Improvement of a Molecular Docking Approach and
Its Applications Using QXP+
Von der Fakultät für Naturwissenschaften
Department Chemie
der Universität Paderborn
zur Erlangung des Grades eines
Doktors der Naturwissenschaften
- Dr. rer. nat. –
genehmigte Dissertation
von
Laleh Alisaraie
aus Teheran, (IRAN)
Paderborn 2005
Eingereicht am: 22.02.05
Mündliche Prüffung am: 17.03.05
Referent: Prof. Dr. G. Fels
Koreferent: Prof. Dr. G. Henkel
ii
The present work was carried out in the Department of Organic Chemistry at University of
Paderborn-Germany, under the supervision of Prof. Dr. Fels from November 2001 until
February 2005.
First and foremost, I thank GOD, the light and the guidance in my life.
It is also a great pleasure to express my gratitude to people, who assisted me during these
three years.
I am especially grateful to my supervisor, Prof. Dr. G. Fels for the useful comments, fruitful
discussions and his continuous support.
I wish to thank Prof. Dr. G. Henkel, Prof. Dr. M. Grote and Prof. Dr. C. Schmidt for willing
to read my thesis and attend in my examination committee.
I am thankful to Dr. D. Lamba, in International Center for Genetic Engineering and
Biotechnology in Trieste-Italy, for generously providing us with the crystallographical data of
BHG and PPG, prior to publication.
I wish to thank Dr. Edgar Luttmann, Jens Krüger and Oliver Stüker for patiently answering
my questions and the technical supports.
I am thankful to Dr. Brigitta Elsässer for reading the correction of this work and being always
ready to help.
My special warm thanks go to my former and present colleagues in research group of Prof.
Dr. Fels for the very friendly workplace atmosphere.
All of my colleagues in Organic Chemistry Department are thanked for their helpfulness and
the pleasant social environment.
Laleh Alisaraie
iii
Contents
Summary………………………………………………………………………..1
1 Introduction............................................................................................2
1.1 Alzheimer’s disease ................................................................................................2
1.2 Cholinergic hypothesis............................................................................................2
1.3 Approaches for treatment of AD’s...........................................................................2
1.4 Cholinergic enhancement therapy............................................................................3
1.5 Structure of AChE...................................................................................................4
1.6 Overview of the three-dimensional structure of AChE.............................................4
1.6.1 The active center and the catalytic triad...........................................................5
1.6.2 Peripheral anionic site .....................................................................................6
1.6.3 Acylation of ACh within the binding pocket....................................................7
1.7 Acetylcholinesterase inhibitors (AChEI) .................................................................8
1.7.1 Synaptic cholinergic drugs...............................................................................8
1.7.2 Pseudo-irreversible AChEIs ............................................................................9
1.7.3 Irreversible AChEIs.......................................................................................10
1.7.4 Transition state analogue inhibitors ...............................................................10
1.7.5 Reversible AChEIs........................................................................................10
1.8 Protein-Ligand docking.........................................................................................14
1.8.1 General procedure of docking........................................................................14
1.8.2 Docking Algorithms implemented in this collection ......................................15
2 Description of the consensus molecular docking approach ...............19
2.1 The initial requirements ........................................................................................19
2.2 Characteristic features of the consensus method....................................................19
2.2.1 Aim of the consensus method........................................................................20
2.2.2 Multi-Step Docking (MSD) procedure...........................................................21
2.3 Application of MSD method for bound docking BHG...........................................22
2.4 Study on the importance of data filtration by bound docking of BHG....................26
2.5 Unbound docking study on BHG...........................................................................28
2.6 Importance of selection a suitable protein for unbound docking ............................31
2.7 A comparison of unbound docking using simulated annealing with using MSD ....33
iv
2.8 Comparison of the MSD result with various other docking methods......................35
3 Evaluation of MSD ..............................................................................37
3.1 Application of MSD for elucidation of structural properties of (BHG)...................37
3.2 Bound docking experiment on BHG-AChE complex.............................................38
3.3 Bound docking of BHG into binding site with waters............................................39
3.4 Docking study on E2020.......................................................................................45
3.4.1 Study on the configuration state of the molecule at C8...................................45
3.4.2 Study on the protonation state at E2020-Nitrogen..........................................46
3.4.3 Effect of the environment on the conformational pose of E2020....................47
3.4.4 Unbound docking study on E2020.................................................................48
3.5 Study on Decamethonium (DECA) .......................................................................50
3.5.1 Bound docking study on Decamethonium (DECA)........................................50
3.5.2 Docking decamethonium into the empty binding site of 1ACL......................50
3.5.3 Docking decamethonium into the binding site of 1ACL including waters......51
3.5.4 Unbound docking study on decamethonium (DECA) ....................................53
3.6 Application of MSD for bound docking study on several inhibitors of other proteins
than AChE........................................................................................................................54
3.6.1 Bound docking of Win51711 and Win52084.................................................54
3.6.2 Unbound docking study on Deoxythymidine.................................................56
3.6.3 Bound docking study of oxindole ..................................................................57
4 Elucidation of structural properties of (PPG)....................................59
4.1 Searching for the protonation state of PPG using bound docking...........................60
4.2 Conformation at galanthamine and piperidine substructures by unbound docking
study 65
4.3 Unbound docking of PPG protonated at the galanthamine-nitrogen (N10).............66
4.4 Unbound docking of PPG protonated at the piperidine-nitrogen (N22)..................67
5 Molecular docking studies on the “Back Door” hypothesis...............71
5.1 Docking study on protonation of DMPO...............................................................73
5.1.1 MSD method.................................................................................................73
5.1.2 G.O.L.D........................................................................................................75
5.2 Study on the location of leaving group in the absence of DMPO ...........................78
5.3 Study on the location of leaving group in presence of DMPO................................79
v
5.3.1 MSD method.................................................................................................79
5.3.2 G.O.L.D........................................................................................................81
5.4 Docking experiment on eseroline in presence of waters.........................................83
6 Binding mode prediction of galanthamine derivatives using MSD...87
6.1 Binding mode prediction of bisgalanthamide analogues into AChE binding site....87
6.2 Binding mode prediction of (IHG) into AChE binding site....................................89
6.3 Docking IHG in binding site of 1EVE...................................................................90
7 References.............................................................................................93
8 Abbreviations .......................................................................................97
1
Summary
Finding the correct binding mode of highly flexible ligands is a demanding approach in
molecular docking studies. With this respect, in some earlier efforts, the docking of these sorts
of inhibitors of AChE was not successfully accomplished. For instance E2020, BHG and
DECA using QXP+, FlexX, Auto Dock and G.O.L.D, which is even true for their docking into
the ligand-bounded protein. Furthermore, even finding the very final possible binding mode
of the ligand that forms the most stable complex with the protein was not confidently
obtainable. In this presented work, the difficulties of docking studies on very flexible ligands
were addressed by improvement of a consensus method that is based on QXP+ and local
Monte Carlo search algorithm to reduce the conformational search space. The method has
been developed using BHG and was successfully applied to other AChE ligands including
E2020 and DECA. Although, the procedure has been particularly developed for study and
prediction of the binding mode of AChE inhibitors, it was successfully employed for docking
the very high flexible inhibitors of other proteins like human Rhinovirus 14, Win52084
(1RUH), and Win51711 (1PIV). The method was also employed for the prediction of binding
mode of a few CDK2 inhibitors through docking them into the ligand unbounded proteins as
well as into their ligand bounded proteins, while the latter was performed to improve the
quality of complexation energy and better scoring of the ligand-protein binding mode to
estimate the inhibitory constant of them as a result of molecular docking study.
The method was also used for detailed study on the various AChE inhibitors such as:
- The possible protonation state for the ligand in its complexed state with AChE.
- Definition of the amino acids that are dominantly involved in binding the ligand.
- Importance and influence of particular water molecules on the conformational pose of the
ligand in the binding site.
- Selection of particular structural posture of the ligand in the case of PPG with two
crystallographicaly possible poses of the ligand in its conformationaly convertible piperidine
moiety.
- Prediction of the binding mode of a new derivative of galanthamine and elucidation of its
structural details in complex state with docking into a non-complexed protein.
- Study on the possibility of leaving the binding site of AChE through an alternative path way
based on “Back door” hypothesis that results finding only one pathway as the most possible
one, among four suggested back and side doors.
1 Introduction
2
1 Introduction
1.1 Alzheimer’s disease
Alzheimer’s disease (AD) is a slow progressive neurodegenerative disorder that clinically is
characterized by cognitive decline and defined by losing memory and learning ability. It also
decreases the ability of performing the basic daily activities. Other arrays of the disease are
apathy, verbal and physical agitation, anxiety, depression, delusions and hallucinations.
Currently, the loss of cholinergic function is an evidential finding responsible for cognitive
decline, hence therapeutical development has focused on this theory. Aging is often regarded
as the main factor in memory impairment and decline in other mental functions. Memory loss
and other neuro psychologics symptoms such as impairments of judgment, language, learning
and abstract thinking, which are descriptive of AD, may be attributed to normal aging. In this
context, the relationship between normal aging and AD is a debatable concept [1]. The fact
that AD increases with advancing the age, nowadays represents a major public health problem
and it is probably becoming the most important pathology of the 21st century in the developed
countries [2].
1.2 Cholinergic hypothesis
In cholinergic hypothesis is simply stated, that the cognitive loss associated with AD is related
to decreasing the cortical cholinergic neurotransmission. Therefore, it is as that increasing
cholinergic transmission may enhance cognitive function [3].
1.3 Approaches for treatment of AD’s
The most AD treatment methods are studied towards two main aims: the ß-amyloid peptide
(Aβ) and the cholinergic neurotransmission. There are thus two major approaches for
treatment of AD. Since Aβ is known as a component of the senile plaques [4], thus the first
approach of the treatment deals with preventing of formation of amyloid peptides or at least
decreasing their generation or deposition. Aβ is produced from proteolytic processing of
amyliod precursor protein (APP) through two pathways: one way is from cleavage of the
protein with α-secretase and generates soluble APP fragments (APPs). The second way is
cleavage of APP by β- or γ- secretases those are responsible of producing Aβ peptides (Figure
1.1).
1 Introduction
3
The trace amount of Aβ has been detected as a normal cellular metabolism of APP, therefore
it should be tried to prevent of more production and consequently deposition of the insoluble
amyloid plaques to prohibition of the disease progress [5].
Figure 1.1 The cleavage of α- β- and γ-Secretase (A. Khalid)
1.4 Cholinergic enhancement therapy
The cholinergic theory has provided the rational basis for therapeutic developments in AD.
Based on this theory and in accord with the mechanism of action in different parts of the
cholinergic neurotransmission system (Figure 1.2), six classes of drugs have been developed
to enhance cholinergic deficit in AD patient. These are [6]:
a. Cholinesterase inhibitors (ChEI), which block the AChE enzyme thereby stimulating
cholinergic activity to enhance cognitive function.
b. Choline precursors, such as phosphatidylcholine, aimed at increasing the
bioavailability of choline.
c. ACh releasers, which should facilitate the release of ACh from presynaptic end
terminals.
d. Nicotinic agonists or substances having nicotinic-like effects, which should enhance
ACh release.
1 Introduction
4
Figure 1.2: The mechanism of cholinergic neurotransmission (R. Bullock)
Among the above pharmacological agents, AChE inhibitors seem to be the most effective
method to improve cholinergic deficit for reducing the symptoms of the disease [6].
1.5 Structure of AChE
The principal biological role of acetylcholinesterase (AChE, acetylcholine hydrolyser, EC
3.1.1.7) is the termination of impulse transmission at the cholinergic synapses by rapid
hydrolysis of the neurotransmitter acetylcholine (ACh) [7]. AChE possesses a remarkably
high specific activity, especially for a serine hydrolyses [8]. Knowledge of the three-
dimensional structure of AChE is essential for understanding its remarkable catalytic efficacy.
1.6 Overview of the three-dimensional structure of AChE
The AChE monomer has an ellipsoidal shape, with dimensions of ca. 45 x 60 x 65 Å. The
subunits contain 11-standard β-sheets surrounded by 15 α-helices and there are also 3 short
standard β-sheets, which are not hydrogen-bonded to the central sheet [9] (Figure 1.3).
1 Introduction
5
Figure 1.3: Schematic ribbon diagram of the 3-D structure of T. californica AChE monomer. 11-standard β-
sheets (yellow) surrounded by 15 α-helices (red) and 3 β-sheets without any hydrogen bond to the central sheet.
ACh in the binding site has been rendered in CPK.
1.6.1 The active center and the catalytic triad
Glu327, His440 and Ser200 are the components of the catalytic triad located at the base of a
narrow gorge with 20 Å depth [10]. The gorge is lined with 14 aromatic residues, some are
deep within the gorge while most others define a large aromatic area on the wall of the gorge.
Two most important anionic amino acids are Asp72 and Glu199. Asp72 is located below the
rim of the gorge, while Glu199 located at the base of the gorge. Also several other anionic
residues are farther from the mouth of the binding site. Glu199 is the closest anionic side to
trimethylamonium group of ACh in its bonded state. Aromatic residues clearly play an
important role in stabilization of the complex. The choline moiety appears to be stabilized by
Trp84 and Phe330 in quaternary ammonium binding site, whose orbital are close to the
trimethylammonium surface as defined by its van der Waals’ radii [11] (Figure 1.4).
The carbonyl oxygen of ACh is stabilized through hydrogen bonding to amide back bone at
position of Gly118, Gly119 and Ala201 in oxyanion hole [10] (Figure1.5).
1 Introduction
6
Figure 1.4: Trp84 and Phe330 are the residues in quaternary ammonium site of the gorge, which interacts with
trimethyamonium moiety of ACh in its bonded state.
1.6.2 Peripheral anionic site
Trp279 and Tyr70 were introduced as the residues of peripheral anionic site (PAS).
Furthermore, two sets of residues (270-278 and 251-266 in TcAChE) contribute to the
peripheral anionic subsite, which are located near the rim of the gorge. Hence, ligand
association with the peripheral site may prevent access of substrate to the gorge by physical
hindrance to restrict entry to the gorge by an allosteric mechanism, in which the active center
conformation is altered [11] [12]. Recently, evidence was presented that AChE accelerates
assembly of amyloid-β-peptides into the amyloid fibrils with involvement of PAS [13].
Figure 1.5 indicates the location of the peripheral site, quaternary ammonium binding site as
well as oxyanion hole and catalytic triad within the gorge of AChE.
1 Introduction
7
Figure 1.5: Shows the binding site gorge of AChE and its most important amino acids (A. Khalid)
1.6.3 Acylation of ACh within the binding pocket
AChE’s physiological task is hydrolization of the cationic neurotransmitter. It acts through an
acylation and deacylation process [14]. As it is shown in Figure 1.6 and 1.7, Ser200 and
His440 in the active site are involved in the reaction with ACh. The acyl-enzyme is produced
after proton transformation from Ser200 to imidazole moiety of His440 and then oxygen of
Ser200 attacks to substrate (ACh), this part is acylation and then the acyl-enzyme is
hydrolyzed with waters of binding pocket. This deacylation process alters the enzyme to its
original form.
AChE + N+O
O
N+OH+ CH3CO-AChE
a
c
y
l
a
t
i
o
n
Figure 1.6: Indicates the involvement of ACh in acylation and deacylation reaction
1 Introduction
8
N
H
N
His
440
O
O
N
+
H
O
Ser
200
Gly
118
NH
Gly
119
HN
Ala
201
HN
H-Bond
ACh
N
Trp
84
Cation interaction
H
N
H
N
His440
OSer200
O
H
O
H
Gly118
NH
Gly119
HN
Ala201
NH
H-Bond
Acyl-AChE
Figure 1.7: Hydrogen-bonding and cation-
π
-interactions
In the acylation process the cation-
π
-interaction takes place between the positively charged
nitrogen in ACh and Trp84. Furthermore, the component of oxyanion holes (Gly118, Gly119
and Ala201) hydrogen bond to carbonyl group of ACh [14] (Figure 1.7).
In the acylation process, both the cation interaction and the H bonds exist during the reaction.
However, these two kinds of interactions indirectly affect the proton transfer from Ser200 to
His440 and the nucleophilic attack of the oxygen atom of Ser200 to the carbonyl of ACh
(Figure 1.7).
1.7 Acetylcholinesterase inhibitors (AChEI)
1.7.1 Synaptic cholinergic drugs
In tissues the most abundant of cholinesterase types are acetylcholinesterase (AChE) and
butyrylcholinesterase (BChE). As it was described earlier, AChE is the predominant one in
brain that is responsible for hydrolyzing ACh to acetate and choline, by which it terminates
the neurotransmitter effect at cholinergic synapses. Therefore inhibition of AChE causes more
bioavailability of ACh at synaptic area and consequently improving neurotransmission
process. Principally this method is more useful for treatment of patients with undamaged
presynaptic neurons those are still active for synthesizing and releasing ACh, thus it works in
early stages of AD and loses effectiveness after usage in a period of time. The activation of
M2 muscarinic receptors that leads to inhibition of presynaptic release of ACh might decrease
1 Introduction
9
the efficacy of acetylcholine inhibitors (AChEIs), through the counteracting effect. Despite of
this AChEIs have shown suitable therapeutic effect. The only drugs currently accepted to treat
the AD are AChEIs (i.e. tacrine, donepezil, rivastigmine and galanthamine) [15].
Based on the mechanism of action in AChE (described in section 1.2.4), different sort of
AChEIs have been designed and classified as pseudo-irreversible, irreversible, transition state
analogue inhibitors and reversible inhibitors.
1.7.2 Pseudo-irreversible AChEIs
This class of AChEIs includes the compounds having carbamates functional group. They are
carbamylated by catalytic triad of AChE binding site. The rate of hydrolization of their
carbamoylated complex with Ser200 is slower than the rate of hydrolization of ACh-AChE
complex. The first AChEI of this class that was studied for treatment of AD was
physostigmine, but because of the lack of efficacy resulting from its short half-life and
variable bioavailability it was rejected. To improve its potency, several analogues of that with
more lipophilic side chains have been designed. The miotine derivative (rivastigmine) is
another AChEIs in carabamate group (Figure 1.8) that is less potent than physostigmine and
inhibits also BChE. However it shows a good combination in brain selectivity, long duration
in vivo activity and its good neuro protective property, which caused it to be accepted as a
drug for AD treatment [19].
O
H
N
N
N
OH
O
H
N
O
N
physostigmine miotine
N
N
H
O
O
H
N
(CH
2
)
8
N
O
ON
O
N
MF268 rivastigmine (SDZ-ENA-713, Exelon)
Figure 1.8: Chemical structure of pseudo-irreversible inhibitors of AchE
1 Introduction
10
1.7.3 Irreversible AChEIs
Organophosphates are included in this group. One of the representatives of this group is
metrifonate. Although its efficacy was acceptable, its application as a drug was withdrawn
due to causing problems such as muscle weakness and respiratory problem in small
proportion of patients.
OPCl
Cl Cl
O
H
O
O
Figure 1.9: Chemical structure of metrifonate
1.7.4 Transition state analogue inhibitors
Trifluoromethylketones are effective inhibitors of this group having a reversible covalent
interaction with Ser200 of the active site forming a tetrahedral-hemiketal transition state [24].
In fact, among AChEIs, m-(N,N,N-trimethylamino)trifluoroacetophenone is a highly potent
reversible inhibitor (Figure 1.10), however its ionic nature prevents its capability to cross
blood brain barrier (BBB), therefore a more lipophilic and non-ionic derivative of that can
function better. One of the representations of this sort of inhibitor is zifrosilone (MDL-73745)
in Figure 1.10 that works as a transition state analogue form, as well.
N+F
FF
O
I
-
Si
F
F
F
O
Figure 1.10: Chemical structure of m-(N,N,N-trimethylamino)trifluoroacetophenone (left) and zifrosilone
(MDL-73745) (right) as transition state analogue AChEIs
1.7.5 Reversible AChEIs
In contrast to the three above described classes of AChEIs, reversible AChEIs binds to the
binding site of the enzyme and inhibit of the activity of substrate. Of this group
aminoacridines, N-benzylpiperidines and alkaloids are known.
