A computational study of the binding of 3-(arylidene) anabaseines to two major brain nicotinic acetylcholine receptors and to the acetylcholine binding protein

Article abstract of DOI:10.1016/j.ejmech.2010.02.027

Nicotinic acetylcholine receptors (nAChRs) have become targets for drug development in recent years. 3-(2,4-dimethoxybenzylidene)-anabaseine (DMXBA), which selectively stimulates the α7 nAChR, has been shown to alleviate some cognitive deficits associated with schizophrenia. In this paper we report an analysis of the interactions between 47 arylidene-anabaseines (including 45 benzylidene-anabaseines) and rat brain α7 and α4β2 nicotinic acetylcholine receptors, using three different modeling techniques, namely 2D-QSAR, 3D-QSAR and molecular docking to the Aplysia californica acetylcholine binding protein (AChBP), a water soluble, homomeric nAChR surrogate receptor with a known crystal structure. Our investigation indicates the importance of: (1) the nitrogen atom of the tetrahydropyridyl (THP) ring for hydrogen bond formation; (2) π-π interactions between the aromatic rings of the ligands and the nAChBP binding site; (3) molecular surface recognition expressed in terms of steric complimentarity. On the basis of the 3D-QSAR results, bulky substituents at positions 2 (and due to the rotational freedom also at position 6) and 4 of the benzylidene moiety, with highly electronegative atoms projecting approximately 3-3.5 A? away from the benzylidene ring at position 4 seem optimal for enhancing binding affinity to the α7 nAChR.

Full text of DOI:10.1016/j.ejmech.2010.02.027

European Journal of Medicinal Chemistry 45 (2010) 2433–2446
Contents lists available at ScienceDirect
European Journal of Medicinal Chemistry
journal homepage: http://www.elsevier.com/locate/ejmech
Original article
A computational study of the binding of 3-(arylidene) anabaseines to two major
brain nicotinic acetylcholine receptors and to the acetylcholine binding protein
Svetoslav H. Slavov a, Maksim Radzvilovits a, Susan LeFrancois b, Iva B. Stoyanova-Slavova a, Ferenc Soti b,
**
,
a,
*
William R. Kem b , Alan R. Katritzky
a Center for Heterocyclic Compounds, Department of Chemistry, University of Florida, Gainesville, FL 32611, USA
b Department of Pharmacology and Therapeutics, College of Medicine, University of Florida, Gainesville, FL 32610, USA
a r t i c l e i n f o
a b s t r a c t
Article history:
Nicotinic acetylcholine receptors (nAChRs) have become targets for drug development in recent years. 3-
Received 29 October 2009
Received in revised form
8 February 2010
Accepted 9 February 2010
Available online 13 February 2010
(2,4-dimethoxybenzylidene)-anabaseine (DMXBA), which selectively stimulates the a7 nAChR, has been
shown to alleviate some cognitive deficits associated with schizophrenia. In this paper we report an
analysis of the interactions between 47 arylidene-anabaseines (including 45 benzylidene-anabaseines)
and rat brain
a7 and a4b2 nicotinic acetylcholine receptors, using three different modeling tech-
niques, namely 2D-QSAR, 3D-QSAR and molecular docking to the Aplysia californica acetylcholine binding
protein (AChBP), a water soluble, homomeric nAChR surrogate receptor with a known crystal structure.
Our investigation indicates the importance of: (1) the nitrogen atom of the tetrahydropyridyl (THP) ring
Keywords:
Nicotinic acetylcholine receptor
QSAR
for hydrogen bond formation; (2)
p–p interactions between the aromatic rings of the ligands and the
nAChBP binding site; (3) molecular surface recognition expressed in terms of steric complimentarity. On
the basis of the 3D-QSAR results, bulky substituents at positions 2 (and due to the rotational freedom also
at position 6) and 4 of the benzylidene moiety, with highly electronegative atoms projecting approxi-
mately 3–3.5 Å away from the benzylidene ring at position 4 seem optimal for enhancing binding affinity
AChBP
Docking
Anabaseine
DMBXA
GTS-21
to the a7 nAChR.
Benzylidene-anabaseine
Ó 2010 Elsevier Masson SAS. All rights reserved.
1. Introduction
diffraction studies or have broadened NMR spectral resonance
peaks. Nicotinic acetylcholine receptors (nAChRs) are such
membrane proteins. The nAChRs belong to a superfamily of pen-
tameric ligand-gated ion channel (LGIC) receptors that also
includes GABA-A, 5HT3 and glycine-activated ionotropic ion
channels. The various nAChRs are composed of a number of
different combinations of subunits. For example, the most abun-
Rational drug design requires knowledge of both the structure
of the receptor and the ligand as isolated species, and also that of
the ligand–receptor complex. With the increasing availability of
crystal structures for many drug receptors, scientists working in the
field of computer aided drug design have changed their focus from
developing simple descriptor–property relationships to the
detailed investigation of the more complex ligand–receptor inter-
actions. The docking procedure responsible for fitting ligand and
receptor together in 3D-space is attracting much attention, and
there is a growing number of software packages available to enable
this important process in drug design [1].
dant subtypes in the brain are the homomeric
bungarotoxin sensitive) and the heteromeric 2 subtype. To date
seventeen different nAChR subunits have been identified: five ( 1,
1, ) expressed in skeletal muscle cells and twelve ( 2-10, 2-
a7 subtype (a-
a4b
a
b
g,
d,
3
a
b
4) predominantly in nerve cells. Vertebrate nAChRs are always
cation channels permeable to Naþ and Kþ and, to varying extents,
Ca2þ ions [2]. A low resolution electron microscopic (4 Å) structure
of a heteromeric muscle nAChR is the best nAChR structure avail-
able at this time [3].
Nicotine binds exclusively to these LGIC receptors, which are
normally activated by the ubiquitous neurotransmitter acetylcho-
line (ACh), making them important targets for the development of
new types of drugs. Understanding the molecular interactions of
nicotinic ligands with some nAChRs may, by facilitating the design
of new drugs for treating nicotine dependence, reduce the preva-
lence of smoking-related diseases, which are responsible for
Although investigation of how drugs work on central nervous
system (CNS) receptors has advanced rapidly in recent years, it is
still hampered by the fact that most of these receptors are large
membrane proteins that are difficult to crystallize for X-ray
* Corresponding author. Tel.: þ1 352 392 0554; fax: þ1 352 392 9199.
** Corresponding author. Tel.: þ1 352 392 0669; fax: þ1 352 392 9696.
E-mail addresses: wrkem@ufl.edu (W.R. Kem), katritzky@chem.ufl.edu (A.R.
Katritzky).
0223-5234/$ – see front matter Ó 2010 Elsevier Masson SAS. All rights reserved.
doi:10.1016/j.ejmech.2010.02.027

2434
S.H. Slavov et al. / European Journal of Medicinal Chemistry 45 (2010) 2433–2446
under argon atmosphere (to protect from moisture) and dried in
a vacuum desiccator at room temperature overnight.
approximately 20% of deaths each year in developed countries [4].
Also, accumulating evidence for abnormal CNS nAChR expression in
neurodevelopmental (schizophrenia) and neurodegenerative
diseases (Alzheimer’s and Parkinson’s diseases) has stimulated
pharmaceutical interest in development of nAChR subtype selec-
tive agonists to enhance mental functions, particularly sensory
gating, attention and cognition [5,6]. Recent studies have demon-
strated that some arylidene derivatives of a marine toxin, anaba-
Method B. If the product did not crystallize, the solution was
evaporated, dissolved in a small amount of water, basified with
solid sodium hydrogen carbonate, extracted with dichloro-
methane, and evaporated. The crude product was subsequently
purified by crystallization or silica gel chromatography. In some
cases semipreparative SG HPLC using a hexane–alcohol (vari-
able)–0.5% diethylamine solvent system with
a linearly
seine, can selectively stimulate the a7 nAChR [7–9]. This receptor
increasing concentration of either isopropanol or ethanol was
used for final compound purification.
has exceptionally high Ca permeability; actually, it can be consid-
ered to be a calcium channel unique in its ability to pass Ca ions
even at normal resting membrane potentials, when other Ca
channels are inactive. The ability of this nAChR to cause localized
and limited elevation of intracellular Ca levels probably explains its
ability to mediate the neuroprotective [10] and cognition
enhancing effects [11–13].
In spite of an ability to express and study mammalian nAChRs in
cultured cells, application of rational drug design to these targets
has been impeded by the unavailability of high resolution 3D
structures of the receptors. However, the crystal structure of
a soluble pentameric nAChR homolog, the acetylcholine binding
protein (AChBP), was reported eight years ago [14]. Subsequently
published crystal structures of this surrogate nAChR and its Aplysia
homolog, occupied by agonists or antagonists have, coupled with
mutagenesis studies on actual nAChRs, supplied a rather precise
picture of how nicotinic ligands bind to the ACh recognition sites of
nAChRs [15,16]. Quite recently, crystalline complexes of Aplysia
AChBP with the marine toxin anabaseine [8] and two of its
benzylidene-adducts, the drug candidate DMXBA and its active
metabolite 4-OH-DMXBA [9], have also been solved at high reso-
lution, affording much insight into the probable interactions of
2.1.1. 3-(4-Aminobenzylidene)anabaseine (1.1)
Method B. 1H NMR (DMSO-d6, TMS, 300 MHz,
d ppm): 8.61–8.57
(2H, m), 7.79 (1H, dt, J ¼ 8.1, 2.1 Hz), 7.42 (1H, ddd, J ¼ 7.8, 5.1,
0.6 Hz), 7.08 (2H, d, J ¼ 8.4 Hz), 6.55 (2H, d, J ¼ 8.4 Hz), 6.36 (1H, s),
3.69 (2H, app t, J ¼ 5.4 Hz), 3.32 (2H, s) 2.77 (2H, app td, J ¼ 6.3,
1.5 Hz), 1.71 (2H, app quintet, J ¼ 6.0 Hz). HRMS [ESI(þ)-TOF-MS]
calcd for C17H18N3 (M þ H)þ, Mr ¼ 264.1495; found 264.1506.
2.1.2. 3-(4-t-Butoxybenzylidene)anabaseine (1.2)
Prepared by reaction of N,N-dimethylformamide di-tert-butyl
acetal with compound 1.5 and purification by column chromatog-
raphy, yield 20%. 1H NMR (CDCl3, TMS, 300 MHz,
d ppm): 8.74 (1H,
dd, J ¼ 2.4, 0.6 Hz), 8.63 (1H, dd, J ¼ 4.8, 1.8 Hz), 7.81 (1H, dt, J ¼ 8.1,
1.8 Hz), 7.32 (1H, ddd, J ¼ 7.5, 4.8, 0.6 Hz), 7.23 (2H, d, J ¼ 8.4 Hz),
6.98 (2H, d, J ¼ 8.4 Hz), 6.60 (1H, s), 3.87 (2H, app t, J ¼ 5.4 Hz), 2.85
(2H, app td, J ¼ 6.6, 2.1 Hz), 1.84 (app quintet, J ¼ 6.0 Hz), 1.37 (9H,
s). HRMS [ESI(þ)-TOF-MS] calcd for C21H25N2O (M þ H)þ,
Mr ¼ 321.1961; found 321.1952.
these ligands with
How various substituents on the arylidene ring affect the
interaction of anabaseine derivatives with the 7 nAChR is the
a7 nAChRs [17].
2.1.3. 3-(4-Trifluoromethoxybenzylidene)anabaseine
bishydrochloride (1.3)
a
Method A, yield 25%. 1H NMR (DMSO-d6, TMS, 300 MHz,
major focus of this study. We have used a variety of computational
methods to determine the 2D- and 3D-QSAR relationships for 47
arylidene-anabaseine (BA) compounds. For comparison, a 2D-QSAR
d
ppm): 8.96–8.90 (2H, m), 8.20 (1H, dt, J ¼ 8.1, 1.8 Hz), 7.79–7.70
(3H, m), 7.51 (2H, d, J ¼ 8.1 Hz), 7.30 (1H, s), 3.85 (2H, app t,
J ¼ 5.7 Hz), 2.97 (2H, app t, J ¼ 6.0 Hz), 2.04 (2H, app quintet,
J ¼ 6.0). Anal. Calcd for C18H17F3Cl2N2O: C, 53.35; H, 4.23; N, 6.91.
Found: C, 53.45; H, 4.33; N, 6.85.
relationship for Kis of a4b2 neuronal receptors which are inhibited
to varying degrees by this type of compounds was also derived. A
number of the BAs were docked with the nAChR receptor surrogate
Aplysia AChBP. The results of our study will hopefully lead to the
design of new arylidene-anabaseine and other compounds that
optimally activate this important brain nAChR.
2.1.4. 3-(4-Methylaminobenzylidene)anabaseine bishydrochloride
(1.4)
Method A, yield 68%. 1H NMR (DMSO-d6, TMS, 300 MHz,
d
ppm): 8.96 (1H, dd, J ¼ 4.8, 1.5 Hz), 8.90 (1H, d, J ¼ 1.8 Hz), 8.23
(1H, dt, J ¼ 8.1, 1.8 Hz), 7.83 (1H, dd, J ¼ 8.1, 5.4 Hz), 7.50 (2H, d,
J ¼ 9.0 Hz), 7.03 (1H, s), 6.66 (2H, d, J ¼ 9.0 Hz), 3.76–3.36 (2H, m),
2.96 (2H, app t, J ¼ 6.0 Hz), 2.05 (2H, app quintet, J ¼ 5.7 Hz). HRMS
[ESI(þ)-TOF-MS] calcd for C18H20N3 (M þ H)þ, Mr ¼ 278.1652;
found 278.1640.
2. Materials and methods
2.1. Synthesis and chemical characterization of Arylidene-
Anabaseines
BAs were generally synthesized by the Kem-Soti laboratory as
previously described [9,18]. In all the reactions the highest purity
chemicals available (generally obtained from Sigma–Aldrich or
Fisher) were used. To a mixture of anabaseine dihydrochloride
monohydrate (101 mg, 0.4 mmole) and the appropriate benzalde-
hyde derivative (0.5 mmole) in dry ethanol (3 ml) one drop of
concentrated hydrochloric acid was added and stirred under a very
weak stream of dry argon gas in an oil bath of 70–75 ꢀC overnight, if
not given a stated time below. The system was protected from
moisture with a Drierite tube. The mixture was then cooled over-
night in a refrigerator. BA product purification proceeded according
to one of these methods, as noted:
2.1.5. 3-(4-Hydroxybenzylidene)anabaseine bishydrochloride (1.5)
Reaction time 9 days, Method A, yield 95%. 1H NMR (DMSO-d6,
TMS, 300 MHz,
d
ppm): 8.91 (1H, dd, J ¼ 5.1, 1.5 Hz), 8.85 (1H, dd,
J ¼ 2.1, 0.9 Hz), 8.14 (1H, dt, J ¼ 7.8, 1.8 Hz), 7.73 (1H, ddd, J ¼ 8.1, 4.8,
0.9 Hz), 7.54 (2H, d, J ¼ 8.7 Hz), 7.13 (1H, s), 6.92 (2H, d, J ¼ 8.7 Hz),
3.78 (2H, app t, J ¼ 5.4 Hz), 2.94 (2H, app t, J ¼ 6.0 Hz), 2.05 (2H, app
quintet, J ¼ 5.7 Hz). Anal. Calcd for C17H18Cl2N2O: C, 60.54; H, 5.38;
N, 8.31. Found: C, 60.32; H, 5.25; N, 8.21.
2.1.6. 3-(4-Methoxybenzylidene)anabaseine bishydrochloride (1.6)
Method A, yield 93%. 1H NMR (DMSO-d6, TMS, 300 MHz,
d
ppm): 8.94 (1H, dd, J ¼ 5.1, 1.8 Hz), 8.91 (1H, d, J ¼ 2.1 Hz), 8.21
(1H, dt, J ¼ 8.1, 1.8 Hz), 7.78 (1H, dd, J ¼ 8.1, 5.1 Hz), 7.64 (2H, d,
Method A. If the hydrochloride salt crystallized, it was filtered
and washed three times with a small amount of dry ethanol
J ¼ 9.0 Hz), 7.20 (1H, s), 7.08 (2H, d, 8.7 Hz), 3.84 (3H, s), 3.80 (2H,

