Influenza A virus is a highly contagious viral pathogen of birds and mammals,
including humans (Webby and Webster, 2001). The most
infamous pandemic, the Spanish flu, manifested a deep impact on world population
mortality. It is estimated that there were at least 50 million mortalities in
1918-1919 (Betakova, 2007). In 2009, swine flu again
attacked various countries in the world. World health Organization (WHO) set
a pandemic influenza A H1N1 virus as a global pandemic on June 11, 2009. At
least, there were approximately 18,449 people worldwide who died from this virus
attack. Then, on August 10, 2010, WHO officially announced that the swine flu
pandemic in the world has ended and changed into post-pandemic phase. Post-pandemic
phase means that the spread of flu activity around the world has returned to
a usual level of treatment as it occurs seasonally. Post pandemic phase is a
phase most appropriate to find antivirals that can overcome this viral infection.
H1N1 virus is a global pandemic, which already affected several countries such
as Malaysia, Iran and Nigeria (Yahyapour et al.,
2007; Sar et al., 2010; Karim
and Razali, 2011). The escalation of this pandemic deserves more concern,
because Alsaif et al. (2010) reported that there
were certain resistances for taking H1N1 vaccination in Hail Community. This
could hamper the process of eradicating the flu pandemic.
Swine flu has been identified as a new strain of influenza virus A H1N1. Influenza
A virus H1N1 is a combination of swine influenza virus genes, avian and human.
Based on genetic characterization, gene Hemagglutinin (HA) is similar to swine
influenza viruses in the United States, while the Neuraminidase gene (NA) and
matrix protein (M) is similar to the swine flu virus isolated from Europe. This
unique genetic combination has not been previously detected anywhere, and no
patient who had direct contact with pigs that could bring the possibility of
transmission of this new strain of influenza virus among humans has been seen
Many antiviral drugs have been developed to overcome these infections, namely
oseltamivir, zanamivir, amantadine and rimantadine, among others. Oseltamivir
and zanamivir are neuraminidase inhibitors. Both antiviral drugs are recommended
by the Centers of Disease Control and Prevention (CDC) and WHO (Rungrotmongkol
et al., 2009), as it is considered effective in treating influenza
virus infection that occurs in humans (Lew et al.,
2000). However, antiviral resistance has been reported lately. Influenza
virus type A (H1N1) resistance against oseltamivir was reported in Europe in
years 2007-2008 (Cheng et al., 2009). In addition,
there are also amantadines and rimantadine which can be used to inhibit M2 ion-channel
proteins (De Clercq, 2006). The U.S. Food and Drug Administration
have approved amantadine and rimantadine in October 1966 as an agent for prophylaxis
against influenza in Asia. Nonetheless, as time went on, there were possible
drug-resistant viruses due to antigenic drift mutations (Beigel
et al., 2005; De Jong et al., 2005).
The level of amantadine and rimantadine resistance increased from below 10%
during 1995-2002, after which it increased to 58% in 2003, 74% in 2004 and 92%
M2 channel protein in influenza A virus is one of the targets of anti-influenza
drug that had been used to control influenza virus infection, namely, amantadine
and rimantadine. However, the recent use of the drugs, amantadine and rimantadine,
is limited because of the drugs resistance. The resistance due to mutations
of the influenza virus strain is a mutation of three residues on influenza A
H5N1 virus, that is, at position V28, S31 and L43, the residues were mutated
into I28, N31 and T43. Among the three mutated residues, the S31N mutation is
most common in the M2 protein channel that can cause resistance to amantadine
(Rungrotmongkol et al., 2009). In addition to
the resistance of amantadine and rimantadine, it was also recently reported
that amantadine also has adverse effects on the central nervous system (De
The harmful amantadine resistance as antiviral drugs was a good motive for
searching for inhibitors of the M2 protein channel that can replace the derivative
compound of adamantine. Before now, several inhibitors of the M2 protein channel
have been reported. They are amantadine, rimantadine, amantadine derivative,
inhibitors of non-adamantane, and Isopinocampheylamine (Hu
et al., 2010). There are some research groups, such as Sohail
et al. (2011), who reported that certain lead compounds from plant
could exhibit anti-Influenza-viral resistance. However, their research still
needs more conclusive results to comprehend the molecular mechanism of the lead
compounds. This is one reason why molecular modeling could play an important
role in this field.
This studys research group has reviewed some notable research on molecular
modeling and successfully conducted a research on H5N1 mutation. Moreover, we
are confident that our previous publication methodology could be utilized for
H1N1 research as well (Tambunan et al., 2008,
Molecular docking and dynamics are commonly used methods in bioinformatics.
Some research groups have successfully utilized it for PB2 of influenza virus,
and anti bacterial resistance towards antibiotics (Lakshmi
et al., 2011; Amir et al., 2011).
Sur et al. (2009) reported that there are high
expressions of pathogenicity related genes which confirm its role as pathogen.
