Accurate prediction of the bound

conformation of galanthamine in the

active site of Torpedo californica
acetylcholinesterase using
molecular docking
Christian Pilger,* Cecilia Bartolucci,† Doriano Lamba,†
Alexander Tropsha,‡ and Gregor Fels*

*Universitaet-GH Paderborn, Chemie und Chemietechnik, Paderborn, Germany †Istituto di Strut-

turistica Chimica. Consiglio Nazionale delle Ricerche, Area della Ricerca di Roma, Roma, and
Italy International Centre for Genetic Engineering and Biotechnology, Area Science Park, Trieste,
Italy ‡The Laboratory for Molecular Modeling, Division of Medicinal Chemistry and Natural
Products, School of Pharmacy, University of North Carolina, Chapel Hill, North Carolina, USA

The alkaloid (⫺)-galanthamine is known to produce significant

INTRODUCTION
improvement of cognitive performances in patients with the Alz- Nicotinic cholinergic neurotransmission is known to be inti-
heimer’s disease. Its mechanism of action involves competitive mately associated with the recognition, learning, and memory.1
and reversible inhibition of acetylcholinesterase (AChE). Herein, Degeneration of cholinergic neurons in the basal forebrain
we correctly predict the orientation and conformation of the causes a neuropathological disorder known as Alzheimer’s
galanthamine molecule in the active site of AChE from Torpedo disease (AD).2 It has been shown that the density/activity of
californica (TcAChE) using a combination of rigid docking and brain nicotinic acetylcholine receptors (nAChR) is substan-
flexible geometry optimization with a molecular mechanics force tially reduced in AD patients compared with a control group of
field. The quality of the predicted model is remarkable, as indi- the same age.3 These abnormalities are closely associated with
cated by the value of the RMS deviation of ⬃0.5Å when compared increased levels of neuritic plaques and neurofibrillary tangles
with the crystal structure of the TcAChE-galanthamine complex. A that are usually considered the primary histopathological
molecular model of the complex between TcAChE and a galan- changes in AD patients.4
thamine derivative, SPH1107, with a long chain substituent on the Based on these well-established findings, the majority of
nitrogen has been generated as well. The side chain of this ligand current drug therapeutic approaches to AD follow the cholin-
is predicted to extend along the enzyme active site gorge from the ergic hypothesis. These approaches are aimed at elevating the
anionic subsite, at the bottom, to the peripheral anionic site, at the transient levels of acetylcholine in the brain that may be
achieved by inhibiting acetylcholinesterase with reversible in-
top. The docking procedure described in this paper can be applied
hibitors.5–7 Inhibition of AChE is considered one of the most
to produce models of ligand-receptor complexes for AChE and
promising strategies for the treatment of AD and related dis-
other macromolecular targets of drug design. © 2001 by Elsevier
eases.8,9 There are also possible therapeutic applications in the
Science Inc.
treatment of Parkinson’s disease,10 aging,10 and myasthenia
gravis.11 Over the years, hundreds of compounds have been
synthesized and tested for anticholinesterase activity, and sev-
eral of them have found clinical applications.12 The chemical
Color Plates for this article are on pages 374–378.
structures of these inhibitors are very different, ranging from
Corresponding author: Dr. Gregory Fels, Universitaet-GH-Paderborn FB bis-quaternary compounds such as decamethonium (DME),13
13-Org. Chemie, Warburgerstr. 100, D-33098 Paderborn, Germany to simple mono cationic compounds such as edrophonium
E-mail address: [email protected] (EDR),14 and formally neutral tricyclic compounds such as

Journal of Molecular Graphics and Modelling 19, 288–296, 2001

© 2001 by Elsevier Science Inc. 1093-3263/01/$–see front matter
655 Avenue of the Americas, New York, NY 10010 PII S1093-3263(00)00056-5
tacrine (THA).13 Due to its important therapeutic role, the others, the abundance of alternative algorithms and methods
enzyme has been a subject of many recent experimental15–20 stresses that there is no single best procedure applicable in all
and theoretical21–26 investigations. cases.46
Among the different classes of AChE inhibitors reported to In this article, we report an accurate prediction of the bound
date,13,14,27–29 galanthamine (Figure 1) stands out as a very conformation of galanthamine (⬃0.5Å RMS deviation from
potent and promising drug for treatment of the cholinergic the X-ray crystal structure of the TcAChE-galanthamine com-
deficit in AD patients. It exhibits very low toxicity and side plex for all heavy atoms) using an ad hoc docking strategy. The
effects and presently is in the final stages of clinical evalua- work reported herein was accomplished well before the crystal
tion.5 structure of the TcAChE-galanthamine complex was elucidat-
Galanthamine is an amaryllidaceae alkaloid that was first ed,36 and is based on structural information derived from avail-
isolated in Russia30 and structurally elucidated in Japan.31 The able crystal structures of a number of TcAChE-drug com-
difficulty of isolating galanthamine from its natural source plexes. We have used a combination of rigid docking
severely hindered its use as a commercial drug or even as a procedure of AUTODOCK44 with flexible optimization of re-
starting material for the synthesis of derivatives. There are, sulting complexes with the Tripos force field.47 In addition, we
however, a number of total synthesis procedures presently have evaluated the ability of the recently developed SCORE
available, of which those via (⫺)-narwedine32 are the most method48 to select most accurate binding orientation of galan-
convenient and have been extensively employed for large-scale thamine from multiple predictions generated from
industrial preparation.33 In addition, a number of galanthamine AUTODOCK runs and optimized with the Tripos forcefield.
derivatives have been synthesized34 and tested against TcAChE Using this ad hoc methodology, we have also predicted the
and human AchE (HuAChE) to understand the underlying bound conformation of a galanthamine derivative, SPH1107,
mechanism of galanthamine action. which exhibits a long chain substituent at the nitrogen atom of
The successful design of novel potent inhibitors of AChE ring D (see Figure 1).
could be greatly facilitated by rational analysis of experimental The results of this study provide an important insight into the
structure–activity relationships among the enzyme inhibitors mode of esterase inhibition by galanthamine and its deriva-
and accurate prediction of their conformations in the enzyme tives. Furthermore, the success of the strategy employed in this
bound form. Recent X-ray crystallographic analysis of AChE work provides a foundation for future docking studies of ex-
from Torpedo Californica (EC 3.1.1.7)35 followed by X-ray isting and de novo designed galanthamine derivatives. Finally,
determination of the complexes of the enzyme with several the docking strategy described in this paper can be applied to
structurally diverse inhibitors, such as THA,13 EDR,14 E2020,28 other classes of ligands and their pharmacological receptors.
and decamethonium,13 provided crucial information with re-
spect to the orientation of these inhibitors in the active site of METHODS
the enzyme. More recently, the structure of TcAChE galan-
thamine complex was also elucidated by X-ray crystallography The protein crystallographic database49 currently contains
to 2.5 Å resolution (1QTI).36 The crystallographic data indi- seven structures of TcAChE-ligand complexes, including
cated that most of these inhibitors have unique, largely non- 1ACJ (tacrine),13 1ACL (decamethonium),13 2ACK (edropho-
overlapping binding orientations in the active site of the en- nium),14 1VOT (huperzine A),27 1EVE [donepezil (E2020)],28
zyme. The results of the structural analysis of these and several 1AMN (the transition state model m-(N,N,N-trimethylamino)-
other AChE-ligand complexes18,27,29 suggest that accurate pre- trifluoroacetophenone),29 and 1OCE (acylated TcAChE by the
diction of the bound conformation and orientation of the AChE physostigmine analogue MF268).19 The catalytic triad, which
ligands represents a challenging task for molecular modelers. is formed by residues Ser200, His440, and Glu327, is located
Several molecular docking strategies have been developed at the bottom of a 20 Å-deep gorge lined by hydrophobic
over the years, following the seminal work of Kuntz et al.37 residues. The active site of the enzyme is very wide, and the
(e.g., DOCK,38 CAVEAT39, FlexX,40 HAMMERHEAD41, structural analysis of existing complexes indicates that all
BUILDER,42 MSMC,43 AUTODOCK,44 and QXP45). Al- inhibitor molecules interact at the bottom of the gorge, with
though some of these methods have been more popular than additional contacts, for some of them, with residues along the