1 Introduction
11
1.7.5.1 Aminoacridines
One of the members of this group is tacrine (Cognex), which was the first AChE inhibitor
approved by FDA in 1993. Interestingly, it is more potent to BChE than AChE, however it
has other features including blocking sodium and potassium channels and has a direct effect
on muscarinic receptors [19]. Some disadvantage of that are short half-life and induction of
hepatotoxicity. However it was the lead compound for synthesizing new derivatives, such as
velnacrine and suronacrine (Figure 1.11), those have shown reduced toxicity [20].
N
R
1
NH
R
3
R
2
R1=R2=R3=H, tacrine
R1=OH, R2=R3=H, velnacrine (HP-029)
R1=OH, R2=H, R3=benzyl, suronacrine (HP-128)
Figure 1.11: Chemical structure of tacrine, velnacrine and suronacrine
1.7.5.2 N-Benzylpiperidines
Donepezil (E2020) is the prototype of this structural class that was the second drug approved
by FDA to treat the mild to moderate type of AD. It is a potent, long-acting and selective
AChEI and shows this selectivity for AChE 1250 times more than for BChE. TAK-147
(Figure 1.12) is another N-benzylpiperidine derivative, which has less potency than donepezil
but its effect on animals showed fewer side effects and is undergoing clinical testing. Other N-
benzylpiperidine derivatives have been introduced, in which indanone moiety of donepezil
has been replaced by different heterocyclic systems [21], such as N-benzylpiperidine
benzisoxazoles (Figure 1.12). One of benzisoxazole derivatives is morpholino substituted one
that showed higher potency and selectivity than donepezil and was effective in animal models
[22].
1 Introduction
12
N
O
O
O
N
N
H
O
donepezil (E2020) TAK-147
N
N
O
N
O
N-benzylpiperidinemorpholinobenzisoxazole
Figure 1.12: Chemical structures of donepezil (E2020), TAK-147 and N-benzylpiperidinemorpholino-
benzisoxazole
1.7.5.3 Alkaloids
Galanthamine (Reminyl) is a tertiary amine alkaloid, isolated from Amaryllidaceae
(Galanthus woronowi, the Caucasian snowdrop), has been approved in several countries for
treatment of AD [23]. It is a reversible and competitive inhibitor of AChE and enhances the
response of nicotinic receptors to ACh, which causes increasing ACh release and other
neurotransmitters plus increasing bioavailability of ACh to inhibit AChE. To develop the
potency of galanthamine derivatives of it have been suggested, which are P11012 and P-
11149 (Figure 1.13) that are 10-fold more potent and 6-fold more selective than galanthamine
[24].
O
R
O
O
H
H
N
R= CH3, galanthamine
R= COCH3, P11012
R= CO (1-adamantly), P11149
Figure 1.13: Chemical structures of galanthamine, P11012 and P11149
1 Introduction
13
(-)-Huperzine A is another alkaloid, isolated from the Chinese medicinal herb Huperzia
serrata, is a very potent, selective and long-acting AChEI with low toxicity. It has been a lead
compound to design derivative such as the 10-methyl substituted that has 8-fold more potency
than (-)-huperzine (Figure 1.14).
C
10
NH
O
H
3
C
C
H
3
NH
2
Figure 1.14: Chemical structure of (-)-huperzine A
1 Introduction
14
1.8 Protein-Ligand docking
Molecular docking can be defined as the prediction of the structure of receptor-ligand
complexes, where the receptor is usually a protein or a protein oligomer and the ligand is
either a small molecule or another protein. Different simplifications are used to make
molecular docking useful in different applications. Initially, molecular docking was used to
predict and reproduce protein-ligand complexes [25].
Docking is defined by placing the putative ligand(s) in appropriate configurations for
interaction with a protein. Therefore in molecular docking, it is attempted to predict the
structure (or structures) of intermolecular complex, which is formed between two or more
molecules to suggest binding modes of protein inhibitors. Most docking algorithms are able to
generate a large number of possible structures and they also require a mean to score each
structure to identify, which are of the most interest. Recent efforts in structural biology have
led and will lead to growth numbers of compounds, which could be potential drug targets.
Because of the recent advances in docking algorithms, in silico screening provides an
attractive alternative to find suitable drug leads in vitro.
1.8.1 General procedure of docking
For a docking program, search algorithm and scoring function are the fundamental parts of the
docking programs. The docking process can be broken down into five phases, which are
modeling of the target, generating the possible conformations of the ligand, docking each
conformation, scoring each docked ligand and selecting candidate ligands for further
investigations [26]. High-resolution X-ray crystallographic structures are routinely used for
ligand docking. In addition, structures derived from NMR experiments or those predicted by
homology modeling may also be used. If the target protein is extracted from a structure, in
which a ligand is bound to the protein, then docking experiments using that structure is
termed “bound docking.” Otherwise, those experiments are termed “unbound docking” [27].
Because of the conformational changes that occur between the liganded and unliganded forms
of many proteins, performing bound docking is preferable to unbound docking that typically
yields better results than of the unliganded state in most of docking algorithms.
Not surprisingly, the most important part of the model is the protein’s binding site. By
homology-modeled protein, if the binding site is modeled poorly, virtual ligand screening will
not yield useful results [27].
1 Introduction
15
In docking of each conformation into the target, the goal is to quickly and correctly predict the
binding geometry of the ligand complexed with the protein. For this purpose, there are several
programs available, which differ in complexity, theoretical orientation, and implementation.
Performance of a docking algorithm is the most time-consuming step in docking process. All
modern docking algorithms used in structure-aided drug design, model the ligand as flexible,
though the receptor does not need to be set flexible.
Flexible docking algorithms can be easily categorized [28]. In this respect, the search
algorithms used in this project are shortly described
1.8.2 Docking Algorithms implemented in this collection
Program Search Algorithm Scoring Function
QXP (FLO+ 0802)
Monte Carlo Amber Force Field
FlexX (1.9.0) Incremental
Construction
Empirical Score
G.O.L.D (2.1.2) Genetic Empirical Score
Table 1.1: The programs used as the docking tools in this presented work
1.8.2.1 Incremental construction algorithm in FlexX
This search algorithm is performed in three steps, which are selecting a base fragment,
placing the base fragment in the binding site and building up the ligand inside the active site.
The base fragment (the ligand core) is selected and is placed into the binding site, using an
algorithmic approach based on a pattern recognition technique called pose clustering. In the
next step, the remainder of the ligand is built up incrementally from the fragments. The
construction method is so that the new fragment is added in all possible conformations to all
placements found in the previous iteration, but only the ‘n’ best placements are taken on to the
next construction step, generating multiple conformations for each fragment and including all
in the ligand building steps. It means that after finding a good set of placements, the
remaining portion of the ligand are divided into small fragments and incrementally grown on
to the base alternatives [29].
1.8.2.2 Genetic algorithm in G.O.L.D
A genetic algorithm (GA) involves a population of possible solutions through genetic
operations, such as mutation, crossover and migration. This is to reach to the final population
1 Introduction
16
of low energy conformations employing the energy function or the fitness function. For the
purpose of conformational sampling, the translational, rotational and the internal degrees of
freedom are encoded into ‘genes’, which are represented by the real number values (codes) of
those degrees of freedom [30]. Each conformation is named a chromosome, which consists of
a collection of genes and is represented by the suitable string of the real numbers. A fitness
value (energy) is assigned to each chromosome. The two most fundamental operators are
schematically shown in Figure 1.15. The mutation operator (a) changes the value of a
randomly selected gene by random value and the crossover operator (b) exchanges a set of
genes between two parent chromosomes, creating new gene. Additional operator is migration
operator, which moves individual chromosomes from one subpopulation to another in
different islands.
(a) (b)
Figure 1.15: Genetic operators used to create a population of children chromosome from a population of parent. (a): mutation
operator; (b): cross over operator
The crossover point is selected randomly, and the genes are exchanged between the two parents. Two children are created,
each having genes from both parents.
The advantage of GA is that it requires less iteration than Monte Carlo to generate a large
population of low energy conformation [31].
1.8.2.3 Monte Carlo Algorithm in QXP+
In a Monte Carlo searching method, starting from any given conformation the program
chooses a random number to decide what will be the next trial. In the case of a molecule, it
randomly selects a bond among several possible rotatable bonds (or torsion angle), by which
the molecular structure could be modified. It then randomly selects a new value for this
torsion angle from a predefined set of values. Multiple-torsion moves as well as Cartesian-
coordinates moves are among the many possible variations on this procedure. Once a new
trial conformation is created, it is necessary to determine whether this conformation will be
1 Introduction
17
accepted or rejected. If rejected, the above procedure will be repeated by randomly creating
string of conformations until one of them is accepted. If accepted, the new conformation
becomes the ‘current’ conformation, and the search process continues from it. The trial
conformation is usually accepted or rejected according to the ‘P’, probability of existence of a
conformation.
min[1,exp ]
U
P
β
− ∆
= Eq. 1.1
β is
( 1/ )
kT
β
=
and ∆U is the change in the potential energy. This means that if the energy of
the new trial conformation is lower than of that in the current conformation,
0
U
∆ <
, the new
conformation could be accepted. But even if the energy of the trail conformation is higher
than the current energy,
0
U
∆ >
, there should be an especial probability, proportional to the
Boltzmann factor, to be accepted. To find whether a higher energy trial conformation is
accepted, a random number r in the range
[
]
0,1
is selected and is compared to the ‘P’ in
equation 1.1. If
r P
<
, the conformation is accepted, otherwise it is rejected. According to this
principle of acceptance, the process continues for a long enough time, till a stationary solution
will be achieved (Equation 1.1).
1.8.2.4 Definition of the energy terms
All the non-bonded interaction energies represent the pair-wise sum of the energies of the
possible interaction between non-bonded atoms i and j of the ligand and protein, respectively
(Figure 1.16).
61 1
12
ijij
i
j j
i i
i j
ij
ijj
A
q q
E
r
B
rr
1 1
=++−
∑∑ ∑∑ Eq. 1.2
van der Walls (attraction)
van der Walls (repulsion)
Electrostatic term
1 Introduction
18
Figure 1.16: The graph indicates the variation of van der Walls interaction and repulsion (left) versus changes in inter-atomic
distances (right)
The non-bonded energy accounts for repulsion, van der Waals attraction and electrostatic
interactions (Equation 1.2). Van der Waals attraction occurs at short range, and rapidly dies as
the interacting atoms move apart by a few Angstroms. Repulsion occurs, when the distance
between interacting atoms becomes slightly less than the sum of their contact radii (Figure
1.16). Repulsion is modeled by an equation that is designed to rapidly blow up at close
distances (with r-12 dependency). These effects as a 6-12 equation are used in QXP+. The plot
in Figure 1.16 displays the circumstances, which cause the attraction and repulsion situation
between a pair of atoms. The "A" and "B" parameters control the depth and position (inter-
atomic distance) of the potential energy well for a given pair of non-bonded interacting atoms.
In fact, "A" determines the degree of "stickiness" of the van der Waals attraction and "B"
determines the degree of "hardness" of the atoms. The "A" parameter can be obtained from
atomic polarizability measurements, or it can be calculated quantum mechanically. The "B"
parameter is typically derived from crystallographic data to reproduce observed average
contact distances between different kinds of atoms in crystals of various molecules. The
electrostatic contribution is modeled using a Columbic potential. The electrostatic energy is a
function of the charge on the non-bonded atoms, their inter-atomic distance and a molecular
dielectric expression, which the later is counted to weaken the electrostatic interaction coming
from the environment (e.g. solvent or the molecule itself). Often, the molecular dielectric is
set to a constant value between 1.0 and 5.0. A linearly changeability and distance-dependency
of the molecular dielectric feature (i.e. 1/r) is sometimes used to account for the increase in
environmental bulk as the separator distance between interacting atoms.
2 Description of the consensus molecular docking approach
19
2 Description of the consensus molecular docking approach
2.1 The initial requirements
Quick eXPlore (QXP+) [32] was employed as a docking program that for the force field
parameters takes advantage of AMBER force field [33]. For docking experiments, proteins
retrieved from Protein Data Bank (PDB) were utilized, which are Torpedo Californica
acetylcholinesterase (TcAChE), kinase and rhinovirus type proteins. Due to the limitation in
QXP+ for using the number of atoms, which must be less than 2000, in the case of AChE a
spherical region of 20.0 Å radius around hydroxyl oxygen of Tyr121 of each protein crystal
structure was cut and using QXP+ the position of the hydrogens on the polar atoms were
optimized. Then the new subsets were utilized as the protein for docking experiments. In
addition, the structure of the ligand was built up and the hydrogens on the polar atoms were
added. Then it was minimized and utilized as the ligand for docking experiments.
2.2 Characteristic features of the consensus method
a. The target is represented by an experimental three-dimensional structure, which is not
necessarily the binding site of the ligand-bounded protein.
b. An additional scoring procedure (consensus method, described later) is combined with
scoring method of the program (QXP+) to filter the output data after each step of the
docking experiment.
c. The data filtration systematically considers all of the possible conformations (output
solutions after docking) with regard to various energy terms, such as total non-bonding
interaction, van der Waals, electrostatic, contact and positive van der Waals energy as
well as the number of hydrophobic interactions.
d. To decrease the calculation time, the protein is kept rigid through out the entire docking
experiment (except for simulated annealing), although there is no general limitation for
flexibility of the receptor active site in MSD.
e. It is particularly applicable to docking experiments involving highly flexible ligands
with large degrees of freedom.
2 Description of the consensus molecular docking approach
20
2.2.1 Aim of the consensus method
For docking a user-defined structure, firstly one step of full Monte Carlo (MC) search with
10,000 search-cycles was carried out. The docking result of this first step includes 25
conformations of the ligand, which are ranked according to the corresponding total estimated
binding energy of their complex with protein. Since among the 25 output answers of each
docking run, there are conformations with better values in different energy terms than of those
in the first rank answer, they thus must be also taken into account. This is to find out, if the
best docking solution, among the 25 generated answers of QXP+ can move to the first rank.
This is because the best solution is often in any rank but the first hit solution. To recognize the
best solution, it is necessary to give more searching time via enough additional runs to access
an energetically better binding mode. This is carried out, through a multi-step docking (MSD)
experiment. Therefore a particular scoring method is required for identification and ending the
last step of docking experiment to achieve:
a. The best binding mode of the ligand with regard to the various factors, such as different
energy terms and the number of hydrophobic interactions.
b. The best conformation at the first rank with the best total energy, by which the ranking
procedure of the main program is also taken into consideration.
And
a. To study whether the first hit conformation of the very final docking step can also reach
to global minima (as a sign of being native-like structure) [34] or it is trapped into local
minima on the energy hyper surface [35]. This is important to be assured that the ligand
in the very last obtained-binding mode is able to reach to better internal energy (ligand
energy).
b. In the case that X-ray structure of the native complex of the ligand is available, it should
be found out that how well the predicted conformation resembles the native structure,
which is done in accordance with calculating the root mean square deviation of the
docking solution (RMSD) with respect to the coordination of the ligand in X-ray
structure
2 Description of the consensus molecular docking approach
21
2.2.2 Multi-Step Docking (MSD) procedure
2.2.2.1 Data filtration
In data filtration the output numerical data of QXP+ are analyzed in a way that each of the 25
answers of a docking run gains one score if it has either the best value (Bv ) in a term or has a
value within an optimal range of that (Bv ± 2). The total score of each hit is calculated
according to equation 2.1.
ass nbd est cnt
Total E LE E vdW E E Nhph
vdW
S S S S S S S S S
+
= + + + + + + + Eq. 2.1
0 8
Total
S
≤ ≤
Where STotal is the final score of the hit, SEass, score of the total estimated binding energy, SLE,
score of the ligand energy relative to the global minimum, SEnbd, score of total non bonding
energy, SvdW, score of the van der Waals energy, SEest, score of the electrostatic energy, SEcnt,
score of the contact energy, SvdW+, score of the positive van der Waals energy; and SNhph, score
of the number of hydrophobic contacts. Finally the criterion for selection of a hit is the STotal.
Then the docking answer with the highest STotal is selected as the starting point of the next
iteration (see Table 2.1 and Table 2.2).
2.2.2.2 Search algorithm
Full Monte Carlo (MC) algorithm is always used in the first docking run of a Multi-Step
procedure (MSD). The resulting 25 answers of the first docking run then are analyzed as
described in data filtration and used as the starting point of the next iteration. In all the
following docking runs, local Monte Carlo searching method (LMCS) is applied [36] [37] in
which the rotational angle varies between 20°-30° degrees. This limits the conformational
search space and increases the probability of success in the random search and quickly
removes atomic clashes between the ligand atoms and the receptor. MSD with LMCS
algorithm is repeated until the first rank answer gains the highest STotal. Figure 2.1 summarizes
MSD procedure. In the case that X-ray structure of the ligand is available the accuracy of the
obtained solution can be tested by RMSD of the atom coordination’s in docking solution,
while
i
R
is the coordination of the atom in docking solution and
'
i
R
is the coordination of the
identical atom in the X-ray structure (Equation 2.2).
2 Description of the consensus molecular docking approach
22
' 2
( )
−
=∑
n
i i
i
R R
RMSD N Eq. 2.2
Figure 2.1: Flow chart of Multi-Step Docking approach
2.3 Application of MSD method for bound docking BHG
As an example of the application of MSD method, the procedure is described here for
reproduction of benzosulfimidohexylgalanthamine (BHG) complexed with AChE [38]. BHG
consists of three main moieties, a galanthamin, which is connected to the benzosulfimide
moiety via a hexyl side chain (Figure 2.2).
NO
YES
Docking experiment based on
full Monte Carlo
A user defined structure
Selection of the optimal hit
as starting point for the next iteration
Docking experiment based on local Monte Carlo
Data filtration based on value of van der Waals, electrostatic
and contact energies as well as the number of hydrophobic contacts
n times
repetition
Does the first rank hit have the highest total
score (S Total)?
Last answer
2 Description of the consensus molecular docking approach
23
O
O
OH
H
NN
S
OO
O
Galanthamine | Hexyl chain | Benzosulfimid
e
Figure 2.2: Chemical structure of BHG
At the first step of the multi-step bound docking of BHG, one step of full Monte Carlo (MC)
search was carried out. In view of the fact that among 25 answers of the first docking run,
there are still conformations with higher STotal than the first rank, these docking solutions
should all thus be taken into consideration. This is to find out whether they are capable to
show up in the first rank with more time given (search-cycle). According to the Monte Carlo
search algorithm the first rank solution has the highest probability with the best potential
energy, ∆U (Equation 1.1). Using this method for data filtration, the selected binding mode is
not necessarily the first rank, but rather the one with the highest possible value of STotal among
all the resulting solutions in the first step of docking (Equation 2.1and Table 2.1, 2nd rank).
From this step on, the local Monte Carlo search (LMCS) with restricted rotational angle
(between 20°-30° degrees) was used until the first rank answer also shows the highest STotal
Table 2.2, 1st rank).
The first step docking using the Full Monte Carlo search algorithm generates a docking
solution with a large magnitude of RMSD with value of 1.62Å, as the difference in the
location of X-ray structure and docking solution is observable in Figure 2.3 (left). Application
of MSD for docking BHG successfully refined the X-ray structure with RMSD of 0.64Å. This
result was obtained after performance of three steps of local Monte Carlo runs. The overlaid
structure of docking solution on the X-ray structure is shown Figure 2.3 (right).
2 Description of the consensus molecular docking approach
24
Table 2.1: Output numbers of the first step of bound docking of BHG using Full Monte Carlo search algorithm.