Moreover, most of the H1N1 basic proteomes are influenced by mutational pressure.
This is one reason why proteomic based research on H1N1 is still considered
From the research of Zhao et al. (2011), it
was reported that compound number 5, the compound (1R, 2R, 3R, 5S)-(-)-isopinocampheylamine,
is the most potent compound as inhibitors of the M2 protein channel and is three
times more active than the M2 inhibitors amantadine as channel protein
(IC50 = 1363 vs. 5960 μM). Therefore, this study created a modification
of the compound (1R, 2R, 3R, 5S)-(-)- isopinocampheylamine to find a more potent
M2 protein channel inhibitor on the H1N1 virus.
The purpose of this research was to design a drug for influenza A H1N1 virus by in silico based compound (1R, 2R, 3R, 5S)-(-)- isopinocampheylamine through molecular docking method. This compound was expected to replace the antivirals amantadine and rimantadine.
MATERIALS AND METHODS
This study was conducted at Bioinformatics Laboratory, Department of Chemistry, Faculty of Mathematics and Science, University of Indonesia. The research duration was between January 2011 and July 2011.
Searching for protein sequences of H1N1 virus M2 channel: The search
for protein sequence M2 H1N1 virus in human and pig host server was done by
GenBank Influenza Virus (IVR-NCBI) online database. Homepage-IVR servers can
be accessed through the NCBI website address at http://ncbi.nlm.nih.gov/genomes/FLU/Database.html.
The format sequence used to store the M2 protein sequence was FASTA, because
it was the most widely used for genome analysis.
Multiple sequence alignment: H1N1 virus M2 protein sequences of human
and pig hosts in FASTA format were incorporated into the display of Multiple
Sequence Alignments (MSA) ClustalW2 server. Its homepage can be accessed online
MSA was performed on the entire H1N1 virus M2 protein. Then, the M2 protein
sequences having the highest bit score was used as input for subsequent analysis.
Three-dimensional protein structure search of the M2 channel: The search
for three-dimensional (3D) structure of the sequence was carried out using the
software of Max Planck Institute Hhpred, which can be accessed through the website
The selected 3D structure has the highest similarity to the receptor binding
protein M2 channel sequences.
Search data of 3D protein GDP-M2 channel structure: three-dimensional
structure of the M2 channel can be downloaded from the PDB database available
at the Research Collaboratory for Structural Bioinformatics Protein Data Bank
(RSCB-PDB) through http://www.rscb.org/pdb/.
Geometry optimization and energy minimization of M2 protein channel: Geometry optimization and energy minimization of the M2 channels 3D structure was performed using Molecular Operating Environment (MOE) software.
Modification of (1R, 2R, 3R, 5S)-(-)- isopinocampheylamine compounds modeled into 3D structures: Modified forms of (1R, 2R, 3R, 5S)-(-)- isopinocampheylamine were performed using ChemSketch software from ACDLabs.
3D structure geometry optimization and energy minimization of ligands: Geometry optimization and energy minimization of the 3D energy structure of (1R, 2R, 3R, 5S)-(-)- isopinocampheylamine (ligand) was performed using MOE software. The algorithm used was the Polak-Ribiere conjugate gradient with a maximum gradient Root Mean Square (RMS) convergence of 0.001 kcal mol-1 Å and molecular mechanics force field parameters MMFFx.
Protein ligand docking with M2 channel: The process begins with the preparation of docking files. It was accomplished using the software contained in the MOE. For both the ligand and the protein, hydrogen was added to both polar and charge forcefield. The M2 channel protein was saved in MOE format for later use in the preparation parameters. Calculations were carried out with the docking parameters of Lamarckian Genetic Algorithm (LGA). These parameters were stored in the MOE format as files used to run the docking process; though the docking process is run using the MOE software.
Determination of protein-ligand complex conformation docking results: The result of docking calculations was saved in text (notepad) format. Determination of protein-ligand conformation docking was done by selecting the ligand conformation that has the lowest binding energy.
Energy association and inhibition constant (Ki): Bond energies and inhibition constants docking results were saved in notepad format. Selected complex protein-ligands which have the smallest value of bond energy and inhibition constants were used for further analysis.
Hydrogen bonding: Hydrogen bonding, which occurs in the M2 channel protein complex best ligand docking results, was identified using MOE software with the input file Mdb.
Residue contacts: Contact residues of complex enzyme-ligand docking results were identified using MOE software.
Analysis of drug scan: Analysis of drug scans was performed on the results of the docking analysis of ligand binding that has a low ΔG0 corresponding to the normal distribution which is 20% of the lowest energy. This analysis was performed by comparing the best ligand docking analysis to the rules of good medicine (Lipinsky's Rule of Five).