Figure 1. Structure of galanthamine and the N-alkyl derivative SPH1107.

J. Mol. Graphics Modell., 2001, Vol. 19, No. 3/4 289

gorge. On the basis of these observations and the fact that DOCK program44 as described in the EXPERIMENTAL sec-
galanthamine inhibits AChE via a competitive like mecha- tion using a subset of 42 amino acids that form the complete
nism,50 we first hypothesized that the galanthamine binding site gorge and the active site of TcAChE. The dimensions of the
is also located at the bottom of the gorge. active site box were set to be 22.5 Å ⫻ 22.5 Å ⫻ 22.5 Å to
A second major postulate was made with respect to the side consider all gorge residues, including the part that connects the
chain orientation of Phe330, which is responsible for substrate active site pocket at the bottom of the gorge with the peripheral
trafficking down the gorge (Color Plate 1). From a comparison anionic site at the top (see Color Plate 4). Similar docking run
of the published crystal structures of various TcAChE com- analyses were carried out for the open, half open, and closed
plexes it appears that this residue is the only site of conforma- AchE structures, for both the (eq)- and (ax)-conformers, yield-
tional flexibility in contrast with the relative rigidity of the ing a total of six different AUTODOCK calculations. Further
other side chains lining the gorge. Three major orientations of refinement of the resulting structures of TcAChE-galanthamine
Phe330 have been observed experimentally. The TcAChE- complexes was done using molecular mechanics optimization
tacrine complex (1ACJ) is characterized by the “closed gate” with the Tripos forcefiled.47 In the course of these calculations,
conformation whereas the TcAChE complexes with decame- the orientation of Phe330 and Tyr121 (which opposes Phe330
thonium (1ACL) and donepezil (1EVE) show an “open gate” in the gorge) was frozen in runs with half-open and closed state
conformation. In other known TcAChE complexes the Phe330 to preserve the respective gate conformations, while all other
side chain is orientated in between these two extreme positions; side chains of the 42 amino acids subset were allowed to relax.
for instance, the half open conformation is observed in the
AChE/edrophonium complex (2ACK14). DOCKING OF GALANTHAMINE
Docking calculations were done both for galanthamine and
its N-alkyl derivative SPH1107. We noticed that if the long In all calculations, galanthamine was initially placed at or near
N-alkyl group of derivative extends into the gorge, only an the catalytic triad (Ser200, His440, Glu327). Five hundred
open gate conformation is possible. Thus, if one assumes that independent AUTODOCK runs for each of the different con-
both galanthamine and its derivative bind via a similar mech- formational states of the gorge were performed.
anism, only this open conformation of the gorge should be
assumed for both molecules. However, in principle, galan- Open Gate Conformation of the Gorge
thamine, which lacks the long side chain, could bind to the
active site in all three conformations of the gorge. Thus, for the As a result of the simulations, only 57 (eq)- and 22 (ax)-
completeness of this study and in light of three dominant galanthamine conformers resided inside the gorge. In all other
orientations of Phe330 side chain (open gate, half open gate, runs galanthamine was located outside of the gorge at or near
and closed gate) observed for other inhibitors of AChE, we the peripheral anionic site. Because of the established compet-
have considered all three possibilities in separate docking runs itive binding behavior of galanthamine, only the former struc-
for galanthamine and only open gate conformation for tures were considered and classified. Based on the energy
SPH1107. As a starting point, we chose the TcAChE coordi- values and galanthamine orientation in the TcAChE-
nates in complexes with decamethonium (1ACL), edropho- galanthamine complexes, AUTODOCK generated six clusters
nium (2ACK), and tacrine (1ACJ) for open, half open, and of binding orientation for the galanthamine (eq)-conformer
closed gate docking calculations, respectively. The crystal (Color Plate 4a) and four clusters for the (ax)-conformer (Color
structure of the TcAChE-galanthamine complex later revealed Plate 4b). The average AUTODOCK energies for all clusters
the orientation of Phe330 to be in an open gate state.36 are given in Table 1.
Two additional assumptions with respect to the galan- Based on the AUTODOCK energy, the Teq1 orientation
thamine conformation were made in our studies. First, although should be chosen as the best prediction since it had the lowest
galanthamine is a fairly rigid molecule, conformational flexi- energy among all clusters. However, one of the major draw-
bility of the cyclohexenol ring (ring C) and of the seven- backs of the AUTODOCK version used in our studies is its
member ring (ring D) cannot be neglected. Thus, it seemed inability to deal with the conformational flexibility of both
reasonable to perform a conformational search on the galan- receptor and ligand. To overcome this limitation, we employed
thamine molecule. To this end we have employed a simulated the Tripos force field (Pullman-charges) and submitted one
annealing protocol that, in our experience, can efficiently over- representative from each of the resulting clusters from
come conformational rotational barriers in cyclic saturated AUTODOCK runs to a geometry optimization. During this
systems.51 This procedure resulted in a conformation with the process, the subset of 42 amino acids employed in the docking
lowest energy, which closely reproduced the crystal structure studies was also kept flexible. The results of these calculations
of (⫺)-galanthaminium bromide52 and therefore was chosen are also given in Table 1.
for the ensuing docking studies. The data in Table 1 show that the models Teq26 and Tax24
Second, galanthamine contains a tertiary methylamine moi- correspond to the orientations of galanthamine in the TcAChE-
ety as part of the seven-member ring (ring D). This group can binding site with the lowest Tripos energy, which is different
adopt either an axial or equatorial orientation (Color Plate 2), from the ranking based on the AUTODOCK energy. Due to
characterized by practically the same value of molecular me- this discrepancy, we have also exploited a recently developed
chanics energy. A priori, both conformers could account for the SCORE method48 to assess the binding free energy of ligands.
biological activity of galanthamine. Therefore, both of them This method uses an empirical scoring function derived from
were considered independently in our docking study and are the analysis of available crystal structures of enzyme-inhibitor
referred to as (eq)- and (ax)- conformers, respectively, in the complexes and the respective experimentally determined bind-
following discussion. ing constants. The highest SCORE energy corresponds to the
Docking calculations were performed with the AUTO- highest free energy of binding. The resulting SCORE ranking