S.H. Slavov et al. / European Journal of Medicinal Chemistry 45 (2010) 2433–2446
2435
app t, J ¼ 5.7 Hz), 3.00 (2H, app t, J ¼ 6.0 Hz), 2.05 (2H, app quintet,
2.1.14. 3-(4-Difluoromethoxybenzylidene)anabaseine
bishydrochloride (1.14)
Method A, yield 67%, mp.: 205–210 ꢀC (decomp.). 1H NMR
(DMSO-d6, TMS, 300 MHz, ppm): 8.99–8.92 (2H, m), 8.25 (1H, dt,
J ¼ 6.6 Hz). Anal. Calcd for C18H20Cl2N2O: C, 61.55; H, 5.74; N, 7.97.
Found: C, 61.54; H, 5.60; N, 7.96.
d
2.1.7. 3-(4-Methylbenzylidene)anabaseine bishydrochloride (1.7)
J ¼ 8.1, 1.8 Hz), 7.80 (1H, dd, J ¼ 7.8, 5.1 Hz), 7.70 (2H, d, J ¼ 8.7, Hz),
7.26 (1H, t, J ¼ 73.2 Hz, OCHF2), 7.31 (2H, d, J ¼ 8.7 Hz), 7.17 (1H, s),
3.83 (2H, app t, J ¼ 5.7 Hz), 2.98 (2H, app t, J ¼ 5.7 Hz), 2.05 (2H, app
quintet, J ¼ 6.0 Hz). Anal. Calcd for C18H18F2Cl2N2O: C, 55.83; H,
4.69; N, 7.23. Found: C, 55.79; H, 4.78; N, 7.17.
Method A, yield 67%.1H NMR (DMSO-d6, TMS, 300 MHz,
d ppm):
8.93 (1H, dd, J ¼ 5.1, 1.8 Hz), 8.89 (1H d, J ¼ 2.1 Hz), 8.19 (1H, dt,
J ¼ 7.8, 2.1 Hz), 7.75 (1H, dd, J ¼ 7.8, 5.1 Hz), 7.52 (2H, d, J ¼ 8.4 Hz),
7.33 (2H, d, J ¼ 8.1 Hz), 7.21 (1H, s), 3.82 (2H, app t, J ¼ 5.4 Hz), 3.00
(2H, app t, J ¼ 5.7 Hz), 2.37 (3H, s), 2.04 (2H, app quintet, J ¼ 6.0 Hz).
HRMS
[ESI(þ)-TOF-MS]
calcd
for
C18H19N2
(M þ H)þ,
2.1.15. 3-(4-Chlorobenzylidene)anabaseine bishydrochloride (1.15)
Mr ¼ 263.1543; found 263.1568.
Method A, yield 69%, 1H NMR (DMSO-d6, TMS, 300 MHz,
d
ppm): 8.97–8.91 (2H, m), 8.24 (1H, dt, J ¼ 8.1, 1.8 Hz), 7.78 (1H, dd,
2.1.8. 3-(4-Methylthiobenzylidene)anabaseine bishydrochloride
J ¼ 7.8, 5.1 Hz), 7.61 (4H, AB q), 7.26 (1H, s), 3.84 (2H, app t,
J ¼ 5.7 Hz), 2.96 (2H, app t, J ¼ 5.7 Hz), 2.04 (app quintet, J ¼ 6.0 Hz).
Anal. Calcd for C17H17Cl3N2: C, 57.40; H, 4.82; N, 7.88. Found: C,
57.21; H, 4.90; N, 7.81.
(1.8)
Method A, yield 95%. 1H NMR (DMSO-d6, TMS, 300 MHz,
d ppm):
9.01–8.93 (2H, m), 8.29 (1H, dt, J ¼ 7.8, 1.8 Hz), 7.84 (1H, dd, J ¼ 7.8,
5.1 Hz), 7.57 (2H, d, J ¼ 8.7 Hz), 7.37 (2H, d, J ¼ 8.7 Hz), 7.21 (1H, s),
3.82 (2H, app t, J ¼ 5.7 Hz), 3.00 (2H, app t, J ¼ 5.7 Hz), 2.53 (3H, s),
2.05 (2H, app quintet, J ¼ 5.7 Hz). Anal. Calcd for C18H20Cl2N2S: C,
58.85; H, 5.49; N, 7.63. Found: C, 58.80; H, 5.37; N, 7.56.
2.1.16. 3-[4-(1H-1,2,4-Triazol-1-yl)-benzylidene]anabaseine
trishydrochloride (1.16)
Method A, yield 71%, mp.: 195–200 ꢀC. 1H NMR (DMSO-d6, TMS,
300 MHz,
d ppm): 9.47 (1H, s), 8.98–8.93 (2H, m), 8.32 (1H, s), 8.25
2.1.9. 3-(4-Cyanobenzylidene)anabaseine (1.9)
(1H, dt, J ¼ 8.1, 1.8 Hz), 8.04 (2H, d, J ¼ 8.7 Hz), 7.86–7.75 (3H, m),
7.31 (1H, s), 3.85 (2H, app t, J ¼ 5.7 Hz), 3.03 (2H, app t, J ¼ 5.7 Hz),
2.07 (2H, app quintet, J ¼ 6.0 Hz), contains w3.8% of ethanol. HRMS
[ESI(þ)-TOF-MS] calcd for C19H18N5 (M þ H)þ, Mr ¼ 316.1557;
found 316.1583.
Reaction time 2 days, Method B, purified by recrystallization
from benzene, yield 20%, mp.: 160.5–161.5 ꢀC. 1H NMR (CDCl3, TMS,
300 MHz,
d
ppm): 8.75 (1H, dd, J ¼ 2.1, 0.6 Hz), 8.65 (1H, dd, J ¼ 4.8,
1.5 Hz), 7.83 (1H, dt, J ¼ 8.1, 1.8 Hz), 7.65 (2H, d, J ¼ 8.4 Hz), 7.39 (2H,
d, J ¼ 8.4 Hz), 7.35 (1H, ddd, J ¼ 7.8, 5.1, 0.9 Hz), 6.65 (1H, s), 3.93
(2H, app t, J ¼ 5.4 Hz), 2.81 (2H, app td, J ¼ 6.6, 2.1 Hz), 1.85 (app
quintet, J ¼ 6.3 Hz). Anal. Calcd for C18H15N3: C, 79.10; H, 5.53; N,
15.37. Found: C, 79.05; H, 5.46; N, 15.41.
2.1.17. 3-(4-Acetamidobenzylidene)anabaseine bishydrochloride
(1.17)
Method A, but concentrated hydrochloric acid was not added,
yield 87%. 1H NMR (DMSO-d6, TMS, 300 MHz,
d ppm): 10.58 (1H, s),
2.1.10. 3-(4-n-Propylbenzylidene)anabaseine bishydrochloride
(1.10)
8.97 (1H, dd, J ¼ 5.4, 1.5 Hz), 8.94 (1H, d, J ¼ 1.8 Hz), 8.26 (1H, dt,
J ¼ 7.8, 1.8 Hz), 7.82 (1H, dd, J ¼ 7.8, 4.8 Hz), 7.77 (2H, d, J ¼ 8.7 Hz),
7.61 (2H, d, J ¼ 8.7 Hz), 7.16 (1H, s), 3.81 (2H, app t, J ¼ 5.4 Hz), 3.01
(2H, app t, J ¼ 5.7 Hz), 2.09 (3H, s), 2.06 (2H, app quintet, J ¼ 5.4 Hz).
Anal. Calcd for C19H21Cl2N3O: C, 60.32; H, 5.60; N, 11.11. Found: C,
59.97; H, 5.76; N, 11.00.
Using 4-n-propylbenzaldehyde diethyl acetal, Method A, yield
57%. 1H NMR (DMSO-d6, TMS, 300 MHz,
d ppm): 8.98–8.92 (2H, m),
8.25 (1H, dt, J ¼ 8.4, 1.8 Hz), 7.80 (1H, ddd, J ¼ 7.8, 5.1, 0.6 Hz), 7.54
(2H, d, J ¼ 8.1 Hz), 7.34 (2H, d, J ¼ 8.4 Hz), 7.22 (1H, s), 3.82 (2H, app
t, J ¼ 5.7 Hz), 3.00 (2H, app t, J ¼ 6.0 Hz), 2.62 (2H, t, J ¼ 7.8 Hz), 2.05
(2H, app quintet, J ¼ 6.0 Hz), 1.60 (2H, sextet, J ¼ 7.5 Hz), 0.89 (3H, t,
J ¼ 7.2 Hz). Anal. Calcd for C20H24Cl2N2: C, 66.12; H, 6.66; N, 7.71.
Found: C, 66.07; H, 6.74; N, 7.75.
2.1.18. 3-[4-bis(2-Hydroxyethyl)aminobenzylidene]anabaseine
bishydrochloride (1.18)
Method A, yield, 59%. 1H NMR (DMSO-d6, TMS, 300 MHz,
d
ppm): 8.93 (1H, dd, J ¼ 5.1, 1.8 Hz), 8.86 (1H, d, J ¼ 1.8 Hz), 8.17
2.1.11. 3-(4-Acetoxybenzylidene)anabaseine (1.11)
(1H, dt, J ¼ 8.1, 1.8 Hz), 7.78 (1H, dd, J ¼ 7.8, 5.1 Hz), 7.51 (2H, d,
J ¼ 9.0 Hz), 7.05 (1H, s), 6.84 (2H, d, J ¼ 9.0 Hz), 3.72 (2H, br s), 3.56
(8H, s), 2.97 (2H, app t, J ¼ 6.0 Hz), 2.05 (2H, app quintet, J ¼ 5.4 Hz).
HRMS [ESI(þ)-TOF-MS] calcd for C21H26N3O2 (M þ H)þ,
Mr ¼ 352.2020; found 352.2017.
Prepared by acetylation of compound 1.5 with acetic anhydride
and purified by column chromatography, yield 72%. 1H NMR (CDCl3,
TMS, 300 MHz,
d
ppm): 8.74 (1H, d, J ¼ 1.5 Hz), 8.64 (1H, dd, J ¼ 4.8,
1.5 Hz), 7.82 (1H, dt, J ¼ 7.8, 1.8 Hz), 7.38–7.28 (3H, m), 7.09 (2H, d,
J ¼ 8.7 Hz), 6.63 (1H, s), 3.89 (2H, app t, J ¼ 5.7 Hz), 2.83 (2H, app td,
J ¼ 5.1,1.8 Hz), 2.31 (3H, s),1.84(app quintet, J ¼ 6.0 Hz). Anal. Calcd for
C19H18N2O2: C, 74.49; H, 5.92; N, 9.14. Found: C, 74.20; H, 6.09; N, 9.22.
2.1.19. 3-(4-Pivaloyloxybenzylidene)anabaseine (1.19)
Prepared by acylation of compound 1.5 with pivalic anhydride
and purified by column chromatography, yield 89%. 1H NMR (CDCl3,
2.1.12. 3-(4-Morpholinobenzylidene)anabaseine bishydrochloride
(1.12)
TMS, 300 MHz,
d
ppm): 8.75 (1H, d, J ¼ 1.8 Hz), 8.65 (1H, dd, J ¼ 4.8,
1.8 Hz), 7.83 (1H, dt, J ¼ 7.8, 1.8 Hz), 7.35 (1H, ddd, J ¼ 7.8, 5.1,
0.9 Hz), 7.31 (2H, d, J ¼ 8.7 Hz), 7.06 (2H, d, J ¼ 8.7 Hz), 6.64 (1H, s),
3.89 (2H, app t, J ¼ 5.7 Hz), 2.82 (2H, app td, J ¼ 6.3, 1.8 Hz),1.84 (2H,
app quintet, J ¼ 6.3 Hz),1.36 (9H, s), contains w2% of ethanol. HRMS
[ESI(þ)-TOF-MS] calcd for C22H25N2O2 (M þ H)þ, Mr ¼ 349.1911;
found 349.1900.
Method A, yield 73%, mp.: 247–250 ꢀC. 1H NMR (DMSO-d6, TMS,
300 MHz,
d
ppm): 8.96 (1H, dd, J ¼ 5.1, 1.5 Hz), 8.91 (1H, d,
J ¼ 2.4 Hz), 8.24 (1H, dt, J ¼ 8.1, 1.8 Hz), 7.82 (1H, dd, J ¼ 8.1, 5.4 Hz),
7.57 (2H, d, J ¼ 9.0 Hz), 7.11 (1H, s), 7.03 (2H, d, J ¼ 9.3 Hz), 3.80–3.67
(6H, m), 3.40–3.31 (4H, m), 2.99 (2H, app t, J ¼ 5.7 Hz), 2.05 (2H, app
quintet, J ¼ 5.7 Hz). HRMS [ESI(þ)-TOF-MS] calcd for C21H24N3O
(M þ H)þ, Mr ¼ 334.1914; found 334.1912.
2.1.20. 3-(4-Acetylbenzylidene)anabaseine bishydrochloride (1.20)
Method A, yield 97%, 1H NMR (DMSO-d6, TMS, 300 MHz,
d ppm):
2.1.13. 3-(4-Dimethylaminobenzylidene)anabaseine
bishydrochloride (1.13)
8.98–8.92 (2H, m), 8.26 (1H, dt, J ¼ 8.1, 2.1 Hz), 8.05 (2H, d,
J ¼ 8.4 Hz), 7.79 (1H, dd, J ¼ 8.1, 5.1 Hz), 7.72 (2H, d, J ¼ 8.4 Hz), 7.32
(1H, s), 3.86 (2H, app t, J ¼ 5.7 Hz), 2.99 (2H, app t, J ¼ 5.7 Hz), 2.05
Known compound [7,8].