Toxicity prediction: The analysis performed on the results of best ligand docking toxicological properties was based on the analysis results of screening and drug scans. Parameters to be seen among others are predicting carcinogenicity and mutagenicity of these ligands by using some software such as ToxTree and Lazar toxicological properties.
RESULTS AND DISCUSSION
Preparation of protein M2 channel of influenza A (H1N1)
Searching for protein sequences of influenza A virus M2 channel (H1N1):
Sequences of the M2 protein channel that will serve as the target protein was
determined by using a database of influenza contained in the official website
of the NCBI (http://www.ncbi.nlm.nih.gov/genomes/flu/).
In this site, Influenza Virus Resources Database allows us to get the influenza
virus data with the desired specifications. This specification covers the type
of influenza, the host of the virus, the isolated virus country of origin, the
desired type of protein sequence, and subtypes of influenza viruses. The used
parameters are, among others, influenza type A, a selected sequence of the H1N1
subtype M2 protein channel between 2009 and 2010 from the whole world and the
human host. Retrieved protein sequences were FASTA format with the full-length
type. The purpose of the download was for facilitating the subsequent analysis.
From this protein sequences determination, 173 M2 channel sequences from various
countries were obtained.
Multiple sequence alignment: One hundred and seventy-three protein sequences of the M2 channel were determined to be targeted using MSA. The resulted best representative sequence will be targeted in this study. The determination was done with online clustalW2 of EBI official site by looking at the similarity score of each compared sequence with the highest value of 100. The selected M2 channel protein sequence that has the highest similarity value (100) is the M2 channel protein of Addis Ababa, Ethiopia, in 2009, with the code matrix protein 2 [Influenza A virus (A/Addis Ababa/WR2848T/2009 (H1N1))] and GenBank code ADM14979.1. NCBI Data which stipulate that sequence> gi | 303385772 | gb | ADM14979.1 | (Attachment 5) was isolated from the human sequence of a 2 year-old girl. These sequences were collected on August 18, 2010 by Watler Reed Army Institute of Research, 503 Robert Grant Ave., Silver Spring, Maryland 20910, USA.
Identification of homology modeling and templates: From the alignment,
protein sequence of the M2 channel of Influenza A H1N1 virus (A/Addis Ababa/WR2848T/2009
(H1N1) was used as target proteins for structure determination of 3D template
channel M2 protein of influenza A (H1N1). Determination of the 3D structure
of the M2 protein channel influenza A H1N1 (A/Addis Ababa/WR2848T/2009 (H1N1)
was done using the software of the Max Planck Institute Hhpred, which could
be accessed through the website http://toolkit.tuebingen.mpg.de/hhpred.
Results of Hhpred showed that the M2 protein channel of Influenza A H1N1 (A/Addis
Ababa/WR2848T/2009 (H1N1) has 100% similarity with the M2 channel protein with
PDB code 2KIH composed of 43 amino acid residues. The existing crystal structure
was the result of Nuclear Magnetic Resonance (NMR) research.
The active visualization of M2 protein channel: 3D structure of the
M2 channel protein was obtained from the PDB. Visualization was conducted to
see the position of residues that have important functions and will be used
as a target residue in the docking process. Based on the publication of Du
et al. (2010), it was known that the M2 channel protein activity
was determined by three important residues, namely, His37, Trp41 and Asp44,
as seen in Fig. 1. Three residues were called the functional
residue because each one of them has a special function. They are, His37 as
a pH sensor, Trp41 as the channel gate, and Asp44 as a channel lock. These three
residues have a fixed position (highly conserved) in influenza A virus M2 channel
(Du et al., 2010).
|| Visualization of M2 channel protein influenza A (H1N1)
In the tertiary structure of proteins, amino acid residues are located on the
hydrophilic exterior (surface), while the hydrophobic residues are generally
located on the interior of the protein (Lehninger and Nelson,
2005). Hydrophobicity and hydrogen bond properties of the channel M2 protein
of influenza virus A (H1N1) was determined using the software MOE 2008.10. Residues
of the M2 channel in Fig. 2 were visualized using the map
program in software MOE 2008-10 and utilizing the surface Gaussian contact parameter.
The hydrophobic residues are shown in green residues that have H bonds in purple
and the polar residues are colored blue.
3D Structure geometry optimization and energy minimization of the M2 channel
protein: The process of geometry optimization and energy minimization was
done using the software MOE 2008-10. The first stage was carried out with 3D
protonation; this was done to alter the enzyme to a protonated state. The application
of this 3D protonation was to change the state of the enzyme's ionization level
and display the position of hydrogen atoms in the crystal structure. The existence
of this hydrogen atom was required in the process of molecular mechanics, dynamics,
or the calculation of electrostatic interactions. Then, the potential parameter
settings in the setup menu were modified. Parameters to be regulated were related
to force field. Selected force field was MMFF94x (Merck Molecular Force Field
94x). It corresponds to peptides, proteins and DNA. MMFF94x was used as a force
field because it is considered better than others (Halgren,
1999). Moreover, the sensitivity of the enzyme with a ligand geometry optimization
was quite high. MMFF94x can also put hydrogen atoms on the most appropriate
position so it is useful also for the validation of the hydrogen atom positions
in the solvation state of water molecules. The selected type of gas phase solvation
due to the molecular docking stage of the enzyme made in the circumstances would
require the removal of rigid solvation energy (Tambunan
et al., 2010). The addition of load (partial charge) by using the
current method of force field parameters was conducted. The purpose of this
addition was to ensure that the proper charge of protonated protein was in the
natural state. Henceforth, the docking will run according to its natural state.