290 J. Mol. Graphics Modell., 2001, Vol. 19, No. 3/4

Table 1. Average energies (kcal/mol), SCORE value, and accuracy of matching (RMSD) the galanthamine x-ray
structure in its complex with TcAChE for different orientation clusters of galanthamine in the active site of
TcAChE in the open gate configuration. The data are shown only for the lowest energy clusters resulting from 500
AUTODOCK runs for each of the (eq)- and (ax) conformers of galanthamine.

Tripos
No of AUTODOCK
Cluster structures Energy of Energy of Energy of Energy of Binding SCORE RMSD
name in cluster complex complex ligand enzyme enthalpy value (Å)

Teq01 7 ⫺49.34 ⫺4001.07 22.10 ⫺3965.35 ⫺57.82 5.65 2.073

Teq02 19 ⫺46.66 ⫺3981.1 20.35 ⫺3964.21 ⫺37.34 4.77 3.493
Teq03 11 ⫺45.54 ⫺3990.97 21.40 ⫺3966.37 ⫺46.01 5.45 3.287
Teq04 5 ⫺43.64 ⫺3983.67 23.62 ⫺3966.52 ⫺40.77 4.97 3.693
Teq26 14 ⫺35.07 ⫺4002.57 22.18 ⫺3963.29 ⫺61.46 7.21 0.581
Teq27 1 ⫺28.32 ⫺3984.36 21.93 ⫺3962.13 ⫺44.16 5.84 3.140

Tax01 4 ⫺44.1 ⫺3989.91 18.53 ⫺3964.30 ⫺44.15 4.81 2.197

Tax02 2 ⫺42.46 ⫺3983.69 20.99 ⫺3962.99 ⫺41.69 6.25 2.222
Tax16 6 ⫺36.61 ⫺3980.72 19.83 ⫺3965.08 ⫺35.47 5.91 3.893
Tax24 10 ⫺32.88 ⫺4003.49 19.62 ⫺3963.02 ⫺60.09 7.07 0.459

of the galanthamine structures optimized with the Tripos force As indicated by the comparison with the X-ray structure,
field is given in Table 1. Again the models Teq26 and Tax24 the use of the half-open and closed gate conformations of
emerged as the most likely orientations of galanthamine in their TcAChE for docking yielded relatively correct prediction of
complexes with TcAChE. the galanthamine orientation in the TcAChE active site as
To evaluate how well the predicted orientations of galan- the lowest Tripos energy structure (cf. Table 2). However,
thamine compare with that found in the crystal structure of the accuracy of the best predictions are almost twice as poor
the TcAChE-galanthamine complex,36 the polypeptide back- as those obtained with the open gate conformation of the
bones of each of the energy minimized models were super- enzyme, with only two clusters (c⫺Teq03 in the closed gate
imposed onto the X-ray structure of the complex by a rigid conformation and ho⫺Teq02 in the half-open conformation)
RMS fit routine. The RMS deviations of the various pre- having RMSD values of just under 1 Å. For each of the gate
dicted orientations of galanthamine from that found in the conformations, only the (eq)-conformer had a reasonable
X-ray crystal structure of the TcAChE-galanthamine com- agreement with the crystal structure while the (ax)-
plex were then calculated independently and are reported in conformer could not adopt an orientation in the active site
the last column of Table 1. Apparently, the galanthamine in similar to the actual structure.
Teq26 and Tax24 models with the highest ranking based The comparison between all calculations (Tables 1 and 2)
both on Tripos and SCORE calculations also had the lowest suggests that only ranking based on the Tripos energies of
RMS deviations from the X-ray structure of 0.58Å and geometry optimized complexes obtained from AUTODOCK
0.46Å, respectively. calculations are in consistent agreement with experimental
results. SCORE ranking agreed with Tripos calculations only
Half Open and Closed Gate Conformations of in the case of open conformation of the enzyme. It is important
the Gorge to note that the Tripos energies of the complexes were higher
than those obtained with the open gate conformation (cf. re-
Both (ax)- and (eq) conformers of galanthamine were placed
spective columns in Tables 1 and 2). Thus, on the basis of the
into the half open and closed gate conformations of TcAChE
Tripos energy values after optimization of AUTODOCK gen-
and the same procedure as described above for the open gate
erated complexes, we would rule out both the half-open and
conformation of TcAChE was applied. AUTODOCK runs
closed conformations of the gorge and choose the open gate
followed by geometry optimization with the Tripos force
field resulted in several lowest energy clusters. These clus- conformation as the correct one. This conclusion is in complete
ters are characterized in Table 2, including the accuracy of agreement with the experimental structure of the TcAChE-
prediction evaluated by RMSD from the crystal structure of galanthamine complex.36
galanthamine-TcAChE complex. Again, the relative ranking
of docking configurations based on the AUTODOCK energy
did not correlate well with the Tripos binding enthalpy. DOCKING OF THE GALANTHAMINE
However, in this case there was no correlation between DERIVATIVE SPH1107.
the Tripos energy and SCORE values as well. In fact, in
several occasions we have observed structures with reason- We have made the assumption that galanthamine derivatives
ably high SCORE values that were clearly outside the active with long chain substituents at the ring-nitrogen (ring D) could
site. only bind to the enzyme in the open gate conformation, with

J. Mol. Graphics Modell., 2001, Vol. 19, No. 3/4 291

Table 2. Average energies (kcal/mol), SCORE value, and accuracy of matching (RMSD) the galanthamine x-ray
structure in its complex with TcAChE for different orientation clusters of galanthamine in the active site of
TcAChE in the “closed” and “half open” gate conformations with fixed positions of Phe330 and Tyr121. The data
are shown only for the three lowest energy clusters resulting from 500 AUTODOCK runs for each of the (eq)- and
(ax) conformers of galanthamine.