Rank number 2 is the optimal hit with the highest score of 4. The best value (BV) has been shown with red color;
numbers at interval value of (Bv±2) have been shown in yellow
*: Eass: Total estimated binding energy kJ/mol; LE: Conformation energy of the ligand kJ/mol; vdW: van der
Waals energy kJ/mol; vdW+: Positive van der Waals energy kJ/mol; Eest: Electrostatic energy kJ/mol; Ecnt:
contact energy of interactions kJ/mol; Enbd: Total energy of non-bonded interactions kJ/mol; Nhph: Number of
hydrophobic contacts; Nhyd: Number of hydrogen bonds; Score: Total score of each hit (STotal)
3 LMCS steps
2 Description of the consensus molecular docking approach
25
Table 2.2: Output numbers after three steps of bound docking of BHG using local Monte Carlo search
algorithm. Rank number 1 is the optimal hit with the highest score (5). The best value (BV) has been shown with
red color; numbers at interval value of (Bv±2) have been shown in yellow
Bound docking of BHG using MSD by employing LMCS generates the solution with RMSD
of 0.76 Å (right in Figure 2.3) that is better than using one step MC with RMSD of 1.72 Å
(left in Figure 2.3).
Figure 2.3: The overlaid structures of docking solution of the first step run (left) and the docking solution of the
last step docking experiment using MSD (right) on the X-ray structure (green).
2 Description of the consensus molecular docking approach
26
2.4 Study on the importance of data filtration by bound docking of BHG
For studying the effect of the selected starting point on the last result of MSD, two further
MSD experiments were performed by docking the ligand (BHG) into its complexed protein.
In experiment A, with Multi-Step Docking of First rank (MSDF), the first rank hit was always
the starting point. In experiment B, with Multi-Step Docking of Selected hit (MSD), the
solution with the highest STotal was always the starting point of the next iteration.
Experiments A and B were performed as MSD run. The value at the final LMCS result as
given in Table 2.3, which proves experiment B not only yields the better RMSD but also has
the higher STotal (red numbers in Table 2.3).
Docking
solution experiment RMSD(Å) E ass LE vdW Ecnt Eest vdW+ Nhph Nhbd
1
A (MSDF) 1.63 -57.1 4.8 -5.5 -23.2 -33.1 9.2 15 2
2 B (MSD) 0.64 -57.7 6.6 -10.4 -15.8 -38.1 7.7 14 3
Table 2.3: Numerical output data of experiment A, with Multi-Step Docking of First rank (MSDF) and
experiment B, with Multi-Step Docking of Selected hit (MSD)
As illustrated in Figure 2.4 (upper), the third step of the experiment B produces the best
answer with the lowest total estimated binding energy. By running additional steps, after third
one, the absolute magnitudes of the energies as well as the rank of the highest scored hit
(STotal) are increased. From the third step on, if the experiment, by purpose additionally be
continued, in the sixth step, it reaches almost to the same value as the third step of the
experiment B (MSD). Figure 2.4 (upper) demonstrates that in experiment B the values of
energies (MSD, blue bars) are decreasing until experiment 3 with very low RMSD 0.64 Å
(lower diagram in Figure 2.4), whereas experiment A (MSDF) reaches to the lowest value of
the energy in the experiment 2 of multi-step docking with a high value of RMSD 1.63 Å
(lower diagram in Figure 2.4).
2 Description of the consensus molecular docking approach
27
-65
-60
-55
-50
-45
-40
-35
-30
1 2 3 4 5 6
Experiment number
Total estimated energy (kJ/mol)
Total estimated MSD
Total estimated MSDF
0
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
Rank number
&
RMSD
1 2 3 4 5 6
Experiment number
Rank number MSD
RMSD MSD
Rank number MSDF
RMSD MSDF
Figure 2.4: The variation of the total estimated binding energy (the upper plot), rank number and RMSD (the
lower plot) when applying multi-step docking of BHG into its corresponding protein, using first rank answer
(Experiment A or MSDF) and selected hit (Experiment B or MSD) for starting point
The differences among the RMSDs in experiment A (MSDF) and B (MSD) at the lower
diagram in Figure 2.4 (red and yellow squares), demonstrate the priority of using MSD with
selected conformation (see height-difference among red and yellow squares in the lower
diagram, Figure 2.4). This means using the first rank answer for MSDF predicts the wrong
2 Description of the consensus molecular docking approach
28
answer. Since in MSDF always the first rank hit is taken for the next iteration, therefore there
is no particular way to recognize the step, which provides the best possible docking solution.
In other word, for MSDF there is no limitation for the number of docking steps
2.5 Unbound docking study on BHG
To study if the Multi-Step Docking procedure is applicable as a predictive method, the
unbound docking experiment of BHG was carried out into 1EVE [39].
1EVE is the X-ray structure of TcAChE complexed with E2020 that was selected as the target
protein. This was because E2020 is a long ligand that occupies a large space in binding site of
AChE, which seems to be suitable enough for accommodation of BHG, as well.
In the case of docking BHG in the binding pocket of 1EVE, a total of 12 steps were
necessary, 1 full and 11 local Monte Carlo runs, to generate the final answer (Figure 2.5). The
detailed energy values and the calculated STotal of all the hits from the first and the twelfth step
are shown in Table 2.4 and Table 2.5, respectively. The docking experiment was stopped,
when the first rank docking solution has the highest STotal and no improvement in the ligand
energy was observed (experiment Nr. 12 in Figure 2.6), such that the hit with the highest STotal
(Table 2.4, 5th rank) moves to the first rank (Table 2.5, 1st rank). In accord with the definition
of the Boltzman factor, the final answer should have the highest population among all
binding-modes. The total estimated binding energy and the ligand energies of the 13 runs are
shown in Figure 2.6 along with the corresponding rank number.
2 Description of the consensus molecular docking approach
29
Table 2.4: Outputs of the first 25 hits, after first step full Monte Carlo. Rank number 5 is the selected hit with
the highest score (5). ; Best value (BV) has been shown with red color; numbers at interval value of (Bv±2) have
been shown in yellow
*: Eass: Total estimated binding energy kJ/mol; LE: Conformation energy of the ligand kJ/mol; vdW: van der
Waals energy kJ/mol; vdW+: Positive van der Waals energy kJ/mol; Eest: Electrostatic energy kJ/mol; Ecnt:
contact energy of interactions kJ/mol; Enbd: Total energy of non-bonded interactions kJ/mol; Nhph: Number of
hydrophobic contacts; Nhyd: Number of hydrogen bonds; Score: Total score of each hit (STotal)
11 LMCS steps
2 Description of the consensus molecular docking approach
30
Table 2.5: Output numbers after eleven runs with local Monte Carlo. Rank number 1 is the optimal hit with the
highest score (4) and the ligand energy (LE) of 0.0 kJ/mol; Best value (BV) has been shown with red color;
numbers at interval value of (Bv±2) have been shown in yellow
*: Eass: Total estimated binding energy kJ/mol; LE: Conformation energy of the ligand kJ/mol; vdW: van der
Waals energy kJ/mol; vdW+: Positive van der Waals energy kJ/mol; Eest: Electrostatic energy kJ/mol; Ecnt:
contact energy of interactions kJ/mol; Enbd: Total energy of non-bonded interactions kJ/mol; Nhph: Number of
hydrophobic contacts; Nhyd: Number of hydrogen bonds; Score: Total score of each hit (STotal)
Figure 2.5: The structure of the first rank answer of one step MC from docking BHG into 1EVE with STotal of 5,
colored in gray (left), the structure of the first hit with the highest score (4) obtained after 12 steps LMC colored
in gray (right) each overlaid on X-ray structure (colored in green)
2 Description of the consensus molecular docking approach
31
-57
-54
-51
-48
-45
-42
-39
-36
-33
-30
-27
-24
-21
-18
-15
-12
-9
-6
-3
0
3
6
9
12
1 2 3 4 5 6 7 8 9 10 11 12 13
Experiment Number
Total estimated energy
Ligand energy
RMSD
Ligand
energy
(kJ/mol)
&
Rnak
number
Total
estimated
binding
energy
(kJ/mol)
Rank number
Total estimated
binding energy
Figure 2.6: Displays the variation of total estimated binding energy and ligand energy with the corresponding
rank number, resulting from multi-step docking of BHG into the binding site of 1EVE. The minimum value was
obtained in the 12th step, where the answer with the highest STotal appeared in the first rank
2.6 Importance of selection a suitable protein for unbound docking
To study the effect of selection a suitable X-ray structure of protein for unbound docking the
next set of experiments were performed, for which, firstly three different TcAChE crystal
structures (1DX6 [40], 1QTI [41], 1EVE [39] were chosen as our target proteins. Among
these three X-ray structures only 1EVE is the one with a long ligand that covers almost the
entire space of the AChE-gorge, while 1QTI and 1DX6 are acetylcholinesterase complexed
with galanthamine, which are located in the bottom of the deep gorge.
Secondly, BHG was built up in two states at galanthamine nitrogen with axial and equatorial
conformations. Then the two prepared ligand structures were individually docked into the
binding site of three different TcAChE crystal structures.
This was to identify the favorable conformation of the galanthamine nitrogen as well as the
conformational pose of the entire ligand, independent of availability of crystal structure of the
ligand-bounded protein. After running six individual MSD, the best solution of each one was
taken as the final answer for the purpose of evaluation. Firstly, comparing the result of the
experiments show that docking result of the ligand with axial conformation has higher STotal
than the ligand with equatorial conformation at galanthamine nitrogen (red numbers in Table
2.6 & 2.7). It means the binding mode, associated with ligand having axial conformation is
more preferable than equatorial one. Therefore three entries, correspond to axial ligand,
2 Description of the consensus molecular docking approach
32
(Docking solutions 3, 5 and 7, Table 2.6, Figure 2.7) should be taken into particular
consideration.
Docking
solution
protein Eass
LE
Enhb
vdW
Eest
Ecnt
vdW+
Nhph Nhb
(3) a 1QTI -60.2 1.2 -61.4 -14.1 -10.7 -36.6 3.9 15 1
(4) e 1QTI -55.6 4.3 -59.8 -8.7 -15.4 -35.8 8.9 9 3
(5) a 1DX6 -64.3 3.9 -68.2 -6.7 -23.8 -37.8 11.5 12 3
(6) e 1DX6 -60.7 4.0 -64.7 -8.1 -18.6 -38.1 10.0 13 3
(7) a 1EVE -55.1 0.0 -55.1 -3.4 -12.1 -39.6 14.7 14 2
(8) e 1EVE -50.1 0.9 -51.0 -4.1 -16.5 -30.4 10.5 9 2
Table 2.6: Numerical data obtained from docking BHG with axial and equatorial conformation
Docking Solution RMSD (Å) Conformation Orienation of saccahrin ring
(3) a 0.85 a Inverted
(4) e 1.94 e Off
(5) a 1.79 a Off
(6) e 2.15 e Off
(7) a 0.76 a Correct
(8) e 3.07 e Off
Table 2.7: Illustrates the connection of the conformational state at galanthamine nitrogen to the ring orientation
in saccharin moiety of BHG
For this class of the unprotonated BHG (all axial), the conformational pose of the ligand from
docking the BHG into 1EVE is suggested as the best docking answer. This solution is the
only one with the minimum value of intra molecular energy of the ligand (0.0 kJ/mol, Table
2.6), having high possibility of being the native-like pose of the ligand in complex form [34]
(red structure in Figure 2.7). RMSD calculation of the final solution with respect to X-ray
structure of the ligand gives a value of 0.76 Å that confirms the accuracy of the unbound
docking result. It is also the only answer with correct orientation of saccharin ring among the
other docking solutions ( Docking solution 7, Table 2.7).
2 Description of the consensus molecular docking approach
33
Figure 2.7: Obtained solutions of docking the axial BHG into 1DX6 (magenta), red (1EVE) and gray (1QTI), X-
ray structure is green
The result of this experiment proves the significant role of the selected X-ray structure of the
target protein, which should have a bounded ligand, whose occupied space in binding site be
large enough for the desired ligand for the purpose of unbound docking.
2.7 A comparison between the result of unbound docking using simulated annealing
(SA) and using MSD
In this step the selected conformation from the first step of the Monte Carlo Search (MCS),
which was the 5th rank answer in Table 2.4 was utilized for the purpose of docking into a
flexible binding site, through which the temperature allowed to be changed. In this
experiment, temperature and flat-well radius (allowed radius for movement of binding site)
must be carefully defined. Since this method mimics physical annealing, the key feature is
reduction of temperature very slowly to achieve optimal solution with global minimum
energy or to be trapped in good minima [35]. For the purpose of having a successful
annealing, the optimal highest temperature should be found. Therefore, four individual
docking runs with different starting temperatures were performed at 600, 500, 400 and 300 K.
Then with a flat well radius of 0.1 Å, the system cooled down to 30 degrees. The results of the
experiments show that the highest STotal was obtained when the annealing temperature was
defined to 600 K (Docking solution 12, Table 2.8). This means the optimal temperature for
further annealing experiments is 600 K.
2 Description of the consensus molecular docking approach
34
Docking
solution
Temp
(K) RMSD (Å) E ass LE Enbd vdW Eest Ecnt vdW+ Nhph Nhbd
9 300 1.90 -59.2 8.5 -67.8 -9.9 -20.6 -37.4 9.5 14 3
10 400 1.85 -59.8 9.2 -68.6 -7.0 -24.7 -36.9 10.6 13 3
11 500 0.88 -60.3 9.2 -69.7 -12.4 -18.3 -39.0 6.1 16 2
12 600 0.85 -60.6 9.1 -70.0 -12.5 -18.4 -39.0 6.4 16 2
Table 2.8: The numerical result of simulated annealing started from four different temperatures
To reveal the effect of a bigger flat-well radius, next experiment was performed with a 0.2 Å
radius. The result again shows the higher STotal for docking solution of 0.1 Å (Docking
solution 12, Table 2.9) than the one with 0.2 Å radius (Docking solution 13, Table 2.9).
Docking
solution
Flat-well
(Å)
Temp
(K) RMSD (Å) E ass LE Enbd vdW Eest Ecnt vdW+ Nhph Nhbd
12 0.1 600 0.85 -60.6 9.1 -70.0 -12.5 -18.4 -39.0 6.4 16 2
13 0.2 600 1.89 -65.2 9.0 -70.9 -7.8 -24.7 -38.4 10.5 13 3
Table 2.9: The numerical result of simulated annealing in 600 K with two different flat-well radiuses
According to the result of this experiment, the best solution was obtained at 600 K with 0.1 Å
flat-well radius. Comparison of this solution with the result of the docking solution using
MSD method reveals that the docking solution 12 in Table 2.9 does not correctly mimic the
X-ray structure of the ligand in the benzosulfimide substructure of the BHG, although its
RMSD obtained from QXP+ has a value of 0.85 Å (left in Figure 2.8), which has a wrong
orientation in the benzosulfimido moiety of the ligand, while the result of MSD is correctly
superimposed on the X-ray structure with RMSD of 0.76 Å (right in Figure 2.8).
Figure 2.8: The answer of SA and MSD, each overlaid on X-ray structure (green). As can be seen the result of
SA (left) has the wrong orientation in benzosulfimide moiety in the upper part of the ligand, whereas MSD
results a correct pose of the ligand (right).
2 Description of the consensus molecular docking approach
35
2.8 Comparison of the MSD result with various other docking methods
As is indicated in Table 2.10 and Figure 2.9, among five different docking method using three
programs, FlexX, G.O.L.D, QXP+ (one step, simulated annealing and multi-step docking), the
best answer is obtained when MSD method was employed (Table 2.10).
Program Search Algorithm Scoring Function *RMSD (Å)
Binding Site
Flexibility
QXP+
(OSD)
Monte Carlo
One Step
Amber Force Field 2.35 No
FlexX Incremental
Construction
Empirical Score 2.82 No
G.O.L.D Genetic Empirical Score 10.13 No
QXP+
(SA)
Monte Carlo
Simulated Annealing
Amber Force Field 1.71 Yes
QXP+
(MSD)
Full and Local
Monte Carlo
Multi-Step Docking
Amber Force Field 0.79 No
Table 2.10: A comparison of docking results from FlexX, G.O.L.D and QXP+
*: RMSDs calculated by “DAG. v. 2” as a reference program
2 Description of the consensus molecular docking approach
36
QXP+ (SA) FlexX
QXP+(OSD) G.O.L.D
QXP+ (MSD)
Figure 2.9: Docking solutions obtained from different docking programs (gray) overlaid on the X-ray structure
of the BHG (green)
3 Evaluation of MSD
37
3 Evaluation of MSD Consensuses Approach
3.1 Application of MSD for elucidation of structural properties of (BHG)
As it was described in chapter 2, BHG is one of the galanthamine derivatives. Glanthamine
was first isolated from snowdrops [42] and has long been known to exhibit esterase-blocking
activity [43]. Jordis and Froehlich [44] have developed a sterospecific synthesis for a large-
scale production of galanthamine that opened the door for therapeutic use of this compound,
so that galanthamine is now commercially available as the fourth anti-Alzheimer drug.
Galanthamine is known not only to block the esterase activity, but also to display an allosteric
potentiating effect on the nicotinic acetylcholine receptor (nAChR) [45]. This finding makes
galanthamine a most valuable drug for Alzheimer treatment as this dual interaction enhances
signal transmission of acetylcholine by twofold.
Inestrosa and coworkers [46] [47] have demonstrated that AChE, besides its established
esterase cleavage activity, is also involved in the formation of β-amyloid plaques, which are
known to play an essential role in the development of Alzheimer’s dementia [48] [49].
Aggregation of this peptide seems to occur specifically at the peripheral anionic site (PAS)
[50] [51] of AChE. More specifically, a 35 amino acids Torpedo derived peptide fragment
corresponding to AChE-sequence position 274-308 is capable of amayloid complex formation
[52]. However, Trp279 and Tyr70 play the most important roles in PAS located at the mouth
of the gorge [53].
As it is shown in Figure 3.1, galanthamine binds to a region of the active site that is not
connected to PAS, therefore there is no inhibition of β-amyloid plaque formation through
application of galanthamine. In this respect, to improve capability of galanthamine to interact
simultaneously with amino acids in the upper and the lower part of the gorge, new derivatives
of galanthamine with long side chain are necessary to cover the entire gorge of AChE. With
these aims molecular docking study have been reported on galanthamine derivatives, such as
bis-galanthamines with varying lengths of a methylen spacer between two galanthamine
moieties [54] as well as benzosulfimidohexylgalanthamine (BHG) [38].
3 Evaluation of MSD
38
Figure 3.1: Galanthamine (green) is located in deep of the gorge without any interaction in rim of the gorge.
Trp279 and Trp84, rendered in CPK, are located in PAS and quaternary ammonium subsite of the AChE binding
site.
In this section the result of multi step docking of BHG will be presented, which was carried
out to study on the possibility of the protonation in galanthamine moiety of the ligand. In this
experiment the ligand is docked into the AChE-BHG complex [38] (bound docking). The
obtained complex energy of the BHG-AChE will be compared to galanthamine and the other
inhibitors of AChE. Furthermore, the most important water molecules will be identified and
their contribution into the docking process will be investigated.
3.2 Bound docking experiment on BHG
Investigation on the BHG’s binding mode was firstly carried out into the empty binding site
of the AChE-BHG complex.
Docking*
solution
Waters RMSD
(Å)
Eass
LE
Eshe Enhb vdW Eest Ecnt vdW+ Nhph Nhb OSR**
1 (p) empty 1.65 -56.2 7.0 0.0 -63.2 -6.6 -20.8 -35.8 10.2 13 3 off
2 (up) empty 0.64 -57.7 6.6 0.0
-64.3 -10.4 -15.8 -38.1 7.7 14 3 Correct
Table 3.1: Numerical results of docking protonated and unprotonated ligand into empty binding site
*: ‘p’ stands for protonated and ‘up’ for unprotonated ligand
**: ‘OSR’; stands for orientation of saccharin ring
3 Evaluation of MSD
39
The result of bound docking of BHG with protonated and unprotonated galanthamine into
empty binding site is more in favor of unprotonated state as it shows that the unprotonated
BHG has a lower RMSD 0.64Å (Docking solution 2, Table 3.1), when the protonated ligand
has much higher RMSD of 1.65Å (Docking solution 1, Table 3.1).