A fixed hydrogen treatment was used to repair the structure if there was a loss
of hydrogen molecule. The next process of protein energy minimization was using
MOE 2008-10 software. Energy minimization was done by the RMS gradient of 0.05
kcal Å-1 corresponding to the protein. The purpose of energy
minimization was to eliminate undesirable interactions (bad contact) of the
structure. Nurbaiti et al. (2010) stated that
the initial coordinates of the biomolecules were generally obtained from X-ray
crystallography or 3D structure modeling and the distance between each atom
could be very close or very far from their equilibrium position. Geometry discrepancies
caused the occurrence of undesirable interactions (bad contact) and high-energy
steric effects that could result in an unstable simulated system. Moreover,
the simulation parameters were adjusted to approximate the real conditions.
Then, the minimization process was carried out. It was necessary for restoring
the positions of atoms that are not suitable for geometry because the resulting
potential energy is too low for the system (Nurbaiti et
Preparation of ligand
Determination of ligands as inhibitors: Determination of protein ligands
as inhibitors of the M2 channel was done by modifying the active group of isopinocampheylamine
compound. According to Zhao et al. (2011), this
compound is an inhibitor for M2 channel protein that can replace amantadine
and rimantadine. The mechanism of M2 channel protein inhibitor that is blocking
the activity of ion channel protein M2 channel of influenza virus A works by
inhibiting viral replication by blocking the flow of protons. The amino group
of amantadine and rimantadine, which are crucial as a barrier towards the proton
flow, also serve as pharmacopore. Pharmacopore is the group that play a role
in the protons interaction, whereas, the adamantyl group serves as a framework
used to maintain the steric effect. Based on the structure-activities relationship
of amantadine, (1R, 2R, 3R, 5S)-(-)- isopinocampheylamine has a structure that
may play a role in inhibiting M2 protein channel (Zhao et
|| Structure of (1R, 2R, 3R, 5S)-(-)-isopinocampheylamine
|| Modification of (1R, 2R, 3R, 5S)-(-)-isopinocampheylamine
Primary amine group in compound (1R, 2R, 3R, 5S)-(-)- isopinocampheylamine
could play the same role with the amino group in compound amantadine which is
a barrier of proton flow in the influenza virus A. The bicyclic ring of compound
(1R, 2R, 3R, 5S)-(-)- isopinocampheylamine also play an equal role with the
adamantyl group in amantadine, that is, a framework that provides steric effects
To increase the potency of the (1R, 2R, 3R, 5S)-(-)- isopinocampheylamine compound, this study carried out modifications on the existing primary amine group. Modification of (1R, 2R, 3R, 5S)-(-)- isopinocampheylamine was done in two forms. The first form modifies primary amine into secondary amine. Modification of the primary amine into a secondary amine group is expected to increase the activity of these compounds, whereas, the second form is the form of imine. The modified form of the C = N bond is expected to have better interaction with the target protein (Fig. 4).
Active groups that were used to modify the (1R, 2R, 3R, 5S)-(-)- isopinocampheylamine
were chosen based on the M2 protein channel inhibitor that was synthesized in
an earlier study by Balannik et al. (2009) and
Zhao et al. (2011). The active groups are listed
in Table 1. This study proved that the active groups were
effective for inhibiting M2 proton channel of influenza A H1N1 virus. From the
results of modifications of the active groups, 52 ligand candidates for isopinocampheylamine
were obtained. A further screening process was carried out to obtain the best
three ligand candidates as an inhibitor of the M2 protein channel.
|| Structure of active groups
Design of the ligands 3D structure: Design of the 3D structure modification of (1R, 2R, 3R, 5S)-(-)- isopinocampheylamine was done using the ACDLabs software. This design was done to prepare the 3D structure modification of isopinocampheylamine compounds as ligands to be used in molecular docking processes. Software ACDlab could describe the position of the active groups of bioactive compounds in accordance with the isolation results.
Storage format of the design modifications was the MDL molfile. Then the format was converted to ligand design using software VEGA MDLmol zz. It means that the software MOE 2008.10 can read the design of these ligands.