No of AUTODOCK Tripos Tripos

Gate Cluster structures energy of energy of binding SCORE RMSD
configuration name in cluster complex complex enthalpy value [Å]

“closed” c_Teq03 40 ⫺43.6 ⫺3828.2 ⫺59.5 6.60 0.96

c_Teq07 8 ⫺40.9 ⫺3827.0 ⫺59.0 5.68 2.04
c_Teq06 2 ⫺41.0 ⫺3823.3 ⫺55.9 6.07 2.71

c⫺Tax21 24 ⫺31.5 ⫺3826.8 ⫺56.4 6.81 1.75

c_Tax10 7 ⫺36.1 ⫺3821.3 ⫺50.7 7.28 3.50
c_Tax03 12 ⫺33.6 ⫺3819.1 ⫺46.1 5.83 3.51

“half open” ho_Teq02 73 ⫺47.0 ⫺3968.5 ⫺54.5 6.81 0.94

ho_Teq01 26 ⫺50.2 ⫺3966.3 ⫺49.3 6.75 2.42
ho_Teq13 8 ⫺31.1 ⫺3962.2 ⫺48.4 4.85 3.64

ho_Tax06 14 ⫺38.5 ⫺3969.7 ⫺48.9 5.62 3.35

ho_Tax03 12 ⫺39.2 ⫺3968.4 ⫺42.9 5.14 2.83
ho_Tax07 15 ⫺38.4 ⫺3962.5 ⫺43.8 5.48 2.98

the side chain extending into the gorge toward the peripheral COMPARISON OF THE MODELING
anionic site. Docking calculations were done with the galan-
RESULTS WITH CRYSTAL
thamine derivative SPH110734 (see Figure 1). The orientation
and conformation of this derivative was predicted using the STRUCTURES.
same procedure described for galanthamine itself. Again, both
(eq)- and (ax)-conformers were considered since no dominant The described modeling procedure discloses that there are only
axial or equatorial orientation of the N-substituents in the two distinct areas in which binding of galanthamine is allowed:
bound state could be deduced from the molecular mechanics the upper and the lower end of the gorge (Color Plate 6). This
calculations. Results were evaluated by both the Tripos binding finding reflects the fact that the gorge narrows quite substan-
energy and the SCORE value, as described above (Table 3). tially from the mouth of the opening down into the protein
The highest scoring models for the (eq)- and (ax)-conformers before it opens up again at the catalytic site on the bottom of
were superimposed onto the crystal structures of decametho- the gorge. The docking procedure clearly distinguishes the
nium, donepezil and MF268, which are known to bind with an active site region from the peripheral anionic site at the mouth
extended conformation along the active site gorge (see Color of the gorge. Galanthamine binding was detected only in these
Plate 5). two locations. The deviation from the orientation observed in

Table 3. Average energy (kcal/mol), SCORE value, and deviation (RMSD) of the galanthamine moiety of
derivative SPH1107 from the respective galanthamine ring structure as observed in the x-ray crystal structure of
the TcAchE-galanthamine complex. Data are shown for the lowest energy clusters resulting from 500 AUTODOCK
runs for each of the (eq)- and (ax) conformers of SPH1107.

No of SYBYL energy of SYBYL binding RMSD

structures in complex enthalpy SCORE (core)
Cluster name cluster [kcal/mol] [kcal/mol] value [Å]

1107ax25 20 ⫺4010.18 ⫺76.27 8.30 0.50

1107ax22 42 ⫺4006.79 ⫺74.45 8.08 0.56

1107eq18 24 ⫺4010.57 ⫺75.22 7.77 0.41

1107eq37 5 ⫺4006.55 ⫺73.19 7.70 0.40

292 J. Mol. Graphics Modell., 2001, Vol. 19, No. 3/4

the X-ray crystal structure of the TcAchE-galanthamine com- of the Phe330 side chain orientation on the docking process,
plex36 as revealed by the RMSD values also clearly separates comparative AUTODOCK runs were performed also with
the structures in the active site (RMSD ⬍4) from those at the fixed orientations of Phe330 and Tyr121 using an enzyme
periphery (RMSD ⬎ 6). No intermediate RMSD values in the structure in the TcAChE-tacrin complex (1ACJ) and the
range of 4 – 6 were found. TcAChE-edrophoinium complex (2ACK) as well as the
A comparison of the orientation and conformation of galan- TcAChE-decamethonium complex (1ACL).
thamine in the crystal structure of the TcAChE-galanthamine Residues with missing side chain atoms were rebuilt using
complex with the best hits from our modeling study revealed an the BIOSYM BIOPOLYMER module.53 Hydrogen atoms were
extremely good agreement for the models Teq26 and Tax24 added to all amino acid residues. The side chains of glutamic
(see Tables 1 and 2, respectively) with a resulting RMSD value and aspartic acids as well as the side chains of arginine, lysine,
of ⬃0.5 Å for both predicted structures. The orientation of the and histidine were assigned formal negative and positive
N-methyl group (ring D) is the only ambiguity present in our charges, respectively. The total formal charge amounted to ⫹4.
modeling studies that could not be resolved. Color Plate 7 The N- and C- terminus of the polypeptide chains were capped
displays an overlay of the corresponding predicted structures with an acetyl moiety (Ser4, Ser 490) and an N-methyl group
clearly showing that the experimental structure indeed has the (Glu484, Thr535), respectively. The resulting structure was
equatorial conformation and thus matches our predicted model subjected to geometry optimization employing the Tripos47
Teq26. force field as implemented in SYBYL 6.5 [Pullman charges,
In addition, the orientation of 42 amino acids that were used Powell Minimizer, 500 steps, convergence criteria: gradient
to represent the TcAChE gorge in the modeling experiments 0.05 kcal/(mol2*Å)]. In the first optimization step, only hydro-
also was in good agreement with the crystal structure (RMS gen atoms were relaxed while all heavy atoms were kept fixed.
deviation for all backbone atoms is 0.12 Å), even though In the second step, side chains of all amino acids were allowed
modeling involved complete relaxation of the side chains in the to move during the optimization process, keeping only the
final optimization steps. backbone atoms at their positions. In the final optimization
The comparison of the hydrogen bonding pattern between step, all atoms were allowed to relax.
galanthamine and TcAChE as revealed by the modeling study The resulting enzyme structure was used as an input for the
with that found in the crystal structure of the TcAChE- AUTOGRID program.44 In the first step, AUTOGRID defines
galanthamine complex is given in Table 4. These interactions a fine-meshed grid (0.375 Å) that surrounds the area of the
include a hydrogen bond between the galanthamine O-CH3 enzyme in which the ligand is expected to bind (size: 22.5 ⫻
oxygen (ring A) acting as an acceptor, and the OH-group of 22.5 ⫻ 22.5 Å3). Since galanthamine is known to be a com-
Ser200 (O␥) acting as a donor, and a second hydrogen bond petitive inhibitor, the center of the box and the initial position
between the OH-group of the cyclohexenol (ring C) acting as of the ligand were placed at the bottom of the gorge.
a donor to the carboxyl-group of Glu199 (O⑀1), acting as an For each ligand atom type AUTOGRID calculates the Van
acceptor. der Waals energies at each single grid point. The program
generates separate grid maps for all atom types occurring in the
EXPERIMENTAL DETAILS ligand structure. A grid map for electrostatic interactions is
computed by moving a point charge within the grid and cal-
Docking studies were performed using the AUTODOCK 2.4 culating the resulting energies. In the case of galanthamine,
program to deduce conformations and orientations of galan- grid maps for C, H, N, O, and point charge were used and the
thamine and the derivative SPH1107 in the TcAChE-binding actual docking process was performed within this grid.
site. AUTODOCK operates within pregenerated grid maps so AUTODOCK translates and rotates the ligand and simulta-
that the conformational flexibility of the receptor is not con- neously modifies user-defined torsion angles. The use of pre-
sidered at all during the docking process. The enzyme- generated grid maps affords a rapid energy evaluation of the
conformation of the TcAChE-decamethonium complex enzyme/ligand complex. AUTODOCK uses a simulated an-
(1ACL) was chosen for our docking study. Since the positions nealing protocol to search for energetically most favorable
of most water molecules in the crystal structures of TcAChE- orientatation of ligand in the active site. One AUTODOCK
decamethonium and TcAChE-galanthamine complexes, re- simulation typically consists of 500 runs, 50 cycles, and 3000
spectively, are unlikely to be conserved, water molecules were accepted or rejected enzyme/ligand alignments, respectively.
excluded before the docking protocols. To study the influence RMS tolerance during the clustering was set to 1 to reduce the
number of clusters. All other values were used as default
settings. Subsequently, AUTODOCK clusters the resulting ori-
Table 4. Distances of heteroatoms in hydrogen entations of the ligand, taking into account the total interaction
bonding of galanthamine with TcAChE active site energy between the ligand and enzyme and an RMSD criterion.
amino acids. Usually up to 30 clusters are obtained, representing a series of
energetically meaningful enzyme/ligand arrangements.
These structures were submitted to a geometry optimization
Distance of heteroatoms [Å]
within the active site of the enzyme employing the Tripos force
H-bond Teq26 Tax24 crystal structure field. During the optimization process, the side chains of 42
amino acids surrounding the active site as well as the single
OH–CLU199 2.51 2.50 2.72 bonds of the ligand (galanthamine and SPH1107, respectively)
COC-HIS440 3.32 3.28 3.20 were kept flexible. When the influence of the orienttion of
OCH3-SER200 2.50 2.50 2.97 Phe330 on the galanthamine docking was investigated, Phe330
and Tyr121 were fixed while the remaining 40 amino acids