3.3 Bound docking of BHG into binding site with waters
Furthermore, using QXP+ six water molecules around BHG molecule was identified in the
AChE gorge as the most important ones that are Wat17, Wat66, Wat121, Wat149, Wat289,
and Wat230. These water molecules mediate interactions between amino acids in binding site
and BHG atoms. In Figure 3.2, these waters are shown as red balls, while small blue balls
represent all other water molecules in the protein. There are a total of 15 water molecules
buried in the binding site of BHG (Figure 3.2, red and yellow balls). For comparison, Figure
3.2 (right) show six water molecules from the AChE complexed with ACh (2ACE) that
correspond to the six most important waters in AChE-BHG complex, Wat603, Wat628,
Wat643, Wat682, Wat719, Wat749 colored in cyan. The average displacement of them has a
value of 1.21Å. Moreover, in the binding site of the native enzyme, 2ACE, there are 36 buried
waters (middle in Figure 3.2) [55]. In Figure 3.2 (left and middle) the waters close to the wall
of the active-site gorge (yellow balls) and their neighboring amino acids in 2ACE have been
shown.
Comparing the position of these buried waters in two crystal structures shows that waters in
2ACE, close to Tyr70, Asp72, Gly118 and Glu199 are approximately in the same region as
are in BHG (left and middle in Figure 3.2).
3 Evaluation of MSD
40
Figure 3.2: BHG with buried waters (yellow) and the six important waters included in docking (red), while the
rest of the crystallographic waters are shown in blue (left); ACh, complexed with AChE (2ACE) with buried
waters (yellow), two important waters interacting with ACh are in red, the rest of waters in X-ray structure have
been colored in blue (the middle); Overlaid six important water molecules of the binding site of BHG (red) on
their comparable waters in 2ACE coloured in cyan (right)
By contribution of the six important water molecules in bound docking of BHG, further steps
of docking study were performed to analyze the effect of them on the possibility of
protonation of galanthamine moiety in BHG. In this context, BHG in two different protonated
and unprotonated states were individually docked into the binding site of the complexed
protein. This means that two independent multi-step docking experiments were performed, in
which either water hydrogens were defined flexible. The numerical results of the docking
experiments are shown in Table 3.2.
From the result of docking the protonated BHG appears that for protonated ligand, in the case
of docking into empty binding site, the docking solution is far from X-ray structure of the
ligand (RMSD= 1.67 Å), while docking solution of the binding site with flexible water
hydrogens has an upside down pose in binding site ( RMSD= 12.26 Å) ( Docking solution 3,
Table 3.2).
In contrast, docking experiments on the unprotonated ligand generates answers with RMSDs
less than 1.0 Å, in which one answer has only an inverted saccharine ring (Docking solution
4, Table3.2). In addition docking of both unprotonated and protonated ligands into the binding
site with flexible waters generate correct answer with a RMSD lower than 1.0 Å for
unprotonated ligand, while docking solution of the protonated BHG has upside down pose in
binding site (data are not shown). Therefore the results are more in favor of the unprotonated
state of the BHG rather than protonated one (Table 3.2).
3 Evaluation of MSD
41
Docking*
solution
Waters RMSD
(Å)
Eass
LE
Eshe Enhb vdW Eest Ecnt vdW+ Nhph Nhb OSR**
3 (p) Flexible
hydrogen
12.26 -40.1 4.1 -0.1 -44.1 -4.9 -21.8 -17.3 3.8 4 2 off
4 (up) Flexible
hydrogen
0.90 -63.3 3.7 7.1 -74.0 -5.5 -26.6 -41.9 15.0 12 5 Inverted
Table 3.2: Numerical results of docking protonated and unprotonated ligand in presence of waters in binding site
*: ‘p’ stands for protonated and ‘up’ for unprotonated ligand
**: ‘OSR’; stands for orientation of saccharin ring
Possibility of protonation of an atom strongly depends on its position in the ligand structure
and on its particular position with respect to the neighboring amino acids in the binding site.
Depending on the residues surrounding the ligand, the pKa value could vary (Figure 3.3) [56]
[57] [58].
Figure 3.3: Impact of the protein environment on the pKa values of a basic ligand group (upper row) and pKa
values of an acidic ligand group (lower row) compared to aqueous solution [58]
Generally, weaken in the basicity of the ligand in its complexed state can occur, because of
the influence of surrounding aromatic amino acids (Figure 3.3). It can be also observed in
Figure 3.4 that galanthamine moiety is surrounded by aromatic amino acids of binding site,
which is very similar condition to the Figure 3.3 (in red square). The pKa value of
galanthamine in aqueous solution is 8.2 [59]. However, according to its position as a
substructure in BHG and location of that in neighborhood of aromatic residues of the gorge, it
loses some basic strength. This confirms the finding in previous step, as a result of docking
study on BHG, the unprotonated BHG is the reliable state.
3 Evaluation of MSD
42
Figure 3.4: Shows the belt-like location of aromatic amino acids surrounding BHG, which affect on the pKb
value of galanthamine and its protonation state
Result of docking the galanthamine into 1QTI also confirms our finding in previous step,
because it shows that unprotonated galanthamine is also the dominated state of the ligand
(Table 3.3).
In docking protonated and unprotonated galanthamine into 1QTI, four most important
crystallographic water molecules, Wat712, Wat756, Wat764, Wat803, were also included, in
which waters were flexible, to consider whether the presence of the waters around the
galanthamine allow that galanthamine becomes protonated at all.
Docking*
solution
Waters RMSD
(Å)
Eass
LE
BE Enhb vdW Eest Ecnt vdW+ Nhph Nhb
5 (up)* Flexible 0.25 -45.9 0.6 0.1 -46.6 -6.3 -18.1 -22.2 5.2 4 2
6 (p) Flexible 17.63 -40.4 0.2 0.0 -40.6 1.4 -36.1 -5.9 4.5 0 3
Table 3.3: Result of docking galanthamine into 1QTI
*: ‘up’ stands for unprotonated and ‘p’ for protonated
In docking experiment on galanthamine with protonated nitrogen, when the waters are
flexible, the docking solution of the ligand is placed completely out of the binding site
3 Evaluation of MSD
43
(Docking solution 6, Table 3.4, whereas unprotonated galanthamine has a docking solution
with a very low RMSD (0.25Å) in docking solution 5 in Table 3.3.
The results of docking galanthamine clear that in presence of the effective water molecules of
binding site, there is no chance for protonated galanthamine to complex with protein within
the active site gorge.
To study the efficacy of different inhibitors of AChE to interact with the binding site, the total
non bonding interaction energies of them can be compared. Since the ligands are different
therefore the ligand energy is subtracted from total estimated binding energy, which results
total nonbonding interaction energy (Enbd). These data are shown in the following plot. The
five different inhibitors of AChE were chosen for the purpose of comparison, which are
E2020, decamethonium, galanthamine, piperidinopropylgalanthamine (PPG) [38] and BHG,
among which the most stable complex corresponds to BHG (Figure3.5).
-70.0
-65.0
-60.0
-55.0
-50.0
-45.0
-40.0
-35.0
1 2 3 4 5
Inhibitor
Enbd (kJ/mol)
Figure 3.5: Displays difference in total non-bonded interaction energy among various inhibitors of AChE,
according to their corresponding docking results.
1; Decamethonium, 2; E2020, 3; Galanthamine, 4; PPG, 5; BHG
The low complexation energy is not the only important feature of an inhibitor, but also it
should be able to cover entire active-site gorge and interact with PAS amino acids in the rim
of the active-site gorge in AChE. Figure 3.6 illustrates the overlaid docking solution of E2020
(see section 3.4), PPG (see chapter 4), BHG and galanthamine. Galanthamine only interacts
with amino acids in the base of the gorge, while another derivative of galanthamine, PPG with
a propylpiperidino substructure has a bent pose that prevents PPG of any access to the rim of
the gorge and PAS, whereas BHG has a suitable ring stacking interaction with Trp279 with -
3 Evaluation of MSD
44
20 kJ/mol interaction energy, which works out better than of that in E2020 with interaction
energy of -16.9 kJ/mol. Since amino acids in PAS (Trp279 & Tyr70) are believed to be
responsible for β-amyloid plaque formation a suitable inhibitor of AChE would be thus more
valuable if it can interact with PAS in the rim of the gorge. Therefore BHG seems to be
valuable inhibitor of AChE.
Figure 3.6: Displays galanthamine (white); E2020 (red), PPG (blue) and BHG (green). Trp279 and Tyr70 are
shown in yellow color
In summary BHG is a more potent inhibitor of AChE than other AChE inhibitors such as
E2020, decamethonium, galanthamine and its derivative PPG. This is according to their
relative binding energies, obtained from docking experiments of each above-mentioned
ligand. Docking solutions in presence the water molecules in the binding site suggest that
unprotonated ligand is the favored state for binding BHG to AChE. Employing MSD for the
purpose of unbound docking led to a successful prediction of the correct conformational pose
by docking BHG into 1EVE. Furthermore, it demonstrated the axial conformation of the
ligand generates better answer than equatorial conformation with respect to STotal and RMSD
values. Comparison of the final docking solution with the crystal structure gives the RMSD
value of 0.76 Å. The successful result of MSD method makes it a promising approach for
further application of that to design more potent inhibitors.
3 Evaluation of MSD
45
3.4 Docking study on E2020
The first two drugs, approved by FDA, for treatment of AD were tacrine (THA) and the more
potent inhibitor of AChE, E2020, which are both reversible inhibitors of AChE.
E2020 with trivial name of donepezil hydrochloride was approved in 1996 [60], which the
three-dimensional structure (3D) of its complex with TcAChE deposited in Protein Data Bank
has been coded with 1EVE [39]. Figure 3.7 indicates that the chemical structure of the
molecule is constructed by three main components, which are dimethoxyindanone, piperidine
and benzyl moieties.
N
14
C
8
O
O
O
Dimethoxyindanone | Piperidine | Benzyl
Figure 3.7: Chemical structure of E2020
Atom at position number 8 is a chiral carbon. The reported pharmacological studies on (R, S)-
E2020 emphasize that both enantiomers are active and have similar pharmacological profiles
[61]. The S configuration of this inhibitor shows five fold more inhibitors constant 17.5 nM
versus 3.35 nM inhibitory constant for R-configuration [62]. Using docking technique and
MSD method, the configurational state of the molecule at C8 as well as the possibility of
protonation at N14 are taken into consideration.
3.4.1 Study on the configuration state of the molecule at C8
For the molecule, two individual series of MSD experiment were carried out, in which the
ligand was considered with two possible configurations (R and S) at C8. It was docked into
the binding site of the enzyme after removing any crystallographic water molecules. The
numerical results of the last step of experiment are shown in Table 3.4.
3 Evaluation of MSD
46
Docking*
Solution
RMSD
(Å)
Eass LE Enbd vdW Eest Ecnt vdW+ Nhph Nhb
7- (up)-R 0.89 -47.1 2.0 -49.1 -12.6 -5.7 -30.8 2.8 11 1
8-(up)-S 2.38 -36.8 4.5 -41.2 -10.8 -1.1 -29.3 3.0 13 0
Table 3.4: Shows the numerical result of MSD of the inhibitors in two individual experiments in which E2020
has either R or S configuration at C8.
*: ‘up’ stands for unprotonated
The calculation results show that the binding mode of E2020 with R configuration has higher
STotal than E2020 with S configuration (red numbers in Table 3.4). The RMSD of the former
(0.89 Å) is also less than the later (2.38 Å), which are docking solution 7 and 8 in Table 3.4,
respectively (Figure 3.8).
Figure 3.8: The overlaid X-ray structure of E2020 (green) on the docking solution of E2020 with S
configuration (left) and the solution with R configuration (right)
Now, it should be studied that what would be the result, if E2020 were protonated at N14.
To answer this question some further experiments were carried out with protonated ligand at
N14.
3.4.2 Study on the protonation state at E2020-Nitrogen
To answer the question of the possibility of protonation at E2020-nitrogen, two series of
individual MSD runs were performed. In these experiments E2020 was protonated at N14 and
the configuration at C8 was set to R or S, respectively. As it can be seen from Table 3.5 the
protonated ligand with S-configuration should be rejected, on the basis of its corresponding
RMSD value (10.48 Å) and even the protonated ligand with R-configuration that shows
RMSD= 2.02 Å (Docking solution 9 in Table 3.5). These docking experiments are well in line
3 Evaluation of MSD
47
with crystallographical data that is confirmed by 0.89 Å RMSD (Docking solution 7 in Table
3.4), however, in addition, it can prove that E2020 will not be protonated in the binding state.
Docking
Solution
RMSD
(Å)
Eass LE Enbd vdW Eest Ecnt vdW++ Nhph Nhb
9-(p)-R 2.02 -51.2 7.4 -58.5 -8.7 -19.6 -30.3 5.3 12 1
10-(p)-S 10.48 -44.2 0.7 -44.8 -2.1 -30.1 -12.6 4.9 5 2
Table 3.5: The numerical result of MSD of the inhibitor in two individual experiments in which the ligand has
either R or S-configuration at C8 with protonated nitrogen;*: ’p’ stands for protonated
3.4.3 Effect of the environment on the conformational pose of E2020
In this step, to consider the effect of the environment of the ligand on its position within the
binding site gorge, the most important crystallographical water molecules were identified.
These waters directly interact with the ligand and have a bridge-like role to mediate the
interactions among the ligand atoms and its environment. The codes of the six most important
waters in the corresponding crystal structure are Wat1159, Wat1160, Wat1249, Wat1254,
Wat1255 and Wat1347, whose position in the binding site are shown in Figure 3.9.
Figure 3.9: The six important water molecules around E2020 in the binding site gorge, rendered in CPK
Due to important effect of waters in binding site, the binding mode of the ligand and its
structural properties were studied in the presence of the six water molecules, which interact
with the ligand in the X-ray structure. For this purpose, the optimization of the hydrogens in
water molecules was also considered, so that the water hydrogens were set to be flexible, by
which the water hydrogens could be energetically minimized during docking process. The
docking solution 11 and 12 in Table 3.6 correspond to E2020 with R-configuration in
unprotonated and protonated states, respectively. According to the results, the STotal of
unprotonated ligand, with ligand energy of 0.0 kJ/mol is higher than protonated ligand (Table
3 Evaluation of MSD
48
3.6), which is consistent with the results of docking into empty binding, saying that E2020
can not be protonated.
Docking*
solution
RMSD
(Å)
Eass LE BE Enhb vdW Eest Ecnt vdW+ Nhph Nhb
11-(up)-R 0.57 -58.7 0.0 3.1 -61.9 -15.2 -12.8 -34.0 2.9 12 3
12-(p)-R 0.56 -53.1 11.5 2.6 -67.2 -15.1 -18.2 -33.9 3.1 13 3
Table 3.6: The result of docking the ligand with R configuration in presence of the six important water
molecules with flexible hydrogens
*: ‘up’ stands for unprotonated, p for protonated.
Figure 3.10 illustrates the amino acids that interact with E2020, in presence of the water
molecules. Although the crystallographical data interestingly indicate the R-configuration at
C8 [39] for protonated ligand, using our docking tool, this experiment shows that E2020 with
R-configuration in its bound state with AChE should have R-configuration in unprotonated
state, by which it energetically forms a more stable complex with AChE.
Figure 3.10: Amino acids that interact with docking solution (gray) in presence of waters with flexible
hydrogen. X-ray structure of the
3.4.4 Unbound docking study on E2020
Independent of the available X-ray data, E2020 was built, minimized and then docked into the
empty binding site of crystal structure of TcAChE complexed with galanthamine (1DX6) [40]
and complexed with decamethonium (1ACL) [63].
3 Evaluation of MSD
49
The first step of docking into 1DX6 generated 25 answers, among which rank number 9 had
the highest STotal (Docking solution 14, Table 3.7). This was selected for the next iteration.
Performance of MSD led to find the final answer (Docking solution 15 in Table 3.7).
Comparison of the result with the available X-ray structure gives a RMSD of 0.96 Å, whereas
the first rank of the very first step has the RMSD of 2.19 Å. Furthermore the total estimated
energy of the complex is improved from-38.0 kJ/mol to-41.6 kJ/mol, respectively.
Docking
solution
Rank RMSD
(Å)
Eass LE Enhb vdW Eest Ecnt vdW+ Nhph Nhb
13 1 Step 1 2.19 -38.0 3.2 -41.2 -9.5 -4.4 -27.3 3.6 13 1
14 9 Step 1 0.97 -37.0 9.4 -46.4 -11.9 -3.7 -30.7 3.2 12 1
15 1 Step 3 0.96 -41.6 4.1 -45.7 -11.8 -3.6 -30.4 3.0 12 1
Table 3.7: Result of the unbound docking of E2020 into empty binding site of 1DX6
Docking E2020 into 1ACL also generated the correct solution with RMSD of 0.92 Å.
(Docking solution 18, Table 3.8).
Table 3.9 compares the results of unbound docking of E2020 into binding site of 1DX6 using
different methods, among which MSD predicts the best solution with regard to RMSD values.
Docking
solution
Rank RMSD
(Å)
Eass LE Enhb vdW Eest Ecnt vdW+ Nhph Nhb
16 1 Step 1 0.92 -44.7 4.4 -49.1 -10.2 -11.7 -27.2 3.9 11 0
17 2 Step 1 1.04 -43.5 4.8 -48.3 -11.2 -10.1 -27.1 3.1 11 0
18 1 Step 2 0.92 -44.8 4.4 -49.2 -10.2 -11.7 -27.3 4.0 11 0
Table 3.8: Result of unbound docking of E2020 into binding site of 1ACL
Program Search Algorithm Scoring Function RMSD (Å)
Binding Site
Flexibility
QXP+
(OSD)
Monte Carlo
One Step
Amber Force Field 2.19 No
FlexX Incremental
Construction
Empirical Score 14.54 No
G.O.L.D Genetic Empirical Score 4.41 No
QXP+
(MSD)
Multi-Step Docking Amber Force Field 0.96 No
Table 3.9: Result of unbound docking of E2020 into binding site of 1DX6 using different method
3 Evaluation of MSD
50
3.5 Study on Decamethonium (DECA)
3.5.1 Bound docking study on Decamethonium (DECA)
Decamethonium is a bisquaternary reversible inhibitor of AChE, with a simple chemical
structure of two trimethyamine, which are connected to each other via a decyl side chain [64]
(Figure 3.11).
N
+
N
+
Figure 3.11: Chemical structure of decamethonium (DECA)
Because of the high flexible structure of DECA, it is a grate challenge for any docking tool.
MSD method successfully performed the bound docking of this ligand. In this section the
detailed structural features of the complex of DECA with AChE, through the bound docking
into 1ACL, will be considered [64].
3.5.2 Docking decamethonium into the empty binding site of 1ACL
The bound docking of DECA into its complexed protein (1ACL) yielded energetic values
shown in Table 3.10.
Docking
solution
RMSD
(Å)
Eass LE Enhb vdW Eest Ecnt vdW+ Nhph Nhb
19 0.99 -40.5 0.0 -40.5 -7.0 -13.1 -20.4 2.6 7 0
Table 3.10: The numerical data of docking DECA into empty binding site of 1ACL
Figure 3.12 illustrates how close the predicted binding mode of the DECA is to the X-ray
structure of the ligand with RMSD of 0.99 Å.
3 Evaluation of MSD
51
Figure 3.12: The overlaid X-ray structure of DECA (green) on the predicted pose of the ligand by docking
(gray)
3.5.3 Docking decamethonium into the binding site of 1ACL including waters
In order to better understand the binding of DECA with AChE, we also looked for water
molecules that mediate the interaction among ligand atoms and enzyme. Using QXP+ five
important water molecules were defined within the binding site of 1ACL that are Wat612,
Wat622, Wat634, Wat640 and Wat642 (Figure 3.13).
Figure 3.13: Indicates the position of the five most important waters in the binding site of 1ACL Wat612,
Wat622, Wat634, Wat640 and Wat642 (red balls
By contribution of these important waters around DECA, other experiment was set up. To
optimize the hydrogen position of waters, their positions were energetically minimized, by
3 Evaluation of MSD
52
definition flexible hydrogens on the waters. The resulting first rank answer (Docking solution
20 in Table 3.11) has an optimal binding energy, which has been improved in all the energy
terms in comparison with docking result of empty binding site (Docking solution 19 in Table
3.10).