2436
S.H. Slavov et al. / European Journal of Medicinal Chemistry 45 (2010) 2433–2446
(app quintet, J ¼ 5.7 Hz). Anal. Calcd for C19H20Cl2N2O: C, 62.82; H,
2.1.27. 3-(3-Hydroxybenzylidene)anabaseine bishydrochloride
(1.27)
5.55; N, 7.71. Found: C, 62.51; H, 5.72; N, 7.60.
Method A, yield 92%. 1H NMR (DMSO-d6, TMS, 300 MHz,
2.1.21. 3-(4-n-Butoxybenzylidene)anabaseine bishydrochloride
(1.21)
d
ppm): 8.94 (1H, dd, J ¼ 4.8, 1.2 Hz), 8.92 (1H, d, J ¼ 2.1 Hz), 8.22
(1H, dt, J ¼ 8.1, 1.8 Hz), 7.78 (1H, dd, J ¼ 7.8, 5.1 Hz), 7.30 (1H, t,
J ¼ 7.8 Hz), 7.16 (1H, s), 7.04–6.90 (3H, m), 3.83 (2H, app t,
J ¼ 5.7 Hz), 2.98 (2H, app t, J ¼ 5.7 Hz), 2.04 (2H, app quintet,
J ¼ 6.0 Hz). Anal. Calcd for C18H20Cl2N2O x H2O: C, 57.47; H, 5.67; N,
7.89. Found: C, 57.41; H, 5.80; N, 7.80.
Method A, yield 64%, 1H NMR (DMSO-d6, TMS, 300 MHz,
d
ppm): 8.94 (1H, dd, J ¼ 5.1, 1.5 Hz), 8.89 (1H, d, J ¼ 2.1 Hz), 8.19
(1H, dt, J ¼ 8.1, 1.8 Hz), 7.77 (1H, dd, J ¼ 8.1, 4.8 Hz), 7.62 (2H, d,
J ¼ 9.0 Hz), 7.19 (1H, s), 7.07 (2H, d, J ¼ 8.7 Hz), 4.06 (2H, t,
J ¼ 6.6 Hz), 3.80 (2H, app t, J ¼ 5.7 Hz), 2.99 (2H, app t, J ¼ 5.7 Hz),
2.05 (2H, app quintet, J ¼ 6.0 Hz), 1.71 (2H, quintet, J ¼ 6.9 Hz), 1.43
(2H, sextet, J ¼ 7.5 Hz), 0.93 (3H, t, J ¼ 7.5 Hz). Anal. Calcd for
C21H26Cl2N2O: C, 64.12; H, 6.66; N, 7.12. Found: C, 63.99; H, 6.95; N,
7.09.
2.1.28. 3-(2,4-Dimethoxybenzyl)anabasine (1.28)
Prepared by hydrogenation (at room temperature and room
pressure) over 10% Pd/C from compound 1.42 in ethanolic solution
followed by purification of the free base by column chromatog-
raphy and recrystallization from n-hexane, yield 14%, mp: 95–
2.1.22. 3-(2-Acetoxybenzylidene)anabaseine (1.22)
96.5 ꢀC. 1H NMR (CDCl3, TMS, 300 MHz,
d ppm): 8.62 (1H, d,
Prepared by acetylation of 3-(2-hydroxybenzylidene)anabaseine
with acetic anhydride and purified by column chromatography,
J ¼ 2.1 Hz), 8.49 (1H, dd, J ¼ 5.1, 1.8 Hz), 7.70 (1H, dm, J ¼ 7.8 Hz),
7.27 (1H, ddd, J ¼ 7.8, 4.8, 0.6 Hz), 6.83–6.77 (1H, m), 6.35–6.30 (2H,
m), 4.05 (1H, d, J ¼ 2.7 Hz), 3.74 (3H, s), 3.61 (3H, s), 3.30 (1H, dm,
J ¼ 11.4 Hz), 2.83 (1H, td, J ¼ 11.7, 2.7 Hz), 2.67 (1H, td, J ¼ 12.6,
2.1 Hz), 2.22–2.09 (2H, m), 1.99–1.76 (2H, m), 1.75–1.63 (1H, m),
1.60–1.38 (2H, m). Anal. Calcd for C19H24N2O2: C, 73.05; H, 7.74; N,
8.97. Found: C, 72.93; H, 8.22; N, 8.95.
yield 78%, 1H NMR (CDCl3, TMS, 300 MHz,
d ppm): 8.66 (1H, d,
J ¼ 1.8 Hz), 8.57 (1H, dd, J ¼ 4.8, 1.5 Hz), 7.73 (1H, dt, J ¼ 8.1, 1.8 Hz),
7.36–7.15 (4H, m), 7.01 (1H, dd, J ¼ 7.8, 1.5 Hz), 6.48 (1H, s), 3.85 (2H,
app t, J ¼ 5.7 Hz), 2.63 (2H, app td, J ¼ 6.6, 1.8 Hz), 2.11 (3H, s), 1.76
(2H, app quintet, J ¼ 6.3 Hz). HRMS [ESI(þ)-TOF-MS] calcd for
C17H17N2O (M þ H)þ, Mr ¼ 265.1335; found 265.1315.
2.1.29. 3-(2-Methoxy-4-glucuronidobenzylidene)anabaseine (1.29)
2.1.23. 3-(2-Trifluoromethoxybenzylidene)anabaseine
bishydrochloride (1.23)
Prepared from compound 1.43 by reaction with acetobromo-
-glucuronic acid methyl ester in methanolic lithium hydroxide
solution, removal of the protecting groups and purification of the
raw product by HPLC. 1H NMR (DMSO-d6, TMS, 300 MHz,
ppm):
8.90 (1H, dd, J ¼ 5.1, 1.8 Hz), 8.83 (1H, d, J ¼ 1.8 Hz), 8.08 (1H, dt,
J ¼ 8.4, 2.1 Hz), 7.69 (1H, dd, J ¼ 7.8, 5.4 Hz), 7.60 (1H, d, J ¼ 9.0 Hz),
7.32 (1H, s), 6.79 (1H, dd, J ¼ 8.7, 2.4 Hz), 6.71 (1H, d, J ¼ 2.4 Hz),
5.24 (1H, d, J ¼ 7.2 Hz), 3.98 (1H, d, J ¼ 9.3 Hz), 3.70 (s), some
multiplets at the water peak, 2.93 (2H, app t), 2.03 (2H, app
quintet). HRMS [ESI(-)-TOF-MS] calcd for C24H25N2O8 (M ꢁ H)ꢁ,
Mr ¼ 469.1616; found 469.1627.
a-
D
Reaction time 10 days, Method A, yield 60%. 1H NMR (DMSO-d6,
TMS, 300 MHz,
d
ppm): 8.96–8.90 (2H, m), 8.21 (1H, dt, J ¼ 8.4,
d
1.8 Hz), 7.79–7.72 (2H, m), 7.70–7.48 (3H, m), 7.18 (1H, s), 3.88 (2H,
app t, J ¼ 5.7 Hz), 2.87 (2H, app t, J ¼ 5.7 Hz), 2.04 (2H, app quintet,
J ¼ 5.7 Hz). Anal. Calcd for C18H17F3Cl2N2O: C, 53.35; H, 4.23; N, 6.91.
Found: C, 53.10; H, 4.55; N, 6.84
2.1.24. 3-(2-Difluoromethoxybenzylidene)anabaseine
bishydrochloride (1.24)
Method A, yield 77%.1H NMR (DMSO-d6, TMS, 300 MHz,
d ppm):
8.96–8.90 (2H, m), 8.21 (1H, dt, J ¼ 8.1, 1.8 Hz), 7.75 (1H, ddd, J ¼ 7.8,
5.1, 0.6 Hz), 7.68 (1H, dd, J ¼ 7.5, 1.8 Hz), 7.60 (1H, td, J ¼ 7.8, 1.8 Hz),
7.41 (1H, app t, J ¼ 7.5 Hz), 7.31 (1H, d, J ¼ 7.8 Hz), 7.24 (1H, s), 7.20
(1H, t, J ¼ 73.5 Hz, OCHF2), 3.87 (2H, app t, J ¼ 5.4 Hz), 2.90 (2H, app
t, J ¼ 5.7 Hz), 2.04 (2H, app quintet, J ¼ 6.0 Hz). Anal. Calcd for
C18H18F2Cl2N2O: C, 55.83; H, 4.69; N, 7.23. Found: C, 55.82; H, 4.82;
N, 7.21.
2.1.30. 3-(2-i-Propoxy-4-methoxybenzylidene)anabaseine (1.30)
Method B, yield 99%. 1H NMR (CDCl3, TMS, 300 MHz,
d ppm):
8.76 (1H, s), 8.62 (1H dm, J ¼ 4.8 Hz), 7.83 (1H, dm, J ¼ 8.1 Hz), 7.34–
7.22 (2H, m), 6.82 (1H, s), 6.49 (1H, dm, J ¼ 8.7 Hz), 6.43 (1H, s), 4.41
(1H, heptet, J ¼ 6.3 Hz), 3.90 (2H, app t, J ¼ 5.1 Hz), 3.80 (3H, s), 2.76
(2H, app tm, J ¼ 6.3 Hz), 1.84 (1H, app quintet, J ¼ 6.3 Hz), 1.22 (6H,
d, J ¼ 6.3 Hz). HRMS [ESI(þ)-TOF-MS] calcd for C21H25N2O2
(M þ H)þ, Mr ¼ 337.1911; found 337.1898.
2.1.25. 3-(2-Pivaloyloxybenzylidene)anabaseine (1.25)
Prepared by acylation of 3-(2-hydroxybenzylideen)anabaseine
with pivalic anhydride and purified by column chromatography,
2.1.31. 3-[2-(Pentan-3-yloxy)-4-methoxybenzylidene]anabaseine
(1.31)
yield 91%, 1H NMR (CDCl3, TMS, 300 MHz,
d
ppm): 8.71 (1H, d,
Method B, yield 99%. 1H NMR (CDCl3, TMS, 300 MHz,
d ppm):
J ¼ 1.5 Hz), 8.62 (1H, dd, J ¼ 4.8, 1.5 Hz), 7.77 (1H, dt, J ¼ 8.1,
1.8 Hz), 7.39 (1H, dd, J ¼ 7.8, 1.8 Hz), 7.35–7.23 (3H, m), 7.02 (1H,
dd, J ¼ 7.8, 1.5 Hz), 6.58 (1H, s), 3.90 (2H, app t, J ¼ 5.7 Hz), 2.76–
2.67 (2H, m), 1.81 (2H, app quintet, J ¼ 6.3 Hz). HRMS [ESI(þ)-
TOF-MS] calcd for C22H25N2O2 (M þ H)þ, Mr ¼ 349.1911; found
349.1903.
8.73 (1H, s), 8.59 (1H, dm, J ¼ 5.1 Hz), 7.80 (1H, dm, J ¼ 7.6 Hz), 7.34–
7.22 (2H, m), 6.86 (1H, s), 6.48 (1H, dm, J ¼ 8.9 Hz), 6.40 (1H, s), 4.03
(1H, quintet, J ¼ 6.3 Hz), 3.89 (2H, app t, J ¼ 6.3 Hz), 3.80 (3H, s),
2.78 (2H, app tm, J ¼ 6.3 Hz), 1.83 (2H, app quintet, J ¼ 6.3 Hz), 1.56
(4H, quintet, J ¼ 6.3 Hz), 0.79 (6H, t, J ¼ 7.2 Hz). HRMS [ESI(þ)-TOF-
MS] calcd for C23H29N2O2 (M þ H)þ, Mr ¼ 365.2224; found
365.2218.
2.1.26. 3-(3-Methoxybenzylidene)anabaseine bishydrochloride
(1.26)
2.1.32. 3-(2-Acetoxy-4-methoxybenzylidene)anabaseine (1.32)
Prepared from compound 1.44 by acetylation with acetic
anhydride and purification by column chromatography, yield 8%.
Method A, yield 80%. 1H NMR (DMSO-d6, TMS, 300 MHz,
d
ppm): 8.97–8.91 (2H, m), 8.24 (1H, dt, J ¼ 8.1, 1.8 Hz), 7.79 (1H, dd,
J ¼ 7.8, 5.1 Hz), 7.44 (1H, t, J ¼ 7.8 Hz), 7.25 (1H, s), 7.19 (1H, d,
J ¼ 7.8 Hz) 7.15–7.06 (2H, m), 3.84 (2H, app t, J ¼ 5.4 Hz), 3.77 (3H, s),
3.00 (2H, app t, 5.7 Hz), 2.04 (2H, app quintet, J ¼ 6.00 Hz). Anal.
Calcd for C18H20Cl2N2O: C, 61.55; H, 5.74; N, 7.97. Found: C, 61.50; H,
5.99; N, 7.96.
1H NMR (CDCl3, TMS, 300 MHz,
d
ppm): 8.71 (1H, d, J ¼ 2.1 Hz),
8.63 (1H, dd, J ¼ 4.8, 1.5 Hz), 7.78 (1H, dt, J ¼ 7.8, 2.1 Hz), 7.37–7.29
(2H, m), 6.81 (1H, dd, J ¼ 8.7, 2.7 Hz), 6.63 (1H, d, J ¼ 2.7 Hz), 6.48
(1H, s), 3.89 (2H, app t, J ¼ 5.7 Hz), 3.81 (3H, s), 2.76–2.68 (2H, m),
2.14 (3H, s), 1.82 (2H, app quintet, J ¼ 5.7 Hz). HRMS [ESI(þ)-TOF-