J. Mol. Graphics Modell., 2001, Vol. 19, No. 3/4 293

again were kept flexible. The rest of the enzyme was hold rigid, ranking the modeling results, the two highest scoring
while long range interactions were still included. Dissociation AUTODOCK clusters were predicted for the (eq)- and (ax)-
constants (pKD) for the obtained ligand/enzyme complexes conformers of galanthamine, respectively.
were estimated with the program SCORE,48 and were the basis As a fortunate coincidence, a crystallographic investigation
for the scoring of the ligand-enzyme-alignments suggested by of the TcAChE-galanthamine complex was recently performed
the docking procedure. by Bartolucci et al36 independently from the modeling studies
reported in this article. This structure enabled us to test the
DISCUSSION accuracy of our predictions with respect to the orientation and
conformation of galanthamine in the enzyme active site. Typ-
The aim of our docking studies was to develop a procedure that ical representatives of these (eq)- and (ax)-clusters nicely over-
would allow the prediction of likely orientations of galan- lay with the galanthamine conformation in the crystal structure
thamine and its derivative, SPH1107, in the active site of of the TcAChE-galanthamine complex, showing RMSD values
TcAChE. This structural information may be valuable for sub- of ⬃0.5 Å.
sequent drug design studies using structure-based or 3D-QSAR Comparably good overlap with the crystal structure was also
methods. Examples of structure-based alignment for 3D QSAR achieved for the (eq)-structure when the closed gate and the
analysis are provided by recent studies from our24 and other open gate protein conformations were employed rather than the
laboratories.25,26 Recently, various galanthamine derivatives open gate state. However, in these cases the presence of the
were synthesized, which provide an excellent starting point for phenyl ring of Phe330 prohibits the (ax)-conformer of galan-
rational design and development of anti-cholinesterase drugs.34 thamine to be oriented in the active site in a position that
Derivatives that contain a long substituent on the nitrogen are matches the crystal structure.
of particular importance. Such structures potentially span the It should be noted that our computational docking experi-
whole binding cavity of TcAChE and interact with several ments do not incorporate water molecules that might be located
amino acids at the peripheral anionic site, which should lead to in the active site. This procedure is always a source of potential
increased potency of AChE inhibition. We have included in our misinterpretation of docking studies as illustrated by recent
study the galanthamine derivative SPH1107, which has an docking calculations with the AChE inhibitor huperzine A.27
N-propyl-piperidine substituent at the nitrogen atom rather than From the crystallographic investigation,36 it is evident that one
the methyl group. water molecule does play an important role in the interaction
Both axial and equatorial orientation of the nitrogen sub- between TcAChE and galanthamine. However, because of the
stituent in galanthamine and its SPH1107 derivative were con- very good overlap shown in this study between the predicted
sidered separately in the AUTODOCK procedure. The incor- galanthamine conformation and that found by X-ray crystal-
poration of the SPH1107 derivative in the present study lography, water molecules do not seem to be essential for the
required the definition of a fairly large active site box in which accurate prediction of the bound ligand conformation.
the ligand is confined. Furthermore, it biased us to consider the The predicted structures show direct hydrogen bonding to
open gate conformation of the active site gorge, found, for the catalytic residue Ser200, which probably leads to the ef-
instance, in the TcAChE-decamethonium complex as the stan- fective blockade of TcAChE by galanthamine. However, the
dard TcAChE structure. However, the closed gate structure (as docking study can not differentiate between the axial and
in the TcAChE-tacrin complex) and the half open gate confor- equatorial conformer, since both conformers are characterized
mation (as in the TcAChE-edrophonium complex) were also by similar score values and comparably good RMSD values.
included in our study. Considering that the two conformers should interchange rap-
Docking studies, especially with AUTODOCK usually iden- idly in the naturally bound ligand, our findings appear reason-
tify a small number of potential binding orientations for a given able.
ligand. It is quite difficult to unambiguously select the most To prove that the gorge interior may be exploited for the
probable binding site and the orientation of the ligand. Addi- docking studies of other inhibitors, we applied the same strat-
tional scoring procedures should be employed to rank the egy to the galanthamine derivative SPH1107 that carries a long
resulting structures. In our study we concentrated on two chain at the nitrogen atom. We found that the AUTODOCK
ranking criteria, the energy of the AChE-ligand complex ob- clusters for either the (eq)- and (ax)-conformer are almost
tained with the Tripos force field, and the energies calculated identical with respect to the orientation of the rigid galan-
with the empirical SCORE method.48 thamine ring structure. The only minor difference is in the
As a consequence of using the large active site box in the orientation of the N-substituent as a consequence of the axial
docking study, a great portion of the resulting galanthamine and equatorial position, respectively. Both structures nicely
structures are found near the peripheral anionic site at the top superimpose on the corresponding galanthamine conformer
of the gorge. However, since galanthamine is known to be a and indeed showed the expected extension of the N-substituent
competitive inhibitor, we assumed that binding actually occurs towards the outside of the gorge. The galanthamine derivative
at the bottom of the gorge, similar to other known inhibitors of SPH1107 displays interactions with the gorge residues that are
TcAChE. Most of the known inhibitors interact with both or very similar to those observed in crystal structures of the
either of the two residues, Ser200 and His440, that belong to TcAChE complex with decamethonium (1ACL) and donepezil
the catalytic triad of the enzyme, with additional interaction (1EVE). It is worth noting that important structural segments of
with Glu199. This assumption substantially decreased the num- both the galanthamine derivative and donepezil have almost
ber of matches in the modeling study to 57 and 22 for the (eq)- exactly the same location in the active site. For instance, the
and (ax)-conformers of galanthamine, respectively, out of 500 nitrogen atom of galanthamine is positioned only 1.26 Å ((eq)-
AUTODOCK results for each configuration. Furthermore, us- conformer) and 1.51 Å ((ax)-conformer) apart from the piper-
ing the binding energies derived from the Tripos force field for idine nitrogen of donepezil. Furthermore, the double bond of