Docking
solution
RMSD
(Å)
Eass LE BE Enhb vdW Eest Ecnt vdW+ Nhph Nhb
20 0.98 -46.7 0.0 1.5 -48.2 -9.0 -15.3 -23.8 2.8 6 0
Table 3.11: The numerical data of docking DECA into the active site of 1ACL, in presence of the five most
important crystallographical waters with flexible hydrogen
In Table 3.12 the amino acids that directly interact with DECA are listed. Analyzing their
interatomic interactions shows that one of the three methyl groups (C16) of DECA in the
upper part of the gorge interacts with carbonyl oxygen of Asp72, the second methyl (C18)
with carbonyl of Tyr334 and the third one (C17) with hydroxyl oxygen of Tyr70. The C7 and
C9 in decyl chain have hydrophobic interactions with the aromatic ring of Phe330 the C4
interacts with carbonyl oxygen in Phe330. N1 atom of DECA in the lower part of the gorge
has a cation-
π
interaction with Trp84 and three-methyl groups of DECA have hydrophobic
interaction with carbonyl oxygen in Glu199.
Figure 3.14: Docking solution of DECA (gray) among the most important waters and interacting amino acids
Table 3.12: List of the interaction energies of amino acids around DECA
Amino acids Interaction energy
(kJ/mol)
Gly335 -4.2
Tyr70 -4.9
Asp72 -6.6
Tyr334 -7.5
Tyr121 -3.5
Phe330 -13.1
Trp84 -9.5
Glu199 -19.9
3 Evaluation of MSD
53
Oxygen of Wat642 interacts with the second and the third carbon of the decyl chain, while the
C10 interacts with oxygen in Wat634 in the lower part of the gorge. In addition from the
magnitude of their corresponding interaction energies can be seen that the most interactive
amino acids are Glu199 and Phe330 with -19.9 kJ/mol and -13.1kJ/mol energies, respectively
(Table 3.12). From these data it is observed that after including waters in docking, Glu199 has
lower interaction energy than Phe330, whereas the result of docking into empty binding site
interact better with Phe330 than with Glu199.
To summarize the results, in this study was found that on the consequence of employing MSD
for docking of DECA into the binding site of 1ACL and the most important water molecules
were recognized. In addition, the correct binding mode with 0.99 Å confirmed the accuracy of
docking procedure, using MSD. The interactive amino acids with docking solution of DECA
presented and their corresponding interaction energies calculated in different circumstances of
the gorge, in absence and in presence of waters with flexible hydrogens. Since in the
corresponding literature of 1ACL [64], the structural data of AChE-DECA complex has not
been described in detailed, therefore the obtained data from these docking studies provide
new information on interatomic interactions between DECA and its surrounding environment
in 1ACL.
The effectiveness of application of MSD method is not limited to docking of AChE’s
inhibitors. To prove this, in the next section the consensus docking method is applied in
bound and unbound docking of the inhibitors of other proteins than AChE.
3.5.4 Unbound docking study on decamethonium (DECA)
To examine capability of the MSD for correct pose prediction of ligands with high flexibility,
unbound docking of decamethonium using MSD was carried out.
This structure caused difficulties in correct pose prediction using several other docking
methods. MSD was able to produce correct answer with unbound docking DECA into 1DX6
[40]. The numerical data of the experiments are given in Table 3.13. The docking solution has
RMSD of 1.27Å (Docking solution 22, Table 3.13) that has 7.5 kJ/mol lower total estimated
energy than first hit answer in the first step docking (Docking solution 21, Table 3.13). For
this ligand it is the first time that a successful docking is reported. As the implementation of
G.O.L.D and FlexX do not produce the correct answer for this ligand (Table 3.14).
3 Evaluation of MSD
54
Docking
solution
Rank RMSD
(Å)
Eass LE Enhb vdW Eest Ecnt vdW+ Nhph Nhb
21 1 Step 1 1.27 -27.5 8.7 -36.2 -8.8 -6.4 -21.0 1.7 7 0
22 1 Step 2 1.27 -35.5 0.7 -36.2 -8.8 -6.4 -21.0 1.6 7 0
Table 3.13: Result of unbound docking of decamethonium into binding site of 1DX6
Program Search Algorithm Scoring Function RMSD (Å)
Binding Site
Flexibility
QXP+
(OSD)
Monte Carlo
One Step
Amber Force Field 1.27 No
FlexX Incremental
Construction
Empirical Score 4.52 No
G.O.L.D Genetic Empirical Score 4.83 No
QXP+*
(MSD)
Multi-Step Docking Amber Force Field 1.27 No
Table 3.14: Different programs utilized for unbound docking of DECA into 1DX6
*: As it has been shown in Table 3.13, despite of the identical RMSDs for docking solution of MSD and OSD,
MSD generates an answer with much better Eass and LE.
The effectiveness of application of MSD method is not limited to docking of AChE’s
inhibitors. To prove this, in the next section the consensus docking method is applied in
bound and unbound docking of the inhibitors of other proteins than AChE.
3.6 Application of MSD for bound docking study on several inhibitors of other
proteins than AChE
3.6.1 Bound docking of Win51711 and Win52084
Surprising binding modes have been observed for two antiviral compounds, Win51711 [65]
(Figure 3.15) and Win52084 (Figure 3.16) [66]. These two compounds possess an almost
identical structure and only differ in absence or presence of one methyl group, respectively.
However, they bind with reverse orientations. This was an encouraging reason for testing
MSD method to see if it is able to produce correct binding mode of both ligands particularly
as they have a long flexible side chain in their structure.
3 Evaluation of MSD
55
O
N
O
N
O
Figure 3.15: Chemical structure of Win51711
O
N
O
N
O
Figure 3.16: Chemical structure of Win52084
MSD result of Win51711 generates a solution with RMSD of 0.80 Å and 0.0 kJ/mol ligand
energy (Docking solution 25, Table 3.15). In addition, result of docking Win52084 is a
reasonable answer with RMSD of 1.06 Å with ligand energy of 0.0 kJ/mol (Docking solution
27, Table 3.15) versus 1.66 Å obtained from first step (Docking solution 26, Table 3.15).
Docking
solution
Rank RMSD
(Å)
Eass LE Enhb vdW Eest Ecnt vdW+ Nhph Nhb
23 1 Step 1 1.10 -44.3 2.3 -46.6 -11.9 -0.7 -34.0 4.0 13 0
24 2 Step 1 0.80 -44.1 2.9 -46.9 -13.3 -0.1 -33.6 3.3 12 0
25 1 Step 2 0.80 -46.9 0.0 -46.9 -13.3 -0.1 -33.6 3.3 12 0
*26 1 Step 1 1.66 -47.2 1.7 -48.9 -13.5 -2.1 -33.3 3.6 12 1
27 1 Step 4 1.06 -48.9 0.0 -48.9 -13.2 -2.5 -33.2 3.7 12 1
Table 3.15: Result of docking Win51711 into 1PIV (Docking solution 23-25) and Win52084 into 1RUH
(Docking solution 26-27)
*: This solution is the first hit with the highest STotal answer of the first step.
Figure 3.17 indicates the isopotential electrostatic surface of Win51711 (right in Figure 3.17)
and Win52084 (left in Figure 3.17). The extra methyl substitution on oxazoline affects on the
electrostatic charge distribution on the molecule and it could be the reason for reverse
orientation of these two ligands in the binding site of the protein.
3 Evaluation of MSD
56
O
N
O
N
O
O
N
O
N
O
Figure 3.17: Difference in isopotential electrostatic surface of Win52084 (left) and Win51711 (right), despite of
their similarity in structural skeleton. Blue/ yellow/ red indicate – n/ 0/ + n charge distribution on the ligands.
3.6.2 Unbound docking study on Deoxythymidine
Antiherpes therapies are principally targeted at viral thymidine kinases (TK) and utilize
nucleoside analogues, which are inhibitors of viral DNA polymerase. The substrate of TK is
deoxythymidine (Figure 3.18), whose X-ray structure (1KIM) has been deposited in protein
data bank [67]. In the next step of examination of MSD, the unbound docking of
deoxythymidine was performed into the uncomplexed proteins of deoxythymidine. These
proteins are KI6, KI7 and 1KI8 (Table 3.16).
PDB ID Protein Ligand
1KIM Thymidine Kinase From Herpes Simplex Virus Type I Deoxythymidine
1KI6 Thymidine Kinase From Herpes Simplex Virus Type I 5-Iodouracil Anhydrohexitol Nucleoside
1KI7 Thymidine Kinase From Herpes Simplex Virus Type I 5-Iododeoxyuridine Phosphotransferase
1KI8 Thymidine Kinase From Herpes Simplex Virus Type I 5-Bromovinyldeoxyuridine
1FVT Cyclin-Dependent Kinase 2 (Cdk2) Oxindole
Table 3.16: Indicates kinase proteins used for docking of deoxythymidine and oxindole
N
N
O
O
O
HO
HO
Figure 3.18: Chemical structure of deoxythymidine
3 Evaluation of MSD
57
In this experiment, MSD was used for unbound docking of deoxythymidine into 1KI7. This
results a solution with RMSD of 1.11 Å (Docking solution 29, Table 3.17), whereas docking
solution of the first step unbound docking has RMSD of 1.82 Å (Docking solution 28, Table
3.17). Furthermore, the total estimated energy of the final answer has -5.3 kJ/mol
improvement in comparison with the first hit in the first step (Table 3.17).
Docking
solution
Rank Protein RMSD
(Å)
Eass LE Enhb vdW Eest Ecnt vdW+ Nhph
Nhb
28 1 Step 1 1KI7 1.82 -44.8 5.1 -49.9 0.7 -29.2 -21.4 12.0 2 2
29 1 Step 4 1KI7 1.11 -50.1 11.0 -61.0 14.4 -54.5 -21.0 23.6 2 3
30 1 Step 1 1KI6 0.43 -56.2 3.9 -60.1 -3.1 -36.4 -20.6 7.8 2 3
31 1 Step 2 1KI6 0.74 -59.1 2.5 -61.6 2.5 -42.0 -22.1 13.7 1 3
32 1 Step 1 1KI8 0.59 -47.0 9.4 -56.4 -6.4 -28.4 -21.6 5.3 2 5
33 1 Step 5 1KI8 0.60 -51.3 1.7 -53.0 -5.6 -27.1 -20.3 5.2 2 4
Table 3.17: Result of unbound docking deoxythymidine
MSD result of docking into 1KI6 (Table 3.17) also generated better answer in comparison
with the first step docking, such that the former has -2.9 kJ/mol lower magnitude of total
estimated energy than the latter one (Docking solution 30 and 31, Table 3.17). In addition, the
ligand energy of MSD answer is 1.4 kJ/mol lower than of the solution in the first step
docking, while the corresponding RMSD is 0.29 Å higher than the docking solution of the
first step docking run (Table 3.17). The worth of using MSD in unbound docking of
deoxythymidine into 1KI8 is also observable, as the result gives -4.3 kJ/mol lower total
estimated energy (Docking solution 32, Table 3.17) than the answer of the first step docking
(Docking solution 33, Table 3.17). It is noteworthy that ligand energy of the former is better
with 7.7 kJ/mol difference, however, the RMSD values of them are almost identical (Table
3.17).
3.6.3 Bound docking study of oxindole
Oxindole (Figure 3.19) is one of the inhibitors of cyclin-dependent kinase 2 that was taken as
a ligand for bound docking into its complexed protein 1FVT [68] (Table 3.16).
3 Evaluation of MSD
58
H
N
O
N
HN
S
H
2
N
O
B
r
O
Figure 3.19: Chemical structure of oxindole
MSD results an answer with better STotal (Docking solution 35, Table 3.18) in comparison
with the first step docking, however the RMSDs are almost identical (Docking solution 34,
Table 3.18).
Docking
solution
Protein RMSD
(Å)
Eass LE Enhb vdW Eest Ecnt vdW+ Nhph
Nhb
34 1FVT 0.46 -39.3 6.0 -45.3 2.8 -26.7 -21.4 13.4 6 5
35 1FVT 0.45 -42.1 4.2 -46.3 3.1 -28.1 -21.4 13.6 6 6
Table 3.18 Numerical data of oxindole bound docking into 1FVT
4 Elucidation of Structural properties of PPG
59
4 Elucidation of structural properties of (PPG)
The aim in this part of the work is application of MSD for docking study on a new derivative
of galanthamine ‘piperidinopropylgalanthamine’ (PPG) [38] (Figure 4.1), which was firstly,
correct placement of the PPG in its complex state with AChE via unbound and bound
docking, secondly estimation of its inhibitory constant based on the calculated total estimated
binding energy of the complex.
O
O
O
H
H
N
10
N
2
2
Figure 4.1: Chemical structure of piperidinopropylgalanthamine (PPG) with two possible inside inverted
piperidine ring
Generally, information of the 3D structure of a ligand-protein complex provides the data on
the arrangement of the atoms in space. Despite of that sometimes it is hard to distinguish
different possible orientations in a substructure of a molecule. For instance, in the case of
PPG, there are two possible ring inversions for piperidine ring in X-ray structure, which are
inside and outside inverted (Figure 4.1). Investigation of this issue has been included as a part
of our molecular docking study in this chapter. Protonation state of nitrogen atoms at
piperidine (N22) as well as at galanthamine nitrogen (N10) has been also of particular
importance in this work, because the corresponding data can lead us to find out whether a
ligand in AChE binding site, at all can have the protonated state and if so, why and in which
particular circumstances.
Since in a biological system of the protein-ligand complex, interactions take place in an
aqueous environment and effect of the waters are reflected in enthalpic and entropic
contributions complex energy of the system [69] [70], therefore, here, the presence of the
important crystallographic water molecules have been also taken into account. The
importance of the waters were recognized by QXP+, according to their ability to either contact
to PPG atoms or mediate interactions among ligand atoms and surrounding amino acids.
4 Elucidation of Structural properties of PPG
60
4.1 Searching for the protonation state of PPG using bound docking
In this section five crystallographic water molecules have been included into the binding site.
These waters were identified by their particular role for mediating the interactions among
ligand atoms and surrounding residues. The codes of these waters are Wat5, Wat9, Wat84,
Wat94, and Wat275. The data given in Table 4.1 should be discussed in two groups. The first
group consists of docking solutions 10-12, in which no water molecules of the binding site
were taken into account. Among these three docking solutions, the protonated ligands show
more favourable total estimated binding energies than the neutral molecule. Of the two
possible protonation sites N10 (Docking solution 11) and N22 (Docking solution 12),
protonation at N10 with a bent shape of the piperidine-propyl side chain is preferable, as it has
more similarity to the known crystal structure (RMSD = 0.46 Å). A protonation at N22 can be
ruled out because of the straightened N-alkyl side chain (RMSD = 1.83 Å). In addition, even
though protonation at N22 yields the lower Eass, the resulting structures with protonation at
N10 gives the higher STotal (red numbers in Table 4.1), including ligand energy of zero,
suggesting a global minimum structure, which usually is assumed to be the native like
structure [34]. The lower total binding energy of docking solution 12 versus solution 11
(Table 4.1) is dominated by an electrostatic interaction between the positively charged
piperidine-nitrogen and (OAc-) of aspartic acid (Asp72) mid way in the gorge. This
electrostatic interaction decreases electrostatic energy (Eest) to -21.2 kJ/mol consequently
affects total binding energy. Figure 4.2 shows the overlapping electrostatic surfaces at the
protonated piperidine and the (OAc-) residue of Asp72.
Figure 4.2: Overlapping electrostatic surfaces around protonated piperidine with positive charge (blue area) and
negatively charged OAc- of Asp72 (red area).
4 Elucidation of Structural properties of PPG
61
Docking
experiment
Ligand* RMSD
(Å)
Eass LE Ebind Enhb vdW Eest Ecnt vdW+ Nhph Nhb
10 Bent-In
(up) a
0.41 -51.4 7.6 0.0 -59.0 -8.5 -18.5 -32.1 6.9 11 2
11 Bent-In
N10 (p) a
0.46 -55.5 0.0 0.0 -55.5 -10.4 -15.5 -29.6 4.2 12 3
12 Staight
N22 (p) a
1.83 -56.6 0.1 0.0 -56.6 -6.5 -21.2 -28.9 6.5 10 2
13 Bent-In
(up) b
0.40 -104.0 7.0 -46.7 -64.4 -10.1 -21.9 -32.3 5.6 12 2
14 Bent-In
N10 (p) b
0.35 -109.0 6.4 -45.7 -69.3 -7.9 -28.9 -32.6 8.1 12 2
15 Straight
N22 (p) b
1.85 -109.0 2.0 -44.5 -66.8 -6.1 -29.1 -31.5 9.0 10 2
Table 4.1: The result of docking into the binding site of AChE-PPG, red numbers indicate the best value in
different terms of docking into empty binding site, Blue numbers indicate the best value in different terms for
docking into binding site with flexible water hydrogen
*: a; Docking was performed into the binding site without water, b; Docking in the presence of the five most
important water molecules with flexible hydrogen atoms
The interaction of the optimal solution of protonated PPG at N10 (Docking solution 11 in
Table 4.1) with its surrounding amino acids is described in detail in Table 4.2. The relevant
amino acids, responsible for non-bonding interaction energy of the protonated PPG [38]
consist of C-H…O, as well as C-H…N, C-H…
π
systems that are known as weakly hydrogen
bonds [71][72], in the more hydrophobic regions of proteins [73]. Furthermore, these
interactions are graphically shown in Figure 4.3. The corresponding structure shows an
excellent RMSD of 0.46 Å and has the highest STotal.
As it can be seen in Figure 4.3, two carbons (C25 and C26) in piperidine ring interact with
carbonyl oxygen in Phe330 (C-H…O) that assists the piperidine ring flip upward to have out
side inversion. The hydroxyl oxygen of Tyr334 also has a (C-H…O) interaction with C27 in
piperidine ring, the same type of interaction take place among alkyl side chain of the ligand
(C20 and C21) with Asp72 oxygen. A (C-H…N) plus a (C-H…
π
) interaction between
Gly118 and cyclohexenyl ring in galanthamine moiety is also observable. A (C-H…O) type
interaction can be seen between Ser122 and a carbon in cycloheptyl ring (C11) of
4 Elucidation of Structural properties of PPG
62
galanthamine (left in Figure 4.3). The nitrogen in Gly118 shows a ‘lone pair-
π
’ interaction
with aromatic ring in galanthamine substructure of PPG. Two amino acids dominate
interaction energies, Glu199 and Ser200, whose involved atoms with PPG have been
indicated in detailed in Figure 4.3 (right). Table 4.2 lists the relevant amino acids that are
responsible for binding of the N10-ptotonated PPG [38].
Amino
Acids
Asp72 Trp84 Tyr121 Ser122 Gly118 Gly119 Glu199 Phe330 Phe331 Ser200 His440 Tyr334
Interaction
energy (kJ/mol)
-8.0 -6.9 -6.0 -2.8 -8.4 -16.8 -26.6 -12.4 -12.6 -27.8 -2.5 -4.5
Table 4.2: Displays amino acids interacting with the neutral PPG and the corresponding interaction energies
Figure 4.3: The amino acids and their specific atoms that interact with protonated PPG at N10 (left) Glu199 and Ser200 are
two amino acids with the lowest interaction energies (right)
The second group of data in Table 4.1 including docking solutions 13-15, show results from
docking studies, in which the five most important waters have been included as part of the
binding site. Due to having the highest STotal, the protonated PPG at N10 (Docking solution 14
in Table 4.1), in this group, turns out to be the most favorable docking result. As it is bound
docking and then comparison of docking results with X-ray structure shows that the
protonated ligand with inside inverted with the lowest RMSD and the lowest total estimated
binding energy is the most reliable answer, however according to its STotal value it has one
score less than unprotonated ligand with inside inverted pose.
Overall, the data from Table 4.2 suggest that PPG binds to the AChE active site in a
protonated state, with protonation at N10 rather than at N22 (Docking solution 13 versus 14)
4 Elucidation of Structural properties of PPG
63
with the highest STotal (blue numbers in Table 4.1). Therefore PPG is one of the rare cases
where the binding of a protonated ligand is made very likely. Crystallographic studies do not
show hydrogens and therefore cannot differentiate between a protonated and a neutral ligand
in the bound state. Modeling studies, however, on the basis of a given crystal structure can
unravel such questions, when important water molecules are included in the docking
experiments and when on the basis of bindings energies and of RMSD-values protonated and
neutral structures can be compared. The finally proposed structure for the protonated PPG has
been shown in Figure 4.4 overlaid on the crystal structure.