Geometry optimization and energy minimization of the ligands' 3D structure:
Ligands with MDL mol format were imported into the database viewer (dv) MOE
2008.10 to do geometry optimization and energy minimization. The process of
geometry optimization begins with washing of the designed ligands contained
in the database viewer MOE 2008-10. The goal is to improve the position of the
hydrogen atoms contained in the ligands and also to improve the structure of
the ligands. Then, the process was performed using the MMFF94x force field optimization.
Partial load settings using the partial charge ligands were conducted with the
parameter method of MMFF94x. The parameter method used was appropriate for peptides,
proteins and DNA. Then, the energy minimization process was done with the aim
of eliminating undesirable interactions. Energy minimization process was performed
on the ligands with RMS gradient 0.001 kcal Å-1 (Singh
et al., 2007). The purpose of the energy minimization process is
to minimize protein interactions in order to eliminate bad contact.
Molecular docking: The M2 channel protein and ligand after optimization
were directed to molecular docking processes. The molecular docking process
was performed using MOE 2008-10 software. Molecular docking was designed to
find the proper conformation bond between the ligand and the receptor. The ligand
docking process can form a complex with the enzyme and iterate it to the most
optimal conformation (Tambunan et al., 2010).
In this study, the docking process was also used for the screening of a number
of candidate inhibitors to obtain the best ones that can be bound to the active
or binding site of proteins (Teodoro et al., 2001),
as well as their potential for development as drugs. Docking process in this
study was conducted between 52 ligand candidates and 3 standard ligands (amantadine,
rimantadine and isopinocampheylamine). The M2 channel protein functional sites
are His37, Trp41 and Asp44. In the process of docking, the protein was pegged
to the rigid condition while the ligand was conditioned on the flexible state,
in order for free movement and rotation.
The parameters set in the docking process are London dG scoring function. Scoring
function is useful in measuring the biological activity based on the bonding
and interactions that occur between the ligand and the target protein (Nylander,
2007). Scoring function in MOE 2008.10 is based on force-field scoring and
is usually calculated with two energies, that is, the ligand-receptor interaction
energy and ligand internal energy (Nylander, 2007). Scoring
function is used to retain the London dG (views) of 100, without duplication
(Mazur et al., 2010). The retain
parameter aims to regulate the amount of the best ligand conformation display.
London dG shows the magnitude of the Gibbs free energy of binding (ΔGbinding)
of each position between the ligand and the enzyme produced by the equation:
where, c is the average rotational and translational entropy gained or released,
Eflex is the energy state that reduced the flexibility of the ligand,
fHB is the size of the imperfection of the hydrogen bonds geometry,
CHB is the energy of an ideal hydrogen bond, FM is the
size of the imperfections of the geometry of the metal ligations, cM
is the energy of an ideal metal ligation and Di is the solvated energy
of atom i (MOE tutorials, 2008).
Other parameters were performed in the docking process through software MOE.2008-10
by setting the triangle matcher. Due to the placement of the MOE, the default
method was used to demonstrate the random motion of ligand in the enzyme active
site to produce an optimal bonding orientation. Triangle matcher was used to
orient the ligand in the active site groups based on the charge and spatial
fit (Cook et al., 2009). The triangle pose matcher
used as much as 1000 which is the default of MOE-Dock program. Stages of refinement
were used to make further improvements. Refinements were performed using the
force field by the results obtained and it was more accurate when compared with
GridMIn using electrostatic calculations on the minimization process (Feher
and Williams, 2009). The default setting of refinement forcefield using
6 Å pocket cut off, or the distance of receptors, was included in the
docking process (Feher and Williams, 2009). Subsequently,
retain (view) parameter of the last refinement results was obtained one by one.
Therefore, it was one of the most optimal conformations of each ligand.
Analysis of docking: There are three important things that are generated in docking simulations. The first is the orientation result and position of a ligand as an inhibitor of the enzyme. The second is to identify compounds that have affinity to the protein from the database of available compounds. The third is to predict the affinity of the enzyme molecule that has a targeted simulation docking. These three are scoring functions (London dG) that estimate the value of the ΔGbinding in kcal mol-1. The docking process was carried out 4 times at the screening stage to get the best of the three ligands.
Free energy association (ΔGbinding) and inhibition constants:
Tendency of the bond strength may be referred to as the affinity of the ligand
to a receptor or enzyme. It can be determined by looking at the value of the
ΔG0 binding affinity (kcal mol-1) generated during
protein-ligand complexes formation. High affinity of a ligand to the protein
is produced from a large intermolecular force between ligands within the protein,
while the low affinity of a ligand to the protein is produced from a small intermolecular
force between the protein-ligand complexes. When binding free energy values
(ΔGbinding) quantified by the KA constant, it is
assumed that the biological activity in the conditions of thermodynamic equilibrium
is in the formation of protein-ligand complex [EI]. There is a relationship
between the values of the bond free energy (ΔGbinding) and those
of the inhibitor constants (Ki) which are shown in the following thermodynamic
equation (Kitchen et al., 2004):
Based on the formula, the lower or higher the negative value of the ΔGbinding, the stronger the protein-ligand complexes. This is because the stability and strength of non-covalent interactions in protein-ligand complexes can be seen from the large free energy released during the interaction on the protein-ligand complexes formation. In the MOE software, Gibbs free energy (ΔG0) is denoted by S, which shows the total amount of the docking final stage. Value of S has the same score with E_refine, where E_refine is the total energy of the complex bond docking. Docking results showed that from the obtained 52 ligand candidate compounds, there are three best ligands with a negative value ΔGbinding.