294 J. Mol. Graphics Modell., 2001, Vol. 19, No. 3/4

the cyclohexenol ring (ring C) of the galanthamine derivative 7 Nordberg, A., and Svensson, A.L. Cholinegic inhibitors
nicely overlays with the aromatic benzyl moiety of donepezil. in the tratment of Alzheimer’s disease - A comparison of
In summary, we have developed a combined docking pro- telerability and pharmacology. Drug Safety 1998, 19,
cedure that uses AUTODOCK rigid docking followed by a 465–480 and 1999, 20, 146
flexible molecular mechanics optimization of resulting struc- 8 Wolmark, Y. Alzheimer’s disease - the therapeutic era,
tures. AUTODOCK is used initially to generate several accept- Drug Discovery Today 1999, 4, 149–151
able orientations of ligands. The resulting structures are sub- 9 Hakansson, L. Mechanism of action of cholinesterase
jected to the geometry optimization with the Tripos forcefield, inhibitors in Alzheimer’s disease. Acta. Neurol. Scand.
and the ligand-receptor interaction energies are used for the Suppl. 1993, 149, 7–9
final ranking of ligand receptor complexes. We showed that 10 Sirvio, J. and Riekkinen, P. J. Brain and cerebrospinal
this procedure provides a reliable method for predicting the fluid cholinesterases in Alzheimer’s disease, Parkin-
orientation and conformation of galanthamine type ligands in son’s disease and aging. A critical review of clinical and
the active site of the TcAChE as judged by very low RMSD experimental studies. J. Neural. Transm. Park. Dis. De-
values between the predicted highest scoring orientations of ment. Sect. 1992, 4, 337–358
11 Linton, D. M.; Philcox, D. Myasthenia gravis. Dis. Mon.
galanthamine and the experimentally determined one. The suc-
1990, 36, 593–637
cess of the reported studies provides an excellent basis for the
12 Taylor, P. Anticholinesterase agents. In: The Pharmaco-
applications of the described docking procedures to a large
logical basis of therapeutics; Gilman, A. G., Rall, T. W.,
number of de novo designed and synthesized galanthamine
Nies, A. S., and Taylor, P., Eds.; Pergamon, New York,
derivatives. The present docking studies may guide the design 1990; pp131–149
process as well as provide structure based 3D alignment of 13 Harel, M., Schalk, I., Ehret-Sabatier, L., Bouet, F.,
galanthamine type AChE inhibitors, which might prove rele- Goeldner, M., Hirth, C., Axelsen, P.H., Silman, I., and
vant for a 3D QSAR analysis. Investigations along these lines Sussman, J.L. Quaternary ligand binding to aromatic
are presently ongoing in our laboratories. Note added in proof. residues in the active-site gorge of acetylcholinesterase.
After completion of this study a second structure of the Proc. Natl. Acad. Sci. USA. 1993, 90, 9031–9035
TcAChE/galanthamine complex was deposited in the PDB 14 Ravelli, R.B.G., Raves, M.L., Ren, Z., Bourgeois, D.,
database (1DX6),54 which shows an identical orientation and Roth, M., Kroon, J., Silman, I., and Sussman, J.L. Static
conformation of galanthamine in the TcAChE active site to the Laue diffraction studies on acetylcholinesterase. Acta
1QTI-structure.36 Cristallogr. D54, 1998, 1359–1366
15 Wagstaff, A.J., and McTavish, D. Tacrine: A review of
ACKNOWLEDGMENTS ist pharmacolodynamic and pharmacokinetic properties,
and therapeutic efficacy in Alzheimer’s diseas. Drugs
Support to C. Pilger by Fonds der Chemischen Industie (FCI) and Aging 1994, 4, 510–540
is gratefully acknowledged. A. Tropsha acknowledges the sup- 16. Inoue, A., Kawai, T., Wakita, M., Iimura, Y., Sugimoto,
port from NIH (SBIR Grant RR10687-02A1). H., and Kawakami, Y., The simulated binding of (⫾)-
2,3-dihydro-5,6-dimethoxy-2-[[1-(phenyl,ethyl)-4-
piperidinyl]methyl]-1H-inden-1-one hydrochloride
REFERENCES (E2020) and related inhibitors to free and acylated
1 Dunnett, S.B., and Fibiger, H.C. Role of forebrain cho- acetylcholinesterases and corresponding structure–
linergic systems in learning and memory: relevance to activity analyses. J. Med. Chem. 1996, 39, 4460–4470
the cognitive deficits of aging and Alzheimer’s demen- 17 Kozikowski, A.P., Miller, C.P., Yamada, F., Pang, Y.-P.,
tia. Prg. Brain. Res. 1993, 98, 413–420 Miller, J.H., McKinney, M., and Ball, R.G., Delineating
2 Babic T., Smith R.J., and Petravic D. Current knowledge the pharmacophoric elements of huperzine-A - Impor-
on Alzheimer’s disease. Neurologia Croatica 1997, 46, tance of the unsaturated 3-carbon bridge to its AChE
55–72 inhibitory activity , J. Med. Chem., 1991, 34, 3399–3402
3 Schroeder, H., Lobron, C., Wevers, A., Maelicke, A., and 18 Bartolucci, C., Perola, E., Cellai, L., Brufani, M.,
Giacobini, E. Cellular acetylcholine receptor expression Lamba, D., “Back door” opening implied by the crystal
in the brain of patients with Alzheimer’s and Parkin- structure of a carbamoylated acetylcholinesterase. Bio-
son’s dementia. In: Advances in Behavioral Biology chemistry 1999, 38, 5714–5719
(Hanin, I., Yoshida, M., Fisher, A., eds.) Plenum Press, 19 Bores, G.M., and Kosley, R.W. Galanthamine deriva-
1995, 44, 63–68. tives for the treatment of Alzheimer’s disease. Drugs of
4 Strittmatter, W.J., Weisgraber, K.H., Goedert, M., Saun- the Future 1996, 21, 621–635
ders, A.M., Huang, D., Corder, E.H., Dong, L.M., Jakes, 20 Han, S.Y., Sweeney, J.E., Bachman, E.S., Schweiger,
R., Alberts, M.J., Gilbert, J.R., Han, S.H., Hulette, C., E.J., Forloni, G., Coyle, J.T., Davis, B.M., and Joullie,
Einstein, G., Schmechel, D.E., Pericakvance, M.A., and M.M. Chemical and pharmacological characterization of
Roses, A.D. Hypothesis - Microtubule instability and galanthamine, an acetylcholinesterase inhibitor, and its
paired helical filament formation in the Alzheimer- derivatives - A potential application in Alzheimers-
disease brain are related to apolipoprotein-E genotype. disease. Eur. J. Med. Chem. 1992, 27, 673–687
Exp. Neurol. 1994, 125, 163–171 21 Gilson, M.K., Straatsma, T.P., McCammon, J.A., Ripoll,
5 Giacobini, E., and Becker, R. (eds.) Alzheimer Disease; D.R., Faerman, C.H., Axelsen, P.H., Silman, I., and
Therapeutic Strategies. Birkhäuser, Cambridge 1994 Sussman, J.L. Open Back Door in a Molecular-
6 Siddiqui, M.F., and Levey, A.I. Cholinergic therapies in Dynamics Simulation of Acetylcholinesterase. Science
Alzheimer’s diseas. Drugs of the Future 1999, 24, 417– 1994, 263, 1276–1278
424 22 Wlodek, S.T., Clark, T.W., Scott, L.R., and McCammon,

