, A.; Sánchez, R.; Melo, F.; Sali, A. Annu. Rev. Biophys. Biomol. Struct. 2000,
29, 291.The crystal structure of AChBP from lymnaea complexed with nicotine
(pdb code 1uw6) was used as template. The derived protein models were
validated using Procheck (Laskowski, R. A.; MacArthur, M. W.; Moss, D. S;
Thornton, J. M. App. Cryst. 1993, 26, 283) and Verify_3D (Eisenberg, D.; Luthy,
23. Binding assays to membrane bound nicotinic receptors were carried out using
standard methods adapted from published procedures, for example, Lippiello,
P. M.; Fernandes, K. G. Mol. Pharmacol. 1986, 29, 448; Davies, A. R.; Hardick, D.
J.; Blagbrough, I. S.; Potter, B. V.; Wolstenholme, A. J.; Wonnacott, S.
Neuropharmacology 1999, 38, 679. Single-point binding data was determined
R.; Bowie, J. U. Methods Enzymol. 1997, 277, 396). Further validation of the
model was carried out by docking diverse library of 493 compounds,
comprised of 127 actives ( 7 Ki 6 500 nM) and 366 decoys ( 7 Ki > 500 nM).
a7
a
a
a
at a competitor concentration of 5 lM and are expressed as the percent
Glide performance to rank these compounds with respect to their binding
affinity, was evaluated by means of the area under the receiver operating
characteristic (ROC) curve. The accuracy obtained was 0.74.
inhibition of control radioligand binding. For IC50 determinations, replicates for
each point of a seven-point dose-response curve were averaged and plotted
against the log of drug concentration. IC50 values (concentration of the
compound that produces 50% inhibition of binding) were determined by least
squares non-linear regression using GraphPad Prism software (GraphPAD, San
Diego, CA). Ki values were calculated using the Cheng–Prusoff equation, Cheng,
Y.; Prusoff, W. H. Biochem. Pharmacol. 1973, 22, 3099.
21. Mullen, G.; Napier, J.; Balestra, M.; DeCory, T.; Hale, G.; Macor, J.; Mack, R.;
Loch, J.; Wu, E.; Kover, A.; Verhoest, P.; Sampognaro, A.; Phillips, E.; Zhu, Y.;
Murray, R.; Griffith, R.; Blosser, J.; Gurley, D.; Machulskis, A.; Zongrone, J.;
Rosen, A.; Gordon, J. J. Med. Chem. 2000, 43, 4045.
22. 5-Phenylspiro[1,3-dihydro-1,4-diazepine-2,30-quinuclidine]-7-ol
(2a):
This
24. (a) Puodzhyunaite, B. A.; Yanchene, R. A.; Terent’ev, P. B. Chem. Heterocycl.
Compd. 1988, 16, 311; (b) Andronati, S. A.; Yur’eva, V. S.; Mazurov, A. A.;
Nivorozhkin, L. E. Chem. Heterocycl. Compd. 1984, 20, 451.
procedure illustrates the general method for preparation of 2a–g. 3-Amino-
3-(aminomethyl)quinuclidine (1) (75 mg, 0.5 mmol) and ethyl benzoylacetate
(100 mg, 0.5 mmol) were dissolved in i-butanol (3 ml). The reaction mixture