Figure 4.4: The final bound docking solution for protonated PPG (gray) overlaid on the X-ray structure (green)
The bent pose of the ligand, protonated at N10, is not only the most populated conformation
as the first hit of docking, but also has the most abundant conformational pose, so that 16
solutions out of 25 resulting structures (64%) are similar to the first hit structure (Figure 4.5).
Figure 4.5: Superimposition of 25 docking solutions of protonated PPG
In view of the pKa-values of the nitrogen-atoms in PPG, one should expect protonation at N22
rather than at N10. While galanthamine is known to have a pKa-value of 8.2 [74], the value
for piperidine is 11.05 [75]. This magnitude is decreased by changing from the secondary
4 Elucidation of Structural properties of PPG
64
amine to a tertiary as in PPG. On the one hand the basicity should increase because of the
inductive effect of alkyl groups [76]. On the other hand the steric hindrance predominates as
solvation of the sterically hindered ammonium-cation is becoming more difficult. As a result
the pKa-value for n-methypiperidine drops to 10.08 [75].
Therefore, in an aqueous solution protonation is more probable at the piperidine-nitrogen than
at the galanthamine-nitrogen by almost a factor of 100. However, a ligand in a binding site is
no longer in an aqueous environment but rather embedded in at least a partially hydrophobic
surrounding. This makes a prediction of the protonation state of a ligand on the basis of pKa-
values impossible. In contrast, docking experiments take into account the entire binding site
area and therefore do reflect much better the actual status of the ligand. The data from Table
4.1, particularly the result from docking solution 14 and from docking solution 11 clearly
suggest that PPG is still protonated in the bound state and N10 is the most probable
protonation site. It should be mentioned that the resolution of the crystal structure of PPG
does not allow to unequivocally defining the orientation of the piperidine-ring with respect to
galanthamine moiety. In other words, from the crystal structure data one cannot decide, if the
piperidine-nitrogen is oriented towards the galanthamine moiety (inside inverted piperidine)
or towards the Tyr334 residue, which is the closest amino acid to the piperidine-ring (outside
inverted piperidine). If we accept that PPG is protonated in the bound state, the structure with
the inside inverted piperidine creates a cage, in which the proton is trapped (right in Figure
4.6) and consequently could be even shared between the two nitrogen-atoms of the ligand (left
in Figure 4.6).
Figure 4.6: Illustrates the hydrogen bond between two nitrogens (left), H+ is trapped into the cage-like shape of
the ligand (right)
4 Elucidation of Structural properties of PPG
65
4.2 Conformation at galanthamine and piperidine substructures by unbound docking
study
In this chapter MSD is used for unbound docking of PPG. It was applied to find the favorable
conformation (equatorial or axial) in two nitrogens of the molecule, at piperidine and
galanthamine substructures with unbound docking of the neutral ligand (deprotonated) into
1DX6 [40]. The docking results indicate the conformational conversion from ‘a-e’ to ‘e-e’
and ‘a-a’ to ‘e-a’, where ‘a’ stands for the axial and ‘e’ for the equatorial. The first letter
shows the conformation in piperidine and the second letter corresponds to the galanthamine.
The conformation of the flexible piperidine is converted to equatorial, by which a six-member
ring is stabilized. In contrast, the conformation at galanthamine-nitrogen, maintains axial
without any conversion (Table 4.3).
As it is shown, in Table 4.3, the final structures with ‘e-e’ conformation are obtained in
entries pose 1 and 4. Particularly, entry Table 4.3 has the highest STotal. Due to this advantage
of the docking result with ‘e-e’ conformation, for further steps of docking PPG, the ‘e-e’
conformation was considered as the starting structure of the ligand.
entry Starting
Conf*
Ending
Conf
RMSD
(Å)
Eass
LE
Enhb vdW Eest Ecnt vdW+ Nhph Nhb
1 a-e e-e 0.49 -52.1 6.3 -58.4 -8.3 -18.5 -31.6 7.1 11 2
2 a-a e-a 1.07 -48.3 4.8 -53.1 -5.4 -17.9 -29.8 8.8 11 2
3 e-a e-a 1.07 -48.7 4.4 -53.1 -5.1 -18.1 -30.0 9.2 11 2
4 e-e e-e 0.57 -51.4 7.6 -59.0 -8.5 -18.4 -32.1 6.9 11 2
Table 4.3: Numerical data obtained from docking the ligand with different conformation at N22 and the N10 in
the starting structure of unbound docking into 1DX6
*: ‘Conf’ stands for conformation, ‘a’ for axial and ‘e’ for equatorial
The inconvertibility of axial conformation to equatorial in galanthamine moiety raises the
question, whether it might have any conformational change in nitrogen moiety of
galanthamine or not. For this purpose besides galanthamine, two derivatives of it (ethyl
galanthamine and demethylated galanthamine) were considered, each of which was studied
with axial and equatorial conformations at galanthamine-nitrogen. The nitrogen displays the
conformational conversion at galanthamine-nitrogen in the case of galanthamine and other
derivatives with small substitution. The reason is that PPG has a longer oligomethylen side
chain, which increases the required energy for conformational conversion.
4 Elucidation of Structural properties of PPG
66
1st rank (Docking solution1) 4th rank (Docking solution2) 8th rank (Docking solution3)
Figure 4.7: Three possible poses of the ligand, among the 25 docking solutions of the neutral ligand extracted
from Table 4.2. The bent pose of the PPG with inside inverted piperidine ring is ranked at the first hit
With respect to the ‘e-e’ conformation of the molecule, three possible shapes can be found,
which are straight, bent with inside and outside inverted piperidine ring. The energetically
best solution of these three conformations has been shown in Figure 4.7, while Table 4.4 lists
all the corresponding docking data.
Docking
solution
Neutral*
Ligand
Rank RMSD
(Å)
Eass LE Enhb
vdW Eest Ecnt vdW+ Nhph Nhb
1 Bent-In
(up)
1 0.57 -51.4 7.6 -59.0 -8.5 -18.4 -32.1 6.9 11 2
2 Straight
(up)
4 1.95 -49.1 2.8 -51.9 -8.4 -15.2 -28.3 5.2 10 2
3 Bent-Out
(up)
8 1.00 -47.9 4.1 -52.0 -5.7 -17.2 -29.1 7.8 12 3
Table 4.4: Numerical data correspond to docking solutions of the neutral ligand.
*: ‘up’ stands for unprotonated (neutral), ‘Bent-In’ for neutral with inside inverted ring, ‘Bent-Out’ neutral with
outside inverted ring.
The unbound docking of the PPG, according to the consensus scoring of MSD method,
suggests that inside inverted bent pose of the PPG (Docking solution 1 in Table 4.4) is the
best pose, as it has the highest STotal among three other docking solutions (red numbers in
Table 4.4).
4.3 Unbound docking of PPG protonated at the galanthamine-nitrogen (N10)
Next, another unbound docking experiment into 1DX6 was set up to find the energetically
favored conformation for a PPG protonated at the galanthamine nitrogen (N10). Again three
different structural poses of the ligand were observed among the first 25 resulting structures,
i.e. a straight and two bent forms, with inside and outside inverted piperidine ring,
respectively.
4 Elucidation of Structural properties of PPG
67
Docking*
solution
Protonated
Ligand*
Rank RMSD
(Å)
Eass LE Enhb vdW Eest Ecnt vdW+ Nhph Nhb
4 Bent-In (p) 1 0.46 -55.6 1.2 -56.8 -8.7 -16.9 -31.2 6.5 12 4
5 Straight (p) 4 1.93 -48.4 1.4 -49.8 -6.9 -14.2 -28.7 6.6 10 3
6 Bent-Out (p) 22 1.16 -42.8 7.6 -50.3 -6.7 -14.4 -29.3 7.1 8 2
Table 4.5: Numerical data correspond to docking solutions of the protonated PPG at galanthamine-nitrogen
*: ‘p’ stands for protonated
1st rank (Docking solution 4) 4th rank (Docking solution 5) 22nd rank (Docking solution 6)
Figure 4.8: The best solution of the different structural poses of PPG with protonation at the galanthamine-
nitrogen, extracted from Table 4.5
The docking experiment yields a bent structure with an inside inverted piperidine ring as the
first hit, which not only is the most populated conformation but shows the lowest energy in all
respects as well as the highest number of hydrophobic interactions and hydrogen bonds, it
thus has the highest STotal among the all three structural poses of the PPG. This structure,
therefore, is suggested to be the most possible conformation of PPG, protonated at nitrogen
N10 (Table 4.5, Figure 4.8).
4.4 Unbound docking of PPG protonated at the piperidine-nitrogen (N22)
Unbound docking of protonated PPG at the piperidine-nitrogen (N22) into 1DX6 also yields
three possible conformational poses among the first 25 hits. The energetically best of each are
shown in Figure 4.9 and Table 4.6.
Docking
solution
Protonated
Rank RMSD
(Å)
Eass LE Enhb vdW Eest Ecnt vdW+ Nhph Nhb
7 Straight (p) 1 1.87 -57.1 0.3 -57.4 -5.6 -22.2 -29.6 8.4 10 2
8 Bent-In (p) 3 0.52 -54.8 0.0 -54.8 -7.5 -15.2 -32.1 7.8 11 3
9 Bent-Out (p) 12 0.97 -46.7 3.0 -49.7 -4.5 -15.2 -30.0 8.7 12 2
Table 4.6: Numerical data correspond to docking solutions of PPG protonated at the piperidine-nitrogen (N22)
4 Elucidation of Structural properties of PPG
68
1st rank (Docking solution 7) 3rd rank (Docking solution 8) 12th rank (Docking solution 9)
Figure 4.9: The best solution of the different structural poses of PPG protonated at the piperidine-nitrogen,
extracted from Table 4.6
In contrast to the related docking run with protonation at the galanthamine-nitrogen (in
previous section), the energetically favored structure with protonation at the piperidine-
nitrogen shows a straight side chain as the energetically structure favored and therefore
deviates massively from the crystal structure. To finally select the ultimate conformation of
PPG, the three first rank answers, i.e. the one from the docking run of unprotonated PPG and
from docking the protonated ligand with protonation at either nitrogen, are given in Table 4.7,
and compared to the crystal structure in Figure 4.10.
Docking
solution
Ligand* RMSD
(Å)
Eass LE Enhb vdW Eest Ecnt vdW++ Nhph Nhb
1 Bent-In
(up)
0.57 -51.4 7.6 -59.0 -8.5 -18.4 -32.1 6.9 11 2
4 Bent-In
N10 (p)
0.46 -55.6 1.2 -56.8 -8.7 -16.9 -31.2 6.5 12 4
7 Straight
N22 (p)
1.87 -57.1 0.3 -57.4 -5.6 -22.2 -29.6 8.4 10 2
Table 4.7: The best answer of all three previous unbound docking unprotonated and protonated at (N10, N22)
*: ‘p’ stands for protonated and ‘up’ for unprotonated
(Docking solution 1, Table 4.2) (Docking solution 4, Table 4.3) (Docking solution 7, Table 4.4)
Figure 4.10: X-ray structure of the neutral ligand (green) overlaid on the first hit answers of unbound docking of
unprotonated and protonated PPG
4 Elucidation of Structural properties of PPG
69
As can be seen form Table 4.7, the protonated structures are energetically preferred,
compared to the neutral PPG. However, the best total estimated binding energy is found for
the straight structure (Docking solution 7 in Table 4.7), which does not agree with the crystals
structure. Here, once more the PPG protonated at galanthamine-nitrogen has the highest STotal
among the solutions in Table 4.7 (Docking solution 4 in Table 4.7), which provides another
reason for the highest possibility of the bent PPG, protonated at galanthamine-nitrogen.
Up to this step, all docking runs were performed into the binding site of the ligand-unbounded
protein (1DX6), reflecting the case that a crystal structure of PPG would not be known.
However as the coordinates of the AChE-PPG complex were available to us, before
deposition in the PDB-database [38], the experiment was additionally run using the
complexed protein (PPG protein).
Despite our finding that the PPG structure with an inside inverted piperidine-ring is the
energetically favored result with the most populated conformation for the protonated and even
for the unprotonated ligand, that structure is further assisted by a detailed analysis of the
interaction energy of the piperidine-moiety with surrounding amino acids. Table 4.8 displays
the particular interaction energies for the two piperidine conformations, and although the
overall number of interacting amino acids is higher in case of the outside inverted ring, the
inside inverted one shows the stronger interaction energies, which sum up in total to better
interaction energy.
Amino Acids Interaction energy (kJ/mol)
for PPGP- N10 Inside inverted piperidine
Interaction energy (kJ/mol)
for PPGP- N10- Out side inverted piperidine
Tyr334 -6.2 -8.6
Phe331 -16.2 -9.7
Asp72 -11.3 -5.2
Phe290 - -3.0
Tyr121 - -4.0
Sum -33.5 -30.0
Table 4.8: The amino acids that interact with piperidine ring of PPG causing piperidine ring inversion.
Summarizing our results for PPG, we therefore would propose a PPG-structure with an inside
inverted ring as shown in Figure 4.6. In terms of our overall aim, to find a galanthamine
derivative that simultaneously can block the esterase- as well as the PAS-site, PPG would not
be a suitable candidate. However the docking experiments with PPG gives valuable
4 Elucidation of Structural properties of PPG
70
information as to the need of a longer N-alkyl side chain in order to block the entire active site
gorge. Such a structure would be the already described BHG.
5 Molecular docking studies on the “Back Door” hypothesis
71
5 Molecular docking studies on the “Back Door” hypothesis
Acetylcholine is the product of the hydrolization process, which is believed to leave the active
site-gorge of AChE through the normal pathway passing the rim of the gorge. From studying
on X-ray structure of AChE arise many questions about substrate solution and product
release. The structure of AChE is characterized by a high negative charge (-14e for TcAChE)
[77], which is distributed asymmetrically over the protein. This negative electrostatic potential
grows along the active-site gorge toward its base and has the largest negative value at the
bottom of the gorge, so that a strong electrostatic dipole momentum is directed toward the
lower part of the binding site [78]. It means this factor steers positively charged ligand down
the gorge [79] that electrostaticly establishes an appropriate location for accommodating the
positively charged portion of the ligand, inside the deep and narrow gorge. This could prevent
the positively charged ACh acylation-product leaves the gorge easily [80]. That is, the
presence of a back door is assumed for the easier clearance of the hydrolization product
through the closer exit doors in the bottom of the gorge than the mouth of that. A thin wall
near the base of the active site, at proximity of residues Met83 and Trp84 possibly offers an
alternative pathway for clearance of the product [78].
Dimethylmorpholine (DMP)
Octyl chain
Eseroline
DMPO
N
N
H
O
O
N
H
(CH
2
)
8
N
O
Figure 5.1: Chemical structure of MF268
MF268 (Figure 5.1) is an analogue of physostigmine with pseudo irreversible inhibitory effect
on AChE, which belongs to the class of carbamate acetylcholinesterase inhibitors. It
carbamoylates the catalytic serine, which afterwards is regenerated by hydrolysis. The amino
acids of catalytic triad of the active site gorge (Glu327, His440, and Ser200) are involved in
carbamoylation process and the ligand is cleaved in two parts during the performance of the
mechanism. One part is dimethylmorpholinoocthyl (DMPO), that covalently binds to Ser200
and the other part is the leaving group eseroline (Figure 5.2).
5 Molecular docking studies on the “Back Door” hypothesis
72
AChE
His
440
-Im
Ser
200
O
H+
carbamoylation
N
N
H+AChE
His
440
-Im
Ser
200
O NH
O
N
O
(CH
2
)
8
N
N
H
O
O
N
H
(CH
2
)
8
N
O
HO
Figure 5.2: Mechanism of carbamoylation
Since the bulky eseroline is not found in the X-ray structure of the ligand-AChE complex
(1OCE) [81], it thus implies the presence of a back door in the bottom of the binding site. In
this approach, for clearance of the eseroline, Bartolucci et al. proposed two possible exit
doors. One is a shutter-like in plane motion of Trp84, Val129 and Gly441 (red colored
channel in Figure 5.3) and a flap-like conformational transition of the Ω-loop stretching from
Cys67 to Cys94 (orange colored loop in Figure 5.3) [81].
Recently, Greenbalatt et al. proposed a “side door” for the clearance of the ACh acylation-
product that might provide the third alternative exit for the acetyl group, rather than for
choline. They suggest a facial rearrangement of the Trp279-Ser291 loop (blue colored region
in Figure 5.3), which may produce a significant increase in the diameter of the gorge,
facilitating a passage for the bulky, rigid inhibitors [82]. Using our MSD method, we have
tried to give further answers to the on going discussions on the presence of a back or side door
for product clearance from binding site of AChE.
The three-dimensional structure of Torpedo Californica acetylcholine esterase (TcAChE)
complexed with DMPO was retrieved from Protein Data Bank with accession code 1OCE
[81]. MSD consensus method of QXP+ besides G.O.L.D with Chem. and Gold Scores were
employed as docking programs.
5 Molecular docking studies on the “Back Door” hypothesis
73
Figure 5.3 : Three alternative exit doors around binding site gorge (upper); the side door (blue, left) alternative channel
(red, right); Ω-loop (orange, middle) and the green arrow shows the direction of exit through mouth of the gorge (green),
white arrows displays the exit direction in each of the pathways.
5.1 Docking study on protonation of DMPO
5.1.1 MSD method
The docking experiments of DMPO in two protonated and unprotonated states were
individually performed in the binding site of 1OCE, using MSD consensuses method. The
obtained total estimated binding energy of the complex for protonated DMPO has -1.7 kJ/mol
lower energy than of that in unprotonated DMPO, as well as a higher STotal (Docking solution
2 in Table 5.1), which is not able to reproduce the ligand structure correctly, because with
respect to the position of the DMPO in X-ray structure, the RMSD value of the unprotonated
5 Molecular docking studies on the “Back Door” hypothesis
74
ligand (1.07 Å) is lower than protonated one (1.63 Å). Figure 5.4 shows the conformational
pose of DMP in both protonated and unprotonated DMP substructure of the ligand. The
unprotonated one has a ring inversion of DMP, as compared to the ring conformation in X-ray
structure (Figure 5.4, the upper). This ring inversion of the docking solution is caused by a
hydrogen bond between DMP oxygen and Tyr70 hydroxyl group (Figure 5.4, the upper). In
contrast, the docking solution of protonated DMP shows a perpendicular position of the
morpholino ring with X-ray structure of the ligand (Figure 5.4, the lower). According to the
resulting RMSDs (Table 5.1), DMPO is not protonated.
Docking*
Solution
RMSD
(Å)
Eass LE Eslb** Enbd vdW Eest Ecnt vdW++ Nhph Nhb
1 (up) 1.07 -28.0 1.9 5.7 -35.6 -2.3 -15.4 -17.8 5.8 7 1
2 (p) 1.63 -29.7 0.0 6.7 -36.4 -3.0 -18.8 -14.5 3.7 5 1
Table 5.1: Results of docking protonated and unprotonated DMPO into the binding site of 1OCE.
*:‘up’ stands for unprotonated and ‘p’ for protonated DMPO.
**: Eslb stands for the covalent bond energy of the ligand
5 Molecular docking studies on the “Back Door” hypothesis
75
Figure 5.4: Illustrates the position of the solutions of docking DMPO (gray) with unprotonated DMP having
inverted morpholino ring (the upper) and with protonated DMP having perpendicular pose of morpholino ring
(the lower). X-ray structure has been colored in green.
To have a more reliable resolution for the possibility of protonation at DMP, we took
advantage of another docking tool. For this purpose G.O.L.D with two scoring functions
(Chem. Score and Gold Score) were utilized.