J. Mol. Graphics Modell., 2001, Vol. 19, No. 3/4 295

 J.A. Molecular dynamics of acetylcholinesterase dimer pedo californica: Implication for the Design of New
complexed with tacrine. J. Am. Chem. Soc. 1997, 119, Anti-Alzheimer Drugs. Proteins (in press)
9513–9522 37 Kuntz, I.D., Blaney, J.M., Oatley, S.J., Langridge, R.,
23 Axelsen, P.H., Haqrel, M. Silman, I., and Sussman, J.L. and Ferrin, T.E. A geometric approach to macromole-
Structure and dynamics of the active site gorge of cule-ligand interactions, J. Mol. Biol. 1982, 161, 269–
acetylcholinesterase: synergistic use of molecular dy- 288
namics simulation and X-ray crystallography. Prot. Sci. 38 Meng, E.C., Shoichet, B.K., and Kuntz, I.D. Automated
1994, 3, 188–197 Docking with Grid-Based Energy Evaluation. J. Com-
24 Cho, S.J., Serrano, M.G., Bier, J. and Tropsha, A. Struc- put. Chem. 1992, 13, 505–524
ture based alignment and comparative molecular field 39 Bartlett, P.A., Shea, G.T., Telfer S.T., and Watermann,
analysis of acetylcholinesterase inhibitors. J. Med. S. CAVEAT: A program to facilitate the structure de-
Chem. 1996, 39, 5064–5071 rived design of biologically active molecules. Roberts,
25 Hasagawa, K., Kimura, T., Funatsu, K. GA strategy for S.M., Ed., Royal Society of Chemsitry: London, 1989,
variable selection in QSAR studies: Application of GA- pp 182–196
based region selection to a 3D-QSAR study of acetyl- 40 Rarey, M. Kramer, B., Lengauer, T., and Klebe, G. A
cholinesterase inhibitors. J. Chem. Inf. Comp. Sci. 1999, fast flexible docking method using an incremental con-
39, 112–120 struction algorithm. J. Biol. Chem. 1996, 261, 470–489
26 Recanatini, M., Cavalli, A., Hansch, C. A comparative 41 Welch, W., Ruppert, J., and Jain, A.N. Hammerhead:
QSAR analysis of acetylcholinesterase inhibitors cur- fast, fully automated docking of flexible ligands to pro-
rently studied for the treatment of Alzheimer’s disease. tein binding sites. Chem. Biol. 1996, 3, 449–462
Chem. Biol. Interact.1997, 105, 199–228 42 Lewis, R.A., Roe, D.C., Huang, C., Ferrin, T.C., Lan-
27 Raves, M.L., Harel, M., Pang, Y., Silman, I., gridge, R., and Kuntz, I.D., Automated Site-Directed
Kozikowski, A.P., and Sussman, J.L. Structure of ace- Drug Design Using Molecular Lattices. J. Mol. Graphics
tylcholinesterase complexed with the nootropic alkyloid, 1992, 10, 66–78
(-)-huperzine A. Nat. Struct. Biol. 1997, 4, 57–63 43 Hart, T.N., and Read, R.J. A Multiple-Start Monte-Carlo
28 Kryger, G., Silman, I., and Sussman, J.L. Three- Docking Method. Proteins 1992, 13, 206–222
dimensional structure of a complex of E2020 with ace- 44 Goodsell, D. S. and Olson, A. J. Automated docking of
tylcholinesterase from Torpedo californica. J. Physiol substrates to proteins by simulated annealing. Proteins:
(Paris) 1998, 92, 191–194
Str. Func. and Genet. 1990, 8, 195–202
29 Harel, M., Quinn, D.M., Nair, H.K., Silman, I., and
45 McMartin, C, and Bohacek, R.S. QXP: Powerful, rapd
Sussman, J.L. The X-ray structure of a transition state
computer algorithms for structure-based drug design.
analog complex reveals the molecular origins of the
J. Comput. Aided Mol. Design 1997, 11, 333–344
catalytic power and substrate specificity of acetylcho-
46 Dixon, J.S., and Blaney, J.M. Docking: Predicting the
linesterase. J. Am. Chem. Soc. 1996, 118, 3240–3246
structure and binding affinity of ligand receptor com-
30 Proskurnia, N.F., Yakovleva, A.P. Alkaloids of Galan-
thus woronwi. II. Isolation of a new alkaloid. Zhur. pexes. Des. Bioact. Mol. 1998, 175–197
Obshchei Khim. 1952, 22, 1899–1902 47 SYBYL 6.5, Tripos Inc., 1699 South Hanley Rd., St.
31 Kobayashi, S., Shingu, T., Uyeo, S. Structure of galan- Louis, Missouri, 63144, USA
thamine and lycoramine. Chem. Ind. 1956, 177–178 48 Wang, R., Liu, L., Lai, L., Tang, Y., SCORE: A new
32 Holton, R.A., Mukund, P.S. and Murphy, W.S. Palladi- empirical method for estimating the binding affinity of a
um-mediated biomimetic synthesis of Narwedine. protein-ligand complex. J. Mol. Model. 1998, 4, 379–
J. Amer. Chem. Soc. 1988, 110, 314–316 33 Kuenburg, 394
B., Czollner, L., Fröhlich, J., Jordis, U. Development of 49 Bernstein, F.C., Koetzle, T.F., Williams, G.J., Meyer Jr.,
a pilot scale process for the anti-Alzheimer drug (⫺)- E., Brice, M.D., Rodgers, J.R., Kennard, O., Shimanou-
galanthamine using large scale phenolic oxidative cou- chi, T., and Tasumi, M. The protein data bank: A com-