From Table 2, there are three ligands (ligands A20, B18 and B20), which have a value of ΔG0 that shows more negativity than that of amantadine, rimantadine and isopinocampheylamine standard. Also from Table 2, the inhibition constant value can also be seen. The dissociation constant of the decomposition herein referred to as Ki may give an idea about the affinity between ligand and decomposition. This constant is reciprocal of the equilibrium constant, thus, Ki is formulated as follows:
where, E is the enzyme, L is the ligand and EL is the enzyme-ligand complex. The smaller the value of Ki, the more the equilibrium reaction tends toward complex formation. Complex protein-ligand binding affinity is said to have a good Ki value if it is in the micromolar scale. Data docking results showed that all ligands have an estimated value of Ki on this scale. Docking data of the software MOE showed inhibition constants in the pKi. This means that the greater the value of the pKi, the smaller the ligands Ki. The value of pKi can be used to determine the level of stability in the formation of protein complexes with ligands. From the docking of ligand A20, pKi value is 6.292. The value indicates that A20 has an affinity with the ligand and interacts more strongly with the M2 protein. When compared to other ligands, it forms a complex with the channel better.
|| Docking simulation results of binding free energy data
Hydrogen bonding and contact residues: Apart from the value of Gibbs
free energy, or ΔG and inhibition constants, parameters that can be analyzed
from the docking process is the interaction between protein and ligand. One
of the observed parameters is hydrogen bond interaction. Hydrogen bond is defined
as an intermolecular or intramolecular force that occurs between atoms that
have a high electronegativity of hydrogen atoms and is covalently bonded to
an electronegative atom (Nurbaiti et al., 2010).
The criterion for the occurrence of this hydrogen bond is that the distance
between the hydrogen with electronegative atoms must be in the range of 2.5-3.5
Å. Hydrogen bonding that occurs in protein-ligand complexes can be identified
and analyzed in ligand interaction using software MOE 2008-10.
From the analysis of hydrogen bonding in the docking process, it could be seen that the hydrogen bonding interaction between the M2 protein channel and the standard ligands is less than the hydrogen bonding between the M2 protein channel and the three best ligands. In this study, the ligand modification compound of (1R, 2R, 3R, 5S-(-)- isopinocampheylamine showed better activity. From the analysis of hydrogen bonds, it can be seen that these ligands tend to bind to Asp44 which is the target residue. Asp44 residue acts as a channel lock on the M2 protein channel. In normal circumstances, Asp44 residue binds to Trp41 residue to maintain a closed ion channel but at a lower pH, Trp41 residue could be protonated, as such, it weakens the bond between the two ion channels. Thus, it could be opened and may help the virus to release its genetic material. Due to the interaction of the Asp44 residue with ligands, it became more difficult for these residues to be protonated in order to maintain a closed conformation. Aside from being a channel lock, Asp44 residue acts as a proton exit channel on the M2 protein and may also impede the flow of protons. With so many hydrogen bonds on Asp44 residues, it is expected that inhibition on ligand modification function compounds (1R, 2R, 3R, 5S)-(-)- isopinocampheylamine will be better. Table 3 shows that most hydrogen bonding interactions with Asp44 residue occurred in ligand A20. This occurrence was observed on three hydrogen bonds with Asp44 residue in the M2 protein channel (two bonds on chain B and one double bond on chain A). In B18 ligands, two hydrogen bonds occurred with Asp44 residue in the chain of the M2 protein channel D, while the B20 ligands occurred in one hydrogen bond with Asp44 residue chain C. From the results of the standard ligand interaction, it was also seen that amantadine and rimantadine lost their effectiveness. In rimantadine, hydrogen bonding did not occur with the M2 protein channel. Contact residues between rimantadine and the M2 channel protein was not on a functional site. On the interaction between the protein isopinocampheylamine, one hydrogen bond was seen with the functional site residues of Asp44 and it contacted only with the Leu40 residue which is not a functional site on the M2 protein channel.
|| M2 protein channels amino acid residue that form hydrogen
bonds with the ligand
|*Standard; residue in bold font are the functional site residue
of M2 channel protein
Apart from hydrogen bonding interactions, the analysis of docking results can
also be viewed from the contact between the protein residues and the ligand.