S.H. Slavov et al. / European Journal of Medicinal Chemistry 45 (2010) 2433–2446
2437
MS] calcd for C20H21N2O3 (M þ H)þ, Mr ¼ 337.1547; found
J ¼ 1.5 Hz), 8.21 (1H, dt, J ¼ 8.1, 1.8 Hz), 7.79 (1H, dd, J ¼ 7.8, 5.1 Hz),
7.24–7.11 (3H, m), 6.99 (1H, d, J ¼ 8.4 Hz), 4.37–4.24 (4H, m), 3.79
(2H, app t, J ¼ 5.7 Hz), 2.98 (2H, app t, J ¼ 5.7 Hz), 2.04 (2H, app
quintet, J ¼ 6.0 Hz). Anal. Calcd for C19H20Cl2N2O2: C, 60.17; H, 5.32;
N, 7.39. Found: C, 59.83; H, 5.39; N, 7.29.
337.1546.
2.1.33. 3-(2-Hydroxy-4-diethylaminobenzylidene)anabaseine
(1.33)
Method B, yield 29%. 1H NMR (DMSO-d6, TMS, 300 MHz,
d ppm):
9.27 (1H br s), 8.59–8.55 (2H, m), 7.78 (1H, dt, J ¼ 7.5, 2.1 Hz), 7.41
(1H, ddd, J ¼ 7.5, 4.8, 0.6 Hz), 7.23 (1H, d, J ¼ 8.7 Hz), 6.73 (1H, s),
6.19 (1H, dd, J ¼ 8.7, 2.4 Hz), 6.12 (1H, d, J ¼ 2.4 Hz), 3.70 (2H, app t,
J ¼ 5.4 Hz), 3.28 (4H, q, J ¼ 6.9 Hz), 2.74 (2H, app t, J ¼ 5.7 Hz), 1.71
(2H, app quintet, J ¼ 6.3 Hz), 1.08 (6H, t, J ¼ 6.9 Hz). HRMS [ESI(þ)-
TOF-MS] calcd for C21H26N3O (M þ H)þ, Mr ¼ 336.2070; found
336.2075.
2.1.40. 3-(3,4-Methylenedioxybenzylidene)anabaseine
bishydrochloride (1.40)
Method A, yield 96%, mp.: 236–238 ꢀC. 1H NMR (DMSO-d6, TMS,
300 MHz,
d
ppm): 8.94 (1H, dd, J ¼ 5.1, 1.5 Hz), 8.90 (1H, d,
J ¼ 1.5 Hz), 8.20 (1H, dt, J ¼ 8.1, 1.8 Hz), 7.78 (1H, dd, J ¼ 8.1, 5.1 Hz),
7.26 (1H, d, J ¼ 1.5 Hz), 7.20 (1H, dd, J ¼ 8.4, 1.2 Hz), 7.16 (1H, s), 7.08
(1H, d, J ¼ 8.1 Hz), 6.15 (2H, s), 3.80 (2H, app t, J ¼ 5.4 Hz), 2.99 (2H,
app t, J ¼ 5.7 Hz), 2.04 (app quintet, J ¼ 5.7 Hz). Anal. Calcd for
C18H18Cl2N2O2: C, 59.19; H, 4.97; N, 7.68. Found: C, 58.79; H, 4.95; N,
7.85.
2.1.34. 3-(2,6-Dimethyl-4-hydroxybenzylidene)anabaseine (1.34)
Method B, purified by column chromatography, yield 64%. 1H
NMR (CDCl3, TMS, 300 MHz,
d
ppm): 8.77 (1H, dd, J ¼ 2.4, 0.6 Hz),
8.64 (1H, dd, J ¼ 4.8, 1.5 Hz), 7.94 (1H, dt, J ¼ 7.8, 1.8 Hz), 7.40 (1H,
ddd, J ¼ 7.8, 5.1, 0.9 Hz), 6.53 (1H, s), 6.50 (2H, s), 3.90 (2H, app t,
J ¼ 5.4 Hz), 2.27–2.18 (2H, m), 2.00 (6H, s), 1.77 (2H, app quintet,
J ¼ 6.6 Hz). Anal. Calcd for C19H20N2O: C, 78.05; H, 6.90; N, 9.58.
Found: C, 77.83; H, 7.29; N, 9.57.
2.1.41. 3-(Benzylidene)anabaseine bishydrochloride (1.41)
Method A, yield 53%, 1H NMR (DMSO-d6, TMS, 300 MHz,
d
ppm): 8.98–8.92 (2H, m), 8.25 (1H, dt, J ¼ 8.4, 1.5 Hz), 7.79 (1H,
ddd, J ¼ 7.8, 5.1, 0.6 Hz), 7.64–7.56 (2H, m), 7.55–7.48 (3H, m), 7.27
(1H, s), 3.84 (2H, app t, J ¼ 5.7 Hz), 3.00 (2H, app t, J ¼ 5.7 Hz), 2.05
(2H, app quintet, J ¼ 6.0 Hz). HRMS [ESI(þ)-TOF-MS] calcd for
C17H17N2 (M þ H)þ, Mr ¼ 249.1386; found 249.1395.
2.1.35. 3-(3,5-Dimethyl-4-hydroxybenzylidene)anabaseine
bishydrochloride (1.35)
Method A, yield 74%. 1H NMR (DMSO-d6, TMS, 300 MHz,
d
ppm):
2.1.42. 3-(2,4-Dimethoxybenzylidene)anabaseine bishydrochloride
8.94 (1H, dd, J ¼ 5.1, 1.8 Hz), 8.88 (1H, dd, J ¼ 5.4, 0.6 Hz), 8.18 (1H,
dt, J ¼ 7.8, 1.8 Hz), 7.78 (1H, ddd, J ¼ 7.8, 5.1 0.9 Hz), 7.30 (2H, s), 7.06
(1H, s), 3.78 (2H, app t, J ¼ 5.4 Hz), 3.01 (2H, app t, J ¼ 6.0 Hz), 2.19
(6H, s), 2.05 (2H, app quintet, J ¼ 6.0 Hz). Anal. Calcd for
C19H22Cl2N2O: C, 62.47; H, 6.07; N, 7.67. Found: C, 62.11; H, 6.00; N,
7.59.
(1.42)
Known compound [9].
2.1.43. 3-(2-Methoxy-4-hydroxybenzylidene)anabaseine (1.43)
Known compound [9].
2.1.44. 3-(2-Hydroxy-4-methoxybenzylidene)anabaseine (1.44)
2.1.36. 3-(2,4,6-Trimethylbenzylidene)anabaseine bishydrochloride
Known compound [9].
(1.36)
Method A. 1H NMR (DMSO-d6, TMS, 300 MHz,
d
ppm): 8.95–
2.1.45. 3-(2,4-Dihydroxybenzylidene)anabaseine (1.45)
8.90 (2H, m), 8.23–8.17 (1H, m), 7.75 (1H, ddd, J ¼ 8.1, 5.1, 0.9 Hz),
7.25 (1H, s), 6.94 (2H, s), 3.85 (2H, app t, J ¼ 5.7 Hz), 2.38–2.30 (2H,
m), 2.25 (3H, s), 2.10 (6H, s), 2.03–1.92 (2H, m). HRMS [ESI(þ)-TOF-
MS] calcd for C20H23N2 (M þ H)þ, Mr ¼ 291.1856; found 291.1861.
Known compound [9].
2.1.46. 3-(6-Methoxynaphth-2-ylmethylene)anabaseine
bishydrochloride(2.1)
Method A, yield 69%, mp.: 223–226 ꢀC (decomp). 1H NMR
2.1.37. 3-(2,4,6-Trimethoxybenzylidene)anabaseine
(DMSO-d6, TMS, 300 MHz, d ppm): 9.04–8.97 (2H, m), 8.34 (1H, dt,
bishydrochloride (1.37)
J ¼ 8.1, 1.8 Hz), 8.19 (1H, s), 7.99–7.83 (3H, m), 7.70 (1H, dd, J ¼ 8.4,
1.2 Hz), 7.41 (1H, d, J ¼ 2.7 Hz), 7.37 (1H, s), 7.25 (1H, dd, J ¼ 9.0,
2.4 Hz), 3.91 (3H, s), 3.85 (2H, app t, J ¼ 5.4 Hz), 3.14 (2H, app t,
J ¼ 5.7 Hz), 2.09 (2H, app quintet, J ¼ 5.7 Hz). Anal. Calcd for
C22H22Cl2N2O: C, 65.84; H, 5.53; N, 6.98. Found: C, 65.46; H, 5.55; N,
6.86.
Method A. 1H NMR (DMSO-d6, TMS, 300 MHz,
d ppm): 8.92 (1H,
dd, J ¼ 4.8, 1.5 Hz), 8.85 (1H, d, J ¼ 2.1 Hz), 8.12 (1H, dt, J ¼ 8.1,
2.1 Hz), 7.74 (1H, dd, J ¼ 7.8, 5.1 Hz), 7.05 (1H, s), 6.33 (2H, s), 3.85
(3H, s), 3.82 (2H, app t, J ¼ 6.0 Hz), 3.79 (6H, s), w2.5 (2H, under the
DMSO signal), 1.97 (2H, app quintet, J ¼ 6.0 Hz). Anal. Calcd for
C20H24Cl2N2O3: C, 58.40; H, 5.88; N, 6.81. Found: C, 58.08; H, 5.86;
N, 6.68.
2.1.47. 3-(Benzofuran-2-ylmethylene)anabaseine bishydrochloride
(2.2)
2.1.38. 3-(3,4,5-Trimethoxybenzylidene)anabaseine
bishydrochloride (1.38)
Method A, yield 63%. 1H NMR (DMSO-d6, TMS, 300 MHz,
d
ppm):
d
8.97 (1H, dd, J ¼ 4.8, 1.5 Hz), 8.94 (1H, d, J ¼ 1.5 Hz), 8.25
Method A, yield 88%. 1H NMR (DMSO-d6, TMS, 300 MHz,
(1H, dt, J ¼ 8.1, 1.8 Hz), 7.85–7.77 (2H, m), 7.73–7.66 (2H, m), 7.52
(1H, td, J ¼ 7.2, 1.2 Hz), 7.36 (1H, td, J ¼ 7.5, 0.6 Hz), 7.16 (1H, s), 3.85
(1H, app t, J ¼ 5.4 Hz), 3.25 (1H, app t, J ¼ 5.7 Hz), 2.14 (1H, app
quintet, J ¼ 5.7 Hz), contains w5.1% of ethanol. HRMS [ESI(þ)-TOF-
MS] calcd for C19H17N2O (M þ H)þ, Mr ¼ 289.1335; found 289.1334.
d
ppm): 8.96 (1H, dd, J ¼ 5.1, 1.5 Hz), 8.92 (1H, d, J ¼ 1.5 Hz), 8.22
(1H, dt, J ¼ 8.1, 1.8 Hz), 7.80 (1H, ddd, J ¼ 7.8, 5.1, 0.6 Hz), 7.21 (1H, s),
6.93 (2H, s), 3.85–3.75 (8H, m), 3.73 (3H, s), 3.06 (2H, app t,
J ¼ 5.7 Hz), 2.05 (2H, app quintet, J ¼ 5.7 Hz), contains w14% of
ethanol. HRMS [ESI(þ)-TOF-MS] calcd for C20H23N2O3 (M þ H)þ,
Mr ¼ 339.1703; found 339.1711.
2.2. Measurements of arylidene-anabaseine affinities for neuronal
nicotinic acetylcholine receptors
2.1.39. 3-(3,4-Ethylenedioxybenzylidene)anabaseine
bishydrochloride (1.39)
Equilibrium inhibition constants (Kis) for receptor binding of the
arylidene-anabaseines were measured in the Kem laboratory as
previously described [9] by their displacement of a radioligand
Method A, yield 90%, mp.: 232–233 ꢀC. 1H NMR (DMSO-d6, TMS,
300 MHz,
d
ppm): 8.95 (1H, dd, J ¼ 5.1, 1.5 Hz), 8.90 (1H, d,