pling and crystallization-induced chiral conversion. Org. puter-based archival file for macromolecular structures.
Process Res. Develop. 1999, 3, 425–431 J. Mol. Biol. 1977, 112, 535–542
34 Czollner, L., Fröhlich, J., Jordis, U., and Küenburg, B. 50 Schuh, F.T., On the molecular mechanism of action of
Neue Benzazepinderivate, diese enthaltende Arzneimit- galanthamine, an antagonist of nondepolarizing muscle
tel und Verwendung derselben zum Herstellen von Ar- relaxants. Der Anaesthesist 1976, 25, 444–448
zneimitteln. Österreichisches Patent, AT 403.803, 51 Mottola, D.M., Laiter, S., Watts, V.J., Tropsha, A.,
15.10.1997. and Czollner, L., Fröhlich, J., Jordis, U., and Wyrick, S.W., Nichols, D.E., and Mailman, R. Confor-
Küenburg, B. New benzazepine derivates, medicaments mational analysis of d1 dopamine receptor agonists:
containing the same and their use to prepare medica- Pharmacophore assessment and receptor mapping.
ments. Eur. Pat. Appl. EU 0897387 A, 24.2.1999 J. Med. Chem., 1996, 39, 285–296
35 Sussman, J.L.; Harel, M.; Frolow, F.; Oefner, C.; Gold- 52 Peeters, O. M., Blaten, N.M., and De Ranter, C. J.
man, A.; Toker, L.; and Silman, I. Atomic structure of (⫺)-Galanthaminium bromide. Acta Cryst. C, 1997, 53,
acetylcholinesterase from Torpedo californica: A proto- 1284–1286
typic acetylcholine-binding protein. Science 1991, 253, 53 BIOSYM BIOPOLYMER module, MSI, San Diego
8872–8879 54 Greenblatt, H.M., Kryger, G., Lewis, T., Silman, I, and
36 Bartolucci, C., Perola, E., Pilger, Ch., Fels, G. Lamba, Sussman, J.L. Structure of Acetylcholinesterase com-
D., Tree-dimensional Structure of a Complex of Galan- plexed with (⫺) galanthamine at 2.4 Å resolution. FEBS
thamine (Nivalin威) with Acetylcholinesterase fom Tor- Lett. 1999, 463, 321–326

296 J. Mol. Graphics Modell., 2001, Vol. 19, No. 3/4

Christian Pilger, Cecilia Bartolucci, Doriano Lamba, Alexander Tropsha, and Gregor Fels

Accurate prediction of the bound conformation of galanthamine in the active site of

Torpedo ealifornica acetylcholinesterase using molecular docking

Color Plate 1. Comparison of the side chain orientation of PHE 330 with a closed gate conformation in the TcAChE-tacrine
complex (left) and an open gate conformation in the TcAChE-decamethonium complex (right).

Color Plate 2. Global minimum structures of galanthamine with tertiary methylamine group in (ax)-conformer (left) and
(eq)-conformer (right).

374 J. Mol. Graphics Modell., 2001, Vol. 19, No. 3/4

Color Plate 3. Location of the active site box (yellow) used in the AUTODOCK procedure to predict galanthamine binding
in the active site. For comparison the structure of decamethonium (white) is placed in the enzyme according to the known
crystal structure (1ACL).

J. Mol. Graphics Modell., 2001, Vol. 19, No. 3/4 375

 13

Color Plate 4. Typical orientations of galanthamine in

the active site of ACHE: (A) (eq)- conformers, (B)
(ax)- conformers) found by AUTODOCK. Structures
are oriented as found in the active site of the enzyme
(not shown). Yellow boxes denote the preferred ori-
entation as found by the modeling study.

376 J. Mol. Graphics Modell., 2001, Vol. 19, No. 3/4

 Color Plate 6. Binding location of all AUTODOCK clusters
found with the (eq)-conformer of galanthamine. For com-
parison, the active site box as used in the AUTODOCK
process and the amino acids Phe330 and Tyrl21 are shown.
The red structure depicts cluster T - e q - 2 6 (see Table 2,
green structures are located in the active site and typically
have RMSD values < 4 compared with the crystal structure,
while blue structures have RMSD values >6 and clearly are
positioned at the mouth of the gorge.

Color Plate 5. Structural overlay of the highest scoring

(ax)-conformer (A) and (eq)-conformer (B) of galanthamine
derivative SPHll07 (white) with the respective galan-
thamine (ax)- and (eq)-conformer as revealed by the mod-
eling study (yellow). For comparison the crystal structures
of decamethonium (1ACL, brown), donapecil (1EVE,
green) and Mf268 (1OCE, yellow) are shown.

J. Mol. Graphics Modell., 2001, Vol. 19, No. 3/4 377

Color Plate 7. Structural overlay of galanthamine models Teq26 (A), and Tax24 (B) as revealed by the docking study witl
that of galanthamine present in the crystal structure of the TcAChE-galanthamine complex45 (green).

378 J. Mol. Graphics Modell., 2001, Vol. 19, No. 3/4