Interaction of non-covalent or non-bonding occurring between the protein and
ligand can increase the ligand affinity to the protein. Non-bonding interactions
represent a flexible interaction between pairs of atoms and particles. Two types
of non-bonding interactions that can lead to the most common change in potential
energy are the electrostatic interaction and van der Waals interactions. Therefore,
there should be an analysis of the docking simulation results of the contact
residues protein-ligand complexes, such that the type of enzyme residue
that interacts with the ligand can be known. The analysis shows that the contact
residues of the ligand modification of compound (1R, 2R, 3R, 5S)-(-)- isopinocampheylamine
are better than those of the standard ligand. In addition to the functional
site of interaction with Asp44, ligand modification of compound (1R, 2R, 3R,
5S)-(-)- isopinocampheylamine also has a contact residue with another functional
site. This is Trp41 residue in ligands A20 and B20. This is expected to increase
the effectiveness of the M2 protein inhibitors of this channel and it will be
displayed following the 2D visualization and 3D complex interactions of the
best candidate ligand and standard ligand with M2 protein channel. 2D visualization
is shown on the left while 3D visualization is shown on the right. In 3D visualization,
hydrogen bonds between ligands with proteins are marked with purple colored
dashed lines (Fig. 5-10).
Drug scan analysis: The three best ligand docking properties analysis was tested for their similarities with an existing drug (drug likeness) using the Lipinski rules (Lipinski's Rule of Five). Lipinski's Rule of Five helps to distinguish the difference between drug-like and non drug-like molecules by taking into account the extent of absorption or permeability of the lipid bilayer present in the human body. The rule predicts that the drug likeness probability is high enough if it has two or more criteria as given thus:
||A molecular weight of less than 500 mg mol-1
||A high lipophilicity (logP less than 5)
||Hydrogen bond donors less than 5
||Hydrogen bond acceptor less than 10
||Refractory molarity between 40 and 130 (optional) (SCFBIO-IITD)
|| Visualization of 2D and 3D interaction of A20 ligand with
M2 channel protein
|| Visualization of 2D and 3D interaction of B18 ligand with
M2 channel protein
|| Visualization of 2D and 3D interaction of B20 ligand with
M2 channel protein
Table 4 shows the result of drug likeness based on ligand
screening using Lipinski's online software filter (SCFBIO-IITD). From Table
4, it could be seen that the best ligand candidates molecular weight
is less than 500 mg mol-1, which fulfills the criteria of Lipinsky's
Rule. The value of the hydrogen bond donor and acceptor was also less than the
maximum extent permitted by Lipinsky's Rule. LogP values of the three best ligands
or ligand candidates also met the standards of Lipinsky's Rule. LogP value is
the partition coefficient that is defined as the ratio of the concentration
of a molecule in octanol and water.
|| Visualization of 2D and 3D interaction of amantadine with
M2 channel protein
|| Visualization of 2D and 3D interaction of rimantadine with
M2 channel protein
|| Drug scan result of the best three ligands and standard ligands
|| Visualization of 2D and 3D interaction of isopinocampheylamine
with M2 channel protein
Analysis of drug scans was performed on the two best ligand dockings which
has the lowest scoring function. LogP value is related to the hydrophobicity
of drug molecules. The larger the logP value, the more hydrophobic the molecule.
Drugs molecule that are too hydrophobic tend to have greater toxicity due to
their ability to be held longer in the lipid bilayer and distributed more widely
in the body. This would cause reduction of the selectivity of binding to the
target enzyme. LogP values that are too negative cannot be recommended because
if the compounds are too hydrophilic, it could not pass through the lipid bilayer
and it is possible to interact with water solvent. The last criterion is the
value of the molar refractivity. According to Lipinsky's Rule of molar refractivity,
the values that were allowed were between 40 and 130. The three candidates met
the rules of the ligand as well as the standard ones. Molar refractivity is
a measure of the total polarizability of a drug molecule which is very dependent
on temperature, refractive index and pressure. Determination of the molar refractivity
can be done using the Lorentz-Lorenz formula:
where, M is the molecular weight, η is the refractive index and r is the density whose value depends only on light waves that are used to measure the refractive index. From the analysis of drug-scan using Lipinsky's Rule of Five, it could be concluded that the three best candidates were possible candidates for oral drugs.
In addition to Lipinsky's Rule of Five, the value of drug likeness can also
be seen from the level of oral bioavailability. High oral bioavailability often
becomes an important consideration for developing bioactive molecules as therapeutic
agents (Veber et al., 2002; Tambunan
and Wulandari, 2010). Oral bioavailability is the extent to which a drug
or other materials are available to the target tissue after administration of
a drug or substance (Tambunan and Wulandari, 2010).
|| Oral bioavailability screening results of standards and ligands
Screening level of ligand's oral bioavailability could be done using Egan's
rules (Egan et al., 2000). Based on Egan's Rules,
a molecule is said to have a good level of oral bioavailability if it meets
the following criteria:
First, the screening was done by determining the ligand properties by using
the online software molinspirations (molinspirations.com). The results of the
oral bioavailability screening of the standards and ligands are shown in Table
From the results of oral bioavailability screening of standards and the ligands using Egan's Rules, it is seen that the three best candidates met the criteria of Egan's Rules. It could be concluded that the three best candidates are ligands A20, B18 and B20 which have high levels of good oral bioavailability.