5.1.2 G.O.L.D
To examine the accuracy of MSD result, in the next step, G.O.L.D with Chem. Score was
employed. Similar to the previous step, RMSD of unprotonated ligand has 1.01 Å value
(Docking solution 3, Table 5.2), which is lower than of that in protonated DMPO with 2.50 Å
RMSD (Docking solution 4, Table 5.2). Due to this fact that G.O.L.D scoring functions have
been trained to count hydrogen-bond energy as an important component of the fitness
function and each possible hydrogen-bonding pair contributes with high weight to the overall
energy of binding [83] [84], therefore the fitness score for protonated ligand is 2.40 more than
unprotonated ligand, because the hydrogen-bonding score of the former is higher.
5 Molecular docking studies on the “Back Door” hypothesis
76
The Ligplot view of the docking solutions of Chem. Score, are shown in Figure 5.5. The
DMPO forms a stable adduct with AChE, so that in vitro enhances an irreversible-like
character.
According to the docking results, in addition to the covalent bond between DMPO and
Ser200, two hydrogen bonds are established between carbamet oxygen and NH of Gly118 and
Gly119 at oxyanion holes. These bonds besides other hydrophobic interactions keep the
ligand for a longer time in the gorge and stabilize the binding of the ligand with active site
(Figure 5.5). In contrast the protonated DMPO makes a hydrogen bond at N+ with OH group
of Tyr121 and establishes three extra hydrogen bonds with catalytic triad at bottom of the
gorge. This hydrogen bond is a driving force to bend the long hexyl chain downward to the
base of the gorge and make a distance from rim of the gorge (Figure 5.5). It means, if the
DMP were protonated, there would be more space close to rim of the gorge, for the leaving
group (eseroline) to leave the binding pocket through the mouth of the gorge, whereas the
unprotonated ligand better blocks the mouth of the gorge (the lower in Figure 5.5). Besides
these two possibilities the backbone of DMPO blocks the space of the gorge and seems to be
the main obstacle for leaving eseroline through the mouth. These factors direct the leaving
group toward other possible pathways (Figure 5.3).
Docking solution Ligand* Total fitness score RMSD (Å) Hydrogen bond
score
3 (up) 27.09 1.01 0.85
4 (p) 29.49 2.50 0.98
Table 5.2: Score and RMSD of docking solutions of protonated and unprotonated DMPO into the binding site of
1OCE using Chem. Score
*:‘up’ stands for unprotonated and ‘p’ for protonated DMPO.
5 Molecular docking studies on the “Back Door” hypothesis
77
Figure 5.5: Solution of docking DMPO, unprotonated (up-left) and protonated (up-right). Overlaid docking
solutions of protonated DMPO (orange) and unprotonated DMPO (gray) indicates that unprotonated DMPO
better blocks the mouth of the gorge (green) than the bent pose of the protonated DMPO.
Furthermore, result of docking using Gold Score also confirms that ligand is not protonated at
DMP moiety and the docking solution of the unprotonated ligand better resembles the X-ray
structure, so that the corresponding solution has RMSD of 1.13 Å (Docking solution 5 in
Table 5.3) versus 2.47 Å (Docking solution 6 in Table 5.3) for protonated ligand. Again, the
much larger score of hydrogen bond for protonated DMPO (11.51, Table 5.3) increases the
fitness score in favor of protonation, which is 43.54 versus 37.43 for the unprotonated one
(Table 5.3). This is due to the extraordinary weight of hydrogen bond in G.O.L.D scoring
functions.
5 Molecular docking studies on the “Back Door” hypothesis
78
Docking solution Ligand* Total fitness score RMSD
(Å)
Hydrogen bond
score
5 (up) 37.43 1.13 0.75
6 (p) 43.54 2.47 11.51
Table 5.3: Score and RMSDs of docking solutions for protonated and unprotonated DMPO in the binding site of
1OCE, using Gold Score.
*: ‘p’ stands for protonated and ‘up’ for unprotonated
Up to this step, in all three experiments, the docking results of only unprotonated DMPO is
close to the X-ray structure. Furthermore it is necessary to find the appropriate position of the
leaving group after cleavage and find out if there is any alternative exit door for its clearance.
5.2 Study on the location of leaving group in the absence of DMPO
Docking experiments on eseroline were carried out, assuming three protonation states for
eseroline with the neutral, protonated at N5 and protonated at N7, respectively (Figure 5.6).
N
7
N
5
H
HO
Figure 5.6: Chemical structure of eseroline
In the absence of DMPO, the eseroline in three different states was individually docked into
the binding site of AChE. According to the docking solutions, probability of protonation at
N5 is rejected, because docking result suggests a location for eseroline, which in the DMPO-
AChE complex (1OCE) is occupied by DMPO (left in Figure 5.7 & Docking solution 9 in
Table 5.4). The location of docking solutions of protonated ligand at N7 and neutral ligand are
very close to each other, as it is demonstrated in Figure 5.7 (right), protonated eseroline
(magenta) is perfectly superimposed on the neutral eseroline, gray (right in Figure 5.7). In
contrast to N5-protonated eseroline, their nitrogens do not overlap with DMPO. The very
close locations of the both states of eseroline (protonated at N7 and unprotonated) suggest that
eseroline takes up this space after split off from MF268 by Ser200.
5 Molecular docking studies on the “Back Door” hypothesis
79
Docking
solution
Ligand* Eass LE Enhb vdW Eest Ecnt vdW+ Nhph Nhb
7 (up) -32.6 0.6 -33.2 -2.0 -15.0 -16.2 5.9 6 2
8 N7 (p) -32.5 1.4 -34.0 -3.4 -14.4 -16.1 4.4 5 2
9 N5 (p) -34.6 0.5 -35.1 -4.1 -15.6 -15.4 3.5 5 2
Table 5.4: Results of docking Eseroline as regard to three different possibilities of protonation states
*: ‘p’ stands for protonated and ‘up’ for unprotonated
Table 5.4 indicates the numerical data obtained from docking eseroline in three different
protonation states. The neutral eseroline has a higher STotal (Docking solution 7 in Table 5.4)
than protonated ligand at N7 (red numbers in Table 5.4). Unprotonated eseroline seems to be
energetically slightly more favored than protonated one, although the locations of both are
approximately identical.
Figure 5.7: Docking solution of N5-protonated eseroline (red) bumps to covalently bonded DMPO in 1OCE
(green) (left), Docking solutions of N7-protonated eseroline (magenta) is superimposed on the docking solution
of the neutral eseroline colored in gray (right), both of which do not interfere with DMPO (green). The docking
solution of DMPO, obtained from QXP+, is in gray color.
5.3 Study on the location of leaving group in presence of DMPO
5.3.1 MSD method
The result of previous step demonstrates that there are only two possibilities for the leaving
group, which could be either neutral or becomes protonated at N7 position, therefore in this
experiment, DMPO, covalently bonded to Ser200, was included in the binding site and again
5 Molecular docking studies on the “Back Door” hypothesis
80
two independent MSD runs of unprotonated and protonated (N7) were performed into the
binding site of 1OCE.
Docking
solution
Ligand* Eass LE Enbd vdW Eest Ecnt vdW+ Nhph Nhb
10 (up) -29.0 0.2 -29.2 -0.5 -6.3 -22.4 10.0 6 1
11 N7 (p) -26.7 1.9 -28.6 1.7 -8.4 -22.0 11.5 7 2
Table 5.5: Results of docking eseroline with neutral and protonated N7
*: ‘p’ stands for protonated and ‘up’ for unprotonated
Once more, the positions of the eseroline in docking results of both neutral and protonated
ligand (N7) are identical. The docking solution of the neutral eseroline has larger STotal than
protonated eseroline (red numbers, Table 5.5). In accord with the priority of the neutral ligand
in different energy terms, at the result of this experiment, the neutral state is suggested as the
answer of docking with MSD.
Figure 5.8: Overlaid docking solutions of protonated (magenta) and neutral (grey) using QXP+
The location of the eseroline was found in the proximity of the channel that confirm
alternative channel at the bottom of the gorge close to Trp84 and His440 (colored in red in
Figure 5.9, left). The leaving group, eseroline, is located in the proximity of the quaternary
anionic site of the binding site having a
π π
−
stacking with aromatic ring of Trp84. The
hydroxyl oxygn of eseroline bonds to carbonyl oxygen in His440. The N7 nitrogen has a
polar-polar interaction with hydroxyl group in Ser122. Also carbonyl group of Glu199
interacts with N5 in eseroline group. The interaction energies of the corresponding amino
5 Molecular docking studies on the “Back Door” hypothesis
81
acids are listed in Table 5.6. The most involved amino acid with eseroline is Trp48 with
interaction energy of -16.4 kJ/mol (Table 5.6).
Amino Acids Interaction energy (kJ/mol)
Ser122 -7.9
Gly118 -13.5
Gly117 -3.4
His440 -14.9
Glu199 -5.1
Trp84 -16.4
Table 5.6: Interaction energies of the amino acids that surround neutral eseroline
.
Figure 5.9: The position of docking solution of eseroline (gray) using MSD method. It is located in front of the
alternative channel, colored in red. Trp84 colored in blue, located in the entrance of the gate (left), the amino
acids of binding site in AChE interacting with the neutral eseroline (right)
5.3.2 G.O.L.D
By individually docking the neutral and protonated eseroline, using Gold Score, resulting
solutions show a larger fitness score for the neutral 177.52 (Docking solution 12, Table 5.7)
versus 124.90 for protonated eseroline (Docking solution 13, Table 5.7). These results are
once more in favor of the neutral state, as the fitness score of that has 51.96 weight more than
of protonated one.
5 Molecular docking studies on the “Back Door” hypothesis
82
Docking solution Docking solution Fitness score
12 (up) 177.52
13 N7 (p) 124.90
Table 5.7: Fitness score of the binding mode of the ligand in neutral and protonated states
Furthermore, Figure 5.10 demonstrates that the location of the neutral ligand (yellow)
obtained from Gold Score, is much closer to the docking solution of the neutral ligand with
QXP+ (gray), whereas Gold places the protonated ligand (magenta) quite far from the
resulting position of the neutral ligand. The identical positions, suggested by Gold Score and
QXP+ provide firstly, a high probability of the neutral state for eseroline. Secondly, due to the
placement of docking solutions of the neutral ligand (yellow and gray) in proximity of Trp84,
Tyr130 and His440 in front of the channel (Figure 5.10), it increase the probability of
clearance of eseroline via this particular alternative channel.
Figure 5.10: Overlaid docking results of Gold Score for the neutral eseroline (yellow) on docking solution of
QXP+ (gray). Docking solution of Gold Score for protonated eseroline (magenta) is far from the neutral ligand
(right), the position of docking solution of Gold Score (yellow) is in front of the alternative exit door, red
channel (left)
Docking of neutral and protonated eseroline was repeated using Chem. Score of G.O.L.D
program. In this experiment again the neutral ligand has better fitness score 177.40 (Docking
solution 14, Table 5.8) versus 125.44 for protonated ligand (Docking15, Table 5.8).
Docking solution Ligand* Fitness score
14 (up) 177.40
15 N7 (p) 125.44
Table 5.8: Fitness score of the binding mode of the ligand in neutral and protonated states
*:‘up’ stands for unprotonated and ‘p’ for protonated DMPO.
5 Molecular docking studies on the “Back Door” hypothesis
83
This experiment proposes that the positions of the ligand in both states, neutral (brown) and
protonated (magenta) are very close to each other as it is observable in Figure 5.11 (left). In
addition, the solution of the neutral ligand with both scoring functions of G.O.L.D are in
identical position, brown and yellow colored ligand, (right in Figure 5.11), with very close
fitness score (177.52 in Gold Score and 177.40 in Chem. Score).
In Figure 5.11 (right), it is obvious that docking solutions of the neutral ligand obtained from
QXP+ (gray), G.O.L.D with Gold Score (yellow) and Chem. Score (brown) are located very
close to each other. This confirms that the most possible location of the leaving group is
where the alternative channel exists, which is created by Trp84, Tyr130 and His 440, whose
shutter like motion permits the eseroline to leave the active site gorge of AChE (Figure 5.11).
Figure 5.11: The docking solutions of the neutral (brown) and protonated ligand (magenta) obtained from
Chem. Score (left), the result of Gold Score (yellow), Chem. Score (brown) and QXP+ (gray) by docking neutral
eseroline (right)
5.4 Docking experiment on eseroline in presence of waters
The effect of water molecules was considered to analyze their possible influence on leading
the leaving group eseroline in the direction of the alternative channel at the bottom of the
gorge. Therefore in next experiment eseroline was immersed into the waters of binding site
(Wat801, Wat805, Wat811, Wat812, Wat819, Wat831, Wat839, Wat840, Wat884, Wat842,
Wat892, and Wat903). This subset was the starting point for next experiment when flexible
waters were included in binding site. Result of this step clearly demonstrates the effect of the
waters to force the ligand toward the back door channel (green ligand in Figure 5.12).
In this experiment, the docking solution of the neutral eseroline interacts with Glu445,
Tyr485, Asn429, Val431, and Leu430 (green, Figure 5.12). This establishes the
5 Molecular docking studies on the “Back Door” hypothesis
84
circumstances, in which leaving group is led toward out of the gorge passing the back door
channel made by Trp84, Tyr130, His440 (Figure 5.12).
Table 5.9: Shows the amino acids that interact with docking solution of the neutral eseroline, in a binding site
with flexible waters.
Figure 5.12: Shows the position of the neutral eseroline proposed by docking into empty binding site (gray) and
with binding site having flexible waters (green), water molecules are shown in blue color. The back door channel
is red that has an entrance in active-site gorge that is open by shutter like motion of Trp84 (dark blue residue).
Table 5.9 indicates that Tyr442 is the most interactive amino acid with -24.7 kJ/mol
interaction energy.
The results of all the experiments provide existence of a back door in bottom of the gorge
where the eseroline can leave the binding site through that specific alternative pathway.
Amino Acids Interaction energy
(kJ/mol)
Glu445 -25.6
Tyr485 -2.5
Asn429 -7.5
Val431 -4.6
Leu430 -6.5
5 Molecular docking studies on the “Back Door” hypothesis
85
Since, eseroline is a leaving group not seen in X-ray structure, it thus stays in active site for a
very short time. Therefore it is reasonable to say that ligand must be neutral. If it were a
positively charged ligand it had to stay longer, due to the electrostatic interaction with high
negatively charged area at the base of the gorge. In that X-ray data should have identified its
location in the gorge.
As it is shown in Figure 5.13 [79] a very strong negative magnetic field in the lower part of
the gorge prevents that the positively charged ligand easily leave the gorge and it will cost
considerable energy to overcome this electrostatic interactions. Also the electrostatic
interactions of the amino acids in the base of the gorge would be in better electrostatic
interaction with positively charged moiety of the ligand than with neutral one. In Figure 5.14,
the red colored area in the base of the gorge shows the highest negatively charge in binding
site of AChE.
Figure 5.13: Shows the negative magnetic filed of the binding site gorge in AChE (red) and positive field (blue).
Trp279 in top of the gorge (magenta), Trp84 (white), Gly118 and Gly119 in brown, Asp72 (yellow) and Ser200
is shown in green [79]
In this investigation was found that DMPO is unprotonated, as the correct structures were
obtained at the result of docking unprotonated ligand using QXP+ and G.O.L.D. Docking
results of eseroline demonstrate that it is neutral and its orientation and location after cleavage
was found by simulated annealing. The result of docking into empty binding site proves that
among four possible exit doors (Figure 5.3) only the alternative channel-like pathway is the
exit door. Trp84, Tyr130 and His440 make this channel. In the entrance part of this channel,
Trp84 is located that permits the clearance of the eseroline by a shutter like motion. Adding
5 Molecular docking studies on the “Back Door” hypothesis
86
flexible waters of binding site, extracted from X-ray structure, led to finding the pathway for
eseroline for exiting the gorge, as the docking result shows that eseroline locates out of the
gorge by the force of waters.
7 Docking study on galanthamine derivatives
87
6 Binding mode prediction of galanthamine derivatives using MSD
In an effort to increase the affinity of galanthamine for AChE, galanthamine derivatives were
designed with the aim to create bisfunctional compounds that interact with two binding sites
on the AChE. The sites targeted on AChE are the anionic subsite of the active site in the
bottom of the gorge, and the peripheral anionic site near top of the gorge.
One of these derivatives of galanthamine is a bisamide analogue, in which two galanthamide
of the molecule are connected via an alkyl side chain. In this work, bisgalanthamides with
different alkyl spacer length are studied, in which the number of methylen groups differentiate
between 1 till 8 (Figure 7.1).
O
O
O
H
H
N
O
O
O
H
H
N
OO
(CH
2
)n
Figure 7.1: Chemical structure of bisgalanthamide, which has 1 till 8 methylen groups, in the alkyl spacer.
6.1 Binding mode prediction of bisgalanthamide analogues into AChE binding site
To find an optimal derivative of galanthamine that covers the entire space of the binding site
in AChE, eight analogues of bisgalanthamide ligands with a spacer of one till eight methylen
groups were considered by docking the ligands into the active site of 1QTI.
The numerical results, from eight individual MSD experiments on these bisgalanthamides, are
shown in Table 7.1. As regards to the highest STotal in bisgalanthamide with heptyl chain
(Docking solution 2 in Table 7.1), the ligand with heptyl spacer is suggested as the optimal
length for connecting amides substructures. Figure 7.2 demonstrates that in all the ligand,
with 8 different spacer-lengths, the locations of the galanthamides are identical at the lower
part of the gorge. The main change is distinguishable in the upper part of the ligand, which
occurs due to the high flexibility of the side chain, as the longer chain has the higher degrees
of freedom (Figure 7.2).
7 Docking study on galanthamine derivatives
88
Docking
solution
Nr.
CH2
Eass
(kJ/mol)
LE
(kJ/mol)
Enhb
(kJ/mol)
vdW
(kJ/mol)
Eest
(kJ/mol)
Ecnt
(kJ/mol)
vdW+
(kJ/mol)
Nhph Nhb
1 8 -62.5 0.0 -62.5 -13.0 -14.5 -35.0 4.2 13 1
2 7 -67.2 0.0 -67.2 -6.9 -21.3 -39.0 11.0 18 3
3 6 -68.0 0.0 -68.0 -11.2 -19.0 -37.8 6.7 13 2
4 5 -55.8 0.0 -55.8 -11.0 -11.7 -33.1 4.9 15 2
5 4 -57.0 0.0 -57.0 -12.1 -11.5 -33.5 4.7 12 2
6 3 -56.2 3.0 -59.2 -10.4 -12.4 -36.4 6.2 15 1
7 2 -52.0 0.0 -52.0 -9.7 -11.9 -30.4 5.3 12 1
8 1 -43.5 2.1 -45.6 -1.0 -10.8 -33.8 13.9 10 2
Table 7.1: Results of docking study on eight derivatives of bisgalanthamide. The ligand with heptyl side chain
has the highest STotal (red numbers); Nr. CH2: Is the number of methylen in spacer
1CH2 2CH2 3CH2
4CH2 5CH2 6CH2
7CH2 8CH2
Figure 7.2: Indicates the position of docking solutions of eight different bisgalanthamide derivatives as regards
to the position of Trp279 and Trp84 at the upper and the lower part of the gorge, respectively.
7 Docking study on galanthamine derivatives
89
The best bisgalanthamide derivative among the eight studied amides is the one with (CH2)7
spacer. The amino acids of the gorge that interact with different part of the ligand and their
corresponding calculated interaction energy are shown in Figure 7.3 and Table 7.2,
respectively.
Figure 7.3: The overlaid X-ray structure of galanthamine (green) on the docking solution of
heptlybisgalanthamide. The bisgalanthamide derivative (gray) interacts with the amino acids directly in the PAS
area and with the mid part of the gorge, while galanthamine has no access to them.
Table 7.2: Lists the interaction energy of the involving amino acids in formation of the complex with
heptylbisgalanthamide.
The overlaid X-ray structure of the galanthamine (green) and heptylbisgalanthamide, in
Figure 7.3, shows that how well the designed ligand interacts in the rim of the gorge with
Trp279 in peripheral anionic subsite, however, among the rest of bisgalanthamide derivatives
there are some which also interact with Trp279, but the optimum ligand (with heptyl spacer)
interacts also with component of the acyl pocket of the gorge at Phe288, Phe290 and Phe331.