2438
S.H. Slavov et al. / European Journal of Medicinal Chemistry 45 (2010) 2433–2446
Table 1
Compound names and structures.
N
R2
N
(1)
(2)
ID
Compound full IUPAC name
Substituent position
1.1
3-(4-Aminobenzylidene)anabaseine
4-NH2
1.2
1.3
1.4
1.5
3-(4-Butoxybenzylidene)anabaseine
4-O-tC4H9
4-OCF3
4-NHCH3
3-(4-Trifluoromethoxybenzylidene)anabaseine
3-(4-Methylaminobenzylidene)anabaseine
3-(4-Hydroxybenzylidene)anabaseine
4-OH
1.6
3-(4-Methoxybenzylidene)anabaseine
4-OCH3
1.7
3-(4-Methylbenzylidene)anabaseine
4-CH3
1.8
1.9
3-(4-Methylthiobenzylidene)anabaseine
3-(4-Cyanobenzylidene)anabaseine
4-SCH3
4-CN
1.10
1.11
1.12
1.13
1.14
1.15
1.16
1.17
1.18
1.19
1.20
1.21
1.22
1.23
1.24
1.25
1.26
1.27
1.28
1.29
1.30
1.31
1.32
1.33
1.34
1.35
1.36
1.37
1.38
1.39
1.40
1.41
1.42
1.43
1.44
1.45
2.1
3-(4-n-Propylbenzylidene)anabaseine
3-(4-Acetoxybenzylidene)anabaseine
4-nC3H7
4-OCOCH3
4-Mop
4-N(CH3)2
4-OCHF2
4-Cl
3-(4-Morpholinobenzylidene)anabaseine
3-(4-Dimethylaminobenzylidene)anabaseine
3-(4-Difluoromethoxybenzylidene)anabaseine
3-(4-Chlorobenzyidene)anabaseine
3-[4-(1H-1,2,4-Triazol-1-yl)benzylidene]anabaseine
3-(4-Acetamidobenzylidene)anabaseine
3-[4-bis(2-Hydroxyethyl)aminobenzylidene]anabaseine
3-(4-Pivaloyloxybenzylidene)anabaseine
3-(4-Acetylbenzylidene)anabaseine
3-(4-n-Butoxybenzylidene)anabaseine
3-(2–Acetoxybenzylidene)anabaseine
3-(2-Trifluoromethoxybenzylidene)anabaseine
3-(2-Difluoromethoxybenzylidene)anabaseine
3-(2-Pivaloyloxybenzylidene)anabaseine
3-(3-Methoxybenzylidene)anabaseine
4-Trz
4-NHCOCH3
4-N[(CH2)2OH]2
4-OCOButt
4-COCH3
4-O-nC4H9
2-OCOCH3
2-OCF3
2-OCHF2
2-OCOButt
3-OCH3
3-(3-Hydroxybenzylidene)anabaseine
3-(2,4-Dimethoxybenzyl)anabasine
3-OH
2,4-(OCH3)2-Ai
2-OCH3, 4-Gluc
2-OiPr, 4-OCH3
2-(pentan-3-yloxy), 4-OCH3
2-OCOCH3, 4-OCH3
2-OH, 4-N(C2H5)2
2,6-(CH3)2, 4-OH
3,5-(CH3)2, 4-OH
2,4,6-(CH3)3
2,4,6-(OCH3)3
3,4,5-(OCH3)3
3,4-EDO
3-(2-Methoxy-4-glucuronidobenzylidene)anabaseine
3-(2-i-Propoxy-4-methoxybenzylidene)anabaseine
3-[2-(Pentan-3-yloxy)-4-methoxybenzylidene]anabaseine
3-(2-Acetoxy-4-methoxybenzylidene)anabaseine
3-(2-Hydroxy-4-diethylaminobenzylidene)anabaseine
3-(2,6-Dimethyl-4-hydroxybenzylidene)anabaseine
3-(3,5-Dimethyl-4-hydroxybenzylidene)anabaseine
3-(2,4,6-Trimethylbenzylidene)anabaseine
3-(2,4,6-Trimethoxybenzylidene)anabaseine
3-(3,4,5-Trimethoxybenzylidene)anabaseine
3-(3,4-Ethylenedioxybenzylidene)anabaseine
3-(3,4-Methylenedioxybenzylidene)anabaseine
3-(Benzylidene)anabaseine
3-(2,4-Dimethoxybenzylidene)anabaseine [DMXBA]
3-(2-Methoxy-4-hydroxybenzylidene)anabaseine
3-(2-Hydroxy-4-methoxybenzylidene)anabaseine
3-(2,4-Dihydroxybenzylidene)anabaseine
3-(6-Methoxynaphth-2-ylmethylene)anabaseine
3-(Benzofuran-2-ylmethylene)anabaseine
3,4-MDO
H
2,4-(OCH3)2
2-OCH3, 4-OH
2-OH, 4-OCH3
2,4-(OH)2
(6-CH3O)-Nph
2-BF
2.2
specific for either
([3H]cytisine) nAChRs; these were purchased from Perkin-Elmer
(Boston, MA). Rat brain membranes (200 g protein per tube)
a
7
([125I]
a
-bungarotoxin, BTX) or
a
4
b
2
filters (presoaked for 45 min in 0.5% polyethyleneimine to reduce
nonspecific binding) using a Brandel Harvester; and quickly
washed three times with 3.0 ml of ice-cold binding saline lacking
BSA. Binding data were analyzed using GraphPad PrismÔ software
(San Diego, CA) [19]. The Ki values were calculated from the Cheng–
Prusoff equation [20], using Kd values for each radioligand (0.32 nM
for BTX, 0.92 nM for cytisine) that we determined by saturation
binding analysis under identical conditions. Each reported Ki esti-
mate is a mean value derived from three independent displacement
curve determinations; the mean standard error for these
m
were suspended in receptor binding assay saline containing
120 mM NaCl, 5 mM KCl, 2 mM CaCl2, 1 mM MgCl2, 50 mM Tris
Buffer (pH ¼ 7.4) and 2 mg/ml BSA to reduce nonspecific binding
[9]. Displacement of 1.0 nM BTX (w150 Ci/mmole) was measured
after incubating at 37 ꢀC for 3 h. Binding of 1.0 nM [3H]cytisine
(w1 Ci/mmole) was performed at 0 ꢀC for 4 h. Nonspecific binding
of each radioligand was measured in the presence of 1 mM nico-
tine. After incubation, membranes with bound radioligand were
rapidly collected by vacuum filtration on Whatman GF/C glass fiber
measurements was 14% of the mean value for the
assays and 22% of the mean for 2 receptor assays.
a7 receptor
a4b

S.H. Slavov et al. / European Journal of Medicinal Chemistry 45 (2010) 2433–2446
2439
2.3. Arylidene-anabaseine pKa and Log P determinations
2.5. Molecular docking
The pKa of the THP ring imine nitrogen was determined by
analysis of the pH dependence of the arylidene-anabaseine
(1.3 mM) absorbance spectrum at room temperature in the pres-
ence of 150 mM NaCl and 50 mM potassium phosphate buffer
(pH ¼ 7.4). The 250–600 nm absorbance spectrum at 13 evenly
spaced pHs (range 4–10) was measured using a Beckman DU650
spectrophotometer. Absorbance values at a wavelength showing
the greatest dependence on pH were then entered into the Enzfitter
software [21] to calculate the pKa value. Log P, a widely used
measure of lipophilicity, was determined and calculated as previ-
ously described, using the experimental pKa estimates to correct
for ionization [9].
Software implementing automated docking has to deal with two
conflicting requirements: i) to be based on a robust and physically
relevant procedure and at the same time; ii) to keep the computa-
tional demands at a reasonable level. Two concurrent approaches for
solving this conflict are currently popular: i) docking the ligands as
rigid bodies (or with torsional flexibility) by scanning the protein
surface for shape complementarity; and ii) building detailed ener-
getic models employing exhaustive searches over a more limited
conformational/structural space [26,27]. The AutoDock package [28]
is optimized to produce reasonable results, imposing two limita-
tions: i) no assurance of finding the global energy minima confor-
mation of the ligand, and ii) a rigid protein target [29].
AutoGrid 4.0 [28] was used to construct a three-dimensional
grid with a dimension 60 ꢂ 60 ꢂ 60 Å centered at the
a7 nAChR
2.4. 2D-QSAR modeling
binding site. A probe atom was used to calculate the interaction
energies at each grid point. Separate tables were also calculated for
each atom type in the ligand, including dispersion/repulsion and
hydrogen bonding energies. In addition, an electrostatic potential
grid map was calculated, using a probe atom carrying a single
positive charge. The electrostatic interaction energies for each
ligand atom were obtained during the docking procedure as
a product of the local value taken from this grid and the partial
Gasteiger charge on that atom.
AutoDock 4.0 [28] was used to perform the actual docking.
During the docking simulation the ligand is subjected to random
rotational/translational and torsional perturbations and the inter-
action energy at each new position is calculated. These positions are
either accepted or rejected probabilistically based on the annealing
temperature of the system. The docking simulation starts at high
temperature, accepting almost every move, thus allowing the ligand
to explore as fully as possible the whole conformational space. With
the decrease of the temperature, the unfavorable positions become
increasingly disallowed. Thus, at the end of the docking procedure
the ligand ultimately finds its optimal position and conformation
within the deepest energetic well it has sampled. Starting from
random initial states and performing multiple docking simulations,
consistent favorable binding conformations could be located.
Three-dimensional conversions and pre-optimizations of the
structures of all 47 compounds (Table 1) were performed using
the molecular mechanics (MMþ) method implemented in the
HyperChem 7.5 package [22]. Final geometry optimization of the
molecule was carried out by using the semi-empirical quantum-
mechanical AM1 parameterization [23]. Optimized geometries of
the compounds were then loaded into CODESSA PRO software
[24]. Overall, more than 800 theoretical descriptors were calcu-
lated. An important stage in use of the multilinear regression
QSAR is the search for the best multilinear equation among a given
descriptor set. In other words, the aim is to find the best corre-
lation of the experimental dissociation constants for binding of
ligand to the nAChR (A) with a certain number, n, of molecular
descriptors (Di) weighted by the regression coefficients bi, as
defined by eq. (1):
n
X
A ¼ b0 þ
biDi
(1)
i ¼ 1
The Best Multilinear Regression (BMLR) method encoded in
CODESSA PRO software was used to select significant descriptors
for building multilinear QSAR models [25]. The treatment starts
with reduction in the number of molecular descriptors. If two
descriptors are highly correlated, then only one descriptor is
selected (all descriptors with insignificant variance are rejected).
This helps to speed up the descriptor selection and reduces the
probability of including unrelated descriptors by chance.
2.6. 3D-QSAR
According to the lock and key principle [30], the most active
(in our case, lowest Ki) compound in the series should have
Fig. 1. Statistical distribution of the Ki and Log Ki data for arylidene-anabaseine binding to rat a7 receptors.