Inhibitor toxicity prediction nature: Testing the toxicity properties
of the three-channel M2 protein inhibitor candidates obtained from the previous
process is important as the basis for drug determination. It is necessary to
predict their adverse effects to living species. One method that can be used
in toxicological studies is qualitative or quantitative structure-activity relationship
(QSARs). The results of studies conducted using animals are the basis of QSARs
in toxicological studies, because it could not guarantee the completeness of
the data required for the toxicological studies. These studies have some limitations
such as funds, requires too much time, availability of adequate laboratory and
the issues about the ethics of using animals as test materials (Gonzalez-Diaz
et al., 2009). To help overcome these problems, several tools have
been developed which can perform quick and inexpensive toxicological studies
or determination of the toxicological properties of a molecule. Some software
developed to assist researchers in determining the toxicological properties
of a molecule are: ToxTree (Ideaconsult Ltd., Bulgaria) and Lazar (http://lazar.in-silico.de/).
Both software are used in determining the toxicological properties of the three-ligand results of previous screening of ligands A20, B18 and B20. Determinations of the toxicological properties of the three ligands are focused on carcinogenicity and mutagenicity because they are an important concern in human health and deal directly with the aim of the drug design.
The difference between the two software is the basis for determining the toxicological
properties of a molecule. The Toxtree is based on rules of Benigni/Bossa for
mutagenicity and carcinogenicity, and was developed by Romualdo Benigni and
Cecilia Bossa from the Instituto Superiore in Sanita, Rome, Italy and approved
by the European Chemical Bureau, Institute for Health and Consumers Protection,
European Commission-Joint Research Centre (JRC) in 2008. Some things to consider
in determining the toxicological properties using Toxtree are the presence or
absence of genotoxic and nongenotoxic structural alerts (SAS) and QSARs determination.
|| Toxtree toxicity prediction results
|| Results of Lazar toxicity prediction
Benigni and Bossa rules are based on the existence of groups which are potentially
seen as having mutagenic and carcinogenic properties of the test compound. Table
6 shows the result of the determination of the toxicological properties
of the software Toxtree.
From the results of the toxicological prediction using software Toxtree, it could be seen that the three ligands have no structural alerts (SAS), which are genotoxic and nongenotoxic and also with this approach, QSARs which are neither mutagenic nor carcinogenic, were not found. As for the software online Lazar, predictions using this software were based on structural equation fragments of a molecule compared with the structural fragment toxicology that exist in databases of known carcinogenicity and mutagenicity and the test results using test animals and microbes. Table 7 shows the result of the determination of the toxicological properties of software Lazar.
The results of software Lazar could predict mutagenic properties of the three ligands in Salmonella Typhimurium by two methods of CPDB and Kazius/Bursi. The three ligands of the test showed negative results. The carcinogen properties were predicted by testing them against animals such as rodents (rodent), rat (rat), murine (mouse) and hamster. The test results using all three animals showed negative carcinogenicity properties; only the B18 ligands showed positive carcinogenic results which could be active against rats. Some parameters in software Lazar did not show any result (not available) for the three ligands. This was caused by the less availability of data in the structural fragment and the inadequate support to predict the outcome of those parameters such as LC50 and the daily dose.
This study obtained 52 ligand modifications for compound (1R, 2R, 3R, 5S)-(-)-isopinocampheylamine. In the process of docking, three of the best ligands were obtained. They have the lowest value of ΔGbinding and the best interaction with the M2 protein channel compared with the standard ligand of amantadine, rimantadine and isopinocampheylamine. The three ligands are A20, B18 and B20. From the obtained results by docking the data, the three ligands form hydrogen bonds with a residual functional M2 channel, that is, Asp44. Analysis of the drug-scan gives good results for the three best ligands based on Lipinskis rules. Toxicological properties of ligands A20, B18 and 820 as a whole are predicted not to be carcinogens and mutagens.
However, further steps like the molecular dynamic simulation to determine the effect of temperature and solvent on ligand interaction with the M2 protein channel and ADME analysis and bioactivity of the ligand which has been designed to determine the treatment of human body systems of ligands, need to be considered.
This research was supported by Hibah Pasca, DIKTI, Indonesia Ministry of Education. The authors are thankful to DRPM, University of Indonesia for facilitating this research. U.S.F.T was supervising this research, R. H was working on the technical details and A.A. P was preparing the English manuscript and re-verifying the data.