In the rim of the gorge it has also hydrophobic interactions with Asp285 and Ser286. The
most valuable interaction energies correspond to the Phe331 (-18.3 kJ/mol) in the acyl pocket
and Trp279 in PAS (-13.1 kJ/mol) (Table 7.2).
6.2 Binding mode prediction of (IHG) into AChE binding site
Another possible inhibitor of galanthamine is indanonehexylgalanthamine (IHG), in which a
galanthamine moiety of the ligand is connected to the indanone via a hexyl side chain.
Chemical structure of IHG indicates that indanone substructure of the ligand can make it
Amino acids Interaction
energy (kJ/mol)
Asp285 -6.0
Trp279 -13.8
Arg289 -5.6
Gly335 -6.4
Phe288 -13.2
Phe330 -6.7
Phe331 -18.3
Tyr121 -4.9
Ser122 -3.2
Ser286 -2.1
7 Docking study on galanthamine derivatives
90
share in inhibitory effect of E2020 at the upper part of gorge. It can take advantage of special
characteristic feature of galanthamine in the lower part of the binding site gorge in AChE. The
hexyl side chain can play similar role of the high flexible side chain of BHG that was
discussed in chapter 3, which can improve inhibitory effect of the galanthamine by its
elongation within the entire 20.0 Å deep and narrow binding site. In order to consider the
possibility of having a more potent ligand than galanthamine, E2020 and BHG,
indanonehexylgalanthamine (IHG) was studied using MSD method. This would enable us to
indicate the possible interaction that can be established by the new ligand as a combination of
three potent inhibitors of AChE (Figure 7.4).
O
O
OH
H
N
12
C
27
O
O
O
Galanthamine | Hexyl | Indanone
Figure 7.4: Chemical structure of indanonehexylgalanthamine (IHG)
Using docking technique, it was tried to identify the conformational state at galanthamine
nitrogen (N12) and configurational state of the molecule in the chiral center (C27) at indanone
moiety of the ligand.
6.3 Docking IHG in binding site of 1EVE
Due to similarity of the IHG to E2020 at indanone moiety, the ligand after minimization was
docked into the binding site of AChE using 3D structure of 1EVE, which is the complexed
protein with E2020. The ligand at the starting point for docking was considered in four
different forms, in which C27 was either in R or S-configuration, when the conformational
state at galanthamine nitrogen was either axial or equatorial. It means that four series of
independent and individual multi-step docking runs were carried out, in which the ligand had
‘a-S’, ‘e-S’, ‘a-R’ and ‘e-R’ poses, where the first letter stands for axial or equatorial
conformation at galanthamine nitrogen and the second letter for S and R configuration at C27
in indanone moiety of the ligand. The obtained numerical results after running MSD
experiments have been shown in Table 7.3.
7 Docking study on galanthamine derivatives
91
Docking
solution
Starting
configmation
Eass
(kJ/mol)
LE
(kJ/mol)
Enhb
(kJ/mol)
vdW
(kJ/mol)
Eest
(kJ/mol)
Ecnt
(kJ/mol)
vdW+
(kJ/mol)
Nhph Nhb
1 e-S -53.8 4.4 -58.2 -2.8 -15.0 -40.4 16.0 13 2
2 a-S -45.5 3.6 -49.1 -6.5 -6.9 -35.8 9.8 14 1
3 e-R -59.5 0.0 -59.5 -2.6 -15.9 -41.0 15.9 14 1
4 a-R -47.1 8.6 -55.7 -2.4 -13.7 -39.6 15.7 14 3
Table 7.3: Shows the numerical results of docking the ligand in different conformational and configurational
states in its starting point.
The highest STotal belongs to the staring structure of equatorial with R-configuration (Docking
solution 3 in Table 7.3), which is also the only one with ligand energy of 0.0 kJ/mol.
Additionally the starting structures with axial conformation, after involving in multi-step
minimization procedure, at the end still keeps their configuration and conformational state and
no axial is changed to equatorial. This might be due to heavy and long side chain substitution
on nitrogen, whose conformational change requires a large amount of conversion energy.
Overlaid X-ray structures of E2020 (orange), galanthamine (green) and BHG on the docking
solution of ‘e-R’-IHG (gray) show that the potential of the ligand is improved in sense of
better interacting with Trp279 at PAS. Its galanthamine nitrogen is closer to aromatic ring of
Phe330 in a magnitude of 0.53 Å. It has also better aromatic-aromatic interaction with Trp84
in the lower part of the ligand than of that in galanthamine. The detailed interaction energies
of the ligand with surrounding amino acids have been listed in Table 7.4. Although, IHG has
the same alkyl chain as in BHG, the slightly bent form of that in IHG makes it to have better
aromatic-aromatic interaction with Trp279 than in BHG (Figure 7.5).
Figure 7.5: Shows the overlaid X-ray structure of galanthamine (green), E2020 (orange) and BHG (white) on
the docking solution of ‘e-R’ (IHG). ‘e-R’ (IHG) has better interaction with Trp279 than E2020. Its galanthamine
nitrogen is closer to Phe330 aromatic and the double bond of its hydroxyl hexenyl ring closer to Trp84 than
galanthamine.
7 Docking study on galanthamine derivatives
92
Figure 7.6: The amino acids interacting with IHG (e-R) in the binding site of AChE (1QTI)
Table 7.4: The interaction energies of IHG with different amino acids in binding site of AChE
Comparing the interaction energy with Trp279 in E2020, BHG and in IHG, which are -16.9
kJ/mol, -20.0 kJ/mol and -44.0 kJ/mol, respectively, reveals that IHG has the best position for
involving with Trp279 among three mentioned ligands of AChE.
IHG, due to its better interaction with residue in PAS and its ability to cover the binding site
gorge in the upper and the lower amino acids of the gorge is suggested as a good candidate to
be synthesized. For this purpose there is strongly emphasis on R-configuration in the chiral
center at indanone substructure of the ligand.
Amino acids Interaction
energies (kJ/mol)
Leu282 -3.4
Trp279 -44.0
Tyr121 -6.0
Tyr334 -8.2
Gly118 -5.7
Ser200 -21.7
Glu199 -23.0
Trp84 -22.7
Phe330 -4.4
Phe331 -8.2
Asp72 -4.8
Ser200 -21.7
9 References
93
7 References
[1] W. Moos, F. Hershenson, Drug News &Perspectives, 1989, 2, 397-409
[2] J. L. Cummings, S. Askin-Edgar, CNS Drugs., 2000, 13, 385-395
[3] D. A. Evans, H. Funkenstein, M. S. Albert, P. A. Scherr, N. R. Cook, M. J. Chown, L.
E. Hebert, C. H. Hennekens, J. O. Taylor, J. Am. Med. Aso., 1989, 262, 2551-2556
[4] G. V. Ferrari, M. A. Canales. L. M. Weiner, I. Silman, N. C. Inestrosa, Biochemistry.,
2001, 40, 10447-10457
[5] C. M. Coughlan, C. K. Brean, Pharmacol. Therapeut., 2000, 86, 111-116
[6] I. Sugimoto, Y. Yamanishi, Y. Iimura, Y. Kawakami, Cur. Med. Chem., 2000, 7, 303-
339
[7] E. A. Barnard, The Peripheral Nervous System, Plenum Press, New York., 1974, pp,
201-224
[8] D. M, Quinn, Chem. Rev., 1987, 87, 955-975
[9] J. L. Sussman, M. Harel, I. Silman. Chem. Biol. Interactions., 1993, 87, 187-197
[10] J. Sussman, M. Harel, F. Frolow, C. Oefner, A. Goldman, Science, 1991, 253, 872-879
[11] P. Taylor, Z. Radic, Annu. Rev. Pharmacol. Toxicol., 1994, 34,281-320
[12] Y. P. Pang, A. Kozikowski, J. Comput,–Aided Mol. Des., 1994, 8, 669-681
[13] G. Kryger, I. Silman, J. Sussman, Structure Fold Des., 1999, 7, 297-307
[14] M. W. Qin, H. L. Jiang, K. X. Chen, R. Y. Ji, YJ. Ye., In. J. Quant. Chem., 1999, 74,
315-330
[15] P. Camps, D. Munoz-Torro, Mini Reviews in Medicinal Chemistry, 2002, 2, 11-25
[16] J. A. Pesti, Chem. Innovation, 2000, 30, 22-28
[17] H. J. Pieniaszek,W. D. Fiske, T. D. Saxton, Y. S. Kim, D. M Garner, M. Xilinas, R.
Martz, J. Clin. Pharmacol., 1995, 35, 22-30
[18] M. W. Qin, H. L. Jiang, K. X. Chen, R. Y. Ji, YJ. Ye, In. J. Quant. Chem., 1999, 74,
315-330
[19] L. S. Schneider, Curr. Opin. Invest. Drugs, 2000, 2, 427-437
[20] E. Rose, J. Aleu, J. Marsal, C. Solsona, Eur. J. Pharmacol., 2000, 390, 7-13
[21] A. Martinez, E. Fernandez, A. Castro, S. Conde, I. Rodriguez-Franco, J. E. Banos, A.
Badia, Eur. J. Med. Chem., 2000, 35, 913-922
[22] I. Villalobos, J. F. Blake, C. K. Biggers, T. W. Butler, D. S. Chapin, Y. L. Chen, J. L.
Ives, S. B. Jones, D. R. Liston, A. A. Nagel, D. M. Nason, J. A. Nilsen, I. A. Shalaby,
W. F. White, J. Med. Chem., 1994, 37, 2721-2728
9 References
94
[23] J. J. Sramek, E. J. Frackiewicz. N. R. Cutler, Exp. Opin. Invest. Drugs, 2000, 9, 2393-
2402
[24] P. Camps, D. Munoz-Torro, Mini Reviews in Medicinal Chemistry, 2002, 2, 11-25
[25] I. D. Kuntz, J. M. Blaney, S. J Oatley, R. Langridge, T. E. Ferrin, J. Mol. Biol., 1982,
161, 269-288
[26] R. Abagyan, M. Totrov, Curr Opin. Chem. Biol., 2001, 4, 375-382
[27] I. Halperin, B. Ma, R. Nussinov, Proteins, 2002, 47, 409-443
[28] Bissantz, G. Folkers, D. Rognan, J. Med. Chem., 2000, 43, 4759-4767
[29] M. Rary, B. Kramer, T. Lengaure, J. Comput,–Aided Mol. Des., 1997, 11, 369-384
[30] M. Vieth, J. D. Hirst, B. N. Domin, J.Comput. Chem., 1998, 19, 1623-1631
[31] O. M. Becker, A.D. MacKerell, Jr. Benoit, M. Watanabe, Computational Biochemistry
and Biophysics, 2001, pp74
[32] C. McMartin, R. S. Bohacek, J. Comp-Aid. Mol. Des.,1997, 11, 333-344
[33] S. J. Weiner, P. A. Kollman, D.A. Case, U.C. Singh, C. Ghio, G. Algona, S. Profeta, P.
Weiner. J. Am. Chem. Soc., 1984, 106, 765-784
[34] J. Trosset, H. Scheraga, Biophysics, 1998, 95, 8011
[35] N. Go, H. Scheraga, Macromolecules, 1978, 11, 552-559
[36] J. Trosset, H. Scheraga, J. Comput. Chem., 1999, 20, 244-252
[37] M. Metzei, J. Chem. Phys., 2002, 8, 3874-3882
[38] [Personal communication with Dr. Lamba], International Center for Genetic
Engineering and Biotechnology, Trieste-Italy
[39] G. Kryger, I. Silman, J. Sussman, Structure Fold Des., 1999, 7, 297-307
[40] H. Greenblatt, G. Kryger, T. Lewis, I., Silman, J., Sussman, FEBS Lett, 1999, 463, 321-
326
[41] M. Harel, I. Schalk, L. Ehret-Sabatier, F. Bouet, M. Goeldner, C. Hirth, P. Axelsen, I.
Silman, J. Sussman, Proc Natl Acad Sci U S A., 1993, 90, 9031-9035
[42] Galli, F. Mori, L. Benini, N. Cacciarelli, Eur. J. Pharmacol., 1994, 270, 189-93
[43] R. L. Irwin, H. Smith, J. Biochem. Pharmacol., 1960, 3, 147-149.
[44] L. Czollner, W. Frantsits, B. Küenburg, J. Fröhlich, Tetrahedron. Lett., 1998, 39, 2987-
2988
[45] A. Maelicke, E. X. Aluquerque, Drug. Design. Thecnol., 1996, 1, 53-59
[46] N. C. Inestrosa, A. Alvarez, CA. Perez, RD. Moreno, M. Vicente, C. Linker, O. I.
Casanueva, C. Soto, J. Garrido, Neuron, 1996, 16, 881-891
9 References
95
[47] A. Alvarez, C. Opazo, R. Alarcon, J. Garrido, N. C. Inestrosa, J. Mol. Biol., 1997, 272,
348-361
[48] J. A. Hardy, G. A. Higgins, Science, 1992, 256, 184-185
[49] D. J. Selkoe, Annu. Rev. Neurosi., 1989, 12, 463-490
[50] F. J. Munoz, R. Aldunate, N. C. Inestrosa, NeuroReport, 1999, 10, 3621-3625
[51] G. Johnson, S. W. Moore, Biochem. Biophys. Res. Commun., 1999, 258, 758-762
[52] C. F. Contreras, M. A. Canales, A. Alvarez, G. V. De Ferrari, N. C. Inestrosa, Protein.
Eng., 1999, 12, 959-966.
[53] V. Giancarlo. I. Shin. I. Sillman, Biochemistry, 2001, 40, 10447-10457.
[54] E. Luttmann, E. Linnemann, G. Fels, J. Mol. Model., 2002, 8, 208-216
[55] G. Koellner, G. Kryger.C. B. Millard. I. Silman, J. L. Sussman.T. Steiner, J.Mol.Biol,
2000, 296, 713-735
[56] J. Antosiewicz, J. A. McCammon, M. K. Gilson, Biochemistry, 1996, 35, 7819-7833
[57] M. G. Nodberg, J. Marelius, S. Ohlsson, A. Persson, G. Swedberg, P. Andersson, S.E.
Aqvist, A.Hallberg. J. Med. Chem., 2000, 43, 3852-3861
[58] H. Gohlke, K. Klebe, Angew. Chem. Int. Ed., 2002, 41, 2645-2676
[59] G. M. Bors, F. P. Huger, W. Petko, A. E. Mutlib, F. Camacho, D. K. Rush, D. E. Selk,
v. Wolf, R. W. Kosley, L. Davis, H. M. Vargas, J. Pharm. Exper. Ther., 1996, 277, 728-
738
[60] K. L. Davis, P. Powchik, J. Am. Med. Aso., 1997, 277, 10-15
[61] H. Sugimoto, Y. Imura, Y. Yamanishi, K. Yamatsu, J. Med. Chem., 1995, 38, 4821-
4829
[62] A. Inoue, T. Kwai, M. Wakita, Y. Imura, H. Sugimoto, Y. Kwakami, J. Med. Chem,
1996, 39, 4460-4470
[63] M. Harel, I. Schalk, L. Ehret-Sabatier, F. Bouet, M. Goeldner, C. Hirth, P. Axelsen, I.
Silman, J. Sussman, Proc Natl Acad Sci U S A., 1993, 90, 9031-9035
[64] M. Harel, I. Schalk, L. Ehert-Sabatie, F. Bouet, M. Goeldner, C. Hirth, P. H. Axelsen, I.
Silman , L. Sussman, Proc. Natl. Acad. Sci., 1993, 90, 9031-9035.
[65] B. G. Benson, H. N. Bramson, S. H. Dickerson, K. M. Dold, D. J. Eberwein, M.
Edelstein, S. V. Frye, R.T. Gampe, R. J. Greiffin, P. A. Harris, A. M. Hassell, W. D.
Holmes, R. N. Hunter, V. B. Knick, K. Lackey, B. Lovejoy, M. Luzzio, D. Murray, P.
Parker, W. J. Rocque, L. Shewsh, Science, 2001, 291, 134
[66] C. N Hiremath, R. AC. Grant, D. J. Filman, J. M. Hogle, Acta Crystallogr D Biol
Crystallogr, 1995, 51, 473-489
9 References
96
[67] N. F. Proskurina, A. P. Yakovleva, J. Gen. Chem., USSR.,1952, 22, 1941-1944
[68] J. N. Champness, M.S. Bennett, F. Wien, R. Visse, W. C. Summers, P. Herdewijn, E. de
Clerq, T. Ostrowski, R. L. Jarvest, M. R. Sanderson, Proteins, 1998, 32, 350-361
[69] M. S. Searle, D. H. Wiliams, U. Gerhard, J. Am. Chem. Soc., 1992, 114, 1097-10704
[70] E. Grunwald, C. Steel, J. Am. Chem. Soc., 1995, 117, 5687-5692
[71] R. Taylor, O. Kennard, J. Am. Chem. Soc., 1982, 104, 5063-5070
[72] S. Tsuzuki, K. Honda, J. Am. Chem. Soc., 2001, 307, 367-377
[73] C. A. Hunter, J. K .M. Sanders, J. Am. Chem. Soc., 1990, 112, 5525-5534
[74] G. M. Bors, F. P. Huger, W. Petko, A. E. Mutlib, F. Camacho, D. K. Rush, D. E. Selik,
V. Wolf, R. W. Kosley, L. Davis, H. M. Vargas, J. Pharmacol. Exp. Ther, 1996, 277,
828-738
[75] H. K. Hall, J. Am. Chem. Soc., 1964, 3135-3138
[76] H. C. Brown, J. Am. Chem. Soc., 1945, 67, 378-381
[77] A. Shafferman, D. Barak, EMBO Journal, 1994, 13, 3448-3455
[78] M. K. Gilso, T. P. Straatsma, J. A. McCammon, D. R. Ripoll, C. H. Faerman, P. H.
Axelse, I. Silman; J. L. Sussman, Science, 1994, 263, 1276-1278.
[79] C. E. Felder, J. Sussman, I. Silman, J. Mol. Graph. Mod., 1997, 15, 318-327
[80] J. Silman, Science, 1991, 253, 872-879
[81] C. Bartolucci, E. Perola, L. Cellia, M. Brufani, D. Lamba, Biochemistry, 1999, 38,
5714-5719,
[82] H. M. Greenblatt, C. Guillo, D. Guenard, A. Arhaman, S. Botti, B. Badet, C. Thal, I.
Silman, JL. Sussman, J. Am. Chem. Soc., 2004, 126, 15405-1541
[83] G. Jones, P. Willet, R. C. Glen, J. Mol. Biol., 1995, 254, 43-53
[84] G. Jones, P. Willet, R.C. Glen, J. Robert, A. R. Leach, R. Taylor, J. Mol. Biol., 1997,
267, 727-748
8 Abbreviations
AChE Acetylcholinesterase
BChE Butyrylchinesterase
E2020 (R,S)-1-benzyl-4-[5,6-dimethoxy-1-indanone)-2-yl] methylpiperidine
FF Force Field
G.O.L.D Genetic Optimization for Ligand Docking
hBChE human butyrylchinesterase
MF268 8-(cis-2,6-dimethylmorpholino) octylcarbamoyleseroline
MM Molecular Mechanics
PAS Peripheral Anionic Site
QXP Quick eXPlore
RMSD Root Mean Square Deviation
Eass Total estimated binding energy
Elig Ligand energy relative to estimated global minimum of free ligand
Eshe Binding site energy relative to local minimum for empty site
Enhb Total energy of non-bonded interactions between ligand and site
vdW van der Waals energy of interactions between ligand and site
Eest Electrostatic energy of interactions between ligand and site
Enhb Non-hydrogen bonding energy
GM Global Minimum energy
LM Local minimum energy
FDA United States Food and Drug Administration
BBB Blood Brain Barrier
AChEI Acetylcholinesterase inhibitors
PDB Protein Data Bank (www.pdb.org)
MSD Multi-Step Docking
DECA Decamethonium
IHG Indanonhexylgalanthamine
BHG Benzosulfimidohexylgalanthamine
PPG
Piperidinopropylgalanthamine