2440
S.H. Slavov et al. / European Journal of Medicinal Chemistry 45 (2010) 2433–2446
a near optimal geometry when bound to the receptor (in our
case, the Aplysia californica AChBP). The THP-Benzylidene and
THP-Pyr angles in 3-(4-Aminobenzylidene)anabaseine, the most
for both fields. When a complete probe map was calculated for
each molecule in the target set, the energy values for each point
in the grid were stored in columns of a study table. In the next
step some or all of the grid data points can be used as descriptors
in generating 3D-QSARs and analyzing structure–activity
relationships.
active compound for the a7 receptor in our binding assays, were
set to correspond to their most optimal values for the so-called A
form of DMXBA [17] when bound to nAChBP. In accordance with
the active analog approach, the so obtained geometry of 3-(4-
Aminobenzylidene)anabaseine (structure 1.1, Table 1) was then
used as a template for the alignment of the remaining molecules.
Structure 2.1 (see Table 1) did not match the template, so it was
not considered further. The flexible fitting procedure imple-
mented in Chem-X [31] was employed to superimpose all
molecules from the series (Fig. 2). Our field comparisons with
flexible fitted structures are based on calculated RMS errors. Only
a single conformer for each molecule was used – the one that
fitted the template best.
The next step was to build a 3D grid around the set of
superimposed molecules using a 1 Å grid constant. To calculate
the steric and electrostatic energies of interaction a fictitious ‘‘H’’
probe atom with charge þ1 was used. Then, the interaction
energies between the probe atom and the aligned molecules
were calculated for each grid point, using 0.95 cutoffs by default
3. Results and discussion
3.1. Chemical and pharmacological diversity of the QSAR data set
A number of independent restrictions imposed by the main
QSAR hypothesis sternly limits chemical diversity of the data set
to the compounds that satisfy the following criteria: i) structural
similarity and presence of the same pharmacophore core, ii) the
same/very similar mode of action, and iii) participation in the
same metabolic pathways. Therefore, the desire to construct
highly accurate and statistically reliable QSAR models necessarily
narrows the domain of applicability down to only 3-(arylidene)
anabaseines.
The data set of the 47 arylidene-anabaseines included exper-
imental equilibrium dissociation constants for receptor (specific)
binding to rat
a7 and a4b2 nAChRs, and additionally, compound
ionization (pKa) and octanol–water distribution (Log P) estimates
(Table 7). The receptor affinities of the compounds varied over
200-fold for the
a7 receptor and 83-fold for the a4b2 subtype
nAChR. Imine nitrogen basicity (pKa) values varied over 3 pH
units and the Log P values varied over 2 logarithmic (base 10)
units.
To improve the normal distribution of the data for the QSARs,
the original Ki values obtained for both neuronal nAChRs were
transformed using a logarithmic function. The Log Kis for both the
a7 (Fig. 1) and a4b2 Ki values possessed distributions close to
normal. The initial data set of 47 compounds was randomly split
into two subsets, conditionally denoted training and test (marked
with an asterisk in Table 7) subsets.
3.2. 2D-QSAR modeling
Previous studies indicated the importance of imine nitrogen
group pKa for understanding of arylidene-anabaseine binding to
the alpha-7 receptor [9,32] and to some extent, variations in the
experimental Kis between compounds in a more limited data set
[33]. For most of the compounds in the training subset, pKa (and
Log P) values were experimentally determined [33]. Hence,
a quantitative structure–property relationship (QSPR) modeling
procedure was applied to estimate the missing values, which are
shown in italics in Table 7. The quality of the estimation was eval-
uated by using an external test set of 10 compounds. Utilizing the
‘‘principle of parsimony’’ (‘‘Ockham’s Razor’’), models with 3
descriptors were preferred as optimal (see Tables 2 and 3 and the
corresponding Figs. 3 and 4).
Table 2
QSPR Log P model.
X
D
X
t-test
4.019
Descriptor
0
1
2
162.9
ꢁ12.19
40.53
1.877
1.693
Intercept
ꢁ6.493
H-donors FPSA (version 2)
Maximum antibonding
contribution of one MO
Maximum e–n attraction
for bond C–C
6.314
3.729
3
ꢁ0.5627
0.1560
ꢁ3.607
Fig. 2. Aligned molecules from the series.

S.H. Slavov et al. / European Journal of Medicinal Chemistry 45 (2010) 2433–2446
2441
N ¼ 32; R2 ¼ 0:713; Rc2v ¼ 0:637; R2test ¼ 0:593;
F ¼ 23:2; s2 ¼ 0:0685
Table 3
QSPR pKa model.
X
D
X
t-test
4.849
Descriptor
0
1
9.693
1.999
Intercept
0.3045
0.06057
5.027
Charged Surface
Area for atom H
Number of O atoms
HOMO energy
2
3
ꢁ0.4013
0.09020
0.1833
ꢁ4.450
0.5438
2.966
N ¼ 34; R2 ¼ 0:757; Rc2v ¼ 0:680; R2test ¼ 0:726;
F ¼ 31:2; s2 ¼ 0:0813
At the second stage of our study, the experimentally available
Log P and pKa values, together with those estimated (using the
models of Tables 2 and 3), were imported into CODESSA Pro as
external descriptors. Using a training set of 36 compounds (35 for
the
generated. To identify the structural characteristics affecting the
binding to both 7 and 2 nAChRs subtypes their corresponding
a4b2 receptor data), models with up to 6 descriptors were
Fig. 3. Predicted vs observed Log P values.
a
a4b
and geometry was also revealed. Comparison of the models in
Tables 4 and 5 reveals slightly different interactions governing the
binding to the two nAChR subtypes. While the descriptors involved
Log Ki values were modeled. While none of the derived BMLR
models contained either Log P or pKa as descriptors, further
examination of the correlation matrix revealed a significant rela-
tionship between pKa and the ‘‘Minimum partial charge (Zefirov)
for atom N’’ descriptor (R ¼ ꢁ0.67) involved in the model
in the
a4
b
2 model are purely electrostatic in nature, those that
7 reflect also the importance of
were most important in the case of
a
steric complimentarity for the stability of the ligand–receptor
complex. This difference is consistent with a recent molecular
modeling analysis of the ACh binding sites in these two receptors,
based on the Lymnaea AChBP crystal structure, which reported that
describing the binding to the a7 nAChR.
Table 4
Best 4-parameter model for Log Ki (a7).
the negative electric charge of the
than for the 7 ACh binding site [34].
In agreement with our primary goal, to understand and even-
tually optimize arylidene-anabaseine interaction with the
nAChR, this article is focused on arylidene-anabaseine binding to
this rat nAChR. A 2D-QSAR model for the 2 subtype receptor is
given for comparative purposes and for possibly providing a guide
to minimizing binding (arylidene-anabaseines generally
a4b2 ACh binding site was higher
X
D
X
t-test
3.908
ꢁ3.944
Descriptor
a
0
1
7.889
2.019
0.02311
Intercept
Structural Information
content (order 2)
Minimum partial charge
(Zefirov) for atom N
Shadow plane ZX
Maximum nucleophilic
reactivity index for atom N
ꢁ0.09115
a7
2
79.83
20.75
3.846
a4b
3
4
0.03835
47.74
0.01157
16.27
3.316
2.934
a4b2
N ¼ 36; R2 ¼ 0:665; Rc2v ¼ 0:557; R2test ¼ 0:398;
F ¼ 15:4; s2 ¼ 0:0911
Table 5
Best 4-parameter model for Log Ki (a4b2).
X
DX
t-test
Descriptor
0
1
2
3
4
ꢁ148.1
43.51
1.502
4.311
0.06814
0.3965
ꢁ3.405
4.747
Intercept
7.128
Max n–n repulsion for bond H–C
Max total interaction for bond H–C
Min atomic state energy for atom N
LUMO þ 1 energy
ꢁ14.21
0.2220
ꢁ1.162
ꢁ3.297
3.258
ꢁ2.930
N ¼ 35; R2 ¼ 0:650; Rc2v ¼ 0:532; R2test ¼ 0:718;
F ¼ 14:0; s2 ¼ 0:0829
Detailed analysis of the derived models (see Tables 4 and 5, and
Figs. 5 and 6) indicates the importance of quantum-chemical
charge distribution-related descriptors to account for the binding
to both
a
7 and
a
4b
2 nAChRs. The significance of the ligand topology
Fig. 4. Predicted vs observed pKa values.

2442
S.H. Slavov et al. / European Journal of Medicinal Chemistry 45 (2010) 2433–2446
Fig. 6. Predicted vs observed a4b2 nAChR Log Ki values.
ꢁ0.0987. Due to the positive regression coefficient sign of the
descriptor, values in the lower range will affect the property
positively. Closer examination revealed that the amino-
benzylidene-anabaseines displayed some of the most negative
nitrogen partial charge values, and thus were highly active. A
possible explanation for their high activity could be their high
propensity for formation of hydrogen bonds. The importance of
the nitrogen net atomic charge was also demonstrated by Sultana
et al. [35] who studied a limited data set of 9 monosubstituted
anabaseines by Hansch-type approach relating pKi (for neuronal
nicotinic receptors of the cockroach, Periplaneta americana) to
electrostatic and steric descriptors. In agreement with our
conclusions it was shown that the affinity of the compounds
increases as the charge of the imine nitrogen atom becomes more
negative.
Fig. 5. Predicted vs observed a7 Log Ki values.
inhibit this receptor subtype; since stimulation of a4b2 receptors
contributes to normal cognitive function, their inhibition would be
undesirable). As already mentioned above, two of the descriptors
(‘‘Structural Information content, order 2’’ and ‘‘Shadow plane ZX‘‘)
[25] are of topological and geometrical origin, respectively. The
former depends in a complex way on the size and branching of
the molecule, its flexibility and spatial mass distribution, while the
latter emphasizes the significance of the orientation towards the
target. The negative regression coefficient sign of the ‘‘structural
information content’’ descriptor indicates that the increase of the
molecular flexibility will result in more efficient binding to nAChR.
All (‘‘Minimum partial charge (Zefirov) for atom N’’) [25]
descriptor values were found to lie in the range ꢁ0.119 to
Fig. 7. Model of arylidene-anabaseine binding to the Aplysia californica AChBP, with all structures superimposed.

S.H. Slavov et al. / European Journal of Medicinal Chemistry 45 (2010) 2433–2446
2443
Table 6
Principal side amino-acids participating in
p
–
p
and H-bonding interactions.
Principal side
(literature [17])
Principal side
(docking)
Frequency of
occurrence in percents
Complimentary side
(literature [17])
Complimentary side
(docking)
Frequency of occurrence
in percents
Tyr93
Tyr91
75
90
85
70
70
Tyr55
Tyr53
70
35
35
80
35
Trp147
Tyr186
Cys190
Tyr193
Trp145
Tyr186
Cys188
Tyr193
Val108
Met116
Ile118
Val106
Met114
Ile116
Asp164
Asp162
According to its t-criterion, the third important descriptor is
‘‘Shadow plane ZX’’. It highlights the importance of the molecular
size and volume, factors crucial for the insertion of the ligand into
the binding pocket. The positive regression coefficient sign implies
that a size over a certain threshold will hinder the access to the
nAChR cleft.
orbitals and the HOMO energy of the molecule. Due to its complex
nature, it was difficult to establish a direct relationship between
certain molecular structural features accounted for by this
descriptor and the experimental Ki values. Aside from any further
speculations, the most unambiguous interpretation suggests the
participation of donor–acceptor interactions.
The least significant descriptor in the model of Table 4 is the
‘‘Maximum nucleophilic reactivity index for atom N’’ [25]. It
depends in a complex way on the ci coefficients of the atomic
Internal and external validation procedures were applied in
order to estimate the predictive power of the derived models. As
can be seen from Tables 4 and 5, both models are characterized by
Table 7
Experimental and calculated binding energies and Ki values.
ID
a
7 (experimental Ki in
m
M)
2D-QSAR (Log Ki) Ki Docking (
a
4
b
2 (experimental Ki in
m
M)
Log P pKa
Ki ( M) Log Ki
m
m
M) Binding E Docking (kCal molꢁ1
)
Ki (
m
M)
Log Ki
2D-QSAR (Log Ki)
1.1
0.05 ꢃ 0.003
2.23 ꢃ 0.48
0.93 ꢃ 0.19
0.05 ꢃ 0.001
0.36 ꢃ 0.02
0.57 ꢃ 0.03
0.436a
0.25 ꢃ 0.003
2.30 ꢃ 0.08
0.28 ꢃ 0.02
0.17 ꢃ 0.006
ꢁ1.301 ꢁ1.141
0.348 ꢁ0.180
ꢁ0.032 ꢁ0.015
ꢁ1.301 ꢁ1.222
ꢁ0.444 ꢁ0.486
ꢁ0.244 ꢁ0.480
ꢁ0.361 ꢁ0.092
ꢁ0.602 ꢁ0.550
0.882
ꢁ8.26
0.12 ꢃ 0.03
2.17 ꢃ 0.24
0.93 ꢃ 0.09
0.05 ꢃ 0.03
0.45 ꢃ 0.04
0.49 ꢃ 0.03
0.99 ꢃ 0.12
0.90 ꢃ 0.03
1.23 ꢃ 0.40
0.89 ꢃ 0.13
0.25 ꢃ 0.01
0.28 ꢃ 0.09
0.17 ꢃ 0.005
1.16 ꢃ 0.25
0.97 ꢃ 0.38
0.52 ꢃ 0.19
0.38 ꢃ 0.013
0.16 ꢃ 0.01
0.83 ꢃ 0.25
4.49 ꢃ 2.3
ꢁ0.921
0.336
ꢁ0.731
ꢁ0.169
0.113
ꢁ0.837
ꢁ0.220
ꢁ0.264
ꢁ0.179
ꢁ0.162
0.098
2.43
3.18
2.44
3.11
2.29
3.17
3.12
3.17
3.11
2.82
2.56
2.72
3.63
3.39
3.50
2.40
2.36
2.57
3.09
2.51
2.68
2.95
2.50
3.07
3.34
3.17
2.72
3.43
1.97
3.33
3.67
2.46
2.96
3.02
2.78
3.12
3.42
3.07
3.16
2.99
3.38
3.86
3.21
4.10
2.46
2.70
1.75
8.33
7.45
6.84
8.54
7.22
7.43
7.33
7.49
6.72
7.39
7.14
7.97
8.66
7.35
7.18
7.06
7.42
8.66
7.12
6.83
7.67
6.28
5.76
6.94
6.48
7.30
7.16
7.67
7.67
7.71
7.48
7.00
7.51
7.14
7.29
7.19
7.15
7.23
7.56
7.64
7.29
7.11
7.34
7.89
7.09
7.07
7.41
1.2*
1.3
0.960
1.970
ꢁ8.21
ꢁ7.78
ꢁ0.032
ꢁ1.301
ꢁ0.347
ꢁ0.310
ꢁ0.004
ꢁ0.046
0.090
ꢁ0.051
ꢁ0.602
ꢁ0.553
ꢁ0.770
0.064
ꢁ0.013
ꢁ0.284
ꢁ0.420
ꢁ0.796
ꢁ0.081
0.652
1.4*
1.5
1.6
1.7*
1.8
1.9
1.10
1.11
0.335
ꢁ8.83
0.362
0.179
ꢁ0.553 ꢁ0.398
ꢁ0.249
ꢁ0.770 ꢁ0.104
0.302
ꢁ8.9
0.037
1.12* 0.30 ꢃ 0.0002 ꢁ0.523 ꢁ0.497
ꢁ0.713
ꢁ0.945
0.037
ꢁ0.079
ꢁ0.519
ꢁ0.911
ꢁ0.759
ꢁ0.169
0.108
ꢁ0.215
ꢁ0.203
0.083
ꢁ0.050
ꢁ0.057
ꢁ0.221
ꢁ0.099
N/A
ꢁ0.237
ꢁ0.776
ꢁ0.228
ꢁ0.163
ꢁ0.905
ꢁ0.370
ꢁ0.231
ꢁ0.437
ꢁ1.107
ꢁ0.142
ꢁ0.205
ꢁ0.027
ꢁ0.174
ꢁ0.839
ꢁ0.165
ꢁ0.511
ꢁ0.716
ꢁ0.020
0.751
1.13
1.14
1.15
0.08 ꢃ 0.0004 ꢁ1.097 ꢁ1.034
0.811
1.170
3.800
ꢁ8.31
ꢁ8.09
ꢁ7.39
1.01 ꢃ 0.18
1.12 ꢃ 0.19
0.004 ꢁ0.011
0.049 0.030
0.072 ꢁ0.033
0.029
1.16* 1.18 ꢃ 0.13
1.17* 0.24 ꢃ 0.03
ꢁ0.620
0.738
1.740
0.450
1.120
0.582
ꢁ8.37
ꢁ7.86
ꢁ8.66
ꢁ8.12
ꢁ8.51
1.18
1.19
1.20
1.21
1.22
1.23
1.24
0.19 ꢃ 0.04
0.69 ꢃ 0.11
0.92 ꢃ 0.04
0.91 ꢃ 0.02
0.57 ꢃ 0.04
3.09 ꢃ 1.88
1.07 ꢃ 0.04
ꢁ0.721 ꢁ1.079
ꢁ0.161 ꢁ0.441
ꢁ0.036 ꢁ0.069
ꢁ0.041 ꢁ0.172
1.96 ꢃ 0.83
0.70 ꢃ 0.36
1.76 ꢃ 0.75
1.41 ꢃ 0.01
1.44 ꢃ 0.05
1.18 ꢃ 0.10
1.17 ꢃ 0.10
N/A
1.25 ꢃ 0.47
0.17 ꢃ 0.02
0.43 ꢃ 0.16
0.77 ꢃ 0.05
0.13 ꢃ 0.04
0.67 ꢃ 0.04
0.36 ꢃ 0.01
0.52 ꢃ 0.22
0.06 ꢃ 0.03
4.67 ꢃ 1.06
0.52 ꢃ 0.01
0.70 ꢃ 0.07
0.85 ꢃ 0.24
0.25 ꢃ 0.04
0.07 ꢃ 0.030
0.39 ꢃ 0.03
0.21 ꢃ 0.016
0.60 ꢃ 0.10
5.92 ꢃ 1.06
0.292
ꢁ0.244
0.490
0.029
0.039
0.200
0.000
ꢁ0.155
5.210
5.350
10.450
1.780
3.360
ꢁ7.21
ꢁ7.19
ꢁ6.8
0.246
0.149
0.158
0.072
0.068
N/A
0.097
1.25* 0.65 ꢃ 0.11
ꢁ0.187 ꢁ0.088
ꢁ0.337 ꢁ0.668
ꢁ0.824 ꢁ0.157
0.637 ꢁ0.143
ꢁ0.745 ꢁ0.729
ꢁ1.155 ꢁ0.921
ꢁ1.301 ꢁ0.801
ꢁ0.523 ꢁ0.397
ꢁ1.301 ꢁ1.272
0.740 ꢁ0.255
ꢁ0.276 ꢁ0.270
0.179 ꢁ0.152
ꢁ0.745 ꢁ0.642
0.542 ꢁ0.068
ꢁ0.699 ꢁ0.636
ꢁ0.638 ꢁ0.735
1.26
1.27
1.28
0.46 ꢃ 0.12
0.15 ꢃ 0.02
4.34 ꢃ 0.63
ꢁ7.84
ꢁ7.47
1.29* 0.18 ꢃ 0.02
1.30
1.31
1.32
1.33
0.07 ꢃ 0.03
0.05 ꢃ 0.002
0.30 ꢃ 0.09
0.05 ꢃ 0.002
ꢁ0.770
ꢁ0.367
ꢁ0.114
ꢁ0.886
ꢁ0.174
ꢁ0.444
ꢁ0.284
ꢁ1.222
0.669
ꢁ0.284
ꢁ0.155
ꢁ0.071
ꢁ0.602
ꢁ1.155
ꢁ0.409
ꢁ0.678
ꢁ0.222
0.772
1.34* 5.49 ꢃ 2.97
1.35
1.36
1.37
0.53 ꢃ 0.11
1.51 ꢃ 0.11
0.18 ꢃ 0.03
1.520
75.910
ꢁ7.94
ꢁ5.62
1.38* 3.48 ꢃ 2.10
155.
0.603
ꢁ5.2
1.39
1.40
1.41
0.20 ꢃ 0.01
0.23 ꢃ 0.002
0.60 ꢃ 0.08
ꢁ8.48
ꢁ0.222
0.114
1.42* 0.13 ꢃ 0.01
ꢁ0.886 ꢁ0.332
ꢁ0.620 ꢁ0.801
ꢁ0.495 ꢁ0.732
ꢁ0.699 ꢁ0.724
ꢁ0.432 ꢁ0.442
ꢁ0.824 ꢁ0.678
1.43
1.44
1.45
2.1
0.24 ꢃ 0.01
0.32 ꢃ 0.07
0.20 ꢃ 0.01
0.37 ꢃ 0.03
0.15 ꢃ 0.003
2.2
0.293
ꢁ8.91
The molecules that are part of the test set are marked with an asterisk; the values listed in italic are estimated on the basis of the models shown in Tables 2 and 3.
a
This mean value was obtained from two experiments with individual Ki values of 0.413 and 0.459 mM.

2444
S.H. Slavov et al. / European Journal of Medicinal Chemistry 45 (2010) 2433–2446
Fig. 8. 3-(4-Difluoromethoxybenzylidene)anabaseine bound to the a7 binding site.
the small differences between R2, Rc2v and Rt2est, which is an indi-
cation of their robustness and potential utility for predicting the
binding affinities of untested compounds.
AChBP (PDB identification code ¼ 2BR7) is almost identical with
the one known from the X-ray crystal structure of DMXBA and 4-
OH-DMXBA bound to the same protein. Ligand interactions with
both sides, principal and complimentary, of the ACh binding site
were studied. The amino acid residues of the principal side
3.3. Molecular docking
involved in
p–p interactions and hydrogen bonding (see Fig. 8),
together with their corresponding frequencies of occurrence, are
shown in Table 6. It can be seen that these derived results are in
reasonable agreement with the crystallographic data for
arylidene-anabaseine binding to the Aplysia AChBP [17] and with
earlier reported docking studies. The results obtained by Huang
et al. [34] using Lymnaea stagnalis AChBP emphasized the
importance of Tyr91, Cys189 and Trp148 on the principal side; at
a qualitative level these agree with our results. Similar positions
were reported by Bisson et al. [37] in their docking study with
Lymnaea stagnalis AChBP. Molecular modeling studies of binding
of DMXBA and its metabolites performed on the same protein
The increased interest towards ligand binding to nAChRs
resulted in significant number of molecular docking related publi-
cations during the past decade [34,36–39]. Due to the unavailability
of X-ray crystallographic data of human nAChRs, different modeling
AChBP systems are currently in use. Despite rather large differences
in the amino acid sequences of the overall structures, sequence
homology within the ligand-binding domain is relatively high.
Mutations in either the nAChR or the AChBP at homologous sites
within the binding sites generally produce very similar effects on
ligand binding, which provides the basis for considering the
AChBPs as surrogate nAChR models.
The available crystal structures of both anabaseine and DMXBA
(compound 1.42, Tables 1 and 7) when bound to the AChBP indicate
that at least one water molecule mediates the binding of the ligand
to the protein [17]. However, the docking procedure we used
recommends the removal of the water molecules from the binding
site. Thus, in our case, the screening was performed for an empty
active site devoid of water. In order to achieve close agreement with
the available experimental X-ray data we focused our efforts on
obtaining a self-consistent picture of the intermolecular ligand–
receptor interactions. Therefore, as demonstrated by Fig. 7, only
those conformers that formed a domain of structures with an
approximately similar axial orientation and occupied the same
preferred spatial position within the ligand binding pocket were
selected. The latter also happened to be the conformers with the
highest binding energy. Due to the excessive computational time
needed, our calculations were restricted to a subset representative
of the general population. The obtained binding energies and the
experimental and calculated affinity constants for binding of the
ligands to the protein are listed in Table 7.
A frequency of occurrence analysis was performed in order to
identify amino acid residues in the Aplysia AChBP which directly
interact with the arylidene-anabaseine ligand (Table 6). It is
important to notice that the arylidene-anabaseine binding site
(Fig. 7) identified by our docking calculations using Aplysia
Fig. 9. Predicted vs experimental Log Ki values for a7 nAChR binding for the electro-
static fields.

S.H. Slavov et al. / European Journal of Medicinal Chemistry 45 (2010) 2433–2446
2445
structure revealed the significance of Tyr93, Trp149, Tyr188,
Cys190 and Tyr195 [39].
Two orientations of DMXBA were observed in its crystal structure
with Aplysia AChBP. The B orientation, being similar to that of the a7
nAChR antagonist methyllycaconitine (MLA) orientation, is thought
to represent a less efficacious orientation [17]. On the other hand, the
A orientation, which is also observed for the more efficacious 4-OH-
DMXBA, is thought to represent the more efficacious orientation of
DMXBA. The complimentary side contacts which were observed in
our docking are for the most part related to the A orientation.
3.4. 3D-QSAR results
Following the procedure described in Section 2.6, the electro-
static and steric energies of interaction were studied. A two-pass-
PLS method was used to obtain the 3D-QSAR model. The first
step in this method calculates the principal components and the
second one yields the model and its statistical parameters. For both
the steric and electrostatic fields, six principal components were
generated.
A separate covariance matrix was constructed from the data
table. PLS determines the full eigensystem of this matrix using
a singular value decomposition (SVD) method. The matrix was
rewritten in a new form, and the coefficients of the transformed
variables (the principal component loadings) were evaluated and
used to determine the principal component scores (a set of new
variables which best explain the variance in the data). The first six
principle components explained 81.3% and 70.0% of the data vari-
ance for the electrostatic and steric fields, respectively.
To evaluate the external predictive power of the model, the
initial data set was divided randomly into two parts (training and
test subsets) having similar distribution of the receptor affinity
data. In our case, the test set contains the following compounds:
1.2, 1.4, 1.7, 1.12, 1.16, 1.17, 1.25, 1.29, 1.34, 1.38 and 1.42.
Cross-validation PLS calculations were performed to validate the
models. The calculations were repeated for 10 sample sets and the
results were then averaged. Each sample set was constructed from
randomly selected compounds in the training subset. The size of
each sample sets was specified as 10% of the total data set and each
compound was omitted only once.
Fig. 10. WLS map of the regions important for the electrostatic fields.
following statistical parameters: R2 ¼ 0.737, Ra2dj ¼ 0.680,
Q2 ¼ 0.399, F ¼ 13.051 and a std. error of estimate: 0.282:
Log Ki ¼ 0:038*PC1 þ 0:032*PC2 þ 0:032*PC3 þ 0:035*PC4
þ 0:020*PC5 þ 0:023*PC6 ꢁ 0:390
ð3Þ
A chart of the predicted vs experimental Log Ki values is shown
in Fig. 11. The external predictive power of the model was again
evaluated by a test series and an external predictive R2test ¼ 0.816
was obtained.
Comparison of the models for the steric and electrostatic fields
suggests that the electrostatic interactions play a more important
role in ligand binding to the
a7 nAChR. The WLS map shown in
Fig. 12 indicates important regions for steric interaction in the
3.4.1. Results for the electrostatic fields
By using the calculated PC components, a PLS model for the
electrostatic interactions with the following statistical parameters
was obtained: R2 ¼ 0.885, Ra2dj ¼ 0.860, Q2 ¼ 0.652, F ¼ 35.939 and
a std. error estimate of 0.186.
Log Ki ¼ 0:078*PC1 þ 0:026*PC2 þ 0:033*PC3 þ 0:021*PC4
þ 0:018*PC5 þ 0:030*PC6 ꢁ 0:397
ð2Þ
A chart of the predicted vs experimental Log Ki values is shown
in Fig. 9. The external predictive power of this model was evaluated
using the above defined test data set. The correlation coefficient of
this prediction was Re2xt ¼ 0.924. By using the WLS (weighted least
squares) [40] method, R2 maps were generated separately for the
electrostatic and steric fields. The portions of the normal coordinate
space where the variation in map data values relates to activity
were visualized (see Fig. 10). An R-squared map was generated by
discarding those points which did not correlate with the activity,
and storing the R-squared value of the correlation for each
remaining point.
3.4.2. Results for the steric fields
Similarly to the electrostatic fields, for the steric fields the
generated PC components produced a QSAR model (Eq. 2) with the
Fig. 11. Predicted vs experimental Log Ki values for a7 nAChR binding for the steric
fields.

2446
S.H. Slavov et al. / European Journal of Medicinal Chemistry 45 (2010) 2433–2446
Acknowledgments
We thank Kelly MacDougall and Renee Gallo for carrying out
some of the nicotinic receptor binding experiments. The research
was funded by NIH grant MH-061412 (R. Freedman and W.R. Kem).
References
[1] D.B. Kitchen, H. Decornez, J.R. Furr, J. Bajorath, Nature Rev. Drug. Discov. 3
(2004) 935–949.
[2] N. Le Nove`re, P.J. Corringer, J.P. Changeux, J. Neurobiol. 53 (2002) 447–456.
[3] N. Unwin, J. Mol. Biol. 346 (2005) 967–989.
[4] R. Peto, A.D. Lopez, J. Boreham, M. Thun Jr., C. Health, Lancet 339 (1992)
1268–1278.
[5] W.R. Kem, Behav. Brain. Res. 113 (2000) 169–181.
[6] L.F. Martin, W.R. Kem, R. Freedman, Psychopharmacology 174 (2004)
54–64.
[7] C.M. de Fiebre, E.M. Meyer, J.C. Henry, S.I. Muraskin, W.R. Kem, R.L. Papke, Mol.
Pharmacol. 47 (1995) 164–171.
[8] W.R. Kem, V.M. Mahnir, R.L. Papke, C.J. Lingle, J. Pharmacol, Exp. Ther. 283
(1997) 979–992.
[9] W.R. Kem, V.M. Mahnir, L. Prokai, R.L. Papke, X. Cao, S. LeFrancois, K. Wildeboer,
K. Prokai-Tatrai, J. Porter-Papke, F. Soti, Mol. Pharmacol. 65 (2004) 56–67.
[10] E.J. Martin, K.S. Panikar, M.A. King, M. Deyrup, B. Hunter, G. Wang, E.M. Meyer,
Drug Dev. Res. 31 (1994) 134–141.
[11] G.W. Arendash, G.J. Sengstock, R. Sanberg, W.R. Kem, Brain Res. 674 (1995)
252–259.
Fig. 12. WLS map of the regions important for the steric fields.
[12] A. Olincy, J.G. Harris, L.L. Johnson, V. Pender, S. Kongs, D. Allensworth, J. Ellis,
G.O. Zerbe, S. Leonard, K.E. Stevens, J.O. Stevens, L. Martin, L.E. Adler, F. Soti,
W.R. Kem, R. Freedman, Arch. Gen. Psychiatry 63 (2006) 630–638.
[13] R. Freedman, A. Olincy, R.W. Buchanan, J.G. Harris, J.M. Gold, L. Johnson,
D. Allensworth, A. Guzman-Bonilla, B. Clement, M.P. Ball, J. Kutnick, V. Pender,
L.F. Martin, K.E. Stevens, B.D. Wagner, G.O. Zerbe, F. Soti, W.R. Kem, Am. J.
Psychiat. 165 (2008) 1040–1047.
[14] K. Brejc, W.J. van Dijk, R.V. Klaassen, M. Schuurmans, J. van Der Oost, A.B. Smit,
T.K. Sixma, Nature 411 (2001) 269–276.
[15] P.H. Celie, S.E. van Rossum-Fikkert, W.J. van Dijk, K. Brejc, A.B. Smit, T.K. Sixma,
Neuron 41 (2004) 907–914.
[16] S.B. Hansen, T.T. Talley, Z. Radic, P. Taylor, J. Biol. Chem. 279 (2004)
24197–24202.
[17] R.E. Hibbs, G. Sulzenbacher, J. Shi, T.T. Talley, W.R. Kem, P. Taylor, P. Marchot,
Y. Bourne, EMBO J. 28 (2009) 3040–3051.
[18] J.A. Zoltewicz, K. Prokai-Tatrai, L.B. Bloom, W.R. Kem, Heterocycles 35 (1993)
171–179.
[19] http://www.graphpad.com/prism/Prism.htm.
unexplored parametric space around the set of superimposed
molecules.
The WLS electrostatic map (Fig. 12) identifies several regions
which are important for ligand binding: positions 2 and 4 of the
benzylidene moiety, and a region around the THP N atom. As can be
seen in the figure, the most important regions for the steric fields
are again located at positions 2 and 4 of the benzylidene moiety.
Simultaneous consideration of the two maps suggests that bulky
substituents at positions 2 (and due to the rotational freedom also
at position 6) and 4 of the benzylidene moiety, with highly elec-
tronegative atoms at approximate distance of 3–3.5 Å projecting
away from the aromatic system (at position 4) will increase binding
affinity. According to the electrostatic map shown in Fig.10, the THP
N atom probably participates in a hydrogen bond with the receptor,
as has been observed in previous experimental studies with
DMXBA, 4-OH-DMXBA [17] and other nicotinic agonists [15,16].
[20] Y. Cheng, W.H. Prusoff, Biochem. Pharmacol. 22 (1973) 3099–3108.
[21] http://www.biosoft.com/w/enzfitter.htm.
[22] Hyperchem, v. 7.5, Hypercube Inc., Gainesville, FL.
[23] M.J.S. Dewar, E.G. Zoebisch, E.F. Healy, J.J.P. Stewart, J. Am. Chem. Soc. 107
(1985) 3902–3909.
[24] CODESSA PRO Software. University of Florida, 2002.
[25] M. Karelson, Melecular Descriptors in QSAR/QSPR. Wiley-Interscience, New
York, 2000.
4. Conclusions
[26] T.P. Lybrand, Curr. Opin. Struct. Biol. 5 (1995) 224–228.
[27] R. Rosenfeld, S. Vajda, C. DeLisi, Annu. Rev. Biophys. Biomol. Struct. 24 (1995)
677–700.
Our QSAR and docking computations indicate the importance of:
i) the nitrogen atom of the THP ring for hydrogen bond formation
with Trp145 in Aplysia AChBP (and by inference, Trp147 in the
a
7
[28] M.F. Sanner, J. Mol. Graphics Mod. 17 (1999) 57–61.
[29] D.S. Goodsell, A.J. Olson, Proteins 8 (1990) 195–202.
[30] H. Kubinyi, Pharm. Acta Helv. 69 (1995) 259–269.
[31] Chem-X. Chemical Design Ltd. (now part of Accelrys), 1994.
[32] T.T. Talley, S. Yalda, K.Y. Ho, Y. Tor, F.S. Soti, W.R. Kem, P. Taylor, Biochemistry 45
(2006) 8894–8902.
[33] S.E. LeFrancois, A structure activity investigation of benzylidene anabaseine
interactions with the alpha7 nicotinic acetylcholine receptor, Ph.D. disserta-
tion, Univ. Florida, 2004, pp. 1–123.
[34] X. Huang, F. Zheng, X. Chen, P.A. Crooks, L.P. Dwoskin, C.-G. Zhan, J. Med.
Chem. 49 (2006) 7661–7674.
[35] I. Sultana, I. Ideda, Y. Ozoe, Bioorg. Med. Chem. 10 (2002) 2963–2971.
[36] X. Huang, F. Zheng, C. Stokes, R.L. Papke, C.-G. Zhan, J. Med. Chem. 51 (2008)
6293–6302.
nAChR); ii) interactions between the two aromatic rings of the
p–p
ligand and the nAChBP binding site; and iii) molecular surface
recognition expressed in terms of steric complimentarity. On the
basis of the 3D-QSAR results, bulky substituents at positions 2 (and
due to the rotational freedom also at position 6) and 4 of the ben-
zylidene moiety, with highly electronegative atoms at approximate
distance of 3–3.5 Å from the benzylidene ring (at position 4) seem
optimal for enhancing the binding affinity to the a7 receptor subtype.
This initial 3-(arylidene) anabaseine QSAR analysis has focused
on identifying chemical properties that determine equilibrium
[37] W.H. Bisson, G. Westera, P.A. Schubiger, L. Scapozza, J. Mol. Model. 14 (2008)
891–899.
binding to the a7 nAChR. Future QSAR investigations will include
a greater diversity of arylidene-anabaseine analogs and will focus
on chemical properties that determine other important pharma-
cological properties in addition to receptor binding affinity, such as
activation potency (EC50) and apparent efficacy (maximal activation
relative to maximal ACh effect).
[38] P.H.N. Celie, S.E. Rossum-Fikkeret, W.J. Dijk, K. Brejc, A.B. Smit, T.K. Sixma,
Neuron 41 (2004) 907–914.
[39] D.-Q. Wei, S. Sirois, Q.-S. Du, H.R. Arias, K.-C. Chou, Biochem. Biophys. Res.
Commun. 338 (2005) 1059–1064.
[40] B. Hudson, D.J. Livingstone, E. Rahr, J. Comp-Aided Mol. Des.
55–65.
3 (1989)